The Urinary Steroid Metabolome As a Novel Non-Invasive Tool to Stage Non-Alcoholic Fatty Liver Disease

Presentation Number: OR09-3
Date of Presentation: April 4th, 2017

Ahmad Moolla*1, Amin Amin2, Beverly Hughes2, Wiebke Arlt2, Zaki Hassan-Smith2, Lorna Gilligan2, Matt Armstrong3, Philip Newsome3, Tahir Shah3, Luc Van Gaal4, An Verrijken4, Sven Francque4, Jane Grove5, Neil Guha5, Guruprasad Aithal5, Ellie Barnes6, Michael Biehl7 and Jeremy W Tomlinson1
1Oxford Centre for Diabetes, Endocrinology and Metabolism, University of Oxford, UK, 2Centre for Endocrinology, Diabetes & Metabolism, University of Birmingham, UK, 3Centre for Liver Research, University of Birmingham, UK, 4University of Antwerp, Belgium, 5NIHR Nottingham Digestive Diseases Biomedical Research Unit, Nottingham University Hospitals NHS Trust and University of Nottingham, Nottingham, UK, 6Translational Gastroenterology Unit, University of Oxford, UK, 7University of Groningen, The Netherlands



Dysregulated glucocorticoid (GC) metabolism is implicated in the pathogenesis of non-alcoholic fatty liver disease (NAFLD). NAFLD is a spectrum ranging from simple steatosis, to inflammation (steatohepatitis/NASH), fibrosis and cirrhosis which currently requires liver biopsy, an invasive and resource intensive procedure, for diagnosis. It is strongly associated with increased cardiovascular mortality and should be regarded as the hepatic manifestation of metabolic syndrome. Changes to GC metabolism, thus far described in small numbers of patients, relate to cortisol and its metabolites. 11β-hydroxysteroid dehydrogenase type 1 (11β-HSD1) regenerates cortisol (F) from inactive cortisone (E), whilst A-ring reductases 5α and 5β reductase (5αR/5βR) inactivate cortisol to tetrahydrocortisol metabolites (THF/5αTHF). We investigated the urinary steroid metabolome as a novel non-invasive marker for NAFLD and compared it with healthy controls and with alcoholic liver disease which is inseparable from NAFLD on liver pathology.


Using gas chromatography / mass spectrometry, we analysed steroid metabolites in spot urine samples (corrected for creatinine) in a large cohort of patients with biopsy proven NASH (n=62), NAFLD cirrhosis (n=29), healthy controls without liver disease (n=45) and those with alcoholic cirrhosis (n=31). We analysed GC metabolites in all groups and used machine learning-based analysis to investigate changes across the complete urinary steroid metabolome of 33 steroid metabolites.


Cortisol regeneration as reflected by 11β-HSD1 activity (THF+5αTHF/THE ratio), differed significantly between controls and NAFLD cirrhosis (p=0.0085), as well as between controls and alcoholic cirrhosis (p<0.0001). Interestingly, cortisol inactivation, as reflected by A-ring reductase activity (THF/5αTHF ratio) differed significantly between controls and NAFLD cirrhosis (p=0.01), but not between controls and alcoholic cirrhosis (p>0.99). Machine learning-based analysis by generalised matrix learning vector quantisation (GMLVQ) achieved excellent separation of controls and NASH groups (AUC ROC: 0.89). In addition, there was near perfect separation of controls from NAFLD cirrhosis (AUC ROC=0.995), controls from alcoholic cirrhosis (AUC ROC=0.98), and NAFLD cirrhosis from alcoholic cirrhosis (AUC ROC=0.98).


Steroid metabolic pathways appear to be differentially regulated across the spectrum of NAFLD highlighting the potential to identify novel treatment targets. Furthermore, through the adoption of an unbiased computational (GMLVQ) approach, we have raised the potential to use this technique to distinguish NAFLD from alcoholic liver disease, as well as a novel biomarker tool to assess the severity of NAFLD.


Nothing to Disclose: AM, AA, BH, WA, ZH, LG, MA, PN, TS, LV, AV, SF, JG, NG, GA, EB, MB, JWT