Background: We investigated tumor regression grading (TRG) as a prognostic marker and individual-level surrogate for disease-free survival (DFS) in patients with rectal carcinoma treated within the Chirurgische Arbeitsgemeinschaft fur Onkologie/Arbeitsgemeinschaft Radiologische Onkologie/Arbeitsgemeinschaft Internistische Onkologie (CAO/ARO/AIO)-04 randomized trial.
Methods: TRG was recorded prospectively using the Dworak classification in 1179 patients after preoperative fluorouracil-based chemoradiotherapy (CRT) with or without oxaliplatin. Multivariable analysis was performed using Cox regression models adjusted for treatment arm, resection status, and pathologic stage. Individual-level surrogacy of TRG for DFS was examined using the four Prentice criteria (PC1-4). All statistical tests were two-sided.
Results: With a median follow-up of 50 months, the addition of oxaliplatin to fluorouracil-based CRT led to statistically significantly improved three-year DFS (75.9%, 95% CI = 72.3 to 79.5, vs 71.3%, 95% CI = 67.6 to 74.9, P = .04, PC 1) and a shift toward more advanced TRG groups ( P < .001, PC 2) compared with CRT with fluorouracil alone. The three-year DFS was 64.6% (95% CI = 57.3 to 71.9), 77.6% (95% CI = 74.5 to 80.7), and 92.3% (95% CI = 88.4 to 96.2) for TRG 0 + 1 (poor regression), TRG 2 + 3 (intermediate regression), and TRG 4 (complete regression), respectively ( P < .001, PC 3). TRG constituted an independent prognostic factor for DFS (TRG 2 + 3 vs TRG 0 + 1, HR = 0.68, 95% CI = 0.51 to 0.90, P = .007). Due to multicollinearity, TRG 4 and pathologic stage could not be tested within the same model. The treatment effect on DFS was captured by TRG, satisfying individual-level PC4.
Conclusions: Higher TRG after preoperative CRT predicted a favorable long-term outcome. At the individual patient level, TRG was a surrogate marker for DFS. Further phase III trials are needed to validate TRG as a surrogate at trial level.
© The Author 2017. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: email@example.com.