Medial gastrocnemius specific force of adult men with spastic cerebral palsy.

INTRODUCTION: Muscle weakness determines functional impairment in spastic cerebral palsy (SCP). Measurement of specific force (SF) allows for strength comparison with unimpaired populations (controls) accounting for neural (activation and coactivation), architectural (fascicle length and pennation angle), and structural differences (moment arm length).
METHODS: Medial gastrocnemius (MG) SF (and its determinants) was assessed in both paretic and non-paretic legs of 11 men with SCP and 11 age-matched controls during plantarflexion maximal voluntary isometric contraction (MVIC).
RESULTS: SCP fascicles were 28% longer than control fascicles (P < 0.05). Pennation angle of SCP patients was 41% smaller than in controls. The physiological cross-sectional area of SCP MG patients was 47% smaller than in controls (P < 0.05). There was no difference in SF between controls and SCP patients.
CONCLUSIONS: Weakness in SCP is primarily attributable to deficits in agonist activation and muscle size; consequently, SF measured in the MG is similar between SCP and controls. Muscle Nerve 56: 298-306, 2017.

anatomical cross‐sectional area

Cerebral Palsy International Sports and Recreation Association

coefficient of variation

dorsiflexion

electromyography

lateral gastrocnemius

medial gastrocnemius

Gross Motor Function Classification System

musculotendinous junction

maximal voluntary isometric contraction

physiological cross‐sectional area

plantarflexion

range of motion

spastic cerebral palsy

soleus

tibialis anterior

Muscle weakness in children with spastic cerebral palsy (SCP) has been shown to originate from impaired neural signaling, smaller muscle size, and altered architecture in the paretic musculature.1, 2, 3, 4 Such weakness of the paretic muscles has been shown to contribute to differences in gait patterns5 and to limit motor control performance.6, 7 Although muscle weakness may limit the performance of daily tasks, only a few studies have addressed the underlying determinants of weakness specifically in adults with SCP.

It has been reported that larger deficits in strength exist in the more distal paretic muscles of the lower limbs in individuals with SCP.8 With this in mind, Elder et al.1 reported that isometric plantarflexion (PF) torque of the paretic limb relative to the anatomical cross‐sectional area (ACSA; in Nm/cm2) in children with hemiplegic SCP was ∼40% lower compared with their non‐paretic limb or the dominant limb of individuals without neurological impairment. Similarly, although such findings are crucial to furthering our understanding of the determinants of muscle weakness, it has been well documented that ACSA measurements underestimate the true physiological cross‐sectional area (PCSA) of pennate muscles.9, 10, 11 In support of these findings, correlations between muscle force during PF maximal voluntary isometric contraction (MVIC) and PCSA have been shown to be considerably higher than correlations with ACSA (r = 0.72 vs. r = 0.92, respectively12).

Although muscle size is the greatest determinant of muscle strength, architectural characteristics of pennate muscles are also known to influence contractile function. Changes in architecture as a result of resistance training13 and bed‐rest interventions14, 15 have been suggested to impact the force output of a muscle in individuals without neurological impairment. In children with spastic hemiplegic SCP, the resting fascicle lengths of the paretic muscle in the gastrocnemius have been reported to be smaller when compared with the muscle of children without neurological impairment2 and the contralateral non‐paretic limb.3 On the other hand, resting fascicle pennation angle of the paretic medial gastrocnemius (MG) did not differ when compared with the non‐paretic muscle of individuals with SCP and the dominant limb of control participants without neurological impairment.2 Conversely, during PF MVIC trials of the MG in young adult men and women, the paretic fascicle length was not different from that of normal control participants.16 It is for this reason that measures of contractile area in SCP should consider the possible morphological differences of the muscle, such as the PCSA. Barber et al.16 showed how MG PCSA can account almost entirely for differences in PF MVIC torque between those with and without CP. However, a more complete assessment of the intrinsic strength of the muscle (and the neural and morphological determinants) would involve the measurement of specific force.

Specific force, defined as the fascicle force / PCSA, is a measure of intrinsic muscle strength, which accounts for these aforementioned architectural and morphological characteristics of the muscle, plus the moment arm length and neural determinants of strength (agonist activation and coactivation).9, 10, 17 Moment arm length is a primary determinant of the effective translation of muscle force to torque.18, 19 Despite the excessive PF and hypothetical impact this may have on the Achilles tendon moment arm,20 particularly given the joint deformation in the ankle,21 there appears to be some preservation of the muscle‐joint configuration, at least in terms of indirect measures of the moment arm in children with SCP.22 There is no information on moment arm lengths in the paretic and non‐paretic limbs of adults with SCP.

In terms of neural impairment in SCP, increased coactivation of the antagonist16 and reduced activation of the agonist23 are known to contribute to strength decrements between individuals with and without SCP. Therefore, given the established neural,23 architectural,16 and possible joint differences20 of individuals with SCP, the aim of this study was to determine whether differences in strength at the fascicle level persist when these morphological and neurological factors are accounted for through the calculation of specific force. Consistent with other neuromuscular conditions that have shown evidence of lower specific force or muscle quality (e.g., sarcopenia9 and disuse24, it was hypothesized that the MG specific force of the paretic limb would be lower than that of the non‐paretic limb and the dominant limb of individuals without neurological impairment (hereafter, this group will be referred to as “controls”).

METHODS

Participants

Twenty‐two active and ambulant men gave written informed consent to participate in our study. Eleven of the participants had spastic hemiplegic CP [mean (SD): age 21.2 (3.0) years; height 1.79 (0.10) m; mass 70.0 (12.5) kg], and 11 control participants had no history of musculoskeletal or neurological impairment [mean (SD): age 21.8 (2.2) years; height 1.81 (0.04) cm; mass 79.0 (8.4) kg]. Each participant with SCP rated between II and III on the modified Ashworth scale and had been formally classified independently by individuals from the Cerebral Palsy International Sports and Recreation Association (CPISRA). All participants with SCP rated as level 1 on the Gross Motor Function Classification System (MGFCS). Both the paretic and non‐paretic legs were tested in each participant with CP, whereas the dominant limb was assessed in the control participants. All participants were free from lower limb injury and had not received any form of medication to reduce the effects of spasticity within the last year. None of the participants had a history of any surgical procedures on their lower limbs that would have affected the data collection from the sites assessed. The study was approved by the local ethics committee at Manchester Metropolitan University and conformed to the standards set by the latest revision of the Declaration of Helsinki.25

Protocol

Participants attended the laboratory on 2 occasions. During the first visit, familiarization was carried out, which included a series of 6 PF MVIC tests followed by a series of submaximal percutaneous electrical stimulations. During the second visit, participants were assessed for resting measures of muscle size and moment arm length, which were then followed by the MVIC tests.

Strength Measurements

The PF MVIC torque was recorded with the participants secured to an isokinetic dynamometer (Cybex Norm; Cybex International, Inc., Medway, Massachusetts) in the seated position with the back angle reclined at 65°. The participant's knee was secured in full extension with Velcro straps, which were positioned proximal to the knee. All participants were able to achieve full knee extension in this position. The medial malleolus was aligned visually with the dynamometer's central axis of rotation, and 2 Velcro straps were used to secure the foot to the footplate to minimize heel displacement. The participant's hips were also secured to the seat to limit extraneous movement during PF MVIC trials. All participants warmed up by performing 3 submaximal isometric contractions with the ankle angle at 0° (the individual's anatomical zero), each of which were separated by a 1‐min rest period. In this instance, 0° was defined as the foot at 90° to the tibia. After the warm‐up, the participant's ankle remained fixed at 0°, and 2 PF MVICs were obtained, separated by a 2‐min rest period. Throughout all MVIC trials, participants were encouraged verbally to exert as much force as possible, and online visual feedback was provided on a monitor. Dorsiflexion (DF) MVIC was completed after the PF MVICs using the same testing posture and protocol in order to calculate tibialis anterior (TA) coactivation. All but 2 of the SCP participants were able to achieve the 0° ankle position [maximum dorsiflexion ROM was –5.30° (0.48°) for SCP participants26]. For these 2 participants, “0°” was measured at 4° and 6° PF. It should be noted that fascicle length was longer in SCP during PF MVIC (see Results), and passive torque at 0° was no different than that of controls [controls 19.5 (6.80) Nm, paretic 22.7 (10.3) Nm]. Furthermore, based on the MVIC torque angle relation described in previous work,16 the influence of this more plantarflexed position in these 2 participants could be estimated to have reduced their PF MVIC by ∼5 Nm, and was considered unlikely to have influenced the significance of the results presented in what follows.

Agonist Activation

To account for any deficit in MVIC torque in the quantification of specific force, 2 supramaximal stimuli were applied to the muscle (pulse width 50 µs). The first stimulus was applied during MVIC. The second was applied approximately 2 s after the first when the torque had returned to levels equivalent to that observed before MVIC, from which voluntary activation levels of gastrocnemius were calculated.17, 23 The stimulus was delivered by applying 2 percutaneous stimuli (DSV Digitimer Stimulator; Digitimer, Ltd., Welwyn Garden City, UK) to the gastrocnemius using rubber stimulation pads (size range: 38 mm × 89 mm to 76 mm × 127 mm; VersaStim; ConMed Corp., Utica, New York), both of which were placed transversely distal to the popliteal crease and myotendinous junction of the soleus. The amplitude of the stimuli was determined before interpolation while the participant was in a relaxed state, administering twitches starting from 50 mA and increasing in increments of 50–100 mA, until no further increase in twitch torque was quantified. The voluntary activation level of each participant was assessed using the trial that produced the highest contractile torque. Agonist activation was calculated by dividing the supramaximal twitch torque during MVIC by the post‐MVIC twitch torque, consistent with Morse et al.9 If there was a deficit in muscle activation (a value < 100%) and assuming a linear relationship between MVIC torque and agonist activation,27, 28 a correction was made with PF MVIC, which was calculated as: (PF MVIC torque / 100) × deficit in voluntary activation. This value was subsequently added to the MVIC torque along with torque contributions in coactivation to estimate PF MVIC net torque.

Coactivation

TA electromyographic (EMG) activity was recorded using 2 pre‐gelled, unipolar, 10‐mm, Ag‐AgCl percutaneous electrodes (Medicotest; Ambu A/S, Ballerup, Denmark). Boundaries of the TA were determined using ultrasonography to ensure accurate placement of each electrode along the midsagittal axis of the muscle and to reduce cross‐talk. Two electrodes were placed distally at two‐thirds of the TA length, and a reference electrode was placed over the lateral epicondyle of the femur. Before placement of the electrodes, the area was shaved and cleaned with an alcohol swab to remove residual skin cells and oils and reduce skin impedance. Raw EMG data were recorded at 2,000 Hz, with high and low bandpass filters set at 10 and 500 Hz, respectively, with a notch filter set at 50 Hz. The integral of the root mean square of the raw signal 0.5 s either side of the MVIC PF torque was used to quantify the level of muscle coactivation. The torque produced by the DF during PF MVIC was estimated by assuming a linear relationship between torque and EMG activity, as previously reported.29 The relative contribution of antagonist coactivation from the DF MVIC was added to estimate PF net torque along with any correction in agonist activation as mentioned previously.

Muscle Volume

B‐mode ultrasonography (AU5; Esaote Biomedica, Genova, Italy) was used to obtain several axial‐plane images of the MG to measure ACSA.30 The MG proximal insertion and the musculotendinous junction (MTJ) were marked to identify the 50% muscle length. Strips of Micropore tape were placed axially across the midline of the MG at approximately 3.5‐cm intervals. These strips of tape were used as echo‐absorptive markers that project a shadow onto the ultrasound image to provide a positional reference into the scanned structures. With the probe in an axial plane, a recording of the probe moving from the medial to the lateral border of the MG was obtained. Individual images were extracted from the recording offline and used to reconstruct the muscle by overlapping anatomical landmarks and external markers, as described in previous work.30 ImageJ software (version 1.34; National Institutes of Health, Bethesda, Maryland) was used to measure the ACSA of the reconstructed MG, from which volume was estimated, as described in what follows.

The use of ultrasonography in the measurement of ACSA has been validated against MRI in the rectus femoris [R 2 = 0.90, coefficient of variation (CV) 6.7%31 and vastus lateralis (R 2 = 0.98, CV 1.7%30]. However, no current data exist on techniques for using a single ACSA measurement to predict MG muscle volume. Based on an approach described previously,32 a retrospective analysis of MRI scans from an adult male population was carried out to allow more accurate predictions of muscle volume for this study. Briefly, MG from 11 adult men [age 24.7 (4.7) years, height 1.79 (0.08) cm, mass 76.9 (12.4) kg] had been scanned previously in the transverse plane in 10% increments, from 10% to 90% of MG muscle length using 0.2‐T MRI (E‐Scan; Esaote Biomedica). At each 10% increment, the ACSA of the MG was measured (OsiriX Medical Imaging Software; OsiriX, Atlanta, Georgia) and presented relative to the maximum ACSA. A third‐order polynomial curve was then fitted through the ACSA at each section relative to the maximum ACSA [see Eq. (1) in what follows]. The MG volume was then estimated by integrating the regression equation over the measured length of the muscle at intervals equivalent to 10% of measured muscle length. Compared with MG volumes acquired from this subgroup using contiguous ACSA measures along the muscle length at 1‐cm intervals (e.g., see Morse et al.33, there was no significant difference in estimated MG volume [301 (65) cm3 and 294 (83) cm3]. The bias (±95% confidence limits) tended toward negative, but was low (–7.63 ± 22.4 cm3), equivalent to 3% of the measured volume. There was also a significant correlation between measured and predicted MG volume (R 2 = 0.86).

Therefore, this regression‐based approach was adopted using the previously validated ultrasonography measure of ACSA to estimate MG muscle volume.

The regression equation used for the estimation of MG volume in the present participants was: where MG ACSA is relative to maximum MG ACSA (y, where ACSA maximum = 1) and expressed relative to muscle length (x, where 100% of muscle length = 1), from which segmental volumes were estimated and summed to calculate MG volume from measured ACSA at 50% of muscle length.

It should be acknowledged that this regression is based on a healthy adult male population and, although the similar muscle length between the participants [controls 25.7 (2.0) cm, SCP 24.5 (3.75) cm] suggests some homogeneity, differences in the distribution of ACSA along the length of the muscle (if present) could not be accounted for.

Muscle Architecture

At the point of peak PF torque during the MVIC trials, real‐time ultrasonography was used to record fascicle length and pennation angle during contraction synchronized with the measured PF torque values. The 5‐cm, 7.5‐Hz linear‐array probe was held on the midsagittal plane of the MG equidistant between the proximal and distal tendon insertions previously established by ultrasonography. In addition, the probe was held perpendicular to the surface of the skin to obtain several visible fasciculi ranging from the superficial to the deep aponeuroses. After the PF MVIC trials were completed, the recording of the highest torque trial was analyzed offline using ImageJ software. Fascicle length was measured as the length between the superficial and deep aponeuroses.34 Pennation angle was defined as the insertion angle of the fascicle into the deep aponeurosis.17 Fascicle length and pennation angle were measured at the time‐point of maximum PF torque, as it has been reported that pennation angle is underestimated and fascicle length is overestimated during rest conditions by 18.1° and 17.0 mm, respectively.34 Thus, to accurately calculate the intrinsic force‐generating capacity of the MG, data must be obtained during contraction, not during rest.9, 17, 35 The dimensions of the window used for analysis were 4.15 cm × 3.5 cm; in some cases, fascicle length was estimated using linear extrapolation if a whole image of the fascicle was not available for direct measurement.

PCSA

The PCSA was estimated as the ratio of MG muscle volume to fascicle length.17, 36

Moment Arm Length

The tendon excursion method was used to estimate moment arm length during a passive stretch trial on an isokinetic dynamometer by passively rotating the ankle to calculate tendon excursion with the subject in a seated position. The medial malleolus was visually aligned with the dynamometer's central axis of rotation. Before the experimental trial, end‐DF range of motion was identified by the experimenter by rotating the ankle at 1°/s, starting from 15° PF, until discomfort caused participants to cease the stretch in DF. This velocity was chosen in relation to previous findings which elicited minimal neural activity throughout passive stretch trials in individuals without neurological impairment.37, 38 During the passive stretch, B‐mode ultrasonography was used to determine the displacement of the MG MTJ throughout the passive stretch. MTJ displacement was measured relative to an acoustically reflective marker (a thin strip of Micropore tape) secured to the skin proximal to the MG MTJ.

The total change in MTJ displacement was divided by the change in ankle range of motion (rad), to predict the moment arm length for each individual. This technique has been validated previously using cadavers when assessing the moment arm length of the Achilles tendon.39 In vivo, the tendon excursion technique has shown high agreement with the center‐of‐rotation approach (R 2 = 0.76), but may underestimate by 2%–8% compared with the MRI‐based measures of the latter.40, 41 As previously mentioned, we observed no significant difference in passive DF end range of motion (ROM) between participant groups [control –8.40° (0.16°), paretic –5.30° (0.48°)26. Based on current measures of the Achilles tendon moment arm over the PF ROM,41 this 3° difference in ROM would be equivalent to an underestimation of the moment arm in SCP by 21 mm.

Achilles Tendon Force

Tendon force was calculated by dividing the net PF torque by the Achilles tendon moment arm length.9, 10

Fascicle Force

To estimate MG fascicle force, PF MVIC net torque was multiplied by the relative contribution of the MG PCSA in the triceps surae muscle group. The relative PCSA of the PF muscles has been used in previous work to determine the relative contribution of each muscle, whereby the relative PCSA of the MG was found to account for 15.4% of the Achilles tendon force.12 Therefore, the force generated by the MG was calculated by determining the ratio of MG contribution to Achilles tendon force. At present, there are no complete data on the relative PCSA in the triceps surae of adults with SCP; therefore, it is not possible to test the assumption that the MG contributes 15% of the Achilles tendon force. However, in the triceps surae of children with SCP, there is a degree of homogeneity to the relative atrophy of these muscles. Compared with age‐matched controls, the MG, soleus (SOL), and lateral gastrocnemius (LG) were found to be 42%, 39%, and 36% smaller, respectively.42 The calculation of specific force is therefore presented with the knowledge that, at least in terms of muscle ACSA, there seems to be some degree of similarity in the relative differences between SCP and controls in the triceps surae.

The force generated by the MG muscle was subsequently divided by the cosine of the pennation angle measured during contraction to determine MG fascicle force.

Specific Force

Specific force was calculated by dividing MG fascicle force by MG PCSA.

Statistics

All statistical analyses were performed using SPSS (version 19; SPSS, Inc., Chicago, Illinois). To ensure the data were parametric, Shapiro–Wilk and Levene tests were utilized to assess the distribution and variance of the data. As there were no breaches of these statistical assumptions, independent t‐tests were used to assess baseline anthropometric data between the SCP and control groups. To minimize type I error of the main outcome measures [as could occur with repeated analysis of variance (ANOVA) tests], a multivariate ANOVA was used to compare the differences and interactions in the joint torque, force, neural, and architectural variables (as listed in Tables 1, 2, 3) of the paretic vs. non‐paretic limbs vs. the dominant limb of control individuals. Statistical significance was accepted at the P < 0.05 level, and all data are presented as mean (SD).

Table 1

Joint torque, moment arm, and neural properties of the paretic and non‐paretic limbs of individuals with SCP and controls

	Paretic limb	Non‐paretic limb	Control group
PF MVIC (Nm)	102 (55.8)a, b	153 (47.7)	190 (26.7)
Net PF MVIC (Nm)	139 (59.5)a	160 (46.9)	198 (27.3)
DF MVIC (Nm)	17.5 (8.58)	21.3 (11.9)	20.8 (10.6)
Moment arm (cm)	6.05 (1.69)	5.08 (0.98)	5.54 (1.56)

Difference between paretic and control groups (P < 0.001).

Difference between paretic and non‐paretic groups (P = 0.039).

Difference between paretic and non‐paretic groups (P < 0.001).

Table 2

Muscle size and architectural characteristics of the MG muscle in the paretic and non‐paretic limbs of individuals with SCP and controls

	Paretic limb	Non‐paretic limb	Control group
Fascicle length (cm)	3.70 (0.62)a	3.14 (0.56)	2.89 (0.47)
Pennation angle (°)	25.7 (4.08)b, d	37.2 (7.59)	43.4 (7.00)
MG length (cm)	24.5 (3.75)	26.8 (3.23)	25.7 (2.00)
MG ACSA (cm2)	12.0 (2.62)c, e	15.0 (2.23)	16.5 (2.90)
MG volume (cm3)	195 (56)a, f	269 (62)	279 (52)
MG PCSA (cm2)	52.3 (11.6)b, g	89.0 (28.1)	98.8 (23.8)

Difference between paretic and control groups (P = 0.0004).

Difference between paretic and control groups (P < 0.001).

Difference between paretic and control groups (P = 0.001).

Difference between paretic and non‐paretic groups (P = 0.001).

Difference between paretic and non‐paretic groups (P = 0.028).

Difference between paretic and non‐paretic groups (P = 0.0005).

Difference between paretic and non‐paretic groups (P = 0.0001).

Table 3

Force measurements in the paretic and non‐paretic limbs of individuals with SCP and controls

	Paretic limb	Non‐paretic limb	Control group
Achilles tendon force (kN)	2.26 (0.57)a	3.34 (1.59)	3.81 (0.32)
MG muscle force (N)	347 (88.2)a	515 (244)	586 (161)
MG fascicle force (N)	388 (104)b, c	662 (317)	814 (205)
Specific force (N/cm2)	7.53 (1.84)	7.37 (2.08)	8.65 (2.99)

Difference between paretic and control groups (P = 0.010).

Difference between paretic and control groups (P < 0.001).

Difference between paretic and non‐paretic groups (P = 0.024).

RESULTS

Adults with SCP and matched control individuals were of similar age (P = 0.575), stature (P = 0.604), and body mass (P = 0.061).

Torque and Moment Arm Properties

The PF MVIC torque produced by the paretic limb was 33% less than that of the non‐paretic limb and 46% lower than the control group (Table 1). No difference in PF MVIC torque was identified between the non‐paretic limb and control groups (P = 0.178). During the PF MVIC trial, net torque from the paretic limb was 30% lower than that of control limbs (Table 1). However, no difference was identified between the paretic group and non‐paretic group net PF MVIC torque (P = 0.892), nor between the non‐paretic limb group and control group (P = 0.193). No differences were identified in DF MVIC torque (P = 0.653) or moment arm length (P = 0.281) between groups (Table 1).

MG Muscle Size and Architecture

There was no difference between the paretic and non‐paretic MG fascicle lengths (P = 0.070) and non‐paretic and control MG fascicle lengths (P = 0.929; Table 2). However, the paretic group fascicles were 28% longer than those of controls (Table 2). The pennation angle at which the fascicles joined the deep aponeurosis during PF MVIC in the paretic MG was 31% less than that of the non‐paretic limb and 41% smaller than that of the control group (Table 2). No difference in pennation angle was identified between the non‐paretic limb and control groups (P = 0.095).

The ACSA of the paretic MG was found to be 20% and 27% smaller than non‐paretic and control group MG, respectively (Table 2). No differences were identified between non‐paretic and control group MG ACSA (P = 0.601). Paretic group MG volume was 28% smaller than that of the non‐paretic group and 30% smaller than in the control group (Table 2). Similarly, the PCSA of the paretic MG was 41% and 47% smaller than the non‐paretic and control groups, respectively (Table 2). However, no difference was identified between non‐paretic limb and control groups when assessing MG volume (P = 0.574) and MG PCSA (P = 0.323). There was no difference between groups with regard to MG length (P = 0.095; Table 2).

Force Measurements

Achilles tendon force and MG force of the paretic limb was 41% lower than for control limb (Table 3). No difference in the non‐paretic Achilles tendon force and MG force was seen when compared with the paretic (both P = 0.100) and control group (both P = 1.000) limbs. Paretic limb MG fascicle force was 41% lower than that of the non‐paretic limb and 52% lower than in controls (Table 3). No difference between non‐paretic and control MG fascicle force was identified (P = 0.370). Last, there was no difference between groups with regard to MG specific force (P = 0.393; Table 3).

DISCUSSION

In this study we have assessed the specific force of the MG in active individuals with SCP. The purpose was to establish whether the specific force of the paretic MG and the variables used in its calculation differ when compared with the non‐paretic limb and control participants. Contrary to the hypothesis, our main finding was that there was no difference between the in vivo specific force of the paretic and non‐paretic MG of active individuals with SCP and muscle of control participants. Although specific force of the MG was the same across all groups, paretic group fascicle force was 41% and 52% lower than that of the non‐paretic and control groups, respectively.

Consistent with the results of our previous work,23 the SCP participants demonstrated significantly lower levels of activation than their unimpaired counterparts. The aim of the electrical stimulation in the present study was not to calculate the level of the activation deficit but to account for any neural contribution to weakness and allow for a more accurate measure of MG specific force. By accounting for the differences in MG activation and TA coactivation across the paretic, non‐paretic, and control groups, the net PF MVIC torque remained 30% lower in the paretic limb group when compared with the control group, but was not different from that of the non‐paretic limb group. Before this correction, the paretic PF MVIC torque was 46% and 33% lower compared with the control group and non‐paretic limb groups, respectively. As described in a previous study,23 the 3‐fold higher coactivation and 38% lower agonist activation in the CP group therefore contributed to about 16% of the difference in PF MVIC strength between CP and controls. The remaining 30% was attributable to morphological or architectural properties of the muscle.

The majority of research concerning muscle weakness in individuals with SCP measured the MVIC torque generated by a muscle or group of muscles.1, 4, 16, 43 A limitation when assessing torque is that moment arm lengths between limbs and/or groups of individuals are not taken into account. As individuals with SCP may have structural deformities in the paretic limb as a result of increased tone of the muscle throughout maturation,44 it is possible that the internal structures between the paretic, non‐paretic, and control limbs may be more prominent in adults, compared with pediatric populations. It has been established that the Achilles tendon moment arm increases in length with PF45 and, although previously hypothesized to be different in the ankle following gait kinematics,20 indirect measures at the wrist suggest some preservation of the moment arm in children with SCP.22 In our study, based on the consistent foot angle of 0°, we observed no significant difference in the Achilles tendon moment arm between the SCP and control groups. Nevertheless, when the in vivo forces were calculated in the paretic Achilles tendon and MG, they were found to be 41% weaker than those of the control group, but with no difference between the paretic and non‐paretic limbs. Although the difference in Achilles tendon moment arm observed in our participants was not significantly different, based on the measured values and the assumed underestimation (see Methods), the paretic Achilles tendon moment arm was between 0.51 and 0.72 cm larger than controls. With all else being equal when calculated using the mean PF MVIC net torque in this SCP population, reducing the moment arm length by 0.51–0.72 cm would theoretically increase the tendon force in the SCP limb by 212–120 N, or approximately 5%–10% of the measured value. As a result, assessment of specific force using joint torque rather than tendon force would overestimate the true force‐producing capacity of the contractile mass. Accounting for moment arm lengths, muscle architecture, and neural properties facilitates assessment of the intrinsic material force‐producing capacity of muscle in vivo.9, 35

In children with SCP, the morphology of the paretic MG during rest showed that fascicle length was 18% shorter compared with the non‐paretic contralateral limb3 and 16% shorter compared with age‐matched controls.2 Based on those findings, deficits in paretic fascicle length would imply that the number of sarcomeres in series is typically lower than in participants without SCP. However, in contrast to previous architectural data, we found that fascicle length during PF MVIC was 8% longer in the SCP group than in the control group. It is likely that the contrasting results we obtained reflect the nature of the measurement technique. Where previous studies have reported muscle architecture at rest, MG fascicle length in this study was measured at peak PF MVIC torque, as is consistent with the calculation of specific force.17 Due to the spasticity in the MG muscle, the paretic foot of participants with SCP is typically in an equinus position. Where previous studies have reported shortened fascicles in SCP, it is likely that this was due to them being measured while in a more plantarflexed or “relaxed” position.2 In addition, as we measured MG fascicle length during PF MVIC, any difference in the tendon properties or force produced would influence the relative shortening experienced by individuals with and without SCP, as has been observed in the elderly.35 At present, however, tendon stiffness comparisons between those with and without SCP have shown no difference in Achilles tendon strain during MVC.16 It should be noted that, in their comparisons, Barber et al.,16 conducted tendon measurements at maximal dorsiflexion (–6° in SCP, –21° in controls) in men and women. Therefore, direct comparisons regarding fascicle shortening may not be possible with the men in our study who performed PF MVC at the 0° ankle angle and who had no difference in maximal dorsiflexion angles.

When neural and architectural factors were accounted for in this study, there was a difference in MG fascicle force between SCP and control participants that was almost entirely accounted for by the difference in PCSA, as evidenced by similar values for specific force between SCP and controls. As PCSA takes into account the volume and fascicle length of the muscle, it provides a more accurate measurement of the true contractile area of pennate muscle.11 We found that paretic MG PCSA was ∼45% smaller than non‐paretic limb and control, similar to the observed difference in MG fascicle force (52% smaller). Such differences indicate that the paretic MG muscle has fewer sarcomeres in parallel compared with the non‐paretic and control MG muscle, and any weakness at the whole muscle level is unlikely to be influenced by a decrease in the quality of the muscle at the fascicle level. No difference in specific force was observed between the SCP and control groups or between the paretic and non‐paretic limbs of SCP participants. This would initially appear to be in contrast to previous work, which established that the size/strength relationship of muscles in individuals with SCP is reduced compared to individuals without neurological impairment (e.g., PF MVIC/ACSA torque1). However, as previously stated, based on the architectural differences between the SCP and control groups, and the substantial neural contribution to reduced joint torque, the estimation of the size/strength relationship (e.g., MVIC/ACSA) may be erroneous unless these factors are considered. Indeed, where PCSA and coactivation have previously been included, the size/strength relationship appears to be no different between SCP and controls.16 It is pertinent to consider, however, that, although lower size/strength indices have previously been reported in SCP even though neural factors are likely to contribute, there is an element that may be attributed to alterations in the collagen content of the extracellular matrix.46 It is possible that inclusion of an elevated collagen content (or other non‐contractile elements) in the calculation of muscle size would result in a diminished size/strength relationship (e.g., MVC/ACSA1). This is particularly relevant, as the participants in the study by Elder et al.1 were likely more impaired than the participants in our study (most of those in the former study had undergone surgical procedures to the Achilles tendon, compared with none in our study), and it has been shown that contracture severity is linked to collagen content.46 This is also reflected in work by Barber et al.,16 who showed no difference in PF MVC/PCSA in young adults with SCP classified on the MGFSC as level 1, or least impaired. At least in our study and that of Barber et al., where no differences in specific force were observed, it is consistent with the direct observations that histological structure is similar between SCP and control muscle.47 Indeed, despite the fact that collagen content has been reported to be elevated in SCP muscle, there is no evidence of its mislocalization within the muscle.46

Although consistent with the observations of others,16 several factors may potentially impact the our calculations of in vivo specific force. In our study, moment arm was estimated during rest, whereas specific force should represent the data obtained during MVIC. The Achilles moment arm length during MVIC is approximately 1.5 cm longer than resting measures in individuals without neurological impairment.45 In addition, fascicle length was measured at a comparable length rather than the angle at which peak torque occurred,35 due to the fact that we used a 0° ankle angle to control for the different resting angles between participants. However, as the plantarflexors are on the ascending limb of the force‐length relationship, it is likely that specific force is underestimated.35 To compare between individuals who have limited dorsiflexion range of motion, a consistent joint angle was chosen for measurement of specific force at 0°. As previously mentioned, compared with SCP, Achilles tendon strain with PF MVIC is no different when compared with controls.16 Nevertheless, the interaction of the muscle and the tendon has yet to be addressed in adult men with SCP at a matched ankle angle, consistent with the measurement of specific force.

In conclusion, in this study we have shown that the paretic MG of physically active individuals with SCP has a similar specific force‐generating capacity to the non‐paretic muscle and the MG of control individuals. Our study has also demonstrated how the pennation angle and fascicle length of paretic muscle at the MVIC is different from the control. Nevertheless, the weakness (while accounting for neural properties, moment arm lengths, and muscle architecture) observed in the paretic MG can be attributed primarily to a smaller PCSA rather than to the intrinsic material properties at the fascicle level.