Establishing and validating predictive nomograms for lateral pelvic lymph node metastasis in patients with rectal cancer based on radiologic factors and clinicopathologic characteristics

Abstract

IntroductionIt is critical to accurately predict the occurrence of lateral pelvic lymph node (LPN) metastasis. Currently, verified predictive tools are unavailable. This study aims to establish nomograms for predicting LPN metastasis in patients with rectal cancer who received or did not receive neoadjuvant chemoradiotherapy (nCRT). Materials and methodsWe carried out a retrospective study of patients with rectal cancer and clinical LPN metastasis who underwent total mesorectal excision (TME) and LPN dissection (LPND) from January 2012 to December 2019 at 3 institutions. We collected and evaluated their clinicopathologic and radiologic features, and constructed nomograms based on the multivariable logistic regression models. ResultsA total of 472 eligible patients were enrolled into the non-nCRT cohort (n = 312) and the nCRT cohort (n = 160). We established nomograms using variables from the multivariable logistic regression models in both cohorts. In the non-nCRT cohort, the variables included LPN short diameter, cT stage, cN stage, histologic grade, and malignant features, and the C-index was 0.930 in the training cohort and 0.913 in the validation cohort. In the nCRT cohort, the variables included post-nCRT LPN short diameter, ycT stage, ycN stage, histologic grade, and post-nCRT malignant features, and the C-index was 0.836 in the training dataset and 0.827 in the validation dataset. The nomograms in both cohorts were moderately calibrated and well-validated. ConclusionsWe established nomograms for patients with rectal cancer that accurately predict LPN metastasis. The performance of the nomograms in both cohorts was high and well-validated.