Altered structural brain network topology in chronic migraine

DOI: https://doi.org/10.1007/s00429- 019- 01994- 7

Abstract-Summary As CM is a complex disorder associated with a range of sensory, cognitive, and affective comorbidities, examining structural network disruption may provide addi- tional insights into CM symptomology beyond studies of focal brain regions.

We compared structural interconnections in patients with CM (n  =  52) and healthy controls (HC) (n  = 48) using MRI measures of cortical thickness and sub- cortical volume combined with graph theoretical network analyses.

The analysis focused on both local (nodal) and global measures of topology to

examine network integration, efficiency, centrality, and segregation.

Our results indicated that patients with CM had altered global network properties that were characterized as less integrated and efficient (lower global and local effi- ciency) and more highly segregated (higher transitivity).

Patients also demonstrated aberrant local network topology that was less inte- grated (higher path length), less central (lower closeness centrality), less efficient (lower local efficiency) and less segregated (lower clustering).

Examining structural correlations between brain areas may be a more sensitive means to detect altered brain structure and understand CM symptomology at the network level.

Extended: As CM is a complex disorder associated with a range of clinical, sen- sory, cognitive, and affective symptoms (Aurora and Brin [373]; Coppola and others [374]; Ferreira and others [375]), examining structural network disruption may pro- vide additional insights into topological patterns underlying CM symptomology beyond studies of focal brain regions.

The degree to which these affective measures (and/or other behavioral measures) are associated with alterations in network topology in CM can be the subject of future studies.

Our findings contribute to an increased understanding of structural connectivity in CM and provide a novel approach to potentially track and predict the progression of migraine disorders.

Introduction Previous studies have assessed brain structure and function using MRI to provide insight into the neurobiological correlates of migraine; however, the majority of these studies have focused on episodic migraine, which is less severe with poten- tially different underlying pathophysiological mechanisms (Sprenger and Borsook [318]; Burstein and others [376]; Chong and others [377]; Chong and others [54]; Schulte and May [378]).

Recent studies have investigated cortical and/or subcortical brain structure in individuals with CM compared to healthy controls or patients with episodic migraine

546

3 Diagnosis

(EM) (Valfrè and others [379]; Schwedt and others [380]; Lai and others [381]; Neeb and others [382]; Coppola and others [374]; Woldeamanuel and others [383]). As CM is a complex disorder associated with a range of clinical, sensory, cogni- tive, and affective symptoms (Aurora and Brin [373]; Coppola and others [374]; Ferreira and others [375]), examining structural network disruption may provide additional insights into topological patterns underlying CM symptomology beyond studies of focal brain regions.

We compare structural interconnections in patients with CM and healthy con- trols (HC) using MRI measures of cortical thickness and subcortical volume com- bined with graph theoretical network analyses.

Methods The cortical thickness and subcortical volumes of the 83 brain regions were extracted and included as nodes in structural network topology analyses using BRAPH (Brain Analysis using Graph theory; http://braph.org) software (Mijalkov and others [384]). Although not a specific aim of the current study, group differences in adjusted

cortical thickness and subcortical volume values were examined exploratorily.

To assess differences between groups in network architecture, we examined local (nodal) and global measures of centrality (connections), segregation (densely inter- connected groups allowing for specialized processing), and integration (the com- bining of specialized information from distributed brain regions) (Rubinov and Sporns [385]; Mijalkov and others [384]).

With regard to global measures, which assess group differences in overall net- work architecture, degree, path length, global and local efficiency were also exam- ined, in addition to modularity, the extent to which a graph can be divided into clearly separated communities; transitivity, the ratio of triangles and connected tri- ples in the graph; and small-worldness, which combines high levels of local cluster- ing among nodes of a network and short paths that globally link nodes of a network (Bullmore and Sporns [386]).

Results The CM and control groups were cohort matched, so while age (p = 0.61) and sex (χ2 = 3.3, p = 0.07) were not statistically significant between groups, raw cortical thickness and subcortical volume values were adjusted by these variables to control for their potential influences.

An exploratory analysis comparing adjusted cortical thickness and subcortical vol- umes between patients with CM and controls revealed no significant differences for any of the 83 brain regions included in the network analyses (p > 0.05, two-tailed).

Local path length and closeness centrality demonstrated the most widespread topological differences with patients showing primarily greater path lengths com- pared to controls, and controls demonstrating primarily greater closeness centrality compared to patients.

Significant differences in clustering were observed in two limbic structures—the left caudal anterior cingulate and the left parahippocampal gyrus, with controls showing higher clustering.

3.4 Neuroimaging

547

Discussion The majority of brain structures demonstrating higher path length, including the insula, frontal pole, cingulate cortex (rostral, posterior, isthmus), medial orbitofron- tal cortex, temporal pole, and transverse temporal gyrus, have previously been shown to contribute to high (86%) classification accuracy of patients with CM ver- sus healthy controls using cortical thickness measures (Schwedt and others [380]). Building upon this important finding, the current study demonstrated weaker

network integration of these brain regions in patients with CM.

Corroborating these findings of lower network integration in CM, the majority of brain regions with longer path length also had lower closeness centrality; a related measure described as the inverse of the average shortest path length from one node to all other nodes in the network (Sporns and others [387]; Rubinov and Sporns [385]).

Conclusions We provide evidence for altered structural network topology in patients with CM compared to healthy controls using MRI measures of gray matter combined with graph theoretical network analysis.

Our results suggest that aberrant structural brain networks in CM are more

weakly integrated, less efficient, less central, and abnormally segregated.

Compared to conventional structural MRI studies of individual brain regions, examining structural correlations between brain areas may be a more sensitive means to detect altered brain structure and can reveal insight into topological pat- terns underlying CM symptomology at the network level.

Acknowledgement A machine generated summary based on the work of DeSouza, Danielle D.; Woldeamanuel, Yohannes W.; Sanjanwala, Bharati M.; Bissell, Daniel A.; Bishop, James H.; Peretz, Addie; Cowan, Robert P. 2019 in Brain Structure and Function.

Evaluation of migraine patients with optical coherence tomography angiography