One enzyme, blocked
Almost everything claimed for this drug traces back to a single chemical step. Understanding that step explains both why the claims are plausible and why so many of them fail.
Interactive · the one mechanism that matters
Tadalafil does not create the signal. It stops the signal draining away.
Step · cGMP
cGMP is the actual currency of the system. How much of it accumulates decides how relaxed the vessel is. Everything tadalafil does traces back to raising this one pool.
Why this framing matters
Tadalafil is a permissive drug, not a stimulant. It blocks the enzyme that clears cGMP, so it can only amplify a signal the body is already producing. If the endothelium has stopped making nitric oxide, there is little for the drug to amplify. That single fact explains both why it works across so many organs and why its effects are modest where the underlying vessel biology is already destroyed.
Interactive · why the half-life is the story
Relative drug level in the blood over three days
Hover the chart
Curves are modelled from each drug's published half-life and time to peak, and every drug is scaled to 100 at its own single-dose peak. This shows the shape of the exposure, not absolute blood levels, which depend on dose and formulation. Drag the slider to see what once-daily dosing does: the short-acting drugs wash out almost completely between doses, while tadalafil accumulates into a near-flat background level. That difference in shape, not a difference in potency, is the entire argument for daily low-dose tadalafil as a vascular intervention rather than an on-demand one.
Why tadalafil specifically
Its half-life is 17.5 hours, against roughly four for sildenafil and vardenafil. On daily dosing it reaches a steady background level by day five, accumulating about 1.6-fold, and food does not affect it. That shape is the entire reason it gets proposed as a chronic vascular therapy rather than an occasional one.
The selectivity trade-off
Tadalafil is over 10,000-fold selective for its target against most related enzymes and 700-fold against the one in the retina, which is why it does not cause the blue-tinted vision sildenafil can. But it is only 40-fold selective over PDE11 in skeletal muscle, against 1,000-fold for sildenafil. That is the leading explanation for its distinctive back pain.
What this cannot do
Because the drug only slows the clearance of a signal the body makes itself, it is amplification, not generation. Where the vessel lining has stopped producing nitric oxide, there is little left to amplify. This is the structural reason the effects are largest in a diseased pulmonary circulation and smallest in the general population that most wants them.
PDE5 is not a penis enzyme
The enzyme sits in smooth muscle and vascular beds throughout the body, which is why one drug plausibly touches this many organs. What differs between them is not whether the mechanism exists, but whether anyone has shown it produces a benefit you would notice.
Brain
Randomized, surrogate endpointBlood flow moves. Thinking does not.
Randomized trials in people with small-vessel disease show measurable gains in perfusion through damaged white matter and in the ability of brain vessels to dilate on demand. Every trial that also tested cognition found nothing, and pooled adverse events more than doubled. The three-month daily-tadalafil trial in exactly this population failed its primary endpoint, which was simply whether patients could tolerate staying on the drug.
Every claim, filed by how good the proof is
Claims are placed at the honest ceiling of their evidence, not at the level of the loudest study supporting them. Filter by organ system, and open any row for the effect size and the reason it might be wrong.
How to read the tiers
- Regulator-approved. A drug regulator reviewed the randomized trials and approved tadalafil for this use. The strongest tier on this page.
- Randomized trials. Humans were randomly assigned to tadalafil or placebo. Randomization is what removes the confounding that sinks observational data.
- Randomized, surrogate endpoint. Randomized trials moved a marker (blood flow, stiffness, a lab value) rather than an outcome anyone feels. Suggestive, not proof of benefit.
- Observational only. Researchers compared people who happened to take the drug with people who did not. Cannot separate the drug from the kind of person who gets prescribed it.
- Animal or lab only. Shown in rodents, cells, or isolated tissue. No human outcome data. Most findings at this tier never replicate in people.
- Tested and failed. Not merely unproven. Someone ran the trial that should have shown a benefit, and it did not. This is the most informative tier on the page and the one that never makes it into a headline.
A drug regulator reviewed the randomized trials and approved tadalafil for this use. The strongest tier on this page.
Humans were randomly assigned to tadalafil or placebo. Randomization is what removes the confounding that sinks observational data.
Randomized trials moved a marker (blood flow, stiffness, a lab value) rather than an outcome anyone feels. Suggestive, not proof of benefit.
Researchers compared people who happened to take the drug with people who did not. Cannot separate the drug from the kind of person who gets prescribed it.
Shown in rodents, cells, or isolated tissue. No human outcome data. Most findings at this tier never replicate in people.
Not merely unproven. Someone ran the trial that should have shown a benefit, and it did not. This is the most informative tier on the page and the one that never makes it into a headline.
The dementia headline, taken apart
This is the claim that sent most people looking. It deserves the most careful treatment on the page, because the supporting evidence is genuinely large and the refuting evidence is genuinely decisive.
What the headlines are built on
- A claims-mining study of 7.23 million records reported a 69% lower Alzheimer's risk with sildenafil, HR 0.31 (95% CI 0.25-0.39). Nothing in dementia prevention comes close to that, which is a reason for suspicion rather than excitement. Its own authors wrote that it does not establish causality.
- A UK cohort of 269,725 men found HR 0.82 (0.72-0.93).
- A half-million-man network analysis reported dementia relative risk 0.68 for tadalafil - alongside near-identical reductions in death, heart attack, stroke and blood clots.
What happened when it was tested properly
- The exact regimen was tested. 5,204 people starting daily low-dose tadalafil, compared with 18,565 starting an alpha-blocker for the same condition: HR 0.99 (0.88-1.12), p = 0.93. A second cohort of 133,336 people over 7.9 years gave HR 0.95.
- A target-trial emulation controlling 76 confounders, comparing against a drug class with the same indication, returned HR 0.99, 1.00, 0.67 and 1.15 across four analyses - none significant.
- Mendelian randomisation, which uses randomly inherited genes and is immune to all of these biases, found no causal association across four datasets.
- Even the supportive UK cohort nulls out at HR 0.93 (0.80-1.08) once three years before diagnosis are excluded - the signature of early dementia stopping the prescription, not the drug stopping dementia.
The one randomized test that exists
Interactive · read this before any headline
Six reasons a drug can look protective without being protective
| Threat to validity | Database cohortthe dementia headlines | Target trial emulationcareful re-analysis | Randomized trialthe only clean answer |
|---|---|---|---|
| Healthy-user bias | ○Fully exposed | ◐Reduced, not removed | ●Removed by design |
| Confounding by indication | ○Fully exposed | ◐Reduced, not removed | ●Removed by design |
| Prevalent-user bias | ○Fully exposed | ●Removed by design | ●Removed by design |
| Immortal time bias | ◐Reduced, not removed | ●Removed by design | ●Removed by design |
| Reverse causation | ○Fully exposed | ◐Reduced, not removed | ●Removed by design |
| Surrogate endpoint failure | ○Fully exposed | ○Fully exposed | ◐Reduced, not removed |
Healthy-user bias
People who fill and keep refilling any prescription are, on average, healthier and more engaged with their health than people who do not. The pill picks out the person as much as the person picks the pill.
A man who asks his doctor for a PDE5 inhibitor, pays for it, and refills it is more likely to exercise, attend check-ups, take his statin, and be in a relationship. Every one of those independently lowers dementia and heart-attack risk.
On the vascular dementia premise itself
Accurate
Vascular dementia is the second leading cause of dementia. In autopsy series, pure vascular disease accounts for about 15% of cases and mixed vascular-plus-degenerative disease for another 16%.
Also accurate
It is substantially preventable. The 2024 Lancet Commission puts 45% of all dementia down to 14 modifiable risk factors.
The leap
Neither fact implies this drug. The Commission's vascular levers are blood pressure, LDL cholesterol, diabetes, smoking, physical activity, body weight and alcohol - each with decades of randomized evidence. No PDE5 inhibitor appears anywhere in it.
The trial that would have answered this enrolled 23 people
Every argument on this page eventually arrives at the same place: the study that would settle it was started, abandoned, and never replaced.
PITCH-HF · NCT01910389
A Phase 3, quadruple-masked, placebo-controlled trial across 63 North American sites, testing tadalafil 20 mg titrated to 40 mg daily for up to three years. Its primary endpoint was the one that matters: cardiovascular death or hospitalisation for heart failure. Its secondary endpoints included all-cause mortality.
It began in November 2013 and ended in February 2014, terminated by its funding agency. Actual enrolment: 23 participants. It produced no usable efficacy data, and in the twelve years since, nobody has run a replacement.
So when a claim appears that this drug prevents heart attacks or extends life, the question to ask is not whether the mechanism is plausible. It is: compared with what, in whom, and who was randomized? For hard cardiovascular outcomes in ordinary people, the answer remains nobody.
What randomized hard-endpoint data exist
- Positive in pulmonary arterial hypertension, on a disease-specific failure endpoint.
- Null in heart failure with preserved ejection fraction.
- Harmful in pulmonary hypertension after valve surgery.
- Never run for heart attack, stroke or cardiovascular death in people without pulmonary hypertension.
The honest expert position
The part that is not negotiable
This drug is broadly safe in the people it was studied in, with one interaction that kills and several that need a prescriber's judgement. The bigger real-world risk for anyone taking it off-label is not the molecule but the supply chain.
Absolute contraindication · nitrates
Interactive · the part that is not optional
Red-flag screener
Tick anything that applies to you. Nothing you tick leaves your browser.
Nothing ticked from this list.
This list covers the label's absolute contraindications and the best-known cautions. It is not a full medical history, it cannot see your blood pressure, your kidneys or the rest of your medication, and taking a prescription drug for an unapproved purpose is a decision for you and a prescriber, not a checklist.
What actually happens, in percent of patients
Side effects against placebo
- Headache20 mg as needed15%5%
- Indigestion20 mg as needed10%1%
- Back pain20 mg as needed6%3%
- Nasal congestion20 mg as needed3%1%
- Flushing20 mg as needed3%1%
- Muscle ache40 mg daily for lung disease14%4%
- Headache40 mg daily for lung disease42%15%
Rates come from the trial tables in the drug's own FDA label. Back pain and muscle ache are tadalafil's signature complaints and are thought to come from its unusually narrow 40-fold selectivity over a related enzyme found in skeletal muscle, against 1,000-fold for sildenafil. In the three-month daily trial in people with small-vessel brain disease, 76% on tadalafil reported an adverse event against 36% on placebo, and enough of them stopped the drug that the trial failed its primary endpoint.
Two things nobody can tell you
What decades of daily use does. The longest datasets run to about two years, and those are of as-needed dosing. For once-daily dosing the label counts just 115 patients treated for two years or more. There is no dataset at all for healthy people taking it preventively for decades, because that has never been studied.
What is in an unregulated tablet. Of 45 samples sold online as tadalafil and analysed forensically, 9 were genuine. Twenty-three contained sildenafil instead - a different drug with a different nitrate washout window - and nine contained neither. Every safety rule above assumes you know which molecule and which dose you took.
What would have to happen for this to be true
Being unproven is not the same as being wrong. Here is what the evidence would need to look like before the popular claims became defensible, and what to watch for.
Established
- Improves exercise capacity and reduces mortality in pulmonary arterial hypertension, at 40 mg daily.
- Modestly relieves the urinary symptoms of an enlarged prostate, at 5 mg daily, without the sexual side effects of the standard alternative.
- Lowers blood pressure by roughly 3/2 mmHg and is safe alongside most blood-pressure drugs, with the nitrate exception.
Plausible, unproven
- Improves the function of the vessel lining, possibly persisting after the drug is stopped - on a measurement never validated as a substitute for events.
- Increases perfusion through damaged small vessels in the brain, with no accompanying cognitive benefit in any trial that looked.
- Improves heart mechanics in men with diabetes, on an imaging measure, with an unexplained absence of effect in women.
Tested and failed
- Dementia prevention, in three independent designs.
- Heart failure with preserved ejection fraction, definitively.
- Insulin sensitivity in type 2 diabetes.
- Muscle function, in a 331-participant trial terminated for lack of efficacy.
- Exercise capacity at altitude.
What would change this page
For the mortality claim: an active-comparator, new-user study against another quality-of-life drug prescribed under similar gatekeeping, with negative control outcomes such as accidental death that no vasodilator should affect. If the signal survived that and the negative controls stayed null, it would deserve a trial. No such study has been published.
For the brain claim: a randomized trial with dementia incidence as an endpoint, in cognitively healthy people, which would need thousands of participants followed for a decade. The nearest thing starting now enrols people who already have impairment, has safety as its primary endpoint, and reports around 2028.
If you are considering this anyway
Check any of it
Every number on this page comes from one of these. Filter by study design to see how much of the case rests on randomized evidence and how much on databases.
- 01
CIALIS (tadalafil) tablets, for oral use - full prescribing information
Eli Lilly and Company (FDA-approved prescribing information) · DailyMed / FDA Structured Product Label, NDA 021368 · 2026
FDA label
The source of record for approved indications, every dose regimen, contraindications and trial-reported adverse-event rates. Approved for erectile dysfunction, the signs and symptoms of benign prostatic hyperplasia, and the two together. There is no boxed warning; the section 4 contraindications carry the same absolute force.
Open source ↗ - 02
ADCIRCA (tadalafil) tablets, for oral use - full prescribing information
United Therapeutics Corporation (FDA-approved prescribing information) · DailyMed / FDA Structured Product Label, NDA 022332 · 2025
FDA label
The same molecule approved at 40 mg once daily for pulmonary arterial hypertension. Source for the phosphodiesterase selectivity figures, the pharmacokinetics, and the instruction that nitrates must not be used within 48 hours of the last dose.
Open source ↗ - 03
Tadalafil pharmacokinetics in healthy subjects
Forgue ST, Patterson BE, Bedding AW, Payne CD, Phillips DL, Wrishko RE, Mitchell MI · British Journal of Clinical Pharmacology · 2006
Pharmacology · n = 237 across 13 clinical pharmacology studies
The definitive pharmacokinetic source: elimination half-life 17.5 hours (5th-95th percentile 11.5-29.6), peak concentration 378 micrograms/L at 2 hours for a 20 mg dose, steady state reached by day 5 of daily dosing with 1.6-fold accumulation, exposure proportional to dose, and no effect of food or time of day.
PubMed 16487221 ↗ - 04
High biochemical selectivity of tadalafil, sildenafil and vardenafil for human phosphodiesterase 5A1 over PDE11A4
Weeks JL, Zoraghi R, Beasley A, Sekhar KR, Francis SH, Corbin JD · International Journal of Impotence Research · 2005
Pharmacology
Measured on purified human enzyme, tadalafil is only 40-fold selective for PDE5 over PDE11A4, against 1,000-fold for sildenafil and 9,300-fold for vardenafil. PDE11 sits in skeletal muscle, prostate and testis, and this narrow margin is the leading explanation for tadalafil's characteristic back pain and muscle ache.
PubMed 15538396 ↗ - 05
Antiproliferative effects of phosphodiesterase type 5 inhibition in human pulmonary artery cells
Wharton J, Strange JW, Møller GM, Growcott EJ, Ren X, Franklyn AP, et al. · American Journal of Respiratory and Critical Care Medicine · 2005
Pharmacology
In human pulmonary artery smooth muscle cells about 80% of all cGMP breakdown is done by PDE5. Blocking it raised cGMP roughly tenfold, against roughly twofold for agents that generate the signal instead. Direct human tissue evidence that PDE5 is the dominant off-switch in a vascular bed.
PubMed 15817798 ↗ - 06
Efficacy and safety of tadalafil 5 mg once daily for lower urinary tract symptoms suggestive of benign prostatic hyperplasia
Porst H, Kim ED, Casabé AR, et al. · European Urology · 2011
Randomized controlled trial · n = 325, 12 weeks
One of the trials behind the prostate approval: urinary symptom score fell 5.6 points on tadalafil vs 3.6 on placebo (p = 0.004), a placebo-adjusted difference of about 2 points. Peak urine flow and post-void residual volume did not improve significantly - the symptom score moves while the plumbing measurements often do not.
PubMed 21871706 ↗ - 07
Efficacy and safety of tadalafil 5 mg once daily in the treatment of benign prostatic hyperplasia: a meta-analysis
Cui Y, Cao J, Bai Y, et al. · Frontiers in Medicine · 2021
Meta-analysis of RCTs · 15 randomized trials
Pooled placebo-adjusted urinary symptom score improvement of 1.97 points (95% CI -2.24 to -1.70). That sits at or below the roughly 3-point difference usually considered the smallest change a patient can notice.
PubMed 34712682 ↗ - 08
Proportion of tadalafil-treated patients with clinically meaningful improvement in lower urinary tract symptoms associated with benign prostatic hyperplasia
Nickel JC, Brock GB, Herschorn S, Dickson R, Henneges C, Viktrup L · BJU International · 2015
Randomized controlled trial · n = 1,499 pooled
71.1% of men on tadalafil improved by at least 3 symptom points - but so did 56.0% of men on placebo. The placebo response in prostate symptoms is enormous, and it is the single most important context number for the whole indication.
PubMed 25195970 ↗ - 09
Comparative effectiveness and safety of monodrug therapies for lower urinary tract symptoms associated with benign prostatic hyperplasia: a network meta-analysis
Yuan JQ, Mao C, Wong SY, et al. · Medicine (Baltimore) · 2015
Meta-analysis of RCTs · 124 trials, 58,548 participants
Head-to-head across the whole drug class, tadalafil's symptom improvement is statistically indistinguishable from tamsulosin, silodosin and dutasteride, and numerically behind doxazosin (-3.67 points) and terazosin (-3.37). Tadalafil is a legitimate option, not a superior one; its real advantage is that it does not cause the ejaculatory dysfunction alpha-blockers do.
PubMed 26166130 ↗ - 10
Efficacy and safety of tadalafil 5 mg once daily in men aged 75 years or older with lower urinary tract symptoms
Oelke M, Wagg A, Takita Y, Büttner H, Viktrup L · BJU International · 2017
Randomized controlled trial · Integrated analysis of 12 phase 2-3 studies
In men aged 75 and over, tadalafil was not significantly better than placebo for urinary symptoms, with a significant treatment-by-age interaction (p = 0.034). The benefit is confined to men under 75.
PubMed 27988986 ↗ - 11
Impact of cardiovascular risk factors and related comorbid conditions and medical therapy on the efficacy of tadalafil in men with lower urinary tract symptoms
Vlachopoulos C, Oelke M, Maggi M, Mulhall JP, Rosenberg MT, Brock GB, et al. · International Journal of Clinical Practice · 2015
Randomized controlled trial · n = 1,498 pooled
In men taking more than one blood-pressure drug the benefit vanished: -1.2 points (95% CI -2.5 to 0.0), against -3.3 in men on a single agent. In diuretic users the effect was -0.2 points, meaning none at all.
PubMed 26299520 ↗ - 12
A phase 3 randomized placebo-controlled trial of tadalafil for Duchenne muscular dystrophy
Victor RG, Sweeney HL, Finkel R, McDonald CM, Byrne B, Eagle M, et al. · Neurology · 2017
Randomized controlled trial · n = 331 boys, 48 weeks, quadruple-masked
The trial the muscle-blood-flow mechanism predicted would work, and the cleanest failure in the literature. Over 48 weeks both tadalafil arms declined more than placebo on the six-minute walk (64.7 m and 59.1 m vs 51.0 m), with no effect on any secondary outcome. Graded Class I evidence of no benefit, and terminated on the registry for lack of efficacy.
PubMed 28972192 ↗ - 13
Dexamethasone but not tadalafil improves exercise capacity in adults prone to high-altitude pulmonary edema
Fischler M, Maggiorini M, Dorschner L, Debrunner J, Bernheim A, et al. · American Journal of Respiratory and Critical Care Medicine · 2009
Randomized controlled trial · n = 23 at 4,559 m
Maximum oxygen uptake was 63% of predicted on tadalafil against 61% on placebo - no meaningful difference, while dexamethasone reached 74%. The authors conclude that corticosteroids, but not PDE5 inhibition, protect exercise capacity under hypoxic pulmonary vasoconstriction.
PubMed 19520908 ↗ - 14
Two decades of the NIA Interventions Testing Program: a comprehensive review of tested agents
Jiang N, Bhatt B, Nikolich-Žugich J, et al. · The Journals of Gerontology Series A · 2025
Systematic review · 54 agents, more than 30,000 mice
The definitive catalogue of what has actually been shown to extend mouse lifespan under rigorous multi-site conditions, covering 54 agents across more than 30,000 mice. No PDE5 inhibitor appears among the reported results.
PubMed 40717358 ↗ - 15
Meta-analysis of the association between phosphodiesterase inhibitors and risk of melanoma
Loeb S, Ventimiglia E, Salonia A, Folkvaljon Y, Stattin P · Journal of the National Cancer Institute · 2017
Meta-analysis of observational studies · 866,049 men, 41,874 melanomas
A small association exists across the class, summary RR 1.11 (95% CI 1.02-1.22), and it fails formal causality criteria: significant only at low exposure rather than high, only at early stage rather than advanced, and equally present for basal cell carcinoma. That is the fingerprint of sun exposure and of men who see doctors more, not of a carcinogen. Where drug-specific data exist the signal is confined to sildenafil; tadalafil was not significantly associated.
PubMed 29117385 ↗ - 16
Phosphodiesterase type 5 inhibitors use and risk of melanoma: a meta-analysis
Han J, Jiang R, Zhang X, et al. · OncoTargets and Therapy · 2018
Meta-analysis of observational studies · 4 cohorts and 3 case-control studies, 1,534,615 men
The drug-specific analysis that matters for this page: the melanoma association is confined to sildenafil, RR 1.26 (95% CI 1.07-1.50, p = 0.007). Tadalafil and vardenafil were not significantly associated. The signal was significant for in-situ melanoma but not for localised or advanced disease, which is the detection-bias signature again.
PubMed 29440918 ↗ - 17
The protective role of phosphodiesterase inhibitors in preventing colorectal cancer and advanced colorectal polyps: systematic review and meta-analysis
Cullinane C, Brett A, Devane L, et al. · Colorectal Disease · 2023
Meta-analysis of observational studies · 995,242 patients
Colorectal cancer or advanced polyps OR 0.88 (95% CI 0.79-0.98, p = 0.02). Every included study is observational, so this carries the same healthy-user problem as the mortality literature. No randomized trial has tested it.
PubMed 37635321 ↗ - 18
Sildenafil and Kidney Function in Heart Failure with Preserved Ejection Fraction
Edmonston D, Sparks M, Rajagopal S, Wolf M · Kidney360 · 2023
Randomized controlled trial
The follow-up to the kidney signal in the big heart-failure trial. Over the full trial the difference in filtration-rate slope was not significant (+0.47 mL/min/1.73m2 per year, 95% CI -6.63 to 7.57): an acute early decline that later recovered, rather than sustained kidney harm.
PubMed 36978225 ↗ - 19
Phosphodiesterase 5 inhibition rescues mitochondrial dysfunction in Leigh syndrome models
Zink A, Conrad J, Telugu NS, et al. · Cell · 2026
Animal study · Brain organoids, small and large animal models, 6 patients treated off-label
The only study in which a PDE5 inhibitor extends lifespan in any animal. It does so in models of Leigh syndrome, a lethal inherited mitochondrial disease, by rescuing a specific defect. That is disease rescue, not slowing of normal ageing, and it must not be cited as longevity evidence.
PubMed 41819105 ↗ - 20
Blueprint protocol - published personal supplement and medication list
Bryan Johnson · protocol.bryanjohnson.com (self-published, retrieved 2026-08-03) · 2026
Self-published protocol
Lists tadalafil 5 mg daily, with the stated rationale that it 'enhances endothelial function, reduces inflammation, and potentially protects against age-related cardiovascular and metabolic decline'. Listed here as the origin of a claim, not as evidence for it: no randomized trial supports the inflammation or age-related-decline parts.
Open source ↗ - 21
Phosphodiesterase Type 5 Inhibition With Tadalafil Changes Outcomes in Heart Failure (PITCH-HF)
Semigran MJ, Assmann SF, Siami FS (Carelon Research / NHLBI / MGH) · ClinicalTrials.gov registry record · 2014
Ongoing trial · 23 enrolled of several thousand planned
The only Phase 3 trial ever designed to test whether tadalafil prevents cardiovascular death or heart-failure hospitalisation. Terminated by its funding agency in February 2014 after enrolling 23 people. No efficacy data. Never replaced.
ClinicalTrials.gov NCT01910389 ↗ - 22
Treatment With Phosphodiesterase 5 Inhibitors and Long-Term Outcomes in Patients Undergoing Coronary Angiography and Cardiac Catheterization
Roy S, Min JJ, McFalls E, Perera D, Salloum FN, Jovin IS · Catheterization and Cardiovascular Interventions · 2025
Observational cohort · n = 4,582 patients (562 on a PDE5 inhibitor)
The crude one-year MACE difference was large and significant (30.4% vs 36.9%, p = 0.003). After propensity matching it disappeared completely: 30.4% vs 31.1%, p = 0.84, adjusted OR 0.99 (95% CI 0.93-1.06). This is the clearest available demonstration of what the mortality signal is actually made of.
PubMed 40421709 ↗ - 23
Association between treatment for erectile dysfunction and death or cardiovascular outcomes after myocardial infarction
Andersson DP, Trolle Lagerros Y, Grotta A, Bellocco R, Lehtihet M, Holzmann MJ · Heart · 2017
Observational cohort · n = 43,145 Swedish men after a first heart attack
All-cause mortality HR 0.67 (95% CI 0.55-0.81) and heart-failure hospitalisation HR 0.60 (0.44-0.82) in men dispensed a PDE5 inhibitor. Alprostadil, which has no systemic cGMP mechanism, showed no mortality association.
PubMed 28280146 ↗ - 24
Association of Phosphodiesterase-5 Inhibitors Versus Alprostadil With Survival in Men With Coronary Artery Disease
Andersson DP, Landucci L, Lagerros YT, Grotta A, Bellocco R, Lehtihet M, Holzmann MJ · Journal of the American College of Cardiology · 2021
Observational cohort · n = 16,548 on a PDE5 inhibitor vs 1,994 on alprostadil
Swapping the comparator from 'no treatment' to an active comparator shrank the mortality signal from HR 0.67 to HR 0.88 (0.79-0.98). The same 'more prescriptions, lower mortality' gradient also appeared among alprostadil users, a drug with no systemic cardiovascular mechanism. The authors state the design does not permit causal inference.
PubMed 33766260 ↗ - 25
The association of tadalafil exposure with lower rates of major adverse cardiovascular events and mortality in a general population of men with erectile dysfunction
Kloner RA, Stanek E, Desai U, Crowe C, Paige Ball D, Haynes V, Rosen R · Clinical Cardiology · 2024
Observational cohort · n = 8,156 tadalafil users vs 21,012 unexposed
MACE HR 0.81 (0.70-0.94) and all-cause mortality HR 0.56 (0.43-0.74). A 44% all-cause mortality reduction from an on-demand drug is larger than statins achieve in primary prevention (roughly 10%), which is a reason to doubt the design rather than celebrate the drug.
PubMed 38377018 ↗ - 26
Benefits of Tadalafil and Sildenafil on Mortality, Cardiovascular Disease, and Dementia
Jehle DVK, Sunesra R, Uddin H, et al. · The American Journal of Medicine · 2025
Observational cohort · n = 509,788 men with erectile dysfunction (TriNetX)
Tadalafil was associated with lower mortality (RR 0.66), heart attack (0.73), stroke (0.66), venous thromboembolism (0.79) and dementia (0.68) at once. Confidence intervals were not published in the abstract. A single exposure appearing to prevent five unrelated outcomes by the same third is the fingerprint of healthy-user bias.
PubMed 39532245 ↗ - 27
Associations between phosphodiesterase type 5 inhibitors and vascular function: a systematic review and meta-analysis on randomized-controlled trials
Zhang Y, Peng H, Xin J, et al. · Systematic Reviews · 2025
Meta-analysis of RCTs · 63 randomized trials, n = 3,242
Systolic BP -2.80 mmHg (95% CI -4.24 to -1.37), diastolic -1.80 mmHg (-2.37 to -1.22), flow-mediated dilation +2.47% (1.24 to 3.71). Carotid wall thickness fell 0.01 mm, which is below the resolution of the ultrasound used to measure it. Augmentation index, brachial diameter and reactive hyperaemia index were all unchanged.
PubMed 41469737 ↗ - 28
Effect of prolonged treatment with phosphodiesterase-5-inhibitors on endothelial dysfunction in vascular diseases and vascular risk conditions
D'Andrea S, Barbonetti A, Martorella A, Necozione S, Francavilla F, Francavilla S · International Journal of Clinical Practice · 2019
Meta-analysis of RCTs · 13 randomized trials, 469 treated vs 463 placebo
Flow-mediated dilation improved (p < 0.0001) but with heterogeneity I-squared = 92% and funnel-plot asymmetry indicating publication bias. No effect in resistance vessels. The endothelin-1 effect vanished once bias was corrected for. The authors call the results 'not convincing' and the evidence quality low.
PubMed 30471172 ↗ - 29
Chronic treatment with tadalafil improves endothelial function in men with increased cardiovascular risk
Rosano GM, Aversa A, Vitale C, Fabbri A, Fini M, Spera G · European Urology · 2005
Randomized controlled trial · n = 32 men at raised cardiovascular risk
Tadalafil 20 mg on alternate days for 4 weeks more than doubled flow-mediated dilation (4.2% to 9.3%, p < 0.01), and the improvement was still present two weeks after stopping (9.1%) - far longer than the drug stays in the body. Small, single-centre, and an effect size at the extreme end of what any intervention produces.
PubMed 15661417 ↗ - 30
Effect of phosphodiesterase-5 inhibition on exercise capacity and clinical status in heart failure with preserved ejection fraction (RELAX)
Redfield MM, Chen HH, Borlaug BA, Semigran MJ, Lee KL, Lewis G, LeWinter MM, et al. · JAMA · 2013
Randomized controlled trial · n = 216 across 26 centres, 24 weeks
Definitively null. Peak oxygen consumption differed by 0.01 mL/kg/min (95% CI -0.60 to 0.61) - a confidence interval tight enough to exclude any meaningful benefit. Six-minute walk favoured placebo (p = 0.92). Serious adverse events 22% on drug vs 16% on placebo.
PubMed 23478662 ↗ - 31
Sildenafil for improving outcomes in patients with corrected valvular heart disease and persistent pulmonary hypertension (SIOVAC)
Bermejo J, Yotti R, García-Orta R, et al. · European Heart Journal · 2018
Randomized controlled trial · n = 200, 6 months
Harm. The clinical composite favoured placebo, OR 0.39 (95% CI 0.22-0.67), p < 0.001; more patients worsened on drug (33) than on placebo (14). Heart-failure-admission-free survival HR 2.0 (1.0-4.0). The authors conclude off-label use for pulmonary hypertension from left heart disease 'should be avoided'.
PubMed 29281101 ↗ - 32
Sildenafil Treatment in Heart Failure With Preserved Ejection Fraction: Targeted Metabolomic Profiling in the RELAX Trial
Wang TJ, Anstrom KJ, Ilkayeva O, et al. · JAMA Cardiology · 2017
Randomized controlled trial
Zero metabolites changed over 24 weeks on placebo; seven changed on sildenafil, in a pattern the authors read as adverse effects on mitochondrial function and endoplasmic reticulum stress that 'could have contributed to the neutral trial results'.
PubMed 28492915 ↗ - 33
Tadalafil therapy for pulmonary arterial hypertension (PHIRST)
Galiè N, Brundage BH, Ghofrani HA, Oudiz RJ, Simonneau G, Safdar Z, et al. · Circulation · 2009
Randomized controlled trial · n = 405, 16 weeks, five arms
The pivotal trial behind the pulmonary hypertension approval. Only the 40 mg dose met significance; placebo-corrected six-minute walk distance improved 33 m (95% CI 15-50). Time to clinical worsening improved (p = 0.041). Change in WHO functional class was not significant.
PubMed 19470885 ↗ - 34
Initial Use of Ambrisentan plus Tadalafil in Pulmonary Arterial Hypertension (AMBITION)
Galiè N, Barberà JA, Frost AE, Ghofrani HA, Hoeper MM, McLaughlin VV, et al. · New England Journal of Medicine · 2015
Randomized controlled trial · n = 500, event-driven
Clinical failure occurred in 18% on ambrisentan plus tadalafil vs 31% on either alone: HR 0.50 (95% CI 0.35-0.72), p < 0.001. The one large event-driven trial in this literature that is clearly positive - and its endpoint is pulmonary-hypertension-specific, not heart attack or stroke.
PubMed 26308684 ↗ - 35
Phosphodiesterase 5 inhibitors for pulmonary hypertension
Barnes H, Brown Z, Burns A, Williams T · Cochrane Database of Systematic Reviews · 2019
Meta-analysis of RCTs · 36 trials, n = 2,999
Pooling the whole PDE5 inhibitor class, not tadalafil alone. In Group 1 pulmonary arterial hypertension: mortality OR 0.22 (95% CI 0.07-0.68) across 8 trials and 1,119 participants, rated high-certainty, number needed to treat 32; six-minute walk +48 m (40-56). In Group 2 pulmonary hypertension from left heart disease the same drugs did worse than placebo on functional class, OR 0.53 (0.32-0.87), which Cochrane reads as possible harm in valvular disease.
PubMed 30701543 ↗ - 36
Feasibility of high-dose tadalafil and effects on insulin resistance in well-controlled patients with type 2 diabetes (MAKROTAD)
Fryk E, Rodrigues Silva VR, Bauzá-Thorbrügge M, Schmelz M, Gan LM, Strindberg L, Jansson PA · eClinicalMedicine · 2023
Randomized controlled trial · 23 randomized, 18 analysed, crossover
Primary endpoint neutral: tadalafil 20 mg daily for 6 weeks did not improve whole-body insulin resistance measured by glucose clamp. HbA1c fell by 2.50 mmol/mol (95% CI -4.20 to -0.78), roughly 0.23%, as a secondary endpoint. Muscle pain, indigestion and headache were all more frequent on the drug.
PubMed 37256099 ↗ - 37
Sex-specific effects of daily tadalafil on diabetic heart kinetics in RECOGITO, a randomized, double-blind, placebo-controlled trial
Pofi R, Giannetta E, Feola T, et al. · Science Translational Medicine · 2022
Randomized controlled trial · n = 122, 20 weeks
Tadalafil 20 mg daily improved cardiac torsion by -3.40 degrees (95% CI -5.96 to -0.84) in men only, with no effect in women, and the sex difference was not explained by drug levels or cGMP. Both sexes improved on albuminuria, renal artery resistance and inflammatory monocyte profiles. Cardiac torsion is an imaging surrogate with no established link to outcomes.
PubMed 35704597 ↗ - 38
The Princeton IV Consensus Recommendations for the Management of Erectile Dysfunction and Cardiovascular Disease
Köhler TS, Kloner RA, Rosen RC, Burnett AL, Blaha MJ, Ganz P, Goldstein I, et al. · Mayo Clinic Proceedings · 2024
Guideline / commission report
The expert panel's firm conclusions are about safety: these drugs can be used alongside antihypertensives and in men with stable cardiovascular disease, with the absolute nitrate exception. Its cardioprotection language is explicitly sourced to retrospective studies.
PubMed 39115509 ↗ - 39
Cardiac Effects of Phosphodiesterase-5 Inhibitors: Efficacy and Safety
Roy S, Kloner RA, Salloum FN, Jovin IS · Cardiovascular Drugs and Therapy · 2023
Systematic review
Reviews why consistent animal cardioprotection (smaller infarcts, less hypertrophy and fibrosis) has not translated, and asks why trials have yet to show efficacy despite no safety concerns. 'Safe, but unproven' is the accurate summary of the field.
PubMed 34652581 ↗ - 40
Endophenotype-based in silico network medicine discovery combined with insurance record data mining identifies sildenafil as a candidate drug for Alzheimer's disease
Fang J, Zhang P, Zhou Y, Chiang CW, Tan J, Hou Y, et al. · Nature Aging · 2021
Case-control · 7.23 million insurance records
The study that started the story: sildenafil use associated with a 69% lower Alzheimer's risk, HR 0.31 (95% CI 0.25-0.39). No intervention in the dementia-prevention literature comes close to that, which is a reason for suspicion. The authors state the association 'does not establish causality or its direction, which requires a randomized clinical trial approach'.
PubMed 35572351 ↗ - 41
Phosphodiesterase Type 5 Inhibitors in Men With Erectile Dysfunction and the Risk of Alzheimer Disease
Adesuyan M, Jani YH, Alsugeir D, Howard R, Ju C, Wei L, Brauer R · Neurology · 2024
Observational cohort · n = 269,725 UK men, median 5.1 years
Headline result HR 0.82 (95% CI 0.72-0.93). The number almost never quoted is the three-year lag analysis, which excludes the period when undiagnosed dementia is already developing: HR 0.93 (0.80-1.08). The association disappears - the signature of the disease suppressing the prescription rather than the drug suppressing the disease.
PubMed 38324745 ↗ - 42
No association between initiation of phosphodiesterase-5 inhibitors and risk of incident Alzheimer's disease and related dementia: results from the DREAM study
Desai RJ, Mahesri M, Lee SB, et al. · Brain Communications · 2022
Target trial emulation · US Medicare, 76 confounders controlled
The best-designed observational test: PDE5 inhibitor starters compared with endothelin receptor antagonist starters inside the same pulmonary hypertension population, so both groups share the indication, the illness and the prescriber contact. Four analyses returned HR 0.99, 1.00, 0.67 and 1.15 - none significant. Cell assays also failed to reproduce the bench half of the original signal.
PubMed 36330433 ↗ - 43
Phosphodiesterase Type 5 Inhibitors and Risk of Dementia
Gronich N, Stein N, Saliba W · Neuroepidemiology · 2025
Observational cohort · n = 5,204 daily low-dose tadalafil users vs 18,565 on alpha-blockers; second cohort n = 133,336
The most directly on-point test of the low-dose daily regimen people are actually taking, using an active comparator with the same indication: HR 0.99 (95% CI 0.88-1.12), p = 0.927. A second cohort of 133,336 patients followed a mean 7.9 years with 8,631 dementia diagnoses gave HR 0.95 (0.86-1.04). The authors conclude PDE5 inhibitor use is not associated with decreased dementia risk.
PubMed 38952132 ↗ - 44
Genetically Proxied Phosphodiesterase-5 Inhibition and Alzheimer's Disease Risk
Alcalde-Herraiz M, Woolf B, Xie J, et al. · Aging Cell · 2025
Mendelian randomisation · Four Alzheimer's genome-wide association studies
Mendelian randomisation uses randomly inherited genetic variants as a natural experiment, so it is immune to healthy-user bias, prevalent-user bias, confounding by indication and reverse causation - the four problems that sink every supportive cohort study. Triangulated across four separate Alzheimer's datasets, it found no causal association. The authors conclude PDE5 inhibitors are unlikely to decrease Alzheimer's risk.
PubMed 41078087 ↗ - 45
A case-control study of phosphodiesterase-5 inhibitor use and Alzheimer's disease and related dementias
Henry DS, Pellegrino RG · PLOS ONE · 2023
Case-control
Reported 54-64% lower PDE5 inhibitor exposure among dementia cases. It also tested endothelin receptor antagonists, a class with no proposed brain mechanism, and found them 63.2% 'protective' too. A failed negative control the authors did not treat as one: the study is measuring who gets prescribed elective chronic medication.
PubMed 37851623 ↗ - 46
Tadalafil in Cerebral Small Vessel Disease: A Randomized, Placebo-Controlled Trial (ETLAS-2)
Ölmestig J, Mortensen KN, Thomas MB, et al. · Stroke · 2025
Randomized controlled trial · n = 76, daily tadalafil 20 mg for 3 months
The closest thing to a real test of daily tadalafil for brain health, run in exactly the population where a vascular mechanism should show up first. The primary endpoint was whether patients could even stay on the drug, and they could not: compliance OR 0.13 (95% CI 0.03-0.69), p = 0.030. Adverse events 76% on tadalafil vs 36% on placebo, OR 5.49 (1.81-18.07). No differences in cognition, wellbeing or blood pressure.
PubMed 40718899 ↗ - 47
Cognitive outcomes of tadalafil treatment in patients with cerebral small vessel disease
Ölmestig J, Mortensen KN, Thomas MB, et al. · Cerebral Circulation - Cognition and Behavior · 2025
Randomized controlled trial · n = 60 per protocol
Pre-registered cognitive sub-study of ETLAS-2. No difference between groups on any cognitive outcome. The single difference found was a slower simple reaction time in the tadalafil group (28.8 ms, p = 0.04), which the authors call possibly spurious. It remains the only randomized cognitive difference ever measured between tadalafil and placebo, and it favours placebo.
PubMed 41332889 ↗ - 48
The PASTIS trial: Testing tadalafil for possible use in vascular cognitive impairment
Pauls MMH, Binnie LR, Benjamin P, Ghatala R, Markus HS, et al. · Alzheimer's & Dementia · 2022
Randomized controlled trial · n = 55, single 20 mg dose, crossover
Primary endpoint not met: subcortical cerebral blood flow rose only non-significantly, the largest effect being +9.8% within white matter lesions (p = 0.096). Blood pressure fell 7.8/4.9 mmHg. The companion cognitive paper found no significant effect on any neuropsychological test.
PubMed 35135037 ↗ - 49
Cerebrovascular Effects of Sildenafil in Small Vessel Disease: The OxHARP Trial
Webb AJS, Birks JS, Feakins KA, Lawson A, Dawson J, Rothwell PM, et al. · Circulation Research · 2024
Randomized controlled trial · n = 75, three-way crossover, 3 weeks per arm
Primary endpoint missed: cerebral pulsatility unchanged, 0.02 (95% CI -0.01 to 0.05), p = 0.18. Secondary blood-flow measures did improve, including perfusion within white matter lesions (+1.82 mL/100 g/min, p = 0.008) and cerebrovascular reactivity. It measured no cognition, no dementia, no stroke - every endpoint was a three-week physiological surrogate.
PubMed 38832504 ↗ - 50
Evaluating the Safety and Efficacy of PDE-5 Inhibitor Tadalafil as a Treatment for Early Stages of Alzheimer's Disease
Imperial College London · ClinicalTrials.gov registry record · 2025
Ongoing trial · 244 planned, amyloid-positive mild impairment and early Alzheimer's
The first serious randomized test, and note what it is not: participants already have cognitive impairment, the primary endpoint is safety and tolerability rather than efficacy, and readout is expected around August 2028. No trial anywhere is testing whether tadalafil prevents dementia in healthy people.
ClinicalTrials.gov NCT07172815 ↗ - 51
Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission
Livingston G, Huntley J, Liu KY, Costafreda SG, Selbæk G, et al. · The Lancet · 2024
Guideline / commission report
Fourteen modifiable risk factors account for an estimated 45% of dementia cases. The vascular levers are blood pressure, LDL cholesterol, diabetes, smoking, physical activity, weight and alcohol - all with decades of randomized evidence. No PDE5 inhibitor appears anywhere in the Commission's risk factors or recommended interventions.
PubMed 39096926 ↗ - 52
Vascular dementia: World Stroke Organization fact sheet 2026
Cai Y, Mok VCT, Markus HS · International Journal of Stroke · 2026
Guideline / commission report
Vascular dementia is the second leading cause of dementia worldwide. In autopsy series pure vascular dementia is about 15% of cases and mixed vascular-plus-degenerative disease adds another 16%, roughly 31% in total. The fact sheet describes it as highly preventable, and attributes that preventability to cerebrovascular risk-factor control.
PubMed 41479247 ↗ - 53
Phosphodiesterase-5 inhibitors for cerebral small vessel disease-related ischemic stroke and cognitive decline
Yan H, Chen Y, Tao J, et al. · Frontiers in Neurology · 2026
Meta-analysis of RCTs · 4 randomized trials, n = 236
Pooling the randomized brain trials: cerebral blood flow within white matter lesions improved by 1.31 mL/100 g/min (95% CI 0.46-2.15) and diastolic blood pressure fell 4.65 mmHg, but there was no cognitive improvement in any domain and adverse events more than doubled, RR 2.36 (1.53-3.65).
PubMed 42051769 ↗ - 54
Sildenafil and risk of Alzheimer disease: a systematic review and meta-analysis
Chua WY, Lim LKE, Wang JJD, et al. · Aging (Albany NY) · 2025
Meta-analysis of observational studies · 5 observational studies, 885,380 patients
Pools the observational literature to a protective-looking summary, but with heterogeneity I-squared = 98% across studies whose estimates run from HR 0.31 to HR 1.00. At that level of disagreement the spread is the finding and the pooled number should not be quoted as a summary of the evidence.
PubMed 40096550 ↗ - 55
Risk of Ocular Adverse Events Associated With Use of Phosphodiesterase 5 Inhibitors in Men in the US
Etminan M, Sodhi M, Mikelberg FS, Maberley D · JAMA Ophthalmology · 2022
Case-control · 213,033 men taking a PDE5 inhibitor
Ischaemic optic neuropathy, a sudden and often permanent loss of vision in one eye, occurred at 3.2 cases per 10,000 person-years with an adjusted incidence rate ratio of 2.02 (95% CI 1.14-3.58). The label puts the background rate at 2.5 to 11.8 cases per 100,000 per year in men over 50, so even taking this estimate at face value the absolute increase is a few cases per 100,000 men per year.
PubMed 35389459 ↗ - 56
The association between phosphodiesterase type 5 inhibitors and ocular adverse events
Belladelli F, Del Giudice F, Glover F, Mulloy E, Muncey W, Basran S, et al. · The Journal of Sexual Medicine · 2023
Observational cohort · 1.94 million men with erectile dysfunction, 2.18 million with prostate symptoms
The largest analysis of the question found no association between PDE5 inhibitor use and serous retinal detachment, retinal vascular occlusion, ischaemic optic neuropathy or any ocular event compared with other erectile dysfunction treatments. Risk tracked the intensity of treatment generally, pointing at the underlying vascular disease rather than the drug.
PubMed 37861186 ↗ - 57
Risk of sudden sensorineural hearing loss in adults using phosphodiesterase type 5 inhibitors
Liu SY, Antonelli PJ, Dahm P, et al. · Pharmacoepidemiology and Drug Safety · 2018
Observational cohort · 377,722 men starting a PDE5 inhibitor vs 1,957,233 non-users
Sudden hearing loss occurred at 4.35 per 10,000 person-years during current use against 2.38 in non-users, adjusted HR 1.25 (95% CI 1.01-1.55). About two extra cases per 10,000 person-years. Sudden hearing loss is an instruction to stop the drug and seek care immediately.
PubMed 29512263 ↗ - 58
Long-term safety and tolerability of tadalafil in the treatment of erectile dysfunction
Montorsi F, Verheyden B, Meuleman E, Jünemann KP, Moncada I, Valiquette L, et al. · European Urology · 2004
Open-label extension · n = 1,173 men, 1,676 patient-years, 24 months
One of the longest safety datasets that exists for this drug, and it runs to two years of as-needed use in an open-label extension. Headache affected 15.8% and indigestion 11.8%; 6.3% stopped for side effects. There is no published safety dataset for taking tadalafil daily for a decade, and none at all in healthy people with no approved indication.
PubMed 15036680 ↗ - 59
Falsified tadalafil tablets distributed in Japan via the internet
Sanada T, Yoshida N, Matsushita R, Kimura K, Tsuboi H · Forensic Science International · 2020
Forensic analysis · 45 tadalafil samples bought online
Of 45 samples sold as tadalafil, only 9 were genuine. 23 were falsified and contained sildenafil instead of tadalafil - a different drug with a different half-life and a different nitrate washout window. Nine more contained neither drug, and some tablets did not dissolve properly. Outside the regulated supply chain, the molecule itself is the first assumption that fails.
PubMed 31931435 ↗
Identifiers are given so you can check any claim on this page against the source rather than trusting the summary. Where a number here disagrees with the source, the source is right.