Degrees of functional connectome abnormality in disorders of consciousness

Abstract

Understanding the neuronal basis of disorders of consciousness can help improve the accuracy of their diagnosis, indicate potential targets for therapeutic interventions, and provide insights into the organization of normal conscious information processing. Measurements of brain activity have been used to find associations of the levels of consciousness with brain complexity, topological features of functional connectomes, and disruption of resting-state networks. However, obtainment of a detailed picture of activity patterns underlying the vegetative state/unresponsive wakefulness syndrome and the minimally conscious state remains a work in progress. We here aimed at finding the aspects of fMRI-based functional connectivity that differentiate these states from each other and from the normal condition. A group of 22 patients was studied (9 minimally conscious state and 13 vegetative state/unresponsive wakefulness syndrome). Patients were shown to have reduced connectivity in most resting-state networks and disrupted patterns of relative connection strengths as compared to healthy subjects. Differences between the unresponsive wakefulness syndrome and the minimally conscious state were found in the patterns formed by a relatively small number of strongest positive correlations selected by thresholding. These differences were captured by measures of functional connectivity disruption that integrate area-specific abnormalities over the whole brain. The results suggest that the strong positive correlations between the functional activities of specific brain areas observed in healthy individuals may be critical for consciousness and be an important target of disruption in disorders of consciousness.

Keywords: consciousness; functional connectivity; magnetic resonance imaging; minimally conscious state; unresponsive wakefulness syndrome.

Figure 1

Scheme of computation of the ICI index measuring the similarity of a patient's connectome to the mean connectome of the reference group of healthy subjects [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 2

Distributions of the index of connectome intactness in healthy controls and DOC patients. Here and below two‐tailed p values are indicated. Effect size is given in Table 2. Note that here and in Figures 3a,b and 4a, an independent healthy control group is used to obtain an unbiased estimate of a healthy individual's connectome similarity to the mean connectome of the reference group [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 3

(a) Distributions of the index of thresholded connectome intactness in healthy controls, MCS, and UWS patients. Boxes correspond to interquartile ranges, whiskers to ranges, and horizontal lines to medians. The threshold value is 1.0 (applied to Fisher‐transformed correlation coefficients). Effect sizes are given in Table 2. (b) Sensitivity to the threshold value of the difference in ITCI between MCS and UWS patients. The lines with shadings show the medians and the interquartile ranges of ITCI in each group as a function of the threshold. Black line without shading: p values of the Mann–Whitney test between the MCS and UWS groups (secondary y‐axis) [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 4

(a) Distributions of the hub disruption index in the subject groups depending on the threshold. The lines with shadings show the medians and the interquartile ranges. Black line without shading: p values of the Mann–Whitney test between the MCS and UWS groups (secondary y‐axis). (b) Distributions of the number of suprathreshold connections. Effect sizes are given in Table 2 [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 5

Distributions of mean connection strengths within resting‐state networks for DOC patients and healthy subjects (reference group). Top left: connections between areas not assigned to any network. Uncorrected p values of the Mann–Whitney test are indicated. For all the networks except the Limbic, the differences remain significant after the Holm–Bonferroni correction [Color figure can be viewed at http://wileyonlinelibrary.com]