Landmark analysis of OS by response status in previously treated patients (pts) with advanced hepatocellular carcinoma (aHCC): Post hoc analysis of the KEYNOTE-394 study.

Abstract

4074 Background: KEYNOTE-224 and KEYNOTE-240 landmark analyses showed that objective response (OR) in pembrolizumab (pembro)-treated pts was prognostic of longer OS after the landmark. Landmark analysis was performed using the KEYNOTE-394 study to determine whether response assessment before landmark is prognostic of longer OS after landmark time in pembro-treated pts. Methods: Adults with confirmed HCC, radiographic progression during or after treatment with or intolerance to sorafenib or oxaliplatin-based chemotherapy were eligible for KEYNOTE-394. Landmark analyses of OS for responders, pts with SD, and pts with other status in these analyses at 6, 12, 18, and 24 wk after randomization were performed on the pembro arm. Tumor imaging was performed every 6 wk after randomization. Responders (R) at each landmark were defined as pts with any response assessment of CR or PR before the landmark date; all other pts were defined as nonresponders (NR). SD at each landmark was defined as pts with a response assessment of SD and no better response before the landmark date. Other was defined as pts with responses of PD or a missing/not evaluable status. Response was assessed per RECIST v1.1 by blinded independent central review. HR and 95% CI for survival after the landmark are based on the Cox regression model with the Efron method for handling ties, with response status by the landmark time as a single covariate. Results: Median time from randomization to data cutoff date (June 30, 2021) was 33.8 mo (range, 18.7-48.8) for pembro. OS after landmark time was numerically longer for R vs NR, R vs pts with SD, and pts with SD vs pts with Other status. Conclusions: KEYNOTE-394 landmark analysis supports the prognostic association between OR with pembro before the landmark and OS after the landmark observed in KEYNOTE-224 and KEYNOTE-240 in previously treated pts with aHCC. Landmark analysis also suggested longer OS after the landmark for pts who achieved a better response to pembro before the landmark. Clinical trial information: NCT03062358 . [Table: see text]