Functional connectivity of the default mode network subsystems in patients with major depressive episodes with mixed features

Study design and participants

This is a cross-sectional study. A total of 74 patients were recruited from the outpatient department of Beijing Anding Hospital, Capital Medical University, between April 2021 and June 2022. Among them, 47 were patients with MDD without mixed features (MDD_noMF) and 27 were patients with MMF. All patients were diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) by two experienced psychiatrists. The patients who were diagnosed with MDD and had fewer than two manic/hypomanic symptoms were considered as patients with MDD_noMF. The patients who met the full diagnostic criteria for MDD and had at least three manic/hypomanic features (eg, an excessively elevated mood, inflated self-esteem or grandiosity, more talkative than usual or feeling pressured to keep talking, flight of ideas or racing thoughts, increase in energy or goal-directed activity, increased or excessive involvement in activities that have a high potential for painful consequences, and decreased need for sleep) during the majority of days of the current or recent major depressive episode were diagnosed as patients with MMF.17 All patients were right-handed and 18–55 years of age. The patients were excluded if they had any co-existing or history of other psychiatric disorders, such as schizophrenia or bipolar disorder, any major medical illnesses, a clinical diagnosis of neurological trauma, any history of substance or alcohol abuse, and any contraindications to magnetic resonance imaging (MRI) scanning. A total of 45 age- and gender-matched healthy controls (HCs) were recruited through advertisements by the Beijing Anding Hospital, Capital Medical University. The study flowchart is presented in figure 1.

Request permission

Clinical assessment

The sociodemographic data and clinical characteristics were collected by the use of a predesigned data collection form. The severity of MDD was assessed with the 17-item Hamilton Depression Rating Scale (HAMD-17).18 The manic/hypomanic symptoms were measured using the Young Mania Rating Scale (YMRS).19

Data acquisition

All the subjects underwent resting-state functional MRI (fMRI) scanning using a Siemens Prisma 3.0 T MRI scanner with a 64-channel phased-array head coil. The structural images were acquired using the T1-weighted magnetization-prepared rapidly acquired gradient-echo sequence with the following parameters: repetition time (TR)=2530 ms; flip angle (FA)=15°; echo time (TE)=1.85 ms; matrix=256×256; field of view (FOV)=256×256 mm²; number of slices=192; slice thickness=1 mm; voxel size=1×1×1 mm³. The resting-state fMRI images were acquired using a gradient-recall echo-planar imaging pulse sequence with the following parameters: TR=2000 ms; TE=30 ms; FA=90°; matrix=64×64; FOV=200×200 mm²; number of slices=33; slice thickness=3.5 mm; gap=0.7 mm; voxel size=3.13×3.13×4.2 mm³; phase encoding direction=anterior to posterior; 200 volumes. During the scanning, the subjects were required to keep their eyes closed, remain awake, relax their minds and keep still without any head motion.

Data preprocessing

Data preprocessing was carried out using Data Processing & Analysis for (Resting-State) Brain Imaging (V.6.0_210501, http://rfmri.org/DPABI). The first five time points were removed to ensure signal stability, and slice timing correction was performed. After the realignment, the T1 images were co-registered to the functional image. Afterwards, linear and quadratic trends, the first five principal components of the individually segmented white matter and cerebrospinal fluid, and Friston’s 24 motion parameters were regressed out as covariates. The images were then normalized to the Montreal Neurological Institute template, and each voxel was resampled to 2×2×2 mm³. Spatial smoothing was conducted using a Gaussian kernel with a full width half-maximum of 4 mm, and band-pass temporal filtering was performed at a range of 0.01–0.1 Hz. To quantify microhead motions, the ‘bad’ time points were removed from the time series by employing a ‘scrubbing’ method with a frame-wise displacement (FD)20 threshold of 0.5 mm. The subjects whose head motion was >2.5 mm maximum translation in any direction of x, y or z or 2.5° of maximum rotation or if their mean FD exceeded 3 SD of the mean value were excluded.21

Definition of ROIs

A total of 24 regions of interest (ROI) were extracted from the 17-network parcellation by Yeo et al.22 They were divided into three DMN subsystems: nine ROIs belong to the core subsystem, nine ROIs belong to the dMPFC subsystem and six ROIs belong to the MTL subsystem (figure 2).

Request permission

Functional connectivity analyses

First, seed-based functional connectivity analyses were conducted to calculate the functional connectivities among ROIs in the three DMN subsystems. We extracted mean time series from each ROI, calculated Pearson’s correlations between each pair of ROIs and applied Fisher’s R-to-Z transformation, forming a 24×24 functional connectivity matrix for each subject. Second, the intranetwork functional connectivity in the three DMN subsystems was individually calculated with the averaged functional connectivity of ROIs within each of the DMN subsystems. The internetwork functional connectivity was calculated with the mean of pairwise connections between the DMN subsystems. The edge functional connectivity within each of the DMN subsystems was the connectivity between pairs of ROIs within this subsystems.

Statistical analysis

Data were analysed using SPSS software, V.24.0 and R, V.4.2.1. The comparisons of sociodemographic and clinical characteristics between MDD_noMF, MMF and HC were performed using χ² tests, one-way analysis of variance (ANOVA) or Kruskal-Wallis H test, as appropriate. The significance level was set as 0.05 (two-tailed).

Analysis of covariance (ANCOVA) was conducted to compare the differences of the intranetwork and internetwork functional connectivity in the DMN subsystems among the MDD_noMF, MMF and HC groups, controlling for age, gender and mean FD. In addition, the edge connectivity within the abnormal DMN subsystems was further compared using the ANCOVA method. The false discovery rate (FDR) was used for multiple comparisons, and the FDR-adjusted p values <0.05 were considered statistically significant. The post hoc least significant difference analyses were applied for multiple group comparisons. The correlations between the abnormal intranetwork functional connectivity of the DMN subsystems and the total score of clinical scales were calculated for the patients using Pearson’s correlation, or Spearman’s correlation method, as appropriate.

Demographic and clinical characteristics comparisons

After excluding some subjects (for details, please see figure 1), a total of 34 patients with MDD_noMF, 25 patients with MMF and 42 HCs were included in the study. The demographic and clinical characteristics are summarised in table 1. The age, sex and mean FD were not significantly different among the groups (all p-values>0.05). There were significant differences in the education level and the total scores of the HAMD-17 and YMRS scales among the three groups. Compared with HC group, the patients with MDD_noMF and MMF had higher education levels. The patients with MDD_noMF and MMF had significantly higher HAMD-17 scores than the HCs, while no significant difference existed between the patients with MDD_noMF and MMF. The patients with MMF had higher YMRS scores than those with MDD_noMF and the HCs.

Differences in functional connectivity of the DMN subsystems

The intranetwork functional connectivity of the DMN core subsystem (F=6.32, p_FDR=0.008) and the MTL subsystem (F=4.45, p_FDR=0.021) showed significant differences among the MDD_noMF, MMF and HC groups. No significant differences were found in the intranetwork functional connectivity of the DMN dMPFC subsystem and in the internetwork functional connectivities between the DMN subsystems.

As shown in figure 3, compared with HCs, the patients with MDD_noMF and MMF showed increased functional connectivity within the DMN MTL subsystem (mean difference (HC–MDD_noMF)=−0.11, SE=0.04, p=0.008, 95% CI: −0.19 to –0.03; mean difference (HC–MMF)=−0.10, SE=0.04, p=0.029, 95% CI: −0.18 to –0.01). Meanwhile, the MMF had increased functional connectivity within the DMN core subsystem (mean difference (HC–MMF)=−0.14, SE=0.04, p=0.001, 95% CI: −0.22 to –0.06). In addition, the patients with MMF also had increased functional connectivity within the DMN core subsystem compared with patients with MDD_noMF (mean difference (MDD_noMF–MMF)=−0.08, SE=0.04, p=0.048, 95% CI: −0.17 to –0.001).

Request permission

Furthermore, we compared the edge functional connectivity within the abnormal DMN subsystems among the MDD_noMF, MMF and HC groups. We found there was a range of abnormal edge connections in the DMN core subsystem, whereas no significant abnormal edge connections existed in the DMN MTL subsystem. In the DMN core subsystem, the edge connectivity of the LH_IPL–LH_PFCm (F=6.85, p_FDR=0.016), LH_IPL–RH_PFCm (F=7.59, p_FDR=0.016), LH_PFCd–RH_PFCd (F=5.41, p_FDR=0.036), LH_PFCd–RH_PFCm (F=6.75, p_FDR=0.016), LH_pCunPCC–RH_PFCm (F=5.21, p_FDR=0.037), LH_PFCm–RH_pCunPCC (F=4.77, p_FDR=0.048), RH_PFCd–RH_PFCm (F=6.49, p_FDR=0.016), RH_pCunPCC–RH_PFCm (F=7.57, p_FDR=0.016) showed significant differences among the MDD_noMF, MMF and HC groups. Except for the functional connectivity of LH_pCunPCC–RH_PFCm, all these edge connections were significantly increased in the patients with MMF compared with the HC group and the patients with MDD_noMF (figure 4).

Request permission

Correlation analysis

The intranetwork functional connectivity in the DMN core or MTL subsystems was not significantly correlated with HAMD-17 or YMRS total scores among the patients with MDD_noMF and MMF.

Main findings

This study investigated the functional connectivity of the DMN subsystems in patients with MDD_noMF and MMF. Results revealed the patients with MMF showed significantly increased connectivity within the DMN core subsystem when compared with patients with MDD_noMF, and they have a wide spread of abnormal edge functional connectivity within the DMN core subsystem. These findings indicate the brain regions within the DMN core subsystems (eg, IPL, PFCm, PFCd and pCunPCC) might play a critical role in the pathological mechanisms underlying MMF. Meanwhile, the MMF also showed increased connectivity within the DMN MTL subsystem compared with the MDD_noMF and HC groups.

This study found significantly increased connectivity within the DMN core subsystem in patients with MMF and MDD_noMF, with higher connectivity in patients with MMF as compared to patients with MDD_noMF. The DMN core subsystem is associated with personally significant affective information, acting as a network hub, linking social and mnemonic processes of the dMPFC and MTL subsystems.9 13 23 Lines of evidence reported increased DMN connectivity in patients with MDD, which may be associated with a predominance of internalised, self-focused mental activity over efficient externally focused, and non-self-related attention.5 10 Abnormal connectivity of the DMN might lead to excessive focusing on external content and psychomotor overexcitement that can manifest in manic symptoms.24 25 The results of this study indicated that the increased connectivity of the DMN core in MMF could be a possible mechanism for the emotional disturbances that are experienced during major depressive episodes accompanying a manic episode.

In addition, our seed-based functional connectivity showed widespread increased edge connectivity within the DMN core subsystem in the patients with MMF compared with HC and patients with MDD_noMF. The mPFC, PCC and pCunPCC are considered as core hubs of the DMN.24 26 27 These regions are central to emotional regulation and self-reference and executive functions--all areas where deficits have been reported in MDD.28 Consistent with previous studies, the patients with MDD_noMF showed increased functional connectivity in these regions compared with HCs.29 30 Moreover, patients with MMF displayed higher functional connectivity than the MDD_noMF group, indicating MMF might be associated with worse emotional regulation skills. Considering the manic/hypomanic symptoms in the MMF, we inferred that increased functional connectivity in these regions could undermine emotional regulation, and may lead to the manic/hypomanic symptoms. As such,our findings raise an intriguing question: Could the connectivity between the regions within the core DMN subsystem be of importance for understanding the increased self-focus in depression and its relationship to difficulties with engaging in emotional regulation among patients with MMF.

Finally, increased connectivity within the MTL subsystem was found in the patients with MMF compared with the HC group; however, there was no significant difference when compared with the patients with MDD_noMF. The MTL is associated with autobiographical memory, episodic future thinking, information retrieval and imagery and navigation.23 31 The functional connectivity of the MTL subsystem is related to the frequency of thinking about the future.32 Previous studies indicated that the patients with depressive symptoms have a lesser cognitive impairment than the patients with manic symptoms.33 As noted, bipolar disorder patients have functional changes in brain regions related to autobiographical memory retrieval.34 Hence, our findings suggest that abnormal connectivity within the MTL subsystem may imply a dysfunctional encoding of episodic memory in patients with MMF, but this may not be specific because the patients with MDD_noMF also showed abnormality in this subsystem.

Limitations

Several limitations of this study should be taken into account. First, the sample size is relatively small. In future studies, we will expand the sample size to improve the reliability of findings. Second, some of the patients were taking antidepressant drugs, which might influence brain function. Future studies with first episodic, drug-naïve patients are encouraged. Third, this study applied a priori ROIs based on the brain altas. However, defining DMN by utilizing individual level approaches also needs to be explored in the future, such as the cortical parcellation approach,35 and the machine learning algorithms. 36

Implications

There are meaningful clinical implications from our findings. Our findings support the association between abnormal connectivity of the DMN subsystems with depression, and provide deeper insight into the pathophysiology associated with mixed features and depression symptoms. A related key finding of the study showed the patients with MMF presented abnormal functional connectivity within the DMN core subsystem, which might help deepen the understanding of the emotion regulation of the MMF. Furthermore, this insight may have key implications for the development and refinement of therapeutic interventions and diagnosis for patients with MMF.