FDA pooled analysis of overall survival according to depth of response as a continuous variable in frontline advanced immuno-oncology renal cell carcinoma trials.

Abstract

e16532 Background: Retrospective analyses of individual studies of immune-oncology-containing combinations (IOC) for advanced renal cell carcinoma (aRCC) suggest that depth of response (DepOR; % reduction from baseline in sum of target lesion diameters) is associated with overall survival (OS). However, these analyses are limited by use of DepOR categories with a small number of patients and guarantee-time bias. We therefore sought to investigate the relationship of week 12 DepOR as a continuous variable with OS, hypothesizing not only complete but also deep partial responses might portend favorable longer-term survival. Methods: We pooled data from patients with treatment (tx)-naïve aRCC enrolled in randomized IO-based frontline aRCC trials submitted to FDA that included a Week 12 imaging assessment. We developed 36-month (mo) overall survival (OS) prediction models based on Week 12 DepOR per Independent Review Committee using Cox proportional hazards with natural spline in the IO-TKI combination (IOC) and sunitinib (SUN) groups. To avoid guarantee-time bias, OS was defined from date of an individual patient’s 12-week imaging, among the subset of patients who were alive and in follow-up at the Week 12 scan. Results: Four trials met inclusion criteria (KEYNOTE-426, JAVELIN Renal 100, CheckMate 9ER, CheckMate 214); in total, there were 1364 patients in the IOC group and 1267 patients in the SUN group. Eligibility criteria, baseline patient characteristics, and endpoints were similar both between the trials and between tx groups. Deepest response occurred at median 31 weeks (interquartile range [IQR], 18-55) in the IOC group and 29 weeks (IQR range, 17-48) in the SUN group. At Week 12, 34.7% of patients had DepOR of at least 30%; median DepOR was 27.6% (interquartile range [IQR] 8.7 to 43.4%) in the IOC and 13.7% (IQR 2.5 to 26.3%) in the SUN group. DepOR and 36-month OS were correlated throughout the entire range of DepOR in both tx groups; at each DepOR, the IOC group had slightly higher 36-mo OS over SUN. We saw similar results modeling 24-month OS compared to 36-month OS. Conclusions: This pooled exploratory analysis suggests that deeper response is associated with better 36-month OS in patients treated with IOC or SUN and slightly higher probability of 36-month OS for any given DepOR for IOC vs. SUN. We saw no plateau in OS as DepOR approached complete response. However, caution should be used when interpreting DepOR at the tails due to sparse data. Further work characterizing the relationship between DepOR and OS at the trial level may advance understanding of the utility of DepOR as an early endpoint in signal-seeking trials and to facilitate efficient drug development. Additionally, identifying patients with favorable long-term prognosis based on DepOR provides hypotheses for new trial designs.