Recurrence-free survival as a surrogate for overall survival in esophageal squamous cell carcinoma: Nationwide real-world data from Japan.

Abstract

269 Background: One of the limitations of the established gold standard, overall survival (OS), is that it requires an extended follow-up period. Addressing this challenge involves investigating appropriate statistically and clinically relevant surrogate endpoints. However, there is a paucity of studies using real-world data (RWD) to explore surrogacy in patients with esophageal squamous cell carcinoma (ESCC). Our study aims to investigate whether recurrence-free survival (RFS) is a valid surrogate endpoint for OS in surgically resectable advanced ESCC patients, utilizing a comprehensive nationwide database in Japan. Methods: The ESCC patients who received neoadjuvant CF (cisplatin and 5-fluorouracil) or DCF (docetaxel, cisplatin and 5-fluorouracil) at 58 Japanese esophageal centers certified by the Japan Esophageal Society were retrospectively reviewed. Kendall's tau between RFS and OS was used at the individual level, and the coefficient of determination was used at the institutional level to assess surrogacy. Additionally, surrogacy between each short-term postoperative endpoint (pathological complete response [pCR] or pathological grade) and OS was investigated using novel statistical methods. Results: Our study encompassed 3154 ESCC patients who underwent subtotal esophagectomy from 2010 to 2015. The cStage I/II/III/IVA/IVB (due to supraclavicular lymph node metastasis) was 224 (7.1%), 1008 (32.0%), 1664 (52.8%), 117 (3.7%), and 140 (4.4%), respectively. The 5-year OS and RFS rate for the entire cohort was 56.6% and 47.7%, respectively. In the primary analysis, a strong correlation between RFS and OS was found (Kendall's tau = 0.797, 95% confidence interval [CI]: 0.782 to 0.812) at the individual level. Subgroup analysis revealed that the better the pathological response, the higher the tau value. An adjusted R2 of 100.0% (95% CI: 40.2 to 100.0) was obtained with the meta-regression model at the institutional level. The surrogate threshold effect was 0.703. Tau between pCR or pathological grade and OS were -0.025 (95% CI: -0.335 to 0.285) and 0.062 (95% CI: -0.086 to 0.209), respectively. Conclusions: The present study was the first nationwide RWD investigation to demonstrate a strong correlation between RFS and OS in surgically resectable ESCC patients who underwent neoadjuvant chemotherapy (NAC). Notably, this correlation was more pronounced in patients who had a more effective response to NAC. These findings hold promise for expediting the development of novel neoadjuvant treatment by shortening the duration of clinical trials.