| Objectives:To explore the correlation between inflammatory composite indicators NLR,MLR,MHR,NHR,SII,SIRI,PNI and diabetic peripheral neuropathy,identify clinical risk factors for diabetic peripheral neuropathy,construct a clinical prediction model for diabetic peripheral neuropathy by using inflammatory composite indicators and the identified clinical risk factors,and evaluate the model,so as to provide a new direction for the prevention and treatment of diabetic peripheral neuropathy.Methods:The clinical data of 460 patients with type 2 diabetes mellitus from June 2021 to June 2022 in the Department of Endocrinology,First Affiliated Hospital of Kunming Medical University were collected and retrospectively analyzed.The subjects were divided into diabetic peripheral neuropathy group(DPN group)with 319 patients and the non-diabetic peripheral neuropathy group(non-DPN Group)with 141 patients.according to the diagnostic criteria of DPN and clinical indicators were compared between the two groups.The whole subjects were randomly divided into training set(n=322)and validation set(n=138)in a 7:3 ratio by R statistical software and the inflammatory composite indicators and other clinical indicators were compared between the two data sets and between the training sets.In the training sets,the trend risk of inflammatory composite indicators and DPN were analyzed using restricted cubic splines(RCS).The risk factors for the development of DPN were identified by univariate logistic regression,and the prediction model 1 were constructed by multivariate logistic regression.Lasso regression and random forest method were used to identify the risk factors of DPN and construct prediction models 2 and 3,respectively.In the training set,10-fold cross-validation and Receiver Operating Characteristic Curve(ROC),Area Under Curve(AUC)were used to evaluate the discrimination of the model,Hosmer-Lemeshow test,calibration curve to evaluate the calibration of the model,Decision Curve Analysis(DCA)to evaluate the clinical utility of the model,and tested it in the validation set,Finally,the best predictive model of this study was demonstrated by predictive equations and nomograms.Results:1.There were 460 patients with type 2 diabetes in this study,322 in the training set and 138 in the validation set.The WHR,the proportion of patients with coronary heart disease,the proportion of patients with atherosclerosis,the proportion of patients with dyslipidemia,Hb A1 c,2-hour postprandial blood glucose,mean glucose,MLR,MHR,NHR and SIRI of patients in the overall study group and the DPN group in the training set were higher than those in the non-DPN group(P<0.05),while the ALB,UACR,TIR and PNI were lower than those in the non-DPN group(P<0.05).Except CHE,there was no significant difference between training set and validation set(P>0.05).2.Restricted cubic spline regression analysis showed a strong relationship between the inflammatory composite indexes MLR,MHR,NHR,SIRI and PNI and the risk of DPN,with the first intersection of MLR,MHR,NHR,SIRI and PNI with the OR=1 curve at 0.207,0.439,3.517,0.714 and 51.725.Respectively,the risk of occurrence of DPN with MLR below 0.207 increased significantly with higher MLR;the risk of occurrence of DPN with MHR above 0.624 increased with higher MHR;the risk of occurrence of DPN with NHR above 5.120 increased with higher NHR;the risk of occurrence of DPN with SIRI below 0.714 increased with higher SIRI;and the risk of occurrence of DPN with PNI below The risk of DPN below 51.725 decreased significantly with increasing PNI.3.Univariate logistic regression analysis identified the age,waist circumference,WHR,history of coronary heart disease,history of atherosclerosis,history of dyslipidemia,Hb A1 c,2h postprandial glucose,NEUT,MONO,ALB,HDL-C,mean glucose,TIR,MLR,MHR,NHR,SIRI,PNI,a total of 19 indicators,with DPN the association was significant(P<0.1).Further multifactorial logistic stepwise regression analysis was performed to establish the predictive model 1(P<0.05),with the history of coronary heart disease,atherosclerosis,dyslipidemia,ALB,average glucose and SIRI.4.The risk factors of DPN were selected by Lasso regression analysis as history of coronary heart disease,history of atherosclerosis,history of dyslipidemia,Hb A1 c,FCP,NEUT,MONO,CHE,Scr,SIRI,PNI.The prediction model 2 was constructed by history of coronary heart disease,history of atherosclerosis,history of dyslipidemia,FCP,NEUT,MONO,Scr,PNI(P<0.05).5.The random forest method was used to identified risk factors for the occurrence of DPN,and a multifactorial regression model was fitted using the risk factors with the top 15 importance scores: history of dyslipidemia,ALB,Scr,NEUT,UACR,PNI,UA,age,history of atherosclerosis,Urea,TG,NLR,CHE,SIRI,BMI,and finally history of atherosclerosis,dyslipidemia,Scr,NEUT,PNI,NLR to established predictive model 3(P<0.05).6.A comparison of the discrimination,calibration and clinical utility of the three models by 10-fold cross-validation,ROC curves,Hosmer-Lemeshow test,calibration curves and DCA suggested that the three prediction models had high discrimination,calibration and clinical utility in both the training and validation sets,and prediction model 2 had the best discrimination,calibration and clinical utility.Conclusions:1.When MLR is higher than 0.207,MHR is higher than 0.439,NHR is higher than 3.517 and PNI is lower than 51.725,the risk of DPN is increased,and early preventive intervention should be carried out.2.History of coronary heart disease,history of atherosclerosis,history of dyslipidemia,average glucose,NEUT,MONO,NLR,SIRI may be independent risk factors for the development of DPN,but ALB,FCP,Scr,PNI may be independent protective factors for the development of DPN.3.The clinical prediction model of DPN constructed by history of coronary heart disease,history of atherosclerosis,history of dyslipidemia,FCP,NEUT,MONO,Scr,and PNI had the highest discrimination and calibration,the strongest clinical utility,and the highest predictive value for the occurrence of DPN. |
PDF Full Text Request