“Puberty, Pregnancy, Menopause: Scientists Scanned 1,095 Brains — and Menopause Was the Odd One Out”
Grey matter contains neuronal cell bodies, dendrites, synapses and supporting cells, and a fall in MRI-measured grey-matter volume does not automatically mean tissue damage. During adolescence, cortical grey-matter volume commonly decreases as the brain matures. Processes such as synaptic remodelling and increasing myelination may contribute, but structural MRI cannot tell us exactly which cellular mechanism is responsible — or whether a particular volume change is beneficial or harmful.
That raises a useful question: do pregnancy and menopause show structural brain patterns that resemble those seen around menarche? Previous studies examined these transitions separately. Direct comparison has been difficult because the cohorts differ in age, scanners and acquisition protocols.
A team led by the Pregnancy Brain Lab at Amsterdam University Medical Center, with colleagues at Leiden University, analysed longitudinal MRI data from 1,095 girls and women across three independent cohorts. The scans came from different studies and scanners, but the researchers applied a consistent FreeSurfer-based morphometric framework and compared transition groups with age-relevant stable controls within each cohort before making cross-cohort comparisons.
What they found
1. Menarche and pregnancy shared a broad pattern of cortical volume reduction. Girls transitioning from pre- to post-menarche showed significant monthly reductions in total and cortical grey matter, and first- and second-time mothers showed significant reductions across pregnancy. The reported total-grey-matter rates were about −0.13 per month in the menarche-transition group and −0.12 per month in first-time mothers in the study's intracranial-volume-adjusted metric. Their relevant stable controls did not show significant total or cortical change. Because the cohorts were scanned on different systems, the authors caution against over-interpreting absolute rate comparisons across cohorts.
2. Puberty and pregnancy converged in 34 cortical regions. After subtracting each cohort's control-group rate, the menarche and first-pregnancy groups did not significantly differ from one another in 34 regions, while both differed from menopause. These convergent regions were concentrated in prefrontal, parietal and temporal association cortex. Those regions support many higher-order functions, but this study did not test social cognition, planning or behaviour, so the functional meaning of the overlap remains uncertain.
3. Second pregnancies showed a similar global pattern. Second-time mothers showed a total-grey-matter rate of about −0.11 per month and did not differ significantly from first-time mothers on the global measures. That suggests the pregnancy-associated structural pattern is not restricted to a first pregnancy, although the study does not establish the cellular mechanism behind it.
4. The differences were at least as important as the similarities. In 33 regions there was a stepwise pattern: menarche showed greater control-subtracted decline than pregnancy, and pregnancy greater decline than menopause. Six additional regions showed greater decline during menarche than during both pregnancy and menopause; several were insular regions. So the study supports both convergence and divergence — not one universal 'female hormone' brain signature.
5. Menopause showed the most surprising pattern — but it is a cautious null result. Women classified as transitioning from pre- to post-menopause showed a total-grey-matter rate of about −0.013 per month; this within-group change did not remain significant after Bonferroni correction. Stable premenopausal and stable postmenopausal groups did show significant total-grey-matter reductions. However, the overall group comparison was not significant (P = 0.122 for total grey matter), so the data do not prove that menopause protects against volume loss or that transitioning women truly lose less than their peers. The authors interpret the pattern as possible attenuation of age-related decline and explicitly call for caution.
6. The study is large for longitudinal neuroimaging, but the strength of evidence differs by cohort. The pregnancy group effect on total grey matter was large (η² ≈ 0.35), while global puberty effects were smaller (around η² 0.06–0.07). In the cross-cohort regional analysis, all 74 cortical regions showed a significant difference among the three transition-specific rates after FDR correction — not that all 74 regions independently 'changed' within every transition. The menopause cohort was much larger (n = 843) than the puberty and pregnancy cohorts, but its transition subgroup contained 120 women.
Now the part the headline can't fit
Brain function was not measured. This study analysed structural MRI. It did not analyse memory, attention, mood or subjective cognitive symptoms. Therefore, the findings cannot tell us whether the observed volume patterns improved, worsened or had no effect on how participants felt or functioned.
The cross-cohort hormone story was not directly tested. The three cohorts did not have harmonised hormone measurements, so this study could not directly relate comparable oestrogen or progesterone trajectories to brain-volume change across all three transitions. The hormonal interpretation is biologically plausible and supported by prior work, but the present cross-cohort analysis does not demonstrate the mechanism.
Three cohorts meant different scanners and acquisition settings. The puberty, pregnancy and menopause datasets came from separate studies. The same analytical framework helps, and subtracting each cohort's own control-group rate reduces some systematic differences, but it cannot fully remove scanner or cohort effects. The authors therefore caution against treating the raw rates as perfectly comparable across cohorts.
The menopause result is the one to read most cautiously. Menopause status was based on self-report rather than detailed clinical staging, so some women labelled premenopausal may already have been perimenopausal. The second scan occurred on average about 2.16 years after the reported final menstrual period, with wide variability, so the most dynamic part of the transition may not have been captured. The groups were also highly imbalanced (120 transitioning versus 674 stable postmenopausal women), which particularly affects the PCA interpretation.
Grey-matter reduction is not automatically good or bad. MRI measures macroscopic structure, not synaptic pruning, myelination, cell number or fluid shifts directly. The authors use the term 'plasticity' descriptively and state that the structural data cannot determine whether the menopause pattern reflects heightened plasticity, resilience or simply a slowing of age-related decline.
This was observational, not experimental. Hormonal contraceptive use was not documented consistently, and socioeconomic status, mental health, stress and reproductive-history variables could not be harmonised across the three cohorts. There was also no male comparison group. The within-cohort controls help account for normal development and ageing, but they cannot isolate hormones from every accompanying biological and life-context factor.
Here's how I'd read this as a doctor.
Don't read 'less grey-matter volume' as 'brain damage.' Structural MRI cannot determine the cellular cause or functional meaning of a volume change. Puberty and pregnancy both showed cortical reductions relative to their controls, with overlap in some association-cortex regions, but the study did not test whether those changes were beneficial, harmful or functionally neutral.
And don't use the menopause result to dismiss symptoms. This study did not measure cognitive symptoms, mood or day-to-day function. A structural MRI null result therefore cannot confirm or refute what a woman experiences during the menopausal transition, and it should not be used to make treatment decisions on its own.
Nothing here is a reason to start or stop hormone therapy. The study did not test hormone therapy as an intervention, and harmonised hormone levels were not available across cohorts. Using these MRI findings to claim that a particular hormone, supplement or 'brain-protective' programme prevents cognitive decline would go beyond the evidence.
What makes this study useful is the direct comparison across life stages. It shows that puberty, pregnancy and menopause should not be assumed to produce one uniform pattern of structural brain change. The next step is prospective work with harmonised hormone measurements, cognitive and emotional outcomes, and more comparable imaging across cohorts. That is what will tell us whether these MRI patterns represent adaptation, vulnerability or simply different trajectories of normal structural change.
This post is a plain-language summary of a published research paper and is not medical advice. Nothing here is a reason to start, stop or change any treatment — talk to your own doctor about that.
Source
van 't Hof SR, Bos MGN, Straathof M, Crone EA, Hoekzema EA. Puberty, pregnancy, and menopause show shared and distinct structural changes across the lifespan. Nature Communications. Published 8 Sep 2026. doi.org/10.1038/s41467-026-76755-2
For the record: longitudinal structural MRI across three independent cohorts (total n = 1,095), all processed with identical FreeSurfer-based morphometry and expressed as change per month. Puberty cohort n = 142 (Leiden BrainLinks / BrainTime / Self-Concept studies; mean age 13.5 ± 1.9 y at T1; mean interscan interval 2.0 y): 49 stable premenarcheal, 34 pre-to-post menarche, 59 stable postmenarcheal. Pregnancy cohort n = 110 (Amsterdam UMC; mean age 31.3 ± 4.1 y; interval 1.4 y): 40 first-time mothers, 30 second-time mothers, 40 nulliparous controls. Menopause cohort n = 843 (UK Biobank; mean age 54.2 ± 3.5 y; interval 8.4 y): 49 stable premenopausal, 120 pre-to-post menopause, 674 stable postmenopausal. Total grey matter: pre-to-post menarche −0.13 mm³/month (t(33) = −5.88, p < 0.001), group difference F(2,139) = 4.79, η² = 0.060; first-time mothers −0.12 (t(39) = −8.24, p < 0.001), second-time mothers −0.11 (t(29) = −8.55, p < 0.001), controls no significant change (p = 0.28), group difference F(2,106) = 28.67, p < 0.001, η² = 0.35; menopause transition −0.013 (t(119) = −2.43, p = 0.017, non-significant after Bonferroni correction) vs stable premenopausal −0.033 (p < 0.001) and stable postmenopausal −0.025 (p < 0.001), group difference F(2,840) = 2.11, p = 0.122 (NS). Cortical volume followed the same pattern. Kruskal–Wallis comparisons across all 74 cortical regions were significant after FDR correction (all p_FDR ≤ 0.001; η² 0.073–0.563, median 0.376), yielding 34 convergent regions (puberty ≈ pregnancy > menopause; prefrontal, parietal and temporal association cortices), 33 stepwise regions (puberty > pregnancy > menopause) and 6 puberty-specific regions (mainly insula). Author-stated limitations: monthly rates may oversimplify nonlinear trajectories; data acquired on different scanners without cross-cohort harmonisation, so absolute cross-cohort rates need caution; no harmonised hormone measures, so hormone–brain associations were not tested; inconsistent documentation of hormonal contraceptive use; no harmonised data on socioeconomic status, mental health or stress; no harmonised reproductive history; menarche and final menstrual period based on single self-report items, so some women classified as premenopausal may already have been perimenopausal; mean 2.16 ± 2.14 years between reported menopause and the second scan, so null findings need cautious interpretation; subcortical analyses limited to aggregate volumes; the menopause principal-component analysis was influenced by the large stable postmenopausal group; and no male control group was included. The authors state that they use "plasticity" descriptively, and that structural data cannot determine whether the menopause finding reflects heightened plasticity or simply a slowing of age-related decline. No cognitive or behavioural outcomes were analysed.
Comments
Post a Comment