Longitudinal Plasma Metabolome Patterns and Relation to Kidney Function and Proteinuria in Pediatric CKD

Abstract

Key Points:

1. Longitudinal untargeted metabolomics.

2. Children with CKD have a circulating metabolome that changes over time.

Background: Understanding plasma metabolome patterns in relation to changing kidney function in pediatric CKD is important for continued research for identifying novel biomarkers, characterizing biochemical pathophysiology, and developing targeted interventions. There are a limited number of studies of longitudinal metabolomics and virtually none in pediatric CKD.

Methods: The CKD in Children study is a multi-institutional, prospective cohort that enrolled children aged 6 months to 16 years with eGFR 30–90 ml/min per 1.73 m². Untargeted metabolomics profiling was performed on plasma samples from the baseline, 2-, and 4-year study visits. There were technologic updates in the metabolomic profiling platform used between the baseline and follow-up assays. Statistical approaches were adopted to avoid direct comparison of baseline and follow-up measurements. To identify metabolite associations with eGFR or urine protein-creatinine ratio (UPCR) among all three time points, we applied linear mixed-effects (LME) models. To identify metabolites associated with time, we applied LME models to the 2- and 4-year follow-up data. We applied linear regression analysis to examine associations between change in metabolite level over time (∆level) and change in eGFR (∆eGFR) and UPCR (∆UPCR). We reported significance on the basis of both the false discovery rate (FDR) <0.05 and P < 0.05.

Results: There were 1156 person-visits (N: baseline=626, 2-year=254, 4-year=276) included. There were 622 metabolites with standardized measurements at all three time points. In LME modeling, 406 and 343 metabolites associated with eGFR and UPCR at FDR <0.05, respectively. Among 530 follow-up person-visits, 158 metabolites showed differences over time at FDR <0.05. For participants with complete data at both follow-up visits (n=123), we report 35 metabolites with ∆level–∆eGFR associations significant at FDR <0.05. There were no metabolites with significant ∆level–∆UPCR associations at FDR <0.05. We report 16 metabolites with ∆level–∆UPCR associations at P < 0.05 and associations with UPCR in LME modeling at FDR <0.05.

Conclusions: We characterized longitudinal plasma metabolomic patterns associated with eGFR and UPCR in a large pediatric CKD population. Many of these metabolite signals have been associated with CKD progression, etiology, and proteinuria in previous CKD Biomarkers Consortium studies. There were also novel metabolite associations with eGFR and proteinuria detected.

Graphical abstract

Figure 1

How 622 metabolites relate to eGFR and time. For visualization, the x axis shows the Spearman correlation of metabolite levels with eGFR. Significance of metabolite–eGFR association was determined by LME models with adjustment for age and sex. The y axis shows the log-scale FDR P value of metabolite level differences over time among follow-up measures. FDR, false discovery rate; LME, linear mixed-effects.

Figure 2

How 622 metabolites relate to UPCR and time. For visualization, the x axis shows the Spearman correlation of metabolite levels with UPCR. Significance of metabolite–UPCR association was determined by LME models with adjustment for age and sex. The y axis shows the log-scale FDR P value of metabolite level differences over time among follow-up measures. UPCR, urine protein-creatinine ratio.

Figure 3

One hundred eighty metabolites have unique associations with eGFR on the basis of CKD etiology. In stratified analyses on the basis of glomerular versus non-glomerular CKD etiology, differences in metabolite–eGFR associations were identified on the basis of LME models with adjustment for age and sex. For visualization, the x axis shows Spearman correlation of metabolite levels with eGFR among the glomerular subgroup. The y axis shows Spearman correlation of metabolites levels with eGFR among the non-glomerular subgroup.

Figure 4

Lipid subpathways are enriched among unique metabolite–eGFR associations on the basis of CKD etiology. Among the 180 metabolites with unique associations with eGFR on the basis of glomerular versus non-glomerular CKD etiology, 66 were unique within the glomerular subgroup and 114 in the non-glomerular subgroup. These overall largely consisted of lipid metabolites. Hypergeometric distribution tests showed unique lipid subpathways represented among these associations.