108P - A meta-analysis of PD-1 inhibitors for the treatment of advanced esophageal squamous cell carcinoma in second-line setting

Background

The purpose of this research was to meta-analyze the survival data using hazard ratios (HR) and Kaplan-Meier (KM) curves available for PD-1 inhibitors versus chemotherapy among advanced or metastatic esophageal squamous cell carcinoma (ESCC) patients receiving second-line treatment.

Methods

A systematic search using Embase and MEDLINE was performed from database inception to July 2022 to identify randomized controlled trials (RCTs) assessing PD-1 inhibitors against chemotherapy in ESCC patients. The direct meta-analysis of HRs was performed using Sidik-Jonkman (SJ) and DerSimonian and Laird (DL) models (random effect). The KM curves were digitized and were used to generate pseudo-individual-level patient data (IPD) using method described by Guyot et al. The IPD data was checked for accuracy by plotting the resulting KM curves against the coordinates from the published graphs. The meta-analysis was performed using the metaSurv package in R. To account for the between-study heterogeneity in the estimation of the pooled conditional survival probabilities, a recent extension of DL methodology for multiple outcomes was used.

Results

The SLR included five RCTs (1970 patients) assessing pembrolizumab (KEYNOTE-181), Nivolumab (ATTRACTION-3), Sintilimab (ORIENT-2), Tislelizumab (RATIONALE-302), and Camrelizumab (ESCORT) versus investigator choice chemotherapy. Second-line PD-1 inhibitors significantly improved the OS (SJ=HR: 0.73, 95% CI: 0.65-0.80; DL= HR: 0.73, 95% CI: 0.66-0.80) compared to conventional chemotherapy among advanced ESCC patients. The results of the Asia-specific subgroup analysis were aligned with the basecase results (SJ&DL=HR: 0.72, 95% CI: 0.64-0.80). The meta-analysis using pseudo-IPD extracted from Kalpan-Meier curves also favored PD-1 inhibitors.

Conclusions

The second-line PD-1 inhibitors significantly improved the OS using both conventional and KM curves-based meta-analysis. The methodology applied for the pooling of KM curves can be extremely useful to validate the conventional HR-based analysis or where the conventional analysis is not feasible.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

Pharmacoevidence Pvt. Ltd.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.