Background
Phase III double-blind RCT, N=794, patients with unresectable HCC, Child-Pugh A, ECOG PS 0–1, no prior systemic therapy. Enrolled at 172 global sites Jan 2019–Apr 2020. Randomized 1:1. Stratified by region, macrovascular invasion/extrahepatic spread, AFP, and ECOG PS. Dual primary endpoints OS and PFS (prespecified one-sided thresholds: OS P≤.019 at final analysis; PFS P≤.002).
Interventions and follow up
Arm A: Lenvatinib (8mg/day if <60kg, 12mg/day if ≥60kg) + pembrolizumab 200mg q3wk
Arm B: Lenvatinib (same dosing) + placebo q3wk
Primary endpoint: OS and PFS (co-primary)
Median follow up: 32.1mo
Arm B: Lenvatinib (same dosing) + placebo q3wk
Primary endpoint: OS and PFS (co-primary)
Median follow up: 32.1mo
Results
OS: 21.2 vs 19.0mo, HR 0.84 (95% CI 0.71–1.00), P=.023 — prespecified threshold P≤.019 not met
PFS: 8.2 vs 8.0mo, HR 0.87 (95% CI 0.73–1.02), P=.047 — prespecified threshold P≤.002 not met
PFS: 8.2 vs 8.0mo, HR 0.87 (95% CI 0.73–1.02), P=.047 — prespecified threshold P≤.002 not met
Adverse events
Cardiovascular/GI: Grade 3–4 hypertension 17% vs 17% (lenva+pembro vs lenva+placebo); diarrhoea 6% vs 4%.
Hepatic/fatal: Grade 3–4 increased AST 7% vs 4%; treatment-related deaths 4 (1%) vs 3 (1%).
Hepatic/fatal: Grade 3–4 increased AST 7% vs 4%; treatment-related deaths 4 (1%) vs 3 (1%).
Conclusions
Adding pembrolizumab to lenvatinib did not statistically improve OS or PFS compared with lenvatinib alone as first-line therapy for advanced HCC. Despite promising phase 1b data (ORR ~46%), LEAP-002 failed to demonstrate added benefit in a randomised setting, likely due to an active comparator (lenvatinib) that is itself highly active as monotherapy.
Key Limitations
An active comparator (lenvatinib + placebo) rather than true placebo limited the absolute OS ceiling. The lenvatinib monotherapy arm performed better than historical sorafenib-era comparators (mOS 19.0 months), potentially attenuating the incremental benefit of adding pembrolizumab. The near-significant OS HR (0.84, P=.023 vs threshold .019) raises the possibility of a marginal benefit not detected due to power constraints. Phase 1b ORR data were not from a randomized comparison.
Clinical Context
LEAP-002 joins a pattern of negative first-line/second-line trials in HCC: LEAP-002 (lenva+pembro), CheckMate 459 (nivo vs sora, OS negative), and KEYNOTE-240 (pembro 2L, negative) all failed prespecified thresholds. The field has consolidated around atezo+bev (IMbrave150), STRIDE/durva+treme (HIMALAYA), and nivo+ipi (CheckMate 9DW) as positive IO regimens. Lenvatinib monotherapy remains a standard first-line option based on REFLECT.