Twenty trained male team sport athletes competing in regional level soccer, rugby, or field hockey competition, aged 26.6 ± 7.8 y (mean ± SD), and with mean body mass of 81.2 ± 11.3 kg, stature of 179 ± 4.4 cm, and maximum oxygen uptake of 61.6 ± 7.1 mL·kg-1·min-1, volunteered to participate in the study. A sample size of 20 was estimated for the expected effect size on Achilles tendon stiffness based on effect size estimate drawn from change in plantarflexion ROM due to AQUA TITAN of 3.7% (effect size 0.2) (Wadsworth et al. 2010) and the coefficient of variation for tendon length using similar procedures (6.1%) (Fletcher et al. 2010), and the method of sample size for meaningful (effect size) magnitude-based inference (Hopkins et al. 2009). Potential participants were interviewed and subsequently excluded if they had a history of Achilles tendinopathy or lower limb trauma, illness or were currently on analgesic medication. All participants were informed in writing about the potential risks of the study and gave written informed consent for their participation in the study, which was performed according to the Declaration of Helsinki and approved by the University’s Research Ethics Committee prior to the start of the investigation.
Procedure
All participants first completed a treadmill-based assessment of O2max (Wadsworth et al. 2010) followed 1-wk later by familiarization of the experimental procedures (Figure 1). Participants performed a double-blind randomised crossover comprising baseline measurements of joint range of motion (ROM), Achilles tendon stiffness, short latency reflex, and isokinetic torque, followed by a 40-min treadmill protocol to simulate the physical demands of intermittent high-intensity sport shown previously to cause leg muscle fatigue and altered contractility (Drust et al. 2000; Rahnama et al. 2006). The treadmill speeds for each activity in the protocol were based upon the speeds observed for each specific movement category during soccer match-play: walking 6 km·h-1, jogging 12 km·h-1, cruising km·h-1 and sprinting 21 km·h-1. The protocol was arranged around two identical cycles separated by a static recovery period of 90 s. Each cycle consisted of 23 discrete bouts of activity (duration): 6 × walking (35.3 s); 6 × jogging (50.3 s); 3 × cruising (51.4 s), and 8 × sprinting (10.5 s). High-intensity exercise (cruise and sprint) bouts were separated by low-intensity recovery (walk and jog), ordered via within-subject randomisation, and replicated in the second arm of the crossover. Recovery mechanics were assessed 48-h following the run after an intervening rest day. There was a 10-d washout between experimental blocks.
All 20 participants completed one trial wearing AQUA TITAN treated tape and one wearing a placebo tape allocated by double-blind randomization. Both the AQUA TITAN and placebo tapes were custom made by Phiten Co. Ltd. (Kyoto, Japan), using an AQUA TITAN concentration of 540 ml·l-1 in the treatment tape. AQUA TITAN is a suspension of titanium nanoparticles in water (Hirata et al. 2004 Patent 522431). Phiten Co. Ltd also funded the study but had no other involvement in, or right to approve or disapprove the current publication. Both tapes were black and covered the entire posterior lower limb from the calcaneus to the proximal attachments of the medial and lateral gastrocnemius (Figure 2a). The tape was applied following the baseline measures after which it remained in place for the duration of the intermittent treadmill protocol and the entire recovery period including all exercise tests and while sleeping. The blinding code was maintained by an external party and revealed to researchers only after the analysis was performed. To prevent any possible mixing, tapes were stored separately.
The effect of the tape was assessed through dynamometry and analysis of the short latency reflex, in that order. Dynamometry was taken with participants reclined in a supine position at a hip angle of 0° (i.e. fully extended). The foot of the dominant leg was placed against the footplate of the dynamometer. The lever arm of the dynamometer was aligned so that the center of rotation of the dynamometer was aligned with the lateral malleolus. The initial ankle position was preset at neutral, 0°, mid plantar/dorsiflexion. Maximal plantar flexion isometric torque was sampled at six predefined angles (0, 5, 10, 15, 20 and 25°). Two contractions were performed with 120 s recovery between efforts. Participants were instructed to impart force to the foot plate in a gradual manner until they reach maximal effort around 1 s into the contraction then to hold that force for the remaining 2 s. Maximal plantar flexion isokinetic torque was sampled at four predefined angular velocities (30, 60, 90 and 120°·s-1) through a predetermined range of motion (25°). Three repetitions at each angular velocity were performed; each velocity was interspersed with 180 s recovery. The isometric conditions were used to estimate tendon stiffness, while the isokinetic data was used to measure the functional changes in the muscle/tendon complex.
Following the dynamometry, changes in the short latency reflex of the medial gastrocnemius muscle were assessed using a tendon tap method. Participants lay prone with the ankle at 90º passive dorsiflexion and were instructed to relax. The Achilles tendon was tapped with an instrumented reflex hammer operated by hand. This test was performed three times.
Apparatus
An isokinetic dynamometer (Biodex Medical Systems System 3, NY, USA) was used to sample isometric tension and isokinetic torque. Participants were videoed while performing the dynamometer test using a Casio Exlim Ex-F1 camera (Casio Computer Co Ltd, Tokyo Japan) at 30 Hz. Black circles were placed on the Achilles insertion at the calcaneus, the medial malleolus and on the head of the ultrasound transducer (Figure 2b). In order to assess tendon length, ultrasound images of the medial gastrocnemius/Achilles tendon musculotendonous junction (MG/AT MTJ) were simultaneously collected at 10 Hz using a Sonosite, MicroMaxx ultrasound (Sonosite Inc., Bothell, USA).
Torque and position for the dynamometer were collected using an ADI power lab system (PowerLab 4/25, ADInstruments) at 1000 Hz. The ultrasound data collection was manually triggered; the trigger also activated a light emitting diode which was used as a synchronization event in the video image, and a signal was simultaneously sent to the power lab system to start data collection. The video was analyzed to find the first frame in which the light was visible, and this frame coincided with the first frame of the ultrasound image, as well as the first sample from the power lab. Using this method, the video was synchronized to within 1/30 of a second of the ultrasound and dynamometer.
The locations of the markers in the video and the MTJ in the ultrasound image were digitized using MaxTRAQ software (version 2.19-012, Innovision systems Inc. Columbiaville, MI, USA). The ultrasound probe was orientated along the longitudinal axis of the MG/AT MTJ. When the location of the MTJ was ascertained, the probe was positioned so that both the superficial and, importantly, the deep aponeurosis between MG and Soleus were apparent, ensuring accurate and reliable identification of the MTJ. To match the sampling frequency of the ultrasound, only every third frame of the video was digitized. To calculate the location of MG/AT MTJ in absolute space, the coordinates of the MTJ in the ultrasound image were added to the centre point of the ultrasound probe digitized in the video.
To examine the short latency reflex, surface electromyographic (sEMG) activities of the GM muscles were recorded from the right leg using bipolar surface electrodes with a 5 mm diameter and a 10 mm fixed inter-electrode distance (Ambu® Blue Sensor N, Ambu A/S, Ballerup Denmark). Skin preparation and electrode placement were performed according to international guidelines for sensor placement (Hermens et al. 2000). sEMG signals were sampled at 1000 Hz during tendon tap. The Achilles tendon was tapped with an instrumented hammer (ADInstruments, Australia). The sEMG and hammer signal were collected using an ADI power lab system (PowerLab 4/25, ADInstruments). The Tendon Hammer contained a piezo-electric sensor within the head to provide a momentary pulse when a surface is struck with the hammer.
Analysis
Video and ultrasound
We pilot tested the procedures used in other studies (Fletcher et al. 2010; Kay and Blazevich, 2009). During this process, it became clear that we needed to modify our procedures to determine tendon length. The previous studies fixed the ultrasound probe to the belly of the medial gastrocnemius to image the attachment of the distal portion of a muscle fascicle into the deep aponeurosis, where changes in tendon length were inferred from displacement of the muscle fascicles, with the assumption that the aponeurosis distal to the measurement site remained a constant length, which we found was not the case. To avoid the fixed length assumption and to image the muscle-tendon junction more accurately, we manually held the ultrasound probe over the MG/AT MTJ and used video to track the location of the ultrasound markers on the probe (Figure 2b), which was digitized using MaxTRAQ software (Figure 2c). This procedure permitted adjustments to the ultrasound coordinates due to any tilt of the probe to be made by multiplying the vertical coordinates of the ultrasound image by the digitized distance between the base of the probe and the midpoint of the top of the probe (Figure 2b) by the actual distance between those points. By doing this we were able to precisely locate the muscle-tendon junction in absolute space while making no assumptions about the length of the aponeurosis. Reliability of thee method was assessed by test-retest of 4 participants across 3 trials with a typical error measurement (within-subject standard deviation) of tendon length of 2.2%.
Tendon length and stiffness
Tendon length (TL) was calculated as the distance between the calcaneous marker and the MG/AT MJT in absolute space. Based on the assumption that the effect of the dorsiflexor muscles was minimal, Tendon Force (TF) was estimated by dividing the torque about the ankle joint, obtained from the Biodex with the torque caused by the weight of the foot subtracted, by the moment arm of the Achilles tendon, d. The moment arm was calculated using the equation
where
is a vector containing the position of the markers in absolute space, with mt being the musculotendinous junction, c being the calcaneus, and m being the medial malleolus. Only the portion of each trial where the tendon force was increasing was analyzed. Tendon stiffness was calculated by fitting the following equation to the estimated tendon length and force:
where TF and TTL are the calculated tendon force and tendon length measure respectively, and F
0, Ae, and λ was fit using the lsqnonlin function in MatLab. The variable λ was used as a measure of the stiffness. The typical error for λ was 16.8%.
To improve precision, only contractions with an R2 >0.3 (large correlation; 697 of 960 contractions) were used in the tendon stiffness analysis. A lower correlation was deemed too variable for the TF/TL slope. The datasets with R2 > 0.5 reduced data points but did not affect the AQUA TITAN tape outcome, but R2 > 0.7 resulted in failure of the mixed model procedure to converge due to insufficient data points. Estimates of tendon stiffness previously used linear models (Lichtwark et al. 2007). However, we and others (Lieber et al. 1991; Magid and Law 1985; Pinto and Fung 1973; Winters 1990) observed an exponential length-tension relationship for tendon dynamics during contraction, illustrated in Figure 3. Our exponential fit approach was therefore similar to Hill type muscle models (Winters and Stark 1987).
Short latency reflex
The sEMG signal was amplified (BioAmp, ADInstruments, Australia), low band-pass filtered (10–500 Hz) and integrated in Chart for Windows (version7). The onset of reflex sEMG activity was defined as the time between the tendon tap (perturbation) and the first deflection from baseline electrical activity and was determined by visual inspection using a cursor on the display (Grey et al. 2002). The typical error was 4.8%.
Statistical analysis
The effect of treatment on outcomes was estimated with mixed modelling (Proc Mixed, SAS Version 9.1; SAS Institute, Cary, NC). All data were log-transformed before modelling to reduce nonuniformity of error and to express outcomes and confidence limits (CL) as percentages (Hopkins et al. 2009). Estimates for the effect of treatment on tendon stiffness were derived from the least-squares mean interaction of the model terms (fixed effects) trial order, treatment, post-pre difference, and contraction number (1 and 2); random effects were subject interacted with contraction number and treatment. Estimates for ROM, short latency reflex, and peak isokinetic torque were derived from a model but without contraction or tap number due to trivial difference in the magnitude of the intra-sample means, that is, the value provided by the model was the average of the contractions or tap number. Peak isokinetic torque was estimated using the model approach as for ROM, but for each of the 4 levels of angular torque and overall. Statistical inference was by magnitude-based clinical inference (Hopkins et al. 2009), with the between-subject standardized difference (modified Cohen’s d) used as the reference to effect size.