Reference table
How many human clinical trials exist for each peptide in this index?
The answer
Far fewer than the counts suggest: four molecules return no trial-type record at all, and on several of the rest the records that are returned measure the body's own peptide rather than the compound anyone is selling.
A trial count is the fastest way to check a claim about a peptide, and it is also the fastest way to publish a false one. Search a name too narrowly and a real literature disappears; search it too broadly and an endogenous molecule's biology is counted as evidence for an injection. Both mistakes were made in the building of this table and both are recorded below rather than quietly corrected.
Every number here comes from a single dated run against the National Library of Medicine's E-utilities interface, through the publication-type filters named in the denominator note. Where a molecule's trial-type search returned a small enough set, every record in it was opened and judged; the final column reports what those records actually are, and on more than one molecule the answer is that they are not studies of the compound at all.
Denominator
Every row was measured on 2026-08-31 against PubMed, using the union of the name variants listed for that molecule. The trial columns count records under "clinical trial"[Publication Type] and "randomized controlled trial"[Publication Type]. Those filters were proved live in the same run: the control term "aspirin" returned 8,904 and 6,113 records under them. The invalid tag clinicaltrial[pt], which returns zero for every term in the database and has produced a false census in this repository before, returned 0 for that same control term. A deliberately invalid record identifier, 99999999, errored in the same run. Counts move as records are indexed; the date above is the date these were true.
Rows are read from CENSUS_ROWS[].variants, CENSUS_ROWS[].total, CENSUS_ROWS[].clinicalTrials, CENSUS_ROWS[].randomizedTrials, CENSUS_ROWS[].note — the committed output of scripts/trial-census.py, one dated PubMed run whose controls are stated in the denominator note above. No count on this page is typed by hand, and the script refuses to write anything at all if its controls misbehave.
| Molecule | Indexed records | Clinical-trial type | Randomized type | What those records are |
|---|---|---|---|---|
| Dihexa | 18 | 0 | 0 | No record of either trial publication type. The indexed set is rodent cognition work and reviews of angiotensin IV analogues; the zero is the finding. |
| AOD-9604 | 27 | 0 | 0 | No record of either trial publication type. All 27 indexed records were opened rather than counted: rodent and cell work, doping-detection analytical chemistry, obesity-pipeline review articles, and three entries in the serial feature "Gateways to clinical trials", which lists development programs rather than reporting them. No primary report of a human trial of this compound is indexed. |
| KPV | 113 | 0 | 0 | No record of either trial publication type; the zero is the finding and it is unchanged from the earlier run. The Humans-tagged records are receptor pharmacology, human cell lines, delivery vehicles and one ex vivo skin study; the heading marks human tissue, not a person given the compound. The total moved: this index's KPV monograph records 97 records for this identical string on 2026-08-31 and the same string returns more on re-measurement the same day. The discrepancy is recorded rather than reconciled by editing either number. |
| BPC-157 | 230 | 0 | 0 | The union returns no record of either trial publication type, so there is no hit to read; the zero is the finding. |
| TB-500 | 31 | 1 | 0 | The single clinical-trial hit was opened and is a logistic-regression study of tuberculosis incidence, matched on the letters TB and unrelated to the peptide. Full-length thymosin beta-4 is a different molecule, counted separately on the monograph, and is not this row. |
| CJC-1295 | 35 | 2 | 1 | Both trial-type records were opened. Both are 2006 reports from the same development program, measuring growth hormone and IGF-1 response to the analogue in healthy adults; neither reports a clinical outcome. |
| Ipamorelin | 54 | 2 | 2 | Both clinical-trial records were opened: the 2014 postoperative ileus study and the 1999 pharmacokinetic-pharmacodynamic study in healthy volunteers. |
| Selank | 89 | 3 | 2 | All three trial-type records were opened. All three are anxiety-disorder studies published in Russian in a single journal between 2008 and 2015; there is no trial of this compound in the English-language literature. |
| GHK-Cu | 197 | 3 | 3 | All three trial-type records were opened. One is conducted in rabbits. The two human records apply the copper tripeptide topically, to a venous stasis ulcer and to laser-resurfaced skin; neither administers the injectable form this index tracks. |
| NAD+ | 22 | 4 | 1 | Scoped to the infused route actually sold, not to the coenzyme. All clinical-trial records were opened; three concern an esophageal cancer chemotherapy sequence, a topical psoriasis preparation, and a 1972 review of nicotinic acid in schizophrenia. The precursor literatures are counted separately on the monograph and are not this row. |
| Semax | 216 | 4 | 1 | All four trial-type records were opened. All four are published in Russian in that country's journals between 1997 and 2018, in ischemic stroke, optic nerve disease, and motor neuron disease; there is no trial of this compound in the English-language literature. |
| Melanotan II | 360 | 4 | 1 | All four trial-type records were opened. Every one is a study of erectile function or sexual desire, published between 1996 and 2000, and the set includes the 1996 pilot phase-I report. None studies the tanning use the compound is sold for. |
| MOTS-c | 254 | 5 | 4 | All five trial-type records were opened and not one administers the peptide. Every one measures a participant's own circulating MOTS-c after exercise, heat stress, metformin, or in a disease cohort. The count is a count of endogenous physiology, not of trials of the injected compound. |
| Glutathione | 206 | 7 | 4 | Scoped to the cosmetic indication actually sold, not to the tripeptide, whose whole-molecule literature runs to six figures. All clinical-trial records were opened; six test an oral, buccal, or topical preparation and the seventh an injectable four-ingredient complex. |
| Retatrutide | 178 | 8 | 7 | All eight trial-type records were opened and all belong to the sponsor's development program, from the phase 1b dose-ascending study to a phase 3 trial. A substantial trial record and no approval are not in tension; they are what an investigational drug looks like. |
| Epitalon | 215 | 9 | 6 | All nine trial-type records were opened, and only one names the synthetic tetrapeptide in its title. The others study epithalamin or an unnamed pineal-gland peptide preparation, which is a glandular extract rather than this molecule. The set spans 1979 to 2011, comes almost entirely from one country's literature, and several records are published in Russian. |
| Kisspeptin-10 | 324 | 9 | 5 | All nine trial-type records were opened, and three do not belong to this molecule in people: one studies kisspeptin-54, one rhesus monkeys, and one female goats. The remaining six are human endocrine-probe studies that measure luteinizing hormone response; none reports a clinical outcome. |
| PEG-MGF | 236 | 12 | 11 | All twelve trial-type records were opened and not one administers the peptide. Every one measures endogenous IGF-1Ec or mechano-growth-factor expression in human muscle after resistance exercise, electrical stimulation, ultrasound, growth hormone, or testosterone. The count is a count of exercise physiology, not of trials of the injected compound. |
| PT-141 | 121 | 16 | 14 | All sixteen trial-type records were opened. They are the development program itself, from the 2003 and 2004 intranasal and subcutaneous dose studies to the two phase 3 trials that supported approval; one 2008 record is marked retracted in its own title. |
| DSIP | 530 | 19 | 6 | All nineteen trial-type records were opened, and three are not studies of this peptide at all: a glucocorticoid-replacement sleep study, a review of the insomnia concept, and a bed-rest hormone survey. Of the sixteen that are, fourteen were published between 1981 and 1998; the remaining two are a 2009 anesthesia-adjunct study and a 2014 Russian report of a DSIP-based preparation in diabetic retinopathy. |
| Tesamorelin | 127 | 28 | 26 | individual records not read; the count is reported as a magnitude |
| Sermorelin | 562 | 71 | 39 | individual records not read; the count is reported as a magnitude |
| LL-37 | 3,076 | 72 | 45 | All 72 trial-type records were retrieved and read individually, out of band because the count exceeds this script's itemization ceiling. Sixty-eight of them administer nothing: they measure a participant's own cathelicidin concentration after vitamin D supplementation, exercise, or a probiotic, or as a biomarker in tuberculosis, cystic fibrosis, chronic obstructive pulmonary disease, sepsis, and periprosthetic joint infection. Four administer the peptide, all by a topical or oral route: randomized placebo-controlled trials in hard-to-heal venous leg ulcers in 2014 and 2021, a randomized trial of an LL-37 cream in diabetic foot ulcer in 2023, and a randomized trial of an oral preparation against a SARS-CoV-2 variant in 2023. Reported as a trial count for the marketed compound, 72 would have overstated the record eighteenfold. |
| Thymosin alpha-1 | 870 | 95 | 65 | individual records not read; the count is reported as a magnitude |
| Tirzepatide | 2,481 | 144 | 133 | individual records not read; the count is reported as a magnitude |
| Semaglutide | 5,670 | 340 | 309 | individual records not read; the count is reported as a magnitude |
| Liraglutide | 5,901 | 632 | 575 | individual records not read; the count is reported as a magnitude |
Two readings of this table are wrong and both are common. The first is that a large count means a molecule works: the counts here belong to publication types, not to conclusions, and the largest belong to approved products whose labels are narrower than their literatures. The second is that a zero means nothing has been studied: a zero here means no record carries a trial publication type under the searched names on the date shown, which is a statement about the published record and not about the world.
What the table does establish is which claims can be checked and which cannot. Where a molecule shows no trial-type record, any assertion about its effect in people rests on preclinical work, on a different molecule, or on nothing; the research-only page covers what that means for supply, and the molecule index carries each row's dated regulatory posture beside its evidence.
Other tables
Evidence reviews
Monographs in this table