Endocrine Abstracts

JOINT2792

Introduction: Androstenedione is a steroid hormone produced in the adrenals and gonads, commonly used in the diagnosis and monitoring of hyperandrogenism and congenital adrenal hyperplasia (CAH) in conjunction with other hypothalamic and pituitary hormones. Excess androgens such as androstenedione are also evaluated in conditions including polycystic ovarian syndrome (PCOS) and adrenal tumours. We present here the comparison of three commercially available immunoassays and one isotope dilution liquid chromatography mass-spectrometry method (ID-LC-MS/MS) for androstenedione in frequently investigated patient groups with changes in steroid biosynthesis.

Methods: We evaluated the correlation between the Elecsys Androstenedione (Roche, Mannheim, Germany) and IDS Androstenedione (IDS, Boldon UK) automated immunoassays in a set of 141 remnant routine serum samples (CHU de Liege). The IDS method was additionally compared to the automated Liaison Androstenedione (DiaSorin, Salluggia, Italy) in a set of 119 routine samples (LMU Munich and CHU Liege), and in a set of 94 samples with clinical diagnosis (n = 20 CAH, n = 20PCOS, n = 13 adrenal carcinoma, n = 21 adrenal insufficiency, n = 20 post-menopausal females). Androstenedione levels were also determined with a validated ID-LC-MS/MS (Chromsystems, Gräfelfing, Germany) in the same samples.

Results: The Roche and IDS Androstenedione methods showed excellent correlation (Passing-Bablok regression IDS = 1.05*Roche – 0.11ng/dl, R²=0.99) with no systematic bias across the range of samples (Bland Altman mean bias 5.5%). Agreement between the IDS and Diasorin assays in general routine samples was acceptable, but results were approximately 33% lower with IDS compared to Diasorin (IDS = 0.67*Diasorin + 3.44ng/dl, R²=0.97, mean bias -33.5%). The overall difference between the two assays was similar in the samples from specific patient groups where excess androgens commonly occur (IDS = 0.74*Diasorin – 5.4ng/dl, R²=0.94, mean bias -35.9%). In the same set of samples where values were additionally determined with LC-MS, the IDS had a mean -0.2% bias (IDS =1.18*LC-MS -12.67, R²=0.79) while mean bias on the Diasorin was 38% (Diasorin = 1.56*LC-MS - 51ng/dl, R²=0.97).

Conclusion: Our investigation into the performance of three commercially available automated immunoassays for Androstenedione revealed a close alignment between measurements from the IDS and Roche methods, the latter being standardised to ID-LC-MS/MS. While there was a good agreement overall, the Diasorin assay measured 33-38% higher than the IDS and ID-LC-MS/MS in routine samples and samples where androgens can often exist in excess. Values generated by the IDS assay were also closely aligned to ID-LC-MS/MS, indicating adequate clinical specificity in this recently available method.