Using Machine Learning to Identify Metabolomic Signatures of Pediatric Chronic Kidney Disease Etiology.

BACKGROUND: Untargeted plasma metabolomic profiling combined with machine learning (ML) may lead to discovery of metabolic profiles that inform our understanding of pediatric CKD causes. We sought to identify metabolomic signatures in pediatric CKD based on diagnosis: FSGS, obstructive uropathy (OU), aplasia/dysplasia/hypoplasia (A/D/H), and reflux nephropathy (RN). METHODS: Untargeted metabolomic quantification (GC-MS/LC-MS, Metabolon) was performed on plasma from 702 Chronic Kidney Disease in Children study participants (n: FSGS=63, OU=122, A/D/H=109, and RN=86). Lasso regression was used for feature selection, adjusting for clinical covariates. Four methods were then applied to stratify significance: logistic regression, support vector machine, random forest, and extreme gradient boosting. ML training was performed on 80% total cohort subsets and validated on 20% holdout subsets. Important features were selected based on being significant in at least two of the four modeling approaches. We additionally performed pathway enrichment analysis to identify metabolic subpathways associated with CKD cause. RESULTS: ML models were evaluated on holdout subsets with receiver-operator and precision-recall area-under-the-curve, F1 score, and Matthews correlation coefficient. ML models outperformed no-skill prediction. Metabolomic profiles were identified based on cause. FSGS was associated with the sphingomyelin-ceramide axis. FSGS was also associated with individual plasmalogen metabolites and the subpathway. OU was associated with gut microbiome-derived histidine metabolites. CONCLUSION: ML models identified metabolomic signatures based on CKD cause. Using ML techniques in conjunction with traditional biostatistics, we demonstrated that sphingomyelin-ceramide and plasmalogen dysmetabolism are associated with FSGS and that gut microbiome-derived histidine metabolites are associated with OU.

The Association of Aging Biomarkers, Interstitial Lung Abnormalities, and Mortality.

Rationale: The association between aging and idiopathic pulmonary fibrosis has been established. The associations between aging-related biomarkers and interstitial lung abnormalities (ILA) have not been comprehensively evaluated.Objectives: To evaluate the associations among aging biomarkers, ILA, and all-cause mortality.Methods: In the FHS (Framingham Heart Study), we evaluated associations among plasma biomarkers (IL-6, CRP [C-reactive protein], TNFR [tumor necrosis factor α receptor II], GDF15 [growth differentiation factor 15], cystatin-C, HGBA1C [Hb A1C], insulin, IGF1 [insulin-like growth factor 1], and IGFBP1 [IGF binding protein 1] and IGFBP3]), ILA, and mortality. Causal inference analysis was used to determine whether biomarkers mediated age. GDF15 results were replicated in the COPDGene (Genetic Epidemiology of Chronic Obstructive Pulmonary Disease) Study.Measurements and Main Results: In the FHS, there were higher odds of ILA per increase in natural log-transformed GDF15 (odds ratio [95% confidence interval], 3.4 [1.8-6.4]; P = 0.0002), TNFR (3.1 [1.6-5.8]; P = 0.004), IL-6 (1.8 [1.4-2.4]; P < 0.0001), and CRP (1.7 [1.3-2.0]; P < 0.0001). In the FHS, after adjustment for multiple comparisons, no biomarker was associated with increased mortality, but the associations of GDF15 (hazard ratio, 2.0 [1.1-3.5]; P = 0.02), TNFR (1.8 [1.0-3.3]; P = 0.05), and IGFBP1 (1.3 [1.1-1.7]; P = 0.01) approached significance. In the COPDGene Study, higher natural log-transformed GDF15 was associated with ILA (odds ratio, 8.1 [3.1-21.4]; P < 0.0001) and mortality (hazard ratio, 1.6 [1.1-2.2]; P = 0.01). Causal inference analysis showed that the association of age with ILA was mediated by IL-6 (P < 0.0001) and TNFR (P = 0.002) and was likely mediated by GDF15 (P = 0.008) in the FHS and was mediated by GDF15 (P = 0.001) in the COPDGene Study.Conclusions: Some aging-related biomarkers are associated with ILA. GDF15, in particular, may explain some of the associations among age, ILA, and mortality.

Plasma Biomarkers of Tubular Injury and Inflammation Are Associated with CKD Progression in Children.

BACKGROUND: After accounting for known risk factors for CKD progression in children, clinical outcomes among children with CKD still vary substantially. Biomarkers of tubular injury (such as KIM-1), repair (such as YKL-40), or inflammation (such as MCP-1, suPAR, TNF receptor-1 [TNFR-1], and TNFR-2) may identify children with CKD at risk for GFR decline. METHODS: We investigated whether plasma KIM-1, YKL-40, MCP-1, suPAR, TNFR-1, and TNFR-2 are associated with GFR decline in children with CKD and in subgroups defined by glomerular versus nonglomerular cause of CKD. We studied participants of the prospective CKiD Cohort Study which enrolled children with an eGFR of 30-90 ml/min per 1.73 m2 and then assessed eGFR annually. Biomarkers were measured in plasma collected 5 months after study enrollment. The primary endpoint was CKD progression, defined as a composite of a 50% decline in eGFR or incident ESKD. RESULTS: Of the 651 children evaluated (median age 11 years; median baseline eGFR of 53 ml/min per 1.73 m2), 195 (30%) had a glomerular cause of CKD. Over a median follow-up of 5.7 years, 223 children (34%) experienced CKD progression to the composite endpoint. After multivariable adjustment, children with a plasma KIM-1, TNFR-1, or TNFR-2 concentration in the highest quartile were at significantly higher risk of CKD progression compared with children with a concentration for the respective biomarker in the lowest quartile (a 4-fold higher risk for KIM-1 and TNFR-1 and a 2-fold higher risk for TNFR-2). Plasma MCP-1, suPAR, and YKL-40 were not independently associated with progression. When stratified by glomerular versus nonglomerular etiology of CKD, effect estimates did not differ significantly. CONCLUSIONS: Higher plasma KIM-1, TNFR-1, and TNFR-2 are independently associated with CKD progression in children.

Hand osteoarthritis in relation to mortality and incidence of cardiovascular disease: data from the Framingham heart study.

OBJECTIVES: To study whether hand osteoarthritis (OA) is associated with increased mortality and cardiovascular events in a large community based cohort (Framingham Heart Study) in which OA, mortality and cardiovascular events have been carefully assessed. METHODS: We examined whether symptomatic (≥1 joint(s) with radiographic OA and pain in the same joint) and radiographic hand OA (≥1 joint(s) with radiographic OA without pain) were associated with mortality and incident cardiovascular events (coronary heart disease, congestive heart failure and/or atherothrombotic brain infarction) using Cox proportional hazards models. In the adjusted models, we included possible confounding factors from baseline (eg, metabolic factors, medication use, smoking/alcohol). We also adjusted for the number of painful joints in the lower limb and physical inactivity. RESULTS: We evaluated 1348 participants (53.8% women) with mean (SD) age of 62.2 (8.2) years, of whom 540 (40.1%) and 186 (13.8%) had radiographic and symptomatic hand OA, respectively. There was no association between hand OA and mortality. Although there was no significant relation to incident cardiovascular events overall or a relation of radiographic hand OA with events, we found a significant association between symptomatic hand OA and incident coronary heart disease (myocardial infarction/coronary insufficiency syndrome) (HR 2.26, 95% CI 1.22 to 4.18). The association remained after additional adjustment for pain in the lower limb or physical inactivity. CONCLUSIONS: Symptomatic hand OA, but not radiographic hand OA, was associated with an increased risk of coronary heart disease events. The results suggest an effect of pain, which may be a possible marker of inflammation.