Time Course Analysis of the Effects of Botulinum Toxin Type A on Elbow Spasticity Based on Biomechanic and Electromyographic Parameters

Methods 

Participants and BTX-A Treatment 

All stroke patients were evaluated before participating in the experiments, which were overseen by a qualified physician in the Department of Rehabilitation at Chi-Mei Hospital, Tainan, Taiwan. The inclusion criteria for subjects included (1) flexor spasticity in the affected elbow joint without contracture, (2) the onset of CVA having occurred at least 6 months previously, (3) the presence of stable spasticity, (4) a Brunnstrom stage of at least level 3,27 and (5) no severe cognitive or affective dysfunction. Subjects provided informed consent for participation in the protocols before undergoing injections and subsequent evaluations, which followed the clinical protocol approved by the local ethics committee. Eight subjects (age range, 38–70y) with previous onsets of CVA (range, 9–69mo) previously completed the 11-week time course observation period. The clinical data of the subjects are summarized in table 1.

Table 1. Summary of the Subjects’ Clinical Data and Dosage of the Injection of BTX-A Into the Biceps Brachii

	Subject (n=8)	Age (y)	Sex	Months Postinjury	Affected Side	Brunnstrom Stage	Dosage (U)
	S1	64	Female	31	Left	V	50
	S2	65	Female	24	Left	V	50
	S3	50	Female	9	Left	IV	50
	S4	61	Male	61	Left	III	100
	S5	68	Male	43	Left	III	50
	S6	38	Female	14	Right	III	50
	S7	70	Male	20	Right	III	50
	S8	55	Female	69	Left	IV	50

Under the guidance of needle electromyography, all subjects were injected with BTX-A (Botox) in the muscles of the forearm and upper arm by the same physician. The dosage and locations of the intramuscular injection were individualized for each patient based on the severity and distribution of the spastic muscles involved. Among the injected muscles, the targeted biceps brachii muscle (one of the main elbow flexors) was always included for injection, at doses ranging from 50 to 100U (see table 1). Other muscles that were injected with intramuscular BTX-A included the flexor digitorum superficialis (50U in 5 subjects), the flexor carpi radialis (50U in 4 subjects), and the flexor carpi ulnaris (50U in 5 subjects). No other medication was prescribed for reducing spasticity during the study, and the rehabilitation program was unchanged during the 11 weeks of observation.

Time Course Evaluation of BTX-A Effects 

Spasticity was evaluated at 2 weeks before and 2, 6, and 9 weeks after the injection of BTX-A. All evaluations were conducted by the same physical therapist. The evaluation procedures comprised the clinical MAS, biomechanic measurements, and electromyographic assessments. The clinical assessment was performed by using the MAS with 6 scores (0–5, modified from the original scores of 0, 1, 1+, 2, 3, and 4).12 In addition, biomechanic and electromyographic data were collected by using a portable measurement system comprising a handheld device and 2 pairs of surface electromyography recording electrodesa (fig 1A). The handheld device consists of wrist cuffs with airbags and a lightweight gyroscope. The airbags held on both the ventral and dorsal sides of the wrist are connected to a differential pressure sensor to record the net resistance (pressure). The joint displacement can be obtained from the integration of gyroscopic measurements of the angular rate. As shown in figure 1B, the phase lag between resistance and derived joint displacement can be used for deriving the velocity-dependent property of spasticity. In addition to biomechanic data, electromyography electrodes were placed in parallel to muscle fibers on muscle bellies of the biceps brachii and triceps brachii to measure stretch reflex responses (see fig 1C).

View Large Image Download to PowerPoint

Fig 1. A schematic setup of biomechanic and electromyographic measurements for quantifying the time course changes of spasticity before and after injections of BTX-A. (A) Portable muscle tone measurement device. The elbow was stretched manually via the wrist cuffs at 4 frequencies. The biomechanic data and electromyographic data were sensed using a portable device (a differential pressure sensor and a lightweight gyroscope) and 2 surface electromyography electrodes, respectively. (B) Biomechanic data. Examples of biomechanic data of reactive resistance and displacement during stretching at 3/2Hz. The dashed lines show the phase lag between resistance and displacement. (C) Electromyographic (EMG) data. Clear reflex electromyographic activities are found in the biceps brachii compared with the triceps brachii during stretching at 3/2Hz.

During the data recording, subjects were asked to relax entirely, as monitored by the electromyography recording. The elbow joint was evaluated by manually stretching the forearm in a back-and-forth manner, approximately sinusoidally. Stretching was performed at 4 frequencies (1/3, 1/2, 1, 3/2Hz) with the assistance of a metronome. The consistency of the stretch frequency can be only confirmed from the frequency response of trajectory. The range of elbow flexion and extension was restricted to −30° and 30° (where 0° indicates a right-angled elbow) by using an elbow limiter (see fig 1A). The reactive resistance, angular rate, and electromyographic signals were digitized simultaneously at a 1000-Hz sampling rate and 12-bit resolution for further processing.

Data Analysis 

Our analytic approach first derived the velocity-dependent viscous component from the biomechanic measurements of reactive resistance and joint displacement. We modeled the spastic joint as a second-order system, which the reactive torque contributes from inertia, viscous, and elastic components when sinusoid stretches were imposed.25 Because the applied force is perpendicular to the forearm, the reactive resistance can be assumed as the reactive torque in following analysis of same subject. Figure 2A shows a representative plot of reactive torque T(t), and joint displacement, X(t), at 1-Hz stretches. The hysteresis loop represents the velocity-dependent damping of the spastic joint model.16 After representing the displacement X(t) in sinusoid fashion, Asin(ωt), in the second-order equation, the averaged complex modulus can be finally derived and defined as T(t)/X(t+θ), which is the overall magnitude of the vector sum of the real part (K-Iω2) and imaginary part (Bω) in figure 2B.25 By shifting the displacement with the phase lag, X(t+θ), the averaged complex modulus can be estimated and then used to derive the viscous component (Bω) (see figs 2B, 2C). Figure 2D shows the Bω estimated at 4 stretching frequencies (ie, 1/3, 1/2, 1, 3/2Hz) denoted as Bω1/3, Bω1/2, Bω1, and Bω3/2, respectively. From the 4 Bω (in N·rad·m−1) estimations, 1 viscosity parameter B can be derived by transforming frequency (in hertz) to angular frequency (in rad/s) and applying a linear fitting process. For each subject, this viscosity B (in N·s·m−1) was used here as a biomechanic index to evaluate the velocity dependence of spasticity. Details of the modeling technique are available elsewhere.25

View Large Image Download to PowerPoint

Fig 2. The processing of biomechanic data. (A) A curve of the reactive torque versus displacement for a single trial of stretches. Phase lag ω was estimated to derive the viscous component. (B) A plot of the reactive torque, T(t) and displacement (shifted by θ). (C) A graphical representation of our analytic approach for deriving the viscous component from the phase lag and averaged complex modulus (ACM). Based on a second-order model, Bω is proportional to the phase lag θ. ACM can be estimated from the curve of T(t) versus X(t+θ) in panel B. (D) Viscous components (Bω1/3, Bω1/2, Bω1, Bω3/2) estimated from the 4 stretching frequencies were pooled to derive the velocity-dependent viscosity (B).

To quantify the length-related property of stretch reflexes, the angle threshold can be derived from the linear-envelope representation of the reflex electromyography. To determine the reflex threshold, full flexion and full extension of the elbow joint from −30° to 30° was defined as 1 stretch range (range between solid lines in fig 3A). In a single-stretch range, the reflex threshold was defined as the angle when the sustained electromyographic activity increased to more than 3 standard deviations above the 100-ms baseline electromyographic signal before the stretch range. Only the 2 highest stretching frequencies (1, 3/2Hz) were sufficient to elicit obvious and consistent stretch responses that were suitable for analysis in our study. For clear representation, only electromyographic thresholds stretched at the highest frequency (3/2Hz) were used to quantify the length-related property in stretch reflexes. The electromyographic thresholds at various stretches were normalized as a percentage of the stretch range, where 0% and 100% represent stretch angles of −30° and 30°, respectively. Five stretch cycles were selected from a single trial of at least 15 cycles to determine the average electromyographic threshold of the stretch reflexes by using the automatic program. The RET of the biceps brachii (represented as percentage form) was then used as a neurophysiologic index to evaluate the length-related property of spasticity, where a higher angle threshold indicates a lower excitability of the stretch reflex in spastic muscles.24

View Large Image Download to PowerPoint

Fig 3. The processing of electromyographic signals to determine the RET of the stretch reflex. (A) One stretch range is indicated with 2 solid lines (lines A, B) in the curve of joint displacement. The linear-envelope formation of the electromyographic raw signal is shown in panels B, C, and D. The vertical dotted line C marks the first time point where the electromyographic linear envelope exceeds 3 standard deviations from the baseline electromyographic activity, as recorded for 100ms before eliciting the stretch (solid line A). The threshold was then calculated as a percentage of the stretch range (RET = 65.3% in panel A).

The response of subjects to BTX-A injections was quantified by observing the peak effect and relapse rate of biomechanic and reflex electromyographic data. The peak effect represents the maximal reduction in B or the maximal increment in RET during the observations. The peak effects on B and RET were represented as percentages relative to the preinjection values. In addition, the relapse rate, representing the relapse of spasticity at the last stage of observation (ie, Blast and RETlast), was calculated from both biomechanic and electrophysiologic data as follows:

where Bpre and RETpre represent the baseline preinjection levels, and Bmin and RETmax represent the minimum B and the maximum RET during the observation stages, respectively. When Bmin or RETmax occurred in the last stage of observation, the injected muscle might not have regained its spasticity and thus the relapse rate was considered undetermined.

One-way analyses of variance (ANOVAs) of the biomechanic and electromyographic parameters (ie, B, RET) from 4 stages in an 11-week period were performed to examine the time effect of BTX-A injection. Post hoc tests were further performed to compare the differences between 2 adjacent stages and between preinjection and at 9 weeks. A similar analysis was applied to the clinical MAS scores by using nonparametric tests (Friedman 2-way ANOVA by ranks for the time factor and Wilcoxon matched-pairs signed-rank test for post hoc investigation). Probability values of P less than .05 were considered indicative of statistically significant results.

Results 

To elucidate the overall effects of BTX-A on spasticity for all subjects, we first evaluated the time factor by 1-way ANOVA of pooled values of the MAS score (table 2), velocity-dependent parameter (B) (table 3), and length-related parameter (RET) (table 4) from the 8 subjects. Comparing spasticity using these 3 parameters across the 4 time stages indicated that time was a significant factor (all P<.05) (see Table 2, Table 3, Table 4). To further quantify the changes in spasticity during the 4 stages, post hoc comparisons were used to assess the differences between 2 adjacent stages and between preinjection and at 9 weeks. A comparison of the levels of preinjection and 2 weeks after the injection (pre vs I; see Table 2, Table 3, Table 4) revealed significant decreases in the averaged MAS score and B as well as a significant increase in RET (all P<.05).

Table 2. MAS Scores for All Subjects at 2 Weeks Before (preinjection) and 2, 6, and 9 Weeks After the BTX-A Injection

	Subject (n=8)	Pre (2wk)	I (post 2wk)	II (post 6wk)	III (post 9wk)
	S1	2	2	2	2
	S2	3	2(−)	2	2
	S3	4	3(−)	3	3
	S4	4	2(−)	2	3(+)
	S5	3	2(−)	2	2
	S6	3	3	3	3
	S7	3	2(−)	2	2
	S8	3	2(−)	2	2
	Mean ± SD	3.13±0.64	2.25±0.46	2.25±0.46	2.38±0.52

	Statistics	Time Factor	Pre vs I	I vs II	II vs III	Pre vs III
	P	.001⁎	.020⁎	NS (.999)†	NS (.317)†	.014⁎

NOTE. The symbols − and + indicate that the MAS score decreased or increased relative to the previous stage. The time factor in averaged MAS scores during the 4 time stages was analyzed statistically by using the Friedman test. Changes between the adjacent stages were then compared by using the Wilcoxon test. Abbreviations: NS, not significant; SD, standard deviation.

⁎ Significant differences. † Nonsignificant differences (P>.05).

Table 3. Time Course Changes in Biomechanic Parameter B for All Subjects

	Subject	Pre (2wk)	I (post 2wk)	II (post 6wk)	III (post 9wk)
	S1	.164	.101(−)	.099(−)	.080(−)
	S2	.364	.206(−)	.092(−)	.124(+)
	S3	.336	.091(−)	.256(+)	.277(+)
	S4	.136	.077(−)	.075(−)	.085(+)
	S5	.170	.138(−)	.132(−)	.128(−)
	S6	.207	.091(−)	.144(+)	.168(+)
	S7	.292	.077(−)	.090(+)	.076(−)
	S8	.137	.128(−)	.126(−)	.135(+)
	Mean ± SD	.226±.092	.114±.043	.127±.057	.134±.066

	Statistics	Time Factor	Pre vs I	I vs II	II vs III	Pre vs III
	P	.020⁎	.008⁎	NS (.646)†	NS (.293)†	.021⁎

NOTE. The symbols − and + indicate that the viscosity decreased or increased relative to the previous stage. In all subjects, the B value decreased after the BTX-A injection. The time factor in averaged B values during the 4 time stages was analyzed statistically by using 1-way ANOVA. Adjacent stages were then tested by using post hoc tests.

⁎ Significant differences. † Nonsignificant differences (P>.05).

Table 4. Time Course Changes in Averaged RET (%) With a Stretching Frequency of 1.5Hz

	Subject	Pre (2wk)	I (post 2wk)	II (post 6wk)	III (post 9wk)
	S1	55.5	68.8(+)	65.4(−)	64.0(−)
	S2	48.9	65.0(+)	66.6(+)	63.3(−)
	S3	65.3	66.7(+)	75.7(+)	66.1(−)
	S4	49.6	52.9(+)	64.2(+)	57.7(−)
	S5	74.0	75.7(+)	81.9(+)	77.0(−)
	S6	51.0	67.0(+)	60.6(−)	55.9(−)
	S7	71.2	72.9(+)	74.3(+)	72.1(−)
	S8	65.9	78.8(+)	76.4(−)	73.5(−)
	Mean ± SD	60.2±10.1	68.5±7.9	70.6±7.4	66.2±7.5

	Statistics	Time Factor	Pre vs I	I vs II	II vs III	Pre vs III
	P	<.001⁎	.011⁎	NS (.366)†	.002⁎	.008⁎

NOTE. The symbols − and + indicate that the electromyographic threshold decreased or increased relative to the previous stage. In all subjects, the RET value decreased after the BTX-A injection. The time factor in the averaged threshold percentage during the 4 time stages was analyzed statistically by using 1-way ANOVA. Adjacent stages were compared by using post hoc tests.

⁎ Significant differences. † Nonsignificant differences (P>.05).

To further elucidate the reduction in spasticity induced by BTX-A, the time course changes in the 3 parameters were compared. Average MAS scores were unchanged between weeks 2 and 6 (average MAS score, 2.25). A slight increase (average MAS score, 2.38) was noted at week 9, but no significant difference was found when compared with the previous stage (P>.05) (see table 2). Similarly, a slight increase in B was noted from weeks 2 to 9 (average B=.114, B=.127, B=.134), but no statistical difference was found (both P>.05) (see table 3). For RET, no significant difference (P>.05) (see table 4) was found between weeks 2 and 6 (average RET=68.5%, RET=70.6%, respectively), but it decreased significantly (P<.05) from weeks 6 to 9 (average RET=70.6%, RET=68.2%). Furthermore, we compared the parameters between preinjection and week 9 to observe the lasting effects of BTX-A in the last stage of our data collection. Both the average MAS score and B were significantly lower at week 9 than before injection (P<.05). Similarly, the average RET was significantly higher at week 9 than before injection (P<.05).

We show the feasibility of observing the individual responses of spasticity to a BTX-A injection during the 4 observation stages by presenting the MAS score, B, and RET for 2 representative cases. As shown in figure 4A, the clinical scale shows no changes during the 4 recording stages (all MAS scores, 3) in subject S6, whereas B decreased significantly from preinjection (.207) to week 2 (.091) and then increased gradually at weeks 2, 6, and 9 (.091, .144, .168, respectively). The peak effect was derived from the ratio of the maximal decrement in B (.116, from .207−.091) to the preinjection level (.207) (ie, 56.0% in this case), and the degree of spasticity relapse was derived from the ratio between the minimal (.077, from .168−.091) and maximal (.116) decrements of B (ie, 66.4% in this case). Similarly, RET increased from preinjection to week 2 and then decreased gradually during the following 3 stages (see fig 4E). The peak effect and relapse rate of RET were 31.4% and 69.3%, respectively.

View Large Image Download to PowerPoint

Fig 4. Two typical examples of time course data based on the clinical scale (MAS), velocity-dependent property (V-D property) B, and length-related property (L-R property) RET. Compared with the clinical scale, our quantitative parameters B and RET successfully reflect the change in spasticity after the BTX-A injection in subject S6 (panels A, C, E) and subject S2 (panels B, D, F).

As the second representative case, a decrease in the MAS score by 1 point was noted in subject S2 between preinjection and 2 weeks after the injection (fig 4B), but it then maintained the same score in the remaining assessments (all MAS scores, 2). It is interesting to see that B decreased gradually from preinjection to weeks 2 and 6 (.364, .206, .092, respectively; peak effect, 74.7%) but was slightly elevated at week 9 (.125; relapse rate, 11.8%). Similarly, RET increased gradually from preinjection up to week 6 and decreased slightly in the last stage (fig 4F); the peak effect and relapse rate were 36.0% and 18.3%, respectively.

Table 5 summarizes the peak effects and their stage of occurrence as well as the relapse rates of the B and RET parameters for all subjects. The maximal reduction in spasticity ranged from 8.0% to 74.7%. The peak effects of B occurred at 2, 6, or 9 weeks after the BTX-A injection. According to the relapse rate of the B values in table 5, 3 subjects (S3, S6, S8) can be considered as exhibiting obvious relapses of spasticity (a relapse rate exceeding 50%) and 2 subjects (S2, S4) showed only minor relapse in spastic limbs (relapse rate, <50%). The other 3 subjects (S1, S5, S7) did not show any relapse of spasticity until the last stage of observation.

Table 5. The Peak Effect, Occurrence Stage, and Relapse Rate of Spasticity After the BTX-A Injection

	Subject	Peak Effect (%)/Occurrence Stage		Relapse Rate (%)
		B	RET	B	RET
	S1	−51.2/week9	24.1/week2	⁎	36.3
	S2	−74.7/week6	36.0/week6	11.8	18.3
	S3	−72.9/week2	15.8/week6	75.9†	92.5†
	S4	−44.9/week6	29.3/week6	16.4	44.6
	S5	−24.7/week9	10.7/week6	⁎	62.0†
	S6	−56.0/week2	31.4/week2	66.4†	69.3†
	S7	−74.0/week9	4.4/week6	⁎	72.1†
	S8	−8.0/week6	19.6/week2	81.8†	41.1

⁎ The subject did not show a relapse of spasticity until the last stages of observation. † The obvious relapse of spasticity from the BTX-A injection in the last stages of observation.

The peak effects of BTX-A injection for RET ranged from 4.4% to 36.0% (table 5), with this occurring in all subjects at either 2 or 6 weeks after the BTX-A injection. Moreover, 4 subjects (S3, S5, S6, S7) were considered as exhibiting obvious relapse (>50% of its initial value).

Discussion 

In this study, we have investigated the time course changes in spasticity after the BTX-A injection from 3 different aspects: clinical MAS, biomechanic measurement of the velocity-dependent property from the viscosity index B, and electrophysiologic assessment of the length-related property from the amplitude of reflex electromyography. The statistical results listed in Table 2, Table 3, Table 4 indicate significant decreases in averaged MAS scores and B and significant increases in averaged RET, from which a reduction in spasticity was obvious at 2 weeks after the BTX-A injection (P<.05). For the average of all 3 parameters, the decrease in spasticity was maintained to the last stage (ie, 9 weeks after the injection [P<.05]).

The benefits of a quantitative analysis based on parameters B and RET are shown by further observations of individual subjects. As indicated in figures 4A and 4B and table 2, the MAS score does not reflect the time course effects of BTX-A on spasticity in most of our subjects, especially in the washout stages (weeks 2 to 9). Only 1 subject in our current MAS assessment (S4 in table 2) exhibited a relapse of spasticity, a 1-point elevation in the MAS score at week 9. In contrast, our biomechanic (B) and electromyographic (RET) parameters (see figs 4C–F, Table 3, Table 4) quantitatively reflect the effects on an injection of BTX-A. Further information including the peak effect and relapse rate can be derived from our quantitative parameters to elucidate the individual response to a BTX-A injection (see table 5). Several studies have shown the lack of reliability and validity of the MAS in the measurement of spasticity,28 especially in longitudinal observations of the effects of interventions on spasticity.5, 29 Our results coincide with the viewpoint that the MAS is a useful screening tool1 but may be not effective in quantifying spasticity, especially subtle changes that occur over time.

More than half of our subjects (S1, S2, S3, S6, S7) responded quite well to a BTX-A injection, with a maximal reduction in B of more than 50% relative to the preinjection level. Four of these subjects (S1, S2, S4, S6) can also be considered as good responders on the basis of RET (peak effect >24%), although the range of the peak effect was much smaller than that for B. The peak effect typically occurred at week 2 or 6 after the injection and may not have reached its maximal effect within our observation period for some subjects (S1, S6, S7 with parameter B). This is in general agreement with previous findings of the peak effect of a BTX-A injection occurring after 2 to 6 weeks,30 although those findings were based on the average value of the clinical MAS score.

The time course of changes in the velocity-dependent and length-related properties were variant, with all subjects exhibiting peak effects in earlier stages (2 or 6 weeks postinjection) based on parameter RET and with 3 subjects (S1, S5, S7) exhibiting peak effects in the last stage based on parameter B. More subjects showed an obvious relapse of spasticity based on parameter RET than based on parameter B. A quantitative comparison of the differences between B and RET might elucidate the differences in the effects of BTX-A on the velocity-dependent and length-related properties of spasticity. In the peripheral nervous system, a block of acetylcholine transmission in the neuromuscular junction of an alpha motoneuron substantially decreases the muscle contraction output of the stretch reflex. Such a decrease in muscle output can be reflected by an obvious delay in the stretch reflex contraction, which corresponds to an increased RET in our study. The effects of BTX-A on the synapses of gamma motoneurons and intrafusal fibers should be also considered. Recent studies provide evidence that BTX-A acts also on the fusimotor system because it reduces the amplitude of the Achilles’ tendon response22 and increases the electromyographic threshold of wrist flexors.31 These effects reduce both the static and dynamic gains of muscle spindles to length and velocity changes to induce a deafferentation of the alpha-motoneuron pool. These are also evident from our observations of a decreased velocity-dependent property and an increased length-related property in the present study. Studies of prolonged motor-evoked potentials and central conduction times in the central nervous system have indicated that a BTX-A injection decreases the size of the excitable alpha-motoneuron pool and thereby reduces the reactive muscle contraction of the stretch reflex.32 Physiologic changes at the level of the alpha-motoneuron pool caused by retrograde transportation of BTX-A along motoneurons might play a role in our finding of variations in the velocity-dependent and length-related properties.32

Table 3, Table 4 suggest that the overall effects of BTX-A lasted for at least 9 weeks in our subjects (P<.05). These results coincide with previous observations that the effects of BTX-A on spastic elbow flexors can last for at least 9 weeks,33 12 weeks,34, 35 and 16 weeks.36 Although the effects of BTX-A were maintained for up to 9 weeks after the injection in the present study, a substantial relapse of spasticity was noted in some of our subjects (based on parameters B and RET). As indicated in table 5, 3 subjects showed an obvious relapse of spasticity (66.4%–81.8% for B). On the basis of RET, 4 subjects were considered as exhibiting similarly obvious relapses of spasticity (range, 62.0%–92.5%). These results suggest that the washout of BTX-A begins as early as 2 or 6 weeks after the injection. With the lack of a quantitative assessment tool, less information is currently available to the clinician about washout progress after a BTX-A injection, especially in the early stages. Our approach provides a quantitative approach to monitor the relapse of spasticity after the BTX-A injection in a clinical setup. The information derived about the early relapse of spasticity after such an injection is valuable for helping the clinician to make decisions to proceed with other related rehabilitation interventions or another BTX-A injection.37

Conclusions 

Our study suggests that biomechanic or electromyographic parameters are more appropriate than the clinical MAS in time course observation of BTX-A effects on spasticity. The change of spasticity in velocity-dependent and length-related properties improves our understanding of the nature of BTX-A and spasticity. The same approach might be used in future studies evaluating the lowest effective dose44 and helping to determine treatments leading to functional improvement of limbs.