Hopeful Therapies

Here to support your little one AND you

Supplements and Autism: How to Read the Evidence Before Spending Money and Hope

Leer este artículo en español

Series: The Whole Child · Part 3 of 4

Folinic acid, N-acetylcysteine, L-carnitine, coenzyme Q10, sulforaphane. These five compounds run through the entire biomedical conversation about autism. This post explains what the trials actually show, what the FDA decided about leucovorin in 2026, and how to evaluate any intervention without relying on subjective impression.

Published by the Hopeful Therapies clinical team · 11 min read

Read this first. Hopeful Therapies is an applied behavior analysis provider. We do not prescribe, recommend, or sell supplements, and we deliberately omit dosages on this page. We wrote this because our families ask about these compounds almost every week, and they deserve an honest read of the evidence rather than promotional material. Every decision about supplementation for a child belongs to their physician.

The question behind the question

When a parent asks us about folinic acid, they are almost never asking about biochemistry. They are asking: am I leaving something untried?

That question deserves respect and a direct answer. Ours is that there is an important difference between “there is a trial with interesting results” and “this works,” and that the supplement industry makes money by blurring that difference. Learning to tell them apart is the most useful skill a family can develop in this field — more useful than any product list.

The frame: two large reviews, two compatible conclusions

In December 2025, Nature Human Behaviour published an umbrella review from Corentin Gosling’s team assessing 248 meta-analyses, roughly 200 clinical trials, and more than 10,000 participants across 19 complementary, alternative, and integrative medicine interventions in autism. Conclusion: there is no high-quality evidence supporting the efficacy of any of them for core or associated autism symptoms. Several showed promising signals, resting on very low-quality evidence [1]. They also published an open, searchable database: ebiact-database.com. If you save only one link from this series, save that one.

Antonio Persico and colleagues’ systematic review covered 115 randomized controlled trials across 133 compounds and reached a somewhat different but compatible formulation: clinical use is fully evidence-based only for melatonin, and for six additional compounds clinical use may be acceptable with caution, though more research is necessary [2]. Those six are folinic acid, N-acetylcysteine, L-carnitine, coenzyme Q10, sulforaphane, and metformin (the last a prescription medication we don’t discuss here).

Editorial accuracy note. This review is frequently cited as though it declared those six compounds “promising and safe” for clinical use. The authors’ phrasing is more cautious, and the difference matters: the only compound they treat as fully evidence-based is melatonin. We flag it because a family reading “promising and safe” makes a different decision than one reading “more research is necessary.”

The five compounds, with their actual evidence

Compound Best available evidence Key limitations
Folinic acid (leucovorin) Frye et al., Molecular Psychiatry 2018: double-blind RCT, 48 children, 12 weeks; improvement in verbal communication versus placebo [3]. Panda et al., European Journal of Pediatrics 2024: RCT, 80 children, 24 weeks [4]. The 2018 trial’s own authors describe the findings as preliminary until assessed in larger, longer multicenter studies. It is a prescription medication.
N-acetylcysteine (NAC) Hardan et al., Biological Psychiatry 2012: pilot RCT, reduced irritability [5]. Wink et al., Molecular Autism 2016: raised blood glutathione but showed no significant effect on the primary social impairment outcome [6]. Reported effects are on irritability, not core traits. Small-sample pilot trials. Mixed results.
L-carnitine Geier et al., Medical Science Monitor 2011: RCT, 30 children, 3 months [7]. Mechanistic basis: Rossignol & Frye review of mitochondrial dysfunction [8]. Very small sample. A single principal randomized trial. Plausibly applicable only in children with documented mitochondrial weakness.
Coenzyme Q10 Mousavinejad et al., Psychiatry Research 2018: 90 children, improvements in oxidative stress markers and in sleep and GI symptoms [9]. Positive outcomes were biochemical markers and associated symptoms, not core autism traits.
Sulforaphane Singh et al., PNAS 2014: RCT in 44 young men (ages 13-27); 34% reduction in Aberrant Behavior Checklist and 17% in Social Responsiveness Scale [10]. The replication trial in children aged 3-12 (Zimmerman et al., Molecular Autism 2021) did not reach significance on its primary outcome [11]. This is exactly the pattern that demands caution.

The last row is worth pausing on, because it teaches you how to read the rest of the table. Sulforaphane produced a striking result in 2014 in adolescents and young adults, and when the same research group tried to replicate it in younger children in 2021, the primary outcome was not significant. That is the normal cycle of science and the reason a single positive trial isn’t enough. When a product is marketed citing only the 2014 study, the omission is not accidental.

The leucovorin case: what actually happened

This is the part of this post we consider most important, because a great deal of imprecise information circulates and because it affects real families’ decisions.

The scientific starting point is solid. There is a condition called cerebral folate deficiency, in which folate does not adequately reach the brain even when blood levels appear normal. One cause is autoantibodies against the folate receptor alpha, reviewed by Daniel Rossignol and Richard Frye in Journal of Personalized Medicine in 2021 [12]. Folinic acid can bypass that blocked transport, and the Frye 2018 and Panda 2024 trials tested precisely that [3][4].

What happened next on the regulatory side has often been told badly. The verifiable sequence is this:

  • In September 2025 the FDA asked GSK to refile an application for leucovorin for cerebral folate deficiency — an unusual move — and public communications linked the condition to “developmental delays with autistic features.”
  • In March 2026 the FDA approved leucovorin (Wellcovorin) exclusively for cerebral folate deficiency with a confirmed FOLR1 gene variant, an ultra-rare genetic form that, per an agency official, affects fewer than 1 in a million people. The approval was not for autism.
  • GSK subsequently withdrew its application. This does not affect the availability of generic leucovorin.

Editorial accuracy note. We have seen this process described as though the FDA recognized leucovorin as an autism treatment, or as though such approval were the natural next step. It is not. The approval covered a genetic condition affecting fewer than 1 in 1,000,000 people, and the agency was explicit that it was not approving an autism indication. Folate receptor autoantibodies — which are indeed common in autism per the 2021 review — are a distinct entity from the FOLR1 variant the approval covers. Conflating them leads a family to believe an approved treatment exists where none does.

None of this means the research line is invalid. It means that as of August 2026, folinic acid for autism remains an off-label use supported by two randomized trials whose own authors describe the results as preliminary. It’s a legitimate conversation with a physician. It is not an established treatment.

Methylation and glutathione

The biochemical basis is real. Jill James and colleagues at the University of Arkansas documented in American Journal of Clinical Nutrition in 2004 that autistic children, as a group, show measurable abnormalities in the methionine-homocysteine-glutathione cycle: lower S-adenosylmethionine, lower cysteine, lower reduced glutathione, and a higher ratio of oxidized to reduced glutathione [13]. In 2009 the same group published a study in which methylcobalamin and folinic acid over three months corrected those biochemical markers [14].

That second study is the one to read carefully. It showed the biochemistry can be normalized. Showing that normalizing the biochemistry improves the child’s life is a different question, requiring randomized trials with functional outcomes, blinding, and control groups. Improving a lab marker is an intermediate outcome, not a clinical one. Medicine has a long history of interventions that corrected the marker without benefiting the patient.

We add a safety consideration that gets little mention: methylation support is not innocuous. Clinical practice reports irritability, hyperactivity, or sleep disruption with certain doses or forms of methyl donors with some regularity. A compound that can worsen a child’s behavior requires medical supervision, not self-experimentation.

What these reviews say does NOT work

Equally useful, and published less. The Persico review explicitly notes that current evidence does not support, for core autism traits, hyperbaric oxygen therapy, high-dose intravenous immunoglobulin outside specific immune indications, secretin, or the older megavitamin protocols using high-dose pyridoxine and magnesium [2].

We mention it because these are expensive interventions, some invasive, and well-meaning families continue to invest finite resources in them.

Clinical note for BCBAs and RBTs

Our usefulness here is not in opining on the compound. It’s in measurement design, which is literally our specialty and which almost no one else on the child’s team can contribute.

Offer structured measurement, not opinion. When a family reports that their physician is starting a biomedical intervention, the appropriate response is neither to endorse nor discourage it, but to offer what we know how to do: establish baseline, operationally define the outcomes the family says matter to them, and measure with the same procedure before, during, and after.

Minimum defensible design. Baseline of at least 2-3 weeks before onset; operationally defined outcomes prioritized by the family; continuous measurement using the same procedure; validated scales administered by the same informant at the same intervals; and explicit logging of concurrent changes in the ABA program, medication, school setting, or sleep. Without that last item, the design isn’t interpretable.

Caution on causal attribution. An AB design in a single case does not permit causal attribution. Maturation, regression to the mean, caregiver expectancy, and concurrent therapy effects remain unruled-out alternatives. Communicate this to the family before starting, not after seeing the data. Where clinical and ethical context permits, a withdrawal or multielement design contributes far more — but that is the prescribing physician’s decision, not ours.

Document adverse behavioral reactions. Increased irritability, sleep disruption, or increased problem behavior following the start of a supplement is clinically relevant information the physician needs. Report it with data and dates, without interpreting mechanism.

Role boundary. We do not recommend compounds or doses, do not interpret lab results, do not validate biomedical protocols, and do not refer to providers in whom we hold any financial interest. BACB standards 2.01, 2.03, 1.11, and 1.13.

For families: five questions before you buy

We find these five questions more useful than any product list:

1. Is there more than one placebo-controlled randomized trial, and did the result replicate? A single positive trial is a starting point. A failed replication attempt — like sulforaphane’s in younger children — is essential information that advertising omits.

2. What exactly was measured? Many positive results are about irritability, sleep, or lab markers, not communication, learning, or independence. That may still be worthwhile, but it’s different from what’s usually promised.

3. Who measured? If the outcome was reported by the caregiver who knew their child was receiving active treatment, the result incorporates expectancy. That’s not fraud; it’s a known design limitation.

4. Who benefits if I buy this? If the article that informed you links to the store selling the product, that doesn’t invalidate the science cited, but it should change how you weight the recommendation. Check it against the EBIACT database, which sells nothing.

5. How will I know if it worked? If you can’t answer this before starting, then six months from now you likely won’t know whether the change you’re seeing came from the supplement, the ABA program, the new school year, or your child’s normal development. Here we can help, and that’s what we offer.

One last point, said with care. None of these interventions has shown effects comparable to what your child gets from well-implemented behavioral and developmental intervention, delivered consistently, by a team that knows them. That’s the best-supported work, and also the least marketed — because no one has a product to sell.

Questions about your child’s program? Our clinical team can meet with you to review your data, discuss goals, and explain how we coordinate with your pediatrician. Call (813) 655-8159 or write to contact@hopefultherapies.com. We serve families in Brandon, Lakeland and Winter Haven, Florida.

About this series. This post discusses and responds to “The Most Evidence-Based Biochemical Therapies for Autism: Folinic Acid, NAC, Methylation Support, and Beyond” by Jill Carnahan, MD, published by the Foundation for Alternative and Integrative Medicine. The text on this page is original to Hopeful Therapies. We verified the primary sources the author cites and, where our reading differs, we say so with the corresponding source. The original article includes pediatric dosages and links to supplement products marketed by the author; we reproduce none of them, because prescribing falls outside our scope of practice and because we hold no commercial relationships with supplement vendors.

References

  1. Gosling CJ, Boisseleau L, Solmi M, et al. Complementary, alternative and integrative medicine for autism: an umbrella review and online platform. Nat Hum Behav. 2025;9(12):2610-2619. nature.com/articles/s41562-025-02256-9
  2. Persico AM, Asta L, Chehbani F, et al. The pediatric psychopharmacology of autism spectrum disorder — Part II: The future. Prog Neuropsychopharmacol Biol Psychiatry. 2025;136:111176. sciencedirect.com
  3. Frye RE, Slattery J, Delhey L, et al. Folinic acid improves verbal communication in children with autism and language impairment. Mol Psychiatry. 2018;23(2):247-256. nature.com/articles/mp2016168
  4. Panda PK, Sharawat IK, Saha S, et al. Efficacy of oral folinic acid supplementation in children with autism spectrum disorder. Eur J Pediatr. 2024;183(11):4827-4835. doi:10.1007/s00431-024-05762-6
  5. Hardan AY, Fung LK, Libove RA, et al. A randomized controlled pilot trial of oral N-acetylcysteine in children with autism. Biol Psychiatry. 2012;71(11):956-961. doi:10.1016/j.biopsych.2012.01.014
  6. Wink LK, Adams R, Wang Z, et al. A randomized placebo-controlled pilot study of N-acetylcysteine in youth with autism spectrum disorder. Mol Autism. 2016;7:26. doi:10.1186/s13229-016-0088-6
  7. Geier DA, Kern JK, Davis G, et al. A prospective double-blind, randomized clinical trial of levocarnitine to treat autism spectrum disorders. Med Sci Monit. 2011;17(6):PI15-PI23. doi:10.12659/msm.881792
  8. Rossignol DA, Frye RE. Mitochondrial dysfunction in autism spectrum disorders: a systematic review and meta-analysis. Mol Psychiatry. 2012;17(3):290-314. doi:10.1038/mp.2010.136
  9. Mousavinejad E, Ghaffari MA, Riahi F, et al. Coenzyme Q10 supplementation reduces oxidative stress in children with autism spectrum disorders. Psychiatry Res. 2018;265:62-69. doi:10.1016/j.psychres.2018.03.061
  10. Singh K, Connors SL, Macklin EA, et al. Sulforaphane treatment of autism spectrum disorder (ASD). Proc Natl Acad Sci U S A. 2014;111(43):15550-15555. doi:10.1073/pnas.1416940111
  11. Zimmerman AW, Singh K, Connors SL, et al. Randomized controlled trial of sulforaphane and metabolite discovery in children with Autism Spectrum Disorder. Mol Autism. 2021;12(1):38. doi:10.1186/s13229-021-00447-5
  12. Rossignol DA, Frye RE. Cerebral folate deficiency, folate receptor alpha autoantibodies and leucovorin (folinic acid) treatment in autism spectrum disorders. J Pers Med. 2021;11(11):1141. PubMed 34834493
  13. James SJ, Cutler P, Melnyk S, et al. Metabolic biomarkers of increased oxidative stress and impaired methylation capacity in children with autism. Am J Clin Nutr. 2004;80(6):1611-1617. doi:10.1093/ajcn/80.6.1611
  14. James SJ, Melnyk S, Fuchs G, et al. Efficacy of methylcobalamin and folinic acid treatment on glutathione redox status in children with autism. Am J Clin Nutr. 2009;89(1):425-430. doi:10.3945/ajcn.2008.26615
  15. On the FDA’s leucovorin decision (March 2026): BioPharma Dive; Fierce Pharma; BioSpace (GSK’s subsequent withdrawal of the application).

Important notice. Hopeful Therapies provides applied behavior analysis (ABA) services. We are not a medical provider and do not diagnose or treat medical conditions, nor do we prescribe or recommend diets, supplements, or medications. Dosages from the cited trials have been deliberately omitted from this page. This content is educational and does not constitute medical advice or substitute for the relationship with your pediatrician or other qualified health professional. Do not start, stop, or change any treatment or supplement based on this post. Several compounds mentioned can produce adverse effects, including behavioral changes, and can interact with medications; their use in minors requires medical supervision. Our services are delivered in accordance with the BACB Ethics Code for Behavior Analysts and applicable Florida Agency for Health Care Administration (AHCA) requirements. We have no commercial relationship with any supplement manufacturer, laboratory, or testing provider mentioned or linked in this series, and we receive no compensation for referrals.