1597 Prognosis of leptomeningeal metastasis diagnosed by magnetic resonance imaging in 329 patients with advanced solid tumors

S236 1594 POSTER Adherence to oral antineoplastic agents: Evaluation of a multi-centre, multidisciplinary training program for doctors and nurses E. Rieder1 , C. Rothermundt2 , A. Margulies3 , I. Bachmann-Mettler4 , 6 1 M. Schmid5 , M. Hafner ¨ . ZHAW, School of Health Profession, Winterthur, Switzerland; 2 Kantonspital, Division of Oncology/Haematology-, St. Gallen, Switzerland; 3 Clinical Oncology Nurse, Zurich, ¨ Zurich, ¨ Switzerland; 4 Swiss Society of Oncology Nursing, Zurich, ¨ Zurich, ¨ 5 Switzerland; Evaluation and Health Research, Glarus, Glarus, Switzerland; 6 Spital Bulach, ¨ Onkologie Bulach, ¨ Bulach, ¨ Switzerland Background: With the steadily increasing number and availability of oral tumour therapies the topic of drug adherence takes on an ever more important role in oncology regimens. Adherence to the therapy is important to maximize treatment efficacy and minimize adverse effects. Patients need correct, understandable information about the drug, the administration, as well as the most frequent, manageable side effects. Learning to recognize early symptoms of side effects, and taking correct measures could lead to greater safety and better adherence. There is growing evidence that the patient needs professional support and assistance. A multidisciplinary group organised by the SGMO (Swiss Society for Medical Oncology) and OPS (Swiss Society for Oncology Nursing), developed a healthcare professional (HCP) training program for safety and adherence with oral tumour therapy in the German-speaking part of Switzerland. The aim of this program is to provide on-site knowledge and skills for doctors and nurses to meet the increasing information and support needs of tumour patients on oral cancer drugs. The training consists of three modules: • Adherence and Self-Management • Cancer Drugs in Practice • Communication and Counselling re. Adherence Methods: To check feasibility, acceptance and use of the program an external evaluation was carried out. The professionals were consulted by a written questionnaire about the training program and asked to give a self-assessment of their skills after a year of implementation. After termination of the consultation patients received a written questionnaire and satisfaction of support and information given was assessed. The evaluation was anonymous. Preliminary results: Eight oncology centres with 82 participants were included in this project. All centres are currently implementing consulting model in the clinical setting. The preliminary analysis show a positive feedback from the training sessions. The majority of the HCP were “satisfied” to “very satisfied” with the course. The evaluation regarding “useful for consultation” was “positive” to “very positive” (agreement: 60%-90% for each item). The patients (n = 48) evaluated the information as “positive” to “very positive” (agreement: 80−90% for each item) and appreciated the quality of consultations by doctors and nurses as “good” to “very good”. The results also underlined the need to include more disease and therapy related psychosocial aspects. Conclusion: The results from the 8 centres demonstrate the need for professional training as well as help to implement the gained knowledge. The HCP’s are altogether satisfied with the training and evaluated the use for the patients as positive. Self-assessment of the newly acquired skills has improved after the first year of imple-mentation. All patient participants benefit from the structured consultation and are satisfied with the quality. No conflict of interest. 1596 POSTER Predictive modeling of the outcomes of chemotherapy-induced (febrile) neutropenia (CIN/FN) prophylaxis with biosimilar filgrastim and development of an outcomes risk calculator (MONITOR-GCSF study) ´ 4 , M. Boccadoro5 , M. Aapro1 , H. Ludwig2 , C. Bokemeyer3 , P. Gascon M. Turner6 , K. Denhaerynck7 , K. MacDonald7 , I. Abraham7,8 . 1 Clinique de Genolier, Institut Multidisciplinaire d’Oncologie, Genolier, Switzerland; 2 Wilhelminenspital, Medizinische Abteilung I − Onkologie und Haematologie, Wien, Austria; 3 Universitaetsklinikum Hamburg Eppendorf, Department of Internal Medicine II, Hamburg, Germany; 4 Hospital Cl´ınic de Barcelona, University of Barcelona, Department of Hematology-Oncology, Barcelona, Spain; 5 Azienda Ospedaliero Universitaria S. Giovanni Battista di Torino, Dipartimento di Oncologia e Ematologia, Torino, Italy; 6 Sandoz Biopharmaceuticals, Holzkirchen, Germany; 7 Matrix45, Tucson, USA; 8 University of Arizona, Center for Health Outcomes and PharmacoEconomic Research, Tucson, USA Background: MONITOR-GCSF is a European prospective observational study of CIN/FN prophylaxis with Sandoz’ filgrastim (EP-2006). We report

Abstracts on the predictive modeling of determinants of outcomes and these models’ use to calculate better-to-worse-case scenarios. Materials and Methods: 1447 evaluable patients (pts) from 140 centers in Europe for up to 6 cycles within a single chemotherapy line including a total of 6213 cycles. Models developed when using (1) pts (“ever CIN/FN experienced”) and (2) cycles (“CIN/FN experienced during cycle”) as analysis unit for the following outcomes: CIN grade 4 (CIN4) episode, FN episode, hospitalization, chemodisturbance, and composite (10 models total). Applying EORTC CIN/FN guidelines, pts were classified as to prophylaxis intensity as under-, correctly-, or over-prophylacted. Prophylaxis initiation was scored per congruence with EORTC guidelines. Better-to-worse pt risk scenarios were constructed by varying presenting determinants in hypothetical pts based from minimal to maximal. Results: Pts (median age 62 y) were mainly female (61%), with solid tumor (77%), ECOG 0/1 (89%), median of 2.5 CIN/FN risk factors, receiving primary prophylaxis (72%); 32% with prior line of chemo, of whom 23% experienced CIN4 episode. Outcome determinants retained across the 5 patient models included: age, female gender, CIN4 history, repeated infections history, ECOG2 ever during study, concomitant antibiotic prophylaxis; prophylaxis intensity. Associated pt risk ranged from <1.0% (FN episode, hospitalization) to 5.7% (chemodisturbance) in “better” scenarios; from 14.7% (hospitalization) to 72.2% (composite) in “worse” scenarios. Outcome determinants retained across the 5 cycle models included: time and duration of prophylaxis initiation; ECOG; concomitant antibiotic prophylaxis; any grade CIN in prior cycle; hematological cancer; history of anemia or CIN4 at enrollment; prophylaxis intensity; center cancer and chemo case mix; center affiliation. Associated cycle risk ranged from <1.0% (all outcomes except composite index where 1.5%) in “better scenarios”; from 21.1% (hospitalization) to 78.3% (composite) in “worse” scenarios. Conclusions: Our study integrates the static approach of examining variables assessed when starting chemotherapy with the dynamic approach of investigating which and how determinants may emerge, disappear, or change in intensity across cycles. Pt-level analyses confirm to clinicians the need for assessing risk factors, determinants, and predictors prior to first chemo cycle. Cycle-level analyses substantiate the need for clinicians to reassess risk at start of each cycle per EORTC guidelines. The risk of pts ever experiencing CIN/FN-related outcome, or such outcome occurring in a given cycle, can be quantified to support clinical practice. Conflict of interest: Advisory Board: MA and PG have attended advisory boards for Sandoz. Corporate-sponsored Research: MA, HL, CB, PG, MB are Study Steering Committee members. KD, KM, IA are employees of Matrix45 who have provided consultancy services to Sandoz,. Other Substantive Relationships: MT is an employee of Sandoz. 1597 POSTER Prognosis of leptomeningeal metastasis diagnosed by magnetic resonance imaging in 329 patients with advanced solid tumors T. Urasaki1 , Y. Naito1,2 , M. Sasaki1 , N. Matsubara1 , A. Hosono1 , T. Kogawa1,2 , H. Kuno3 , T. Kobayashi3 , M. Kusumoto3 , S. Niho4 , K. Goto4 , T. Yoshino5 , H. Mukai1 . 1 National Cancer Center Hospital East, Breast and Medical Oncology, Chiba, Japan; 2 National Cancer Center Hospital East, Experimental Therapeutics, Chiba, Japan; 3 National Cancer Center Hospital East, Diagnostic Radiology, Chiba, Japan; 4 National Cancer Center Hospital East, Thoracic Oncology, Chiba, Japan; 5 National Cancer Center Hospital East, Gastrointestinal Oncology, Chiba, Japan Background: Leptomeningeal metastasis (LM) was a sign of poor prognosis in patients (pts) with solid tumors, with the median survival time (MST) of approximately 3 months as previously reported. Since recent rapid progress in various medical treatments has led to a significant improvement of overall survival, we investigated the prognosis and related prognostic factors in pts with LM. Methods: All MRI reports taken from May 2008 to May 2014 at our institute were reviewed by radiologists. Corresponding clinicopathological factors and genetic features including genetic alterations and gene amplification were retrieved from medical records. Correlation of those factors and overall survival was analyzed. Results: Data on 7,006 MRI reports were available in 4,299 pts. LM was detected in 329 pts (7.7%) with solid tumors. The characteristics of 329 pts were as follows: median age was 64 (26−82), 162 were female (49.2%), ECOG PS 0−1/2/3−4 was 106 (32.2%)/119 (36.2%)/104 (31.6%), respectively. Consciousness disturbance was observed in 28 pts (8.5%) at the time of LM diagnosis. The primary site of tumors were lung (N = 214; 65.0%), breast (N = 52; 15.8%), stomach (N = 17; 5.2%), and others (N = 46; 14.0%). Among 171 pts with non-small cell lung cancer (NSCLC), 59 pts with EGFR mutations, 32 pts with wild-type EGFR and the remaining 80 pts with unknown EGFR status. The MST of the whole pts was 3.3 months. Univariate analysis demonstrated that tumor types, ECOG PS, and consciousness disturbance were significantly

Abstracts

S237

associated with survival. In NSCLC, EGFR mutation was also correlated with survival (MST for EGFR mutant and wild-type were 9.6 and 3.5 months, respectively). In breast cancer neither ER status nor HER2 status was correlated with survival. On multivariate analysis, tumor types, ECOG PS, consciousness disturbance were significantly correlated with survival (Table). We also evaluated whether RTOG recursive partitioning analysis (RPA) could predict survival of pts with LM. The MST of RTOG RPA class I (N = 33; 10.0%), II (N = 191; 58.1%), and III (N = 105; 31.9%) were 8.7, 5.9 and 1.1 months, respectively. The MST of class I and II was significantly better than that of class III (p < 0.001 each). Variable

Number of pts

ECOG PS (vs. PS 0/1) PS 2 PS 3/4 Gender (male vs. female) Age Types of tumor (vs. EGFR mutant NSCLC) GI cancers Other cancer (including EGFR wild NSCLC) Consciousness disturbance (Yes vs. No)

106 119 104 167 vs. 162 329 59 37 233 28 vs. 301

Hazard ratio

P value

95% CI

1.427 3.747 1.188 1.007

0.020 <0.001 0.181 0.213

1.058–1.925 2.728–5.147 0.923–1.529 0.996–1.019

3.951 1.943 2.092

<0.001 <0.001 0.001

2.447–6.381 1.394–2.707 1.362–3.216

Cycle

Phase

1 Acute Delayed Overall 2 Acute Delayed Overall 3

Conclusions: Survival of pts with LM was poor across the cancer types except for the EGFR mutated NSCLC. Tumor type, ECOG PS, and consciousness disturbance predict survival. RTOG RPA could be utilized for stratification of pts with LM. No conflict of interest. 1598 POSTER A multicycle phase III study evaluating palonosetron vs ondansetron at preventing chemotherapy-induced nausea and vomiting in pediatric patients E. Kabickova1 , A. Wachtel2 , E. Basharova3 , T. Spinelli4 , P. Nicolas4 , G. Kovacs5 . 1 Department of Pediatric Hematology and Oncology, Charles University 2nd Medical School, Prague, Czech Republic; 2 Department ´ of Pediatrics, Instituto Nacional de Enfermedades Neoplasicas, Lima, Peru; 3 Oncohematology Center, Chelyabinsk Pediatric Regional 4 Clinical Hospital, Chelyabinsk, Russian Federation; Corporate Clinical Development, Helsinn Healthcare SA, Lugano, Switzerland; 5 Second Department of Pediatrics, Semmelweis University, Budapest, Hungary Background: Palonosetron is non-inferior to ondansetron at preventing chemotherapy-induced nausea and vomiting (CINV) in adult patients receiving moderately/highly emetogenic chemotherapy (MEC/HEC). This multinational double-blind study evaluated efficacy and safety of two palonosetron doses vs ondansetron in pediatric patients receiving up to four cycles of MEC/HEC. Materials and Methods: Patients scheduled to receive MEC/HEC were randomized to receive intravenous palonosetron (10, 20 mg/kg) or ondansetron (3×150 mg/kg). For the primary objective statistics were used to show non-inferiority for palonosetron (d=-15%) vs ondansetron from complete response rates (CR: no vomiting/retching/rescue medication) during the acute phase (0−24h after the first MEC/HEC dose) of cycle 1. Secondary objectives included CR rate during the delayed (>24–120h) and overall (0–120h) phases, the proportion of patients without vomiting and safety. Results: Of 502 patients randomized, 493 aged 2.1 months to 16.9 years were included in the full analysis set. CR rates were highest in the palonosetron 20 mg/kg group during all phases of cycles 1, 3 and 4 with non-inferiority shown for this dose vs ondansetron during the acute phase of cycle 1 (97.5% CI −11.7–12.4; p = 0.0022). In cycle 1 reports of no vomiting (see table) were higher in the palonosetron groups vs the ondansetron group during all phases with proportions in the palonosetron 20 mg/kg group >10% higher. In cycles 2−4 reports of no vomiting were consistently higher in the palonosetron 20 mg/kg group vs the ondansetron group during all phases. In cycles 1−4 treatment-emergent adverse events (TEAEs, %) were fewer in the palonosetron 20 mg/kg group (69.3, 64.4, 55.9, 48.4) vs the palonosetron 10 mg/kg (80.2, 76.2, 72.1, 75.0) and ondansetron (81.7, 82.6, 68.2, 72.2) groups. All withdrawals/fatal TEAEs were not deemed drugrelated. Electrocardiogram evaluations raised no concerns. Conclusions: In pediatric patients receiving up to four cycles of MEC/HEC, intravenous palonosetron 20 mg/kg has shown non-inferiority vs ondansetron during the acute phase of cycle 1, numerical superiority vs ondansetron during all phases of cycles 1−4 for no vomiting rates, and presented no significant safety risks. Conflict of interest: Other Substantive Relationships: Edita Kabickova and Antonio Wachtel have had travel, accommodation and/or other expenses paid for by Helsinn Healthcare SA. Tulla Spinelli and Pierre Nicolas are employees of Helsinn Healthcare SA. Gabor Kovacs has received honoraria from Helsinn Healthcare SA.

Acute Delayed Overall 4 Acute Delayed Overall

Patients without vomiting, n (%) Palonosetron Palonosetron 10 mg/kg 20 mg/kg

Ondansetron 3×150 mg/kg

N = 166 133 (80.1) 113 (68.1) 98 (59.0) N = 82 69 (85.2) 66 (81.5) 59 (72.8) N = 43 33 (76.7) 35 (81.4) 29 (67.4) N = 19 15 (78.9) 15 (78.9) 14 (73.7)

N = 162 119 (73.5) 94 (58.0) 83 (51.2) N = 86 66 (76.7) 65 (75.6) 54 (62.8) N = 44 36 (81.8) 31 (70.5) 28 (63.6) N = 19 13 (68.4) 14 (73.7) 10 (52.6)

N = 165 138 (83.6) 122 (73.9) 114 (69.1) N = 90 79 (87.8) 75 (83.3) 70 (77.8) N = 59 56 (94.9) 52 (88.1) 52 (88.1) N = 31 27 (87.1) 27 (87.1) 25 (80.6)

1599 POSTER Current practice in the detection of cardiovascular toxicity in breast cancer patients over 70 years treated with anthracyclines A. Gobert1 , C. Mateescu2 , D. Lassoued1 , C. Falandry3 , C. Sajous3 , A. Bruyas3 , S. Guita4 , S. Edehry5 , J.P. Spano1 , G. Freyer3 . 1 GHPS-CFX-APHP-IUC/UPMC, Medical oncology, Paris, France; 2 Pitie´ ˆ ere ` University Hospital, Medical oncology, Paris, France; 3 Hospital Salpetri Center Lyon Sud, Medical oncology, Lyon, France; 4 Teva laboratory, ´ Medical Head of Oncology, Paris, La Defense, France; 5 Saint Antoine Hospital, Cardiology department, Paris, France Background: Anthracyclines may cause acute or delayed cardiac toxicity. In breast cancer (BC), they are administered in neoadjuvant (NAd), adjuvant (Ad) or metastatic settings. The surveillance of baseline and subsequent cardiac function in patients receiving anthracycline-based chemotherapy (ABC) is poorly documented in routine practice. Methods: Observational, prospective, cardiac surveillance, French multicenter study. Patients over 70 years with HER2 negative BC scheduled to receive ABC were enrolled before starting chemotherapy. Clinical examination, laboratory assays, electrocardiogram (ECG) and left ventricular ejection fraction (LVEF) were collected at baseline, 6 weeks (V1), 18 weeks (V2), 6 months (V3) and 1 year (V4). Results: From January 2012 to July 2014, 114 patients were included by 29 investigators (private clinics 41.4%, teaching hospitals17.2%, anticancer centers 10.3%). Median age was 74 years (75−85 in 39.8% of cases). Nine and 2 patients had had prior ABC in the Ad and NAd settings respectively. They received 6.8±3.2 cycles of ABC (median: 6, range: 1−16) as Ad/Nad (68.4%) or metastatic (31.6%) treatment. Liposomal doxorubicin was prescribed in 37.7% of cases. Most patients (91.4%) had ECOG 0 or 1 at inclusion, 73.5% had at least one cardiovascular (CV) risk factor (mainly hypertension or dyslipidemia) and 13.9% a history of CV event. ECG, LVEF and brain natriuretic peptide (BNP) measurement were performed in 54.9%, 93.8% and 5.3% of patients at baseline, 17.3%, 22.0% and 3.6% at V1, 16.7%, 20.6% and 1.0% at V2; 6.7%, 10.2% and 1.1% at V3; and 10.0%, 16.3% and 1.3% at 1 year. They were performed at least once during follow-up in 31.2%, 43.5% and 4.6% of patients. Among those who had appropriate examinations, de novo hypertension, dyspnea, ECG abnormalities, LVEF 50% and congestive heart failure occurred in 5, 13, 4, 2, and 1 patients respectively. A cardioprotective treatment (dexrazoxane) was administered in 2 patients. Conclusions: The evaluation of cardiac function is incomplete before administration of ABC and cardiac tests are insufficiently repeated during treatment, even in a population of elderly patients with cardiovascular risk factors No conflict of interest. 1600 POSTER Incidental pulmonary embolism in cancer: A prognostic score derived from a prospective cohort with uniform management G. Bozas1 , D. Ramanujam-Venkatachala1 , N. Jeffery1 , G. Avery2 , A. Stephens2 , V. Algar3 , P. June1 , M. Elliott1 , M. Anthony1 . 1 Castle Hill Hospital, Queen’s Centre for Oncology and Haematology, Cottingham, United Kingdom; 2 Castle Hill Hospital, Department of Radiology, Cottingham, United Kingdom; 3 Hull York Medical School, Medical Statistics, York, United Kingdom Background and Aims: We analysed our prospective cohort of patients with active cancer and incidental pulmonary embolism (IPE) uniformly