Cooling has been widely utilized for the treatment of acute musculoskeletal injuries. In acute treatment, it is very important to take measures against the secondary hypoxic injury subsequent to the initial trauma (10,14,22). Cooling has been suggested to cause a decrease in cell metabolism by decreasing the skeletal muscle’s temperature and consequently allow the uninjured cells around the initial trauma to survive the period of hypoxia (6,10,14,22). In addition, cooling has been commonly utilized as an effective therapeutic modality after various sports activities. Increased H+ concentration due to strenuous muscle activity leads to decreased intracellular pH (acidosis) in the skeletal muscle (1,9,15,19,23,26), which may result in a decrease in the sports performance or a delay in the recovery of skeletal muscle after exercise. It was demonstrated that the intracellular pH increased with decreasing temperature due to the cooling (8,18,21,29). However, most investigators have demonstrated that cooling resulted in a decrease in cell metabolism or an increase in intracellular pH using experimental animals. Yoshioka et al. (29) first demonstrated the relation between increased intracellular pH and decreased skeletal muscle temperature in human subjects, but not after exercise for these skeletal muscles. In sports science and medicine, therefore, it is strongly desired that the effects of cooling on the intracellular environment be investigated noninvasively and quantitatively in human skeletal muscle after exercise.
31P MR spectroscopy has been used extensively to investigate energy metabolism in human skeletal muscle. In particular, relative concentrations of adenosine triphosphate (ATP), inorganic phosphate (Pi), phosphocreatine (PCr), and intracellular pH have been utilized for noninvasively evaluating oxidative and glycolytic metabolism in skeletal muscle during and after exercise (1,9,13,15,19,23,26). Therefore, 31P MR spectroscopy should be suitable for evaluating the effects of cooling on the intracellular environment in human skeletal muscle after exercise. To the best of our knowledge, there has been no study that tried to investigate, using 31P MR spectroscopy, the effects of cooling on cell metabolism in human skeletal muscle after exercise.
On the other hand, cooling has been shown to decrease muscle edema formation and muscle cell damage (2,10,14,22). Eccentric exercise is apt to educe greater damage to the muscle fibers and is sometimes accompanied by secondary muscle edema (5). It has been demonstrated using T2 relaxation time (T2) in MR imaging that delayed muscle edema occurs after strenuous exercise including eccentric muscular contraction (3,4,12,17,19,24). T2 value has been suggested to reflect not only the intramuscular water content but also the extent of muscle injury after exercise (3,4,12,17,19,24). Thus, we applied MR imaging to help clarify whether cooling relieves the extent of the muscle damage and edema after strenuous exercise. In addition, cooling was reported to have positive effects on decreasing muscle pain and spasms (22), but further investigations are necessary to examine the delayed onset muscle soreness (DOMS), which frequently occurs after eccentric exercise.
In this study, we intended to investigate, using 31P MR spectroscopy and MR imaging, the effects of cooling on the intracellular environment, intramuscular water content, and muscle fiber damage with DOMS in human skeletal muscle after strenuous exercise.
MATERIALS AND METHODS
Fourteen healthy untrained male subjects volunteered for this study and were randomly divided into two groups (control group: seven subjects; mean ± SD, age 23.9 ± 2.8 yr, height 172.4 ± 5.7 cm, weight 62.7 ± 5.8 kg; cooling group: seven subjects; mean ± SD, age 23.7 ± 1.9 yr, height 171 ± 2.8 cm, weight 67.4 ± 5.9 kg). Each subject was instructed to refrain from participating in any physical exercise and performing private physical therapeutic treatment throughout this study. Informed written consent was obtained from all the subjects. In addition, this study was approved by the Ethical Committee of the University of Tsukuba.
Cooling by cold-water immersion.
The 15-min cold-water immersion was given immediately after the completion of the exercise and initial spectral measurements in the cooling group. The cooling consisted of submerging the subjects’ legs into a cold-water immersion tank cooled to 5°C. The temperature of the tank was monitored using a thermometer and adjusted accordingly during the application to maintain the desired temperature. The legs were immersed so that the head of the fibula was completely covered. Ankle supporters were worn to protect the dorsum of the feet from cold hypersensitivity.
The subjects performed the ankle plantar flexion exercise concentrically and eccentrically using a specially designed rig by shouldering the bar with a weight, straightening the lower extremity at the knee joint, paralleling both sides of the toe, putting the metatarsal bone on the edge of the hole made in the floor of the rig (Fig. 1). The hole was designed to elicit the full eccentric contractions of the triceps surae muscle. Before exercise, all subjects were tested for maximal isotonic contraction (MIC) in concentric mode for the ankle plantar flexion using free weights. All subjects performed the ankle plantar flexion exercise using a constant load (kg) equal to 30% of each subject’s MIC throughout the full joint range of motion. This exercise consisted of five sets of 12 repetitions and 1-min rest between sets. In addition, this exercise was performed to a cadence of 60 counts per minute using the metronome. On count 1 the subject raised the weight; on counts 2 and 3, the subject lowered the weight. This type of exercise forces the triceps surae muscle to undergo full eccentric contractions. The exercise was performed at the side of the MR imaging room.
31P MR spectroscopy.
The 31P-spectra were collected from the medial gastrocnemius (the closest area to the subcutaneous fat) using a 1.5-T superconducting MR imaging system (Gyroscan Intera, Philips Medical Systems, The Netherlands) with a 6-cm-diameter surface coil. The subjects assumed the supine position, and the coil was fixed firmly on the belly of the medial gastrocnemius using elastic tape. Magnetic field homogeneity was optimized by shimming on the proton signal, and phosphorus signals were collected with an optical pulse to produce the maximal signal intensity per pulse. The shimming at immediate measurements was skipped to perform the spectral measurements quickly, but the subjects lay on similar positions as for the measurement at rest using the markers written on both the scan table and each subject’s skin. The acquisition parameters were: repetition time = 3000 ms; spectral bandwidth = 2000 Hz; number of data point = 2048; 96 averages. Acquisition time was 5 min. Spectral measurements were performed before and immediately, 30, 60 min, 24, 48, 96, and 168 h after exercise. The collected spectra were multiplied by an exponential of the decay, which was chosen to correspond to the 5-Hz line broadening before fast Fourier transformation. For quantification of energy metabolism, the ratio of inorganic phosphate to phosphocreatine (Pi/PCr) was determined by integrating each peak area. Muscle intracellular pH was calculated from the chemical shift of Pi relative to PCr using the following equation (25):MATH
where s represents the chemical shift of Pi compared with PCr in parts per million. The same person performed both the measurement and the analysis for 31P MR spectroscopy.
Transaxial T2-weighted MR images were also obtained using the same MR imaging system with the body coil in the supine position before and 24, 48, 96, and 168 h after exercise. The MR sequence was as follows: spin echo technique: repetition time/echo time = 3000 ms/20, 40, 60, 80 ms; 256 × 256 matrix; two excitations: 300-mm field of view; 10-mm slice thickness; a single slice. Acquisition time was 3 min 14 s; scan position was at 75% (proximal side) of the length between the head of the fibula and the lateral malleolus on the dominant leg throughout this study and was marked on each subject’s skin with semipermanent ink to ensure that images and spectra were collected at the same site at each set time. The region of interest (ROI) was set by tracing along the inner surface of the medial head of the gastrocnemius muscle on T2-weighted images at each measurement time. Special care was taken to avoid inclusion of subcutaneous fat, fascia, blood vessels, or bony anatomy into ROI. The T2 values at each time were determined for the ROI using software provided by Philips Medical Systems. Calculation of T2 was carried out three times, and an averaged value was adopted as representative of T2. The same person performed both the MR imaging scan and the T2 calculation.
Perception of muscle soreness of the calf was assessed using a questionnaire on a scale of one (normal) to five (very, very sore) before and immediately, 30, 60 min, 24, 48, 96, and 168 h after exercise. The perception when the subjects walked was assessed on the dominant side.
Means and standard deviations (±SD) were calculated for all measurements. The changes in all measurements over time were analyzed with a repeated measures ANOVA. Tukey post hoc test was used to confirm the results by ANOVA. The unpaired t-test was used to determine the differences in response between the two groups at each measurement time. In addition, correlations among variables (T2 vs intracellular pH, vs Pi/PCr ratio, vs muscle soreness level; intracellular pH vs Pi/PCr ratio, vs muscle soreness level; Pi/PCr ratio vs muscle soreness level) were assessed by determining Pearson’s correlation coefficient. P < 0.05 was considered significant in all analyses.
T2 relaxation time.
The time courses of the T2 change in the medial gastrocnemius are shown in Figure 2. There was no significant difference in the T2 value at rest between the two groups. The control group showed significant T2 increase at 48 h after exercise, but the cooling group showed no significant T2 change over the course of this study. No significant difference was observed between the two groups at each measurement time. In addition, Figure 3 shows the T2-weighted images of one typical subject’s triceps surae muscle in the control or cooling group before and at 48 h postexercise, respectively. The control group showed elevated signal intensity in the medial gastrocnemius at 48 h after exercise, but there was little change after exercise in the cooling group.
The time courses of intracellular pH are shown in Figure 4. No significant difference between the two groups was observed at rest. Two groups showed significantly decreased intracellular pH in the medial gastrocnemius immediately after exercise. After that, the cooling group showed a significantly increased value at 60 min after exercise. For comparisons between the two groups at each measurement time, the cooling group showed significantly greater values than the control group at 30 and 60 min after exercise.
Figure 5 displays the time courses of Pi/PCr change of the medial gastrocnemius in both groups. There was no significant difference in the Pi/PCr value at rest between both groups. No significant change was observed until the end of measurements in the two groups.
The time courses of the muscle soreness level are shown in Figure 6. Significant changes in the muscle soreness level in both groups were found immediately and 24–48 h after exercise. For comparisons between two groups, no significant difference was observed throughout this study.
There was no significant correlation among all variables (T2 vs intracellular pH, vs Pi/PCr ratio, vs muscle soreness level; intracellular pH vs Pi/PCr ratio, vs muscle soreness level; Pi/PCr ratio vs muscle soreness level) throughout this study.
The present study showed a significant T2 elevation at 48 h postexercise in the control group. In addition, increased signal intensity on T2-weighted image was observed in the medial gastrocnemius in the control group. These MR findings are nearly consistent with those of previous study, which used a similar exercise with MR imaging (4). It has been suggested that postexercise T2 elevation is well associated with an increase in intramuscular water (1,3,4,11,12,17,19,20,24,28), though the detailed mechanism remains unclear. Above all, eccentric exercise causes an elevation in the T2 value and/or signal intensity on T2-weighted images at 48–168 h postexercise, which may be partly attributed to the muscle damage due to the eccentric exercise (3,4,12,17,19,20,24). Several studies reported that the T2 elevation at a few days after eccentric exercise should be related to the delayed increases in muscle enzymes such as serum creatine kinase (CK) (3,4,12,17, 19,20,24) or in plasma concentrations of myosin heavy chain fragments that indicate the degradation of myofilaments and leakage of the plasma membrane by eccentric exercise (12,20). It is likely that this delayed muscle edema resulted from the changed osmotic pressure between the vascular space and the interstitium due to the increased muscle enzymes and/or degraded protein components, and partly reflects the ultrastructural muscle fiber damage.
It was suggested that excessive edema may make the transport route for the oxygen to the cells longer and also compress the capillaries by its high extracellular pressure, which may cause further cell damage (5,22). Thus, it is interesting that the cooling group showed no significant T2 elevation throughout this study. Eston and Peters (2) demonstrated the effectiveness of cooling on relieving CK activity, but the detailed mechanism is unclear. They suggested that the amount or the rate of the damaged muscle cells might be reduced by cooling. It was also suggested that the cooling decreases the cell metabolism, which should contribute to protect the uninjured cells against secondary hypoxic death due to the ischemia and/or secondary enzymatic damage due to the dying and inflammatory cells (10,14,22).
Several studies reported a decreased PCr/(PCr+Pi) (increased Pi/PCr) ratio and/or decreased intracellular pH that resulted from exercise (1,9,13,15,19,23,26). However, the present study found no significant change immediately after exercise in the Pi/PCr ratio in either group. During recovery after exercise, PCr is resynthesized immediately through a CK reaction (23). It was demonstrated that exercise-induced PCr/(PCr+Pi) change reverted to the rest value at 0.75–1.4 min (9). Because the exercise was not performed in the bore of the MR imaging system, at least a few minutes passed away until the start of the spectral measurements. Presumably this time loss affected the findings of the present study. On the other hand, this study showed significantly decreased intracellular pH immediately after exercise, suggesting muscle acidosis. It was suggested that the acidosis is well related to the lactate acid and muscle [H+] produced by exercise (1,15,23). Judging from the finding that there was no significant difference between the two groups concerning the postexercise Pi/PCr ratio and intracellular pH, it is quite likely that the postexercise cellular environments were similar in both groups.
Myrer et al. (16) reported that the 20-min ice bag applied to the triceps surae muscle significantly decreased the intramuscular temperature at 1 cm below the subcutaneous fat by approximately 7°C. Zemke et al. (30) also demonstrated that an ice massage or the ice bag induced −4.3°C or −2.3°C change from the resting temperature to the intramuscular temperature at 1 cm into the medial aspect of the gastrocnemius at the end of a 15-min cooling period. We obtained 31P-spectra from the most superficial area in the medial gastrocnemius and confirmed from MR images that there was no subject with excessive subcutaneous fat around the calf. Judging from these findings, it is suggested that the effect of cooling reached the region of interest for 31P-spectra in the medial gastrocnemius. Although no significantly positive changes after cooling were observed in the Pi/PCr ratio or muscle soreness level in the cooling group throughout this study, intracellular pH showed a significant increase at 60 min postexercise compared with the value at rest. In addition, the cooling group showed significantly greater pH values than the control group at 30 and 60 min after exercise. H+ washout due to the increased blood flow may contribute to the inhibition of decreased intracellular pH (23). Ho et al. (6) observed a paradoxical increase in arterial blood flow after a 10-min ice wrap to the knee joint and suggested a possible reflex vasodilation in the arterial blood vessels in response to cooling. However, the present study did not examine the blood flow change, and thus it is not clear whether the blood flow increased by cooling. In addition, several studies demonstrated an increase in intracellular pH with decreasing temperature of the skeletal muscle using laboratory animals (8,18,21). Stinner and Hartzler (21) suggested that based on the change in the intracellular ion concentration with a decrease in the muscle temperature, the ion-exchange mechanism may play an important role for adjusting intracellular pH with changes in temperature. It is most likely that the cooling had positive effects on the intracellular buffering systems or the cellular H+ extrusion mechanisms responding to the lactic acid produced during exercise. Yoshioka et al. (29) first demonstrated a significantly inverse relation between the temperature and intracellular pH in human skeletal muscle and this slope (−0.005 pH units·°C−1) was smaller than those of animal skeletal muscle, suggesting the presence of high intracellular buffering systems in human skeletal muscle. However, no studies have investigated the relation between the temperature and intracellular pH in human skeletal muscle cooled after exercise. Therefore, further investigations will be necessary to clarify the mechanism.
Several studies reported that the Pi/PCr ratio remained above the preexercise value for several days after strenuous eccentric exercise (13,19). McCully et al. (13) observed that patients with various destructive neuromuscular diseases had an elevated Pi/PCr ratio and serum CK activity above normal levels at rest, and found that the increased Pi/PCr ratio in these patients was similar to those in normal subjects seen at a few days after eccentric exercise. Therefore, they suggested that the increased Pi/PCr ratio seen after strenuous exercise could reflect an elevated cellular metabolism to repair the small tears and breaks in the myofilaments of injured muscle, an increased ion transport pumping activity to compensate for leaky membranes, or mitochondrial alterations due to the muscle cell damage. However, the present study showed no change in the Pi/PCr ratio in either group after exercise. Judging from the muscle soreness level in the present study, it is possible that this exercise had less severe intensity and consequently resulted in less muscle damage compared with previous studies (13,19).
The subjects in the present study complained of DOMS in their calves, especially at the proximal insertion site of the medial gastrocnemius, after exercise. McCully et al. (13) observed a significantly increased Pi/PCr ratio above the resting value in the presence of DOMS after eccentric exercise. Rodenburg et al. (19) also reported the increased Pi/PCr ratio followed the same time course as the occurrence of DOMS. However, the findings of the present study were not in agreement with those of the previous studies (13,19). Some patients with muscle fiber injury have an increased Pi/PCr ratio at rest without muscle soreness (13). Therefore, the relation between the increased Pi/PCr ratio and DOMS after exercise is unclear. In previous studies, it was demonstrated that T2 increase did not correlate significantly with DOMS (3,12). On the other hand, some studies suggested a positive correlation between DOMS and T2 increase (4,19,24). Fridén et al. (5) reported that muscle fiber swelling as a result of myofibrillar damage was a major factor of DOMS. Perhaps an elevated intramuscular pressure due to the edema may stimulate muscle soreness receptors. However, both groups in this study showed no significant correlation between T2 change and the muscle soreness level throughout this study. Fleckenstein et al. (4) reported that a positive correlation exists between T2 and DOMS in the exercised muscle, but T2 continued to be elevated for approximately 2 wk despite the cessation of symptoms of DOMS. Therefore, some factors other than muscle edema may be associated with DOMS in this study.
Cooling has been shown to have an analgesic effect on muscle soreness, muscle spasm, or other injury-related pain (22). However, several studies (2,7,27) in addition to the present findings showed that the cooling alone was not effective in reducing the symptoms of DOMS. In this study, the cooling group showed a decreased tendency in the muscle soreness level at 24–48 h compared with the control group, but there was no significant difference between them. Although the cooling may result in an increase in the pain threshold with resulting pain relief immediately after its application, there may be no positive effect on suppressing the several factors that produced DOMS.
In conclusion, the findings of this study suggest that the cooling prevents the delayed muscle edema and decreases the extent of muscle cell damage resulting from exercise. In addition, the cooling appears to have a positive effect on intracellular pH after exercise. These findings should be useful for athletic trainers to manage an athlete’s physical condition.
The authors would like to thank the medical staff in the Department of Radiology, University of Tsukuba, for their helpful comments and assistance. Our sincere thanks are extended to Drs. Yoshichika Yoshioka, Iwate Medical University for helpful comments. Finally, the authors appreciate the University of Tsukuba Research Project Grant that supported this study.
1. Cheng, H. A., R. A. Robergs, J. P. Letellier, A. Caprihan, M. V. Icenogle, and L. J. Haseler. Changes in muscle proton transverse relaxation times and acidosis during exercise and recovery. J. Appl. Physiol. 79: 1370–1378, 1995.
2. Eston, R., and D. Peters. Effects of cold water immersion on the symptoms of exercise-induced muscle damage. J. Sport. Sci. 17: 231–238, 1999.
3. Evans, G. F. F., R. G. Haller, P. S. Wyrick, R. W. Parkey, and J. L. Fleckenstein. Submaximal delayed-onset muscle soreness: correlations between MR imaging findings and clinical measures. Radiology 208: 815–820, 1998.
4. Fleckenstein, J. L., P. T. Weatherall, R. W. Parkey, J. A. Payne, and R. M. Peshock. Sports-related muscle injuries: evaluation with MR imaging. Radiology 172: 793–798, 1989.
5. Fridén, J., P. N. Sfakianos, A. R. Hargens, and W. H. Akeson. Residual muscular swelling after repetitive eccentric contractions. J. Orthopaed. Res. 6: 493–498, 1988.
6. Ho, S. S. W., R. L. Illgen, R. W. Meyer, P. J. Torok, M. D. Cooper, and B. Reider. Comparison of various icing times in decreasing bone metabolism and blood flow in the knee. Am. J. Sports Med. 23: 74–76, 1995.
7. Isabell, W. K., E. Durrant, W. Myrer, and S. Anderson. The effects of ice massage, ice massage with exercise, and exercise on the prevention and treatment of delayed onset muscle soreness
. J. Athl. Train. 27: 208–217, 1992.
8. Johnson, D. C., C. T. Burt, W. Perng, and B. M. Hitzig. Effects of temperature on muscle pHi
and phosphate metabolites in newts and lungless salamanders. Am. J. Physiol. 265: R1162–R1167, 1993.
9. Kemp, G. J., D. J. Taylor, G. K. Radda, and B. Bajagopalan. Bio-energetic changes in human gastrocnemius muscle 1–2 days after strenuous exercise. Acta Physiol. Scand. 146: 11–14, 1992.
10. Knight, K. L. Effects of hypothermia on inflammation and swelling. J. Athl. Train. 11: 7–10, 1976.
11. Le Rumeur, E., J. De Certaines, P. Toulouse, and P. Rochcongar. Water phases in rat striated muscles as determined by T2
proton NMR relaxation times. Magn. Reson. Imaging 5: 267–272, 1987.
12. Mair, J., A. Koller, E. Artner-Dworzak, et al. Effects of exercise on plasma myosin heavy chain fragments and MRI of skeletal muscle. J. Appl. Physiol. 72: 656–663, 1992.
13. Mccully, K. K., Z. Argov, B. P. Boden, R. L. Brown, W. J. Bank, and B. C. Chance. Detective of muscle injury in humans with 31-P magnetic resonance spectroscopy. Muscle Nerve 11: 212–216, 1988.
14. Merrick, M. A., J. M. Rankin, F. A. Andres, and C. L. Hinman. A preliminary examination of cryotherapy and secondly injury in skeletal muscle. Med. Sci. Sports Exerc. 31: 1516–1521, 1999.
15. Morikawa, S., T. Inubushi, K. Kito, and R. Tabata. Imaging of phosphoenergetic state and intracellular pH
in human calf muscles after exercise by 31
P NMR spectroscopy. Magn. Reson. Imaging 12: 1121–1126, 1994.
16. Myrer, J. W., G. Measom, E. Durrant, and G. W. Fellingham. Cold- and hot-pack contrast therapy: subcutaneous and intramuscular temperature change. J. Athl. Train. 32: 238–241, 1997.
17. Nosaka, K., and P. M. Clarkson. Variability in serum creatine kinase response after eccentric exercise
of the elbow flexors. Int. J. Sports Med. 17: 120–127, 1996.
18. Osterman, A. L., R. B. Heppenstall, A. A. Sapega, M. Katz, B. Chance, and D. Sokolow. Muscle ischemia and hypothermia: a bioenergetic study using 31
phosphorus nuclear magnetic resonance spectroscopy. J. Trauma 24: 811–817, 1984.
19. Rodenburg, J. B., R. W. De Boer, P. Schiereck, C. J. A. Van Echteld, and P. R. Bär. Changes in phosphorus compounds and water content in skeletal muscle due to eccentric exercise
. Eur. J. Appl. Physiol. 68: 205–213, 1994.
20. Sorichter, S., J. Mair, A. Koller, et al. Creatine kinase, myosin heavy chains and magnetic resonance imaging after eccentric exercise
. J. Sport. Sci. 19: 687–691, 2001.
21. Stinner, J. N., and L. K. Hartzler. Effect of temperature on pH and electrolyte concentration in air-breathing ectotherms. J. Exp. Biol. 203: 2065–2074, 2000.
22. Swenson, C., L. Swärd, and J. Karlsson. Cryotherapy in sports medicine. Scand. J. Med. Sci. Sports 6: 193–200, 1996.
23. Takahashi, H., M. Inaki, K. Fujimoto, et al. Control of the rate of phosphocreatine resynthesis after exercise in trained and untrained human quadriceps muscles. Eur. J. Appl. Physiol. 71: 396–404, 1995.
24. Takahashi, H., S. Kuno, T. Miyamoto, et al. Changes in magnetic resonance image in human skeletal muscle after eccentric exercise
. Eur. J. Appl. Physiol. 69: 408–413, 1994.
25. Taylor, D. J., J. Borep, P. Styles, D. G. Gadian, and G. K. Radda. Bioenergetics of intact human muscle: a 31
P nuclear magnetic resonance study. Mol. Biol. Med. 1: 77–94, 1983.
26. Vandenborne, K., G. Walter, L. Ploutz-Snyder, G. Dudley, M. A. Elliott, and K. De Meirleir. Relationship between muscle T2*
relaxation properties and metabolic state: a combined localized 31
P-spectroscopy and 1
H-imaging study. Eur. J. Appl. Physiol. 82: 76–82, 2000.
27. Yackzan, L., C. Adams, and K. T. Francis. The effects of ice massage on delayed muscle soreness. Am. J. Sports Med. 12: 159–165, 1984.
28. Yoshioka, H., and Y. Itai. Magnetic resonance imaging of isolated skeletal muscles: osmotic influence. Invest. Radiol. 31: 359–363, 1996.
29. Yoshioka, Y., H. Oikawa, S. Ehara, et al. Noninvasive estimation of temperature and pH in human lower leg muscles using 1
H nuclear magnetic resonance spectroscopy. Spectroscopy 16: 183–190, 2002.
30. Zemke, J. E., J. C. Andersen, W. K. Guion, J. Mcmillan, and A. B. Joyner. Intramuscular temperature responses in the human leg to two forms of cryotherapy: ice massage and ice bag. J. Orthop. Sports Phys. Ther. 27: 301–307, 1998.