High gamma oscillations (60-90 Hz) in the human primary motor cortex (M1) are elevated at the onset of active movements (movement-related gamma synchronization, MRGS) [1,2]. While subcortical MRGS has been shown to be related to movement velocity, effort, or force level [3-5], the relationship between MRGS in M1 and motor performance measures remains less clear.

We recorded magnetoencephalography data from 29 healthy young participants, 14 well-recovered chronic stroke patients and 15 age-matched control participants during a movement speed task. All participants were right-handed. The task consisted of alternately pressing two buttons four times with the thumb of the performing hand (the left hand in the young group, the affected hand in stroke patients, the hand matched to the stroke group in the control group), as quickly as possible. Movement speed was defined as the inverse of the duration between the first and fourth button press. Each participant performed 6 blocks of 40 trials each. Participants received feedback regarding their task performance and their improvement relative to previous trials.

Movement speed served as a significant predictor of MRGS in the young and control participant groups, as trials with higher movement speed were associated with greater MRGS in M1 and premotor cortex (PMC) (Fig. 1, young participants: M1: χ²(1) = 24.6, p_cor < 0.001, PMC: χ²(1) = 39.5, p_cor < 0.001; control participants: M1: χ²(1) = 15.3, p_cor < 0.001, PMC: χ²(1) = 10, p_cor = 0.003). In stroke patients, the association of movement speed and MRGS did not reach significance (M1: p_cor = 0.14, PMC: p_cor = 0.14). Stroke patients had significantly lower MRGS than control participants (Fig. 2, M1: p_cor = 0.04, PMC: p_cor = 0.04), and control participants had significantly lower MRGS than young participants (M1: p_cor = 0.04, PMC: p_cor = 0.03). When matching movement speed and handedness in stroke patients and age-matched controls, stroke patients still had a significantly lower MRGS than control participants (Figure 2C, M1: p_cor = 0.04, PMC: p_cor = 0.04).

Our findings suggest a strong link between movement speed and MRGS in the human M1 and PMC. In addition, we find that MRGS is decreased in stroke patients compared to age-matched control participants, even if corrected for movement speed. We propose the exploration of non-invasive brain stimulation in the high gamma range as a strategy to improve the rehabilitation of stroke patients.

Figure 1: Effect plots of the effect of the predictor movement speed in trial-level linear mixed-effects models that model the relationship between MRGS and movement speed. Movement speed was a significant predictor of MRGS in young and control participant groups in both M1 and PMC. ***p < 0.001

Figure 2: A: Group-averaged MRGS topography (flipped to the right hemisphere). B: Distribution of participant averages of MRGS in virtual sensors in M1 and PMC. C: Group difference in MRGS when matched for movement speed. Boxplots show the distribution of mean MRGS for each participant of 1000 repetitions of the matching process. *p < 0.05