Validation Of Risk Models For Contrast-Induced Nephropathy Following Percutaneous Coronary Intervention In Chinese Population

Objective:Our study aims to assess the application of risk scores of National Cardiovascular Data Registry(NCDR),Mehran and AGEF for contrast-induced nephropathy(CIN)following Percutaneous Coronary Intervention(PCI)in Chinese population.Methods:A total of 636 patients with coronary heart disease who underwent PCI in Ningde Multiple Hospital from March 2017 to December 2019 were included in our stduy.The general clinical data,examination results and drug use of the patients were collected.The incidences of CIN was calculated according to the definition of CIN recommended by Contrast Media Safety Committee,European Society of Urogenital Radiology(ESUR CMSC)and Kidney Disease: Improving Global Outcome(KDIGO).Risk factors were analysed in the definition of CIN that was recommend by KDIGO.Risk models performance was evaluated via area under receiver-operating-characteristic curves(discrimination)and graphical analysis/ goodness of fit test(calibration).Results:1.The incidence of CIN was 16.67% when CIN defined by ESUR CMSC,while the incidence of CIN was 6.13% when CIN defined by KDIGO.2.The proportions of age > 75 years old,hypotension,acute heart failure,congestive heart failure,hypertension,CKD and anemia in CIN group were significantly higher than those in non-CIN group.However,left ventricular ejection fraction in CIN group was lower than in non-CIN group.Diuretic and recombinant human brain natriuretic peptide(rh BNP)were more largely used in CIN group compared with non-CIN group.3.Multivariate logistic regression showed hypertension(OR = 2.540,95% CI:1.031-6.262,p = 0.043),hypotension(OR = 4.367,95% CI: 1.479-12.897,p = 0.008),age >75 years old(OR = 2.315,95% CI: 1.048-5.117,p = 0.038),CKD(OR = 3.044,95% CI: 1.173-7.896,p = 0.022),diuretic(OR = 2.369,95% CI: 1.046-5.366,p = 0.039)and rh BNP(OR = 2.818,95% CI: 1.319-6.022,p = 0.007)were independent predictors of CIN.4.Discrimination of 3 risk models improved a lot in the definition of CIN that was recommend by KDIGO(NCDR: AUC = 0.753,Mehran: AUC = 0.718,AGEF: AUC =0.706)than in the definition of CIN that was recommend by ESUR CMSC(NCDR:AUC = 0.679,Mehran: AUC = 0.630,AGEF: AUC = 0.569).5.Mehran model and AGEF model exhibited good discrimination.However,NCDR model obtained Hosmer-Lemeshow p value of 0.024(CIN defined by ESUR CMSC)and Hosmer-Lemeshow p value of 0.118(CIN defined by KDIGO).Conclusion:1.The incidence of CIN is overestimated for patients with normal renal function or with less damage in the definition of serum creatinine increased by 25%.However,the CIN definition recommended by KDIGO is more suitable for these patients.2.CKD,rh BNP,hypertension,anemia,hypotension,diuretics,old age,and cardiac insufficiency are risk factors related to the occurrence of CIN.Controlling these risk factors may protect kidney and reduce the occurrence of CIN.3.Mehran model and AGEF model show favourable performance in contemporary Chinese interventional cardiology when CIN is defined by KDIGO.While NCDR score shows good discrimination,but due to its limitation of calibration,its application in Chinese population needs to be further evaluated.