新型炎症评分系统与胃癌延迟性术后肠麻痹的关系及预测模型构建

Relationship between new inflammation scoring system and delayed postoperative intestinal paralysis of gastric cancer and construction of prediction model

目的分析胃癌延迟性术后肠麻痹(PPOI)与炎症反应标志物的关系并构建新型炎症评分系统,结合其他临床危险因素构建Nomogram预测模型。方法选取2019年1月—2021年8月在南昌大学第二附属医院行根治术的299例胃癌患者。分为训练集199例与验证集100例。采用单因素和多因素Logistic回归模型分析胃癌患者PPOI发生的危险因素。绘制受试者工作特征(ROC)曲线分析中性粒细胞与淋巴细胞比值(NLR)、血小板与淋巴细胞比值(PLR)、淋巴细胞与单核细胞比值(LMR)及C反应蛋白与白蛋白比值(CAR)4种炎症反应标志物并获取最佳截断值。根据最佳截断值构建新型炎症评分系统。将独立危险因素作为构建Nomogram模型的预测指标并在训练集与验证集数据中进行效能验证。结果训练集和验证集中无PPOI和PPOI患者的年龄、ASA分级、CCI、肿瘤大小、TNM分期、手术方式、NLR、PLR、LMR及CAR比较,差异均有统计学意义(P<0.05)。NLR、PLR、LMR及CAR的敏感性分别为78.43%、86.27%、68.63%和76.47%,特异性分别为69.59%、61.49%、79.73%和73.65%,ROC曲线下面积(AUC)分别为0.775、0.776、0.778和0.808,最佳截断值分别为>2.8、>113.71、≤3.68、>0.28。年龄≥65岁[O^R=4.102(95%CI:1.042,16.149)]、ASA分级Ⅲ、Ⅳ级[O^R=3.061(95%CI:0.885,10.586)]、TNM分期Ⅲ、Ⅳ期[O^R=3.825(95%CI:2.698,6.033)]、开放式手术[O^R=4.063(95%CI:3.263,8.268)]、NLR≥2.67[O^R=2.171(95%CI:1.368,3.445)]及CAR≥0.28[O^R=1.028(95%CI:1.011,1.046)]是胃癌患者术后PPOI发生的独立危险因素(P<0.05)。训练组校正曲线分析显示一致性指数(C-index)为0.892,ROC曲线分析结果显示AUC为0.889,当预测PPOI发生风险阈值>0.185时,Nomogram模型提供显著的临床净收益;验证组校正曲线分析显示C-index为0.817,AUC为0.806,当预测PPOI发生风险阈值>0.056时,Nomogram模型提供显著的临床净收益。结论基于NLR-CAR评分构建的Nomogram预测模型能有效地对胃癌患者PPOI发生风险机进行早期识别,为医护人员采取适当干预措施提供理论依据。

Objective To analyze the relationship between delayed postoperative intestinal paralysis(PPOI)in gastric cancer patients and inflammatory markers,and to construct a new inflammatory scoring system.Combining other clinical risk factors,a Nomogram prediction model was also established.Methods A total of 299 gastric cancer patients who underwent radical surgery at the Second Affiliated Hospital of Nanchang University from January 2019 to August 2021 were selected.They were divided into a training set of 199 cases and a validation set of 100 cases.Single-factor and multiple-factor Logistic regression models were used to analyze the risk factors for PPOI in gastric cancer patients.Receiver operating characteristic(ROC)curves were drawn to analyze four inflammatory markers:neutrophil-to-lymphocyte ratio(NLR),platelet-to-lymphocyte ratio(PLR),lymphocyte-to-monocyte ratio(LMR),and C-reactive protein-to-albumin ratio(CAR),and to obtain the best cutoff values.A new inflammatory scoring system was constructed based on the optimal cutoff values.Independent risk factors were used as predictive indicators to construct the Nomogram model,which was then validated for performance in both the training and validation sets.Results There were statistically significant differences(P<0.05)in age,ASA classification,CCI,tumor size,TNM staging,surgical approach,NLR,PLR,LMR,and CAR between patients with and without PPOI in both the training and validation sets.The sensitivity for NLR,PLR,LMR and CAR was 78.43%、86.27%、68.63%and 76.47%,respectively,The specificity for NLR,PLR,LMR and CAR was 69.59%、61.49%、79.73%and 73.65%,respectively,The AUC values for NLR,PLR,LMR and CAR was 0.775,0.776,0.778 and 0.808,respectively,and the best cutoff values were>2.8,>113.71,≤3.68 and>0.28,respectively.Age(≥65 years)[O^R=4.102(95%CI:1.042,16.149)],ASA classification(Ⅲ,Ⅳ)[O^R=3.061(95%CI:0.885,10.586)],TNM staging(Ⅲ,Ⅳ)[O^R=3.825(95%CI:2.698,6.033)],surgical approach(open)[O^R=4.063(95%CI:3.263,8.268)],NLR(≥2.67)[O^R=2.171(95%CI:1.368,3.445)],and CAR(≥0.28)[O^R=1.028(95%CI:1.011,1.046)]were independent risk factors for PPOI in gastric cancer patients(P<0.05).The calibration curve analysis in the training group showed a consistency index(C-index)of 0.892,and the ROC curve analysis showed an AUC of 0.889.When the predicted risk threshold for PPOI occurrence was>0.185,the Nomogram model provided a significant clinical net benefit.In the validation group,the calibration curve analysis showed a C-index of 0.817,and the AUC was 0.806.When the predicted risk threshold for PPOI occurrence was>0.056,the Nomogram model also provided a significant clinical net benefit.Conclusion The Nomogram prediction model based on the NLR-CAR score can effectively identify the risk of PPOI occurrence in gastric cancer patients,providing a theoretical basis for healthcare providers to take appropriate intervention measures.