4.1 Subjects
Fourteen subjects volunteered to participate in the study (7 males, and 7 females, age: 33 ± 5 year, height: 1.71 ± 0.93 m, weight: 72.8 ± 14.2 kg, and BMI: 24.6 ± 3.5). None of the subjects had any history of neuromuscular or orthopedic diseases and all subjects were informed about the procedures and gave written informed consent. Subjects were fully introduced to the protocol, and they had the opportunity to withdraw from the study at any time. An ethical statement (1267/13.00.04.00/2021) was given by the ethics board of the University and the study was performed in conformity with the declaration of Helsinki (2013).
4.2 Experimental design
Three perturbation sessions (PS1, PS2, PS3) were conducted within 48 hours intervals. In PS1 and PS3, after EMG electrode set-up and 5 min cycling warm-up (70W) on the fitness cycle (Monark, 282E, Varberg, Sweden), 16 balance perturbations (without any stimulations) were used to collect center-of-pressure (COP) and EMG activity data. After this, subjects were positioned in a custom-built ankle dynamometer (University of Jyväskylä, Jyväskylä, Finland) to measure their isometric maximal voluntary contraction force (MVC) of the right leg. The TMS coil was set on a hotspot and the active motor threshold (aMT) was tested when subjects sat in the ankle dynamometer. With a TMS coil attached to the head and held by the custom-built helmet (Hu et al., 2022), subjects moved carefully to the balance platform, and 10 TMS pulses were given to measure MEPs during standing rest. In the balance perturbation task with the stimulation, MEPs were evoked two times after the onset of ankle movement during the balance perturbation in a random order. H-reflexes were elicited at the same two times and in a random order. The stimulations were delivered during each balance perturbation (anterior and posterior directions), however regardless of the direction, only MEPs and H-reflexes during posterior perturbations were analyzed. In PS2, 200 random balance perturbations (16 perturbations in one trial) were given to subjects with 1–2 min rest between perturbation trials.
4.3 Electromyography (EMG)
EMG was measured by bipolar electrodes (Blue Sensor, Ag/AgCl, 28 mm2, Ambu A/S, Ballerup, Denmark) placed 2 cm below the gastrocnemius on the line of the Achilles tendon for soleus (SOL), tibialis anterior (TA) and gastrocnemius (GM) muscles according to SENIAM guidelines (Hermens et al., 1999). MEPs during TMS measurements were collected from SOL only and using a pseudo-monopolar setup considering potential discomfort and tension caused by high-intensity stimulation, especially at 140 ms (voluntary activation) time-onset of the balance perturbation. The pseudo-monopolar setup allowed MEPs of higher amplitude to be recorded compared with 20 mm bipolar arrangement, which in turn also decreased the intensity of the stimulus needed to evoke a detectable MEP (Kirk et al., 2019). In addition, according to our practical experience, the shape of the MEP is more consistent with the pseudo-monopolar setup, which is important for dynamic tasks. A disadvantage of this electrode montage is that the signal-to-noise ratio can be compromised, however, this was not a problem in the current study. One electrode was placed 2 cm below the gastrocnemius on the line of the Achilles tendon and the reference electrode was placed on the tibia at the same level. The skin was shaved, carefully abraded with sandpaper, and cleaned with alcohol. Skin target impedance was less than 5 kΩ and if this was not the case, skin preparation was repeated. All EMG data were collected using the Neurolog EMG system (CED ltd., Cambridge, England), with a gain of 1000. Data were band-passed (15–500 Hz) filtered and further collected using CED 1401 A/D-converter (CED ltd., Cambridge, England) and Spike 2 (8.0) software (CED ltd., Cambridge, England) with a sampling rate of 5 kHz.
4.4 Isometric maximal voluntary contraction
Isometric maximal voluntary contraction (MVC) was used to investigate possible muscle fatigue between sessions and to monitor muscle contraction level during the measure of aMT (10% MVC). After EMG setup and a 5 min warm-up, subjects were positioned in a custom-built ankle dynamometer (University of Jyväskylä, Jyväskylä, Finland) to assess the MVC with the right foot on the pedal at 100° hip angle, 180° knee angle (leg fully extended) and 90° ankle angle. After the positioning procedure, the subject performed 5–7 submaximal plantarflexion trials to practice performance. MVC was performed at least three times at one-minute intervals and the highest force value was considered as the MVC. If the last trial was > 5% higher than the second-best, single additional trials were performed until no further improvement was observed. The typical number of required maximum trials was 3–5. Force from the dynamometer pedal was measured by a strain gauge transducer sampled at 1 kHz in Spike2 software and the maximum MVC amplitude was analyzed.
4.5 TMS and H-reflex measurement setups
TMS was delivered using a single-pulse Magstim 2002 stimulator with a double-cone coil (Magstim, Whitland, UK). A skin-tight (swimming) cap was placed on the head of the subject to increase friction between the coil and the scalp. The optimal TMS stimulation site for the right SOL was located on average 1 cm lateral (left) and 1 cm posterior to the cranial apex. Several stimulations were delivered to determine optimal coil placement and it was then marked by a marker pen on the cap. The aMT was the lowest stimulus intensity to elicit clear MEPs in three out of five stimulations from right ankle plantarflexion with 10% MVC. After confirmation of aMT, a second swimming cap with a hole in the middle of the vertex (Orca High Visibility Neoprene Swim Cap, Orca, Auckland, New Zealand) was placed over the coil to reduce the gap and relative movement between the coil and the head. Then, a custom-made helmet (modified from an ice-hockey helmet; CCM TACK 710 JK-K, CCM Hockey, Montreal, Canada) was attached to the subject’s head with a chin strap. In the balance perturbation system, the TMS cable was placed on a conveyor adjacent to the safety belt conveyor on the roof and connected with the balance platform by a firm handle, which was the same as in our previous study (Hu et al., 2022). Single-pulse TMS with 110% intensity of aMT was delivered during standing rest and balance perturbation tasks to investigate corticospinal excitability, and 110% intensity would cause less discomfort in our pilot study.
For H-reflex measurements, subjects stood relaxed during the electrical stimulation set-up. Electrical stimulation was administrated to the tibial nerve in the popliteal fossa. A cathode (1.5 cm × 1.5 cm) was placed over the tibial nerve, and an anode (5 cm × 8 cm) was placed above the patella. Rectangular stimulation pulse (DS7AH, Digitimer Ltd., Hertfordshire, UK) with a duration of 0.2 ms was delivered at 10s intervals. Once the optimal site of stimulation was found, the site was marked by a marker pen, and an electrode (Blue Sensor, Ag/AgCl, 28 mm2, Ambu A/S, Ballerup, Denmark) was placed and strapped around the subject’s knee with an elastic band. An increasing intensity interval (1–5 mA) was chosen to measure the H-M recruitment curve with at least 30 data points up to the maximal M-wave. The stimulus intensity was adjusted to 5% (± 2%) of the maximum M-wave, which was used during balance perturbations to control the stimulation intensity in H-reflex measurements.
4.6 Dynamic balance perturbations with TMS and H-reflex
Balance perturbation tasks utilized a custom-built dynamic balance device (University of Jyväskylä, Jyväskylä, Finland) modified from Piirainen et al. (2013) and Hu et al. (2022). The balance perturbation system operated at 0.25 m/s, accelerating at 2.5 m/s2 over a 0.3 m displacement. During balance perturbation tasks, 16 balance perturbations were delivered in anterior and posterior directions in a random order with 6–12 s intervals. A fixation point was set on the wall 3 m from the subjects at eye level to stabilize the subjects’ visual attention during measurements.
During balance perturbation tasks, the COP displacement and velocity in anterior and posterior directions were collected by a custom-designed balance platform, with one strain gauge sensor in each of the four corners of the force plate (BT4 balance platform; HUR Labs, Tampere, Finland) and saved and analyzed using the Coachtech-feedback system (University of Jyväskylä, Jyväskylä, Finland). COP in anterior-posterior direction was calculated using the formula \(\text{C}\text{O}\text{P}\text{y}=\left(\right(\text{F}\text{l}\text{f}+\text{F}\text{r}\text{f})\times 0.26-(\text{F}\text{l}\text{r}+\text{F}\text{r}\text{r})\times 0.26)/(\text{F}\text{l}\text{f}+\text{F}\text{r}\text{f}+\text{F}\text{r}\text{r}+\text{F}\text{l}\text{r})\), where lf = left front, rf = right front, or = right rear, lr = left rear, and 0.26(m) is a sensor distance from the middle line.
In our previous study, a constant delay (25 ms) between the platform control signal and the onset of ankle movement was observed (Hu et al., 2022). Therefore, the 40 ms and 140 ms after ankle movement were defined as SLR and voluntary activation timing. Using the same protocol as in TMS trials, H-reflex was measured in standing rest and during balance perturbations. In the dynamic balance perturbation tasks with the stimulation, 16 perturbations were performed in one set of trials, with 8 anterior and 8 posterior perturbations in a random order, which ensured that subjects were not able to anticipate the direction of the perturbation. Two-min rest periods were given after every perturbation set to minimize possible muscle fatigue (Piirainen et al., 2013). During H-reflex balance perturbation trials, a successful trial was defined as an M-wave response of 5% (± 2%) of the M-max value. The intensity of electrical stimulation was adjusted during perturbation trials to obtain at least five successful trials. If less than five successful backward trials in a 16-trial perturbation set were achieved, an extra 8-trial balance perturbation set, four backward and four forward, with a random order were performed. For each perturbation task, five successful trials were usually completed within 16–24 perturbations, followed by two minutes of rest.
4.7 Data and Statistical analysis
Perturbation trials were performed at 6–12 s intervals and triggered when COP was at least 1 s within ± 5 mm level from zero level. With this approach, the subject stood straight without anticipating any perturbations that might occur. COP values were analyzed in perturbation trials without stimulations. The mean standard deviation of the COP displacement curve was calculated to evaluate the general body sway (COP_SD). Peak to peak COP displacement (dCOP) was analyzed in the time window of 1 s before platform movement (Preparation-phase; Pre), during platform movement (Active-phase; Act), and 1 s from the end of platform movement (Recovery-phase; Rec) (see Error! Reference source not found.). The COP velocity curve was calculated by differentiating the COP curve by using 20 ms windows, and then the average COP velocity of the velocity curve (vCOP) was analyzed in the same time window as dCOP (see Fig. 1). Both dCOP and vCOP were normalized by individual subject’s height × weight (dCOP: mm/(m*kg); vCOP: (mm/s)/(m*kg)) according to the recommendation of Chiari et al. (2002).
In the balance perturbation trials with no stimulations, average of all subjects’ full wave rectified EMG data from 100ms before the perturbation onset to 400 ms after onset was analyzed shown in Fig. 2A. Furthermore, EMG activity was analyzed using root mean square (RMS) with 20 ms time windows from the perturbation onset (0 ms) to 180 ms after onset (Fig. 2B).
MEPs and H-reflex were measured during standing rest and balance perturbation. In the standing rest, a clear MEP was defined to start when EMG was above the mean + 2SD level recorded 100 ms before the TMS trigger and ending when below the mean − 2SD level (Hirano et al., 2016). The average MEP latency and the MEP duration were analyzed during the standing rest and then used to trigger TMS during the balance perturbation. Average MEP values were determined with peak-to-peak amplitude (in mV) from 10 TMS stimulations. Outliers were identified from the ten trials (± 2.5 SD) and removed before analysis (Avenanti et al., 2006). In balance perturbation trials, MEP amplitude from 7–8 trials was selected when the platform moved backward and averaged after excluding outliers (± 2.5 SD), which has been proved good reliability in our previous study (Hu et al., 2022). All MEPs were normalized to the peak-to-peak value of the maximal M-wave (Mmax) and presented as %Mmax in the results. H-reflex was determined as peak-to-peak amplitude and averaged from all successful trials (within 3% − 7% Mmax) in standing rest and balance perturbation tasks. H-reflex was normalized to the Mmax as well. RMS of background EMG (BGemg) was also analyzed with a 30 ms window before TMS and H-reflex trigger and normalized to Mmax (monopolar and bipolar).
In order to evaluate neural excitability changes between corticospinal and spinal level MEP/H-reflex ratio (MEP/H ratio) was calculated. To explore the relationship between the changes in balance performance and corticospinal modulation from PS1 to PS3, ∆MEP and ∆H-reflex and their correlation with ∆ dCOP were analyzed by Person correlation. Delta values were calculated (i.e., MEP, H-reflex, and dCOP respectively) using the formula: (value (PS1) - value (PS3)) / value (PS1) × 100%.
Statistical analyses were conducted using JASP (Version 0.17.1). Result visualizations were performed using Prism (V9, GraphPad Software, San Diego, California USA). All variables of the MEP/H ratio were processed by log transformation before statistical analyses following Nielsen’s suggestion (Nielsen, 1996) since the original data was not normally distributed, which resulted in data being normally distributed as assessed by Shapiro-Wilk’s W tests.
Mmax, MVC, and Hmax/Mmax values were assessed by paired t-test. Since dCOP and vCOP were analyzed by different time windows (i.e., 1 s for Pre and Rec phase, but 1.3 s for the Act phase), between-session differences of dCOP, vCOP, and COP_SD were examined by paired t-test. To assess modulation in corticospinal excitability during balance perturbations, MEPs, H-reflex, BGemg, and EMG activity were assessed by two-way (2 × 3) repeated-measures ANOVA with the factors SESSION (PS1 and PS3) and TIME (standing rest, 40 ms, 140 ms). When a significant F-value was observed, Mauchly’s test was used to evaluate sphericity, and where the assumption was valid F-values were reported with sphericity-assumed degrees of freedom and df error (i.e. F (sphericity assumed df, df error)). Effect sizes for the ANOVA main effects are reported as partial eta squared (ηp2), where 0.02, 0.13, and 0.26 are considered small, medium, and large, respectively. If significance for TIME was revealed, Bonferroni post hoc analysis was used for pairwise comparisons between levels (standing rest, 40 ms, 140 ms). The significance level was set at p < 0.05 and all results were displayed as Mean ± SD in text and figures.