Background: Chronic kidney disease (CKD) is an increasing public health issue. It is therefore potentially highly advantageous to identify patients at risk of accelerated renal progression and death. Neutrophil gelatinase-associated lipocalin (NGAL) is an established urinary biomarker for acute kidney injury, but it is not known whether adding urinary NGAL (uNGAL) measurements to conventional risk factors will improve risk assessment in the setting of chronic disease.
Methods: This is a prospective observational cohort study of 158 patients with Stage 3 or 4 CKD. The ability of baseline uNGAL to improve prediction of outcome was assessed by multivariate modelling and a number of metrics including net reclassification analysis. A primary composite endpoint of all-cause mortality or progression to end-stage renal disease (ESRD) requiring renal replacement therapy (RRT) at 2 years and a secondary endpoint of ≥5 mL/min/1.73 m(2) decline in the estimated glomerular filtration rate (eGFR) after 1 year were considered.
Results: Forty patients (25%) reached the primary composite endpoint, 20 of whom died. Twenty-seven patients (19%) reached the secondary endpoint of a ≥5 mL/min/1.73 m(2) decline in the eGFR. The baseline uNGAL-to-creatinine ratio (uNCR) was associated with the combined endpoint of death or initiation of RRT (HR per 5 µg/mmol increase 1.27, 95% CI: 1.01-1.60, P = 0.036) independent of conventional cardiovascular and renal risk factors, including proteinuria. In separate analysis of these two competing endpoints, however, uNCR only remained associated with increased risk of progression to ESRD requiring RRT. Higher baseline uNCR was also independently predictive of rapid renal decline over 1 year (HR per 5 µg/mmol increase 1.47, 95% CI: 1.06-2.06, P = 0.022). Addition of uNCR to the base model resulted in a significant increase in discrimination for the secondary (C-statistic 0.76-0.85, P = 0.001) but not the primary endpoint (P = 0.276). Reclassification analysis on the other hand, demonstrated an improvement in risk predication of both primary and secondary endpoints by incorporating uNCR into the base model, but only in those with low-level urine protein excretion (<28 mg/mmol), with category-free net reclassification improvement (NRI) scores of 64% (95% CI: 8-70; P = 0.019) and 79% (95% CI: 12-83; P = 0.009), respectively.
Conclusion: The utilization of uNCR in addition to conventional established cardiovascular and renal risk factors may improve the prediction of disease progression in elderly Caucasian pre-dialysis CKD patients with low-grade proteinuria.
Keywords: CKD; NGAL; renal decline; risk prediction.