LH: the test that tells you where the problem is

This is the test that tells you where the problem is, and it is measured worse than any other number on the panel.

What it measures

Luteinizing hormone is released by the pituitary and drives the Leydig cells of the testis to make testosterone. When testosterone falls, LH is supposed to rise. Whether it does is the whole diagnostic question.

High LH with low testosterone means the pituitary is asking and the testis is not answering — a testicular problem, called primary hypogonadism.

Low or unremarkable LH with low testosterone means the pituitary is not asking — a hypothalamic or pituitary problem, called secondary hypogonadism. That one has a differential that includes things you would very much want to find: a prolactinoma, hemochromatosis, opioids, a pituitary mass.

Why it matters more than most of the panel

Because it changes what happens next rather than just describing what happened.

Every major body requires it. The Endocrine Society recommends distinguishing primary from secondary hypogonadism by measuring LH and FSH, as a strong recommendation on moderate-quality evidence. The AUA goes further on LH specifically: “In patients with low testosterone, clinicians should measure serum luteinizing hormone levels” — Strong Recommendation, Evidence Level Grade A, the AUA’s highest evidence level. The EAU rates it Strong. And the VA’s Recommendations for Use — formulary guidance rather than a clinical practice guideline — requires it once low testosterone is confirmed.

A man given testosterone on the basis of two low levels without an LH has skipped the step that would have found the pituitary tumour.

Evidence: Established.

The pulsatility problem

LH is not secreted steadily. It comes in bursts, and the numbers behind that are worth knowing because they explain why one value is weak evidence.

In eight healthy young men studied by deconvolution, LH secretory bursts occurred roughly every 72 minutes, each secretory burst having a half-duration of about eight minutes — smeared, by the time it reaches the blood, into a concentration peak lasting about an hour — and 95% of the day’s total LH release occurred in about five hours. A separate study sampling twenty men every ten minutes found interpulse intervals ranging from 30 minutes to eight hours, with mean LH concentration and mean pulse amplitude each varying fourfold between individuals.

The consequence was quantified in 1973 and has never needed revising. Sampling normal men repeatedly, the investigators found that pooling over six hours reduced the 95% confidence limit on an LH estimate from ±50–90% down to ±12%. Read that in reverse: a single LH in a man carries a 95% confidence interval of roughly plus or minus half to nearly double the value printed.

And the clinical consequence, in their own words from that same paper — repeated sampling “allowed normal subjects to be distinguished from patients with low or moderately elevated LH levels in whom gonadotropin levels in single samples were often in the ‘normal range.’”

That sentence is fifty years old and it is still the best argument on this page.

Evidence: Established.

Biological variation

Separate from the minute-to-minute bursts, LH varies substantially week to week.

Three independent studies land in the same region — roughly 22 to 28%. In 38 men sampled weekly for ten weeks the within-person coefficient of variation was 22%. A Norwegian replication gave 24 to 28% depending on the statistical method. A 1993 study of 20 men over a year gave 24%, though that one is a median of individual coefficients rather than a pooled estimate and is not directly interchangeable with the other two. Compare any of them with total testosterone at 10 to 11%.

The between-person variation is larger still — around 33 to 37% — which gives LH an index of individuality below 1. In plain terms: the population reference interval is a weak instrument for judging any individual man’s LH. A value inside the range does not establish that his own LH is where it should be, and a value slightly outside it may be where he always sits.

The standardization problem, which is worse than for testosterone

Testosterone has a CDC standardization program with an accuracy criterion and a published list of certified laboratories. Estradiol has one too.

LH and FSH have neither. The CDC Hormone Standardization Program covers testosterone and estradiol only. There is no reference measurement procedure and no accuracy-based certification for either gonadotropin.

WHO International Standards do exist. That is not the same thing. The Director of the UK’s external quality assurance scheme for peptide hormones put it plainly in print: “Differences in calibration, antibody specificities and assay design mean that gonadotropin results obtained in different methods are still not comparable. Comparing patient results obtained in different methods therefore remains problematic.”

So an LH of 9.0 on one platform is not interchangeable with 9.0 on another. We have not been able to find a published figure quantifying how large that gap is in men, and we will not guess at one.

Evidence: Established (that the problem exists); the magnitude is unquantified.

What “inappropriately normal” actually means

The phrase appears in all four documents and none of them defines it numerically. It is a judgement: an LH that would be unremarkable in a man with a normal testosterone becomes abnormal in a man whose testosterone is 220, because it should have risen and did not.

The only quantitative version in the literature comes from the European Male Ageing Study, which classified 3,369 men aged 40 to 79 on a two-by-two grid:

	Testosterone ≥ 10.5 nmol/L	Testosterone < 10.5 nmol/L
LH ≤ 9.4 U/L	Eugonadal (about 77%)	Secondary — 11.8%
LH > 9.4 U/L	Compensated — 9.5%	Primary — 2.0%

Note how the categories fall out. Secondary hypogonadism was six times commoner than primary, and it was strongly associated with obesity — a body mass index of 30 or above carried a relative risk ratio of 8.74.

The important caveat: 9.4 U/L is a research cut-point from one European cohort on one assay. No guideline adopts it. The EAU’s chapter gives no numerical LH or FSH threshold at all. If you have been quoted a precise LH cutoff, it is a convention, not a standard.

Compensated hypogonadism, and why we are cautious about it

The top-left quadrant above — normal testosterone with a raised LH — has a name, and a growing literature, and we would grade the concept lower than most sites do.

The mechanism is real and rather elegant. Insulin-like peptide 3, a direct marker of Leydig cell capacity, declines about 15% per decade from age 40, while testosterone declines about 1.9% per decade — because rising LH is covering the gap. Compensated hypogonadism is what that compensation looks like on a report.

But whether it is a clinical entity is genuinely contested, and the contest is in the primary literature rather than in opinion. In 4,173 consecutive men attending for sexual dysfunction, 4.1% met compensated criteria, and after adjustment no specific sexual symptom was associated with it. The authors concluded: “The present data do not support the concept that compensated (subclinical) hypogonadism represents a new clinical entity.”

More telling: in a direct comparison, compensated hypogonadism was more common in a general-population sample (8.1%) than in a sexual dysfunction clinic (4.4%). That is the wrong direction for a disease.

The EAU’s own summary is one sentence: “the clinical significance of this condition is unclear.”

And nobody has published a longitudinal answer to the obvious question. The same cohort produced natural-history papers for primary and for secondary hypogonadism. We have not been able to find one for compensated. Whether it progresses to overt testicular failure is, as far as we can establish, unstudied.

Evidence: Limited.

The number that should change how you think about a low testosterone

From the same cohort’s longitudinal data: among men who met the biochemical definition of secondary hypogonadism, 42.9% recovered to eugonadal within 4.3 years without any treatment.

Nearly half. Untreated. That is the strongest single argument for the two-sample rule, for looking for a cause, and for not starting lifelong therapy on the strength of one bad year.

Evidence: Moderate.

What raises LH

Aging, and this is the best-documented pattern. In 3,200 men, age was associated with falling free testosterone together with a rising LH — the signature of a testis that is failing while the pituitary compensates.

Primary testicular failure of any cause: Klinefelter syndrome, prior orchitis, chemotherapy, testicular radiation, undescended testes, testicular surgery. Testicular sperm extraction itself raises FSH measurably; the accompanying LH rise did not reach significance.

Smoking, which raises SHBG and LH together — total testosterone rises but free testosterone does not, which the investigators read as the feedback loop resetting around a higher binding capacity.

Androgen receptor blockade, which is a useful proof of mechanism: flutamide roughly doubled bioactive LH entirely by increasing the mass and amplitude of each burst, with no change at all in burst frequency. Negative feedback in men acts on amplitude, not on timing.

Gonadotropin-secreting pituitary adenoma, rarely. In one surgical series of 22 clinically non-functioning adenomas, 5 showed hormone hypersecretion in life — and all five were men.

What lowers LH

Exogenous testosterone and anabolic steroids, profoundly. See below.

Obesity. In the 3,200-man cohort, a BMI of 30 or above lowered total testosterone by 5.09 nmol/L and free testosterone by 53.72 pmol/L with LH unchanged or reduced — which the authors interpreted as genuine hypothalamic-pituitary dysfunction rather than a binding artefact.

Opioids, dose-relatedly. In 54 men on sustained-release oral opioids, testosterone was subnormal in 74%, and every one of the 28 men taking 100 mg or more of methadone equivalent daily had a subnormal testosterone or estradiol.

Energy deficit, and the mechanism here is worth stating because it is usually described wrongly. In eight healthy men fasted for five days with sampling every five minutes, mean LH fell 30% — but burst frequency did not change at all (12.9 to 12.6 bursts per day). What halved was the mass released per burst. FSH was unaffected entirely, and the pituitary’s response to injected GnRH was enhanced, placing the lesion above the pituitary. The “fasting slows GnRH pulse frequency” picture comes from female athlete physiology and does not transfer to men.

Hyperprolactinemia, pituitary disease, congenital hypogonadotropic hypogonadism, hemochromatosis and severe illness. In hemochromatosis specifically, testicular volume correlated with LH at r = 0.71 in 30 affected men, and cirrhosis added a further central effect.

Glucocorticoids — with a caveat. This is routinely listed as a cause of central suppression, and the direct evidence in men is thinner than that implies. Dexamethasone given to GnRH-deficient men did not change LH, GnRH-stimulated LH amplitude or FSH — no direct pituitary effect. And in men with Cushing’s syndrome, testosterone fell further on dexamethasone regardless of cortisol suppression, which the investigators attributed to direct suppression of the Leydig cell. The effect is real; the location of it is less settled than the textbooks suggest.

Suppression and recovery on testosterone therapy

Exogenous testosterone shuts LH down almost completely. In men currently using androgens, mean LH was 0.5 IU/L, against 5.2 in non-users.

Recovery after stopping is slower than most men are told, and unevenly slower across the axis. In the same series, mean time to recovery was 10.7 months for LH, 19.6 months for FSH, 7.3 months for anti-Müllerian hormone, 14.1 months for sperm output, and 31.7 months for inhibin B. Full recovery took somewhere between 6 and 18 months for most measures, with testicular volume and SHBG the exceptions that did not fully return.

One caution about a citation you will see attached to this question. The large 2006 analysis of recovery after hormonal male contraception is frequently quoted as showing how long gonadotropins take to come back. It does not measure that. Its outcome is sperm concentration — median 3.4 months to reach 20 million/mL, 100% by 24 months — and LH appears in it only as a baseline predictor. It is also a study of young healthy men in supervised contraceptive trials, not of men who took supraphysiological doses for years.

Evidence: Moderate.

Assay artefacts

Macro-LH — LH bound into a large complex with immunoglobulin, producing a spuriously high result in a person whose gonadal function is clinically normal — is documented. Every published case we could find, however, is in a female patient. The mechanism is not sex-specific and it should be on the differential for an unexplained isolated LH elevation in a man, but we should be honest that no male case is on record.

Biotin interferes with the streptavidin-based immunoassays used for LH and FSH; high-dose biotin produced abnormal results across a panel of hormones that included both, correcting once the biotin was stripped out of the sample, and absent on assays that do not use the streptavidin-biotin system. If you take a high-dose biotin supplement, stop it before the draw and tell the laboratory.

The hook effect — where a very high concentration paradoxically produces a low reading — is a documented problem for hCG and prolactin. We found no documented case for LH or FSH, and will not assert one.

Related

The Lab Library · FSH · Total testosterone · Why is testosterone low?

Questions patients ask