Acute ischaemic stroke alters the brain's preference for distinct dynamic connectivity states

Abstract

Acute ischaemic stroke disturbs healthy brain organization, prompting subsequent plasticity and reorganization to compensate for the loss of specialized neural tissue and function. Static resting state functional MRI studies have already furthered our understanding of cerebral reorganization by estimating stroke-induced changes in network connectivity aggregated over the duration of several minutes. In this study, we used dynamic resting state functional MRI analyses to increase temporal resolution to seconds and explore transient configurations of motor network connectivity in acute stroke. To this end, we collected resting state functional MRI data of 31 patients with acute ischaemic stroke and 17 age-matched healthy control subjects. Stroke patients presented with moderate to severe hand motor deficits. By estimating dynamic functional connectivity within a sliding window framework, we identified three distinct connectivity configurations of motor-related networks. Motor networks were organized into three regional domains, i.e. a cortical, subcortical and cerebellar domain. The dynamic connectivity patterns of stroke patients diverged from those of healthy controls depending on the severity of the initial motor impairment. Moderately affected patients (n = 18) spent significantly more time in a weakly connected configuration that was characterized by low levels of connectivity, both locally as well as between distant regions. In contrast, severely affected patients (n = 13) showed a significant preference for transitions into a spatially segregated connectivity configuration. This configuration featured particularly high levels of local connectivity within the three regional domains as well as anti-correlated connectivity between distant networks across domains. A third connectivity configuration represented an intermediate connectivity pattern compared to the preceding two, and predominantly encompassed decreased interhemispheric connectivity between cortical motor networks independent of individual deficit severity. Alterations within this third configuration thus closely resembled previously reported ones originating from static resting state functional MRI studies post-stroke. In summary, acute ischaemic stroke not only prompted changes in connectivity between distinct networks, but it also caused characteristic changes in temporal properties of large-scale network interactions depending on the severity of the individual deficit. These findings offer new vistas on the dynamic neural mechanisms underlying acute neurological symptoms, cortical reorganization and treatment effects in stroke patients.

Keywords: dynamic functional network connectivity; functional integration; functional segregation; hand motor deficits; sliding window analysis.

Figure 1

Intrinsic connectivity networks and between-network analysis. (A) Spatial maps of the 13 independent components computed for the entire sample (31 ischaemic stroke patients and 17 healthy controls). These maps were organized in three domains: sensorimotor (SMN, eight components, framed in dark blue), subcortical (SC, three components, framed in light blue) and cerebellar (CB, two components, framed in orange). Components were back-reconstructed based on the independent components of the cortical and subcortical sensorimotor as well as cerebellar domains presented in Allen et al. (2014). (B) Static functional network connectivity between independent components resulting in a total of 78 connectivity pairs. Connectivity values correspond to the Fisher’s Z-transformed Pearson correlation, averaged over the entire group of healthy controls and ischaemic stroke patients. Red colour indicates positive correlation, blue colour indicates negative correlation. Thus, the connectivity matrix illustrates high positive intra-domain connectivity within the sensorimotor, subcortical and cerebellar domains as well as neutral to negative inter-domain connectivity: SMN-SC, SMN-CB and SC-CB. Asterisks indicate significantly altered static connectivity between the three subgroups: healthy controls, moderately, and severely affected stroke patients (one-way ANOVA, P < 0.05). The components ‘left ipsilesional sensorimotor’ and ‘bilateral postcentral gyri’ were both characterized by the highest number of disturbed static connectivity pairs (six each). L = left; R = right; SMA = supplementary motor area.

Figure 2

Overlap maps of DWI lesions (A–D) and within-network connectivity (E). (A) Entire sample of stroke patients. The majority of lesions were located subcortically, with the maximum overlap (n = 18/31) being in the posterior limb of the internal capsule. (B) Illustration of the lesion overlap with the corticospinal tract (CST, in red, median overlap for the entire stroke patient sample: 6.8%). (C) Subgroup of moderately affected stroke patients (n = 18). (D) Subgroup of severely affected stroke patients (n = 13). Both stroke patient subgroups primarily presented with subcortical lesions. Importantly, while subgroups were defined based on their motor function (cut-off: ARAT 28/29), they neither differed in lesion volume, nor corticospinal tract overlap (see Supplementary Table 1). (E) Moderately and severely affected stroke patients exhibited significantly reduced within-network connectivity in comparison to healthy controls. Contrasting the patient subgroups resulted in a small cluster of voxels with reduced signal intensity in case of moderate symptoms (P < 0.05, FDR-corrected for multiple comparisons; left: moderately affected versus healthy controls; middle: severely affected versus healthy controls; right: moderately versus severely affected patients).

Figure 3

Circle plots of significant static functional connectivity differences between the subgroups (post hoc t-tests, P < 0.05, FDR-corrected for multiple comparisons). (A) Moderately affected stroke patients versus healthy controls. Connectivity strength in stroke patients was found to be decreased between the pre- and postcentral areas and between the supplementary motor area and bilateral cerebellar components as well as the left, ipsilesional precentral gyrus and the right cerebellum, yet comparably increased between the ipsilesional precentral gyrus and putamen. (B) Severely affected stroke patients versus healthy controls. In contrast to the previous group comparison, only two significantly altered connectivity pairs emerged: the connectivity between both of the putamen components and the more anterior putamen component with the thalamic component were increased in stroke patients. Note that there were no significant connectivity differences between moderately and severely affected stroke patients. SMA = supplementary motor area.

Figure 4

Dynamic functional network connectivity analysis. (A) Three resulting connectivity states as well as their across-group frequencies. The first state was characterized by a highly positive intra-domain connectivity in all of the domains [sensorimotor (SMN), subcortical (SC), cerebellar, (CB)] and highly negative inter-domain connectivity. It was the state resembling the static connectivity matrix the most, measured in Manhattan distance. The second and most frequent state featured comparably lower connectivity within the sensorimotor domain, particularly between the ventral precentral component and the paracentral lobule to further sensorimotor components. Inter-domain connectivity was mostly neutral. The third state comprised positive intra-domain connectivity, negative inter-domain connectivity between both the sensorimotor and subcortical domains to the cerebellar domain and no connectivity between the sensorimotor and subcortical domains. (B) Elbow criterion. Trajectory of the cluster validity index with increasing numbers of clusters, i.e. k in k-means clustering (top) and cluster distributions for a given number of clusters (bottom). The cluster validity index was computed as the ratio between the within-cluster distance to between-cluster distance. As the steepness of the decline in the clustering validity index decreased markedly after three and four clusters, yet the four-cluster solution included a state with a frequency of <10%, k = 3 combined the lowest cluster validity index and most well-balanced solution. This choice was additionally enforced by the highest silhouette measure for k = 3 (see Supplementary material). (C) Connectivity states separately for each of the three subgroups. Please note that some subjects only entered one or two of the defined three connectivity states, resulting in varying numbers of subjects within a specific state (cf., stated absolute numbers of subjects entering the state as well as the percentage of the entire subgroup). Connectivity state frequencies did not differ significantly between subgroups.

Figure 5

DFNC analysis. Fraction and dwell times as well as the number of transitions for the three groups: healthy controls, moderately, and severely affected patients. (A) Fraction times. Over the entire scan duration, moderately affected stroke patients spent significantly more time in State 2 than healthy control subjects. Higher fraction times in State 2 in the case of moderate symptoms compared to severe symptoms were significant only at uncorrected thresholds (P = 0.047, asterisk in parentheses). State 2, the generally most frequent connectivity state, was characterized by comparable low positive intra-sensorimotor domain connectivity. (B) Dwell times. Once again, moderately affected stroke subjects differed from healthy controls and spent significantly more time in State 2 at any one time. (C) Number of transitions. The absolute number of transitions did not differ significantly between the three groups. Subjects switched between states five to seven times on average. Asterisks indicate statistically significant group differences based on significant post hoc t-tests (P < 0.05, FDR-corrected for multiple comparisons).

Figure 6

Transition matrices displaying the differences in likelihood of changing from one connectivity state to another between the subgroups. In general, subjects tended to stay in their current connectivity state. Thus, if they were in State 1 at any one time t, they would most likely be in the same connectivity state at t + 1, i.e. the next window. The same was true for States 2 and 3 (likelihood of remaining in the same state: range 87–97%). Transitions from any one state to one of the other two states were less likely (likelihood range 1–10%). There were statistically significant group differences for transitions from State 2 to itself as well as to State 1 (one-way ANOVAs, P < 0.05). Moderately affected stroke patients had a significantly higher likelihood of staying in State 2 than healthy controls and severely affected stroke patients (post hoc t-tests, P < 0.05, FDR-corrected for multiple comparisons). In contrast, severely affected stroke patients had a higher likelihood of not staying in State 2, but switching from this state to State 1 (post hoc t-tests, P < 0.05, FDR-corrected for multiple comparisons). Therefore, they preferably transitioned to the regionally densely connected State 1 and left the weakly connected State 2.

Figure 7

Differences in dynamic functional connectivity between the various subgroups. Subgroup-specific connectivity matrices as well as numbers and percentages of subjects within a group entering this state (left) and circle plots of significant functional dynamic connectivity differences (right, post hoc t-tests, P < 0.05, FDR-corrected for multiple comparisons. Significant differences are also marked by an asterisk in the connectivity matrices). (A and B) Differences between moderately affected stroke patients and healthy controls. Moderately affected stroke patients presented higher connectivity strength between the ipsilesional sensorimotor area and the more posterior putamen component, yet lower connectivity between a bilateral, more ventral precentral component and the same posterior putamen component in State 2 (A). Furthermore, connectivity differences in State 3 resembled those in the static functional connectivity analysis. Stroke patients featured significantly lower connectivity between the ipsi- and contralesional sensorimotor areas as well as lower connectivity between the ipsilesional sensorimotor area and the paracentral lobule and supplementary motor area (B). (C) Differences between healthy controls and severely affected patients in State 3. Stroke patients presented with a decreased connectivity between the contralesional, right sensorimotor and paracentral areas, while connectivity between the more anterior putamen component and both cerebellar components was increased. (D) Differences between moderately and severely affected stroke patients in State 3. Several connectivity pairs were reduced in the case of moderately affected stroke patients. These pairs were: ipsilesional, left sensorimotor cortex–putamen, paracentral cortex–putamen and the more anterior putamen component–left cerebellum. SMA = supplementary motor area.