Magn Reson Mater Phy DOI 10.1007/s10334-014-0461-4
RESEARCH ARTICLE
Characterizing blood oxygen level-dependent (BOLD) response following in-magnet quadriceps exercise Jessica E. Caterini • Alyaa H. Elzibak • Emilie Jean St. Michel Brian W. McCrindle • Andrew N. Redington • Sara Thompson Michael D. Noseworthy • Greg D. Wells
• •
Received: 6 May 2014 / Revised: 11 August 2014 / Accepted: 9 September 2014 Ó ESMRMB 2014
Abstract Object There have been no studies to investigate the effects of cycling exercise protocols, as well as repeated bouts of exercise, on the blood oxygen level-dependent (BOLD) response in the quadriceps muscles. This study characterized BOLD signal recovery following non-ischemic bouts of exercise in the quadriceps muscles of healthy adults in order to provide a basis for application of a protocol for clinical populations. Materials and methods Healthy male subjects (23.7 ± 2.0 years of age, n = 10) completed three cycles of one-minute exercise (65 % of maximum workload), with two minutes of rest between each bout, on an MRIcompatible ergometer. The BOLD responses during recovery were fitted to a sigmoid model, and response
kinetics (post-exercise intensity [S0]), response time (a), change in baseline BOLD signal (j), and inflection point (b)] were measured. Results The sigmoid function fit well to the post-exercise BOLD data (r2 = 0.95 ± 0.04). The mean response time was 10.5 ± 3.8 seconds, change in baseline BOLD intensity was 0.15 ± 0.068, and time to half-peak was 20.2 ± 8.6 seconds. Conclusion The proposed sigmoid model is a robust method for quantifying quadriceps BOLD response postexercise without induced ischemia. Extension of this model to evaluate microvascular responses in patients with chronic disease could improve our understanding of exercise intolerance. Keywords
J. E. Caterini S. Thompson Department of Exercise Sciences, University of Toronto, Toronto, Canada A. H. Elzibak G. D. Wells Physiology and Experimental Medicine, The Hospital for Sick Children, Toronto, Canada E. J. St. Michel B. W. McCrindle A. N. Redington Division of Cardiology, The Hospital for Sick Children, Toronto, Canada M. D. Noseworthy Department of Electrical and Computer Engineering, School of Biomedical Engineering, McMaster University, Hamilton, Canada G. D. Wells (&) Faculty of Kinesiology and Physical Education, The University of Toronto, 555 University Avenue, Toronto, ON M5G 1X8, Canada e-mail: [email protected]
BOLD imaging Exercise Skeletal muscle
Introduction The use of functional MRI techniques such as blood oxygen level-dependent (BOLD) imaging allows for noninvasive evaluation of skeletal muscle tissue oxygenation status [1–3]. BOLD exploits the difference in magnetic susceptibility between oxyhemoglobin (oxyHb) and deoxyhemoglobin (deoxyHb) of water molecules [4]. As oxyHb is diamagnetic and deoxyHb paramagnetic, a change in the bulk magnetic susceptibility of local water content occurs when the ratio of these molecules is altered. Thus the measured BOLD MR signal is heavily influenced by the local concentration of these molecules. A number of stimuli for inducing measurable BOLD signal changes in skeletal muscles have been introduced, including cuff compression [5–10] and exercise challenges [11–13]. Most studies utilizing these protocols have been
123
Magn Reson Mater Phy
performed to gain insight into muscles of the calf region. Recently, models have been developed to quantify calf muscle BOLD recovery following exercise and postocclusion hyperemia [13–15]. To our knowledge, there are no studies in the literature that have modeled the BOLD recovery of exercised quadriceps muscles, most likely due to the challenges associated with exercising the quadriceps inside the MR scanner. In addition, the muscular region chosen for evaluation is dependent upon the disease and population being investigated, as well as the functional end organ [16]. In most clinical studies, the focus has been on diseases that manifest through microvascular alterations in the calf region [16]. Cycle ergometers that are both non-ferrous and fit in the MRI bore are generally limited to major hospitals with ongoing research programs. Previous experimenters looking at T2-based changes had participants perform exercise outside of the MRI, and then data acquisition occurred once participants were repositioned inside the scanner [17]. The resulting lag time between end-exercise and onset of data acquisition eliminates a significant portion of early exercise recovery measurements. Thus, similar to BOLD studies that use MRI-compatible ergometers to model the BOLD signal following calf exercise, there is a need to characterize quadriceps BOLD recovery immediately after an exercise protocol. In addition, numerous diseases are known to result in changes in muscle microcirculation. Analysis of immediate post-exercise muscle metabolite recovery has already been performed in several patient populations using phosphorus31 magnetic resonance spectroscopy (31P-MRS) and cycle ergometers that fit in the magnet bore [18, 19]. Although the cuff-compression BOLD paradigm for measuring calf muscle response has been useful in distinguishing between healthy and diseased subjects, some diseases exhibit impaired oxidative function during exercise and recovery [17–19]. Characterizing the immediate BOLD recovery of quadriceps muscles in addition to the existing 31P-MRS data following a submaximal exercise bout has the potential to further our understanding of vascular reactivity and perfusion in these diseased populations. In this study, we employed an exercise recovery paradigm that has been used in the past to characterize phosphocreatine and pH recovery [18, 20]. As such, we investigated an exercise recovery protocol for quantifying BOLD recovery over a time course that could be used in conjunction with phosphocreatine recovery. This would allow us to assess global changes in skeletal muscle oxygen delivery and oxidative metabolism. The aim of this study was to assess whether quadriceps exercise alone, without an ischemic paradigm, was sufficient to induce a quantifiable BOLD signal in the quadriceps of healthy subjects following submaximal exercise. We used an MRI-
123
Table 1 Descriptive characteristics of the subjects that took part in the study (n = 10)
* Physical activity hours are the total hours completed in the week preceding the study
Characteristic
Mean ± SD
Age (years)
23.7 ± 2.0
Height (cm)
175.9 ± 5.5
Weight (kg)
81.8 ± 16.0
BMI
27.4 ± 4.7
BP systolic (mmHg)
128.5 ± 5.4
BP diastolic (mmHg)
72.4 ± 5.2
Physical activity* (h)
3.1 ± 2.9
Fig. 1 Experimental setup depicting MRI-compatible ergometer used to exercise the quadriceps muscles inside the scanner. a The participant positioning on the magnet table outside the ergometer. During the entire protocol (imaging and exercise), the magnet table and ergometer (b) move into the MRI such that the quadriceps are at the center of the magnet. The receive-only coil is not shown in this figure
compatible ergometer to induce BOLD signal changes in quadriceps muscles, and we quantified the post-exercise BOLD response using a sigmoid function.
Materials and methods Subjects The study was approved by our research ethics board, and our experiment was performed in accordance with the ethical standards laid down by the Declaration of Helsinki and its later amendments. All subjects read and signed informed consent documentation prior to inclusion in the study. Ten healthy, non-smoking, post-pubescent male subjects between the ages of 20 and 27 years took part in the study. None of the subjects were athletes (no more than five exercise sessions per week) or had major medical conditions, surgery, medical intervention, or illness in the
Magn Reson Mater Phy
48 hours preceding the study. Participants were instructed to avoid caffeine and alcohol for 24 hours prior to partaking in the study. The mean and standard deviation for the subjects’ height, weight, blood pressure, age, and level of exercise in the week prior to the study are shown in Table 1. Experimental design An up/down MRI-compatible ergometer that fit inside the magnet bore (Lode; AEI, The Netherlands) was used to exercise the quadriceps muscles (quadriceps extensions) inside the MR scanner while subjects were in the supine position (Fig. 1). Prior to data collection, subjects performed a 30-second maximal initial exercise test, which was used to determine their maximal workload (Wmax) in order to establish the intensity of subsequent exercise. To allow for recovery, the testing session was followed by a 10-minute resting period prior to the imaging session. During the imaging session, participants completed three one-minute cycles of exercise at 65 % of Wmax, with two minutes of rest between each bout of exercise. The participant’s non-dominant quadriceps muscle was established based on handedness as stated by the participant. The non-dominant quadriceps was imaged, and data was acquired at baseline, during exercise, and during the twominute recovery immediately following each exercise bout. To minimize motion during imaging, the leg was secured to the ergometer at the ankle and the knee. The ergometer automatically controlled power output by adjusting resistance in relationship to the participant’s freely chosen movement frequency. Thus, exercise was controlled for power output, as relative work rate is an important factor in the determination of muscle metabolism during exercise and recovery. Magnetic resonance imaging All images were collected using a 1.5T MRI system (TwinSpeed EXCITETM III 12.0, GE Healthcare, Milwaukee, WI) and a multi-purpose receive-only flexible surface coil wrapped around the quadriceps of the nondominant leg. MR images were acquired from the midquadriceps region, approximately 50 % of the distance from the patella to the femoral neck. Subjects were in the supine position and were positioned head-first in the scanner. Velcro straps attached to the coil helped to secure the coil in position and minimize the motion of the muscle relative to the coil during imaging. Following localization, T1-weighted anatomical images were acquired axially from the mid-quadriceps region (spin echo, 8 slices, 10 mm thick, FOV 250 mm, TE/TR 15/350 ms, 256 9 192 matrix). T2*-weighted BOLD images were
Fig. 2 Sample T2*-weighted BOLD slice of the quadriceps showing the location of the three ROIs investigated in the study. Each ROI covered a region of 88 mm2. ROIs were positioned in the following muscles: 1-vastus medialis, 2-biceps femoris, and 3-vastus intermedius
then obtained using a gradient echo EPI sequence (slice thickness 10 mm, FOV 250 mm, TE/TR 40/250 ms, a 90°, 64 9 64 matrix, 2,400 temporal points, 1 slice, total scan duration 10 min). Data analysis Images were analyzed offline. BOLD scans were initially corrected for motion using the MCFLIRT script [21] from the FMRIB Software Library FSL (FMRIB Analysis Group) [22]. Regions of interest (ROIs) were drawn in the vastus medialis, biceps femoris, and vastus intermedius muscles using Analysis of Functional Neuroimages (AFNI) software (National Institute of Mental Health) [23]. Care was taken to exclude resolved vessels. Each ROI covered a region of 88 mm2 (Fig. 2). The average time course of signal intensity changes in each ROI was used in the fitting process. The BOLD time course was initially normalized to the mean baseline intensity. In muscles that were recruited in the exercise (vastus medialis and vastus intermedius), BOLD signal recovery following exercise was fitted to a four-parameter sigmoid function, S(t): SðtÞ ¼ S0 þ
k 1 þ eðbtÞ=a
ð1Þ
where S(t) is the BOLD signal intensity at a given time t, S0 is the baseline BOLD signal intensity and point of minimum tissue oxygenation (immediately after the
123
Magn Reson Mater Phy
Fig. 3 Sample sigmoid function as represented by Eq. (1). Shown are the four parameters used in the fitting. S0 is the baseline BOLD signal intensity (immediately after the exercise, 1.001 in this example), j is the change in BOLD signal intensity (0.1314 as shown here), a indicates the response of the sigmoid function (the higher the value, the slower the response, a = 8.556 s in this example), and b is the half-time of the BOLD signal recovery (22.17 s as shown here)
Fig. 4 Sample time course of BOLD signal intensity from the vastus medialis muscle in one of the subjects. Grey boxes represent the exercise portion. A similar response was found in the vastus intermedius muscle
exercise), j is the change in baseline BOLD signal intensity representing the maximum increase in reoxygenation and hyperemia following exercise, a indicates the response of the sigmoid function or washout time of deoxyhemoglobin, and b measures recovery half-time of (s) (Fig. 3). Recovery curves were fitted using the Curve Fitting Toolbox with MATLAB (The MathWorks, Natick, MA). Parameter estimates were computed using a trust-region fitting algorithm. Statistical analysis To determine whether the recovery characteristics varied among the three sequential exercise bouts, test–retest repeatability was evaluated between recoveries 1 and 2, 1 and 3, and 2 and 3, using Pearson’s correlation coefficient. All statistical analyses were performed using GraphPad Prism (Version 5.0c, GraphPad Software, Inc.). The level of significance was set to p \ 0.05 for statistical procedures.
Results Participants were involved in physical activity ranging from 1 to 9 hours per week, and were recreationally active. Systolic and diastolic blood pressures of all subjects were within normal range (Table 1). All subjects were able to successfully complete the exercise protocol. A sample time course of BOLD signal intensity from the vastus medialis muscle from one participant is shown in Fig. 4. It was noted that the vastus medialis and vastus intermedius muscles were both recruited during the exercise challenge, and they showed similar recovery patterns.
123
Fig. 5 Sample time course of BOLD signal intensity from the biceps femoris muscle in one of the subjects. This muscle was not recruited in the exercise, as evidenced by the lack of BOLD recovery following exercise. Grey boxes represent the exercise bouts
In contrast, the biceps femoris muscle (part of the hamstring group) was not recruited in the exercise, as evidenced by the lack of change in BOLD signal during recovery (Fig. 5). A total of 30 recovery curves were collected from each muscle, as each participant completed three bouts of exercise (n = 10 subjects). Of these 30 curves, four could not be fitted due to motion contamination that was still present after motion correction. As such, these curves were not included in the data analysis. To characterize BOLD signal during recovery following exercise in the muscles that were activated by the exercise (vastus medialis and vastus intermedius), a four-parameter sigmoid function was generated (Fig. 3; Eq. 1). Sample BOLD recovery data and the fitted sigmoid function are shown in Fig. 6. The sigmoid function fit the data well, with mean r2 (SD) of 0.95 (0.04) (Table 2). Parameter estimates from the sigmoid curve were used to characterize BOLD post-exercise response kinetics. The mean response time for the vastus medialis (a) was 10.5 ± 3.8 seconds,
Magn Reson Mater Phy Table 3 Test–retest correlation among all three recoveries for vastus medialis and vastus intermedius muscle VM
1 versus 2
2 versus 3
1 versus 3
S0
0.93**
0.72
0.78
j
0.96**
0.88*
0.92*
b
0.86*
0.58
0.73
a
0.59
0.58
0.54
S0
0.56
0.63
0.94**
j b
0.82** 0.94**
0.81* 0.77*
0.96** 0.84*
a
0.89**
0.77*
0.97**
VI
* Represents a significant correlation of p \ 0.05, ** p \ 0.01
Fig. 6 Fitted curves of mean parameter estimates for vastus medialis (VM, top panel) and vastus intermedius (VI, lower panel) from all participants (parameters shown in Table 2) for recoveries 1, 2, and 3
Table 2 Parameters of the BOLD recovery curves for the two muscles that were recruited in the exercise as obtained from the fitted sigmoid function (Fig. 3) Parameter
Recovery 1 (n = 10)
Recovery 2 (n = 10)
Recovery 3 (n = 6)
Discussion
Vastus medialis muscle Baseline intensity, S0 (au)
0.97 ± 0.06
Change in baseline intensity (j)
0.14 ± 0.06
Inflection point b (s)
22.1 ± 8.5
18.5 ± 7.4
19.8 ± 11.2
Response time a (s)
9.6 ± 3.6
10.8 ± 3.7
11.7 ± 4.6
0.95 ± 0.03
0.95 ± 0.04
0.94 ± 0.06
Goodness of fit r2
0.97 ± 0.07 0.16 ± 0.07
0.94 ± 0.08 0.15 ± 0.09
Vastus intermedius muscle Baseline intensity, S0 (au)
0.97 ± 0.05
0.98 ± 0.07
0.98 ± 0.05
Change in baseline intensity j Inflection point b (s)
0.19 ± 0.07
0.20 ± 0.07
0.16 ± 0.05
24.6 ± 13.4
20.9 ± 13.4
20.5 ± 10.1
Response time a (s)
10.0 ± 4.4
10.3 ± 4.7
7.2 ± 3.4
0.96 ± 0.02
0.96 ± 0.02
0.92 ± 0.05
Goodness of fit r
2
the mean change in baseline BOLD intensity (j) was 0.15 ± 0.068, and the inflection point (b) was at 20.2 ± 8 seconds (Table 2). The mean response time for vastus intermedius (a) was 9.4 ± 4.3 seconds, the mean change in baseline BOLD intensity (j) was 0.18 ± 0.07, and the inflection point (b) was at 22.2 ± 12.3 seconds. Intraparticipant test–retest repeatability was significantly correlated (p \ 0.05) for all recoveries for parameters j, b, and a in the vastus intermedius muscle (Table 3). Baseline signal (S0) in the vastus intermedius correlated only for recoveries 1 and 3. Vastus medialis recovery was significantly correlated among all recoveries for signal recovery range j, and only in recoveries 1 and 2 for baseline signal intensity S0 and half-time for recovery b.
Values are mean ± SD (n = 10 for the first two recoveries; n = 6 for the third recovery, as four curves were contaminated by motion and could not be fitted)
This study was designed to characterize quadriceps BOLD response following submaximal bouts of leg extension exercise performed using an MRI-compatible ergometer. Consistent temporal changes in BOLD SI were observed in the 10 participants who performed submaximal exercise during MRI. These changes fit well to a four-parameter sigmoid curve (r2 = 0.95 ± 0.04) in both the vastus medialis and vastus intermedius muscles (Table 2; Fig. 6). There are only a few studies in the literature involving quantification of BOLD response following an applied stimulus [13–15, 24], and all of these have modeled the BOLD signal in the calf region, either following exercise [13, 14, 24] or during the reperfusion after cuff-induced ischemia [15]. To our knowledge, no study has modeled the BOLD response in the quadriceps region post-exercise without induced ischemia. Our BOLD time course was well-characterized by a four-parameter sigmoid function
123
Magn Reson Mater Phy
similar to that of McGrath and colleagues, who used a fourparameter sigmoid curve to fit the T1 signal response in skeletal muscles in the trunk region to 100 % oxygen inhalation [25]. It is interesting to note that previous BOLD modeling studies performed in the calf region have utilized a gamma variate function [14, 15], either in combination with sigmoidal and linear functions [15] or in addition to a linear curve [14]. Our post-exercise quadriceps BOLD response (Figs. 4, 6) did not demonstrate the sharp increase and rapid dampening characteristic of the gamma function commonly used in calf studies [14, 15]. It is possible that the quadriceps muscle exhibits a different BOLD signal recovery following an oxygen-depletion stimulus from the response demonstrated by the calf muscle [15]. However, our fitted parameters were similar to those in the calf region demonstrated in previous studies. For example, the change in baseline BOLD intensity (j) was 0.16 in both vastus medialis and vastus intermedius muscles (Table 2), which was similar to that observed in the lateral gastrocnemius muscle (0.18) following 2.5 minutes of plantar flexion exercise [14]. Since the total measurement time in our study was 120 seconds following each bout of exercise, our model did not require the addition of a linear function to account for the signal drift that has been observed over longer measurement durations [14, 15]. A longer period of BOLD MRI data acquisition following exercise could result in a dampening of BOLD SI to baseline values, as Fig. 4 demonstrates that signal recovery increases to a plateau approximately 15 % higher than baseline. Nevertheless, our data demonstrate that 120 seconds of imaging following submaximal exercise without induced ischemia may be sufficient to evaluate the acute recovery of the BOLD SI response following a short bout of exercise, as all of our participants’ BOLD SI recovery reached a clear plateau within this time frame. Quantifying the BOLD MRI response immediately following exercise is crucial, as physiological responses such as increased blood flow and tissue perfusion can be impaired in chronic diseases. A consequence of reduced perfusion is a reduction in the ability of tissue to remove waste products, resulting in decreased pH, a more rapid onset of fatigue, and longer recovery times from exercise. Research has demonstrated that patients with respiratory diseases such as cystic fibrosis exhibit longer phosphocreatine recovery times following quadriceps exercise, contributing to exercise intolerance [18]. In addition, individuals with Turner syndrome exhibit differences in pH following short bursts of quadriceps exercise in comparison to healthy controls [19]. In this study, we selected an exercise recovery protocol that was similar to the paradigms that have been applied to study 31P magnetic resonance spectroscopy in children
123
with cystic fibrosis [18] and Turner syndrome [19]. Since consistent temporal changes were observed in the current study, we anticipate that comparisons between BOLD MRI signal responses and phosphocreatine and pH recovery can be carried out in subsequent studies. The use of 31P-MRS in conjunction with BOLD MRI will provide data to elucidate the pathophysiology of skeletal muscle vascular or metabolic dysfunction. Thus, important questions related to impairment of oxygen delivery versus oxygen utilization could be answered using this paradigm. Approaches that combine MR imaging techniques, such as the simultaneous use of 31P-MRS and arterial spin labeling (ASL) to assess muscle energy metabolism and perfusion, respectively [26– 30], and31P and 1H spectroscopy to investigate muscle oxygenation supply and utilization [30–32], have been carried out to investigate pathological conditions [28]. The BOLD response is modulated by numerous parameters and is dependent upon tissue oxygenation, blood fractional volume, capillary network architecture, and orientation relative to the main magnetic field. Since a complex set of factors modulates the BOLD response, it is difficult to identify the physiological factors behind our observed quadriceps recovery following submaximal exercise without the use of complementary imaging approaches. To examine intracellular oxygenation, for instance, myoglobin desaturation can be probed using 1HMRS, as others have done previously [33–35]. In future studies, we plan to use a combination of simultaneous BOLD signal acquisition with NIRS data to measure hemoglobin saturation in order to quantify oxygen consumption and reoxygenation from vascular inflow. There were several limitations to this study. Subsequent experiments should include longer post-exercise recovery following the third exercise bout in order to test whether BOLD SI returns to baseline levels. Furthermore, significant intra-subject variability as measured by non-significant test–retest reliability was present among recoveries in the vastus medialis muscle, while significant reliability was present in the vastus intermedius (Table 3). This suggests that the vastus intermedius muscle may be a more reliable region of interest to evaluate BOLD response in the quadriceps following acute exercise. Additionally, there was high variance in the four parameters of BOLD SI signal recovery response among the different participants The change in signal intensity and maximal hyperemic value from baseline (j) could be affected by vascular volume differences due to increased capillarization caused by variations in aerobic activity levels [36]. Physically active subjects have demonstrated a greater BOLD hyperemic response in the anterior tibialis muscle following maximal ankle dorsiflexion compared to inactive subjects [37]. We observed a high variability in physical activity levels (1–9 hours) among participants in the week prior to
Magn Reson Mater Phy
testing, as demonstrated in Table 1. While none of our participants were trained athletes, if some subjects were more physically active than others, this could have contributed to the variability in the observed BOLD response. In addition, the muscle fiber composition of each subject may be different, leading to differences among subjects in the metabolic cost to generate ATP. Differences in ATPgenerating capacity of the mitochondria may affect the half-time to recovery for the BOLD response (b). Increased ATP regeneration efficiency would result in more rapid reoxygenation of oxyhemoglobin, as less oxygen would be used to regenerate ATP for synergistic muscle groups. As such, further experiments are needed to characterize the difference in quadriceps BOLD response between trained and untrained subjects in order to understand the interindividual variability inherent in these measurements. Conclusion In conclusion, this study presents a method to evaluate quadriceps BOLD response following submaximal bouts of leg extension exercise without accompanying ischemia. A four-parameter sigmoidal function was employed to model quadriceps BOLD time course in response to an inmagnet exercise protocol using an MRI-compatible ergometer. In future studies, we propose a comparison of simultaneous BOLD signal acquisition with NIRS data to measure hemoglobin saturation. These measurements will enable calculation of the post-exercise relative contribution of vascular inflow versus tissue oxygenation in the quadriceps muscle in our sigmoid model. This has already been done in the calf muscles of healthy volunteers and patients using an ischemia-hyperemia paradigm [13]. Application of our model may aid in understanding diseases that exhibit impaired skeletal muscle function during exercise and recovery, such as peripheral arterial occlusive disease, metabolic disorders, and mitochondrial myopathies. Conflict of interest conflict of interest.
The authors each declare that they have no
Ethical standards Our hospital research ethics board approved the study, and all human study was performed in accordance with the ethical standards created by the Declaration of Helsinki and its later amendments. Prior to study inclusion, all participants read and signed informed consent documentation.
References 1. Jacobi B, Bongartz G, Partovi S et al (2012) Skeletal muscle BOLD MRI: from underlying physiological concepts to its usefulness in clinical conditions. J Magn Reson Imaging 35(6): 1253–1265
2. Noseworthy MD, Davis AD, Elzibak AH (2010) Advanced MR imaging techniques for skeletal muscle evaluation. Semin Musculoskelet Radiol 14(2):257–268 3. Partovi S, Karimi S, Jacobi B et al (2012) Clinical implications of skeletal muscle blood-oxygenation-level-dependent (BOLD) MRI. Magn Reson Mater Phy 25:251–261 4. Noseworthy MD, Bulte DP, Alfonsi J (2003) BOLD magnetic resonance imaging of skeletal muscle. Semin Musculoskelet Radiol 7(4):307–315 5. Ledermann HP, Schulte AC, Heidecker HG et al (2006) Blood oxygenation level-dependent magnetic resonance imaging of the skeletal muscle in patients with peripheral arterial occlusive disease. Circulation 113:2929–2935 6. Schulte AC, Aschwanden M, Bilecen D (2008) Calf muscles at blood oxygen level-dependent MR imaging: aging effects at postocclusive reactive hyperemia. Radiology 247(2):482–489 7. Partovi S, Schulte AC, Aschwanden M et al (2012) Impaired skeletal muscle microcirculation in systemic sclerosis. Arthritis Res Ther 14:R209 8. Jacobi B, Schulte AC, Partovi S et al (2013) Alterations of skeletal muscle microcirculation detected by blood oxygenation level-dependent MRI in a patient with granulomatosis with polyangiitis. Rheumatology 52:579–581 9. Partovi S, Aschwanden M, Jacobi B et al (2013) Correlation of muscle BOLD MRI with transcutaneous oxygen pressure for assessing microcirculation in patients with systemic sclerosis. J Magn Reson Imaging 38(4):845–851 10. Partovi S, Schulte AC, Staub D et al (2013) Correlation of skeletal muscle blood oxygenations level-dependent MRI and skin laser Doppler flowmetry in patients with systemic sclerosis. J Magn Reson Imaging. doi:10.1002/jmri.24503 11. Damon BM, Wadington MC, Hornberger JL, Lansdown DA (2007) Absolute and relative contributions of BOLD effects to the muscle functional MRI signal intensity time course: effect of exercise intensity. Magn Reson Med 58:335–345 12. Meyer RA, Towse TF, Reid RW, Jayaraman RC, Wiseman RW, McCully KK (2004) BOLD MRI mapping of transient hyperemia in skeletal muscle after single contractions. NMR Biomed 17:392–398 13. Towse TF, Slade JM, Ambrose JA, DeLano MC, Meyer RA (2011) Quantitative analysis of the postcontractile blood-oxygenation-level-dependent (BOLD) effect in skeletal muscle. J Appl Physiol 111(1):27–39 14. Davis AD, Noseworthy MD (2013) Consistency of post-exercise skeletal muscle BOLD response. In: Proceedings of the 21st Annual Meeting of ISMRM, Salt Lake City, 6496 15. Schewzow K, Andreas M, Moser E, Wolzt M, Schmid AI (2013) Automatic model-based analysis of skeletal muscle BOLD-MRI in reactive hyperemia. J Magn Reson Imag 38:96–969 16. Aschwanden M, Patrovi S, Jacobi B et al (2014) Assessing the end-organ in peripheral arterial occlusive disease-from contastenhanced ultrasound to blood-oxygen-level-dependent MR imaging. Cardiovasc Diagn Ther 4(2):165–172 17. Reid RW, Foley JM, Jayaraman RC, Prior BM, Meyer RA (2001) Effect of aerobic capacity on the T2 increase in exercised skeletal muscle. J Appl Physiol 90(3):897–902 18. Wells GD, Wilkes DL, Schneiderman JE et al (2011) Skeletal muscle metabolism in cystic fibrosis and primary ciliary dyskinesia. Pediatr Res 69(1):40–45 19. Wells GD, O’Gorman CS, Rayner T et al (2013) Skeletal muscle abnormalities in girls and adolescents with turner syndrome. J Clini Endocrinol Metab 98(6):2521–2527 20. Massie BM, Conway M, Rajagopalan B et al (1988) Skeletal muscle metabolism during exercise under ischemic conditions in congestive heart failure. Evidence for abnormalities unrelated to blood flow. Circulation 78(2):320–326
123
Magn Reson Mater Phy 21. Jenkinson M, Bannister P, Brady JM, Smith SM (2002) Improved optimisation for the robust and accurate linear registration and motion correction of brain images. Neuroimage 17(2):825–841 22. Jenkinson M, Beckmann CF, Behrens TE, Woolrich MW, Smith SM (2002) FSL Neuroimage 62:782–790 23. Cox RW (1996) AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Comput Biomed Res 29(3):162–173 24. Davis AD, Noseworthy MD (2013) Haemoglobin-derived curve fitting to post-exercise muscle BOLD data. In: Proceedings of the 21st Annual Meeting of ISMRM, Salt Lake City. p 5311 25. McGrath DM, Naish JH, O’Connor JPB et al (2008) Oxygeninduced changes in the longitudinal relaxation times in skeletal muscle. Magn Reson Imaging 26(2):221–227 26. Carlier PG, Brillault-Salvat C, Giacomini E, Wary C, Bloch G (2005) How to investigate oxygen supply, uptake, and utilization simultaneously by interleaved NMR imaging and spectroscopy of the skeletal muscle. Magn Reson Med 54:1010–1013 27. Carlier PG, Bertoldi D (2005) In vivo functional NMR imaging of resistance artery control. Am J Physiol Heart Circ Physiol 288:H1028–H1036 28. Vidal G, Giacomini E, Wary C et al (2003) Combined ASL perfusion imaging, BOLD imaging and 31P NMR spectroscopy of the leg in a rat model of peripheral arteriopathy. Proc 11th Meeting Int Soc Magn Reson Med, Toronto p. 1692 29. Baligand C, Gilson H, Me´nard JC et al (2010) Functional assessment of skeletal muscle in intact mice lacking myostatin by concurrent NMR imaging and spectroscopy. Gene Ther 17:328–337
123
30. Duteil S, Bourrilhon C, Raynaud JS et al (2004) Metabolic and vascular support for the role of myoglobin in humans: a multiparametric NMR study. Am J Physiol Regul Integr Comp Physiol. 287:R1441–R1449 31. Vanderthommen M, Duteil S, Wary C et al (2003) A comparison of voluntary and electrically induced contractions by interleaved 1 H- and 31P-NMRS in humans. J Appl Physiol 94:1012–1024 32. Brillault-Salvat C, Giacomini E, Wary C et al (1997) An interleaved heteronuclear NMRI-NMRS approach to non-invasive investigation of exercising human skeletal muscle. Cell Mol Biol 43:751–762 33. Richardson RS, Newcomer SC, Noyszewski EA (2001) Skeletal muscle intracellular PO2 assessed by myoglobin desaturation: response to graded exercise. J Appl Physiol 91:2679–2685 34. Richardson RS, Noyszewski EA, Kendrick KF, Leigh JS, Wagner PD (1995) Myoglobin O2 desaturation during exercise. Evidence of limited O2 transport. J Clin Invest 96:1916–1926 35. Richardson RS, Noyszewski EA, Leigh JS, Wagner PD (1998) Lactate efflux from exercising human skeletal muscle: role of intracellular Po2. J Appl Physiol 85:627–634 36. Shono N, Urata H, Saltin B et al (2002) Effects of low intensity aerobic training on skeletal muscle capillary and blood lipoprotein profiles. J Atheroscler Thromb 9(1):78–85 37. Towse TF, Slade JM, Meyer RA (2005) Effect of physical activity on MRI-measured blood oxygen level-dependent transients in skeletal muscle after brief contractions. J Appl Physiol 99(2):715–722
RESEARCH ARTICLE
Characterizing blood oxygen level-dependent (BOLD) response following in-magnet quadriceps exercise Jessica E. Caterini • Alyaa H. Elzibak • Emilie Jean St. Michel Brian W. McCrindle • Andrew N. Redington • Sara Thompson Michael D. Noseworthy • Greg D. Wells
• •
Received: 6 May 2014 / Revised: 11 August 2014 / Accepted: 9 September 2014 Ó ESMRMB 2014
Abstract Object There have been no studies to investigate the effects of cycling exercise protocols, as well as repeated bouts of exercise, on the blood oxygen level-dependent (BOLD) response in the quadriceps muscles. This study characterized BOLD signal recovery following non-ischemic bouts of exercise in the quadriceps muscles of healthy adults in order to provide a basis for application of a protocol for clinical populations. Materials and methods Healthy male subjects (23.7 ± 2.0 years of age, n = 10) completed three cycles of one-minute exercise (65 % of maximum workload), with two minutes of rest between each bout, on an MRIcompatible ergometer. The BOLD responses during recovery were fitted to a sigmoid model, and response
kinetics (post-exercise intensity [S0]), response time (a), change in baseline BOLD signal (j), and inflection point (b)] were measured. Results The sigmoid function fit well to the post-exercise BOLD data (r2 = 0.95 ± 0.04). The mean response time was 10.5 ± 3.8 seconds, change in baseline BOLD intensity was 0.15 ± 0.068, and time to half-peak was 20.2 ± 8.6 seconds. Conclusion The proposed sigmoid model is a robust method for quantifying quadriceps BOLD response postexercise without induced ischemia. Extension of this model to evaluate microvascular responses in patients with chronic disease could improve our understanding of exercise intolerance. Keywords
J. E. Caterini S. Thompson Department of Exercise Sciences, University of Toronto, Toronto, Canada A. H. Elzibak G. D. Wells Physiology and Experimental Medicine, The Hospital for Sick Children, Toronto, Canada E. J. St. Michel B. W. McCrindle A. N. Redington Division of Cardiology, The Hospital for Sick Children, Toronto, Canada M. D. Noseworthy Department of Electrical and Computer Engineering, School of Biomedical Engineering, McMaster University, Hamilton, Canada G. D. Wells (&) Faculty of Kinesiology and Physical Education, The University of Toronto, 555 University Avenue, Toronto, ON M5G 1X8, Canada e-mail: [email protected]
BOLD imaging Exercise Skeletal muscle
Introduction The use of functional MRI techniques such as blood oxygen level-dependent (BOLD) imaging allows for noninvasive evaluation of skeletal muscle tissue oxygenation status [1–3]. BOLD exploits the difference in magnetic susceptibility between oxyhemoglobin (oxyHb) and deoxyhemoglobin (deoxyHb) of water molecules [4]. As oxyHb is diamagnetic and deoxyHb paramagnetic, a change in the bulk magnetic susceptibility of local water content occurs when the ratio of these molecules is altered. Thus the measured BOLD MR signal is heavily influenced by the local concentration of these molecules. A number of stimuli for inducing measurable BOLD signal changes in skeletal muscles have been introduced, including cuff compression [5–10] and exercise challenges [11–13]. Most studies utilizing these protocols have been
123
Magn Reson Mater Phy
performed to gain insight into muscles of the calf region. Recently, models have been developed to quantify calf muscle BOLD recovery following exercise and postocclusion hyperemia [13–15]. To our knowledge, there are no studies in the literature that have modeled the BOLD recovery of exercised quadriceps muscles, most likely due to the challenges associated with exercising the quadriceps inside the MR scanner. In addition, the muscular region chosen for evaluation is dependent upon the disease and population being investigated, as well as the functional end organ [16]. In most clinical studies, the focus has been on diseases that manifest through microvascular alterations in the calf region [16]. Cycle ergometers that are both non-ferrous and fit in the MRI bore are generally limited to major hospitals with ongoing research programs. Previous experimenters looking at T2-based changes had participants perform exercise outside of the MRI, and then data acquisition occurred once participants were repositioned inside the scanner [17]. The resulting lag time between end-exercise and onset of data acquisition eliminates a significant portion of early exercise recovery measurements. Thus, similar to BOLD studies that use MRI-compatible ergometers to model the BOLD signal following calf exercise, there is a need to characterize quadriceps BOLD recovery immediately after an exercise protocol. In addition, numerous diseases are known to result in changes in muscle microcirculation. Analysis of immediate post-exercise muscle metabolite recovery has already been performed in several patient populations using phosphorus31 magnetic resonance spectroscopy (31P-MRS) and cycle ergometers that fit in the magnet bore [18, 19]. Although the cuff-compression BOLD paradigm for measuring calf muscle response has been useful in distinguishing between healthy and diseased subjects, some diseases exhibit impaired oxidative function during exercise and recovery [17–19]. Characterizing the immediate BOLD recovery of quadriceps muscles in addition to the existing 31P-MRS data following a submaximal exercise bout has the potential to further our understanding of vascular reactivity and perfusion in these diseased populations. In this study, we employed an exercise recovery paradigm that has been used in the past to characterize phosphocreatine and pH recovery [18, 20]. As such, we investigated an exercise recovery protocol for quantifying BOLD recovery over a time course that could be used in conjunction with phosphocreatine recovery. This would allow us to assess global changes in skeletal muscle oxygen delivery and oxidative metabolism. The aim of this study was to assess whether quadriceps exercise alone, without an ischemic paradigm, was sufficient to induce a quantifiable BOLD signal in the quadriceps of healthy subjects following submaximal exercise. We used an MRI-
123
Table 1 Descriptive characteristics of the subjects that took part in the study (n = 10)
* Physical activity hours are the total hours completed in the week preceding the study
Characteristic
Mean ± SD
Age (years)
23.7 ± 2.0
Height (cm)
175.9 ± 5.5
Weight (kg)
81.8 ± 16.0
BMI
27.4 ± 4.7
BP systolic (mmHg)
128.5 ± 5.4
BP diastolic (mmHg)
72.4 ± 5.2
Physical activity* (h)
3.1 ± 2.9
Fig. 1 Experimental setup depicting MRI-compatible ergometer used to exercise the quadriceps muscles inside the scanner. a The participant positioning on the magnet table outside the ergometer. During the entire protocol (imaging and exercise), the magnet table and ergometer (b) move into the MRI such that the quadriceps are at the center of the magnet. The receive-only coil is not shown in this figure
compatible ergometer to induce BOLD signal changes in quadriceps muscles, and we quantified the post-exercise BOLD response using a sigmoid function.
Materials and methods Subjects The study was approved by our research ethics board, and our experiment was performed in accordance with the ethical standards laid down by the Declaration of Helsinki and its later amendments. All subjects read and signed informed consent documentation prior to inclusion in the study. Ten healthy, non-smoking, post-pubescent male subjects between the ages of 20 and 27 years took part in the study. None of the subjects were athletes (no more than five exercise sessions per week) or had major medical conditions, surgery, medical intervention, or illness in the
Magn Reson Mater Phy
48 hours preceding the study. Participants were instructed to avoid caffeine and alcohol for 24 hours prior to partaking in the study. The mean and standard deviation for the subjects’ height, weight, blood pressure, age, and level of exercise in the week prior to the study are shown in Table 1. Experimental design An up/down MRI-compatible ergometer that fit inside the magnet bore (Lode; AEI, The Netherlands) was used to exercise the quadriceps muscles (quadriceps extensions) inside the MR scanner while subjects were in the supine position (Fig. 1). Prior to data collection, subjects performed a 30-second maximal initial exercise test, which was used to determine their maximal workload (Wmax) in order to establish the intensity of subsequent exercise. To allow for recovery, the testing session was followed by a 10-minute resting period prior to the imaging session. During the imaging session, participants completed three one-minute cycles of exercise at 65 % of Wmax, with two minutes of rest between each bout of exercise. The participant’s non-dominant quadriceps muscle was established based on handedness as stated by the participant. The non-dominant quadriceps was imaged, and data was acquired at baseline, during exercise, and during the twominute recovery immediately following each exercise bout. To minimize motion during imaging, the leg was secured to the ergometer at the ankle and the knee. The ergometer automatically controlled power output by adjusting resistance in relationship to the participant’s freely chosen movement frequency. Thus, exercise was controlled for power output, as relative work rate is an important factor in the determination of muscle metabolism during exercise and recovery. Magnetic resonance imaging All images were collected using a 1.5T MRI system (TwinSpeed EXCITETM III 12.0, GE Healthcare, Milwaukee, WI) and a multi-purpose receive-only flexible surface coil wrapped around the quadriceps of the nondominant leg. MR images were acquired from the midquadriceps region, approximately 50 % of the distance from the patella to the femoral neck. Subjects were in the supine position and were positioned head-first in the scanner. Velcro straps attached to the coil helped to secure the coil in position and minimize the motion of the muscle relative to the coil during imaging. Following localization, T1-weighted anatomical images were acquired axially from the mid-quadriceps region (spin echo, 8 slices, 10 mm thick, FOV 250 mm, TE/TR 15/350 ms, 256 9 192 matrix). T2*-weighted BOLD images were
Fig. 2 Sample T2*-weighted BOLD slice of the quadriceps showing the location of the three ROIs investigated in the study. Each ROI covered a region of 88 mm2. ROIs were positioned in the following muscles: 1-vastus medialis, 2-biceps femoris, and 3-vastus intermedius
then obtained using a gradient echo EPI sequence (slice thickness 10 mm, FOV 250 mm, TE/TR 40/250 ms, a 90°, 64 9 64 matrix, 2,400 temporal points, 1 slice, total scan duration 10 min). Data analysis Images were analyzed offline. BOLD scans were initially corrected for motion using the MCFLIRT script [21] from the FMRIB Software Library FSL (FMRIB Analysis Group) [22]. Regions of interest (ROIs) were drawn in the vastus medialis, biceps femoris, and vastus intermedius muscles using Analysis of Functional Neuroimages (AFNI) software (National Institute of Mental Health) [23]. Care was taken to exclude resolved vessels. Each ROI covered a region of 88 mm2 (Fig. 2). The average time course of signal intensity changes in each ROI was used in the fitting process. The BOLD time course was initially normalized to the mean baseline intensity. In muscles that were recruited in the exercise (vastus medialis and vastus intermedius), BOLD signal recovery following exercise was fitted to a four-parameter sigmoid function, S(t): SðtÞ ¼ S0 þ
k 1 þ eðbtÞ=a
ð1Þ
where S(t) is the BOLD signal intensity at a given time t, S0 is the baseline BOLD signal intensity and point of minimum tissue oxygenation (immediately after the
123
Magn Reson Mater Phy
Fig. 3 Sample sigmoid function as represented by Eq. (1). Shown are the four parameters used in the fitting. S0 is the baseline BOLD signal intensity (immediately after the exercise, 1.001 in this example), j is the change in BOLD signal intensity (0.1314 as shown here), a indicates the response of the sigmoid function (the higher the value, the slower the response, a = 8.556 s in this example), and b is the half-time of the BOLD signal recovery (22.17 s as shown here)
Fig. 4 Sample time course of BOLD signal intensity from the vastus medialis muscle in one of the subjects. Grey boxes represent the exercise portion. A similar response was found in the vastus intermedius muscle
exercise), j is the change in baseline BOLD signal intensity representing the maximum increase in reoxygenation and hyperemia following exercise, a indicates the response of the sigmoid function or washout time of deoxyhemoglobin, and b measures recovery half-time of (s) (Fig. 3). Recovery curves were fitted using the Curve Fitting Toolbox with MATLAB (The MathWorks, Natick, MA). Parameter estimates were computed using a trust-region fitting algorithm. Statistical analysis To determine whether the recovery characteristics varied among the three sequential exercise bouts, test–retest repeatability was evaluated between recoveries 1 and 2, 1 and 3, and 2 and 3, using Pearson’s correlation coefficient. All statistical analyses were performed using GraphPad Prism (Version 5.0c, GraphPad Software, Inc.). The level of significance was set to p \ 0.05 for statistical procedures.
Results Participants were involved in physical activity ranging from 1 to 9 hours per week, and were recreationally active. Systolic and diastolic blood pressures of all subjects were within normal range (Table 1). All subjects were able to successfully complete the exercise protocol. A sample time course of BOLD signal intensity from the vastus medialis muscle from one participant is shown in Fig. 4. It was noted that the vastus medialis and vastus intermedius muscles were both recruited during the exercise challenge, and they showed similar recovery patterns.
123
Fig. 5 Sample time course of BOLD signal intensity from the biceps femoris muscle in one of the subjects. This muscle was not recruited in the exercise, as evidenced by the lack of BOLD recovery following exercise. Grey boxes represent the exercise bouts
In contrast, the biceps femoris muscle (part of the hamstring group) was not recruited in the exercise, as evidenced by the lack of change in BOLD signal during recovery (Fig. 5). A total of 30 recovery curves were collected from each muscle, as each participant completed three bouts of exercise (n = 10 subjects). Of these 30 curves, four could not be fitted due to motion contamination that was still present after motion correction. As such, these curves were not included in the data analysis. To characterize BOLD signal during recovery following exercise in the muscles that were activated by the exercise (vastus medialis and vastus intermedius), a four-parameter sigmoid function was generated (Fig. 3; Eq. 1). Sample BOLD recovery data and the fitted sigmoid function are shown in Fig. 6. The sigmoid function fit the data well, with mean r2 (SD) of 0.95 (0.04) (Table 2). Parameter estimates from the sigmoid curve were used to characterize BOLD post-exercise response kinetics. The mean response time for the vastus medialis (a) was 10.5 ± 3.8 seconds,
Magn Reson Mater Phy Table 3 Test–retest correlation among all three recoveries for vastus medialis and vastus intermedius muscle VM
1 versus 2
2 versus 3
1 versus 3
S0
0.93**
0.72
0.78
j
0.96**
0.88*
0.92*
b
0.86*
0.58
0.73
a
0.59
0.58
0.54
S0
0.56
0.63
0.94**
j b
0.82** 0.94**
0.81* 0.77*
0.96** 0.84*
a
0.89**
0.77*
0.97**
VI
* Represents a significant correlation of p \ 0.05, ** p \ 0.01
Fig. 6 Fitted curves of mean parameter estimates for vastus medialis (VM, top panel) and vastus intermedius (VI, lower panel) from all participants (parameters shown in Table 2) for recoveries 1, 2, and 3
Table 2 Parameters of the BOLD recovery curves for the two muscles that were recruited in the exercise as obtained from the fitted sigmoid function (Fig. 3) Parameter
Recovery 1 (n = 10)
Recovery 2 (n = 10)
Recovery 3 (n = 6)
Discussion
Vastus medialis muscle Baseline intensity, S0 (au)
0.97 ± 0.06
Change in baseline intensity (j)
0.14 ± 0.06
Inflection point b (s)
22.1 ± 8.5
18.5 ± 7.4
19.8 ± 11.2
Response time a (s)
9.6 ± 3.6
10.8 ± 3.7
11.7 ± 4.6
0.95 ± 0.03
0.95 ± 0.04
0.94 ± 0.06
Goodness of fit r2
0.97 ± 0.07 0.16 ± 0.07
0.94 ± 0.08 0.15 ± 0.09
Vastus intermedius muscle Baseline intensity, S0 (au)
0.97 ± 0.05
0.98 ± 0.07
0.98 ± 0.05
Change in baseline intensity j Inflection point b (s)
0.19 ± 0.07
0.20 ± 0.07
0.16 ± 0.05
24.6 ± 13.4
20.9 ± 13.4
20.5 ± 10.1
Response time a (s)
10.0 ± 4.4
10.3 ± 4.7
7.2 ± 3.4
0.96 ± 0.02
0.96 ± 0.02
0.92 ± 0.05
Goodness of fit r
2
the mean change in baseline BOLD intensity (j) was 0.15 ± 0.068, and the inflection point (b) was at 20.2 ± 8 seconds (Table 2). The mean response time for vastus intermedius (a) was 9.4 ± 4.3 seconds, the mean change in baseline BOLD intensity (j) was 0.18 ± 0.07, and the inflection point (b) was at 22.2 ± 12.3 seconds. Intraparticipant test–retest repeatability was significantly correlated (p \ 0.05) for all recoveries for parameters j, b, and a in the vastus intermedius muscle (Table 3). Baseline signal (S0) in the vastus intermedius correlated only for recoveries 1 and 3. Vastus medialis recovery was significantly correlated among all recoveries for signal recovery range j, and only in recoveries 1 and 2 for baseline signal intensity S0 and half-time for recovery b.
Values are mean ± SD (n = 10 for the first two recoveries; n = 6 for the third recovery, as four curves were contaminated by motion and could not be fitted)
This study was designed to characterize quadriceps BOLD response following submaximal bouts of leg extension exercise performed using an MRI-compatible ergometer. Consistent temporal changes in BOLD SI were observed in the 10 participants who performed submaximal exercise during MRI. These changes fit well to a four-parameter sigmoid curve (r2 = 0.95 ± 0.04) in both the vastus medialis and vastus intermedius muscles (Table 2; Fig. 6). There are only a few studies in the literature involving quantification of BOLD response following an applied stimulus [13–15, 24], and all of these have modeled the BOLD signal in the calf region, either following exercise [13, 14, 24] or during the reperfusion after cuff-induced ischemia [15]. To our knowledge, no study has modeled the BOLD response in the quadriceps region post-exercise without induced ischemia. Our BOLD time course was well-characterized by a four-parameter sigmoid function
123
Magn Reson Mater Phy
similar to that of McGrath and colleagues, who used a fourparameter sigmoid curve to fit the T1 signal response in skeletal muscles in the trunk region to 100 % oxygen inhalation [25]. It is interesting to note that previous BOLD modeling studies performed in the calf region have utilized a gamma variate function [14, 15], either in combination with sigmoidal and linear functions [15] or in addition to a linear curve [14]. Our post-exercise quadriceps BOLD response (Figs. 4, 6) did not demonstrate the sharp increase and rapid dampening characteristic of the gamma function commonly used in calf studies [14, 15]. It is possible that the quadriceps muscle exhibits a different BOLD signal recovery following an oxygen-depletion stimulus from the response demonstrated by the calf muscle [15]. However, our fitted parameters were similar to those in the calf region demonstrated in previous studies. For example, the change in baseline BOLD intensity (j) was 0.16 in both vastus medialis and vastus intermedius muscles (Table 2), which was similar to that observed in the lateral gastrocnemius muscle (0.18) following 2.5 minutes of plantar flexion exercise [14]. Since the total measurement time in our study was 120 seconds following each bout of exercise, our model did not require the addition of a linear function to account for the signal drift that has been observed over longer measurement durations [14, 15]. A longer period of BOLD MRI data acquisition following exercise could result in a dampening of BOLD SI to baseline values, as Fig. 4 demonstrates that signal recovery increases to a plateau approximately 15 % higher than baseline. Nevertheless, our data demonstrate that 120 seconds of imaging following submaximal exercise without induced ischemia may be sufficient to evaluate the acute recovery of the BOLD SI response following a short bout of exercise, as all of our participants’ BOLD SI recovery reached a clear plateau within this time frame. Quantifying the BOLD MRI response immediately following exercise is crucial, as physiological responses such as increased blood flow and tissue perfusion can be impaired in chronic diseases. A consequence of reduced perfusion is a reduction in the ability of tissue to remove waste products, resulting in decreased pH, a more rapid onset of fatigue, and longer recovery times from exercise. Research has demonstrated that patients with respiratory diseases such as cystic fibrosis exhibit longer phosphocreatine recovery times following quadriceps exercise, contributing to exercise intolerance [18]. In addition, individuals with Turner syndrome exhibit differences in pH following short bursts of quadriceps exercise in comparison to healthy controls [19]. In this study, we selected an exercise recovery protocol that was similar to the paradigms that have been applied to study 31P magnetic resonance spectroscopy in children
123
with cystic fibrosis [18] and Turner syndrome [19]. Since consistent temporal changes were observed in the current study, we anticipate that comparisons between BOLD MRI signal responses and phosphocreatine and pH recovery can be carried out in subsequent studies. The use of 31P-MRS in conjunction with BOLD MRI will provide data to elucidate the pathophysiology of skeletal muscle vascular or metabolic dysfunction. Thus, important questions related to impairment of oxygen delivery versus oxygen utilization could be answered using this paradigm. Approaches that combine MR imaging techniques, such as the simultaneous use of 31P-MRS and arterial spin labeling (ASL) to assess muscle energy metabolism and perfusion, respectively [26– 30], and31P and 1H spectroscopy to investigate muscle oxygenation supply and utilization [30–32], have been carried out to investigate pathological conditions [28]. The BOLD response is modulated by numerous parameters and is dependent upon tissue oxygenation, blood fractional volume, capillary network architecture, and orientation relative to the main magnetic field. Since a complex set of factors modulates the BOLD response, it is difficult to identify the physiological factors behind our observed quadriceps recovery following submaximal exercise without the use of complementary imaging approaches. To examine intracellular oxygenation, for instance, myoglobin desaturation can be probed using 1HMRS, as others have done previously [33–35]. In future studies, we plan to use a combination of simultaneous BOLD signal acquisition with NIRS data to measure hemoglobin saturation in order to quantify oxygen consumption and reoxygenation from vascular inflow. There were several limitations to this study. Subsequent experiments should include longer post-exercise recovery following the third exercise bout in order to test whether BOLD SI returns to baseline levels. Furthermore, significant intra-subject variability as measured by non-significant test–retest reliability was present among recoveries in the vastus medialis muscle, while significant reliability was present in the vastus intermedius (Table 3). This suggests that the vastus intermedius muscle may be a more reliable region of interest to evaluate BOLD response in the quadriceps following acute exercise. Additionally, there was high variance in the four parameters of BOLD SI signal recovery response among the different participants The change in signal intensity and maximal hyperemic value from baseline (j) could be affected by vascular volume differences due to increased capillarization caused by variations in aerobic activity levels [36]. Physically active subjects have demonstrated a greater BOLD hyperemic response in the anterior tibialis muscle following maximal ankle dorsiflexion compared to inactive subjects [37]. We observed a high variability in physical activity levels (1–9 hours) among participants in the week prior to
Magn Reson Mater Phy
testing, as demonstrated in Table 1. While none of our participants were trained athletes, if some subjects were more physically active than others, this could have contributed to the variability in the observed BOLD response. In addition, the muscle fiber composition of each subject may be different, leading to differences among subjects in the metabolic cost to generate ATP. Differences in ATPgenerating capacity of the mitochondria may affect the half-time to recovery for the BOLD response (b). Increased ATP regeneration efficiency would result in more rapid reoxygenation of oxyhemoglobin, as less oxygen would be used to regenerate ATP for synergistic muscle groups. As such, further experiments are needed to characterize the difference in quadriceps BOLD response between trained and untrained subjects in order to understand the interindividual variability inherent in these measurements. Conclusion In conclusion, this study presents a method to evaluate quadriceps BOLD response following submaximal bouts of leg extension exercise without accompanying ischemia. A four-parameter sigmoidal function was employed to model quadriceps BOLD time course in response to an inmagnet exercise protocol using an MRI-compatible ergometer. In future studies, we propose a comparison of simultaneous BOLD signal acquisition with NIRS data to measure hemoglobin saturation. These measurements will enable calculation of the post-exercise relative contribution of vascular inflow versus tissue oxygenation in the quadriceps muscle in our sigmoid model. This has already been done in the calf muscles of healthy volunteers and patients using an ischemia-hyperemia paradigm [13]. Application of our model may aid in understanding diseases that exhibit impaired skeletal muscle function during exercise and recovery, such as peripheral arterial occlusive disease, metabolic disorders, and mitochondrial myopathies. Conflict of interest conflict of interest.
The authors each declare that they have no
Ethical standards Our hospital research ethics board approved the study, and all human study was performed in accordance with the ethical standards created by the Declaration of Helsinki and its later amendments. Prior to study inclusion, all participants read and signed informed consent documentation.
References 1. Jacobi B, Bongartz G, Partovi S et al (2012) Skeletal muscle BOLD MRI: from underlying physiological concepts to its usefulness in clinical conditions. J Magn Reson Imaging 35(6): 1253–1265
2. Noseworthy MD, Davis AD, Elzibak AH (2010) Advanced MR imaging techniques for skeletal muscle evaluation. Semin Musculoskelet Radiol 14(2):257–268 3. Partovi S, Karimi S, Jacobi B et al (2012) Clinical implications of skeletal muscle blood-oxygenation-level-dependent (BOLD) MRI. Magn Reson Mater Phy 25:251–261 4. Noseworthy MD, Bulte DP, Alfonsi J (2003) BOLD magnetic resonance imaging of skeletal muscle. Semin Musculoskelet Radiol 7(4):307–315 5. Ledermann HP, Schulte AC, Heidecker HG et al (2006) Blood oxygenation level-dependent magnetic resonance imaging of the skeletal muscle in patients with peripheral arterial occlusive disease. Circulation 113:2929–2935 6. Schulte AC, Aschwanden M, Bilecen D (2008) Calf muscles at blood oxygen level-dependent MR imaging: aging effects at postocclusive reactive hyperemia. Radiology 247(2):482–489 7. Partovi S, Schulte AC, Aschwanden M et al (2012) Impaired skeletal muscle microcirculation in systemic sclerosis. Arthritis Res Ther 14:R209 8. Jacobi B, Schulte AC, Partovi S et al (2013) Alterations of skeletal muscle microcirculation detected by blood oxygenation level-dependent MRI in a patient with granulomatosis with polyangiitis. Rheumatology 52:579–581 9. Partovi S, Aschwanden M, Jacobi B et al (2013) Correlation of muscle BOLD MRI with transcutaneous oxygen pressure for assessing microcirculation in patients with systemic sclerosis. J Magn Reson Imaging 38(4):845–851 10. Partovi S, Schulte AC, Staub D et al (2013) Correlation of skeletal muscle blood oxygenations level-dependent MRI and skin laser Doppler flowmetry in patients with systemic sclerosis. J Magn Reson Imaging. doi:10.1002/jmri.24503 11. Damon BM, Wadington MC, Hornberger JL, Lansdown DA (2007) Absolute and relative contributions of BOLD effects to the muscle functional MRI signal intensity time course: effect of exercise intensity. Magn Reson Med 58:335–345 12. Meyer RA, Towse TF, Reid RW, Jayaraman RC, Wiseman RW, McCully KK (2004) BOLD MRI mapping of transient hyperemia in skeletal muscle after single contractions. NMR Biomed 17:392–398 13. Towse TF, Slade JM, Ambrose JA, DeLano MC, Meyer RA (2011) Quantitative analysis of the postcontractile blood-oxygenation-level-dependent (BOLD) effect in skeletal muscle. J Appl Physiol 111(1):27–39 14. Davis AD, Noseworthy MD (2013) Consistency of post-exercise skeletal muscle BOLD response. In: Proceedings of the 21st Annual Meeting of ISMRM, Salt Lake City, 6496 15. Schewzow K, Andreas M, Moser E, Wolzt M, Schmid AI (2013) Automatic model-based analysis of skeletal muscle BOLD-MRI in reactive hyperemia. J Magn Reson Imag 38:96–969 16. Aschwanden M, Patrovi S, Jacobi B et al (2014) Assessing the end-organ in peripheral arterial occlusive disease-from contastenhanced ultrasound to blood-oxygen-level-dependent MR imaging. Cardiovasc Diagn Ther 4(2):165–172 17. Reid RW, Foley JM, Jayaraman RC, Prior BM, Meyer RA (2001) Effect of aerobic capacity on the T2 increase in exercised skeletal muscle. J Appl Physiol 90(3):897–902 18. Wells GD, Wilkes DL, Schneiderman JE et al (2011) Skeletal muscle metabolism in cystic fibrosis and primary ciliary dyskinesia. Pediatr Res 69(1):40–45 19. Wells GD, O’Gorman CS, Rayner T et al (2013) Skeletal muscle abnormalities in girls and adolescents with turner syndrome. J Clini Endocrinol Metab 98(6):2521–2527 20. Massie BM, Conway M, Rajagopalan B et al (1988) Skeletal muscle metabolism during exercise under ischemic conditions in congestive heart failure. Evidence for abnormalities unrelated to blood flow. Circulation 78(2):320–326
123
Magn Reson Mater Phy 21. Jenkinson M, Bannister P, Brady JM, Smith SM (2002) Improved optimisation for the robust and accurate linear registration and motion correction of brain images. Neuroimage 17(2):825–841 22. Jenkinson M, Beckmann CF, Behrens TE, Woolrich MW, Smith SM (2002) FSL Neuroimage 62:782–790 23. Cox RW (1996) AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Comput Biomed Res 29(3):162–173 24. Davis AD, Noseworthy MD (2013) Haemoglobin-derived curve fitting to post-exercise muscle BOLD data. In: Proceedings of the 21st Annual Meeting of ISMRM, Salt Lake City. p 5311 25. McGrath DM, Naish JH, O’Connor JPB et al (2008) Oxygeninduced changes in the longitudinal relaxation times in skeletal muscle. Magn Reson Imaging 26(2):221–227 26. Carlier PG, Brillault-Salvat C, Giacomini E, Wary C, Bloch G (2005) How to investigate oxygen supply, uptake, and utilization simultaneously by interleaved NMR imaging and spectroscopy of the skeletal muscle. Magn Reson Med 54:1010–1013 27. Carlier PG, Bertoldi D (2005) In vivo functional NMR imaging of resistance artery control. Am J Physiol Heart Circ Physiol 288:H1028–H1036 28. Vidal G, Giacomini E, Wary C et al (2003) Combined ASL perfusion imaging, BOLD imaging and 31P NMR spectroscopy of the leg in a rat model of peripheral arteriopathy. Proc 11th Meeting Int Soc Magn Reson Med, Toronto p. 1692 29. Baligand C, Gilson H, Me´nard JC et al (2010) Functional assessment of skeletal muscle in intact mice lacking myostatin by concurrent NMR imaging and spectroscopy. Gene Ther 17:328–337
123
30. Duteil S, Bourrilhon C, Raynaud JS et al (2004) Metabolic and vascular support for the role of myoglobin in humans: a multiparametric NMR study. Am J Physiol Regul Integr Comp Physiol. 287:R1441–R1449 31. Vanderthommen M, Duteil S, Wary C et al (2003) A comparison of voluntary and electrically induced contractions by interleaved 1 H- and 31P-NMRS in humans. J Appl Physiol 94:1012–1024 32. Brillault-Salvat C, Giacomini E, Wary C et al (1997) An interleaved heteronuclear NMRI-NMRS approach to non-invasive investigation of exercising human skeletal muscle. Cell Mol Biol 43:751–762 33. Richardson RS, Newcomer SC, Noyszewski EA (2001) Skeletal muscle intracellular PO2 assessed by myoglobin desaturation: response to graded exercise. J Appl Physiol 91:2679–2685 34. Richardson RS, Noyszewski EA, Kendrick KF, Leigh JS, Wagner PD (1995) Myoglobin O2 desaturation during exercise. Evidence of limited O2 transport. J Clin Invest 96:1916–1926 35. Richardson RS, Noyszewski EA, Leigh JS, Wagner PD (1998) Lactate efflux from exercising human skeletal muscle: role of intracellular Po2. J Appl Physiol 85:627–634 36. Shono N, Urata H, Saltin B et al (2002) Effects of low intensity aerobic training on skeletal muscle capillary and blood lipoprotein profiles. J Atheroscler Thromb 9(1):78–85 37. Towse TF, Slade JM, Meyer RA (2005) Effect of physical activity on MRI-measured blood oxygen level-dependent transients in skeletal muscle after brief contractions. J Appl Physiol 99(2):715–722
