© 2024 MJH Life Sciences™ and Center for Biosimilars®. All rights reserved.
Oncology trials often report on progression-free survival or overall survival, but the authors of an analysis presented during a poster session today at the European Society of Medical Oncology (ESMO) Congress argue that such endpoints are less sensitive then short-term overall response rate.
Oncology trials often report on progression-free survival (PFS) or overall survival (OS), but the authors of an analysis presented during a poster session today at the European Society of Medical Oncology (ESMO) Congress argue that such endpoints are less sensitive then short-term overall response rate (ORR) in confirming biosimilarity when long median PFS or OS are expected.
The authors presented considerations on interpreting survival data as illustrated by a sensitivity analysis from a recent study of biosimilar rituximab. In a confirmatory phase 3 study, treatment-naïve patients with follicular lymphoma received either European Medicines Agency-approved GP2013 (n = 314) or reference rituximab (n = 315).
Patients who responded to an induction treatment with cyclophosphamide, vincristine, and prednisone chemotherapy received either GP2013 or reference rituximab as maintenance monotherapy.
The primary endpoint for equivalence was ORR at the end of induction. Secondary endpoints included PFS and OS.
As of December 2016, the median follow-up was 23.6 months for the GP2013 arm and 24.2 months for the reference rituximab arm. The study met its primary endpoints to confirm biosimilarity.
Median PFS and OS could not be estimated. Hazard rations for PFS and OS were 1.31 (90% confidence interval [CI]; range, 1.02 to 1.69) and 0.77 (90% CI; range, 0.29 to 1.22), respectively.
A Kaplan-Meier analysis showed that PFS survival curves diverged between 12 and 24 months, yet the curves ran parallel outside of that period, contrary to the proportional hazard assumption of the Cox model. Complete response (CR) rates were similar between treatments at all time points, including at 33 months.
The researchers concluded that a small sample size, a low event rate, data immaturity, and other aspects of study design can decrease a study’s sensitivity in biosimilarity assessments, as demonstrated by the fact that hazard ratios for PFS and OS had opposite directions, while CR rates remained similar across time.
PFS and OS results, the authors say, should be interpreted with caution, as they may not reflect either a difference or a lack of a difference between treatments.