Clinical risk scores for the prediction of CIN before primary PCI

Background: Several scores for risk stratification have been developed in candidates to percutaneous or surgical myocardial revascularization. These scores have been recently validated even in different settings than the ones where they were originally developed. We prospectively assessed the relative accuracy of AGEF score, EuroSCORE and Mehran Risk Score (MRS) for the prediction of Contrast-Induced Nephropathy (CIN) in 481 consecutive patients undergoing primary PCI for STEMI. Methods: CIN was defined as an absolute increase in serum creatinine ≥0.5 mg/dL or an increase ≥25% from baseline within 72 hours after the administration of contrast medium. AGEF score was calculated by adding 1 point to the Age/EF(%) ratio if the eGFR was <60 mL/min per 1.73 m2. Logistic regression analysis, receiver-operating characteristic (ROC) curve analysis and Hosmer-Lemeshow χ2 statistic were performed to assess accuracy and calibration of AGEF score, EuroSCORE and MRS as predictors of CIN, with the AUC as a measurement of accuracy. The best cutoff value for each score was identified according to the Youden index. Results: Overall, the incidence of CIN was 5.2%. AGEF score was an accurate (OR 5.19, 95% CI 3.13-8.62, p<0.0001, AUC 0.88) and calibrated (Hosmer-Lemeshow χ2=6.24, p=0.62) predictor of CIN with a 100% sensitivity for AGEF score >1.5 points; all patients developing CIN were in the highest tertile of AGEF score (p<0.0001). When considered linear, continuous variables MRS (OR1.27, 95% CI 1.17-1.39, p<0.0001, Hosmer-Lemeshow χ2=3.18, p=0.53) and EuroSCORE (OR 1.61, 95% CI 1.36-1.91, p<0.0001, Hosmer-Lemeshow χ2=5.39, p=0.50) predicted the risk of CIN as well. Both MRS (AUC 0.80, p=0.15 Vs AGEF score) andEuroSCORE (AUC 0.82, p=0.14 Vs AGEF score) were less accurate, though not significantly, than AGEF score. The cutoff for MRS was 5, with 72% sensitivity and 73.5% specificity, and coincided with the upper boundary of the lowest risk category in the original Mehran study. The cutoff for EuroSCORE was 6, with 92% sensitivity and 59.2% specificity, and coincided with the lower boundaryof the high risk category. Conclusions: In patients undergoing primary PCI for STEMI, a linear risk score based on age, ejection fraction and eGFR can predict the risk of CIN at least as accurately as more complex non-linear risk scores. Simple models based on pre-procedural, readily obtainable objective variables, such as the AGEF score, are well fitted to the acute settings. Complex risk models may be over fitted, at least in populations with a low rate of events.