Download pdf - Pharmacotherapy, clinical pharmacology and biomarker ... · geriatric patients2,3. Evaluation of medication in the elderly should aim to reduce polypharmacy on the one hand and prevent

Pharmacotherapy, clinical pharmacology and biomarker research in geriatric patients

ISBN-10: 90-393-4419-1 ISBN-13: 978-90-393-4419-4 © 2006 Suzanne V. Frankfort, Utrecht Printed by: Ponsen & Looijen BV, Wageningen, The Netherlands Cover design: “Proefschrift Memory”, Martijn van Dalen, Utrecht

Pharmacotherapy, clinical pharmacology and biomarker research in geriatric patients

Farmacotherapie, klinische farmacologie en onderzoek naar biologische markers in geriatrische patiënten

(met een samenvatting in het Nederlands)

PROEFSCHRIFT

ter verkrijging van de graad van doctor aan de Universiteit Utrecht

op gezag van rector magnificus, prof. dr. W.H. Gispen, ingevolge het besluit van het college voor promoties

in het openbaar te verdedigen op woensdag 10 januari 2007 des middags te 2.30 uur

door

Suzanne Viola Frankfort

geboren op 31 juli 1978 te Gendringen

Promotor: Prof. Dr J.H. Beijnen Publication of this thesis was financially supported by: Janssen-Cilag BV, Tilburg, The Netherlands Stichting KNMP fondsen, Den Haag, the Netherlands Department of Pharmaceutical Sciences, Utrecht, The Netherlands Netherlands Laboratory for Anticancer Drug Formulation (NLADF), Amsterdam, The Netherlands Internationale Stichting Alzheimer Onderzoek, Maastricht, The Netherlands Wyeth Pharmaceuticals BV, Hoofddorp, The Netherlands Lundbeck BV, Amsterdam, The Netherlands AstraZeneca BV, Zoetermeer, The Netherlands

The studies described in this thesis were performed at the department of Geriatric Medicine and the department of Pharmacy & Pharmacology of the Slotervaart Hospital, Amsterdam, The Netherlands.

Contents Preface 9 Chapter 1 Pharmacotherapy of geriatric patients 13

1.1 Evaluation of pharmacotherapy in geriatric patients after performing 15 Complete Geriatric Assessment (CGA) at a diagnostic day-clinic

1.2 The effect of admission to a geriatric ward on medication use: 2002 25 versus 1985

Chapter 2 Digoxin 33

2.1 Role of ABCB1 genotypes and haplotypes in digoxin steady state 35 pharmacokinetics in geriatric patients

Chapter 3 Rivastigmine 49 Chapter 3.1 Bioanalysis of rivastigmine 51

3.1.1 A simple and sensitive assay for the quantitative analysis of 53 rivastigmine and its metabolite NAP 226-90 in human EDTA plasma using coupled liquid chromatography and tandem mass spectrometry

Chapter 3.2 Rivastigmine pharmacotherapy and clinical pharmacology 71

3.2.1 Discontinuation of rivastigmine in routine clinical practice 73 3.2.2 Treatment effects of rivastigmine on cognition, performance of daily 81 living activities and behaviour in Alzheimer’s disease in an outpatient geriatric setting 3.2.3 Identification of responders and reactive domains to rivastigmine in 99 Alzheimer’s disease

3.2.4 Population pharmacokinetics and pharmacodynamics of 111 rivastigmine and its metabolite NAP 226-90 in patients with dementia

Chapter 4 Biomarkers for dementia 129 Chapter 4.1 Genetic biomarkers 131

4.1.1 APOE genotype and allele distribution in geriatric outpatients and 133 healthy volunteers 4.1.2 ABCB1 genotypes and haplotypes in patients with dementia and 141 age-matched non-demented control patients

Chapter 4.2 Proteomic biomarkers 149

4.2.1 Amyloid beta protein and tau in cerebrospinal fluid and plasma as 151 biomarkers for dementia: a review of recent literature 4.2.2 Cerebrospinal fluid biomarkers found for Alzheimer’s disease and 169 Vascular Dementia using Surface-Enhanced Laser Desorption Ionisation-Time of Flight Mass Spectrometry (SELDI-TOF MS) 4.2.3 Serum amyloid beta peptides in patients with dementia and age- 185 matched non-demented controls as detected by Surface-Enhanced

Laser Desorption Ionisation- Time of Flight Mass Spectrometry (SELDI-TOF MS)

Conclusions and Perspectives 201 Summary 205 Samenvatting 211 Dankwoord 217 Curriculum Vitae 221 List of Publications 222

9

Preface Optimisation of pharmacotherapy and research into the diagnosis of different types of dementia (Alzheimer’s disease, vascular dementia, Lewy body disease, frontotemporal disease) and treatment of dementia are of primary importance within the field of geriatric medicine. Challenges in the dementia field are optimisation of the current symptomatic therapy with cholinesterase inhibitors, the development of disease-modifying therapies and diagnosing dementia as early as possible before the biological processes have already started. Biomarker research in body fluids, for example cerebrospinal fluid and serum/plasma, plays an important role in early diagnostics. First, this thesis deals with optimisation of general pharmacotherapy and clinical pharmacological investigations of digoxin, a frequently used drug, in geriatric patients. Second, this thesis aims to gain further insight into rivastigmine therapy, an example of a symptomatic cholinesterase inhibitor, in outpatients with Alzheimer’s disease. Third, the thesis deals with the development of biological markers for the diagnosis of dementia. The number of elderly patients that visit the department of Geriatric Medicine is growing due to the increase of the number of elderly in the population. Geriatric patients often suffer from complex co-morbidity and subsequent use of multiple drugs. This may lead to an increase in drug-drug interactions and incidence of adverse events1. It is thus important to reduce polypharmacy, but it is recognised, however, that undertreatment is also apparent in geriatric patients2,3. Evaluation of medication in the elderly should aim to reduce polypharmacy on the one hand and prevent undertreatment on the other. This thesis deals with pharmacotherapy in geriatric patients when they visit the diagnostic day-clinic (chapter 1.1) and the geriatric ward (chapter 1.2) and with the changes that are made after pharmacotherapy evaluation by a geriatrician. Clinical pharmacological research in the geriatric population is still in its infancy as patients aged over 65 are frequently excluded from this type of clinical research. This thesis deals with the role that ABCB1 genotypes may play in the steady-state pharmacokinetics of digoxin in geriatric patients using digoxin as part of their therapeutic regimen (chapter 2.1). Rivastigmine is one of the registered cholinesterase inhibitors for the treatment of mild-to-moderate Alzheimer’s disease4,5. It is known, however, that patients using rivastigmine often discontinue therapy and that treatment effects of rivastigmine differ substantially between individuals6. The rate of and reason for discontinuation in clinical practice and possible risk factors for discontinuation within the first 6 months of treatment were investigated (chapter 3.2.1). In addition, research investigating treatment effects of rivastigmine during 30 months (chapter 3.2.2) and reactive domains and characteristics of responders (chapter 3.2.3) are described. A population pharmacokinetic model was developed for rivastigmine and its metabolite NAP226-90 and pharmacokinetic parameters

10

were estimated and related to outcome measures (cognition, activities in daily living, behaviour) to investigate whether differences in pharmacokinetics may explain differences in treatment effects (chapter 3.2.4). With the advance of new disease-modifying therapies, for example the secretase-inhibitors7, it becomes important to establish an early dementia diagnosis. Biological markers of genetic origin may add to the dementia diagnosis (chapter 4.1.1 and 4.1.2). Chapter 4.2.1 summarises recent developments in the use of amyloid β and tau protein in cerebrospinal fluid and plasma as biomarkers for dementia. Proteomics can be used to identify possible biomarkers in biological fluids. Using the new method Surface Enhanced Laser Desorption/Ionisation-Time of Flight Mass Spectrometry (SELDI-TOF MS), biomarkers for Alzheimer’s disease and vascular dementia were investigated in cerebrospinal fluid (chapter 4.2.2) and in serum (chapter 4.2.3).

References 1. Stewart RB, Cooper JW. Polypharmacy in the aged: practical solutions. Drugs Aging 1994;4: 449-61. 2. Mendelson G, Aronow WS. Underutilization of warfarin in older persons with chronic nonvalvular atrial

fibrillation at high risk for developing stroke. J Am Geriatr Soc 1998; 46: 1423-1424. 3. Ghosh S, Ziesmer V, Aronow WS. Underutilization of aspirin, beta blockers, angiotensin-converting enzyme

inhibitors, and lipid-lowering drugs and overutilization of calcium channel blockers in older persons with coronary artery disease in an academic nursing home. J Gerontol A Biol Sci Med Sci 2002; 57: 398-400.

4. Corey-Bloom J, Anand R, Veach J. A randomized trial evaluating the efficacy and safety of ENA 713 (rivastigmine tartrate), a new acetylcholinesterase inhibitor, in patients with mild to moderately severe Alzheimer’s disease. International Journal of Geriatric Psychopharmacology 1998; 1:55-65.

5. Rösler M, Anand R, Cican-Sain A, et al. Efficacy and safety of rivastigmine in patients with Alzheimer’s disease: international randomised controlled trial. BMJ 1999; 318:633-40.

6. Rockwood K, MacKnight C. Assessing the clinical importance of statistically significant improvement in anti-dementia drug trials. Neuroepidemiology 2001;20:51-56.

7. Siemers ER, Quinn JF, Kaye J, et al. Effects of a gamma-secretase inhibitor in a randomized study of patients with Alzheimer disease. Neurology 2006;66:602-4.

11

12

13

CHAPTER 1

PHARMACOTHERAPY OF GERIATRIC PATIENTS

14

15

CHAPTER 1.1 Evaluation of pharmacotherapy in geriatric patients after

performing Complete Geriatric Assessment (CGA) at a diagnostic day clinic

S.V. Frankfort, C.R. Tulner, J.P.C.M. van Campen, C.H.W. Koks, J.H. Beijnen

Abstract

Background: Elderly patients often take multiple drugs. It is known that polypharmacy, i.e. use of five or more drugs, may lead to drug interactions and adverse events. However, undertreatment of conditions or illnesses is also a concern in geriatric patients. A centralised review of both diagnoses and medication may play a key role in optimising pharmacotherapy in geriatric patients. The aims of this study were to evaluate quality and appropriateness of medication after performing complete geriatric assessment (CGA) and medication review at a diagnostic geriatric day clinic, to investigate reasons for drug changes, and to determine whether medication review leads to a reduction in the number of drugs used. Methods: A chart review was performed in 702 patients (mean age 82.0 years, range 57.1-104.1) who underwent a CGA at a diagnostic geriatric day clinic. Medication at admission, changes in medication and reasons for changes were noted. Results: Vitamins, for example folic acid and vitamin B12 (cyanocobalamin), and trimethoprim for urinary tract infections were the most frequently started medications after CGA and medication review. The number of drugs used was reduced in only a minority of patients (11.7%); reasons for discontinuation were a diagnosis that was not longer relevant (38.8%), adverse events (33.2%) and identification of better pharmacotherapeutic options (21.9%). In 69.2% of the cases, a new diagnosis was the reason for starting new medication, followed by osteoporosis prophylaxis (15.0%) and improvement in pharmacotherapy (10.6%). At admission, patients were taking a mean number of 4.6 drugs (range 0-17). A mean of 0.8 drugs (range from reduction of 5 to addition of 7) had been added per patient resulting in a mean number of 5.4 (range 0-18) prescribed drugs at discharge. Conclusion: Evaluation of medication in patients after performing CGA at the geriatric day clinic resulted in relevant medication changes. The main reason for prescribing new drugs was a new diagnosis. Absence of a relevant medical indication was the main reason for stopping drugs. CGA and medication review resulted in a mean net addition of 0.8 drugs per patient.

Clinical Drug Investigation 2006;26:169-74

Chapter 1.1

16

Introduction Healthcare for older patients has undergone many changes during the last decade. In The Netherlands, the emphasis has shifted from inpatient to ambulatory care. The diagnostic day clinic, for example, has been a unique part of the department of geriatric medicine in our hospital since 1995 and has been introduced into 30 hospitals in The Netherlands since then. At such clinics, a complete geriatric assessment (CGA) is carried out by a geriatrician, predominantly in outpatients, most of whom have been referred by a general practitioner. These patients are referred to a geriatrician because they have both acute and chronic problems, mostly in combination with multiple morbidities and medication use and often accompanied by psychiatric illnesses or functional decline. Review of medical history, physical examination, laboratory tests, medication review, chest x-ray, testing of functional state, cognition and mood are all carried out over a period of one day. At the end of this day, diagnoses are made and medication is sometimes changed or pharmacotherapeutic advice is given to the general practitioner. Further treatment can take place through the general practitioner or the geriatrician and in some patients involving other specialists as well. From the patient’s point of view, it is of great value that all diagnostic tests are conducted over one day so that the patient does not need to visit the outpatient department several times. From the geriatrician’s standpoint, the day clinic is a practical setting to perform CGA and thereby obtain insight into patients’ diagnoses and drug use. Frail elderly patients are often diagnosed with multiple diseases and visit several specialists besides their general practitioner. This may lead to multiple prescriptions and specialists not knowing which drugs are being prescribed by their colleagues. A centralised review of both medications and diagnoses may therefore play a key role in optimising pharmacotherapy in geriatric patients. The diagnostic process can lead to new diagnoses and prescriptions, but may also reveal inappropriate drug use. This should of course be minimised in the geriatric population, for whom there is strong evidence of a sizeable and consistent negative effect of inappropriate drug use on patients’ health status1 and an associated increase in use of outpatient services and more rapid hospitalisation2. It is preferable to reduce polypharmacy, defined as the use of five or more drugs3, because this can lead to drug-drug interactions and adverse events4, and the number of drugs is inversely correlated with adherence to the therapeutic regimen5. On the other hand, however, undertreatment has also been described6,7. Therefore, evaluation of medication in geriatric patients should aim to reduce polypharmacy on the one hand and prevent undertreatment on the other. To our knowledge, this is the first study describing medication evaluation at a geriatric diagnostic day clinic. The aims of this study were: (a) to evaluate the quality and appropriateness of pharmacotherapy after performing a CGA and medication review at a geriatric day clinic;

Pharmacotherapy at a geriatric day clinic

17

(b) to investigate reasons for changes in pharmacotherapy; and (c) to investigate if medication review is associated with a reduction in drugs used.

Patients and methods Patients and data collection This study was carried out at the geriatric day clinic of a general hospital (Slotervaart Hospital) in Amsterdam, The Netherlands, in 2002. Patients were included and underwent CGA if it was their first visit to the day clinic. Patients were excluded if their medical records were incomplete. A retrospective chart review was performed that included date of visit, date of birth, sex, medication at admission and medication that was started and discontinued, reasons for changes in medication and new diagnoses after performing CGA. Medication at admission was counted as the medication that was brought to the day clinic and that was actually ingested by the patient. Reasons for changes were sometimes explicitly noted. In other cases, reasons were assumed by the person doing the chart review. In addition, changes are sometimes based upon protocols that are created and used at the geriatric department. Only changes in medication were counted; changes in dosage were not taken into account. Data were incorporated in a database in Visual FoxPro 6.0 (Microsoft Corporation, Redmond, WA). Design Illnesses or conditions diagnosed after geriatric assessment at the day clinic were classified according to the International Classification of Diseases and Related Health Problems, 10th revision (ICD-10)8. Drugs were classified according to the Anatomical Therapeutical Chemical (ATC) classification index9. Descriptive analyses of the medication used on admission to the geriatric day clinic and changes in medication after therapy evaluation were carried out. The change in number of drugs was also counted in each patient to determine whether medication review was associated with a reduction in the number of drugs.

Chapter 1.1

18

Results Patients Seven hundred and two patients were included in this study; 11 patients were excluded because their medical records were incomplete. The cohort consisted of 464 women (66.1%) and 238 men (33.9%). Mean age of these patients at admission was 82.0 years (SD 6.8 years) with a range of 57.1-104.1 years. Diagnoses The most frequently made diagnosis was dementia, including mild cognitive impairment, (45.0% of patients), followed by vitamin deficiencies as confirmed by laboratory tests (26.4%), constipation (16.0%), depression (15.8%), urinary tract infection (13.0%), osteoporosis (9.5%), anaemia from chronic diseases (7.8%), arthrosis (6.8%), diabetes mellitus (6.6%) and electrolyte disorders (5.4%), Parkinson disease (5.4%) and atrial fibrillation (5.4%). Medication at admission The ATC codes for drug groups that were frequently used on admission to the geriatric day clinic were: (A) gastrointestinal tract and metabolism; (B) blood and blood-forming organs; (C) cardiovascular agents; and (N) central nervous system agents. Table 1 lists the top ten of medications being taken at admission. Acetylsalicylic acid (25.8%), paracetamol (acetaminophen) (20.9%) and furosemide (18.0%) were the most frequently used medications. Evaluation of medication Table 1 shows the top ten medications started and the medications discontinued after evaluation at the day clinic. Hydrochlorothiazide was discontinued in 3.6% of patients, haloperidol in 2.6% of participants, pipamperone (an antipsychotic agent) in 2.1% and ferrous fumarate in 1.6% of patients. Vitamins, for example folic acid, vitamin D and vitamin B12 (cyanocobalamin), were started frequently, in 18.9%, 13.4% and 7.3% of participants, respectively. Trimethoprim was added to drug regimens in 4.6% of subjects, and risperidone in 4.2%.


19

Tabel 1. Top ten medications on admission to and top ten medications started or discontinued at the geriatric day clinic (% of total patients; n=702).

Reason for change of medication after pharmacotherapy evaluation The reasons for changes in pharmacotherapy after CGA and medication evaluation are presented in table 2. In 69.2% of started medication, a new diagnosis was the reason for starting new medication, followed by osteoporosis prophylaxis (15.0%) and improvement in pharmacotherapy (10.6%). Examples of medications started for new diagnoses were trimethoprim, for urinary tract infections, and citalopram for depression, whereas calcium carbonate and vitamin D were started for osteoporosis prophylaxis. Risperidone, usually to replace the typical antipsychotic haloperidol, is an example of a drug that was started to improve pharmacotherapy. No longer existing or relevant diagnosis accounted for 38.8% of all discontinued drugs, followed by adverse events (33.2%) and availability of better pharmacotherapeutic options (21.9%). Hydrochlorothiazide, ferrous fumarate and betahistine are examples of medications that were often discontinued because they were prescribed for diagnoses that no longer appeared to be significant after geriatric assessment. Pipamperone was discontinued because of cholinergic adverse events. Haloperidol was discontinued because a better pharmacotherapeutic alternative was available.

Medication

On admission [n (%)]

Discontinued at day clinic [n (%)]

Started at day clinic [n (%)]

1 Acetylsalicylic acid [181 (25.8)] Hydrochlorothiazide [25 (3.6)] Folic acid [133 (18.9)] 2 Paracetamol (acetaminophen) [147 (20.9)] Haloperidol [18 (2.6)] Vitamin D [94 (13.4)] 3 Furosemide [126 (18.0)] Betahistine [18 (2.6)] Magnesium oxide [82 (11.7)] 4 Lactulose [86 (12.3)] Pipamperone [15 (2.1)] Vitamin B12 [51 (7.3)] 5 Temazepam [81(11.5)] Lactulose [13 (1.9)] Calcium carbonate [41 (5.8)] 6 Digoxin [79 (11.2)] Ferrous fumarate [11 (1.6)] Citalopram [38 (5.4)] 7 Acenocoumarol [77 (11.0)] Triamterene [11 (1.6)] Trimethoprim [32 (4.6)] 8 Isosorbide dinitrate [70 (10.0)] Furosemide [10 (1.4)] Risperidone [29 (4.2)] 9 Oxazepam [65 (9.3)] Cinnarizine [8 (1.1)] Paracetamol [28 (4.0)] 10 Hydrochlorothiazide [63 (9.0)] Diclofenac [7 (1.0)]

Tramadol [7 (1.0)] Paracetamol [7 (1.0)]

Acetylsalicylic acid [27 (3.8)]

Chapter 1.1

20

Table 2. Reasons for medication changes at the geriatric day clinic (% of total discontinued or total started medication).

Change in number of drugs after pharmacotherapy evaluation The mean number of drugs used per patient at admittance was 4.6 (range 0-17). A mean addition of 0.8 drugs (range from reduction of 5 to addition of 7) per patient resulted in a total of 5.4 (range 0-18) drugs used per patient after CGA and medication review at the day clinic. The number of drugs was reduced only in a minority of patients (11.7%). Reduction of one drug was achieved in 52 patients (7.4%), whereas a reduction of two drugs was achieved in only 20 patients (2.8%). In 246 patients (35.0%), the number of added drugs equalled the number of discontinued drugs or medication remained unchanged. However, in a majority (53.3%) of the day clinic patients, the number of drugs increased. Addition of one, two and three drugs occurred in 183 (26.1%), 106 (15.1%) and 57 (8.1%) patients, respectively.

Discussion After CGA and medication review at our diagnostic geriatric day clinic, primarily vitamins, magnesium oxide, calcium carbonate, trimethoprim, citalopram and risperidone were added to the pharmacotherapeutic regimen, whereas diuretics, antipsychotics (pipamperone, haloperidol) and medication for dizziness/vertigo (betahistine, cinnarizine) were discontinued most frequently. New diagnoses, including folic acid and vitamin B12 deficiencies, accounted for most of the newly prescribed drugs. Discontinuation of drugs, although relatively uncommon, was mostly because the indications for prescription no longer appeared to be relevant. Evaluation of pharmacotherapy at our geriatric day clinic thus resulted primarily in addition of medication with only a minority of patients experiencing discontinuation of medications, resulting in a mean net addition of 0.8 drugs per patient.

Reason Started [no.(%)] Discontinued [no. (%)]

Adverse events 130 (33.2) Better pharmacotherapeutical options 101 (10.6) 86 (21.9) Diagnosis not relevant anymore 152 (38.8) Insufficient effect 15 (3.8) New diagnosis (including folic acid and vitamin B12 deficiencies) 659 (69.2) Osteoporosis prophylaxis (vitamin D, calcium) 142 (15.0) Preventive cardiac medication (acetylsalicylic acid, acenocoumarol) 26 (2.7) Prevention adverse event other medication 12 (1.3) Other 12 (1.2) 9 (2.3) Total 952 (100) 392 (100)


21

Optimisation of geriatric pharmacotherapy involves initiating new therapy in case of a new diagnosis, improving drug use or discontinuing medication when its use is no longer appropriate. Vitamin supplementation was frequently started at the day clinic; two diagnosed deficiencies were treated (folic acid and vitamin B12) and purely preventive supplementation with vitamin D was also started. It is known that an inadequate folate status is associated with increased risk for chronic diseases that may have a negative impact on the health of the aging population10. A moderately reduced vitamin B12 level is associated with vascular disease and neurocognitive disorders such as depression and impaired cognitive performance. Furthermore, poor vitamin B12 status is assumed to be involved in the development and progression of dementia11. Vitamin D and calcium carbonate were often started at the day clinic because these reduce the risk of hip fracture and other nonvertebral fractures among elderly women12. Depression was one of the most frequent diagnoses made at the day clinic and is one of the leading causes of poor quality of life in the elderly13. If medical treatment is warranted, then citalopram is one of the better alternatives in the elderly, because it is efficacious in this age group, has negligible cholinergic and histaminergic adverse events14 and has a low potential for pharmacokinetic interactions because of its low affinity for metabolising cytochrome P450 enzymes15. Discontinuation of drugs if they are no longer useful is necessary, particularly in older people who are often taking multiple drugs. Discontinuation of drugs at the day clinic was, however, achieved only in a minority of patients. Iron supplementation, for example, was discontinued in patients demonstrating a regular iron state. This is important as iron may induce constipation, the incidence of which increases with age16. Optimisation of drug treatment is also possible when better pharmacotherapeutic options are available, for example in our cohort with the change of haloperidol to risperidone for treatment of behavioural and psychological symptoms of dementia. In this condition, for which long-term antipsychotic treatment is expected, risperidone is preferred because of possibly fewer adverse events17,18.

Conclusion Evaluation of medication in patients after CGA at our geriatric day clinic resulted in many changes in pharmacotherapy. In a majority of patients, new drugs were prescribed and preventive medicine played an important role in these changes. The most frequent reason for prescribing additional drugs was the diagnosis of a new condition and only in a minority of day clinic visitors was medication discontinued because diagnoses no longer appeared relevant. In The Netherlands, general practitioners play a key role in healthcare as they refer patients to specialists. We therefore recommend that if a general practitioner sees a geriatric patient with multiple morbidities and associated medication use that is possibly causing psychiatric

Chapter 1.1

22

illnesses, functional decline or other adverse events, the patient should be referred to a geriatric day clinic for CGA and medication review.

References 1. Fu AZ, Liu GG, Christensen DB. Inappropriate medication use and health outcomes in the elderly. J Am

Geriatr Soc 2004; 52: 1934-9. 2. Fillenbaum GG, Hanlon JT, Landerman LR, et al. Impact of inappropriate drug use on health services

utilization among representative older community-dwelling residents. Am J Geriatr Pharmacother 2004; 2: 92-101.

3. Kruse W, Rampmaier J, Frauenrath-Volkers C, et al. Drug-prescribing patterns in old age: a study of the impact of hospitalisation on drug prescriptions and follow-up survey in patients 75 years and older. Eur J Clin Pharmacol 1991;41:441-7.

4. Stewart RB, Cooper JW. Polypharmacy in the aged.: practical solutions. Drugs Aging 1994;4: 449-61. 5. Hemminki E, Heikkila K. Elderly people's compliance with prescriptions, and quality of medication. Scand J

Soc Med 1975; 3: 87-92. 6. Mendelson G, Aronow WS. Underutilization of warfarin in older persons with chronic nonvalvular atrial

fibrillation at high risk for developing stroke. J Am Geriatr Soc 1998; 46: 1423-1424. 7. Ghosh S, Ziesmer V, Aronow WS. Underutilization of aspirin, beta blockers, angiotensin-converting enzyme

inhibitors, and lipid-lowering drugs and overutilization of calcium channel blockers in older persons with coronary artery disease in an academic nursing home. J Gerontol A Biol Sci Med Sci 2002; 57: 398-400.

8. World Health Organisation (WHO) International Classification of Diseases, 10th edition. Available from URL: http://www3.who.int/icd/vol1htm2003/fr-icd.htm/ [Accessed 2005 Jun 30].

9. World Health Organisation Collaborating Centre for Drug Statistics Methodology. Anatomical Therapeutical Chemical (ATC) classification index including defined daily doses (DDDs) for plain substances. Oslo: World Health Organisation Collaborating Centre for Drug Statistics Methodology, 1994.

10. Rampersaud GC, Kauwell GP, Bailey LB. Folate: a key to optimizing health and reducing disease risk in the elderly. J Am Coll Nutr 2003; 22: 1-8.

11. Wolters M, Ströhle A, Hahn A. Cobalamin: A critical vitamin in the elderly. Prev Med 2004; 39: 1256-66. 12. Chapuy MC, Arlot ME, Duboeuf F, et al. Vitamin D3 and calcium to prevent hip fractures in the elderly

women. New Engl J Med 1992;327: 1637-1642. 13. Mecocci P, Cherubini A, Mariani E, et al. Depression in the elderly: new concepts and therapeutic

approaches. Aging Clin Exp Res 2004; 16: 176-89. 14. Solai LK, Mulsant BH, Pollock BG. Selective serotonin reuptake inhibitors for late-life depression: a

comparative review. Drugs Aging 2001;18: 355-368. 15. Spina E, Scordo MG. Clinically significant drug interactions with antidepressants in the elderly. Drugs

Aging 2002;19: 299-320. 16. Muller-Lissner S. General geriatrics and gastroenterology: constipation and faecal incontinence. Best Pract

Res Clin Gastroenterol 2002; 16: 115-33. 17. De Deyn PP, Rabheru K, Rasmussen A, et al. A randomized trial for risperidone, placebo and haloperidol for

behavioural symptoms of dementia. Neurology 1999; 53: 946-55. 18. Jeste DV, Lacro JP, Bailey A, et al. Lower incidence of tardive dyskinesia with risperidone compared with

haloperidol in older patients. J Am Geriatr Soc 1999; 47: 716-9.

23

24

25

CHAPTER 1.2 The effect of admission to a geriatric ward on medication

use: 2002 versus 1985

S.V. Frankfort, C.R. Tulner, J.P.C.M. van Campen, C.H.W. Koks, J.H. Beijnen

Abstract Aim: To investigate the changes in pharmacotherapy of patients during and after admission to a geriatric ward in 2002 and to investigate if this goes along with reduction of drugs. To describe the differences of the admitted patients and their medication in 2002 compared to 1985. Methods: Included patients were admitted to the geriatric ward of a general hospital in the Netherlands during 2002 (n= 258, mean age 84.2 years). Medication at admission, during admission and at discharge was described after retrospective reviewing of medical charts. A comparable study was performed at the same ward in 1985. Results:In 2002, most frequently used medication at admission was acetylsalicylic acid (30.2%). Pantoprazole was during admission used in 38.8% of patients and at discharge in 31.8%. Folic acid that was at admission used by 11.6% of patients was at discharge increased to 23.4%. At discharge, vitamin D was used in 21.5% of patients, whereas lisinopril was used in 17.8% of patients. Both in 1985 and 2002 vitamins were added and use of antibiotics was increased during admission. A mean addition of 1.0 drug in 1985 and of 0.7 drugs in 2002 was observed. Conclusions: Geriatric hospital admission resulted both in 1985 and 2002 in addition of medication. In both periods reductions in medication were nullified by addition of medication for reason of therapy optimisation. Compared to 1985 admitted patients receive more medication resulting from new insights into pharmacotherapy and more use of preventive medicine.

Pharmacoepidemiology and Drug Safety 2006;15:602-606

Chapter 1.2

26

Introduction Polypharmacy is an important issue in geriatric medicine1,2. The frail elderly are often diagnosed with multiple diseases and may thus have to use several drugs, resulting in an increased risk of drug interactions and adverse events3. In addition, the number of drugs prescribed is inversely correlated with adherence to the therapeutic regimen4. Therefore, a balance between optimising pharmacotherapy and reduction in number of drugs used is essential. During hospital admission changes in drug use may result from optimisation of drug regimen by addition of medication for new diagnoses and discontinuation of inappropriate medication. During the last decennium, health care in the Netherlands has changed. Long admissions to geriatric wards have been partly substituted by admissions of shorter duration and by more visits to diagnostic day clinics, where complete geriatric assessment (CGA) is carried out. These changes may possibly result in different characteristics of admitted patients and their medication use. The primary aims of this study are to investigate the changes in pharmacotherapy of newly admitted patients in 2002 during and after admission to a geriatric ward and to investigate if this goes along with reduction of drugs. The secondary aim of this study is to describe the historical differences of the admitted patients and their medication in 2002 compared to 1985.

Methods Patients and data collection This retrospective chart study was carried out at the geriatric ward of a general hospital in the Netherlands during 2002. Patients were excluded if their medical records were incomplete. Origin of patient (i.e. home, nursing home), date of birth, gender, date of admission and discharge, medication at admission, during admission and at discharge were anonymously collected from the medical records and incorporated in a database in Visual FoxPro 6.0 (Microsoft Corporation, Redmond, WA). Design The cohort of admitted patients is defined as cohort 1 (n1) and the cohort of discharged patients is defined as cohort 2 (n2). Drugs were classified according to the anatomical therapeutical chemical (ATC) classification index5. A descriptive analysis was performed for demographic variables, medication used at admittance to, during admission at and at discharge from the geriatric ward. Because in

Pharmacotherapy at a geriatric ward

27

1985 a comparable study6 has been performed at the same geriatric ward changes in pharmacotherapy between 1985 and 2002 could be described.

Results Patients In 2002, a total of 258 patients were included. One patient was excluded, because of incomplete medical information. Table 1 shows patient demographics and characteristics of hospital admission of the studied populations in 2002 (n=258) and 1985 (n=724). Both cohorts consisted of more women, patients lived primarily at their own homes and comparable percentages of deaths during admission were shown. Mean duration of admission was remarkably shorter in 2002 (25 days) compared to 1985 (52 days). Table 1. Descriptives of demographics, hospital admission and medication use in 1985 and 2002. Characteristics 19856 (n1=724) 2002 (n1=258) Demographics Gender, Male (%) / Female (%) 521 (72.0) / 203 (28.0) 162 (62.8) / 96 (37.2) Age at admission (years), mean ± SD (range) 83 (64-104) 84.2 ± 6.6 (65.2-99.4) Place of living, Own home (%) / Other (%) 474 (65.5) / 250 (34.5) 189 (73.3) / 69 (28.3) Hospital admission Discharged patients (n2) (%) / Deaths (%) 622 (85.9) / 102 (14.1) 214 (83.0) / 44 (17.0) Inpatient days, mean ± SD (range) 52 (1-144) 25.1 ± 26.4 (1-166) Medication use At admission, mean ± SD (range) 2.8 ± 2.3 5.2 ± 3.3 (0-16) During admission, mean ± SD (range) 6.4 ± 3.4 10.6 ± 5.0 (0-25) At discharge, mean ± SD (range) 3.8 ± 2.2 5.9 ± 3.1 (0-22) Drugs added during hospital admission, mean (range) 1.0 (-6-+8) 0.7 (-7-+10)

Medication at admission, during admission and at discharge ATC code groups A, gastrointestinal tract and metabolism; group B, blood and blood-forming organs; group C, cardiovascular agents; and group N, central nervous system agents, were frequently used at admission. Hospital admission resulted in a strong increase in almost all ATC code groups, except oncolytics (L). At discharge a slight increase in group A and a slight decrease in group C and N was noticed, whereas groups B, J (systemic antimicrobial drugs) and R (respiratory tract agents), were almost comparable to admission. Table 2 shows the top 10s of individual medication at admission, during admission and at discharge in 2002 and the top 5s of medication groups in 1985 and 2002, all expressed as percentage of patients. In 2002, most frequently used medication at admission was

Tab

le 2

. M

edic

atio

n us

e at

and

dur

ing

adm

issi

on to

and

at d

isch

arge

from

the

geria

tric

war

d in

198

5 an

d 20

02. N

umbe

rs b

etw

een

squa

re

brac

kets

refle

ct th

e ra

nk a

t adm

issi

on.

20

02

2002

19

856

To

p 10

indi

vidu

al m

edic

atio

n (%

of p

atie

nts)

To

p 5

med

icat

ion

grou

ps (%

of p

atie

nts)

To

p 5

med

icat

ion

grou

ps (%

of p

atie

nts)

A

t adm

issi

on (n

1)

Dur

ing

adm

issi

on (n

1)

At d

isch

arge

(n

2)

At a

dmis

sion

(n

1)

Dur

ing

adm

issi

on

(n1)

A

t dis

char

ge

(n2)

A

t adm

issi

on

(n1)

D

urin

g ad

mis

sion

(n

1)

At d

isch

arge

(n

2)

1 A

cety

lsal

icyl

ic

acid

(30.

2)

Para

ceta

mol

(41.

9) [3

] Pa

ntop

razo

le

(31.

8) [2

1]

Diu

retic

s (6

0.9)

A

ntic

oagu

lant

s (1

40.3

) [2]

A

ntic

oagu

lant

s (9

3.0)

[2]

Diu

retic

s (3

4.0)

La

xativ

es (5

0.8)

[7

] La

xativ

es (4

2.1)

[7

] 2

Furo

sem

ide

(28.

7)

Furo

sem

ide

(40.

7) [2

] A

cety

lsal

icyl

ic

acid

(30.

0) [1

] A

ntic

oagu

lant

s (5

6.9)

A

ntib

iotic

s (1

03.1

) [10

] V

itam

ins (

44.9

) [5

] A

nalg

esic

s (2

2.4)

A

ntib

iotic

s (4

5.8)

[8]

Vita

min

s (31

.4)

[5]

3 Pa

race

tam

ol

(21.

3)

Frax

ipar

ine

(39.

9) [8

1]

Furo

sem

ide

(25.

7) [2

] A

nalg

esic

s (2

7.5)

N

euro

lept

ics

(68.

2) [4

] Pr

oton

Pum

p In

hibi

tors

(34.

6)

[13]

Seda

tives

/ H

ypno

tics

(22.

0)

Vita

min

s (44

.4)

[5]

Diu

retic

s (29

.6)

[1]

4 D

igox

in (1

5.5)

Pa

ntop

razo

le (3

8.8)

[2

1]

Folic

aci

d (2

3.4)

[7]

Neu

role

ptic

s (2

7.5)

A

nalg

esic

s (5

6.2)

[3]

AC

E in

hibi

tor

(28.

5) [6

] D

igox

in

(14.

9)

Diu

retic

s (42

.1)

[1]

Seda

tives

/ H

ypno

tics

(24.

8) [3

] 5

Ace

noco

umar

ol

(13.

2)

Ace

tyls

alic

ylic

aci

d [1

] (3

6.4)

Pa

race

tam

ol

(22.

4) [3

] V

itam

ins

(22.

9)

Vita

min

s (49

.2)

[5]

Iron

/ M

iner

als

(27.

6) [7

] V

itam

ins

(13.

7)

Seda

tives

/ H

ypno

tics

(39.

4) [3

]

Ana

lges

ics

(22.

5) [2

]

6 Te

maz

epam

(1

2.4)

H

epar

ine

(35.

7) [-

] V

itam

in D

(2

1.5)

[11]

7 Fo

lic a

cid

(11.

6)

Tem

azep

am (3

0.6)

[6]

Mag

nesi

um

Oxi

de (2

1.0)

[2

5]

8 Is

osor

bide

di

nitra

te (1

0.9)

A

mox

icill

in (2

9.1)

[-]

Ace

noco

umar

ol

(20.

5) [5

]

9 Sp

irono

lact

one

(10.

9)

Folic

Aci

d (2

8.7)

[7]

Dig

oxin

(19.

2)

[4]

10La

ctul

ose

(10.

0)

Dig

oxin

(24.

8) [4

] Li

sino

pril

(17.

8) [1

2]


29

acetylsalicylic acid (30.2%), followed by furosemide (28.7%). During admission therapy with proton-pump-inhibitors (PPIs), pantoprazole, was used in 38.8% of patients and 31.8% of discharged patients used pantoprazole. Folic acid that was at admission used by 11.6% of patients, was at discharge increased to 23.4%. At discharge vitamin D was used in 21.5% of patients, whereas lisinopril, an angiotensin converting enzyme (ACE) inhibitor, was used in 17.8% of patients. In 1985, the percentage of patients using diuretics at admission was lower compared to 2002, however at discharge this percentage was higher in 1985 compared to 2002. Introduction of ACE inhibitors in routine clinical practice was shown in 2002. In 2002, more than half of the admitted patients already used anticoagulants and this was further increased at discharge. Both in 1985 and 2002, vitamins were added and use of antibiotics was increased during admission. In 2002, use of analgesics was markedly increased during admission, whereas in 1985 use of laxatives was increased during admission, which was not shown in 2002. Use of neuroleptics was increased in 2002 compared to 1985. Change in number of drugs at admission, during admission and at discharge The mean number of drugs used per patient at admission, during admission and at discharge both in 1985 and 2002 is shown in table 1. Patients received more medication in 2002 compared to 1985. In both periods admission resulted in addition of drugs, partly discontinued at discharge, but a substantial part is continued and thus resulted in a mean addition of 1.0 drug (range from reduction of 6 to addition of 8) in 1985 and of 0.7 drugs (range from reduction of 7 to addition of 10) in 2002. Reduction of medication has been achieved in a minority (30.4%) of patients. In 47 patients (22.0%) medication was not changed or number of added drugs equalled discontinued ones. In 47.6% of patients addition has been experienced.

Discussion Geriatric hospital admission results in addition of drugs. During admission more antibiotics, analgesics, anticoagulants and neuroleptics are used. Heparin and amoxicillin are discontinued at discharge, but for example pantoprazole, folic acid and vitamin D are continued. These changes in pharmacotherapy result in a mean addition of 0.7 drugs at discharge and only in 30.4% of patients medication has been reduced. Comparable numbers of added drugs were shown in 2002 and 1985. During both periods, antibiotics and vitamins were added during hospital admission. Laxatives, sedatives and diuretics were more used at discharge in 1985 compared to 2002. Use of ACE inhibitors and PPIs in routine clinical practice is apparent nowadays.

Chapter 1.2

30

During admission to the geriatric ward an important goal is the optimisation of pharmacotherapy and we showed a primary role for preventive therapy, for example vitamin supplementation. It is known that an inadequate folate status is associated with an increased risk for chronic diseases that may have a negative impact on the health of the aging population7. Studies showed a relative risk of 6.8 for folate deficiencies in elderly living at home versus elderly living in institutions8 and its prevalence increases with age9. Elderly populations aged over 65 and residents of nursing homes showed prevalence of vitamin D deficiencies between 25 and 54%. This deficiency is an important risk factor for osteopenia and bone fractures. Supplementation of 800IE vitamin D daily showed substantially reduced risks of osteoporotic fractures10. The risk of developing gastroduodenal ulcers or complications is recognised as a problem in patients using NSAIDs. Important risk factors are increasing age and a past history of upper gastrointestinal ulcer or bleeding11. Once daily pantoprazole that showed efficacy in preventing peptic ulcers12, is often co-prescribed with NSAIDs in our setting. Optimisation by means of an ACE inhibitor, for example lisinopril, is favourable in type 1 diabetes mellitus with microalbuminuria, because blood pressure control is a key element in slowing the progression of diabetic nephropathy13. ACE inhibitors are also of primary importance after acute decompensated heart failure because of proven prolonged life expectance14. Geriatric health care has changed. In 1985, the geriatric ward of our hospital counted 108 beds6 compared to only 24 beds in 2002. However, the geriatric day clinic counted 702 first-time visitors in 2002 and plays an important role in diagnostics resulting in less and shorter durations at the ward. In 2002, patients used almost twice as much medication as the studied population in 1985. This can be explained by the introduction of new medication groups since 1985, introduction of more guidelines and guidelines advising drug combinations. For example, patients with heart failure resulting from left ventricular dysfunction are now often treated with a combination of an ACE inhibitor, diuretic, a beta blocker and spironolactone15.

Conclusions Geriatric hospital admission resulted in both 1985 and 2002 in addition of medication. In both periods, reductions in medication were nullified by addition of medication for reason of therapy optimisation. Changes in healthcare structure resulted in less and shorter admissions. Compared to 1985 admitted patients receive more medication resulting from new insights into pharmacotherapy and more use of preventive medicine.


31

References 1. Linjakumpu T, Hartikainen S, Klaukka T, et al. Use of medications and polypharmacy are increasing among

the elderly. J Clin Epidemiol 2002; 55: 809-17. 2. Rollason V, Vogt N. Reduction of polypharmacy in the elderly: a systematic review of the role of the

pharmacist. Drugs Aging 2003; 20: 817-32. 3. Stewart RB, Cooper JW. Polypharmacy in the aged. Practical solutions. Drugs Aging 1994; 4: 449-61. 4. Hemminki E, Heikkila J. Elderly people’s compliance with prescriptions and quality of medication. Scand J

Soc Med 1975; 3: 87-92. 5. Anatomical Therapeutical Chemical (ATC) classification index including defined daily doses (DDD) for

plain substances. Oslo: World Health Organisation Colaborating Centre for Drug Statistics Methodology, 1994.

6. Olde Rikkert MG, Brouwer E, Thijssen C, de Ruijter GN. Polypharmacy in patients of the geriatric department of a general hospital (in dutch). Tijdschr Gerontol Geriatr 1990; 21: 51-9.

7. Rampersaud GC, Kauwell GP, Bailey LB. Folate: a key to optimizing health and reducing disease risk in the elderly. J Am Coll Nutr 2003; 22: 1-8.

8. Figlin E, Chetrit A, Shahar A et al. High prevalences of vitamine B12 and folic acid deficiency in elderly subjects in Israel. Br J Haematol 2003; 123: 696-701.

9. Clarke R, Grimley Evans J, Schneede J, et al. Vitamin B12 and folate deficiency in later life. Age Ageing 2004; 33: 34-41.

10. Thomas MK, Lloyd-Jones DM, Thadhani RI, et al. Hypovitaminosis D in medical inpatients. N Engl J Med 1998; 338: 777-83.

11. Goldstein JL. Challenges in managing NSAID-associated gastrointestinal tract injury. Digestion 2004; 69 Suppl 1: 25-33.

12. Bianchi Porro G, Lazzaroni M, Imbesi V, et al. Efficacy of pantoprazole in the prevention of peptic ulcers induced by non-steroidal anti-inflammatory drugs: a prospective, placebo-controlled, double-blind, parallel-group study. Dig Liver Dis 2000; 32: 201-8.

13. Vilarrasa N, Soler J, Montanya E. Regression of microalbuminuria in type 1 diabetic patients: results of a sequential intervention with improved metabolic control and ACE inhibitors. Acta Diabetol 2005; 42: 87-94.

14. Southworth MR. Treatment options for acute decompensated heart failure. Am J Health Syst Pharm 2003; 60 Suppl 4: S7-15.

15. Guidelines for the diagnosis and treatment of chronic heart failure. Eur Heart J 2001; 22: 1527-60.

32

33

CHAPTER 2

DIGOXIN

34

35

CHAPTER 2.1 Role of ABCB1 genotypes and haplotypes in digoxin steady

state pharmacokinetics in geriatric patients

S.V. Frankfort, R.J. Keizer, A.D.R. Huitema, V.D. Doodeman, C.R. Tulner, J.P.C.M. van Campen, P.H.M. Smits, J.H.M. Schellens, J.H. Beijnen

Abstract

Objectives: To evaluate the influence of ABCB1 polymorphism on digoxin steady state pharmacokinetics in geriatric patients using digoxin for the indications atrial fibrillation or congestive heart failure. Methods: ABCB1 genotype was determined by sequencing at positions C1236T (exon 21), G2677T/A (exon 21) and C3435T (exon 26) and haplotypes were subsequently inferred. Non-linear mixed effect modelling (NONMEM) was used to model the steady-state pharmacokinetics of digoxin in geriatric patients. In co-variate analyses the influence of ABCB1 genotypes and haplotypes on the clearance of digoxin after oral ingestion was investigated. Results: 140 patients were included with a median age of 84.8 years and a median daily digoxin dose of 125 µg. A subpopulation of 40 prospectively recruited patients was genotyped for polymorphisms of ABCB1. A typical value for the apparent digoxin clearance of 5.68 L/h was found. Identified ABCB1 genotypes and haplotypes were not significant covariates for digoxin clearance. Conclusion: Digoxin steady state pharmacokinetics were not significantly influenced by the identified ABCB1 polymorphism or haplotypes of these polymorphisms in a population of geriatric patients.

Submitted

Chapter 2.1

36

Introduction Digoxin is used in the treatment of atrial fibrillation and congestive heart failure in the elderly1,2. Both inotropic and arrhythmogenic effects of digoxin are related to its effects on the sodium-potassium pump3. Sodium-calcium exchange is promoted and thereby the intracellular calcium concentration is increased and eventually results in increased force of cardiac contraction3,4. The pharmacokinetics of digoxin are characterised by a bioavailability exceeding 90%, a volume of distribution between 4 and 7 L/kg and a long elimination half-life between 30 and 40 hours4. Earlier research investigated age, serum creatinine level, co-administration of spironolactone, gender and presence of congestive heart failure as variables that may influence digoxin pharmacokinetics. Female gender, higher age, higher level of serum creatinine, and use of spironolactone resulted in significantly lower oral digoxin clearances5-7. Digoxin is a substrate for the drug transporter P-glycoprotein (P-gp), a 170 kDa membrane bound efflux pump. This efflux pump is located amongst others in the epithelial layer of the intestine, renal tubules, placenta and blood-brain-barrier. It enhances transport from blood into the gut lumen via the intestinal epithelium and from the renal tubules into urine8,9. The Multi Drug Resistance gene (ABCB1) encodes for P-gp. It is known that the ABCB1 gene is highly polymorphic10. The three most frequently occurring Single Nucleotide Polymorphisms (SNPs) are C1236T in exon 12, G2677T/A in exon 21 and C3435T in exon 2611. A common haplotype containing these three SNPs simultaneously was found by Kim et al.12 The synonymous SNP C3435T was the first variant to be associated with altered protein function. As a linkage disequilibrium exists between this SNP and the nonsynonymous G2677T/A it has been suggested that functional differences associated with the C3435T polymorphism may be the result of the associated SNP at G2677T/A10. Studies investigating single SNPs resulted in conflicting results; both increases, no significant differences or decreases in maximal plasma concentrations and area under the plasma concentration versus time curve (AUC) of digoxin were reported in heterozygous and homozygous mutants at positions G2677T/A and C3435T13-15. ABCB1 haplotypes composed of different SNPs may better represent changes in P-gp function, as found in digoxin and ciclosporin pharmacokinetic studies10,14,15. Digoxin has been used as a probe drug in studies regarding the functional relevance of P-gp16,17. Clinical studies regarding the functional relevance of ABCB1 SNPs were mostly performed in young healthy volunteers as either a single-dose study or a multiple dose study during a short period13-15, 18-21. This study aims to investigate the correlation between ABCB1 SNPs and their haplotypes and the steady state pharmacokinetics of digoxin in a cohort of geriatric patients that use digoxin as part of a regular therapeutic regimen for atrial fibrillation and/or congestive heart failure.

Role of ABCB1 genotypes in digoxin pharmacokinetics

37

Methods Patients This study consisted of two parts. First, a retrospective part was performed in patients who were admitted to the geriatric diagnostic day-clinic of the Slotervaart Hospital, Amsterdam, The Netherlands between May 2000 and May 2006. In these patients digoxin plasma concentrations were determined. Second, a prospective study was carried out at the department of Geriatric Medicine of the Slotervaart Hospital between January 2004 and May 2006. In this prospective study, digoxin plasma concentrations were determined in patients treated with digoxin and patients were genotyped at positions C1236T in exon 12, G2677T/A in exon 21, and C3435T in exon 26 of ABCB1. All patients used digoxin for the indications atrial fibrillation or congestive heart failure and plasma digoxin concentrations were regularly measured within the therapeutic drug monitoring service of the Department of Pharmacy & Pharmacology. Patients were only included if digoxin was used at the current dose for a minimum of two weeks to assure that pharmacokinetics were measured at steady state levels. The study protocol was approved by the Institutional Review Board of the Slotervaart Hospital, Amsterdam, The Netherlands. Written informed consent was obtained from each participant for ABCB1 genotyping in this study. Sampling, bioanalysis and pharmacogenetic analysis From all patients, a 5 ml EDTA blood sample was obtained by venous puncture. We considered untimed samples as steady state levels due to the very long elimination half-life of digoxin. In the prospective part of this study, 2 ml whole blood was stored at -20 °C pending genotype analysis. After isolation of plasma by centrifuging, the samples were stored at –20 °C until drug analysis. Digoxin concentrations were assayed with the TDx digoxin assay kit (Abbott Laboratories Ltd) for fluorescence polarisation immunoassay (FPIA) technology. The lower limit of quantification was 0.2 ng/mL. The precision and accuracy of this assay were within 8% (as mentioned in the manufacturer manual)22. Genomic DNA was extracted from EDTA whole blood using the Qiagen QIAamp® DNA Mini Kit (Qiagen, Leusden, The Netherlands) according to the manufacturer protocol. ABCB1 genotypes at positions C1236T, G2677T/A and C3435T were analysed as previously described23. Haplotypes were inferred with the software package HPlus65v2.1.1 (http://qge.fhcrc.org/hplus/) using the Expectation-Maximisation (EM) algorithm.

Chapter 2.1

38

Pharmacokinetic analysis The nonlinear mixed effect modelling program (NONMEM) Version V (double precision, level 1.1, GloboMax LLC, Hanover MD, USA)24 was used to perform the analysis. The first-order conditional estimation procedure (FOCE) with interaction was used throughout. The adequacy of the tested models was evaluated using statistical and graphical methods. The minimal value of the objective function (OFV, equal to minus twice the log likelihood) provided by NONMEM was used as a goodness of fit characteristic to discriminate between hierarchical models using the log likelihood ratio test24. A p-value of 0.001, representing a decrease in OFV of 10.8 points (degrees of freedom (df)= 1) or 13.8 points (df= 2) was considered statistically significant. Standard errors for all parameters were calculated with the COVARIANCE option in NONMEM, individual Bayesian pharmacokinetic parameter estimates were obtained using the POSTHOC option24. The program PDx-POP (version 1.1, release 4, Globomax LLC, Hanover MD, USA) was used as a tool for expediting population pharmacokinetic analysis with NONMEM and for graphical model diagnostics. Pharmacokinetic model and co-variate analysis The data were fitted to the following one-compartment steady state deterministic model: Css,i= Di/(CLi/F*τi) (equation 1) where Css,i is the serum steady state concentration of digoxin (µg/L) measured in the ith patient, Di is the dose of digoxin (µg), CLi/F is the oral clearance of the ith individual (L/h) and τi is the interval between daily doses for the ith individual. Variability was modelled according to the following equation: Ci=Ĉi+εi (equation 2) where Ci is the observed digoxin concentration of individual i, Ĉi is the corresponding predicted concentration by the model and εi is the difference between observed and predicted concentrations and consists of the interindividual variability and residual variability since only one digoxin concentration was available per patient. To identify possible relationships between the clearance and patient characteristics, the following co-variates were collected: age (years), gender, creatinine level in serum, use of spironolactone, use of diuretics and presence of congestive heart failure. Genotypic data were not available of the retrospectively included patients. Therefore, in the analyses investigating genotypes as possible co-variates, separate typical values of CL/F were estimated for the retrospectively included patient population and for prospectively included


39

patients including the influence of genotypes. Genotypic data were investigated at the positions C1236T (exon 12), G2677T/A (exon 21) and C3435T (exon 26) as three groups: homozygous wild-type, heterozygous mutants and homozygous mutants. The three investigated genotypes were separately introduced as co-variates according to the following equations: CL/F= θ1 (equation 3) CL/F= θ2* θ3

heterozygous * θ4homozygous (equation 4)

where θ1 represents the apparent clearance in the retrospectively included patients, θ2

represents the apparent clearance in the prospectively included wild-type patients, θ3 is the fractional change in heterozygous mutants and θ4 is the fractional change in homozygous mutants. Haplotypes were investigated as inferred haplotypes at positions in exon 12, exon 21 and exon 26. Haplotype data were divided into dichotomous variables: presence or absence of the TTT haplotype and presence or absence of the CGC haplotype. Inferred haplotypes at position C1236T, G2677T/A and C3435T were introduced in the model as described in the following equations: CL/F = θ1 (equation 5) CL/F = θ2* θ3

HAPLOTYPE (equation 6)

where θ1 represents the apparent clearance in retrospectively included patients, θ2 represents the apparent clearance in prospectively included patients and θ3 is 1 for carriers of the TTT haplotype and 0 for non-carriers of the TTT haplotype and in a separate analysis 1 for carriers of the CGC haplotype and 0 for non-carriers of the CGC haplotype. A covariate was considered statistically significant when the inclusion was associated with a decrease in minimal value of the OFV associated with a p-value of <0.001 (log-likelihood ratio test). Results Patients A total of 140 patients were included in this study of whom 40 were prospectively included and who gave informed consent for ABCB1 genotyping. Demographic characteristics were not significantly different between prospectively and retrospectively included patients (results not shown). The total population had a median age of 84.8 years, 64% were females and they used a median daily digoxin dose of 125 µg (table 1). Plasma concentrations of

Chapter 2.1

40

digoxin are shown in figure 1. In table 1 the frequencies of the genotypes at positions C1236T, G2677T/A and C3435T and the frequencies of the inferred haplotypes at those three positions are listed. Table 1. Baseline characteristics of digoxin users. Baseline characteristic Digoxin users (n=140)* Age (years), median (range) 84.8 (69.5-96.8) Gender, male / female (%) 12/ 21 (36 / 64) Dose (µg), median (range) 125 (62.5-500) Prospectively included patients, n 40 C1236 T (Exon 12), n (% of prospective patients) CC 12 (30.0) CT 21 (52.5) TT 7 (17.5) G2677T/A (Exon 21), n (% of prospective patients) GG 11 (27.5) GT 23 (57.5) TT 6 (15.0) C3435T (Exon 26), n (% of prospective patients) CC 8 (20.0) CT 24 (60.0) TT 8 (20.0) Haplotype, n (% of (prospective patients*2‡))† CGC 38 (47.5) CGT 6 (7.5) CTT 1 (1.3) TGC 1 (1.3) TTC 1 (1.3) TTT 33 (41.3) Presence of minimal 1 TTT haplotype, n (% of prospective patients) †

28 (70.0)

Presence of minimal 1 CGC haplotype, n (% of prospective patients) †

31 (77.5)

Serum creatinine (µmol/L), median (range) 93 (54-482) Use of spironolactone, n (%) 29 (20.7) Use of diuretics, n (%) 89 (83.6) Presence of congestive heart failure, n (%) 51 (36.4) * total included patients: n=140 (retrospective study: n=100; prospective study, including ABCB1 genotyping: n=40), ‡ each patient carriers two haplotype alleles, † haplotype C1236T-G2677T-C3435T


41

Figure 1. Steady state plasma concentration (µg/L) versus daily digoxin dose (µg).

00.5

11.5

22.5

33.5

44.5

5

0 100 200 300 400 500

Daily digoxin dose (µg)

Conc

entra

tion

digo

xin

(µg/

L)

Population pharmacokinetics and co-variate analyses The basic pharmacokinetic model based upon equation 1 and modelled as a one-compartment model resulted in a population clearance of 5.68 L/h with a relative standard error (RSE) of 5.76%. Age, gender, serum creatinine level, use of spironolactone and presence of congestive heart failure were not identified as significant co-variates for the steady-state clearance of digoxin, as introduction of these co-variates resulted in decreases in OFV of less than 10.8 points. Table 2. Results of analyses investigating genotypes as possible co-variates for digoxin steady-state pharmacokinetics. SNP Typical value CL/F (% RSE) Retrospectively

included patients

Prospectively included wild types

Fractional change in Cl/F in heterozygous mutants (% RSE)

Fractional change in Cl/F in homozygous mutants (% RSE)

p-value

C1236T 5.61 (6.52) 4.75 (14.4) 1.32 (23.6) 1.41 (21.6) >0.05 G2677T/A 5.61 (6.52) 4.38 (13.2) 1.44 (22.0) 1.63 (21.8) >0.05 C3435T 5.61 (6.52) 4.77 (13.3) 1.28 (21.6) 1.37 (24.7) >0.05 SNP = Single Nucleotide Polymorphism, CL/F= apparent digoxin clearance (L/h), RSE= Relative standard error, Retrospectively included patients: CL/F= θCL/F , Prospectively included patients: CL/F=θCL/F, WILD TYPES * θHETEROZYGOUS * θHOMOZYGOUS

Chapter 2.1

42

An increase in CL/F was shown in both heterozygous and homozygous mutants. However, all relations between CL/F and genotypes of ABCB1 were non-significant. The results of the analyses investigating ABCB1 genotypes as covariates are summarised in table 2. Presence of the TTT or CGC haplotype was not revealed as a significant co-variate for CL/F as the decreases in OFV were smaller then 10.8 points (table 3). Table 3. Results of analyses investigating haplotypes as possible co-variates for digoxin steady-state pharmacokinetics. Haplotype Typical value CL/F (% RSE) Retrospectively

included patients Prospectively included patients

Fractional change in CL/F in patients carrying the haplotype (% RSE)

p-value

TTT 5.61 (6.52) 4.75 (14.4) 1.34 (20.7) >0.05 CGC 5.61 (6.52) 6.69 (15.7) 0.854 (21.9) >0.05 Haplotype of C1236T-G2677T-C3435T. CL/F= apparent digoxin clearance (L/h), RSE= Relative standard error. Prospectively included patients: CL/F= θCL/F * θTTT, where TTT=0 for non-carriers and TTT=1 for carriers or CL/F= θCL/F * θCGC, where CGC=0 for non-carriers and CGC=1 for carriers

Discussion Our basic pharmacokinetic model revealed a typical value of 5.68 L/h for CL/F. Yukawa et al.5 gave an overview of estimated digoxin clearances in various studies and these ranged between 1.56 L/h when calculated for a creatinine clearance of 5 mL/min to 10.3 L/h as calculated for a creatinine clearance of 100 mL/min. The population clearance from our results is comparable to the range in the described population clearances. In our population we did not find age, gender, serum creatinine, presence of congestive heart failure and use of spironolactone as significant co-variates in an intermediate model. The frequencies of ABCB1 genotypes in our population are comparable to those earlier described in a geriatric population25 and in younger populations12,23,26. Haplotype frequencies were also comparable to those earlier described by us in another geriatric population25. No significant correlations were found between individual genotypes at positions C1236T, G2677T/A and C3435T and digoxin clearance. We did, however, show a non-significant increase in the apparent clearance of digoxin in patients bearing mutant alleles compared to wild-types. This was most pronounced for the homozygous mutants and for heterozygous mutants intermediate values were found. An increase in the apparent clearance of digoxin results in lower steady state concentrations and lower AUC values. We performed analyses in geriatric patients that use digoxin for medical conditions. Thus far, only investigations into the role of ABCB1 polymorphisms into digoxin pharmacokinetics were performed in


43

healthy volunteers. These volunteers are often of a younger age14,18,19 or receive only a single oral dose of digoxin18,19. Our results for C3435T in exon 26 are comparable to those of Sakaeda et al.20 In that study a lower maximum plasma concentration and lower AUC0-4h of digoxin were found in heterozygous and homozygous mutants. On the other hand, our results are in contradiction to the results of Johne et al.14 who found the highest AUC0-24h in homozygous mutants at position C3435T (not statistically significant) and steady-state plasma concentrations were significantly higher in homozygous mutants compared to wild-types. Our results for G2677T/A in exon 21 are comparable to the results of Kim et al.12 who showed non-significantly higher fexofenadine, also a P-gp substrate, AUC0-4h in homozygous wild-type subjects compared to homozygous mutants. In the study by Verstuyft et al.18 no significant differences in digoxin AUC0-24h were found between different genotypes at position G2677T/A in exon 21. As earlier suggested12,13-15, correlations between ABCB1 genotypes and pharmacokinetics show conflicting results. This study adds to those conflicting results. A possible explanation for the conflicting results may be that the function and/or expression of P-gp is not solely influenced by a single ABCB1 genotype, but instead by different SNPs at other loci in the ABCB1 gene13. In addition, it could be possible that different SNPs that are currently unknown, but may be important functional SNPs, are in a strong linkage disequilibrium with C3435T. This may result in the linkage of the mutant T allele at position 3435 to different SNPs, resulting in differences in function of P-gp27. Haplotype analysis may thus be superior to genotype analysis. In our geriatric population the TTT and CGC haplotypes inferred at positions C1236T, G2677T/A and C3435T were most frequent among the different haplotypes that were present. We investigated whether presence of at least one TTT haplotype or the presence of at least one CGC haplotype was significantly related to the apparent clearance in our population, but did not show significant results. In the study by Johne et al.14 carriers of the TT haplotype (position G2677T linked to C3435T) showed significantly higher digoxin trough concentrations compared to non-carriers. Those results are contradictory to our results as we showed a (non-significantly) higher apparent clearance, corresponding to lower digoxin plasma concentrations. Johne et al.14 also demonstrated that individuals with at least one haplotype GT (position G2677T linked to C3435T) had significant higher digoxin plasma levels. In our population of geriatric patients we included 6 patients with at least 1 haplotype CGT (position C1236T linked to G12677T and C3435T) and haplotypes consisting of TGT were not found in the population. The presence of at least one haplotype CGT resulted in not significantly reduced apparent clearance, corresponding to higher digoxin plasma concentrations (results not shown). The study by Kim et al.12 investigated fexofenadine concentrations in patients with different ABCB1 haplotypes. Patients homozygous for CGC showed higher AUC values compared to patients with one TTT haplotype and one CGC

Chapter 2.1

44

haplotype. In our population we did not find such significant differences in clearance between those latter two groups (results not shown). Our results did not show that haplotype analysis is superior to genotype analysis in the correlation between SNPs and steady state digoxin pharmacokinetics and are thus not in agreement with the results of Johne et al.14 and Kim et al.12. We do not have clear explanations for these differences. We have included a relatively low number of participants of whom we obtained genotypic data, however, numbers of included patients were also small in those studies12,14. In general, to obtain additional knowledge into the functional relevance of ABCB1 haplotypes, future studies should include larger number of patients and if possible haplotypes consisting of more genotypes. These studies may reveal new insights into haplotypes that are less frequent and into haplotypes composed of different genotypes. In healthy volunteer studies, complete pharmacokinetic curves were sampled in participants. In the study by Kurata et al.15 heterozygous and homozygous mutants had significantly lower apparent tubular secretary clearances and they suggested that both reductions in intestinal secretion of digoxin and renal secretion into the urine occur in subjects. The study by Johne et al.14 suggested that the terminal elimination does not proceed more slowly in homozygous mutants compared to wild-types. This study showed increased absorption and higher levels of steady state exposure in homozygous mutants at position C3435T compared to wild-types and suggest the absorption phase is particularly affected by ABCB1 polymorphism. Our study only investigated effect of genotypes and haplotypes on steady state concentrations. In the long-term steady-state pharmacokinetics the trough concentration is probably less influenced by effects of genotype or haplotype on absorption as compared to the single dose studies or short-term multiple dose studies. The non-significant influences of genotypes and haplotypes on clearance in our study may be explained by these differences. Earlier described studies investigating the relation between ABCB1 genotypes and digoxin pharmacokinetics in healthy volunteers and our study differ in design and this may also explain different results between studies. Those studies intended to investigate the functional relevance of ABCB1 SNPs by using digoxin as a probe drug. Our study was designed to investigate the role of these SNPs in the steady state pharmacokinetics of digoxin in geriatric patients. We did not exclude patients if they used co-medication that may influence P-gp function, which is contrary to the healthy volunteer studies. In addition, our study was performed in routine clinical practice and possible medication non-adherence may have influenced our results. In conclusion, digoxin steady state pharmacokinetics were not significantly influenced by ABCB1 polymorphism or haplotypes of these polymorphisms in a population of geriatric patients using digoxin as part of their therapeutic drug regimens for the indications atrial fibrillation or congestive heart failure.


45

Acknowledgements Dieuwke Meier, Eric Calla, Remko Harms, Jan Nelis and Kees de Goey are kindly acknowledged for their technical assistance in digoxin measurements.

References 1. Cheap G, Girard K, Vincent JP. Atrial fibrillation in the elderly. Facts and management. Drugs Aging

2002;19:819-846. 2. Dec GW. Digoxin remains useful in the management of chronic heart failure. Med Clin N Am 2003;87:317-

337. 3. Eichhorn EJ, Gheorghiade M. Digoxin. Prog Card Dis 2002;44:251-266. 4. Hanratty CG. McLinchey P, Johnston GD, Passmore AP. Differential pharmacokinetics of digoxin in elderly

patients. Drugs Aging 2000;17:353-362. 5. Yukawa E, Honda T, Ohdo S, Higuchi S, Aoyama T. Population-based investigation of relative clearance of

digoxin in Japanese patients by multiple through screen analysis: an update. J Clin Pharmacol 1997;37:92-100.

6. Yukawa E, Suematu F, Yukawa M, et al. Population pharmacokinetics of digoxin in Japanese patients. A 2-compartment pharmacokinetic model. Clin Pharmacokinet 2001;40:773-781.

7. Yukawa E, Mine H, Higuchi S, Aoyama T. Digoxin population pharmacokinetics from routine clinical data: role of patient characteristics for estimating dosing regimens. J Pharm Pharmacol 1992;44:761-765.

8. Schinkel AH and Jonker JW. Mammalian drug efflux transporters of the ATP binding cassette (ABC) family: an overview. Adv Drug Del Rev 2003;55:3-29.

9. Borst P, Oude Elferink R. Mammalian ABC transporters in health and disease. Ann Rev Biochem 2002;71:537-92.

10. Marzolini C, Paus E, Buclin T, Kim RB. Polymorphisms in human MDR1 (P-glycoprotein): Recent advances and clinical relevance. Clin Pharmacol Ther 2004;75:13-33.

11. Bosch TM, Meijerman I, Beijnen JH, Schellens JHM. Genetic polymorphisms of drug-metabolising enzymes and drug transporters in the chemotherapeutic treatment of cancer. Clin Pharmacokinet 2006;45:253-85.

12. Kim RB, Leake BF, Choo EF, et al. Identification of functionally variant MDR1 alleles among European Americans and African Americans. Clin Pharmacol Ther 2001; 70:189-99.

13. Chowbay B, Li Huihua, David M, et al. Meta-analysis of the influence of MDR1 C3435T polymorphism on digoxin pharmacokinetics and MDR1 gene expression. Br J Clin Pharmacol 2005;60:159-171.

14. Johne A, Kopke K, Gerloff T, et al. Modulation of steady state kinetics of digoxin by haplotypes of the P-glycoprotein MDR1 gene. Clin Pharmacol Ther 2002;72:584-594.

15. Kurata Y, Ieiri I, Kimura M, et al. Role of human MDR1 gene polymorphism in bioavailability and interaction of digoxin, a substrate of P-glycoprotein. Clin Pharmacol Ther 2002;72:209-219.

16. Mayer U, Wagenaar E, Beijnen JH, et al. Substantial excretion of digoxin via the intestinal mucosa and prevention of long-term digoxin accumulation in the brain by the mdr 1a P-glycoprotein. Br J Pharmacol 1996;119:1038-1044.

17. Schinkel AH, Wagenaar E, van Deemter L, et al. Absence of the mdr1a P-glycoprotein in mice affects tissue distribution and pharmacokinetics of dexamethasone, digoxin and cyclosporin A. J Clin Invest 1995;96:1698-1705.

18. Verstuyft C, Schwab M, Schaeffeler E, et al. Digoxin pharmacokinetics and MDR1 genetic polymorphisms. Eur J Clin Pharmacol 2003;58:809-812.

19. Gerloff T, Schaefer, Johne A, et al. MDR1 genotypes do not influence the absorption of a single oral dose of 1 mg digoxin in healthy white males. Br J Clin Pharmacol 2002;54:610-616.

Chapter 2.1

46

20. Sakaeda T, Nakamura T, Horinouchi M, et al. MDR1 genotype related pharmacokinetics of digoxin after single oral administration in healthy Japanese subjects. Pharm Res 2001;18:1400-1404.

21. Horinouchi M, Sakaeda T, Nakamura T, et al. Significant linkage of MDR1 polymorphisms at positions 3435 and 2677: functional relevance to pharmacokinetics of digoxin. Pharm Res 2002;19:1581-5.

22. Abbott Laboratories. Digoxin TDx (No 9511). August 2000. 23. Bosch TM, Doodeman VD, Smits PHM, Meijerman I, Schellens JHM, Beijnen JH. Pharmacogenetic

screening for polymorphisms in drug-metabolizing enzymes and drug transporters in a Dutch population. Mol Diagn Ther 2006; 10:175-85.

24. Beal SL, Sheiner LB. NONMEM User´s guides. NONMEM Project Group, University of California at San Francisco, 1998.

25. Frankfort SV, Doodeman VD, Bakker R, et al. ABCB1 genotypes and haplotypes in patients with dementia and age-matched non-demented control patients. Mol Neurodegener 2006;1:13.

26. Kroetz DL, Pauli-Magnus C, Hodges LM, et al. Pharmacogenetics of membrane transporters investigators: Sequence diversity and haplotype structure in the human ABCB1 (MDR1, multi drug resistance transporter) gene. Pharmacogenetics 2003;13:481-494.

27. Tang K, Ngoi S, Gwee P, et al. Distict haplotype profiles and strong linkage disequilibrium at the MDR1 multidrug transporter gene locus in three ethnic Asian populations. Pharmacogenetics 2002;12:437-450.

47

48

49

CHAPTER 3

RIVASTIGMINE

50

51

CHAPTER 3.1

BIOANALYSIS OF RIVASTIGMINE

52

53

CHAPTER 3.1.1 A simple and sensitive assay for the quantitative analysis of

rivastigmine and its metabolite NAP 226-90 in human EDTA plasma using coupled liquid chromatography and tandem

mass spectrometry

S.V. Frankfort, M. Ouwehand, M.J. van Maanen, H. Rosing, C.R. Tulner, J.H. Beijnen

Abstract

A sensitive and specific LC-MS/MS assay for the determination of rivastigmine and its major metabolite NAP 226-90 is presented. A 100 µL plasma aliquot was spiked with a structural analogue of rivastigmine as internal standard (PKF214-976-AE-1) and proteins were precipitated by adding 200 µL of methanol. After centrifugation 100 µL of the clear supernatant was mixed with 100 µL of methanol-water (30:70, v/v) and 25 µL-volumes were injected onto the HPLC system. Separation was acquired on a 150 x 2.0 mm ID Gemini C18 column using a gradient system with 10 mM ammonium hydroxide and methanol. Detection was performed by using a turboionspray interface and positive ion multiple reaction monitoring by tandem mass spectrometry. The assay quantifies rivastigmine from 0.25 to 50 ng/mL and its metabolite NAP 226-90 from 0.50 to 25 ng/mL, using 100 µL human plasma samples. Validation results demonstrate that rivastigmine and metabolite concentrations can be accurately and precisely quantified in human EDTA plasma. This assay is now used to support clinical pharmacologic studies with rivastigmine.

Rapid Communications in Mass Spectrometry 2006;20:3330-3336

Chapter 3.1.1

54

Introduction Rivastigmine, ((S)- N- ethyl-3- [(1- dimethylamino) ethyl] –N –methyl - phenylcarbamate (Exelon®)), an acetylcholinesterase inhibitor, has shown efficacy in the symptomatic treatment of mild to moderately severe Alzheimer’s dementia (AD)1,2. Rivastigmine interacts with acetylcholinesterase resulting in a carbamylated complex that slowly breaks down to form free enzyme. This temporary inhibition of acetylcholinesterase (AChE), leads to an increased availability of acetylcholine in cholinergic neurons of the brain3. Cholinergic deficits lead to cognitive and behavioural disturbances in Alzheimer’s disease4 and increasing acetylcholine availability ameliorates these disturbances5. Patients respond very differently to rivastigmine. Efficacy ranges from continuation of deterioration or maintaining baseline levels to a clear clinical effect6,7. In addition, many patients discontinue therapy within 6 months due to experience of adverse events8, most frequently nausea, vomiting and diarrhoea9. Therefore, it would be ideal to support rivastigmine treatment in clinical practice with therapeutic drug monitoring (TDM) by measuring plasma concentrations and relating those to adverse events and efficacy. Rivastigmine is extensively metabolised by the target enzyme acetylcholinesterase to the decarbamylated metabolite NAP 226-90 ((S)-3-(1-dimethylamino-ethyl)-phenol). In an ex-vivo experiment in rat brain samples and plasma it was shown that increase in NAP 226-90 concentration correlated well with enzyme inhibition, so the concentration of NAP 226-90 reflects the extent of enzyme inhibition10. Two methods are described in literature that measured rivastigmine and NAP 226-90 in plasma by liquid chromatography coupled to the tandem mass spectrometry (LC-MS/MS). Pommier et al.11 developed a method which uses a volume of 0.5 mL heparinised plasma, stable isotopically labelled internal standards for both rivastigmine and NAP 226-90, derivatisation of NAP 226-90 and extraction from plasma using liquid-liquid extraction (LLE). Enz et al.10 also described a method in heparinised plasma. The internal standard in this method was not isotopically labelled and smaller volumes of plasma (0.1 mL) were used and rivastigmine and metabolite were extracted by a laborious LLE with ethylacetate from plasma. To support clinical studies in our hospital regarding TDM of rivastigmine, a sensitive LC-MS/MS assay was developed. Our aim was to develop a simple, rapid and sensitive method for measuring rivastigmine and NAP 226-90 in a matrix of EDTA plasma and to simplify the procedure for the extraction of the analytes from plasma. We achieved our goals successfully by developing a simple, sensitive and reproducible assay with protein precipitation (PP) as sample pre-treatment, using a structural analogue as internal standard instead of a non-commercial available isotopically labelled one. The assay has been successfully applied in clinical pharmacologic studies with rivastigmine.

Bioanalysis of rivastigmine and NAP 226-90

55

Experimental Materials Rivastigmine (C14H22N2O2; Figure 1A), its metabolite NAP 226-90 (C10H15NO, Figure 1B) and the internal standard PKF214-976-AE-1 (C15H25N2O2, Figure 1C) were kindly supplied by Novartis Pharma AG (Basel, Switzerland). Methanol (LC gradient grade) was obtained from Bissolve Ltd. (Amsterdam, The Netherlands). Ammonia 25% (analytical grade) was from Merck (Amsterdam, The Netherlands). Distilled water was used throughout the analyses. Drug free human EDTA plasma was obtained from the Central Laboratory for Blood Transfusion (Sanquin, Amsterdam, The Netherlands). Figure 1. Chemical structures of rivastigmine (A), NAP 226-90 (B) and internal standard PKF214-976-AE-1 (C).

N O

N

O

(C)(A)

O

O

N

N OH

N

(B)

Liquid Chromatograpy The HPLC system comprised an HP1100 (Agilent Technologies, Palo Alto, CA) binary pump, degasser and HP1100 (Agilent Technologies) autosampler. Separation was performed using a Gemini C18 column (150 x 2.0 mm ID, particle size 5 µm, Phenomenex, Torrance, CA, USA) with gradient elution. The mobile phase was set at a mixture of 10 mM ammonium hydroxide – methanol (50:50, v/v) for 1 minute followed by a block gradient to 10 mM ammonium hydroxide in water – methanol (5:95, v/v) in 0.1 minute, at which it was kept for 7 minutes. The flow-rate was set at 0.2 mL/min and 25 µL was injected. The autosampler temperature was set at 10 oC.

Chapter 3.1.1

56

Mass spectrometry An API 2000 triple quadrupole MS equipped with an turbo-ionspray (Sciex, Thornhill, ON, Canada) was used. Positive ion multiple reaction monitoring (MRM) was performed for selective and sensitive detection. Data were acquired and processed using the proprietary software application Analyst™ (version 1.2, Sciex, Canada). Mass transitions of m/z 251 to 206 and 166 to 121 were optimised for rivastigmine and NAP226-90, respectively. Dwell times of 700 ms were used. Mass transition of m/z 263 to 218 was optimised for the internal standard PKF-214-976-AE-1 with a dwell time of 700 ms. Nebulizer and turbo gas (both compressed air) operated at 40 psi and 30 psi, respectively. The curtain gas was set at 35 psi and the collision gas (N2) at 3 psi. The ionspray voltage was set at 5500 V, with a source temperature of 350 oC. Preparation of stock and working solutions Two sets of stock solutions of both rivastigmine and NAP 226-90 were prepared from independent weightings in methanol-water (30:70, v/v) at a concentration of 0.1 mg/mL. One stock solution was used to prepare calibration standards, the other to prepare quality control samples. Stock solutions were further diluted in methanol-water (30:70, v/v) to working solutions in a range of 10 to 10,000 ng/mL for preparing calibration standards for rivastigmine and in a range of 20 to 10,000 ng/mL for NAP 226-90. Additionally, stock solutions were further diluted in methanol-water (30:70, v/v) to working solutions in a range of 100 to 10,000 ng/mL for preparing quality control samples for rivastigmine and NAP 226-90. A stock solution of PKF214-976-AE-1 internal standard was prepared in water at a concentration of 0.1 mg/mL. The stock solution was further diluted with water to obtain a working solution of 100 ng/mL. Then this solution was further diluted in methanol to a final concentration of 5 ng/mL. All solutions were stored at 2-8 oC. Preparation of calibration standards and quality control samples in human plasma Control human EDTA plasma was centrifuged for approximately 5 minutes at 1,000 g. Calibration standards were prepared freshly in EDTA plasma in a range from 0.25 to 50 ng/mL for rivastigmine and from 0.50 to 25 ng/mL for NAP 226-90, and vortex mixed for approximately 30 seconds before processing. Standards were processed and analysed in duplicate. Validation samples for rivastigmine and NAP 226-90 were spiked separately to control human EDTA plasma and stored at -20 oC. Concentrations of 0.25, 0.75, 15 and 40 ng/mL for rivastigmine and 0.5, 1.5, 10 and 20 ng/mL for NAP 226-90 were prepared.


57

Sample preparation Plasma proteins were precipitated by adding 200 µL of methanol, which contained the internal standard in a concentration of 5 ng/mL, to 100 µL of sample. The samples were vortexed for 10 seconds and mixed for 10 min at 1250 rpm. Next, samples were centrifuged for 10 min at 23,100 g and 100 µL of the clear supernatant was mixed with 100 µL methanol-water (30:70, v/v) and transferred to a glass autosampler vial with insert. A volume of 25 µL was injected onto the HPLC column. Validation procedures Validation of the assay for quantitation of rivaStigmine and NAP226-90 in EDTA plasma included linearity, accuracy, precision, specificity, selectivity, ion suppression, recovery and stability12. We prepared and analysed calibration standards in duplicate in three analytical runs. In order to establish the best weighting factor back-calculated calibration concentration was determined. The model with the lowest total bias and most constant bias across the range was considered the best fit. The linearity was evaluated by means of back-calculated concentrations of the calibration standards. The deviations from the nominal concentrations should be within ± 20% for the Lower Limit Of Quantitation (LLOQ) and within ± 15% for other concentrations with coefficient of variation (C.V.) values less than 20% and 15% for both the LLOQ and the other concentrations, respectively12. Five replicates of the independently prepared QC samples in plasma were analysed together with calibration standards in three analytical runs. Accuracies were determined as the percentage difference of the measured concentration from the nominal concentration and the coefficient of variation was used to report the precision. The intra and inter-assay accuracies (% bias) should be within ±20% at the LLOQ level and within ±15% at the other concentrations12. The intra and inter-assay precisions should be ±20% at the LLOQ level and ±15% at the other concentrations12. To investigate whether endogenous matrix constituents interfered with the assay, six individual batches of control drug-free plasma samples containing neither analyte nor internal standard (double blank), samples containing only internal standard (blank), and LLOQ samples were prepared. Samples were processed according to the described procedures and analySed. Peak areas of compounds co-eluting with the analyte or internal standard should not exceed 20% of the analyte peak area at the LLOQ or 5% of the internal standard area. Deviations from the nominal concentrations should be within ± 20%12. For the determination of ion suppression, control drug-free plasma was processed and dry extracts were dissolved with working solutions, containing the analytes and internal standard in methanol-water (1:1, v/v), that represented 100% recovery. Ion-suppression was determined by comparing the analytical response of these samples to that of the working

Chapter 3.1.1

58

solutions, the loss of signal represents the ion-suppression. Recovery was determined by comparing the analytical response of processed QC samples with the analytical response of blank samples reconstituted with working solutions as described above. These experiments were performed in triplicate at three concentration levels. Overall recovery corresponded to the net response after subtraction of the ion-suppression and signal loss due to the extraction. Ion suppression and recovery experiments for the internal standard were performed in a similar way. The stability of rivastigmine and metabolite in spiked human EDTA plasma samples after 3 freeze-thaw cycles from nominally –20 °C to ambient temperature was assessed in triplicate at 0.75 and 40 ng/mL for rivastigmine and 1.5 and 20 ng/mL for NAP 226-90. The mean rivastigmine and metabolite concentrations after 3 freeze-thaw cycles were compared against freshly prepared extracts. The stability of rivastigmine and its metabolite in spiked human EDTA plasma samples maintained at ambient temperature for 2 and 24 hours was evaluated at 0.750 and 15 ng/mL for rivastigmine and 1.5 and 10 ng/mL for the metabolite. Stability was evaluated by comparing the rivastigmine and NAP226-90 concentrations against freshly prepared extracts. Additionally, processed sample stability was assessed at 0.75 and 40 ng/mL for rivastigmine at 10 °C in the autosampler after 3 days of storage and after 8 days of storage at 2-8 °C by comparing the mean rivastigmine concentration against freshly prepared extracts. For the metabolite, processed sample stability was assessed at 1.5 and 20 ng/mL under the same conditions as described for rivastigmine. Finally, re-injection reproducibility of rivastigmine and its metabolite in human plasma was determined in duplicate at three concentration levels (0.75, 15 and 40 for rivastigmine; 1.5, 10 and 20 ng/mL for NAP 226-90) after 24 hours at nominally 10 °C by comparing the mean concentrations against freshly prepared and analysed extracts. Rivastigmine and its metabolite are considered stable in biological matrix or extracts when 80-120% of the initial concentrations are found12. Clinical Study The analytical method described in this article has been used to support a clinical study in patients with Alzheimer’s disease or Lewy Body Dementia. In this study rivastigmine is twice-daily orally administrated. Daily doses of rivastigmine ranged between 3 and 12 mg. Blood samples were collected up to 7h after intake. After collection, blood samples were immediately centrifuged (1,000g for 10 min) and the plasma layer was stored at – 20 oC until analysis.


59

Results and discussion Method development LLE with ethyl acetate, based on the sample pre-treatment method described by Enz et al. 10, was tested. Plasma was mixed with 0.1 M sodium carbonate (pH 11) and extracted with ethyl acetate. Recoveries of approximately 60% were obtained. To increase the extraction recovery, diethyl ether and methyl-tert.-butyl ether were tested. With diethyl ether the recovery increased to 80%. However, a high variation in extraction recovery was observed for all tested extraction solvents, resulting in poor accuracies and precisions. Then protein precipitation with trichloro acetic acid, acetonitrile and methanol was tested. The use of trichloro acetic acid resulted in inclusion of the analytes in the protein precipitate. Protein precipitation with acetonitrile and methanol resulted in recoveries of approximately 60%. As the accuracy and precision after protein precipitation with methanol was better, it was chosen as sample pre-treatment for the extraction of rivastigmine and NAP 226-90 from human EDTA plasma. To obtain high sensitivity and selectivity MS/MS detection was chosen. The Q1 mass spectra of rivastigmine, NAP 226-90 and internal standard showed molecular ions [M+H]+ at m/z 251, 166 and 263, respectively. Fragment ions, resulting from cleavage of the tertiary dimethyl amine moiety, were seen for rivastigmine, NAP 226-90 and internal standard at m/z 206, 121 and 218, respectively. Figure 2. Product ion spectrum of rivastigmine (precursor ion m/z 251).

0.0E+00

2.0E+07

4.0E+07

6.0E+07

50 75 100 125 150 175 200 225 250 275 300m/z

Inte

nsity

(cps

)

O

O

N

N

206

206

251

The molecular ion of rivastigmine at m/z 251, of NAP 226-90 at m/z 166 and of internal standard at m/z 263 were used as precursor ions to generate the product ion spectra presented in figures 2, 3 and 4, for rivastigmine, NAP 226-90 and IS, respectively. The

Chapter 3.1.1

60

main fragment ions of rivastigmine (m/z 206), NAP 226-90 (m/z 121) and internal standard (m/z 218) were most abundant for the analytes and were used for MRM. The proposed fragmentation patterns for rivastigmine, NAP 226-90 and internal standard are presented in figures 2, 3 and 4, respectively. Figure 3. Product ion spectrum of NAP 226-90 (precursor ion m/z 166).

0.0E+00

5.0E+06

1.0E+07

1.5E+07

2.0E+07

50 75 100 125 150 175 200 225 250m/z

Inte

nsity

(cps

)

OH

N

121

121

166

Figure 4. Product ion spectrum of PKF-214-976-AE-1 (precursor ion m/z 263).

0.0E+00

5.0E+06

1.0E+07

100 125 150 175 200 225 250 275 300m/z

Inte

nsity

(cps

)

263

N O

N

O

218

218

Two assays have described the quantitative analysis of rivastigmine and NAP 226-90 in biological fluids using LC-MS techniques10,11. In these assays the analytes are chromatographically separated from matrix components using isocratic conditions and with mobile phases methanol-0.02 M ammonium acetate (55:45, v/v)11 and acetronitrile-water (80:20, v/v) containing 0.1% formic acid.10 In our laboratory, however, the best results were


61

obtained with gradient elution with a mobile phase of 10 mM ammonium hydroxide in water – methanol (50:50, v/v to 5:95, v/v). A basic mobile phase can be well suited for bioanalysis of basic compounds in combination with positive ionisation as shown earlier in our department for paclitaxel, an anticancer drug, and its metabolites13,14. Higher intensities were shown for the main fragment ions of both rivastigmine and NAP 226-90 under basic conditions compared to acidic conditions (table 1). Due to the use of protein precipitation as sample pre-treatment it was necessary to use gradient elution in order to elute contaminants from the column separately from the analytes to prevent ion-suppression from endogenous compounds. Using this system, the retention time of rivastigmine was 6.2 min, of NAP 226-90 5.5 min and for the internal standard 6.0 min. Representative HPLC-MS/MS chromatograms of an LLOQ sample for rivastigmine, NAP 226-90 and the internal standard from control human plasma are depicted in figure 5. Table 1. Peak intensities for rivastigmine and NAP 226-90 and their product ions in acidic and basic mobile phase conditions. Compound (mass) Intensity (cps) Acidic (Formic Acid; 1µg/mL*) Basic (Ammonium Hydroxide; 0.1 µg/mL*) Rivastigmine (251) 241060 199000 Product ion (206) 49000 167000 NAP 226-90 (166) 908720 3129240 Product ion (121) 409000 940000 * concentration of analytes (rivastigmine or NAP 226-90)

Validation of the assay The assay was linear over a concentration range from 0.25 to 50.0 ng/mL for rivastigmine and 0.50-25.0 ng/mL for the metabolite in human plasma. The linear regression of peak area ratio versus the concentration 1/x2 (the reciprocal of the squared concentration) was weighted to obtain the lowest total bias and the most constant bias across the range. Correlation coefficients of the calibration curves for rivastigmine and metabolites respectively were always better than 0.99. At all concentration levels deviation of measured concentrations from nominal concentration were between -7.0 and 2.8 % for rivastigmine and between -3.5 and 5.0% for NAP 226-90. CV values less than 8.8 and 10.8 % were obtained for rivastigmine and NAP 226-90, respectively. Rivastigmine and NAP 226-90 proved to be stable in the supernatant after protein precipitation for 8 days when stored at 2-8 °C. Processed samples were stable for 72h when stored at 10 °C. Finally, re-injection reproducibility was established, the analytical run can be re-injected after at least 24 hours of storage in the autosampler (10°C).

Chapter 3.1.1

62

Figure 5. Representative HPLC-MS/MS chromatograms of an LLOQ sample for rivastigmine (A, 0.25 ng/mL), NAP 226-90 (B, 0.5 ng/mL) and the internal standard from control plasma (C, 10.0 ng/mL).

01020304050

0 2 4 6 8 10 12 14time (min)

Inte

nsity

(cps

) 251 →206

01020304050

0 2 4 6 8 10 12 14time (min)

Inte

nsity

(cps

)

166 →121

(B)

0

400

800

1200

0 2 4 6 8 10 12 14time (min)

Inte

nsity

(cps

)

(C)

(A)

Assay performance data for rivastigmine and its metabolite are summarised in tables 2A and 2B, respectively. The intra-assay accuracies (% bias) for rivastigmine were within ± 14.3% for the LLOQ and within ± 9.1% for other concentrations and found to be acceptable12. The intra-assay accuracies (% bias) for NAP226-90 were within ± 8.1% for the LLOQ and ±14.0 % for all concentrations. The intra-assay precisions for rivastigmine were less than 15.4 % at the LLOQ level and less than 10.4% for all other concentrations. The intra-assay precisions for its metabolite NAP226-90 were less than 14.0 % at the

263 218


63

LLOQ level and less than 14.4% for all concentrations. Thus, based upon acceptable accuracy and precision, the validated range for rivastigmine based on 100 µL of human plasma is from 0.25 to 50.0 ng/mL for rivastigmine and for the metabolite NAP 226-90 from 0.25 to 25.0 ng/mL. MRM chromatograms of six batches of control drug-free plasma contained no co-eluting peaks >20% of the rivastigmine and its metabolite area at the LLOQ level, and no co-eluting peaks >5% of the area of internal standard. Deviations from the nominal concentrations at the LLOQ level were within ±19.6 % for rivastigmine. Deviations from the nominal concentrations at the LLOQ level for NAP 226-90 were within ± 15.3 %. The mean ion-suppression for rivastigmine and its metabolite were 9.4 and 19.9%, respectively. The mean ion-suppression for the internal standard was 17.3%. The mean recovery after protein precipitation was 70.9, 85.7, and 68.2% for rivastigmine, NAP 226-90 and the internal standard, respectively. Total recoveries obtained for rivastigmine, NAP 226-90 and internal standard were 63.8, 68.4, and 56.4 %, respectively. Stability data of rivastigmine and NAP 226-90 are presented in table 3A and 3B, respectively. Rivastigmine in not stable in EDTA plasma when kept at ambient temperatures for more than 2 hours. Less than 80% rivastigmine was recovered at low concentrations after 2 hours at ambient temperatures, whereas an increase of more than 20% was observed for NAP 226-90. Rivastigmine and NAP 226-90 are stable during 1 freeze/thaw cycle, however a decrease in rivastigmine and increase in NAP 226-90 was seen after 2 and 3 freeze/thaw cycles (data not shown). NAP 226-90 is formed by enzyme- catalysed hydrolysis of rivastigmine. These cholinesterases are present in plasma and explain the degradation of rivastigmine in EDTA plasma when kept at ambient temperatures. Pommier et al. 11, added a competitive cholinesterase inhibitor (physostigmine hemisulphate) to heparin plasma to prevent degradation of rivastigmine. The ability to stabilise EDTA plasma with a competitive cholinesterase inhibitor should be further investigated. To prevent the ex-vivo hydrolysis of rivastigmine, samples obtained in the clinical study were immediately centrifuged and stored at -20°C within 30 minutes after collection and during analysis samples were processed directly after thawing.

Chapter 3.1.1

64

Table 2A. Assay performance data for rivastigmine Run Nominal

concentration (ng/mL)

Mean calculated concentration (ng/mL)

Accuracy (% deviation)

Precision (% CV)

Number of replicates

1 0.250 0.214 -14.3 10.5 5 2 0.250 0.217 -9.1 13.1 5 3 0.250 0.254 1.7 15.4 5 Average 0.250 0.228 -7.6* 14.7# 15 1 0.749 0.739 -1.3 4.7 5 2 0.749 0.706 -5.8 7.1 5 3 0.749 0.780 4.1 8.9 5 Average 0.749 0.742 -0.9* 7.9# 15 1 15.0 15.2 1.5 9.7 5 2 15.0 15.3 2.1 5.1 5 3 15.0 14.4 -4.0 8.5 5 Average 15.0 14.9 0.1* 7.9# 15 1 39.9 43.1 8.0 3.1 5 2 39.9 39.5 -1.1 6.0 5 3 39.9 36.3 -9.0 10.4 5 Average 39.9 39.6 -0.8* 9.6# 15 *=Inter-assay accuracy(%), # =Inter-assay precision (%)

Table 2B. Assay performance data for NAP226-90 Run Nominal

concentration (ng/mL)

Mean calculated concentration (ng/mL)

Accuracy (% deviation)

Precision (% CV)


1 0.537 0.563 4.8 10.7 5 2 0.537 0.580 8.0 11.5 5 3 0.537 0.551 2.5 14.0 5 Average 0.537 0.565 5.8* 11.4# 15 1 1.61 1.63 -12.0 13.2 5 2 1.61 1.42 8.1 13.6 5 3 1.61 1.74 13.2 8.1 5 Average 1.61 1.60 -0.9* 13.9# 15 1 10.7 10.3 -3.8 8.2 5 2 10.7 9.78 -8.6 14.9 5 3 10.7 10.9 2.2 5.7 5 Average 10.7 10.3 -3.7* 10.4# 15 1 21.5 20.8 -3.3 11.0 5 2 21.5 20.0 -7.1 14.1 5 3 21.5 24.5 14.0 9.5 5 Average 21.5 21.8 1.3* 14.4# 15 *=Inter-assay accuracy(%), # =Inter-assay precision (%)


65

Table 3A. Stability data for rivastigmine Conditions Matrix Initial

conc (ng/mL)

Found conc (ng/mL)

Dev (%)

CV (%)


Ambient, 2 h Plasma 0.791 0.615 -22.3 8.5 3 15.0 13.5 -10.0 3.7 3 Ambient 24 h Plasma 0.772 0.100 -87.0 33.6 3 38.7 26.9 -30.6 11.5 3 1 Freeze (-20°C)- thaw cycle Plasma 0.715 0.669 -6.4 14.6 3 40.8 35.1 -13.9 8.5 3 2-4°C, 8 days Supernatant 0.669 0.524 -21.7 5.0 3 after PP 37.9 35.6 -5.90 9.3 3 Reinjection reproducibility Processed 0.587 0.657 12.0 2.2 2 10°C, 24 h sample 14.2 12.6 -11.3 2.2 2 33.7 37.3 10.8 6.8 2 Autosampler, 10°C, 72 h Processed 0.669 0.636 -5.0 2.1 3 sample 38.4 33.7 -12.3 5.2 3 PP=protein precipitation

Table 3B. Stability data for NAP226-90 Conditions Matrix Initial

conc (ng/mL)

Found conc (ng/mL)

Dev (%)

CV (%)


Ambient, 2 h Plasma 1.60 2.03 26.9 21.6 3 10.0 12.4 23.8 2.9 3 Ambient 24 h Plasma 1.58 1.58 0.42 4.2 3 20.3 23.7 16.7 3.8 3 1 Freeze (-20°C)- thaw cycle Plasma 1.47 1.52 3.4 12.8 3 20.8 23.0 10.6 3.8 3 2-4°C, 8 days Supernatant 1.92 1.86 -3.3 1.1 3 after PP 25.4 23.3 -8.4 7.7 3 Reinjection reproducibility Processed 1.72 1.82 5.8 1.2 2 10°C, 24 h sample 13.2 12.7 -3.4 10.0 2 21.5 27.3 13.3 3.6 2 Autosampler, 10°C 72 h Processed 1.84 1.77 -4.2 3.3 3 sample 23.5 25.5 8.5 2.7 3 PP=protein precipitation

Chapter 3.1.1

66

Figure 6. Concentration vs. time profiles of rivastigmine and NAP 226-90 in a patient treated orally with twice daily 3 mg rivastigmine (A) and in a patient treated orally with twice daily 6 mg rivastigmine (B).

0

10

20

30

40

0 2 4 6 8

Time (h)

Conc

entr

atio

n (n

g/m

L) RivastigmineNAP-226-90

(B)

0

5

10

15

20

0 2 4 6 8

Time (h)

Conc

entr

atio

n (n

g/m

L) RivastigmineNAP-226-90

(A)


67

Clinical study In figure 6, the concentration vs. time plots are presented for rivastigmine and NAP 226-90 from a patient treated orally with twice daily 3 mg rivastigmine and of a patient treated orally with twice daily 6 mg rivastigmine. A maximum rivastigmine concentration of 18.4 ng/mL was reached 0.5 hours after administration of the drug in the patient treated with twice daily 3 mg; the highest plasma level for NAP 226-90 was 6.2 ng/mL. The maximum concentration of rivastigmine (36.3 ng/mL) was reached 2 hours after administration of the drug in the patient treated with twice daily 6 mg; the highest plasma level for NAP 226-90 was 8.0 ng/mL. It is clear that rivastigmine levels in the patient treated with twice daily 6 mg are approximately twice the level compared to that in the patient treated with twice daily 3 mg. This, however, is not shown for the metabolite NAP 226-90 that reached maximum levels between 6 and 8 ng/mL for those 2 patients.

Conclusions For the quantitation of rivastigmine and NAP 226-90 in human plasma, a simple, accurate, reproducible and selective LC-MS/MS assay has been developed. The assay quantifies a range for rivastigmine from 0.25 ng/mL to 50 ng/mL and for NAP 226-90 from 0.50 ng/mL to 25 ng/mL using 100 µL human plasma aliquots. Validation results demonstrate that the rivastigmine and metabolite concentrations can be accurately and precisely quantified in human plasma. This assay is now used to support clinical pharmacologic studies with rivastigmine.

References 1. Corey-Bloom J, Anand R, Veach J. A randomized trial evaluating the efficacy and safety of ENA 713

(rivastigmine tartrate), a new acetylcholinesterase inhibitor, in patients with mild to moderately severe Alzheimer’s disease. International Journal of Geriatric Psychopharmacology 1998;1:55-65.


3. Jann MW, Shirley KL, Small GW. Clinical pharmacokinetics and pharmacodynamics of cholinesterase inhibitors. Clin Pharmacokinet 2002;41:719-739.

4. Francis PT, Palmer AM, Snape M, Wilcock G. The cholinergic hypothesis of Alzheimer's disease: a review of progress. J Neurol Neurosurg Psychiatry 1999;66:137-147.

5. Davis KL, Mohs RC, Marin D, et al. Cholinergic markers in elderly patients with early signs of Alzheimer’s disease. JAMA 1999;281:1401-6.


7. Frankfort SV, Appels BA, de Boer A, et al. Treatment effects of rivastigmine on cognition, performance of daily living activities and behaviour in Alzheimer’s disease in an outpatient geriatric setting. Int J Clin Pract 2006;60:646-654.

8. Frankfort SV, Appels BA, de Boer A, et al. Discontinuation of rivastigmine in routine clinical practice. Int J Geriatr Psychiatry 2005;20:1167-71.

Chapter 3.1.1

68

9. Gauthier S. Cholinergic adverse effects of cholinesterase inhibitors in Alzheimer’s disease. Drugs Aging 2001;18: 853-62.

10. Enz A, Chappuis A, Dattler A. A simple, rapid and sensitive method for simultaneous determination of rivastigmine and its major metabolite NAP 226-90 in rat brain and plasma by reversed-phase liquid chromatography coupled to electrospray ionization mass spectrometry. Biomed Chromatogr 2004;18:160-166.

11. Pommier F, Frigola R. Quantitative determination of rivastigmine and its major metabolite in human plasma by liquid chromatography with atmospheric pressure chemical ionization tandem mass spectrometry J Chromatogr B Analyt Technol Biomed Life Sci 2003;784:301-313.

12. Rosing H, Man WY, Doyle E, Bult A, Beijnen JH. Bioanalytical liquid chromatographic method validation. A review of current practices and procedures. J Liq Chrom Rel Technol 2000; 23: 329-354.

13. Vainchtein LD, Thijssen B, Stokvis E, et al. A simple and sensitive assay for the quantitative analysis of paclitaxel and metabolites in human plasma using liquid chromatography /tandem mass spectrometry. Biomed Chromatogr 2006;20:139-148.

14. Stokvis E, Ouwehand M, Nan LGAH, et al. A simple and sensitive assay for the quantitative analysis of paclitaxel in human and mouse plasma and brain tumor tissue using coupled liquid chromatography and mass spectrometry. J Mass Spectr 2004;39:1506-1512.

69

70

71

CHAPTER 3.2

RIVASTIGMINE PHARMACOTHERAPY AND CLINICAL PHARMACOLOGY

72

73

CHAPTER 3.2.1 Discontinuation of rivastigmine in routine clinical practice

S.V. Frankfort, B.A. Appels, A. de Boer, C.R. Tulner,

J.P.C.M. van Campen, C.H.W. Koks, J.H.Beijnen

Abstract Background: Rivastigmine is used for symptomatic treatment of mild-to-moderately severe Alzheimer’s Dementia (AD). We investigated the frequency of and reasons for rivastigmine discontinuation in clinical practice and possible predictive factors for discontinuation within the first six months after starting therapy. Methods: A retrospective cohort study was performed in rivastigmine users, who started therapy in a naturalistic setting. A nurse supported a part of the studied cohort, as this was introduced during the study period. Reasons for discontinuation were investigated, including therapy discontinuation if the Maximum Achieved Dose (MAD) was below 6 mg daily. Predictors of discontinuation within the first half year were investigated by logistic regression analysis. Results: Baseline Mini-Mental-State-Examination (MMSE) of included patients (n=154) was 20.1, mean age was 78.4 years and 70% was female. Within 6 months, 61 users (39.6%) discontinued therapy, primarily (59.0%) for adverse events. Thereafter, the main reason for discontinuation was non-response according to clinimetrics. A MAD during the titration phase of 1.5-4.5 mg/day and absence of nurse support are significantly related to discontinuation within 6 months. Conclusions: Rivastigmine is primarily discontinued within the first six months for intolerable adverse events and thereafter mainly for ongoing deterioration. A MAD of 1.5-4.5 mg/day and the absence of nurse support are independently related to discontinuation of rivastigmine within the initial 6 months.

International Journal of Geriatric Psychiatry 2005;20:1167-1171

Chapter 3.2.1

74

Introduction Rivastigmine (Exelon®), an acetylcholinesterase inhibitor, has shown efficacy in the symptomatic treatment of mild to moderately severe Alzheimer’s dementia (AD)1,2. Known major occurring adverse events of rivastigmine include nausea, vomiting and diarrhoea, whereas bradycardia, dizziness, muscle cramps and weakness are of minor occurrence3. These adverse events are in certain cases reason for discontinuation of rivastigmine, whereas in others it is not continued because of ongoing decline in cognition. Geriatricians started prescribing rivastigmine in 1998 in our hospital and since April 2001 it is the policy at the geriatric department that these patients are supported by a nurse. Support consists of intensive telephone contact during the titration phase and regular telephone contact thereafter. This study aims describing reasons for discontinuation of rivastigmine in routine clinical practice and investigating predictive variables of discontinuation within the first half year after starting rivastigmine. Patients and Methods Patients This study was carried out in patients with mild-to-moderate severe AD, diagnosed according to the NINCDS-ADRDA criteria4, using rivastigmine via the geriatric outpatient department of a general hospital in the Netherlands. Only patients who had responsible relatives or friends who could monitor drug intake were included. Patients were excluded if their data were incomplete. Included patients were followed between May 1998 and September 2004. Dose titration Patients started rivastigmine at 1.5 mg twice daily. If tolerated for minimal 2 weeks doses were increased to 3.0 mg twice daily (until the end of 2001) or to once daily 3.0 mg and once daily 1.5 mg (after 2001). If tolerated for another 2 weeks, then doses were titrated to 4.5 mg twice daily (until the end of 2001) or to 3.0 mg twice daily (after 2001). Patients were further titrated by dose increments of 3.0 mg (until the end of 2001) or by 1.5 mg (after 2001), after tolerating therapy for an interval of 2 weeks at each subsequent dose level, to the individual Maximum Achieved Dose (MAD), up to a maximum of 6 mg twice daily. Rivastigmine was discontinued if daily doses of 6 mg were not achieved in case of adverse events, because lower doses are considered to be associated with less efficacy in retaining cognition, performance and behaviour1,2.

Discontinuation of rivastigmine

75

Neuropsychological assessment At baseline and at 6-months intervals effectiveness was measured in three domains: cognition, performance in daily living activities and behaviour. Decline or improvement was investigated compared to the previous 6-monthly visit. Rivastigmine had to be discontinued if one of the domains showed major deterioration or if minor decline in two domains without improvement in the third domain was shown. Discontinuation criteria are based upon score differences in a historical control cohort of AD patients. This cohort did not use cholinesterase inhibitors and was tested after an interval of 6 months with the same scales as used in our cohort. Design and Statistical analysis A retrospective analysis, describing reasons for and time window of discontinuation was carried out. If patients discontinued therapy because of adverse events, the MAD of rivastigmine was also considered and in these patients multiple reasons for discontinuation were counted. The Pearson Chi-square test for categorical data and independent sample t-tests for continuous data were used to compare patient characteristics who discontinued rivastigmine use within the first six months and who did not. Logistic regression analysis was performed to investigate possible risk factors for discontinuation within the first half year after starting therapy. Age was dichotomised to the mean of the population, ≤ 78.4 versus > 78.4, number of concomitant drugs to none versus ≥ 1, as we hypothesised taking other medication could enhance compliance, baseline MMSE score to ≤ 23 versus > 23, MAD of rivastigmine to < twice daily 3 mg versus ≥ twice daily 3 mg, level of education to low, level 1 through 4, versus high, level 5 through 7 as we used a seven-point scale, ranging from less than 6 years of elementary school (score 1) to a university degree (score 7)5, and titration schedule as 1.5 mg versus 3.0 mg dose increments, because both titration schedules were used during the study period. Gender, involvement of nurse support and place of living were examined as dichotomous variables. When multiple significant (p<0.1) covariates were identified univariately, multivariate logistic analysis was performed. Statistical calculations were performed with SPSS for Windows (version 11.0, SPSS Inc., Chicago, IL, USA). A p-value of 0.05 or less was considered statistically significant.

Chapter 3.2.1

76

Results Patient characteristics Two patients were excluded because their medical records were incomplete. The 154 included rivastigmine users lived primarily at home, had a mean age of 78.4 years, a mean baseline MMSE score of 20.1 and used a mean daily rivastigmine dose of 7.7 mg (Table 1). Table 1. Baseline characteristics.

Within the first half year Baseline characteristics Total population Continueda Discontinueda p-value

n=154 (100%) n= 85 (55.2%) n= 61 (39.6%) Age (years), mean ± SD (range)

78.4± 5.8 (56-89) 78.1 ± 6.0 (56-89) 78.8± 5.6 (66-89) 0.479

Education Level 1-4 (%)/Level 5-7 (%)

110 (71.4)/44 (28.6) 55 (64.7)/30 (35.3) 49 (80.3)/12 (19.7) 0.040

Gender, Female (%) 109 (70.8 57 (67.1) 46 (75.4 0.275 MMSE, mean ±SD (range)

20.1 ± 4.2 (8-28) 20.5± 4.3 (8-28) 19.7± 4.0 (10-28) 0.258

MADb, mean ± SD (range)

7.7 ±3.1 (3.0-12.0) 9.2 ± 2.7 (3.0-12.0) 5.8 ± 2.5 (3.0-12.0) <0.001

No. of drugsc, mean ± SD (range)

2.3 ±2.0 (0-9) 2.4 ±2.0 (0-9) 2.1 ± 2.1 (0-8) 0.422

Nurse support, Yes (%) 85 (55.2)) 54 (63.5) 26 (42.6) Patients place of living At home (%)/ Otherd (%)

140 (90.9)/14 (9.1) 77 (90.6)/8 (9.4) 57 (93.4)/4 (6.6) 0.536

Titration Scheme: Increases Incre 1.5mg (%)/ Incr 3.0mg (%)

86 (55.8)/68 (44.2) 48 (56.5)/37 (43.5) 34 (55.7)/27 (44.3) 0.930

an = 8 (5.2%): unknown if patients discontinued therapy, bMAD= Maximum Achieved Dose, c In addition to rivastigmine, de.g. nursing home, eIncrements

Discontinuation during follow-up Within the first six months 61 users (39.6% of total users) discontinued therapy, primarily for adverse events (n=36, 59.0%) and in 23 patients these adverse events resulted in not achieving a MAD of 6 mg daily. Between 6 and 12 months, 17 users (20.0%) discontinued therapy. Seven patients (41.2%) discontinued for adverse events and 7 (41.2%) because of non-response. Between 12 and 18 months, 11 patients (21.2%) discontinued therapy, primarily because of significant decline in test results (n=9, 81.8%). Other reasons for discontinuation (n=16) included refusing to take rivastigmine, malignancies and transfer to a nursing home, where rivastigmine is frequently discontinued, because of financial limitations (Table 2).


77

Up to 24 months, users were lost to follow-up because of death (n=5) and loss of contact (n=11). Of 154 started patients, 23 users were still continuing therapy by September 2004, ranging from less than 12 months to more than 42 months of treatment. Table 2. Discontinuation during follow-up Number of patients < 6

months 6 - 12 months

12-18 months

18-24 months

24-30 months

30-36 months

36-42 months

42-48 months

At start interval 154 85 52 34 18 7 3 3 Still using rivastigmine 11 5 1 3 2 - 1 Lost to follow-up Died 3 1 1 - - - - - Loss of contact 5 4 1 1 - - - - Discontinued during intervala

61 (39.6) 17 (20.0) 11 (21.2) 14 (41.2) 8 (44.4) 2 (28.5) - 2 (66.7)

Decline Noted by caregiversb 10 (16.4) 2 (11.8) - 1 (7.1) - - - 1 (50.0) Cognitive test resultsb - 7 (41.2) 9 (81.8) 6 (42.9) 5 (62.5) 1 (50.0) - - Adverse eventsb 36c(59.0) 7 (41.2) 1 (9.1) 3c(21.4) 1 (12.5) - - - MAD < 6mg dailyb 23 (37.1) 1 (5.9) 3 (27.3) - - - - - Not compliantb 7 (11.5) - - - - - - - Unknownb - 1 (5.9) - - 1 (12.5) - - - Otherb 8 (11.6) - 1 (9.1) 4 (28.6) 1 (12.5) 1 (50.0) - 1 (50.0) anumber (% of total users at start interval), b number (% of total discontinued users during interval) cincludes 1 bradycardia

Table 3. Factors examined in univariate and multivariate logistic regression

Univariate Multivariate* Independent variable OR 95% CI p-value OR 95% CI p-value

Age < 78.5 years 0.76 0.39-1.48 0.418 Baseline MMSE score <24 1.70 0.78-3.73 0.185 Education level 1-4 2.22 1.03-4.82 0.042 2.21 0.90-5.39 0.082 Living at home 1.48 0.43-5.16 0.538 Male gender 0.66 0.32-1.39 0.276 MAD 1.5-4.5 mg/day 12.26 3.96-37.92 <0.001 11.6 3.65-37.0 <0.001 No co-medication 1.67 0.78-3.60 0.187 No nurse support 2.35 1.20- 4.60 0.013 2.22 1.05-4.73 0.038 Titration: 1.5 mg/day increments 0.93 0.50-1.88 0.930 *corrected for age and gender, CI= Confidence Interval, MMSE = Mini Mental State Examination, MAD = Maximum Achieved Dose, OR = Odds Ratio

Chapter 3.2.1

78

Risk factors for discontinuation within the first half year As can be observed in table 1, the 61 patients who discontinued therapy within the first 6 months had a lower MAD (p<0.001), were less educated and were less frequently supported by a nurse (p<0.05) in comparison to the patients who continued therapy. Results of the logistic regression analysis are shown in table 3. The multivariate analysis showed that a MAD of 1.5-4.5 mg daily (OR 11.6, 95% CI 3.65-37.0 p<0.001) and no nurse support (OR 2.22, 95% CI 1.05-4.73, p=0.038) appeared to be independent predictors of discontinuation within the first half year.

Discussion Rivastigmine was frequently discontinued, because of intolerable adverse events during the first 6 months of treatment and thereafter mainly for ongoing deterioration. Discontinuation within the first 6 months was significantly related with a MAD of 1.5-4.5 mg daily and absence of nurse support. The comparison of our results to those described in literature has limitations. It is difficult to compare cholinesterase studies, because designs are substantially different6. Rivastigmine was evaluated in two large randomised placebo-controlled trials during 26 weeks which were subdivided into low (<6 mg/day) and high (6-12 mg/day) dose groups1,2. In the Newcastle (UK) study, four of 26 patients (15.4%) discontinued rivastigmine therapy because of side effects within the first 4 weeks of treatment7. As we did not investigate discontinuations in this period direct comparisons fall short. An Austrian study8, however, followed 529 patients in usual care during 24 weeks. This period is comparable to our cohort, which was followed for 26 weeks. In the Austrian study there were 67 drop-outs (12.7%) of whom 40.3% experienced side effects in that study. 39 patients were able to continue treatment although not achieving doses of 6 mg daily, which was not possible in our design and partly explains differences in number of discontinuations. As earlier described, patients are urged not to continue therapy if titration to 6 mg daily failed and explains a daily MAD <6 mg as an independent predictor of discontinuation in our clinical setting. Absence of nurse support, also an independent predictor, can be explained because adverse events and changes in titration-rate are discussed in regular telephone calls between the nurse and relatives or close friends of rivastigmine users. Strengths of our study are a relatively large population in a naturalistic setting, a total follow-up time of 42 months and accessibility to all relevant clinical data. In conclusion, initially discontinuation for intolerable adverse events is of major concern. Support by a nurse is important in this first period. After the first period major reason for discontinuation is an ongoing decline in cognition, performance or behaviour.


79


(rivastigmine tartrate), a new acetylcholinesterase inhibitor, in patients with mild to moderately severe Alzheimer’s disease. International Journal of Geriatric Psychopharmacology 1998; 1:55-65.


3. Gauthier S. Cholinergic adverse effects of cholinesterase inhibitors in Alzheimer’s disease. Drugs Aging 2001; 18:853-62.

4. McKahnn G, Drachmann D, Folstein M, et al. Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA work group under the auspices of department of health and human services task force on Alzheimer’s disease. Neurology 1984;34:939-44.

5. Verhage F. Intelligence and age. Van Gorcum: Assen 1964. (in Dutch). 6. Anand R, Hartman R, Sohn H, et al. Impact of study design and patient population on outcomes from

cholinesterase inhibitors trials. Am J Geriatr Psychiatry 2003;11:160-8. 7. Pakrasi S, Mukaetova EB, McKeith IG, et al. Clinical predictors of response to acetylcholinesterase

inhibitors: experience from routine clinical use in Newcastle. Int J Geriatr Psychiatry 2003;18:879-86. 8. Schmidt R, Lechner A, Petrovic K. Rivastigmine in outpatient services: experiences of 114 neurologists in

Austria. International Clinical Psychopharmacology 2002;17:81-85.

80

81

CHAPTER 3.2.2 Treatment effects of rivastigmine on cognition, performance

of daily living activities and behaviour in Alzheimer’s disease in an outpatient geriatric setting

S.V. Frankfort, B.A. Appels, A. de Boer, C.R. Tulner, J.P.C.M. van Campen,

C.H.W. Koks, J.H. Beijnen

Abstract We investigated rivastigmine effectiveness in 84 Alzheimer outpatients, with a special focus on behavioural problems. Cognition, activities in daily living (ADL) and behaviour were assessed during 30 months. Changes in test results between 6 months and baseline were compared with a historical control cohort of Alzheimer patients (n=69) by performing t-tests and calculation of Cohen’s d and standardised response mean (SRM). During 6 months rivastigmine showed effect on cognition (p<0.001, Cohen’s d = 0.33, SRM = 0.78), ADL (p<0.001, Cohen’s d = -0.43, SRM = -0.54) and memory related behaviour (p=0.006, Cohen’s d = -0.28, SRM = -0.28). Depressive behaviour worsened (p=0.001, Cohen’s d = 0.30, SRM = 0.37) and disruptive behaviour (p=0.369, Cohen’s d = -0.07, SRM = -0.09) was not effected by rivastigmine. During 30 months, a gradual decline was shown in most domains. Most RMBPC items showed stabilisation during 30 months. Improvement on disruptive behaviour items and depression items was shown after 6 months of treatment in a large proportion of patients in whom behavioural problems were present at baseline. In conclusion, a huge discontinuation rate is experienced within the first half year of treatment. In the subpopulation of patients who continued rivastigmine for 6 months, it shows modest effectiveness on cognition, functionality and memory associated behaviour compared with historical control patients. Unfortunately, disruptive behaviour is not altered by rivastigmine therapy, and depressive behaviour worsened slightly after initial treatment. During 30 months, rivastigmine showed stabilisation on numerous behaviour items as measured by the RMBPC.

International Journal of Clinical Practice 2006;60:646-654

Chapter 3.2.2

82

Introduction Rivastigmine (Exelon®), an acetylcholinesterase inhibitor, has shown efficacy in the symptomatic treatment of mild-to-moderately severe Alzheimer’s dementia (AD) in randomised placebo-controlled trials. These trials were rather liberal regarding inclusion criteria, for example co-morbidity1,2. Therefore, effectiveness in routine clinical practice would expect to be in a similar range like the trial outcomes. However, patients in trials might respond differently to drugs than patients in routine care. Complementary research in a clinical setting may reveal additional data on the effectiveness of rivastigmine in different domains. To our knowledge, a number of clinically based studies have investigated the effects of rivastigmine on cognition and activities in daily living (ADL) in routine clinical practice3,4. For example, López-Pousa et al.3 investigated cognition by MMSE and Mossello et al.4 investigated cognition by MMSE and performance in daily living activities by ADL and IADL scores. In addition, an open label extension phase of the trials investigated cognition, using the MMSE, up to 5 years of treatment5. Most trials lasted for up to 26 or 52 weeks, some were extended as open-label studies1,2,6. However, the long-term effects of cholinesterase inhibitor therapy remain largely unestablished. Although it is becoming clearer that cholinergic deficits are involved in the behavioural symptoms present in AD7 and that butyrylcholinesterase mediates behaviour8, effect of rivastigmine regarding behaviour is less described as compared with cognition. In a German open-label extension study, B 305, behaviour was measured with the CIBICplus in 34 rivastigmine users9. Meta analyses of three 6-month, double blind, placebo-controlled, regulatory trials investigated behavioural responses with the CIBICplus in rivastigmine users with mild-to-moderate severe AD10. In the clinical setting, research has predominantly been performed in nursing home residents taking rivastigmine, where behaviour was assessed using the NPI-NH. Aupperle et al.11 investigated long term effects during a 52 week open-label study in moderate-to-severe rivastigmine users. Hatoum et al.12 used the occupational disruptive scale of the NPI-NH to investigate the impact of rivastigmine on the disruptive behaviour of nursing home residents. Cummings et al.13 described the effects of rivastigmine treatment on neuropsychiatric and behavioural disturbances also in nursing home residents with moderate to severe probable AD. In the Netherlands, rivastigmine may only be prescribed if evaluation is regularly performed. Patients who are willing to start therapy are tested at baseline and at designated follow-up intervals during treatment. In our hospital, rivastigmine is prescribed via the geriatric outpatient department since 1998, and every 6 months three domains are assessed: cognition, behaviour and performance in daily living activities. In this cohort of patients, we performed retrospective analyses regarding the effectiveness of rivastigmine during 30 months.

Treatment effects of rivastigmine

83

The study aims investigating treatment effects of rivastigmine on different domains, with a special attention to behaviour, in a cohort of outpatients suffering from Alzheimer’s disease.

Patients and methods Patients This retrospective study was carried out in patients with mild-to-moderate, probable or possible AD according to the NINCDS-ADRDA criteria14 and using rivastigmine via the geriatric outpatient department of a general hospital. Only patients who had relatives or friends who could monitor drug intake and patients in whom therapy was evaluated after 6 months were included. Patients were excluded if baseline cognitive test results were incomplete. Dose titration Patients started rivastigmine at 1.5 mg twice daily, and doses were titrated up after intervals of minimal 2 weeks at each dose level, until the individual Maximum Achieved Dose (MAD) up to 6 mg twice daily. During titration, it was possible to omit a dose, to postpone dose increments, or to return to a lower dose level if adverse events required this. Assessment of domains At baseline and at 6-months intervals rivastigmine use was evaluated by assessment of three domains. Cognition was measured with MMSE15 and Cambridge Cognitive Examination (CAMCOG)16. CAMCOG consists of 60 items covering orientation, language, memory, praxis and calculation/attention, abstract reasoning and perception. Total sum scores range from 0 to 107. CAMCOG can be subdivided into a memory (maximum score 37) and a non-memory (maximum score 70) section17. Functional disability was measured with the performance subscale of the Interview for Deterioration in Daily living activities in Dementia (IDDD), a caregiver based paper-and-pencil questionnaire, which consists of 11 items with sum scores ranging from 0 to 4418. Behaviour was measured with the Revised Memory and Behavioural Problems Checklist (RMBPC). It is also a caregiver based paper-and-pencil questionnaire and consists of three subsections. First, a 7-item memory subscale (score per item 0-4, maximum score 28) that includes forgetting recent events, repeated questions, losing things, forgetting the day, forgetting past events, reduced concentration and not finishing tasks. Second, an 8-item disruptive behaviour subscale (score per item 0-4, maximum score 32) including verbal aggression, threats to hurt others, destroying property, to behave dangerous to self or others, talking loudly and rapidly, embarrassing

Chapter 3.2.2

84

behaviour, arguing and waking caregiver up. Third, a 9-item depression subscale (score per item 0-4, maximum score 36) includes comments about hopelessness, comments about being a burden, appearing sad or depressed, comments about dead, comments about being a failure, crying, comments about loneliness, appearing anxious and suicidal threats19. Lower scores on MMSE and CAMCOG and higher scores on the IDDD and subscales of the RMBPC reflect a worse functioning. Data collection All assessment results and demographic variables like gender, age and level of education20 were prospectively incorporated in a database. Place of living, number of concomitant drugs at baseline and MAD, all according to the medical records, were retrospectively obtained. Historical control cohort A research project was carried out at the memory clinic of an academic hospital in Amsterdam. In this project, 69 Alzheimer patients, who did not take rivastigmine, were tested at baseline and after 6 months with MMSE, CAMCOG, IDDD and RMBPC21. These 69 patients represent the historical control group in the present study. Statistical analysis Statistical calculations were performed with SPSS for Windows (version 11.0, SPSS Inc., Chicago, IL, USA). A p-value of 0.05 or less was considered statistically significant. Mean, standard deviation (SD) and range of baseline test results and differences between test results at 6-monthly follow-up visits and baseline were tabulated. For each of the RMBPC subitems, the percentages of patients who improved, deteriorated or remained stable after 6 months were calculated in a subpopulation of patients in whom that particular subitem was present at baseline. Deterioration was defined as an increase of equal or higher than one point and an improvement was defined as a decrease of equal or higher than one point on the RMBPC in this additional analysis. Differences in test results between the first follow-up at 6 months and baseline were compared to the historical control cohort by performing a one-sample t-test. Cohen’s d and standardised response means (SRM), were calculated to assess whether the differences are large enough to be clinically detectable. This is accomplished by dividing the raw effect size by the standard deviation of the measures in their respective populations, as shown in equation 1.

Effect Size (ES) = σ

)Y(Y-)X(X t1-t2t1-t2 (1)


85

To calculate a SRM, σ is the root mean square of the standard deviations of the change scores and to calculate Cohen’s d, σ is the root mean square of the baseline deviations22,23. In our calculations, X is the mean of test results in the rivastigmine group, Y is the mean of test results in the historical control cohort, t2 is after 6 months treatment and t1 is at baseline. Effect sizes greater than 0.20 are conventionally held to be clinically detectable. Effect sizes can be roughly subdivided into small (0.2), medium (0.5) and large (0.8)22. Linear regression analysis were performed to investigate relationships between effectiveness and different dosages.

Results Patient characteristics From the original population of 154 rivastigmine users, 61 patients discontinued treatment within the first 6 months because of adverse events (n=36), decline as noted by caregivers (n=10), incompliance (n=7) or for other reasons (n=8). Eight patients were lost to follow-up, because of death or losing contact within 6 months. After 6 months of treatment, patients mainly discontinued because of ongoing decline in cognition, performance or behaviour. Discontinuation of rivastigmine in this cohort of patients has been described in more detail in a separate publication24. One, six and eight patient(s) were excluded from MMSE/CAMCOG, IDDD and RMBPC analyses, respectively, because baseline data were lacking. In total, 84, 79 and 77 Alzheimer patients were included in this study for analyses on cognition, performance and behaviour, respectively. Table 1. Baseline characteristics. Baseline characteristics (n=84) n (%)* Age, mean ± SD (range) 78.2± 6.0 (56-89) Baseline MMSE, mean ±SD (range) 20.4 ±4.4 (8-28) Education < 6 years elementary school 2 (2.4) 6 years elementary school 19 (22.6) >6 years elementary school 14 (16.7) Domestic science school, junior technical school 20 (23.8) Secondary school for lower educational level 19 (22.6) Secondary school for higher educational level 10 (11.9) University degree 0 (0.0) Gender, female 58 (69.0) MADa, mean ± SD (range) 8.8 ±2.8 (1.5-12.0) Number of concomitant drugs, mean ± SD (range) 2.4 ±2.0 (0-9) Patients place of living, at home 76 (90.5) *unless otherwise noted, aMAD= Maximum Achieved Dose

Chapter 3.2.2

86

Baseline characteristics are summarised in table 1. Included patients had a mean age of 78.2 years, a mean baseline MMSE score of 20.4, 69% was female and patients lived primarily at home. Mean MAD of rivastigmine was 8.8 mg (range: 1.5 –12.0 mg). Table 2. Test results at baseline and differences in test results compared with baseline at subsequent 6-monthly follow-up visits. Test Statistics Baseline* 6 monthsa 12 monthsa 18 monthsa 24 monthsa 30 monthsa

N 84 83j 52 34 18 7 Mean± SD 20.4 ±4.4 0.1±3.1 -0.9±3.3 -1.8±3.8 -1.5±3.3 -2.9±2.3

MMSEb

range 8-28 -11-+7 -8-+9 -9-+6 -7-+3 -6-0 N 84 83j 52 34 15 7 Mean± SD 68.8±13.1 0.5±6.3 -1.2±8.3 -1.8±9.8 -3.4±10.0 -7.9±6.9

CAMCOGc

range 33-92 -15-+13 -17-+17 -20-+20 -27-+15 -17-+4 CAMCOG N 84 83j 52 34 15 7 Memd Mean ± SD 17.4±5.5 -0.5±3.4 -2.0±4.2 -1.6±4.7 -2.4±5.3 -4.9±4.8 range 5-29 -10-+8 -9-+8 -12-+10 -13-+5 -11-+2 CAMCOG N 84 83j 52 34 15 7 Nonmeme Mean ± SD 51.5±9.3 0.9±4.6 0.7±5.8 -0.1±6.7 -1.3±6.8 -2.9±2.5 range 25-67 -9-+16 -13-+15 -18-+15 -21-+10 -6-+2 IDDDf N 79 79 45 30 16 6 Mean± SD 14.0±8.8 0.7±5.9 2.7±7.8 4.5±10.1 5.7±10.1 3.7±11.6 range 0-36 -18-+13 -16-+21 -18-+22 -9-+28 -14-+19 RMBPC N 77 77 46 29 14 5

memg Mean± SD 17.1±3.9 -0.4±4.3 -0.5±5.6 +0.4±5.3 +1.0±4.3 -1.2±1.6 range 8-26 -10-+9 -13-+12 -14-+9 -7-+9 -3-+1 RMBPC N 77 77 46 29 14 5 dish Mean± SD 4.3±3.6 -0.5±2.6 -0.4±3.5 -0.6±2.7 -0.8±2.9 -1.0±2.1 range 0-18 -9-+7 -9-+8 -6-+5 -4-+7 -4-+2 RMBPC N 77 77 46 29 14 5 depi Mean±SD 7.7±5.7 +0.2±4.6 -0.8±4.8 -0.5±4.5 1.9±4.2 -4.0±4.8 range 0-21 -10-+12 -12-+9 -12-+11 -3-+10 -10-+1 *Lower baseline MMSE/CAMCOG scores / higher baseline IDDD/RMBPC scores represent worse cognitive performance, functional performance or behaviour, aNegative differences reflect improvement on IDDD/RMBPC and deterioration on MMSE/CAMCOG, brange 0-30, crange 0-107, dmemory subsection: range 0-37, enon-memory subsection: range 0-70, frange 0-44, gmemory subscale: range 0-28, hdisruptive behaviour subscale: range 0-32, idepression subscale: range 0-36, jresult of 1 patient is missing

Neuropsychological assessment during follow-up Results of neuropsychological assessment at baseline and differences compared to baseline, during follow-up until 30 months, are summarised in table 2. It is shown that group sizes diminish gradually at subsequent follow-up visits. Discontinuation of therapy was mostly due to ongoing decline measured by clinimetrics. Other reasons for discontinuation included occurrence of adverse events, transfer to a nursing home and refusing to take rivastigmine.


87

In table 2, it is shown that therapy resulted in an initial stabilisation (MMSE) or even improvement (total and non-memory section of CAMCOG) of cognition, but thereafter a slow and gradual decline during follow-up. However, the memory section of the CAMCOG showed no initial stabilisation or improvement. Performance of daily living activities (IDDD) showed a gradual decline during follow-up. The memory subsection of the RMBPC showed an initial stabilisation and thereafter a gradual decline. The disruptive subsection of the behaviour scale showed an initial improvement, which stabilises during follow-up. The depression subscale, however, showed an initial increase of depressive symptoms and thereafter an decrease of these symptoms. In table 3 are shown the test results at baseline and differences in test results compared to baseline during 30 months for each RMBPC item. Memory related behaviour disturbances are most apparent, followed by depression and disruptive behaviour. Remarkable are the items “Appears sad or depressed”, “ Appears anxious or worried” and “Arguing, irritability”. From this table, it is clear that during 30 months, numerous items remain stable. However, improvement is shown during 30 months on “losing or misplacing things”, “waking you or your family members at night”, “ arguing, irritability”, “ crying and tearfulness”, “comments about feeling like a failure” and “appears anxious or worried” A separate analysis was carried out in subgroups of patients in whom behavioural problems were present at baseline. The results of this analysis after 6 months are summarised in table 4. The percentage of patients who deteriorated is small for the disruptive behaviour items; a large proportion of patients improved actually. For the depression items also, a minor part of the patients deteriorated; exceptions here are “expressing feelings of hopelessness or sadness about the future” and “comments about feeling worthless or being a burden to others”. For the depression subsection, large number of patients remained stable. For the memory-related behaviour items, around a half of patients remained stable, whereas approximately a quarter deteriorated and a quarter improved at 6 months. Of the patients who initially did not show behavioural problems, the memory-associated behaviour items deteriorated in approximately 50% of patients, the disruptive items deteriorated in around 10% of patients with exception for “doing things that embarrasses you” (25%) and “aggressive to others verbally”(20%) and “arguing” (50%). The depression subitems deteriorated in approximately 25%, with exceptions for “comments about feeling like a failure” (10%) and “appears sad or depressed” (50%) and threats to hurt oneself (7%).

Tab

le 3

.Tes

t res

ults

at b

asel

ine

and

diff

eren

ces i

n te

st re

sults

com

pare

d to

bas

elin

e at

subs

eque

nt 6

-mon

thly

follo

w-u

p vi

sits

, for

eac

h in

divi

dual

R

MB

PC it

em

RM

BPC

Item

B

asel

ine*

N=

77

6 m

onth

sa

N=7

7 12

mon

thsa

N=4

6 18

mon

thsa

N=2

9 24

mon

thsa

N=1

4 30

mon

thsa

N=5

Mem

ory

rela

ted

beha

viou

r

A

skin

g th

e sa

me

ques

tion

over

and

ov

er

2.90

± 0

.77

-0.0

1± 0

.85

-0.0

7 ±

0.80

0.

10 ±

0.8

2 0.

07 ±

0.7

3 0.

00 ±

0.0

0

Trou

ble

rem

embe

ring

rece

nt e

vent

s 2.

81 ±

0.8

1 -0

.09

± 1.

03

-0.0

9 ±

1.15

0.

10 ±

1.1

8 0.

07 ±

0.9

2 -0

.40

± 1.

67

Trou

ble

rem

embe

ring

sign

ifica

nt

past

eve

nts

1.62

± 1

.00

-0.0

8 ±

1.06

0.

15 ±

1.3

0 0.

07 ±

1.0

3 0.

00 ±

1.1

8 -0

.20

± 0.

84

Losi

ng o

r mis

plac

ing

thin

gs

2.91

± 0

.69

-0.2

2 ±

0.94

-0

.28

± 1.

09

-0.1

0 ±

0.90

-0

.07

± 1.

21

-1.4

0 ±

1.14

Fo

rget

ting

wha

t day

it is

2.

74 ±

0.8

4 -0

.04

±0.9

4 0.

00 ±

1.1

0 0.

28 ±

0.9

2 -0

.07

± 0.

73

0.20

± 0

.45

Star

ting

but n

ot fi

nish

ing

thin

gs

1.82

± 1

.06

0.14

± 1

.33

-0.0

2 ±

1.31

0.

14 ±

1.3

0 0.

29 ±

1.6

4 0.

00 ±

0.7

1 D

iffic

ulty

con

cent

ratin

g on

a ta

sk

2.13

± 1

.0

0.13

± 1

.24

0.02

± 1

.13

0.10

± 1

.21

0.07

± 0

.92

0.60

± 0

.90

Dis

rupt

ive

beha

viou

r

D

estro

ying

pro

perti

es

0.22

± 0

.53

-0.0

1 ±

0.60

-0

.02

± 0.

58

0.00

± 0

.54

0.14

± 0

.36

0.20

± 0

.45

Doi

ng th

ings

that

em

barr

asse

s you

0.

77 ±

0.9

2 -0

.08

± 1.

00

-0.0

7 ±

1.00

-0

.21

± 0.

94

0.14

± 0

.77

0.60

± 0

.55

Wak

ing

you

or f

amily

mem

bers

at

nigh

t 0.

39 ±

0.9

1 -0

.10

± 0.

80

-0.1

1 ±

0.88

-0

.21

± 1.

15

-0.1

4 ±

1.29

-0

.40

± 2.

07

Talk

ing

loud

ly a

nd ra

pidl

y 0.

52 ±

0.9

4 -0

.09

± 0.

83

0.02

± 1

.30

-0.1

0 ±

0.98

0.

57 ±

1.1

6 0.

00 ±

0.0

0 A

rgui

ng. i

rrita

bilit

y

1.56

± 1

.09

-0.2

2 ±

1.11

-0

.13

± 1.

05

0.00

± 1

.13

-0.2

1 ±

1.05

-0

.80

± 0.

84

Tab

le 3

.Tes

t res

ults

at b

asel

ine

and

diff

eren

ces i

n te

st re

sults

com

pare

d to

bas

elin

e at

subs

eque

nt 6

-mon

thly

follo

w-u

p vi

sits

, for

ea

ch in

divi

dual

RM

BPC

item

(con

tinue

d)

RM

BPC

Item

B

asel

ine*

N=

77

6 m

onth

sa

N=7

7 12

mon

thsa

N=4

6 18

mon

thsa

N=2

9 24

mon

thsa

N=1

4 30

mon

thsa

N=5

En

gagi

ng

in

beha

viou

r th

at

is

pote

ntia

lly

dang

erou

s to

se

lf or

ot

hers

0.42

± 0

.68

-0.0

1 ±

0.66

-0

.02

± 0.

65

0.07

± 0

.59

-0.0

7 ±

0.62

-0

.20

± 0.

45

Agg

ress

ive

to o

ther

s ver

bally

0.

39 ±

0.7

5 0.

00 ±

0.8

0

0.07

± 0

.85

-0.1

0 ±

1.11

-0

.07

± 1.

39

-0.4

0 ±

0.55

Thre

ats t

o hu

rt ot

hers

0.

06 ±

0.2

5 0.

03 ±

0.4

0 0.

00 ±

0.3

0 -0

.03

± 0.

19

0.07

± 0

.62

0.00

± 0

.00

Dep

ress

ive

beha

viou

r

Th

reat

s to

hurt

ones

elf

0.09

± 0

.33

0.03

± 0

.43

0.11

± 0

.64

0.03

± 0

.19

0.07

± 0

.27

0.20

± 0

.45

App

ears

sad

or d

epre

ssed

1.

49 ±

1.0

3 -0

.08

± 1.

07

-0.0

9 ±

1.05

0.

17 ±

0.8

9 0.

21 ±

0.8

0 -0

.20

± 0.

84

Expr

essi

ng f

eelin

gs o

f ho

pele

ssne

ss

or sa

dnes

s abo

ut th

e fu

ture

1.

06 ±

1.0

4 0.

03 ±

1.0

8 -0

.04

± 0.

97

0.10

± 1

.08

0.64

± 1

.15

-0.2

0 ±

1.10

Cry

ing

and

tear

fuln

ess

0.95

± 1

.08

-0.1

0 ±

0.97

-0

.20

± 0.

91

0.00

± 0

.96

0.07

± 1

.33

-0.8

0 ±

0.84

C

omm

entin

g ab

out

dead

of

self

or

othe

rs

0.66

± 1

.08

0.09

± 0

.71

-0.0

7 ±

0.68

0.

03 ±

0.9

4 0.

29 ±

0.7

3 -0

.80

± 1.

10

Talk

ing

abou

t fee

ling

lone

ly

0.84

± 1

.14

0.06

± 0

.95

0.07

± 1

.06

-0.2

9 ±1

.12

0.00

± 0

.56

-0.4

0 ±

0.89

C

omm

ents

abo

ut f

eelin

g w

orth

less

or

bei

ng a

bur

den

to o

ther

s 0.

65 ±

0.9

0 0.

22 ±

1.0

5 0.

11 ±

0.9

7 0.

10 ±

1.1

5 0.

29 ±

0.9

1 -0

.60

± 1.

14

Com

men

ts

abou

t fe

elin

g lik

e a

failu

re

0.35

± 0

.72

-0.0

4 ±

0.75

-0

.17

± 0.

68

-0.0

3 ±

0.50

-0

.21

± 0.

58

-0.4

0 ±

0.89

App

ears

anx

ious

or w

orrie

d 1.

56 ±

1.1

3 -0

.14

± 1.

07

-0.3

3 ±

1.16

0.

00 ±

0.9

6 -0

.21

± 1.

48

-0.8

0 ±

0.84

Chapter 3.2.2

90

Table 4. Number of patients with behavioural problems at baseline and number of patients (% of total patients experiencing problems at baseline) who showed stabilisation, deterioration or improvement at 6 months. RMBPC-item Baseline Stable Deteriorated Improved

Memory related behaviour Asking the same question over and over 76/77 44 (57.8) 16 (21.1) 16 (21.1) Trouble remembering recent events 76/77 33 (43.4) 22 (29.0) 21 (27.6) Trouble remembering significant past events 66/77 23 (34.8) 14 (21.2) 29 (44.0) Losing or misplacing things 77/77 38 (49.4) 14 (18.2) 25 (32.4) Forgetting what day it is 76/76 32 (42.1) 21 (27.6) 23 (30.3) Starting but not finishing things 68/77 30 (44.1) 18 (26.5) 20 (29.4) Difficulty concentrating on a task 73/77 29 (39.7) 25 (34.3) 19 (26.0) Disruptive behaviour Destroying properties 13/77 2 (15.4) 3 (23.1) 8 (61.5)

Doing things that embarrasses you 38/77 16 (42.1) 4 (10.5) 18 (47.4) Waking you or family members at night 16/77 4 (25.0) 2 (12.5) 10 (62.5) Talking loudly and rapidly 23/77 10 (43.5) 3 (13.0) 10 (43.5) Arguing, irritability 60/77 21 (35.0) 12 (20.0) 27 (45.0) Engaging in behaviour that is potentially dangerous to self or others

25/77 10 (40.0) 3 (12.0) 12 (48.0)

Aggressive to others verbally 19/77 5 (26.3) 1 (5.3) 13 (68.4) Threats to hurt others 5/77 2 (40.0) 0 (0.0) 3 (60.0) Depressive behaviour Threats to hurt oneself 6/77 3 (50.0) 0 (0.0) 3 (50.0) Appears sad or depressed 58/77 28 (48.3) 9 (15.5) 21 (36.2) Expressing feelings of hopelessness or sadness about the future

45/77 12 (26.7) 14 (31.1) 19 (42.2)

Crying and tearfulness 39/77 18 (46.2) 3 (7.7) 18 (46.2) Commenting about dead of self or others 25/77 13 (52.0) 4 (16.0) 8 (32.0) Talking about feeling lonely 32/77 15 (46.9) 5 (15.6) 12 (37.5) Comments about feeling worthless or being a burden to others

32/77 7 (21.8) 11 (34.4) 14 (43.8)

Comments about feeling like a failure 17/77 4 (23.5) 3 (17.6) 10 (58.9) Appears anxious or worried 59/77 20 (34.0) 13 (22.0) 26 (44.0)

Effectiveness of rivastigmine during 6 months as compared with historical control cohort In table 5, the baseline characteristics and baseline test results of both the experimental group and the historical control cohort are summarised. The experimental and control group did not significantly differ in age. The control cohort, however, consisted of more women. MMSE and CAMCOG baseline scores were significantly lower in the control cohort (p<0.001). Baseline IDDD and RMBPC scores differed not significantly. For both the experimental and the control group, the differences between test results at 6 months and


91

baseline and effect sizes are also presented in table 5. All domains, except disruptive behaviour showed significant and clinical detectable differences between rivastigmine users and controls. Rivastigmine users showed stabilisation on MMSE and a slight improvement on the CAMCOG, whereas controls significantly deteriorated on both tests (p<0.001). Performance (IDDD) showed significantly less deterioration in rivastigmine users compared to the control group (p<0.001). Rivastigmine significantly improved memory-related behaviour (p<0.05) and non-significantly disruptive behaviour. Depression, however, significantly worsened during rivastigmine use compared with controls where this diminished (p<0.001). Table 5. Baseline characteristics and assessment results of rivastigmine users and historical controls, differences in assessment results between 6 months and baseline of rivastigmine users and historical controls and effect sizes of rivastigmine treatment after 6 months.

Baseline characteristics /Assessment scores (mean (SD))*

Differences in assessment scores between six months and baseline (mean (SD))a

Effect sizea Characteristics / Assessment scales

RIVAi

(n=84) Controls (n=69)

p-value

RIVAi (n=84)

Controls (n=69)

p-value

Cohen’s d

SRMb

Age 78.2 (6.0) 78.3 (6.2) 0.773 Gender, female 69% 81% MMSEc 20.4 (4.4) 17.5 (5.4) <0.001 0.1 (3.1) -1.6 (3.1) <0.001 0.35 0.55 CAMCOGd 68.8 (13.1) 61.0 (16.6) <0.001 0.5 (6.3) -4.5 (6.5) <0.001 0.33 0.78 IDDD-performancee

14.0 (8.8) 12.5 (8.5) 0.146 0.7 (5.9) 4.4 (7.7) <0.001 -0.43 -0.54

RMBPC memf 17.1 (3.9) 17.4 (5.8) 0.468 -0.4 (4.3) 1.0 (5.4) 0.006 -0.28 -0.28 RMBPC disg 4.3 (3.9) 4.2 (4.3) 0.786 -0.5 (2.6) -0.2 (3.9) 0.369 -0.07 -0.09 RMBPC deph 7.7 (5.7) 8.4 (6.1) 0.259 0.2 (4.6) -1.6 (5.0) 0.001 0.30 0.37 *Lower baseline MMSE/CAMCOG scores / higher baseline IDDD/RMBPC scores represent worse cognitive performance, functional performance or behaviour, aNegative differences/effect sizes reflect improvement on IDDD/RMBPC and deterioration on MMSE/CAMCOG, Standardized Response Mean, crange 0-30, drange 0-107, erange 0-44, f memory subscale: range 0-28, g disruptive behaviour subscale: range 0-32, h depression subscale: range 0-36, I Rivastigmine

If the follow-up results of the treated cohort are compared with those of the historical control cohort, it can be seen that decline as shown in 6 months in untreated controls is comparable with decline in treated patients in 18 months as measured on scales for cognition (MMSE) and activities of daily living (ADL) or 24 months as measured on scales for cognition (CAMCOG) and memory associated disturbances in behaviour.

Discussion In a cohort of Alzheimer patients, rivastigmine shows a significant but modest clinically detectable effect on cognition, performance in daily living and the memory part of the behaviour scale during 26 weeks compared with a historical control group of Alzheimer

Chapter 3.2.2

92

patients. Disruptive behaviour was not significantly altered and depressive behaviour worsened slightly. During 30 months, rivastigmine users showed stabilisation compared with baseline test results on numerous RMBPC items. Improvement was shown on a number of depression and disruptive behaviour subitems in particular. Baseline assessment results were not significantly different for the IDDD and all subsections of the RMBPC, and therefore rivastigmine effectiveness after 6 months can be easily interpreted. On the contrary, baseline assessment results differed significantly for both MMSE and CAMCOG. Although rivastigmine showed a significant and clinical detectable effect after 6 months on cognition, this should be interpreted with caution. MMSE scores can be subdivided into no cognitive impairment (24-30), mild cognitive impairment (18-23) and severe cognitive impairment (0-17)25. The control cohort may present a more advanced state of cognitive impairment and this might result in a different rate of cognitive decline compared with the experimental group of rivastigmine users. Therefore, the shown effectiveness can possibly not be attributed solely to rivastigmine use. A review by Grossberg described effects of rivastigmine on behaviour. A significant reduction in the frequency of aggressiveness as compared with placebo was shown after 6 months treatment26. In nursing home patients significant improvement was shown after 52 weeks for hallucinations, anxiety, euphoria, apathy, disinhibition, irritability, aberrant motor behaviour, nighttimes behaviour and appetite/eating change among patients with these symptoms at baseline11. In another study, in a population of nursing home patients 26 weeks of rivastigmine showed improvement of anxiety in 56% of patients showing disturbances at baseline, improvement of irritability in 66% of patients and also nighttime disturbances improved in 82% of patients13. Our cohort study showed after that 26 weeks of rivastigmine treatment, around 45% of patients with symptoms present at baseline improved for irritability and anxiety and in 62% of patients improvement of nighttime behaviour was shown. It must be pointed out also that in the studied cohort, only six patients did not receive daily doses above 6 mg. A meta-analysis showed a relation between efficacy and rivastigmine dose27. The shown effectiveness in our cohort may therefore be only representative for doses of rivastigmine above 6 mg daily. In addition, in the studied cohort, no relation was found between dose of rivastigmine and effectiveness in the three studied domains (results not shown). Our experimental cohort shows initial improvement of cognition as well as initial worsening of depressive symptoms. A possible relation could be hypothesised. Lopez et al. 28 did not find a difference in presence of depression between patients who were aware of their cognitive deficits and those who were not. Harwood et al.29 however, found a positive relation between greater insight and more depressive symptoms.


93

Dementia may put a burden on relatives and the health system30, and therefore, it is of primary importance if symptomatic therapy could delay deterioration in cognition, performance and behaviour. In this study, rivastigmine therapy shows a 12-18 months delay in deterioration of ADL, cognition and memory-associated behaviour compared with the historical control cohort. However, a positive bias is introduced. Responders to rivastigmine continue therapy after the first follow-up evaluation at 6 months, while others discontinue therapy, because of non-response. Therefore, only in initial responders therapy could postpone cognitive and functional deterioration with 1 - 1.5 years. In addition, because only 6-month data are available of the control cohort, a direct comparison cannot be made. However, a dropout study suggested an effect of rivastigmine on disease progression31. Moreover, a huge bias is introduced during the first period of 6 months of treatment. Rivastigmine shows more adverse events as compared to donepezil32. A large drop-out rate is the result, which is also apparent in our cohort. So only in initial survivors up to the first follow-up visit after 6 months, rivastigmine effectiveness is measured and in the sub cohort of responders, therapy is continued.

Our results show a gradual decline up to 30 months after treatment initiation. However, the long-term results of our cohort are purely descriptive in nature. Long-term effects of rivastigmine and other cholinesterase inhibitors remain largely unknown. Studies following patients beyond 12 months are scarce and descriptive in nature and placebo responses are sometimes modelled6,33. Lopez-Lousa et al.3 reported great variability in treatment response between rivastigmine users. Our results confirm this. A wide range of differences in scores between follow-up assessments and baseline scores was shown during the complete follow-up period. Rockwood and MacKnight34 proposed cholinesterase inhibitor users to be subdivided into a group of responders, a group of non-responders and a group remaining at equivocal results at retesting. In the future, more research should be performed to identify these subgroups and make it ideally possible to identify individual patients in routine clinical care as a responder or not. If it is possible to identify patients as responders or non-responders, therapy should be initiated only in patients who fulfil the characteristics for a responder. The non-responders will be spared then from the frequently occurring adverse events. Moreover, even a part of the identified responders will experience adverse events and some patients will possible discontinue therapy before the initial follow-up visit at 6 months. However, it is currently unclear which characteristics define the (long-term) rivastigmine responders, and therefore all patients should be offered therapy, as it is unethical to withdraw a possible effective therapy from patients. Strengths of our study are a relatively large population in a naturalistic setting, a total follow-up time of 30 months and full accessibility to all relevant clinical data. This study is

Chapter 3.2.2

94

an example of rivastigmine monitoring in clinical practice and could serve as a model for other memory clinics. The scales that are used are easy to apply and cover a wide range of domains; both memory and non-memory cognitive function, performance in daily living, memory, disruptive and depressive behaviour are assessed. Monitoring rivastigmine effectiveness in different domains is an advantage over only using the MMSE as instrument to monitor treatment effects. The use of CAMCOG, IDDD and RMBPC in clinical practice is not highly time-consuming, as CAMCOG screening takes about on average 30 minutes only, and in the meantime, proxies are able to fill in the IDDD and RMBPC questionnaires. It is important that the same proxy accompanies the patient every follow-up visit. An unacceptable low level of agreement was found between primary and secondary informants on the RMBPC. The RMBPC, however, is very useful for longitudinal follow-up as it showed an acceptable test-retest agreement and did not show a ceiling effect35. The CAMCOG evaluates a broad range of cognitive functions that are often affected in dementia and thus an advantage over brief screening tests16. Previous research showed this instruments’ utilities for assessing and monitoring cognitive decline in moderate and moderately severe dementia patients36. Ceiling effects are not supposed when using CAMCOG for cognitive evaluation in AD, as it showed little ceiling effect when used in the non-demented elderly37 and appeared to be sensitive to the early stages of dementia38. The IDDD, developed for community-dwelling dementia patients, has a high internal consistency and all the activities that are mentioned in this instrument are relevant and applicable to both men and women18. Our results show that the non-memory section of the CAMCOG, consisting of items regarding for example attention, praxis, abstract reasoning and perception, initially improved, whereas the memory section initially deteriorated in the rivastigmine group. As we performed a cohort study without a placebo group the results may be shaded in this descriptive open-label setting and future research should focus on investigating and confirming this further. Identification of responders is also of primary importance and should be involved in future research. In conclusion, a huge discontinuation rate is experienced within the first half year of treatment. In the subpopulation of patients who continued rivastigmine for 6 months, it shows modest, significant effectiveness on cognition, functionality and memory associated behaviour as compared with historical control patients. Unfortunately, disruptive behaviour is not significantly altered by rivastigmine therapy and depressive behaviour increased slightly after initial treatment. During 30 months, rivastigmine showed stabilisation on numerous behaviour items as measured by the RMBPC.


95


(rivastigmine tartrate), a new acetylcholinesterase inhibitor, in patients with mild to moderately severe Alzheimer’s disease. Int J Geriatr Psychopharmacol 1998; 1:55-65.

2. Rösler M, Anand R, Cicin-Sain A, et al. Efficacy and safety of rivastigmine in patients with Alzheimer’s disease: international randomized controlled trial. BMJ 1999;318:633-638.

3. Lopez-Pousa S, Turon-Estrada A, Garre-Olmo J, et al. Differential efficacy of treatment with acetylcholinesterase inhibitors with mild and moderate Alzheimer’s disease over a 6-month period. Dement Geriatr Cogn Disord 2005;19:189-195.

4. Mossello E, Tonon E, Caleri V, et al. Effectiveness and safety of cholinesterase inhibitors in elderly subjects with Alzheimer’s disease: a “real world” study. Arch Gerontol Geriatr 2004; 9:S297-S307.

5. Small GW, Kaufer D, Mendiondo MS, et al. Cognitive performance in Alzheimer’s disease patients receiving rivastigmine for up to 5 years. Int J Clin Pract 2005;59:473-77.

6. Johanssen P. Long-term cholinesterase inhibitor treatment of Alzheimer’s disease. CNS Drugs 2004; 18:757-768.

7. Wynn ZJ, Cummings JL. Cholinesterase inhibitor therapies and neuropsychiatric manifestations of Alzheimer’s disease. Dement Geriatr Cogn Disord 2004;17:100-108.

8. Lane RM, Potkin SG, Enz A. Targeting acetylcholinesterase and butyrylcholinesterase in dementia. Int J Neuropsychopharmacol 2006;9:101-124.

9. Rösler M, Retz W, Retz-Junginger P, Dennler HJ. Effects of two-year treatment with the cholinesterase inhibitor rivastigmine on behavioural symptoms in Alzheimer’s disease. Behav Neurol 1998;11:211-216.

10. Finkel SI. Effects of rivastigmine on behavioral and psychological symptoms of dementia in Alzheimer’s disease. Clin Ther 2004; 26:980-990.

11. Aupperle PM, Koumaras B, Chen M, et al. Long-term effects of rivastigmine treatment on neuropsychiatric and behavioural disturbances in nursing home residents with moderate to severe Alzheimer’s disease: results of a 52-week open-label study. Curr Med Res Opin 2004; 20:1605-1612.

12. Hatoum HT, Lin SJ, Arcona S, et al. The use of the occupational disruptiveness scale of the neuropsychiatric inventory-nursing home version to measure the impact of rivastigmine on the disruptive behaviour of nursing home residents with Alzheimer’s disease. J Am Med Dir Assoc 2005; 6:238-245.

13. Cummings JL, Koumaras B, Chen M, et al. Effects of rivastigmine treatment on the neuropsychiatric and behavioral disturbances of nursing home residents with moderate to severe probable Alzheimer’s disease: a 26-week, multicenter, open-label study. Am J Geriatr Pharmacother 2005;3:137-148.

14. McKhann G, Drachman D, Folstein M, et al. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA work group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s disease. Neurology 1984;34:939-944.

15. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state” a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-198.

16. Roth M, Tym E, Mountjoy CQ, et al. CAMDEX A standardized instrument for the diagnosis of mental disorder in the elderly with special reference to early detection of dementia. Br J Psychiatry 1986;149:698-709.

17. Schmand B, Walstra G, Lindeboom J, Teunisse S, Jonker C. Early detection of Alzheimer’s disease using the Cambridge Cognitive Examination (CAMCOG). Psychol Med 2000;30:619-627.

18. Teunisse S, Derix MM. Measuring functional disability in community-dwelling dementia patients: development of a questionnaire (in Dutch). Tijdschr Gerontol Geriatr 1991;22:53-59.

19. Teri L, Truax P, Logsdon R, Uomoto J, Zarit S, Vitaliano PP. Assessment of behavioral problems in dementia: the revised memory and behavior problems checklist. Psychol Aging 1992;7:622-631.

20. Verhage F. Intelligence and Age (in Dutch). Assen, van Gorcum, 1964.

Chapter 3.2.2

96

21. Public Health Board. Diagnostics in Dementia (in Dutch). Amstelveen, Public Health Board, 1998. 22. Cohen J. Statistical power analysis for the behavioral sciences. New York, Academic Press, 1969. 23. Rockwood K. Size of the treatment effect on cognition of cholinesterase inhibition in Alzheimer’s disease. J

Neurol Neurosurg Psychiatry 2004;75:677-685. 24. Frankfort SV, Appels BA, de Boer A, et al. Discontinuation of rivastigmine in routine clinical practice. Int J

Geriatr Psychiatry 2005;20:1167-71. 25. Tombaugh TN, McIntyre NJ. The mini-mental state examination: a comprehensive review. J Am Geriatr Soc

1992;40:922-935. 26. Grossberg GT. Effect of rivastigmine in the treatment of behavioral disturbances associated with dementia:

review of neuropsychiatric impairment in Alzheimer’s disease. Curr Med Res Opin 2005;10:1631-1639. 27. Ritchie CW, Ames D, Clayton T, Lai R. Metaanalysis of randomized trials of the efficacy and safety of

donepezil, galantamine, and rivastigmine for the treatment of Alzheimer’s disease. Am J Geriatr Psychiatry 2004;12:358-369.

28. Lopez OL, Becker JT, Somsak D, et al. Awareness of cognitive deficits and anosognosia in probable Alzheimer’s disease. Eur Neurol 1994;34:277-282.

29. Harwood DG, Sultzer DL, Wheatly DV. Impaired insight in Alzheimer’s disease: association with cognitive deficits, psychiatric symptoms, and behavioral disturbances. Neuropsychiatry Neuropsychol Behav Neurol 2000;13:83-88.

30. Trinh NH, Hoblyn J, Mohanty S, Yaffe K. Efficacy of cholinesterase inhibitors in the treatment of neuropsychiatric symptoms and functional impairment in Alzheimer’s disease: a meta analysis. JAMA 2003;289:210-216.

31. Farlow M, Potkin S, Koumaras B, et al. Analysis of outcome in retrieved dropout patients in a rivastigmine vs placebo, 26 week, Alzheimer disease trial. Arch Neurol 2003;60:843-848.

32. Birks J. Cholinesterase inhibitors for Alzheimer’s disease. Cochrane Database Syst Rev 2006;25:CD005593. 33. Grossberg G, Irwin P, Satlin A, et al. Rivastigmine in Alzheimer’s disease: efficacy over two years. Am J

Geriatr Psychiatry 2004;12:420-431. 34. Rockwood K, MacKnight C. Assessing the clinical importance of statistically significant improvement in

anti-dementia drug trials. Neuroepidemiology 2001;20:51-56. 35. Teunisse S, de Haan R, Walstra GJM, et al. Behavioural problems in mild dementia: clinical relevance and

methodological evaluation of the Revised Memory and Behavioural Problems Checklist. In: Teunisse S. Clinimetrics in Dementia [Thesis]. Amsterdam: University of Amsterdam, 1997.

36. Garre-Olmo J, López-Pousa S, Vilalta-Franch J, Turon-Estrada A, Lozano-Gallego, Hernández-Ferràndiz M, et al. Neuropsychological profile of Alzheimer’s disease in women: moderate and moderately severe cognitive decline. Arch Womens Ment Health 2004;7:27-36

37. Huppert FA, Brayne C, Gill C, Paykel ES, Beardsall L. CAMCOG- A concise neuropsychological test to assist dementia diagnosis: Socio-demographic determinants in an elderly population sample. Br J Clin Psychol 1995;34:529-41.

38. Williams JG, Huppert FA, Matthews FE, Nickson J. Performance and normative values of a concise neuropsychological test (CAMCOG) in an elderly population sample. Int J Geriatr Psychiatry 2003;18:631-44

97

98

99

CHAPTER 3.2.3 Identification of responders and reactive domains to

rivastigmine in Alzheimer’s disease

S.V. Frankfort, B.A. Appels, A. de Boer, C.R.Tulner, J.P.C.M. van Campen, C.H.W. Koks, J.H. Beijnen, B.A. Schmand

Abstract Purpose Presently, it is unclear which patients suffering from Alzheimer’s Disease (AD) respond to rivastigmine and if rivastigmine acts on specific cognitive domains. The aims of this study are thus to investigate treatment effects of rivastigmine on specific cognitive domains and to find possible responsive subpopulations to rivastigmine cognitive effects. Methods Mini Mental State Examination (MMSE) and Cambridge Cognitive Examination (CAMCOG) were administered at baseline and after 6 months in 83 rivastigmine users and 96 historical controls, representing natural decline. Treatment effects on different subsections of the CAMCOG and in different subpopulations were investigated by linear regression analyses. Results Rivastigmine showed effectiveness on total CAMCOG (p<0.001), CAMCOG non-memory subsection (p<0.001) and subscales of language (p=0.002), attention/calculation (p=0.043), abstract thinking (p<0.001) and perception (p=0.031). In patients with baseline MMSE ≤19 rivastigmine showed significant and favourable effects compared to historical controls on total CAMCOG (p<0.001) and both non-memory (p<0.001) and memory subsections (p=0.002). Conclusion Rivastigmine showed primarily effectiveness on the non-memory section of the CAMCOG and patients with a baseline MMSE ≤19 appeared to show greater responses to rivastigmine compared to patients with baseline MMSE ≥20.

Pharmacoepidemiology and Drug Safety (in press)

Chapter 3.2.3

100

Introduction Rivastigmine, an acetylcholinesterase inhibitor, showed efficacy in the symptomatic treatment of mild-to-moderately severe Alzheimer’s disease (AD) in randomised placebo-controlled trials1,2. However, patients respond very differently to therapy, ranging from continuation of deterioration or maintaining baseline levels to a clear clinical effect3. Previous research showed improvement in APOE ε4 carriers treated with rivastigmine4,

more benefit of therapy in patients with a rapid rate of disease progression5,6 and an association between the occurrence of hallucinations and response to therapy, as defined by an increase of two or more points on the MMSE7. Starting therapy only in those patients in whom effect is expected would be ideal in clinical practice considered the fact that rivastigmine users often experience adverse events8.

It is presently unclear which characteristics identify rivastigmine responders. It is also uncertain if rivastigmine exerts treatment effects on certain cognitive subdomains in AD as, to our knowledge, no studies regarding this issue have been published. However, attention is one of the cognitive subdomains that indeed responded to rivastigmine therapy in a study performed in Lewy Body Disease (LBD) patients9. Therefore, attention might also specifically respond to therapy in AD patients. The aims of this study are to investigate treatment effects of rivastigmine on specific cognitive domains and to find possible responsive subpopulations to rivastigmine cognitive effects.

Methods Patients This prospective study was carried out in patients with mild-to-moderate, probable or possible AD according to the NINCDS-ADRDA criteria10. The index group consisted of patients using rivastigmine (Exelon®) via the geriatric outpatient department of a Dutch hospital. Only patients who had relatives or friends who could monitor drug intake and patients in whom therapy was evaluated after six months were included. The historical control cohort consisted of Alzheimer patients who did not take rivastigmine and were followed during a period of six months as part of a research project regarding the utility of diagnostics procedures in a memory clinic in The Netherlands11. This historical control group represents natural decline in Alzheimer’s disease. Patients were excluded if cognitive test results were incomplete. Education was scored on a seven-point scale, ranging from less than six years of elementary school (1) to a university degree (7)12.

The review board of the Slotervaart Hospital, Amsterdam, The Netherlands, approved the use of routine anonymous neuropsychological assessment data for research purposes.

Reactive domains and responders to rivastigmine

101

Assessment of domains At baseline, i.e. when starting rivastigmine in case of the index group, and after six months patients were evaluated by cognitive assessment including Mini Mental State Examination (MMSE; max score 30)13 and Cambridge Cognitive Examination (CAMCOG)14. CAMCOG consists of 60 items and total sum scores range from 0-107. CAMCOG can be subdivided into a memory (maximum score 37) and a non-memory (maximum score 70) section15. The memory subsection can be further subdivided into subscales regarding recent memory (max score: four), remote memory (max score: six), learning (max score: 17) and orientation (max score: 10). The non-memory subsection can be subdivided into subscales assessing language (max score: 30), praxis (max score: 12), attention/calculation (max score: nine), abstract thinking (max score: eight) and perception (max score: 11). One item in the CAMCOG perception subscale (asking if the respondent recognised two people in the room) was omitted and always scored as one point. Lower scores on MMSE and CAMCOG reflect more severe disease. Statistical analysis A) We performed linear regression analyses to investigate rivastigmine effectiveness on cognition compared to the historical control cohort during 6 months as measured by MMSE, CAMCOG and subsections and subscales of the CAMCOG. The dependent variable was the test result after 6 months and the independent variable was rivastigmine use. B) We investigated whether effect modification plays a role by introducing interaction terms in linear regression analyses regarding CAMCOGtotal, non-memory and memory subsections and the attention subscale. Therefore, we sequentially performed multivariate linear regression analyses including the interaction term rivastigmine by dichotomised disease severity, in addition to both terms separately. As an indicator for disease severity, the baseline MMSE score was dichotomised to the median of the whole population, resulting in MMSE ≤19 versus ≥ 20. If the interaction terms showed significance, we performed additional linear regression analyses to investigate rivastigmine effects in subgroups. We corrected analyses for age, gender and level of education, as these variables may influence cognitive performance, and for baseline test results. All statistical calculations were performed with SPSS for Windows (version 11.0, SPSS Inc. ,Chicago, IL, USA).

Chapter 3.2.3

102

Results Patient characteristics In total, 179 patients were included in the study of whom 83 used rivastigmine during 6 months and 96 were historical control patients with complete cognitive screening results. 24 patients were excluded from the historical control cohort and one patient was excluded from the index group because cognitive test results were incomplete. Table 1 shows baseline characteristics and baseline test results of included patients. Median baseline MMSE score of all included patients was 20 (IQR: 7.00, range 1-28), median age was 79.0 (IQR: 8.00, range 56-89), almost 63% were women and the median education level was 3 (IQR: 3.00, range 1-7). Table 1. Baseline characteristics, Baseline Characteristics/ Test results Rivastigmine users (n=83)* Historical controls (n=96)*

Age (years) 78.1± 6.0 (56-89) 77.8± 5.9 (65-89) Education, median, IQRa (range) 4.0, IQR 3.0 (1-6) 2.0, IQR 2.0 (1-7) Female gender, n (%) 57 (68.7) 55 (57.3) MMSE 20.5± 4.3 (8-28) 18.4± 5.2 (1-28) CAMCOG total 69.0± 13.0 (33-92) 62.8± 16.0 (12-94) CAMCOG memory 17.5± 5.5 (5-29) 16.6± 7.6 (0-32) Orientation 6.6 ± 2.1 (2-10) 6.0± 2.5 (0-10) Recent memory 1.8± 1.1 (0-4) 1.6±1.3 (0-4) Remote memory 3.0± 1.7 (0-6) 2.7± 1.9 (0-6) Learning 6.1± 2.6 (0-11) 6.2± 3.5 (0-14) CAMCOG non-memory 51.6± 9.3 (25-67) 46.3± 10.1 (12-64) Language 23.1± 3.5 (12-29) 21.4± 4.1 (7-28) Calculation/Attention 6.7± 2.4 (1-9) 5.7± 2.2 (1-9) Praxis 9.4± 2.0 (4-12) 7.9± 2.5 (1-12) Abstract thinking 5.1± 2.3 (0-8) 3.8± 2.3 (0-8) Perception 7.3± 2.4 (1-11) 7.5± 1.8 (3-10) a IQR: Inter Quartile Range, * Expressed as Mean ±SD (range), unless otherwise stated Rivastigmine effectiveness during 6 months Table 2 presents the results of the linear regression analyses investigating if rivastigmine use exerts significant effects as measured by MMSE, CAMCOG total score and subscales assessing specific cognitive domains and adjusted for baseline cognitive test results, age, gender and level of education. Rivastigmine use showed a significant effect on MMSE (Mean difference compared to controls (MD) = 1.7), CAMCOG total score (MD=4.4) and the non-memory section of this instrument (MD=3.5). On the contrary, no significant


103

effects on the scores of the memory section were shown. By investigating subscales of the non-memory subsection, rivastigmine did show significant effects on language (MD=1.5), attention/calculation (MD=0.6), abstract thinking (MD=1.3) and perception (MD= 0.5). Praxis, orientation and subscales of remote, recent and learning memory did not show significant effects of rivastigmine use. Table 2. Treatment effects of rivastigmine as measured on MMSE, CAMCOG total score, subsections and subscales assessing specific cognitive domains and expressed as corrected mean differences from linear regression analyses. Linear regression analysesa

(Sub) Test Mean difference 95% CI* P-value

MMSE 1.7 0.7-2.6 0.001 CAMCOG total 4.4 2.3-6.4 <0.001 CAMCOGnonmemory 3.5 1.9-5.2 <0.001 Language 1.5 0.6-2.4 0.002 Attention/calculation 0.6 0.0-1.1 0.043 Praxis 0.3 -0.3-0.8 0.333 Abstract thinking 1.3 0.8-1.9 <0.001 Perception 0.5 0.0-1.0 0.031 CAMCOGmemory 0.8 -0.2-1.8 0.108 Memory: recent 0.1 -0.2-0.4 0.565 Memory: remote 0.1 -0.2-0.5 0.568 Memory: learning 0.5 -0.2-1.1 0.170 Orientation 0.4 -0.2-1.0 0.152 a corrected for age, gender, education level and baseline scores; * CI= Confidence Interval

Predictors of rivastigmine effectiveness Only the interaction terms that showed significance in different analyses are presented in this section. The interaction term of rivastigmine by disease severity, indicating differential responses to rivastigmine in patients with baseline MMSE ≤19 compared to MMSE≥20 showed significance on CAMCOGtotal (p=<0.001), CAMCOG-nonmemory (p=0.001), attention/calculation (p=0.003) and CAMCOGmem (p=0.004). Differential effectiveness in subpopulations Table 3 shows treatment effects in subpopulations as defined by significant interaction terms as described above. It is clear from the table that the described subpopulations showed differential rivastigmine effectiveness, concerning both statistically significant effects as well as size of the treatment effect expressed as the mean difference (MD) of test scores in the index group compared to the reference group. Disease severity (MMSE≤19) is

Chapter 3.2.3

104

an important identifying variable as in the studied population rivastigmine showed significant improvement compared to historical controls on total CAMCOG (MD= 8.2), non-memory subsection (MD=6.3), attention/calculation (MD=1.2) and the memory subsection (MD=2.2). The described results mean that patients with more severe AD respond better to rivastigmine. Table 3. Differential treatment effects of rivastigmine in subpopulations expressed as mean differences from linear regression analysis corrected for age, gender, education level and baseline scores.

Linear regression analyses a Test Subpopulation Mean Difference 95% CI* p-value

CAMCOG total MMSE≤19 8.2 5.0-11.3 <0.001 MMSE ≥20 0.7 -1.9-3.3 0.582 CAMCOGmemory MMSE≤19 2.2 0.8-3.5 0.002 MMSE ≥20 -0.5 -2.0-1.0 0.534 CAMCOGnonmem MMSE≤19 6.3 3.7-9.0 <0.001 MMSE ≥20 0.6 -1.4-2.6 0.531 Attention/calculation MMSE≤19 1.2 0.4-2.0 0.05 MMSE ≥20 0.2 -0.5-0.9 0.596 a corrected for age, gender, education level and baseline scores; * CI = Confidence Interval

Discussion Rivastigmine shows favourable cognitive effects as measured by MMSE, total CAMCOG and the total score on the non-memory subsection of this instrument. The non-memory subscales of language, attention/calculation, abstract thinking and perception also showed improvement as compared to a historical control cohort of Alzheimer patients. However, total scores on the memory section and all subscales of memory, including orientation, as well as the non-memory subscale of praxis did not show significant favourable rivastigmine effects as compared to historical control patients. Baseline MMSE score can identify a subpopulation of responders to rivastigmine. Patients with a baseline MMSE score ≤19 show significant and favourable effects as compared to historical controls on MMSE, total CAMCOG, memory and non-memory subsection scores and the cognitive subscale of attention/calculation. Patients with a baseline MMSE ≥20, however, do not show significant effectiveness on these domains in this clinically based study. A review16 criticised the methodology of cholinesterase inhibitor trials performed during the last years and questioned whether small clinical effects and insufficient methodology are enough scientific basis for prescribing these drugs in Alzheimer’s disease. However, in our study, we did show clinical effects, as measured by CAMCOG, and these effects were


105

more pronounced in different subpopulations. For example, in the subpopulation of more severe disease our results show a mean difference compared to control patients of 8 points on total CAMCOG scores, which is regarded a large difference in clinical practice. Our study also confirms the hypothesis of differential efficacy in cholinesterase inhibitor users3.

Our results show cognitive subdomains to respond differently to rivastigmine. Attention responds significantly and favourably and may possibly play a mediating role in effectiveness in other cognitive domains. Acetylcholine plays an important role in attentional processing and cholinergic dysfunction interferes probably indirectly with cognitive functioning via attention17. A study by Sahakian and Coull18 showed that tacrine improved attentional functions in patients with AD. In addition, rivastigmine studies performed in LBD patients, where the cholinergic deficit is probably even larger than in AD19, showed effectiveness on attention9. To our knowledge, this is the first study reporting rivastigmine differential effects in Alzheimer’s disease. However, in a trial with tacrine, a first generation cholinesterase inhibitor, patients responded significantly better compared to those receiving placebo on Alzheimer’s disease Assessment Scale-Cognition section (ADAS-Cog) items assessing recall, naming, language and word finding20,21. The differential effectiveness on cognitive domains, as our investigations showed, should be most ideally confirmed by placebo-controlled trials and/or by research using tests that are more specific. Previous research investigated rivastigmine use in more advanced Alzheimer’s disease22,23. Burns et al.23 suggested subjects with more severe disease might also benefit from rivastigmine and Kurz et al22 showed cognitive benefits to be more marked in moderate and moderately severe cohorts than the mild AD cohort. For the cholinesterase inhibitor galantamine, it was observed that patients with MMSE scores lower than 18 had a more robust response to therapy24 and similarly for another cholinesterase inhibitor, donepezil, a more robust response was noticed in patients with MMSE scores of 20 or below25. Recently, the National Institute for Health and Clinical Excellence, UK, (NICE) has issued a draft of a revised guideline that recommends donepezil, galantamine or rivastigmine to be considered only in the treatment of persons with AD of moderate severity only (MMSE 10-20)26. It is clear that disease severity may serve as a variable in identifying responsive subpopulations. A possible explanation could be a larger cholinergic deficit in more severe stages of the disease27 suggesting this subpopulation to be more responsive to cholinergic enhancement. As attention may play a key role in cognitive functioning17 and differential attentional deficits were shown in different stages of AD28, this may also attribute to a possible explanation why patients with MMSE ≤19 respond significantly better to rivastigmine therapy. Perry et al.28 investigated sustained, divided and selective attention in different severity stadia of AD. In early AD (MMSE 24-30), only selective attention showed deficits, while patients with moderately severe AD (MMSE 18-23) showed

Chapter 3.2.3

106

impairment in sustained, divided and selective attention. Finally, the rate of disease progression, as measured on scales, does not show linearity during the course of AD. A more marked change occurred in the moderate range29-31. Farlow et al.5 showed a more rapid rate of disease progression to be a predictor of rivastigmine effectiveness. This could possibly attribute to our results showing more effectiveness in more severe AD. CAMCOG evaluates a broad range of cognitive functions that are often affected in dementia.14 Although previous research showed this instruments’ utilities for assessing and monitoring cognitive decline in moderate and moderately severe patients32 and reliability of individual subscales has been reported as acceptable33, one may question the use of the CAMCOG as a monitoring instrument. Lindeboom et al.34 showed the CAMCOG to be a non-linear scale. Thus, observed score differences may possibly not represent equal differences in changes in cognition across the test’s range. In addition, it is important to keep in mind possible floor and ceiling effects of cognitive tests, which could influence the less marked changes in early and late disease. Ceiling effects, however, are not supposed when using CAMCOG for cognitive evaluation in AD, as it showed little ceiling effect when used in the non-demented elderly35 and appeared to be sensitive to the early stages of dementia36. Possible floor effects of CAMCOG screening are in our investigations of minor importance as we included few patients with low baseline CAMCOG scores. However, these results should be interpreted with caution. We did not conduct a randomised clinical trial. Our study was designed in clinical practice and included patients treated with rivastigmine and historical control patients who were untreated. Bias may have been introduced by this study design. The historical control cohort, for example, was evaluated years before the index group. As dementia care has changed over the years, this may be a confounder in our analysis. Therefore, our results should be confirmed preferably by prospective randomised controlled trials. The Investigation in the Delay to Diagnosis of AD with Exelon (InDDEx) study is currently conducted and will provide us with knowledge regarding efficacy of rivastigmine in Mild Cognitive Impairment37.

Conclusion Rivastigmine showed differential effectiveness on cognitive subdomains that is more pronounced on the non-memory subsection. In addition, disease severity can be used to identify responsive subpopulations. More severe Alzheimer’s disease, as scored by a baseline MMSE ≤19, appeared to be an important subpopulation showing more effectiveness to rivastigmine compared to patients with baseline MMSE≥20.


107

Acknowledgements The authors would like to thank Gerard Walstra, neurologist, and Saskia Teunisse, neuropsychologist, for their permission to use the data of the historical control cohort in the described study.


(rivastigmine tartrate), a new acetylcholinesterase inhibitor, in patients with mild to moderately severe Alzheimer’s disease. Int J Geriatr Psychopharmacol 1998;1:55-65.

2. Rösler M, Anand R, Cicin-Sain A, et al. Efficacy and safety of rivastigmine in patients with Alzheimer’s disease: international randomised controlled trial. BMJ 1999;318:633-38.


4. Farlow MR, Lane R, Kudaravalli S, He Y. Differential qualitative responses to rivastigmine in APOE ε4 carriers and non-carriers. Pharmacogenomics J 2004;4:332-5.

5. Farlow MR, Hake A, Messina J, et al. Response of patients with Alzheimer’s disease to rivastigmine treatment is predicted by the rate of disease progression. Arch Neurol 2001;58:417-22.

6. Farlow MR, Small GW, Quarg P, Krause A. Efficacy of rivastigmine in Alzheimer’s disease patients with rapid disease progression: results of a meta-analysis. Dement Geriatr Cogn Disord 2005;20:192-97.

7. Pakrasi S, Mukaetova-Ladinska EB, McKeith IG, O’Brien JT. Clinical predictors of response to acetyl cholinesterase inhibitors: experience from routine clinical use in Newcastle. Int J Geriatr Psychiatry 2003; 18:879-86.

8. Gauthier S. Cholinergic adverse events of cholinesterase inhibitors in Alzheimer’s disease. Epidemiology and management. Drugs Aging 2001;18:853-62.

9. McKeith I, Del Ser T, Spano P, et al. Efficacy of rivastigmine in dementia with Lewy Bodies: a randomized, double-blind, placebo-controlled international study. Lancet 2000;356:2031-6.


11. Walstra GJ, Teunisse S, van Gool, van Crevel H. Reversible dementia in elderly patients referred to a memory clinic. J Neurol 1997;244:17-22.

12. Verhage F. Intelligence and Age. van Gorcum: Assen. (in Dutch) 1964. 13. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A practical method for grading the cognitive

state of patients for the clinician. J Psychiatr Res 1975;12:189-198. 14. Roth M, Tym E, Mountjoy CQ, et al. CAMDEX. A standardised instrument for the diagnosis of mental

disorder in the elderly with special reference to early detection of dementia. Br J Psychiatry 1986;149:698-709.

15. Schmand B, Walstra G, Lindeboom J, et al. Early detection of Alzheimer's disease using the Cambridge Cognitive Examination (CAMCOG). Psychol Med. 2000;30:619-27.

16. Kaduszkiewicz H, Zimmermann T, Beck-Bornholdt HP, van den Bussche, H. Cholinesterase inhibitors for patients with Alzheimer’s disease: systematic review of randomised clinical trials. BMJ 2005;331:321-3.

17. Francis PT, Palmer AM, Snape M, Wilcock G. The cholinergic hypothesis of Alzheimer’s disease: a review of progress. J Neurol Neurosurg Psychiatry 1999;66:137-47.

18. Sahakian BJ, Coull JT. Nicotine and tetrahydroaminoacradine: evidence for improved attention in patients with dementia of the Alzheimer type. Drug Dev Res 1994;31:80-8.

Chapter 3.2.3

108

19. Tiraboschi P, Hansen LA, Alford M, et al. Cholinergic dysfunction in diseases with Lewy bodies. Neurology 2000;54:407-11.

20. Farlow M, Gracon SI, Hershey LA, et al. A controlled trial of tacrine in Alzheimer’s disease. JAMA 1992;268:2523-29.

21. Cummings JL. Use of cholinesterase inhibitors in clinical practice: evidence-based recommendations. Am J Geriatr Psychiatry 2003;11:131-145.

22. Kurz A, Farlow M, Quarg P, Spiegel R. Disease stage in Alzheimer’s disease and treatment effects of rivastigmine. Alzheimer Dis Assoc Disord 2004;18:123-8.

23. Burns A, Spiegel R, Quarg P. Efficacy of rivastigmine in subjects with moderately severe Alzheimer’s disease. Int J Geriatr Psychiatry 2004;19:243-9.

24. Wilcock GK, Lilienfeld S, Gaens E. Efficacy and safety of galantamine in patients with mild to moderate Alzheimer’s disease: multicentre randomised controlled trial. BMJ 2000;321:1445-9.

25. McLendon B, Murali Doraiswamy P. Defining meaningful change in Alzheimer’s disease trials: the donepezil experience. J Geriatr Psychiatry Neurol 1999;12:39-48.

26. NICE 2006/001: NICE consults on revised first draft guidance on the use of drugs to treat Alzheimer’s disease. http://www.nice.org.uk/pdf2006_001_Alz_Press_release_AD_App_ Jan06.pdf. [Accessed March 28, 2006].


28. Perry RJ, Watson P, Hodges JR. The nature and staging of attention dysfunction in early (minimal and mild) Alzheimer’s disease: relationship to episodic and semantic memory impairment. Neuropsychologica 2000;38:252-71.

29. Morris JC, Edland S, Clarck C, et al. The consortium to establish a registry for Alzheimer’s disease (CERAD). Part IV: Rates of cognitive change in the longitudinal assessment of probable Alzheimer’s disease. Neurology 1993;43:2457-65.

30. Stern Y, Liu X, Albert M, et al. Application of a growth curve approach to modelling the progression of Alzheimer’s disease. Journal of Gerontology 1996;51A:M179-184.

31. Mendiondo MS, Wesson Ashford J, Kryscio RJ, Schmitt FA. Modelling Mini-Mental State Examination changes in Alzheimer’s disease. Statist Med 2000;19:1607-16.

32. Garre-Olmo J, López-Pousa S, Vilalta-Franch J, et al. Neuropsychological profile of Alzheimer’s disease in women: moderate and moderately severe cognitive decline. Arch Womens Ment Health 2004;7:27-36

33. Cullum S, Huppert FA, McGee M, et al. Decline across different domains of cognitive function in normal ageing: results of a longitudinal population-based study using CAMCOG. Int J Geriatr Psychiatry 2000;15:853-62.

34. Lindeboom R, Schmand B, Holman R, et al. Improved brief assessment of cognition in aging and dementia. Neurology 2004;63:543-6.

35. Huppert FA, Brayne C, Gill C, et al. CAMCOG- A concise neuropsychological test to assist dementia diagnosis: Socio-demographic determinants in an elderly population sample. Br J Clin Psychol 1995;34:529-41.

36. Williams JG, Huppert FA, Matthews FE, Nickson J. Performance and normative values of a concise neuropsychological test (CAMCOG) in an elderly population sample. Int J Geriatr Psychiatry 2003;18:631-44.

37. Farlow MR, He Y, Tekin S, et al. Impact of APOE in mild cognitive impairment. Neurology 2004;63:1898-1901.

109

110

111

CHAPTER 3.2.4 Population pharmacokinetics and pharmacodynamics of

rivastigmine and its metabolite NAP226-90 in patients with dementia

S.V. Frankfort, A.D.R. Huitema, C.R. Tulner, J.P.C.M. van Campen, J.H. Beijnen

Abstract

Objectives: To develop and evaluate a population pharmacokinetic model that describes the pharmacokinetics of rivastigmine and its metabolite NAP226-90 in elderly outpatients with dementia and to investigate the influence of demographic variables that are involved in the variability of pharmacokinetic parameters. In addition, this study also aimed to investigate the relation between pharmacokinetics and treatment efficacy. Methods: Alzheimer’s disease or Lewy Body disease outpatients using rivastigmine were included in this prospective study. Blood samples were drawn during the 6-monthly therapy evaluation visits. In a subcohort of the patients, blood samples were collected to accomplish a complete pharmacokinetic curve. Using non-linear mixed effect modelling (NONMEM), a population pharmacokinetic model was developed that described the pharmacokinetics of rivastigmine and metabolite. A pharmacokinetic-pharmacodynamic analysis was performed to investigate whether relationships exist between pharmacokinetics and therapy response. Results: From 33 patients (mean age 78.4 years), 163 plasma concentrations of both rivastigmine and NAP226-90 were available. The pharmacokinetics of both rivastigmine and metabolite were described by one-compartment models with Michaelis-Menten elimination for rivastigmine and first-order elimination for the metabolite. Volume of distribution (V1/F) was 177 L for rivastigmine and its elimination was characterised by a maximal elimination rate (Vmax/F) of 1.01 mg/h and a Michaelis Menten constant (Km) of 0.00698 mg/L. Clearance (CL/F/fm) and volume of distribution (V2/F/fm) were 126 L/h and 39.4 L, respectively, for NAP226-90, where fm represents the fraction of rivastigmine that is metabolised into NAP226-90. Interindividual variability was estimated in V1/F (70.6%), Vmax/F (14.1%), Km (65.8%), V2/F/fm (122%) and in relative bioavailability of rivastigmine (F;17.3%). Females had a 25% higher Vmax/F compared to males (p<0.01). No differences in exposure to rivastigmine or NAP226-90 were observed between responders and non-responders. Conclusion: A model was developed and evaluated that described the population pharmacokinetics of rivastigmine and NAP226-90 in geriatric outpatients with dementia. Gender was the only factor involved in the variability of pharmacokinetic parameters. No correlation could be demonstrated between the exposure to rivastigmine or NAP226-90 and effectiveness on cognition, activities in daily living or behaviour.

Submitted

Chapter 3.2.4

112

Introduction Rivastigmine (Exelon®), an acetylcholinesterase inhibitor, has shown efficacy in the symptomatic treatment of mild-to-moderate severe Alzheimer’s Disease (AD) in randomised placebo-controlled trials1,2. In Lewy Body Disease (LBD) that is clinically characterised by fluctuating cognitive impairment, attention deficits, visual hallucinations and parkinsonism, rivastigmine showed clinically improvement in behavioural symptoms3 and in memory and attention4. In LBD patients, the cholinergic deficit is probably even larger than in AD5 and this may explain why LBD patients respond more favourable to rivastigmine therapy. In addition, in the last years it is recognised that cholinergic deficits are also involved in the behavioural symptoms present in AD6 and therapy efficacy should ideally be investigated in different domains. Rivastigmine interacts with acetylcholinesterase resulting in a carbamylated complex that slowly dissociates to release active enzyme. This temporary inhibition of acetylcholinesterase (AChE), leads to an increased availability of acetylcholine in cholinergic neurons of the brain7. Increased acetylcholine availability ameliorates cognitive and behavioural disturbances8. Rivastigmine is rapidly absorbed and earlier pharmacokinetic studies revealed a bioavailability of 40% and a short elimination half-life of 2 hours. Rivastigmine is not significantly metabolised by hepatic CYP isoenzymes, but is converted by the target enzyme acetylcholinesterase into the decarbamylated metabolite NAP226-90. Thereafter, NAP226-90 may undergo N-demethylation and/or sulphate conjugation7. Although NAP226-90 itself is not pharmacodynamically active9, its concentrations correlated well with enzyme inhibition in an ex-vivo experiment in rat brain samples and plasma10. The concentration of NAP226-90 may thus better reflect the extent of enzyme inhibition than parent rivastigmine. AD patients show a highly variable response to therapy, ranging from continuation of deterioration, stabilisation of symptoms to a clear clinical beneficial effect11. It is presently unclear which characteristics identify rivastigmine responders and if therapy response is influenced by pharmacokinetic parameters. A study by Gobburu et al.12 did not find significant correlations between cognitive neuropsychological test battery scores and rivastigmine or metabolite concentrations in patients treated for 11 weeks with 1 to 6 mg bid. For donepezil, another cholinesterase inhibitor, a significant correlation was found between plasma drug concentrations and changes in the Alzheimer’s Disease Assessment Scale-Cognition subscale (ADAS-COG), Mini Mental State Examination (MMSE) and the quality of life, as reported by the patient13. In order to establish the pharmacokinetics of rivastigmine and NAP226-90 and to identify if certain patient characteristics influence the pharmacokinetic profile, a study was conducted in patients treated with rivastigmine in routine clinical practice. Furthermore, the developed

Population pharmacokinetics and pharmacodynamics of rivastigmine

113

pharmacokinetic model was used to calculate individual pharmacokinetic parameters. These pharmacokinetic parameters were linked to efficacy measures in different domains (cognition, performance of daily living activities and behaviour) in a pharmacokinetic-pharmacodynamic analysis in patients treated for a minimum of 6 months. Methods Patients This prospective study was performed at the geriatric department of the Slotervaart Hospital, Amsterdam, The Netherlands. Geriatricians prescribed rivastigmine for patients suffering from AD, diagnosed according to the NINCDS-ADRDA criteria14, or LBD as diagnosed according to the consensus criteria15. Patients started rivastigmine at 1.5 mg twice daily and doses were titrated up after intervals of minimal 2 weeks at each dose level, until the individual Maximum Achieved Dose (MAD) with a maximum of 6 mg twice daily. However, as it was shown that doses below twice daily 3 mg are associated with lower efficacy in retaining cognition, performance and behaviour1,2, doses should be equal or higher than twice daily 3 mg. In the clinical setting rivastigmine was discontinued within the first 6 months of treatment if titration to doses equal or higher than 3 mg bid failed because of intolerance. Rivastigmine effects on cognition, activities of daily living and behaviour was evaluated after 6 months of therapy, when patients were on their individual MAD, and subsequently at 6-months intervals. Patients were recruited at those 6-monthly follow-up visits at the hospital. Only patients who had relatives or friends who could monitor drug intake were included. The study protocol was approved by the Institutional Review Board of the Slotervaart Hospital, Amsterdam, The Netherlands. Written informed consent was obtained from each participant in this study. Sampling and bioanalysis Patients were admitted to the hospital for the assessment of a pharmacokinetic curve of rivastigmine and NAP226-90 in plasma during a single dose interval. During maximal 8 hours, 11 EDTA blood samples were drawn. In addition to the complete pharmacokinetic curves, blood samples at random time points were drawn at the 6-monthly follow-up visits to the clinic for the determination of the plasma concentrations of rivastigmine and NAP226-90. The time of dose and sampling were recorded. After isolation of plasma by centrifugation, the samples were stored at –20 °C until analysis. The concentrations of rivastigmine and NAP226-90 were simultaneously quantified by a validated method, using high-performance liquid chromatography coupled with tandem mass spectrometry16. Briefly, sample pre-treatment consisted of protein precipitation with

Chapter 3.2.4

114

methanol using 100 µl plasma. Chromatographic separation was performed on a 150x2.0 mm ID Gemini C18 column using a gradient system with 10 mM ammoniumhydroxide and methanol. Accuracies were within ± 14.3% for rivastigmine and NAP226-90. The inter-assay precision was less than 14.7% for both compounds. The intra-assay precision was less than 15.4% for rivastigmine and less than 14.0% for NAP226-90 at the lower limit of quantification level. The intra-assay precision of quality control samples at other concentration levels was less than 14.9% for both compounds. The validated concentration ranges were 0.00025 to 0.050 mg/L for rivastigmine and 0.00050 to 0.025 mg/L for NAP226-90. Pharmacokinetic analysis The nonlinear mixed effect modelling program (NONMEM) Version V (double precision, level 1.1, GloboMax LLC, Hanover MD, USA)17 was used to perform the analysis. The first-order estimation procedure (FO) was used throughout in combination with logarithmic transformation of the data. The adequacy of the tested models was evaluated using statistical and graphical methods. The minimal value of the objective function (OFV, equal to minus twice the log likelihood) provided by NONMEM was used as a goodness of fit characteristic to discriminate between hierarchical models using the log likelihood ratio test17. Standard errors for all parameters were calculated with the COVARIANCE option in NONMEM, individual Bayesian pharmacokinetic parameters were obtained using the POSTHOC option17. The program PDx-POP (version 1.1, release 4, Globomax LLC, Hanover MD, USA) was used as a tool for expediting population pharmacokinetic analysis with NONMEM and for graphical model diagnostics. Basic pharmacokinetic model For rivastigmine absorption, both first-order absorption models with and without absorption lag-time were tested. Also a chain of transition compartments was tested to describe the absorption process, as earlier investigated for the antiretroviral drug efavirenz18. To describe the distribution kinetics of rivastigmine and NAP 226-90, single and multiple compartment models with linear and non-linear elimination were investigated. Interindividual variability in the pharmacokinetic parameters and in the relative bioavailability were estimated from an exponential model19. For instance, variability in clearance was determined from the equation Cl/Fi= θ1*exp(ηi) in which CL/Fi represents the clearance of the ith individual, θ1 is the typical value of apparent clearance, ηi is the interindividual random effect with a mean of 0 and variance ω2. Residual variability was modelled with a combined additive and proportional error model.20 Larger residual variability was likely to be associated with samples collected closer to the time of intake for instance because of irregular absorption. A residual error model


115

accounting for this increased variability was used and is represented by Ci=Ĉi+ θip x εi in which Ci and Ĉi are, respectively, the measured and model predicted drug concentration of the ith individual, εi is the residual random error with mean 0 and variance σ2 and θip is an increment proportion (ip) whose value is to be estimated but fixed to 1 for time points exceeding 2h after intake. Covariate pharmacokinetic model To identify possible relationships between the pharmacokinetics of rivastigmine or NAP 226-90 and patient characteristics, the following co-variates were collected: age (years), gender and body weight (in kg). A covariate was included in an intermediate model when the inclusion of this covariate was statistically significant. A covariate was considered statistically significant when the inclusion was associated with a decrease in minimal value of the OFV of 3.84 points associated with a p-value of <0.05 (log-likelihood ratio test, chi-square distribution, degrees of freedom (df)=1). Finally, a backward elimination procedure was carried out. A parameter was only retained in the model when the influence of this parameter was statistically significant (log likelihood ratio test: p≤0.01, corresponding to an increase in OFV of 6.63 points). Model evaluation The model was evaluated using the jackknife method that investigates the influence of a individual patient on the population model21. A number of new datasets were made by deleting subsequently data of one patient from the primary dataset. Pharmacokinetic parameters were estimated according to the final pharmacokinetic model in each of the new datasets. The median and range were tabulated for all pharmacokinetic parameters and compared with the parameters as obtained during model building with the original full dataset. Pharmacokinetic-Pharmacodynamic analysis Individual pharmacokinetic parameters Individual values for the area under the plasma concentration versus time curve during 12 hours (AUC12) for both rivastigmine and NAP226-90 were calculated as a measure of exposure. Bayesian analysis was used to estimate maximal plasma concentrations (Cmax) and AUC12 of rivastigmine and NAP226-90.

Chapter 3.2.4

116

Efficacy Measurements At baseline and at 6-months intervals rivastigmine use was evaluated by assessment of three domains. Cognition was measured with MMSE22 and Cambridge Cognitive Examination (CAMCOG)23. CAMCOG consists of 60 items covering orientation, language, memory, praxis and calculation/attention, abstract reasoning and perception. Total sum scores range from 0-107. CAMCOG can be subdivided into a memory (maximum score 37) and a non-memory (maximum score 70) section24. Functional disability was measured with the performance subscale of the Interview for Deterioration in Daily living activities in Dementia (IDDD), a caregiver based paper-and-pencil questionnaire, which consists of 11 items with sum scores ranging from 0 to 4425. Behaviour was measured with the Revised Memory and Behavioural Problems Checklist (RMBPC). It is also a caregiver based paper-and-pencil questionnaire and consists of three subsections. First, a 7-item memory subscale (score per item 0-4, maximum score 28,) that includes forgetting recent events, repeated questions, losing things, forgetting the day, forgetting past events, reduced concentration and not finishing tasks. Second, an 8-item disruptive behaviour subscale (score per item 0-4, maximum score 32) including verbal aggression, threats to hurt others, destroying property, to behave dangerous to self or others, talking loudly and rapidly, embarrassing behaviour, arguing and waking caregiver up. Third, a 9-item depression subscale (score per item 0-4, maximum score 36) includes comments about hopelessness, comments about being a burden, appearing sad or depressed, comments about dead, comments about being a failure, crying, comments about loneliness, appearing anxious and suicidal threats26. Lower scores on MMSE and CAMCOG and higher scores on the IDDD and subscales of the RMBPC reflect a worse functioning. Statistics Alzheimer’s disease patients were included in the pharmacokinetic-pharmacodynamic analysis. Patients were divided into responders and non-responders as measured on the different domains: memory related cognition, non-memory related cognition, performance in daily living activities, memory related behaviour, disruptive behaviour and depressive behaviour. For all these test results a median increase or decrease between current test results and the test results at the previous 6-monthly visit was calculated for the total population. Patients were regarded responders if their differences in test results measured on the CAMCOG were equal or larger compared to the median of the population. Patients were also regarded as responders if their differences in test results as measured on the IDDD and subscales of the RMBPC were equal or lower than the median of the population. Mann-Whitney-U testing or Chi-square testing was performed to investigate whether patient characteristics between responders and non-responders were significantly different. As a measure of exposure to rivastigmine and NAP226-90 the maximal plasma


117

concentration (Cmax) and area under the plasma concentration versus time curve (AUC12) were compared between responders and non-responders by Mann-Whitney-U testing. Statistical calculations were performed with SPSS for Windows, version 11.0 (SPSS Inc., Chicago, IL, USA)

Results Patients A total of 33 patients were included with a mean age of 78.4 years, a mean MMSE score of 21.5, the median education level was 4 (on a seven-point scale, ranging from less than 6 years of elementary school (score 1) to a university degree (score 7))27 and they used a mean daily rivastigmine dose of 9.6 mg in different bid dosing regimens (table 1). From these 33 patients, 13 pharmacokinetic curves and 43 plasma concentrations at single time points were available, resulting in a database of 163 rivastigmine and 163 NAP226-90 concentrations. Plasma concentrations of rivastigmine and NAP226-90 are shown in figures 1A and 1B. The concentrations varied between 0.000833 and 0.148 mg/L for rivastigmine and between 0.000544 and 0.0115 mg/L for NAP226-90. Table 1. Baseline characteristics of rivastigmine users. Baseline characteristic Rivastigmine users (n=33) Age (years), mean ± SD (range) 78.4 ± 4.8 (71-94) Gender Male / Female (%) 12/ 21 (36 / 64) Education median, IQR (range) 4, IQR: 2 (2-6) MMSE, mean ± SD (range) 21.5 ± 3.9 (14-29) Weight (kg), mean ± SD (range) 72.4 ± 13.1 (51-110) Clinical Diagnosis Alzheimer’s Disease / Lewy Body Dementia (%)

30 / 3 (91 / 9)

Maximum Achieved Dose, n (%) 3 mg bid 5 (15) 3 mg qd and 4.5 mg qd 6 (18) 4.5 mg bid 5 (15) 4.5 mg qd and 6 mg qd 5 (15) 6 mg bid 12 (37) IQR = interquartile range, MMSE = Mini Mental State Examination

Population pharmacokinetics Both rivastigmine and NAP226-90 data were best described by one-compartment models. The absorption process of rivastigmine was best described by first-order absorption.

Chapter 3.2.4

118

Figure 1. Plasma concentration versus time data for rivastigmine (panel A) and NAP226-90 (panel B). Solid diamonds ( ) connected with lines represent full pharmacokinetic curves, solid squares (■) represent plasma concentrations at a single time point.

(A)

0.00001

0.0001

0.001

0.01

0.1

1

0 2 4 6 8

Time after ingestion (h)

Riv

astig

min

e C

once

ntra

tion

(mg/

L)

(B)

0.0001

0.001

0.01

0.1

1

0 2 4 6 8

Time after ingestion (h)

NA

P226

-90

conc

entra

tion

(mg/

L)

However, due to limited data in the absorption phase, it was not possible to estimate both ka and volume of distribution as was previously shown by Wade et al28. Therefore, the ka was fixed at a value of 0.7 h-1 in subsequent analyses. Description of rivastigmine absorption with an additional lag time or described as time-dependent absorption did not improve the goodness-of-fit. We introduced bioavailability dose-dependently in our model. This did not result in optimisation of the plots or a decrease in OFV. Rivastigmine metabolism into NAP226-90 was best described as a non-linear process and it was possible to estimate both Vmax/F and Km as population parameters. NAP226-90 clearance was modelled with first-order elimination. The model was further described with the parameters V1/F representing


119

he apparent volume of distribution of rivastigmine and V2/F/fm representing the apparent volume of distribution of NAP226-90 divided by the fraction of rivastigmine metabolised to NAP226-90 (fm). The apparent maximum elimination rate (Vmax/F) of rivastigmine, the oral clearance (CL/F/fm) of NAP226-90 and volumes of distribution of both rivastigmine and NAP226-90 were allometrically scaled for bodyweight as follows29: Vmax/Fi=θ1*(weight/70)0.7, CL/F/fm,i=θ2*(weight/70)0.7, V1/Fi=θ3*(weight/70)1, V2/F/fm,i=θ4*(weight/70)1

In which Vmax/Fi, CL/F/fm,i, V1/Fi and V2/F/fm,i represent Vmax/F, Cl/F/fm, V1/F and V2/F/fm of the ith individual, θ1 is the typical value of Vmax/F, θ2 is the typical value of CL/F/fm, θ3 the typical value of V1/F and θ4 is the typical value of V2/F/fm. With this code, the estimated values of Vmax/F, CL/F/fm. V1/F and V2/F/fm represent the value for patients weighing 70 kg. The introduction of an increment proportion (θip) in the error model resulted in an increase in goodness-of-fit (OFV decrease of 39.8 points). Interindividual variability could be estimated in V1/F, V2/F/fm, Km, Vmax/F and F, although all these estimations showed high relative standard errors. Both an additive and proportional error component could be estimated20. The results of the basic pharmacokinetic model are summarised in table 2. The different covariates were introduced separately in the basic model for Vmax/F, Km and CL2/F/fm. Only gender showed a significant relation with Vmax/F by decreasing the OFV with 7.5 points (p<0.01). Females showed a higher maximal elimination rate of rivastigmine compared to males (θGENDER =1.25 (95% confidence interval: 1.04 – 1.46)). The results of the final pharmacokinetic model are summarised in table 2. Figure 2 shows the logarithm of model predicted concentrations (PRED; panel A and C) and the individual model predicted concentrations (IPRED; panel B and D) from the final model versus the observed concentrations of rivastigmine (DV; panel A and B) and NAP226-90 (DV; panel C and D). The following equation describes the final model for Vmax/F: Vmax/F = 1.0 * 1.25GENDER * (weight/70)0.7 where GENDER is 0 for males and 1 for females.

Chapter 3.2.4

120

Table 2. Final parameter estimates of basic and final pharmacokinetic model of rivastigmine and its metabolite NAP226-90.

Basic model Final model

Jackknife evaluation

Estimation RSE (%)

Estimation RSE (%)

Median (range)

V1/F (L) 178 18.9 177 17.9 177 (168-193) Vmax/F (mg/h) 1.17 14.1 1.01 11.5 1.01 (0.94-1.07) θ GENDER* 1.25 8.56 1.26 (1.15-1.29) Km(mg/L) 0.00766 26.9 0.00698 26.8 0.00701 (0.00580-

0.00781) V2/F (L) 39.2 21.2 39.4 15.8 39.5 (34.0-45.6) CL2/F/fm (L/h) 122 6.5 126 5.9 126 (123-130) Increment proportion (< 2h)

2.33 29.4 2.54 28.9 2.54 (2.22 – 3.52)

Interindividual variability V1/F (%)

56.5 37.6 70.6 35.7 71.4 (62.2-74.4)

Interindividual variability Vmax/F (%)

11.4 264 14.1 144 14.0 (0.00-20.3)

Interindividual variability Km (%)

61.7 56.7 65.8 62.6 66.8 (36.7-70.2)

Interindividual variability V2/F/fm (%)

113 79.5 122 60.6 123 (93-142)

Interindividual variability F (%)

26.1 39.6 17.3 67.6 17.3 (14.1-20.5)

Additive error RIV (mg/L)

0.000862 72.5 0.000548 33.4 0.000544 (0.000000-0.000702)

Proportional error RIV (%)

38.2 30.1 36.6 25.9 36.6 (26.0-39.0)

Additive error NAP (mg/L)

0.000928 36.7 0.000574 31.5 0.000573(0.000185-0.000710)

Proportional error NAP (%)

11.6 79.7 17.1 15.1 17.1 (15.4-20.0)

RIV = Rivastigmine, NAP = NAP226-90, V1/F = apparent volume of distribution of rivastigmine, Vmax/F = apparent maximal elimination rate of rivastigmine, Km = Michaelis-Menten constant, V2/F/fm = apparent volume of distribution of NAP226-90, CL2/F/fm = apparent clearance of NAP226-90, F= relative bioavailability, fm = fraction rivastigmine metabolised into NAP226-90, RSE= relative standard error (as calculated with COVARIANCE option of NONMEM) * Vmax/F=θVmax/F * θGENDER

GENDER * (weight/70)0.7, in which GENDER is 1 for women, 0 for men


121

Figure 2. Logarithm of model predicted concentrations (PRED; panel A and C) and the individual model predicted concentrations (IPRED; panel B and D) from the final model versus the observed concentrations of rivastigmine (DV; panel A and B) and NAP226-90 (DV; panel C and D) in mg/L.

(A)

-11

-9

-7

-5

-3

-1

1

-11 -9 -7 -5 -3 -1 1

DV

PR

ED

(B)

-11

-9

-7

-5

-3

-1

1

-11 -9 -7 -5 -3 -1 1

DV

IPR

ED

(C)

-11

-9

-7

-5

-3

-1

1

-11 -9 -7 -5 -3 -1 1

DV

PR

ED

(D)

-11

-9

-7

-5

-3

-1

1

-11 -9 -7 -5 -3 -1 1

DV

IPR

ED

Chapter 3.2.4

122

Pharmacokinetic-Pharmacodynamic correlations From the 33 included patients in the study, 6-monthly test results of 30 AD patients could be used in the pharmacokinetic-pharmacodynamic analysis. The results of these analyses for the different tests are summarised in table 3. No significant differences were shown between responders and non-responders on the different tests for the pharmacokinetic parameters. Table 3. Results of the pharmacokinetic-pharmacodynamic analysis. CAMCOGmem CAMCOGnonmem R (n=17) NR (n=13) p-value* R (n=13) NR (n=17) p-value* Age (years) 78.5 79.8 0.786 77.4 79.8 0.391 Gender, % male 29.4 30.8 1.000 38.5 23.5 0.443 Education 4 4 0.881 4 4 0.864 Baseline 18 19 0.283 54 56 0.600 Riva Cmax (mg/L) 0.00926 0.0112 0.917 0.00866 0.0140 0.346 Riva AUC12 (mg*h/L) 0.0439 0.0556 0.917 0.0395 0.0698 0.439 NAP Cmax (mg/L) 0.00618 0.00502 0.917 0.00519 0.00618 0.722 NAP AUC12 (mg*h/L) 0.0396 0.0352 0.346 0.0352 0.0396 0.884 IDDD RMBPCmem R (n=13) NR (n=13) p-value* R (N=18) NR (=10) p-value* Age (years) 79.8 76.9 0.427 79.0 79.9 0.962 Gender, % male 30.8 23.1 1.000 27.8 40.0 0.677 Education 3 4 0.527 4 4 0.571 Baseline 14 11 0.511 17 15.5 0.323 Riva Cmax (mg/L) 0.0103 0.00926 0.817 0.00953 0.0168 0.270 Riva AUC12 (mg*h/L) 0.0400 0.0439 0.701 0.0397 0.0808 0.164 NAP Cmax (mg/L) 0.00660 0.00490 0.397 0.00566 0.00618 0.145 NAP AUC12 (mg*h/L) 0.0396 0.0265 0.369 0.0317 0.0436 0.057 RMBPCdis RMBPCdep R (n=13) NR (n=16) p-value* R (n=22) NR (n=6) p-value* Age (years) 78.5 79.7 0.759 79.1 77.3 0.314 Gender, % male 15.4 43.8 0.130 31.8 33.3 1.000 Education 4 4 0.892 4 4.5 0.372 Baseline 5 1 0.009 4.5 3 0.612 Riva Cmax (mg/L) 0.00866 0.0105 0.599 0.0100 0.00665 0.145 Riva AUC12 (mg*h/L) 0.0395 0.0570 0.430 0.0497 0.0256 0.117 NAP Cmax (mg/L) 0.00638 0.00535 0.456 0.00628 0.00484 0.145 NAP AUC12 (mg*h/L) 0.0352 0.0351 0.693 0.0374 0.0239 0.057 R=responder , NR=nonresponder, CAMCOGmem= memory subsection of camcog cognitive screening, CAMCOGnonmem= non-memory subsection of camcog cognitive screening, IDDD= performance in daily living activities in dementia, RMBPCdis= disruptive subscale of the RMBPC, RMBPCmem= memory subscale of the RMBPC, RMBPCdep= depressive subscale of RMBPC, Riva Cmax= rivastigmine maximal plasma concentration, Riva AUC12= Area under the plasma concentration time curve (0-12 h) of rivastigmine, NAP Cmax= NAP226-90 maximal plasma concentration, NAP AUC12= Area under the plasma concentration time curve (0-12h) of NAP226-90, *based upon Mann-Whitney-U testing or Chi-square statistics when appropriate


123

Discussion Observed plasma concentrations of rivastigmine and its metabolite NAP226-90 in our relatively unselected outpatient population were in the same range as earlier described in rivastigmine controlled clinical trials in Alzheimer patients12,30. Pronounced non-linear elimination of rivastigmine was found in our study, which in agreement with Hossain et al.30 but not with the study of Gobburu et al.12. Final parameter estimates for V1/F and Vmax/F were comparable to those estimated by Hossain et al.30 in 11 AD patients (mean age 69.5 years). The typical values of Km and volumes of distribution of rivastigmine and NAP226-90 were lower compared to those earlier described studies12,30. We showed a non-linear elimination of rivastigmine in the treated outpatients with a value for Km of 0.00698 mg/L. The concentrations that we measured in our study are in a wide range around this value and indicate that there is variability in rivastigmine elimination between individuals. Concentrations of rivastigmine increase thus non-lineary with higher doses. Our results indicate that increasing doses may lead to disproportional increases in exposure of rivastigmine. Higher doses of rivastigmine are often not tolerated by patients because of gastrointestinal adverse events31. It is known that daily doses above twice daily 3 mg are required for optimal efficacy1,2, but it is questionable if it should be intended to increase rivastigmine dose up to twice daily 6 mg in all patients, because that may result in disproportionally higher rivastigmine concentrations and an increased risk of adverse events. Interindividual variability for both rivastigmine and NAP226-90 pharmacokinetics were large in our population. For example the interindividual variability of V/F for rivastigmine was 70.6%. Goburru et al.12 did not show any correlation between pharmacokinetic parameters with age, gender or body weight. The maximal elimination rate of rivastigmine was, however, significantly increased in females in our study, which results in a lower exposure to rivastigmine in females. The basal acetylcholinesterase activity is significantly larger in females compared to men12. Because rivastigmine is metabolised by acetylcholinesterase into NAP226-90 this possibly explains why females showed an increased maximal rate of rivastigmine elimination. In clinical practice often only the MMSE is used to evaluate effect of rivastigmine therapy. We measured the effects of rivastigmine in clinical practice extensively in the domains cognition (CAMCOG), activities of daily living (IDDD) and behaviour (RMBPC). No relation between both rivastigmine and NAP226-90 exposure and effect parameters in the different domains was found in this population of outpatients treated with rivastigmine for minimal 6 months. An earlier study in rats indicated that the concentration NAP226-90 was probably a better predictor of cholinesterase inhibition than rivastigmine itself10. Our results in the pharmacokinetic-pharmacodynamic analysis, however, do not reveal differences in the relation between exposure to rivastigmine or NAP226-90 and effectiveness.

Chapter 3.2.4

124

Rivastigmine inhibits both acetylcholinesterase and butyrylcholinesterase. Plasma butyrylcholinesterase activity has not been related to cognitive screening results while cerebrospinal fluid butyrylcholinesterase activity was related32. Therefore, it can be questioned whether inhibition of plasma cholinesterase and the concentration of NAP226-90 are useful as surrogate markers for brain cholinesterase inhibition and therapy effectiveness. Rivastigmine treatment effects on cognition, activities of daily living and behaviour show large variability between individual patients treated in a outpatient setting33. This variability in treatment effects was also found in patients included in the currently described pharmacokinetic-pharmacodynamic analysis. It is largely unknown which patients respond to cholinesterase-inhibitor treatment. Unfortunately, we did not found a relation between rivastigmine pharmacokinetics and treatment effects. This may be explained because the period of enzyme inhibition is substantially longer compared to the appearance of rivastigmine in plasma7. Therefore, differences in enzyme inhibition between individuals may be better predictors of therapy effects compared to pharmacokinetic variables. However, plasma cholinesterase activity may not be useful at all as a surrogate marker as described above. Clinical or neurophysiologic parameters may better predict therapy efficacy than differences in pharmacokinetics between individuals. We have shown that patients with more severe AD, as defined by a MMSE score ≤ 19, responded better to rivastigmine therapy compared to AD patients with a baseline MMSE score ≥ 20 in a study performed in patients treated via our geriatric outpatient department34. Adler et al.35 have shown that treatment responders had a greater decrease in theta power as measured with a electro encephalogram (EEG). In conclusion, we developed a population pharmacokinetic model that simultaneously described rivastigmine and its metabolite NAP226-90 plasma pharmacokinetics. This study is the first to describe a pharmacokinetic model in geriatric outpatients suffering from Alzheimer’s disease or Lewy body dementia and treated for minimal 6 months. We found no relation between plasma pharmacokinetics of both rivastigmine and NAP226-90 and therapy effectiveness of rivastigmine as measured on cognition, activities of daily living and behaviour, which indicates that other factors may be more important in the observed variability in treatment response.

Acknowledgements Edith Vermaat is kindly acknowledged for her help in including patients when they attended the geriatric outpatient department for 6-monthly cognitive screening visits.


125


(rivastigmine tartrate), a new acetylcholinesterase inhibitor, in patients with mild to moderately severe Alzheimer’s disease. Int J Geriatr Psychopharmacol 1998; 1:55-65.

2. Rösler M, Anand R, Cicin-Sain A, et al. Efficacy and safety of rivastigmine in patients with Alzheimer’s disease: international randomized controlled trial. BMJ 1999;318:633-638.

3. McKeith IG, Del Ser T, Spano P, et al. Efficacy of rivastigmine in dementia with Lewy Bodies: a randomised, double-blind, placebo-controlled international study. Lancet 2000;356:2031-36.

4. Wesnes KA, McKeith IG, Ferrar R, et al. Effects of rivastigmine on cognitive function in dementia with Lewy Bodies: a randomised placebo-controlled international study using the cognitive drug research computerised assessment system. Dement Geriatr Cogn Disord 2002;13:183-192.

5. Tiraboschi P, Hansen LA, Alford M, Sabbagh MN, Schoos S, Masliah E, et al. Cholinergic dysfunction in diseases with Lewy bodies. Neurology 2000;54:407-11.

6. Wynn ZJ, Cummings JL. Cholinesterase inhibitor therapies and neuropsychiatric manifestations of Alzheimer’s disease. Dement Geriatr Cogn Disord 2004;17:100-108.

7. Jann MW, Shirley KL, Small GW. Clinical pharmacokinetics and pharmacodynamics of cholinesterase inhibitors. Clin Pharmacokinet 2002;41:719-739.


9. Nordberg A, Svensson AL. Cholinesterase inhibitors in the treatment of Alzheimer´s disease: a comparison of tolerability and pharmacology. Drug Saf 1998;19:465-80.

10. Enz A, Chappuis A, Dattler A. A simple, rapid and sensitive method for simultaneous determination of rivastigmine and its major metabolite NAP 226-90 in rat brain and plasma by reversed-phase liquid chromatography to electrospray ionization mass spectrometry. Biomed Chromatogr 2004;18:160-166.


12. Gobburu JVS, Tammara V, Lesko L, et al. Pharmacokinetic-pharmacodynamic modeling of rivastigmine, a cholinesterase inhibitor, in patients with Alzheimer´s disease. J Clin Pharm 2001;41:1082-1090.

13. Rogers SL, Friedhoff LT. The efficacy and safety of donepezil in patients with Alzheimer´s Disease: results of a US multicentre, randomized, double blind, placebo controlled trial. Dementia 1996;7:293-303.


15. McKeith IG, Tatemichi TK, Erkinjutti T, et al. Consensus guidelines for the clinical and pathological diagnosis of dementia with Lewy Bodies (DLB): report of the consortium on DLB international workshop. Neurology 1996;47:1113-1124.

16. Frankfort SV, Ouwehand M, van Maanen MJ, et al. A simple and sensitive assay for the quantitative analysis of rivastigmine and its metabolite NAP 226-90 in human EDTA plasma using coupled liquid chromatography and tandem mass spectrometry. Rapid Comm Mass Spectr 2006; 20:3330-3336.

17. Beal SL, Sheiner LB. NONMEM User´s guides. NONMEM Project Group, University of California at San Francisco, 1998.

18. Kappelhoff BS, Huitema ADR, Yalvac Z, et al. Population pharmacokinetics of efavirenz in an unselected cohort of HIV-1-infected individuals. Clin Pharmacokinet 2005;44:849-861.

19. Karlsson MO, Sheiner LB. The importance of modelling interoccassion variability in population pharmacokinetic analyses. J Pharmacokinet Biopharm 1993;21:735-750.

20. Beal SL. Ways to fit a PK model with some data below the quantification limit. J Pharmacokinet Pharmacodyn 2001;28:481-504.

Chapter 3.2.4

126

21. Armitage P, Berry G, Matthews JNS. Statistical methods in medical research. Blackwell Publishing: Berlin Germany, 2002.

22. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state” a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12:189-198.

23. Roth M, Tym E, Mountjoy CQ, et al. CAMDEX A standardized instrument for the diagnosis of mental disorder in the elderly with special reference to early detection of dementia. Br J Psychiatry 1986;149:698-709.

24. Schmand B, Walstra G, Lindeboom J, Teunisse S, Jonker C. Early detection of Alzheimer’s disease using the Cambridge Cognitive Examination (CAMCOG). Psychol Med 2000;30:619-627.

25. Teunisse S, Derix MM. Measuring functional disability in community-dwelling dementia patients: development of a questionnaire (in Dutch). Tijdschr Gerontol Geriatr 1991;22:53-59.

26. Teri L, Truax P, Logsdon R, Uomoto J, Zarit S, Vitaliano PP. Assessment of behavioral problems in dementia: the revised memory and behavior problems checklist. Psychol Aging 1992;7:622-631.

27. Verhage F. Intelligence and Age (in Dutch). Assen, van Gorcum, 1964. 28. Wade JR, Kelman AW, Howie CA, Whiting B. Effect of misspecification of the “absorption process” on

subsequent parameter estimates in population analysis. J Pharmacokinet Biopharm 1993;21:209-222. 29. Hu TM, Hayton WL. Allometric scaling of xenobiotic clearance: uncertainty versus universality. AAPS

PharmSci2001;3:E29. 30. Hossain M, Jhee SS, Shiovitz T, et al. Estimation of the absolute bioavailability of rivastigmine in patients

with mild-to-moderate dementia of the Alzheimer´s type. Clin Pharmacokinet 2002;41:225-234. 31. Gauthier S. Cholinergic adverse effects of cholinesterase inhibitors in Alzheimer’s disease. Drugs Aging

200;118:853-862. 32. Giacobini E, Spiegel R, Enz A, et al. Inhibition of acetyl- and butyrylcholinesterase in the cerebrospinal fluid

of patients with Alzheimer’s disease by rivastigmine: correlation with cognitive benefit. J Neural Transm 2002;109:1053-1065.

33. Frankfort SV, Appels BA, de Boer A, et al. Treatment effects of rivastigmine on cognition, performance of daily living activities and behaviour in Alzheimer’s disease in an outpatient geriatric setting. Int J Clin Pract 2006;60:646-654.

34. Frankfort SV, Appels BA, de Boer A, et al. Identification of responders and reactive domains to rivastigmine in Alzheimer’s disease. Pharmacoepidemiol Drug Saf 2006; in press.

35. Adler G, Brassen S, Chwalek K, et al. Prediction of treatment respons to rivastigmine in Alzheimer’s dementia. J Neurol Neurosurg Psychiatry 2004;75:292-294.

127

128

129

CHAPTER 4

BIOMARKERS FOR DEMENTIA

130

131

CHAPTER 4.1

GENETIC BIOMARKERS

132

133

CHAPTER 4.1.1 APOE genotype and allele distribution in geriatric

outpatients and healthy volunteers

S.V. Frankfort, R. Bakker, C.R. Tulner, J.P.C.M. van Campen, P.H.M. Smits, J.H. Beijnen

Abstract

Background: The Apolipoprotein E (APOE) ε4 allele has been recognised as a risk factor in Alzheimer ’s disease (AD) and more recently also for the transition from cognitively healthy to Mild Cognitive Impairment (MCI) to AD. Aim: To describe APOE genotypes and allele frequencies in different types of dementia and age-matched non-demented control patients diagnosed in an outpatient setting, and to compare APOE genotypes and allele frequencies in geriatric patients with healthy younger volunteers. Methods: APOE genotyping was performed using Real Time Polymerase Chain Reaction in patients with AD, Vascular Dementia (VaD), a heterogeneous group of other dementias, MCI, and age-matched non-demented controls. Also a group of healthy younger volunteers was screened for APOE genotype. A p-value of 0.01 or less was considered statistically significant, due to corrections for multiple testing. Results: 113 patients (mean age 81.6 years (range: 63.3-94.8)) and 41 age-matched controls (mean age 81.9 years (69.5-94.5)) were included. In AD compared to age-matched controls a significant difference was shown in total APOE allele frequencies (p=0.005). APOE ε4 allele was significantly more frequent in MCI compared to age-matched controls (p=0.004) and trends for higher frequencies in AD compared to age-matched controls (p=0.014) and MCI compared to other dementias (p=0.015) were shown. A trend for the APOE ε2 allele to be more frequent in controls compared to AD was found (p=0.011). No significant differences in APOE genotypes between the different geriatric subpopulations were found. No significant differences were found between APOE genotypes and allele frequencies between the total geriatric population (MCI, all dementia patients and age-matched controls) and 99 younger healthy volunteers. Conclusions: APOE allele frequencies were significantly different between AD and age-matched non-demented control patients. The APOE ε4 allele may be a genetic marker for the development of MCI and AD. Geriatric patients attending the day-clinic do not belong to a selected population based upon APOE genotypes as compared with younger healthy volunteers.

Submitted

Chapter 4.1.1

134

Introduction The Apolipoprotein E (APOE) ε4 allele has been recognised as a risk factor in Alzheimer ’s disease (AD)1. Furthermore, it was recently shown that the presence of an APOE ε4 allele affects transition from cognitively healthy to Mild Cognitive Impairment (MCI) to AD2. Research, however, yielded conflicting results regarding APOE ε4 allele as a risk factor for Vascular Dementia (VaD)3. Presently, dementia diagnosis is based upon clinical examination, in combination with neuropsychological testing and neuroimaging. We conducted analyses into APOE genotypes in the population patients attending our geriatric diagnostic day-clinic. This study aims to describe APOE genotypes and allele frequencies in different types of dementia and age-matched non-demented control patients diagnosed in an outpatient setting. We wondered if the patients that attend our geriatric diagnostic day-clinic belong to a selected population carrying the APOE ε4 allele. Therefore we compared the results of the total geriatric population to a group of healthy volunteers.

Patients and methods Patients and diagnosis APOE genotyping was performed at the geriatric diagnostic day-clinic of the Slotervaart Hospital, a teaching hospital in Amsterdam, the Netherlands, between July 2004 and August 2005. Complete geriatric assessment was performed including: Mini Mental State Examination (MMSE)4, the 7-Minute Neurocognitive Screening Test5 and laboratory testing. Patients underwent, if necessary, computerised tomography or magnetic resonance imaging was performed. Current guidelines were used for diagnosing AD6, Lewy Body Disease (LBD)7, VaD8, frontotemporal lobar degeneration (FTD)9 and MCI10,11. We categorised patients as unspecified dementia if the underlying process could not be diagnosed. Age-matched controls (n=41) were recruited from the same diagnostic day-clinic. These participants did not show any cognitive impairment. Most of these geriatric patients were presented at the day clinic for a somatic screening. In addition to the geriatric patients included at the geriatric day-clinic, a number of 99 healthy volunteers (mean age 41.2 years) were also screened for APOE genotypes. The study protocol was approved by the Institutional Review Board of the Slotervaart Hospital, Amsterdam, The Netherlands. Written informed consent was obtained from each participant in this study.

APOE genotypes in geriatric outpatients

135

Blood sampling and DNA isolation From each participant a 2 ml EDTA blood sample was obtained by venous puncture and stored at -20 °C until genotype analysis. Genomic DNA was extracted from EDTA whole blood using the Qiagen QIAamp® DNA Mini Kit (Qiagen, Leusden, The Netherlands) according to the manufacturer protocol. APOE: Real Time-Polymerase Chain Reaction (RT-PCR) APOE was screened by RT-PCRs for SNPs at position T334C and C472T. The combinations 334T/472T, 334T/472C and 334C/472C constitute the known isoform-specific APOE alleles ε2, ε3 and ε4, respectively12. RT-PCR, based upon the method by Koch et al.12 was carried out in a final volume of 25 µl with 300 nmol forward and reverse primers (Eurogentec, Belgium), 100 nmol of FAM labelled probe (Eurogentec, Belgium) and 100 nmol of VIC labelled probe (Eurogentec, Belgium), 12.5 µl of TaqMan® Universal Mastermix with AmpErase (Applied Biosystems, Foster City, USA) and 250 ng DNA. At least one negative control sample and five positive control samples (2/2, 2/3, 3/3, 3/4 and 4/4) were systematically included in each analysis. PCR was carried out in a real-time thermocycler (ABI PRISM Sequence Detection System 7000, Applied Biosystems). After the AmpErase step (2 minutes 50 °C) and enzyme activation for 10 minutes at 95 °C, 40 two-cycle steps were performed: 15 seconds of denaturation at 95 °C and 1 minute of annealing and elongation at 60 °C. Allelic discrimination was performed by measuring fluorescence intensity at the endpoint after 1 minute at 60 °C. The results of the measurement were analysed using SDS software version 1.1 (Applied Biosystems) and the genotype was determined. Statistical analysis The Pearson Chi-square test and Fisher’s exact testing, when appropriate, were used to compare gender and APOE genotypes and allele frequencies between different geriatric subgroups. A one-way ANOVA was used to compare mean age and MMSE between different groups and Kruskal-Wallis testing was performed to compare education level, as scored on a seven-point scale, ranging from less than 6 years of elementary school (score 1) to a university degree (score 7)13. Significant differences in genotype and allele frequencies between the geriatric population and the healthy volunteer population were investigated by Chi-square testing. A p-value of 0.01 or less was considered statistically significant, due to corrections for multiple testing. Statistical calculations were performed with SPSS for Windows (version 12.0, SPSS Inc., Chicago, IL, USA).

Chapter 4.1.1

136

Results Patient characteristics A total of 161 geriatric patients signed informed consent of which seven were excluded: three patients refused further medical examination, so their diagnoses could not be established and four patients were diagnosed delirious. The 154 included participants consisted of 113 patients (48 AD, 19 VaD, 26 other dementia (OD), 20 MCI) and 41 age-matched controls. The group of OD included 10 patients with a mixed type of dementia, three with LBD, three with alcohol induced dementia, two with FTD and eight “unspecified” dementia subjects. Age (p=0.459), gender (p=0.084) and education level (p=0.281) were not significantly different between the subgroups. Mean MMSE score was significantly different between the groups, post hoc analysis showed significant differences between the dementia subgroups and the control group (p<0.001) and between MCI and AD/VaD (p<0.001) (table 1). In addition, a total of 99 healthy younger volunteers (mean age 41.2 years) were separately included. Table 1. Demographic characteristics, genotypes and allele frequencies (%) of APOE in geriatric patients. Controls

(n=41) AD (n=48)

VaD (n=19)

OD (n=26)

MCI (n=20)

Demographics Age, mean ± SD (range)

81.9±5.7 (69.5–94.5)

81.0 ± 5.5 (71.9-93.3 )

83.6±5.6 (67.6-89.7 )

82.1 ± 5.2 (69.6-94.8)

80.5±6.9 (63.3-90.7 )

Gender, n (%) female 27 (65.9) 33 (68.8) 8 (42.1) 11 (42.3) 13 (65.0) Education, median IQR (range)

4 IQR:3 (1-6)&

4 IQR:3 (1-7)§

5 IQR:2 (4-6)§

4 IQR: 3

(2-7) § 4 IQR:3 (2-6) &

MMSE, mean ± SD (range)

26.6 ± 2.2 (21- 29)~

18.2 ± 4.4* #

(6- 26)’ 18.1 ± 7.5* # (3-29 )

20.9 ± 5.2*

(12-30)§ 25.3 ± 2.7

(19-30)§ Genotype frequency, n (%) ε2/ ε2 1 (2.4) 0 (0.0) 0 ( 0.0) 0 (0.0) 0 (0.0) ε2/ ε3 9 (22.0) 2 (4.2) 3 (15.8 ) 2 (7.7) 2 (10.0) ε2/ ε4 0 (0.0) 1 (2.1) 1 (5.3) 0 (0.0) 1 (5.0) ε3/ ε3 21 (51.2) 22 (45.8) 9 (47.4 ) 16 (61.5) 6 (30.0) ε3/ ε4 8 (19.5) 18 (37.5) 3 (15.8) 8 (30.8) 8 (40.0) ε4/ ε4 2 (4.9) 5 (10.4) 3 (15.8 ) 0 (0.0) 3 (15.0) Allele frequency**, n (%) ε2## 11 (13.4) 3 (3.1) 4 (10.5) 2 (3.8) 3 (7.5) ε3 59 (72.0) 64 (66.7) 24 (63.2) 42 (80.8) 22 (55.0) ε4§§ 12 (14.6) 29 (30.2) 10 (26.3) 8 (15.4) 15 (37.5) AD= Alzheimer’s Disease, VaD= Vascular Dementia, OD= Other dementias, MCI= Mild Cognitive Impairment, IQR = Inter Quartile Range, MMSE = Mini Mental State Examination, § missing for 1 patient,& missing for 2 patients, ‘missing for 3 patients, ~ performed in 22 controls ;* p<0.001 vs controls, # p<0.001 vs MCI, **p=0.005 AD vs controls, ## p=0.011 AD vs controls, §§p=0.004 MCI vs controls, p=0.014 AD vs controls, p=0.015 OD vs MCI


137

APOE genotypes and allele distribution APOE genotype and allele frequencies in the geriatric subpopulations are shown in table 1. APOE allele distribution was significantly different between the different geriatric subpopulations (p=0.009). After post-hoc analyses a significant difference was shown between AD and controls (p=0.005). No significant differences were reached between AD and VaD (p=0.246), AD and MCI (p=0.298), VaD and controls (p=0.323), MCI and OD (p=0.023) and MCI and controls (p=0.022). APOE ε4 allele frequencies differed significantly between MCI and controls (p=0.004) and a trend was shown in AD compared to controls (p=0.014) and in other dementias compared to MCI (p=0.015). A trend for higher APOE ε2 allele frequencies in controls compared to AD was shown (p=0.011). APOE genotypes were not significantly different between the geriatric subpopulations (p=0.252). Table 2. APOE genotypes and allele frequencies (%) in healthy younger volunteers and geriatric patients. Healthy volunteers (n=99) Total geriatric population (n=154) p-value Genotype frequency, n (%) 0.735 ε2/ ε2 1 (1.0) 1 (0.6) ε2/ ε3 13 (13.1) 18 (11.7) ε2/ ε4 1 (1.0) 3 (2.0) ε3/ ε3 52 (52.5) 74 (48.0) ε3/ ε4 29 (29.5) 45 (29.2) ε4/ ε4 3 (3.0) 13 (8.4) Allele frequency, n (%) 0.296 ε2 16 (8.1) 24 (7.8) ε3 146 (73.7) 210 (68.2) ε4 36 (18.2) 74 (24.0)

The APOE genotype and allele distributions were not significantly different between the total geriatric population and the younger volunteer population (table 2). In the geriatric population a not significant increase in patients homozygous for the ε4 allele was found compared to the healthy volunteers.

Discussion We described the APOE genotype and allele frequencies in geriatric outpatients suffering from different types of dementia and age-matched non-demented control patients. The distribution was comparable to those earlier described for other caucasian populations 3,14,15. APOE allele frequencies were significantly different between AD and age-matched non-demented controls, as earlier described3,15. APOE ε2 allele frequencies were more frequent in age-matched controls versus AD, which is also as earlier described16. Our results did not

Chapter 4.1.1

138

show a significant difference in all allele frequencies, or more specific in APOE ε4 allele frequencies between VaD and age-matched controls and this study thus strengthens earlier reported conflicting results3,17-20. APOE ε4 allele frequencies did differ between MCI and age-matched controls and as Kryscio et al.2 described this allele to affect the transition from control to MCI to AD, it may serve an important role in supporting the clinical diagnosis. We showed higher APOE ε4 allele frequencies in MCI compared to other dementias. Engelborghs et al.15 showed no significant differences between FTD and controls and between LBD and controls. They did not investigate differences between those entities and patients with MCI. Whether the APOE ε4 allele has a future role in diagnosing FTD or LBD deserves further attention in larger studies. We did not show significant differences between the APOE genotype and allele frequencies in geriatric patients and healthy younger volunteers. Thus, the population of geriatric patients who attend the diagnostic day-clinic is apparently not a selected population based upon their APOE genotypes. A non-significant increase in patients homozygous for the ε4 allele was shown in geriatric patients compared to healthy volunteers. This can be explained by the higher frequencies of patients homozygous for the ε4 allele in the subpopulation of patients with Alzheimer’s dementia or vascular dementia. Our results show comparable frequencies of APOE genotypes in different types of dementia. APOE allele frequencies were significantly different between AD and age-matched non-demented control patients. The APOE ε4 allele may be a genetic marker for the development of MCI and AD. Additional research is warranted into the role of the APOE ε4 allele in VaD, LBD and FTD. In our study, geriatric patients attending the day-clinic do not belong to a selected population based upon APOE genotypes, as compared to younger healthy volunteers.

References 1. Wesson Ashford J. APOE genotype effects on Alzheimer’s disease onset and epidemiology. J Mol

Neuroscience 2004;23:155-63. 2. Kryscio RJ, Schmitt FA, Salazar JC, et al. Risk factors for transitions from normal to mild cognitive

impairment and dementia. Neurology 2006;66:828-32. 3. Davidson Y, Gibbons L, Purandare N, et al. Apolipoprotein E ε4 allele frequency in vascular dementia.

Dement Geriatric Cogn Disord 2006;22:15-9. 4. Folstein MF, Folstein SE, McHugh PR. Mini-Mental State: a practical method for grading cognitive state of

patients for the clinician. Journal of Psychiatric Research 1975;12:189-198. 5. Solomon PR, Hirschoff A, Kelly B, et al. A 7 Minute Neurocognitive Screening Battery highly sensitive to

Alzheimer´s disease. Arch Neurol 1998;55:349-55. 6. McKahnn G, Drachmann D, Folstein M et al. Clinical diagnosis of Alzheimer´s disease: Report of the

NINCDS-ADRDA work group under the auspices of department of health and human services task force on Alzheimer´s disease. Neurology 1984;34:939-944.


139

7. McKeith IG, Tatemichi TK, Erkinjutti T, et al. Consensus guidelines for the clinical and pathological diagnosis of dementia with Lewy Bodies (DLB): report of the consortium on DLB international workshop. Neurology 1996;47:1113-1124.

8. Erkinjunnti T. Clinical criteria for vascular dementia: the NINDS-AIREN criteria. Dementia 1994;5:189-92. 9. Neary D, Snowden JS, Gustafson L, et al. Frontotemporal lobar degeneration: a consensus on clinical

diagnostic criteria. Neurology 1998;51:1546-54. 10. Petersen RC, Doody R, Kurz A, et al. Current concepts in mild cognitive impairment. Arch Neurol 2001;

58:1985-92. 11. Petersen RC, Smith GE, Waring SC, Ivnik RJ et al. Mild cognitive impairment: clinical characterisation and

outcome. Arch Neurol 1999;56:303-308. 12. Koch W, Ehrenhaft A, Griesser K, et al. TaqMan systems for genotyping of disease-related polymorphisms

present in the gene encoding apolipoprotein E. Clin Chem Lab Med 2002;40:1123-31. 13. Verhage F. Intelligence and age. Van Gorcum, Assen, 1964 (in Dutch). 14. Corbo RM, Scacchi R. Apolipoprotein E (APOE) allele distribution in the world. Is APOE*4 a “thrifty”

allele? Ann Hum Genet 1999;63:301-10. 15. Engelsborghs S, Dermaut B, Goeman J, et al. Prospective Belgian study of neurodegenerative and vascular

dementia: APOE genotype effects. J Neurol Neurosurg Psychiatry 2003;74:1148-1151. 16. Talbot C, Lendon C, Craddock N, et al. Protection against Alzheimer’s disease with APOE ε2. Lancet

1994;343:1432-3. 17. Ji Y, Urakami K, Adachi Y, et al. Apolipoprotein E polymorphism in patients with Alzheimer’s disease,

vascular dementia and ischaemic disease. Dementia Geriat Cogn Disord 1998;9:243-245. 18. Slooter AJ, Tang MX, van Duijn CM, et al. Apolipoprotein E epsilon 4 and the risk of dementia with stroke:

a population based study. JAMA 1997;277:818-821. 19. Pirttila T, Lehtimaki T, Rinne J, et al. The frequency of apolipoproteine E epsilon 4 allele is not increased in

patiens with probable vascular dementia. Acta Neurol Scand 1996;93:352-354. 20. Frank A, Diez-Tejedor E, Bullido MJ, et al. APOE genotype in cerebrovascular disease and vascular

dementia. J Neurol Sci 2002;203:173-176.

140

141

CHAPTER 4.1.2 ABCB1 genotypes and haplotypes in patients with dementia

and age-matched non-demented control patients

S.V Frankfort, V.D Doodeman, R. Bakker, C.R Tulner, J.P.C.M van Campen, P.H.M Smits, J.H Beijnen

Abstract Amyloid β is an in vitro substrate for P-glycoprotein (P-gp), an efflux pump at the blood brain barrier (BBB). The Multi Drug Resistance (ABCB1) gene, encoding for P-gp, is highly polymorphic and this may result in a changed function of P-gp and may possibly interfere with the pathogenesis of Alzheimer’s disease. This study investigates to what extent ABCB1 Single Nucleotide Polymorphisms (SNPs; C1236T in exon 12, G2677T/A in exon 21 and C3435T in exon 26) and inferred haplotypes exist in an elderly population and if these SNPs and haplotypes differ between patients with dementia and age-matched non-demented control patients. ABCB1 genotype, allele and haplotype frequencies were neither significantly different between patients with dementia and age-matched controls, nor between subgroups of different types of dementia nor age-matched controls. This study shows ABCB1 genotype frequencies to be comparable with described younger populations. To our knowledge this is the first study on ABCB1 genotypes in dementia. ABCB1 genotypes are presently not useful as a biomarker for dementia, as they were not significantly different between demented patients and age-matched control subjects.

Molecular Neurodegeneration 2006;1:13

Chapter 4.1.2

142

Findings P-glycoprotein (P-gp), a 170 kDa membrane bound efflux pump at the apical membrane of endothelial cells, functions as part of the blood brain barrier (BBB)1,2 and is also expressed at the blood-cerebrospinal fluid (BCSF) barrier, formed by the choroid plexus3. The Multi Drug Resistance gene (ABCB1) encodes for P-gp. It is known that the ABCB1 gene is highly polymorphic4. The three most frequently occurring Single Nucleotide Polymorphisms (SNPs) are C1236T in exon 12 (dbSNP: rs1128503), G2677T/A in exon 21 (dbSNP: rs2032582) and C3435T in exon 26 (dbSNP: rs1045642)5. ABCB1 haplotypes composed of different SNPs may better represent changes in P-gp function4. The “amyloid hypothesis” states that accumulation of beta amyloid peptides in the brain is the key event in the pathogenesis of Alzheimer´s Disease (AD)6. Amyloid deposits in plaques in brain parenchyma and along the vascular system7. Amyloid β is an in vitro substrate for P-gp8 and recent research found that P-gp deficiency at the BBB increases β amyloid deposition in an AD mouse model9. Vogelgesang et al.10 showed P-gp expression at the BBB to be inversely correlated to the number of amyloid plaques in the medial temporal lobe in 243 non-demented elderly. Thus, the efflux pump P-gp possibly plays a role in the pathogenesis of late-onset dementia by interfering with the amyloid clearance, as late-onset AD would result from inefficient clearance of beta amyloid from the brain11. We hypothesised ABCB1 genotypes to be related to dementia occurrence as amyloid load in the brain is possibly inversely related to P-gp expression at the BBB and ABCB1 SNPs and haplotypes may be related to P-gp expression and function. This study aimed to test this hypothesis in an elderly population consisting of patients suffering from dementia and age-matched non-demented control patients. This prospective study was carried out at the geriatric diagnostic day-clinic of the Slotervaart Hospital, a teaching hospital in Amsterdam, the Netherlands. Dementia was diagnosed or excluded after performing complete geriatric assessment including: Mini Mental State Examination (MMSE)12, the 7-Minute Neurocognitive Screening Test13 and laboratory testing, including thyroid function, levels of folic acid, thiamine and vitamin B12. Thereafter, patients underwent more extensive neuropsychological assessment and, if necessary, computerised tomography or magnetic resonance imaging was performed. Mild Cognitive Impairment (MCI) and different types of dementia were diagnosed according to the current guidelines14-19. We categorised patients as unspecified dementia if the underlying process could not be diagnosed. Age-matched controls were recruited from the same diagnostic day-clinic. These participants did not show any cognitive impairment. Most of these geriatric patients were presented at the day clinic for a somatic screening. The study protocol was approved by the Institutional Review Board of the Slotervaart Hospital, Amsterdam, The Netherlands. Written informed consent was obtained from each participant in this study.

ABCB1 genotypes in geriatric patients

143

From each participant a 2 ml EDTA blood sample was obtained by venous puncture and genomic DNA was extracted using the Qiagen QIAamp® DNA Mini Kit (Qiagen, Leusden, The Netherlands) according to the manufacturer protocol. ABCB1 was screened for C1236T in exon 12 (dbSNP: rs1128503), G2677T/A in exon 21 (dbSNP: rs2032582) and C3435T in exon 26 (dbSNP: rs1045642) by sequencing, as earlier described20. APOE genotype was determined using real-time polymerase chain reactions based upon Koch et al.21. The Chi square statistic test was used to calculate whether the alleles are in Hardy-Weinberg Equilibrium (HWE). Linkage disequilibrium (LD) between ABCB1 SNPs was performed by Graphical Overview of Linkage Disequilibrium (GOLD) software V1.1.0.022 and haplotype analysis with the software package HPlus65v2.1.123. The Pearson Chi-square test was used to compare categorical variables and a one-way ANOVA for continuous variables. Kruskal-Wallis testing was performed to compare education level, scored on a seven-point scale, ranging from less than 6 years of elementary school (score 1) to a university degree (score 7)24. Logistic regression was performed to investigate the role of the APOE ε4 allele as a possible confounder, with the different types of dementia compared to controls as dependent variables and the described ABCB1 SNPs (as wild-type, heterozygous and homozygous mutants) and APOE (as ε4 allele carriers vs. non-carriers) as independent variables. A p-value of 0.05 or less was considered statistically significant. Bonferroni corrections were made in case of multiple testing. Statistical calculations were performed with SPSS for Windows (version 12.0, SPSS Inc., Chicago, IL, USA). A power analysis, using NQUERY advisor version 5.0, was performed for Chi-square testing between two groups (AD vs. controls) comparing proportions in three categories (wild-type, heterozygous and homozygous mutants). In total, 161 patients signed informed consent. Seven patients were excluded, because of refusing further medical examination (n=3) or because patients were diagnosed delirious (n=4). The 154 included participants consisted of 113 patients (48 AD, 19 Vascular Dementia (VaD), 26 other dementia (OD), and 20 MCI) and 41 age-matched controls. The group of OD included 10 patients with a mixed type of dementia, 3 with Lewy Body Disease, 3 with alcohol induced dementia, 2 with Frontotemporal Dementia and 8 “unspecified” dementia syndromes. Baseline characteristics are presented in the table. The total population (n=154) had a mean age of 81.7 ± 5.9 (63.3-94.8) years and almost 60% was female. Age, gender and education level were not significantly different between the subgroups. Mean MMSE score was significantly different between the dementia subgroups and the control group (p<0.001), which is as expected. Only in the MCI group allele frequencies for G2677T/A did not apply to Hardy Weinberg Equilibrium, although the other allele frequencies in all other groups did apply to HWE (results not shown). Strong linkage was observed between C1236T and G2677T/A (ρ2= 0.831, p<0.000001), between C1236T and C3435T (ρ2=0.424, p<0.000001) and between

Chapter 4.1.2

144

Table. Demographic characteristics and ABCB1 genotype, allele and haplotype frequencies (n, (%)) Total POP


AD (n=48)

VaD (n=19)

OD (n=26)

MCI (n=20)

Age, Mean ± SD (range)

81.7±5.9 (63.3-94.8 )

81.9±5.7 (69.5–4.5)

81.0 ± 5.5 (71.9-93.3 )

83.6±5.6 (67.6-89.7 )

82.1 ± 5.2 (69.6-94.8)

80.5±6.9 (63.3-90.7 )

Gender, n (%) female 92 (59.7) 27 (65.9) 33 (68.8) 8 (42.1) 11 (42.3) 13 (65.0) Education, median IQR (range)

4 IQR: 3 (1-7)

4 IQR:3 (1-6)*

4 IQR:3 (1-7) †

5 IQR:2 (4-6) †

4 IQR: 3

(2-7) † 4 IQR:3 (2-6) *

Baseline MMSE, Mean ± SD (range)

21.2±5.7 (3-30 )

26.6 ± 2.2 (21- 29) ¶

18.2 ± 4.4†,§

(6- 26)# 18.1 ± 7.5†,§

(3-29 ) 20.9 ± 5.2†

(12-30) † 25.3 ± 2.7

(19-30) † Ethnicity, n (%) CAU 151 (98.1) 40 (97.6) 48 (100) 19 (100) 25 (96.2) 19 (95.0) SNP C1236T (12) CC 44 (28.6) 12 (29.3) 12 (25.0) 3 (15.8) 7 (26.9) 10 (50.0) CT 75 (48.7) 20 (48.8) 25 (52.1) 10 (52.6) 14 (53.8) 6 (30.0) TT 35 (22.7) 9 (22.0) 11 (22.9) 6 (31.6) 5 (19.2) 4 (20.0) Allele freq T (%) 0.471 0.463 0.490 0.579 0.462 0.350 SNP G2677T/A (21) GG 46 (29.9) 12 (29.3) 13 (27.1) 4 (21.2) 7 (26.9) 10 (50.0) GT 67 (43.5) 16 (39.0) 24 (50.0) 9 (47.4) 13 (50.0) 5 (25.0) TA 3 (1.9) 2 (4.9) 0 (0.0) 0 (0.0) 1 (3.8) 0 (0.0) TT 38 (24.7) 11 (26.8) 11 (22.9) 6 (31.6) 5 (19.2) 5 (25.0) Allele freq T (%) 0.474 0.488 0.479 0.553 0.462 0.375 Allele freq A (%) 0.010 0.024 0.000 0.000 0.019 0.000 SNP C3435T (26) CC 32 (20.8) 9 (22.0) 5 (10.4) 2 (10.5) 8 (30.8) 8 (40.0) CT 70 (45.5) 18 (43.9) 26 (54.2) 9 (47.4) 10 (38.5) 7 (35.0) TT 52 (33.8) 14 (34.1) 17 (35.4) 8 (42.1) 8 (30.8) 5 (25.0) Allele freq T (%) 0.565 0.561 0.625 0.658 0.500 0.425 Haplotype Total POPa


AD (n=48)

VaD (n=19)

OD (n=24)

MCI (n=19)

T-T-T 130 (0.439) 33 (0.434) 43 (0.448) 20 (0.526) 22 (0.458) 12 (0.316) C-G-C 114 (0.385) 29 (0.382) 34 (0.354) 11(0.289) 21 (0.438) 19 (0.500) C-G-T 38 (0.128) 10 (0.132) 15 (0.156) 5 (0.132) 4 (0.083) 4 (0.105) C-T-T 5 (0.017) 3 (0.039) 1 (0.010) 0 (0.000) 0 (0.000) 1 (0.026) T-G-C 4 (0.014) 1 (0.013) 2 (0.021) 1 (0.026) 0 (0.000) 0 (0.000) T-T-C 5 (0.017) 0 (0.000) 1 (0.001) 1 (0.026) 1 (0.021) 2 (0.053) * missing for 2 patients.† missing for 1 patient.‡ p<0.001 vs. controls. § p<0.001 vs. MCI ¶performed in 22 controls.#missing for 3 patients. Haplotype of the different SNPs, i.e. number of alleles and frequencies of the haplotype mentioned. C1236T-G2677T-C3435T;a 3 patients with an ethnicity different from Caucasian and 3 patients bearing an A allele at the 2677 (exon 21) position were excluded from haplotype analysis POP=Population, AD= Alzheimer’s Disease, VaD= Vascular Dementia, OD= Other dementias, MCI= Mild Cognitive Impairment, SD= Standard Deviation, IQR = Inter Quartile Range, MMSE = Mini Mental State Examination, CAU= Caucasian race, SNP= Single Nucleotide Polymorphism.


145

G2677T/A and C3435T (ρ2=0.456, p<0.000001). In the table the ABCB1 genotype, allele and haplotype frequencies are presented. Haplotype data were inferred from genotype data only for Caucasian participants not possessing an A allele at position G2677T/A in exon 21. No statistical differences were observed for genotype data, allele frequencies and haplotype data between patients with dementia and age-matched controls, nor between patients with different types of dementia and age-matched controls. The logistic regression analyses did not show the APOE ε4 allele as confounder for the ABCB1 genotypes as possible risk factors for dementia (results not shown). Frequencies of ABCB1 genotypes of the SNPs C1236T, G2677T/A, C3435T in this elderly population are comparable to earlier reports on younger populations20,25,26. We did not find a relation between ABCB1 SNPs and different types of dementia. Whether ABCB1 SNPs and haplotypes result in different function of P-gp at the BBB is not clear. In a study in 10 healthy volunteers who were homozygous for the TTT haplotype and in 10 healthy volunteers who where homozygous for the CGC haplotype, no differences in 11C-verapamil kinetics, as measured by Positron Emission Tomography, were apparent27. This could point out that ABCB1 SNPs and/or haplotypes are not related to P-gp function at the BBB. This first study on ABCB1 genotypes in dementia has 27% power to detect differences in C3435T genotypes between AD and control patients. Based upon our preliminary results, 173 patients should be included in both the AD and the control group to obtain an ideally 80% power. This study and possible future ones may be combined in a meta-analysis to achieve more power to detect differences in ABCB1 genotypes between the different groups. In conclusion, our study suggests that frequencies of ABCB1 genotypes and haplotypes are not significantly different between demented patients and age-matched control subjects and are presently not useful as biomarker for (different types of) dementia.

List of abbreviations used ABCB1: ATP-Binding Cassette Subfamily B member 1; APOE: Apolipoprotein E

Acknowledgements The authors want to thank Markus Joerger of the department of Pharmacy & Pharmacology, Slotervaart Hospital, for his help with inferring haplotypes from the genotypes. Bregje Appels of the department of Medical Psychology, Slotervaart Hospital, is kindly acknowledged for her assistance in coding the education level of all patients. Ninja Antonini of the department of Biometrics, Antoni van Leeuwenhoek Hospital/Netherlands Cancer Institute, is kindly acknowledged for performing the power calculation for this study.

Chapter 4.1.2

146

References 1. Schinkel AH, Jonker JW. Mammalian drug efflux transporters of the ATP binding cassette (ABC) family: an

overview. Adv Drug Del Rev 2003;55:3-29. 2. Borst P, Oude Elferink R. Mammalian ABC transporters in health and disease. Ann Rev Biochem 2002;

71:537-92. 3. Lee G, Bendayan R. Functional expression and localization of P-glycoprotein in the central nervous system:

relevance to the pathogenesis and treatment of neurological disorders. Pharm Res 2004; 21:1313-30. 4. Marzolini C, Paus E, Buclin T, Kim RB. Polymorphisms in human MDR1 (P-glycoprotein): Recent

advances and clinical relevance. Clin Pharmacol Ther 2004; 75:13-33. 5. Bosch TM, Meijerman I, Beijnen JH, Schellens JH. Genetic polymorphisms of drug-metabolizing enzymes

and drug transporters in the chemotherapeutic treatment of cancer. Clin Pharmacokinet 2006; 45:253-85. 6. Hardy J, Selkoe D. The amyloid hypothesis of Alzheimer’s disease: progress and problems on the road to

therapeutics. Science 2002; 297:353-356. 7. Walsh D, Klyubin I, Fadeeva J, Cullen WK, Anwyl R, Wolfe MS, Rowan MJ, Selkoe DJ. Naturally secreted

oligomers of beta amyloid potently inhibit hippocampal long-term potentation in vivo. Nature 2002; 416:535-539.

8. Lam FC, Liu R, Lu P, Shapiro AB, Renoir JM, Sharom FJ, Reiner PB. Beta-amyloid efflux mediated by p-glycoprotein. J Neurochem 2001; 76:1121-1128.

9. Cirrito JR, Deane R, Fagan AM, Spinner ML, Parsadanian M, Finn M, Jiang H, Prior JL, Sagare A, Bales KR, Paul SM, Zlokovic BV, Piwnica-Worms D, Holtzman DM. P-glycoprotein deficiency at the blood-brain-barrier increases beta-amyloid deposition in an Alzheimer’s disease mouse model. J Clin Invest 2005; 115:3285-3290.

10. Vogelgesang S, Cascorbi I, Schroeder E, Pahnke J, Kroemer HK, Siegmund W, Kunert-Keil C, Walker LC, Warzok RWL. Deposition of Alzheimer’s beta amyloid is inversely correlated with P-glycoprotein expression in the brains of elderly non-demented humans. Pharmacogenetics 2002; 12:535-41.

11. Zlokovic BV. Clearing amyloid through the blood-brain-barrier. J Neurochem 2004; 89:807-11. 12. Folstein MF, Folstein SE, McHugh PR. Mini-Mental State: a practical method for grading cognitive state of

patients for the clinician. Journal of Psychiatric Research 1975; 12:189-198. 13. Solomon PR, Hirschoff A, Kelly B, Relin M, Brush M, DeVeaux RD, Pendlebury WW. A 7 Minute

Neurocognitive Screening Battery highly sensitive to Alzheimer´s disease. Arch Neurol 1998; 55:349-55. 14. McKahnn G, Drachmann D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical diagnosis of Alzheimer´s

disease: Report of the NINCDS-ADRDA work group under the auspices of department of health and human services task force on Alzheimer´s disease. Neurology 1984; 34:939-944.

15. McKeith IG, Galasko D, Kosaka K, Perry EK, Dickson DW, Hansen LA, Salmon DP, Lowe J, Mirra SS, Byrne EJ, Lennox G, Quinn NP, Edwardson JA, Ince PG, Bergeron C, Burns A, Miller BL, Lovestone S, Collerton D, Jansen EN, Ballard C, de Vos RA, Wilcock GK, Jellinger KA, Perry RH. Consensus guidelines for the clinical and pathological diagnosis of dementia with Lewy Bodies (DLB): report of the consortium on DLB international workshop. Neurology 1996; 47:1113-1124.

16. Erkinjunnti T. Clinical criteria for vascular dementia: the NINDS-AIREN criteria. Dementia 1994; 5:189-92. 17. Neary D, Snowden JS, Gustafson L, Passant U, Stuss D, Black S, Freedman M, Kertesz A, Robert PH,

Albert M, Boone K, Miller BL, Cummings J, Benson DF. Frontotemporal lobar degeneration: a consensus on clinical diagnostic criteria. Neurology 1998; 51:1546-54.

18. Petersen RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, Kokmen E. Mild cognitive impairment: clinical characterization and outcome. Arch Neurol 1999; 56:303-308.

19. Petersen RC, Doody R, Kurz A, Mohs RC, Morris JC, Rabins PV, Rictchie K, Rossor M, Thal L, Winblad B. Current concepts in mild cognitive impairment. Arch Neurol 2001;58:1985-92.


147

20. Bosch TM, Doodeman VD, Smits PH, Meijerman I, Schellens JH, Beijnen JH. Pharmacogenetic screening for polymorphisms in drug-metabolizing enzymes and drug transporters in a Dutch population. Mol Diagn Ther 2006; 10:175-85.

21. Koch W, Ehrenhaft A, Griesser K, Pfeufer A, Muller J, Schomig A, Kastrati A. TaqMan systems for genotyping of disease-related polymorphisms present in the gene encoding apolipoprotein E. Clin Chem Lab Med 2002;40:1123-31.

22. GOLD homepage [http://www.sph.umich.edu/csg/abecasis/GOLD]. 23. HPLUS [http://qge.fhcrc.org/hplus/]. 24. Verhage F. Intelligence and age. Van Gorcum, Assen, 1964 (in Dutch). 25. Kroetz DL, Pauli-Magnus C, Hodges LM, Huang CC, Kawamoto M, Johns SJ, Stryke D, Ferrin TE,

DeYoung J, Taylor T, Carlson EJ, Herskowitz I, Giacomini KM, Clark AG. Pharmacogenetics of membrane transporters investigators: Sequence diversity and haplotype structure in the human ABCB1 (MDR1, multi drug resistance transporter) gene. Pharmacogenetics 2003;13:481-494.

26. Kim RB, Leake BF, Choo EF, Dresser GK, Kubba SV, Schwarz UI, Taylor A, Xie HG, McKinsey J, Zhou S, Lan LB, Schuetz JD, Schuetz EG, Wilkinson GR. Identification of functionally variant MDR1 alleles among European Americans and African Americans. Clin Pharmacol Ther 2001; 70:189-99.

27. Brunner M, Langer O, Sunder-Plassman R, Dobrozemsky G, Muller U, Wadsak W, Krcal A, Karch R, Mannhalter C, Dudczak R, Kletter K, Steiner I, Baumgartner C, Muller M. Influence of functional haplotypes in the drug transporter gene ABCB1 on central nervous system drug distribution in humans. Clin Pharmacol Ther 2005; 78:182-90.

148

149

CHAPTER 4.2

PROTEOMIC BIOMARKERS

150

151

CHAPTER 4.2.1 Amyloid beta protein and tau in cerebrospinal fluid and plasma as biomarkers for dementia:

a review of recent literature

S.V. Frankfort, C.R. Tulner, J.P.C.M. van Campen, M.M. Verbeek, R.W.M.M. Jansen, J.H. Beijnen

Abstract

This review addresses recent developments in amyloid β (Aβ), total tau (t-tau) and phosporylated tau (p-tau) of Aβ42 and t-tau in CSF can discriminate between patients with stable MCI and patients with progressive MCI into AD or other types of dementia with a sufficient sensitivity and protein analysis, in cerebrospinal fluid (CSF) and plasma as biomarkers for dementia. Recent research focused on the progression of patients with mild cognitive impairment (MCI) into dementia and the differential diagnosis of Alzheimer´s Disease (AD). A combination of Aβ42 and t-tau in CSF can discriminate between patients with stable MCI and patients with progressive MCI into AD or other types of dementia with a sufficient sensitivity and specificity. Regression analyses demonstrated that pathological CSF (with decreased Aβ42 and increased tau levels) is a very strong predictor for the progression of MCI into AD. Furthermore, CSF measurements of p-tau and Aβ42 can assist in diagnosing vascular dementia or frontotemporal dementia in the differential diagnosis of AD indicated by a reasonable sensitivity and specificity. Whether tau in combination with Aβ42 or in combination with the Aβ37/Aβ42 or Aβ38/Aβ42 ratio aids in the discrimination between AD and Lewy Body dementia remains to be elucidated. Cross-sectional research could not demonstrate significant differences for Aβ40 and Aβ42 in plasma between AD and controls. However, a recently published longitudinal study showed high baseline Aβ40 levels, especially when combined with low baseline Aβ42 levels, as a strong risk factor for the development of dementia. This emphasises the importance of performing longitudinal studies in addition to cross-sectional ones. The origin of plasma Aβ and its transport between CSF and plasma, however, needs further clarification. In conclusion, progress has been made regarding Aβ and tau as biomarkers for dementia both for differentiation between stable MCI and progressive MCI patients and for the differential diagnosis of AD. Future research should aim to validate these recently published results, preferably in pathologically confirmed AD patients. In addition, it is important to standardise research in terms of study design (longitudinal, minimal follow-up period of 5 years), type of researched parameters (total or p-tau, type of Aβ peptides), type of matrix (CSF and plasma) and data analysis (establishment of predefined cut-off values, type of ratio, type of marker combination).

Submitted

Chapter 4.2.1

152

Introduction Dementia is a rapidly growing health and socioeconomic problem worldwide, as it is estimated that 24 million people have dementia nowadays and that this will increase to 42 million by 2020 and 81 million by 20401. Alzheimer’s Disease (AD) is the most common dementia, accounting for 50-60% of cases2 and Vascular dementia (VaD) is the second in the developed world3. Lewy Body Dementia (LBD), that was initially thought to be a rare disease is currently recognised as another important degenerative dementia accounting for 10-15% of cases at autopsy4. Frontotemporal Dementia (FTD) is the most common type of dementia associated with frontotemporal lobar degeneration (FTLD)5. The clinical diagnosis of dementia is preceded by a period of Mild Cognitive Impairment (MCI). Only a subset of patients suffering from MCI, however, progresses to dementia6, at a rate of approximately 15% of the patients per year7. It is known that the pathogenesis of dementia starts years before the clinical onset of dementia8. With the advent of new future therapies, e.g. secretase inhibitors9, disease modifying therapy could then be initiated before the clinical onset of the disease. Thus, it is very important if a diagnosis can be made at a very early moment in the dementia process, for example by biomarker diagnostics in body fluids. Currently, most biomarker research has been executed in cerebrospinal fluid (CSF)7. In addition, several research groups investigated biomarkers in plasma10. The most frequently studied markers are amyloid beta (Aβ) peptides and tau, as phosphorylated tau or total tau, as these markers are related to the two hallmarks of the pathogenesis in AD, the senile plaques and neurofibrillary tangles2. In the past, several reviews7,10-13 have been published about biomarker research in dementia. Therefore, this review only addresses the recent developments into Aβ and tau as biomarkers for dementia. We focused on Aβ and tau (in CSF or plasma) as biomarkers for progression of Mild Cognitive Impairment into dementia and biomarker research applied in the differential diagnosis of AD. Our search strategy was limited to a PubMed search of published articles between 2004 and September 2006 in English language and used keywords were “amyloid beta” or “tau” and either of the following terms: “Alzheimer’s disease” “Lewy Body Disease”, “Vascular Dementia”, “Frontotemporal Dementia” and “Mild Cognitive Impairment”.

Molecular background The “amyloid hypothesis” assumes that accumulation of Aβ peptides in the brain is the key event in the pathogenesis of AD14. Aβ deposits in plaques in brain parenchyma and along the vascular system. Soluble Aβ oligomers may eventually lead to synaptic dysfunction in the brains of AD patients15. Amyloid precursor protein (APP) and the amyloidogenic pathway resulting in formation of Aβ peptides is depicted in figure 1. APP has a large

Amyloid beta protein and tau in cerebrospinal fluid and plasma as biomarkers for dementia

153

extracellular and a smaller intracellular domain. It is degraded by β-secretases and γ-secretases into Aβ peptides2. C-terminally truncated Aß peptides can be generated by γ-secretase that has multiple potential cleavage sites16-18 represented as arrows in figure 1 resulting in the formation of Aβ1-37, Aβ1-38, Aβ1-39, Aβ1-40 and Aβ1-42. Figure 1. Amyloid Precursor Protein and the amyloidogenic secretase pathway resulting in formation of amyloid beta (Aβ) peptides. The letters in the Aβ sequence refer to the amino acid composition. Tangles that are composed of hyperphosphorylated tau protein are the second hallmark in the pathogenesis of Alzheimer´s dementia. Tau is a normal protein in axons that binds to microtubules and thereby promotes assembly and stability of microtubules. There are six different isoforms ranging in size from 352 to 441 amino acids and tau has numerous potential phosphorylation sites11. Phosphorylation of tau starts intracellularly and is regulated by the balance between kinases and phosphatases and eventually leads to disassembly of microtubules and impaired axonal transport thereby compromising neuronal function2.

Amyloid β and tau measurements in CSF or plasma CSF is in direct contact with the extracellular space of the brain and cerebral processes can be reflected in CSF7. In CSF, total tau can be measured using ELISA with antibodies that detect all isoforms of tau proteins independent of phosphorylation or specific phosphorylation sites19. The total concentration of tau protein in CSF probably indicates the intensity of the neuronal damage and degeneration. Tau phosphorylated at specific sites (p-tau) can be measured, for example at positions threonine 181, serine 199, threonine 231, serine 235, serine 396 and serine 4047. Aβ peptides can be measured in CSF and plasma. As CSF is in direct contact with the extracellular space of the brain, changes in Aβ concentration in CSF may better reflect

AβExtracellular Intracellular

Amyloid Precursor Protein

β-secretase γ-secretase

D A E F R H D S G Y I AE V H H Q K L V F F A E D V G S N K G A I I G L M V G G V V

1 37 38 39 40 42

Tab

le 1

. Tau

and

am

yloi

d be

ta a

s bio

mar

kers

in c

ereb

rosp

inal

flui

d: re

sear

ch fo

cuse

d on

Mild

Cog

nitiv

e Im

pairm

ent (

MC

I)

Des

ign

Gro

ups

Stat

istic

al c

ompa

riso

ns b

etw

een

grou

ps

Cut

of

f va

lue

(pg/

mL

)

Sens

itivi

ty (%

)

Spec

ifici

ty (%

) R

efer

ence

Long

itudi

nal

Follo

w-u

p 19

mon

ths

26 sM

CI

33 p

MC

I

Tau:

pM

CI >

sMC

I (p<

0.05

) Aβ4

2: p

MC

I < sM

CI (

p<0.

01)

NA

N

A

NA

B

ouw

man

et

al.

22

Long

itudi

nal

55 sM

CI

23 p

MC

I 46

CO

N

Tau:

pM

CI >

sMC

I(p=

0.00

11)

p-ta

u181

: pM

CI >

sMC

I(p=

0.00

03)

Aβ4

2: p

MC

I < sM

CI (

p=0.

0007

)

Aβ4

2: 4

52

p-ta

u 181

: 70

Aβ4

2/ p

-tau 1

81:

61 (p

MC

I)

Aβ4

2/ p

-tau 1

81:

91 (C

ON

) 87

(sM

CI)

Her

ukka

et

al.

23

Long

itudi

nal

Follo

w-u

p m

edia

n 5.

2 ye

ars

56 sM

CI

57 M

CI

AD

21

MC

IO

D

39 C

ON

Tau

/p-ta

u181

: M

CI

AD

> sM

CI (

p<0.

0001

);

MC

IA

D >

MC

IO

D (p

<0.0

01)

Aβ4

2: M

CI

AD

< sM

CI (

p<0.

0001

);

M

CI

AD

< M

CI

OD

(p<0

.001

);

M

CI

OD

< C

ON

(p=0

.002

)

Aβ4

2: 3

50

Tau:

530

Aβ4

2 an

d t-t

au:

95 (M

CI

AD

) Aβ4

2 an

d t-t

au:

92 (C

ON

) 83

(con

vers

ion

AD

)

Han

sson

et

al.

24

Long

itudi

nal

Mea

n fo

llow

up

3.94

ye

ars

46 sM

CI

33 p

MC

I 60

CO

N

Aβ4

2: p

MC

I < sM

CI(

p<0.

0001

) Ta

u: p

MC

I > sM

CI(

p<0.

0001

) p-

tau:

pM

CI >

sMC

I(p<

0.00

01)

NA

N

A

NA

H

eruk

ka

et a

l. 25

Long

itudi

nal

Mea

n fo

llow

up

8.4

mon

ths

23 sM

CI

29 p

MC

I 93

A

D,

10

CO

N

Aβ4

2: p

MC

I < sM

CI (

p=0.

071)

;

AD

< M

CI (

p<0.

01);

Ta

u: p

MC

I vs s

MC

I (N

S);

A

D >

MC

I (p<

0.02

5);

Aβ4

2: 6

79

Tau:

479

Aβ4

2: 8

3

(pM

CI v

s sM

CI)

ta

u: 9

0

(pM

CI v

s sM

CI)

Aβ4

2:57

(p

MC

I vs s

MC

I)

tau:

48

(pM

CI v

s sM

CI)

Ham

pel

et a

l. 26

sMC

I=st

able

Mild

Cog

nitiv

e Im

pairm

ent,

pMC

I= p

rogr

essi

ve M

ild C

ogni

tive

Impa

irmen

t, C

ON

= co

ntro

ls, A

D=A

lzhe

imer

’s D

isea

se, M

CI

AD

= M

ild C

ogni

tive

Impa

irmen

t pat

ient

s pr

ogre

ssed

to

Alz

heim

er’s

Dis

ease

, MC

I O

D=

Mild

Cog

nitiv

e Im

pairm

ent p

atie

nts p

rogr

esse

d to

oth

er d

emen

tias


155

brain processes7. However, Zlokovic20 developed a transport-clearance model for Aβ regulations of the brain and for transport of plasma-derived Aβ across the blood-brain-barrier (BBB). Transport of Aβ at the BBB is regulated by the Receptor for Advanced Glycation Endproducts (RAGE) that transports Aβ from blood into the central nervous system and by Low-density Receptor-related Protein (LRP) that is responsible for transport from brain interstitial fluid to blood. Studies in animal models found that Aβ was significantly transported from blood into the brain and a study in AD patients suggested that circulating Aβ could be a precursor of brain Aβ20. It remains, however, the question where the Aβ-peptides in plasma originate from. Plasma Aβ can be transported from the brain or can be produced in platelets. Platelets have the highest level of APP expression among peripheral tissue and are able to produce all APP fragments found in neurons. Thus platelets can process APP via the amyloidogenic pathway (involving β-secretase and γ-secretase) resulting in Aβ formation21.

Recent biological marker studies in CSF It is important to perform longitudinal studies in MCI patients with a follow-up time that is sufficient for diagnosing the incipient dementia cases. Around 15% of MCI patients progress to dementia each year. A follow-up time should ideally last for more than 5 years to ascertain that control patients will not progress into dementia7. Longitudinal studies in MCI patients (table 1) showed that it is possible to differentiate between progressive MCI (pMCI) (to AD) and both stable MCI (sMCI) and control patients using CSF measurements with rather sufficient sensitivities and specificities. Almost all these studies showed that CSF Aβ42 was significantly decreased and both total tau and p-tau were significantly increased in pMCI versus sMCI and in pMCI versus control patients. These results are consistent with earlier studies in AD11 and can be explained by deposition of Aβ42 in plaques and by axon degeneration and tau hyperphosphorylation and tangle formation. In the study by Hansson et al.24 a Cox regression analysis was performed to investigate the hazard ratio for conversion from MCI to AD. This showed pathological CSF (combination of increased tau and decreased Aβ42 levels) as a very strong risk factor (HR=17.7; p<0.0001) after adjusting for age, gender, education level and APOE genotype. In general, combinations between Aβ42 and tau or p-tau are superior to either of the markers alone in differentiating sMCI patients from pMCI patients. However, the described studies included relatively few patients and different types of markers have been measured combined with variable follow-up periods (mean or median follow-up time between 8.4 months and 5.2 years). The results of these studies show progress in the field of CSF diagnostics in MCI patients, but it has to be recognised that important study characteristics have to be optimised. Most ideally patients clinically diagnosed as MCI patients should be followed for a period of at least 5 years. Follow-up to other types of dementia besides AD

Tab

le 2

. Tau

and

am

yloi

d be

ta a

s bio

mar

kers

in c

ereb

rosp

inal

flui

d: c

ross

-sec

tiona

l res

earc

h fo

cuse

d on

diff

eren

tiatin

g be

twee

n de

men

tias.

Gro

ups

Stat

istic

al c

ompa

riso

ns

betw

een

grou

ps

Cut

-off

val

ue

(pg/

mL

) Se

nsiti

vity

(%)

Sp

ecifi

city

(%)

Ref

eren

ce

FTD

73 F

TD

17 A

D

13 C

ON

Tau:

FTD

<AD

(p<0

.000

1)

Aβ4

2: A

D<F

TD (p

<0.0

4)

Aβ4

2: 5

5.2

Tau:

275

.8

Tau:

74

(FTD

vs A

D)

Aβ4

2: 3

7 (F

TD v

s AD

) Ta

u: 8

2 (F

TD v

s AD

) Aβ4

2: 5

9 (F

TD v

s AD

) G

ross

man

et

al.27

22 F

TLD

47

AD

(ear

ly)

21 C

ON

Tau:

FTD

<AD

(p<0

.001

) p-

tau 1

81: F

TD<A

D (p

<0.0

01);

Aβ4

2: A

D<F

TD (p

<0.0

01)

Aβ4

2: 4

13

Tau:

377

p-

tau 1

81: 5

4

Aβ4

2: 8

5 (A

D v

s FTL

D)

Tau:

85

(AD

vs F

TLD

) p-

tau 1

81:8

5 (A

D v

s FTL

D)

Aβ4

2:75

(AD

vs F

TLD

) Ta

u:74

(AD

vs F

TLD

) p-

tau 1

81:8

2 (A

D v

s FTL

D)

Scho

onen

- bo

om

et a

l.28

35 F

LTD

51

AD

27

CO

N

Tau:

FTD

<AD

(p=0

.002

) Aβ4

2: A

D<F

TD (p

<0.0

01)

Aβ4

2: 3

15

Tau:

908

Aβ4

2: 8

6 (F

TLD

vs A

D)

Tau:

86

(FTL

D v

s AD

) Aβ4

2: 5

9 (F

TLD

vs A

D)

Tau:

26

(FTL

D v

s AD

) Pi

jnen

burg

et

al.29

VaD

20 V

aD,

31 A

D+W

MC

35

AD

24

ON

D

Aβ4

2: A

D<V

aD (p

<0.0

001)

;

AD

+WM

C<V

aD (p

<0.0

001)

; p-

tau 1

81: A

D>V

aD (p

<0.0

5);

AD

+WM

C>V

aD (p

<0.0

5);

Aβ4

2: 4

93

Aβ4

2: 7

50

Aβ4

2: 7

7 (A

D v

s. V

aD)

Aβ4

2:>9

0 (A

D+W

MC

vs.

VaD

) Aβ4

2: 9

6 (A

D/A

D+W

MC

vs V

aD)

Aβ4

2: 8

0 (A

D v

s. V

aD)

Aβ4

2: 6

0 (A

D+W

MC

vs.

VaD

) Aβ4

2:

60

(AD

/AD

+WM

C

vs

VaD

)

Stef

ani

et a

l.30

38 V

aD

39 A

D

35 C

ON

Aβ4

2: A

D<V

aD (p

<0.0

1);

p-ta

u 181

:AD

>VaD

(p<0

.01)

N

A

NA

N

A

Jia

et a

l.31

25 V

aD

61 A

D

30 C

ON

Aβ4

2: A

D<V

aD (p

<0.0

01);

Pt

au18

1: A

D>V

aD (p

<0.0

01)

Aβ4

2/p 1

81ta

u: A

D<V

aD (p

<0.0

01);

Aβ4

2/pt

au18

1: 12

.7

Aβ4

2/pt

au18

1:100

(AD

vs V

aD))

Aβ4

2/pt

au18

1:85

(AD

vs V

aD)

De

Jong

et

al.32

Tab

le 2

. Ta

u an

d am

yloi

d be

ta a

s bi

omar

kers

in

cere

bros

pina

l flu

id:

cros

s-se

ctio

nal

rese

arch

foc

used

on

diff

eren

tiatin

g be

twee

n de

men

tias

(con

tinue

d).

Gro

ups

Stat

istic

al c

ompa

riso

ns

betw

een

grou

ps

Cut

-off

val

ue

(pg/

mL

) Se

nsiti

vity

(%)

Sp

ecifi

city

(%)

Ref

eren

ce

LB

D

71

LB

D

67 A

D

41 C

ON

Tau:

AD

>LB

D (p

=0.0

03)

NA

N

A

NA

M

olle

nhau

er

et a

l.33

20 L

BD

34

AD

20

CO

N

Tau:

AD

>LB

D (p

=0.0

02)

NA

N

A

NA

M

olle

nhau

er

et a

l.34

21 L

BD

21

PD

D

23 A

D

23 C

ON

Aβ4

2/ Aβ3

7: A

D<L

BD

(p=3

.2*1

0-3);

AD

<PD

D (p

=1.1

*10-4

) Aβ4

2/ Aβ3

7:

0.65

9 0.

772

Aβ4

2/ Aβ3

7:

74 (A

D v

s LB

D)

83 (A

D v

s PD

D)

Aβ4

2/ Aβ3

7:

71 (A

D v

s LB

D)

76 (A

D v

s PD

D)

Bib

l et a

l.35

25 L

BD

18

AD

14

CO

N

Aβ4

2/ Aβ3

7: A

D<L

BD

(p=6

.6*1

0-6)

Aβ4

2/ Aβ3

8: A

D<L

BD

(p=7

*10-6

) Aβ4

2/ Aβ4

0: A

D<L

BD

(p=7

.4*1

0-5)

Tau:

AD

> LB

D (4

*10-7

)

Aβ4

2/ Aβ3

7/ta

u:

1.70

9

Aβ4

2/ Aβ3

7/ta

u:

100

(AD

vs L

BD

+ C

ON

) Aβ4

2/ Aβ3

7/ta

u:

100

(AD

vs L

BD

)

Aβ4

2/ Aβ3

7/ta

u:

92 (A

D v

s LB

D+

CO

N)

Aβ4

2/ Aβ3

7/ta

u:

92 (A

D v

s LB

D)

Bib

l et a

l.36

VA

RIO

US

10

9 A

D

25 M

CI

15 F

TD

41 V

aD

58 C

ON

Aβ-

ratio

: AD

<VaD

(p<0

.05)

Ta

u: 3

89

p-ta

u 181

: 62.

5 Aβ4

2: 6

12

Aβ-

ratio

: 1.1

5

Tau:

68

(AD

vs F

TD+V

aD)

p-ta

u181

:66

(AD

vs F

TD+V

aD)

Aβ4

2:82

(AD

vs F

TD+V

aD)

Aβ-

ratio

:93

(AD

vs F

TD+V

aD)

Tau:

65 (A

D v

s FTD

+VaD

) p-

tau1

81:7

2 (A

D v

s FTD

+VaD

) Aβ4

2:46

(AD

vs F

TD+V

aD)

Aβ-

ratio

:44

(AD

vs F

TD+V

aD)

Bre

tt-sc

hnei

der

et a

l.37

Tab

le 2

. Ta

u an

d am

yloi

d be

ta a

s bi

omar

kers

in

cere

bros

pina

l flu

id:

cros

s-se

ctio

nal

rese

arch

foc

used

on

diff

eren

tiatin

g be

twee

n de

men

tias

(con

tinue

d).

Gro

ups

Stat

istic

al c

ompa

riso

ns

betw

een

grou

ps

Cut

-off

val

ue

(pg/

mL

) Se

nsiti

vity

(%)

Sp

ecifi

city

(%)

Ref

eren

ce

VA

RIO

US

27

CO

N

23 A

D

5 FT

LD

10 A

RC

D

p-ta

u181

/ Aβ4

2:

AD

>FTL

D (p

=0.0

04)

AD

>AR

CD

(p<0

.001

)

Not

men

tione

d p1

81ta

u/ Aβ4

2 86

(AD

vs F

TLD

) 91

(AD

vs A

RC

D)

p181

tau/

Aβ4

2 91

(AD

vs F

TLD

) 10

0 (A

D v

s AR

CD

)

Bla

sko

et

al.38

76 A

D

39 C

ON

48

non

-AD

(1

7 FT

D, 4

PP

A, 6

LB

D,

13 u

nspe

c,

8 ra

re d

em)

Tau:

non

-AD

<AD

(p<0

.001

) p-

tau1

81: n

on-A

D<A

D (p

<0.0

01)

Aβ4

2: n

on-A

D>A

D (p

=0.0

1)

p181

tau/

Aβ4

2: n

on-A

D<A

D

(p<0

.001

) t-t

au/ A

β42:

non

-AD

<AD

(p<0

.001

)

tau:

400

p-

tau 1

81:6

9 Aβ4

2: 5

40

ptau

181/A

β42:

0.

130

tau/

Aβ4

2: 0

.78

Tau:

72 (n

on-A

D v

s AD

) p-

tau 1

81:7

4 (n

on-A

D v

s AD

) Aβ4

2:71

(non

-AD

vs A

D)

ptau

181/

Aβ4

2:78

(non

-AD

vs A

D)

tau/

Aβ4

2:75

(non

-AD

vs A

D)

Tau:

71 (n

on-A

D v

s AD

) p-

tau1

81:7

4 (n

on-A

D v

s AD

) Aβ4

2:57

(non

-AD

vs A

D)

p181

tau/

Aβ4

2:74

(n

on-A

D v

s AD

) t-t

au/ A

β42:

74

(non

-AD

vs A

D)

Ibac

h

et a

l.39

148

CO

N

201

AD

33

AD

+CV

D

27 F

TD

22 L

BD

Aβ4

2: F

TD>A

D (p

<0.0

5);

L

BD

>AD

(p<0

.01)

Ta

u: F

TD<A

D (p

<0.0

1);

LB

D<A

D (p

<0.0

01)

p-ta

u181

: AD

+CV

D<A

D (p

<0.0

5);

F

TD<A

D (p

<0.0

5);

L

BD

<AD

(p<0

.01)

NA

N

A

NA

En

gelb

orgh

s et

al.40

AD

=Alz

heim

er’s

Dis

ease

, M

CI=

Mild

Cog

nitiv

e Im

pairm

ent,

FTD

=Fro

ntot

empo

ral

Dem

entia

, V

aD=V

ascu

lar

Dem

entia

, C

ON

=con

trol

patie

nts,

FTLD

= Fr

onto

tem

pora

l Lo

be D

egen

erat

ion,

A

RC

D=A

lcoh

ol R

elat

ed C

ogni

tive

Dis

orde

rs, n

on-A

D=

Dem

entia

not

of

the

Alz

heim

er ty

pe, P

PA=p

rimar

y pr

ogre

ssiv

e ap

hasy

, LB

D=L

ewy

Bod

y D

isea

se, u

nspe

c=un

spec

ified

type

of

dem

entia

, rar

e de

m=r

are

type

of d

emen

tia, A

D+C

VD

=Alz

heim

er’s

dis

ease

with

con

com

itant

cer

ebro

vasc

ular

dis

ease

, AD

+WM

C=

Alz

heim

er’s

Dis

ease

and

Whi

te M

atte

r Cha

nges

, ON

D=O

ther

neu

rolo

gica

l dis

ease

s,

PDD

=Par

kins

on’s

Dis

ease

Dem

entia

.


159

should also be incorporated in these studies and CSF measurements and data analysis should be standardised. Progress has been made into biomarkers for the differential diagnosis of AD (table 2). A combination of p-tau and Aβ42 serves an important role, especially for the differentiation of AD from VaD or FT(L)D. In the study by de Jong et al.32 the ratio of Aβ42 and ptau181 showed an optimal sensitivity (100%) and specificity (95%) for the differentiation of VaD and AD. In the study by Blasko et al.38, for example, the most effective separation of AD from FTLD and alcohol related dementia was obtained with the ratio of p-tau and Aβ42, resulting in sensitivities and specificities equal or higher than 80%. In the study by Ibach et al.39 non-AD dementias could be best discriminated from AD using the ratio of p-tau/Aβ42 resulting in a sensitivity of 78% and a specificity of 74%. Future research should incorporate longitudinal studies with longer clinical follow-up of patients progressing to different types of dementia and results from small studies may be combined when study design and markers that are measured have been standardised. Recent biologic marker studies in plasma If a biomarker in plasma meets the criteria of the “working group” consensus report (1998)41, this would be an enormous advantage, as a venapuncture is evidently less invasive than lumbar puncture. A number of studies investigated Aβ in plasma in MCI, dementia and control patients (table 3). Most cross-sectional studies did not reveal significant differences in plasma Aβ levels between AD and controls which is concordant with studies published before 200410. However, in a recent study by van Oijen et al.45 including large numbers of participants, high baseline plasma Aβ40, especially when in combination with low baseline plasma Aβ42, was indicated as a risk factor for dementia. An earlier longitudinal study by Mayeux et al.47, that included less participants, showed high baseline plasma Aβ40 and high baseline plasma Aβ42, in individuals who developed dementia within 5 years. After correcting for age, body mass index and Aβ40 levels, only the concentration of Aβ42 remained significantly different between progressors to AD and non-progressors. In the longitudinal study by van Oijen et al.45, plasma was drawn when the participants were included, years before the onset of dementia. It is known that during the disease process peripheral Aβ levels decrease47. This may explain the discrepancies between longitudinal studies and cross-sectional studies and points out the importance of performing longitudinal studies, i.e. performing longer clinical follow-up of patients until progression to dementia, in addition to cross-sectional ones. As earlier studies described correlations between age48, creatinine levels49 and plasma Aβ levels it is very important to adjust for these co-variates in analyses.

Chapter 4.2.1

160

To fully understand the principle of plasma Aβ diagnostics, further insights into the dynamic equilibrium between Aβ in CSF and blood are needed. A cross-sectional study described by Mehta et al.50 did not show congruent results between both Aβ40 and Aβ42 in plasma and CSF. In the future it will be necessary to conduct longitudinal studies investigating CSF and plasma Aβ concomitantly to obtain more insight into the correlation between CSF and plasma Aβ over a certain time period. Aβ measurements in plasma should also be initiated to investigate the relation between plasma Aβ and the risk of progression to dementia. Plasma Aβ biomarkers may become very important in clinical practice. It is necessary, however, that insights into the Aβ balance between brain, CSF and blood are elucidated and that plasma Aβ is significantly different between patients progressing to AD and non-progressors or controls and most ideally also between AD and other types of dementia. Table 3. Biomarker studies performed in plasma. Design Analysis Groups

(N, Type) Result Reference

Cross-sectional ELISA

88 MCI 72 CON

Aβ40 and Aβ42: MCI vs. CON (NS) Aβ42 in women: MCI > CON (P<0.05)

Assini et al.42

Cross-sectional SDS-PAGE/ immunoblot

8 AD 9 CON

Aβ37, Aβ38, Aβ39, Aβ40, Aβ42: AD vs. CON (NS)

Lewczuk et al.43

Cross-sectional ELISA

32 AD 25 CON

Aβ38,Aβ40,Aβ42: AD vs. CON (NS)

Mehta et al.44

Longitudinal Case-cohort Mean follow up 8.6 years

ELISA

392 dementia 289 AD 54 VaD 17 PDD 32 other 1594 controls

Baseline Aβ40: developed dementia > free of dementia Ratio Aβ42/ Aβ40: developed dementia < free of dementia

Oijen et al.45

Cross sectional ELISA

146 AD 89 MCI 89 CON

Aβ42: AD < MCI/CON (p<0.001)

Pesaresi et al.46

MCI=Mild Cognitive Impairment, CON=control patients, AD=Alzheimer’s Disease, VaD= Vascular Dementia, PDD=Parkinson’s Disease Dementia

Introduction of other techniques and newly investigated Aβ peptides With Surface Enhanced Laser Desorption Ionisation- Time of Flight Mass Spectrometry (SELDI-TOF MS), a new analytical technique, and gel electrophoresis that were applied

Tab

le 4

. Am

yloi

d be

ta a

s bio

mar

kers

in c

ereb

rosp

inal

flui

d: re

sear

ch fo

cuse

d on

oth

er Aβ

pept

ides

in A

lzhe

imer

’s D

emen

tia (A

D).

Des

ign

Gro

ups

Stat

istic

al c

ompa

riso

ns b

etw

een

grou

ps

Sens

itivi

ty (%

)

Spec

ifici

ty (%

) R

efer

ence

Cro

ss-

sect

iona

l 32

AD

25

CO

N

Aβ3

8/Aβ4

2: A

D >

CO

N (p

=0.0

01)

Aβ4

0/Aβ4

2: A

D>C

ON

(p=0

.001

) Aβ4

2: A

D<C

ON

(p=0

.001

)

NA

N

A

Meh

ta e

t al.44

Cro

ss-

sect

iona

l 30

AD

26

CO

N

Aβ4

2: A

D<C

ON

(p<0

.001

);

Aβ3

8 an

d Aβ4

0: A

D v

s CO

N (N

S)

Aβ4

2/Aβ3

8: A

D<C

ON

(p<0

.001

) Aβ4

2/Aβ4

0: A

D<C

ON

(p<0

.001

)

Aβ4

2: 8

8 Aβ4

2/Aβ4

0: 8

7 Aβ4

2/Aβ3

8: 8

7

Aβ4

2: 8

7 Aβ4

2/Aβ4

0: 8

1 Aβ4

2/Aβ3

8: 7

7

Scho

onen

boom

et

al.51

Cro

ss-

sect

iona

l 10

AD

9

CO

N

Aβ3

7, Aβ3

8 Aβ3

9 Aβ4

0: A

D v

s CO

N (N

S)

Aβ4

2, Aβ3

-44

and

Aβ X

-X: A

D<C

ON

(p<0

.01)

Aβ3

-47:

AD

>CO

N (p

<0.0

5)

NA

N

A

Lew

czuk

et

al.52

Cro

ss-

sect

iona

l 24

AD

24

CO

N

Aβ3

8, Aβ4

0: A

D<C

ON

(p<0

.01)

Aβ3

8/Aβ4

0: A

D>C

ON

(p=0

.01)

N

A

NA

M

adda

lena

et

al.53

AD

=Alz

heim

er’s

Dis

ease

, CO

N=

Con

trol p

atie

nts,

NA

= n

ot a

pplic

able

Chapter 4.2.1

162

complementary to the conventional ELISA method, other Aβ peptides (Aβ37, Aβ38, Aβ39) were quantified in CSF of AD and control patients (table 4), LBD and PDD patients (table 2) and in plasma studies in AD patients (table 3). Recently, also an ELISA directed at Aβ3844,51 was applied in addition to the commercially available ELISA kits for Aβ40 and Aβ42. These Aβ-peptides have been investigated in cross-sectional studies in different types of dementia both in CSF and plasma. In most of the described CSF studies Aβ38 and Aβ40 did not significantly differ between AD and control patients. However, Aβ42 and ratios of Aβ42 and either Aβ40 and Aβ38 were significantly different in CSF. As Schoonenboom et al.51 showed comparable sensitivities and specificities for both Aβ42 and the ratios of Aβ42 and either Aβ40 and Aβ38 it is questionable if these ratios are superior to measurement of Aβ42 alone. However, these ratios have not been applied before in longitudinal MCI patients to assist in diagnosing progression to AD. In addition, the ratio of Aβ peptides alone or in combination with tau could play a more pronounced role in the differential diagnosis of AD. Bibl et al.36 applied the combination of these peptides in the differentiation of AD from LBD. The ratios Aβ42/Aβ37 and Aβ42/Aβ38 showed better sensitivities (83%) compared to Aβ42 alone (48%). However, the ratio Aβ42/Aβ40 showed better sensitivity (100%) but less specificity (68%) in differentiating between AD and LBD compared to the ratios Aβ42/Aβ37 and Aβ42/Aβ38 (sensitivity: 83%; specificity: 84%). When using combinations of the Aβ ratios and tau measurements, ratios of Aβ42/Aβ37/tau, Aβ42/Aβ38/tau and Aβ42/Aβ40/tau showed sensitivities of 100% to distinguish between AD and LBD, however, specificity was slightly increased when using Aβ42/Aβ37/tau or Aβ42/Aβ38/tau (92%) compared to Aβ42/Aβ40/tau (88%). These results demonstrate that the ratios of Aβ peptides were not advantageous in diagnosing AD versus control patients, but the ratios of Aβ peptides alone or in combination with tau may serve a future role in the differential diagnosis of AD. This was shown in the differential diagnosis of LBD and further research may be initiated in FTD or VaD.

Discussion and future perspectives In general, progress has been made in the field of biomarker diagnostics in CSF and plasma. More specifically, in longitudinal studies it was shown that pMCI can be distinguished from sMCI and from control patients based upon Aβ and tau measurements in CSF. Earlier studies showed that CSF Aβ and tau can assist in differentiating AD from control patients. From the results of the longitudinal MCI studies it is clear that the subset of patients that progresses to AD can be predicted by CSF biomarker analysis with rather good sensitivity and specificity. This is a large step forward in the field of early dementia diagnostics.


163

Research into the differentiation of AD from other types of dementia by CSF biomarkers resulted in a number of studies performed in prevalent dementia cases and in a longitudinal study by Hansson et al.24 in MCI patients progressing to other types of dementias besides AD. Earlier reported studies investigated other dementias often as an additional group, but the main focus was on differentiation between AD and controls. The currently described studies aimed to investigate CSF biomarkers for the differential diagnoses of AD. The combination of p-tau and Aβ42 plays an important role and their ratio may have a function in the differential diagnosis of AD. Optimal sensitivity and specificity was reported by De Jong et al.32 for differentiation between AD and VaD and Schoonenboom et al.28 reported sufficient sensitivity and specificity for differentiation between AD and FTD. Besides Aβ42 and Aβ40, other Aβ peptides (Aβ37, Aβ38, Aβ39) can be measured by gel electrophoresis or SELDI-TOF MS. Differentiation between AD and LBD was investigated by Bibl et al.36 with the different ratios Aβ42/Aβ40, Aβ42/Aβ38 and Aβ42/Aβ37 alone or in combination with tau. Although these ratios were not superior to Aβ42 in differentiating AD from controls, they showed to be better discriminators between AD and LBD than Aβ42 alone. In conclusion, the described studies in this review show that progress in CSF biomarkers for the differential diagnosis of AD has been made during recent years. Biomarker analysis in plasma will be an enormous advantage compared to CSF diagnostics as a lumbar puncture is more invasive compared to a venapuncture. A recently published and relatively large longitudinal study by van Oijen et al.45 found that high baseline plasma Aβ40 combined with low baseline plasma Aβ42 is a risk factor for the development of dementia. Although this study revealed interesting new findings, uncertainties regarding plasma diagnostics are evident. The origin of plasma Aβ and its equilibrium with CSF and brain Aβ remains to be further elucidated. If more insights into transport mechanisms and origin of plasma Aβ are obtained, plasma Aβ can possibly serve a future role in dementia diagnosis. Progress has been made during recent years in CSF biomarkers for AD and the differential diagnosis of AD and results show promise for the future. However, this review also showed that design differs largely between studies. Studies executed in different settings showed a diversity in cut-off levels for performing specificity and sensitivity analyses and a number of different ratios have been examined as a possible biomarker. These differences make it difficult to converge the data obtained in these studies. In addition, in case of studies in MCI patients it is necessary to accomplish a follow-up period for more than 5 years to ascertain that control patients do not progress to AD or another type of dementia. A few studies in MCI patients do have an extensive follow-up period, but optimisation of follow-up time is still necessary. Before biomarker diagnostics can be really applied in routine clinical practice certain points should be further investigated or optimised. First, it is of great importance that future

Chapter 4.2.1

164

studies are prospective, longitudinal and coherent in study design, type of measured markers and data analysis. Longitudinal studies in MCI patients in CSF or plasma with a minimal follow-up period of 5 years are superior to cross-sectional studies. When these studies are carried out, it is needed to standardise the sample pre-treatment, the matrix to investigate (CSF and plasma), the type of measurements (type of Aβ-peptide, total tau, p-tau), the data analysis (type of ratio, type of combination, predefined cut-off values), and practical issues, e.g. the length of the clinical follow-up period and investigating CSF and plasma concomitantly at certain timepoints. Results from different studies can then be combined and will give more insights into the performance of the diagnostic markers in the future and the role of plasma biomarkers herein. Second, most studies were performed in research settings and the applicability of markers should also be validated in patients diagnosed in regular clinical practice. Third, patients in the reported studies are clinically diagnosed. Although in specialised centres a 85% sensitivity for AD diagnosis based upon the NINDS-ADRDA criteria is obtained, it is not a definitive diagnosis as made at autopsy41. As neuropathological studies have shown that a high proportion (40-80%) of clinically diagnosed VaD patients have concomitant AD pathology and that there is a large overlap between AD and LBD11 the results as obtained in research in clinically diagnosed patients, should ideally be validated in definitive diagnosed patients in the future. In conclusion, progress has been made in research investigating Aβ and tau as biomarkers for their potential to differentiate AD from controls, to differentiate pMCI patients from sMCI patients and for the differential diagnoses of AD. However, we plea for more standardisation of study design and data analysis in addition to longitudinal investigations to gather more insights into the diagnostic performance of Aβ and tau in CSF and plasma.

References 1. Ferri CP, Prince M, Brayne C, Brodaty H, et al. Global prevalence of dementia: a delphi consensus study.

Lancet 2005;366:2112-17. 2. Blennow K, de Leon MJ, Zetterberg H. Alzheimer´s Disease. Lancet 2006;368:387-403. 3. Martinez-Villa E, Murie-Fernandez M, Perez-Larraya J, Irimia P. Neuroprotection in vascular dementia.

Cerebrovasc Dis 2006;21 (suppl 2):106-117. 4. McKeith I, Mintzer J, Aarsland D, et al. Dementia with Lewy Bodies. Lancet Neurol 2004;3:19-28. 5. Neary D, Snowden J, Mann D. Frontotemporal Dementia. Lancet Neurol 2005;4:771-80. 6. Gauthier S, Reisberg B, Zaudig M, et al. Mild cognitive impairment. Lancet 2006;367:1262-1270. 7. Blennow K. CSF biomarkers for mild cognitive impairment. J Inter Med 2004;256:224-34. 8. Morris JC, Quaid KA, Holtzman DM, et al. Role of biomarkers in studies in presymptomatic Alzheimer´s

disease. Alzheimer´s and Dementia 2005;1:145-151. 9. Siemers ER, Quinn JF, Kaye J, et al. Effects of a gamma-secretase inhibitor in a randomized study of

patients with Alzheimer disease. Neurology 2006;66:602-4. 10. Irizarry MC. Biomarkers of Alzheimer Disease in plasma. NeuroRx 2004;1:226-234. 11. Blennow K and Hampel H. CSF markers for incipient Alzheimer´s Disease. Lancet Neurol 2003;2:605-13.


165

12. Sjögren M, Andreasen N, Blennow K. Advances in the detection of Alzheimer´s Disease-use of cerebrospinal fluid biomarkers. Clin Chim Acta 2003;332:1-10.

13. Teunissen CE, de Vente J, Steinbusch HW, de Bruijn C. Biochemical markers related to Alzheimer’s dementia in serum and cerebrospinal fluid. Neurobiol Aging 2002;23:485-508.

14. Hardy J and Selkoe D. The amyloid hypothesis of Alzheimer’s disease: progress and problems on the road to therapeutics. Science 2002;297:353-356.

15. Walsh D, Klyubin I, Fadeeva J, et al. Naturally secreted oligomers of beta amyloid potently inhibit hippocampal long-term potentation in vivo. Nature 2002;416:535-539.

16. Murphy MP, Hickman LJ, Eckman CB. Gamma-secretase, evidence for multiple proteolytic activities and influence of membrane positioning of substrate on generation of amyloid beta peptides of varying length. J Biol Chem 1999;247:11914-11923.

17. Beher D, Wrigley JD, Owens AP, Shearman MS. Generation of C-terminally truncated amyloid-beta peptides is dependent on gamma-secretase activity. J Neurochem 2002;82:563-575.

18. Fluhrer R, Multhaup G, Schlicksupp A, et al. Identification of beta-secretase activity, which truncates amyloid beta peptide after its presenilin-dependent generation. J Biol Chem 2003;278:5531-5538.

19. Hampel J, Teipel SJ. Total and phosporylated tau proteins: evaluation as core biomarker candidates in frontotemporal dementia. Dement Geriatr Cogn Disord 2004;17:350-354.

20. Zlokovic BV. Clearing amyloid through the blood-brain barrier. J Neurochem 2004;89:807-811. 21. Evin G, Zhu A, Holsinger D et al. Proteolytic processing of the Alzheimer’s Disease amyloid precursor

protein in brain and platelets. J Neurosci Res 2003;74:386-392. 22. Bouwman FH, Schoonenboom SNM, van der Flier WM, et al. CSF biomarkrs and medial temporal lobe

atrophy predict dementia in mild cognitive impairment. Neurobiol Aging 2006; in press. 23. Herukka S, Hallikainen M, Soininen H, et al., CSF Aβ42 and tau or phosphorylated tau and prediction of

progressive mild cognitive impairment. Neurology 2005;64:1294-1297. 24. Hansson O, Zetterberg H, Buchhave P, et al. Association between CSF biomarkers and incipient

Alzheimer´s Disease in patients with mild cognitive impairment: a follow-up study. Lancet Neurol 2006;5:;228-34.

25. Herukka S, Helisalmi S, Hallikainen M, et al. CSF Aβ42 and phosphorylated Tau, APOE ε4 allele and MCI type in progressive MCI. Neurobiol Aging 2006; in press.

26. Hampel H, Teipel SJ, Fuchsberger T, et al. Value of CSF Aβ42 and tau as predictors of Alzheimer´s Disease in patients with mild cognitive impairment. Mol Psychiatry 2004;9:705-710.

27. Grossman M, Farmer J, Leight S, et al. Cerebrospinal fluid profile in frontotemporal dementia and Alzheimer’s disease. Ann Neurol 2005;57:721-729.

28. Schoonenboom NSM, Pijnenburg YAL, Mulder C, et al. Aβ42 and phosphorylated tau in CSF as markers for early-onset Alzheimer’s disease. Neurology 2004;62:1580-1584.

29. Pijnenburg YAL, Schoonenboom NSM, Rosso SM, et al. CSF tau and Aβ42 are not useful in the diagnosis of frontotemporal lobar degeneration. Neurology 2004;62:1649.

30. Stefani A, Bernardini S, Panella M, et al. AD with subcortical white matter lesions and vascular dementia: CSF markers for differential diagnosis. J Neurol Sci 2005;237:83-88.

31. Jia JP, Meng R, Sun YX, et al. Cerebrospinal fluid tau, Aβ42 and inflammatory cytokines in patients with Alzheimer’s disease and vascular dementia. Neurosci Letters 2005;383:12-16.

32. De Jong D, Janssen RWMM, Kremer BPH, Verbeek MM. Cerebrospinal fluid Aβ42/phosphorylated tau ratio discriminates between Alzheimer’s disease and vascular dementia. J Gerontol 2006;61A:755-758.

33. Mollenhauer B, Cepek L, Bibl M, et al. Tau protein, Aβ42 and S-100B protein in cerebrospinal fluid of patients with dementia with lewy bodies. Dement Geriatr Cogn Disord 2005;19:164-170.

Chapter 4.2.1

166

34. Mollenhauer B, Bibl M, Wiltfang J, et al. Total tau protein, phosphorylated tau (181p) protein, Aβ42 and Aβ40 in cerebrospinal fluid of patients with dementia with lewy bodies. Clin Chem Lab Med 2006;44:192-195.

35. Bibl M, Mollenhauer B, Esselmann H, et al. CSF amyloid-β-peptides in Alzheimer’s disease, dementia with Lewy bodies and Parkinson’s dementia. Brain 2006;129:1177-1187.

36. Bibl M, Mollenhauer B, Esselmann H, et al. CSF diagnosis of Alzheimer’s Disease and dementia with Lewy Bodies. J Neural Transm 2006;113:1771-8.

37. Brettschneider J, Petzold A, Schottle D, et al. The neurofilament heavy chain (NfHSMI35) in the cerebrospinal fluid diagnosis of Alzheimer’s disease. Dement Geriatr Cogn Disord 2006;21:291-295.

38. Blasko I, Lederer W, Oberbauer H, et al. Measurement of thirteen biological markers in CSF of patients with Alzheimer’s disease and other dementias. Dement Geriatr Cogn Disord 2006;21:9-15.

39. Ibach B, Binder H, Dragon M, et al. Cerebrospinal fluid tau and β-amyloid in Alzheimer patients, disease controls and an age-matched random sample. Neurobiol Aging 2006;1202-1211.

40. Engelborghs S, Maertens K, Vloeberghs E, et al. Neuropsychological and behavioural correlates of CSF biomarkers in dementia. Neurochem Int 2006;48:286-295.

41. Working group on “Molecular and biochemical markers of Alzheimer’s disease”. Consensus report. Neurobiol Aging 1998;19:109-116.

42. Assini A, Cammarata S, Vitali A, et al. Plasma levels of Aβ42 are increased in women with mild cognitive impairment. Neurology 2004;63:828-831.

43. Lewczuk P, Esselmann H, Bibl M, et al. Electrophoretic separation of amyloid β peptides in plasma. Electrophoresis 2004;25:3336-3343.

44. Mehta PD, Prittila T. Increased cerebrospinal fluid Aβ38/ Aβ42 ratio in Alzheimer´s Disease. Neurodegenerative Dis 2005;2:242-245.

45. Van Oijen M, Hofman A, Soares HD, et al. Plasma Aβ40 and Aβ42 and the risk of dementia: a prospective case-cohort study. Lancet Neurol 2006;5:655-660.

46. Pesaresi M, Lovati C, Bertora P, et al. Plasma levels of Aβ42 in Alzheimer’s disease and mild cognitive impairment. Neurobiol Aging 2006;27:904-905.

47. Mayeux R, Honig LS, Tang MX et al. Plasma Aβ40 and Aβ42 and Alzheimer’s Disease: relation to age, mortality and risk. Neurology 2003;61:1185-90.

48. Fukumoto H, Tennis M, Locascio JJ, et al. Age but not diagnosis is the main predictor of plasma amyloid β-protein levels. Arch Neurol 2003;60:958-964.

49. Arvanitakis Z, Lucas JA, Younkin LH, et al. Serum creatinine levels correlate with plasma amyloid beta protein. Alzheimer Dis Assoc Disord 2002;16:187-190.

50. Mehta PD, Pirtitila T, Mehta SP et al. Plasma and cerebrospinal fluid levels of amyloid β proteins 1-40 and 1-42 in Alzheimer’s Disease. Arch Neurol 2000;57:100-105.

51. Schoonenboom NS, Mulder C, van Kamp GJ, et al. Amyloid β 38, 40 and 42 species in cerebrospinal fluid: more of the same? Ann Neurol 2005;58:139-142.

52. Lewczuk P, Esselmann H, Wolfgang T, et al. Amyloid beta peptides in cerebrospinal fluid as profiled with surface enhanced laser desorption/ionization time-of-flight mass spectrometry: evidence of novel biomarkers in Alzheimer’s disease. Biol Psychiatry 2004;55:524-30.

53. Maddalena AS, Papassotiropoulos A, Gonzalez-Agosti C, et al. Cerebrospinal fluid profile of amyloid beta peptides in patients with Alzheimer’s disease determined by protein biochip technology. Neurodegenerative Dis 2004;1:231-35.

167

168

169

CHAPTER 4.2.2 Cerebrospinal fluid biomarkers found for Alzheimer’s

disease and Vascular Dementia using Surface-Enhanced Laser Desorption Ionisation-Time of Flight Mass

Spectrometry (SELDI-TOF MS)

S.V. Frankfort, M.M. Verbeek, J.Y.M.N. Engwegen, J.P.C.M. van Campen, R.W.M.M. Jansen, J.H. Beijnen

Abstract

Background: Biomarker analysis in cerebrospinal fluid (CSF) may be an attractive tool to diagnose Alzheimer’s disease (AD) before clinical onset of the disease. By using surface enhanced laser desorption/ionisation- time of flight mass spectrometry (SELDI-TOF MS) discriminatory proteins in CSF of patients with AD compared to controls (CON) can be identified. Objective: Our study aimed to verify the results of previous small studies in larger populations of AD and CON, as well as in Vascular Dementia (VaD) and to find possible new discriminatory proteins. Methods: SELDI-TOF MS analysis, using Q10 arrays and 6E10 monoclonal Antibody (mAb) against ß-amyloid (Aβ) peptides covalently coupled to PS20 arrays, was performed in duplicate in 52 CON, 39 AD and 31 VaD subjects. Results: The results of Carrette et al. (2003) and Lewczuk et al. (2004) were confirmed in CSF of AD and CON. In addition, those protein profiles were also found in CSF of VaD. Intensities of the β-2 microglobulin and cystatin C peaks were significantly higher in AD and VaD compared to CON. Peak intensities corresponding to the Aβ fingerprint profile (Aβ1-37, Aβ1-38, Aβ1-39, Aβ1-40, Aβ1-42) were significantly different between the three groups (AD<VaD<CON). New peaks were found, possibly representing oxidised, truncated and aggregated amyloid β peptides. Several of these peaks were significantly different in AD versus VaD and may thus serve as potential biomarkers for AD. Conclusions: We showed protein profiles measured by SELDI-TOF MS in CSF of AD, VaD and CON. Interestingly, different profiles of Aβ aggregates and truncated or oxidised Aβ peptides were observed in CSF of AD, VaD and CON subjects. Future research is needed to obtain further insight into the applicability of those markers in early diagnostics and their role in the pathogenesis of dementia.

Submitted

Chapter 4.2.3

170

Introduction Dementia is a rapidly growing health and socioeconomic problem worldwide, as it is estimated that 24 million people have dementia nowadays and that this will increase to 42 million by 2020 and 81 million by 20401. An accurate and early diagnosis is essential for appropriate support and treatment of dementia patients, as symptomatic drugs are specifically available for Alzheimer’s disease (AD) patients2 and neuroprotective drugs based on altered amyloid β (Aβ) metabolism are being developed3. Biomarker analysis in cerebrospinal fluid (CSF) is one of the possible instruments that can be used in early diagnostics4. More specifically, tau and amyloid β proteins may serve as biomarkers for AD, which can be analysed by Enzyme Linked Immunosorbent Assay (ELISA)4,5. In addition, also “general” biomarkers in the fields of cholesterol homeostasis, oxidative stress and inflammation have been studied6. Besides AD, research focuses also on Vascular Dementia (VaD) that is the second most common cause of dementia in the developed world. Moreover, it is supposed that cerebrovascular disease worsens the development of AD and it is more and more recognized that VaD and AD may co-exist in mixed dementia syndrome7. Results from new proteomic techniques in dementia research possibly add to current knowledge regarding biomarkers. One of these techniques is the Protein Chip® array based Surface Enhanced Laser Desorption/Ionisation in combination with Time of Flight Mass Spectrometry (SELDI-TOF MS). Arrays with different chromatographic surfaces enhance on chip purification and retained proteins are analysed on the same chip resulting in a profile of proteins characterised by mass and charge8. Carrette et al.9 described a study in CSF of 9 patients with AD and 10 controls using strong anion exchange (SAX) arrays in combination with SELDI-TOF MS. This study resulted in 5 discriminatory biomarkers of which four were identified as ß-2-microglobulin, cystatin C, Vaccinia Growth Factor (VGF), a neurosecretory protein, and one unknown polypeptide. In addition, Lewczuk et al.10 described a study in CSF of 10 patients with AD and 9 controls using 6E10 antibody, specific for the N-terminus of ß-amyloid peptides, covalently coupled to PS20 arrays and SELDI-TOF MS. This study resulted in a profile of multiple ß-amyloid peptides. The aims of our study are: 1) to verify the results of both studies in larger populations of patients with AD and controls. 2) To extend these findings to VaD patients. This will provide insight into the potency of the CSF protein profiles to distinguish patients with AD from VaD. 3) To identify new additional discriminatory peaks in our patient groups.

Cerebrospinal fluid biomarkers for Alzheimer’s disease and Vascular Dementia

171

Methods Participants Clinical investigation, including medical history, physical, neurological and psychiatric examination, screening laboratory tests, CT or MRI of the brain, Mini Mental State Examination (MMSE)11, neuropsychological assessment, and routine CSF investigations were performed in each participant. All patients were recruited at the Memory Clinic, Department of Geriatric Medicine, Radboud University Nijmegen Medical Centre (RUNMC). AD was diagnosed according to the NINCDS-ADRDA criteria12 and VaD according to the NINDS/AIREN criteria13. As (neurological) controls (CON) we included subjects over age 50 years who visited our outpatient clinic for various reasons but turned out not to suffer from a neurological disorder. Their CSF had normal leukocyte and erythrocyte counts, normal total protein, bilirubin, hemoglobin, glucose and lactate concentrations, and no oligoclonal IgG bands. In total, 122 participants were included: 52 (CON), 39 AD and 31 VaD. Baseline characteristics are presented in table 1. Gender was not significantly different between the three groups, however mean age was significantly different between AD and CON (p<0.001) and between VaD and CON (p<0.001). Table 1. Baseline characteristics of subjects with Alzheimer’s Disease (AD), Vascular Dementia (VaD) and (neurological) controls (CON). Baseline characteristics

AD VaD CON P value

N 39 31 52 Age, y, mean ± SD (range)

69.1± 7.8 (53-85)

71.8 ± 9.6 (49-87)

61.4± 7.8 (50-80)

AD vs CON <0.001 VaD vs CON <0.001

Male, n (%) 20 (51.3) 21 (67.6) 26 (50.0) 0.250

The study was approved by the institutional review board of the RUNMC. All patients gave informed consent prior to lumbar puncture. Sample collection CSF samples were obtained by lumbar puncture and collected in polypropylene tubes. These samples were centrifuged at 860 g for 5 min to remove cells and other insoluble material. Supernatants were divided into 0.5 ml aliquots, stored at -80 ºC and shortly before analysis these 0.5 ml aliquots were once more divided into smaller aliquots, as SELDI-TOF MS analysis requires only small volumes. SELDI-TOF MS analysis was therefore performed with CSF exposed to 1 freeze-thaw cycle.

Chapter 4.2.3

172

Sample pre-treatment and SELDI-TOF MS analysis First, the method as described by Carrette et al.9 was executed in duplicate in our CSF sample set. Q10 (anion exchange) Protein-Chip arrays (Ciphergen Biosystems Inc., Fremont, CA, USA) were equilibrated for 5 min with 5 µl of binding buffer (100 mM Na-acetate, pH 4.0). Afterwards, the buffer was removed and 2 µl of binding buffer was added to each spot. Crude CSF samples (3 µl) were also added to each spot and incubated for 20 minutes at room temperature in a humidity chamber on a platform shaker at 350 r/m, which was used during all further wash steps. CSF was removed from all spots and these were washed individually with 10 µl of binding buffer for 5 minutes. The arrays were then placed in 10 ml polypropylene tubes and washed twice with the binding buffer for 5 minutes. Finally, the chip was rinsed with distilled water. As energy absorbing matrix a volume of 0.5 µl of Sinapic Acid (SPA) in 50% v/v acetonitrile and 0.5% v/v trifluoroacetic acid (TFA) was added twice to each spot. Protein profiling of each spot was performed using the PBS-IIC Protein Chip Reader (Ciphergen Biosystems). Data were collected between 0 and 15 000 Da. Data collection was optimised, resulting in an average of 65 laser shots per spectrum at laser intensity 154 and detector sensitivity 7. Second, the method as described by Lewczuk et al.10 was performed in duplicate in our CSF sample set by using Ciphergen’s ProteinChip® ß-amyloid Multipeptide Kit. 2 µl of a 6E10 monoclonal antibody (mAb) solution in phosphate buffered saline (PBS) (0.25 mg/ml) was added to each spot of a PS20 Protein Chip array used for detection of patient CSF samples. Additionally, 2 µl of a bovine IgG solution in PBS (0.25 mg/ml) was added on spots used for analysis of negative controls. The arrays were incubated for 2 hours at room temperature in a humidity chamber on a platform shaker at 350 r/m. Afterwards 6E10 antibody or IgG were removed from each spot and the arrays were placed in 10 ml polypropylene tubes containing deactivation buffer (0.5M ethanolamine in PBS, pH 8.0) to block the residual sites on the array. Tubes were agitated on a shaker for 30 minutes at room temperature. The arrays were then washed twice with wash buffer (0.5% Triton-X-100 in PBS) for 10 minutes and once with PBS for 5 minutes at room temperature. The arrays were transferred to a 96-well format bioprocessor (Ciphergen Biosystems) and 50 µl of crude CSF was applied to each well. The bioprocessor was placed on a platform shaker and agitated overnight at 4 ºC. The samples were removed and the arrays were washed in the bioprocessor three times with wash buffer for 15 minutes and three times with PBS for 5 minutes and finally rinsed with 1.0 mM HEPES buffer. As energy absorbing matrix a volume of 0.5 µl of a 20% saturated solution of α-cyano-4-hydroxy-cinnamic acid (CHCA) in 50% v/v acetonitrile and 0.5% v/v TFA was added twice to each spot. Protein profiling of each spot was performed using the PBS-IIC Protein Chip Reader (Ciphergen Biosystems). Data were collected between 0 and 15 000 Da. Data collection was optimized,


173

resulting in an average of 100 laser shots per spectrum at laser intensity 150 and detector sensitivity 7. M/z values for detected proteins in both above described methods were calibrated externally with a standard peptide mixture (Ciphergen Biosystems) containing [Arg8] vasopressin (1084.3 Da), somatostatin (1637.9 Da), dynorphine (2147.5 Da), ACTH (2933.5 Da), insuline ß-chain (bovine) (3495.9 Da) and insulin (human recombinant) (5807.7 Da). Statistics and Bioinformatics Data analysis followed the same protocol for both described methods. Data were analysed with ProteinChip software (version 3.1, Ciphergen Biosystems) and all acquired spectra were compiled and analysed as a whole experiment. Spectra were baseline subtracted and normalised to the total ion current between 1500 and 15 000. The Biomarker Wizard application (BMW) (Ciphergen Biosystems) was used to autodetect m/z peaks with a signal-to-noise ratio of at least 5. Peak clusters were completed with peaks with a signal-to-noise ratio of at least 2 in a 0.3% cluster mass window. Peak data per spectrum were exported to Excel (Microsoft Inc., Redmond, WA, USA) and mean intensities were calculated from duplicate spectra for each participant. Statistical calculations were performed with SPSS for Windows (version 11.0, SPSS Inc., Chicago, IL, USA). The Pearson Chi-square test for categorical data and a one-way ANOVA for continuous data were used to compare patient characteristics. Non-parametric Kruskal-Wallis followed by Mann-Whitney-U testing was performed to compare peak data between groups of patients with Alzheimer’s disease, vascular dementia and (neurological) controls. A p-value of 0.01 or less (to correct for multiple testing) was considered statistically significant.

Results Protein profiling using Q10 arrays We were able to confirm the presence of four discriminatory peaks corresponding to VGF, ß-2-microglobulin (two peaks) and cystatin C at masses (Da) of 4814, 11709, 11916 and 13358, respectively (figure 1). The fifth marker described by Carrette et al.9 at 7691 Da was not explicitly found. The four markers were replicated in AD and CON and, in addition, were also detected in VaD. Unfortunately, we were not able to show a significant difference with VGF as seen in the study by Carrette et al.9 However, we found significantly higher intensities in AD compared to CON for the ß-2-microglobulins and cystatin C (p<0.001). The peak intensities corresponding to the ß-2-microglobulins were also significantly higher in VaD compared to CON in our study (p<0.001). However, the mean peak intensities do not differ between AD and VaD.

Chapter 4.2.3

174

With the method as described by Carrette et al.9 we found additional discriminatory peaks, listed in table 2, in CSF of AD, VaD and CON. Mean peak intensities were significantly higher in AD compared to controls for peaks with molecular masses of 2465 Da, 6818 Da, 6898 Da, 7053 Da. Mean peak intensity was significantly lower in AD compared to CON for the peak at m/z 7255. Mean peak intensities were significantly higher in VaD compared to CON for peaks at m/z 3486, 3910 and 3925. Mean peak intensities were lower in VaD compared to CON for peaks at m/z 5318, 6627 and 7255. Intensities of peaks at masses of 8049 and 8129 Da were significantly higher in AD compared to VaD. Table 2. Results of the SELDI-TOF MS analyses using Q10 arrays. M/z value (Da) Peptide Intensity (mean ± SD)

AD (n=39)

VaD (n=31)

CON (n=52)

P-value

Verification method Carrette et al.

4814 VGF 25.5 ± 9.4 24.1 ± 9.8 27.1 ± 10.8 NS

11709 ß-2-microglobulin 1.6 ± 0.9 1.8 ± 1.4 0.8 ± 0.7 <0.001 (AD vs CON) <0.001 (VaD vs CON)

11916 ß-2-microglobulin 0.5 ± 0.2 0.6 ± 0.4 0.3 ± 0.3 <0.001 (AD vs CON) <0.001 (VaD vs CON)

13358 Cystatin C 1.3 ± 0.6 1.2 ± 0.5 0.9 ± 0.5 <0.001 (AD vs CON) New significant m/z values 2465 18.3 ± 4.8 16.3 ± 4.0 14.9 ± 4.4 <0.001 (AD vs CON) 3486 4.3 ± 2.6 4.8 ± 2.8 2.9 ± 2.0 0.002 (VaD vs CON) 3910 6.9 ± 4.4 9.6 ± 4.3 5.7 ± 4.2 <0.001 (VaD vs CON) 3925 2.5 ± 2.2 3.7 ± 2.2 2.2 ± 2.2 <0.001 (VaD vs CON) 5318 2.5 ± 1.8 1.9 ± 1.2 3.4 ± 2.2 0.001 (VaD vs CON) 6627 9.0 ± 4.2 7.4 ± 3.8 10.3 ± 5.2 0.004 (VaD vs CON) 6818 6.3 ± 2.7 5.5 ± 2.0 4.3 ± 2.4 <0.001 (AD vs CON) 6898 3.0 ± 1.5 2.5 ± 2.2 2.2 ± 1.2 0.005 (AD vs CON) 7053 6.1 ± 3.0 4.9 ± 1.9 4.5 ± 2.5 0.006 (AD vs CON) 7255 7.6 ± 3.1 7.5 ± 3.2 10.5 ± 4.3 0.001 (AD vs CON)

0.001 (VaD vs CON) 8049 3.9 ± 1.3 3.0 ± 1.3 3.9 ± 1.5 0.004 (VaD vs AD) 8129 2.6 ± 1.0 1.9 ± 0.9 2.4 ± 1.1 0.002 (VaD vs AD) AD= Alzheimer’s Disease, VaD=Vascular Dementia, CON=(neurological) controls Protein profiling using 6E10 mAb covalently coupled to PS20 arrays We confirmed the profile of Aß1-37, Aß1-38, Aß1-39, Aß1-40 and Aß1-42, described by Lewczuk et al.10,14 in CSF of patients with AD and CON in the validation population. This profile is depicted in figure 2 and in table 3 the masses are listed corresponding to these peptides, their intensities and differences in intensities between groups.


175

Table 3. Results of the SELDI-TOF MS analyses using 6E10 antibodies coupled to PS20 arrays. Mass (measured)

Peptide ID*

Mass* (calculated)

Intensity (mean ± SD) P-value

AD (n=39)

VaD (n=31)

CON (n=52)

AD vs CON

VaD vs CON

VaD vs AD

Verification method Lewczuk et al. 4077 1-37 4074 5.2 ± 4.6 9.0 ± 4.6 15.1 ± 5.0 < 0.001 < 0.001 0.001 4135 1-38 4131 11.3 ± 8.7 20.0 ± 9.2 29.0 ± 7.6 < 0.001 < 0.001 < 0.001 4233 1-39 4230 3.5 ± 2.0 5.9 ± 2.6 9.6 ± 3.9 < 0.001 < 0.001 < 0.001 4333 1-40 4329 18.9 ± 9.8 27.6 ± 10.1 35.2 ± 8.5 < 0.001 0.001 < 0.001 4519 1-42 4514 0.9 ± 0.8 1.8 ± 1.0 2.3 ± 1.2 < 0.001 NS < 0.001 4822 1-45

3-47 4827 4846

1.7 ± 2.3 1.0 ± 1.1 0.4± 0.6 < 0.001 0.002 NS

7760 (X-X)2 1.9 ± 1.3 1.8± 1.5 1.7± 2.7 NS NS NS Additional significant m/z values 3657 9-43 3657 1.8 ± 1.3 2.1 ± 1.2 2.8 ± 1.7 0.002 NS NS 3676 1-33 3673 2.9 ± 2.4 4.8 ± 2.0 6.1 ± 2.2 < 0.001 0.007 0.001 3701 6-40 3711 1.5 ± 1.0 2.0 ± 0.8 2.6 ± 0.9 < 0.001 0.003 NS 3789 1-34 3789 2.7 ± 2.0 4.1 ± 2.0 5.4 ± 1.9 < 0.001 0.006 0.006 4093 1-37-O

6-44 4092 4095

4.5 ± 3.7 6.7 ± 3.7 11.6 ± 4.7 < 0.001 < 0.001 NS

4151 3-40 1-38-O 5-43

4143 4149 4152

8.3 ± 6.3 12.8 ± 6.1 19.8 ± 6.4 < 0.001 < 0.001 NS

4172 1-38-2O 4165 4.3 ± 3.2 7.7 ± 3.7 12.9 ± 4.6 < 0.001 < 0.001 < 0.001 4250 1-39-O

5-44 4246 4251

3.6 ± 1.4 4.3 ± 1.9 6.7 ± 3.4 < 0.001 0.001 NS

4283 1-39-3O 4-43

4278 4299

5.4 ± 1.8 6.8 ± 2.6 9.5 ± 3.3 < 0.001 < 0.001 NS

4350 1-40-O 5-45

4345 4365

16.2 ± 7.8 22.3 ± 8.4 27.2 ± 7.1 < 0.001 0.005 0.003

4370 1-40-2O 5-45

4361 4365

10.0 ± 3.7 14.8 ± 5.6 18.3 ± 5.3 < 0.001 0.006 < 0.001

5534 (X-X)2 2.3 ± 2.2 2.1 ± 1.7 0.9 ± 1.1 < 0.001 < 0.001 NS 5677 (X-X)2 2.4 ± 1.2 2.7 ± 1.5 1.7 ± 0.8 0.003 0.001 NS 5761 (X-X)2 4.2 ± 1.8 3.1 ± 1.4 2.3 ± 1.3 < 0.001 0.004 NS 6248 (X-X)2 2.0 ± 1.0 1.6 ± 1.1 1.2 ± 0.7 < 0.001 NS NS 7664 (X-X)2 2.5 ± 1.7 3.7 ± 1.9 3.4 ± 1.8 0.009 NS 0.003 8214 (X-X)2 2.7 ± 1.9 2.1 ± 1.3 0.9 ± 0.9 < 0.001 < 0.001 NS 8602 (X-X)2 6.0 ± 2.5 4.1 ± 1.9 3.2 ± 1.2 < 0.001 NS < 0.001 8678 (X-X)2 3.9 ± 1.6 3.0 ± 1.9 2.4 ± 0.8 < 0.001 NS 0.002 11507 (X-X)3 3.0 ± 1.5 1.4 ± 0.6 1.3 ± 0.6 < 0.001 NS < 0.001 11573 (X-X)3 2.3 ± 1.3 1.2 ± 0.5 1.0 ± 0.5 < 0.001 NS < 0.001 12196 (X-X)3 2.3 ± 2.0 1.2 ± 1.0 0.8 ± 0.5 0.005 NS NS * Proposed peptide ID based upon calculated mass. NS= not significant, -O = + 1 oxygen (MW+16), -O2 = +2 oxygen (MW+32), -O3 = +3 oxygen (MW+48), (X-X)2 = dimeric form of unknown amyloid peptide, (X-X)3 = trimeric form of unknown amyloid peptide

Chapter 4.2.3

176

In our population, the intensities of Aß1-37, Aß1-38, Aß1-39, Aß1-40 and Aß1-42 were significantly lower in AD compared to CON. With the exception of Aß1-42, all other amyloid peptide intensities were also significantly lower in VaD compared to CON. In addition, the five described amyloid peptides differed significantly between the both dementias. Figure 1. Representative m/z spectra for a patient with Alzheimer’s Disease (AD), Vascular Dementia (VaD) and a control subject (CON) as obtained with Q10 arrays.

010203040

5000 7500 10000 12500

AD

0

102030

40

VaDInte

nsity

0

10203040

CON

5000 7500 10000 12500

m/z value

1

2 3

1

1

2

2

4

4

Numbers represent identified peptides of Carrette et al. in the validation populatation and their m/z values: 1 = 4814 (VGF), 2 = 11709 (β2- microglobulin), 3 = 11916 (β2- microglobulin), 4 = 13358 (Cystatin C)

Lewczuk et al.10 reported in their study also three new Aß peptides: 3-44 (4528 Da), 3-47 (4853 Da) and a probable dimer (7756 Da). Peaks corresponding to those molecular masses could also be found in CSF of our AD, VaD and CON groups, although a peak corresponding to 3-44 could not conclusively established and it was difficult to assign the peak at m/z 4822 to either 3-47 (4853 Da) or to 1-45 (4827 Da). The intensity of the peak corresponding to the putative dimer was not significantly different between AD, VaD and CON. The peak (4822 Da) corresponding to 1-45 or 3-47, however, was significantly higher in AD and VaD compared to CON.


177

As Lewczuk et al.10 reported Signal to Noise (S/N)-values in their study, we also performed S/N analysis in our population for all the described amyloid peptides in this section (results not shown). Significant differences calculated for intensities were highly comparable to those calculated for S/N in our population. Aß1-42 is the only exception, as its S/N is significantly lower in VaD compared to CON and its intensity is non-significantly lower in VaD compared to CON. In addition to the fingerprint and new peptides described by Lewczuk et al.10, we have also found additional Aβ peptides in CSF in AD, VaD and CON subjects. In general, peak intensity increased in the following order: AD<VaD<CON. These peptides are listed in table 3 and based upon their molecular masses peptide IDs are proposed. Aß1-33 and Aß1-34 were identified and are smaller peptides that belong to the amyloid peptide fingerprint. Peaks corresponding to fingerprint amyloid peptides are often accompanied by a peak with a molecular mass +16 or molecular mass +32, possibly due to oxidation. However, it cannot be excluded that these molecular masses correspond to amino-truncated Aß at positions 4, 5 or 6. These peptides are elongated at the C-terminal, comparable to 3-47 as proposed by Lewczuk et al.10 Intensities of the peaks at molecular masses possibly corresponding to oxidised or truncated Aß are significantly lower in AD compared to CON and in VaD compared to CON. Intensity of the peaks at m/z values 4172 (Aβ1-38-2O), 4350 (Aβ1-40-O) and 4370 (Aβ1-40-2O or Aβ5-45) differed significantly between AD and VaD. Discriminatory peak clusters at molecular masses above 5534 may correspond to dimers or trimers of small Aß peptides. Almost all these dimers and trimers, except for the one corresponding to m/z 7664, are upregulated in AD compared to CON. Also in VaD, some of these peaks are upregulated as compared to controls and some are downregulated compared to AD. Discussion We successfully replicated the CSF protein profiles described by Carrette et al.9 and Lewczuk et al.10 in CSF of larger populations of patients with AD and controls. In addition, both profiles were also detected in patients with VaD. Using both methods also additional discriminatory peak clusters were found. We were unable to replicate significant differences in intensities for VGF between AD and CON, as described by Carrette et al.9 Additionally, for VGF no significant differences between VaD and CON and between VaD and AD were found. For cystatin C and ß-2-microglobulin we showed highly significant differences between AD and CON and for the ß-2-microglobulin also between VaD and CON, although peak intensities, in general, were smaller in our analysis.

Chapter 4.2.3

178

Figure 2. Representative amyloid beta fingerprint in spectra of a patient with Alzheimer’s Disease (AD), Vascular Dementia (VaD) and a non-demented age-matched control participant (CON).

Numbers represent identified fingerprint peptides and their m/z values: 1 = 3676 (Aβ1-33), 2 = 3789 (Aβ1-34), 3 = 4077 (Aβ1-37), 4 = 4093 (Aβ1-37-O), 5 = 4135 (Aβ1-38), 6 = 4151 (Aβ1-38-O), 7 = 4172 (Aβ1-38-2O), 8 = 4233 (Aβ1-39), 9 = 4250 (Aβ1-39-O), 10 = 4333 (Aβ1-40), 11 = 4350 (Aβ1-40-O), 12 = 4370 (Aβ1-40-2O), 13 = 4519 (Aβ1-42)

3500 4000 4500 5000

3500 4000 4500 5000

AD

VaD

CON

0

10

20

30

40

0

10

20

30

40

0

10

20

30

40

35

8

11

1012

1 2

3 47 8 9

10511

12

1 2

10

11

126

5

3

8

m/z value

Inte

nsity

3500 4000 4500 5000

3500 4000 4500 5000

AD

VaD

CON

0

10

20

30

40

0

10

20

30

40

0

10

20

30

40

35

8

11

1012

1 2

3 47 8 9

10511

12

1 2

10

11

126

5

3

8

3500 4000 4500 5000

3500 4000 4500 5000

AD

VaD

CON

0

10

20

30

40

0

10

20

30

40

0

10

20

30

40

3500 4000 4500 5000

3500 4000 4500 5000

AD

VaD

CON

0

10

20

30

40

0

10

20

30

40

0

10

20

30

40

35

8

11

1012

1 2

3 47 8 9

10511

12

1 2

10

11

126

5

3

8

35

8

11

1012

35

8

11

1012

1 2

3 47 8 9

10511

121 2

3 47 8 9

10511

12

1 2

10

11

126

5

3

8

1 2

10

11

126

5

3

8

m/z value

Inte

nsity


179

These proteins were proposed as biomarkers for AD by Carrette et al.9. Interestingly, these markers are also significantly higher in patients with VaD as compared to CON, but no significant differences between those markers were found between VaD and AD. It can thus be hypothesised that these markers reflect processes in the brain that underlie both AD and VaD. Rüetschi et al.15 identified a fragment of neurosecretory protein VGF (amino acids 1-3 deleted) and truncated cystatin C (amino acids 1-8 deleted) as potential biomarkers for frontotemporal dementia (FTD) using SELDI-TOF MS. The fragment of VGF was significantly less abundant in FTD and truncated cystatin C was significantly increased in FTD patients compared to controls. Moreover, in a small study in patients with Creutzfeldt-Jakob Disease (CJD) and using SELDI-TOF MS it was shown that cystatin C levels were significantly decreased16. Clearly, as differences between the groups are small and a large overlap is apparent it will be difficult to define cut-off values of the markers appropriate for discriminating dementia syndromes from other diseases and dementia syndromes from each other. We were also able to replicate the Aβ fingerprint in CSF of AD and CON as described previously by others10,17,18. We also found peaks corresponding to molecular masses of Aß 1-33 and Aß 1-34 that were reported in other studies as well17,18. The seven C-terminally truncated Aß peptides, the fingerprint, can be generated by γ-secretase that has seven potential cleavage sites19-21. Aß1-38 and Aß1-40 are the most abundant species in CSF and those results are comparable with other studies10,14,17,18. For all the Aß fingerprint peptides we observed significantly lower intensities in AD compared to CON. This is in contrast with Lewczuk et al.10 as they showed significant differences for Aß1-42 but not for the other four fingerprint peptides identified in their analysis. Maddalena et al.17, however, showed significant differences between AD and CON for Aß1-38 and Aß1-40. Using ELISA it has also been shown that CSF Aß1-42 levels are significantly lower in AD compared to CON22. To our knowledge, the fingerprint has not yet been described in CSF of patients with VaD. Using ELISA it has been shown that CSF levels of Aß1-42 were significantly lower in AD compared to VaD23-25 and that Aß1-42 could discriminate AD from VaD23. In contrast, however, another study suggests that the Aβ42/phosphorylated-tau ratio, but not Aβ42 alone, differentiates AD from VaD26. CSF levels of Aß1-42, are intermediate between AD and CON24, which is in agreement with our results obtained with SELDI-TOF MS. The amyloid fingerprint is thus present in CSF of both VaD and AD, although less abundant in AD. This is an interesting result of this study, as nowadays it is more recognised that pathological processes overlap and/or interact with each other in VaD and AD. Aß production in AD may lead to cerebrovascular dysfunction and the microvasculature alteration caused by amyloid deposits increases the risk of VaD. And the other way around, development of AD-pathology may partly be due to cerebral ischemia.7 In addition,

Chapter 4.2.3

180

research regarding cerebral amyloid angiopathy (CAA), which is considered to frequently co-exist in AD and more recently also in VaD, suggested that distinct mechanisms are responsible for the differential deposition of beta amyloid in CAA associated with AD and associated with cerebrovascular disease27.

In addition to our results in VaD, Bibl et al.28 recently described the amyloid peptide fingerprint in CSF of patients with Lewy Body Disease (LBD), Parkinson’s Disease Dementia, AD and CON. Aß1-42 was significant less abundant in AD and LBD as compared to CON. The amyloid fingerprint is thus present in different types of dementia and future research should reveal whether amyloid peptides are suitable biomarkers to distinguish between these different syndromes. Using the method of Lewczuk et al.10 we found extra discriminating peaks and did match those with calculated masses of either oxidised or amino-terminally truncated Aß peptides. Several amino-terminally truncated beta amyloid peptides have been identified in temporal cortex from patients with AD, VaD and controls14,29. Two amino-terminally truncated peptides, Aß8-42 and Aß11-42, were shown in CSF of AD and CON30. We have theoretically assigned peaks to different truncated Aß-peptides based upon their molecular masses. To what extent these peptides play a role in the pathogenesis and if they can be used as biomarkers to differentiate between dementia and control patients and between different dementia subtypes should receive further attention in future research. Fingerprint Aß peptides accompanied by an additional peak, with the mass plus 16 Da, were seen for Aß1-37, Aß1-38, Aß1-39 and Aß1-40 in CSF of non-demented controls and are probably oxidised Aß peptides. Oxidation is possible at methionine at residue 35 and thus affects all fingerprint Aß peptides31. In addition to the single oxidised forms we hypothesised also double or triple oxidised Aß peptides. Oxidation is also possible at histidine residues at positions 6, 13 and 1431. Aß peptide oxidation may be catalysed by metal-ions and reactive oxygen species, the result of oxidative stress, and the oxidized peptides are prone to aggregate and form deposits in the brain31. Our results show that peaks theoretically responding to 1-38-2O, 1-40-O and 1-40-2O are significantly different between AD and VaD. These makers may serve a possible future role as biomarker. Although our results are only hypothesis generating, as we have not unambiguously identified the peaks corresponding to the oxidised species or truncated peptides, it is shown that intensities of those peaks are significantly lowered in AD and VaD as compared to CON. In theory, this could be explained by the fact that oxidised31 and/or N-truncated peptides are prone to plaque formation32. In addition, at higher m/z values we identified peaks that we proposed to be dimers and trimers of small Aß peptides that remained in solution. Assuming those masses represent aggregates of Aß peptides in our analysis, we did show that these are more abundant in AD compared to CON. Only a few of these peaks showed to be more abundant in VaD


181

compared to CON. However, peaks at m/z values 7664, 8602, 8678, 11507 and 11573 differed significantly between AD and VaD an may in future be used as biomarkers after further investigation. The CON subjects in our study were significantly younger compared to both AD and VaD subjects. Whether this difference has influenced our results needs more research. In conclusion, we showed protein profiles measured by SELDI-TOF MS in CSF of AD, VaD and CON. Interestingly, different profiles of Aβ aggregates and truncated or oxidised Aβ peptides were observed in CSF of AD, VaD and CON subjects. Future research is needed to obtain further insight into the applicability of those markers in early diagnostics and their role in the pathogenesis of dementia.

References 1. Ferri CP, Prince M, Brayne C, Brodaty H, et al. Global prevalence of dementia: a delphi consensus study.

Lancet 2005;366:2112-17. 2. Ringmann JM, Cummings JL. Current and emerging pharmacological treatment options for dementia. Behav

Neurol 2006;17:5-16. 3. Siemers ER, Quinn JF, Kaye J, et al. Effects of a gamma-secretase inhibitor in a randomized study of

patients with Alzheimer disease. Neurology 2006;66:602-4. 4. Blennow K, Hampel H. CSF markers for incipient Alzheimer’s Disease. Lancet Neurol 2003;2:605-13. 5. Andreasen N, Blennow K. CSF biomarkers for mild cognitive impairment and early Alzheimer’s disease.

Clin Neurol Neurosurg 2005; 107: 165-73. 6. Teunissen CE, de Vente J, Steinbusch HW, de Bruijn C. Biochemical markers related to Alzheimer’s

dementia in serum and cerebrospinal fluid. Neurobiol Aging 2002;23:485-508. 7. Martinez-Villa E, Murie-Fernandez M, Perez-Larraya J, Irimia P. Neuroprotection in vascular dementia.

Cerebrovasc Dis 2006;21 (suppl 2):106-117. 8. Engwegen JYMN, Gast MCW, Schellens JHM, Beijnen JH. Clinical proteomics: searching for better tumour

markers with SELDI-TOF mass spectrometry. Trends Pharmacol Sci 2006;27:251-9. 9. Carrette O, Dermalte I, Scherl A, et al. A panel of cerebrospinal fluid potential biomarkers for the diagnosis

of Alzheimer’s disease. Proteomics 2003;3:1486-94. 10. Lewczuk P, Esselmann H, Wolfgang T, et al. Amyloid beta peptides in cerebrospinal fluid as profiled with

surface enhanced laser desorption/ionization time-of-flight mass spectrometry: evidence of novel biomarkers in Alzheimer’s disease. Biol Psychiatry 2004;55:524-30.

11. Folstein MF, Folstein SE, McHugh PR. Mini-Mental State: a practical method for grading cognitive state of patients for the clinician. Journal of Psychiatric Research 1975;12:189-198.

12. McKahnn G, Drachmann D, Folstein M et al. Clinical diagnosis of Alzheimer´s disease: Report of the NINCDS-ADRDA work group under the auspices of department of health and human services task force on Alzheimer´s disease. Neurology 1984;34:939-944.

13. Erkinjunnti T. Clinical criteria for vascular dementia: the NINDS-AIREN criteria. Dementia 1994;5:189-92. 14. Lewczuk P, Esselmann H, Meyer M, et al. The amyloid beta peptide pattern in cerebrospinal fluid in

Alzheimer’s disease: evidence of a novel carboxyterminally elongated Aß peptide. Rapid Commun Mass Spectrom 2003;17:1291-6.

15. Ruetschi U, Zetterberg H, Podust V, et al. Identification of CSF biomarkers for frontotemporal dementia using SELDI-TOF. Exp Neurol 2005;196:273-281.

Chapter 4.2.3

182

16. Sanchez JC, Guillaume E, Lescuyer P, et al. Cystatin C as a potential cerebrospinal fluid marker for the diagnosis of Creutzfeldt-Jakob disease. Proteomics 2004;4:2229-33.

17. Maddalena AS, Papassotiropoulos A, Gonzalez-Agosti C, et al. Cerebrospinal fluid profile of amyloid beta peptides in patients with Alzheimer’s disease determined by protein biochip technology. Neurodegenerative Dis 2004;1:231-35.

18. Portelius E, Westman-Brinkmalm A, Zetterberg H, Blennow K. Determination of beta amyloid peptide signatures in cerebrospinal fluid using immunoprecipitation-mass spectrometry. J Prot Res 2006;5:1010-16.

19. Beher D, Wrigley JD, Owens AP, Shearman MS. Generation of C-terminally truncated amyloid-beta peptides is dependent on gamma-secretase activity. J Neurochem 2002;82:563-575.

20. Murphy MP, Hickman LJ, Eckman CB. Gamma-secretase, evidence for multiple proteolytic activities and influence of membrane positioning of substrate on generation of amyloid beta peptides of varying length. J Biol Chem 1999;247:11914-11923.


22. Sunderland T, Linker G, Mirzna N et al. Decreased beta-amyloid1-42 and increased tau levels in cerebrospinal fluid of patients with Alzheimer disease. JAMA 2003; 2094-2103.

23. Stefani A, Bernardini S, Panella M, et al. AD with subcortical white matter lesions and vascular dementia: CSF markers for differential diagnosis. J Neurol Sci 2005;237:83-8.

24. Andreasen N, Minthon L, Davidsson P, et al. Evaluation of CSF tau and CSF abeta 42 as diagnostic markers for alzheimer’s disease in clinical practice. Arch Neurol 2001;58:373-79.

25. Parnetti L, Lanari A, Saggese E, et al. Cerebrospinal fluid biochemical markers in early detection and in differential diagnosis of dementia disorders in routine clinical practice. Neurol Sci 2003;24:199-200.

26. de Jong D, Jansen RWMM, Kremer HPH, Verbeek MM. CSF Amyloid β42 and phosphorylated tau can discriminate between dementia disorders. J Gerontol;61A:755-758.

27. Haglund M, Kalaria R, Slade JY, Englund E. Differential deposition of amyloid ß peptides in cerebral amyloid angiopathy associated with Alzheimer’s disease and vascular dementia. Acta Neuropathol 2006;111:430-435.

28. Bibl M, Mollenhauer B, Esselmann H, et al. CSF amyloid-beta peptides in Alzheimer’s disease, dementia with Lewy bodies and Parkinson’s disease dementia. Brain 2006;129:1177-1187.

29. Lewis H, Beher D, Cookson N, et al. Quantification of Alzheimer pathology in ageing and dementia: age related accumulation of amyloid beta 42 peptide in vascular dementia. Neuropathol Appl Neurobiol 2006;32:103-118.

30. Vanderstichele H, de Meyer G, Andreasen N, et al. Amino-truncated beta amyloid 42 peptides in cerebrospinal fluid and prediction of progression of mild cognitive impairment. Clin Chem 2005;51:1650-60.

31. Inoue K, Garner C, Ackermann B, et al. Liquid chromatography/mass spectrometry characterization of oxidized amyloid beta peptides as potential biomarkers of Alzheimer’s disease. Rapid Commun Mass Spectrom 2006;20:911-18.

32. Sergeant N, Bombois S, Ghestem A, et al. Truncated beta-amyloid peptide species in pre-clinical Alzheimer’s disease as new targets for the vaccination approach. J Neurochem 2003;85:1581-91.

183

184

185

CHAPTER 4.2.3 Serum amyloid beta peptides in patients with dementia and

age-matched non-demented controls as detected by Surface-Enhanced Laser Desorption Ionisation-Time of Flight Mass

Spectrometry (SELDI-TOF MS)

S.V. Frankfort, J.P.C.M. van Campen, C.R. Tulner, J.H. Beijnen

Abstract Background: By using surface enhanced laser desorption/ionisation- time of flight mass spectrometry (SELDI-TOF MS) an amyloid ß (Aß) profile was shown in cerebrospinal fluid (CSF) of patients with dementia. This method, however, has not been used to investigate an Aß profile in serum or plasma. Objective: To investigate the Aβ-profile in serum with SELDI-TOF MS, to evaluate if this profile resembles CSF profiles and to investigate if the intensity of the peaks corresponding to Aβ-peptides in serum are correlated to clinical, demographical and genetic variables. Methods: Profiling of Aß by an Aß immunocapture assay using SELDI-TOF MS was performed in duplicate in 106 included patients, suffering from Alzheimer´s Disease or Vascular Dementia and age-matched non-demented control patients. Linear regression analyses were performed to investigate the intensities of four selected Aβ peaks as dependent variables in relation to the independent variables type of clinical diagnosis, creatinine level, age, gender, presence of APOE ε4 allele and ABCB1 genotypes at positions C1236T, G2677T/A and C3435T, encoding for P-glycoprotein (P-gp), an Aβ-transporter . Results: Aß37, Aß38 and Aß40 were found among additional unidentified Aß peptides, with the most pronounced Aß peak at a molecular mass of 7752 and this profile partly resembled that found earlier in CSF. The clinical diagnosis was not a predictive independent variable, however ABCB1 genotypes C1236T, G2677T/A, age and creatinine level showed to be related to Aß peak intensities in multivariate analyses. Conclusions: We found an Aß profile in serum by an Aß immunocapture assay using SELDI-TOF MS that partly resembled the profile as earlier obtained in CSF of demented patients. This protein profile was, however, not associated with any investigated clinical diagnosis. Age, creatinine levels, presence of the APOE ε4 allele and ABCB1 genotypes (C1236T and G2677T/A) were correlated with the Aβ serum profile. The role of P-gp as an Aß transporter and the role of ABCB1 genotypes deserves further research. The investigated serum Aβ profile is probably not useful in the diagnosis of dementia.

Submitted

Chapter 4.2.3

186

Introduction Dementia is a rapidly growing health and socioeconomic problem worldwide, as it is estimated that 24 million people have dementia nowadays and that this will increase to 42 million by 2020 and 81 million by 20401. Alzheimer’s Disease (AD) is the most common form of dementia2 and Vascular Dementia (VaD) is the second in the developed world3. Amyloid precursor protein (APP) is degraded by β-secretases and γ-secretases into amyloid beta (Aβ) peptides that play a key role in the pathogenesis of AD2 and are also related to VaD4. Aβ40 and Aβ425,6 have been frequently investigated and more recently also Aβ387,8 was found in cerebrospinal fluid (CSF) by using Enzyme Linked Immunosorbent Assay (ELISA). Aβ38, Aβ40 and Aβ42 are candidate biomarkers for AD and Aβ42 for its potential to differentiate from VaD9-12. Aβ38, Aβ40 and Aβ42 are also found in plasma of patients with AD13,14,15 and Aβ40 and Aβ42 in patients with VaD15 as investigated with ELISA. Recently, van Oijen et al.15 showed in a longitudinal follow-up study that elevated baseline Aβ40 plasma levels, especially when combined with low baseline plasma levels of Aβ42, indicate an increased risk for AD. Results from new proteomic techniques in dementia research may add to current knowledge regarding biomarkers obtained with ELISA. One of these techniques is the Protein Chip® array based Surface Enhanced Laser Desorption/Ionisation in combination with Time of Flight Mass Spectrometry (SELDI-TOF MS). Arrays with different chromatographic surfaces enhance on chip purification and retained proteins are analysed on the same chip resulting in a profile of proteins characterised by mass and charge16. SELDI-TOF MS has been applied in CSF of AD17,18 and in VaD (unpublished results Frankfort et al.) and revealed novel Aβ-peptides besides Aβ38, Aβ40 and Aβ42 resulting in an Aβ-fingerprint, consisting of Aβ37, Aβ38, Aβ39, Aβ40 and Aβ42, in CSF. To our best knowledge, only Lewczuk et al.19 described an Aβ-fingerprint in plasma of AD patients, using gel electrophoresis. SELDI-TOF MS analysis in plasma or serum of patients suffering from AD or VaD has not yet been described. An interesting question is whether any Aβ-profile in serum is comparable to CSF. In addition, it would be relevant to investigate whether significant differences in the Aβ serum profile exist between patients with AD, VaD, and age-matched non-demented control patients and how known variables correlate to Aβ-levels, e.g. age20, serum creatinine level21 and presence of the apolipoprotein E (APOE) ε4 allele22. Aβ peptides can thus be measured both in CSF and plasma, however the source of the peptides remains unknown. P-glycoprotein (P-gp, ABCB1), a 170 kDa membrane bound efflux pump, is located at the Blood Brain Barrier (BBB) at the apical membrane of endothelial cells and is proposed to act as a flippase and distributes substrates into the extracellular space23,24. P-gp is also expressed at the blood-cerebrospinal fluid (BCSF) barrier, that is formed by the choroid plexus25. A possible role of P-gp in mechanisms

Serum amyloid beta peptides in geriatric patients

187

underlying the pathology of Alzheimer’s disease (AD) has been hypothesised. Aβ is an in vitro substrate for P-gp26 and P-gp deficiency at the BBB increases beta amyloid deposition in an AD mouse model27. Vogelgesang et al.28 showed P-gp expression at the BBB to be inversely correlated to the number of amyloid plaques in the medial temporal lobe in 243 non-demented elderly. The Multi Drug Resistance gene (ABCB1), encoding for P-gp, is highly polymorphic29. The three most frequently occurring Single Nucleotide Polymorphisms (SNPs) are C1236T in exon 12, G2677T/A in exon 21 and C3435T in exon 2630. The synonymous SNP C3435T was the first variant to be associated with altered protein expression29. As ABCB1 SNPs may be related to P-gp expression and function we hypothesised that ABCB1 genotypes are possibly correlated to serum amyloid levels. This study aims a) to investigate the Aβ-profile in serum of elderly patients with SELDI-TOF MS, b) to evaluate if the serum profile resembles the Aβ CSF profile as investigated with SELDI-TOF MS and c) to investigate if the intensities of the peaks corresponding to Aβ-peptides in serum are correlated to clinical (dementia diagnosis, serum creatinine level), demographic (age, gender) and genetic (APOE ε4 allele, ABCB1 SNPs C1236T, G2677T/A, C3435T) variables.

Methods Patients and Diagnosis This prospective study was executed at the geriatric diagnostic day-clinic of the Slotervaart Hospital, a teaching hospital in Amsterdam, the Netherlands. In each patient complete geriatric assessment including Mini Mental State Examination (MMSE)31, the 7-Minute Neurocognitive Screening Test32 and laboratory testing, including thyroid function, levels of folic acid, thiamine and vitamin B12, was performed. To assist in dementia diagnosis, patients underwent more extensive neuropsychological assessment and, if necessary, computerised tomography or magnetic resonance imaging was performed. This study included patients suffering from AD or VaD and age-matched non-demented control patients. AD was diagnosed according to the NINCDS-ADRDA criteria33 and VaD according to the NINDS/AIREN criteria34. Controls were recruited from the same diagnostic day-clinic. These participants did not show any cognitive impairment. Most of these geriatric patients were presented at the diagnostic day-clinic for a somatic screening. The study protocol was approved by the Institutional Review Board of the Slotervaart Hospital, Amsterdam, The Netherlands. Written informed consent was obtained from each participant in this study.

Chapter 4.2.3

188

Sample collection From each participant a 2 ml EDTA whole blood sample was obtained by venous puncture and stored at -20 °C until genotype analysis. Serum samples (5 ml) were obtained by venapuncture and directly after clotting centrifuged at 1000 g for 10 min. Supernatants were divided into 0.5 ml aliquots and stored at -20 ºC. After collection of all samples these were stored at -80 ºC. SELDI-TOF MS analysis was performed with serum unexposed to any freeze-thaw cycle. Sample pre-treatment and SELDI-TOF MS analysis SELDI-TOF MS analysis using an antibody-capture assay was performed in duplicate in our serum sample set. 2 µl of 6E10 monoclonal antibody (Chemicon International Inc., Temecula, CA, USA) solution in phosphate buffered saline (PBS) (0.25 mg/ml) was added to each spot of a PS20 Protein Chip array (Ciphergen Biosystems Inc., Fremont, CA, USA) used for detection of patient serum samples. Additionally, 2 µl of a bovine IgG solution in PBS (0.25 mg/ml) (Ciphergen Biosystems Inc.) was added on spots used for analysis of negative controls. The arrays were incubated for 2 hours at room temperature in a humidity chamber on a platform shaker at 350 r/m. Afterwards 6E10 antibody or IgG were removed from each spot and the arrays were placed in 10 mL polypropylene tubes containing deactivation buffer (0.5M ethanolamine (Sigma, St. Louis, MO, USA) in PBS (Sigma), pH 8.0) to block the residual sites on the array. Tubes were agitated on a shaker for 30 minutes at room temperature. The arrays were then washed twice with wash buffer (0.5% Triton-X-100 (Sigma) in PBS) for 10 minutes and once with PBS for 5 minutes at room temperature. The arrays were transferred to a 96-well format bioprocessor (Ciphergen Biosystems) and 50 µl of serum was applied to each well. The bioprocessor was placed on a platform shaker and agitated overnight at 4 ºC. The samples were removed and the arrays were washed in the bioprocessor three times with wash buffer for 15 minutes and three times with PBS for 5 minutes and finally rinsed with 1.0 mM HEPES buffer (Ciphergen Biosystems). As energy absorbing matrix a volume of 0.5 µl of a 20% saturated solution of α-cyano-4-hydroxy-cinnamic acid (CHCA) (Ciphergen Biosystems) in 50% v/v acetonitrile and 0.5% v/v TFA was added twice to each spot. Protein profiling of each spot was performed using the PBS-IIC Protein Chip Reader (Ciphergen Biosystems). Data were collected between 0 and 15,000 Da. Data collection was optimised, resulting in an average of 100 laser shots per spectrum at laser intensity 150 and detector sensitivity 7. M/z values for detected proteins were calibrated externally with a standard peptide mixture (Ciphergen Biosystems) containing [Arg8] vasopressin (1084.3 Da), somatostatin (1637.9 Da), dynorphine (2147.5 Da), ACTH (2933.5 Da), insuline ß-chain (bovine) (3495.9 Da) and insulin (human recombinant) (5807.7 Da).


189

Bioinformatics All acquired SELDI-TOF MS spectra were compiled and analysed as a whole experiment with ProteinChip software (version 3.1, Ciphergen Biosystems). Spectra were baseline subtracted and normalised to the total ion current between 1500 and 15,000. The Biomarker Wizard application (BMW) (Ciphergen Biosystems) was used to autodetect m/z peaks with a signal-to-noise ratio of at least 5. Peak clusters were completed with peaks with a signal-to-noise ratio of at least 2 in a 0.3% cluster mass window. Peak data per spectrum were exported to Excel (Microsoft Inc., Redmond, WA, USA) and mean intensities were calculated from duplicate spectra for each participant. DNA isolation and genotype analysis Genomic DNA was extracted from EDTA whole blood using the Qiagen QIAamp® DNA Mini Kit (Qiagen, Leusden, The Netherlands) according to the manufacturer protocol. APOE genotype was determined using real time polymerase chain reactions (RT-PCR) based upon Koch et al.35 ABCB1 was screened for C1236T in exon 12, G2677T/A in exon 21 and C3435T in exon 26 by sequencing, as earlier described36. Statistics Univariate linear regression analyses were performed with intensities of selected peaks obtained for each patient by SELDI-TOF MS as dependent variable and age, gender, serum creatinine level, presence or absence of an APOE ε4 allele, ABCB1 genotypes C1236T (CC, CT, TT), G2677T/A (GG, GT, TT/TA), C3435T (CC, CT, TT) and diagnostic group (age-matched non-demented controls, AD or VaD) as independent variables. When multiple significant (p<0.1) covariates were identified univariately, multivariate linear regression analyses were performed and corrected for age and gender. Statistical calculations were performed with SPSS for Windows (version 11.0, SPSS Inc., Chicago, IL, USA).

Results Participants In total, 108 patients signed informed consent. In 2 patients venous puncture failed and 106 patients were thus included. Baseline characteristics are presented in table 1. The mean age of the included patients was 81.8 years (SD 5.6), mean MMSE score was 20.5 (SD 5.9), the median education level was 4 (on a seven-point scale, ranging from less than 6 years of elementary school (score 1) to a university degree (score 7)37, and 63.2% was female.

Chapter 4.2.3

190

Table 1. Demographic characteristics of included participants. Characteristic Included participants(n=106) Age, mean ± SD (range) 81.1±5.6 (67.6–94.5) Baseline MMSE, mean ± SD (range) 20.5 ± 5.9 (3-29) ¶

Clinical Diagnosis, Alzheimer’s Disease, n (%) 48 (45.3) Vascular Dementia, n (%) 18 (17.0) Age-matched non-demented controls, n (%) 40 (37.7) Creatinine level (µmol/L), mean ± SD (range) 98.8 ± 50.6 (52-504)*

Education, median IQR (range) 4 IQR:3 (1-7)* #

Gender, n (%) female 67 (63.2) APOE ε4 allele present, n (%) 41 (38.7) ABCB1 exon 12 SNP C1236T CC, n (%) 27 (25.5) CT, n (%) 53 (50.0) TT, n (%) 26 (24.5) ABCB1 exon 21 SNP G2677T/A GG, n (%) 29 (27.4) GT, n (%) 47 (44.3) TA, n (%) 2 (1.9) TT, n (%) 29 (26.4) ABCB1 exon 26 SNP C3435T CC, n (%) 16 (15.1) CT, n (%) 51 (48.1) TT, n (%) 39 (36.8) SD= Standard Deviation, IQR = Inter Quartile Range, MMSE = Mini Mental State Examination.missing for 3 patients; ¶ in 22 controls performed; # scored on scale 1 (less than 6 years of elementary school) to 7 (a university degree)37

Aβ peptide profile in serum The Aβ profile obtained in serum is depicted in figure 1 and masses and their corresponding intensities (mean ± SD) are summarised in table 2. Aβ37, Aβ38 and Aβ40 were detected in serum with SELDI-TOF MS, however, Aβ39 and Aβ42 were undetectable using this method. Besides the Aβ fingerprint, masses possibly corresponding to other Aβ-peptides were found. Prominently apparent was the huge peak at a molecular mass of 7752 and other peaks at masses 3766, 3891, 4623, 5102, 7567, 8143 and 8931. Peaks detected at those masses could theoretically correspond to truncated or dimeric forms of Aβ peptides, because they are selectively retained on the amyloid antibody chip array. In table 2 is also given which of the peaks found in serum were also detected in a separate analysis in CSF (unpublished results Frankfort et al.). The data regarding serum and CSF SELDI-TOF MS analyses are from different patient populations (serum: Slotervaart Hospital; CSF: Radboud University Nijmegen Medical Centre). Included patients in both the cerebrospinal fluid study and the serum study suffered from AD or VaD or were control participants. Besides the Aβ fingerprint peptides, peaks at masses of 1019, 3891, 5102, 5675, 5805, 7752, 8143, 8608 and 8678 were also detected in CSF. Reported intensities were higher in CSF for Aβ37, Aβ38, Aβ40 and peaks at masses of 5675, 8608 and 8678 (p<0.001). Peak intensities at masses of 1019, 3891, 5102, 7752 and 8143, were lower in CSF than in serum (p<0.001).


191

Table 2. Results of the SELDI-TOF MS Aβ profile analysis in serum. Measured mass (Peptide ID*) Intensity (Mean ± SD) Cerebrospinal fluid detected?#

Measured mass (Mean intensity ± SD) 1019 2.0 ± 2.8 1021 (0.7 ± 0.8) 3049 2.9 ± 2.3 3268 1.9 ± 1.6 3374 3.2 ± 3.0 3444 4.1 ± 4.8 3487 1.9 ± 1.8 3505 1.7 ± 1.3 3766 9.8 ± 7.6 3891 9.7 ± 5.0 3885 (0.3 ± 2.1) 3979 3.3 ± 2.6 4073 (Aβ37) 2.4 ± 1.3 4077 (10.3 ± 6.4) 4123 (Aβ38) 0.4 ± 1.1 4135 (21.0 ± 11.2) 4331 (Aβ40) 3.0 ± 2.1 4333 (28.1 ± 11.5) 4480 6.1 ± 3.5 4623 7.4 ± 3.6 5102 9.8 ± 6.9 5102 (1.2 ± 1.0) 5675 1.1 ± 0.8 5677 (2.2 ± 1.5) 5805 1.7 ± 1.0 5821 (1.9 ± 1.2) 5864 2.3 ± 1.1 5912 2.0 ± 1.2 6657 4.2 ± 2.6 6856 1.2 ± 1.5 6975 2.4 ± 2.2 7183 3.2 ± 2.3 7249 2.9 ± 2.2 7432 3.2 ± 2.7 7472 4.0 ± 2.9 7567 8.0 ± 3.9 7752 43.7 ± 9.9 7760 (1.8 ± 1.8) 8143 7.2 ± 2.0 8145 (0.7 ± 0.8) 8608 1.9 ± 1.8 8602 (4.3 ± 2.5) 8678 2.0 ± 1.5 8678 (3.0 ± 1.6) 8931 7.5 ± 3.4 9208 4.9 ± 3.8 9323 4.3 ± 3.5 9395 2.5 ± 2.2 10544 1.3 ± 0.7 13349 1.2 ± 1.1 * Proposed peptide ID based upon calculated mass, # based upon CSF SELDI-TOF MS analysis in a different patient population recruited at the Radboud University Nijmegen Medical Centre (unpublished results Frankfort et al)

Chapter 4.2.3

192

Figure 1. Representative amyloid β-profile in serum of a participant with Alzheimer’s Disease as obtained with SELDI-TOF MS

0

10

20

30

40

50

Inte

nsity

80004000 5000 6000 7000

4000 5000 6000 7000 8000

m/z value

6

7

1 5

2 3 4

Numbers represent identified fingerprint peptides and their m/z values: 1 = 3891, 2 = 4073 (Aβ37), 3 = 4123 (Aβ38), 4 = 4331 (Aβ40), 5 = 5102, 6 = 7752, 7=8143.

Predictors of Aβ peak intensities measured by SELDI-TOF MS Four peaks were chosen to identify possible predictors of Aβ peak intensities in serum. Peaks corresponding to m/z 4073 (Aβ37) and m/z 4331 (Aβ40) were chosen as these are Aβ peptides known from ELISA and earlier SELDI-TOF MS experiments in CSF. Peaks at m/z 3891 and m/z 7752 were chosen as these were prominent in this serum experiment and also major peaks, besides the Aβ-fingerprint, in the SELDI-TOF MS analysis in CSF as described by Lewczuk et al.18 Results of univariate and multivariate regression analyses for those four peaks are summarised in table 3. Multivariate regression analyses showed that the type of clinical diagnosis, AD/VaD/age-matched non-demented control, was not significantly related to the peak intensities for either of the four chosen peaks. After correcting for age and gender, creatinine level showed a significant relation with Aβ37 peak intensity (B= -0.006, p=0.024) and for the ABCB1 genotype C1236T a trend towards significance was shown (B=0.318, p=0.064). For Aβ40, significant relations between its intensity, age (B=0.093, p=0.012) and ABCB1 genotype G2677T/A (B=0.597, p=0.034) were found. The intensity of the peak corresponding to m/z 3891 was significantly related with the presence of an APOE ε4 allele (B=2.040, p=0.040) and for the ABCB1 genotype C1236T (B=1.352, p=0.047). For the peak at m/z 7752 neither of the tested variables were significantly related with its intensities.


193

Table 3. Results of univariate and multivariate linear regression analyses. Peak Variable Univariate Multivariate* B 95% CI p-value B 95% CI p-value 4073 Age 0.003 -1.33 – 5.63 0.903 (Aβ37) Gender -0.149 -0.64 – 0.64 0.548 Creatinine -0.005 -0.01 – 0.00 0.041 -0.006 -0.01 – 0.00 0.024 APOE ε4 0.345 -0.14 – 0.83 0.158 C1236T 0.281 -0.05 – 0.61 0.094 0.318 -0.02 – 0.65 0.064 G2677T/A 0.237 -0.07 – 0.55 0.131 C3435T 0.215 -0.13 – 0.56 0.215 Diagnosis -0.025 -0.35 – 0.30 0.880 4331 Age 0.094 0.02 – 0.17 0.010 0.093 0.02 – 0.16 0.012 (Aβ40) Gender -0.668 -1.50 – 0.17 0.116 Creatinine 0.007 0.00 – 0.02 0.091 0.009 0.00 - 0.03 0.224 APOE ε4 0.406 -0.43 – 1.24 0.337 C1236T 0.241 -0.33 – 0.82 0.407 G2677T/A 0.496 -0.06 – 1.05 0.077 0.597 0.05 – 1.15 0.034 C3435T 0.431 -0.16 – 1.02 0.149 Diagnosis 0.465 -0.08 – 1.01 0.095 0.211 -0.35 – 0.77 0.455 3891 Age -0.05 -0.22 – 0.13 0.604 Gender -1.31 -3.31 – 0.68 0.195 Creatinine -0.006 -0.03 – 0.01 0.550 APOE ε4 2.119 0.17 – 4.07 0.034 2.040 0.10 – 3.98 0.040 C1236T 1.297 -0.05 – 2.65 0.060 1.352 0.02 – 2.68 0.047 G2677T/A 0.966 -0.35 – 2.28 0.149 C3435T 0.617 -0.79 – 2.03 0.387 Diagnosis 0.006 -1.31 – 1.32 0.993 7752 Age -0.248 -0.59 – 0.91 0.150 Gender 2.704 -1.23 – 6.64 0.176 Creatinine -0.01 -0.05 – 0.03 0.577 APOE ε4 -1.168 -5.09 – 2.75 0.556 C1236T -0.560 -3.26 – 2.14 0.682 G2677T/A -0.773 -3.39 – 1.85 0.559 C3435T -1.091 -3.87 – 1.69 0.438 Diagnosis -1.77 -4.34 – 0.81 0.178 * Corrected for age en gender

Chapter 4.2.3

194

Discussion Our results show an Aβ profile in serum consisting of Aβ37, Aβ38 and Aβ40, whereas Aβ39 and Aβ42 were undetectable with SELDI-TOF MS. Most pronounced other peaks are at molecular masses 3766, 3891, 4623, 5102, 7567, 7752, 8143 and 8931 Da and may theoretically represent truncated or dimeric forms of Aβ peptides. The Aβ peptide profile in serum partly resembles the one we found earlier in CSF. The selected Aβ peaks were, however, unrelated with clinical diagnosis, but age, creatinine level and ABCB1 genotypes appeared as significant co-variates. We showed the Aβ37, Aβ38 and Aβ40 in serum of patients suffering from AD, VaD or age-matched non-demented controls. Lewczuk et al.19 showed also Aβ39 and Aβ42 in addition to Aβ37, Aβ38 and Aβ40 in plasma of patients with AD or controls using gel electrophoresis. Aβ40 was the most abundant Aβ-peptide among the fingerprint peptides in our serum set and this is in concordance with the sample set as described by Lewczuk et al.19 using gel electrophoresis and as described by Mehta et al.8 measuring Aβ38, Aβ40 and Aβ42 using ELISA. Differences in methodology may explain why Aβ39 and Aβ42 were not detected in our sample set. The peak intensities of Aβ39 and Aβ42 are the lowest among the fingerprint peptides in CSF and ELISA experiments have learned us that the concentration of Aβ1-42 is a 120-fold lower in plasma than in CSF8. This indicates that the concentrations may be too low to be detected by SELDI-TOF MS. In our serum sample set the peak at a mass of 7752 was most pronounced and its intensity was approximately 14 times higher as compared to Aβ40. To our knowledge this is the first report on this peptide in serum or plasma. This peak could theoretically correspond to a possible dimeric form of an Aβ peptide. The serum Aβ fingerprint profile thus coincides partly with CSF measurements with SELDI-TOF MS in AD and control patients17,18 and in VaD patients (unpublished results Frankfort et al.). Aβ peaks can thus be measured in CSF and serum, but it remains the question where the Aβ-peptides originate from and if and how they are transported between CSF and blood. Platelets have the highest level of APP expression among peripheral tissue and are able to produce all APP fragments found in neurons. Thus platelets can process APP via the amyloidogenic pathway (involving β-secretase and γ-secretase) resulting in Aβ formation. Platelets produce mainly Aβ40 and cell experiments did not reveal production of Aβ42 thus far.38 The existence of the Aβ-fingerprint in blood can thus partly be explained by the fact that γ-secretase has seven potential cleavage sites39-41 and it is possibly active in platelets38. However, as cell experiments did not reveal production of Aβ42 in platelets38 and Aβ42 is detected in blood by ELISA, it might originate from the brain. Aβ-ApoE and Aβ-ApoJ complexes may cross the BBB42 and plasma Aβ may thus theoretically be incorporated into brain amyloid plaques. Whether there is a dynamic equilibrium between Aβ in CSF and


195

blood remains to be proven. A study described by Mehta et al.14 did not show congruent results between both Aβ40 and Aβ42 in plasma and CSF. Our results could not differentiate between clinical diagnoses on the basis of serum Aβ levels, which is in agreement with other cross-sectional studies that investigated differences between AD and control patients. Lewczuk et al.19 did not find significant differences in Aβ levels of AD and controls as measured by gel electrophoresis. Mehta et al.8 did not show significant differences between AD and controls in Aβ38, Aβ40 and Aβ42 as measured by ELISA. Fukomoto et al.20 did not find differences between AD and controls for Aβ40 and Aβ42 as measured by ELISA. We also investigated a group of patients suffering from VaD. Serum Aβ levels were not associated with this diagnosis. In the study of Gurol et al.43 plasma Aβ40 levels were independently associated with the extent of white matter hyperintensity in subjects with AD, Mild Cognitive Impairment or cerebral amyloid angiopathy and thus appeared as a possible risk factor or biomarker for microvascular damage in these patients. In addition, van Dijk et al.22 showed higher Aβ40 and Aβ42 levels with more lacunar infarcts and white matter lesions in elderly subjects carrying an APOE ε4 allele. Although we did not observe a correlation between VaD and serum Aβ levels, peripheral Aβ may still play a role in VaD as it was related to the extent of white matter hyperintensity and lacunar infarcts. Most cross-sectional studies did not reveal significant differences in peripheral Aβ levels between AD and controls44. However, recently published results of a longitudinal study indicate increased baseline plasma Aβ40, especially when combined with decreased baseline Aβ42 levels, to be a risk factor for the development of AD15. In this study, plasma was drawn when the participants were included, years before the onset of dementia. It is known that during the disease process peripheral Aβ levels decrease13 and this may explain the discrepancies between longitudinal studies and cross-sectional studies. It also points out the importance of performing longitudinal studies in addition to cross-sectional ones. Earlier studies described correlations between age20, creatinine levels21 and Aβ levels. Our results from the linear regression analyses confirmed these results. We were also interested in the relation between ABCB1 genotypes and Aβ peak intensities. The ABCB1 gene, encoding for P-gp, is highly polymorphic and our results showed the SNP G2677T/A to be a significant co-variate for the intensity of the peak at a mass of 4331 (Aβ40). The SNP C1236T was a significant co-variate for the intensity of the peak at a mass of 3891. Peak intensities increased with the number of mutant alleles for this SNPs. Thus far investigations showed homozygous T-allele carriers at position C3435T had on average more than twofold lower intestinal P-gp expression levels compared to the CC genotype and a 1.5 fold lower P-gp expression level was shown in human kidney45. As this silent polymorphism is in linkage disequilibrium with G2677T/A29, this possibly explains differences in function and may explain why G2677T/A and not C3435T was related to

Chapter 4.2.3

196

peak intensities in our patient samples. However, relationships between ABCB1 SNPs and P-gp expression and function remain unclear. In addition, in the study by Vogelgesang et al.28 no correlation was shown between ABCB1 G2677T/A and C3435T and P-gp expression at the BBB, assayed with immunostaining. This points out that these ABCB1 SNPs are possibly not related to P-gp expression or function at the BBB. Thus, P-gp can interact in two ways with serum Aβ levels. It can act as a transporter at the BBB, from brain into the capillary blood vessel, and it can act as a renal transporter, located at the luminal membrane of proximal tubule cells in the kidney, with transport from blood into urine46. Our results may reflect renal processes instead of a transporter function at the BBB. Both processes may be involved with a net effect of an increase in serum Aβ level. Our results confirm a possible role for P-gp as an Aβ transporter, however it remains uncertain where it exerts its major effects (kidney or BBB). How P-gp is involved in the pathogenesis of dementia requires further research. As longitudinal studies did show plasma Aβ as a risk factor for the development of Alzheimer’s Disease13,15 and the origin of peripheral Aβ and the equilibrium with CSF and brain Aβ is yet unknown, future studies are warranted into this subject. The roles of newly identified Aβ peptides by SELDI-TOF MS, P-gp as an Aβ transporter and the role of ABCB1 genotypes in this function deserve further attention in longitudinal studies in people at risk for dementia. In conclusion, this study shows an Aβ-profile in serum, consisting of Aβ37, Aβ38 and Aβ40 among additional unidentified Aβ peptides, partly resembling the Aβ-profile as earlier determined in CSF. We found no significant correlation between type of clinical diagnosis and serum Aβ intensities, however, age, creatinine levels, the presence of the APOE ε4 allele and certain ABCB1 genotypes may predict serum Aβ levels. P-gp may serve a role in Aβ transport, possibly located at the kidney, the BBB or at both sites, as indicated by a significant correlation between ABCB1 genotypes and serum Aβ levels.

Acknowledgements Valerie Doodeman and Remco Bakker are kindly acknowledged for their help in genotyping ABCB1 and APOE. References 1. Ferri CP, Prince M, Brayne C, Brodaty H, et al. Global prevalence of dementia: a delphi consensus study.

Lancet 2005;366:2112-17. 2. Blennow K, de Leon MJ, Zetterberg H. Alzheimer’s Disease. Lancet 2006;368:387-403. 3. Martinez-Villa E, Murie-Fernandez M, Perez-Larraya J, Irimia P. Neuroprotection in vascular dementia.

Cerebrovasc Dis 2006;21 (suppl 2):106-117.


197

4. Lewis H, Beher D, Cookson N, et al. Quantification of Alzheimer pathology in ageing and dementia: age related accumulation of amyloid beta 42 peptide in vascular dementia. Neuropathol Appl Neurobiol 2006;32:103-118.

5. Blennow K, Hampel H. CSF markers for incipient Alzheimer’s Disease. Lancet Neurol 2003:2:605-13. 6. Andreasen N, Blennow K. CSF biomarkers for mild cognitive impairment and early Alzheimer’s disease.

Clin Neurol Neurosurg 2005;107: 165-73. 7. Schoonenboom NS, Mulder C, van Kamp GJ, et al. Amyloid β 38, 40, and 42 species in cerebrospinal fluid:

more of the same. Ann Neurol 2005;58:139-142. 8. Mehta PD, Pirtilla T. Increased cerebrospinal fluid Aβ38/Aβ42 ratio in Alzheimer’s Disease.

Neurodegenerative Dis 2005;2:242-245. 9. Andreasen N, Minthon L, Davidsson P, et al. Evaluation of CSF tau and CSF abeta 42 as diagnostic markers

for alzheimer’s disease in clinical practice. Arch Neurol 2001;58:373-79. 10. Parnetti L, Lanari A, Saggese E, et al. Cerebrospinal fluid biochemical markers in early detection and in

differential diagnosis of dementia disorders in routine clinical practice. Neurol Sci 2003;24:199-200. 11. de Jong D, Jansen RWMM, Kremer HPH, Verbeek MM. CSF Amyloid β42 and phosphorylated tau can

discriminate between dementia disorders. J Gerontol 2006;61A:755-758. 12. Stefani A, Bernardini S, Panella M, et al. AD with subcortical white matter lesions and vascular dementia:

CSF markers for differential diagnosis. J Neurol Sci 2005;237:83-8. 13. Mayeux R, Honig LS, Tang MX et al. Plasma Aβ40 and Aβ42 and Alzheimer’s Disease: relation to age,

mortality and risk. Neurology 2003;61:1185-90. 14. Mehta PD, Pirtitila T, Mehta SP et al. Plasma and cerebrospinal fluid levels of amyloid β proteins 1-40 and

1-42 in Alzheimer’s Disease. Arch Neurol 2000;57:100-105. 15. van Oijen M, Hofman A, Soares HD et al. Plasma Aβ40 and Aβ42 and the risk of dementia: a prospective

case-cohort study. Lancet Neurol 2006;5:655-60. 16. Engwegen JYMN, Gast MCW, Schellens JHM, Beijnen JH. Clinical proteomics: searching for better tumour

markers with SELDI-TOF mass spectrometry. Trends Pharmacol Sci 2006;27:251-9. 17. Maddalena AS, Papassotiropoulos A, Gonzalez-Agosti C, et al. Cerebrospinal fluid profile of amyloid beta

peptides in patients with Alzheimer’s disease determined by protein biochip technology. Neurodegenerative Dis 2004;1:231-35.

18. Lewczuk P, Esselmann H, Wolfgang T, et al. Amyloid beta peptides in cerebrospinal fluid as profiled with surface enhanced laser desorption/ionization time-of-flight mass spectrometry: evidence of novel biomarkers in Alzheimer’s disease. Biol Psychiatry 2004;55:524-30.

19. Lewczuk P, Esselmann H, Bibl M et al. Electrophoretic separation of amyloid β peptides in plasma. Electrophoresis 2004;25:3336-3343.

20. Fukumoto H, Tennis M, Locascio JJ, et al. Age but not diagnosis is the main predictor of plasma amyloid β-protein levels. Arch Neurol 2003;60:958-964.

21. Arvanitakis Z, Lucas JA, Younkin LH, et al. Serum creatinine levels correlate with plasma amyloid beta protein. Alzheimer Dis Assoc Disord 2002;16:187-190.

22. van Dijk E, Prins ND, Vermeer SE et al. Plasma amyloid β, apolipoprotein E, lacunar infarcts and white matter lesions. Ann Neurol 2004;55:570-575.

23. Schinkel AH and Jonker JW. Mammalian drug efflux transporters of the ATP binding cassette (ABC) family: an overview. Adv Drug Del Rev 2003;55:3-29.

24. Borst P, Oude Elferink R. Mammalian ABC transporters in health and disease. Ann Rev Biochem 2002;71:537-92.

25. Lee G and Bendayan R. Functional expression and localization of P-glycoprotein in the central nervous system: relevance to the pathogenesis and treatment of neurological disorders. Pharm Res 2004;21:1313-30.

Chapter 4.2.3

198

26. Lam FC, Liu R, Lu P, et al. Beta-amyloid efflux mediated by p-glycoprotein. J Neurochem 2001;76:1121-1128.

27. Cirrito JR, Deane R, Fagan AM, et al. P-glycoprotein deficiency at the blood-brain-barrier increases beta-amyloid deposition in an Alzheimer’s disease mouse model. J Clin Invest 2005;115:3285-3290.

28. Vogelgesang S, Cascorbi I, Schroeder E, et al. Deposition of Alzheimer’s beta amyloid is inversely correlated with P-glycoprotein expression in the brains of elderly non-demented humans. Pharmacogenetics 2002;12:535-41.

29. Marzolini C, Paus E, Buclin T, Kim RB. Polymorphisms in human MDR1 (P-glycoprotein): Recent advances and clinical relevance. Clin Pharmacol Ther 2004;75:13-33.

30. Bosch TM, Meijerman I, Beijnen JH, Schellens JHM. Genetic polymorphisms of drug-metabolising enzymes and drug transporters in the chemotherapeutic treatment of cancer. Clin Pharmacokinet 2006;45:253-85.

31. Folstein MF, Folstein SE, McHugh PR. Mini-Mental State: a practical method for grading cognitive state of patients for the clinician. Journal of Psychiatric Research 1975;12:189-198.

32. Solomon PR, Hirschoff A, Kelly B, et al. A 7 Minute Neurocognitive Screening Battery highly sensitive to Alzheimer´s disease. Arch Neurol 1998;55:349-55.

33. McKahnn G, Drachmann D, Folstein M et al. Clinical diagnosis of Alzheimer´s disease: Report of the NINCDS-ADRDA work group under the auspices of department of health and human services task force on Alzheimer´s disease. Neurology 1984;34:939-944.

34. Erkinjunnti T. Clinical criteria for vascular dementia: the NINDS-AIREN criteria. Dementia 1994;5:189-92. 35. Koch W, Ehrenhaft A, Griesser K et al. TaqMan systems for genotyping of disease-related polymorphisms

present in the gene encoding apolipoprotein E. Clin Chem Lab Med 2002;40:1123-31. 36. Bosch TM, Doodeman VD, Smits PH, et al. Pharmacogenetic screening for polymorphisms in drug-

metabolising enzymes and drug transporters in a Dutch population. Mol Diagn Ther 2006;10:175-85. 37. Verhage F. Intelligence and age. Van Gorcum, Assen, 1964 (in Dutch). 38. Evin G, Zhu A, Holsinger D et al. Proteolytic processing of the Alzheimer’s Disease amyloid precursor

protein in brain and platelets. J Neurosci Res 2003;74:386-392. 39. Beher D, Wrigley JD, Owens AP, Shearman MS. Generation of C-terminally truncated amyloid-beta

peptides is dependent on gamma-secretase activity. J Neurochem 2002;82:563-575. 40. Murphy MP, Hickman LJ, Eckman CB. Gamma-secretase, evidence for multiple proteolytic activities and

influence of membrane positioning of substrate on generation of amyloid beta peptides of varying length. J Biol Chem 1999;247:11914-11923.


42. Zlokovic BV. Cerebrovascular transport of Alzheimer’s amyloid β and apolipoproteins J and E: possible anti-amyloidogenic role of the blood-brain-barrier. Life Sci 1996;59:1483-1497.

43. Gurol ME, Irizarry MC, Smith EE et al. Plasma β amyloid and white matter lesions in AD, MCI and cerebral amyloid angiopathy. Neurology 2006;66:23-29.

44. Irizarry MC. Biomarkers of Alzheimer Disease in plasma. NeuroRx 2004;1:226-234. 45. Schwab M, Eichelbaum M, Fromm M. Genetic polymorphisms of the human MDR1 drug transporter. Annu

Rev Pharmacol Toxicol 2003;43:285-307. 46. Fromm M. Importance of P-glycoprotein at blood-tissue barriers. Trend Pharmacol Sci 2004;25:423-429.

199

200

Conclusions and Perspectives

201


Pharmacotherapy and clinical pharmacology in geriatric patients Evaluation of pharmacotherapy in geriatric patients is important. This patient group is more vulnerable for adverse events and pharmacokinetic interactions and polypharmacy should be reduced. Undertreatment of illnesses, however, is also recognised in geriatric populations during the last years. Tempering the number of drugs and starting medication when a condition is not (optimal) treated should be balanced. Geriatric pharmacotherapy can be evaluated when patients are admitted to a diagnostic day-clinic or to a geriatric ward. In our research projects pharmacotherapy of geriatric patients was evaluated at both departments. Our retrospective investigations show that pharmacotherapy evaluation results in a mean net addition of medication per patient. Vitamin supplementation, treatment of urinary tract infections and proton-pump-inhibitor therapy are the most frequently started therapies after evaluation. Although less frequent, medication was discontinued mostly because diagnoses were not longer relevant. Future research should ideally be prospective in design and may include the use of algorithms, for example the Medication Appropriateness Index (MAI)1, to obtain insights into the appropriateness of medication. Patients aged above 65 years are often excluded from clinical pharmacological research. We showed selected polymorphism in exon 12, 21 and 26 of the ABCB1 gene, encoding for the efflux pump P-glycoprotein, to be unrelated to the steady-state digoxin clearance in geriatric patients. This study is an example of clinical pharmacological research in geriatric patients. As the percentage of elderly in the total population increases, this patient group should also be included in future studies regarding clinical pharmacology.

Pharmacotherapy and clinical pharmacology in dementia The group of cholinesterase inhibitors is currently registered for the indication of mild-to-moderate severe Alzheimer’s disease2. Adverse events of cholinesterase inhibitors, primarily of gastrointestinal order, may lead to discontinuation. In addition, treatment effects differ substantially between individuals3. Before therapy inititation it is unknown which patients respond to cholinesterase inhibitor therapy. Rivastigmine is one of the cholinesterase inhibitors. This thesis describes studies that investigate discontinuation of rivastigmine, treatment effects of rivastigmine on cognition, performance of activities in daily living, and identification of characteristics of responders to rivastigmine. These studies were all performed in outpatients treated with rivastigmine in routine clinical practice. Rivastigmine is primarily discontinued within the first 6 months of therapy because of adverse events, after 6 months mainly for therapy ineffectiveness. Absence of nurse support and not tolerating the minimal effective daily dose of twice daily 3 mg results


202

in an increased risk for discontinuation of rivastigmine within 6 months after starting therapy. Rivastigmine shows a modest effect on cognition, activities of daily living and memory-related behaviour. Rivastigmine does not have clinically relevant treatment effects on disruptive behaviour and depressive behaviour worsened during the first 6 months of treatment. Patients with a baseline Mini Mental State Examination score ≤ 19, however, show significant and larger responses to rivastigmine therapy. Differences in pharmacokinetics were not related to rivastigmine treatment outcome in Alzheimer’s disease patients and could not explain treatment variability in individuals. Future research will aim to develop disease-modifying therapies for dementia in addition to the currently available symptomatic therapies. Disease-modifying therapies interact with biological mechanisms that underlie the clinical diagnosis of dementia, for example β-secretase and γ-secretase. β-secretase has been recognised as a therapeutic target, and the challenge is to generate β-secretase inhibitors as a compound for therapeutic purposes4. Currently, clinical studies are conducted with γ-secretase-inhibitors. The γ-secretase-inhibitor MK-0752 significantly lowered cerebrospinal fluid concentration of amyloid β-40 in humans5. Another γ-secretase-inhibitor LY450139 showed decreases in plasma and cerebrospinal fluid amyloid β-40 concentrations in humans6. Another therapeutic target that is currently investigated is the amyloid β immunotherapy. A phase II study with AN1792 had to be terminated because of meningoencephalitis in a subset of patients. In this study, however, it was shown that cerebrospinal fluid tau levels decreased during treatment with AN17927. Future research is necessary to obtain more insights into these new therapies for dementia.

Biomarker research in dementia Dementia is a rapidly growing health and socioeconomic problem worldwide, as it is estimated that 24 million people have dementia nowadays and that this will increase to 42 million in 2020 and 81 million by 20408. With the future possibilities of disease-modifying therapies it is becoming important to diagnose dementia preferably in a pre-clinical stage or as early as possible in the clinical process, when pathological mechanisms have already been started. We showed that selected genotypes in exon 12, 21 and 26 of the ABCB1 gene, encoding for the efflux drug-transporter P-glycoprotein, are not useful as biological markers for different types of dementia. Amyloid beta is a substrate for this efflux pump and this pump may play a possible role in the dementia pathogenesis. These ABCB1 single nucleotide polymorphism (SNP), however, were not significantly different between dementia patients and age-matched non-demented control patients, nor between different types of dementia. Furthermore, we used Surface-Enhanced Laser Desorption/Ionisation-Time of Flight Mass Spectrometry (SELDI-TOF MS) and immunocapture to investigate amyloid β profiles in


203

cerebrospinal fluid and serum of patients with Alzheimer’s disease, vascular dementia and non-demented control patients. Certain peaks in cerebrospinal fluid, as obtained with SELDI-TOF MS, were significantly different between those patient groups. In serum, however, no significantly different peaks intensities were obtained between the three patient groups. Investigations in serum showed that the SNP in exon 21 (G2677T/A) of ABCB1 was significantly related to the peak intensity of amyloid β-40 and the SNP in exon 12 (C1236T) of ABCB1 was significantly related to a peak at molecular mass 3891, corresponding to an yet unidentified amyloid-peptide. Amyloid β protein and (phosphorylated) tau are the most frequently researched proteins in cerebrospinal fluid and their combination or ratio may result in the development of a diagnostic test for Alzheimer’s disease. Prospective studies should follow mild cognitive impairment patients for a minimum of 5 years, be cohorent in study design and these markers should be applicable not only in clinical trial settings, but also in daily clinical care. Future research is necessary to investigate the role of newly identified amyloid β peptides as possible biomarkers for different types of dementia. Biomarker analysis in serum/plasma would be an enormous advantage, as a venapucture is evidently less invasive than lumbar puncture. Analysis in blood may be a future opportunity for the development of a diagnostic test as the amyloid β profile in serum partly resembles the profile obtained in cerebrospinal fluid. More insights should be gained in the future into the transport mechanisms and the dynamic equilibrium of amyloid β between brain, cerebrospinal fluid and blood. We showed certain ABCB1 genotypes to be related to peak intensities in experiments in serum of included patients. P-glycoprotein is an amyloid β- transporter. The role of P-glycoprotein in the pathogenesis of dementia and the role of ABCB1 genotypes needs to be further elucidated. Positron Emission Tomography (PET), using 11C-verapamil, can be used to investigate the functional relevance of P-glycoprotein9. In the future it may be possible to image amyloid by PET. Currently, tracers are evaluated for amyloid imaging10-12 and it may become possible to measure both amyloid load and P-gp function at the blood brain barrier and obtain more insights into the role of P-glycoprotein in amyloid β transport and the pathogenesis of dementia. In conclusion, research into the balance between reducing polypharmacy and preventing undertreatment, and treatment of and biomarker research in dementia are important in geriatric medicine. Future research should intend to further optimise pharmacotherapy in geriatric patients and to develop disease-modifying therapy and diagnostic tests for diagnosing dementia in an early stage.


204

References 1. Hanlon JT, Schmader KE, Samsa GP, Weinberger M, Uttech KM, Lewis IK, et al. A method for assessing

drug therapy appropriateness. J Clin Epidemiol 1992;45:1045-51. 2. Blennow K, de Leon MJ, Zetterberg H. Alzheimer´s Disease. Lancet 2006;368:387-403. 3. Rockwood K, MacKnight C. Assessing the clinical importance of statistically significant improvement in

anti-dementia drug trials. Neuroepidemiology 2001;20:51-56. 4. Citron MO. Beta-secretase as a therapeutic target. Alzheimer’s Dementia 2006;2:3 (Suppl 1) [abstract O1-

06-04]. 5. Rosen LB, Stone JA, Plump A, et al. The gamma secretase inhibitor MK-0752 acutely and significantly

reduces CSF Abeta40 concentrations in humans. Alzheimer’s Dementia 2006;2:3 (Suppl 1) [abstract O4-03-02]

6. Siemers ER, Quinn JF, Kaye J, et al. Effects of a γ secretase inhibitor in a randomised study of patients with Alzheimer’s disease. Neurology 2006;66:602-604.

7. Schenk DB, Seubert P, Grundman M, Black R. Aβ immunotherapy: lessons learned for potential treatment of Alzheimer’s disease. Neurodegenerative Dis 2005;2:255-260.

8. Ferri CP, Prince M, Brayne C, Brodaty H, et al. Global prevalence of dementia: a delphi consensus study. Lancet 2005;366:2112-17.

9. Sasongko L, Link JM, Muzi M, et al. Imaging P-glycoprotein transport activity at the human blood-brain barrier with positron emission tomography. Clin Pharmacol Ther 2005;77:503-14.

10. Wu C, Pike VW, Wang Y. Amyloid imaging: from benchtop to bedside. Curr Top Dev Biol 2005;70:171-213.

11. Ono M, Kawashima H, Nonaka A. Novel benzofuran derivatives for PET imaging of beta-amyloid plaques in Alzheimer’s disease brains. J Med Chem 2006;49:2725-30.

12. Zeng F, Southerland JA, Voll RJ, et al. Synthesis and evaluation of two 18F-labeled imidazo[1,2-a] pyridine analogues as potential agents for imaging beta-amyloid in Alzheimer’s disease. Bioorg Med Chem Lett 2006;16:3015-18.

Summary

205

Summary

Introduction The number of elderly patients that visit the department of Geriatric Medicine is growing due to the increase of the number of elderly in the population. Geriatric patients can be characterised by complex co-morbidity and subsequent use of multiple drugs. This may lead to an increase in drug-drug interactions and incidence of adverse events. It is thus important to reduce polypharmacy. It is currently recognised, however, that undertreatment is also apparent in geriatric patients. Evaluation of medication in the elderly should aim to reduce polypharmacy on the one hand and prevent undertreatment on the other. Geriatric medicine can be subdivided into somatic and psychogeriatric divisions. Clinical pharmacological research in the geriatric population is still in its infancy. Patients aged over 65 are frequently excluded from this type of clinical research. Research into the diagnostics of different types of dementia (Alzheimer’s disease, vascular dementia, Lewy body disease, frontotemporal disease) and treatment of dementia are of primary importance within the field of psychogeriatric medicine. Challenges in this field are optimisation of the current symptomatic therapy with cholinesterase inhibitors, the development of disease-modifying therapies and diagnosing dementia as early as possible when the biological processes have already started. Biomarker research in body fluids, for example cerebrospinal fluid and serum/plasma, plays an important role in early diagnostics. This thesis deals with pharmacotherapy (polypharmacy and undertreatment; rivastigmine treatment in dementia) and clinical pharmacological investigations in geriatric patients (digoxin, rivastigmine) on the one hand and the development of biological markers in body fluids (based upon genetics or proteomics) for early dementia diagnostics on the other hand. Pharmacotherapy of geriatric patients Patients visit the geriatric diagnostic day clinic for a complete geriatric assessment including physical examination, cognitive screening and pharmacotherapy evaluation. In chapter 1.1 a retrospective research is described that investigated 702 patients in 2002 for their first visit to the day-clinic. Vitamin supplementation was predominantly started at the day-clinic. Reasons for discontinuation of medication were a not longer relevant diagnosis (39% of discontinued medication), adverse events (33%) and better pharmacotherapeutical options (22%). 69% of the started medication was for reason of a newly diagnosed illness/condition. A mean net addition of 0.8 drug per patient was seen after visiting the day-clinic. In chapter 1.2, 258 patients admitted to the geriatric ward in 2002 are retrospectively analysed. Pharmacotherapy and patient characteristics were compared to 724 patients admitted to the same ward in 1985. Both in 1985 and 2002 an increase in the number of

Summary

206

drugs used was seen during admittance, mainly because of vitamin supplementation. Compared to 1985 admitted patients receive more medication resulting from increases in the use of preventive medicine, introduction of new medication groups, and guidelines advising combination therapies. Digoxin Chapter 2.1 describes the relation between polymorphism of ABCB1, encoding for the efflux pump P-glycoprotein, and the pharmacokinetics of digoxin. Selected genotypes and haplotypes were not significantly related with steady-state digoxin clearance in a geriatric population using digoxin for the medical indications atrial fibrillation or congestive heart failure. Rivastigmine The “cholinergic” hypothesis aims that cognitive deterioration and behavioural problems in dementia are due to a cholinergic deficit. Use of rivastigmine, an acetylcholinesterase inhibitor, leads to an increased availability of acetylcholine in cholinergic neurons of the brain. Rivastigmine is a symptomatic drug registered for the treatment of mild-to-moderate Alzheimer’s disease. In clinical practice, however, it is recognised that patients often have to discontinue therapy because of adverse events. In addition, a large variability in treatment effects is apparent in routine clinical practice. It is of primary importance to optimise symptomatic therapy in Alzheimer’s disease and to identify responders to rivastigmine therapy. Pharmacokinetic differences may play a role in the differences in effectiveness between individual patients. Therapy effectiveness should be measured in different domains (cognition, activities in daily living, behaviour). In additon, the long term effects of rivastigmine remain largely unknown. A validated and sensitive bioanalytical method is necessary to measure plasma drug concentrations and are important for further clinical pharmacological research. Chapter 3.1.1 describes the development and validation of a liquid chromatography assay coupled to tandem mass spectrometry to measure rivastigmine and its metabolite NAP226-90. This method is used in the clinical pharmacological investigations described in chapter 3.2.4. In chapter 3.2.1 discontinuation of rivastigmine in routine clinical practice is described. Within 6 months after starting rivastigmine, 40% of patients discontinued therapy, primarily because of adverse events (59%). In addition, not tolerating the minimal effective dose of twice daily 3 mg resulted in a 12.3 times higher chance of discontinuation and a 2.4 times higher chance of discontinuation was found when patients were not supported by a nurse. After the first 6 months, patients primarily discontinued because of insufficient therapy effectiveness. Chapter 3.2.2 describes rivastigmine effectiveness on cognition, activities of daily living and behaviour (memory related behaviour, disruptive behaviour,

Summary

207

depressive behaviour) in 84 Alzheimer outpatients after 6 months of therapy. Cohen’s d (effect size) was calculated to investigate the clinical relevance in addition to the significance level. In clinical practice rivastigmine showed a modest effect on cognition, activities of daily living and memory related behaviour. Unfortunately, rivastigmine showed no effect on disruptive behaviour and depressive behaviour worsened after 6 months of therapy. Patients that contined rivastigmine therapy after 6 months were evaluated each 6 months. Most individual items on the behavioural checklist showed stabilisation during 30 months. Cognition and activities of daily living showed a gradual decline during 30 months. Chapter 3.2.3 describes investigations into reactive cognitive subdomains and responders to rivastigmine therapy. Rivastigmine showed effectiveness in non-memory related cognitive decline (language, attention, abstract thinking, perception). Rivastigmine showed to be more effective in more severe Alzheimer’s disease. Chapter 3.2.4 describes the development of a population pharmacokinetic model for rivastigmine and its metabolite NAP226-90 in patients suffering from dementia. Data were used of a prospective pharmacokinetic study in patients with Alzheimer’s disease or Lewy body dementia in whom therapy was evaluated every 6 months. Rivastigmine and NAP 226-90 pharmacokinetics were best described with one-compartment models and with Michaelis-Menten elimination for rivastigmine and first-order elimination for the metabolite. This model was used to relate the pharmacokinetics of both rivastigmine and metabolite to therapy effectiveness. No relation was found between maximal plasma concentrations as well as the area under the concentration versus time curve of both rivastigmine and NAP 226-90 and test results in the domains cognition, activities of daily living and behaviour. Biomarkers for dementia An early diagnosis of dementia is important for further treatment of the disease. Currently available medication is symptomatic, but research into disease-modifying therapy is ongoing. When these therapies become available it may be possible to start treatment when biological processes underlying the clinical diagnosis of dementia have already been started. Biomarkers (genetic or proteomic) may serve a future role in dementia diagnosis. The pathogenesis of Alzheimer’s disease is characterised by amyloid plaques, existing of amyloid β peptides, and tangles existing of phosphorylated tau protein. Chapter 4.1.1 describes the distribution of APOE genotypes and allele frequencies in geriatric outpatients. The frequency of the APOE ε4 allele, a risk factor for Alzheimer’s disease, was significant higher in patients with mild cognitive impairment compared to age-matched control patients; a trend to significant higher frequenties of the APOE ε2 allele, that would be preventive for Alzheimer’s dementia, was shown in age-matched control patients compared to Alzheimer patients. In chapter 4.1.2 frequencies of the ABCB1 gene, encoding for the efflux pump P-glycoprotein, are described in the same population. ABCB1 genotype and

Summary

208

haplotype frequencies in patients with a mean age of 81.7 years are comparable to earlier described younger populations. Amyloid β peptide is a substrate for the efflux pump P-glycoprotein. Frequencies of ABCB1 genotypes en haplotypes were not significantly different between patients with dementia and age-matched control patients and between patients with different types of dementia. Chapter 4.2.1 provides an overview of the recent literature into amyloid β protein and tau as biomarkers for dementia. Measuring both proteins in cerebrospinal fluid may assist in distinguishing between patients with a stable or a progressive type of mild cognitive impairment with a reasonable sensitivity and specificity. In addition, progress has been made in the use of both proteins in the differential diagnosis of Alzheimer’s disease. A lumbar puncture is not regularly performed in clinical practice and therefore amyloid β has also been investigated in plasma as a biomarker for dementia. Chapter 4.2.2 describes investigations into fragments of amyloid β in cerebrospinal fluid as possible biomarkers for dementia by using a new technique (Surface-Enhanced Laser Desorption/Ionisation Time of Flight Mass Spectrometry). Cerebrospinal fluid of groups of Alzheimer’s disease patients, patients with vascular dementia and (neurological) control patients has been investigated. New peaks, corresponding to amyloid β fragments, were found and some of these intensities were different between patients with Alzheimer’s disease and vascular dementia. Fragments corresponding to those peaks may serve as future biomarkers. Additional research and validation are necessary. Chapter 4.2.3 describes prospective research into patients with Alzheimer’s disease, vascular dementia and non-demented age-matched control patients. Using the same method as described in chapter 4.2.2, several fragments of amyloid β in serum were investigated as biomarker. The amyloid β profile in serum was partly comparable to the profile obtained in cerebrospinal fluid. No significant differences in peak intensities between different types of dementia were shown in serum. A significant relation was shown, between presence of polymorphism of the ABCB1 gene, encoding for the efflux pump P-glycoprotein, and the intensities of certain peaks, corresponding to fragments of amyloid β. Hopefully, this thesis has contributed to insights into the balance of reducing polypharmacy and prevention of undertreatment, and into clinical pharmacological investigations in geriatric patients. Research into dementia constitutes a primary part of research in geriatric patients. This thesis provides further insights into rivastigmine effectiveness in clinical practice and in patients that respond to this symptomatic therapy. Future research into disease-modifying therapy stays necessary. Studies into biomarkers, genetics or proteomics based, described in this thesis may add to current knowledge in the field of early diagnostics of dementia.

209

210

Samenvatting

211

Samenvatting Introductie Het aantal oudere patiënten dat gezien wordt op een geriatrische afdeling blijft stijgen door de toename van 65-plussers in de bevolking, de zogenaamde vergrijzing. Geriatrische patiënten worden gekenmerkt door veelal complexe co-morbiditeit en tengevolge daarvan het gebruik van meerdere geneesmiddelen. Het is bekend dat gebruik van meerdere geneesmiddelen kan leiden tot stijging van het aantal geneesmiddeleninteracties en bijwerkingen. Het is dus belangrijk om het aantal voorgeschreven geneesmiddelen te beperken. Echter, het wordt steeds duidelijker dat er tevens sprake kan zijn van onderbehandeling. De uitdaging in de medicamenteuze behandeling van deze patiëntengroep wordt dan ook gekenmerkt door het vinden van de juiste balans tussen saneren van medicatie en preventie van onderbehandeling. Binnen de geriatrische geneeskunde wordt een onderscheid gemaakt in de somatische geriatrie en psychogeriatrie. Klinisch farmacologisch onderzoek bij geriatrische patiënten staat nog in de kinderschoenen. Frequent worden patiënten boven de leeftijd van 65 jaar geëxcludeerd van dit type onderzoek. Binnen de psychogeriatrie neemt onderzoek naar diagnostiek van verschillende soorten dementie (o.a. ziekte van Alzheimer, vasculaire dementie, dementie met Lewy lichaampjes, frontotemporale dementie) en behandeling van dementie een belangrijke plaats in. Uitdagingen in dit onderzoeksveld zijn het optimaliseren van de huidige symptomatische therapie met cholinesterase remmers, het ontwikkelen van ziekte-modificerende therapie en diagnostiek van de verschillende typen dementie in een vroeg stadium, bijvoorbeeld door middel van biologische markers in lichaamsvloeistoffen zoals liquor of bloed. Het onderzoek dat beschreven wordt in dit proefschrift richt zich enerzijds op de farmacotherapie (polyfarmacie en onderbehandeling; rivastigmine behandeling bij ziekte van Alzheimer) en klinische farmacologie (digoxine, rivastigmine) in een geriatrische populatie en anderzijds richt het onderzoek zich op de ontwikkeling van biologische markers (gebaseerd op genetica of eiwitten) in lichaamsvloeistoffen voor de vroegtijdige herkenning van dementie. Farmacotherapie van geriatrische patiënten Patiënten die de afdeling geriatrie bezoeken voor een volledig geriatrisch onderzoek worden gezien op de dagkliniek. Hier wordt in één dag een lichamelijk onderzoek en een cognitieve screening uitgevoerd en wordt een overzicht van het medicatie gebruik verkregen. In hoofdstuk 1.1 is een retrospectief onderzoek beschreven onder 702 patiënten die in 2002 de dagkliniek bezochten voor een eerste afspraak. Met name vitamine suppletie werd gestart na een dagkliniekbezoek. Redenen om medicatie te stoppen waren o.a. een niet

Samenvatting

212

(langer) relevante diagnose (39% van gestopte medicatie), bijwerkingen (33%) en betere farmacotherapeutische opties (22%). Van de nieuw gestarte medicatie was dat in 69% van de gevallen vanwege een nieuw gestelde diagnose. Na een dagkliniek bezoek werd gemiddeld 0.8 geneesmiddel per patiënt toegevoegd aan het medicatie beleid. In hoofdstuk 1.2 is een retrospectief onderzoek beschreven onder 258 patiënten die werden opgenomen op de verpleegafdeling geriatrie in 2002. Farmacotherapie en karakteristieken van deze patiënten zijn vergeleken met 724 patiënten die in 1985 werden opgenomen op dezelfde geriatrische verpleegafdeling. Zowel in 1985 als in 2002 nam het aantal geneesmiddelen toe bij opname, voornamelijk vanwege vitamine suppletie. In vergelijking met 1985 gebruikten patiënten in 2002 meer geneesmiddelen, te verklaren door een toename van preventieve geneesmiddelen, het beschikbaar komen van nieuwe geneesmiddelgroepen en het bestaan van richtlijnen die combinaties van geneesmiddelen adviseren. Digoxine In hoofdstuk 2.1 wordt een onderzoek beschreven naar de invloed van polymorfismen (genetische variatie) van P-glycoproteïne (een geneesmiddelpomp) op de farmacokinetiek, in het bijzonder de klaring, van digoxine. Verschillende genotypes en haplotypes van P-glycoproteïne lieten geen significante relatie zien met de klaring van digoxine op steady-state niveau in een geriatrische patiëntenpopulatie die chronisch digoxine gebruiken voor de indicatie boezemfibrilleren en/of hartfalen. Rivastigmine De “cholinerge” hypothese stelt dat de cognitieve achteruitgang en gedragsproblematiek bij dementie patiënten te wijten zijn aan een cholinerg tekort. Rivastigmine is een acetylcholinesterase remmer en zorgt ervoor dat acetylcholine minder snel wordt afgebroken waardoor er langer acetylcholine in de synaptische spleet aanwezig blijft. Rivastigmine is een symptomatisch geneesmiddel dat geregistreerd is voor de behandeling van milde-tot-matig ernstige vormen van Alzheimer dementie. In de praktijksituatie blijkt echter dat veel patiënten de bijwerkingen niet kunnen verdragen en genoodzaakt zijn om te stoppen met therapie. Daarnaast is er veel variatie in effect van cholinesterase remmers in het algemeen. De uitdaging is dus om de symptomatische therapie van Alzheimer patiënten te optimaliseren en te onderzoeken wat de patiëntenkarakteristieken van een “responder” zijn en of farmacokinetische aspecten een verklaring kunnen zijn voor verschillen in effectiviteit tussen individuen. Therapie effectiviteit moet bij voorkeur gemeten worden in meerdere domeinen met behulp van vragenlijsten t.a.v. gedragsproblematiek en uitvoeren van activiteiten in het dagelijks leven en een cognitieve screening. Daarnaast is er beperkte kennis met betrekking tot de lange-termijn effectiviteit van rivastigmine.

Samenvatting

213

De basis van klinisch farmacologisch onderzoek wordt gevormd door de beschikbaarheid van een gevalideerde en gevoelige bioanalytische methode om nauwkeurig concentraties te meten in plasma. In hoofdstuk 3.1.1 wordt de ontwikkeling en validatie beschreven van een methode om rivastigmine en de metaboliet NAP226-90 te meten in plasma met behulp van vloeistofchromatografie gekoppeld aan massaspectrometrische detectie. Deze methode is toegepast voor het klinisch farmacologische onderzoek dat beschreven wordt in hoofdstuk 3.2.4. In hoofdstuk 3.2.1 wordt een onderzoek beschreven naar het stoppen van rivastigmine in de klinische praktijk. Binnen 6 maanden na start van rivastigmine therapie stopten 40% van de patiënten met therapie, hoofdzakelijk vanwege het optreden van bijwerkingen (59%). Daarnaast bleek het niet behalen van de minimale effectieve dosering van 2 maal daags 3 mg te resulteren in een 12.3 maal zo grote kans om te stoppen en een 2.4 keer zo hoge kans om te stoppen werd gevonden indien patiënten niet begeleid werden door een gespecialiseerde verpleegkundige. Na de eerste periode van 6 maanden stopten patiënten omdat onvoldoende effectiviteit (cognitie, uitvoeren activiteiten in het dagelijks leven, gedrag) werd behaald. In hoofdstuk 3.2.2 worden rivastigmine effecten beschreven op de cognitie, uitvoering van activiteiten in het dagelijks leven en gedrag (geheugen gerelateerd gedrag, storend gedrag en depressief gedrag) bij 84 poliklinische Alzheimer patiënten die na 6 maanden therapie werden geëvalueerd. Cohen’s d (effect grootte) werd berekend om een inzicht te verkrijgen in de klinische relevantie in aanvulling op het significantie niveau. In de klinisch praktijk situatie laat rivastigmine gedurende 6 maanden een middelmatig effect zien op cognitie, uitvoeren van activiteiten en geheugen gerelateerd gedrag t.o.v. niet behandelde Alzheimer patiënten. Rivastigmine had helaas geen effect op storend gedrag en in de eerste 6 maanden van therapie nam het depressieve gedrag zelfs toe. Patiënten die na 6 maanden rivastigmine therapie continueerden werden gevolgd door middel van 6-maandelijkse evaluaties. De meeste individuele items van de gedragsvragenlijst lieten stabilisatie zien gedurende 30 maanden rivastigmine therapie. Voor de domeinen cognitie en uitvoering van dagelijks leven activiteiten werd gedurende 30 maanden rivastigmine therapie een langzame achteruitgang gezien. In hoofdstuk 3.2.3 wordt een onderzoek beschreven naar reactieve cognitieve subdomeinen en patiënten die responderen op rivastigmine therapie. Rivastigmine is met name effectief in de behandeling van niet-geheugen gerelateerde cognitieve achteruitgang (taal, aandacht, abstract denken en perceptie). Tevens bleek rivastigmine met name effectief bij meer ernstigere vormen van Alzheimer dementie. In hoofdstuk 3.2.4 wordt de ontwikkeling beschreven van een populatie farmacokinetisch model voor plasmaconcentraties van rivastigmine en metaboliet NAP226-90 bij patiënten met dementie. Data werden gebruikt van een prospectieve kinetiek studie, waarin patiënten met Alzheimer dementie en dementie met Lewy lichaampjes geïncludeerd werden en waarin therapie elke 6 maanden werd geëvalueerd. De farmacokinetiek van rivastigmine en NAP226-90 werd beschreven met één-compartiment

Samenvatting

214

modellen met een niet-lineaire omzetting van rivastigmine in NAP226-90 en een lineaire eliminatie van de metaboliet NAP226-90. Het ontwikkelde model is gebruikt om de farmacokinetiek van zowel moeder stof als metaboliet te relateren aan de respons op rivastigmine therapie. Er werd geen relatie gevonden tussen zowel maximale plasma concentraties en de oppervlakte onder de plasma concentratie versus tijd curves van rivastigmine en NAP 226-90 en de diverse testresultaten voor cognitie, uitvoeren van activiteiten in het dagelijks leven en gedrag. Biologische markers voor dementie Een vroegtijdige herkenning van dementie is zeer belangrijk voor het vervolgtraject van behandeling van de ziekte. De huidige medicatie mogelijkheden zijn symptomatisch van aard, maar onderzoek naar ziekte modificerende therapie is in grote omvang aanwezig. Met het beschikbaar komen van deze nieuwe groepen geneesmiddelen binnen het dementie onderzoek is het wellicht mogelijk om in de toekomst in een vroeg stadium wanneer biologische processen reeds gestart zijn te starten met therapie. Biologische markers kunnen zowel van genetische aard als van eiwit aard zijn. De pathogenese van Alzheimer dementie wordt gekenmerkt door amyloid plaques die bestaan uit amyloid β peptiden en uit kluwens die bestaan uit het gefosforyleerde tau eiwit. In hoofdstuk 4.1.1 wordt een onderzoek beschreven naar de distributie van APOE genotypes en allel frequenties in geriatrische poliklinische patiënten. Dit onderzoek liet zien dat het APOE ε4 allel, een risicofactor voor de ziekte van Alzheimer, significant vaker voorkomt in patiënten met milde cognitieve stoornissen vergeleken met controle patiënten; een trend werd gezien voor significant hogere frequenties van het APOE ε2 allel, dat een beschermende werking zou hebben, in Alzheimer patiënten versus niet-demente controle patiënten. In hoofdstuk 4.1.2 worden frequenties van het ABCB1 gen, dat codeert voor de efflux pomp P-glycoproteïne, beschreven in dezelfde populatie. Dit onderzoek liet zien dat frequenties van ABCB1 polymorfismen overeenkomen in patiënten met een gemiddelde leeftijd van 81.7 jaar met eerder beschreven jongere populaties. Het amyloid β peptide dat een belangrijke rol speelt in de pathogenese van dementie is een substraat voor de efflux pomp P-glycoproteïne. Het gen dat codeert voor deze pomp is bijzonder polymorf. De resultaten van dit onderzoek lieten geen verschillen zien in frequenties van ABCB1 genotypes en haplotype tussen dementen en niet-demente patiënten als tussen verschillende soorten dementie onderling. In hoofdstuk 4.2.1 wordt een overzicht gegeven van de recente literatuur met betrekking tot de karakteristieke eiwitten amyloid β en tau als biologische markers voor dementie. Het meten van beide eiwitten in combinatie in liquor laat zien dat het mogelijk is om progressieve patiënten met milde cognitieve stoornissen (een mogelijk voorstadium van dementie) met een redelijke sensitiviteit en specificiteit te onderscheiden van stabiele patiënten met milde cognitieve stoornissen. Daarnaast wordt steeds meer vooruitgang

Samenvatting

215

geboekt om met een combinatie van beide eiwitten in liquor onderscheid te kunnen maken tussen Alzheimer dementie en andere typen dementie. Liquor diagnostiek echter is geen algemeen goed voor de dagelijkse klinische praktijk. Daarom worden er in minder mate ook studies in plasma uitgevoerd naar amyloid β als biologische marker voor dementie. In hoofdstuk 4.2.2 wordt een onderzoek beschreven in liquor waarin met behulp van een nieuwe techniek (Surface-Enhanced Laser Desorption/Ionisation Time of Flight Mass Spectrometry) diverse fragmenten van amyloid β als biologische marker voor dementie zijn onderzocht. Liquor van groepen Alzheimer patiënten, patiënten met vasculaire dementie en (neurologische) controle patiënten werd onderzocht. Nieuwe pieken werden gevonden tijdens de massa spectrometrische detectie en de intensiteit van deze pieken was verschillend tussen patiënten met Alzheimer dementie en vasculaire dementie. Deze pieken (eiwitten) zouden dus kunnen dienen als een toekomstige biologische marker en daarvoor is aanvullend onderzoek en validatie noodzakelijk. In hoofdstuk 4.2.3 wordt een prospectief onderzoek beschreven in patiënten met Alzheimer dementie, vasculaire dementie en niet-demente controle patiënten in dezelfde leeftijdscategorie. In deze studie werd met behulp van dezelfde methode als in liquor, beschreven in hoofdstuk 4.2.2, diverse fragmenten van amyloid β in serum onderzocht als biologische marker. Het profiel in serum was gedeeltelijk overeenkomstig met het profiel in liquor. In het serum onderzoek werden echter geen statistisch significante verschillen gevonden in piekintensiteiten tussen de verschillende soorten dementie. Dit onderzoek liet wel een relatie zien tussen het aanwezig zijn van polymorfismen in het ABCB1 gen, coderend voor de efflux pomp P-glycoproteïne, en de intensiteit van bepaalde pieken, die corresponderen met fragmenten van amyloid β. Hopelijk hebben de diverse studies uitgevoerd in geriatrische patiënten bijgedragen aan inzichten in de balans met betrekking tot enerzijds sanering van medicatie en anderzijds preventie van onderbehandeling en klinisch farmacologisch onderzoek in geriatrische patiënten. Dementie onderzoek maakt een belangrijk deel uit van onderzoek in een geriatrische populatie en door de onderzoeken die beschreven zijn in dit proefschrift is aanvullend inzicht verkregen in de effectiviteit van rivastigmine in de klinische praktijk en in de patiënten die respons vertonen op deze symptomatische therapie. Toekomstig onderzoek blijft nodig naar therapie die ingrijpt in de biologische processen die ten grondslag liggen aan de klinische diagnose van dementie. De studies naar biologische markers op het gebied van genetica en eiwitten kunnen wellicht een startpunt zijn voor toekomstig onderzoek naar verbeterde (vroeg)diagnostiek en behandeling van dementie.

216

Dankwoord

217

Dankwoord

Allereerst gaat mijn dank uit naar mijn promotor Jos Beijnen. Jos, je gaf me de mogelijkheid om een geheel nieuw project op te starten binnen de ruime grenzen van jouw onderzoeksapotheek met de diverse beschikbare technieken. Je gaf me veel vrijheid bij het opzetten en uitvoeren van mijn onderzoek waardoor ik in de afgelopen 4 jaar veel heb kunnen leren. De leden van de beoordelingscommissie, Prof. Dr Ph. Scheltens, Prof. Dr J.H.M. Schellens, Prof. Dr A. de Boer, Prof. Dr M.G.M. Olde Rikkert en Dr R.W.M.M. Jansen ben ik zeer erkentelijk voor het beoordelen van het manuscript. Een nieuw klinisch project opzetten is één, maar zonder de medewerking van patiënten zou dit project niet hebben kunnen slagen. Ik wil de patiënten van de dagkliniek, polikliniek en verpleegafdeling Klinische Geriatrie van het Slotervaartziekenhuis bedanken dat ze mee hebben willen werken aan de diverse prospectieve klinische onderzoeken. Veelal hield dit in dat er een extra bloedafname noodzakelijk was, maar voor veel patiënten bleek dit een kleine moeite en was het mogelijk om in een relatief korte tijd behoorlijke patiëntenaantallen te includeren. Een nieuw project opstarten, implementeren en het uiteindelijke onderzoek doen was een behoorlijke opgave. Ik had soms het gevoel er alleen voor te staan, maar gelukkig waren er altijd mensen vanuit diverse disciplines binnen en buiten het ziekenhuis bij wie ik terecht kon met de meest uitlopende vragen. Linda Tulner en Jos van Campen, klinisch geriaters, wil ik bedanken voor hun klinische ondersteuning. Ik heb veel van jullie geleerd op het klinische vlak en met name op het voor mij zeer onbekende gebied van de dementie diagnostiek. Linda, je chocoladekoekjes tijdens de visite en onze overleg-uurtjes zal ik me nog lang herinneren! Bregje Appels, neuropsycholoog, en Ton de Boer van de disciplinegroep Farmacoepidemiologie en Farmacotherapie, Universiteit Utrecht, wil ik bedanken voor de samenwerking tijdens het onderzoek naar rivastigmine gebruik en effectiviteit in de klinische setting. Bregje, bedankt dat je me wegwijs hebt gemaakt in de neuropsychologische testen die gebruikt worden om de effectiviteit van rivastigmine te monitoren. Bedankt voor je interesse in mijn overige onderzoeken en je hulp bij het coderen van opleidingsgegevens van prospectief geïncludeerde patiënten. Daarnaast waren onze overleg uurtjes aan het eind van de vrijdagmiddag ook nog eens erg gezellig. Ton, bedankt dat ik diverse keren even bij je langs kon komen voor overleg over de methodologische achtergronden van dit onderzoek. Na een korte uitleg van mij, tekende je het probleem schematisch en was het altijd snel helder waar het om ging, wat het probleem was en wat de beste oplossing zou zijn. Ben Schmand, neuropsycholoog in het AMC, wil ik bedanken voor het meedenken met de analyse naar rivastigmine responders en respons op cognitieve

Dankwoord

218

subdomeinen en voor de positieve en snelle reacties op mijn emails en manuscript. Kees Koks, ziekenhuisapotheker, wil ik bedanken voor de ondersteuning tijdens de beginfase van mijn onderzoek en voor de antwoorden op de vele farmacotherapeutische vragen waarmee ik altijd bij je binnen kon lopen. Alwin Huitema, ziekenhuisapotheker, wil ik bedanken voor de hulp bij de NONMEM analyses en voor je snelle en grondige reacties op mijn manuscripten. De afdeling moleculaire biologie wil ik bedanken voor de hulp bij de genetica analyses. Remco bedankt voor je APOE genotyperingswerk, Valerie bedankt voor het genotyperen van P-gp en Paul Smits bedankt voor je snelle commentaar op mijn manuscripten. Liquor diagnostiek neemt een belangrijke plaats in binnen het biomarker onderzoek. Ik wil Marcel Verbeek, neurochemicus, en René Jansen, klinisch geriater, Radboud Universiteit Nijmegen Medisch Centrum, bedanken voor het beschikbaar stellen van liquor om samen biomarker onderzoek te doen met behulp van SELDI-TOF MS. Harm van Tinteren en Ninja Antonini, AvL statistici, bedankt dat ik bij jullie terecht kon voor vragen en hulp. Vele dagen heb ik doorgebracht op de dagkliniek geriatrie om patiënten te includeren en bloed af te nemen. Christa, Shinta, Earl, Dirkje, Math, Aleid, Edith, Hilly, Winny en Carol bedankt voor jullie logistieke hulp op z’n tijd, de gezelligheid en interesse in mijn onderzoek. Daarnaast wil ik Edith heel erg bedanken voor haar inzet bij het includeren van patiënten in de prospectieve rivastigmine studie. Ik kwam vooral op het lab om monsters te verwerken en later om SELDI-TOF analyses te doen. Lianda, Michel, Matthijs, Ciska, Joke, Selma, Carolien, Mariët, Rianne, Bas, Abadi, Dieuwke, Jan, Kees, Remko, Eric en Anissa bedankt voor jullie hulp en antwoorden op mijn vragen wanneer ik op het lab aan het werk was. Daarnaast wil ik Mariët en Rianne bedanken voor het opzetten van de rivastigmine analyse methode. Bas en Abadi bedankt voor jullie “bestel” hulp. Dieuwke, Jan, Kees, Remko en Eric bedankt voor jullie hulp bij het verzamelen van monsters voor de digoxine studie. Hilde bedankt voor je interesse in mijn onderzoek en het grondig nakijken van mijn analyse manuscript. Roel bedankt voor je PC hulp en andere praktische zaken. Olaf en Tessa bedankt voor de hulp bij het overnacht schudden van monsters in de koude kamer op H8. Joyce, Esther en Henny bedankt voor de hulp van het secretariaat. Naast het harde werken, heb ik de afgelopen 4 jaar een hele gezellige tijd gehad met de groep OIO’s van het Slotervaartziekenhuis en AvL/NKI. Natalie, Marie-Christine, Judith, Jeany, Tessa, Anthe, Isa, Milly, Markus, Marjolein, Monique, Elke, Sabien, Liesbeth, Roos, Corine, Robert, Carola, Joost, Susanne, Stijn, Jolanda, Annemieke, Ron, David, Maarten, Sander, Nienke en Marleen bedankt voor de gezelligheid bij gezamenlijke wijntjes, borrels in “die Rooie”, etentjes en OIO-weekendjes/uitjes, naast de serieuze zaken zoals inhoudelijke hulp en overleg over het onderzoek. Een gedeelte van mijn OIO tijd bracht ik door op kamer 15 op de oude geriatrie gang met Kristel, Bregt, Jan-Hendrik, Ellen, Liia en

Dankwoord

219

Wandena. Het was soms een dolle boel bij ons op de kamer, we hebben veel gelachen en vooral de kamer 15-borrels op vrijdagmiddag waren een verademing. Bedankt voor de gezelligheid op onze kamer en de hulp die ik van jullie kreeg in het begin bij het opzetten van onderzoek en later bij de dataverwerking. Daarna zat ik op kamertje 6 in de keet met mijn roomies Rob en Ly en sinds kort met Claudia. We hadden onszelf al gelijk een imago aangemeten, door overal op de kamer voorraad wijn neer te zetten! Succes met jullie onderzoeken!!! Gelukkig was er ook nog een leven naast het Slotervaart en onderzoek. Ik wil mijn vrienden en vriendinnen bedanken voor etentjes, drankjes in de kroeg, de leuke kerst en oud & nieuw feestjes, wijnavondjes, etc. Bedankt voor de belangstelling in mijn werk! Papa en mama bedankt dat jullie mij hebben geleerd om na te streven om datgene te bereiken wat binnen je mogelijkheden ligt, maar tegelijkertijd van het leven te genieten. Mijn broer en zussen (partners en kinderen) wil ik bedanken voor de interesse in mijn onderzoek, ondanks dat we heel verschillend zijn en totaal andere wegen zijn ingeslagen. Oma Brinkman heel erg bedankt voor je interesse in mijn onderzoek en ik vind het altijd leuk om in Dodewaard eventjes met je bij te praten. Arie en Jetty, oma Bazuin en de rest van de schoonfamilie, bedankt voor jullie belangstelling in mijn onderzoek de afgelopen paar jaren. Liia en Petra bedankt dat jullie op deze dag mijn paranimfen willen zijn! Liia, bedankt voor je zorgzaamheid, vooral in het afgelopen jaar dat ik het heel druk had. Ik zal je missen in een volgende baan, wie moet me dan uitleggen hoe LC/MS/MS werkt, en dat je wodka niet moet drinken, maar moet gebruiken om mee schoon te maken??? Petra, geheel onverwacht kwam jij als psycholoog in het klinisch geneesmiddel onderzoek terecht! Bedankt voor je steun en voor de vele ontspannende kopjes thee die we de afgelopen paar jaren samen hebben gedronken. Martijn, jij bent mijn solide, maar tevens zeer flexibele thuisbasis en dat heeft een belangrijke bijdrage geleverd aan het onderzoek en dit proefschrift. Bedankt voor je continue support! Suzanne, november 2006

220

Curriculum Vitae

221

Curriculum Vitae

Suzanne Frankfort werd geboren op 31 juli 1978 te Gendringen. In 1996 behaalde zij het VWO diploma aan het Christelijk Lyceum Veenendaal (CLV) in Veenendaal. Aansluitend werd gestart met de studie Farmacie aan de Universiteit Utrecht. Tijdens de doctoraalopleiding werd bij de afdeling Farmacie & Farmacologie van het Slotervaartziekenhuis onder leiding van Prof. Dr. J.H. Beijnen onderzoek gedaan naar medicatiegebruik in een populatie van HIV-geïnfecteerde patiënten. In december 2002 behaalde ze haar apothekersdiploma. In januari 2003 werd gestart met het onderzoek dat beschreven is in dit proefschrift. Het onderzoek werd uitgevoerd als een samenwerkingsverband tussen de afdelingen Farmacie & Farmacologie en Klinische Geriatrie van het Slotervaartziekenhuis, onder leiding van Prof. Dr. J.H. Beijnen. Tegelijkertijd werd de opleiding tot klinisch farmacoloog gevolgd onder auspiciën van de Nederlandse Vereniging voor Klinische Farmacologie en Biofarmacie. The author of this thesis, Suzanne Frankfort, was born on July 31, 1978 in Gendringen, the Netherlands. In 1996 she graduated after attending the Christelijk Lyceum Veenendaal (CLV) in Veenendaal for 6 years at pre-university level. In the same year she commenced her studies at the faculty of pharmacy of the Utrecht University. As part of the Master’s degree, she did research into the pharmacotherapy of HIV-infected patients at the department of Pharmacy & Pharmacology of the Slotervaart Hospital under supervision of Prof. Dr. J.H. Beijnen. In December 2002 she obtained her pharmacist degree. In January 2003 she started with the investigations that are described in this thesis, under supervision of Prof. Dr. J.H. Beijnen. The research was performed as a cooperation between the departments of Pharmacy & Pharmacology and Geriatric Medicine. At the same time she was trained as a clinical pharmacologist under the supervision of the Dutch Society of Clinical Pharmacology and Biopharmacy.

List of publications

222


De Maat MMR, Frankfort SV, Mathot RA, Mulder JW, Meenhorst PL, van Gorp EC, Koks CHW, Hoetelmans RM, de Boer A, Beijnen JH. Discrepancies between medical and pharmacy records for patients on anti-HIV drugs. Ann Pharmacother 2002;36:410-5. Frankfort SV, Roos JC, Franssen EJF. Iodine prophylaxis to prevent radiation damage following nuclear disasters [review in Dutch]. Ned Tijdschr Geneeskd 2003;147:1641-4. Frankfort SV, Roos JC, Franssen EJF. Iodine prophylaxis to prevent radiation damage following nuclear disasters [review in Dutch]. NVS nieuws 2003;28:18-21. Frankfort SV, Tulner CR, Knol W, van Campen JPCM, Schellens JHM, Beijnen JH. Prescription error resulting in valproic acid intoxication. J Am Geriatr Soc 2004;52:2142-3. Frankfort SV, Appels BA, de Boer A, Tulner CR, van Campen JPCM, Koks CHW, Beijnen JH. Discontinuation of rivastigmine in routine clinical practice. Int J Geriatr Psychiatry 2005;20:1167-71. Frankfort SV, Tulner CR, van Campen JPCM, Koks CHW, Beijnen JH. Evaluation of pharmacotherapy in geriatric patients after performing complete geriatric assessment at a diagnostic day clinic. Clin Drug Invest 2006;26:169-74. Frankfort SV, Appels BA, de Boer A, Tulner CR, van Campen JPCM, Koks CHW, Beijnen JH. Treatment effects of rivastigmine on cognition, performance of daily living activities and behaviour in Alzheimer’s disease in an outpatient geriatric setting. Int J Clin Pract 2006;60:646-54. Frankfort SV, Tulner CR, van Campen JPCM, Koks CHW, Beijnen JH. The effect of admission to a geriatric ward on medication use: 2002 versus 1985. Pharmacoepidemiol Drug Saf 2006;15:602-606. Frankfort SV, Doodeman VD, Bakker R, Tulner CR, van Campen JPCM, Smits PHM, Beijnen JH. ABCB1 genotypes and haplotypes in patients with dementia and age-matched non-demented control patients. Mol Neurodegener 2006;1:13.


223

Frankfort SV, Ouwehand M, van Maanen MJ, Rosing H, Tulner CR, Beijnen JH. A simple and sensitive assay for the quantitative analysis of rivastigmine and its metabolite NAP226-90 in human EDTA plasma using coupled liquid chromatography and tandem mass spectrometry. Rapid Communic Mass Spectrom 2006;20:3330-6. Frankfort SV, Appels BA, de Boer A, Tulner CR, van Campen JPCM, Koks CHW, Beijnen JH, Schmand BA. Identification of responders and reactive domains to rivastigmine in Alzheimer’s disease. Pharmacoepidemiolog Drug Saf; in press. Frankfort SV, Verbeek MM, Engwegen JYMN, van Campen JPCM, Jansen RWMM, Beijnen JH. Cerebrospinal fluid biomarkers found for Alzheimer’s disease and Vascular Dementia using Surface-Enhanced Laser Desorption Ionisation-Time of Flight Mass Spectrometry (SELDI-TOF MS). Submitted. Frankfort SV, Bakker R, Tulner CR, van Campen JPCM, Smits PHM, Beijnen JH. APOE genotype and allele distribution in geriatric outpatients and healthy volunteers. Submitted. Frankfort SV, van Campen JPCM, Tulner CR, Beijnen JH. Serum amyloid beta peptides in patients with dementia and age-matched non-demented control patients as detected by Surface-Enhanced Laser Desorption Ionisation-Time of Flight Mass Spectrometry (SELDI-TOF MS). Submitted. Frankfort SV, Tulner CR, van Campen JPCM, Verbeek MM, Jansen RWMM, Beijnen JH. Amyloid beta protein and tau in cerebrospinal fluid and plasma as biomarkers for dementia: a review of recent literature. Submitted. Frankfort SV, Huitema ADR, Tulner CR, van Campen JPCM, Beijnen JH. Population pharmacokinetics and pharmacodynamics of rivastigmine and its metabolite NAP226-90 in patients with dementia. Submitted. Frankfort SV, Keizer RJ, Huitema ADR, Doodeman VD, Tulner CR, van Campen JPCM, Smits PHM, Schellens JHM, Beijnen JH. Role of ABCB1 genotypes and haplotypes in digoxin steady state pharmacokinetics in geriatric patients. Submitted.

224