Comparison of Neoadjuvant Rectal Score and MRI-Based Nomogram Model as Disease-Free Survival Predictors for Locally Advanced Rectal Cancer Patients

Abstract

To compare the prognostic value of neoadjuvant rectal (NAR) score and MRI-based nomogram model for locally advanced rectal cancer (LARC) treated with neoadjuvant therapy.Retrospective analysis was performed for 233 patients with LARC (T3-4 and/or N1-2, M0), who were initially treated in our institution from Mar.2015 to Mar.2018. All patients had baseline MRI assessments and received neoadjuvant therapy and TME surgery. The neoadjuvant rectal (NAR) score was calculated according to the baseline MRI factor (cT) and pathological stage (pT, pN). NAR = [5pN-3(cT-pT)+12]2/9.61. The patients were sequentially allocated to two cohorts (training cohort and validation cohort) in a ratio of 4:3 based on the date of image examination. The nomogram model was constructed based on the results of univariate logistic regression analysis and multivariable cox regression analysis of the training cohort for disease-free survival (DFS). The nomogram and NAR score were evaluated by Harrell's concordance index (C-index), calibration plots, receiver operating characteristic curve (ROC) analysis, and decision curve analysis (DCA) in both training and validation cohort.With a median follow-up of 43.2 months (13.3-61.3months) in the training cohort and 32.0 months (12.3-39.5 months) in the validation cohort. Multivariate cox regression analysis identified the MRI-detected extramural vascular invasion (mrEMV), pathological T stage (ypT) and perineural invasion (PNI) as independent predictors. According to previous studies, lymph vascular invasion (LVI) (which was founded to be significant after univariate regression analysis) and three other independent predictors are included in the nomogram model. The univariate survival analysis showed that the nomogram model and NAR score were correlated with 3-year DFS (P < 0.05). For comparison, the nomogram performed better over the NAR score (C-index 0.76 (95% CI 0.70-0.84) versus 0.66 (95% CI 0.60-0.71) for the training cohort, and 0.77 (95% CI 0.70-0.85) versus 0.69(0.60-0.78) for the validation cohort).The nomogram model achieved a better 3-year DFS predictive capacity (AUC = 0.84 in the training cohort; AUC = 0.77 in the validation cohort) than NAR score (AUC = 0.70 in the training cohort; AUC = 0.73 in the validation cohort). DCA revealed that the use of the nomogram model was associated with better benefit gains relative to the prediction of 3-year DFS compared to NAR score in training cohort (threshold probability range between 0.06 and 0.54) and validation cohort (threshold probability range between 0.08 and 0.50).We developed and validated a novel nomogram model based on MRI factor and pathological factors for predicting DFS in LARC treated with neoadjuvant therapy. Compared with NAR score, nomogram model shows the better potential predictive value of prognosis, which could improve the risk stratification and individual treatment for LARC patients.