Evaluation of event-free survival (EFS) as a surrogate for overall survival (OS) in patients with locally advanced esophageal carcinoma (LA EC) receiving definitive chemoradiotherapy (dCRT).

412 Background: Drug development in esophageal cancer is rapidly evolving, and identifying surrogates for patient outcomes to accelerate clinical drug development remains an important unmet medical need. Although early clinical endpoints such as EFS, disease free survival (DFS) are proven to be valid surrogates for overall survival (OS) in early esophageal (Ajani JA et al. 2022), gastric (Oba K et al. 2013; Wainberg ZA et al. 2023), and gastroesophageal junction tumors (Ronellenfitsch U. et al. 2013) after disease resection, there are no such surrogacy data evaluating patients who underwent non-operative management. Specifically, there has been no prior systemic evaluation of the predictive value of EFS as a surrogate for OS among LA EC patients treated with dCRT. This study evaluated the trial-level correlation between EFS and OS in unresectable LA EC patients receiving dCRT. Methods: A systematic literature review was conducted to identify randomized controlled trials (RCTs) of dCRT interventions in unresectable LA EC (search cutoff date: April 2023). Studies reporting EFS (or similar outcomes such as progression free survival (PFS), defined as time from randomization to local, regional, or distant recurrence, or death) and OS were included. EFS and OS hazard ratios (HRs) of each treatment comparison were extracted. Correlation between the log(HR) of EFS and OS was assessed using weighted linear regression (WLR). The coefficient of determination (R 2 ) was used to evaluate the strength of the correlation. A good surrogate was defined as an R 2 ≥ 0.65 (Ciani et al . 2017) or an R 2 ≥ 0.75 (Lassere et al . 2008). Leave-one-out cross-validations and sensitivity analyses were conducted to assess the robustness of the WLR model and correlation analyses, respectively. Results: The final analysis included 14 between-treatment comparisons from 11 RCTs (total number of patients = 2,812). We found a strong statistical correlation between EFS and OS, with an estimated slope coefficient of 0.91 ( p < 0.001) and R 2 of 0.80 (95% confidence interval [CI]: 0.38, 0.96). In the leave-one-out cross validation, 71.4% of observed OS HRs fell within the predicted 95% CIs. Sensitivity analyses showed R 2 of 0.75 among studies conducted in China and R 2 of 0.73 after including studies that did not report EFS definition. Conclusions: In a pooled analysis of 11 clinical trials corresponding to over 2800 patients, we observed a strong statistical correlation between EFS and OS outcomes. This analysis supports the utility of EFS as a valid surrogate of OS in patients with unresectable LA EC receiving dCRT.