Though the research is still preliminary, scientists appear to be closing in on a molecular explanation for at least some cases of autism. Research in lab animals, cadavers, and now in a small clinical trial in children seems to support the idea that autism is caused by a lack of chemical regulation in synapses, the junction between brain cells.

The latest evidence for this comes from a clinical trial of 25 autistic children ages six to 17. The trial was conducted by Seaside Therapeutics of Cambridge, Massachusetts. In unpublished data released last week, the company says the children seemed to respond well to the drug arbaclofen and suffer minimal side effects. Arbaclofen is related to the drug baclofen, a commonly used muscle relaxer and antispastic agent. The children took the drug for eight weeks, and most saw significant improvements in measures of irritability and communication, two common problems in autism.

The finding is considered preliminary because the study was not placebo-controlled, meaning scientists couldn’t compare children taking the drug to those taking a placebo. This is especially important in studies of autism because many of the tests used to assess patients are subjective, including behavioral evaluations from parents and clinicians, says Jeremy Veenstra-VanderWeele, assistant professor of psychiatry, pediatrics, and pharmacology at Vanderbilt University. About a half-dozen of his patients were involved in the trial.

Although more study is needed, Veenstra-VanderWeele says,he was pleasantly surprised by the drug’s effectiveness. One teenage patient who had benefited little from other medications was quite agitated at the beginning of the study, unable to stay in the room with investigators for more than a few minutes, Veenstra-VanderWeele says. Eight weeks later, the patient was writing notes to the investigators and seemed far less anxious and aggressive–an unusual and particularly encouraging response to a medication, says Veenstra-VanderWeele, who has no financial relationship with Seaside outside of the trial. The trial itself was funded by Seaside.

Overall, the subjects seemed to make eye contact more easily, and were less agitated and anxious, says Paul Wang, Seaside’s vice president for clinical development and a developmental-behavioral pediatrician. Some were able to interact with peers more readily and were less likely to engage in “stimming,” the repetitive rocking or banging behaviors often characteristic of autism.

Seaside has also tested arbaclofen in children with Fragile X syndrome, a genetic disorder that is often linked to autism. That trial, which was larger and placebo-controlled, showed similar positive results, according to the company. The success of the drug trials, building on other research, suggests that Fragile X is the first good molecular model for autism, says Randi Hagerman, professor of pediatrics at the University of California, Davis, and medical director of the school’s MIND Institute. Hagerman worked with Seaside on the Fragile X trial, but not on this new one.