基于微炎症指标构建老年维持性血液透析患者合并骨质疏松的动态列线图预测模型

朱晶晶; 付俞; 周雅萍

doi:10.3969/j.issn.1671-2390.2025.04.008

基于微炎症指标构建老年维持性血液透析患者合并骨质疏松的动态列线图预测模型

Construction of a dynamic nomogram to predict osteoporosis in elderly maintenance hemodialysis patients based on microinflammatory indicators

摘要

摘要: 目的基于微炎症指标超敏C反应蛋白（high-sensitivity C reactive protein，hs-CRP）和降钙素原（procalcitonin，PCT）构建老年维持性血液透析患者合并骨质疏松的动态列线图（dynamic nomogram，DynNom）预测模型。方法分别选取2020年1月1日至2023年1月31日和2023年2月1日至4月30日南京市溧水区人民医院（东南大学附属中大医院溧水分院）收治的老年维持性血液透析患者作为训练集（n=161）和验证集（n=37）。将训练集患者根据骨质疏松发生情况分为骨质疏松组（n=71）和非骨质疏松组（n=90）。收集患者临床资料，采用单因素和多因素回归分析老年维持性血液透析患者合并骨质疏松的影响因素。采用受试者工作特征（receiver operating characteristic，ROC）曲线评估hs-CRP和PCT对老年维持性血液透析患者合并骨质疏松的预测价值，并据此建立DynNom动态评分模型。结果骨质疏松组和非骨质疏松组患者性别、年龄、吸烟史、原发病、慢病肾衰竭病程、血钙、血磷、血钾、全段甲状旁腺素、肌酐、尿酸和碱性磷酸酶比较，差异无统计学意义（P>0.05）。骨质疏松组患者的体重指数（body mass index、BMI）<18.5 kg/m²的占比（23.94%比8.89%）、透析龄≥24个月的占比（32.39%比14.44%）、白蛋白<35 g/L的占比（22.54%比7.78%）、血红蛋白<110 g/L的占比（47.89%比25.56%）、hs-CRP［（7.30±1.67）mg/L比（6.82±1.20）mg/L］和PCT［（19.87±3.28）ng/L比（18.16±3.34）ng/L］水平高于非骨质疏松组（P<0.05）。BMI<18.5 kg/m²、透析龄较长、白蛋白<35 g/L、血红蛋白<110 g/L、hs-CRP和PCT较高是老年维持性血液透析患者合并骨质疏松的独立危险因素（P<0.05）。hs-CRP和PCT预测老年维持性血液透析患者合并骨质疏松的ROC曲线的曲线下面积（area under curve，AUC）分别为0.597（95%CI：0.506~0.688）和0.636（95%CI：0.551~0.722）。DynNom预测模型验证结果显示，训练集和验证集的一致性指数分别为0.801（95%CI：0.767~0.835）和0.793（95%CI：0.752~0.821），两集的校准曲线趋近于理想曲线，两集的ROC曲线的AUC分别为0.807（95%CI：0.872~0.842）和0.797（95%CI：0.868~0.837），两集的决策曲线均显示0~100%预测范围内，模型净获益值均≥0。结论 BMI<18.5 kg/m²、透析龄较长、白蛋白<35 g/L和血红蛋白<110 g/L与老年维持性血液透析患者合并骨质疏松的风险呈正相关关系。hs-CRP和PCT对老年维持性血液透析患者合并骨质疏松具有一定的预测价值。基于hs-CRP和PCT构建的DynNom预测模型能够有效预测老年维持性血液透析患者合并骨质疏松的风险。

Abstract: Objective To construct a dynamic nomogram （DynNom） to predict osteoporosis in elderly maintenance hemodialysis patients based on the microinflammatory indicators of high-sensitivity C reactive protein （hs-CRP） and procalcitonin （PCT）.Methods Elderly maintenance hemodialysis patients admitted to Nanjing Lishui People's Hospital （Zhongda Hospital Lishui Branch， Southeast University） from January 1， 2020 to January 31， 2023 and from February 1 to April 30， 2023 were selected and enrolled in the training set （n=161） and validation set （n=37）. Patients in the training set were divided into the osteoporosis group （n=71） and non-osteoporosis group （n=90） according to the occurrence of osteoporosis. Clinical data of the patients were collected. Univariate and multivariate regression analyses were performed to identify factors affecting osteoporosis in elderly patients with maintenance dialysis. The values of hs-CRP and PCT in predicting osteoporosis in elderly maintenance hemodialysis patients were assessed using receiver operating characteristic （ROC） curves， and the DynNom was established accordingly.Results There were no significant differences in the gender， age， smoking history， primary disease， duration of chronic renal failure， blood calcium， blood phosphorus， blood potassium， whole segment parathyroid hormone， creatinine， uric acid， and alkaline phosphatase between the osteoporosis group and non-osteoporosis group （P>0.05）. The percentage of patients in the osteoporosis group with body mass index （BMI） <18.5 kg/m² （23.94% vs 8.89%）， dialysis duration≥24 months （32.39% vs 14.44%）， albumin <35 g/L （22.54% vs 7.78%）， hemoglobin <110 g/L （47.89% vs 25.56%）， hs-CRP ［（7.30±1.67）mg/L vs （6.82±1.20）mg/L］ and PCT ［（19.87±3.28）ng/L vs （18.16±3.34）ng/L］ were significantly higher than those of the non-osteoporotic group （P<0.05）. BMI <18.5 kg/m²， and longer duration of dialysis， albumin <35 g/L， hemoglobin <110 g/L， and higher hs-CRP and PCT were independent risk factors for osteoporosis in elderly maintenance hemodialysis patients （P<0.05）. The area under the curve （AUC） of hs-CRP and PCT in predicting osteoporosis in elderly maintenance hemodialysis patients was 0.597 （95%CI： 0.506-0.688） and 0.636 （95% CI： 0.551-0.722）， respectively. Validation of the DynNom showed that the consistency of the training and validation sets was 0.801 （95%CI： 0.767-0.835） and 0.793 （95%CI： 0.752-0.821）， respectively， with the two sets of calibration curves converged to the ideal curve. The AUC was 0.807 （95%CI： 0.872-0.842） in the training set and 0.797 （95%CI： 0.868-0.837） in the validation set. The decision curves of the two sets showed that the net benefit value of the model was ≥0 within the range of 0 to 100% prediction.Conclusion BMI <18.5 kg/m²， longer duration of dialysis， albumin <35 g/L and hemoglobin <110 g/L are positively correlated with the risk of osteoporosis in elderly maintenance hemodialysis patients. hs-CRP and PCT have some predictive value for osteoporosis in elderly maintenance hemodialysis patients. The DynNom created based on hs-CRP and PCT can effectively predict the risk of osteoporosis in elderly maintenance hemodialysis patients.

HTML全文

参考文献(23)

施引文献

资源附件(0)