For people who carry two copies of the APOE4 gene, the message from clinical medicine over the past three decades has been blunt: your risk of developing Alzheimer’s disease is dramatically elevated, the elevation is genetic, and there is not much you can do about it. About 2 percent of the human population falls into this category. They have inherited one copy of the APOE4 variant from each parent, which combine to produce a 12-fold higher lifetime risk of Alzheimer’s disease compared with people who carry no APOE4 copies. The dogma, until recently, was that this risk was essentially built in. Lifestyle changes might shift the odds at the margins. Nothing was thought to substantially overcome the genetic loading. In August 2025, a research team led by investigators at Mass General Brigham, the Harvard T.H. Chan School of Public Health, and the Broad Institute published a paper in Nature Medicine suggesting that the dogma was wrong — or at least incomplete — and that a specific dietary pattern can substantially modulate even the highest-risk genetic profile.
According to the Mass General Brigham press release accompanying the study, the research team analysed data from approximately 5,700 participants in two long-running US cohorts — the Nurses’ Health Study (4,215 women, followed from 1989 to 2023) and the Health Professionals Follow-up Study (1,490 men, followed from 1993 to 2023). Combined, the data covered more than three decades of dietary records, genetic information, blood metabolomic profiles, and tracking of dementia outcomes across the participants’ lives. The study population was large enough, and the follow-up long enough, to provide statistically meaningful comparisons across genetic-risk subgroups — including the small but critical APOE4 homozygous subgroup, which most previous dietary studies had been too small to analyse on its own.
What the data showed
The headline finding was specific and striking. Per CNN Health’s coverage of the study, participants who most closely followed a Mediterranean dietary pattern — high in vegetables, fruits, nuts, whole grains, legumes, fish, and olive oil; low in red meat, processed meat, and alcohol — showed reduced dementia risk across all genetic-risk categories. The protective effect was modest in people with no APOE4 copies. It was somewhat larger in people with one APOE4 copy. And it was substantially largest in people with two APOE4 copies, in whom close Mediterranean diet adherence was associated with at least a 35 percent reduction in dementia risk compared with APOE4 homozygotes who did not follow the dietary pattern. The protective effect, in other words, scaled with the genetic risk it was working against — the diet helped most in the people who needed it most.
Yuxi Liu, the study’s lead author and a research fellow at Brigham and Women’s Hospital and Harvard Medical School, summarised the finding for CNN: “We followed over 5,700 people for 34 years and found those who followed a baseline Mediterranean diet with little alcohol, red and processed meat but full of vegetables, fruits, nuts, whole grains, legumes, fish and olive oil all reduced the risk of dementia. But the benefit was highest for people with the APOE4 gene, especially those with two copies of APOE4.” Richard Isaacson, an Alzheimer’s specialist not involved in the research, framed the implications more directly: “Hopefully this will change the current dogma that having two copies of the APOE4 is a ‘fait accompli’ for developing Alzheimer’s. Genes do not have to be your destiny.”
The metabolomic mechanism
What distinguishes the 2025 study from earlier dietary research on Alzheimer’s is that the team also measured the participants’ blood metabolomes — the full set of small molecules circulating in their plasma — and tracked how the metabolite profiles varied by genotype and by diet. As documented in the original paper by Yuxi Liu and colleagues in Nature Medicine, the integrated analysis identified 57 specific metabolites whose associations with dementia risk varied by APOE4 genotype or by other Alzheimer’s-related genetic variants. Cholesteryl esters and sphingomyelins, for example, were most strongly associated with increased dementia risk specifically in APOE4 homozygotes. Inverse associations with glycerides — meaning higher glyceride levels were associated with lower dementia risk — were also specific to the APOE4 homozygous genotype.
The Mediterranean diet, in the data, was unusually effective at modulating exactly the metabolites that mattered most in APOE4 homozygotes. The diet pushed cholesteryl ester and sphingomyelin levels downward (reducing dementia-associated metabolites) while supporting glyceride profiles associated with protection. The mechanism is not yet fully characterised, but the integrated metabolomic data provide what the authors call “biological plausibility” for the observed dietary effect — meaning the diet appears to be working through measurable biochemical pathways rather than through some unspecified general health benefit. The combination of long-term outcome data and detailed metabolomic profiling makes the 2025 finding substantially more mechanistically grounded than earlier dietary observational studies that established correlation without identifying the underlying biochemistry.
Why the APOE4 finding matters
The clinical significance of the APOE4 result is hard to overstate. Alzheimer’s disease has a substantial genetic component — heritability has been estimated at up to 80 percent — and APOE is by some distance the strongest single genetic risk factor for the common sporadic form of the disease. The 12-fold elevated risk that comes with carrying two APOE4 copies is among the largest known gene-disease effect sizes in medicine. For decades, the consensus framing was that this genetic load was essentially fixed. Patients who learned through genetic testing that they were APOE4 homozygotes typically received the news as a sentence rather than a probability — something to brace against rather than something to act on.
The 2025 finding does not, by itself, eliminate the elevated risk. A 35 percent reduction in dementia risk among APOE4 homozygotes following a Mediterranean diet still leaves these individuals at substantially elevated risk compared with the general population. What the finding does establish is that the risk is modifiable — that a daily food pattern, sustained over decades, can meaningfully shift the trajectory in a population that previous research had largely written off. As reported in Harvard Health Publishing’s coverage of the study, the practical recommendation for APOE4 carriers is not radically different from what cardiologists and neurologists have been telling everyone for decades — eat more vegetables, more fish, more olive oil, less red meat, less processed food — but the new evidence suggests that for the APOE4 homozygous subgroup specifically, the recommendation may carry substantially more weight than it does for the general population.
Limitations and what comes next
The Liu et al. 2025 paper is an observational study rather than a randomised controlled trial, and the authors acknowledged several limitations that warrant honest framing. The cohort consisted of well-educated individuals of European ancestry, which means the generalisability to other populations — different genetic backgrounds, different food cultures, different socioeconomic contexts — is not yet established. The dietary adherence was self-reported through periodic food-frequency questionnaires rather than objectively measured. The participants who closely followed a Mediterranean diet may have differed from non-followers in unmeasured ways that could partly account for the dementia difference. The 2025 finding is not, on its own, proof that Mediterranean diet directly causes reduced dementia risk in APOE4 homozygotes; it is strong observational evidence consistent with that conclusion.
What the paper adds to the existing literature is the genetic-stratified analysis and the metabolomic mechanism, both of which provide convergent support for the broader claim. Earlier work — the 2023 brain-autopsy study showing 40 percent lower odds of Alzheimer’s pathology in Mediterranean and MIND diet adherents, the FINGER multi-domain lifestyle trial in Finland, the long line of MIND diet research at Rush University — had built the case that diet matters for dementia risk in general populations. The 2025 paper extends the case specifically to APOE4 homozygotes, who had been the most pessimistically counselled subgroup and now appear to be the subgroup with the most to gain from dietary intervention. Whether this translates into changed clinical recommendations, changed genetic-counselling practice, and changed daily behaviour in the 2 percent of the population carrying two APOE4 copies will be the central question for the next several years of Alzheimer’s prevention research. The biological case for the diet has now been made. The cultural case for actually following it is, as always, harder.
