A 2006 study found that THC inhibits an enzyme implicated in Alzheimer’s pathology, in a test tube. Twenty years later there are randomised trials of a cannabinoid in Alzheimer’s patients — and they are not testing that at all. They are testing agitation. Understanding why the second question replaced the first is more useful than either finding on its own.
Nothing here is medical advice, and nothing here suggests cannabis prevents, slows or treats Alzheimer’s disease. It does not. In Thailand cannabis flower is a controlled herb dispensed only on prescription form PT 33 by an authorised practitioner, and prescribing for a person with dementia raises consent questions that belong with the treating clinician and family.
1. The Laboratory Finding
Eubanks LM et al., “A molecular link between the active component of marijuana and Alzheimer’s disease pathology,” Molecular Pharmaceutics 2006;3(6):773–777, doi:10.1021/mp060066m — work from the Scripps Research Institute in California.
The finding: THC competitively inhibits acetylcholinesterase (AChE), and — the more interesting part — prevents AChE-induced aggregation of amyloid beta peptide, the protein whose accumulation is a defining feature of Alzheimer’s pathology.
That is a genuine result in a serious journal. It identified a specific molecular interaction, and it is why anyone has taken this line of enquiry seriously since.
It is also a computational and in vitro study. There is no patient in it.
2. Three Reasons It Does Not Translate
Acetylcholinesterase inhibition is not a new idea, and it is not disease modification. Three AChE inhibitors are already licensed for Alzheimer’s — donepezil, rivastigmine, galantamine. They have been in use for decades. They produce modest symptomatic benefit for a period, and they do not alter the course of the disease. Even if THC’s inhibition worked in a person, the ceiling on that mechanism is already known, and it is low.
The amyloid hypothesis is contested. The idea that amyloid aggregation is the driver of Alzheimer’s rather than a marker of it has had a difficult two decades: many drugs designed to clear amyloid have failed, and the ones that reached approval have produced modest clinical effects with significant safety monitoring requirements. A compound that interferes with amyloid aggregation in a dish is aiming at a target whose clinical importance is itself under argument.
Concentration and access. Laboratory studies of cannabinoids often use concentrations far higher than human dosing achieves in blood, and blood is not brain tissue. That objection applies here with full force.
So: a real finding, at the very beginning of a pipeline, aimed at a target of disputed importance, through a mechanism whose licensed exemplars produce only modest benefit. Preclinical work has continued since 2006, including low-dose THC studies in cell and mouse models. Twenty years on, there is still no human evidence that cannabinoids modify the disease.
3. The Caution That Holds Up
One observation often made about cannabis in older people holds up well: THC alone can worsen confusion in older patients. It is often suggested that CBD be given alongside THC, in equal or greater quantity, to reduce THC’s psychoactive effect in elderly patients.
The reasoning is pharmacological. Older patients are more sensitive to psychoactive effects and more vulnerable to orthostatic hypotension and falls, and the risk is greater where cognitive impairment already exists. Keeping the THC dose low is the cautious starting point.
It should be labelled for what it is: a reasonable inference from pharmacology, not a finding from a trial comparing THC-only against balanced preparations in dementia patients. Whether CBD reliably blunts THC’s effects in people is not settled, and no trial has established a ratio for this population.
4. What Is Actually Being Tested Now
The clinical research programme in this area is not about amyloid. It is about agitation — and that shift deserves explaining, because it is not a retreat.
Agitation in dementia — restlessness, aggression, distress, resistance to care — is one of the most consequential symptoms in the whole condition. It is a leading driver of caregiver exhaustion and of admission to residential care, and it is frequently what determines whether a person can continue living at home. The drugs most used for it, antipsychotics, carry a boxed warning for increased mortality in elderly patients with dementia. So a treatment that works with a better safety profile would matter a great deal, independently of anything to do with disease pathology.
This question did not start recently. Earlier randomised trials tested nabilone, a synthetic cannabinoid, for agitation in Alzheimer’s disease — for example Herrmann et al., American Journal of Geriatric Psychiatry 2019;27(11):1161–1173, a 39-patient crossover trial that found less agitation on nabilone and more sedation. Trials like this make up the small evidence base that later reviews draw on.
The Johns Hopkins-led dronabinol trial. Seventy-five patients with agitation in Alzheimer’s disease at five US centres were randomised to dronabinol (synthetic THC) 5 mg twice daily or placebo for three weeks. Results were first presented at the International Psychogeriatric Association conference in September 2024. The trial has since been published: Rosenberg PB, … Forester BP, American Journal of Geriatric Psychiatry 2026;34(2):167–179, doi:10.1016/j.jagp.2025.10.011.
- On the Pittsburgh Agitation Scale, one of two co-primary outcomes, dronabinol reduced agitation more than placebo (effect size 0.53, p = 0.015).
- On the other co-primary outcome, the NPI-C agitation/aggression score, the difference was not statistically significant (p = 0.094).
- Somnolence was the only adverse effect linked to dronabinol.
The October 2024 press release said the dronabinol group’s average Pittsburgh Agitation Scale score fell from 9.68 to 7.26, a drop of about 25%, and that placebo scores did not change. The researchers planned longer studies with larger samples. They also stressed that their findings “are not intended to encourage or inform the use of other forms of medical marijuana” (Johns Hopkins Medicine press release, October 2024). The study used only the FDA-approved synthetic form.
The pooled picture (2026). Pan TJ, … Lanctôt KL, “Efficacy and Safety of Cannabinoids for Neuropsychiatric Symptoms of Dementia: A Systematic Review with Meta-analysis,” CNS Drugs 2026;40(5):669–685, doi:10.1007/s40263-026-01277-w — 10 randomised controlled trials, 328 participants:
- A statistically significant reduction in agitation across all included studies, with high heterogeneity between studies — but after excluding studies at high risk of bias, the result was no longer statistically significant.
- In subgroup analyses, benefit was seen at doses above 10 mg THC-equivalent and in severe dementia.
- No significant reduction in total neuropsychiatric symptoms compared with placebo.
- Roughly double the risk of sedation versus placebo (risk ratio 2.09). No significant difference in overall adverse events.
- Authors’ conclusion: efficacy for agitation, evidence scarce for other symptoms, more research needed.
That sensitivity analysis — significant overall, not significant once the weaker studies are removed — is the single most informative line in this article. It is the signature of a real but fragile finding, and it is exactly what gets lost when a result is reported as “cannabis helps dementia agitation.”
5. Two Questions, Kept Apart
| The laboratory question | The clinical question | |
|---|---|---|
| What is being tested | Does THC interfere with Alzheimer’s pathology? | Does a cannabinoid reduce agitation in people who have Alzheimer’s? |
| Evidence | In vitro AChE inhibition and anti-aggregation, 2006; later preclinical work | 10 RCTs, 328 participants (2026 meta-analysis): modest and fragile signal for agitation |
| What a positive result would mean | A target worth pursuing in drug development | A symptomatic option for a specific, distressing problem |
| Current status | Still no human evidence of disease modification after twenty years | Promising, small, sedation-limited, more research needed |
| What it does not mean | That cannabis prevents or slows Alzheimer’s | That cannabis treats Alzheimer’s |
Anyone quoting the Scripps study as evidence that cannabis helps Alzheimer’s patients is answering the clinical question with the laboratory one. The clinical question has its own answer now, it is more modest, and it is about a different thing entirely.
6. The Consent Question Nobody Skips
A person with moderate or severe dementia may not have capacity to consent to a treatment decision. That does not make the decision impossible — it moves it, to a legally recognised substitute decision-maker acting in the person’s best interests, with the treating clinician.
It also means the ordinary safeguards matter more, not less. In this population the specific risks are sedation (doubled in the pooled analysis), falls, worsening confusion, and interaction with the several other medicines most such patients take. A trial of any cannabinoid here belongs inside a documented plan with a defined endpoint and a stopping rule — not as an open-ended addition.
7. In Thailand
Cannabis flower is a controlled herb under the Notification of the Ministry of Public Health Re: Controlled Herbs (Cannabis), B.E. 2568, in force since 26 June 2025. Flower must be prescribed on form PT 33, for no more than 30 days per prescription, with no refills, and must come from a GACP-certified cultivation site. Online sale and advertising of flower are not permitted. The Cannabis Control Act is still only a draft: the Cabinet approved it on 22 September 2026 and sent it to the House, so it is not yet law. Rules and enforcement are changing, so check the current requirements with the prescriber or dispensary.
Whether dronabinol is registered and available in Thailand is a separate question and should be checked with the treating physician or hospital pharmacy; a foreign approval is not a Thai registration.
If you care for someone with dementia in Thailand or the region: what has actually been offered for agitation, and what worked? The comparison that matters clinically is not cannabinoid versus nothing — it is cannabinoid versus the antipsychotics that carry a mortality warning, and almost nobody has data on how that choice gets made here.
This article is general educational information, not medical advice. Cannabis does not prevent, slow or cure Alzheimer’s disease.