Androstenedione

What Androstenedione actually measures — the analyte, and the assay

The analyte is androstenedione (A4): a 19-carbon Δ4-3-ketosteroid, 286 Da, with a ketone rather than a hydroxyl at C17. That single missing hydroxyl is why it is a precursor and not an androgen in its own right — it has to be reduced at C17 by 17β-hydroxysteroid dehydrogenase to become testosterone, or aromatized to estrone. It is made in two places at once: the adrenal zona reticularis under ACTH, and the ovarian theca or testicular Leydig cell under LH. A single number is therefore a sum of two signals, which is the first thing that makes it hard to read.

The second thing is the instrument. A4 is measured either by immunoassay — ELISA, RIA, or an automated electrochemiluminescence platform — or by isotope-dilution LC-MS/MS, which separates the molecule by mass before counting it. Those are not two routes to the same number. In a five-site evaluation of the Roche Elecsys androstenedione assay, the same sera regressed against isotope-dilution LC-MS/MS with a slope of 1.040 and r = 0.996 (n = 332), against the LIAISON assay at a slope of 0.625 (r = 0.984, n = 327), and against the IMMULITE assay at a slope of 0.459 with r = 0.856 (n = 320) Obermayer-Pietsch 2022. Read that last line slowly: two commercially available androstenedione assays, run on the same 320 samples, disagreed by more than a factor of two.

An older head-to-head found the same thing in a harsher form. Against LC-MS/MS, an ELISA produced a regression with r² = 0.103 — effectively no relationship — while RIA managed r² = 0.371, and the paper's conclusion was that LC-MS/MS agrees with radioimmunoassay but not with ELISA Yucel 2018. The mechanism is cross-reactivity: an antibody raised against a steroid nucleus cannot cleanly ignore the dozen other steroids in serum that differ from A4 by one hydroxyl, and the ones present at higher concentration — DHEA-S above all — contribute most of the error Ghazal 2022.

So the practical rule: an androstenedione result is not a number, it is a number plus a platform. If your report does not name the method, you cannot compare it with last year's, and you cannot compare it with any cut-off you read anywhere else.

Androstenedione: what changes the blood, and what only changes the reading

What changes the steroid in your blood — largest effect first:

1. A 21-hydroxylase block. In classic and non-classic congenital adrenal hyperplasia the pathway backs up above the block and A4 is one of the products that accumulates; it is used alongside 17-OH progesterone both to detect the condition and to monitor treatment Sarafoglou 2023. This is the largest single biological driver and the only one with a formal interpretive framework behind it.
2. Ovarian theca output. PCOS raises A4, and in some women it is the androgen that is raised when testosterone is not.
3. Menstrual cycle phase. A4 tracks the cycle, so a value drawn on day 20 and one drawn on day 3 are not the same measurement, and a result with no cycle day on the form is close to uninterpretable.
4. Time of day. The adrenal share follows ACTH, which peaks early morning. Afternoon draws read lower.
5. Oral DHEA. DHEA is one enzymatic step upstream; supplementing it raises A4 directly and by an amount nobody can predict from the dose.
6. Glucocorticoids suppress the adrenal share, which is the basis of dexamethasone suppression when the question is adrenal versus ovarian.

What changes only the reading:

1. Which assay ran it. Slope 0.459 against slope 1.040 Obermayer-Pietsch 2022 is a larger effect than most of the physiology above. This belongs at the top of any ranked list of what moves an androstenedione, and it is on no other marker page we can find.
2. Cross-reacting steroids in immunoassay — DHEA-S, 17-OH progesterone, testosterone. This is worst in exactly the patients who get tested: someone with a suspected 21-hydroxylase block has a 17-OH progesterone that may be tenfold normal sitting in the same tube.
3. Biotin, on any streptavidin-based platform, and heterophile antibodies, which bridge the capture and detection antibodies with no steroid involved Ghazal 2022.

Reference interval or decision threshold — which kind of number Androstenedione is

A reference interval, and a method-specific one. It is the central 95% of whichever population that laboratory measured on whichever analyzer it owns — not a value tied to an outcome. Nobody has published a concentration of androstenedione above or below which anything happens to you.

The clearest evidence that the interval belongs to the machine rather than to the species is that laboratories keep having to re-derive it. A 2023 study exists solely to establish reference intervals for androstenedione and DHEA-sulfate in women on the Roche Cobas Bokulić 2023, and the 2022 multicenter evaluation of a new automated assay had to determine its own reference ranges as part of the launch Obermayer-Pietsch 2022. Both are ordinary, correct laboratory practice. Both also mean that quoting somebody else's androstenedione cut-off at your own result is a category error.

The one place a genuine decision threshold lives is inside a diagnostic algorithm — A4 read together with 17-OH progesterone in 21-hydroxylase deficiency, where the numbers guide both diagnosis and the adequacy of treatment Sarafoglou 2023. Those thresholds are assay-specific too, and they are not thresholds for ‘optimal’ anything.

How you would know your Androstenedione was wrong — and when to redraw

The analyte clears in hours, so the retest interval is not set by the steroid. Androstenedione is a small unconjugated steroid with a circulating half-life measured in tens of minutes; what you are holding is this morning's combined adrenal and gonadal output, not this quarter's. That means a redraw tomorrow would be a valid measurement of a different morning — and it is exactly why the retest interval is set by the intervention (8–12 weeks for anything acting on the adrenal or the ovary), not by the molecule.

Four conditions have to match or the second number means nothing: the same laboratory and the same analyzer; a draw between 7 and 9am; in cycling women, the same cycle phase, conventionally days 2–5; and no DHEA in the preceding month.

What would have to change for the retest to mean something. A move of less than roughly a third is inside the range that two manufacturers' assays disagree by on the same serum Obermayer-Pietsch 2022, so if the platform changed between draws, a ‘fall’ is not evidence of anything. A real change should carry its neighbors with it: 17-OH progesterone if the driver is adrenal, DHEA-S if it is adrenal and chronic, total testosterone and SHBG if the driver is ovarian. Androstenedione moving alone, with 17-OH progesterone and DHEA-S flat, is more likely to be the assay than the adrenal.

What Androstenedione cannot tell you

It cannot tell you where the androgen is coming from. That is the single most common thing people ask it to do. A4 is made by both the adrenal and the gonad and rises with either; localizing the source needs the pattern — DHEA-S for the adrenal, testosterone and the LH:FSH relationship for the ovary, 17-OH progesterone for the enzyme block.

It cannot diagnose PCOS. PCOS is a clinical and ultrasound diagnosis with an androgen component; no androstenedione value establishes or excludes it.

A normal A4 does not exclude non-classic congenital adrenal hyperplasia. The marker that answers that question is 17-OH progesterone, and in borderline cases an ACTH-stimulated one Sarafoglou 2023.

And it cannot be compared across laboratories. Given a slope of 0.459 between two real assays Obermayer-Pietsch 2022, a patient who moves clinic can appear to have halved their androstenedione while nothing whatsoever has happened to them. The wrong inference readers draw from this page's marker is that a high number means an adrenal problem; the right first question is which machine produced it.

Sources read for these sections

• Obermayer-Pietsch B, et al. Multicenter Evaluation of a New, Fully Automated Androstenedione Electrochemiluminescence Immunoassay: Precision Analysis, Method Comparison, and Determination of Reference Ranges. Journal of Applied Laboratory Medicine 2022 · PMID 34662384
• Yucel K, et al. Comparison of Immunoassay and Liquid Chromatography-Tandem Mass Spectrometry Methods in the Measurement of Serum Androstenedione Levels. Clinical Laboratory 2018 · PMID 29479885
• Bokulić A, et al. Androgens in women: Establishing reference intervals for dehydroepiandrostenedione sulphate and androstenedione on the Roche Cobas. Biochemia Medica 2023 · PMID 37324111
• Sarafoglou K, et al. Interpretation of Steroid Biomarkers in 21-Hydroxylase Deficiency and Their Use in Disease Management. Journal of Clinical Endocrinology and Metabolism 2023 · PMID 36950738
• Ghazal K, et al. Hormone Immunoassay Interference: A 2021 Update. Annals of Laboratory Medicine 2022 · PMID 34374345