bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022.

The copyright holder for this preprint

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.

Menopause, Brain Anatomy, Cognition and Alzheimer’s Disease

Manuela Costantino (1), Grace Pigeau (1,2), Olivier Parent (1,3), Justine Ziolkowski (1,3), Gabriel A.
Devenyi (1,4), Nicole J. Gervais (5), M. Mallar Chakravarty (1,2,4)
1. Cerebral Imaging Center, Douglas Mental Health University Institute, Verdun, Canada
2. Department of Biological and Biomedical Engineering, McGill University, Montreal, Canada
3. Integrated Program in Neuroscience, McGill University, Montreal, Quebec, Canada
4. Department of Psychiatry, McGill University, Montreal, Canada
5. Krembil Centre for Neuroinformatics, Centre for Addiction and Mental Health, Toronto, Ontario, Canada

Abstract
The menopause transition has been repeatedly associated with decreased cognitive performance
and increased incidence of Alzheimer’s Disease (AD), particularly when it is induced surgically 1,2 or takes
place at a younger age 3,4. However, there are very few studies that use neuroimaging techniques to examine
the effects of these variables in aggregate and in a large sample. Here, we use data from thousands of
participants from the UK Biobank to assess the relationship between menopausal status, menopause type
(surgical or natural), and age at menopause with cognition, AD, and neuroanatomical measures derived
from magnetic resonance imaging. We find that for brain and cognitive measures, menopausal status,
menopause type and age at surgical menopause do not impact the brain; but that there is a positive
correlation between anatomy, cognition and age at non-surgical menopause. These results do not align with
previous reports in the literature with smaller samples. However, we confirm that both early and surgical
menopause are associated with a higher risk of developing AD, indicating that early and abrupt ovarian
hormone deprivation might contribute to the development of the disorder.

Introduction
Menopause is associated with a decrease been used to model the impact that ovarian
in circulating ovarian hormones, which impacts hormone deprivation may have on the ageing
multiple physiological systems, including brain process. Critically, lower cognitive performance
5,6
structure and function . Surgical menopause and increased incidence of Alzheimer’s Disease
(where ovaries are removed via bilateral (AD), have been observed in women who had
oophorectomy) as opposed to spontaneous (after their ovaries removed prior to non-surgical
7
one year without menstruation) leads to a faster menopause . Similarly, early non-surgical
decrease in hormone production. The timing of menopause has been associated with accelerated
the transition is also believed to impact women’s neurodegeneration, measured with magnetic
health, thus, surgical and early menopause have resonance imaging (MRI), in addition to worse
 bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.

3,4
cognition . Adding to this critical body of Socioeconomic status (SES) was quantified
literature are mixed findings related to improved based on total household income and missing
cognition in women taking hormone therapy data was imputed by the mean. HT was encoded
(HT), exogenous female hormones, during the as a binary variable, distinguishing those with a
8
transition . Although these vary in type, dose, history of HT during the menopause transition
duration and timing, they contain estrogens with and those without it. Unfortunately, no additional
or without progestin 2. Thus, we investigated how information about HT was available in UK
both surgical and spontaneous menopause, as Biobank records. Date of AD diagnosis was
well as age at menopause, related to multiple obtained from hospital admissions data linked to
measures of cognitive performance, brain the dataset. 9
anatomy, and AD incidence in a large cohort of
women from the UK Biobank, while robustly Cognitive Data
controlling for age. Cognitive tests were performed at the UK
Biobank’s assessment centers 10. Seven domains
Methods were selected, based on data availability: fluid
Sample intelligence score, numeric memory, pairs
Cross-sectional demographic, cognitive, matching, prospective memory, reaction time,
MRI and health-outcome data was obtained from symbol digit substitution and tower
the UK Biobank. Based on self-report data, rearrangement. To maximize power while
female participants were divided into avoiding missing data, participants were included
premenopausal women (PRE), women who if they had participated in all of the first five tests
underwent a bilateral oophorectomy (with or (cognitive group 1) or in both of the last two
without a hysterectomy) prior to menopause (cognitive group 2) (Table 1).
(SURG) and women who underwent menopause
without surgical interference (POST). MRI Data
Participants who reported uncertainty of their T1-weighted structural MRI scans were
menopausal status or who underwent a processed with CIVET 2.1.1 and four
hysterectomy without a bilateral oophorectomy morphometric outputs were selected: total brain
were excluded from the study. Different volume (TBV), total gray matter volume (GMV),
subsamples of these groups were included in each total white matter volume (WMV) and vertex-
analysis, based on data availability (Table 1). wise cortical thickness (CT) measurements at
Participants were excluded if self- 77,122 points across the cortical surface.
reported data was missing for menopausal status Participants were further excluded based on
or type, age at menopause or use of HT.
 bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.

manual quality control of raw scans and cortical were carried out on a vertex-wise basis, with CT
reconstructions. as a response variable and false detection rate
(FDR) correction. A chi squared test was
Statistical Analysis performed to compare the prevalence of AD in
Groups were age matched using the the POST and SURG groups and models 2 and 3
nearest neighbors algorithm with the MatchIt were used to examine the effects of age at
package in R 4.1.2 and the supplementary menopause on age at AD diagnosis. Logistic
methods can be found in the online-only regressions with the same predictors as models 2
materials. Multiple regressions were performed and 3 were performed with AD diagnosis as a
with the seven cognitive tests and the three response variable. Analyses were run in R v4.1.2
whole-brain anatomical measures as response and using RMINC v.1.5.3.0.
variables. For each of them, three models were
used: Supplementary analyses
Alternative models were attempted,
Model 1: (cognitive/anatomical measure) ~ age including the encoding of age at menopause as a
+ group + SES categorical variable (premature: <40, early: <45,
Model 2: (cognitive/anatomical measure) ~ age spontaneous: 45+) (Supplementary fig 1), the
+ age at menopause + HT + SES, POST only inclusion of a group by age at menopause
Model 3: (cognitive/anatomical measure) ~ age interaction term (Supplementary fig 2), the
+ age at menopause + HT + SES, SURG only removal of all participants under the age of 60
(Supplementary fig 3) and the removal of all
Multiple comparisons were accounted for using participants who with a history of HT
Bonferroni correction. These same three models (Supplementary fig 4)

COGNITION (Group 1)

Unmatched Matched

Surgical Postmenopausal Surgical Postmenopausal Premenopausal

Sample size 1,778 14,526 567 1134 567

Age 58.83 (6.63) 60.11(5.40) 52.07 (5.82) 52.55 (5.58) 53.08 (5.25)

Age at
menopause 44.89 (6.90) 49.84 (5.06) 42.38 (6.94) 47.33 (5.43)

SES 437/437/353/178/4 3,463/3,685/2,780/1 114/121/157/97/ 199/234/301/214/56 74/102/141/126/29

 bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.

2 ,633/370 23

HT users 915 (51.4%) 2,799 (19.2%) 222 (39.1%) 98 (8.64%)

COGNITION (Group 2)

Unmatched Matched

Surgical Postmenopausal Surgical Postmenopausal Premenopausal

Sample size 877 8,350 391 782 391

Age 57.05 (6.07) 58.34 (5.14) 51.70 (4.00) 51.83 (3.87) 51.95 (3.74)

Age at
menopause 45.77 (6.00) 50.07(4.50) 44.15 (5.47) 47.84 (4.66)

107/219/223/163/4 1,080/1,968/2,163/1 32/77/110/112/2

SES 0 ,707/457 3 64/129/238/218/66 21/51/107/133/40

HT users 428 (48.8%) 1,580 (18.9%) 144 (36.8%) 81 (10.4%)

ANATOMY

Unmatched Matched

Surgical Postmenopausal Surgical Postmenopausal Premenopausal

Sample size 979 13,237 223 446 223

Age 63.68 (6.91) 63.42 (6.63) 54.11(2.95) 53.78 (2.67) 53.42 (2.39)

Age at
menopause 46.33 (6.32) 50.48 (4.65) 44.89(6.22) 49.29 (4.33)

123/268/254/160/4 1,694/3,467/3,487/2
SES 6 ,382/688 24/37/69/57/22 29/68/112/141/52 14/28/71/72/28

HT users 477 (48.7%) 1,808 (13.7%) 62 (27.8) 16 (3.5%)

AD DIAGNOSIS

Unmatched Matched

Surgical Postmenopausal Surgical Postmenopausal Premenopausal

Sample size 16,960 149,399 16,960 16,960

Age 58.88 (6.49) 60.09 (5.45) 58.88 (6.49) 58.88 (6.49)

# Cases 105 (0.62%) 514 (0.34%) 105 (0.62%) 91 (0.54%)

Age at
menopause 45.58 (6.71) 49.78 (5.05) 45.58 (6.71) 49.17 (5.44)

4,109/3,890/3,190/ 34,636/35,516/28,3 4,109/3,890/3,19 3,857/3,997/3,343/2

SES 1,893/318 79/17,597/4,275 0/1,893/318 048/516

HT users 8,189 (48.3%) 29,020 (19.4%) 8,189 (48.3%) 3,165 (18.7%)

 bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.

AGE AT AD DIAGNOSIS

Unmatched Matched

Surgical Postmenopausal Surgical Postmenopausal Premenopausal

Sample size 105 514

Age 64.64 (4.44) 64.70 (3.44)

Age at
menopause 44.25 (8.28) 49.68 (5.90)

SES 31/18/15/3/1 169/114/51/19/6

HT users 37 (35.2%) 220 (42.8%)

Age at
diagnosis 73.08 (5.41) 73.57 (4.52)

Table 1: Basic demographic information for the participants in each analysis. For age and age at
menopause, the mean and standard deviation, in parentheses, are shown. The number of participants who
used HT as well as the proportion of the total sample that they represent is also reported. For SES, the
values represent the number of participants in each of the following income categories, in pounds sterling:
less than 18,000/18,000-30,999/31,000-51,999/52,000-100,000/greater than 100,000. Over 90% of the
sample identifies as ethnically white.

Results p<0.001), white matter (𝛽=0.047, p<0.001), and

A full description of the samples can be grey matter volumes (𝛽=0.033, p<0.001) being
found in Table 1. 3,373 MRIs were discarded relatively preserved in those with later onset of
after quality control, namely 151(14.7%) in the menopause in the POST group (see Fig 1a). All
PRE group, 3,091(18.9%) in the POST group and the continuous variables were z-scored to obtain
313(23.8%) in the SURG group. standardized beta coefficients. Further still, age at
non-surgical menopause was positively
Anatomy and Cognition associated with CT in many regions across the
We observed that later onset of cortical surface (Fig 1b). Outside of these
menopause in the POST group only was findings age of menopause, age and HT status
positively associated with fluid intelligence were not associated with cognitive performance
(𝛽=0.095, p<0.001), numeric memory (𝛽=0.064, or brain anatomy measures.
p<0.001) and pairs matching (𝛽=0.040, p<0.001).
These observations are further supported by
results that demonstrate total brain (𝛽=0.046,
 bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.
 bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.

Alzheimer’s Disease urgent need to increase sample sizes and improve

The rates of conversion to AD (see Fig 1c and d) age-correction as a means of understanding sex-
further supports a cognitive and brain specific factors that may underly dementia.
preservation effect of a later onset of menopause. Our findings of fewer AD diagnoses in
Logistic and multiple regressions and chi squared those with a later menopause onset further
tests suggest that women in the SURG group are support its protective effect in women with
more likely to have AD (𝛘2 = 1.0058, p = 0.3159) surgical menopause, suggesting that early and
and that an earlier age of menopause is related to abrupt hormone deprivation predisposes women
higher rates of AD diagnosis (𝛽=-0.1091, to AD later in life without impacting their
p<0.001 for POST, 𝛽=-0.2879 , p<0.001 for cognition or brain health at midlife 16. Any of the
SURG). In the SURG group, women who did not endocrine, immune, or cardiovascular changes
consume HT are more likely to develop AD (𝛽=- associated with menopause could be interacting
0.6003, p=0.004). with AD patopathology to alter its progression,
as opposed to directly worsening general brain
Discussion health.
Overall, our findings suggest that The question of whether the relationship
menopause status does not have notable effects between menopause and the brain’s structural
on the brain or cognition, for both surgical and and functional changes is causal remains
non-surgical groups. These critical findings do unanswered. On the one hand, there is evidence
not agree with previous reports of brain atrophy from animal studies that ovarian hormone
and reduced cognition 11,12, particularly for those deprivation causes changes in the brain and in
undergoing surgical menopause 1,13
. This behavior and there are multiple proposed
highlights the importance of larger sample sizes pathways 17. Yet, our results do not show an
in association studies to prevent incidental impact of menopausal status and the positive
findings. Nontheless, our findings suggest that correlation between age at menopause and brain
those undergoing non-surgical menopause at an health might simply be explained by generally
earlier age show domain-specific cognitive healthy women having both healthier brains and
decline and brain atrophy patterns, as previously later menopause. In either case, our study
reported 3,14
. Spontaneous, but not surgical, confirms that there are relevant interactions
menopause depends on multiple factors such as between the endocrine and nervous systems,
genetics, lifestyle factors and comorbidities 15
, which supports the idea that brain aging and
which could explain this difference and may be neurodegenerative diseases should not be
intertwined as causal factors that impact this studied in isolation, but rather by integrating
relationship. Overall, our results underscore the data from multiple physiological systems.
 bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.

Some limitations to the study are the Thus, our study is the first to
lack of a perimenopausal group and of comprehensively analyze well-powered MRI,
distinction between HT regimens, due to data cognitive, and clinical data in the context of
availability. Most demographic and reproductive menopause with robust age correction. We
variables were self-reported, which is known to observed that, in healthy populations,
introduce noise and bias, such as incorrect menopause was not associated with changes in
determination of menopausal status. the brain, except in women who underwent early
Furthermore, we do not have any measure of spontaneous menopause. However, both surgical
verbal memory, which is known to decrease transitions and early age at menopause of any
during perimenopause 18. Finally, the proportion type increased the risk of AD. Further research
of the cohort that was diagnosed with AD is is needed to understand the mechanisms behind
limited, particularly in the SURG group. the interactions between these two systems.
Therefore, it will be critical to monitor the AD-
conversion rates as UK Biobank continues to
collect data. Since this study only examines the
long term effects of the transition, no
conclusions can be drawn on how
perimenopause affects the brain.
 bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.

References

1. Rocca WA, Bower JH, Maraganore DM, et al. Increased risk of cognitive impairment or dementia in
women who underwent oophorectomy before menopause. Neurology. 2007;69(11):1074-1083.

2. Rocca WA, Grossardt BR, Shuster LT. Oophorectomy, estrogen, and dementia: a 2014 update. Mol
Cell Endocrinol. 2014;389(1-2):7-12.

3. Kuh D, Cooper R, Moore A, Richards M, Hardy R. Age at menopause and lifetime cognition:
Findings from a British birth cohort study. Neurology. 2018;90(19):e1673-e1681.

4. Schelbaum E, Loughlin L, Jett S, et al. Association of Reproductive History With Brain MRI
Biomarkers of Dementia Risk in Midlife. Neurology. 2021;97(23):e2328-e2339.

5. Maki PM, Thurston RC. Menopause and Brain Health: Hormonal Changes Are Only Part of the
Story. Front Neurol. 2020;11:562275.

6. Jabbour HN, Sales KJ, Catalano RD, Norman JE. Inflammatory pathways in female reproductive
health and disease. Reproduction. 2009;138(6):903-919.

7. Georgakis MK, Beskou-Kontou T, Theodoridis I, Skalkidou A, Petridou ET. Surgical menopause in

association with cognitive function and risk of dementia: A systematic review and meta-analysis.
Psychoneuroendocrinology. 2019;106:9-19.

8. Joffe H, Hall JE, Gruber S, et al. Estrogen therapy selectively enhances prefrontal cognitive
processes: a randomized, double-blind, placebo-controlled study with functional magnetic resonance
imaging in perimenopausal and recently postmenopausal women. Menopause. 2006;13(3):411-422.

9. Sudlow C, Gallacher J, Allen N, et al. UK biobank: an open access resource for identifying the
causes of a wide range of complex diseases of middle and old age. PLoS Med. 2015;12(3):e1001779.

10. Fawns-Ritchie C, Deary IJ. Reliability and validity of the UK Biobank cognitive tests. PLoS One.
2020;15(4):e0231627.

11. Mosconi L, Berti V, Dyke J, et al. Menopause impacts human brain structure, connectivity, energy
metabolism, and amyloid-beta deposition. Sci Rep. 2021;11(1):1-16.
 bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.

12. Pertesi S, Coughlan G, Puthusseryppady V, Morris E, Hornberger M. Menopause, cognition and

dementia – A review. Post Reprod Health. 2019;25(4):200-206.

13. Zeydan B, Tosakulwong N, Schwarz CG, et al. Association of Bilateral Salpingo-Oophorectomy

Before Menopause Onset With Medial Temporal Lobe Neurodegeneration. JAMA Neurol.
2019;76(1):95-100.

14. Bove R, Secor E, Chibnik LB, et al. Age at surgical menopause influences cognitive decline and
Alzheimer pathology in older women. Neurology. 2014;82(3):222-229.

15. Ceylan B, Özerdoğan N. Factors affecting age of onset of menopause and determination of quality of
life in menopause. Turk J Obstet Gynecol. 2015;12(1):43-49.

16. Scheyer O, Rahman A, Hristov H, et al. Female Sex and Alzheimer’s Risk: The Menopause
Connection. J Prev Alzheimers Dis. 2018;5(4):225-230.

17. Tao X, Yan M, Wang L, et al. Effects of estrogen deprivation on memory and expression of related
proteins in ovariectomized mice. Ann Transl Med. 2020;8(6):356.

18. Maki PM. Verbal memory and menopause. Maturitas. 2015;82(3):288-290.

 bioRxiv preprint doi: https://doi.org/10.1101/2022.10.18.512730; this version posted October 21, 2022. The copyright holder for this preprint
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
available under aCC-BY 4.0 International license.