Non-subtype-selective opioid receptor antagonism in treatment of levodopa-induced motor complications in Parkinson's disease

Brief Reports

Mitochondrial Complex I and IVActivities in Leukocytes FromPatients With Parkin Mutations

Meltem Muftuoglu, PhD,1 Bulent Elibol, MD, PhD,2

Ozlem Dalmızrak, MSc,1 Ayse Ercan, MSc,1

Gulnihal Kulaksız, MD,1 Hamdi Ogus, MD, PhD,1

Turgay Dalkara MD, PhD,2 and Nazmi Ozer, PhD1*

1Department of Biochemistry, Hacettepe University,Faculty of Medicine, Ankara, Turkey; 2Department ofNeurology, Hacettepe University, Faculty of Medicine,

Ankara, Turkey

Abstract: The parkin protein functions as a RING-typeubiquitin protein ligase. Considering the possibility thatimpaired ubiquitin-proteosomal system activity may impairantioxidant defenses and enhance oxidative stress, we haveinvestigated the activity of mitochondrial respiratory en-zymes in patients with parkin gene mutations. A significantdecrease in the leukocyte complex I activity was found bothin patients with parkin mutations (62.5%) and idiopathicPD (64.5%) compared with age-matched controls (P <0.001). Complex IV activity was also decreased significantlyin idiopathic PD patients (60%), but no difference wasdetected between controls and patients with parkin muta-tions. © 2003 Movement Disorder Society

Key words: Parkinson’s disease; complex I; complex IV;mtDNA

Although relatively rare, inherited forms of Parkin-son’s disease (PD) provide a unique opportunity to dis-cover molecular mechanisms underlying this selectiveneuronal loss. Mutations in genes encoding �-synucleinand enzymes in the ubiquitin-proteosomal system, par-kin, and ubiquitin carboxy-terminal hydrolase L1, aremainly responsible for monogenic forms of familial PD,and mutations in the parkin gene account for the most

common familial form, PARK2, the autosomal-recessiveearly-onset PD.1,2 Patients with parkin mutations displayclinicopathological features similar to idiopathic PD, al-though earlier age of onset and relatively slow progres-sion with absence of Lewy body pathology vary fromthat in idiopathic PD.1,2

The parkin protein is made up of 465 amino acids, hasa molecular weight of 52 kDa,3 and has amino acidsequence homology with ubiquitin at the N-terminal endand two RING finger motifs at the C-terminal end, whichcould be responsible for adding ubiquitin to proteins.Because of these structural features, parkin functions asan E3-ubiquitin ligase. Although molecular mechanismsare yet to be disclosed, failure in proteosomal degrada-tion of specific target protein(s) may cause selectiveneuronal death by accumulation of toxic intermediates.4

Systemic mitochondrial complex I deficiency haslong been implicated in pathogenesis of idiopathic PD,possibly by generating additional oxidative stress innigral neurons,5 and there is growing evidence to linkoxidative damage to pathogenetic events induced bymutations in parkinsonism-related genes.6 Oxidativestress and glutathione depletion have been shown todecrease ubiquitin-protein conjugation, hence ubiq-uitin-dependent proteolysis. Oxidative stress has alsobeen proposed to provoke aggregation of �-synuclein.As similar oxidative mechanisms might also be oper-ating when parkin is dysfunctional, we have investi-gated the activity of mitochondrial respiratory chainenzymes in leukocytes from patients with parkin genemutations and idiopathic PD patients.

PATIENTS AND METHODS

Patients

Venous blood was collected from 10 patients from 6families with parkin gene mutations, 20 patients withidiopathic PD, and 17 age-matched control subjects with-out neurological disorders. Parkin mutations were mainlyhomozygous single deletions of exons 4 to 8, exceptcompound heterozygous for nonsense mutation in exon 8(C1032T) and deletions of exons 2 to 5 in 1 patient.7 Allpatients were taking antiparkinsonian-medication (levo-dopa [L-dopa], selegiline, and/or pergolide). Clinical fea-tures of the subjects are given in Table 1.

*Correspondence to: Prof. Dr. Nazmi Ozer, Hacettepe University,Faculty of Medicine, Department of Biochemistry, 06100, Sıhhiye,Ankara, Turkey. E-mail: [email protected]

Received 5 June 2003; Revised 23 September 2003; Accepted 23September 2003

Published online 12 December 2003 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/mds.10695

Movement DisordersVol. 19, No. 5, 2003, pp. 544–579© 2003 Movement Disorder Society

544

Isolation of Leukocyte Mitochondria

Venous blood samples (20 ml) were collected inEDTA-containing tubes and leukocytes were isolatedusing Histopaque-1119 (Sigma Diagnostic, St. Louis,MO) according to the manufacturer’s instructions. Leu-kocyte mitochondria were isolated by the method of Pichand colleagues,8 with modifications. Briefly, the leuko-cyte suspensions from control and PD patients werehomogenized with a glass-Teflon homogenizer. The ho-mogenate was centrifuged at 6,000g for 10 minutes at4°C. Mitochondria were collected from the supernatantby centrifugation at 20800g for 20 minutes, and the pelletwas resuspended in 150 mM KCl, 50 mM Tris-HCl, and1 mM EDTA, pH 7.4. The mitochondrial suspension wassonicated five times (10-second periods with 50-secondintervals). Mitochondrial membrane fragments were di-luted 1:1 in 0.25 M sucrose, 30 mM Tris, and 1.0 mMEDTA, pH 7.7, and centrifuged at 33,000g for 10 min-utes. The supernatant was then ultracentrifuged at100,000g for 60 minutes and the pellet was resuspendedin the same buffer. The final suspension was assayed forcomplex I and IV enzyme activities.

Complex I and Complex IV Enzyme Assays

Complex I activity was measured at 340 nm by mon-itoring the oxidation of NADH for 3 minutes and therotenone-insensitive form calculated. Complex IV activ-ity was measured by monitoring the oxidation of reducedcytochrome c at 550 nm.9 Each activity assay included50 �g mitochondrial protein. Protein concentrationswere determined by the Lowry method.10 Lactate dehy-drogenase enzyme activity was measured at 340 nmaccording to Beutler’s method.11 All spectrophotometricassays were carried out on a Milton Roy Spectronic 3000Array (Milton Roy, Ivyland, PA) spectrophotometer. Re-sults were analyzed statistically by one-way analysis ofvariance (ANOVA) and P values smaller than 0.01 wereconsidered to be significant.

RESULTS

Various factors such as the use of homogenates orpurified mitochondria or age differences between patientand control group may affect the mitochondrial respira-tory chain enzyme activities, and yield variable results.Because the amount of mitochondrial enzymes in bloodcells is relatively low, pure mitochondrial suspensionsmust be used when detecting these activities. Therefore,we used lactate dehydrogenase enzyme activity measure-ment (a cytosolic marker) in purified mitochondrial prep-arations to demonstrate the absence of cytosolic contam-ination. Each enzyme activity assay was carried out atleast three times on each sample to establish the linearityof the assay. Intravariability between triplicate valuesvaried between 15 and 18%.

For accurate evaluation of the complex I activity locatedin the inner mitochondrial membrane, samples with rote-none sensitivity (�74%) were taken into account. A signif-icant decrease (62.5%) in leukocyte complex I activity wasfound in patients with parkin mutations (5.95 � 3.17nmol/mg protein) compared with that in the controls(15.87 � 2.42 nmol/mg protein), and the mean activity ofcomplex I in idiopathic PD patients was 64.5% of that of thecontrol with a significant decrease (60%) in leukocyte com-plex IV activity in idiopathic PD patients. No statisticallysignificant difference was found, however, in complex IVactivity between the control group and patients with parkinmutations. Complex IV levels in controls and patients withparkin mutations were found as 13.25 � 8.95 and 10.89 �5.88 nmol/mg protein, respectively (Table 2). All patientswith parkin mutations and idiopathic PD were clearly out ofthe activity range of the control group and had low complexI activity (Fig. 1).

Both complex I (r � 0.671) and IV (r � 0.739)activities in leukocytes decrease with age in healthysubjects (Fig. 2 and 3). In contrast, no significant corre-lation between age, age of onset, and disease durationwith complex I and IV activities were found in patientswith parkin mutations and idiopathic PD patients.

TABLE 1. Clinical features of PD patients with parkin mutations, idiopathic PD patients and controls

Control Patients with parkin mutations Idiopathic PD patients

Patients (n) 17 10 20Age, yr (range) 43.1 � 11.0 (29–61) 42.7 � 10.4 (28–61) 68.4 � 8.7 (55–83)Age of onset, yr (range) 30.1 � 11.2 (16–49) 61.6 � 9.9 (47–79)Duration of the disease (yr) 13.8 � 9.8 (1–29) 7 � 2.9 (2–13)H&Y scale 2.4 � 0.6 2.3 � 0.4

Values are expressed as mean � SD.PD, Parkinson’s disease; H&Y; Hoehn and Yahr.

MITOCHONDRIAL ENYZMES IN PARKINSON’S DISEASE 545

Movement Disorders, Vol. 19, No. 5, 2003

DISCUSSION

There is convincing evidence that oxidative damage isinvolved in the pathogenesis and progression of neuro-degeneration in idiopathic PD.5 Markers of oxidativedamage have been detected in the substantia nigra of PDpatients and were found to be associated with selectivedeficiency in complex I activity and loss of reducedglutathione (GSH). Although GSH depletion is thoughtto have a major role in oxidative stress in nigral dopa-minergic neurons, the defect in complex I seems to be theprimary event, because inhibition of complex I alonecauses depletion of GSH over time12,13 and complex Ideficiency affects other tissues as well. A modest loss ofcomplex I activity is found in platelets and, althoughvariable, also in muscle and lymphocytes.14–17 Severalstudies suggest that both genetic and environmental fac-

tors may account for the systemic defect in complex Ithat, possibly by contribution of additional factors, leadsto selective nigral neuronal death in PD.14–19

Recent identification of monogenic forms of familialPD has shed light on novel aspects of pathogeneticmechanisms in nigral degeneration, such as accumula-tion of toxic protein products and dysfunction of ubiq-uitin-proteosomal system (UPS).20 Interestingly, there isgrowing evidence linking oxidative damage to the puta-tive pathogenetic events induced by mutations in parkin-sonism-related genes, such as �-synuclein and par-kin.21–23 Oxidative stress may also impede proteinubiquitination and lead to malfunction of the UPS; ubiq-uitin-protein conjugation has been shown to decrease inneuronal cells exposed to oxidative stress and GSH de-pletion. It was suggested that changes in the thiol oxida-tion state of E1 and E2 enzymes might decrease theiractivity, and GSH depletion decreases E1 enzyme activ-ity by oxidation of the sulfhydryl residue at its active sitein PC12 dopaminergic cell lines.12,13,24 On the other

FIG. 2. The relationship between the age and the activities of complexI in patients with parkin mutations, late-onset idiopathic PD patients,and control. Diamonds, control; solid circles, patients with parkinmutations; open circles, idiopathic PD patients.

TABLE 2. Complex I and Complex IV activities in leukocytes of PD patients with parkinmutations, idiopathic PD patients and controls

Activity (U/mg protein) Control PD with parkin mutations Idiopathic PD

Complex I 15.87 � 2.42 5.95 � 3.17a 5.64 � 1.79a

Complex IV 13.25 � 8.95 10.89 � 5.88 5.3 � 3.0b

Mean � SD values were calculated from three independent experiments.aP � 0.001; bP � 0.01.PD, Parkinson’s disease.

FIG. 1. Complex I and Complex IV activities in patients with parkinmutations, late-onset idiopathic PD patients and controls. Dots repre-sent individual subjects, bars are standard deviations and lines are meanvalues.

546 M. MUFTUOGLU ET AL.


hand, proteosomal inhibition was found to increase oxi-dative damage and protein nitration in neuronal cell lineswhereas antioxidant defenses were reduced.25 In linewith these findings, signs of decreased proteosomal en-zyme activity have been found in the substantia nigra ofidiopathic PD patients.26

The parkin protein, which is found to be defective inthe most common form of autosomal recessive PD, func-tions as E3-ubiquitin ligase in UPS and targets specificproteins such as O-glycosylated form of �-synuclein,synphilin-1, CDCrel-1, a synaptic vesicle-associated pro-tein, and insoluble Pael receptor protein. Various muta-tions in the parkin gene may impair substrate recognitionand disrupt interaction of these target proteins with ubiq-uitin-carrier protein (E2). These, in turn, may possiblyresult in accumulation of non-ubiquitinated substratesand overload of UPS.27–29 Considering that impairedUPS activity can impair antioxidant defenses, parkinmutations might also enhance oxidative stress and in-creased iron accumulation in the substantia nigra ofpatients with parkin mutations supports this view.21

Accordingly, we have measured complex I and com-plex IV activities in idiopathic PD patients and patientscarrying parkin gene mutations and the results demon-strate clearly that leukocyte complex I activity is de-creased significantly (P � 0.001). No significant de-crease was found, however, in complex IV activity inpatients with parkin mutations, whereas idiopathic PD

patients showed a significant decrease (P � 0.01). Bothcomplex I and IV activities in leukocytes showed atendency to decrease with age in healthy subjects withoutneurological disorders; however, no significant correla-tion between age, age of onset, and disease duration withcomplex I and IV activities were found in patients withparkin mutations and idiopathic PD patients. In thisstudy, all PD patients were receiving L-dopa at the timeof enzyme activity measurements. It has been shown thatL-dopa treatment does not affect mitochondrial electrontransport chain enzyme activities.30,31 Thus, the decreasein complex I and IV activities are related to PD ratherthan to treatment with L-dopa.

This selective decrease in leukocyte complex I activityin patients with parkin gene mutations could be due toenvironmental factors, as was suggested for idiopathicPD. Alternatively, the parkin gene, which is expressednormally in leukocytes but probably as a smaller tran-script, might directly increase oxidative stress when mu-tated.32 It has been shown that transfection of neuronalcell lines with DNA encoding several types of patho-genic mutant parkin caused an increase in oxidativestress as indicated by decreased levels of GSH and ele-vated protein carbonyls and lipid peroxidation.21 It hasalso been suggested that mutant proteins might compet-itively inhibit or reversibly overload proteasome, leadingto proteosomal dysfunction, which in turn may be re-sponsible for enhanced oxidative stress. Likewise, theselective decrease in complex I activity in patients withparkin mutations might be due to a reduction in mito-chondrial GSH levels. Depletion of antioxidant GSHcauses inactivation of mitochondrial complex I activityvia thiol oxidation and a decrease in complex I activitywill cause mitochondrial respiratory chain failure, whichnot only generates reactive oxygen species but also impairsenergy metabolism, leading to a bioenergetic defect.

To our knowledge, this is the first clinical study put-ting forward evidence for a mitochondrial dysfunction inrelation to parkin mutations. If the deficit detected inleukocytes is also relevant for the brain, the suggestedrole of oxidative stress in the pathogenesis of neuronaldeath in idiopathic PD should also be considered forparkinsonism due to parkin mutations, possibly more asa secondary factor.

Acknowledgments: We thank Dr. F.B. Atac from BaskentUniversity, Ankara, Turkey, and Drs. N. Hattori and Y. Mizuno,Juntendo University, Tokyo, Japan, for their invaluable contribu-tions in detecting parkin mutations. This work was supported bythe Scientific and Technical Research Council of Turkey (grantnumbers SBAG-2596 and TBAG-1754) and by Hacettepe Uni-versity Scientific Research Unit (grant number 0002101001).

FIG. 3. The relationship between the age and the activities of complexIV in patients with parkin mutations, late-onset idiopathic PD patientsand control. Diamonds, control; solid circles, patients with parkinmutations; open circles, idiopathic PD patients.

MITOCHONDRIAL ENYZMES IN PARKINSON’S DISEASE 547


REFERENCES

1. Kruger R, Eberhardt O, Riess O, Schulz JB. Parkinson’s disease:one biochemical pathway to fit all genes? Trends Mol Med2002;8:236–240.

2. Vaughan JR, Davis MB, Wood NW. Genetics of parkinsonism: areview. Ann Hum Genet 2001;65:111–126.

3. Kitada T, Asakawa S, Hattori N, et al. Mutations in the parkingene cause autosomal recessive juvenile parkinsonism. Nature1998;392:605–608.

4. Tanaka K, Suzuki T, Chiba T, et al. Parkin is linked to theubiquitin pathway. J Mol Med 2001;79:482–494.

5. Greenamyre JT, Sherer TB, Betarbet R, Panov AV. Complex Iand parkinson’s disease. IUBMB Life 2001;52:135–141.

6. Andersen JK. What causes the build-up of ubiquitin-containinginclusions in Parkinson’s disease? Mech Ageing Dev 2000;118:15–22.

7. Atac FB, Elibol B, Hattori N, et al. Molecular analysis of Turkishfamilies with parkin gene mutations. Mov Disord 2000;15(Suppl.):199.

8. Merlo Pich M, Bovina C, Formiggini G, et al. Inhibitor sensitivityof respiratory complex I in human platelets: a possible biomarkerof ageing. FEBS Lett 1996;380:176–178.

9. Trounce IA, Kim YL, Jun AS, Wallace DC. Assessment ofmitochondrial oxidative phosphorylation in patient muscle biop-sies, lymphoblasts, and transmitochondrial cell lines. MethodsEnzymol 1996;264:495–496.

10. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein mea-surement with the folin phenol reagent. J Biol Chem 1951;193:265–275.

11. Beutler E. Red cell metabolism, 2nd ed. New York: Grune andStratton; 1975.

12. Jha N, Jurma O, Lalli G, et al. Glutathione depletion in PC12results in selective inhibition of mitochondrial complex I activity.J Biol Chem 2000;275:26096–26101.

13. Bharat S, Cochran BC, Hsu M, et al. Pre-treatment with R-lipoicacid alleviates the effects of GSH depletion in PC12 cells: im-plications for Parkinson’s disease therapy. Neurotoxicology2002;23:479–486.

14. Benecke R, Strumper P, Weiss H. Electron transfer complexes Iand IV of platelets are abnormal in Parkinson’s disease butnormal in Parkin-plus syndromes. Brain 1993;116:1451–1463.

15. Bindoff LA, Birch Machin MA, Cartlidge NE, et al. Respiratorychain abnormalities in skeletal muscle from patients with Parkin-son’s disease. J Neurol Sci 1991;104:203–208.

16. Krige D, Carroll MT, Cooper JM, et al. Platelet mitochondrialfunction in Parkinson’s disease. Ann Neurol 1992;32:782–788.

17. Yoshino H, Hattori Y, Kondo T, Mizuno Y. Mitochondrial com-plex I and II activities of lymphocytes and platelets in Parkinson’sdisease. J Neural Transm 1992;4:27–34.

18. Mizuno Y, Yoshino H, Ikebe S, et al. Mitochondrial dysfunctionin Parkinson’s Disease. Ann Neurol 1998;44(Suppl.):99–109.

19. Schapira AHV, Gu M, Taanman J, et al. Mitochondria in theetiology and pathogenesis of Parkinson’s disease. Ann Neurol1998;44(Suppl.):89–98.

20. Mouradian MM. Recent advances in the genetics and pathogen-esis of Parkinson disease. Neurology 2002;58:179–185.

21. Hyun DH, Lee M, Hattori N, et al. Effect of wild-type or mutantparkin on oxidative damage, nitric oxide, antioxidant defenses,and the proteosome. J Biol Chem 2002;277:28572–28577.

22. Rajagopalan S, Andersen JK. Alpha synuclein aggregation: is ittoxic gain of function responsible for neurodegeneration in Par-kinson’s disease? Mech Ageing Dev 2001;122:1499–1510.

23. Hashimoto M, Hsu LJ, Rockenstein E, et al. �-Synuclein protectsagainst oxidative stress via inactivation of the c-Jun N-terminalkinase stress-signaling pathway in neural cells. J Biol Chem2002;277:11468–11472.

24. Jha N, Kumar MJ, Boonplueang R, Andersen JK. Glutathionedecreases in dopaminergic PC12 cells interfere with the ubiquitinprotein degradation pathway: relevance for Parkinson’s disease.J Neurochem 2002;80:555–561.

25. Lee MH, Hyun DH, Jenner P, Halliwell B. Effect of proteosomeinhibition on cellular oxidative damage, antioxidant defences andnitric oxide production. J Neurochem 2001;78:32–41.

26. McNaught KS, Belizaire R, Isacson O, et al. Altered proteosomalfunction in sporadic Parkinson’s disease. Exp Neurol 2003;179:38–46.

27. Zhang Y, Dawson VL, Dawson TM. Oxidative stress and genet-ics in the pathogenesis of Parkinson’s disease. Neurobiol Dis2000;7:240–250.

28. Benoit I, Lee VM. Parkin and molecular pathways of Parkinson’sdisease. Neuron 2001;31:885–888.

29. Chung KKK, Dawson VL, Dawson TM. The role of the ubiq-uitin-proteosomal pathway in Parkinson’s disease and other neu-rodegenerative disorders. Trends Neurosci 2001;24(Suppl.):7–14.

30. Dagani F, Ferrari R, Anderson JJ, Chase TN. L-dopa does notaffect electron transfer chain enzymes and respiration of ratmuscle mitochondria. Mov Dis 1991;6:315–319.

31. Haas RH, Nasirian F, Nakano K, et al. Low platelet mitochondrialcomplex I and complex II/III activity in early untreated parkin-son’s disease. Ann Neurol 1995;37:714–722.

32. Sunada Y, Saito F, Matsumura K, Shimuzu T. Differential ex-pression of the parkin gene in the human brain and peripheralleukocytes. Neurosci Lett 1998;254:180–182.

Neuronal Globus PallidusActivity in Patients With

Generalised Dystonia

Marcelo Merello, MD, PhD,1,2*Daniel Cerquetti, PhD,1 Angel Cammarota, MD,1,2

Eduardo Tenca, MD,3 Carlos Artes, PhD,3

Julio Antico, MD,3 and Ramon Leiguarda, MD2

1Movement Disorders Section, Raul Carrea Institute forNeurological Research, FLENI, Buenos Aires, Argentina;

2Neurology Department, Raul Carrea Institute forNeurological Research, FLENI, Buenos Aires, Argentina;

3Functional Neurosurgery Department, Raul Carrea Institutefor Neurological Research, FLENI, Buenos Aires, Argentina

Abstract: We studied 516 globus pallidus neurons in dys-tonic patients. The firing rate was analysed. We classifiedthe burst activity into tonic, burst, and pause patterns.

*Correspondence to: Dr. Marcelo Merello, Movement DisordersDepartment, FLENI, Montaneses 2325, C1428AQK Buenos Aires,Argentina. E-mail: [email protected]

Received 26 April 2003; Revised 18 September 2003; Accepted 29September 2003


548 M. MERELLO ET AL.


Mean � SD firing rates and tonicity score for internalglobus pallidus (GPi) and external globus pallidus (GPe)were 54.6 � 28.6; 58.01 � 39.1 and 1.18 � 0.55; 0.95 � 0.43,respectively. Differences in percentage appearance of tonic,burst, or paused neurons were not statistically significantfor GPi versus GPe. GPi firing features in dystonic patientswere closely similar to those of GPe. This could suggest thatthe abnormally patterned output from GPi would not resultfrom increased differential inhibitory/excitatory input aris-ing from the direct/indirect pathway but rather be trans-mitted from GPe, striatum, or either centromedian nucleus.© 2003 Movement Disorder Society

Key words: dystonia; pallidotomy; DBS; firing analysis

Functional disruptions of the basal ganglion thalamo-cortical motor circuit lead to errors in movement scalingand focusing.1–4 Depending on the particular abnormal-ities in the circuit, this disruption could result in eitherhypokinetic or hyperkinetic movement disorders.5,6

Although there are no data from normal humans forcomparison, the mean discharge rate of neurons in theexternal globus pallidus (GPe) and internal globus palli-dus (GPi) in dystonic patients has been reported to bereduced compared to rates recorded in Parkinson’s dis-ease (PD) patients.7,8 Based on early models of hyperki-netic movement disorders,1 it could not have been pre-dicted that GPi lesions would improve dystonia; in fact,one would predict that GPi lesions would make it worse.Strikingly, a dramatic positive effect is observed in-stead,9–13 so that alterations in firing rates alone appear tobe insufficient to explain the beneficial effect of a singleprocedure such as posteroventral pallidotomy on twoantagonistic disorders such as PD and dystonia.14 Wedescribe globus pallidus (GP) neuronal activity in dys-tonic patients, focusing on the discharge pattern.


Patients

Performance of microrecording-guided bilateral pos-teroventral pallidotomy or unilateral pallidotomy withcontralateral implantation of deep brain stimulation(DBS) system for dystonia treatment was approved bythe local ethics committee. Informed consent for patientinclusion was signed by parents because all cases wereminors. Patients were 3 boys and 5 girls with a mean ageof 12 years (range, 9–16 years). Mean duration of dis-ease was 6 years (range, 4–7 years). All presented gen-eralised idiopathic dystonia, 4 of whom carried theDYT1 gene. DYT1-negative patients presented a normal

magnetic resonance imaging (MRI) scan, a negative re-sponse trial to levodopa and an extensive, negative workoutsearching for causes of secondary dystonia. Patients wereall treated with a combination of trihexyphenidyl, diaze-pam, and baclofen, and were evaluated by means of theBurke–Fahn–Marsden Dystonia Movement (BFMDS-M)and Disability (BFMDS-D) scales15 at basal preoperativeconditions, then at 1, 6, and 12 months.

Microrecording and Surgical Technique

Microrecording was carried out by means of platinum/iridium (80%/20%) microelectrodes with nominal im-pedance of 0.8 to 1.2 M� (measured at 1 kHz) with glassinsulation, sheathed inside a 26-gauge stainless steeltube, epoxy sealed, and insulated with polyamide tubingwith an outer diameter of 0.625 mm (FHC mTSPBN-LX1); each electrode is electrically shielded with anouter stainless steel protective tube.

A preamplifier (remote probe) mounted onto a motor-ised microdrive (FHC 65-00-1 Stepper Drive and ST-M0-00 TMS Controller), located near the electrode tip tominimise pickup of electrical noise, was connected to adifferential amplifier with a built-in impedance meter(FHC IS-AM-00-01 Iso-Xcell 3� Amplifier), and anisolated stimulus generator (FHC IS-PL-06 Isolated Bi-polar Pulsar Stimulator). The signal was amplified, fil-tered and led to a 16-channel acquisition system (CED1401plus), audio equipment (Optimus) with a 10-bandgraphic equaliser and external loudspeakers, dual win-dow discriminator (BAK DDIS1), dual time base oscil-loscopes, and analogue recording system. The signal wasdigitised, processed, visualized, and stored using dedi-cated software (CED Spike2) running in a PC, and alsorecorded in a HI FI videocassette using a 16-channelPCM multiplexor (VETTER) and a HI-FI VCR (SONYSLV770HF).

On-line recording of the raw signal and digitised rasterdisplay (filtering and masking by means of dynamictemplates to discriminate single units) was monitored.The analogue video/audio system allowed recording ofthe complete registering process, while the stored digi-tised signal was used for on-line and off-line analysis,allowing discrimination and isolation of single cells,which met amplitude and acceptable signal to noise ratiocriteria. While the acquisition software allowed visual-isation of the signal in different formats, the oscillo-scopes and window discriminator were used to helpanalyse raw data with slow and fast time base settings,being able to trigger and display isolated spikes, showing

GP DISCHARGE IN DYSTONIA 549


FIG. 1. Sample from raw data, its raster display and frequency histograms showing typical discharges corresponding to three different firing patternsdescribed: (A) tonic TS � 0.5, disclosing sustained patterns, regular and uniform firing rate with isolated or no gaps; (B) burst TS 0.5 to 1.0, showingpatterns with frequent gaps, small bursts, or noticeable variations in firing rate; and (C) paused neurons TS � 1, with dominant presence of large gapsand formation of clustered discharges.


polarity, shape, and duration. Data stored in the analoguevideo/audio system were played back into the same sys-tem after surgery. The number of recording tracks per-formed at each operation depended on (1) the ability tounequivocally identify putamen and GP, (2) correct iden-tification of internal capsule applying constant currentmicrostimulation protocols, and (3) correct identificationof the optic tract by microstimulation and signal record-ing during visual stimulation. In this particular protocol,examination of receptive field properties that requirespassive manipulation and active movements of the limbsand orofacial structures was not undertaken so as to keepthe subject awake, neither sedated nor anaesthetised, fora more accurate analysis of discharge frequency andpattern. Then GPi macrostimulation (2–300 Hz) wasperformed to identify side effects and at the site deter-mined by the electrophysiological findings, either a ther-molesion was performed or a DBS (#3389 lead,Medtronic, Minneapolis, MN) was implanted. The leadwas connected to an implantable pulse generator (SO-LETRA Medtronic) at the same surgical procedure.

Signal Analysis

Exhaustive off-line analysis was carried out with thecollected data. Firing rate was analysed jointly withinterspike interval, and a composite score of burst activ-ity (tonicity score, TS) was evaluated for each case.

Tonicity Score.

By analysing a digitised and postprocessed firing pat-tern over a certain interval of time (TI), the interspikeinterval (ISI) is calculated as the time elapsing betweentwo consecutive spikes, along such TI. ISI mean (mISI)and standard deviation (sISI) are then computed for thewhole interval under evaluation TI.

We define the TS as the relationship between ISIstandard deviation and ISI mean, calculated along theinterval of time TI, where the number of computed ISIsshould be greater than 100 to be regarded as a represen-tative index. This last condition [(TI)/(mISI) � 100],thus, would satisfy our definition of a “representativetonicity score” and for the firing patterns under study inthis paper, an interval TI � 10 sec would normally

FIGURE 1. (Continued)



satisfy this condition by delivering a mISI ranging from8 to 60 msec, which corresponds to patterns containingfrom 150 to 1,300 interspike intervals. The tonicity scoregives us an idea of the uniformity or “tonicity” of thepattern under study, so that, for a tonic pattern, thecorresponding ISIs would be similar to one another,approaching a periodic and sustained signal; this resultsin a small sISI rendering small values of TS.

On the other hand, in a burst pattern where the ISIsdiffer greatly from one another, the deviation increasesand the TS becomes higher. By applying this concept inanalysing ISI distribution in a certain pattern, we mayevaluate their dispersion with respect to their meanvalue, i.e., we may analyse the content of very small ISIs(high instantaneous frequency discharges) or very largeISIs (pauses) related to the mean frequency of suchpattern. The greater the relative difference or dispersionof the ISIs, the more uneven or “less tonic” the patternwill be, i.e., it will show a great content of ISIs muchsmaller and much larger than a central average value.Considering the above definition of the tonicity score,simplification and transformation yields an equivalentformula:

TS � �nISI2

ISI�2 � 1 or TS � �nISI2

TI2 � 1

“n” being the size of the sample or the number of ISIscontained in the time interval TI. Patterns are classifiedaccording to their corresponding TS, into three groups:tonic [TS � 0.5], sustained pattern, regular and uniformfiring rate with isolated or no gaps; burst [0.5 � TS � 1],patterns with frequent gaps, small bursts, or noticeablevariations in firing rate; and paused [TS � 1], dominantpresence of large gaps, and formation of clustered dis-charges (Fig. 1). Computing the TS for the differentstreams helps to describe, classify, and compare firingpatterns among the recorded units.

RESULTS

Clinical results correspond to 7 of 8 patients reported,as 1 patient underwent a unilateral procedure, while therest received unilateral posteroventral pallidotomy withDBS applied to the contralateral GPi in 5 patients andbilateral pallidotomy in 2. Surgery was performed in onestep, and the decision of combined technique or bilaterallesion was not randomised but based on economic rea-sons. Mean � SD scores for BFMDS-M at basal, 1, 6,and 12 months were 62.3 � 5.6, 35.6 � 9.5, 21.0 � 2.3,and 21.0 � 2.3, respectively; F(3,15) � 63,11; P �

0.0001. Mean � SD scores for BFMDS-D at basal, 1, 6,and 12 months were 22.1 � 4.1; 17.6 � 7.7; 11 � 1.5;and 9.6 � 3.1, respectively; F(3,15) � 9,81; P � 0.0008.

Electrophysiological Findings

A total of 516 neurons were recorded along 49 trackswithin 15 GPi. An average of 34.4 neurons per GP and10.5 per track were recorded. Because of the absence ofdifferential characteristics between neurons and the de-cision not to explore the sensorimotor response of re-corded neurons, the main criterion to decide whether agiven neuron belonged to the GPi or GPe was based onits distance to the optic tract in each particular track. Thisprocess was evenly performed across the cases. A total of45% of recorded neurons were arbitrarily classified asbelonging to GPi, and 35% to GPe, while 20%, eventhough individually analysed, were not classified as partof the external or internal globus pallidus, being at dis-tances to the optic tract not conclusive for a clear clas-sification. Mean � SD firing rates for GPi and GPe were54.6 � 28.6 and 58.01 � 39.1, respectively (P � 0.7;Fig. 2A). When the frequency of discharge and tonicityscore of the whole group of neurons was plotted againstthe distance to the optic tract, there was no significantcorrelation for either parameter.

Mean GPi TS was 1.18 (0.55) and mean GPe TS 0.95(0.43), respectively (P � 0.09; Fig. 2B). Differences inpercentage appearance of tonic, burst, or paused neuronswere not statistically significant for GPi versus GPe(�2 � 0.631; df � 2: P � 0.7).

DISCUSSION

We found that GPi firing features in dystonic patientsclosely resemble those of GPe and hinder on-line differ-entiation mainly because of similar rates and firing pat-terns within both nuclei. In addition to reductions inmean discharge rates and pattern changes in GPi, previ-ous work on spontaneous neuronal activity in patientswith dystonia and hemiballismus has disclosed an in-creased degree of synchronisation and somatosensoryresponsiveness.7–9,16 We decided not to examine recep-tive fields that act as stimuli for dystonic movements andjerks with the patient harnessed to the stereotactic frame,to keep the subject awake for more accurate recording, aspropofol, the most widely used anaesthetic for this kindof surgery, has been noted to interfere with the record-ing.17–19 Frequency of discharge in our patients washigher than in previous published literature,7–9 and weassume the use of general anaesthesia in published workwas responsible for this difference. Even though thefiring rate we found was lower than those described in

552 M. MERELLO ET AL.


Parkinson’s disease patients,20 the reduction was not ofthe expected magnitude according to the rate-based basalganglion model. Nevertheless, the firing rate we foundwas slightly lower than the one recently reported byHutchison and colleagues,17 and perhaps the heteroge-neous population under their study may be responsiblefor the differences between their study and ours. Vitekand coworkers7 were the first to focus the pattern ofdischarge as a critical factor for understanding the basalganglion model of functioning.

Although we were able to determine different patternsof firing discharge, our score was unable to differentiatethe internal from the external globus pallidus based ondischarge pattern. In agreement with findings reported byVitek and colleagues, who documented irregularlygrouped discharges with intermittent pauses in 19 of 26of GPi and 25 of 44 of GPe neurons,7 taking advantage

of our so-called TS, we classified neurons into tonic,burst, and paused, which occurred with a similar inci-dence in GPe and GPi, thus representing a parallel be-haviour of both nuclei.

Apomorphine-induced dyskinesias likewise featurechanges in GPi neuronal pattern, varying from tonic tosporadic burst-like activity with a lower dischargerate.21,22 Such alterations seem closely related to dyski-nesias, or perhaps essential for their development.

Changes in discharge pattern within the GPi couldeither represent the abnormal pattern of discharge trans-mitted from afferent nuclei such as GPe, STN, striatum,and centromedian nucleus,23,24 or could be internallygenerated within GPi as a result of abnormal enhance-ment of antagonistic excitatory and inhibitory effectsexerted by the indirect and direct pathway, respectively.7

That we found that both nuclei discharge in such asimilar manner suggests that the abnormally patternedoutput from GPi would not result from increased differ-ential inhibitory/excitatory input arising from the direct/indirect pathway, but rather be transmitted from GPe,striatum, or either centromedian nucleus. This hypothesisshould be clarified by future recordings arising fromthese nuclei. Alternatively, it is possible that the origin ofthe abnormal pattern signal could depend of the type ofdystonia under study. Even though all of our eight pa-tients presented idiopathic dystonia, that only 4 carriedthe DYT1 dystonia gene makes our population hetero-geneous, and further studies should be performed toconfirm these findings.

The explanation why a lesion/inhibition within GPiimproves clinically opposite disorders such as PD anddystonia exceeds the rate hypothesis and could resultfrom an interruption of abnormal pattern of pallidal out-put. Present findings lend further support to the proposedpattern-based basal ganglion model.16

Acknowledgments: We thank Prof. F. Micheli and Prof O.Gershanik who kindly referred two of the patients for treatmentat our centre.

REFERENCES

1. DeLong MR. Primate models of movement disorders of basalganglia origin. Trends Neurosci 1990;13:281–285.

2. Alexander GE, Crutcher M, De Long MR. Functional architectureof basal ganglia circuits: parallel substrates for motor, oculomotor,“prefrontal” and “limbic” functions. Prog Brain Res 1990;85:119–146.

3. Albin RL, Young AB, Penney JB. The functional anatomy of basalganglia disorders. Trends Neurosci 1989;12:366–375.

4. Smith Y, Wichmann T, DeLong, MR. The external pallidum andthe subthalamic nucleus send convergent inputs onto single neu-rons in the internal pallidal segment in the monkey: anatomicalorganization and functional significance. In: Percheron G, Mc

FIG. 2. Box and whiskers plots for discharge frequency (in Hertz, A)and tonicity score (TS, B) display no significant mean differences forinternal globus pallidus (GPi) versus external globus pallidus (GPe).Error bars indicate �SD, large boxes indicate standard error; smallboxes represent the mean.



Kenzie FS, Feger J, editors. The basal ganglia IV: new ideas anddata on structure and function. New York: Plenum Press; 1994. p51–61.

5. Hoover JE, Strick PL. Multiple output channels in the basal gan-glia. Science 1993;259:819–821.

6. Jinnai K, Nambu A, Yoshida S, Tanibuchi I. The two separateneuron circuits through the basal ganglia concerning the prepara-tory or execution processes of motor control. In: Mamo N, HamadaI, DeLong MR, editors. Role of the cerebellum and basal gangliain voluntary movement. Amsterdam: Elsevier Science; 1993. p153–161.

7. Vitek JL, Chocklan V, Zhang JY, et al. Neuronal activity in basalganglia in patients with generalized dystonia and hemiballismus.Ann Neurol 1999;46:22–35.

8. Lenz FA, Suarez JL, Metman LV, et al. Pallidal activity duringdystonia: somatosensory reorganization and changes with severity.J Neurol Neurosurg Psychiatry 1998;65:767–770.

9. Lozano AM, Kumar R, Gross RE, et al. Globus pallidus internuspallidotomy for generalized dystonia. Mov Disord 1997;12:864–870.

10. Ondo W, Desaloms J, Jankovic J, Grossman R. Pallidotomy forgeneralized dystonia. Mov Disord 1998;13:693–698.

11. Vitek JL, Evatt M, Zhang J, et al. Pallidotomy and deep brainstimulation as a treatment of dystonia. Neurology 1999;52:S46.006.

12. Kumar R, Dagher A, Hutchison WD, Lang AE, Lozano AM.Globus pallidus deep brain stimulation for generalized dystonia:clinical and PET investigation. Neurology 1999;53:871.

13. Krack P, Vercueil L. Review of the functional surgical treatment ofdystonia. Eur J Neurol 2001;8:389–399.

14. Marsden CD, Obeso JA. The functions of the basal ganglia and theparadox of stereotaxic surgery in Parkinson’s disease. Brain 1994;117:877–897.

15. Burk R, Fahn S, Marsden C, et al. Validity and reliability of arating scale for the primary torsion dystonias. Neurology 1985;35:73–77.

16. Vitek JL. Pathophysiology of dystonia: a neuronal model. MovDisord 2002;17(Suppl. 3):S49–S62.

17. Hutchison WD, Lang AE, Dostrovsky JO, Lozano AM. Pallidalneuronal activity: implications for models of dystonia. Ann Neurol2003;53:480–488.

18. Krauss JK, Akeyson EW, Giam P, Jankovic J. Propofol-induceddyskinesias in Parkinson’s disease. Anesth Analg 1996;83:420–422.

19. Anderson BJ, Marks PV, Futter ME. Propofol-contrasting effectsin movement disorders. Br J Neurosurg 1994;8:387–388.

20. Merello M, Lees A, Balej J, et al. GPi firing rate modificationduring beginning-of-dose motor deterioration following acute ad-ministration of apomorphine. Mov Disord 1999;14:481–483.

21. Merello M, Balej J, Delfino M, Cammarota A, Betti O, LeiguardaR. Apomorphine induces changes in GPi spontaneous outflow inpatients with Parkinson’s disease. Mov Disord 1999;14:45–49.

22. Hutchison WD, Levy R, Dostrovsky JO, Lozano AM, Lang AE.Effects of apomorphine on globus neurons in Parkinsonian pa-tients. Ann Neurol 1997;42:767–775.

23. Sadikot AF, Parent A, Francois C. Efferent connections of thecentromedian and parafascicular thalamic nuclei in the squirrelmonkey: a PHA-L study of subcortical projections. J Comp Neurol1992;315:137–159.

24. Cooper IS. 20-Year follow-up study of the neurosurgical treatmentof dystonia musculorum deformans. In: Eldridge R, Fahn S, edi-tors. Advances in neurology. New York: Raven Press; 1976. p423–452.

Non-Subtype-Selective OpioidReceptor Antagonism in

Treatment of Levodopa-InducedMotor Complications in

Parkinson’s Disease

Susan Fox, MCRP, PhD,1*Montague Silverdale, MRCP,

Mark Kellett, MCRP, MD, Rhys Davies, MRCP,Malcolm Steiger, FRCP, MD,1

Nicholas Fletcher, FRCP, MD,1 Alan Crossman, DSc,2

and Jonathan Brotchie, PhD1The Walton Centre for Neurology and Neurosurgery,Liverpool, United Kingdom; 2Manchester Movement

Disorders Laboratory, Universityof Manchester, Manchester, United Kingdom

Abstract: Opioid peptide transmission is enhanced in the stria-tum of animal models and Parkinson’s disease (PD) patients withlevodopa-induced motor complications. Opioid receptor antago-nists reduce levodopa-induced dyskinesia in primate models ofPD; however, clinical trials to date have been inconclusive. Adouble-blind, placebo controlled, crossover design study in 14patients with PD experiencing motor fluctuations was carriedout, using the non-subtype-selective opioid receptor antagonistnaloxone. Naloxone did not reduce levodopa-induced dyskinesia.The duration of action of levodopa was increased significantly by17.5%. Non-subtype-selective opioid receptor antagonism mayprove useful in the treatment of levodopa-related wearing-off inPD but not in dyskinesia. © 2003 Movement Disorder Society

Key words: Parkinson’s disease; dyskinesia; motor fluctua-tions; opioids; naloxone

Long-term treatment of Parkinson’ disease (PD) withlevodopa (L-dopa) results in the development of motorcomplications, including end-of-dose wearing-off, on–off fluctuations and dyskinesia. The incidence of all

Current address for Montague Silverdale and Mark Kellett: theDepartment of Neurology, Hope Hospital, Salford, United Kingdom.

Current address for Rhys Davies: the University Neurology Unit,Addenbrooke’s Hospital, Cambridge, United Kingdom.

Current address for Jonathan Brotchie: Toronto Western ResearchInstitute, Toronto Western Hospital, Toronto, Ontario, Canada.

*Correspondence to: Dr. Susan H. Fox, Movement Disorders Clinic,Toronto Western Hospital, 399 Bathurst St., MP11, Toronto, ON,Canada M5T-2S8. E-mail: [email protected]

Received 23 May 2003; Revised 18 September 2003; Accepted 30September 2003


554 S. FOX ET AL.


motor complications rises with duration and cumulativedose of L-dopa given, as well as with disease duration.1,2

The neural mechanisms underlying L-dopa–inducedmotor complications are thought to involve changes inneurotransmitter signalling pathways within the basalganglia as a result of long-term dopamine receptor stim-ulation.3,4 One such system thought to be involved isopioid transmission. Striatal GABAergic projection neu-rones within the basal ganglia use opioid peptides asco-transmitters.5–8 Enkephalins derived from the highmolecular weight precursor, pre-proenkephalin A(PPE-A) activate opioid receptors and opioid peptidesderived from pre-proenkephalin B (PPE-B) activate , �,and �-subtype of opioid receptors.9,10

After long-term L-dopa treatment and the developmentof motor complications, the levels of opioid peptides andthe mRNA encoding their precursors are elevated inanimal models of parkinsonism.11–17 This does not occurafter treatment that does not induce dyskinesia inman.13,17,18 Furthermore, in post-mortem tissue from PDpatients with L-dopa–induced motor complications, thereis increased striatal PPE-B19 and PPE-A20,21 expression.In addition, positron emission tomography (PET) studieshave shown abnormalities in opioid receptor function,consistent with enhanced opioid transmission in the stri-atum of PD patients with dyskinesia compared with thatin non-dyskinetic patients.22

These findings suggest that opioid receptor antagonistsmight be useful in treatment of L-dopa–induced motorcomplications. In animal models of PD, or �, but not�-subtype selective opioid receptor antagonists can re-duce L-dopa–induced motor complications.23 Non-sub-type-selective opioid antagonists have been shown toeither decrease23,24 or not modify L-dopa–induced dyski-nesia in the MPTP-primate.25 To date, no preclinicalstudies have addressed the effect of opioid antagonismon wearing-off or on-off fluctuations.

Clinical studies of opioid antagonists in PD patientshave thus far been inconclusive. One small-scale, dou-ble-blind, placebo-controlled study in 6 patients, usingthe non-subtype–selective opioid receptor antagonist,naloxone, showed a significant reduction in total dailydyskinesia and a reduction in rigidity.26 Case reportsusing naloxone have shown either a reduction27 or noeffect.28 In all these studies, naloxone was administeredas a bolus injection. Oral administration of naltrexone, alonger-acting analogue of naloxone, is ineffective at lowdose29,30 and at higher doses, has minimal effect inalleviating L-dopa–induced dyskinesias with no effect onthe parkinsonian scores.31

This study was designed to test definitively the hy-pothesis that enhanced opioid neurotransmission under-lies L-dopa–induced motor complications in patients withPD. We carried out a randomized, double-blind, placebo-controlled, crossover acute challenge study using thenon-subtype-selective opioid receptor antagonist nalox-one in PD patients experiencing L-dop–induced motorcomplications. Naloxone was selected for the study as ahighly selective opioid receptor antagonist, active at allopioid receptor subtypes with a good safety record andside-effect profile. Furthermore, intravenous (i.v.) nalox-one has been employed widely to demonstrate opioidactions in man and administration regimens that provideopioid receptor antagonism over several hours gave noreported adverse effects.32–36


Twenty patients with a clinical diagnosis of PD37 andexperiencing stable motor fluctuations and dyskinesiawere recruited from a regional movement disordersclinic. Patient characteristics are shown in Table 1. Allpatients were on stable medication for 3 months beforethe study; L-dopa (mean daily dose 727 � 338 mg, n �20), pergolide (3.8 � 1.6 mg, n � 10), ropinirole (23 �1.7 mg, n � 3), cabergoline (2 mg, n � 1), bromocriptine(35 mg, n � 1), apomorphine (10.1 � 6.4 mg, n � 7),selegiline (5 mg, n � 1), entacapone (600 � 500 mg, n �6), and amantadine (175 � 95 mg, n � 4). Patientstaking amantadine were asked to stop taking the medi-cation for a minimum of 4 weeks before the study.Exclusion criteria were opioid dependence, liver disease,or abnormal liver function tests, history of cardiac ar-rhythmia or heart block, renal impairment, premeno-

TABLE 1. Patient characteristics

Characteristic Value

Gender 14M/6FMean age, yr (range) 61.3 (45–71)Mean disease duration, yr (range) 10.1 (5–21)H&Y (range) 3.0 (1.5–3.0)On period UPDRS III motor score, median (range) 20.5 (6–49)On total UPDRS, median (range) 49.9 (18–78)Median dyskinesia duration, yr (range)a 2 (1–3)Median dyskinesia disability, yr (range)b 2 (1–3)Patients experiencing wearing-off, n (%)c 18 (90)Patients experiencing sudden/unpredictable offs,

n (%)d 13 (65)

aUPDRS IV, item 32.bUPDRS IV, item 33.cUPDRS IV item 36.dUPDRS IV items 37, 38.H&Y, Hoehn and Yahr; UPDRS, Unified Parkinson’s Disease Rat-

ing Scale.

NALOXONE AND PARKINSON’S DISEASE 555


pausal females without adequate contraception, cognitiveimpairment, or any history of psychiatric disease. Thelocal health authority ethics committee approved thestudy and all patients gave written informed consent.

Patients were admitted for a L-dopa challenge in thepractically-defined off state,38 i.e., after a 12-hour with-drawal of antiparkinsonian medication and a light break-fast. The dose of L-dopa used was the normal morningL-dopa dose (given as Madopar dispersible) plus 100 mgif on a dopamine receptor agonist.

Commencing 10 minutes before the L-dopa treatment,the patient received either naloxone, or vehicle, double-blind, as a bolus i.v. injection of 25 �g/kg followed bycontinuous infusion at a rate of 0.4 �g/kg/min through-out the duration of the L-dopa challenge, to maintainblockade of opioid receptors. Randomisation for nalox-one/placebo was carried out using standard random num-ber tables and retained in the hospital pharmacy untilcompletion of the study. Naloxone and placebo were ofidentical colour and prepared in the hospital pharmacy.Labeling was standard with the patient initials and num-ber of visit (1 or 2) to maintain blinding.

Parkinsonian disability was assessed by the examineron the day of the study, using a modified motor score partIII of the Unified Parkinson’s Disease Rating Scale (UP-DRS). Assessments were carried out in the practicallydefined off state, at 20 minutes and then 60-minute in-tervals after L-dopa administration, with additional as-sessments when the patients reported they were switch-ing-on, best-on and switching-off. The gait subsectionwas omitted due to practical difficulties with the infusionpump. The on period was defined as the time duringwhich the UPDRS III improved by at least 30% com-pared with the practically defined off period. The latencyto switching-on (time from administration of L-dopa tofirst improvement in UPDRS III by 30%) and the per-centage of the on period during which any dyskinesiawas present was also recorded. Adverse events werenoted.

Patients were videotaped and assessment of dyskinesiacarried out by post-hoc video analysis, by a neurologistblinded to the treatment given. Dyskinesia was assessedin the practically defined off period and then every 20minutes after L-dopa administration until the patient

FIG. 1. Progress of patients through the trial.

556 S. FOX ET AL.


switched off, i.e., UPDRS was within 10% of off-periodUPDRS. Dyskinesia was measured using a 5-point ob-jective dyskinesia intensity rating scale, which rated 7body parts, maximum possible total score 28 (scores:0 � no dyskinesia present; 1 � questionable or milddyskinesia; 2 � moderate amplitude but quite apparentabnormal postures or movements that are not intrusive;3 � large amplitude movements or postures that maydistort and mildly or moderately disturb voluntary move-ments; 4 � severe and grotesque postures or movementsthat markedly disturb posture or ongoing voluntary ac-tivities) (J. Nutt, personal communication). Activation ofdyskinesia was achieved by mental calculation and dur-ing assessment of UPDRS III. Total dyskinesia wasobtained by area under the curve summation. Peak dosedyskinesia was determined when the patient was in the“best-on” state, i.e., when the patient subjectively feltthat they were experiencing maximal benefit from L-dopaand had the lowest UPDRS III motor score.

After a period of 1 to 2 weeks, the experiment wasrepeated as a crossover design. Given the short half-lifeof naloxone, this time is more than sufficient to allow forwashout of drug.

Statistical Analysis

Given the repeated measure design, scores for dyski-nesia and parkinsonism were expressed as median �range and compared using the Wilcoxon matched pairstest. The latency, duration of on period and percentageon period dyskinesia were expressed as mean � SEMand compared using a paired t test. All patients who

completed both parts of the study were included in theanalysis.

RESULTS

Fourteen patients completed the study (Fig. 1). Onepatient developed abdominal pain, diarrhoea, lethargy,and cold-cyanosed peripheries that occurred within 1hour of the infusion and lasted 6 hours and was a pre-sumed opioid withdrawal reaction. Subsequent unblind-ing revealed that the patient had received naloxone onthat day and the reaction was thought to be secondary toundisclosed long-term codeine use. None of the excludedpatients were included in the final analysis. There wereno other reported adverse events.

There was no significant difference in total L-dop–induced dyskinesia intensity between placebo (mediantotal dyskinesia, 23, range, 1–123) and naloxone (mediantotal dyskinesia, 34; range, 5–171; P � 0.05, n � 14;Fig. 2A,B). Peak dose dyskinesia was not significantlydifferent between placebo (median dyskinesia, 6.5;range, 1–24) and naloxone (median dyskinesia, 5.0;range, 1–23) (P � 0.05). To ensure no possible long-term antidyskinetic effects of amantadine in those pa-tients previously on this medication, subgroup analysisexcluding these patients (n � 4) was carried out. Nosignificant difference in either total or peak dose L-dopa–induced dyskinesia was seen (data not shown).

The duration of action of L-dopa was increased signif-icantly with naloxone. Thus, the duration of the onperiod with placebo was 111.7 � 8.1 minutes comparedwith 131.3 � 10 minutes with naloxone (P � 0.05; Fig.

FIG. 2. A: Time course of dyskinesia intensity after acute L-dopa challenge and co-treatment with naloxone/placebo infusion in patients with PD andmotor fluctuations. Each data point represent median score (ranges omitted for clarity); n � 14. B: Total accumulated dyskinesia intensity score afteracute L-dopa challenge and co-treatment with naloxone/placebo infusion in patients with PD and motor fluctuations. Each data points represents anindividual patient, line represents median score; n �14.



3). There was no significant effect on either baseline offperiod UPDRS III scores before each L-dopa challenge orthe time to switch on between the treatment groups (datanot shown) (Fig. 3). Although the on period was ex-tended with naloxone, there was no significant differencein the percentage of on period during which dyskinesiawas present (80.5 � 8.1% with placebo compared with83.7 � 6.5% with naloxone; P � 0.05).

DISCUSSION

This study demonstrates that the non-subtype-selec-tive opioid receptor antagonist naloxone significantlyextends the duration of L-dopa action, suggesting thatnon-subtype-selective opioid receptor antagonists maybe useful in the treatment of wearing-off in PD patients.Naloxone, however, does not reduce L-dopa–induceddyskinesia.

Methodological Considerations

The lack of an antidyskinetic effect of naloxone can-not be due simply to an ineffective dose being adminis-tered, as naloxone significantly extended the action ofL-dopa. The dosing regime of naloxone used is similar toprevious protocols employed to produce blockade ofopioid receptors for between 2 and 24 hours.32–36 Inaddition, the patient who unfortunately developed anopioid withdrawal-like reaction suggests that efficientblockade of opioid receptors occurred at the dose em-ployed. Naloxone is lipid soluble and is distributed rap-idly from serum to tissues. Animal studies have demon-strated high levels of naloxone within the brain within 1to 2 minutes of administration.39,40 In addition, compar-ative brain:serum levels of naloxone reveal brain-to-

serum ratios of approximately 5:1.41 Continuous infusionof naloxone at the dose employed would therefore haveensured blockade of central opioid receptors for theduration of the L-dopa challenge.

Phase II “proof of concept” clinical trials for newtreatments in PD are an important prelude to larger PhaseIII clinical trials from both an efficacy and a safetyperspective. Acute L-dopa challenges have been a tradi-tional method of carrying out these studies. The L-dopachallenge in this study was carried out using oral L-dopaas a soluble preparation. This is a standard method forassessing dyskinesia and has been employed in previoustrials42–45 and as part of the presurgical assessment ofpatients undergoing surgery for PD.38 There is debateregarding the dose of L-dopa that should be employed,either the patient’s normal morning dose with an addi-tional supplementation if also taking a dopamine recep-tor agonist, or a “supramaximal” dose. The former optionwas chosen in this study as we have found, in nonhumanprimates, that some agents with antidyskinetic potentialonly show such activity when treatment is optimal withrespect to antiparkinsonian activity. Furthermore, if asupramaximal dose of L-dopa were given to maximisedyskinesia, any potential exacerbation of dyskinesia bynaloxone may have been missed. In some patients, thisdosing regime resulted in a low intensity of dyskinesiaduring the L-dopa challenges. An alternative to preventdose failure or low intensity of dyskinesia is to use i.v.L-dopa. This has been employed previously in Phase IItrials of antidyskinetic agents.46,47 As naloxone was be-ing administered i.v., however, then technically, i.v. L-dopa was not a logistically viable option in this study.

Mechanism of Action

Effect on Duration of Action of L-Dopa.

Naloxone infusion significantly increased the durationof action of L-dopa without a significant increase indyskinesia. It is thus unlikely that the effect on on time isdue to enhanced dopamine levels or an effect on L-dopametabolism, as a concomitant increase in dyskinesiaswould be expected. Two patients did experience en-hanced dyskinesia after naloxone treatment (Fig. 2B);however, statistical analysis between the two groupsshowed no significant difference in dyskinesia.

The effects of naloxone on increasing on time proba-bly reflects blockade of neural mechanisms responsiblefor wearing-off. In the absence of intracerebral applica-tion of naloxone to specific regions of the basal gangliacircuitry and administration of subtype-selective agents,the precise mechanism of action can only be speculated.

FIG. 3. Time course of parkinsonian disability (UPDRS III) afteracute L-dopa challenge and co-treatment with naloxone/placebo infu-sion in patients with PD and motor fluctuations. Each data pointrepresent median score (ranges omitted for clarity); n � 14.

558 S. FOX ET AL.


We have demonstrated recently, however, that levels ofthe � opioid ligand, dynorphin A1–13 are enhanced spe-cifically in the external globus pallidus in MPTP-le-sioned primates with L-dopa–induced motor complica-tions (Zhou, personal communication). It is unlikely thatsuch a rise contributes to L-dopa–induced dyskinesia, as� opioid receptor antagonists do not have antidyskineticactions,23 but it is possible that it may contribute tomechanisms of wearing-off. Thus blockade of this pro-cess may underlie the extension of on time by naloxone,as shown here.

Effect on L-dopa–Induced Dyskinesia.

The lack of effect of naloxone on L-dopa–induceddyskinesia may be due to the non-subtype–selective na-ture of naloxone. In the current study, the naloxoneadministration protocol was chosen to ensure completeblockade of all opioid transmission. Blockade of , �,and � opioid receptors within the basal ganglia, however,may have conflicting effects, antidyskinetic but also pro-dyskinetic depending on the site of action. These opioidreceptors modulate release of GABA, glutamate, anddopamine at different sites within the basal ganglia cir-cuitry, with possible conflicting effects.

Thus, we propose that or � subtype-selective opioidreceptor antagonism, in the absence of � opioid receptorantagonism, may be required to demonstrate an action ofopioid antagonism in L-dopa–induced dyskinesia. Therelationship between single nucleotide polymorphisms inthe � opioid receptor and the propensity for patients todevelop dyskinesia is consistent with this idea.48 Further-more, in the MPTP-lesioned primate model of L-dopa–induced dyskinesia, the levels of the opioid �-neoendor-phin are increased in the basal ganglia output regions.48

�-Neoendorphin preferentially targets � and opioidreceptor subtypes. Thus, the neural mechanisms under-lying L-dopa–induced dyskinesia may involve overstimu-lation of � and opioid receptor subtypes in the internalglobus pallidus (GPi) and substantia nigra pars reticulata(SNR). At present, subtype-selective opioid antagonistsare not available for clinical use. In conclusion, thecurrent study supports the notion that non-selective opi-oid antagonism is an effective and safe method for re-ducing the problem of wearing-off and is worthy offurther study.

Acknowledgments: This study was funded by the Move-ment Disorders Society/Roche Parkinson’s Disease Award (toS.H.F.). We thank Prof. O. Rascol for advice. We also thankMrs. L. Owen, Mrs. A. Dennis, the nursing staff on the ClinicalTrials Unit, the neurologists at the Walton Centre, and ofcourse the enthusiasm of the patients is greatly appreciated.

REFERENCES

1. Marsden CD, Parkes JD. “On-off” effects in patients with Parkinson’sdisease on chronic levodopa therapy. Lancet 1976;1:292–296.

2. Rajput AH, Fenton ME, Birdi S, et al. Clinical-pathological studyof levodopa complications. Mov Disord 2002;17:289–296.

3. Bezard E, Brotchie JM, Gross CE. Pathophysiology of levodopa-induced dyskinesia in Parkinson’s disease: opportunities for noveltreatments. Nat Rev Neurosci 2001;2:577–588.

4. Crossman AR. A hypothesis on the pathophysiological mecha-nisms that underlie levodopa- or dopamine agonist-induce dyski-nesias in Parkinson’s disease: implications for future strategies intreatment. Mov Disord 1990;5:100–108.

5. Cuello AC, Paxinos G. Evidence for a long Leu-enkephalin stria-topallidal pathway in rat brain. Nature 1978;271:178–180.

6. Del Fiacco M, Paxinos G, Cuello AC. Neostriatal enkephalin-immunoreactive neurones project to the globus pallidus. Brain Res1982;231:1–17.

7. Vincent SR, Hokfelt T, Christensson I, Terenius L. Immunohisto-chemical evidence for a dynorphinergic immunoreactive strion-igral pathway. Eur J Pharmacol 1982;85:251–252.

8. Gerfen CG, Young WS III. Distribution of striatonigral and stria-topallidal peptidergic neurons in both patch and matrix compart-ments: an in situ hybridisation histochemistry and fluorescentretrograde tracing study. Brain Res 1988;460:161–167.

9. Seizinger BR, Grimm C, Hollt V, Herz A. Evidence for aselective processing of proenkephalin B into different opioidpeptide forms in particular regions of rat brain and pituitary.J Neurochem 1984;42:447– 457.

10. Breslin MB, Lindberg I, Benjannet S, Mathis JP, Lazure C,Seidah NG. Differential processing of proenkephalin by pro-hormone convertases 1(3) and 2 and furin. J Biol Chem 1993;268:27084 –27093.

11. Gerfen CR, Engber TM, Mahan LC, Susel Z, Chase TN, MonsmaF Jr, Sibley DR. D1 and D2 dopamine receptor-regulated geneexpression of striatonigral and striatopallidal neurons. Sci-ence1990;250:1429–1432.

12. Engber TM, Susel Z, Kuo S, Gerfen CR, Chase TN. Levodopareplacement therapy alters enzyme activities in striatum and neu-ropeptide content in striatal output regions of 6-hydroxydopaminelesioned rats. Brain Res.1991;552:113–118.

13. Henry B, Brotchie JM. Potential of opioid antagonists in thetreatment of levodopa-induced dyskinesias in Parkinson’s disease.Drugs Aging1996;9:1–10.

14. Duty S, Brotchie JM. Enhancement of the behavioral response toapomorphine administration following repeated treatment in the6-hydroxydopamine-lesioned rat is temporally correlated with arise in striatal preproenkephalin-B, but not preproenkephalin-A,gene expression. Exp Neurol 1997;144:423–432.

15. Andersson M, Hilbertson A, Cenci MA. Striatal fosB expression iscausally linked with l-DOPA-induced abnormal involuntary move-ments and the associated upregulation of striatal prodynorphinmRNA in a rat model of Parkinson’s disease. Neurobiol Dis1999;6:461–474.

16. Cenci MA, Lee CS, Bjorklund A. L-DOPA-induced dyskinesia inthe rat is associated with striatal overexpression of prodynorphin-and glutamic acid decarboxylase mRNA. Eur J Neurosci 1998;10:2694–2706.

17. Henry B, Crossman AR, Brotchie JM. Effect of repeated L-DOPA,bromocriptine, or lisuride administration on preproenkephalin-Aand preproenkephalin-B mRNA levels in the striatum of the 6-hy-droxydopamine-lesioned rat. Exp Neurol 1999;155:204–220.

18. Jenner P. Factors influencing the onset and persistence of dyskinesiain MPTP-treated primates. Ann Neurol 2000;47(Suppl.):90–104.

19. Henry B, Duty S, Fox SH, Crossman AR, Brotchie JM. Increasedstriatal pre-proenkephalin B expression associated with dyskinesiain Parkinson’s disease. Exp Neurol 2003;183:458–468.



20. Nisbet AP, Foster OJ, Kingsbury A, Eve DJ, Daniel SE, MarsdenCD, Lees AJ. Preproenkephalin and preprotachykinin messengerRNA expression in normal human basal ganglia and in Parkinson’sdisease. Neuroscience 1995;66:361–376.

21. Calon F, Birdi S, Rajput AH, Hornykiewicz O, Bedard PJ, Di PT.Increase of preproenkephalin mRNA levels in the putamen ofParkinson disease patients with levodopa-induced dyskinesias.J Neuropathol Exp Neurol 2002;61:186–196.

22. Piccini P, Weeks RA, Brooks DJ. Alterations in opioid receptorbinding in Parkinson’s disease patients with levodopa-induceddyskinesia. Ann Neurol 1997;42:720–726.

23. Henry B, Fox SH, Crossman AR, Brotchie JM. �- and -Opioidreceptor antagonists reduce levodopa-induced dyskinesia in theMPTP-lesioned primate model of Parkinson’s disease. Exp Neurol2001;171:139–146.

24. Klintenberg R, Svenningsson P, Gunne L, Andren PE. Naloxonereduces levodopa-induced dyskinesias and apomorphine-inducedrotations in primate models of parkinsonism. J Neural Transm2002;109:1295–1307.

25. Gomez-Mancilla B, Bedard PJ. Effect of nondopaminergic drugson L-DOPA-induced dyskinesias in MPTP-treated monkeys. ClinNeuropharm 1993;16:418–427.

26. Trabucchi M, Bassi S, Frattola L. Effects of naloxone on the“on-off” syndrome in patients receiving long-term levodopa ther-apy. Arch Neurol 1982;39:120–121.

27. Sandyk R, Snider SR. Naloxone treatment of L-dopa-induceddyskinesias in Parkinson’s disease. Am J Psychiatry 1986;143:118.

28. Price P, Baxter RC, Parkes JD, Marsden CD. Opiate antagonistsand Parkinson’s disease. Arch Neurol 1979;36:661.

29. Nutt JG, Rosin AJ, Eisler T, Calne DB, Chase TN. Effect of an opiateantagonist on movement disorders. Arch Neurol 1978;35:810–811.

30. Rascol O, Fabre N, Blin O, et al. Naltrexone, an opiate antagonists,fails to modify motor symptoms in patients with Parkinson’sdisease. Mov Disord 1994;9:437–440.

31. Manson AJ, Katzenschlager R, Hobart J, Lees AJ. High dosenaltrexone for dyskinesias induced by levodopa. J Neurol Neuro-surg Psychiatry 2001;70:554–556.

32. Delitala G, Giusti M, Mazzocchi G, Granziera L, Tarditi W,Giordano G. Participation of endogenous opiates in regulation ofthe hypothalamic-pituitary-testicular axis in normal men. J ClinEndocrinol Metab 1983;57:1277–1281.

33. Bonuccelli U, Piccini P, Del Dotto P, Rossi G, Corsini GU,Muratorio A. Naloxone partly counteracts apomorphine side ef-fects. Clin Neuropharmacol 1991;14:442–449.

34. Limone P, Calvelli P, Altare F, Ajmone-Catt P, Lima T, MolinattiGM. Evidence of an interaction between alpha-MSH and opioidsin the regulation of gonadotropin secretion in man. J EndocrinolInvest 1997;20:207–210.

35. Coiro V, Volpi R, Capretti L, et al. Different effects of naloxone onthe growth hormone response to melatonin and pyridostigmine innormal men. Metabolism 1998;47:814–816.

36. Tomasi PA, Fanciulli G, Palermo M, Pala A, Demontis MA,Delitala G. Opioid receptor blockade blunts growth hormone (GH)secretion induced by GH-releasing hormone in the male. HormMetab Res 1998;30:34–36.

37. Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of clinicaldiagnosis of idiopathic Parkinson’s disease: a clinico-patholog-ical study of 100 cases. J Neurol Neurosurg Psychiatry 1992;55:181–184.

38. Langston JW, Widner H, Goetz CG, Brooks D, Fahn S, FreemanT, Watts R. Core assessment program for intracerebral transplan-tations (CAPIT). Mov Disord 1992;7:2–13.

39. Evans JM, Hogg MI, Lunn JN, Rosen M. Degree and duration ofreversal by naloxone of effects of morphine in conscious subjects.Br Med J 1974;2:589–591.

40. Berkowitz BA. The relationship of pharmokinetics to pharmaco-logical activity, morphine, methadone and naloxone. Clin Pharma-cokinet 1976;1:219–230.

41. Ngai SH, Berkowitz BA, Yang JC, Hempstead J, Spector S. Pharma-cokinetics of naloxone in rats and in man: basis for its potency andshort duration of action. Anesthesiology 1976;44:398–401.

42. Manson AJ, Schrag A, Lees AJ. Low-dose olanzapine for levodopainduced dyskinesias. Neurology 2000;55:795–799.

43. Sieradzan KA, Fox SH, Hill M, Dick J, Crossman AR, BrotchieJM. Cannabinoids reduce levodopa-induced dyskinesia in Parkin-son’s disease: a pilot study. Neurology 2001;57:2108–2111.

44. Rascol O, Nutt JG, Blin O, et al. Induction by dopamine D1receptor agonist ABT-431 of dyskinesia similar to levodopa inpatients with Parkinson disease. Arch Neurol 2001;58:249–254.

45. Rascol O, Arnulf I, Peyro-Saint Paul H, et al. Idazoxan, an alpha-2antagonist, and L-DOPA-induced dyskinesias in patients with Par-kinson’s disease. Mov Disord 2001;16:708–713.

46. Verhagen Metman L, Del Dotto P, Natte R, van den MunckhofP, Chase TN. Dextromethorphan improves levodopa-induceddyskinesias in Parkinson’s disease. Neurology 1998;51:203–206.

47. Verhagen Metman L, Del Dotto P, van den Munckhof P, Fang J,Mouradian MM, Chase TN. Amantadine as treatment for dyski-nesias and motor fluctuations in Parkinson’s disease. Neurology1998;50:1323–1326.

48. Strong JA, Dalvi A, Samaha FJ, Gong J, Xu K, Yue X, Yu L. Muopioid receptor polymorphisms and L-Dopa induced dyskinesia inParkinson’s disease. Proc Soc Neurosci 1999;25:846.2.

Grip Force Abnormalities inDe Novo Parkinson’s Disease

Stuart J. Fellows, PhD,* and Johannes Noth, MD

Neurologische Klinik, Universitatsklinikumder RWTH Aachen

Abstract: In recent years it has been shown that a variety ofmovement disorders are associated with abnormalities ofthe fine motor control of the hand. In Parkinson’s disease(PD), these changes consist of a slowing of the rate of gripforce development and the use of abnormally large gripforces both during lifting and static holding of an object. Ithas been suggested, however, that these changes are a directeffect of the patient’s levodopa medication or associatedwith levodopa induced dyskinesias. Accordingly, we exam-ined the performance of de novo Parkinson patients in aprecision lifting task. All patients (n � 6) were newly diag-nosed and showed rigidity, bradykinesia, or both, but wereunaffected by tremor or dyskinesia. None of the patients

*Correspondence to: Dr. Stuart Fellows, Neurologishe Klinik, Uni-versitatsklinikum der RWTH Aachen, Pauwelsstr. 30, D-52074Aachen, Germany. E-mail: [email protected]

Received 1 August 2003; Revised 25 September 2003; Accepted 13October 2003


560 S.J. FELLOWS AND J. NOTH


had received antiparkinson medication. Grip force wasabnormally high in both the lifting and hold phases. Thisexaggeration was equal in magnitude to that observed pre-viously in medicated patients. Thus we conclude that theabnormalities in grip force observed here are intrinsic fea-tures of PD and not the result of dopamine medication or itsside effects. © 2003 Movement Disorder Society

Key words: Parkinson’s disease; de novo; precision grip

In recent years it has been shown that a variety ofmovement disorders are associated with abnormalities offine motor control of the hand.1–7 In Parkinson’s disease(PD), these changes consist of a slowing of the rate ofgrip force development8 and the use of abnormally largegrip forces, during both lifting and static holding of anobject.9,10 It has been suggested, however, that thesechanges were a direct effect of the patient’s levodopa(L-dopa) medication.11 This claim is somewhat surpris-ing, given the improved quality of movement generallyreported by the patients themselves, and indeed, L-dopamedication has been shown to markedly improve reach-to-grasp movements in patients with PD.12 A more likelysuggestion was that the exaggerated grip force levelsresulted from L-dopa induced dyskinesias.13 Accord-ingly, we examined the performance of de novo Parkin-son patients in a precision lifting task. These patientswere in the early stages of the disease and did not exhibittremor or dyskinesia as part of their symptoms. They hadhad no exposure to L-dopa or other dopaminergic medi-cation, and so their performance clearly could not beinfluenced, directly or indirectly, by effects of L-dopa.We show that they demonstrated grip force abnormalitiescompatible with those of a group of parkinsonian pa-tients on a stable L-dopa regime,9 indicating that theabnormalities are an intrinsic feature of the pathophysi-ology of PD. An alternative explanation for these deficitsis discussed.

SUBJECTS AND METHODS

The study involved 6 patients who were referred to ouroutpatient clinic with a suspected and subsequently con-firmed diagnosis of PD (Table 1). In cases of hemipar-kinsonism, the affected hand was studied, whereas in theother cases the dominant hand was used. None of thepatients was receiving or had received parkinsonianmedication. A control group comprised 12 age-matchedsubjects (6 men, 6 women; mean age, 61 � 3 years) withno history of neurological disorder. All subjects gavetheir informed consent to the procedures, which had beenapproved previously by the local ethics committee.

Details of the apparatus and methods employed havebeen fully described elsewhere.9 Briefly, the investiga-tion was carried out in a quiet room with subdued light-ing. The subject was seated in a stable chair that sup-ported the back (but not the head) before a table on whichwas situated the lifting device. Subjects were positionedso that they were able to grip the object between theirforefinger and thumb and lift and hold the object at thewrist while their elbow remained fully supported on apadded rest. The measuring instruments built into thedevice registered the grip force exerted on the object(9301b; Kistler, Winterhur, Switzerland) and its verticalposition (T60500; VAC, Munchen, Germany). Thesesignals were amplified and then passed to the analogue-to-digital converter board (NI-PCI-MIO-16XE; NationalInstruments, Austin, TX) of a laboratory computer(Macintosh PPC 7600/132; Apple, Cupertino, CA) sam-pling each channel at 2.5 kHz.

The subjects were required, without visual feedbackconcerning hand position, to grip and lift the object 4 to6 cm above the table, then hold it steady for 6 to 8seconds before replacing the object on the table andreleasing it. The contact pads on the object for thumb andforefinger were covered with sandpaper (extra-fine, corn400). A second laboratory computer (Macintosh IIVX;Apple) was used to control the load of the object via aservo-device. A torque motor attached via a nonelasticband to the object was used to alter object load betweenlifts without the subject’s knowledge in a pseudo-randommanner between two levels, namely 3.3 N (light) and 7.8N (heavy), such that five lifts could be selected for eachload where the load remained unaltered from the preced-ing lift. A 10- to 15-second pause was allowed betweeneach lift.

TABLE 1. Clinical details of the patients

Patientno. Gender

Age(yr) Main symptoms

H&Ystagea

1 F 61 Akinesia, rigor 12 M 70 Akinesia, bradykinesia 1.53 F 46 Akinesia, bradykinesia, rigor 2.54 M 49 Akinesia, bradykinesia 15 F 60 Akinesia, bradykinesia 16 M 73 Akinesia, bradykinesia 1.5

aStage 0, no signs of disease; Stage 1, unilateral disease; Stage 1.5,unilateral plus axial involvement; Stage 2, bilateral disease withoutimpairment of balance; Stage 2.5, mild bilateral disease with recoveryon pull test; Stage 3, mild to moderate bilateral disease, some posturalinstability, physically independent; Stage 4, severe disability, still ablewalk or stand unassisted; Stage 5, wheelchair-bound or bedriddenunless aided.

H&Y, Hoehn and Yahr.

GRIP FORCE ABNORMALITIES IN DE NOVO PD 561


The grip force curves obtained from each of the liftscarried out was measured subsequently (see Fig. 1) toyield a series of parameters: (1) IGL, the time betweenthe onset of grip force and the lift-off of the object(msec); (2) TPGF, the time taken to reach the peak gripforce (msec); (3) PGF, peak grip force magnitude (N);and (4) SGF, the stable grip force adopted while holdingthe object steady above the table (N). The IGL may beconsidered to provide a measure of the co-ordinationbetween the fingers gripping the object and more prox-imal arm muscles responsible for the actual horizontallift of the object. TPGF provides information about therate of grip force development at the fingers. PGF pro-vides information on the largely automatic processes ofthe selection from memory of motor sets matched toobject properties,14 whereas the SGF is the result ofmodification of these stored commands by actual sensoryfeedback concerning object properties obtained duringthe lift itself.15

Statistical analysis was carried out using the Statview5.0 package (SAS Institute, Cary, NC). For this purpose,the median value obtained from five lifts with a givenload were obtained for each parameter and comparedbetween subjects using MANOVA analysis with clinical

status and object load as the main factors. Post-hoctesting was carried out using the Tukey-Kramer test.

RESULTS

All 6 de novo PD patients displayed obvious abnor-malities in their grip force curves. Figure 2 shows 5 gripforce profiles obtained while lifting a light load for arepresentative control subject (upper traces) and a patientwith PD (lower traces). It is apparent that the patientdeveloped grip force markedly slower than did the con-trol subject, and consistently employed exaggerated lev-els of grip force, in both the dynamic and static phases ofthe lift.

The group values for the four lifting parameters aredisplayed in Figure 3. Each dot represents the value of asingle parkinsonian patient, whereas the grey boxes rep-resent the mean value for the control group (�SEM).The mean values for the patients with PD are shown asfilled triangles. Figure 3A shows the data for the IGL. Itmay be seen that the parkinsonian patients all demon-strated timings outside or at the upper end of the range ofvalues shown by the control group. On a group basis, thisprolongation was highly significant (P � 0.01) and itsmagnitude was comparable with that observed in an

FIG. 2. The 5 grip force profiles obtained while lifting the light loadby a typical control subject (upper traces) and a patient with PD (4,lower traces). Grip force development clearly was slower in the patient,who also developed consistently excessive force levels during both thedynamic lifting phase and the static hold phase.

FIG. 1. The grip force and object position curves for a typical controlsubject lifting the light load. The parameters obtained from these curvesare numbered. 1, Time between onset of grip force development andobject lift-off (IGL), a measure of finger/wrist co-ordination; 2, timetaken to achieve peak grip force (TPGF), a measure of the rate for gripforce development; 3, peak grip force (PGF) developed in the dynamiclifting phase, a measure of pre-planned matching of grip force to objectproperties; and 4, static grip force (SGF), the grip force developedwhile holding the object steady above the table, a measure of grip forceadaptation to actual conditions based on cutaneous afferent feedback.



earlier study9 involving patients at a later stage of thedisease who were on a stable regime of L-dopa medication.IGL modulation with load remained significant (P � 0.05)in the parkinsonian patients. The relative increase overcontrol values (�60%) was equal for both loads.

TPGF values for the parkinsonian patients (Fig. 3B)were also prolonged relative to the control group (P �0.05). This prolongation was less marked, however, thanthat observed previously in patients with a longer diseaseduration.9 The modulation of timing with load observedin the latter group of patients and in the control groupwas not significant in the group of de novo patients.

The most pronounced abnormalities shown by the denovo patients were observed in the exaggerated levels ofgrip force employed in both the dynamic and staticphases of the lift. PGF values (Fig. 3C) for patients weresignificantly higher than control values (P � 0.01) forboth the light and the heavy load, although the scaling of

grip force to load was retained (P � 0.01). It is interest-ing to note that the exaggeration was more marked forthe light load (on average, twice the mean level of thecontrol group) than for the heavy load (on average, anincrease of two-thirds). The exaggeration was also moremarked than that observed previously in the patients witha longer duration of disease,9 although the fact that thelatter were receiving L-dopa medication must be borne inmind. Figure 3D shows the data for SGF and it can beseen that a significant exaggeration of grip force levels inde novo PD patients occurred (P � 0.01), even moremarked than in the dynamic phase of the lift, particularlyfor the light load, where the exaggeration over controlvalues was (in relative terms) almost twice as great asthat observed with the heavy load. Scaling of grip forceto load was retained (P � 0.01). Once again, the exag-geration was more marked than that observed previouslyin patients with a longer duration of disease.9

FIG. 3. Group values for the four lifting pa-rameters. Each dot represents the value of asingle parkinsonian patient, whereas grey boxesrepresent the mean value for the control group(�SEM). The parkinsonian mean is representedby a filled triangle. All four parameters showeda significant increase over control values in PD(P � 0.01, except TPGF, P � 0.05). The adap-tation of the values according to load seen in thecontrol group was maintained in PD, with theexception of TPGF (IGL, P � 0.05; PGF, SGF,P � 0.01). The percentage values under eachload give the increase in the parkinsonian meanrelative to the control mean. Abnormalities aremore evident with the lighter load, and gripforce is exaggerated especially in the staticholding phase.

GRIP FORCE ABNORMALITIES IN DE NOVO PD 563


DISCUSSION

An unequivocal result of this study is that grip forceabnormalities were present in the early stages of PD, inpatients with no exposure to L-dopa medication. Further-more, the greatest abnormalities were observed in gripforce magnitude. Thus the hypothesis put forward byGordon and Reilmann,11 that exaggerated grip forces area side effect of L-dopa medication, is contradicted by ourfindings. Another suggestion, namely that exaggeratedgrip force levels result from L-dopa-induced dyskine-sia,13 can also be ruled out as an explanation for ourfindings, as none of the patients showed dyskinesia andas de novo patients had had clearly no chance to developL-dopa-induced dyskinesia. Indeed, the grip force pro-files of parkinsonian patients with L-dopa-induced dys-kinesia (Wenzelburger and associates,13 Fig. 2C) resem-ble much more closely those of patients withHuntington’s disease (see Hermsdorfer and colleagues,5

Fig. 2A) than the markedly slowed profiles obtained inthe present study and in a group of PD patients on astable L-dopa regime,9 indicating that a different patho-physiology may underlie the two phenomena.

Although this is more a matter of interpretation, wewould also argue that the present results provide strongsupport for the hypothesis that grip force abnormalities in avariety of basal ganglia disorders result from a disturbanceof sensorimotor processing.5,9,16 This idea arose in partfrom the similarities between exaggerated grip force pro-files seen in basal ganglia disorders and those of neurolog-ically normal subjects with a local anesthesia-induced blockof the cutaneous receptors of the hand and wrist,17 or inpatients with sensory neuropathy of the fingers.18 Peripheralsensory function, however, is largely normal in basal gan-glia disorders, implying that disordered central processingof sensory input is the likely cause of any abnormalities.Reports of abnormal sensorimotor function in PD haveappeared steadily in recent years. In particular, parkinsonianpatients performing arm movements without visual guid-ance regularly underestimate the extent of a movement,either passively imposed or self-performed,19–22 and havedifficulty in judging arm or finger position on the basis ofproprioceptive information alone.23 A similar deficit hasbeen observed when patients replicating a movement im-posed on the other hand must rely solely on kinestheticinformation or receive visual cues designed to distract themfrom relevant kinesthetic input.24 It was found recently thatpatients with PD lack the inhibition of responses in wristmuscles to transcortical magnetic stimulation seen normallyduring passive movement of the joint at which the targetmuscles operate.25 Taken together, these results indicate

that patients with PD suffer a decreased sensitivity to sen-sory input acting on structures at a cortical level. We wouldargue that this is supported further by our findings. Firstly,although the scaling of grip force levels to load was main-tained, a general shift to larger values was found. Secondly,the extent of the exaggerated grip force relative to thecontrol values was significantly higher for the light loadthan for the heavy load. As more afferent input would beexpected in the latter case, it could be argued that theexaggerated grip force levels result from a degree of insen-sitivity to afferent input caused by an upward shift in thethreshold level at which sensory input can act effectively.The finding that the relative grip force abnormalities aremost pronounced for the SGF, the control of which reliesheavily on cutaneous feedback information, further sup-ports this hypothesis.

In summary, we conclude that exaggerated grip forceobserved in patients with PD are intrinsic features of thepathophysiology of the disease, and not the result ofdopamine medication or its side effects. Rather, we sug-gest that abnormalities arise from decreased efficiency inthe utilization of sensory input concerning object prop-erties and the performance of the motor apparatus.

Acknowledgments: This work was supported by grantsfrom the Deutsche Forschungsgemeinschaft as part of the pro-gram SPP 1001 “sensomotorische Integration”. We thank Pro-fessor M. Schwarz, Dr. R. Topper, and Dr. P. Weiss-Blanken-horn for fruitful discussions of our findings.

REFERENCES

1. Fellows SJ, Ernst J, Schwarz M, Topper R, Noth J. Precision gripdeficits in cerebellar disorders in man. Clin Neurophysiol 2001;112:1793–1802.

2. Fellows SJ, Schwarz M, Schaffrath C, Domges F, Noth J. Senso-rimotor deficits and the delay of compensatory responses to slip ofan object held in the fingers of patients with Huntington’s disease.Electroencephalograph Clin Neurophysiol 1998;106:41–42P.

3. Gordon A, Quinn L, Reilmann R, Marder K. Coordination ofprehensile forces during precision grip in Huntington’s disease.Exp Neurol 2000;163:136–148.

4. Hermsdorfer J, Wessel K, Mai N, Marquardt C. Perturbation ofprecision grip in Friedrich’s ataxia and late-onset cerebellar ataxia.Mov Disord 1994;9:650–654.

5. Schwarz M, Fellows SJ, Schaffrath C, Noth J. Deficits in sensori-motor control during precise hand movements in Huntington’sdisease. Clin Neurophysiol 2001;112:95–106.

6. Serrien D, Burgunder JM, Wiesendanger M. Disturbed sensorimo-tor processing during control of precision grip in patients withwriter’s cramp. Mov Disord 2000;15:965–972.

7. Serrien D, Wiesendanger M. Grip-load force co-ordination incerebellar patients. Exp Brain Res 1999;128:76–80.

8. Ingvarsson P, Gordon A, Forssberg H. Coordination of manipula-tive forces in Parkinson’s disease. Exp Neurol 1997;145:489–501.

9. Fellows SJ, Noth J, Schwarz M. Precision grip and Parkinson’sdisease. Brain 1998;121:1771–1784.

10. Muller F, Abbs J. Precision grip in Parkinsonian patients. In:Streifler M, Korczyn A, Melamed E, Youdim M, editors. Advances



in neurology, Vol.53. Parkinson’s disease: anatomy, pathology andtherapy. New York: Raven Press; 1990. p 191–195.

11. Gordon A, Reilmann R. Getting a grasp on research: does treatmenttaint testing of parkinsonian patients? Brain 1999;122:1597–1598.

12. Castiello U, Bennett K, Bonfiglioli C, Peppard R. The reach-to-grasp movement in Parkinson’s disease before and after dopami-nergic medication. Neuropsychologia 2000;38:46–59.

13. Wenzelburger R, Zhang BR, Pohle S, et al. Force overflow andlevodopa-induced dyskinesias in Parkinson’s disease. Brain 2002;125:871–879.

14. Gordon A, Westling G, Cole K, Johanson RS. Memory represen-tations underlying motor commands used during manipulation ofcommon and novel objects. J Neurophysiol 1993;69:1789–1796.

15. Westling G, Johansson RS. Factors influencing the force controlduring precision grip. Exp Brain Res 1984;53:277–284.

16. Noth J, Fellows SJ, Schwarz M, Topper R. Disturbances of preci-sion grip in movement disorders. Pflugers Arch 2001;441:R117.

17. Macefield V, Hager-Ross C, Johansson RS. Control of grip forceduring restraint of an object held between finger and thumb:responses of cutaneous afferents from the digits. Exp Brain Res1996;108:155–171.

18. Thonnard JL, Detrembleur C, Van den Bergh P. Assessment ofhand function in a patient with chronic sensory demyelinatingneuropathy. Neurology 1997;49:253–257.

19. Demirci M, Grill S, McShane L, Hallet M. A mismatch betweenkinesthetic and visual perception in Parkinson’s disease. Ann Neu-rol 1997;41:781–788.

20. Jobst E, Melnick M, Byl N, Dowling G, Aminoff M. Sensoryperception in Parkinson disease. Arch Neurol 1997;34:450–454.

21. Klockgether T, Borutta M, Rapp H, Spieker S, Dichgans J. A defectof kinesthesia in Parkinson’s disease. Mov Disord 1995;10:460–465.

22. Schneider J, Diamond S, Markham C. Parkinson’s disease: sensoryand motor problems in arms and hands. Neurology 1987;37:951–956.

23. Zia S, Cody F, O’Boyle D. Joint position sense is impaired byParkinson’s disease. Ann Neurol 2000;47:218–228.

24. Byblow WD, Lewis GN, Stinear JW. Effector-specific visual in-formation influences kinesthesis and reaction time performance inParkinson’s disease. J Mot Behav 2003;35:97–102.

25. Lewis GN, Byblow WD. Altered sensorimotor integration in Par-kinson’s disease. Brain 2002;125:2089–2099.

Verb and Noun Generation Tasksin Huntington’s Disease

Patrice Peran, MSc,*Jean-Francois Demonet, MD, PhD, Cyril Pernet, MSc,

and Dominique Cardebat, PhDInstitut National de la Sante et de la Recherche Medicale U

455, Federation de Neurologie, Centre HospitalierUniversitaire Purpan, Toulouse, France

Abstract: We compared noun- and verb-generation tasks ina demented group (n � 9, Dementia Rating Scale < 129)

and in a non-demented group (n � 17, Dementia RatingScale > 129) of Huntington’s disease (HD) patients com-pared to 26 matched normal subjects. We did not find aspecific deficit for verb production in non-demented pa-tients who had a performance similar to but weaker thanthat of the controls across the four tasks. The profile ofresults was different in the demented group because, apartfrom a global deficit whatever the task in comparison withboth non-demented and control groups, the demented pa-tients exhibited increased difficulties in the two tasks im-plying verb production. The deficit of verb production ob-served in demented HD patients is discussed in relation tothe damage to the motor loop in HD patients at later stagesof disease. © 2003 Movement Disorder Society

Key words: Huntington’s disease; language; word genera-tion; verb

Besides motor deficiencies, Huntington’s disease(HD) is characterized by several cognitive deficits inmemory and executive functions such as problem solv-ing, visuoperceptive processing, and spatial or arithmet-ical reasoning.1 Language disorders mainly concern“frontal” aspects such as some features of syntactic abil-ities2,3 and verbal fluency.4 Ho and colleagues4 specifiedthat this fluency deficit seems to concern specifically theability to switch across subcategories, reflecting impair-ment of frontostriatal circuits, rather than the size ofclusters within subcategories.

The frontal nature of language impairment in HD canbe envisaged in another paradigm that concerns a differ-ential deficit between verb and noun processing. Clinicalevidence5–9 shows a relationship between object-namingdeficit and damage to the left temporal lobe, and betweenaction-naming deficit and large lesions in the left frontalcortex. Damasio and Tranel7 formulated the hypothesisthat in the left hemisphere, noun retrieval is mediatedpreferentially by temporal regions, whereas verb re-trieval is subserved by a large network including theprefrontal cortex; this hypothesis has received furthersupport in studies devoted to degenerative diseases af-fecting the frontal cortex. A verb deficit was observed infrontotemporal dementia10 and in motor neuron diseaseassociated with pathological changes in two frontal ar-eas, namely Brodmann areas 44 and 45.11 Languagestudies in Parkinson’s disease (PD) have also revealedthe existence of such dysfunction; Grossman and col-leagues12 showed a verb learning impairment in a groupof early PD patients. Our group has shown recently thatword generation tasks may be useful to unveil in non-demented PD patients a specific impairment of verbproduction compared with noun generation.13

*Correspondence to: Patrice Peran, INSERM U 455, CHU Purpan, 31059Toulouse Cedex 3, France. E-mail: [email protected]

Received 26 April 2003; Revised 23 July 2003, 8 October 2003;Accepted 14 October 2003


VERB PROCESSING IN HUNTINGTON’S DISEASE 565


The same protocol was used here in HD patientscompared with matched controls to show a possibledeficit of verb production in a disease that leads to deeperand earlier dysfunction of frontal cortex than that ob-served in PD. The word generation tasks we used seemappropriate to test this hypothesis in HD. Indeed, thetasks involve auditory input, bypassing the visual deficitsemphasized in earlier studies on HD,2 and they requirethe controlled production of a single response per item,thus alleviating articulatory, planning, and workingmemory load that HD patients may encounter in fluencytasks.

SUBJECTS AND METHODS

Subjects

A series of 26 right-handed French HD patients wereenrolled over a 1-year period at a dedicated clinic for HDin the Neurology Department at Toulouse UniversityHospital. All patients had a positive family history of HDand more than 36 CAG repeats upon genetic testing forthe IT-15 mutation.14 None of the patients had any his-tory of neurological or psychiatric disease other than HD.Their motor disability was evaluated using the motor partof the Unified Huntington’s Disease Rating Scale (UH-DRS) and we only included in the study patients that didnot present severe dysarthria (mean UHDRS dysarthriascore, 1 � 0.4). Cognitive status was assessed using theMini-Mental State Examination (MMSE; mean score,25 � 2.6; range, 21–29). In addition, a global assessmentof cognitive functions was obtained using the DementiaRating Scale (DRS; mean score, 131.9 � 7.6; range,120–144).

Two groups of patients were considered based on DRSscores with a cut-off of 129, which indicates normalcognitive functioning in the elderly:15 (1) a non-de-

mented group (ND) including 17 patients with DRSscores �129; and (2) a demented group (D) including 9patients with DRS scores �129.

A control group (C) of 26 right-handed French sub-jects, closely matched to HD patients for age, gender,and education, was included (MMSE mean score, 29.1 �0.8). Demographic and clinical data of the three groupsare reported in Table 1.

Methods

Word generation was assessed with 40 concrete nounsand 40 action verbs matched for lexical frequency16 andlength (di- and trisyllabic). The protocol, published pre-viously,17 consisted of two intra-category tasks, noun tonoun (NN) and verb to verb (VV) generation, in whichsubjects were instructed to produce a semantically re-lated noun or verb when listening to a noun or a verb, andtwo inter-category tasks, noun to verb (NV) and verb tonoun (VN) generation, in which subjects were presentedwith either a noun or a verb and instructed to produce asemantically related item from the other category. Eachtask included 40 items. The same list of nouns compris-ing 20 biological items (e.g., abeille, bee; champignon,mushroom; coton, cotton) and 20 manufactured items(e.g., marteau, hammer; chapeau, hat; pinceau, paintbrush) was used for NN and NV tasks. Similarly, a list of40 action verbs (e.g., laver, to wash; coudre, to sew;piloter, to drive) was used for VN and VV tasks.

Subjects were not asked to generate new responses toeach stimulus within a condition. In each group, the fourtasks were presented in the order NN, VN, NV, and VVfor half of the subjects and in reverse order for the secondhalf. The task had to be carried out within a period of 4seconds for control subjects and 6 seconds for HD pa-tients to account for the slowed articulatory rate often

TABLE 1. Demographic and clinical data

ParameterControl subjects

(n � 26)Non-demented HD patients

(n � 17)Demented HD patients

(n � 9)

Age (yr) 45.6 � 10.7 43.2 � 10.8 51.7 � 8.1Education (�9 yr/�9 yr) 14/12 10/7 4/5Gender (M/F) 15/11 8/9 7/2MMSE 29.1 � 0.8 25.5 � 2.1 24.3 � 3.2Disease duration (yr) — 4.8 � 1.6 8.9 � 2.4DRS score (of 144) — 136.4 � 5.1 123.9 � 3.1a

UHDRS motor score — 30.2 � 10.3 44.2 � 12.3b

Values are means � SD.aP � 0.0001.bP � 0.005.HD, Huntington’s disease; MMSE, Mini Mental State Examination; DRS, Dementia Rating Scale; UHDRS,

Unified Huntington’s Disease Rating Scale.

566 P. PERAN ET AL.


present in HD. Before each task, an example was given(e.g., for VV task, “If I tell you boire [drink], you mayanswer manger [eat] or avaler [swallow]”).

We calculated the number of correct responses, i.e.,semantically and grammatically adequate productionswithin the given time period, accepting responses that in-cluded a determiner followed by a noun for nouns or apronoun followed by a verb for verbs. The errors wereclassified as no-production errors, grammatical errors (i.e.,production of an item semantically but not grammaticallyadequate like a verb or, more rarely, an adjective when anoun is required), semantic errors, and others (for example,neologisms). In French, some words can be ambiguous interms of grammatical category, for example brike can be aconjugated or infinitive form of the verb briquer (to scrub)or a noun i.e., briquet (lighter). In case of ambiguity, theresponse was scored as correct if the semantic relationshipcould disentangle the grammatical ambiguity. For example,in verb/verb condition, the response briquer (to scrub) wasconsidered as correct for the target cirer (to polish) based onthe semantic vicinity. The semantic relationship was scoredby two independent judges and only congruent judgmentswere considered.

The item lists may involve some subsets in which theword stimuli belonged to the same semantic category,therefore leading subjects to produce repeatedly the sameresponse. For instance, subjects may respond to the wordgrapes, orange, or olive with the same verb, to eat.Although repeated, such responses were considered ap-propriate. These responses were computed separately ineach subject group. In addition, perseverative responsescould be observed, i.e., inappropriate repetition of agiven response, and perseverations were computed also.

Statistical Analyses

Scores were compared using analysis of variance(ANOVA) and planned comparisons with group (ND, D,and C), task (NN, NV, VN, and VV), and error types asfactors. Nonparametric correlation analyses were used toassess the relationships between significantly impairedscores in HD and UHDRS motor scores.

RESULTS

Comparisons Between Groups for GlobalGeneration Performance

The scores on the four generation tasks are reported inFigure 1. A two-factor ANOVA (group and task) withrepeated measures showed a significant main effect forgroup (F[2,49] � 42.23, P � 10�4) with significantdifferences of performance between the three groups(controls � non-demented, P � 0.002; controls � de-mented, P � 10�4; non-demented � demented, P �10�4), a significant main effect for the task (F[3,147] �21.4, P � 10�4) with significant difference between VVand the other three tasks (Scheffe tests: VV/NN, P �0.001; VV/VN, P � 0.001; VV/NV, P � 0.0006) andbetween NV and NN (P � 0.003). Moreover, a signifi-cant group � task interaction was observed (P � 10�4).

To explore further this interaction, we assessed acrossthe three groups the influence of the noun/verb categoryconsidering either the required output (e.g., verb in verb/verb and noun/verb conditions) or the given stimulus(e.g., noun in noun/noun and noun/verb conditions).Three-factor ANOVA (group, required output, and givenstimulus), in which output and stimulus were consideredas repeated factors, showed significant main effects for:

FIG. 1. Global performance measured as errorrate (mean � SD) in HD patients and controlsubjects on word generation tasks.



(1) group (F[2,49] � 42.24; P � 10�4), with controlsubjects performing better than non-demented patients(P � 10�4) and non-demented patients performing betterthan demented patients (P � 10�4); (2) for requiredoutput (F[1,49] � 43.42; P � 10�4), with more errors inverb production than in noun production; and (3) forgiven stimulus (F[1,49] � 28.06, P � 10�4), with verbstimulus eliciting more errors than that elicited by nounstimulus.

A significant interaction between required output andgroup (F[2,49] � 13.52, P � 10�4) was found. Thisinteraction was induced by a larger error rate in verbproduction tasks (mean error rate, 26.8 � 9.7) than innoun production tasks (mean error rate, 16.8 � 8.5) indemented patients compared with both non-dementedpatients (verb production mean error rate, 12.1 � 6.7;noun production mean error rate, 9.9 � 5.6) and thecontrol group (verb production mean error rate, 5.75 �4.6; noun production mean error rate, 4.5 � 4.0). Pair-wise differences between noun and verb production errorrates were significant between demented and non-de-mented groups (post-hoc Scheffe test, P � 0.0005), andbetween demented and control groups (post-hoc Scheffetest, P � 10�4), whereas such differences were notobserved between non-demented and control subjects(P � 0.78).

Repeated Responses

Two-factor ANOVA showed that appropriate repeti-tions did not differ across groups but differed acrosstasks (F[3,147] � 33.24, P � 10�6) without group �task interaction. A larger number of repeated responses

was observed for NV in the 3 groups (post-hoc Scheffetests: NN/NV, P � 10�6; VN/NV, P � 10�6; VV/NV,P � 10�6).

No statistical analysis was carried out for persevera-tions because such responses were very rare (mean re-sponse less than 0.5 of 40 items, in any group and task).

Influence of Error Type

Because semantic errors, perseverations, and errorsqualified as “other” were very rare in the production ofthe subjects (less than one error on average, whatever thegroup or the task), we only included no-production andgrammatical errors in the analyses (Fig. 2). A three-factor ANOVA (group, task, and error type) showedsignificant differences for all main effects and interac-tions (main effects: group F[2,49] � 37.7, P � 10�6;task F[3,147] � 18.54, P � 10�6; error type F[1,49] �105.01, P � 10�6; interactions: group � task F[6,147] �4.67, P � 0.0003; group � error type F[2,49] � 14.63,P � 10�3; task � error type F[3,147] � 38.44, P �10�6; group � task � error type F[6,147] � 5.16, P �10�3). Further two-factor ANOVA (group, error type)was carried out for each task, again showing significantdifferences for all main effects and interactions (for NN,main effects: group F[2,49] � 20.86, P � 10�6; errortype F[1,49] � 35.22, P � 10�6; interactions: group �error type F[2,49] � 5.61, P � 0.007; for VN, maineffects: group F[2,49] � 11.6, P � 10�4; error typeF[1,49] � 34.46, P � 10�6; interactions: group � errortype F[2,49] � 8.63, P � 0.0007; for NV, main effects:group F[2,49] � 34.63, P � 10�6; error type F[1,49] �82.42, P � 10�6; interactions: group � error type

FIG. 2. No-production and grammatical errors (mean � SD) on word generation tasks. Diamonds, control subjects; squares, non-demented HDpatients; triangles, demented HD patients.

568 P. PERAN ET AL.


F[2,49] � 10.98, P � 0.0002; for VV, main effects:group F[2,49] � 33.28, P � 10�6; error type F[1,49] �137.98, P � 10�6; interactions: group � error typeF[2,49] � 16.98, P � 10�5).

For no-production errors, post-hoc comparisons(Scheffe tests) showed significantly more errors in de-mented patients than in the two other groups regardlessof the task (non-demented patients: NN, P � 0.007; VN,P � 0.0007; NV, P � 0.0001; VV, P � 0.0001; controlsubjects: NN, P � 0.0001; VN, P � 0.0001; NV, P �0.0001; VV, P � 0.0001). Comparisons between non-demented patients and control subjects showed that no-production errors were not more frequent for the twotasks requiring a noun production (NN and VN), whereassignificantly poorer performance was found in patientscompared with performance of control subjects for VVgeneration (P � 0.05), with a similar trend observed forNV (P � 0.09).

For grammatical errors, significant impairments in de-mented patients compared with control subjects wereobserved in all tasks (NN, P � 0.03; VN, P � 0.0001;NV, P � 0.0001), except in VV in which only a trendwas noted (P � 0.07). Demented patients performedsignificantly worse than non-demented patients only forNV generation (P � 0.0001), whereas no significantdifference was found for the three other tasks (NN, VN,and VV). Indeed, non-demented patients produced veryfew grammatical errors (1.18 � 1.13 on average) in NVgeneration although their grammatical errors reached thesame level as those of demented patients in the othertasks. Finally, non-demented patients made significantlymore grammatical errors than did control subjects forNN (P � 0.0006) and VN (P � 0.03) tasks, whereas nodifference was noted for verb generation tasks.

Language Performance and Competing Alternativesin HD Groups

Because the tasks required subjects to generate a verbor a noun in response to a noun or a verb, the stimuligiven can provide high or low selection constraint for therequired output choice, and the degree of competition forselection among different possible responses can accountfor abnormal performance (e.g., Thompson-Schill et al.,199818).

We then calculated, based on correct responses ofcontrols for each task, a ratio of the relative frequency ofthe most common response to the relative frequency ofthe second-most common response as a measure of re-sponse strength. This ratio was based on the methoddescribed by Thompson-Schill and colleagues.18

We assessed the influence of increased selection de-mand on performance in HD groups for the four tasks.We failed to find any significant correlation (Spearmanrank correlation) between the computed ratio and themean global error rate for all tasks and groups.

Language Performance and Motor Status

We sought to relate motor impairment measured byUHDRS motor scores and impaired language perfor-mance in patients.

In the whole group of patients, a trend was observedbetween UHDRS motor global score and two-verb pro-duction tasks (NV Spearman rank correlation, � � 0.37,P � 0.06; VV Spearman rank correlation, � � 0.34, P �0.09). Analyses using UHDRS subscores (motor dys-function, chorea, dystonia, and oculomotor 1 and 2) asdefined by the results of factorial analyses19 were morerevealing, as correlations were found only between per-formance on verb production tasks and subscores inmotor dysfunction (NV Spearman rank correlation, � �0.63, P � 0.0005; VV Spearman rank correlation, � �0.58, P � 0.002) and dystonia (NV Spearman rankcorrelation, � � 0.4, P � 0.05; VV Spearman rankcorrelation, � � 0.51, P � 0.008).

Correlation analyses were also carried out in eachgroup of patients. In demented patients, we did not findany significant correlation between UHDRS motor scoreand error rate regardless of the task. In non-dementedpatients, we observed only a trend in the correlationbetween NN error rate and UHDRS global motor score(Spearman rank correlation, � � �0.456, P � 0.065),indicating that some patients may perform poorly on thistask in the absence of severe motor deficit. Nevertheless,even for the latter task, no significant correlation wasfound between UHDRS subscores and performance.

DISCUSSION

Very few studies have been devoted specifically to theanalysis of language performance in HD patients, andthese addressed impairments of verbal fluency known tobe affected in this pathology. Two studies4,20 emphasizedin fluency tasks the importance of switching deficit be-tween categories as a marker of frontostriatal dysfunc-tion. Our study is the first to explore impairments ofword generation tasks in HD. Word generation sharescognitive features with fluency tasks and even in theirswitching dimension, because the serial presentation ofcue words is likely to make subjects explore differentsemantic fields as the task goes on. The global impair-ment observed in HD patients might be viewed as aswitching deficit; however, one might have expected a



more marked deficit in tasks implying switching betweencategories (VN and NV) than in purely verb- or noun-based tasks (VV and NN). In accordance with Randolphand colleagues,21 another interpretation would be im-paired lexical access due to the time-limited feature ofgeneration tasks and a lack of flexibility across semanticfields. These factors relate to the general framework ofexecutive functions, intervene in all generation tasks, andare likely to account for global impairment of HD pa-tients whatever the task.

We showed previously13 in non-demented PD patientsa specific deficit for verb production in generation tasksthat we interpreted as the presence of prefrontal dysfunc-tion in this neurodegenerative disease. We tested thishypothesis in HD by applying exactly the same method-ology in a demented and a non-demented group of HDpatients.

Our hypothesis was confirmed only partially. Indeed,considering the total errors, we found a specific deficitfor verb production in demented patients because theyexhibited increased difficulties in the two tasks implyingverb production, as shown by the larger pair-wise differ-ence for verb production than for noun production com-pared with both non-demented and control groups. Con-trary to our hypothesis, however, this effect was notfound in non-demented patients who performed worsethan control subjects but in a similar pattern across thefour tasks.

Although this finding should be considered with cau-tion due to the small sample size, a verb productiondeficit in demented HD patients is congruent with thatfound in non-demented PD patients.13 A possible expla-nation of this may be the relationship between verbproduction processes and the representation of actions7

that is probably impaired in pathologies affecting basalganglia circuitry involved in motor loops, such as in HDand PD.22 As proposed earlier,13 the verb productiondeficit in non-demented parkinsonian patients would re-late to dysfunction of the putaminal direct motor loop. Inthe earlier stages of HD, pathology begins in the dorsalcaudate nucleus, which is part of the dorsolateral pre-frontal cortex loop, and spreads gradually throughout thefrontostriatal system, with only minimal impairment ofthe temporal lobes.23 Early frontal/caudate dysfunctionmay account for the well-known frontal features in thecognitive profile in these patients, and for the generalimpairment observed in all generation tasks in this study.Lesions affecting the putaminal motor loop are likely tooccur at later stages, in which the cognitive deficitevolves to demented status. The marked and significantdifference of UHDRS motor score between demented

(44.2 � 12.3) and non-demented (30.2 � 10.3) groups inthe present study indicates an important motor deficit indemented HD patients and a possible involvement ofputaminal structures in the lesion pattern. We proposethat the verb production deficit observed in dementedHD patients reflects damage to the motor loop at laterstages of disease.

Degenerative diseases such as HD do not consist ofall-or-none phenomena but rather induce progressivechanges in terms of both lesion intensity and extent.Moderate damage to the motor loop in our non-dementedHD group is most likely, and therefore an incipientdeficit for verb production might exist. In non-dementedHD patients, however, results are complex and do notreadily support our hypothesis. A verb-specific effectwas not found in analysis of the global rate of errors;however, the pattern of errors, when considering theabsence of production, may be interpreted as an incipientverb processing deficit in this patient group. No differ-ence was noted between patients and controls in nounproduction tasks, whereas in verb production tasks thepatients seemed to have marked difficulties in accessingverb lexicon, as shown by an increase of no-productionerrors compared with that in control subjects. The re-verse pattern observed for grammatical errors (more er-rors than controls in noun production tasks and no dif-ference in verb production tasks) seems to detract fromthe hypothesis of a slight verb disturbance in this group.When considering both no-production errors (increasefor verb production) and grammatical errors (increase fornoun production) in non-demented patients, however, theinterplay between these two types of errors might sug-gest a hierarchical level of disturbance. When verb pro-duction is required, the deficit becomes important andleads to a failure of the lexical search, whereas whennoun production is required, patients tend to be moreproductive but with erroneous responses that are mainlygrammatical.

Finally, correlation analyses between UHDRS motorscores and language performance were consistent overallwith the motor hypothesis of a verb production deficit inHD patients. Significant correlations were found betweenverb production tasks and UHDRS subscores linkedtightly to action performance, i.e., motor dysfunction anddystonia. Indeed, the motor dysfunction subscore in-volved all items exploring forearm movements, and dys-tonia may interfere with such movements. The absenceof such correlations in the demented group is likely torelate to lowered statistical power in a small sample andreduced variance as all patients presented with severemotor symptoms. The absence of such correlations when

570 P. PERAN ET AL.


considering the group of non-demented patients is lessclear but could be related to important intersubject het-erogeneity in early disease stages, where some subjectspresent with marked impairment in some cognitive domainsand few if any motor symptoms, and vice versa.1,24

In conclusion, our results suggest a verb productiondeficit in word generation tasks at least in demented HDpatients, possibly related to damage to the motor system,that parallels the deficit we observed previously in par-kinsonian patients presenting mainly motor symptoms.Further studies with larger number of patients might helpconsolidate the hypothesis of a relationship between mo-tor and verb impairment in Huntington’s disease.

Acknowledgments: This work was supported by MENRTCognitique-Action 2000, and is part of the Language, Image,Movement and Cognition Project. We thank the staff of thePurpan Hospital Huntington’s Disease Clinic, Prof. P. Calvas,C. Pigois-Gayo, M.C. Deneuville, and H. Delabaere.

REFERENCES

1. Bachoud-Levi AC, Maison P, Bartolomeo P, et al. Retest effectsand cognitive decline in longitudinal follow-up of patients withearly HD. Neurology 2001;56:1052–1058.

2. Podoll K, Caspary P, Lange HW, Noth J. Language functions inHuntington’s disease. Brain 1988;111:1475–1503.

3. Murray LL, Lenz LP. Productive syntax abilities in Huntington’sand Parkinson’s diseases. Brain Cogn 2001;46:213–219.

4. Ho AK, Sahakian BJ, Robbins TW, Barker RA, Rosser AE,Hodges JR. Verbal fluency in Huntington’s disease: a longitudinalanalysis of phonemic and semantic clustering and switching. Neu-ropsychologia 2002;40:1277–1284.

5. Zingeser LB, Berndt RS. Retrieval of nouns and verbs in agram-matism and anomia. Brain Lang 1990;39:14–32.

6. Caramazza A, Hillis AE. Lexical organization of nouns and verbsin the brain. Nature 1991;349:788–790.

7. Damasio AR, Tranel D. Nouns and verbs are retrieved with dif-ferently distributed neural systems. Proc Natl Acad Sci USA1993;90:4957–4960.

8. Breedin SD, Saffran EM, Schwartz MF. Semantic factors in verbretrieval: an effect of complexity. Brain Lang 1998;63:1–31.

9. Daniele A, Giustolisi L, Silveri MC, Colosimo C, Gainotti G.Evidence for a possible neuroanatomical basis for lexical process-ing of nouns and verbs. Neuropsychologia 1994;32:1325–1341.

10. Cappa SF, Binetti G, Pezzini A, Padovani A, Rozzini L, TrabucchiM. Object and action naming in Alzheimer’s disease and fronto-temporal dementia. Neurology 1998;50:351–355.

11. Bak TH, O’Donovan DG, Xuereb JH, Boniface S, Hodges JR.Selective impairment of verb processing associated with patholog-ical changes in Brodmann areas 44 and 45 in the motor neuronedisease-dementia-aphasia syndrome. Brain 2001;124:103–120.

12. Grossman M, Stern MB, Gollomp S, Vernon G, Hurtig HI. Verblearning in Parkinson’s disease. Neuropsychology 1994;8:413–423.

13. Peran P, Rascol O, Demonet JF, Celsis P, Nespoulous JL, DuboisB, Cardebat D. Deficit of verb generation in nondemented patientswith Parkinson’s disease. Mov Disord 2003;18:150–156.

14. The Huntington’s Disease Collaborative Research Group. A novelgene containing a trinucleotide repeat that is expanded and unsta-ble on Huntington’s disease chromosomes. Cell 1993;72:971–983.

15. Monsch AU, Bondi MW, Salmon DP, et al. Clinical validity of theMattis Dementia Rating Scale in detecting Dementia of the Alz-

heimer type. A double cross-validation and application to a com-munity-dwelling sample. Arch Neurol 1995;52:899–904.

16. Baudot J. Frequences d’utilisation des mots en francais contempo-rain. Montreal: Presses de l’Universite de Montreal; 1992.

17. Cardebat D, Candelon N, Kuznierek L, Viallard G, Demonet JF,Lambert J. [Generation of isolated words in normal subjects]. ActaNeurol Belg 2000;100:24–33.

18. Thompson-Schill SL, Swick D, Farah MJ, D’Esposito M, Kan IP,Knight RT. Verb generation in patients with focal frontal lesions:a neuropsychological test of neuroimaging findings. Proc NatlAcad Sci USA 1998;95:15855–15860.

19. Shannon K, Raman R, Leurgans S. Primary clinical components ofHuntington’s disease: factor structure of the motor subscale of theUnified Huntington’s Disease Rating Scale. Neurology 1999;52(Suppl. 2):A120.

20. Rich JB, Troyer AK, Bylsma FW, Brandt J. Longitudinal analysisof phonemic clustering and switching during word-list generationin Huntington’s disease. Neuropsychology 1999;13:525–531.

21. Randolph C, Braun AR, Goldberg TE, Chase TN. Semantic fluency inAlzheimer’s, Parkinson’s, and Huntington’s disease: dissociation ofstorage and retrieval failures. Neuropsychology 1993;7:82–88.

22. Joel D. Open interconnected model of basal ganglia-thalamocorti-cal circuitry and its relevance to the clinical syndrome of Hunting-ton’s disease. Mov Disord 2001;16:407–423.

23. Hedreen JC, Folstein SE. Early loss of neostriatal striosome neu-rons in Huntington’s disease. J Neuropathol Exp Neurol 1995;54:105–120.

24. Soliveri P, Monza D, Piacentini S, et al. Cognitive and psychiatriccharacterization of patients with Huntington’s disease and theirat-risk relatives. Neurol Sci 2002;23(Suppl.):105–106.

Cerebrospinal Fluid AnalysisDifferentiates Multiple System

Atrophy From Parkinson’s Disease

W. Farid Abdo, MD,1,2 Danielle de Jong, MD,1

Jan C.M. Hendriks, MSc, PhD,3

Martin W.I.M. Horstink, MD, PhD,1

Berry P.H. Kremer, MD, PhD,1

Bastiaan R. Bloem, MD, PhD,1

and Marcel M. Verbeek, MSc, PhD1,2*1Department of Neurology, University Medical Center,

Nijmegen, The Netherlands; 2Laboratory of Pediatrics andNeurology, University Medical Center, Nijmegen, The

Netherlands; 3Department of Epidemiology and Biostatistics,University Medical Center, Nijmegen, The Netherlands

Abstract: We investigated whether cerebrospinal fluid (CSF)analysis discriminates between idiopathic Parkinson’s disease

*Correspondence to: Dr. M.M. Verbeek, University Medical CenterNijmegen, Department of Neurology, 319 LKN, PO Box 9101, 6500HB Nijmegen, The Netherlands. E-mail: [email protected]

Received 6 July 2003; Revised 4 October 2003; Accepted 15 Octo-ber 2003


CSF ANALYSIS IN MSA AND PD 571


(PD; n � 35) and multiple system atrophy (MSA; n � 30). Themedian CSF concentration of the neurotransmitter metabo-lites 5-hydroxyindolacetic acid (5-HIAA) and 3-methoxy-4-hydroxyphenylethyleneglycol (MHPG) was reduced signifi-cantly (49-70%) in MSA compared to PD. In contrast, severalbrain-specific proteins (tau, neuron-specific enolase, myelinbasic protein) were elevated (130–230%) in MSA comparedwith those in PD. A combination of CSF tau and MHPGdiscriminated PD from MSA (adjusted odds ratios: tau, 27.2;MHPG, 0.14). Our data suggest that the more progressive andwidespread neurodegenerative nature of MSA, as comparedwith PD, is reflected in the composition of CSF. We proposethat CSF analysis may become part of the diagnostic work-upof patients with parkinsonian syndromes. © 2003 MovementDisorder Society

Key words: multiple system atrophy; Parkinson’s disease;cerebrospinal fluid; neurotransmitters; brain-specific proteins

An important disorder to consider in the differentialdiagnosis of idiopathic Parkinson’s disease (PD) is mul-tiple system atrophy (MSA). MSA is characterized clin-ically by different combinations of parkinsonian, cere-bellar, pyramidal, and autonomic features.1,2 It is morerapidly progressive than PD with a mean survival be-tween 5 and 10 years3–7 and in contrast to PD where thesubstantia nigra is affected predominantly, MSA showsmuch more widespread neurodegeneration.3,5

The initial presentation and progression of signs andsymptoms of MSA varies widely. In some patients par-kinsonism develops at onset, and cerebellar and auto-nomic features develop later, whereas in others a differ-ent pattern is observed.8 MSA patients presenting withparkinsonian features may occasionally have a good ini-tial levodopa (L-dopa) response or asymmetric parkin-sonism, features that are both considered typical forPD.4,9 Therefore, it is often difficult to differentiate PDfrom MSA based on clinical grounds alone. Indeed, up to25% of patients with a parkinsonian syndrome may havebeen misdiagnosed during life.3,10–13

Reasonable diagnostic sensitivity and specificity (bothapproximately 90%) to distinguish MSA from PD could

be obtained by multivariate analysis of clinical symp-toms. These numbers are achieved by highly-specializedmovement disorder specialists, however, generally after3 to 5 years of clinical follow-up.9,14,15 Even then, thefalse-positive rate of the diagnosis of MSA can be ashigh as 14 to 20%,11,14 yet adequate prognosis, rationaltherapy, and future therapeutic trials for both PD andMSA require a stringent diagnostic accuracy in the ear-liest possible stages of the disease. As clinical criteriaalone seem insufficient to achieve such accuracy andtiming, a biochemical marker that aids in the discrimi-nation of MSA and PD would be useful.

We analyze the composition of cerebrospinal fluid(CSF) from PD and MSA patients to disclose potentialbiomarkers that might aid in a differential diagnosis.


Patients

The study included 65 patients referred to the Depart-ment of Neurology at the UMC Nijmegen who under-went lumbar puncture between January 1996 and June2001 as part of the analysis of an (at that time) unclearcause of a parkinsonian syndrome (Table 1). In May2002, the clinical diagnosis of all patients was reassessedby two independent neurologists who were not involvedin the previous process, in strict accordance with the UKParkinson’s Disease Society Brain Bank criteria forPD,13 and the criteria proposed by Gilman and col-leagues2 for probable or possible MSA. The clinicalfeatures required for applying the diagnostic criteriawere abstracted from clinical records. Diagnostic evalu-ation included a detailed history of disease, a neurolog-ical examination, and routine laboratory blood testing.Autonomic dysfunction was rated according to the crite-ria defined by Gilman and associates.2 Ancillary inves-tigations such as CSF analysis, 123IBZM-SPECT, com-puted tomography (CT) scan or anal sphincterelectromyogram (EMG) were not taken into account.

TABLE 1. Demographic characteristics by patient group

Characteristic MSA-C MSA-P PD P

n 14 15 35Age (yr)a 60.0 (54.5–65.5) 63.0 (58.0–65.0) 53.0 (47–61) 0.004b,c

Age at onset (yr)a 55.0 (51.0–60.5) 59.0 (55.5–62.5) 50.0 (41–56) 0.005b

Disease duration (mo)a 48.0 (25.5–60.5) 42.0 (34.5–68.5) 39.0 (15–57) 0.36Number of men (%) 12 (86) 10 (67) 21 (60) 0.22

Median (25th–75th percentile range) values are given.aAt the time of lumbar puncture.bMSA-C vs. PD; cMSA-P vs. PD; P � 0.05 (Dunn’s post-hoc correction for multiple comparisons).MSA-C, multiple system atrophy, predominant cerebellar features; MSA-P, multiple system atrophy, predominant parkin-

sonian features, PD, Parkinson’s disease.

572 W.F. ABDO ET AL.


MSA patients with predominant parkinsonian featureswere designated as having MSA-P, whereas those withpredominant cerebellar features were classified asMSA-C. Patients not fulfilling the clinical criteria for PDor MSA were excluded. Results of CSF analysis wereknown to the managing clinician. In case patients weretaking antiparkinsonian medication and a lumbar punc-ture was planned as part of the diagnostic work-up,medication was stopped at least 1 week before the lum-bar puncture and CSF neurotransmitter metabolite datawere included in the study. Neurotransmitter metabolitedata were excluded from the analysis; however, in casepatients were taking antiparkinsonian (MSA, n � 3; PD,n � 4) or antidepressive (MSA, n � 5; PD, n � 4)medication at the time of lumbar puncture.

For determination of reference values of the brain-specific proteins in CSF, we included all 62 patients aged45 to 75 years (median age, 52.5 years; range, 48.3–60.1years; 43% men) who had undergone lumbar puncture aspart of the diagnostic process but had been determined asnot suffering from a neurological disease. For determi-nation of reference values of the neurotransmitter metab-olites in the correct fraction of CSF (see below), weincluded 26 patients aged 45 to 75 years (median age,55.3 years; age range, 48.0–64.6 years; 38% men) whounderwent a lumbar puncture as part of the diagnosticprocess but who did not suffer a parkinsonian or adementia disorder, nor any other disorder expected toaffect neurotransmitter metabolism.

CSF Variables

The following CSF variables were measured: lactate,pyruvate, total protein, albumin ratio (CSF/serum), neu-ron-specific enolase (NSE), S-100B, glial fibrillaryacidic protein (GFAP), myelin basic protein (MBP), tau,homovanillic acid (HVA), 5-hydroxyindolacetic acid (5-HIAA), and 3-methoxy-4-hydroxyphenylethyleneglycol(MHPG). Lactate concentration was determined by theenzymatic conversion of nicotinamide adenine dinucle-otide (NAD) into NADH, measured at 340 nm, in thepresence of lactate dehydrogenase (LDH). Pyruvate con-centration was determined after fixation of CSF withcitric acid and baxilin at pH 4, in a reduction reactionwith 2,4-dinitrophenylhydrazine, which results in theformation of the brownish p-hydroxyphenylpyruvate.Absorption was measured at 483 nm. Total protein con-centration in CSF was determined with the Lowry reac-tion and absorption is measured at 720 nm. Lactate,pyruvate, and total protein were analyzed with an auto-mated analyzer (Mira Plus; ABX, Eindhoven, The Neth-erlands). Albumin concentration in both CSF and serum

is determined by nephelometry (Image; BeckmanCoulter, Mijdrecht, The Netherlands). Both NSE andS-100B concentrations in CSF are analyzed in an immu-noluminometric assay (Byk Sangtec, Dietzenbach, Ger-many) by using the Liaison automated analyzer (BykSangtec). The assays are linear up to 200 �g/L (NSE)and 30 �g/L (S-100b). The interassay variation coeffi-cients are �5.3 % (NSE) and �11% (S-100b). MBPconcentrations in CSF were analyzed by using a com-mercial ELISA (DSL, Webster, Texas; linearity up to 10�g/L; interassay variation coefficient �10%), CSFGFAP by using a homemade sandwich ELISA16,17 (lin-ear up to 250 �g/L; interassay variation coefficient�14%), and CSF tau by using the Innotest hTau assay(Innogenetics, Gent, Belgium; linearity up to 1,200 pg/L;interassay variation coefficient �6.0%).

HVA and 5-HIAA in CSF were measured according topreviously described methods.18 Because the concentra-tions of HVA and 5-HIAA vary in the different fractionsof CSF,18 we always used the 9th to the 11th ml of CSF.In brief, 1 mL CSF from this fraction was adjusted to pH2.5 with formic acid and applied to a Sephadex G-10(Sigma, St. Louis, MO) column. After successive wash-ing with formic acid and phosphate solution the metab-olites were eluted with an ammonia solution into a tubecontaining both formic acid and ascorbic acid. High-performance liquid chromatography (HPLC) was carriedout using a mobile phase of phosphate-solution, citricacid and methanol on a Hypersil ODS-1 column (Ther-moQuest, Breda, The Netherlands) with an amperomet-ric detection. MHPG was measured by using the samemethod with some minor modifications.18 The assayswere linear within the following ranges: HVA, 0 to 4�M; 5-HIAA, 0 to 2 �M; MHPG, 0 to 125 nM. Inter-assay variation coefficient was �4.8% in all three assays.

Statistical Analysis

The Kolmogorof test was used to analyze for normal-ity of the CSF variables; this test did not reject theassumption of a normal distribution, except for MBP.For statistical analysis of the data, we made a conserva-tive choice and consistently used nonparametric statisti-cal analyses. The Kruskall-Wallis test was used to testthe differences between MSA-P, MSA-C, and PD pa-tients for statistical significance in case of quantitativevariables and the �2 test was used in case of 3 � 2 tables.The Mann-Whitney test was used to test the differencesbetween two groups for statistical significance. Spear-man’s rank correlation test was used to test correlationsbetween quantitative variables. Univariate logistic re-gression was used to evaluate the prognostic ability of



the individual biochemical variables to discriminate be-tween PD and MSA. Because in particular the brain-specific proteins are expected to be non-linearly relatedin the logistic model, reasonable cut-off values for thesevariables to discriminate PD from MSA were con-structed. At this point the condition of equal “costs” ofmisclassification of patients and non-patients was usedby calculation of the maximal Youden index (� sensi-tivity � specificity � 1.0). In other words, the optimalcut-off value for a particular protein was chosen so thatthe sum of the sensitivity and the specificity to discrim-inate PD from MSA was maximal. Multivariate logisticregression with forward selection procedures was used toidentify variables that contributed independently to dis-criminate PD from MSA. Because forward selectionprocedures do not identify other important variables, Pvalues for entry into the model were considered to findclose alternatives to the variables selected. Adjusted oddsratios with 95% confidence intervals are presented.

RESULTS

Patients

Of 65 patients included, 26 had probable MSA, 4 hadpossible MSA, and 35 had PD. Among all 30 MSApatients, 15 had predominantly parkinsonian symptoms(MSA-P), 14 were categorized as MSA-C, and 1 hadboth prominent parkinsonian and cerebellar features.None of the patients showed clinical evidence of cogni-tive dysfunction. Both the mean age at onset of symp-toms and the age at time of the lumbar puncture weresignificantly higher in the two MSA subgroups comparedwith that in the PD group (Table 1). Disease duration andgender distribution were comparable in all three groups.

Results of CSF Analysis

Table 2 and Figure 1 show several differences in CSFvariables between the MSA-C and MSA-P subgroups andPD. The one MSA patient with both prominent parkinso-

TABLE 2. CSF variables by patient group

CSF variable

Median (n) and range

P*MSA-C MSA-P PD Control

Neurotransmitter metabolitesHVA (nM) 99.0 (11) 116 (9) 157 (29) 185 (26) 0.06

87.5–143 86.0–159 97–205 119–3155-HIAA (nM) 52.0 (11) 54.0 (9) 106 (29) 122 (26) 0.0001a,d

41.5–70.5 52.5–107 84–123 73–166MHPG (nM) 30 (11) 31.0 (9) 45 (29) 46 (23) 0.0005a,c

27.5–40.0 25.0–37.0 41–50 41–56Brain-specific proteins

NSE (�g/L) 7.3 (14) 8.7 (15) 5.6 (35) 7.9 (48) 0.008d

5.9–9.8 6.6–10.5 4.9–7.1 6.1–12.5MBP (�g/L) 0.8 (13) 1.0 (15) 0.4 (35) 0.5 (38) �0.0001a,c

0.6–1.3 0.7–1.4 0.3–0.6 0.2–0.8S-100B (�g/L) 2.5 (13) 2.5 (15) 2.1 (35) 2.5 (48) 0.062

1.9–3.6 2.0–3.2 1.5–2.6 2.1–3.1GFAP (�g/L) 2.0 (9) 1.3 (9) 1.1 (16) 0.9 (47) 0.61

0.6–3.7 0.8–2.1 0.9–2.1 0.5–1.9Tau (ng/L) 242 (10) 259 (11) 105 (30) 163 (36) �0.0001a,c

210–386 193–320 68–127 127–226Miscellaneous

Lactate (mM) 1.8 (14) 1.7 (14) 1.6 (35) 1.6 (62) 0.091.7–1.9 1.5–2.0 1.5–1.7 1.5–1.8

Pyruvate (�M) 125 (13) 108 (12) 109 (35) 114 (43) 0.07112–129 95–136 97–119 100–125

Q-albumin 11.0 (11) 8.2 (10) 5.2 (33) 5.0 (55) 0.004a

6.9–13.7 6.0–11.1 4.1–7.9 4.1–6.6

Median levels (25th–75th percentile range) are given.*P value for differences using the Kruskal-Wallis test on the MSA-C, MSA-P, and PD groups only. Dunn’s

post-hoc test for multiple comparisons was used to identify between-group differences.aP � 0.01, bP � 0.05 for MSA-C vs. PD; cP � 0.01, dP � 0.05 for MSA-P vs. PD.MSA-C, multiple system atrophy, predominant cerebellar features; MSA-P, multiple system atrophy, pre-

dominant parkinsonian features, PD, Parkinson’s disease; n, number of patients; HVA, homovanillic acid;5-HIAA, 5-hydroxyindolacetic acid; MHPG, 3-methoxy-4-hydroxyphenylethyleneglycol; NSE, neuron-specificenolase; MBP, myelin basic protein; GFAP, glial fibrillary acidic protein.



nian and cerebellar features was not included in this anal-ysis. The median levels of the neurotransmitter metabolites5-HIAA and MHPG were decreased significantly in bothMSA groups as compared with PD. A similar but non-significant trend was observed for HVA. CSF concentra-tions of the brain-specific proteins tau, MBP and NSE, butnot that of GFAP or S-100B, were elevated significantly inboth MSA groups compared with that in PD. Lactate andpyruvate concentrations did not differ between all threegroups. The ratio between albumin in CSF and serum(Q-albumin), an indicator of blood-CSF barrier dysfunc-tion, was significantly higher in the MSA-C group com-

pared with that in PD. We did not observe any differencesbetween the MSA-C and MSA-P groups. Because the me-dian age was different between the MSA and PD groups,we repeated the above analyses with age as a covariate andreached comparable conclusions.

A strong correlation in all samples was observed amongthe neurotransmitter metabolites HVA, 5-HIAA, andMHPG (r � 0.56–0.76; P � 0.001; Table 3). We observeda statistically significant correlation between all brain-spe-cific proteins (r � 0.23–0.55; P � 0.05). Furthermore, weobserved a statistically significant correlation between dis-ease duration and HVA (r � �0.36; P � 0.01).

FIG. 1. Scatterplots of the concentrations of HVA, 5-HIAA, MHPG (in nM), MBP (�g/L), and tau (ng/L) in CSF from MSA-C, MSA-P, and PDpatients. Horizontal lines represent median levels.

TABLE 3. Spearman correlation coefficient between cerebrospinal fluid variables and disease duration

CSFvariables

Correlation withdisease duration

CSF variables

HVA 5-HIAA MHPG NSE MBP S-100B

HVA �0.36 (0.01, 50) — — — — — —5-HIAA �0.21 (0.15, 50) 0.75 (�0.01, 50) — — — — —MHPG �0.08 (0.56, 50) 0.56 (�0.01, 50) 0.76 (�0.01, 50) — — — —NSE �0.12 (0.33, 65) �0.076 (0.60, 50) �0.19 (0.18, 50) �0.19 (0.18, 50) — — —MBP 0.003 (0.98, 64) �0.20 (0.16, 49) �0.42 (�0.01, 49) �0.45 (�0.01, 49) 0.55 (�0.01, 64) — —S-100B �0.09 (0.46, 64) �0.16 (0.28, 49) �0.32 (0.03, 49) �0.24 (0.10, 49) 0.36 (�0.01, 64) 0.23 (0.07, 63) —Tau 0.24 (0.08, 52) �0.29 (0.07, 42) �0.47 (�0.01, 42) �0.39 (0.01, 42) 0.37 (�0.01, 52) 0.30 (0.03, 52) 0.33 (0.02, 52)

Values in parentheses are (P, n).CSF, cerebrospinal fliud; HVA, homovanillic acid; 5-HIAA, 5-hydroxyindolacetic acid; MHPG, 3-methoxy-4-hydroxyphenylethyleneglycol; NSE, neuron-specific

enolase; MBP, myelin basic protein; GFAP, glial fibrillary acidic protein.



Because no differences were observed between theMSA-C and MSA-P groups, we carried out a univariatelogistic regression analysis to discriminate MSA fromPD. We identified cut-off values of those variables thatwere significantly different (P � 0.05) between the PDand combined MSA groups. Cut-off values were definedwhen an optimal combination of diagnostic sensitivityand specificity for the differentiation between MSA andPD was reached (Table 4). This analysis revealed that,within the entire group of CSF variables, tau offered thebest combination of sensitivity (95%) and specificity(77%) for the detection of MSA. Other CSF variablesthat discriminated PD from MSA with reasonable sensi-tivity (�62%) and specificity (�72%) were 5-HIAA,MHPG, and MBP.

Univariate analysis did not identify a single CSF variablethat discriminated MSA from PD with both sensitivity andspecificity exceeding 85%, criteria usually defined for ap-plication in clinical practice. Therefore, a multivariate lo-gistic regression model was used for a better discriminationbetween MSA and PD. Because our data sets had missingvalues, we added the CSF variables groupwise to the modelbased on the correlations described above. First, a groupconsisting of brain-specific proteins (tau/NSE/MBP/S-100B) was used in the selection procedure. In the secondstep, the neurotransmitter metabolites (HVA/5-HIAA/MHPG) were entered. Not surprisingly, in the first selectionstep, tau was entered into the model (Table 5) to discrimi-nate PD from MSA; however, tau was the only brain-specific protein that was selected. From the second group ofvariables, only MHPG was selected for addition into themultivariate analysis. Although MBP discriminated PD

from MSA, this was a nonsignificant addition to the model.Furthermore, NSE did not provide additional information tothis discrimination model, presumably because it is corre-lated with tau. In an alternative model, we omitted tau fromthe first variable group to estimate the contribution of MBP.This approach resulted in inclusion of MBP and MHPG foroptimal discrimination of PD from MSA.

Figure 2 shows the occurrence of autonomic dysfunctionin MSA and PD groups and the possible relation with

TABLE 4. Cut-off values of CSF variables for optimal combination ofdiagnostic sensitivity and specificity for differentiation between MSA and PD

CSF variablesCut-offvalue Sensitivity Specificity

Youdenindex*

HVA (nM)a 116 0.62 0.72 0.345-HIAA (nM)a 62 0.71 0.90 0.61MHPG (nM)a 34 0.71 0.90 0.61NSE (�g/L)b 8.4 0.47 0.91 0.38MBP (�g/L)b 0.7 0.72 0.86 0.58S-100B (�g/L)b 1.9 0.79 0.45 0.25Tau (ng/L)b 128 0.95 0.77 0.72Q-albuminb 5.8 0.90 0.58 0.48

Optimal cut-off values of CSF variables to discriminate between MSA and PD usingunivariate logistic regression with ROC analysis.

*Youden index � sensitivity � specificity �1.00.aSensitivity calculated for a value below the cut-off point.bSensitivity calculated for a value above the cut-off point.CSF, cerebrospinal fluid; MSA, multiple system atrophy; PD, Parkinson’s disease;

HVA, homovanillic acid; 5-HIAA, 5-hydroxyindolacetic acid; MHPG, 3-methoxy-4-hydroxyphenylethyleneglycol; NSE, neuron-specific enolase; MBP, myelin basic protein.

TABLE 5. Adjusted odds ratio of the CSF variables todiscriminate between MSA and PD using multivariate

logistic regression with selection procedures

Variable

Adjusted odds radio*

Model 1** Model 2***

Tau 27.2 (3.7–570.7) —MBP — 43.51 (3.77–999.00)MHPG 0.14 (0.01–0.96) 0.023 (0.001–0.201)r2 0.64 0.74n 42 47

*Odds ratio adjusted for all other variables in the model.**Model 1. In the first step, a group consisting of brain-specific

proteins (tau/NSE/MBP/S-100B/A�42) was entered in the multivariateanalysis. In the second selection step the neurotransmitter metabolites(HVA/5-HIAA/MHPG) were entered. This approach resulted in theinclusion of tau and MHPG for the optimal discrimination of PD fromMSA.

***Model 2. Tau was omitted from the first variable group toestimate the contribution of MBP in the first selection step. In thesecond selection step, the neurotransmitter metabolites (HVA/5-HIAA/MHPG) were entered. This approach resulted in the inclusion of MBPand MHPG for the optimal discrimination of PD from MSA. CSF,cerebrospinal fluid; MSA, multiple system atrophy; PD, Parkinson’sdisease; MBP, myelin basic protein; MHPG, 3-methoxy-4-hydroxy-phenylethyleneglycol; NSE, neuron-specific enolase; HVA, ho-movanillic acid; 5-HIAA, 5-hydroxyindolacetic acid.



neurotransmitter metabolite levels. Both MHPG and5-HIAA levels were significantly lower in MSA patientswith autonomic dysfunction compared with MSA patientswithout signs of autonomic dysfunction. There was nodifference in the concentrations of HVA in both the MSAand PD groups, and of 5-HIAA or MHPG in the PD groups.

DISCUSSIONOur biochemical analysis of CSF from patients with

PD and MSA may have both biological and diagnosticrelevance. First, the biochemical composition of CSF ofMSA-C and MSA-P seemed similar. Second, the con-centrations of three neurotransmitter metabolites (HVA,5-HIAA, and MHPG) were lower in CSF from both

MSA-P and MSA-C patients than in that from PD pa-tients. Third, the concentrations of some brain-specificproteins (tau, NSE, and MBP) were elevated in CSFfrom MSA patients compared with that from PD patients.Finally, a multivariate logistic regression analysis re-vealed that the combination of tau and MHPG may havepotential in discriminating MSA from PD.

In both PD and MSA-C/P, the CSF concentration ofHVA is reduced compared with a neurological controlgroup, confirming previous studies19–22 and supporting cur-rent practice of trial of L-dopa in MSA-P. HVA concentra-tions, however, did not differentiate between MSA and PD.Furthermore, our data of decreased CSF MHPG levels in

FIG. 2. Scatterplots of the concentrations of HVA, 5-HIAA, and MHPG (in nM) in CSF from MSA and PD patients, either with (“yes”) or without(“no”) features of autonomic dysfunction. Horizontal lines represent median levels. A significant decrease was observed for the levels of 5-HIAA (P �0.018) and MHPG (P � 0.0078) in the MSA patients with autonomic dysfunction as compared with MSA patients without autonomic dysfunction(Mann-Whitney test). The differences were not significant in HVA concentration between the MSA and PD groups as well as in 5-HIAA or MHPGin the PD groups.



MSA, particularly in those patients with autonomic failure,are in line with other reports.22,23 In PD patients, MHPGlevels have been reported variably as either normal ordecreased.19,20 Finally, our data are in line with observa-tions that CSF concentrations of 5-HIAA are reduced inMSA21,22 but not in PD.19 In addition, we observed that5-HIAA levels are affected particularly by the occurrenceof autonomic dysfunction in MSA patients. Because diseaseduration is comparable, our data suggest that neuronal dam-age is more progressive and widespread in MSA than inPD, and support earlier findings that both MHPG and5-HIAA levels may be lower in MSA patients with auto-nomic dysfunction as compared with patients without suchfeatures.

Our observation that the CSF concentrations of tau, NSE,and MBP were elevated in MSA compared with PD alsoimplicate a more widespread neural degeneration in MSAand may reflect the more progressive nature of this disease.These variables may reflect ongoing neuronal (tau andNSE) damage or demyelination (MBP), and myelin degra-dation has indeed been demonstrated in MSA brains.24

Because disease duration did not correlate with any CSFvariable except for HVA, this suggests that the CSF com-position had stabilized during the first few years of thedisease or, alternatively, disease duration may have beentoo short to observe significant differences.

The relatively high concentrations of tau in CSF fromMSA patients compared with PD patients may seemsurprising. Both MSA and PD do not belong to the groupof so-called “tauopathies”,25 wherein tau accumulates asintraneuronal filamentous aggregates which, in turn, arereflected by increased CSF concentrations of tau. Otherstudies reported tau concentrations for PD patients sim-ilar to those that we observed.26–29 We are the first toreport CSF tau levels in MSA. Because tau is a micro-tubule-binding protein with a function in stabilizing mi-crotubules that are essential for axonal transport or cy-toplasmic organelles in neurons, the relatively high CSFconcentration of tau in MSA may reflect more wide-spread axonal degeneration relative to that seen in PD.Alternatively, the CSF tau levels in PD may also beinterpreted as relatively low, because it was shown inanother control population aged 51 to 70 years that taulevels may reach up to 450 ng/L.30 The cause for thispossible decrease remains unclear.

Holmberg and colleagues31 described a significant de-crease in the concentration of the amyloid �42 protein(A�42) in CSF from MSA patients as compared with thatfrom PD patients. Analysis of CSF A�42 levels in ourpatient groups could not confirm this decrease in MSA.32

In another study, an increase in CSF neurofilament light

chain (NFL) in MSA compared with that in PD wasdescribed.33 Like tau, NFL is an axonal protein; there-fore, the relative high tau levels in MSA patients areconsistent with these data.

To our knowledge, this is the first systematic multicom-ponent analysis of CSF from MSA and PD patients. Froma clinical perspective, our data suggest that a combinationof CSF variables, especially tau, MHPG, and possiblyMBP, may discriminate MSA from PD at a time of ongoingdiagnostic uncertainty. In clinical practice, application ofCSF analysis might therefore help to separate PD fromMSA-P, possibly as part of a more extensive test battery(e.g., SPECT, magnetic resonance imaging [MRI], andsphincter EMG). Because no significant difference betweenthe MSA-P and MSA-C groups was observed, the aboveanalysis is not only restricted to a comparison between theentire MSA and PD groups, but may also apply to thecomparison of the MSA-P subgroup with PD.

The retrospective nature of our study has some draw-backs. First, the data set contained some missing data,resulting in a suboptimal multivariate analysis. There-fore, sensitivity and specificity for the multivariate mod-els were not calculated. Second, bias in the inclusion ofpatients may have occurred, as only patients with anunclear initial clinical presentation were included. Pre-sumably, this promoted selection of patient groups withan atypical initial presentation of PD as is reflected by,for example, the relatively young age of the PD patients.The final clinical diagnosis, however, was made afterextended follow-up of the patients in strict accordancewith accepted clinical criteria for both MSA and PD.2,13

Therefore, this study design closely reflects the actualclinical situation where patients present to the clinicianwith an unclear parkinsonian syndrome that often re-quires prolonged follow-up before a final diagnosis canbe reached. Our study suggests that analysis of CSF, at atime when clinical examinations remain inconclusive,may help to establish a clinical diagnosis of either MSAor PD. Prospective follow-up studies and independentstudies are needed to confirm our results.

Acknowledgments: This work was supported by grantsfrom Hersenstichting Nederland (to M.M.V.) and the Interna-tional Parkinson Foundation (to B.R.B.). We thank Dr. J.Lenders for analysis of autonomic dysfunction, the techniciansfrom the Department of Clinical Chemistry for albumin anal-ysis, and technicians from the Laboratory of Pediatrics andNeurology for all other CSF analyses.

REFERENCES

1. Quinn N. Multiple system atrophy—the nature of the beast. J Neu-rol Neurosurg Psychiatry 1989;(Suppl.):78–89.



2. Gilman S, Low PA, Quinn N, et al. Consensus statement on thediagnosis of multiple system atrophy. J Neurol Sci 1999;163:94–98.

3. Wenning GK, Ben-Shlomo Y, Magalhaes M, Daniel SE, QuinnNP. Clinicopathological study of 35 cases of multiple systematrophy. J Neurol Neurosurg Psychiatry 1995;58:160–166.

4. Wenning GK, Ben Shlomo Y, Magalhaes M, Daniel SE, QuinnNP. Clinical features and natural history of multiple system atro-phy. An analysis of 100 cases. Brain 1994;117:835–845.

5. Wenning GK, Tison F, Ben Shlomo Y, Daniel SE, Quinn NP.Multiple system atrophy: a review of 203 pathologically provencases. Mov Disord 1997;12:133–147.

6. Fearnley JM, Lees AJ. Striatonigral degeneration. A clinicopath-ological study. Brain 1990;113:1823–1842.

7. Gouider-Khouja N, Vidailhet M, Bonnet AM, Pichon J, Agid Y.“Pure” striatonigral degeneration and Parkinson’s disease: a com-parative clinical study. Mov Disord 1995;10:288–294.

8. Gilman S, Quinn NP. The relationship of multiple system atrophy tosporadic olivopontocerebellar atrophy and other forms of idiopathiclate-onset cerebellar atrophy. Neurology 1996;46:1197–1199.

9. Wenning GK, Ben-Shlomo Y, Hughes A, Daniel SE, Lees A,Quinn NP. What clinical features are most useful to distinguishdefinite multiple system atrophy from Parkinson’s disease? J Neu-rol Neurosurg Psychiatry 2000;68:434–440.

10. Hughes AJ, Daniel SE, Lees AJ. Improved accuracy of clinicaldiagnosis of Lewy body Parkinson’s disease. Neurology 2001;57:1497–1499.

11. Litvan I, Goetz CG, Jankovic J, et al. What is the accuracy of theclinical diagnosis of multiple system atrophy? A clinicopathologicstudy. Arch Neurol 1997;54:937–944.

12. Rajput AH, Rozdilsky B, Rajput A. Accuracy of clinical diagnosisin parkinsonism—a prospective study. Can J Neurol Sci 1991;18:275–278.

13. Hughes AJ, Daniel SE, Kilford L, Lees AJ. Accuracy of clinicaldiagnosis of idiopathic Parkinson’s disease: a clinico-pathologicalstudy of 100 cases. J Neurol Neurosurg Psychiatry 1992;55:181–184.

14. Hughes AJ, Daniel SE, Ben-Shlomo Y, Lees AJ. The accuracy ofdiagnosis of parkinsonian syndromes in a specialist movementdisorder service. Brain 2002;125:861–870.

15. Hughes AJ, Ben-Shlomo Y, Daniel SE, Lees AJ. What featuresimprove the accuracy of clinical diagnosis in Parkinson’s disease:a clinicopathologic study. Neurology 1992;42:1142–1146.

16. van Geel WJA, de Reus HPM, Nijzing H, Verbeek MM, Vos PE,Lamers KJ. Measurement of glial fibrillary acidic protein in blood:an analytical method. Clin Chim Acta 2002;326:151–154.

17. Herrmann M, Vos P, Wunderlich MT, de-Bruijn CH, Lamers KJ.Release of glial tissue-specific proteins after acute stroke: a com-parative analysis of serum concentrations of protein S-100B andglial fibrillary acidic protein. Stroke 2000;31:2670–2677.

18. Brautigam C, Wevers RA, Jansen RJ, Smeitink JA, de Rijk-vanAndel JF, Gabreels FJ, Hoffmann GF. Biochemical hallmarks oftyrosine hydroxylase deficiency. Clin Chem 1998;44:1897–1904.

19. Chia LG, Cheng FC, Kuo JS. Monoamines and their metabolites inplasma and lumbar cerebrospinal fluid of Chinese patients withParkinson’s disease. J Neurol Sci 1993;116:125–134.

20. Mann JJ, Stanley M, Kaplan RD, Sweeney J, Neophytides A.Central catecholamine metabolism in vivo and the cognitive andmotor deficits in Parkinson’s disease. J Neurol Neurosurg Psychi-atry 1983;46:905–910.

21. Polinsky RJ, Brown RT, Burns RS, Harvey-White J, Kopin IJ.Low lumbar CSF levels of homovanillic acid and 5-hydroxyin-doleacetic acid in multiple system atrophy with autonomic failure.J Neurol Neurosurg Psychiatry 1988;51:914–919.

22. Mathias CJ, Williams AC. The Shy-Drager syndrome (and multi-ple system atrophy). In: Calne DB, editor. Neurodegenerativediseases. Philadelphia: WB Saunders; 1994. p 743–767.

23. Polinsky RJ, Jimerson DC, Kopin IJ. Chronic autonomic failure:CSF and plasma 3-methoxy-4-hydroxyphenylglycol. Neurology1984;34:979–983.

24. Matsuo A, Akiguchi I, Lee GC, McGeer EG, McGeer PL, KimuraJ. Myelin degeneration in multiple system atrophy detected byunique antibodies. Am J Pathol 1998;153:735–744.

25. Goedert M, Spillantini MG, Serpell LC, Berriman J, Smith MJ,Jakes R, Crowther RA. From genetics to pathology: tau andalpha-synuclein assemblies in neurodegenerative diseases. PhilosTrans R Soc Lond B Biol Sci 2001;356:213–227.

26. Sjogren M, Davidsson P, Tullberg M, et al. Both total and phos-phorylated tau are increased in Alzheimer’s disease. J NeurolNeurosurg Psychiatry 2001;70:624–630.

27. Vanmechelen E, Vanderstichele H, Davidsson P, et al. Quantifi-cation of tau phosphorylated at threonine 181 in human cerebro-spinal fluid: a sandwich ELISA with a synthetic phosphopeptidefor standardization. Neurosci Lett 2000;285:49–52.

28. Molina JA, Benito-Leon J, Jimenez-Jimenez FJ, et al. Tau proteinconcentrations in cerebrospinal fluid of non-demented Parkinson’sdisease patients. Neurosci Lett 1997;238:139–141.

29. Verbeek MM, de Jong D, Kremer HPH. Brain-specific proteins incerebrospinal fluid for the diagnosis of neurodegenerative diseases.Ann Clin Biochem 2003;40:25–40.

30. Sjogren M, Vanderstichele H, Agren H, et al. Tau and Abeta42 incerebrospinal fluid from healthy adults 21–93 years of age: estab-lishment of reference values. Clin Chem 2001;47:1776–1781.

31. Holmberg B, Johnels B, Blennow K, Rosengren L. Cerebrospinalfluid Abeta42 is reduced in multiple system atrophy but normal inParkinson’s disease and progressive supranuclear palsy. Mov Dis-ord 2003;18:186–190.

32. Verbeek MM, Abdo WF, de Jong D, Horstink MWIM, KremerHPH, Bloem BR. Cerebrospinal fluid A�42 levels in multiplesystem atrophy. Mov Disord 2004;DOI 10.1002/mds.10687.

33. Holmberg B, Rosengren L, Karlsson JE, Johnels B. Increasedcerebrospinal fluid levels of neurofilament protein in progressivesupranuclear palsy and multiple-system atrophy compared withParkinson’s disease. Mov Disord 1998;13:70–77.



Documents

Non-subtype-selective opioid receptor antagonism in treatment of levodopa-induced motor complications in Parkinson's disease