Shared and Disorder-Specific Alterations of Brain Temporal Dynamics in Obsessive-Compulsive Disorder and Schizophrenia

Abstract

Background

Obsessive-compulsive disorder (OCD) and schizophrenia have distinct but also overlapping symptoms. Few studies have examined the shared and disorder-specific disturbances in dynamic brain function in the 2 disorders.

Study Design

Resting-state functional magnetic resonance imaging data of 31 patients with OCD and 49 patients with schizophrenia, all untreated, and 45 healthy controls (HCs) were analyzed using spatial group independent component (IC) analysis. Time-varying degree centrality patterns across the whole brain were clustered into 3 reoccurring states, and state transition metrics were obtained. We further explored regional temporal variability of degree centrality for each IC across all time windows.

Study Results

Patients with OCD and patients with schizophrenia both showed decreased occurrence of a state having the highest centrality in the sensorimotor and auditory networks. Additionally, patients with OCD and patients with schizophrenia both exhibited reduced dynamics of degree centrality in the superior frontal gyrus than controls, while dynamic degree centrality of the cerebellum was lower in patients with schizophrenia than with OCD and HCs. Altered dynamics of degree centrality nominally correlated with symptom severity in both patient groups.

Conclusions

Our study provides evidence of transdiagnostic and clinically relevant functional brain abnormalities across OCD and schizophrenia in neocortex, as well as functional dynamic alterations in the cerebellum specific to schizophrenia. These findings add to the recognition of overlap in neocortical alterations in the 2 disorders, and indicate that cerebellar alterations in schizophrenia may be specifically important in schizophrenia pathophysiology via impact on cerebellar thalamocortical circuitry.

Introduction

Obsessive-compulsive disorder (OCD) and schizophrenia have distinct clinical symptoms, but also have overlapping symptom features and cognitive impairments. These disorders also have overlaps in cognitive functions,¹ neurobiology, and gene associations,² often have a chronic course and have similar age distribution of onset.³ Perhaps most importantly, both disorders have the relatively distinct feature of rigidly holding onto beliefs despite low credibility and maladaptive consequences. Despite their established overlapping features, the overlapping and distinctive alterations in brain function associated with the disorder remain poorly delineated.

Previous neuroimaging studies have found that schizophrenia and OCD have overlapping as well as distinct associated alterations in brain structure and function. Of interest given the reduced flexibility of thinking associated with the disorders, both have been shown to have decreased gray matter volume in dorsal anterior cingulate cortex (ACC).⁴ At the network level, one study reported that patients with schizo-obsessive comorbidity (SOC), schizophrenia, and OCD all exhibited increased static functional connectivity (sFC) between subregions of the default mode network (DMN) and executive control networks (ECN) relative to healthy controls (HCs).⁵ Patients with OCD and schizophrenia share a failure in internal source monitoring that may contribute to the phenomenological overlap between delusions/hallucinations and intrusive obsessional thoughts.⁶ At the neurobiological level, reduced sFC and structural connectivity involving auditory network (AUD) and sensorimotor network (SMN)^7–9 might contribute to these alterations as they provide important inputs for self-monitoring and the discrimination of self and other-initiated events.^10,11 Disorder-specific brain imaging alterations have also been identified. Qin et al. explored white matter network topological characteristics and found that patients with schizophrenia exhibited reduced nodal strength and efficiency in the striatum and visual/auditory cortices relative to OCD.¹² Yu et al. reported dissimilarities of spontaneous brain activity in multiple regions (indicated by amplitude of low-frequency fluctuation, regional homogeneity [ReHo], and static degree centrality) between schizophrenia and OCD.¹³ Sustained attention deficit in schizophrenia has been correlated with reduced sFC between left medial prefrontal cortex (mPFC) and bilateral ACC, while in OCD it was related to decreased sFC between right mPFC and left superior frontal gyrus (SFG).¹⁴ Additionally, increased sFC between DMN and middle temporal gyrus has been reported in patients with SOC and schizophrenia, while decreased sFC between these regions was reported in patients with OCD.⁵ More severe brain structural^15,16 and functional abnormalities⁵ as well as clinical symptom severity¹⁷ have been reported in patients with SOC than schizophrenia or OCD alone, suggesting potential additive factors of comorbidity on brain features.

Connectome models have driven home the point that brain regions are highly interactive, and that the brain nodes function together as components of differentiated networks.^18,19 The role of brain regions in the broader brain connectome has been evaluated using graph-theory-based analysis. Using this approach, disrupted sFC topology has been shown in both schizophrenia and OCD, and related to clinical symptoms of the 2 disorders.^20,21 Degree centrality is a graph-theory-based metric that reflects the relative “importance” of nodes in brain networks. A brain region with a high degree of centrality has more robust sFC than other regions. Aberrant static degree centrality has been documented in schizophrenia²² and correlated with the reduction in Positive and Negative Syndrome Scale (PANSS) scores after antipsychotic treatment.²³ Degree centrality alterations have also been found in OCD, with the presence of such alterations allowing differentiation from HCs with an accuracy of 88%.²⁴ These points highlight the novelty and importance of graph/connectome modeling approach and its potential to advance understanding of brain alterations in psychiatric disorders, and the need for studies of untreated patients to limit the diverse effects of psychotropic medications in studies aiming to understand illness-related alterations in brain function.^25,26

In recent years, neuroimaging researchers have focused more on dynamic than static (or average) connectivity, based on the recognition that the brain network is a dynamic, time-varying system.^27–30 Most previous FC studies comparing schizophrenia and OCD evaluated static connectivity, which evaluated average connectivity rather than its dynamic organization over time. But studies of dynamic connectivity in individual disorders have been informative. Patients with schizophrenia have been shown to spend more time in states with weak connections between networks,^31,32 and the reduced transitions between functional brain states were associated with clinical symptom severity.³³ Patients with OCD have been reported to spend more time in a state characterized by strong within-network connections and less time in an integrated state characterized by the predominance of between-network connections.³⁴ Wang et al. reported that dynamic degree centrality can help differentiate responders and nonresponders to antipsychotic treatment in schizophrenia.³⁵ And, Linda et al. found higher dynamic coherence of dorsolateral prefrontal cortex (DLPFC) with other brain regions before treatment was related to greater distress reduction after repetitive transcranial magnetic stimulation treatment in patients with OCD.³⁶

Given these considerations and the lack of previous studies comparing dynamic FC (dFC) alterations of patients with schizophrenia and OCD directly, the present study used graph modeling and dFC analysis to identify transdiagnostic and disorder-specific alterations of the functional brain connectome in OCD and schizophrenia using resting-state functional magnetic resonance imaging (rfMRI). We studied untreated patients with either disorder, and patients without psychiatric comorbidities, to clarify their similarities and differences without potential confounding from these features. We focused on 2 aspects of brain dynamics: The time-varying degree centrality patterns of the whole brain and temporal variability of degree centrality for each brain region. Based on previous findings with studies of one of the target disorders,^37,38 we hypothesized that subcortical alterations would be more specific to schizophrenia, and that neocortical alterations of ACC, DMN, AUD, and SMN would be observed in both schizophrenia and OCD.

Methods

Participants

This study was approved by the research ethics committee of West China Hospital of Sichuan University, and informed written consent was obtained from all participants. Diagnoses of OCD and schizophrenia were determined by experienced clinical psychiatrists using Structured Clinical Interview for DSM-IV (Patient Edition). Patients with OCD and schizophrenia had no psychiatric comorbidities. Only participants who completed MRI scanning, met criteria of minimal head motion, and whose brain MRI passed image quality control procedures were included. In OCD, 17 patients were drug-naïve and 14 patients had previously received psychiatric medications but had been medication free for at least 2 weeks. All 49 patients with schizophrenia were treatment-naïve and 45 HCs were recruited. The 3 groups did not differ significantly in age (P = .386), sex (P = .843), or years of education (P = .140) (table 1). Additional information is presented in supplementary material.

Table 1.

The Demographic and Clinical Information of All Participants

	OCD (n = 31)	SZ (n = 49)	HC (n = 45)	F/χ2/t value	P value
Gender (number)	19M, 12F	27M, 22F	25M, 20F	0.341a	.843
Age (years)	27.1(9.4)	24.6(8.0)	26.6(9.3)	0.960b	.386
Education(years)	13.7(2.8)	12.3(2.9)	12.5(3.5)	1.995b	.140
Duration (years)	6.0(5.4)	1.6(1.9)	—	27.819	<.001*
YBOCS	22.9(5.2)	—	—	—	—
HAMA	7.9(3.1)	—	—	—	—
HAMD	9.8(2.7)	—	—	—	—
PANSS total	—	88.3(18.3)	—	—	—
PANSS positive	—	23.4 (7.2)	—	—	—
PANSS negative	—	19.2 (8.3)	—	—	—
PANSS general	—	45.7 (10.1)	—	—	—

^achi-square test. ^bANOVA, ^c2-sample t-test; values are given as mean (SD).

Abbreviations: OCD, obsessive-compulsive disorder; SZ, schizophrenia; HC, healthy controls; YBOCS, Yale-Brown Obsessive-Compulsive Scale; HAMA, Hamilton Anxiety Rating Scale; HAMD, Hamilton Depression Rating Scale; PANSS, Positive and Negative Syndrome Scale.

MR Data Acquisition and Preprocessing

High-resolution 3-dimensional structural MRI and rfMRI data were collected using a 3.0T MRI system (EXCITE, GE) with an 8-channel phased array head coil. Parameters of MRI scanning sequences and data preprocessing are detailed in supplementary material. To reduce the influence of head motion, participants with a maximum displacement of more than 2mm, maximum rotation greater than 2°, and a mean framewise displacement larger than 0.2mm were excluded. The mean framewise displacement did not significantly differ among the 3 groups (F = 0.581, P = .561).

Independent Component Analysis

After MR data preprocessing, we performed spatial group independent component analysis (GICA) using the GIFT toolbox (mialab.mrn.org/software/gift/) to identify intrinsic connectivity networks (details in supplementary material). Reliability and stability of the Informax GICA algorithm in ICASSO³⁹ were ensured by repeating the algorithm 20 times, and the most central run was used to reconstruct subject-specific time courses and spatial maps of each independent component (IC) with the GICA back reconstruction algorithm.⁴⁰ The group independent components (ICs) of the 20 runs were clustered, and then the quality index (Iq) of IC was calculated (supplementary table S1). ICs with Iq > 0.7 were selected as prior work suggested.⁴¹ We selected 55 ICs and sorted them into 8 functional networks (figure 1): AUD, basal ganglia, DMN, ECN, language network, SMN, salience attention network (SAN), and visual network. The coefficient of determination (R2) between the ICs and the Stanford functional template as a criterion for assigning ICs to 1 of the 8 functional networks in this study, using the highest R2 of nodes with the 8 networks to assign nodes to networks. The peak coordinates and R² values of all selected ICs are provided in supplementary table S1. Fifty-five ICs were defined as functionally independent nodes, with the sFC between pairs of ICs (Pearson’s correlation coefficient) as edges used for the calculation of degree centrality.

Fig. 1.

The 55 independent components (ICs) derived from all subjects in the present study were divided into 8 functional networks: Auditory network (AUD), default mode network (DMN), executive control network (ECN), language network (LAN), salience network (SAN), subcortical network (SC), sensorimotor network (SMN), and visual network.

Degree Centrality of Each Component for All-time Windows

The sliding-window approach was used to measure time-varying changes in degree centrality for the 55 identified ICs. We chose a 22TR window (44s) because it has been suggested that time windows of 30 to 60 seconds can successfully capture patterns of resting-state fluctuations of FC.⁴² To explore the dynamics of degree centrality for each IC, we measured the fluctuation of degree centrality across all sliding windows using GRETNA software (www.nitrc.org/projects/gretna). Instead of using a single sparsity threshold, we applied a wide range of sparsity levels, and computed the area under the curve of estimated degree centrality across all sparsity thresholds. The selected thresholds ranged from 0.07 to 0.32 (with an interval of 0.01) and were used for the calculation of area under the curve of degree centrality of the ICs for each time window. For every sparsity threshold, the degree centrality for each IC was measured by the number of edges connected to this IC in the binarized FC matrices. Methodological details are included as supplementary material.

Dynamics of Degree Centrality

Measuring the time-varying degree centrality patterns evaluates how the whole-brain degree centrality pattern changes over time, and measuring temporal variability of degree centrality for each IC makes it possible to understand each brain region’s dynamic fluctuation of nodal centrality.

To assess time-varying degree centrality patterns, k-means clustering was performed on the degree centrality pattern across all time windows for all participants. The k value was varied from 2 to 10 to identify the optimal number of brain states. To reduce the potential bias of the initial random selection of cluster centroids, the k-means clustering algorithm was iterated 100 times for each k value. The elbow method was used to evaluate the number of meaningful clusters of FC patterns that comprise brain states. This approach calculates the sum of squared errors against each cluster number k (here, k = 2–10), and then plots the curve of sum of squared errors against each corresponding cluster number to identify the elbow point of the curve, which basically represents the point where additional clusters are less robust.⁴³ Using the “elbow criteria,” the optimal k value was determined to be 3 in the present study (supplementary figure S2). We used the cluster centroids of all participants to represent the 3 reoccurring degree centrality states during the resting-state study (see results and supplementaryfigure 2A). To characterize each degree centrality state, we calculated the mean degree centrality for each network of the 3 cluster centroids (figure 2B). Then we assessed 3 different state transition metrics to quantify the temporal properties of the degree centrality states: (1) fractional time, which represents the proportion of time window belonging to each state, (2) mean dwell time, which measures the average length of time spent in each state before switching into other states, and (3) number of transitions, which counts the total number of switches between different states over time. To quantify temporal variability of degree centrality for each IC, the coefficient of variation of regional degree centrality across all time windows was calculated. Higher coefficient of variation represents higher fluctuation of brain region centrality. To evaluate the consistency and validity of the dynamic degree centrality analysis at different window sizes relative to the 22TR window used, we also examined the data using a 30TR window size repeating the above analyses. Findings using 22TR window are well replicated in this validation analysis (details in supplementary material).

Fig. 2.

Cluster centroids of time-varying degree centrality patterns in all subjects (A). The radar map of mean degree centrality of each network for 3 states (B). Cluster centroids of time-varying degree centrality patterns in obsessive-compulsive disorder (OCD), schizophrenia (SZ), and healthy controls (HC), respectively (C). Abbreviations of brain networks see figure 1.

Statistical Analysis

Due to the non-normal distribution of dynamic degree centrality measures, the independent samples Kruskal–Wallis one-way ANOVA and post hoc pair-wise tests were used to assess group differences of all metrics. Significance was set at P < .05 for these statistical analyses, with the false discovery rate correction used to correct for multiple comparisons of degree centrality state transition measures. To test for the significance of group differences in the larger number of nodal degree centrality dynamic properties, the false positive adjustment approach was used to control the false-positive rate to 1 per analysis⁴⁴ and Bonferroni correction was used for post hoc tests. There were no differences in both dynamic degree centrality states and temporal variability of degree centrality between drug-naïve patients and drug-washout patients with OCD. Thus, we pooled the 2 subgroups for analysis. To explore the relationships between altered dynamic degree centrality measures and clinical symptom severity within each patient group, we performed Spearman partial correlation analyses, with age, sex, and education years as covariates. We set P < .05 (nominal significance) for these exploratory correlational analyses given a large number of correlations examined, and describe them as suggestive effects rather than significant effects.

Results

Group Differences in Temporal Properties of Degree Centrality States

We obtained 3 time-varying degree centrality states in an analysis of all participants based on k-means clustering: State 1 occurred most frequently (43.5%), which was characterized by the highest degree centrality in DMN; state 2 occurred less frequently (30.4%), and showed the highest degree centrality in SAN and ECN; state 3 (26.1% of the time), and showed the highest degree centrality in SMN and AUD (figure 2A–B). For OCD, HCs, and schizophrenia, the total occurrences of the 3 states were 46.2%, 41.6%, and 43.6% for state 1, 36.7%, 24.7%, and 31.8% for state 2, and 17.1%, 33.7%, and 24.6% for state 3, respectively (figure 2C). There were significant differences in the occurrences of state 3 among the 3 groups (P = .006, P_FDR = .018). The occurrences of the other 2 states were not different among the 3 groups.

For state transition metrics, patients with OCD showed decreased fractional time in state 3 compared with HCs (F = 10.380, P = .006; post hoc test for OCD vs HCs: P = .001, P_FDR = .021). OCD patients exhibited decreased dwell time in state 3 relatives to controls (F = 10.518, P = .005; post hoc test for OCD vs HCs: P = .001, P_FDR = .021; post hoc test for schizophrenia vs HCs: P = .045, P_FDR = .315, respectively), though the P value of the schizophrenia group was significant with a nominal significance threshold. We did not find significant group differences in other state transition metrics (figure 3).

Fig. 3.

Group comparisons of state transition metrics. Patients with obsessive-compulsive disorder (OCD) showed decreased fractional time in state 3 compared with healthy controls (HC) (A). Patients with OCD and schizophrenia (SZ) both showed decreased dwell time in state 3 compared with HC (B). No significant group difference in number of transitions (C). The * represents uncorrected P < .05, and the ** represents corrected P < .05.

Group Differences in Temporal Variability of Degree Centrality

We observed significant group differences in temporal variability of degree centrality in IC38 (located in SFG of right DLPFC that belonged to SAN, F = 9.532, P = .009) and IC76 (located in the cerebellum and belonged to SMN, F = 8.083, P = .018). Post hoc analysis showed that patients with OCD (P = .003, P_Bonferroni = .009) and nominally in schizophrenia patients (P = .034, P_Bonferroni = .102) showed lower temporal variability of degree centrality in IC38 than controls. Patients with schizophrenia also showed significantly lower temporal variability in IC76 compared with HCs (P = .032, P_Bonferroni = .096) and patients with OCD (P = .009, P_Bonferroni = .027) (Figure 4A–B). There was no difference in dynamic degree centrality in IC76 between OCD and controls.

Fig. 4.

Group comparisons of temporal variability of degree centrality (A). Spatial maps of independent component (IC) 38 and IC76 (B). Significant correlations between temporal variability of degree centrality for IC38 and obsessive-compulsive score in patients with obsessive-compulsive disorder (OCD) (C). Significant correlation between temporal variability of degree centrality for IC76 and positive symptom score of Positive and Negative Syndrome Scale (PANSS) in schizophrenia (SZ) (D). The * represents uncorrected P < .05, and the ** represents corrected P < .05.

Correlation Analysis

Altered state transition metrics did not show significant correlations with clinical symptom severity, illness duration, or age of patients with OCD or schizophrenia. Temporal variability of degree centrality for IC38 showed a nominally significant negative correlation with symptom severity in OCD (r = −0.477, P = .010, figure 4C), and temporal variability of degree centrality for IC76 showed a positive nominally significant relationship with positive symptom score of PANSS in schizophrenia patients (r = 0.424, P = .003, figure 4D). Altered temporal variability of degree centrality did not correlate with illness duration, or age of patients with OCD or schizophrenia.

Discussion

In this study, patients with OCD and schizophrenia both spent less time in a state characterized by the highest degree centrality in sensory and sensorimotor systems and shared decreased temporal variability of degree centrality in the SFG than controls. Patients with schizophrenia showed significantly lower temporal variability of degree centrality in the cerebellum than HCs and patients with OCD. Nominally significant correlations with symptom severity were observed in both disorders. We did not demonstrate alterations of dynamic degree centrality in DMN or ACC, which may be ascribed to the difference in sample size, drug effect in prior studies, illness status, or analytic method. One possibility is that average FC is more reduced in these regions than the dynamic variability of FC, which would suggest a general reduction in connectivity with greater preservation of dynamic covariation of activity across brain nodes in the identified brain states.

Temporal Properties of Degree Centrality States Indicating Shared Alterations

Using k-means clustering, we identified 3 reoccurring degree centrality states. Compared with HCs, both patients with OCD and schizophrenia spent less time in state 3, characterized by the highest degree of centrality in SMN and AUD. Sensorimotor and auditory processing deficits are established in both disorders.^45–49 Thus, alterations in these networks, both from past neurophysiological studies and our current dFC findings, suggest shared pathophysiological alterations in the 2 disorders.

In our study, decreased dwell time in the state characterized by the highest centrality of brain regions of SMN and AUD was a common abnormality for the 2 disorders. Patients with schizophrenia have shown decreased sFC between sensory networks and decreased dwell time in the state characterized by strong connections between these networks.³¹ Patients with OCD have displayed decreased sFC in sensory networks^50,51 and lower centrality in the sensorimotor network⁵² compared with HCs. A recent brain dynamic study showed patients with OCD with higher symptom severity spend less time in the state with substantial between-network dFC (eg, between AUD, SMN, and visual network).³⁴ As a shared endophenotype for OCD and their unaffected first‐degree relatives, this FC state has been considered a neurophysiological biomarker of OCD.³⁴ Thus, previous brain sFC studies, previous dynamic findings with each single disorder, and our direct dFC comparison study converge in supporting the idea that a lower occurrence of the functional brain state characterized by high centrality in sensorimotor and primary sensory networks may represent a shared neurophysiological alteration in OCD and schizophrenia.

Shared and Disorder-Specific Alterations of Temporal Variability of Nodal Degree Centrality

In the analysis of nodal temporal variability of degree centrality, both patients with OCD and schizophrenia showed decreased temporal variability of degree centrality in the right DLPFC’s SFG (belonging to SAN) that is involved in the execution of multiple higher cognitive functions.⁵³ The SAN is crucial for the detection of salient stimuli and orienting attention to them, and switching between default mode and executive states based on task demands.⁵⁴ Previous studies have demonstrated that patients with schizophrenia and OCD manifest abnormal cortical thickness and volume^55,56 and sFC^57,58 in SFG. Similar to our results, previous studies suggested that patients with schizophrenia exhibited decreased dFC in dorsal prefrontal cortex during tasks⁵⁹ and during the resting state.⁶⁰ Our findings now demonstrate a similar severity of alteration in this region in schizophrenia and OCD.

Recent brain dFC studies of OCD or schizophrenia have demonstrated decreased brain dynamics involving areas of SAN,^61,62 consistent with findings of our direct comparison of the 2 disorders. Decreased temporal variability of nodal centrality in these regions in our study suggests reduced efficiency or flexibility in the integration of SAN with other brain networks. This may contribute to impaired behavioral flexibility and reduced insight in both OCD and schizophrenia.^61,63 In the case of OCD, our findings indicated that decreased temporal variability of degree centrality in this network is nominally related to level of obsessive-compulsive symptoms and behaviors.

We also observed that patients with schizophrenia exhibited lower dynamics of cerebellar degree centrality than were observed in OCD and HCs. Via reciprocal connections with the cerebral cortex through cerebellar-thalamic-cortical circuitry, this region is now recognized to play a vital role in cognitive and affective processes.^64,65 The role of cerebellar dysfunction in the psychopathology of schizophrenia was highlighted by the “cognitive dysmetria theory,” which postulated that impaired coordination and sequencing of both sensorimotor and mental processes in schizophrenia results from alterations in cerebellar circuitry.^66,67 Studies of individuals at high risk for psychosis indicated that cerebellar-cortical dysconnectivity predicts the onset of psychosis.⁶⁸ Reduced gray matter volume of the cerebellum has been associated with decreased dFC between the cerebellum and the frontal-parietal network in patients with schizophrenia.⁶⁹ The nominal association of decreased dynamics of cerebellar degree centrality with lower PANSS positive symptom scores in the present study reflects their clinical significance in schizophrenia.

Despite the differences between schizophrenia and OCD patients in cerebellar function observed in our study, some previous findings suggest some cerebellar alterations in OCD.⁷⁰ Also, we note that in our study, OCD patients did show a nominally significant trend of increased temporal variability of cerebellar centrality compared with HCs. A task-based study using the Stroop response conflict task reported reduced activation in patients with OCD compared with HCs.⁷¹ A meta-analysis provided a quantitative overview of ReHo findings in OCD and reported lower ReHo in the cerebellum in patients with OCD.⁷² Thus, while alterations in cerebellar dFC appear to be greater in schizophrenia as our findings demonstrate, future studies are needed to clarify whether there are different or more subtle cerebellar alterations in OCD as well.

Limitations

First, the sample size in our study was not large, and some effects of interest were significant with a nominal threshold but not full correction for multiple comparisons, most notably the correlational analyses of brain-behavior associations. Second, while replication is required, we note that we validated our results using a different window size and obtained similar results, suggesting that our findings were stable regardless of the choice of window length. Third, the mean illness duration of patients with OCD was longer than the first-episode patients with schizophrenia. Illness duration differences or prior treatment in OCD patients may affect brain dynamics. Furthermore, whether the identified abnormalities persist after recovery from acute illness remains to be determined. Fourth, although the sliding-window method has been widely used to measure brain dynamics, controversy remains regarding the optimal strategy to best capture dynamic fluctuations.⁷³ Evaluating brain dynamics with different methods and metrics might contribute to future advances and clarifications. Fifth, we noted that the scanning time of the resting-state studies was briefer than is now recommended,⁷⁴ although, this is somewhat allayed by the fact that similar results were found for the 2 sliding-window sizes. Sixth, as with all studies using ICA, there is a risk of overfitting and independent replication is needed to confirm our ICA solution. Additionally, it should be noted that while the lack of psychiatric comorbidities and untreated status of patients are limitations concerning the generalizability of our findings to treated patients with diverse comorbidities, these features are also a strength of our study with regard to confidence of inferences about illness-related biology because confounding influences of comorbidity and drug effects do not impact our group comparisons. Furthermore, our study aimed to identify similarities and differences in brain function between OCD and schizophrenia, and we thus are not able to address the issue of patients meeting criteria for both disorders. Finally, we performed disorder-specific clinical ratings of both patient samples, using rating scales widely used for each disorder. Future studies incorporating identical clinical ratings in both groups may better clarify symptom/brain function correlations across the disorders.

Conclusion

In summary, we found that patients with OCD and schizophrenia both showed a decreased occurrence of a state manifested as the highest centrality in sensory and sensorimotor systems. Some alterations of regional degree centrality dynamics have nominally significant associations with symptom severity. Our results extend previous studies by providing a dynamic brain network-level view of shared and unique brain dysfunction in OCD and schizophrenia using the same scanner and analytic framework. Our findings provide the novel demonstration that the ability of neocortical regions to shift their dFC with other regions is altered in generally similar ways in OCD and schizophrenia, but that the dynamic integration of cerebellum into the brain connectome is particularly altered in schizophrenia.

Funding

This study was supported by the National Natural Science Foundation of China (Grant Nos. 61876032 to Fei Li, 82102007 to Li Yao, 82001795 to Youjin Zhao, 81621003, 81761128023, and 82027808 to Qiyong Gong, and 81820108018 to John A. Sweeney and Qiyong Gong), National Key R&D Program of China (2022YFC2009900), and Sichuan Science and Technology Program (2021YFS0077 to Li Yao, 2022YFS0069 to Youjin Zhao, and 2023YFS0226 to Lizhou Chen).

Author Contributions

Fei Li conceptualized and designed the study. Lekai Luo, Qian Li, Yaxuan Wang, Ning He, Yuxia Wang, Wanfang You, Qian Zhang, Fenghua Long, Lizhou Chen, Youjin Zhao, and Li Yao contributed to literature search, data acquisition and analysis, and draft the manuscript. Lekai Luo, Qian Li, Yaxuan Wang, John A. Sweeney, Qiyong Gong, and Fei Li critically revised the manuscript, and contributed to the data analysis strategy and interpretation. All authors approved the final version of the manuscript.

Data Availability

The data that support the findings of this study are available from the corresponding authors upon reasonable request.