Neural networks underlying adolescent anxiety
Anxiety disorders are not due to deficits of a single brain structure. Instead, there are plenty of studies suggesting anxiety-related neural networks. This section reviews previous findings that link deficits in neural networks to anxiety, especially in adolescents.
Anxiety disorders have abnormalities in many aspects of psychological processes, such as cognitive control, fear conditioning, uncertainty anticipation, motivation bias and stress regulation. Regarding cognitive control, anxiety disorders are characterised by dysfunctional cognitive control of the projection from the prefrontal cortex to the amygdala.13 Patients with anxiety disorders show alterations in the fear condition, including abnormalities in the network involving the ventral hippocampus, basolateral amygdala and mPFC.22 Regarding uncertainty anticipation, neural pathways involving the BNST mediate the over-reaction to uncertain anticipations in patients with anxiety disorders.33 As for motivation, the striatum is associated with unbalanced reward function in anxiety disorders, and the failed regulation of the striatum, amygdala and prefrontal cortex is an important neural underpinning of anxiety disorders.34 Regarding stress, abnormal recruitment of the hypothalamus, amygdala and prefrontal cortex is associated with the failure of stress regulation. Based on the mentioned associations, we assume that psychopathological symptoms of anxiety disorders in adolescence may be underlain with abnormalities in brain structures relevant to cognitive control, fear conditioning, uncertainty anticipation, motivational processing and stress regulation.
Figure 1 summarises previous findings regarding the deficits of neural networks underlying anxiety disorders, characterised by the five components. We will further review the findings relevant to these components in the following subsections.
Figure 1Schematic presentation of anxiety-related functional systems and brain structures. The arrows indicate the neural projections between two brain regions. The pathological interactions of the six critical regions in the anxiety network bring about functional abnormality for patients with anxiety disorders. Amyg, Amygdala; BNST, Bed nucleus of the stria terminalis; Hipp, Hippocampus; Hyp, Hypothalamus; PFC, Prefrontal cortex; Str, Striatum.
Cognitive control: amygdala and mPFC
The neural circuit between the mPFC and the amygdala is closely related to cognitive control. The amygdala projects to different areas of the mPFC; specifically, basal amygdala neurons project to the prelimbic and the infralimbic subdivision of the mPFC, completing the fear expression and extinction.35 The activation of this pathway corresponds to an increase in individual anxiety-like behaviours and a decrease in social interaction.36 The mPFC integrates inputs from various areas and projects them back to the amygdala to achieve top-down inhibitory control. The number of basal amygdala neurons projecting to the mPFC tends to become stable during adolescence.4 In adolescence, the maturation of the mPFC is later than that of the amygdala, and the ability of top-down regulation is immature and weak.37 Hence, insufficient inhibition of the amygdala neurons may provide a potential neural basis for adolescent anxiety disorders.
Fear conditioning: ventral hippocampus, basolateral amygdala and mPFC
Deficiency in fear extinction is one of the characteristics of anxiety disorders. The ventral hippocampus, basolateral amygdala and the mPFC form an interconnected circuit that plays a vital role in fear learning and extinction.38 39 The mPFC receives outputs from the ventral hippocampus. Glutamatergic inputs from the ventral hippocampus enhance the plasticity of mPFC neurons and promote maturation.40 When the ventral hippocampus projects to the mPFC, neurons synchronise to the theta frequency (4 to 12 Hz), forming and maintaining anxiety-like behaviours.39 41 In contrast, inhibiting the projections from the ventral hippocampus to the mPFC reduces the synchrony of the theta frequency and reduces the probability of anxiety disorders.39
The ventral hippocampus and the basolateral amygdala rely on each other coding fear-related memories.42 Glutamatergic inputs from the basolateral amygdala to the ventral hippocampus pyramidal neurons increase individual anxiety, while inhibition of this projection reduces anxiety-related behaviours.43 The pathway projected from the basolateral amygdala to the ventral hippocampus regulates social interactions as well.42 43 The dual function of this pathway in modulating anxiety and social behaviours possibly explains the high comorbidity rate of anxiety disorders and autism spectrum disorders.44 The projections from the ventromedial prefrontal cortex to the amygdala inhibit fear expression, and this process is modulated by the hippocampus.45
Uncertainty anticipation: BNST, amygdala and prefrontal cortex
The neural networks that include the BNST, amygdala and prefrontal cortex participate in the anticipation of uncertain threats.33 46 47 There is a functional separation between the basolateral amygdala and the BNST. The basolateral amygdala mediates immediate responses to threats and is related to panic disorder and specific phobia in anxiety disorders. The BNST, in contrast, mediates sustained responses to unpredictable threat information and is related to generalised anxiety disorder and post-traumatic stress disorder.48 The excessive and persistent anxiety response of patients with anxiety disorders to uncertain information may be underlain with neural projections from the basolateral amygdala to the BNST.33 The BNST receives the inputs from the mPFC.7 The nerve connection between the mPFC and the BNST has a modulating effect on anxiety.49
Motivation processing: striatum, amygdala and prefrontal cortex
The nucleus accumbens of the striatum evaluates stimuli in terms of motivational values, and the dorsal striatum integrates and transmits information to the prefrontal and motor cortices.50 Reward representations are available to the ventral striatum that participates in forming motivational and goal-oriented behaviours. The amygdala actively regulates the striatum. The direct projection from the amygdala to the striatum supports ‘fight or flight’ motor responses, as well as avoidance learning.33 In addition, the ventral striatum participates in emotion and motivation processing, driving action outputs from the basal ganglia.51 The projection from the prefrontal cortex to the striatum contributes to cognitive functions, such as decision-making.52 The mPFC achieves a balance between anxiety-like and motivated behaviours by the inputs to the amygdala and the striatum.53
Stress regulation: hypothalamus, amygdala and prefrontal cortex
The hypothalamus is sensitive to stressors and plays a regulatory role. Under stressful conditions, amygdala-to-hypothalamic outputs maintain anxiety behaviours.54 The neurotransmissions from the hypothalamus to the amygdala build fear expression and generalisation, and the deactivation of hypothalamic orexin neurons alleviates excessive fear.55 The direct projection from the prefrontal cortex to the hypothalamus involves regulating emotional stress, emotional arousal and control of aggressive behaviours.56 In anxiety-inducing situations, the prefrontal cortex inputs dopamine to the amygdala, which further excites the hypothalamic neurons and initiates the HPA axis, triggering a somatic response of sympathetic excitation.57 In patients with anxiety disorders, overactivation of the amygdala leads to overexpression of corticotropin-releasing factors, promoting HPA axis hyperactivity and manifesting as an over-reaction to stress.58
Differences in brain networks across subtypes of anxiety disorders
Morphologically, patients with social anxiety disorder (SAD) have a larger grey matter volume in the dorsal striatum.59 They have reduced frontal lobe volume and increased amygdala volume relative to healthy controls.60 61 Patients with generalised anxiety disorder (GAD) have reduced ventromedial prefrontal cortex volume and hypothalamus volume.62–64 Patients with panic disorder (PD) have smaller grey matter volumes in the amygdala, the hippocampus, the prefrontal cortex and the bilateral striatum.65 66
During emotion-related tasks, the amygdala and the parahippocampal gyrus are overactive in patients with SAD, with enhanced functional connectivity between the amygdala and the prefrontal cortex.67 68 In contrast, patients with GAD have insufficient activation in the prefrontal cortex and weak functional connectivity between the amygdala and the prefrontal cortex.69 In gambling games, BNST activity is increased while amygdala activity is suppressed in patients with GAD, where anxiety experience is triggered by high uncertainty.70 In fear conditioning, patients with PD have hyperactivation in the hypothalamus and abnormal recruitments of the hippocampal-prefronto-amygdala network with hippocampal excitation, manifesting as excessive fear learning.67 71
Brain network differences exist in different subtypes of anxiety disorders and are reflected differentially in various brain structures. Notably, current treatments for anxiety disorders also have different efficacy for different subtypes; for example, cognitive-behavioural therapy to treat patients with SAD tends to be less effective for other anxiety disorders.61 Possible neural underpinnings are that the aberrance patterns of the anxiety network are different for different subtypes of anxiety disorders: hypersensitivity to emotional stimuli in patients with SAD, inadequate top-down control in patients with GAD and excessive fear learning in patients with PD. With future verifications of this hypothesis, changes in the anxiety network may help to predict the efficacy of treatments.72