Metabolome-derived kidney function scores as prognostic readouts of cardiovascular-kidney-metabolic health.

Kidney disease is a rapidly growing global health concern, tightly linked to cardiovascular and metabolic disease within the recently defined cardiovascular-kidney-metabolic (CKM) syndrome. Routine clinical assessment of kidney function primarily relies on creatinine-based estimated glomerular filtration rates (eGFRcr) which can be affected by non-renal factors and declines only after substantial kidney damage. Using data from the MetaCardis study (N=1,723), we derived metabolomic kidney function gaps (mKFGs)-the difference between metabolome-predicted eGFRcr and measured eGFRcr. A physiology-guided feature-selection strategy - anchored on three cardiometabolic markers (haemoglobin, albumin and the Neutrophil-to-Lymphocyte Ratio) that associate with mKFGs but not with eGFRcr - yielded compact sex-stratified metabolite panels (40-98 out of 1,514 metabolites) that tracked cardiovascular-kidney-metabolic risk. Machine-Learning models revealed distinct metabolic signatures, including biomarkers of meat consumption and androgens, highlighting potential sex-specific pathways linked to CKM health. External validation in the general-population Canadian Longitudinal Study on Aging (N=9,958) and German Chronic Kidney Disease Study (N=5,113) cohorts revealed that mKFGs added incremental predictive value for future CKM events and all-cause mortality beyond eGFRcr and established risk equations. Our findings open new avenues for personalised risk stratification for CKM disease progression and targeted interventions based on metabolically defined kidney health.