A study says people eating the most 'plant-type' protein had 63% lower odds of epilepsy. They also ate about 1,300 more calories a day than the people they were compared with, and the paper's summary reports differences its own tables do not show

Protein is built from about twenty amino acids, and some of them double as raw material for the brain's chemical messengers. That is the reason researchers have wondered whether the mix of amino acids in a diet has anything to do with seizures. A paper published online in Experimental Physiology on 2 October 2026 says it is the first to look at this across a whole population, and its headline result is a big one: 63% lower odds of having epilepsy among people who followed a 'plant amino acid pattern' most closely.

The data come from Sabzevar, a city in north-east Iran, where one neighbourhood is part of a national health study called the PERSIAN cohort. Between 2017 and 2020, adults aged 35 to 70 were weighed, examined and interviewed about how often they had eaten each of 115 foods over the previous year. Of the 4,063 people analysed, 112 already had epilepsy diagnosed by a specialist. The researchers converted each person's food list into grams of 18 amino acids and let a statistical method sort those into two bundles that tend to rise and fall together. One bundle, led by lysine, they called the animal pattern. The other, led by proline and glutamic acid, they called the plant pattern.

Everyone was then ranked into four equal groups on each pattern. Compared with the bottom quarter on the plant pattern, the top quarter had odds of epilepsy of 0.37, which is where the 63% comes from, after adjusting for age, sex, education, body mass index, diabetes, high blood pressure, marital status, employment and cancer. The animal pattern pointed the same way, at 0.62, but that result could have been chance.

Now look at who was in those quarters. People at the top of the plant pattern reported eating about 3,250 calories a day and 555 g of carbohydrate; people at the bottom reported about 1,940 calories and 304 g. Protein went from 61 g to 103 g. Nearly three in four of the top quarter were men, against fewer than one in four at the bottom, and almost two thirds had a job, against under a third. Every one of the 18 amino acids rose from the bottom quarter to the top on both patterns. So the 'plant pattern' is not a kind of protein that some people chose over another. It is, to a large degree, a measure of how much someone eats, and the model behind the 63% did not adjust for calories.

The names deserve a second look too. The paper does not show which foods supplied each pattern. The two amino acids that lead the plant bundle, proline and glutamic acid, are the two that wheat protein is famously rich in, and bread is the staple of the Iranian table. That is our reading, not the authors', but a pattern that climbs with carbohydrate and is led by wheat's signature amino acids looks more like 'eats a lot of bread' than like beans and lentils.

Then there is the order of events. Everyone with epilepsy had it before the food interview, and the questionnaire covered only the previous year. The study cannot tell whether eating less came first or the illness did. In this sample, people with epilepsy were more likely to have had a stroke (6 of 112, against about 1 in 100 of the others), to have diabetes or a history of cancer, and they were a little less active. People who are unwell often eat less. Stroke, which is itself a known cause of epilepsy, was not in the model. Neither were anti-seizure medicines, seizure frequency, the type of epilepsy or how long people had had it, because the cohort did not record them. The authors list these gaps themselves.

The summary at the top of the paper does not match the tables beneath it. It says participants with epilepsy were more likely to be male, less educated, married and unemployed, and ate less of every amino acid, and the discussion calls that last gap significant. In the tables, none of these differences passes the paper's own test for chance: the amino acid comparisons run from P = 0.11 to P = 0.35, and total protein was 76 g a day against 79 g. The only dietary difference that did pass was vitamin D intake. The steps are uneven as well: the second quarter on the plant pattern had odds of 0.59, the third went back up to 0.81, and only the fourth dropped to 0.37. The authors disclose that they used an AI chatbot to edit the text of the manuscript.

None of this means diet is irrelevant to epilepsy. Medically supervised ketogenic diets are an established treatment for some people whose seizures do not respond to medicines, and that is a different thing from a food questionnaire in a general population. What this study shows is narrower: in one Iranian town, the 112 adults with epilepsy were found more often among the people who reported eating least. The authors say clearly that a study like this cannot show cause, and then go on to suggest that amino acid advice could become part of epilepsy prevention. On these numbers that is a long way ahead of the evidence. Nobody should change how they eat, or how they take their medicines, because of it.