In this preliminary study, women with a history of severe major depression (in remission) demonstrated significant hypoactivations in stress response circuitry that were, in part, associated with gonadal hormone dysregulation, as evidenced during the late follicular/midcycle peak of the menstrual cycle. These findings provide initial evidence that hormonal dysregulation and brain activity deficits in response to stress contribute to trait characteristics in women with MDD. Results demonstrated hypoactivation in MDD women across the circuitry (anterior hypothalamus, amygdala, hippocampus, ACC, OFC, and sgACC), with significant hypoactivations in the OFC, ACC, and hippocampus.
Integration of these results with the broader neuroimaging literature on MDD is limited by the dearth of studies examining response to emotional stimuli in MDD women in remission
. However, hypoactivation in the dorsal ACC in response to negative stimuli was previously reported in MDD individuals in the remitted state (Hooley et al., 2009
). In contrast to Hooley and colleagues, who reported an elevated amygdala response to negative stimuli, our analyses showed hypoactivation in the amygdala in MDD women. However, this inconsistency is likely related to the impact of state
anxiety and mood on brain activity in the amygdala. That is, Hooley et al. (2009)
reported a significant increase in negative mood compared to baseline. In our study, despite an elevated trait
anxiety reported by MDD women, they did not differ from HC women on current state
of mood or anxiety before or after scanning. This implies that hypoactivation in our ROIs cannot be explained by variation in transient anxiety levels and may reflect a trait characteristic of the illness.
MDD women in our study showed decreased levels of estradiol and higher progesterone compared to HC women during the midcycle phase. Abnormal gonadal functioning has been previously demonstrated in women in an acute
episode of MDD (Young et al., 2000
). Similar to Young et al. (2000)
who reported 30% lower estradiol levels in the follicular phase in MDD women, serum estradiol in our MDD group was approximately 25% lower than the HCs. Importantly, our subjects were in remission, and thus this implies that decreased estradiol is likely a trait characteristic in women with MDD. Further, MDD women had elevated serum progesterone, even during midcycle when progesterone levels are relatively low. This is consistent with previous reports demonstrating elevated progesterone in the luteal phase in MDD women in remission (Hardoy et al., 2006
), and previous findings that effective ECT treatment did not alter progesterone levels in MDD women (Baghai et al., 2005
Extending previous neuroendocrine work, we showed that gonadal hormone dysfunction, in part, accounted for variation in brain activity differences in anterior hypothalamus, left amygdala, left hippocampus, and subgenual ACC, in response to stress comparing MDD and HC women. This is not wholly surprising given that in healthy individuals, stress response circuitry regions, such as anterior hypothalamus, amygdala, and hippocampus, are governed by the coordinated action of HPA and HPG-axis hormones. They are regions dense in estrogen α and/or β receptors and progesterone receptors (Donahue et al., 2000
; Guerra-Araiza et al., 2002
; Kato et al., 1994
; Osterlund et al., 2000a
; Osterlund et al., 2000b
). In fact, in previous imaging studies, estradiol and progesterone levels have been significantly associated with brain activity in response to reward (Dreher et al., 2007
), emotional expressions (van Wingen et al., 2008
), and fear extinction and learning (Milad et al., 2009
; Milad et al., 2010
Our results extend these findings to women with MDD and underlying hormonal dysfunction. For example, when unopposed by progesterone, estradiol was negatively associated with activation across the stress response circuitry in healthy control and MDD women, extending our previous findings (Goldstein et al., 2010
; Goldstein et al., 2005
) demonstrating in healthy women that higher estradiol at ovulation was associated with lower activation. Exceptions to this trend were seen in the right amygdala and hippocampus, in which activations were positively associated (at the trend level) with estradiol unopposed by progesterone in the MDD women. In contrast, progesterone, unopposed by estradiol, had stimulatory effects on almost all regions of interest in healthy control women, but inhibitory effects in MDD, with significant effect sizes when comparing the correlations between MDD and HC women. Further work (currently underway) that includes the assessment of the adrenal response to stress in MDD is necessary in order to fully interpret these results, given that high levels of adrenal response, previously found in MDD, can inhibit gonadal hormone levels in women. In addition, progesterone can be released by the adrenal cortex and thus a full understanding of the affected hormonal pathways in MDD must include pituitary, adrenal and gonadal hormone responses to stress.
In the remitted state, medication status serves as a potential confound in interpretation of differences in the brain’s response to stress and the impact of hormones. However, although four of the MDD women were taking antidepressant medication, exclusion of these subjects did not change the findings, demonstrating that hypoactivations in MDD were not driven by medication status.
The generalizability of our findings is limited by small sample size and restriction to female subjects. However, MDD women were carefully made comparable to healthy women on a number of potential confounds. They were in remission so we were investigating trait effects, and the majority was unmedicated. Further, although these women had gonadal hormone abnormalities, they were all cycling. Thus, we in fact sampled against our hypothesis investigating hormonal dysregulation and the brain in MDD, underscoring our results. Further, although generalizability may be limited, this would not negate the internal validity of the results. In addition, even though we had a small sample size, we still demonstrated FWE-corrected results significant in two regions. Further, using analyses to obtain signal intensity changes in our regions of interest based on anatomy, we demonstrated substantial effect size differences in hypoactivations (half to almost one standard deviation) comparing MDD and HC women (see ).
In conclusion, results of our study revealed significant associations between hormonal dysregulation and brain activity deficits in response to stress in women with a history of severe major depression in remission. We demonstrated hypoactivations in MDD women across the stress response circuitry, unrelated to medication status. Further, gonadal hormone abnormalities were evidenced by lower estradiol and higher progesterone levels in MDD which were significantly associated with decreased activations in MDD in the anterior hypothalamus, amygdala, sgACC, and OFC. Findings indicate that hormonal dysregulation and stress response circuitry dysfunction in MDD may be trait characteristics, given that only subjects who were in remission of MDD symptoms were included in the sample. We would argue further that these findings have important implications for understanding the pathophysiology of sex differences in MDD (given sex differences in gonadal hormones), a hypothesis currently under investigation. Moreover, our approach has critical implications for the design of studies of MDD, underscoring the importance of attending to the gender of subjects, the women’s hormonal status, and clinical status of state versus trait characteristics of the illness.