Prolactin finds the tumours — and is wrong more often than anything else on the panel.

This test finds the tumours. It is also wrong more often than anything else on the panel, and it can be wrong in both directions.

Why it belongs in the workup

The standard account is that a high prolactin suppresses the pulsatile GnRH signal driving LH, and that without LH the testis stops making testosterone. The clearest human demonstration is small — four hyperprolactinemic men in whom LH pulse frequency was subnormal in three, and in whom bromocriptine normalised prolactin and restored testosterone, libido and LH pulses together.

Four patients is a thin foundation for a mechanism this widely taught, and it has not been uniformly reproduced. A separate series of men with macroprolactinoma found no significant difference in mean LH, LH pulse frequency or pulse amplitude, and concluded that hypothalamic-pituitary function was preserved. The direction of the effect is not in doubt. The mechanism is less settled than it is usually presented.

Evidence: Promising — for the mechanism specifically.

The reason to measure it is not the mechanism anyway. It is what you find. In 164 men with erectile dysfunction and repeatedly low testosterone, imaging the pituitary found a potentially serious lesion in 11 — 6.7%: five microadenomas, four macroadenomas, two hypothalamic lesions. Mean testosterone in the men with a lesion was 121 ng/dL against 177 in those without.

That is the case for the test. One man in fifteen with genuinely low testosterone and low gonadotropins has something in his head that needs finding.

Evidence: Established.

When each body says to order it

Body	What it says
AUA (2018, confirmed 2024)	“Serum prolactin levels should be measured in patients with low testosterone levels combined with low or low/normal luteinizing hormone levels.” Strong Recommendation, Evidence Level Grade A — and a second Grade A statement that persistently high prolactin of unknown cause needs endocrine evaluation
Endocrine Society (2018)	In men with secondary hypogonadism, “clinicians should perform serum prolactin and iron saturation measurements.” This is a technical remark, not a graded recommendation, so it carries no GRADE rating of its own
EAU (2026)	“Measure prolactin (PRL) levels if evidence of low sexual desire (or other suggestive signs/symptoms) and secondary hypogonadism is present.” Strong. Note the extra condition — the EAU asks for symptoms as well
VA (January 2026)	“In cases of low or inappropriately normal LH/FSH concentrations, further pituitary evaluation includes serum prolactin, iron studies and in some cases pituitary imaging”

One correction worth making because it circulates widely. The Endocrine Society’s 150 ng/dL figure is a threshold for pituitary imaging, not for measuring prolactin. In that guideline prolactin is measured in every man with secondary hypogonadism, whatever his testosterone. The 150 ng/dL number belongs to a different sentence — about who gets an MRI — and the AUA uses the same figure the same way, adding that men below it with a low LH “should undergo a pituitary MRI regardless of prolactin levels, as non-secreting adenomas may be identified.”

The specialist guidance has also moved. The Endocrine Society’s 2011 hyperprolactinemia guideline has been functionally superseded for prolactinoma by a 2023 Pituitary Society international consensus statement — which reverses one of its central recommendations, as below.

Three ways the number is wrong

This is the part of the page that matters most.

Macroprolactin — falsely high

Prolactin can circulate bound into a large complex with immunoglobulin. The complex is measured by the assay and is biologically inactive. The result reads high; the man is fine.

How often? Primary audits give 22% of hyperprolactinemic samples in a five-year series of 2,089, and 24% in a series of 300. A guideline sentence puts it at “approximately 40%” — that is the highest figure in circulation and it is a guideline statement rather than a measurement, so we quote the audits.

Two received rules about macroprolactin are wrong.

It is not confined to mild elevations. In the 2,089-sample audit, macroprolactin explained 18%, 21%, 19% and 17% of samples in the 700–1,000, 1,000–2,000, 2,000–3,000 and above-3,000 mU/L bands respectively — essentially flat across the whole range. The 2023 consensus nonetheless restricts macroprolactin testing to results below 200 ng/mL, and we would flag that as a genuine conflict between the consensus and the audit data rather than resolve it.

It is not confined to asymptomatic men. The same audit found macroprolactinemic patients “could not be differentiated from true hyperprolactinemic patients on the basis of clinical features alone.” In a series of 160 patients, symptoms overall were less frequent in macroprolactinemia — but erectile dysfunction occurred at the same rate in both groups, as did galactorrhea. Imaging showed a microadenoma-like appearance in 11% of the macroprolactinemic patients.

The consequence of not checking is measurable. Among 21 retrospectively identified macroprolactinemic patients at one centre — all of them women — pituitary imaging had been performed in 93% and a dopamine agonist prescribed in 87%. Introducing routine screening produced a net cost saving.

The screen is simple: precipitate the serum with polyethylene glycol and re-assay the supernatant. The conventional cutoff is recovery below 40%, though there is a published argument that reporting the absolute post-precipitation concentration against a monomeric reference interval misclassifies fewer people.

Evidence: Established.

The hook effect — falsely low, and it is a male problem

At extremely high concentrations, some sandwich immunoassays saturate and report a low number. A man with a giant prolactinoma can have a prolactin reported as mildly raised — indistinguishable from a stalk effect — and be sent for surgery for a tumour that would have melted on a tablet.

It is not a curiosity. In one consecutive surgical series it affected 4 of 69 patients with a pituitary macroadenoma (5.8%); in another, 4 of 28 (14.2%). The reported values were 2,110 mU/L at the median in one series and 97 µg/L in the other; the true values after dilution were 384,720 mU/L and 5,795 µg/L — underestimates of roughly 180-fold and 60-fold.

And the sex distribution is striking. In the first series the hook-effect group was 4 out of 4 male, against 9.9% male among macroprolactinomas generally. In the second, 75% male. This artefact selects for exactly the population this site is written for.

All four of the hooked macroprolactinomas in the second series shrank by more than half on a dopamine agonist. The artefact, not the tumour, was what would have sent them to theatre.

The fix is serial dilution — 1:100 per both guidelines, 1:1 and 1:10 per the older series. Modern platforms claim resistance to it. Whether yours is resistant is a question for your laboratory, not for a website.

Evidence: Established.

The needle itself — falsely high, but less than you were told

Prolactin rises with the stress of venepuncture, which is why laboratories tell men to rest before the draw. The best data suggest the instruction is right and the reasoning is wrong.

In 235 men and women referred with a single raised prolactin — with macroprolactin, dopamine antagonists and known pituitary disease excluded — 64 (27%) normalised on retesting through a cannula. But of those 64, 41 normalised on the immediate repeat sample alone, and only 23 normalised after a further 120 minutes of rest. Most of the benefit came from drawing a second tube, not from lying down.

A useful number from that study: a referral prolactin of 2,000 mU/L (about 94 ng/mL) or above had 97% specificity for true hyperprolactinemia. Below that, repeat it.

The ubiquitous instruction to rest for 30 minutes has no traceable primary source. That study used 120 minutes. The 2011 guideline says only “without excessive venipuncture stress,” with no duration. The 2023 consensus recommends cannulated sampling “if an influence of stress is suspected,” also with no duration. Thirty minutes is a laboratory convention.

The 2023 consensus reverses the older position outright: “Patients with hyperprolactinaemia but serum levels of prolactin less than five times the upper limit of normal should undergo repeat prolactin testing.” The 2011 guideline had recommended a single measurement.

Evidence: Moderate.

And a fourth: biotin, in the direction nobody expects

High-dose biotin interferes with streptavidin-based immunoassays, and prolactin is among the analytes affected. The direction depends on the assay format — and prolactin on the common platforms is a sandwich assay, which biotin biases downward, not upward. The frequently repeated claim that biotin falsely raises prolactin has the sign backwards for those platforms. If you take a high-dose biotin supplement, stop it and tell the laboratory.

The reference range, and the bands

There is no single male prolactin reference interval. Prolactin is standardised to a common international standard but not harmonised: across six analysers, the upper limit of the male monomeric reference interval ranged from 196 to 301 mU/L — and in the presence of macroprolactin, the same serum read 2.3 to 7.8 times differently across nine assay systems.

The familiar bands — this level means a drug, that level means a stalk effect, that one means a microadenoma — are convention rather than measurement, and we could not trace them to a derivation study. The figures that do have a guideline behind them are that drug-induced hyperprolactinemia usually runs 25 to 100 µg/L, that metoclopramide, risperidone and phenothiazines can exceed 200, and that above 250 µg/L usually indicates a prolactinoma. The 2023 consensus gives no numeric bands at all.

The one boundary derived from primary data is better than any of them. In 226 histologically verified non-functioning pituitary macroadenomas, of which 146 were male, prolactin was below 2,000 mU/L in 98.7%. The authors’ conclusion: “serum PRL > 2000 mU/l is almost never encountered in nonfunctioning pituitary macroadenomas.” That is a real, usable discriminator. Below it, the number does not tell you what you have.

What raises it

Drugs, first and most often — antipsychotics above all, with metoclopramide and domperidone close behind, and and, in feminising hormone therapy, cyproterone acetate — which is where most of that rise comes from, since estrogen with spironolactone raised prolactin by only about 3 µg/L. Review the medication list before ordering an MRI. The 2023 consensus makes this a strong recommendation in its own right.

Then: a prolactinoma, any sellar mass compressing the stalk, and macroprolactin.

The rest of the standard list — primary hypothyroidism, chronic kidney disease, cirrhosis, chest wall lesions, seizures, and some antidepressants — we carry on the authority of the guidelines and textbooks rather than on primary male data. In the case of chest wall lesions we could not find a primary source at all, which is worth knowing about an item that appears in every review.

Two things commonly listed that the male data do not support. Opioids — two male datasets found no prolactin difference, and opioid hypogonadism appears to be mediated through GnRH and LH rather than prolactin. Cannabis — controlled human male data show a non-significant fall, or no effect at all.

Does treating it restore testosterone?

Partly, in about half to two-thirds, mostly in the first year — and a substantial minority never recover. This is worth knowing before anyone promises otherwise.

In 97 men with prolactinoma on a dopamine agonist, testosterone rose from 5.2 nmol/L at baseline to 9.6 at six months and 13.1 at twelve, then plateaued. Recovery from hypogonadism was 43% at six months, 50% at twelve, and 54% at twenty-four — rising to 69% among the men whose prolactin actually normalised. The best single predictor of persistent hypogonadism was a testosterone still below 7.4 nmol/L at six months, with 91% sensitivity and 94% specificity.

The sharper finding comes from a study restricted to men with macroprolactinoma who all achieved a normal prolactin on cabergoline. Only 8 of 30 (26.7%) reached eugonadism. And the pattern of the residual deficit is informative: those men had “suppression of the luteinizing hormone-testosterone axis with sparing of other anterior pituitary hormonal axes, including follicle-stimulating hormone-inhibin B” — an isolated, selective LH suppression that the authors read as functional rather than structural gonadotroph damage.

At the extreme end, among 32 published patients with prolactin above 10,000 ng/mL — 81% of them men — only one achieved a normal testosterone, even with surgery or radiotherapy added.

There is also a small open-label finding that cuts against the assumption that prolactin must be normalised first: clomiphene restored testosterone in 10 of 14 men with prolactinoma on dopamine agonists, including five who remained hyperprolactinemic. Fourteen men, single arm, no control — take the direction, not the number.

Evidence: Moderate.

On testosterone therapy

No guideline recommends routine prolactin monitoring on testosterone therapy. We checked all four. The Endocrine Society’s monitoring plan is testosterone, hematocrit and prostate cancer risk. The AUA’s prolactin statements are both pre-treatment. The EAU’s is in initial evaluation only. The VA’s appears once, in the secondary-hypogonadism workup.

In ordinary hypogonadal men, testosterone appears not to move prolactin — two small studies, one in men with Klinefelter syndrome and one in 24 men with late-onset hypogonadism, found no change.

There is one exception with primary evidence behind it. In a cohort of 53 men with prolactinoma, five had a significant prolactin rise on testosterone therapy; in the four analysed, prolactin rose by 231% to 396%, and adding letrozole brought it down by 61% to 93%. The implied mechanism is aromatization of the testosterone to estradiol driving lactotroph secretion. Four men, retrospective, in a resistant-prolactinoma population — so this is a reason to re-check prolactin in a man with a known prolactinoma who starts testosterone, and not a reason to check it in anyone else.

Related

The Lab Library · LH · Why is testosterone low? · When testosterone isn’t the answer

Questions patients ask