PURPOSE

The neoadjuvant rectal cancer (NAR) score is a prognostic tool for locally advanced rectal cancer (LARC) treated with total neoadjuvant therapy (TNT). It has been previously validated as an endpoint that predicts survival more accurately than pathologic complete response (pCR) and is the primary endpoint of the ongoing NRG-GI002 Phase II trial. Using the National Cancer Database (NCDB), we aimed to validate the NAR score's ability to predict survival in a large hospital-based dataset.

METHODS

We queried the NCDB to identify locally advanced rectal cancer patients from 2004 to 2015 that received TNT followed by surgical resection. Overall survival (OS) was calculated using Kaplan-Meier curves evaluating NAR score and pCR separately. A multivariable Cox proportional hazards model was used to identify factors associated with survival. Multivariate regression was used to evaluate characteristics associated with a favorable (< 14.98) NAR score.

RESULTS

From > 264,000 patients diagnosed with rectal adenocarcinoma in the NCDB, our final cohort yielded 209 patients with a median age of 62 years. Factors associated with worse survival included age > 62 years old (p = 0.04), lower income (p = 0.03), and unfavorable (≥ 14.98) NAR score (p = 0.04). On multivariate regression, tumors with perineural invasion and a higher comorbidity score (> 1) were less likely to have a favorable NAR response (p = 0.01 and p = 0.01). pCR was not associated with improved survival (p = 0.09).

CONCLUSIONS

Our study validates the NAR score as a prognostic tool in patients receiving TNT for LARC. Tumors with perineural invasion and patients with a higher comorbidity score had worse NAR scores.