SHBG: why total and free testosterone disagree
The same total testosterone can mean nearly double the free hormone in one man and half of it in another. What moves SHBG, and why lowering it is not a treatment.
This is the test that explains the disagreements on the rest of the panel.
What it measures
Sex hormone-binding globulin is a glycoprotein made by the liver that carries sex steroids in the blood. It binds dihydrotestosterone most tightly, testosterone next, and estradiol several-fold more weakly. Hormone bound to SHBG is not available to tissue while it is bound.
Its production is controlled by the transcription factor HNF-4α, and the thing that suppresses HNF-4α is hepatic de novo lipogenesis. In cell and animal work, glucose and fructose lowered SHBG production by driving lipogenesis; blocking lipogenesis prevented the fall. That mechanism is why a low SHBG is a reasonably sensitive readout of the metabolic state of the liver, and it is the best reason to think low SHBG mostly reports the company it keeps — obesity, insulin resistance, fatty liver — rather than driving it. That reading is not settled; see the risk-marker section below for why.
Why it matters
SHBG is why total and free testosterone disagree, and the size of that effect is larger than the numbers on a report suggest.
Working the Vermeulen equation at a total testosterone held constant at 500 ng/dL, with albumin at 4.3 g/dL:
| SHBG (nmol/L) | Calculated free T | % free |
|---|---|---|
| 20 | 13.2 ng/dL (456 pmol/L) | 2.6% |
| 30 | 10.9 ng/dL (376 pmol/L) | 2.2% |
| 40 | 9.2 ng/dL (317 pmol/L) | 1.8% |
| 60 | 6.9 ng/dL (238 pmol/L) | 1.4% |
| 80 | 5.5 ng/dL (189 pmol/L) | 1.1% |
Same total testosterone. The man with SHBG 20 has 91% more free testosterone than the man with SHBG 60.
Run it the other way and the consequence for thresholds becomes hard to ignore. Here is the total testosterone required to reach a free testosterone of 6.4 ng/dL — about 220 pmol/L, the one figure any guideline body has adopted:
| SHBG (nmol/L) | Total T needed |
|---|---|
| 20 | 260 ng/dL |
| 30 | 313 ng/dL |
| 40 | 365 ng/dL |
| 60 | 470 ng/dL |
| 80 | 574 ng/dL |
A 300 ng/dL cutoff is internally consistent at an SHBG of about 30 and nowhere else. At SHBG 80, a man needs a total testosterone of 574 ng/dL — comfortably normal by every threshold in use — to reach the same free concentration. At SHBG 20, a man at 300 ng/dL is already above it.
That arithmetic is ours, computed from the published equation with the inputs stated above rather than taken from a calculator, and it carries the equation’s known limitations: the calculation runs about 19% high against equilibrium dialysis, and the fixed binding constant is not right for every man (see the genetics section below).
Evidence: Established.
When it’s ordered
Not routinely. SHBG’s role in every current guideline is as an input, not as a finding.
The Endocrine Society recommends determining free testosterone — by equilibrium dialysis or by a calculation using total testosterone, SHBG and albumin — in men whose total testosterone is near the lower limit or who have a condition that alters SHBG. It does not recommend SHBG as a standalone test.
The EAU is the most explicitly in favour, with a Strong recommendation to “measure sex hormone-binding globulin and free-testosterone calculation when indicated.”
The AUA has no graded statement mentioning SHBG at all. Its diagnostic statements rest on total testosterone below 300 ng/dL plus symptoms. SHBG content attributed to “the AUA” generally comes from a separate white paper, not from the guideline, and should be cited that way.
The VA puts the reasoning most plainly: if SHBG is decreased, total testosterone “may underrepresent a patient’s androgen status”; if elevated, it “may overestimate” it.
What raises SHBG
Aging. In the Massachusetts Male Aging Study, SHBG rose 1.6% per year — and that figure is cross-sectional. The paper’s longitudinal within-subject slopes are for testosterone, not SHBG, and the total testosterone longitudinal slope happens also to be 1.6%, which makes the two easy to confuse. In trial baseline data, men aged 59 to 75 had a mean SHBG of 52 nmol/L against 33 in men aged 19 to 35.
Hyperthyroidism — and this is the cleanest demonstration of the whole page. Seven healthy lean men given low-dose T3 for 63 days on a metabolic ward: SHBG rose 150% (18.0 to 44.9 nmol/L), total testosterone rose 42%, and free testosterone did not change at all (142.8 to 137.3 pmol/L). The same pattern appears clinically in Graves’ disease, where SHBG and total testosterone are both elevated and free testosterone is not significantly different from the euthyroid state.
Smoking. In the European Male Ageing Study, smokers had SHBG 5.96 nmol/L higher and total testosterone 1.31 nmol/L higher — with no difference in free testosterone, which the authors read as the feedback loop resetting around the free level. Smoking is a genuine SHBG-raising exposure and it appears on none of the Endocrine Society, AUA or VA lists.
Enzyme-inducing anticonvulsants. Carbamazepine and phenytoin, prospectively, over five years: SHBG rose and the free androgen index fell, in proportion to duration of treatment. Untreated men with epilepsy did not differ from controls, so it is the drug rather than the disease. Valproate behaves differently and should not be lumped in with them.
Liver disease — but not all of it. SHBG was raised in alcoholic cirrhosis and normal in idiopathic hemochromatosis, in a study where both groups had testosterone reduced by more than half by different mechanisms. “Liver disease raises SHBG” is aetiology-specific.
Weight loss. After bariatric surgery, pooled across before-and-after studies, SHBG rose by 22 units and total testosterone by 8.1 nmol/L, with resolution of obesity-associated hypogonadism in 87% of men. That evidence base is uncontrolled before-and-after data, not trials.
Endurance training. Fifteen Masters runners aged 60 to 70 had SHBG of 124 nmol/L against 68 in minimally exercising men, with total testosterone higher and the free androgen index lower — and directly measured free testosterone identical between the groups. Habitual exercise, the authors concluded, does not influence free testosterone; it amplifies the age-associated rise in SHBG.
Also listed by the guidelines, on their authority rather than a primary study we have read: HIV, and estrogen use.
What lowers SHBG
Obesity and insulin resistance, by the lipogenesis mechanism described above. Worth stating carefully, because it is routinely oversimplified: in the European Male Ageing Study, obesity lowered free testosterone by 53.7 pmol/L at BMI 30 or above, with LH unchanged or reduced — a pattern the authors interpreted as genuine hypothalamic-pituitary dysfunction, not an SHBG measurement artefact. “Obese men only look hypogonadal because their SHBG is low” is half-wrong.
Type 2 diabetes, on the same mechanism.
Exogenous androgens — at supraphysiologic doses. In men who had used androgens for bodybuilding, SHBG was one of only two measures (with testicular volume) that did not fully recover after stopping, with recovery taking 6 to 18 months. But at replacement-level exposure the effect may be absent: 400 mg and 800 mg testosterone pellets did not lower SHBG in a controlled study. We have not found formulation-stratified figures for SHBG on testosterone therapy and will not print any.
Also listed by the guidelines, on their authority: glucocorticoids, nephrotic syndrome, hypothyroidism, acromegaly, and SHBG gene polymorphisms. We have not read a primary human study for the first three; the acromegaly evidence is a single study of 15 men, confounded by prolactin.
Genetics — and why it argues against a popular idea
Two variants at the SHBG locus dominate the genetics, in a study of 14,429 men. Per copy of the minor allele:
| Variant | SHBG | Total testosterone | Calculated free testosterone |
|---|---|---|---|
| rs12150660 (23% of alleles) | +3.9 nmol/L | +31.8 ng/dL | no effect (p = 0.39) |
| rs6258 (2% of alleles) | −7.4 nmol/L | −82.3 ng/dL | no effect (p = 0.065) |
Read that last column again. Genetic variants that move total testosterone by up to 82 ng/dL — enough to move a man across the AUA’s diagnostic threshold, and they do: carrying three or more risk alleles raised the prevalence of a total testosterone under 300 ng/dL from 4.6% to 30.1% — move calculated free testosterone not at all.
Two footnotes for anyone tracing these numbers. The variant usually quoted, rs1799941, was not the one measured; it sits in near-complete linkage with rs12150660 and, in the authors’ own words, “both SNPs represent the same signal.” And it is rs6258, not rs6259, that interferes with SHBG immunoassays: in rs6258 carriers the immunoassay overestimates functional binding capacity, the dissociation constant for testosterone is nearly doubled, and the equilibrium-dialysis free fraction runs 22% higher than in men without the variant. rs6259 is a different variant that adds a glycosylation site and slows clearance; its binding affinity is normal. The two are constantly conflated, and the conflation inverts the biology.
The common pitfalls
SHBG does not have a meaningful circadian rhythm. The roughly 12% swing that gets described that way is an ambulant-versus-supine plasma-volume effect: it tracks albumin, which moved 10.5% in the same experiment, almost exactly. It is posture, not secretion.
SHBG does not require a fasting draw. A 550-calorie meal dropped total testosterone by 30% at 60 to 120 minutes with no change in SHBG or LH. The fasting rule is a testosterone rule.
SHBG is not the noisy input to calculated free testosterone. Within-person coefficients of variation, measured in the same 20 men over a year: SHBG 12.1%, total testosterone 10.9%. The instability of a calculated free testosterone is inherited from the testosterone, not from the SHBG.
There is no harmonized SHBG reference range. Testosterone has one; SHBG does not. Population means in large cohorts run around 42 to 44 nmol/L with standard deviations of 19 to 22 — which is the quantitative reason to say the range is wide and to interpret an SHBG against the performing laboratory’s own interval. SHBG’s between-person variation (43%) is more than triple its within-person variation (12%), which makes population reference intervals a weak tool for judging any individual’s value.
SHBG as a risk marker
Low SHBG is one of the strongest single biochemical predictors of incident type 2 diabetes in men. In the Massachusetts Male Aging Study, the odds of future diabetes rose 1.89-fold per standard-deviation fall in SHBG (16 nmol/L) — independently of, and alongside, free testosterone at 1.58-fold. A later nested case-control study with a Mendelian randomization component estimated an odds ratio of 0.29 per standard-deviation rise in SHBG in men.
We would grade the association as robust and the causal claim as unsettled, for three reasons. That Mendelian randomization used two instruments both sitting in the same gene, so it cannot exclude that the locus does something else metabolic. The male replication rested on 340 people. And the mechanism runs the other way: if hepatic lipogenesis suppresses SHBG, then low SHBG is a downstream readout of the liver’s metabolic state, which is exactly what you would want in a risk marker and is not the same as a cause.
Evidence: Mixed — Established as a predictor, Limited as a cause.
Claims that are more complicated than they sound
“Boron lowers SHBG.” Everything traces to one study of eight men with no control group and no blinding, in which SHBG fell over a six-hour window on a single day. A free testosterone rise was reported a week later, from a single 8am draw. The same research group’s earlier four-week study found that boron supplementation significantly raised estradiol, which is not the direction the supplement is sold on. A better-controlled inpatient metabolic study, in women, found no effect on steroids at all. Directionally not impossible; the evidence base is one small uncontrolled study.
Evidence: Limited.
“Stinging nettle root frees up testosterone.” Every supporting study is a test-tube study. Lignans from nettle root bind SHBG in vitro, and an aqueous extract blocks SHBG binding to its prostatic membrane receptor — a different phenomenon entirely — at extract concentrations between 0.6 and 10 mg/mL, which are not plausibly reached in human serum after swallowing a capsule. The clinical trials that exist are trials of prostate symptoms, not of testosterone. The most authoritative review of the literature concluded the mechanism is still unclear and it is unclear whether the in vitro data are a surrogate for clinical effect at all.
Evidence: Unsupported for the androgen claim.
“Low SHBG is good, because it means more free testosterone.” The arithmetic in the first half is right — the table at the top of this page shows it. The inference is wrong twice over.
First, low SHBG is a marker of the hepatic metabolic state you would not choose, and it predicts incident diabetes independently of testosterone.
Second, the free testosterone gain does not persist, because free testosterone is what the feedback loop defends. This is visible three separate ways in the primary data. Thyroid hormone moved SHBG 150% and free testosterone not at all. The SHBG genotypes that move total testosterone by up to 82 ng/dL move calculated free testosterone not at all. And smokers’ higher SHBG raised their total testosterone but not their free testosterone — which the investigators read as the testosterone-LH feedback loop resetting around the free level.
The honest summary: SHBG changes total testosterone and leaves free testosterone roughly where the axis wants it. That is precisely why SHBG matters for interpreting the test — and precisely why manipulating SHBG is not a testosterone therapy.
How it relates to treatment
Before starting, SHBG is worth measuring when the total testosterone is borderline or when any of the conditions above is present, because it determines whether the total is telling you the truth.
There is a second reason that is less often mentioned: baseline SHBG predicts how fast a man clears testosterone. In men on graded testosterone doses, the metabolic clearance rate correlated inversely with SHBG and SHBG was an independent predictor of it. Two men on the same dose will not reach the same concentration, and this is part of why.
Related
The Lab Library · Free testosterone · Total testosterone · How testosterone should actually be tested
Questions patients ask
My SHBG is high, which is why my testosterone doesn't work.
High SHBG raises your total testosterone and leaves your free testosterone about where your body was holding it. It changes how the test reads, not usually how you feel.
What the evidence showsGiving healthy men thyroid hormone for nine weeks raised SHBG by 150% and total testosterone by 42% while free testosterone did not move. In Graves' disease the same pattern appears clinically. Genetic variants that shift total testosterone by up to 82 ng/dL do not shift calculated free testosterone at all.
What remains uncertainWhether there is a level of SHBG high enough to defeat that compensation. Nobody has defined one.
Bottom lineHigh SHBG is mostly a measurement problem, and the fix is to measure free testosterone.
Moderate
Low SHBG is good because more testosterone is free.
More complicated than it sounds. The arithmetic is right; the conclusion is not.
What the evidence showsLow SHBG is produced by hepatic de novo lipogenesis suppressing the transcription factor that drives SHBG synthesis — it is a readout of the metabolic state of the liver. It predicts incident type 2 diabetes with an odds ratio of 1.89 per standard-deviation fall, independently of free testosterone. And the free testosterone advantage does not persist, because the feedback loop resets around the free level.
What remains uncertainWhether low SHBG causes anything or only reports it. The Mendelian randomization evidence is single-locus and thin.
Bottom lineLow SHBG is not a benefit. It is usually a message about your liver and your insulin.
Mixed — Strong on the association, Limited on causation
A total testosterone of 300 is the line, whatever my SHBG.
It cannot be, and the arithmetic shows why.
What the evidence showsWorking the standard equation at a fixed free testosterone of 6.4 ng/dL, the total testosterone needed to reach it is 260 ng/dL at SHBG 20 and 574 ng/dL at SHBG 80. A single total-testosterone cutoff is internally consistent at about one SHBG value.
What remains uncertainWhat should replace it. No body has published an SHBG-adjusted threshold, and the free-testosterone reference ranges that would be needed disagree by more than twofold.
Bottom lineThe threshold is a convention. It is a reasonable one near the middle of the SHBG distribution and a poor one at the tails.
Moderate
Boron supplements lower SHBG and raise free testosterone.
This rests on one study of eight men, with no control group.
What the evidence showsThe SHBG fall was observed over six hours on a single day, in eight unblinded volunteers who acted as their own controls the previous day. The free testosterone rise came from a single 8am draw a week later. The same group's earlier four-week study found estradiol rose significantly, and a controlled inpatient study found no steroid effects at all.
What remains uncertainWhether there is any effect. Nobody has run a randomized controlled trial.
Bottom lineNot impossible, and vastly oversold relative to the evidence behind it.
Limited
SHBG has to be drawn fasting and first thing, like testosterone.
No. Those rules exist for testosterone.
What the evidence showsA 550-calorie meal dropped total testosterone by 30% while producing no change in SHBG or LH. The apparent daily rhythm in SHBG is about 12% and tracks albumin almost exactly — a plasma-volume effect from standing up rather than a secretory rhythm.
What remains uncertainNothing material.
Bottom lineDraw it with the testosterone because that is convenient, not because SHBG requires it.
Moderate
My SHBG is outside the reference range, so it needs treating.
SHBG is not a treatment target. It is a lens for reading the rest of the panel.
What the evidence showsThere is no harmonized SHBG reference range at all — unlike testosterone, which has one. Between-person variation in SHBG is more than triple within-person variation, which makes a population interval a weak instrument for judging any individual. And no guideline body sets an SHBG target or recommends treating the value.
What remains uncertainNothing that would change the answer.
Bottom lineAsk what it changes about the interpretation. If the answer is nothing, it changes nothing.
Strong
Ready for testosterone care built on a diagnosis?
The first step is a full endocrine evaluation, not a prescription. We see patients across San Diego County and welcome referrals from other physicians.
