Combining steroid and global metabolome profiling by mass spectrometry with machine learning to investigate metabolic risk in benign adrenal tumours with mild autonomous cortisol secretion

Background: Benign adrenal tumours are discovered in 3-10% of adults and can be non-functioning (NFAT) or associated with adrenal hormone excess, most frequently mild autonomous cortisol secretion (MACS) defined by the failure to suppress cortisol after 1 mg dexamethasone overnight but lack of distinct signs of Cushing’s syndrome (CS). We found that MACS increases the prevalence and severity of type 2 diabetes and hypertension and primarily affects women (Ann Int Med. 2022 Doi:10.7326/M21-1737).Objectives: We prospectively recruited 1305 patients with benign adrenal tumours to assess their steroid and global metabolomes and determine links to type 2 diabetes and hypertension.Methods: We analysed 24-h urine samples from 1305 patients (649 NFAT, 591 MACS, 65 CS) using a 17-steroid liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay. We also performed untargeted serum metabolome analysis in a representative sub-cohort of 290 patients (104 NFAT, 140 MACS, 47 CS) employing HILIC and C18-lipidomics LC-MS assays. The data were analysed by two supervised machine learning approaches, generalized matrix learning vector quantization and ordinal regression, to identify the most relevant metabolic changes.Results: Urine steroid metabolome analysis revealed increased glucocorticoid metabolite excretion from NFAT over MACS to CS, whereas androgen metabolite excretion decreased. Similarly, increased glucocorticoid metabolites were observed in patients with type 2 diabetes and hypertension. Lipidome analysis revealed gradual progression towards lipotoxicity with increasing cortisol excess. Patients with type 2 diabetes showed additional changes in acylcarnitines, bioactive lipids, and triacylglycerides.Conclusions: We provide mechanistic insights into the metabolic consequences of cortisol excess. Increased cortisol was linked to a change in the lipidome towards lipotoxicity. Patients with type 2 diabetes and hypertension had increased glucocorticoid output and more adverse changes in the lipidome, indicative of a causative contribution of cortisol excess to their higher cardiometabolic burden. Observed changes may hold promise for risk stratification in MACS, a highly relevant and previously largely overlooked metabolic risk condition