Relationships Among Plasma Dehydroepiandrosterone Sulfate and CortisolLevels, Symptoms of Dissociation, and Objective Performance in Humans Exposedto Acute Stress FREE

Dehydroepiandrosterone (DHEA) is an endogenous hormone secreted by theadrenal cortex in response to adrenocortiocotropin-releasing hormone (ACTH).¹ Dehydroepiandrosterone and its sulfated derivative,DHEA-S, were originally thought to be produced in situ in brain tissue andwere referred to as neurosteroids.² At present,however, it is believed that the only source of brain DHEA-S is from the periphery.³ A growing body of research has provided evidence thatDHEA-S is involved in an organism's response to stress and that it may providebeneficial behavioral and neurotrophic effects.

In mice, DHEA-S release has been shown to be triggered by stress-inducedrelease of ACTH and also exhibits memory-enhancing, antidepressant, anxiolytic,and antiaggression properties.^3- 6 Neuronaland glial survival and differentiation have also been shown to be enhancedby DHEA-S in dissociated cultures of mouse embryo brain and intact rats.^7- 8 Hippocampal neurotoxicity induced bycorticosterone, oxidative stressors, and the glutamate agonist N-methyl-D-aspartate (NMDA) is prevented by DHEA-S.^7- 8 In addition, DHEA-S prevents corticosterone-inducedperformance decrements.^9- 10 Takentogether, these findings suggest that DHEA-S may play a significant role inmodulating vulnerability of the organism to negative consequences of stress.

In humans, levels of DHEA-S peak around ages 20 to 25 years and thendecline to values of 20% to 30% at approximately 70 to 80 years of age.^11- 14 Levelsof DHEA-S are also reduced in a number of medical illnesses such as end-stagerenal disease,¹⁵ liver disease,¹⁶ type2 diabetes mellitus,¹⁷ coronary artery disease,¹⁸ and rheumatoid arthritis.¹⁹ Levelsof DHEA-S have also been noted to be reduced in individuals with chronic fatiguesyndrome,²⁰ depression,^21- 23 anxiety,²⁴ anorexia nervosa,²⁵ andschizophrenia.²⁶ In posttraumatic stress disorder(PTSD), levels of DHEA or DHEA-S have been variable. Levels have been notedto be increased in Israeli soldiers and Kosovo refugees with PTSD, but lowin Kosovo refugees with PTSD who also had comorbid depression. Recently, increasedrelease of DHEA was found in response to ACTH administration in women withPTSD but a negative relationship between the degree of ACTH-induced DHEA releaseand PTSD symptomatology (A.R., C.A.M., S.S., and D.C., unpublished data, 2004).

Consistent with preclinical data regarding the balance of DHEA-S toglucocorticoid levels, the DHEA-S–cortisol ratio in humans has beensignificantly associated with the degree of functional impairment^27- 29 or the response toclinical intervention.³⁰ Taken together, thesepreclinical and clinical studies provide indirect evidence that the DHEA-Slevel and DHEA-S–cortisol ratio may play a role in modulating the impactof stress on a variety of processes in humans.

The present study was designed to evaluate levels of DHEA-S and cortisol,the DHEA-S–cortisol ratio, and the relationship of these hormone indexeswith stress-induced symptoms of dissociation and objective performance inmilitary personnel participating in survival school training. Previous investigationshave demonstrated that survival school represents a valid, reliable modelfor the study of acute, uncontrollable stress in humans.^31- 34 Theresults of these studies demonstrate that the stress experienced by subjectsduring survival school activates biological threat-response systems and elicitspsychological symptoms of dissociation on a scale of magnitude comparableto the degree of arousal and dissociation noted in humans responding to real-world,threat-to-life experiences. Clinical studies have provided evidence that peritraumaticsymptoms of dissociation represent a risk factor for the development of PTSD³⁵ and that such symptoms are positively associatedwith the stress-induced release of glucocorticoids.³³ Onthe basis of this and the preclinical and clinical literature suggesting thatan increased DHEA-S–cortisol ratio may buffer against the negative impactof stress,^7- 8,30 wehypothesized that individuals with a higher DHEA-S–cortisol ratio duringstress exposure would be protected from the negative impact of survival schoolstress as evidenced by fewer symptoms of dissociation and superior militaryperformance.

Twenty-six consecutively recruited active duty military personnel werethe subjects of this study. One female subject was removed from the data set,resulting in a study population of 25. As designated by their military operationalspecialty, 8 subjects were naval aviators, and 17 were nonaviating marines.The mean age was 25 years (SD, 4.4 years). Six subjects (24%) were married,and 19 (76%) were single. The average number of years in the service was 4.6(SD, 4.1).

The methods used in this study have been reported in detail elsewhere.^31- 34 Inbrief, before enrollment in this investigation, each participant completedin-processing into the survival training course. Recruitment of subjects wasconducted by the principal investigator (C.A.M.). All subjects gave written,informed consent. As per survival training course requirements, all subjectsprovided documentation of physical examination and medical and psychiatricclearance before enrollment. All subjects were free of illicit substancesas documented by results of urine toxicologic screening.

Five days before stress exposure, baseline saliva samples were obtainedat 4 PM on the second day of didactic (classroom) activities.Immediately following the collection of salivary samples, baseline plasmasamples were collected by one of the current investigators and Vladimir Coric,MD. Baseline salivary samples were again collected at 7:45 AM.Subjects then completed a modified self-report version of the Clinician-AdministeredDissociative States Scale (CADSS) to rate their symptoms of dissociation duringthe classroom phase.³⁶

The CADSS assesses the frequency and intensity of state symptoms ofdissociation. The items of the instrument are designed to assess how perceptuallyin touch (or out of touch) an individual is vis-à-vis his or her environmentduring specific conditions (nonstressed and stressed). Although some of theitems on the scale ask about one's sense of physical self (eg, "Do you feelas if you are looking at things outside of your body?" and "Do you feel asif you are watching the situation as an observer or spectator?"), other itemsask about cognitive or perceptual distortions (eg, "Do colors seem to be diminishedin intensity?" "Do sounds almost disappear or become much stronger than youwould have expected?" "Do you space out or in some way lose track of whatis going on?" and "Do you see things as if you were in a tunnel, or lookingthrough a wide-angle photographic lens?"). The self-report scale contains19 items, each of which is rated by subjects on a Likert scale of 0 (not atall) to 4 (extremely). A total score of 76 is possible.

At the conclusion of the didactic phase of the training, soldiers participatedin an experiential phase of survival training. During this phase, they wereconfined in a mock prisoner of war camp (POWC). In the POWC, each subjectexperienced various types of psychological stress. Broadly speaking, theseincluded interrogations and problem-solving dilemmas designed to test thetrainees' ability to use the information they learned during the didacticphase. As noted in previous publications,^31,33 theseinterrogations result in robust increases in cortisol and catecholamine levels,heart rate, and subjective distress and significant reductions in testosteronelevel. During the POWC stage, subjects also underwent uniform food and sleepdeprivation. Before interrogation stress, all subjects had been deprived offood for approximately 8 hours and were physically inactive. During the 30-minuteexposure to interrogation stress, subjects remained standing and relativelyimmobile; they did not engage in exercise or physical exertion. Immediatelyafter interrogation, subjects were moved to a second room identical in appearanceto the first, where their blood and saliva samples were collected by the researchteam (C.A.M. and G.H.) between 4:30 and 5 PM. Subjects providedthe saliva samples before the venipuncture. After this, subjects continuedto undergo uniform sleep and food deprivation until their release from thePOWC. All participants were monitored by the survival school medical staffduring the POWC stage and received water on a uniform schedule.

Survival school instructors performed an objective appraisal of observablemilitary-relevant performance of each participant during the POWC phase ofsurvival school. These performance assessment scores are part of the survivalschool program and are not available to the public. The overall rating score,however, is designed to reflect how well a participant in training is ableto demonstrate specific behaviors and problem-solving abilities while experiencingacute stress. The performance ratings are scored on a scale that ranges from0 (no skills demonstrated) to a maximum score of 4 (excellent demonstrationof skills). Because these performance scores were generated independent ofthe research team and the measures collected by the research team, they representa double-blind opportunity to assess the relationship between operationallyrelevant military performance and the psychobiological measures that wereof interest to the research team.³³

Twenty-four hours after the conclusion of POWC stress, recovery plasmaand saliva samples were collected in all subjects. Because of programmaticconstraints within the Navy survival school program, the collection of recoverysamples occurred at 7:45 AM and not 4 PM asin our previous investigations of the Army survival school program. In additionto providing blood and saliva samples, subjects once again completed the CADSS.Subjects were asked to complete the CADSS using the stress they experiencedduring the interrogation phase of the POWC as their reference point.

Plasma samples were spun down in a refrigerated centrifuge, pipettedinto microtubules, and frozen within 40 minutes of venipuncture. Samples werestored at −70°C from the time of initial collection until analyseswere performed. Salivary samples were frozen and shipped with plasma samplesto our laboratory within 24 hours of collection.

Saliva was collected in Salivette tubes (Sarstedt Inc, Newton, NC),centrifuged, and pipetted into two 1.5-mL plastic vials. The samples wereshipped on dry ice to our laboratory and stored at −70°C until assayed.Salivary cortisol levels were analyzed by means of radioimmunoassay (IncStarCorp, Stillwater, Minn). The intra-assay and interassay coefficients of variationwere 4.2% and 6.1%, respectively.

Frozen plasma samples were used and processed in batch by means of DHEA-Scommercial radioimmunoassay kits (Diagnostic Systems Laboratories, Inc, Webster,Tex). Determinations were performed on duplicate 10-mL plasma samples accordingto the manufacturer's recommendations, using supplied rabbit polyclonal anti-DHEA-Santibody–coated tubes. Plasma DHEA-S concentrations were measured witha sensitivity (detectability) of approximately 2 µg/dL and intra-assayand interassay coefficients of variation of 9% and 15%, respectively.

Separate repeated-measures analyses of variance (ANOVAs) using the timefactors baseline-stress and baseline-recovery were performed to detect whetherexposure to stress significantly affected levels of plasma DHEA-S and plasmaand salivary cortisol. To control for diurnal variation, the cortisol samplescollected at the 4 PM point at baseline were compared withthe cortisol samples collected at the stress time point. Similarly, the ANOVAsexamining whether salivary cortisol levels had returned to the reference rangeafter the conclusion of the training used the 7:45 AM salivarycortisol samples, because these were closely associated with the time of salivaassessment on the day of recovery (7:45 AM).

Pearson correlation analyses were used to evaluate the relationshipsamong the assessed hormone levels (DHEA-S and cortisol) and the relationshipsamong the hormones and the independent variables of age and weight.

Spearman rank correlation analyses were used to compare the DHEA-S–salivarycortisol ratios from the baseline, stress, and recovery time points with psychologicalsymptoms of dissociation (CADSS total scores) and military performance scores.Exploratory post hoc analyses were planned for individual CADSS items in theevent that a significant relationship was observed between DHEA-S–cortisolratios and CADSS total score.

Finally, separate stepwise linear regression analyses were used to examinewhether or how well the independent variables of age, rank, time in service,or the DHEA-S–salivary cortisol ratio would explain the variance instress-induced symptoms of dissociation (the CADSS total score) and in objectivelyassessed military performance.

The variables of age, rank, and time in the service did not contributeto variance in the hormone or the psychological data, and thus were removedfrom the analyses.

The mean CADSS score at baseline was 1.0 (SD, 1.6), whereas the poststressCADSS score was 17.4 (SD, 13.0). This increase was statistically significant(F_1,24 = 40.8 [P<.001]). In the publishedliterature this would be considered a moderate level of dissociation.³⁷

The mean, objectively assessed military performance rating for subjectswas 2.3 (SD, 0.7), with a range of 1.0 to 3.8.

Compared with baseline, there was a significant increase in DHEA-S levelin response to stress (baseline, 27.81 µg/dL [SD, 11.06 µg/dL][SI units {calculated with a conversion factor of 0.02714}, 0.755 µmol/L{SD, 0.30 µmol/L}]; stress, 60.12 µg/dL [SD, 26.15 µg/dL][1.63 µmol/L {SD, 0.71 µmol/L}]; F_{1, 21} = 76.1 [P<.001]), and it remained significantly increased comparedwith baseline at the recovery time point (27.81 µg/dL [SD, 11.06 µg/dL][0.755 µmol/L {SD, 0.30 µmol/L}] vs 37.32 µg/dL [SD, 16.98µg/dL] [1.01 µmol/L {SD, 0.46 µmol/L}]; F_1,18 =22.4 [P<.001]). The DHEA-S values at recoverywere significantly reduced compared with the stress time point (F_1,17 = 68.2 [P<.001]). Plasma cortisol levelwas also significantly increased by exposure to survival school stress andremained significantly elevated compared with baseline at the recovery timepoint (baseline, 8.6 µg/dL [237.3 nmol/L] [SD, 3.8 µg/dL {104.8nmol/L}]; stress, 31.1 µg/dL [858.0 nmol/L] [SD, 5.8 µg/dL {160.0nmol/L}]; recovery, 19.9 µg/dL [549.0 nmol/L] [SD, 7.0 µg/dL {193.1nmol/L}]) (F_1,22 = 168.2 [P<.001] andF_1,23 = 55.8 [P<.001], baseline vsstress and baseline vs recovery, respectively). Similarly, compared with baselinelevels, salivary cortisol level was significantly increased by exposure tosurvival school stress (baseline, 0.14 µg/dL [3.9 nmol/L] [SD, 0.04µg/dL {1.1 nmol/L}]; stress, 0.87 µg/dL [24.0 nmol/L] [SD, 0.45µg/dL {12.4 nmol/L}]) (F_1,24 = 62.5 [P<.001]). However, compared with baseline morning values, morningsalivary cortisol levels collected at recovery (recovery value, 0.49 µg/dL[13.5 nmol/L] [SD, 0.20 µg/dL {5.5 nmol/L}]) tended to be significantlyhigher only at the recovery time point (F_1,24 = 3.7 [P = .06]).

No significant correlations were observed between plasma levels of DHEA-Sand cortisol or between plasma levels of DHEA-S and salivary levels of cortisolat any of the assessment time points (baseline and stress or recovery). Similarly,Spearman rank correlation analyses did not reveal a significant correlationbetween plasma cortisol level and poststress CADSS scores (r = 0.16; P = .41). However, there was a significantpositive relationship between salivary cortisol level during stress exposureand the poststress CADSS scores (r = 0.4; P<.05) and a trend for a significant negative relationship betweenstress-induced levels of DHEA-S and poststress CADSS scores (r = −0.4; P = .08). Finally, as shownin Figure 1, there was a significantnegative correlation between the DHEA-S–salivary cortisol ratio duringstress and the poststress CADSS scores (r = −0.63; P = .002).

Analysis of individual CADSS items indicated that there were significantnegative relationships between the DHEA-S–salivary cortisol ratio duringstress and CADSS items 2 (feeling unreal as if in a dream; r = −0.44; [P<.04]), 6 (feeling disconnectedfrom one's body; r = −0.54 [P = .009]), 10 (colors appearing to be diminished in intensity; r = −0.6 [P = .004]), 11(feeling as if one is viewing the world in a tunnel or looking through a wide-anglelens; r = −0.6 [P =.003]), 15 (feelings of being spaced out or of losing track of what is goingon; r = −0.44 [P =.04]), 16 (sounds disappearing or becoming much stronger than expected; r = −0.51 [P = .01]), 17(things having a special sense of clarity; r = −0.49[P = .02]), and 18 (feeling as if one is lookingat world as if in a fog, people appearing far away or unclear; r = −0.72 [P<.001]).

With regard to the relationship between objective military performanceand the indexes of DHEA-S and cortisol levels and dissociation, as shown in Figure 2, there was a significant positivecorrelation between the DHEA-S–salivary cortisol ratio during stressand the military performance scores (r = 0.61 [P = .008]) and a significant negative correlation betweenstress-induced levels of salivary cortisol and military performance (r = −0.51 [P<.01]). Inaddition, and similar to the findings of our previous investigations, therewas a significant negative relationship between stress-induced symptoms ofdissociation (CADSS scores) and military performance (r = −0.51 [P<.01]).

No significant relationships were observed between baseline hormonevalues, recovery hormone values, DHEA-S–cortisol ratios at baselineor at recovery, and the outcome measures of dissociative symptoms and militaryperformance scores in response to stress.

Stepwise linear regression analysis using poststress CADSS dissociationscores as the dependent variable and the DHEA-S–salivary cortisol ratioand plasma levels of DHEA-S and cortisol during stress as the independentvariables showed that the model was significant (F_1,19 = 7 [P = .02]). The adjusted multivariate coefficient of determination(R²) for the model was 0.24 for the predictorDHEA-S–salivary cortisol ratio during stress. The model did not improvewhen the variables plasma levels of DHEA-S and cortisol during stress wereadded. The standardized β coefficient value for DHEA-S–salivarycortisol ratio during stress was −0.53, with a t valueof −2.6 (P = .02).

Stepwise linear regression analysis using military performance scoresas the dependent variable and the DHEA-S–salivary cortisol ratio andplasma levels of DHEA-S and cortisol during stress as the independent variablesshowed that the model was significant (F_1,16 = 5.6 [P = .03]). The adjusted multivariate coefficient of determination (R²) for the model was 0.23 for the predictorDHEA-S–salivary cortisol ratio during stress. The model did not improvewhen the variable plasma level of DHEA-S or cortisol during stress was added.The standardized β coefficient value for DHEA-S–salivary cortisolratio during stress was 0.52, with a t value of 2.4(P = .03).

The principal finding of this study is that individuals with a higherDHEA-S–salivary cortisol ratio during stress experienced fewer symptomsof dissociation and exhibited superior military performance. Because indexesof dissociation and military performance presumably index central processes,the data herein provide support that in healthy humans, the ratio or balancebetween circulating levels of DHEA-S to unbound cortisol may help buffer againstcentrally mediated, negative effects of stress.

One implication of the present findings is that a low DHEA-S–cortisolratio may be associated with vulnerability to stress-induced symptoms of dissociation.In the future it may be fruitful to conduct clinical trials designed to prospectivelyevaluate whether augmentation of DHEA-S levels in humans, before the timeof their exposure to stress, will confer a protective effect, as evidencedby diminished peritraumatic dissociation and improved cognitive performance.

Military performance was significantly and positively associated withDHEA-S–cortisol ratios. Because the rating scales used by the militaryhave not been made available to the public, the degree to which these scalesrelate to more traditional cognitive or psychological measures has not yetbeen fully established. However, in 3 separate investigations, psychologicalsymptoms of dissociation—as measured by the CADSS—have been shownto be associated with poor military performance. Because the military performancescores reflect the capacity of soldiers' cognitive and decision-making abilitiesduring stress, the present finding of a positive relationship between militaryperformance scores and the DHEA-S–cortisol ratio adds weight to theidea that increased levels of DHEA-S are associated with an antistress effecton human cognition. These findings are consistent with those of previous clinicalstudies linking superior cognitive performance with higher concentrationsof DHEA.³⁸

Although no preclinical studies deal specifically with symptoms of dissociation,much is known about the relationship between levels of cortisol, DHEA-S, andglutamate and stress-induced neurotoxicity. However, it is not known at presentwhether stress-induced dissociation in humans is related to stress-inducedneurotoxicity, as reported in the preclinical literature.

A host of preclinical investigations have shown that glucocorticoidscan be neurotoxic and that DHEA-S exerts a potent antiglucocorticoid effectperipherally and centrally.⁷ For example, prolongedexposure to high levels of circulating corticosterone in rats increases theage-related rate at which hippocampal pyramidal neurons are lost.³⁹ Glucocorticoids have also been shown to potentiateneurodegeneration induced by anoxia and glutamate analogues. The neurotoxiceffects of glucocorticoids can be attenuated or blocked by in vivo and invitro administration of DHEA-S.^7- 8

Preclinical investigations suggest several possible mechanisms by whichDHEA-S or DHEA may protect against dissociation or improve cognition in humans.For example, antiglucocorticoid effects of DHEA have been demonstrated inmany tissues, including brain.^40- 45 Withinthe brain, region-specific metabolism of DHEA may ultimately control the natureof DHEA effects on cognition and behavior.⁴⁶ Forinstance, 7α-hydroxylated metabolites of DHEA have been shown to interferewith the nuclear uptake of activated glucocorticoid receptors in the neuronsof the hippocampus.⁴⁶ Dehydroepiandrosteronealso protects against excitatory amino acid– and oxidative stress–induceddamage, restores cortisol-induced decrements in long-term potentiation, regulatesprogrammed cell death, and promotes neurogenesis in the hippocampus.^{7- 8,47- 50} Thusit is possible that the military personnel exhibiting higher DHEA-S–cortisolratios during extreme stress in this study came to survival training withbrain structures and functional capacities that protected them from dissociationduring the interrogation stress.

Preclinical and clinical data suggest that the reduced levels of dissociationin subjects with an increased DHEA-S–salivary cortisol ratio duringstress may be due, in part, to the action of DHEA-S at NMDA receptors and/orat the γ-aminobutyric acid–benzodiazepine receptor complex, perhapsat the level of the hippocampus or frontal cortex. Dehydroepiandrosteronesulfate serves as a negative modulator of the γ-aminobutyric acid–benzodiazepinereceptor complex,^40,51- 52 andthere is evidence of altered benzodiazepine receptor modulation and sensitivityin stress-related disorders characterized by symptoms of dissociation, suchas PTSD.⁵³ Furthermore, DHEA-S also positivelymodulates NMDA receptors, which have been implicated in dissociative phenomenain humans as assessed by the CADSS.³⁷ Thus,it is reasonable to speculate that the reduced levels of dissociation in subjectswith a high DHEA-S–cortisol ratio may be due, in part, to the effectof DHEA on NMDA receptors and/or the effect of DHEA at the γ-aminobutyricacid–benzodiazepine receptor complex, perhaps at the level of the hippocampus.

There are several limitations to this study. First, all subjects insurvival school training experienced food deprivation before stress exposure.Because dieting in clinically obese individuals has been reported to resultin an increase in DHEA-S level,⁵⁴ it is theoreticallypossible that the increase in DHEA-S levels noted in the present subjectsmay have been, in part, influenced by this factor. However, because food deprivationwas kept uniform across subjects, this factor cannot account for the significantrelationships between the DHEA-S–cortisol ratios and the symptoms ofdissociation and military performance.

All subjects also experienced sleep deprivation, raising the possibilitythat alterations in the diurnal variation in DHEA-S and cortisol levels mayhave contributed to the findings of this study. Although possible, this isunlikely to explain the present findings for 2 reasons. First, the findingsof previous studies in military subjects exposed to high stress have shownthat the diurnal variation of hormones is extremely small relative to thevery large alterations of these hormones induced by exposure to acute stress.³¹ Second, sleep deprivation was kept uniform acrossall subjects.

Another limitation is the relatively small number of subjects in thisstudy. It is possible that the large amount of the variance explained in theregression analyses may be due to the relative homogeneity of these militarypersonnel, and the present data may not generalize to the general population.However, the data may be relevant for civilian groups (eg, firefighters, policeofficers, and emergency personnel) who are at increased risk for stress exposureand stress-related illness.

Correspondence: Charles A. Morgan III, MD, MA, c/o 116A, NationalCenter for Post-Traumatic Stress Disorder, Veterans Affairs New England HealthcareSystem, 950 Campbell Ave, West Haven, CT 06516 (charles.a.morgan@yale.edu).

Submitted for publication July 22, 2003; final revision received February13, 2004; accepted February 18, 2004.

This study was supported by US Army Research Institute for EnvironmentalMedicine, Natick, Mass; the Robert Mitchell Center for Repatriated POW Studies,Pensacola, Fla; and the National Center for Post-Traumatic Stress Disorder,West Haven, Conn.

We thank Jeremy Cordova, MS; CPO John Burkhart, USN; and Vladimir Coric,MD, whose assistance greatly facilitated the completion of this project.

The views expressed in this article reflect those of the authors anddo not represent those of the US government or the funding agencies.