The hydrocortisone-responsive urinary metabolome of premature infants

Abstract

Background: Extremely premature infants are at risk for circulatory collapse or respiratory failure that are often treated with hydrocortisone (HC); however, there is no information on the metabolic consequences of this therapy.

Methods: Longitudinal urine samples from infants <28 weeks gestation in the Trial of Late Surfactant were analyzed by untargeted UHPLC:MS/MS. Fourteen infants who received a tapering course of HC beginning at 3 mg/kg/day for ≥9 days were compared to 14 matched control infants. A secondary cross-sectional analysis by logistic regression used urines from 314 infants.

Results: Of 1145 urinary metabolites detected, abundance of 219, representing all the major biochemical pathways, changed at p < 0.05 in the HC-treated group with 90% decreasing; 3 cortisol derivatives increased ~2-fold with HC therapy. Only 11% of regulated metabolites remained responsive at the lowest HC dose. Regulated metabolites included two steroids and thiamin that are associated with lung inflammation in infants. HC responsiveness was confirmed in 57% of metabolites by cross-sectional analysis.

Conclusions: HC treatment of premature infants influenced in a dose-dependent manner abundance of 19% of identified urinary metabolites of diverse biochemical systems, primarily reducing concentrations. These findings indicate that exposure to HC reversibly impacts the nutritional status of premature infants.

Impact: Hydrocortisone treatment of premature infants with respiratory failure or circulatory collapse alters levels of a subset of urinary metabolites representing all major biochemical pathways. This is the first description of the scope, magnitude, timing and reversibility of metabolomic changes in infants in response to hydrocortisone, and it confirms corticosteroid regulation of three biochemicals that are associated with lung inflammatory status. The findings indicate a dose-dependency of hydrocortisone for metabolomic and anti-inflammatory effects, that prolonged therapy may lower the supply of many nutrients, and that monitoring concentrations of cortisol and inflammation markers may be a useful clinical approach during corticosteroid therapy.

Fig. 1. Excreted cortisol and cortisol derivatives and effect of HC treatment on cortisol glucuronide.

a Relative abundance of urinary cortisol and derivatives normalized to cortisone 21-sulfate. Total urine samples = 332; number of urines with steroids detected are shown in each bar. b Cortisol glucuronide vs cortisone sulfate in HC-treated infants. n = 64 for 14 infants while on HC (range 15–63 days); r = 0.70, p = 8.7e–11 by linear regression. c Cortisol glucuronide vs cortisone urinary levels in control infants. n = 80 for 14 infants during a comparable time period to HC-treated infants (range 10–45 days); r = 0.69, p = 4.0e–10. d Comparison of cortisol glucuronide abundance in HC-treated and control infants. Data are for urine samples collected over 10 days before starting HC (Pre) and 10 days during HC therapy (post). Levels in HC-treated infants increased 1.9- (mean)/2.6-fold (median) and were 2.0-fold higher vs control infants. *p = 0.31; **p = 0.006; n = 14/14 infants and 44/53 (post/pre) samples in control infants and 52/38 in HC-treated infants.

Fig. 2. Examples of urinary metabolite abundance over time in HC-treated and control infants.

a 17hydroxypregnanolone glucuronide in an infant treated with HC days 25–81 and receiving full oral feeds at day 44; b 17hydroxypregnanolone glucuronide in a control infant; c tigloylglycine, a metabolite of isoleucine, in a HC-treated infant days 24–52; d tigloylglycine in a control infant; e 6-sialyl-N-acetyllactosamine, an oligosaccharide, in a HC-treated infant days 14–48; f 6-sialyl-N-acetyllactosamine in a control infant. g Concentrations of erythronate and mevalonolactone in an infant receiving a 10-day course of HC; levels decrease on starting HC and increase 1.6- and 1.9-fold by off/on analysis on stopping HC. h Concentrations of erythronate and mevalonolactone remain lower until transitioning to oral feeds in an infant on HC days 25–81.

Fig. 3. HC suppression in post/pre analysis of named metabolites by Super Pathway assignment.

Box plots with number of suppressed metabolites in each pathway. Significance between pathways by ANOVA with Tukey’s HSD: nucleotide vs lipid p = 0.0008, peptide vs amino acid p = 0.0003, peptide vs cofactor/vitamin p = 0.022, carbohydrate vs lipid p = 0.003. HC suppresses metabolites in each of 8 Super Pathways with significant differences in magnitude.

Fig. 4. Effect of HC on the urinary metabolome by cross-sectional analysis.

a PCA of 1010 metabolites with <30% imputed values. PC1 accounts for 22% of the variance in metabolite abundance between HC-treated and untreated infants with the majority reduced by HC. b Box plot. Significance of PC1 (p = 8.3e–7) and not PC2 (p = 0.65) in PCA analysis. c Volcano plot. Significant differences (above dashed red line) with HC are primarily negative (left of 0) and include metabolites of all super pathways.