Peptides for dementia prevention: hype check on TikTok's latest brain claims

What did @drafroese actually say?

The core argument is this: dementia is a metabolic disease first and a brain disease second. The creator frames Alzheimer's as "type three diabetes," describing insulin resistance in the brain as the upstream driver of plaques, tangles, and cognitive decline. From there, the video pivots to peptides, specifically GLP-1 receptor agonists like semaglutide and a research compound called dihexa, as more promising interventions than current FDA-approved drugs. The creator is candid that dihexa evidence is animal-only so far, and explicitly tells viewers not to stop any medication. Credit where it is due: those disclaimers matter.

The claim that gets the most airtime is dihexa: it "crosses the blood-brain barrier," activates the HGF/c-Met pathway, and is "up to a million times more potent than BDNF" in animal models. The video cuts off before finishing the sentence about why there is hesitancy to test it in humans, which leaves viewers with a dangerously incomplete picture.

Does the science back this up?

The metabolic framing is genuinely supported by a growing body of research, but the dihexa claims need serious qualification. The "type 3 diabetes" hypothesis has real traction in neuroscience literature. De la Monte and Wands (2008, Journal of Alzheimer's Disease) first coined the term, and subsequent work has confirmed that brain insulin signaling is impaired in Alzheimer's patients independent of peripheral diabetes. That part of the video is largely accurate.

On dihexa: the potency claim traces back to McCoy et al. (2013, Journal of Pharmacology and Experimental Therapeutics), which did show dramatic effects on spatial learning in scopolamine-impaired rats. The HGF/c-Met pathway activation is real. However, the reason human trials have stalled is not just regulatory caution. There are concerns about HGF/c-Met signaling being a known oncogenic pathway. Amplifying it systemically in humans carries theoretical cancer risk that has never been resolved. The video never gets to that part.

What did they get wrong (or right)?

The creator gets the metabolic hypothesis mostly right. Insulin resistance, chronic inflammation, and mitochondrial dysfunction are all legitimate upstream contributors to neurodegeneration. The 2024 NEJM trial of lecanemab (van Dyck et al.) showed that even successful amyloid clearance produces modest clinical benefit, which does support the argument that plaques alone are not the whole story.

Where things get shaky is the framing around GLP-1 drugs. Saying "I've seen them do more good than harm" as a defense for dementia use is not evidence, it is anecdote. The ongoing EVOKE and REWIND trials are studying semaglutide and dulaglutide in cognitive outcomes respectively, but results are not yet conclusive. Presenting GLP-1 agents and dihexa as equivalent categories of "peptide therapy" is also misleading. One class has thousands of human trials behind it; the other has rodent data and a theoretical cancer concern. Lumping them together flattens a distinction that matters a lot to patients.

The disclaimer about not stopping medications is appreciated, but it follows several minutes of describing those same medications as things the creator "couldn't get behind." That framing will erode adherence regardless of the legal disclaimer that follows.

What should you actually know?

The metabolic model of dementia is a legitimate and active area of research, not fringe thinking. Addressing insulin resistance, inflammation, and cardiovascular risk factors in midlife is supported by the 2020 Lancet Commission on dementia prevention (Livingston et al.), which identified 12 modifiable risk factors accounting for roughly 40% of dementia cases. Exercise, sleep, blood pressure control, and treating diabetes are the interventions with the strongest evidence.

Dihexa is not available as an approved therapeutic. It is sold through unregulated peptide suppliers with no standardized dosing, no human safety data, and unresolved concerns about oncogenic risk given its mechanism. Anyone encountering it framed as a "supplement" should know it is nothing of the sort. GLP-1 receptor agonists are a different story, with active clinical trials in dementia, but those trials have not yet reported results that would justify use specifically for cognitive protection outside a study protocol.

• The "type 3 diabetes" framing is scientifically grounded but still a hypothesis, not a consensus diagnosis.
• Dihexa's potency claim comes from animal models only. No human safety or efficacy data exists.
• GLP-1 drugs are promising but not yet proven for dementia prevention or treatment.
• The HGF/c-Met pathway dihexa activates is also a known oncogenic pathway. That risk has never been studied in humans.

Bottom line

This video asks the right questions about dementia. The metabolic framing is not wrong, and the skepticism toward amyloid-only treatment models is shared by a lot of serious researchers. But the leap from "current drugs underperform" to "try this unregulated research peptide" skips over a canyon of missing safety data. Viewers deserve to hear both halves of that story.