Introduction
Association football (soccer) is a team based, high-intensity, intermittent-sprint sport typically played over 90 min. However, in certain knockout tournament scenarios (e.g., FIFA World Cup or UEFA Champions League) when a match is tied at 90 min, but requires an outright winner, an additional 30 min period of play termed extra time (ET) is required. Recently, negative impacts of this additional period of play have been shown on technical (Harper et al.
2014) and physical (Penas et al.
2015; Russell et al.
2015) performance, as well as aspects of metabolism and hydration status (Harper et al.
2016a,
b,
d). These negative consequences are concurrent with the greatest occurrence of contact related injuries during this time (Aoki et al.
2012). Participation in soccer results in high levels of metabolic (Rampinini et al.
2011), mechanical (Akenhead et al.
2013), and perceptual stress (Impellizzeri et al.
2004). The aetiology of soccer-specific fatigue, which manifests transiently during simulated and actual match-play, has been hypothesised to be due to several putative mechanisms including, compromised excitation–contraction coupling (Clarke et al.
2015; Rampinini et al.
2011), depletion of endogenous fuel sources (Bendiksen et al.
2012), ionic disturbances (Bangsbo et al.
2006), and dehydration (Laitano et al.
2014). Despite these investigations, the precise mechanisms of fatigue are yet to be delineated.
Fatigue in soccer has been the subject of several reviews (Bangsbo et al.
2007; Mohr et al.
2005; Nedelec et al.
2012) and experimental study (Andersson et al.
2008; Oliver et al.
2008; Rahnama et al.
2006; Robineau et al.
2012), however, a limited number of investigations have attempted to quantify the neuromuscular fatigue response, with equivocal results (Girard et al.
2015; Marshall et al.
2014; Nybo et al.
2013; Rampinini et al.
2011). Fatigue is classically defined as an exercise-induced reduction in the ability of a muscle or muscle group to generate maximal force (Gandevia
2001), which stems from peripheral and central mechanisms. Peripheral fatigue is the loss in muscle force caused by disturbances in sites at or distal to the neuromuscular junction, whereas central fatigue is defined as a progressive, exercised-induced reduction in the voluntary activation (VA) of muscle (Gandevia
2001). Simulated and actual soccer match-play has been shown to elicit substantial peripheral fatigue (Clarke et al.
2015; Girard et al.
2015; Rampinini et al.
2011) likely attributable to alterations in excitation–contraction coupling. Soccer match-play also results in significant central fatigue; a reduced VA of the knee-extensors (~ 8%) following 90 min of football match-play was first reported by Rampinini et al. (
2011). Smaller reductions of ~ 1.5% in VA of the plantar flexors have also been reported following 90 min matches in hot (43 °C) and temperate (~ 20 to 21 °C) environments (Girard et al.
2015; Nybo et al.
2013). However, in these investigations, the post-match fatigue assessments were recorded 30–40 min following the match, a time in which the degree of fatigue would have dissipated. Transcranial magnetic stimulation (TMS) can be used to stimulate neural structures (such as the primary motor cortex) to further investigate the central nervous system responses to exercise, and the presence of a supraspinal contribution to central fatigue (Goodall et al.
2014). Of relevance to soccer, TMS has recently been used to demonstrate how maximal repeated-sprint running exercise elicits central fatigue that is partly attributable to sub-optimal output from the motor cortex (Goodall et al.
2015b). Although these data provide some indication of the responses to repeated-sprint activity, akin to soccer, the use of TMS to examine the pattern of fatigue during soccer-specific exercise has not been investigated. Further research is required to elucidate the aetiology of fatigue during soccer, both during regulation 90 min games, and for tournament scenarios where ET periods are common. The potential accumulation of fatigue incurred by ET might explain the previously observed performance reductions (Harper et al.
2014) and increased injury incidence (Aoki et al.
2012). Furthermore, practitioners working in professional soccer have recently highlighted that understanding fatigue responses following ET performance is an important area for future research (Harper et al.
2016c).
While there is value in studying the mechanisms of neuromuscular fatigue, the usefulness of such study is dependent on the data demonstrating acceptable reliability. Reliability refers to measurement stability when a testing protocol is undertaken repeatedly (Hopkins
2000). Knowledge of measurement reliability for neuromuscular responses over time is important as these data are rarely provided. Accordingly, the primary aim of this study was to investigate neuromuscular fatigue in response to 120 min of simulated soccer-specific exercise. A secondary aim was to investigate the reliability of the fatigue response.
Discussion
The primary aim of this study was to investigate the development of neuromuscular fatigue during a 120 min soccer match simulation. Our data demonstrate that 90 min of simulated soccer elicits reductions in the force generating capabilities of the knee-extensors, and this fatigue is a combination of both central and peripheral factors. An additional 30 min period of extra-time induced further fatigue that was primarily of central origin. A secondary aim of the study was to assess the consistency of fatigue development on repeat trials of the 120 min SMS. The development of fatigue was reliable across the two trials with the most variable responses noted following the ET period. Collectively, these data are the first to profile the neuromuscular fatigue response to 120 min of soccer-specific exercise and can help to explain the previously reported reductions in technical performance and physical performance that have been shown to occur during this extended period.
The development of fatigue throughout 120 min of soccer simulation was progressive, with decrements in the ability to generate maximum force evident at successive time-points. Knee-extensor MVC, decrements in which are considered as a global measure of fatigue involving peripheral and central components, was reduced by 11% after 45 min of the simulated match. After 90 min, the ability to generate maximal force was further reduced, and this reduction in strength was similar to the results of simulated and actual intermittent exercise performance (~15%; Clarke et al.
2015; Robineau et al.
2012), but larger than others (Andersson et al.
2008; Ascensao et al.
2008; Ispirlidis et al.
2008; Rampinini et al.
2011; Thorlund et al.
2009). Extra time elicited further reductions in MVC compared to FT (Fig.
2a), a finding which might offer some insight as to why technical performance and injury risk are also known to be affected during this period (Aoki et al.
2012; Harper et al.
2014). In a separate investigation, the loss in maximal force generating capacity of the knee-extensors following the performance of a simulated protocol was not recovered 72 h post-exercise (Thomas et al.
2017). Specifically, the MVC reduction in that study at FT was 16%, similar to that of this study (20%), but following ET, this reduction was further exacerbated (27%). Thus, the fatigue observed following the ET protocol is likely to have persisted for several days’ post-exercise.
The impairment in maximal force production was accompanied by reductions in the
Q
tw,pot, demonstrative of a contribution from peripheral mechanisms of fatigue (Fig.
2b). The
Q
tw,pot was reduced from baseline by 15% at HT, and thereafter, no further reduction was observed at FT or ET demonstrating a plateau in the peripheral fatigue response. Such a plateauing of the peripheral fatigue response has previously been demonstrated following self-paced isokinetic exercise (Froyd et al.
2013), intermittent high-intensity cycling (Decorte et al.
2012), and repeated-sprint exercise (Goodall et al.
2015b; Hureau et al.
2014). In line with this study, these previous investigations show a similar biphasic pattern of peripheral fatigue development, whereby most of the decrements in muscle function are manifest early in the exercise bout and are then small thereafter. Such a regulated development of peripheral fatigue can be explained by the recently proposed model based on task-dependency (Thomas et al.
2016). During the first half of the SMS, participants would have met the exercise demand by preferentially exhausting the higher threshold motor units, which are most susceptible to fatigue and change in response to peripheral stimulation. The remaining, smaller degree of fatigue observed at FT and ET was likely attributable to change in the more fatigue-resistant motor units, which exert a smaller reduction in the peripheral twitch but also reduce physical performance (Harper et al.
2016a,
d). As with many investigations, the unchanged
M
max values (Table
1) throughout exercise suggest maintenance of sarcolemmal excitability and a preserved neuromuscular propagation of the action potential. Thus, the peripheral fatigue elicited by simulated soccer performance was likely related to disturbances in the process of excitation–contraction coupling. Specifically, impairments to intracellular Ca
2+ regulation in the sarcoplasmic reticulum might reduce Ca
2+ sensitivity, leading to a reduction in mechanical output and such muscle fatigue (MacIntosh et al.
2012).
A significant development of central fatigue was also observed, voluntary activation measured with motor nerve stimulation was reduced from baseline throughout the protocol, confirming the previous work showing competitive soccer match-play elicits central fatigue (Rampinini et al.
2011). Following 90 min of match-play reductions in VA of <2% (Girard et al.
2015; Nybo et al.
2013) and ~8% (Rampinini et al.
2011) have been previously reported which is less than observed in this study (~16%, Fig.
2c). A likely explanation for the lack of fatigue in these aforementioned studies is partly due to the investigation of different muscle groups, and the timing of post-exercise measures which might allow some aspects of central fatigue to dissipate (Taylor et al.
1996). In this study, we also quantified VA using TMS of the motor cortex; reductions in VA measured with TMS indicate that some of the observed central fatigue is attributable to supraspinal factors (Gandevia
2001). There was a significant reduction in VA measured with TMS, indicating a reduced capacity for the motor cortex to drive the knee-extensors during, and immediately following 120 min of soccer-specific exercise (Fig.
2c). Over the 120 min simulated soccer match, central fatigue tended to be exacerbated, and this duration-dependent contribution of central processes to fatigue is broadly evident across a range of exercise modes (Lepers et al.
2002; Place et al.
2004; Thomas et al.
2015). In this study, there was a pattern of a progressive decrease in voluntary activation across 120 min of the SMS (pre vs. HT; HT vs. ET), which provides further evidence that central fatigue becomes progressively more limiting as the exercise duration extends.
It is, perhaps, surprising that the ability to produce maximal knee-extensor force dropped following the period of ET compared to FT, but the period of ET did not induce any additional reductions in the
Q
tw,pot or voluntary activation. The effect sizes for the change in both measurements of VA, and VA
TMS, between FT and ET were small (
d = 0.24 and 0.21), whereas the
Q
tw,pot showed no effect at all (
d = 0.01). Thus, we consider it likely that the additional reductions in MVC following ET are related to central fatigue, which were not detectable by the measurement tools of the study. Taken together, these data support the previous conclusions regarding central fatigue and soccer performance (Rampinini et al.
2011) and, in part, can offer an explanation for the reduced technical and physical performance (Harper et al.
2016b,
2014), and increased risk of injury (Aoki et al.
2012), known to occur during ET.
To substantiate the neuromuscular fatigue responses observed in this study, it is necessary to evaluate the magnitude of change against the reliability of the measurements. Due to there being no preferred, or single statistical approach, the evaluation of measurement reliability is somewhat problematic (Hopkins
2000). In this regard, and in line with other investigations evaluating reliability of responses from the knee-extensors (Bachasson et al.
2013; Rainoldi et al.
2001), we used two approaches to evaluate measurement reliability, the CV and ICC, which provide an absolute and relative assessment, respectively. An excellent level of reliability was evident for measures of neuromuscular function pre-exercise (Table
2), which is in line with the previous work from our laboratory (Goodall et al.
2015b; Thomas et al.
2015), and importantly, enabled us to detect significant changes throughout the soccer-specific exercise. Similar reliability coefficients have been demonstrated in unfatigued states for both upper (Lee et al.
2008; Madsen
1996; Taylor et al.
1996) and lower limb (Amann et al.
2013; Bachasson et al.
2013; Place et al.
2007; Todd et al.
2004) muscle groups, but the reliability of the fatigue response following locomotor exercise is unknown. The fatigue response was consistent across repeated trials of the SMS, though the variability in the response tended to increase with exercise duration, with most variable responses found at the ET assessment point. The change in MVC at each time point demonstrated a good level of reliability (CV range 6.3–10.8%). The fatigue response identified with peripheral stimulation showed a moderate level of reliability (CV range 5.2–17.8%), whilst the data obtained with motor cortex stimulation showed an excellent level of reliability (CV range 3.0–5.7%). As such, our results demonstrate the fatigue response to the SMS is consistent on repeated trials under the present testing conditions. Furthermore, these results are important for future investigations as they could be used to calculate appropriate sample sizes and ascertain worthwhile changes for the variables studied during this mode of exercise.
Limitations and future directions
There are some limitations worthy of consideration in the current study. Most important is the performance of a simulated activity and the level of ecological validity. Participants covered ~14 km during the SMS, which is in line with the distance covered during an actual game (Russell et al.
2015); however, the prescribed nature of the current, and other simulated protocols, differs to that of a real match scenario. Participants exercised to the sound of audio cues throughout the SMS, whereas the intensity of an actual game would fluctuate according to individual motivation and physiological capacity beyond that assessed by a simple aerobic capacity test, and hence potentially impact the fatigue response. Moreover, participants knew that they were partaking in a 120 min exercise protocol, not a 90 min performance then an unbeknown period of ET. Ergometer-based investigations have shown that the physiological and perceptual responses to exercise are different when the duration is known vs. unknown (Baden et al.
2005; Eston et al.
2012). Notwithstanding, the SMS protocol does serve as a valid laboratory based stimulus that allows the assessment of demands akin to soccer (Russell et al.
2011), and the strict control of the activity profile affords a more reproducible exercise stimulus compared to the variable nature of competitive soccer (Carling et al.
2016). To address these limitations, the neuromuscular fatigue response should be determined following actual match-play, and in a way that the period of ET can be blinded. As with all neuromuscular fatigue investigations, we are aware that aspects of corticospinal function have been shown to recover within 1 min following exercise (Taylor et al.
1996). Thus, the present experimental design might not have elucidated the full extent of central fatigue elicited by the SMS. However, our measurement methods were consistent at each time point and the finding that central fatigue was evident at all-time-points demonstrates the robust and reliable nature of the data.