Metabolite profiling of CKD progression in the chronic renal insufficiency cohort study

Abstract

BACKGROUNDMetabolomic profiling in individuals with chronic kidney disease (CKD) has the potential to identify novel biomarkers and provide insight into disease pathogenesis.METHODSWe examined the association between blood metabolites and CKD progression, defined as the subsequent development of end-stage renal disease (ESRD) or estimated glomerular filtrate rate (eGFR) halving, in 1,773 participants of the Chronic Renal Insufficiency Cohort (CRIC) study, 962 participants of the African-American Study of Kidney Disease and Hypertension (AASK), and 5,305 participants of the Atherosclerosis Risk in Communities (ARIC) study.RESULTSIn CRIC, more than half of the measured metabolites were associated with CKD progression in minimally adjusted Cox proportional hazards models, but the number and strength of associations were markedly attenuated by serial adjustment for covariates, particularly eGFR. Ten metabolites were significantly associated with CKD progression in fully adjusted models in CRIC; 3 of these metabolites were also significant in fully adjusted models in AASK and ARIC, highlighting potential markers of glomerular filtration (pseudouridine), histamine metabolism (methylimidazoleacetate), and azotemia (homocitrulline). Our findings also highlight N-acetylserine as a potential marker of kidney tubular function, with significant associations with CKD progression observed in CRIC and ARIC.CONCLUSIONOur findings demonstrate the application of metabolomics to identify potential biomarkers and causal pathways in CKD progression.FUNDINGThis study was supported by the NIH (U01 DK106981, U01 DK106982, U01 DK085689, R01 DK108803, and R01 DK124399).

Keywords: Chronic kidney disease; Nephrology.

Figure 1. CKD progression across the study cohorts.

Flow diagrams depicting the total number of study participants, the number of study participants selected for metabolomic profiling, exclusions from data analysis, and the number of study participants who did or did not have CKD progression in the CRIC Study (left), AASK (middle), and the ARIC Study (right). ESKD, end-stage kidney disease.

Figure 2. Metabolite profiling identifies markers of CKD progression in the CRIC Study.

(A–D) Volcano plots depicting the HR (x axis) and P value (y axis) for CKD progression for each metabolite in Cox proportional hazards models adjusted for Model 1: age, sex, race, and study center (A); Model 2: Model 1 + BMI, systolic blood pressure, diabetes, cardiovascular disease, smoking, alcohol use, and APOL1 genotype (B); Model 3: Model 2 + log PCR (C); and Model 4: Model 3 + eGFR (D). The 10 metabolites significantly associated with CKD progression at FDR < 0.05 in Model 4 are labeled in panel D (n = 1,773).

Figure 3. Metabolite predictors of CKD progression are correlated with eGFR in CRIC.

Histogram showing frequency distribution of Spearman correlations of all measured metabolites with eGFR in CRIC (n = 1,773), with select metabolites associated with CKD progression labeled.