Introduction

Post-traumatic stress disorder (PTSD), which may develop following a traumatic experience, is a debilitating illness characterized by emotional distress associated with physiological arousal and poor concentration (Atwoli et al. 2015). The traumatic experience may also lead to stress-induced functional changes in multiple domains of the brain. In order to develop precise treatments for the PTSD symptoms, identifying the biological targets of treatment is imperative (Bremner 2007).

The neural correlates of the emotional dysregulation in PTSD likely involve the fronto-limbic system, given the corresponding symptoms of anger, nervousness, fear, intrusive, recurrent recollections, flashbacks, and nightmares. However, only a few publications have utilized cognitive tasks that probe the poor concentration and other related executive function difficulties that are commonly noted in PTSD (Moores et al. 2008; Sailer et al. 2008; Felmingham et al. 2009; Cisler et al. 2015; Shaw et al. 2009; Elman et al. 2009; Astur et al. 2006; Strigo et al. 2010; Geuze et al. 2008; Bryant et al. 2005; Falconer et al. 2008; Jovanovic et al. 2013). Given that emotions influence cognition, it is hypothesized that there would be whole-brain involvement, with combined emotional and cognitive domain dysfunction and corresponding network abnormalities in both the cognitive and emotional brain networks. Therefore, the present meta-analysis included published studies that examined two brain domains, namely cognition and emotion, based on different levels of the domain hierarchy from the brainmap coding scheme (Laird et al. 2009). We also added a third domain, i.e., the interface between emotion and cognition, since the underlying networks show close connectivity, as informed by our prior work (Pavuluri et al. 2010), and hence may be involved in the pathophysiology of PTSD (Hayes et al. 2012b).

The majority of individual functional neuroimaging studies on PTSD are proof-of-concept trials with small sample sizes. One way to amplify the power of these studies is to perform a meta-analysis, which integrates the data from all published studies on PTSD. One of the most common algorithms for performing coordinate-based meta-analyses that offer robust and reliable findings is the recently developed activation likelihood estimation (ALE) technique (Eickhoff et al. 2009). This method assumes that the peak coordinates reported by each study represent the activation maps from which they are derived and uses the reported coordinates in voxel-wise analyses to assess the consistency of activation in any given set of studies (Eickhoff et al. 2009; Kober and Wager 2010; Turkeltaub et al. 2012). By using this type of quantitative voxel-wise meta-analysis of already published results we can compare the findings from the PTSD population to those from the non-clinical control (NC) group and yield objective, unbiased, and statistically based quantified evidence of aberrant brain activation during cognitive, emotional and interface of both processing in PTSD. As far as we know, there has been no previous study about the meta-analysis of PTSD using the ALE method for those three domains – cognition, emotion, and interface of both.

Here, we conducted a separate meta-analysis for each individual domain of cognition, emotion processing, and the interface between cognition and emotion, as they relate to the PTSD diathesis. We hypothesized that emotional systems and circuits, in individuals with and without PTSD, would be closely linked to cognitive circuits. By separating which probe or domain dysfunction, cognition, emotion, and/or the interface of both, contributes to the activity in any given coordinate in this meta-analysis, we can quantitatively evaluate the results of neuroimaging studies performed in individuals with PTSD and understand that cognition and emotion are separate but interacting sub-systems of the brain in PTSD.

Methods

Search strategy

We performed a systematic literature search following the flow suggested by the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement (Moher et al. 2009). We identified primary studies through a comprehensive literature search of the MEDLINE (using both free-text and Medical Subject Headings searches) and PsychINFO databases using the following keywords: PTSD or traumatic or post-traumatic, and functional magnetic or fMRI. In addition, manual searches of the reference sections of review articles and individual studies were conducted to check for any missing studies not identified by the computerized searches. There were no language restrictions, as all included manuscripts were written in English. An initial list of studies was produced that included any report of functional magnetic resonance imaging (fMRI) studies of PTSD published in print or online by December 31, 2017. The selection process for the final list of primary studies for the planned meta-analyses in this study was very clear and specific. Accordingly, the inclusion of specific study could not be misinterpreted.

The first-level literature search with further manual search yielded 179 unique published articles with 73 studies meeting the following selection criteria. All reports included in the meta-analysis satisfied the following criteria: (1) included an NC comparison group; (2) conducted whole-brain analyses with or without region of interest (ROI) analyses; (3) provided standard Talairach or Montreal Neurological Institute (MNI) spatial coordinates for the key findings; (4) patient groups consisted of individuals diagnosed with PTSD according to the specific diagnostic criteria such as Diagnostic and statistical manual of mental disorders; (5) each subject group consisted of at least five subjects; (6) reported activation foci as three-dimensional coordinates in stereotactic space; (7) utilized active task constructs; and (8) presented the findings at the disorder vs. NC group level. Excluded manuscripts consisted of the following: (1) reviews or other meta-analyses; (2) studies with subject overlap; and (3) studies using other MRI modalities (e.g., structural imaging, spectroscopy, diffusion tensor imaging, and functional connectivity studies).

After a second-level review of these 73 studies, only 56 studies reported the coordinates that were essential for inclusion in this meta-analysis (Table 1). Any ambiguity in inclusion was resolved through discussion by the authors of this manuscript until consensus was reached. Study data (e.g., coordinates, subject numbers) were entered and cross-checked by the participating authors. The grouping of the studies based on the task type, i.e., emotion, cognition, and the interface of emotion and cognition, was also determined through independent classification decision-making by researchers, with any differences in classification being resolved through discussion until consensus (Fig. 1).

Table 1 Primary fMRI studies of participants with post-traumatic stress disorder (PTSD), and non-clinical controls included in meta-analysis
Full size table
Fig. 1
figure 1

PRISMA Flow Diagram of the Literature Search

Full size image

ALE methods and pairwise ALE meta-analysis

The GingerALE software (version 2.3.6) from the BrainMap Project was used to conduct ALE meta-analyses of eligible studies (Eickhoff et al. 2009; Turkeltaub et al. 2012; Eickhoff et al. 2012). We separately analyzed the data by the type of the task or brain domain probed, i.e., cognition, emotion, and the interface between cognition and emotion. Activation coordinates originally reported in MNI space were converted to Talairach coordinates using the Lancaster transform (icbm2tal) program the in GingerALE software. Our meta-analyses were conducted in Talairach space. For uniformity, Talairach coordinates expressed by the previous Brett transformation (Brett et al. 2002) were converted into MNI space and re-transformed into Talairach space.

We performed pairwise ALE meta-analyses that included the following comparisons in each domain (cognition, emotion, interface between cognition and emotion): greater activation in PTSD vs. NC and greater activation in NC vs. PTSD. The input coordinates were weighted to form estimates of activation likelihood for each intracerebral voxel. The activation likelihood of each voxel in standard space was then combined to form a statistic map of the ALE score at each voxel. Statistical significance of the ALE scores was determined by a permutation test controlling the false discovery rate at p < 0.05 (Laird et al. 2005). The statistic maps were thresholded by default at this critical value, and a recommended minimum cluster size was suggested from the cluster statistics. By using this minimum cluster size of supra-threshold voxels, we obtained the thresholded ALE image. The results of the pairwise ALE analyses are reported at p = 0.05 and are whole-brain corrected. The Talairach daemon was used to determine the anatomical locations of significant clusters.

Analyses were performed in two steps. In step one, only whole-brain analyses were included. In step two, both the whole-brain analyses and the ROI analyses were included.

Results

Our meta-analyses included 56 fMRI studies with PTSD and NC groups (emotion: 28 studies, cognition: 12, interface between cognition and emotion: 16 studies). No overlap in subjects that completed the same task across the selected studies was noted. The primary studies included in the meta-analyses are listed in Table 1. Findings are summarized in Table 2 and Fig. 2.

Table 2 Activation likelihood estimation (ALE) meta-analysis findings for fMRI studies comparing participants with post-traumatic stress disorder (PTSD), and non-clinical controls
Full size table
Fig. 2
figure 2

Results from pairwise activation likelihood estimation analysis. Each coordinate represents center coordinate of each cluster. A) Emotional task, PTSD>normal control, combined results of whole brain plus ROI analysis. Amygdala, x = 23.6, y = −6.7, z = −13.3, cluster size = 168 mm3, extreme value = 0.020 (B) Emotional task, normal control>PTSD, whole brain only analysis. Anterior cingulate cortex, x = −0.6, y = 34, z = −4.3, cluster size = 144 mm3, extreme value = 0.023 (C) Emotional task, normal control>PTSD, whole brain plus ROI analysis. Anterior cingulate cortex, x = −0.6, y = 34, z = −4.3, cluster size = 104 mm3, extreme value = 0.023 (D) interface between cognition and emotion, PTSD>normal control, whole brain only analysis. Lentiform nucleus, lateral globus pallidus, x = −18.6, y = −2.2, z = −7.2, cluster size = 240 mm3, extreme value = 0.024 (E) interface between cognition and emotion, PTSD>normal control, whole brain plus ROI analysis. Lentiform nucleus, lateral globus pallidus, x = −18.7, y = −2.2, z = −7.2, cluster size = 248 mm3, extreme value = 0.024

Full size image

Cognition

No significant differences were found between the PTSD and NC groups for the cognitive task results either in the whole-brain only analyses or in the whole-brain plus ROI analyses.

Emotion

The PTSD group did not show any clusters with significantly greater activation than the NC group in the whole-brain only analyses. However, in the whole-brain plus ROI analyses, the PTSD group showed greater activation than the NC group in one cluster, i.e., the limbic lobe including the amygdala and parahippocampal gyrus within the right cerebrum (Fig. 2, Table 2).

In the whole-brain only analyses, the NC group showed significantly greater activation than the PTSD group in one cluster, namely the bilateral anterior cingulate cortex (ACC), at the interface of both the dorsal and the ventral ACC. In whole-brain plus ROI analyses, the NC group also showed significantly greater activation than the PTSD group in the same cluster (Fig. 2).

Interface between cognition and emotion

The PTSD group showed significantly greater activation than the NC group in both the whole-brain analyses and in the whole-brain plus ROI analyses in one cluster, i.e., the caudate, putamen, and globus pallidus (Fig. 2).

When comparing the NC group relative to the PTSD group, no clusters of significantly greater activation were found in either the whole-brain analyses or the whole-brain plus ROI analyses.

Discussion

The present meta-analysis is the first to collectively evaluate studies that used tasks probing emotion, cognition, and the interface of emotion and cognition. Given the small number of studies, we employed the broad approach of using whole-brain and ROI analyses to assess the affective circuitry’s operations in individuals with PTSD. Our findings illustrate that PTSD predominantly affects the emotional circuitry regions either directly or in association with the cognitive circuitry. Our results are particularly important as extreme affect dysregulation in the PTSD patients when faced with exterrnal triggers may often be confused with that of bipolar disorder. This study is one of the many steps to discern the underlying brain pathophysiology of emotional and cognitive systems in the PTSD.

Fear, negativity, avoidance, and hyperarousal are known clinical features of PTSD, which may explain the increased amygdala activation that we observed in individuals with PTSD compared to NC individuals (emotion domain, PTSD>NC, Right parahippocampal gyrus and amygdala in whole brain + region of interest analysis). Indeed, increased activation of the limbic system, especially the amygdala, has been observed in individual studies of patients with PTSD (Lanius et al. 2010). The amygdala is known to play a key role in fear conditioning (Rogan et al. 1997), including regulating learned fear during Pavlovian fear conditioning, and individuals with PTSD show fear associated with such conditioning (Mahan and Ressler 2012). The parahippocampal gyrus is also related to memory encoding (Alkire et al. 1998; Brewer et al. 1998). Interestingly, the amygdala reportedly modulates memory processes linked with the hippocampus-parahippocampus complex (Packard et al. 1994). In line with these studies, a hyperactive limbic lobe, including the amygdala in individuals with PTSD, modulates memory through its connectivity with the parahippocampal gyrus and the frontal regions during emotionally arousing learning situations that mimic traumatic stress (Kilpatrick and Cahill 2003). Our results support the hypothesis that individuals with PTSD may show an exaggerated response to fearful stimuli owing to hyperresponsiveness of the amygdala and related structures including the parahippocampal gyrus.

In contrast, individuals with PTSD showed less activation in the bilateral ACC, the key interface region involving the dorsal and ventral ACC (emotion domain, NC > PTSD, both ACC in whole brain analysis). Activation of the ACC is associated with decreased limbic activity, perhaps explaining the resolution of emotional conflict at the cortical level coupled with the top-down inhibition of limbic activity (Etkin et al. 2006). Indeed, the ACC relays projections from higher-order sensory areas to other regions of the prefrontal cortex and subcortical striatum (Johnson et al. 2003; Gunaydin and Kreitzer 2016), as well as the amygdala (Etkin et al. 2006).

It is quite important to understand the dynamic interplay between emotion and cognition. The lentiform nucleus (globus pallidus and putamen), parahippocampal gyrus and amygdala all showed greater activity in the PTSD than NC group when both the emotional and cognitive systems were jointly probed the interface domain (interface domain, PTSD>NC, lentiform nucleus, lateral & medial globus pallidus, parahippocampal gyrus, amygdala, putamen in whole brain analysis & whole brain + region of interest analysis). The PTSD group showed greater activity in amygdala than the normal control group. Other recent meta-analysis of imaging studies in the PTSD subjects showed that amygdala and mid-anterior cingulate cortex is hyperactive, whereas lateral and medial prefrontal cortex is hypoactive in that group for negative emotional stimuli vs. neutral or positive stimuli (Hayes et al. 2012a). A neurocircuitry model of PTSD shows that dysfunction of the ventromedial prefrontal cortex results in failure to inhibit an overactive amygdala, leading to an exaggerated fear response and impaired fear extinction learning (Hayes et al. 2012b).

The basal ganglia, especially the striatum, is involved in concentration and emotional processing along with the amygdala (Wise et al. 1996; Hollerman et al. 2000). The striatum is also a critical component of the motor and reward systems. Motor activity and planning, as moderated by the globus pallidus and striatum including putamen, where significant activity differences were shown in this meta-analysis, may explain why individuals with PTSD are unable to take action or avoidance measures in the context of emotional arousal (Stark et al. 2015). Unfortunately, our data could not afford to confirm this in the present study, as we could not address the correlation with specific PTSD clinical symptoms.

The small number of individual studies that were available for inclusion in this meta-analysis posed some limitations. First, the control group consisted of both healthy volunteers and trauma-exposed controls without clinical symptoms. Second, we presented the combined results of the ROI analyses and whole-brain analyses to assess the involvement of the hypothesized regions as simply mixing the ROI activation results with whole-brain scanning can bring the overrepresentation of specific ROI activations. However, we presented the combined results of whole brain and ROI analysis only as an option in addition to the whole brain analysis results. Finally, we could not provide a developmental perspective, as only two of the included studies specifically evaluated youths with PTSD.

In conclusion, the findings from this meta-analysis imply that individuals with PTSD process emotion and cognition in the subcortical regions, including the striatum, and in the limbic regions, including the amygdala, when the emotional system is probed with and without the cognitive effort, respectively. We also found that individuals with PTSD exhibited prominent underactivity in the ACC, especially when processing emotions, suggesting that PTSD may be an emotional disorder. These findings underscore the mechanistic dysfunction in key subcortical regions that potentially explain the emotional and cognitive distress experienced by clinically ill individuals diagnosed with PTSD.