|Home | About | Journals | Submit | Contact Us | Français|
Understanding sex differences in stress regulation has important implications for understanding basic physiological differences in the male and female brain and their impact on vulnerability to sex differences in chronic medical disorders associated with stress response circuitry. In this fMRI study, we demonstrated that significant sex differences in brain activity in stress response circuitry were dependent on women's menstrual cycle phase. Twelve healthy Caucasian premenopausal women were compared to a group of healthy men from the same population, based on age, ethnicity, education, and right-handedness. Subjects were scanned using negative valence/high arousal versus neutral visual stimuli that we demonstrated activated stress response circuitry (amygdala, hypothalamus, hippocampus, brainstem, orbitofrontal and medial prefrontal cortices (OFC and mPFC), and anterior cingulate gyrus (ACG). Women were scanned twice based on normal variation in menstrual cycle hormones (i.e., early follicular (EF) compared with late follicular-midcycle menstrual phases (LF/MC)). Using SPM8b, there were few significant differences in BOLD signal changes in men compared to EF women, except ventromedial (VMN) and lateral (LHA) hypothalamus, left amygdala, and ACG. In contrast, men exhibited significantly greater BOLD signal changes compared to LF/MC women on bilateral ACG and OFC, mPFC, LHA, VMN, hippocampus, and periaqueductal gray, with largest effect sizes in mPFC and OFC. Findings suggest that sex differences in stress response circuitry are hormonally regulated via the impact of subcortical brain activity on the cortical control of arousal, and demonstrate that females have been endowed with a natural hormonal capacity to regulate the stress response that differs from males.
Although animal studies support the notion that hormonal differences contribute to explaining sex differences in stress response circuitry, no one has directly tested this in the human brain. This has widespread implications for clinical medicine, given that response to stress has been implicated in the vulnerability for numerous disorders with known sex differences in incidence, such as depression and anxiety disorders, cardiovascular disease and other chronic medical disorders. Further, an understanding of sex differences in the regulation of stress has important implications for understanding basic physiological differences in the male and female brain.
Brain regions in stress response circuitry include subcortical regions, central amygdala, hypothalamus [paraventricular (PVN), ventromedial (VMN) nuclei and lateral hypothalamic area (LHA)], hippocampus, and brainstem (e.g., periacqueductal gray (PAG)), and cortical regions, orbitofrontal and medial prefrontal cortex (OFC; mPFC), and anterior cingulate gyrus (ACG), referred to as playing a role in the cortical control of arousal (Keverne, 1988; McEwen and Magarinos, 1997; Price, 1999). These regions are sexually dimorphic in animal and human studies and regulate the hypothalamic-pituitary-adrenal (HPA) and –gonadal (HPG) axes (Keverne, 1988; Tobet and Hanna, 1997; Goldstein et al., 2001; Ostlund et al., 2003; Swaab, 2004; Bao et al., 2005b), suggesting hormonal regulation of stress response circuitry.
Functional brain imaging studies provide a unique opportunity to test the impact of hormones on explaining sex differences in this circuitry in in-vivo human paradigms. Behavioral and imaging studies with negative/high arousal stimuli have demonstrated significantly increased brain activations in men and women in stress response regions. Although inconsistent (Schienle et al., 2005), women demonstrated greater arousal to aversive cues than men (Lane et al., 1997; Lang et al., 1998; Bradley et al., 2001; Caseras et al., 2007), and greater signal intensity changes in AMYG, ACG, and/or OFC (Wrase et al., 2003; McClure et al., 2004), even when both sexes had similar subjective stimuli ratings (McClure et al., 2004). However, no imaging studies have directly tested the full circuitry or explanatory models for sex differences in the stress response.
We previously demonstrated in an fMRI paradigm of aversive arousal in healthy women scanned twice, increased arousal and brain activity in stress response circuitry during the early follicular (EF) menstrual phase (when estrogen and progesterone were low), compared with attenuated brain activity during late follicular/midcycle phase (LF/MC) (when estrogen was high and progesterone still relatively low), which was attributed to attenuating effects of estrogen (Goldstein et al., 2005). The purpose of the current study was to test whether hormonal effects on activations of the stress response circuitry in the women explained sex differences in activations of this circuitry. We predicted that men will look more similar to women in EF than during LF/MC, given the greater similarity of men to women’s hormonal status at EF compared to LF/MC. In this study, we demonstrated that females are endowed with a natural hormonal capacity to regulate the stress response that differs from males, and suggest that these hormones act through the impact of subcortical brain activity on the cortical control of the stress response.
Participants in the present study were selected from a community sample of women drawn by the Harvard Mood and Cycles Study (Harlow, 1999) and men from the same geographic area. Subjects were aged 35–46, right-handed, Caucasian, had at least high school education, English as their first language, no current psychiatric disorder and free of Axis I psychiatric disorders during their lifetime. The exclusion criteria also required absence of: a) substance abuse within the last six months; b) history of head injury with any documented cognitive sequelae or loss of consciousness greater than five minutes; c) neurologic disease; d) mental retardation; and e) medical illness that may significantly impair neurocognitive function. Further, women were required to have regular menstrual cycles for at least one year prior to inclusion in the present study. They knew their cycles well given that they had kept daily diaries of their cycles in the Harvard Moods and Cycles study for approximately 5 years charting their menstrual cycles, validated by blood assays obtained regularly (Harlow, 1999).
Twelve right-handed Caucasian women were systematically ascertained from the above community population study, and reported in a previous analysis in (Goldstein et al., 2005). All participants were unmedicated and free of medical, neurological, and psychiatric illness, and sensory impairments, had regular monthly cycles greater than one year, and reported no changes in the description of the flow of their menstrual cycles over the last year as well as other cycle characteristics, such as pain or menstrual discomfort. All gave informed consent and received an honorarium for participation. The mean age for the women was 43.9 years (SD = 1.4 years), ranging from 41– 46 years old. All women were normally cycling and no one was on oral contraception.
With regard to the number of weeks between scans, all women were scanned within a two-week interval between the beginning of their menstrual cycle and midcycle, even for those women whose first visit was not the beginning of her cycle. The range of menstrual cycle days on which women were scanned were days 2–14 of their menstrual cycles. There was a median menstrual cycle time of “day 3 ± 2 days” (representing early follicular when scanning occurred on days 2–5) and “day 12 ± 2 days” representing mid-cycle (when scanning was conduced on menstrual cycle days 10–14). We understand, given the lack of blood assays at the time of scanning, that some women, particularly the older women, may have been in the luteal phase rather than in midcycle, given the shorter follicular phase of some older women. However, all women had charted their cycles for three months prior to scanning to provide accurate cycle timing by self-report, given that there was a one-year period between the daily diaries of their cycle obtained in Harlow’s study and the scanning in our study. The range of time between scans in days was 10–14 days.
In the current study, thirteen men were systematically sampled from the same geographic area as the community sample of women and made comparable on age, ethnicity, education and right-handedness. The mean age of the men was 39.1 years (SD = 2.3) ranging from 35 to 43 years old. There were no significant differences between the age of the men and women and they were comparable on education, i.e., all had at least a high school education. Adjusting for age in our general linear models did not affect the results presented here.
Prior to and after each scan, each participant was administered the Profile of Mood States (POMS) and the Spielberger State-Trait Anxiety Inventory (STAI). Participants were asked to rate their current mood. The POMS is an adjective checklist of 72 items, rated by the participant on a scale from 0 to 4 (0 = does not apply; 4 = the adjective describes emotional state extremely well). Scales include Anxiety, Vigour, Fatigue, Confusion, Anger, and Depression, out of which an overall mood state score is derived as well.
The Spielberger State-Trait Anxiety Inventory is a questionnaire designed to assess an individual’s perception of current and “usual” levels of anxiety. The individual rates forty statements regarding feelings of anxiety on a scale from 1 to 4 (1 = statement poorly reflects feelings of anxiety; 4 = statement accurately reflects feelings of anxiety). Statements reflect how the individual feels in general (reflecting trait-level anxiety) and current feelings of anxiety (reflecting state-level anxiety), out of which an overall rating is also calculated. The STAI was administered pre- and post-scanning to assess state and trait-level anxiety and to test for similarities across the two menstrual cycle phases. All interviews were carried out by one of the authors (M.J.). Overall POMS and STAI scores (presented in Table 1) have been standardized to clinical populations. Any score <50 is in the low normative range. Thus, as seen in Table 1, all subjects were similar in their clinical ratings and were in the low/normal range of anxiety or mood-related symptoms (see Table 1). Further, there were no significant differences found in the change in mood state pre- to postscan, indicating that the stimuli had similar effects on all groups. Thus, controlling for mood state did not change the results.
Stimulus materials were drawn from the International Affective Picture System (IAPS) (CSEA-NIMH, 1999), a system of color pictures that were rated by large groups of subjects along the dimensions of affective valence (pleasant-unpleasant) and arousal (calm-aroused). Pictures were systematically drawn from the set according to affective valence (unpleasant and neutral) and arousal (high and low) based on normative ratings developed by Lang et al., 1993 (Lang et al., 1993). Two sets of pictures, each containing 72 images, were selected – one of unpleasant valence and high arousal and the other of neutral valence and low arousal. Faces and scenes were counterbalanced between high arousal and neutral slides. To create the 72 fixation slides, the neutral valence/low arousal slides were transformed using Fourier transforms to create a set of slides with the same physical properties of the original but without content that was readily recognizable.
During the functional scans, participants were presented with three blocks of stimuli in a counterbalanced method. One block of stimuli consisted of fixation slides presented at the rate of one every five seconds for 30 seconds. The other two blocks consisted of six arousing negative affect images and six neutral in valence and low arousal. Six different pictures were presented for five seconds in each block, which lasted 30 seconds, with blocks repeated four times during each six-minute functional scan. Three functional scans were taken, resulting in 12 blocks of each condition. In all conditions, the participant was asked to press the button under her or his index finger each time the picture being presented changed in order to ensure attention to the presented stimuli.
The women were brought in for two sessions, during early follicular menstrual cycle phase (at onset of menstrual cycle) and during midcycle. In a counterbalanced design, half of the women came in for their first session within two to three days of their predicted date of ovulation. This group then returned for the second session all within two weeks after their first visit. The other group came in for their first session within two weeks after their predicted date of ovulation and then returned approximately two weeks after that first visit. Counterbalancing visit with menstrual cycle phase controlled for potential confounding effects of “time” and/or “novelty”. The men were compared to these women at two different points in their menstrual cycle.
On arrival, the study was explained in detail, informed consent was obtained, and participants were administered the mood measures. After entry into the magnet room, participants were given the arousal task instructions and fit with earplugs to attenuate scanner noise. Participants lay prone on the scanner gurney and foam padding was placed across the forehead to prevent head motion. Physiological monitoring sensors were placed on the fingers of the left hand of the participant and the participant was given a response box with two buttons for the right hand. The head coil was then placed over the head and the participant placed into the scanner.
A Macintosh G4 Computer (Apple Computer, Cupertino, CA) running MacStim experimental presentation program (WhiteAnt Occasional Publishing, West Melbourne, Australia) was used to produce the visual stimuli. Visual stimuli were presented via an LCD projector through a custom lens onto a screen situated in the magnet bore. These stimuli were viewed through a mirror attached to the head coil. The computer was in the control room outside the Faraday shield of the scanner. Stimuli were presented as described previously. When scanning was completed, the participant was again administered the POMS and state portion of the STAI. This procedure was repeated during the second visit.
Scanning was performed with a quadrature full head coil and a 1.5 T Sonata MR scanner (Siemens Corp., Germany) modified for echo-planar imaging. A single slice 2D spoiled gradient recall (SPGR) axial localizer scan was acquired, followed by a 60 slice sagittal localizer scan (conventional T1-weighted MP-RAGE gradient echo sequence; FOV = 23×17.25 cm; matrix = 256×192; in-plane resolution = 0.90 mm; slice thickness = 2.8 mm) to orient; for subsequent scans, 15 contiguous axial slices along the AC-PC line and covering the whole brain. This was followed by an automated shim procedure to improve B0 field homogeneity and a MP-RAGE T1-weighted flow-compensated scan (FOV = 40×20 cm; matrix = 512×256; in-plane resolution = 0.78 mm; slice thickness = 7 mm; 15 slices coplanar with the functional slices). The next scan was a 15-slice T1-weighted echo planar inversion recovery sequence (TR = 20 s; TE = 40 ms; TI = 1100 ms; FOV = 40×20 cm; matrix = 256×128; in-plane resolution = 1.57 mm; slice thickness = 7mm) coplanar to the functional images for anatomic localization. Finally, a series of functional scans were acquired, using an asymmetric spin echo, T2*-weighted sequence (TR = 2000 ms; TE = 70 ms; refocusing pulse offset by −25 ms; 1 excitation; FOV = 40×20 cm; matrix = 128×64; in-plane resolution = 3.125 mm; slice thickness = 7 mm; 15 contiguous slices along the AC-PC line). This pulse sequence has excellent sensitivity to parenchymal signal changes concurrent with experimental perturbation, and reduced macrovascular sensitivity. Functional scans were acquired for 184 time points per experimental run (4 time points of scanner shimming and 180 time points of A-B-A-B blocks). Three sets of functional scans were acquired.
FMRI data were preprocessed using Statistical Parametric Mapping (SPM8b) (Wellcome Department of Cognitive Neurology, 2008) and using custom routines in MATLAB (Mathworks, Inc., 2000). Preprocessing included correction for acquisition timing across slices, bulk-head motion, and spatial smoothing with a Gaussian filter (8mm at FWHM). No individual runs exhibited head motion greater than 2.8mm across all runs. After motion correction and spatial smoothing, images for each subject were spatially normalized using nonlinear volume-based spatial normalization techniques within SPM. The template used by SPM is the standard brain template developed at the Montreal Neurological Institute (MNI). Following preprocessing, statistical analysis was performed at the single-subject level using SPM. SPM treats each voxel’s BOLD time series according to a general linear model. As this was a block design, each epoch of trials was modeled using a boxcar function convolved with a canonical hemodynamic response function. Low-frequency components of the fMRI signal were modeled as confounding covariates using a set of cosine basis functions in order to increase sensitivity to signals of interest. Specific comparisons of interest (negative valence/high arousal versus neutral valence/low arousal) were tested using linear contrasts, and SPM maps were created based on these contrasts. These contrast values (which represent estimates of the mean signal change at each voxel) were used in statistical analyses.
Results from the individual subject level were submitted to a second level analysis in which subjects were treated as a random effect. For a single group, activation was assessed using a one-sample t-test to compare the value of the contrast images against zero. For multiple groups, independent sample t-tests were used to compare the size of a particular effect between groups. To correct for multiple comparisons on the voxel-wise error rate, we adopted methods implemented by SPM that employ the theory of Gaussian random fields (Friston et al., 1994b) in order to obtain statistical maps with a single alpha level across the search volume of interest. Given that we had specific hypotheses about a set number of brain regions, we used the approach in SPM8b referred to as small volume correction (SVC), which limits voxel-wise analyses to voxels within our apriori regions of interest. Anatomically-defined regions of interest included the amygdala, hippocampus, hypothalamus, midbrain regions, orbitofrontal and ventromedial prefrontal cortices (OFC and vmPFC), and anterior cingulate gyrus (ACG). False positives were controlled using a combination of a voxel-wise height threshold (p<.005 uncorrected) and an extent threshold, that jointly resulted in a cluster-level false positive level of p<.05, corrected for multiple comparisons within the search volume using family-wise error (FWE) correction based on random field theory. The FWE rate is a more conservative correction for multiple comparisons within fMRI data. Anatomic borders were defined using overlays created using the Wake Forest University (WFU) Pickatlas (Maldjian et al., 2003) toolbox for SPM to match the borders of the hypothesized regions on the canonical brain provided in SPM8b. Using the SVC tool in SPM, activations within the borders of the overlays were identified.
Functional ROIs were defined from the main effect (across both sexes and both female time points) of the negative - neutral contrast. The statistical maps of the negative > neutral condition, including by menstrual phase, were examined at the uncorrected p <.01 level. Effect sizes were reported for all activations within our ROIs (see below). Activated clusters within the ROIs were labeled by visual inspection of our neuroanatomist (NM) and comparison to the MNI neuroanatomical atlas. The neuroanatomist (NM) was blind to study hypotheses and specific nuclei of interest prior to identification of the significant BOLD signal intensity differences in activation, thus underscoring the validity of the neural network activated by our stimuli.
After identifying clusters within the ROIs, WFU Pickatlas was used to build spherical overlays with the center of the sphere at the SPM-identified peak of activation within each cluster. These overlays were then used on the statistical maps of each individual to acquire signal change values across a specific region of interest, which was then used in general models for later statistical comparisons. Values indicate the degree of change in MR signal detected between the negative arousal condition and neutral condition and are expressed in terms of percent signal change. Average effect sizes (percentage BOLD signal change averaged across all voxels within a sphere) comparing the negative versus neutral conditions were obtained for each ROI using the REX toolbox for SPM8b (Whitfield-Gabrieli, 2009). If more than one cluster was found within a single anatomical region of interest, the signal change values for all spheres within the ROI were averaged. The effect sizes combine voxel-level data across an entire region of interest (see Table 3 results), regardless of whether the change in signal for that voxel passes the voxel-level threshold (reported in Table 2). Thus, the effect sizes are “more representative” of signal intensity values within an anatomically-identified region of interest as they represent an analysis of a priori anatomic regions of interest that do not rely only on voxel- or cluster-level significance.
The formula for calculating these effect sizes (ES) is: ES = (negative condition signal change - neutral condition signal change)/standard deviation of signal intensity value of the whole sample across all conditions. The effect sizes (standard deviation units) were then compared between the three groups (men, EF women, LF/MC women) using PROC MIXED in SAS. This procedure fits various mixed linear models and can appropriately handle repeated measurements, a function which is necessary to take into account the variance of the two different measurements, EF and LF/MC, for each woman in the current study.
Comparisons of men versus EF women and men versus LF/MC women are shown in Table 2 and Figure 1. Negative-to-neutral-stimuli in men compared to EF women showed few significant sex differences in our hypothesized regions of interest (ROIs), with none significant at a cluster level FWE-corrected p<.05 level (Table 2a). There were no significant differences in comparisons of EF women versus men.
In contrast, comparisons of negative-to-neutral-stimuli for men versus women during the late follicular/midcycle phase showed a greater number and extent of differential regions activated in men (see Table 2b). BOLD signal changes in men compared to women were greater in the anterior cingulate gyrus, OFC, medial and ventromedial PFC, medial PFC/superior frontal gyrus, amygdala, and hippocampus. Medial PFC, OFC and hippocampus remained significant based on the more conservative small volume correction using the family-wise error rate. Comparing LF/MC women to men, there were no significant signal intensity changes in the stress response circuitry.
Table 3 presents sex differences in tests of signal intensity changes in our set of functionally-defined ROIs (chosen based on our previous significant findings in women (Goldstein et al., 2005) and findings presented here). Figure 2 illustrates these differences comparing effect sizes of signal intensity changes in these regions in men versus EF women and men versus LF/MC women. As seen here, men significantly differed from EF women on bilateral ACG, amygdala, and hypothalamus (lateral hypothalamic area (LHA) and ventromedial hypothalamic nucleus (VMN)). Men significantly differed from LF/MC women on bilateral ACG and OFC, medial and ventromedial PFC, amygdala, hippocampus, LHA, VMN, and the brainstem. Findings were most pronounced (i.e, largest effect sizes) comparing cortical arousal regions, in particular medial and ventromedial PFC (with significant effect sizes of .60 and .61; see Table 3, column 7 & Figure 2).
The strongest evidence for sex differences is reflected in the formal testing of whether there were “different differences” in signal intensity comparisons between men versus EF women and men versus LC/MC women (see Table 3, last column). These tests demonstrated significant effect size differences for bilateral OFC and medial and ventromedial PFC, right hippocampus and amygdala, and PAG (see Table 3). The effect sizes were particularly large for medial and ventromedial PFC and OFC, critical brain regions implicated in the cortical control of arousal (Dougherty et al., 2004; Shin et al., 2004; Milad et al., 2007).
In this study, we provide compelling evidence that sex differences in brain activity in the stress response circuitry are accounted for by women’s hormonal status. We suggest that gonadal hormones play a role in the regulation of arousal via the impact of subcortical brain activity on the cortical control of arousal. This has significant implications for elucidating the neurobiological mechanism of the control of stress as well as understanding basic physiological differences in the male and female brain in response to stress.
We demonstrated that, in the face of stressful stimuli, subcortical arousal circuitry, i.e. hypothalamic nuclei and left amygdala, along with anterior cingulate gyrus (ACG), showed greater activation in men than women, regardless of women’s cycle timing. That is, in comparing men to EF women, there was significantly greater activation in men of the left amygdala, hypothalamic nuclei (lateral area (LHA) and ventromedial nucleus (VMN)) and bilateral ACG. All other stress response regions activated similarly in men compared with EF women. In contrast, compared to LF/MC women, men exhibited greater activation in ACG, OFC, medial PFC, left amygdala, hippocampus, LHA and VMN, and brainstem regions (e.g., PAG and midbrain). Thus, we demonstrated that stress response circuitry activated in the male brain is more similar to women in the EF menstrual phase compared with women during LF/MC. The strongest evidence for this is seen in Table 3 in which we averaged signal intensity values in voxels across a priori defined anatomic regions in the stress response circuitry. These findings suggest that compared to men, there is an attenuation of stress response circuitry activation in women, particularly evident during midcycle, in subcortical arousal regions that couple with attenuation of regions implicated in the cortical control of arousal. In fact, the largest sex difference effect sizes were in the ventral and medial prefrontal cortices.
Given that EF (when estrogen and progesterone are low) is hormonally more similar to men compared with LF/MC (when estrogen is high), we would argue that it is likely estrogen or ratio of estrogen: progesterone that accounts for activation differences. A recent study of reward circuitry, which shares some brain regions with the stress response (e.g., amygdala, hypothalamus, OFC and ACG), demonstrated region-specific gonadal hormonal regulation, particularly by estrogen unopposed by progesterone (Dreher et al., 2007). Findings in Dreher (Dreher et al., 2007) are not directly comparable, given they tested women at different points in the menstrual cycle than our study and did not directly test for hormonal effects on sex differences. However, the Dreher study does underscore the importance of hormonal regulation in related brain circuitry and the role of estrogen unopposed by progesterone (as would be similar to women during midcycle in our study).
Sex differences in negative affect were directly tested in two recent fMRI studies (Schienle et al., 2005; Caseras et al., 2007), suggesting greater activation of amygdala in men (Schienle et al., 2005) and left ventrolateral PFC in women (Caseras et al., 2007). However, gender groups were only matched on age and handedness, and stimulus and contrast images were not matched on faces and scenes, producing potential confounding influences on sex differences in brain activity. Further, lack of control for menstrual status may have attenuated sex differences in both studies. Finally, subjective valence and arousal ratings of images differed in men and women (Schienle et al., 2005; Caseras et al., 2007) and accounted for findings in (Caseras et al., 2007), in contrast to our study demonstrating sex differences in brain activations given similar subjective ratings of arousal and valence. Our findings importantly suggest sex differences in the brain’s response to stress even in the context of similar clinical state and subjective behavioral ratings, suggesting a role for hormones in regulating homeostasis in the brain in response to stress.
Previous animal studies demonstrated that testosterone (T) regulated norepinephrine (NE) levels during prenatal development and differentially by sex (Stewart and Rajabi, 1994), suggesting an association between HPG and HPA circuitry. That is, inhibiting aromatase (the enzyme that converts T to estrogen) during development produced a rise in NE in anterior frontal, insula and cingulate cortices (Stewart and Rajabi, 1994). This is consistent with our findings in men compared with LF/MC women, in which we demonstrated a large sex difference in the LHA, a key region in the noradrenergic pathway (Luiten et al., 1987), coupled with regions implicated in the cortical arousal circuitry (ACG, OFC, and vmPFC).
LHA has a primary role in arousal and is connected with cortical circuitry implicated in the inhibitory control of arousal and motivational shifts in behavior. Thus it contributes to maintaining homeostasis in the face of the fight/flight arousal response (Swanson, 1987). Some of the densest connections of the LHA are with mPFC and OFC, demonstrated in rats (Luiten et al., 1987) and cats (Room and Groenewegen, 1986), and it is part of the reticular system regulating the autonomic nervous system (ANS) and somatomotor circuitry (Swanson, 1987). In fact in our study, BOLD signal intensity values in activation of LHA correlated significantly and positively with those of vmPFC and OFC (r=.55 and .66 (p≤.005), respectively).
LHA is also a primary region in the noradrenergic pathway with the PVN (Luiten et al., 1987) (hypothalamic nuclei with the greatest concentration of corticotropin releasing hormone), thus providing an anatomical basis of prefrontal control of the HPA axis. In fact, LHA and VMN were qualitatively different in the male and female response circuitry, i.e., LHA and VMN were significantly activated in men, and, compared with women, not dependent on cycle phase. Sex differences in hypothalamic activations may be one reason for sex differences in cortisol levels in response to stress, regulated in part by the PVN in releasing adrenocorticotropin hormone (Schaeffer and Baum, 1984) which interacts with LHA and VMN.
Increases in hypothalamic activations in men may have necessitated compensatory increases in prefrontal circuitry in order to provide cortical control and homeostasis, in part reflected in the positive correlations of LHA with vmPFC and OFC (noted above). In fact, BOLD signal intensity values in activations of vmPFC and OFC were also significantly and positively correlated with the amygdala (r=.61 and .65 (p≤.001), respectively) and hippocampus (r=.72 and .65 (p≤.005), respectively). Previous literature has argued that BOLD signal is mainly associated with excitatory inputs to cells rather than their output spikes (Logothetis, 2008), and BOLD changes can be associated with increased inhibition depending on the region being inhibited and experimental conditions (Buzsaki et al., 2007; Logothetis, 2008). In our study, BOLD activity in subcortical regions was expected to represent both an early increase in excitatory inputs, signaling arousal, followed by an increase in cortical inhibitory inputs to inhibit arousal. This would result in an effect that was consistent with a net increase in BOLD activity (i.e., positive correlations between cortical and subcortical activity). Positive correlations were also found between vmPFC and hippocampus and amygdala during recall of fear (Milad et al., 2007), which was interpreted as consistent with the top-down inhibitory control of subcortical arousal by vmPFC (Milad et al., 2007). In fact, the associations of cortical arousal circuitry to regulation of subcortical arousal regions have been disrupted in disorders such as anxiety and major depression (Shin et al., 2001; Dougherty et al., 2004) and autism (Bachevalier and Loveland, 2006), for which stress response and fear circuitries have been implicated.
The modulatory role of cortical arousal circuitry fits with our previous findings (Goldstein et al., 2005) and others (Protopopescu et al., 2005) in women and with studies demonstrating an inhibitory role of estradiol on arousal circuitry (Best, 1992; Lindheim et al., 1994; Kirschbaum et al., 1996). Although some preclinical studies have demonstrated excitatory roles for estradiol (Woolley and McEwen, 1993; Segal and Murphy, 2001; Lund et al., 2005), this does not negate an inhibitory role for estradiol, which has been associated with estrogen receptor beta (Ostlund et al., 2003; Shansky et al., 2004; Bao et al., 2005b; Lund et al., 2005) while the excitatory role may be associated with estrogen receptor alpha (Lund et al., 2006). These opposing effects of ERα; and ERβ in response to stress have been demonstrated at the level of behavior (Lund et al., 2005) and the level of the anterior hypothalamus (Lund et al., 2006).
Thus, our findings present a potential mechanism for the regulation of the stress response by circulating hormones in women. We also showed that there is not a general effect of hormones on overall blood flow or brain activity, but hormonal effects are region-specific. This makes sense given that these brain regions control the HPA and HPG axes, involve brainstem regions implicated in ANS function and provide frontal cortical influence over autonomic and endocrine function (Price, 1999). They are also dense in sex steroid receptors (McEwen, 1981; Keverne, 1988; Ostlund et al., 2003; Bao et al., 2005a; Lund et al., 2005), noradrenergic receptors (among other monoaminergic receptors), and vasopressin and oxytocin (Keverne, 1988; De Vries and Al-Shamma, 1990; Feldman et al., 1995; Pacak et al., 1995; Tobet and Hanna, 1997; Price, 1999; McEwen, 2000; Swaab, 2004), thus underscoring their adrenal and ovarian functions.
In summary, findings in this study demonstrated that the male and female brain differs in brain activity in response to stressful stimuli, even given the same clinical or behavioral response. Hormonal changes in women during midcycle result in an attenuation of subcortical arousal coupled with attenuation in cortical arousal circuitry that differs from men. This may have important clinical implications for understanding sex differences in clinical disorders associated with stress response circuitry (McEwen, 2000) and high rates of co-morbid endocrine disorders.
From an evolutionary point of view, it is important for the female during midcycle to have a heightened cortical capacity, unencumbered by excessive arousal, to optimally judge whether a potentially threatening stimulus, such as an approaching male, is an opportunity for successful mating or for fight or flight. Thus, females have been endowed with a natural hormonal capacity to regulate the stress response that differs from males. This mechanism may have been maladaptive or unnecessary from an evolutionary point of view for the male, who had primary responsibility for protection of the species thus necessitating a constant fight or flight behavioral response. Historically, these complementary sex-specific social roles have been dynamic and this may reflect sex-specific plasticity in these arousal circuitry neural systems. Thus, although there are sex differences in these neurobiologic systems, they may now support some of the same social functions.
This work was supported by grants from the National Institute of Mental Health to JMG, NIMH-ORWH P50 MH082679, NIMH RO1 MH56956, and the Athinoula Martinos Center for Biomedical Imaging at Massachusetts General Hospital Collaborative Start-up Award. The authors would also like to thank Bernard Harlow, Sc.D. for access to the female population sample, Russell Poldrack, Ph.D. for statistical help on the design of earlier versions of this study, Deborah Walder, Ph.D. and Gabe Daly for help on some earlier stages of data analyses, George Papadimitriou for help in image production, Drs. Anne Klibanski and Stuart Tobet for comments on earlier aspects of the study, and Lisa Cushman-Daly for manuscript preparation.