The effects of long‐term muscle disuse on neuromuscular function in unilateral transtibial amputees

What is the central question of this study? The effects of long‐term muscle disuse on neuromuscular function are unclear because disuse studies are typically short term. In this study, we used a new model (unilateral transtibial amputees) to investigate the effects of long‐term disuse on quadriceps neuromuscular function. What is the main finding and its importance? Kinetic analysis (knee‐extension moments during gait) indicated habitual disuse of the amputated limb quadriceps, accompanied by lower quadriceps muscle strength (60–76%) and neural activation (32–44%), slower contractile properties and altered muscle architecture in the amputated limb, which could not be predicted from short‐term disuse studies.

a default position for many clinical populations (Brown, Friedkin, & Inouye, 2004) and the sedentary, but it is unclear how both MVT and RTD change with long-term, habitual disuse, because typical disuse study models last <90-120 days for logistical and ethical reasons.
Individuals with unilateral transtibial amputations (ITTAs; belowknee amputation on one limb) might provide a useful model for studying the effects of long-term, habitual disuse. These individuals adopt an asymmetrical loading pattern characterized by considerably lower vertical ground reaction forces and knee-extensor moments on the amputated compared with the intact limb, during movements such as walking (Fey & Neptune, 2012), jumping (Schoeman, Diss, & Strike, 2012) and stair ascent/descent (Schmalz, Blumentritt, & Marx, 2007). This suggests that the quadriceps of the amputated limb in ITTAs are chronically disused, which would explain observations of considerably lower (∼50%) quadriceps size (Moirenfeld, Ayalon, Ben-Sira, & Isakov, 2000) in the amputated compared with the intact limb.
Further, studies comparing strength in ITTAs and healthy controls have found significantly reduced amputated limb (∼50%) MVT (Isakov, Burger, Gregoric, & Marincek, 1996;Lloyd, Stanhope, Davis, & Royer, 2010;Pedrinelli, Saito, Coelho, Fontes, & Guarniero, 2002) compared to the intact and control limbs. Comparison of quadriceps neuromuscular function in the amputated versus intact limb of ITTAs, coupled with comparison of limb loading during gait as an estimation of typical use, might therefore offer new insight into the long-term effects of habitual disuse. However, it is currently unclear whether the intact limb provides an internal control that is unaffected by the amputation and comparable to the limb of an able-bodied control subject, which would support the efficacy of ITTAs as a study model of long-term disuse. Despite similar peak vertical ground reaction forces and knee moments during walking gait (Lloyd et al., 2010;Nolan et al., 2003;Sanderson & Martin 1997), previous studies in ITTAs have shown lower MVT in the intact limb compared with the limbs of able-bodied participants (Isakov et al., 1996;Lloyd et al., 2010;Pedrinelli et al., 2002;Powers, Boyd, Fontaine, & Perry, 1996). However, the latter studies did not control between the groups for other factors known to have independent effects on muscle strength, such as ageing, health and sedentary lifestyle (Narici & de Boer 2011;Sacchetti et al., 2013).

New Findings
• What is the central question of this study?
The effects of long-term muscle disuse on neuromuscular function are unclear because disuse studies are typically short term. In this study, we used a new model (unilateral transtibial amputees) to investigate the effects of longterm disuse on quadriceps neuromuscular function.
• What is the main finding and its importance?
Kinetic analysis (knee-extension moments during gait) indicated habitual disuse of the amputated limb quadriceps, accompanied by lower quadriceps muscle strength (60-76%) and neural activation (32-44%), slower contractile properties and altered muscle architecture in the amputated limb, which could not be predicted from short-term disuse studies.
of the muscle, such as the RTD relative to peak torque recorded during electrically evoked involuntary contractions (e.g. twitch or octets; Folland, Buckthorpe & Hannah, 2014), and short-term disuse causes a shift towards faster contractile properties (Lambertz et al., 2001). Finally, the maximal force-generating potential of a muscle is dependent upon its architecture (Blazevich, Cannavan, Horne, Coleman, & Aagaard, 2009), and 21-30 days of disuse have elicited changes such as declines in muscle size (≤10%), pennation angle (≤13%) and fascicle length (≤9%; Campbell et al., 2013;de Boer et al., 2007). Determining the degree of change in these neuromuscular determinants of muscle strength with long-term habitual disuse might allow better targeting of preventative and rehabilitative interventions for populations subject to muscle disuse.
The first aim of the present study was to assess the efficacy of unilateral ITTAs as a model to study long-term habitual disuse, by comparing knee-extensor strength (MVT and RTD) and loading (kneeextensor moments and impulse) during walking gait of the intact limb with a control able-bodied population, where both groups are healthy, young and active. The second aim was to assess MVT and RTD, and the neuromuscular determinants of these (neural drive, intrinsic contractile properties and vastus lateralis muscle architecture), in the disused quadriceps muscles of ITTAs, in comparison to both the intact limb and an able-bodied control limb.

Ethical approval
Participants provided written informed consent before their involvement in the study, which complied with the standards set by the 2013 Declaration of Helsinki (except for registration in a database) and was approved by the University of Roehampton Ethics Committee (LSC 16/176) and the National Health Service Health Research Authority (17/NW/0566).

Participants
Nine male ITTAs and nine male control subjects took part. Before data analysis, groups were matched to ensure similar group means and variability in age, height, body mass (BM) and physical activity.
Physical activity was assessed using the International Physical Activity

Overview
Participants visited the laboratory on three separate occasions, with each visit 3-7 days apart, to complete a familiarization (visit 1; identical to visit 2), neuromuscular function assessment of the quadriceps muscles of both limbs (visit 2) and a gait assessment (visit 3 filtered using a fourth-order low-pass Butterworth filter with a cut-off frequency of 10 Hz and corrected for the passive weight of the limb.

Electromyography
Electromyography signals were recorded from the superficial knee  able-bodied control subjects using the same methods as described above. The coefficient of variation was 4.4, 10.9 and 9.3% for muscle thickness, fascicle length and pennation angle, respectively.

Electrical stimulation
Square-wave (0.2 ms duration) electrical impulses were delivered percutaneously to the femoral nerve, via a constant-current, variablevoltage stimulator (model DS7AH; Digitimer, Ltd, Welwyn Garden F I G U R E 1 Static B-mode ultrasound image of the vastus lateralis (VL) and vastus intermedius (VI) muscles at rest for the amputated (AMP) and intact (INT) limb of one individual with unilateral transtibial amputation and one control limb (CON). Architectural measures taken included the pennation angle (ϴ) relative to the deep aponeurosis and extrapolated fascicle length, which were each determined from three fascicles; and muscle thickness, measured between the superficial and deep aponeuroses at three separate points (the centre and either end of each image, as indicated by numbered circles in the middle image). A significant reduction in the amputated limb VL muscle thickness is evident, and similarities in pennation angle in all three limbs, and in muscle thickness between INT and CON, can be seen clearly The mean M-wave peak-to-peak amplitude of the three supramaximal twitch contractions was defined as the maximal M-wave (M max ) for each muscle. Torque measurements from the evoked contractions were twitch and octet peak torque (PT) and peak RTD (calculated using a 15 ms moving time window), presented as absolute and relative to PT. These variables were averaged across the three supramaximal twitch and octet contractions recorded.

Knee-extension MVCs
Participants performed a series of ∼20 warm-up contractions of 3 s duration at progressively higher intensities before completing six

Voluntary activation
The second, fourth and sixth MVCs had a single doublet superimposed at the plateau of the torque-time curve, and two further doublets evoked at rest immediately after the MVC. The difference between superimposed and resting potentiated doublet torque was used to determine the voluntary activation (VA; a measure of neural drive at MVT), using eqn (1): where D s and D c are the superimposed and control doublets, respectively.
The root mean square (RMS) of the EMG signal for each quadriceps muscle was calculated over the 500 ms window centred on or nearest to MVT, which was not influenced by the stimulation artefact (EMG MVT ). The EMG MVT was normalized to M max of the same muscle and averaged across the three quadriceps muscles. Time since amputation (years) 12.2 ± 11.5 1.5-29.0 --Data are mean values ± SD and are presented for n = 9 subjects for both groups, except for walking speed [n = 8 for individuals with unilateral transtibial amputations (ITTAs) and n = 9 for control subjects]. The cause of amputation was trauma for all ITTAs.

Walking gait
Kinematic data were collected using 12 Vicon Vantage V5 ( Standard inverse dynamics techniques were used to calculate net internal joint moments, normalized by body mass (Winter & Sienko, 1988).
Internal peak knee-extension moment and total impulse (calculated as the integral of internal knee-extension moment with respect to time) for the entire stance phase were extracted for each limb and averaged across the trials selected for analysis.

Statistical analysis
Student's paired t tests revealed no differences in either MVT or peak

RESULTS
Owing to an injury to one ITTA occurring between visits 2 and 3 (neuromuscular and gait assessment), data are for nine and eight ITTAs, respectively. One control subject withdrew from octet and doublet stimulation; therefore, control data for VA and octet variables are presented for eight control subjects, but all other variables are for nine control subjects. The groups had similar age, height, body mass and physical activity scores (P ≥ 0.354; g = 0.10-0.64; Table 1). There was a large effect size for the control subjects to walk faster (g = 1.21), although this difference was not statistically significant (P = 0.616; Table 1).

Contractile properties
The PT and absolute RTD in both the twitch and the octet (Table 2) were lower in AMP than in both INT and CON (−72 to −50%, TA B L E 2 Knee-extensor kinetics and neuromuscular determinants of strength in the amputated (AMP) and intact (INT) limbs of unilateral transtibial amputees and in an able-bodied control limb (CON) Data are presented as the mean value ± SD for n = 9 (AMP and INT) and n = 9 (CON). Data in italics correspond to those variables where n = 8 owing to participant withdrawal. Differences compared with AMP are denoted by * (P < 0.05) or ** (P < 0.001). Abbreviations: subscript BM, relative to body mass; M max , maximal M-wave; MVT, relative to maximal voluntary torque; PT, peak torque; RMS EMG MVT , root mean squared electromyography at MVT; RMS EMG 0-100 , root mean squared electromyography from 0 to 100 ms of an explosive voluntary contraction; RTD, rate of torque development. P = 0.001-0.004, g = 1.97-2.84), but similar between INT and CON (P ≥ 0.284, g = 0.40-0.68).
When expressed relative to PT, twitch RTD was 18% lower (P = 0.006, g = 1.35) and octet RTD 25% lower (P < 0.001, g = 2.60) in AMP when compared with INT (Table 2). Relative twitch and octet RTD were also both 14% lower in AMP compared with CON (twitch RTD, P = 0.036, g = 1.59; and octet RTD, P = 0.037, g = 1.63). Despite being statistically similar, there was a large effect for relative octet RTD to be greater in INT than in CON (P = 0.120, g = 1.03; Table 2), whilst relative twitch RTD was similar between INT and CON (P = 1.000, g = 0.18).

Neural drive
Both VA and RMS EMG MVT (Table 2)

DISCUSSION
In this study, we compared quadriceps strength and neuromuscular function in the amputated limb of ITTAs with their intact limb and a control group limb, providing a new model for studying the longterm (>1.5 years) effects of chronic disuse. Long-term disuse of the amputated limb in ITTAs was evidenced from the ∼60% lower peak knee-extensor moments during walking compared with the intact and control limbs. This disuse was accompanied by ∼60% lower MVT and ∼75% lower RTD in the amputated limb, which are much greater differences than might be predicted from short-term disuse studies.
Declines in MVT appeared to be largely attributable to reduced muscle size (evidenced by lower muscle thickness in AMP) and neural drive (evidenced by lower VA and EMG MVT in AMP). Declines in RTD appeared to be attributable primarily to declines in MVT and a shift towards slower intrinsic contractile properties, with neural drive in explosive contractions being unaffected in AMP.

Transtibial amputees as a model for long-term disuse
In the present study, there were large effects for knee-extensor kinetics during gait to be lower in amputated than intact or control limbs which, coupled with the considerable reductions in knee-extensor strength in the amputated limb, suggests that the amputated limb undergoes substantially less habitual loading during ambulation. The reduced knee-extensor moments in gait may also be attributable, in part, to increased co-contraction at the knee of the amputated limb during gait (Culham et al., 1986;Isakov et al., 2001).
Future research should therefore aim to quantify internal loading of the knee extensors for a more direct estimation of disuse and its association with changes in strength in the amputated limb. Consistent with our results, previous studies have reported decreased knee moments (Powers, Rao, & Perry, 1998;Winter & Sienko, 1988), power (Powers et al., 1998;Winter & Sienko, 1988) and work (Silverman & Neptune, 2012) in the amputated limb in walking. In contrast to these previous studies, however, the ITTAs of the present study were young, healthy and moderately to highly active. As a result, the effects of the evident disuse on strength and neuromuscular function could be isolated from factors such as ageing, disease and sedentary behaviour, which are known independently to affect muscle strength and function (Narici & de Boer 2011;Sacchetti et al., 2013).
The knee extensors of the intact limb in the ITTAs did not differ from those of an able-bodied control population for kinetics during walking, MVT, RTD or any of the neuromuscular determinants of strength.
This suggests that, for these parameters, the intact limb of the ITTAs provides an ideal internal control for comparison with the amputated limb, from which to draw conclusions about the effects of chronic disuse.

Changes in strength
The declines in MVT found in the amputated limb compared with the intact limb (−59%) are comparable to, albeit at the high end of, differences observed in previous studies of amputees (−33 to −57%;Isakov et al., 1996;Lloyd et al., 2010;Moirenfeld et al., 2000;Pedrinelli et al., 2002), but considerably greater than the reduction in strength typically observed after a period of short-term disuse of up to 120 days (∼23%; Narici & de Boer, 2011 (Bamman et al., 1998) and42% (de Boer et al., 2007) decreases in RTD after 16 days of bed rest and 23 days of unilateral lower-limb suspension, respectively. The considerable reductions in peak RTD (−75%) in the amputated versus intact limb are important, because RTD is considered more functionally relevant than MVT in many sportsspecific and daily tasks, such as sprinting, jumping and recovery of balance (Behan et al., 2018;Pijnappels et al., 2008;Tillin, Pain, & Folland, 2013).
When expressed relative to MVT, peak RTD was significantly reduced in the amputated compared with the non-amputated limbs, although the differences between limbs were considerably smaller for relative than absolute peak RTD. Thus, the reduction in MVT appears to be a large contributing factor to reduced absolute RTD in the amputated limb; however, this only contributed to the reduction in peak RTD, which was probably also influenced by the slowing of the contractile properties (discussed in more detail below, Section 4.3.3).

Neural drive
A broad suppression in neuromuscular activity at maximal force production, indicated by a reduction in VA (∼44%) and EMG MVT (∼38%) in the amputated compared with the non-amputated limb, is likely to contribute to the reduction in amputated limb MVT. Previous studies have reported reduced quadriceps EMG amplitude (−16 to −35%; Deschenes et al., 2002) and VA (−7%; Kawakami et al., 2001), whereas others have not observed changes in these measurements (Campbell et al., 2013;de Boer et al., 2007;Horstman et al., 2012), after periods of disuse of ≤89 days. Thus, the large limb effects on VA and EMG responses observed in the present study suggest that reductions in neural drive with disuse become more pronounced and observable over time. Of note is the specificity of the neural deficits in the ITTAs to the amputated limb. Evidence from unilateral injury and training studies suggests a crossover effect of neural function, in that neural drive adaptations occur in the contralateral limb in addition to the injured/trained limb (Bogdanis et al., 2019;Tillin, Pain & Folland, 2011). In the present study, however, there was no evidence that reduced neural drive on the amputated side had affected neural drive on the intact side, which was similar to the control limb. Perhaps this is because ITTAs rely more on the intact limb for most activities of daily living and exercise (e.g. Winter & Sienko 1988), which might negate any crossover effects of reduced neural drive from the amputated to the intact limb.
Despite the substantial differences between the amputated and non-amputated limbs evident in neural drive during maximal force production, no such differences were observed in the present study in explosive-phase EMG amplitude (Table 2). This suggests that altered neural drive does not explain the lower peak RTD in the amputated limb, which is interesting given that neural drive is a key determinant of RTD (Folland et al., 2014;Vecchio et al., 2019). The large variability in EMG, even after normalization to M max (Buckthorpe, Hannah, Pain, & Folland, 2012), greater variability in RTD compared with MVT (Folland et al., 2014;Tillin et al., 2013) and small sample sizes (n = 9 per limb) might have reduced the chances of observing a significant effect. Alternatively, the role of the amputated limb in ambulation might explain the lack of differences in neural drive during the explosive contractions. Specifically, although the knee extensors of the amputated limb experience reduced load compared with the intact side during ambulation, the amputated side does contribute to stability and postural correction, for which RTD appears to be important (Behan et al., 2018). Thus, typical physical activity in the amputees might provide sufficient stimulus to maintain the neural drive during short, rapid contractions, which typically underpins RTD.

Muscle architecture
The VL muscle was 41% thinner in the amputated limb compared with the intact side, which is a larger difference than the declines in magnetic resonance imaging and computed tomography scanner measurements of muscle size (−3 to −18%) observed in short-term disuse studies Campbell et al., 2013;de Boer et al., 2007;Dirks et al., 2016). Thus, similar to the changes observed for strength and neural drive, reductions in muscle size with longterm disuse are much greater than could be predicted from short-term disuse studies. Muscle size is considered an important determinant of MVT (Blazevich et al., 2009), and thus the reduction in muscle thickness is likely to contribute to the declines in both MVT and, by association, RTD, in the amputated limb.
Fascicle length was reduced by 36% in the amputated limb compared with the intact limb. Again, this difference is considerably greater than the decline in knee-extensor fascicle length (6-9%) typically observed with short-term unloading (Campbell et al., 2013;de Boer et al., 2007). Individuals with unilateral transtibial amputations walk with a comparatively stiff knee joint on the amputated limb (Powers et al., 1998;Winter & Sienko 1988), which would theoretically isolate loading to shorter fascicle lengths and limit the stimulus likely required to maintain longer fascicle lengths. Decreases in fascicle length might reduce maximal shortening velocities and power (Blazevich & Sharp 2005) and shift the torque-angle relationship towards more extended knee positions (Blazevich et al., 2009). Given that our strength measurements were made at a typical plateau region of the torque-angle relationship (Chow, Darling, & Ehrhardt, 1999), a shift away from this region in the amputated limb might have contributed to the observed differences in MVT and RTD.
In contrast to the results of previous research, which demonstrated decreases in pennation angle during short periods of unilateral lowerlimb suspension (Campbell et al., 2013;de Boer et al., 2007), our results appear to suggest that pennation angle does not change with long-term disuse. In healthy populations, angles of pennation of the VL muscle have been reported to be 6-27 deg (Blazevich, Gill, & Zhou, 2006;Rutherford & Jones 1992); the pennation angle of all three groups of limbs in the present study (∼12-14 deg) falls within this range. This suggests that the structural remodelling that seems to take place in the early phases of disuse is not representative of long-term adaptations. It is possible that muscle thickness declines at a faster rate than fascicle length with short-term disuse, causing a decline in pennation angle, whereas over longer periods of disuse, reductions in fascicle length 'catch up' with muscle thickness loss, causing a return to baseline pennation angle, but this hypothesis cannot be tested with our data.

Intrinsic contractile properties
The significant reductions in evoked (twitch and octet) contractile peak torque in the amputated compared with the intact and control limbs ( Table 2) are reflective of the reduced capacity of the amputated limb knee extensors for torque production. These changes were accompanied by reductions in RTD, both absolute and relative to peak torque, reflecting a shift towards slower contractile properties in the intact limb. This is in contrast to the results of short-term disuse studies in both healthy control subjects and pathological populations, which have reported a shift towards faster contractile properties owing to a greater expression of fast-contracting myosin heavy-chain (MHC) isoforms (Bamman et al., 1998;Kapchinsky et al., 2018;Trappe et al., 2004). The results of the present study therefore provide new evidence that changes in intrinsic contractile properties with longterm disuse are more characteristic of ageing muscle, which also displays a slowing of the contractile properties (Roos, Rice, Connelly, & Vandervoort, 1999). This slowing might be attributable to preferential atrophy of type II muscle fibres and, potentially, also to an increased dominance of type I MHC in fibres co-expressing MHCs commonly seen in old age (Lexell, Taylor, & Sjöström, 1988). The slower contractile properties in the amputated limb probably contributed to the lower voluntary peak RTD also observed in the amputated limb, because twitch and octet RTD are important determinants of voluntary RTD (Folland et al., 2014).

Conclusion
This study was the first to use ITTAs as a study model to investigate the effects of long-term muscle disuse on strength and neuromuscular function, in young, healthy, active adults. Strength, neuromuscular function and loading during gait of the intact limb of ITTAs were comparable to those of a control able-bodied limb, suggesting that the intact limb provides a suitable internal control for comparison with the amputated limb for these parameters. The quadriceps muscles of the amputated limb displayed considerably less habitual loading during gait than the intact side. This disuse of the amputated limb was accompanied by larger reductions in MVT and RTD than could be predicted from short-term disuse studies. The reductions in MVT were probably attributable to the declines in muscle size and neural drive, whereas the reductions in RTD appeared to be attributable to the decline in MVT coupled with a slowing of the contractile properties.