Epimer-Specific Diagnostic Contribution: α- and β-Cortolone as Independent Predictors in Adrenal Tumor Classification Outperform Cortisol Alone
In a study of 260 subjects (48 lean controls, 143 obese controls, 43 benign adrenal tumor patients, 26 Cushing's disease patients), a logistic regression model retaining cortisol, THF, and α- and β-cortolone as significant independent analytes generated an ROC AUC of 0.927 for discriminating secreting adrenal tumors (AT+) from controls—a 27.2% absolute improvement over cortisol alone (AUC 0.729; P < 0.0001). For silent adrenal tumors (AT−), a model retaining cortisol, allo-THF, and α- and β-cortolone yielded AUC 0.910 vs. cortisol-alone AUC 0.635 (P < 0.0001), a 43.3% relative improvement [1]. Both epimers contributed independently to model performance; exclusion of either epimer was not tested in the reported model, but the epimers entered as separate analytes in the stepwise logistic regression, confirming non-redundant information content [1]. This establishes that the 20β-epimer carries diagnostic information distinct from the 20α-epimer.
| Evidence Dimension | Diagnostic classification performance (ROC AUC) for adrenal tumor detection |
|---|---|
| Target Compound Data | Model containing cortisol + THF + α-cortolone + β-cortolone: AUC 0.927 (AT+ vs. controls); model containing cortisol + allo-THF + α-cortolone + β-cortolone: AUC 0.910 (AT− vs. controls) |
| Comparator Or Baseline | Cortisol alone: AUC 0.729 (AT+ vs. controls); AUC 0.635 (AT− vs. controls) |
| Quantified Difference | Δ AUC: +0.198 (AT+); +0.275 (AT−). P < 0.0001 for both comparisons. Relative improvement: 27.2% (AT+); 43.3% (AT−). |
| Conditions | Multiplex LC-MS/MS urinary steroid metabolite quantification; logistic regression with stepwise variable selection; n = 260 subjects (48 Nl, 143 Ob, 43 AT, 26 CD) |
Why This Matters
Procurement of defined 20β-cortolone-3-glucuronide—rather than an undefined epimeric mixture—is essential for replicating the validated multi-analyte diagnostic panel, as the independent contribution of each epimer to the ROC model was confirmed by their retention as separate significant variables.
- [1] Brossaud J, Ducint D, Corcuff JB. Urinary glucocorticoid metabolites: biomarkers to classify adrenal incidentalomas? Clin Endocrinol. 2015;82(2):236-243. doi:10.1111/cen.12717 View Source
