Abstract
Background: The diagnosis of nonclassic 21-hydroxylase deficiency (N21OHD) is considered in children with precocious pubarche, in the differential diagnosis of polycystic ovary syndrome, and in first-degree relatives of patients with 21OHD. Currently, the diagnosis of N21OHD requires cosyntropin (ACTH)-stimulated serum 17α-hydroxyprogesterone (17OHP4) measurements; in addition, 17OHP4 also derives from the ovaries, and basal 17OHP4 is often inconclusive. Objectives: To determine if a panel of steroid biomarkers can diagnose N21OHD in a single blood draw, circumventing the need for dynamic testing. Methods: Patients undergoing evaluation for N21OHD with ACTH stimulation at two tertiary referral centers were included in this study. The diagnosis of N21OHD was based on stimulated serum 17OHP4 concentrations of >1,000 ng/dl (30 nmol/L). Baseline serum samples were used for the quantitation of 22 steroids (18 unconjugated and 4 sulfated steroids) by liquid chromatography-tandem mass spectrometry. Mann Whitney U test was used for two-group steroids comparison. Logistic regression modeling with lasso penalty, combined with clinical knowledge (relevant due to high correlation between biomarkers) were implemented for the selection of a small set of steroids that best discriminate patients with N21OHD from controls. Results: A total of 72 patients (63 females, 9 males) who underwent ACTH stimulation testing, median age 28 (range 6-70 years) were included in this study. Of these, 24 patients (4 males) were diagnosed with N21OHD; the other 48 patients (5 males) in whom N21OHD was excluded served as controls. Age and sex distribution were similar between the two groups (p = 0.9 and 0.5, respectively). Steroids displaying the largest differences between the two groups included: 11-ketotestosterone, 11-ketoandrostenedione (11KA4), 21-deoxycortisol (21dF), 17OHP4 and 16α-hydroxyprogesterone (all higher in N21OHD), plus 11-deoxycorticosterone (DOC) and corticosterone (lower in N21OHD), p <0.0001 for all. Notably, testosterone, androstenedione and all four sulfated steroids were similar between N21OHD patients and controls, including when data analysis was restricted to females. Logistic regression modeling showed that a three-steroid panel incorporating 21dF, 11KA4 and DOC was highly accurate in diagnosing N21OHD in baseline serum samples (AUC, 0.98) and superior to basal 17OHP4. Other combinations of adrenal-derived steroids also provided better discrimination than basal 17OHP4 (AUC >0.97 vs. 0.93). Conclusion: A limited number of adrenal-specific steroids measured in a single baseline blood draw can diagnose N21OHD with high accuracy. Such steroid panels might circumvent the need of dynamic testing in most patients to diagnose or exclude N21OHD.