Stress hormones and post-traumatic stress disorder in civilian trauma victims: a longitudinal study. Part II: The adrenergic response

Abstract

The aim of the study was to prospectively evaluate the association between the occurrence of post-traumatic stress disorder (PTSD) and the adrenergic response to the traumatic event, and additionally, to explore the link between PTSD and the initial norepinephrine:cortisol ratio. Plasma levels and urinary excretion of norepinephrine (NE) were measured in 155 survivors of traumatic events during their admission to a general hospital emergency room (ER) and at 10 d, 1 month and 5 months later. Symptoms of peri-traumatic dissociation, PTSD and depression were assessed in each follow-up session. The Clinician-Administered PTSD Scale (CAPS) conferred a diagnosis of PTSD at 5 months. Trauma survivors with (n=31) and without (n=124) PTSD had similar levels of plasma NE, urinary NE excretion, and NE:cortisol ratio in the ER. Plasma NE levels were lower in subjects with PTSD at 10 d, 1 month, and 5 months. There was a weak but significant positive correlation between plasma levels of NE in the ER and concurrent heart rate, and a negative correlation between NE in the ER and dissociation symptoms. Peripheral levels of NE, shortly after traumatic events, are poor risk indicators of subsequent PTSD among civilian trauma victims. Simplified biological models may not properly capture the complex aetiology of PTSD.

Introduction

Stress hormones modulate the acquisition and the retention of fear-conditioned responses (see below). A companion paper (Shalev et al., 2007) addresses the contribution of initial levels of the hypothalamic–pituitary–adrenal (HPA) axis hormones to post-traumatic stress disorder (PTSD). This paper describes the early adrenergic response, as well as the contribution of the norepinephrine:cortisol ratio to PTSD.

The occurrence of post-traumatic stress disorder (PTSD) has been equated with the acquisition of a conditioned fear response (Elzinga and Bremner, 2002; Pitman, 1989). Accordingly, the traumatic event is a conditioning stimulus (CS), the immediate response is the unconditioned response (UCR), and the acquired reactions to reminders of the traumatic event are the conditioned responses (CR). Findings of elevated heart rate (HR) shortly after exposure, in survivors who developed PTSD (Bryant et al., 2000; Shalev et al., 1998b; Zatzick et al., 2005), were interpreted as an excessive UCR, possibly reflecting an underlying adrenergic activation (Pitman and Delahanty, 2005).

Previous research has shown that circulating catecholamines facilitate the acquisition of fear-conditioned responses (Gold and Van Buskirk, 1975; Kobayashi et al., 2001; McGaugh, 1988, 1990). This effect requires an intact amygdala in both animals (Roozendaal et al., 1997) and humans (Cahill et al., 1995). It can be blocked by the β-adrenergic blocking agent propranolol, injected into the amygdala in animals (Roozendaal et al., 1997) or given systemically in humans (Cahill et al., 1994). The magnitude of the adrenergic enhancement of fear conditioning is sensitive to concurrent levels of adrenal cortical hormones (Bohus, 1984; Borrell et al., 1984; Roozendaal et al., 1997).

Studies of pharmacological interventions have explored the practical implications of the hypothesized role of catecholamines in the aetiology of PTSD. Cahill et al. (1994) showed that propranolol reduces the retention of stressful recollections in healthy volunteers. Pitman et al. (2002) showed that administering propranolol to survivors of traumatic events, shortly after exposure, reduces the magnitude of physiological responses to reminders 6 months later (see also Vaiva et al., 2003). However, Orr et al. (2006) and van Stegeren et al. (2002) failed to show an effect of β-adrenergic blockade on the acquisition and the retention of conditioned responses to aversive stimuli in humans.

Studies of the initial adrenergic response in PTSD yielded inconclusive results. Delahanty et al. (2000) did not show higher urinary NE levels in adult trauma survivors who developed acute PTSD. Delahanty et al. (2005) described a positive correlation between initial urinary levels of epinephrine, NE and dopamine and acute PTSD symptoms in children. Both studies evaluated the prediction of acute and not chronic PTSD. The combined effect of NE and cortisol has not been evaluated.

This study tests the main hypothesis that PTSD is associated with higher adrenergic response to the traumatic event. In this model, the main outcome variable is PTSD at 5 months (‘Chronic PTSD’ according to DSM-IV criteria) and the main predictive variables are (a) plasma NE in the emergency room (ER) and (b) hourly urinary excretion of NE during ER admission. We also evaluate the prediction of 5-month PTSD symptoms from plasma NE levels in the ER.

The study's secondary hypothesis, namely, that PTSD follows the co-occurrence of elevated plasma NE and relatively low cortisol levels, shortly after the traumatic event, is evaluated by an assessment of the association between PTSD and the NE:cortisol ratio in the ER.

The study also evaluates the relationship between PTSD and resting (unprovoked) levels of plasma and urinary NE, during the months that follow a traumatic event.

Methods

Subjects

The study's sample is same as that of the companion HPA axis paper (Shalev et al., 2007). Its recruitment and follow-up are described there, and briefly summarized here. Subject candidates for this study were recruited from consecutive admissions to a general hospital ER following traumatic events. Subjects were eligible for the study if they were 16–65 years old and had experienced a traumatic event meeting DSM-IV Criterion ‘A’. Subjects were not included if their traumatic event included physical injury requiring medical or surgical intervention, if they had an intravenous line inserted, lost consciousness, sustained head injury with amnesia, or reported an ongoing victimization (e.g. domestic violence). Pregnant women, subjects with lifetime history of endocrine disorder and subjects with chronic PTSD were not included.

A total of 270 survivors were examined in the ER, 1.91±3.1 h after a traumatic event, 182 agreed to participate in the follow-up portion of the study, and 155 (91 males, 64 females) completed the study. Subsequent assessments took place 10.6±3.9, 38.4±3.0, and 159.3±44.7 d after the traumatic event. Traumatic events among participants included road traffic accidents (n=125), terror attacks (n=19) and other incidents (n=11, for details see Shalev et al., 2007).

The study was approved by the Committee on Research Involving Human Subjects (Helsinki Committee) of the Hebrew University – Hadassah Medical School. All subjects signed an informed consent to participate in the ER portion of the study. Those who continued signed an informed consent for the follow-up portion of the study on the first (10 d) assessment session.

Follow-up assessments took place in the morning. Blood and urine samples were collected upon arrival at the hospital, following which subjects completed self-report questionnaires and were interviewed by clinical psychologists.

Instruments

The Clinician-Administered PTSD Scale (CAPS; Blake et al., 1995) conferred a diagnosis of PTSD and a continuous measure of PTSD symptoms. The Structured Clinical Interview for DSM-IV (First et al., 1995) evaluated current and lifetime Axis I disorders. The Impact of Events Scale – Revised (IES-R; Weiss, 1996) evaluated symptoms of intrusion, avoidance and hyperarousal. The Peritraumatic Dissociation Questionnaire (PDEQ; Marmar et al., 1997) assessed dissociation symptoms during the traumatic event. The Trauma History Questionnaire (THQ; Green, 1996) evaluated lifetime exposure to traumatic events. An ad-hoc Trauma Severity Instrument, administered in the ER, evaluated objective and subjective measures of trauma severity.

HR and blood pressure levels were obtained, manually, upon arrival at the ER, and, again, by trained research associates, upon subjects' consent to participate in the study. We report average values.

Biological sampling and analyses

Blood samples were collected following subjects' consent in the ER, and upon subjects' arrival at the hospital on subsequent sessions. Blood samples were spun immediately in a cold centrifuge, and frozen for subsequent analysis. Plasma NE was isolated by alumina extraction and measured by HPLC with electrochemical detection. HPLC–EC analysis was performed with a Bioanaylytical (BAS) system (BAS 200A chromatograph and CMA240 autosampler; Carnegie Medicine, Stockholm, Sweden). Urine was collected (a) during 4 h in the ER (initial void and subsequent collection) and (b) at the beginning of each follow-up assessment session (single void). Hourly urine excretion was determined by multiplying the concentration by the volume and dividing the result by the time, in hours, since the last void.

Square-root-transformed data were plotted in box-whisker diagrams. Data-points that were more than three times the interquartile range from the top of the box were seen as outliers and treated as missing values. For NE values, there was one outlier in the ER urine assessment, two outliers in 10-d plasma assessment, and two outliers at 1 month (one plasma and one urine). Samples that could not be analysed for technical reasons inclued one plasma and 12 urine samples in the ER, one plasma and one urine sample at 10 d, and one plasma and one urine sample at 5 months.

Statistical analysis

Analyses of variance (ANOVA) were used for group comparisons. Pearson's product-moment correlations examined associations between continuous variables. Biological measures were square-root-transformed for the analyses, but not in the tables. The main analyses are carried, separately, for all trauma victims, and for those of road traffic accidents (a subgroup homogeneous for trauma type).

Results

Test of the study's main hypothesis

Thirty-one subjects (20%, n=155) had PTSD at 5 months (PTSD group) and 124 did not have PTSD (Non-PTSD group). PTSD was associated with higher levels of symptoms at all time- points (see Table 2 of Shalev et al., 2007).

The study groups had similar mean age, gender distribution, ER HR, and ER blood pressure levels (Table 1). The study groups had similar levels of plasma NE in the ER [ANOVA F(1, 119)=1.58, p=0.21]. Figure 1 illustrates the distribution of individual plasma NE levels at all time-points. The study groups had similar hourly urinary excretion of NE in the ER [F(1, 77)=0.36]. ER plasma NE did not correlate significantly with PTSD symptoms at 5 months (i.e. r=−0.11, p=0.22 for IES total scores; r=−0.06, p=0.53 for CAPS total scores).

Group comparisons for plasma NE in the ER remained non-significant after controlling for each and all of the following: age, body mass index (BMI), number of previous traumatic events, number of childhood traumatic events, number of cigarettes smoked daily, time from the event to ER arrival and trauma severity. ANOVA using gender and PTSD as grouping factors and ER plasma NE as a dependent variable showed no main effects of the groups and no interaction.

A separate analysis of road traffic accident victims showed no difference in ER plasma NE levels [280.4±120.1 pg/ml in PTSD vs. 313.0±149.2 pg/ml in non-PTSD; F(1, 100)=0.67]. The study groups had similar hourly urinary excretion of NE in the ER (and at all time-points; all F values <1).

Test of the study's secondary hypothesis

The study groups had similar plasma NE:cortisol ratios in the ER [F(1, 116)=0.6, p=0.71]. Adjusting for confounds, as above, did not change the results.

Hormone levels and PTSD during follow-up

The PTSD group had significantly lower levels of plasma NE at 10 d [F(1, 110)=5.57, p=0.02], 1 month [F(1, 101)=10.43, p=0.002] and 5 months [F(1, 145)=3.09, p=0.02]. When road traffic accidents were analysed separately, the PTSD group still had significantly lower NE levels at 10 d [F(1, 87)=4.09, p=0.05], 1 month [F(1, 79)=10.39, p=0.002], and 5 months [F(1, 116)=5.77, p=0.018], but not in the ER [F(1, 100)=0.7, p=0.42].

The study groups had similar plasma NE:cortisol ratios at 10 d [F(1, 109)=0.93, p=0.12], and at 5 months [F(1, 142)=0.17, p=0.7]. At 1 month, however, NE:cortisol ratios were lower in PTSD (n=18) than in non-PTSD subjects [n=81; F(1, 97)=6.11, p=0.01]. The difference is due to higher NE:cortisol ratios in control subjects (Table 1).

Correlation between plasma NE and continuous measures

Symptom measures

Plasma NE levels in the ER correlated negatively with peri-traumatic dissociation symptoms, as measured by PDEQ (r=−0.28, p=0.003).

Biological measures

ER plasma NE levels correlated significantly with subsequent plasma NE levels (for 10 d, 1 month and 5 months, respectively, r=0.241, p=0.011; r=0.221, p<0.035; r=0.208, p<0.026). There was no significant correlation between ER plasma levels and both concurrent and subsequent HPA axis measures.

Physiological measures

ER plasma NE levels correlated with ER HR (r=0.208, p=0.007), but not with systolic or diastolic blood pressure (r=0.07, r=0.06, respectively). PTSD subjects did not have higher HR levels in the ER [F(1, 144)=2.23, p=0.13].

Likelihood of Type II error

The likelihood of a Type II error in this study is 87.3% for ER plasma NE, 60% for ER urine NE and 52.3% for 5-month NE. The number of subjects required to have a statistical power of 90% to detect the current mean differences in ER plasma NE levels, at α=0.05 is 724.

Discussion

The results of this study do not support the hypothesized association between NE levels, and NE excretion, shortly after traumatic events and PTSD. PTSD was not associated with a higher NE:cortisol ratio in the ER. Trauma survivors who developed PTSD had lower plasma NE levels at 10 d, 1 month and 5 months.

Our negative ER findings extend to chronic PTSD a previous negative observation of a relationship between initial NE levels and acute PTSD (Delahanty et al., 2000). Lower plasma NE was found in chronic PTSD in one study (Murburg et al., 1995). Other studies have failed to show a difference between PTSD subjects and controls (Jensen et al., 1997; Marshall et al., 2002; McFall et al., 1992; Mellman et al., 1995), and some studies have shown higher NE levels in PTSD (De Bellis et al., 1999; Glover and Poland, 2002; Lemieux and Coe, 1995; Liberzon et al., 1999; Pitman and Orr, 1990; Yatham et al., 1996; Yehuda et al., 1992, 1998; Young and Breslau, 2004). Together, our results, and those of previous studies, point to the inherent limitations of relying on resting (unprovoked) hormone levels in the study of PTSD.

The absence of a simple association between PTSD and NE in the ER may also be interpreted as reflecting a complex relationship between the initial ‘impact’ of the traumatic event and PTSD. To the extent that PTSD reflects a learning process, that process follows an acquisition phase, withstands subsequent extinction, and is subject to consolidation (Davidson et al., 2004; Yehuda et al., 2006). Accordingly, the initial hormonal reaction to the traumatic event (a putative UCR) initiates a bio-psychological condition that has the potential to become PTSD. This ‘potential’ to become a disorder may be attested by the fact that most trauma survivors who express early PTSD symptoms recover with time (Shalev et al., 1998a). Consequently, the portion of the total causation of PTSD that early hormonal responses may carry could be limited.

The finding of a correlation between plasma NE and HR in the ER is in line with the previously expressed view that an early HR increase reflects an adrenergic response to a traumatic event (Pitman and Delahanty, 2005; Shalev et al., 1998b). However, the weak correlation observed in this study, suggests that adrenergic activation is not the only modulator of HR responses to traumatic events. Other putative modulators of HR variations during emotional events, include vagal afference. For example, Watkins et al. (1998) found an impairment of the baroreceptor reflex under stress, and such impairment could reduce the inhibitory effect of the vagus nerve on HR.

This study did not replicate previous findings of higher initial ER HR in survivors who develop PTSD (Bryant et al., 2000; Shalev et al., 1998b; Zatzick et al., 2005). A possible explanation of this non-replication may reside in the use, in this study, of DSM-IV PTSD Criterion ‘A’ to define a ‘traumatic event’ (i.e. exposure and strong reaction of fear or horror). This might have excluded from this sample, subjects who might have been included in previous, DSM-III-R based, studies, where a ‘strong reaction’ was not used to define a traumatic event.

In line with previous publications (Delahanty et al., 2003; Simeon et al., 2003), this study found a negative correlation between dissociation symptoms and ER NE. Early dissociation has been conceived as protective against experiencing fear and horror during trauma (Griffin et al., 1997). Its occurrence may equally dampen the physiological responses to trauma. Conversely, higher adrenergic response might be associated with over-focused attention or time distortion (Aston-Jones et al., 1999). The PDEQ counts these symptoms as ‘dissociation’.

This study is limited by not having evaluated other catecholamines and their degradation products. Plasma NE is more reflective of the peripheral nerve synapse than the adrenal medulla. Plasma NE is absorbed from arterial blood, and conversely cleared into the veins from peripheral tissue. About half of the NE obtained via venepuncture may originate in the forearm muscles (Hjemdahl, 1993), or reflect the stress of venepuncture. Geracioti et al. (2001) found elevated baseline NE in the CSF of combat veterans with PTSD, but no difference in plasma levels. The half-life of NE in plasma is only several minutes (Hjemdahl, 1993), and therefore, plasma NE may not accurately reflect a summation of the adrenergic response to the traumatic event. Future studies might consider measuring plasma 3-methoxy-4-hydroxyphenylglycol (MHPG), which might better reflect the brain's sympathetic activity (Kopin et al., 1984), or aim to obtain arterial blood, and extend the time lag between line insertion and sampling. Our study is also limited by the significant ‘noise’ that characterizes a naturalistic follow-up study (e.g. sample and trauma heterogeneity).

Given these potential confounds, our finding should not be read as rejecting the putative role of initial adrenergic drive in the aetiology of PTSD, or excluding the possibility of preventive effect of propranolol. Thus, a conservative reading of this study's results is that peripheral levels of NE are poor risk indicators of subsequent PTSD among civilian trauma victims. Along with the joint study of early HPA axis responses and PTSD (Shalev et al., 2007), our observations point to the difficulties of finding robust predictors of PTSD in single biological systems.

Acknowledgements

The authors acknowledge the contribution to data collection of Rivka Tuval-Mashiach, Ph.D., Sara Freedman Ph.D., Neta Bargai, Ph.D., Yair Banet, M.D., and that of Iouri Makotkine to processing and analyses of biological samples. The study was supported by PHS Research grant no. MH 50379 to Dr Shalev.

Statement of Interest

None.

References