NTI164 and Autism: What the New Cannabis Trial Really Shows

The short answer

A positive study does not need to be dismissed. It needs to be put in proportion. The responsible response is “this deserves confirmation”, not “parents should find a similar oil now”.

What is NTI164?

NTI164 is a proprietary full-spectrum medicinal-cannabis extract developed in Australia. It contains several plant cannabinoids, including CBDA, CBD, CBGA and CBDV, with very low THC. It is an unregistered investigational product, not a standard bottle of cannabidiol.

The theory is that its combination of cannabinoids may affect inflammatory, signalling or endocannabinoid pathways. The trial can test whether the product changes measured outcomes. It cannot prove which ingredient or biological theory caused a change.

How did the Harmony trial work?

The study recruited children and teenagers aged 8 to 17 from a tertiary paediatric neurology clinic in Australia. The paper classified participants as needing substantial or very substantial support (DSM-5 levels 2 or 3). These broad labels do not describe every child's individual needs. Participants were randomly assigned to NTI164 or placebo for eight weeks, followed by an open-label period in which participants could receive the active product.

• 61 participants enrolled.
• 54 completed the blinded phase and were included in the published efficacy analysis: 26 on NTI164 and 28 on placebo.
• The average age was just over 12; 29 completers were male and 25 female.
• The main outcome was the clinician-rated Clinical Global Impression–Severity score.
• Caregiver questionnaires, adaptive functioning and quality-of-life measures were also collected.

Randomisation and placebo control are strengths. The single site, short duration and analysis of completers rather than every randomised participant are important weaknesses.

What did the study find?

After eight weeks, the main clinician-rated severity score favoured NTI164 by 1.65 points, with a 95% confidence interval from 1.01 to 2.29 points in favour of treatment and p<0.001. The numerical difference is relatively large for that scale, but this small completer analysis cannot tell families how much everyday change to expect.

Some Vineland adaptive-functioning domains, affective measures and parts of the Social Responsiveness Scale also favoured the active group. Caregivers reported improvements in aspects of family experience and quality of life.

Not every measure was positive. The SRS total score narrowly missed conventional statistical significance (p=0.055), and its social-communication and social-motivation domains were not significantly different. That does not erase the primary result; it does show that the paper's broad claim about “core autism features” is less simple than its title suggests.

Participants who switched from placebo later improved in a similar direction, but because families and clinicians knew the active product was being given, this open-label phase cannot confirm that NTI164 caused the change.

Why is the result still uncertain?

1. The blinded result is based on 54 completers. Seven of 61 enrolled participants did not complete that phase. Excluding people after randomisation can make a treatment look better if drop-out relates to tolerability or response.
2. It came from one clinic. A result is more reliable when independent teams reproduce it across different services and populations.
3. Treatment lasted eight weeks. We do not know whether any benefit lasts, plateaus or changes with longer exposure.
4. Many outcomes were tested. Some positive secondary findings may occur by chance when researchers examine many scales and subscales.
5. Most outcomes relied on clinician or caregiver ratings. These are useful, but future studies should also show concrete changes in everyday participation, communication, safety and independence.
6. The study had product-company involvement. Funding came from Neurotech International and Fenix Innovation Group. Several authors were affiliated with Fenix, and the paper reports equity or advisory relationships. That does not invalidate the study, but independent replication becomes especially important.

The registry has not been updated since November 2022 and contains no results. CGI-S at week 8 was the registered primary outcome, but Vineland, SRS-2 and ADAMS were not registered for the week-8 assessment used in the paper. Without a published protocol or statistical-analysis plan resolving this discrepancy—and with numerous unadjusted secondary tests—the secondary findings should be treated as exploratory.

What did the trial find about safety?

The paper reports 41 adverse events among 26 participants and describes them as mild, non-serious and temporary. No serious adverse event was reported. Seven participants did not complete the blinded phase: four because of taste, one because of oil intolerance, one after consent was withdrawn and one after an event described as unrelated.

No serious adverse event was reported in this short study, but the small completer sample cannot establish long-term safety, uncommon harms or medicine interactions. The study did not collect detailed seizure, genetic or IQ data, limiting subgroup safety conclusions.

Why NTI164 is not the same as CBD oil

“Cannabis”, “CBD” and “NTI164” are not interchangeable labels:

• NTI164 is a fixed, proprietary full-spectrum trial product containing several cannabinoids and very low THC.
• Prescription cannabinoid medicines have their own formulations, indications and monitoring. Epidyolex, for example, is licensed for certain severe epilepsies—not autism.
• Over-the-counter CBD products vary in composition and quality and may contain different ingredients or amounts from the label.
• High-THC or unregulated cannabis products bring additional psychiatric, developmental and legal risks.

A positive result for NTI164 is evidence about NTI164 in that study. It does not transfer to an oil sold by a wellness shop, a private clinic's different formulation or a product imported online.

Can UK families access NTI164?

The paper says NTI164 is unregistered, available in Australia through clinical trials or compassionate use, and not available outside Australia. It is not a NICE-recommended or routine NHS autism treatment.

On the NHS, cannabis-based medicines are routinely considered only for a small number of indications. Private specialist prescribing exists, but a prescription for another formulation is not access to NTI164 and does not show that autism itself is being treated.

Our broader CBD and medical cannabis autism guide explains UK law, product differences, interactions and clinic red flags.

What happens next?

A larger Phase 3 study is registered. It is expected to use a longer blinded period and more participants. The questions to watch are:

• Does the primary result repeat in a larger intention-to-treat analysis?
• Do independent sites find the same effect?
• Are changes visible in everyday functioning, not only rating scales?
• How many participants stop treatment, and why?
• What happens over months rather than eight weeks?
• Are benefits and harms different by age, epilepsy, intellectual disability or other medicines?

It is fair to call this an early positive result that needs confirmation. “Proven”, “available in the UK” or “CBD treats autism” is not supported.

The bottom line

The Harmony study reported a positive primary outcome, but its small completer sample, single site, short duration and company involvement leave substantial uncertainty. It does not justify UK clinical use or substituting another cannabinoid product for NTI164.

Families can follow the evidence without becoming the experiment. For the whole pipeline, see new autism treatment research in 2026.

Sources and further reading

• Keating et al. — NTI164 Harmony randomised controlled trial, 2026
• Harmony trial — journal full text
• ClinicalTrials.gov — original Harmony study record
• ClinicalTrials.gov — registered NTI164 Phase 3 study
• NHS — medical cannabis
• NICE CG170 — autism support and management recommendations

Disclaimer: Information only, not medical advice. Evidence and source status were checked on 29 August 2026. Do not give a child an unregulated CBD or cannabis product, or change a prescribed medicine, based on an online article.