Mesorectal fat area-based nomogram for predicting the difficulty of minimal invasive surgery in mid to low rectal cancer

Purpose : This study aims to develop a mesorectal fat area-based nomogram, covering preoperative baseline characteristics and other pelvic MRI data, to predict the difficulty of robotic or laparoscopic-assisted total mesorectal excision (TME)in patients with mid to low rectal cancer. Method : 378 patients were retrieved in our hospital and divided into non-difficult and difficult groups based on five criteria. Factors independently associated with the difficulty were identified by univariate and multivariate logistic regression analysis and then were used to develop a nomogram model to visualize the risk of surgical difficulty. Result : Tumor distance from anal verge, intertuberous distance, pelvic depth, anorectal angle and mesorectal fat area independently predicted difficulty level. A nomogram model which combines these predictors including mesorectal fat area was developed and constructed. An area under the ROC curve (AUC) of 0.8668 was obtained for the training data set and 0.9134 for the internal validation one. Discrepancy in surgical approach (laparoscopic or robotic) was not the independent predictive factor of the surgical difficulty ( P ＞0.05). Conclusions : The mesorectal fat area-based nomogram model is feasible for predicting the difficulty level of rectal surgery, combined other MRI-based pelvimetry parameters and clinical factors in mid-low RC cases.