Two New Studies Point to Positive Benefits of HRT for Dementia Prevention

Ask ten physicians whether menopausal hormone therapy protects the brain and you will get a split room. Some will point to the Women’s Health Initiative and tell you it raises dementia risk. Others will tell you that trial studied the wrong women at the wrong age with the wrong hormones and that the picture looks different when estrogen is started near menopause. Both groups are reading the same literature. The disagreement has lasted more than twenty years because the studies point in different directions depending on who was treated, when, and with what.

The biological rationale for estrogen as protection. Women develop Alzheimer’s disease at about twice the rate of men, and roughly two thirds of people living with the disease are women. Longer female lifespan explains part of that gap but not all of it. The main event that separates women’s biology from men’s in midlife is menopause, when estrogen production from the ovaries drops sharply over a few years. Estrogen acts directly on neurons. It supports the growth and maintenance of synapses, the connections between brain cells, helps brain cells take up and use glucose for energy, restrains inflammation, and supports blood flow through the small vessels of the brain. Laboratory studies have also shown that estrogen influences how the brain processes amyloid, the protein that accumulates in Alzheimer’s disease. Put together, the reasoning was that the loss of estrogen at menopause removes a form of protection the female brain had relied on for decades, and that replacing it might restore that protection.

The history in brief. Through the 1990s, observational studies of women who chose to take estrogen reported lower rates of Alzheimer’s disease, in the range of 30 to 40 percent lower. Then in 2003 the Women’s Health Initiative Memory Study, a randomized trial of conjugated equine estrogen plus a synthetic progestin in 4,532 women aged 65 and older, found that the treated group developed dementia at roughly twice the rate of placebo. In absolute terms that was 45 cases per 10,000 women per year on treatment compared with 22 on placebo, so an extra 23 cases per 10,000 women each year. Hormone therapy prescriptions collapsed. Ten years later the same investigators reported the WHIMSY trial, which followed 1,326 women who had started the same hormones between ages 50 and 55, and found no difference in cognition either way about seven years after treatment ended. A Danish registry study in 2023 reported a 24 percent higher rate of dementia in women who had used estrogen plus a progestin, including women treated at 55 or younger, which would move a woman’s lifetime dementia risk from roughly 48 percent to about 55 percent if the association were causal. Then in December 2025 a World Health Organization commissioned meta-analysis in The Lancet Healthy Longevity pooled the ten studies its authors judged rigorous enough, covering just over one million women, and found no association with dementia in either direction. The WHO currently advises against prescribing hormone therapy to prevent dementia and the FDA removed its dementia warning from hormone therapy labels in November 2025. That is where things stood at the start of this summer.

Two large studies published in August report lower Alzheimer’s risk with hormone therapy

The Stanford autopsy study measured Alzheimer’s pathology directly in brain tissue. On August 12, a team led by Dr. Hadi Hosseini at Stanford published in Neurology an analysis of women in two large research databases, the National Alzheimer’s Coordinating Center and the Alzheimer’s Disease Neuroimaging Initiative. The autopsy comparison included 258 women who had used estrogen-only hormone therapy and 2,701 who had never used any hormone therapy. Pathologists score autopsied brains for the amount and spread of amyloid plaques and tau tangles, the two protein deposits that define Alzheimer’s disease. Women who had used estrogen-only therapy had 35 percent lower odds of Alzheimer’s pathology in their brain tissue. They also had lower odds of having received a clinical dementia diagnosis during life, lower rates of blood vessel damage in the brain, and more favorable blood and spinal fluid amyloid markers in the second cohort. Estrogen-only therapy is generally prescribed to women who have had a hysterectomy, so this is a specific population and the study cannot say anything about the more common estrogen plus progesterone combination. Most of these women would also have been treated decades ago with older formulations. The authors describe the effect as modest and stress that an observational study cannot prove cause and effect. What the study adds is that the protective signal appears in the physical hallmarks of the disease, whereas earlier studies relied on clinical labels, which are known to be inaccurate.

The UK Biobank study identified which women seemed to benefit. On August 26, Dr. Anne Marie Minihane and Dr. David Llewellyn published in Alzheimer’s & Dementia an analysis of 183,450 postmenopausal women followed for an average of 13.3 years, during which 3,948 developed dementia. Women who had used hormone therapy for at least one year had 10 percent lower all-cause dementia risk and 16 percent lower Alzheimer’s risk than never users. Using the roughly 48 percent lifetime dementia risk that a 2025 study estimated for American women after age 55, a 10 percent relative reduction would move a woman from 48 percent to about 43 percent. The associations were larger in particular groups. Women who went through surgical menopause after removal of the uterus or both ovaries had a 26 percent lower dementia risk and a 32 percent lower Alzheimer’s risk, which as an illustration would move lifetime dementia risk from about 48 percent to about 36 percent. Women carrying the APOE4 gene variant, the most common genetic risk factor for Alzheimer’s disease, had a 13 percent lower risk, and women whose bodies produced less estrogen over their lifetime because they started their periods late or reached menopause early had a 16 percent lower risk. For women outside those groups the study found little benefit and no harm. Age at initiation mattered. The association with lower risk was concentrated in women who started hormone therapy between 46 and 56, meaning during perimenopause or the first few years after menopause. The study could not separate pills from patches or different doses. As Dr. Aimee Spector of University College London pointed out in the Science Media Centre reaction, women who take hormone therapy tend to have fewer of the other known dementia risk factors such as lower education and poorer physical health.

Timing, formulation, and the woman’s starting risk explain most of the disagreement

Timing explains most of the apparent contradiction. The Women’s Health Initiative treated women who were on average more than a decade past menopause. The UK Biobank benefit was confined to women who started before 56. The WHIMSY women who started at 50 to 55 showed no harm. A 2023 imaging study from Massachusetts General Hospital found that women who began hormone therapy more than five years after menopause had higher tau levels on PET scans while women who started earlier did not. A 2025 follow-up from the same group found that hormone therapy use predicted faster tau accumulation in women over 70 but not in younger women. The one randomized trial built to test timing directly, ELITE, assigned 567 healthy women who were either within six years of menopause or ten or more years past it to oral estradiol with vaginal micronized progesterone or placebo, and after about five years found no cognitive difference from placebo in either group. That trial measured cognitive test scores over a limited period and so cannot rule out effects on dementia, but it did not find the late-initiation harm that the Women’s Health Initiative reported with conjugated equine estrogen and medroxyprogesterone acetate.

One hypothesis for the pattern, which has not been tested directly, is that estrogen given to a brain that has already spent years without it behaves differently from estrogen that maintains continuity through the transition. The proposed biology is that estrogen supports how brain cells use energy, restrains inflammation, and helps maintain blood flow, and that the abrupt loss of it at menopause is a stress the brain adapts to over time. Reintroducing estrogen after that adaptation, particularly when blood vessels have aged, might add vascular stress. Whether that is what happened in the Women’s Health Initiative, or whether the older formulation explains the result, is unknown.

Formulation is the second variable. The Women’s Health Initiative tested conjugated equine estrogen with the synthetic progestin medroxyprogesterone acetate. The Danish registry data reflect that country’s prescribing, which is mostly oral estradiol with the synthetic progestin norethisterone. The Stanford autopsy signal came from estrogen alone. The KEEPS trial, which randomized women within three years of their final period to oral conjugated estrogen, a transdermal estradiol patch, or placebo, with micronized progesterone in both hormone arms, found no cognitive harm and no cognitive benefit at four years and again about ten years later. That trial is reassuring on safety but was small, ran for four years, and measured cognitive test scores in women who were still young enough that dementia would have been rare regardless.

The woman’s starting point is the third variable. The UK Biobank pattern is that women with the least estrogen exposure, whether from surgical menopause, early menopause, or late menarche, and women with the highest genetic risk gained the most. This matches an earlier European Prevention of Alzheimer’s Disease cohort analysis from Dr. Minihane’s group, which found that APOE4 carriers on hormone therapy had better delayed memory scores and larger volumes in the entorhinal cortex and amygdala, two of the first brain regions Alzheimer’s disease damages. For a woman with an average hormonal history and no APOE4 copy, the newer data suggest hormone therapy started near menopause is neutral for dementia risk. For a woman with surgical menopause or an APOE4 copy, the same data suggest a meaningful reduction.

The Danish registry study is the main finding this frame does not account for. It reported a higher dementia rate in women treated with estrogen plus a progestin at 55 or younger, inside the window the timing hypothesis calls safe. Two features of that study reduce the weight it carries. In an accompanying editorial, Dr. Kejal Kantarci and Dr. JoAnn Manson pointed out that the excess risk was present even in women who used hormones for a year or less, a pattern that suggests the women seeking treatment differed from those who did not, since one year of estrogen is unlikely to cause dementia decades later. The same study found no association for estrogen-only therapy or progestin-only therapy. The result remains an unexplained signal in the combined-therapy group.

The null meta-analysis and the two new positive studies are not in conflict. The Lancet Healthy Longevity review pooled studies with different formulations, ages at initiation, and populations, and rated the certainty of most of the evidence as low or very low. Averaging a benefit in some women with a harm in others produces a null. The two August studies were built to look inside that average, one by examining tissue and one by splitting women into subgroups. Both found the benefit was concentrated where the timing hypothesis predicts it should be.

The evidence still lacks a randomized trial designed to answer the question

No trial has tested modern hormone therapy against dementia as the outcome. Every positive finding above is observational. Observational studies of hormone therapy have a specific weakness. Women who are prescribed it tend to be healthier, wealthier, and more engaged with medical care. Those traits lower dementia risk on their own. The only randomized dementia data come from the Women’s Health Initiative, which tested older hormones in older women. A trial that recruits women in perimenopause, randomizes them to transdermal estradiol with micronized progesterone or to placebo, and follows them for dementia would settle the question.

Blood biomarkers could shorten the wait. Blood tests for phosphorylated tau and amyloid now detect Alzheimer’s pathology many years before symptoms. A randomized trial that used change in blood p-tau217 over five to seven years as its primary endpoint would be feasible in size and duration. Dr. Minihane has said she hopes exactly this kind of test will make it possible to determine whether starting hormone therapy earlier keeps the brain healthier for longer. The KEEPS continuation study is also collecting dementia and imaging outcomes from its original participants as they age into the years when Alzheimer’s becomes common.

The specific subgroups need their own trials. The largest signals in the UK Biobank data came from surgical menopause and APOE4 carriers. A trial that enrolled women by APOE genotype, as Dr. Minihane’s group proposed in 2023, would test the hypothesis in the population most likely to show an effect and most in need of an option. Women who lose their ovaries before natural menopause are a second population where the biology is clearest and the ethical case for treatment is already strong, since those women face higher rates of heart disease and bone loss without estrogen replacement.

Late initiation with modern hormones needs its own trial. The evidence that starting hormone therapy more than ten years after menopause harms the brain comes from the Women’s Health Initiative formulation and from observational imaging of women treated decades ago. ELITE found no cognitive harm from oral estradiol started ten or more years after menopause over five years. A trial of transdermal estradiol with micronized progesterone in women in their sixties, with brain biomarkers as the endpoints, would show whether the timing window is a property of estrogen itself or of the older hormones that were tested. It may turn out that older women can start safely.

Formulation and route remain untested for brain outcomes. Nearly all the dementia data involve oral conjugated estrogen. Whether transdermal estradiol, which avoids the liver and carries lower clot risk, produces a different brain outcome is unknown. Whether the choice of progestogen matters is also unknown, though micronized progesterone and synthetic progestins act differently in the brain and the Stanford finding in estrogen-only users hints that the progestin component deserves attention.

My take

Dementia prevention is one of the reasons I take hormone therapy.

The evidence has limits. There is no randomized trial showing that modern hormone therapy prevents dementia. What the evidence does show, consistently across the Women’s Health Initiative, WHIMSY, KEEPS, the imaging studies, and now the UK Biobank and Stanford data, is that hormone therapy started within the first few years of menopause does not harm the brain and may help it. The harm seen with later initiation comes from older formulations in older women, and whether modern estradiol started late carries the same risk has not been established. That asymmetry matters for a decision that has to be made in real time. A woman in her late forties or early fifties cannot wait for the trial. The window in which hormone therapy appears safe for the brain is the same window in which she is deciding.

There are additional reasons to take HRT. Hormone therapy is the most effective treatment we have for hot flashes and night sweats, it protects bone, and for me it resolved a sleep problem that had not responded to anything else. I wrote about that in my article on sleep and the brain. Poor sleep is itself a dementia risk factor. It also lifted my mood and gave me back the energy to exercise consistently. Both of those feed back into brain health. Perimenopause raises the risk of depression and anxiety, estrogen loss drains energy in a way many women mistake for aging or overwork, and depression and physical inactivity are each established dementia risk factors on their own. So the brain case for hormone therapy in early menopause runs through several channels at once, better sleep, better mood, more energy, better vascular health, and whatever direct effect estrogen has on amyloid and tau. The new studies strengthen the last of those without changing the others.

My APOE4 allele is another reason I take it. APOE4 carriers were one of the groups with a larger benefit in the UK Biobank data.

Hormone therapy is not for everyone. A history of breast cancer or a clotting disorder changes the calculation. The decision belongs in a conversation with a physician who knows the current data. My own view, after twenty years of studying how the brain ages and a few years of taking estrogen myself, is that for a woman near menopause without those contraindications the brain evidence has now shifted from a reason to hesitate to a reason to consider it. The field owes women the trial that would turn that consideration into an answer.