Hormonal mechanisms of women's risk in the face of traumatic stress

Abstract

Women are underrepresented in biomedical research, limiting understanding of their disproportionate rates of stress-related disorders. Although men experience more trauma, women are twice as likely to develop posttraumatic stress disorder (PTSD). We tested whether ovarian hormones contribute to this disparity via their influence on threat-related brain activity. In a randomized, double-blind crossover study of 110 young women, we examined how exogenous estradiol (E2) modulates threat neurocircuitry, timed precisely to ovarian phase. E2 increased engagement of the ventromedial prefrontal cortex and, in the early luteal phase, reduced central and corticomedial amygdala reactivity to threat in women with little trauma history, but not those with PTSD. Fluctuations in E2 therefore modulate neural responses to threat, and sensitivity to these fluctuations may underlie women's heightened vulnerability to stress-related disorders.

Keywords: anxiety; fMRI; hormones; posttraumatic stress disorder; women’s health.

Fig. 1.

Study design. (A) Timing of patch administration and study procedures. (B) FMRI task, including whole-brain responses across full set of 203 scan visits, at cluster-corrected p_FWE < 0.05. Fearful faces, relative to neutral faces, were associated with greater activation of bilateral amygdala (Left Z = 5.16, xyz = −32, −10, −16, k = 131; Right Z = 4.55, xyz = 24, −10, −16, k = 187), bilateral visual cortex (Left Z = 6.44, xyz = −42, −60, −16, k = 2,651; Right Z = 5.75, xyz = 46, −42, 9, k = 637), and Right ventrolateral prefrontal cortex (vlPFC; Z = 4.76, xyz = 48, 18, 19, k = 269), and lesser activation of posterior cingulate cortex (PCC; Z = 4.55, xyz = 1, −18, 36, k = 374) and dorsal anterior cingulate cortex (dACC; Z = 3.91, xyz = 11, 38, 16, k = 142). The amygdala clusters overlapped both BLA and CeA. Color scale: Red-yellow shows t-score range 2 to 5; blue-green shows range −2 to −5. (C) The E2 patch produced an increase in serum estradiol relative to placebo, with similar levels of increase observed in both the early follicular phase and the early luteal phase [E2 patch M(SD) = 176.06(89.28) pg/mL; PB patch: 88.66(63.58) pg/mL; t = 6.52, P < 0.001]. There was no interaction of patch condition with cycle phase, P = 0.51. Serum estradiol levels were higher in the cohort scanned in the early luteal vs. the early follicular phase of the cycle [t = 4.19, P < 0.001; luteal: M(SD) = 171.59(84.96) pg/mL; follicular: 105.82(82.31) pg/mL].

Fig. 2.

Effect of exogenous E2 on the vmPFC response. The E2 patch produced a preferential response to neutral vs. fearful faces in the vmPFC (E2 main effect t = −3.17, P = 0.002, d = −0.60; E2 × phase P = 0.06). On the y-axis, negative values indicate greater vmPFC response to neutral vs. fearful faces. Disaggregated analyses showed that the E2 effect was specific to the early follicular phase of the cycle (t = −3.48, P < 0.001, d = −0.66). E2 had no effect on the vmPFC response in the early luteal phase (P = 0.83).

Fig. 3.

Group differences in the response to social threat cues, under endogenous hormonal conditions (placebo). (A) Schematic diagram of study visits relative to prototypical hormonal fluctuation over the ovarian cycle. B. The three groups all showed right CeA reactivity to social threat cues (fearful > neutral faces) which were higher in the early follicular phase (follicular vs. luteal phase t = 3.16, P = 0.002, group effect in follicular phase P = 0.30). However, in the early luteal phase, the PTSD+ and NLT groups showed greater right CeA threat reactivity than TC (NLT vs. TC t = 3.71, P = 0.0006, PTSD+ vs. TC: t = 2.48, P = 0.02). (C) In the luteal phase, endogenous estradiol was negatively related to right CeA reactivity in the TC and NTC groups, but positively related in PTSD+ (endogenous E2 × group t = 2.26, P = 0.04; overall model R²Δ = 0.55). TC = Trauma-exposed control, NLT = No- and low-trauma control, PTSD = PTSD and subthreshold.

Fig. 4.

Group-specific effects of E2 administration. (A) Effects of exogenous E2 administration on the right amygdala responses to fearful>neutral faces, separated by ovarian cycle phase. There was an E2 patch × phase × group interaction (t = −2.86, P = 0.005) in right CeA. In the early follicular phase, right CeA responses were not influenced by E2 patch (E2 P = 0.68; E2 × group P = 0.79). In the early luteal phase, exogenous E2 dampened the right CeA response in the NLT group (E2 × NLT group t = −4.01, P < 0.001; E2 effect in NLT: b(SE) = −0.12(0.05), P = 0.03, d = −0.85), and magnified the response in the TC group (E2 effect in TC: b(SE) = 0.16(0.03), P = 0.0002, d = 1.90). In the PTSD+ group, E2 had a different effect than in TC (E2 × PTSD+ group t = −2.27, P = 0.02), such that there was no E2 effect in PTSD+: −0.02(0.05), P = 0.79, d = −0.10). Similar effects were observed for the right ComA (NLT: b(SE) = −0.14(0.06), P = 0.02, d = −0.84, TC: b(SE) = 0.15(0.06), P = 0.02, d = 0.93, PTSD+: b(SE) = 0.05(0.06), P = 0.48, d = 0.28). B. In the luteal phase, lower P4/E2 ratio in serum was linked with higher right CeA reactivity in the TC group, whereas it was linked with lower right CeA reactivity in the PTSD+ group (E2 patch × P4/E2 ratio × PTSD+ group t = 2.11, P = 0.04). There was no association of either P4 or the P4/E2 ratio with right ComA reactivity. (C) Group moderated the effects of E2 administration on state anxiety (E2 patch × PTSD+ group t = 2.43, p = 0.02), but post hoc tests showed no significant effects of E2 in any group (ps > 0.05).