Two things with one name
The first thing to establish about p21 is which p21 is meant, because the two are unrelated and one of them is thousands of times better studied.
p21 the protein is the product of the CDKN1A gene, also written p21^Waf1/Cip1^. It stops the cell cycle, sits downstream of p53, and is one of the most-studied molecules in cancer biology and in the study of cellular senescence. It is not a drug, not sold, and not taken by anyone.
P021 the compound is a short experimental molecule derived from ciliary neurotrophic factor, published as Ac-DGGL(A)G-NH₂. It is four residues long, with blocking groups at each end that let it be given by mouth. It is studied for memory, neurogenesis and tau pathology, in animals.
The scale of the mismatch:
| Search | PubMed records |
|---|---|
p21 |
61,731 |
CDKN1A |
11,878 |
p21 WAF1 |
3,820 |
| the compound (P021, neurotrophic) | 17 |
Queried 2026-09-20. The search that looks most like a search for this compound is the one that finds it least.
What the compound's record actually holds
Seventeen records, running from about 2014 to 2026, and it is worth being precise about what is in them.
Zero are of clinical-trial publication type. Five carry a human MeSH tag, and those are reviews and discussion pieces rather than administrations.
The primary research is animal and laboratory work, reported findings including:
- Chronic oral treatment in a triple-transgenic Alzheimer's mouse model — twelve months of P021 in the diet, starting at 9–10 months of age, reported to reduce abnormal tau hyperphosphorylation at major neurofibrillary sites, with a smaller effect on amyloid confined to soluble levels, and rescued deficits in cognition, neurogenesis and synaptic plasticity (Neurobiology of Disease, 2014).
- Lowered cerebrospinal-fluid tau in aged rats (Journal of Alzheimer's Disease, 2015).
- Rescued developmental delay and memory deficits in the Ts65Dn mouse model of Down syndrome (Scientific Reports, 2017).
- Prevention of Alzheimer-like behaviour when treatment began prenatally or in early postnatal life (Alzheimer's Research & Therapy, 2020).
- Brain microstructure changes detected by diffusion MRI in the same Alzheimer's model, the most recent work in the set (Magnetic Resonance Imaging, 2026, with a corrigendum the following month).
Almost all of this comes from one research tradition — principally the laboratory of Khalid Iqbal at the New York State Institute for Basic Research in Developmental Disabilities. That is not a criticism of the work; it is a statement about the shape of the evidence. A finding reproduced twelve times inside one group and never outside it is in a different position from the same finding reproduced twice by two groups.
There is one clear test from outside: a 2024 paper in the Journal of Neurodevelopmental Disorders examined the compound in cell and mouse models of CDKL5 deficiency disorder, a condition unrelated to the originating programme's interests. That is the most informative single record in the set, precisely because of who ran it.
The registry trap, which this compound demonstrates perfectly
A search of ClinicalTrials.gov for P021 returns five studies. None of them has anything to do with this compound:
| Record | What it is |
|---|---|
| NCT06622395 | telepsychiatry in autism assessment |
| NCT03649919 | rare neurological disease management (withdrawn) |
| NCT03747107 | pharmacist-led quality improvement in primary care |
| NCT03894709 | a care model for elderly hip-fracture patients |
| NCT07168447 | transfusion and cognition in sickle cell disease |
The string matches their own protocol identifiers — the internal numbers sponsors assign to their studies — not an intervention.
This library says this often enough that it is worth putting on a page: a hit is not a record. A count of "five trials" for this compound would be wrong by five, and it would be wrong in a way that no amount of counting could detect, because the only fix is to open the records and read what is being given to whom.
Where it stands
A small, coherent, entirely preclinical body of work on a defined molecule, with one independent test outside the laboratory that produced it, no registered trial, and no published administration to a human being. The name it carries belongs, in almost all published usage, to something else entirely.
If someone is reading about p21 and what they are reading concerns cancer, senescence or the cell cycle, they are reading about the protein and it has no bearing on the compound. If what they are reading concerns memory in mice, they are reading about a very small literature.
Related entries in this library: noopept is the other nootropic in this library that is not a peptide in the marketed sense; cerebrolysin comes at the same question through a tissue-derived mixture rather than a defined molecule; C-peptide and the natriuretic peptides are the library's other entries that exist mainly to separate one meaning of a name from another. For the vocabulary problem itself, see the index of what peptides are used for.
Last checked
2026-09-20. PubMed counts were taken through E-utilities and the ClinicalTrials.gov records through the v2 API on that date, with the five P021 registry hits opened individually to establish that none administers the compound.