Metabolic and Structural Changes in Lower-Limb Skeletal Muscle Following Neuromuscular Electrical Stimulation: A Systematic Review

Background Transcutaneous neuromuscular electrical stimulation (NMES) can be applied as a complementary intervention to regular exercise training programs. A distinction can be made between high-frequency (HF) NMES and low-frequency (LF) NMES. In order to increase understanding of the mechanisms of functional improvements following NMES, the purpose of this study was to systematically review changes in enzyme activity, muscle fiber type composition and muscle fiber size in human lower-limb skeletal muscles following only NMES. Methods Trials were collected up to march 2012 and were identified by searching the Medline/PubMed, EMBASE, Cochrane Central Register of Controlled Trials, CINAHL and The Physical Therapy Evidence Database (PEDro) databases and reference lists. 18 trials were reviewed in detail: 8 trials studied changes in enzyme activities, 7 trials studied changes in muscle fiber type composition and 14 trials studied changes in muscle fiber size following NMES. Results The methodological quality generally was poor, and the heterogeneity in study design, study population, NMES features and outcome parameters prohibited the use of meta-analysis. Most of the LF-NMES studies reported significant increases in oxidative enzyme activity, while the results concerning changes in muscle fiber composition and muscle size were conflicting. HF-NMES significantly increased muscle size in 50% of the studies. Conclusion NMES seems to be a training modality resulting in changes in oxidative enzyme activity, skeletal muscle fiber type and skeletal muscle fiber size. However, considering the small sample sizes, the variance in study populations, the non-randomized controlled study designs, the variance in primary outcomes, and the large heterogeneity in NMES protocols, it is difficult to draw definitive conclusions about the effects of stimulation frequencies on muscular changes.


Introduction
Regular exercise training programs consist of a combination of aerobic and strengthening exercises for developing and maintaining muscular endurance and strength, respectively [1]. Indeed, combined training modalities result in improvements in body composition and cardiorespiratory fitness. These improvements can partially be explained by intramuscular changes, such as an increased enzyme activity and an increased muscle fiber size [2,3].
These intramuscular changes are dependent on the type of exercise training. Generally, aerobic exercise training results in increased levels of oxidative enzymes [4] and only a marginal increase in percentage type I fibers [5]; whereas resistance training results in increased levels of glycolytic enzymes [6] and an increase in percentage and size of type II fibers [7][8][9][10].
Neuromuscular electrical stimulation (NMES) can be applied as an complement intervention to voluntary exercise training [11]. NMES involves the application of an electric current through electrodes placed on the skin over the targeted muscles, thereby depolarizing motor endplates via the motor nerve and, in turn, inducing skeletal muscle contractions [12,13]. NMES is composed of stimulation-rest cycles situated in regard to muscle motor points [14]. In contrast to voluntary muscle actions, NMES activates the muscle to a greater extent under identical technical conditions [15]. At identical levels of workload (10% of the quadriceps maximum isometric voluntary torque), the muscle reaches higher values in blood flow and oxygen consumption during NMES compared with voluntary contractions [15]. Moreover, a single session of NMES is sufficient to stimulate molecular-level responses, which are indicative of the initiation of myogenic processes in skeletal muscle, while an additional NMES session (a total of 14 minutes spread over 2 days), was sufficient to induce an increase in the concentration of total ribonucleic acid (RNA) [16], most likely representing an increase in muscle protein synthesis. There is sufficient evidence that NMES induced contractions differ physiologically compared to voluntary contractions [17]. In human studies contradictory findings on motor unit recruitment order have been found [18]. Some studies suggest preferential or selective activation of fast motor units with NMES [19,20], whereas others suggest that motor unit recruitment during NMES reflects a non-selective, spatially fixed, and temporally synchronous pattern rather than in a reversal of the physiological voluntary recruitment order [17]. These diverse results could have been related to differences in protocols and stimulated muscles [14].
In daily clinical practice, lower-limb NMES improves skeletal muscle mass and function, exercise capacity and health status [21][22][23], particularly in subjects who are unable to perform or complete volitional exercise training programs. Therefore, NMES may be valuable in dyspneic and deconditioned patients with chronic organ failure due to the low metabolic load on the impaired cardio-respiratory system [23,24].
NMES training sessions generally last 10-30 minutes during a 4-to 5-week period that involves 20-25 sessions to increase peripheral muscle function [14]. 2 types of NMES frequencies can be distinguished: high-frequency NMES (HF-NMES, .50 Hertz); and low-frequency NMES (LF-NMES, ,20 Hertz) [13,14,[25][26][27]. Frequencies of 50 Hertz and above induce a fused tetanus [28,29] and generate higher torques than low frequencies [30]. The mechanisms by which NMES results in increased muscle strength or endurance are poorly understood. In isolated muscles in rats HF-NMES induces anabolic processes similar to resistance training (e.g. increased PKB-TSC2-mTor and protein synthesis) and LF-NMES similar to endurance training (AMPK-PG C1a activation) [31]. In humans, it is unknown which stimulation frequency is involved in the specific physiological and biochemical processes [13].
To date, narrative reviews have been published about the effects of NMES on gains in muscle performance, activation of motor units and/or muscle energetics [14,26,32,33]. However, there is a broad diversity in NMES programmes, populations and outcomes which makes it difficult to interpret the conclusions. The effects of NMES on intramuscular changes have not been systematically reviewed yet. The purpose of this study is to systematically review changes in enzyme activity, muscle fiber type composition and muscle fiber size in human lower-limbs following a NMES programme. A distinction will be made between HF-NMES and LF-NMES, as well as in healthy volunteers, patients with chronic organ failure or orthopedic problems. Our hypothesis is that LF-NMES (,20 Hz) will primarily induce endurance training-like adaptations such as increased oxidative enzyme capacity and fiber type I proportion, whereas HF-NMES (.50 Hz) will primarily induce adaptations comparable to resistance training such as an increased glycolytic capacity, fiber type II proportion and muscle fiber size. Safety and the methodological quality of the trials will also be assessed.

Data sources and searches
We followed the procedures described in the PRISMA statement for reporting systematic reviews (online supplement) [34]. A broad computerized literature search was performed to identify relevant trials reported in the English language. We used the following databases: Medline/PubMed (from 1966), EMBASE (from 1974), Cochrane Central Register of Controlled Trials (from 1898), CINAHL (from 1982), and The Physical Therapy Evidence Database (PEDro) (from 1982). Trials were collected up to March 2012. Search terms were combinations of keywords related to neuromuscular electrical stimulation, lower-limb muscles, muscle mass and muscle metabolic profile. The exact search algorithm for Medline/PubMed can be found in Appendix 1. Similar search algorithms were used for the other databases. In addition, reference lists and citations of original articles were also scanned to identify additional articles that may contain information on the topic of interest.

Data extraction
A pre-designed data abstraction form was used to obtain data on trial design and relevant results. For each article, characteristics of the study subjects were noted: a) the condition of the study population (healthy or primary diagnosis), gender and age; b) study design and NMES features (i.e. pulse duration, pulse frequency, duty cycle and pulse amplitude of the used current, training intensity, session time and duration in weeks); c) outcome measures, such as muscle enzyme activity (i.e. changes in oxidative and glycolytic enzymes), changes in muscle fiber type, changes in muscle fiber size and d) safety.

Article selection
Articles were used for further analyses when they met the following eligibility criteria: Types of studies. Randomized controlled trials (RCTs), controlled clinical trials (CCTs) and clinical trials were included. A priori, congress abstracts, reviews, editorials and case reports were considered ineligible.
Study subjects. Included were trials in which human lowerlimb muscles were electrically stimulated. Reasons for exclusion were studies with subjects suffering from neurological disorders (e.g., hemiplegia or lesion of the spine) or smooth muscle problems (e.g., period of bladder dysfunction) Types of interventions. Included were trials in which the muscles were stimulated transcutaneously at the muscle motor points with a stimulation frequency of ,20 Hertz (LF-NMES) or .50 Hertz (HF-NMES), a minimum total session time of 120 minutes, a minimum of 3 sessions per week in a minimum of 4 weeks [13,14,18]. Trials were not excluded based on pulse duration, pulse amplitude or training intensity.
Types of outcome measures. In the reviewed publications the outcome measures were muscular activities of enzymes involved in oxidative or glycolytic energy metabolism, changes in fiber type composition and/or muscle fiber size following NMES.

Assessment of methodological quality
The methodological quality of the identified trials was scored using the PEDro scale and is based on the Delphi list and ''expert consensus'' [35]. The PEDro scale consists of 11 criteria which receives either a ''yes'' or a ''no''. Criterion 1 ('Eligibility criteria') is not used in the calculation of the PEDro score. All ''yes'' scores were summed resulting in a maximum score of 10 points [35]. A k coefficient was used to measure the level of interrater reliability, using a method for comparing the level of reliability with categorical data along with their respective 95% confidence intervals [36]. Consensus was sought in case of disagreement. Trials with a PEDro score of $6 points were classified as ''highquality trials'', while trials with a PEDro score #5 points were classified as ''low-quality trials'' [37].

Data analysis
The use of meta-analytic techniques for data-pooling was not possible, because of the heterogeneity in study types, study populations, wide diversity in NMES protocols (e.g., frequency, pulse duration, session time, total number of sessions) and/or outcome parameters (e.g., activity of different enzymes). Also, the technique for measuring muscle fiber size differed among the included studies. Therefore, the present authors were only able to systematically review the available peer-reviewed literature and to critically appraise the methodological quality and the overall findings.
All trials used biphasic impulse current forms ranging from 8 to 20 and 50 to 120 Hertz. Pulse duration, not reported in 1 study [53], ranged between 200 and 700 ms. Duty cycle, not reported in 1 study [53], ranged between 3 seconds on, 30 seconds off to 55 seconds on, 2 seconds off. Pulse amplitude, not reported in 1 study [39], varied between 10 mA until the individual's maximum level of toleration. Session time varied between 10 minutes and 8 hours, 1 to 2 times a day. The total number of sessions varied between 12 and 140 (table 2) between 4 and 10 weeks. The total session time ranged from 2 to 384 hours.

Safety
Safety was not reported in 13 trials. In 3 trials [38,47,49] no relevant side effects or adverse events were reported. Only once [49] a delayed onset muscle soreness was reported and one trial explicitly reported the absence of serious discomfort in the stimulated subjects [41]. Finally, in 1 trial [54] 1 study subject withdrew because of discomfort during NMES.

Changes in oxidative enzymes in healthy volunteers
Levels of oxidative enzymes generally increased following LF-NMES (table 3) and following HF-NMES (table 4).
Isocitrate dehydrogenase. Isocitrate dehydrogenase, another enzyme that participates in the tricarboxylic acid cycle, increased significantly following HF-NMES compared to baseline [40].
Enoyl CoA hydratase. Enoyl CoA hydratase, an enzyme that participates in the ß-oxidation of fatty acids, increased significantly following HF-NMES compared to baseline [40].
NADH-ubiquinone oxidoreductase. NADH-ubiquinone oxidoreductase, complex I of the electron transport chain, increased significantly following HF-NMES compared to baseline [40].
Succinate dehydrogenase. Succinate dehydrogenase, an enzyme that participates in both the tricarboxylic acid cycle and in complex II of the electron transport chain, increased significantly following HF-NMES compared to baseline and increased 16% compared to controls [48].
Ubiquinol-cytochrome c reductase. Ubiquinol cyt C reductase, complex III of the electron transport chain, increased significantly following HF-NMES compared to baseline [40].
Pyruvate dehydrogenase. Pyruvate dehydrogenase, an enzyme which contributes to linking the glycolysis metabolic pathway to the citric acid cycle and releasing energy via NADH, increased significantly following HF-NMES compared to baseline [40].

Changes in oxidative enzymes in patients with CHF or COPD
Levels of CS increased following LF-NMES (15 Hertz) compared to sham-stimulation in patients with severe CHF [47] and did not change following HF-NMES (50 Hertz) in patients with severe COPD [54].Levels of HADH did not change following HF-NMES (50 Hertz) in patients with severe COPD [54].

Changes in glycolytic enzymes in healthy volunteers
Levels of glycolytic enzymes generally did not change or decreased following LF-NMES or HF-NMES respectively (tables 3 and 4).
Phosphofructokinase (PFK). Levels of PFK, a glycolytic enzyme that catalyses the phosphorylation of fructose phosphate, decreased or did not change compared to baseline following LF-NMES or HF-NMES (variation from baseline was between 211 and 0%) [39,40,53].

Glyceraldehyde
3-phosphate dehydrogenase (GAPDH). Concentrations of GAPDH, a marker enzyme of anaerobic energy metabolism by catalysing the sixth step of glycolysis, decreased significantly in 1 LF-NMES trial [46] compared with sham-stimulation (variation from baseline was -       15%). Levels of GAPDH did not change in 2 LF-NMES trials [39,53] compared to baseline. Hexokinase. Hexokinase, a key glycolytic enzyme, increased significantly in females (36%) and did not change in males following LF-NMES compared to baseline [39].
B-enolase. B-enolase, which catalyses the glycolysis of 2phosphoglycerate to phosphoenolpyruvate, did not change in a sedentary group of healthy young men following HF-NMES compared to baseline, but increased in an active group of healthy young men following HF-NMES compared to baseline [40].

Changes in glycolytic enzymes in patients with CHF
In patients with severe CHF levels of GAPDH decreased significantly in 1 LF-NMES trial [47] compared with shamstimulation (variation from baseline was 215%).
Type IIa proportions did not change in patients with CHF following LF-NMES [47] and these fiber type proportions remained unchanged compared to controls in patients with COPD following HF-NMES [54]. Type IIx fibers decreased following LF-NMES (19%) [47] and did not change following HF-NMES compared to a control group [54].
Patients with orthopedic problems. Whole muscle CSA increased significantly in 1 HF-NMES trial [55] and did not change in two other HF-NMES trials [50,51] (table 8).

Discussion
This is the first systematic review on the effects of lower-limb NMES on intramuscular changes in the human lower-limb muscles. Most of the studies reported a significant increase in oxidative enzymes following LF-NMES. There are obvious changes in skeletal muscle fiber type composition following NMES. Indeed, LF-NMES seems to increase percentage of type I and IIa fibers, whereas fiber type composition following HF-NMES shows conflicting results. Both NMES protocols showed conflicting results in changes in muscle fiber size and total muscle volume. Heterogeneity in study design, study population, NMES features and outcome parameters prohibits the use of metaanalysis.

Methodological considerations
Overall, the methodological quality of the included trials was poor (median score 4 points). None of the 18 included trials had a perfect score on the PEDro scale (table 2). In fact, only 7 trials (39%) were of high-quality. Eligibility criteria were not specified in 4 trials (22%) and a control group was lacking in 6 trials (33%). Other methodological considerations were the limited number of study subjects (n = 8 to n = 40), the low mean age (38 years) and the fact that most subjects studied were men. Elderly subjects may respond differently on anabolic training stimulus compared to younger subjects [56]. Moreover, gender-differences exist in fiber type distribution and mean CSA [57]. So, the internal and external validity of the findings of the reviewed trials were limited.

NMES protocols
NMES protocols varied tremendously among the included trials. Pulse duration, if reported, ranged between 200 and 700 ms. Pulse duration of 300-400 ms is recommended for large muscle groups, such as the quadriceps muscles and calf muscles [58]. It remains currently unknown which duty cycle is optimal for effective treatment.
While throughout the literature a wide variety of protocols are used, there seems to be at least some agreement on the use of biphasic symmetrical pulses that last between 100 and 500 ms and are delivered at a pulse rate of 10-100 Hertz. Pulse rates between 10-50 Hz are used in patients with CHF and in patients with COPD with positive improvements in exercise capacity and health status [23,59] whereas pulse rates between 50-100 Hz are mostly recommended for gains in muscle performance [14,18]. Such pulses are widely accepted as being well tolerated.
The stimulus intensity varied among the included trials, from a comfortable stimulus till maximum tolerance level. These diverse results could have been related to differences in protocols and stimulated muscles. It is strongly recommended that pulses are delivered at the highest tolerable pulse amplitude [22]. Another common procedure is to quantify isometric maximal voluntary contraction (MVC) force at the beginning of a NMES session, and subsequently express the level of each electrically elicited contraction as a percentage of the MVC force [18].
The duration of the NMES programmes varied between 10 minutes to 8 hours/day, 1 to 2 sessions/day, 3 to 7 days/week for 4 to10 weeks. The minimum total duration of the NMES in the included studies are in line with the studies which show significant improvements in peripheral muscle function [13,14].

Changes in enzyme activity following NMES
Activity of oxidative enzymes generally increased significantly following 6 weeks of LF-NMES. The increase in oxidative enzyme activity was accompanied with an improved resistance to fatigue [52,53] and an improvement in functional exercise capacity [47].
Compared with a minimum of 6 weeks of endurance cycling training [60,61], the absolute and relative improvement in CS activity after NMES is lower.
Levels of glycolytic enzymes did not change or decreased following LF-NMES. These results are comparable with endurance training in healthy young men [62] and in patients with COPD [63]. In healthy volunteers levels of oxidative enzymes increased following HF-NMES [40,48]. These adaptations are more endurance-specific. Collins and colleagues recently showed that the use of a wide pulse (1 ms), high frequency (80-100 Hertz) and a low intensity might favour the recruitment of fatigueresistant motor units (according to the Henneman's size principle) [64]. This combination of stimulation parameters could also be relevant for increasing oxidative capacity. However, they used a wide pulse and low stimulation intensity whereas the included trials in the present review [40,48] used narrow pulses and intensities at the maximum toleration level. As the consequence, the corresponding increase of oxidative enzymes is likely due to the non-selective recruitment of both type IIx and type I fibers during HF-NMES [17]. Another study limitation is that only one study (in healthy volunteers) with a very small sample size [40] studied glycolytic enzymes following HF-NMES. In the sedentary group the downregulation of glycolytic enzymes is highly consistent with the fast-to-slow MHC isoform shift as slow fibers mostly have an oxidative metabolism and type IIx fibers mostly have a glycolytic metabolism. Additionally, glycolytic enzyme content is known to increase in the order of slow oxidative, fast oxidative, glycolytic, fast glycolytic fibers [65]. However, the small sample size is too limited (n = 10) to provide an answer on the hypothesis that HF-NMES increases glycolytic capacity.
Changes in skeletal muscle fiber type composition following NMES Changes in type I and IIa fiber proportion were variable following HF-NMES. Resistance training resulted in no changes in type I fibers and an increase of type II fibers [7,66]. However, based on the results of the present systematic review changes in type IIa and type IIx fibers following regular resistance training programs [67] cannot be compared with HF-NMES. Following LF-NMES type I and IIa fibers increased. Endurance training programs also resulted in an increase of type I and type IIa fibers and a decrease of type IIx fibers [4,5,68]. However, considering the small number of studies and heterogeneity in NMES protocols and study populations it is difficult to draw relations between LF-NMES and endurance training.

Changes in muscle size following NMES
In 50% of the HF-NMES studies muscle fiber size increased significantly. The increased CSA was accompanied with an increased muscle strength [40][41][42]49]. Changes in total muscle fiber size following LF-NMES are conflicting. Differences in the LF-NMES studies which could possible explain the conflicting results are the study population (age, healthy volunteers versus patients with chronic organ failure), measurement of muscle fiber size or intensity of NMES. These results are in line with previous studies concluding that muscle fiber size increased less in subjects who performed endurance training than in strength training [69,70]. Conflicting results for the changes in muscle fiber size in HF-NMES could be related to the intensity of the training. In two trials [38,51] reporting no changes in muscle fiber size, the stimulus intensity varied from ''a comfortable stimulus'' [51] to 25 mA [38] compared with a stimulus at the maximum tolerable level in the trials with an increased muscle fiber size [40][41][42]48]. Moreover, Vivodtzev and colleagues showed that gains in muscle strength were proportional to the increase in pulse amplitude during the training program and to the final pulse amplitude of training [54]. The impact of NMES is also dependent on the training duration. It is well known that long training duration is needed to induce muscle hypertrophy [71], however the total duration time in the HF-NMES trials ranged from 2 [45] to 54 hours [50].

Recommendations
Overall, LF-NMES seems to improve oxidative phenotype (oxidative enzyme capacities, type I/IIa fibers). However, some of the results of the included trials are difficult to compare and cannot be generalized. Besides the heterogeneity in NMES protocols and study designs, the number of study populations is limited and varies among the trials, from well-trained healthy volunteers [40] to patients with severe chronic organ failure [38,54] or severe orthopedic problems [50,55]. Stimulation variables (i.e. pulse amplitude, session time and number of sessions) might have influenced the number of muscle fibers recruited during NMES, the motor unit recruitment order and the types of the recruited muscle fibers [18,72].
Therefore, future trials are needed to determine the optimal settings of NMES, such as stimulation frequency (HF-NMES or LF-NMES), session time, pulse amplitude and electrodes (number, size and location) in healthy (i.e. athletes) and in diseased people (i.e. COPD, CHF, orthopedic problems). These trials should not only study the effects of NMES versus volitional training, but also study the superimposed effects of NMES on volitional training.
Based on the results of the present systematic review, randomized controlled trials using concealed allocation, blinded therapists, blinded participants and blinded outcome assessors are recommended. Additionally, studies should focus on larger study populations, including both genders and a broad range in age. These studies should not include only healthy people but also people who are unable to perform or complete volitional exercise training programs. Finally, safety should be added to new randomized controlled trials as secondary outcome.

Conclusion
NMES seems to be a training modality resulting in changes in oxidative enzyme activity, skeletal muscle fiber type and skeletal muscle fiber size. A more formal meta-analysis would be a more rigorous way to analyze the current data, but is not possible at this time. Indeed, the small sample sizes, the variance in study populations, the non-randomized controlled study designs, the variance in primary outcomes and the large heterogeneity in NMES protocols are major methodological limitations which may limit the external validity of the current findings. Therefore, it is difficult to draw definitive conclusions about the effects of stimulation frequencies on muscular changes. This systematic review, however, will help generate discussion in the field that would lead to a consensus in study design that would permit a meta-analysis in the future.
A better understanding of metabolic and structural changes following NMES is of particular clinical interest as it will increase its applicability in specific populations who are not able to perform regular exercise training. Therefore, future well-designed, randomized controlled trials with larger study samples are needed to determine the optimal NMES settings (i.e. electrode placement, stimulation frequency and pulse amplitude) to achieve endurance or resistance training-like adaptations. The actual stimulation parameters, session time, total time and changes in NMES pulse intensity over time should be reported to enable comparisons between studies and to facilitate the further development and implementation of NMES.