Scientists map a blood protein fingerprint of menopause tied to brain health

A multi-cohort study indicates that the transition through menopause is accompanied by distinct shifts in blood proteins tied to inflammation, nerve cell connections, and brain aging. These molecular changes track more closely with reproductive hormone levels than with chronological age, and older women whose blood shows more of this protein pattern tend to experience faster cognitive decline and a modestly higher risk of Alzheimer’s disease. The findings, published in Nature Medicine, suggest that blood-based markers could help identify biological pathways linking reproductive aging to long-term brain health.

Menopause is a midlife transition for people with ovaries, occurring at an average age of 51, when ovarian reproductive function ends and menstrual cycles cease. It involves a steep drop in ovarian hormones like 17-beta-estradiol (the primary reproductive estrogen) and progesterone, alongside a sustained rise in follicle-stimulating hormone, a pituitary gland messenger. Although dementia symptoms typically emerge in the mid-70s, Alzheimer’s brain pathology begins decades earlier, coinciding with midlife. Many menopause symptoms, including hot flashes, night sweats, sleep problems, and subtle memory difficulties, originate in the central nervous system.

Scientists have increasingly examined whether these hormonal and immune shifts leave the brain vulnerable to neurodegeneration. The authors of a 2020 review in Journal of Neuroinflammation argued that perimenopause acts as a systemic inflammatory phase because estrogen loss disrupts immune regulation and cellular energy production in the brain.

In addition, a 2023 study of midlife women in the American Journal of Obstetrics and Gynecology found that more frequent hot flashes during sleep, recorded with skin monitors, were linked to a lower ratio of two amyloid proteins in the blood, a pattern suggestive of more amyloid in the brain. A 2025 clinical and autopsy study by four of the same authors in Science Advances indicated that earlier menopause strengthened the link between degraded synapses (nerve cell communication junctions) and both tau tangles and cognitive decline.

To map the broader molecular network behind this transition, lead author Madeline Wood Alexander, a doctoral candidate in the Hurvitz Brain Sciences Program at Sunnybrook Research Institute and the Temerty Faculty of Medicine at the University of Toronto, conducted the study with co-senior authors Rowan Saloner, an assistant professor of neurology, and Kaitlin B. Casaletto, an associate professor of neurology. Both Saloner and Casaletto are based at the Edward and Pearl Fein Memory and Aging Center in the Weill Institute for Neurosciences at the University of California, San Francisco, and all three researchers are affiliated with the Ann S. Bowers Women’s Brain Health Initiative at the University of California.

Free daily newsletter

“Women are more likely than men to develop dementia, especially Alzheimer’s disease, but scientists still do not understand the reasons for this sex difference,” Wood Alexander, Saloner, and Casaletto told PsyPost in a joint response. “One contributing factor may be menopause.” Pointing to earlier research, they added, “Other studies (including work from our team) have shown that factors such as earlier age at menopause or more severe menopause symptoms are associated with greater risk for dementia in later life.”

“All this evidence positions menopause as a key transition for the brain that may be relevant for women’s dementia risk, but very limited research has investigated the brain-relevant biology of the menopause transition in humans,” the researchers explained. “This is a major gap because understanding what happens to the brain during menopause could help us identify which women may be at greater risk of neurodegenerative disease later in life, and ultimately, when and how we can best intervene to protect brain health. Our study aimed to start to address this gap by looking at blood levels of molecules related to brain health in a unique cohort of midlife women with hormone levels and detailed menopause staging.”

Using blood proteomics, the large-scale measurement of proteins circulating in the bloodstream, the researchers first examined 80 healthy women aged 43 to 58 years and 36 age-matched men from the University of California, Santa Barbara Healthy Aging Study. Based on menstrual cycle regularity and hormone levels, the team classified the women into 30 premenopausal, 26 perimenopausal, and 24 postmenopausal participants, all with an intact uterus, at least one ovary, and no current hormone therapy use. Fasted morning blood serum was tested for six sex hormones and analyzed on an ultrasensitive assay measuring 118 neurological, inflammatory, and metabolic proteins. Perimenopausal and postmenopausal women also reported recent hot flashes, night sweats, vaginal dryness, and irritability.

Adjusting for age, 16 of the 118 proteins were elevated in postmenopausal compared to premenopausal women at an uncorrected statistical threshold, spanning eight inflammatory proteins, four synaptic proteins, two metabolic proteins, and two Alzheimer’s-related proteins: beta-secretase 1 (an amyloid-producing enzyme) and phosphorylated tau-231 (a tangle-related protein). Other standard Alzheimer’s markers, including phosphorylated tau-181 and amyloid-beta 42, did not show statistically significant differences after age adjustment.

To reduce noise and multiple comparisons, the researchers combined the 16 elevated proteins into a composite menopause proteomic score, which increased stepwise across the transition, with perimenopausal women scoring 1.36 standard deviations higher and postmenopausal women scoring 3.25 standard deviations higher than premenopausal women in unadjusted comparisons.

Chronological age initially correlated with higher menopause proteomic scores in women, but this link was no longer statistically significant and reversed direction after adjusting for menopause stage, and the protein differences held even when the sample was restricted to 54 women aged 47 to 53. Among the 36 age-matched men, chronological age did not correlate with the score, and men overall had higher composite scores than women after adjusting for age.

In women, when estradiol, follicle-stimulating hormone, progesterone, and age were modeled together, only follicle-stimulating hormone remained a statistically significant predictor of the overall score. Regression analyses of individual proteins showed that lower estradiol tracked with four inflammatory proteins and higher follicle-stimulating hormone tracked with five proteins spanning synaptic, metabolic, Alzheimer’s-related, and inflammatory biology, and a network analysis showed these two sets clustering separately.

Among perimenopausal and postmenopausal women, those reporting night sweats showed a borderline association with higher protein scores and had elevated levels of four inflammatory markers, including CCL2 (also called MCP-1), an immune signaling protein that draws white blood cells toward inflammation. Vaginal dryness was unexpectedly associated with lower scores, a finding whose meaning the authors say is unclear, while links with hot flashes and irritability were not statistically significant.

To test whether these serum shifts replicated on a broader plasma platform and to reduce age confounding, the team turned to the UK Biobank, a health database of about 500,000 British adults. They matched 1,407 pre- or perimenopausal women one-to-one with 1,407 postmenopausal women of nearly identical ages (45 to 60 years), excluding those with prior hysterectomy, bilateral ovary removal, or current hormone therapy. Across 2,923 plasma proteins, 1,300 (44%) differed by menopause status (1,146 elevated and 154 reduced after menopause), nine of 13 overlapping discovery proteins replicated, and the menopause proteomic score remained higher in postmenopausal women.

In 2,340 of these age-matched women, the team also applied proteomic aging clocks, models that estimate an organ’s or cell type’s age from the proteins in blood relative to a person’s chronological age. Compared to pre- and perimenopausal women of the same chronological age, postmenopausal women had older estimated ages across 11 of 13 organs and 36 of 38 cell types. Beyond reproductive cells, the largest gaps appeared in the arteries, intestines, overall brain, and oligodendrocyte precursor cells, which mature into the brain cells that produce myelin insulation around nerve fibers.

“It is important to acknowledge that while we saw increases in molecules reflecting Alzheimer’s disease biological processes in postmenopausal women relative to pre/perimenopausal women, these are subthreshold elevations that do not mean these women have Alzheimer’s disease,” the researchers noted. “At this stage we need much more research before we can understand the clinical implications of these elevations in healthy midlife women.”

“However, what we can say is that we observed widespread molecular shifts across menopause that suggest accelerated biological aging across many organ systems, consistent with other recent studies in this area,” the authors added. “This supports that menopause is an important inflection point in the aging process that may contribute to increases in risk for age-related diseases among postmenopausal women. However, there is likely a lot of variability between women in the size or practical significance of these effects, and we need more work before we can predict what these changes mean for an individual woman’s health or disease risk.”

To test whether the same protein pattern relates to cognition later in life, the team applied the score to blood samples from four cohorts of older women totaling 11,925 participants, whose blood was drawn years to decades after menopause, at average ages of 61 to 72. They analyzed 89 women (mean age 69.2 years) with retrospective menopause symptom reports and 93 women with cross-sectional cognitive tests from the Wisconsin Registry for Alzheimer’s Prevention, a cohort enriched for people who had a parent with Alzheimer’s disease (64.5% amyloid-positive).

They also tracked longitudinal cognition over roughly five years in 673 women from the Alzheimer’s Disease Neuroimaging Initiative (mean age 71.5 years; 54.3% with mild cognitive impairment or dementia; no menopause data collected) and 100 women from the University of California, San Francisco Brain Aging Network for Cognitive Health (mean age 72.1 years; 1% with mild cognitive impairment). Finally, they followed 11,059 dementia-free women aged 50 and older (mean age 60.7 years) in the UK Biobank for an average of 15.7 years.

In the Wisconsin cohort, after adjusting for age, hormone therapy, and ovary removal, women who reported a history of hot flashes, sleep problems, or a higher total symptom count had higher menopause proteomic scores years to decades later, with CCL2 linked to temperature-regulation symptoms in both midlife and older cohorts. Unlike in the midlife cohort, however, past night sweats were not associated with the score among these older women, a mismatch the authors suggest may stem from recall bias in self-reported symptoms.

Adjusting for age and education, higher protein scores were also associated with lower cross-sectional cognitive performance in the Wisconsin cohort and faster cognitive decline over time in both the Alzheimer’s Disease Neuroimaging Initiative and the San Francisco cohort. Over the 15.7-year UK Biobank follow-up, in which 219 of the 11,059 women (about 2%) developed Alzheimer’s dementia, each one-unit increase in the baseline score was associated with a 15% higher relative hazard of Alzheimer’s dementia (hazard ratio = 1.15). Associations with vascular (65 cases, hazard ratio = 1.15), frontotemporal (23 cases, hazard ratio = 0.90), and all-cause dementia (387 cases, hazard ratio = 1.08) were not statistically significant.

These cognitive associations held after controlling for available cardiovascular and metabolic risk factors, and after excluding phosphorylated tau-231 or participants with baseline cognitive impairment. In exploratory analyses, carrying the APOE4 gene variant, an Alzheimer’s genetic risk factor, modified the results only in the Alzheimer’s Disease Neuroimaging Initiative, where it strengthened the link between higher protein scores and faster cognitive decline.

Because individual protein links with cognition varied across cohorts, the authors suggest this variation may reflect differences in cohort makeup and measurement platforms, or that the links with cognition depend on a broader coordinated shift rather than any single protein. They also propose that follicle-stimulating hormone may directly affect brain vulnerability, as its receptors populate Alzheimer’s-vulnerable brain regions.

“To us, the major contribution of our study is proof-of-concept that we can detect brain-related molecules in the blood of midlife women that show relevance to cognitive aging trajectories,” Wood Alexander, Saloner, and Casaletto said. “The key takeaways are (1) menopause is an important transition for brain health, involving not just changes to the reproductive system but shifts in immune, metabolic, vascular, and nervous systems (among others); and (2) we can use this window as an opportunity to understand women’s brain aging and dementia risk in its very early stages.”

These findings align with other research on female reproductive aging. For example, a prior neuroimaging study of pre- and postmenopausal women, which examined brain structure via MRI rather than blood-based molecular markers, found that postmenopausal women had lower gray matter volume in key brain regions, slower reaction times, and higher levels of anxiety and depression.

Likewise, an earlier cohort study of older women, which examined clinical markers of estrogen exposure rather than blood proteomic signatures, found that women with greater lifetime estrogen exposure, such as reaching menopause at a later age, tended to have better cognitive performance, larger brain volumes, and a lower risk of dementia.

Additionally, the association between past hot flashes and elevated inflammatory proteins, measured years later in the Wisconsin cohort, complements an earlier observational study of postmenopausal women. That study, which assessed self-reported perimenopausal symptom burden rather than blood protein profiles, found that women who recalled experiencing more symptoms during perimenopause tended to show more cognitive and behavioral changes later in life.

As with all research, there are some caveats to consider. The midlife analyses compared different women at a single time point rather than tracking the same individuals across their final menstrual period, and the scores in the four aging cohorts were calculated from blood drawn when women were already older, meaning the study does not show that midlife protein levels predict future cognitive decline.

At the same time, because men in the discovery cohort had higher composite scores than women, the score may also capture general aging and inflammation biology, not only menopause. In addition, the UK Biobank Alzheimer’s risk increase was modest, representing a 15% relative increase per one-unit rise in the score on a base rate of roughly 2% (219 of 11,059 women diagnosed over about 16 years).

“We do not believe menopause ’causes’ dementia,” the researchers emphasized. “All midlife individuals with ovaries will undergo menopause, but the majority do not go on to develop dementia. Furthermore, in our study, the levels of brain-related molecules in postmenopausal women did not exceed those of age-matched men, suggesting women may experience a more ‘compressed’ window of biological aging during menopause. In this way, we think that menopause may act as a ‘stress test’ for the brain that could give us clues into a woman’s future brain health risk or resilience.”

Methodological factors also shape how the protein and symptom data can be interpreted. Because the 16 discovery proteins were selected using an uncorrected statistical threshold across 118 comparisons for differing between pre- and postmenopausal women, a few chance hits would be expected and the score gap between those two groups is partly built in, making the intermediate position of perimenopausal women and the UK Biobank replication the more independent tests.

Circulating blood proteins can also originate from multiple organs rather than exclusively the brain, and three of the 13 overlapping discovery proteins did not replicate in the UK Biobank, which the authors suggest could reflect differences between assay platforms, chance findings in the smaller discovery sample, or the UK Biobank’s cruder self-reported measure of menopause status.

Finally, the 80-woman discovery sample was nearly 90% white and menopause symptoms were assessed via self-reported questionnaires rather than objective skin monitors, so longitudinal studies tracking more diverse cohorts with repeated blood draws, objective symptom monitoring, and brain imaging will be needed to map these trajectories.

“A major barrier impeding our understanding of how menopause shapes brain aging is the lack of longitudinal studies tracking changes in hormones, brain structure and function, and broader multi-system health in women across this critical midlife window,” the researchers said. “While we believe our Nature Medicine paper is an important first step, our analyses were all cross-sectional (i.e., comparing women of different menopause stages), and we need careful and detailed studies that follow the same women through menopause and beyond to start to understand how these neurobiological changes unfold and how/why some women show more resilient versus vulnerable responses to this major transition.”

“Our teams at UCSF, University of Toronto, and the Ann S. Bowers Women’s Brain Health Initiative (including with collaborators at UCSB, UCSD, and others) have recently launched multi-site cohort studies to address this gap, including the Longitudinal Menopause Project and the Women’s Brain Health Collaborative.”

The study, “Blood proteomics of menopause map to brain aging and dementia risk,” was authored by Madeline Wood Alexander, Jennifer S. Rabin, Michelle Caunca, Allesandra Iadipaolo, Louisa Cornelis, Ria Warrier, Keenan A. Walker, Nina Miolane, Veronica Augustina Bot, Brendan Wood, Hamilton Se-Hwee Oh, Tony Wyss-Coray, Albert Pham, Julia Borger, Valentina Diaz, Emily W. Paolillo, Joel Kramer, Laura Pritschet, Caitlin Taylor, Matthew S. Panizzon, Ramiro Eduardo Rea Reyes, Marisa N. Denkinger, Nicholas J. Ashton, Sterling C. Johnson, Emily G. Jacobs, Rowan Saloner, and Kaitlin B. Casaletto.