Key takeaways
- PSA reflects prostate tissue activity and volume - it is specific to the prostate, not to prostate cancer.
- Low testosterone suppresses PSA, so the rise on treatment is usually tissue re-normalising rather than a new problem.
- Prostate androgen receptors saturate near the bottom of the normal range, which is why replacement does not behave like the old linear-fuel model predicted.
- Without a baseline drawn before the first dose, no later PSA result can be interpreted.
- Velocity, the free fraction and a confirmatory repeat carry more information than any single absolute value.
Most of the anxiety around PSA on testosterone therapy comes from reading a single number instead of a pattern. A man starts treatment, his PSA moves from 0.6 to 0.9, and he is told — or decides — that something has gone wrong. In the majority of cases nothing has. Understanding why requires two things: knowing what PSA is actually measuring, and knowing what happens to a prostate that has been running on low androgen for years and is suddenly given a normal supply.
What PSA measures, and what moves it
Prostate-specific antigen is an enzyme made by prostate epithelial cells and secreted into semen, where its job is to liquefy the ejaculate. A small fraction leaks into the bloodstream, and that is what the test picks up. PSA is therefore a marker of prostate tissue activity and volume — not a cancer test. It is specific to the prostate, not to prostate cancer, and the difference is the source of most of the confusion around it.
Things that raise it besides cancer: benign prostatic hyperplasia, which is simply more tissue making more enzyme; prostatitis and urinary infection, which disrupt the barrier between gland and blood; recent ejaculation; vigorous cycling; a digital rectal examination or any instrumentation; and, importantly, androgen status. Things that lower it: 5-alpha-reductase inhibitors, which suppress it substantially, and low testosterone.
Why treatment produces an early rise
Prostate epithelium is androgen-dependent tissue. Sustained low testosterone downregulates it: less secretory activity, less PSA production, a lower reading. That reading is not evidence of a healthy prostate — it is evidence of an under-stimulated one. Restoring testosterone to a normal range restores normal secretory function, and the PSA comes back to where it would have been all along.
The typical size of that change is modest, commonly around 0.2 to 0.5 ng/mL, arriving over roughly the first three to six months and then settling (Kang et al., Medicine 2015). After that adjustment most men are stable, and many show no further change at all. The rise is a re-baselining, not a trend — which is exactly why the follow-up measurement at six months matters more than the one at three.
The saturation model, and why the old fear was wrong
The historical position was that androgens fuel prostate cancer, therefore any added testosterone is dangerous. That belief came from observations in men with advanced disease whose testosterone was surgically or chemically removed — tumours regressed dramatically. The inference drawn was that the relationship is linear: more androgen, more growth.
It is not linear. Prostate androgen receptors saturate at relatively low testosterone concentrations, near the bottom of the normal range. Below that point the tissue is genuinely androgen-limited and responds steeply to change, which is why castration produces such marked effects. Above it, the receptors are already fully occupied and additional testosterone has nowhere further to act (Morgentaler et al., Eur Urol 2009). Moving a man from deficient to normal crosses the steep part of the curve. Moving him from normal to high-normal does not.
That single piece of receptor pharmacology reconciles findings that looked contradictory for decades: castration works, and replacement in deficient men does not accelerate disease. Both are true because the curve is flat over the range that replacement operates in.
What the trial evidence added
Mechanistic reasoning is not proof, and the field needed outcome data. The large randomised prostate-safety analysis conducted alongside the modern cardiovascular safety trial reported no increase in high-grade prostate cancer, prostate cancer overall, or acute urinary retention in treated men with confirmed hypogonadism (Bhasin et al., JAMA Netw Open 2023). Meta-analyses of earlier randomised work point the same way.
Two honest caveats belong with that. Men with known prostate cancer or markedly abnormal baseline PSA were excluded from those trials, so the result applies to the population that is actually being treated, not to everyone. And a trial of a few years cannot speak to a cancer that takes a decade to become clinically relevant. What the evidence supports is that monitored replacement in appropriately screened men does not appear to cause prostate cancer — not that monitoring can be dropped.
Reading the pattern rather than the number
Three properties of a PSA result carry information: the absolute value, the rate of change, and the free fraction. An isolated absolute number is the weakest of the three, which is why a baseline before starting is non-negotiable — without it, every subsequent result is uninterpretable.
| Pattern | What it suggests | Action |
|---|---|---|
| Modest 0.2-0.5 rise in the first six months | Tissue re-normalising on treatment | Expected; continue monitoring |
| Rise >0.75 ng/mL in a year | Velocity outside the expected pattern | Investigate |
| Absolute PSA >4 ng/mL | Conventional referral threshold | Investigate |
| PSA >3 ng/mL with an abnormal examination | Two findings agreeing | Investigate |
| Sudden jump from a stable baseline | Often infection, inflammation or recent activity | Repeat in six weeks before acting |
| Family history plus any rise | Higher pre-test probability | Lower threshold for evaluation |
The repeat-before-acting rule earns its place. Prostatitis, a recent ride, recent ejaculation and a recent examination can all move a result enough to trigger a referral that a second test makes unnecessary. Abstaining from ejaculation and hard cycling for a couple of days before the draw removes two of those variables for free.
A monitoring schedule that makes sense
The pattern is dense early, then sparse: a baseline before the first dose, a check at around three months, another at six once the re-baselining should have finished, and annually after that. Digital rectal examination follows general men's health guidance rather than anything specific to treatment. Testing at the same laboratory each time is more useful than it sounds, because assay-to-assay variation is large enough to imitate a trend.
PSA is one line on a panel that should also be tracking haematocrit, estradiol and lipids — why bloodwork comes before treatment covers what a proper baseline includes, and haematocrit management deals with the marker that actually changes management most often in the first year.
What happens if something is flagged
A flagged result does not go straight to biopsy. The usual sequence is a confirmatory repeat after around six weeks to exclude transient causes, then a free PSA percentage — a lower free fraction shifts probability toward cancer — and where available a refined risk score. Multiparametric MRI has substantially changed this step, because it can identify who genuinely needs a biopsy and where to target it, rather than sampling blindly. Urology referral follows if imaging or risk scoring supports it.
Treatment is commonly paused during evaluation, and can generally be resumed if cancer is excluded. A pause is a precaution taken to keep the picture clean, not a verdict.
The clinical pearl: the two failures on this topic are opposite and equally avoidable. One is refusing treatment to a deficient man on a prostate fear the evidence no longer supports. The other is treating him without a baseline PSA, so that when a number does move, nobody can tell whether it moved. Get the baseline, expect a small early rise, then read velocity rather than single values.
Bottom line
PSA reflects prostate tissue activity, and androgen status is one of the things that sets it. A man whose testosterone was low was running an artificially suppressed PSA; restoring normal levels usually produces a small early rise that then stabilises. The saturation model explains why physiologic replacement does not behave like the old linear-fuel model predicted, and randomised prostate-safety data in screened, monitored men supports that. None of which removes the need to monitor — it changes what monitoring is for. A baseline, a check at three and six months, then annual testing, with velocity and repeat testing rather than a single value driving any decision (Bhasin et al., JCEM 2018).
Educational content, not medical advice. Laboratory interpretation and any treatment decision are made by a licensed physician after individual evaluation. Individual results vary.
