DSIP (Delta Sleep-Inducing Peptide)

Identity and nomenclature

Delta sleep-inducing peptide carries a second registered name. The FDA Global Substance Registration System files it under UNII YN28Z5YZ73, where the primary name is emideltide, the International Nonproprietary Name entry is INN 7202, and the registered subunit sequence is WAGGDASGE. PubChem holds a single record for it: CID 68816, CAS 62568-57-4, molecular formula C35H48N10O15, molecular weight 848.8 g/mol, InChIKey ZRZROXNBKJAOKB-GFVHOAGBSA-N. ChEMBL indexes it as CHEMBL2104403. Both registries agree on the nine-residue chain Trp-Ala-Gly-Gly-Asp-Ala-Ser-Gly-Glu. The naming aside, there is no ambiguity about which molecule the name denotes.

A second CAS registry number, 69431-45-4, sits on the same GSRS record flagged as having no structure given, and it circulates on catalogue documentation alongside the primary number. Both numbers appear on that one registry entry. The FDA evaluation published in May 2026 went further and judged the substance not well characterised, citing naming conventions across the literature that do not follow USAN, INN or IUPAC standards, and the absence from the published record of specific tests for impurities, aggregates and endotoxins.

The May 2026 evaluation also recorded that the reported water solubility is limited enough that it is unclear how an injectable preparation at 1 mg/mL could be formulated without a co-solvent, and raised aggregation and immunogenicity as open questions for a nine-residue peptide delivered by injection. No published study addresses either question, and the evaluation records both as gaps in the publicly available scientific literature. The two nominations proposed a subcutaneous route, for which the same document identified no pharmacokinetic or toxicokinetic study at all.

What the 1977 isolation established

Between 1963 and 1970 Monnier's group in Basel kept donor rabbits asleep by electrical stimulation of the intralaminar thalamus, dialysed blood drawn from the confluence of the sinuses, and infused the dialysate into the cerebral ventricles of recipient animals, which showed increased delta electroencephalogram activity. Isolation and sequencing of the responsible fraction followed. Schoenenberger and Monnier published the nonapeptide sequence in 1977, in a paper that also carried the specificity test for the synthetic material (PMID 265572).

That paper tested specificity carefully. Nine synthetic peptides were infused intraventricularly under double-blind conditions across 58 rabbits including controls: the nonapeptide, five candidate metabolic fragments, two analogues with two residues exchanged, and a related tripeptide. Only the full nonapeptide produced significant enhancement of delta and spindle patterns. The companion report in Pflugers Archiv, using 61 rabbits at 6 nmol/kg, put the mean increase in delta activity at 35 per cent in neocortex and limbic cortex against controls, and recorded that only the pure alpha-aspartyl peptide was highly active, in contrast to its beta-Asp isomer; that report does not state that the beta-Asp isomer was wholly inactive (PMID 568769).

Two features of the founding work matter downstream: the route was infusion into the brain ventricles, and the species was rabbit. Schoenenberger's 1984 review of the whole programme recorded that intracerebroventricular, intravenous and subcutaneous administration each yielded a parabolic rather than monotonic dose-response curve (PMID 6548966). That review was written by the originating investigator, and several figures that circulate as measurements of this peptide trace back to its summary paragraphs, a point taken up in the unsourced claims recorded below.

Where the animal record stopped agreeing with itself

Tobler and Borbely tested the same effect in rats. Systemic doses of 40 to 160 nmol/kg intraperitoneally did not significantly reduce motor activity over 24 hours, and neither injection nor infusion of 7 to 24 nmol into the lateral or third ventricle significantly increased sleep or delta-band power in the two hours afterwards. Administration of 7 nmol into the third ventricle reduced delta-band power. Their stated conclusion was that the peptide does not qualify as a specific sleep-promoting substance (PMID 7405185).

Reports from other groups did not line up with each other either. The 2026 FDA review assembled the disagreement: intraventricular treatment of cats at 7 nmol/kg selectively prolonged non-REM sleep in one report, while intravenous treatment at 30 nmol/kg prolonged both non-REM and REM sleep in another; in rats, 7 nmol/kg intraventricularly prolonged non-REM sleep only in one study (Ursin and Larsen 1983, PMID 6688864) while a slightly lower dose, recorded by FDA as 6 nmol/kg, prolonged both stages in a second (Young and Key 1984, PMID 6549349). The Young and Key report itself describes four intracerebroventricular doses of 5 micrograms per animal, so the per-kilogram figure and the primary description are both set down here. The review noted that the reasons underlying the discrepant findings have not been discussed or elucidated.

Iyer and colleagues deprived male rats of sleep for four hours on a rotating wheel, which raised both slow-wave sleep and plasma growth hormone afterwards, and reported that microinjection of a specific antiserum into the third ventricle blocked both increases where normal rabbit serum did not (PMID 3368469). Naloxone blocked the sleep-prolonging effect in a separate rat study, in which four intracerebroventricular doses of 5 micrograms increased both REM and non-REM sleep and pre-treatment with the opiate antagonist abolished that increase (PMID 6549349). In-vitro work found no displacement of diprenorphine binding at any opioid receptor subtype, pointing at indirect action through calcium-dependent Met-enkephalin release (PMID 2706459).

The human sleep trials, and what happened outside the originating group

First administration to people was six normal volunteers. Schneider-Helmert and colleagues infused 25 nmol/kg intravenously in the morning under a double-blind crossover design and reported median total sleep time in the following 130 minutes 59 per cent higher than placebo, with shorter sleep onset and better sleep efficiency the following night (PMID 6895513). A series of studies in insomniac patients came out of the same collaboration, among them 18 chronic psychophysiological insomniacs given six doses across a week (PMID 3792404) and 14 middle-aged insomniacs treated on seven successive nights under placebo-controlled double-blind conditions (PMID 3622582), both reporting normalisation of sleep.

Independent groups ran the design again. Monti and colleagues in Montevideo gave 25 nmol/kg intravenously over four nights to six patients with severe chronic insomnia in a double-blind crossover. Total sleep time and non-REM sleep rose, the increase was carried by stage 2, and stage 1, slow-wave sleep and REM sleep were not modified. Where differences against placebo reached significance, the same differences were already present at baseline. The authors concluded that the sleep improvement was of little clinical significance (PMID 3583493).

Bes and colleagues in Amsterdam studied 16 chronic insomniacs in a matched-pairs parallel-groups design, half receiving 25 nmol/kg intravenously before each of three nights. Sleep efficiency and sleep latency favoured the peptide, but the authors judged the significant effects weak and in part attributable to an incidental change in the placebo group, and recorded no change in subjective sleep quality (PMID 1299794). The most recent human study points in the other direction. In a randomised study of 24 female ASA I-II surgical patients, twelve received saline and twelve received a single intravenous bolus of 25, 50 or 100 nmol/kg. At 25 nmol/kg during isoflurane anaesthesia the treated group showed significantly reduced delta rhythm, reduced burst suppression and increased bispectral index, with heart rate increased and heart-rate variability decreased, which the authors read as a lightening of anaesthetic depth (PMID 19142086).

Endocrine claims, tested directly in people

Rodent experiments from Iyer and McCann are the origin of most endocrine claims made for this molecule. Injection of 5 micrograms into the third ventricle of long-term ovariectomised Sprague-Dawley rats raised luteinising hormone within 30 minutes and left follicle-stimulating hormone unchanged, with no response from dispersed pituitary cells in vitro, which placed the action above the pituitary (PMID 3121137). A companion paper reported dose-related growth hormone elevation after third-ventricle injection, blocked by the dopamine antagonist pimozide (PMID 3575154). A third report measured somatostatin release from rat median eminence fragments in vitro and found dose-dependent inhibition, maximal at 10^-8 M and blocked by pimozide at 10^-6 M (PMID 2886936). All three used rodent tissue or rodent intraventricular delivery.

Human tests of the same claims were negative. Giusti and colleagues studied eight healthy women: five received 25 micrograms per kilogram intravenously over 30 minutes, alone and against an arginine challenge, and three received the same dose infused from 21:30 to 22:30 for the circadian arm. Neither growth hormone nor prolactin responded in any condition (PMID 8475226). Spath-Schwalbe and colleagues infused total doses of 3 and 4 milligrams into healthy young men and found ACTH and cortisol responses to corticotropin-releasing hormone almost identical to placebo, with the meal-related midday surge also unaffected (PMID 7777652). Friedman and colleagues, sampling plasma every 30 minutes in Cushing syndrome patients and controls, found the correlation between delta sleep and morning peptide immunoreactivity negative rather than positive, and said in print that this was against the notion of a causal relationship (PMID 7700506).

No gene, no receptor, and what DSIP-LI measures

Kovalzon and Strekalova reviewed the field in 2006 and stated that the link to sleep had never been further characterised, in part because the gene, the protein and any related receptor had not been isolated, and described the sleep-factor hypothesis as extremely poorly documented and still weak (PMID 16539679). No precursor structure has been published since. A molecule whose name asserts a mechanism has, five decades after its isolation, no identified receptor and no identified transcript.

Graf and colleagues detected DSIP-like immunoreactivity in the breast milk of two women, at about 30 ng/mL in colostrum falling to about 10 ng/mL in mature milk, and gel chromatography showed that most of the immunoreactive material occurred in a form larger than the nonapeptide (PMID 6547144). The nonapeptide itself was demonstrated by high pressure liquid chromatography, alongside additional antibody-reactive peptides in the same sample. Radioimmunoassay of this kind measures antibody binding, and the paper reports the two measurements separately.

Kovalzon and Strekalova drew the inference that follows and proposed that a distinct, unidentified peptide accounts for a share of both the immunoreactivity and the reported biological activity. They also noted that several synthetic structural analogues promoted slow-wave sleep in rabbits and rats where the parent nonapeptide did not. Published figures for tissue and fluid concentrations of this peptide derive from radioimmunoassay of DSIP-like immunoreactivity; no structure-specific quantification of the nonapeptide in a human body fluid has been published.

Regulatory position as of August 2026

The compound is not an approved medicine in the United States and carries no marketing authorisation there for any indication. A ClinicalTrials.gov search across delta sleep-inducing peptide, DSIP and emideltide returned zero registered studies in August 2026. The clinical record consists of small investigator-run studies, most of them from the 1980s, none of them registered. The most recent human study located in this survey was published in 2009 and examined the peptide as an adjunct during isoflurane anaesthesia.

In 2026 the molecule received a formal regulatory read for the first time. Two nominations were filed to have emideltide-related bulk drug substances added to the list of substances that may be used in compounding under section 503A, and FDA published an 83-page evaluation dated 11 May 2026. Its findings were that the substances are not well characterised physically and chemically; that clinical studies for chronic insomnia appear inconclusive and at best preliminary, limited by poor control selection and small samples; that no data at all support the nominated subcutaneous route; and that a search of the FDA Adverse Event Reporting System through 3 March 2024 retrieved zero reports. Across the studies the agency could identify, intravenous doses of 25 to 150 nmol/kg had been given to 209 human subjects for one to fifteen days.

The Pharmacy Compounding Advisory Committee voted on 24 July 2026. Legal and trade reporting of that meeting records emideltide as the only one of seven peptides reviewed that the committee did not recommend, on a tally of six in favour to seven against with one abstention; FDA has not posted minutes, so that tally comes from secondary coverage rather than the agency's own record. FDA's written proposal, which is primary, was not to add either the free base or the acetate. A committee recommendation is advice, and no final determination has issued.

What is not known

No registered clinical trial of this compound exists. A ClinicalTrials.gov search across delta sleep-inducing peptide, DSIP and emideltide returned zero studies in August 2026, so nothing about it has been subject to the registration, protocol publication or results-reporting discipline that applies to registered research. The human record is 209 subjects given intravenous doses of 25 to 150 nmol/kg for one to fifteen days, spread across small studies from the 1980s and 1990s, by FDA's own count. There is no human pharmacokinetic study, no dose-ranging study, no study of chronic administration, and no data of any kind on the subcutaneous route, which is the route both nominations proposed. FDA identified no nonclinical toxicity studies to inform potential clinical use, no developmental or reproductive toxicity work, and no assessment of abuse potential despite the proposed mechanism running through endogenous opioid release. Immunogenicity has not been characterised. The gene, the precursor and the receptor have never been isolated, so no claim about endogenous physiology can be grounded in a molecular pathway. A search of the FDA Adverse Event Reporting System through 3 March 2024 retrieved zero reports. FDA's evaluation states that result without interpreting it, and also records that it identified no relevant case reports in the medical literature.

Questions