Neuromuscular electrical stimulation (NMES) that consists of evoking contractions by applying an electrical current over the muscles via surface electrodes is a successful method to enhance force (3,15,16,23,31) or to provide contractile activity in paralyzed or immobilized muscles (14,22,31,40,42). However, the acute effects of a single NMES session may limit its wide utilization, at least in a rehabilitation context. Indeed, in addition to the perceived discomfort, NMES has been associated to a greater muscle fatigue and exaggerated metabolic response compared with voluntary exercise performed at the same intensity (42,45). Thus, by varying stimulation parameters such as pulse duration and stimulation frequency (10,17,31,34,35), previous studies aimed to determine the optimal NMES protocol, that is, which would minimize muscle fatigue and maximize muscle performance.
It has been recently shown that a stimulation protocol combining long pulse durations (1 ms) with a high frequency (100 Hz) delivered at low stimulation intensities (evoking 10% of maximal voluntary contraction [MVC]) induced a lower evoked contraction force compared with a stimulation protocol using low-frequency (25 Hz), short pulse duration (50 μs), and higher pulse amplitude (34). This result led the authors to conclude that the 1 ms–100 Hz combination induced greater muscle fatigue despite a similar decline in MVC force induced by both protocols (34). Similarly, Papaiordanidou et al. (35) examined muscle fatigue development during NMES protocols using two different stimulation frequencies (30 vs 100 Hz) and two different pulse durations (500 μs vs 1 ms). The authors showed that independently of the pulse duration, the protocol using higher stimulation frequencies induced a greater evoked torque decline, whereas MVC torque loss was identical after both protocols. The greater torque decline during NMES was not attributed to a greater neuromuscular fatigue, but rather to changes in the excitability threshold of the activated axons, leading to a decreased number of active motor units (33). This phenomenon related to activity-dependent changes in axonal excitability, more prominent for high stimulation frequencies (7), could account for the higher decrease in evoked torque during high as compared with low-frequency stimulation, whereas subjects' maximal force-generating capacity is similar altered. However, the stimulation frequency specifically alters the evoked torque during repeated stimulation trains (18,34,35). Therefore, applying the same number of evoked contractions when testing protocols with different stimulation frequencies may induce different total muscle work, which may lead to different effects on the neuromuscular system. To overcome this problem, the total muscle work has to be matched by adapting the number of evoked contractions for each frequency of stimulation, to obtain a similar total torque time integral (TTI) value. This methodology would help to draw valid conclusions concerning the effects of stimulation frequency modulation on neuromuscular fatigue.
The use of different combinations of pulse durations, amplitudes and stimulation frequencies may modulate the neural pathway by which the muscle is activated, because NMES does not activate directly the muscle fibers but the terminal branches of a mixed nerve (28), that is, containing afferent and efferent fibers. The activation of afferent fibers, mainly Ia, may induce an indirect recruitment of the muscle via a reflex pathway. In contrast, the activation of efferent fibers, that is, the motor axons, may induce a direct recruitment of the motor unit (MU). The reason why NMES induces the activation of direct or indirect pathways mainly depends on the specific excitability threshold of afferent fibres and motor axons (27). Because excitability changes in response to repeated stimulation also differ between afferent and efferent fibers (7), modulating the NMES parameters will affect the relative contribution of direct and indirect motor unit activation, but also the neural mechanisms (spinal and/or supraspinal) underlying neuromuscular fatigue.
In a translational perspective, refining our understanding of the neural mechanisms that subserve such electro-induced muscle contractions should widely help to optimize NMES programs. Indeed, given the popularity of NMES in both rehabilitation and sport performance (31), using optimal stimulation parameters is of great clinical importance. The purpose of the present study was to investigate the acute effects of submaximal NMES protocols delivered at different stimulation frequencies (20, 60, and 100 Hz) on the neuromuscular fatigue development of triceps surae muscles during both evoked NMES contractions and during MVC. Particular care was taken in matching a similar total muscle work by modulating the pulse amplitude and the number of evoked contractions. We hypothesized that increasing stimulation frequency from low to high would induce a greater decrease in both NMES-induced torque and MVC torque, due to a greater indirect activation of the muscle. To test this hypothesis, electrophysiological and mechanical responses of the solicited motor units during each protocol were recorded by means of single electrical pulses, that is, single twitches, evoked at the pulse amplitude used during each protocol (32).
Ten healthy subjects (seven men: age, 24.6 ± 4.2 yr; height, 1.76 ± 0.1 m; weight, 73.1 ± 10.8 kg), with no history of neurological or muscular disorder, gave written informed consent to perform three experimental sessions with different NMES protocols combining a pulse duration of 1 ms with three different stimulation frequencies of 20, 60, and 100 Hz. The sessions were separated by at least 7 d and randomly performed. The subjects committed not to engage in any unusual training or exercise program during the duration of the study and should avoid any intense exercise before and after each session. The experimental design of the study was approved by the regional ethic committee (Comité de Protection des Personnes de la Région Grand Est) and conducted in conformity with the latest version of the Declaration of Helsinki.
Overview of the experimental design
After skin preparation and electrode positioning, subjects sat on an isokinetic dynamometer (Biodex System 3, Shirley, NY). Recruitment curves were first established to determine nerve stimulation intensities needed to evoke H-reflexes and M-wave responses of the triceps surae muscles. Experiments started with a warm-up consisting of 8 to 10 submaximal contractions. Then, the subjects were asked to perform eight MVC separated by at least 1 min rest to record four superimposed H-reflexes and four superimposed M-waves. For each session, the pulse amplitude of muscular stimulation (IES) was then determined to obtain an evoked torque corresponding to 20% of MVC. Single pulses were also evoked over the muscle at rest and at IES to provide triceps surae mechanical response and soleus electromyographic response associated to myostimulation.
Each session consisted of intermittent evoked contractions of the triceps surae muscle (6 s ON/6 s OFF) at constant NMES pulse amplitude (IES). Independent of the protocol tested, 40 contractions were always evoked in the first session. To keep constant muscular solicitation across the sessions, the total TTI of the 40 contractions reached during the first session was matched during the two following sessions by adjusting the number of evoked contractions.
Immediately after the end of the last train, muscular responses (evoked by myostimulation at IES) and nerve responses (evoked by nerve stimulation) were evoked. Subjects had to perform two MVC; one with superimposed maximal M-wave and one with superimposed H-reflex. At rest, one supramaximal nerve stimulation (to record maximal M-wave) and four single myostimulations at IES were evoked. All POST measurements were randomized.
The axis of the dynamometer was aligned with the right external malleolus. Subjects were placed with the hip and knee joints at 90° (180°, full extension), and ankle joint at 90° (angle between the leg and the sole of the foot). During all conditions, particular care was taken to avoid trunk and head rotations to maintain constant corticovestibular influences on the excitability of the motor pool (39). The foot was firmly strapped to the pedal of the dynamometer. The trunk was stabilized by two crossover shoulder harnesses. A summary of the experimental design is depicted in Figure 1.
Torque signal was recorded continuously during the protocol and integrated on line by a custom electronic device. Then, the TTI of each NMES contraction were calculated and summed. The total value of cumulated TTI was displayed online so that during the second and third sessions, NMES stimulations were stopped once the value of the total TTI obtained in the first session was reached.
EMG activity was recorded from the three triceps surae muscles (soleus [SOL], medial gastrocnemius [MG], lateral gastrocnemius [LG]) and from one muscle of the tibial compartment (tibialis anterior, TA). After shaving and dry cleaning, the skin with alcohol to keep low impedance (<5 kΩ), EMG signals were obtained by using two silver-chloride surface electrodes (8-mm diameter; center-to-center distance, 2 cm) placed over the muscle bellies. The common reference electrode was placed in a central position between the stimulation and recording sites. Electrode placements were marked on the skin with indelible ink to ensure same positioning between sessions.
For SOL, MG, LG, and TA muscles, respectively, the electrodes were positioned 2 cm below the insertions of the gastrocnemii over the Achilles tendon, over the mid belly of the gastrocnemii muscles, and at one-third of the distance on the line between the fibula and the tip of the medial malleolus (see Fig. 2).
EMG signals were amplified (gain = 1000) and filtered with a bandwidth frequency ranging from 15 to 1 kHz. EMG and mechanical signals were digitized online (sampling frequency, 5 kHz) and stored for analysis with Tida software (Heka Elektronik, Lambrecht/Pfalz, Germany).
Trains of monophasic rectangular electrical stimuli were delivered to the triceps surae muscles by using a high-voltage (maximal voltage 400 V) constant-current stimulator (digitimer stimulator, model DS7A; Hertfordshire, UK). Two large electrodes (5 × 10 cm; Medicompex SA, Ecublens, Switzerland) were placed over the gastrocnemii (∼5 cm below the popliteal fossa) and soleus (∼10 cm above the calcaneus) muscles, as described by Collins et al. (11,12). The electrode placements were marked on the skin with indelible ink to ensure the same positioning between sessions. The upper NMES electrode (cathode) was placed proximally to the EMG electrodes of gastrocnemii muscles and the lower one (anode) proximally to the EMG of soleus muscle. Such an electrode arrangement allows to record soleus electromyographic responses to myostimulation (Fig. 2).
Pulse duration was always set at 1 ms. Frequencies of stimulation were randomly set at 20, 60, and 100 Hz, respectively, for the three sessions. The pulse amplitude (IES) was set to evoke an initial force level corresponding to 20% MVC with testing trains of 1 s using frequencies of 20, 60, or 100 Hz.
We also evoked single NMES stimulations at IES before and after each protocol to analyze M and H waves. These electromyographic responses were analyzed for SOL muscle, noted HES for H-reflex and MES for M-wave (see Fig. 2 for experimental setup). These electrophysiological responses were useful to characterize muscle activation induced by NMES, that is, to identify the relative contribution of direct (MES) and indirect—reflexive activation (HES); and to quantify the total muscle activation before and after the protocols (HES + MES). The concomitant analysis of electromyographic (HES + MES) and mechanical (PtES) responses associated with single NMES pulse also provided clues about the mechanisms underlying the decrease of evoked torque (33).
Posterior tibial nerve stimulation
The present study used specific investigation tools for each type of muscle contraction, because evoked torque and maximal voluntary force can change differently with NMES-induced fatigue, resulting from different mechanisms (Neyroud 2014). Here, the analysis of muscle electromyographic responses to single percutaneous tibial nerve stimulations (H-reflexes, M- and V-waves), evoked during, pre- and post-MVC, provided information about the mechanisms responsible for the decline of the maximal force generation, that is, at neuromuscular junction (M-wave), spinal (H-reflex) or supraspinal (V-wave) levels (5,25).
Single rectangular pulses (1 ms duration) were delivered to the posterior tibial nerve with a Digitimer stimulator (model DS7) to evoke H-reflex of the triceps surae muscles. A self-adhesive cathode (8-mm diameter, Ag-AgCl) was placed in the popliteal fossa and an anode (5 × 10 cm, Medicompex SA, Ecublens, Switzerland) over the patella. The optimal stimulation site was first located by a handheld cathode ball electrode (0.5-cm diameter). Then, the stimulation electrode was firmly fixed with straps. The intensity of stimulation was then increased from H-reflex threshold to maximal M-wave (Mmax), with 1-mA increment. The Mmax intensity was increased by 50% to ensure supramaximal stimulation, so that no variation of maximal M-wave amplitude could be the result of a variation in axonal excitability (27).
The intensity used to record H-reflex was the one that elicited maximal superimposed H-reflex (Hsup) in triceps surae muscles, which is generally associated with a small M-wave, noted MatHsup. Intensity of stimulation was optimized to record the greatest SOL responses, but variations in SOL, MG, and LG MatHsup amplitudes were useful to identify any modification in posterior tibial nerve stimulation during the experiment (20). The responses of the three muscles (SOL, MG, LG) to nerve stimulation were elicited before and after each protocol during the recording of MVC. They were noted Msup for maximal M-wave, Hsup for maximal H-reflex, and MatHsup (see Fig. 2 for experimental setup).
During the 6-s NMES trains, the mechanical signal was time-integrated to determine the total TTI for each protocol. The mean torque was analyzed and compared with PRE value, that is, mean torque as assessed during a 1-s trains elicited at IES before each NMES protocol. Mean torques, expressed in percentage of PRE value, were then plotted against the time during the NMES protocol. Given the discrepancies of numbers of evoked contractions between each session, the duration of each protocol (20, 60, and 100 Hz) was expressed as a percentage of total duration. Thus, mean torque during each NMES protocol was analyzed by an average of three to six responses according to the total number of evoked contractions. This provided 10 average values per subject, noted in percentage of total time of the protocol (from 10% to 100%). Group data of these 10 values are depicted in Figure 3 for each protocol.
Before (PRE) and after (POST) each protocol, peak-to-peak amplitudes of mechanical twitches were also analyzed and compared. This analysis was performed for single stimulations at IES (PtES) and for nerve stimulation at Mmax (PtM). The ratio PtES/PtM reflecting the balance between the mechanical response of plantar flexor muscles activated by NMES and the maximal mechanical response of the entire pool of MU was calculated.
For each session and for each muscle, the average of peak-to-peak amplitude of Mmax, Msup, Hsup,MatHsup, and V waves were calculated. Electromyographic activity of each tested muscles was quantified with root mean square (RMS) values of the raw signal over a 500-ms period before the stimulation. Electrophysiological responses and RMS were normalized to maximal M-wave (Msup) evoked in the same condition. RMS/Msup, Hsup/Msup, V/Msup, and MatHsup/Msup were then considered as dependent variables and compared between the three protocols.
For each session, SOL electromyographic responses to single NMES pulses at IES (MES and HES) were identified according to their latencies, measured from the onset of stimulation artefact to the first peak of the maximal recorded response during PRE measurement. The sum of HES and MES was also calculated to provide the total muscle activation induced by NMES. These responses, as well as HES + MES were then normalized by Mmax as follows (HES + MES)/Mmax, MES/Mmax, and HES/Mmax.
All data are expressed by their mean ± SD. Their normality was verified by the Shapiro–Wilk test (P < 0.05) to ensure that classical ANOVA could be used.
Differences between protocols in pulse amplitude, number of contractions, and total TTI were assessed by one-way repeated-measures ANOVA with factor frequency of stimulation (20, 60, and 100 Hz). A two-way repeated-measures ANOVA was conducted for each variable measured PRE and POST with factors frequency of stimulation (20, 60, and 100 Hz) and PRE versus POST. This analysis was performed on electrical and mechanical data (RMS/Msup, MVC, PtES, PtM, PtES/PtMmax), evoked responses with nerve stimulation on the three muscles of the triceps surae (Mmax, Msup,MatHsup/Msup, Hsup/Msup, V/Msup) and evoked responses of SOL muscle to myostimulation at IES (MES/Mmax, HES/Mmax, MES + HES/Mmax). A two-way repeated-measures ANOVA was also conducted over mean torque, with factors frequency of stimulation (20, 60, and 100 Hz) and time (from PRE value and 10% to 100% of total time).
When a main effect or an interaction was found, a post hoc analysis was made (HSD, Tukey test). Statistical analysis was performed using STATISTICA (8.0 version; Statsoft, Tulsa, OK). Significance level was set at P < 0.05. Effect sizes were calculated by the partial eta-squared method (31). For each session Pearson correlations (r coefficient) were assessed with P obtained in the Bravais–Pearson table (degree of freedom = 8) between the relative PRE–POST evolutions of the following variables: mean evoked torque, PtES, HES + MES, MVC, V/Msup, Hmax/Mmax.
Evolution of evoked contractions during NMES
A significant frequency effect was found on the number of evoked contractions (F2,18 = 38.96, P < 0.001, ηP2 = 0.81) which were: 34 ± 3 at 20 Hz, 41 ± 3 at 60 Hz, and 45 ± 5 at 100 Hz for a similar total TTI among the three protocols (4455 ± 303 N·m·s−1 on average). A significant interaction effect was found between the factors time and frequency of stimulation on mean evoked torque (F18,162 = 69.46, P < 0.001, ηP2 = 0.89) suggesting that the three NMES protocols induced different torque development kinetics throughout the session, with a greater reduction in evoked torque at 60 Hz and 100 Hz compared with 20 Hz (Fig. 3). Mean torque increased compared with PRE value during the first 20% time duration (extra-force phenomena) for both 60- and 100-Hz protocols (P < 0.001), whereas it remained unchanged for 20 Hz (P = 0.82). The mean torque progressively decreased thereafter at 60 Hz and 100 Hz to reach a steady state around 50% of the total time, while at 20-Hz mean torque significantly decreased only during the latest contractions (see Fig. 3). Interestingly, the declines in mean evoked torque and PtES were highly correlated for the 100-Hz protocol (r = 0.90, P < 0.01) and the 60-Hz protocol (r = 0.97, P < 0.01), but not for the 20-Hz protocol (r = 0.28).
Mechanical and electrophysiological responses to single myostimulation
A significant frequency effect was found on NMES pulse amplitude IES among the three protocols (F2,18 = 19,2, P < 0.001, ηP2 = 0.68): 18.5 ± 1.9 mA at 20 Hz, 15.1 ± 1.5 mA at 60 Hz and 12.9 ± 1.4 mA at 100 Hz, respectively. The post hoc analysis revealed that the pulse amplitude was significantly lower as the stimulation frequency was high (all P < 0.001). An interaction effect between the factors PRE–POST and frequency of stimulation was observed for the amplitudes of the initial peak twitch associated with IES, that is, PtES (F2,18 = 35.49, P < 0.001, ηP2 = 0.79). PtES was statistically lower at the higher frequency than at low frequency (all P < 0.001), ranging from 9.46 ± 1.42 N·m at 20 Hz to 6.87 ± 1.45 N·m at 100 Hz (Fig. 4A). PtES significantly decreased after the NMES protocol in the three conditions (all P < 0.001), by −11.6% ± 2.6% for 20 Hz, −17.9% ± 3.41% for 60 Hz, and −26.1% ± 2.6% for 100 Hz, respectively (Fig. 4A).
When analyzing electrophysiological responses associated with single myostimulation at IES (HES and MES of SOL muscle), differences were observed between the three protocols. A significant interaction effect between the factors frequency and PRE–POST was found on the sum of HES and MES (HES + MES) evoked at IES (F2,18 = 2.18, P = 0.03, ηP2 = 0.31). This sum represents the total amount of SOL electrical activation after one single NMES pulse, whether MU are activated directly (MES) or reflexively HES). In both PRE and POST measurements, HES + MES was significantly lower for 100 and 60 Hz compared with 20 Hz protocol (Fig. 4B). This sum decreased only after 60-Hz (P = 0.012) and 100-Hz protocols (P = 0.046) but not after 20 Hz (P = 0.93). This decline was correlated to the decline in mean evoked torque and the decline in PtES for 60-Hz protocol (for mean torque: r = 0.69, P < 0.05; for PtES: r = 0.68, P < 0.05) and 100-Hz protocol (for mean torque: r = 0.88, P < 0.001; for PtES: r = 0.79, P < 0.01), but not at 20 Hz for both mean torque (r = 0.24) and PtES (r = 0.29).
According to H and M amplitudes recorded at IES, the relative contribution of direct and indirect MU activation differed between protocols (see Fig. 5 for typical recordings). First, a significant interaction effect between the factors frequency and PRE–POST (F2,18 = 3.33, P = 0.012, ηP2 = 0.21) was found on MES/Mmax. Greater MES/Mmax was observed for the 20-Hz protocol compared with the two other protocols (all P < 0.001) and this ratio decreased significantly after all protocols (Fig. 4C). Second, significant interaction effect between the factors frequency and PRE–POST (F2,18 = 3.27, P = 0.013, ηP2 = 0.20) was also found on HES/Mmax. In PRE conditions, HES/Mmax was significantly greater at 100 Hz than at 20 Hz (P = 0.002) (Fig. 4D). After NMES, HES/Mmax significantly decreased after 60 Hz (PRE vs POST: P = 0.041) and 100 Hz (PRE vs POST: P = 0.036) but not after 20-Hz protocol (PRE vs POST: P = 0.96).
Pre- to post-MVC
A significant interaction effect between the factors frequency and PRE–POST (F2,18 = 8.11, P = 0.003, ηP2 = 0.47) was found on MVC torques (Fig. 6A). The MVC torque significantly decreased after NMES by 9.6% ± 3.3% for 20 Hz (PRE vs POST: P = 0.012), 10.7% ± 3.2% for 60 Hz (PRE vs POST: P = 0.007) and 16.3% ± 2.7% for 100 Hz protocol (PRE vs POST: P < 0.001). While PRE MVC were not statistically different between protocols (all P > 0.90), POST MVC torque was significantly lower at 100 Hz than at 20 Hz (P < 0.001) and 60 Hz (P < 0.001). In the three conditions, the decrease in MVC torque was significantly correlated (P < 0.01) with the decrease in mean evoked torque (Fig. 3).
The electromyographic activities (RMS/Msup) analyzed during the PRE and the POST MVC evolved differently for the three protocols. Indeed, significant interaction between the factors frequency and PRE–POST were found on RMS/Mmax of the SOL muscle (F2,18 = 6.78, P = 0.006, ηP2 = 0.43), MG muscle (F2,18 = 4.76, P = 0.0218, ηP2 = 0.35) and LG muscle (F2,18 = 5.46, P = 0.013, ηP2 = 0.37). The post hoc analysis revealed that for the 20-Hz protocol, no significant changes were found between PRE (SOL: 0.024 ± 0.004, MG: 0.028 ± 0.004, LG: 0.036 ± 0.003) and POST values (SOL: 0.023 ± 0.003, MG: 0.029 ± 0.004 and LG: 0.033 ± 0.002, respectively). However, for the 100-Hz protocol PRE RMS/Msup (SOL: 0.025 ± 0.003, MG: 0.035 ± 0.005 and LG: 0.043 ± 0.009, respectively) were significantly greater than POST values (SOL: 0.016 ± 0.002, MG: 0.026 ± 0.006 and LG: 0.037 ± 0.001). For the 60-Hz protocol, PRE ratios (SOL: 0.024 ± 0.005, MG: 0.035 ± 0.007 and LG: 0.037 ± 0.005) were also significantly greater than POST RMS/Msup (SOL: 0.020 ± 0.003, MG: 0.027 ± 0.005 and LG: 0.027 ± 0.004).
Electrophysiological responses to nerve stimulation
No main effect or interaction were found for maximal superimposed M-wave (Msup) amplitude for the three muscles (SOL: F2,18 = 1.93, P = 0.17, ηP2 = 0.18; MG: F2,18 = 0.13, P = 0.87, ηP2 = 0.01; LG: F2,18 = 0.009, P = 0.90, ηP2 = 0.01). However, an interaction effect was found on Hsup/Msup (SOL: F2,18 = 4.84, P < 0.001, ηP2 = 0.35; MG: F2,18 = 5.60, P = 0.005, ηP2 = 0.38; LG: F2,18 = 6.54, P = 0.002, ηP2 = 0.42) which, according to post hoc analysis significantly decreased by 37.7% ± 10.5% after 100-Hz protocol but was not altered after 20-Hz and 60-Hz protocols (Fig. 6). Similarly, an interaction effect was found on V/Msup (SOL: F2,18 = 4.51, P = 0.013, ηP2 = 0.33 ; MG: F2,18 = 4.04, P = 0.047, ηP2 = 0.31; LG: F2,18 = 6.86, P = 0.008, ηP2 = 0.43). V/Msup significantly decreased by 19.7% ± 7% after 60 Hz and by 25.8% ± 6.2% after 100-Hz protocol (PRE vs POST: P = 0.032, P = 0.023 and P = 0.026 for SOL, MG and LG respectively), but was not significantly reduced after 20-Hz protocol (−4.8% ± 1.5%, −6.0% ± 6.7%, +5.9% ± 2.6% for SOL, MG and LG, respectively). In addition, the decrease observed in SOL V/Msup PRE – POST was highly correlated to the decrease observed in MVC for 60 Hz (r = 0.83, P < 0.01) and 100 Hz (r = 0.93, P < 0.01). This last correlation was also found for both MG muscle (r = 0.70 and r = 0.91 for 60 and 100 Hz, respectively) and LG muscle (r = 0.65 and r = 0.80 for 60 and 100 Hz, respectively).
This study investigated the acute effects of NMES protocols with different frequencies (20 Hz, 60, and 100 Hz) but a similar total torque–time integral on the neuromuscular system. The main findings were: i) increasing the stimulation frequency induced greater decrease in both mean evoked torque and MVC torque, ii) NMES pulse amplitude used for high-frequency protocols (60 and 100 Hz) induced larger reflexive responses (HES) compared with a low frequency (20 Hz), iii) force decrease in high-frequency protocols was accompanied by a substantial reduction in both direct and indirect responses of the muscle, as well as in the neural activation during MVC, as evidenced by a decrease in both H- and V-wave amplitudes.
Evoked and voluntary torque losses
In the present study, particular care was taken to match the total TTI between fatiguing protocols. TTI was shown to reflect the muscle energy expenditure of intermittent or sustained isometric contractions (5,38). It has been shown that sustained isometric voluntary contractions at different force levels but with a similar total TTI have similar alterations of the maximal voluntary force production capacity (37). Regarding electrically evoked contraction, a standardized TTI across several protocols may induce similar metabolic demands regardless of the modulation of stimulation parameters such as the frequency, as shown in animal experiments (17). In humans, regardless of the total muscle work, it has been shown that different NMES protocols may induce similar changes in muscle metabolism, that is, intracellular pH or phosphocreatine depletion assessed through phosphore magnetic resonance spectroscopy (32). Thus, different frequencies of NMES would likely have the same effects at the intracellular muscle level. Here, the discrepancies observed between voluntary and evoked force loss after the three NMES protocols may involve, in addition to peripheral factors, central factors. In contrast with previous studies using a fixed number of NMES contractions among protocols (e.g., 20), a greater decrease of MVC was found here after high-frequency NMES than after mid- and low frequencies. Thus, despite the fact that different NMES protocols may induce similar changes in muscle metabolism, the neural mechanisms involved may differ between evoked and voluntary torque losses. More specifically, different muscle activation pathways could be a possible mechanism responsible for such differences in neuromuscular fatigue among the protocols.
Stimulation frequencies and muscle activation
To reach the same initial level of evoked force during the 1-s NMES trains, the pulse amplitude had to be lowered as the frequency increased. This reduction in NMES pulse amplitude led to a lower initial amount of total mechanical response of the solicited MU (4) as assessed by the amplitude of the twitch evoked at IES, that is, PtES (34–36), which was lower during the protocol using the highest frequency. This was also accompanied, at least for the SOL muscle, by a greater reduction in the total electrophysiological amplitude response, quantified by HES + MES sum, at high compared with low frequency. Such variations in mechanical and electrical responses can be mostly attributed to different amount of recruited MU among protocols, that is, more MU were recruited at the NMES pulse amplitude used for low-frequency protocols.
As observed in previous studies (34,47), a greater loss of evoked force was observed at high stimulation frequencies. Here, the number of evoked trains at high frequency had to be increased to reach a similar TTI compared with low frequencies protocols. Therefore, despite a similar total TTI between protocols, a low-stimulation frequency associated with a greater muscle activation, as shown by the greater mechanical (PtES) and electrophysiological (HES + MES) responses, seems less fatiguing than lower muscle activation associated with high frequency.
The decline of evoked force was correlated to the decrease in PtES after 60–Hz and 100–Hz protocols, whereas no significant correlation was observed for the 20–Hz protocol. The decline in twitch amplitude at high frequencies was accompanied by a decrease in HES + MES, suggesting a reduction in the total amount of solicited MU (32). This loss reduction could be related to axonal threshold changes with repeated NMES trains (7,13,35). Indeed, the combination of high stimulation frequency and low pulse amplitude may have induced a loss of solicited MU which have an excitability threshold close to the pulse amplitude, thus contributing to the greater evoked torque decrease (33) and therefore an increase in the number of evoked trains needed to reach a similar total TTI.
Furthermore, the relative contribution of the HES and MES waves to the sum amplitudes among the three NMES protocols showed that direct and indirect pathways of MU recruitment were differently involved. For instance, the HES observed at 100 Hz was greater than that 20 Hz. This strongly suggests that the pulse amplitude used for the 100–Hz protocol will recruit MU via an indirect pathway, that is, inducing a reflexive activation of the Mus; while the one used for 20–Hz protocol preferentially activates MU via an efferent pathway. The great reflexive activation at 100 Hz can be mainly attributed to the weakness of the stimulation used, which activates preferentially Ia afferents due to their lower excitability threshold than motor axons (27). A strong correlation between the decline in evoked force and spinal excitability has already been demonstrated during high-frequency protocols (21). In contrast, the greater MES observed at 20–Hz protocol indicated that a greater proportion of motor axons were activated at the pulse amplitude used during low-frequency protocol, leading to more direct activation of the MU and a greater antidromic collision decreasing the reflexive activation of the MU (4).
The decrease of MES observed after all three protocols may therefore be attributed to a decrease in motor axons excitability while the decrease of HES observed after 60 Hz and even more pronounced after 100 Hz may involve several other mechanisms, such as homosynaptic postactivation depression (15) and/or presynaptic inhibition processes, mediated by primary afferent depolarization interneurons affecting the Ia afferent-Mn efficacy transmission (2). Whatever the mechanisms, that is, axonal excitability threshold changes and/or Ia-Mn synaptic transmission failure, increasing the stimulation frequency seems to decrease the number of activated MU after NMES.
Neural alterations with NMES-induced fatigue
Although the analysis of electrophysiological responses associated with myostimulation (HES and MES) would help to identify the neural mechanisms associated with the decrease in evoked force, analyzing the electromyographic activities and the responses to superimposed peripheral nerve stimulation allowed to identify the mechanisms involved in MVC decrease.
Above all, only the 60- and 100-Hz protocols induced a substantial decrease in electromyographic activity—assessed through RMS/Msup ratios, suggesting an impact on the neuromuscular function at these frequencies. The decrease in EMG activity may be partly explained by modulations originating at the spinal level. Indeed, a significant change in Hsup/Msup was observed only after the 100-Hz protocol. Previous studies showed that such long-pulse, high-frequency protocol involved the spinal loop during motor unit recruitment (6,12,29,47). These authors attributed the decrease of nerve-evoked H-reflex amplitude to factors originated at the junction between Ia afferent and alpha motoneurons. For instance, such change in H-reflex amplitude may be attributed to changes in motoneuronal excitability, presynaptic and/or postsynaptic inhibitions of the Ia afferent terminals (48). Similar to HES, the depression of the superimposed H-reflex amplitude may also originate from homosynaptic postactivation depression, which characterizes a decrease in neurotransmitters’ release from Ia afferent terminals after a repetitive activation (24).
The use of long pulse duration (1 ms) is known to initially recruit the sensory axons, because of a longer strength-duration and lower rheobase than the motor axons (46). Therefore, all NMES protocols performed in our study were susceptible to induce spinal excitability changes. However, despite the fact that Ia afferents were also theoretically activated during low-frequency and midfrequency protocols, H-reflexes recorded POST MVC (Hsup/Msup) did not decrease compared with those recorded PRE MVC. First, the stimulation frequency of Ia afferents may be insufficient to induce a decrease of the Ia-alpha motoneuron transmission efficiency. Second, during low-frequency protocols, supraspinal mechanisms involved during voluntary contraction might also compensate for the spinal excitability changes observed at POST MVC measurements.
The decrease in both V-wave and RMS/Mmax after the 60- and 100-Hz protocols suggested a central contribution to the reduction in MVC. The V-wave, originating from a collision in motor axons between the antidromic volleys from maximal nerve stimulation and the volley from voluntary neural drive, is a tool commonly used to assess the amount of descending command (19). Its modulation after electrical stimulation protocols could therefore be attributed to changes in motoneuron firing frequency and/or recruitment (16), originating from the descending neural drive (1). Both mechanisms induce different probability for the descending neural drive to collide with antidromic impulse from the stimulation. Previous studies showed that high motor centers in the cortex can be activated during electrically evoked contraction in upper and lower limb muscles (17,33), and that NMES can have acute effects on specific cortical pathways, such as interhemispheric inhibition (21). Indeed, it has been shown that corticospinal excitability can be increased during NMES (26). Furthermore, variations in V-wave amplitude were shown to be highly correlated to motor evoked potentials changes (19). Thus, in addition to the significant correlation between V-wave amplitude decrease and MVC torque decline, the present findings suggest that high-frequency protocols induced substantial alterations at supraspinal level.
Nevertheless, after the 100-Hz protocol, the higher decrease in voluntary neural drive (V-wave amplitude) did not allow to compensate for the axonal excitability changes, as shown by the significant decrease in Hsup/Msup. On the contrary, 60-Hz protocol induced a decrease in V-wave amplitude, as well as in RMS/Msup, but not in Hsup amplitude. These findings suggest that both spinal and supraspinal mechanisms may account for the decrease in MVC torque for the 100-Hz protocol whereas only supraspinal mechanisms were affected after 60-Hz protocol.
To summarize, all protocols likely activated Ia afferent pathway, but the highest frequencies, combined with the lowest pulse amplitude, induced the greatest solicitation of the spinal loop. This solicitation of the sensory feedback at high frequencies involved changes at spinal and supraspinal levels. In addition to the similarities between the modulations of nerve-evoked reflexes (Hsup) and muscle evoked responses (HES), the positive correlations between the modulations of evoked and MVC force after the three protocols (Fig. 3) suggested a link between the mechanisms involved during NMES and those involved during MVC force loss. However, these mechanisms cannot be exhaustive, because during voluntary contraction, as well as during NMES, other mechanisms may be involved in the decrease of maximal force generation capacity such as postsynaptic circuitry or changes in background motoneuron excitability. Further experiments would help identifying postsynaptic mechanisms involved in voluntary force loss, such as assessing recurrent inhibition level (25).
Optimizing the stimulation parameters to provide neuromuscular gains through NMES sessions may help to individualize training and/or rehabilitation programs. Long-pulse high-frequency NMES has been shown to improve paretic limb performances in poststroke patients (8). Based on the present results and those of the literature, we suggest that the use of repeated long pulse duration and high-frequency NMES trains could also be worthwhile for neurological disorders such as incomplete spinal cord injuries (9). Indeed, activating afferent pathway with NMES could be of interest for patients who have paresis (43).
The present study also showed that the use of high-frequency NMES could lead to high neuromuscular impairment for a lesser pulse amplitude than at low frequencies (34,47). On one hand, this is an advantage given that the discomfort induced by high pulse amplitude stimulation is frequently considered as a drawback that impairs the utilization of NMES (31). But on the other hand, using high-frequency protocol may also lead to exaggerated muscle fatigue. If the aim is to minimize muscle fatigue, low frequencies may be therefore recommended. Interestingly, mid-frequency of stimulation (e.g., 60 Hz) could represent a compromise between the great peripheral solicitation induced by low-frequency and high-pulse amplitude and the great central contribution induced by high-frequency and low-pulse amplitude.
Regarding methodological considerations, the present study raised the importance to standardize the total muscle work in experimental settings. Indeed, given the interindividual and intraindividual variability of the force evoked through such high frequencies and long pulse duration NMES (47), we recommend to carefully set the number of duty cycles to reach the desired amount of total muscle work.
Finally, further studies are required to determine if the present findings can be extended to paralyzed muscles and overall to other lower and upper limb muscles. Indeed, because the amount of afferent fibers may vary from one muscle group to another (44), the impact of NMES on indirect muscle activation could be different. Nevertheless, the similar results found here between SOL and gastrocnemii muscles, that have different fiber typology and afferent proportion (44), is encouraging to generalize our findings.
By matching the total torque time for NMES protocols using different frequencies (20, 60, and 100 Hz), our study avoided confounding factors that could influence the mechanisms of muscle fatigue. When the total muscle work load was standardized, the neuromuscular fatigue was greater with high-frequency and lower-pulse amplitude protocols. The pulse amplitude corresponding to the 20-Hz protocol seemed rather to activate the muscles directly through motor axon depolarization, as shown by larger MES. To generate a given submaximal force level with NMES-evoked contractions, low frequencies of stimulation involve greater pulse amplitude than high frequencies, leading to greater direct contribution. On the contrary, the pulse amplitude used for 60 and 100 Hz protocols likely activated the muscles both directly and indirectly, through Ia afferents depolarization and reflexive pathways (larger HES). Regardless of their initial values, the different evolution of SOL HES and MES after each protocol suggests that increasing stimulation frequency induces greater decrease in the number of recruited MU, probably linked to changes in axonal excitability under the stimulation electrodes or to alterations of the transmission at the Ia-Alpha synaptic level. Low frequencies (20 Hz) induced alterations mainly at the muscle level, whereas higher frequencies (60–100 Hz) rather induced modulations at both spinal and supraspinal levels. The present study demonstrates that there is not a unique gold standard NMES protocol, but that stimulation parameters may be modulated according to the context in which electrical stimulation is used.
This research work was supported by the Région de Bourgogne (Contract 9201AAO050S02953) and the Fonds Européen de Développement Régional (FEDER).
No conflicts of interest, financial or otherwise, are declared by the authors. The results of the present study do not constitute endorsement by the American College of Sports Medicine. The authors declare that the results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation.
1. Aagaard P, Simonsen EB, Andersen JL, Magnusson P, Dyhre-Poulsen P. Neural adaptation to resistance training: changes in evoked V-wave
and H-reflex responses. J Appl Physiol
2. Achache V, Roche N, Lamy JC, et al. Transmission within several spinal pathways in adults with cerebral palsy. Brain
. 2010;133(Pt 5):1470–83.
3. Bax L, Staes F, Verhagen A. Does neuromuscular electrical stimulation strengthen the quadriceps femoris? A systematic review of randomised controlled trials. Sports Med
4. Bergquist AJ, Wiest MJ, Collins DF. Motor unit recruitment when neuromuscular electrical stimulation is applied over a nerve trunk compared with a muscle belly: quadriceps femoris. J Appl Physiol
5. Bergström M, Hultman E. Energy cost and fatigue during intermittent electrical stimulation of human skeletal muscle. J Appl Physiol
6. Blouin JS, Walsh LD, Nickolls P, Gandevia SC. High-frequency submaximal stimulation over muscle evokes centrally generated forces in human upper limb skeletal muscles. J Appl Physiol
7. Bostock H, Bergmans J. Post-tetanic excitability changes and ectopic discharges in a human motor axon. Brain
. 1994;117(Pt 5):913–28.
8. Clair-Auger JM, Collins DF, Dewald JP. The effects of wide pulse neuromuscular electrical stimulation on elbow flexion torque in individuals with chronic hemiparetic stroke. Clin Neurophysiol
9. Clair-Auger JM, Lagerquist O, Collins DF. Depression and recovery of reflex amplitude during electrical stimulation after spinal cord injury. Clin Neurophysiol
10. Collins DF. Central contributions to contractions evoked by tetanic neuromuscular electrical stimulation. Exerc Sport Sci Rev
11. Collins DF, Burke D, Gandevia SC. Large involuntary forces consistent with plateau-like behavior of human motoneurons. J Neurosci
12. Collins DF, Burke D, Gandevia SC. Sustained contractions produced by plateau-like behaviour in human motoneurones. J Physiol
. 2002;538(Pt 1):289–301.
13. Doix AC, Matkowski B, Martin A, Roeleveld K, Colson SS. Effect of neuromuscular electrical stimulation intensity over the tibial nerve trunk on triceps surae muscle fatigue. Eur J Appl Physiol
14. Gibson JN, Smith K, Rennie MJ. Prevention of disuse muscle atrophy by means of electrical stimulation: maintenance of protein synthesis. Lancet
15. Gondin J, Cozzone PJ, Bendahan D. Is high-frequency neuromuscular electrical stimulation a suitable tool for muscle performance improvement in both healthy humans and athletes? Eur J Appl Physiol
16. Gondin J, Duclay J, Martin A. Soleus- and gastrocnemii-evoked V-wave
responses increase after neuromuscular electrical stimulation training. J Neurophysiol
17. Gondin J, Giannesini B, Vilmen C, et al. Effects of stimulation frequency and pulse duration on fatigue and metabolic cost during a single bout of neuromuscular electrical stimulation. Muscle Nerve
18. Gregory CM, Dixon W, Bickel CS. Impact of varying pulse frequency and duration on muscle torque production and fatigue. Muscle Nerve
19. Grosprêtre S, Martin A. Conditioning effect of transcranial magnetic stimulation evoking motor-evoked potential on V-wave
response. Physiol Rep
. 2014;2(12): e2191.
20. Grosprêtre S, Papaxanthis C, Martin A. Modulation of spinal excitability by a sub-threshold stimulation of M1 area during muscle lengthening. Neuroscience
21. Gueugneau N, Grosprêtre S, Stapley P, Lepers R. High-frequency neuromuscular electrical stimulation modulates interhemispheric inhibition in healthy humans. J Neurophysiol
22. Horstman AM, Beltman MJ, Gerrits KH, et al. Intrinsic muscle strength and voluntary activation of both lower limbs and functional performance after stroke. Clin Physiol Funct Imaging
23. Hortobágyi T, Maffiuletti NA. Neural adaptations to electrical stimulation strength training. Eur J Appl Physiol
24. Hultborn H, Illert M, Nielsen J, Paul A, Ballegaard M, Wiese H. On the mechanism of the post-activation depression of the H-reflex in human subjects. Exp Brain Res
25. Hultborn H, Pierrot-Deseilligny E. Changes in recurrent inhibition during voluntary soleus contractions in man studied by an H-reflex technique. J Physiol
26. Khaslavskaia S, Ladouceur M, Sinkjaer T. Increase in tibialis anterior motor cortex excitability following repetitive electrical stimulation of the common peroneal nerve. Exp Brain Res
27. Kiernan MC, Lin CS, Burke D. Differences in activity-dependent hyperpolarization in human sensory and motor axons. J Physiol
. 2004;558(Pt 1):341–9.
28. Lagerquist O, Collins DF. Influence of stimulus pulse width on M-waves, H-reflexes, and torque during tetanic low-intensity neuromuscular stimulation. Muscle Nerve
29. Lagerquist O, Walsh LD, Blouin JS, Collins DF, Gandevia SC. Effect of a peripheral nerve block on torque produced by repetitive electrical stimulation. J Appl Physiol
30. Levine TR, Hullett CR. Eta squared, partial eta squared, and misreporting of effect size in communication research. Hum Commun Res
31. Maffiuletti NA. Physiological and methodological considerations for the use of neuromuscular electrical stimulation. Eur J Appl Physiol
32. Martin A, Grosprêtre S, Vilmen C, et al. The etiology of muscle fatigue differs between two electrical stimulation protocols. Med Sci Sports Exerc
33. Matkowski B, Lepers R, Martin A. Torque decrease during submaximal evoked contractions of the quadriceps muscle is linked not only to muscle fatigue. J Appl Physiol
34. Neyroud D, Dodd D, Gondin J, Maffiuletti NA, Kayser B, Place N. Wide-pulse-high-frequency neuromuscular stimulation of triceps surae induces greater muscle fatigue compared with conventional stimulation. J Appl Physiol
35. Papaiordanidou M, Stevenot JD, Mustacchi V, Vanoncini M, Martin A. Electrically induced torque decrease reflects more than muscle fatigue. Muscle Nerve
36. Regina Dias Da Silva S, Neyroud D, Maffiuletti NA, Gondin J, Place N. Twitch potentiation induced by two different modalities of neuromuscular electrical stimulation: implications for motor unit recruitment. Muscle Nerve
37. Rozand V, Cattagni T, Theurel J, Martin A, Lepers R. Neuromuscular fatigue following isometric contractions with similar torque time integral. Int J Sports Med
38. Russ DW, Elliott MA, Vandenborne K, Walter GA, Binder-Macleod SA. Metabolic costs of isometric force generation and maintenance of human skeletal muscle. Am J Physiol Endocrinol Metab
39. Schieppati M. The Hoffmann reflex: a means of assessing spinal reflex excitability and its descending control in man. Prog Neurobiol
40. Sheffler LR, Chae J. Neuromuscular electrical stimulation in neurorehabilitation. Muscle Nerve
41. Snyder-Mackler L, Ladin Z, Schepsis AA, Young JC. Electrical stimulation of the thigh muscles after reconstruction of the anterior cruciate ligament. Effects of electrically elicited contraction of the quadriceps femoris and hamstring muscles on gait and on strength of the thigh muscles. J Bone Joint Surg Am
42. Theurel J, Lepers R, Pardon L, Maffiuletti NA. Differences in cardiorespiratory and neuromuscular responses between voluntary and stimulated contractions of the quadriceps femoris muscle. Respir Physiol Neurobiol
43. Thompson CK, Lewek MD, Jayaraman A, Hornby TG. Central excitability contributes to supramaximal volitional contractions in human incomplete spinal cord injury. J Physiol
. 2011;589(Pt 15):3739–52.
44. Tucker KJ, Tuncer M, Türker KS. A review of the H-reflex and M-wave in the human triceps surae. Hum Mov Sci
45. Vanderthommen M, Duteil S, Wary C, et al. A comparison of voluntary and electrically induced contractions by interleaved 1H- and 31P-NMRS in humans. J Appl Physiol
46. Veale JL, Mark RF, Rees S. Differential sensitivity of motor and sensory fibres in human ulnar nerve. J Neurol Neurosurg Psychiatry
47. Wegrzyk J, Fouré A, Vilmen C, et al. Extra Forces induced by wide-pulse, high-frequency electrical stimulation: occurrence, magnitude, variability and underlying mechanisms. Clin Neurophysiol
48. Zehr EP. Considerations for use of the Hoffmann reflex in exercise studies. Eur J Appl Physiol