Kinematic and kinetic benefits of implantable peroneal nerve stimulation in people with post-stroke drop foot using an ankle-foot orthosis

Abstract

Background:

Contralesional ‘drop foot’ after stroke is usually treated with an ankle-foot orthosis (AFO). However, AFOs may hamper ankle motion during stance. Peroneal functional electrical stimulation (FES) is an alternative treatment that provides active dorsiflexion and allows normal ankle motion. Despite this theoretical advantage of FES, the kinematic and kinetic differences between AFO and FES have been scarcely investigated.

Objective:

To test whether walking with implanted FES leads to improvements in stance stability, propulsion, and swing initiation compared to AFO.

Methods:

A 4-channel peroneal nerve stimulator (ActiGait^®) was implanted in 22 chronic patients after stroke. Instrumented gait analyses were performed during comfortable walking up to 26 weeks (n = 10) or 52 weeks (n = 12) after FES-system activation. Kinematics of knee and ankle (stance and swing phase) and kinetics (stance phase) of gait were determined, besides spatiotemporal parameters. Finally, we determined whether differences between devices regarding late stance kine(ma)tics correlated with those regarding the swing phase.

Results:

In mid-stance, knee stability improved as the peak knee extension velocity was lower with FES (β = 18.1°/s, p = 0.007), while peak ankle plantarflexion velocity (β = –29.2°/s, p = 0.006) and peak ankle plantarflexion power (β = –0.2 W/kg, p = 0.018) were higher with FES compared to AFO. With FES, the ground reaction force (GRF) vector at peak ankle power (i.e., ‘propulsion’) was oriented more anteriorly (β = –1.1°, p = 0.001). Similarly, the horizontal GRF (β = –0.8% body mass, p = 0.003) and gait speed (β = 0.03 m/s, p = 0.015) were higher. An increase in peak ankle plantarflexion velocity and a more forward oriented GRF angle during late stance were moderately associated with an increase in hip flexion velocity during initial swing (r_s = 0.502, p = 0.029 and r_s = 0.504, p = 0.028, respectively).

Conclusions:

This study substantiates the evidence that implantable peroneal FES as a treatment for post-stroke drop foot may be superior over AFO in terms of knee stability, ankle plantarflexion power, and propulsion.

Keywords

Functional electrical stimulation peroneal nerve ankle-foot orthosis stroke gait rehabilitation

1 Introduction

A common gait problem after supratentorial stroke is the occurrence of a hemiparetic ‘drop foot’ (Kottink et al., 2004). Paresis of the ankle dorsiflexors, sometimes accompanied by plantarflexor spasticity, causes an equinus or equinovarus position of the foot during the swing phase and early stance. As a result, patients with drop foot have problems with foot clearance and pre-positioning for loading, which makes them prone to tripping and falling (Weerdesteyn et al., 2008). An ankle-foot orthosis (AFO) is commonly used to compensate post-stroke drop foot. Typically, AFOs provide passive dorsiflexion during the swing phase (Mulroy et al., 2010), thereby reducing the risk of tripping (Cakar et al., 2010). However, depending on their design and stiffness, most AFOs hamper normal ankle motion and push-off during the stance phase (Mulroy et al., 2010; Nair et al., 2010; Vistamehr et al., 2014). An alternative for AFO is functional electrical stimulation (FES) of the common peroneal nerve (Liberson et al., 1961). Peroneal FES provides ‘active’ ankle dorsiflexion during the swing phase and supports a gentle foot landing during the first ankle rocker without interfering with ankle motion during mid-stance or push-off. However, despite this theoretical advantage of FES over AFO, both interventions have been found to be equally effective for improving common gait-related outcomes such as gait speed and physical activity (Prenton et al., 2016). Possibly, these outcomes are not sensitive to capture the biomechanical benefits of FES over AFO.

In line with the theoretical advantage of FES over AFO, several groups have shown that ankle kinematics can indeed be improved by replacing AFO with FES (Ernst et al., 2013; Kottink et al., 2012; Sheffler et al., 2013). In the same vein, we recently reported that late stance ankle plantar flexion was enhanced while walking with implantable peroneal FES, which led to a substantial increase in peak plantarflexion power during push-off and better step-length symmetry (Schiemanck et al., 2015). However, in the latter paper we limited our kinematic and kinetic analysis to the ankle joint. While late stance ankle plantarflexion power is indeed highly important for generating propulsion (Liu et al., 2008; Neptune et al., 2008; Neptune et al., 2004; Zajac et al., 2003), through interjoint coupling, it is also a key mechanism for providing knee flexion and initiating leg swing (Goldberg et al., 2004; Neptune et al., 2008). Enabling people with stroke to utilize all of their residual ankle plantarflexion power may, therefore, improve propulsion as well as knee and hip flexion during (pre-) swing (Campanini et al., 2013; Peterson et al., 2010). In our previous single-case report of a successful peroneal nerve implantation we have been able to demonstrate improvements both in paretic ankle kine(ma)tics and in knee and hip flexion during swing while walking with FES compared to AFO (van Swigchem et al., 2011), but this result has not yet been replicated.

Besides a lack of propulsion during late stance and limited knee flexion during swing, a subset of stroke patients with drop foot also suffer from a knee (hyper)extension thrust during mid-stance (Tani et al., 2016). In these patients, a lack of knee control and/or troublesome ankle plantarflexor spasticity limits the ability to smoothly dorsiflex the ankle during mid-stance (i.e., tibial progression during 2nd rocker) (Bleyenheuft et al., 2010), which is detrimental for stance phase stability and gait efficiency. Depending on the stiffness, AFOs may also hamper normal dorsiflexion during mid-stance and, thus, facilitate this knee instability towards extension. Conversely, through replacing an AFO by peroneal FES, knee stability may be improved. In a recent case study, peroneal FES indeed showed to be a promising tool in management of knee (hyper)extension during stance (Chantraine et al., 2016), but this effect has never been investigated in a group of stroke patients using an AFO.

With this study, we extend our previous work on the benefits of implantable peroneal nerve stimulation (Schiemanck et al., 2015) by investigating kinematic differences (stance and early swing phase) and kinetic differences (stance phase) between FES and AFO in a larger sample of patients with chronic drop foot after supratentorial stroke. We tested the hypothesis that walking with FES would lead to improvements in the kinematics and kinetics of the ankle and knee during the stance phase of the paretic leg, thereby improving the knee and hip kinematics of the paretic swing phase as well as the spatiotemporal symmetry of gait. In this study, we specifically focus on knee stability in mid-stance, propulsion, and swing initiation.

2 Methods

2.1 Participants

Data of two longitudinal observational cohorts were pooled. Our first Actigait cohort (Group I) consisted of 10 patients who were followed-up for 26 weeks (Schiemanck et al., 2015). Our second Actigait cohort (Group II) comprised another 12 patients with a follow-up of 52 weeks. All participants were recruited from the outpatient clinic of the departments of Rehabilitation at the Radboud University Medical Center in Nijmegen and at the Academic Medical Center in Amsterdam. Only patients with an established clinical diagnosis of unilateral supratentorial stroke (either ischemic or hemorrhagic) and who were at least 6 months post stroke were eligible. Patients were included when they used an AFO for the compensation of foot dorsiflexion weakness (Medical Research Council scale <5) and showed a positive response to surface-based peroneal nerve stimulation (NESS L300, Bioness inc, Valencia, California). All patients needed to be able to walk independently for 10 minutes without supervision or walking aid, except an AFO (Functional Ambulation Categories 4 or 5). For other inclusion and exclusion criteria we refer to our previous publication (Schiemanck et al). At inclusion, balance capacity (Berg Balance Scale), motor impairment of the paretic leg (Motricity Index and Fugl-Meyer Assessment), calf muscle spasticity of the paretic leg (Modified Ashworth Scale), and vibration threshold at the affected ankle and forefoot (graduated Rydel-Seiffer tuning fork) were evaluated by the treating physician.

2.2 Study design

A within-subjects repeated measures design was used for both cohorts. Gait analyses were performed at inclusion (T0) as well as 2 weeks (T1), 8 weeks (T2), 26 weeks (T3), and 52 weeks (T4, Group II only) after activation of the ActiGait^® system. At T0, these gait measures were obtained only with the AFO. From T1, gait was always assessed with AFO and FES on the same day, see Fig. 1. At these instances, walking with AFO and FES was tested separately and in a balanced order across participants to neutralize possible order effects, e.g. related to fatigue. Patients always used their own AFO and wore the same footwear throughout all measurements.

Fig.1

Timeline of the two study cohorts. The start of activation of the ActiGait system took place three weeks after implantation. Thereafter, the system was increasingly used for an extra 15 minutes per day until whole day use. AFO = ankle-foot orthosis; FES = implanted functional electrical stimulation.

2.3 Intervention

The ActiGait^® system (Neurodan, Otto Bock, Germany) is a 4-channel peroneal nerve stimulator. The implantable parts consist of a stimulator body at the proximal end and a cuff electrode with four separate electrodes at the distal end, connected by a lead wire. The stimulator body selectively controls 4 electrodes embedded in the cuff, which allows differential activation of nerve fibers to the tibialis anterior, peroneus longus/brevis, and toe extensor muscles. We refer to our previous publication for a more detailed description of the system, surgical procedure and activation procedure (Schiemanck et al., 2015).

2.4 Measurements

Instrumented gait analyses were performed to obtain spatiotemporal, kinematic, and kinetic outcome measures. Subjects walked at a self-selected, comfortable speed along a walkway of approximately 10 m. A 3D-movement analysis system (Vicon Motion Systems, Inc., Lake Forest, CA,USA) was used to measure the trajectories of 16 reflective markers placed on key anatomic landmarks at a sampling rate of 100 Hz. Markers were bilaterally placed on the pelvis and on both legs: anterior superior iliac spine, posterior superior iliac spine, lateral epicondyle of the femur, lateral malleolus of the ankle, heel, dorsal head of the second metatarsal bone, and two markers on the lateral aspect of the thigh and lower leg. Trajectories of these markers were recorded within the measurement volume of the 3D-movement analysis system, which was approximately from 2 to 8 m on the walkway. Position and displacement of the markers were analysed using the manufacturer’s ‘PlugInGait’ model. Simultaneously, ground reaction forces were recorded from two force plates at a sampling rate of 1000 Hz (Advanced Mechanical Technology (AMTI), Inc., Watertown, MA, USA). Five walking trials were captured per condition (AFO/FES) for each subject.

2.5 Data analysis

Raw motion capture data was filtered using a Woltring filtering routine (Woltring, 1986). Gait events (heel strike and toe-off), joint kinematics and kinetics were derived from the Vicon PlugInGait model and processed with Matlab (The Mathworks Inc. Natick, MA, USA). The values of all outcome measures were averaged across the five trials.

2.5.1 Mid-stance knee kinematics (stance stability)

A lack of knee control can cause instability during mid-stance. Such instability usually presents as either excessive knee flexion or rapid knee extension (i.e., knee (hyper)extension thrust). Excessive knee flexion is often the result of calf muscle weakness, whereas knee extension thrust is typically caused by ankle plantarflexor spasticity. To quantify the knee extension thrust on the paretic side during mid-stance, peak knee extension velocity was determined (in patients without excessive knee flexion). Knee angular velocity was computed by taking the first derivative of the knee angle. Then the maximum value between each ipsilateral heel strike and toe-off was established. For each step, the point where peak extension velocity was reached was visually controlled by the primary researcher (FB) to make sure that 1) maximum knee flexion was smaller than 15° (to eliminate participants with an excessive knee flexion pattern) and 2) the point of peak extension velocity occurred during a knee extension motion of minimally 5° (to avoid the identification of very small, meaningless extension peaks).

2.5.2 Late stance ankle kinematics and kinetics (propulsion)

To quantify the push-off during late stance, maximum ankle plantarflexion angle, maximum ankle plantarflexion velocity (first derivative of the plantarflexion angles), and maximum ankle plantarflexion power were established. Maximum values were identified between 30 and 100% of the stance phase (heel strike to ipsilateral toe-off) to avoid the identification of ankle plantarflexion during early stance. In addition, we determined the orientation of the ground reaction force at the instant of maximum plantarflexion power, expressed as the angle relative to the vertical (GRF angle), as well as the maximum ground reaction force in the anteroposterior direction (AP-GRF).

2.5.3 Swing hip and knee kinematics (swing initiation)

Maximum knee flexion angle, maximum knee flexion velocity (first derivative of the knee flexion angles), maximum hip flexion angle, and maximum hip flexion velocity (first derivative of the hip flexion angles) were calculated. Since knee flexion starts during pre-swing, these values were identified between 40% and 100% of the gait cycle.

2.5.4 Spatiotemporal characteristics (speed and symmetry)

We also determined gait speed and step-length asymmetry as basic spatiotemporal variables. Gait speed was calculated as the displacement (meters) of the heel marker in the anteroposterior direction between the first and last heel strike divided by the time from first to last heel strike (seconds). Step length (SL) on one body side was determined as the distance from the ipsilateral heel to the contralateral heel. Step-length asymmetry was expressed in a symmetry index (SI) (Patterson et al., 2010): $\begin{matrix} {SI}_{SL} = | ({SL}_{paretic} - {SL}_{nonparetic}) \\ / (0.5 * ({SL}_{paretic} + {SL}_{nonparetic})) | * 100 % \end{matrix}$

2.6 Statistics

In previous work maximum ankle plantarflexion angle, peak ankle plantarflexion power, step-length asymmetry and gait speed were reported for Group I (Schiemanck et al., 2015). For the current research report data from Group I was re-analysed, analyses were extended, and results were combined with the data from Group II. All outcome measures were tested using Generalized Estimated Equations modeling (GEE, autoregressive correlation structures). As we were primarily interested in the differences between FES and AFO, a model with the (time-integrated) factor Device (T1–T4) was tested first (Step 1). Then the factor Time (T1–T3) and its interaction with Device were added to the model (Step 2) to assess whether there were any learning effects during the study period, and whether time was of influence on the effects of Device. For the assessment of Time effects we neglected the final assessment (T4) of Group II to avoid finding Time effects attributable to group differences. Finally, we determined whether (time-integrated) Device effects on late stance characteristics (maximum ankle plantarflexion velocity and power, AP-GRF, GRF-angle) were correlated with those on swing characteristics (maximum knee and hip flexion velocity). To this end, differences between FES and AFO (FES-AFO) were averaged over time for all subjects. We calculated Spearman’s rank correlation coefficients, because we did not expect the differences between FES and AFO to be normally distributed. Level of significance was set at p < 0.05. All statistical analysis were performed in SPSS 22.0 (IBM Corp. Armonk, NY, USA).

3 Results

3.1 Patient inclusion and characteristics

Twenty-two participants were included of whom three were removed from the analysis, see Table 1. One participant (Group I) suffered peroneal nerve damage after surgery (which showed full recovery after one-and-a-half year). One participant (Group II) died shortly after activation of the ActiGait system, the cause of death being unrelated to the study. No follow-up data of these two participants could be collected. The third participant (Group I) had severe calf muscle clonus in reaction to FES, which needed to be treated with repeated intramuscular injections with botulinum toxin. Because of this additional treatment, this participant deviated too much from the study protocol. The ActiGait implant of one participant (Group II) failed after 26 weeks, yet sufficient data was collected to include this participant in the analysis. Hence, data of 19 participants from the two combined study cohorts was analysed.

Table 1
Demographic and clinical characteristics of both study cohorts obtained at baseline (T0)

n = 19

Age (mean yrs±SD) 54.4±12.3

Body weight (mean kg±SD) 83.3±15.2

Body height (mean cm±SD) 177.0±7.2

Time since stroke (mean months±SD) 59.9±44.0

Sex (men/women) 14/5

Paretic side (left/right) 11/8

Type of stroke (ischemic, hemorrhagic) 14/5

Motricity index – paretic leg (median (range)) 77 (28–99)

Fugl-Meyer Assessment – paretic leg (median % (range)) 71 (25–88)

Berg Balance Scale (median (range)) 53 (42–56)

Modified Ashworth Scale calf (median (range)) 1 (0–2)

Vibration threshold paretic forefoot (median (range)) 4 (0–8)

Vibration threshold paretic ankle (median (range)) 4 (0–8)

Type of AFO used (fixed/hinged) 17/2

Baseline comfortable walking speed using AFO (mean m/s±SD) 0.94 (0.2)

	n = 19
Age (mean yrs±SD)	54.4±12.3
Body weight (mean kg±SD)	83.3±15.2
Body height (mean cm±SD)	177.0±7.2
Time since stroke (mean months±SD)	59.9±44.0
Sex (men/women)	14/5
Paretic side (left/right)	11/8
Type of stroke (ischemic, hemorrhagic)	14/5
Motricity index – paretic leg (median (range))	77 (28–99)
Fugl-Meyer Assessment – paretic leg (median % (range))	71 (25–88)
Berg Balance Scale (median (range))	53 (42–56)
Modified Ashworth Scale calf (median (range))	1 (0–2)
Vibration threshold paretic forefoot (median (range))	4 (0–8)
Vibration threshold paretic ankle (median (range))	4 (0–8)
Type of AFO used (fixed/hinged)	17/2
Baseline comfortable walking speed using AFO (mean m/s±SD)	0.94 (0.2)

3.2 Mid stance knee kinematics (stance stability)

Our analysis focused on the difference between AFO and FES on knee instability towards extension on the paretic side during mid-stance. Two subjects showed instability towards knee flexion, i.e. excessive (>15°) knee flexion in mid-stance, and were thus discarded from this analysis. Peak knee extension velocity was significantly different between the two devices and was on average 14% lower with FES compared to AFO (Device, β = 18.1°/s, p = 0.007) (see Table 2 and Fig. 2a). The Time effect was also significant (p = 0.039). Knee extension velocity was slightly higher at 26 weeks after system activation compared to 2 weeks. There was no significant Device*Time interaction.

Fig.2

Estimated mean (time integrated) values and standard errors for AFO and FES for each of the stance-phase parameters (n = 19). AFO = ankle-foot orthosis; FES = implanted functional electrical stimulation. GRF angle = angle of ground reaction force with the vertical at the instance of peak ankle power; AP-GRF = ground reaction force in anteroposterior direction. n.s = not significant difference; * = p < 0.05; ** = p < 0.01; *** = p < 0.001.

Table 2

Estimated marginal means for both study cohorts combined (T1-3) (n = 19)

Outcome	Time	Mean±SE	Statistics
		AFO	FES	Device	Time
Maximum knee extension velocity (degrees/s)	T1	134±18	116±16	p = 0.007	p = 0.039
	T2	133±16	122±14
	T3	146±17	133±21
Maximum ankle plantarflexion angle (degrees)	T1	–4.09±1.32	–1.51±2.06	p = 0.07	p = 0.125
	T2	–3.12±1.34	–1.56±1.73
	T3	–1.31±1.33	–0.13±1.53
Maximum ankle plantarflexion velocity (degrees/s)	T1	133±11	171±25	P = 0.006	P = 0.312
	T2	121±8	149±13
	T3	134±9	161±17
Maximum ankle plantarflexion power (Watt/kg)	T1	1.30±0.15	1.45±0.15	P = 0.018	P = 0.970
	T2	1.22±0.15	1.57±0.16
	T3	1.28±0.15	1.47±0.17
GRF angle (degrees)	T1	7.25±0.80	8.37±0.97	P = 0.001	P = 0.887
	T2	7.20±0.80	8.77±0.90
	T3	7.35±0.80	8.37±0.95
Maximum GRF in AP direction (% body weight)	T1	8.80±0.95	9.81±1.01	P = 0.003	P = 0.740
	T2	8.60±0.88	9.65±0.88
	T3	8.62±0.98	9.45±0.89
Maximum knee flexion angle (degrees)	T1	40.61±2.98	40.62±2.96	P = 0.798	P = 0.028
	T2	44.59±3.23	43.92±3.48
	T3	42.13±3.20	44.30±3.16
Maximum knee flexion velocity (degrees/s)	T1	244±23	247±22	P = 0.819	P = 0.066
	T2	253±23	262±22
	T3	256±23	270±26
Maximum hip flexion angle (degrees)	T1	32.44±1.22	33.70±1.38	P = 0.637	P = 0.002
	T2	33.89±1.22	34.48±1.05
	T3	35.67±1.13	36.17±1.24
Maximum hip flexion velocity (degrees/s)	T1	142±16	137±16	P = 0.696	P = 0.687
	T2	136±11	132±9
	T3	140±13	144±18
Gait Speed (m/s)	T1	0.93±0.05	0.95±0.05	P = 0.015	P = 0.345
	T2	0.91±0.05	0.96±0.05
	T3	0.94±0.05	0.98±0.04
Step-length asymmetry index (% )	T1	10.21±2.38	8.78±1.76	P = 0.852	P = 0.717
	T2	9.30±2.37	8.33±2.30
	T3	8.91±1.72	9.15±2.02

AFO = ankle-foot orthosis. FES = implanted functional electrical stimulation.

3.3 Late stance ankle kinematics and kinetics (propulsion)

There was a trend towards a larger maximum ankle plantarflexion angle with FES compared to AFO (Device, β = –1.6°, p = 0.07) (see Table 2). On average, both devices did not provide true plantarflexion in late stance as can be seen by the negative (i.e., dorsiflexion) values for maximum plantarflexion in Fig. 2b. Yet, there was a significant difference between the two devices for both peak ankle plantarflexion velocity (Device, β = –29.2 °/s, p = 0.006) and peak ankle plantarflexion power (Device, β = –0.2 W/kg, p = 0.018). With FES, plantarflexion velocity and plantarflexion power were on average 22% and 17% higher, respectively (see Table 2 and Figs. 2c and 2d). GRF angle at the instant of peak ankle power significantly changed towards a more forward orientated angle with FES compared to AFO (Device, β = –1.1°, p = 0.001). In the same vein, maximum AP-GRF was significantly larger with FES compared to AFO (Device, β = –0.8 % body weight, p = 0.003). On average, GRF angle and maximum AP-GRF increased by 14% and 10% with FES, respectively (see Table 2 and Figs. 2e and 2f). There were no other significant (interaction) effects for maximum ankle plantarflexion angle, peak ankle plantarflexion velocity, peak ankle plantarflexion power, GRF angle or maximum AP-GRF.

3.4 Swing hip and knee kinematics (swing initiation)

There was no significant difference between AFO and FES for maximum knee flexion angle (Device, β = 0.4°, p = 0.798) or maximum knee flexion angular velocity (Device, β = 19.0°/s, p = 0.819). Likewise, there was no significant difference between AFO and FES for maximum hip flexion angle (Device, β = –0.2°, p = 0.637) or maximum hip flexion angular velocity (Device, β = 15.0°/s, p = 0.696) (see Table 2). Time was a significant factor for both peak knee flexion angle (p = 0.028) and peak hip flexion angle (p = 0.002) during swing. Compared to the first follow-up measurements (T1), peak knee flexion angles were significantly larger after 8 weeks (Time_t2, β = 4.0°, p = 0.006) and peak hip flexion angles were significantly larger after 26 weeks (Time_t3, β = 3.2°, p = 0.006).

3.5 Spatiotemporal characteristics (speed and symmetry)

There was a significant difference between AFO and FES for comfortable walking speed, indicating that participants walked faster with FES (Device, β = –0.03, p = 0.015). On average subjects walked 3% faster with FES compared to walking with their AFO (see Table 2). Further analysis revealed no other significant (interaction) effects for walking speed. There was no significant difference between AFO and FES for step-length asymmetry (Device, p = 0.852). In our previous report (Schiemanck et al., 2015), step-length asymmetry was significantly different between devices. As Group II presented with fairly low values of step-length asymmetry (see Supplementary Table), we corrected for possible floor effects. To this end, step-length asymmetry at baseline and its interaction with Device was added to the model. This additional analysis revealed a trend for a significant interaction between Device and baseline step-length asymmetry (p = 0.077), indicating that participants with greater step-length asymmetry at baseline tended to have more benefit from FES.

3.6 Relationship between Device effects on late stance versus swing characteristics

Correlations between the (time-integrated) Device effects in late stance and those in the swing phase are presented in Table 3. An increase in peak ankle plantarflexion velocity during late stance with FES was significantly correlated with an increase in peak hip flexion velocity during swing (r_s = 0.502, p = 0.029). A more forward oriented GRF angle during late stance with FES was also significantly correlated with an increase in peak hip flexion velocity during swing (r_s = 0.504, p = 0.028). An increase in ankle plantar flexion power and an increase in AP-GRF during late stance with FES tended to correlate with an increase in peak hip flexion velocity (r_s = 0.453, p = 0.052) and knee flexion velocity (r_s = 0.440, p = 0.059) during swing, respectively.

Table 3
Relationship between changes in stance-phase characteristics (FES-AFO) with changes in swing-phase characteristics (n = 19)

Δ Maximum knee flexion velocity (degrees/s) Δ Maximum hip flexion velocity (degrees/s)

Δ Maximum ankle plantarflexion velocity (degrees/s) r_s = 0.135 r_s = 0.502

p = 0.581 p = 0.029

Δ Maximum ankle plantarflexion power (Watt/kg) r_s = 0.318 r_s = 0.453

p = 0.185 p = 0.052

Δ GRF angle (degrees) r_s = 0.349 r_s = 0.504

p = 0.143 p = 0.028

Δ Maximum GRF in AP direction (% body weight) r_s = 0.440 r_s = 0.153

p = 0.059 p = 0.533

	Δ Maximum knee flexion velocity (degrees/s)	Δ Maximum hip flexion velocity (degrees/s)
Δ Maximum ankle plantarflexion velocity (degrees/s)	r_s = 0.135	r_s = 0.502
	p = 0.581	p = 0.029
Δ Maximum ankle plantarflexion power (Watt/kg)	r_s = 0.318	r_s = 0.453
	p = 0.185	p = 0.052
Δ GRF angle (degrees)	r_s = 0.349	r_s = 0.504
	p = 0.143	p = 0.028
Δ Maximum GRF in AP direction (% body weight)	r_s = 0.440	r_s = 0.153
	p = 0.059	p = 0.533

Δ= individual mean difference between FES and AFO, averaged over time. For example: (GRF angle_FES^T1 - GRF angle_AFO^T1) + … (GRF angle_FES^Tx - GRF angle_AFO^Tx) / x. AFO = ankle-foot orthosis. FES = implanted functional electrical stimulation. x = number of measurements.

4 Discussion

With this report we extended our previous work on the benefits of implantable peroneal nerve stimulation (FES) compared to ankle-foot orthosis (AFO) including a larger sample of patients (n = 22) with chronic drop foot after supratentorial stroke. We focused on the possible kinematic, kinetic, and spatiotemporal improvements with FES throughout the stance and swing phase of walking as well as on possible relationships between stance-phase and swing-phase improvements. As such, it is currently the largest cohort to investigate the beneficial effects of implantable peroneal nerve stimulation compared to AFO on gait characteristics, including kinetics. Our results showed that implantable peroneal FES might be superior over AFO in terms of knee stability, plantarflexion power, and propulsion, but we found no swing-phase related benefits at group level.

4.1 Stance-phase kinematics and kinetics and spatiotemporal characteristics

The results confirmed our hypothesis that implantable peroneal FES would improve knee stability compared to AFO in patients with knee (hyper)extension during mid-stance on the paretic side. Many of these patients show a knee (hyper)extension thrust, which is promoted by an AFO that limits tibial progression during the 2nd rocker (Arch et al., 2016). Peroneal FES does not promote such a knee (hyper)extension thrust. The average improvement was a 14% lower maximum knee extension velocity, which difference may well be clinically relevant. Yet, there is no literature on the minimal clinically relevant difference in this outcome for supporting its interpretation. Furthermore, no previous study compared maximum knee extension velocity between FES and AFO. Generally, maximum knee extension velocity was slightly higher at 26 weeks compared to 2 weeks after system activation, which may be related to a small increase in walking speed over time.

The results also confirm our hypothesis that implantable peroneal FES would enhance propulsion compared to AFO, by improving ankle kinematics and kinetics during late stance. We found a borderline significant improvement of the maximum ankle plantarflexion angle and a significant improvement of peak ankle plantarflexion velocity (22% ) and power (17% ). These findings are in agreement with our previous study (Schiemanck et al., 2015), although the presently reported mean differences in peak plantarflexion power between FES and AFO were lower (0.21 W/Kg vs 0.48 W/Kg). This reduction in device effect is remarkable as the average scores for the Motricity Index and Fugl-Meyer Assessment of the paretic leg were somewhat better for Group II compared to Group I (see Table 1). Since well recovered participants seem more likely to suffer from limitations imposed by AFO, it was expected that the difference in peak plantarflexion power between FES and AFO would have increased with the addition of this group. Nevertheless, the observed difference in peak plantarflexion power between FES and AFO closely matches the values observed by Sheffler et al (0.18 W/Kg) (Sheffler et al., 2013). As expected, the observed improvements in peak ankle plantarflexion power with FES coincided with improved propulsion, which was indicated by a larger maximum anteroposterior component (10% ) and more anteriorly orientated angle (14% ) of the ground reaction force. Overall, these results suggest that FES allowed participants to better utilize their residual calf muscle strength and ankle power for push-off. Indeed, the calf muscles provide approximately 75% of the propulsion power during normal gait. This propulsion can be limited by using an AFO, especially in patients with substantial residual calf muscle strength.

In our previous report, an improvement in step-length asymmetry with FES was found (Schiemanck et al., 2015). We explained this finding on the basis of an improved peak ankle power on the paretic side, leading to a larger step length with the nonparetic leg. In the pooled results of Group I and Group II, the improvement in step-length asymmetry with FES was no longer significant. Yet, when the presence of step-length asymmetry at baseline was considered in the statistical model, we still found a borderline significant interaction with Device, indicating that participants with the greatest step-length asymmetry at baseline indeed tended to improve their gait symmetry with FES. Overall, the pooled data set showed less step-length asymmetry at baseline compared to Group I reported in our previous study, allowing less room for improvement, which may well explain the absence of a significant Device effect. Consistent with the improved peak ankle power and propulsion, we did find a significant Device effect for walking speed, albeit fairly small (3% ). This improvement in mean comfortable walking speed does not surpass the minimal clinically important difference of 0.10 –0.17 m/s (Bohannon et al., 2014). Nevertheless, this is an interesting finding in view of a recent meta- analysis by Prenton et al. that did not find significant differences between FES and AFO for gait speed in patients after stroke (Prenton et al., 2016).

4.2 Swing-phase kinematics and associations with stance-phase characteristics

In contrast with our hypothesis, the beneficial effects of FES on stance-phase characteristics such as ankle power and propulsion did not coincide with improved swing-phase kinematics in the group at large. The use of implantable peroneal FES did not lead to a larger or faster knee or hip flexion during the swing phase compared to AFO. This finding is in contrast to our previous case study that showed normalization of knee and hip kinematics during swing with implanted FES compared to AFO (van Swigchem et al., 2011). This observation was explained by an improved individual capacity to utilize residual calf muscle strength with FES, leading to larger knee and hip angles during swing through inter-joint coupling. Yet, the presently observed lack of differences in knee and hip kinematics with FES is in line with previous studies on both implantable (Kottink et al., 2012) and surface-based FES (Sheffler et al., 2013). Collectively, these results indicate that a translation of stance-phase benefits into swing-phase benefits may not be expected for the majority of patients. It is possible that the observed moderate average increase in paretic peak ankle power (17% ) from AFO to FES was generally insufficient to improve swing-phase kinematics. Nevertheless, the fact that FES-induced improvements of ankle plantarflexion velocity, ankle plantarflexion power, and orientation angle of the ground reaction forces during the stance phase were moderately associated (R² = 25% ) with improvements in hip flexion velocity during the swing phase may point towards some type of functional coupling between these phenomena as a result of peroneal stimulation, but this needs to be corroborated by further research. The observed small increase across time in peak knee and hip flexion angles during the swing phase, irrespective of Device, may well be related to the parallel (modest) increase in walking speed.

4.3 Study limitations

Despite the combination of two study cohorts, this study is still limited by a small sample size, which is inherent in our focus on implantable peroneal FES and in the complexity of the instrumented gait assessments. Patients were compared to themselves instead of a control group using an AFO and not receiving implanted FES, yet the applied within-subjects design was considered a better option to identify FES-induced improvements than a parallel group design with larger (between-subjects) variability. The latter option would also have required many more participants. Selection bias has probably occurred due to the fact that patients needed to be motivated for a surgical intervention and had to show a positive response to surface-based peroneal stimulation, which limits the generalizability of our data. Blinding of participants and investigators was not possible due to the visibility of the devices. To minimize assessor bias, the approach by the investigators (e.g., instructions, encouragement) was kept constant during all data collection. Each patient used his own AFO during the course of the study, which may not necessarily have been the optimal device, but it does reflect regular clinical practice. Most participants preferred the use of FES over AFO during the follow-up, which might have led to an ‘unlearning’ effect. However, we did not find any evidence for worsening of walking with AFO over time.

Fig.3

Estimated mean (time integrated) values and standard errors for AFO and FES for each of the spatiotemporal and swing-phase parameters (n = 19). n.s = not significant difference; * = p < 0.05. AFO = ankle-foot orthosis; FES = implanted functional electrical stimulation.

5 Conclusions

This study substantiates the scientific evidence that implantable peroneal FES as a treatment for post-stroke drop foot may have kinematic and kinetic advantages compared to AFO. In particular, by allowing full freedom of motion at the ankle joint, patients are able to optimally use any residual calf muscle strength for generating ankle plantarflexion power and propulsion during push-off, which may optimize step-length symmetry in asymmetric patients. In addition, knee instability (towards extension) during mid-stance may be less with FES compared to AFO due to optimal tibial progression at the ankle. Whether these stance-phase benefits translate into improved swing-phase characteristics remains questionable. Changes in swing-phase kinematics are probably strongly dependent on individual patient characteristics such as the timing and strength of the push-off and compensation strategies used. Future research should focus on more complex gait tasks in which stance-phase stability and push-off regulation are challenged in order to fully understand the potential benefits of (implantable) peroneal FES compared to AFO in patients with post-stroke drop foot.

Conflict of interest

None of the authors have potential conflicts of interest to be disclosed.

Footnotes

Acknowledgments

The first study cohort (Group I) was funded by the Otto Bock Group. They provided the implantable systems used in this study and an allowance per patient was given. The second study cohort (Group II) was financed by TWIN Institute for Neuromodulation. We thank P. van den Munckhof, J. de Vries, and T. Beems^† for their neurosurgical contribution and R. Keukenkamp, J. den Boer and R. van Swigchem for their help during the measurements.

The supplementary material is available in the electronic version of this article: .

References

Arch

E.S

, Stanhope

S.J

, & Higginson

J.S.

(2016). Passive-dynamic ankle-foot orthosis replicates soleus but notgastrocnemius muscle function during stance in gait: Insights fororthosis prescription. Prosthetics and OrthoticsInternational, 40(5), 606–616. doi:10.1177/0309364615592693

Bleyenheuft

, Bleyenheuft

, Hanson

, & Deltombe

(2010). Treatment of genu recurvatum in hemiparetic adult patients: A systematic literature review. Annals of Physical and Rehabilitation Medicine, 53(3), 189–199. doi:10.1016/j.rehab.2010.01.001

Bohannon

R.W

, & Glenney

S.S.

(2014). Minimal clinically important difference for change in comfortable gait speed of adults with pathology: A systematic review. Journal of Evaluation in Clinical Practice, 20(4), 295–300. doi:10.1111/je12158

Cakar

, Durmus

, Tekin

, Dincer

, & Kiralp

M.Z.

(2010). The ankle-foot orthosis improves balance and reduces fall risk of chronic spastic hemiparetic patients. European Journal of Physical and Rehabilitation Medicine, 46(3), 363–368.

Campanini

, Merlo

, & Damiano

(2013). A method todifferentiate the causes of stiff-knee gait in stroke patients. Gait & Posture, 38(2), 165–169. doi:10.1016/j.gaitpost.2013.05.003

Chantraine

, Schreiber

, Kolanowski

, & Moissenet

(2016). Control of Stroke-Related Genu Recurvatum With Prolonged Timing of Dorsiflexor Functional Electrical Stimulation: A Case Study. Journal of Neurologic Physical Therapy, 40(3), 209–215. doi:10.1097/NPT.0000000000000137

Ernst

, Grundey

, Hewitt

, von Lewinski

, Kaus

, Schmalz

, … Liebetanz

(2013). Towards physiological ankle movements with the ActiGait implantable drop foot stimulator in chronic stroke. Restorative Neurology and Neuroscience, 31(5), 557–569. doi:10.3233/RNN-120283

Goldberg

S.R

, Anderson

F.C

, Pandy

M.G

, & Delp

S.L.

(2004). Muscles that influence knee flexion velocity in double support: Implications for stiff-knee gait. Journal of Biomechanics, 37(8), 1189–1196. doi:10.1016/j.jbiomech.2003.12.005

Kottink

A.I

, Oostendorp

L.J

, Buurke

J.H

, Nene

A.V

, Hermens

H.J

, & MJ

I.J.

(2004). The orthotic effect of functional electrical stimulation on the improvement of walking in stroke patients with a dropped foot: A systematic review. Artificial Organs, 28(6), 577–586. doi:10.1111/j.1525-1594.2004.07310.x

10.

Kottink

A.I

, Tenniglo

M.J

, de Vries

W.H

, Hermens

H.J

, & Buurke

J.H.

(2012). Effects of an implantable two-channel peroneal nerve stimulator versus conventional walking device on spatiotemporal parameters and kinematics of hemiparetic gait. Journal of Rehabilitation Medicine, 44(1), 51–57. doi:10.2340/16501977-0909

11.

Liberson

W.T

, Holmquest

H.J

, Scot

, & Dow

(1961). Functional electrotherapy: Stimulation of the peroneal nerve synchronized with the swing phase of the gait of hemiplegic patients. Archives of Physical Medicine and Rehabilitation, 42, 101–105.

12.

Liu

M.Q

, Anderson

F.C

, Schwartz

M.H

, & Delp

S.L.

(2008). Muscle contributions to support and progression over a range of walking speeds. Journal of Biomechanics, 41(15), 3243–3252. doi:10.1016/j.jbiomech.2008.07.031

13.

Mulroy

S.J

, Eberly

V.J

, Gronely

J.K

, Weiss

, & Newsam

C.J.

(2010). Effect of AFO design on walking after stroke: Impact of ankle plantar flexion contracture. Prosthetics and Orthotics International, 34(3), 277–292. doi:10.3109/03093646.2010.501512

14.

Nair

P.M

, Rooney

K.L

, Kautz

S.A

, & Behrman

A.L.

(2010). Stepping with an ankle foot orthosis re-examined: A mechanical perspective for clinical decision making. Clinical Biomechanics, 25(6), 618–622. doi:10.1016/j.clinbiomech.2010.03.001

15.

Neptune

R.R

, Sasaki

, & Kautz

S.A.

(2008). The effect of walking speed on muscle function and mechanical energetics. Gait & Posture, 28(1), 135–143. doi:10.1016/j.gaitpost.2007.11.004

16.

Neptune

R.R

, Zajac

F.E

, & Kautz

S.A.

(2004). Muscle force redistributes segmental power for body progression during walking. Gait & Posture, 19(2), 194–205. doi:10.1016/S0966-6362(03)00062-6

17.

Patterson

K.K

, Gage

W.H

, Brooks

, Black

S.E

, & McIlroy

W.E.

(2010). Evaluation of gait symmetry after stroke: A comparison of current methods and recommendations for standardization. Gait & Posture, 31(2), 241–246. doi:10.1016/j.gaitpost.2009.10.014

18.

Peterson

C.L

, Hall

A.L

, Kautz

S.A

, & Neptune

R.R.

(2010). Pre-swing deficits in forward propulsion, swing initiation and power generation by individual muscles during hemiparetic walking. Journal of Biomechanics, 43(12), 2348–2355. doi:10.1016/j.jbiomech.2010.04.027

19.

Prenton

, Hollands

K.L

, & Kenney

L.P.

(2016). Functionalelectrical stimulation versus ankle foot orthoses for foot-drop: Ameta-analysis of orthotic effects. Journal of RehabilitationMedicine, 48(8), 646–656. doi:10.2340/16501977-2136

20.

Schiemanck

, Berenpas

, van Swigchem

, van den Munckhof

, de Vries

, Beelen

, … Geurts

A.C.

(2015). Effects of implantable peroneal nerve stimulation on gait quality, energy expenditure, participation and user satisfaction in patients with post-stroke drop foot using an ankle-foot orthosis. Restorative Neurology and Neuroscience, 33(6), 795–807. doi:10.3233/RNN-150501

21.

Sheffler

L.R

, Bailey

S.N

, Wilson

R.D

, & Chae

(2013). Spatiotemporal, kinematic, and kinetic effects of a peroneal nerve stimulator versus an ankle foot orthosis in hemiparetic gait. Neurorehabilitation and Neural Repair, 27(5), 403–410. doi:10.1177/1545968312465897

22.

Tani

, Otaka

, Kudo

, Kurayama

, & Kondo

(2016). Prevalence of Genu Recurvatum during Walking and Associated Knee Pain in Chronic Hemiplegic Stroke Patients: A Preliminary Survey. Journal of Stroke and Cerebrovascular Diseases, 25(5), 1153–1157. doi:10.1016/j.jstrokecerebrovasdis.2016.01.028

23.

van Swigchem

, Weerdesteyn

, van Duijnhoven

H.J

, den Boer

, Beems

, & Geurts

A.C.

(2011). Near-normal gait pattern with peroneal electrical stimulation as a neuroprosthesis in the chronic phase of stroke: A case report. Archives of Physical Medicine Rehabilitation, 92(2), 320–324. doi:10.1016/j.apmr.2010.10.038

24.

Vistamehr

, Kautz

S.A

, & Neptune

R.R.

(2014). The influence of solid ankle-foot-orthoses on forward propulsion and dynamic balance in healthy adults during walking. Clinical Biomechanics, 29(5), 583–589. doi:10.1016/j.clinbiomech.2014.02.007

25.

Weerdesteyn

, de Niet

, van Duijnhoven

H.J

, & Geurts

A.C.

(2008). Falls in individuals with stroke. Journal of Rehabilitation Research & Development, 45(8), 1195–1213.

26.

Woltring

H.J.

(1986). A Fortran package for generalized, cross-validatory spline smoothing and differentiation. Advances in Engineering Software, 8(2), 104–113. doi: https://doi.org/10.1016/0141-1195(86)90098-7

27.

Zajac

F.E

, Neptune

R.R

, & Kautz

S.A.

(2003). Biomechanics and muscle coordination of human walking: Part II: Lessons from dynamical simulations and clinical implications. Gait & Posture, 17(1), 1–17.