Urinary metabolomics profile in children with ureteropelvic junction obstruction

Fetal hydronephrosis (HN) is detected in about 0.25-1% of fetuses, where ureteropelvic junction obstruction (UPJO) is the most commonly found congenital urinary tract anomaly, remaining the leading cause of kidney failure in infants and children. Since it is a spectral disease, UPJO represents a particularly challenging management. Severe UPJO must be treated surgically to avoid impairment of renal function, but children with non-obstructive hydronephrosis can be treated conservatively. There has been controversy regarding the indication for surgical intervention in asymptomatic patients.

This pilot study aimed at investigating urine samples from children with UPJO using gas- and high-performance liquid-chromatography coupled to mass spectrometry (GC- and HPLC-MS) via untargeted metabolomics to prospect discriminatory metabolites indicative of hydronephrosis spectral progression that may contribute to optimize clinical decision-making.

Thirty-seven patients, whose clinical characteristics were recorded upon visits, had urinary samples collected, processed and analyzed by both GC- and HPLC-MS analytical platforms. Three distinct age-matched groups (2-12 years old) were inspected: 10 children with obstructive HN (OHN) established at initial imaging diagnosis, 15 asymptomatic children with non-obstructive HN (NOHN), and 12 children without any urinary tract problems as control group (CTR). Univariate analysis (including distribution checking and goodness-of-fit testing using ANOVA, Tukey test, and Benjamini-Hochberg procedure for FDR estimation) and multivariate analysis (PLS-DA, using cross-validation and permutation testing) were applied to the metabolomics data. Exploratory ROC curve analyses were performed using combinations of PLS-DA ranked features with VIP > 1 to assess their discriminatory performance.

Metabolomics profiling revealed distinct expression patterns upon selected group comparisons. The inspection of discriminatory metabolites across the different study groups allowed a clear visualization of the disease progression at the molecular level. Moreover, the discrimination of asymptomatic patients (NOHN group) from those requiring surgical intervention (OHN group) is of upmost relevance and could also be delineated here. All univariate and multivariate methods applied to both GC- and HPLC-MS data exhibited good statistical performance (adjusted p-values < 0.05, 0.04 < q-values < 0.1, and 0.662 < AUC < 0.882 for univariate analysis; p-value < 0.03, 3 < F < 11.4, 0.984 < R²<0.996, 0.714 < Q²<0.859, and 0.923 < AUC < 0.995 for multivariate analysis) resulting in a total of 94 discriminatory metabolites, suggesting strong potential clinical value as a diagnostic/prognostic signature. Worth mentioning cystine and methionine sulfoxide exhibiting the largest positive fold change scores, 156% and 152%, respectively (both from OHN vs. CTR group comparison). From the total of 94 discriminatory metabolites, 17 metabolites had particular statistical relevance since they were confirmed by both univariate and multivariate methods combined. Three metabolites, 1-methylhistidine-3-methylhistidine, 1-methyhistamine-3methylhistamine, and glutamyl-hydroxyproline (dipeptide), discriminate NOHN vs. CTR groups whereas nine metabolites, mandelic acid, pantothenic acid, furoylglicine, gluconic acid, 3-hydroxyphenylacetate, 4-hydroxyphenylacetate, glutamylhydroxyproline, 3-hydroxy-3-methylglutaric acid, ribitol/arabitol/xylitol (pentose alcohols) and tagalose discriminate OHN vs. CTR groups; those metabolites may serve as diagnostics purposes, being the first group comparison (NOHN vs. CTR) indicative of the initial metabolic perturbation caused by the UPJO and the latter group comparison (OHN vs. CTR) indicative of the impact UPJO had exerted on the children metabolism at long term. More importantly, five metabolites, ascorbic acid, furoylglycine, gluconic acid, and ribose/arabinose/xylose (pentoses), already discriminating OHN vs. CTR groups, appear again in the OHN vs. NOHN group comparison, in addition to threitol; those metabolites might be useful to support clinical decisions whether a patient should undergo surgery. Metabolomic pathway analyses revealed an alteration of the beta-alanine metabolism at early stages of HN, progressing to further compromising of tyrosine, phenylalanine, alanine/aspartate/glutamate, and amino sugar/nucleotide sugar metabolisms. Several other important metabolic pathways were further compromised at advanced stages of HN revealing the overall impact of UPJO at the basal metabolism of healthy children.

Despite the limited size and age spanning of the cohort evaluated here, the statistical quality of the present study, that certified 17 metabolites by both univariate and multivariate methods among 94 discriminant metabolites, deriving from NOHN, OHN, and CTR group comparisons, may contribute to pave the way into searching valuable biomarkers for the obstructive HN setting up and progression.