Phase II Single-Arm Study of Durvalumab and Bevacizumab Following Transarterial Radioembolisation in Unresectable Hepatocellular Carcinoma (EMERALD-Y90): Interim Safety Analysis with Contextualisation Using an External Comparator Arm

Background/Objectives: EMERALD-Y90 (NCT06040099) is a phase II, open-label, single-arm study evaluating Y90 transarterial radioembolisation (TARE) plus durvalumab and bevacizumab for unresectable hepatocellular carcinoma (uHCC) amenable to TARE, without extrahepatic spread. Interim safety was compared with an EMERALD-1 external comparator arm (ECA). Methods: Participants received TARE with dosimetry, durvalumab (1500 mg) ≥ 14 days later, then durvalumab (1120 mg) plus bevacizumab (15 mg/kg) every 21 days from ≥ 14 days later. Safety was assessed across treatment intervals. The ECA included EMERALD-1 participants who received transarterial chemoembolisation, durvalumab and bevacizumab, and met EMERALD-Y90 eligibility. Inverse probability of treatment weighting balanced prespecified baseline characteristics before comparisons. Results: At the 9 January 2026 cut-off, 59 participants received TARE (safety set), 58 received durvalumab and 57 received bevacizumab. Treatment-emergent adverse events (TEAEs) occurred in 58 (98.3%) participants, including Grade 3/4 TEAEs in 30 (50.8%); 53 (89.8%) had treatment-related adverse events (TRAEs); 18 (30.5%) had serious adverse events and 8 died (6 unrelated to treatment). Common TEAEs were gastrointestinal (89.8%), musculoskeletal (54.2%), respiratory (54.2%), skin-related (52.5%), and nervous system-related (50.8%). After weighting, EMERALD-Y90 (N = 58) and EMERALD-1 ECA (N = 45) had balanced baseline characteristics and descriptively similar safety; TRAEs occurred in 52 (86.5%) and 40 (87.5%) participants, respectively. Conclusions: In this interim safety analysis of EMERALD-Y90, TARE followed by durvalumab and bevacizumab showed an encouraging safety profile, consistent with the known component profiles, without new safety signals versus these profiles. Observed safety was broadly similar to the EMERALD-1 ECA, supporting further investigation of this approach in embolisation-eligible uHCC.