تعامل نورآدرنالین و کورتیزول خاطرات مزاحم منفی در اختلال استرس پس از سانحه را پیش بینی می کند

Abstract Recent evidence suggests that an interaction of noradrenaline (NE) and cortisol (CORT) during encoding leads to greater consolidation of emotional memories. Convergent models of posttraumatic stress disorder (PTSD) suggest the release of CORT and NE lead to greater intrusive memories in PTSD. This study examined the effect of NE and CORT during encoding on recall and intrusive memories in PTSD. Fifty-eight participants (18 participants with PTSD, 20 trauma-exposed controls, and 20 non-trauma exposed controls) provided saliva samples of NE (indexed by salivary alpha amylase; sAA) and CORT at (a) baseline and (b) after viewing negative emotional stimuli. Delayed memory recall and number of intrusive memories of negative, neutral and positive stimuli were recorded two days after this initial testing session. The PTSD group had greater NE levels to negative stimuli and reported greater numbers of intrusive memories of negative stimuli than controls. Regression analyses revealed that the interaction of CORT and NE significantly predicted negative intrusive memories in the PTSD group. The trauma-exposed group reported significantly greater recall of negative images compared to controls, but did not differ significantly from the PTSD group. The PTSD group reported greater levels of suppression of negative images during encoding compared to the other groups. Our results confirm that the interaction of NE and CORT significantly predicts greater negative intrusive memories, but this occurs specifically in the PTSD group. This suggests that a level of heightened arousal is required for the relationship between stress hormones and emotional memory to manifest in PTSD.

Introduction A central model of emotional memory is the memory modulation hypothesis which predicts that memories of threatening events are better recalled due to the interaction of noradrenaline (NE) and cortisol (CORT) in the basolateral nucleus of the amygdala (BLA) during encoding (McGaugh, 2004). Emotionally arousing experiences release adrenaline and glucocorticoids from the adrenal gland and induce the release of noradrenaline in the BLA by activating vagal afferents to the nucleus of the solitary tract. Glucocorticoids freely enter the brain and bind directly to glucocorticoid receptors in brainstem noradrenergic neurons to potentiate NE release in the BLA, as well as postsynaptically in BLA neurons to facilitate the NE signalling cascade (Roozendaal, Barsegyan, & Lee, 2008). Lesion and pharmacological challenge studies in animals reveal that the effects of stress hormones in enhancing memory consolidation depend on the integrity of the amygdala noradrenergic system (Roozendaal et al., 2008 for review). The release of NE and CORT is thought to strengthen the memory trace by activating the amygdala, which strengthens the storage of emotionally arousing information via it’s modulation of different brain regions involved in learning and memory (including the prefrontal cortex, hippocampus, caudate nucleus and nucleus accumbens (Roozendaal et al., 2008). Roozendaal, Quirarte, and McGaugh (2002) hypothesized that activation of glucocorticoid receptors in the BLA may facilitate memory consolidation by facilitating the noradrenergic signal cascade. Indeed, administration of the noradrenergic antagonist atenolol to rats prior to inhibitory avoidance training prevented the memory enhancing effect of the glucocorticoid agonist RU28362 which was administered immediately post-training (Roozendaal et al., 2002). There is also recent evidence for these interactive effects in humans. Participants with higher endogenous CORT had increased amygdala response to emotional images compared to participants with lower CORT levels, however administration of propranolol (a noradrenergic antagonist) blocked this CORT-dependent amygdala activation (Van Stegeren, 2008). Kukolja et al. (2008) examined amygdala responses when participants viewed fearful, happy and neutral faces after taking either a placebo, reboxetine (noradrenergic re-uptake inhibitor which increases noradrenergic concentration), hydrocortisone (elevating CORT levels), or both reboxetine and hydrocortisone. Increased amygdala activity was found in the reboxetine/cortisol group in response to fearful faces. Using the same design, a subsequent study reported that the simultaneous elevation of CORT and NE enhanced hippocampal activity during encoding of negative images (Kukolja, Klingmuller, Maier, Fink, & Hurlemann, 2012). These studies confirm that the interaction of CORT and NE elevates amygdala and hippocampal function, regions known to be involved in emotional processing and episodic memory. Kukolja et al. (2012) proposed this interaction of NE and CORT may lead to hyperconsolidation of episodic memories of trauma. This prediction converges with a model of PTSD proposed by Pitman and Delahanty (2005) which suggests that traumatic events cause intense arousal, triggering the release of NE and CORT. This surge of stress hormones results in increased encoding of environmental stimuli at the time of trauma, and an over-consolidation of the trauma memory. Over-consolidated memories are more easily triggered as they are associated with greater arousal, and have a stronger memory trace – these factors lead to greater conditioning and associations with a wider range of stimuli, which leads to greater priming and triggering from stimuli in the environment. These over-consolidated memories lead to the development of the intense (intrusive) memories and reactivity to trauma reminders that are the core symptoms of PTSD (Pitman & Delahanty, 2005). Supporting this model, in a recent analogue study with healthy participants, intrusive memories of previously presented aversive stimuli during administration of a cold pressor stress were predicted in male participants by the increased interaction of NE and CORT following the cold pressor manipulation (Bryant, McGrath, & Felmingham, 2013). No studies have directly examined the relationship between NE and CORT, and emotional memory consolidation in PTSD. This study provides the first examination to our knowledge of the relationship between NE and CORT during encoding of negative images and subsequent recall and intrusive memories in participants with PTSD. To control for the relative effects of trauma exposure, we compared PTSD participants with trauma-exposed controls as well as non-trauma exposed controls. We hypothesized that compared to controls, the PTSD group would display greater levels of CORT and NE. Given that memory recall reflects the strength of memory consolidation, we predicted that the PTSD group would display greater recall of negative images than the control groups. In addition, models of PTSD propose that greater arousal results in greater intrusive memories, therefore, we predicted that the PTSD group would display greater numbers of negative intrusive memories than the control groups. Finally, in line with existing models, we hypothesized that the interaction of CORT and NE would predict negative recall and intrusions, and this relationship would be particularly evident in the PTSD group.

Results 3.1. Demographic and clinical data Table 1 shows the mean scores and standard deviations for demographic and clinical measures. There were no significant differences in the numbers of males and females in each group, however there were significant differences between groups on age, DASS depression, anxiety and stress scales, PTSD symptoms on the PCL and number of traumatic events experienced on the TEQ (Table 1). Sidak post hoc tests revealed that participants in the PTSD group were significantly older than the control group (p = .04, 95% CI [0.44, 17.40]) however there were no significant differences in age between the PTSD and TE groups (p = .95) or the TE and control groups (p = .10). These analyses revealed the PTSD group had significantly higher scores compared to the TE and control groups for DASS depression, anxiety and stress as well as PCL and TEQ measures. All subsequent analyses were repeated taking age and DASS depression scores as a covariate, but this did not affect any findings and these analyses will not be reported here. There were no significant differences on all clinical scores between the TE and control groups apart from greater TEQ scores in the TE group. There were no significant differences in the years since the index traumatic event in the TE or PTSD groups, and there were no significant differences in the type of trauma experienced in the TE or PTSD groups. The PTSD group had significantly more participants taking anti-anxiety or anti-depressant medication than the TE or control groups. Table 1. Mean scores (standard deviations) of demographic, clinical and salivary measures for participants in the control, TE and PTSD groups. Measures Control (n = 20) TE (n = 20) PTSD (n = 18) Test statistic p Effect size View the MathML sourceηp2 Age 21.80 (5.72) 29.00 (11.46) 30.72 (13.42) F = 3.87 0.03 .12 DASS – depression 6.10 (6.18) 6.50 (7.49) 20.33 (10.83) F = 17.79 <.001 .39 – anxiety 3.55 (3.53) 3.70 (4.78) 13.78 (9.55) F = 16.31 <.001 .37 – stress 10.80 (7.63) 8.80 (5.63) 23.11 (9.41) F = 19.15 <.001 .41 PCL 23.75 (6.72) 25.90 (7.30) 52.50 (11.78) F = 62.17 <.001 .69 TEQ 0.35 (0.59) 3.70 (1.84) 5.17 (2.20) F = 42.13 <.001 .61 Gender 12F, 8M 14F, 6M 12F, 6M χ2 = 0.46 0.80 Months post – 10.2 (10.3) 13.2 (14.1) F = .57 0.45 Trauma type – 0 8 MVA, 12 ass 3 MVA, 15 ass χ2 = 2.5 0.11 Medications 2 7 χ2 = 11.6 <.01 Base CORT Post CORT Base SAA Post SAA Note: n = number of participants in group. Months post = months post index trauma. MVA = motor vehicle accident, ass = assault, Table options 3.2. Salivary data 3.2.1. sAA The mean levels of sAA across the three groups at baseline and after viewing the negative IAPS images are presented in Fig. 1. The 3 (Group) × 2 (Time) repeated measures ANOVA of sAA levels indicated a significant main effect of group [F(2, 55) = 3.27, p = .045, View the MathML sourceηp2=.11], with post hoc tests revealing sAA levels in the PTSD group (M = 102.97, SD = 4.16) were significantly higher than controls (M = 77.89, SD = 3.20, p = .04, g = 6.81); however there was no significant difference in sAA levels between the PTSD and TE (p = .26, g = 5.18) groups, or between the TE and control groups (p = .78, g = 3.04). The difference in sAA level between PTSD and TE groups did not achieve significance, however there was a large effect size between these two groups that was similar in magnitude to the difference between the PTSD and control groups. It is likely that this failure to reach significance was due to the high within-group variability in the sAA measure. There was no significant main effect of condition View the MathML source[p=.35,ηp2=.02], and no significant group × condition interaction View the MathML source[p=.60,ηp2=.02]. Mean levels of sAA (U/mL) in the control, TE and PTSD groups at baseline and ... Fig. 1. Mean levels of sAA (U/mL) in the control, TE and PTSD groups at baseline and after viewing the negative IAPS images (error bars: 95% CI). Figure options 3.2.2. CORT The 3 (Group) × 2 (Time) repeated measures ANOVA of CORT levels indicated no significant main effect of group View the MathML source[p=.23,ηp2=.05], however there was a significant main effect of time [F(1, 55) = 8.30, p = .006, View the MathML sourceηp2=.13], such that CORT levels were significantly higher at baseline (M = 0.34, SD = 0.10) than post-IAPS (M = 0.32, SD = 0.10, p = .006, g = 0.20) for all groups. There was no significant group by condition interaction View the MathML source[p=.91,ηp2=.003]. 3.3. Memory data 3.3.1. Delayed recall data Fig. 2 displays the mean number of negative, neutral and positive images deliberately recalled in each group. The 3 (Group) × 2 (Time) repeated measures ANOVA of recall scores levels revealed a significant main effect of valence [F(2, 110) = 120.9, p < .001, View the MathML sourceηp2=.69], however this was superseded by a significant group by valence interaction [F(4, 110) = 3.7, p = .008, View the MathML sourceηp2=.12]. Mean recall values revealed that both TE and PTSD groups recalled more negative IAPS images than controls, but SIDAK pairwise comparisons showed that while the TE group recalled significantly more negative images than controls (p = .02, g = 0.99), there were no significant differences in the number of negative images recalled between PTSD and controls. Mean number of negative, neutral and positive memories deliberately recalled for ... Fig. 2. Mean number of negative, neutral and positive memories deliberately recalled for the control, TE and PTSD groups (error bars: 95% CI). Figure options 3.3.2. Proportion of suppressors during encoding A 3 × 2 Chi Square test of independence revealed a significant association between type of group and suppression of images, Fisher’s exact test χ2(2) = 20.10, p < .001, Phi = .59. Post hoc 2 × 2 Chi Square tests of independence revealed that the proportion of suppressors to non-suppressors was significantly greater in the PTSD group (72%) compared to the TE [(25%) Fisher’s exact test χ2(1) = 8.47, p = .004, phi = .47] and the control groups [(5%) Fisher’s exact test χ2(1) = 18.40, p < .001, phi = .70], but there was no significant difference between the proportion of individuals who were self-reported suppressors and non-suppressors in the TE and control groups [Fisher’s exact test χ2(1) = 3.14, p = .18, phi = .28]. 3.3.3. Intrusive memory data Fig. 3 displays the mean number of negative, neutral and positive intrusive memories recalled across the groups. The 3 (Group) × 2 (Time) repeated measures ANOVA of intrusion scores indicated a significant effect of group [F(2, 55) = 10.95, p < .001, View the MathML sourceηp2=.29] and a significant effect of valence [F(1.59, 87.34) = 32.57, p < .001, View the MathML sourceηp2=.37]. These were superseded by a significant group by valence interaction [F(3.18, 87.34) = 12.93, p < .001, View the MathML sourceηp2=.32]. Sidak pairwise comparisons revealed the PTSD group recalled significantly more negative intrusive memories than controls (p < .001, g = 1.84) and the TE group (p < .001, g = 1.15), but there were no significant differences in the number of negative intrusive memories recalled between the control and TE groups (p = .49, g = 0.48). Mean number of negative, neutral and positive intrusive memories recalled for ... Fig. 3. Mean number of negative, neutral and positive intrusive memories recalled for the control, TE and PTSD groups (error bars: 95% CI). Figure options 3.4. Relationship between sAA, CORT and memory data To test predictions that the interaction of NE and CORT predicted greater emotional memory, separate multiple regressions were conducted using group membership, gender, sAA, CORT and the interaction term of sAA × CORT as the predictor variables. Table 2 presents the regression findings. There were no significant predictors of negative memory recall. For intrusive memories, group membership was a significant predictor of negative intrusive memories, and the interaction term sAA × CORT was a significant predictor of negative intrusive memories beyond that of group. Beta weights revealed that the increase in sAACORT was associated with an increase in the number of negative intrusive memories. The interaction term sAA × CORT significantly predicted the number of negative intrusive memories in the PTSD group only. Fig. 4 presents a scatterplot of this significant positive relationship between sAA × CORT and number of negative intrusive memories in the PTSD group. Table 2. Multiple regression analyses for negative memory recall and negative intrusive memories. Predictor B Std. err. Beta t Sig. Whole sample: Negative memory recall Group .577 .350 .229 1.647 .106 Gender −.464 .593 −.107 −.783 .438 CORT 3.234 1.374 .231 1.425 .161 sAA −.070 .183 −.055 −.381 .705 sAACORT .027 2.159 .002 .012 .990 Whole sample: Negative intrusive memories Group .612 .098 .676 6.249 .000 Gender .056 .166 .036 .336 .738 CORT 1.780 1.223 .183 1.455 .152 sAA .030 .051 .065 .580 .564 sAACORT 1.865 .604 .411 3.088 .003⁎⁎ PTSD group: Negative intrusive memories Gender .343 .322 .246 1.065 .306 CORT −.059 3.101 −.005 −.019 .991 sAA .103 .085 .273 1.213 .247 sAACORT 5.087 2.035 .678 2.500 .027⁎ TC group: Negative intrusive memories Gender −.403 .409 −2.76 −.984 .343 CORT 2.441 4.160 .335 .587 .567 sAA −.081 .161 −.145 −.502 .624 sAACORT 1.940 2.393 .461 .811 .432 NC group: Negative intrusive memories Gender .314 .165 .411 1.905 .079 CORT .487 1.282 .100 .380 .710 sAA .008 .046 −.037 −.166 .871 sAACORT .879 .508 .501 1.732 .107 ⁎ p < .05. ⁎⁎ p < .01. Table options Scatterplot of the significant positive relationship between sAA*cortisol and ... Fig. 4. Scatterplot of the significant positive relationship between sAA * cortisol and number of negative intrusive memories in the PTSD group. Figure options As seen in Fig. 4, multiple regression analyses revealed that the interaction of CORT and sAA significantly predicted the recall of negative intrusions in the PTSD group (B = 10.32, 95% CI [1.36, 18.89], β = 0.64, t = 2.48, p = .027), with 34% of variance in negative intrusive memories explained by the sAA * CORT interaction (R2 = 0.34). There was also a trend toward a significant positive relationship between sAA * CORT and negative memory recall in PTSD group (B = 14.26, 95% CI [−0.46, 28.98], β = 0.58, t = 2.08, p = .057). There were no significant relationships between sAA, CORT and sAA * CORT and negative recall or intrusions in the TE or control groups.