Grey matter abnormalities in methcathinone abusers with a Parkinsonian syndrome

Abstract Background A permanent Parkinsonian syndrome occurs in intravenous abusers of the designer psychostimulant methcathinone (ephedrone). It is attributed to deposition of contaminant manganese, as reflected by characteristic globus pallidus hyperintensity on T1‐weighted MRI. Methods We have investigated brain structure and function in methcathinone abusers (n = 12) compared to matched control subjects (n = 12) using T1‐weighted structural and resting‐state functional MRI. Results Segmentation analysis revealed significant (p < .05) subcortical grey matter atrophy in methcathinone abusers within putamen and thalamus bilaterally, and the left caudate nucleus. The volume of the caudate nuclei correlated inversely with duration of methcathinone abuse. Voxel‐based morphometry showed patients to have significant grey matter loss (p < .05) bilaterally in the putamina and caudate nucleus. Surface‐based analysis demonstrated nine clusters of cerebral cortical thinning in methcathinone abusers, with relative sparing of prefrontal, parieto‐occipital, and temporal regions. Resting‐state functional MRI analysis showed increased functional connectivity within the motor network of patients (p < .05), particularly within the right primary motor cortex. Conclusion Taken together, these results suggest that the manganese exposure associated with prolonged methcathinone abuse results in widespread structural and functional changes affecting both subcortical and cortical grey matter and their connections. Underlying the distinctive movement disorder caused by methcathinone abuse, there is a more widespread pattern of brain involvement than is evident from the hyperintensity restricted to the basal ganglia as shown by T1‐weighted structural MRI.


| INTRODUCTION
Due to the ease of synthesis from readily available components methcathinone (ephedrone) still presents a continuing public health hazard in many Eastern European countries. Intravenous abuse of this designer psychostimulant results in a levodopa-unresponsive Parkinsonian syndrome (Stepens et al., 2014). Home preparation of methcathinone by oxidation of the pseudoephedrine and ephedrine in common cold medications, in the presence of potassium permanganate, results in a high overload of manganese. The movement disorder closely resembles that of chronic manganism, involving combinations of hypokinesia, dysarthria, dystonia, and postural instability which manifest as facial impassivity, slowed movements, low volume speech, micrographia, lurching gait with absent arm swing, and falls (Stepens et al., 2008). This methcathinone/manganese-associated movement disorder can be severely disabling, is permanent, and shows no improvement despite abstinence (Selikhova et al., 2008;Sikk et al., 2007;Stepens et al., 2008).
In active users of methcathinone, manganese blood levels are markedly elevated and associated with a distinctive T1-weighted MRI signal hyperintensity in the basal ganglia (Selikhova et al., 2008;Sikk et al., 2007;Stepens et al., 2008), which is attributable to manganese deposition (Klos et al., 2006). Diffusion-weighted MR tractography shows abusers to have diffuse white matter abnormalities throughout the brain, with severe focal damage in tracts underlying the right ventral premotor cortex and the medial prefrontal cortex. These findings point to a widespread neuropathology underlying this disorder of higher level motor programming (Stepens et al., 2010). So far no studies have investigated grey matter pathology either within the basal ganglia or elsewhere in the brain. This MRI study investigates subcortical and cortical grey matter integrity in intravenous methcathinone abusers.

| Image acquisition
All MRI data were acquired using a MR scanner 3.0T Achieva (Philips Medical Systems) at Tartu University Hospital. Whole-brain T1weighted scans were acquired using a 3D Fast Field Echo sequence with 146 axial slices (TR = 12 ms, TE = 3.2 ms, 1.0 mm isotropic resolution).
Resting-state fMRI data were acquired with dynamic T2*-weighted gradient echo single shot EPI-BOLD sequence (TR = 3,000 ms, TE = 35 ms, acquisition voxel size 3.0 mm isotropic). Sixty volumes were acquired with 47 axial slices.
Furthermore, for each subject, brain tissue volume, normalized for subject head size, was estimated with SIENAX , part of FSL. SIENAX starts by extracting brain and skull images from the single whole-head input data. The brain image is then affine registered to MNI152 space (using the skull image to determine the registration scaling); this is primarily in order to obtain the volumetric scaling factor, to be used for normalization for head size. Tissue-type segmentation with partial volume estimation is carried out in order to calculate total volume of brain tissue (including separate estimates of volumes of total grey matter, white matter, cortical grey matter, and ventricular CSF).
The results of each step of the image processing, most importantly the subcortical segmentation, were carefully examined to ensure accuracy of the results. As expected, the signal hyperintensities in the left and right pallidum prevented correct automated (as well as reliable manual) segmentation in the sub-group of patients who were active users; the pallidum was therefore excluded from volumetric analysis.
Before conducting statistical analyses, the volumes of each subcortical region of interest were adjusted for inter individual head size differences via multiplication by the volumetric scaling factor derived from SIENAX. All statistical analyses were carried out using IBM SPSS Statistics (Version 20). Statistical comparisons were carried out separately for each hemisphere using independent samples t-tests. Pearson correlation coefficients were calculated to investigate the relationship between adjusted subcortical volumes and clinical rating scale scores.

| Voxel-based morphometry
T1-weighted MPRAGE data were analyzed using a standard approach with FSL voxel-based morphometry (VBM; part of FSL), a voxelbased morphometry style analysis (Ashburner & Friston, 2000;Good et al., 2001). First, structural images were brain extracted (Smith, 2002). Next, tissue-type segmentation was carried out using FAST4 (Zhang, Brady, & Smith, 2001). The resulting grey matter partial volume images were then aligned to MNI152 standard space using the affine registration tool FLIRT (Jenkinson, Bannister, Brady, & Smith, 2002;Jenkinson & Smith, 2001), followed by nonlinear registration using FNIRT (Anderson, Andersson, Jenkinson, & Smith, 2007). The resulting images were averaged to create a study-specific template, to which the native grey matter images were then nonlinearly reregistered. We then multiplied the registered partial volume images of all subjects by the Jacobian of the warp field ("modulation") to correct for local expansion or contraction. The modulated segmented images were smoothed with an isotropic Gaussian kernel with a sigma of 3 mm.
To investigate group differences, a voxel-wise GLM was applied using permutation-based nonparametric testing with correction for multiple comparisons (family-wise error [FWE]; Smith & Nichols, 2008).

| Cerebral cortical thickness analysis
Cortical reconstruction and volumetric segmentation were performed via a semiautomated process using the FreeSurfer image analysis suite. This processing includes removal of nonbrain tissue using a hybrid watershed/surface deformation procedure (Segonne, Pacheco, & Fischl, 2007), automated transformation to Talairach space, segmentation of the subcortical white matter and deep grey matter volumetric structures , intensity normalization (Sled, Zijdenbos, & Evans, 1998), tessellation of the grey matter/ white matter boundary, automated topology correction (Fischl, Liu, & Dale, 2001;Segonne et al., 2007), and surface deformation following intensity gradients to demarcate the grey matter/white matter and grey matter/CSF borders at the location where the greatest shift in intensity defines the transition to the other tissue class (Dale et al., 1999;Fischl & Dale, 2000). For two subjects, manual adjustment of watershed parameters for the skull stripping procedure was required; four subjects needed additional control points for complete tracing of brain tissue; and for one subject, tracing of the pial surface had to be corrected. Finally, completed cortical models were resampled into a common space (Fischl & Dale, 2000). Computations were carried out in the High Performance Computing Center of the University of Tartu.
Anatomical landmarks encompassed by clusters with statistically significant changes were labeled according to a complete parcellation of the cortical surface that uses internationally accepted standard nomenclature and criteria (Destrieux, Fischl, Dale, & Halgren, 2010).
Graphical reconstructions were created using PySurfer.
Cortical thickness was compared firstly between patients and controls and then between active and abstinent users using mri_glmfit, part of the FreeSurfer toolkit. Inference was performed using permutation testing and the maximum cluster size as the test statistic, allowing correction for multiple comparisons across surface vertices. A z-threshold corresponding to p < .05 was chosen for both the clusterforming threshold and cluster-wise significance threshold.

| Resting-state functional MRI analysis
Resting-state fMRI analysis was carried out using the Multivariate

Exploratory Linear Optimised Decomposition into Independent
Components tool (Beckmann, DeLuca, Devlin, & Smith, 2005), part of FSL. First, standard preprocessing steps were performed on each individual subject's fMRI images, consisting of motion correction, brain extraction, and spatial smoothing using a Gaussian kernel of full-width at half-maximum of 6 mm, and high-pass temporal filtering equivalent to 150 s. Next, fMRI images were registered to the individual's structural images using boundary-based linear registration and then to standard MNI space images using FNIRT (Anderson et al., 2007).
Preprocessed functional data containing 60 time points for each subject were temporally concatenated across subjects to create a single 4D dataset. Concatenated fMRI multiple datasets were decomposed using independent component analysis (ICA) to give 25 spatial components. Subsequently, components corresponding to the eight canonical resting-state networks (RSN) were selected (Beckmann et al., 2005). A dual-regression approach was used to identify, within each subject's fMRI dataset, subject-specific temporal dynamics, and associated spatial maps. This involved spatial regression of group-ICA spatial maps against individual fMRI datasets, resulting in matrices describing temporal dynamics for each component and subject, followed by temporal regression of time-course matrices against fMRI data to estimate subject-specific spatial maps. Finally, different component maps (corresponding to a specific RSN) in each subject were concatenated into single 4D files.
To investigate group differences, a voxel-wise GLM was applied using permutation-based nonparametric testing, with FWE correction for multiple comparisons (Smith & Nichols, 2008). We specifically studied the motor network, using the default-mode network (DMN) as an anatomical control. The DMN is distinct from the motor network and there is no evidence of its involvement in this clinical syndrome.
In order to test for altered strength of functional coupling between the deep grey structures, functional connectivity was investigated based on regional correlations of the BOLD signal time courses.
The representative time course of activity for each structure (pallidum, putamen, caudate, and thalamus in MNI space) was extracted.
Functional connectivity was calculated using Pearson correlation coefficients between different structures, converted into Z statistics using Fisher's r-to-Z transformation, and these resulting statistics were used to test for group differences between patients and controls using a Repeated Measures ANOVA.

| Patient characteristics
Clinical rating scale results are summarized in Table 1

| Voxel-based morphometry
Whole-brain VBM analysis (excluding the globus pallidus) revealed areas of significant grey matter loss in patients compared to controls in the left and right putamen and caudate, as well as in the deep left temporal lobe ( Fig. 1C and D). No correlations with duration of methcathinone use were demonstrated.

| Cerebral cortical thickness
When compared to controls, patients displayed cortical thinning in a total of nine clusters (Fig. 2, Table 2). Notably, given the preserved MMSE scores, large areas of prefrontal, parietal, and temporal cortex were relatively spared. There was no significant difference in the duration of methcathinone usage between active and abstinent users.

| Resting-state functional connectivity
We investigated functional connectivity within the motor network while the patients were at rest. Voxel-wise comparison revealed several regions with significantly increased coactivation with the motor RSNs in patients compared to controls (p < .05), mainly within the right primary motor cortex (Fig. 3)

| DISCUSSION
This is the first comprehensive study examining structural integrity and functional connectivity of cortical and subcortical grey matter structures in intravenous methcathinone abusers. Subcortical structure volumetry and whole-brain voxel-based grey matter morphometry both showed significant striatal differences bilaterally between patients and controls. Furthermore, volumetric comparison revealed reduced thalamic volume in methcathinone abusers. In addition to these changes affecting subcortical structures, surface-based analysis demonstrated widespread areas of cerebral cortical thinning in patients. Resting-state functional MRI analysis revealed significantly increased functional connectivity at rest within the motor network in methcathinone abusers, particularly within the primary motor cortices bilaterally.
The observed pattern of subcortical grey matter loss is most likely to reflect the manganese toxicity associated with intravenous Damage to these basal ganglia structures is in keeping with the patients' Parkinsonian clinical features. In particular, damage to the putamen, the most consistent site of grey matter loss in our study, is associated with motor symptomatology (Middleton & Strick, 2000). Unfortunately, the MRI T1-weighted signal hyperintensity attributable to manganese deposition within the globi pallidi precluded reliable segmentation within the patient group, preventing assessment of morphometric group differences in this structure.
Notably, our previous study of methcathinone abusers demonstrated increased mean diffusivity in the globus pallidus and a decrease in fractional anisotropy within globus pallidus-cortical connections (Stepens et al., 2010).
Functional coupling between brain regions, in part reflecting underlying anatomical connectivity, can be measured using restingstate functional connectivity. The cause of the increased functional This pattern of structural atrophy associated with increased functional connectivity has been described in a number of motor disorders, including Parkinson's disease (Poston & Eidelberg, 2012) and amyotrophic lateral sclerosis (Douaud, Filippini, Knight, Talbot, & Turner, 2011). This increase in functional connectivity in conjunction with structural atrophy may reflect either a compensatory adaptation response to injury, or a loss of inhibitory neuronal influences resulting from structural damage (Douaud et al., 2011).
Does methcathinone itself, in addition to the manganese toxicity, contribute to the observed brain changes, as well as the clinical  Gygi, Gibb, & Hanson, 1996). Human imaging studies showed normal dopamine levels and intact presynaptic nigrostriatal neurons in cases of methcathinone/manganese-induced movement disorder (Sikk et al., 2010); suggesting manganese rather than methcathinone as the main causative agent of movement disorder in our patients. Methcathinone is an "amphetamine-like" drug and methcathinone and amphetamine have similar chemical structures (Glennon, Yousif, Naiman, & Kalix, 1987). The most consistently observed change associated with chronic amphetamine abuse is reduced cerebral cortical grey matter density and volume, along with enlargement of the globus pallidus and putamen (Berman, O'Neill, Fears, Bartzokis, & London, 2008). It seems that the structural changes caused by chronic amphetamine abuse mainly affect the cortex, whereas manganese toxicity predominantly produces subcortical damage. It may be, therefore, that intravenous methcathinone abuse is a "double toxin" with manganese toxicity affecting the subcortical structures and the methcathinone component leading to the widespread changes in cerebral cortical thickness. It is worth noting, however, that a similar pattern of widespread cortical thinning has been seen in progressive supranuclear palsy (Worker et al., 2014), a disease that shares many clinical features with those seen in our cohort of methcathinone abusers. This finding would suggest that the cortical thinning observed in our patients may be a result of neurodegeneration secondary to the subcortical damage caused by manganese toxicity rather than the direct effect of methcathinone per se, although a histopathological study would be required to confirm or refute these hypotheses.

| CONCLUSION
We have demonstrated widespread atrophy and increased functional connectivity within the brains of methcathinone abusers with a manganese-induced movement disorder. These abnormalities occurred both within and outside the basal ganglia. The damage localized to subcortical structures within the motor loop of the basal ganglia is likely to be attributable to manganese toxicity and probably underlies the distinctive nature of this stereotyped motor syndrome.
This notion is supported by the resting-state fMRI analysis revealing increased functional connectivity within the motor network. Our identification of diffuse cerebral cortical thinning may reflect an additional toxic effect of methcathinone, by analogy with the effects of amphetamine, a similar stimulant. Overall our results illustrate that in intravenous methcathinone/manganese abusers there is a more widespread pattern of grey matter damage than demonstrable by conventional imaging techniques.

ACKNOWLEDGMENTS
The study was supported by Grants GARNR9199 and GARLA0148P of