INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

THE SUPERIMPOSED TWITCH TECHNIQUE is frequently used to assess the level of activation of a muscle or muscle group (e.g., Refs. 4, 12). It consists of superimposing an electrical stimulation (typically a single or double twitch) to a muscle or its nerve while the muscle is voluntarily activated (1, 3, 10). If applied properly, the superimposed electrical stimulus fully activates all motor units of a muscle, and, in case of incomplete motor unit activation, the stimulus produces an increment in force. The force produced by the electrical stimulation on top of the voluntary force is called the interpolated twitch force, often measured as the interpolated twitch torque (ITT). The ITT decreases as voluntary muscle activation increases (e.g., Refs. 4, 12). The ITT is typically normalized to the resting twitch torque, i.e., the force evoked by the same twitch applied to the resting muscle. This normalized value is referred to as muscle inhibition (MI). Others have normalized the ITT to the voluntary torque and used the ratio to estimate central activation failure (16). Using the interpolated twitch technique, the ability to fully activate a muscle group has been investigated in healthy subjects and patients with musculoskeletal problems (4, 12, 15, 21, 24, 25).

Although the superimposed twitch technique has proven to be useful in estimating the functional status of specific muscle groups, the technique has been limited by its variability. The ITT, or the corresponding MI, may easily exceed a 10-15% difference for consecutive tests in a given subject and for the same contractile conditions (e.g., Refs. 1, 10, 18). Because of this variability in ITT, repeat measurements are necessary, comparisons between groups require a large number of subjects, and interpretation of individual data is virtually impossible. Therefore, understanding the source of the variability in ITTs is necessary, and being able to reduce these variations would be of scientific and clinical importance.

The large range in the ITT for given experimental conditions could be caused by a "random measurement error" or by variations that are an inherent part of the interpolated twitch technique. For a variety of reasons, we believe the latter to be true. the application of an electrical twitch to a stimulation pulse train produces an additional force (14), provided the pulse train frequency is submaximal (below ~50 Hz, see Fig. 2). It is also known that the magnitude of the ITT differs as a function of the timing of the superimposed twitch in relation to the pulses of the ongoing stimulation train. For example, a superimposed twitch that occurs within 5-10 ms of a pulse from the stimulation train produces a nonlinear increase in ITT because of the so-called catchlike property or doublet effect (cf. Refs. 2, 8, 26) compared with a superimposed twitch that does not produce the doublet effect. During a voluntary contraction, hundreds of motor units may be activated in a muscle or muscle group (19). The occurrence of the superimposed twitch constitutes a stochastic event relative to the stimulation trains of the voluntary contraction. Therefore, it is reasonable to hypothesize that the variations in the ITT are primarily a reflection of the stochastic nature of the superimposed twitch. If this hypothesis is correct, two conditions must be satisfied: 1) the ITT must vary systematically as a function of the timing of the superimposed single twitch in relation to a stimulation train, and 2) the variability of the ITT must decrease as the single twitch is replaced by two or more twitches that follow each other in short succession (i.e., 5-10 ms).

The purpose of the study was to perform two specific experiments aimed at testing whether these two conditions are satisfied.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Two experiments were performed to test the proposed hypotheses. All experimental protocols were approved by the Conjoint Medical Ethics Committee of the University of Calgary. Subjects gave written, informed consent for participation.

Experiment 1

Purpose. The first experiment was aimed at assessing the magnitude of the ITT as a function of timing of the superimposed twitch. This was tested by electrically stimulating the triceps surae group with a train and superimposing a single twitch at different time points with respect to the train firing frequency. Electrical stimulation was required to control precisely and repeatedly the superposition of a single twitch on an ongoing excitation train.

Subjects. Twelve healthy subjects, seven men and five women with a mean age of 30.8 ± 6.7 (SD) yr, height of 174 ± 13 cm, and mass of 69 ± 13 kg, participated in the first experiment. All subjects were moderately active and were recruited from among members of the Faculty of Kinesiology.

Force measurements. Subjects were seated on a device designed to measure isometric ankle torque at different joint angles (17). A footplate that rotates about a plantar-dorsiflexion axis was aligned with the medial malleolus of the left leg. The knee joint was flexed at 90°, and the leg was stabilized with pads over the distal thigh and proximal tibia. Subjects wore their normal exercise shoes. The foot was rigidly fixed to the footplate at 10° of dorsiflexion. This angle was chosen to apply a small stretch to the muscle, which increases twitch size (4). Torque signals were amplified and collected at 2,000 Hz using an analog-to-digital board with a resolution of 12 bits. Data were stored on an IPC 486DX for further analysis.

Stimulation. The tibialis nerve was stimulated in the popliteal fossa using two Grass S88 stimulators in combination with isolation units approved for human use (Quincy, MA). A movable stimulation electrode was used to locate the best point for stimulation of the tibialis nerve. This point was identified as the strongest contraction of the triceps surae group that did not produce a cocontraction of the dorsiflexors. The point for stimulation was marked and covered by a carbon-impregnated rubber electrode (5 × 4 cm). The nonstimulating electrode was placed on the posterior aspect of the shank 10 cm superior to the heel. Both electrodes were thinly coated with a conductive gel and secured to the calf with adhesive tape.

Protocol. The influence of timing of twitch application on the ITT amplitude was tested by applying electrical stimulation trains with a superimposed single twitch to the triceps surae muscle. Stimulation trains of 1,200 ms (square-wave pulses of 0.8-ms pulse width) were applied at frequencies of 4, 8, 12, 16, 20, 35, and 50 Hz and two different voltages to simulate a range of motor unit firing conditions. One trial was performed for each condition. The low voltage was set in such a way that a 4-Hz train would just evoke a measurable force; the mean voltage used was 117 ± 21 (SD) V. The high voltage was limited by how much a subject could tolerate during the stimulation at 50 Hz; the mean voltage used was 140 ± 24 (SD) V. All train stimulations were submaximal. Approximately 750-800 ms into the stimulation train, a single supramaximal twitch (square-wave pulse of 240 V and 0.8-ms pulse width) was superimposed. The timing of the superimposed twitch was systematically varied as illustrated in Fig. 1 and occurred 5 ms after a train stimulus, which corresponded to 755 ms into the train for frequencies from 4 to 16 Hz and to 805 ms for frequencies from 20 to 50 Hz (trial A); in the middle of two train stimuli (trial C); and 5 ms before a train stimulus (trial E). For trials B and D, the superimposed twitches were placed in the middle of trials A and C, and trials C and E, respectively. For stimulations at 35 and 50 Hz, only trials A, C, and E were performed because of the short interpulse intervals at the high frequencies. For each subject, the order of trial sequence was randomized. The ITT evoked by the single twitch was calculated as the difference between the highest torque amplitude within 200 ms of twitch application and the train torque immediately before twitch application. For stimulation frequencies that did not cause a fused contraction (i.e., 4 and 8 Hz), the highest torque within 200 ms before the single twitch was used to calculate the train force. A synchronization pulse at the instant of the superimposed twitch was recorded with the force trace for reference of the data analysis. ITTs for the different timings of the superimposed twitch relative to the periodic pulse train were compared using repeated-measures ANOVA. A significance level of  = 0.05 was used for all tests.

View larger version (12K): [in this window] [in a new window]   Fig. 1.   Schematics illustrating the 5 different time points at which the single superimposed twitch was applied relative to 2 consecutive train pulses (Px and Px+1) occurring within the time interval t. 't = t/2; t" = (t  10 ms)/4.

Experiment 2

Purpose. The second experiment was aimed at testing the hypothesis that variability of the ITT magnitude decreases as a function of the number of superimposed twitches applied. For this experiment, subjects performed submaximal knee extensor contractions during which single and multiple electrical twitches were applied and the evoked ITT was measured.

Subjects. Five men and seven women with a mean age of 25.1 ± 4.4 (SD) yr participated in the second experiment. All subjects were moderately active and were recruited from among members of the Faculty of Kinesiology.

Force measurements. Isometric knee extensor moments were measured using a Cybex dynamometer (Cybex Norm, Testing and Rehabilitation System, Lumex). Subjects were seated with a knee angle of 30° from full extension and a hip angle of 90°. The lateral epicondyle of the femur was aligned with the axis of rotation of the dynamometer, and the thighs and upper body were firmly secured with straps. Subjects were asked to perform a series of maximal and submaximal isometric knee extensor contractions. Subjects were instructed to slowly build up force to reach the required force level and then hold the contraction steady for 3-4 s. The knee extensor moment produced by the subjects was displayed on-line for visual feedback.

Stimulation. Femoral nerve stimulation was performed by applying electrical twitches to the femoral nerve ~1 s after the subjects reached the required force plateau (4, 23). A Grass S88 muscle stimulator in combination with an isolation unit approved for human use was used for nerve stimulation. Carbon-impregnated rubber electrodes (4.5 × 10 cm) were thinly coated with a conductive gel and secured to the thigh with adhesive tape. The stimulating electrode was placed over the femoral nerve just distal to the inguinal ligament. The second electrode was placed over the distal portion of the quadriceps muscle. Square-wave pulses of 240 V (maximum) and 0.8-ms duration were applied either as a single twitch, doublets, triplets, or quadruplets. The interpulse interval for the multiple twitches was 8 ms. Before testing, electrical stimuli of increasing voltage up to 240 V were applied to the nerve to familiarize the subjects with the stimulation procedure and to ensure maximal nerve stimulation.

Protocol. Subjects were required to perform three maximal-effort isometric knee extensor contractions with 2 min of rest between contractions. The contraction with the highest force was taken as the maximum from which 50% was calculated. A target line corresponding to 50% of the maximal voluntary contraction was displayed on an oscilloscope. This line had to be matched by a second line, which moved as a result of subjects' contractions. Subjects were asked to hold the contraction as steady as possible. During the force plateau, either a single twitch, doublet, triplet, or quadruplet was applied. Five trials were performed for each condition in a random order. The ITT during a specific condition was normalized to the torque produced by the same twitch (i.e., single twitch, doublet, triplet, or quadruplet) applied to the resting muscle, referred to as the resting twitch torque (RTT). MI was calculated as (ITT/RTT) × 100%. Mean MI (average of the 5 trials) and variability in ITT, calculated as the standard deviation of five corresponding trials, were determined for each subject. Differences in MI and differences in the ITT variability (i.e., SD) between conditions were tested for statistical significance using ANOVA. A significance level of  = 0.05 was used for all tests.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experiment 1

View larger version (20K): [in this window] [in a new window]   Fig. 2.   Torque (open bars) and interpolated twitch torque (solid bars) for the different stimulation frequencies. Values are means ± SE.

View larger version (12K): [in this window] [in a new window]   Fig. 3.   Interpolated twitch torque for the 5 different timing conditions. Values are means ± SE for frequencies from 4 to 20 Hz. *P < 0.05.

Experiment 2

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Superimposing an electrical twitch on a voluntarily contracted muscle has been the primary method for estimating activation levels of muscles (e.g., Refs. 1, 3-5, 7, 13, 18, 23, 25). With the use of this technique, the maximal activation potential has been assessed in patients with joint pathologies (e.g., Refs. 15, 21, 24), in aging populations (e.g., Ref. 9), and during muscle fatigue (e.g., Ref. 6). In healthy subjects, results regarding muscle activation potential remain ambiguous. Although it was found in some studies that maximal voluntary activation is possible in specific muscle groups (e.g., Refs. 1, 4), others claim that a small ITT can usually be evoked during maximal effort contractions, suggesting that motor unit activation is incomplete (e.g., Refs. 10, 22, 23, 27).

It has been argued that the absence of a twitch torque found in some studies may result from a lack of sensitivity to discover small twitch torques during near-maximal contractions rather than reflect true maximal muscle activation (16). Furthermore, differences in the stimulation protocols may contribute to the different findings. Train stimulations have been found to more often evoke an ITT than single twitches (16, 20, 22). Other authors, however, found that although multiple twitches evoke larger ITTs, when normalized to a corresponding resting twitch muscle activation was similar when using single, double, or up to five twitches (1, 3). Also, the ability to maximally activate a muscle group seems to depend on the group tested (e.g., Ref. 4).

Although the effects of the number of stimuli and the sensitivity of the recording devices on estimates of muscle activation have been discussed in the literature, the issue of variability in the ITT magnitude has not been considered. The present study was motivated by the observation that the interpolated twitch technique is associated with a large variation in the elicited twitch torque for repeated measurements at similar levels of voluntary contraction (see Fig. 5 in Ref. 18). This variability affects the ability to predict muscle activation levels from single measurements and limits the use of this technique for diagnostic purposes in patient populations. It was hypothesized that the variability in the ITT magnitude is a function of the random timing of the superimposed twitch with respect to the pulses of voluntary excitation trains. By systematically altering the time point of twitch application in a simulated pulse train, as done here, a systematic difference in the ITT magnitude as a function of timing should be obtained. Furthermore, if the random timing of the twitch application is the source for the variability in ITT, the variability should be decreased when part of the randomness is eliminated, for example, by increasing the number of superimposed twitches.

The results of this study suggest that at least part of the variability in the superimposed twitch technique is caused by the stochastic nature of the timing of the superimposed electrical twitch relative to the voluntary stimulation trains. The ITT was shown to depend critically on the occurrence of the additional twitch relative to the pulse train (Fig. 3). A superimposed (supramaximal) twitch given 5 ms before a (submaximal) pulse of the periodic stimulation train produced a significantly larger ITT than the superimposed twitches applied at the remaining times. Therefore, it appears that the time interval between the superimposed twitch and the train stimulus is crucial for the magnitude of the resulting ITT. By knowing the approximate number of active motor units and the distribution of stimulation frequencies of the active motor units, it should be possible to estimate theoretically the variability in the ITT.

One possibly surprising result was that the superimposed twitch preceding the pulse of a stimulation train by 5 ms had an effect on the ITT different from a superimposed twitch following a train pulse by 5 ms. The catchlike properties of a doublet stimulation have been described in the literature (e.g., cf. Refs. 2, 8, 26). That is, if a superimposed twitch occurs within 5-10 ms of a pulse of the ongoing stimulation train, a nonlinear increase in ITT occurs because of the ability of the muscle to hold tension and sustain it for the duration of the train (26). Therefore, we would have expected that a superimposed twitch applied 5 ms before a train pulse would have a similar effect than a superimposed twitch applied 5 ms after a train pulse. However, it has to be remembered that the pulses of the stimulation train were submaximal, whereas the superimposed twitches were supramaximal. Therefore, the results of this study suggest that a doublet stimulation with a small (first) and a large (second) pulse produces a smaller ITT than a doublet stimulation in which the large pulse is given first. Pilot studies on four additional subjects using 18 systematic small-large and large-small doublet stimulations showed that, indeed, a larger ITT is produced when the large stimulus precedes the small one. This result, in turn, may be explained by the idea that the first pulse of a doublet sensitizes the activated motor units to a doublet force; therefore, the size of the first pulse of any doublet may be more important than the size of the second one. This speculation needs detailed examination.

If much of the variability of the ITT is caused by the stochastic nature of the timing of the superimposed twitch relative to the voluntary pulse trains, then this variation should decrease by removing the randomness of the timing of the superimposed twitch. One way of doing this is by adding further electrical stimuli to the single twitch. By increasing the number of superimposed twitches, the stimulation environment becomes increasingly more similar, the randomness decreases, and the variation of the ITT should decrease in parallel. In fact, by adding a large number of electrical twitches to the single pulse, the variability should disappear completely. However, adding a long superimposed stimulation train onto a voluntary contraction is often impractical because of pain and the risk of producing injury.

In accordance with our expectation, the variability in the ITT decreased continuously by going from a single-twitch to a quadruplet stimulation (Table 1), indicating that multiple twitches may provide more reliable estimates of MI than single twitches. Practical considerations may limit the number of superimposed twitches that can be applied in an experimental setting without interfering with the required task or the clinical goal.

To summarize, the results of this study suggest that the ITT of a muscle that is treated like a single motor unit by virtue of the synchronized activation varies as a function of the timing of the superimposed twitch relative to the pulses of the ongoing stimulation train. If we can assume that this result is also obtained for single motor units in a voluntarily contracting muscle, and if we further assume that these effects on single motor units can be summed across all motor units of an active muscle, then large variations in ITT should be observed during voluntary contraction. Although it is hard (if not impossible at present) to directly verify this mechanism of ITT variability during voluntary muscle contraction, a theoretical model similar to that by Fuglevand et al. (11) with randomly applied superimposed twitches gives the variations in ITT one would expect based on the findings in this study.