基于WIfI分级的慢性下肢重度缺血患者预后列线图构建和验证

Construction and validation of a prognostic nomogram for patients with chronic limb-threatening ischemia based on WIfI classification

目的构建并验证基于创面-缺血-足部感染(wound-ischemia-foot infection,WIfI)分级的列线图,以预测慢性下肢重度缺血(chronic limb-threatering ischemia,CLTI)患者在腔内治疗后1年内无截肢生存(amputation free survival,AFS)预后。方法收集2017年1月~2023年1月南方医科大学顺德医院223例经下肢血管腔内治疗后的CLTI患者资料,将数据随机划分为训练集(156例)和验证集(67例),以无截肢生存(AFS)为阳性结局(161例,占比72.2%)。在训练集中利用Logistic回归分析筛选建模变量并构建包含WIfI分级变量的二分类Logistics回归模型,绘制列线图。通过受试者工作特征曲线、Hosmer-Lemeshow拟合优度检验、校准曲线和决策曲线分别对训练集和验证集的区分度、校准度及临床适用度作出评估,计算净重新分类改善指数、综合区分改善指数比较新开发模型与WIfI分级风险预测模型的预测性能。结果性别、空腹血糖、估算肾小球滤过率、WIfI分级、创面病原学是血管腔内治疗后CLTI患者1年内AFS的独立影响因素。基于这些变量构建列线图,新模型在训练集中的曲线下面积(area under curve,AUC)为0.902[95%CI(0.8514,0.9531)];验证集中AUC为0.856[95%CI(0.7586,0.9542)]。行H-L检验,得到训练集χ^(2)为7.6399(P=0.4694),验证集χ^(2)为9.2647(P=0.3205),校准曲线拟合良好。决策曲线显示当阈值概率大于0.25时,模型具有较高的临床净收益。两模型比较示当截断值取0.64时,净重新分类改善指数(分类变量)=0.1876[95%CI(0.0904,0.2849)],P<0.001;综合区分改善指数=0.1765[95%CI(0.1226,0.2303)],P<0.001。结论新开发列线图相关模型与原WIfI分级风险预测模型在预测CLTI患者腔内治疗后1年内AFS的区分度、校准度及临床适用度均有较积极的表现,相比之下,新模型的预测能力更准确,新模型预测能力优于旧模型,在个体化诊疗上具备临床推广潜力。

Objective To develop and validate a wound-ischemia-foot infection(WIfI)-based nomogram for predicting amputation-free survival(AFS)in patients with chronic limb-threaten-ing ischemia(CLTI)after endovascular treatment within 1 year.Methods A total of 223 CLTI patients who underwent lower extremity endovascular treatment at our hospital between January 2017and January 2023 were included.The dataset was randomly divided into a training set(156 cases)and a validation set(67 cases).The positvive outcome was AFS(161 cases,72.2%).Logistic regression analysis was performed on the training set to identify significant variables associated with AFS,including WIfI grading variables.Subsequently,a binary logistic regression model incorporating these variables was constructed and visualized as a nomogram.Discrimination,calibration,and clinical applicability of the model were evaluated using receiver operating characteristic(ROC)curve analysis,Hosmer-Lemeshow test(H-L test),calibration curve analysis,and decision curve analysis(DCA),respectively.The net reclassification improvement index(NRI)and the integrated discrimination improvement(IDI)were calculated to compare the prediction performance of the newly developed model and the WIfI classification model.Results Gender,fasting blood glucose level,estimated glomerular filtration rate(eGFR),WIfI classification score,and wound etiology were identified as independent factors influencing AFS in CLTI patients within one year after endovascular treatment.Based on these variables,we developed a nomogram that demonstrated good predictive performance in both the training set area under curve(AUC)=0.902[95%CI(0.8514,0.9531)],and the validation cohort AUC=0.856[95%CI(0.7586,0.9542)].H-L tests indicated satisfactory goodness-of-fit for both sets(training setχ^(2)value=7.6399;P=0.4694;validation setχ^(2)value=9.2647;P=0.3205),while calibration curves confirmed excellent agreement between predicted probabilities from the model and observed outcomes in both sets.The DCA curve demonstrated that the model exhibits substantial clinical net benefits when the threshold probability exceeds 0.25.The two models indicated that with a binary NRI cut value of 0.64,the NRI(classification variable)=0.1876[95%CI(0.0904,0.2849)],with P<0.001.The IDI=0.1765[95%CI(0.1226,0.2303)],with P<0.001.Conclusion The newly developed WIfI-based nomogram model and the original WIfI risk classification model demonstrate superior discrimination,calibration,and clinical applicability in predicting AFS within one year after endovascular treatment of patients with chronic limb-threatening ischemia(CLTI).In contrast,the new model exhibits enhanced accuracy and predictive ability compared with the old model,thereby holding potential for clinical implementation in personalized diagnosis and treatment.