The Honest Dose #8

I posted something on a peptide research community a few days ago that I'd been chewing on since I finished scoring my fortieth compound. The short version: I was wrong about something I didn't even realize I believed.

When I started this, I assumed FDA approval and research quality were basically the same signal. Approved meant well-studied. Unapproved meant thin evidence. The grey market was where the unproven stuff lived. Clean, intuitive, and after grading 40 compounds across seven research-quality dimensions, I don't think it holds at all. Approval and evidence quality are two separate axes, and they come apart in both directions.

The clearest way I can put it: an approval is tied to one specific indication and one sponsor who paid to run the trials. Tesamorelin, which I covered last issue, is FDA approved, but for HIV-associated lipodystrophy, not for the general growth hormone goals most people running GHRH peptides actually care about. The stamp is real. It just isn't validating the use case it gets borrowed for. GLP-1s are the opposite: approved and genuinely strong evidence, both axes lined up. And then collagen peptides have a real body of human trial data behind them, mostly on skin and joint endpoints, with no approved-drug status at all because they sit as a supplement. Same disconnect, flipped the other way.

I'm not saying approval is meaningless. It clears a safety and manufacturing bar that an unapproved compound never has to clear, and that's worth real weight. What I'm noticing is that "approved" mostly answers "did a company find it commercially worth running trials for one specific use," which is a different question from "how good is the research for the thing I'd actually want to do with it." I'd been quietly collapsing those into a single signal.

Which brings me to this issue's compound, because DSIP is an almost perfect illustration of the gap. It's not FDA approved for anything. And yet it has human trial data going back to the 1970s that some approved products would envy. The approval axis and the evidence axis, once again, refusing to line up.

Let's get into it.

DSIP, delta sleep-inducing peptide, was first isolated in 1977 by Swiss researchers (Schoenenberger-Monnier group in Basel), from the cerebral blood of rabbits who had been electrically stimulated into a sleep-like state. The name is almost comically on the nose: they were hunting for a substance that produces deep, delta-wave sleep, they found a candidate, and they named it after the thing they were looking for.

Nearly fifty years later, it remains one of the most intriguing and genuinely mysterious compounds in the catalog. It is not FDA approved. It sits in the broader peptide compounding conversation and is sold online as a research chemical, used in sleep and recovery circles. But unlike most compounds in that bucket, DSIP has an actual base of human clinical data, and that data is both more interesting and more limited than the marketing around it suggests.

Here's what makes DSIP scientifically strange in a way almost nothing else in the catalog is: after five decades, we still don't fully understand what it is or how it works.

The gene that would produce DSIP in the human body has never been identified. We don't know with certainty where it's synthesized. In the bloodstream it has a half-life of roughly 15 minutes before a specific enzyme breaks it down, which has led researchers to suspect it normally travels bound to a carrier protein or exists as part of a larger precursor molecule that hasn't been found. For a substance discovered in 1977 and studied continuously since, that's a remarkable number of open questions about its basic biology.

Its proposed mechanisms are sprawling and not cleanly mapped. It may act on NMDA receptors. It appears to influence slow-wave sleep, though even that, its defining and namesake property, has produced conflicting results across studies. It's been linked to stress hormone modulation, thermoregulation, growth hormone release, antioxidant activity in mitochondria, and opioid receptor interactions. When a compound is reported to do this many unrelated things, it usually means one of two things: it's a genuinely fundamental regulatory molecule, or the research base is noisy enough that effects haven't been cleanly separated. With DSIP, it's plausibly some of both.

DSIP has something most research-chemical peptides don't: actual human sleep trials, including double-blind, placebo-controlled work. The catch is that most of it is old, small, and used intravenous administration, which complicates translating it to how people use the compound today.

The foundational human studies are genuinely interesting. A double-blind crossover study in healthy volunteers found that an IV dose produced an immediate feeling of sleep pressure, with total sleep time increasing meaningfully versus placebo, plus carryover effects on the following night's sleep efficiency. A separate double-blind study in chronic insomniacs found higher sleep efficiency and shorter sleep latency on DSIP compared to placebo. Notably, across these studies, the researchers were careful to describe DSIP as a sleep-promoting substance rather than a sedative. It didn't knock people out the way a sleeping pill does. The behavioral and EEG analyses showed no classic sedation.

That distinction, which I'll come back to in Claim vs Reality, is the single most important thing to understand about this compound, and the thing most marketing gets wrong.

The limitations are substantial and worth being direct about. The strongest human studies are from the late 1970s and early 1980s. Sample sizes were tiny, often well under twenty subjects. The administration was intravenous in controlled settings, not subcutaneous self-injection. And the results, even on slow-wave sleep, were inconsistent enough across the literature that DSIP's status as "the" endogenous sleep factor was never established. More recent work has largely been in animal models, exploring stroke recovery and addiction applications rather than advancing the human sleep evidence.

DSIP is a fascinating compound with a real but dated and limited human evidence base. It's not snake oil; there are genuine double-blind trials showing effects on sleep efficiency in disturbed sleepers. But it's also nowhere near the level of evidence you'd want before treating it as a reliable sleep intervention, and the gap between the controlled IV studies of 1980 and the subcutaneous vials sold today is real. It belongs to a small group of compounds where the science is genuinely interesting, genuinely incomplete, and unlikely to be resolved soon because the research momentum has moved elsewhere.

RQS: 37/100 - Weak Evidence

Read the full DSIP profile

The claim: Melatonin, and the broader category of sleep supplements built around it, work like gentle sedatives. Take it, get drowsy, fall asleep. It's marketed as a natural sleeping pill, and an enormous wellness category, magnesium, melatonin gummies, and stacks of the two, rests on that mental model.

The reality: Melatonin is not a sedative. It's a timing signal. And understanding that difference explains both why it works for some problems and why it disappoints for others.

Melatonin is the hormone your body uses to tell itself it's nighttime. It regulates the timing of your sleep-wake cycle rather than directly inducing sleep. This is why it's genuinely useful for problems of timing, jet lag, shift work, delayed sleep phase, where you're trying to shift when your body thinks bedtime is. It's also why it often underdelivers for classic insomnia, where the problem isn't that your clock is misaligned but that you simply can't stay asleep. Using a timing signal to solve a sleep-maintenance problem is using the wrong tool, and a lot of disappointed melatonin users are quietly discovering that mismatch.

The dosing situation makes it worse. Most melatonin products are dramatically overdosed relative to what the evidence supports. The studies suggesting benefit often use low doses, around 0.5 to 1 mg, while retail gummies frequently contain 5 or 10 mg. More is not better here; higher doses can actually blunt the effect and produce grogginess. The American Academy of Sleep Medicine recommends against routine melatonin use for chronic insomnia, which surprises most people given how the category is marketed.

Magnesium is the other pillar of the sleep-supplement world, and the evidence there is genuinely split. A 2025 trial found magnesium bisglycinate improved sleep quality; a 2026 study in the European Journal of Nutrition found no notable effect. The most honest read is that magnesium probably helps people who are actually deficient, and roughly half of American adults fall short of recommended intake, but does little for those with normal levels. It's a correction of a deficiency, not a universal sleep aid.

The honest version of the claim: The most popular sleep supplements aren't sedatives, and treating them like one sets you up to be disappointed. Melatonin is a clock-setter that shines for timing problems and underperforms for staying asleep. Magnesium helps mainly if you're deficient. Neither is the natural Ambien the category markets itself as. And here's the thread back to this issue's lead: DSIP, the actual sleep peptide, was described by its own researchers in exactly these terms, a sleep-promoting modulator, not a sedative, with its clearest effects in people whose sleep was already disturbed. The pattern keeps repeating. The compounds that help sleep mostly do it by nudging a system back toward normal, not by chemically flattening you. The "knock you out" model is the wrong model almost across the board.

DSIP at 37 is a useful score to sit with, because the number alone undersells what's actually unusual about this compound: its evidence is simultaneously better and worse than its band would suggest, in different dimensions.

On Study Design, DSIP scores better than most research-chemical peptides that land near it. It actually has double-blind, placebo-controlled human trials, the real thing, not just animal models and open-label observation. That alone separates it from a lot of the compounds scoring in the 20s and low 30s on thin or anecdotal data. If the framework only looked at whether controlled human trials exist, DSIP would score meaningfully higher.

But the RQS doesn't stop there, and this is what keeps it in the Weak band. The Replication dimension is hurt by the fact that the strongest findings are decades old and weren't consistently reproduced, even on the compound's signature slow-wave sleep effect. The Sample Size dimension takes a hit because those trials enrolled tiny cohorts, often under twenty people. And there's an implicit recency problem: the human evidence essentially stopped advancing, with research attention shifting to animal stroke and addiction models rather than larger, modern human sleep trials.

So you get a compound that has genuinely better study design than its score-neighbors but can't climb out of the Weak band because the design quality isn't backed by the replication, sample size, and recency to support it. That's not a flaw in the scoring, it's the score doing exactly its job, refusing to let one strong dimension paper over several weak ones. A compound can have real double-blind data and still not have enough of it, recently enough, replicated well enough, to earn confidence. 37 captures that tension better than any single label like "proven" or "unproven" could.

FDA's PCAC briefing docs for the July 23-24 meeting just dropped, and all 7 peptides under review (BPC-157, KPV, TB-500, MOTS-c, Emideltide/DSIP, Epitalon, Semax, plus their acetate forms) got a "do not add to the 503A Bulks List" recommendation. This made the rounds online fast, mostly framed as a peptide ban. It isn't one.

What's actually in the documents is more interesting than the headline: across all 7, the rejection logic is nearly identical, inconsistent naming with no standardized chemical identity, no real impurity or endotoxin testing, no demonstrated history of compounding use, and zero published human studies for the proposed indications. This isn't "we found these are dangerous." It's "nobody has done the work to show whether they are or aren't."

One detail worth flagging on its own: Epitalon's FDA writeup raises a mechanistic cancer risk angle that we highlighted in Issue #6. It activates telomerase and lengthens telomeres, and longer telomeres are independently linked to elevated cancer risk in the literature. The animal studies cited weren't long enough to actually test for this.

PCAC recommendations are advisory only. FDA still has to go through formal rulemaking before anything is final, so this isn't a done deal yet.

New magnesium-for-sleep data lands on the null side. A 2026 study in the European Journal of Nutrition found that magnesium supplementation did not produce a notable effect on sleep quality, adding to a genuinely split literature. The most defensible interpretation now is that magnesium helps people who are deficient, and about half of American adults fall short of recommended intake, but offers little for those with normal levels. It's worth knowing which group you're in before spending money on it. This is the kind of "it depends on your baseline" finding that rarely survives the trip to a supplement label.

Glycine quietly has some of the better small-trial sleep evidence. Often overlooked next to melatonin and magnesium, glycine, an amino acid that lowers core body temperature, improved subjective sleep quality at around 3 grams before bed in small Japanese trials. The evidence is genuinely promising but limited to well under 100 total participants, so it sits in the same "interesting, underpowered" zone as a lot of what we cover. I mention it because it's a good example of evidence quality and marketing budget being completely uncorrelated: one of the better-supported options gets a fraction of the attention.

Staying in the neuropeptide world that DSIP lives in, Selank is a natural companion compound and a useful contrast, because it shares DSIP's international research lineage and its anxiolytic overlap while sitting in a slightly different lane.

Selank is a synthetic heptapeptide developed in Russia at the Institute of Molecular Genetics, derived from a fragment of the natural immunomodulatory peptide tuftsin. It's primarily studied as an anxiolytic (a compound for reducing anxiety), with secondary interest in cognition, stress resilience, and immune modulation. Where DSIP's headline is sleep, Selank's is calm without sedation, and the two get discussed together in stress-and-recovery contexts.

Its evidence profile will feel familiar if you read Issue #3 on Semax, its close cousin from the same research program. There's a real body of Russian clinical literature suggesting anxiolytic effects without the dependency profile of benzodiazepines, but it carries the same Russian Literature Note caveat: lower journal impact factor scores that reflect the publishing ecosystem, and limited independent replication outside its country of origin. The mechanism, modulation of GABA and the balance of certain neurotransmitters and brain-derived signaling, is plausible and partially supported, but the Western RCT base is thin.

It's a compound where the honest read is "promising and genuinely studied, but mostly in a research tradition that's hard to verify through standard frameworks," which is a sentence I keep finding myself writing about this whole category.

RQS: 36/100 - Weak Evidence

Read the full Selank profile

Hopefully that was helpful/interesting, see you next week.

-Emeka

You're receiving this because you signed up at peptideclear.com. No sponsors. No affiliate links. Just the research, translated.