Effects of menopause on brain structural changes in schizophrenia


Correspondence: Yoshiro Okubo, MD, PhD, Department of Neuropsychiatry, Nippon Medical School, 1-1-5 Sendagi, Bunkyo-ku, Tokyo 113-8603, Japan. Email: okubo-y@nms.ac.jp



The aim of this study was to investigate the influences of menopause on brain morphological changes in schizophrenia using magnetic resonance imaging (MRI).


Forty female schizophrenia patients, 20 premenopausal and 20 postmenopausal, and 50 female controls underwent cerebral MRI. Optimized voxel-based morphometry was performed with Statistical Parametric Mapping version 5.


Compared with controls, regional gray matter reductions in schizophrenia patients were observed in the insula, superior temporal gyrus, anterior cingulate, parahippocampal gyrus, and thalamus. Direct comparison between the patient groups showed that the gray matter of postmenopausal patients was significantly smaller when compared with premenopausal patients in the left middle frontal gyrus, and no region had significantly lower gray matter volume in premenopausal patients relative to postmenopausal patients. Significant negative correlation between gray matter volume and the interval after menopause was found in the right superior frontal gyrus in the postmenopause patient group.


Differential morphological alterations between postmenopausal and premenopausal schizophrenia patients were observed, suggesting that the female hormone plays a protective role against schizophrenia.

MANY STUDIES HAVE indicated the vulnerability to the pathological process of male schizophrenia patients relative to female patients. Female schizophrenia patients have lower antipsychotic dosage,[1] and a less malignant course of illness[2] than male patients. In addition, several studies have reported that schizophrenia has a later onset in female patients, with the first peak of onset in female patients occurring at age 25–29 years, in contrast to the peak of onset in male patients at age 20–24 years.[3] In addition, a second smaller peak of onset occurs in women after age 44 years, around perimenopause and menopause.[4, 5] Based on these findings, it had been hypothesized that estrogen exerts a protective effect against the pathological process in schizophrenia.[5]

Studies regarding gender differences of brain morphology in schizophrenia had demonstrated more vulnerability in male patients than female patients. Magnetic resonance imaging (MRI) and postmortem studies have reported a tendency of greater abnormalities in male patients,[6] and male patients had larger lateral ventricle[6] and smaller medial temporal volume, that is, hippocampus and amygdala,[7] superior temporal gyrus,[7-9] and frontal[9] and temporal[10] lobe volumes than normal controls. Although these findings related to gender differences cannot be explained only by the degree of estrogen level, it could be hypothesized that the effect of estrogen partly reduces brain morphological changes in schizophrenia.

To our knowledge the relationship between estrogen and brain morphology in schizophrenia has not been reported in previous studies, although in postmenopausal female subjects without schizophrenia, several studies have reported the effect of estrogen therapy on brain atrophy.[11-13]

If female patients are under hormonal protection that can reduce the particular brain morphological changes of this disease, it could be suggested that premenopausal patients are under similar protection relative to postmenopausal patients. This would explain why only female patients have a second peak of onset in the paramenopausal phase.

In this study, we examined the influences of menopause on brain morphological changes of schizophrenia patients by classifying them into two subgroups, postmenopausal and premenopausal, using MRI. Although there is no previous study that investigated the effect of menopause on brain morphology of schizophrenia, studying gray matter (GM) volume by voxel-based morphometry (VBM)[14] has helped detect the localized effects of clinical characteristics, such as symptoms,[15] duration of illness,[16] and antipsychotic treatment,[17] and we considered VBM suitable for this investigation. We also examined whether the volumetric changes of postmenopausal patients were related to years elapsed after menopause.



Twenty female schizophrenia patients whose interval between last menstruation and the present MRI was at least 12 months were recruited and classified as the postmenopausal patient group. For the premenopausal patient group, 20 female schizophrenia patients matched to the postmenopausal patient group by age, and who had an interval between their last menstruation and the present MRI <12 months, were recruited.

The menstrual state of all patients was checked by interview. Twelve consecutive months of amenorrhea is the epidemiological definition of menopause according to Treloar.[18]

All patients fulfilling the diagnosis of schizophrenia as defined by DSM-IV criteria were recruited from the inpatient and outpatient facilities of Asai Hospital, Chiba, Japan. Diagnoses were made by the attending psychiatrists on the basis of a review of their charts and a conventional, semi-structured interview. In the 20 premenopausal patients, schizophrenia diagnosis was as follows: paranoid subtype, n = 12; disorganized subtype, n = 1; catatonic subtype, n = 1; undifferentiated subtype, n = 1; and residual subtype, n = 5. In the 20 postmenopausal patients the diagnosis was as follows: paranoid subtype, n = 13; disorganized subtype, n = 0; catatonic subtype, n = 2; undifferentiated subtype, n = 2; and residual subtype, n = 3, based on DSM-IV criteria.

The comparison group consisted of 50 healthy female volunteers matched with the patient groups by age, exclusion being based on a history of DSM-IV axis I or axis II disorder. Patients and comparison subjects were excluded if they had a history of head injury, neurological illness, or a diagnosis of substance abuse or dependence. Demographic and clinical characteristics are listed in Table 1.

Table 1. Subject characteristics
 Schizophrenia patientsNormal controls
  1. BPRS, Brief Psychiatric Rating Scale; GAF, Global Assessment for Functioning; (N), negative symptoms; (P), positive symptoms.
Variable (mean ± SD)n = 20n = 20n = 50
Age (years)44.4 ± 8.046.8 ± 8.245.0 ± 8.7
Period after menopause (years).NA5.0 ± 5.1Unknown
Age at onset (years)22.7 ± 10.624.9 ± 8.4NA
Duration of illness (years)21.0 ± 11.625.0 ± 13.0NA
No. episodes5 ± 3.97 ± 6.2NA
GAF score46 ± 11.042 ± 10.4NA
Total BPRS score49 ± 15.747 ± 18.2NA
BPRS (P)11.7 ± 3.910.7 ± 3.9NA
BPRS (N)9.8 ± 2.910.2 ± 4.0NA
Education (years)11.2 ± 1.911.6 ± 2.1Unknown
Antipsychotics (n)   
Combined typical and atypical86NA
Dosage (mg/day, chlorpromazine equivalent)959.5 ± 618.5746.8 ± 463.5NA
Cumulative dosage (kg, chlorpromazine equivalent)7.4 ± 7.36.3 ± 4.5NA

All subjects were within the limited age range of 30–59 years. The mean age of premenopausal patients was 44.4 ± 8.0 years, postmenopausal patients, 46.8 ± 8.2 years; and normal controls, 45.0 ± 8.7 years. There were no significant differences in age between the groups according to diagnosis and menopause (analysis of variance, P = 0.63).

All patients were taking neuroleptic medication. Neuroleptic dosage in the premenopausal group was 959.5 ± 618.5 mg (chlorpromazine equivalent: atypical, n = 7; typical, n = 5; combined typical and atypical, n = 8). The dosage in the postmenopausal group was 746.8 ± 463.5 mg (chlorpromazine equivalent: atypical, n = 10; typical, n = 4; combined typical and atypical, n = 6). There were no significant differences in dosage or ratio of atypical, typical and combined therapy. Further, none of the patients was taking any hormonal therapy or other medication that could influence the level of sex hormones.

Patient symptoms were rated by Global Assessment of Functioning (GAF) and the Brief Psychiatric Rating Scale (BPRS).[19] BPRS total scores, as well as positive symptom (conceptual organization, unusual thought content, hallucinatory behavior) and negative symptom (emotional withdrawal, motor retardation, blunted affect) subscales,[20, 21] were used in the present analysis.

There were no significant differences between the patient subgroups in age at onset (two-tailed t-test, P = 0.50), duration of illness (P = 0.41), drug dosage (P = 0.24), GAF score (P = 0.45), total BPRS score (P = 0.71), BPRS positive symptom subscale (P = 0.48), negative symptom subscale (P = 0.75), or duration of education (P = 0.46). The average number of episodes, which represented the times they had been hospitalized because of worsening of their mental conditions, was five for the premenopausal patients and seven for the postmenopausal patients. This difference in number was not statistically significant (P = 0.28). Any characteristics except menopause were not significantly different between the patient subgroups.

This study was approved by the ethics committees of Nippon Medical School and Asai Hospital. After complete description of the study, all subjects gave written informed consent.

MRI acquisition

T1-weighted MRI was carried out on a 1.5-T GE Signa scanner (GE Medical Systems, Milwaukee, WI, USA). T1-weighted images were acquired in a coronal plane using an SPGR 3-D imaging sequence with the following parameters: TE = 9 ms, TR = 22 ms, flip angle = 30°, slice thickness = 1.5 mm, field of view = 25 cm, matrix = 256 × 192, voxel dimensions = 0.98 × 0.98 × 1.5 mm.

Image analysis

All images were organized for preprocessing and analyzed using SPM5 (Wellcome Department of Imaging Neuroscience, London, UK. http://www.fil.ion.ucl.ac.uk/spm/software/spm5/) running in MATLAB 2008a (MathWorks, Natick, MA, USA). We used the optimized protocol as detailed by Good et al.[22]

We used the customized template of Asai Hospital. It was created from the MRI of 120 healthy subjects (60 male, 60 female) recruited from the local community. All the subjects were within the limited age range of 30–59 years, and were physically healthy at the time of scanning, and none had a history of current or past psychiatric illness, serious head injury, serious medical or neurological illness, or substance abuse.

Each structural MRI from all present schizophrenia and healthy control subjects was segmented into GM, white matter (WM), and cerebral spinal fluid (CSF) compartments using the SPM priors. An automated brain extraction procedure that incorporated a segmentation step was used to remove non-brain tissue.[22] The extracted GM images were normalized to the created GM template. The normalization parameters obtained from this step were then applied to the original structural images in native space to reduce any contribution from non-brain voxels and afford optimal spatial normalization of GM. These normalized images were segmented into GM, WM, and CSF partitions.

Finally, all normalized, segmented GM images were smoothed with a 12-mm full width at half maximum (FWHM) isotropic Gaussian kernel.

Data analysis

The processed images were analyzed using SPM5. Volumes were compared according to two linear contrasts (more or less GM volume between subject groups).[23]

The resulting set of voxel values for each contrast constituted a statistical parametric map of the t-statistic (SPM-t). Comparisons were made between subject groups, organized by menopause and diagnosis. At first, the GM volume difference between female schizophrenia patients and female controls was investigated using SPM-t maps thresholded at P < 0.05 corrected by the family-wise error rate (FWE). Comparison was performed using an analysis of covariance (ANCOVA) model,[24] with age as covariate. We could not check the menopausal state of each of the healthy female subjects. Therefore, this first analysis was to demonstrate the overall tendency of brain morphology in female patients relative to healthy female subjects, without relation to menopause.

Then, the GM volume difference between postmenopausal patients and premenopausal patients was investigated using multivariate analysis of variance with age, antipsychotic medication, duration of illness, and BPRS negative symptoms as covariates (MANCOVA), at P < 0.001 uncorrected. The numerical value of the antipsychotic medication was the cumulative exposure calculated by the accumulation of the representative daily dosage of each year of their treatment histories. The representative dosage used was from the longest period of each year. The dosage was multiplied by the treated days of each year, and then accumulated through their treatment histories. The sum of the calculation for premenopausal patients was 7.4 ± 7.3 kg (chlorpromazine equivalent, mean ± SD), and the sum for postmenopausal patients was 6.3 ± 4.5 kg. There was no significant difference between the patient groups (P = 0.57).

Further, the relation of brain morphology of postmenopausal patients to years elapsed after menopause was examined using correlation analysis with age as covariate, at P < 0.001 uncorrected. Then we performed a region of interest analysis to investigate volume differences of the region significantly correlated with the interval after menopause in each postmenopausal patient, using the Marsbar toolbox.[25]

In the first analysis, we performed a comparison between 40 schizophrenia patients and 50 controls thresholded at P < 0.05 corrected. We used a different threshold at P < 0.001 uncorrected for the next two analyses, because these were done using a smaller sample size, that is, 20 postmenopausal patients and 20 premenopausal patients for the second analysis and 20 postmenopausal patients for the third analysis.

For visualization of group differences, the coordinates of significant voxels were converted from Montreal Neurological Institute space to Talairach and Tournoux coordinates[26] using a MATLAB conversion program written by Matthew Brett (MRC Cognition and Brain Sciences Unit, UK). Coordinates were then entered into Talairach Daemon[27] to localize results.


Brain regions with significant GM volume change in all analyses are shown in Tables 2-4 and Figs 1, 2.

Figure 1.

Voxel-based morphometry (VBM) analysis of T contrasts showing gray matter reductions in (a) female schizophrenia patients (n = 40) compared with female controls (n = 50; threshold P < 0.05, corrected); and (b) female postmenopausal patients (n = 20) compared with female premenopausal patients (n = 20; threshold P < 0.001, uncorrected). (c) VBM analysis of T contrasts showing the negative correlation between gray matter volume and the period of time after menopause in female postmenopausal patients (n = 20; threshold P < 0.001, uncorrected).

Figure 2.

Correlation between gray matter volume of the right superior frontal gyrus and the period of time after menopause in female postmenopausal patients (n = 20).

Table 2. SPM5 regional GM reduction in patients relative to controls
 Schizophrenia patients < Controls
Anatomical region[Area]aTCoordinateCluster volume (mm3)
  1. aBrodmann area. GM, gray matter; Lt., left hemispheric; Rt., right hemispheric.
Superior temporal gyrus[41]Lt.5.04−46−248390
Parahippocampal gyrus[34]Lt.4.98−184−18130
Anterior cingulate[25]Rt.6.2722−62380
Table 3. SPM5 regional GM reduction vs menopause in schizophrenia
 Postmenopausal patients < Premenopausal patients
TCoordinateCluster volume (mm3)
Anatomical region[Area]axyz
  1. aBrodmann area. GM, gray matter; Lt., left hemispheric; Rt., right hemispheric.
Middle frontal gyrus[10]Lt.4.24−364010120
Table 4. SPM5 negative correlation between GM volume and period of time after menopause
 CoordinateCluster volume (mm3)
Anatomical region[Area]aTxyz
  1. aBrodmann area. GM, gray matter; Rt., right hemispheric.
Superior frontal gyrus[8]Rt.4.83203838340

Female schizophrenia patients vs female controls

The regional GM of female schizophrenia patients was significantly smaller when compared with female control subjects (P < 0.05 corrected). Significant volume reductions were observed in the bilateral insula, left superior temporal gyrus, bilateral anterior cingulate, left parahippocampal gyrus, and left thalamus (Table 2; Fig. 1a).

No region had significantly lower GM volume in female controls relative to female schizophrenia patients.

Premenopausal patients vs postmenopausal patients

The regional GM of postmenopausal patients was significantly smaller when compared with premenopausal patients at the left middle frontal gyrus (Table 3; Fig. 1b; P < 0.001 uncorrected). In contrast, no region had significantly lower GM volume in premenopausal patients compared to postmenopausal patients.

Correlation between brain morphology of postmenopausal patients and interval after menopause

Correlation between brain morphology of postmenopausal patients and the interval after menopause is shown in Table 4 and Figs 1(c),2. The regional brain volume of the right superior frontal gyrus was significantly correlated with the interval after menopause: the longer the interval, the greater the volume deficit in this region (P < 0.001 uncorrected). There was no significant correlation between longer postmenopausal interval and less volume deficit.


As a first step, we examined regional GM changes in female schizophrenia patients in relation to healthy female subjects by conducting VBM. The regions with remarkable GM volume reduction in patients were the insula, superior temporal gyrus, anterior cingulate, parahippocampal gyrus, and thalamus. The changes were predominantly found on the left side.

A meta-analysis reviewing 31 studies of VBM for GM volume in schizophrenia showed that patients had reduced GM density relative to control subjects in the bilateral insula, anterior cingulate, left postcentral gyrus, left parahippocampal gyrus, left middle frontal gyrus, and thalamus.[28] Another meta-analysis reviewed 15 studies of VBM for GM and WM volume in schizophrenia,[29] reporting that the areas significantly reduced in schizophrenia patients in >50% of the studies were the bilateral superior temporal gyrus, left medial frontal gyrus, left inferior frontal gyrus, and left parahippocampal gyrus.[29]

These two meta-analyses identified the morphology trait of schizophrenia using VBM. Generally, the present findings regarding patients versus controls are consistent with those of these meta-analyses in terms of locations and laterality.

We then examined the GM volume difference between postmenopausal patients and premenopausal patients. Direct comparison between the patient groups indicated morphological change in the left middle frontal gyrus. In addition, a significant correlation between brain morphology and interval after menopause was found in the right superior frontal gyrus. These regions are not identical, but both are located in the prefrontal area.

Volume reduction in the prefrontal area in schizophrenia patients has been observed in past studies, and morphological change in the prefrontal area has been indicated in relation to negative symptoms and cognitive impairment.[30] Additionally, a study regarding the relationship between illness duration and brain volume indicated that the right middle frontal cortex is particularly vulnerable to the long-term effect of schizophrenia.[16] According to these reports, the present results may indicate that postmenopausal patients are at a more advanced stage of progression.

A study on gender differences of brain morphology in schizophrenia reported that GM reductions in prefrontal areas were found predominantly in male patients.[9] It may be assumed that the duration of the low estrogen condition played a role in the morphological change in this area.

In the studies on menopause age of healthy subjects, the estimated median age at last menstruation period lies between 50 and 52 years.[31-33] A cross-sectional study concerning natural menopause of Japanese women reported that the median age of menopause was 50.54 years.[34]

The age of menopause of the postmenopausal patients in the present study, 41.8 ± 9.0 years, seemed considerably earlier than that of the healthy subjects. A major reason for this might be an antipsychotic medication-induced side-effect. Because of the very early age at menopause, the postmenopausal patient group in this study may not be a representative sample. This is a limitation of the study.

A study concerning the effect of antipsychotic-induced hyperprolactinemia in schizophrenia patients on antipsychotic medication showed that there was a significant inverse relationship between the prolactin levels in female patients and estradiol levels, and when female patients with hyperprolactinemia were compared to those with a normal range of prolactin, only estradiol levels were significantly different between the two groups with regard to measured hormone levels (estradiol, testosterone, progesterone, luteinizing hormone, follicle-stimulating hormone, and thyroid-stimulating hormone).[35]

According to that study, it could be assumed that the young patients in the postmenopausal group with probable hyperprolactinemia had low estrogen levels in the present study. The gradual decline of ovarian estrogen production begins in the years prior to menopause, although a dramatic decline in plasma estrogen level occurs at the final menstrual cycle.[36] Research concerning gender differences and age of onset in schizophrenia has shown a second peak of onset in female patients in the age range 45–54 years, and it was hypothesized that the vulnerability threshold for schizophrenia is raised in women until menopause due to the effect of estrogen.[37] Therefore, a gradual loss of the protective role of estrogen against brain morphological change may also begin prior to menopause.

In healthy subjects, according to studies investigating the effect of estrogen on brain morphology in postmenopausal women, some reported adverse effects of hormone therapy associated with greater atrophy in the hippocampal region[38] or in the putamen,[39] while several others demonstrated that estrogen therapy slows age-related GM loss in frontal cortices,[11, 13] temporal, parietal, and occipital cortices,[11] cerebellum,[11, 13] and the hippocampal region,[11, 12] findings partially in agreement with the present ones. Therefore, it is suggested that the effect of menopause on brain structural changes in schizophrenia patients could be attributable to the loss of the protective effect of estrogen against the pathophysiology of schizophrenia.

A limitation of the present study is that blood concentrations of sex hormones were not measured. Therefore, the present results do not represent the effects of sex hormones but rather the effects of menopause. In addition, because neither hormone levels nor menstrual state were checked in the healthy controls, the main purpose the patients versus controls analysis was to demonstrate that the overall tendency of the present patients would reflect the previous studies concerning the brain morphology of schizophrenia.

In conclusion, volumetric comparisons showed differential morphological alterations due to female hormonal change in schizophrenia. Postmenopausal patients had more GM volume reduction than premenopausal patients at the left middle frontal gyrus. In addition, there was significant correlation between brain morphology and interval after menopause in the right superior frontal gyrus. These results support the hypothesis of the protective role of estrogen against schizophrenia.


Professor Hidenori Suzuki of the Department of Pharmacology, Nippon Medical School and Noriaki Yahata, PhD of the Department of Neuropsychiatry, Graduate School of Medicine, University of Tokyo are gratefully acknowledged. This study was supported by a Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Culture, Sports, Science and Technology (19390308), and a Health and Labor Sciences Research Grant for Research on Psychiatric and Neurological Diseases and Mental Health (H22-seishin-ippan-002) from the Japanese Ministry of Health, Labor and Welfare. All authors have no conflicts of interest.