Voxel-based asymmetry of the regional gray matter over the inferior temporal gyrus correlates with depressive symptoms in medicated patients with major depressive disorder

The number of patients suffering from major depressive disorder (MDD) is increasing worldwide. Imbalanced hemispherical brain activity may be an underlying factor of MDD; however, whether structural asymmetry also contributes to the symptoms experienced in MDD has been scarcely investigated. In this study, we aimed to examine cortical asymmetry in association with the severity of depressive and cognitive symptoms observed in MDD during stable medication. The association between the affective and cognitive symptoms and gray matter asymmetry was evaluated in 17 MDD patients using voxel-wise gray matter asymmetry analysis on high-resolution T1-weighted MR images. Asymmetry index values in the inferior temporal gyrus (ITG) correlated with the scores of the 17-item Hamilton Depression Rating Scale (HDRS), but no association was found with the Beck Hopelessness Scale, and performance on the 1-, 2and 3-back task. Our results indicate that the asymmetry of gray matter content in the ITG might be associated with higher depression severity. Our findings might help to better understand how structural changes contribute to depression severity in patients with MDD.


Introduction
Major depressive disorder (MDD) is a debilitating condition affecting mood, cognition, sleep, appetite, and libido (Kennedy 2008). The regulation of these functions involves several brain areas as well as brain circuitries, and consequently, structural abnormalities and functional disruptions have been reported across the entire brain in MDD (Drevets et al. 2008;Suh et al., 2020;Ye et al., 2016). The frontal areas have been implied to show imbalanced brain activity and metabolism with right hemispheric hyperactivity compared to the relative hypoactivity of the left side (for a review see Hecht, 2010). The presence of imbalance has also been supported by EEG (electroencephalography) measures in the alpha frequency range (Grunewald et al. 2018). Moreover, frontal alpha asymmetry seems to correlate with depressive symptoms including symptom severity, psychomotor retardation, and suicidal behavior (Cantisani et al., 2015;Diego et al., 2001;Park et al., 2019). Whether gray matter structural asymmetries are also present in MDD has only been investigated directly in four studies (see Table 1 for a summary), suggesting abnormal frontal and temporal asymmetries (Kumar et al., 2000;Liu et al., 2016;Zuo et al., 2019) or unilateral volume differences compared to healthy controls van Tol et al., 2014

rightlateralization
A recent large-scale asymmetry analysis conducted by the ENIGMA Consortium comparing gyral-based regions of interest (ROIs), however, has not supported the presence of any cortical asymmetries as compared to healthy participants (de Kovel et al., 2019;Kong et al., 2020). Although this study has high statistical power and sample size, its results do not exclude the possibility that subregions within the examined ROIs and/or subgroups of MDD patients show structural asymmetry that might contribute to the clinical (either cognitive or affective) manifestations of MDD. Convergence of asymmetric cortical volumes has been suggested when analyzing clinical subgroups of patients (distinguished based on medication status or the presence of comorbid disorders) (Gray et al. 2020). Furthermore, voxel-based analyses may be performed to test if certain gray matter subregions (instead of predefined ROIs) show abnormal asymmetry.
Structural asymmetries in MDD have been linked to depressive symptoms, indicating their potential relevance in the pathomechanism of MDD. For instance, higher structural asymmetry of the dorsolateral prefrontal cortex (DLPFC) correlated with self-reported depressive symptoms in subclinical and major depression . Strikingly, associations between structural asymmetry and other core aspects of MDD (e.g., cognitive symptoms or suicidal ideations) have not been explored previously, despite cognitive deficit being a prominent feature of MDD (Gärtner et al., 2018;Mohn & Rund, 2016).
In the present study, we aimed to investigate voxel-based cortical asymmetry in association with the severity of both depressive and cognitive symptoms observed in MDD patients with stable medication status. The interaction of voxel-based asymmetry with depressive symptoms, hopelessness, attention, and working memory were examined to identify brain areas with abnormal cortical asymmetry linked to potential residual symptoms. The voxel-based approach we are using in the current investigation provides higher spatial resolution and does not suffer from the inherent bias of ROIbased methods (Astrakas and Argyropoulou 2010). A protocol for voxel-wise gray matter asymmetry calculation was performed to investigate the association between cortical asymmetries and the cognitive symptoms and MDD severity (Kurth et al. 2015).

Participants
Seventeen patients with MDD diagnosed by expert clinicians were recruited from the Department of Psychiatry of the Albert Szent-Györgyi Health Centre, University of Szeged (see Table 2 for sample characteristics). The medication status of all patients was kept stable at least two weeks prior to the neuroimaging recordings.
Antidepressant treatment was combined with benzodiazepines in the case of 8 participants, 4 patients received antidepressants only, and 5 participants took benzodiazepines without antidepressant medication. The antidepressants were classified as follows: serotonin and norepinephrine reuptake inhibitors (n = 4), selective serotonin reuptake inhibitors (n = 2), atypical antidepressants (n = 2), tetracyclic antidepressants (n = 2), monoamine oxidase inhibitor (n = 1), and tricyclic antidepressant (n = 1). The study was approved by the Ethics Committee of the University of Szeged (Ref.No.: 165/2014). All participants gave written informed consent in accordance with the Declaration of Helsinki.

Assessment of affective and cognitive symptoms
To assess depression-related symptoms, a half-structured interview, the 17-item HDRS was administered. The HDRS comprises the assessment of several symptoms including affective state, suicidal thoughts, somatic symptoms, sleeping and eating behavior, and sexual symptoms (Hamilton, 1986). The HDRS is a frequently used tool in clinical research (Behera et al., 2017). The short, 4-item version of the Beck Hopelessness Scale (BHS) was also administered to further explore negative expectations and hopelessness (Beck & Steer, 1988).
Three levels of the n-back task (1-back, 2-back, and 3-back) were performed in ascending order by each participant. The task was presented using PsychoPy (version: presented consecutively on the screen for 1500 ms with 500-ms-long interstimulus intervals. During the 1-back task, participants were asked to press the spacebar if the target stimulus presented on the screen was the same as the previous one. During the 2back and 3-back tasks, the spacebar had to be pressed if the second or third letter preceding the target stimulus was identical with the letter presented, respectively. At each level, a total of 100 trials were completed, and 20% of all presented stimuli were target stimuli to which participants were expected to respond. Sensitivity index (d' score) was calculated for all levels based on the signal detection theory by subtracting the false alarm rate from the hit rate, both expressed in z-scores: d′ = Z(hit rate) − Z(false alarm rate) (Haatveit et al. 2010). d' scores of the 1-back task were interpreted in the attention domain, while the results of the 2-back and 3-back tasks were interpreted in the working memory domain (Martin et al., 2016).

Data acquisition
Magnetic resonance imaging (MRI) was performed using a 1. voxel, a step-by-step guideline published by Kurth and colleagues was followed (Kurth et al. 2015). All non-brain parts were removed from the T1-weighted images and then the brain was segmented into gray matter, white matter, and cerebrospinal fluid. All gray and white matter segments were flipped along the midline prior to creating the symmetric Diffeomorphic Anatomical Registration using Exponentiated Lie Algebra (DARTEL) template from the original and flipped gray and white matter segments (Ashburner 2007). Afterwards, the images were registered to the mean DARTEL template and averaged, and a binarized right hemisphere mask was created in the symmetric template space to limit statistical calculations to the right hemisphere. Then, we calculated an asymmetry index based on the following equation:

Results
The result of voxel-based gray matter asymmetry analysis was visualized as an overlay on the MNI152 2 mm standard brain. Gray matter asymmetry analysis yielded a significant negative correlation between the gray matter AI values of the inferior temporal gyrus (ITG) and the individual HDRS scores (Figure 1), as determined by the MNI152 standard space coordinates (x=18, y=55, z=17). No significant association was found between the gray matter AI values and the BHS and d' scores derived from n-back task. The higher the HDRS scores, the lower or more negative the AI values were, as indicated by the AI values extracted from the significant cluster (R=-0.879, p0.001) (Figure 2).
Thus, in patients with more severe depressive symptoms, a higher leftward asymmetry in the ITG (i.e., lower gray matter content in the right hemisphere, higher gray matter content in the left hemisphere) can be observed.

Discussion
Brain abnormalities in MDD have been widely investigated to better understand the neural background of this debilitating disorder. The aim of the present study was to examine the association of affective and cognitive symptoms with cortical asymmetries in MDD patients under stable medication. Voxel-based cortical asymmetry calculations were carried out indicating a leftward asymmetry within the ITG in patients with more severe depressive symptoms (i.e., the higher the depression severity, the less gray matter volume in the right ITG as compared to the left homologue area). No similar association or any asymmetry within other brain regions was found with respect to hopelessness and cognitive measures. While previous ROI-based studies have reported no abnormal asymmetry or results limited to specific MDD subgroups (de Kovel et al., 2019;Kong et al., 2020), our analysis showed a cluster of voxel-wise asymmetries within the ITG that were in association with depressive symptoms present in medicated MDD patients. Such small changes might be overlooked if imaging data is analyzed using predefined ROIs.
The ITG has previously been reported to show an asymmetric reduction of gray matter volume , especially in medication naïve (Guo et al. 2014a, Gray et al. 2020) and first-episode MDD patients (Peng et al. 2011). Moreover, the volume of the ITG has had a positive correlation with the 17-item HDRS scores and with longer reaction times on an attention task (Li et al. 2010). The ITG has also been found to be affected in other disorders, such as schizophrenia (Onitsuka et al. 2004) and bipolar affective disorder (Gong et al. 2019). Summarizing all these results, the ITG appears to be affected in several disorders including MDD, which casts light on its potential importance in the pathomechanism of these conditions.
The ITG seems to be engaged in cognitive processes such as semantic processing and concept retrieval (Faber et al. 2014), but also contributes to social cognition (Gallagher & Frith, 2003), self-referential processing (Herold et al. 2016) and the processing of affective stimuli (Hu et al. 2017). On a network-level, the ITG seems to be involved in the extended default mode network (DMN) (Allen & Williams 2011, Guo et al. 2014b, which is reported to show an imbalance in MDD (Hamilton et al. 2015). More specifically, an abnormally decreased network homogeneity (Guo et al. 2014b) and reduced amplitude of low-frequency fluctuation (indicating the absolute intensity of spontaneous brain activity) of the right ITG has been observed within the DMN as compared to healthy individuals (Guo et al. 2014b). Our results linking the structural features of the ITG to the clinical symptoms of depression supplement the findings on functional disruptions and suggest that the gray matter content of the ITG might contribute to a larger-scale network-level abnormality relating to negative affect and rumination.
In light of these results, the characteristics of the ITG in MDD should be explored more thoroughly, such as the extent of its white matter network and its relationship with functional or behavioral factors associated with depression. Furthermore, ITG may be suitable as a target for non-invasive brain stimulation (e.g., transcranial magnetic stimulation, transcranial electrical stimulation). These interventions have been utilized to mitigate the affective and cognitive symptoms in MDD (Iimori et al. 2019, Holczer et al. 2021 and to supplement the effects of the ongoing medication (Berlim, Van den Eynde et al. 2013, Plewnia et al. 2014. By stimulating the ITG, the residual symptoms might be further mitigated; however, this requires empirical support. Limitations of the present study include sample size and imbalanced sex ratio. However, we corrected for the effects of sex and age by performing a covariate analysis. From the present study, we cannot clearly determine if and which pharmaceutical preparation has an effect on the structural changes of the ITG. In our sample, different types of medication (e.g., selective serotonin reuptake inhibitors, serotonin and norepinephrine reuptake inhibitors) were administered to the patients. Thus, future longitudinal-design studies should explore the antidepressant effects on the gray matter volume of the ITG.
An additional limitation may be the validity of HDRS-17 among patients undergoing different drug therapies as different therapies may induce divergent responses in terms of the final score and its subcomponents. It would be necessary to take additional samples assessed with another psychometric tool measuring depression (e.g., the Beck Depression Inventory). However, antidepressants regardless of the exact type seem to decrease depressive symptoms, and their network-level effect mainly act on the connection strength between symptom-domains (Berlim et al., 2020). Since our results are cross-sectional, longitudinal studies are needed to examine whether depression severity is consistently associated with the structural changes of the ITG.

Conclusion
In the present study, we examined the association of affective and cognitive symptoms with cortical asymmetries in medicated MDD patients and found that the leftward shift of the ITG was associated with higher depression severity. The understanding of cerebral pathology is essential and may facilitate more targeted approaches for the prevention and management of MDD.