INTRODUCTION

The different pathophysiological processes occurring in multiple scIerosis (MS) including demyelination of white matter and grey matter¹ have been proven to induce structural brain disconnection². Intriguingly, such structural disconnection is, at least at the very first stage of MS, accompanied by functional hyperconnectivity ^3,4. Such apparent contrintuitive phenomena have been recently confirmed by the means of neural computational models constrained by virtual structural damage of white matter, deep grey matter and cortical areas compatible with real MS related processes ^5,6. We aimed here at demonstrating that 'The Virtual Brain' (TVB) ⁷, a computational neural mass model fed with structural connectome derived from individual diffusion MRI can successfully generate reorganized individual resting-state functional connectomes in early MS patients with similar topologies that those obtained with real rs-fMRI data.

METHODS

Thirty-four early MS patients (24 women, median age: 30y [19-49], disease duration <3y, median EDSS: 1 [0-3], median MSFC: -0;02 [-1.37-1.46]) and 19 age-, sex- matched healthy controls (15 women, median age: 31y [21-51]) were explored using a 3T Magnetom Verio system (Siemens, Erlangen Germany) to obtain high resolution MPRAGE T₁-w MRI (voxel size: 1x1x1mm³), DTI (64 gradient directions, b:1000s/mm², voxel size: 2x2x2mm3) and rs-fMRI (300 volumes of single shot EPI, TE/TR: 27ms/3s, voxel size: 2x2x2.5mm³). Tractography and structural connectomes were obtained using the MRTrix software using the Destrieux atlas parcellisation (FSL). Resting state functional connectomes were determined by Pearson correlations between signal times courses from the same parcels used for structural connectomes. Structural connectomes were thresholded to get a sparcity of 15%, and structural connectivity was expressed for each remaining links with a weighting parameters accounting for the numbers of reconstructed fibers (N), the mean lengths of bundles (L) and the average FA of bundles (FA). This weighted connectivity matrices were used as input for the TVB platform to generate 15 min (equivalent of real recorded rs-fMRI data) of functional signals from the different nodes before convolution with the hrf in order to obtain 'BOLD' like signals. Functional connectomes of virtual functional data were reconstructed using the same procedure used for real rs-fMRI data. Optimization of conduction speeds and coupling strengths used as priors in the TVB was conducted for each subject by determining the best couple of values to get the highest correlation between the upper part of connectivity matrices obtained from simulated and real functional data. Topology of the real and simulated functional connectomes were compared by quantifying EGlob, Average ELoc, Average Betweeness Centrality and hub disruption indexes K_Degree, K _ELoc, K _BC. Finally, Abnormal parameters in MS patients were correlated with clinical scores.

RESULTS

Real rs-fMRI data showed no significant differences in terms of Eglob, Eloc and BC between early MS patients and controls. In contrast, MS patients showed significant negative hub disruption indexes (p<0.0001) for KDegree, KEloc and KBC. TVB fitting performances and optimal couples of conduction speeds and coupling strengths were not significant different between MS patients and controls (p>0.1). In addition, graph parameters obtained with real and simulated data were significantly correlated (p<0.0001) (Figure 1). Comparing simulated parameters between MS patients and controls, we obtained the same profiles as for real data: no differences in simulated Eglob, Eloc and BC (Figure 2) and significant negative values of K_Degree, K_Eloc and K_BC in patients (Figure 3). Finally, the real and simulated KDegree were both correlated with PASAT scores in MS patients (Figure 4).

DISCUSSION

Inter-individual variability of structural connectome as defined by numbers and lengths of fibers as well as FA provides sufficient information to enable TVB to generate functional connectome reflecting the subtle functional resting state reorganization in early MS patients at the individual level.

CONCLUSION

This proof of concept opens new perspectives to better predict the functional reorganization in MS patients in relation to structural tissue damage or repair at the subject level.

Acknowledgements

No acknowledgement found.

References

1. Lassmann H, van Horssen J, Mahad D. Progressive multiple sclerosis: pathology and pathogenesis. Nat Rev Neurol. 2012 Nov 5;8(11):647-56. 2. Dineen RA, Vilisaar J, Hlinka J, Bradshaw CM, Morgan PS, Constantinescu CS, Auer DP. Disconnection as a mechanism for cognitive dysfunction in multiple sclerosis. Brain. 2009 Jan;132(Pt 1):239-49. 3 Schoonheim MM, Geurts JJ, Barkhof F. The limits of functional reorganization in multiple sclerosis. Neurology. 2010 Apr 20;74(16):1246-7. 4. Faivre A, Robinet E, Guye M, Rousseau C, Maarouf A, Le Troter A, Zaaraoui W, Rico A, Crespy L, Soulier E, Confort-Gouny S, Pelletier J, Achard S, Ranjeva JP, Audoin B.Depletion of brain functional connectivity enhancement leads to disability progression in multiple sclerosis: A longitudinal resting-state fMRI study. Mult Scler. 2016 Nov;22(13):1695-1708. 5. Patel KR, Tobyne S, Porter D, Bireley JD, Smith V, Klawiter E. Structural disconnection is responsible for increased functional connectivity in multiple sclerosis.. Brain Struct Funct. 2018 Jun;223(5):2519-2526. 6. Tewarie P, Steenwijk MD, Brookes MJ, Uitdehaag BMJ, Geurts JJG, Stam CJ, Schoonheim MM. Explaining the heterogeneity of functional connectivity findings in multiple sclerosis: An empirically informed modeling study. Hum Brain Mapp. 2018 Jun;39(6):2541-2548. 7. Falcon MI, Jirsa V, Solodkin A. A new neuroinformatics approach to personalized medicine in neurology: The Virtual Brain. Curr Opin Neurol. 2016 Aug;29(4):429-36.