Lesions causing post-stroke spasticity localize to a common brain network

Abstract

Objective: The efficacy of clinical interventions for post-stroke spasticity (PSS) has been consistently unsatisfactory, probably because lesions causing PSS may occur at different locations in the brain, leaving the neuroanatomical substrates of spasticity unclear. Here, we investigated whether heterogeneous lesions causing PSS were localized to a common brain network and then identified the key nodes in this network.

Methods: We used 32 cases of PSS and the Human Connectome dataset (n = 1,000), using a lesion network mapping method to identify the brain regions that were associated with each lesion in patients with PSS. Functional connectivity maps of all lesions were overlaid to identify common connectivity. Furthermore, a split-half replication method was used to evaluate reproducibility. Then, the lesion network mapping results were compared with those of patients with post-stroke non-spastic motor dysfunction (n = 29) to assess the specificity. Next, both sensitive and specific regions associated with PSS were identified using conjunction analyses, and the correlation between these regions and PSS was further explored by correlation analysis.

Results: The lesions in all patients with PSS were located in different cortical and subcortical locations. However, at least 93% of these lesions (29/32) had functional connectivity with the bilateral putamen and globus pallidus. These connections were highly repeatable and specific, as compared to those in non-spastic patients. In addition, the functional connectivity between lesions and bilateral putamen and globus pallidus in patients with PSS was positively correlated with the degree of spasticity.

Conclusion: We identified that lesions causing PSS were localized to a common functional connectivity network defined by connectivity to the bilateral putamen and globus pallidus. This network may best cover the locations of lesions causing PSS. The putamen and globus pallidus may be potential key regions in PSS. Our findings complement previous neuroimaging studies on PSS, contributing to identifying patients with stroke at high risk for spasticity at an early stage, and may point to PSS-specific brain stimulation targets.

Keywords: functional connectivity; lesion network mapping; lesions; post-stroke spasticity; resting-state networks.

FIGURE 1

Lesion network mapping method. In step 1, lesions were manually traced to standard brain atlas (MNI 152 template). In step 2, using a resting-state functional connectome dataset (n = 1,000), the lesion location was used as a seed point for functional connectivity analysis with the whole brain. The functional connectivity map of each lesion was obtained. In step 3, functional connectivity maps of all lesions were thresholded and binarized, and then overlapped to identify the common brain regions.

FIGURE 2

Lesion locations causing PSS. The lesions of 32 patients with PSS were manually located on standard brain atlas. All lesions were tracked according to their laterality.

FIGURE 3

Lesion network mapping of PSS. (A) > 93% (30/32) of lesion locations causing PSS were functionally positively connected to the bilateral putamen, pallidum, and thalamus. Using two different statistical methods, two-sample t-test (B) and Liebermeister test (C), lesion network maps of patients with PSS were compared with those of patients with post-stroke non-spastic motor dysfunction to identify regions specific to PSS (FWE-corrected p<0.05). (D) The conjunction analyses determined bilateral putamen and pallidum as regions associated with PSS that were both sensitive and specific.

FIGURE 4

Lesion network mapping results are independent of thresholds and are highly repeatability. (A) The overlapping regions of peak lesion networks in the bilateral putamen, pallidum and thalamus were similar in different network thresholds. (B) Patients with PSS were randomly divided into two subgroups for comparison of lesion network mapping results. The lesion network overlap regions in both subgroups were found to be highly repeatability and consistent with our results above.

FIGURE 5

Defining the lesion network of PSS. Positive connectivity with our right putamen and globus pallidus ROI (red) (A) and left putamen and globus pallidus ROI (blue) (B) were identified, respectively. (C) The overlap of these two networks (purple) defines a distributed brain network, which is the lesion network of PSS.

FIGURE 6

Lesion network involves the location of lesions causing PSS. Our lesion network of PSS involves the location of lesions in 31/32 patients identified in the PSS group. Lesions of PSS (yellow) are the same as in Figure 2. The lesion network of PSS is shown in purple.

FIGURE 7

Correlation between spasticity and the putamen and globus pallidus. The functional connectivity of the right putamen and globus pallidus ROI (A), the left putamen and globus pallidus ROI (B), and the bilateral putamen and globus pallidus (C) to lesions causing PSS was positively correlated with the MAS score.