建立预测直肠癌术后前切除综合征的列线图模型

目的:分析腹腔镜直肠癌低位前切除术后发生前切除综合征（low anterior resection syndrome，LARS）的独立危险因素，并构建列线图预测模型。方法:采用回顾性病例对照研究，纳入并分析210例患者的临床病理资料。根据症状评分将患者分为LARS组和无LARS组，对临床资料进行单因素和多因素Logistic回归分析。将筛选出的LARS独立危险因素引入R软件，构建LARS发生概率的列线图预测模型。收集模型建立后的病例81例，采用时段验证法进行外部验证；模型的区分度通过计算C-index评价；模型的一致性通过计算校正后的C-index，并使用Hosmer-Lemeshow检验评价。结果:单因素分析中，新辅助治疗、体质量指数（BMI）、肿瘤下缘距肛缘距离、吻合口瘘与术后发生LARS相关；多因素Logistic分析显示新辅助治疗（P=0.003）、BMI≥30 kg·m-2（P=0.035）、肿瘤下缘距肛缘<5 cm（P＜0.001）和吻合口瘘（P=0.007）是LARS的独立危险因素；本模型C-index为0.808，校正后C-index为0.794；经Hosmer-Lemeshow拟合优度检验，预测模型与实际观察值之间无差异（χ2=3.368 1，P=0.909 2）。结论:肥胖、低位肿瘤，术前新辅助治疗，存在吻合口瘘的患者发生LARS的风险较高。本研究构建的列线图对术后发生LARS有较高的预测价值，可为个体化选择手术方式、制定治疗方案提供依据。

Establishment of a nomogram predicting low anterior resection syndrome after low anterior resection of rectal cancer

Objective:To investigate the risk factors of low anterior resection syndrome(LARS) after laparoscopic low anterior resection(LAR) for rectal cancer,and establish a nomogram prediction model.Methods:A retrospective case-control study was conducted.We analyzed clinicopathological data of 210 patients.Patients were classified into LARS and non-LARS group according to the LARS questionnaires scores.Clinicopathological data were analyzed by univariate and multivariate analysis.Factors which have statistical significance were then screened by multivariate Logistic regression to identify independent risk factors of LARS.Then the R software was introduced to construct nomogram prediction model.After establishing the model,clinicopathological data of 81 patients were collected for external validation.The discrimination of this model was evaluated by calculating the C-index,and calibration of it was evaluated by calculating the bias-corrected C-index and using the Hosmer-Lemeshow test.Results:In univariate analysis,the occurrence of LARS was related to neoadjuvant therapy,BMI,distance from tumor to anus and anastomotic leak.Multivariate analysis showed neoadjuvant therapy(P=0.003),BMI≥30 kg·m-2(P=0.035),distance from tumor to anus<5 cm(P＜0.001) and anastomotic leak(P=0.007) were the independent risk factors of LARS.The C-index of this model was 0.808.After internal verification by the Bootstrap method,the bias-corrected C-index was 0.794.The Hosmer-Lemeshow goodness of fit test showed there was no difference between the predicted model and the actual observed value(χ2=3.368 1,P=0.909 2).Conclusion:Patients with obesity,low distance tumor,neoadjuvant therapy before surgery,and anastomotic leakage are at high risk of LARS.This nomogram model has a good predictive ability for LARS after low anterior resection and can help making clinical treatment strategy individually.