Predictors of Hyporesponsiveness to Erythropoiesis-Stimulating Agents in Patients with Non-Dialysis-Dependent Chronic Kidney Disease (RADIANCE-CKD Study).

ABSTRACT

INTRODUCTION: Hyporesponsiveness to erythropoiesis-stimulating agents (ESAs) has been associated with increased mortality and cardiovascular events in patients with chronic kidney disease. We hypothesized that the prediction of ESA resistance during ESA administration would be very useful in deciding on a treatment plan. METHODS: Patients enrolled in a randomized controlled trial to evaluate renal prognosis in anemic patients with non-dialysis-dependent chronic kidney disease with hyporesponsiveness to ESA were included; the patients had different target hemoglobin levels. A landmark analysis was performed at 3 months into the study. To construct a predictive model for the severe ESA hypo-responder group, in which there was no increase in hemoglobin even with active treatment, background factors and serum test items that affect anemia at study entry were included in a logistic regression model, the area under the curve (AUC) and 95% confidence intervals (CI) were estimated, and sensitivity and specificity were calculated. This study was a post hoc sub-analysis of a randomized controlled trial. RESULTS: The AUC for the 19 existing risk factors as predictors was 0.783 (95% CI 0.711-0.855). Among the 19 risk factors, the combination of six factors (hemoglobin level, systolic blood pressure, weight, gender, smoking status, and hypertensive retinopathy) with the largest χ2 statistics were selected by multiple logistics regression. The AUC for these 6 predictors was 0.716 (95% CI 0.634-0.799). To the six existing risk factors, five serum test items that affect anemia (vitamin B12, vitamin B6, folic acid, parathyroid hormone, and 25-hydroxyvitamin D) were added, for a total of 11 risk factors, with a similar AUC of 0.736 (95% CI 0.655-0.817), sufficient to predict ESA resistance. CONCLUSIONS: Our results suggest that existing risk factors and serum test items can be used to predict ESA resistance in patients with non-dialysis-dependent chronic kidney disease on ESA.