Improvement in the Assessment of Response to Preoperative Chemoradiotherapy for Rectal Cancer Using Magnetic Resonance Imaging and a Multigene Biomarker

Abstract

Preoperative chemoradiotherapy is now the gold standard for treating locally advanced rectal cancer and has been demonstrated to decrease local recurrence and promote sphincter preservation. Therefore, developing methods to accurately predict patients’ response to chemoradiation is imperative for choosing the best surgical option after chemoradiation and predicting the oncologic outcomes of patients. Radiological tools and endoscopy are the most commonly used tools for post-treatment response assessment. In addition, examining the expression levels of genes correlated with treatment response could provide clinicians with more power to gauge each patient’s potential to respond. In this study, we explored how biological and radiological tools can be used together to provide a more tailored and multidimensional representation of patients’ response status.The response to preoperative chemoradiotherapy (PCRT) is correlated with oncologic outcomes in patients with locally advanced rectal cancer. Accurate prediction of PCRT response before surgery can provide crucial information to aid clinicians in further treatment planning. This study aimed to develop an evaluation tool incorporating a genetic biomarker and magnetic resonance imaging (MRI) to improve the assessment of response in post-CRT patients with locally advanced rectal cancer. A total of 198 patients who underwent PCRT followed by surgical resection for locally advanced rectal cancer between 2010 and 2016 were included in this study. Each patient’s response prediction index (RPI) score, a multigene biomarker developed in our previous study, and magnetic resonance tumor regression grade (mrTRG) score were added to create a new predictive value for pathologic response after PCRT, called the combined radiation prediction value (cRPV). Based on the new value, 121 and 77 patients were predicted to be good and poor responders, respectively, showing significantly different cRPV values (p= 0.001). With an overall predictive accuracy of 84.8%, cRPV was superior to mrTRG and RPI for the prediction of pathologic chemoradiotherapy response (mrTRG, 69.2%; RPI, 77.3%). In multivariate analysis, cRPV was found to be the sole predictor of tumor response (odds ratio, 32.211; 95% confidence interval, 14.408–72.011;p= 0.001). With its good predictive value for final pathologic regression, cRPV may be a valuable tool for assessing the response to PCRT before surgery.