Abstract / Summary
To quantify overall-survival effects of immune-checkpoint-inhibitor (ICI)-containing regimens by treatment line and to evaluate whether clinical characteristics or PD-L1 definitions modify relative treatment benefit in advanced esophageal squamous cell carcinoma (ESCC). PubMed, Embase, the Cochrane Central Register of Controlled Trials, and Web of Science were searched from inception through 1 August 2026. Two reviewers independently screened records, assessed full texts, extracted data, and evaluated risk of bias. Reports were linked to their parent randomized trial family. First-line single-checkpoint blockade plus chemotherapy and later-line checkpoint monotherapy were synthesized separately using REML random-effects models and modified Hartung-Knapp inference. Treatment-effect modification was quantified using within-trial ratios of hazard ratios (RHRs). Twenty-one interactions with at least two independent trial families entered Holm correction; Bonferroni correction was used as a sensitivity analysis. We identified 14 independent randomized trial families represented by 31 reports. The primary overall-survival analysis included 7 first-line trials (N = 4,151; pooled HR 0.70, 95% CI 0.64-0.77; P < 0.001; I 2 = 0%) and 5 later-line trials (N = 1,970; pooled HR 0.74, 95% CI 0.64-0.85; P = 0.004; I 2 = 0%). Of 30 post hoc line-specific interaction hypotheses, 28 had at least one eligible trial-family contrast: 21 met the k ≥ 2 pooling threshold and were included in multiplicity-adjusted analyses, 7 were single-trial descriptive contrasts, and 2 had no eligible estimate. None of the 21 pooled interactions remained significant after Holm correction; all Holm- and Bonferroni-adjusted P values were 1.000. The smallest unadjusted P value was 0.092 for first-line CPS ≥ 1 versus < 1. ICI-containing regimens improved overall survival in both first- and later-line settings, but aggregate within-trial interaction analyses did not identify a reliable clinical or PD-L1-defined treatment-effect modifier. Sparse subgroup reporting and the predominantly Asian evidence base limit the precision and generalizability of these findings. https://www.crd.york.ac.uk/PROSPERO/, identifier CRD42024551725.