Neoadjuvant rectal cancer (RC) score to predict survival: Potential surrogate endpoint for early-phase trials.

Abstract

384 Background: Valentini, et al. (J Clin Oncol 29: 2011, 3163-72) developed nomograms for predicting overall survival (OS) based on clinical factors available after neoadjuvant therapy (tx). Pathologic T-stage (pT), N-stage (pN), and clinical T (cT) were the most important independent predictors of OS. We developed a neoadjuvant RC score (NAR score) using pN and downstaging of T (cT – pT) based on relative weights suggested by the nomograms. NSABP’s R-04 trial presents an opportunity for independent validation of the NAR score. Methods: Pts with clinical stage II/III RC undergoing preoperative RT (4,500cGy in 25 fractions over 5 wks + boost of 540-1,080cGy in 3-6 daily fractions) were randomized to one of four regimens in a 2x2 design: CVI 5-FU (225mg/m2 5 days/wk), with or without oxaliplatin (Ox) (50mg/m2 /wk x 5) or oral capecitabine (825 mg/m2 BID 5 days/wk), with or without Ox. The NAR score is computed as [5 pN – 3 (cT – pT) + 12]2/ 9.61 where: cT in {1, 2, 3, 4}, pT in {0, 1, 2, 3, 4}, and pN in {0, 1, 2}. The NAR score takes values from 0 to 100; higher scores indicate poorer prognosis. Analyses based on the score should be stratified by cT. NAR score is compared to pathologic complete response (ypCR) by Akaike’s information criterion (AIC) to determine the better predictor of OS. Results: 1,479 pts had data for the NAR score and follow-up for OS. Continuous NAR score was significantly associated with OS (HR/unit 1.04 95% CI 1.03-1.05, p<0.0001). Pts were grouped into low, intermediate, and high risk of death categories based on tertiles of the NAR score. Categories were significantly associated with OS (p<0.0001) with 5yr OS values of 92, 89, and 68%, respectively. ypCR was also significantly associated with OS (HR 0.37 95% CI 0.24-0.58, p<0.0001). Continuous NAR score had a much lower AIC (2,902.49) than ypCR (2,987.08) suggesting stronger association with OS. Conclusions: The NAR score has been validated as a predictor of OS after neoadjuvant tx for RC. AIC of the continuous NAR score suggests it is a better predictor of OS than dichotomous ypCR. NAR score could be used in place of ypCR as a surrogate endpoint for OS in early phase neoadjuvant rectal trials. Support: U10-CA-12027, U10-CA-37377, U10-CA-69651, U-10-CA-69974; Sanofi; Roche. Clinical trial information: NCT00058474.