Corticomuscular and intermuscular coherence are correlated after stroke: a simplified motor control?

Abstract

During movement, corticomuscular coherence is a measure of central-peripheral communication, while intermuscular coherence is a measure of the amount of common central drive to the muscles. Although these two measures are modified in stroke subjects, no author has explored a correlation between them, neither in stroke subjects nor in healthy subjects. Twenty-four chronic stroke subjects and 22 healthy control subjects were included in this cohort study, and they performed 20 active elbow extension movements. The electroencephalographic and electromyographic activity of the elbow flexors and extensors were recorded. Corticomuscular and intermuscular coherence were calculated in the time-frequency domain for each limb of stroke and control subjects. Partial rank correlations were performed to study the link between these two variables. Our results showed a positive correlation between corticomuscular and intermuscular coherence only for stroke subjects, for their paretic and non-paretic limbs (P < 0.022; Rho > 0.50). These results suggest, beyond the cortical and spinal hypotheses to explain them, that stroke subjects present a form of simplification of motor control. When central-peripheral communication increases, it is less modulated and more common to the muscles involved in the active movement. This motor control simplification suggests a new way of understanding the plasticity of the neuromuscular system after stroke.

Keywords: common central drive; motor control; neuromuscular network; stroke.

Graphical Abstract

Figure 1

Description of the different steps involved in the calculation of coherences, illustrated for the corticomuscular coherence. Panel Ai represents the electroencephalographic activity of the C3 electrode (left) and Panel Aii, the electromyographic activity of the brachioradialis muscle (right) typical of a healthy subject (n = 1). Panels Bi and Bii represent the average auto-spectrum calculated from the respective electrophysiologic signals. Panel C represents the cross-spectrum and the red contours identify the areas in the time–frequency plane where the correlation between the electrophysiologic signals is significant. Panel D represents the wavelet magnitude-squared coherence between the two electrophysiologic time series. Panel E represents the wavelet magnitude-squared coherence where the correlation between the electrophysiologic signals is significant. For the calculation of intermuscular coherence, the processing steps are identical but with the input data in A being an electromyographic signal (for an illustration, see Charissou et al., Figure 1).

Figure 2

Mean coherence maps of the paretic limb (first row) and non-paretic limb (second row) of stroke subjects (n = 24) and the dominant limb (third row) and non-dominant limb (fourth row) of control subjects (n = 22). Intermuscular coherence (IMC) is represented for the biceps brachii–brachioradialis (BB-BR) (first line) and triceps brachii–brachioradialis (TB-BR) (second line) muscle pairs. The mean corticomuscular coherence (CMC) is represented between the biceps brachii and brachioradialis muscles (BB-BR) (third line) and triceps brachii and brachioradialis muscles (TB-BR) (fourth line). Please refer to Delcamp et al.^, for our previous investigation of the differences in CMC and IMC between stroke and control subjects. In the last line, the mean active range of motion during the movement is plotted with the standard deviation. The dot represents the velocity peak and the vertical lines the temporal quantification window.

Figure 3

Partial rank Spearman’s correlation plot of (A) corticomuscular and intermuscular coherence for stroke subjects for the biceps brachii–brachioradialis [Rho = 0.50 (0.10–0.78); P = 0.022*] and (B) triceps brachii–brachioradialis muscle pairs [Rho = 0.72 (0.36–0.91); P ≤ 0.001*] of their paretic limb and (C) the biceps brachii–brachioradialis [Rho = 0.78 (0.48–0.91); P ≤ 0.001*] and (D) triceps brachii–brachioradialis muscle pairs [Rho = 0.59 (0.20–0.86); P = 0.005*] for their non-paretic limb. The data plotted are the residuals of the Spearman’s rank correlation performed between: intermuscular coherence and the co-variables, and corticomuscular coherence and the co-variables. The asterisks represent a significant correlation with respect to the Benjamini-Hochberg critical P-value.

Figure 4

Partial rank Spearman’s correlation plot of (A) corticomuscular and intermuscular coherence for control subjects for the biceps brachii–brachioradialis [Rho = 0.12 (−0.48 to 0.61); P = 0.63] and (B) triceps brachii–brachioradialis muscle pairs [Rho = 0.06 (−0.36 to 0.61); P = 0.82] of their dominant limb and (C) the biceps brachii–brachioradialis [Rho = 0.35 (−0.21 to 0.68); P = 0.14] and (D) triceps brachii–brachioradialis muscle pairs [Rho = 0.57 (−0.27 to 0.90); P = 0.10] for their non-dominant limb. The data plotted are the residuals of the Spearman’s rank correlation performed between: intermuscular coherence and the co-variables, and corticomuscular coherence and the co-variables.