Disease-free survival as surrogate for overall survival in esophageal cancer: An individual patient data meta-analysis of neoadjuvant chemotherapy and chemoradiotherapy

BACKGROUND: The use of surrogate endpoints may expedite the reporting of study outcomes of clinical trials. The validity of disease-free survival (DFS) as a surrogate for overall survival (OS) in the neoadjuvant treatment of esophageal (E) or gastroesophageal junctional (GEJ) carcinomas remains uncertain. OBJECTIVE: To evaluate DFS as a surrogate end-point for OS in E/GEJ using the meta-analytical approach DESIGN, SETTING, AND PARTICIPANTS: individual patient data from an international meta-analysis on operable locally advanced E/GEJ, which including randomized trials comparing at least two of the neo-adjuvant treatment strategies: upfront surgery (S), chemotherapy followed by surgery (CS), and/or chemoradiotherapy followed by surgery (CRS). MAIN OUTCOMES AND MEASURES: Individual (Kendall's tau) and trial-level (R2) correlations between DFS and OS were estimated using a Clayton copula. RESULTS: DFS and OS data were available for a total of 4518 pts: 2222 pts included in CS vs S, 1908 pts in CRS vs S, and 388 in CS vs CRS comparisons. 3440 patients had a DFS event and 3303 patients died. Kendall's tau was 0.73 [95 % CI 0.71 - 0.75] and R2 trial-level correlation was 0.95 [0.84 - 0.99] for CS vs S, Kendall's tau was 0.76 [0.74 - 0.77] and R2 was 0.96 [0.87 - 0.99] for CRS vs S, Kendall's tau was 0.87 [0.78 - 0.92] and R2 was 0.93 [0.43 - 1] for CRS vs CS. In a multistate model, the median time in the recurrence state was shorter in older vs more recent trials: mean time of 10.8 [10.2 - 11.4] vs 16.5 months [15.4-17.6]. CONCLUSIONS AND RELEVANCE: DFS is a validated surrogate endpoint for OS in trials evaluating neoadjuvant chemotherapy or chemoradiotherapy in E/GEJ. DFS may be more useful as an endpoint when delays between recurrences and death become larger.