--- canonical_name: BPN14770 alternate_names: Zatolmilast, BPN-14770 canonical_topic: BPN14770 for Health & Longevity short_topic_lc: bpn14770 creation_date: 2026-0904-1200 creator_ai_fullname: Opus 5 ep_keywords: PDE4D Inhibitors, Phosphodiesterase-4D Inhibitors, PDE4 Inhibitors, Phosphodiesterase-4 Inhibitors --- # BPN14770 for Health & Longevity

Evidence Review created on 09/04/2026 using [AI4L](https://github.com/forever-healthy/AI4L) / Opus 5 **Also known as:** Zatolmilast, BPN-14770 ## Motivation BPN14770 is an experimental oral drug that partially blocks one narrow form of an enzyme brain cells use to clear away a short-lived signalling molecule involved in learning and memory. By slowing that clearance, it is meant to strengthen the chemical messages that help nerve cells build and keep their connections. It was deliberately designed to bind at an unusual site and to switch the enzyme down only part of the way, in the hope of avoiding the nausea that halted earlier drugs of the same family. The compound grew out of work on fragile X syndrome, an inherited condition that causes intellectual disability, and it carries a special regulatory status for that use. It has since been tested in people with early Alzheimer's disease, in healthy older volunteers, and in a second rare genetic condition. Interest beyond those groups rests on the claim that the same signalling pathway supports memory in ordinary aging. This review examines what the human and animal evidence actually shows: what the drug does, how large and how repeatable the reported effects are, what harms have been recorded, and what is still unknown. **[Benefits](#expected-benefits) - [Risks](#potential-risks--side-effects) - [Protocol](#therapeutic-protocol) - [Conclusion](#conclusion)** ## Recommended Reading High-level background on BPN14770, spanning the developer's own primary reports — Tetra Discovery Partners / Tetra Therapeutics, now owned by Shionogi, funded and co-authored nearly every one of them — an independent academic review, and the patient-advocacy foundation that co-funded and continues to champion the program. * [Shionogi Update on Zatolmilast Fragile X Clinical Trial Findings](https://www.fraxa.org/shionogi-update-on-zatolmilast-fragile-x-clinical-trial-findings/) - FRAXA Research Foundation A plain-language readout of the 2026 phase 3 topline. FRAXA co-funded and publicly advocates for this program, so its framing of the result is not a disinterested one. * [Inhibition of phosphodiesterase-4D in adults with fragile X syndrome: a randomized, placebo-controlled, phase 2 clinical trial](https://pubmed.ncbi.nlm.nih.gov/33927413/) - Berry-Kravis et al., 2021 The single positive controlled human trial, reported in full with per-endpoint effect sizes. Authored and funded by the developer, which is why the later independent replications matter so much. * [Design and Synthesis of Selective Phosphodiesterase 4D (PDE4D) Allosteric Inhibitors for the Treatment of Fragile X Syndrome and Other Brain Disorders](https://pubmed.ncbi.nlm.nih.gov/31013090/) - Gurney et al., 2019 The discovery paper. It explains the structural trick behind the compound and why a primate-specific amino acid forced the entire preclinical program into genetically humanized mice. * [The Emerging Role of Phosphodiesterase Inhibitors in Fragile X Syndrome and Autism Spectrum Disorder](https://pubmed.ncbi.nlm.nih.gov/41155624/) - Thomas et al., 2025 A narrative review by authors with no stake in the compound, placing BPN14770 alongside competing phosphodiesterase targets and naming the unresolved questions of isoform selectivity and dose timing. * [Memory enhancing effects of BPN14770, an allosteric inhibitor of phosphodiesterase-4D, in wild-type and humanized mice](https://pubmed.ncbi.nlm.nih.gov/30131563/) - Zhang et al., 2018 The clearest account of the animal memory data and of the vomiting-surrogate testing that underpins the tolerability claim; its value lies in the dose-response shape, not the headline. Coverage note: none of the six priority expert platforms has published anything on BPN14770 or zatolmilast. Both a general web search and a search restricted to those six domains returned no relevant content, which is unsurprising for a compound that has never been commercially available. ## Grokipedia No Grokipedia article exists for BPN14770. A direct site search returns only passing mentions inside articles on other subjects, and no dedicated, primary page for the compound. ## Examine No Examine article exists for BPN14770. This is expected: Examine covers supplements and nutrition, and BPN14770 is an unapproved investigational prescription-track drug, a category Examine does not typically cover. ## ConsumerLab No ConsumerLab article exists for BPN14770. ConsumerLab tests marketed dietary supplements and does not typically cover investigational prescription medications, which is the regulatory category BPN14770 occupies. ## Systematic Reviews No systematic reviews or meta-analyses for BPN14770 were found on PubMed as of September 04, 2026. The trade-off at the centre of this review is unrepresented on both sides: the evidence base contains neither a systematic review or meta-analysis of the claimed cognitive and functional benefit, nor one of the principal risk, which is gastrointestinal and neuropsychiatric tolerability. Every quantitative statement below therefore rests on individual trials. ## Mechanism of Action BPN14770 inhibits PDE4D (phosphodiesterase-4D, an enzyme that breaks down cAMP — cyclic adenosine monophosphate, the "second messenger" carrying signals inside a cell). Raising cAMP activates PKA (protein kinase A, which passes the signal on) and then CREB (cAMP response element-binding protein, a switch that turns memory-related genes on), whose targets include BDNF (brain-derived neurotrophic factor, a protein supporting synapse growth). In humanized mice this cascade improved LTP (long-term potentiation, the cellular basis of memory) and object-recognition memory ([Zhang et al., 2018](https://pubmed.ncbi.nlm.nih.gov/30131563/)). The design is unusual twice over. It binds not the catalytic site but UCR2 (upstream conserved region 2, a regulatory flap over the active site), capping inhibition near 80–90% rather than shutting the enzyme down, the basis of the reduced-nausea claim ([Burgin et al., 2010](https://pubmed.ncbi.nlm.nih.gov/20037581/)). That flap carries a tyrosine in primates but a phenylalanine in rodents, so ordinary mice barely respond. A competing reading exists: inhibiting the closely related PDE4B isoform protects rodent memory just as well ([Zhao et al., 2024](https://pubmed.ncbi.nlm.nih.gov/38056709/)). Pharmacology: IC50 (half-maximal inhibitory concentration) is roughly 7–8 nM at PDE4D3 and PDE4D7; half-life is 11–13 hours, peaking at 1.5–3 hours, with linear kinetics from 5–100 mg. It is brain-penetrant, with PET (positron emission tomography) binding highest in prefrontal and temporal cortex and hippocampus ([Wakabayashi et al., 2020](https://pubmed.ncbi.nlm.nih.gov/32212718/)). Clearance is low-extraction hepatic oxidation: about 16% extraction in human liver cells, one minor metabolite, roughly 99.5% protein binding. Interaction work used midazolam, the standard probe for CYP3A4 (a liver enzyme that clears many drugs). ## Historical Context & Evolution PDE4 (phosphodiesterase-4) inhibition is not new. Rolipram, developed in the 1970s as an antidepressant, showed antidepressant and memory-enhancing activity but was abandoned because nausea and vomiting appeared at or below therapeutic doses. That dose-limiting nausea, not a lack of efficacy, is what stalled the class for three decades. The route around it came from structural biology. Co-crystal structures published in 2010 showed the enzyme's regulatory domain closing across its own active site, and that insight allowed the design of molecules that stabilize the closed form and therefore inhibit only partially ([Burgin et al., 2010](https://pubmed.ncbi.nlm.nih.gov/20037581/)). Mark Gurney, an author on that work, founded Tetra Discovery Partners to develop the chemistry; BPN14770 was the lead that emerged ([Gurney et al., 2019](https://pubmed.ncbi.nlm.nih.gov/31013090/)). The FRAXA Research Foundation, which had already pointed to phosphodiesterase inhibition for fragile X syndrome, partnered with Tetra to test the compound in fragile X mice and then co-funded the first human trial ([Gurney et al., 2017](https://pubmed.ncbi.nlm.nih.gov/29116166/)). Two phase 1 studies in 109 healthy adults ran in 2015–2016; a phase 2 Alzheimer's trial missed its primary endpoint in 2020; Shionogi acquired Tetra the same month. The 2021 fragile X phase 2 result was widely read as a breakthrough. Two phase 3 trials reported in 2026 did not reproduce its primary cognitive finding. Which reading survives is not yet settled: the developer and the foundation both argue the endpoint, not the drug, may be at fault, and the open-label extension continues. ## Expected Benefits ### High 🟩 🟩 🟩 #### Caregiver-Rated Daily Functioning in Fragile X Syndrome Caregivers rating adult men on language use and everyday functioning scored them better on drug than on placebo. In the 30-man crossover trial this reached statistical significance on both visual-analogue ratings, and the effect reappeared in the much larger adult phase 3 trial on the fragile-X numerical rating scale, making this the only BPN14770 finding replicated across two controlled trials ([Berry-Kravis et al., 2021](https://pubmed.ncbi.nlm.nih.gov/33927413/)). The adolescent phase 3 trial found nothing comparable. Caregiver ratings are unblinded in practice, and expectancy inflates them. **Magnitude:** In the phase 2 trial, +14.04 mm for language (P = 0.0051; P is the probability a difference this large would arise by chance alone) and +14.53 mm for daily functioning (P = 0.0017) on 100 mm visual-analogue scales; the adult phase 3 trial reported a statistically significant advantage on its caregiver rating scale without releasing a figure. ### Medium 🟩 🟩 No benefit reaches Medium: every single-trial human result here is internally conflicted — a dose-dependent reversal within the phase 1 study in healthy older adults, and a phase 2 cognitive endpoint that two subsequent phase 3 trials failed to reproduce — so each falls to Low. ### Low 🟩 #### Directly Tested Crystallized Cognition in Fragile X Syndrome ⚠️ Conflicted Direct computerized testing of vocabulary and oral reading — the "crystallized" domain, meaning accumulated knowledge — improved in the phase 2 trial, then failed in both phase 3 trials built around exactly that endpoint ([Berry-Kravis et al., 2021](https://pubmed.ncbi.nlm.nih.gov/33927413/)). Net reading: the early signal did not survive powered replication. **Magnitude:** Phase 2 crystallized composite +5.31 points (P = 0.0018), oral reading recognition +2.81 (P = 0.0157), picture vocabulary +5.81 (P = 0.0342); both phase 3 trials reported no significant difference from placebo on the same composite. #### Working Memory in Cognitively Normal Older Adults ⚠️ Conflicted The only data in the target audience. In 45 older volunteers, pooled 10 mg and 20 mg beat placebo on a working-memory task while 40 mg worsened — post hoc, unpublished ([Alzforum database entry](https://www.alzforum.org/therapeutics/bpn14770)); humanized-mouse work shows the same narrow window ([Zhang et al., 2018](https://pubmed.ncbi.nlm.nih.gov/30131563/)). Net reading: suggested, not established. **Magnitude:** A reported effect size above 0.70 on two measures of working memory, including a one-card-back task, in the pooled 10 and 20 mg groups (20 on drug, 15 on placebo; P < 0.05), set against worse measured performance at 40 mg twice daily in the same study. #### Memory and Global Function in Early Alzheimer's Disease ⚠️ Conflicted The 255-participant [PICASSO AD trial](https://clinicaltrials.gov/study/NCT03817684) found no memory benefit at either dose, despite the amyloid-model rationale behind it ([Cui et al., 2019](https://pubmed.ncbi.nlm.nih.gov/31488603/)); the sponsor later highlighted a post-hoc signal on a global dementia rating at 25 mg ([Alzforum database entry](https://www.alzforum.org/therapeutics/bpn14770)). Net reading: null, with a hypothesis-generating subgroup. **Magnitude:** Delayed-memory index change at week 13 was +10.1, +9.9 and +9.7 points for 10 mg, 25 mg and placebo respectively — differences of well under one point on the scale. ### Speculative 🟨 #### Normalization of Cortical Hyperexcitability Auditory evoked-response amplitude and resting peak alpha frequency — markers of cortical hyperexcitability (nerve cells firing too readily) — shifted with drug exposure ([Norris et al., 2024](https://pubmed.ncbi.nlm.nih.gov/39488698/)). Unvalidated electrophysiological biomarkers, not outcomes. #### Synaptic Plasticity and Neurotrophic Signalling Chronic dosing raised brain cAMP, phospho-CREB and BDNF and improved the shape of nerve-cell connection points (dendritic spines) in mice, and protected against amyloid-induced deficits ([Cui et al., 2019](https://pubmed.ncbi.nlm.nih.gov/31488603/)). The basis is animal and mechanistic. #### Sleep Duration and Social Behavior Dietary dosing lengthened sleep and increased social interaction in both fragile-X-model and wild-type mice ([Rosenheck et al., 2021](https://pubmed.ncbi.nlm.nih.gov/34411704/)). No human sleep or social outcome has been measured; the basis is animal only. ## Benefit-Modifying Factors * **PDE4D genotype:** Common PDE4D variants alter enzyme expression and have been linked to stroke risk in pooled analyses ([Yoon et al., 2011](https://pubmed.ncbi.nlm.nih.gov/21677445/)). Whether they shift response to a PDE4D inhibitor of this type has never been tested. * **Baseline cAMP signalling tone:** Reduced cAMP is documented in fragile X cells and models. Someone whose signalling is already intact has less capacity for improvement, which is the simplest explanation for why the healthy-volunteer gains were small and dose-restricted. * **Baseline cognitive performance:** Every positive human signal came from impaired populations or from tasks with measurable headroom. Ceiling effects on crystallized measures make gains hard to detect in high-functioning adults. * **Sex:** No sex-based analysis exists. Both fragile X trial programs enrolled males only, so the entire replicated efficacy signal comes from men; the Alzheimer's and phase 1 studies enrolled both sexes without reporting a split. * **Pre-existing conditions:** Severe kidney or liver impairment changes exposure, which dedicated 2025 studies were run to quantify. Depression or anxiety may interact with a class known for mood effects. * **Age:** Phase 1 dosing covered young and elderly volunteers, and the adverse cognitive effect at 40 mg twice daily appeared specifically in the older cohort — the age band most of this review's audience occupies. ## Potential Risks & Side Effects ### High 🟥 🟥 🟥 #### Gastrointestinal Adverse Events (Nausea, Vomiting and Diarrhea) Nausea and diarrhea are the signature of PDE4 inhibition and were the reason the older compounds failed ([Burgin et al., 2010](https://pubmed.ncbi.nlm.nih.gov/20037581/)). Partial inhibition reduces but does not remove them: in the Alzheimer's trial nausea occurred on both active doses and in nobody on placebo, and diarrhea roughly doubled. In the fragile X trial vomiting was slightly more common on drug ([Berry-Kravis et al., 2021](https://pubmed.ncbi.nlm.nih.gov/33927413/)). Events were mild and none led to withdrawal. **Magnitude:** Nausea in 4/80 (5.0%) at 10 mg and 5/85 (5.9%) at 25 mg twice daily versus 0/86 on placebo; diarrhea 5/80 and 6/85 versus 3/86; vomiting 3/30 versus 2/30 in the fragile X crossover. ### Medium 🟥 🟥 #### Dose-Dependent Cognitive Worsening at Higher Doses The same phase 1 study that produced the low-dose memory signal recorded worse cognitive performance in the 40 mg twice-daily group of older volunteers ([Alzforum database entry](https://www.alzforum.org/therapeutics/bpn14770)). Inverted-U dose-response — benefit rising, then reversing as dose climbs — is a known feature of cAMP-raising drugs, and animal work shows benefit concentrated in a narrow window ([Zhang et al., 2018](https://pubmed.ncbi.nlm.nih.gov/30131563/)). For anyone self-dosing without a ceiling, this is the most consequential finding in the human record. **Magnitude:** Direction is a decline in measured cognition at 40 mg twice daily in adults around age 60, appearing within a week of dosing; the source study was never published and reports no outcome figure for the decrement. #### Headache Headache was the most frequently reported adverse effect in the seven-day multiple-ascending-dose study in older volunteers, concentrated in the highest dose group and consistent with the vasodilatory headache seen across PDE4 inhibitors ([Blauvelt et al., 2023](https://pubmed.ncbi.nlm.nih.gov/37924462/), a narrative review of the selective PDE4B/D class). It was not described as severe, though the one participant who left that cohort early, by choice, was among those affected. **Magnitude:** Headache in 5 of 10 volunteers in the 40 mg twice-daily elderly cohort of that seven-day study, where it was the most frequently reported adverse event; the study is unpublished and no placebo comparator figure has been released. ### Low 🟥 #### Neuropsychiatric Effects: Depressed Mood, Anxiety and Suicidal Ideation Approved PDE4 inhibitors carry psychiatric signals, including depression and suicidal ideation in post-marketing surveillance ([Ren et al., 2025](https://pubmed.ncbi.nlm.nih.gov/40195166/)). Nothing similar has been reported for BPN14770, yet its sponsor treats it as live: suicidality is a named safety endpoint in the [Jordan's syndrome trial](https://clinicaltrials.gov/study/NCT06717438). The evidence is class-extrapolated, therefore indirect. **Magnitude:** Not quantified in available studies. No BPN14770 trial has reported a psychiatric adverse-event rate, and the class figures derive from a different molecule, dose range and patient population. #### Weight Loss Roflumilast causes dose-related weight loss in controlled trials ([Yuan et al., 2016](https://pubmed.ncbi.nlm.nih.gov/27418821/), a meta-analysis in chronic obstructive pulmonary disease). Whether a partial, subtype-selective inhibitor does the same is untested; no BPN14770 trial has reported weight as an outcome. For an audience already managing lean mass, unintended loss would matter. **Magnitude:** Direction is weight reduction under sustained PDE4 inhibition, most pronounced in the first six months and at lower starting weight; no outcome figure exists for BPN14770 itself. #### Insomnia and Disturbed Sleep ⚠️ Conflicted Insomnia is a labelled adverse reaction of roflumilast, the closest approved comparator ([Yuan et al., 2016](https://pubmed.ncbi.nlm.nih.gov/27418821/), a meta-analysis in chronic obstructive pulmonary disease), and raised brain cAMP is a plausible route. Rodent work with this compound points the other way, toward longer sleep. Net reading: the direction is unresolved. **Magnitude:** Not quantified in available studies. No BPN14770 trial has reported sleep or insomnia as an outcome, and the class figures come from a different molecule in a respiratory population. ### Speculative 🟨 #### Raised Exposure to Transporter-Substrate Drugs A 2025 [interaction study](https://clinicaltrials.gov/study/NCT07011992) tested whether BPN14770 raises rosuvastatin levels via two drug transporters. Results are not posted. The concern — statin accumulation and muscle injury — is mechanistic only. #### Mesenteric Vasculopathy Older PDE4 inhibitors caused inflammatory damage to the blood vessels supplying the intestine in rats and dogs. The developer reports reduced vascular toxicity for BPN14770 ([Gurney et al., 2019](https://pubmed.ncbi.nlm.nih.gov/31013090/)). No human data exist. ## Risk-Modifying Factors * **PDE4D genotype:** Variants altering PDE4D expression could plausibly change both effect and tolerability. No pharmacogenetic analysis has been published for this compound, so this remains a theoretical modifier. * **Baseline liver and kidney function:** Dedicated phase 1 studies in severe renal impairment and in hepatic impairment were run in 2025 precisely because exposure is expected to rise. Impaired clearance shifts a narrow dose window upward. * **Sex:** Both fragile X programs enrolled males only. No sex-stratified safety data exist, so nothing is known about whether women experience the gastrointestinal or psychiatric effects differently. * **Pre-existing mood disorder:** A history of depression, anxiety or suicidal ideation is the clearest amplifier of the class psychiatric signal, and is an exclusion criterion in the sponsor's current trials. * **Age:** The cognitive decrement at 40 mg twice daily was seen in volunteers around age 60. Older users have less renal and hepatic reserve and are the group in which the adverse dose signal actually appeared. * **Baseline weight and lean mass:** Lower starting weight predicts larger PDE4-inhibitor-associated weight loss, so leaner older adults carry more downside from a class effect never measured for this compound. ## Key Interactions & Contraindications * **Statins and other transporter substrates (rosuvastatin, atorvastatin):** Caution. BPN14770 was formally tested for inhibition of two drug transporters using rosuvastatin. Consequence is statin accumulation and myopathy (muscle damage). Mitigation: dose separation, lowest effective statin dose, creatine kinase monitoring. * **Strong CYP3A4 inhibitors (ketoconazole, ritonavir, clarithromycin, grapefruit juice):** Caution. Interaction work used midazolam as its probe for this enzyme. Consequence is raised BPN14770 exposure, pushing into the dose range associated with cognitive worsening. Mitigation: avoidance rather than dose adjustment. * **Strong CYP3A4 inducers (rifampicin, carbamazepine, St John's wort):** Caution. Consequence is loss of exposure and of any effect. Mitigation: avoidance of concurrent use, since the dose ceiling is set by tolerability rather than by efficacy. * **Cholinesterase inhibitors (donepezil, rivastigmine — memory drugs that slow the breakdown of acetylcholine, a nerve-signalling chemical):** Monitor. Both converge on the same cascade; a dedicated interaction study used donepezil. Consequence is additive gastrointestinal upset. Mitigation: staggered introduction. * **Over-the-counter caffeine and theophylline:** Monitor. Both are non-selective phosphodiesterase inhibitors. Consequence is additive cAMP elevation with jitteriness, nausea and headache. Mitigation: separation of high-dose caffeine from dosing by several hours. * **Over-the-counter acid suppressants and antacids (omeprazole, calcium carbonate):** Monitor. BPN14770 is essentially insoluble in water. Consequence is erratic absorption and unpredictable exposure. Mitigation: a stable rather than intermittent schedule of acid-modifying products. * **Supplements that raise cAMP (forskolin from *Coleus forskohlii*, luteolin, artichoke leaf extract, resveratrol):** Caution. All inhibit phosphodiesterases or stimulate cyclase. Consequence is additive gastrointestinal upset and headache. Mitigation: no stacking; one variable at a time. * **Supplements affecting absorption (high-dose fish oil and other fat-based carriers):** Monitor. A lipophilic drug taken with a variable fat load gives variable exposure. Consequence is inconsistent effect. Mitigation: consistent timing relative to meals. * **Alcohol:** Caution. PDE4 inhibition alters alcohol intake and response in preclinical and early clinical work. Consequence is an unpredictable subjective effect and worse nausea. Mitigation: abstention during dose-finding. * **Other interventions:** Monitor. Any concurrently trialled nootropic or cognitive protocol makes the inverted-U dose window impossible to locate, since the only usable feedback is a change in measured cognition. **Populations who should avoid BPN14770:** * Anyone outside a clinical trial, since no pharmaceutical-grade product exists * Pregnancy and breastfeeding — no reproductive toxicology is public * Severe renal impairment (estimated glomerular filtration rate below 30 mL/min/1.73 m²) * Moderate to severe hepatic impairment (Child-Pugh Class B or C, a standard grading of liver disease severity) * Active major depressive disorder, bipolar disorder, or any history of suicidal ideation * Body weight under 25 kg, the sponsor's own floor for weight-adjusted dosing * Anyone taking a strong CYP3A4 inhibitor that cannot be stopped ## Risk Mitigation Strategies * **Dose ceiling below the phase 1 harm threshold:** Protocols derived from trial data stay at or under 20 mg twice daily, since the only recorded cognitive decrement in healthy older adults occurred at 40 mg twice daily. * **Low starting dose with slow escalation:** Starting at 10 mg once daily for a week before moving to twice daily limits first-exposure nausea, the class-defining adverse effect, and surfaces individual sensitivity early. * **Dosing with food:** Taking the capsule with a meal blunts nausea and vomiting, the highest-graded risk here, and stabilizes absorption of a compound that is effectively water-insoluble. * **Objective cognitive testing rather than impression:** A validated computerized battery run at baseline and every four weeks makes the inverted-U dose response visible; without it, a drug-induced decrement is indistinguishable from normal variation. * **Formal mood screening and tracking:** A depression and anxiety questionnaire completed before starting and every four weeks, with any new suicidal ideation treated as a stop signal, targets the psychiatric signal documented for approved PDE4 inhibitors. * **Creatine kinase monitoring alongside statins:** A baseline and six-week check targets the transporter-mediated statin accumulation and muscle injury that the 2025 interaction study was designed to quantify. * **Weekly weighing:** Unintended loss beyond 5% of body weight is the conventional stop signal for the class weight-loss effect, which matters most for leaner older adults with limited reserve. * **No stacking of cAMP-raising compounds:** Adding another phosphodiesterase inhibitor while tolerance is being established makes any adverse effect unattributable and compounds the gastrointestinal and headache burden. ## Therapeutic Protocol * **No clinical practice exists:** BPN14770 is unapproved everywhere. There is no prescribing physician, clinic or compounding pharmacy protocol; everything below is reconstructed from registered trial regimens. * **Trial regimen in adults:** 25 mg twice daily by mouth, the dose used in the fragile X phase 2 trial, in the adult phase 3 trial, and as the higher of two arms in the Alzheimer's trial. * **Lower-dose alternative:** 10–20 mg twice daily. This is the range that improved working memory in healthy older volunteers, and the range below the dose linked to cognitive worsening in that same study. * **Competing approaches:** Neither dosing philosophy is the default. The sponsor optimized 25 mg for a disease population with reduced signalling; the low-dose reading comes from the only data in cognitively normal adults. * **Best time of day:** Morning and early afternoon. With an 11–13 hour half-life, evening dosing carries exposure through the night, and rodent work shows altered sleep duration. * **Half-life:** 11–13 hours in humans, with peak concentration at 1.5–3 hours and dose-proportional exposure from 5 to 100 mg as a single dose. * **Split versus single dose:** Every trial used twice-daily dosing. Splitting keeps peak concentration lower, which matters when the adverse effects — nausea, headache — track peak rather than average exposure. * **Genetic polymorphisms:** No pharmacogenetic guidance exists. PDE4D variants and CYP3A4 variability are the plausible candidates, and neither has been studied for this compound. * **Sex-based differences:** Unknown. Both fragile X programs enrolled males only; no trial has reported a sex-stratified dose or response analysis. * **Age considerations:** Dosing scales downward with age. The 40 mg twice-daily cognitive decrement occurred specifically in volunteers around 60, and renal and hepatic clearance both decline across that range. * **Baseline biomarkers:** No biomarker predicts response. The exploratory candidates — evoked-response amplitude, resting peak alpha frequency — require electroencephalography and remain unvalidated. * **Pre-existing conditions:** Reduced kidney or liver function raises exposure and shifts dosing down; the sponsor ran dedicated impairment studies in 2025 rather than assuming no effect. ## Discontinuation & Cycling * **Intended duration:** Open-ended in the disease setting. The extension study doses participants for up to four years, and no trial has framed BPN14770 as a short course. * **Withdrawal effects:** None documented. The sponsor nonetheless administers a discontinuation-emergent signs and symptoms questionnaire after stopping in its current trial, so the question is treated as open. * **Tapering:** No taper protocol exists. Trials stopped dosing abruptly at the end of each period without reported rebound, which argues a taper is unnecessary rather than proven safe. * **Persistence after stopping:** The phase 2 fragile X benefit was reported to persist up to 12 weeks after the last dose, and the animal behavioural benefit persisted two weeks after washout. * **Cycling:** Never studied. There is no tolerance signal to cycle against, but neither is there data showing continuous dosing keeps working, so both patterns are unevidenced. * **Why persistence matters:** If effects outlast exposure, a crossover design contaminates its own second period — a plausible contributor to the ambiguity in the trial record. ## Sourcing and Quality * **No medicinal product exists:** BPN14770 is available only inside a clinical trial. There is no pharmacy formulation, no generic, and no legal route to a medicinal-grade supply. * **What is actually sold:** Research chemical, catalogued under the chemical registry (CAS) number 1606974-33-7 and sold explicitly for laboratory use only by suppliers such as Cayman Chemical, MedChemExpress and Selleck Chemicals. * **What to look for:** A batch-specific certificate of analysis with a stated purity figure, an NMR (nuclear magnetic resonance) spectrum confirming identity, and a safety data sheet. Reputable catalogue suppliers publish all three; marketplace resellers typically do not. * **Third-party testing:** Independent verification falls to the purchaser. Catalogue purity claims are self-reported, and no third-party certification programme covers research chemicals, unlike the marketed supplement category. * **Compounding pharmacies:** Not an option. Compounding requires an approved active ingredient, so no legitimate pharmacy can prepare BPN14770 in any jurisdiction. * **Formulation:** The trial product is a capsule. The raw powder is practically insoluble in water and soluble in ethanol and dimethyl sulfoxide, which makes accurate low-dose home preparation genuinely difficult. ## Practical Considerations * **Time to effect:** Rapid where it appears at all. The healthy-volunteer working-memory signal emerged after the first dose; the fragile X trials measured their endpoints at 12–13 weeks. * **Common pitfall — assuming more is better:** The single most consequential error. The only dose above 25 mg twice daily ever tested over a week produced worse cognition, not better. * **Common pitfall — trusting subjective impression:** Every replicated human signal here is a caregiver rating, and every direct measurement in adequately powered trials was null. Self-assessment is exactly the instrument that failed. * **Common pitfall — ignoring the population gap:** All efficacy data come from people with a specific inherited signalling deficit. Extrapolation to intact signalling has no empirical support. * **Regulatory status:** Investigational worldwide, with orphan drug designation (a regulatory status for rare-disease treatments) for fragile X syndrome. Use outside a trial is unapproved use of an unapproved drug, not off-label prescribing. * **Cost and accessibility:** Legitimate access is limited to trial participation. Research-grade powder is inexpensive relative to any approved cognitive medication, which is precisely what makes unsupervised dosing likely. * **Payer incentives:** Should BPN14770 reach approval as an orphan drug, institutional payers would face a branded price against the generic off-label agents currently used in fragile X, giving them a structural incentive to favour the incumbents in coverage and guideline decisions. ## Interaction with Foundational Habits * **Sleep:** Direct and bidirectional. Chronic dosing lengthened sleep in both fragile-X-model and wild-type mice ([Rosenheck et al., 2021](https://pubmed.ncbi.nlm.nih.gov/34411704/)), and PDE4D inhibition prevented sleep-deprivation-induced memory deficits ([Zhao et al., 2025](https://pubmed.ncbi.nlm.nih.gov/40490297/)). Practical consequence: trial dosing falls in the morning, and the sleep-deprivation data describe protection in an impaired state. * **Nutrition:** Indirect, via absorption. The compound is effectively water-insoluble, so a variable fat load produces variable exposure, and trial dosing was standardized against meals. Stacking dietary phosphodiesterase inhibitors — high-dose caffeine, artichoke leaf extract, luteolin — adds gastrointestinal upset without adding evidence. * **Exercise:** None known, in either direction. No trial measured strength, endurance or hypertrophy. One practical caution: strenuous training raises creatine kinase and confounds the muscle-injury monitoring relevant to anyone combining this compound with a statin. * **Stress management:** Potentiating, mechanistically. The cAMP-to-CREB cascade this drug amplifies is the same one that antidepressants and chronic stress both act on, which is why the class carries mood effects in both directions. Practical consequence: formal mood tracking is more informative than impression. ## Monitoring Protocol & Defining Success Before a first dose, a baseline capable of distinguishing real change from noise includes liver enzymes (ALT and AST), kidney function, a complete blood count, creatine kinase, a fasting lipid panel, body weight, and formal depression and anxiety questionnaires (PHQ-9 and GAD-7). Because the claimed benefit is cognitive and the recorded harm is also cognitive, a validated computerized battery is run at least twice before starting, so that practice effects are exhausted and the second run becomes the true comparator. Thereafter, mood questionnaires and cognitive testing repeat every four weeks, weight weekly, and bloodwork at six weeks, three months, and then every six months. Success is a measured gain on the cognitive battery exceeding the spread of the baseline runs, held without weight loss, mood change or persistent nausea. | Biomarker | Optimal Functional Range | Why Measure It? | Context/Notes | |---|---|---|---| | Body weight | Within 2% of personal baseline | Class-wide weight loss with sustained PDE4 inhibition | Fasted, same time of day; unintended loss beyond 5% is the conventional stop signal | | ALT and AST | ALT below 25 U/L (men), below 20 U/L (women) | Liver strain from an unapproved, hepatically cleared compound | ALT and AST are liver enzymes; conventional labs flag only above roughly 40 U/L; no fasting needed | | Creatinine and eGFR | eGFR above 90 mL/min/1.73 m² | Reduced kidney filtering raises drug exposure | eGFR is the estimated glomerular filtration rate; cystatin C is better at high muscle mass; conventional cut-off is 60 | | Creatine kinase | Below 150 U/L | Muscle breakdown (rhabdomyolysis) occurred once on drug, and transporter inhibition may raise statin levels | Strenuous exercise within 48 hours of the draw makes the result uninterpretable | | Fasting lipid panel with LDL cholesterol | LDL cholesterol below 70 mg/dL when taking a statin | Detects changed statin exposure from the transporter interaction | LDL is low-density lipoprotein, the cholesterol-carrying particle statins lower; 12-hour fast; best paired with apolipoprotein B, which tracks particle number | | Complete blood count | Within conventional reference range | General safety screen for a compound with no published long-term record | No known effect on blood counts; the value is a personal comparator | | PHQ-9 and GAD-7 scores | Both below 5 | The psychiatric class signal, including suicidal ideation | Self-administered depression and anxiety questionnaires; any new suicidal thought is a stop signal | | Computerized cognitive battery | No established target exists; track change from the individual's own baseline | Both the claimed benefit and the documented dose-dependent harm | At least two pre-start runs exhaust practice effects; same time of day, same sleep state | Qualitative markers worth tracking alongside the numbers: * Word-finding fluency and ease of conversation, the domain in which every positive human signal appeared * Mental clarity and sustained concentration during demanding work * Nausea, appetite and stool consistency, the earliest indicators of dose excess * Headache frequency and character * Sleep duration and how rested mornings feel * Mood stability and irritability, assessed by someone in the household rather than by self-report alone ## Emerging Research * **Open-label extension:** [NCT05367960](https://clinicaltrials.gov/study/NCT05367960) is dosing 314 fragile X participants for up to four years and will be the first long-term human safety dataset for this compound. Shionogi has confirmed it continues despite the phase 3 misses. * **Second indication:** [NCT06717438](https://clinicaltrials.gov/study/NCT06717438) is a 30-participant phase 2 trial in disorder of PPP2R5D (a gene encoding a regulator of a key cellular enzyme), also called Jordan's syndrome. Its primary efficacy endpoint is a caregiver language rating. Completion is estimated for 2027. * **Special-population pharmacokinetics:** Phase 1 studies in severe renal impairment ([NCT07012005](https://clinicaltrials.gov/study/NCT07012005)) and hepatic impairment ([NCT07018492](https://clinicaltrials.gov/study/NCT07018492)) completed in late 2025. Results will define how much reduced clearance capacity changes exposure. * **Transporter interaction:** [NCT07011992](https://clinicaltrials.gov/study/NCT07011992) tested the effect of repeated BPN14770 dosing on rosuvastatin in 12 healthy adults. Results are unposted; a positive finding would establish a real statin interaction rather than a theoretical one. * **Biomarker work that could strengthen the case:** Resting peak alpha frequency separated drug from placebo in a machine-learning reanalysis of the phase 2 electroencephalography data ([Norris et al., 2025](https://pubmed.ncbi.nlm.nih.gov/40877251/)). A drug-responsive physiological marker would make future trials harder to dismiss. * **Evidence that could weaken the case:** Comparative work found that inhibiting PDE4B protects rodent memory as effectively as inhibiting PDE4D ([Zhao et al., 2024](https://pubmed.ncbi.nlm.nih.gov/38056709/)). If subtype selectivity is not what delivers the effect, the design premise behind BPN14770 loses its rationale. * **The unresolved endpoint question:** Shionogi and FRAXA both argue the crystallized cognition composite, not the drug, failed ([FRAXA Research Foundation, 2026](https://www.fraxa.org/shionogi-update-on-zatolmilast-fragile-x-clinical-trial-findings/)). Both have a direct interest in that reading, and it stays untested until a trial powered on a functional endpoint is run. * **Regulatory next step:** Shionogi has said it will complete exploratory analyses and continue discussions with regulators before deciding the program's future ([Maia, 2026](https://fragilexnewstoday.com/news/fragile-x-drug-zatolmilast-misses-main-goal-phase-3-trials/)). No further trial has been registered. ## Conclusion BPN14770 is an experimental oral drug that partially blocks one narrow form of an enzyme brain cells use to clear a short-lived signalling molecule tied to learning and memory. Nearly all of the evidence about it comes from the company that owns it and from a patient foundation that helped pay for and champion the studies, so the published picture is not an independent one, and both parties have argued publicly that the disappointing results reflect the measuring instrument rather than the drug. The clearest human finding is that caregivers of adult men with an inherited intellectual disability rated language and everyday functioning as better on the drug than on placebo, and that has now happened twice. Against it, the direct tests of thinking and vocabulary that the studies were actually designed around improved in one small early study and then failed to improve in two much larger ones. In people with early Alzheimer's disease, memory did not improve. In healthy older volunteers, the only reported gain came at low doses, while a higher dose went with worse performance. Side effects so far are mild and mostly digestive, with nausea appearing in a small minority on the drug and in nobody on placebo. Almost nothing is known about taking it for years, and nothing at all about taking it while healthy. It is not approved anywhere, and what circulates outside trials is laboratory chemical rather than medicine. **[Top](#top) - [Benefits](#expected-benefits) - [Risks](#potential-risks--side-effects) - [Protocol](#therapeutic-protocol)**