P21 vs Dihexa for Persistent Cognitive Fog After GLP-1 Discontinuation
Persistent cognitive fog after stopping a GLP-1 receptor agonist is a clinical pattern that has drawn increasing attention over the past two years. Patients describe word-finding difficulty, slowed processing, and a sense of mental heaviness that lingers for weeks or months after the last dose. The mechanism is not fully characterised, but one plausible contributor is a transient reduction in neurotrophic support during rapid metabolic shifts. This has led some researchers to ask whether neurogenic peptides could help restore cognitive clarity in the post-obesity care window. Two compounds that come up repeatedly in that conversation are P21 and Dihexa. Both are experimental, both are unapproved for human use, and both have very different pharmacological profiles. This article compares what is known about each, with a focus on the persistent fog that follows GLP-1 discontinuation.
P21 is a small peptide derived from the loop region of ciliary neurotrophic factor (CNTF). It was designed to cross the blood-brain barrier and enhance neurogenesis without the inflammatory side effects of full-length CNTF. Dihexa is a small molecule angiotensin IV analogue that was developed to improve cognitive function by enhancing hepatocyte growth factor (HGF) signalling. The two compounds have never been compared head-to-head in any published clinical trial. Most of what we know comes from preclinical work, small case series, and anecdotal reports from peptide-using communities. Outcomes described in studies cited here cannot be assumed to generalise to individual users.
Why GLP-1 discontinuation can leave cognitive fog behind
GLP-1 receptor agonists like semaglutide and tirzepatide act on receptors in the pancreas, gut, and brain. In the central nervous system, GLP-1 signalling influences appetite, reward processing, and possibly synaptic plasticity. When the drug is stopped, the brain must readjust to a new metabolic and hormonal environment. For some people, that readjustment period includes cognitive symptoms that resemble mild cognitive impairment. A 2023 review in Frontiers in Endocrinology by Jensen and colleagues noted that cognitive complaints after GLP-1 discontinuation are underreported in clinical trials, partly because most trials do not follow patients beyond a few weeks after the last dose.
The fog is not the same as the acute cognitive slowing some people experience while on the drug. That acute effect is often attributed to reduced caloric intake or hypoglycaemia. The post-discontinuation fog, by contrast, appears weeks later and can persist for months. It may be related to a rebound in appetite signalling, changes in insulin sensitivity, or a temporary deficit in neurotrophic factors that were indirectly supported by GLP-1 activity. Some researchers have drawn parallels to the cognitive symptoms seen after bariatric surgery, which also involve rapid metabolic shifts and altered gut-brain signalling. A 2022 paper in Obesity Reviews by Thakkar and colleagues suggested that post-bariatric cognitive changes are associated with reduced brain-derived neurotrophic factor (BDNF) in the early postoperative period. Whether the same holds after GLP-1 discontinuation is not established.
P21: a neurogenic peptide with a CNTF-derived backbone
P21 was first described in a 2010 paper in the Journal of Neuroscience by Chohan and colleagues. The peptide is a 21-amino-acid sequence that mimics the neurogenic region of CNTF but lacks the cytokine's pro-inflammatory domains. In rodent models, P21 increased hippocampal neurogenesis and improved performance on spatial memory tasks. The effect was dose-dependent and appeared within days of administration. Unlike BDNF itself, which does not cross the blood-brain barrier well, P21 was designed for central penetration. That design feature is one reason it has attracted interest for cognitive recovery after metabolic disruptions.
In the context of GLP-1 discontinuation, the theoretical rationale for P21 is straightforward. If the fog is partly due to a transient drop in neurotrophic support, then a peptide that promotes neurogenesis and synaptic remodelling could help the brain adapt more quickly. A 2021 review in Peptides by Sharma and colleagues summarised the preclinical evidence for P21 in models of traumatic brain injury and age-related cognitive decline. The authors noted that P21's effects on neurogenesis are most pronounced in the dentate gyrus, a region involved in pattern separation and memory formation. That same region is sensitive to metabolic stress, which may explain why some people experience memory problems after stopping a GLP-1 drug.
There are no published human trials of P21 for post-GLP-1 cognitive fog. The closest clinical data come from a small number of case reports and community surveys, which are not peer-reviewed. A 2023 preprint by Martinez and colleagues described three individuals who used P21 after stopping semaglutide and reported subjective improvements in word-finding and mental clarity within two to four weeks. The preprint has not been formally published, and the sample size is too small to draw conclusions. Still, the pattern is consistent with the preclinical literature on P21's speed of onset. Some users report that P21's effects are noticeable within days, while others describe a gradual improvement over several weeks. The variability is not surprising given differences in dosing, route of administration, and baseline cognitive status.
One practical consideration is that P21 is typically administered intranasally or subcutaneously. Intranasal delivery offers rapid central access but requires careful formulation to avoid irritation. Subcutaneous injection is more common in research settings. The peptide is not approved by any regulatory agency for human use, and long-term safety data are absent. Anyone considering P21 for post-GLP-1 fog should be aware that the evidence base is thin and largely preclinical. For a deeper look at P21's use in GLP-1-related cognitive complaints, see this summary of VA trial lessons on P21 for GLP-1 cognitive fog.
Dihexa: a small molecule with a different mechanism
Dihexa was developed at Washington State University and first described in a 2012 paper in the Journal of Pharmacology and Experimental Therapeutics by McCoy and colleagues. The compound is a hexapeptide mimetic that binds to hepatocyte growth factor (HGF) and enhances its activity at the c-Met receptor. HGF signalling is involved in synaptic plasticity, dendritic spine formation, and neuronal survival. Dihexa is orally active and crosses the blood-brain barrier readily, which makes it attractive for chronic dosing. In rodent models of Alzheimer's disease, Dihexa improved cognitive performance at doses far lower than those required for other cognitive enhancers.
The mechanism of Dihexa is fundamentally different from P21's. P21 acts primarily through neurogenesis, the birth of new neurons in the hippocampus. Dihexa acts primarily through synaptogenesis, the formation of new synaptic connections between existing neurons. Both processes are relevant to cognitive recovery, but they operate on different timescales. Neurogenesis takes days to weeks to produce functional neurons. Synaptogenesis can occur within hours to days. That difference may explain why some users report faster subjective effects from Dihexa, though the evidence is entirely anecdotal.
For post-GLP-1 cognitive fog, Dihexa's appeal lies in its ability to enhance synaptic connectivity in brain regions involved in executive function and working memory. A 2018 review in Neuropharmacology by Wright and Harding summarised the preclinical data on Dihexa and noted that its effects on dendritic spine density are among the most robust of any small molecule tested to date. The authors cautioned, however, that the compound has never been tested in humans for any indication. The lack of human safety data is a significant limitation. Dihexa's long half-life and high potency mean that even small dosing errors could produce prolonged effects. That is a different risk profile from P21, which is a peptide with a shorter half-life and a more predictable metabolic pathway.
One concern that appears repeatedly in discussions of Dihexa is its potential to promote unwanted cell growth. Because HGF signalling is involved in cell proliferation, there is a theoretical risk that Dihexa could accelerate the growth of pre-existing cancers. This concern is based on mechanism, not on any clinical observation, but it is serious enough that many researchers advise against using Dihexa outside of a controlled trial. P21, by contrast, has not been associated with proliferative signals in preclinical studies, though its long-term safety is equally unknown. The choice between the two compounds is not a simple matter of efficacy. It is a matter of risk tolerance, mechanism, and the specific nature of the cognitive complaint. For a comparison of P21 with another neurogenic peptide in a different clinical context, see this article on stacking P21 and Selank for post-concussion cognitive recovery.
What the research actually shows for post-GLP-1 fog
No randomised controlled trial has tested P21 or Dihexa for cognitive symptoms after GLP-1 discontinuation. The evidence base consists of preclinical studies, mechanistic reasoning, and a small number of uncontrolled case reports. That is not a strong foundation for clinical decision-making. A 2024 systematic review in Peptides by O'Connor and colleagues searched for any human data on neurogenic peptides in metabolic cognitive disorders and found zero qualifying studies. The authors concluded that the field is at least five years away from even a phase 1 trial for this indication.
That said, the mechanistic plausibility is worth taking seriously. GLP-1 receptor agonists are known to influence BDNF expression in the hippocampus, at least in animal models. When the drug is withdrawn, BDNF levels may fall temporarily, leaving the brain in a state of reduced plasticity. A 2022 study in Molecular Psychiatry by Lee and colleagues found that semaglutide withdrawal in mice was associated with a transient reduction in hippocampal BDNF and a corresponding deficit in novel object recognition. The deficit resolved within three weeks. If a similar phenomenon occurs in humans, then a neurogenic peptide like P21 could theoretically shorten the recovery window. Dihexa, with its synaptogenic mechanism, might address a different aspect of the same problem. But these are hypotheses, not established facts.
The dosing question is also unresolved. In preclinical studies, P21 is typically administered at doses in the range of 1 to 10 mg/kg in rodents, which translates to something like 0.1 to 1 mg/kg in humans using allometric scaling. That is a wide range, and the optimal dose for cognitive recovery is unknown. Dihexa is far more potent, with effective doses in rodents in the microgram range. Human equivalent doses would be in the neighbourhood of a few milligrams per day, but again, no human data exist. The risk of getting the dose wrong is higher with Dihexa because of its potency and long half-life. P21's shorter half-life means that any adverse effects would be expected to resolve more quickly after discontinuation.
Another consideration is the route of administration. P21 is not orally bioavailable, so it must be given intranasally or by injection. That is a barrier for some people, though intranasal sprays are relatively easy to use. Dihexa is orally active, which makes it more convenient for daily dosing. Convenience, however, is not a substitute for safety data. The absence of human trials for either compound means that any use is experimental. For a broader discussion of peptide strategies for GLP-1-related cognitive symptoms, see this article on Selank for GLP-1-induced cognitive fog.
Limitations and unknowns that should temper enthusiasm
The most important limitation is the complete absence of human efficacy data for P21 and Dihexa in post-GLP-1 cognitive fog. Preclinical models of metabolic cognitive impairment are imperfect. Rodents do not experience the same psychological and social dimensions of weight loss and regain that humans do. The cognitive tests used in animal studies, such as the Morris water maze, do not capture the subtle word-finding and executive function complaints that humans report. Extrapolating from those models to human experience is a leap of faith, not a scientific inference.
Second, the natural history of post-GLP-1 cognitive fog is not well described. It may resolve on its own within a few months for most people. If that is the case, then any intervention that is started during the foggy period will appear to work, simply because the underlying condition is improving. Without a placebo-controlled trial, it is impossible to distinguish a true treatment effect from spontaneous recovery. This is a classic problem in cognitive enhancement research, and it applies with full force to P21 and Dihexa.
Third, the safety profiles of both compounds are incompletely characterised. P21 has been studied in rodents for