Upon ethical approval from the local clinical Research Ethics Committee (CPP N° 0104), 12 male soccer players volunteered to participate in this study (mean ± SD; age: 22.9 ± 0.9 years; body mass: 71.3 ± 6.4 kg; height: 1.84 ± 0.06 cm, Maximal aerobic speed: 16.4 ± 2.3 km/h, MVC: 829.3 ± 165.5 N, SJ: 36.6 ± 1.94 cm, CMJ: 37.5 ± 3.7 cm, 20 m sprint: 2.8 ± 0.37s). All participants were playing in the third Tunisian National League and trained at least 5 days per week for∼1.5 h per day during the competitive period. Experimental testing sessions took place in the late off-season to early pre-season phase of the players’ training year, while players were still resting.

Normality of every dependent variable and homogeneity of distribution variances (equal variance) were confirmed using Shapiro-wilk test and the Levene test, respectively. Two-way ANOVA (Condition [CWI or TWI] × Time [Baseline,post-match, 24h, 48h and 72 h recovery]) with repeated measures was used to test differences in neuromuscular function, physical performance, and enzymes activity. When a significant difference was found, multiple comparison analysis was performed with Bonferroni post hoc test. Effect sizes were reported as Cohen’s d. Data presented in the results section are expressed as mean ± standard deviation and were analyzed using Statistica for Windows software (version 10, StatSoft, Inc, Tulsa, OK). Statistical significance was set at p < 0.05.

The decrease in MVC from baseline (100%) to immediately post-match intervention (i.e., fatigue) was significantly lower (p < 0.001, d = 1.3) in CWI (-9.9 ± 3%) compared to TWI (-23.7 ± 14.7%). Later, MVC recovered and returned to baseline values at 24 h in CWI and at 72 h in TWI (Figure 2A). The decrease in VA from baseline (100%) to post-match was significantly lower (p < 0.001, d = 1.3) in CWI (-3.7 ± 4.9%) compared to TWI (-15.4 ± 5.6%). After that, VA recovered and returned to baseline values at 24 h in CWI and at 48 h in TWI (Figure 2B). The decrease in Q_tw,pot from baseline (100%) to post-match was significantly lower (p < 0.05, d = 1.15)in CWI (-15.7 ± 5.9%) compared to TWI (-24.8 ± 9.5%). Also, the decrease in Q_tw,pot from baseline to 72 h was significantly lower (p < 0.05, d = 0.98) in CWI (-0.2 ± 7.7%) compared to TWI (-8.8 ± 9.6%). Then, Q_tw,pot recovered and returned to baseline values at 24 h in CWI, yet, it did not fully recover in 72 h in TWI (Figure 2C). Maximum M-wave (VL, ∼9.44 ± 0.6 mV, p > 0.05, VM, ∼ 6.22 ± 0.8mV, p > 0.05, RF, ∼ 5.3 ± 0.4 mV, p > 0.05) and RMS/M_max (∼0.05 ± 0.003 mV, p > 0.05) did not differ from baseline values throughout recovery period in both conditions.

The decrease in SJ from baseline (100%) to post-match was significant, yet not different between conditions (CWI, -7.4 ± 5.7%; TWI,-6.9 ± 4.3%, p > 0.05). Afterwards, SJ recovered and returned to baseline values at 72 h in CWI but did not fully recover in 72 h in TWI (Figure 3A). The decrease in CMJ from baseline (100%) to post-match was significant, however, not different between conditions (CWI, -8.8 ± 2.8%; TWI, -8.9 ± 10.7%, p > 0.05). Posteriorly, CMJ recovered and returned to baseline values at 48 h in CWI and at 72 h in TWI (Figure 3B). The increase in 20 m time from baseline (100%) to post-match was significantly lower (p < 0.05, d = 2.17) in CWI (+11.1 ± 3.2%) compared to TWI (+18.5 ± 3.6%). Then, 20 m performance recovered and returned to baseline values at 24 h in CWI but did not fully recover in 72 h in TWI (Figure 3C).

The increase in CK from baseline (100%) to post-match was significant, but not different between conditions (CWI, +76.5 ± 29.7%; TWI, +132.3 ± 70.4%, p > 0.05). Then, CK peaked at 24 h in both CWI (+154.5 ± 113%) and TWI (+216.5 ± 147.5%) conditions. CK returned to baseline values at 72 h in both conditions (Figure 4A). The increase in LDH from baseline (100%) to post-match was significant, and different between conditions at 48 h (CWI, 11.2 ± 11.3%; TWI, +25.4 ± 22.3%, p < 0.05, d = 0.8). Later, LDH recovered and returned to baseline values at 72 h in CWI but did not fully recover during the 72-h period in TWI (Figure 4B).

When compared to TWI, CWI attenuated the development of post-match peripheral fatigue (Q_tw,pot) by 9.1% at post-match and 8.6% at 72h, resulting in a faster recovery (Figure 2). Since M-waves were unaltered throughout recovery in the two experimental conditions, the greater peripheral fatigue in TWI was not linked to a reduced membrane excitability, and indicates that factors located beyond the sarcolemma, such as alterations in the excitation–contraction coupling, were responsible for the reduction in Q_tw,pot (Balog and Fitts, 1996).

The faster recovery of Q_tw,pot in CWI might be associated with a lower production of reactive oxygen species (ROS) and reactive nitrogen species (RNS) in the skeletal muscle, as they occur ∼3 days following intense exercise in mice (McArdle et al., 1999) and humans (Close et al., 2004). Indeed, an accumulation of ROS/RNS during exercise can affect the sarcolemma Ca²⁺ release, and has been associated with redox dependent modification of ryanodine receptors as well as a reduction in calcium sensitivity of myofilaments (Allen et al., 2008). Therefore, given the potential of CWI to contain inflammation and inhibit the production of ROS/RNS (White and Wells, 2013), it is likely that the faster CWI-induced recovery of peripheral fatigue following a simulated soccer match-play observed in this study is related to reduced ROS/RNS (Andrade et al., 1998). However, it is important to note that ROS/RNS were not collected in the present study and, hence, further explorations are required to confirm this assumption.

In the present study, the interpolated twitch technique and VA calculation were used to assess central fatigue and its recovery kinetics (Merton, 1954). When compared to TWI, CWI attenuated post-match VA by 11.7% and accelerated its recovery (Figure 2). This result could be explained by the ability of CWI to reduce muscle damage, induced by the simulated soccer match-play. This ability was evidenced by the attenuated increase in LDH but, quite surprisingly, not in CK (Figure 4). However, similar discrepancies in blood markers were previously observed after CWI (Vaile et al., 2008). Overall, finding from previous studies investigating effects of CWI after a simulated soccer match-play (Bouzid et al., 2018) or jiu-jitsu training (Fonseca et al., 2016) confirm the positive effects of this therapy on muscle damage. The mechanism(s) responsible for the lower exercise-induced intracellular protein release to plasma following CWI remains unclear. Eston and Peters (1999) suggested that cryotherapy could reduce the post-exercise protein efflux from the muscle into the lymphatic system or reduce the amount of post-exercise damage. This indirect indication of lower muscle damage could be associated with decreased vessel permeability probably due to cryotherapy-induced attenuation of the inflammatory response (Eston and Peters, 1999). Interestingly, muscle damage, induced by prior eccentric exercise, was found to promote central fatigue during subsequent exercise (Endoh et al., 2005). Mechanistically, muscle damage/soreness stimulates the inhibitory feedback from group III/IV muscle afferents, through hormones involved in pain signaling such as bradykinin and prostaglandins (Kaufman et al., 1982), leading to increased central fatigue (i.e., reduction in VA). Similarly, greater ROS/RNS production in TWI compared to CWI, as discussed earlier, would also stimulate group III and IV muscle afferents (Delliaux et al., 2009). Finally, the direct implications of peripheral fatigue, in terms of promoting central fatigue by triggering group III and IV muscle afferent feedback (Amann et al., 2009; Hureau et al., 2018), may have contributed to the blunted VA in CWI compared to TWI.

Additionally, it has been shown that a passive (Morrison et al., 2004) or exercise-induced hyperthermia (Nybo and Nielsen, 2001) is associated with the development of central fatigue. Given the potential of CWI to reduce body temperature (Peiffer et al., 2010), it is likely that the attenuation of central fatigue observed following the CWI in the present study was also associated with a faster reduction in body temperature.

The present study showed that 20 m sprint performance was less impaired immediately after post-match intervention and throughout the time course of recovery in CWI compared to TWI (Figure 3), and this is consistent with previous investigations (Fonseca et al., 2016; Bouzid et al., 2018). It should be noted that post-match jump performances were not less impaired immediately in CWI compared to TWI, but they fully recovered 1 day earlier. Interestingly, the recovery kinetics of physical performance indices were slightly different than those of neuromuscular fatigue, indicating that performance is not a surrogate for fatigue measurement, and that other mechanisms than fatigue might explain the recovery of physical performance. A number of possible explanations could account for the discrepancies between recovery of neuromuscular fatigue and physical measures. Namely, there is an inherent level of variability associated with measures of voluntary performance, with individuals able to alter their jump or sprint mechanics in an attempt to maximize performance (Ratel et al., 2006). For example, it has previously been suggested that individuals alter their jump mechanics when fatigued in order to help maintain jump height (Gathercole et al., 2015). In addition, measures of neuromuscular function target the knee extensors under isometric conditions, while jump performance involves multi-joint dynamic movements, which could further contribute to the discrepancies. In the same line, a previous study found no correlation between the recovery of peripheral and central fatigue indices and physical performance indices following a competitive soccer match-play (Brownstein et al., 2017). Since CWI had a positive impact on the recovery of physical performance in the present study, it is highly important to emphasize that this impact was confirmed after a simulated, and not a competitive soccer match-play. Even so, this does not undermine the reliability of our reasoning, as we strongly believe that the simulated match-play adopted in this study had accurately simulated the physical demands of a competitive soccer match-play (Nicholas et al., 2000). Over and above that, in methodological terms, the use of this exercise modality had certainly contributed to a consistent simulation of physiological demands from one experimental visit to another (i.e., CWI vs. TWI); a crucial condition that unpredictable competitive soccer match-plays could have failed to achieve.