Disrupted association between structural and functional coupling of the supplementary motor area and neurocognition in major depressive disorder

To the Editor: Major depressive disorder (MDD) is a common mood disorder that contributes considerably to disability worldwide. Abnormalities in either structural connectivity or functional connectivity in the brains of patients with MDD have been widely reported, which greatly extends our knowledge of the pathophysiology of MDD. The human brain is a complex network that consists of structurally and functionally linked regions that continuously communicate with each other. Previous studies have suggested that functional connectivity patterns were largely constrained by structural connectivity. The relationship between structural connectivity and functional connectivity is associated with brain development. For example, the structural-functional connectivity (SC-FC) relationship is strengthened with age, which is consistent with the finding that significant correlations along intra-hemispheric tracts are observed between structural connectivity and functional connectivity in adults but not in children. Moreover, the maturation of some functional connections in the default-mode network precedes that of structural connectivity. Based on the large number of investigations of the relationship between structural connectivity and functional connectivity, the concept of SC-FC coupling was proposed. SC-FC coupling may reflect the relationship between structural connectivity and functional connectivity in an individual's brain.[1] Cognitive impairment frequently occurs in patients with MDD and could be considered one of the core features of depression. Impairments in multiple aspects of neurocognitive functions, such as attention, visual memory, and working memory, have been reported in patients with MDD.[2] The SC-FC interaction in human brain is involved in the basis of cognitive function and behaviors.[1] Therefore, the application of SC-FC coupling in the study of MDD may provide novel insights into the pathophysiology underlying MDD and facilitate the identification of biological markers. A previous publication reported that SC-FC coupling of intra-hemispheric connections was significantly decreased and positively correlated with disease severity in patients with MDD. However, the conventional SC-FC coupling metric represents the global and whole-brain correlation between structural connectivity and functional connectivity, ignoring regional properties of SC-FC coupling.[3] A recent study developed an algorithm for the calculation of regional-specific SC-FC coupling based on regional structural and functional connectivity profiles,[4] which could provide detailed insights into SC-FC coupling in the brain. In this study, we compared regional-specific SC-FC coupling between patients with MDD and healthy controls (HCs) and explored the correlation between neurocognitive function and regional-specific SC-FC coupling. Han Chinese participants aged 16 to 55 years who met the criteria of MDD as specified in the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) were recruited from the Mental Health Center of West China Hospital of Sichuan University. The Chinese revised version of Wechsler Adult Intelligence Scale was used to calculate intelligence quotient for all participants. The Hamilton Depression Scale (HAMD) and Hamilton Anxiety Scale (HAMA) were used for evaluation of depressive symptoms and anxiety symptoms in patients with MDD, respectively. The patients were required to be anti-depressant- or anti-psychotic-naïve or anti-depressant- or anti-psychotic-free for at least 3 months. Healthy Han Chinese volunteers aged 16 to 55 years with Wechsler's intelligence test scores ≥90 were recruited via online advertisements. Individuals were excluded from this study if they met the following criteria: (1) having DSM-IV Axis I or II disorders; (2) having neurodegenerative diseases, serious endocrine diseases or metabolic disorders; (3) taking hormone medication; (4) being pregnant or breastfeeding. The study was approved by the Institutional Ethics Committee of West China Hospital, Sichuan University (No. 2020319). All participants provided written informed consent. The study procedures were in accordance with the Declaration of Helsinki. The participants were scanned by a 3-Tesla whole-body magnetic resonance (MR) scanner (Achieva, Philips, Netherlands) with an eight-channel phased-array head coil. Detailed information on image acquisition and preprocessing procedures is presented in the Supplemental File, https://links.lww.com/CM9/B454. We defined each region of interest (ROI) from the automated anatomical labeling atlas consisting of 90 brain regions as a network node. For construction of the functional network, the Pearson correlations were calculated on the averaged functional magnetic resonance imaging time series between all paired nodes, which was used to define the edge of the functional network. For construction of the structural network, all 90 ROIs were registered on the individual T1 image for brain segmentation, and the fractional anisotropy was used as the edge in the structural network. Regional SC-FC coupling was calculated according to the method reported in a previous publication.[4] Each column of a structural or functional connectivity matrix was defined as the regional connectivity of a specific region, which represents the connectivity from a single node (i.e., region) to all the other nodes in the structural or functional network of the participant. Then, the Fisher Z transformation of Pearson's correlation coefficients was calculated between the columns of the functional connectivity matrix and the non-zero counterparts of the white matter connectivity matrix. The regions with >50% of the data missing were excluded from the following analysis. The calculation generated a vector of metrics that represents the SC-FC coupling of corresponding brain regions for each participant. Neurocognitive function was assessed using the delayed matching to sample (DMS) subtest of computerized Cambridge Neurocognitive Test Automated Battery (CANTAB, Cambridge Cognition Ltd., Bottisham Cambridge, United Kingdom). Statistical analysis was performed using the R software (version 4.0.3, https://www.r-project.org/). Demographic characteristics are presented as median (interquartile range) or n (%). The chi-squared test was applied to compare gender differences between HCs and patients with MDD. The Wilcoxon rank sum test was used to compare age, years of education, and neurocognitive function between MDD patients and HCs. The difference in SC-FC coupling between MDD patients and HCs was compared using a linear model controlling for age, gender, and years of education. Partial Spearman's correlation, using age, gender, and education years as covariates, was employed to assess the relationship between SC-FC coupling and clinical characteristics or neurocognitive function. The test for differences in correlation coefficients between patients with MDD and HCs was performed according to the method described in https://www.ibm.com/support/pages/differences-between-correlations. The Benjamini–Hochberg procedure was used to control false discovery rates (FDRs) of multiple tests in comparing SC-FC coupling and correlation coefficients between patients with MDD and HCs. The FDR was set at the 0.05 level. q values (adjusted P values) <0.05 were considered statistically significant. A total of 135 HCs (55 males and 80 females) and 115 patients with MDD (42 males and 73 females) were included in this study. There was no statistically significant difference in median age (25.0 [12.5] years vs. 25.0 [8.0] years, Z = −0.75, P = 0.64) and gender (male: 36.5% [42/115] vs. 40.7% [55/135], χ2 = 0.30, P = 0.58) between the MDD and HC groups. The years of education of HCs were significantly higher than those of patients with MDD (16.0 [3.0] years vs. 15.0 [4.0] years, Z = −3.81, P < 0.01). The median of onset age, total disease duration, and current disease duration of patients with MDD were 22.0 (8.0) years, 24.0 (48.5) months, and 3.0 (6.0) months, respectively. The median HAMD and HAMA scores for patients with MDD were 21.0 (6.0) and 15.0 (8.5), respectively [Supplementary Table 1, https://links.lww.com/CM9/B454]. Mean percent correct in DMS simultaneous task (DMS_PC_A) was significantly lower in MDD patients than in HCs after correction for multiple testing (Z = −2.79, P = 0.01, q = 0.04). No significant differences in other DMS variables were found between the MDD and HC groups [Supplementary Table 2, https://links.lww.com/CM9/B454]. After matching the functional network to the non-zero structural network, the coupling metrics of 74 regions for each participant were generated. SC-FC coupling of the right supplementary motor area (SMA_R) in MDD patients was significantly lower than that in HCs (P = 6.1 × 10−4, q = 0.045) [Supplementary Figure 1, https://links.lww.com/CM9/B454]. SC-FC coupling of SMA_R was negatively and positively correlated with onset age (r = −0.25, P = 0.01, q = 0.02) and total duration (r = 0.26, P < 0.01, q = 0.02) of MDD after correction for multiple testing, respectively. There was no significant correlation between other clinical characteristics of MDD and SC-FC coupling of SMA_R [Supplementary Table 3, https://links.lww.com/CM9/B454]. There was a statistically significant correlation between SC-FC coupling of SMA_R and DMS_MCL_S (r = −0.32, P < 0.01, q = 0.01) in HCs after correction for multiple testing. There was a statistically significant difference in correlation coefficients between SC-FC coupling of SMA_R and DMS_MCL_S in patients with MDD and HCs (P < 0.01, q = 0.02) after correction for multiple testing [Table 1]. Table 1 - Correlation analysis between neurocognitive function and SMA_R in MDD patients and HCs. Correlation between neurocognitive function and SMA_R Difference of HC and MDD in correlation coefficients Items All HC MDD r values P values q values r values P values q values r values P values q values P values q values DMS_PEGC <−0.01 0.93 0.93 0.07 0.50 0.66 0.01 0.95 0.95 0.66 0.73 DMS_PEGE 0.03 0.69 0.93 −0.05 0.66 0.66 0.16 0.14 0.50 0.17 0.34 DMS_MCL −0.05 0.52 0.93 −0.21 0.04 0.18 0.11 0.29 0.50 0.02 0.12 DMS_MCL_A −0.02 0.77 0.93 −0.17 0.09 0.30 0.12 0.26 0.50 0.04 0.15 DMS_MCL_S −0.11 0.13 0.93 −0.32 <0.01 0.01 0.11 0.29 0.50 <0.01 0.02 DMS_PC −0.01 0.85 0.93 −0.10 0.35 0.59 −0.02 0.84 0.93 0.60 0.73 DMS_PC_A −0.04 0.58 0.93 −0.11 0.28 0.57 −0.06 0.55 0.68 0.74 0.74 DMS_PC_S 0.08 0.29 0.93 0.05 0.65 0.66 0.17 0.11 0.50 0.41 0.58 DMS: Delayed matching to sample; DMS_MCL: Mean latency in DMS; DMS_MCL_A: Mean latency in DMS delayed task; DMS_MCL_S: Mean latency in DMS simultaneous task; DMS_PC: Mean percent correct in DMS task; DMS_PC_A: Mean percent correct in DMS delayed task; DMS_PC_S: Mean percent correct in DMS simultaneous task; DMS_PEGC: DMS probability error given correct; DMS_PEGE: DMS probability error given error; HC: Healthy control; MDD: Major depressive disorder; SMA_R: Right supplementary motor area. SMA locates in the superior frontal gyrus and is traditionally considered a region for controlling movement. Recent findings suggested that SMA is involved in various cognitive domains, such as spatial processing, numerical cognition, and working memory. Moreover, a number of publications reported abnormalities in SMA in patients with MDD.[5] Considering that a wide range of cognitive impairments occur in patients with MDD, research on how SMA contributes to cognitive impairment in MDD would be helpful for a better understanding of the pathophysiology of MDD in the future. In this study, we found that SC-FC coupling of SMA_R was decreased in patients with MDD, positively correlated with the total duration of disease, and negatively correlated with the onset age of MDD. These findings appear counterintuitive. A similar correlation was reported previously between SC-FC coupling of intra-hemispheric connections and clinical severity in patients with MDD. In fact, counterintuitive network-behavioral relationships of the brain have been reported in multiple psychiatric disorders. The underlying mechanism is poorly understood, but it was hypothesized that compensatory effects of the brain in response to abnormal conditions may be involved.[3] There were some limitations in this study. First, the sample size was relatively small. Second, the long-term influence of anti-depressant medication, which may have affected both the structural and functional properties of the brain, was not excluded. Third, only DMS was used to evaluate the neurocognitive function of participants. Including more subtests of the CANTAB would provide more detailed information about the relationship between SC-FC coupling and neurocognitive impairments in patients with MDD in the future. In conclusion, our results suggested that the SC-FC decoupling of SMA_R and the disrupted association between the SC-FC coupling of SMA_R and neurocognition in patients with MDD might be involved in the pathophysiology of MDD. Funding This study was supported by grants from the Key Research and Development Program of Science and Technology Department of Sichuan Province (Nos. 22ZDYF1531 and 22ZDYF1696), the Program of Chengdu Science and Technology (No. 2021-YF05-00272-SN), the National Natural Science Foundation of China (No. 82001432), the China Postdoctoral Science Foundation (Nos. 2020TQ0213 and 2020M683319), the Open Project Program of the National Laboratory of Pattern Recognition (No. 202000034), and the West China Hospital Postdoctoral Science Foundation (No. 2020HXBH104). Conflicts of interest None.