Scand J Med Sci Sports 2014: ••: ••–•• doi: 10.1111/sms.12234

© 2014 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd

Assessment of magnetic resonance techniques to measure muscle damage 24 h after eccentric exercise J. Fulford1, R. G. Eston2,3, A. V. Rowlands2, R. C. Davies3 Exeter NIHR Clinical Research Facility, University of Exeter Medical School, University of Exeter, Exeter, UK, 2School of Health Sciences, University of South Australia, Adelaide, Australia, 3College of Life and Environmental Sciences, University of Exeter, Exeter, UK Corresponding author: Jonathan Fulford, MRI Unit, University of Exeter Medical School, University of Exeter, Exeter, EX1 2LU, UK. Tel: +44 1392 262982, Fax: +44 1392 262926, E-mail: [email protected]

1

Accepted for publication 17 March 2014

The study examined which of a number of different magnetic resonance (MR) methods were sensitive to detecting muscle damage induced by eccentric exercise. Seventeen healthy, physically active participants, with muscle damage confirmed by non-MR methods were tested 24 h after performing eccentric exercise. Techniques investigated whether damage could be detected within the quadriceps muscle as a whole, and individually within the rectus femoris, vastus lateralis (VL), vastus medialis (VM), and vastus intermedius (VI). Relative to baseline values, significant changes were seen in leg and muscle cross-sectional areas and volumes and the resting inorganic phosphate concentration. Significant time effects

over all muscles were also seen in the transverse relaxation time (T2) and apparent diffusion coefficient (ADC) values, with individually significant changes seen in the VL, VM, and VI for T2 and in the VI for ADC. A significant correlation was found between muscle volume and the average T2 change (r = 0.59) but not between T2 and ADC or Pi alterations. There were no significant time effects over all muscles for magnetization transfer contrast images, for baseline pH, phosphocreatine (PCr), phosphodiester, or ATP metabolite concentrations or the time constant describing the rate of PCr recovery following exercise.

Bouts of high-intensity or atypical exercise often lead to modifications in the muscle, leading to general soreness (Armstrong, 1984), increases in inflammatory mediators (MacIntyre et al., 1996), and mitochondrial swelling (Armstrong et al., 1983). These changes are indicative of exercise-induced muscle damage (EIMD), a term used to collectively describe the response. Concomitant with these symptoms, increases in muscle dysfunction become evident, leading to decreases in exercise performance or capacity, such as reductions in force output (Stupka et al., 2001; Clarkson & Hubal, 2002; Byrne et al., 2004), or the time to exhaustion (Davies et al., 2011). In an attempt to gain insight in to the underlying mechanisms associated with EIMD, as well as to provide indications of severity, a number of studies have previously utilized magnetic resonance imaging (MRI) techniques. By far, the most common approach has focused on alterations in the transverse relaxation time (T2; Shellock et al., 1991) following exercise. T2 represents a fundamental magnetic resonance (MR) property of a tissue and reflects the rate at which the generated MR signal decays. It is a parameter that can be altered by a wide range of physiological modifications and although the dominant mechanism behind such changes is unclear, it seems apparent that they are at least in part

associated with osmotic changes, potentially indicative of inflammation and/or edema. When measured in a range of muscle groups, following different exercise studies, including EIMD protocols, significant T2 increases have been observed. These have included, after downhill running in mice (Marqueste et al., 2008; Mathur et al., 2011), hand grip exercise (Morvan & Leroy-Willig, 1995), biceps curl (Foley et al., 1999), short-duration eccentric exercise (Nurenberg et al., 1992; LeBlanc et al., 1993), and extended duration concentric leg contractions (Larsen et al., 2007). Given that inflammation and/or edema are proposed mechanistic explanations for T2 modifications, changes in muscle volume might be anticipated to accompany T2 changes and be detectable themselves within MRI images, particularly given other methodologies have been utilized to record volume changes associated with EIMD (Bobbert et al., 1986; Whitehead et al., 1998). Indeed, MRI has been used extensively to measure limb and muscle volumes (Fukunaga et al., 1992; LeBlanc et al., 2000; Tothill & Stewart, 2002), including assessments of edema (Haaverstad et al., 1992). However, with the exception of a small number of studies (Rodenburg et al., 1994), MRI volume techniques have not been utilized to assess changes resulting from EIMD.

1

Fulford et al. A somewhat more specialized MRI methodology is one that examines the movement of water molecules, via diffusive processes, thereby gaining an insight into cell wall and general tissue integrity. Diffusion-weighted MRI sequences provide scans where the signal intensity within the image is dependent upon the amount of diffusion within the tissue being assessed, with the greater the degree of diffusion, the greater the signal attenuation. The freedom of water to diffuse is expressed via a parameter known as the apparent diffusion coefficient (ADC; Morvan & Leroy-Willig, 1995; Yanagisawa et al., 2003; Yanagisawa & Fukubayashi, 2010; McMillan et al., 2011). In tissues where water is free to move around without restriction, high ADC values are recorded. In contrast, low ADC values are seen in highly organized tissue where the movement of water is limited by structurally intact cell walls or membranes. Thus, any alterations in the water distribution between the microvascular and extravascular muscle compartments will potentially lead to measurable alterations in the ADC value (Hazlewood et al., 1991; Le Bihan, 1991) as will general water content changes or alterations in cellular architecture leading to modifications in the diffusion restrictions. Hence, muscle damage, which may lead to adaptations in all of these characteristics, has the potential to elicit measureable changes in ADC (Cermak et al., 2012). A similarly specialized MRI methodology is known as magnetization transfer contrast (MTC) imaging. An MTC image can be generated as a result of the interaction and subsequent exchange of energy between protons, which are bound within a tissue and those that are free to diffuse. The signal contrast is created via the application of an MR excitation pulse at a frequency that will cause excitation of the bound protons but leaves the free protons unaffected. Although this bound pool does not contribute to the MR signal directly, transfer of magnetization from the bound to free pools leads to a decrease in signal when a subsequent excitation pulse at the free proton frequency is applied. Thus, if two separate scans are obtained, one with, and one without, the bound pool excitation pulse, the signal intensity difference between them provides an indication of the degree of coupling between the two separate pools. As the bound pool is associated with protons that are present within macromolecules, any alteration in the population of this pool may be detected via MTC methods. Although this technique has not previously been used within EIMD protocols, McDaniel et al. (1999) have successfully utilized MTC methods to examine differences in muscle characteristics between controls and muscular dystrophy patients while Yoshioka et al. (1994) have examined acute exercise-induced effects via the technique. 31 P magnetic resonance spectroscopy (MRS) techniques have been widely used within exercise studies to monitor changes in muscle metabolite concentrations.

2

Of these, a number have specifically examined alterations associated with EIMD. These have reported increased baseline values for the inorganic phosphate (Pi) to phosphocreatine ratio (Pi : PCr) (Lund et al., 1998a,b) as well as for the baseline Pi concentration (Davies et al., 2011). Increases in Pi have previously been implicated in inhibiting force generating ability within the muscle (Westerblad et al., 2002; Debold et al., 2006) and may impact on cross-bridge cycling. Thus, any increase in Pi associated with EIMD might readily impact on muscle function. As well as the examination of metabolite concentrations, other nondamaging exercise protocols have utilized 31P recovery kinetics to assess mitochondrial oxidative capacity (Kemp et al., 1993; Vanhatalo et al., 2011). Although no EIMD studies have utilized this approach, eccentric exercise has been associated with mitochondrial swelling and damage to the sarcoplasmic reticulum (Armstrong et al., 1983). Given that any degree of mitochondrial dysfunction, either directly, or indirectly from impaired vascular function, may impact on mitochondrial oxidative capacity, such alterations arising from EIMD could potentially be examined via assessments of PCr recovery kinetics. Peak decrements in markers of static and dynamic muscle function (isometric and dynamic strength, power, speed, jump height) have been noted to occur between immediately post to 24 h post-eccentric exercise protocols (Byrne & Eston, 2002; Byrne et al., 2004; Marginson et al., 2005). The aim of the present study was to utilize a range of MR techniques, to determine which are most sensitive to the early signs of muscle damage, 24 h after eccentric exercise. By doing so, insights into the principal mechanistic basis of muscle damage at that time may be provided. Subsequently, leg and muscle cross-sectional areas and volumes, T2, MT images, ADC, phosphorous metabolite concentrations, intracellular pH and maximum oxidative capacity were each calculated prior to and 24 h after the muscledamaging protocol.

Methods Seventeen healthy, physically active participants (eight males, nine females, age, 20.7 ± 1.3 years; mass, 69.2 ± 12.8 kg, height, 1.72 ± 0.10 m) attended the MR center immediately prior to and 24 h after an eccentric muscle-damaging protocol. All participants were asymptomatic of illness and preexisting injuries and had not performed any resistance training of the lower limbs within the previous 6 months. Participants provided written informed consent to participate in the study, which had been approved by the Institutional Ethics Committee. Before the data collection period, participants were fully familiarized with all facets of the protocol and given the opportunity to practice until they were comfortable with the requirements, thereby minimizing any possible learning effects and allowing an estimation of suitable working forces for the study. Participants were instructed to drink sufficient fluids to ensure they were fully

MR techniques and muscle damage detection hydrated when tested and were requested to desist from undertaking strenuous physical exertion for 48 h prior to the testing procedure.

Experimental procedures Eccentric, muscle-damaging exercise protocol Participants completed 100 (Smith) squats, performed as 10 sets of 10 repetitions with the load on the bar corresponding to ∼70% of each participant’s body mass. Prior to commencing, all participants were instructed in correct and safe lifting technique. The bar was positioned on the participant’s shoulders and feet were positioned under the bar, with the back straight and legs fully extended (knee = 180°). The descent phase involved eccentric action of the knee extensors to lower the bar to a knee angle of just past 90°. The lifting phase involved concentric action to return the bar to the starting position. This protocol has previously been used to induce significant EIMD at 24 h (Byrne & Eston, 2002; Davies et al., 2008, 2009). Non-MR assessment of muscle damage In order to assess the relative level of muscle damage induced by the protocol, a number of non-MR measures were also untaken. These consisted of perceived muscle soreness, creatine kinase (CK) activity, squat jump height, and isokinetic peak torque (30 deg/s), measured in the order listed, immediately before, and 24 h after performing the eccentric, muscle-damaging exercise protocol. In both cases, these measures were obtained after completion of all MR measurements in order to avoid acute exercise-induced modification of the MR parameters. Perceived soreness of the knee extensors was assessed using a blank 0–10 visual analog scale (VAS). The VAS consisted of a 10-cm line labeled from left (no soreness) to right (worst soreness ever). Participants squatted to 90° knee flexion with hands on hips and then placed a mark on the VAS to indicate their level of soreness. Perceived pain was quantified by measuring the distance to the mark on the line to the nearest 0.1 cm. Plasma CK activity was assessed from fingertip capillary samples. The sample was centrifuged at 4000 rpm (2000 g) for 5 min and two 20 μL samples of plasma were then added to 1 mL of reagents (Randox CK-NAC 110, Randox Laboratories Ltd, Crumlin, Co. Antrim, UK). The solution was incubated at 37 °C and absorbance at 340 nm was recorded by spectrophotometry (Jenway 6310 spectrophotometer, Jenway, Essex, UK) at 1, 2, 3, and 4 min. CK values were calculated using the formula CK(U/L) = 8095 × Δ absorbance 340 nm/min. The mean CK value of the two samples was calculated and used for subsequent analysis. Normal serum values of 24–195 and 24–170 U/L are reported for men and women, respectively, using this method (Szasz et al., 1976).

Squat jump height was assessed using an infrared jump system with microgate optojump software, which utilizes the method described by Bosco et al. (1983) to calculate the rise in the center of gravity (Marginson et al., 2005). The squat jump involved a rapid preparatory downward eccentric movement, thereby incorporating the stretch-shortening cycle and hence increasing the performance sensitivity to muscle damage. Before the assessment of jump height, all participants received a standardized warm-up of five submaximal continuous jumps and five maximal continuous jumps, to minimize the risk of injury. Participants performed three maximal jumps, separated by a 1-min rest. Visual feedback was provided throughout all testing. Participants were encouraged to perform to their maximal capacity and to try to jump higher than their previous jump with the highest jump height was recorded. Isokinetic peak torque was measured using a Biodex B-2000 isokinetic dynamometer (Biodex Corp, Shirley, New York, USA), which was calibrated prior to each data collection session in accordance with the manufacturer’s guidelines. Following familiarization sessions, participants performed five maximal voluntary contractions (MVCs) at 30 deg/s with a rest period of 30 s between contractions. The highest of the five MVCs was recorded. Visual feedback, displaying real-time force, was used to encourage maximal effort. These procedures have been used previously (Davies et al., 2008, 2009). MR measurements All MR measurements were performed in the University of Exeter Magnetic Resonance Research Centre using a 1.5-T superconducting MR scanner (Philips Gyroscan Clinical Intera, Philips Medical Systems, Amsterdam, the Netherlands). MRI Participants were positioned within the scanner feet first, in a supine position, with a four-element wrap around coil positioned such that it covered the subject’s legs from below the knee to above the hip. To ensure consistency in measurement location between visits, the location where analysis was undertaken was based upon its position relative to anatomical references. For the baseline visit, scout images were initially acquired followed by a high-resolution stack in the coronal plane to allow identification of the lateral and medial condyles. The coronal images were obtained via a T1 (longitudinal relaxation time) weighted turbo spin echo (TSE) sequence with an echo time (TE) of 15 ms and a repetition time of 282 ms (TR). Twenty slices were obtained with an in-plane resolution of 1.04 × 1.04 mm, a slice thickness of 5 mm and a slice gap of 0.5 mm, with three signal averages. A location within the center of the thigh was then selected based upon the coronal images and the

3

Fulford et al. position of this ‘central slice’ relative to the condyles was determined via measurement tools contained within the scanner software. On the 24 h post-exercise visit, scout and coronal images were again acquired and the central slice position was replicated by placing the slice at the same distance from the condyles as for the baseline visit. Other than for general leg volume and area measurements, all other MRI techniques generated muscle group specific information. In these cases, measurements were carried out for the four muscles within the quadriceps, namely, rectus femoris (RF), vastus lateralis (VL), vastus medialis (VM), and vastus intermedius (VI). For each scanning methodology, a region of interest (ROI) was manually drawn around each muscle within the right leg and the MRI information within that region recorded. Muscle volume and cross section In order to examine the variation in area and volume within the leg, a T1-weighted TSE sequence with a TE of 15 ms and a TR of 421 ms was used to generate slices in the transverse plane. Due to the chosen sequence parameters, fat appeared with high signal intensity within the images, in contrast to cortical bone that appeared dark, with muscle having an intermediate signal intensity. A stack of 29 slices centered on the “central slice” location were obtained with in-place resolution of 0.98 × 0.98, a slice thickness of 5 mm, and a slice gap of 0.5 mm, with three signal averages. For the central slice, the crosssectional area was determined based upon an image intensity thresholding method that resulted in a tracing of the outer limits of the leg with the exclusion of the surrounding background. Any internal structures with low signal intensity such, as cortical bone, which had been excluded by the high-intensity thresholding method used, were then included by manually modifying the ROI selected. A sub-volume of the leg, a length of 11.6 cm, comprising 21 continuous slices in the foothead direction, covering the majority of the quadriceps muscle, was calculated by the same intensity selection method as for the single slice. Total volume was calculated based on a sum of the area of each slice multiplied by its thickness. Separate area and volume measurements were carried out for both legs, with the values reported an average of the two. In order to determine total muscle size, both area and volume measurements were repeated with the same data set and employing the same methodology, with the only modification being that both upper and lower intensity thresholds were set so as to exclude high-intensity regions such as fat and lowintensity regions such as cortical bone, thereby leaving only muscle. For all participants, the same thresholding values were selected for both visits having normalized for the intensity range displayed within the images. An illustration of an image where the outer perimeter of the leg and the area of the muscle within are defined is shown in Fig. 1.

4

As the volume and area determinations are particularly sensitive to the reproducibility of measurement location, consistency was assessed by undertaking repeat measurements on six participants not involved in the muscle damage protocol who attended on two occasions using the techniques as described above. Leg volume measurements were determined with a mean difference for the second visit relative to the first of − 0.08%. Reliability measurements have previously determined a coefficient of variation (CV), for leg volume of 0.6% and for cross-sectional area of 0.8%. T2 The T2 values were determined using a multi-echo spin echo sequence with a TR of 1000 ms and eight echoes with TE of n × 13 ms. A single central slice was examined with an in-plane resolution of 0.89 × 0.89 mm and a slice thickness of 5 mm with two signal averages. The signal intensity (S) for each echo time, for each muscle, was then determined and the values of T2 calculated, based on the relationship:

S = S0 exp(− TE T 2) Where S0 is a constant. A plot of the natural log of S was then fitted against TE for all TE values, such that the gradient of the best fit line was given by − 1/T2. Reliability measurements have previously determined a CV, averaged over all four muscles of the quadriceps, of 3.1%. T2 weighted Rather than merely undertaking a calculation of T2, a T2-weighted imaging sequence was also run to assess any T2 based signal enhancement post damage. A dual echo TSE sequence was utilized with TE values of 10 and 100 ms and the ratio of signal intensity at the two TE

Fig. 1. Example image illustrating how the outer perimeter of the leg and the area of the muscle within are defined, based upon image intensity thresholding, which results in all pixels with intensity values within a specified range being included in area determinations.

MR techniques and muscle damage detection values used to define signal enhancement. A single central slice was examined with an in-plane resolution of 1.98 × 2.12 mm and a slice thickness of 5 mm with three signal averages. Reliability measurements have previously determined a CV, averaged over all four muscles of the quadriceps, of 5.4%. Magnetization transfer To assess magnetization transfer contrast, a TSE sequence was used with a TR of 1837 ms and a TE of 100 ms. A single central slice was examined with an in-plane resolution of 0.66 × 0.84 mm and a slice thickness of 3 mm with three signal averages. The sequence was run twice, once with and once without an offresonance MT pulse and the images subtracted to yield an MT contrast image to determine the increase in signal resulting from the absence of the off-resonance pulse. Reliability measurements have previously determined a CV, averaged over all four muscles of the quadriceps, of 7.8%. ADC To assess the ADC, a single-shot echo planar imaging sequence was used with a TR of 2000 ms and a TE of 56 ms. A single central slice was examined with an in-plane resolution of 4 × 4 mm and a slice thickness of 10 mm with six signal averages. Five different diffusion weightings were used (b = 0, 200, 400, 600, 800 mm2/s) and an ADC was calculated based on the b = 200 and 800 values such that the ADC was given by:

ADC800 − 200 = − 1 600 ln (signal intensity at b = 800 signal intensity at b = 200 ) The b = 200 value was selected rather than b = 0 value as the former is insensitive to flow from large vessels present within the ROI, whereas the latter will include signal from vessels, and thus when compared with the b = 800 may overestimate the ADC value. Reliability measurements have previously determined a CV, averaged over all four muscles of the quadriceps, of 6.7%. MRS After completion of the MRI measures, participants were removed from the scanner to allow equipment appropriate for spectroscopy and exercise within the scanner to be set up. Resting phosphorous metabolite concentrations Absolute baseline concentrations of metabolites were established via a technique similar to that described by

Kemp et al. (2007). Participants were positioned within the scanner head first in a prone position with a 6 cm 31P transmit/receive surface coil placed within the scanner bed and positioned such that the subject’s right quadriceps muscle was centered directly over it and a phosphoric acid source was directly beneath it. After initially acquiring images to confirm the muscle was positioned correctly relative to the coil, spatially localized spectroscopy was undertaken to determine the relative signal intensities obtained from the phosphoric acid source and inorganic phosphate from the subject’s quadriceps. On completion of the exercise protocol to determine PCr recovery constants, as described in the following section, and after the participant had been removed from the scanner, subsequent scans were obtained comparing the signals obtained from the same phosphoric acid standard and an external inorganic phosphate solution of known concentration, where the localized voxel sampled within the external solution were of the same dimensions and distance from the coil as from the muscle previously, allowing the calculation of muscle Pi concentration following corrections for relative coil loading. Absolute values of PCr, phosphodiester and ATP concentrations were subsequently calculated via the ratio of the area of the Pi peak relative to the areas of the PCr, phosphodiester, and ATP peaks, respectively, within the acquired muscle spectrum. From this, resting ratios of PCr : Pi were calculated. Reliability measurements have previously determined a CV for PCr, Pi, ATP, and PDE concentrations of 1.4%, 6.7%, 3.5%, and 14.8%, respectively. PCr recovery constant The rate of PCr recovery post-exercise was assessed by performing single-legged, knee-extension exercise tests. Exercise was conducted in the prone position within the MRI system with the 6-cm 31P transmit-receive surface coil placed within the subject bed directly beneath the quadriceps muscle of the leg to be exercised. Subjects were then secured to the ergometer bed with Velcro straps at the thigh, buttocks, and lower back to minimize extraneous movement during the protocol. The foot of the leg to be exercised was connected to a pulley system that permitted a nonmagnetic weight to be lifted and lowered and work rate to be calculated. Exercise was performed in two bouts of 24 s separated by a recovery period of 4 min with the subjects lifting and lowering the mass over a distance of ∼0.22 m at a rate of 0.66 Hz in accordance with a visual cue projected onto the front wall of the scanner room. The contraction phase of the knee extensors and the 31P-MRS interrogation of the quadriceps occurred in unison. The period of 24 s was selected, as exercise with a heavy weight (1 kg less the maximum each subject was able to attain determined during a previous practice session) for that period results in a significant PCr depletion (∼40%), thereby

5

Fulford et al. maximizing the accuracy of the recovery data, without leading to a measureable decrease in pH, which is known to affect the rate of PCr recovery (Jubrias et al., 2003; van den Broek et al., 2007). Data were continuously acquired every 1.5 s, with a spectral width of 1500 Hz and 1 K data points. Phase cycling with four phase cycles was employed, leading to a spectrum being acquired every 6.0 s. 31

P data analysis

The acquired spectra were quantified via peak fitting, assuming prior knowledge, using the jMRUI (version 3) software package employing the AMARES fitting algorithm (Vanhamme et al., 1997). Spectra were fitted assuming the presence of the following peaks: Pi, phosphodiester, PCr, α-ATP (2 peaks, amplitude ratio 1 : 1), γ-ATP (2 peaks, amplitude ratio 1 : 1), and β-ATP (3 peaks, amplitude ratio 1 : 2 : 1). Intracellular pH was calculated using the chemical shift of the Pi spectral peak relative to the PCr peak (Taylor et al., 1983). For the PCr values following the 24 s exercise period, PCr recovery was fitted with Prism 5 software (GraphPad Software Inc, La Jolla, California, USA) by a single exponential of the form;

PCr(t ) = PCrend + PCr(0) (1 − e(− t τ) ) Where PCrend is the value at the end of exercise, PCr(0) if the difference between the PCr at end exercise and fully recovered, t is the time from exercise cessation, and τ is the time constant for the exponential recovery of PCr. Each 24 s exercise bout recovery was fitted individually and the time constants determined from each, before being averaged to give the value quoted for the trial. Reliability measurements have previously determined a CV for τ of 4.0%. Statistics Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS, ver. 21, SPSS Inc, Chicago, Illinois, USA). All data were assessed to examine whether they conformed to a normal distribution by running Shapiro– Wilk tests with a significance threshold of P < 0.05. For muscle soreness, plasma CK activity, isokinetic peak torque, and squat jump height muscle damage, main effects with time were determined by running repeated-measures analysis of variance (ANOVA) within a general linear model (GLM). If significant main effects were found, two-tailed paired t-tests with a Bonferroni correction to the alpha level were run for each individual parameter to confirm that the muscledamaging protocol had been effective in terms of functional changes. A significant intervention was accepted at a P-value < 0.0125.

6

To check for any gender effects, independent samples t-tests were run for all pre- and post-MR measures, as well as for pre-post changes, grouping for gender. When assessing for statistical significance for localized MR techniques where parameter changes were examined in individual muscles within the quadriceps, main effects with time were determined by running repeated-measures ANOVA within a GLM. If significant main effects were found, two-tailed paired t-tests with a Bonferroni correction to the alpha level were run for each individual muscle, comparing pre- and postmeasures, where significance was defined as a value P < 0.0125, thereby correcting for an elevated type I error rate arising from the multiple comparisons. Likewise for the assessment of changes in the four phosphorous metabolites measured, significance was also defined as occurring for P < 0.0125. For cross-sectional area, volume and PCr recovery comparisons, two-tailed paired t-tests were run comparing pre- and post-measures where significance was accepted for P-values < 0.05. For leg and muscle cross-sectional areas and volumes, to account for variable limb dimensions between participants, changes in area and volume post-damage were also expressed as a percentage of the pre-damage values, thereby normalizing for limb size. To assess the significance in these changes, one-sample t-tests were run, where significance was accepted for P-values < 0.05. For all parameters, average values were determined along with standard deviations and 95% confidence intervals. For parameters that showed significant differences between visits, difference values (visit 1–visit 2) along with the associated 95% confidence intervals in the difference were also expressed. In order to estimate the sample size (SS) required to obtain statistically valid results for each parameter, calculations were undertaken based upon the effect size and assuming an α-level of 0.05 and a power (1 − β) of 0.8. Reciprocally, the power values for the results obtained within the present study with a SS of 17 were calculated assuming an α-level of 0.05. In the event of a specific muscle region showing significant changes with multiple localized techniques (T2, ADC, and MT), Pearson correlation coefficients were determined, examining the correlation between the changes in value for the variables between the two time points (pre- and post-damage), correcting for elevated type I error rate in order to determine significance, to an extent dependent upon the number of multiple comparisons, which were run. Similarly, correlations were run between any techniques, which showed significant main effects with time with the differences between time points averaged over all four muscle groups examined. As it has previously been suggested that a similar mechanism might be responsible for any muscle damage-induced changes seen in T2 and muscle volume, a correlation was run between the changes seen in

MR techniques and muscle damage detection muscle volume in the whole leg and the average T2 changes seen over all four muscles in the quadriceps. Results All participants successfully completed the protocol and none reported any adverse effects other than the soreness associated with the muscle-damaging protocol. All of the data assessed was found to be normally distributed. Significant changes were seen in all markers of muscle damage following eccentric exercise. Muscle soreness increased from 1.4 ± 1.0 to 7.2 ± 1.3 arbitrary units (P < 0.001). Plasma CK activity increased from 159 ± 84 to 291 ± 155 U/L (P < 0.001) isokinetic peak torque (45 deg/s) decreased from 190 ± 44 to 149 ± 48 N.m (P < 0.001) and squat jump height decreased from 28 ± 5 to 22 ± 6 cm (P < 0.001). Assessments of all pre- and post-MR measures, and for pre-post changes, revealed no significant gender effects. As a result, all data were pooled into a single group for subsequent analysis. For T2 measurements, a significant main effect with time was seen across the muscle groups (P = 0.003). Significant individual effects were seen in VL (P = 0.001), VI (P < 0.001), and VM (P = 0.004), but not in RF (P = 0.221). Pre values for RF, VL, VI, and VM were 37.74 ± 0.45 (CI: 36.78–38.69), 40.86 ± 0.55 (CI: 39.70–42.02), 40.10 ± 0.39 (CI: 39.27 ± 40.94), and 39.63 ± 0.54 (CI: 38.50–40.77) ms, respectively. Equivalent post-damaging protocols were 38.52 ± 0.51 (CI: 37.44–39.60), 42.77 ± 0.37 (CI: 41.99–43.54) (difference − 1.90 ± 0.49, CI: − 2.95, − 0.86), 42.76 ± 0.39 (CI: 41.93 ± 43.59) (difference − 2.65 ± 0.56, CI: − 3.84, − 1.46) and 41.50 ± 0.34 (CI: 40.77–42.22) (difference − 1.86 ± 0.55, CI: − 3.04, − 0.69) ms. Power calculation and SS requirements to provide sufficient power to detect a statistically significant effect if such an effect exists were given as: RF: (1 − β) = 0.164, SS = 127. VL: (1 − β) = 0.926, SS = 13. VI: (1 − β) = 0.990, SS = 9. VM: (1 − β) = 0.825, SS = 17. For T2-weighted images, significant main effects with time were observed (P < 0.001). Significant effects were seen in VL (P < 0.001), VI (P < 0.001), and VM (P < 0.004), but not in RF (P = 0.923). Pre values for RF, VL, VI, and VM, were 8.12 ± 0.18 (CI: 7.75–8.49), 9.41 ± 0.22 (CI: 8.94–9.87), 9.29 ± 0.16 (CI: 8.96–9.62), 7.76 ± 0.14 (7.46–8.05), and post 8.10 ± 0.24 (CI: 7.60– 8.60), 10.43 ± 0.19 (CI: 10.03–10.83) (difference − 1.02 ± 0.20, CI: − 1.45, − 0.59), 10.67 ± 0.17 (CI: 10.30–11.04) (difference − 1.38 ± 0.23, CI: − 1.88, − 0.88), and 9.06 ± 0.22 (CI: 8.61–9.52) (difference − 1.31 ± 0.19, CI: − 1.71, − 0.90). Power calculation and SS values were RF: (1 − β) = 0.051, SS = 13 989. VL: (1 − β) = 0.997, SS = 8. VI: (1 − β) = 1.000, SS = 7. VM: (1 − β) = 1.000, SS = 6. For ADC values, significant main effects with time were observed (P = 0.004). Significant between visit

effects were seen in VI (P = 0.005) but not in RF (P = 0.221), VL (P = 0.123) and VM (0.030). Values for RF, VL, VI, and VM prior to the damaging protocol were 1.524 ± 0.023 (CI: 1.476–1.572), 1.579 ± 0.022 (CI: 1.532–1.626), 1.565 ± 0.026 (CI: 1.510–1.620), and 1.451 ± 0.032 (1.384–1.519) × 10−3 mm2/s, respectively. Equivalent post-damaging protocols were 1.511 ± 0.022 (CI: 1.465–1.557), 1.617 ± 0.023 (CI: 1.568 − 1.666) (difference: − 0.087 ± 0.027, CI: − 0.145, − 0.0298), 1.652 ± 0.019 (CI: 1.612–1.693) and 1.550 ± 0.032 (CI: 1.483–1.618) × 10−3 mm2/s. Power calculation and SS values were RF: (1 − β) = 0.119, SS = 205. VL: (1 − β) = 0.331, SS = 53. VI: (1 − β) = 0.855, SS = 15. VM: (1 − β) = 0.607, SS = 26. For MT, no significant main effects were seen with time (P = 0.941). Values for RF, VL, VI, and VM prior to the damaging protocol were 8.97 ± 2.7 (CI: 3.25–14.69), 5.34 ± 1.37 (CI: 2.43–8.25), 7.39 ± 1.80 (CI: 3.58– 11.21), 9.03 ± 1.78 (CI: 5.27–12.79)%. Equivalent post-damaging protocols were 9.87 ± 2.61 (CI: 4.34– 15.39), 5.95 ± 1.09 (CI: 3.64–8.26), 6.44 ± 0.99 (CI: 4.35–8.54), 9.20 ± 1.21 (6.64–11.76)%. Power calculation and SS values were RF: (1 − β) = 0.054, SS = 3034. VL: (1 − β) = 0.062, SS = 1187. VI: (1 − β) = 0.065, SS = 911. VM: (1 − β) = 0.051, SS = 29 265. Significant changes were seen in the leg single slice cross-sectional area following the damaging protocol (P = 0.011), with pre and post values of 21 752.0 ± 810.7 (CI: 20 033.3, 23 470.6) and 22 225.5 ± 823.9 mm2 (CI: 20 478.8, 23 972.2) (difference: − 473.6 ± 165.0, CI: − 823.5, − 123.7 mm2) recorded. Likewise, when leg volume changes were assessed, a significant modification (P = 0.029) was determined with pre and post values of 2498.1 ± 92.1 (CI: 2302.8, 2693.4) and 2544.4 ± 92.4 cm3 (CI: 2348.4, 2740.3) (difference: − 46.3 ± 19.3, CI: − 87.2, − 5.3 cm3) found. Equivalent assessments of muscle changes showed significant changes in the muscle single slice cross-sectional area following the damaging protocol (P = 0.003), with pre and post values of 14 119.2 ± 692.0 (CI: 12 652.2–15 586.3) and 14 457.7 ± 711.5 mm2 (CI: 12 949.5–15 965.9) (difference: − 338.5 ± 98.2, CI: − 546.6, − 130.4 mm2) recorded. Muscle volume changes were also significant (P = 0.009) with pre and post values of 1587.3 ± 78.5 (CI: 1420.8, 1753.8) and 1620.4 ± 81.1 cm3 (CI: 1448.5, 1792.4) (difference: − 33.1 ± 11.2, CI: − 56.8, − 9.5 cm3) determined. Power calculation and SS values were the following: single slice leg cross-sectional area: (1 − β) = 0.768, SS = 19; leg volume: (1 − β) = 0.614, SS = 26; single slice muscle cross-sectional area: (1 − β) = 0.899, SS = 14; muscle volume: (1 − β) = 0.797, SS = 18. When assessing the percentage change in areas and volumes relative to pre-damage values significant changes were found for leg single slice cross-sectional area (P = 0.010), with an increase of 2.24 ± 3.14% (CI:

7

Fulford et al. 0.63, 3.86%), for leg volume (P = 0.025), with an increase of 1.94 ± 3.24% (CI: 0.27, 3.61%), for muscle single slice cross-sectional area (P = 0.001), with an increase of 2.38 ± 2.56% (CI: 1.06, 3.70%) and for muscle volume (P = 0.005), with an increase of 2.04 ± 2.58% (CI: 0.72, 3.38%). Power calculation and SS values were single slice leg cross-sectional area: (1 − β) = 0.878, SS = 14. Leg volume: (1 − β) = 0.763, SS = 19. Single slice muscle cross-sectional area: (1 − β) = 0.978, SS = 9. Muscle volume: (1 − β) = 0.930, SS = 12. When phosphorous metabolites were examined no significant changes were seen in PCr (P = 0.400), ATP (P = 0.510) or PDE (P = 0.698) concentrations. However, significant changes were recorded for Pi concentration (P < 0.001). Pre and post values were PCr 32.86 ± 0.96 (CI: 30.83, 34.90) and 33.26 ± 1.08 mM (CI: 30.98, 35.55), Pi 3.82 ± 0.19 (CI: 3.42, 4.22) and 4.91 ± 0.22 mM (CI: 4.45, 5.38) (difference: − 1.09 ± 0.16, CI: − 1.43, − 0.75 mM), ATP 8.66 ± 0.36 (CI: 7.90, 9.42) and 8.50 ± 0.35 mM (CI: 7.77, 9.24), PDE 1.95 ± 0.16 (CI: 1.61, 2.29) and 1.87 ± 0.23 mM (CI: 1.39, 2.35). No significant changes were found for the rate of PCr recovery with values of 27.88 ± 1.93 (CI: 23.79, 31.97) and 28.82 ± 2.17 s (CI: 24.23, 33.41) found for pre and post, respectively. Power calculation and SS values were PCr: (1 − β) = 0.129, SS = 180. VL: (1 − β) = 1.000, SS = 6. VI: (1 − β) = 0.097, SS = 297. VM: (1 − β) = 0.065, SS = 909. PCr recovery: (1 − β) = 0.211, SS = 91. As significant changes were found for both T2 and ADC in VI, correlation coefficients were run between the changes found for both variables, but no significant correlation was found (r = 0.143, P = 0.584). Likewise, given both T2 and ADC changes showed main effects with time, correlations were run between the changes for the two parameters averaging over individual muscles. However, no significant correlation was seen (r = 0.283, P = 0.271). The correlation run between the changes seen in muscle volume in the whole leg and the average T2 changes seen over all four muscles in the quadriceps revealed a significant interaction with r = 0.587 (P = 0.013). In contrast, neither the correlation between average T2 and Pi or average ADC and Pi changes showed significant interactions (r = 0.125, P = 0.633 and r = 0.187, P = 0.473, respectively). Discussion Assessments of the non-MRI measures of EIMD confirmed the exercise protocol had successfully induced a degree of muscle damage that could be recorded at 24 h post-event. Perceived soreness, CK activity, squat jump height, and isokinetic peak torque measures were all markedly changed after the damaging exercise protocol. As a result, although muscle damage has many different

8

components associated with it, each with their own individual time course, some of which do not peak until several days after the muscle-damaging protocol, it would be anticipated that some of the MR techniques utilized would have capacity to detect early onset changes, if the specific technique was attuned to any of the tissue parameters that EIMD alters. This is particularly the case given that the techniques used have previously resulted in a high degree of reliability. Previous work has reported that changes in T2 reach a peak 3–5 days post-damage induction (Shellock et al., 1991). Subsequently, at the time of analysis within the current study, T2 may not have reached maximum amplitude. However, in line with previous work (Marqueste et al., 2008; Mathur et al., 2011), significant changes were still detectable for T2 within the VI, VL, and VM. However, no significant changes were visible in the RF. The form of the damage-inducing exercise will inevitably dictate the spatial damage pattern that is subsequently seen. For the squat exercise used in the present study, all four muscles within the quadriceps group will be challenged. However, they will not be equally susceptible to the eccentric form of exercise that typically induces muscle damage. The RF is the only muscle within the quads, which crosses the hip, whereas the others originate on the proximal femur. During knee flexion, as the body is lowered, the RF is not as taxed as the VF, VL, and VI as the portion of it at the proximal end is not as extensively stretched (Basmajian, 1976). In contrast, the other muscles are working at longer relative lengths, resulting in greater damage being induced. The best way to damage the RF would be for the participant to lie prone with the hip hyper-extended and then, with the knee in a flexed position, force knee flexion with the participant providing muscular resistance. This would tax both proximal and distal parts of the RF with the entire muscle at a long length, unlike in the present study when the participant performed half squats. Thus, the absence of a significant T2 increase in the RF, in contrast to VL, VM, and VI is in line with the demands of the exercise. Such results are also similar to those found by Takahashi et al. (1994) where significant T2 increases were reported 24 h after an exercise-damaging protocol involving repeated bouts of raising and lowering the body from a sitting position in the VL and VI, whereas no changes were found in the RF. As would be anticipated, similar results were found when comparing the spatial distribution of significant muscle changes determined from T2 calculations and from the ratio of signal intensities from T2-weighted images. Given that both are dictated by tissue T2 value, such agreement has a basic underlying mechanism. However, it does confirm that changes in T2 associated with muscle damage can be recognized by employing a simple dual echo T2-weighted sequence rather than having to calculate T2 values, allowing a recognition of muscle changes using a scanning sequence that takes

MR techniques and muscle damage detection less time and requires less time to analyze. That it is possible to discriminate the changes manifested in specific muscles, as seen in the T2 analysis, highlights one of the strengths of some MRI techniques, namely that they can provide spatially localized information, something that is not possible with more generalized methodologies such as measurements of exercise performance or CK concentrations. Although there are upper limits to the practical degree of resolution as a result of signal to noise constraints, it is nevertheless possible to compare, not only different muscles, but different regions within the same muscle, down to volumes of the order of 1 mm3. This is particularly pertinent when specific small muscle masses have been damaged, as these will almost certainly not produce sufficient disturbances from normal levels in whole body plasma CK values to be detectable or impact significantly on performance, unless a tailored exercise regime is selected that can assess the muscle in isolation from surrounding muscles. The origin of increases in the value of T2 remain unclear, in part due to it almost certainly having multiple components, such as edema, inflammatory responses, and direct muscle fiber damage, with their relative importance being dependent upon the exact exercise modality. It has been suggested (Meyer & Prior, 2000) that increased T2 values may be due, in part, to osmotically mediate fluid shifts into the intracellular space resulting from the accumulation of inorganic phosphate and lactate. Alternatively, there may be an increased extracellular water component resulting from the buildup of degraded protein components (Shellock et al., 1991; Clarkson & Hubal, 2002). However, whatever the exact underlying mechanism causing the T2 changes, it has been shown to be a reliable indicator of muscle damage, with alterations being detected in a range of conditions unrelated to exercise but which have an impact on muscle characteristics, such as Duchenne muscular dystrophy (Kim et al., 2010) and stroke (Ploutz-Snyder et al., 2006) and sometimes, although not always, it correlates well with other markers of damage, such as increased CK blood levels (LeBlanc et al., 1993; Rodenburg et al., 1994, 1995). In support of general edema and muscle swelling being at least in part responsible for the recorded T2 changes, significant differences were found for both leg and muscle volume and cross-sectional areas. This is in line with previous assessments of muscle volume following a damaging protocol albeit not employing an MR-based volumetric approach (Bobbert et al., 1986; Whitehead et al., 1998). Examinations of the crosssectional area of specific muscle groups undertaken by Takahashi et al. (1994) revealed increases in the VL, VI, and VM to be approximately 5–7%, 24 h after exercise. In contrast, the RF showed a 2% decrease in area at the same time point. Longitudinal volumetric assessments are inevitably very susceptible to accurate participant repositioning. However, within the present study, as well

as in previous assessments, its viability via the utilization of either anatomical or external makers has been well demonstrated. That there was a significant correlation between the average T2 of the muscles examined, and the overall changes in muscle size, provides support to them having a common modifying cause, although, inevitably, causality cannot be confidently inferred by correlation, and it may be that the two parameters are dependent upon different underlying causes, which share a similar time response. Like the T2 measures, an overall significant main time effect was found for ADC values. However, a significant effect was only found within the VI when individual pre and post values were assessed. In contrast, no significant effects were seen in RF or VL or VM. Such a finding is generally in line with those generated from the T2 examinations, in particular as regards the lack of modification within the RF. Previous work by Cermak et al. (2012) looking at ADC values 24 h after eccentric exercise revealed increased ADC values, although as for the present study, only in select muscles in the quad (VL following knee extensor exercises). Likewise, Yanagisawa et al. (2011) reported only selective muscles in the calf showing significant ADC changes following ankle plantar flexion. Previous work has demonstrated a close correlation between changes in ADC and T2 (Morvan & Leroy-Willig, 1995) immediately after exercise, with the perceived underlying mechanism for both being changes in intramuscular water content. However, for muscle injuries in mice, examined over longer periods, the time courses of T2 and ADC were not correlated (Heemskerk et al., 2007). When a correlation between ADC and T2 was examined within VI for the present study, no significant correlation was found, suggestive of the two parameters having different underlying mechanisms for the changes that were observed, and with potentially different time scales associated with them. Thus, given that overall, both T2 and ADC resulted in significant time effects in the quadriceps as a whole, combining the two measurements may potentially increase the sensitivity to detecting damage-induced muscle changes particularly if different types of muscledamaging protocols are considered. Previously, magnetization transfer contrast has been successfully utilized to show acute changes in exercising muscle (Yoshioka et al., 1994). Given that MT contrast arises from the interchange of energy between macromolecules and the bulk water proton pool, alterations in the concentrations of either of these would be anticipated to result in detectable EIMD modifications. Thus, amendments in the concentration of macromolecules either via breakdown or accumulation or of the water pool in contact with the macromolecules though edema or hyperemia would all give rise to detectable signal changes. That no significant changes were seen in the present study is either indicative of there being no underlying physiological changes or a failure in the technique

9

Fulford et al. to detect them. One significant problem with MT methods is that, as a result of them being generated via the subtraction of two separate scans, they are extremely sensitive to movement in the time either during or between those scans. In the present study indeed, very noisy results were obtained with the large standard deviation obtained contributing to the lack of significance seen. Thus, there may be value in undertaking further studies utilizing MT methods where the muscle under investigation is under significant restraint. However, it appears to be a technique that has limited potential in the early detection of muscle damage. Previous work (Rodenburg et al., 1995; Davies et al., 2011) has suggested an increased baseline level of Pi : PCr post-EIMD. However, rather than relying on a ratio, in the present study, absolute molarities of the phosphorous metabolites were determined via a calibration procedure, thereby allowing an assessment of whether the Pi : PCr ratio was being influenced by alterations in resting PCr values rather than it being largely dependent on Pi concentration modifications, as is assumed. In the event, no concentration differences were found for PCr following EIMD confirming the validity of assessing alterations in Pi via the PCr : Pi ratio, which is much faster in terms of scanning time than calculating absolute molarities. As previously, significant increases were found for baseline Pi concentrations (and as a result the PCr : Pi ratio) following the muscle damage protocol. Given the reported implications of Pi on force generating ability within the muscle (Westerblad et al., 2002; Debold et al., 2006) potentially via cross-bridge cycling (Fitts, 2008), such an increase may have an important impact on muscle function. The decrease in both jump height and isokinetic peak torque is in line with this theoretical decrease in muscle function. However, no significant differences were seen in the PCr recovery time constants. The hypothesis behind any anticipated decrease in the time constant, which represents oxidative capacity, would be based on either modifications in mitochondrial function or alterations in the oxygen delivery. Previous work has suggested there may be some alterations in local blood flow following EIMD (Davies et al., 2008). That such a finding was not reported in the present study may be as a result of no vascular impairment having resulted or due to the participant characteristics in the present study. Given they were all young and comprised a physically active group, and the exercise only challenged a relatively small muscle mass, any small alterations in maximal O2 delivery might not result in a measureable impact on mitochondrial function. In addition, other MR measures have tended to indicate the focus of any muscle damage is within the VL, VI, and VM muscles with the RF being spared. However, as a result of the position of the participant, lying on top of the coil with it being centered on the quadriceps, and the nonlocalized nature of the technique, a large proportion of the MRS signal will be

10

generated from the RF. The fact that significant changes were seen for Pi concentrations when sampling from an approximately equivalent muscle volume though, suggests that if significant changes were induced in PCr recovery times they would be detectable, despite any bias toward sampling from the RF. Thus, the finding does provide some support to the suggestion that the decrease in performance seen following EIMD does not result from either impaired vascular or mitochondrial function and hence, utilizing techniques to examine those attributes will not be sensitive to detecting muscle damage at 24 h post-event. It should be noted that no gender effects were seen for the MR measures used in the present study and thus all data were pooled into a single group. However, gender differences may have an age or physical activity dependence that could result in a divergence in the MR results found in response to muscle damage, if a study group of different characteristics to those for the present study were examined. Likewise, there is an assumption that the population examined responded in a homogeneous fashion to the damaging protocol, in respect of the MR parameters measured. One option for attempting to normalize for the degree of damage induced, and thus account for any variability in response would be to take the non-MR functional and performance measures as individual damage scores. However, given the multicomponent nature of EIMD, a direct relationship between any two outcome measures cannot be assumed. Of all of the measures undertaken, those that gave statistically significant changes 24 h after EIMD included well-established techniques such as T2 values and cross-sectional areas/volumes. However, less commonly used methodologies such as ADC coefficients and Pi concentration determinations also provided insightful indications of muscle alterations resultant from muscledamaging exercise. In contrast, measures of oxidative metabolism and magnetization transfer did not reveal significantly measureable effects. Although muscle damage has been demonstrated to be a multicomponent process, each aspect evolving with a different time course and different exercise modalities may cause different forms and degree of EIMD, the present study nevertheless gives a good indication of those techniques which are most likely to detect muscle damage at an early stage after the potentially damaging procedure. Future work would aim to extend the time course the variables were measured over, in particular, assessing the correlation between different measures over multiple time points to examine the potential interaction between them and thereby provide further mechanistic insight. Perspectives Early indications of exercise-induced muscle damage have the potential to allow prompt intervention and management, so as to minimize any detrimental performance

MR techniques and muscle damage detection impact. Of the MR techniques available, the present study reveals significant effects can be observed at 24 h post-damage, via T2, ADC, muscle volume, and Pi measurements. That there are no significant correlations seen between changes in T2, ADC and Pi, illustrates the multicomponent nature of muscle damage and the importance of adopting a multiple technique approach to accurately detect its presence. It also highlights the potential of MR to provide insight into the mechanistic basis of any damage arising, as well as illustrating its capability to provide highly localized

information. Such capacity means that much greater precision in determining the site of muscle damage is possible, in conjunction with the increased sensitivity this leads to, relative to whole body or large muscle mass markers of damage. Key words: MRI, MRS, EIMD, methods.

Acknowledgements Jonathan Fulford’s salary was supported via an NIHR grant.

References Armstrong RB. Mechanisms of exercise-induced delayed onset muscular soreness: a brief review. Med Sci Sports Exerc 1984: 16: 529–538. Armstrong RB, Ogilvie RW, Schwane JA. Eccentric exercise-induced injury to rat skeletal muscle. J Appl Physiol 1983: 54: 80–93. Basmajian JV. Primary anatomy. 7th edn. Baltimore: Williams & Wilkins, 1976. Bobbert MF, Hollander AP, Huijing PA. Factors in delayed onset muscular soreness of man. Med Sci Sports Exerc 1986: 18: 75–81. Bosco C, Luhtanen P, Komi PV. A simple method for measurement of mechanical power in jumping. Eur J Appl Physiol 1983: 50: 273–282. Byrne C, Eston R. Maximal-intensity isometric and dynamic exercise performance after eccentric muscle actions. J Sports Sci 2002: 20: 951–959. Byrne C, Twist C, Eston RG. Neuromuscular function after exercise-induced muscle damage: theoretical and applied implications. Sports Med 2004: 34: 49–69. Cermak NM, Noseworthy MD, Bourgeois JM, Tarnopolsky MA, Gibala MJ. Diffusion tensor MRI to assess skeletal muscle disruption following eccentric exercise. Muscle Nerve 2012: 46: 42–50. Clarkson PM, Hubal MJ. Exercise-induced muscle damage in humans. Am J Phys Med Rehabil 2002: 81: S52–S69. Davies RC, Eston RG, Fulford J, Rowlands AV, Jones AM. Muscle damage alters the metabolic response to dynamic exercise in humans: a 31P-MRS study. J Appl Physiol 2011: 111: 782–790. Davies RC, Eston RG, Poole DC, Rowlands AV, Dimenna F, Wilkerson DP, Twist C, Jones AM. The effect of eccentric exercise-induced muscle damage on the dynamics of muscle oxygenation and pulmonary oxygen

uptake. J Appl Physiol 2008: 105: 1413–1421. Davies RC, Rowlands AV, Eston RG. Effect of exercise-induced muscle damage on ventilatory and perceived exertion responses to moderate and severe intensity cycle exercise. Eur J Appl Physiol 2009: 107: 11–19. Debold EP, Romatowski J, Fitts RH. The depressive effect of Pi on the force calcium relationship in skinned single muscle fibres is temperature dependent. Am J Physiol Cell Physiol 2006: 290: C1041–C1050. Fitts RH. The cross-bridge cycle and skeletal muscle fatigue. J Appl Physiol 2008: 104: 551–558. Foley JM, Roop C, Prior BM, Pivarnik JM, Meyer RA. MR measurements of muscle damage and adaptation after eccentric exercise. J Appl Physiol 1999: 87: 2311–2318. Fukunaga T, Roy RR, Shellock FG, Hodgson JA, Day MK, Lee PL, Kwong-Fu H, Edgerton VR. Physiological cross-sectional area of human leg muscles based on magnetic resonance imaging. J Orthop Res 1992: 10: 928–934. Haaverstad R, Nilsen G, Myhre HO, Saether OD, Rinck PA. The use of MRI in the investigation of leg oedema. Eur J Vasc Surg 1992: 6: 124–129. Hazlewood CF, Rorschach HE, Lin C. Diffusion of water in tissues and MRI. Magn Reson Med 1991: 19: 214–216. Heemskerk AM, Strijkers GJ, Drost MR, van Bochove GS, Nicolay K. Skeletal muscle degeneration and regeneration after femoral artery ligation in mice: monitoring with diffusion MR imaging. Radiology 2007: 243: 413–421. Jubrias SA, Crowther GJ, Shankland EG, Gronka RK, Conley KE. Acidosis inhibits oxidative phosphorylation in contracting human skeletal muscle in vivo. J Physiol 2003: 553: 589–599. Kemp GJ, Meyerspeer M, Moser E. Absolute quantification of phosphorus metabolite concentrations in human

muscle in vivo by 31P MRS: a quantitative review. NMR Biomed 2007: 20: 555–565. Kemp GJ, Taylor DJ, Radda GK. Control of phosphocreatine resynthesis during recovery from exercise in human skeletal muscle. NMR Biomed 1993: 6: 66–72. Kim HK, Laor T, Horn PS, Racadio JM, Wong B, Dardzinski BJ. T2 mapping in Duchenne muscular dystrophy: distribution of disease activity and correlation with clinical assessments. Radiology 2010: 255: 899–908. Larsen RG, Ringgaard S, Overgaard K. Localization and quantification of muscle damage by magnetic resonance imaging following step exercise in young women. Scand J Med Sci Sports 2007: 17: 76–83. Le Bihan D. Molecular diffusion nuclear magnetic resonance imaging. Magn Reson Q 1991: 7: 1–30. LeBlanc A, Lin C, Shackelford L, Sinitsyn V, Evans H, Belichenko O, Schenkman B, Kozlovskaya I, Oganov V, Bakulin A, Hedrick T, Feeback D. Muscle volume, MRI relaxation times (T2), and body composition after spaceflight. J Appl Physiol 2000: 89: 2158–2164. LeBlanc AD, Jaweed M, Evans H. Evaluation of muscle injury using magnetic resonance imaging. Clin J Sport Med 1993: 3: 26–30. Lund H, Vestergaard-Poulsen P, Kanstrup IL, Sejrsen P. The effect of passive stretching on delayed onset muscle soreness, and other detrimental effects following eccentric exercise. Scand J Med Sci Sports 1998a: 8: 216–221. Lund H, Vestergaard-Poulsen P, Kanstrup IL, Sejrsen P. Isokinetic eccentric exercise as a model to induce and reproduce pathophysiological alterations related to delayed onset muscle soreness. Scand J Med Sci Sports 1998b: 8: 208–215. MacIntyre DL, Reid WD, Lyster DM, Szasz IJ, McKenzie DC. Presence of WBC, decreased strength, and delayed

11

Fulford et al. soreness in muscle after eccentric exercise. J Appl Physiol 1996: 80: 1006–1013. Marginson V, Rowlands AV, Gleeson NP, Eston RG. A comparison of the symptoms of exercise-induced muscle damage following an initial and repeated bout of eccentric exercise in men and boys. J Appl Physiol 2005: 99: 1174–1181. Marqueste T, Giannesini B, Le Fur Y, Cozzone PJ, Bendahan D. Comparative MRI analysis of T2 changes associated with single and repeated bouts of downhill running leading to eccentric-induced muscle damage. J Appl Physiol 2008: 105: 299–307. Mathur S, Vohra RS, Germain SA, Forbes S, Bryant ND, Vandenborne K, Walter GA. Changes in muscle T2 and tissue damage after downhill running in mdx mice. Muscle Nerve 2011: 43: 878–886. McDaniel JD, Ulmer JL, Prost RW, Franczak MB, Jaradeh S, Hamilton CA, Mark LP. Magnetization transfer imaging of skeletal muscle in autosomal recessive limb girdle muscular dystrophy. J Comput Assist Tomogr 1999: 23: 609–614. McMillan AB, Shi D, Pratt SJP, Lovering RM. Diffusion tensor MRI to assess damage in healthy and dystrophic skeletal muscle after lengthening contractions. J Biomed Biotechnol 2011. doi: 10.1155/2011/970726 Meyer RA, Prior BM. Functional magnetic resonance imaging of muscle. Exerc Sport Sci Rev 2000: 28: 89–92. Morvan D, Leroy-Willig A. Simultaneous measurements of diffusion and transverse relaxation in exercising skeletal muscle. Magn Reson Imaging 1995: 13: 943–948. Nurenberg P, Giddings CJ, Stray-Gundersen J, Fleckenstein JL, Gonyea WJ, Peshock RM. MR imaging-guided muscle biopsy for correlation of increased signal intensity with ultrastructural change and delayed-onset muscle soreness after exercise. Radiology 1992: 184: 865–869.

12

Ploutz-Snyder LL, Clark BC, Logan L, Turk M. Evaluation of spastic muscle in stroke survivors using magnetic resonance imaging and resistance to passive motion. Arch Phys Med Rehabil 2006: 87: 1636–1642. Rodenburg JB, de Boer RW, Schiereck P, van Echteld CJ, Bar PR. Changes in phosphorous compounds and water content in skeletal muscle due to eccentric exercise. Eur J Appl Physiol Occup Physiol 1994: 68: 205–213. Rodenburg JB, Degroot MCH, Vanechteld CJA, Jongsma HJ, Bar PR. Phosphate-metabolism of prior eccentrically loaded vastus medialis muscle during exercise in humans. Acta Physiol Scand 1995: 153: 97–108. Shellock FG, Fukunaga T, Mink JH, Edgerton VR. Exertional muscle injury: evaluation of concentric versus eccentric actions with serial MR imaging. Radiology 1991: 179: 659–664. Stupka N, Tarnopolsky MA, Yardley NJ, Phillips SM. Cellular adaptation to repeated eccentric exercise-induced muscle damage. J Appl Physiol 2001: 91: 1669–1678. Szasz G, Gruber W, Bernt E. Creatine-kinase in serum: 1. determination of optimum reaction conditions. Clin Chem 1976: 22: 650–656. Takahashi H, Kuno S, Miyamoto T, Yoshioka H, Inaki M, Akima H, Katsuta S, Anno I, Itai Y. Changes in magnetic resonance images in human skeletal muscle after eccentric exercise. Eur J Appl Physiol 1994: 69: 408–413. Taylor DJ, Bore PJ, Styles P, Gadian DG, Radda GK. Bioenergetics of intact human muscle. A 31P nuclear magnetic resonance study. Mol Biol Med 1983: 1: 77–94. Tothill P, Stewart AD. Estimation of thigh muscle and adipose tissue volume using magnetic resonance imaging and anthropometry. J Sports Sci 2002: 20: 563–576. van den Broek NMA, De Feyter HMML, Graaf LD, Nicolay K, Prompers JJ.

Intersubject differences in the effect of acidosis on phosphocreatine recovery kinetics in muscle after exercise are due to differences in proton efflux rates. Am J Physiol 2007: 293: C228–C237. Vanhamme L, van den Boogaart A, Van Huffel S. Improved method for accurate and efficient quantification of MRS data with use of prior knowledge. J Magn Reson 1997: 129: 35–43. Vanhatalo A, Fulford J, Bailey SJ, Blackwell JR, Winyard PG, Jones AM. Dietary nitrate reduces muscle metabolic perturbation and improves exercise tolerance in hypoxia. J Physiol 2011: 589: 5517–5528. Westerblad H, Allen DG, Lännergren J. Muscle fatigue: lactic acid or inorganic phosphate the major cause? News Physiol Sci 2002: 17: 17–21. Whitehead NP, Allen TJ, Morgan DL, Proske U. Damage to human muscle from eccentric exercise after training with concentric exercise. J Physiol 1998: 512: 615–620. Yanagisawa O, Fukubayashi T. Diffusion-weighted magnetic resonance imaging reveals the effects of different cooling temperatures on the diffusion of water molecules and perfusion within human skeletal muscle. Clin Radiol 2010: 65: 874–880. Yanagisawa O, Kurihara T, Kobayashi N, Fukubayashi T. Strenuous resistance exercise effects on magnetic resonance diffusion parameters and muscle-tendon function in human skeletal muscle. J Magn Reson Imaging 2011: 34: 887–894. Yanagisawa O, Niitsu M, Takahashi H, Goto K, Itai Y. Evaluations of cooling exercised muscle with MR imaging and 31 P MR spectroscopy. Med Sci Sports Exerc 2003: 35: 1517–1523. Yoshioka H, Takahashi H, Onaya H, Anno I, Niitsu M, Itai Y. Acute change of exercised muscle using magnetization transfer contrast MR imaging. Magn Reson Imaging 1994: 12: 991–997.