INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

MUSCLE CRAMP IS A SUDDEN, involuntary painful muscle contraction, accompanied by a palpable knotting of the muscle. Cramp is self-limiting within minutes and can often be relieved by passive stretch or massage of the muscle (16, 39). According to a survey of sport academy students, the prevalence of ordinary cramp can be as high as 95% (32). According to a more recent study, one-third of the Dutch adult population has at least one muscle cramp a year and 2% suffer from muscle cramp every week (19). Although cramp is common among healthy people, muscle cramp may also appear as a consequence of neuromuscular diseases like amyotrophic lateral sclerosis, radiculopathy, and neuropathy (18, 24, 29).

Although the origin of muscle cramp is generally considered to be of a neurogenic nature (e.g., Ref. 25), there are still two different views about its precise origin. The first theory proposes that cramp is the result of abnormal excitation of the terminal branches of motor axons (2, 7, 23), whereas the second theory states that cramp may result from some form of hyperexcitability or bistability of motoneurons (22, 32, 38). Each theory fails to explain all observations and to provide conclusive evidence. For example, the fact that cramp can be elicited by electrical stimulation of the peripheral nerve, even after proximal nerve block (2), is not in agreement with the motoneuron theory, whereas the observation of whole motor unit potentials (MUPs) during cramp (38) is in correspondence with that theory only. In addition, we think that even an origin at the level of the muscle fiber cannot be excluded. First, the supposed neurogenic origin is only based on observations that cramp may occur in individuals with no obvious pathophysiology and that cramp cannot be induced after the administration of curare (26). Second, single muscle fibers are capable of manifesting repetitive discharges independently (6). The normal insertional activity recorded during needle electromyogram (EMG) investigations (9) is another indication that muscle fibers under normal physiological conditions are not that far from presenting independent activity.

Ross and Thomas (38) concluded that cramp elicits whole motor unit activity. Such a result would imply an origination of cramp at the motoneuron level. They came to this conclusion from needle EMG recordings with electrodes having very small recording areas. These recordings did not show any differences between the shape of the action potentials during cramp and that during maximal voluntary contraction (MVC) (32, 38). Because of the small recording electrode used in these studies, it can be argued that potentials from single fibers or a very small number of single fibers were recorded (11). Therefore, we think that from these results it is difficult to conclude on a motor unit level, as a large part of the motor unit is not "seen." Another methodological disadvantage of using a needle electrode for studying muscle cramp is the local and unpredictable character of the cramp (7, 23, 32, 38).

We aimed to study how cramp is distributed and how it spreads along the muscle and whether indeed whole motor units are active during cramp. A technique by which activation of superimposed patterns of whole motor units can be studied is surface EMG. During normal motor unit activation, like in voluntary contraction, the distant view of surface EMG guarantees a view of the whole motor unit (35, 36). Spectral analysis can be used to obtain information about the firing process and the shape of the average MUPs (17, 21, 33, 46). Changes in muscle fiber membrane properties are also known to show up in the surface EMG (47). Furthermore, the noninvasiveness of surface EMG allows a topographical view, which facilitates the analysis of that part of the muscle that is acting most strongly in a cramp, and allows the study of spread and propagation of the cramp. In the present study, 64-channel surface EMG obtained during muscle cramp is analyzed in time and frequency domain and compared with results obtained during MVC of the same muscle.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects. Eight healthy adult (age 47 ± 14 yr) cramp-prone subjects (3 men, 5 women) participated in this experiment after giving informed consent. The subjects were selected from a pool of 104 subjects who previously participated in a study concerning the efficacy of hydroquinine in preventing frequent muscle cramp (20). All selected subjects were able to voluntarily induce cramp in the triceps surae, considered themselves as healthy, did not suffer from a known neuromuscular disease at the moment of selection, and did not use quinine- or hydroquinine-containing products. Although we intended to compare the cramp characteristics with those of MVCs, only four of the selected subjects could generate an MVC without having a muscle cramp almost instantaneously. The Committee on Experiments in Humans of the Faculty of Medicine at the University of Nijmegen approved the experimental protocol.

Electrodes. Sixty-four Beckman biopotential electrodes (Sensormedics), with diameters of 7 mm, were situated as an eight-by-eight matrix in a square-shaped rubber sheet with center-to-center distances of 2 cm (both row- and column-wise). Before the electrodes were positioned, the skin surface was rubbed with Skinpure (Nihon Kohden) and cleaned with alcohol, and all electrodes were filled with electrode paste (Elefix; Nihon Kohden). The holder with the 64 surface electrodes was placed over the lower leg such that the triceps surae were completely covered with electrodes (see Fig. 1). Recordings were made in a unipolar montage. A common reference electrode was placed on the contralateral knee, and a ground electrode was placed on the ipsilateral ankle.

View larger version (137K): [in this window] [in a new window]   Fig. 1.   Overview of the electrode configuration over the calf muscles.

Data acquisition. The myoelectric signals from the skin surface were amplified (Neurotop, Nihon Kohden; input impedance of 100 M) over a frequency range of 5-800 Hz with a gain of 1,000. The EMG signals and a stage marker were analog-to-digital converted (DT2839 and DT2896, Data Translation; 12 bits, maximal total sample frequency of 224 kHz) and continuously stored on the hard disk of a computer with a sample frequency of 2 kHz/channel.

Experimental protocol. After the surface electrode array was placed over the triceps surae of the left leg or over the leg that most frequently cramped for the subject, the EMG signals were inspected visually. The subjects were then asked to perform an MVC of at least 20 s, with feedback of delivered force. Each subject lay on their belly on a bench with knee bent at ~120 degrees, ankle positioned at 90 degrees, and foot placed in a force transducer. When a cramp appeared during the MVC measurement, the MVC measurement was repeated after 10-min rest. After one to three MVC trials, the position of the subject's leg was changed into a preferable position for generating muscle cramp with the knee bent at 160 degrees and the foot completely plantar flexed. In this position, no external force was delivered. The subject was then asked to generate a muscle cramp in the calf muscle and was instructed to report immediately when and where he or she felt a muscle cramp, to stop the voluntary contraction, and to let the cramp develop until it spontaneously disappeared. Subjective information from the subject on the time envelope of the muscle cramp (voluntary contraction, cramp, termination, disappearing cramp) was added to the EMG signals by the examiner by using a stagemarker. Five to ten minutes after each trial of cramp generation (successful or nonsuccessful), a new trial was performed. The cramp trials were stopped after 1 h or sooner on indication of the subject. Thereafter, the leg was replaced into the original position, and the MVC trial was repeated.

Parameterization. Data processing was performed with software written in Matlab version 5.2. After visual inspection of the data, the root-mean-square (RMS) amplitude of all 64 EMG channels was calculated in blocks of 2.048 s (4,096 data points). Then, with the information from the stagemarker, the average RMS amplitude during cramp or MVC was determined for the 64 channels. The percentage of electrodes having average values higher than 60% of the maximally averaged RMS in the recording was taken to represent the spatial size (S) of the cramp or MVC. A relative threshold was chosen so that even small-amplitude contractions had a final size of at least the domain of one electrode. The threshold value of 60% gives the largest variation in size parameter between the different contractions.

Statistics. The data are presented as means ± SD. Statistical tests were applied with SPSS for windows software, release 6.1 (1989-1994). The data were analyzed twice. First, to use all obtained data, a one-way ANOVA model was used with subject and contraction type (cramp or MVC) as factors. Second, paired t-tests were used to compare MVC and cramp contractions of the four subjects (subjects 2, 4, 7, and 8) with both MVC and cramp recordings. In three (subjects 2, 4, and 7) of these four subjects, the cramp recording with the same leg position as the MVC recording was used for this purpose. The fourth subject (subject 8) had only one cramp contraction with the leg completely extended. For each of these four subjects, the MVC recording with the V2 value occurring closest to the cramp recording was chosen. The significance level was set at P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In total, 38 muscle cramp events and 8 MVCs were recorded in the triceps surae of cramp-prone subjects. Four subjects (subjects 1, 3, 5, and 6) were not able to perform an MVC test without developing a muscle cramp. For the development of a muscle cramp, maximal contraction was often not a prerequisite. A specific leg position could already trigger a cramp. Sometimes, the cramp or the MVC trial failed because of a muscle cramp in the foot, the hamstrings, or the contralateral leg.

Spatiotemporal development of cramp. Figure 2 shows an example of the development of a cramp in terms of a change in RMS amplitude distribution over time. First, the subject contracted the muscle voluntarily (5th s) and then tried to relax the muscle, but a cramp developed in a small area (7th s). The cramp spread over a larger region, and its focus changed position (9th to 29th s) and then decreased in size and intensity until it disappeared (31st to 51st s). Such a rather slow change in position (a few centimeters per second) over the muscle was observed in ~50% of the cramps; in the other cramps, the position did not change. The direction of spreading was without preference. In Table 1, the individual and group averages of the muscle cramp parameters are shown together with the group averages of the MVCs. The spatial size parameter shows that, as in the example shown in Fig. 2, cramp often involves a part of the muscle that is smaller than that involved in a voluntary contraction. One-way ANOVA analysis of the whole data set and paired t-tests of the selected MVC and cramp contraction of four subjects showed that this difference in spatial size was significant. The subject's sensation of the (shifting) location of the muscle cramp corresponded to sites with maximal EMG amplitudes.

View larger version (44K): [in this window] [in a new window]   Fig. 2.   Typical example of spatiotemporal development of a muscle cramp. Each graph represents the electromyographic (EMG) amplitude distribution over the skin surface determined over 1 s. Intensity of the gray scale of each small square represents the root-mean-square (RMS) amplitude recorded by a specific electrode on the skin (see scale bar). Bottom left corner of each graph depicts the most medial, distal position, whereas top right corner represents the most lateral, proximal position. Distance between the proximal and distal electrodes and the medial and lateral electrodes is 14 cm. Number above each graph indicates which period of 1 s after onset of contraction is used for that particular graph. This subject (subject 4) reported the cramp 6 s after onset of contraction; 45 s later (in the 51st s after contraction onset), subject 4 reported cramp disappearance.

View larger version (30K): [in this window] [in a new window]   Fig. 3.   EMG representation of 2 muscle cramps is visualized in 2 ways for 3 different subjects (rows I, II, and III). A and C: RMS value at the focus of the muscle cramp as a function of time. RMS values are obtained in blocks of 2 s. At time 0, the subjects felt a muscle cramp. B and D: RMS distribution over the skin surface at maximal cramp in time and space. Intensity of the gray scale corresponds to RMS values (ranges are plotted above each graph). Lightest blocks represent the electrodes with RMS values lower than 60% of the maximum (the complement of spatial size). dist, Distal; prox, proximal; med, medial; lat, lateral.

Time and frequency analysis in focus of activity. Figure 4 shows four typical examples of one-channel unipolar EMG and its power spectrum and the average MUP with its power spectrum. Rows I and III in Fig. 4 show MVC recordings from subjects 4 and 2, and rows II and IV show cramp recordings from the same subjects. In Fig. 4A, it is obvious that the prominent peaks have smaller durations and appear more frequently during the cramp than during MVC. When the EMG power spectra of cramp and MVC were compared, cramp was shown to have less power at low frequencies and more power at high frequencies. The differences were quantified with the parameters P10 and P50 (Table 2). One-way ANOVA analysis of the whole data set showed that the P10 and P50 during the beginning and the maximum of the MVC (V1 and V2) were significantly lower than those during C1 and C2. Paired t-tests of four subjects with MVC and cramp contractions showed similar results. The P50 during V1 and V2 is significantly lower than during C1 and C2, respectively (95% confidence interval of the difference between the P50 during V1 and C1 is between 61 and 3 Hz; the 95% confidence interval of the difference between the P50 during V2 and C2 is between 36 and 10 Hz). The P10 during V1 and V2 was not significantly lower than during C1 and C2, respectively (95% confidence interval of the difference between the P10 during V1 and C1 is between 42 and +3 Hz; the 95% confidence interval of the difference between the P10 during V2 and C2 is between 38 and +6 Hz).

View larger version (33K): [in this window] [in a new window]   Fig. 4.   Four examples (rows I-IV) of 500-ms one-channel unipolar EMG (A), power spectrum of 2-s EMG (B), and an average motor unit potential (MUP; C) with its power spectrum (D). Horizontal dotted lines in the EMG recordings (A) represent trigger level for obtaining MUPs. The 50th percentile of power spectrum distribution (P50) frequencies (in Hz) are shown in B. Row I and III were obtained during maximal voluntary contraction of cramp-prone subjects [subjects 4 (S4) and 2 (S2), respectively], and rows II and IV were obtained during cramp of the same subjects.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Thirty-eight muscle cramps in the triceps surae of eight healthy subjects were compared with eight MVCs of the same muscle in four of these subjects. The results showed that the location of cramp in the muscle can vary considerably between and within subjects. Cramp can start at several places simultaneously and can spread out slowly over the muscle. Compared with the surface EMG measured during MVC, the power spectrum of the EMG in the cramp regions contained less low frequencies and more high frequencies. This was observed at the onset of cramp vs. submaximal voluntary contraction as well as at full cramp compared with MVC. In our view, the results provide evidence against normal motor unit activity during cramp and therefore against the spinal origin of cramp. In addition, our results are well in accordance with the concept that the action potentials during cramp are generated at the level of the terminal branches of motor axons or at the level of the muscle fibers themselves. This viewpoint will be elucidated below.

Spatiotemporal development of cramp. The amplitude of the EMG signals, expressed in the RMS value, was distributed more equally over the muscle during MVC than during cramp (Fig. 2). During MVC, the complete muscle group is active, whereas often only a part of the muscle is involved in cramp. This local character of muscle cramp was previously reported (7, 23, 32, 38), but the 64-channel recordings allowed a precise study of the distribution and especially the propagation of many cramp episodes. In an individual subject, a cramp could occur in any of the outer boundaries or centrally in the muscle. Sometimes the cramp was even multifocal (Figs. 2 and 3). The cramp could slowly change its region, but it never jumped from one part of the muscle to another. Apparently, a cramp spreads through activation of neighboring (groups of) muscle fibers. Consequently, the hypothesis of the spinal origin of muscle cramp can only hold when the motoneurons that innervate neighboring muscle fiber populations are also neighbors. It is known that the most distal muscles are supplied by motor nuclei located in the dorsolateral part of the ventral horn, whereas more proximal muscles are innervated by motor nuclei in the ventrolateral ventral horn, with a more or less continuous change in between (somatotopic spinal motor unit pools; Refs. 37, 48, 50). This indicates that a strong topographical correlation between motoneurons and the motor unit muscle fibers cannot be completely excluded. Nevertheless, a generation in the closely packed motoneurons in longitudinal and cross-sectional level and the large innervation ratio of motor units in the calf muscles are at odds with the focal character of cramps, often limited to a small part of one muscle.

Time and frequency analysis in focus of activity. The EMG power spectrum can be considered as the spectrum of the average MUP multiplied by the spectrum of the average process of central innervation (21, 33). The power spectrum of a regular and stationary process of motoneuron firing displays almost no power up to frequencies equal to the average firing rate, with a large peak at this average firing rate and smaller peaks at the first and higher harmonics of this frequency. The steep peaks flatten with increasing firing rate irregulation (12, 46). A possible explanation for part of the higher frequency content of the cramp EMG could thus be higher firing rates of the contributing motor units. Ross and Thomas (38) reported higher firing rates in the first 20 s of cramp [~23 pulses per second (pps)] with incidentally very high firing rates (up to 80 pps) compared with those recorded during MVC (~12 pps). However, such a difference of, on average, 10 pps cannot cause the 20-Hz difference in P10 and the 30-Hz difference in P50 between the EMG during MVC and cramp, especially not since a change in firing rate or firing stationarity mainly affects the low-frequency region (P10) and hardly affects the median frequency (P50; Ref. 12). Moreover, during cramp, the firings are supposed to be less stationary and more irregular than during MVC (38), which can even slightly reduce the median frequency (12). In addition, the spectrum of the average MUP recorded during cramp and MVC changed similarly to the frequency contents of the EMG data. Therefore, the spectral difference must mainly be attributed to a difference between the active building blocks of the signal, being action potentials with short vs. longer duration.