Potential correlations between asymmetric disruption of functional connectivity and metabolism in major depressive disorder

,


Introduction
Major Depressive Disorder (MDD) is a prevalent and intricate psychiatric condition marked by enduring sadness, diminished energy, slowed cognition, and reduced voluntary activity.In severe instances, it may escalate to contemplation of suicide or actual attempts (Yuan et al., 2020).Depression globally affects around 4.4 % of the population, impacting over 300 million individuals worldwide.Projections indicate that by 2030, MDD is anticipated to emerge as the foremost global burden of disease (Mathers and Loncar, 2006).The COVID-19 pandemic has exacerbated the incidence of severe depression due to factors such as prolonged isolation, fear of infection, lack of information, feelings of shame, or economic loss (Bueno-Notivol et al., 2021).Despite this, less than 10 % of individuals afflicted with MDD receive efficacious treatment annually (Kessler et al., 2003).Consequently, enhancing our comprehension of the pathogenesis of MDD holds the potential to ameliorate treatment outcomes and alleviate the societal burden.
The exploration of the relationship between hemispheric asymmetries and the onset of psychiatric disorders, including MDD, has captivated researchers for an extended duration.Studies demonstrate that hemispheric asymmetry plays a contributing role in various cognitive and emotional processes, encompassing language, attention, perception, and emotional regulation (Raimondo et al., 2021).Moreover, they correspond to functional specialization in domains such as cognition, language, emotion, behavior, vision, and hearing (Hellige, 1990).These disparities could lead to changes in specialization function, resulting in cognitive, emotional, and behavioral impairments, particularly evident in MDD (Bruder et al., 2012).However, the debate continues regarding the pattern of lateralization in emotional processing.Modern studies suggest that functional lateralization in emotional processes may be restricted to specific brain regions exclusively (Sheline, 2003).Bruder et al. conducted a review examining evidence related to electroencephalography and functional MRI asymmetry.Their findings indicated that both alpha rhythm asymmetry (FAA) and lateralization information in the brain were influenced by factors such as gender, age, and comorbidity (Bruder et al., 2017).Research has established links between asymmetries and changes in lateralization, particularly in response to emotional stimuli (Forscher and Li, 2012).
Resting-state functional magnetic resonance imaging (rs-fMRI) is a valuable tool for investigating hemispheric asymmetry in connectivity, offering valuable insights into the integration of functional neural networks across diverse brain regions.Studies suggest that MDD may stem from aberrant FC within brain networks (Broyd et al., 2009).Prior investigations employing voxel-mirrored homotopic connectivity (VMHC) and the parameter of asymmetry (PAS) reveal a robust association between hemispheric functional asymmetry and mood/cognitive impairments in MDD.These measures hold promise as valuable diagnostic markers for assessing resting-state brain activity (Ding et al., 2021;Fu et al., 2021;Zhang et al., 2023).The PAS is an innovative quantitative voxel metric derived from computing functional connectivity (FC) differences within and between hemispheres.This computation effectively illustrates both intra-and inter-hemispheric asymmetries (Zhu et al., 2018).The research further highlights that individuals with MDD exhibiting comorbid gastrointestinal symptoms tended to display higher PAS scores, indicative of lower hemispheric specialization.This trend was observed in various areas, including the default mode network, control network, attention network, cerebellum, and visual cortex.Additionally, the study emphasizes that demographics and clinical factors significantly contribute to these observed abnormalities (Fu et al., 2021).
The etiology of MDD is considered multifactorial, with connections identified between MDD and thyroid dysfunction, as well as abnormal lipid metabolism (Almeida-Montes et al., 2000;Hussain et al., 2020;Natesan and Kim, 2021;Peng and Li, 2017;Petrlová et al., 2004;Trifu et al., 2020).Thyroid-stimulating hormone (TSH), originating from the pituitary gland, governs thyroid function.Acting as vital metabolic regulators, thyroid hormones influence various physiological processes, including metabolism, growth, and development.Changes in TSH levels, a pivotal hormone facilitating the synthesis and release of thyroid hormones, can impact the thyroid hormone levels in the brain, thereby influencing its function.An investigation revealed a correlation between TSH levels and cognitive function along with neuropsychological symptoms (Schraml et al., 2011).Fountoulakis et al. identified an elevated occurrence of thyroid dysfunction in individuals with MDD, noting a higher prevalence of abnormal TSH and thyroxine levels in this cohort compared to the general population (Fountoulakis et al., 2006).Furthermore, research has shown a positive correlation between serum TSH levels and changes in the gray matter volume pattern of the middle frontal gyrus (MFG) and executive functional subregions (Zhao et al., 2021).However, there has been no investigation into whether TSH levels are related to PAS, which has sparked our interest.
Cholesterol (CHOL), a crucial component essential for brain function due to its prevalence in cell membranes and its imperative role in synapse formation and maturation (van Vliet, 2012), is potentially associated with MDD (Dietschy and Turley, 2004b).Various studies propose that reduced serum CHOL levels may impact both brain function and structure, particularly concerning symptoms associated with MDD (Ergün et al., 2004;Jia et al., 2020;Rabe-Jabłońska and Poprawska, 2000;Shin et al., 2008;Wagner et al., 2019).Additionally, diminished levels of serum high-density lipoprotein cholesterol (HDL-C) have been observed to inversely correlate with the severity of depressive symptoms (Wang and Shen, 2023).Nevertheless, conflicting perspectives have been presented in certain studies, suggesting a lack of association between CHOL levels and MDD (Almeida-Montes et al., 2000;Brown et al., 1994;Dietschy and Turley, 2004b).Previous research has found that elevated serum CHOL levels are associated with disruption in the FC of the salience network (SN) in non-demented elderly individuals (Zhang et al., 2016).However, there is currently no research indicating whether MDD is associated with changes in PAS values.
While studies have demonstrated an association between TSH, CHOL, and MDD, the precise mechanism governing this relationship requires further elucidation.In this investigation, the PAS method was employed to scrutinize resting-state FC (Rs-FC) asymmetry in individuals with MDD.Additionally, an exploration was conducted to determine if abnormal PAS scores exhibited correlations with blood lipids and thyroid hormone levels in these patients.

Participant
The study included 46 individuals diagnosed with MDD from the Third People's Hospital of Foshan.The diagnosis was meticulously confirmed by two qualified psychiatrists using the Diagnostic and Statistical Manual of Mental Disorders-5 (DSM-5) patient version (Diagnostic and statistical manual of mental disorders: DSM-5™, 5th ed, 2013).Prior to enrollment, relapsed patients independently ceased their medication for a minimum of two weeks by themselves leading to relapse this time.The cohort comprised 34 individuals with first-episode MDD and 12 with recurrent MDD, featuring a gender distribution of 17 males and 29 females, all within the age range of 18-60 years.To ensure comparability, 44 healthy controls, matched for both gender and years of education, were recruited from the local community.Notably, all participants were right-handed.
All subjects were uniformly excluded based on shared exclusion criteria: (1) individuals with concurrent neuropsychiatric or severe physical illnesses; (2) those with a history of substance and alcohol abuse; (3) individuals with significant physical impairment hindering the completion of the follow-up study; (4) those with contraindications for MRI examinations or whose head movement during rs-fMRI resulted in non-compliant images; (5) pregnant and lactating women.Each participant underwent a thorough assessment concurrent with their rs-fMRI scan, involving the completion of diverse tests and questionnaires, including the Hamilton Depression Inventory (HAMD-17) and Hamilton Anxiety Inventory (HAMA).Additionally, assessments of blood lipid and thyroid hormone levels were conducted to gauge physiological parameters.
All participants signed an informed consent at the time of enrollment, and this study was approved by the Ethics Committee of Foshan Third People's Hospital (Ethics approval number: FSSY-LS202106).

Data acquisition
FMRI data were acquired using a GE 3.0 T Signa Pioneer scanner.Participants were instructed to lie flat with closed eyes, maintaining a quiet and awake state.To minimize scanner noise and head motion, soft earplugs and foam pads were employed.Rs-fMRI scans utilized a gradient echo-planar imaging (EPI) sequence with the following parameters: repetition time/echo time = 2000 ms/30 ms, 33 slices, 64 × 64 matrix, 90 • flip angle, 22 cm field of view, 4 mm slice thickness, no slice gap, and 250 volumes.Each scan duration was 480 s per participant.

Data preprocessing
The rs-fMRI data underwent preprocessing through the Data Y. Yang et al.Processing Assistant for Resting-State fMRI (DPARSF) software (5.1) in MATLAB (Chao-Gan and Yu-Feng, 2010).Initially, correction for slice timing and head motion was performed.To maintain imaging data quality, strict criteria were applied, mandating a maximum displacement of 2 mm and an angular motion of 2 • in the x, y, or z direction for all subjects.Subsequently, the functional images were resampled to a voxel size of 3 × 3 × 3 mm 3 and normalized using the methodology advocated by Chao-Gan and Yu-Feng.The preprocessed images underwent smoothing using an isotropic Gaussian kernel with a full width at half-maximum of 8 mm.Subsequently, a linear trend removal and bandpass filter (0.01-0.08 Hz) were applied to the images (Song et al., 2011).Following this, linear regression was employed to eliminate various spurious covariates and their temporal derivatives from the data.These covariates included the signal from a ventricular region of interest, the signal from a region centered in the white matter, and 24 head motion parameters determined through rigid body correction (Fox et al., 2005).
To address the impact of head movement, we computed the Framewise Displacement (FD) following the methodology of a prior study (Jia et al., 2020) and integrated it into our analysis (Dosenbach et al., 2017).The averaged FD served as an unconnected variable in group analyses.Additionally, we implemented scrubbing, eliminating time points where FD exceeded 0.2 mm, as a stringent motion control measure.Notably, we refrained from eliminating the global signal, as its removal remains a topic of ongoing debate in rs-fMRI analysis.

Data analysis and calculation of the PAS values
As in previous studies (Ding et al., 2021;Fu et al., 2021;Zhu et al., 2018), the PAS score was calculated using the following formula.PAS = FC inter − FC intra FC inter and FC intra denote functional connections between and within brain hemispheres, respectively.FC inter represents the average functional connection strength between voxels in the opposite hemisphere for a given voxel in a specific space (obtained through Fisher Z transform).Conversely, FC intra signifies the average functional connection strength between voxels in the ipsilateral hemisphere for a given voxel in the same space.To maintain reliability, negative correlations are excluded from the calculation (Wang et al., 2011).A threshold of r > 0.2 is established to filter out weak correlations potentially attributable to signal noise (Wang et al., 2014).

Detection of blood lipids
Blood lipids were assessed using the Mindray CL6000i Shenzhen, a fully automated biochemical detection analyzer, following standardized procedures.Peripheral blood samples of 5 ml were collected from each participant, allowed to clot, and then separated from blood cells through centrifugation.The serum was subsequently transferred to clean and sterile tubes, with assay reagents prepared according to the manufacturer's guidelines.The CL6000i employed an endpoint method, catalyzing a reaction between cholesterol esterase, cholesterol oxidase, and peroxidase with cholesterol in the sample, leading to the formation of hydrogen peroxide.This compound then reacted with a chromogenic substrate, generating a colored compound quantified by the CL6000i.CHOL levels were reported in distinct ranges: a suitable range of ≤ 5.6 mmol/l, a high-hazard critical value range of 5.6 ~ 6.6 mmol/l, and a high-hazard range of > 6.6 mmol/l.

The thyroid hormone detection
The Mindray CL6000i Shenzhen, an automated immunoassay analyzer, along with its corresponding reagents, was employed.Fasting peripheral blood samples of 10 ml were drawn from each participant at 8 am on the second day of enrollment.After centrifugation, the samples were placed on the CL6000i carrier shelf, and the barcodes on the sample tubes were scanned to link the samples to both the tester and patient information.These samples were then introduced into designated reagent trays within the CL6000i, where the instrument executed automated reactions, encompassing sample handling, enzyme labeling, and light emission, utilizing a chemiluminescence method to gauge the optical signal and calculate thyroid hormone content.The CL6000i documented the measurement outcomes.The laboratory kit's normal reference value ranged from 0.35 to 5.1 mIU/l, featuring a minimum detection limit of ≤ 0.005 μIU/ml and a functional sensitivity of ≤ 0.02 μIU/ml.

Statistical analyses
Demographic and clinical data underwent analysis through twosample t-tests and a Chi-square test (p < 0.05).Additional two-sample t-tests were performed to explore distinctions in PAS maps between patients and healthy controls, with age, sex, years of education, and mean FD considered as covariates.Significance was established at p < 0.05 using Gaussian Random Field theory (voxel significance: p < 0.001, cluster significance: p < 0.05) in REST, a software framework developed by Song et al. (2011).
Mean PAS scores were extracted from clusters identified as abnormal in group comparisons.Pearson's or Spearman's correlation analyses (p < 0.05) were conducted to evaluate the correlation between mean PAS values in regions exhibiting group differences and CHOL as well as TSH levels.The significance threshold was established at p < 0.05, Bonferroni corrected.

Characteristics of the subjects
Scanning data from 6 participants (2 HCs, 4 MDD) were discarded due to excessive head movements.Therefore, 42 patients, and 42 HCs were finally enrolled.Comparing the demographic data from the two groups, there were significant differences in age (p < 0.05) between the MDD and the HC group, no significant differences in gender, years of education, TG, CHOL, HDL, LDL, fasting blood glucose (FBG) and free thyroxine (FT4) level (p > 0.05).The patients show significant differences in HAMD (p < 0.001), HAMA (p < 0.001), TSH3UL (p < 0.01), and FT3 (p < 0.01) between the two groups (Table 1).

PAS difference between MDD patients and HCs
Table 2 and Fig. 1 showed that patients with MDD had significantly lower PAS score values in the left inferior frontal gyrus (T-value = − 4.4552, p-value = 0.000012) and higher in Bi-PCC)/precuneus (Tvalue = 3.6528, p-value = 0.000089) than HCs.
The significantly negative correlations were observed between abnormal PAS scores in the left IFG and CHOL values (r = − 0.335, p = 0.030), and TSH3UL values (r = − 0.310, p = 0.046) in the patients with MDD, and the correlations between abnormal PAS was no longer significant after the Bonferroni correction.IFG = Inferior Frontal Gyrus; PAS = parameter of asymmetry; CHOL = total cholesterol; TSH3UL = third-generation ultra-sensitive thyroidstimulating hormone; MDD = Major depressive disorder.

Discussion
This study aimed to investigate the relationship between MDD and asymmetrical abnormalities in Rs-FC by calculating PAS scores.Our results revealed significant differences in PAS scores between MDD patients and HCs, particularly in the left IFG and Bi-PCC regions.A decrease in the PAS score of the left IFG indicated reduced hemispheric asymmetry, whereas an increase in the PAS score of the Bi-PCC suggested enhanced hemispheric asymmetry.These findings emphasize the association between irregular hemispheric FC asymmetry in the left IFG and Bi-PCC regions and MDD.Notably, the PAS score of the left IFG negatively correlated with both TSH and total CHOL levels.However, these correlations lose significance after the Bonferroni correction.
Extensive research underscores the connection between MDD and anomalies in both structure and function within the fronto-limbic system (Bell-McGinty et al., 2002;Biaggi et al., 2016;MacMaster et al., 2008;Murphy et al., 2007).This intricate network holds pivotal significance in the regulation of emotions, motivation, and the processing of reward-related stimuli (Sheline et al., 2010;Tekin and Cummings, 2002).Encompassing key brain regions like the prefrontal cortex (PFC), amygdala, hippocampus, and striatum, aberrations within this system align with hallmark symptoms of MDD, including enduring sadness, diminished interest in pleasurable activities, cognitive impairments, and disruptions in sleep patterns (Sheline, 2003;Taki et al., 2005).
The prefrontal and cingulate cortex play pivotal roles in cognitive control and emotional regulation, and disruptions in neural circuits within these regions can significantly influence cognition (Koenigs et al., 2008).Individuals with hypofunction in the prefrontal cortex often manifest cognitive and emotional impairments (Henriques and Davidson, 1991).Consequently, it is typical for MDD patients to demonstrate reduced executive control abilities, including compromised memory and attention (Jenkins et al., 2018).The left IFG is frequently designated as a 'multifunctional' region integral to emotional, behavioral, and cognitive control.It assumes critical functions in semantic understanding, production, and inhibition (Gabrieli et al., 1998).Multiple studies propose that dysfunctional semantic perception, potentially signaling compromised functionality in the left IFG, could be linked to depressive symptoms, including low mood, disinterest, and social withdrawal (Padmanabhan et al., 2019).A meta-analysis study of the IFG indicates a strong leftward bias in semantic perception, while emotional perception did not show lateralization (Belyk et al., 2017).Therefore, the exacerbated leftward lateralization of the IFG, manifested by decreased PAS scores in the left IFG, may suggest dysfunction in emotional perception when perceiving emotional semantics.Our previous research has also confirmed this assertion (Fu et al., 2021).
The Posterior Cingulate Cortex (PCC), situated within Brodmann's area 23/31, establishes numerous structural connections with various brain regions and forms an integral part of the default mode network (Hagmann et al., 2008).Particularly noteworthy, the PCC stands as one of the most active regions in the human brain during periods of rest, playing a crucial role in recalling episodic memories, processing selfrelevant information, and regulating emotions (Cheng et al., 2018;Gabrieli et al., 1998;Leech and Smallwood, 2019).Research, exemplified by the work of Maddock et al., suggests that aberrant FC in the PCC may be associated with feelings of hopelessness commonly observed in depression (Maddock et al., 2001).This implies that asymmetrical abnormalities in the connectivity of these brain regions could potentially contribute to the onset and progression of depressive symptoms.However, diverse perspectives on these findings also exist (Gray et al., 2020;Horato et al., 2022;Kong et al., 2018).
This study revealed that TSH levels in individuals with MDD were notably lower compared to those in HCs, consistent with findings from several prior investigations (Hage and Azar, 2012;Joffe and Levitt, 2008;Williams et al., 2009).Our study found a negative correlation between PAS values in the left IFG and TSH levels, although these correlations lost significance after the Bonferroni correction.This may primarily be due to our relatively small sample size, resulting in the loss  of significance after the correction.Indeed, the relationship between changes in TSH levels and altered brain connectivity is likely multifaceted, influenced by diverse biological and environmental factors.TSH has an impact on FC which is typically mediated through thyroxine, suggesting a possibly indirect relationship (Bernal, 2007;Zoeller and Rovet, 2004).Alterations in FC may encompass diverse biological mechanisms, incorporating neurotransmitter levels, neuronal activity, and synaptic function, highlighting that a singular biomarker, like TSH, may not comprehensively encapsulate the spectrum of modified brain FC (Bauer et al., 2008).A post-mortem study conducted by Naicker et al. demonstrated the expression of TSH receptors in the prefrontal cortex (Naicker and Naidoo, 2018), suggesting that serum TSH levels may influence the IFG through mechanisms mediated by TSH receptors.Plasma CHOL necessitates binding to apolipoprotein to form lipoprotein, facilitating its crossing of the blood-brain barrier.The effectiveness of CHOL is consequently influenced by lipoproteins of diverse densities, including high-density lipoprotein (HDL) and low-density lipoprotein (LDL) (Müller et al., 2015).Our study revealed no significant disparity in serum CHOL levels between MDD patients and healthy controls, aligning with Huang's findings (Huang, 2005).Our study found a negative correlation between PAS values in the left IFG and CHOL levels, although these correlations lost significance after Bonferroni correction.Numerous studies have reported typical CHOL dysregulation in MDD patients, marked by decreased HDL, increased LDL, and elevated LDL/HDL ratios (Horsten et al., 1997;Olusi and Fido, 1996;Zhang et al., 2022).Research suggests that FC within the default mode network mediates the association between gray matter volume in the ventromedial prefrontal cortex (VMPFC) and serum HDL-C levels.Furthermore, the integrity of white matter in the genu of the corpus callosum mediates the link between serum HDL-C levels and the severity of depressive symptoms (Zhang et al., 2020).Normal lipid metabolism is crucial for myelin formation, synaptic plasticity, and receptor function in the central nervous system (Dietschy and Turley, 2004a;Muse et al., 2001).Abnormal CHOL levels may lead to changes in PAS values in the left IFG.
Several limitations must be acknowledged when interpreting the results of this study.Firstly, due to the relatively small sample size, it may lower the statistical power and the generalizability of the research results.Secondly, the cross-sectional design limits our ability to determine the causal relationship between brain asymmetry and severe depression.Additionally, we should consider the influence of medication status on the observed results because antidepressant treatment may confound measurements of brain connectivity.If possible, efforts should be made to recruit treatment-naive patients.

Conclusion
This study revealed that MDD patients exhibited reduced PAS scores in the fronto-limbic system, indicating disrupted Rs-FC asymmetry in these regions.This finding contributes to a further understanding of the neuroimaging mechanisms associated with MDD.Additionally, we observed uncorrected correlations between abnormal PAS scores and TSH and CHOL values.This suggests that changes in TSH and CHOL levels may lead to abnormal PAS scores in the left IFG, offering a new avenue for neurological research on MDD.
HAMD = Hamilton Depression Rating Scale; HAMA = Hamilton Anxiety Rating Scale; TG = Triglycerides; CHOL = Total Cholesterol; HDL = High-Density Lipoprotein; LDL = Low-Density Lipoprotein; FBG = Fasting Blood Glucose; TSH3UL = Third-generation ultra-sensitive thyroid-stimulating hormone; FT3 = Free Triiodothyronine; FT4 = Free Thyroxine.a The p-values were obtained by two samples t-tests.b The p-value for sex distribution was obtained by a Chi-square test.

Fig. 1 .
Fig. 1.PAS significant difference between MDD and HCs.Red and blue denote increased and decreased PAS respectively, and the color bar represents the T values of the group analysis, patients with MDD had increased PAS values in the bilateral posterior cingulate cortex (Bi-PCC) and decreased increased PAS values in the left Inferior Frontal Gyrus (IFG).PAS = parameter of asymmetry; MDD = Major depressive disorder.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 1
Characteristics of participants.

Table 2
Brain regions with abnormal PAS in the patients.