Neurobiology of Disease 73 (2015) 327–333
Contents lists available at ScienceDirect
Neurobiology of Disease journal homepage: www.elsevier.com/locate/ynbdi
Dynamic cortical gray matter volume changes after botulinum toxin in cervical dystonia Cathérine C.S. Delnooz, Jaco W. Pasman, Bart P.C. van de Warrenburg ⁎ Radboud University Nijmegen Medical Centre, Donders Institute for Brain, Cognition and Behaviour, Department of Neurology, The Netherlands
a r t i c l e
i n f o
Article history: Received 15 May 2014 Revised 15 September 2014 Accepted 20 October 2014 Available online 28 October 2014 Keywords: Cervical dystonia Motor planning Botulinum toxin Structural plasticity Spatial computation
a b s t r a c t Previous electrophysiological and functional imaging studies in focal dystonia have reported on cerebral reorganization after botulinum toxin (BoNT) injections. With the exception of microstructural changes, alterations in gray matter volume after BoNT have not been explored. In this study, we sought to determine whether BoNT influences gray matter volume in a group of cervical dystonia (CD) patients. We analyzed whole brain gray matter volume in a sample of CD patients with VBM analysis. In patients, scans were repeated immediately before and some weeks after BoNT injections; controls were only scanned once. We analyzed 1) BoNT-related gray matter volume changes within patients; 2) gray matter volume differences between patients and controls; and 3) correlations between gray matter volume and disease duration and disease severity. The pre- and post-BoNT treatment analysis revealed an increase of gray matter volume within the right precentral sulcus, at the lateral border of the premotor cortex. In comparison to healthy controls, CD patients had reduced gray matter volume in area 45 functionally corresponding to the left ventral premotor cortex. No gray matter volume increase was found for CD patients in comparison to controls. Gray matter volume of the left supramarginal gyrus and left premotor cortex correlated positively with disease duration, and that of the right inferior parietal lobule correlated negatively with disease severity. We have identified structural, yet dynamic gray matter volume changes in CD. There were specific gray matter volume changes related to BoNT injections, illustrating indirect central consequences of modified peripheral sensory input. As differences were exclusively seen in higher order motor areas relevant to motor planning and spatial cognition, these observations support the hypothesis that deficits in these cognitive processes are crucial in the pathophysiology of CD. © 2014 Elsevier Inc. All rights reserved.
Introduction Primary cervical dystonia (CD) is characterized by involuntary abnormal movements and postures of the head and neck due to sustained contractions of the cervical musculature. Injecting the involved cervical muscles with botulinum toxin (BoNT) is an effective and evidencebased treatment of CD. In contrast to an identifiable cause in secondary dystonia, e.g. infarction of the basal ganglia, primary dystonia – including primary CD – has no evidence of an underlying cause, except for a possible genetic background as for example ANO3, GNAL and CIZ1. Although CD patients have several abnormalities in motor system physiology (Cakmur et al., 2004), its precise pathophysiology still remains to be clarified. In recent years, several functional cerebral abnormalities have ⁎ Corresponding author at: Department of Neurology (943), Radboud University Nijmegen Medical Centre PO box 9101, 6500 HB Nijmegen, The Netherlands. E-mail address: [email protected] (B.P.C. van de Warrenburg). Available online on ScienceDirect (www.sciencedirect.com).
http://dx.doi.org/10.1016/j.nbd.2014.10.013 0969-9961/© 2014 Elsevier Inc. All rights reserved.
been found in CD by means of functional magnetic resonance imaging (fMRI) (Naumann et al., 2000; Delnooz et al., 2013a; de Vries et al., 2012), which may originate from underlying structural abnormalities. Structural differences have been shown using diffusion tensor imaging, demonstrating altered structural integrity in the striatum, corpus callosum and the cortical motor circuitry (Blood et al., 2006; Fabbrini et al., 2008). Studies using voxel-based morphometry (VBM) reported alterations in gray matter (GM) volume in the basal ganglia (Draganski et al., 2003; Pantano et al., 2011; Prell et al., 2013), the cortical motor circuitry (Draganski et al., 2003; Pantano et al., 2011; Prell et al., 2013), and the cerebellum in CD (Draganski et al., 2003; Prell et al., 2013). Despite several reports on structural differences between dystonia patients and healthy controls at baseline, few studies exist on longitudinal structural changes. Interestingly, several electrophysiological and functional imaging studies in focal dystonia have reported on cerebral reorganization after BoNT treatment (Delnooz et al., 2013a; Kojovic et al., 2011; Thickbroom et al., 2003). Even more, Blood et al. reported on structural white matter alterations after BoNT injections
328
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333
(Blood et al., 2006). Following these previous reports, we sought to investigate the possible influence of BoNT on structural gray matter volume in CD patients. Since baseline differences in GM volume have not been consistent across studies, probably due to small sample sizes, clinical heterogeneity of the patients studied, and different methodological approaches (e.g. region of interest analysis versus whole brain analysis, and with or without correction for non-stationarity), it remains unclear which GM abnormalities in CD are sufficiently consistent to be regarded as the structural fingerprint of CD. VBM of T1-weighted structural MRI images provides information about the regional differences in volumetric organization of the brain by measurement of the quantity of tissue within a voxel (Ashburner and Friston, 2000). Regional GM alterations have been demonstrated in several movement disorders. VBM has also been of value in the identification of longitudinal GM alterations in neurological diseases, reflecting for example disease progression or treatment response (Pantano et al., 2011). In this study, we tested the hypothesis that BoNT treatment modifies the ‘dystonic brain’, by evaluating wholebrain GM volume (GMV) by means of VBM methodology in a longitudinal pre- and post-treatment design within CD patients. Also, we investigated GMV in a sample of CD patients and a group of matched, healthy controls, and the relationships between GMV and disease severity, disease duration and duration of BoNT treatment. Methods Subjects Twenty-three primary CD patients (14 women; mean age 57.2 years, 21 right-handed) and 22 healthy controls (12 women; mean age 54.5 years, 22 right-handed) participated after giving informed consent. The genetic status was not evaluated. The study was approved by the Ethics Committee of the Radboud University Nijmegen Medical Centre. The exclusion criteria included age under
18 years, severe head tremor or dystonia outside the cervical region, and absence of regular BoNT treatment. In two thirds of the patients, torticollis was the dominant feature, of which 9 patients presented with torticollis to the left. In the other patients, laterocollis was the predominant symptom (Table 1). Patients were all being treated with BoNT type A (Dysport®) every 2 to 4 months (mean duration of BoNT treatment: 7.6 years). No other neurotropic medication was used. Dystonia severity was measured with the Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS) severity subscore. The results of TWSTRS scores ranged from 6 to 22 (median 19) before injection (t = 0) and from 1 to 17 (median 11) after injection (t = 1), demonstrating a significant improvement due to BoNT (related-samples Wilcoxon signed rank test TWSTRS t = 0 vs. t = 1: Z = −4.11, p = .00). Image acquisition To investigate treatment-related changes, patients were scanned three times: before BoNT injection (t = 0), after 4–5 weeks (t = 1), and just before the next BoNT treatment (t = 2). Healthy controls were only scanned at baseline. Time point t = 2, which is comparable to t = 0 for the following cycle of BoNT treatment, was added in order to verify that possible BoNT-related effects were consistent over time. The timing of t = 1 was chosen because a maximal effect of the BoNT injections could be expected at this point. The mean delay between BoNT treatment and the MRI scan was 93.2 ± 13.9 days for the t = 0 scan and 93.2 ± 17.9 days for the t = 2 scan (related-samples Wilcoxon signed rank test interval scan-BoNT t = 0 vs. t = 2: Z = −.21, p = .83). Subjects laid supine with their eyes closed. The necessity of head immobility was emphasized to each subject, while head movements were minimized by an adjustable padded head holder. None of the subjects had dystonic symptoms during scanning, as confirmed in a post-scanning briefing. Images were acquired on a 3-T Siemens Magnetom Allegra Scanner (Erlangen, Germany) equipped with a 32-channel head coil. All scans were made with the same scanner; no scanner updates were
Table 1 Patient details. Patient no.
Sex
Age (years)
Handedness
Duration symptoms (years)
Duration BoNT treatment (years)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Mean (SD) or median (IQR)a
M M F F F F F M F F F F M F F M M F M F M M F F M Total
46 45 53 58 60 58 48 40 82 63 60 50 57 71 61 50 60 69 47 49 67 59 64 60.4 (9.3) 52.3 (8.8) 57.3 (9.8)
R R R R R R R R R R R R R R R L L R R R R R R
2 25 15 2 20 10 15 20 22 9 3 21 18 5 9 3 4 10 8 12 16 20 16 12.8 (6.9) 12.6 (8.1) 12.7 (7.2)
2 15 1 5 12 10 2 10 21 8 0.25 11 15 5 8 2 0.25 2 3 20 15 17 1 7.6 (6.7) 9.0 (6.9) 8.0 (6.7)
TWSTRS t=0 6 21 15 14 19 17 20 22 17 19 17 19 17 20 20 21 19 19 22 18 20 20 20 19 (17–19.8) 20 (19–21) 19 (17–20)
t=1 1 8 12 8 13 9 14 17 12 15 7 9 3 3 9 14 10 15 14 X 12 16 8 9 (8–13) 12 (8–14) 11 (8–14)
t=2 8 20 20 20 20 20 19 22 19 20 17 17 20 17 19 21 18 20 22 X 20 21 19 19 (19–20) 20 (20–21) 20 (19–20)
F = female; IQR = interquartile range (first quartile–third quartile); L = left; M = male; no. = number; R = right; TWSTRS = Toronto Western Spasmodic Torticollis Rating Scale; X = only MRI scan at t = 0. a Mean used for age and duration of symptoms/BoNT treatment; median used for TWSTRS scores.
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333
performed during the study. High-resolution anatomical images were acquired using an MP-RAGE sequence (TE/TR 3.03/2300 ms, 192 sagittal slices, voxel size 1.0 × 1.0 × 1.0 mm3, FOV 256 mm2, GRAPPA with acceleration factor 2 and 24 reference lines). Image pre-processing Before further analysis of the structural images, we visually confirmed that all images were clear of major head motion. The raw structural images were pre-processed using the default settings of the VBM8 toolbox (Structural Brain Mapping group, Jena, Germany, http://dbm/ neuro.uni-jena.de/vbm) incorporated in SPM8 (Wellcome Trust Centre for Neuroimaging, London, UK; http://www.fil.ion.ucl.ac.uk/spm) running under Matlab 7.9 (Mathworks, Natick, MA, USA). We performed three different analyses. We evaluated longitudinal GMV differences related to BoNT treatment within patients (Analysis 1); baseline GMV differences between CD patients' t = 0 and healthy controls (Analysis 2); and relationships between patients' t = 0 GMV and disease characteristics (Analysis 3). The longitudinal analysis (Analysis 1) required customized pre-processing that considered the characteristics of intra-subject analysis. For this purpose, we applied the default longitudinal pre-processing approach integrated in the VBM8 toolbox requiring the following pre-processing steps: a) twostep realignment to the mean image of the three time points for each subject, b) intra-subject bias correction, c) segmentation of the different tissue classes, d) spatial normalization using linear and DARTEL nonlinear registration, and e) smoothing with a 10 mm full-width at half-maximum isotropic Gaussian kernel. For Analyses 2 and 3 raw images were: a) bias corrected for MRI inhomogeneities and noise; b) normalized using linear and non-linear transformations using a diffeomorphic non-linear registration tool (diffeomorphic anatomical registration using exponentiated lie algebra [DARTEL]) to improve inter-subject registration (Ashburner, 2007), c) segmented into three tissue types (GM, white matter, and cerebrospinal fluid), and d) modulated through multiplication with non-linear components derived from the normalization matrix to preserve tissue volume after warping, and e) smoothed with a 10 mm full-width at half-maximum isotropic Gaussian kernel.
329
Results Analysis 1 First, we examined possible longitudinal changes in GMV between the patients' three time points. We observed that after BoNT treatment (Pt = 1 N Pt = 0) GMV was increased in an area within the right precentral sulcus at the lateral border of the premotor cortex (Fig. 1A, Table 2) (Mayka et al., 2006). A similar pattern, slightly more caudally, functionally corresponding to the right ventral premotor cortex (PMv) was found for Pt = 1 N Pt = 2 (Fig. 1B, Table 2) (Mayka et al., 2006). No differences were found between Pt = 0 and Pt = 2; these last two analyses were done to check internal consistency of the data. Analysis 2 Second, VBM was used to explore differences in regional GMV across the whole brain between CD patients' t = 0 and healthy controls. CD patients exhibited reduced cortical GMV in area 45, functionally corresponding to the left PMv (Fig. 1C, Table 2) (Mayka et al., 2006). In a post-hoc analysis in patients after BoNT treatment (patients' t = 1 vs. healthy controls) we observed that the GMV was decreased in the same area. No GMV increase was found for CD patients in comparison to controls. Analysis 3 Lastly, we assessed whether the GMV at patients' t = 0 related to disease characteristics. We observed a positive correlation between disease duration and GMV of the left supramarginal gyrus (SMG) within the inferior parietal cortex, r = .48, p b .05, and the left premotor cortex (PMC), r = .49, p b .05 (Eickhoff et al., 2005). A negative correlation was found between disease severity and GMV of the left inferior parietal lobule (IPL), r = −.60, p b .05. (Fig. 2, Table 2) We observed no significant correlation between BoNT duration and GMV of the lateral border of the right PMC. In a post-hoc analysis, no significant effect was found between GMV of the lateral border of the right PMC nor the left SMG and left PMC at patients' t = 1 and disease characteristics. Discussion
Statistical analysis As outlined above we performed three analyses using SPM8. First, we examined longitudinal differences in GMV between the patients' three time points using a repeated-measures ANOVA (flexible factorial model) with within-subject factor time (t = 0, t = 1, and t = 2), again adding age and gender as nuisance effects (Analysis 1). For this analysis, only 22 patients could be included, as one patient did not complete the full experimental procedure due to non-dystonia-related issues. Second, we examined whole-brain statistical differences in GMV between patients' t = 0 and controls using an independent ttest correcting for age and gender (Analysis 2). Last, we assessed the relationship between GMV at patients' t = 0 and disease duration, GMV at patients' t = 0 and disease severity, and GMV at patients' t = 0 and BoNT duration by multiple regression correcting for age and gender (Analysis 3). In a post-hoc analysis we evaluated whether the dystonia-related GMV changes and correlations with disease characteristics were also present in patients after BoNT-treatment. We therefore performed an independent t-test between healthy controls and patients' t = 1 and a multiple regression analysis for patients' t = 1 as outlined above. To avoid possible edge effects between different tissue types, absolute threshold masking was used to exclude voxels with GM values b.2. The cluster-level significance was assessed using a non-stationary cluster extent correction at p b .05 to correct for multiple comparisons (Hayasaka et al., 2004).
In this study of CD, we explored alterations in GMV and the modifying effect of BoNT injections, by means of VBM methodology in CD patients. An intriguing result was the observation that BoNT treatment was related to GMV increase in the right precentral sulcus at the lateral border of the premotor cortex. Also, we found evidence for loss of GMV in area 45 functionally corresponding to the left ventral premotor cortex in cervical dystonia patients. BoNT-related GMV alterations Interestingly, within patients we observed a GMV increase in the right precentral sulcus at the border of the premotor cortex after BoNT treatment. BoNT-related structural alterations have been reported before in a small selection (n = 6) of focal dystonia patients, demonstrating restoration of white matter changes after treatment with BoNT (Blood et al., 2006). Our finding of fast, treatment-dependent changes in GMV provides further evidence for ‘stimulus’-dependent structural plasticity of the adult healthy and diseased brain. This type of structural plasticity, has recently been reported after exercise, immobilization and learning in healthy adults and dystonia patients (Granert et al., 2011; Ceccarelli et al., 2009; Draganski et al., 2004). It may be displayed on the biological level through spine growth or retraction leading to synapse formation or elimination and with that altering synapse density, as shown in several animal studies (Trachtenberg et al., 2002). Following stimulus-dependent plasticity, we propose that the observed dynamic
330
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333
Fig. 1. Gray matter volume differences in cervical dystonia. Depicted are the gray matter volume abnormalities in cervical dystonia, controlled for age and gender (p b .05 corrected). A. The left panel shows the spatial distribution of GMV increase after BoNT treatment in CD patients: right precentral sulcus, at the lateral border of the PMC (Analysis 1; t = 1 N t = 0). In the right panel, the histogram shows the beta values of the GMV averaged over the significant cluster. B. The left panel shows the spatial distribution of GMV difference between t = 1 N t = 2 as control for the results presented at A: right PMv (Analysis 1; t = 1 N t = 2). In the right panel, the histogram shows the beta values of the GMV averaged over the significant cluster. C. The left panel shows the spatial distribution of GMV decrease for CD patients in comparison to healthy controls: left PMv (Analysis 2). In the right panel, the histogram shows the beta values of the GMV averaged over the significant cluster, for each group. Post-hoc analysis evaluating patients' t = 1 and healthy controls showed a similar effect. Images are tstatistics overlaid on the MNI-152 standard brain. The left hemisphere of the brain corresponds to the right side in this image.
structural GMV change could be caused by functional cerebral reorganization after peripheral treatment with BoNT injections, in which functional reorganization acts as the necessary stimulus. This is supported by the absence of a correlation between GMV in the right precentral sulcus and BoNT duration, which makes a more “permanent” GMV alteration induced by BoNT treatment unlikely. It has been postulated that BoNT exerts an indirect influence on cortical organization through the alteration of muscle spindle afferent input to the central nervous system, i.e. a neuroplastic response to peripheral sensory alterations (Curra et al., 2004). Indeed, functional reorganization after BoNT treatment has been demonstrated in primary sensorimotor and lateral premotor areas, both electrophysiologically and by means of fMRI
(Delnooz et al., 2013a; Kojovic et al., 2011). Based on our own fMRI data, we recently hypothesized that the BoNT-related functional organization of these higher order motor areas reflects a “reorganization” of primary motor planning defects in CD (Delnooz et al., 2013a). Since there is partial overlap between the functionally altered regions identified earlier and the regions that exhibit GMV alteration in our longitudinal BoNT-related analysis, we suggest that the observed GMV changes here may be a structural reflection of this previously hypothesized “reorganization”. The observation of only a significant GMV alteration after BoNT treatment in the right precentral sulcus and not in V2 and the motor cortex (in contrast to our earlier results (Delnooz et al., 2013a)), is due to a threshold effect as at an uncorrected statistical
Table 2 Local maxima of regions with altered gray matter volume. Analysis/contrast
Area (probability)
Side
X
Y
Z
p-value
Z-score
Cluster size (voxels)
1. Pt = 1 N Pt = 0 2. Pt = 1 N Pt = 2 3. C N Pt = 0 4. C N Pt = 1 5. Pt = 0 vs. duration
Lateral PMC (NA)a PMv (NA)a PMv (NA)a PMv (NA)a IPL PF (85%)b PMC (56%)b IPL PGa (55%)b
Right Right Left Left Left Left Right
45 48 −50 −50 −58 −44 51
6 15 21 20 −37 40 −60
48 28 24 24 37 51 34
.010 .032 .019 .030 .019 0.041 0.013
4.82 4.37 4.61 4.60 4.05 3.88 4.06
496 352 1134 977 1389 952 1625
6. Pt = 0 vs. severity
C = controls, IPL = intraparietal lobule, NA = not available, P = patients, PMC = premotor cortex, PMv = ventral premotor cortex. Coordinates are given in the Montreal Neurological Institute standard space. Statistical inference is set at corrected p b .05. a (Mayka et al., 2006) (precise probabilities are not available). b (Eickhoff et al., 2005).
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333
331
Fig. 2. Gray matter volume alterations correlated to disease characteristics. Depicted are the gray matter volume alterations in cervical dystonia, controlled for age and gender (p b .05 corrected). A. The left panel shows the cluster of voxels where GMV increased with disease duration: the left PMC and left SMG (Analysis 3). In the right panel, individual values are plotted separately for patients' t = 0 (left PMC: black diamonds; left SMG: gray squares). The solid line (black: PMC; gray: SMG) represents the linear relation between the beta values of GMV averaged over the significant cluster and disease duration. B. The left panel shows the cluster of voxels where GMV increased with disease severity: the right inferior parietal cortex (Analysis 3). In the right panel, individual values are plotted separately for patients' t = 0. The solid line represents the linear relation between the beta values of GMV averaged over the significant cluster and disease severity. Images are t-statistics overlaid on the MNI-152 standard brain. The left hemisphere of the brain corresponds to the right side in this image.
threshold these areas also show a treatment related GMV alteration. Our finding of BoNT-related GMV differences in the right hemisphere in contrast to the baseline GMV differences in the left hemisphere may be due to a threshold effect. A larger patient sample is needed to confirm this lateralized effect. Dystonia-related GMV alterations Our finding of dystonia-related GMV decrease in higher order motor areas is consistent with earlier functional imaging and also electrophysiological studies that indicate the involvement of prefrontal and premotor areas in CD. In several connectivity studies, abnormal coupling has been demonstrated between higher order motor areas, such as the ventral and dorsal PMC and prefrontal cortex, and several other sensorimotor cortical and subcortical areas (Delnooz et al., 2012, 2013a; Ibanez et al., 1999; Moore et al., 2012). Also, in a number of task-related imaging studies, abnormal activation of these regions has been demonstrated in focal dystonia (Naumann et al., 2000; Delnooz et al., 2013b). Although we need to be cautious in interpreting our VBM data mechanistically, we think that our findings further support the concept that abnormalities of higher order motor areas are crucial to the pathophysiology of dystonia, and in particular the areas that serve motor control and spatial computation (Hallett, 2000; Fiorio et al., 2007). The question is, are the structural alterations we have found here a primary deficit, a secondary phenomenon caused by longstanding functional alterations, or perhaps the result of compensatory efforts? Because results of previous studies in various forms of genetic and non-genetic dystonia have not been consistent in terms of which brain areas withheld GMV differences, we find it difficult to lump
all these data to come up with a summarizing interpretation. In our view, the observed base line GMV decrease of the left PMv is likely caused by functional deprivation of premotor and prefrontal areas that cause a persistent decline in neuronal processing (Granert et al., 2011; Taubert et al., 2012). Our observation of loss instead of gain of GMV supports this idea, as an increase of GMV has been shown to occur after e.g. training of motor sequences, which implies functional gain of these areas (Draganski et al., 2003). As regression analysis failed to show a significant correlation between disease characteristics and GMV differences in the left PMv and the GMV of this area is not influenced by BoNT treatment, we believe that this GMV difference reflects a rather static change, perhaps occurring early in the development of cervical dystonia. Nevertheless, it cannot be excluded that the GMV changes found are (in part) caused by differences in cognitive function between controls and patients as the observed changes were found in an brain area that is also important for e.g. language and memory (area 45). It would be of interest to further investigate this with the use of additional cognitive and language tests in future studies. For both the left SMG and the left PMC, this is different as we found a positive correlation between the GMV in these regions and disease duration. The same applies to the left inferior parietal lobule for which we observed a negative correlation between the GMV and disease severity. Both structural and functional abnormalities in these regions, which contribute to the integration of multimodal sensory information and motor planning, were recently demonstrated in focal dystonia (Delnooz et al., 2013a; de Vries et al., 2012; Opavsky et al., 2012). Still, this dynamic, disease characteristic-related alteration of GMV suggests an adaptation process of motor control and could be compatible with a role of these areas as either a pathological driver of dystonia or a more compensatory force.
332
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333
Interpretational issues
Contributorship statement
Our cross-sectional observations are partially in agreement with earlier, smaller sized studies concerning VBM in CD that also demonstrated decreased GMV in higher order motor areas, specifically the prefrontal cortex, premotor cortex, and supplementary motor area (Draganski et al., 2003; Pantano et al., 2011; Prell et al., 2013). By contrast, we did not find GMV alterations in the basal ganglia, the cerebellum, or the primary sensorimotor cortex (Draganski et al., 2003; Pantano et al., 2011; Granert et al., 2011). We attribute these discrepancies to methodological factors including differences in study population and the method used for data analysis. More specifically, imaging parameters and scanner characteristics influence GM segmentation. Also, the use of newer VBM analysis procedures (see Methods section) and the application of particular statistical methods (correction for non-stationarity of structural images and the use of whole-brain analysis without predefined cortical anatomical regions of interest or masks) may provide an explanation for the discrepancies with these previous, but recent studies (Draganski et al., 2003; Pantano et al., 2011; Prell et al., 2013). Alternatively, differences in population size, age, and disease duration or severity are important for GMV differences. Despite the inconsistent detection of GMV change in the basal ganglia in previous reports, the evaluation of basal ganglia GMV in our study could be considered a minor limitation. Our analysis may have been less sensitive for subcortical structures because of 1) the possible reduced reliability of basal ganglia segmentation due to the combination of common high iron content and the automated tissue classification algorithm in VBM8, and 2) the lower signal-to-noise ratio in the basal ganglia caused by our imaging parameters (32-channel head coil and GRAPPA imaging). It should be noted however, that alterations in subcortical GMV have been shown in several recent studies using similar methodology (Ryan et al., 2013; Lindholm et al., 2009). When interpreting the results it has to be taken into account that the patients' t = 0 scan is not a baseline scan in the strict sense, i.e. not a scan before any form of treatment, but a scan preceding BoNT injection within the regular BoNT treatment regime of the individual patient. It can therefore not be excluded that the GMV differences observed between patients and controls (Analysis 2) are not only caused by the disease process itself, but also may be influenced by long-term BoNT treatment. A similar issue could play a role in the longitudinal BoNT-related analysis (Analysis 1). Due to the absence of a BoNTnaive CD group, we cannot fully rule out that the GMV differences after BoNT treatment are related to the natural course of the disease itself rather than being purely BoNT-related. Our observation, however, that GMV did not differ between patients' t = 0 and t = 2 and that similar regions were related to BoNT-treatment when comparing the pre-treatment scans with the post-treatment scan (Pt = 1 N Pt = 0 and Pt = 1 N Pt = 2) support the thought that these dynamic GMV changes within patients following BoNT-treatment are not related to the disease itself but indeed to the given treatment. There were no differences in the quality of the scans between the separate time points, excluding variability in image quality as alternative explanation for the longitudinal BoNT-related effect. Nevertheless, it would be of great interest to try and replicate our findings in a group of BoNT-naive CD patients.
Planning of the study: CD and BvdW. Data collection: CD and JP. Statistical analysis: CD. Drafting the manuscript: CD. Critical review of the manuscript: JP and BvdW.
Conclusion We have used VBM to detect GMV abnormalities in patients with cervical dystonia. An intriguing finding was the increase in GMV related to BoNT-treatment in the right precentral sulcus, at the lateral border of the premotor cortex. We observed evidence for loss of GMV in higher order motor areas in contrast to healthy controls. These observations support the hypothesis that areas involved in motor planning and spatial cognition are crucial to the pathophysiology of cervical dystonia and that these areas are modified by BoNT indirectly, probably through peripheral sensory alterations.
Funding This study was funded by the Netherlands Brain Foundation, which had no further involvement in this study. Competing interests Cathérine Delnooz reports no disclosures. Jaco Pasman reports no disclosures. Bart van de Warrenburg receives research support from the Netherlands Brain Foundation, the Prinses Beatrix Fonds, the Gossweiler Foundation, the Radboud University Nijmegen Medical Centre, and the Royal Dutch Society for Physical therapy. References Cakmur, R., Donmez, B., Uzunel, F., et al., 2004. Evidence of widespread impairment of motor cortical inhibition in focal dystonia: a transcranial magnetic stimulation study in patients with blepharospasm and cervical dystonia. Adv. Neurol. 94, 37–44. Naumann, M., Magyar-Lehmann, S., Reiners, K., et al., 2000. Sensory tricks in cervical dystonia: perceptual dysbalance of parietal cortex modulates frontal motor programming. Ann. Neurol. 47, 322–328. Delnooz, C.C., Pasman, J.W., Beckmann, C.F., et al., 2013a. Task-free functional MRI in cervical dystonia reveals multi-network changes that partially normalize with botulinum toxin. PLoS One 8, e62877. de Vries, P.M., de Jong, B.M., Bohning, D.E., et al., 2012. Reduced parietal activation in cervical dystonia after parietal TMS interleaved with fMRI. Clin. Neurol. Neurosurg. 114, 914–921. Blood, A.J., Tuch, D.S., Makris, N., et al., 2006. White matter abnormalities in dystonia normalize after botulinum toxin treatment. Neuroreport 17, 1251–1255. Fabbrini, G., Pantano, P., Totaro, P., et al., 2008. Diffusion tensor imaging in patients with primary cervical dystonia and in patients with blepharospasm. Eur. J. Neurol. 15, 185–189. Draganski, B., Thun-Hohenstein, C., Bogdahn, U., et al., 2003. “Motor circuit” gray matter changes in idiopathic cervical dystonia. Neurology 61, 1228–1231. Pantano, P., Totaro, P., Fabbrini, G., et al., 2011. A transverse and longitudinal MR imaging voxel-based morphometry study in patients with primary cervical dystonia. AJNR Am. J. Neuroradiol. 32, 81–84. Prell, T., Peschel, T., Kohler, B., et al., 2013. Structural brain abnormalities in cervical dystonia. BMC Neurosci. 14, 123. Kojovic, M., Caronni, A., Bologna, M., et al., 2011. Botulinum toxin injections reduce associative plasticity in patients with primary dystonia. Mov. Disord. 26, 1282–1289. Thickbroom, G.W., Byrnes, M.L., Stell, R., et al., 2003. Reversible reorganisation of the motor cortical representation of the hand in cervical dystonia. Mov. Disord. 18, 395–402. Ashburner, J., Friston, K.J., 2000. Voxel-based morphometry—the methods. NeuroImage 11, 805–821. Ashburner, J., 2007. A fast diffeomorphic image registration algorithm. NeuroImage 38, 95–113. Hayasaka, S., Phan, K.L., Liberzon, I., et al., 2004. Nonstationary cluster-size inference with random field and permutation methods. NeuroImage 22, 676–687. Mayka, M.A., Corcos, D.M., Leurgans, S.E., et al., 2006. Three-dimensional locations and boundaries of motor and premotor cortices as defined by functional brain imaging: a meta-analysis. NeuroImage 31, 1453–1474. Eickhoff, S.B., Stephan, K.E., Mohlberg, H., et al., 2005. A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data. NeuroImage 25, 1325–1335. Granert, O., Peller, M., Gaser, C., et al., 2011. Manual activity shapes structure and function in contralateral human motor hand area. NeuroImage 54, 32–41. Ceccarelli, A., Rocca, M.A., Pagani, E., et al., 2009. Cognitive learning is associated with gray matter changes in healthy human individuals: a tensor-based morphometry study. NeuroImage 48, 585–589. Draganski, B., Gaser, C., Busch, V., et al., 2004. Neuroplasticity: changes in grey matter induced by training. Nature 427, 311–312. Trachtenberg, J.T., Chen, B.E., Knott, G.W., et al., 2002. Long-term in vivo imaging of experience-dependent synaptic plasticity in adult cortex. Nature 420, 788–794. Curra, A., Trompetto, C., Abbruzzese, G., et al., 2004. Central effects of botulinum toxin type A: evidence and supposition. Mov. Disord. 19 (Suppl. 8), S60–S64. Ibanez, V., Sadato, N., Karp, B., et al., 1999. Deficient activation of the motor cortical network in patients with writer's cramp. Neurology 53, 96–105. Moore, R.D., Gallea, C., Horovitz, S.G., et al., 2012. Individuated finger control in focal hand dystonia: an fMRI study. NeuroImage 61, 823–831. Delnooz, C.C., Helmich, R.C., Toni, I., et al., 2012. Reduced parietal connectivity with a premotor writing area in writer's cramp. Mov. Disord. 27, 1425–1431.
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333 Delnooz, C.C., Helmich, R.C., Medendorp, W.P., et al., 2013b. Writer's cramp: increased dorsal premotor activity during intended writing. Hum. Brain Mapp. 34, 613–625. Hallett, M., 2000. Disorder of movement preparation in dystonia. Brain 123 (Pt 9), 1765–1766. Fiorio, M., Tinazzi, M., Ionta, S., et al., 2007. Mental rotation of body parts and noncorporeal objects in patients with idiopathic cervical dystonia. Neuropsychologia 45, 2346–2354. Taubert, M., Villringer, A., Ragert, P., 2012. Learning-related gray and white matter changes in humans: an update. Neuroscientist 18, 320–325.
333
Opavsky, R., Hlustik, P., Otruba, P., et al., 2012. Somatosensory cortical activation in cervical dystonia and its modulation with botulinum toxin: an FMRI study. Int. J. Neurosci. 122, 45–52. Ryan, N.S., Keihaninejad, S., Shakespeare, T.J., et al., 2013. Magnetic resonance imaging evidence for presymptomatic change in thalamus and caudate in familial Alzheimer's disease. Brain 136, 1399–1414. Lindholm, T.L., Botes, L., Engman, E.L., et al., 2009. Parallel imaging: is GRAPPA a useful acquisition tool for MR imaging intended for volumetric brain analysis? BMC Med. Imaging 9, 15.
Contents lists available at ScienceDirect
Neurobiology of Disease journal homepage: www.elsevier.com/locate/ynbdi
Dynamic cortical gray matter volume changes after botulinum toxin in cervical dystonia Cathérine C.S. Delnooz, Jaco W. Pasman, Bart P.C. van de Warrenburg ⁎ Radboud University Nijmegen Medical Centre, Donders Institute for Brain, Cognition and Behaviour, Department of Neurology, The Netherlands
a r t i c l e
i n f o
Article history: Received 15 May 2014 Revised 15 September 2014 Accepted 20 October 2014 Available online 28 October 2014 Keywords: Cervical dystonia Motor planning Botulinum toxin Structural plasticity Spatial computation
a b s t r a c t Previous electrophysiological and functional imaging studies in focal dystonia have reported on cerebral reorganization after botulinum toxin (BoNT) injections. With the exception of microstructural changes, alterations in gray matter volume after BoNT have not been explored. In this study, we sought to determine whether BoNT influences gray matter volume in a group of cervical dystonia (CD) patients. We analyzed whole brain gray matter volume in a sample of CD patients with VBM analysis. In patients, scans were repeated immediately before and some weeks after BoNT injections; controls were only scanned once. We analyzed 1) BoNT-related gray matter volume changes within patients; 2) gray matter volume differences between patients and controls; and 3) correlations between gray matter volume and disease duration and disease severity. The pre- and post-BoNT treatment analysis revealed an increase of gray matter volume within the right precentral sulcus, at the lateral border of the premotor cortex. In comparison to healthy controls, CD patients had reduced gray matter volume in area 45 functionally corresponding to the left ventral premotor cortex. No gray matter volume increase was found for CD patients in comparison to controls. Gray matter volume of the left supramarginal gyrus and left premotor cortex correlated positively with disease duration, and that of the right inferior parietal lobule correlated negatively with disease severity. We have identified structural, yet dynamic gray matter volume changes in CD. There were specific gray matter volume changes related to BoNT injections, illustrating indirect central consequences of modified peripheral sensory input. As differences were exclusively seen in higher order motor areas relevant to motor planning and spatial cognition, these observations support the hypothesis that deficits in these cognitive processes are crucial in the pathophysiology of CD. © 2014 Elsevier Inc. All rights reserved.
Introduction Primary cervical dystonia (CD) is characterized by involuntary abnormal movements and postures of the head and neck due to sustained contractions of the cervical musculature. Injecting the involved cervical muscles with botulinum toxin (BoNT) is an effective and evidencebased treatment of CD. In contrast to an identifiable cause in secondary dystonia, e.g. infarction of the basal ganglia, primary dystonia – including primary CD – has no evidence of an underlying cause, except for a possible genetic background as for example ANO3, GNAL and CIZ1. Although CD patients have several abnormalities in motor system physiology (Cakmur et al., 2004), its precise pathophysiology still remains to be clarified. In recent years, several functional cerebral abnormalities have ⁎ Corresponding author at: Department of Neurology (943), Radboud University Nijmegen Medical Centre PO box 9101, 6500 HB Nijmegen, The Netherlands. E-mail address: [email protected] (B.P.C. van de Warrenburg). Available online on ScienceDirect (www.sciencedirect.com).
http://dx.doi.org/10.1016/j.nbd.2014.10.013 0969-9961/© 2014 Elsevier Inc. All rights reserved.
been found in CD by means of functional magnetic resonance imaging (fMRI) (Naumann et al., 2000; Delnooz et al., 2013a; de Vries et al., 2012), which may originate from underlying structural abnormalities. Structural differences have been shown using diffusion tensor imaging, demonstrating altered structural integrity in the striatum, corpus callosum and the cortical motor circuitry (Blood et al., 2006; Fabbrini et al., 2008). Studies using voxel-based morphometry (VBM) reported alterations in gray matter (GM) volume in the basal ganglia (Draganski et al., 2003; Pantano et al., 2011; Prell et al., 2013), the cortical motor circuitry (Draganski et al., 2003; Pantano et al., 2011; Prell et al., 2013), and the cerebellum in CD (Draganski et al., 2003; Prell et al., 2013). Despite several reports on structural differences between dystonia patients and healthy controls at baseline, few studies exist on longitudinal structural changes. Interestingly, several electrophysiological and functional imaging studies in focal dystonia have reported on cerebral reorganization after BoNT treatment (Delnooz et al., 2013a; Kojovic et al., 2011; Thickbroom et al., 2003). Even more, Blood et al. reported on structural white matter alterations after BoNT injections
328
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333
(Blood et al., 2006). Following these previous reports, we sought to investigate the possible influence of BoNT on structural gray matter volume in CD patients. Since baseline differences in GM volume have not been consistent across studies, probably due to small sample sizes, clinical heterogeneity of the patients studied, and different methodological approaches (e.g. region of interest analysis versus whole brain analysis, and with or without correction for non-stationarity), it remains unclear which GM abnormalities in CD are sufficiently consistent to be regarded as the structural fingerprint of CD. VBM of T1-weighted structural MRI images provides information about the regional differences in volumetric organization of the brain by measurement of the quantity of tissue within a voxel (Ashburner and Friston, 2000). Regional GM alterations have been demonstrated in several movement disorders. VBM has also been of value in the identification of longitudinal GM alterations in neurological diseases, reflecting for example disease progression or treatment response (Pantano et al., 2011). In this study, we tested the hypothesis that BoNT treatment modifies the ‘dystonic brain’, by evaluating wholebrain GM volume (GMV) by means of VBM methodology in a longitudinal pre- and post-treatment design within CD patients. Also, we investigated GMV in a sample of CD patients and a group of matched, healthy controls, and the relationships between GMV and disease severity, disease duration and duration of BoNT treatment. Methods Subjects Twenty-three primary CD patients (14 women; mean age 57.2 years, 21 right-handed) and 22 healthy controls (12 women; mean age 54.5 years, 22 right-handed) participated after giving informed consent. The genetic status was not evaluated. The study was approved by the Ethics Committee of the Radboud University Nijmegen Medical Centre. The exclusion criteria included age under
18 years, severe head tremor or dystonia outside the cervical region, and absence of regular BoNT treatment. In two thirds of the patients, torticollis was the dominant feature, of which 9 patients presented with torticollis to the left. In the other patients, laterocollis was the predominant symptom (Table 1). Patients were all being treated with BoNT type A (Dysport®) every 2 to 4 months (mean duration of BoNT treatment: 7.6 years). No other neurotropic medication was used. Dystonia severity was measured with the Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS) severity subscore. The results of TWSTRS scores ranged from 6 to 22 (median 19) before injection (t = 0) and from 1 to 17 (median 11) after injection (t = 1), demonstrating a significant improvement due to BoNT (related-samples Wilcoxon signed rank test TWSTRS t = 0 vs. t = 1: Z = −4.11, p = .00). Image acquisition To investigate treatment-related changes, patients were scanned three times: before BoNT injection (t = 0), after 4–5 weeks (t = 1), and just before the next BoNT treatment (t = 2). Healthy controls were only scanned at baseline. Time point t = 2, which is comparable to t = 0 for the following cycle of BoNT treatment, was added in order to verify that possible BoNT-related effects were consistent over time. The timing of t = 1 was chosen because a maximal effect of the BoNT injections could be expected at this point. The mean delay between BoNT treatment and the MRI scan was 93.2 ± 13.9 days for the t = 0 scan and 93.2 ± 17.9 days for the t = 2 scan (related-samples Wilcoxon signed rank test interval scan-BoNT t = 0 vs. t = 2: Z = −.21, p = .83). Subjects laid supine with their eyes closed. The necessity of head immobility was emphasized to each subject, while head movements were minimized by an adjustable padded head holder. None of the subjects had dystonic symptoms during scanning, as confirmed in a post-scanning briefing. Images were acquired on a 3-T Siemens Magnetom Allegra Scanner (Erlangen, Germany) equipped with a 32-channel head coil. All scans were made with the same scanner; no scanner updates were
Table 1 Patient details. Patient no.
Sex
Age (years)
Handedness
Duration symptoms (years)
Duration BoNT treatment (years)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Mean (SD) or median (IQR)a
M M F F F F F M F F F F M F F M M F M F M M F F M Total
46 45 53 58 60 58 48 40 82 63 60 50 57 71 61 50 60 69 47 49 67 59 64 60.4 (9.3) 52.3 (8.8) 57.3 (9.8)
R R R R R R R R R R R R R R R L L R R R R R R
2 25 15 2 20 10 15 20 22 9 3 21 18 5 9 3 4 10 8 12 16 20 16 12.8 (6.9) 12.6 (8.1) 12.7 (7.2)
2 15 1 5 12 10 2 10 21 8 0.25 11 15 5 8 2 0.25 2 3 20 15 17 1 7.6 (6.7) 9.0 (6.9) 8.0 (6.7)
TWSTRS t=0 6 21 15 14 19 17 20 22 17 19 17 19 17 20 20 21 19 19 22 18 20 20 20 19 (17–19.8) 20 (19–21) 19 (17–20)
t=1 1 8 12 8 13 9 14 17 12 15 7 9 3 3 9 14 10 15 14 X 12 16 8 9 (8–13) 12 (8–14) 11 (8–14)
t=2 8 20 20 20 20 20 19 22 19 20 17 17 20 17 19 21 18 20 22 X 20 21 19 19 (19–20) 20 (20–21) 20 (19–20)
F = female; IQR = interquartile range (first quartile–third quartile); L = left; M = male; no. = number; R = right; TWSTRS = Toronto Western Spasmodic Torticollis Rating Scale; X = only MRI scan at t = 0. a Mean used for age and duration of symptoms/BoNT treatment; median used for TWSTRS scores.
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333
performed during the study. High-resolution anatomical images were acquired using an MP-RAGE sequence (TE/TR 3.03/2300 ms, 192 sagittal slices, voxel size 1.0 × 1.0 × 1.0 mm3, FOV 256 mm2, GRAPPA with acceleration factor 2 and 24 reference lines). Image pre-processing Before further analysis of the structural images, we visually confirmed that all images were clear of major head motion. The raw structural images were pre-processed using the default settings of the VBM8 toolbox (Structural Brain Mapping group, Jena, Germany, http://dbm/ neuro.uni-jena.de/vbm) incorporated in SPM8 (Wellcome Trust Centre for Neuroimaging, London, UK; http://www.fil.ion.ucl.ac.uk/spm) running under Matlab 7.9 (Mathworks, Natick, MA, USA). We performed three different analyses. We evaluated longitudinal GMV differences related to BoNT treatment within patients (Analysis 1); baseline GMV differences between CD patients' t = 0 and healthy controls (Analysis 2); and relationships between patients' t = 0 GMV and disease characteristics (Analysis 3). The longitudinal analysis (Analysis 1) required customized pre-processing that considered the characteristics of intra-subject analysis. For this purpose, we applied the default longitudinal pre-processing approach integrated in the VBM8 toolbox requiring the following pre-processing steps: a) twostep realignment to the mean image of the three time points for each subject, b) intra-subject bias correction, c) segmentation of the different tissue classes, d) spatial normalization using linear and DARTEL nonlinear registration, and e) smoothing with a 10 mm full-width at half-maximum isotropic Gaussian kernel. For Analyses 2 and 3 raw images were: a) bias corrected for MRI inhomogeneities and noise; b) normalized using linear and non-linear transformations using a diffeomorphic non-linear registration tool (diffeomorphic anatomical registration using exponentiated lie algebra [DARTEL]) to improve inter-subject registration (Ashburner, 2007), c) segmented into three tissue types (GM, white matter, and cerebrospinal fluid), and d) modulated through multiplication with non-linear components derived from the normalization matrix to preserve tissue volume after warping, and e) smoothed with a 10 mm full-width at half-maximum isotropic Gaussian kernel.
329
Results Analysis 1 First, we examined possible longitudinal changes in GMV between the patients' three time points. We observed that after BoNT treatment (Pt = 1 N Pt = 0) GMV was increased in an area within the right precentral sulcus at the lateral border of the premotor cortex (Fig. 1A, Table 2) (Mayka et al., 2006). A similar pattern, slightly more caudally, functionally corresponding to the right ventral premotor cortex (PMv) was found for Pt = 1 N Pt = 2 (Fig. 1B, Table 2) (Mayka et al., 2006). No differences were found between Pt = 0 and Pt = 2; these last two analyses were done to check internal consistency of the data. Analysis 2 Second, VBM was used to explore differences in regional GMV across the whole brain between CD patients' t = 0 and healthy controls. CD patients exhibited reduced cortical GMV in area 45, functionally corresponding to the left PMv (Fig. 1C, Table 2) (Mayka et al., 2006). In a post-hoc analysis in patients after BoNT treatment (patients' t = 1 vs. healthy controls) we observed that the GMV was decreased in the same area. No GMV increase was found for CD patients in comparison to controls. Analysis 3 Lastly, we assessed whether the GMV at patients' t = 0 related to disease characteristics. We observed a positive correlation between disease duration and GMV of the left supramarginal gyrus (SMG) within the inferior parietal cortex, r = .48, p b .05, and the left premotor cortex (PMC), r = .49, p b .05 (Eickhoff et al., 2005). A negative correlation was found between disease severity and GMV of the left inferior parietal lobule (IPL), r = −.60, p b .05. (Fig. 2, Table 2) We observed no significant correlation between BoNT duration and GMV of the lateral border of the right PMC. In a post-hoc analysis, no significant effect was found between GMV of the lateral border of the right PMC nor the left SMG and left PMC at patients' t = 1 and disease characteristics. Discussion
Statistical analysis As outlined above we performed three analyses using SPM8. First, we examined longitudinal differences in GMV between the patients' three time points using a repeated-measures ANOVA (flexible factorial model) with within-subject factor time (t = 0, t = 1, and t = 2), again adding age and gender as nuisance effects (Analysis 1). For this analysis, only 22 patients could be included, as one patient did not complete the full experimental procedure due to non-dystonia-related issues. Second, we examined whole-brain statistical differences in GMV between patients' t = 0 and controls using an independent ttest correcting for age and gender (Analysis 2). Last, we assessed the relationship between GMV at patients' t = 0 and disease duration, GMV at patients' t = 0 and disease severity, and GMV at patients' t = 0 and BoNT duration by multiple regression correcting for age and gender (Analysis 3). In a post-hoc analysis we evaluated whether the dystonia-related GMV changes and correlations with disease characteristics were also present in patients after BoNT-treatment. We therefore performed an independent t-test between healthy controls and patients' t = 1 and a multiple regression analysis for patients' t = 1 as outlined above. To avoid possible edge effects between different tissue types, absolute threshold masking was used to exclude voxels with GM values b.2. The cluster-level significance was assessed using a non-stationary cluster extent correction at p b .05 to correct for multiple comparisons (Hayasaka et al., 2004).
In this study of CD, we explored alterations in GMV and the modifying effect of BoNT injections, by means of VBM methodology in CD patients. An intriguing result was the observation that BoNT treatment was related to GMV increase in the right precentral sulcus at the lateral border of the premotor cortex. Also, we found evidence for loss of GMV in area 45 functionally corresponding to the left ventral premotor cortex in cervical dystonia patients. BoNT-related GMV alterations Interestingly, within patients we observed a GMV increase in the right precentral sulcus at the border of the premotor cortex after BoNT treatment. BoNT-related structural alterations have been reported before in a small selection (n = 6) of focal dystonia patients, demonstrating restoration of white matter changes after treatment with BoNT (Blood et al., 2006). Our finding of fast, treatment-dependent changes in GMV provides further evidence for ‘stimulus’-dependent structural plasticity of the adult healthy and diseased brain. This type of structural plasticity, has recently been reported after exercise, immobilization and learning in healthy adults and dystonia patients (Granert et al., 2011; Ceccarelli et al., 2009; Draganski et al., 2004). It may be displayed on the biological level through spine growth or retraction leading to synapse formation or elimination and with that altering synapse density, as shown in several animal studies (Trachtenberg et al., 2002). Following stimulus-dependent plasticity, we propose that the observed dynamic
330
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333
Fig. 1. Gray matter volume differences in cervical dystonia. Depicted are the gray matter volume abnormalities in cervical dystonia, controlled for age and gender (p b .05 corrected). A. The left panel shows the spatial distribution of GMV increase after BoNT treatment in CD patients: right precentral sulcus, at the lateral border of the PMC (Analysis 1; t = 1 N t = 0). In the right panel, the histogram shows the beta values of the GMV averaged over the significant cluster. B. The left panel shows the spatial distribution of GMV difference between t = 1 N t = 2 as control for the results presented at A: right PMv (Analysis 1; t = 1 N t = 2). In the right panel, the histogram shows the beta values of the GMV averaged over the significant cluster. C. The left panel shows the spatial distribution of GMV decrease for CD patients in comparison to healthy controls: left PMv (Analysis 2). In the right panel, the histogram shows the beta values of the GMV averaged over the significant cluster, for each group. Post-hoc analysis evaluating patients' t = 1 and healthy controls showed a similar effect. Images are tstatistics overlaid on the MNI-152 standard brain. The left hemisphere of the brain corresponds to the right side in this image.
structural GMV change could be caused by functional cerebral reorganization after peripheral treatment with BoNT injections, in which functional reorganization acts as the necessary stimulus. This is supported by the absence of a correlation between GMV in the right precentral sulcus and BoNT duration, which makes a more “permanent” GMV alteration induced by BoNT treatment unlikely. It has been postulated that BoNT exerts an indirect influence on cortical organization through the alteration of muscle spindle afferent input to the central nervous system, i.e. a neuroplastic response to peripheral sensory alterations (Curra et al., 2004). Indeed, functional reorganization after BoNT treatment has been demonstrated in primary sensorimotor and lateral premotor areas, both electrophysiologically and by means of fMRI
(Delnooz et al., 2013a; Kojovic et al., 2011). Based on our own fMRI data, we recently hypothesized that the BoNT-related functional organization of these higher order motor areas reflects a “reorganization” of primary motor planning defects in CD (Delnooz et al., 2013a). Since there is partial overlap between the functionally altered regions identified earlier and the regions that exhibit GMV alteration in our longitudinal BoNT-related analysis, we suggest that the observed GMV changes here may be a structural reflection of this previously hypothesized “reorganization”. The observation of only a significant GMV alteration after BoNT treatment in the right precentral sulcus and not in V2 and the motor cortex (in contrast to our earlier results (Delnooz et al., 2013a)), is due to a threshold effect as at an uncorrected statistical
Table 2 Local maxima of regions with altered gray matter volume. Analysis/contrast
Area (probability)
Side
X
Y
Z
p-value
Z-score
Cluster size (voxels)
1. Pt = 1 N Pt = 0 2. Pt = 1 N Pt = 2 3. C N Pt = 0 4. C N Pt = 1 5. Pt = 0 vs. duration
Lateral PMC (NA)a PMv (NA)a PMv (NA)a PMv (NA)a IPL PF (85%)b PMC (56%)b IPL PGa (55%)b
Right Right Left Left Left Left Right
45 48 −50 −50 −58 −44 51
6 15 21 20 −37 40 −60
48 28 24 24 37 51 34
.010 .032 .019 .030 .019 0.041 0.013
4.82 4.37 4.61 4.60 4.05 3.88 4.06
496 352 1134 977 1389 952 1625
6. Pt = 0 vs. severity
C = controls, IPL = intraparietal lobule, NA = not available, P = patients, PMC = premotor cortex, PMv = ventral premotor cortex. Coordinates are given in the Montreal Neurological Institute standard space. Statistical inference is set at corrected p b .05. a (Mayka et al., 2006) (precise probabilities are not available). b (Eickhoff et al., 2005).
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333
331
Fig. 2. Gray matter volume alterations correlated to disease characteristics. Depicted are the gray matter volume alterations in cervical dystonia, controlled for age and gender (p b .05 corrected). A. The left panel shows the cluster of voxels where GMV increased with disease duration: the left PMC and left SMG (Analysis 3). In the right panel, individual values are plotted separately for patients' t = 0 (left PMC: black diamonds; left SMG: gray squares). The solid line (black: PMC; gray: SMG) represents the linear relation between the beta values of GMV averaged over the significant cluster and disease duration. B. The left panel shows the cluster of voxels where GMV increased with disease severity: the right inferior parietal cortex (Analysis 3). In the right panel, individual values are plotted separately for patients' t = 0. The solid line represents the linear relation between the beta values of GMV averaged over the significant cluster and disease severity. Images are t-statistics overlaid on the MNI-152 standard brain. The left hemisphere of the brain corresponds to the right side in this image.
threshold these areas also show a treatment related GMV alteration. Our finding of BoNT-related GMV differences in the right hemisphere in contrast to the baseline GMV differences in the left hemisphere may be due to a threshold effect. A larger patient sample is needed to confirm this lateralized effect. Dystonia-related GMV alterations Our finding of dystonia-related GMV decrease in higher order motor areas is consistent with earlier functional imaging and also electrophysiological studies that indicate the involvement of prefrontal and premotor areas in CD. In several connectivity studies, abnormal coupling has been demonstrated between higher order motor areas, such as the ventral and dorsal PMC and prefrontal cortex, and several other sensorimotor cortical and subcortical areas (Delnooz et al., 2012, 2013a; Ibanez et al., 1999; Moore et al., 2012). Also, in a number of task-related imaging studies, abnormal activation of these regions has been demonstrated in focal dystonia (Naumann et al., 2000; Delnooz et al., 2013b). Although we need to be cautious in interpreting our VBM data mechanistically, we think that our findings further support the concept that abnormalities of higher order motor areas are crucial to the pathophysiology of dystonia, and in particular the areas that serve motor control and spatial computation (Hallett, 2000; Fiorio et al., 2007). The question is, are the structural alterations we have found here a primary deficit, a secondary phenomenon caused by longstanding functional alterations, or perhaps the result of compensatory efforts? Because results of previous studies in various forms of genetic and non-genetic dystonia have not been consistent in terms of which brain areas withheld GMV differences, we find it difficult to lump
all these data to come up with a summarizing interpretation. In our view, the observed base line GMV decrease of the left PMv is likely caused by functional deprivation of premotor and prefrontal areas that cause a persistent decline in neuronal processing (Granert et al., 2011; Taubert et al., 2012). Our observation of loss instead of gain of GMV supports this idea, as an increase of GMV has been shown to occur after e.g. training of motor sequences, which implies functional gain of these areas (Draganski et al., 2003). As regression analysis failed to show a significant correlation between disease characteristics and GMV differences in the left PMv and the GMV of this area is not influenced by BoNT treatment, we believe that this GMV difference reflects a rather static change, perhaps occurring early in the development of cervical dystonia. Nevertheless, it cannot be excluded that the GMV changes found are (in part) caused by differences in cognitive function between controls and patients as the observed changes were found in an brain area that is also important for e.g. language and memory (area 45). It would be of interest to further investigate this with the use of additional cognitive and language tests in future studies. For both the left SMG and the left PMC, this is different as we found a positive correlation between the GMV in these regions and disease duration. The same applies to the left inferior parietal lobule for which we observed a negative correlation between the GMV and disease severity. Both structural and functional abnormalities in these regions, which contribute to the integration of multimodal sensory information and motor planning, were recently demonstrated in focal dystonia (Delnooz et al., 2013a; de Vries et al., 2012; Opavsky et al., 2012). Still, this dynamic, disease characteristic-related alteration of GMV suggests an adaptation process of motor control and could be compatible with a role of these areas as either a pathological driver of dystonia or a more compensatory force.
332
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333
Interpretational issues
Contributorship statement
Our cross-sectional observations are partially in agreement with earlier, smaller sized studies concerning VBM in CD that also demonstrated decreased GMV in higher order motor areas, specifically the prefrontal cortex, premotor cortex, and supplementary motor area (Draganski et al., 2003; Pantano et al., 2011; Prell et al., 2013). By contrast, we did not find GMV alterations in the basal ganglia, the cerebellum, or the primary sensorimotor cortex (Draganski et al., 2003; Pantano et al., 2011; Granert et al., 2011). We attribute these discrepancies to methodological factors including differences in study population and the method used for data analysis. More specifically, imaging parameters and scanner characteristics influence GM segmentation. Also, the use of newer VBM analysis procedures (see Methods section) and the application of particular statistical methods (correction for non-stationarity of structural images and the use of whole-brain analysis without predefined cortical anatomical regions of interest or masks) may provide an explanation for the discrepancies with these previous, but recent studies (Draganski et al., 2003; Pantano et al., 2011; Prell et al., 2013). Alternatively, differences in population size, age, and disease duration or severity are important for GMV differences. Despite the inconsistent detection of GMV change in the basal ganglia in previous reports, the evaluation of basal ganglia GMV in our study could be considered a minor limitation. Our analysis may have been less sensitive for subcortical structures because of 1) the possible reduced reliability of basal ganglia segmentation due to the combination of common high iron content and the automated tissue classification algorithm in VBM8, and 2) the lower signal-to-noise ratio in the basal ganglia caused by our imaging parameters (32-channel head coil and GRAPPA imaging). It should be noted however, that alterations in subcortical GMV have been shown in several recent studies using similar methodology (Ryan et al., 2013; Lindholm et al., 2009). When interpreting the results it has to be taken into account that the patients' t = 0 scan is not a baseline scan in the strict sense, i.e. not a scan before any form of treatment, but a scan preceding BoNT injection within the regular BoNT treatment regime of the individual patient. It can therefore not be excluded that the GMV differences observed between patients and controls (Analysis 2) are not only caused by the disease process itself, but also may be influenced by long-term BoNT treatment. A similar issue could play a role in the longitudinal BoNT-related analysis (Analysis 1). Due to the absence of a BoNTnaive CD group, we cannot fully rule out that the GMV differences after BoNT treatment are related to the natural course of the disease itself rather than being purely BoNT-related. Our observation, however, that GMV did not differ between patients' t = 0 and t = 2 and that similar regions were related to BoNT-treatment when comparing the pre-treatment scans with the post-treatment scan (Pt = 1 N Pt = 0 and Pt = 1 N Pt = 2) support the thought that these dynamic GMV changes within patients following BoNT-treatment are not related to the disease itself but indeed to the given treatment. There were no differences in the quality of the scans between the separate time points, excluding variability in image quality as alternative explanation for the longitudinal BoNT-related effect. Nevertheless, it would be of great interest to try and replicate our findings in a group of BoNT-naive CD patients.
Planning of the study: CD and BvdW. Data collection: CD and JP. Statistical analysis: CD. Drafting the manuscript: CD. Critical review of the manuscript: JP and BvdW.
Conclusion We have used VBM to detect GMV abnormalities in patients with cervical dystonia. An intriguing finding was the increase in GMV related to BoNT-treatment in the right precentral sulcus, at the lateral border of the premotor cortex. We observed evidence for loss of GMV in higher order motor areas in contrast to healthy controls. These observations support the hypothesis that areas involved in motor planning and spatial cognition are crucial to the pathophysiology of cervical dystonia and that these areas are modified by BoNT indirectly, probably through peripheral sensory alterations.
Funding This study was funded by the Netherlands Brain Foundation, which had no further involvement in this study. Competing interests Cathérine Delnooz reports no disclosures. Jaco Pasman reports no disclosures. Bart van de Warrenburg receives research support from the Netherlands Brain Foundation, the Prinses Beatrix Fonds, the Gossweiler Foundation, the Radboud University Nijmegen Medical Centre, and the Royal Dutch Society for Physical therapy. References Cakmur, R., Donmez, B., Uzunel, F., et al., 2004. Evidence of widespread impairment of motor cortical inhibition in focal dystonia: a transcranial magnetic stimulation study in patients with blepharospasm and cervical dystonia. Adv. Neurol. 94, 37–44. Naumann, M., Magyar-Lehmann, S., Reiners, K., et al., 2000. Sensory tricks in cervical dystonia: perceptual dysbalance of parietal cortex modulates frontal motor programming. Ann. Neurol. 47, 322–328. Delnooz, C.C., Pasman, J.W., Beckmann, C.F., et al., 2013a. Task-free functional MRI in cervical dystonia reveals multi-network changes that partially normalize with botulinum toxin. PLoS One 8, e62877. de Vries, P.M., de Jong, B.M., Bohning, D.E., et al., 2012. Reduced parietal activation in cervical dystonia after parietal TMS interleaved with fMRI. Clin. Neurol. Neurosurg. 114, 914–921. Blood, A.J., Tuch, D.S., Makris, N., et al., 2006. White matter abnormalities in dystonia normalize after botulinum toxin treatment. Neuroreport 17, 1251–1255. Fabbrini, G., Pantano, P., Totaro, P., et al., 2008. Diffusion tensor imaging in patients with primary cervical dystonia and in patients with blepharospasm. Eur. J. Neurol. 15, 185–189. Draganski, B., Thun-Hohenstein, C., Bogdahn, U., et al., 2003. “Motor circuit” gray matter changes in idiopathic cervical dystonia. Neurology 61, 1228–1231. Pantano, P., Totaro, P., Fabbrini, G., et al., 2011. A transverse and longitudinal MR imaging voxel-based morphometry study in patients with primary cervical dystonia. AJNR Am. J. Neuroradiol. 32, 81–84. Prell, T., Peschel, T., Kohler, B., et al., 2013. Structural brain abnormalities in cervical dystonia. BMC Neurosci. 14, 123. Kojovic, M., Caronni, A., Bologna, M., et al., 2011. Botulinum toxin injections reduce associative plasticity in patients with primary dystonia. Mov. Disord. 26, 1282–1289. Thickbroom, G.W., Byrnes, M.L., Stell, R., et al., 2003. Reversible reorganisation of the motor cortical representation of the hand in cervical dystonia. Mov. Disord. 18, 395–402. Ashburner, J., Friston, K.J., 2000. Voxel-based morphometry—the methods. NeuroImage 11, 805–821. Ashburner, J., 2007. A fast diffeomorphic image registration algorithm. NeuroImage 38, 95–113. Hayasaka, S., Phan, K.L., Liberzon, I., et al., 2004. Nonstationary cluster-size inference with random field and permutation methods. NeuroImage 22, 676–687. Mayka, M.A., Corcos, D.M., Leurgans, S.E., et al., 2006. Three-dimensional locations and boundaries of motor and premotor cortices as defined by functional brain imaging: a meta-analysis. NeuroImage 31, 1453–1474. Eickhoff, S.B., Stephan, K.E., Mohlberg, H., et al., 2005. A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data. NeuroImage 25, 1325–1335. Granert, O., Peller, M., Gaser, C., et al., 2011. Manual activity shapes structure and function in contralateral human motor hand area. NeuroImage 54, 32–41. Ceccarelli, A., Rocca, M.A., Pagani, E., et al., 2009. Cognitive learning is associated with gray matter changes in healthy human individuals: a tensor-based morphometry study. NeuroImage 48, 585–589. Draganski, B., Gaser, C., Busch, V., et al., 2004. Neuroplasticity: changes in grey matter induced by training. Nature 427, 311–312. Trachtenberg, J.T., Chen, B.E., Knott, G.W., et al., 2002. Long-term in vivo imaging of experience-dependent synaptic plasticity in adult cortex. Nature 420, 788–794. Curra, A., Trompetto, C., Abbruzzese, G., et al., 2004. Central effects of botulinum toxin type A: evidence and supposition. Mov. Disord. 19 (Suppl. 8), S60–S64. Ibanez, V., Sadato, N., Karp, B., et al., 1999. Deficient activation of the motor cortical network in patients with writer's cramp. Neurology 53, 96–105. Moore, R.D., Gallea, C., Horovitz, S.G., et al., 2012. Individuated finger control in focal hand dystonia: an fMRI study. NeuroImage 61, 823–831. Delnooz, C.C., Helmich, R.C., Toni, I., et al., 2012. Reduced parietal connectivity with a premotor writing area in writer's cramp. Mov. Disord. 27, 1425–1431.
C.C.S. Delnooz et al. / Neurobiology of Disease 73 (2015) 327–333 Delnooz, C.C., Helmich, R.C., Medendorp, W.P., et al., 2013b. Writer's cramp: increased dorsal premotor activity during intended writing. Hum. Brain Mapp. 34, 613–625. Hallett, M., 2000. Disorder of movement preparation in dystonia. Brain 123 (Pt 9), 1765–1766. Fiorio, M., Tinazzi, M., Ionta, S., et al., 2007. Mental rotation of body parts and noncorporeal objects in patients with idiopathic cervical dystonia. Neuropsychologia 45, 2346–2354. Taubert, M., Villringer, A., Ragert, P., 2012. Learning-related gray and white matter changes in humans: an update. Neuroscientist 18, 320–325.
333
Opavsky, R., Hlustik, P., Otruba, P., et al., 2012. Somatosensory cortical activation in cervical dystonia and its modulation with botulinum toxin: an FMRI study. Int. J. Neurosci. 122, 45–52. Ryan, N.S., Keihaninejad, S., Shakespeare, T.J., et al., 2013. Magnetic resonance imaging evidence for presymptomatic change in thalamus and caudate in familial Alzheimer's disease. Brain 136, 1399–1414. Lindholm, T.L., Botes, L., Engman, E.L., et al., 2009. Parallel imaging: is GRAPPA a useful acquisition tool for MR imaging intended for volumetric brain analysis? BMC Med. Imaging 9, 15.
