直肠LARS危险因素分析及风险预测模型构建

ANALYSIS OF RISK FACTORS FOR LOW ANTERIOR RESECTION SYNDROME AND CONSTRUCTION OF RISK PREDICTION MODEL

目的探讨直肠恶性肿瘤保肛术后低位前切除综合征(LARS)的危险因素,并进行风险预测模型的构建及验证。方法回顾性收集并分析2019年9月—2020年9月青岛大学附属医院胃肠外科诊断为直肠癌并行直肠癌保肛手术的242例病人的临床资料,进行LARS量表评分,根据评分结果分为LARS组(包括轻度LARS及重度LARS)和无LARS组,比较两组病人临床资料的差异;应用多因素Logistic回归分析方法筛选出保肛根治术术后LARS发生的独立危险因素,以此构建Nomogram风险预测模型,通过校准曲线、一致性指数(C-index)及受试者工作特征曲线(ROC曲线)评价Nomogram模型的准确度,通过决策曲线分析(DCA)对模型进行临床有效性评价。结果LARS组病人76例,无LARS组病人166例。多因素Logistic回归分析显示,BMI≥24 kg/m^(2)(OR=3.616,95%CI=1.757~7.442,P<0.05)、肿瘤位置(OR=5.305,95%CI=2.568~10.956,P<0.05)、新辅助治疗(OR=5.082,95%CI=2.267~11.393,P<0.05)和吻合口瘘(OR=14.265,95%CI=4.706~43.244,P<0.05)是直肠恶性肿瘤保肛根治术后LARS发生的独立危险因素。利用上述危险因素构建预测术后LARS发生风险的列线图模型,训练集评价模型的C-index为0.834(95%CI=0.684~0.953),ROC曲线下面积(AUC)为0.868(95%CI=0.737~0.936);验证集评价C-index为0.893(95%CI=0.697~0.964),AUC值为0.819(95%CI=0.767~0.945);训练集和验证集的校准曲线与理想曲线校准良好,经Hosmer-Lemeshow拟合优度检验,预测模型与实际观察值之间差异无显著意义(P>0.05);DCA曲线显示了很大的正向净收益,表明其拥有良好的临床应用价值。结论本研究所构建的Nomogram预测模型对直肠癌保肛根治术后LARS的发生具有较好的预测能力。

Objective To investigate risk factors for low anterior resection syndrome(LARS)after sphincter-preserving surgery for malignant rectal tumors,and to construct and verify a risk prediction model.Methods We retrospectively analyzed the clinical data of 242 patients who had been diagnosed with rectal cancer and undergone sphincter-preserving surgery at the Department of Gastrointestinal Surgery,The Affiliated Hospital of Qingdao University from September 2019 to September 2020.Based on the LARS scale score,these patients were divided into LARS group(including mild and severe LARS)and non-LARS group.Clinical data were compared between the two groups.Multivariable Logistic regression analysis was used to determine the independent risk factors for the occurrence of LARS after radical sphincter-preserving surgery.The risk factors would be used to construct a nomogram prediction model for the risk of LARS.The accuracy of the nomogram model was assessed through the calibration curve,Harrell’s concordance index(C-index),and receiver operating characteristic(ROC)curve.Decision curve analysis(DCA)was used to evaluate the clinical value of the model.Results There were 76 patients in the LARS group and 166 patients in the non-LARS group.The multivariable Logistic regression analysis showed that body mass index≥24 kg/m^(2)(OR=3.616,95%CI=1.757-7.442,P<0.05),tumor location(OR=5.305,95%CI=2.568-10.956,P<0.05),neoadjuvant therapy(OR=5.082,95%CI=2.267-11.393,P<0.05),and anastomotic leakage(OR=14.265,95%CI=4.706-43.244,P<0.05)were independent risk factors for LARS after radical sphincter-preserving resection for rectal cancer.The performance of the nomogram mo-del derived from the above risk factors for predicting the risk of postoperative LARS was as following:for the training set,the C-index was 0.834(95%CI=0.684-0.953),and the area under the ROC curve(AUC)was 0.868(95%CI=0.737-0.936);for the validation set,the C-index was 0.893(95%CI=0.697-0.964),and the AUC was 0.819(95%CI=0.767-0.945);the calibration curvesof the training and validation sets were well calibrated to the ideal curve,and the Hosmer-Lemeshow goodness-of-fit test showed that there was no significant difference between the predicted and actual values(P>0.05);the DCA curve showed a great positive net benefit,indicating good clinical application value.Conclusion The nomogram prediction model constructed in this study has a good ability to predict the occurrence of LARS after radical sphincter-preserving resection for rectal cancer.