Association between Uncinate Fasciculus Integrity and Agoraphobia Symptoms in Female Patients with Panic Disorder.

OBJECTIVE: Although neural correlates of sub-clinical agoraphobia (AG) symptoms have been previously suggested, only a few studies evaluating structural changes of the brain have been conducted in agoraphobic patients with panic disorder (PD). We investigated and compared white matter (WM) micro-structural alterations between PD patients with AG (PD ＋ AG) and those without AG (PD - AG).
METHODS: Our study included 56 female PD patients, of which 25 were diagnosed with AG and 31 were diagnosed without AG. Diffusion tensor imaging was performed to investigate micro-structural changes in the WM tracts related to fronto-temporo-occipital areas (uncinate fasciculus, cingulum bundle, inferior longitudinal/fronto-occipital fasciculus, fornix column and body, and fornix/stria terminalis). All participants were subjected to the Anxiety Sensitivity Inventory-Revised (ASI-R), Beck Depression Inventory-II (BDI-II), and Albany Panic and Phobia questionnaires.
RESULTS: The fractional anisotropy values of the right uncinate fasciculus in PD ＋ AG were significantly lower than that of PD - AG and showed significant correlations with BDI-II and ASI-R total scores. Mean diffusivity and radial diffusivity values of the right uncinate fasciculus were significantly higher in PD ＋ AG as compared to PD - AG.
CONCLUSION: Our findings suggest that the uncinate fasciculus may be associated with AG symptoms in PD, possibly through demyelination. Our findings may contribute to the neurobiological evidence regarding the association between AG and WM structural changes in PD.

INTRODUCTION

Agoraphobia (AG) is defined as an anxiety disorder characterized by marked fear in situations where the person perceives their environment to be unsafe with no way to escape [1]. With the revision of the Diagnostic and Statistical Manual of Mental Disorders, fourth edition text revision (DSM-IV-TR) to the DSM-V (fifth edition), AG was re-classified as a sole psychological disorder separate from panic disorder (PD). Based on this revision, the emphasis was on AG being an independent psychological entity; hence, it was defined as a separate disorder, despite being strongly associated with PD [2]. The prevalence rate of AG without PD was reported to be 0.8%, whereas that of AG with PD was 1.1% [3]. Younger age, female sex, and comorbidity with PD and other phobias were reported to be risk factors for AG [4]. Furthermore, AG has been associated with substantial comorbidity with other DSM disorders, with higher rates of comorbidity observed in PD ＋ AG as compared to PD − AG [3].

While providing the classical hypothesis on the neuroanatomical basis of PD, Gorman et al. [5] had emphasized the role of the fear network involving frontal and limbic regions. The uncinate fasciculus is one of the white matter (WM) tracts connecting fronto-limbic structures an association tract that connects parts of the limbic system such as the parahippocampus and amygdala in the temporal lobe with portions of the frontal lobe such as the orbitofrontal cortex [6]. Recent imaging studies have discovered extended areas of the fear network model (FNM), such as the sensory regions of the temporal and occipital lobes [7], to be strongly associated with PD. The visuo- spatial dysregulation and false threat alarm related to the temporal lobe has been well discussed by many literatures on PD [8,9]. Dysfunction in the processing of sensory information from fear-related stressors in the visuo-spatial system stimulates structures in the temporal region to exchange information with the autonomic nerve system and neuroendocrine system [10]. Since the occipital lobes are associated with awareness of fearful stimuli and the surrounding environment through visuo-spatial information, they are involved in transmitting sensory information to the FNM to process fear and environmental changes, which may contribute to panic attacks [11].

There are not many reports discussing the brain’s structural role in AG, but structures related to the occipital and temporal lobes has been suggested to mediate AG symptoms as it shares common disease characteristics with PD, such as sensory-related fear, anxiety response, and visuo- spatial regulation dysfunction [12]. Anticipatory anxiety commonly observed in AG and PD could be mediated by brain activity dysfunction of the occipital lobe [13]. Hyperactivation of the temporal lobe was reported in anticipating AG-specific stimuli [14]. A recent network study on sub-clinical AG demonstrated a lower level of efficiency and clustering of the visuo-spatial-emotional network, including the amygdala, primary visual cortex, visual stream, parietal lobes, and prefrontal cortices. Hence, this lower connectivity suggests that sub-clinical AG symptoms may not organize all of this information into high-level executive function, and may react more intuitively to AG stimuli [15].

Recently, in patients with sub-clinical AG symptoms, positive correlations between symptom severity and cortical volumes of temporo-occipital areas, parts of the extended fear network, including bilateral calcarine sulci, right lingual gyrus, left superior, middle and inferior temporal gyri areas, have been reported [12]. We suggest that WM tracts associated with the extended fear network and fronto-temporo-occipital areas may underline the neural correlates of AG. Imaging techniques associated with diffusion tensor imaging (DTI) have made it possible to investigate alterations in the micro-structural integrity of the WM tracts [16].

Given the above considerations, we investigated and compared the WM micro-structural alterations between PD ＋ AG and PD − AG. First, we hypothesized that PD ＋ AG would exhibit altered integrity in the WM tracts related to fronto-temporo-occipital areas as compared to PD − AG. Second, we hypothesized that AG symptom severity will show significant correlations with WM integrity in PD ＋ AG.

METHODS

Participants and Clinical Assessment

We studied 56 female patients with PD who were recruited from the outpatient psychiatric clinic of CHA Bundang Medical Center. All participants were between 18 and 65 years of age, right-handed, and of Korean ethnicity. The medical histories of the participants were recorded through interviews and questionnaires. Diagnoses of PD ＋ AG and PD − AG were determined by experienced psychiatrists according to the DSM-IV-TR, using the Structured Clinical Interview for DSM-IV Axis I disorders [17]. Only patients with primary diagnosis of PD were included. The exclusion criteria were as follows: (1) Primary or comorbid psychiatric diagnoses on Axis I or Axis II (based on DSM-IV-TR criteria) other than PD; (2) history of serious or unstable medical illness; (3) abnormal findings in physical examination and routine laboratory tests; (4) primary neurological illness; (5) pregnancy; and (6) any contraindications for brain magnetic resonance (MR) scanning. Fifty-four patients were consuming psychotropic medications, including selective serotonin reuptake inhibitors such as escitalopram and paroxetine (escitalopram equivalent dosage, 8.33 ± 7.09 mg/day), and benzodiazepines such as alprazolam and clonazepam (alprazolam equivalent dosage, 1.43 ± 1.17 mg/day). Brain MR scans of PD ＋ AG and PD − AG were obtained after initiation of medication after 3.92 ± 3.81 and 5.87 ± 5.71 days, respectively. Two PD ＋ AG did not consume any type of psychotropic medication within the study period.

To evaluate symptom severities, the Anxiety Sensitivity Inventory-Revised (ASI-R), Albany Panic and Phobia Questionnaire (APPQ), Beck Depression Inventory-II (BDI-II), and Beck Anxiety Inventory (BAI) were conducted by all participants. Anxiety sensitivity levels were assessed using the Korean version of the ASI-R [18,19], which consists of fear of a respiratory symptom, fear of a cardiovascular symptom, fear of cognitive dyscontrol, and fear of a publicly observable anxiety reaction. The internal consistency coefficient of the Korean version is 0.92 and its test-retest reliability is 0.82. The APPQ was developed to assess fear of activities that may induce physical sensation in PD patients. The Korean version of the APPQ, showed good internal consistency (Cronbach’s alpha = 0.95) and high test-retest reliability (r = 0.77) [20]. APPQ has three sub-scales: interoceptive avoidance, AG, and social phobia [21].

This study was approved by the CHA Bundang Medical Center Ethics Committee (no. 2011-11-164, 2018-06-029, 2019-05-030). All procedures involved in this study complied with CHA Bundang Medical Center Institutional Review Board regulations, Declaration of Helsinki principles, and Good Clinical Practice principles. All participants fully understood the protocol of the research and provided their oral and written consents.

MRI Acquisition

All participants underwent MR imaging on the 3.0 T GE Signa HDxt scanner (GE Healthcare, Milwaukee, WI, USA) comprising of an eight-channel phase-array head coil at the CHA Bundang Medical Center, CHA University, and diffusion data were acquired on a 3.0 T GE Signa HDxt scanner (GE Healthcare). Diffusion-weighted images were processed using an echo planar imaging (EPI) sequence with the following parameters: repetition time (TR) of 17,000 ms, echo time (TE) of 108 ms, field of view (FOV) of 24 cm, 144 × 144 matrix, 1.7 mm slice thickness, and voxel size of 1.67 × 1.67 × 1.7 mm3. A double-echo option was applied to minimize the effect of eddy current. An eight-channel coil and an array of spatial sensitivity encoding techniques (ASSET, GE Healthcare) with a sensitivity encoding (SENSE) speed-up factor of two were used to reduce the impact of EPI spatial distortions. Seventy axial slices parallel to the anterior commissure−posterior commissure (AC-PC) line covering the whole brain were acquired in 51 directions with b-value = 900 s/mm2. Eight baseline scans with b = 0 s/mm2 were also acquired. DTIs were approximated from the diffusion-weighted images using the least-squares method (approximate scan time = 17 min).

Image Processing

Voxel-wise analysis of DTI metrics data was evaluated using Tract-Based Spatial Statistics (TBSS version 1.2) applied in the Oxford functional MRI of the brain (FMRIB) Software Library (FSL version 4.1, Oxford, UK; https://fsl. fmrib.ox.ac.uk/fsl/fslwiki) according to the standard procedure [22]. First, DTI preprocessing, including skull stripping using the Brain Extraction Tool (BET) and eddy current correction, were performed using the FSL. Ac-cordingly, FA images were constructed by fitting a tensor model to the raw diffusion data [23]. The FA data of all the participants were aligned in the standard space (Montreal Neurologic Institute 152 standard) using the FMRIB’s nonlinear image registration tool (FNIRT). All transformed FA images were combined and applied to the original FA map, resulting in a standard-space version of the FA map. All transformed FA images were averaged to create a mean FA image, which was thinned (skeletonized) to create a mean FA skeleton, including only the centers of the WM tracts. The skeleton was thresholded using an FA > 0.2 (TBSS default) to include only major fiber bundles. To compare the axial diffusivity (AD), radial diffusivity (RD), and mean diffusivity (MD), we used FSL of the FA images to achieve non-linear registration and skeletonization stages, as well as to estimate the projection vectors from each individual participant onto the mean FA skeleton. The non-linear warps and skeleton projection can also be applied to other images.

For the analysis of neural correlates of PD ＋ AG vs. PD − AG, mean FA skeletons were multiplied using the Johns Hopkins University (JHU) DTI-based probabilistic tractography atlas. Subsequently, the aforementioned regions of interests (ROIs) were extracted using the three-dimensional Slicer version 3.6 to create a mask (threshold: p = 0.05) and perform voxel-wise statistical analysis [24]. The selected WM ROIs included the fornix, stria terminalis, cingulum bundle, uncinate fasciculus, and inferior longitudinal/fronto-occipital fasciculus from the JHU WM atlas, based on previous studies [12,15]. These fronto-temporo-occipital regions were involved in the extended fear network linked with AG symptoms (Fig. 1).

Fig. 1

3D reconstructions of the regions of interest.

The FA index is the most commonly and widely used parameter of DTI, as it detects the integrity of WM fibers [25]. FA indicates various characteristics of the axon fibers, with increased values representing either a greater number and size of the axon fibers or a decrease in the density of the axon fibers [26]. Previous studies have reported AD to be sensitive to the axonal pathogenesis, MD to be sensitive to necrosis and edema, and RD to be sensitive to myelination, which may correspond to the WM micro-integrity that in turn affects the functional connection of the axonal fibers [27]. Therefore, values for AD, RD, and MD could supplement FA values in order to analyze the potential underlying pathogenesis in tissue micro-structure [26,27].

Statistical Analysis

To compare the sociodemographic and clinical data between PD ＋ AG and PD − AG, the independent ttest was used. The independent ttest was applied with multiple test correction using post-hoc Benjamini-Hochberg false discovery rate to compare the FA values between PD ＋ AG and PD − AG. To ensure the results were not confounded by other variables, such as age and intracranial volume, analysis of covariance (ANCOVA) was performed. Spearman correlation was used in each PD group for non-hypothetical exploratory analysis.

All statistical analyses were performed using the Statistical Package for the Social Sciences (IBM SPSS statistics 25.0 software; IBM Co., Armonk, NY, USA). For all tests, p < 0.05 was considered statistically significant.

RESULTS

Sociodemographic Characteristics

Table 1 summarizes the sociodemographic and clinical characteristics of PD ＋ AG and PD − AG There were no statistically significant differences between the two diagnostic groups, in terms of age at the time of scan, years of education, and intracranial volume (ICV). APPQ, BAI, BDI-II, and ASI-R total scores, along with APPQ sub-scale scores, were significantly higher in PD ＋ AG as compared to PD − AG.

Table 1

Demographic and clinical characteristics of the panic disorder participants

Variable	PD − AG (n = 31)	PD ＋ AG (n = 25)	Statistics
			t	pvalue
Age (yr)	35.52 ± 10.83	38.44 ± 10.53	−1.02	0.314
Education (yr)	13.32 ± 2.99	13.08 ± 3.17	0.30	0.766
Intracranial volumes (mm3)	1,400.27 ± 119.08	1,361.63 ± 87.28	1.14	0.262
BDI-II, total score	15.06 ± 7.58	22.96 ± 10.75	−3.19	0.002
BAI, total score	23.10 ± 11.52	36.17 ± 13.57	−3.86	0.000
ASI-R, total score	44.97 ± 19.20	68.46 ± 29.22	−3.38	0.002
APPQ, total score	31.94 ± 21.38	105.58 ± 35.03	−9.08	0.000
Agoraphobia	10.77 ± 7.11	44.16 ± 13.06	−11.48	0.000
Social phobia	10.00 ± 8.85	30.40 ± 17.83	−5.23	0.000
Interoception	11.16 ± 9.98	31.29 ± 12.37	−6.68	0.000
SSRI escitalopram equivalent dosage (mg/day)a	7.5 ± 7.33	9.40 ± 6.78	−0.98	0.330
Benzodiazepine equivalent dosage (mg/day)b	1.46 ± 1.45	1.39 ± 0.73	−0.21	0.830

Values are presented as mean ± standard deviation.

PD, panic disorder; AG, agoraphobia; BDI-II, Beck Depression Inventory-II; BAI, Beck Anxiety Inventory; ASI-R, Anxiety Sensitivity Index-Revised; APPQ, Albany Panic and Phobia Questionnaire; SSRI, selective serotonin reuptake inhibitor.

aThe approximate equivalent oral doses to 10 mg escitalopram are given. bThe approximate equivalent oral doses to 1 mg lorazepam are given.

Group Differences in FA, AD, MD, and RD Values

In the comparison between PD ＋ AG and PD − AG, significant difference in the FA values of the uncinate fasciculus was observed (Table 2). PD ＋ AG showed significantly lower FA values as compared to PD − AG (t= 4.21, Benjamini-Hochberg False Discovery Rate [FDR]- corrected p = 0.021). The MD and RD values of PD ＋ AG were significantly higher than PD − AG, while no significant difference was observed in the AD values between the two groups (MD, corrected p = 0.041; RD, corrected p = 0.034). No significant difference in the other ROIs was observed between PD − AG and PD ＋ AG (Table 2). Moreover, no significant correlation was shown between the FA values of the right uncinate fasciculus and age, years of education, duration of medication, medicine dosage, and ICV. Furthermore, the difference in FA values of the uncinate fasciculus remained significant when ANCOVA was performed with the aforementioned covariates. There was no significant interaction effect between the ROIs according to laterality. That is, there was no significant difference between the FA values of the right and left uncinate fasciculus. In addition, there was no interaction effect between laterality and groups

Table 2

Comparison of FA values of ROIs between PD ＋ AG and PD − AG

Regions of interest	PD − AG (n = 31)	PD ＋ AG (n = 25)	Statistics
			t	pvaluea
Fornix (column and body)	0.66 ± 0.13	0.62 ± 0.11	0.981	0.328
Left
Inferior longitudinal/ fronto-occipital fasciculus	0.63 ± 0.03	0.63 ± 0.04	0.381	0.705
Fornix (cres/stria terminals)	0.65 ± 0.03	0.65 ± 0.04	0.809	0.374
Uncinate fasciculus	0.63 ± 0.04	0.62 ± 0.06	3.481	0.069
Cingulum bundle	0.73 ± 0.05	0.73 ± 0.04	0.388	0.669
Right
Inferior longitudinal/ fronto-occipital fasciculus	0.67 ± 0.03	0.66 ± 0.03	1.158	0.252
Fornix (cres/stria terminalis)	0.71 ± 0.03	0.70 ± 0.04	0.001	0.97
Uncinate fasciculus	0.65 ± 0.04	0.62 ± 0.06	4.206	0.021
Cingulum bundle	0.77 ± 0.04	0.77 ± 0.04	0.322	0.541

Values are presented as mean ± standard deviation.

FA, fractional anisotropy; ROI, region of interest; PD, panic disorder; AG, agoraphobia.

apvalues were corrected for multiple comparisons using a Benjamini-Hochberg false discovery rate correction method.

Exploratory Correlation Analysis between FA Values of the Uncinate Fasciculus and Clinical Symptom Severity in Patients with PD

In PD ＋ AG, the FA values of the uncinate fasciculus showed significant correlations with BDI-II and ASI-R total scores, as well as the ASI-R cognitive sub-scale scores (r = 0.476, p = 0.02; r = 0.405, p = 0.049; and r = 0.413, p = 0.04, respectively). In PD-AG, FA values of the uncinate fasciculus significantly correlated with the total APPQ and APPQ interoception sub-scale scores (r = 0.452, p = 0.011; r = 0.389, p = 0.031, respectively). Nevertheless, after multiple comparisons with Bonferroni correction, the pvalues did not survive (Figs. 2, 3).

Fig. 2

Exploratory correlation analysis between FA value of right uncinate fasciculus and clinical symptom severity in PD ＋ AG. PD, panic disorder; AG, agoraphobia; ASI-R, Anxiety Sensitivity Inventory-Revised; BDI-II, Beck Depression Inventory-II.

Fig. 3

Exploratory correlation an-alysis between FA value of right un-cinate fasciculus and clinical symptom severity in PD − AG. PD, panic disorder; AG, agoraphobia; APPQ, Albany Panic and Phobia Questionnaire.

Association Analysis Using Multiple Linear Regression

In PD ＋ AG, the multiple linear regression analysis showed FA value of the uncinate fasciculus to be significantly associated with ASI-Cog (p = 0.026), ASI-SUM (p = 0.018), and BDI-SUM (p = 0.014), with age, ICV, education, and medication effect included as covariates in the analysis.

In PD − AG, the multiple linear regression analysis showed FA value of the uncinate fasciculus to be significantly associated with APPQ sum (p = 0.039), but not with APPQ interoception, with age, ICV, education, and medication effect included as covariates in the analysis.

DISCUSSION

To the best of our knowledge, this is the first study to examine and compare WM structural changes between PD ＋ AG and PD − AG. The present study observed that the FA values of the uncinate fasciculus in PD ＋ AG were significantly lower than that of PD − AG. Furthermore, FA values of the uncinate fasciculus in each group showed positive correlations with symptom severities.

The results of the present study indicate decreased integrity of the uncinate fasciculus in PD ＋ AG as compared to PD − AG. The uncinate fasciculus connects the prefrontal lobe and amygdala, which are central in controlling amygdala-related fear responses [28]. Generally, negative correlations have been observed between structural integrity and anxiety trait levels (i.e., higher FA values predicted lower anxiety levels) [29]. Furthermore, the uncinate fasciculus has also been associated with executive functions, such as associative and episodic memory functions and social-emotion functions [30].

High AG-related anxiety is closely associated with spatial and visual processing and activation of autobiographical memory to interpret the surrounding environ-ment. Commonly, patients describe this arousal as more distressing and anxiety inducing in daily life than confronting and staying in the agoraphobic situation [31]. Functionally, impaired integrity of the uncinate fasciculus could contribute to the disconnectivity between the prefrontal cortex and amygdala. This may lead to the dysfunction in inhibitory control of the prefrontal cortex on the amygdala, and influence amygdala-related fear responses. Episodic memory functions and social-emotion functions may also be impaired, affecting the anticipatory anxiety related to AG and misinterpreting the surrounding environment to be safe or not from previous memory. Impaired uncinate fasciculus-related executive functions could result in prominent susceptibility to phobic situations and greater risk for developing phobic symptoms, which may contribute to the development of AG symptoms.

Similar fronto-temporal structural hypo-connectivity has also been observed in patients with social anxiety disorder and generalized anxiety disorder, which indicates an association between uncinate fasciculus abnormalities and emotional regulation deficits [32,33]. Decreased FA values of WM around the frontal lobe were observed in patients with PD, which indicated impaired micro-connectivity of the frontal WM in PD [34]. Furthermore, antidepressant therapy was shown to increase micro-integrity of the fronto-occipital fasciculus and uncinate fasciculus in PD [35]. Therefore, phobic and anxiety symptoms of AG could be related to impaired micro-connectivity of the uncinate fasciculus, which connects the frontal lobe structures including the orbitofrontal cortex and anterior cingulate cortex, and the anterior temporal lobe, including portions of the amygdala and hippocampus, which are critical structures in fear and memory modulation [36].

The exact role of the right and left uncinate fasciculus is still unknown, but one may speculate from the results of previous studies which have focused on hemispheric lateralization in psychiatric disorders [37]. Depression has often been associated with left brain damage [38], and panic anxiety has been associated with right cortical disturbance [39]. In rodents, Andersen et al. found increased serotonin levels in the right, but not the left amygdala in relation to anxiety, which also suggested anxiety to be associated with right hemispheric function [40].

Numerous case reports have reported associations between right sided brain lesions such as tumors and anxiety [41,42], and certain studies have suggested left cortical lesions to be associated with anxiety combined with depression, and right hemisphere lesions to be associated with anxiety alone [43].

Furthermore, our results showing decreased FA values of the right uncinate fasciculus are compatible with previous studies reporting acute and chronic stress to result in greater involvement of the right hemisphere [44] and the effect of life stress on reduced FA of right uncinate fasciculus, and higher level of anxiety symptoms [45]. Such results of previous studies suggested anxiety and stress to be associated more with right hemispheric function compared to the left, and this may be why our results showed only significant results in the right uncinate fasciculus and not the left uncinate fasciculus.

Second, the degree of directionality of diffusion and the diffusivity in a tissue can be measured by FA and MD, respectively. Extending the knowledge from the evaluation of FA and MD, recent studies showed AD (presumably linked with directional diffusivity along the axon) and RD (presumably linked with diffusivity perpendicular to the axon) underpin the biological alterations such as axonal and myelin changes [46,47]. Our findings of increases in both RD and MD values of the right uncinate fasciculus suggest a possible contribution of demyelination to the AG-related reduction in the right uncinate fasciculus integrity in the PD group. Impairment of myelin could contribute to frontal lobe inhibition and emotional dysregulation related to AG. Furthermore, this trend of FA and RD values was observed in many other DTI studies [48,49].

Third, significant correlations were observed between the FA values of the uncinate fasciculus and the BDI-II, ASI-R total, and ASI-R cognitive sub-scale scores in PD ＋ AG. Correlations between FA values of the uncinate fasciculus and APPQ scores were observed only in PD − AG in our study. Although there was no significant difference in the prevalence of major depressive disorder between patients with (15.3%) and without AG (12.7%) in PD [50], agoraphobia is reported to be one of the risk factors for major depressive disorder [51]. Also, the importance of anxiety sensitivity has been emphasized as a predictor of agoraphobia symptoms [52,53]. We considered our results to be similar to the results of previous studies reporting the presence of agoraphobia to be associated with depression and anxiety vulnerability, as BDI-II assesses depression symptoms and ASI-R assesses anxiety sensitivity. Only APPQ scores were associated with FA values in PD − AG, with the APPQ assessing fear of activities that may elicit physical sensation in panic patients [21].

However, the correlations between FA values of the uncinate fasciculus and symptom severities were positive rather than negative. A little research is done to explain such a finding but we may speculate that decreased integrity of the uncinate fasciculus in PD ＋ AG may influence g-aminobutyric acid GABAnergic inhibitory transmission of the uncinate fasciculus [54]. Dysfunction of GABAnergic inhibitory transmission can influence the local axonal environment and mediate myelin dysfunction and overactivation, which may lead to increased FA values [55]. Patients with higher symptom severities may have further dysfunction in GABAnergic inhibitory transmission, hence present paradoxically increased FA values. Such paradoxic increases in FA values, due to possible disruption to the inhibitory GABAnergic system, have been reported in previous studies [56-58]. Further studies are needed to explain speculative assumption.

This study had several limitations. First, only females were included in this study. Although we purposely included only female subjects in our study due to the higher prevalence of AG in females compared to males, the fact that all subjects were females may be considered as a limitation in the context of generalizing our findings [59]. Second, we relied on a relatively small sample size. While our sample size was similar to that of recent neuroimaging studies, future studies including a larger sample size would be helpful in demonstrating confident results [60,61]. Third, 55 patients were consuming antidepressants during the study, and several studies have found a relationship between structural changes of the brain and antidepressant treatment [62]. However, our results remained statistically significant even after correcting for duration of medication as a covariate.

In conclusion, we provide evidence for decreased integrity of the uncinate fasciculus in PD ＋ AG as compared to PD − AG. Additionally, our results suggest an association between uncinate fasciculus integrity and depression and anxiety vulnerability in PD ＋ AG. Our study provides further neurobiological evidence on structural brain changes in AG.