A methods-forward indirect comparison of IO-IO versus IO-VEGF as first-line therapy for unresectable hepatocellular carcinoma.

Abstract

e16214 Background: Many immune-based regimens are approved for first-line unresectable hepatocellular carcinoma (uHCC), yet real-world treatment selection frequently centers on choosing between IO–IO and IO–VEGF approaches. While prior comparative analyses have ranked available therapies, few studies have performed a focused indirect comparison between these two classes using transparent assumptions or assessed treatment discontinuation due to toxicity as a clinically meaningful outcome. We conducted a targeted indirect comparison to inform this common clinical decision. Methods: We performed a pairwise Bucher indirect comparison using sorafenib as a common comparator, anchored on two phase III randomized trials: HIMALAYA (tremelimumab + durvalumab [STRIDE] vs sorafenib) and IMbrave150 (atezolizumab + bevacizumab vs sorafenib). Published hazard ratios (HRs) for overall survival (OS) and progression-free survival (PFS) were extracted without imputation. Indirect estimates were calculated on the log scale with variance from reported confidence intervals. Safety outcomes included grade ≥3 treatment-related adverse events (TRAEs). Treatment discontinuation due to adverse events was summarised descriptively as definitions varied across trials. Results: STRIDE improved OS versus sorafenib (HR 0.78), while atezolizumab + bevacizumab demonstrated a greater OS benefit (HR 0.66). The indirect comparison yielded an OS HR of 1.18 (95% CI 0.88–1.59) for IO–IO versus IO–VEGF. For PFS, the indirect HR was 1.38 (95% CI 1.06–1.80), favoring IO–VEGF. STRIDE demonstrated lower rates of grade ≥3 TRAEs compared with sorafenib (25.8% vs 36.9%), whereas atezolizumab + bevacizumab showed higher rates (43%), resulting in an indirect RR of 0.74 (95% CI 0.55–1.00) favoring IO–IO. Discontinuation outcomes were summarized descriptively. Conclusions: In this focused indirect comparison, IO–VEGF was associated with improved PFS and a numerically greater OS benefit compared with IO–IO, whereas IO–IO showed a more favorable severe toxicity profile. Rather than offering another broad ranking of regimens, this analysis addresses the practical clinical choice between IO–IO and IO–VEGF using transparent assumptions and clinically relevant tolerability outcomes. These results suggest that first-line treatment selection may be guided by bleeding risk, immune tolerance, and comorbidities. Trial-level inputs and indirect comparison (Bucher; common comparator = sorafenib). Endpoint HIMALAYA (STRIDE vs sorafenib) IMbrave150 (Atezo+Bevvs sorafenib) Indirect (STRIDE vs Atezo+Bev) OS (HR) 0.78 (96.02% CI 0.65–0.93) 0.66 (95% CI 0.52–0.85) 1.18 (95% CI 0.88–1.59) PFS (HR) 0.90 (95% CI 0.77–1.05) 0.65 (95% CI 0.53–0.81) 1.38 (95% CI 1.06–1.80) Grade ≥3 TRAEs 25.8% 43.0% RR 0.74 (95% CI 0.55–1.00) Discontinuation 8.2% (TRAEs) 15.5% (any AE) Descriptive only