SO-13 Pembrolizumab in patients with advanced hepatocellular carcinoma (aHCC) previously treated with sorafenib: Updated data from the open-label, phase 2 KEYNOTE-224 study

Abstract

Pembrolizumab demonstrated efficacy and a manageable adverse event profile in sorafenib-treated patients with aHCC in KEYNOTE-224, leading to accelerated approval in the United States. KEYNOTE-240 showed a similar ORR and clinically meaningful improvements in OS and PFS, although prespecified statistical significance criteria for OS and PFS were narrowly missed. Recently in KEYNOTE-394, pembrolizumab plus BSC demonstrated statistically significant and clinically meaningful benefit in OS, PFS, and ORR versus placebo plus BSC in patients from Asia with aHCC who were previously treated with sorafenib or oxaliplatin-based chemotherapy. We report efficacy and safety from cohort 1 of KEYNOTE-224 for patients with sorafenib-treated aHCC after ∼5 years of follow-up. Adults with pathologically confirmed HCC who experienced progression after or intolerance to sorafenib treatment received pembrolizumab 200 mg every 3 weeks for ≤35 cycles or until confirmed progression, unacceptable toxicity, withdrawal of consent, or investigator decision. Primary end point was ORR assessed by BICR per RECIST v1.1. Secondary end points included DOR, DCR, TTP, and PFS, all assessed by BICR per RECIST v1.1; OS; and safety and tolerability. A post hoc landmark analysis of OS among responders at first scan (days 40-77), and nonresponders at first scan was performed. A total of 104 patients were assessed for efficacy and safety. Median time from first dose to data cutoff (October 1, 2021) was 59.1 months (range, 55.4-63.3) and median duration of exposure was 4.2 months (range, 0.0-38.4). ORR was 18.3% (95% CI, 11.4-27.1), and DCR was 61.5%. Best overall responses were CR in 3.8% (n = 4) of patients, PR in 14.4% (n = 15), SD in 43.3% (n = 45), and PD in 32.7% (n = 34). Median DOR was 21.0 months (range, 3.1 to 52.6+); 77.0% and 60.6% of patients had a response lasting ≥12 months and ≥18 months (Kaplan-Meier), respectively. Median TTP was 4.8 months (95% CI, 3.9-7.0) and median PFS was 4.9 months (95% CI, 3.5-7.0). The estimated PFS rate at 24 months was 12.5%. Median OS was 13.2 months (95% CI, 9.7-15.3); estimated OS rates at 24 and 36 months were 30.8% and 20.2%, respectively. In a landmark analysis, median OS from the first scan was 53.1 months (95% CI, 10.7-not reached) among responders (n = 11) and 10.3 months (7.3-12.5) among nonresponders (n = 87); estimated OS rates among responders versus nonresponders at the first scan at 24 and 36 months were 81.8% versus 26.4% and 63.6% versus 16.1%, respectively. Of 104 patients, 76 (73.1%) reported treatment-related AEs; most were low grade in severity (grade 3/4, n = 26 [25.0%]; grade 5, n = 1 [1.0%]). No new treatment-related deaths, immune-mediated hepatitis events, or viral-induced hepatitis flares occurred with extended follow-up. Longer-term follow-up from KEYNOTE-224 demonstrated that pembrolizumab provides durable efficacy and manageable AEs in patients with sorafenib-treated aHCC. These data, together with previously reported findings from KEYNOTE-240 and positive results from KEYNOTE-394, reinforce the long-term efficacy and favorable benefit-risk profile of pembrolizumab in the aHCC setting.