RESUMO
(1) Background: Numerous variables could influence the risk of rectal cancer recurrence or metastasis, and machine learning (ML)-based algorithms can help us refine the risk stratification process of these patients and choose the best therapeutic approach. The aim of this study was to assess the predictive performance of 4 ML-based models for the prediction of local recurrence or distant metastasis in patients with locally advanced low rectal adenocarcinomas who underwent neoadjuvant chemoradiotherapy and surgical treatment; (2) Methods: Patients who were admitted at the first Oncologic Surgical Clinic from the Regional Institute of Oncology, Iasi, Romania were retrospectively included in this study between November 2019 and July 2023. Decision tree (DT), naïve Bayes (NB), support vector machine (SVM), and random forest (RF) were used to analyze imagistic, surgical, and pathological data retrieved from the medical files, and their predictive performance was assessed; (3) Results: The best predictive performance was achieved by RF when used to predict disease recurrence (accuracy: 90.85%) or distant metastasis (accuracy: 89.63%). RF was closely followed by SVM (accuracy for recurrence 87.8%; accuracy for metastasis: 87.2%) in terms of predictive performance. NB and DT achieved moderate predictive power for the evaluated outcomes; (4) Conclusions: Complex algorithms such as RF and SVM could be useful for improving the prediction of adverse oncological outcomes in patients with low rectal adenocarcinoma.
RESUMO
Background and Objectives: A positive pathological circumferential resection margin is a key prognostic factor in rectal cancer surgery. The point of this prospective study was to see how well different MRI parameters could predict a positive pathological circumferential resection margin (pCRM) in people who had been diagnosed with rectal adenocarcinoma, either on their own or when used together. Materials and Methods: Between November 2019 and February 2023, a total of 112 patients were enrolled in this prospective study and followed up for a 36-month period. MRI predictors such as circumferential resection margin (mCRM), presence of extramural venous invasion (mrEMVI), tumor location, and the distance between the tumor and anal verge, taken individually or combined, were evaluated with univariate and sensitivity analyses. Survival estimates in relation to a pCRM status were also determined using Kaplan-Meier analysis. Results: When individually evaluated, the best MRI predictor for the detection of a pCRM in the postsurgical histopathological examination is mrEMVI, which achieved a sensitivity (Se) of 77.78%, a specificity (Sp) of 87.38%, a negative predictive value (NPV) of 97.83%, and an accuracy of 86.61%. Also, the best predictive performance was achieved by a model that comprised all MRI predictors (mCRM+ mrEMVI+ anterior location+ < 4 cm from the anal verge), with an Se of 66.67%, an Sp of 88.46%, an NPV of 96.84%, and an accuracy of 86.73%. The survival rates were significantly higher in the pCRM-negative group (p < 0.001). Conclusions: The use of selective individual imaging predictors or combined models could be useful for the prediction of positive pCRM and risk stratification for local recurrence or distant metastasis.