If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Over the past three decades of psychiatric research, abnormalities in the noradrenergic system have been identified in particular anxiety disorders such as panic disorder. Simultaneously, neuroscience research on fear pathways and the stress response have delineated central functions for the noradrenergic system. This review focuses on the noradrenergic system in anxiety spectrum disorders such as panic disorder, generalized anxiety disorder, and phobias for the purpose of elucidating current conceptualizations of the pathophysiologies. Neuroanatomic pathways that are theoretically relevant in anxiogenesis are discussed and the implications for treatment reviewed.
It has long been recognized that particular somatic symptoms in pathological anxiety, such as hyperventilation, palpitations, and perspiration, are indicative of altered autonomic activity. In 1871, when characterizing the pathology observed in soldiers with “irritable heart,” Da Costa referred to the “hyperaesthesia of cardiac nerve centres” (
). Almost a century later Klein reported that the mostly noradrenergic tricyclic antidepressant imipramine was effective in blocking panic attacks in agoraphobic patients. Klein’s seminal observation along with his suggestion that drug response may be useful for the delineation of psychiatric syndromes ushered in the psychopharmacological era that has, in large part, led to our current neurobiological understanding of pathological anxiety (
). With the operationalization of the American Psychiatric Association’s Diagnostic and Statistical Manual in 1980, biological research in anxiety has benefited from a rational approach based on illness categories. Administration of particular pharmacological agents in the laboratory setting, known as pharmacological “probes,” has permitted the identification of abnormal behavioral, endocrine, and autonomic responses between patients with specific anxiety disorders and in comparison to healthy controls. The use of such agents as probes has allowed further dissection of autonomic abnormalities in these disorders. Altered activity in noradrenergic systems has emerged as a key element of many aspects of pathological anxiety.
In certain respects, anxiety disorders research has reached a new crossroads that has demanded a paradigm shift to one which incorporates preclinical neuroscience research. Despite the major benefits of DSM-directed, criteria-based research, many studies now point to abnormalities that cut across several DSM diagnoses and adhere best to behavioral phenomena common to several anxiety disorders. Also, comorbidity and clinical overlap in anxiety disorders are extensive (
). Hence, in this new paradigm it is now expected that focus on individual symptom phenomena—such as separation distress, phobic avoidance, anticipatory anxiety, and panic attacks—rather than on the diagnoses in which these symptoms occur may better reveal the brain mechanisms relevant to forms of anxiety. We and others believe that such pathological fear symptoms have underpinnings directly related to preclinical “fear circuitry” identified by neuroscience research. This article focuses on abnormalities in noradrenergic systems, including the autonomic nervous system (ANS), relevant to pathological anxiety. In particular, we focus on panic disorder (PD) as it includes symptom phenomena that are common to several anxiety disorders. Also, particular aspects of PD make the disorder particularly amenable to biological study, and there is a wealth of biological data that can be drawn upon for the current theoretical framework. It is hoped that such a focus will shed light on other anxiety spectrum disorders, such as generalized anxiety disorder (GAD), and phobic disorders (social phobia [SP], specific phobia, and agoraphobia), and relevant data on these disorders are incorporated. Every attempt is made to bridge clinical and preclinical theory for the heuristic purpose of building a cogent model indicating the potential pathogenic mechanisms in such disorders.
Noradrenergic function in pathological anxiety
Some of the most notable findings involving pharmacological probes have pointed to a role for noradrenergic dysregulation in particular anxiety disorders. Yohimbine is an α2-adrenergic receptor antagonist that is generally believed to increase synaptic availability of norepinephrine (NE) by antagonism at autoreceptors on the presynaptic neurons. Clonidine, on the other hand, is an α2-adrenergic agonist that may decrease activity at noradrenergic synapses by agonism at presynaptic neurons. In PD patients, administrations of yohimbine and clonidine both elicit abnormal responses compared with the respective responses in controls (
Abnormal regulation of noradrenergic function in panic disorders Effects of clonidine in healthy subjects and patients with agoraphobia and panic disorder.
). Yohimbine elicits high rates of paniclike anxiety in PD patients, accompanied by greater cardiovascular response and increase in serum level of the principal noradrenergic metabolite, 3-methoxy-4-hydroxyphenylglycol (MHPG). The sensitivity of yohimbine appears relatively specific to PD, although in posttraumatic stress disorder (PTSD) about a third of yohimbine-challenged subjects experience paniclike anxiety (
). Clonidine, which has acute anxiolytic and antifear effects in humans and animals, replicably results in a blunted growth hormone (GH) response in PD patients (
). Such a response has been theorized to be due to presynaptic autoreceptor supersensitivity in PD, though the possibility of postsynaptic subsensitivity has also been raised. To date only one group has reported a negative blunted GH finding to clonidine in PD patients (
In addition to blunted GH response to clonidine, other types of challenge have also indicated abnormal autonomic activity in GAD and SP. Females with GAD were shown to have an attenuated skin conductance response to stress as well as a slower recovery to baseline skin conductance poststress (
), suggesting a hyporesponsive and prolonged autonomic response in GAD. Orthostatic challenge in SP has demonstrated elevated supine and standing NE levels compared with PD patients and controls (
Heart rate and plasma norepinephrine responsivity to othostatic challenge in anxiety disorders Comparison of patients with panic disorder and social phobia and normal control subjects.
). Similarly, increased blood pressure response to Valsalva’s maneuver and exaggerated vagal withdrawal in response to isometric exercise has also been identified in SP (
). Orthostatic challenge, isometric exercise, and Valsalva’s maneuver in PD patients do not appear to result in differences in cardiorespiratory or plasma catecholamine response in comparison to normal controls (
). Noradrenergic volatility describes the magnitude of plasma MHPG oscillatory activity, utilizing the root of the mean square successive differences statistic. Compared with healthy controls, untreated PD patients showed markedly increased within-subject MHPG volatility, and only a limited group overlap was observed (patients, mean ± SD, 1.96 ± 0.92 per hour vs. controls, 0.519 ± 0.28 ng/mL per hour; Mann–Whitney U test, z = −4.7; p < .001; effect size, 2.18) (Figure 1). Figure 2 indicates the excessive plasma MHPG volatility by showing each subject’s baseline and successive postclonidine plasma MHPG levels. Note the large oscillations in MHPG in the PD group in contrast to the relatively horizontal values observed in controls during clonidine challenge. It is important to consider that all 16 controls had values of plasma MHPG change per hour of less than 1.25 ng/ml, whereas only 3 of 17 patients showed values below this post hoc cutoff, indicating 100% specificity and 82% sensitivity (p < .001, Fisher’s exact) (Figure 1).
Figure 1Comparison of the means and SDs of within-subject plasma 3-methoxy-4-hydroxyphenylglycol (MHPG) volatility during clonidine challenge between untreated patients with panic disorder (PD) and healthy volunteers. The scatterplot depicting the individual values is on the left of the group mean and SD. Mann–Whitney Utest, z = −4.7; p < .001. Application of cutoff is post hoc, p < .001, Fisher’s exact test. (Reproduced with permission from Coplan et al 1997.)
Figure 2Successive plasma 3-methoxy-4-hydroxyphenylglycol (MHPG) levels following administration of clonidine (indicated by Clonidine and arrow) in patients with panic disorder (PD) and in healthy volunteers. The mean group differences in patterns of plasma MHPG production are shown in Figure 1. Plasma MHPG volatility is determined using the root of the mean square successive difference (Siever et al 1992) from the four postclonidine time points (indicated by Volatility and arrows). (Reproduced with permission from Coplan et al 1997.)
The links between stress and anxiety have long been assumed, but the neurobiological bases of such links are still only beginning to emerge. Animal models of stress have delineated major components of the stress response. In order for an organism to direct attention rapidly to a threatening situation and prepare for defensive action, there is a relatively uniform cascade of events resulting in a state of increased mental and physical arousal necessary for survival actions in response to the challenge (
). The central systems responsible for the response to stress appear to be the NE system and the corticotropin-releasing factor (CRF) system. Norepinephrine neurons originating in the locus ceruleus (LC), as well as in other nuclei in the medulla and pons, are activated during stress and send projections to cortical and subcortical regions also believed to be important in mediating fear and fear responses (
). Projection sites include the prefrontal and entorhinal cortices, the amygdala, the bed nucleus of the stria terminalis (BNST), the hippocampus, the periaqueductal gray (PAG), the hypothalamus, the thalamus, and the nucleus tractus solitarius (NTS) (see Figure 3). The LC is innervated by areas, such as the amygdala, known to process fear-relevant sensory stimuli and from areas such as the medullary nucleus paragigantocellularis, which receives viscerosensory stimuli relayed by the NTS (
Amygdaloid corticotropin-releasing factor targets locus coeruleus dendrites Substrate for the co-ordination of emotional and cognitive limbs of the stress response.
Light and electron microscopic evidence for topographic and monosynaptic projections from neurons in the ventral medulla to noradrenergic dendrites in the rat locus coeruleus.
). Therefore it appears the LC is uniquely positioned to integrate both external sensory and internal visceral data and influence a wide distribution of stress- and fear-related neural structures, including specific cortical areas.
Figure 3Widespread projections of the pontine locus ceruleus to telencephalic, diencephalic, mesencephalic, and medullary structures implicated in fear, stress, and anxiety. AMYG = amygdala, HIP = hippocampus, HYP = hypothalamus, LC = locus ceruleus, NTS = nucleus tractus solitarius, PAG = periaqueductal gray, PFC = prefrontal cortex, TH = thalamus.
Localization of novel corticotropin-releasing factor receptor (CRF2) mRNA expressiojn to specific subcortical nuclei in rat brain Comparison with CRF1 receptor mRNA expression.
). Corticotropin-releasing factor, like NE, has emerged as a neurotransmitter that plays a central role not only in stress but also in anxiety and depression. The cell bodies, terminals, and receptors for CRF have a pattern of distribution that anatomically maps onto key structures involved in response to adversity (
). Regions in which CRF neurons are predominantly found include the paraventricular nuclei (PVN) of the hypothalamus, the central nucleus of the amygdala (CNA), and the lateral BNST. The activity of CRF is clearly altered during an adaptive response (
), and when CRF is artificially administered to the ventricles of experimental animals, the characteristic physical and behavioral adaptations of the stress response are produced (
). Further, CRF is now hypothesized to serve as a facilitator of both the cognitive and physical symptoms of anxiety by enhancing transmission through key structures in the stress response system.
There is much evidence that the CRF and LC–NE systems cross-regulate their activities. Paraventricular nuclei CRF neurons also project to NE neurons of the brain stem, such as those in the LC, which reciprocally project back to the PVN, linking the two main central constituents of the stress system (
). Stress results in increased CRF concentrations in the LC, and when infused into the LC, CRF increases LC neuronal firing rate, cortical NE levels, and cortical electroencephalographic activity (
Activation of the locus coeruleus noradrenergic system by intracoerulear microinfusion of corticotropin-releasing factor Effects on discharge rate, cortical norepinephrine levels and cortical electroencephalographic activity.
). Recent tracer studies indicate that the CNA is a source of CRF projections to the LC, and the CNA–LC pathway may play a role in the integration of emotional and cognitive responses to stress (
Light and electron microscopic evidence for topographic and monosynaptic projections from neurons in the ventral medulla to noradrenergic dendrites in the rat locus coeruleus.
). Figure 4 is a schematic diagram of key inputs and outputs of the amygdala involved in the processing of emotional stimuli and the production of the fear response.
Figure 4Key inputs and outputs of the amygdala involved in the processing of emotional stimuli and the production of the fear response. AB = accessory basal nucleus of the amygdala, AC = auditory cortex, B = basal nucleus of the amygdala, CNA = central nucleus of the amygdala, EC = entorhinal cortex, HIP = hippocampus, LA = lateral nucleus of the amygdala, LC = locus ceruleus, LH = lateral hypothalamus, PAG = periaqueductal gray, PBN = parabrachial nucleus, PFC = prefrontal cortex, PVN = paraventricular nuclei of the hypothalamus (which received inputs from the CNA directly and via the bed nucleus of the stria terminalis), SC = somatic cortex, TH = thalamus, VC = visual cortex.
). Corticotropin-releasing factor released by the PVN of the hypothalamus activates the pituitary–adrenal response, while the LC–NE system activates the ANS, including both the sympathetic-adrenomedullary branch and the parasympathetic branch. The endocrine response to the stress-induced activity of these two limbs includes the adrenal gland release of glucocorticoids and epinephrine, respectively. Activation of the ANS has direct effects on multiple organ systems including the cardiovascular, pulmonary, urogenital, and gastrointestinal systems. Components of the both the CRF system axis and the ANS also have direct effects on the immune system (
reported that lesions of the temporal lobe in the nonhuman primate, among other effects, virtually abolished both fear of natural dangers, such as snakes, and typical fear behavior when confronted with novelty. Since then there has been extensive elucidation of fear neural circuitry, and most evidence points to the amygdala as the central component (
). In auditory fear conditioning, the pairing of an innocuous tone, the conditioned stimulus (CS), with a mild yet aversive foot shock, the unconditioned stimulus (US), leads to cellular activation and synaptic plastic changes in fear circuits. Even after only one such pairing the rat will later respond to the CS alone with behavioral, neuroendocrine, and autonomic changes indicative of fear of an impending shock. This response to the tone alone includes freezing, stress hormone release, cardiovascular and pulmonary activation, and somatic reflex potentiation. Contextual fear conditioning involves the acquisition of an association between the foot shock and a representation of the environment (the context) in which the foot shock was received. After such conditioning, the rat responds to the context alone with all the same fear responses described. A conditioned fear response to a discrete cue, such as a tone, or a context can therefore be elicited by stimuli indicative of real danger or stimuli that are simply predictive of danger. Fear associations can also be made without the occurrence of adversity (i.e., no US). Such “higher order” conditioning involves the pairing of a new innocuous stimulus (CS2) with a past CS such that the new stimulus will elicit a fear response when later presented alone.
The amygdala appears to process fear-relevant information in series and in parallel through an organized array of intraamygdaloid circuits formed by individual amygdalar nuclei (
). In the case of auditory fear conditioning, auditory CS information proceeds from the ear through the brain stem to the auditory relay in the thalamus, known as the medial geniculate body (
). From there the information is relayed to the amygdala by two parallel pathways: a quick monosynaptic thalamus–amygdala pathway and a slower polysynaptic thalamus–cortex–amygdala pathway, which both converge on the lateral nucleus of the amygdala (LA). The former pathway is believed to provide a fast yet unprocessed representation of stimuli that is necessary for immediate action. The latter pathway contains cortically processed information that may provide for a more appropriate action when the stimuli is measured against cortical memory from past experience. Complex intraamygdala pathways may allow a stimulus representation to be modulated by different functional systems such as hormonal and homeostatic systems and those mediating memories of past experiences of the organism (
Cue and contextual fear conditioning appear to occur by different intraamygdala pathways. Auditory information, relayed through the thalamus, converges with information about the shock on cells of the LA. Such information is then relayed to the CNA. In contrast, contextual information appears to require the hippocampus to first create a representation of the environment utilizing information received from the entorhinal cortex and subiculum (
). Such information is relayed through the basal/accessory basal (B/AB) nuclei of the amygdala and then to the CNA. Thus, lesions of the LA block auditory fear conditioning (
). The CNA is conceived of as the main output station of the amygdala as it projects to and activates regions critical for the expression of the conditioned fear response (
). Central nucleus of the amygdala projections to the BNST and PVN are involved in the hypothalamic–pituitary–adrenal (HPA) axis activation, those to the parabrachial nucleus, lateral hypothalamus, and dorsal motor nucleus of the vagus are involved in cardiorespiratory stimulation, and those to the PAG are involved in freezing behavior and pain modulation (
) (see Figure 4). In addition to the discussed projection of the CNA to the noradrenergic LC, the CNA also projects to the ventral tegmental area and the raphe nuclei, thus connecting it with key dopaminergic and serotonergic loci, respectively (
Damasio theorizes, due to studies with humans who have damage to their prefrontal region, that the ventromedial PFC is necessary for the association of new sensory input with the memory of the type of emotional state usually associated with that class of situation in prior experience (
). The PFC may establish a somatosensory activity pattern that marks the scenario as good or bad and therefore constrains the decision-making space. In this model, the amygdala is seen as a central autonomic effector activating appropriate somatic responses. In the rodent, an intact medial PFC has been shown to be critical to the normal extinction of fear response to repeated presentations of a CS (
), and it is hypothesized that LC–NE activation of the medial PFC increases the amplitude of the PFC neuronal response to incoming signals by decreasing their basal firing (
). Amygdala–PFC–LC interactions may therefore be critical for the establishment of the appropriate emotional valence to a given situation and are thus implicated in pathological fear and anxiety.
Stress and fear conditioning
Several lines of evidence indicate stress activation of LC–NE and CRF systems can influence functioning of various components of the fear system. Corticosteroid administration on the day of testing of fear behavior to a CS alone results in potentiated learned fear behavior to a given explicit auditory cue (
). Social isolation, a stressful situation in the rat, impairs auditory fear conditioning. Normally, response occurs only to the exact frequency of the tone CS, but stress results in generalization of responsiveness to other frequencies (
The effects of AMPA-induced lesions of the septo-hippocampal cholinergic projection on aversive conditioning to explicit and contextual cues and spatial learning in the water maze.
). In fact, NE release in the amygdala upon stress, such as with foot shock, may be a critical link between activation of the stress system, including the effects of corticosteroids and epinephrine, and the induction of fear memory for associated cues and contexts (
). Infusions of β-adrenergic antagonists into the amygdala as well as lesions of the amygdala completely block the fear memory enhancement seen with stressful stimuli or fear-enhancing pharmacological manipulation. Thus there is evidence that stress and related stress molecules, including noradrenergic molecules, modulate the acquisition and expression of both cue and contextual fear memory, influencing both the magnitude of later response and the degree of generalization.
The role of the amygdala in human fear and anxiety
There is mounting evidence of a phylogenetic conservation of function across mammals with respect to the amygdala’s role in establishing emotional valence to stimuli, predicting adversity, and coordinating the fear response. Stereotaxic electrical stimulation of the amygdala in conscious humans has been reported to elicit symptoms of anxiety disorders including fear, anxiety, depersonalization, and changes in autonomic function (
). Neuroimaging studies indicate amygdala involvement in human conditioned fear as well as conscious and unconscious responses to fearful facial expressions (
). As neuroimaging techniques rapidly progress, there will be further opportunity to compare findings from neuroscience with human neuroanatomy and function, and individuals with localized damage to “fear circuitry” structures offer rare opportunities to make such comparisons. One report indicated that a patient with bilateral amygdala damage could not acquire conditioned autonomic responses to visual or auditory cues, while another patient with bilateral hippocampal damage could condition but could not acquire declarative facts about which cues were paired with the US, and a patient with both amygdala and hippocampal damage could acquire neither conditioning nor facts (
). In a separate investigation, individuals with bilateral amygdala damage demonstrated poor ability to judge unfamiliar faces as untrustworthy or unapproachable, compared to such judgments of trustworthiness and approachability by normal controls (
Early environmental influences on stress response, the fear system, and anxious behavior
When considering the factors involved in the pathogenesis of anxiety disorders, it is helpful to divide development into two discrete periods in which both the types of threats and the architectonics of the response systems are very different. It is assumed, from a wealth of evidence not reviewed here, that there is a strong genetic component to fearful traits, anxious behavior, and the development of anxiety disorders. Animal studies have revealed that during the postnatal period the thresholds for future activation of the stress response system are sculpted by the environment. For example, rat pups that are handled by their human caretakers for short periods of time and then returned to their dams have long been noted to appear more resilient in the face of stress and demonstrate less anxietylike behavior. Recent work has demonstrated that dams of handled pups increase the time licking and grooming their pups while nursing (
). Even in the absence of handling there are variations among rats in the level of maternal licking and grooming while nursing. Adult rats that received high levels of such maternal behavior during the first 10 days of life show reduced plasma corticotropin and cortisol in response to restraint stress. It was further demonstrated that CRF messenger RNA expression in the PVN negatively correlated, and glucocorticoid receptor (GR) messenger RNA in the hippocampus positively correlated, with the frequency of such licking and grooming. Increased sensitivity to glucocorticoid feedback at the level of the hippocampus is believed to be at least partially responsible for the decreased levels of corticotropin and CRF observed in response to stress. The adult offspring of the handling effect have also been shown to have increased benzodiazepine receptor density in the amygdala and LC, decreased CRF receptor density in the LC, and increased α2-adrenergic receptor density in the LC (
In a maternal deprivation paradigm, the period rat pups are away from the dams is extended to 180 min (which is longer than the separation time would normally be in the wild) (
Ladd CO, Huot RL, Thrivikraman KV, Plotsky PM (1998, November): Persistent alterations in the negative feedback regulation of the hypothalamic-pituitary-adrenal (HPA) axis in maternally-separated adult Long Evans hooded rats. Abstract presented at the meeting of the Society for Neuroscience, Los Angeles, CA.
Ladd CO, Stowe ZN, Plotsky PM (1999, March): Maternal separation alters NA and CRF neurocircuits: Reversal by antidepressants. Abstract presented at “Norepinephrine: New Vistas for an Old Neurotransmitter,” Key West, FL.
Stowe ZN, Tang ZL, Plotsky PM, Nemeroff CB (1998, November): The effects of maternal separation on the neuronal activity in locus coeruleus and nucleus accumbens. Abstract presented at the meeting of the Society for Neuroscience, Los Angeles, CA.
). Control groups receive 15 min normal handling. The latency to retrieval of the first pup after handling by the dams is shorter in the 15-min group than in the 180-min group. Though baseline firing rates of neurons in the LC are similar between groups, LC neuronal excitation to paw pinch is markedly elevated in the 180-min group. When the 180-min group is stressed by restraint or air puff startle, the corticosterone response is greater, but the response is similar to controls in response to hemorrhagic stress. The 180-min group has increased CRF messenger RNA in the PVN, the CNA, and the BNST and increased CRF in the cerebrospinal fluid (CSF); increased tyrosine hydroxylase and decreased α2-adrenoreceptor binding in the LC; and decreased GRs and increased mineralocorticoid receptors (MRs) in the hippocampus. Remarkably, both the selective serotonin reuptake inhibitor (SSRI) paroxetine and the selective NE reuptake inhibitor (SNRI) reboxetine, administered for 21 days, normalize the HPA axis response to stress as well as the CRF level in the CSF; CRF messenger RNA in the PVN and the CNA; and the GR/MR ratio in the hippocampus. These data from the handling and deprivation paradigms indicate that the early environment has dramatic effects on the CRF system, the amygdala, and the LC–NE system, thus directly shaping future response to stress.
Work by our group in the nonhuman primate suggests an even closer connection between adversity in the early developmental environment and subsequent development of anxiety disorders. Earlier work by our collaborators indicated that subtle manipulations of the early psychosocial environment of infant bonnet macaques result in anxietylike behavioral profiles in youth and adult life (
). In this paradigm, nursing mothers of such infant monkeys are subjected to unpredictable demands when foraging for food, as opposed to predictable low-demand and predictable high-demand control environments, which results in adversely altered behavior toward the infant (
). As a result of this unpredictable environment, the infants have heightened anxietylike behavior throughout development including increased behavioral inhibition in response to separation, to fear stimuli, and to new social groups and environments (
). A close analogy can be made between this behavior and what Jerome Kagan described in certain children as “behaviorally inhibited to the unfamiliar” (
). When challenged with the α2-antagonist yohimbine, these monkeys have been shown to be hyperresponsive, similar to responses seen in humans with PD and PTSD. In fact, the CSF of these monkeys has CRF levels that are persistently elevated while CSF cortisol levels are depressed (
). This pattern is similar to the elevated CSF CRF and depressed peripheral cortisol observed in humans with PTSD, although abnormalities in CSF CRF levels have not been conclusively identified in other anxiety disorders. There are also significant correlations in these monkeys, not seen in controls, between CSF CRF and heightened CSF levels of serotonin and dopamine metabolites and the GH axis peptide somatostatin. A theoretical explanation for such lasting biochemical abnormalities is that increased adversity in the mother–infant interaction results in CRF overexpression and this, in turn, results in alterations in other systems, such as the LC–NE system, relevant to stress and anxiety. In separate but related work, electroencephalogram recording in abnormally fearful rhesus macaques has demonstrated relatively higher right frontal lobe activity, supporting the notion that a fearful phenotype is related to grossly altered regional activity (
These macaque studies provide important links between fear and stress research in rodents and anxiety disorders research in humans. Further, prospective human studies by Kagan and coworkers have linked inhibited behavioral response to the unfamiliar in early development with subsequent anxious behavior. Four-month-old infants who manifest a low threshold to becoming distressed and motorically aroused to unfamiliar stimuli are more likely than others to become fearful and subdued during early childhood (
). Moreover, children who were identified as behaviorally inhibited at 21 months and remain inhibited at follow-ups (ages 4, 5.5, and 7.5 years old) have higher rates of anxiety disorders than children who were not consistently inhibited (
Construction of a conceptual framework of anxiogenesis and anxiety pathology
From the foregoing developmental studies it would appear that adversity in the early environment, whether due to unpredictability of caretaker nurturing, neglect, or prolonged separation, has an interactive effect with the genetic predispositions such that general thresholds of activation are set for the CRF system, the LC–NE system, and the amygdala. In this framework, the genotype can be conceived of as on a spectrum from facilitatory to ameliorating for the development of maladaptive responses to stress. Indeed, with the handling model of rats and the unpredictable psychosocial environment model in monkeys there are the unexpectedly resilient and the unexpectedly poorly adapted outlying individuals whose reactivity does not correspond with the group response to the environment.
For the later environment, after such thresholds have been set, it can be assumed that stressful life events continue to elicit stress responses and may cumulatively add to fear memory. Integrating stress and conditioning findings, conditioned memory might, theoretically, also have effects on future stress response. Thus, even though thresholds have been set, there would be the potential for facilitated responses due to memory of past adverse experience. Therefore, theoretically, a response to a stressful situation after early development is the sum of the severity of the current stress plus the memory of cues and contexts from past stresses. Such a response may be appropriate, excessive, or prolonged.
Cumulative fear memory and activation of fear pathways may be inherently related to anxiety. This hypothesis posits that acute anxiety is elicited when cues in the present environment have been previously associated with stress-inducing experience. Chronic anxiety in this model is a result of repetitive cuing. For example, a person who has lost several close relatives to cancer may become acutely anxious solely at the mention of the word cancer. When this person is told of an abnormal cancer-screening test, several visits to the doctor for the work-up may induce a state of chronic anxiety that continues even after the work-up definitively demonstrates no evidence of cancer. In this case, the emotional expectation of a deadly outcome continues despite sound evidence to the contrary.
A sudden autonomic discharge as occurs during a panic attack, whether occurring spontaneously or seemingly evoked by a cue or a context, leads to avoidance behavior that attractively fits with avoidance paradigms of fear conditioning, especially conditioned avoidance. Therefore, vulnerability to disorders such as GAD, PD, and SP may be explained by the culmination of faciliatory genetics, early environmental sculpting, and current environmental stressful life events. Thus, the stress response, though clearly necessary for survival in an environment full of challenges to homeostasis, may not be finely tuned for appropriately gauged responses to all permutations of the human environment. The combination of “jumpy” CRF and LC–NE systems due to early environmental sculpting and increased amygdala-mediated cue and contexual fear memories due to past adversity may make future responses much less specific, such that anxious and/or avoidant behavior predominates.
Pathological effects of anxiety on individual organ systems
The term allostatic load has been developed to refer to excessive activity of a response system resulting in pathophysiological damage in an organism (
Induction of corticotropin-releasing hormone gene expression by glucocorticoids implications for understanding the states of fear and anxiety and allostatic load.
). For example, monkeys that live for several years in a colony in which they are socially subordinate to a dominant male develop stomach ulcers and have marked degeneration of the hippocampus, both evidence of allostatic load (
). In anxiety disorders such as GAD, PD, and SP there is much evidence for increased autonomic and central arousal and marked susceptibility to psychosocial stress. The expectation of negative events—whether manifest as worry, as in the cancer work-up example, or as fear of the autonomic surge of a panic attack—can be theoretically conceived of as allostatic load in which cognition has been adversely altered such that catastrophic misappraisals persist.
Particular symptoms in panic attacks, such as numbness and tingling, dizziness, and altered visual perception, are similar to symptoms observed in cerebral ischemia. One neurology service reported that 5% of referrals for evaluation of focal neurological symptoms were due to PD without clinical evidence for comorbid neurological disease (
). The authors noted that hyperventilation reproduced the specific focal findings in 42% of these PD patients. Hyperventilation results in hypocapnia, and hypocapnia in turn causes cerebral blood vessels to constrict and cerebral blood flow (CBF) to decrease (
). To date, two studies using Doppler measurement of CBF have demonstrated that PD patients have greater decrease in basilar artery blood flow during hyperventilation compared with controls (
). Several functional brain-imaging studies involving anxious subjects reveal some degree of decreased CBF compared with controls. For example, with xenon single photon emission computed tomography (SPECT) measurement during lactate infusion, normal controls manifested the expected increase in CBF, but subjects who panicked had either a significantly blunted increase or even a decrease in global CBF (
). With Tc-HMPAO SPECT imaging, PD patients challenged with yohimbine demonstrated increased anxiety and decreased frontal CBF when compared to controls (
Yohimbine is generally believed to increase availability of synaptic NE by α2 antagonism at autoreceptors on the presynaptic neurons. Mathew has suggested that LC–NE activation in anxiety can bring about reductions in intracranial blood flow (
). Not only do sympathetic fibers innervate larger cerebral vessels via the superior cervical and stellate ganglia, but the LC appears to have fibers projecting to intraparenchymal cerebral microvessels. Hyperventilation, particularly during stress imposed by the laboratory setting, may therefore result in greater hypocapnia-induced vasoconstriction in PD patients through the added contribution of LC activation. Our group recently presented blood-flow data from quantitative positron emission tomography neuroimaging studies supporting the idea that PD patients tend to converge on a lower CBF value in response to hypocapnia, and the degree of such CBF decreases does not appear to be accounted for by the degree of hypocapnia achieved during hyperventilation (
Kent JM, Coplan JD, Abi-Dargham A, et al (1998): Changes in cerebral blood flow during hyperventilation in panic disorder. Presented at the 37th annual meeting of the American College of Neuropsychopharmacology, Las Croabas, Puerto Rico.
). Since this effect appears to be global rather than local, it is plausible that the added vasoconstrictive effect is accounted for by activation of other, far-reaching systems. The LC, with its extensive innervation of the cerebral microvasculature, is a likely candidate because of its known activation during both stress and fear response.
Cardiac rate abnormalities have emerged as a feature of pathological anxiety. Preclinical work on the fear network has shown that the cardiovascular response to a CS appears to depend on both sympathetic and parasympathetic components, with vagal influences serving to suppress heart rate acceleration to a certain degree (
). Multiple clinical studies have shown that PD patients have faster heart rates at baseline in the laboratory and that panic attacks, at least in the laboratory and possibly in a natural setting, are routinely accompanied by increases in heart rate (
). Several years ago we showed that the tachycardia developed during laboratory-induced panic is probably not mediated by stimulation of peripheral β-adrenergic receptors by circulating catecholamines (
). In more recent work with cardiac transplant patients we have demonstrated that direct innervation of the heart by the brain via sympathetic and parasympathetic (vagal) fibers, and not circulating catecholamines, is crucial for heart-rate response to psychological stress (
). One method for measuring direct innervation of the heart by the brain is heart period variability (HPV) determination. Normal cardiac rhythm varies on a beat-to-beat basis with respiration and depends on intact parasympathetic innervation via the vagus nerve. Pharmacological or surgical disruption of parasympathetic nerves to the heart eliminates normal beat-to-beat HPV. Spectral analysis of HPV shows a low frequency peak mediated by both sympathetic and parasympathetic systems related to baroreceptor control, and a high frequency peak mediated by parasympathetic mechanisms related to respiratory sinus arrhythmia (
). Consequently, many believe that the ratio of low-frequency HPV to high-frequency HPV gives an index of sympathetic influence. In general, loss of parasympathetic control of the heart leads to a decrease in opposition to the sympathetic influence. This results in more rapid heart rate and may leave the heart prone to the proarrhythmic influence of the sympathetic system.
Several studies indicate that patients with high levels of “phobic anxiety” and PD have increased rates of sudden death as well as coronary artery disease (
). This may be due to a decrease in cardiac vagal tone or an increase in cardiac sympathetic responsiveness, both of which have been associated with an increase in cardiovascular mortality (
). Also, decreased ultralow-frequency and very low-frequency spectral domains have been shown to be independent predictors of cardiac morbidity and sudden death in cardiac patients (
). Increased heart rate during laboratory panic also appears secondary to decreased parasympathetic innervation, and several studies show reduced HPV in subjects with PD (
) there appears to be a normalization of HPV in PD patients, with a decrease in sympathetic and an increase in parasympathetic influence.
Gastrointestinal disorders of unclear etiology such as irritable bowel syndrome have been shown to be associated with affective and anxiety disorders (
). Such pathways may serve to coregulate the sacral parasympathetic nervous system and the brain noradrenergic system. Barrington’s nucleus appears to have complex interconnections with the LC, the PAG, and the forebrain; and evidence is emerging that this center may be involved in particular stress- and anxiety-related dysfunction in the gastrointestinal and genitourinary systems (
Direct projections from the periaqueductal gray to pontine micturition center neurons projecting to the lumbosacral cord segments An electron microscopic study in the rat.
Impact of corticotropin-releasing hormone on gastrointestinal motility and adrenocorticotropic hormone in normal controls and patients with irritable bowel syndrome.
). Corticotropin-releasing factor infusion into the LC/subceruleus nuclei also affects both gastric acid secretion and colonic motor function via spinal pathways (
Microinfusion of corticotropin-releasing factor into the locus coeruleus/subcoeruleus nuclei inhibits gastric acid secretion via spinal pathways in the rat.
). These pathways are now considered to coordinate pelvic visceral functions with forebrain activity, with the LC serving as a key relay. Thus the neural basis for common urogenital and gastrointestinal symptoms in anxiety states is becoming clearer.
Treatment implications
Preclinical studies utilizing agents with clinical anxiolytic efficacy have indicated that these agents alter the LC–NE system and its projection sites. For example, chronic systemic imipramine was shown to decrease α2-receptor density in the LC and the A2 region of the NTS, without significant changes observed in the amygdala, the pyriform cortex, the PAG, and the BNST (
). This may suggest that chronic imipramine treatment preferentially modulates the α2-receptor population localized in the brain stem NE-rich nuclei rather than at areas that receive the terminal projections. Systemic administration of either the tricyclic desipramine (DMI) or the tetracyclic maprotiline, both with relatively selective noradrenergic activity, was shown to inhibit the firing rate of noradrenergic LC neurons and induce excitation of the PFC neurons (
). This effect was blocked by intracerebroventricular 6-hydroxydopamine, indicating that the prefrontal activity may be due to noradrenergic neurons of LC. Systemic and intra-LC administration of DMI each had effects on NE levels in the cingulate cortex and increased LC NE levels by an α2-dependent mechanism (
Somatodendritic alpha2-adrenoceptors in the locus coeruleus are involved in the in vivo modulation of cortical noradrenaline release by the antidepressant desipramine.
). Bupropion was also shown to inhibit firing rates of NE cells in the LC at doses significantly lower than those that inhibit activity of midbrain DA cells or dorsal raphe serotonin cells, while cathecholamine depletion by reserpine pretreatment greatly reduced this inhibition of LC firing (
). The SSRI fluoxetine, when given systemically, increased both tissue levels of MHPG and microdialysate concentrations of NE and serotonin in the hypothalamus. This was not observed with fluoxetine administered locally to the hypothalamus, indicating noradrenergic innervation from another site such as the LC (
In various stress models, there is also the suggestion that chronic administration of clinically efficacious anxiolytics may work, at least in part, by countering the effects of CRF activation of the LC. For example, in one stress model chronic imipramine increased the density of noradrenergic neurons in the frontal cortex, which otherwise decreases as an effect of stress (
), and whereas intracerebroventricular CRF alters the LC response to repeated sciatic nerve stimulation, chronic administration of phenelzine and sertraline produces the opposite effect. Rats of a strain very susceptible to stress-induced gastric ulcers were shown upon repeated daily stress to have decreased NE transporter sites in the LC, amygdala, and hypothalamus (
The effects of alprazolam on corticotropin-releasing factor neurons in the rat brain Implications for a role for CRF in the pathogenesis of anxiety disorders.
In our study identifying elevated noradrenergic volatility in PD patients during challenge with clonidine the subjects were rechallenged after 12 weeks of treatment with the SSRI fluoxetine. Fluoxetine treatment response was associated with a significant reduction in MHPG volatility during clonidine administration (see Figure 5). Also, the PD patients who received fluoxetine and had the greatest between-visit reductions of the basal MHPG were more likely to show clinical improvement. These results suggest SSRIs may have a noradrenergic stabilizing effect. This is consistent with the attenuating effects of fluvoxamine on yohimbine-induced anxiety in PD patients (
Figure 5Individual successive plasma 3-methoxy-4-hydroxyphenylglycol (MHPG) levels following clonidine administration before and after fluoxetine treatment in 11 responders with panic disorder. Fluoxetine response was associated with a significant reduction of plasma MHPG volatility. (Reproduced with permission from Coplan et al 1997.)
We have previously reviewed both the use of SSRIs in the spectrum of anxiety disorders and possible mechanisms of action of the SSRIs, which include enhancement of inhibitory serotonergic afferents of the dorsal raphe nuclei that project to the LC (see
Aguiar LM, Haskins T, Rudolf RL, et al (1998, May–June): Double-blind, placebo-controlled study of once-daily venlafaxine extended release in outpatients with GAD. Abstract presented at the annual meeting of the American Psychiatric Association, Toronto, Canada.
), though it is not yet clear if its noradrenergic properties play a role. It will be interesting to see what the efficacies of the new, better tolerated SNRIs such as reboxetine are in this spectrum of anxiety disorders. In depression, reboxetine appears to have potency equal to that of tricyclic antidepressants and, more importantly, may be superior to SSRIs, especially in assessment of depression-related social functioning and negative self-perception (
). And it was recently reported that reboxetine, in a randomized, placebo-controlled trial, has demonstrable efficacy in PD with significant reductions in major attacks, phobic symptoms, and illness severity scores (
Brown MT, Dubini A, DiNicolo P, et al (1999, March): Double-blind, placebo-controlled study of reboxetine in panic disorder. Abstract presented at “Norepinephrine: New Vistas for an Old Neurotransmitter,” Key West, FL.
For SP, the monoamine oxidase inhibitors (MAOIs), including the reversible inhibitor brofaramine and the irreversible inhibitor phenelzine, have produced robust response rates (
). Considering alternate strategies to monoaminergic manipulation, clearly benzodiazepines have been a mainstay in the treatment of anxiety syndromes over many decades. They have shown beneficial results in PD, GAD, and, in a study with the high-potency benzodiazepine clonazepam, SP as well. Yet it may be the case that the γ-aminobutyric acid system has less potential for specific manipulation of anxiety pathways than monoaminergic approaches. On the other hand, newer approaches, such as CRF antagonists currently in the pipeline, offer exciting possibilities for modulating the stress and anxiety systems by very specific mechanisms.
Conclusion
The noradrenergic system appears to play a key role in the pathophysiology of anxiety spectrum disorders. Historically, clinical response to noradrenergic agents along with the successful utilization of noradrenergic pharmacological probes in the laboratory setting focused much attention on this system. During the same period, neuroscience research has identified important roles for noradrenergic transmission in both fear circuitry and the stress response. Clearly, anxiety disorders research is at an exciting stage in which the clinical and preclinical findings are beginning to converge. This has allowed the development of more cogent hypotheses about the pathophysiology of anxiety and related medical disorders and offers the potential for newer therapies specifically directed to modulate or correct the pathophysiology at more basic levels. In light of the current convergence of data from different disciplines, refocusing attention on the function of the noradrenergic in anxiety disorders highlights both the immense complexity of anatomic structures and circuitry involved and how far indeed the field has come in elucidating the neurobiology.
Acknowledgements
This work was presented at the conference, “Norepinephrine: New Vistas for an Old Neurotransmitter,” held in Key West, Florida in March 1999. The conference was sponsored by the Society of Biological Psychiatry through an unrestricted educational grant provided by Pharmacia & Upjohn.
References
Abelson J.L.
Glitz D.
Cameron O.G.
et al.
Blunted growth hormone response to clonidine in patients with generalized anxiety disorder.
Aguiar LM, Haskins T, Rudolf RL, et al (1998, May–June): Double-blind, placebo-controlled study of once-daily venlafaxine extended release in outpatients with GAD. Abstract presented at the annual meeting of the American Psychiatric Association, Toronto, Canada.
Brown MT, Dubini A, DiNicolo P, et al (1999, March): Double-blind, placebo-controlled study of reboxetine in panic disorder. Abstract presented at “Norepinephrine: New Vistas for an Old Neurotransmitter,” Key West, FL.
Impact of corticotropin-releasing hormone on gastrointestinal motility and adrenocorticotropic hormone in normal controls and patients with irritable bowel syndrome.
Kent JM, Coplan JD, Abi-Dargham A, et al (1998): Changes in cerebral blood flow during hyperventilation in panic disorder. Presented at the 37th annual meeting of the American College of Neuropsychopharmacology, Las Croabas, Puerto Rico.
Ladd CO, Huot RL, Thrivikraman KV, Plotsky PM (1998, November): Persistent alterations in the negative feedback regulation of the hypothalamic-pituitary-adrenal (HPA) axis in maternally-separated adult Long Evans hooded rats. Abstract presented at the meeting of the Society for Neuroscience, Los Angeles, CA.
Ladd CO, Stowe ZN, Plotsky PM (1999, March): Maternal separation alters NA and CRF neurocircuits: Reversal by antidepressants. Abstract presented at “Norepinephrine: New Vistas for an Old Neurotransmitter,” Key West, FL.
Somatodendritic alpha2-adrenoceptors in the locus coeruleus are involved in the in vivo modulation of cortical noradrenaline release by the antidepressant desipramine.
The effects of AMPA-induced lesions of the septo-hippocampal cholinergic projection on aversive conditioning to explicit and contextual cues and spatial learning in the water maze.
Microinfusion of corticotropin-releasing factor into the locus coeruleus/subcoeruleus nuclei inhibits gastric acid secretion via spinal pathways in the rat.
Stowe ZN, Tang ZL, Plotsky PM, Nemeroff CB (1998, November): The effects of maternal separation on the neuronal activity in locus coeruleus and nucleus accumbens. Abstract presented at the meeting of the Society for Neuroscience, Los Angeles, CA.
Light and electron microscopic evidence for topographic and monosynaptic projections from neurons in the ventral medulla to noradrenergic dendrites in the rat locus coeruleus.
00246-2&id=gr1.gif){kind=link}
00246-2&id=gr2.gif){kind=link}
00246-2&id=gr3.gif){kind=link}
00246-2&id=gr4.gif){kind=link}
00246-2&id=gr5.gif){kind=link}