I order a follicle-stimulating hormone (FSH) on nearly every woman who walks into my medical practice, and I do it for a reason that most of my colleagues do not know about.
Welcome to THE NUMBER, the series every Friday where I describe a lab value or measurement. We began the series with cortisol, so that you know where you stand with stress (Issue No. 01).
Last week was estradiol, which I check every three to six months (Issue No. 02).
FSH rides in the same blood draw and until a few years ago, I ordered it as a formality to check fertility in a woman in her 30s or 40s, or to assess for menopause in a woman with a hysterectomy. Early in my career, I would check it occasionally as a proxy for how estrogen was getting absorbed in the tissues, but based on emerging research in the female brain, I’m ordering FSH more and keeping it lower, even though the data is early and mostly associational.
Summary of this post:FSH isn’t just the hormone that clocks out your ovaries, it sneaks into your hippocampus and cozies up to Alzheimer’s proteins. Short version: High FSH may not be good for the brain. The mouse data suggests causation, the human data is compelling but early, and nobody has handed us a magic cutoff yet. So I check my FSH anyway, because I'd rather have an ambiguous data point and be early to the party.
For three decades, FSH has had exactly one job in clinical medicine. It confirms that the ovaries have stopped responding. It rises as they go quiet, and once it climbs into the postmenopausal range, it tells us what we already suspected. NICE and the Choosing Wisely campaign both say the same thing about the perimenopausal years: the number swings too widely, within a single cycle and across cycles, to mean anything reliable, and the diagnosis belongs to her symptoms, not her labs. [1][2] I was taught that. I ran the panel for years believing it was settled science.
What I know now is that FSH may not be a bystander marker of a woman’s reproductive timeline. It may be a driver of what happens to her brain once that timeline ends.
Here is what surprised even the researchers studying FSH. Mone Zaidi, MD/PhD, working alongside imaging science from Lisa Mosconi PhD’s menopausal brain imaging research, went looking for FSH in the brain.
They found it. FSH was sitting on neurons in the hippocampus and cortex, the exact regions that Alzheimer’s attacks first.
FSH does not clock out once it leaves the pituitary brain. We used to think its entire job was to travel through the blood and talk to the ovaries. One way signal, out and done. What Zaidi’s group found is that FSH does not stop there, it also acts directly on the neurons in the hippocampus and cortex, because these neurons carry their own FSH receptors, sitting there waiting for a signal nobody knew was meant for them.
Once FSH binds in these regions, it flips a molecular switch that seems to increase the buildup of two proteins Alzheimer’s is built from: amyloid and tau.
The researchers tested this in mice bred to develop Alzheimer’s. When they blocked FSH from reaching the brain, the mice got sharper. Memory improved, thinking improved. [3]
Then came the twist. A second study found FSH doesn’t do this damage alone. It teams up with ApoE4, the gene variant already known to raise Alzheimer’s risk. Researchers gave an FSH-blocking antibody to two groups of genetically identical mice, siblings from the same litter, differing only in which version of that one gene they carried. In the mice with ApoE4, the antibody reversed the brain damage. In their littermates carrying the more common ApoE3 version, it did nothing at all. Same treatment, same lab, same day. The only thing that changed the outcome was the gene. [4]
A third study pushed further, and the pattern held with almost eerie precision. Researchers dialed down how much FSH receptor the mice’s brain cells were allowed to produce, in careful steps. Less receptor, better memory, every single time, like turning down a dial and watching the fog lift in real time. [5]
The human data followed in 2025. Wang and colleagues studied 884 postmenopausal women, age 60 and older, spread across cognitively normal, mild cognitive impairment, and Alzheimer’s dementia groups. FSH rose step by step with disease severity: lowest in the cognitively normal women, higher in mild impairment, highest in dementia. It tracked with amyloid burden on PET imaging, and amyloid burden explained the link between FSH and cognitive performance. Estradiol, measured in the same women, showed no such pattern. [6] That last finding is the one I sit with. For two decades, the working theory in menopause medicine was that falling estrogen explained the cognitive changes of this transition. This study points at a different hormone.
Other lines of evidence converge on the same signal. A Mendelian randomization study, which uses genetic variants as a stand-in for lifetime FSH exposure, found evidence of a causal link to Alzheimer’s disease, a stronger claim than correlation allows. [7]
Longitudinal imaging data from the Alzheimer’s Disease Neuroimaging Initiative found that higher baseline FSH predicted faster cognitive decline and greater temporal lobe volume loss across the cohort, with the association reaching significance in women but not in men. [8]
A 2026 review now proposes FSH-blocking antibodies as a dual-target therapy for osteoporosis and Alzheimer’s disease at once, on the premise that a single hormone drives both. [9]
I want to be honest about the limits of this evidence before anyone orders a test based on this post.
Wang’s study analyzed FSH as a continuous variable and by diagnostic group. It did not define a cutoff, and no one has validated one in postmenopausal women. [6]
The figure that circulates in the preclinical literature, an FSH under 8 ng/mL associated with better memory in mice, is a mouse benchmark, not a human target. [5]
A cross-sectional study that did generate a cutoff, 28.4 IU/L for predicting dementia, drew its cohort from adults over 80, most of them men, which makes it a poor fit for a woman in her 50s reading this over coffee. [10]
The human studies are not even uniform. One memory-clinic cohort found FSH positively correlated with cognition in women who were cognitively normal, and a separate cohort found no relationship between FSH and subjective cognitive decline at all. [11][12]
The mainstream position, and I think it is a defensible one, is that this remains associational and preclinical, that no diagnostic or treatment threshold exists, and that FSH should not be used to stratify dementia risk outside of a research setting.
I hold a different position for my own practice, and here’s why.
Free subscribers: annual subscription just dropped to $99, and it’s only staying there a week or so before it climbs back. If you’ve been on the fence about going paid for the full protocols and deeper stuff, this is the easiest that math is ever going to look. $99 for the year. Do it before I change my mind (I won’t, but pretend I might).
I believe the biology here is specific enough, and the mechanism clean enough across mouse and human data, that waiting for a consensus cutoff means waiting way too long for permission I do not need in order to have a more informed conversation with my patients.
A result without a validated threshold is not useless. It’s a data point I can track over years, weigh against a woman’s ApoE4 status if she has had it checked, and read alongside her estradiol, her cognitive complaints, and her family history.
None of that requires a number engraved into a guideline. Instead, it requires a physician willing to follow where the evidence points before the committees finish arguing about how to word it.
Here is the action: *If you are over 40, ask your physician to run FSH alongside estradiol, not as a menopause-staging tool but as a baseline you track over time. *
If you carry an ApoE4 allele, the case for tracking it is stronger still, since that is the population where the Zaidi group’s antibody work showed the clearest effect. [4]
There is no target number to hit today, and no treatment protocol attached to a high result. The value of this test right now is longitudinal: a marker you and your physician watch move across years, not a single draw that settles the question.
I check my own FSH every quarter, the same morning I run my estradiol in my blood test. I do not yet know what my number means for my brain in 30 years, but I aim to keep my FSH around 30 or less. Given the early evidence and my own concern about brain health, my take is that I would rather have the data and no certainty than no data and false reassurance.
What would you want to know about your own FSH?
Would you ask your physician to run it before your next visit, or wait for the guidelines to catch up?
xo, Dr. Sara
Notes
Lumsden MA et al. JAMA Internal Medicine, 2016.
Crandall CJ et al. JAMA, 2023.
Xiong J et al. Nature, 2022.
Xiong J et al. Nature Communications, 2023.
Korkmaz F et al. Molecular Psychiatry, 2025.
Wang SM et al. Frontiers in Aging Neuroscience, 2026.
Ding C et al. Annals of Human Genetics, 2025.
Zhao M et al. Frontiers in Neuroscience, 2025
Sims S et al. Current Osteoporosis Reports, 2026.
Zhu Y et al., Frontiers in Aging Neuroscience, 2025.
Oh DJ et al. Journal of Korean Medical Science, 2025.
Wang Q et al. Clinical Endocrinology, 2024.
