Criteria for Rationally Evaluating Animal Models of Posttraumatic Stress Disorder

Rachel Yehuda & Seymour M. Antelman (1993)

Originally appeared in Biological Psychiatry, 1993, 33, 479-486. Note that this online version may have minor differences from the published version.
From the Department of Psychiatry (RY), Mount Sinai School of Medicine, and Posttraumatic Stress Program, Bronx Veterans Administration Medical Center, New York, New York; and Department of Psychiatry (SMA), Western Psychiatric Institute and Clinic, and Pittsburgh Cancer Institute, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
Address reprint requests to Rachel Yehuda, PhD., Psychiatry Department, Bronx VAMC, 130 W. Kingsbridge Road, Bronx, 10468.
Support for this work was provided by grants MH49536-OI (RY), MH 49555-01 (RY), MH24114 (SA), and NIAAA PSOAA08746 (SA).

Abstract

Animal models of stress have the potential to provide information about the course and etiology of posttraumatic stress disorder (PTSD). To date, however, there have been no systematic approaches for evaluating the relevance of animal models of stress to PTSD. It has been established in the animal literature that different types of stress paradigms lead to different biobehavioral consequences and that many different factors contribute to differential responsivity to stress. It becomes important therefore to differentiate between factors that are essential to the induction of PTSD-like symptoms and those that influence their manifestations. In the present commentary, we present five criteria that must be fulfilled by animal models of stress for them to be useful to understanding the induction of PTSD. We then evaluate two potential animal models of stress -- inescapable shock-learned helplessness and time-dependent sensitization -- to illustrate how to more successfully pair animal models of stress with the specific clinical syndrome of PTSD.

Key Words: Stress, animal models, posttraumatic stress disorder, inescapable shock, learned helplessness, time-dependent sensitization

Introduction

Posttraumatic stress disorder (PTSD) as defined by DSM- III-R [American Psychiatric Association (APA) 1987] is an anxiety disorder involving both somatic and psychological symptoms that occur in response to severe trauma, including combat, natural and manmade disasters, hostage situations, rape, and assault. Recent statistics suggest that more than one-third of combat Vietnam veterans develop PTSD at some time following exposure to war-zone stressors (Kulka et al, 1990). Equally alarming is that the lifetime prevalence of PTSD among civilians is 1%, which is comparable to that of schizophrenia (Helzer et al 1987; Malloy et al 1983). Given the debilitating and often chronic nature of the symptoms, PTSD clearly constitutes a mental health problem of considerable magnitude.

The core symptoms of PTSD are intrusive, involuntary reexperiencing of the trauma (e.g., through nightmares, flashbacks, unwanted memories); avoidance symptoms (e.g., attempts to actively avoid reminders of the trauma; emotional unresponsiveness and numbing); and hyperarousal (e.g., hypervigilance, insomnia, startle). These symptoms can develop immediately following the trauma or may be delayed for several months or years; symptoms may be acute or chronic in nature (APA 1987; Kardiner, 1941).

Because the formal diagnosis of PTSD is relatively new, the literature examining biological aspects of PTSD is still small. It is impressive, nonetheless, in demonstrating consistent disturbances in neuroendocrine and neurotransmitter systems involved in mediating stress responsivity (see, for review, Giller 1991; Friedman 1991; Yehuda 1992; Yehuda et al 1991a). The biological findings have supported the notion of a distinct pathophysiology of PTSD and have prompted investigators to search for potential animal models of PTSD in the hope of increasing our understanding of the biological underpinnings of this disorder.

Rationale for Animal Modeling in PTSD

Animal models of human processes and disorders are desirable for several reasons. First, they offer the possibility of simulating a human condition under controlled circumstances, with large numbers of subjects, in a simpler more readily understandable system. Second, in contrast to human disorders, which can be studied only after they become clinically manifest, animal models are observable as they evolve, permitting the study of symptoms as they develop. Third, they allow the testing of pharmacological and other prospective "treatments" that might be difficult in humans.

Animal models of psychiatric disorders have been useful in elucidating connections between behavioral symptoms and biological abnormalities and in suggesting possible treatment strategies for psychiatric disease. Most animal models of psychiatric disorders have been derived by trying to simulate aspects of the clinical disturbance (e.g., administering reserpine to rats to mimic the decreased catecholamines in depression) or by noting similarities in neurochemical or behavioral manifestations between a clinical phenomenon and a manipulation in animals (e.g., catecholamine depletion in both human depression and "learned helplessness" in rats); however, because these models have not been based on a knowledge of the primary etiologic agents(s) producing the disorder (i.e., catecholamine depletion may not be a primary cause of depression), they have usually been much more simplistic than the disease state they have purported to reflect, and their utility has therefore been limited.

In the case of PTSD, there is a greater potential to accurately model the disorder because the major precipitating factors are known (i.e., PTSD occurs in response to severe and unusual stressful or traumatic situations). Even so, attempts to compare the clinical findings in PTSD and results of animal studies of stress require consideration of several points. First, PTSD, which is a fairly circumscribed biobehavioral syndrome, can be induced by a wide range of stressors, whereas animal studies of stress have shown marked biobehavioral differences depending on the type of stressor studied (e.g., controllable, escapable, acute, chronic, predictable, physiological, psychological). Second, differential responsivity to stress can be influenced by factors other than the actual stressor, such as the state of the organism during stress, past stress history of the organism, and even genetic makeup. Third, the stressor itself may be only one of many important variables contributing to the development of PTSD. That is, whereas exposure to a stressor is certainly a necessary condition for PTSD, this factor alone may not be sufficient. These considerations make the task of animal modeling more complex than simply establishing prima facie comparisons between PTSD and the sequelae of stress.

In considering relevant animal models of PTSD, it is critical to differentiate between factors that can influence the manifestations or course of PTSD and those that are essential for its induction. Whereas many factors (such as those mentioned above) appear to contribute to the form and or intensity with which the stress response eventually manifests itself, they are almost certainly not all central to its induction. By establishing criteria that can help in this differentiation and then systematically eliminating animal models that do not fulfill these criteria, it is possible to weed out studies that are uninformative with regard to the induction of core PTSD symptoms (i.e., intrusive reexperiencing of trauma, avoidance, and hyperarousal) even though they may be relevant to other aspects of stress responsivity.

To date, the model of PTSD that has received the greatest acceptance has been that of "inescapable (uncontrollable) shock," or "learned helplessness" (van der Kolk et al 1985; van der Kolk 1987; Krystal et al 1989, 1991; Southwick et al 1992). The appeal of this experimental paradigm (discussed below) is that it appears to have "face validity," because traumas known to give rise to PTSD, such as rape, natural disasters, or combat, occur without control of the individual; however, as Robbins and Sahakian (1980) pointed out in discussing animal models of mania, "a superficial face validity is only the first step in establishing a model; a stringent behavioral analysis is necessary to determine in which ways the behavior is similar and in which it differs" from the human disorder. When this caution is heeded, the "inescapable shock-learned helplessness" (IS-LH) model loses much of its luster, as will be described. The dangers inherent in the uncritical acceptance of a model based principally on a superficial face validity graphically point up the need for establishing indisputable criteria for evaluating the appropriateness of proposed animal models for PTSD. In this commentary, we present criteria that must be fulfilled by animal models of stress for them to be useful to understanding the induction of PTSD. The criteria were derived by paring down PTSD phenomenology to its most basic components and identifying relevant counterparts for these clinical characteristics based on animal studies.

Criteria for Evaluating the Relevance of Individual Stress Paradigms to the Phenomenology of PTSD

1. Even Very Brief Stressors Should be Capable of Inducing Biological & Behavioral Sequelae Of PTSD.

According to DSM-III-R, the clinical syndrome of PTSD can develop in response to different types of stressors, both acute and chronic in nature (APA 1987). In many cases, such as natural disasters, violent assaults, or traffic accidents, the actual traumatic event may be very brief in nature, often lasting only seconds to minutes; however, the ensuing clinical disorder may be severe or prolonged (vanDyke et al 1987). The major defining characteristic of the stressor, then, does not appear to be its actual duration, but, rather, the extent to which it is traumatic enough to elicit the intrusive reexperiencing, avoidance/numbing, and hyperarousal symptoms of PTSD.

For this reason, a useful animal model of PTSD must be one in which the long-term behavioral or neurochemical correlates are capable of induction by a stressor imposed on the organism for no more than seconds to a very few minutes. Models that depend exclusively on long-lasting stressors or training or learning on the part of the animal are therefore not relevant to the induction of PTSD.

2. The Stressor Should be Capable of Producing The PTSD-like Sequelae in a Dose-Dependent Manner.

Because PTSD can occur in response to a wide range of stressors, choosing an animal model of stress based solely on the characteristics of a particular stressor (i.e., acute versus chronic, controllable versus uncontrollable) and/or the "unique" behavioral and neurochemical aftermath conferred by it is of limited value in understanding what causes PTSD. Theoretically, any type of stressor of sufficient magnitude should be able to induce the biobehavioral sequelae of PTSD. Therefore, although duration of the stressor (i.e., acute versus chronic) does not seem to be directly relevant to predicting the clinical and, possibly, the biological consequences of PTSD, as established above, the dosage or intensity of stressor is relevant to this issue.

In humans, PTSD is believed to be produced by a threshold "dose" of stress that is "outside the range of normal experience" or at least traumatic enough to produce the requisite symptoms of PTSD. Additionally, there is evidence for a relationship between stressor intensity and severity of PTSD symptoms once the diagnosis has been established (Yehuda et al 1992). In laboratory animals, one may hypothesize that any experimentally induced stressor might be considered "outside of the range of [the animal's] usual experience." Such a hypothesis would widen the net of potentially relevant animal models of PTSD; however, given the relationship between magnitude of trauma and severity of PTSD, a useful animal model for PTSD would be one in which behavioral and biological sequelae of the stressor(s) have been studied in a dose-dependent manner and in which changes in relevant biological systems show differential responsivity to different levels of similar stressors.

3. The Stressor Should Produce Biological Alterations that Persist over Time or Become More Pronounced with the Passage of Time.

According to DSM-III-R, even though core symptoms can appear immediately following the trauma, the diagnosis of PTSD can only be made if core symptoms have endured for at least I month following the trauma. Thus, biological changes that occur immediately following the trauma and return to baseline without ever showing signs of persistence or reappearance are not relevant to the biological and behavioral correlates of PTSD. Furthermore, given that the onset of PTSD is often delayed for months or even years in many individuals, one would expect to see a gradual worsening of biological abnormalities with the passage of time that culminate in larger scale defects as the behavioral symptoms worsen. By the same token, following clinical expression of the full "PTSD-like syndrome," a good animal model would also theoretically allow for the possibility of recovery or attenuation of symptoms in addition to relapse or late onset.

In considering paradigms that might model biological aspects of PTSD, one would choose those which identify long-term biological changes following stressors. It has been well documented that several biological systems are altered immediately following stress; however, many of these systems show a rapid recovery from stress. The mechanisms involved in maintaining short-term homeostasis in response to stress are probably not relevant to the induction of the more chronic sequelae that lead to the development of PTSD. For this reason, paradigms in which animals are only studied immediately following exposure to stress do not produce information directly relevant to the biology of PTSD unless they are also followed by longterm studies.

Short-term or acute biological responses to trauma can be relevant to the subsequent development of PTSD if they establish or sensitize stress-adaptive systems toward further (i.e., future) dysregulation and therefore should be measured. Furthermore, certain acute biological or behavioral alterations may only occur in individuals who will subsequently move on to develop a more long-term syndrome. As such, short-term responses to stress may be decisive predictors of the full-blown chronic syndrome in some individuals. By identifying these immediate biological changes, it is theoretically possible to explore the therapeutic and prophylactic effects of pharmacological interventions at earlier time points; however, to demonstrate a relationship between immediate biological changes in response to stress and the subsequent development of PTSD, it is essential that studies of acute biological responses be followed up by studies of long- term responses to stress.

4. The Stressor Should Induce Biobehavioral Alterations that have the Potential for Bidirectional Expression.

As noted above, the core symptoms of PTSD include the bidirectional manifestations of both enhanced (intrusive reexperiencing) and reduced (avoidance and/or numbing) responsiveness to environmental stimuli that recall the initial trauma. Although the avoidance symptoms are thought to be invoked in early stages following the trauma and then as a coping response to the intrusive symptoms that subsequently appear (Arnold 1985), both symptom clusters must be present concurrently to fulfill the diagnostic criteria for PTSD. It is therefore essential that any animal model of PTSD exhibit the capacity for bipolar, or bidirectional, manifestations. Moreover, the intensity of avoidance symptoms tends to alternate with that of intrusive symptoms and may in fact result from the neurobiological activation or disinhibition that may underlie intrusive symptoms. Therefore, an ideal animal model would additionally show cycling between excitatory and inhibitory biobehavioral changes. To the extent that this may be a more difficult criterion to achieve than others, it could serve as a form of Occam's Razor to discriminate among potential models.

5. Interindividual Variability in Response to a Stressor Should be Present either as a Function of Experience (e.g., Prior Stress History & Poststressor Adaptations), Genetics, or an Interaction of the Two.

Here we suggest that several factors other than the trauma may contribute to the induction of PTSD. This criterion is borne out by the fact that although exposure to trauma remains the most salient predictor of who develops PTSD, not all persons who are exposed to extreme trauma develop symptoms. This observation has prompted a wealth of studies examining the effect of prior stress history, cognitive appraisal of the trauma, premorbid psychiatric functioning and character structure, family studies, and coping skills on PTSD phenomenology. It is now clear that many of the above factors may influence both the induction and the manifestations of this disorder.

The major implication of interindividual variability in response to stress for animal modeling in PTSD is in establishing the fact that other variables clearly do play a role in modulating the response to stress and trauma. An appropriate animal model for PTSD, then, might be one in which the stressor cannot always elicit PTSD-like sequelae in all organisms. The demonstration of interindividual variability and the elucidation of the factors that give rise to these differences will ultimately allow us to tackle important etiologic issues related to PTSD, such as vulnerability versus resistance to stress.

Reexamining the Utility of "Inescapable Shock-Learned Helplessness" as a Model for PTSD

In the animal stress literature, uncontrollable stress such as "inescapable shock" describes a specific behavioral paradigm in which animals are tested with a series of several shocks (sometimes as many as 60-180 trials) from which they cannot escape ("control" rats can escape) and then are placed in an alternative stress paradigm from which it is possible to escape. Animals that have been previously exposed to the inescapable shock have trouble learning to escape even when this option is presented and often exhibit marked passivity and apparent numbing. The behavioral responses of animals subjected to uncontrollable stress, such as inescapable shock, have been referred to as "learned helplessness," which refers (in an anthropomorphosized way) to the "despair" than ensues following the "realization" that attempts to escape stress will not be helpful. It should be noted that the cognitive appraisal implicit in the IS-LH model has been disputed in favor of neurochemical explanations by several noted investigators (Weiss et al 1976, 1981, 1984; Anisman 1984). Nevertheless, the terms "inescapable shock" and "learned helplessness" are often used interchangeably (Krystal et al 1989). The hallmark of the IS-LH paradigm is that through repeated conditioning, the organism learns that attempts to escape the extreme physical stress imposed on it will fail and as such develops the well-documented maladaptive behaviors (e.g., passivity, numbing, inability to learn subsequent escape tasks) and resultant biological manifestations (i.e., catecholamine depletion).

The model of "inescapable shock" or"learned helplessness" was initially proposed and is still studied as a model of depression (Mater and Seligman 1976; Weiss et al 1970; Anisman 1984). Both the neurochemical finding of norepinephrine depletion and the behavioral passivity appeared to reflect clinical aspects of this disorder (Weiss et al 1976, 1981, 1984; Seligman 1975; Anisman 1984). Although major depression and PTSD have now been shown to be distinct from one another from a clinical (Southwick et al 1991) and biological (Mason et al 1986; Yehuda et al 1990, 1991a, 1991b) perspective, particularly with regard to catecholamine alterations (Kosten et al 1987; Perry et al 1991; Yehuda et al 1993; Southwick et al 1991b), the paradigm of IS-LH continues to be accepted by many investigators as an animal model for PTSD (van der Kolk 1987; Krystal et al 1989, 1991; Southwick et al 1992).

How Does the IS-LH Paradigm Fit the Proposed Criteria?

Because PTSD can develop after a relatively brief or even instantaneous exposure to trauma, this would tend to mitigate the importance of learning and conditioning to the development of the core PTSD symptoms (i.e., intrusive reexperiencing, avoidance, hyperarousal). After a brief exposure to trauma, an organism does not have the opportunity to "learn" a maladaptive response or learn to obtain "controllability" over its environment. The types of stimuli that induce PTSD can be over long before an individual can even come close to developing control (or helplessness) over the situation. Yet, that is the time in which the seeds are presumably planted that later result in the persistent effects of trauma. Given that the IS-LH paradigm does not account for the possibility of developing symptoms after exposure to single, very brief or acute stress, nor does it provide a way of evaluating behavioral manifestations in an untrained animal in the presence of a novel exposure to trauma, the model fails to satisfy criterion 1. Furthermore, because the biobehavioral manifestations can only be induced in rats that cannot escape shock, compared with those that can escape shock, the model fails to satisfy criterion 2, requiring that any stressor of sufficient intensity should be capable of producing the PTSD-like sequelae.

A further limitation of the IS-LH model is that most of biobehavioral alterations induced do not appear to be "chronic" but, rather, appear to return to normal within 72 hr (Seligman 1975; Weiss et al 1976). In contrast, PTSD symptoms intensify over time (Archibald and Tuddenham 1965). This would suggest that criterion 3 is not satisfied by this model. It may be, however, that these alterations sensitize the organism toward future stress and as such may contribute to subsequent biobehavioral consequences (i.e., similar to those of PTSD). For example, Anisman and colleagues have shown that some long-term behavioral and neurochemical deficits can be produced in the IS-LH paradigm if the uncontrollable stress is modest and persistent, and they have argued that the conditioning or sensitization of neurochemical changes induced by the initial stressor may be responsible for the long-term behavioral effect (Anisman et al 1978; Anisman and Zacharcko 1990).

Another criterion that is not met by the model is that of bidirectionally (criterion 4). It should be noted at the outset that one of the major limitations of animal modeling is that the behavioral repertoire of animals is limited, and only externally manifest behaviors can be meaningfully measured. This makes it impossible to address questions concerning many "intrusive" symptoms, such as spontaneous "reliving" of the trauma. Nevertheless, it is possible to measure the response to a trauma mimetic stimulus or recall stressor in an experimental paradigm. Moreover, one can expect to see at least some evidence of the physiological hyperarousal that is characteristic of the sequelae of stress and trauma. Unlike intrusive symptoms, hyperarousal symptoms, such as irritability, hypervigilence, increased autonomic arousal, or explosive, impulsive behaviors, do lend themselves to behavioral analysis, allowing for the assessment of physiological "bidirectionality. "

In IS-LH the passivity observed following uncontrollable stress does appear to parallel the numbing and avoidance seen in PTSD; however, the manipulations characteristic of the classic IS-LH paradigm do not seem to elicit many behaviors that resemble intrusive or hyperarousal symptoms. It should be noted that in cases where a more moderate stressor is used, a short period of hyperexcitability may occur. This arousal is thought to act against the appearance of behavioral deficits associated with passivity. Of importance, however, the passivity and hyperexcitability do not occur simultaneously or as alternating behaviors as they do in PTSD.

Finally, because the IS-LH paradigm accounts well for interindividual variability in stress responsivity (criterion 5), it may be relevant to exploring vulnerability factors associated with the development and course of PTSD. It has recently been shown that IS-LH provoked a wide range of neurochemical and behavioral effects in different strains of mice. That some strains appeared to be more vulnerable than others to the behavioral alterations in response to stress is a finding that can potentially contribute greatly to our understanding of individual differences to stressors, which is a cornerstone question of PTSD (Anisman and Zacharko 1990); however, this information should be evaluated in the context of the fact that the model does not meet the other criteria described herein.

Certainly, factors such as controllability or helplessness may influence how PTSD is manifested and should be examined; however, the identification of factors that influence such manifestations of the disorder need to be clearly differentiated from those that directly contribute to the induction of core PTSD symptoms because they may be limited to certain types of stress responses and not be generalizable to the syndrome of PTSD. The IS-LH paradigm, for example, does appear to simulate the specific trauma of combat exposure (i.e., basic training followed by subsequent chronic high-intensity stressors from which there may often be no escape and/or followed by experiences from which escape is possible). Therefore, examining the role of factors such as "controllability" could theoretically contribute to an understanding of the significant differences that have been observed across clinical groups in manifestations of PTSD, such as form, intensity, chronicity of illness, level of impairment, or comorbidity (that appear to be particularly common among war veterans but not necessarily other traumatized groups). Of note, however, although the IS-LH paradigm may simulate aspects of the stressor of combat, the behavioral aftermath of combat in war veterans, which consists of highly aggressive, explosive, and impulsive behaviors (Keane et al 1985), is not consistent with the behavioral passivity observed by laboratory animals. This may be yet another instance of an "apparent" face validity that is not borne out on closer scrutiny.

A Promising Animal Model for PTSD: Time-Dependent Sensitization

Another paradigm that has been proposed as modeling the clinical syndrome of PTSD is time-dependent sensitization (TDS) (Antelman 1988; Antelman and Yehuda 1993; Rosen and Fields, 1988). TDS refers to the fact that one exposure to a stressor (e.g., injection of pharmacologic agent, immobilization stress) can induce an extremely longlasting alteration in the subsequent responsiveness of the organism to pharmacological or nonpharmacological stressors. In this paradigm, rats receive one typically very brief exposure to an inducing stressor and are later tested with the same or another recall stressor. Physiological and behavioral responsivity to the second stressor are significantly altered in animals previously exposed, compared with those receiving the stressor for the first time. In those instances where more than one interval between the inducing and recall stimuli have been measured, it has been shown that this effect progresses with time since the first stressor (Antelman 1988). In other words, the influence of the first stressor strengthens entirely as a function of the increased passage of time.

The time-dependent effects on behavioral and physiological systems are ubiquitous. Alterations have been found in indices of four different neurotransmitter systems (dopamine, norepinephrine, serotonin, and y-aminobutyric acid); endocrine systems (e.g., corticosterone and ACTH) (Caggiula et al 1989; Antelman et al 1991a, 1991b); the immune system (Antelman et al 1990); the cardiovascular system (Antelman et al 1989a); and in carbohydrate metabolism (Antelman et al 1991a).

TDS meets the proposed criteria. First, it can be induced by stressful events imposed on the organism for only seconds or, alternatively, by more chronic and severe stressors, such as inescapable shock. Such persistent TDS effects lasting at least 1 month have been shown after a single jab with an empty syringe needle or one injection of saline (Antelman et al 1988, 1989a, 1989b). Second, it is "dose dependent," having recently been shown after the strongest (defined in terms of corticosterone levels) of a graded series of psychological stressors (S.M. Antelman et al 1992). Also consistent with the second criterion, and s noted above, it is induced by all types of different stressors, including physical (both controllable and uncontrollable), psychological, pharmacological, and metabolic agents (Antelman 1988; Antelman et al 1991a, 1991b). Third, the consequences of TDS both persist for extremely long periods and, as mentioned, grow stronger with the increased passage of time, similar to what is observed in chronic or delayed PTSD. Fourth, the effects of TDS can be either excitatory or inhibitory (Antelman et al 1991a, 1991b), thus meeting the criterion for bipolarity. For example, it has very recently been found that the behavioral effect of a "lower intensity" stressor is progressively enhanced, whereas the effect of "higher intensity" stressors are progressively inhibited with the passage of time (Antelman et al 1991b). Fifth, as is the case with PTSD, TDS shows marked interindividual variability (Robinson 1988; Antelman et al 1992).

Conclusion

The animal stress literature provides a rich source of information about PTSD; however, given the number of stress paradigms and the unique behavioral and biochemical consequences that accompany them, it is important to choose judiciously among the various options. We believe that the above criteria will allow investigators to more successfully pair animal models of stress with the specific clinical syndrome of PTSD. The paradigm of time-dependent sensitization illustrates that there appear to already be potential animal models of stress that fulfill the necessary criteria for successfully modeling the onset of PTSD in response to trauma. Other animal models of stress, such as fear-potentiated startle (Davis 1986), experimental neurosis (Foa et al 1992), and long-term potentiation (LeDoux et al 1989), have also been suggested as reflecting important aspects of PTSD. To the extent that these models satisfy the proposed criteria, they, too, may contribute substantially to our understanding of the biological aspects of PTSD.

References

American Psychiatric Association (APA) (1987): American Psy chiatric Association: Diagnostic and Statistical Manual of Mental Disorder, 3rd ed. rev. Washington, DC: APA Press.

Anisman H (1984): Vulnerability to depression: Contribution of stress. In Post RM and Ballenger JC (eds), Neurobiology of Mood Disorders. Baltimore: Williams & Wilkins, pp 407-431.

Anisman H, Sklar LS (1979): Catecholamine depletion upon reexposure to stress: mediation of the escape deficits produced by inescapable shock. J Comp Psychol 93:610625.

Anisman H, Zacharko RM (1990): Multiple neurochemical and behavioral consequences of stressors: implications for depression. Pharmacol Ther 46:119-136.

Anisman H, DeCantanzaro D, Remington G (1978): Escape performance following exposure to inescapable shock: deficits in motor response maintenance. J Exp Psychol (Anim Behav) 4:197- 218.

Antelman SM (1988): Time-dependent sensitization as the cornerstone for a new approach to pharmacotherapy: drugs as foreign/stressful stimuli. Drug Dev Res 14:1-30.

Antelman SM, Kocan D, Knopf S, Edwards DJ, Caggiula AR (1992): One brief exposure to a psychological stressor induces long-lasting, time-dependent sensitization of both the cataleptic and neurochemical responses to haloperidol. Life Sciences 51:261-266.

Antelman SM, Yehuda R (1993): Time-dependent change following acute stress: relevance to the chronic and delayed aspects of posttraumatic stress disorder. In Murburgh MM (ed), Catecholamines and PTSD: Emerging Concepts. Washington, DC: APA Press.

Antelman SM, Knopf S, Kocan D, Edwards DJ, Ritchie JC, Nemeroff CB (1988): One stressful event blocks multiple actions of diazopam for up to at least a month. Brain Res 445:380-385.

Antelman SM, DeGiovanni LA, Kocan D (1989a): A single exposure to cocaine or immobilization stress provides extremely longlasting selective protection against sudden cardiac death from tetracaine. Life Sci 44:201-207.

Antelman SM, Kocan D, Edwards DJ, Knopf S (1989b): Anticonvulsant and other effects of diazepam grow with time after a single treatment. Pharmacol Biochem Behav 33:3139.

Antelman SM, Cunnick JE, Lysle DT, et al (1990): Immobilization 12 days (but not I hour earlier) enhanced to 2-deoxyd-glucose- induced immunosuppression: Evidence for stressor-induced time- dependent sensitization of the immune system. Prog Neuropsychopharmacol Biol Psychiatry 14:579-590.

Antelman SM, Caggiula AR, Knopf S, Kocan DJ, Barry H III (1991a): Exposing rats to a single brief stressor two weeks earlier modifies the response of plasma. ACTH and glucose to ethanol (ETOH) (abstract). Neuroscience 17:494.96.

Antelman SM, Caggiula AR, Kocan D, et al (199lb): One experience with "lower" or "higher" intensity stressors, respectively enhances or diminishes responsiveness to haloperidol weeks later: implications for understanding drug variability. Brain Res 566:276- 283.

Archibald HC, Tuddenham RO (1965): Persistent stress reaction after combat: a twenty-year follow-up. Arch Gen Psychiatry 12:475-481.

Arnold AL (1985): Diagnosis of post-traumatic stress disorder in Vietnam veterans. In Sonnenberg SM, Blank AS Jr, Talbott JA (eds), The Trauma of War: Stress and Recovery in Vietnam Veterans. Washington, DC: APA Stress, pp 99-123.

Caggiula AR, Antelman SM, Aul E, Knopf S, Edwards DJ (1989): Prior stress attenuates the analgesic response but sensitizes the corticosterone and cortical dopamine responses to stress 10 days later. Psychopharmacology 99:233-237.

Davis M (1986): Pharmacological and anatomical analysis of fear conditioning using the fear potentiated startle paradigm. Behav Neurosci 100:814-824.

Foa E, Zinbarg R, Rothbaum B (1992): Uncontrollability and unpredictability in Posttraumatic stress disorder: an animal model. Psychol Bull 112:218-237.

Friedman MJ (1991): Biological approaches to the diagnosis and treatment of posttraumatic stress disorder. J Traum Stress 4:67-91.

Giller EL (1991): Biological Assessment and Treatment of Posttraumatic Stress Disorder Washington, DC: APA Press.

Helzer JE, Robins LN, McEvoy L (1987): Post-traumatic stress disorder in the general population: findings of the epidemiological catchment area survey. New Engl J Med 317:163 1634.

Kardiner A (1941): The Traumatic Neuroses of War. New York: P. Hoeber.

Keane TM, Fairbanks JA, Cadell JM, Zimmering RT, Taylor KL, Mora CA (1985): A behavioral approach to assessing and treating post- traumatic stress disorder in Vietnam veterans. In Figley CR (ed), Trauma and Its Wake: The Study and Treatment of Post-traumatic Stress Disorder. New York: Brunner/Mazel.

Kosten TR, Mason JW, Giller EL, Ostroff RB, Harkness L (1987): Sustained urinary norepinephrine elevation in posttraumatic stress disorder. Psychoneuroendocrinology 12:1330.

Krystal JH, Kosten TR, Perry BD, et al (1989): Neurobiological aspects of PTSD: review of clinical and preclinical studies. Behav Ther 20:177-198.

Kulka RA, Schlenger WE, Fairbank JA, et al (1990): Trauma and the Vietnam War Generation. New York: Brunner/Mazel.

LeDoux JE, Romanski L, Xagoraris A (1989): Indelibility of subcortical emotional memories. J Cog Neuroscience I: 238-243.

Maier SF, Seligman MEP (1976): Learned helplessness: theory and evidence. J Exp Psychol: I :3-46.

Malloy PF, Fairbank JA, Keane TM (1983): Validation of a multimethod assessment of posttraumatic stress disorders in Vietnam veterans. J Consult Clin Psychol 51 :488-494.

Mason JW, Giller EL, Kosten TR, et al (1986): Urinary free cortisol levels in post-traumatic stress disorder patients. J Nerv Ment Dis 174:145-149.

Mason JW, Giller EL, Kosten TR, Yehuda R (1990): Psychoendocrine approaches to the diagnosis and pathogenesis of PTSD. In Giller EL (ed), Biological Assessment and Treatment of PTSD. Washington, DC: APA Press, pp 65-86.

Perry BD, Southwick SM, Yehuda R, Giller EL (1990): Adrenergic receptor regulation in post-traumatic stress disorder. In Giller EL (ed), Biological Assessment and Treatment of Posttraumatic Stress Disorder. Washington, DC: APA Press, pp 87-114.

Rosen J, Fields R (1988): The long-term effects of extraordinary trauma: A look beyond PTSD. Journal of Anxiety Disorders 2:179- 191.

Robbins TW, Sahakian BJ (1980): Animal models of mania. In: Belmaker RH, van Praag HM (eds), Mania: An Evolving Concept. New York: Spectrum Publications, pp 143-216.

Robinson TE (1988): Stimulant drugs and stress: factors influencing individual differences in the susceptibility to sensitization. In Kalivas PW and Barnes CD (eds), Sensitization in the Nervous System. Telford, NJ: pp 119-144.

Seligman MEP (1975): Helplessness: On Depression, Development and Death. San Francisco: Freeman.

Southwick SM, Krystal JH, Charney DS (1990): Yohimbine in PTSD. (abstract). New Res 143:NR478.

Southwick SM, Yehuda R, Giller EL (1991): Characterization of depression in war-related posttraumatic stress disorder. Am J Psychiatry 148:179-182.

Southwick SM, Krystal J, Johnson D, Charney DS (1992): Neurobiology of PTSD. Annual Review of Psychiatry. Washington, DC: APA Press, 1992.

van der Kolk B (1987): The drug treatment of post-traumatic stress disorder. J Affective Disord 13:203-213.

van der Kolk B, Greenberg M, Boyd H, Krystal J (1985): Inescapable shock, neurotransmitters, and addiction to trauma: toward a psychobiology of posttraumatic stress disorder. Biol Psychiatry 20:314-325.

van Dyke C, Zilberg NJ, McKinnon JA (1987): Am J Psychiatry 144:51-55.

Weiss J, Goodman PA, Losito BG, Corrigan S, Charry JM, Bailey WH (1981): Behavioral depression produced by an uncontrollable stressor: relationship to norepinephrine, dopamine, serotonin levels in various regions of rat brain. Brain Res Rev 3: 167-205.

Weiss JM, Simpson PA (1984): Neurochemical mechanisms underlying stress-induced depression. In Field T, McCabe P, Schneiderman N (eds), Stress and Coping New Jersey: Lawrence Erlbaum, pp 93-116.

Weiss JM, Glazer Hl, Pohorecky LA (1976): Coping behavior and neuchemical changes: An alternative explanation for the original "learned helplessness" experiments. In Sehan G, Kling A (eds), Animal Models in Human Psychobiology. New York: Plenum, pp 141-174.

Willner P (1986): Validation criteria for animal models of human mental disorders: learned helplessness as paradigm case. Prog Neuropsychopharmacol Biol Psychiatry 10:677-690.

Yehuda R (1992): Neuroendocrinology of PTSD. In McClelland RJ, Nemeroff CB (eds), Current Topics in Psychiatry. London: Current Science Ltd, February, 5:109-112.

Yehuda R, Southwick SM, Nussbaum G, Wahby V, Giller EL, Mason JW (1990): Low urinary cortisol excretion in patients with posttraumatic stress disorder. J Nerv Ment Dis 178:366369.

Yehuda R, Giller EL, Southwick SM, Lowy MT, Mason JW (1991a): Hypothalamic-pituitary-adrenal dysfunction in PTSD. Biol Psychiatry 30:1031-1048.

Yehuda R, Lowy MT, Southwick SM, Shaffer D, Giller EL (199lb): Lymphocyte glucocorticoid receptor number in posttraumatic stress disorder. Am J Psychiatry 148:499-504.

Yehuda R, Southwick SM, Krystal JH, Charney DS, Mason JW (1993): Enhanced suppression of cortisol following dexamethasone administration in posttraumatic stress disorder. Am J Psychiatry 150:83-86.

Yehuda R, Southwick SM, Giller EL (1992): Exposure to atrocities and symptom severity in combat-related PTSD. Am J Psychiatry 149:333-336.

Zacharko RM, Anisman H (1989): Pharmacological, biochemical and behavioral analysis of depression. Animal Models. In Koob G, Ehlers C, Kupfer D (eds), Animal Models of Depression. Boston, Birkhauser.