Significant advances in our understanding of the mediating psychobiology and the development of effective treatments for anxiety disorders have been made in recent years. Modern brain imaging techniques have proved useful in exposing specific albeit overlapping neurocircuitry that underlies individual anxiety disorders 12. However, relatively little work has focused on the extent to which the anxiety disorders overlap with respect to changes in brain perfusion that accompany response to first-line treatment that is after all pharmacologically similar for different disorders.

The selective serotonin reuptake inhibitors (SSRIs) are currently recommended as first line medications for most anxiety disorders, including obsessive-compulsive disorder (OCD) 3, posttraumatic stress disorder 4 and social anxiety disorder 5. A number of imaging studies have now examined the effects of SSRI's on brain perfusion in individual anxiety disorders. In OCD, attenuation of pre-treatment regional activation has been shown to correlate with treatment response in the anterolateral orbitofrontal cortex (OFC), caudate nucleus, thalamus, and temporal regions 67891011. Results for studies assessing pre-treatment cerebral perfusion as a predictor of response, have, however, yielded mixed results. In some, an inverse relationship appears to exist with pre-treatment regional activation of the OFC 12, anterior cingulate, caudate 6 and subsequent responses to treatment. Conversely findings of higher prefrontal, cingulate and basal ganglia activation correlating with subsequent treatment response have also been reported 1314. In OCD co-morbid with depression, substrates of response to the SSRI, paroxetine, appear to differ based on pretreatment 15 activation patterns as well as changes that accompany treatment response when an SSRI is given in identical doses for either of the two conditions separately 16.

In social anxiety disorder SSRI treatment response accompanies attenuation of frontal, anterior and lateral temporal cortex, cingulate, and thalamic activity 1718. Higher anterior and lateral temporal cortical perfusion at baseline correlated with subsequent treatment response in the former study. The latter study also demonstrated some overlap of regions demonstrating attenuation of activity for both cognitive and pharmacotherapy interventions.

In PTSD, a single study by our group demonstrated medial temporal lobe deactivation with treatment irrespective of clinical response and medial prefrontal cortex activation correlated with treatment response. In addition, no baseline differences distinguished responders and non-responders to subsequent SSRI treatment 19.

In this present study, we hypothesised firstly, that response to SSRI treatment in this combined group of subjects with anxiety disorders (OCD, PTSD, SAD) would effect shared changes in rCBF affecting primarily limbic and related prefrontal regions and thus suggest some overlap between disorders in the mechanism of their response to effective treatment with SSRI's. Secondly, pre-treatment differences in regional perfusion would likely differentiate responders to subsequent treatment with citalopram across the anxiety disorders.

Twenty-two males and fifteen females with a mean age of 33.5 years (SD 9.8) completed the study. Clinical changes with pharmacotherapy for each disorder are provided in Table 1. This shows that for each of the anxiety disorders being studied, citalopram was effective in significantly reducing clinical measures of severity as determined by a CGI change score of 2 or less (much or very much improved). As such, 20 of 37 patients (54%) were responders to citalopram.

Comparison of pre- and post-treatment scans for the whole group showed decreased activity in 4 significant clusters in grey matter (Figure 1): These included the superior cingulate, right thalamus, anterior cingulate, and the left hippocampus (Table 2). Comparison of pre- and post-medication scans showed no significant areas of activation.

Comparison of responders with non-responders demonstrated that responders had a significantly greater decrease of activity in 4 clusters (Figure 2). These clusters were localised to the left precentral, right middle frontal, right inferior frontal and, left prefrontal regions (Table 3). Comparison of baseline scans of responders and non-responders did not reveal any significant differences.

The main finding of this paper was that citalopram pharmacotherapy resulted in significant deactivation within anterior and superior cingulate cortex, the left hippocampus and the right thalamus in a combined group of patients with different anxiety disorders (OCD, PTSD, and SAD). Furthermore, deactivation was significantly more apparent in responders than in non-responders to SSRI treatment within precentral, right inferior, middle frontal and left prefrontal regions. Interestingly, no pre-treatment differences in regional perfusion between subsequent treatment responders vs non-responders were found.

Although there are important differences in the symptomatology of the anxiety disorders, these conditions do share certain aspects of their phenomenology, including heightened anxiety and avoidance behaviour. Furthermore, previous functional brain imaging work has demonstrated overlapping neurocircuitry across different anxiety disorders with activation of paralimbic circuitry and right inferior frontal cortex in a combined group comprising subjects with OCD, PTSD, and specific phobia 1. Results in the present study now also point to an overlap in the functional neuroanatomy, primarily implicating paralimbic neurocircuitry, in treatment response to the same SSRI, citalopram, across anxiety disorders. In citalopram responders, effects across disorders were most pronounced in the mid, inferior and prefrontal cortex. In other regions, such as the striatum, data on treatment response and symptom provocation seems to indicate less overlap across anxiety disorders, which may suggest only partial and regionally specific overlap between disorders 12.

Specific limbic regions are well-known to play a role in broadly mediating anxiety. Early observations of epileptogenic cingulate lesions support its role in regulating affect 27. Furthermore, recent work has suggested a role for the anterior cingulate in integrating cognitive and motivational processes. These include evaluating environmental cues and monitoring performance 28 On the other hand, a central role for the hippocampus in contextual aspects of fear conditioning has been demonstrated 2930.

The findings here complement previous studies of OCD, PTSD, and SAD that have demonstrated a specific role for the cingulate and hippocampus in these conditions. Studies in OCD have shown increased anterior cingulate activity at baseline, or deactivation during pharmacotherapy with serotonergic agents 31. In PTSD, anterior cingulate activity is also increased in some, although not all, studies of PTSD 3233 Further, the anterior cingulate is deactivated during citalopram treatment of SAD patients 17. Dysfunction of the hippocampus, as indicated by smaller hippocampal volume and declarative memory deficits, may play an important role in PTSD 34.

The medial prefrontal cortex comprises several related areas including anterior cingulate cortex. Lesions of this area are associated with suboptimal responses to stress, and the area has important inhibitory inputs to the amygdala which mediate extinction of fear conditioning 29. The middle and inferior frontal cortex, on the other hand, is involved in encoding and retrieval of verbal memories. Our finding that the right inferior frontal cortex was more deactivated in responders is perhaps consistent with previous findings showing increased activity pre-treatment in this region across different anxiety disorders 2 and in some, but not all, studies of PTSD 35.

Serotonergic circuits innervate the medial prefrontal cortex and other limbic structures, and chronic administration of a serotonin reuptake inhibitor may lead to an increase in their neurotransmission. It is possible that the medial prefrontal cortex deactivation during serotonergic pharmacotherapy indicates that a compensatory increase of activity in this region is no longer needed after symptom improvement. Along these lines, a number of functional and electrophysiological imaging studies of depression have found that anterior cingulate hyperactivity predicts a positive response to pharmacotherapy, a finding that has also been interpreted as indicating the baseline presence of an adaptive compensatory response 36. In addition changes in cognitive processing of frontal cortex may be secondary to symptom reduction caused by primary drug-induced changes within the limbic system. We have previously demonstrated similarly higher pre-treatment prefrontal perfusion in subsequent responders relative to non-responders using inositol in OCD 37. Interestingly, inositol responsive disorders overlap with those responsive to SSRI's which may suggest that it is serotonergic components of these disorders that account for at least some of the overlap in perfusion patterns demonstrated here.

In contrast, however, increased activity in anterior cingulate or orbitofrontal region in OCD has also been shown to predict a poorer response to pharmacotherapy 9.

Perhaps increased activity in particular limbic circuits plays a different functional role in different psychiatric disorders. Only limited functional imaging studies of pharmacotherapy effects have involved provocation paradigms 38 and such differences in design may account for certain inconsistencies across studies. Alternatively, it is feasible that different effects in different disorders may also help explain inconsistencies. In the current dataset, however, we were unable to demonstrate any associations between baseline activity and pharmacotherapy response for the combined group.

This study is limited by the slightly different inclusion criteria (inclusion of secondary depression in PTSD group) and pharmacotherapy duration for different disorders. While the absence of untreated controls may to some extent limit the conclusions we can draw, comparing non-responders to responders we believe serves as a reasonable evaluation of changes that result from treatment response. The lower spatial resolution of SPECT may be considered a limitation nevertheless this study usefully emphasizes the importance of limbic regions (amygdala, hippocampus) in mediating anxiety. Furthermore, deactivation within these regions as well as richly connected frontal regions following SSRI treatment, particularly in responders, is clearly demonstrated. Further research combining pharmacological interventions and functional methodologies, and using tracers tailored to specific neurotransmitter receptors, will undoubtedly lead to increased understanding of the pathogenesis of the anxiety disorders and the mechanisms of response to treatment in the future.

The authors declare that they have no competing interests.

The authors contributed to the work in the following ways: drafting of manuscript (PDC, JW, DJS), data analysis (JW, PDC), psychiatric evaluations (DJHN, GVDL, SS, DJS), SPECT imaging (JW, BBVH). All authors have read and approved the final manuscript.