grant/research support from: American Diagnostica, Consultant for:Genomic Health (advisory board).

Keyword: Breast cancer

doi:10.1016/j.jgo.2014.09.165

Track 1 - Solid Tumours in the ElderlyBreast cancer in elderly patients

O16UNFAVOURABLE PREDICTIVE FACTORS IN OLDER WOMEN WITHCLINICALLY FAVOURABLE BREAST CANCERM. Kiderlen1,*, L.J. de Gruiter2, E. Bastiaannet1, A. Witteveen3,A.J. de Craen2, C.J. van de Velde2, G.-J. Liefers2, S. Siesling31Surgical Oncology/Geriatrics & Gerontology, Netherlands2Leiden University Medical Center, Leiden, Netherlands3University of Twente, Enschede, Netherlands

Introduction: Forty percent of all breast cancer cases occuramong older women (age 65 years and older). This patient group isunderrepresented in the currently available studies, which obligesphysicians to take treatment decisions based on guidelines andmodels that are not validated for older patients. With this study weaim to provide insight in predictive factors for recurrence in clinicallyfavourable breast cancer, which can aid in individualizing treatmentfor older breast cancer patients.

Objectives: To establish a predictive model for recurrence inolder breast cancer patients with a small breast tumour and aclinically negative axilla.

Methods: From the Netherlands Cancer Registry, all patients aged65 years and older with non-metastatic breast cancer, diagnosedbetween 2003 and 2006, with a tumour with a maximum size of5 cm, a clinical stage N0, and primary breast surgery, were included inthe analyses. First, we defined a set with most important predictors forrecurrence based on literature and expert-knowledge, comprising age,T-stage (T1 or T2), histological grade (1 to 3), morphology (ductal,lobular, mixed or other), oestrogen receptor (ER), progesteronereceptor (PR), Her2Neu receptor and multifocality. Primary endpointwas 5-year recurrence, a combinedmeasure of locoregional and distantrecurrence. To take account for the competing risk of mortality, thepredictive value on the primary endpoint was analysed with Fine &Gray analyses. Using backward elimination, predictors without statis-tically significant predictive value were eliminated stepwise from themodel, until a remaining model was constituted with predictors thathad a statistically significant predictive value. All models were analysedon discrimination by calculating the Area Under the Curve (AUC) of theROC-curve. Finally, all analyses were repeated in two age strata (65–74and ≥75 years), to find out if predictors are the same for each agecategory. Missing values were filled in using multiple imputation.

Results: Overall, 9183 patients were included in this study. Afterbackward eliminationwe remainedwith a predictivemodel comprisingfour variables: age, T-stage, grade and ER. This model was able topredict 5-years recurrencewith amean AUC of 0.69 (internal validationafter bootstrapping with 1000 replications). This AUC was the same asfor the full model. Stratification on age groups yielded the samevariables in the model for both strata.

Conclusion: In this national population-based study among olderbreast cancer patients, we created a predictive model in which age,T-stage, grade and ER can accurately predict 5-years recurrence risk.Interestingly, the potential predictors morphology, Her2, multi-focality and PR could be excluded from the model without losingpredictive value. This results in a model in which a physician can

predict the risk of recurrence based on a pre-operative biopsy, andtake treatment decisions on that basis. Future research shoulddetermine the external validity of this prediction model, also takingother factors such as comorbidity into account.

Disclosure of interest: None declared.

Keywords: Breast cancer, Epidemiology

doi:10.1016/j.jgo.2014.09.166

Track 1 - Solid Tumours in the ElderlyBreast cancer in elderly patients

O17BREAST CANCER SCREENING IN OLDER WOMENN. De Glas1,*, A.J.M. de Craen2, E. Bastiaannet1, E. Op 't Land1, M.Kiderlen1, W. van de Water1, S. Siesling3, J.E.A. Portielje4, H.M.Schuttevaer5, G.H. de Bock6, C.J.H. van de Velde1, G.-J. Liefers11Surgery, Leiden University Medical Center, Leiden, Netherlands2Gerontology & Geriatrics, Leiden University Medical Center, Leiden,Netherlands3Research, Comprehensive Cancer Center the Netherlands, Utrecht,Netherlands4Medical Oncology, Haga Hospital, Den Haag, Netherlands5Radiology, RIjnland hospital, Leiderdorp, Netherlands6Epidemiology, University of Groningen, Groningen, Netherlands

Introduction: In upcoming decades, an increasing proportion ofbreast cancer patients will be elderly. It has been assumed thatdiagnosis at an earlier stage through screening programs could improveprognosis. However, elderly may be at risk for over diagnosis due toscreening programs, and consequently unnecessarily at risk for possibleharmful effects of cancer treatment. In The Netherlands, the upper agelimit of the screening program was extended from 69 to 75 years in1998. However, it remains unclear whether the mass screeningprogram has a beneficial effect in women aged 70 years and older.

Objectives: If a screening program is effective, it can be expectedthat the incidence of early stage breast cancer increases, while theincidence of advanced stage cancers decreases. According to severalstudies, this is the most appropriate method to investigate the efficacyof a screening program in population-based data, as studying mortalityrates as an indicator for the effect of screening programs can lead toseveral forms of bias. Therefore, we investigated the effect of theimplementation of the screening program on the stage distribution ofincident breast cancer in women aged 70–75 years in the Netherlands.

Methods: The Netherlands Cancer Registry was used to includeall patients aged 70–75 years who were diagnosed between 1995and 2011 with invasive or in situ breast cancer. Time trends ofincidence rates of different tumor stages were analyzed in linearregression analyses with the incidence rate of both early stage (0, Iand II) and advanced stage (III and IV) breast cancer as the outcome,and year of diagnosis as the independent variable.

Results: Overall, we included 25,414 patients aged 70–75 years atdiagnosis. The incidence of early stage tumors significantly increasedafter extension of the upper age limit to 75 years in 1998 (260 cases per100,000women in 1995 up to 382 cases per 100,000women in 2011, pfor trend= 0.03), while the number of advanced stage breast cancersdid not significantly change (59 cases per 100,000women in 1995 to 53cases per 100,000 women in 2011, p for trend= 0.2).

Conclusion: The extension of the upper age limit to 75 years has notled to a decrease of advanced stage breast cancer, while the number ofearly stage tumors strongly increased. This implies that the effect ofscreening in elderly women is limited and leads to a large proportion of

AbstractsS20

overdiagnosis. Until new studies in this specific group have beenperformed, we propose that the decision to participate in the screeningprogram should be personalized based on remaining life expectancy,functional status and patients' preferences.

Disclosure of interest: None declared.

Keywords: Breast cancer, Epidemiology

doi:10.1016/j.jgo.2014.09.167

Track 4 - Nursing, Supportive Care and Geriatric AssessmentNursing

O18A MULTIDISCIPLINARY EDUCATIONAL COURSE FOR ONCOLOGYNURSES TO ADVANCE EXPERTISE IN THE CARE OF OLDERPATIENTS WITH CANCERH. Yulico1, B. Korc-Grodzicki1,*, L. McEvoy21Geriatrics2Nursing,Memorial Sloan Kettering Cancer Center, NewYork, United States

Introduction: Older cancer patients have a higher prevalence ofchronic health conditions that can have negative effects on functionalstatus and quality of life. Effects of multiple comorbidities in conjunctionwith a diagnosis of cancerwill require increased surveillance and specificsafety protocols to prevent iatrogenic events. As care in cancer centersshifts out of the hospital and into outpatient settings, all nurses andsupport staff (inpatient and outpatient) must be adequately educatedand trained to meet the specialized needs of the geriatric oncologypatient across the continuum of care.

Objectives: Educate nurses on geriatric syndromes and increasecompetency in caring for the older adult with cancer across thecontinuum of care.

Methods: A multidisciplinary task force was assembled to create atwo day course that is offered twice a year to all nurses at our center andto nurses from outside facilities. It covers topics on: age relatedphysiologic changes, carcinogenesis, advanced care planning, geriatricpharmacology, cancer and nutrition, depression and anxiety, painmanagement and palliative care, and physical and occupationaltherapy. Faculty includes clinicians from: Geriatrics, Nursing, Rehabil-itation, Pharmacy, Pain and Palliative Care, and Psychiatry. Data wascollected from participants: pre and post-course assessment of geriatriconcology knowledge as well as overall educational experience.

Results: Over the past four years, 284 nurses completed this course.A 10 item questionnaire was used to evaluate knowledge gained. Testscores recorded for 235 participants improved from 64% (95% CI 62.28–66.18) to 80% (95% CI 79.01–81.83)with the p value of b0.0001. This is astatistically significant improvement in knowledge. 100%of participantsrate the course as very good or excellent on a 5 item Likert-scale inmeeting the overall purpose and achieving the objectives of the course.

Conclusion: Our center has successfully created a multidisciplin-ary course to educate all oncology nurses about care of the oldercancer patient. This course can serve as a model for other institutionslooking to implement educational programs on nursing care of thisrapidly increasing population.

Disclosure of interest: None declared.

Keyword: Nursing

doi:10.1016/j.jgo.2014.09.168

Track 4 - Nursing, Supportive Care and Geriatric AssessmentGeriatric Assessment

O19EVALUATION OF A PHARMACIST-LED MEDICATION ASSESSMENTUSED TO IDENTIFY PREVALENCE AND ASSOCIATIONS WITHPOTENTIALLY INAPPROPRIATE MEDICATION (PIM) USE AMONGAMBULATORY SENIOR ADULTS WITH CANCERG. Nightingale1,*, E. Hajjar1, J. Andrel-Sendecki2, K. Swartz3, A. Chapman41Department of Pharmacy Practice, Jefferson School of Pharmacy, ThomasJefferson University, United States2Department of Pharmacology and Experimental Therapeutics, Division ofBiostatistics, Thomas Jefferson University, United States3Department of Family and Community Medicine, Division of Geriatrics andPalliative Medicine, United States4Department of Medical Oncology, Thomas Jefferson University Hospital,Philadelphia, United States

Introduction: Potentially inappropriate medication (PIM) use inseniors is a significant public health problem linked to billions inhealth expenditures; cancer-related therapies further escalate itsprevalence and complexity. Existing studies are limited by antiquat-ed criteria and inherent pitfalls of patient self-report/chart extraction(versus a pharmacist-led assessment).

Objectives: Objectives: 1) Identify PIM associations with patientcharacteristics; 2) Determine the most inclusive PIM criteria.

Methods: This retrospective study involved 248 patients(aged≥ 60 years) referred for a geriatric oncology multidisciplinaryassessment in January 2011 through June 2013 (data from physician/pharmacist electronic notes). Patients brought medications (prescrip-tion, non-prescription, herbals) for the pharmacist-led assessment. PIMuse was determined by 2012 Beers, the Screening tool of older persons'potentially inappropriate prescriptions (STOPP) and the Healthcare anddata information set (HEDIS) criteria. PIM associations with age, gender,polypharmacy, comorbidities, ECOG and functional statuswere analyzed.

Results: Mean age was 79.9 years [range 61–98]; 64% women, 74%Caucasian, 87% solid tumor,mean comorbidities, 7.69; only 234patients(evaluated by pharmacists) were included in the final analysis. PIMprevalence was 51% (n= 119). Beers, STOPP and HEDIS classified 173PIMs equivalent to 8% of total medications (n= 2163). Mean PIM usewas 0.74 (SD 0.89), range [(0–4)]. The proportion of PIMs used (0, 1, 2,3, 4) was 49%, 34%, 12%, 4% and 1%, respectively. Associations with PIMuse (versus no-PIM) were polypharmacy (P b 0.001) and increasedcomorbidities (P = 0.005), specifically cardiovascular (P = 0.014),gastrointestinal (P= 0.013), neurologic (P = 0.020), and psychiatric(P b 0.001) conditions. STOPP and BEERsweremost inclusive capturing119 (69%) and 118 (68%) PIMs versus HEDIS [58 (34%)]. Beers andSTOPP mutually identified 67 (39%) PIMs.

Conclusion: A pharmacist-led medication assessment utilizingBeers and STOPP identified a high prevalence of PIM use andassociations with increased comorbidity count. A modified PIM toolthat integrates Beers and STOPP and considers cancer diagnosis,prognosis and cancer-related therapy is needed to identify andprevent PIM use in this vulnerable population.

References: The NCCN Clinical Practice Guidelines in OncologySenior Adult Oncology (version 2.2014).

American Geriatrics Society Updated Beers Criteria for PotentiallyInappropriate Medication Use in Older Adults. The American GeriatricsSociety 2012 Beers Criteria Update Expert Panel. AGS updated BeersCriteria for potentially inappropriate medication use in older adults. JAm Geriatr Soc 2012; doi: 10.1111/j.1532-5415.2012.03923.x.

O'Mahony D, Gallagher P, Ryan C, Byrne S, Hamilton H et al.STOPP & START criteria: A new approach to detecting potentially

Abstracts S21