基于血清标志物构建预测老年重症肺炎预后的Nomogram模型

Nomogram model for predicting the prognosis of senile severe pneumonia based on serum markers

目的:探究血清标志物Nomogram预测模型对老年重症肺炎(SP)预后的预测价值。方法:选取2022年1月至2023年1月武汉市红十字会医院收治的310例老年SP患者,按7∶3比例随机分为建模人群(n=217)与验证人群(n=93)。比较建模人群、验证人群入院28 d内预后情况,血清可溶性髓系细胞表达的触发受体-1(sTREM-1)、基质金属蛋白酶抑制剂-1(TIMP-1)、可溶性白细胞分化抗原14亚型(Presepsin)、N末端脑钠肽前体(NT-proBNP)、C反应蛋白(CRP)、饥饿素(Ghrelin)、降钙素原(PCT)、中性粒细胞与淋巴细胞比值(NLR)、肿瘤坏死因子-α(TNF-α)和白介素-6(IL-6)水平,Lasso-logistic回归分析老年SP预后不良的预测因素,并构建预后不良Nomogram预测模型,在验证人群中对Nomogram预测模型进行外部验证。结果:建模人群入院28 d内死亡78例(35.94%),验证人群入院28 d内死亡34例(36.56%),两组病死率比较无统计学差异(P>0.05)。建模人群、验证人群中,不同预后患者血清sTREM-1、NT-proBNP、TIMP-1、Presepsin、PCT、Ghrelin、CRP、IL-6、NLR、TNF-α水平比较,差异有统计学意义(P<0.05)。Lasso回归筛选预测因素,logistic回归分析显示,血清sTREM-1、TIMP-1、NT-proBNP、Presepsin、Ghrelin、PCT、NLR水平为老年SP预后不良的影响因素(P<0.05)。基于Lasso-logistic回归预测因素构建预测模型,验证人群受试者工作特征(ROC)曲线、临床决策曲线(DCA)显示,该预测模型具有良好的临床效用。结论:血清sTREM-1、TIMP-1、NT-proBNP、Presepsin、Ghrelin、PCT、NLR水平为老年SP患者预后不良的预测因子,基于以上因素构建Nomogram预测模型具有一定的临床价值。

Objective:To investigate the predictive value of Nomogram prediction model based on serum markers for senile severe pneumonia(SP).Methods:A total of 310 senile patients with SP admitted to Wuhan Red Cross Hospital from January 2022 to January 2023 were selected and randomly divided into modeling population(n=217)and validation population(n=93)according to a ratio of 7:3.The prognosis within 28 days after admission was compared between the modeling population and the validation population.Serum soluble triggering receptor expressed on myeloid cells-1(sTREM-1),tissue inhibitor of metalloproteinase-1(TIMP-1),soluble leukocyte differentiation antigen 14 subtype(Presepsin),N-terminal pro-brain natriuretic peptide(NT-proBNP),C-reactive protein(CRP),Ghrelin,procalcitonin(PCT),neutrophils lymphocytes ratio(NLR),and the levels of tumor necrosis factor-α(TNF-α)and interleukin-6(IL-6)were compared.Lasso-logistic regression analysis was used to analyze the predictive factors of poor prognosis in senile patients with SP,and a Nomogram prediction model of poor prognosis was constructed.The Nomogram prediction model was externally validated in the validation population.Results:Seventy-eight patients(35.94%)died within 28 days after admission in the modeling group,and 34 patients(36.56%)died within 28 days in the validation group.There was no significant difference in the case fatality rate between the two groups(P>0.05).There were significant differences in serum sTREM-1,NT-proBNP,TIMP-1,Presepsin,PCT,Ghrelin,CRP,IL-6,NLR and TNF-αlevels among patients with different prognosis in modeling population and validation population(P<0.05).Logistic regression analysis showed that serum levels of sTREM-1,TIMP-1,NT-proBNP,Presepsin,Ghrelin,PCT and NLR were the influencing factors for the poor prognosis of senile SP(P<0.05).Based on Lasso-logistic regression prediction factors,the prediction model was constructed to verify the receiver operating characteristic(ROC)curve and clinical decision curve(DCA)of the population.The results showed that the prediction model had good reference clinical utility.Conclusion:Serum sTREM-1,TIMP-1,NT-proBNP,Presepsin,Ghrelin,PCT and NLR levels are predictive factors of poor prognosis in senile patients with SP,and the Nomogram prediction model based on the above factors has certain clinical value.