WASHINGTON (AP) -- When President Donald Trump's administration announced it would repurpose an old, generic drug as a new treatment for autism, it came as a surprise to many experts -- including the physician who suggested the idea to the nation's top health officials.

Dr. Richard Frye told The Associated Press that he'd been talking with federal regulators about developing his own customized version of the drug for children with autism, assuming more research would be required.

"So we were kinda surprised that they were just approving it right out of the gate without more studies or anything," said Frye, an Arizona-based child neurologist who has a book and online education business focused on the experimental treatment.

It's another example of the haphazard rollout of the Trump administration's Monday announcement on autism, which critics say has elevated an unproven drug that needs far more study before being approved as a credible treatment for the complex brain disease.

A spokesperson for the Republican administration did not immediately respond to a request for comment Wednesday morning.

The nation's leading autism groups and researchers quickly distanced themselves from the decision on leucovorin, a derivative of vitamin B, calling the studies supporting its use "very weak" and "very small."

"We have nothing resembling even moderate evidence that leucovorin is an effective treatment for autism symptoms," said David Mandell, a psychiatrist at the University of Pennsylvania.

Mandell and other researchers say the evidence suggests autism is mostly rooted in genetics, with input from other factors, including the age of the child's father.

Nevertheless, a growing number of doctors are prescribing the medication, repurposing versions used for chemotherapy or ordering new formulations from compounding pharmacies.

Many researchers agree the drug warrants additional study, particularly for patients with a deficiency of folate, or vitamin B9, in the brain that may play a role in autism. But for now, they say, it should only be taken in carefully controlled clinical trials.

"We often say our job is to stay between the yellow lines," said Dr. Lawrence Gray, a pediatric developmental specialist at Northwestern University. "When people just decide to go outside of current guidelines, then they're outside of that. And nobody knows what's going to happen out there."

The case for leucovorin's use in autism begins with established science but quickly veers into uncertain terrain.

When metabolized, the drug turns into folate, which is essential for healthy prenatal development and is recommended before and during pregnancy. But far less is known about its role after birth.

The issue caught the attention of Frye and others more than 20 years ago, when research suggested some people with autism had low levels of folate in the brain due to antibodies blocking the vitamin's absorption.

The theory linking autism to folate levels was mostly abandoned, however, after research showed that the siblings of people with autism can also have low folates without any symptoms of the condition.

"I honestly thought this had died out as a theory for autism and was shocked to see its reemergence," Mandell said.

In 2018, Frye and his colleagues published a study of 48 children in which those taking leucovorin performed better on several language measures than those taking a placebo.

Four small studies in other countries, including China and Iran, showed similar results, albeit using different doses, metrics and statistical analyses, which researchers say is problematic.

Frye struggled to get funding to continue within the traditional academic system.

"I decided to move out of academia to be more innovative and actually do some of this stuff," he said.

Earlier this year, Frye and several other researchers formed a new entity, the Autism Discovery Coalition, to pitch their work to Trump administration officials including Health Secretary Robert F. Kennedy Jr.

"After Kennedy got in, we thought they'd hopefully be friendly to autism scientists," he said.

An August meeting with National Institutes of Health Director Jay Bhattacharya quickly led to further discussions with the Food and Drug Administration about testing a proprietary, purified version of leucovorin.

A new formulation of the decades-old drug would mean new patents, allowing Frye and his yet-to-formed drug company to charge far more than the cheap generics currently on the market.

"We have a lot of investors who are excited about leucovorin and want to do something high quality for kids with autism," he said.

But the FDA's announcement Monday may have scuttled that plan. Instead of previewing a new version, the agency said it would simply update the label on the generic drug to mention use in boosting folate brain levels, including for patients with autism. That's expected to encourage more doctors to prescribe it and insurers to cover it.

Specialists who have spent decades treating autistic patients say it's important to proceed carefully and methodically.

Gray recalls other experimental treatments that initially looked promising only to fail in larger studies.

"Small studies often find populations that are very motivated," Gray said. "But when those therapies are moved into larger studies, the initial positive findings often disappear."

Among the challenges facing leucovorin: There isn't agreement about what portion of autism patients have the folate-blocking antibodies supposedly targeted by the drug.

Frye screens his patients for the antibodies using a test developed at a laboratory at the State University of New York. Like many specialty tests, it has not been reviewed by the FDA.

Gray says the only way to definitively test for the antibodies would be by extracting cranial fluid from children with autism through a spinal tap.

"That's a big limiting factor in having these large, randomized controlled trials," Gray said.

While the Trump administration discusses fast-tracking leucovorin, interest in the drug continues to swirl online, including in forums and social media groups for parents of children with autism.

Brian Noonan, of Phoenix, found out about the drug earlier this year after asking ChatGPT for the best autism drug options for his 4-year-old son.

The FDA has never approved any drug for the underlying causes of autism, but the chatbot directed Noonan to Frye's research.

After an evaluation and confirmatory blood test, the boy started on a formulation of the drug from a compounding pharmacy in June.

Within days, Noonan says, he saw improvement in his son's ability to make eye contact and form sentences.

"He's not cured, but these are just areas of improvement," Noonan said. "It's been a big thing for us."

___

The Associated Press Health and Science Department receives support from the Howard Hughes Medical Institute's Science and Educational Media Group and the Robert Wood Johnson Foundation. The AP is solely responsible for all content.