Effect of early antibiotic exposure on survival of patients receiving atezolizumab plus bevacizumab but not sorafenib for unresectable HCC: A sub-analysis of the phase III IMbrave150 study.

Abstract

597 Background: The combination of atezolizumab plus bevacizumab (A+B) is the standard of care for patients (pts) with unresectable/advanced hepatocellular carcinoma (HCC). Emerging evidence suggest a detrimental role of early antibiotics (ATB) exposure in immunotherapy recipients. However, it is unclear whether ATB affect outcomes of pts receiving A+B. Methods: We conducted a patient-level analysis of subjects receiving either A+B or Sorafenib (Sor) within the IMbrave150 trial to assess the impact of early ATB exposure (ATB+, i.e. +/-30 from treatment initiation) on overall (OS) and progression-free survival (PFS). Differences in median (m) OS and mPFS were estimated by the Kaplan-Meier method. Cox proportional regression models were used to assess the prognostic impact of ATB+ adjusting for race, age, baseline AFP, baseline ALBI grade, continent of origin and neutrophil-to-lymphocyte ratio, a measure of systemic inflammation. Results: Data of 420 IMbrave150 participants were made available by Vivli. Among them, 278 received A+B and 142 Sor. Most pts were of ECOG performance status 0 (65%); 47% had extrahepatic spread, 40% had macrovascular invasion and 35% had AFP>400 ng/mL. Viral aetiology (39% HBV and 24% HCV) was the most common risk factor for HCC. Overall, 60 pts in the A+B arm (21.5%) and 31 (21.8%) in the Sor group were ATB+. Baseline characteristics were well balanced across ATB groups, except for region of origin, with a higher proportion of Asian pts in ATB+ (60% vs 45%; p=0.01). After a median follow-up of 10.38 months (95%CI:8.6-12.5), OS and PFS rates at 12 months were 72.3% and 9.7% in A+B; 64.8%, and 1.4% in Sor, respectively. ATB+ exposure was associated with increased risk of death both in univariate (HR 1.67; 95%CI 1.04-2.69) and multivariate models (HR 1.62; 95%CI 1.40-1.89) in the A+B (p for interaction <0.001), but not in the Sor arm (univariate HR 0.63; 95%CI 0.32-1.25; multivariate HR 0.70; 95%CI 0.31-1.60). Similarly, the risk of progression was increased for ATB+ pts in the A+B arm (univariate HR 1.68; 95%CI 1.03-2.71; multivariate HR 1.59; 95%CI 1.40-1.81), but not in the Sor arm (univariate HR 0.51;95%CI 0.26-1.01; multivariate HR 0.53; 95%CI 0.21-1.30) with positive p for interaction (p=0.002). Objective response rate (ORR) according to RECIST v.1.1 did not differ between ATB+ (28.0%) and ATB- (30.0%) in A+B (p=0.68). Conclusions: In a post-hoc sub-analysis of IMbrave150 we demonstrate for the first time the negative impact of early antibiotics exposure on outcomes of patients receiving A+B but not in those treated with Sor. Prospective translational studies should evaluate the role of ATB-mediated gut dysbiosis as a proposed mechanism underlying the adverse outcome in immunotherapy-recipients.