The overlapping neurobiology of induced and pathological anxiety: a meta-analysis of functional neural activation

Abstract

Objective

Anxiety can be an adaptive response to unpredictable threats, while pathological anxiety disorders occur when symptoms adversely impact daily life. Whether or not adaptive and pathological anxiety share mechanisms remains unknown, but if they do, induced (adaptive) anxiety could be used as an intermediate translational model of pathological anxiety to improve drug-development pipelines. Meta-analyses of functional neuroimaging studies of induced and pathological anxiety were therefore compared.

Methods

A systematic search was conducted in June 2019 on the PUBMED database for whole-brain functional magnetic resonance imaging articles. Eligible articles contrasted anxious patients to controls, or an unpredictable-threat condition to a safe condition in healthy participants. Five anxiety disorders were included: post-traumatic stress disorder, social anxiety disorder, generalized anxiety disorder, panic disorder, and specific phobia. 3433 records were identified, 181 met criteria and the largest subset of task type was emotional (N=138). Seed-based d-mapping software was used for all analyses.

Results

Induced anxiety (n=693 participants) and pathological anxiety (n=2554 patients and 2348 controls) both showed increased activation in the bilateral insula (xyz=44,14,-14 and xyz=-38,20,-8;k=2102 and k=1305 respectively) and cingulate cortex/medial prefrontal cortex (xyz=-12,-8,68;k=2217). When split by disorders, specific phobia appeared the most, and generalized anxiety disorder the least, similar to induced anxiety.

Conclusions

This meta-analysis indicates a consistent pattern of activation across induced and pathological anxiety, supporting the proposition that some neurobiological mechanisms overlap and that the former may be used as a model for the latter. Induced anxiety might, nevertheless, be a better model for some anxiety disorders than others.

Introduction

Anxiety disorders are the most prevalent mental health condition (1) with a lifetime prevalence of 17% (2), resulting in significant individual and social impairment (1) and a considerable overall burden of disease, ranked 9^th cause of years lived with disability in the world in 2015 (3). Response rates to existing treatments usually range between 40 and 60% (4) which leaves a large number of people with debilitating symptoms and/or high probability of relapse (5).

Development of new treatments for symptoms of anxiety has, however, stagnated for several decades (6), partly due to the difficulty of establishing robust translational links between models of fear and anxiety in rodents and clinical anxiety in humans. It has recently been argued, therefore, that models of anxiety [as defined by aversive anticipation and apprehension of perceived potential, but unpredictable, threats] in healthy humans, could help us bridge this gap and facilitate therapeutic progress (7).

More precisely, using the same techniques to induce anxiety in healthy individuals and animal models should enable us to both better understand the neurobiological basis of anxiety, as well as provide an intermediate route to screen the efficacy of candidate interventions prior to full clinical trial (8). This experimental approach is possible because, perhaps unique amongst psychiatric symptoms, anxiety is also an adaptive behavior with a benefit to survival. Anxiety enhances vigilance to threat and primes defense mechanisms (9) which allows the individual to react faster in dangerous situations. It occurs naturally in every individual, when walking down a dark alley at night for instance. This adaptive anxiety can be reliably induced in healthy individuals in the lab by exposing them to unpredictable threat of rare electrical shocks. This approach is a well-validated (10), reliable both for self-report and task performance (11) and, critically, it is also fully translational – a close paradigm is used in animal models (12). A growing body of literature shows that, in addition to clear increases in subjective and physiological reports of anxiety (13), threat of shock results in cognitive and psychophysiological changes mirroring pathological anxiety (14–16).

Induced and pathological anxiety therefore overlap at the level of symptoms, as both promote functions and states that promote harm avoidance. What remains insufficiently explored, however, is the extent to which underlying neurobiological mechanisms overlap, or whether ostensibly similar symptoms are driven by dissociable underlying mechanisms. Critically, for the ‘experimental psychopathology’ (7) approach to be valid then the assumption that induced anxiety evokes (at least some of) the same neurobiological mechanisms as pathological anxiety must be met, particularly on emotion-related paradigms where the literature suggests they lead to similar changes in cognitive performance (14).

Induced anxiety via unpredictable threat paradigms has been shown to involve brain regions involved in emotional processing, decision-making and reward circuitry, such as the amygdala, anterior cingulate cortex (ACC), medial prefrontal cortex (PFC), bed nucleus of the stria terminalis (BNST), insula or striatum (17–21) but this has not been systematically meta-analyzed. Meta-analyses and systematic reviews exploring the neural circuitry of different anxiety disorders suggest that many of the same regions have been implicated (22–25), however differences across disorders have not been investigated in the past decade (26,27). Meta-analyses of fear conditioning studies, a related experimental model that focuses on predictable (rather than unpredictable) threats, also reported resulting hyperactivation in dorsal ACC and bilateral anterior insula (28,29) and one meta-analysis that came out whilst this paper was in submission investigated shared neural correlates across mood and anxiety disorders (30). What is lacking, however, is an up-to-date systematic meta-analysis directly assessing the extent to which the neural activation in anxiety disorders in emotion-related paradigms overlaps with that evoked by unpredictable threat-induced anxiety (as opposed to fear conditioning) in the general population. That is, in other words, a systematic assessment of the neurobiological links between induced and pathological anxiety; and a quantitative assessment of the ‘experimental psychopathology’ approach to anxiety.

This meta-analysis therefore aims to 1) investigate the common functional neural activity pattern across induced anxiety studies, 2) update disorder-specific maps for 5 anxiety disorders and examine commonalities across all pathological anxiety brain activity in emotion-related paradigms, 3) compare neural patterns of induced anxiety to pathological disorder(s). We used a coordinate-based meta-analytic (CBMA) whole-brain approach (31) to test the broad prediction that activation patterns overlap across induced and pathological anxiety. The CBMA approach has important strengths over conventional activation likelihood meta-analyses as it uses the effect sizes and enables investigation of voxelwise publication bias (32). Unthresholded groups maps were also collected where possible to ensure that the results were as precise as possible. In addition, we explored overlap of our induced anxiety results with a recent fear-conditioning meta-analysis.

Methods

Literature search and inclusion

A systematic search was conducted on the PUBMED database (all studies published before 11^th of June 2019, including studies in press) for fMRI whole-brain BOLD activity papers reporting anxious or depressed patients vs. controls contrasts, or an unpredictable threat vs. safe conditions contrasts (flowchart in Fig. S1, see Supplementary methods for full details).

181 publications were identified comprising 2911 anxious patients and 2685 controls. To improve consistency across the paradigms used for the patient vs. controls contrasts, articles were then split into broad tasks categories: emotion (exposure to phobic [e.g., spider images], traumatic [e.g., combat films], socioemotional [e.g., faces] or general strongly aversive stimuli [e.g., loud noises]), attention (sensory detection and focus, Go/NoGo), decision (strategic planning and calculus, monetary decision-making) and memory (working memory encoding and retrieval, learning tasks). The main patients vs. controls analyses were focused on the emotion category (138 articles) which includes 2554 patients and 2348 controls (of which 27 were a depressed (but not anxious) control sample and the rest healthy) because it was the largest paradigm subset. 693 participants undergoing induction of anxiety were included. Post-traumatic stress disorder (PTSD) articles using traumatized controls (22 articles, 325 patients, 353 controls) were not included in the main analysis. Unthresholded maps for 17 of the 138 included articles were obtained (one included induced anxiety study provided unthresholded maps but did not report coordinates (15 subjects)). See Tables S1 for a full description of the samples and included articles.

SDM meta-analysis procedure

Activation and deactivation coordinates, as well as the t-threshold and t-values, were collected from each article for the contrast of interest and entered into the SDM-PSI (32) software (version 6.12).

See Supplementary methods for full details, but briefly, for each article group, coordinate-based maps were reconstituted and preprocessed with default parameters (FWHM 20mm, gray matter mask). This led to the following analyses: 1) a meta-analysis of all induced anxiety articles, 2) a meta-analysis of all pathological anxiety articles using emotional tasks, 3) convergence analysis of induced anxiety vs. pathological anxiety, 4) separate meta-analyses of the PTSD, social anxiety disorder, generalized anxiety disorder, panic disorder and specific phobia diagnostic groups of pathological anxiety, and 5) separate convergence analyses of induced anxiety vs. each of the 5 main diagnostic groups.

Publication bias was assessed for each cluster with the Egger’s test implemented in SDM using, for each cluster, the mean effect size from each study. See also Supplementary materials for exploratory analyses of the non-emotion tasks in pathological anxiety and of PTSD patients with traumatized controls.

An exploratory similarity analysis between our induced anxiety results and a recent fear-conditioning meta-analysis (28) was also conducted via the Neurovault comparison tool in the similarity search (chosen map: CS+ vs. CS-, pseudo Z scores). Regional correlations were calculated from a brain-masked, 4mm transformation of the original images.

Results

Findings from 138 papers and 5595 participants are presented here. See Table S2 in Supplement for full list of included papers. All collected coordinates and t-value files are available online at https://osf.io/9s32h/. All unthresholded whole-brain activation and convergence maps reported below are available online at https://neurovault.org/collections/6012/.

Pathological anxiety-associated brain activity

Anxious patients across disorders (N=2554) vs controls (N=2348) demonstrated increased activation bilaterally in a cluster encompassing the middle and superior temporal gyri, insula and inferior frontal gyrus (IFG), the left part extending to the amygdala, parahippocampal gyrus, hippocampus, bilateral lingual and fusiform gyri and thalamus (z=5.413 and z=6.156 for left and right clusters respectively). Increased activation was also found in the anterior and mid-cingulate and superior medial frontal gyrus (z=3.951). Other clusters of increased activation include left middle occipital, left postcentral gyrus, bilateral caudate, bilateral calcarine fissure, bilateral precuneus, right supramarginal, bilateral superior parietal and superior occipital gyri, right parahippocampal gyrus, left middle frontal gyrus, and supplemental motor area. No clusters of reduced activation were significant. No significant publication bias was revealed by the Egger’s test for any peak, including the left (bias=0.34,p=0.519) and right (bias =0.46,p=0.355) superior temporal gyrus (STG)/insula/IFG clusters and the cingulate/medial frontal clusters (bias=0.25, bias=0.14 and bias = 0.05, p=0.666, p=0.780 and p=0.927 respectively) (see Figure 1A and Table 1 for full information). Upon specific examination, bilateral peri-acqueductal grey (PAG) increased activation was also found.

Figure 1. Functional activation and convergence for induced and pathological anxiety.

A- Brain regions differing significantly between threat vs. safe conditions in induced anxiety studies (693 participants). B- Brain regions differing significantly between 2554 anxious patients vs. 2348 controls across pathological anxiety studies. SDM Z-value of activation in red-yellow gradient, deactivation in blue-green. C- Convergence of brain regions between induced vs. pathological anxiety. Converging activation in purple.

Table 1. Whole-brain meta-analysis of induced anxiety articles in threat vs. safe conditions, and of pathological anxiety articles across disorders.

p ≤ 0.005, k ≥ 10. Exploratory Egger’s tests are reported for meta-analytic clusters. Egger’s test not applicable to convergences, as those are pairwise comparisons of meta-analyses. ACC: anterior cingulate cortex; MCC: midcingulate cortex; MidFG: middle frontal gyrus; MedFG: Medial frontal gyrus; Orb. MedFG: orbital medial frontal gyrus; SMA: Supplementary motor area; STG: superior temporal gyrus; MTG: Middle temporal gyrus; ITG: inferior temporal gyrus; SPG: superior parietal gyrus; IPG: inferior parietal gyrus; SOG: superior occipital gyrus; MOG: middle occipital gyrus; IOG: inferior occipital gyrus

MNI coordinates	Voxels	Z value	Description	Egger’s bias	Egger’s p value
Pathological anxiety
-36,6,-14	12836	5.413	L. Insula, IFG all, Putamen, Pallidum, Rolandic operculum, Precentral g., Postcentral g., MidFG, Heschl’s g., STG pole, STG, MTG pole, MTG, Amygdala, Hippocampus, bilateral Parahippocampal g., bilateral Vermis 3-7, bilateral Cerebellum 3-6, Cerebellum 8, bilateral Lingual g., bilateral Fusiform g., bilateral thalamus	0.34	0.519
48,4,-14	5701	6.159	R. Insula, IFG all, STG pole, STG, MTG pole, MTG, Heschl’s g., Rolandic operculum, ITG	0.46	0.355
-10,-2,68	1203	3.951	Bilateral MCC, SMA, MedFG, ACC, paracentral lobule	0.25	0.666
-36,-74,30	565	3.589	L. MOG, IPG, SOG, SPG	-0.17	0.722
4,-90,12	333	3.362	Centre Calcarine, bilateral Cuneus	0.39	0.397
-26,22,40	280	3.149	L. MidFG, SFG	0.22	0.698
-16,12,14	202	2.975	L. Caudate, Thalamus	0.09	0.863
18,-56,50	180	2.667	R. SPG, Precuneus, IPG	0.09	0.859
20,-76,34	130	2.992	R. SOG, Cuneus	0.11	0.828
22,-16,-26	112	3.146	R. Parahippocampal g., Hippocampus	0.20	0.685
-14,-58,58	64	3.034	L. Precuneus, SPG	0.26	0.611
8,-66,18	65	2.574	R. Calcarine, Cuneus	0.01	0.992
14,10,18	41	2.536	R. Caudate	0.17	0.737
-44,-20,56	37	2.557	L. Postcentral g.	0.14	0.790
34,12,44	30	2.279	R. MidFG	0.03	0.948
58,-32,40	22	2.308	R Supramarginal g.	0.08	0.875
10,-2,66	21	2.433	R. SMA	0.05	0.916
12,-28,40	14	2.201	R. MCC	0.14	0.780
-8,6,38	10	2.114	L. MCC	0.05	0.927
Induced anxiety
4,38,38	6538	6.415	ACC, MCC, Sup. MedFG	1.48	0.110
50,22,2	4537	5.183	R. Insula, IFG all, Rolandic operculum, STG pole	1.91	0.148
-30,18,-14	1811	5.067	L. Insula, IFG all, Putamen, Rolandic operculum	1.30	0.199
60,-46,36	1373	4.458	R. Supramarginal g., STG, Angular g.	0.78	0.553
46,2,48	336	3.590	R. Precentral g., MidFG	1.53	0.161
-56,-44,28	303	3.354	L. Supramarginal	1.28	0.199
35,52,18	153	2.500	R. MidFG, SFG	2.65	0.222
-10,-34,-48	36	3.019	Possible cerebellum 9,10	0.30	0.778
-48,-62,-6	1251	-4.117	L. ITG, MTG, IOG	-0.32	0.752
-56,-22,46	636	-4.492	L. Postcentral g., IPG	-0.38	0.726
54,-60,-10	559	-4.102	R. ITG, IOG, MTG, MOG	-0.14	0.891
-8,52,-22	478	-3.267	L. Orb. MedFG	-0.11	0.932
42,-44,-16	188	-3.391	R. Fusiform g.	-0.25	0.826
-12,-64,12	185	-3.132	L. Calcarine, Lingual g.	-0.12	0.912
-20,-46,0	121	-3.063	L. Lingual g. Fusiform g.	-0.32	0.769
26,-62,-8,	47	-2.525	R. Fusiform g., Lingual g.	-0.34	0.744
-20,-70,-8	44	-2.697	L. Fusiform g., Lingual g.	-0.38	0.728
-32,-26,-20	42	-2.675	L. Fusiform g., Parahippocampal g.	-0.42	0.695
12,-70,16	41	-2.691	R. Calcarine	-0.09	0.940
-22,-16,-22	37	-3.052	L. Parahippocampal g., Hippocampus	-0.62	0.608
28,-24,-24	13	-2.448	R. Parahippocampal g., Fusiform g.	0.08	0.945
18,-78,14	12	-2.119	R. Calcarine	-0.21	0.842
Convergence
-12,-8,68	2217		SMA, MCC, MedFG sup., ACC
44,14,-14	2102		R. Insula, IFG all, STG pole, Rolandic operculum, STG, Putamen
-38,20,-8	1305		L. Insula, IFG all, Putamen, Rolandic operculum, STG pole
-4,-22,-10	615		R. Thalamus, Vermis 3
50,2,44	183		R. Precentral g., MidFG
12,-26,40	43		R. MCC

Diagnostic group analyses

Breaking down analyses into diagnostic groups (Figure 2A), specific phobia (414 patients) shows the three increased activation clusters in cingulate and bilateral IFG/insula. Panic disorder (263 patients) and PTSD (436 patients) also show more activation in bilateral insula/STG but no activation or deactivation in the mid- and anterior cingulate cortex. Social anxiety disorder (805 patients) shows activation in the right insula/IFG/STG and left amygdala but no activation or deactivation in the cingulate as well. In contrast to the other disorders, generalized anxiety disorder (233 patients) shows deactivation in the cingulate cortex and in bilateral insula. See Table S3 Supplement for full disorder-specific peak information. No significant publication bias was revealed by the Egger’s test for any peak. PAG increased activation was found bilaterally in specific phobia and in the left hemisphere in panic disorder.

Figure 2. Functional activation and convergence with induced anxiety for each anxiety disorder.

A- Brain regions differing significantly between anxious patients vs. controls for each anxiety disorder (Post-traumatic stress disorder - PTSD: 436 patients vs. 411 controls; Social anxiety disorder - SAD: 805 vs. 741; Panic disorder - PD: 263 vs. 268; Generalized anxiety disorder - GAD: 233 vs. 218; Specific phobia - SpP: 414 vs. 373). SDM Z-value of activation in red-yellow gradient, deactivation in blue-green. B- Convergence of brain regions between induced anxiety and each anxiety disorder. Converging activation in purple.

Induced anxiety-associated brain activity

Across participants (N=693), induced anxiety in threat vs. safe conditions demonstrated greater activation in the cingulate and medial frontal cortices (z = 6.415), and bilaterally in the inferior frontal gyrus (IFG)/anterior insula/Rolandic operculum (z = 5.183, z = 5.067 for right and left clusters respectively). Other areas of increased activation include bilateral supramarginal, right superior temporal (STG), right middle frontal, and right precentral gyri. Reduced activation was found in bilateral parahippocampal gyrus, fusiform and lingual gyri, as well as in bilateral calcarine fissure, bilateral inferior temporal, middle temporal, inferior occipital bilateral, left postcentral and left orbital medial frontal gyri. The Egger’s test for publication bias was not significant for any clusters, including the cingulate/medial frontal (bias=1.5, p=0.11) the left (bias=1.3, p=0.20) and right (bias=1.91, p=0.15) IFG/anterior insula clusters (See Figure 1B, Table 1 for full information). Upon specific examination, BNST and PAG increased activation was also found. Restricting the analysis to induction of anxiety via threat-of-shock did not affect primary outcome (see Table S4 for details).

Comparison between induced and pathological anxiety

When comparing all pathological anxiety with induced anxiety (Figure 1C) we see convergence for increased activation in bilateral insula/IFG and in ACC/mid-cingulate cortex (MCC)/superior medial frontal cortex. These clusters were also present, both for activity and convergence, in the complementary FWHM 10mm analysis. Convergence was also found in bilateral PAG activation. Excluding articles reporting any medicated patients or articles using a youth patient sample did not affect primary outcomes (see Table S5).

Diagnostic group analyses

When compared with induced anxiety (Figure 2B), specific phobia shows convergence for increased activation in cingulate/medial prefrontal and in bilateral insula/IFG/putamen/STG pole. Panic disorder shows convergence for bilateral insula/IFG hyperactivation whereas PTSD is only convergent with induced anxiety for increased activation in insula/IFG opercular, but not for IFG triangular or orbital. Social anxiety disorder converges in the right insula/IFG orbital and triangular. Generalized anxiety disorder shows very limited overlap with induced anxiety. These findings are illustrated in Figure 2B. See Table S6 in Supplement for full pairwise convergence peaks. All clusters mentioned above were also present in the FWHM 10mm analysis for convergence with induced anxiety, although the left insula contribution to activity in panic disorder was absent. Specific phobia also converged with induced anxiety for bilateral BNST and PAG activation.

Overlap of induced anxiety with fear-conditioning

Induced anxiety showed a whole-brain Pearson correlation coefficient of r=0.66 with a fear-conditioning meta-analysis. Regional correlations were r=0.76 for the putamen, r=0.75 for the insula, r=0.73 for the frontal lobe, r=0.65 for the parietal lobe, r=0.57 for the caudate and r=0.54 for the thalamus.

Discussion

Consistent with the hypothesis that induced anxiety may be an experimental psychopathological model of anxiety disorders, induced and pathological anxiety show overlapping neurobiological activations. Specifically, induced anxiety and pathological anxiety both converged in increased activation in the cingulate cortex/medial PFC, bilateral insula/IFG and PAG. However, there were also some important dissociations, especially when pathological anxiety was broken down into component disorders, perhaps suggesting that induced anxiety overall might be a closest model for specific phobia and furthest from generalized anxiety disorder.

Induced anxiety as a model for pathological anxiety

The first thing to note is that induced anxiety evokes ACC, MCC, medial PFC and insula activation, as well as activation in the BNST and PAG. The insula and cingulate regions have been argued to form part of a ‘fear-conditioning’ circuitry (28) and/or a ‘salience’ network (33) which drives interoception in particular (34). In fact, Neurovault similarity analysis reveals that our induced anxiety map shows reasonably high (r~0.7) correlation with a recent meta-analysis investigating Pavlovian fear-conditioning neural correlates (28). The overlapping regions perhaps therefore reflect a shared circuitry that responds to the threats common to anxiety- and fear-conditioning, with the non-overlapping circuits perhaps being specific to the spatial/temporal predictability of these threats. MCC electroencephalographic activity has also been reported to play a key role in adapting behaviour to uncertainty and to be modulated by anxiety (35). It is therefore possible that these regions contribute to circuitry which (in the case of the cingulate) detect salient environmental stimuli and then promote behavioural avoidance of threats (via connections to motor cortex), or (in the case of the insula) detect salient internal change that require some kind of homeostatic response (e.g. heart rate increases). The overall effect being to reduce the negative impact of potential harms (perhaps in concert as part of a putative ‘salience’ network).

Critically, the same insula and cingulate activations are seen across pooled anxiety disorders in our data, as well as in older meta-analyses (26,36–38). They may therefore play the same role in pathological anxiety disorders – promoting avoidance responses to salient negative stimuli. Indeed, insula and midcingulate response is thought to be a promising predictor of psychotherapy response (39), which suggests that this circuitry is also important for clinical response (which is largely defined as a reduced avoidance/response to threats).

Thus, induced anxiety holds promise as an intermediate translational model of anxiety disorders (7). In other words, promising new candidate medications might be shown to first modify the effects of threat of shock in animal models (i.e., subjecting animal to unpredictable shocks) (12), and then the effects of threat of shock in healthy humans, before being rolled out in a full-scale clinical trial in anxiety disorders. This would provide greater confidence that the candidate medication targets relevant symptoms and mechanisms and therefore improve the (currently very poor) hit rate of psychiatric drug development (4,5). This is important because it has been suggested that fear and anxiety in humans can and should both be conceptually segregated across two systems with separate but interacting circuitry: the behavioral and physiological response on the one hand, and the conscious feeling and state on the other hand (40). Conceptually, induction of anxiety via unpredictable threat spans both systems in humans: a conscious but diffuse feeling of anxiety as well as avoidance and physiological defensive arousal. The overlapping activity we observe may therefore be involved in both of these facets, but future work (ideally with identical cognitive tasks across both induced and pathological anxiety) is needed to truly disentangle these important distinctions. At the same time, if very similar manipulations can be used in animal models, it (to a certain extent) circumvents the problem that it is not possible to measure subjective feeling/states in animal models. In other words, consistent translational manipulations provide a more direct bridge from animal models to human clinical work as well as a means of eventually reconciling disparate anxiety-related systems (40).

Specificity across disorders

However, it is important to recognize that although similarities were found with induced anxiety when all the pathological anxiety studies were pooled together, some differences became apparent when studies were split by disorder. These results may be confounded by biases in sample sizes and/or cognitive tasks (see limitations) thus we must refrain from excessive interpretation, but specific phobia was revealed to be most similar to induced anxiety, showing significant increased activation convergence in the cingulate cortex/medial PFC and bilateral insula/IFG as well as BNST and PAG. PTSD and panic disorder converged with induced anxiety for increased activation in bilateral insula but not for cingulate hyperactivation. Finally, both social anxiety disorder and generalized anxiety disorder had a more complex pattern, the former only converging in the right insula and the latter actually failing to show convergence for bilateral insula and cingulate hyperactivation. As such, it may also be that induced anxiety is a better model for some sub-types of pathological anxiety than others.

Overall, the disorder-specific findings might indicate that pathological anxiety mechanisms are diverse and that we should not always assume similarities across disorders. Indeed, with the detection power allowed by the current literature, induction of anxiety – mainly by threat of unpredictable shock – appears to be a very good model for specific phobia, relatively good for panic disorder, PTSD and possibly social anxiety disorder, and less relevant for generalized anxiety disorder at the functional activation level. However, future direct comparison with the exact same tasks and the same power in all groups is needed to be confident in this prediction.

Limitations

To our best knowledge, this is the first meta-analysis to investigate functional activity in anxiety induced via unpredictable threat paradigms and compare it with up-to-date meta-analyzed functional activity of pathological anxiety disorders. However, it is important to recognize several limitations.

Firstly, in order to collect sufficient samples in each group of eligible articles, we did not exclude any articles based on male/female ratio, age, as well as potential medication, individual comorbidities and clinical severity – all of which could potentially confound the functional correlates of anxiety in patients or in healthy controls.

Secondly, splitting by anxiety type and disorders (and restricting to emotional tasks) resulted in varied group sizes, leading to differences in power between diagnostic-specific meta-analyses as well as an overabundance of some task types in some disorders (e.g. symptom provocation in specific phobia). Similarly, at a more global level, the tasks within the induced anxiety sample are more consistent than the diverse emotion tasks in the pathological anxiety sample. Ultimately, this likely makes all interpretation of the differences between groups less solid than the observed common/shared effects.

Thirdly, although we did restrict analyses to broad task categories, we did not filter our systematic analysis by precise task, because we wanted to examine as many aspects of anxiety as the body of literature allowed. As a result, the tasks used in eligible articles are somewhat diverse. For example, the criteria of our broad emotion task category were: exposure to threatening or strongly aversive (electrical shocks, loud noises, etc), phobic (images of spider, sounds of dental care, etc), traumatic (combat-related movies, etc) or socioemotional (faces or words with or without emotion, etc) stimuli. Since all induced anxiety papers included strongly aversive threat, they all qualified by definition into this category. Coincidentally, all the eligible specific phobia papers used phobic stimuli in their specific tasks and as a result also qualified for the emotion category. Hence, one explanation for the consistency between specific phobia and induced anxiety may be that the symptom provocation tasks used (i.e. those designed to provoke symptoms using shocks or phobic stimuli) were most similar across these studies. In a broader sense, our inclusive approach where we pool articles across different diagnoses and paradigms will inevitably lead to biases that limit inference. In an ideal world, we would restrict analyses to single tasks, but this would severely limit our detection power. As it stands, our inference is perhaps stronger for the conjunction analyses (where we are seeing similarities in spite of confounds) rather than our difference analyses (where discrepancies might simply be driven by confounds).

Fourthly, it is worth noting that we restricted our meta-analysis to articles reporting a whole-brain analysis. Unfortunately, for a small number of articles, it was unclear as to whether the analysis was carried out with homogenous thresholding across the whole brain or whether some regions of interest were singled out (again, we would recommend increased clarity in reporting of future research). Notably, some key structures, the amygdala in particular, as well as the BNST and PAG, often do not emerge in whole-brain analyses, most of which have a comparatively large cluster threshold. Thus, our results for the most part did not reflect amygdala activation or deactivation, which are often reported in region-of-interest analyses only.

Conclusion

This meta-analysis demonstrates that induced anxiety evokes activation of cingulate and insular regions in common with pathological anxiety, which (at least partially) validates the former as an intermediate translational model of the latter. Nevertheless, our findings also indicate functional differences between anxiety disorders, suggesting that induced anxiety might be a better model for some disorders than others.