Raising Each Other
Menu
How leucovorin works, and when it doesn't
Learn · folate, antibodies, and a drug that uses the side door

How leucovorin works, and when it doesn’t

This is the page we looked for when we first heard the word, and couldn’t find. One picture, honest about the evidence, and exactly what to ask the paediatrician. It isn’t advice; our son gets it on prescription, under his doctor’s plan, and nothing here is a dose for another child.

In three sentences

The brain doesn’t make folate; it has to import it from the blood, and for that it uses one main door where blood meets the brain’s fluid. In some children with autism the immune system makes antibodies that lock that door. Folate in the blood is normal; folate in the brain is not enough. Leucovorin is a form of folate that can get in through a second, smaller door the antibodies don’t see, if there’s enough of it.

BLOODfolate normalCHOROID PLEXUSthe wall between blood and brain fluidCSF → BRAINtoo little folateFRα: the main doorlocked by antibodies (FRAA)RFC: the side doorlets folate through only at high concentrationfolate from foodleucovorin, high doseFRAA
Figure 1. There is folate in the blood; it doesn't reach the brain because the main door is locked. High-dose folinic acid gets through the side door.

What folate is and why the brain depends on it

Folate (vitamin B9) is raw material for three things the brain does constantly: making DNA for new cells, supplying the methyl groups that switch genes on and off (via SAM), and helping make neurotransmitters. A three-year-old’s brain is building myelin (the insulation around nerve fibres) faster than it ever will again, and that work consumes folate in quantity.

Blood and brain are not one space. Between them is a wall, and folate crosses that wall only through doors with carriers.

Three doors

FRα (folate receptor alpha): the main door. It sits on the choroid plexus, the tissue inside the brain’s ventricles that produces cerebrospinal fluid. FRα catches folate from the blood with high precision at normal concentrations, pulls it into the cell and releases it into the fluid, which carries it through the brain. This is how it works in everyone.

RFC (reduced folate carrier): the side door. Found on the choroid plexus and on the brain’s blood vessels. It catches folate weakly: for anything to pass, folate in the blood has to be much higher than normal. But the antibodies don’t recognise it.

PCFT: a third carrier. It works in acidic conditions, more important in the gut than in the brain. Mentioned because it appears in the literature on “boosters”.

What goes wrong

In some people the immune system makes antibodies against their own FRα receptor. They’re called FRAA (folate receptor alpha autoantibodies). Two kinds: blocking ones sit exactly where folate should attach; binding ones attach to the side and can trigger destruction of the cell carrying the receptor.

The result is called cerebral folate deficiency (CFD): the blood folate result is fine, the fluid folate is low. It was first described in 2005 (Ramaekers, NEJM) in children with epilepsy, developmental delay and autistic features. This is why an ordinary blood test doesn’t see the problem. Serum folate is normal, and that is the biggest trap in the whole story.

Where do the antibodies come from? The leading hypothesis is cow’s milk: the protein that binds folate in milk (bovine FBP) resembles human FRα closely enough that the gut’s immune system, having made antibodies to milk, makes antibodies to the receptor too. The evidence is small but direct: in twelve children the antibody titre fell on a milk-free diet and rose when milk returned (Ramaekers, 2008). That’s why nearly every protocol that includes leucovorin also removes dairy. But twelve children are twelve children; this is a reason to try it, not proof that it’s true for everyone.

Cow's milkbovine FBP, the proteinthat binds folateGut (GALT)the immune system makesantibodies to FBPThe antibodies can't tell apartmilk FBP ≈ human FRαFRα lockedcerebral folate deficiencyno milk: antibody titre falls over months (12 children, Ramaekers 2008)
Figure 2. The leading hypothesis for where the antibodies come from. A small study, direct evidence; a reason to drop dairy, not proof for everyone.

How many children with autism have FRAA? In the most-cited series (Frye, 2013): 75% of 93 children. Other groups found much less, and the antibodies are found in some children without autism too. Take it this way: not rare, not universal. And the test that measures it is itself in question (see below).

How leucovorin gets around the block

Leucovorin is folinic acid, a reduced, stable form of folate. It is not the folic acid from the pharmacy shelf: folic acid is the oxidised synthetic form that the body still has to convert, through an enzyme that is slow in humans. Folinic acid skips that step.

The logic of the high dose is simple: if the main door is locked, the only way in is the side door, and the side door lets things through only when the blood concentration is high. That’s why studies use a per-kilogram dose far above anything “vitamin-like”. It isn’t a supplement; it’s a prescription medicine, and that’s why a doctor runs it.

Once inside the cell, folinic acid converts into the other folate forms: for DNA, for methylation (5-MTHF), for BH4, which is needed to make dopamine and serotonin. That’s why parents and doctors watch speech and interaction first: those are the functions where studies showed an effect, and they’re what’s monitored as the dose goes up.

Two things that go with it in most protocols: no dairy (so the antibody titre falls over time and the main door recovers), and vitamin D (in mice without FRα, active vitamin D boosts RFC and imports more folate into the brain (Alam, PNAS, 2019); not shown in humans, but safe and given anyway).

What the science actually says

Precision matters here, because in the past year there has been more noise around this drug than in the previous ten.

Mechanism in genetic CFD: strong. In children with a mutation in the FRα gene (FOLR1), leucovorin is standard treatment, and the FDA approved it for exactly that on 10 March 2026: cerebral folate deficiency with a confirmed FOLR1 variant. That’s an ultra-rare condition, about one in a million.

Mechanism in FRAA-positive autism: plausible, not proven. That the antibodies exist, that they’re associated with lower fluid folate, and that folinic acid raises fluid folate: that has been measured. That this is the cause of symptoms in most of those children has not.

Effect on autism symptoms: one solid study and several weak ones. The best: Frye et al., Molecular Psychiatry, 2018, with 48 children with autism and language impairment, double-blind, placebo, 12 weeks. Verbal communication improved with a medium-to-large effect (d = 0.70), larger in FRAA-positive children (d = 0.91), with no difference in adverse effects from placebo. Alongside it: a small French trial (EFFET, 19 children) and two more small ones. The largest study (77 children, India, 2024) was retracted in January 2026 over data errors nobody could reproduce. Four small randomised trials remain, and in March 2026 the FDA said explicitly that for autism the evidence is insufficient.

Honestly: that means leucovorin in autism is neither a “miracle” nor a “scam”. It’s a drug with a reasonable mechanism, one good small trial, and no large trial to confirm or overturn it. A parent considering it is doing so on logic and one study, not on consensus. And, since we first published this page, against the stated advice of three professional bodies. The next section is about that. We are one of those parents, so we are not writing it from a distance.

Safety. In the trials, no difference from placebo. In practice: hyperactivity, irritability, worse sleep in some children, which is why the dose is raised gradually and lowered if that happens. Behavioural worsening has been described in children taking risperidone. It isn’t stopped abruptly.

What changed since we published this

We published this page on 11 March 2026. Two things have changed since, and one of them means part of our advice was wrong. We would rather say so here than quietly edit it out.

We told you to ask about the FRAT test. We no longer think you should lean on it. In March 2026 the Child Neurology Society stated that folate receptor autoantibody testing “is not recommended as a basis for clinical decision-making”, warning that the current radioligand method may produce a high false-positive rate through non-specific binding. In February 2026 the American Academy of Pediatrics said plainly that there are no FDA-cleared or approved tests for these antibodies, and told clinicians to avoid over-reliance on the results. A positive result may mean considerably less than it appears to.

It is worth knowing who sells it. The test costs around US$295, is not covered by insurance, and is sold commercially under licence from the laboratory of a researcher who also co-authored the foundational papers arguing that these antibodies matter in autism. Being CLIA-certified means the laboratory is regulated; it does not mean the test’s accuracy has been validated by the FDA. None of this proves the idea is wrong. You simply deserve to know it before you pay.

Three professional bodies have now said the same thing. The Child Neurology Society: leucovorin “should not be prescribed as standard of care for individuals with autism”. The American Academy of Pediatrics, interim guidance, February 2026. The American College of Medical Toxicology, August 2026, which described the FDA approval as “a departure from evidentiary standards”. All three point at genetic testing first: chromosomal microarray, fragile X, and exome sequencing where indicated. That is where the provable, sometimes treatable things are found.

One safety point we did not make strongly enough. High doses of folinic acid can mask the blood signs of a vitamin B12 deficiency while nerve damage carries on underneath. If a child is on a high dose, B12 should be checked.

None of this changes the mechanism sections above; the biology is the same as it was. What changed is how confident anyone should be, and what we now think is worth asking for.

What to ask the paediatrician

This is the part the page exists for. You don’t need to know more than this:

  1. “Given that FRAT isn’t recommended, what would you look at instead?” We used to suggest asking for the antibody test first. We don’t any more (see the section above). The professional bodies point at genetic testing, which is where a rare but genuinely treatable cause like a FOLR1 mutation would actually show up.
  2. “If we try folinic acid anyway, how will we know whether it’s working, and when do we stop?” Ask before starting: what exactly is being watched (speech, interaction), over how many weeks, and what result means stop. Agreeing the stopping rule in advance is the best protection there is against fooling yourself. Children change on their own, and hope is a powerful lens.
  3. “Should we try a month without dairy before anything else?” Cheap, safe, and part of the hypothesis.
  4. “What else do we check?” For our son the first things were a blood count, iron, vitamin D, genetics (fragile X, microarray) and hearing, to rule out what’s more common.

If the paediatrician doesn’t know about this, that isn’t a reason for a fight. The mechanism was described in 2005, it isn’t in the textbooks, and most paediatricians meet it for the first time when a parent brings it in. Print the page. Bring the 2018 study.

Print it: the questions on one page, with room for notes (PDF, A4).

When not

  • Not without a doctor. Don’t buy “leucovorin” or “folinic acid” online and dose it yourself. The dose is per kilogram, raised gradually, and someone has to watch the child.
  • Folic acid is not folinic acid. Large doses of ordinary folic acid don’t do the same thing.
  • High doses of either can hide a B12 deficiency, normalising the blood picture while nerve damage continues underneath. Get B12 checked.
  • If the child has epilepsy or takes psychiatric medication, that is a separate conversation, especially around risperidone.
  • If the child gets worse on leucovorin (hyperactivity, sleep, behaviour), that’s information, not failure: dose down, and tell the doctor.
  • It isn’t for every child with autism. The professional bodies do not recommend it as standard treatment, and no reliable test tells you in advance whom it would help.

Our case, briefly

Maxi started folinic acid in August 2025, as drops, raised gradually; capsules from January; from May 2026 on a lower dose, which held. Dairy- and gluten-free from the start. First finger pointing: April 2026. The whole record, with dates and results, is on the journey page. As far as we know, he never had the FRAT test. When we first wrote this page we called that an open question we were carrying to the next appointment. Since then the Child Neurology Society has said the test isn’t a sound basis for decisions, so we have stopped chasing it. We are asking about genetic testing instead. This is what changing your mind in public looks like.

Sources

First published 11 March 2026. Revised and re-verified 13 September 2026. Two sentences that still leaned a decision on the FRAT corrected 21 September 2026.