Neural circuits in anxiety and stress disorders: a focused review

Abstract

Anxiety and stress disorders are among the most prevalent neuropsychiatric disorders. In recent years, multiple studies have examined brain regions and networks involved in anxiety symptomatology in an effort to better understand the mechanisms involved and to develop more effective treatments. However, much remains unknown regarding the specific abnormalities and interactions between networks of regions underlying anxiety disorder presentations. We examined recent neuroimaging literature that aims to identify neural mechanisms underlying anxiety, searching for patterns of neural dysfunction that might be specific to different anxiety disorder categories. Across different anxiety and stress disorders, patterns of hyperactivation in emotion-generating regions and hypoactivation in prefrontal/regulatory regions are common in the literature. Interestingly, evidence of differential patterns is also emerging, such that within a spectrum of disorders ranging from more fear-based to more anxiety-based, greater involvement of emotion-generating regions is reported in panic disorder and specific phobia, and greater involvement of prefrontal regions is reported in generalized anxiety disorder and posttraumatic stress disorder. We summarize the pertinent literature and suggest areas for continued investigation.

Keywords:

• fear
• anxiety
• neuroimaging

Introduction

Anxiety and stress disorders are among the most prevalent categories of mental illnesses, with a median onset at age 11 years and a lifetime prevalence of 28%.¹ When left untreated, anxiety symptoms persist and are associated with significant impairments in functioning, poor quality of life, and a huge economic burden.^2–4 Anxiety disorders are of particular importance in the context of recent and ongoing world conflicts, as environmental factors can have a strong impact on anxiety and stress disorder development, particularly posttraumatic stress disorder (PTSD).⁵ Given the high prevalence rates, negative effects on many aspects of functioning, and environmental factors associated with trauma and stress, it is imperative that we continue to improve our understanding of the mechanisms underlying anxiety disorder presentations in an effort to improve existing treatments.

Although anxiety disorders have been extensively studied, the literature examining underlying neural mechanisms remains scarce, with relatively little evidence identifying specific deficits for various anxiety disorders. Despite the lack of concrete knowledge regarding the specific mechanisms underlying anxiety, both pharmacologic (selective serotonin reuptake inhibitors) and psychotherapeutic (cognitive behavioral therapy) treatments for anxiety management have been developed. These treatments are effective for many patients suffering from anxiety, but the exact mechanisms of action are not well known. Moreover, many patients do not have access to or do not experience complete symptom relief with the existing evidence-based treatments. It is thus imperative that we continue striving for an improved understanding of the specific neural deficits underlying anxiety disorder presentations and the mechanisms of action by which effective treatments reduce anxiety symptomatology. The widely accepted categorization of anxiety disorders is based solely on behavioral and subjective experiences. However, it is possible that the neural mechanisms underlying anxiety symptomatology overlap across a number of discretely defined disorders. In fact, arguments are emerging that anxiety disorders may fall along a continuum ranging from specific fear-based reactivity to more diffuse and prolonged stress or apprehension. Identifying mechanisms will allow for more accurate diagnostic categories, improved ability to predict individual treatment response, and will provide specific targets for new and improved treatments aimed at correcting aberrant neuronal functioning.

Clinical categorization of anxiety disorders

According to the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition,⁵ anxiety and stress disorders are characterized by an excessive fear response and/or worry that interferes with functioning or causes significant distress. This class of disorders includes panic disorder (PD), specific phobia (SP), social anxiety disorder (SAD), PTSD, and generalized anxiety disorder (GAD).

Panic disorder

PD is characterized by sudden panic attacks, often occurring unexpectedly, followed by a month or more of worrying about having another attack or the consequence of the attack (eg, heart attack, stroke). Common symptoms during a panic attack include racing heart, shortness of breath, tightness in chest, paresthesia, gastrointestinal distress, sweating, hot/cold flashes, fear of dying, and fear of losing control. PD sufferers often develop agoraphobia, avoiding places or situations where they think they might have a panic attack.

Specific phobia

SP is characterized by excessive fear triggered by a specific object or situation. SP falls into four subtypes, including animal, natural environment (eg, heights, storms), blood-injection-injury, and other (choking, vomiting, illness, costumed characters, etc). The excessive fear brought on by the phobic object or situation leads to intense distress, anxious anticipation, panic attacks, and/or avoidance of the feared object or situation.

Social anxiety disorder

SAD is characterized by persistent fear of social or performance situations resulting from the possibility of negative judgment, embarrassment, or humiliation. Cognitive distortions and self-monitoring in social situations, involving hyperawareness of internal cues and behaviors, are often associated with SAD. Feared social situations are avoided or tolerated with dread.

Posttraumatic stress disorder

PTSD can develop after exposure to serious injury, death, or a potential threat to the physical integrity of self or others. Symptom clusters include intrusive reexperiencing symptoms (memories of trauma, nightmares, flashbacks); avoidance of trauma-related thoughts, memories, contexts or cues; negative mood and cognition; and hyperarousal/hypervigilance.

Generalized anxiety disorder

The core symptom in GAD is excessive and continuous worry, anxiety, and apprehensive expectation in multiple contexts. The ambiguity and diversity of the sources of stress and anxiety distinguishes this disorder from cue-related anxiety disorders such as PD, SP, SAD, and PTSD. There is a cognitive component to GAD that is characterized by worrisome thoughts and cognitive errors.

Fear versus anxiety

An important but controversial distinction in the anxiety and stress disorder literature is between the constructs of fear and anxiety. Fear and anxiety have considerable overlap with respect to subjective, behavioral, physiological, and neurological characteristics. However, some key differences separate the two. Fear is typically defined as a phasic and abrupt fight-or-flight response accompanied by intense arousal in response to an immediate and identifiable threat. Alternatively, anxiety is often defined by a more prolonged state of tension, worry, and apprehension regarding uncertain, and potentially negative, future events.⁶ Although fear and anxiety serve important evolutionary functions to keep us safe, fear allows us to combat or avoid immediate threats or danger, whereas anxiety increases vigilance and improves our ability to identify uncertain or potential threats. Still, anxiety disorders can develop when anxiety or fear responding is excessive or occurs in the absence of true threat, either immediate or future.

Although fear and anxiety both play a role in all anxiety disorder presentations, some argue that anxiety and stress disorders can be placed along a fear–anxiety continuum, with PD and phobias classified as fear-based disorders and PTSD and GAD as primarily anxiety-based disorders.⁵ In fact, evidence from rodent models of fear and anxiety suggests that different neural mechanisms may underlie fear and anxiety states. For example, amygdala lesions block fear responding to predictable and identifiable threat but do not affect prolonged anxiety states in response to uncertain future threat. In contrast, lesions of the bed nucleus of the stria terminalis diminish anxiety states while leaving phasic fear responses intact.⁷ A study examining physiological responses to threat in humans demonstrated a gradient of fear reactivity across the anxiety disorder spectrum, with exaggerated fear reactivity in SP, diminished fear reactivity in GAD and PTSD, and intermediate levels of fear reactivity in the other anxiety disorders.⁸ These findings support the notion that fear and anxiety might be somewhat distinct processes, playing different roles across the spectrum of anxiety and stress disorders.

To identify neural circuits underlying fear and anxiety symptomatology, specific experimental paradigms have been developed.⁹ Both rodent and human studies of fear typically employ a fear conditioning paradigm, which pairs an aversive stimulus (eg, electric shock) with a neutral stimulus (eg, a blue light), resulting in the neutral stimulus becoming a signal of imminent threat. This experimental manipulation permits dissociation of brain circuits that imbue “fear” from those involved in the processing of the same or identical stimuli without fear content. A number of key structures have been identified that generate and modulate fear responses to the conditioned threat signal. Specifically, thalamus integrates sensory input from the primary sensory cortices and sends output to the amygdala. The amygdala and dorsal anterior cingulate cortex (dACC) process aversive signals and send output to the hypothalamus, basal ganglia, and brainstem to produce defensive behaviors.^10,11 The hippocampus is responsible for encoding contextual information associated with the threat cue.¹¹ In this way, the hippocampus has been implicated in the extinction of fear, playing a role in down-regulating amygdala response in safe contexts. The medial prefrontal cortex (mPFC) provides top-down regulatory control of fear responding, receiving input from the hippocampus and thalamus and projecting to the amygdala to modulate fear behavior on the basis of complex environmental information.¹⁰ Overall, these findings support involvement of thalamus, amygdala, dACC, hypothalamus, hippocampus, and mPFC in fear circuitry.¹² These regions work in concert to both generate and modulate fear responses to imminent and identifiable threat.

Similarly, both rodent and human studies have been used to identify regions involved in anxiety. Typically, these studies involve unpredictable presentation of an aversive stimulus, leading to the development of vigilance, tension, anticipation, and worry. Many of the same regions making up the fear circuitry also underlie anxiety. Some overlapping regions implicated in both fear and anxiety circuitry include the thalamus, amygdala, and dACC.¹³ In addition, the insula^14,15 is implicated in vigilance during unpredictable threat, highlighting its role in anxiety circuitry. The bed nucleus of the stria terminalis receives input from the hippocampus, amygdala, and mPFC and mediates anxiety-related behaviors.¹⁶ Emotion regulation regions common to both fear and anxiety include the mPFC and hippocampus. One key difference is the involvement of additional regions in anxiety involved in emotion regulation and attention modulation, including rostral anterior cingulate cortex (rACC) and dorsolateral PFC.^12,17

Three reviews recently summarized structural and functional abnormalities in brain circuitry across the anxiety and stress disorders.^13,18,19 Functional imaging tasks used to probe neurocircuitry of anxiety disorders mostly include symptom provocation through presentation of disorder-related visual or auditory cues, emotional faces, or fear conditioning. Connectivity analyses examine the co-activation of different brain regions in response to the above-mentioned tasks. The purpose of the current work was to review the recent literature and summarize knowledge gained since 2010. We used PubMed to search the literature published from January 2010 until August 2014, with key terms similar to and including combinations of the following: anxiety, social anxiety disorder, social phobia, panic disorder, specific phobia, posttraumatic stress disorder, generalized anxiety disorder, fMRI (functional magnetic resonance imaging), MRI, PET (positron emission tomography), DTI (diffusion tensor imaging), and neuroimaging. Articles were included if they examined brain structure or function in anxiety on the basis of information presented in the abstract section. On the basis of the articles reviewed, we present a breakdown of the structural and functional abnormalities identified for various anxiety disorders, as well as results pertaining to connectivity between brain regions (see Table 1 for summary of articles reviewed). Our main aims were to synthesize the current literature to identify the neural networks underlying anxiety disorder presentations and to determine whether brain structure, function, and connectivity findings point to similar or independent neural networks across anxiety and stress disorders. We expected to find a consistent pattern of hyperactivation of brain areas underlying fear generation (amygdala, insula, dACC) in more fear-related anxiety disorders (PD, SP) and the same pattern coupled with deficits in emotion modulation areas (mPFC and rACC) in disorders with more prominent anxiety and cognitive components (GAD, PTSD, SAD).

Table 1 Summary of studies reviewed reporting structural and functional differences across anxiety and stress disorders

Study	Thalamus	Amygdala	Insula	Hippocampus	Medial prefrontal cortex
Neural volume
Panic disorder		↓ Del Casale et al (2013)^Citation38 ↓ Kim et al (2012)^Citation39 ↓ Pannekoek et al (2013)^Citation40		↓ Del Casale et al (2013)^Citation38 ↓ Pannekoek et al (2013)^Citation40 –
Specific phobia  Social anxiety disorder		↓ Fisler et al (2013)^Citation41 ↓ Irle (2010)^Citation45 ↑ Machado-de-Sousa et al (2014)^Citation47 – Syal et al (2012)^Citation46	– Rosso et al (2010)^Citation67 ↑ Bruhl et al (2014)^Citation66 ↓ Syal et al (2012)^Citation46	↓ Irle (2010)^Citation45 – Syal et al (2012)^Citation46 ↑ Machado-de-Sousa et al (2014)^Citation47 ↑ Talati et al (2013)^Citation108 ↓ Liao et al (2011)^Citation90 ↓ Qiu et al (2011)^Citation104	↑ Liao et al (2011)^Citation90
Generalized anxiety disorder		↑ Schienle et al (2011)^Citation43		↓ Hettema et al (2012)^Citation106	– Cha et al (2014)^Citation91
Posttraumatic stress disorder		↑ Kuo et al (2014)^Citation44 ↓ Rajendra et al (2012)^Citation42	↓ Chao et al (2013)^Citation69 ↓ Kroes et al (2011)^Citation70	↓ Chao et al (2013)^Citation69 ↓ Kuhn et al (2013)^Citation80 ↓ Rajendra et al (2012)^Citation42 ↓ Thomaes et al (2013)^Citation107 ↓ Woon et al (2010)^Citation105	↓ Chen et al (2012)^Citation92 ↓ Kuhn et al (2013)^Citation80 ↓ Nardo et al (2013)^Citation77 ↓ Weber et al (2013)^Citation93
Neural activation
Panic disorder		↑ Kim et al (2012)^Citation39 ↓ Ottaviani et al (2012)^Citation64
Specific phobia	↑ Caseras et al (2010)^Citation26 ↑ Hilbert et al (2014)^Citation27 ↑ Lueken et al (2011)^Citation29 ↑ Schweckendiek et al (2011)^Citation28	↑ Lipka et al (2011)^Citation48 ↑ Lueken et al (2013)^Citation49 ↑ Schweckendiek et al (2011)^Citation28 ↑ Caseras et al (2010)^Citation26	↑ Hilbert et al (2014)^Citation27 ↑ Lipka et al (2013)^Citation71 ↑ Lueken et al (2011)^Citation29 ↑ Schweckendiek et al (2011)^Citation28	↑ Lueken et al (2013)^Citation49
Social anxiety disorder	↑ Bruhl et al (2011)^Citation24 ↑ Gimenez et al (2012)^Citation25	↑ Ball et al (2012)^Citation50 ↑ Bruhl et al (2011)^Citation24 – Klumpp et al (2013)^Citation65 ↑ Klumpp et al (2010)^Citation51 ↑ Labuschangne et al (2010)^Citation55 ↑ Mansson et al (2013)^Citation56 ↑ Phan et al (2013)^Citation57 ↑ Schmidt et al (2010)^Citation52 ↑ Taylor et al (2013)^Citation58	↑ Ball et al (2012)^Citation50 ↑ Carre et al (2014)^Citation72 ↑ Gimenez et al (2014)^Citation31 ↑ Klumpp et al (2012)^Citation73 ↑ Klumpp et al (2013)^Citation65 ↑ Schmidt et al (2010)^Citation52 ↑ Taylor et al (2013)^Citation58	↑ Bruhl et al (2011)^Citation24	↑ Blair et al (2010)^Citation96 ↓ Sripada et al (2009)^Citation97 ↓ Taylor et al (2013)^Citation58
Generalized anxiety disorder		↑ Holzel et al (2013)^Citation60 – Palm et al (2011)^Citation62			↓ Etkin et al (2011)^Citation98 ↓ Palm et al (2011)^Citation62 ↓ Price et al (2011)^Citation99 ↓ Schlund et al (2012)^Citation100
Posttraumatic stress disorder	↑ Patel et al (2012)^Citation30	↑ Milad et al (2010)^Citation53 ↑ Shvil et al (2013)^Citation54 ↑ Zantvoord et al (2013)^Citation59	↑ Fonzo et al (2010)^Citation74 ↑ Garrett et al (2012)^Citation75 ↑ Nardo et al (2013)^Citation77 ↑ Patel et al (2012)^Citation30 ↑ Rougemont-Bucking et al (2011)^Citation76 ↑ Shvil et al (2013)^Citation54	↑ Brohawn et al (2010)^Citation109 ↑ Felmingham et al (2010)^Citation95 ↑ Garrett et al (2012)^Citation75 – Hayes et al (2012)^Citation110 ↓ Milad et al (2009)^Citation53 ↑ Patel et al (2012)^Citation30 – Shvil et al (2013)^Citation54	↑ Felmingham et al (2010)^Citation95 ↑ Fonzo et al (2010)^Citation74 ↑ Garrett et al (2012)^Citation75 ↓ Jovanovic et al (2011)^Citation94 ↓ Patel et al (2012)^Citation30 ↓ Rougemont-Bucking et al (2011)^Citation76 ↓ Zantvoord et al (2013)^Citation59

Note: Arrows preceding citations indicate greater or less volume or activation in anxiety compared with healthy controls.

Sensory processing regions

According to rodent models and human imaging studies, a number of regions responsible for taking in and processing sensory information, such as the occipital cortex, fusiform gyrus, and thalamus, have been implicated in anxiety disorder neurocircuitry.^11,20 The main finding reported is increased activation in these regions in response to threatening stimuli in anxious patients compared to healthy controls.

Occipital cortex and fusiform gyrus

The occipital cortex, a region responsible for processing visual stimuli, is more active in response to threatening images in anxious patients, particularly those with SAD, compared with controls. In patients with SAD, greater activation in occipitotemporal regions predicted D-cycloserine treatment response, with greater pretreatment activation associated with a greater decrease in symptoms posttreatment.²¹ The fusiform gyrus contains neurons specific to face perception, known as the fusiform face area,²² which is more active in response to threat faces in SAD compared with healthy controls.²³

Thalamus

The thalamus is implicated in sensory integration, and functional imaging studies revealed increased activation in the thalamus in response to phobic-related and threat stimuli in patients with SAD,^24,25 blood-injection-injury phobia,²⁶ dental phobia,²⁷ spider phobia,²⁸ snake phobia,²⁹ and PTSD.³⁰ Contrary to these findings, one study reported that dental phobics did not demonstrate increased activation in thalamus to phobic images.²⁹ Activation in the thalamus was correlated with the degree of anxiety and disgust in blood-injury-injection phobia,²⁶ as well as autonomic arousal in snake phobia, but not in dental phobia.²⁹ Treatment with paroxetine reduced activation in thalamus compared with placebo treatment in response to a recorded performance task in SAD.³¹

Emotion generating/processing regions

Ample evidence from basic and human imaging studies suggests that regions such as the striatum,³² amygdala,^10,33,34 insula,^11,14,15 and dACC³⁵ play a large role in identifying fear stimuli and generating fear responses. These areas often have structural abnormalities and are hyperactive in anxiety compared to controls. Extensive prior evidence¹³ suggests hyperactivation in the amygdala across all anxiety disorders. Hyperactivation was also reported in insula in SP, SAD, PTSD, and GAD, whereas activation differences were less consistent in PD. Hyperactivation in dACC was reported in SP and PTSD, with mixed findings in PD and SAD and limited evidence in GAD.

Striatum

Less activation in ventral striatum has been reported in SAD while anticipating giving a speech, with greater levels of anticipatory anxiety predicting less activation.³⁶ Although striatal activation is typically modulated in response to social cooperation, this was not observed in SAD, suggesting abnormalities in reward circuitry related to the initiation and maintenance of social relationships.³⁷

Amygdala

Studies examining structural differences in the amygdala in anxiety patients compared with healthy controls find decreased amygdala volume and density in PD,^38–40 SP,⁴¹ and PTSD,⁴² with symptom severity predictive of smaller amygdala volume.⁴¹ In contrast, some studies reported larger amygdala volume in anxiety, specifically GAD⁴³ and PTSD.⁴⁴ The picture for SAD is more complex, with reports of reduced amygdala volume,⁴⁵ no differences in amygdala volume,⁴⁶ and larger amygdala volume⁴⁷ in SAD compared with controls.

Recent functional imaging studies report amygdala hyperactivation in response to threatening stimuli in PD,³⁹ SP,^28,48,49 SAD,^{24,36,50–52} and PTSD^53,54 compared with healthy controls. The degree of amygdala activation was positively correlated with symptom severity in SP⁴⁸ and SAD.⁵⁰ Moreover, treatment with medication and psychotherapy often results in decreased amygdala hyperactivation to threat from pre- to posttreatment in SP,⁴⁸ SAD,^55–58 and PTSD.⁵⁹ Patterns of amygdala activation in GAD are more complex, with studies reporting increased^60,61 or no difference⁶² in activation in GAD compared with healthy controls. Null and opposing findings suggesting no difference or decreased amygdala activation in anxiety compared with controls have also been reported in PD,^63,64 SP,⁴⁹ and SAD,⁶⁵ with some evidence that medication treatment (paroxetine) increases amygdala activation in SAD compared with healthy controls.³¹

Insula

Structural imaging results revealed conflicting reports of increased volume in SAD,⁶⁶ no difference in SP,⁶⁷ and decreased volume in SAD⁴⁶ and PTSD^68–70 compared with controls.

Recent functional imaging studies revealed insula hyperactivation to threat in patients with SP,^27–29,71 SAD,^{36,50,52,65,72,73} and PTSD^{30,54,74–77} compared with controls. The degree of insula hyperactivation was positively correlated with symptom severity in SAD.⁷² Both medication and psychotherapy have been shown to decrease insula hyperactivation from pre-to posttreatment in SAD.^31,58,78 In contrast, one study found that patients with dental phobia did not show hyperactivation in insula during exposure to phobic stimuli compared with healthy controls.²⁹

Dorsal anterior cingulate cortex

Literature suggests decreased dACC gray matter and white matter volume in PD,⁴⁰ SP,⁷⁹ and PTSD^80–82 compared with healthy controls. In contrast, some studies have reported increased volume in dACC in SAD⁶⁶ and GAD.⁴³

Functional imaging studies reported dACC hyperactivation to threat in SP,^26–29,71 SAD,^24,50,83 and PTSD.^30,84–86 Increased dACC activation was correlated with greater autonomic arousal²⁹ and subjective anxiety levels^26,71 in SP. Cognitive behavioral therapy was found to decrease ACC activation in SP,⁷¹ and pretreatment ACC activation predicted positive treatment response in patients with GAD.⁶¹ Despite relatively consistent findings of hyperactivation in dACC, some idiosyncrasies exist in the literature. In contrast to other types of phobia, patients with dental phobia²⁶ and blood-injection-injury phobia²⁹ did not exhibit increased ACC activation to phobic stimuli relative to healthy controls, suggesting dACC may not be involved in threat processing in some anxiety symptom presentations. Moreover, one study reported that dACC was less active to threat in SAD compared with healthy controls.⁸⁷

Emotion modulation regions

Regions involved in regulating threat responding are particularly important in anxiety, as they can decrease activation in threat-processing regions such as the amygdala, insula, and dACC. These have been identified using basic science models and human imaging studies. The mPFC, hippocampus, dorsolateral prefrontal cortex (dlPFC), and rACC have been implicated in modulating fear responding. Although the mPFC^10,33,34,88 and rACC⁸⁹ are primarily involved in modulating emotion, the dlPFC⁸⁹ has been implicated in both emotion modulation and attention control. The hippocampus is primarily involved in encoding contextual information and modulating fear responding within the context of threat and safety signals.¹¹ As such, these regions underlie different functions that may work in concert to modulate threat response. Evidence reported before the scope of the current review¹³ suggests hyperactivation in the hippocampus in PD and PTSD but little evidence for hippocampus involvement in other anxiety disorders. Hypoactivation in the mPFC has been reported in PTSD and GAD, with less consistent results seen in PD, SP, and SAD. Evidence for dlPFC and rACC is less consistent and less studied, with both hyper- and hypoactivation reported in PD, SP, SAD, and PTSD.

Medial prefrontal cortex

Structural imaging studies report differences in mPFC, with increased volume in SAD,⁹⁰ decreased volume in PTSD,^77,80 and no difference in GAD,⁹¹ with decreased volume associated with greater symptom severity in PTSD.^92,93

Functional imaging studies, including two recent meta-analyses and a literature review,^30,76,94 primarily report decreased mPFC activation in PTSD compared with healthy controls. Some other studies of PTSD patients reported increased mPFC activation in response to fearful faces.^74,75,95 Similar findings are reported in SAD, with both increased⁹⁶ and decreased⁹⁷ mPFC activation in response to threat and social tasks. Results are somewhat more consistent in GAD, with the majority reporting decreased mPFC activation.^62,98–100 It has been suggested that hyperactivation and hypoactivation in mPFC may be associated with different symptom profiles.¹⁰¹ It is also possible that although hypoactivation indicates a deficit in emotion regulation, hyperactivation indicates an overcompensatory response in an effort to decrease excessive fear responding. Regardless of these discrepancies, the treatment literature is quite consistent, suggesting that both pharmacotherapy and psychotherapy produce increases in mPFC activation in SAD^{57,58,102,103} and PTSD,⁵⁹ which is related to symptom improvements.

Hippocampus

Although the hippocampus is often considered part of the limbic system responsible for fear generation, the majority of evidence in rodent and human models examining hippocampal function suggests its primary role is in context learning and fear modulation in the presence of safety and threat contexts.¹¹ Structural imaging studies of the hippocampus suggest decreased volume and density in PD,^38,40 SAD,^45,104,90, GAD,¹⁰⁶ and PTSD.^42,69,80,105 There is also evidence for increased hippocampal volume after treatment with selective serotonin reuptake inhibitors in PTSD.¹⁰⁷ However, findings are somewhat mixed, with studies also reporting no differences⁴⁶ and larger^47,108 hippocampal volume in SAD compared with healthy controls.

Functional imaging studies report increased hippocampal activation to threat in SP,⁴⁹ SAD,²⁴ and PTSD^30,75,95 compared with healthy controls. Increased hippocampal activation was related to defensive reactivity in response to threat stimuli in SP.⁴⁹ Differences in hippocampus activation in PTSD during memory tasks^53,109,110 and emotional activation tasks⁵⁴ have been less consistent, with both increased and decreased activation reported.

Other modulatory regions

Evidence exists suggesting that other regions, including the dlPFC and subgenual/rACC, play a role in emotion modulation.⁸⁹ Structural imaging studies report increased dlPFC volume and thickness in SP,¹¹¹ with larger volume predicting more severe symptoms and arousal levels.

Functional imaging studies report dlPFC hyperactivation to threat in SP¹¹¹ compared with controls. In addition, dlPFC mediated the influence of SAD on laughter perception, which was related to symptom severity.¹¹² Hypoactivation in rACC has been reported in SAD,⁶⁵ PTSD,^94,113 and GAD.^62,100,114 However, treatment with computerized attention training⁵⁸ and paroxetine³¹ reduced activation in subgenual ACC in SAD.

Connectivity between regions

More recent studies have begun to focus on connectivity between brain regions in anxiety patients. Results overwhelmingly suggest decreased connectivity between emotion processing (amygdala, insula) and emotion modulation (mPFC, rACC) regions. This finding is consistent both during rest and while performing a variety of cognitive and emotional tasks and is often interpreted as a deficit in regulating fear responding. Studies examining the structural connectivity between medial-frontal and basal-limbic areas, including the amygdala, measured via volume of the uncinated fasciculus, revealed smaller volume in the left uncinated fasciculus in patients with SAD than in controls. This finding suggests communication deficits between emotion-generating and emotion-modulation regions in SAD.¹¹⁵

Functional connectivity analyses reveal less connectivity between amygdala and mPFC in SAD,¹¹⁶ PTSD,⁹⁴ and GAD.^117–120 There is also evidence for reduced connectivity between the amygdala and ventrolateral prefrontal cortex (vlPFC) in GAD,^60,121 and the amygdala and subgenual/rACC in SAD¹²² and GAD.^98,114,117 Connectivity improves after treatment for GAD^60,123 and SAD.¹¹⁶ Some conflicting findings do exist, however, with reports of increased connectivity between dorsomedial prefrontal cortex (dmPFC) and amygdala¹⁰³ and between mPFC and amygdala¹²⁴ in patients with SAD compared with controls.

Differences in connectivity within emotion-processing regions are also reported in anxiety compared with controls in a small number of studies, with the majority reporting decreased connectivity. Specifically, reports show decreased connectivity between insula and dACC in SAD⁷³ and between amygdala and insula in GAD.^120,125 However, increased connectivity between amygdala and insula was reported in PTSD.¹²⁶

More global differences in connectivity have been observed in SAD, with differences in gray matter volume across the whole brain¹²⁷ and deficits in global brain networks including the default mode network and the central-executive network compared with healthy controls.¹⁰⁴ Deficits in connectivity were also observed in and between regions involved in general arousal and attention.²⁵

Discussion

This review of recent literature suggests the presence of both common and distinct neural circuits involved in anxiety disorders. It partially supports the idea of a spectrum, with more cue-related and fear-based disorders (PD, SP) on one end and anxiety-based disorders (GAD, SAD, and PTSD) on the other. In general, all disorders involved deficits in both emotion-generating regions and modulatory regions, suggesting fear and anxiety both play key roles across the anxiety spectrum. What differentiates disorders appears to be the degree of dispersion of functional changes across the brain. Although PD and SP are characterized by deficits primarily in emotion-generating regions, SAD is characterized by deficits in a number of additional regions involved in sensory processing and attentional control, GAD findings are the least consistent in the emotion-generating circuit and most consistent in the frontal emotion regulatory circuit, while PTSD had a relatively consistent pattern in both circuits.

Common neurocircuitry

Changes in amygdala, ACC, and hippocampus are implicated across all anxiety disorders by multiple studies. Structural imaging studies report overall mixed results, with reports of both increased and decreased regional volumes in anxiety patients. Functional neuroimaging findings are more consistent, however, with the majority of studies reporting hyperactive amygdala response to threat across anxiety disorders. The hippocampus is reported as hyperactive across multiple anxiety disorders; however, what aspect of hippocampal function this hyperactivity reflects is less clear. Although hyperactivation in hippocampus seems to parallel hyperactivation observed in amygdala, the prototypic emotion-generating structure, basic rodent and human models of anxiety suggest the hippocampus is primarily a regulatory region. It is possible that the hippocampus is involved in multiple processes related to both fear generation and modulation, consistent with the view of “neural reuse,” arguing that brain regions can be used for multiple functions.¹²⁸

ACC activation varied as a function of subregions: dorsal, rostral, and subgenual. In general, it seems that across anxiety disorders, activation in dACC is increased in response to threat, whereas subgenual/rACC activation patterns are less consistent, although hypoactivation was the most common finding in SAD, PTSD, and GAD. Although differences in activation patterns may depend on the task and stimuli used, the different functions of ACC subregions and the connectivity of these subregions to other brain structures likely contribute to these findings. Specifically, although dACC appears to be involved in threat processing and fear generation, subgenual/rACC is primarily involved in emotion modulation.

Connectivity analysis suggests a common mechanism for relationships between PFC and limbic structures across anxiety disorders. Specifically, decreased connectivity between emotion-generating areas (amygdala, insula) and cortical regulatory regions (mPFC, rACC) have been consistently found across anxiety disorders. Connectivity was inversely related to symptom severity, and in a number of studies employing pre-post design, it increased after anxiety treatment. Overall, these findings point to a potential deficit in fear regulation circuitry in anxiety consisting of hyperactivation of emotion-generating regions coupled with dysfunction in emotion-regulation regions. This suggests the possibility of a shared network underlying the spectrum of anxiety and stress disorders.

Distinct neurocircuits

Activation patterns in amygdala and hippocampus were less consistent in GAD. In contrast, hypoactivation in the subgenual/rACC and mPFC are more consistent in this disorder. The primary clinical presentation of GAD is constant and unfocused worries about ambiguous and potentially negative future outcomes, rather than a focus on a specific threat. This cognitive component of GAD, including distorted beliefs and ruminations, may explain consistent deficits in emotion regulatory areas. In other words, GAD may be more “anxiety related” than “fear related,” where impaired function of the prefrontal cortical areas has a “permissive” role in increased anxiety. In SAD, the changes in neural activation patterns also appeared more distributed (beyond the set of limbic and prefrontal regions discussed earlier) compared with other anxiety disorders. SAD studies report increased activation in the fusiform face area and occipital cortex, indicating enhanced stimulus processing associated with anxiety. The increased hypersensitivity to social cues and self-reference typical in SAD could account for differences in these brain regions that may be less relevant in other anxiety disorders.

More consistent reports of hypoactivation in mPFC regions in disorders that are closer to the anxiety than to the fear end of the spectrum (ie, GAD, SAD, and PTSD) may relate to the cognitive processes and diffuse anxiety states associated with these disorders, leading to more anxiety-like cognitions and ruminations. Decreased activation in mPFC was observed consistently in PTSD and GAD, with mixed findings in SAD. In contrast, dlPFC hyperactivation to threat was observed in SP, and this region also appears to regulate responses to an ambiguous cue (laughter) in SAD. Hypoactivation in emotion regulation regions such as mPFC is typically interpreted as a deficit in ability to appropriately inhibit threat responding in anxiety. However, a number of studies also report hyperactivation in mPFC in anxiety. One possible explanation for hyperactivation is that these regions are overcompensating in an effort to down-regulate activity in amygdala and other emotion-generating regions. We may also be observing a time delay with respect to when these regulatory regions are coming online in anxious compared with healthy controls, resulting in activation differences. In other words, although the main changes observed in fear-related anxiety disorders such as SP or PD may reflect hyperreactivity of the emotion-generating system to threat/feared cues, distorted anxiety-related cognitions may play a larger role in GAD, SAD, and PTSD.

The existing literature examining structural and functional deficits in fear and anxiety circuitry is quickly growing, and just in the last 5 years, a considerable amount of knowledge regarding common and distinct patterns of neural function across the anxiety and stress disorder spectrum has been gained. However, much remains to be learned about specific differences in brain function across both emotion-generating and emotion-regulating regions. Future studies should focus on striving for consistency in paradigms used and patient populations studied to improve our ability to compare and contrast findings across experiments. It is also imperative to continue investigating connectivity between brain regions across the spectrum of anxiety and stress disorders to fully identify neural circuitry underlying symptom presentations, rather than simple activation patterns. It is clear that threat processing and emotion regulation circuits interact with one another, making more global and interactive analyses key in understanding these complex processes. Additional treatment studies are also indicated in an effort to better understand how pharmacologic and psychotherapeutic approaches can be used and modified to best target and correct aberrant brain function underlying anxiety.

Disclosure

The authors report no conflicts of interest in this work.