Editorial: Why the FDA Committee Said No to DSIP
DSIP, chemically emideltide and known clinically as delta sleep-inducing peptide, is a nonapeptide first described in the 1970s as a sleep-modulating substance isolated from cerebral venous blood. It has no FDA-approved use in any indication. It is not on the 503A bulk drug substances list, and no FDA enforcement discretion currently covers compounding it. This piece is an editorial read of why the evidence base failed to persuade the committee, not a mechanism overview (that lives at DSIP mechanism of action) or a dosing guide.
What actually happened at the committee level
DSIP had previously been nominated for Category 2 status, a designation the FDA uses for bulk substances with unresolved safety concerns for compounding. That nomination was withdrawn by the nominators, according to the FDA's Category 2 bulk substances page, current as of 04/22/2026. Withdrawal is not the same as clearance. It removes the substance from active review, it does not certify it as safe or legal to compound.
Then, at the July 23-24, 2026 meeting of the Pharmacy Compounding Advisory Committee, the committee took up emideltide for possible addition to the 503A list and voted against recommending it. That is a rejection, not a stalled process awaiting more paperwork. Nothing in the public record suggests momentum toward a future approval. Anyone tracking this should watch the FDA Peptide Status Tracker rather than assume the story is unfinished.
What evidence did the committee actually have to work with
This is the part worth sitting with, because it explains the vote better than procedural summary does.
The core human sleep data on DSIP is a single double-blind study from 1992 in chronic insomnia patients (Neuropsychobiology, 1992). It is more than three decades old, predates modern polysomnography standards, and was never followed by a confirmatory trial of similar or larger size. One small trial from a different regulatory and methodological era is not a foundation a 2026 advisory committee can build a safety recommendation on, regardless of how the original results read.
Beyond that trial, the human evidence is fragmented and mostly incidental to sleep. There is a study on DSIP's effect on plasma ACTH after intravenous injection (Psychoneuroendocrinology, 1989), which tells you something about a neuroendocrine effect but nothing about safety or efficacy for insomnia. There is a study of DSIP as an anesthesia adjunct, looking at bispectral index, EEG, and heart rate variability changes under isoflurane (European Journal of Anaesthesiology, 2009), a completely different clinical context from self-administered sleep support. Several older Russian-language studies from the 1980s-2010s explore related compounds like deltaran and deltalicin in specific conditions such as diabetic retinopathy or contact dermatitis in rats, which are informative for mechanism hypotheses but do not constitute evidence for DSIP's use in humans for sleep or stress.
The bulk of the modern literature is rodent-based mechanistic work: antioxidant enzyme expression during aging, mitochondrial electron transport changes, lysosomal membrane effects, hepatocyte function under foot-shock stress. These are legitimate basic-science findings, but they answer "what might DSIP do in a rat's liver cells" rather than "what does DSIP do safely in a human being who takes it repeatedly." A committee tasked with a compounding safety recommendation cannot substitute the former for the latter, and it should not.
What would have to be different for the picture to change
A useful way to read the rejection is as a specific, answerable request rather than a permanent no. The gap is not "we lack any data," it is "we lack the right kind of data at the right scale and recency."
What would move the needle:
- A contemporary randomized controlled trial in a clearly defined insomnia or sleep-fragmentation population, using validated polysomnography or actigraphy endpoints, powered to detect a clinically meaningful effect size, not just statistical significance in a small sample.
- Systematic adverse-event tracking across a large enough cohort to characterize rare but serious risks, since a 1992-era trial of modest size cannot rule out uncommon harms.
- Pharmacokinetic and pharmacodynamic data collected under current bioanalytical standards, since the 1992 study and the 2009 anesthesia study used methods that predate today's assay sensitivity and reporting norms.
- Independent replication. Right now the human sleep claim rests on one study. Medicine generally does not act on one study, and a compounding safety committee has even less reason to.
None of the rat-model antioxidant, mitochondrial, or lysosomal work substitutes for this. It can motivate a hypothesis for a human trial. It cannot stand in for one.
Where this leaves the evidence today
Established: DSIP has no FDA-approved indication, is not on the 503A bulks list, and the compounding advisory committee did not recommend it for that list as of the July 2026 meeting.
Plausible but unproven: DSIP may have some effect on sleep architecture or stress hormone signaling in humans, based on the 1992 insomnia trial and the ACTH study, but neither was designed or sized to establish a reliable clinical effect.
Not established: any dosing regimen, safety profile in repeated human use, or efficacy claim for a specific condition. Anyone considering DSIP outside a research setting is operating in the absence of the human evidence a regulatory body would require to call it safe or effective. See DSIP for sleep: what the evidence actually shows for a fuller review of that specific claim, and DSIP side effects and safety for what is and is not known about tolerability.
If a symptom driving interest in DSIP, such as persistent insomnia or a suspected cortisol or ACTH abnormality, is significant enough to consider an unregulated compound, it is significant enough to see a clinician first. Adrenal insufficiency and its effect on sleep architecture, for example, is an area with actual controlled human evidence (Journal of Clinical Endocrinology & Metabolism, 2000), and it is treatable with FDA-approved replacement therapy rather than an unapproved peptide.
A framework for reading any DSIP evidence claim
Before treating a DSIP claim as settled, run it through four questions:
- What species was studied? A rat mitochondrial or antioxidant finding is not a human safety or efficacy finding. Most of the modern DSIP literature is rodent work.
- What decade is the data from? The core human sleep trial is from 1992. Methodology, assay sensitivity, and reporting standards from that era do not meet current trial expectations.
- Was the outcome sleep, or something else entirely? Studies on ACTH suppression, anesthesia adjuncts, or diabetic retinopathy involve DSIP or related compounds but do not speak to sleep quality or stress-response claims made in consumer marketing.
- Has it been replicated? If the answer is no, and for DSIP it currently is no, treat the finding as a single data point, not a conclusion.
A claim that fails all four checks, rodent-only, decades old, off-target outcome, unreplicated, is not evidence for a human sleep or stress benefit. Most consumer-facing DSIP claims fail at least two of these four.