A Prognosis Model Construction And Chinese Herbal Medicine Analysis For Early Type 2 Diabetic Kidney Disease Based On TCM Syndromes

ObjectiveTo analyze risk factors from Western medicine and TCM syndromes in the progression of early type 2 diabetic kidney disease(T2DKD),build different prognosis models based on different factors,evaluate the prediction performance among models,develop the optimal one using nomogram for clinical prevention and treatment.To explore Prof Fan Guanjie’s understanding of pathogenesis,medication characteristics and syndromes differentiation for early T2DKD based on the risk factors of TCM syndromes,enrich the experience of TCM for early T2DKD treatment and provide references for clinical application.Methods1.A prognosis model construction and verification for early T2DKD based on risk factors from Western medicine and TCM syndromes.This study was a retrospective cohort study.The patients who were diagnosed as early T2DKD in Department of Endocrinology from Guangdong Provincial Hospital of Traditional Chinese Medicine were selected as subjects.Patients can be divided into two cohorts based on different period,and one cohort was development cohort and the other was verification cohort.The general data,past diseases,combined medication,TCM syndromes and laboratory examinations of patients were recorded.The disease progression was taken as the composite end points,and the follow-up study was carried out.Univariate Cox regression analysis was used to screen relevant variables.Youden’s index of receiver operating characteristic curve(ROC)and HR value of restricted cubic spline(RCS)were used to explore the optimal cutoffs of continuous variables.These continuous variables were transformed into categorical variables based on different cutoffs.Multivariate Cox regression models were constructed using different forms of variables combination.The kaplan-meier survival curve was applied to evaluate the progression free rate of T2DKD patients between different groups,and the log-rank test was used to compare the difference between groups.The optimal model was obtained and visualized using nomogram after comparing the prediction performance among different models using C-index,the area under ROC(AUC),net reclassification index(NRI),integrated discrimination improvement(IDI)and decision curve analysis(DCA).Finally,the calibration charts were used to verify the accuracy of the best model by bootstrap method.The external verification was also conducted to evaluate the stability and accuracy of the model.2.Medication analysis for early T2DKD based on the risk factor from TCM syndromes.The medical records of patients with early T2DKD with the TCM risk syndromes in Professor Fan Guanjie’s outpatient department were collected as subjects.According to the inclusion and exclusion criteria,the medical records and prescriptions were screened.Some data such as the basic information of patients,tongue and pulse,TCM syndromes,TCM prescriptions were recorded.The data were normalized and standardized before analysis using Ancient and Modern Medical Case Cloud Platform.All data were analyzed using frequency statistic analysis.The relation between TCM syndromes were detected using correlation analysis and complex network.The compatibility relation between traditional herbal medicines were investigated using correlation analysis and cluster analysis.The corresponding relation between TCM syndromes and traditional herbal medicines were detected using correlation analysis and complex network.Through these analysis,the thoughts for the treatment of early T2DKD of Prof.Fan were obtained.Through correlation analysis and cluster analysis,the common herbal combinations for T2DKD treatment were obtained.By setting the confidence degree and edge weighted number,the core herbal combination for the risk symptom was obtained.Results1.A total of 244 cases were used as development cohort.The median follow-up time was 6.89[6.19,7.59]years.By the end of follow-up,137 patients had endpoint events.A total of 159 cases were used as verification cohort,the median follow-up time was 6.50[5.43,7.57]years and finally there were 81 positive events.Given the results of Cox univariate analysis and collinearity analysis in development cohort,13 variables,including age>60 years old,smoking,BMI>24kg/m2,DR,DPN,hypertension,hyperlipidemia,blood-stasis syndrome,damp-heat syndrome,Hb<120/130g/L,UAER,eGFR and NLR were used to perform multivariate Cox regression.UAER,eGFR and NLR were continuous variables,and others were categorical variables.Model 1 showed that smoking,BMI>24kg/m2,hyperlipidemia,damp-heat syndrome,Hb<120/130g/L,UAER and NLR were the independent risk factors for outcome events.According to results of ROC and RCS analysis,model 2 and 3 were built based on different cutoffs of continuous variables,respectively.Model 2 showed that smoking,BMI≥24kg/m2,hyperlipidemia,hypertension,damp-heat syndrome,Hb<120/130g/L,UAER≥71.80mg/24h and NLR≥2.54 were the independent risk factors for outcome events,while model 3 showed the similar results of model 2,but the cutoff of UAER was 70.83mg/24h and NLR was 2.33.In terms of discrimination evaluation,the C-index of model 1,model 2 and model 3 were 0.699,0.711 and 0.712,respectively.The AUC of 3-year progression free rate of model 1,model 2 and model 3 were 0.749,0.746 and 0.755,5-year progression free rate were 0.717,0.731 and 0.632,and 9-year progression free rate were 0.639,0.733 and 0.692,respectively.In terms of diagnostic value evaluation,compared with model 1,with predicting 3-year,5-year and 9-year progression free rate,the continuous NRIs of model 2 were 0.360,0.303 and 0.335,respectively(P<0.05),and the IDIs were 0.0314,0.3551 and 0.5327,respectively(P<0.05).Compared with model 1,with predicting 3-year progression free rate,the continuous NRI of model 3 was 0.277(P<0.05),and the IDI was 0.0228(P<0.05).However,with predicting 5-year and 9-year progression free rate,the continuous NRIs of model 3 were 0.216 and 0.190(P<0.05),the IDIs were 0.0225 and 0.0198(P<0.05).In terms of clinical benefits,DCA showed that model 2 and model 3 were similar but higher than model 1 within the threshold range of 10%～58%,model 2 was significantly higher than model 3 and model 1 within the threshold range of 58%～78%,while model 2 was similar to model 1 and slightly lower than model 3 within the threshold range of 78%～90%.Finally,a prognosis model for early T2DKD was built and displayed with nomogram using variables in model 2(smoking,BMI≥24kg/m2,hyperlipidemia,hypertension,damp-heat syndrome,Hb<120/130g/L,UAER≥71.80mg/24h and NLR≥2.54).The C-index of verification cohort was 0.721.The AUC of 3-year,5-year and 9-year progression free rate of verification cohort were 0.749,0.757 and 0.789,respectively.The 3-year,5-year and 9-year calibration curves of development cohort and verification cohort were close to the reference curves.Both internal and external validation suggested that the stability and accuracy of the prognostic model were good.2.Damp-heat was an independent risk factor for early T2DKD progression.Cleaning damp and heat are the main strategies for retarding early T2DKD.Based on Ancient and Modern Medical Case Cloud Platform,137 cases of early T2DKD with damp-heat syndrome from Professor Fan’s outpatient department were analyzed and the results showed that damp,heat and blood stasis were the top TCM syndrome factors and related closely.In terms of four natures of traditional herbal medicines,slightly cold was used 526 times,which accounted for the highest proportion(33.33%),followed by cold 270 times(17.11%)and warm 261 times(16.54%).In terms of five flavors of traditional herbal medicines,bitter was used 877 times,which accounted for the highest proportion(55.58%).followed by 655 times of pungent(41.51%),576 times of sweet(36.50%)and 125 times of light(7.92%).Channel tropisms sorted by quantity were liver and spleen,which were 849 times(53.80%)and 698 times(44.23%).According to the function,traditional herbal medicines were mainly classified into four kinds,including inhibiting-damp drugs,clearing-heat drugs,blood-activating and stasis-resolving drugs and reinforcing drugs.The most common inhibiting-damp drugs were Yinchen,Yiyiren and Cheqiancao,which were used 108 times(78.83%),86 times(62.77%)and 65 times(47.45%),respectively.The most common clearing-heat drugs were Huangbai,Mudanpi and Shengdihuang,which were used 84 times(61.83%),76 times(55.57%)and 43 times(31.35%),respectively.The most common blood-activating and stasis-resolving drugs were Danshen,Eshu and Yimucao,which were used 98 times(71.53%),52 times(37.96%)and 33 times(24.09%),respectively.The most common reinforcing drugs were Baishao,Huangqi,Gancao and Baizhu,which were used 56 times(40.88%),41 times(29.93%),37 times(27.01%),and 37 times(27.01%),respectively.According to the results of cluster analysis,correlation analysis and the efficacy of traditional herbal medicines,different traditional herbal medicines can be divided into six drug combinitions.Combinition 1 included Yinchen,Huangbai and Dansen,which was used in pitients with mild demp-heat sydrome.Combinition 2 contained Yiyiren,Cheqiancao,Cangshu,Eshu and Buzhaye,which was used in patients with severe dampness and stasis.Combinition 3 included Mudanpi,Baishao,Chaihu and Bohe,which was used in patients with liver depression and spleen deficiency.Combinition 4 had Yimucao,Baishu,Honghua,Shanzha and Dahuang,which was used in patients with damp-heat in the gastrointestinal and severe heat and stasis.Combinition 5 had Shengdihuang,Huangqi,Gancao and Zelan,which was used in patients with deficiency of qi and yin.Combinition 6 included Nvzhenzi,Hanliancao,Gouji and Xuduan,which was used in patients with yin deficiency of liver and kidney.A multi-scale backbone network was constructed using TCM syndromes and different traditional herbal medicines.With the confidence degree 0.6 and the edge weighted number 70,the core network was established with eight nodes and twelve edges,which consisted of three TCM syndromes including dampness,heat and blood stasis,and a core group composed of Yinchen,Dansen,Yiyiren,Huangbai and Mudanpi.This herbal combination mainly targeted the pathogenesis of damp,heat and blood stasis,which was used as the core prescription for early T2DKD treatment.The core prescription and common herbal combinations can be modified for clinical application based on the evolution of pathogenesis of early T2DKD.Conclusion1.Smoking,BMI≥24kg/m2,hyperlipidemia,hypertension,damp-heat syndrome,Hb<120/130g/L,UAER≥71.80mg/24h and NLR≥2.54 were the independent risk factors affecting the progression of early T2DKD.Among these variables,damp-heat syndrome was the risk factor from TCM syndromes.With these risk factors from Western medicine and damp-heat syndrome,a prognosis model was finally established and visualized with nomogram,which may provide reference for prognosis evaluation and treatment strategies selection of early T2DKD.2.Damp-heat was the core TCM syndrome in the progression of early T2DKD.For this special pathogenesis,Prof.Fan usually used Yinchen,Dansen,Yiyiren,Huangbai and Mudanpi as the basic prescription for treatment.According to the degree and location of damp-heat syndrome or the evolution rules of damp-heat and healthy qi of early T2DKD,using the medication flexibly based on the basic prescription,may provide TCM treatment directions and ideas for delaying early T2DKD.