Association Between Three-Dimensional Transrectal Ultrasound Findings And Tumor Response To Neoadjuvant Chemoradiotherapy In Locally Advanced Rectal Cancer: An Observational Study

Background There is a significant demand for the development of non-surgical methods for the evaluation of complete response to tumor therapy. Predicting ability and image quality of routine imaging has not been satisfactory. To avoid the deficiencies, we assessed the capability of three-dimensional transrectal ultrasound in predicting the response to neoadjuvant chemoradiotherapy in rectal cancer patients. Methods The inclusion criteria were patients with locally advanced rectal adenocarcinoma, receiving capecitabine-based neoadjuvant chemoradiotherapy, distance from anal verge (<= 6 cm), clinical stage T3-4 and/or N+ without evidence of distant metastasis, and restaging ycT0-3a (T3a <5 mm) after the end of neoadjuvant chemoradiotherapy. Three-dimensional transrectal ultrasound was performed 7 weeks after neoadjuvant chemoradiotherapy to discern the patients with complete response from the others. Eight main parameters were obtained from three-dimensional transrectal ultrasound: thickness of muscularis on the residual side, thickness of contralateral muscularis, angle of residual arc, regularity of the shape, integrity of the mucosal layer, blurring of the margin, internal echo, and posterior echo. The association between tumor response and three-dimensional transrectal ultrasound parameters was analyzed, and a model was developed by logistic regression. Results Between 2014 and 2019, 101 patients were recruited; 72 cases received total mesorectal excision, and 29 cases underwent watch-and-wait. Among the three-dimensional transrectal ultrasound parameters, the adjusted-thickness of the muscularis (P<0.01), angle of the residual arc (P<0.01), and regularity of the residual shape (P<0.01) were strongly associated with tumor response. In the dataset with total mesorectal excision cases (TME dataset), the residual adjusted-thickness (odds ratio [OR]=4.88, 95% confidence interval [CI]=1.44-16.6, P=0.01) and regularity of the residual shape (OR=5.00, 95% CI=1.13-22.2, P=0.03) were kept in the final logistic model. The area under the curve of the logistic model was 0.84. Among these parameters, residual adjusted-thickness correlated significantly with tumor response. Additionally, we observed similar results in the whole population of 101 cases (whole dataset) and in the cross-validation. Conclusion Three-dimensional transrectal ultrasound model is a valuable method for predicting tumor response in rectal cancer patients undergoing neoadjuvant chemoradiotherapy, which should be included as a factor for evaluating clinical complete response.