Altered brain morphology and functional connectivity in postmenopausal women: automatic segmentation of whole-brain and thalamic subnuclei and resting-state fMRI

Subjects

Twenty-one postmenopausal women (mean age: 55.3 ± 2.5 years) and 21 premenopausal women (mean age: 39.8 ± 7.6 years) underwent MRI. All women with right-handedness were recruited vis advertisements.

The premenopausal group was selected based on several criteria [1]: 1) absence of a menopause diagnosis as determined by the Stages of Reproductive Aging Workshop (STRAW) +10 criteria; and 2) regular menstrual bleeding. Ovulation day was estimated using the rhythm method. Participants with a history of psychiatric or neurological illnesses were excluded. Those who had undergone any hormonal or steroid treatment and those who had used oral contraceptives in the month preceding the study were also excluded. We included 21 premenopausal women in the premenstrual phase, 10 to 19 days prior to their expected period.

The postmenopausal group was selected based on the following inclusion criteria [1, 29]: 1) those with a confirmed diagnosis of menopause according to STRAW +10 criteria; and 2) an absence of menstrual bleeding for more than one year. Participants who had a history of a hysterectomy or bilateral oophorectomy were excluded. Those whose follicle-stimulating hormone (FSH) levels were less than 40 μg/mL were also excluded. Similar to the premenopausal group, these who had a history of neurological illnesses and those who had received hormonal or steroid treatments or used oral contraceptives in the month prior to the study were excluded. The average duration since the onset of menopause in these women was 4.8 ± 2.5 years. In terms of demographics, the majority of the premenopausal women, 18 out of 21, were married with an average of 1.8 ± 1.0 children, while 3 were single. In contrast, all postmenopausal women were married with an average of 2.8 ± 0.7 children.

This study was approved by the Institutional Review Board (IRB) of Chonnam National University Hospital (IRB-CNUH). The experimental procedure was explained to all Participants and written informed consent was obtained. All experimental procedures and methods were performed in accordance with relevant guidelines and regulations approved by the IRB-CNUH.

Sex hormones

Levels of sex hormones including estradiol (E2), FSH, and luteinizing hormone (LH) were measured by chemiluminescent immunoassays using an ADVIA Centaur System (Bayer Healthcare, Chicago, IL, USA). Levels of estrogen, estriol (E3), free testosterone (free-T), and sex hormone-binding globulin (SHBG) were measured by radioimmunoassay using a gamma counter (Cobra 5010 Quantum, Packard Instrument Co, Meriden, CT, USA).

MRI data acquisition

MRI and fMRI images were collected on a 3.0-T Magneton Tim Trio MR Scanner (Siemens Medical Solutions, Erlangen, Germany) with an 8-channel head coil. For structural imaging, T1-weighted sagittal images were acquired using a refined three-dimensional magnetization-prepared rapid-acquisition gradient echo (3D-MPRAGE) pulse sequence with the following parameters: a repetition time (TR) of 1900 ms and an echo time (TE) of 2.35 ms. The field of view (FOV) was maintained at 256 × 256 mm² with a matrix size of 256 × 256. This protocol was designed to yield a comprehensive set of 176 slices per scan, with the number of excitations (NEX) set to 1.

Two 1-minute resting-state blood oxygen level-dependent (BOLD) scans were collected. These scans utilized a gradient echo-planar pulse sequence with the following parameters: a TR/TE of 2000 ms/30 ms, a flip angle of 90°, an FOV of 220 × 220 mm², and a matrix size of 64 × 64. The slice gap was meticulously set to 0 mm to ensure contiguous coverage and optimal signal quality.

Data processing and analysis

Anatomical (surface area and subcortical volume) and resting-state functional MR data were post-processed using FreeSurfer 6.0 software package (http://surfer.nmr.mgh.harvard.edu). Thalamic subnuclei were calculated using FreeSurfer 7.2 software package. Resting-state fMRI data analysis included runs with a temporal signal-to-noise ratio greater than 125 and a relative head motion value of less than 1 mm between one or more pairs of consecutive TRs [30].

Structural analysis

T1-weighted images were analyzed with the procedure described in our previous studies [30, 31]. We incorporated advanced motion correction techniques and intensity non-uniformity adjustments. Talairach transformation was performed for each subject’s brain, followed by removal of non-brain tissues. We conducted a detailed segmentation for cortical gray matter, subcortical white matter, and deep gray matter volumetric structures [29, 30]. The cortical surface was reconstructed through triangular tessellation at the gray matter/white matter interface and the gray matter/cerebrospinal fluid boundary, followed by topological correction [29, 30].

Thirty brain regions of interest (ROIs) were selected from both hemispheres based on previous studies [1, 6, 19, 20, 32] focused on postmenopausal women, including the superior/middle/inferior frontal cortex, superior/middle/inferior temporal cortex, superior/inferior parietal lobule, lateral/medial orbitofrontal cortex, insula, hippocampus, thalamus, putamen, and amygdala. Twenty-three thalamic subnuclei (46 ROIs) were extracted for the left and right hemispheres, including the antero-ventral (AV), latero-dorsal (LD), lateral posterior (LP), ventral anterior (VA), ventral anterior magnocellular (VAmc), ventral lateral anterior (VLa), ventral lateral posterior (VLp), ventral posterolateral (VPL), central medial (CeM), central lateral (CL), paracentral (Pc), centromedian (CM), parafascicular (Pf), reuniens/medial ventral (MV-re), mediodorsal medial magnocellular (MDm), mediodorsal lateral parvocellular (MDl), lateral geniculate (LGN), medial geniculate (MGN), limitans/suprageniculate (L-SG), pulvinar anterior (PuA), pulvinar medial (PuM), pulvinar lateral (PuL), and pulvinar inferior (PuI) (Figure 1) [29].

Resting-state fMRI analysis

Preprocessing of resting-state fMRI data was conducted using FS-FAST (the FreeSurfer Functional Analysis Stream) [30, 33]. First, functional data were realigned to correct for rigid head motion and corrected for slice time correction. These preprocessed functional data were then normalized to a common space (fsaverage). Normalized functional data were smoothed using a 3D Gaussian kernel (5-mm full-width-half-maximum). Spurious variance was reduced by nuisance regression derived from rigid body motion correction, signal regression in the white matter and cerebrospinal fluid, and global signal regression [30]. Functional data underwent a specialized low-pass filtering process at 0.08 Hz, effectively isolating neural signals from physiological-related noise. Based on reduced cortical surface areas in postmenopausal women, we conducted a voxel-wise resting-state connectivity analysis using the left medial orbitofrontal cortex (mOFC), right lateral OFC (lOFC), and right superior temporal cortex (STC) as seed regions. Resting-state functional connectivity between the seed region and the whole brain was determined via multiple regression analysis of the average time-series in each voxel [30].

Statistical analyses

Demographic characteristics and sex hormones

Independent samples t-test was used to compare the age and sex hormone levels of the premenopausal and postmenopausal women using SPSS (version 28.0, IBM, Armonk, NY, USA).

Structural analysis

The Shapiro-Wilk test was used to assess the normality of the MR data. Multivariate analysis of variance, with age and whole brain volume as covariates, was used to compare cortical surface area and subcortical volume between the two groups. The significance level was set to 0.05 (p < 0.0017) after Bonferroni’s correction for 30 ROIs to adjust for multiple comparisons. Multivariate analysis of variance, with age and whole brain volume as covariates, was used to evaluate thalamic subnuclei volumes. The significance level was set to 0.05 (p = 0.001) after Bonferroni’s correction for 46 ROIs. A Spearman’s correlation test was used to evaluate the relationship between female sex hormone levels and surface area (or thalamic subnuclei volumes).

Resting-state fMRI analysis

Multivariate analysis of variance, with age as a covariate, was used to compare functional connectivity between the two groups, which was corrected for multiple comparisons with Monte Carlo simulation using a voxel height threshold of p < 0.005 with a cluster-wise p-value (CWP) threshold of p < 0.05 [30]. A partial correlation adjusted for age was used to evaluate the relationship between left OFC-right thalamus functional connectivity and female sex hormone levels.

Age and serum sex hormone levels

There was a significant difference in age between the 2 groups (p < 0.001). Postmenopausal women had lower total estrogen (p < 0.001) and E2 levels (p < 0.001) and higher FSH (p < 0.001) and LH levels (p = 0.005) than premenopausal women (Table 1). However, there were no significant differences in E3, free-T, or SHBG between the two groups (p > 0.05, Bonferroni-corrected; Table 1).

Cortical surface area and subcortical volume

As shown in Figures 2, 3, the surface area of the left mOFC, right lOFC, and right STC was significantly reduced in postmenopausal women compared to premenopausal women (p < 0.05, Bonferroni-corrected: Table 2). None of the other 27 ROIs were significantly different between the two groups (all p > 0.05; Figure 3 and Table 2).

Although postmenopausal women did not show volume atrophy in the right thalamus, the volume of the right PuA, which is one of thalamic subnuclei, was significantly decreased (p < 0.05, Bonferroni-corrected; Figure 4 and Table 3). There was also a significant volume difference in the right PuA between the two groups (Mann-Whitney U-test; p < 0.001). None of the other 45 ROIs were significantly different between the two groups (all p > 0.05; Figure 4 and Table 3).

Correlation between structural changes and sex hormone levels

Estrogen levels were positively correlated with the surface area of the left mOFC (p = 0.35, p = 0.041), right STG (p = 0.31, p = 0.045), and right lOFG (p = 0.41, p = 0.007) and the volume of the right PuA (p = 0.32, p = 0.037), respectively (Figure 5). E2 levels were positively correlated with the surface area of the left mOFC (p = 0.42, p = 0.006) and right lOFC (p = 0.40, p = 0.008) and the volume of the right PuA (p = 0.41, p = 0.008), respectively (Figure 5). Note that no correlation between the E2 level and the surface area of the right STG was observed.

Functional connectivity analysis

Significantly lower functional connectivity between the left mOFC and right thalamus was found in postmenopausal women compared to premenopausal women (p < 0.005, Monte-Carlo corrected; Figure 6). However, no significant difference (p > 0.005) was detected in whole-brain connectivity of the other two ROIs (right lOFC or right STC) between the two groups.

Summary of main findings

In this study, postmenopausal women exhibited lower total estrogen and E2 levels and higher FSH and LH levels compared to premenopausal women. In the structural analysis, the surface areas of the left mOFC, right lOFC, and right STC were significantly lower in postmenopausal women compared to premenopausal women. In addition, the functional connectivity between the left mOFC and the thalamus in postmenopausal women was significantly lower than premenopausal women. Also, the volume of the right PuA, which is one of the important thalamic subnuclei, was significantly reduced in postmenopausal women compared to premenopausal women (Figure 4). Given that decreases in female sex hormones are considered as a sign of menopause [2, 9, 34], our findings suggest that diminished mOFC size in postmenopausal women is closely associated with the correspondent functional abnormalities.

Reduced surface areas in postmenopausal women

Postmenopausal women showed significantly lower levels of surface areas in the left mOFC and right lOFC compared to premenopausal women. The OFC is implicated in multiple cognitive processes, including inhibitory control, context memory, recency judgment, and behavior choice [35]. Decreases in female sex hormones following menopause may be linked to the left mOFC and right lOFC, further supporting the hypothesis that menopause-related hormonal changes can lead to alterations in brain morphology and cognitive dysfunction [10, 19]. The effects of estrogen on the structure of the prefrontal cortex, including the OFC, are relevant to the kinds of cognitive symptoms reported by women undergoing natural, medical, and surgical menopause [8, 32, 34, 36, 37]. The PFC has a relatively high density of estrogen receptors, and estrogen receptor mRNA is also highly expressed throughout the PFC [38–40]. A previous study [40] found a positive correlation between E2 levels and cortical thickness in the orbitofrontal cortex. In our study, the surface area of the left mOFC was positively correlated with estrogen and E2 levels, and the surface area of the right lOFC was positively correlated with estrogen levels. A morphometric study [6] on ET found increased volume in the OFC of postmenopausal women receiving ET compared to postmenopausal women. A similar study [32] reported significantly lower gray matter concentrations in the bilateral orbitofrontal cortex of postmenopausal women receiving ET than in postmenopausal women without ET.

We found lower surface areas in the right STC of postmenopausal women than in premenopausal women. This finding is consistent with prior evidence reporting that the STC volume was positively correlated with female sex hormones [1]. The STC has been implicated in language processing and social perception and is widely recognized as being sensitive to estrogen changes [1, 37, 38]. Estrogen exposure was positively correlated with metabolism in the STC [41, 42]. Concerning the close connection between the estrogen level and STC volume, our findings support a potential role of decreases in sex hormones following menopause due to the correspondent brain structural atrophy. However, further study is needed to elucidate the specific cognitive and emotional implications in connection with these structural changes.

Functional connectivity between mOFC and thalamus

Together with the surface area analysis, we analysis revealed significantly decreased functional connectivity between the left mOFC and right thalamus in postmenopausal women compared to premenopausal women. This finding reinforces the hypothesis that the left orbitofrontal-bilateral thalamus connectivity is associated with cognitive impairment [43]. Several studies [44–46] demonstrated key roles for orbitofrontal-thalamic connections in the occurrence and development of psychiatric symptoms and cognitive impairment. In addition, OFC has long been known to play a central role in female sex hormones. For example, a study [47] reported that activation of the OFC from sexually relevant stimuli was positively correlated with the estradiol-to-progesterone ratio in women. Also, other studies [8, 48] demonstrated positive correlations between ET and brain activity in the OFC during emotion identification tasks and negative emotional image presentations, which suggests a potential link between the OFC and female sex hormones. Diminished OFC size and functional connectivity in postmenopausal women may be associated with cognitive dysfunction caused by reduced sex hormone levels in connection with neurophysiological change following menopause.

Decreased thalamic subnuclei volume

Although postmenopausal women did not show volume atrophy in the right thalamus, the volume of right PuA, which is one of thalamic subnuclei, was significantly decreased. In addition, the volume of the right PuA was positively correlated with E2 levels. Although the association between cognitive function and estrogen in postmenopausal women is well known, little is known about thalamic subnuclei volumes. The thalamus, with its cortical and subcortical connections, is a critical node in networks supporting cognitive functions known to decline in normal aging [49]. Conversion from MCI to AD dementia was associated with reduced gray matter volume in the right thalamus, suggesting an association with worse cognitive performance [24]. Decreased left mOFC-right thalamus functional connectivity in postmenopausal women may affect specific thalamic subnuclei volumes, such as the PuA. Specifically, the pulvinar is a thalamic nucleus, which has long been considered a key structure for sensory processing and attention [50]. There is another evidence on morphological change of the PuA in patients with attention-deficit disorder [51]. Decreased PuA volume in postmenopausal women is closely related to decreases in female sex hormone levels following menopause. Our findings provide novel insight into the structural and functional changes in the brain associated with menopause.

Limitations and future directions

This study had some limitations. Firstly, our sample size was relatively small, with only 21 postmenopausal women included in the study. This small sample size may limit the generalizability of our findings and potentially introduce bias. To overcome this limitation, we used multiple-comparison correction to analyze brain volumes between the two groups. Secondly, our study groups were not matched for age, probably bringing about age-related volume alterations. Our findings should be interpreted cautiously because of the limitations related to age differences and the statistical methods used. Future studies should consider matching participants for age to control the potential effects of aging on brain volume and functional connectivity. Lastly, our study did not include a cognitive test to assess the cognitive abilities of the subjects. Given that menopause and the associated decreases in sex hormone levels have been linked to cognitive decline, the inclusion of cognitive testing would provide a more comprehensive understanding of the relationship between menopause, brain structure and function, and cognitive ability. Future research should incorporate cognitive assessments to elucidate the potential cognitive implications of the observed structural and functional brain changes in postmenopausal women.