PD-1/PD-L1 inhibitors as second-line therapy for advanced, unresected esophageal squamous-cell carcinoma: A series of patient-level meta-analyses.

Abstract

e16052 Background: Patients with esophageal squamous cell carcinoma (ESCC) who fail definitive treatment are candidates for second-line therapy. They often present with more aggressive disease and suboptimal responses to definitive treatments. In this individual patient-level meta-analysis, we aim to evaluate the efficacy of PD-1/PD-L1 inhibitors as second-line treatment in advanced ESCC. Methods: A systematic search of PubMed, Scopus, and Cochrane databases (December 5, 2024) identified randomized controlled trials (RCTs) comparing PD-1/PD-L1 inhibitors to chemotherapy as second-line treatment for advanced ESCC. Kaplan-Meier (KM) curves were reconstructed, and individual patient data (IPD) were extracted using R scripts. A patient-level meta-analysis was performed using the reconstructed IPDs and the Cox proportional hazards regression model. The Schoenfeld residuals test and Restricted Mean Survival Time (RMST) were also applied. The following R packages were used: MASS, Splines, survival, survRM2, survminer, and dbplyr . Results: Out of 3,930 records, five RCTs met the criteria: four phase 3 trials and one phase 2 trial (ORIENT-2), all were open-label, comparing PD-1 inhibitors to chemotherapy. Overall survival (OS) improved significantly with PD-1/PD-L1 inhibitors compared to chemotherapy (HR: 0.718, 95% CI: 0.650–0.793, p < 0.001). For patients with CPS ≥ 10, OS further improved (HR: 0.626, 95% CI: 0.446–0.878, p = 0.007), ), indicating an additional beneft in those with high PD-1 expression. The proportional hazards assumption was met for both OS and CPS ≥ 10 (Schoenfeld residuals test: p > 0.05). Progression-free survival (PFS) showed no significant difference (HR: 0.913, CI: 0.828–1.007, p = 0.069). However, the proportional hazards assumption was violated (Schoenfeld residuals test p < 0.001); therefore, these results should be interpreted with caution. PFS was further evaluated using RMST at 6, 12, 18, and 24 months. At 6 months, no significant difference was observed (RMST difference: -0.047, p = 0.593). However, PD-1 inhibitors showed significant differences at 12 months (0.37 months, p = 0.022), 18 months (0.74 months, p = 0.001), and 24 months (0.97 months, p < 0.001). Despite these significant differences, the magnitude of the difference is relatively small and may not be clinically meaningful. Conclusions: PD-1/PD-L1 inhibitors as second-line therapy improve overall survival in patients with advanced esophageal squamous cell carcinoma, compared to chemotherapy. However, the improvements in progression-free survival (PFS) were modest, and the clinical significance of these PFS benefits may be limited.