Abstract: Analysis of risk factors of anastomotic leakage after rectal surgery and construction of a nomogram prediction model        LIN Zhen-yu*，ZHONG Bin，ZHANG Jian-xin，et al. *The First School of Clinical Medicine，Southern Medical University，Guangzhou 510515，China 
Corresponding authors：JIN Wei-dong，E-mail: jwdong1972@163.com；ZHANG Jian-xin，E-mail: mai_andy0@126.com
LIN Zhen-yu and ZHONG Bin are the first authors who contributed equally to the article
Abstract    Objective    To explore the preoperative risk factors of anastomotic leakage after rectal surgery，construct a preoperative prediction model and guide clinical decision-making. Methods    We retrospectively analysed the preoperative data of 330 patients who underwent radical resection of rectal carcinoma and primary anastomosis in the Central Warring Area General Hospital of the Chinese People's Liberation Army (formerly Wuhan General Hospital of Guangzhou Military Region) between January 1，2007 and December 31，2016. To analyse the risk factors for anastomotic leakage. R software was used to complete the drawing of the nomogram prediction model. The receiver operating characteristic curve(ROC) and calibration plots were used to assess predictive model performance. There were 57 patients from January 1，2017 to December 31，2018 underwent radical resection of rectal carcinoma and primary anastomosis for external validation. Results    There are 42 cases of anastomotic leakage that occurred in this study. The incidence of anastomotic leakage was 12.73% (n=330). Univariate analysis showed nutritional support before the operation，distance from the tumor to the anus，body mass index(BMI)，nutritional risk screening 2002(NRS 2002 score)，American Society of Anesthesiologists (ASA) score，hemoglobin and prealbumin were significantly correlated with anastomotic leakage (P<0.05). Multivariate analysis showed BMI≥25，NRS 2002 score>3，distance from the tumor to the anus≤10cm and hemoglobin<120g/L were independent risk factors for anastomotic leakage. According to the multivariate logistic regression analysis results，the nomogram prediction model was constructed. The area under the ROC curve is 0.874（95%CI：0.823~0.925）. Through internal validation，the concordance index value of the model was 0.869. In the external validation，statistical analysis showed that the AUC was 0.787（95%CI：0.632~0.942）.  Conclusion    BMI≥25，NRS 2002 score>3，distance from the tumor to the anus≤10cm，and hemoglobin<120g/L are independent risk factors for anastomotic leakage. The nomogram based on these variables can help clinical decision-making and support future research.

Key words: rectal carcinoma, anastomotic leakage, risk factors, prediction model

摘要： 目的    探讨直肠癌术后发生吻合口漏的术前危险因素并构建预测模型。方法    回顾性分析2007-01-01至2016-12-31中国人民解放军中部战区总医院普通外科收治的行直肠癌根治术并作一期吻合的330例病人的临床资料，分析发生吻合口漏的术前危险因素。应用R软件完成列线图预测模型。通过受试者工作特征曲线（ROC）和校准曲线评估列线图预测模型的能力。另收集2017-01-01至2018-12-31收治的行直肠癌根治术并作一期吻合的57例病人资料作为外部验证数据。结果    330例中有42例术后发生吻合口漏（12.7%）。单因素分析结果显示，吻合口漏发生与术前营养支持、肿瘤距肛门距离、BMI、营养风险筛查评分（NRS2002）、美国麻醉师协会（ASA）评分、血红蛋白、前白蛋白相关（P均＜0.05）。多因素Logistic回归分析结果显示，BMI≥25、NRS2002评分>3分、肿瘤距肛门距离≤10 cm、血红蛋白<120 g/L是术后发生吻合口漏的独立危险因素。根据多因素回归分析结果构建列线图预测模型，受试者工作特征曲线下面积（AUC）为0.874（95%CI 0.823-0.925）。内部验证显示，模型的一致性指数为0.869。外部验证采用时段验证，AUC为0.787（95%CI 0.632-0.942）。结论    BMI≥25、NRS2002评分>3分、肿瘤距肛门距离≤10 cm及血红蛋白<120 g/L是吻合口漏的独立危险因素，基于这些变量的列线图可作为临床决策参考。

关键词: 直肠癌, 吻合口漏, 危险因素, 预测模型