Panic disorder (PD) is defined as episodic, unexpected panic attacks with no clear trigger that includes worry about further attacks and modifications to behavior in maladaptive ways to avoid them (1). Panic disorder is a common mental illness with a lifetime prevalence rate of 0.5%; the China Mental Health Survey reports estimates a 12-month prevalence of 0.3% (2).

The pathogenesis of PD is complex, comprised of an interaction of biological factors, particularly genetic factors (heritability estimates: 48%) (3) and brain structure (4). Gamma-aminobutyric acid (GABA) is the main inhibitory neurotransmitter of the central nervous system, and its dysfunction is considered to be one of the main neurobiological pathomechanisms of anxiety (5), especially patients with anxiety and panic disorder (6). GAD is the key rate-limiting enzyme that catalyzes the decarboxylation of glutamate to synthesize Gamma -aminobutyric acid, it has two isozyme forms–GAD67 (GAD1) and GAD65 (GAD2). The GAD1 gene is the rate-limiting enzyme for glutamate synthesis of GABA in the brain, GAD2 gene is present in membranes and nerve endings, so it is more sensitive to the effects of GABA levels (7). GAD’s role in the pathogenesis of PD has been well-established. One study analyzed the association of the GAD1 and GAD2 genes with genetic risk for a range of diseases. They found that variations in the GAD1 gene may affect susceptibility across a range of anxiety disorders (8). Another study suggested that patients with panic disorder exhibited significantly lower average GAD1 methylation than HCs, though no methylation alterations were observed for the GAD2 gene (9). A third study tested the genetic association of 93 single nucleotide polymorphisms (SNPs) with anxiety disorders in the Finnish population-based Health 2000 sample (282 cases and 575 matched controls), they showed that several SNPs in the GAD1 gene (rs769401, rs3791851, and rs769395) were associated with PD, with GAD1 having the most obvious association (10). Gorman (11) proposed a fear network model (FNM) of panic disorder based on an animal model of fear. FNM is centered in the amygdala and includes the thalamus (THA), hypothalamus, and hippocampus, as well as the locus ceruleus, periaqueductal gray area, and other brainstem sites. However, more extended regions of FNM have been identified in recent imaging studies. For example, Lai (12) suggests FNM also includes sensory regions of occipital lobe, temporal cortex, insula, and parietal cortex.

A previous meta-analysis of PD conducted by our research group revealed significant volume reductions in the right insula [extending to the PoCG, right inferior frontal gyrus, rolandic operculum, superior temporal gyrus (STG) and putamen], median cingulate/paracingulate gyrus, and SFG (13). In addition, some researchers have reported significant volume reductions in parietal regions (14) and the THA (15) in PD patients, whereas significant increases in volume have been found in the left inferior frontal gyrus (IFG), midbrain (16) and cuneus (17). The PoCG has been associated with functions of receiving, integrating, and interpreting most of the sensory information in the human body (18). Gorman et al. (11) showed that abnormalities in PoCG could potentially lead to misinterpretation of somatosensory information, thus allowing inappropriate activation of the FNM through misleading excitatory inputs to the amygdala (11). Another region that has gained attention is the ANG. ANG is important in empathic response, emotional regulation, mood, anxiety, and is associated with meditation and calmness (19). Previous studies reported that not only do PoCG and ANG play an important role in panic disorder, a correlation between GMV of those regions and clinical symptoms in PD exists as well (20).

Although evidence suggests that epigenetic alterations of GAD1 gene and gray matter volumes alterations of PoCG and ANG play an important role in the onset and maintenance of panic disorder, whether GAD1 methylation influences GMV of PoCG and ANG in PD has not yet been confirmed. Thus, the current study sought to explore the differences in whole brain gray matter volume and GAD1 gene promoter methylation levels between patients with panic disorder and the normal population. And furthermore, to explore the correlation between GAD1 methylation, brain gray matter volume and Panic Disorder Severity Scale (PDSS) scores. We hypothesized that GMV of PoCG and ANG mediates the relationship between GAD1 methylation and panic disorder severity.

The present study investigated whether GAD1 gene methylation impacted panic disorder severity and whether GMV played a mediating role in this relationship. Results point to a significant yet indirect relationship between GAD1 gene methylation and severity of PD. Importantly, the effects of GAD1 gene methylation on PDSS scores were mediated by gray matter volumes.

The question of whether and how GAD1 gene methylation affects PD has tremendous importance for improving our understanding of the pathogenic mechanism of panic disorder. We found that PD patients had lower gene methylation in ten GAD1 CPG sites compared to HCs. In general, DNA hypomethylation probably promotes gene expression (27). GAD1 gene hypomethylation increases GABA levels, it is possible that GAD1 hypomethylation of PD patients is not the cause of the panic behavior but a compensation for it. In view of the above results, we examined the link between methylation of ten CPG sites and PD severity. Results showed a significant negative association between methylation of cg12/171674146 site of GAD1 with clinical severity. This finding is supported by a recent study suggesting that GAD1 gene hypermethylation may play a compensatory role in panic disorder and may mediate the effect of negative life events on panic disorder (9). These results provide further evidence for the belief that GAD1 gene methylation plays an important role in the occurrence and development of PD.

The differences in GMV in the brains of PD and HCs imply that many structures are involved in the cognitive regulation of anxiety, the fear response, and the generation of threat perception, while also replicating and extending previous findings. Abnormal gray matter volumes in panic disorders has been reported in several studies (28, 29), and their results are consistent with our findings of decreased GMV in the regions of the left PHG, left SMG, left STG, right SPG, and right TPOsup, and bilateral THA in PD subjects compared to HCs. As we all know, the THA regulates emotional and cognitive functions (30). FNM suggests that changes in the temporal lobe may affect the transmission of sensory information to the THA for further filtering and subsequent “top-down” modulation of the frontal system (11). The THA interacts with temporal, parietal, subcortical limbic structures, and other FNM structures to modulate the noradrenergic system response toward fear (19, 31, 32, 33). Sensory areas of the brain, such as the occipital and temporal lobes, transmit sensory information to the FNM in order for the FNM to recognize and process facial and physical fear signals (34). Thus, the reduction of gray matter volumes in these brain regions affects the processing of fear information and the response to fear in patients with panic disorder. The results of multiple regression analysis also showed a negative correlation between cg12 site methylation with GMV of left PHG, the bilateral MTG, the left ANG, the left PoCG, and so on. Although the relationship between GAD1 gene methylation with gray matter volumes in PD remains unclear, the reduction in GABA receptor binding or reduced GABA activity occurs mainly in the frontal, limbic, temporal, and insula regions (35, 36). Consequently, anticipatory anxiety and panic attacks may be triggered by reduced GABAergic inhibition in different brain areas (37). Therefore, aberrant methylation of cg12 may be an important molecular mechanism underlying the anatomical changes in PD.

The results of the multiple regression analysis showed a significant positive correlation between right GMV of MOG, left PoCG, left anterior cingulate, and paracingulate gyri with panic disorder severity, and a negative correlation between gray matter volumes of left PCUN, the right middle frontal gyrus, and the right cuneus with PD severity. Interestingly, further mediation analysis demonstrated that GMV in the left PoCG totally mediated the association between cg12 methylation and PD severity. The PoCG has been associated with functions of receiving, integrating, and interpreting most of the sensory information in the human body (18). Abnormalities in PoCG could potentially lead to misinterpretation of somatosensory information, resulting in misleading excitatory input to the amygdala and inappropriate activation of the FNM (11). Notably, postcentral abnormalities were also seen when panic disorder was compared with MDD and generalized anxiety, suggesting that brain areas involved in somatosensory processing are more likely to be specific to patients with somatic symptoms, primarily in perceptual processing, rather than the general chronic and excessive feelings of anxiety and worry which is core features of generalized anxiety (38). Indeed, structural alterations in the PoCG may explain the specific symptoms of PD, which include abnormal perception of body signals and extreme sensation of the heartbeat. Thus, we suggest that a reduction in the volume of PoCG gray matter leads to more severe alterations in PD and this process is associated with cg12 methylation.

Path analysis showed that GMV of the ANG mediated the relationship between cg12 methylation and PD severity. The ANG plays an important role in affective regulation, empathic response, anxiety, and mood and is associated with meditation and calmness (19). In patients with panic disorder, abnormal gray matter volumes of ANG (part of parietal lobe) have been reported. For example, one study on voxel-based morphometry found reduced gray matter volumes in the parietal and temporal lobes of patients with PD (14, 15). These results provide further evidence for the idea that GMV in the right ANG potentially mediated the association between GAD1 gene methylation and PD severity.

The present study has the following limitations. First, epigenetic studies using peripheral blood do not allow for direct conclusions about the respective methylation patterns in brain tissue. Second, the study is cross-sectional and the direction of the relationship between brain structure and diseased state is uncertain. Third, the relatively small size might affect the significance of our results.

In conclusion, the results of this study confirmed a significant negative relationship between cg12 methylation and PD severity. Further, we revealed that gray matter volumes of PoCG and ANG mediated this relationship, which has not been found in previous PD research. Our findings suggest that cg12 site methylation of GAD1 may affect the development of PoCG and ANG GMV and further participate in the pathophysiology of panic disorder. Future PD research should combine epigenetics with neuroimaging, which is conducive to a more comprehensive understanding of the pathogenesis of panic disorder from the epigenetic level to the neural network level.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: NCBI, PRJNA793355.

The studies involving human participants were reviewed and approved by the Institute of Ethics Committee of the Affiliated Brain Hospital of Nanjing Medical University. The patients/participants provided their written informed consent to participate in this study.

CW and NL designed the study and supervised the conduct of the study. GL and HD contributed to the data collection. HX, YW, SY, and YZ provided the methodological advice. HW and HX performed the data analysis and results interpretation. HW and CW drafted the manuscript, which all authors reviewed and approved for publication. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.