Efficacy and safety of first-line PD-1/PD-L1 inhibitor–based therapy in esophageal squamous cell carcinoma: A systematic review and meta-analysis.

Abstract

113 Background: First-line treatment of esophageal squamous cell carcinoma (ESCC) has evolved with PD-1/PD-L1 inhibitor–based regimens, but survival benefit estimates vary across trials due to differences in checkpoint agents, chemotherapy backbones, and PD-L1 subgroup definitions. Additional phase III trials and updated follow-up data now available prompted this updated meta-analysis to re-estimate efficacy with greater precision. Methods: PubMed, Embase, Scopus, and the Cochrane Library were searched from inception through 2 February 2026 for randomized controlled trials comparing PD-1/PD-L1 inhibitor–based therapy versus chemotherapy in first-line ESCC. Eligible trials reported OS and/or PFS as hazard ratios (HRs) with 95% CIs. Two reviewers independently screened and extracted data. Pooled HRs were calculated using random-effects models; heterogeneity was quantified by I² and τ². Publication bias was assessed by funnel plot. Robustness was evaluated using influence diagnostics (studentized residuals, DFFITS, Cook's distance, covariance ratio), Baujat, radial (Galbraith), and GOSH analyses. Results: Nine randomized phase III trials contributed 13 HR comparisons (N=7,713). PD-1/PD-L1 inhibitor–based therapy improved OS (pooled HR 0.70, 95% CI 0.66–0.74), with all individual comparisons favoring immunotherapy (HR range 0.58–0.77). PFS was also improved (pooled HR 0.64, 95% CI 0.60–0.68); most comparisons were statistically significant (HR range 0.56–0.67), with one estimate crossing unity (HR 0.81, 95% CI 0.64–1.03). Funnel plots showed no clear small-study effect. Influence diagnostics, Baujat/radial plots, and GOSH analyses confirmed the pooled estimates were stable and not driven by any single comparison. Conclusions: PD-1/PD-L1 inhibitor–based first-line therapy significantly improved OS and PFS in ESCC across 13 trial comparisons, with consistent benefit and stable pooled estimates. Further trials are needed to refine patient selection and optimize regimen choice across biomarker subgroups.