Longevity & Aging · Metabolic & Cardiometabolic
testosterone therapy does not increase prostate cancer risk
In plain terms: Is testosterone therapy safe for the prostate?
Strong support Longevity & Aging 💰 Industry COI noted
Part of: 💊 Testosterone therapy (TRT)
RefutedContestedStrong support
consensus score 1.00
On the cancer question, the modern evidence is reassuring: the old fear that testosterone feeds prostate cancer is not supported. Randomized trials, large registries, and even prostate-cancer-survivor studies show no increased risk, and some show lower risk. The main caveat is that the trials weren't long enough to be the final word, so men on testosterone should still get routine PSA/prostate monitoring.
📅 Last reviewed: 2026-07-14 ⓘ
Evidence ladder
How far up the ladder this claim has climbed. A high consensus on a low rung means "consistent so far," not "proven in people."
Top evidence so far: All trials, pooled (Meta-analysis)
MechanismIn-vitroAnimalObservationalRCTMeta-analysis
How the studies fall
16 support 0 contradict 0 tested null 0 mixed · 16 sources, 16 independent groups
What the evidence shows
The old fear that testosterone therapy causes or fuels prostate cancer is not supported by the evidence — a genuine reversal of decades of dogma.
The evidence (22)
| Source | Grade | Stance | Quality | Finding |
|---|---|---|---|---|
| Nigro N, Christ-Crain M 2012 · Swiss Med Wkly | observational # was mechanism — narrative review of human data, nothing measured; role review carries the review-ness (reread #316) | mixed | low | Narrative review, no new data (publisher section: 'Review article'). Restates others' work on both sides: 'Although until now no clear evidence has been found that testosterone replacement therapy has a causative role in prostate cancer or indeed in changes of the biology of the prostate, in a recent meta-analysis a 4-fold increased risk of prostate-associated event rates in testosterone treated elderly men sounds a note of caution.' The 4-fold figure is secondhand and its endpoint is prostate-associated EVENTS, not prostate-cancer incidence. Overall verdict is an ignorance statement, not a finding: 'Importantly, so far the long-term safety and efficacy of testosterone replacement therapy has not been established.' |
| Chandra MS et al 2026 · Endocr Pract | observational | supports | low | Single-arm retrospective cohort of 33 men with pathology-confirmed prostate cancer treated with radiation +/- ADT and then given TRT (median age 75): median testosterone rose 66.0 to 299.3 ng/dL (P<.001) and 'One patient (3%) developed metastatic disease 3 years after starting TRT'; a companion narrative review of published case reports found a 'weighted mean biochemical recurrence rate was 3.3% during mean 42.6 months of follow-up'. Reassuring in direction but there is NO comparator arm, so the 3% cannot be contrasted with an untreated rate, and PSA itself rose within-patient: 'Prostate-specific antigen rose from 0.04 (IQR 0.02-0.17) to 0.17 ng/mL (IQR 0.04-0.44) (P = .018).' The authors conclude only that the findings 'provide the ethical and scientific rationale for a randomized controlled trial'. Senior author reports AbbVie research grants, Besins Healthcare consulting fees, and equity in Xyone. |
| Debruyne FMJ et al. 2017 · BJU Int | observational | supports | low | RHYME prospective registry, 999 hypogonadal men, 750 on TRT (23 900 person-months), max 36 months: 'the proportion of positive biopsies was nearly identical in men on TRT (37.5%) compared to those not on TRT (37.0%)' — but that proportion is over the 67 biopsied men only ('55 biopsies... for suspected prostate cancer, and 12 non-cancer related'), not the registry; non-randomised, and the incidence rates per 100 000 person-years the methods promise are not reported in the abstract. |
| Merlino L et al 2026 · World J Mens Health | observational | supports | low | TriNetX propensity-matched cohort of hypogonadal men with CKD stages 3-5 (1,545 per arm, median follow-up 60 months treated vs 45 months untreated), with prior prostate cancer, dysplasia and carcinoma in situ excluded at baseline: 'Secondary outcomes, including VaD, AD, stroke, myocardial infarction, prostate-related conditions, prostate cancer, and heart failure, did not differ significantly between the TRT and non-TRT cohorts.' A true incident-cancer null, but a weak one: no prostate-cancer hazard ratio or event count is published outside an image-only table, the endpoint is a composite of malignancy with dysplasia and carcinoma in situ, competing mortality is high (26-36% on dialysis), and the authors concede 'power was limited for non-mortality endpoints' - 'results for AD and prostate neoplasia were suppressed by TriNetX because of low event counts' in the advanced-CKD subgroup and prostate cancer 'had too few events for reliable estimation' among dialysis patients. Supports the claim in a narrow, high-risk population; quality low. |
| Kardoust Parizi M et al. 2019 · Urol Oncol | meta-analysis | supports | low | SR/meta-analysis, 21 eligible studies, PRISMA-guided search closed Nov 2018, in men with nonmetastatic prostate cancer who received TRT after curative-intent local therapy. The pooled endpoint is a SINGLE-ARM PROPORTION, not a comparative effect: 'The pooled BCR rate in CaP men treated with TRT after definitive local therapy with curative intent was calculated using a random effects model.' Result: 'The overall pooled BCR rate was 0.01 (95%CI 0.00-0.02) suggesting a lack of association between TRT and BCR; there was no heterogeneity among included studies (I2 = 24.34%, P = 0.15)', with subgroups 0.00 (0.00-0.02) after radical prostatectomy and 0.02 (0.00-0.04) after EBRT/brachytherapy/cryotherapy/HIFU. Conclusion: 'we did not observe higher rate of BCR after TRT for nonmetastatic CaP patients after definitive local therapy.' Directionally supports the claim, but at the quality floor: there is no untreated comparator and no relative risk, so a 1% pooled rate cannot be read against any background recurrence rate; the pooled studies are small uncontrolled series of men selected as low-risk enough to be offered TRT; the endpoint is biochemical recurrence in survivors rather than incident prostate-cancer risk; and the authors themselves treat the base as merely sufficient to justify a future trial - 'others and we have outlined a phase I/II trial assessing the safety and benefits of TRT'. Paywalled; abstract-grade, so the study list, participant totals, follow-up and COI could not be checked. |
| Morgentaler A 2011 · Urol Clin North Am | mechanism | supports | low | Sole-author narrative review (Morgentaler A, Urol Clin North Am 2011;38(2):119-24), no primary data: 'Modern studies indicate no increased risk of PCa among men with serum T in the therapeutic range', with 'Studies have failed to show increased risk of PCa in men with higher serum T, and supraphysiologic T fails to increase prostate volume or prostate-specific antigen in healthy men.' Explained by 'the Saturation Model, which posits a finite capacity of androgen to stimulate PCa growth.' Direction is right and on-topic, but the paper measures nothing — review role; the underlying studies are graded on their own notes. |
| Abdelgadir O et al. 2024 · Cancer Epidemiol | observational | supports | moderate | SEER-Medicare 2007-2015 retrospective cohort, 105,690 men >=65 y (82,578 White, 10,256 Black), four pre-diagnostic exposure groups (neither, statins alone, TTh alone, dual users). The discussion states 'Pre-diagnostic use of statins and TTh were, independently and jointly, associated with reduced risks of HRC and specific cancer sites at three years of follow-up' - no increase in prostate-cancer risk. But every reported estimate is for the JOINT statins+TTh contrast ('inverse joint associations of statins and TTh with incident HRCs before (aHR: 0.39; 95 % CI: 0.35-0.44) and after 3 years of follow-up (aHR: 0.74; 95 % CI: 0.67-0.82)'), with 'similar joint associations ... with incident PCa, aggressive PCa' holding 'only <3 years follow-up'; no TTh-alone prostate-cancer effect estimate is available. The sharp attenuation with longer follow-up is the signature of detection/healthy-user bias in a prescription-claims cohort, and the authors state 'further studies are needed to validate these findings'. |
| Morgentaler A, Traish AM 2026 · J Urol | meta-analysis | supports | low | Narrative review, no new data: 'A MEDLINE review of the literature was performed.' It asserts the claim's direction — 'Large randomized clinical trials reveal identical PCa rates in men receiving TTh vs placebo' and 'TTh in men with known PCa has not shown increased rates of recurrence or progression' — and proposes the androgen adequacy-vs-inadequacy/saturation-point framework ('Adequate T concentrations for optimal PCa growth occur at a low concentration called the saturation point'). Restates trials the vault already counts, by the saturation model's own originators, with two published Comments in the same issue; retained for transparency at zero weight per CONVENTIONS §5. |
| Morgentaler A 2009 · ScientificWorldJournal | observational | supports | low | Single-author narrative review by the Saturation Model's own originator; no new data. On the claim's therapy arm: 'This has special applicability to T-deficient men, since this means that T therapy may not be nearly as risky as once assumed', and 'growing acceptance of the use of T therapy in men with a prior history of PCa, with early data indicating minimal risk of cancer recurrence or progression'. Separately asserts low (not high) serum T tracks 'high Gleason score, advanced stage of presentation, and increased risk of biochemical recurrence after surgery'. [originator's own review - down-weighted] |
| Naelitz BD et al 2026 · Urology 2026 Jul;213:151-156 | observational | supports | low | Retrospective Merative MarketScan claims cohort (n=149,848 men >40, varicocele vs benign-scrotal-pathology controls). In adjusted Cox models 'varicocele alone was not associated with prostate cancer risk (HR: 1.04, P = .18), whereas untreated hypogonadism was associated with a modest but statistically signifiant increase in incident prostate cancer risk (HR: 1.27, P<.01)'; the excess therefore falls on UNTREATED hypogonadism, and the authors conclude 'Testosterone therapy and possibly varicocelectomy mitigate this effect.' Supporting only indirectly: testosterone therapy enters solely as a model covariate and NO hazard ratio for it is reported. Treated-vs-untreated status is a post-baseline exposure held fixed (immortal time, unacknowledged), exposures and outcome are billing codes with no serum testosterone or PSA, both cohorts are defined by urological contact (detection bias), and the senior author is 'a paid consultant to Tolmar Pharmaceuticals.' |
| Corona G et al 2026 · Andrology | meta-analysis | supports | low | Systematic review + random-effects meta-analysis of testosterone therapy after radical prostatectomy: 'Seven retrospective studies met the inclusion criteria, encompassing 398 patients (mean age: 63.9 years; mean follow-up: 29.6 months).' The 'pooled BCR rate in TTh-treated patients was 3.5% (95% CI: 1.5-8.5).' and 'In studies with untreated controls, TTh was associated with a significantly lower BCR risk.' Supports the no-increased-risk direction, but graded low quality on the authors' own limits: 'the absence of randomized controlled trials and incomplete reporting of key clinical variables preclude definitive conclusions.' Outcome is biochemical recurrence in prostate-cancer survivors, not incident prostate cancer, and confounding by indication (TTh offered to men with favourable pathology) is unaddressed. |
| Hussain MR et al. 2024 · Andrology | observational | supports | moderate | SEER-Medicare 2007-2015, 41,707 men aged 65+ ALREADY DIAGNOSED with a hormone-related cancer (31,097 prostate). Outcome is survival, not incidence: 'To examine the independent and joint effects of pre-diagnostic use of statins and testosterone replacement therapy on the risk of all-cause and cause-specific mortality among older men diagnosed with hormone-related cancers.' Primary models were null — 'No significant associations were found in Cox-proportional hazard models for hormone-related cancers' — with a prostate-cohort subgroup signal of lower all-cause death (HR 0.76, 95% CI 0.59-0.97) and lower prostate-cancer-specific death (HR 0.43, 95% CI 0.24-0.75) for testosterone therapy alone. The paper never assesses whether testosterone therapy raises prostate cancer risk, so it cannot vote on that claim. |
| Wallis CJD et al. 2016 · Lancet Diabetes Endocrinol | observational | supports | moderate | Ontario matched cohort, 10,311 TRT vs 28,029 controls, median 5.3 y: 'Risk of prostate cancer diagnosis was decreased for those with the highest tertile of exposure (HR 0.60, 95% CI 0.45-0.80) compared with controls, but not for those with the shortest exposure.' No excess at any exposure level. Authors caution: 'In view of the limitations of observational data and the potential for selection bias, these results warrant confirmation in a randomised trial.' Registry-diagnosis outcome, exposure-tertile gradient open to healthy-adherer bias; abstract-grade. |
| Isbarn H, Pinthus JH, Marks LS, Montorsi F, Morales A, Morgentaler A, Schulman C 2009 · Eur Urol | mechanism | supports | low | Narrative Medline review (PubMed type: Review; 'A Medline search was conducted to identify original articles, review articles, and editorials'), no new data. Conclusion: 'The available evidence indicates that TT neither increases the risk of PCa diagnosis nor affects the natural history of PCa in men who have undergone definitive treatment without residual disease' - explained by the saturation model. The authors disclaim their own base in the same breath: 'It must, however, be recognized that the literature remains limited regarding the effect of TT on PCa risk' and 'Although no controlled studies have yet been performed and there is a paucity of long-term data...'. Zero-weighted per CONVENTIONS section 5 (review is not evidence); Morgentaler co-authors and is the saturation model's originator. |
| Cui Y et al. 2014 · Prostate Cancer Prostatic Dis | meta-analysis | supports | low | Meta-analysis of 22 placebo-controlled RCTs (n=2351; 11 short-term <12 mo, 11 long-term 12-36 mo): no route reached significance for prostate cancer - short-term OR 0.39 (0.06-2.45) injection, 1.10 (0.26-4.65) transdermal; long-term 2.09 (0.18-24.73) injection, 3.06 (0.12-76.70) transdermal, 0.19 (0.01-4.03) oral. 'None of these reached or even approached statistical significance (all P>0.10).' Short-term TRT did raise PSA vs placebo (P<0.00001). Supports the claim's direction, but the CIs are so wide that a large increase is not excluded - absence of evidence, not evidence of absence; authors: 'long-term data are warranted with justifiable end points.' |
| Bhasin S et al. 2023 · JAMA Netw Open | RCT | supports | moderate | TRAVERSE prostate substudy (NCT03518034), n=5198 safety set, 14 304 person-years, mean treatment 21.8 mo: adjudicated high-grade prostate cancer 5 of 2596 (0.19%) TRT vs 3 of 2602 (0.12%) placebo, HR 1.62 (95% CI 0.39-6.77), P=.51; any prostate cancer 12 (0.46%) vs 11 (0.42%), HR 1.07 (0.47-2.42). Null on cancer, and biopsy rates were equal (16 vs 14) so it is not a biopsy-ascertainment artifact. NOT equivalence — the authors state '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.' Two prostate endpoints did move: PSA rose more on TRT (+0.15 ng/mL at 12 mo, 95% CI 0.08-0.21; no further widening after month 12) and invasive BPH surgery was nearly double (0.89% vs 0.46%, HR 1.91, 95% CI 0.95-3.84, P=.07). Men with PSA >3.0, prior prostate cancer or IPSS >19 were excluded; funded by an AbbVie-led manufacturer consortium that approved the final manuscript. |
| Kang DY et al. 2015 · Medicine (Baltimore) | meta-analysis | supports | low | PRISMA meta-analysis, 15 controlled trials (13 RCT + 2 prospective nonrandomized), 739 testosterone vs 385 control hypogonadal men, 8 weeks-12 months, search closed Feb 2014. On the CANCER outcome the paper is thin and says so: 'Only 3 studies provided data with respect to the development of prostate cancer, and rates were similar between those that received treatment and controls' — Shigehara 0/23 vs 0/23, Legros 1/237 across three oral doses vs 0/79, Marks 2/21 (9.5%) vs 4/19 (21.1%), i.e. 3 events on testosterone vs 4 on control, never pooled and no OR calculated. The authors' own verdict: 'Data of the included studies were not sufficient to evaluate the risk of prostate cancer with testosterone replacement therapy; however, based on evidence in the literature it does not appear that the risk of prostate cancer is affected by testosterone replacement therapy' — note that the reassuring half of that sentence cites OTHER literature, not this synthesis. What the paper DID pool is the PSA surrogate: +0.154 ng/mL vs control (95% CI 0.069-0.238, P<0.001; I2=0%), driven by the IM route (+0.271, 95% CI 0.117-0.425, P=0.001) with transdermal null (+0.085, 95% CI -0.021 to 0.190, P=0.116), and no difference in the rate of threshold PSA elevation (OR 1.02, 95% CI 0.48-2.20, P=0.953). PSA is 'prostate specific, but not prostate cancer specific'. Supports the negated claim in direction, but at the quality floor: 7 total cancer events, <=12 months (far shorter than prostate-cancer latency), Marks' rates in both arms are protocol-biopsy detection rather than incidence, and medical writing was funded by MSD China alongside a 'no funding or conflicts of interest' declaration. |
| Santucci J, Stapleton P, Perera M, Ischia J, Murphy D, Bolton D, Lawrentschuk N, Frydenberg M, Sathianathen N 2025 · BJU Int | observational # was meta-analysis — SR of 19 retrospective studies, pooling explicitly declared invalid by the authors (reread #317) | supports | low | Systematic review of 19 observational studies (5 active surveillance, 10 post-RP, 6 post-radiotherapy) in men with localised prostate cancer; 'Pooled meta-analysis was not valid due to insufficient number of comparative studies and significant clinical and methodological heterogeneity.' Active-surveillance progression 0-32% and 'did not differ significantly to non-exposed controls on retrospective comparison'; biochemical recurrence 0-7% post-RP and 0-6% post-radiotherapy (uncontrolled single-arm rates, <=60 months). Authors: 'Overall certainty of TRT oncological safety is limited by lack of long-term, prospective, controlled comparative data and lack of assessment of survival outcomes.' Speaks to progression/recurrence in existing localised disease, not to de-novo prostate-cancer incidence. |
| Elliott J et al. 2017 · BMJ Open | meta-analysis | supports | low | Pairwise meta-analysis of 13 RCTs, 12 weeks to 36 months: 'Prostate cancer 13 12 weeks to 36 months Placebo: 11/1649 TRT: 12/1877 0.97 (0.43 to 2.23) 0%'. Not from the network meta-analysis — 'Owing to the low event rates for most outcomes, network meta-analysis was only possible for serious adverse events and withdrawals due to adverse events.' 10 NRS also reported prostate cancer, none showing excess risk (RHYME registry incidence rate ratio 0.52, 95% CI 0.22 to 1.26). Quality low: 23 events total, credible interval reaching OR 2.23, 'Few RCTs or NRS were at low risk of bias', 'publication bias could not be ruled out for most outcomes', and 'The median treatment duration of the RCTs was 6 months (range: 3–36 months)'. Authors' own caveat: 'We found no increased risk of major harms; however, this must be viewed in light of the high risk of bias of the included studies, the rare nature of serious harms, and the short treatment duration and follow-up of most studies.' |
| Bhasin S et al. 2026 · JAMA Intern Med | RCT | supports | low | Phase-2 RCT, n=136 post-prostatectomy survivors of low-grade prostate cancer (Gleason 6-7, undetectable PSA >=2y), testosterone cypionate 100 mg IM weekly vs placebo for 12 weeks: 'No participant in either group experienced biochemical recurrence.' Zero events in BOTH arms, and the authors cap it themselves - 'The trial was neither long enough nor large enough to evaluate clinical recurrence or long-term safety' - and exclude high-grade cancer, ADT and radiotherapy patients. Directionally reassuring, near-zero discriminating weight. |
| Garcia-Becerra CA et al. 2026 · Int J Impot Res | meta-analysis | supports | moderate | Meta-analysis of 41 RCTs (n = 11,161): 'TTh was not associated with a statistically significant increase in ... PCaE (OR 0.88; 95% CI: 0.52-1.51; I² = 0.0%), or CsPcE (OR 1.13; 95% CI: 0.39-3.26; I² = 0.0%)'. Supports the negated claim, at moderate quality rather than high: the clinically-significant-cancer interval spans a threefold increase, I² = 0.0% on both endpoints reflects near-zero event counts rather than agreement ('sensitivity analyses were performed using continuity correction for zero-event trials'), and the trials are short- to mid-term — the same not-powered-and-not-long-enough caveat the claim already carries. Paywalled; abstract-only. |
| Baik SH et al 2026 · J Clin Endocrinol Metab | observational | supports | high | Medicare claims cohort 2007-2020, men 65+ with diagnosed hypogonadism, 1:1 propensity-matched (546,964 men in the prostate-cancer analysis): 'TRT use was linked to a significant 16% reduction in the hazard of PCa (HR = 0.84; 95% CI: 0.82-0.86), including reductions observed across long-term (15%), short-term (16%), parenteral (12%), and topical (13%) use.' Sensitivity analyses excluding men with elevated PSA and family history of PCa showed an 18% reduction. The same cohort found the opposite direction for benign disease: 'TRT use exhibited a 13% increase in the hazard of BPH (HR = 1.13; 95% CI: 1.11-1.14).' |
Disagree, or know a study we missed?
We grade by evidence, not opinions. The way to weigh in is to point us to a study we haven't cited (check the evidence table above first), or to flag a problem with one we have. Every submission is reviewed; if it holds up, the grade updates and shows in Science Changes Its Mind.
Opens a short form. You'll sign in with Google so submissions are tied to a real account — we don't display your identity, and we only accept a link we can verify (PubMed, DOI, ClinicalTrials.gov).
Educational only, not medical advice. Grades and scores reflect published evidence weighted by study design and quality; see the methodology.