A Novel Kidney Failure Prediction Model in Individuals With CKD: Impact of Serum Bilirubin Levels

Abstract

Abstract Context Predicting the progression of chronic kidney disease (CKD) to end-stage kidney disease (ESKD) is crucial for improving patient outcomes. Objective To reveal the highly predictive activity of serum bilirubin levels for the progression of CKD to ESKD, and to develop and validate a novel ESKD prediction model incorporating serum bilirubin levels. Methods We assessed the relative importance of 20 candidate predictors for ESKD, including serum bilirubin levels, in a CKD cohort (15 ≤ estimated glomerular filtration rate [eGFR] < 60 mL/min/1.73 m2), and subsequently developed a prediction model using the selected variables. The development cohort comprised 4103 individuals with CKD who underwent follow-up at Kyushu University Hospital, Japan, from 2008 to 2018. The primary outcome was incident ESKD, defined as an eGFR < 15 mL/min/1.73 m2, chronic dialysis, or renal transplantation. Results The mean follow-up time was 7.0 ± 4.2 years, during which 489 individuals (11.9%) progressed to ESKD. The Cox proportional hazard model selected eGFR, serum bilirubin, proteinuria, age, diabetes, gender, hypertension, serum albumin, and hemoglobin in order of their importance. The predictive performance of the model was optimized by incorporating these 9 variables in discrimination evaluated by time-dependent area under the curve (AUC). This model also demonstrated excellent calibration. Additionally, this model exhibited excellent predictive performance in both discrimination (2-year AUC: 0.943, 5-year AUC: 0.935) and calibration in a validation cohort (n = 2799). Conclusion Serum bilirubin levels were strong predictors for the progression of CKD to ESKD. Our novel model that incorporates serum bilirubin levels could accurately predict ESKD in individuals with CKD.