Fabian Finkelmeier

Atezolizumab plus bevacizumab (AB) and lenvatinib can be alternatively used as first‐line systemic treatment of unresectable hepatocellular carcinoma (HCC). However, no direct comparison of the two regimens has been performed in randomized clinical trials, making the identification of baseline differential predictors of response of major relevance to tailor the best therapeutic option to each patient. Baseline clinical and laboratory characteristics of real‐world AB‐treated HCC patients were analyzed in uni‐ and multivariate analyses to find potential prognostic factors of overall survival (OS). Significant variables were incorporated in a composite score (α‐FAtE) and it was tested for specificity and sensitivity in receiver operating characteristic (ROC) curve and in multivariate analysis for OS. The score was applied in uni‐ and multivariate analyses for OS of a comparable lenvatinib‐treated HCC population. Finally, comparison between treatments was performed in patients with low and high α‐FAtE scores and predictivity estimated by interaction analysis. Time‐to‐progression (TTP) was a secondary endpoint. OS of AB‐treated HCC patients was statistically longer in those with α‐fetoprotein <400 ng/mL (HR 0.62, p = .0407), alkaline phosphatase (ALP) <125 IU/L (HR 0.52, p = .0189) and eosinophil count ≥70/μL (HR 0.46, p = .0013). The α‐FAtE score was generated by the sum of single points attributed to each variable among the above reported. In ROC curve analysis, superior sensitivity and specificity were achieved by the score compared to individual variables (AUC 0.794, p < .02). Patients with high score had longer OS (HR 0.44, p = .0009) and TTP (HR 0.34, p < .0001) compared to low score if treated with AB, but not with lenvatinib. Overall, AB was superior to lenvatinib in high score patients (HR 0.55, p = .0043) and inferior in low score ones (HR 1.75, p = .0227). At interaction test, low α‐FAtE score resulted as negative predictive factor of response to AB ( p = .0004). In conclusion, α‐FAtE is a novel prognostic and predictive score of response to first‐line AB for HCC patients that, if validated in prospective studies, could drive therapeutic choice between lenvatinib and AB.