Cognitive efficacy and neural mechanisms of music-based neurological rehabilitation for traumatic brain injury

Abstract

Traumatic brain injury (TBI) causes lifelong cognitive deficits, most often in executive function (EF). Both musical training and music-based rehabilitation have been shown to enhance EF and neuroplasticity. Thus far, however, there is little evidence for the potential rehabilitative effects of music for TBI. Here, we review the core findings from our recent cross-over randomized controlled trial in which a 10-week music-based neurological rehabilitation (MBNR) protocol was administered to 40 patients with moderate-to-severe TBI. Neuropsychological testing and structural/functional magnetic resonance imaging were collected at three time points (baseline, 3 months, and 6 months); one group received the MBNR between time points 1 and 2, while a second group received it between time points 2 and 3. We found that both general EF and set shifting improved after the intervention, and this effect was maintained long term. Morphometric analyses revealed therapy-induced gray matter volume changes most consistently in the right inferior frontal gyrus, changes that correlated with better outcomes in set shifting. Finally, we found changes in the between- and within-network functional connectivity of large-scale resting-state networks after MBNR, which also correlated with measures of EF. Taken together, the data provide evidence for concluding that MBNR improves EF in TBI; also, the data show that morphometric and resting-state functional connectivity are sensitive markers with which to monitor the neuroplasticity induced by the MBNR intervention.

Keywords: brain morphometry; executive function; music-based neurological rehabilitation; resting-state networks; traumatic brain injury.

FIGURE 1

The three constituent modules of the music‐based neurological rehabilitation. (A) Rhythmical training with djembe. (B) Structured cognitive‐motor training with drum set. (C) Assisted music playing using figure notes

FIGURE 2

Behavioral test results from the intention‐to‐treat (ITT) analysis. (A) Frontal Assessment Battery (FAB) score, (B) Number–Letter Task (NLT) switching cost error (panel taken from Ref. 53), and (C) BRIEF‐A Behavioral Regulation Index (panel taken from Ref. 54). The bar plots (mean SEM) show changes in test scores over the three time points (TP) presented group‐wise (AB/BA) from the imputed data set (depicting the mean of 20 imputations). Significant time–group interactions are shown with solid gray lines, and significant within‐group time main effects are shown with dashed gray lines. Abbreviation: SEM, standard error of the mean

FIGURE 3

VBM results from the mixed‐model ANOVA, including 25 patients with TBI and three time points (TP). (A) Group (AB > BA) Time (TP2 > TP1) interaction in gray matter volume (GMV). (B) Group (BA > AB) Time (TP3 > TP2) interaction in GMV. (C) Conjunction analysis between A and B in GMV. (D–F) Bar plots (mean SEM) for GMV from the local maxima in the right IFG Tri cluster in A, B, and C contrasts, respectively. The results are reported in MNI coordinates and at an uncorrected p < 0.005 threshold at the voxel level for visualization purposes. Abbreviations: ACC, anterior cingulate gyrus; CB, cerebellum lobule 8; FusG, fusiform gyrus; IFG Tri, inferior frontal gyrus triangular part; L, left; MCgG, middle cingulate gyrus; MFG Orb, middle frontal gyrus orbital part; MNI, Montreal Neurological Institute; MTG, middle temporal gyrus; R, right; SEM, standard error of the mean; SFG, superior frontal gyrus; SPG, superior parietal gyrus; TBI, traumatic brain injury; VBM, voxel‐based morphometry. Figure taken from Ref.

FIGURE 4

Changes during the intervention versus control period pooled across groups in 25 patients with TBI. (A) VBM results from the intervention versus control comparison (AB:TP2 > TP1 & BA:TP3 > TP2 vs. AB:TP3 > TP2 & BA:TP2 > TP1). The results are reported in MNI coordinates and at an uncorrected p < 0.001 threshold at the voxel level. The right IFG Tri cluster shown survived a p < 0.05 FWE‐corrected threshold at the cluster level. (B) Bar plots (mean SEM) for the Number–Letter Task switching cost score (errors, %) change (left) and the GMV change from the local maxima in the right IFG Tri cluster shown in A (right) during the intervention (INT) and control (CON) periods. (C) Pearson correlation between changes in the Number–Letter Task switching cost (errors, %) and in the GMV of the right IFG Tri cluster from A during the intervention period. Abbreviations: FWE, familywise error; GMV, gray matter volume; IFG Tri, inferior frontal gyrus triangular part; INS, insula; MNI, Montreal Neurological Institute; MOFC, medial orbitofrontal gyrus; R, right; SEM, standard error of the mean; TBI, traumatic brain injury; VBM, voxel‐based morphometry. Figure taken from Ref.

FIGURE 5

Changes in between‐network connectivity induced by the music‐based neurological rehabilitation. Nodes are overlaid on a rendered semitransparent brain generated using CONN. Connectivity matrices display the mean post‐ minus pre‐intervention (A–C) Fisher‐transformed Z‐score correlation values for each node. The bar plots (D) show the effect size of the AB>BA and TP2>TP1 interaction represented by the Fisher‐transformed Z‐score correlation values for each node. Abbreviations: DAN, dorsal attention; DMN, default mode network; FEF, frontal eye field; FPN, frontoparietal; IPS, intraparietal sulcus; L, left; LAT, lateral; LPFC, lateral prefrontal cortex; MED, medial; MPFC, medial prefrontal cortex; OCC, occipital; PCC, posterior cingulate cortex; PPC, posterior parietal cortex; R, right; SM, sensorimotor; Sup, superior; VIS, visual. Figure taken from Ref.

FIGURE 6

Within‐ and between‐network functional connectivity changes associated with cognitive recovery induced by the music‐based neurological rehabilitation. (A) The behavior and functional connectivity values are derived from the pre–post intervention comparison, which was significant for the within‐network connectivity changes in the FPN. (B) The behavior and functional connectivity values are derived from the AB > BA and TP2 > TP1 interaction, which was significant for the network connectivity changes between the DMN and the SM network. The scatter plots represent the bivariate Pearson correlation, and shaded areas represent the 95% CI prediction bounds. Abbreviations: DMN, default mode network; FPN, frontoparietal; L, left; LPFC, lateral prefrontal cortex; MPFC, medial prefrontal cortex; R, right; SM, sensorimotor; Sup, superior. Figure taken from Ref.