سطوح پایین استروژن با افزایش پاسخ شرطی در طول انقراض ترس و با خاطرات مزاحم در زندگی روزمره در ارتباط است

Abstract Posttraumatic stress disorder (PTSD) can be conceptualized as a disorder of emotional memory showing strong (conditioned) responses to trauma reminders and intrusive memories among other symptoms. Women are at greater risk of developing PTSD than men. Recent studies have demonstrated an influence of ovarian steroid hormones in both fear conditioning and intrusive memory paradigms. However, although intrusive memories are considered non-extinguished emotional reactions to trauma reminders, none of the previous studies has investigated effects of ovarian hormones on fear conditioning mechanisms and intrusive memories in conjunction. This may have contributed to an overall inconsistent picture of the role of these hormones in emotional learning and memory. To remedy this, we exposed 37 healthy women with a natural menstrual cycle (during early follicular or luteal cycle phase) to a novel conditioned-intrusion paradigm designed to model real-life traumatic experiences. The paradigm included a differential fear conditioning procedure with short violent film clips as unconditioned stimuli. Intrusive memories about the film clips were assessed ambulatorily on subsequent days. Women with lower levels of estradiol displayed elevated differential conditioned skin conductance responding during fear extinction and showed stronger intrusive memories. The inverse relationship between estradiol and intrusive memories was at least partially accounted for by the conditioned responding observed during fear extinction. Progesterone levels were not associated with either fear acquisition/extinction or with intrusive memories. This suggests that lower levels of estradiol might promote stronger symptoms of PTSD through associative processes.

Introduction Posttraumatic stress disorder (PTSD) and other anxiety disorders are highly disabling conditions, with women being at greater risk of PTSD than men (Olff et al., 2007 and Tolin and Foa, 2006). Contemporary cognitive-behavioral theories assume that PTSD constitutes a disorder of emotional memory with strong responding to trauma reminders and intrusive memories belonging to its most prominent symptoms (Ehlers and Clark, 2000, Mineka and Oehlberg, 2008 and VanElzakker et al., 2014). According to the fear conditioning approach, a traumatic event can be seen as an unconditioned stimulus (UCS) that elicits an unconditioned response (UCR) characterized by strong arousal and fear. Co-occurrence with the traumatic event turns stimuli such as sounds, sights or smells into conditioned stimuli (CSs), signalling impending danger and eliciting a conditioned fear response (CR) even when being presented alone later. Despite sometimes being experienced as ‘coming out of the blue’, intrusive memories are frequently triggered by such CSs. Based on this account, intrusive memories can be regarded as conditioned emotional reactions to trauma reminders (Foa et al., 1992 and Mineka and Oehlberg, 2008). Ehlers and Clark (2000) suggested that associative learning is particularly strong in patients with PTSD, thus making the triggering of intrusive memories even more likely. Similarly, Orr et al. (2000) proposed that the propensity to acquire larger as well as more persistent autonomic responses to an aversive CS, i.e. stronger fear acquisition and/or weaker extinction (higher fear conditionability), might render individuals more likely to develop symptoms of PTSD. This notion was supported by findings of heightened conditioned responding during acquisition or extinction in patients with PTSD (Blechert et al., 2007, Lommen et al., 2013, Norrholm et al., 2011, Orr et al., 2000, Peri et al., 2000 and Wessa and Flor, 2007). Owing to the higher risk for PTSD in women, research has recently begun to study fear acquisition and extinction (e.g. Glover et al., 2012 and Zeidan et al., 2011) as well as intrusive memories (e.g. Bryant et al., 2011 and Ferree et al., 2011) specifically in women and in relation to the ovarian steroid hormones estradiol and progesterone. Localization of estradiol and progesterone receptors is not restricted to the hypothalamus, but receptors have also been identified in a multitude of other brain areas, including the hippocampus, amygdala, and prefrontal cortex (Brinton et al., 2008, Cui et al., 2013, McEwen et al., 2012 and Montague et al., 2008). Accordingly, estradiol and progesterone are not only involved in the control of reproductive physiology and behavior, but have been found to more broadly modulate cognitive as well as emotional processes (Farage et al., 2008, Lebron-Milad and Milad, 2012 and van Wingen et al., 2011). The latter is, for example, illustrated by symptoms of anxiety and depression being particularly common during periods with extreme changes in ovarian hormone levels, such as during postpartum and perimenopausal periods (Altshuler et al., 1998, Moses-Kolko et al., 2009 and Schmidt and Rubinow, 2009). However, effects of ovarian hormones on cognitive-emotional processing have also become evident on the basis of more subtle fluctuations of these hormones as they occur during the menstrual cycle (Sacher, Okon-Singer, & Villringer, 2013). Estradiol levels are low during the early follicular cycle phase, peak before ovulation in the late follicular phase and decrease to a moderate level during the luteal phase, whereas progesterone levels are low during the follicular phase and peak at the mid-luteal phase (Sacher et al., 2013). Crucially, with respect to potential relationships between ovarian hormones and cognitive-emotional processing as revealed during fear conditioning, an initial study by Milad et al. (2010) reported that naturally cycling women with low as compared to high levels of estradiol exhibited reduced extinction recall (i.e. more fear in an extinction recall session that was conducted one day after initial fear acquisition and extinction training) (but see also Milad et al., 2006). Zeidan et al. (2011) replicated this finding and additionally found higher activation of brain regions such as the ventromedial prefrontal cortex during extinction recall but also during initial extinction learning in women with higher levels of estradiol. Furthermore, it has been shown that the use of hormonal contraceptives – which are known to decrease ovarian production of estradiol and progesterone – impairs extinction recall and that, on the other hand, administration of estradiol has the potential to enhance women’s ability to recall extinction memory ( Graham & Milad, 2013). Importantly, a role of estradiol in fear extinction is not only supported by studies in healthy participants, but has also been demonstrated in traumatized women ( Glover et al., 2012). Furthermore, Glover et al. (2012) found average PTSD symptoms to be higher in traumatized women with low as compared to high estradiol levels. In summary, studies reported above agree on an association between estradiol and fear extinction processes and have led to the conclusion that low levels of estradiol might constitute a vulnerability factor for symptoms of PTSD (respective findings are reviewed by Lebron-Milad, Graham, & Milad, 2012, and Hiroi & Neumaier, 2011). A separate line of research has investigated associations between estradiol, progesterone, or menstrual cycle phase and intrusive memories. Ferree and Cahill (2009) found that women in the luteal phase reported more spontaneous intrusive memories of film clips than women in the follicular phase. Similarly, Bryant et al. (2011) reported that traumatized women were more likely to experience flashback memories if they were in the luteal phase either at the time of trauma or at the time of assessment. However, neither of the two studies assessed ovarian hormone concentrations. This gap was closed by Ferree et al. (2011) observing a positive correlation between progesterone levels and spontaneous intrusive memories that women reported following the viewing of emotional films. More recently, Cheung, Chervonsky, Felmingham, and Bryant (2013) failed to replicate the latter finding. They found that levels of estradiol rather than progesterone were positively related to intrusive memories of negative images. However, it should be noted that Cheung et al. (2013) neither excluded women on hormonal contraceptives nor assessed women’s menstrual cycle phase, which could potentially account for the different study outcomes. Finally, a recent study by Soni, Curran, and Kamboj (2013) found a negative relationship between intrusive memories and the estradiol-to-progesterone ratio rather than observing correlations with either estradiol or progesterone alone. Thus, compared to the emerging picture in fear conditioning research, the intrusive memory literature remains inconsistent. Importantly, despite the theoretical considerations that fear conditioning and intrusive memories are functionally connected (see above), none of the reviewed studies has investigated the two processes in conjunction, thereby possibly contributing to the discrepant findings between these two lines of research. It is likely that methodological differences between these two paradigms precluded their integrated study: Fear conditioning studies mostly rely on UCSs such as electrical stimulation (e.g. Milad et al., 2006 and Milad et al., 2010) or air blasts (Glover et al., 2012) that only partially depict the typical features (e.g. dynamic, multimodal events) of situations involved in fear learning in real life. Such UCSs are unlikely to generate the kind of complex memories that could later give rise to intrusive memories and are thus inappropriate to investigate fear conditioning and intrusive memories in conjunction. To overcome this problem, we developed a conditioned-intrusion paradigm to model the fear learning process in a more naturalistic manner: Short aversive film clips depicting severe violence serve as UCS and are predicted by neutral sound clips as CSs. Using the paradigm, reliable differential fear conditioning as well as intrusive memories on subsequent days were obtained. Importantly, residual conditioned responding during extinction (indicating both stronger and more persistent conditioned responding, hence higher ‘fear conditionability’ as defined by Orr et al., 2000) was positively correlated with subsequent intrusive memory strength, thereby encouraging further investigation of intrusive memories within a fear conditioning framework ( Wegerer, Blechert, Kerschbaum, & Wilhelm, 2013). The current study extends this work toward investigating the role of estradiol and progesterone in fear conditioning and intrusive memories, studied together in the conditioned-intrusion paradigm. Lower levels of estradiol were expected to be accompanied by stronger conditioned responding during fear extinction (Glover et al., 2012 and Zeidan et al., 2011), whereas acquisition response levels should be unaffected. Furthermore, inasmuch as intrusive memories represent non-extinguished conditioned responding to trauma reminders, low estradiol should also be associated with higher intrusive memory strength and this link should be mediated by the degree of conditioned responding observed during fear extinction. For progesterone predictions were less clear since the reviewed evidence failed to document an effect on fear conditioning. Similarly, predictions for intrusive memories are complicated by inconsistent findings (see above; Cheung et al., 2013, Ferree et al., 2011 and Soni et al., 2013) and by the fact that some studies did not assess ovarian hormone levels but analyzed data only with respect to self-reported menstrual cycle phase (Bryant et al., 2011 and Ferree and Cahill, 2009). Yet, a related line of research, focusing on deliberate (rather than spontaneous, intrusive) emotional memory recall, found progesterone levels to predict recall and recognition memory for negative images (Ertman et al., 2011 and Felmingham et al., 2012). Together with the initial evidence for an association between progesterone and intrusive memories (see above; Ferree et al., 2011) these latter studies point to a more general effect of progesterone on enhancing aversive (declarative) memory. Based on this background, we made the more tentative prediction that progesterone levels are positively associated with intrusive memories (in the absence of an association with – or mediation by – conditioned responding during extinction).

. Results 3.1. Demographic and psychometric characteristics Participating women’s mean age was 23.9 years (SD = 3.4) and their trait anxiety (M = 38.9, SD = 10.3; Laux et al., 1981) as well as depressive symptoms (M = 13.2, SD = 8.4; Hautzinger & Bailer, 1993) were in the normal range. Participants reported regular menstrual cycles with a mean cycle length of 28.9 days (SD = 2.9). On the day of testing, 16 participants (43.2%) were in their early follicular phase and the other 21 participants (56.8%) were in their luteal phase. All further analyses in the main text are based on continuous levels of estradiol (M = 6.23 pg/mL, SD = 3.90) and progesterone (M = 191.17 pg/mL, SD = 89.01) whereas Appendix A contains an exploratory analysis comparing women in the two cycle phases. Estradiol and progesterone levels were unrelated to age (largest ρ(37) = .06, smallest p = .719), trait anxiety (largest r(35) = −.17, smallest p = .323) or depressive symptoms (largest r(35) = −.11, smallest p = .529). 3.2. Physiological baseline activation and reactions to unconditioned stimuli (film clips) Neither estradiol nor progesterone levels were significantly related to preconditioning baseline SCL (M = 5.28 μS, SD = 2.06; largest r(33) = .17, smallest p = .350; note that two participants had to be excluded from this analysis due to technical problems during baseline measurement) or electrodermal UCR (M = 0.90 μS, SD = 0.59; largest r(35) = −.09, smallest p = .619). 7 Mean valence of film clips was 87.2 (SD = 11.6; 0 = very pleasant, 100 = very unpleasant) indicating that films were experienced as highly unpleasant by participants. Likewise, participants’ ratings of film clips were not significantly correlated with levels of estradiol or progesterone (largest ρ(37) = .14, smallest p = .427). Thus, neither estradiol nor progesterone levels covaried with participants’ physiological baseline activation or unconditioned film responding. 3.3. Fear conditioning 3.3.1. Manipulation check for the fear conditioning procedure Participants demonstrated reliable differential electrodermal fear acquisition: The repeated-measures ANOVA for the whole conditioning task displayed a significant main effect of CS-type (F(1, 36) = 7.84, p = .008, η2 = .18) and Conditioning phase (F(2, 72) = 13.94, p < .001, η2 = .28, ε = .634), as well as a significant CS-type × Conditioning phase interaction (F(2, 72) = 4.25, p = .025, η2 = .11, ε = .828). Post-hoc tests revealed that participants’ SCRs did not significantly differ between CS+ and CS− during habituation (t(36) = −1.39, p = .174). However, a significant difference between SCRs to CS+ and CS− could be observed during acquisition (t(36) = 2.20, p = .035, d = 0.20) as well as during extinction (t(36) = 2.81, p = .008, d = 0.23). Thus, consistent with our previous report on the larger sample ( Wegerer, Blechert, Kerschbaum, et al., 2013), these data indicate that the film-based conditioning paradigm can induce fear learning in healthy (naturally cycling) women. Fig. 1 displays means and standard errors for SCRs over the whole conditioning procedure and all participants. Means (and standard errors) of skin conductance responses (SCRs) to CS+ and CS− ... Fig. 1. Means (and standard errors) of skin conductance responses (SCRs) to CS+ and CS− across habituation (Hab), acquisition (Acq), and extinction (Ext) phases of the fear conditioning task. Asterisks indicate significant difference in skin conductance responses (SCR) between CS+ and CS− trials (p < .05). Figure options 3.3.2. Effects of estradiol and progesterone on differential fear conditioning Assessing estradiol and progesterone effects on differential SCRs, ANCOVA for habituation revealed neither a significant main effect of Time (F(1, 34) = 0.33, p = .569), nor a main or interaction effect involving estradiol (all Fs(1, 34) < .0.38, ps > .544) or progesterone (all Fs(1, 34) < 0.02, ps > .896). Similarly, during acquisition, no significant main effect of Time (F(1, 34) = 0.17, p = .683) and no main or interaction effects involving estradiol (all Fs(1, 34) < 0.37, ps > .545) or progesterone (all Fs < 0.15, ps > .706) were observed. During extinction, however, the main effect for estradiol was significant (F(1, 34) = 5.43, p = .026, η2 = .14), indicating a linear relationship between estradiol levels and conditioned responding during extinction in the sense that participants with lower levels of estradiol displayed larger differential responses during extinction (see Fig. 2). There was no significant main effect of Time (F(1, 34) = 0.93, p = .343) or other main or interaction effects involving estradiol (F(1, 34) = 0.87, p = .359) or progesterone (all Fs(1, 34) < 0.14, ps > .713) during extinction. 8 Including UCR to film clips, anxiety, and depressive symptoms as further control variables into ANCOVAs did not change the significant main effect of estradiol for conditioned responding during extinction (F(1, 31) = 4.81, p = .036, η2 = .13). Furthermore, other main and interaction effects involving estradiol or progesterone during habituation, acquisition or extinction remained non-significant (all Fs(1, 31) < 0.94, ps > .339). Relationship between estradiol and conditioned responding during fear ... Fig. 2. Relationship between estradiol and conditioned responding during fear extinction. See ANCOVA main effect of estradiol reported in Section 3.3.2. Differential SCR was calculated subtracting ln (1 + SCR) for CS− trials from ln (1 + SCR) for CS+ trials. ρ: Spearman correlation coefficient (see Footnote 5). Figure options 3.4. Ambulatory assessment of intrusive memories Table 1 displays means and standard deviations for intrusive memories as measured by the three items of the IMQ on days 0–2 after the laboratory session. Importantly, we observed a significant negative correlation between levels of estradiol and intrusive memory strength reported on days 0–2 (ρ(37) = −.36, p = .030), indicating that women with lower levels of estradiol reported overall larger intrusive memory strength relating to the film clips used in the conditioning procedure (see Fig. 3). However, intrusive memories did not significantly correlate with progesterone (ρ(37) = −.08, p = .656). Partialling out UCR to film clips, anxiety, and depressive symptoms did not change these results much (estradiol: ρ(32) = −.38, p = .026, progesterone: ρ(32) = −.09, p = .605). Relationship between estradiol and intrusive memory strength. Values of ... Fig. 3. Relationship between estradiol and intrusive memory strength. Values of intrusive memory strength represent sum scores of t-scores of Intrusive Memory Questionnaire (IMQ) items (number, duration, and distress of intrusive memories) averaged across days 0–2. ρ: Spearman correlation coefficient. Figure options 3.5. Mediation analyses: does conditioned responding during fear extinction mediate the association between estradiol and intrusive memories? Estradiol was associated with differential responding during extinction (see Section 3.3.2. above) and with intrusive memory strength (see Section 3.4. above) but was the latter correlation actually mediated by individual differences in conditioned responding during extinction? In fact, further correlational analyses confirmed that conditioned responding during extinction was significantly correlated with intrusive memory strength (ρ(37) = .41, p = .012) and remained so when partialling out estradiol (ρ(34) = .33, p = .050) (ρ represents coefficient for partial Spearman correlation; Kendall & Gibbons, 1990). The correlation between estradiol and intrusive memory, on the other hand, fell below significance when partialling out extinction responding (ρ(34) = −.26, p = .130). Thus, results suggest that conditioned responding during fear extinction was a relevant mediating mechanism in the inverse correlation between estradiol and intrusive memory strength. 9