Muscle volume alterations in spastic muscles immediately following botulinum toxin type-A treatment in children with cerebral palsy

Authors


Abstract

Aim

With evidence for an atrophic effect of botulinum toxin type A (BoNT-A) documented in typically developing muscles, this study investigated the immediate morphological alterations of muscles in children with cerebral palsy (CP) after BoNT-A treatment.

Method

Fifteen children (10 males, five females; age range 5–11y, mean age 8y 5mo, SD 1y 10mo) with spastic diplegic CP [Gross Motor Function Classification System Levels I (n=9) and II (n=6)] receiving BoNT-A injections for spasticity management were included. None of the children was a first-time receiver of BoNT-A. Magnetic resonance imaging and Mimics software assessed muscle volume, timed 2 weeks before and 5 weeks after injection. All participants received BoNT-A bilaterally to the gastrocnemius muscle, and five participants also received BoNT-A bilaterally to the medial hamstring muscles. Functional assessment measures used were the 6-Minute Walk Test (6-MWT), the Timed Up and Go (TUG) test, and hand-held dynamometry.

Results

Whilst total muscle group volume of the injected muscle group remained unchanged, a 4.47% decrease in the injected gastrocnemius muscle volume (p=0.01) and a 3.96% increase in soleus muscle volume (p=0.02) was evident following BoNT-A. There were no statistically significant changes in function after BoNT-A as assessed by the TUG. There was also no statistically significant change in distance covered in the 6-MWT. Muscle strength, as assessed using hand-held dynamometry was also not statistically different after BoNT-A treatment.

Interpretation

Muscle volume decreases were observed in the injected muscle (gastrocnemius), with synergistic muscle hypertrophy that appeared to compensate for this decrement. The 4% to 5% decrease in the volume of BoNT-A injected muscles are not dramatic in comparison to reports in recent animal studies, and are a positive indication for BoNT-A, particularly as it also did not negatively alter function.

What this paper adds

  • The first known documentation of the morphological alterations of pathological muscles in response to BoNT-A.
  • Documentation of morphological response in synergist and antagonist muscle to BoNT-A treatment in spastic muscle.
  • Increased understanding of the effect of BoNT-A treatment on muscle strength and functional ability in children with CP.

Spasticity, muscle weakness, and muscle co-contraction are common motor problems associated with cerebral palsy (CP). Botulinum toxin type A (BoNT-A) is widely used for the management of spasticity in children with CP.[1, 2] As well as its respectable safety profile,[3-5] the literature describes many positive outcomes from BoNT-A treatment, including a reduction in muscle tone,[6] increase in joint range of motion,[6] improved gait patterns,[6] functional improvements,[7, 8] and delayed and reduced requirement for surgical interventions to treat musculoskeletal deformities when combined with conservative treatments.[9] However, generalised muscle weakness[5, 10, 11] and weakness in neighbouring non-targeted muscles[12, 13] have been reported as undesirable effects of BoNT-A.

Recent publications have raised concern regarding the effect of BoNT-A on muscle size and morphology.[14, 15] Schroeder et al.[15] measured neurogenic atrophy in the injected lateral gastrocnemius in two healthy adults after BoNT-A injection and reported a reduction of 14% to 19% in the cross-sectional area after 3 months and reductions still seen at 6, 9 and 12 months post BoNT-A treatment, there were no changes in the contralateral placebo-injected muscle. Dunne et al.[16] supplemented this finding with reports of prolonged denervation of BoNT-A-injected muscles up to 5 months post injection. Rare reports in the cosmetic industry provide us with some observable examples of BoNT-A-related atrophy, in its use in facial contouring for reducing masseter muscle thickness,[17, 18] and its use in reduction of muscle size in women's legs.[19] Animal studies have also revealed BoNT-A's effect on muscle. Fortuna et al.[13] compared injected limbs to non-injected limbs of rabbits and found reductions in muscle mass of up to 50% in injected muscles 1 month after injection. Ma et al.[20] found muscle mass in injected limbs in rats to reduce by 32% of the control side 2 weeks after the injection, and still to be reduced by 24% at 3 months post injection. Considering the frequent use of BoNT-A to treat spasticity in children with CP, it is necessary to understand the impact of this treatment in this population, particularly with regard to muscle function and structure.

It is well recognised that muscle structure and size is associated with muscle strength in the adult and adolescent population.[21, 22] Children with CP have been shown to have smaller and weaker muscles.[23] Over the last decade, weakness has been increasingly recognised as a significant motor impairment,[22-24] purportedly affecting functional ability in children with CP.[25, 26] A recent review of the literature on spastic CP found consistent evidence for small muscle size as indicated by reduced muscle volume, cross-sectional area, thickness, and belly length in comparisons of paretic muscles with non-paretic and typically developing muscles.[24] Barber et al.[25] described volumetric deficits of 22% in the medial gastrocnemii of young children (2–5y) with spastic CP compared with typically developing children. For a population already predisposed to decrements in muscle size and strength, a treatment that potentially leads to further atrophy and weakening of the muscle should be well understood.

The concern is that whilst children appear to do well functionally with BoNT-A treatment, there is evidence for an atrophic effect of BoNT-A, documented in typically developing muscles in human and animal research, that, to the authors' knowledge, has not yet been investigated in pathological muscles. This study is the first to investigate the immediate morphological alterations in muscles of the lower limbs of children with CP following BoNT-A treatment for spasticity management. To report the comprehensive effect of the neurotoxin we looked at the alterations of the injected synergist and antagonist muscles after BoNT-A injections, and included measures of strength and functional ability. It was hypothesised that the injected muscle would display a reduction in muscle volume that could also result in a reduced strength capacity.

Method

Participants

Ten males and five females with spastic diplegic CP classified in Gross Motor Function Classification System (GMFCS) levels I (n=9) and II (n=6) were recruited via the spasticity management service at Princess Margaret Hospital (PMH) in Perth, Australia. The children ranged from 5 to 11 years (mean age 8y 5mo, [SD 1y 10mo]; mean height 129.87cm [SD 10.92cm]; mean weight 27.97kg [SD 7.43kg], and a body mass index of 16.34 [SD 2.4]). No child had undergone serial casting in the previous 6 months and no child had a history of lower limb surgery. Informed written consent was obtained from the parents of all the participants. Ethical approval for the study was obtained from the Ethics Committee of PMH (1766) and from The University of Western Australia.

Spasticity was assessed bilaterally using the modified Ashworth Scale (MAS)[26] by the same assessor, blinded to timing of BoNT-A treatment. All 15 children were receiving BoNT-A for spasticity management bilaterally in their lower limbs, and had already received a minimum of two injection series of BoNT-A before the series included in the study (maximum series 15, mean series 8.93). The muscle(s) selected for injection were determined by clinical assessment and functional goals, the total dose of BoNT-A (Botox®; Allergan, Irvine, CA, USA) was empirically selected for each muscle, and injections were guided by ultrasound. All participants received BoNT-A bilaterally to the gastrocnemius (30 legs injected; 2–6U/kg) and five participants also received BoNT-A to bilateral medial hamstring muscles (semimembranosus and semitendinosus; 10 legs; 2–4U/kg). Other muscles injected included the soleus (4 legs; 1–2U/kg) adductors (2 legs; 1U/kg), rectus femoris (2 legs; 1U/kg) and tibialis posterior (1 leg; 1U/kg). No child had more than three muscles injected per leg.

Of the five children who received BoNT-A to both the medial hamstrings and the gastrocnemius, the mean age was 7 years, 10 months (SD 2y 7mo), three were classified in GMFCS level I and two in GMFCS level II.

Design and procedures

The study was a before and after design. Children completed two assessments: (1) timed approximately 2 weeks before their scheduled BoNT-A (pre), and (2) on average 5 weeks (SD 1wk, range 3–6wk) post injection (post). Children were a minimum of 6 months post-previous BoNT-A injection at the initial assessment.

Muscle volume

Muscle volume of the lower limbs was assessed using magnetic resonance imaging (MRI). Efforts were made to standardise the time of day that each participant completed their scan. Axial spin-echo T1-weighted MRI scans were acquired bilaterally from the level of the ankle malleoli to the iliac crest while participants lay prone in a 1.5T whole-body MR unit (Magnetom Sonata Maestro Class; Siemens Medical Solutions, Erlangen, Germany). The children were positioned in neutral hip rotation, maintained passively using standard patient positioning with foam pads. Images of the thigh and lower leg were collected using a repetition time of 572ms, echo time of 13ms, slice thickness of 5mm, and mean inter-slice gap between 5 and 7mm. A matrix size of 256 × 160mm was used for all thigh scans and 256 × 144 for the lower leg, and the field of view (280–300mm) was varied to maximise in-plane resolution for each scan. The mean number of axial slices for the thigh was 30.27 (SD 1.53) and for the lower leg was 28.60 (SD 1.99).

Magnetic resonance images were transferred to an independent workstation for digital reconstruction. Isotropic voxel size was obtained using a trilinear interpolation routine. Muscles were manually traced (see Fig. 1) and segmented for all participants using a digitisation tablet (Intuos2, Wacom Technology Corp., Vancouver, WA, USA) and Mimics software (version 9.0, Materialise, Leuven, Belgium). The segmented muscles included the semitendinosus, semimembranosus, biceps femoris, rectus femoris, vastus lateralis and medialis, medial gastrocnemius, lateral gastrocnemius, soleus, and the tibialis anterior. Muscle volume was calculated by summing the number of voxels contained within each muscle and multiplying by the voxel dimension (1mm3). To account for the large variation of age and height, and to account for confounding effects of skeletal growth, muscle volume was normalised to femur and tibia length as determined from MRI scans with Mimics software. Muscle volume was then presented as a percentage of muscle volume (cm3) per femur (for thigh, cm) or tibial length (for shank, cm).

Figure 1.

Creating a 3-dimensional model of the leg from an MRI scan using Mimics software (version 9.0, Materialise, Leuven) to determine muscle volume.

We have previously reported the intraclass correlation coefficient (ICC) of muscle volume using this method,[23] reporting that both intra- and interrater reliability were high with ICC values consistently greater than 0.92 and 0.94 (CI 95%) respectively, for quadriceps and hamstring muscle volume.[23] In the present study, intrarater reliability was high with an ICC value of 0.97, tested in a random selection of five MRI scans of the hamstring muscles pre and post injection, wherein the investigator was blinded to the scan time point. In this study, all data were processed and analysed by one investigator (SW).

Muscle strength

A Biodex System-3 dynamometer (Biodex Medical Systems, Inc., Shirley, NY) was employed to assess isometric strength of the knee flexors and knee extensors. Children performed three maximum isometric contractions of the knee flexors and knee extensors bilaterally, with randomisation of the order of which side was tested. Trials evaluated muscle peak torque normalised to body weight (PT/BW) in a static posture with the knee flexed at 90°. Standard straps constrained the upper body and pelvis to avoid the contribution of other muscles to the assessment. The lower limb segment was attached to the Biodex arm using the standard velcro straps, leaving the ankle joint unconstrained during the knee flexion/extension tasks. Continual verbal encouragement was provided throughout the assessment with adequate rest and recovery in between contractions to minimise muscle fatigue.

A hand-held dynamometer (HHD; Model 01163; Lafayette Instrument Company, Lafayette, IN, USA) was used to determine maximal isometric strength for plantar flexion and dorsiflexion by a trained and experienced physiotherapist. For plantar flexion, the child lay supine with straight legs, the HHD was placed under the padding of the foot with the foot positioned as close to 90° as achievable. The child was instructed to ‘point the toes down’, pushing against the HHD, whilst another tester constrained the child at the knee and the hip. Dorsiflexion was measured with the child seated upright, knees flexed at 90° and the HHD positioned on top of the foot. The child was constrained at the knee, hips and torso, with arms crossed over the chest. For consistency of results, the same assessor recorded all measurements at both assessments.

Functional ability

The 6-minute walk test (6-MWT) was used to measure functional walking capacity.[27] In addition, children performed the Timed Up and Go (TUG) assessment from an adjustable-height chair in accordance with the test protocol.[28]

Statistical analysis

To estimate the effect of BoNT-A, we determined whether the difference in measures of muscle volume and strength was significantly different from zero, accounting for possible within-person correlations by using a mixed model with a random intercept. For functional ability measures, a comparison of the pre- and post-BoNT-A group means were analysed using repeated measures, two-tailed t-tests with a confidence interval of 95%, and an alpha level of 0.05. Effect sizes were determined using Cohen's d equation using the mean standard deviation of the two scores being compared. Each individual's percentage change was grouped and averaged to provide a mean percentage change for measures presented in the included tables. A post-hoc power analysis using the sample size of 15 revealed a power of 0.64.

Results

Spasticity

All of the children in this study had a clinically appropriate response to BoNT-A treatment, demonstrating improvements in spasticity. Total scores from the MAS were summated to provide a representation of spasticity, with a lower score indicating less spasticity in the lower limbs. Spasticity scores for the entire sample significantly decreased immediately after BoNT-A injection, from a mean score of 10.07 (SD 3.94) to 8.27 (SD 2.12), [t (14)=2.36, p=0.03, effect size (ES)=1.17].

Plantar/dorsiflexors

This section on plantar/dorsiflexor muscle volume and strength changes relates to all 15 children in the study as they all received BoNT-A to the gastrocnemius muscles.

Muscle volume

There were no significant changes in the muscle volume for the total plantar flexion muscle group (the combination of the soleus, and lateral and medial gastrocnemius) across the two time points [t(14)=−0.195, p=0.848, ES=0.01]. The combined volumes of the lateral and medial gastrocnemius showed a significant decrease in gastrocnemius volume [t(14)=−2.42, p=0.03, ES=0.17] of 4.47% (SD 8.57%), whilst the soleus volume had a significant increase of 3.96% (SD 8.05%), [t(14)=2.20, p=0.04, ES=0.10]. The prime antagonist muscle of the dorsiflexion group, the tibialis anterior, remained constant with no significant change in muscle volume [t(14)=0.64, p=0.53, ES=0.05] (see Table 1).

Table 1. Muscle volumes and strength (mean [SD])of the lower leg for 15 children (30 legs) receiving BoNT-A to the gastrocnemius muscle group
 Pre treatmentPost treatmentMean individual change%p (effect size)
  1. All values for muscle volume are expressed as % of muscle volume (cm3) divided by tibia length (cm). Strength values are in kilogrammes. Pre and post group means and the means of each individuals percentage change are presented. aSignificance at p<0.05.

Gastrocnemius3.00 (0.93)2.84 (0.85)−4.47 (8.57)0.01a (0.17)
Soleus4.32 (1.42)4.46 (1.38)3.96 (8.05)0.04a (0.10)
Total PF7.32 (2.25)7.30 (2.12)0.45 (7.15)0.85 (0.01)
Total DF1.27 (0.34)1.29 (0.32)2.64 (12.02)0.53 (0.05)
PF strength29.80 (7.62)30.46 (8.71)2.45 (17.92)0.28 (0.24)
DF strength12.04 (4.73)12.23 (3.85)1.23 (20.97)0.79 (0.25)

Muscle strength

There were no significant differences reflected in the measures of strength as measured by the HHD for the plantar flexors or dorsiflexors (see Table 1).

Knee flexors/extensors

In reporting the measured morphological response of the hamstring muscles (knee flexors) and quadriceps (knee extensor) muscle, the results of the 10 children (20 legs) who received BoNT-A to the gastrocnemius muscles are reported separately from the five children (10 legs) who received medial hamstring muscle injections in addition to the gastrocnemius muscles.

Muscle volume

BoNT-A treatment in the gastrocnemius (10 children, 20 legs)

There was a non-significant change of 2.85% (SD 6.17%) in the muscle volume of the total hamstring muscle group [t(9)=1.79, p=0.11, ES=0.10] in the 10 children who received BoNT-A in the gastrocnemius. When considering the individual muscles of the hamstring group, we found that whilst the biceps femoris approached a significant increase of 4.67% (SD 8.54%), [t(9)=−2.22, p=0.05, ES=0.19], the muscle volume of the medial hamstring muscles (semitendinosus and semimembranosus) remained relatively constant [t(9)=0.53, p=0.61, ES=0.04].

The antagonist to the hamstring muscles, the quadriceps muscle (knee extensor) group, increased its total muscle volume significantly by 4.23% (SD 5.84%), [t(9)=2.66, p=0.03, ES=0.17].

BoNT-A treatment in the gastrocnemius and medial hamstring muscles (5 children, 10 legs)

In the group of five children (10 legs) who had BoNT-A injections in both the gastrocnemius muscle group and the medial hamstring muscles, there was no significant change in total hamstring (knee flexor) muscle volume [t(4)=−1.45, p=0.22, ES=0.29]. However the medial hamstring muscles did show a decrease of 5.89% (SD 9.23%) in muscle volume but this was not statistically significant [t(4)= −2.01, p=0.08, ES=0.45]. The prime antagonists, the quadriceps muscle group (knee extensor), displayed no significant change in muscle volume [t(9)=0.19, p=0.86, ES=0.01; see Table 2].

Table 2. Pre and post group means (SD), with the grouped means of each individuals percentage change in muscle volumes of the thigh for 10 children (20 legs) receiving BoNT-A to the gastrocnemius muscle group and for five children (10 legs) receiving BoNT-A to the medial hamstring and gastrocnemius muscles
Muscle group MVBoNT-A treatment gastrocnemius muscles only n=10 children (20 legs)BoNT-A treatment gastrocnemius and medial hamstring muscles n=5 children (10 legs)
Pre treatmentPost treatmentMean individual % changep (effect size)Pre treatmentPost treatmentMean individual % changep (effect size)
  1. All values are expressed as a percentage of muscle volume (cm3)/femur length (cm). aSignificance at p<0.05, bApproaching significance at p<0.06. BoNT-A, botulinum toxin type A.

Medial hamstring3.52 (0.94)3.56 (0.84)1.80 (6.86)0.61 (0.04)2.87 (0.47)2.68 (0.38)−5.89 (9.23)0.08 (0.45)
Biceps femoris2.60 (0.56)2.71 (0.55)4.67 (8.54)0.05b (0.19)2.34 (0.39)2.31 (0.40)−1.14 (8.34)0.67 (0.08)
Total hamstring6.12 (1.43)6.26 (1.34)2.85 (6.17)0.11 (0.10)5.21 (0.82)4.99 (0.74)−3.84 (8.07)0.22 (0.29)
Total quadriceps12.51 (2.79)12.96 (2.57)4.23 (5.84)0.03a (0.17)12.41 (2.21)12.44 (2.49)−0.14 (2.71)0.86 (0.01)

Muscle strength and function

BoNT-A treatment in the gastrocnemius muscles (10 children, 20 legs)

From the children receiving BoNT-A injections in the gastrocnemius muscles only, there was no statistical change in the knee flexor muscles (−2.19%, SD 42.34%, t(9)=−1.15, p=0.28, ES=0.26), or the knee extensor muscles (11.70%, SD 33.72%, t(9)=0.39, p=0.71, ES=0.05; see Table 3).

Table 3. Pre and post treatment mean (SD) values of knee flexor and knee extensor strength (isometric peak torque), with the grouped mean of individual percentage changes in 10 children (20 legs) receiving BoNT-A to the gastrocnemius muscle group, and five children (10 legs) receiving BoNT-A to the medial hamstring and gastrocnemius muscles
Isometric Pk Torque (Nm/Kg)BoNT-A treatment gastrocnemius muscles only n=10 children (20 legs)BoNT-A treatment gastrocnemius and hamstring muscles n=5 children (10 legs)
Pre treatmentPost treatmentMean individual % changep (effect size)Pre treatmentPost treatmentMean individual% changep (effect size)
  1. Values are normalised to body weight. aSignificance at p<0.05. BoNT-A, botulinum toxin type A.

Knee flexion83.86 (47.79)72.36 (40.20)−2.19 (42.34)0.28 (0.26)62.68 (7.94)56.09 (15.12)−9.75 (25.46)0.34 (0.57)
Knee extension205.86 (112.74)211.85 (108.44)11.70 (33.72)0.71 (0.05)207.10 (35.13)222.24 (44.17)8.70 (18.27)0.67 (0.38)

In the 10 children having BoNT-A to the gastrocnemius muscles only, there were no statistically significant changes in function after BoNT-A as assessed by the TUG, completing the task in 4.50s (SD 0.95 s) before BoNT-A, compared with 4.66s (SD 0.98s) after BoNT-A, a 3.94% increase in time (SD 12.54%), [t(9)=−0.92, p=0.38, ES=0.16]. There was also no statistically significant change in the distance covered in the 6-MWT: 577.49m (SD 73.53m) to 559.89 m (SD 72.14m), a 2.89% decrease in distance (SD 6.45%), (t(9)=1.44, p=0.18, ES=0.24).

BoNT-A treatment in the gastrocnemius and medial hamstring muscles (5 children, 10 legs)

In the measures of muscle strength, isometric peak torque showed no statistically significant change in the knee flexor [t(4)=−1.09, p=0.34, ES=0.57] or knee extensor muscles (t(4)=−0.45, p=0.67, ES=0.38; see Table 3).

In the five children who had gastrocnemius and medial hamstring muscles injected, there were no statistically significant changes in the TUG after BoNT-A, completing the task in 4.76s (SD 0.91s) before and 4.68s (SD 0.91s) after BoNT-A, a 0.89% decrease in time (SD 11.79%), [t(4) = 0.36, p=0.74, ES=0.09]. Distance covered in the 6-MWT also did not change statistically after BoNT-A, from 512.82m (SD 103.65m) to 517.55 m (SD 99.77m), a 1.34% increase (SD 8.78%) [t(4)=−0.23, p=0.83, ES=0.05].

Discussion

This study has, for the first time, provided an in-depth investigation into the immediate morphological responses to BoNT-A injections of muscles of the lower limbs of children with CP. In agreement with Schroeder et al.'s.[15] research on BoNT-A in the healthy muscle, reports from the cosmetic industry,[17-19] and animal research,[13, 20] we also found evidence of muscular atrophy in BoNT-A injected muscles. An average reduction of 4.5% was measured in the gastrocnemius muscle of 15 children undergoing BoNT-A for spasticity management, whilst five of those children who also had the medial hamstring muscles injected had an average decrease of 5.9% in medial hamstring muscle volume. These results are reassuring in that whilst we did measure atrophy, it appears to be much smaller than the concerning reports of an almost 20% reduction as seen in healthy adult muscles of the gastrocnemius,[15] the 22 to 25% reduction in masseter muscle thickness,[17, 18] and the 32 to 50% decreases seen in animal research[13, 20] These responses, however, should be viewed in perspective and compared with caution. Previous reports of muscle atrophy use a variety of methods to measure muscle morphology, making it difficult to draw true comparisons of results. In addition to this, the research in healthy muscles reports the response of muscles to the first series BoNT-A injection.

In our study in pathological muscles, children had already received a minimum of two series of BoNT-A; the responses reported here are likely to not be as strong as they may have been if it were to their first injection of BoNT-A. Fortuna et al.'s[13] animal study on repeat injections of BoNT-A reported that the greatest amount of atrophy occurred as the muscles first response to BoNT-A. The animal studies also indicated that muscle atrophy was not homogenous within a muscle group exposed to BoNT-A, and could be affected by the muscle fibre type,[13] a factor that could be highly variant in individuals with spastic muscle. The reality is that muscles in children with CP are variable, and the use of BoNT-A for spasticity management is repeated over many years. This study provides the first snapshot into morphological alterations in this population. Research now needs to look at alterations after the muscles first BoNT-A injection, and delve further into the muscles' response following a sequential series of BoNT-A.

In exploring the alterations of each of the injected, synergist and antagonist muscles after BoNT-A injections, an interesting outcome emerged. Whilst the volume of the injected gastrocnemius muscle decreased by 4.5%, the soleus muscle matched the decrease with a nearly 4% increase in volume, with a consequential total volume for the plantar flexion muscle group that is relatively stable after the BoNT-A. This is a potential indication of a compensatory reaction to synergist hypertrophy. The prime antagonist muscle, the tibialis anterior, also did not show any changes in muscle volume, whilst increases in muscle volume were also measured in the hamstring muscles (changes in the medial hamstring muscles were not significant while the changes in the biceps femoris were significant) and quadriceps muscles (out of the 10 children not receiving BoNT-A in the hamstrings).

The quadriceps muscle volume increase was complemented by an 11.7% increase in knee extensor strength, in accordance with evidence that muscle size is related to muscle strength.[21, 22] Knee flexor strength, however, indicated more of a decreasing trend, possibly in part as a result of the contribution of the gastrocnemius in knee flexion, weakened as a response to BoNT-A. However, with no statistical significance, and small effect sizes, we cannot extrapolate our observations of strength but merely speculate. The HHD measures of strength for plantar flexors and dorsiflexors also showed no significant changes, and large variations in scores. Indeed a slight increase in isometric strength was measured in the plantar flexors and dorsiflexors after BoNT-A, possibly attributed to the participants' ease in achieving range after BoNT-A treatment. Unfortunately, our measures of strength are limiting in that they cannot separate out the individual contributions of each muscle, but instead provide us with a more holistic assessment of the torque about a joint. The changes in muscle volume resulted in no significant difference in the strength of the participants at different joints; this is further supported by the functional tests, which demonstrated that the potential detriments in strength that have been highlighted in the literature may not be such a great cause for concern – at least for this patient group, in the short-term and after multiple injections.

It was a slightly different story for the five children who had BoNT-A injected to both the medial hamstring and gastrocnemius muscles. As previously mentioned, there was a decreasing trend in medial hamstring volume that approached significance; the synergist biceps femoris muscle did not increase in response, but rather showed a small decrease in muscle volume. This resulted in an overall decrease of 3.8% for the total hamstring group muscle volume that corresponded with a decrease in knee flexor strength, which, although not statistically significant, had a moderate effect size of 0.57. The antagonist quadriceps muscle volume showed no notable change, whilst there was no change in knee extensor strength. Whilst these minor alterations may be explained by the muscle strength–size relationship,[21, 22] we cannot rule out the possibility of any neurological alterations as a result of BoNT-A altering innervation patterns.

A possible confounding limitation in our study was that four other muscles in the leg in addition to the gastrocnemius or hamstring muscles were targeted for injection for five of the children included in this study. This was determined by a clinical decision and was out of the researcher's control and may have compromised results. However, doses were minimal and only two children received injections into an additional muscle in the gastrocnemius- and hamstring-injected group. The soleus muscle, as well as the gastrocnemius muscle, was injected in two of the children in this study; when we removed their results, the increase of the soleus muscle was even greater at 4.4% (with the gastrocnemius muscle volume decrease changing to 5.4%), which supports our premise of synergist hypertrophy of the soleus. Instituting MRI analysis in research designs is costly and time consuming; unfortunately the addition of a control group (no treatment or saline injected) was not feasible for this study. Our measurements of change could be queried as many of our statistics include small effect sizes, for example the gastrocnemius decrease in muscle volume was significant but had a small effect size, whilst the drop in the injected hamstring muscle approached significance yet had a moderate effect size. In addition to this, interpretations of our results are limited by our small sample size, and statistics have not been protected against the possibility of returning false positives (see Tables SI–SVIII, online supporting information). With this in mind, it is important that findings within this study be viewed with caution. Nonetheless, we believe the new information presented in this study to be clinically important.

The morphological alterations in spastic muscle in our sample of children with CP did not show atrophy as severe as that shown in animal studies and healthy muscle.[13] This is certainly optimistic for the continued future BoNT-A use; however, as mentioned earlier, the initial response to BoNT-A in spastic muscle may be different. Despite this, this study also appeared to uncover the possibility of compensatory hypertrophy in synergist muscles as a response to atrophy of the injected muscles. It should be borne in mind for the future application of BoNT-A that the morphological response of muscle to BoNT-a treatment may not be limited to the injected muscle alone. The hypertrophy is probably a product of an increased physiological demand on the synergistic muscle, the increasing work and load of the synergistic muscle, leading to work-induced hypertrophy.[29] In terms of muscle strength, our results showed no significant changes after BoNT-A treatment but may indicate potential changes, in particular in children who had multiple muscles injected. It is also important to note that the morphological alterations following BoNT-A did not have any significant detrimental effects on the children's functional performance. In the present study the effects of BoNT-A injections where measured to correspond with the BoNT-A taking peak effect; to get a deeper understanding, follow-up assessments would provide us with a clearer picture as to how the muscles are responding to the BoNT-A in the longer term, considering that pharmacological active period of BoNT-A stretches beyond the period measured within this study.

Conclusion

This is the first study, to our knowledge, to report upon the immediate morphological alterations in spastic muscles following BoNT-A treatment. It has demonstrated reductions in muscle volume in the injected muscle, and hypertrophy in synergistic muscle as a compensatory reaction in children with CP. This potentially has a role in explaining the good clinical and functional response in children after BoNT-A injections. The atrophy measured in our sample of children with CP was not as large as that reported in research in animal and healthy muscle. Whilst strength deficits were not seen following a single site injection in this study, attention should be paid to muscle alterations of children undergoing treatment to multiple sites on repeated occasions.

Acknowledgements

This project was supported by the Princess Margaret Hospital (PMH) Foundation Grant as well as by the University of Western Australia's (UWA) Research and Development Awards. The authors thank the Department of Paediatric Rehabilitation, PMH, for their support and assistance with recruitment, the Department of Diagnostic Imaging, PMH, for their assistance and knowledge in data collection, and the contribution of the School of Sport Science, Exercise & Health at UWA for the use of equipment and facilities. They also thank Tania Shillington, Bree Dwyer and Martin Spits for their role with the data collection, and Associate Professor Eve Blair and Peter Jacoby (Telethon Institute for Child Health Research) for their statistical advice in the preparation of this manuscript.

Ancillary