预测胃癌D2根治术后感染风险列线图模型的建立与验证

Construction and validation of a nomogram model for predicting the risk of infection after D2 radical resection of gastric cancer

摘要:

目的 调查胃癌D2根治术后感染发生的危险因素, 构建能够准确预测术后感染发生风险的列线图模型并进行评价。方法 回顾性分析安徽某医科大学附属医院2019年1月—2021年12月胃肠外科接受胃癌D2根治术患者的临床资料, 按照3∶1比例随机分为训练集和验证集。将训练集病例分为感染组和非感染组, 采用单因素及多因素logistic回归分析胃癌D2根治术后感染的独立危险因素。基于胃癌D2根治术后感染独立危险因素构建列线图风险预测模型, 并分别在训练集和验证集数据中进行验证、评价, 采用受试者工作特征(ROC)曲线、Hosmer-Lemeshow拟合优度检验、校准曲线评估模型的效能, 采用临床决策曲线(DCA)评估模型的临床效用。结果 共纳入477例患者, 其中训练集358例, 验证集119例; 185例胃癌D2根治术患者发生术后感染, 以下呼吸道感染为主(98例, 53.0%), 其次为器官或腔隙感染(48例, 25.9%)、血流感染(17例, 9.2%)。多因素logistic回归分析显示, 腹部手术史(OR=1.922, 95%CI: 1.048~3.523)、NRS2002≥3分(OR=3.525, 95%CI: 2.178~5.707)、PNI < 49.6(OR=1.662, 95%CI: 1.061~2.602)、高ASA分级(2级VS 1级: OR=3.513, 95%CI: 1.883~6.557;3级VS 1级: OR=10.219, 95%CI: 1.097~95.216)、联合器官切除(OR=2.115, 95%CI: 1.105~4.049)、入住重症监护病房(OR=15.927, 95%CI: 1.847~137.330)是胃癌D2根治术后感染的独立危险因素(均P＜0.05)。基于以上独立危险因素构建列线图预测模型, 训练集ROC曲线下面积(AUC)为0.768(95%CI: 0.718~0.818), 验证集AUC为0.750(95%CI: 0.664~0.837), 提示该模型具有良好的区分度和判别能力。Hosmer-Lemeshow拟合优度检验提示该模型拟合度较好(P>0.05), 校准曲线显示该模型在预测胃癌D2根治术后感染具有较好的一致性。DCA显示该模型的临床价值高。结论 基于胃癌D2根治术后感染独立危险因素构建的列线图预测模型, 可为临床早期评估胃癌D2根治术后感染的发生概率提供定量、直观的参考工具。

Abstract:

Objective To investigate the risk factors for postoperative infection after D2 radical resection for gastric cancer, and construct a nomogram model that can accurately predict the risk of postoperative infection.Methods Clinical data of patients underwent D2 radical resection for gastric cancer in the gastrointestinal surgery department of the hospital affiliated to a Medical University in Anhui from January 2019 to December 2021 were retrospectively analyzed. Patients were randomly divided into the training set and the validation set in a ratio of 3 ∶1. Patients in the training set were divided into the infected group and the non-infected group. Univariate and multivariate logistic regression were used to analyze the independent risk factors for postoperative infection after D2 radical resection for gastric cancer. A nomogram model of risk prediction was constructed based on the independent risk factors after D2 radical resection for gastric cancer, validated and assessed in the training set and validation set data, respectively. Receiver operating characteristic (ROC) curve, Hosmer-Lemeshow goodness-of-fit test and calibration curve were used to assess the efficiency of the model. Clinical decision curve analysis (DCA) was used to evaluate clinical validity of the model.Results A total of 477 patients were included in analysis, with 358 in the training set and 119 in the validation set. Among them, 185 patients underwent D2 radical resection for gastric cancer developed postoperative infection, with lower respiratory tract infection being the main infection (n=98, 53.0%), followed by organ or cavity infection (n=48, 25.9%) and bloodstream infection (n=17, 9.2%). Multivariate logistic regression analysis showed the history of abdominal surgery (OR=1.922, 95%CI: 1.048-3.523), NRS 2002≥3 points (OR=3.525, 95%CI: 2.178-5.707), PNI < 49.6(OR=1.662, 95%CI: 1.061-2.602), high ASA classification (Grade 2 vs Grade 1: OR=3.513, 95%CI: 1.883-6.557; Grade 3 vs Grade 1: OR=10.219, 95%CI: 1.097-95.216), combined with organ resection (OR=2.115, 95%CI: 1.105-4.049), and admission to the intensive care unit (OR=15.927, 95%CI: 1.847-137.330) were independent risk factors for postoperative infection after D2 radical resection for gastric cancer (all P < 0.05). A nomogram model of risk prediction was constructed based on the above independent risk factors. The area under the curve (AUC) in the training set and validation set were 0.768 (95%CI: 0.718-0.818) and 0.750 (95%CI: 0.664-0.837), respectively, indicating good discrimination ability of the model. Hosmer-Lemeshow goodness-of-fit test suggested a good fit for the model (P>0.05), and the calibration curve showed good consistency in predicting infection after D2 radical resection for gastric cancer. DCA curve demonstrated the high clinical value of the model.Conclusion The nomogram prediction model based on independent risk factors for postoperative infection after D2 radical resection for gastric cancer can provide a quantitative and intuitive reference tool for early clinical evaluation of the probability of infection after D2 radical resection for gastric cancer.