Study design
Four centres participated in this retrospective clinical study, three centres in France (Institut régional du Cancer de Montpellier [ICM], Montpellier, Institut Claudius Regaud, Toulouse, and Centre Georges-François Leclerc, Dijon), and one centre in Switzerland (CHU Vaudois, Lausanne). An Access database was created to collect retrospective data using different panels: patients’ identification, tumour histology, previously delivered treatments and regimens, initial assessment, initial biology, EM chemotherapy-related data (delivered EM dose and cause of diminution, number of cycles administered, toxicities and side effects), efficacy and follow-up data. The tumour response was assessed every 3 cycles of treatment. Tumours were considered as ER and PR positive when > 10 % tumour cells were stained by immunohistochemistry (IHC). HER2 status was determined based on HER2 protein expression level by IHC. Tumours with HER2 scores of 0 and 1+ were considered as HER2 negative. In tumours with equivocal HER2 IHC test results (2+), gene amplification was evaluated using fluorescent (FISH) or chromogenic (CISH) in situ hybridization. Specimens with HER2 3+ scores were considered as HER2 positive. The Database locking and patients’ follow-up was scheduled for April 7th, 2014. This study was reviewed and approved by the respective Institutional Review Boards (Dijon, Toulouse and Montpellier Cancer Centres, and Lausanne CHUV, ID number ICM-URC-2014/73). Considering the retrospective, non-interventional nature of this study evaluating an approved drug, no consent was deemed necessary by the clinical research review board of Montpellier cancer centre (sponsor of the study).
Patients
All patients affected by a metastatic or locally advanced breast cancer treated with EM between March 28th, 2011 and January 15th, 2014 in one of the participating centres, were included in our retrospective analysis. Patients with EM treatment initiated in other centres, and who received only one EM injection or cycle in a participating centre, were not considered suitable for this study due to the lack of data.
Treatment
Eribulin mesylate was administrated intravenously over 2 to 5 minutes on days 1 and 8 of a 21-day chemotherapy cycle, according to the product guidelines. The EM treatment was generally administered at the standard dose of 1.4 mg/m2. It was reduced in case of hepatic or moderate renal impairments: a -20 % reduction of the recommended EM dose (1.1 mg/m2) was applied for patients with mild hepatic impairment (Child-Pugh A) or moderate renal impairment (creatinine clearance of 30–50 mL/min), and a −50 % reduction (0.7 mg/m2) was administered in patients with moderate hepatic impairment (Child-Pugh B). Clinical and radiological assessment of the tumour response was performed according to each centre standard of care, most of the time every 3 cycles of treatment (every 9 weeks). Clinical and biological assessment of toxicity was performed at each clinical visit, i.e. at day 1 and day 8 of each 21-day cycle. Primary and secondary granulocyte-colony stimulating factor (G-CSF) prophylaxis was delivered according to each centre’s practice.
Efficacy assessment
Clinical and radiological efficacy assessment was performed every 3 cycles by a medical oncologist during the whole treatment period. Response and progression evaluations were performed using the RECIST version 1.1 criteria [12]. For each evaluation, treatment response was determined as such: complete response (CR), partial response (PR), stable disease (SD), progressive disease (PD), or not established (NE). Clinical proposal following assessment (continued treatment, dose reduction or treatment discontinuation) was recorded, together with the reasons of treatment discontinuation (toxicity, evolution, death or other).
Safety assessment
Clinical and biological toxicities were retrospectively identified and graded according to the Common Terminology Criteria grid for Adverse Events (CTCAE) version 4.03 at each clinical visit, using the patients’ clinical charts. Interdose complications were recorded: treatment delay (duration and cause), treatment cancellation and reason, hospitalization (duration and cause), duration of antibiotic treatment, and need and number of Red Blood Cells (RBC) and/or platelet concentrates in case of transfusion.
Statistical analysis
For the descriptive analysis, quantitative variables were presented for each group and for the overall population as mean, variance, standard deviation, minimum, maximum, and median. Quantitative criteria were compared using the Kruskal-Wallis test. Qualitative variables were presented as numbers and frequencies for each category of the variable. Qualitative criteria were compared using the chi2 or by the Fisher exact tests when the chi2 test was not applicable. The primary objective of the study was to assess overall survival until progression. Survival estimates were calculated using the Kaplan-Meier method. The time to progression was defined for each patient as the time from the first cycle until objective tumour progression (TTP does not include deaths). The secondary objective was overall survival (OS), defined as the time from the first cycle of treatment until death from any cause. For the clinicopathological features, univariate analyses to compare clinical benefit and no clinical benefit were performed using Pearson’s 2 or Fisher’s exact tests for categorical variables, and the two-sample Wilcoxon test for all continuous variables. Categorical covariates analysis were ECOG Performance Status (0–1 vs. 2–3), hormone receptors (HR- vs. HR+), and age (≤50 vs. > 50 years), HER2+ over-expression (no/yes), number of prior chemotherapy (≤4 vs. > 4 lines) and different metastasis localization (visceral metastases, liver, bone, lymph nodes, lung, brain, serous or skin metastases), and response under taxane chemotherapy (CR/PR/SD vs. PD/NE). Differences were considered statistically significant when p < 0.05. Significant factors in the univariate analyses were included in a multivariate logistic regression analysis to identify independent predictors of the clinical benefit.
For the multivariate analysis using the Cox’s proportional hazard model to define independent prognostic factors for PFS, the variables included in the logistic regression were used. The hazard ratio and the 95 % confidence interval (CI) were also estimated.
Different bases transfers were made through STAT-TRANSFER version 9 and data were analysed using the STATA® version 13.0 software.