(B-008) Comparing approaches to real-world progression-free survival estimation among patients with advanced melanoma on first-line immunotherapy

Background: Although overall survival (OS) is generally considered the gold standard measure for treatment effectiveness in oncology studies, real-world progression-free survival (rwPFS) remains an important outcome as it can address some of the challenges of OS, such as immature OS due to event frequency and follow-up requirements. However, the assessment of rwPFS can raise questions around definition, evaluation/assessment bias, and missing data.

Objectives: To compare two approaches to evaluate real-world progression-free survival among patients with advanced melanoma initiating first-line (1L) treatment with immunotherapy (IO) regimens.

Methods: We conducted a cohort study among patients with advanced melanoma initiating 1L systemic therapies between January 2016 to October 2023 in the US using the ConcertAI Oncology Database. The index date was defined as the date of 1L initiation of an IO regimen. RwPFS was defined as the time from the index date to the date of disease progression or death, whichever occurred first. Censoring was defined in two ways. For definition 1 (D1): RwPFS was censored at the earliest of the following: second-line (2L) initiation, loss-to-follow-up (LTFU), or end of the study period. For definition 2 (D2): rwPFS was censored at the earliest of LTFU date or end of the study period. Median rwPFS and 95% confidence intervals (CI) were estimated with the Kaplan-Meier estimator.

Results: We identified 1,388 eligible patients, of whom 62.1% initiated IO monotherapy, and 37.9% IO combination therapy. The mean age was 66 years (standard deviation = 13) and 34.5% of patients were female. Median follow-up for the full cohort was 12 months. We observed a difference in median rwPFS estimates according to the two definitions for patients on monotherapy regimens. Among patients initiating IO monotherapy, the D1 median rwPFS was 11.7 months (95% confidence interval (CI): 9.5 – 14.3) vs 13.0 months (95% CI: 10.5 – 15.6) for D2. However, for combination therapy, the median rwPFS point estimates were the same for both definitions. For IO combination therapy, the D1 median rwPFS was 8.0 months (95% CI): 6.2 – 10.3) vs 8.0 months (95% CI: 6.2 – 10.2) for D2.

Conclusions: In this analysis on two rwPFS approaches following 1L IO treatment, we observed a differential impact on median rwPFS for monotherapy regimens compared to combination regimens when different censoring rules were applied. This finding may be related to differences in the distribution of progression and censoring events across the regimens, as well as differential disease evaluation or monitoring.