Tadalafil (IC-351)

Identity, and a regulatory position that inverts the usual one

Tadalafil ran through development under three codes, IC-351, GF-196960 and LY450190, and all three sit in the synonym list attached to a single PubChem entry: CID 110635, CAS 171596-29-5, molecular formula C22H19N3O4, molecular weight 389.4 g/mol, InChIKey WOXKDUGGOYFFRN-IIBYNOLFSA-N. The molecule came out of ICOS and was developed with Eli Lilly. The FDA global substance registry carries it as UNII 742SXX0ICT with a calculated mass of 389.4048, and ChEMBL indexes the same structure as CHEMBL779. Where the identity does divide is stereochemistry, and that is not incidental here. The registry holds separate records under separate UNIIs for the (6R,12aS) and (6S,12aR) diastereomers. The marketed substance is the (6R,12aR) form.

Approval history is a matter of record rather than inference. The product approved under NDA 021368 was approved on 21 November 2003, first for erectile dysfunction and later for the signs and symptoms of benign prostatic hyperplasia. The 20 mg product for pulmonary arterial hypertension was approved under NDA 022332 on 22 May 2009. Two fixed-dose combinations followed: finasteride with tadalafil, under NDA 215423 on 9 December 2021, now carried as discontinued in the Drugs@FDA product listing, and macitentan with tadalafil, under NDA 218490 on 22 March 2024. This page names those products by application number rather than by trade name.

That position inverts the usual one on this site. Most compounds indexed here hold no approval anywhere and many have no registered trial at all. A ClinicalTrials.gov intervention search for tadalafil returned 286 studies: 181 recorded as completed, 71 carrying a phase 3 designation, 23 recruiting, 16 terminated, 11 withdrawn and 38 with a status of UNKNOWN. The useful question is therefore not whether human data exist. It is which of the claims in circulation those data actually cover.

The selectivity table, and the row that gets dropped

Selectivity figures are printed in full on the label for the product approved under NDA 021368, which makes the omissions elsewhere easy to see. The label states that tadalafil is more than 10,000-fold more potent for PDE5 than for PDE1, PDE2, PDE3, PDE4 and PDE7, more than 9,000-fold more potent than for PDE8, PDE9 and PDE10, and 700-fold more potent than for PDE6, the retinal enzyme of phototransduction. Two rows break the pattern. Tadalafil is given as 14-fold more potent for PDE5 than for PDE11A1 and 40-fold more potent than for PDE11A4, and the label adds that it inhibits recombinant PDE11A1, and to a lesser degree PDE11A4, at concentrations within the therapeutic range.

Published sources do not agree on what that means. Weeks and colleagues at Vanderbilt expressed and purified human PDE11A4, measured PDE5A1 over PDE11A4 selectivities of 40-fold for tadalafil against 9,300-fold for vardenafil and 1,000-fold for sildenafil, and read the 40-fold figure as evidence that cross-reaction in patients was unlikely (PMID 15538396). Bischoff, writing the previous year, described the same molecule as extremely selective for PDE5 but also a potent inhibitor of PDE11, an enzyme he characterised as having unknown physiological function (PMID 15224129). Pomara and Morelli filed a comment on the Weeks paper taking the opposite reading, under the heading that PDE11 inhibition impacts on sperm quality; the authors replied in the same issue (PMID 15995718).

Nothing published since has settled it. The label's own language is that the physiological role and clinical consequence of PDE11 inhibition in humans have not been defined, and it lists PDE11 as present in human prostate, testes, skeletal muscle and adrenal cortex. A PubMed search combining tadalafil with PDE11 and either back pain or myalgia returned no records, which is worth weighing against how often those two adverse reactions are explained by PDE11 on secondary pages.

Pharmacokinetics, and what the 36 hours measures

The largest pooled pharmacokinetic analysis on this file is Forgue and colleagues' 2006 integrated analysis of thirteen clinical pharmacology studies in healthy subjects, run at Lilly Research Laboratories (PMID 16487221). Pooling 237 subjects given single doses of 2.5 to 20 mg, it reported a mean terminal half-life of 17.5 hours, with 5th and 95th percentiles of 11.5 and 29.6 hours. Mean Cmax after 20 mg was 378 micrograms per litre at a median 2 hours. Oral clearance averaged 2.48 litres per hour and apparent volume of distribution 62.6 litres. Diurnal variation, food effects and dose proportionality were analysed in three of the thirteen studies rather than across the pooled 237. A separate multiple-dose arm of fifteen subjects per group reached steady state by day five with 1.6-fold accumulation.

Several parameters that circulate as facts are absent from the record. The NDA 021368 label states plainly that absolute bioavailability following oral dosing has not been determined, so any percentage quoted for it comes from somewhere other than the regulatory file. What the label does supply: 94 per cent plasma protein binding at therapeutic concentrations, predominant CYP3A4 metabolism to a catechol whose methylcatechol glucuronide is the major circulating metabolite, excretion mainly in faeces at roughly 61 per cent of dose and in urine at roughly 36 per cent, and less than 0.0005 per cent of an administered dose recovered in semen.

That 36-hour figure is real and routinely misdescribed. It comes from Porst and colleagues' 2003 randomised trial in 348 men, in which participants attempted intercourse approximately 24 or 36 hours after a 20 mg dose: 59.2 per cent of attempts at 36 hours were reported successful against 28.3 per cent on placebo, and 52.9 against 29.1 per cent at 24 hours (PMID 12837435). This is a duration-of-response endpoint captured by patient diary. It is not a half-life and it does not describe how long the molecule is present.

The trials behind the approved indications

Erectile dysfunction was the first indication and carries the largest dataset. Brock and colleagues published integrated analyses of five randomised, double-blind, placebo-controlled twelve-week trials totalling 1,112 men of mean age 59, with erectile dysfunction of mixed aetiology and severity (PMID 12352386). On 20 mg taken as needed, the erectile function domain of the International Index of Erectile Function rose by a mean 7.9 points from baseline against placebo, 75 per cent of intercourse attempts were completed on question 3 of the Sexual Encounter Profile, and 81 per cent reported improved erections at endpoint against 35 per cent on placebo. PubMed links an erratum at J Urol 2005;173(2):664 correcting a dosage error in the published abstract.

Prostate symptom trials produced a more mixed measurement set. Yokoyama and colleagues randomised 612 Asian men with lower urinary tract symptoms suggestive of benign prostatic hyperplasia to placebo, tadalafil 2.5 mg, tadalafil 5 mg or tamsulosin 0.2 mg for twelve weeks (PMID 22958078). Total International Prostate Symptom Score fell by 4.8 and 4.7 least-squares mean points on the two tadalafil arms against 3.0 on placebo. No significant improvement in peak urinary flow rate was observed on either tadalafil dose, nor in that trial on the tamsulosin control. The label states separately that the mechanism for reducing these symptoms has not been established.

PHIRST supplied the pulmonary arterial hypertension data. Galie and colleagues randomised 405 patients, treatment-naive or on background bosentan, to placebo or tadalafil at 2.5, 10, 20 or 40 mg once daily for sixteen weeks, with change in six-minute walk distance as the primary endpoint (PMID 19470885). Only the 40 mg arm met the prespecified significance level. The mean placebo-corrected treatment effect was 33 metres (95% CI 15 to 50), splitting into 44 metres in bosentan-naive patients against 23 metres (CI -2 to 48) in those already on bosentan. Change in WHO functional class was not statistically significant. PubMed links an erratum at Circulation 2011;124(10):e279 for a dosage error in the article text.

Endpoints the trials did not reach

Duchenne muscular dystrophy is the clearest case of a mechanism that held and an outcome that did not. Nelson and colleagues studied ten boys with the disease and ten age-matched healthy controls, found functional sympatholysis impaired in the patient group, and reported that single oral doses of tadalafil or sildenafil alleviated exercise-induced muscle ischaemia dose-dependently and normalised the blunted exercise-induced rise in forearm muscle blood flow (PMID 24808022). The journal graded it Class IV evidence, and graded it specifically for the proposition that PDE5 inhibition restores functional sympatholysis, which is narrower than the blood-flow finding reported alongside it. Three years later Victor and colleagues randomised 331 boys aged 7 to 14, all on glucocorticoids, to tadalafil at 0.3 or 0.6 mg per kilogram per day or placebo for 48 weeks (PMID 28972192).

That phase 3 trial missed. Declines in six-minute walk distance at 48 weeks were 51.0 plus or minus 9.3 metres on placebo, 64.7 plus or minus 9.8 on the low dose and 59.1 plus or minus 9.4 on the high dose, with p values of 0.307 and 0.538 against placebo, and no effect on any secondary outcome. The paper is graded Class I evidence that tadalafil does not slow ambulatory decline in that age band, and its registry record is carried as terminated for lack of efficacy. A prespecified exploratory upper-limb score declined less on the low dose in boys over ten, which the authors described as hypothesis-generating and nothing more.

Two further null results belong on the record because both are quoted in the affirmative elsewhere. The PASTIS trial gave 20 mg or placebo in crossover to 55 people with symptomatic cerebral small vessel disease, mean age 66.8, and measured subcortical cerebral blood flow by arterial spin labelling; flow rose in all subcortical regions without reaching significance, the largest effect being 9.8 per cent within white matter hyperintensities at P = .0960 (PMID 35135037). In healthy male athletes, Di Luigi and colleagues gave 20 mg or placebo in double-blind crossover to fourteen subjects and found VO2max and both the ventilatory and anaerobic thresholds unchanged in normoxia (PMID 17614028).

Small trials, and the claims built on them

Some of the smallest studies on this file carry the heaviest secondary traffic. Maggiorini and colleagues randomised 29 adults with a history of high-altitude pulmonary oedema to tadalafil 10 mg twice daily, dexamethasone 8 mg twice daily or placebo during an ascent from 490 metres and a two-night stay at 4,559 metres (PMID 17015867). Oedema developed in seven of nine on placebo and one of the eight remaining on tadalafil (P = 0.007), and systolic pulmonary artery pressure rose 13 mmHg on tadalafil against 28 on placebo. Acute mountain sickness incidence did not differ from placebo, and two tadalafil participants withdrew with severe acute mountain sickness on arrival.

Weed and colleagues ran a three-arm randomised study in patients undergoing definitive resection of oral and oropharyngeal squamous cell carcinoma, giving 10 mg daily, 20 mg daily or placebo for at least twenty days before surgery, with a registered enrolment of 35 on NCT00843635 (PMID 25320361). Blood and tumour myeloid-derived suppressor cells and regulatory T cells fell, and CD8-positive T cells reactive to autologous tumour antigen rose, each at P less than 0.05. The immunomodulatory effect was maximal at the intermediate dose and not at the higher one, which the authors attributed to an off-target action on PDE11.

Hormone claims trace to two small studies, neither of which reports what is attributed to it. Di Luigi and colleagues gave a single dose or placebo to nine healthy male athletes in double-blind crossover and sampled salivary steroids around a maximal exercise test: salivary testosterone rose after exercise on the tadalafil arm only, but salivary cortisol rose more on tadalafil than on placebo, and both the testosterone-to-cortisol and DHEAS-to-cortisol ratios fell (PMID 18559908). Francomano and colleagues reported a rise in the testosterone-to-oestradiol ratio across four weeks in twenty men with multiple sclerosis given 5 mg once daily, single-blind, with ten untreated men as controls and no placebo (PMID 27752863).

An observational signal, and material outside the supply chain

Fung and colleagues at the National Library of Medicine followed 1.1 million male Medicare enrollees aged 65 and over, diagnosed with benign prostatic hyperplasia and prescribed one of the study drugs between 2007 and 2020, for a mean 3.1 years. Taking tamsulosin as the reference, they reported tadalafil associated with risk reductions of 27 to 40 per cent across all three outcomes examined: Parkinson's disease, Alzheimer's disease and mortality (PMID 39172922). This is an observational cohort with an active drug as its comparator rather than placebo, the per-outcome hazard ratios for tadalafil are not given in the abstract, and the authors asked for work on the mechanism. ClinicalTrials.gov lists one phase 2 PDE5-inhibitor study in Alzheimer's disease, NCT07172815, enrolment 244, status not yet recruiting.

A separate literature concerns material that is not what it says. Twohig and colleagues applied an atmospheric solids analysis probe with time-of-flight mass spectrometry to counterfeit tablets and to five herbal products sold as natural alternatives, and found three patterns across the five: tadalafil or sildenafil added, both added together, and analogues of the approved molecules added (PMID 20878886). The analogues are the part that matters here. An analogue carries the marketing claim and none of the pharmacology, toxicology or regulatory file summarised above, and no figure on this page transfers to one.

Every pharmacokinetic and efficacy number recorded here was generated with a characterised pharmaceutical product of known content, in a supervised trial, in a defined population. Tadalafil is a prescription medicine in the United States and in the other jurisdictions where it holds a marketing authorisation. Material circulating outside that route has no established relationship to the studies described.

What is not known

The human file is large and narrow at the same time. Almost all of it was generated in men with a diagnosed condition, and the trials that stepped outside those populations largely failed to reach their endpoints: the phase 3 Duchenne muscular dystrophy trial in 331 boys returned Class I evidence of no effect on ambulatory decline and is carried on ClinicalTrials.gov as terminated for lack of efficacy, PASTIS did not meet its cerebral blood flow endpoint in 55 participants, and a crossover study in fourteen healthy athletes found no change in VO2max or either threshold. No trial has characterised chronic administration in healthy adults for any non-clinical purpose, so the exposure most relevant to unsupervised use is the one least studied. Absolute oral bioavailability has never been determined; the NDA 021368 label states so directly, which means every percentage figure quoted for it originates outside the regulatory file. The physiological role and clinical consequence of PDE11 inhibition remain undefined on the current label, twenty-one years after the exchange between Weeks and colleagues and their commenters, and nothing published has connected it to the back pain and myalgia that appear in the adverse-reaction tables — reactions that, on the as-needed erectile dysfunction table, run at 3, 3, 5 and 6 per cent for back pain and 1, 1, 4 and 3 per cent for myalgia against placebo rates of 3 and 1 per cent, equalling placebo at 5 mg, separating from it only at 10 and 20 mg, and running non-monotonically for myalgia. Long-term exposure data are thinner than the size of the programme suggests: the label reports 1,434, 905 and 115 subjects treated for at least six months, one year and two years respectively on once-daily dosing. Women are barely represented outside the pulmonary arterial hypertension work. Nothing here establishes a dose, a schedule or a route for any individual, and none of the trials described enrolled healthy people for chronic administration.

Questions