PSA: a small rise, in the first year, that then stops
What each guideline actually requires, where the referral thresholds came from, and why one of them was quietly dropped from the guideline it is still attributed to.
This is the test that decides whether a man gets to start testosterone, and the one most likely to interrupt it.
What it measures
Prostate-specific antigen is made by prostate epithelium. It is specific to the prostate and not to cancer — a distinction established in the paper that introduced it to clinical practice, which found PSA raised in 86% of men with benign prostatic hyperplasia and concluded that “neither marker is specific.”
Its serum half-life is about 2.2 days, which is why anything that disturbs the prostate takes days to weeks to wash out.
What moves it that isn’t cancer
Ejaculation. In 64 men aged 49 to 79, PSA rose after ejaculation in 87% — by a mean of 41% at one hour, falling to 9% at six hours and 8% at 24. It had returned to baseline in 92% of men by 24 hours and 97% by 48. That paper’s closing recommendation is the origin of the rule you have been given: “It is recommended that men abstain from ejaculation for 48 hours prior to having a serum PSA determination.” A second study found 40% of men still above baseline at 24 hours, which is why 48 rather than 24 is the number.
A digital rectal exam does not. This is worth stating because the belief is near-universal. In the definitive study, there was “no clinically significant difference between serum PSA levels obtained immediately before and at 5 or 90 minutes after rectal examination.” Prostatic massage raised PSA above 4 ng/mL in 1 of 17 men; transrectal ultrasound in 3 of 27.
A biopsy does, enormously. In 100 men, needle biopsy raised PSA immediately in 92. With four or more cores the mean rise was about six-fold, lasting roughly two weeks — and in about one man in three the level had not come back down at the rate the half-life predicts.
Prostatitis does, and for longer than you would guess. In 31 men with acute prostatitis, total PSA rose to day three and was still falling at day 30 — while C-reactive protein had normalised by day ten. Free PSA was still low at a month. One month is not long enough for a PSA to normalise after prostatitis. And the reflex of giving antibiotics and repeating is weaker than it looks: in 98 young men retested after empiric antibiotics, PSA fell in 54% and rose in 46%.
Cycling essentially does not. Across eight studies in 912 men, six of which could be pooled, the pooled effect was +0.027 ng/mL, p = 0.74 — not different from zero. One outlier study that drew blood within five minutes of finishing a long ride found a 9.5% rise; that is not how PSA is drawn in practice, and no study has shown habitual cycling raises it.
Urinary retention and catheterisation are more equivocal than usually stated. Men in retention do have higher PSA — 64% above 4.0 versus 38% in a comparison group — but that is cross-sectional, and chronic prostatic inflammation was also commoner in that group and explains part of it. The one randomised before-and-after study of catheterisation itself found the change “not clinically important in any of the patients.”
Evidence: Established.
How much change is a real change
Two figures, from two designs, and they agree.
Sampling 24 men on ten consecutive weekdays, the biological variation was such that “an increase between 2 consecutive PSA levels that is less than 20 to 46% may be due to biological and analytical variation.” A survey of 27 studies by the European Group on Tumour Markers put it more simply: “during monitoring of serial measurements, the change needed for significance is approximately 50%.” A single PSA result carries a one-sided 95% confidence interval of roughly 33%.
At a baseline PSA of 1.0, that means a rise to 1.5 is noise.
Two lab problems that get ignored
Assays disagree. Comparing seven methods on 157 samples in the range that matters, relative differences exceeded 10% at a PSA of 4.0 for four of them. For free PSA the disagreement was worse — one platform ran about 17% high against another.
Calibration disagrees, by a quarter. PSA can be calibrated to the traditional Hybritech standard or to the WHO standard. WHO-calibrated results run about 25% lower. Applied in silico to 5,865 men, switching to WHO calibration at the same nominal threshold produced about 20% fewer biopsies and about 20% fewer cancers detected. The recommendation from that literature is that a WHO-calibrated assay needs a cutoff nearer 3.0 to behave like 4.0 on the traditional one.
No guideline states which calibration its threshold assumes. One laboratory in the published literature explicitly declined to switch calibration precisely so that patients being followed over years would not have a step change appear in their numbers. That is the right instinct, and it is the practical lesson: track PSA on one assay, in one laboratory.
If you take finasteride
Finasteride roughly halves PSA, and the correction factor is not constant.
The origin of the “multiply by two” rule is a randomised trial of 3,040 men: using an upper limit of 2.0 on finasteride against 4.0 on placebo preserved sensitivity and improved specificity, and the authors concluded that “multiplying PSA by 2 and using normal ranges for untreated men preserves the usefulness of PSA.”
But over longer follow-up in a much larger trial, “the adjustment factor required to preserve median PSA increased from 2 at 24 months to 2.5 at 7 years.” So the doubling rule is right early and increasingly wrong later. If you have been on a 5-alpha reductase inhibitor for years, two is not enough.
What testosterone therapy actually does to it
A small rise, in the first year, that then stops.
The meta-analyses are consistent and unimpressive: a difference in PSA change of 0.154 ng/mL in one pooling of 15 randomised trials; 0.10 ng/mL and not significant across 26 trials in another; not significant across 16 placebo-controlled trials in a third. A 51-study review found “no significant effect on mortality, prostate, or cardiovascular outcomes.”
The best distribution data — not just the mean — come from the Testosterone Trials. In 790 hypogonadal men over 65 treated for 12 months, PSA rose from a baseline of 1.14 by 0.47 ng/mL on testosterone against 0.06 on placebo. Crucially: 5% of treated men had a rise of 1.7 ng/mL or more, 2.5% had a rise of 3.4 or more, and a confirmed PSA above 4.0 at 12 months occurred in 1.9% on testosterone versus 0.3% on placebo.
TRAVERSE, the largest trial ever run, followed 5,198 men for a mean of 33 months. The between-group PSA difference was 0.11 ng/mL at three months and 0.15 (0.08 to 0.21) at twelve — and then it stopped. In the trial’s own words: “There was no significant between-group difference in PSA levels after month 12.”
Its prostate event table:
| Endpoint | Testosterone | Placebo | Hazard ratio |
|---|---|---|---|
| High-grade prostate cancer | 5 (0.19%) | 3 (0.12%) | 1.62 (0.39–6.77) |
| Any prostate cancer | 12 (0.46%) | 11 (0.42%) | 1.07 (0.47–2.42) |
| Acute urinary retention | 20 (0.77%) | 16 (0.61%) | 1.25 (0.65–2.41) |
| Prostate biopsy | 16 (0.62%) | 14 (0.54%) | 1.13 (0.55–2.31) |
| Invasive prostate surgery for BPH | 23 (0.89%) | 12 (0.46%) | 1.91 (0.95–3.84) |
What that trial does and does not establish. It enrolled men with a PSA of 3.0 or below, no prostate nodule, an IPSS of 19 or below, and no cancer history — and it mandated digital rectal examination at baseline, 12 months, 36 months and the end of the study. Its own authors wrote: “Because of the small number of prostate cancer events, these findings should not be interpreted to imply that the risk of prostate cancer in the testosterone and placebo groups was similar,” and “these findings should not be applied to patients with known prostate cancer, those with higher PSA values, or men who do not have confirmed hypogonadism.”
One observation from its data that is easy to miss: of the 23 prostate cancers, 14 occurred in men whose baseline PSA was between 2 and 3 — a band containing only 9.3% of the trial. Almost the entire signal sat in the top sliver of an already-restricted range.
Evidence: Established (that the rise is small and front-loaded); Limited (on long-term cancer risk).
The saturation model
The idea is that the prostate’s response to androgen saturates at concentrations well below the normal male range, so that adding testosterone to a man who already has some does very little.
What supports it is real. The strongest human evidence is mechanistic: in 44 men whose serum testosterone was more than doubled over six months, prostate tissue testosterone and DHT did not change significantly, and neither did histology, androgen receptor expression, proliferation markers or the PSA gene. A pooled analysis of 18 prospective studies found no association between prostate cancer and testosterone, free testosterone, DHT, or any other androgen measured. In the placebo arm of a trial where 3,255 men were biopsied per protocol regardless of PSA, cancer detection was 25.5% in low-testosterone men and 25.1% in normal-testosterone men.
What contests it is also real, and newer. Mendelian randomisation using genetic instruments in nearly 200,000 men found free testosterone associated with aggressive prostate cancer, odds ratio 1.23 per standard deviation — while total testosterone and SHBG were null. The TRAVERSE authors themselves cite that finding, and note that testosterone promotes the growth of metastatic prostate cancer.
Our reading: the saturation model is a coherent framework, well supported for total testosterone and for tissue androgen levels, consistent with the trial and cohort data, never tested by a study designed in advance to test it, and weakest exactly where the newest genetic data point — free testosterone and aggressive disease. Its own supporting analysis says the results “must be confirmed in future studies using an a priori defined hypothesis.”
Evidence: Promising.
Where the guidelines disagree, precisely
| Baseline PSA | Baseline DRE | PSA bar to starting | On-treatment schedule | Referral trigger | |
|---|---|---|---|---|---|
| Endocrine Society (2018) | Only for men who choose monitoring after shared decision-making: ages 55–69, or 40–69 if high risk | Yes, within that opt-in pathway | >4 ng/mL; >3 if high risk | PSA and DRE at 3–12 months; after a year, revert to ordinary screening guidance | Confirmed rise >1.4 ng/mL in the first 12 months, or confirmed PSA >4.0, or an abnormal DRE |
| AUA (2018, confirmed 2024) | Yes — every man over 40, unconditional | Not mentioned anywhere | None stated | No statement at all | None stated |
| EAU (2026) | Yes | “in all subjects” before therapy | Locally advanced or metastatic cancer is an absolute contraindication | — | “a significant rise or increase in PSA velocity or total PSA” — rated Strong, with no number attached |
| VA (January 2026) | Yes, then 12 months, then annually | Not mentioned | >4 ng/mL unevaluated; >3 with risk factors, which for the VA include Agent Orange exposure | Baseline, 12 months, annually | Rise >1.4 ng/mL within 12 months, or PSA >4.0 at any time |
| FDA label (2025) | “Evaluate patients for prostate cancer prior to initiating” | Not specified | Known or suspected prostate cancer is a contraindication | “periodically” | None |
Four real conflicts fall out of that. The age to start testing: an AUA-following clinician checks a PSA on a 42-year-old; an Endocrine Society-following one does not. Whether monitoring on therapy is required at all: the VA fixes a schedule, the Endocrine Society requires it only for the first year and only in men who opted in, and the AUA is silent. The rectal exam: mandatory for the EAU in every man, conditional for the Endocrine Society, unmentioned by the AUA and the VA. And the referral numbers, which is the next section.
Where the referral thresholds came from
“PSA above 4.0.” Stated by the Endocrine Society and the VA. Neither cites an empirical derivation; 4.0 is the historical biopsy threshold, and it is calibration-dependent — on a WHO-calibrated assay the equivalent is about 3.0. The Endocrine Society’s own grading of that recommendation is weak, on low-quality evidence, despite the sentence beginning with the word “recommend.”
“A rise above 1.4 ng/mL in twelve months.” Stated by the Endocrine Society in both 2010 and 2018 and by the VA — with a wording drift worth noticing. The 2010 version said “within any 12-month period”; the 2018 version narrowed it to “during the first 12 months”; the VA reverted to the looser form. We have not been able to find any citation or derivation for 1.4 in either Endocrine Society document. It is functionally a consensus number. The nearest empirical anchor was published afterwards and so cannot be its source: 5% of testosterone-treated men had a 12-month rise of 1.7 ng/mL or more. In TRAVERSE the 1.4 criterion fired in 0.96% of treated men and 0.46% of placebo men — so roughly half the men it flags would have been flagged without treatment.
“PSA velocity above 0.4 ng/mL per year.” This one is misattributed twice over.
It appears in the 2010 Endocrine Society guideline and only there. It was dropped from the 2018 revision — the phrase “PSA velocity” does not appear in that document — and it is absent from the AUA guideline, the VA document, the FDA label and the TRAVERSE protocol. The EAU still names PSA velocity, rated Strong, but attaches no number to it.
It is also not the urological literature’s number. The published velocity thresholds from the group that developed the concept are 0.75 ng/mL per year and 0.35 ng/mL per year. 0.4 matches neither.
And the concept itself has been tested and found wanting. In 5,519 men biopsied regardless of clinical indication, adding PSA velocity moved the area under the curve from 0.702 to 0.709, and biopsying on velocity alone would biopsy close to one man in seven. The authors’ conclusion: “We found no evidence to support the recommendation that men with high PSA velocity should be biopsied in the absence of other indications; this measure should not be included in practice guidelines.”
Evidence: Unsupported — for PSA velocity as a standalone trigger.
The harm you inherit by being monitored
Men on testosterone get more PSA tests than other men, and they inherit the downstream consequences of that.
TRAVERSE quantifies it cleanly. 57 of 2,596 treated men (2.2%) met urological referral criteria, against 28 of 2,602 on placebo (1.1%) — roughly double the referrals, for essentially identical cancer incidence. The meta-analysis usually cited for “more prostate events on testosterone” found a composite odds ratio of 1.78, while cancer, PSA above 4, and biopsies were each individually non-significant. That composite is a detection signal, not a disease signal. One newer pooling found the pattern more starkly still: fewer cancers and more than twice as many biopsies.
Biopsy is not free. In 2,049 men undergoing standard twelve-core biopsy: haematuria in 66%, haematospermia in 39%, rectal bleeding in 28%, infection in 6% — and urosepsis in 0.5%, retention in 0.3%.
And the underlying screening test buys less than most men assume. At sixteen years of follow-up in the largest screening trial, the absolute prostate cancer mortality difference was 0.18%, with 18 men diagnosed per death prevented. Modelling puts overdiagnosis at 23% to 42% of screen-detected cancers, and higher in some analyses.
None of this is an argument against measuring PSA before testosterone. It is an argument for what TRAVERSE actually did: confirm every elevation by repeating it, estimate the individual’s risk formally, and use a shared decision aid before biopsy. In that trial, only 16 of the 85 men who met referral criteria chose to be biopsied — and the authors noted that the small number of biopsies and the high proportion that were positive supported the approach.
Related
The Lab Library · Testosterone, the prostate and the heart · Testosterone replacement therapy · Should you start testosterone?
Questions patients ask
Testosterone raises PSA, so it must be feeding the prostate.
It raises it slightly, in the first year, and then the rise stops.
What the evidence showsIn TRAVERSE, the between-group difference was 0.11 ng/mL at three months and 0.15 at twelve, with no significant difference at any point afterwards. Meta-analyses give 0.10 to 0.15 ng/mL. Mechanistically, in 44 men whose serum testosterone more than doubled, prostate tissue testosterone and DHT did not change significantly.
What remains uncertainLong-term risk. TRAVERSE followed men a mean of 33 months, and its own authors wrote that the small number of events means you cannot conclude the risks were similar.
Bottom lineExpect a small early rise. Do not expect it to keep climbing.
Strong on the short-term measurement; Limited on long-term risk
My PSA went from 1.0 to 1.5, which is a 50% jump.
That is inside the noise.
What the evidence showsSampling men on ten consecutive weekdays, an increase of less than 20 to 46% between two values can be biological and analytical variation alone. The European tumour marker group put the change needed for significance in serial monitoring at about 50%. A single PSA carries a one-sided 95% confidence interval of about 33%.
What remains uncertainNothing about the arithmetic.
Bottom lineRepeat it before anyone acts on it — which is what every guideline says to do.
Strong
I should skip the rectal exam because it will raise my PSA.
It does not.
What the evidence showsIn the definitive study of 199 men, there was no clinically significant difference between PSA measured immediately before and at 5 or 90 minutes after a rectal examination. What does raise it is a biopsy — about six-fold for two weeks — and prostatic massage or ultrasound, modestly. Ejaculation raises it 41% at one hour, which is why the 48-hour rule exists.
What remains uncertainNothing on this point.
Bottom lineAbstain for 48 hours. The exam is not the problem.
Strong
My doctor should be tracking my PSA velocity.
That recommendation was withdrawn, and the concept did not survive testing.
What the evidence showsThe 0.4 ng/mL-per-year figure appears only in the 2010 Endocrine Society guideline. The phrase "PSA velocity" does not appear in the 2018 revision, in the AUA guideline, in the VA document, or in the FDA label. And in 5,519 men biopsied regardless of indication, adding velocity moved predictive accuracy from 0.702 to 0.709; the authors wrote that it "should not be included in practice guidelines."
What remains uncertainThe EAU still names PSA velocity as a Strong recommendation — without attaching a number to it.
Bottom lineThe absolute level, confirmed by repeating it, is what the current guidance runs on.
Strong
A PSA of 4.0 means the same thing everywhere.
No. It depends on which standard your laboratory calibrates to.
What the evidence showsWHO-calibrated PSA results run about 25% lower than traditionally calibrated ones. Applied to 5,865 men, the same nominal threshold produced about 20% fewer biopsies and about 20% fewer cancers detected. Assays also disagree with one another by more than 10% at a PSA of 4.0. No guideline states which calibration its threshold assumes.
What remains uncertainWhich calibration your own laboratory uses. It is worth asking.
Bottom lineStay with one laboratory. A change of assay mid-follow-up can look like a change in you.
Strong
If my PSA rises on testosterone I'll have to stop.
No guideline says to stop. They say to investigate.
What the evidence showsThe Endocrine Society directs you to confirm the elevation by repeating it first, "to exclude the possibility of transient rises," and then to urological consultation. The VA says referral is warranted. Neither instructs discontinuation. In TRAVERSE, 85 men met referral criteria and only 16 chose biopsy after formal risk estimation and a shared decision aid.
What remains uncertainWhether continuing testosterone during an evaluation is safe. Nobody has studied it directly.
Bottom lineA rise triggers a conversation and a repeat test, not an automatic stop.
Moderate
I take a 5-alpha reductase inhibitor, so I just double my PSA.
Roughly, for the first couple of years. After that, doubling is not enough.
What the evidence showsThe doubling rule comes from a randomised trial of 3,040 men and works well early. But over longer follow-up in a much larger trial, "the adjustment factor required to preserve median PSA increased from 2 at 24 months to 2.5 at 7 years."
What remains uncertainWhat the right factor is beyond seven years.
Bottom lineTell whoever reads your PSA that you take it, and how long for. And note that all of this comes from finasteride trials — we could not trace an equivalent correction factor to a primary dutasteride publication.
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.
