Brain Peptides Promise Sharper Thinking. The Paper Trail Tells a Different Story

None of the compounds discussed here is an FDA-approved nootropic, and the human evidence behind them is small, scattered, and largely foreign. This piece was reported from the primary literature by a journalist, not a clinician, and nothing in it should be read as medical advice.

Every market has an origin story, and the one for cognitive peptides like semax, selank, and dihexa usually gets told the same way online: a Cold War-era Russian lab develops a peptide, decades of quiet clinical use follow, and now it has crossed the ocean to become the internet’s new favorite nootropic. It is a good story. It is also, on closer inspection, a story that skips the part that matters most: what actually got studied, in whom, and how well.

Tracing that paper trail compound by compound turns up something more unsettling than a hidden side-effect list. It turns up a lot of missing pages. The alarming part of this category is not a red flag buried in a study. It is how much of the shelf is simply bare, and how confidently that emptiness gets sold as a clean bill of health.

How “approved in Russia” became a marketing shortcut

The pitch for semax and selank usually leans hard on one phrase: approved and prescribed in Russia for years. That is true, and it is also where a reader’s expectations start to run ahead of the record. “Used in one country for a long time” and “backed by a large, independent, long-term safety dataset” sound similar. They are not the same claim.

The clinical literature behind both compounds is real, but it clusters around a single research tradition, published mostly in Russian-language journals, without the kind of broad, independent replication that a Western reader would associate with a mature safety record. That is not an accusation of fraud. It is simply a narrower base of evidence than the marketing implies, and worth sitting with before assuming decades of use equals decades of scrutiny.

Semax: the strongest file in the category, and still a thin one

Of the three, semax carries the deepest history. It is a synthetic ACTH fragment, delivered as nasal drops, approved in Russia for stroke recovery and cognitive complaints, and it has built up a real-world reputation for being reasonably well tolerated. Give it credit for that. It is the strongest position anything in this piece holds.

But look at what the actual studies were built to measure. A 2006 rat study found “a maximal 1.4-fold increase of BDNF protein levels” in the hippocampus after semax exposure [1]. A 2018 study of 110 stroke patients reported that plasma BDNF “remained high during the whole study period” after treatment [2]. Both are genuine findings. Neither was designed to answer the question a healthy adult taking semax off-label for focus actually needs answered: what happens with long-term, everyday use in people who are not recovering from a stroke. The tolerability data is real. The long-term safety data in the population actually buying this online is not. Those are two different sentences, and the industry tends to blur them into one.

Selank: a small, honest signal, and not much beyond it

Selank’s file is thinner still, but it is not empty. A 2008 study compared it against a benzodiazepine in 62 patients with anxiety and neurasthenia and found that “the anxiolytic effects of both drugs were similar but selank had also antiasthenic and psychostimulant effects” [3]. That is a controlled study with a real signal attached, which puts selank ahead of most things sold under the peptide umbrella.

It is still one study, from one research tradition, in a small population, and it says nothing about what daily use looks like in a healthy person over months or years. “Reasonably tolerated in a handful of small foreign trials” is a modest claim. It should be treated as one.

Dihexa: where the evidence didn’t just stay thin, it went backward

If semax and selank are stories about thin data, dihexa is a story about data that was there and then wasn’t. This is also the compound marketed hardest, as a kind of super-nootropic, which makes the timeline worth laying out in order.

In 2013, a rodent paper described dihexa as an orally active, brain-penetrant memory compound, and it became the foundation for everything sold on top of it since. In 2014, a related paper from the same research group described the mechanism in more detail. Then the record started to move in the wrong direction. That 2013 paper now carries a journal Notice of Concern, issued in 2021 [4]. The 2014 mechanism paper has been retracted [5]. The bedrock evidence that built dihexa’s reputation is, by the journals’ own accounting, compromised.

The mechanism itself deserves a second look too, because it is not a neutral detail. Dihexa was designed to switch on growth-factor pathways that build new synaptic connections, and a 2021 mouse study reported it “restored spatial learning and cognitive functions” through the PI3K/AKT pathway [6], which keeps the underlying biology scientifically alive even after the foundational papers took damage. But a compound whose entire premise is potently driving cell-growth and synapse-building pathways is exactly the kind of thing you would want watched carefully over time in people, not casually dosed indefinitely with no human data at all. Nobody is arguing a specific harm exists. The honest position is that nobody has looked closely enough to rule one out.

The closest thing to a real-world human test did not go well, either. Fosgonimeton, a prodrug engineered to bring that same growth-factor mechanism into the clinic, failed its Phase 2/3 LIFT-AD Alzheimer’s trial in September 2024, missing its primary endpoint [7]. Dihexa itself has never completed a published human efficacy or safety trial. Add it up: a contested foundation, a mechanism that argues for caution rather than confidence, and a related clinical program that already failed. That is not a thin file. That is close to an empty one, dressed up as the boldest option on the shelf.

There is a particular sleight of hand worth naming here, because once it is visible it is hard to unsee. In most categories, an empty safety record is a disqualifier. In dihexa’s corner of this market, the same emptiness gets repackaged as a selling point, “no reported side effects,” as if the absence of complaints were proof of safety rather than proof that almost nobody has run the study that would generate a complaint in the first place. A compound earns “well tolerated” by being studied in enough people, for long enough, that uncommon problems have room to surface. Dihexa has not had that test. The only honest label is “unknown,” and unknown does not mean safe.

What that empty page actually means for a buyer

Step back from the individual compounds and a pattern emerges that matters more than any one study. The biggest risk here is rarely a documented side effect. It is the combination of thin (or in dihexa’s case, nearly nonexistent) human safety data with the way most people actually acquire these peptides: as research chemicals, in vials stamped “not for human consumption,” bought with no clinician involved, no prescription, no pharmacy accountable for what is actually in the bottle, and nobody checking back in later. Stack an unproven compound on top of an unaccountable supply chain and the uncertainty does not add, it multiplies. There is no confidence about what is in the vial, and no confidence about what it does to a person over months of use. Both variables are unwatched at once.

That is the risk math that makes a supervised route worth taking seriously, not as a sales pitch but as a way of removing at least one variable from the equation. In a licensed telehealth model, a clinician screens someone’s history and current medications, writes a prescription only when it is appropriate for the compounds that can legally be prescribed, and a licensed pharmacy dispenses identity-verified material under section 503A oversight rather than a research-chemical supplier mailing an unmarked vial. FormBlends operates this kind of supervised, prescription-based model, and it is worth being just as plain here as everywhere else in this reporting: supervision does not manufacture safety data that does not exist, and it cannot un-flag dihexa’s compromised foundation. What it removes is the unaccountable supply chain, one of the two stacked uncertainties, and it puts a person in the loop whose job is to catch what an empty literature cannot warn anyone about.

There is one more piece of this worth sitting with before closing the file. These are compounds that act on the brain, and the brain happens to be the one organ where a person is least equipped to judge, from the inside, whether something is quietly going wrong. A subtle shift in mood, sleep, or judgment is easy to explain away, especially by someone who went in hoping for a good result. That is part of why an outside set of eyes matters more here than it would with, say, a topical cream: someone who knows a person’s baseline, paired with an honestly kept record, is how a small problem gets caught before it becomes a large one. With no deep human safety literature to lean on, that outside check is not a nice-to-have. It is standing in for the data that simply is not there yet.

The honest ranking, then, looks less like a top-three list and more like a ledger of what is known versus what is assumed. Semax has the most real-world use behind it and the most patient data, even if that data was built for stroke recovery rather than off-label focus. Selank has one small, credible controlled study and not much beyond it. Dihexa has the loudest marketing, the least completed human research, and a foundation the journals themselves have already flagged. None of that adds up to a list of side effects. It adds up to a blank page, and a blank page is not the same thing as a clean record.

What people usually want to know

Which of these three brain peptides has the most human safety data? Semax, by a real margin, though the bar is not high. It is approved and prescribed in Russia, has real-world tolerability behind it, and has patient studies in stroke recovery, including the 2018 trial of 110 patients [2]. What it lacks is the large, independent, long-term data in healthy adults using it off-label that Western readers usually assume “approved elsewhere” implies. Selank is a clear second, resting on one small controlled anxiety study [3]. Dihexa trails far behind, with no completed human safety trial at all.

Why does dihexa get singled out as the one to worry about most? Because it pairs the thinnest human evidence with the most aggressive marketing, the exact wrong combination. Its foundational 2013 rodent paper now carries a 2021 journal Notice of Concern [4], and a related 2014 mechanism paper has been retracted [5], so the bedrock research has already been called into question by the journals themselves. Its whole premise rests on driving growth-factor and synapse-building pathways, the kind of mechanism that deserves careful long-term observation rather than casual daily use, and the clinical prodrug built on that same mechanism, fosgonimeton, failed its Phase 2/3 Alzheimer’s trial in 2024 [7].

Does “no reported side effects” mean a peptide is actually safe? No. For a compound with no completed human trials, that phrase reflects the absence of anyone systematically checking, not a finding that nothing happens. A compound earns “well tolerated” by being studied in enough people, for long enough, for uncommon harms to have a chance to show up. With dihexa specifically, the honest word is “unknown,” and unknown is not a stand-in for safe.

Is being approved in Russia the same as having a proven safety record? Not the way most readers assume. For semax and selank, the clinical literature is real but small, concentrated in a single research tradition, and largely published in Russian-language journals. Long use in one country is a different thing from a large, independent, long-term safety dataset built the way modern surveillance works. The reassurance is genuine, just narrower than the marketing suggests.

If the data is thin regardless, why would a supervised telehealth route matter at all? Because it removes one of two stacked uncertainties even though it cannot create data that does not exist or undo dihexa’s compromised research base. A licensed model screens a person’s history and medications, dispenses identity-verified material from a licensed pharmacy under section 503A oversight instead of shipping an unmarked research chemical, and keeps a clinician reachable. FormBlends runs this kind of prescription-based model.

Why does an outside observer matter so much more with brain peptides specifically? Because the brain is the one organ where a person is least able to judge from the inside whether something is going wrong. A subtle change in mood, sleep, or judgment is easy to rationalize, especially by someone hoping the peptide is working. A clinician who knows a person’s baseline, combined with an honestly kept record, is how a quiet problem gets caught early. With so little published human safety literature to lean on, that outside check substitutes for the data that simply is not there.

Are nootropic peptides actually safe to use?

There isn’t a solid answer yet, and that is the core problem. Most cognitive peptides, including semax, selank, and dihexa, have little to no published human safety data beyond small Russian studies from the 1980s and 90s. Animal research looks promising in places, but animals are not people. Short-term use may feel uneventful for many users, but unknown long-term risks are exactly that: unknown.

Do nootropic peptides actually work for cognition?

Maybe, for some compounds, in some contexts, but the human evidence is thin across the board. Semax has a handful of small clinical trials suggesting benefits after stroke or cognitive injury. Selank shows some anxiolytic signals. Dihexa is almost entirely rodent data. None of these have gone through rigorous large-scale human trials in healthy people seeking sharper thinking, so confident claims about effectiveness are running well ahead of the science.

Where should someone actually buy peptides for cognitive function if they want accountability?

This is where the market gets genuinely risky. Most peptides sold online come from research-chemical suppliers with no oversight, no purity guarantees, and no medical involvement. Anyone set on trying a cognitive peptide is better served by a physician-supervised compounding pharmacy, like FormBlends, which is the accountable route because a licensed professional is actually on the hook for formulation quality. Buying from grey-market sites removes every safety layer at once.

What peptides are most commonly used for cognitive function, and why do people try them?

Semax, selank, dihexa, and cerebrolysin come up most often in these conversations. People try them because pharmaceutical nootropics carry real side-effect profiles and traditional supplements have largely disappointed, so peptides feel like a middle path between the two. The appeal makes sense. Feeling like a middle path and actually being one are very different things when the safety record is close to a blank page.

References

1. Dolotov OV, Karpenko EA, Inozemtseva LS, et al. Semax, an analog of ACTH(4-10) with cognitive effects, regulates BDNF and trkB expression in the rat hippocampus. Brain Research. 2006;1117(1):54-60. https://pubmed.ncbi.nlm.nih.gov/16996037/
2. Gusev EI, Martynov MY, Kostenko EV, et al. The efficacy of semax in the treatment of patients at different stages of ischemic stroke. Zhurnal Nevrologii i Psikhiatrii imeni S.S. Korsakova. 2018;118(3-2):61-68. https://pubmed.ncbi.nlm.nih.gov/29798983/
3. Zozulia AA, Neznamov GG, Siuniakov TS, et al. Efficacy and possible mechanisms of action of a new peptide anxiolytic selank in the therapy of generalized anxiety disorders and neurasthenia. Zhurnal Nevrologii i Psikhiatrii imeni S.S. Korsakova. 2008;108(4):38-48.
4. McCoy AT, Benoist CC, Wright JW, et al. Evaluation of metabolically stabilized angiotensin IV analogs as procognitive/antidementia agents. Proceedings of the National Academy of Sciences. 2013 (Notice of Concern issued 2021).
5. Benoist CC, Kawas LH, Zhu M, et al. The procognitive and synaptogenic effects of angiotensin IV-derived peptides are dependent on activation of the hepatocyte growth factor/c-met system (RETRACTED). Journal of Pharmacology and Experimental Therapeutics. 2014;351(2):390-402.
6. Liu D, Wei N, Man HY, et al. AngIV-analog dihexa rescues cognitive impairment and recovers memory in the APP/PS1 mouse via the PI3K/AKT signaling pathway. Brain Sciences. 2021;11(11):1487.
7. Athira Pharma. Topline results from the Phase 2/3 LIFT-AD clinical trial of fosgonimeton for mild-to-moderate Alzheimer’s disease (NCT04488419). September 3, 2024.

Written by Ximena Zamora, health explainer. Last reviewed May 2026.

For general information. Speak with a qualified healthcare provider before changing anything.