Mild kidney dysfunction affects the predictive accuracy of blood-based biomarkers for neuropsychological and neuroimaging outcomes over a 9 year follow-up period

Abstract

INTRODUCTION: The impact of chronic kidney disease (CKD) on Alzheimer's disease (AD) plasma biomarkers is poorly understood. We tested whether kidney function decline affects the predictive accuracy of plasma biomarkers on neuropsychological or neuroimaging outcomes. METHODS: Three hundred thirty-three non-demented older adults were included. Linear regressions related plasma glial fibrillary acidic protein, neurofilament light chain (NfL), amyloid beta 42, and phosphorylated tau231, with a blood-based biomarker x estimated glomerular filtration rate (eGFR) interaction term, to cross-sectional and longitudinal (mean follow-up time = 6.4 ± 2.8 years) neuropsychological and neuroimaging outcomes. RESULTS: Plasma NfL interacted with eGFR on the longitudinal trajectory of nearly all neuropsychological outcomes (p values < 0.02) and several neuroimaging outcomes (p values < 0.02). Associations were stronger in individuals with no CKD/stage 1 and stage 2 CKD, while associations were weaker or not significant in individuals with stage 3 CKD. DISCUSSION: Among older adults free of severe CKD, the ability of plasma NfL to predict key AD-related biomarker outcomes was moderated by renal function. Highlights: Reduced kidney function was associated with increased blood biomarker levels. The predictive accuracy of neurofilament light chain was reduced in participants with kidney dysfunction. Even mild kidney dysfunction can affect blood biomarkers of Alzheimer's disease.