Investigating the relationship between multiple grip forces and BOLD signal in the Cerebellum and dentate nuclei of MS subjects

• ¹ UCL Institute of Neurology, Queen Square MS Centre, University College London, United Kingdom
• ² Department of Diagnostic Radiology, Faculty of Applied Medical Science, KAU, Saudi Arabia
• ³ Department of Neurosciences, Rehabilitation, Ophthalmology, Genetics and Maternal and Child Health, University of Genoa, Italy
• ⁴ Department of Brain and Behavioural Sciences, University of Pavia, Italy
• ⁵ Brain Connectivity Centre, C. Mondino National Neurological Institute, Italy
• ⁶ Wellcome Centre for Imaging Neuroscience, University College London,, United Kingdom
• ⁷ Brain MRI 3T Mondino Research Center, C. Mondino National Neurological Institute, Italy

Introduction: The cerebellum has been initially recognized as an important region in the control and coordination of movement. However, recent studies have shown that the cerebellum is also implicated in cognitive and higher-level functions (Stoodley and Schmahmann 2009; Spraker et al. 2012). Blood-oxygen-level-dependent (BOLD) signal response as measured by functional magnetic resonance imaging (fMRI) is a non-invasive technique, which allows the measurement of changes in neuronal activity; hence, reflecting changes in functional activation. Our recent work has established complex linear and nonlinear relationships between BOLD signals and applied grip force (GF) in the cortex and in the cerebellum of healthy volunteers during a complex motor task (Alahmadi et al. 2016). The complexities of the signals were also observed in the cerebellum of healthy subjects. Crucially, these relationships do not reflect pure motor monotonic behaviours and are localized not only in related motor areas but also in associative and sensory areas. How these signals are altered in pathology is unknown. Multiple sclerosis (MS) is a chronic inflammatory disease characterized by axonal loss and demyelination as well as inflammation, all affecting blood perfusion and flow, as well as the effectiveness of the motor and visual systems integration. Here, using a visually guided motor task, we aimed to characterise linear and nonlinear responses in MS to a range of applied GFs, focusing on the cerebellum and its dentate nucleus (DN), and investigate how these non-linear behaviours are altered in the presence of MS pathology. Methods: Subjects: 16 right-handed (RH) healthy volunteers (12 female; mean age 32 (±4.75) years) and 16 right-handed relapsing-remitting MS (RRMS) patients (11 female; 36 (±5.21) years; median expanded disability status (EDSS) 4, range (1.5-6.5)). MRI protocol: A 3.0 T Philips Achieva MRI system (Philips-Healthcare, Best, The Netherlands) with a 32-channel head coil was used. The imaging protocol included: 1) BOLD sensitive T2*-weighted EPI: TE/TR=35/2500ms, voxel size =3×3×2.7mm3, inter-slice gap=0.3mm, SENSE =2, 46 slices acquired with descending order, FOV=192×192mm2, 200 volumes, flip angle=90°; 2) 3D anatomical T1-weighted 1x1x1mm3 reference scan (for normalization). FMRI paradigm: A dynamic event-related power grip paradigm was designed using an MR-compatible squeezeball. The task was performed using the whole right (dominant) hand. 75 squeeze trials (3 seconds each) were divided equally into 5 GF targets (20, 30, 40, 50, and 60 % of the subject’s maximum voluntary contraction (MVC)). Thus each trial was repeated 15 times. The squeeze trials were randomised with 75 rests in counter-balanced and jittered orders. A visual cue was used for online instructions and to display gripping targets. Image pre-processing and statistical analyses: Pre-processing steps used the spatially unbiased infratentorial template (SUIT) software (Diedrichsen 2006) for the cerebellum and its DN implemented within SPM12. The pre-processing included slice timing, realignment, and co-registration to T1-weighted volumes. Then, the within-subject first level analysis model included five regressors of interest comprising polynomial functions of GF (up to the 4th order), specified by the integral of GFs. Each order contains unique information about the behaviour of the BOLD signal in each voxel. The 0th order term represents the main effect of gripping compared to rest (i.e. regardless of the applied GF). The 1st order models BOLD linear effects across GFs; higher orders (2nd order and above) model complex nonlinear relationships (e.g. U-shaped captured by +2nd order or more complicated neurometric functions that can be approximated by 3rd and 4th polynomial orders). Movement parameters for each subject were added as regressors of no interest. At this level (the first-level model), t-statistics were used to test for the effects of each polynomial coefficient. Then, the SUIT cerebellar normalization procedures were performed. The normalized cerebellum functional contrast images (of each polynomial order) from each subject were smoothed (kernel =8mm). A (between-subjects) second level random effects analysis was tested for increasingly higher-order nonlinear effects with one or two sample t-tests for within- and between-group effects, respectively. Significance was corrected (FWE) at the cluster level (P<0.05). The resulting SPMs were visualized on the flattened cerebellum template using Caret (Diedrichsen and Zotow 2015). In addition, data was subsequently normalised using a lower smoothing kernel (4mm) together with the SUIT DN template to investigate the effect in the DN. Correlations with the EDSS were explored. Results: Figures 1 and 2 show activations projected onto the flattened cerebellum for HV and MS. Figures 3 and 4 show the main effect of gripping in the DN. The preliminary results of this study show: In both groups, a main effect (i.e. irrespective of GF) was detected in the anterior and posterior cerebellum. Positive 1st order linear effects were seen only in MS within the anterior cerebellum and lobule VI (medially). Areas following a positive 2nd order term were seen in both groups in the superior cerebellum (lobules V-VI bilaterally). Negative 3rd order effects were observed in lobule VI bilaterally in both groups (although with a smaller effect size in MS) and in lobule VII in HV only. The ventral part of the DN was activated (0th order only) in both groups, whereas the dorsal part was primarily activated in the healthy group. In the 0th, 1st and 2nd orders, the BOLD effects were negatively correlated with the EDSS in most of the anterior cerebellar lobules. Conclusion: In this study, we investigated the effects of MS pathology on cerebellar (and DN) responses to a complex visuo-motor task. Comparing the main effect of gripping in MS and HV, we showed increased recruitment of the anterior lobe and lobule VI, which could be, at first glance, related to compensatory mechanisms linearly depending on applied force. However, the negative correlations with EDSS suggest a possible unsuccessful compensatory mechanism or maladaptation. +2nd and -3rd order effects were observed in MS but with lesser extent and lower effect size than HV, possibly indicating a loss of efficiency. Interestingly, the dorsal part of DN, which is more involved in motor functions (Dum and Strick 2003; Küper et al. 2011), was silent in MS compared to HV. These findings suggest a complex cerebellar reorganization of the functional anatomy of visuo-motor integration.