An investigation of the association between focal damage and global network properties in cognitively impaired and cognitively preserved patients with multiple sclerosis

• Published in: Frontiers in neuroscience 2023-02, Vol.17, p.1007580-1007580
• Main Authors: Wenger, A L, Barakovic, Muhamed, Bosticardo, Sara, Schaedelin, Sabine, Daducci, Alessandro, Schiavi, Simona, Weigel, Matthias, Rahmanzadeh, Reza, Lu, Po-Jui, Cagol, Alessandro, Kappos, Ludwig, Kuhle, Jens, Calabrese, Pasquale, Granziera, Cristina
• Format: Article
• Language: English
• Subjects: Cognitive ability; connectomics; Efficiency; Information processing; information processing speed; Lesions; Magnetic resonance imaging; Memory; Multiple sclerosis; multiple sclerosis (MS); neuropsychological test; Neuropsychology; Neuroscience; Pathology; Statistical analysis; structural connectivity; Substantia alba
• ISSN: 1662-4548; 1662-453X; 1662-453X
• DOI: 10.3389/fnins.2023.1007580

The presence of focal cortical and white matter damage in patients with multiple sclerosis (pwMS) might lead to specific alterations in brain networks that are associated with cognitive impairment. We applied microstructure-weighted connectomes to investigate (i) the relationship between global network metrics and information processing speed in pwMS, and (ii) whether the disruption provoked by focal lesions on global network metrics is associated to patients' information processing speed. Sixty-eight pwMS and 92 healthy controls (HC) underwent neuropsychological examination and 3T brain MRI including multishell diffusion (dMRI), 3D FLAIR, and MP2RAGE. Whole-brain deterministic tractography and connectometry were performed on dMRI. Connectomes were obtained using the Spherical Mean Technique and were weighted for the intracellular fraction. We identified white matter lesions and cortical lesions on 3D FLAIR and MP2RAGE images, respectively. PwMS were subdivided into cognitively preserved (CPMS) and cognitively impaired (CIMS) using the Symbol Digit Modalities Test (SDMT) z-score at cut-off value of -1.5 standard deviations. Statistical analyses were performed using robust linear models with age, gender, and years of education as covariates, followed by correction for multiple testing. Out of 68 pwMS, 18 were CIMS and 50 were CPMS. We found significant changes in all global network metrics in pwMS vs HC ( < 0.05), except for modularity. All global network metrics were positively correlated with SDMT, except for modularity which showed an inverse correlation. Cortical, leukocortical, and periventricular lesion volumes significantly influenced the relationship between (i) network density and information processing speed and (ii) modularity and information processing speed in pwMS. Interestingly, this was not the case, when an exploratory analysis was performed in the subgroup of CIMS patients. Our study showed that cortical (especially leukocortical) and periventricular lesions affect the relationship between global network metrics and information processing speed in pwMS. Our data also suggest that in CIMS patients increased focal cortical and periventricular damage does not linearly affect the relationship between network properties and SDMT, suggesting that other mechanisms (e.g. disruption of local networks, loss of compensatory processes) might be responsible for the development of processing speed deficits.