Prediction Of Anastomotic Leakage After Anterior Rectal Resection:the LASSO Method

Background: Due to changes in eating habits,colorectal cancer is becoming the most common malignant tumor of the digestive tract in the world.There are approximately 2 million new cases of colorectal cancer in the world each year,of which approximately 600,000 patients die from colorectal cancer.With the low rectal cancer resection(LAR)becomes the standard procedure,the incidence of postoperative complications,including postoperative anastomotic leakage(AL),Postoperative sexual dysfunction and other surgical site infections.It may cause some serious consequences such as pelvic abscesses,peritonitis,sepsis,septic shock and even death if it is not treated properly.Therefore,recognition of the incidence and risk factors related to postoperative complications after anterior rectal resection is essential for both the informed consent provided to the patients and the safe performance of this procedure.And the pathogenesis of anastomotic leakage is not clear,so anastomotic leakage has always been difficult to accurately predict,and there is no uniform standard or consensus for the surgeon to choose whether to prevent the patient's prophylactic stoma,this fact to the colorectal surgeon Seriously troubled.In the past,they always decided whether to perform prophylactic stoma for this patient based on their own clinical experience combined with intraoperative conditions and whether the patient was at high risk.However,decisions based solely on the clinician's own clinical experience have no way to be scientific and standardized.This will lead to two inevitable serious consequences,including some patients who do not need stoma originally suffer the pain of colon or ileostomy surgery and the re-stomach rectal surgery,and some high-risk groups,who had not undergone reflux stoma,had anastomotic leaks.Objective: This study will use the LASSO regression model to accurately predict the occurrence of anastomotic leakage after rectal cancer.Materials and methods: Data from 188 patients who underwent anterior resection of rectal cancer performed by the same surgical team in our hospital from 2010 to 2016 were collected for retrospective statistical analysis.To build the prediction model,we first randomly selected half of the samples as the training data set.We included the predictors: age,sex,preoperative chemoradiotherapy,tumor size,degree of tumor differentiation,stage,TNM stage,lymph vascular invasion,distance,anastomotic method,diabetes,intraoperative time,intraoperative bleeding,smoking as the candidates for variable selection.The variables : preoperative chemoradiotherapy,degree of tumor differentiation,anastomotic method,distance,intraoperative bleeding,smoking and diabetes were selected in the LASSO model.We used the remaining samples as the validation data set to construct the ROC curves to verify the accuracy of the prediction model.Results:The study included 188 patients who had undergone DIXON surgery without prophylactic colostomy.Anastomotic leakage took place in 20 of188 patients(10.6%).The retrospective analysis showed that preoperative chemoradiotherapy(p=0.044715),middle differentiation(p=0.039485),anastomotic method(p=0.007199),intraoperative bleeding ?400ml(p=0.001071),smoking(p=0.000869),diabetes(p=0.000299)were significantly related to the anastomotic leakage.The area under the ROC curve of the prediction model of anastomotic leakage is 0.952.Conclusion: This study used the LASSO model to screen out the independent influencing factors of anastomotic leakage after anterior rectal resection,which helps to accurately manage postoperative anastomotic leakage.