Hemodynamic responses are abnormal in isolated cervical dystonia

Abstract

Neuroimaging studies using functional magnetic resonance imaging (fMRI), which measures brain activity by detecting the changes in blood oxygenation levels, are advancing our understanding of the pathophysiology of dystonia. Neurobiological disturbances in dystonia, however, may affect neurovascular coupling and impact the interpretability of fMRI studies. We evaluated here whether the hemodynamic response patterns during a behaviorally matched motor task are altered in isolated cervical dystonia (CD). Twenty-five CD patients and 25 healthy controls (HCs) underwent fMRI scanning during a paced finger tapping task (nondystonic task in patients). Imaging data were analyzed using a constrained principal component analysis-a statistical method that combines regression analysis and principal component analysis and enables the extraction of task-related functional networks and determination of the spatial and temporal hemodynamic response patterns associated with the task performance. Data from three patients and two controls were removed due to excessive movement. No significant differences in demographics or motor performance were observed. Three task-associated functional brain networks were identified. During task performance, reduced hemodynamic responses were seen in a sensorimotor network and in a network that included key nodes of the default mode, executive control and visual networks. During rest, reductions in hemodynamic responses were seen in the cognitive/visual network. Lower hemodynamic responses within the primary sensorimotor network in patients were correlated with the increased dystonia severity. Pathophysiological disturbances in isolated CD, such as alterations in inhibitory signaling and dopaminergic neurotransmission, may impact neurovascular coupling. Not accounting for hemodynamic response differences in fMRI studies of dystonia could lead to inaccurate results and interpretations.

Keywords: BOLD; cervical dystonia; fMRI; finger tapping; hemodynamic response.

Figure 1.

Results of traditional blocked-design voxel-based analyses showing mean sensorimotor network activation maps during a finger-tapping task in A) healthy controls and B) patients with cervical dystonia. C) No significant differences in response were identified between controls and patients. Voxel-based findings were family wise error corrected for multiple comparisons at an overall p <0.05.

Figure 2.

Three functional networks were identified with CPCA and are displayed here with the strongest 5% of component loadings shown. Component 1 (A) consisted of activations (positive loadings, threshold = 0.31, max = 0.53) within a sensorimotor network, Component 2 (B) consisted of deactivations (negative loadings, threshold = −0.20, max = −0.30) of key nodes of the default mode, executive control, and visual networks, and Component 3 (C) consisted of activations (positive loadings, threshold = 0.18, max = 0.31) within sensorimotor network structures including the sensorimotor cortex and cerebellum. Spatial patterns for each functional network are overlaid on a standard Montreal Neurological Institute template with activation displayed using warm colors (red/yellow) and deactivation using cool (blue) colors.

Figure 3.

Predictor weight time courses averaged over cervical dystonia (CD) patients and healthy controls showing estimated task- and rest-related hemodynamic response functions for A) Component 1 (sensorimotor network), B) Component 2 (default mode, executive control and visual networks), and C) Component 3 (sensorimotor/cerebellar network). Error bars shown are standard deviations. * denotes significance (p<0.05) observed with univariate t-tests.

Figure 4.

The hemodynamic response function within the sensorimotor network (Component 1) during an unaffected finger-tapping task as measured by area under the curve negatively correlated with the severity of cervical dystonia in patients assessed using the motor score of the Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS). Residuals are shown as data were adjusted for age, sex and scanner.