AST-001 and Autism: Why the Phase 3 Result Matters

The short answer

It is reasonable to be interested in AST-001. It is not reasonable for a seller to present it—or an L-serine supplement—as a breakthrough. The most important evidence is now the failed Phase 3 primary result, even though the full numerical findings have not yet been peer reviewed.

What is AST-001?

AST-001 is a proprietary syrup developed by the South Korean company Astrogen. Its active basis is L-serine, an amino acid the body uses in many processes, including brain development and cell signalling.

The scientific idea is that changing L-serine availability might affect pathways involved in brain function. A plausible idea is only a starting point. To show a useful treatment effect, the exact product must perform better than placebo in well-designed trials and produce changes that matter in daily life.

What did the Phase 2 trial find?

The published Phase 2 study enrolled 151 autistic children aged 2 to 11 in South Korea. For the first 12 weeks, children were randomly assigned to high-dose AST-001, low-dose AST-001 or placebo. The groups and assessors were blinded.

The main outcome was the Korean version of the Vineland Adaptive Behavior Scales composite—a parent interview covering communication, daily living, socialisation and motor skills.

• The high-dose group improved by an adjusted 1.43 points more than placebo. The reported 90% confidence interval was 0.28 to 2.58, with p=0.042.
• The low-dose group did not meet the primary outcome.
• Clinician-rated severity favoured both doses by about a quarter of a point.
• The Social Responsiveness Scale and Aberrant Behavior Checklist did not show a significant difference.

That is a signal, but it is modest. The trial also used a 90% confidence interval for the main comparison rather than the more usual 95% interval. Most participants were boys, most had an intellectual disability and around six in ten were reported as not using functional spoken language (the paper used the term “nonverbal”). Those details matter when deciding whom the result might apply to.

The study's longer extension does not solve the placebo question. During the extension, the placebo group switched to active treatment, so everyone knew the study was no longer a simple blinded comparison. Improvement over time can reflect development, expectations, repeated testing or other support as well as the medicine.

What happened in Phase 3?

The Phase 3 study was designed to provide stronger confirmation. Its ClinicalTrials.gov record lists a randomised, double-blind comparison in young autistic children. The sponsor's August 2026 statement said that active participants improved, but the placebo group improved more than expected and the study did not meet its pre-specified primary endpoint at 12 weeks.

A South Korean regulatory advisory committee also concluded that efficacy could not be accepted from the evidence submitted. Reporting at the time said a proposed result in a narrower 2-to-5-year-old group and motor-domain findings were post-hoc—chosen after looking at the data—and could be confused with ordinary early-childhood development.

The sponsor says it is preparing a new confirmatory trial. That is exactly what should happen when the evidence is uncertain. But until detailed results are posted and a further trial succeeds, AST-001 should not be described as effective.

Why can Phase 2 look positive and Phase 3 fail?

This pattern is common across medicine, and especially important in autism research:

• Small effects can be unstable. A 1.43-point difference may disappear when tested in a new group.
• Caregiver ratings provide important information about daily life. Scores can also be affected by expectations, extra trial contact, natural development and ordinary fluctuation. This does not mean parents are mistaken or dishonest.
• Young children develop quickly. Both groups may improve during a study, making a small drug effect hard to separate from normal development.
• Autistic children are diverse. A broad trial may average together children who respond differently—but a subgroup must be predicted and confirmed, not discovered after failure.
• Larger confirmatory trials reduce uncertainty. A signal that does not repeat should not be treated as established.

Our guide to why autism trials fail explains primary outcomes, placebo response and subgroup claims in more detail.

What do we know about safety?

In the Phase 2 trial, diarrhoea was the most common treatment-related reaction. Treatment-related adverse reactions were not more frequent in the active groups than in placebo, and the serious events recorded were judged unrelated to the product.

A later uncontrolled 52-week study enrolled 61 children selected as prior “responders”; 49 completed. Two adverse drug reactions—reduced appetite and enteritis—and three serious adverse events, including one fatal septic-shock event, were reported. Investigators judged the serious events unrelated to AST-001. The selected population and lack of a concurrent placebo group mean this study cannot establish effectiveness or rule out uncommon harms.

“No safety concern detected” in a small study is not the same as “known to be safe for every child”. Rare harms, interactions and long-term effects need larger numbers and routine monitoring.

Can families access AST-001 in the UK?

AST-001 is not a NICE-recommended or routine NHS treatment for autism. We found no evidence that it is available as an authorised autism medicine in the UK. Trial status can change, but the published and registered studies have been based in South Korea.

NICE says medication should not be offered to manage the core features of autism in children and young people. That does not stop clinicians treating a separate condition, such as epilepsy, ADHD, pain or sleep problems, where an ordinary evidence-based treatment is indicated.

Be cautious if a seller says:

• “The Phase 2 trial proved L-serine works.”
• “Phase 3 only failed because the placebo group improved.”
• “Our supplement is basically AST-001.”
• “There is no downside because L-serine is natural.”
• “Your child fits the responder subgroup.”

Each claim goes beyond what the current evidence can establish.

The bottom line

AST-001 was worth taking into a Phase 3 trial. That trial did not meet its main outcome. Until transparent, peer-reviewed confirmatory evidence changes the picture, AST-001 remains experimental and L-serine supplements should not be presented as an autism treatment.

Parents do not need to choose between blind optimism and giving up. The evidence-based middle is to follow the research, avoid self-dosing, and focus current support on the child's actual communication, health, sensory, learning and wellbeing needs.

For the wider pipeline, see new autism treatment research in 2026.

Sources and further reading

• Kim et al. — AST-001 versus placebo, Phase 2 randomised trial
• AST-001 Phase 2 — free full text
• ClinicalTrials.gov — AST-001 Phase 3 study, NCT06333964
• Astrogen — sponsor's Phase 3 statement, August 2026 (Korean)
• Seoul Economic Daily — independent report on the regulatory review
• AST-001 52-week open-label follow-up of prior responders
• NICE CG170 — 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 L-serine or another supplement at a research dose without advice from an appropriately qualified clinician.