Intraspinal microstimulation preferentially recruits fatigue-resistant muscle fibres and generates gradual force in rat

Animal treatment and care

All animal procedures were carried out under the guidelines of the Canadian Council for Animal Care and approved by the University of Alberta Animal Welfare Committee. Female Sprague-Dawley rats were housed in pairs in a controlled environment with 50–55% humidity, at 20°C, with alternating 12 h light–12 h dark cycles and received food and water ad libitum. Animals were randomly assigned to experimental groups for either glycogen depletion or force recruitment experiments. For all surgical experiments animals were anaesthetized using 2.5% isofluorane delivered in gaseous form. Upon completion of the experiments all animals were immediately killed by cardiac excision whilst under anaesthesia. Fifty-four animals were used for this study with a mean mass of 349.6 ± 8.7 g (± s.e.m.).

Glycogen depletion experiments

Glycogen depletion was accomplished by stimulating one hindlimb while the contralateral hindlimb served as a sham control. For NCS experiments a bipolar nerve cuff with 1–2 mm interelectrode separation was placed around the femoral nerve, further insulated from surrounding muscles with mineral oil and sutured securely closed. For ISMS experiments a laminectomy was performed at T12–T13 to expose spinal cord segments L3–L4. Following this, 30 μm diameter, Teflon-insulated microwires (stainless steel 304, California Fine Wire Company, Grover Beach, CA, USA) were inserted 2–3 mm past the dura mater of the spinal cord. Specifically, microwire tips exposed for 30–60 μm were inserted into the quadriceps motoneurone pool in the ventral grey matter. Mono-polar stimulation with the return electrode placed in the back musculature was used for all ISMS experiments. The ISMS target area has been mapped by Mushahwar & Horch (1998, 2000b) in cats and is thought to be maintained in other species such as rabbit and humans (Sharrard, 1955; Romanes, 1964; Portal et al. 1991). Placement of microwires in the quadriceps motoneurone pool was confirmed with practice twitches and ultimately by the glycogen-depleted fibres themselves. A stimulator (STG1008, Multi-channel Systems Inc, Reutlingen, Germany) was used to deliver 400 μs, biphasic, cathodic-first, charge-balanced pulses through the electrodes at each frequency and amplitude. For both NCS and ISMS, stimulation consisted of five bouts of 5-min stimulation periods (1 s on, 1 s off) interspersed with 2-min rest periods between each bout. Since the duty cycle was 50%, total stimulation time was 12.5 min. Stimulation procedures were designed to produce depletion of glycogen in the activated quadriceps muscle fibres. Our own preliminary experiments showed that glycogen depletion of slow-twitch fibres was undetectable for up to 1 h of stimulation. Similarly, Rafuse & Gordon (1996) found that slow-twitch motor units required as long as 3 h of stimulation to produce fatigue and glycogen depletion. Previous reports demonstrated that glycogen depletion can be accelerated and enhanced with ischaemia of the selected muscles (Totosy de Zepetnek et al. 1992; Hudlicka et al. 1994). Therefore, in the present experiments, ischaemia was produced by placing sutures around the femoral artery of both stimulated and contralateral sham control hindlimbs. Muscles were made ischaemic at the beginning of the stimulation period by tightening the sutures around the femoral arteries. Sham hindlimbs were made ischaemic for the same duration. This procedure did not disrupt blood flow to the femoral nerve, which is supplied by the internal iliac artery.

Previous reports conducted ISMS at a range of amplitudes, including near-threshold levels (Mushahwar et al. 2003; Saigal et al. 2004). For near-threshold stimulation both NCS and ISMS protocols were carried out with the amplitude held at 1.2 × threshold (T). Threshold was defined as the lowest stimulation amplitude that produced a visible twitch. In light of the possibility that interneuronal pathways are differentially activated according to stimulus frequency (Minassian et al. 2004; Jilge et al. 2004) we performed stimulation at 1, 20 or 50 s⁻¹ to determine the effect of frequency on muscle fibre type recruitment by ISMS. NCS experiments were carried out with matching frequencies as a control. Each NCS and ISMS group at the 1.2T condition contained six animals for a total of 36. High-amplitude stimulation was also performed at 3.0T to determine the recruitment properties of ISMS and NCS at stimulation amplitudes more likely to be used in clinical FNS applications. The stimulation frequency of 20 s⁻¹ was chosen for these experiments because it represents a frequency that is similar to the maximum natural rate of motor unit firing in humans (Fuglevand & Johns, 2004 [www.colorado.edu/intphys/news/summaries.html]; Taylor & Enoka, 2004). This stimulation rate has also been successfully used in the past for ISMS experiments from our group (Saigal et al. 2004) and is likely to be the most clinically relevant frequency for ISMS therapy. Six animals were included in each of the 3.0T groups (ISMS and NCS) for a total of 12.

Following stimulation, the vastus lateralis and rectus femoris muscles were quickly excised as a group from experimental and control legs and placed into ice-cold saline for 30 s with gentle stirring. Rectus femoris and vastus lateralis muscles were chosen because they represent a large portion of the quadriceps muscle mass and they contain mixed MHC isoform based fibre types which allowed for identification of the full spectrum of fibre types. Handling and separation of these muscles was performed on a metal platform embedded in ice. Once separated, muscles were placed in a slightly stretched position on aluminium foil and frozen in melting isopentane (−156°C) cooled in liquid nitrogen. Muscles were stored at −80°C until sectioned to a thickness of 10 μm.

To determine which fibres were activated by stimulation, frozen muscle cross-sections of the rectus femoris and vastus lateralis were stained for glycogen using the periodic acid-Schiff (PAS) reaction. Activated fibres were identified as those lacking glycogen, a technique commonly known as the glycogen depletion method (Kugelberg & Edstrom, 1968; Kim et al. 1995). The absence of glycogen depletion in any of the contralateral sham control muscles confirmed that neither ischaemia due to occlusion of the femoral artery nor the muscle collection procedure induced glycogen degradation.

Force recruitment experiments

Isometric twitch force was measured during application of NCS or ISMS to generate recruitment curves for these two stimulation protocols. Nerve cuffs were placed around the femoral nerve and a laminectomy was performed to expose spinal cord segments L3–L4. The animal was fixed at the iliac crest and at the epicondyles of the femur with a stereotaxic array to prevent movement. The patellar tendon was dissected from its point of insertion and attached to a force transducer (Interface MB-5, Interface Inc., Scottsdale, AZ, USA). Stimulation (single, 400 μs, biphasic, cathodic-first, charge-balanced pulses) proceeded with the nerve cuff over the femoral nerve or with intraspinal microwires inserted within the quadriceps motoneurone pool in the ventral spinal cord (Mushahwar & Horch, 2000a). Data were amplified 100-fold and captured at a sampling rate of 1000 s⁻¹ using a data acquisition interface (Power 1401, Cambridge Electronic Design, Cambridge, UK) with associated software (Signal v. 2.13, Cambridge Electronic Design). Stimulation amplitude was varied upwards from subthreshold levels in quasi-random order and the force of the evoked twitch was measured with rest periods of approximately 30 s between pulses to avoid potentiation between twitches. Isometric force was recorded from the same hindlimb using both NCS and ISMS methods to allow paired comparisons. In some cases multiple ISMS curves obtained by stimulating different locations in the same pool were gathered on the same hindlimb to compare against a single NCS curve. Recruitment curves were gathered from six animals providing six NCS and seven ISMS curves.

To compare twitch forces evoked by NCS and ISMS, forces were normalized to the maximum twitch force obtained by the nerve cuff for each hindlimb and plotted against pulse amplitude (μA). Linear regression was used to calculate slopes for the recruitment curves between threshold and 250 μA or between threshold and the onset of force plateau if this occurred before 250 μA.

Statistical analysis

The same statistical analysis was used for both glycogen depletion and force recruitment experiments. All analyses were performed with a computerized software statistics package (SPSS 11.0, SPSS Inc., Chicago, IL, USA). Differences between group means were determined using Student's t test with independent samples or one-way ANOVA with Scheffe's post hoc test. All planned comparisons were undertaken with one-tailed analysis, whereas unplanned comparisons were analysed using a two-tailed test. Differences were considered significant at P < 0.05. All results are presented as means ± s.e.m.

Threshold levels for ISMS were significantly lower than those for NCS

Following either the glycogen depletion or force recruitment experiments, spinal cords were extracted, frozen, and sectioned on a cryostat. The purpose of this procedure was to confirm that ISMS wires were properly located within the ventral spinal cord. Figure 1 summarizes, on a single cross-section, the tip locations for 17 of 33 intraspinal microwires used in this investigation. Mean stimulation threshold in the glycogen depletion experiments was 12.1 ± 1.0 μA and 25.0 ± 3.0 μA for ISMS and NCS, respectively, a highly significant difference (P < 0.0002). Differences in mean threshold generated during the force recruitment experiments were larger than for the glycogen depletion experiments. Mean threshold was 7.7 ± 0.1 μA and 39.1 ± 5.7 μA for ISMS and NCS, respectively (P < 0.0006).

ISMS caused greater depletion of fatigue-resistant fibres at high amplitudes of stimulation

Glycogen depletion experiments were carried out in order to determine which fibres were activated by either NCS or ISMS under different stimulation conditions. Figure 2 shows an example of the method used to identify glycogen-depleted fibres in one ISMS animal and one NCS animal, both stimulated at 1 s⁻¹ 1.2T. Depleted fibres were identified by the PAS stain (Fig. 2A) and then characterized according to their MHC content in serial sections (type-I, -IIA, -IIB in Fig. 1B–D, respectively). In the example shown, ISMS primarily depleted type-I and -IIA fibres while NCS depleted type-IIB and -IID fibres. PAS staining of control muscles demonstrated that neither the ischaemia, nor the extraction procedure produced detectable glycogen depletion (Fig. 2E). In the 1.2T condition, the mean number of depleted fibres per animal was 176.6 ± 27.5 and 181.2 ± 44.6 for ISMS and NCS, respectively. In the 3.0T condition, the mean number of fibres depleted per animal was 150.5 ± 49.5 and 1215.7 ± 363.6 for ISMS and NCS, respectively. The sum of all fibres depleted and analysed in this study was 14 637.

Figure 3 shows the cumulative proportion of each fibre type depleted in each stimulation condition from all animals. The total number of fibres depleted in the 1 s⁻¹ 1.2T, 20 s⁻¹ 1.2T, 50 s⁻¹ 1.2T and 20 s⁻¹ 3.0T were 726, 1481, 1055 and 7294, respectively, for NCS, and 1249, 1178, 751 and 1188, respectively, for ISMS. NCS stimulation at 3.0T depleted significantly more fibres than any other condition (P < 0.004). All other conditions were not significantly different in terms of total fibres depleted. The only condition that resulted in greater depletion of type I, I/IIA and IIA fibres by ISMS was 20 s⁻¹ 3.0T. Figure 4 shows the proportion of glycogen-depleted fibres that were FR type (type I + type I/IIA + type IIA) in each experimental group (n = 6 animals per group). At 1.2T, the proportion of FR fibres recruited by ISMS was 1.7- (P < 0.23), 2.8- (P < 0.059) and 0.67-fold (P < 0.24) that of NCS at 1, 20 and 50 s⁻¹ frequencies, respectively. Although both 1 and 20 s⁻¹ ISMS recruited proportionally more FR fibres than NCS, this trend did not reach statistical significance in either case. Specifically, in the 1.2T conditions, some ISMS samples did not display depleted FR fibres, producing large variances in the data (e.g. min 0.0%, max 97.8% in the 1 s⁻¹ condition). In contrast, the 3.0T condition produced consistent and highly significant increases (P < 0.0001) in the proportion of FR fibres recruited by ISMS over NCS. Only 0.4% of fibres recruited by 20 s⁻¹ 3.0T NCS were FR fibres, whereas 44.4% of the total fibres depleted by ISMS at 20 s⁻¹ 3.0T were FR.

Fibre type recruitment by ISMS and NCS in mixed fast-twitch muscle

Electrophoresis was performed to measure the MHC-isoform contents of activated muscles. This allowed us to determine whether the number of FR fibres activated by ISMS or NCS was proportional to their distribution within the vastus lateralis and rectus femoris, or if these fibres were selectively recruited. The MHC-isoform contents of the vastus lateralis and rectus femoris are shown in Fig. 5. There was no difference in MHCI content between rectus femoris and vastus lateralis (P = 0.99), but the rectus femoris displayed significantly more MHCIIa (P < 0.001), MHCIId/x (P < 0.003), and less MHCIIb (P < 0.003). For the purposes of this study the MHC-isoform contents of the rectus femoris and vastus lateralis were grouped into FR and FF categories. FR MHC content was 22.8 ± 2.0% and 16.0 ± 1.4% in the rectus femoris and vastus lateralis, respectively (P < 0.02). These data were further summarized to compare with the grouped rectus femoris and vastus lateralis glycogen depletion results. Using these broad classifications, the proportion of FR and FF fibres were 19.4 ± 1.5% and 80.6 ± 1.5%, respectively. Comparison of the cumulative proportion of FR fibres depleted in each stimulation condition revealed that ISMS recruited FR fibres at 2.9-, 1.9-, 1.7- and 2.5-fold the FR MHC content at 1, 20 and 50 s⁻¹ 1.2T and 20 s⁻¹ 3.0T, respectively. In contrast, NCS recruited FR fibres at 1.2, 1.0, 2.1 and 0.0 times the FR MHC content at 1, 20 and 50 s⁻¹ 1.2T and 20 s⁻¹ 3.0T, respectively (Fig. 5C).

ISMS recruited force more gradually than NCS

Force recruitment experiments were performed to compare the functional characteristics of ISMS and NCS with the results from the glycogen depletion experiments. Twitch recruitment curves obtained via intraspinal and nerve cuff stimulation from 0 to 250 μA are shown in Fig. 6. The mean slope of the recruitment curves for NCS was 0.020 ± 0.006 and was thus 4.9-fold steeper (P < 0.02) than the mean slope of recruitment curves for ISMS (0.0041 ± 0.0008).

A charge level of 90 nC has previously been identified as the maximum level for safe charge injection in central nervous system tissue (Agnew & McCreery, 1990). Since the pulse width for all stimulation experiments was 400 μs, the maximum safe charge of 90 nC corresponds to 225 μA. Mean peak twitch force at 225 μA was 1.56 ± 0.49 N and 0.34 ± 0.06 N for NCS and ISMS, respectively (P < 0.03). At 225 μA, ISMS through a single microwire recruited a mean of 36.8 ± 11.1% of twitch force recruited by NCS at the same charge level.

Overview

The purpose of this study was to compare the muscle recruitment characteristics of NCS, a peripheral method of stimulation, with the novel rehabilitative therapy, ISMS. In the first set of experiments, muscle fibres activated at three different frequencies and two different amplitudes by each stimulation protocol were identified by the glycogen depletion method and classified according to their corresponding MHC-based fibre types. In the second set of experiments, data were gathered from ISMS and NCS twitch force recruitment curves to assess force recruitment characteristics. During high-amplitude stimulation ISMS activated a significantly larger proportion of fatigue-resistant fibres than NCS. In accordance with this result we found that NCS twitch force recruitment was steeper than ISMS. ISMS through a single electrode produced a considerable amount of force with more gradual recruitment. Whole nerve stimulation through the nerve cuff required significantly more charge at threshold but produced higher forces at the same charge level. Other work from our laboratory has shown that ISMS probably recruits motoneurones through interneurones or afferent pathways (Mushahwar et al. 2003). This would explain the more gradual recruitment of force and increased activation of fatigue-resistant muscle fibres by ISMS compared with NCS. This study corroborates and enhances previous evidence for the efficacy of ISMS and is the first work to compare directly the muscle recruitment properties of ISMS and NCS.

ISMS causes greater depletion of fatigue-resistant fibres at high amplitudes of stimulation

In this study, glycogen depletion was enhanced using ischaemia. Our preliminary experiments, as well as the work of others (Rafuse & Gordon, 1996), showed that stimulation alone required up to 3 h to produce detectable glycogen depletion in slow-twitch muscle fibres. Therefore, ischaemia was induced in the quadriceps muscles, but not in the femoral nerve, by occlusion of arterial flow through the femoral artery. PAS staining in control muscles showed that ischaemia alone did not induce detectable glycogen depletion, thus proving the validity of this technique. Glycogen depletion via ISMS or NCS resulted in differences in the proportions of FR fibres depleted. In the 1.2T condition these differences approached significance at 20 s⁻¹ (P < 0.059) but not in the 1 or 50 s⁻¹ conditions.

In all of the 1.2T conditions, there were at least a few ISMS samples that did not deplete FR fibres. Although fibre recruitment resulting from low-amplitude stimulation may be especially sensitive to electrode placement, at 3.0T small differences in electrode placement are less likely to affect the outcome due to increased current spread. Indeed, the differences between ISMS and NCS at 3.0T were particularly consistent, where the former recruited a significantly higher proportion of FR fibres (Fig. 3). It is interesting that the 3.0T ISMS condition did not deplete a significantly larger number of fibres than other ISMS conditions. This suggests a limited spread of current as has been shown in the past (Mushahwar & Horch, 1997; Pikov & McCreery, 2004; Lemay & Grill, 2004; Snow et al. 2005).

When comparing the proportion of FR fibres depleted by each condition it is illuminating to consider the contents of MHC isoforms in rectus femoris and vastus lateralis muscles. Electrophoretic analyses revealed that 19.4% of the total MHC content can be placed in the FR category. Thus, the cumulative proportion of FR fibres recruited by ISMS represents a 2.9-, 1.9-, 1.7- and 2.5-fold increase over the FR MHC content at 1, 20 and 50 s⁻¹ 1.2T and at 20 s⁻¹ 3.0T, respectively (Fig. 4). In contrast, NCS recruited FR fibres at 1.2-, 1.0-, 2.1- and 0.0-fold the FR MHC content at 1, 20 and 50 s⁻¹ 1.2T and at 20 s⁻¹ 3.0T, respectively. Although ISMS preferentially recruited FR fibres in the rectus femoris and vastus lateralis, only the 50 s⁻¹ 1.2T NCS condition did the same.

Surprisingly, the 50 s⁻¹ condition not only recruited FR fibres out of proportion to their MHC isoform content, but 50 s⁻¹ NCS depleted a larger proportion of FR fibres than the 50 s⁻¹ ISMS condition (Figs 3 and 4). The proportion of FR fibres depleted by ISMS did not decrease significantly compared with other frequencies. Instead, there was an increase in the number of FR fibres depleted by NCS at 50 s⁻¹. Other authors have used peripheral, low amplitude stimulation at 100 s⁻¹ to produce contractions which may contain both peripheral and central components (Collins et al. 2001; Collins et al. 2002). These authors suggested that their results were due to motoneurone activation through Ia afferents and the generation of plateau potentials. Although the frequencies used in this study were not as high as the 100 s⁻¹ used by Collins et al. (2001, 2002), it is possible that the higher proportion of FR fibres at the 50 s⁻¹ 1.2T NCS condition is due to activation of Ia afferents. These afferents make physiologically normal connections to motoneurones and could recruit fibres in a near-normal order. Although our data suggest that it is possible for NCS-type systems at 50 s⁻¹ to recruit FR fibres, they would probably still suffer from problems encountered by peripheral stimulation methods (Popovic, 1992). Furthermore, this result would only be present at lower stimulation amplitudes where low levels of force are produced. Indeed, our results suggest that at 3.0T stimulation this effect is abolished due to the overriding influence of whole nerve stimulation through motor axons at higher amplitudes.

There is one report of surface stimulation recruiting fibres in a mixed manner (Thomas et al. 2002), but the results of this experiment are difficult to interpret (Enoka, 2002). It is likely that the authors showed neither normal recruitment nor truly random recruitment. Rather, it is more likely that the authors showed pseudorandom recruitment order due to the spread of current from the point-source surface electrode used in these experiments. Although the largest axons are characterized by their low recruitment thresholds (Henneman et al. 1965), the spatial orientation of axons in the nerve trunk could result in a situation where the smallest axons receive current above threshold level while larger axons, located further away do not. This ‘depth effect’ phenomenon could also explain the pseudorandom recruitment order reported by Singh et al. (2000) when using intramuscular BION™ electrodes.

ISMS recruits force more gradually than NCS

Similar to other work in our lab (Snow et al. 2005) we found that the mean slope for ISMS force recruitment curves was 4.9-fold lower than for NCS. A range of force recruitment curves generated by ISMS has previously been reported for electrodes targeting feline quadriceps motoneurone pools (Mushahwar & Horch, 2000a) and intermediate grey matter (Aoyagi et al. 2004). It has been proposed that differences in electrode placement within a motoneurone pool may activate different subsets of motoneurones targeting different compartments in skeletal muscle (English et al. 1993; Lieber & Friden, 2000). The fibre type distribution of these compartments may vary, and thus the force recruitment characteristics can change depending on microwire placement. This may point to another advantage of activating motoneurones in the ventral horn. Not only is it possible to activate muscles selectively with ISMS (Mushahwar & Horch, 1998, 2000a), this selectivity may also extend to subsets of motoneurone pools composed of different types of motor units. Thus, ISMS seems capable of activating the muscles in a more gradual manner, as required for postural support for extended periods, and in a steep manner as needed for standing from a chair. A multielectrode array has been proposed for ISMS (Mushahwar & Horch, 1997; Snow et al. 2005) and our results suggest that such a system could be used not just to activate individual muscles selectively but populations of fibre types within the muscle. This would represent a significant improvement over current forms of peripheral nerve stimulation.

Although NCS recruited higher forces, ISMS through a single wire recruited mean force approaching 37% of that normalized to the nerve cuff at 225 μA. This result is surprising given that it was produced with only one electrode and a single pulse. Other authors have compared single pulses with 50 s⁻¹ stimulation trains and concluded that tetanic force levels produced by single ISMS wires can match those of whole nerve stimulation through peripheral nerve cuffs (Snow et al. 2005). The use of higher frequencies, multiple electrodes and/or interleaved stimulation protocols might allow ISMS to activate the muscle fully without producing damaging effects. This approach has already proven successful in creating consistent, fatigue-resistant weight-bearing stepping in spinalized animals (Saigal et al. 2004).

Potential mechanisms for ISMS stimulation results

At the 1.2T condition the differences between NCS and ISMS did not reach significance. In fact the NCS and ISMS results from near-threshold stimulation were so similar that it is tempting to speculate that both stimulation procedures might have been acting through similar mechanisms. It has already been shown that ISMS activates afferent fibres before motoneurones (Mushahwar et al. 2003) and may also activate interneurones contained within the ventral spinal cord. In the NCS condition, Ia afferents were probably responsible for the twitch response at near-threshold conditions, otherwise known as the Hoffmann (H) reflex (Zehr, 2002). Ia afferents directly synapse on motoneurones, and thus the resulting synaptic activation of motoneurones produces a normal or near-normal physiological recruitment order. Reversed recruitment order is commonly seen at higher amplitudes with peripheral methods such as NCS or surface stimulation (Levy et al. 1990). As charge levels increase, ISMS is expected to activate more interneurones and afferents thereby recruiting force more gradually than NCS. In contrast, direct stimulation of motoneurone axons during NCS leads to rapid recruitment of fast-twitch motor units and abolition of the slow-twitch response at higher amplitudes. Neurophysiologically, this is confirmed by the increasing dominance of the muscle-wave response and disappearance of the H-reflex as stimulus amplitude increases beyond the maximal H-reflex response (Zehr, 2002). If stimulus amplitude is increased further, F-waves will be produced by NCS because of antidromic activation of motor axons in possible combination with the Ia afferent reflex arc (Mesrati & Vecchierini, 2004). In the present study, as stimulation amplitude increased, NCS reached peak twitch force quickly due to the preferential activation of FF fibres. In contrast, ISMS twitch force continued to rise gradually due to the delayed recruitment of large, fast-twitch fatigable motor units. Our results with 3.0T stimulation demonstrated that ISMS preferentially recruited FR fibres in rectus femoris and vastus lateralis muscle while NCS almost exclusively recruited FF fibres (Fig. 3). Ultimately, the pattern of ISMS activation is similar to the physiologically normal activation of motor units in which the smallest, most fatigue-resistant motor units are activated first, holding the larger units in reserve for activities requiring large force or power (Henneman 1957, 1965).

Previous work has provided insight into how muscle fibre type is specified over the long term. Periods of tonic motoneurone activity promote a slow-fibre program while periods of inactivity interspersed with bursts of high frequency activity promote the fast-fibre program (Pette & Vrbova, 1992). Evidence for this position has been enhanced by the discovery of the calcineurin signalling pathway, which transduces this activity pattern signal into the nucleus and alters gene expression (Schiaffino & Serrano, 2002). In accordance with this theory there is a dramatic slow-to-fast transformation of muscle fibre type following SCI (Burnham et al. 1997; Castro et al. 1999). Due to the plastic nature of skeletal muscle (Pette & Staron, 1997), previous authors have performed chronic low-frequency stimulation through a peripheral FNS system to rehabilitate skeletal muscle following SCI. The ultimate goal of these interventions is the up-regulation of FR fibre types and restoration of fatigue resistance (Martin et al. 1992; Andersen et al. 1996).

Although fatigue-resistant movements are an important goal for FNS it is also important to ensure that the potential for high force production in stimulated muscle is preserved over the long-term. The mixed fibre type content of intact quadriceps muscle (Fig. 5) represents the capacity of this muscle to undertake both fatigue-resistant and high-force movements. Thus, maintaining mixed muscle fibre types should remain the goal for FNS therapies following SCI. Unfortunately, long-term stimulation with peripheral forms of FNS entails the possibility of complete transformation of the muscle such that no fast fibres remain to produce high-force movements. Andersen et al. (1996) performed 30 min of stimulation through surface electrodes three times a week at 60 s⁻¹ and approximately 3.25–7.25T for up to 12 months in five male spinal cord injured subjects. Following stimulation, fibres expressing pure IID/X MHC (the fastest isoform expressed in humans) dropped from 37.2 ± 15.5% to 2.3 ± 1.4% of the total fibre content with concomitant increases in pure IIA fibre content from 21.2 ± 13.9% to 91.2 ± 2.8% (Andersen et al. 1996). This dramatic transformation is to be expected from peripheral forms of stimulation, which entail a reversed recruitment order of motor unit firing. The fastest fibres, which undergo a phasic duty cycle in intact muscle, are also the first activated by peripheral systems, thus being highly susceptible to transformation. In contrast, ISMS through a single electrode allowed for fibre type selectivity and recruited a mixed fibre type at both 1.2T and 3.0T conditions. The capacity of ISMS to recruit slow fibres while leaving fast fibres in reserve should aid in maintaining a mixed fibre type and should prove helpful in rescuing the normal mixed-muscle phenotype in chronic SCI cases. Further studies with ISMS should be designed to address the capacity of this novel rehabilitative therapy to rescue the normal mixed-muscle phenotype in a chronic SCI model.