Abstract
Objective:
The role of circulating sex hormones on structural brain ageing is yet to be established. This study explored whether concentrations of circulating sex hormones in older women are associated with the baseline and longitudinal changes in structural brain ageing, defined by the brain-predicted age difference (brain-PAD).
Design:
Prospective cohort study using data from NEURO and Sex Hormones in Older Women; substudies of the ASPirin in Reducing Events in the Elderly clinical trial.
Patients:
Community-dwelling older women (aged 70+ years)
Measurements:
Oestrone, testosterone, dehydroepiandrosterone (DHEA), and sex-hormone binding globulin (SHBG) were quantified from plasma samples collected at baseline. T1-weighted magnetic resonance imaging was performed at baseline, one and three years. Brain age was derived from whole brain volume using a validated algorithm.
Results:
The sample comprised of 207 women not taking medications known to influence sex hormone concentrations. A statistically higher baseline brain-PAD (older brain age relative to chronological age) was seen for women in the highest DHEA tertile compared with the lowest in the unadjusted analysis (p=0.04). This was not significant when adjusted for chronological age, and potential confounding health and behavioural factors. Oestrone, testosterone and SHBG were not associated with brain-PAD cross-sectionally, nor were any of the examined sex hormones or SHBG associated with brain-PAD longitudinally.
Conclusion:
No strong evidence of an association between circulating sex hormones and brain-PAD. Given there is prior evidence to suggests sex hormones may be important for brain ageing, further studies of circulating sex hormones and brain health in postmenopausal women are warranted.
Keywords: ageing, biomarkers, brain, hormones, neuroimaging, magnetic resonance imaging, postmenopause
INTRODUCTION
Biological ageing is broadly defined by the gradual accumulation of molecular and cellular damage, and subsequent loss of physiological functions 1. The brain is particularly susceptible to the effects of ageing, undergoing progressive changes over the lifespan. This includes brain atrophy (the loss of tissue volume), which has been associated with cognitive decline, and neurodegenerative disorders 2. However, the rate of brain aging varies, and is strongly influenced by endogenous and exogenous factors.
Several biomarkers have been developed to capture the heterogeneity of biological brain ageing. These include ‘brain age’; which is a neuroimaging-based measure derived by modelling the relationship between brain structure and chronological age in a large normative dataset. The model can then be used to predict the biological brain age of each individual in a new dataset relative to the population norm. The difference between predicted and chronological age is defined as an individual’s “brain-predicted age difference” (brain-PAD) 3. An older brain age relative to chronological age (i.e., a positive brain-PAD) is considered a sign of advanced brain ageing, and has been linked to a weaker grip strength, poor cognitive function 3 and dementia 4. Using a brain age model based on structural magnetic resonance imaging (MRI) 3, we previously reported that brain-PAD is negatively associated with cognitive performance in older community-dwelling individuals, and observed a faster rate of brain ageing in men compared with women, over three years of follow-up 5,6.
Sex differences in brain ageing may arise from differences in the concentration of circulating sex hormones 7. Oestrogens have received considerable attention in this field, especially as there are changes in the production of oestrogens after menopause 8. However, evidence linking oestrogens with cognitive ageing remains equivocal 9. The roles of other hormones in older women, such as testosterone and its precursor dehydroepiandrosterone (DHEA), are yet to be established, though some studies suggest they may be neurotrophic 10. Indeed, testosterone has androgen receptor-mediated neuroprotective actions in the brain, including a resilience to pathological processes contributing to dementia, namely soluble beta-amyloid toxicity 10,11. Endogenous gonadotropins, follicle stimulating hormone (FSH) and luteinizing hormone (LH), have been implicated as influencing cognition in human and animal studies, with mixed findings 12,13. There is a paucity of research examining the influence of circulating sex hormones on brain structures, with most studying the indirect actions of hormone therapies 14.
The current brain-PAD literature follows a similar trend, with studies mainly focused on the role of oestrogens. Of these, one study reported a negative cross-sectional association between oestradiol and the deviation of brain age from chronological age; measured in middle aged and older women who were not carriers of the APOE e4 genotype 15. Two studies investigated brain age longitudinally, one across the menstrual cycle, while the other during the post-partum period 16,17. The ASPirin in Reducing Events in the Elderly (ASPREE)-NEURO study was a prospective neuroimaging sub-study of the ASPREE clinical trial 18. Women from ASPREE-NEURO also had their sex hormone profile measured, as a part of the Sex Hormones in Older Women (SHOW) study 19. These studies have provided a unique opportunity to determine whether circulating sex hormones in older women are cross-sectionally associated with brain-PAD, and the longitudinal change in brain-PAD over three years.
MATERIALS AND METHODS
Study design
ASPREE was a double-blind, randomized, placebo-controlled trial investigating the risks and benefits of low dose aspirin (100mg daily) on health outcomes in older men and women from Australia and the US 20. Eligibility criteria into ASPREE are detailed elsewhere 20. In brief, participants were excluded if they had experienced a cardiovascular event or atrial fibrillation, or were diagnosed with a chronic illness that had a high risk of a major bleed or death within 5 years of randomisation 20. Eligible participants were also cognitively unimpaired at randomisation (score >77 on the Modified Mini-Mental State [3MS] 21), and had no history of dementia 20.
ASPREE-NEURO substudy recruited 572 consenting ASPREE men and women from general practitioners distributed around the Melbourne district 18. Eligible participants were free of any contraindications to MRI (e.g., foreign bodies not known to be safe for scanning), and claustrophobia 18. To understand the effects of sex hormones on brain ageing, the present study was limited to 238 NEURO women who completed a baseline and one or more follow-up brain scans, and who had sex steroids quantified at baseline, as part of the Sex Hormones in Older Women (SHOW) study 19. A further 31 women were excluded as they were using systemic sex steroid therapy, an anti-oestrogen or anti-androgen, or systemic glucocorticoid therapy, which could directly influence sex hormone concentrations. This left a total of 207 participants. The SHOW study was approved by the Monash Human Research Ethics Committee (MUHREC; CF16/10–2016000001) and the Alfred Hospital Human Research Ethics Committee (616/15). NEURO is independently registered with the Australian New Zealand Clinical Trial Registry (ACTRN12613001313729) and received MUHREC approval (CF12/2271–2012001223). ASPREE is registered with the International Standard Randomized Controlled Trial Number Register (ISRCTN83772183) and Clinicaltrials.gov (NCT01038583). The current study was approved by the Monash University Human Research and Ethics Committee (2021–29311-58845); all participants provided written informed consent.
Sex hormone measurement
Sex hormones and sex-hormone binding globulin (SHBG) were quantified from a single EDTA plasma sample collected at recruitment or within 12 months. Details of sex hormone quantification are provided elsewhere 19,22. In brief, oestrone, testosterone and DHEA were quantified within a single run, without derivatisation, using liquid chromatography-tandem mass spectrometry (LC-MS/MS) located at the ANZAC Research Institute, University of Sydney 19. Limits of hormone detection (LOD), quantification, and within-run and between-run coefficients of variation were as following: E1 (3.7 pmol/L, 11 pmol/L, 4.7%, 4.6–7.5%), testosterone (35 pmol/L, 0.09 nmol/L, 2.0%, 3.9–6.5%) and DHEA (0.07 nmol/L, 0.17 nmol/L, <10%, <10%) 19. SHBG was measured in batches by automated immunoassay (Roche Diagnostics Australia). Quantification was limited to a coefficient of variation of 1 to 2% 19.
Neuroimaging data collection and quality control
Structural MRI was performed at baseline, one and three years of follow-up using a 3 Tesla Siemens Skyra scanner (Siemens Erlangen, Germany), with a 32-channel head coil, located at the Monash Biomedical Imaging Centre 18. High resolution 3D magnetization-prepared rapid gradient-echo (MPRAGE) images were acquired in the sagittal orientation, with a 1mm isotropic resolution (192 slices, TR=2300 ms, TE=2.07 ms, TI=900, flip angle=9°, and a 256 × 240 mm3 field of view).
Images were assessed for quality using the MRI Quality Control tool (MRIQC) 23, as previously described 5. Poor quality images were identified as outliers (1.5 times the inter-quartile range) for one or more metrics defined by the Pre-processed Connectomes Project Quality Assessment Protocol (http://preprocessed-connectomes-project.github.io/quality-assessment-protocol). These outliers were independently inspected by three study investigators (JW, PW, IHH), and retained if all investigators rated the images as being acceptable. A total of 8, 2 and 3 images were excluded from a total of 11 participants at baseline, year one and three, respectively.
Brain-predicted brain age difference (brain-PAD)
Brain age was derived from raw T1-weighted MRI using Cole and colleagues’ trained model (https://github.com/james-cole/brainageR) 3. The model first initiates voxel-level pre-processing using the Statistical Parametric Mapping (SPM12) toolbox (University College London, London, UK). This involves segmenting the raw images into grey matter (GM), white matter (WM) volume, and cerebrospinal fluid, and non-linear spatial normalisation to a template image, using the DARTEL (“Diffeomorphic Anatomical Registration using Exponentiated Lie algebra” 24) algorithm. Tissue volume encoding was preserved in the template-normalised images by multiplying each voxel with the Jacobian determinant (relative volume of tissue before and after warping 25. Images were resampled and smoothed using a voxel size of 1.5mm and a Gaussian spatial smoothing kernel of 4mm at full-width at half-maximum (FWHM), respectively 3.
Normalised tissue maps were combined, reduced to 435 principal components (which account for 80% of the total variance of chronological age), and used to train a brain age model using a cohort of 3377 healthy adults (aged 18–92 years), sourced from seven publicly available datasets (further details provided https://github.com/james-cole/brainageR). The model was then inverted to estimate brain age for each of our study participants at each timepoint using a Gaussian process regression algorithm. Brain-PAD was defined as the deviation of brain age from chronological age. A positive brain-PAD (i.e., older brain age relative to one’s chronological age) represents advanced brain ageing, while a negative brain-PAD (i.e., younger brain age relative to chronological age), is considered a sign of preserved brain ageing. To account for a potential age bias on brain age estimates (i.e., overestimation in younger, while underestimated in older participants), we include chronological age as a covariate in the adjusted model (refer to ‘Statistical analysis’) 26.
Covariates
On enrolment into the parent ASPREE clinical trial, participants completed a number of questionnaires and physical assessments 27. Hypertension was defined by a systolic blood pressure (BP) ≥ 140mmHg and/or a diastolic BP ≥ 90mmHg, or participant self-report of treatment for high BP 27. Diabetes mellitus was defined by a self-reported diagnosis, treatment for diabetes, or a fasting blood glucose ≥ 126mg/dL 27. Women diagnosed with dyslipidaemia were taking cholesterol-lowering medications or had a serum cholesterol ≥ 212 mg/dL 27. Impaired renal function was defined by an estimated glomerular filtration rate (eGFR) <60mL/min/1.73m2. The eGFR values were calculated according to the Chronic Kidney Disease Epidemiology (CKD-EPI) collaboration equation 28.
Statistical analysis
The mean and standard deviation (SD) were used to describe normally distributed continuous data; otherwise the median and interdercile range (IDR) are reported. Categorical data are summarised using the frequency and percentage. The minimum values were imputed for participants with serum concentrations below the assay limit of detection [oestrone (n=2), testosterone (n=8) and DHEA (n=3)], or missing data (n=13), using the LOD/√2 29. Sex hormones concentrations were divided into tertiles for interpretability. At baseline, with 80% power and a sample size of 69 per tertile, we can detect a minimal difference in brain-PAD of 0.34 standardized beta coefficient.
Linear mixed modelling was used to investigate the association between sex hormones and brain-PAD. To examine group differences at baseline, models included a fixed effect of hormones, while the rate of change in brain-PAD was analysed using a fixed effect of time (i.e., annual visits of 0 [baseline], 1 and 3 years, treated on a continuous scale), and the interaction between time and hormones. Models included random intercept and random slope.
Multivariable mixed modelling was performed to control for potential confounding due to chronological age, body mass index (BMI), smoking, diabetes, hypertension, dyslipidaemia, impaired renal function, and aspirin (placebo vs. aspirin). Aspirin was chosen due to ASPREE being a randomised control trial of low-dose aspirin, while all other covariates were selected based on prior knowledge relating to postmenopausal hormones 19,30 and brain age 31. Models were not adjusted for polynomial age as this was not significant in the final models, nor did it improve the overall model fit of the models oestrone (likelihood ratio [LR] χ2(2)=1.73, p=0.42), testosterone (LR χ2(2)=1.81, p=0.41), DHEA (LR χ2(2)=1.91, p=0.38), and SHBG (LR χ2 (2)=1.04, p=0.79). Analyses were performed using Stata software, version 17.0 (StataCorp).
RESULTS
Study participants
The characteristics and sex hormone concentrations of the 207, mainly white, women (97.1%) included in this study are summarised in Table 1. As shown, 75% were aged 70–74 years, and 32% had ever smoked. Over half had dyslipidaemia and/or hypertension (Table 1). Compared with the remaining SHOW participants who were excluded from this analysis due their therapy at the time of their blood draw, our analyses included fewer women with dyslipidaemia (69.6 vs. 77.4%, p=0.009), hypertension (59.9 vs. 73.7%, p<0.001), and impaired renal function (16.4 vs. 26.6%, p=0.002). Included participants were also younger than the excluded participants (median age of 72.5 years compared to 74.0 years, respectively [z=5.4, p<0.001]).
Table 1.
Baseline characteristics of women included in the study sample.
| Participant characteristics: | n=207a |
|---|---|
|
| |
| Chronological age (years), n (%): | |
| 70–74 | 156 (75.4) |
| 75+ | 51 (24.6) |
| a BMI (kg/m2), n(%): | |
| <24 | 69 (33.7) |
| 25–29 | 82 (40.0) |
| 30+ | 54 (26.3) |
| Ever smoked, n (%): | 67 (32.4) |
| Diabetes, n (%) | 13 (6.3) |
| Hypertension, n (%) | 124 (59.9) |
| Dyslipidaemia, n (%) | 144 (69.6) |
| Impaired renal function, n (%) | 32 (16.4) |
| b Statin, n (%) | 62 (30.0) |
| Metformin, n (%) | 5 (2.4) |
| Aspirin, n (%) | 104 (50.2) |
| Oestrone (pmol/L), median (IDR) | 151.6 (51.8, 310.6) |
| Tertile 1 (n=71) | 77.7 (2.6, 110.9) |
| Tertile 2 (n=68) | 153.5 (125.7, 192.3) |
| Tertile 3 (n=68) | 251.5 (207.1, 369.8) |
| Testosterone (nmol/L), median (IDR) | 0.3 (0.02, 0.7) |
| Tertile 1 (n=74) | 0.1 (0.02, 0.2) |
| Tertile 2 (n=65) | 0.3 (0.2, 0.4) |
| Tertile 3 (n=68) | 0.6 (0.4, 1.0) |
| DHEA (nmol/L), median (IDR) | 2.4 (0.6, 5.3) |
| Tertile 1 (n=70) | 0.9 (0.1, 1.4) |
| Tertile 2 (n=68) | 2.4 (1.8, 2.6) |
| Tertile 3 (n=69) | 4.3 (3.3, 5.4) |
| SHBG (nmol/L), median (IDR) | 41.8 (24.4, 73.0) |
| Tertile 1 (n=65) | 29.2 (20.0, 34.0) |
| Tertile 2 (n=65) | 42.0 (37.2, 47.6) |
| Tertile 3 (n=64) | 63.8 (52.0, 86.8) |
Abbreviations: BMI=body mass index; DHEA=dehydroepiandrosterone; IDR=interdecile range; SHBG=sex hormone binding globulin.
Missing data from BMI (n=2), impaired renal function (n=12).
Statin use is specific to HMG CoA reductase inhibitors, such as simvastatin, lovastatin, pravastatin, fluvastatin, atorvastatin, rosuvastatin, and pitavastin.
Brain age and brain-predicted age difference (brain-PAD)
Baseline, year one and year three brain scans were acquired at a median time point after enrolment in the ASPREE clinical trial of 14 (IDR=8, 28), 364 (IDR=347, 432) and 1,085 (IDR=1058, 1126) days, respectively. The mean brain age at the time of the baseline scan was 72.3 years, and had increased by 1.15 (95% confidence interval[CI]: 0.88, 1.42, p<0.001) and 3.10 years (95% CI: 2.78, 3.42, p<0.001) at the 1 and 3 year follow-up scans, respectively (Figure 1a). Brain age had a moderate positive association with chronological age, collected at the baseline (n=207, rho=0.50), year one (n=203, rho=0.47) and year three scan (n=187, rho=0.44; all p<0.001; Supplementary Figure 1a, c & e).
Figure 1.
Dot plot showing the distribution of brain age (a) and the brain-predicted age difference (brain-PAD; b) at baseline, one and three year follow-up.
Footnote: Mean brain age and brain-PADs at baseline, one and three years are represented by the orange line. Standard deviations are indicated by the green lines. A brain-PAD of zero defines a brain age that is equivalent to the chronological age, and is represented by the red line. Each blue dot represents a single observation.
Mean baseline and follow-up brain-PAD are summarised in Figure 1b. As shown, brain-PAD remained relatively stable over the one year (mean difference=0.17 [−0.10, 0.44]years, p=0.218) and three year follow-up periods (mean difference=0.15 [−0.16, 0.47] years, p=0.338). However, at the individual level, variability in brain ageing was observed. Brain-PAD was evaluated for age bias as brain age may be underestimated and overestimated for older and younger individuals, respectively. There was no linear association between brain-PAD and chronological age (baseline: rho=−0.02, p=0.721; year one: rho=−0.02, p=0.752; year three: rho=−0.11, p=0.124; Supplementary Figure 1b, d & f).
Association between sex hormones and brain-PAD at baseline, and over time
Results of the linear mixed models investigating the cross-sectional and longitudinal associations between sex hormones and brain-PAD are shown in Table 2. Unadjusted analyses indicated women in the highest DHEA tertile had a higher baseline brain-PAD compared with women in the lowest tertile (p=0.043). This association was not significant after adjustment for age, BMI, smoking, diabetes, hypertension, dyslipidaemia, impaired renal function, and treatment allocation. Testosterone, oestrone, and SHBG were not associated with brain-PAD at baseline, nor were any of the sex hormones or SHBG associated with the change in brain-PAD over 3 years (Table 2).
Table 2.
Linear mixed models analysing the association between baseline sex hormones and longitudinal brain-predicted age difference (brain-PAD) (n=193).
| Baseline | Longitudinal | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| aUnadjusted | bAdjusted | aUnadjusted | bAdjusted | |||||||
| Sex hormone | b (95% CI) | β | p | b (95% CI) | β | p | b (95% CI) | p | b (95% CI) | p |
|
| ||||||||||
| Oestrone: | ||||||||||
| Tertile 2 | 0.55 (−1.20, 2.30) | 0.05 | 0.537 | 0.87 (−0.90, 2.64) | 0.08 | 0.337 | −0.02 (−0.27, 0.24) | 0.884 | −0.02 (−0.27, 0.23) | 0.877 |
| Tertile 3 | −0.16 (−1.91, 1.60) | −0.01 | 0.860 | 0.58 (−1.23, 2.38) | 0.05 | 0.532 | 0.02 (−0.24, 0.27) | 0.901 | 0.02 (−0.24, 0.28) | 0.878 |
| Testosterone: | ||||||||||
| Tertile 2 | −0.90 (−2.65, 0.85) | −0.08 | 0.314 | −0.40 (−2.21, 1.42) | −0.04 | 0.668 | 0.03 (−0.22, 0.28) | 0.807 | −0.03 (−0.29, 0.23) | 0.843 |
| Tertile 3 | 0.22 (−1.51, 1.95) | 0.02 | 0.804 | 0.70 (−1.08, 2.48) | 0.06 | 0.440 | 0.02 (−0.23, 0.27) | 0.859 | 0.01 (−0.25, 0.27) | 0.951 |
| DHEA: | ||||||||||
| Tertile 2 | 0.76 (−0.99, 2.50) | 0.07 | 0.395 | 0.85 (−0.92, 2.62) | 0.08 | 0.346 | −0.13 (−0.38, 0.12) | 0.313 | −0.15 (−0.41, 0.10) | 0.236 |
| Tertile 3 | 1.79 (0.06, 3.53) | 0.16 | 0.043* | 1.65 (−0.13, 3.43) | 0.15 | 0.069 | 0.07 (−0.18, 0.32) | 0.586 | 0.05 (−0.21, 0.30) | 0.704 |
| SHBG: | ||||||||||
| Tertile 2 | −0.44 (−2.28, 1.39) | −0.04 | 0.637 | −0.54 (−2.41, 1.32) | −0.05 | 0.569 | −0.01 (−0.27, 0.25) | 0.951 | −0.04 (−0.31, 0.35) | 0.789 |
| Tertile 3 | −1.23 (−3.07, 0.61) | −0.11 | 0.189 | −1.41 (−3.30, 0.47) | −0.13 | 0.141 | 0.20 (−0.07, 0.46) | 0.140 | 0.21 (−0.06, 0.48) | 0.127 |
p<0.05.
Abbreviations: DHEA=dehydroepiandrosterone; SHBG=sex hormone binding globulin. Linear mixed models include a fixed effect of time (baseline, one and three year visit, not shown in the table above) and baseline sex hormone (as represented by ‘Baseline’). The 2-way interactions between sex hormones and time were also included, and represent the effects of hormones on the change in brain-PAD (as indicated by ‘Longitudinal’). Coefficients for tertiles 2 and 3 are relative to women with the lowest values (tertile 1).
Total sample size of 207 women were included in each independent model, with the exception of SHBG, which includes data from 194 participants. Models were fully adjusted for chronological age, smoking, body mass index, diabetes, hypertension, dyslipidaemia, impaired renal function and aspirin as covariates.
Missing data were from BMI (n=2) and impaired renal function (n=12). As such, adjusted models include 193 women for all sex hormones, except SHBG, which includes 180.
DISCUSSION
This study explored whether circulating sex hormone concentrations in older women were associated with brain-PAD at baseline, and with longitudinal change in brain-PAD over three years. From our cohort of initially cognitively unimpaired women, we observed no robust, statistically significant associations between sex hormones and brain-PAD.
To our knowledge this is the first study to explore the association between oestrone and structural brain age in older women. An earlier cross-sectional study reported higher serum oestradiol to be associated with a younger brain age in a large cohort of middle aged and older women, who were not carriers of the APOE e4 genotype 15. However, oestradiol was measured by immunoassay, which lacks precision for the measurement of the very low concentrations of oestradiol in postmenopausal women, rendering this finding uncertain 32. Added to this, the number of women with an oestradiol level below the LOD in that study was not reported 15. Over two-thirds of our sample had an oestradiol concentration below the LOD using LC-MS/MS whereas 99% had measurable serum oestrone 19. Therefore, oestrone, as the primary circulating oestrogen in postmenopausal women and a robust surrogate for oestradiol in older women, was used in the present study 8.
Our null finding for testosterone corroborates the few neuroimaging studies that also found no association between testosterone and whole brain volume 33 or regional brain volume 11,33, irrespective of the approach used to quantify concentrations from blood serum (radio- and enzyme-linked-immunoassays). Conversely, two studies that investigated the effects of SHBG did find an association with brain ageing 33,34, which does not align with our results. However, unlike our study, Kim et al. 33 compared concentrations of plasma SHBG between a small sample of young and older women, cross-sectionally, while Xu and colleagues 34 found a statistically significant longitudinal association between higher plasma SHBG concentrations and a faster rate of decline in hippocampal volume in a pooled sample of older men and women.
The lack of significant association between sex hormone concentrations and brain-PAD should not be interpreted as evidence of sex hormones having no role in brain ageing. Rather our results may simply reflect the tenuous relationship between blood sex hormone concentrations and their target tissue effects. Concentrations of circulating DHEA and its sulphate (DHEAS) do not change at menopause 35, and are in abundant supply. These hormones, along with androstenedione from the adrenal glands, are essential for oestrogen and testosterone biosynthesis after menopause 36. Although we found an association between DHEA and brain ageing in unadjusted analysis, this did not remain after taking into account confounding factors such as age and metabolic covariates (e.g. body mass index), which are associated with DHEA, brain ageing or both 19,31,36. In contrast, postmenopausal oestrogen and testosterone concentrations are much lower than those of DHEA and their production is dependent on the local activity of biosynthetic enzymes in peripheral tissues, including the brain. Therefore, the blood concentrations of oestrone, oestradiol and testosterone are a consequential spill-over from their production in peripheral tissues 37, so that the circulating concentrations of these hormones may not reliably reflect their effects in target tissues.
This study has a number of strengths, including the longitudinal design. Data were collected from a large cohort of 207 cognitively unimpaired, older women (aged 70 to 88 years), recruited from the community 38. The brain age model was trained on multiple datasets that cover a range of geographical locations, scanner strengths and data acquisitions 6, and was validated on a larger independent test cohort 3. Images underwent rigorous quality assessments to reduce bias. Sex steroids were measured by the gold standard LC-MS/MS 19, which enables accurate measurement of several sex steroids simultaneously in a single sample 39. This is particularly important for the measurement of testosterone as immunoassays lack sensitivity and precision at the low concentrations in women, relative to men 40. Lastly, women known to be taking therapies that interfere with endogenous sex hormone concentrations were excluded from the analysis 19.
There are several limitations to be addressed. Despite the longitudinal design, this study was limited to a short follow-up period of three years. The high proportion of women of European descent, who were both relatively healthy, and cognitively normal at inclusion, limits the generalisability of our findings to other racial or ethnic populations. We did not consider key aspects to postmenopausal health, including age at menopause onset, or the length of reproductive lifespan on brain ageing, which have been considered a primary explanation for inconsistencies among observational and clinical studies 15. Moreover, we were unable to explore the potential associations between gonadotropins and brain ageing, as FSH and LH concentrations were not measured in ASPREE participants 11,12,33. We did not consider other medications or comorbidities (including depression), which may also influence sex hormone levels, and could in turn be associated with brain ageing.
CONCLUSION
This study found no evidence of an association between circulating sex hormones and brain ageing. However, the relatively small sample size, and short follow-up period of three years do limit this study. Given existing evidence suggests sex hormones may play a role in brain and cognitive ageing 10,11,33, further studies of circulating sex hormones and brain ageing in postmenopausal women are warranted.
Supplementary Material
ACKNOWLEDGEMENTS
In addition to the general practitioners and medical clinics who supported participants during the trial process, the authors would like to thank the participants of the Australian ASPREE, NEURO and SHOW sub-studies for volunteering their time. We would also like to acknowledge the work of the ASPREE field staff, and those at the Monash Biomedical Imaging Facility.
ASPREE received funding from the National Institute on Aging and the National Cancer Institute at the National Institutes of Health (U01AG029824), Monash University, the Victorian Cancer Agency, and the Australian National Health and Medical Research Council (NHMRC, grant numbers 334047 and 1127060). NEURO received funding from NHMRC (1086188) and support from Monash Biomedical Imaging. Jo Wrigglesworth is the recipient of a Research Training Program stipend, awarded by Monash University and the Australian government. Ian H. Harding is supported by an NHMRC Fellowship (APP1106533). John J. McNeil receives an investigator grant funded through a NHMRC Leadership Fellowship (APP1173690). Joanne Ryan is funded by a NHMRC Dementia Research Leader Fellowship (APP1135727). Funders did not direct the conduct of this study, nor the decision to publish these findings.
Footnotes
CONFLICT OF INTERESTS
Dr Davis reports having received honoraria from Besins Healthcare, BioFemme, Biosyent, Southern Star Research, Lawley Pharmaceuticals, and Que Oncology. She served on Advisory Boards for Mayne Pharma, Astellas Pharmaceuticals, Roche Diagnostics, Theramex and Abbott Pharmaceuticals, and has been an institutional investigator for Que Oncology and OvocaBio. All other authors declare no competing interests.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
REFERENCES
- 1.Lopez-Otin C, Blasco MA, Partridge L, Serrano M, Kroemer G. The hallmarks of aging. Cell. 2013;153(6):1194–1217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Coupe P, Manjon JV, Lanuza E, Catheline G. Lifespan Changes of the Human Brain In Alzheimer’s Disease. Sci Rep. 2019;9(1):3998. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Cole JH, Ritchie SJ, Bastin ME, et al. Brain age predicts mortality. Mol Psychiatry. 2018;23(5):1385–1392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Franke K, Ziegler G, Kloppel S, Gaser C, Alzheimer’s Disease Neuroimaging I. Estimating the age of healthy subjects from T1-weighted MRI scans using kernel methods: exploring the influence of various parameters. Neuroimage. 2010;50(3):883–892. [DOI] [PubMed] [Google Scholar]
- 5.Wrigglesworth J, Yaacob N, Ward P, et al. Brain-predicted age difference is associated with cognitive processing in later-life. Neurobiology of Aging. 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wrigglesworth J, Harding IH, Ward P, et al. Factors Influencing Change in Brain-Predicted Age Difference in a Cohort of Healthy Older Individuals. Journal of Alzheimer’s Disease Reports. 2022(Preprint):1–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Armstrong NM, An Y, Beason-Held L, et al. Sex differences in brain aging and predictors of neurodegeneration in cognitively healthy older adults. Neurobiol Aging. 2019;81:146–156. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Davis SR, Martinez-Garcia A, Robinson PJ, et al. Estrone Is a Strong Predictor of Circulating Estradiol in Women Age 70 Years and Older. The Journal of Clinical Endocrinology & Metabolism. 2020;105(9):e3348–e3354. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hogervorst E, Combrinck M, Smith AD. Testosterone and gonadotropin levels in men with dementia. Neuro Endocrinol Lett. 2003;24(3–4):203–208. [PubMed] [Google Scholar]
- 10.Pike CJ, Carroll JC, Rosario ER, Barron AM. Protective actions of sex steroid hormones in Alzheimer’s disease. Front Neuroendocrinol. 2009;30(2):239–258. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Lee JH, Byun MS, Yi D, et al. Sex-specific association of sex hormones and gonadotropins, with brain amyloid and hippocampal neurodegeneration. Neurobiology of aging. 2017;58:34–40. [DOI] [PubMed] [Google Scholar]
- 12.Udeh-Momoh C, Watermeyer T, Female Brain H, Endocrine Research c. Female specific risk factors for the development of Alzheimer’s disease neuropathology and cognitive impairment: Call for a precision medicine approach. Ageing Res Rev. 2021;71:101459. [DOI] [PubMed] [Google Scholar]
- 13.Xiong J, Kang SS, Wang Z, et al. FSH blockade improves cognition in mice with Alzheimer’s disease. Nature. 2022;603(7901):470–476. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ryan J, Artero S, Carriere I, et al. Brain volumes in late life: gender, hormone treatment, and estrogen receptor variants. Neurobiol Aging. 2014;35(3):645–654. [DOI] [PubMed] [Google Scholar]
- 15.de Lange AG, Barth C, Kaufmann T, et al. Women’s brain aging: Effects of sex-hormone exposure, pregnancies, and genetic risk for Alzheimer’s disease. Hum Brain Mapp. 2020;41(18):5141–5150. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Franke K, Hagemann G, Schleussner E, Gaser C. Changes of individual BrainAGE during the course of the menstrual cycle. Neuroimage. 2015;115:1–6. [DOI] [PubMed] [Google Scholar]
- 17.Luders E, Gingnell M, Poromaa IS, Engman J, Kurth F, Gaser C. Potential Brain Age Reversal after Pregnancy: Younger Brains at 4–6Weeks Postpartum. Neuroscience. 2018;386:309–314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ward SA, Raniga P, Ferris NJ, et al. ASPREE-NEURO study protocol: A randomized controlled trial to determine the effect of low-dose aspirin on cerebral microbleeds, white matter hyperintensities, cognition, and stroke in the healthy elderly. Int J Stroke. 2017;12(1):108–113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Davis SR, Bell RJ, Robinson PJ, et al. Testosterone and Estrone Increase From the Age of 70 Years: Findings From the Sex Hormones in Older Women Study. J Clin Endocrinol Metab. 2019;104(12):6291–6300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Aspree Investigator Group. Study design of ASPirin in Reducing Events in the Elderly (ASPREE): a randomized, controlled trial. Contemp Clin Trials. 2013;36(2):555–564. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Teng EL, Chui HC. The Modified Mini-Mental State (3MS) examination. J Clin Psychiatry. 1987;48(8):314–318. [PubMed] [Google Scholar]
- 22.Islam RM, Bell RJ, Handelsman DJ, et al. Longitudinal changes over three years in sex steroid hormone levels in women aged 70 years and over. Clin Endocrinol (Oxf). 2021;94(3):443–448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Esteban O, Birman D, Schaer M, Koyejo OO, Poldrack RA, Gorgolewski KJ. MRIQC: Advancing the automatic prediction of image quality in MRI from unseen sites. PLoS One. 2017;12(9):e0184661. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Ashburner J A fast diffeomorphic image registration algorithm. Neuroimage. 2007;38(1):95–113. [DOI] [PubMed] [Google Scholar]
- 25.Ashburner J Computational anatomy with the SPM software. Magn Reson Imaging. 2009;27(8):1163–1174. [DOI] [PubMed] [Google Scholar]
- 26.de Lange AG, Cole JH. Commentary: Correction procedures in brain-age prediction. Neuroimage Clin. 2020;26:102229. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.McNeil JJ, Woods RL, Nelson MR, et al. Baseline Characteristics of Participants in the ASPREE (ASPirin in Reducing Events in the Elderly) Study. J Gerontol A Biol Sci Med Sci. 2017;72(11):1586–1593. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Levey AS, Stevens LA, Schmid CH, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009;150(9):604–612. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Handelsman DJ, Ly LP. An Accurate Substitution Method To Minimize Left Censoring Bias in Serum Steroid Measurements. Endocrinology. 2019;160(10):2395–2400. [DOI] [PubMed] [Google Scholar]
- 30.Sabbatini AR, Kararigas G. Estrogen-related mechanisms in sex differences of hypertension and target organ damage. Biol Sex Differ. 2020;11(1):31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Cole JH. Multimodality neuroimaging brain-age in UK biobank: relationship to biomedical, lifestyle, and cognitive factors. Neurobiol Aging. 2020;92:34–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Santen RJ, Pinkerton JV, Liu JH, et al. Workshop on normal reference ranges for estradiol in postmenopausal women, September 2019, Chicago, Illinois. Menopause. 2020;27(6):614–624. [DOI] [PubMed] [Google Scholar]
- 33.Kim GW, Park K, Jeong GW. Effects of Sex Hormones and Age on Brain Volume in Post-Menopausal Women. J Sex Med. 2018;15(5):662–670. [DOI] [PubMed] [Google Scholar]
- 34.Xu W, Su BJ, Shen XN, et al. Plasma sex hormone-binding globulin predicts neurodegeneration and clinical progression in prodromal Alzheimer’s disease. Aging (Albany NY). 2020;12(14):14528–14541. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Davison SL, Bell R, Donath S, Montalto JG, Davis SR. Androgen levels in adult females: changes with age, menopause, and oophorectomy. J Clin Endocrinol Metab. 2005;90(7):3847–3853. [DOI] [PubMed] [Google Scholar]
- 36.Davis SR, Panjari M, Stanczyk FZ. Clinical review: DHEA replacement for postmenopausal women. J Clin Endocrinol Metab. 2011;96(6):1642–1653. [DOI] [PubMed] [Google Scholar]
- 37.Simpson ER, Davis SR. Minireview: aromatase and the regulation of estrogen biosynthesis--some new perspectives. Endocrinology. 2001;142(11):4589–4594. [DOI] [PubMed] [Google Scholar]
- 38.Szucs D, Ioannidis JP. Sample size evolution in neuroimaging research: An evaluation of highly-cited studies (1990–2012) and of latest practices (2017–2018) in high-impact journals. Neuroimage. 2020;221:117164. [DOI] [PubMed] [Google Scholar]
- 39.Walters KA, Eid S, Edwards MC, et al. Steroid profiles by liquid chromatography-mass spectrometry of matched serum and single dominant ovarian follicular fluid from women undergoing IVF. Reprod Biomed Online. 2019;38(1):30–37. [DOI] [PubMed] [Google Scholar]
- 40.Rosner W, Vesper H, Endocrine S, et al. Toward excellence in testosterone testing: a consensus statement. J Clin Endocrinol Metab. 2010;95(10):4542–4548. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.