Early bevacizumab dose and time modifications may affect efficacy of atezolizumab plus bevacizumab for advanced hepatocellular carcinoma treatment

Background: Atezolizumab (1200 mg) plus bevacizumab (15 mg/kg) every 3 weeks (AtezBev) became a standard-of-care first-line treatment for advanced hepatocellular carcinoma (aHCC) following IMbrave150. However, real-world data suggest milder efficacy. Early bevacizumab interruption due to adverse events (AEs) is a frequent occurrence in real-world scenario associated with poor prognosis. Additionally, early bevacizumab dose/time modifications (eBEVmod) may negatively impact outcome. Materials and methods: Data from AtezBev-treated aHCC patients (n = 100) in five Italian institutions were retrospectively analyzed. Cumulative bevacizumab dose (mg/kg) received in the first 3 months of treatment was analyzed by receiver operating characteristic (ROC) to identify a cut-off value to estimate survival and dichotomize the variable. Baseline clinical and laboratory characteristics were analyzed with uni-/multivariate models to explore potential differences on overall survival (OS), objective response rate (ORR) and disease control rate (DCR) based on eBEVmod. Progression-free survival (PFS) was a secondary endpoint. Results: In the overall population, the median (m) follow-up was 11.4 months, mOS was 20.5 months, mPFS was 10.4 months, ORR was 31% and DCR was 75%. ROC on 3-month cumulative bevacizumab dose revealed an area under the curve of 0.74 (P = 0.001) and eBEVmod cut-off of <45 mg/kg/3 months (i.e. 10.5 mg/kg/3 weeks) having 73% sensitivity and specificity in predicting death. Twenty-three percent of patients had eBEVmod (of which 39.1% had treatment delay, 39.1% discontinuation and 21.7% dose reduction), with no differences in baseline characteristics nor second-line treatments compared with non-eBEVmod, except for sex. eBEVmod patients had inferior ORR/DCR (14%/48% versus 36%/82%) and increased risk of death [hazard ratio (HR) 4.2, P = 0.0049], with a 12-month survival probability of 53.3% versus 77.4%. eBEVmod was an independent negative prognostic factor of survival at multivariate analysis (HR 3.3, P = 0.0125). Patients with eBEVmod also had a trend in worse PFS, although not statistically significant (mPFS 3.8 versus 12.6 months, HR 1.8, P = 0.0774). Conclusions: eBEVmod is an independent unfavorable prognostic factor of response and survival in AtezBev-treated aHCC patients. Optimization of bevacizumab AE management to reduce eBEVmod may substantially improve treatment outcome in real-world practice.