The comparative efficacy of atezolizumab and bevacizumab versus lenvatinib in patients with unresectable hepatocellular carcinoma (uHCC).

Abstract

e16151 Background: Hepatocellular carcinoma (HCC) is the most common type of liver cancer, comprising 90% of all cases. In the phase III IMbrave150 trial, combination of atezolizumab and bevacizumab (atezo+bev) resulted in improved overall survival (OS) and progression-free survival (PFS) compared with sorafenib for patients with unresectable hepatocellular carcinoma (uHCC). In the phase III REFLECT trial, lenvatinib met the primary endpoint of non-inferiority in OS versus sorafenib, and demonstrated superiority in secondary endpoints; PFS, time to progression and objective response rate (ORR). Furthermore, post-hoc analyses strongly suggest that the original non-inferiority trial likely underestimated the true effect of lenvatinib on OS due to imbalances in baseline prognostic covariates and the comparatively greater use of post-treatment anti-cancer therapies in the sorafenib arm. In the absence of a head-to-head comparison between atezo+bev and lenvatinib, the objective of this analysis was to estimate their comparative efficacy as first-line treatments for patients with uHCC. Methods: A network meta-analysis (NMA) was conducted based on a de novo systematic literature review which identified evidence for licensed therapies in first-line uHCC. This was supplemented with data from the extended IMbrave150 study findings (clinical cut-off on 31 August 2020). Median hazard ratios (HR) with corresponding 95% credible intervals (CrI) were estimated for OS and PFS. HRs for OS and PFS from REFLECT were taken from multivariable analyses which adjusted imbalances in baseline prognostic covariates. All analyses were conducted in a Bayesian generalised linear model framework. Model selection was based on deviance information criterion (DIC). Results: Four eligible phase III studies (REFLECT, SHARP, Asia-Pac and IMbrave150) were included. Using a fixed-effects model, the estimated median HRs for atezo+bev compared with lenvatinib were 0.81 (95%CrI: 0.61, 1.08; posterior p=0.078) and 0.99 (95%CrI: 0.76, 1.28; posterior p=0.454) for OS and PFS, respectively. The results obtained were similar to those observed by Vogel et\xa0al, who reported an OS HR of 0.63 (95% CI: 0.32, 1.25), and a PFS HR of 0.91 (95% CI: 0.23, 3.65). Conclusions: While the NMA respects the randomization of the included studies, it does not represent direct evidence and should therefore be considered exploratory in nature. The results of this NMA suggest that the benefit of atezo+bev is not statistically significant compared to lenvatinib, as the differences between atezo+bev and lenvatinib fell within the Bayesian range of equivalence (upper limit of Crl ≥ 1 for OS and PFS) and were subject to a high degree of uncertainty. The results of the NMA were consistent for all the publicly available data cuts from IMbrave150.