Background and Aims:

To investigate resting state (RS) functional connectivity (FC) and white matter (WM) microstructural abnormalities underlying executive function (EF) impairment in multiple sclerosis (MS).

Methods:

One-hundred and sixteen MS patients and 65 age- and sex-matched healthy controls (HC) underwent 3T brain T1-weighted, RS and diffusion-weighted sequences and Wisconsin Card Sorting Test (WCST) to test EF. The main large-scale brain RS cognitive networks were derived with independent component analysis. Mean fractional anisotropy (FA) was calculated from a priori-selected WM tracts. Associations between WCST scores and RS FC and FA abnormalities were investigated with multivariable models.

Results:

In MS patients, independent predictors of working memory/updating (WCST achieved categories) were: lower corpus callosum (CC) genu FA, lower left working-memory network (WMN) (precuneus), right WMN (middle temporal gyrus) RS FC for worse performance; lower executive control network (ECN) (superior temporal gyrus), higher default-mode network (DMN) (superior parietal lobule) and salience network (SN) (superior frontal gyrus, SFG) RS FC for better performance (R2=0.35). Predictors of attention (WCST errors) were lower CC genu FA, lower left WMN (precuneus) and DMN (anterior cingulate gyrus) RS FC for worse performance; higher left WMN (cerebellum lobule IX) and ECN (SFG) RS FC for better performance (R2=0.24). Predictors of inhibition (WCST perseverative errors/responses) were lower CC genu and superior cerebellar peduncle (SCP) FA, lower left WMN (precuneus) RS FC for worse performance; and higher ECN (SFG) RS FC for better performance (R2=0.24).

Conclusions:

CC and SCP microstructural damage and RS FC abnormalities in cognitive networks underlie EF frailty in MS.

Background and Aims:

Altered corpus callosum integrity has been associated to motor impairment in multiple sclerosis (MS), but its contribution has not been evaluated with multiparametric-MRI approaches on large populations. We investigated structural and functional inter-hemispherical substrates of global disability at different milestones and upper-limb motor impairment in a large cohort of MS patients.

Methods:

In this cross-sectional study, 340 MS patients and 130 age- and sex-matched healthy controls underwent a clinical assessment including Expanded disability status scale (EDSS) rating, 9-Hole-Peg-Test (9HPT) and electronic-finger-tapping-rate (EFTR). Structural and resting-state functional-MRI scans were used to perform: probabilistic-tractography of hand-corticospinal tracts (CSTs) and transcallosal-fibers between hand-motor cortices (hand-M1), supplementary-motor areas (SMAs) and premotor cortices (PMCs); voxel-mirror homotopic connectivity (VMHC) analysis between the same cortical regions. Random forest analyses identified MRI-predictors of clinical disability at different EDSS-milestones (3.0, 4.0, 6.0), and upper-limb motor impairment (defined as z9HPT and zEFRT scores below the 5th percentile).

Results:

Predictors of EDSS-3.0 were global atrophy and lesion measures together with damage of CSTs and PMCs and SMAs transcallosal-fibers (out-of-bag-accuracy [OOB]=0.86, p-range=<0.001-0.03). For EDSS-4.0 similar predictors were found, in addition to hand-M1 transcallosal-fibers damage (OOB-accuracy=0.90, p-range=<0.001-0.045). No MRI-predictors were identified for EDSS-6.0 milestone. Impaired upper-limb motor impairment was predicted by SMAs and PMCs transcallosal-fibers damage (OOB-accuracy-range=0.71-0.76; p-range=<0.001-0.05). No VMHC abnormalities contribute to explain clinical outcomes.

Conclusions:

In MS, transcallosal pre-motor and motor white matter-fibers abnormalities predict global disability and upper-limb motor impairment severity. A plateau-effect for higher levels of disability is present.

Funding: This study was partially supported by Fondazione Italiana Sclerosi Multipla (FISM2018/R/5).

Background and Aims:

There is evidence suggesting that both the hypothalamus and dorsal pons could be putative drivers of migraine attacks. Here, we explored whether longitudinal resting state (RS) functional connectivity (FC) changes of the pons and hypothalamus might influence migraine patients’ disease progression.

Methods:

RS functional magnetic resonance imaging (MRI) scans were acquired from 91 headache-free episodic migraine patients and 73 controls. Twenty-three patients and 23 controls underwent a clinical and MRI follow-up evaluation after 4 years. A seed-based correlation approach was used to study RS FC changes of the pons and hypothalamus, separately. The association between the hypothalamic and pontine RS FC networks was investigated using partial correlations analyses.

Results:

After 4 years, migraine patients developed an increased FC between the bilateral hypothalamus and bilateral orbitofrontal gyrus (OFG), as well as between the left pons and left cerebellum. They also experienced decreased RS FC between the right hypothalamus and the ipsilateral lingual gyrus. At baseline, the decreased hypothalamic-lingual gyrus RS FC correlated with higher migraine attack frequency. While, at follow-up, higher hypothalamic-OFG RS FC correlated with lower migraine attack frequency and higher pontine-cerebellar RS FC correlated with an increased number of migraine attacks over the years. A significant negative association between the pontine-cerebellar RS FC and the hypothalamic-lingual RS FC was found in migraine patients.

Conclusions:

Our findings support the presence of a common functional framework comprising the hypothalamic, pontine, cerebellar and visual networks that might influence migraine disease progression, as measured by changes in migraine frequency.

Background and Aims:

Resting state (RS) functional connectivity (FC) and graph theoretical analysis shed light into functional reorganization in multiple sclerosis (MS). Advanced network-based methods exploiting machine learning on RS FC may support MS patients’ classification. Here, we developed advanced methods of RS FC analysis to classify MS patients according to disease phenotype.

Methods:

RS fMRI scans were obtained from 46 healthy controls (HC) and 113 MS patients (62 relapsing-remitting [RR] and 51 progressive [P]MS). Dominant sets clustering was used to group covariance-based RS FC matrices into clusters of subjects sharing some similarities in their network configuration. Support vector machines (SVMs) were used to classify disease phenotypes exploiting a representation of networks based on their geodesic distance from cluster means. Finally, a sensitivity analysis on the trained classifier identified features relevant for classification.

Results:

The described machine learning tool was able to classify RRMS patients from HC (accuracy=72.5%), PMS patients from HC (accuracy=85.2%) and PMS from RRMS patients (accuracy=76%). The sensitivity analysis on trained SVMs found that increased connectivity within the basal ganglia sub-network and decreased RS FC within the temporal sub-network contributed to an accurate classification of both RRMS and PMS patients from HC. Moreover, decreased RS FC within the occipital and parietal sub-networks contributed to differentiate PMS patients from HC.

Conclusions:

A combination of different machine learning principles allowed to detect specific RS FC configurations of our study subjects, which allowed to classify MS patients with different clinical phenotypes from HC with a good accuracy.

Funding. Partially supported from Fondazione Italiana Sclerosi Multipla (FISM2018/R/5)