Structural Disconnection of the Tool Use Network after Left Hemisphere Stroke Predicts Limb Apraxia Severity

Abstract

Producing a tool use gesture is a complex process drawing upon the integration of stored knowledge of tools and their associated actions with sensory-motor mechanisms supporting the planning and control of hand and arm actions. Understanding how sensory-motor systems in parietal cortex interface with semantic representations of actions and objects in the temporal lobe remains a critical issue and is hypothesized to be a key determinant of the severity of limb apraxia, a deficit in producing skilled action after left hemisphere stroke. We used voxel-based and connectome-based lesion-symptom mapping with data from 57 left hemisphere stroke participants to assess the lesion sites and structural disconnection patterns associated with poor tool use gesturing. We found that structural disconnection among the left inferior parietal lobule, lateral and ventral temporal cortices, and middle and superior frontal gyri predicted the severity of tool use gesturing performance. Control analyses demonstrated that reductions in right-hand grip strength were associated with motor system disconnection, largely bypassing regions supporting tool use gesturing. Our findings provide evidence that limb apraxia may arise, in part, from a disconnection between conceptual representations in the temporal lobe and mechanisms enabling skilled action production in the inferior parietal lobule.

Keywords: connectome-based lesion-symptom mapping; dorsal stream; limb apraxia; tool use; voxel-based lesion-symptom mapping.

Figure 1

Distribution of cortical and subcortical damage in each LCVA participant. A depiction of the 57 LCVA participants’ lesions is presented on the ch2bet template. Lesions are projected on the cortical surface but include subcortical voxels using a 12-voxel search depth. Axial slices begin at the Montreal Neurological Institute (MNI) coordinate [0,0,0] and increase in 10-mm increments superiorly. Note that Sub041’s lesion rendering includes voxels below the origin.

Figure 2

Lesion overlap and SVR-VLSM results. (A) Voxelwise lesion overlap among 57 participants. Only voxels with at least 5 lesions are included in the analysis of Tool Use Gesturing. (B) Voxels associated with reduced performance in tool use gesturing (greater hand posture errors in tool use gesturing controlling for hand posture errors in meaningless imitation; red-to-white scale). (C) Voxels associated with reduced right-hand grip strength (weaker grip strength with the right hand controlling for grip strength of the left hand; blue-to-green scale). Whole-brain results are rendered in MNI space in 5-mm increments. SVR-VLSM maps are set to a voxelwise threshold of P < 0.05 with 10 000 iteration Monte Carlo style permutation analysis; clusters are included if they are greater than or equal to 500 contiguous 1 mm³ voxels.

Figure 3

SVR-CLSM of hand posture errors in tool use gesturing, controlling for imitation of meaningless gestures and total lesion volume. (A) Maximally disconnected edges associated with reduced performance in tool use gesturing (greater hand posture errors in tool use gesturing controlling for hand posture errors in meaningless imitation and total lesion volume; red-to-yellow scale). (B) Cumulative node-level disconnection associated with reduced tool use gesturing projected on the cortical surface. (C) A heatmap of the disconnected subgraph depicting ROI-to-ROI disconnection patterns projected in Figure 3A.

Figure 4

SVR-CLSM of reduced right-hand grip strength, controlling for left-hand grip strength and total lesion volume. (A) Maximally disconnected edges associated with reduced right-hand grip strength (weak grip strength with the right hand controlling for grip strength of the left hand and total lesion volume; blue-to-green scale). (B) Cumulative node-level disconnection associated with reduced right-hand grip strength projected on the cortical surface. (C) A heatmap of the disconnected subgraph depicting ROI-to-ROI disconnection patterns projected in Figure 4A.

Figure 5

Analysis of IPL disconnection associated with hand posture errors in tool use gesturing, controlling for imitation of meaningless gestures and total lesion volume. (A) Edge-level disconnection identifies the left ventral temporal cortex and medial frontal regions as sites that exhibit a disconnection to the left inferior parietal lobule. (B) Cumulative node-level disconnection count associated with reduced tool use gesturing performance is projected on the cortical surface.