Dysconnectivity of the cerebellum and somatomotor network correlates with the severity of alogia in chronic schizophrenia

Recent fMRI resting-state findings show aberrant functional connectivity within somatomotor network (SMN) in schizophrenia. Moreover, functional connectivity aberrations of the motor system are often reported to be related to the severity of psychotic symptoms. Thus, it is important to validate those findings and confirm their relationship with psychopathology. Therefore, we decided to take an entirely data-driven approach in our fMRI resting-state study of 30 chronic schizophrenia outpatients and 30 matched control subjects. We used independent component analysis (ICA), dual regression, and seed-based connectivity analysis. We found reduced functional connectivity within SMN in schizophrenia patients compared to controls and SMN hypoconnectivity with the cerebellum in schizophrenia patients. The latter is strongly correlated with the intensity of alogia, i.e. poverty of speech and reduction in spontaneous speech, demonstrated by patients. Our results are consistent with the recent knowledge about the role of the cerebellum in cognitive functioning and its abnormalities in psychiatric disorders, e.g. schizophrenia. In conclusion, the presented results, for the first time clearly showed the involvement of the cerebellum hypoconnectivity with SMN in the persistence and severity of alogia symptoms in schizophrenia.


Introduction
Alongside prominent positive and negative psychotic symptoms and cognitive impairment, most schizophrenia patients also experience a range of distressing motor symptoms (e.g., involuntary movement, catatonia, psychomotor slowing; Walther and Strik, 2012).Although motor symptoms have been considered primarily as side effects of antipsychotic treatment, they are also present in drug naïve and first-episode patients (Peralta et al., 2010).To investigate neural underpinnings of the above-mentioned phenomenon, resting-state functional magnetic resonance imaging (fMRI) has often been used, focusing on functional connectivity, i.e. measure of synchronization between remote neural assemblies (Fingelkurts et al., 2005).So far, evidence of resting-state functional connectivity changes compared to neurotypicals in various parts of the motor system has been found, including increased functional connectivity within basal ganglia (Duan et al., 2015), decreased functional connectivity of the cerebellum with multiple cortical regions (Andreasen and Pierson, 2008;Collin, 2011), and increased functional connectivity of primary motor cortex with the thalamus (Li et al., 2016).
Furthermore, decreased functional connectivity within somatomotor network (SMN; Keyvanfard et al., 2023) consisting of precentral (PRG) and postcentral (POG) gyri and supplementary motor area (SMA; Uddin et al., 2019) has been found in schizophrenia patients.What is more, the hypoconnectivity of the SMN and other resting state networks (Huang et al., 2020;Keyvanfard et al., 2023) is also present in this clinical population.These reported connectivity measures are often correlated with the severity of both positive and negative symptoms or impairment of cognitive performance (Bernard et al., 2017;Harikumar et al., 2023).Recent findings additionally suggest that dysconnectivity (a widely used term that can mean both hyper-and hypoconnectivity; Pettersson-Yeo et al., 2011) within that network can be a transdiagnostic marker of schizophrenia, bipolar disorder, and major depression (Huang et al., 2020;Magioncalda et al., 2020).All in all, current empirical research suggests that, alongside aberrations in other resting-state networks (e.g.hyperconnectivity within default mode network; Li et al., 2019;Mingoia et al., 2012), the motor system connectivity disturbances seem to play an important role in the clinical image of schizophrenia.
The primary aim of this study was to validate between-group differences in functional connectivity of the somatomotor network (SMN) in chronic schizophrenia outpatients compared to neurotypical controls.Additionally, we sought to explore the relationship between aberrant brain connectivity and the severity of psychopathological symptoms, with the goal of inferring the origins or implications of the observed correlations between brain networks and psychiatric symptomatology in these patients.Recognizing the heterogeneous nature of schizophrenia, we avoided arbitrary selection of regions of interest (ROIs) based on previous studies or atlases.Instead, we adopted a data-driven approach to investigate functional connectivity patterns.Specifically, probabilistic independent component analysis (PICA) was employed to identify the SMN location and analyze within-network connectivity pattern differences between patients and healthy controls.Following this, seedbased functional connectivity analysis was conducted using brain regions identified in the previous step as seeds to detect between-group differences in their connectivity with other brain areas.To further elucidate these findings, we examined the strength of connectivity in relation to individual psychiatric scale scores, including measures of positive, negative, and disorganization symptoms.In assessing negative symptoms, we utilized the Brief Negative Symptoms Scale (BNSS), which, unlike previous resting-state studies that treated negative symptoms as a single variable (Bernard et al., 2017), allows for a more detailed examination of symptoms across different domains (cognitive, emotional, and motivational).

Subjects
The study involved two groups: 30 chronic schizophrenia patients (SCH) and 30 controls (CON) matched with respect to age and sex (Table I; the sample is described in detail elsewhere, Adamczyk et al., 2021).All participants signed the informed consent for the participation in the study.The study protocol was accepted by the Research Ethics Committee at the Institute of Psychology, Jagiellonian University, Krakow, Poland; following standards of the World Medical Association Declaration of Helsinki (2013).
Three subjects (two SCH and one CON) were removed from the further analysis due to excessive head movements in the magnetic resonance imaging (MRI) scanner.The main selection criterion was number of invalid scans exceeding 3rd Q + 3 IQR threshold.The groups did not differ significantly in mean head motion (p = 0.252)

MRI resting-state procedure
Before entering MRI scanner, participants were instructed about the safety issues in MRI and asked to relax, lie still, refrain from any head movement, and focus their gaze on the fixation cross during 10 min of the protocol.

Data preprocessing
Data were preprocessed using CONN software version 22a (Nieto-Castanon and Whitfield-Gabrieli, 2022).Functional and anatomical data were preprocessed using a flexible preprocessing pipeline (Nieto-Castanon, 2020): 1) removal of 30 initial scans in each subjects' functional run to account for time when written instruction was displayed and a short period of adaptation to the scanner for the subjects, 2) realignment with susceptibility distortion correction using fieldmaps (realign & unwarp procedure (Andersson et al., 2001) integrating fieldmaps for susceptibility distortion correction; all scans were coregistered to a reference image (first scan of the first session) using a least squares approach and a 6 parameter (rigid body) transformation), 3) slice timing correction, 4) outlier detection (ART toolbox -framewise displacement above 0.5 mm or global BOLD signal changes above 3 standard deviations), 5) direct segmentation (Tissue Probability Maps), 6) MNI-space normalization, and 7) smoothing with a Gaussian kernel of 6 mm FWHM.In addition, functional data were denoised using a standard denoising pipeline (Nieto-Castanon, 2020) including the regression of potential confounding effects characterized by white matter timeseries (5 CompCor noise components), CSF timeseries (5 CompCor noise components), motion parameters and their first order derivatives (12 factors), outlier scans (below 168 factors), session effects and their first order derivatives (2 factors), and linear trends (2 factors) within each functional run, followed by bandpass frequency filtering of the BOLD timeseries between 0.008 Hz and 0.09 Hz.CompCor noise components within white matter and CSF were estimated by computing the average BOLD signal as well as the largest principal components orthogonal to the BOLD average, motion parameters, and outlier scans within each subject's eroded segmentation masks.

Defining SMN and other networks
Probabilistic ICA (Beckmann and Smith, 2004) as implemented in MELODIC (Multivariate Exploratory Linear Decomposition into Independent Components) Version 3.15, part of FSL (FMRIB's Software Library, www.fmrib.ox.ac.uk/fsl) was performed to identify the general location and temporal characteristics of resting-state brain networks.The number of components was estimated to be 14 using the minimum description length criterium (Rissanen, 1978).

Recognizing regions of interest
The set of spatial maps from the group-average analysis was used to generate subject-specific versions of the spatial maps, and associated timeseries, using dual regression (Nickerson et al., 2017).We then tested between-group differences using FSL's randomise permutation-testing tool (Winkler et al., 2014) with Family-wise Error (FWE) correction on threshold-free cluster enhancement at alpha set to 0.05.

Seed-based connectivity
Regions that differed significantly between groups in the previous step were taken as seeds for further analysis in the CONN software.Seedbased connectivity maps (SBC) were estimated using the Fishertransformed bivariate correlation coefficients from a weighted General linear model (weighted-GLM; Nieto-Castanon, 2020).In each SBC map, coefficients were estimated separately for each seed and target voxels, modelling the association between their BOLD signal time-series.In order to compensate for possible transient magnetization effects at the beginning of each run, individual scans were weighted by a step function convolved with an SPM canonical hemodynamic response function and rectified.
Group-level analyses were performed using the General linear model (GLM).Voxel-level hypotheses were evaluated using multivariate parametric statistics with random-effects across subjects and sample covariance estimation across multiple measurements.Inferences were performed at the level of individual clusters.Cluster-level inferences were based on parametric statistics from Gaussian random field theory (Worsley et al., 1996).Results were thresholded using a combination of a cluster-forming p < 0.001 voxel-level threshold, and a False Discovery Rate (FDR) p < 0.05 cluster-size threshold (Chumbley et al., 2010).

Relationship between connectivity and psychiatric measures
Parameter estimates from dual regression and connectivity values between a seed and significant cluster were extracted from each subject and imported to JASP (Version 0.18.3;JASP Team, 2024) to calculate Spearman correlation coefficients with PANSS and BNSS subscales.We decided to correct for a number of clinical scales that we tested for PANSS and BNSS factorial models separately (7 comparisons for BNSS and 6 comparisons for PANSS).
Effect sizes for mean cluster connectivity between-group differences were estimated by calculating rank-biserial correlation (r rb ) as variables' distributions deviated from normal.

Recognizing SMN
All PICA group components were inspected by experienced neuroscientists (M.B., P.A.) One component closely resembled SMN.It displayed the strongest activation in PRG, POG, and SMA regions.We used the Personode software (Pamplona et al., 2020) to determine the similarity of networks we obtained with the predefined SMN template.The correlation between our IC components and the template was the highest for the component we identified as SMN (58 %).

Recognizing regions of interest for seed-based connectivity analysis
Between-group comparisons revealed significantly decreased functional connectivity of the left SMA within the SMN in SCH group (T = 4.64, p = 0.013, r rb = 0.98, cluster size = 172; peak MNI coordinates = − 2 − 18 44; Fig. 1A).This one cluster was selected for performing the seed-based correlations.

Relationship between connectivity and the clinical symptoms measures
When controlling for the illness duration and chlorpromazine equivalent of medication, the SMA connectivity within SMN from dual regression analysis did not correlate significantly with the clinical scales.Meanwhile, the mean cluster connectivity values between the left SMA and left cerebellar lobule VI from the seed-based analysis were significantly related to the clinical scores in SCH group (Table II).The SMA-VI lobule connectivity score correlated with a total score of BNSS, as well as the alogia, blunted affect, distress BNSS subscale scores, and the score obtained in the PANSS negative symptoms scale.However, only the BNSS alogia result survived the FDR correction for multiple comparisons (Fig. 1D).

Discussion
In the present paper, we investigated intrinsic functional connectivity of SMN, using the whole-brain data-driven approach.Moreover, we explored the relationship between seed-based connectivity of regions within the SMN and the severity of symptoms from various domains.
By applying the analytical pipeline involving ICA decomposition, dual regression and seed-based correlations analysis, we found hypoconnectivity within the SMN and its dysconnectivity with the cerebellum in a sample of patients with schizophrenia when compared with the neurotypical control group.
More specifically, we found that disrupted SMN-cerebellum connectivity (i.e., between the medial SMA and the left cerebellar lobule VI) was correlated significantly with the BNSS alogia subscale.Interestingly, it may indicate that disruption of SMN-cerebellum connectivity may be recognized as related to motor-based problems, resulting in enhancement of manifestation of specific negative symptoms, such as pronounced reduction in spontaneous speech, a narrowing of the range of speech, and deficiencies in speech content.
Our results are consistent with the current knowledge about the involvement of the cerebellum in language processing (De Smet et al., 2007;Mariën and Beaton, 2014;Zhang et al., 2023).The left VI cerebellar lobule that we report is known to be functionally connected to SMN at rest and involved in motor tasks (Guell et al., 2018).The general role of the cerebellum in cognitive processing and its abnormal functioning in schizophrenia has been an influential idea in the past decade.
The cerebellum is mainly responsible for the prediction of sensorimotor input and the correction of errors that arise during the planning and execution of movement (Albus, 1971;Blakemore et al., 2000;Marr, 1969).In the recent models of cerebellar function, predicting and detecting errors is extended beyond the motor domain to the cognitive domain, e.g.predicting the next word in a sentence (D'Mello et al., 2020;Leiner et al., 1991).We know that the cerebellum contributes to verbal fluency (Molinari and Leggio, 2016) which is also commonly impaired in schizophrenia (Gourovitch et al., 1996).Thus, the dysconnectivity of the cerebellum may correspond to abnormal adaptive feedback to the cortex concerning pattern changes and error information.Therefore, this error monitoring process seems to be extremely important in various language functions, which is in line with the direct cerebellar contribution to language hypothesis (Daum and Ackermann, 1995;De Smet et al., 2007).The reported SMN-cerebellum dysconnectivity requires future experimental studies, which might reveal the importance of such a phenomenon and its impact on alogia manifestation and persistence.Noteworthy, in light of our findings, we may consider alogia as a severe motor-based impairment of verbal fluency.Furthermore, our results are consistent with previous findings on the dysconnectivity of the cerebellum with cortical regions and the altered connectivity within cerebellum, both linked to the severity of negative symptoms in previous studies (Choi et al., 2023;Kim et al., 2014).Importantly, it has been shown that stimulation of the cerebellum with transcranial magnetic stimulation (TMS) in patients with altered cerebellar-prefrontal network connectivity was reported to cause a reduction in negative symptoms (Brady et al., 2019;Hua et al., 2022).One of the reasons why TMS therapy is effective in reducing negative symptoms may be the establishment of synchronization between SMN and the cerebellum.
Finally, it is worth mentioning that, most previous studies in their correlation analysis used just one general scale to measure negative symptoms (PANSS), so they were not able to provide a more precise insight into aetiology of its specific symptomatology.Therefore, our work provides a novel insight into the specificity of negative symptoms subscales and their relationships with altered brain connectivity, i.e.SMA-cerebellum and alogia.The small sample size is a main limitation of this study.It would be prudent to reanalyze the second step using a different dataset.Performing interconnected analyses on a single dataset may raise issues related to circular analysis which can inflate the risk of type one error.Although this is not a guaranteed outcome, it is important to be mindful of this possibility when evaluating the results.As such, our findings should be seen as preliminary, prompting additional research on bigger samples with specific clinical assessment.

Conclusions
Our study highlights the hypoconnectivity within the SMN and SMN dysconnectivity with the cerebellum.The latter connection is linked to alogia.The involvement of the cerebellum in language processing and its disrupted connectivity with the SMN underscore its potential role in the clinical manifestation of alogia.Further research on interventions targeting cerebellar connectivity, such as TMS, holds promise for alleviating negative symptoms in schizophrenia and revealing in more detail neural underpinnings of alogia.

Fig. 1 .
Fig. 1.Abnormal SMN-cerebellum connectivity and its relationship with BNSS alogia in schizophrenia.A) Between-group differences in SMN connectivity.FWEcorrected results at contrast CON > SCH with no threshold.Slices presented at MNI: x=− 2; y= − 18; z = 44.A statistically significant cluster at p < 0.05 is highlighted in white.The color bar represents 1 -p values.B) The group differences in SMN-cerebellum seed-based connectivity at contrast SCH > CON.Results with no threshold.A statistically significant cluster at p < 0.05 within grey matter is highlighted in white.The color bar represents T values.Slices presented at MNI: x=− 30; y= − 68; z= − 26.C) SMA-cerebellar lobule VI connectivity values violin plot in both groups.SCH connectivity clearly oscillates around zero while CON subjects show tendency toward negative functional connectivity.D) Scatterplot of SMA-cerebellar lobule VI connectivity values and BNSS alogia scores with density distribution of both variables.Abbreviations: SCHchronic schizophrenia outpatients, CONcontrols, BNSS -Brief Negative Symptoms Scale.

Table 1
Demographic and clinical data.

Table 2
SMA-cerebellar lobule VI mean cluster connectivity and clinical scales correlation analysis results.Spearman's rho partial correlation coefficients with bootstrapped (10,000 samples) confidence intervals are shown.Illness duration and chlorpromazine equivalent of medication are controlled.Abbreviations: CIconfidence intervals, FDRfalse discovery rate.* p < 0.05; ** p < 0.01.