The FreMRI Study: Advanced MRI on Migraine Patients Treated With Fremanezumab

Fremanezumab is a monoclonal antibody targeting calcitonin C-reactive peptide (CGRP) that has proven to be effective in the treatment of episodic and chronic migraine. Due to the molecular weight of the drug, it is not supposed to cross the blood-brain barrier, acting peripherally. Patients with migraine have shown gray and white matter changes that can be evaluated by magnetic resonance imaging (MRI). In the present study, we aim to evaluate the presence of MRI changes in patients treated with Fremanezumab, 12 weeks after the treatment onset, compared with the baseline. Number of Subjects: 87 patients. Study Duration per Subject Study duration per patient will be 8 months. Study procedures: In the first visit, the screening visit, patients will receive a detailed explanation of the study and will sign informed consent form. Inclusion-exclusion criteria will be reviewed and evaluation of vital signs, general and neurological examination will be done. Patients will receive a headache diary and will be trained in its use. First MRI scan will be acquired within 0-14 days, prior to Fremanezumab injection. Patients will do four monthly additional hospital visits. During each visit, occurrence of any adverse event will be interrogated; clinical situation will be analyzed and Fremanezumab will be administered. Visits will be performed at weeks 0, 4 and 8. Last visit will be performed after 12 weeks, without Fremanezumab injection. The second MRI will be acquired at 12 ± 1 weeks after the first Fremanezumab injection. Intervention: MRI acquisition Images will be acquired during interictal periods, defined as at least 24 hours from last migraine attack. High-resolution 3D T1-weighted, diffusion-weighted and resting-state functional MRI data will be acquired using a Philips Achieva 3T MRI unit (Philips Healthcare, Best, The Netherlands) with a 32-channel head coil in the MRI facility at the Universidad de Valladolid (Valladolid, Spain). For the anatomical T1-weighted images, the following acquisition parameters will be used: Turbo Field Echo (TFE) sequence, repetition time (TR) = 8.1 ms, echo time (TE) = 3.7 ms, flip angle = 8º, 256 x 256 matrix size, 1 x 1 x 1 mm3 of spatial resolution and 170 sagittal slices covering the whole brain. Diffusion-weighted images (DWI) will be obtained using the next parameters: TR = 9000 ms, TE = 86 ms, flip angle = 90º, 61 gradient directions, two baseline volumes with opposite phase encoding direction, b-value = 1000 s/mm2, 128 x 128 matrix size, 2 x 2 x 2 mm3 of spatial resolution and 66 axial slices covering the whole brain. Resting-state functional MRI (rs-fMRI) will be acquired with the following parameters: TR = 3000 ms, TE = 30 ms, flip angle = 80º, 80 x 80 matrix size, 3 x 3 x 4 mm3 of spatial resolution, 35 axial slices covering the whole brain and 197 volumes. During this acquisition, the patient will close the eyes but remain awake. All the scans will be acquired during the same session, starting with the T1-weighted scan, followed by the diffusion-weighted scan and ending with the rs-fMRI scan. Total acquisition time for a single subject is approximately 28 minutes, divided in the following periods of time: six minutes for the T1-weighted scan, 12 minutes for the diffusion-weighted scan and 10 minutes for the rs-fMRI scan. If we consider patient preparation, obtainment of documents and informed consent form, the whole process will take about 50-70 minutes. Image processing T1-weighted morphometric parameters MRI images will be processed in order to obtain cortical curvature, cortical thickness, gray matter volume and surface area of the different gray matter regions. Grey matter volume will be obtained for all the 84 grey matter regions from the Desikan-Killiany atlas (Reuter et al., 2012). Also, cortical curvature, cortical thickness and area will be calculated for the 68 regions from the atlas that are cortical regions. DWI processing DWI will be processed to carry out two types of analysis: Tract-Based Spatial Statistics (TBSS) and structural connectomics. Prior to the beginning of both processing pipelines, diverse preprocessing procedures will be implemented on the DWI data. Diffusion-weighted images will be denoised, using "dwidenoise" tool from MRtrix (www.mrtrix.org), eddy currents, motion and B0 field inhomogeneity corrected, using "dwipreproc" tool from MRtrix, and B1 field inhomogeneity corrected, using "dwibiascorrect" tool with the "-fast" option from MRtrix. Four diffusion descriptors from Diffusion Tensor Imaging will be obtained: Fractional Anisotropy (FA), Mean Diffusivity (MD), Radial Diffusivity (RD) and Axial Diffusivity (AD). We will also be using measures non based on the diffusion tensor, like the return-to-origin probability (RTOP), which reflects cellularity and restrictions better than MD. Once the DWI data are preprocessed, a whole brain mask for each image will be generated using "dwi2mask" tool from MRtrix and, next, diffusion tensors at each voxel will be estimated using the "dtifit" tool from FSL, also obtaining FA, MD and AD maps. RD will be manually calculated by obtaining the mean of the second and the third eigenvalues, which will also be previously computed with "dtifit". RTOP will be estimated using the method called "Apparent Measures Using Reduced Acquisitions" (AMURA). Structural connectomics The analysis of structural connectivity will use the segmentation results from T1-weighted processing pipeline. Anatomically-Constrained Tractography (ACT) will be implemented. Previously, five-tissue-type (5TT) segmented images for each subject will be obtained from the T1-weighted images using the "5ttgen" tool from MRtrix to have suitable images for ACT. 5TT image and the automated parcellation from FreeSurfer will be inearly registered to the FA image using the FLIRT tool from FSL. Finally, structural connectivity matrices will be computed from the filtered tractography output and the registered cortical segmentation volumes. 84 × 84 connectivity matrices, corresponding to the 84 cortical and subcortical regions from the Desikan-Killiany atlas, will be obtained using mean FA and the number of streamlines in each connection as connectome metrics. Connectivity matrices constructed in this manner are symmetric, so only a half of each matrix will be employed for further analysis. Due to the tractography method employed, it is possible that streamlines start and end in different points belonging to the same gray matter region from the Desikan-Killiany atlas. For this reason, these connections ("self-connections") will also be included in the analysis. fMRI processing The fMRI processing pipeline will be fully implemented in the software CONN. Firstly, some preprocessing steps will be implemented. fMRI volumes will be realigned and unwarpped to estimate and correct subject motion. To identify outliers, ARtifact detection Tools (ART) based outlier detection algorithm will be implemented. Then, fMRI volumes will be directly coregistered to the corresponding structural volumes (T1-weighted images) using a rigid body transformation. T1-weighted images will be segmented to detect different types of tissue, i.e., gray matter, white matter and cerebrospinal fluid (CSF). Finally, the functional connectivity matrix for each case will be computed considering whole-brain region-to-region connections, using the 84 cortical and subcortical regions from the Desikan-Killiany atlas in each subject as regions of interest. The matrices will contain the Z-values from the Fisher's r-to-z transformation. Safety Variables Safety and side effects will be evaluated according to reported adverse events, both spontaneously by patients and systematically analyzed during study visits. Vital signs (blood pressure, pulse, temperature and respiratory rate), physical examination and 12-lead electrocardiography will be done prior to the enrollment. Presence of any adverse event or local injection-site reaction will be systematically addressed. Columbia-Suicide Severity Rating Scale will assess suicidal ideation and behavior at baseline. Statistical Analysis Parameter evolution in time and relationship with change in monthly migraine days for each of the parameters, i.e., morphometric (gray matter) parameters, diffusion descriptors (white matter), structural connectivity and resting-state functional connectivity, longitudinal evolution will be assessed. n order to analyze the structural connectivity matrices, firstly, the mean number of streamlines in each connection (cell from the connectivity matrix) will be computed for each group. Next, connections with less than 500 streamlines (group mean) will be discarded in order to exclude weak connections, for which results could be unreliable, from further analysis. In the case of functional resting-state connectivity matrices, a similar rejection of non-important connections will be selected, connections with absolute Z-value lower than 0.1 are excluded, i.e., a Pearson correlation value of r = 0.1 following the Fisher's r-to-z transformation. In the previous two cases, if the thresholds are too high or too low, i.e., almost all the connections are rejected, or almost no connections are rejected, the thresholds will be changed. A model will be implemented for each of the four types of analysis described in the document. In each of the models, the significant effect of the covariates and the predictive capability of the model using the lowest number of covariates will be considered. To compare the different models, Akaike's Information Criterion (AIC) will be used (67). In the case of very similar AIC values, the model with a lower number of regressors will be chosen. For each of the four types of analysis, the predictive model will be computed for a single regressor. Regressors with p-value equal or higher than 0.05 in the singular predictive model will not be further considered to be included in the final model; while the remaining regressors will be considered for the final model of the corresponding type of analysis. Relationship between MRI parameters evolution and clinical response A linear mixed-effect model will be prepared using the same strategy that was employed for the analysis of change in monthly migraine days and MRI parameters. To evaluate the clinical response to treatment, a dichotomic variable showing positive or negative response to treatment will be used, instead of change in monthly migraine days. Positive response to treatment is considered as 50% reduction in number of monthly migraine days. Exploratory variables analysis The final model used in the analysis of the relationship between the diverse MRI parameters and change in monthly migraine days will be used. The variable representing the change in monthly migraine days will be replaced by the change in monthly intense headache days, in monthly days of acute treatment medication. In the case of response to treatment, the same final model from the analysis of clinical response considered as 50% reduction in number of monthly migraine days will be used. Response to treatment, in this case, will be represented by reduction of 75%, 100% and 30% of monthly migraine days. Analysis of adverse events Categorical adverse events will be compared using Fisher's exact test. A p-value \< 0.05 will be considered as statistically significant.