RESEARCH
ARTICLE
Modifications of Resting State Networks in Spinocerebellar Ataxia Type 2 , MD,2 Amedeo Cervo, MD,1 Angela Marsili, Mbiol,2 Cinzia Valeria Russo, MD,2 Sirio Cocozza, MD,1* Francesco Sacca Sara Maria delle Acque Giorgio, MD,1 Giuseppe De Michele, MD,2 Alessandro Filla, MD,2 Arturo Brunetti, MD,1 and Mario Quarantelli, MD3 1
Department of Advanced Biomedical Sciences, University “Federico II,” Naples, Italy Department of Neurosciences and Reproductive and Odontostomatological Sciences, University “Federico II,” Naples, Italy 3 Biostructure and Bioimaging Institute, National Research Council, Naples, Italy
2
A B S T R A C T : Purpose: We aimed to investigate the integrity of the Resting State Networks in spinocerebellar ataxia type 2 (SCA2) and the correlations between the modification of these networks and clinical variables. Methods: Resting-state functional magnetic resonance imaging (RS-fMRI) data from 19 SCA2 patients and 29 healthy controls were analyzed using an independent component analysis and dual regression, controlling at voxel level for the effect of atrophy by co-varying for gray matter volume. Correlations between the resting state networks alterations and disease duration, age at onset, number of triplets, and clinical score were assessed by Spearman’s coefficient, for each cluster which was significantly different in SCA2 patients compared with healthy controls. Results: In SCA2 patients, disruption of the cerebellar components of all major resting state networks was present, with supratentorial involvement only for the default
Spinocerebellar ataxias (SCAs) are a set of autosomal dominant progressive neurodegenerative disorders. Their classification is essentially based on the underlying
-----------------------------------------------------------*Correspondence to: Sirio Cocozza, MD, Department of Advanced Biomedical Sciences, University “Federico II,” Via Pansini, 5—80131— Naples, Italy, E-mail: [email protected]
Funding agencies: The study was supported by a grant from ‘‘Associazione Italiana per la lotta alle Sindromi Atassiche (AISA) sez. Campania’’ to FS, and a grant from the Italian Ministry of Education, University and Research, project PRIN 2010-2011 20108WT59Y_007 to G.D.M. Relevant conflicts of interest/financial disclosures: Nothing to report. Author roles may be found in the online version of this article. Received: 2 February 2015; Revised: 13 April 2015; Accepted: 11 May 2015 Published online 12 June 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/mds.26284
1382
Movement Disorders, Vol. 30, No. 10, 2015
mode network. When controlling at voxel level for gray matter volume, the reduction in functional connectivity in supratentorial regions of the default mode network, and in cerebellar regions within the default mode, executive and right fronto-parietal networks, was still significant. No correlations with clinical variables were found for any of the investigated resting state networks. Conclusions: The SCA2 patients show significant alterations of the resting state networks, only partly explained by the atrophy. The default mode network is the only resting state network that shows also supratentorial changes, which appear unrelated to the cortical gray matter volume. Further studies are needed to assess the clinical sigC nificance of these changes. V 2015 International Parkinson and Movement Disorder Society
K e y W o r d s : functional MRI; spinocerebellar ataxia; default mode network; resting state networks; independent component analysis
genetic defect. Spinocerebellar ataxia type 2 (SCA2), the second most frequent ataxia worldwide after SCA3,1 is associated with a CAG repeat expansion on chromosome 12q23-24.1 in the gene encoding for ataxin-2, a protein involved in RNA splicing.2 SCA2 is characterized by progressive olivo-pontocerebellar atrophy. SCA2 patients clinically present ataxia, ophthalmoplegia, corticospinal involvement, cognitive dysfunction, and peripheral neuropathy.2 Onset of symptoms in SCA2 typically occurs in the fourth decade, with a disease duration between 10 and 15 y, and clinical presentation appears to be in close relationship to the CAG repetition length.3 Magnetic resonance imaging (MRI) studies have allowed in vivo assessment of the atrophy pattern in SCA2, using both voxel-based morphometry (VBM)
R E S T I N G
TABLE 1. Subjects’ demographic and clinical variables
Age (mean 6 SD) Sex (M/F) SARA (median) Triplets (median) DD (mean 6 SD) AAO (mean 6 SD)
HC
SCA2
40.3 6 10.8 (range, 18-83) 15/14 N/A N/A N/A N/A
37.3 6 16.1 (range, 19-62) 10/9 14 (range, 9-24) 42 (range, 35-51) 11.4 6 8.2 28.5 6 11.4 (range, 3-50)
HC, healthy controls; SCA2, spinocerebellar ataxia type 2; SD, standard deviation; SARA, Scale for the Assessment and Rating of Ataxia; N/A, not applicable; DD, disease duration; AAO, age at onset. Ages and DD are in years.
and tensor-based morphometry (TBM), which have shown significant gray matter (GM) loss, mainly affecting the cerebellar structures.4-12 Recently, TBM proved to be capable of demonstrating the progression of ponto-cerebellar atrophy in SCA2, supporting a possible role of MRI as biomarker in future trials.13 Conversely, studies of white matter microstructural alterations by mean of diffusion MRI have quite consistently shown both cerebellar and supratentorial involvement.5,7,14-16 Resting state functional MRI (RS-fMRI) is a powerful method for evaluating regional interactions that occur between brain structures when a subject is not performing an explicit task. Data-driven analysis of RS-fMRI data has consistently demonstrated the presence of preferential connectivity between specific cerebral structures, which are organized in stable and simultaneously operating networks. Abnormalities of many of these resting state networks (RSN), as detected by RS-fMRI, have been demonstrated in several neurodegenerative disorders, such as Alzheimer’s disease,17 Parkinson’s disease,18 depression,19 schizophrenia,20 Huntington’s disease,21,22 and epilepsy.23 To the best of our knowledge, only a few studies have assessed in SCA patients the functional connectivity (FC) integrity,24-27 although it is limited to specific structures through seed-based analysis. These studies suggested the presence of FC alterations, showing, besides the obvious cerebellar involvement, modifications in several supratentorial (both cortical and deep GM) structures. However, no studies have assessed systematically the alterations of the major RSNs in SCA patients using a model-free approach. Furthermore, FC alterations in SCA have not been assessed independently of the underlying atrophic changes. The aim of this exploratory study was to investigate the integrity of the major RSNs, as detected by independent component analysis (ICA; a fully data-driven method applied to RS-fMRI data), in a group of SCA2 patients, compared with a group of age- and
S T A T E
N E T W O R K S
I N
S C A 2
sex-matched healthy controls (HC), also assessing possible correlations of RSN alterations with clinical data. To discriminate between the changes caused by functional and structural alterations in the involved areas, FC strength within each of the detectable networks was assessed with and without controlling for the effect of atrophy by covarying for GM volume.
Methods Subjects Nineteen symptomatic SCA2 patients and 29 ageand sex-matched HCs were enrolled. All subjects were right-handed. The molecular test showed 32 or more CAG triplets in the ataxin-2 gene in all SCA2 patients. Disease duration (DD) and age at onset were retrospectively measured, based on the first occurrence of symptoms clearly related to the disease. In addition, within 1 week from the MRI acquisition, an expert neurologist examined the patients and recorded the scale for the assessment and rating of ataxia (SARA), consisting of eight items (range, 0-40, with higher score indicating more severe clinical outcome).28 The HCs had no history of neurological or psychiatric disorders, nor were they receiving treatment with medications active on the central nervous system. Demographic information of both SCA2 patients and HCs are listed in Table 1. This study was conducted on patients enrolled in a previous trial,29 approved from the local Ethics Committee (151/09), and registered at www.clinicaltrials. gov (NCT00998634) and EUDRACT (2009-01631720). All patients gave written informed consent.
MRI Data Acquisition All MRI studies have been carried out on a 3 Tesla MRI scanner (Trio, Siemens Medical Systems, Erlangen, Germany). Sequences included a morphological T1w volumetric acquisition and T2*-weighted images for RS-fMRI analysis, acquired within the same scanning session. Structural T1w volumes were acquired using a three-dimensional magnetization-prepared rapid gradient-echo sequence; axial planes; TR 5 1900 ms; TE 5 3.4 ms; TI 5 900 ms; Flip Angle 5 98; voxel size 5 1 3 1 3 1 mm3; number of slices 5 160). The T2*-weighted volumes were acquired using an echo-planar imaging sequence (axial planes; TR 5 2500 ms; TE 5 40 ms; 64 3 64 acquisition matrix; 30 slices; voxel size 5 3 3 3 3 4 mm3; gap 1 mm; 128 time points; acquisition time 50 2000 ). During the MRI study, the subjects were laying supine with the head lightly fixed by straps and foam pads to minimize head movement, and were asked to relax with eyes closed, without falling asleep.
Movement Disorders, Vol. 30, No. 10, 2015
1383
C O C O Z Z A
E T
A L
MRI Data Analysis Pre-processing of the functional data was performed using Data Processing Assistant for Resting-State fMR I (http://www.restfmri.net),30 which is based on Statistical Parametric Mapping (SPM8) (http://www.fil.ion. ucl.ac.uk/spm) and Resting-State fMRI Data Analysis Toolkit (http://www.restfmri.net).31 Briefly, it included removal of the first five time points (to allow for instability of the initial MRI signal, leaving 123 time points), motion correction, slicetiming correction, band-pass filtering (0.01 Hz < f < 0.1 Hz), co-registration to structural scan, spatial normalization to the Montreal Neurological Institute template, and spatial smoothing (6-mm Gaussian kernel). For all pre-processing steps, default SPM8 parameters were used. The SPM motion correction procedure realigns the subsequent volumes of each study to the first one, iteratively finding the translation and rotation parameters that minimize a least-squares cost function derived from the voxel-by-voxel intensity difference from the reference image, a least squares approach, and a sixparameter (rigid body) spatial transformation.32 This approach proved to be accurate in realigning fMRI volumes for motion correction purposes.33 From the motion correction procedure, the mean displacement for each brain volume was computed as the root-mean-square (RMS) of the translation parameters at each time point. Studies with a mean relative RMS of 0.15 or higher (according to Van Dijk et al.34) or with 1.5 mm or more or 1.5 degrees rotation along any axis were discarded. Five SCA2 patients and four HCs were excluded by the analysis because of excessive movements during the fMRI scan. In the remaining 39 studies, a residual significant difference in the RMS was present between the two groups (0.035 6 0.024 in HC and 0.061 6 0.029 in SCA2, P < 0.01). Accordingly, in the following analysis RMS was added as nuisance covariate. Preprocessed volumes were then analyzed by using a temporal-concatenation spatial ICA approach,35 implemented in MELODIC (multivariate exploratory linear decomposition into independent components) Version 3.12, part of FSL (FMRIB’s Software Library, www. fmrib.ox.ac.uk/fsl). Pre-processed data were whitened and projected into a 3-dimensional subspace by using probabilistic principal component analysis in which the number of dimensions was estimated by using the Laplace approximation to the Bayesian evidence of the model order.35,36 The whitened observations were decomposed into sets of vectors that describe signal variation across the temporal domain (time-courses), the session/subject domain, and across the spatial domain (maps) by optimizing for nonGaussian spatial source distributions using a fixed-point iteration technique.37 Estimated component maps were
1384
Movement Disorders, Vol. 30, No. 10, 2015
divided by the standard deviation of the residual noise and thresholded by fitting a mixture model to the histogram of intensity values.35 MELODIC provided 20 independent components, among which those best matching the 10 major RSN templates published in Smith et al.38 were used for subsequent analysis. The correspondence between the components and the RSNs was defined based on cross-correlation of voxel values using the fslcc routine. Only the networks providing r > 0.5, and further confirmed by visual inspection to spatially overlap published templates, based on the descriptions available from reference studies,38,39 were further considered for analysis. The selected components were subsequently fed into a dual-regression analysis.40 Accordingly, for each subject, each RSN-specific mean time course is extracted and then used as voxelwise regressors in a linear model against the individual fMRI sets, to obtain subjectspecific spatial z-score maps for each RSN. The RSN maps were then compared between SCA patients and HCs by using a general linear model with permutation testing (5,000 permutations), including for each subject as nuisance covariates age, sex, and the average RMS to remove potential residual movement effects. This second-level analysis was carried out both with and without correction for atrophy by covarying at a voxel level, using the GM probability maps of each patient as voxel-dependent covariates. The GM probability maps were obtained by using the fast diffeomorphic registration algorithm (Diffeomorphic Anatomical Registration using Exponentiated Lie algebra; DARTEL),41 implemented in the Statistical Parametric Mapping software package (SPM8 - http://www.fil.ion. ucl.ac.uk/spm Wellcome Trust Centre for Neuroimaging, University College London). For all of the DARTEL preprocessing steps, the default SPM8 parameters were used. The normalized GM maps (2 3 2 3 2 mm3 voxel size) provided by DARTEL were visually assessed, to ensure good quality of the segmentation and normalization, and then smoothed using a 6–mm FWHM isotropic Gaussian kernel to match the resolution of the RS-fMRI images used for RSN definition. To characterize the distribution of GM atrophy in this cohort of patients, a VBM analysis was carried out on normalized GM maps (see Supplemental Data). When comparing the two groups, both contrasts (HC > SCA2 and HC < SCA2) were probed. Differences between the two groups were considered significant for P < 0.05, corrected for familywise error by threshold-free cluster enhancement.42 Because of the exploratory nature of study, we did not apply at this stage a restrictive multiple-comparison correction across the tested RSNs. Possible correlations between GM loss detected by VBM and FC alterations with clinical variables were
R E S T I N G
S T A T E
N E T W O R K S
I N
S C A 2
FIG. 1. Functional connectivity changes for the five RSNs that showed only cerebellar alterations at the dual regression analysis, including age as covariate. Clusters of significantly reduced functional connectivity in SCA2 patients, compared with HC, are superimposed for anatomical reference to the average T1-weighted volume provided by SPM8 in the standard Montreal Neurological Institute stereotactic space. No cluster of significantly increased connectivity was present. Patient’s right is at the observer’s right. Axial planes are sampled every 14 mm, starting at Z 5 –42 mm. SMN, sensorimotor network; ECN, executive control network; RFPN, right fronto-parietal network; LFPN, left fronto-parietal network; VIS, visual network.
assessed by nonparametric correlation analysis by using Spearman’s coefficient. To this end, for each cluster that was significantly different in SCA2 patients compared with HC at VBM or the dual regression analysis, the corresponding mean values of GM volume (for VBM) and of the Z-maps (for fMRI) were recorded. Tested clinical variables included DD, age at onset, number of triplets, and SARA score. In addition, for networks showing supratentorial alterations, an ancillary analysis was performed to assess the potential role of the neuronal loss in cerebellar nodes in determining alterations in the FC of the supratentorial nodes of a network. For this purpose, the mean values of the Z-maps of supratentorial clusters where probed for correlation with the mean GM volume of the cerebellar clusters, masked by the atrophy map derived from VBM, by Spearman’s correlation coefficient.
Results The SCA2 and HC groups were not significantly different for age and sex. The VBM analysis showed a
significant GM loss throughout both cerebellar hemispheres and vermis (Supplemental Data Fig. S1, Table S1), without significant clusters of atrophy in the supratentorial compartment. When assessing the HCSCA2 contrast, without correcting for GM values, all of the investigated RSNs showed cerebellar clusters of significantly reduced connectivity strength (Fig. 1). Only the DMN showed, in addition to the cerebellar FC loss, supratentorial clusters of significant connectivity reduction (Fig. 2). When controlling for atrophy, the HC>SCA2 contrast resulted in cerebellar clusters of significantly
Movement Disorders, Vol. 30, No. 10, 2015
1385
C O C O Z Z A
E T
A L
FIG. 2. Functional connectivity changes of the default mode network (DMN) at the dual regression analysis, including age as covariate. Clusters of significantly reduced functional connectivity in SCA2 patients, compared with HC, are superimposed for anatomical reference to the average T1weighted volume provided by SPM8 in the standard Montreal Neurological Institute stereotactic space. No cluster of significantly increased connectivity was present. Patient’s right is at the observer’s right.
reduced FC only in the DMN, ECN, and RFPN. In addition, the DMN maintained a similar pattern of supratentorial involvement (Fig. 3; Table 2). In all cases, the significant clusters of connectivity changes fell within the positive components of the RSNs (depicted in red in Supplemental Data Fig. S2). When assessing the HCSCA between group contrast, after controlling for atrophy Cluster Volume
DMN
19,344 mm
3
3,760 mm3 2,840 mm3
ECN RFPN
1,752 mm3 672 mm3 296 mm3 264 mm3 64 mm3 8 mm3 16 mm3 928 mm3 456 mm3
X
Y
Z
4 24 10 26 212 246 226 228 8 18 28 24 24 26 8 16 16 52 230 212 24 236
30 20 36 28 40 30 16 8 4 10 20 6 8 32 34 15 272 224 258 276 270 272
217 211 11 1 21 23 29 221 5 1 211 211 23 59 59 71 223 27 229 225 221 231
Right rectal gyrus Left rectal gyrus Right anterior cingulate cortex Left anterior cingulate cortex Left mid orbital gyrus Left inferior frontal gyrus (p. orbitalis) Left insula lobe Left temporal pole Right caudate nucleus Right putamen Right inferior frontal gyrus (p. orbitalis) Right olfactory cortex Left caudate nucleus Left superior medial gyrus Right superior medial gyrus Right SMA Right cerebellum—lobule VI (Hem) Right cerebellum—lobule I–IV (Hem) Left cerebellum—lobule VI (Hem) Left cerebellum—lobule VI (Hem) Cerebellar vermis Left cerebellum—lobule VIIa crus I (Hem)
DMN, default mode network; EN, executive control network; RFPN, right fronto-parietal network. Anatomical labeling is according to Tzourio-Mazoyer et al.55
ganglia, thalamus, superior frontal gyri, parietal cortex, cerebellum, and pons, and increased connectivity with the M1 cortex. Although we did not test specifically putaminal and SMA connectivity by seed-based analysis, consistent with the lack of extrapyramidal signs in our patients, connectivity of all of these structures was not significantly altered in the networks that we analyzed. The only exception are the superior frontal gyri, which show an overlap with one of the clusters of significantly reduced connectivity within the DMN in our data, extending into the right SMA. Further studies are needed to clarify whether these changes in the DMN have a significant relationship with alterations of the extrapyramidal circuits, previously shown in parkinsonian SCA2 patients. The functional connections, as derived from a metaanalysis of the FC literature, of the cerebellar regions that are atrophic in SCA1725 have shown a dichotomy of these connections. In particular, the anterior cerebellar clusters of atrophy proved to be mainly functionally connected to motor-related areas, whereas the posterior cerebellar atrophic regions show FC with more cognitive/emotional-related fronto-temporoparietal areas. These results mirror the spectrum of motor and neuropsychiatric deficits that is present in SCA17, hinting at a possible role for disruption of cerebellar-(thalamo)-cerebral connections in determining the clinical phenotype, which is characterized by a cognitive decline, finally resulting in dementia.47,48 Accordingly, an alteration of the FC networks
1388
Movement Disorders, Vol. 30, No. 10, 2015
including these regions could have been hypothesized also in SCA2, which shares a similar pattern of cerebellar atrophy with SCA17. Instead, consistent with the reduced severity of cognitive deficits in SCA2 compared with SCA17 (in which dementia/psychiatric symptoms, associated with cerebral atrophy, are a major feature48,49), we did not find significant supratentorial alterations of executive and both frontoparietal networks in our group of patients. Conversely, the DMN alterations that we found in SCA2 patients may very well be in agreement with the preferential involvement of lobule IX in SCA2 (with uvula being the most significant GM cerebellar cluster detected when assessing regionally the atrophy in SCA2,12 with higher progression of atrophy over time, compared with normal subjects13). A reduced FC within the DMN has been associated with a larger Stroop effect,50 which may be involved in the subtle deficits at Stroop test that have been found in this pathology.51 Possible correlations between the alterations of the RSNs and cognitive deficits need to be explored in future studies, possibly with a longitudinal design. In particular, patient groups including milder or preclinical phases of the disease should be examined, to avoid plateauing effects that are likely to occur in more advanced stages of the disease, considering that cognitive alterations correlate to some extent with the severity of the ataxia symptoms.52 Taken together, these data suggest that, although the GM loss in cerebellar structures may result in a
R E S T I N G
derangement of the corresponding RSNs, as is the case for DMN in our data, this does not occur necessarily, possibly because of compensatory phenomena. We did not find any correlation between RS-fMRI results and the tested clinical variables. The lack of correlation between RSNs clusters and SARA score is not surprising, given the nature of the SARA score, which is essentially based on an assessment of motor deficits, hardly related to the altered RSNs, which did not include the SMN, not significantly altered in these patients. These results do not conflict with the few available data on the correlations between FC and clinical data (ie, disease severity scores, CAG repeats, and DDs). In particular, in SCA7, the number of CAG repetitions and DDs correlated with the FC between right anterior cerebellum and supratentorial structures (left superior frontal gyrus and left parahippocampal gyrus), which, however, show some degree of atrophy in this pathology.24 Conversely, in SCA1, DDs and severity of the disease showed a linear correlation only to changes in functional organization of the thalamus,26 a structure not included in the investigated RSNs, as previously discussed. More in general, the scarce correlation between imaging and clinical data in SCA2 4,7-9,11,13 has been hypothesized to be attributable to the presence of extensive cerebellar GM loss before the appearance of clinical symptoms, with a relatively slower progression of the neurodegenerative phenomena in the symptomatic phase, compared with other SCAs.5,10,53,54 This plateauing effect also may explain the lack of correlation between the degree of cerebellar atrophy and supratentorial alterations in the DMN, as shown by the ancillary analysis in these patients. To avoid any bias caused by head motion during the scans, we corrected the data for movement by rigid body registration, excluded from the analysis the studies with movements above 50% of voxel size, and included as nuisance in the analysis the mean head displacement of each study. The latter procedure rests on an assumption of linearity of the effect of motion on FC strength, which may not hold entirely true, because a nonlinear effect of movements has been shown on fMRI temporal signal/noise ratio.34 Further studies are needed to fully characterize the effects of motion on the apparent strength of FC, which may help develop new strategies to completely remove the effects of motion from the analysis. In conclusion, our results suggest the presence of alterations of several RSNs in SCA2 patients, explained only in part by the cerebellar atrophy. When taking into account the degree of atrophy, the DMN is the only RSN that shows both cerebellar and supratentorial changes. Further studies, including specific neurocognitive scores, are needed to confirm
S T A T E
N E T W O R K S
I N
S C A 2
the results of this exploratory study, and to assess the clinical significance of these modifications.
References 1.
Auburger GW. Spinocerebellar ataxia type 2. Handb Clin Neurol 2012;103:423-436.
2.
Magana JJ, Velazquez-Perez L, Cisneros B. Spinocerebellar ataxia type 2: clinical presentation, molecular mechanisms, and therapeutic perspectives. Mol Neurobiol 2013;47:90-104.
3.
Pulst SM, Nechiporuk A, Nechiporuk T, et al. Moderate expansion of a normally biallelic trinucleotide repeat in spinocerebellar ataxia type 2. Nat Genet 1996;14:269-276.
4.
Brenneis C, Bosch SM, Schocke M, Wenning GK, Poewe W. Atrophy pattern in SCA2 determined by voxel-based morphometry. Neuroreport 2003;14:1799-1802.
5.
Guerrini L, Lolli F, Ginestroni A, et al. Brainstem neurodegeneration correlates with clinical dysfunction in SCA1 but not in SCA2: a quantitative volumetric, diffusion and proton spectroscopy MR study. Brain 2004;127:1785-1795.
6.
Della Nave R, Ginestroni A, Tessa C, et al. Brain structural damage in spinocerebellar ataxia type 2. A voxel-based morphometry study. Mov Disord 2008;23:899-903.
7.
Della Nave R, Ginestroni A, Tessa C, et al. Brain white matter tracts degeneration in Friedreich ataxia: an in vivo MRI study using tract-based spatial statistics and voxel-based morphometry. Neuroimage 2008;40:19-25.
8.
D’Agata F, Caroppo P, Boghi A, et al. Linking coordinative and executive dysfunctions to atrophy in spinocerebellar ataxia 2 patients. Brain Struct Funct 2011;216:275-288.
9.
Goel G, Pal PK, Ravishankar S, et al. Gray matter volume deficits in spinocerebellar ataxia: an optimized voxel based morphometric study. Parkinsonism Relat Disord 2011;17:521-527.
10.
Jacobi H, Reetz K, du Montcel ST, et al. Biological and clinical characteristics of individuals at risk for spinocerebellar ataxia types 1, 2, 3, and 6 in the longitudinal RISCA study: analysis of baseline data. Lancet Neurol 2013;12:650-658.
11.
Mercadillo RE, Galvez V, Diaz R, et al. Parahippocampal gray matter alterations in spinocerebellar ataxia type 2 identified by voxel based morphometry. J Neurol Sci 2014;347:50-58.
12.
Jung BC, Choi SI, Du AX, et al. MRI shows a region-specific pattern of atrophy in spinocerebellar ataxia type 2. Cerebellum 2012; 11:272-279.
13.
Mascalchi M, Diciotti S, Giannelli M, et al. Progression of brain atrophy in spinocerebellar ataxia type 2: a longitudinal tensorbased morphometry study. PLoS One 2014;9:e89410.
14.
Della Nave R, Foresti S, Tessa C, et al. ADC mapping of neurodegeneration in the brainstem and cerebellum of patients with progressive ataxias. Neuroimage 2004;22:698-705.
15.
Mandelli ML, De Simone T, Minati L, et al. Diffusion tensor imaging of spinocerebellar ataxias types 1 and 2. AJNR Am J Neuroradiol 2007;28:1996-2000.
16.
Salvatore E, Tedeschi E, Mollica C, et al. Supratentorial and infratentorial damage in spinocerebellar ataxia 2: a diffusion-weighted MRI study. Mov Disord 2014;29:780-786.
17.
Damoiseaux JS. Resting-state fMRI as a biomarker for Alzheimer’s disease? Alzheimers Res Ther 2012;4:8.
18.
Prodoehl J, Burciu RG, Vaillancourt DE. Resting state functional magnetic resonance imaging in Parkinson’s disease. Curr Neurol Neurosci Rep 2014;14:448.
19.
Dutta A, McKie S, Deakin JF. Resting state networks in major depressive disorder. Psychiatry Res 2014;224:139-151.
20.
Argyelan M, Ikuta T, DeRosse P, et al. Resting-state fMRI connectivity impairment in schizophrenia and bipolar disorder. Schizophr Bull 2014;40:100-110.
21.
Dumas EM, van den Bogaard SJ, Middelkoop HA, Roos RA. A review of cognition in Huntington’s disease. Front Biosci (Schol Ed) 2013;5:1-18.
22.
Quarantelli M, Salvatore E, Giorgio SM, et al. Default-mode network changes in Huntington’s disease: an integrated MRI study of functional connectivity and morphometry. PLoS One 2013;8: e72159.
Movement Disorders, Vol. 30, No. 10, 2015
1389
C O C O Z Z A
E T
A L
23.
Cataldi M, Avoli M, de Villers-Sidani E. Resting state networks in temporal lobe epilepsy. Epilepsia 2013;54:2048-2059.
41.
Ashburner J. A fast diffeomorphic image registration algorithm. Neuroimage 2007;38:95-113.
24.
Hernandez-Castillo CR, Alcauter S, Galvez V, et al. Disruption of visual and motor connectivity in spinocerebellar ataxia type 7. Mov Disord 2013;28:1708-1716.
42.
Smith SM, Nichols TE. Threshold-free cluster enhancement: addressing problems of smoothing, threshold dependence and localisation in cluster inference. Neuroimage 2009;44:83-98.
25.
Reetz K, Dogan I, Rolfs A, et al. Investigating function and connectivity of morphometric findings–exemplified on cerebellar atrophy in spinocerebellar ataxia 17 (SCA17). Neuroimage 2012;62: 1354-1366.
43.
Tang L, Ge Y, Sodickson DK, et al. Thalamic resting-state functional networks: disruption in patients with mild traumatic brain injury. Radiology 2011;260:831-840.
44.
26.
Solodkin A, Peri E, Chen EE, Ben-Jacob E, Gomez CM. Loss of intrinsic organization of cerebellar networks in spinocerebellar ataxia type 1: correlates with disease severity and duration. Cerebellum 2011;10:218-232.
Gwinn-Hardy K, Chen JY, Liu HC, et al. Spinocerebellar ataxia type 2 with parkinsonism in ethnic Chinese. Neurology 2000;55: 800-805.
45.
Lu CS, Wu Chou YH, Kuo PC, Chang HC, Weng YH. The parkinsonian phenotype of spinocerebellar ataxia type 2. Arch Neurol 2004;61:35-38.
46.
Shan DE, Soong BW, Sun CM, Lee SJ, Liao KK, Liu RS. Spinocerebellar ataxia type 2 presenting as familial levodopa-responsive parkinsonism. Ann Neurol 2001;50:812-815.
47.
Mariotti C, Alpini D, Fancellu R, et al. Spinocerebellar ataxia type 17 (SCA17): oculomotor phenotype and clinical characterization of 15 Italian patients. J Neurol 2007;254:1538-1546.
48.
Rolfs A, Koeppen AH, Bauer I, et al. Clinical features and neuropathology of autosomal dominant spinocerebellar ataxia (SCA17). Ann Neurol 2003;54:367-375.
49.
Manto MU. The wide spectrum of spinocerebellar ataxias (SCAs). Cerebellum 2005;4:2-6.
50.
Duchek JM, Balota DA, Thomas JB, et al. Relationship between Stroop performance and resting state functional connectivity in cognitively normal older adults. Neuropsychology 2013;27:516528.
51.
Rodriguez-Labrada R, Velazquez-Perez L, Aguilera-Rodriguez R, et al. Executive deficit in spinocerebellar ataxia type 2 is related to expanded CAG repeats: evidence from antisaccadic eye movements. Brain Cogn 2014;91:28-34.
52.
Ma J, Wu C, Lei J, Zhang X. Cognitive impairments in patients with spinocerebellar ataxia types 1, 2 and 3 are positively correlated to the clinical severity of ataxia symptoms. Int J Clin Exp Med 2014;7:5765-5771.
53.
Jacobi H, Bauer P, Giunti P, et al. The natural history of spinocerebellar ataxia type 1, 2, 3, and 6: a 2-year follow-up study. Neurology 2011;77:1035-1041.
54.
Tezenas du Montcel S, Charles P, Goizet C, et al. Factors influencing disease progression in autosomal dominant cerebellar ataxia and spastic paraplegia. Arch Neurol 2012;69:500-508.
55.
Tzourio-Mazoyer N, Landeau B, Papathanassiou D, et al. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 2002;15:273-289.
27.
28.
Wu T, Wang C, Wang J, Hallett M, Zang Y, Chan P. Preclinical and clinical neural network changes in SCA2 parkinsonism. Parkinsonism Relat Disord 2013;19:158-164. Hering S, Achmuller C, Kohler A, Poewe W, Schneider R, Boesch SM. Phenotype variability in spinocerebellar ataxia type 2: a longitudinal family survey and a case featuring an unusual benign course of disease. Mov Disord 2009;24:774-777.
29.
Sacca F, Puorro G, Brunetti A, et al. A randomized controlled pilot trial of lithium in spinocerebellar ataxia type 2. J Neurol 2015; 262:149-153.
30.
Chao-Gan Y, Yu-Feng Z. DPARSF: A MATLAB toolbox for “pipeline” data analysis of resting-state fMRI. Front Syst Neurosci 2010;4:13.
31.
Song XW, Dong ZY, Long XY, et al. REST: a toolkit for restingstate functional magnetic resonance imaging data processing. PLoS One 2011;6:e25031.
32.
Friston KJA, Ashburner J, Frith CD, Poline JB, Heather JD, Frackowiak RSJ. Spatial registration and normalization of images. Human Brain Mapping 1995;3:165-189.
33.
Ardekani BA, Bachman AH, Helpern JA. A quantitative comparison of motion detection algorithms in fMRI. Magn Reson Imaging 2001;19:959-963.
34.
Van Dijk KR, Sabuncu MR, Buckner RL. The influence of head motion on intrinsic functional connectivity MRI. Neuroimage 2012;59:431-438.
35.
Beckmann CF, Smith SM. Probabilistic independent component analysis for functional magnetic resonance imaging. IEEE Trans Med Imaging 2004;23:137-152.
36.
Minka T. Automatic choice of dimensionality for PCA. Technical Report 514, MIT Media Lab Vision and Modeling Group, 2000.
37.
Hyvarinen A. Fast and robust fixed-point algorithms for independent component analysis. IEEE Trans Neural Netw 1999;10:626634.
38.
Smith SM, Fox PT, Miller KL, et al. Correspondence of the brain’s functional architecture during activation and rest. Proc Natl Acad Sci U S A 2009;106:13040-13045.
39.
Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. A default mode of brain function. Proc Natl Acad Sci U S A 2001;98:676-682.
40.
Filippini N, MacIntosh BJ, Hough MG, et al. Distinct patterns of brain activity in young carriers of the APOE-epsilon4 allele. Proc Natl Acad Sci U S A 2009;106:7209-7214.
1390
Movement Disorders, Vol. 30, No. 10, 2015
Supporting Data Additional Supporting Information may be found in the online version of this article at the publisher’s web-site.
ARTICLE
Modifications of Resting State Networks in Spinocerebellar Ataxia Type 2 , MD,2 Amedeo Cervo, MD,1 Angela Marsili, Mbiol,2 Cinzia Valeria Russo, MD,2 Sirio Cocozza, MD,1* Francesco Sacca Sara Maria delle Acque Giorgio, MD,1 Giuseppe De Michele, MD,2 Alessandro Filla, MD,2 Arturo Brunetti, MD,1 and Mario Quarantelli, MD3 1
Department of Advanced Biomedical Sciences, University “Federico II,” Naples, Italy Department of Neurosciences and Reproductive and Odontostomatological Sciences, University “Federico II,” Naples, Italy 3 Biostructure and Bioimaging Institute, National Research Council, Naples, Italy
2
A B S T R A C T : Purpose: We aimed to investigate the integrity of the Resting State Networks in spinocerebellar ataxia type 2 (SCA2) and the correlations between the modification of these networks and clinical variables. Methods: Resting-state functional magnetic resonance imaging (RS-fMRI) data from 19 SCA2 patients and 29 healthy controls were analyzed using an independent component analysis and dual regression, controlling at voxel level for the effect of atrophy by co-varying for gray matter volume. Correlations between the resting state networks alterations and disease duration, age at onset, number of triplets, and clinical score were assessed by Spearman’s coefficient, for each cluster which was significantly different in SCA2 patients compared with healthy controls. Results: In SCA2 patients, disruption of the cerebellar components of all major resting state networks was present, with supratentorial involvement only for the default
Spinocerebellar ataxias (SCAs) are a set of autosomal dominant progressive neurodegenerative disorders. Their classification is essentially based on the underlying
-----------------------------------------------------------*Correspondence to: Sirio Cocozza, MD, Department of Advanced Biomedical Sciences, University “Federico II,” Via Pansini, 5—80131— Naples, Italy, E-mail: [email protected]
Funding agencies: The study was supported by a grant from ‘‘Associazione Italiana per la lotta alle Sindromi Atassiche (AISA) sez. Campania’’ to FS, and a grant from the Italian Ministry of Education, University and Research, project PRIN 2010-2011 20108WT59Y_007 to G.D.M. Relevant conflicts of interest/financial disclosures: Nothing to report. Author roles may be found in the online version of this article. Received: 2 February 2015; Revised: 13 April 2015; Accepted: 11 May 2015 Published online 12 June 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/mds.26284
1382
Movement Disorders, Vol. 30, No. 10, 2015
mode network. When controlling at voxel level for gray matter volume, the reduction in functional connectivity in supratentorial regions of the default mode network, and in cerebellar regions within the default mode, executive and right fronto-parietal networks, was still significant. No correlations with clinical variables were found for any of the investigated resting state networks. Conclusions: The SCA2 patients show significant alterations of the resting state networks, only partly explained by the atrophy. The default mode network is the only resting state network that shows also supratentorial changes, which appear unrelated to the cortical gray matter volume. Further studies are needed to assess the clinical sigC nificance of these changes. V 2015 International Parkinson and Movement Disorder Society
K e y W o r d s : functional MRI; spinocerebellar ataxia; default mode network; resting state networks; independent component analysis
genetic defect. Spinocerebellar ataxia type 2 (SCA2), the second most frequent ataxia worldwide after SCA3,1 is associated with a CAG repeat expansion on chromosome 12q23-24.1 in the gene encoding for ataxin-2, a protein involved in RNA splicing.2 SCA2 is characterized by progressive olivo-pontocerebellar atrophy. SCA2 patients clinically present ataxia, ophthalmoplegia, corticospinal involvement, cognitive dysfunction, and peripheral neuropathy.2 Onset of symptoms in SCA2 typically occurs in the fourth decade, with a disease duration between 10 and 15 y, and clinical presentation appears to be in close relationship to the CAG repetition length.3 Magnetic resonance imaging (MRI) studies have allowed in vivo assessment of the atrophy pattern in SCA2, using both voxel-based morphometry (VBM)
R E S T I N G
TABLE 1. Subjects’ demographic and clinical variables
Age (mean 6 SD) Sex (M/F) SARA (median) Triplets (median) DD (mean 6 SD) AAO (mean 6 SD)
HC
SCA2
40.3 6 10.8 (range, 18-83) 15/14 N/A N/A N/A N/A
37.3 6 16.1 (range, 19-62) 10/9 14 (range, 9-24) 42 (range, 35-51) 11.4 6 8.2 28.5 6 11.4 (range, 3-50)
HC, healthy controls; SCA2, spinocerebellar ataxia type 2; SD, standard deviation; SARA, Scale for the Assessment and Rating of Ataxia; N/A, not applicable; DD, disease duration; AAO, age at onset. Ages and DD are in years.
and tensor-based morphometry (TBM), which have shown significant gray matter (GM) loss, mainly affecting the cerebellar structures.4-12 Recently, TBM proved to be capable of demonstrating the progression of ponto-cerebellar atrophy in SCA2, supporting a possible role of MRI as biomarker in future trials.13 Conversely, studies of white matter microstructural alterations by mean of diffusion MRI have quite consistently shown both cerebellar and supratentorial involvement.5,7,14-16 Resting state functional MRI (RS-fMRI) is a powerful method for evaluating regional interactions that occur between brain structures when a subject is not performing an explicit task. Data-driven analysis of RS-fMRI data has consistently demonstrated the presence of preferential connectivity between specific cerebral structures, which are organized in stable and simultaneously operating networks. Abnormalities of many of these resting state networks (RSN), as detected by RS-fMRI, have been demonstrated in several neurodegenerative disorders, such as Alzheimer’s disease,17 Parkinson’s disease,18 depression,19 schizophrenia,20 Huntington’s disease,21,22 and epilepsy.23 To the best of our knowledge, only a few studies have assessed in SCA patients the functional connectivity (FC) integrity,24-27 although it is limited to specific structures through seed-based analysis. These studies suggested the presence of FC alterations, showing, besides the obvious cerebellar involvement, modifications in several supratentorial (both cortical and deep GM) structures. However, no studies have assessed systematically the alterations of the major RSNs in SCA patients using a model-free approach. Furthermore, FC alterations in SCA have not been assessed independently of the underlying atrophic changes. The aim of this exploratory study was to investigate the integrity of the major RSNs, as detected by independent component analysis (ICA; a fully data-driven method applied to RS-fMRI data), in a group of SCA2 patients, compared with a group of age- and
S T A T E
N E T W O R K S
I N
S C A 2
sex-matched healthy controls (HC), also assessing possible correlations of RSN alterations with clinical data. To discriminate between the changes caused by functional and structural alterations in the involved areas, FC strength within each of the detectable networks was assessed with and without controlling for the effect of atrophy by covarying for GM volume.
Methods Subjects Nineteen symptomatic SCA2 patients and 29 ageand sex-matched HCs were enrolled. All subjects were right-handed. The molecular test showed 32 or more CAG triplets in the ataxin-2 gene in all SCA2 patients. Disease duration (DD) and age at onset were retrospectively measured, based on the first occurrence of symptoms clearly related to the disease. In addition, within 1 week from the MRI acquisition, an expert neurologist examined the patients and recorded the scale for the assessment and rating of ataxia (SARA), consisting of eight items (range, 0-40, with higher score indicating more severe clinical outcome).28 The HCs had no history of neurological or psychiatric disorders, nor were they receiving treatment with medications active on the central nervous system. Demographic information of both SCA2 patients and HCs are listed in Table 1. This study was conducted on patients enrolled in a previous trial,29 approved from the local Ethics Committee (151/09), and registered at www.clinicaltrials. gov (NCT00998634) and EUDRACT (2009-01631720). All patients gave written informed consent.
MRI Data Acquisition All MRI studies have been carried out on a 3 Tesla MRI scanner (Trio, Siemens Medical Systems, Erlangen, Germany). Sequences included a morphological T1w volumetric acquisition and T2*-weighted images for RS-fMRI analysis, acquired within the same scanning session. Structural T1w volumes were acquired using a three-dimensional magnetization-prepared rapid gradient-echo sequence; axial planes; TR 5 1900 ms; TE 5 3.4 ms; TI 5 900 ms; Flip Angle 5 98; voxel size 5 1 3 1 3 1 mm3; number of slices 5 160). The T2*-weighted volumes were acquired using an echo-planar imaging sequence (axial planes; TR 5 2500 ms; TE 5 40 ms; 64 3 64 acquisition matrix; 30 slices; voxel size 5 3 3 3 3 4 mm3; gap 1 mm; 128 time points; acquisition time 50 2000 ). During the MRI study, the subjects were laying supine with the head lightly fixed by straps and foam pads to minimize head movement, and were asked to relax with eyes closed, without falling asleep.
Movement Disorders, Vol. 30, No. 10, 2015
1383
C O C O Z Z A
E T
A L
MRI Data Analysis Pre-processing of the functional data was performed using Data Processing Assistant for Resting-State fMR I (http://www.restfmri.net),30 which is based on Statistical Parametric Mapping (SPM8) (http://www.fil.ion. ucl.ac.uk/spm) and Resting-State fMRI Data Analysis Toolkit (http://www.restfmri.net).31 Briefly, it included removal of the first five time points (to allow for instability of the initial MRI signal, leaving 123 time points), motion correction, slicetiming correction, band-pass filtering (0.01 Hz < f < 0.1 Hz), co-registration to structural scan, spatial normalization to the Montreal Neurological Institute template, and spatial smoothing (6-mm Gaussian kernel). For all pre-processing steps, default SPM8 parameters were used. The SPM motion correction procedure realigns the subsequent volumes of each study to the first one, iteratively finding the translation and rotation parameters that minimize a least-squares cost function derived from the voxel-by-voxel intensity difference from the reference image, a least squares approach, and a sixparameter (rigid body) spatial transformation.32 This approach proved to be accurate in realigning fMRI volumes for motion correction purposes.33 From the motion correction procedure, the mean displacement for each brain volume was computed as the root-mean-square (RMS) of the translation parameters at each time point. Studies with a mean relative RMS of 0.15 or higher (according to Van Dijk et al.34) or with 1.5 mm or more or 1.5 degrees rotation along any axis were discarded. Five SCA2 patients and four HCs were excluded by the analysis because of excessive movements during the fMRI scan. In the remaining 39 studies, a residual significant difference in the RMS was present between the two groups (0.035 6 0.024 in HC and 0.061 6 0.029 in SCA2, P < 0.01). Accordingly, in the following analysis RMS was added as nuisance covariate. Preprocessed volumes were then analyzed by using a temporal-concatenation spatial ICA approach,35 implemented in MELODIC (multivariate exploratory linear decomposition into independent components) Version 3.12, part of FSL (FMRIB’s Software Library, www. fmrib.ox.ac.uk/fsl). Pre-processed data were whitened and projected into a 3-dimensional subspace by using probabilistic principal component analysis in which the number of dimensions was estimated by using the Laplace approximation to the Bayesian evidence of the model order.35,36 The whitened observations were decomposed into sets of vectors that describe signal variation across the temporal domain (time-courses), the session/subject domain, and across the spatial domain (maps) by optimizing for nonGaussian spatial source distributions using a fixed-point iteration technique.37 Estimated component maps were
1384
Movement Disorders, Vol. 30, No. 10, 2015
divided by the standard deviation of the residual noise and thresholded by fitting a mixture model to the histogram of intensity values.35 MELODIC provided 20 independent components, among which those best matching the 10 major RSN templates published in Smith et al.38 were used for subsequent analysis. The correspondence between the components and the RSNs was defined based on cross-correlation of voxel values using the fslcc routine. Only the networks providing r > 0.5, and further confirmed by visual inspection to spatially overlap published templates, based on the descriptions available from reference studies,38,39 were further considered for analysis. The selected components were subsequently fed into a dual-regression analysis.40 Accordingly, for each subject, each RSN-specific mean time course is extracted and then used as voxelwise regressors in a linear model against the individual fMRI sets, to obtain subjectspecific spatial z-score maps for each RSN. The RSN maps were then compared between SCA patients and HCs by using a general linear model with permutation testing (5,000 permutations), including for each subject as nuisance covariates age, sex, and the average RMS to remove potential residual movement effects. This second-level analysis was carried out both with and without correction for atrophy by covarying at a voxel level, using the GM probability maps of each patient as voxel-dependent covariates. The GM probability maps were obtained by using the fast diffeomorphic registration algorithm (Diffeomorphic Anatomical Registration using Exponentiated Lie algebra; DARTEL),41 implemented in the Statistical Parametric Mapping software package (SPM8 - http://www.fil.ion. ucl.ac.uk/spm Wellcome Trust Centre for Neuroimaging, University College London). For all of the DARTEL preprocessing steps, the default SPM8 parameters were used. The normalized GM maps (2 3 2 3 2 mm3 voxel size) provided by DARTEL were visually assessed, to ensure good quality of the segmentation and normalization, and then smoothed using a 6–mm FWHM isotropic Gaussian kernel to match the resolution of the RS-fMRI images used for RSN definition. To characterize the distribution of GM atrophy in this cohort of patients, a VBM analysis was carried out on normalized GM maps (see Supplemental Data). When comparing the two groups, both contrasts (HC > SCA2 and HC < SCA2) were probed. Differences between the two groups were considered significant for P < 0.05, corrected for familywise error by threshold-free cluster enhancement.42 Because of the exploratory nature of study, we did not apply at this stage a restrictive multiple-comparison correction across the tested RSNs. Possible correlations between GM loss detected by VBM and FC alterations with clinical variables were
R E S T I N G
S T A T E
N E T W O R K S
I N
S C A 2
FIG. 1. Functional connectivity changes for the five RSNs that showed only cerebellar alterations at the dual regression analysis, including age as covariate. Clusters of significantly reduced functional connectivity in SCA2 patients, compared with HC, are superimposed for anatomical reference to the average T1-weighted volume provided by SPM8 in the standard Montreal Neurological Institute stereotactic space. No cluster of significantly increased connectivity was present. Patient’s right is at the observer’s right. Axial planes are sampled every 14 mm, starting at Z 5 –42 mm. SMN, sensorimotor network; ECN, executive control network; RFPN, right fronto-parietal network; LFPN, left fronto-parietal network; VIS, visual network.
assessed by nonparametric correlation analysis by using Spearman’s coefficient. To this end, for each cluster that was significantly different in SCA2 patients compared with HC at VBM or the dual regression analysis, the corresponding mean values of GM volume (for VBM) and of the Z-maps (for fMRI) were recorded. Tested clinical variables included DD, age at onset, number of triplets, and SARA score. In addition, for networks showing supratentorial alterations, an ancillary analysis was performed to assess the potential role of the neuronal loss in cerebellar nodes in determining alterations in the FC of the supratentorial nodes of a network. For this purpose, the mean values of the Z-maps of supratentorial clusters where probed for correlation with the mean GM volume of the cerebellar clusters, masked by the atrophy map derived from VBM, by Spearman’s correlation coefficient.
Results The SCA2 and HC groups were not significantly different for age and sex. The VBM analysis showed a
significant GM loss throughout both cerebellar hemispheres and vermis (Supplemental Data Fig. S1, Table S1), without significant clusters of atrophy in the supratentorial compartment. When assessing the HCSCA2 contrast, without correcting for GM values, all of the investigated RSNs showed cerebellar clusters of significantly reduced connectivity strength (Fig. 1). Only the DMN showed, in addition to the cerebellar FC loss, supratentorial clusters of significant connectivity reduction (Fig. 2). When controlling for atrophy, the HC>SCA2 contrast resulted in cerebellar clusters of significantly
Movement Disorders, Vol. 30, No. 10, 2015
1385
C O C O Z Z A
E T
A L
FIG. 2. Functional connectivity changes of the default mode network (DMN) at the dual regression analysis, including age as covariate. Clusters of significantly reduced functional connectivity in SCA2 patients, compared with HC, are superimposed for anatomical reference to the average T1weighted volume provided by SPM8 in the standard Montreal Neurological Institute stereotactic space. No cluster of significantly increased connectivity was present. Patient’s right is at the observer’s right.
reduced FC only in the DMN, ECN, and RFPN. In addition, the DMN maintained a similar pattern of supratentorial involvement (Fig. 3; Table 2). In all cases, the significant clusters of connectivity changes fell within the positive components of the RSNs (depicted in red in Supplemental Data Fig. S2). When assessing the HCSCA between group contrast, after controlling for atrophy Cluster Volume
DMN
19,344 mm
3
3,760 mm3 2,840 mm3
ECN RFPN
1,752 mm3 672 mm3 296 mm3 264 mm3 64 mm3 8 mm3 16 mm3 928 mm3 456 mm3
X
Y
Z
4 24 10 26 212 246 226 228 8 18 28 24 24 26 8 16 16 52 230 212 24 236
30 20 36 28 40 30 16 8 4 10 20 6 8 32 34 15 272 224 258 276 270 272
217 211 11 1 21 23 29 221 5 1 211 211 23 59 59 71 223 27 229 225 221 231
Right rectal gyrus Left rectal gyrus Right anterior cingulate cortex Left anterior cingulate cortex Left mid orbital gyrus Left inferior frontal gyrus (p. orbitalis) Left insula lobe Left temporal pole Right caudate nucleus Right putamen Right inferior frontal gyrus (p. orbitalis) Right olfactory cortex Left caudate nucleus Left superior medial gyrus Right superior medial gyrus Right SMA Right cerebellum—lobule VI (Hem) Right cerebellum—lobule I–IV (Hem) Left cerebellum—lobule VI (Hem) Left cerebellum—lobule VI (Hem) Cerebellar vermis Left cerebellum—lobule VIIa crus I (Hem)
DMN, default mode network; EN, executive control network; RFPN, right fronto-parietal network. Anatomical labeling is according to Tzourio-Mazoyer et al.55
ganglia, thalamus, superior frontal gyri, parietal cortex, cerebellum, and pons, and increased connectivity with the M1 cortex. Although we did not test specifically putaminal and SMA connectivity by seed-based analysis, consistent with the lack of extrapyramidal signs in our patients, connectivity of all of these structures was not significantly altered in the networks that we analyzed. The only exception are the superior frontal gyri, which show an overlap with one of the clusters of significantly reduced connectivity within the DMN in our data, extending into the right SMA. Further studies are needed to clarify whether these changes in the DMN have a significant relationship with alterations of the extrapyramidal circuits, previously shown in parkinsonian SCA2 patients. The functional connections, as derived from a metaanalysis of the FC literature, of the cerebellar regions that are atrophic in SCA1725 have shown a dichotomy of these connections. In particular, the anterior cerebellar clusters of atrophy proved to be mainly functionally connected to motor-related areas, whereas the posterior cerebellar atrophic regions show FC with more cognitive/emotional-related fronto-temporoparietal areas. These results mirror the spectrum of motor and neuropsychiatric deficits that is present in SCA17, hinting at a possible role for disruption of cerebellar-(thalamo)-cerebral connections in determining the clinical phenotype, which is characterized by a cognitive decline, finally resulting in dementia.47,48 Accordingly, an alteration of the FC networks
1388
Movement Disorders, Vol. 30, No. 10, 2015
including these regions could have been hypothesized also in SCA2, which shares a similar pattern of cerebellar atrophy with SCA17. Instead, consistent with the reduced severity of cognitive deficits in SCA2 compared with SCA17 (in which dementia/psychiatric symptoms, associated with cerebral atrophy, are a major feature48,49), we did not find significant supratentorial alterations of executive and both frontoparietal networks in our group of patients. Conversely, the DMN alterations that we found in SCA2 patients may very well be in agreement with the preferential involvement of lobule IX in SCA2 (with uvula being the most significant GM cerebellar cluster detected when assessing regionally the atrophy in SCA2,12 with higher progression of atrophy over time, compared with normal subjects13). A reduced FC within the DMN has been associated with a larger Stroop effect,50 which may be involved in the subtle deficits at Stroop test that have been found in this pathology.51 Possible correlations between the alterations of the RSNs and cognitive deficits need to be explored in future studies, possibly with a longitudinal design. In particular, patient groups including milder or preclinical phases of the disease should be examined, to avoid plateauing effects that are likely to occur in more advanced stages of the disease, considering that cognitive alterations correlate to some extent with the severity of the ataxia symptoms.52 Taken together, these data suggest that, although the GM loss in cerebellar structures may result in a
R E S T I N G
derangement of the corresponding RSNs, as is the case for DMN in our data, this does not occur necessarily, possibly because of compensatory phenomena. We did not find any correlation between RS-fMRI results and the tested clinical variables. The lack of correlation between RSNs clusters and SARA score is not surprising, given the nature of the SARA score, which is essentially based on an assessment of motor deficits, hardly related to the altered RSNs, which did not include the SMN, not significantly altered in these patients. These results do not conflict with the few available data on the correlations between FC and clinical data (ie, disease severity scores, CAG repeats, and DDs). In particular, in SCA7, the number of CAG repetitions and DDs correlated with the FC between right anterior cerebellum and supratentorial structures (left superior frontal gyrus and left parahippocampal gyrus), which, however, show some degree of atrophy in this pathology.24 Conversely, in SCA1, DDs and severity of the disease showed a linear correlation only to changes in functional organization of the thalamus,26 a structure not included in the investigated RSNs, as previously discussed. More in general, the scarce correlation between imaging and clinical data in SCA2 4,7-9,11,13 has been hypothesized to be attributable to the presence of extensive cerebellar GM loss before the appearance of clinical symptoms, with a relatively slower progression of the neurodegenerative phenomena in the symptomatic phase, compared with other SCAs.5,10,53,54 This plateauing effect also may explain the lack of correlation between the degree of cerebellar atrophy and supratentorial alterations in the DMN, as shown by the ancillary analysis in these patients. To avoid any bias caused by head motion during the scans, we corrected the data for movement by rigid body registration, excluded from the analysis the studies with movements above 50% of voxel size, and included as nuisance in the analysis the mean head displacement of each study. The latter procedure rests on an assumption of linearity of the effect of motion on FC strength, which may not hold entirely true, because a nonlinear effect of movements has been shown on fMRI temporal signal/noise ratio.34 Further studies are needed to fully characterize the effects of motion on the apparent strength of FC, which may help develop new strategies to completely remove the effects of motion from the analysis. In conclusion, our results suggest the presence of alterations of several RSNs in SCA2 patients, explained only in part by the cerebellar atrophy. When taking into account the degree of atrophy, the DMN is the only RSN that shows both cerebellar and supratentorial changes. Further studies, including specific neurocognitive scores, are needed to confirm
S T A T E
N E T W O R K S
I N
S C A 2
the results of this exploratory study, and to assess the clinical significance of these modifications.
References 1.
Auburger GW. Spinocerebellar ataxia type 2. Handb Clin Neurol 2012;103:423-436.
2.
Magana JJ, Velazquez-Perez L, Cisneros B. Spinocerebellar ataxia type 2: clinical presentation, molecular mechanisms, and therapeutic perspectives. Mol Neurobiol 2013;47:90-104.
3.
Pulst SM, Nechiporuk A, Nechiporuk T, et al. Moderate expansion of a normally biallelic trinucleotide repeat in spinocerebellar ataxia type 2. Nat Genet 1996;14:269-276.
4.
Brenneis C, Bosch SM, Schocke M, Wenning GK, Poewe W. Atrophy pattern in SCA2 determined by voxel-based morphometry. Neuroreport 2003;14:1799-1802.
5.
Guerrini L, Lolli F, Ginestroni A, et al. Brainstem neurodegeneration correlates with clinical dysfunction in SCA1 but not in SCA2: a quantitative volumetric, diffusion and proton spectroscopy MR study. Brain 2004;127:1785-1795.
6.
Della Nave R, Ginestroni A, Tessa C, et al. Brain structural damage in spinocerebellar ataxia type 2. A voxel-based morphometry study. Mov Disord 2008;23:899-903.
7.
Della Nave R, Ginestroni A, Tessa C, et al. Brain white matter tracts degeneration in Friedreich ataxia: an in vivo MRI study using tract-based spatial statistics and voxel-based morphometry. Neuroimage 2008;40:19-25.
8.
D’Agata F, Caroppo P, Boghi A, et al. Linking coordinative and executive dysfunctions to atrophy in spinocerebellar ataxia 2 patients. Brain Struct Funct 2011;216:275-288.
9.
Goel G, Pal PK, Ravishankar S, et al. Gray matter volume deficits in spinocerebellar ataxia: an optimized voxel based morphometric study. Parkinsonism Relat Disord 2011;17:521-527.
10.
Jacobi H, Reetz K, du Montcel ST, et al. Biological and clinical characteristics of individuals at risk for spinocerebellar ataxia types 1, 2, 3, and 6 in the longitudinal RISCA study: analysis of baseline data. Lancet Neurol 2013;12:650-658.
11.
Mercadillo RE, Galvez V, Diaz R, et al. Parahippocampal gray matter alterations in spinocerebellar ataxia type 2 identified by voxel based morphometry. J Neurol Sci 2014;347:50-58.
12.
Jung BC, Choi SI, Du AX, et al. MRI shows a region-specific pattern of atrophy in spinocerebellar ataxia type 2. Cerebellum 2012; 11:272-279.
13.
Mascalchi M, Diciotti S, Giannelli M, et al. Progression of brain atrophy in spinocerebellar ataxia type 2: a longitudinal tensorbased morphometry study. PLoS One 2014;9:e89410.
14.
Della Nave R, Foresti S, Tessa C, et al. ADC mapping of neurodegeneration in the brainstem and cerebellum of patients with progressive ataxias. Neuroimage 2004;22:698-705.
15.
Mandelli ML, De Simone T, Minati L, et al. Diffusion tensor imaging of spinocerebellar ataxias types 1 and 2. AJNR Am J Neuroradiol 2007;28:1996-2000.
16.
Salvatore E, Tedeschi E, Mollica C, et al. Supratentorial and infratentorial damage in spinocerebellar ataxia 2: a diffusion-weighted MRI study. Mov Disord 2014;29:780-786.
17.
Damoiseaux JS. Resting-state fMRI as a biomarker for Alzheimer’s disease? Alzheimers Res Ther 2012;4:8.
18.
Prodoehl J, Burciu RG, Vaillancourt DE. Resting state functional magnetic resonance imaging in Parkinson’s disease. Curr Neurol Neurosci Rep 2014;14:448.
19.
Dutta A, McKie S, Deakin JF. Resting state networks in major depressive disorder. Psychiatry Res 2014;224:139-151.
20.
Argyelan M, Ikuta T, DeRosse P, et al. Resting-state fMRI connectivity impairment in schizophrenia and bipolar disorder. Schizophr Bull 2014;40:100-110.
21.
Dumas EM, van den Bogaard SJ, Middelkoop HA, Roos RA. A review of cognition in Huntington’s disease. Front Biosci (Schol Ed) 2013;5:1-18.
22.
Quarantelli M, Salvatore E, Giorgio SM, et al. Default-mode network changes in Huntington’s disease: an integrated MRI study of functional connectivity and morphometry. PLoS One 2013;8: e72159.
Movement Disorders, Vol. 30, No. 10, 2015
1389
C O C O Z Z A
E T
A L
23.
Cataldi M, Avoli M, de Villers-Sidani E. Resting state networks in temporal lobe epilepsy. Epilepsia 2013;54:2048-2059.
41.
Ashburner J. A fast diffeomorphic image registration algorithm. Neuroimage 2007;38:95-113.
24.
Hernandez-Castillo CR, Alcauter S, Galvez V, et al. Disruption of visual and motor connectivity in spinocerebellar ataxia type 7. Mov Disord 2013;28:1708-1716.
42.
Smith SM, Nichols TE. Threshold-free cluster enhancement: addressing problems of smoothing, threshold dependence and localisation in cluster inference. Neuroimage 2009;44:83-98.
25.
Reetz K, Dogan I, Rolfs A, et al. Investigating function and connectivity of morphometric findings–exemplified on cerebellar atrophy in spinocerebellar ataxia 17 (SCA17). Neuroimage 2012;62: 1354-1366.
43.
Tang L, Ge Y, Sodickson DK, et al. Thalamic resting-state functional networks: disruption in patients with mild traumatic brain injury. Radiology 2011;260:831-840.
44.
26.
Solodkin A, Peri E, Chen EE, Ben-Jacob E, Gomez CM. Loss of intrinsic organization of cerebellar networks in spinocerebellar ataxia type 1: correlates with disease severity and duration. Cerebellum 2011;10:218-232.
Gwinn-Hardy K, Chen JY, Liu HC, et al. Spinocerebellar ataxia type 2 with parkinsonism in ethnic Chinese. Neurology 2000;55: 800-805.
45.
Lu CS, Wu Chou YH, Kuo PC, Chang HC, Weng YH. The parkinsonian phenotype of spinocerebellar ataxia type 2. Arch Neurol 2004;61:35-38.
46.
Shan DE, Soong BW, Sun CM, Lee SJ, Liao KK, Liu RS. Spinocerebellar ataxia type 2 presenting as familial levodopa-responsive parkinsonism. Ann Neurol 2001;50:812-815.
47.
Mariotti C, Alpini D, Fancellu R, et al. Spinocerebellar ataxia type 17 (SCA17): oculomotor phenotype and clinical characterization of 15 Italian patients. J Neurol 2007;254:1538-1546.
48.
Rolfs A, Koeppen AH, Bauer I, et al. Clinical features and neuropathology of autosomal dominant spinocerebellar ataxia (SCA17). Ann Neurol 2003;54:367-375.
49.
Manto MU. The wide spectrum of spinocerebellar ataxias (SCAs). Cerebellum 2005;4:2-6.
50.
Duchek JM, Balota DA, Thomas JB, et al. Relationship between Stroop performance and resting state functional connectivity in cognitively normal older adults. Neuropsychology 2013;27:516528.
51.
Rodriguez-Labrada R, Velazquez-Perez L, Aguilera-Rodriguez R, et al. Executive deficit in spinocerebellar ataxia type 2 is related to expanded CAG repeats: evidence from antisaccadic eye movements. Brain Cogn 2014;91:28-34.
52.
Ma J, Wu C, Lei J, Zhang X. Cognitive impairments in patients with spinocerebellar ataxia types 1, 2 and 3 are positively correlated to the clinical severity of ataxia symptoms. Int J Clin Exp Med 2014;7:5765-5771.
53.
Jacobi H, Bauer P, Giunti P, et al. The natural history of spinocerebellar ataxia type 1, 2, 3, and 6: a 2-year follow-up study. Neurology 2011;77:1035-1041.
54.
Tezenas du Montcel S, Charles P, Goizet C, et al. Factors influencing disease progression in autosomal dominant cerebellar ataxia and spastic paraplegia. Arch Neurol 2012;69:500-508.
55.
Tzourio-Mazoyer N, Landeau B, Papathanassiou D, et al. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 2002;15:273-289.
27.
28.
Wu T, Wang C, Wang J, Hallett M, Zang Y, Chan P. Preclinical and clinical neural network changes in SCA2 parkinsonism. Parkinsonism Relat Disord 2013;19:158-164. Hering S, Achmuller C, Kohler A, Poewe W, Schneider R, Boesch SM. Phenotype variability in spinocerebellar ataxia type 2: a longitudinal family survey and a case featuring an unusual benign course of disease. Mov Disord 2009;24:774-777.
29.
Sacca F, Puorro G, Brunetti A, et al. A randomized controlled pilot trial of lithium in spinocerebellar ataxia type 2. J Neurol 2015; 262:149-153.
30.
Chao-Gan Y, Yu-Feng Z. DPARSF: A MATLAB toolbox for “pipeline” data analysis of resting-state fMRI. Front Syst Neurosci 2010;4:13.
31.
Song XW, Dong ZY, Long XY, et al. REST: a toolkit for restingstate functional magnetic resonance imaging data processing. PLoS One 2011;6:e25031.
32.
Friston KJA, Ashburner J, Frith CD, Poline JB, Heather JD, Frackowiak RSJ. Spatial registration and normalization of images. Human Brain Mapping 1995;3:165-189.
33.
Ardekani BA, Bachman AH, Helpern JA. A quantitative comparison of motion detection algorithms in fMRI. Magn Reson Imaging 2001;19:959-963.
34.
Van Dijk KR, Sabuncu MR, Buckner RL. The influence of head motion on intrinsic functional connectivity MRI. Neuroimage 2012;59:431-438.
35.
Beckmann CF, Smith SM. Probabilistic independent component analysis for functional magnetic resonance imaging. IEEE Trans Med Imaging 2004;23:137-152.
36.
Minka T. Automatic choice of dimensionality for PCA. Technical Report 514, MIT Media Lab Vision and Modeling Group, 2000.
37.
Hyvarinen A. Fast and robust fixed-point algorithms for independent component analysis. IEEE Trans Neural Netw 1999;10:626634.
38.
Smith SM, Fox PT, Miller KL, et al. Correspondence of the brain’s functional architecture during activation and rest. Proc Natl Acad Sci U S A 2009;106:13040-13045.
39.
Raichle ME, MacLeod AM, Snyder AZ, Powers WJ, Gusnard DA, Shulman GL. A default mode of brain function. Proc Natl Acad Sci U S A 2001;98:676-682.
40.
Filippini N, MacIntosh BJ, Hough MG, et al. Distinct patterns of brain activity in young carriers of the APOE-epsilon4 allele. Proc Natl Acad Sci U S A 2009;106:7209-7214.
1390
Movement Disorders, Vol. 30, No. 10, 2015
Supporting Data Additional Supporting Information may be found in the online version of this article at the publisher’s web-site.
