NMR-based Urinary Biomarkers in Pediatric Primary Mitochondrial Disorders and Chronic Kidney Disease: Shared Mitochondrial Dysfunction, Diverging Biosignatures

Background: Renal involvement is a frequent manifestation of primary mitochondrial disorders (PMD), either as a presenting feature or during the disease course. Simultaneously, the metabolomic profile of chronic kidney disease (CKD) is often associated with underlying mitochondrial dysfunction. This study aimed to characterize urinary metabolic signatures in genetically confirmed paediatric PMD without chronic kidney disease, comparing them to healthy controls, suspected (unconfirmed) mitochondrial disease (SMD), and non-mitochondrial CKD. Methods: We performed untargeted ¹ H-NMR metabolomic profiling of 76 urine samples from 61 paediatric patients. Outlier samples and patients undergoing acute decompensation were excluded from the main analyses. Final comparisons included genetically confirmed PMD without CKD (n = 13), SMD (n = 10), non-mitochondrial CKD (n = 28; 17 at stages 1–2 and 9 at stages 3–5), and healthy controls (n = 10). Spectral data were analyzed using multivariate techniques—including principal component analysis (PCA) and partial least squares–discriminant analysis (PLS-DA)—as well as univariate methods following normalization to urinary creatinine or total spectral area. Results: Urinary metabolic profiles of PMD patients differed markedly from those of healthy controls and CKD patients. Multivariate analysis (PLS-DA) revealed a high discriminative ability for separating PMD from controls (Q² = 0.52) and CKD (Q² = 0.69). Several metabolites were significantly elevated in PMD, including 4-aminohippurate, homovanillic acid (HVA), cis -aconitate, and fumarate. These remained discriminative when comparing PMD to both CKD and control groups. A multimetabolite panel comprising 4-aminohippurate, HVA, histidine, and cis-aconitate achieved high diagnostic performance in distinguishing PMD from advanced CKD (stage 3–5), with an area under the curve (AUC) of 0.944. This biosignature integrated metabolites involved in distinct functional domains - energy metabolism, amino acid handling, and renal tubular function. Correlations between normalization strategies (creatinine vs. total spectral area) varied across metabolites (Pearson r ranging from –0.32 to 0.98), highlighting the potential impact of normalization choice on quantitative interpretation and biomarker reproducibility. Conclusion: Urinary metabolomic profiling by ¹ H-NMR revealed a distinct biosignature in pediatric PMD patients without renal involvement, characterized by elevated levels of 4-aminohippurate, HVA, and tricarboxylic cycle intermediates. The consistent increase in 4-aminohippurate and HVA—both reliant on energy-dependent proximal tubular excretion—points to impaired mitochondrial modulation of tubular function, even in the absence of overt renal disease. These findings support the use of urinary ¹ H-NMR metabolomics as a non-invasive tool for biomarker discovery in PMD and highlight the potential of integrated, multiparametric metabolic fingerprints for diagnostic refinement and patient stratification.