| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the University of Pittsburgh (L.W.H., A.L.D., A.B.), Pittsburgh, PA; and Uniformed Services University of the Health Sciences (R.J.U.), Bethesda, MD.
Address reprint requests to: Larry Hawk, PhD, Department of Psychology, Park Hall, State University of New York, Buffalo, NY 14260. Email: LHAWK{at}acsu.buffalo.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: Levels of catecholamines and cortisol in 15-hour urine samples were examined in 55 adults who had been in serious motor vehicle accidents and in 22 age-matched control.
RESULTS: Catecholamines were related to PTSD diagnosis and symptoms, but only among men: PTSD-symptomatic men who had been in an accident exhibited elevated levels of epinephrine and norepinephrine 1 month after the accident and had higher epinephrine levels 5 months later. Intrusive thoughts predicted catecholamine levels at 1 month, and avoidance of trauma-relevant stimuli was associated with higher epinephrine levels 5 months later. These effects were not significant among women. Urinary cortisol was also elevated among PTSD-symptomatic men, but not women, and only immediately (1 month) after the accident. For men and women, greater emotional numbing predicted a lower cortisol level 6 months after the accident.
CONCLUSIONS: These findings were interpreted as limited support for the generalizability of findings in men with chronic, combat-related PTSD and indicate the need for additional research on psychoendocrine assessment of traumatized women and specific dimensions of PTSD symptomatology.
Key Words: posttraumatic stress disorder • cortisol • catecholamines • psychoendocrine • gender differences
Abbreviations: ANCOVA = analysis of covariance; DSM = Diagnosticand Statistical Manual of Mental Disorders; HPA =hypothalamic-pituitary-adrenal; IES = Impact of Event Scale; MVA = motor vehicle accident; PTSD = posttraumatic stressdisorder; SCID = Structured Clinical Interview for DSM-III-R.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
Epinephrine and norepinephrine are catecholamines secreted by the adrenal medulla that play a major role in sympathetic arousal and metabolic, circulatory, and regulatory changes in the body during stress. Cortisol, produced in the adrenal cortex, is the primary human glucocorticoid and has a range of metabolic and antiinflammatory properties that are also important during stress. Together, catecholamines and cortisol regulate or extend most of the systemic changes that occur during stress, and they can be measured in several ways. Assessment in urine provides a measure of the average level of hormone production over many hours (9). Assessment and interpretation of these hormones has improved over the past 50 years, and they have become common and useful indices of stress.
Despite the potential importance of this research and advances in measurement and interpretation of hormone levels, relatively few studies of endocrine activity and PTSD have been reported. Of these, the majority have focused on male veterans with longstanding PTSD subsequent to combat exposure in Vietnam. In a widely cited series of reports by Mason et al. (10–12), hormonal profiles were assessed among small groups of men hospitalized with PTSD, major depressive disorder, bipolar disorder, and paranoid and undifferentiated schizophrenia (diagnoses were based on DSM-III and Research Diagnostic Criteria). In general, men with combat-related PTSD exhibited higher rates of urinary epinephrine and norepinephrine excretion, supporting the hypothesis of chronic sympathetic hyperarousal in PTSD. Some studies have replicated this finding (13), but others have not (14, 15).
In contrast to many studies of stress, Kosten et al. (11) found that PTSD was characterized by low urinary cortisol. Subsequent research has consistently suggested altered activity of the HPA axis, the endocrine pathway that produces cortisol. Low cortisol has been found among male combat veterans with PTSD using urinary (16, 17) and blood sampling (18). These findings have been incorporated in a model of enhanced negative feedback of the HPA axis in chronic PTSD in which cortisol is tonically suppressed but sensitized to acute challenge (19, 20).
These findings are intriguing, but they are based entirely on studies of men with combat-related PTSD. Despite the apparent greater risk of chronic PTSD among women (21), only two studies have examined gender differences in hormonal activity and PTSD. In one study (22), Holocaust survivors with PTSD exhibited lower urinary cortisol levels than did Holocaust survivors without PTSD or control subjects who were not in the Holocaust. The effect was similar for men and women. However, this was an older sample, all of the women were postmenopausal, and observed PTSD had persisted for many years, limiting interpretation of this finding. More recently, Maes et al. (23) reported higher urinary cortisol levels among a group of 3 men and 7 women with PTSD symptoms (ie, threshold and subthreshold diagnoses) 6 to 9 months after an MVA or accidental fire than among a group of 17 control subjects (6 men and 11 women). Although the group effect did not vary with gender, the small sample size placed an obvious limit on the statistical power of the test. However, the results from this predominantly female sample are consistent with those from a study of younger women, in which urinary cortisol was higher among those with PTSD than in similarly traumatized women without PTSD and nonabused control subjects (24). Overall, the equivocal nature of the findings suggests the need for caution in generalizing results from studies of men with combat-related PTSD.
As we have noted, virtually all of these studies were of chronic PTSD, with initiating stressors occurring decades earlier. Little is known about neuroendocrine function within the first months and years after the onset of PTSD, when comorbidity and confounding factors, such as substance abuse, might be minimal. The work of Maes et al. (23) is an exception, and the authors hypothesize that hyperactivity of the HPA axis soon after PTSD onset may give way to enhanced negative feedback over time. Although a history of prior assault was recently associated with both lower plasma cortisol levels within a few days after rape and greater likelihood of subsequent development of PTSD, cortisol levels were not directly associated with the PTSD diagnosis 1 to 6 months later (25). Thus, the time course of the hypothesized downregulation of HPA axis function in PTSD (20) remains uncertain.
The present study addressed these issues by examining urinary catecholamine and cortisol levels among men and women during the first year after their involvement in an MVA. The automobile accident model was chosen because MVAs are extremely common in the United States, with an estimated 6.5 million accidents in 1994, leading to 3.2 million injuries and more than 40,000 fatalities (26). They are also associated with substantial psychiatric morbidity: Up to one-third of those involved in MVAs meet diagnostic criteria for PTSD several months after the accident, and another third may exhibit subthreshold, or subsyndromal, PTSD (27, 28). Because MVAs are a common stressor affecting both men and women, results of these evaluations should be generalizable. The primary hypothesis tested in this research was that urinary levels of catecholamines and cortisol would be higher among MVA victims exhibiting PTSD symptomatology. For cortisol, the direction of the prediction was more tentative, because chronic stress has been associated with high cortisol levels (29) but chronic PTSD has often been associated with low cortisol levels.
Relationships between stress hormones and specific symptom dimensions were also examined, because recent work has emphasized the multidimensional nature of PTSD (30). Most cognitive and/or behavioral models of PTSD emphasize intrusive reexperiencing and hyperarousal soon after trauma, with avoidance developing later and operantly reinforced as a means of coping with the recurrent intrusions (31–33). Failure of active cognitive-behavioral avoidance may lead to emotional numbing (34). Research has consistently indicated a positive relationship between catecholamines and intrusive symptoms, but evidence of links to avoidance symptoms have been equivocal (13, 24, 29). A lower cortisol level has been associated with more avoidance symptoms (22), but this effect was not replicated in a study of women with abuse-related PTSD (24). Given the separation of cognitive-behavioral avoidance and numbing in PTSD theory (35, 36), as well as the untested hypothesis that low cortisol may be specifically related to emotional numbing (37), relationships to hormones were examined separately for cognitive-behavioral avoidance and emotional numbing in this study.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
Participants
Participants were recruited from either a regional trauma center at a large suburban hospital (75% of the sample, including all control subjects; participation rate 
50%) or local police reports involving MVAs (participation rate = 25%). All participants were between 18 and 65 years of age. MVA participants (n = 55, 31 female) were typically driving (78%) alone (65%) in passenger cars (81%) on their way to or from work or home at the time of the accident. Most MVA participants reportedly wore passenger restraints (86%) and were involved in accidents involving one other vehicle (61%), with 22% involving no other vehicle and 17% involving two or more other vehicles. MVA participants had been seen at an emergency room or the trauma center, and most had been hospitalized for up to 10 days. Potential participants were excluded if there was evidence of organic brain syndrome from neurological or mental status examinations or a computed tomographic scan. Control participants (n = 22, 13 women and 9 men) had sustained minor injuries (eg, cuts or sprains), from minor (non–motor vehicle) accidents, that were treated at the emergency room of the hospital housing the trauma center. They did not require hospitalization. As anticipated, injuries were objectively worse among the MVA participants than among the controls (injury severity index (38): mean = 5.1 and 1.7, respectively; F(1,70) = 6.93, p < .01). Only one MVA participant reported a fatality as a result of the accident.
A total of 165 MVA victims and control subjects agreed to participate in the study. To provide a PTSD-free comparison group, four control subjects with current threshold or subthreshold PTSD were excluded. Of the remaining 161 participants, only 77 provided complete urine samples and questionnaire data and completed clinical interviews at both time points. The average age of participants with complete data was 39.1 years (SD = 13.6). "Completers" tended to be married (48%), but 37% reported never being married. More than three-quarters (78%) were white, 14% were African American, 5% were Hispanic, and 3% were of other races. Fifty-one percent had earned a college degree, and 43% percent had a high school degree. Median household income was approximately $35,000 per year. These demographics were generally similar to those of the larger sample (28), but a few differences between completers and noncompleters emerged: Noncompleters were younger (mean age = 33.0 years, SD = 12.2) than completers (F(1,159) = 9.12, p < .01), and a lower percentage of noncompleters were married (30%) (
2(1) = 5.04, p <.03) and had college degrees (35%) (2(1) = 5.34, p < .03). However, noncompleters did not differ from completers with respect to race (2(3) < 1) or income (F(1,142) = 1.43, p = .24). Most importantly, although men represented a greater percentage of the noncompleters (61%) than of the completers (44%) (2(1) = 4.42, p < .04), there was no differential attrition across groups. That is, regardless of whether group was based on MVA and PTSD status at 1 or 6 months or was limited to men, women, or considered the entire sample, the groups of interest (control, MVA/PTSD-free, and MVA/PTSD-symptomatic) were similarly represented among completers and noncompleters (all 2(2) values < 2.64, all p values > .27).
Procedure
All assessments were conducted in the participant’s home. During the initial meeting, the study design, purpose (examination of the thoughts, feelings, and reactions of persons involved in MVAs), and participant responsibilities were described, and informed consent was obtained. For both sessions, participants were given instructions for completion of questionnaires and 15-hour urine collection. Questionnaires and the urine sample were picked up by a researcher the next day. Approximately 1 week later, one of two psychiatric social workers administered the diagnostic interview. Participants were paid $25 per session.
Measures
Diagnosis and symptoms.
The SCID (39) and supplemental PTSD module were used to make DSM-III-R diagnoses of current and lifetime mood and anxiety disorders as well as other Axis I disorders. The PTSD module of the SCID has good interrater reliability and convergent validity (for a review, see Ref. 40).
For this study, the interviewers were trained to 90% interrater reliability on practice interviews involving a variety of diagnoses. Interviews with study participants were audiotaped, and half of these tapes were reviewed by one of the authors (R.J.U.) on an ongoing basis to ensure agreement on PTSD symptoms and diagnosis. Ultimately, diagnoses were determined by consensus between the two interviewers and the supervising psychiatrist (R.J.U.).
To receive the diagnosis of PTSD, a participant had to meet all DSM-III-R diagnostic criteria. The SCID also allows for subthreshold diagnosis, referred to as partial PTSD in the National Vietnam Veterans Readjustment Study (41) and as subsyndromal PTSD in prior studies involving MVAs (27). This diagnosis can be applied when reexperiencing, avoidance, or arousal symptoms (diagnostic criteria B, C, and D) are present but not of sufficient frequency, intensity, or duration to reach threshold levels. For most subthreshold diagnoses in the present study (100% at 1 month, 82% at 6 months), a participant meeting subthreshold criteria met full diagnostic criteria for two of the symptom clusters and was subthreshold on the third. The remaining subthreshold diagnoses were assigned to participants who met criteria for one cluster and were just shy of the criteria on the other two. Because of the relative similarity in symptoms as well as other indicators of distress (27, 41) between subthreshold and threshold PTSD, and to increase the minimum cell size (ie, at 6 months, there were four men with subthreshold PTSD and five with threshold PTSD), diagnoses of subthreshold and threshold PTSD were combined. This allowed a comparison of participants who were PTSD-symptomatic with those who were relatively free of PTSD symptomatology and doubled the minimum cell size. The resulting groups at 1 month consisted of 22 control participants (9 men and 13 women), 21 MVA/PTSD-free participants (12 men and 9 women), and 34 MVA/PTSD-symptomatic participants (12 men and 22 women). At 6 months, there were 27 MVA/PTSD-symptomatic participants (9 men and 18 women) (22 had been PTSD-symptomatic at 1 month and 5 had been PTSD-free at the earlier assessment). Although there were unequal numbers of men and women in each group at each time, these differences were not statistically significant in gender-by-group tests at either time point (for both, 2(2) < 3.9, p values > .14).
The IES (42) is a 15-item self-report measure of reactions to a specific traumatic event. It assesses the frequency of intrusive (eg, "Any reminder brought back feelings about it") and avoidant (eg, "I stayed away from reminders of it") symptoms experienced over the past week (0 = not at all, 1 = rarely, 3 = sometimes, 5 = often). The intrusion and avoidance subscales were developed on the basis of theory (43), and the distinction has been empirically supported by factor analyses and examination of internal consistency reliability (44). The IES has been used to measure intrusive thoughts and avoidance behavior in the majority of prior examinations of relationships between symptom severity and hormones (24, 29).
Emotional numbing has been vaguely defined (36), and there is no widely recognized self-report instrument for measuring it. However, a relatively consistent set of PTSD diagnostic criteria are considered to reflect the construct. Restricted affect, loss of interest, difficulty concentrating, and detachment loaded highly on a numbing factor, but not on other factors, in both prior research (34) and a preliminary principal components analysis of the SCID scores in the present sample.1 Therefore, numbing was operationally defined as the sum of the SCID scores for these four criteria (yielding a scale range of 4–12).
Stress hormones.
A 15-hour urine sample was collected by each participant for assessment of epinephrine, norepinephrine, and cortisol. Although a 15-hour collection period limits comparability to other collection periods and does not allow computation of 24-hour excretion rates (because of variation in excretion over the day), this time frame was selected to reduce adherence and refrigeration problems and activity confounds that can occur with collection during the workday (9). Participants were provided with a vessel containing 1 g of sodium metabisulfite (a nontoxic preservative) and asked to collect all urine excreted between 6 PM and 9 AM (typically during a weeknight, unless this was not feasible for the participant) and to keep the sample refrigerated or on ice during this period. The participant also completed a checklist for foods and activities (eg, cigarette smoking) that could affect hormone levels. The urine sample was picked up by a researcher the next morning and brought to the laboratory, where a 10-ml aliquot was obtained and immediately frozen at -20°C until assayed for free epinephrine, norepinephrine, and cortisol by radioenzymatic and radioimmune assays (45, 46). The concentration of each hormone (in ng/ml) was converted to rate of hourly excretion as follows: (Concentration x Urine Volume)/15.
Data Analysis
In preliminary correlational analyses, several variables were considered as potential covariates. Older participants tended to have higher catecholamine excretion rates at month 6 (r = +.24 and +.21, p values < .04 and .07 for norepinephrine and epinephrine, respectively). In addition, endorsement of greater numbers of foods on the food checklist was associated with higher rates of excretion of epinephrine at month 6 and cortisol at months 1 and 6 (r values = +.23, +.22, and +.21, p values < .03, .05, and .07, respectively). However, cigarette smoking, caffeine consumption, and injury severity scores were not reliably related to hormone levels (all |r| values < .18, all p values > .11). In addition, the percentage of regular smokers did not vary reliably with group membership at either time point (all 2(2) values < 3.5, all p values > .18). Although there was a tendency (p = .06) for a greater percentage of control participants to be taking cardiovascular medication at month 6, there were no other group differences in medication use at either time point (all other 2(2) values < 3.2, p values > .20). More importantly, participants taking medications for cardiovascular problems (n = 14 at 1 month and 11 at 6 months) or psychological distress (n = 5 at both time points) did not differ from those not taking medications on any of the endocrine measures at either time point (all F values < 1.9, p values > .17). Therefore, only age and number of hormone-influencing foods consumed during the collection period were entered as covariates in the analyses.
Lifetime diagnosis of PTSD, comorbid depression, and substance abuse were also considered in preliminary analyses (see also Ref. 47). The most robust association with group membership occurred for lifetime history of PTSD (threshold and subthreshold combined, as for current diagnosis). That is, participants with prior PTSD symptomatology were far more likely to be in the MVA/PTSD-symptomatic group (n = 11) than in either the MVA/PTSD-free or control group (n = 1 each) at month 1 (2(2) = 10.4, p = .01). This tendency was no longer statistically significant at month 6 (n = 7, 5, and 1, respectively; 2(2) = 4.0, p = .14). Concurrent depression tended to be more frequent among PTSD-symptomatic persons at month 1 (n = 6, 0, and 1, respectively; 2(2) = 5.7, p = .06), but the trend was not reliable at month 6 (n = 3, 1, and 0, respectively; 2(2) = 3.3, p = .20). Rates of SCID-diagnosed substance abuse were relatively low. Alcohol abuse was most common, with two alcohol abusers (1 man and 1 woman) at month 1 and three (2 men and 1 woman) at month 6. All of these individuals were PTSD-symptomatic, resulting in a nearly significant alcohol abuse-by-group effect (2(2) = 5.8, p = .06) at month 6. Somewhat unexpectedly, cannabis use was marginally greater among MVA/PTSD-free individuals (1 man and 1 woman) compared with the other groups (no diagnoses) at month 1 (2(2) = 5.5, p = .07). No other substance abuse diagnosis was given to more than one individual at each time point.
To determine the degree to which PTSD symptomatology was associated with different catecholamine and cortisol levels, two sets of analyses were performed. The first set focused on diagnostic group. Because diagnosis varied between assessments for nearly one-third of MVA participants (17 of 55), a repeated-measures analysis was not plausible. Rather, separate ANCOVAs were performed for each hormone (epinephrine, norepinephrine, and cortisol) at each assessment point (1 and 6 months). Age and food checklist scores were covariates, and gender, group, and their interaction were entered as between-subjects predictors (entered hierarchically in this order). Significant interactions were followed-up using simple main effects analyses. Significant effects involving group were followed with comparisons of 1) the control vs. MVA/PTSD-free group (to assess the influence of exposure to trauma) and 2) the MVA/PTSD-free vs. MVA/PTSD-symptomatic group (to assess the impact of PTSD diagnosis). Although comparisons of the control vs. MVA/PTSD-symptomatic group may be most likely to result in significant effects, such tests confound differences due to trauma exposure and PTSD.
The second set of analyses, a series of hierarchical multiple regressions, was performed to examine relations between hormones and specific dimensions of PTSD symptoms (IES intrusions and avoidance and SCID-derived numbing). These analyses were conducted only among participants who had been in an MVA. Separate regressions were conducted for each hormone at each assessment point. As in prior research, the symptom dimensions were moderately to highly intercorrelated at both time points (Table 1). Because of interpretative problems posed by this shared variance, all three symptom scores were included in each regression. These analyses provided tests of whether a given symptom dimension accounted for unique variance in hormone excretion when the other two dimensions were taken into account statistically. For each regression, three blocks of predictors were entered into the equation in the following order: 1) covariates (age and food checklist score) and gender, 2) symptom dimensions, and 3) interactions between gender and each symptom dimension. To reduce problems of multicollinearity, the mean was subtracted from all values for each predictor to produce predictor variables, with means of 0, before analysis (48).
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
|
Month 6.
Figure 1 (right) presents the mean rate of excretion for all gender-by-group conditions when diagnoses and hormones were reassessed 5 months later. As at 1 month, group and gender combined to determine epinephrine levels (gender x group: F(2,69) = 4.65, p < .02). For men, but not women, epinephrine tended to be higher for PTSD-symptomatic participants than for those who experienced an MVA but were PTSD-free (men: F(1,69) = 3.67, p = .06; women: F < 1). However, this effect was no longer evident for cortisol (F(2,69) < 1) or norepinephrine (F(2,69) = 2.16, p = .13). Averaged across genders, there was no evidence of differences in hormone excretion among the groups (all three F values < 1.48, p values > .23).
Both lifetime history of PTSD (threshold and subthreshold combined) and concurrent depression were relatively common in the present sample (see Data Analysis), and prior trauma and depression have both been associated with endocrine changes (23, 25, 47). To determine whether the group differences in hormone levels were influenced by these comorbid psychiatric diagnosis, we performed two sets of ANCOVAs, covarying lifetime PTSD in one set and comorbid depression in the other. In both sets of analyses, all previously significant interaction effects and simple main effects for each gender remained statistically significant.
Symptom Dimensions and Hormones
Month 1.
The results of the regression analyses for month 1 are summarized in Table 2. Paralleling the findings for diagnosis, relationships between symptoms and hormones at 1 month were evident only among men. For epinephrine, the IES intrusions x gender interaction was significant (t(45) = -2.72, p < .01), indicating that intrusions positively predicted excretion rates among men (t(19) = 2.28, p < .04, ß = +0.53) but not women (t(24) < 1). Levels of epinephrine were not related to avoidance or numbing symptoms (all t values < 1). An identical pattern emerged in the analysis of NE, intrusions-by-gender interaction for norepinephrine (t(45) = -2.36, p < .03). That is, more frequent intrusions were associated with higher norepinephrine excretion rates among men (t(19) = 3.07, p < .01, ß = +0.73) but not women (t < 1). Neither avoidance nor numbing symptoms predicted norepinephrine levels (all t values < 1.4, all p values > .16). For cortisol, excretion rate was not related to any of the symptom dimensions or their interactions with gender (all |t| value < 1.34, all p values > .19).
|
|
Importantly, these analyses tested increments in variance, with all three symptom dimensions entered simultaneously. As noted earlier, the advantage of this approach was the ability to test whether specific symptom dimensions could account for unique variance in hormones over and above general symptom severity. However, the analyses were prone to a different interpretational difficulty. Because of the high degree of multicollinearity, particularly between IES subscales at 6 months (Table 1), the findings could have been due to suppression effects (49). If suppression was operating, then the observed findings would disappear in regressions that tested each symptom dimension separately. However, when separate regressions were conducted, the same pattern of effects emerged. The only major difference was one additional finding: at 1 month, avoidance symptoms were associated with higher norepinephrine among men but not women (gender x avoidance: t(49) = -2.09, p < .05). Because this effect was not reliable when intrusions were taken into account (Table 2), it was likely due to variance common to both IES subscales. Thus, as intended, the primary multiple regressions seemed to simplify the findings, and suppression did not account for any of the observed effects.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
Urinary catecholamines were assessed as indices of activity in the sympathetic branch of the autonomic nervous system, the portion of the nervous system that would underlie the presumed hyperarousal in PTSD.2In the present study, catecholamines were associated with PTSD diagnosis and symptom severity among men but not women. Specifically, excretion rates for epinephrine and norepinephrine were higher among men who exhibited significant PTSD symptomatology 1 month after an MVA than among men who also experienced trauma but were asymptomatic. Five months later, PTSD in men continued to be associated with higher epinephrine excretion. These findings are consistent with biopsychological models of sympathetic hyperarousal in PTSD and extend prior findings in longstanding, combat-related PTSD (11) to recent-onset PTSD after a different stressor.
Catecholamines were also related to symptom severity. The integration of biological and psychological processes in PTSD is perhaps best evidenced by the finding, among men, of higher catecholamine excretion among those experiencing more intrusive symptoms at 1 month. This is consistent with cognitive-behavioral theories of PTSD (32, 33) that emphasize the early role of intrusive reexperiencing and associated hyperarousal. Moreover, it is consistent with a model of overconsolidation of traumatic memory that would promote maintenance of PTSD (50, 51). Specifically, intrusive thoughts may lead to heightened levels of stress hormones that facilitate overconsolidation of memory and may make subsequent intrusive thoughts more likely. In addition, cognitive-behavioral formulations of PTSD suggest that avoidance develops later as a means of reducing the anxiety and hyperarousal associated with recurrent intrusions. Interestingly, 6 months after the MVA, epinephrine was associated with higher avoidance symptoms among men, perhaps reflecting the effort necessary for active avoidance.
More generally, the present results converge with findings for plasma catecholamines and psychophysiological reactivity to yield a consistent picture of sympathetic activity among men with PTSD. Psychophysiological (particularly heart rate) reactivity has been repeatedly found to be greater among people with PTSD (for a review, see Ref. 52), including PTSD after MVAs (53, 54). This hyperreactivity has been generally restricted to trauma-relevant information, such as combat films or imagery, and has not been found for other stressors. In contrast, plasma catecholamines have consistently failed to distinguish those with PTSD from control participants (55–57). This seeming inconsistency may well indicate that sympathetic activation in PTSD is chronic but not tonic. That is, the sympathetic nervous system is not simply in overdrive, always pumping out excess catecholamines; rather, the data suggest that the sympathetic nervous system is repeatedly activated by recurrent intrusive thoughts (50). Urinary measures, which assess catecholamines over a much longer time frame than plasma measures, are more likely to "catch" these intermittent, relatively low-frequency events, averaging them with less reactive periods and providing general estimates of differences across time. More acute measures, such as plasma catecholamines, may miss these bursts of responding and underestimate group differences in catecholamines.
It remains to be seen whether psychophysiological reactivity to trauma-relevant stimuli and urinary catecholamines provide convergent or unique information, as well as the degree to which this information aids in the assessment and diagnosis of PTSD. The present data suggest that the greatest impediment to their clinical use may be the complete absence of PTSD-related differences in catecholamines among women. Unfortunately, prior data for women are sparse. In the only other study of catecholamines among women with PTSD (24), epinephrine and norepinephrine were higher among women with chronic PTSD subsequent to childhood sexual abuse than among nonabused control participants, but neither group differed reliably from victims of childhood sexual abuse who did not have PTSD. Thus, the meager data available for women do not indicate strong relationships between PTSD and catecholamines.
There are several possible explanations for these gender effects. Although measurement issues (eg, combining subthreshold and threshold PTSD) may be important, it may also be the case that the relationships between PTSD and sympathetic activity simply do not generalize to women. This hypothesis is consistent with findings for acute reactivity: men exhibit larger epinephrine responses to acute behavioral stress (see the meta-analysis in Ref. 58). Although the biopsychological mechanism for this difference remains unclear (59–61), the findings for acute reactivity may be relevant to gender differences in posttraumatic stress. As discussed above, PTSD-related catecholamine differences seem to be related to recurrent intrusions. To the extent that sympathetic reactivity to these intrusive episodes parallels that to acute stressors, men would be expected to exhibit larger epinephrine responses to intrusive thoughts. If this logic is correct, it could explain why PTSD diagnosis and symptoms would be associated with greater epinephrine excretion in men but not women.
Regardless of the mechanism, gender differences in these hormonal correlates of PTSD have implications for theory and practice. For example, it may not be appropriate to refer to the symptoms under PTSD diagnostic criterion D (difficulty falling or staying asleep, irritability or outbursts of anger, difficulty concentrating, hypervigilance, and exaggerated startle) as "symptoms of increased arousal" (4), because the women with PTSD did not actually exhibit increased arousal, operationally defined in the present study as heightened urinary catecholamines. Moreover, in our analysis of symptom severity, hyperarousal symptoms were not related to catecholamine excretion.2 Interestingly, research has already resulted in the reclassification of physiological reactivity from a symptom of arousal in DSM-III-R to one of intrusive reexperiencing (criterion B) in DSM-IV (52). It may be that some other arousal symptoms merit redistribution. At present, data on this issue remain preliminary, and the only firm conclusions that may be drawn are that future research on catecholamines and PTSD should include women and should test for gender differences.
Paralleling the effects on catecholamines in the present research, urinary cortisol was elevated among PTSD-symptomatic men, but not women, 1 month after MVA. This is consistent with the hypothesis that PTSD is associated with early hyperactivation of the HPA axis and elevated cortisol secretion from the adrenals (20, 23). However, according to this model, the HPA axis adapts over time, establishing a strong negative feedback loop, resulting in tonically low cortisol but maintaining heightened sensitivity to challenge (19, 20). Interestingly, at 6 months, the mean for MVA/PTSD-symptomatic men was (nonsignificantly) below that of MVA/PTSD-free men. This could be taken to indicate the beginning of a transition to the proposed downregulation of adrenal cortical activity (20). Of course, the null hypothesis is a weak platform for arguing that cortisol levels show a biphasic response to extreme stress. However, this interpretation is bolstered by the findings for symptom dimensions. As hypothesized, at 6 months, more numbing predicted a lower cortisol level.
The numbing-cortisol association was the only relationship between hormones and PTSD diagnosis or symptoms that was reliable for men and women, and it is important for several reasons. First, it suggests that the hypothesized downregulation of HPA output (see Ref. 20) may begin relatively soon after trauma (but see also Ref. 23). Second, these data provide support for the hypothesis (37) that this shift in HPA function may be specifically related to the development of emotional numbing, also referred to as a "shutdown" of the affective system (34). That is, whereas catecholamine response may be closely related to the effort required to cope with a stressor, cortisol may be more closely related to such dimensions as distress or controllability (61–64). To the extent that numbing involves a disengagement of distress and relinquishing of control, one might expect lower cortisol. Although the present finding is preliminary and awaits replication with a psychometrically tested measure of numbing, it points out the importance of attending to psychological dimensions in general, in addition to diagnosis (65), and suggests that additional research on numbing is warranted. Third, the current data may extend prior findings in persons with longstanding PTSD (22), suggesting that low cortisol is consistent across gender, type of trauma, and duration of posttraumatic stress. Indeed, it is interesting to speculate that low urinary cortisol levels during the first months after trauma, associated with emotional numbing in the present study, may be a useful prognostic indicator.
Interpretation of the results of the present study is limited by several factors. Fewer than half of those contacted agreed to participate in the study, and it is possible that those who declined were either more or less distressed than our participants. Relatedly, fewer than 50% of initial participants provided complete data for both sessions. Although completers and noncompleters were similar on many dimensions, including gender, we cannot rule out possible selection biases that may have contributed to our results. In addition, even among completers, there were several potential psychosocial mediators of endocrine responding among men and women that were not evaluated. We did not have adequate instruments or sample size to examine all possible factors related to occupational stress and family roles. However, these factors are important in understanding psychoendocrine relationships (61, 66, 67) and should be examined in future work. Relatedly, women in the control group exhibited higher catecholamine levels than did control men, making it difficult to gauge the effect of exposure to trauma among women. It may also be advantageous to consider the time frame for endocrine collection. Although work in men with PTSD suggests that nocturnal sampling, as performed in the present study, may be most sensitive to group differences (Ref. 68; see also Ref. 14; cf, Ref. 69), comparable research with women is warranted (see Ref. 70).
The present study did not directly address whether psychoendocrine relationships mediate PTSD-disease associations. However, the present data lead to several relevant hypotheses. First, people with PTSD may be at risk for physical health problems that are common to chronic stress (via heightened catecholamines) as well problems that are relatively unique (related to altered HPA function). Second, the health risks of men and women with PTSD should have both overlapping (related to cortisol) and unique (related to catecholamines) components. Third, health risks are likely more closely associated with specific dimensions of symptomatology, such as reexperiencing symptoms and numbing, than to overall diagnosis. Tests of these hypotheses will be most powerful to the extent that larger numbers of participants can be assessed repeatedly with instruments designed to capture the dimensional nature of PTSD (71, 72).
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
|
NOTES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
This raises the question of whether endocrine and other measures of hyperarousal converged. In post hoc analyses, the sum of the SCID scores for the hyperarousal criteria was used to predict catecholamines and cortisol. When added to the regressions presented in Results, hyperarousal symptoms failed to account for incremental variance in any hormone at either time point.
Received for publication March 24, 1999.
Revision received September 29, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONNOTESACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. L. F. Shaver, S. K. Johnston, M. J. Lentz, and C. A. Landis Stress Exposure, Psychological Distress, and Physiological Stress Activation in Midlife Women With Insomnia Psychosom Med, September 1, 2002; 64(5): 793 - 802. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. H. Powell, W. R. Lovallo, K. A. Matthews, P. Meyer, A. R. Midgley, A. Baum, A. A. Stone, L. Underwood, J. J. McCann, K. Janikula Herro, et al. Physiologic Markers of Chronic Stress in Premenopausal, Middle-Aged Women Psychosom Med, May 1, 2002; 64(3): 502 - 509. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Lamprecht and M. Sack Posttraumatic Stress Disorder Revisited Psychosom Med, March 1, 2002; 64(2): 222 - 237. [Abstract] [Full Text]
|
|
|
|
|
|
J. W. Mason, S. Wang, R. Yehuda, H. Lubin, D. Johnson, J. D. Bremner, D. Charney, and S. Southwick Marked Lability in Urinary Cortisol Levels in Subgroups of Combat Veterans With Posttraumatic Stress Disorder During an Intensive Exposure Treatment Program Psychosom Med, March 1, 2002; 64(2): 238 - 246. [Abstract] [Full Text]
|
|
|
|
 |
|
 L. M. Diamond Contributions of Psychophysiology to Research on Adult Attachment: Review and Recommendations Personality and Social Psychology Review, November 1, 2001; 5(4): 276 - 295. [Abstract] [PDF]
|
|
|
|
|
|
L. Cohen, C. de Moor, D. Devine, A. Baum, and R. J. Amato Endocrine Levels at the Start of Treatment Are Associated With Subsequent Psychological Adjustment in Cancer Patients With Metastatic Disease Psychosom Med, November 1, 2001; 63(6): 951 - 958. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
