Leg joint power output during progressive resistance FES-LCE cycling in SCI subjects: developing an index of fatigue

Six spinal cord injured subjects participated in the study. All subjects had previous experience with FES cycle ergometry, were between the ages of twenty and fifty years old, and possessed either a complete or incomplete spinal cord injury at or below the fourth cervical vertebra. Two subjects had incomplete injuries, four had complete injuries. All subjects signed a consent form, explaining the terms and conditions of the study in agreement with the Institutional Review Board of the University.

The ERGYS I™ (Therapeutic Alliances^®, Inc., Fairborn, OH) semi-reclined cycle ergometer was used in this study. Resistance was produced by increasing the tension of a friction-induced band applied to the perimeter of the flywheel and secured to the ergometer frame. Tensions required to produce ergometer power outputs of 0, 6.25 and 12.5 Watts were determined assuming a constant cadence of 50 rpm. 6.25W of ergometer power is equivalent to a resistance of 1/8^th kp. A digital speedometer was attached to the front of the ergometer, allowing the subject to monitor their current cadence levels. If cadence levels fell below 10 rpm, stimulation was terminated and subjects were assisted in a passive cycling cool-down. Stimulation was supplied by the ergometer in the form of a sinusoidal, biphasic waveform with a pulse duration of 500 μsec, a phase duration of 1000 μsec and frequency of 50 Hz. Maximum stimulation intensity was 140 mA. Since each muscle contracted during a different phase of cycle rotation, pre-programmed sensors were used to stimulate appropriate muscle groups at specific crank angles (Figure 1). Seat depth of the ergometer was adjusted horizontally for each rider so that the rider's knee was not fully extended when the pedal reached an angle of 110° with respect to top dead center (TDC). The subject's feet were secured within boots attached to the ergometer pedal and the thighs were secured with Velcro straps to restrict movement perpendicular to the sagittal plane. A complete setup of the subject in the FES-LCE system can be viewed in Figure 2.

Stimulation was delivered to the quadriceps, hamstrings and gluteus maximus muscles of each leg using two oval, 2.5" × 3.25"self adhesive surface electrodes for each muscle group, figure 3. Each active/ground electrode set was independently stimulated and grounded, eliminating the potential for co-contraction. Surface electrodes were arranged so that the quadriceps ground electrode was placed a distance of approximately 6 cm superior to the patella. The active electrode was placed approximately 10 cm superior and slightly lateral so that it rested over the muscle belly of the rectus femoris and vasti. The hamstring ground electrode was placed a distance of approximately 6 cm superior to the posterior crease of the knee joint; situated approximately over the semitendinosus and short head of the femoral bicep. The active electrode was placed a distance of 10 cm superior so that the muscle bellies of the semitendinosus, semimembranosus and long head of the femoral biceps were covered. The gluteus maximus ground electrode was placed at the gluteal fold. The active electrode was placed approximately 4 cm superior and anterior so as to rest on the muscle belly of the gluteus maximus. To maintain consistency, electrode placement was performed by the same person for all subjects.

Adhesive reflective markers were placed over the humeral head of the shoulder, approximately 1 cm anterior and superior to the tip of the greater trochanter, at the center of the lateral femoral epicondyle, the lateral tip of the malleolus, and on the lateral side of the ergometer boot at the approximate location of the fifth metatarsal. Additional markers were placed at the center of the pedal spindle, on the lateral side of the pedal force sensor, at the heal of the ergometer boot, and the ergometer crank center, figure 2. A reference was placed on the frame of the ergometer, vertically aligned with the crank center. The reflective markers were illuminated using a flood light and continuously video recorded, using the image from a video camera oriented perpendicular to the rider's sagittal plane. A video-based motion capture software system (Peak Motus^® System, Peak Performance Technologies Inc., Denver, CO) was used to measure the displacement of the crank tip, hip, knee, and ankle joints during cycling (Figure 4). Displacement data were measured at a frequency of 60 Hz. The kinematic data were filtered using a 5^th-order, zero-lag, low pass Butterworth filter with a cutoff frequency of 4 Hz. All data were expressed as a function of crank angle. One complete rotation was the trajectory of the crank tip as it moved from TDC to bottom dead center (BDC) and back to TDC (Figure 5). All calculations were averaged over the last 10 rotations completed at each resistance level to represent the steady-state values of each parameter of interest.

A 4-pin, triax, ICP^® piezo-electric force transducer (PCB Piezotronics, Inc., NY) with a full-scale measurement range (45 to 22 k N compression, 2200 N tension) was mounted underneath the boot of the right ergometer pedal to record pedal forces in normal and tangential directions with respect to the pedal plane. The force data were collected at a frequency of 180 Hz using a LABView program (National Instruments Corp^®, Austin, TX), developed for the study. A synchronizing signal was transmitted to the kinematic video recording at the onset of pedal force acquisition, allowing visual synchronization of the pedal force and video-recorded kinematic data.

A progressive intensity test was conducted in order to investigate stimulation intensity, joint power outputs, cadence, resultant pedal force (RPF) and resistance as possible indicators of fatigue. In the context of this test, fatigue was defined as the point when the subject could no longer cycle at or above 10 rpm. The subjects were passively pedalled through a 2 minute warm-up without stimulation. Following the warm-up, stimulation intensity was applied so that subjects maintained a constant cadence of 50 rpm throughout the exercise. 50 rpm was chosen at the target cadence since this was the cadence at which the tension of the friction belt was determined in order to equal 6.25W or 1/8^th kp. Resistance was then increased by 1/8^th kilo-pound (kp) (6.25 W) every two minutes. A gradient increase of 2 minutes was chosen to minimize muscular fatigue, but still be considered long enough to achieve steady-state conditions. Ninety seconds after each resistance adjustment, force data were recorded for 30 seconds. Stimulation intensity (mA) and cadence were also recorded at this time. 140 mA was considered 100% stimulation. The subject's heart rate, blood pressure and pulse oxygen concentration (SPO2) were also recorded to ensure that the subject maintained normal metabolic behaviours. Resistance was increased only to a level that each subject felt comfortable. If a rider's cadence dropped below 10 rpm, stimulation was automatically terminated and the subject was passively cycled through a 2-minute cool-down.

Final values of stimulation intensity and HPO were significantly higher than initial values at 0/8^th kp (116.2 ± 28.1 mA vs. 66.6 ± 20.3 mA, p = 0.03, and 27.1 ± 30 mW vs. -88.5 ± 30 W, p < 0.01, respectively). Stimulation intensity increased 23% between the first two levels of resistance and an additional 13% by the final resistance level of 2/8^th kp. HPO increased 59% from 0/8^th to 1/8^th kp, and increased by 41% at 2/8^th kp. 100% HPO corresponded to the maximum power output observed at 2/8^th kp. (Figure 6)

Mean cadence remained close to the target cadence of 50 rpm for the first two resistance levels, experiencing only a slight decrease of 0.2 rpm (Table 1). Mean cadence decreased by 7 rpm between 1/8^th and 2/8^th kp. Changes observed in mean cadence were not statistically significant between any of the resistance levels. (p = 0.12)

Mean APO and KPO did not change significantly with increased resistance. Mean APO decreased slightly from 0/8^th to 1/8^th kp, and decreased by 35% by 2/8^thkp (Figure 7). KPO decreased by 20% between the first two levels of resistance and decreased another 59% between 1/8^th and 2/8^th kp (Figure 8).

RPF did not change significantly with increasing resistance. From 0/8^th to 1/8^th kp and 1/8^th to 2/8^th kp, RPF increased by 3.5% and 6.1%, respectively (Figures 9, 10).

Regression lines for APO and KPO were developed to calculate indices of fatigue. APO was predicted by cycling cadence (R² = 0.75, p < 0.001). APO = 0.002C - 0.074. Considering a cadence of 10 rpm as fatigue, ankle fatigue occurs at -0.054 W. As a result, the index of fatigue represents the proportion of power present at the ankle joint with respect to fatigue:

Where APO _M is ankle power output measured during cycling. As APO approaches fatigue, the index will approach 0.

A multiple linear regression of cycling cadence (C), stimulation intensity (S) and RPF (R) provided a significant predictor of KPO = 0.006C - 0.002S + 0.004R - 0.17 (R² = 0.94, p < 0.05). At fatigue, cadence and stimulation intensity were defined as 10 rpm and 100%, respectively. Since RPF does not have a definitive point of fatigue that can be measured, the mean value of RPF at 2/8^th kp was used.

Incorporating this value into the regression equation for KPO indicates that if fatigue occurs at 2/8^th kp, KPO will be -209.88 mW, absorbing power. The index of fatigue developed for KPO presents the proportion of existing KPO with respect to fatigue:

where KPO _M is measured during cycling and KPO _F is the value at fatigue.

Therefore, as KPO approaches fatigue, the index will approach a value of 0. The fatigue indices were applied to the group mean APO and KPO values (Table 3). Application of the ankle fatigue index indicates that at 0/8^th kp, the ankle joint has 1.7 times as much power than levels calculated for fatigue. Additionally, the ankle power index decreases to 1.45 by 2/8^th kp. The knee fatigue index at 0/8^th kp was 1.6 times greater than fatigue. The decrease in knee fatigue index to 1.1 at 2/8^th kp indicates a greater reduction in knee power capacity than that experienced at the ankle. The regression equation for HPO was HPO = 0.009R - 0.085 (p < 0.001), whereby resistance was the only predictor (R² = 0.65). Therefore, no factors define HPO at fatigue and an index of hip fatigue cannot be developed.

The main findings of this study were: (1) APO and KPO decreased with increasing resistance whereas HPO increased with resistance; (2) cadence, stimulation intensity, and RPF were significant predictors of KPO; and (3) knowing the value of these predictors at 10 rpm, an index of fatigue can be developed. This index quantitatively expresses the power generated at the knee joint with respect to a baseline power level defined as fatigue.

The absence of lower leg stimulation in this exercise suggests that changes observed in APO must be attributed to changes in ankle angular velocity. Studies have suggested that plantar-flexion of the ankle joint during late recovery phase in both able-bodied and SCI subjects was a direct result of boot-pedal inertial forces [29]. Such inertial forces would be augmented by changes in cadence [30]. The subsequent changes in mean APO observed in this exercise demonstrated a strong link to changes in cadence, supporting the identification of cadence as the only significant predictor of APO. The inclusion of lower leg stimulation to such muscles as the tibialis anterior, soleus and gastrocmenius would likely improve cycling performance and potentially increase the circulatory and cardiovascular benefits to the user [27, 31‐34].

The increase in stimulation intensity and subsequent decrease in knee power by 1/8th kp, 3.5 minutes into cycling, are most likely due to premature fatiguing of the quadriceps and/or hamstrings [12‐14]. However, the exact muscle group that fatigued cannot be identified since all muscle groups received identical increases in stimulation when cadence dropped below target. Despite potential fatigue, 86% of the power generated within the leg was produced at the knee at 1/8^th kp. Research involving able-bodied subjects has reported the quadriceps to remain active over the largest range of the cycling period [23]. The quadriceps muscles also exhibit the greatest strength in both paralyzed and able-bodied riders and make a larger contribution to forward cycling than the hamstrings and gluteus muscles [18, 20, 23, 27]. Therefore, the quadriceps may be primarily responsible for knee power output in this exercise, but additional studies would be required to confirm this conclusion. Cadence-dependent inertial forces may have contributed to knee power output as well. The strong correlation between ankle and knee power outputs (r = 0.889, Table 2) may reflect the boot-pedal's contribution to knee joint moment. Also, moments created by thigh and shank inertial forces may have further increased knee power output. The decrease in KPO at 2/8^th kp was likely due not only to a decrease in knee angular velocity, but also from a decline in inertial force associated with decreased cadence, subsequently compromising knee joint moment.

The hip absorbed power for the first two resistance levels (Tables 1 and 3). Power absorption is specified by a negative power value and indicates that the direction of muscle force production opposes the direction of crank motion. Power absorption may result from fatigue, improper coordination of flexor and extensor muscle contractions, or more likely, from opposing inertial forces produced by the thigh segment that intensify at higher cadence levels [21, 30]. The transition to hip power exertion at 2/8^th kp indicates a concurrence in both force and crank directions, and is most likely due to an increase in muscle forces resulting from a decrease in ergometer cadence [35]. The significant increase observed in HPO is consistent with Ericson's findings involving able-bodied subjects [23, 33]. The higher cadences observed at 0/8^th and 1/8^th kp, may have produced thigh inertial forces that were larger in magnitude and opposite in direction to the force produced by the gluteus muscles. As a result, the hip joint absorbed power. At 2/8^th kp, the decrease in ergometer cadence resulted in a lower thigh inertial force. While the inertial force still opposed the direction of muscle force generation, the magnitude of the gluteus muscle force was sufficiently larger, promoting power exertion. The transition from power absorption to exertion at the hip with decreased cadence supports previous findings that suggest lower ergometer cadences to be favourable to higher muscle forces [35]. From a mechanical perspective, increased ergometer resistance incurred an increasingly larger moment about the hip. To compensate for this increase, the gluteal muscles increased force production, subsequently increasing HPO. Ergometer cadence and moment arm (pedal to hip) did not change significantly, therefore their influence on HPO is negligible. Since the hip was the only joint to exhibit an increase in power output in this exercise, the gluteal muscles may be the primary source of both power and continual force generation during FES-cycling in SCI subjects [21].

The index of fatigue developed in this study assesses a subject's capacity to pedal as described by their KPO. The knee was selected for the definition of this index due to its close proximity to the quadriceps and hamstring muscle groups. As demonstrated by the knee's regression equation, different combinations of stimulation and cadence may be examined to investigate riding conditions that may augment riding time and increase power output. The fatigue index value at 0/8^th kp, measured 90 seconds into active, unassisted pedaling was 1.6. This indicates initial power capacity at the knee to be 1.6 times greater than fatigue. The fatigue index decreased to 1.1 at 2/8^th kp, representing approximately a 30% decrease in the knee's power capacity within a 4 minute time span. These findings suggest that the present cycling protocol is not sufficient for a rider to gain the benefits of FES and thus raises speculation as to whether or not progressive resistance cycling is an appropriate protocol for SCI subjects. The index values derived from these five subjects, should be expanded and generalized to a larger population. If similar results emerge in future studies, this index may be useful within a clinical or experimental setting.

Mixing the results of both complete and incomplete SCI subjects was not considered a significant influence on the final outcome of this study. One of the incomplete SCI subjects produced results comparable to, and in some cases, better than complete SCI subjects. From this observation, it was determined that degree and frequency of ergometer use was a greater influence on the study results than extent of injury. Furthermore, using a mixed subject group of complete and incomplete SCI subjects is a truer representation of the FES-LCE user population. Variation existed within the subject group due to differences in age, time since injury (1.5 to 21 years), frequency and extent of cycling experience, as well as a likely difference in proportion of fast- to slow-twitch muscle fibers present in the stimulated muscle groups associated with variable atrophy. Previously, female SCI subjects have scored significantly higher on an endurance index than their male counterparts whereas no gender differences were noted between able-bodied control subjects in that particular study [31]. Other studies have suggested that the properties of muscle fiber in females may differ from males [36]. Future studies may investigate how the presently developed indices of fatigue may differ with gender. Additionally, alterations in stimulation timing and intensity could produce significantly different muscle synergies during FES cycling. Resulting joint power outputs and the subsequent development of a fatigue index would ultimately be affected. Therefore, the results of this study are valid for the presently employed stimulation protocol only. Additional research would be required to verify whether the present fatigue index is valid under different stimulation parameters.