Relation of multi-marker panel to incident chronic kidney disease and rapid kidney function decline in African...

Few investigations have evaluated the incremental usefulness of multiple biomarkers representing varying physiological pathways for predicting risk of renal outcomes in African Americans.

We related a multi-marker panel to incident chronic kidney disease (CKD) and rapid kidney function decline (RKFD) in 2813 Jackson Heart Study participants without prevalent CKD at exam 1 (2000-2004) and with complete assays at exam 1 for 9 biomarkers: adiponectin, aldosterone, B-natriuretic peptide [BNP], cortisol, high sensitivity C-reactive protein (hsCRP), endothelin, homocysteine, plasma renin activity and mass. Incident CKD was defined as estimated glomerular filtration rate (eGFR) < 60 mL/min/1.73 m² at exam 3 while RKFD was defined as eGFR ≥30% loss between exams 1 and 3 (8.2 median years). We employed multiple logistic regression model to describe association between the panel and incident CKD and RKFD and used backward elimination strategy to estimate the most parsimonious biomarker model while controlling for conventional risk factors.

The multi-marker panel predicted the risk for both incident CKD (odds ratios [OR], 2.72; 95% confidence intervals [CI], 1.63, 4.56; P = 0.001) and RKFD (2.61; 95% CI, 1.67, 4.08; P < 0.001). Per standard deviation increase in log biomarker concentrations were significantly (multivariable adjusted odds ratios, [95% confidence interval], p-value) associated with incident CKD: plasma adiponectin (1.24 [1.07, 1.44], p = 0.005) and leptin (1.3 [1.06, 1.61], p = 0.011), and with RKFD: plasma adiponectin (1.22 [1.06, 1.40], p = 0.006); hsCRP (1.17 [1.01, 1.36], p = 0.031) and aldosterone (0.85 [0.74, 0.96], p = 0.012). Moderate levels (3rd quartile) of aldosterone were inversely associated with incident CKD (0.54 [0.35, 0.82], p = 0.004) while leptin was associated with RKFD (1.64 [1.10, 2.44], p = 0.015). Biomarkers improved CKD risk prediction (P = 0.003) but not RKFD risk prediction (P = 0.10).

In this community-based sample of African Americans, a multi-marker panel added only moderate predictive improvement compared to conventional risk factors.