Altered Cerebellum Spontaneous Activity in juvenile Autism Spectrum Disorders Associated with Cognitive Functions

Background Autism Spectrum Disorders (ASD) is a neurodevelopment disorder. The cerebellum has been reported to be one of the key regions involved in ASD. However, the associations between the cerebellum and clinical traits remain unclear. Methods Here we performed Amplitude of Low Frequency Fluctuations (ALFF) analysis to detect the alterations of brain spontaneous activity in ASDs and explore the associations between spontaneous brain activity and clinical traits. Results Compared with normal controls, cerebellum crus 2 showed signicantly weaker average ALFF values. Other regions such as left cerebellum 6, cerebellum vermis 4 5, putamen, SMA and thalamus showed increased mean ALFF values. In ASD patients with SRS total score T above 59, the mean ALFF values of cerebellum vermis 4 5 was signicantly correlated with SRS total score T (r=0.175, P=0.031), SRS cognition score T (r=0.169, P=0.036) and SRS motivation score T (r=0.176, P=0.028). These ndings were not observed in other brain regions and in normal controls. Our study suggests a role of cerebellum in cognitive impairments in ASD and supports a mechanistic basis for the targeted treatment of ASD disorders.

The cerebellum has for a long time been thought as one important brain region functioning in motor learning and coordination (De Zeeuw and Ten Brinke, 2015; Manto, et al., 2012). There is accumulating evidence supporting the cerebellum as one of the key brain regions involved in ASD. Genetic factors in ASD were estimated to be approximately 80% (Bai, et al., 2019). Meanwhile,a few of risk genes have been recognized to be related to cerebellum abnormality in ASD (Aldinger, et al., 2013;Wang, et al., 2019). Moreover the ASD patients showed alterations in cerebellum volume and Purkinje Cell density (Skefos, et al., 2014;Webb, et al., 2009). The resting-state brain functional connectivity (FC) analysis also showed altered cerebro-cerebellar and within-network of cerebellum FCs in ASD (Khan, et al., 2015;Stoodley, et al., 2017). These ndings from genetic, anatomical and neuroimaging studies strongly suggest that the cerebellum plays an important role in neuropathophysiological mechanisms of ASD. However, the small sample size in most ASD studies limited the reliability of these ndings. Results from different studies were not exactly consistent. More importantly, few studies focused on the associations between brain regions and clinical traits (e.g., cognition, communication, motivation) in ASD.
The question how the cerebellum in ASD is associated with clinical symptoms has been a matter of interest in recent studies. A previous neuroimaging study has indicated that reductions of gray matter in ASD children in cerebellar lobule VII (Crus I/II) correlated with the severity of symptoms in social interaction and communication (D'Mello, et al., 2015). Regional Homogeneity (ReHo) analysis found that the ReHo value in cerebellum in ASD is signi cantly correlated to clinical trait scored by Social Communication Questionnaire (SCQ) (Dajani and Uddin, 2016). Some other studies, however, failed to detect the correlations between FC or spontaneous activity of cerebellum and symptoms of ASD (Carper, et al., 2015;Padmanabhan, et al., 2013). Therefore, the relationship between cerebellum and symptoms of ASD remains to be clari ed.
The aim of the present study was to explore the associations between spontaneous activity of cerebellum and the clinical traits of ASD. For that purpose, we applied Amplitude of Low Frequency Fluctuations (ALFF) analysis on resting state functional magnetic resonance imaging (fMRI) to explore the spontaneous brain activity divergences between ASD patients and healthy controls. Then, the ALFF values of region of interest (ROI), used to depict the spontaneous activity were extracted, and their correlations with symptoms of ASD were analyzed. Finally, the probable relations between cerebellumthalamus circuits and clinical traits were detected.

Participants
The data included in present study were acquired from the Autism Brain Imaging Data Exchange II (ABIDE II, http://fcon_1000.projects.nitrc.org/indi/abide/abide_II.html) (Di Martino, et al., 2017) dataset collected by 19 independent sites. We downloaded the whole dataset including 1114 subjects with resting-state functional magnetic resonance imaging(fMRI) data and their phenotypic data. Then, juvenile ASD patients (aged less than or equal to 18 years) were selected for the study. Age-and sex-matched healthy subjects were included as controls. The clinical severity of ASD was scored using the Social Responsiveness Scale (SRS) (T scores) (Constantino, 2013) where available and included ve subscales: awareness, cognition, communication, motivation, mannerism.

Data preprocessing
The scan protocols of the resting state fMRI vary across the 19 sites. Therefore, we preprocessed the neuroimages site by site using standardized pipeline with corresponding parameters. The preprocessing steps included removing the rst four volumes, slice-timing correction, motion realignment, spatial normalization using the EPI template to the stereotactic space of the Montreal Neurological Institute (MNI) with voxel size of 3 × 3 × 3 mm. Then, nuisance covariates including head motion parameters and linear trends were regressed out from the BOLD signals. Finally, we performed temporal bandpass ltering (0.01-0.08 Hz) across time series. These steps were conducted using a MATLAB toolbox Data Processing & Analysis for Brain Imaging (DAPABI) (Yan, et al., 2016).

Amplitude of Low Frequency Fluctuations (ALFF) analysis
After data preparation, the ALFF analysis on each subject's data was performed using DAPABI (Yan, et al., 2016). The ALFF analysis on fMRI data has been widely applied to measure the magnitude of the energy from Blood oxygenation level dependent (BOLD) signal intensity and indirectly depict the intensity of regional spontaneous brain activity (Deng, et al., 2016;Lu, et al., 2014;Tu, et al., 2015) in resting state.
Such relative activity of the local brain area is caused by the rhythmic activity of brain region functionally related to other brain regions. Power spectra of the time series were calculated and the sum of amplitudes within the low frequency band (.01-.08 Hz) was computed for each voxel. Then the square root was obtained at each power spectrum frequency. The ALFF value of each voxel was taken as average square root across the interest band and was divided by the mean ALFF value within the brain mask to obtain a standardized value (Yu-Feng, et al., 2007). The location assignments of ALFF differences between ASDs and NCs were done by using Anatomical Automatic Labeling template 2(AAL2) (Rolls, et al., 2015).
The ALFF values of regions of interest (ROIs) were extracted to characterize the spontaneous activity of ROIs. The ALFF values of ROIs for each subject were extracted to depict spontaneous brain activity.

Correlation analysis
To explore the associations between clinical severity and brain spontaneous activity, Pearson correlation analysis between the extracted ALFF values of ROIs and SRS T scores were performed.

Statistical analysis
To test the spontaneous brain activity differences between ASDs and NCs, we carried out ALFF analysis and used two-sample t test and two tailed threshold free cluster enhancement (TFCE) correction with Permutation Analysis of Linear Models (PALM) permutation test (P < 0.005, number of permutations 1000) (Chen, et al., 2018;Winkler, et al., 2016). The edge cluster connectivity criterion, rmm = 5, cluster size > = 50 voxels.

ALFF changes of cerebellum regions between ASDs and NCs
To test the spontaneous brain activity differences between ASDs and NCs, ALFF analysis was performed.
Six clusters were obtained and the locations were assigned by using AAL2 template. Signi cant differences were located in extensive cerebellar regions involving two clusters: cluster 1(cerebellum_crus2_R, crebelum_crus2_L) and cluster 2 (cerebellum 4 5 L, cerebellum 6L and vermis 4 5).

ALFF values of spontaneous brain activity was correlated with Clinical trait
Considering that four subregions of cerebellum (cerebellum crus 2, left cerebellum 4 5, left cerebellum 6, cerebellum vermis 4 5),putamen, SMA and thalamus showed signi cantly altered ALFF in ASDs, we hypothesized that these changes in ALFF, depicting the spontaneous brain activity might be associated with clinical traits. Therefore, to examine the relationship between alteration of ALFF and clinical trait, Pearson correlation analyses were performed between SRS T scores including SRS total T score, SRS awareness T score, SRS cognition T score, SRS communication T score, SRS motivation T score and SRS mannerisms T score and mean ALFF values of these ROIs. ROIs regions including left cerebellum crus 2, left cerebellum 4 5, left cerebellum 6, cerebellum vermis 4 5, putamen, thalamus and SMA were de ned using AAL2 template (Rolls, et al., 2015). The mean ALFF values of each ROI for individuals were extracted from corresponding ALFF results.
We noted that in juvenile ASD patients with SRS total T scored above 59 (n = 258), the mean ALFF value of cerebellum vermis 4 5 was signi cantly correlated with SRS Total T(r = 0.175,P = 0.031, FDR correction), SRS cognition T (r = 0.169,P = 0.036) and SRS motivation T(r = 0.176, P = 0.028), These ndings were not obtained in NC group. After FDR correction, the ALFF values of other regions including left cerebellum crus 2, left cerebellum 4 5, and left cerebellum 6, Putamen, Thalamus and SMA showed no signi cant correlations with SRS scores. Moreover, in NC group, we found the ALFF values of left cerebellum crus 2 was dramatically negatively related to SRS motivation score T (r = − 0.158, P = 0.014), which was not found in ASD patients (Fig. 2, Supplementary Table 2,3).

Medication effect and ALFF values of ROIs
Considering that medication might have potential impacts on spontaneous brain activity of ROIs, we divided the ASDs into two groups according to their medication status. We found that the ALFF differences between ASDs with medications (n = 84) and ASDs without medication (n = 172) did not survive after FDR adjusted.

Reproducibility
We assessed reproducibility through a simple strategy. We randomly selected (with Matlab) another two sex-and age-matched groups and carried out ALFF analysis. The results of ALFF analysis remained the same with same settings as previous analysis (Supplementary Fig. 1). Then, we de ned the ROIs based on AAL2 atlas, not based on our own ALFF clusters.

Discussion
In the present study, we applied ALFF analysis to detect the alterations of spontaneous brain activity in ASDs and explored the associations between spontaneous brain activity and clinical trait. The different ALFF pattern can be related to different brain spontaneous activity patterns of the ASDs and NCs. The regions with different ALFF pattern were used as ROIs to investigate if the ALFF was related to the clinical severity of ASD. For that purpose, Pearson's correlation analysis was employed to demonstrate the probable correlations between spontaneous brain activity and clinical traits. Furthermore, subgroup analysis of ASDs suggested that spontaneous activity of the brain might be impacted by medication status. Finally, ALFF analysis with another randomly selected sample produced a similar ALFF map, which con rmed the high reproducibility of the present study. The combined ndings are expected to provide evidence for a functional role of cerebellum in ASD from the functional imaging level.
Based on our ndings from functional imaging data, we con rmed the important role of cerebellum in ASD, which is in accordance with previous fMRI studies (Itahashi, et al., 2015;Jack, et al., 2017). Importantly, the decreased ALFF values of cerebellum crus 2 and increased ALFF values of cerebellum 4 5 and cerebellum vermis 4 5 in ASD patients might suggested that the subregions of cerebellum have heterogeneous roles in ASD. That means the dysfunction of ASD might be partly attributed to the enhanced brain spontaneous activity in cerebellum 4 5, cerebellum vermis 4 5, cerebellum 6 and declined brain spontaneous activity in cerebellum crus 2. The subsequent correlation analysis supported this assumption: the ALFF of cerebellum vermis 4 5 was positively correlated with SRS total score T. Interestingly, we noted that the spontaneous activity of left cerebellum crus 2 was negatively related to SRS motivation score T in NCs, which might be caused by decreased spontaneous activity of left cerebellum crus 2 in ASDs. These ndings support a mechanistic basis for the targeted treatment of ASD related disorders. Excessive or insu cient spontaneous activity of subregions in cerebellum could induce disorder in ASD.
As we expected, the cerebellum spontaneous activity was associated with clinical severity and functional de cits, which is in good agreement with earlier ALFF analysis (Guo, et al., 2017) and task-dependent fMRI study (Murphy, et al., 2017). Speci cally, we found that SRS cognition score T was only positively related to cerebellum vermis 4 5, which were not obtained in NCs. This result, has not been presented in previous studies, indicated that excessive spontaneous brain activity in the cerebellum subregion: vermis 4 5 might play a key role in the changes of cognitive function in ASDs. Notably, the ALFF changes of other ROIs such as cerebellum 4 5 in ASD patients were not signi cantly correlated with any clinical trait scored with SRS, though cerebellum 4 5 showed a signi cantly increased ALFF in ASDs, compared with NCs. This result further supported the notion that no single pattern could be drawn to characterize the role of cerebellum in ASD.
The present ndings about the role of cerebellum subregions in cognitive function were in accordance with previous studies. Clinical study has con rmed that patients with cerebellum lesions might experience cerebellar cognitive affective syndrome (Argyropoulos, et al., 2020). The abnormal cerebellum activity has also been found in other disease with cognitive impairment such as Parkinson's Disease (Solstrand, et al., 2020). However, the organization of cognitive function was still an unsolved issue. The cerebellum crus 2 has been found to send and receive projections to (from) prefrontal area 46 (Bostan, et al., 2013). In addition, imaging studies have indicated that some subregions of cerebellum might functionally couple networks of the cerebral cortex (Buckner, 2013). Our results con rmed this speculation: the left Cerebellum Crus 2 showed decreased spontaneous activity, while the left cerebellum 6 and cerebellum vermis 4 5 showed increased spontaneous activity, compared with NCs. Notably, only the spontaneous activity of vermis 4 5 was found to be signi cantly correlated with SRS cognition score T. These ndings indicated that different patterns might be drawn for different subregions of cerebellum in cognitive function. We speculated that the anatomically and functionally heterogeneity together play a role in the cognitive function of cerebellum.
Besides cerebellum, some other brain regions including SMA, thalamus and putamen were also found to show signi cant changes in ALFF. According to previous studies, these brain regions might function in the form of functional circuits, such as cortico-basal ganglia-thalamic circuits (Nair, et al., 2013;Schuetze, et al., 2016). To explore the relations between FC of these regions and clinical traits, we have calculated the functional connectivity between the six ROIs (results not shown). However, we did not nd signi cant associations between functional connections of ROIs and clinical traits. Therefore, further studies are needed to investigate how these brain regions showing signi cant changes of ALFF, interact with each other in ASDs.
In subgroup analysis, it was noted that ASD patients with medication and ASD patients without medication showed no differences in brain spontaneous activities of ROIs after critical FDR correction, which means that the targets of drugs present used might be not linked to the ROIs including cerebellum, thalamus and SMA. However, there was study found that medication use might affect brain FC in ASDs (Linke, et al., 2017). Considering that the signi cant changes of brain spontaneous activity in ASDs, regulating brain spontaneous activity might be one important treatment mechanism of ASD medications. However, longitudinal studies should be devoted to investigating whether ALFF changes over the course of ASD treatments.

Conclusion
In conclusion, in this functional imaging approach based on ALFF analysis and Pearson's correlation analysis, we were able to demonstrate that ASD patients showed signi cantly alterations in spontaneous activity of cerebellum regions involving cerebellum crus 2, cerebellum 4 5, cerebellum 6 and cerebellum vermis 4 5. Moreover, the changes of spontaneous activities of cerebellum vermis 4 5 were signi cantly correlated with cognitive functions in ASD. Our study suggests a role of cerebellum in cognitive impairments in ASD and supports a mechanistic basis for the targeted treatment of ASD related disorders.