A multimodal analysis of magnetic resonance imaging for the study of brain abnormalities in migrainegray matter morphometry, white matter integrity and structural connectivity

  1. Planchuelo Gómez, Álvaro
Dirigée par:
  1. Santiago Aja Fernández Directeur
  2. Rodrigo de Luis García Directeur

Université de défendre: Universidad de Valladolid

Fecha de defensa: 01 juin 2021

Jury:
  1. Silvia De Santis President
  2. Juan Pablo Casaseca de la Higuera Secrétaire
  3. Julio Pascual Gómez Rapporteur
Département:
  1. Teoría de la Señal y Comunicaciones e Ingeniería Telemática

Type: Thèses

Résumé

Migraine is a primary disease characterized by recurrent headache attacks. Despite the high migraine prevalence and its consequences, currently there are no migraine biomarkers and the diagnosis is exclusively based on the description of the symptoms by the patient. Furthermore, the migraine pathophysiology is not completely understood yet. In order to find a migraine biomarker and better understand the migraine pathophysiology, Magnetic Resonance Imaging (MRI) has been employed thanks to its excellent tissue contrast and spatial resolution using non-ionizing radiation. Multiple studies that have assessed gray matter and white matter structure in patients with migraine have shown conflicting results, although some patterns such as loss of gray matter volume have been widely described. From the clinical point of view, other important assessments like the comparison between the two current main migraine types, Chronic Migraine (CM) and Episodic Migraine (EM), have been barely carried out. Considering the technical perspective, specific evaluations of the structural connections between gray matter regions and the relationships between the MRI findings from different modalities have not been performed. OBJECTIVES The main objective was the characterization of gray matter and white matter structural properties of patients with CM and EM. With regard to both migraine groups, the comparison between the two migraine types through the employment of diverse MRI processing techniques is also included in the main objective. The use of advanced and novel diffusion measures not employed previously in the migraine literature was also considered to provide an additional strategy for the assessment of white matter. An evaluation of the diffusion MRI (dMRI) acquisition parameters in association with the sample size was carried out to identify possible sources of the variability. Moreover, the structural connections between gray matter regions through the white matter tracts were assessed, bearing in mind their possible relationship with gray matter morphometry. MATERIALS AND METHODS On the one hand, the analysis of the migraine patients was performed using methods previously employed in the literature. These methods include morphometry measures for gray matter and diffusion tensor imaging (DTI) parameters in combination with tract-based spatial statistics (TBSS) for white matter. On the other hand, more advanced techniques were implemented to assess migraine patients. These techniques cover methods employed in previous studies, such as tractography and connectomics. Moreover, novel methods recently developed such as Apparent Measures Using Reduced Acquisitions (AMURA) to evaluate diffusion properties were applied. To analyze the relationship between the changes identified in the diverse MRI modalities, the fusion method multimodal Canonical Correlation Analysis followed by joint Independent Component Analysis (mCCA-jICA) was used. This method including was adapted from previous proposals for the assessment of morphometry and connectomics variables. RESULTS The sample of the diverse assessments with dMRI data included 56 patients with CM, 54 patients with EM and 50 healthy controls, and 57 and 52 subjects in both migraine groups and the control group, respectively, in the analysis of gray matter morphometry. In the morphometry comparisons, higher cortical curvature values (seven regions) and lower cortical thickness (nine regions), gray matter volume (15 regions) and surface area (25 regions) values were found in migraine patients (one or both groups) compared to controls. These gray matter results are statistically significant with p < 0.05 after False Discovery Rate correction for multiple comparisons and posthoc tests. Regarding the white matter analysis with the DTI descriptors, different trends were obtained in CM with respect to controls. Higher (10 regions, including duration of migraine, aura and medication overuse headache as covariates) and lower fractional anisotropy (other 12 regions, including time from onset of CM as covariate), and lower radial diffusivity (14 regions, adjusted by three covariates) were found in CM compared to controls. Higher axial diffusivity in EM compared to controls was identified in eight regions (adjusted by three covariates). With AMURA, additional differences between EM and controls that could not be identified with conventional DTI parameters were found. Lower return-to-origin probability values were observed in 24 regions in EM compared to controls (no covariates). Additionally, it was observed that lower sample size may be counterbalanced with higher number of diffusion orientations in TBSS studies. The TBSS results are statistically significant with p < 0.05 after Family Wise Error or False Discovery Rate correction for multiple comparisons. With regard to connectomics, two different trends were found. The first trend was composed of debilitated structural connections between regions within the diverse lobes (seven connections with lower number of streamlines in patients), and the second trend included strengthened connections with pain processing regions (10 connections with higher number of streamlines in patients). These results were recognized using the connectomics analysis and the mCCA-jICA approach, with specific results provided by each method. The changes of structural connectivity were associated with cortical curvature alterations. These structural connectivity results are statistically significant with p < 0.05 after False Discovery Rate correction for multiple comparisons and post-hoc tests. Diverse statistically significant differences were detected between CM and EM. With respect to gray matter morphometry, surface area differences between both migraine groups in 28 regions were identified, with lower values in CM. In the white matter analysis, lower axial diffusivity values in CM compared to EM were found in 38 regions, obtaining similar results with AMURA and other DTI parameters. Weakened structural connectivity (lower number of streamlines) in CM was detected connections between the caudate nucleus and regions from the orbitofrontal cortex, while strengthened specific altered connectivity (higher number of streamlines) was found in connections with the hippocampus in CM patients. The statistical significance is the same with respect to the comparisons between migraine groups and controls. CONCLUSIONS Migraine (CM and EM) is characterized by a series of structural gray and white matter changes, including connections between specific gray matter regions. Considering the comparison between both migraine groups, CM follows a set of alterations different from the changes between EM and controls. The results suggest that CM is an entity with unique properties and substantially different from EM. The surface area could work as a CM biomarker. The findings demonstrated that the DTI model may be insufficient to provide a complete description of the white matter state. AMURA can be employed to complement the results of the classical DTI approach, even in suboptimal conditions for the extraction of advanced diffusion measures. The use of sophisticated fusion approaches such as mCCA-jICA allows the identification of the relationship between the changes observed in diverse MRI modalities and also the identification of additional alterations. Future research should be focused on the analysis of possible migraine subgroups bearing in mind the migraine heterogeneity, longitudinal studies to assess the evolution of CM and EM, and the analysis of the relationships between structural, functional and non-MRI modalities.