Significance of worsening renal function and nuclear cardiology for predicting cardiac death in patients with known or suspected coronary artery disease

AbstractBackgroundEstimated glomerular filtration rates (eGFRs) at baseline are useful to determine the severity of renal function and to predict cardiac events. However, no studies aimed to demonstrate significance of eGFRs measured during follow-up and usefulness of combination with nuclear cardiology for prediction of cardiac death in patients with coronary artery disease (CAD).MethodsWe retrospectively investigated 1739 patients with known/suspected CAD who underwent myocardial perfusion single photon emission computed tomography (SPECT), who had eGFRs measured at baseline and after one year and who underwent a three-year follow-up. The SPECT images were analyzed with the visual scoring model to estimate summed defect scores. Reduction in eGFRs (ΔeGFR) was defined as the difference between eGFRs measured after one year and at baseline. The endpoint of the follow-up was cardiac deaths within three years after the SPECT, which were identified with medical records or responses to posted questionnaires.ResultsCardiac death was observed in 54 of 1739 patients during the follow-up period (45.6±9.1 months). The multivariate Cox regression analysis showed baseline eGFRs, ΔeGFR, and summed stress scores to be significant independent variables for prediction of cardiac death. The area under receiver operating characteristic curves for detection of cardiac death was 0.677 for the baseline eGFR and 0.802 for the follow-up eGFR. Sensitivity of detection of cardiac death was significantly higher in the follow-up eGFR than in the baseline eGFR (p=0.0002). Combination of the best cut-off values, i.e. 9 for the summed stress scores and 10 for the ΔeGFR, which were suggested by receiver operating characteristic analysis, was useful for risk stratification of cardiac death both in patients with and without chronic kidney disease.ConclusionBaseline and follow-up eGFRs as well as nuclear variables are useful to predict cardiac death in patients with known/suspected CAD