MSACL 2015 EU Abstract

Strategy to Identify and Validate Urinary Steroid Biomarkers Obtained by Untargeted LC-MS/MS Metabolomics: Application to Human Cases of Dioxin Exposure
Fabienne Jeanneret
University of Geneva

Authorship:
Fabienne Jeanneret (1, 2), David Tonoli (1, 2), Denis Hochstrasser (1, 3), Jean-Hilaire Saurat (1, 2), Olivier Sorg (1, 2), Julien Boccard (1), Serge Rudaz (1, 2)
(1) University of Geneva, (2) Swiss Centre for Applied Human Toxicology, (3) Geneva University Hospitals.

Short Abstract

In vitro metabolic reactions were investigated to improve the identification rate from urinary biomarkers obtained by untargeted metabolomic approaches. A previous study performed with UHPLC-QTOF highlighted a subset of 24 urinary steroid and bile acid biomarkers in human cases of acute dioxin exposure. Biosynthesis of glucuronide and sulfate conjugates of biomarker candidates, respectively produced in human liver microsomes and cytosolic fractions, was demonstrated as a successful approach to identify urinary metabolites when authentic chemical standards are not commercially available. Analysis of the biomarkers subset in an independent human cohort exposed to dioxins strengthened the hypothesis of dysregulated profiles of steroids and bile acids.

Long Abstract

Introduction

A previous metabolomic study performed with ultra-high pressure liquid chromatography coupled to quadrupole time-of-flight mass spectrometry highlighted a dysregulation of urinary steroids and bile acids in human cases of acute dioxin exposure. A subset of 24 compounds was highlighted as relevant biomarkers to discriminate controls from dioxin exposed cases, but an unambiguous identification of these compounds was missing. The aim of the current study was i) to evaluate the 24 biomarkers in an independent human cohort exposed to dioxins and, ii) to identify them by comparison with authentic chemical standards and biosynthesized products obtained with in vitro metabolic reactions.

Methods

Urine samples were extracted on HLB cartridges and analysed with UHPLC-QTOF in ESI ion negative mode (Acquity UPLC, Xevo QTOF, Waters). Signals corresponding to the 24 biomarkers were integrated, normalized to the sum of the 24 compounds and submitted to multivariate data analysis (Markerlynx, Waters; SIMCA-P, Umetrics). In vitro metabolic reactions were performed in liver fractions to generate potential matching candidates that were analysed on the same instrument; glucuronide and sulfate conjugates were respectively synthetized in human pooled liver microsomes (HLM) and in human pooled cytosolic fractions.

Results

The 24 biomarkers were studied in a population (Maincy cohort, n= 24) that was exposed to release of dioxins and furans in the incineration fumes of a municipal solid waste incinerator. Unit variance scaling was applied to the dataset and an OPLS-DA model was built to distinguish biomarker profiles of the intoxicated cohort from a control group matched for age. The two groups were separated with reported values of 93.8%, 100% and 87.5% respectively for global accuracy, sensitivity and specificity. These results corroborated the 24 compounds as exposure biomarkers. The profiles obtained for this cohort were compared to those obtained in the previous metabolomic study (cases of acute dioxin exposure). The different types of exposure were highlighted by a co-clustering analysis, but a definite identification was necessary for a better understanding of dioxin toxicity.

Identification process in untargeted metabolomics go through the comparison of two or more orthogonal properties of the biomarker with an authentic chemical standard analysed under identical analytical conditions. The majority of steroids excreted in urine are phase I and II metabolites. As numerous stereoisomers and structural isomers exist, their identification remains challenging. Moreover, source of conjugated steroids and bile acids is scarce, and only a few putative candidates were available for comparison with the proposed biomarkers. Nevertheless, five biomarkers were successfully identified by matching retention time, exact mass and fragmentation spectrum with purchasable products: DHEAS, androsterone 3-glucuronide, androsterone 3-sulfate, pregnanediol 3-glucuronide and oxo-androsterone glucuronide. To further improve the identification rate, the next step was to perform in vitro metabolic reactions for generating standards. Glucuronidation and sulfation reactions were carried out on postulated parent compounds. A match to molecules detected in urine was obtained for glucuronide conjugates of 11-β hydroxyandrosterone, glycochenodeoxycholic acid and glycocholic acid produced in HLM and for glycoursodeoxycholic acid sulfate generated in cytosolic fraction. Numerous other candidates were synthesised. Expected exact masses and fragmentation spectra were obtained, without matching the biomarker retention times. Several synthesised conjugates were observed in urine but no alteration due to dioxin exposure was reported. This illustrates the complexity of the identification process in urinary metabolomics, which is principally due to the presence of numerous isomers.

Conclusions

The 24 biomarker panel was applied successfully to the Maincy validation cohort. Dioxin exposure was shown to be linked with perturbed steroid profiles in urine. In vitro reactions in human liver microsomes and cytosolic fractions was shown as a reliable alternative for urinary biomarker identification when commercial source of authentic chemical standards is missing. Identification studies highlighted a perturbation in the androsterone metabolism. Combining metabolomics with high-resolution mass spectrometry and in vitro metabolic syntheses allowed the discovery and the evaluation of human biomarkers of dioxin exposure. This strategy could be applied to different toxic exposures or diseases.