Serum metabolites are associated with all-cause mortality in chronic kidney disease

Abstract

Chronic kidney disease (CKD) involves significant metabolic abnormalities and has a high mortality rate. Because the levels of serum metabolites in patients with CKD might provide insight into subclinical disease states and risk for future mortality, we determined which serum metabolites reproducibly associate with mortality in CKD using a discovery and replication design. Metabolite levels were quantified via untargeted liquid chromatography and mass spectroscopy from serum samples of 299 patients with CKD in the Modification of Diet in Renal Disease (MDRD) study as a discovery cohort. Six among 622 metabolites were significantly associated with mortality over a median follow-up of 17 years after adjustment for demographic and clinical covariates, including urine protein and measured glomerular filtration rate. We then replicated associations with mortality in 963 patients with CKD from the African American Study of Kidney Disease and Hypertension (AASK) cohort over a median follow-up of ten years. Three of the six metabolites identified in the MDRD cohort replicated in the AASK cohort: fumarate, allantoin, and ribonate, belonging to energy, nucleotide, and carbohydrate pathways, respectively. Point estimates were similar in both studies and in meta-analysis (adjusted hazard ratios 1.63, 1.59, and 1.61, respectively, per doubling of the metabolite). Thus, selected serum metabolites were reproducibly associated with long-term mortality in CKD beyond markers of kidney function in two well characterized cohorts, providing targets for investigation.

Keywords: chronic kidney disease; metabolomics; mortality; uremic toxins.

Figure 1

Hazard ratio and p-values for 622 metabolite associations with mortality in the Modification of Diet in Renal Disease Study (N=299) * The 6 metabolites with associations with mortality at p<0.001 are identified in red. The model adjusted for age, sex, smoking history, history of coronary heart disease, body mass index, systolic blood pressure, log urinary protein, measured glomerular filtration rate, study arms, diabetes and race.

Figure 2

P-values of the 622 metabolites’ adjusted associations with mortality in the Modification of Diet in Renal Disease study (N=299) on the x axis and in the African American Study of Kidney Disease on the y axis (N=963) * The 3 metabolites with associations with mortality at p<0.008 (.05/6) are identified in red. Analyses were adjusted for age, sex, smoking history, history of coronary heart disease, body mass index, systolic blood pressure, log urinary protein, measured glomerular filtration rate, and study arms.

Figure 3

Kaplan-Meier curves of survival in the Modification of Diet in Renal Disease study (top row) and the African American Study of Kidney Disease and Hypertension (bottom row) stratified by median levels of the replicated metabolites (solid line: ≥median; dotted line,

Figure 4

Hazard ratios (HR) and 95% confidence intervals (CI) for associations between metabolites and mortality in the Modification of Diet and Renal Disease and African American Study of Kidney Disease and Hypertension individually as well as across both studies using fixed effects meta-analysis. HRs were estimated in Cox regression models with adjustment for age, sex, smoking history, history of coronary heart disease (cardiovascular disease, coronary artery disease, or peripheral vascular disease), body mass index (BMI), systolic blood pressure (SBP), urine protein, measured glomerular filtration rate (mGFR), and study arms. HRs reflect risk per doubling of metabolite. In the MDRD study, adjustment was also done for diabetes status and race.