A new randomized clinical trial is adding significant evidence to the debate over whether carefully standardized cannabis-derived medicines may have a role in treating symptoms associated with autism spectrum disorder. Published in Neurotherapeutics, the Phase II/III Harmony Study tested NTI164, a low-THC, full-spectrum medicinal cannabis extract, against placebo in children and adolescents with moderate-to-severe autism. Researchers reported improvements across several clinician- and caregiver-rated measures, including overall clinical severity, adaptive functioning, social responsiveness, anxiety, emotional regulation and family quality of life.
The results are notable because much of the previous research surrounding cannabis and autism has consisted of observational studies, small trials or studies using different cannabinoid formulations that are difficult to compare. The Harmony Study used a randomized, double-blind, placebo-controlled design, making it stronger evidence than uncontrolled reports. Even so, it does not establish cannabis as a proven general treatment for autism. The study was relatively small, conducted at a single center and had only eight weeks of blinded treatment. What it provides is a promising signal that one highly specific pharmaceutical cannabis preparation deserves larger and more independent testing.
What Is NTI164?
NTI164 is not ordinary cannabis oil or a commercially available CBD supplement. It is a standardized full-spectrum medicinal cannabis extract developed for pharmaceutical research. The preparation contains several cannabinoids, including cannabidiolic acid, or CBDA; cannabidiol, or CBD; cannabigerolic acid, or CBGA; and cannabidivarin, or CBDV. Its concentration of delta-9 THC is extremely low, below 0.3 percent. Researchers designed the extract around a reproducible cannabinoid profile rather than the highly variable compositions found in many retail cannabis products.
That distinction matters when interpreting the results. It would be inaccurate to conclude from the Harmony Study that smoking cannabis, using high-THC products or giving children consumer CBD oils would produce the same effects. NTI164 was administered as an oral investigational medicine under clinical supervision, with dosing gradually increased toward a maximum of 20 milligrams per kilogram per day. The original Phase II/III trial registry describes an initial dose of 5 mg/kg/day followed by weekly increases, allowing clinicians to stop at a participant’s maximum tolerated dose.
How the Harmony Study Was Conducted
The study involved children and adolescents between 8 and 17 years old who had Level II or Level III autism spectrum disorder, classifications generally associated with substantial or very substantial support needs. Sixty-one participants entered the study, and 54 completed the randomized eight-week double-blind portion. The average age was approximately 12.2 years. Participants received either NTI164 or placebo twice daily while researchers assessed changes using established clinical and caregiver-reported instruments.
The primary endpoint was the Clinical Global Impression-Severity scale, or CGI-S, a seven-point clinician assessment designed to provide an overall judgment of symptom severity. Researchers also used measures including the Vineland Adaptive Behavior Scales, Social Responsiveness Scale, Anxiety Depression and Mood Scale and autism-specific anxiety assessments. After the blinded phase, participants initially assigned to placebo could switch to NTI164 during an open-label treatment period, giving investigators another opportunity to see whether improvements appeared once the active medicine was introduced.
Researchers Reported a Large Change in Overall Clinical Severity
The primary result favored NTI164. At the beginning of the study, participants receiving the extract averaged a CGI-S score of approximately 5.5, placing them in the markedly-to-severely affected range used by the scale. After eight weeks, the average had fallen to approximately 3.8. The placebo group changed far less. The estimated treatment effect on CGI-S was about 1.65 points and reached strong statistical significance, with a reported p-value below 0.001.
Results on the Clinical Global Impression-Improvement scale supported the primary finding. Researchers also reported significant improvements in adaptive functioning measured by Vineland-3 and in portions of the Social Responsiveness Scale. Improvements involved practical domains such as communication, daily living skills and socialization, while measures related to social cognition and restricted or repetitive behavior also showed favorable changes. Participants who initially received placebo and later crossed over to NTI164 demonstrated similar improvement patterns during the open-label portion, although open-label results are inherently less reliable because participants and caregivers know treatment is being administered.
Anxiety, Emotional Regulation and Family Life Also Improved
Some of the most interesting secondary findings involved associated symptoms that can substantially affect everyday functioning. On the Anxiety, Depression and Mood Scale, the NTI164 group showed improvements involving anxiety, depressed mood, social avoidance and compulsive behavior. Autism-specific anxiety measurements also favored the cannabis-derived medicine in selected areas. Caregivers reported improvements in family experiences and quality of life, suggesting that perceived benefits were not confined to scores recorded during clinician assessments.
One result helps complicate a simple explanation for these changes: sleep did not significantly improve compared with placebo. Cannabinoid therapies can sometimes affect sleep, and better sleep could theoretically make daytime behavior, anxiety and emotional regulation appear improved. In the Harmony Study, however, the Sleep Disturbance Scale for Children did not show a significant treatment advantage. That does not prove that NTI164 directly altered every behavioral domain, but it makes improved sleep alone an unlikely explanation for the broader pattern of results.
Safety Results Were Encouraging but Still Preliminary
The researchers described NTI164 as well tolerated during the trial. Reported adverse events were mild, transient and non-serious, and no serious adverse events were reported during the eight-week randomized period. Symptoms reported during the study included gastrointestinal complaints, headache and anxiety, with adverse-event rates generally comparable between the treatment and placebo groups. Laboratory monitoring did not reveal clinically significant kidney or liver abnormalities in the reported trial results.
Calling a treatment “safe,” however, requires context. An eight-week trial involving several dozen children cannot establish the long-term neurological, psychiatric, hormonal or developmental safety of repeated cannabinoid exposure. NTI164 also contains extremely little THC, meaning these findings should not be extrapolated to high-THC cannabis. Longer trials with larger populations will be needed to identify uncommon adverse effects, drug interactions and potential differences related to age, other medications or underlying medical conditions. The encouraging short-term safety signal is important, but it is not the same as demonstrating long-term safety for widespread pediatric use.
The Study Builds on Earlier Cannabis and Autism Research
The Harmony results did not emerge in isolation. A 2021 randomized trial published in Molecular Autism studied 150 participants between 5 and 21 years old using either a whole-plant CBD-rich cannabis extract, purified CBD and THC or placebo. Both active preparations used roughly a 20:1 CBD-to-THC ratio. Results were mixed: one primary measure of disruptive behavior did not significantly differ between treatment and placebo, while another clinician-rated global measure showed evidence of improvement with the whole-plant extract. The researchers characterized the findings as preliminary rather than definitive.
Another randomized, double-blind trial involving children with autism tested a CBD-rich cannabis extract and also reported potential improvements in selected symptoms. Meanwhile, biological research has found differences in endocannabinoid signaling among some children with autism. A 2019 Molecular Autism study, for example, reported lower circulating levels of several endocannabinoids in children with ASD compared with controls. That provides a biological reason to investigate the endocannabinoid system, but it does not establish that cannabinoid supplementation corrects autism or that an endocannabinoid deficiency causes the condition.
The Broader Evidence Remains Much Less Certain Than the New Trial May Suggest
A 2025 systematic review in the Journal of Clinical Pharmacology illustrates why caution remains necessary. Researchers screened more than 1,200 references and identified 11 randomized trials evaluating cannabis derivatives or related compounds in autism, but only four had available results involving children or adolescents. Five different cannabinoid preparations had been investigated, making direct comparisons difficult. The review concluded that one whole-plant cannabis study suggested improvement in global symptoms, while evidence for other outcomes remained uncertain. Overall certainty was rated from very low to low.
The Harmony Study strengthens that evidence base because it produced placebo-controlled improvements across several different outcome measures rather than relying only on caregiver observations. Yet it also introduces familiar limitations. Fifty-four completing participants remain a small population for a heterogeneous condition such as autism. The blinded treatment period was short, most outcomes relied to some degree on human ratings, and the study involved investigators and commercial interests connected with development of NTI164. Independent replication will therefore be especially important before the treatment can be regarded as established therapy.
A Larger Phase III Trial Will Be the More Important Test
The development program is already moving toward a larger confirmatory study. A Phase III trial registered as HarmonyPlus, NCT07257939, is designed to evaluate NTI164 against placebo for 16 weeks. The study plans to enroll approximately 98 participants between 6 and 25 years old with Level II or III autism and will again evaluate core and associated symptoms while expanding safety data. Researchers also plan biological analyses intended to investigate potential mechanisms underlying treatment response.
That trial may determine whether the striking improvements reported in the Harmony Study survive testing in a larger group over a longer period. Until then, the most appropriate interpretation is neither dismissal nor celebration of cannabis as an autism treatment. The evidence instead points toward something more specific: a carefully standardized, very-low-THC, multi-cannabinoid medicine produced encouraging results in a controlled pediatric study and now warrants rigorous confirmation. Nothing in the trial shows that conventional marijuana products can substitute for NTI164, and the findings do not support unsupervised cannabinoid treatment of children.
What the NTI164 Study Could Mean for Cannabis Medicine
The Harmony Study represents an important shift in cannabis research. Rather than treating “medical marijuana” as a single intervention, investigators are increasingly testing precisely characterized cannabinoid formulations at controlled doses for specific medical indications. That approach makes it possible to determine whether a particular medicine actually works rather than trying to draw conclusions from cannabis products whose cannabinoid concentrations vary dramatically. NTI164’s combination of CBDA, CBD, CBGA, CBDV and very low THC also raises questions about whether minor cannabinoids may contribute effects that would not appear with CBD alone. The present trial cannot establish which compounds are responsible for the clinical results.
For families and clinicians, the study is promising but not a green light for routine treatment. It provides unusually strong evidence compared with much of the previous cannabis-autism literature, particularly because the improvement appeared across clinician assessments, adaptive functioning measures, social measures and caregiver reports. At the same time, the small sample, short treatment period and need for independent replication remain substantial limitations. If the larger Phase III program reproduces these findings, NTI164 could become one of the most clinically significant cannabinoid medicines investigated for a neurodevelopmental condition. For now, the Harmony Study is best viewed as an important advance in the evidence—not the final answer on cannabis and autism.
