Neoadjuvant rectal-tumor regression grade combined score as surrogate endpoint for disease-free survival in locally advanced rectal cancer patients after neoadjuvant chemoradiotherapy

Abstract

Background: Existing prognostic models, such as tumor regression grade (TRG) and neoadjuvant rectal (NAR) score, have been validated as important indicators for assessing the efficacy of neoadjuvant therapy in locally advanced rectal cancer (LARC) and predicting disease-free survival (DFS). However, both models have inherent limitations in prognostic prediction. This study aims to construct a composite NAR-TRG score to predict DFS in LARC patients treated with chemoradiotherapy (CRT) followed by radical surgery.

Patients and methods: A total of 952 consecutive LARC patients between December 2010 and July 2018 at Sun Yat-sen University Cancer Center and Fujian Cancer Hospital, were enrolled in this study. After calculating the NAR score, patients were categorized into NAR low, medium, and high groups; TRG was dichotomized into TRG low and high groups; the NAR-TRG combined score was then determined based on both NAR and TRG groupings. Survival outcomes were analyzed using Kaplan-Meier, Cox regression. Nomograms were developed to forecast patient DFS, with the area under the curve values of time-dependent receiver operating characteristic (timeROC) and c-index utilized to assess the accuracy and reliability of the nomograms.

Results: Significant differences in 5-year DFS were observed among the NAR-TRG score from 1 to 3 (91.4% vs 79.9% vs 72.3%, P < .001). NAR-TRG score was identified as an independent predictor of DFS in multivariate analysis (HR = 1.577, 95% CI: 1.298-1.915, P < .001). The comparison of timeROC AUCs revealed that the NAR-TRG score consistently outperformed both the NAR score and TRG group at various time points (Main cohort: NAR-TRG score vs TRG, P = .002; NAR-TRG score vs NAR, P = .002; Validation cohort: NAR-TRG score vs TRG, P = .003; NAR-TRG score vs NAR, P = .002). The nomogram model including the NAR-TRG score demonstrated a superior c-index and area under the timeROC for DFS compared to models excluding the NAR-TRG score both in the main cohort and validation cohort.

Conclusions: The NAR-TRG score effectively stratifies LARC patients receiving neoadjuvant CRT, which can serve as a surrogate endpoint for DFS, contributing to the optimization of decisions related to postoperative therapy and subsequent follow-up strategies.

Keywords: disease-free survival; neoadjuvant rectal score; preoperative chemoradiotherapy; rectal cancer; tumor regression grade.

Figure 1.

Relationship Between TRG and NAR. *P < .05, **P ≤ .01, ***P < .001.

Figure 2.

Kaplan–Meier survival analysis of DFS in LARC patients (A) based on different subgroups based on TRG and NAR score classifications. (B) categorized by NAR-TRG score in the main cohort. (C) categorized by NAR-TRG score in the validation cohort.

Figure 3.

Comparison of disease-free survival-related time-dependent areas under the curve (AUCs) among the NAR-TRG score, NAR, and TRG (A) in the main cohort (B) in the validation cohort.

Figure 4.

Model performance in DFS prediction by incorporating the NAR-TRG score. (A) Nomogram model construction based on NAR-TRG score, tumor differentiation, and baseline CA19-9 level for predicting 3-year and 5-year DFS. (B) Comparison of c-index with 95% CI in predicting DFS in nomogram model with NAR-TRG score, without NAR-TRG score, NAR alone, and TRG alone in the main cohort. (C) Comparison of c-index with 95% CI in predicting DFS in nomogram model with NAR-TRG score, without NAR-TRG score, NAR alone, and TRG alone in the validation group. (D) Comparison of time-dependent ROC curves for 3-year DFS prediction between the model with and without NAR-TRG score in the main cohort. (E) Comparison of time-dependent ROC curves for 5-year DFS prediction between model with and without NAR-TRG score in the main cohort.