Introduction
The latest nationwide survey of mental disorders showed the lifetime prevalence rates of major depressive disorder (MDD) at 3.4% in the general population of China [95% confidence interval (CI): 2.9–3.9%] (Huang et al., Reference Huang, Wang, Wang, Liu, Yu, Yan and Wu2019). Meanwhile, the World Health Organization has predicted MDD to become the first-leading cause of the global burden of disease in 2030 (Collins et al., Reference Collins, Patel, Joestl, March, Insel, Daar and Stein2011). Due to the inherently heterogeneous of MDD, the neurobiological mechanisms underlying the pathophysiology become quite complex. Melancholic features, a subtype of MDD in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) with highly homogeneities, characterized by lack of reactivity, psychomotor retardation, agitation, weight loss and inappropriate guilt (Duckworth, Reference Duckworth2015). Tondo et al., found a prevalence of DSM-5 melancholic features in MDD patients at intake measured with the 21-item version of the Hamilton Depression Rating Scale (21-HDRS) of 33.9% (Tondo, Vazquez, & Baldessarini, Reference Tondo, Vazquez and Baldessarini2020). Moreover, patients with the melancholic depressive subtype demonstrated a higher risk of suicidality, greater depression severity and worsen cognitive performance than those MDD patients without melancholic features (Caldieraro et al., Reference Caldieraro, Baeza, Pinheiro, Ribeiro, Parker and Fleck2013; Jeon et al., Reference Jeon, Peng, Chua, Srisurapanont, Fava, Bae and Hong2013; Roca et al., Reference Roca, Monzon, Vives, Lopez-Navarro, Garcia-Toro, Vicens and Gili2015). In addition to differences in psychological characteristics, these two major subtypes of depression may also differ in pathophysiological mechanisms including inflammatory, metabolic, hypothalamic-pituitary-adrenal (HPA) axis, hypothalamic-pituitary-thyroid (HPT) axis and brain function (Duval et al., Reference Duval, Mokrani, Monreal-Ortiz, Fattah, Champeval, Schulz and Macher2006; Lamers et al., Reference Lamers, Vogelzangs, Merikangas, de Jonge, Beekman and Penninx2013; Shan et al., Reference Shan, Cui, Liu, Li, Huang, Tang and Xie2021; Soriano-Mas et al., Reference Soriano-Mas, Hernandez-Ribas, Pujol, Urretavizcaya, Deus, Harrison and Cardoner2011; Vassilopoulou et al., Reference Vassilopoulou, Papathanasiou, Michopoulos, Boufidou, Oulis, Kelekis and Lykouras2013). Therefore, it is important to elucidate the pathogenic mechanisms underlying melancholic MDD, developing precise treatment strategies matching their unique biological characteristics and achieving the goal of ‘cognitive remission’.
Accumulating evidence has implicated that the structural changes of the hippocampus may be associated with the neural physiopathology of melancholic MDD. A recent study found a reduced left hippocampal volume in Met66 carriers in MDD patients with melancholic features when comprised of Val66 homozygotes (Cardoner et al., Reference Cardoner, Soria, Gratacos, Hernandez-Ribas, Pujol, Lopez-Sola and Soriano-Mas2013), as well as the reduced hippocampal volumes in the older population with melancholic depression (Hickie et al., Reference Hickie, Naismith, Ward, Turner, Scott, Mitchell and Parker2005). However, the inconsistent results suggested that there were no significant differences in hippocampal volume between groups of melancholic and other subtype depressed participants (Greenberg, Payne, MacFall, Steffens, & Krishnan, Reference Greenberg, Payne, MacFall, Steffens and Krishnan2008; Rusch, Abercrombie, Oakes, Schaefer, & Davidson, Reference Rusch, Abercrombie, Oakes, Schaefer and Davidson2001; Vasilopoulou et al., Reference Vasilopoulou, Papathanasiou, Michopoulos, Boufidou, Oulis, Nikolaou and Lykouras2011; Vassilopoulou et al., Reference Vassilopoulou, Papathanasiou, Michopoulos, Boufidou, Oulis, Kelekis and Lykouras2013). This inconsistency could arise from most neuroimaging studies of MDD patients with or without melancholic features who have considered the hippocampus as a single homogeneous structure. Based on the cytoarchitectonic characteristics of the hippocampal, Fan and colleagues suggested that it can be divided into the following major subregions: rostral and caudal hippocampus nuclei (Fan et al., Reference Fan, Li, Zhuo, Zhang, Wang, Chen and Jiang2016). The rostral hippocampus is closely linked to general memory processes including both learning, memory, encoding and retrieval, meanwhile, the caudal hippocampus is implicated more in spatial processing (Carr, Rissman, & Wagner, Reference Carr, Rissman and Wagner2010; Robinson et al., Reference Robinson, Barron, Kirby, Bottenhorn, Hill, Murphy and Fox2015; Zeidman & Maguire, Reference Zeidman and Maguire2016). Abnormalities in hippocampus subregion-based networks or volume have been found in post-traumatic stress disorder (Lazarov, Zhu, Suarez-Jimenez, Rutherford, & Neria, Reference Lazarov, Zhu, Suarez-Jimenez, Rutherford and Neria2017; Malivoire, Girard, Patel, & Monson, Reference Malivoire, Girard, Patel and Monson2018; Suarez-Jimenez et al., Reference Suarez-Jimenez, Zhu, Lazarov, Mann, Schneier, Gerber and Markowitz2020), Alzheimer's disease (Bai et al., Reference Bai, Xie, Watson, Shi, Yuan, Wang and Zhang2011) and chronic stress population (Chen et al., Reference Chen, Wei, Han, Jin, Xu, Dong and Peng2019). However, very few studies investigate hippocampus subregion-based dysfunction in melancholic MDD. Due to the underlying functional differences between the anterior and posterior hippocampus, further studies are encouraged to investigate hippocampus dysfunction at a subregional level in melancholic MDD.
Recently, investigations of depression-related differences using resting-state functional connectivity (FC) have begun to emerge. Regarding static FC, previous studies have revealed the aberrant FC between the hippocampal subregions and cortical and subcortical regions or associated neural circuits in MDD (Cao et al., Reference Cao, Liu, Xu, Li, Gao, Sun and Zhang2012; Fateh et al., Reference Fateh, Long, Duan, Cui, Pang, Farooq and Chen2019). However, most static FC studies on MDD implicitly assumed that FC was stationary throughout the entire resting scan period. It has been shown that human brain connectivity is dynamic and associated with ongoing rhythmic activity over time rather than stationarity(Allen et al., Reference Allen, Damaraju, Plis, Erhardt, Eichele and Calhoun2014; Reinen et al., Reference Reinen, Chen, Hutchison, Yeo, Anderson, Sabuncu and Holmes2018). The dynamic functional connectivity (dFC) analysis could provide abundant information about the time-varying functional architecture of specific regions, and could be a powerful supplement to static FC (Han et al., Reference Han, Wu, Wang, Sun, Ding, Cao and Zhou2018; Kaiser et al., Reference Kaiser, Whitfield-Gabrieli, Dillon, Goer, Beltzer, Minkel and Pizzagalli2016). dFC is sensitive to behavioral performance and emotional measures, and also be a sensitive prognostic indicator of disease progression in neuropsychiatric disorders including Alzheimer's disease, depression, and schizophrenia (Greicius, Reference Greicius2008; Liao et al., Reference Liao, Li, Duan, Cui, Chen and Chen2018; Shirer, Ryali, Rykhlevskaia, Menon, & Greicius, Reference Shirer, Ryali, Rykhlevskaia, Menon and Greicius2012). Therefore, the investigation of dFC may provide a nuanced view of the disrupted brain communication in MDD and a better understanding of the pathological mechanisms underlying this disorder. Of note, previous studies revealed abnormal dFC variability between the medial prefrontal cortex (mPFC) and insular regions (Kaiser et al., Reference Kaiser, Whitfield-Gabrieli, Dillon, Goer, Beltzer, Minkel and Pizzagalli2016; Wang et al., Reference Wang, Wang, Huang, Jia, Zheng, Zhong and Huang2020), between the mPFC and posterior cingulate cortex (Wise et al., Reference Wise, Marwood, Perkins, Herane-Vives, Joules, Lythgoe and Arnone2017), between mPFC and parahippocampal gyrus (Kaiser et al., Reference Kaiser, Whitfield-Gabrieli, Dillon, Goer, Beltzer, Minkel and Pizzagalli2016), and between the default mode network (DMN) and central executive network (Demirtas et al., Reference Demirtas, Tornador, Falcon, Lopez-Sola, Hernandez-Ribas, Pujol and Deco2016), as well as the greater dFC strengths in the precentral gyrus (Pang et al., Reference Pang, Zhang, Cui, Yang, Lu, Chen and Chen2020). These results consistently suggested that the alterations of dFC between regions of the DMN and areas of the prefrontal cortex or insula or hippocampus are believed to play important roles in emotional regulation and also may underlie the neural mechanisms of MDD (Long et al., Reference Long, Cao, Yan, Chen, Li, Castellanos and Liu2020). However, we are only beginning to understand the precise anatomy of the hippocampus in humans, little is known about the abnormal dFC variability at its subregional level of melancholic MDD patients.
Cognitive impairment is acknowledged as a core feature of clinical manifestations of all MDD subtypes. Patients with melancholic MDD showed prominent cognitive deficits including verbal memory, executive function, visual learning, attention, working memory and processing speed (Bosaipo, Foss, Young, & Juruena, Reference Bosaipo, Foss, Young and Juruena2017; Lin et al., Reference Lin, Xu, Lu, Ouyang, Dang, Lorenzo-Seva and Lee2014; Withall, Harris, & Cumming, Reference Withall, Harris and Cumming2010; Zaninotto et al., Reference Zaninotto, Solmi, Veronese, Guglielmo, Ioime, Camardese and Serretti2016). Previous studies indicated that patients fail to regain full functional recovery even in a euthymic state, which may be partly attributed to cognitive deficits (Pan et al., Reference Pan, Park, Brietzke, Zuckerman, Rong, Mansur and McIntyre2019; Woo, Rosenblat, Kakar, Bahk, & McIntyre, Reference Woo, Rosenblat, Kakar, Bahk and McIntyre2016). The hippocampal formation is heterogeneous and consists of different subregions that are complexly interacted with diverse brain areas, which form the neuroanatomical network of emotion regulation and cognitive processing (Bremner, Vythilingam, Vermetten, Vaccarino, & Charney, Reference Bremner, Vythilingam, Vermetten, Vaccarino and Charney2004; Drevets, Reference Drevets2000; Fateh et al., Reference Fateh, Long, Duan, Cui, Pang, Farooq and Chen2019; Rive et al., Reference Rive, van Rooijen, Veltman, Phillips, Schene and Ruhe2013). For instance, reduced hippocampal volumes were associated with visual and verbal memory deficit in patients with melancholic depression (Hickie et al., Reference Hickie, Naismith, Ward, Turner, Scott, Mitchell and Parker2005), and executive dysfunction in MDD (Frodl et al., Reference Frodl, Schaub, Banac, Charypar, Jager, Kummler and Meisenzahl2006; Khan et al., Reference Khan, Ryali, Bhat, Prakash, Srivastava and Khanam2015). A previous study suggested that the longitudinal changes in FC between the left cornu ammonis of the hippocampus and posterior cingulate cortex/precuneus were positively correlated with cognitive impairment in remitted late-onset depression (Wang et al., Reference Wang, Yuan, Bai, Shu, You, Li and Zhang2015). And the less posterior-DMN-hippocampal connectivity was associated with higher cognitive reactivity and rumination in MDD (Figueroa et al., Reference Figueroa, Mocking, van Wingen, Martens, Ruhe and Schene2017). A recent study also found the abnormal resting-state FC of hippocampal subfields may be related to the impairment of working memory in MDD patients (Hao et al., Reference Hao, Zhong, Ma, Xu, Kong, Wu and Wang2020). Moreover, the FC between the rostral hippocampus and the inferior part of the lateral occipital cortex mediated the negative relationship between cortisol and visuospatial memory in healthy young adults (Hakamata et al., Reference Hakamata, Komi, Sato, Izawa, Mizukami, Moriguchi and Tagaya2019). Taken together, early works suggested that changes in FC of the hippocampus can be related to the cognitive deficits in depression, but much remains unknown about the neurocognitive significance of dFC. Brain dynamics reflect the neural system's functional capacity and the spontaneous fluctuations in moment-to-moment behavioral variability (Kucyi, Hove, Esterman, Hutchison, & Valera, Reference Kucyi, Hove, Esterman, Hutchison and Valera2017), and these fluctuations may involve in a wide range of cognitive processes and emotional regulation. Previous studies have indicated that decreased FC variability in the DMN is associated with slower processing speed and executive function impairment in bipolar disorder patients (Nguyen et al., Reference Nguyen, Kovacevic, Dev, Lu, Liu and Eyler2017). Unfortunately, little is known about the alerted dFC variability in hippocampal subregions that may underlie the melancholic MDD-relevant cognitive impairment.
To address these questions, we collected several resting-state functional magnetic resonance imaging (rs-fMRI) data from 42 unmedicated melancholic MDD patients and 55 matched controls to detected the dFC alterations in hippocampal subregions in the present study. Meanwhile, a cognitive assessment was conducted using the Chinese version of the Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) Consensus Cognitive Battery (MCCB). We hypothesized that melancholic MDD patients would exhibit an abnormal dFC variability of the hippocampal subregions, and hope to probe the neurobiological signature of this refined depression subtype. And our second objective was to explore the association between the abnormal dFC variability and the cognitive performance of this disorder.
Materials and methods
Participants
A total of 55 right-handed, unmedicated, melancholic features MDD patients between the ages of 17 and 35 years from the psychiatry department of First Affiliated Hospital of Jinan University, Guangzhou, China enrolled in this cross-sectional trial. Two experienced psychiatrists (YJ and SZ, with 23 years and 6 years of experience in clinical psychiatry, respectively) followed the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) diagnostic criteria for the presence of melancholic features in major depressive episodes of MDD, based on eight items selected from the 24-item Hamilton Depression Rating Scale (24-item HDRS). In accord with DSM-5 criteria, all patients met the criteria for MDD with melancholic features, which required (1) pervasive anhedonia and/or nonreactive mood and (2) three (or more) of the following: characteristic depressive mood, regularly worse in the morning, early morning awakening, marked psychomotor agitation or retardation, significant anorexia or weight loss, and excessive or inappropriate guilt.
All subjects included in comparisons with melancholic features had intake 24-item HDRS total scores of ⩾ 20 (moderate-severe depression) and the Young Mania Rating Scale (YMRS) total score <7 was able to participate. Participants were excluded if they met one of the following criteria: (1) other serious psychiatric disorders and symptoms (with the exception of MDD and anxiety disorders/symptoms); (2) a history of the use of any psychotropic medication, psychotherapy, or electroconvulsive therapy; (3) a history of neurological or organic brain disorder; (4) a history of alcohol/substance abuse or dependence; and (5) any physical illness demonstrated by personal history or clinical or laboratory examinations, pregnancy, or postpartum depression. Finally, 10 patients were excluded due to the following reasons: other psychiatric disorders rather than melancholic MDD (n = 4) based on a Chinese version of the Structured Clinical Interview for DSM-IV (SCID), confirmed medical diseases (n = 1), inability to comprehend consent procedures or refusal to provide consent forms (n = 2), and later switch to bipolar disorder patients in the 12-month longitudinal follow-up (n = 3).
Besides, 55 right-handed volunteers who participated as healthy controls (HCs) were recruited from Jinan University and the community. They were carefully screened through a diagnostic interview, the Structured Clinical Interview for DSM-IV Nonpatient Edition (SCID-NP), to rule out the presence of current or past psychiatric illness in self or first-degree relatives or past substance abuse/dependence.
We employed the 24-item HDRS and YMRS to obtain a comprehensive measure of depressive and mania symptoms. Meanwhile, the 14-item Hamilton Anxiety Rating Scale (HAMA) was used to assess the severity of anxiety symptoms. The two psychiatrists with 23 years and 6 years of experience in clinical psychiatry attended a training session on the use of the 24-item HDRS, YMRS and HAMA before the start of the current study. After training, the inter-rater correlation coefficient of 24-item HDRS, YMRS and HAMA total scores between two raters was over 0.8.
The study was approved by the Ethics Committee of First Affiliated Hospital of Jinan University, China. All participants signed informed consent forms after reviewing a full written and verbal explanation of the study. And the neuropsychological assessment, and MRI scanning was completed within 48 h of initial contact.
Cognitive assessments
Cognitive function was qualified using the MCCB (Shi et al., Reference Shi, Kang, Yao, Ma, Li, Liang and Yu2015). The final MCCB battery requires approximately 70 min to administer and it consists of Trail Making Test Part A; Brief Assessment of Cognition in Schizophrenia: Symbol coding; Hopkins Verbal Learning Test (HVLT); Wechsler Memory Scale Spatial span; Neuropsychological Assessment Battery (NAB): Mazes; Brief Visuospatial Memory Test; Category fluency; Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT): Managing Emotions; and the Continuous Performance Test: Identical Pairs. MCCB was used to evaluate seven domains of information processing speed, attention/alertness, working memory, verbal learning, visual learning, reasoning and problem-solving and social cognition, with a global composite score. Of note, the clinical validity and test-retest reliability were established in both healthy controls and MDD patients (Liang et al., Reference Liang, Yu, Ma, Luo, Zhang, Sun and Zhang2020; Shi et al., Reference Shi, Kang, Yao, Ma, Li, Liang and Yu2015). Specifically, the effect size for test-retest reliability of nine cognitive subtests varied from 0.73 to 0.94 and the Cronbach's alpha of each item in the internal consistency analysis was ranged from 0.78 to 0.83. And two graduate students attended a training session on the use of the MCCB cognitive battery. After training, the inter-rater correlation coefficient of MCCB between two raters was over 0.8.
Image acquisition and preprocessing
All MRI data were gathered on a GE Discovery MR750 3.0T System with an 8-channel phased-array head coil. The participants were scanned in a supine, head-first position with symmetrically placed cushions on both sides of the head to decrease motion. During the scanning, the participants were instructed to relax with their eyes closed without falling asleep. After the experiment, each participant confirmed not having fallen asleep.
The rs-fMRI data were acquired using a gradient-echo echo-planar imaging sequence with the following parameters: time repetition (TR)/time echo (TE) = 2000/25 ms; flip angle = 90°; voxel size = 3.75 × 3.75 × 3 mm3; field of view (FOV) = 240 × 240 mm2; matrix = 64 × 64; slice thickness/gap = 3.0/1.0 mm; 35 axial slices covering the whole brain; and 210 volumes acquired in 7 min. In addition, a three-dimensional brain volume imaging (3D-BRAVO) sequence covering the whole brain was used for structural data acquisition with the following parameters: TR/ TE = 8.2/3.2 ms; flip angle = 12°; bandwidth = 31.25 Hz; slice thickness/gap = 1.0/0 mm; matrix = 256 × 256; FOV = 240 × 240 mm2; NEX = 1; and acquisition time = 3 min 45 s. Routine MRI examination images were also collected for excluding any anatomic abnormality. All participants were found by two experienced neuroradiologists (ZQ and ZL, with 5 and 3 years of experience in neuroimaging, respectively) to confirm that there were no brain structural abnormalities.
Functional image data preprocessing
The preprocessing was conducted using Data Processing Assistant for Resting-State fMRI (DPABI_V3.0, http://restfmri.net/forum/DPABI) (Yan, Wang, Zuo, & Zang, Reference Yan, Wang, Zuo and Zang2016) which is based on Statistical Parametric Mapping (SPM12, http://www.fil.ion.ucl.ac.uk/spm/). For each subject, the first 10 images of the rs-fMRI dataset were discarded to ensure steady-state longitudinal magnetization. The remaining 200 images were first slice-time corrected and then were realigned to the first image for correcting for inter-TR head motion. This realignment correction provided a record of the head motion within the rs-fMRI scan. All subjects should have no more than 2 mm maximum displacement in any plane, 2° of angular motion as well as 0.2 mm in mean frame-wise displacement (FD) (Jenkinson, Bannister, Brady, & Smith, Reference Jenkinson, Bannister, Brady and Smith2002). The individual T1 structural images were segmented (white matter, gray matter, and cerebrospinal fluid) using a segmentation toolbox. Then, the DARTEL toolbox was used to create a study-specific template for accurate normalization. Then, resting-state functional images were co-registered to the structural images and transformed into standard Montreal Neurological Institute (MNI) space, resliced to a voxel size of 3 × 3 × 3 mm3 resolution. The data were removed linear trend and passed through a band-pass filter of 0.01–0.1 Hz. Several spurious covariates and their temporal derivatives were then regressed out from the time course of each voxel, including the signals of the brain global mean, white matter, and cerebrospinal fluid as well as the Friston-24 parameters of head motion.
Dynamic functional connectivity variability analysis
Following previous work (Fan et al., Reference Fan, Li, Zhuo, Zhang, Wang, Chen and Jiang2016), seed-based dFC analyses were performed by placing regions of interest (ROIs) within four masks (bilateral rostral hippocampus and bilateral caudal hippocampus) using Brainnetome atlas (http://atlas.brainnetome.org/bnatlas.php) (Fig. 1). The dFC variability characteristics of the hippocampus were calculated using the sliding-window method based on the Temporal Dynamic Analysis (TDA) toolkits integrated into the DPABI software (http://rfmri.org/DPABI). The Hamming sliding window was selected for the whole-brain blood oxygenation level-dependent (BOLD) signal time series; 50 TRs window length and step width of 1 TRs were selected for dFC analysis. The minimum window length should be no less than 1/f min (1/0.01 s = 100 s) according to previous studies (Leonardi & Van De Ville, Reference Leonardi and Van De Ville2015; Li, Duan, Cui, Chen, & Liao, Reference Li, Duan, Cui, Chen and Liao2019); the f min was defined as the minimum frequency of time series. Shorter window lengths might increase the risk of introducing spurious fluctuations in the observed dFC. The window length of 50 TRs (100 s) was selected to compute the temporal variability of FC because a longer window length might hinder the description of the temporal variability dynamics. Also, other window lengths (30 TRs and 70 TRs) and shifting steps (1 TRs) were tried to further examine their possible effects on dFC results (Liao et al., Reference Liao, Wu, Xu, Ji, Zhang, Zang and Lu2014). In total, 151 sliding windows of dFC were obtained (each sliding window matrix is 61 × 73 × 61 × 100s). For each sliding window, correlation maps were produced by computing the temporal correlation coefficient between the truncated time series of the hippocampus subregions and all the other voxels. Consequently, 151 sliding window correlation maps were obtained for each individual. To improve the normality of the correlation distribution, each correlation map was converted into z-value maps using Fisher's r-to-z transformation. Then, the dFC maps were computed by calculating the standard deviation of 151 sliding-window z-value maps. Then, z-standardization was applied for the dFC maps. Finally, all the dFC maps were smoothed using a 6 mm full width at half maximum Gaussian kernel.
Fig. 1. Four seeds of the hippocampus in the bilateral hemisphere; L (R), left (right) hemisphere.
Statistical analysis
All indicators (i.e. demographics, and cognitive function) were measured for normal distributions by goodness-of-fit testing (Kolmogorov–Smirnov test, Levine's test of equality of error variances) using SPSS 24.0 software (SPSS, Chicago, IL, USA). When comparing group differences in terms of demographics and clinical data, t test was used if continuous variables were normal; likewise, the Mann–Whitney U test was used if continuous variables were skewed. A χ2 test was used to compare the gender differences between the two groups.
We tested for group differences on the seven cognitive domains plus the MCCB composite score using a multivariate analysis of covariance with a subject type (melancholic MDD v. healthy controls) as a fixed factor and including age and education levels as a covariate. Bonferroni correction was applied to account for multiple testing, with the threshold for significance was set to p < 0.006 (adjusted α = 0.006, 0.05/8).
The one-sample t test was performed to demonstrate the within-group dFC variability distribution of each subregion in patients with melancholic MDD and HCs (p < 0.05, uncorrected). To further examine the difference in dFC variability patterns between patients with melancholic MDD and HCs, a two-sample t test was performed on the standard deviation in the z value at each voxel within the union mask of one-sample t test results of the two groups. Age, gender and years of education were included as nuisance covariates in the comparisons. The cluster-level multiple comparison correction was conducted based on the Gaussian random field (GRF) theory (voxel p < 0.005; cluster p < 0.05/4 = 0.0125, corrected).
Once the significant group differences in dFC variability were observed in each subregion of the hippocampus, the Spearman correlation coefficient was calculated between the dFC variability values and MCCB T-scores (overall and specific domains) in patients with melancholic MDD. Also, we calculated the Spearman correlation coefficient between the demographic and clinical variables (age, education levels, onset age of illness, total number of MDD episodes, number of previous MDD episodes, duration of illness, 24-item HDRS score, and HAMA score), abnormal dFC variability values and MCCB T-scores (overall and specific domains) in the melancholic MDD group. All tests were two-tailed, and the significant level was set at a p value less than 0.05.
Validation analysis
Another 2 supplementary window lengths (30 TRs and 70 TRs) were applied to validate the main results of dFC with the window length of 50 TRs.
Results
Demographic information
Table 1 shows the demographic and clinical information for all the study participants. Three patients with melancholic MDD and none control participants were excluded from further analyses because of excessive head motion during the image acquisition. Finally, the participants were 42 patients with melancholic MDD and 55 healthy controls. There were no significant differences in age, sex, or education levels between the melancholic MDD group and HCs group (all p > 0.05).
Table 1. Demographic and clinical data of participants
MDD, major depressive disorder; HCs, healthy controls; 24-item HDRS, 24 item Hamilton Depression Rating Scale; YMRS, Young Manic Rating Scale; HAMA, 14-item Hamilton Anxiety Rating Scale (HAMA); MCCB, the Chinese version of the Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS) Consensus Cognitive Battery; s.d., standard deviation.
Values are reported as mean (s.d., standard deviation).
a χ2 test.
b Mann–Whitney U test.
c Multivariate analysis of covariance.
*p < 0.05, **p < 0.01, ***p < 0.001.
Group differences of cognitive function
The cognitive performance results of MCCB scores in patients with melancholic MDD and HCs are shown in Table 1. Compared with the HCs, patients with melancholic MDD showed significantly lower overall composite T-score (F = 25.347, p < 0.001), speed of processing (F = 5.663, p = 0.02), working memory (F = 9.457, p = 0.003), verbal learning (F = 14.229, p < 0.001), visual learning (F = 9.944, p = 0.002) and social cognition (F = 28.347, p < 0.001) at p < 0.05. After Bonferroni correction (adjusted α = 0.006, 0.05/8), those results were still significantly different between groups except for the domain of processing speed.
Dynamic functional connectivity variability of the hippocampal subregions
The one-sample t test revealed the dFC variability patterns for each hippocampal subregion in two groups (Fig. 2). The dFC spatial distribution in the melancholic MDD group were similar to those of the HCs group (p < 0.05, uncorrected for visual inspection). In both groups, the dFC of the bilateral rostral hippocampus were mainly located in the superior and middle frontal gyrus, cingulate gyrus, parahippocampal gyrus, temporal lobe, parietal lobe, postcentral, occipital regions and anterior and posterior lobe of the cerebellum, and the dFC of the bilateral caudal hippocampus were mainly located in the precuneus, temporal lobe, anterior and middle cingulate, hippocampus, parahippocampal, insula and anterior lobe of the cerebellum. However, statistical analysis revealed that compared with the HCs group, the melancholic MDD group exhibited decreased dFC variability between the left rostral hippocampus and left anterior lobe of the cerebellum (voxel p < 0.005, cluster p < 0.0125, GRF corrected). No significant differences were found in the whole dFC of the right rostral hippocampus and bilateral caudal hippocampus between the melancholic MDD group and the HCs group (Table 2; Fig. 3).
Fig. 2. dFC patterns of the bilateral rostral hippocampus (rHipp) and the bilateral caudal hippocampus (cHipp) in melancholic MDD patients and HCs (p < 0.05, uncorrected). The color bar represents a dynamic functional connection. dFC, dynamic functional connectivity; MDD, major depressive disorder; HCs, healthy controls.
Fig. 3. Significant dFC differences between the two groups for hippocampus seed, respectively (voxel p < 0.005, cluster p < 0.0125, GRF corrected). The color bar indicates the t values from the two-sample t test analysis. dFC, dynamic functional connectivity; rHipp, the rostral hippocampus; GRF, Gaussian random field; L (R), left (right) hemisphere.
Table 2. The areas of significantly different dFC between the melancholic MDD patients and the HCs (voxel p < 0.005, cluster p < 0.0125, GRF corrected)
dFC, dynamic functional connectivity; MDD, Major depressive disorder; HCs, healthy controls; GRF, Gaussian random field.
Correlation analyses
A significant positive correlation was observed between the dFC variability values of the left rostral hippocampus- left anterior lobe of the cerebellum and working memory T -score (r = 0.338, p = 0.029) only in patients with melancholic MDD (Fig. 4). After correcting for age and education levels, this correlation was still existing (r = 0.329, p = 0.038). But there were no significant correlations between demographic and clinical characteristics and dFC variability values between the left rostral hippocampus- left anterior lobe of the cerebellum in patients with melancholic MDD (all p > 0.05). Additionally, the verbal learning was negatively correlated with the 24-HDRS scores (r = −0.403, p = 0.008), but there were no significant correlations between other demographic and clinical characteristics (age, education levels, onset age of illness, total number of MDD episodes, number of previous MDD episodes, duration of illness, and HAMA scores) and MCCB cognitive domains in patients with melancholic MDD (all p > 0.05).
Fig. 4. Positive correlation between the abnormal dFC variability values and working memory T-score (r = 0.338, p = 0.029). dFC, dynamic functional connectivity; rHipp, the rostral hippocampus; L (R), left (right) hemisphere.
Validation results
The validation results in 30 TRs sliding window length between the two groups also showed melancholic MDD patients exhibited decreased dFC variability values between the left rostral hippocampus and left anterior lobe of the cerebellum (online Supplementary Table S1 and Fig. S1). However, there were no significant differences in 70 TRs sliding window length and static FC between the two groups using the hippocampus as the ROIs.
Discussion
To the best of our knowledge, this study was the first to investigate the whole-brain dFC of hippocampal subregions in unmedicated patients with melancholic MDD, as well as to explore the relationship between the abnormal dFC variability values and the cognitive performance of this disorder. The main findings of this study showed melancholic MDD have decreased dFC variability values between the left rostral hippocampus and left anterior lobe of cerebellum than that in healthy controls. Our results also indicated that the melancholic MDD patients may have a profile of widespread cognitive impairments, showing in the domains of working memory, verbal learning, visual learning and social cognition, as well as MCCB composite scores. And verbal learning was negatively correlated with the 24-HDRS scores. Furthermore, correlation analysis showed that the decreased dFC variability values between the left rostral hippocampus and left anterior lobe of the cerebellum were positively correlated to working memory impairment.
Decreased dFC variability values of the subgenual hippocampus in melancholic MDD
A growing number of studies suggested that the cerebellum and hippocampus play an important role in cognition processing and emotional regulation (Anacker & Hen, Reference Anacker and Hen2017; Guo et al., Reference Guo, Liu, Dai, Jiang, Zhang, Yu and Xiao2013; Reshetnikov et al., Reference Reshetnikov, Kovner, Lepeshko, Pavlov, Grinkevich and Bondar2020; Xu et al., Reference Xu, Xu, Liu, Ji, Wu, Wang and Yu2017). In our current study, patients with melancholic MDD showed decreased dFC variability between the left rostral hippocampus and left anterior lobe of the cerebellum relevant to healthy controls. The previous study has indicated that the melancholic group had a greater number of early life stress (ELS) events than the non-melancholic patients (Quinn, Dobson-Stone, Outhred, Harris, & Kemp, Reference Quinn, Dobson-Stone, Outhred, Harris and Kemp2012), and the left hippocampus is more sensitive to stressful events than the right hippocampus (Saleh et al., Reference Saleh, Potter, McQuoid, Boyd, Turner, MacFall and Taylor2017; Teicher, Anderson, & Polcari, Reference Teicher, Anderson and Polcari2012). Interesting, the previous study has also pointed out that the left hippocampus is typically more affected than the right hippocampus in depression and other psychiatric disorders (Small, Schobel, Buxton, Witter, & Barnes, Reference Small, Schobel, Buxton, Witter and Barnes2011). Moreover, the anterior hippocampus mainly contributes to emotional reactions (Therriault et al., Reference Therriault, Wang, Mathotaarachchi, Pascoal, Parent, Beaudry and Alzheimer's Disease Neuroimaging2019), and this hippocampus subregion also differed between the depressed patients and controls (Ballmaier et al., Reference Ballmaier, Narr, Toga, Elderkin-Thompson, Thompson, Hamilton and Kumar2008; Posener et al., Reference Posener, Wang, Price, Gado, Province, Miller and Csernansky2003). Collectively, the dysfunction of the left anterior hippocampus may explain why the melancholic MDD patients suffered greater depression severity than typical MDD.
Moreover, previous studies found that the melancholic MDD patients were associated with lower BDNF levels (Patas et al., Reference Patas, Penninx, Bus, Vogelzangs, Molendijk, Elzinga and Oude Voshaar2014; Primo de Carvalho Alves & Sica da Rocha, Reference Primo de Carvalho Alves and Sica da Rocha2018) and the Met66 carriers in melancholic MDD patients paralleled with a reduced left hippocampal volume (Cardoner et al., Reference Cardoner, Soria, Gratacos, Hernandez-Ribas, Pujol, Lopez-Sola and Soriano-Mas2013). These changes may cause the left hippocampus to reduce its functional connections to other cortical regions due to synaptic depletion. Meanwhile, recent studies have also reported important functional interactions between the cerebellum and the hippocampal formation (Igloi et al., Reference Igloi, Doeller, Paradis, Benchenane, Berthoz, Burgess and Rondi-Reig2015; Onuki, Van Someren, De Zeeuw, & Van der Werf, Reference Onuki, Van Someren, De Zeeuw and Van der Werf2015; O'Reilly, Beckmann, Tomassini, Ramnani, & Johansen-Berg, Reference O'Reilly, Beckmann, Tomassini, Ramnani and Johansen-Berg2010; Watson et al., Reference Watson, Obiang, Torres-Herraez, Watilliaux, Coulon, Rochefort and Rondi-Reig2019). Functional brain networks demonstrate significant temporal variability and dynamic reconfiguration, and this ability of specific regions to dynamically connectivity may play an important role in cognitive flexibility and behavioral adaptability (Bray, Arnold, Levy, & Iaria, Reference Bray, Arnold, Levy and Iaria2015; Zhang et al., Reference Zhang, Cheng, Liu, Zhang, Lei, Yao and Feng2016). The melancholic MDD showed decreased dFC in hippocampus-cerebellum signifies a lack of flexibility changes in spontaneous brain activity and neural communication over time, which may be a neural maker of melancholic features. The reduced hippocampal volumes have been found in the melancholic depression (Cardoner et al., Reference Cardoner, Soria, Gratacos, Hernandez-Ribas, Pujol, Lopez-Sola and Soriano-Mas2013; Hickie et al., Reference Hickie, Naismith, Ward, Turner, Scott, Mitchell and Parker2005), which can predict a slower recovery after treatment initiation (Soriano-Mas et al., Reference Soriano-Mas, Hernandez-Ribas, Pujol, Urretavizcaya, Deus, Harrison and Cardoner2011). The evidence showed an inverse correlation between the volume of the deep white matter hyperintensities and hippocampal volume, as well as a direct influence on the connectivity properties of this important cerebral region (Porcu et al., Reference Porcu, Operamolla, Scapin, Garofalo, Destro, Caneglias and Saba2020), and thus influence the state and dynamic connectivity between hippocampal and cerebellum. Furthermore, a recent study suggested modulating the function of the hippocampus–cerebellum circuit may be a potential therapeutic strategy for depressive symptoms in epilepsy patients (Peng et al., Reference Peng, Mao, Yin, Sun, Wang, Zhang and Wang2018). Taken together, these findings suggest that impaired hippocampus-cerebellum circuits function might contribute to the pathogenesis of melancholic MDD and may provide a potential target for therapeutic intervention.
Impairments of MCCB cognitive performance in melancholic MDD
In the current study, we found the melancholic MDD have significantly lower scores in the cognitive domains of working memory, verbal learning, visual learning and social cognition than that in healthy controls, which are consistent with most of the previous studies evaluating cognitive performance in patients with melancholic MDD (Day et al., Reference Day, Gatt, Etkin, DeBattista, Schatzberg and Williams2015; Quinn, Harris, Felmingham, Boyce, & Kemp, Reference Quinn, Harris, Felmingham, Boyce and Kemp2012; Withall et al., Reference Withall, Harris and Cumming2010). These findings suggested that the melancholic MDD showed a profile of widespread cognitive impairments relative to healthy controls, and that are independent of the severity of symptoms (Linden, Jackson, Subramanian, Healy, & Linden, Reference Linden, Jackson, Subramanian, Healy and Linden2011).
Our results indicated that the melancholic MDD patients were mainly involved in memory deficit, in line with other studies (Bosaipo et al., Reference Bosaipo, Foss, Young and Juruena2017; Zaninotto et al., Reference Zaninotto, Solmi, Veronese, Guglielmo, Ioime, Camardese and Serretti2016). And the effect size of these memory domains was medium to large (Cohen's d range: 0.72 to 1.05) confirmed similar findings in previous studies (Austin et al., Reference Austin, Mitchell, Wilhelm, Parker, Hickie, Brodaty and Hadzi-Pavlovic1999; Withall et al., Reference Withall, Harris and Cumming2010). Verbal learning was assessed by the Hopkins Verbal Learning Test (HVLT) also defined as verbal memory. Similarly, a previous study found worse performance in tests measuring verbal working memory with melancholic MDD compared with healthy controls (Austin, Mitchell, & Goodwin, Reference Austin, Mitchell and Goodwin2001). Meanwhile, in the domains of visual-spatial memory, and verbal working memory, melancholic depressives also performed significantly worse than healthy controls (Lin et al., Reference Lin, Xu, Lu, Ouyang, Dang, Lorenzo-Seva and Lee2014). Moreover, Linden and colleagues found an emotional bias on working memory performance in the melancholic depression group (Linden et al., Reference Linden, Jackson, Subramanian, Healy and Linden2011). According to the cognitive theories (Mathews & MacLeod, Reference Mathews and MacLeod2005; Ridout, Astell, Reid, Glen, & O'Carroll, Reference Ridout, Astell, Reid, Glen and O'Carroll2003), the patients with melancholic depression posit a bias for negative or sad information, the capacity-limited memory system was full of unrelated negative emotional materials, resulting in the brain dysfunction in shifting processing, updating and inhibiting, and thus damaged the working memory. After remission from depression, melancholic depression patients could recover their visual-spatial memory and verbal working memory function to the level of healthy controls (Lin et al., Reference Lin, Xu, Lu, Ouyang, Dang, Lorenzo-Seva and Lee2014). Indeed, our results also revealed a negative correlation between verbal learning and 24-HDRS scores. Therefore, changing the sad bias in working memory may become a potential direction or measure for future targeted treatment and psychological interventions.
Correlations between abnormal dFC variability values and cognitive deficits
In our present study, the decreased dFC variability values between the left rostral hippocampus and left anterior lobe of the cerebellum was positively correlated with working memory deficit in patients with melancholic MDD. Previous numerous fMRI and positron emission tomography (PET) studies have been reliably demonstrated that cerebellar engagement in working memory tasks (Beneventi, Barndon, Ersland, & Hugdahl, Reference Beneventi, Barndon, Ersland and Hugdahl2007; Guell, Gabrieli, & Schmahmann, Reference Guell, Gabrieli and Schmahmann2018; Hautzel, Mottaghy, Specht, Muller, & Krause, Reference Hautzel, Mottaghy, Specht, Muller and Krause2009; Hayter, Langdon, & Ramnani, Reference Hayter, Langdon and Ramnani2007; Ng et al., Reference Ng, Kao, Chan, Chew, Chuang and Chen2016). Of note, the rostral hippocampus is closely linked to general memory processes including both learning, memory, encoding and retrieval (Carr et al., Reference Carr, Rissman and Wagner2010; Robinson et al., Reference Robinson, Barron, Kirby, Bottenhorn, Hill, Murphy and Fox2015; Zeidman & Maguire, Reference Zeidman and Maguire2016). And accumulating evidence suggested that functionally intact cerebellar-hippocampal interactions underlie spatial memory processing and coding (McNaughton, Battaglia, Jensen, Moser, & Moser, Reference McNaughton, Battaglia, Jensen, Moser and Moser2006). Importantly, the persistent neural activity in the hippocampus is critical for working memory processing (Boran et al., Reference Boran, Fedele, Klaver, Hilfiker, Stieglitz, Grunwald and Sarnthein2019; Yonelinas, Reference Yonelinas2013). And working memory processing also relies on persistent neural activity in a widespread neural network of brain areas (Boran et al., Reference Boran, Fedele, Klaver, Hilfiker, Stieglitz, Grunwald and Sarnthein2019; Kim, Reference Kim2019). In the human neural network, human brain connectivity is dynamic and associated with ongoing rhythmic activity over time and these dynamic properties provide high-level flexibility in cognitive function processing (Dehaene et al., Reference Dehaene, Naccache, Cohen, Bihan, Mangin, Poline and Riviere2001; Yu & Dayan, Reference Yu and Dayan2005). Recent studies have also indicated that the dFC are indeed related to cognitive function, and may supersede traditional neuroimaging measures in explaining cognitive variance (Douw, Wakeman, Tanaka, Liu, & Stufflebeam, Reference Douw, Wakeman, Tanaka, Liu and Stufflebeam2016; Hellyer, Jachs, Clopath, & Leech, Reference Hellyer, Jachs, Clopath and Leech2016). In addition, two recent studies on the multiple sclerosis brain discovered that the disrupted cerebellar dFC was related to worse working memory and processing speed (Schoonheim et al., Reference Schoonheim, Douw, Broeders, Eijlers, Meijer and Geurts2021), and the lower (both left and right) hippocampus dFC was also correlated with memory dysfunction (van Geest et al., Reference van Geest, Hulst, Meijer, Hoyng, Geurts and Douw2018). Furthermore, the decreased FC between the left hippocampus and the left anterior cerebellum were well as correlated with cognitive dysfunction in patients with obstructive sleep apnea (Zhou et al., Reference Zhou, Liu, Luo, Li, Peng, Zong and Ouyang2020). Whereas, the higher levels of brain dynamics are an important indicator for better cognitive performance in healthy subjects, such as working memory, cognitive flexibility, executive function and processing speed (Braun et al., Reference Braun, Schafer, Walter, Erk, Romanczuk-Seiferth, Haddad and Bassett2015; Cole et al., Reference Cole, Reynolds, Power, Repovs, Anticevic and Braver2013; Douw et al., Reference Douw, Wakeman, Tanaka, Liu and Stufflebeam2016; McIntosh, Kovacevic, & Itier, Reference McIntosh, Kovacevic and Itier2008; Nomi et al., Reference Nomi, Vij, Dajani, Steimke, Damaraju, Rachakonda and Uddin2017). Therefore, these findings would explain the potential mechanism of decreased hippocampus-cerebellum dynamic paralleled with working memory impairment in patients with melancholic MDD.
The hippocampus and cerebellum are functionally connected in a bidirectional manner and helping to the formation of spatial memory (McNaughton et al., Reference McNaughton, Battaglia, Jensen, Moser and Moser2006; Passot, Sheynikhovich, Duvelle, & Arleo, Reference Passot, Sheynikhovich, Duvelle and Arleo2012; Rochefort, Lefort, & Rondi-Reig, Reference Rochefort, Lefort and Rondi-Reig2013). Critically, studies have demonstrated an important dependence of the cerebellum on the hippocampus that the cerebellum supplied by the hippocampus in the coding of time and, declarative and episodic memory (Burgess, Maguire, & O'Keefe, Reference Burgess, Maguire and O'Keefe2002; Eichenbaum, Reference Eichenbaum2014; Zeidler, Hoffmann, & Krook-Magnuson, Reference Zeidler, Hoffmann and Krook-Magnuson2020). Evidence from rats' experiments indicated a crucial role for the cerebellum in hippocampus-dependent spatial memory (Netrakanti et al., Reference Netrakanti, Cooper, Dere, Poggi, Winkler, Brose and Ehrenreich2015). Meanwhile, Bohne and colleagues also confirmed the connectivity map between the hippocampus and cerebellum in mice and strengthen the notion of the cerebellum's involvement in cognitive functions (Bohne, Schwarz, Herlitze, & Mark, Reference Bohne, Schwarz, Herlitze and Mark2019). The left hippocampus and cerebellar interact during the prediction of spatio-temporal aspects of voluntary movements which are related more closely to spatial cognition (Onuki et al., Reference Onuki, Van Someren, De Zeeuw and Van der Werf2015; Stoodley, Valera, & Schmahmann, Reference Stoodley, Valera and Schmahmann2012). Therefore, despite the cerebellum's hypothesized role in working memory, without hippocampal support, the cerebellum appears unable to keep information about the conditioned stimulus ‘on-line’ (Kuper et al., Reference Kuper, Kaschani, Thurling, Stefanescu, Burciu, Goricke and Timmann2016; McNaughton et al., Reference McNaughton, Battaglia, Jensen, Moser and Moser2006) and thus disrupts working memory (Zeidler et al., Reference Zeidler, Hoffmann and Krook-Magnuson2020).
Finally, our results also found the melancholic MDD patients performed worse social cognitive performance compared to healthy participants. Recent evidence indicated that both the acute and remission stage of MDD illness exhibited social cognitive deficits (Knight & Baune, Reference Knight and Baune2019; LeMoult, Joormann, Sherdell, Wright, & Gotlib, Reference LeMoult, Joormann, Sherdell, Wright and Gotlib2009), which also contribute to functional deficits in occupational functioning, interpersonal relationships, and self-perceived quality of life (Weightman, Air, & Baune, Reference Weightman, Air and Baune2014). Moreover, the effect size of social cognition impairment was large (Cohen's d = 1.23) in melancholic MDD patients, suggesting that acute social cognitive deficits may be greater in currently melancholic depressed individuals. This aspect of social cognition should be considered a prime target in adjunctive cognitive and psychosocial treatments (Knight & Baune, Reference Knight and Baune2017; McIntyre & Lee, Reference McIntyre and Lee2016), achieving the goal of full functional recovery.
Limitations
However, some limitations to the present study should also be considered. First, this study was designed as a cross-sectional study, the progressive changes did not be observed. Second, there is no comparison of intelligence quotient (IQ) differences between the two groups. And it would be helpful that take the premorbid IQ into account when assessing the differences in cognitive performance between the two groups in future studies. Third, we compared these cognitive and dFC variability differences between the melancholic MDD patients and healthy controls only, but no non-melancholic MDD patients were included. A previous study found that the atypical MDD patients exhibited significantly decreased dynamic FC of the cerebellar subregions connecting with the superior temporal gyrus, dorsal lateral prefrontal cortex, ventral medial prefrontal cortex and visual area (Zhu et al., Reference Zhu, Yang, Zhang, Wang, Wang, Zhang and Zhu2020). Another study suggested that melancholic depression exhibited decreased effective connectivity between the right frontoparietal and insula networks compared with no-melancholic depression (Hyett, Breakspear, Friston, Guo, & Parker, Reference Hyett, Breakspear, Friston, Guo and Parker2015). Consequently, a direct comparison of dynamic connectivity between melancholic and typical MDD using a larger homogeneous sample is encouraged and the present findings might not apply to other MDD subtypes. Furthermore, previous studies reported that melancholic depressed patients may demonstrate different serum concentrations of inflammatory cytokine and cortisol in comparison with non-melancholic features (Kaestner et al., Reference Kaestner, Hettich, Peters, Sibrowski, Hetzel, Ponath and Rothermundt2005; Karlovic, Serretti, Vrkic, Martinac, & Marcinko, Reference Karlovic, Serretti, Vrkic, Martinac and Marcinko2012; Primo de Carvalho Alves & Sica da Rocha, Reference Primo de Carvalho Alves and Sica da Rocha2020). Accumulating evidence suggests that the cortisol levels, thyroid hormones and inflammatory cytokine levels may be associated with the functional connectivity of depressed-related brain regions (Felger et al., Reference Felger, Li, Haroon, Woolwine, Jung, Hu and Miller2016; Peters et al., Reference Peters, Jenkins, Stange, Bessette, Skerrett, Kling and Langenecker2019; Wang et al., Reference Wang, Chen, Zhong, Jia, Xia, Lai and Liu2018), and visuospatial memory (Hakamata et al., Reference Hakamata, Komi, Sato, Izawa, Mizukami, Moriguchi and Tagaya2019). However, it remains unclear how hippocampal connectivity is involved in the relationship between cortisol and visuospatial memory in melancholic MDD. Next, we can also explore the underlying complex interactions between these blood biomarkers and the brain functional abnormalities in patients with melancholic MDD.
Conclusions
In summary, our results indicate that the decreased dFC variability values between the left rostral hippocampus and left anterior lobe of the cerebellum may signify an underlying neural substrate of working memory impairment in melancholic MDD. And mapping subregional hippocampal abnormalities and their cognitive correlates may provide a potential direction for future interventions of this MDD subtype.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291721004906.
Acknowledgements
The authors thank the patients, volunteers, and their families whose participation made this work possible.
Financial support
Funding for this work was provided by the National Natural Science Foundation of China (No: 81801347; 81971597; 81671351; 81671670 and 82102003), Planned Science and Technology Project of Guangdong Province, China (No: 2017B020227011), National Key Research and Development Program of China (2020YFC2005700), Project in Basic Research and Applied Basic Research in General Colleges and Universities of Guangdong, China (2018KZDXM009) and Natural Science Foundation of Guangdong Province, China (No: 2021A1515011034). The founders have not played any roles in study design, data collection, analysis, manuscript writing and decision to publish.
Author contributions
Yanbin Jia and Ying Wang designed the trial and prepared the manuscript. Lai Shunkai and Ting Su contributed equally to this work, and are the first co-authors. Yanbin Jia, Shunkai Lai, Shuming Zhong, Ying Wang, Ting Su, Yiliang Zhang, Hui Zhao, Guanmao Chen, Pan Chen, Guixian Tang, Zhangzhang Qi, Jiali He, Yunxia Zhu, Sihui Lv, Zijing Song, Haofei Miao, Yilei Hu, and Hanglin Ran acquired the data. Shunkai Lai and Ting Su carried out the statistical analyses, drafted the initial manuscript. All authors interpreted the data, revised the paper critically for important intellectual content, approved the final version, and agreed to be accountable for all aspects of the work.
Conflict of interest
Each author has declared that there are no conflicts of interest in relation to the study presented here.
Data availability statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
