Physiological responses of human skeletal muscle to acute blood flow restricted exercise assessed by multimodal MRI

Abstract

Important physiological quantities for investigating muscle hypertrophy include blood oxygenation, cell swelling, and changes in blood flow. The purpose of this study was to compare the acute changes of these parameters in human skeletal muscle induced by low-load (20% 1-RM) blood flow-restricted (BFR-20) knee extensor exercise compared with free-flow work-matched (FF-20_WM) and free-flow 50% 1-RM (FF-50) knee extensor exercise using multimodal magnetic resonance imaging (MRI). Subjects (n = 11) completed acute exercise sessions for each exercise mode in an MRI scanner, where interleaved measures of muscle R₂ (indicator of edema), $R_2^{\prime }$ (indicator of deoxyhemoglobin), macrovascular blood flow, and diffusion were performed before, between sets, and after the final set for each exercise protocol. BFR-20 exercise resulted in larger acute decreases in R₂ and greater increases in cross-sectional area than FF-20_WM and FF-50 (P < 0.01). Blood oxygenation decreased between sets during BFR-20, as indicated by a 13.6% increase in $R_2^{\prime }$ values (P < 0.01)), whereas they remained unchanged for FF-20_WM and decreased during FF-50 exercise. Quadriceps blood flow between sets was highest for the heavier load (FF-50), averaging 305 mL/min, and lowest for BFR-20 at 123 ± 73 mL/min until post-exercise cuff release, where blood flow rates in BFR-20 exceeded both FF protocols (P < 0.01). Acute changes in diffusion rates were similar for all exercise protocols. This study was able to differentiate the acute exercise response of selected physiological factors associated with skeletal muscle hypertrophy. Marked differences in these parameters were found to exist between BFR and FF exercise conditions, which contribute to explain the anabolic potential of low-load blood flow restricted muscle exercise.

NEW & NOTEWORTHY Acute changes in blood flow, diffusion, blood oxygenation, cross-sectional area, and the “T₂ shift” are evaluated in human skeletal muscle in response to blood flow-restricted (BFR) and conventional free-flow knee extensor exercise performed in an MRI scanner. The acute physiological response to exercise was dependent on the magnitude of load and the application of BFR. Physiological variables changed markedly and established a steady state rapidly after the first of four exercise sets.

INTRODUCTION

Despite the vital role that skeletal muscle mass plays in metabolic health, longevity, and the ability to perform activities of daily living, the fundamental triggering mechanisms that stimulate muscle hypertrophy are not fully understood. Presently, it is recommended to apply exercise loads corresponding to or above 70% one-repetition maximum (1-RM) to induce muscle hypertrophy based on numerous training studies (3). The higher degree of mechanical tension, metabolic fatigue, microdamage, and inflammation associated with the use of heavier exercise loads has been argued to stimulate or upregulate the humoral and myocellular signaling pathways to hypertrophy. However, there has been compelling evidence that performing low-load strength exercise with concomitant blood flow restriction (BFR) to the working muscle can elicit marked increases in muscle strength and skeletal muscle mass similar to that demonstrated by heavy-load strength exercise (32, 34, 61). Specifically, gains in muscular strength and muscle mass have been observed using lighter resistance loads combined with BFR in young (35, 43, 63), old (48, 64), and rehabilitating individuals (28, 45). The addition of BFR to resistance exercise has been observed to reduce blood flow, reduce muscle blood oxygenation, increase myocellular swelling, and increase acute muscle edema (6, 16, 36, 43, 49), which are key physiological parameters deemed relevant to the stimulation of muscle hypertrophy.

In recent years, muscle functional magnetic resonance imaging (mfMRI) techniques have been devised to evaluate oxygenation, blood flow, and edema/cell swelling (see Table 1 for overview). Yet mfMRI has rarely been applied during exercise paradigms known to elicit hypertrophy. Examining the specific conditions under which the triggering of skeletal muscle hypertrophy is known to occur is important to better understand the factors responsible for exercise-induced gain in muscle mass. This, in turn, is a prerequisite to use mfMRI as a tool to optimize the design of effective exercise-based rehabilitation protocols in clinical patient groups where maintenance of muscle tissue is impaired and use of heavy exercise loads often is contraindicated. MRI measures of particular interest include measurement changes of R₂ (also called “T₂ shift”), which is sensitive to edema/cell swelling and pH (10, 18), alterations in R₂ prime ($R_2^{\prime }$) that reflect blood oxygenation (14, 15, 65), phase contrast imaging (PC-MRI), which measures macrovascular blood flow (15), and diffusion tensor imaging (DTI), which measures the movement of tissue water (46). R₂ has been shown to provide a surrogate measure of skeletal muscle activity, having a linear correlation to skeletal muscle contractile work (18) and muscle glucose uptake (22). Although several biological factors may theoretically affect R₂, the acute changes induced in muscle have been demonstrated to primarily be due to changes in intracellular water concentrations and, to a lesser degree, acidification (10, 20). When monitoring myocellular swelling, R₂ measures can be combined with measures of cross-sectional area (CSA) obtained from the raw MRI images and diffusion. Diffusion by itself, on the other hand, is less specific, as it is affected by several biological factors that have been associated with exercise, including microfiber damage, membrane permeability, metabolite concentrations, and levels of inflammation (49, 66).

Table 1.

General overview of parameters quantified with magnetic resonance imaging

Parameters
R2 Irreversible dephasing of MRI signal (s) • ↑ Cell swelling ↓R2 (10, 11) • ↑ pH ↓R2 (11, 38) • ↑ Extracellular fluid ↓R2 (20) • ↑ Inflammation ↓R2 (39)	Diffusion: MD (mm2/s) • ↑Permeability ↑MD • ↓Viscosity ↑MD • ↑Inflammation ↑MD (67) • ↑Micro damage ↑MD • ↑Temperature ↑MD (42)
$R_2^{\prime }$ Reversible dephasing of MRI signal (s) • ↑Deoxyhemoglobin ↑ $R_2^{\prime }$ (14, 65)	Blood flow: Flow in veins and arteries within muscle tissue (mL/min) • ↑Blood flow ↑Macrovascular flow (15)

MD, mean diffusivity. The 4 quantitative MRI parameters measured in the present study are R₂, $R_2^{\prime }$, diffusion, and blood flow. The physiological parameters of interest in the present study are blood flow, tissue oxygenation, and edema. Macrovascular blood flow is measured and tissue oxygenation interpreted from $R_2^{\prime }$ values that are sensitive to deoxyhemoglobin concentrations. Low oxygenation can also affect R₂ values as pH decreases. Edema is evaluated by a combination of R₂ values (predominantly sensitive to intracellular fluid shifts) with diffusion and cross-sectional area (CSA). ↑Increase; ↓decrease.

This study aimed to monitor acute changes in muscle blood flow, blood oxygenation, cell swelling/edema, and diffusion in response to low-load BFR exercise, low-load exercise, and high-load exercise. It was hypothesized that, compared with control exercise, blood flow-restricted exercise would be expected to induce greater responses in terms of cell swelling/edema, tissue hypoxia, diffusion, and blood flow.

METHODS

Subjects.

Eleven recreationally active young men (age: 26 ± 4.5 yr; exercise/wk: 4.8 ± 2.7 h) volunteered to participate in the study. The Ethics Committee of the Capital Region of Denmark approved the study (H-1-2013-146), and the participants gave written, informed consent to participate, in accordance with the principles of the Declaration of Helsinki.

Study design.

When positioned in the scanner (Philips 3T Achieva dStream, Best, The Netherlands), and following the completion of baseline scans, subjects performed three protocols of unilateral knee extension exercise [range of motion (ROM): 10–70° flexion, 0° = full extension], wearing a custom-made boot designed for applying lead weight blocks to provide the external exercise load (see setup illustration in Fig. 1). First, four sets of BFR exercise were each performed to task failure (unable to complete a full ROM) using a load of 20% of 1-RM (BFR-20). BFR of the working musculature was accomplished using an 11-cm-wide pneumatic cuff applied around the proximal thigh and inflated to 110 mmHg for partial occlusion (29) using a computerized tourniquet system (Zimmer, Dover, OH). The cuff was inflated immediately before the first set and deflated immediately after the fourth exercise set. The weight boot was then moved to the contralateral leg for the remaining two exercise protocols. The second protocol was free-flow work-matched knee extensor contractions (FF-20_WM), where participants completed the same number of repetitions and used the same relative load (20% of 1-RM) as in the BFR condition. The third protocol involved four sets of free-flow exercise, each performed to task failure with 50% of 1-RM (FF-50). Overall, knee extensor repetitions were performed at the same tempo, and all sets were interspaced by a 50-s rest in which MRI scans to assess macrovascular blood flow, R₂ and $R_2^{\prime }$ were performed. After each exercise protocol, the same measures along with diffusion scans were repeated. The right-left assignment of exercise protocols was balanced such that half of the subjects (n = 6) performed the BFR-20 exercise with the right leg and both free-flow protocols (FF-20_WM and FF-50) on the left leg while the other half (n = 5) performed BFR-20 on the left leg and the free-flow protocols on the right leg. On the experimental day, participants completed BFR-20, FF-20_WM, and FF-50 in that order. All participants agreed to refrain from vigorous activity for ≥24 h before participating in the present study procedures.

Fig. 1.

Test setup in the scanner. Three exercise sessions, each consisting of 4 sets, were performed in the scanner by lifting a boot containing lead weights. Subjects were equipped with an inflatable cuff around the thigh of 1 leg to perform blood flow-restricted exercise (BFR-20) with an exercise weight corresponding to 20% 1-repetition maximum (1-RM) continuing until failure each set. The 2nd exercise protocol [free-flow work-matched (FF-20_WM)] was performed on the contralateral leg without cuff (i.e., free-flow exercise) and was work matched to the BFR-20 exercise. The 3rd exercise protocol (FF-50) involved an exercise load of 50% 1-RM performed until failure for the free-flow (FF) leg, i.e., without cuff application. MRI scans to acquire a $R_2^{*}$, a R₂, and a phase contrast (PC) data set were performed between each set of exercise while the subject rested with a total scan time of ∼1 min. All measures, including diffusion data (DWI), were collected at baseline, between the 3 exercise protocols, and after completion of all exercise protocols.

Maximal muscle strength testing and familiarization.

Participants visited the laboratory for evaluation of five-repetition maximum (5-RM) muscle strength, from which 1-RM loads were estimated (7), and familiarization to the BFR exercise protocol ≥48 h before the experimental day. Following a standardized 5-min warm-up on a cycle ergometer, subjects completed a unilateral (estimated from right leg) 5-RM test in the supine position (Fig. 1). After completing the RM test, all subjects completed two sets of BFR exercise to failure, with a load corresponding to 20% of 1-RM with their right leg for familiarization.

MRI sequences.

Diffusion, R₂ and $R_2^{\prime }$ data were acquired over the same field of view (475 × 215 mm) covering a transaxial slice of both legs at mid-femur and included fat suppression with spectral adiabatic inversion recovery (SPAIR). Scans were performed on a Philips 3T Achieva dStream scanner with a 16-channel anterior coil strapped over the thighs of both legs. After each exercise set, the placement of the coil was controlled to avoid displacement.

R₂ data were acquired using a turbo spin echo sequence (TSE factor 20) with a repetition time (TR) of 1,500 ms and two echo times (TE1 = 9.7 ms, TE2 = 80 ms). One 8-mm slice was acquired with a 68 × 160 matrix and reconstructed to a voxel size of 1.85 × 1.85 mm² and a sense factor of 1.5 giving a scan time of ∼12 s.

R₂ maps were calculated as

$$R_2=\frac{\text{ln}\left({\text{S}}_1/{\text{S}}_2\right)}{\Delta TE},$$

where S₁ and S₂ indicate the signal from the first and second echoes, respectively.

R₂ prime ($R_2^{\prime }$) data were calculated from R₂ and R₂ star ($R_2^{*}$) maps. $R_2^{*}$ maps were obtained using a gradient-echo sequence with a TR of 266 ms, 10 echo times of TE1 = 8.0 ms and ΔTE = 7.4 ms, a flip angle of 40°, and a fast-field echo readout with an EPI factor of 5 with a scan time of ∼2 s per dynamic. Three slices of 8 mm covering both legs at the mid-femur with a field of view of 475 × 215 mm² were acquired in a 160 × 69 matrix and reconstructed to a pixel size of 1.85 mm. SPAIR fat suppression was applied. $R_2^{*}$ maps were calculated by least-squares minimization of the following equation:

$${\text{S}}_{\text{i}}={\text{S}}_0\times {\text{e}}^{-R_2^{*}\cdot T{E}_{\text{i}}}$$

The central slice of the $R_2^{*}$ maps was used to calculate $R_2^{\prime }$, where the extra slices were to ensure the ability to coregister $R_2^{*}$ and R₂ maps in case of movement. $R_2^{\prime }$ was calculated as the difference between the successive $R_2^{*}$ and R₂ measures:

$$R_2^{\prime }=R_2^{*}-R_2$$

Muscle blood flow was calculated using phase contrast MRI (PC-MRI). A single slice was acquired using a TR/TE of 10.7/6.5 ms and turbo field echo factor of 50 with a 256 × 132 acquisition matrix reconstructed to voxel dimensions of 1.6 × 1.6 × 6 mm³. Velocity encoding was 100 cm/s. Macrovascular arterial and venous vessels within the quadriceps muscles were identified using an adaptive threshold of 10% above average absolute water velocity of the quadriceps combined with k means clustering [5 bins and unrestricted cluster size using velocity maps and magnitude images as input (13, 69)]. The acquisition time was ∼10 s. To correct for differences in muscle size between subjects, all values for macrovascular blood flow were normalized to a quadriceps muscle mid-thigh cross-sectional area (CSA) of 100 cm² by multiplying flow values by a factor of 100 cm²/CSA.

Diffusion-weighted data were acquired with six directions and 10 b values (0, 100, 200, 250, 300, 350, 400, 450, 550, 700, 850, 1,000, and 1,300 s/mm²) using a TR/TE of 1,100 ms/98 ms and gradient (G) of 48 mT/m. Data were acquired with an EPI factor of 69 in a matrix of 160 × 69 reconstructed to a 256 matrix with 1.86 × 1.86 mm² voxels in three 5-mm slices. The signal-to-noise ratio (SNR) was tested for all regions of interest (ROIs) using the signal from two subsequent measures (S₁ and S₂) in the following equation (53).

$$\text{SNR}={|}_{\text{ROI}}\frac{\text{mean}\left({\text{S}}_1+{\text{S}}_2\right){|}_{\text{ROI}}}{\sqrt{2}\text{std}\left({\text{S}}_1-{\text{S}}_2\right){|}_{\text{ROI}}}$$

S_ROI is the diffusion weighted signal for a region of interest in arbitrary units. Maps of the mean apparent diffusion (MD) and fractional anisotropy (FA) were calculated from a single fit to the diffusion images consisting of b values from 200 to 700 s/mm² using the Diffusion Toolbox version 3.0 from FSL (5). Data from the b values >700 were acquired for an analysis outside the scope of the present study.

Statistical analysis.

Values for R₂, $R_2^{\prime }$, MD, and FA were obtained from regions of interest (ROI) drawn in the vastus lateralis, vastus medialis, vastus intermedius, and rectus femoris muscles of the quadriceps. Reported quadriceps R₂ and $R_2^{*}$ values are the mean of the four obtained ROI mean values from parametric maps. ROIs were drawn by the same observer, avoiding boundaries with fat, bone, and blood vessels on the DTI and $R_2^{*}$ data sets, transposing the latter onto R₂ images. Blood flow and CSA were calculated from a single ROI encompassing the entire exercising quadriceps muscle group. A mixed-effects model method was used to perform an ANOVA test for differences of measured changes in R₂, $R_2^{\prime }$, MD, FA CSA, and blood flow between the three exercise protocols.

A second analysis was performed to test for correlations between the acute changes of the measured parameters after exercise using linear regression, where goodness of fit was evaluated with a Pearson’s adjusted R² value. Linear regressions having an adjusted R² value >0.2 and a P value of <0.05 (corrected for 24 comparisons) were considered a minimal criterion for reporting correlations. A Bonferroni correction for multiple comparisons was applied to P values. Image coregistration, ROI analysis, calculations, and statistical analysis were performed with software created in MATLAB 2016b (MathWorks, Natick, MA).

RESULTS

All subjects completed each set of the FF-50 and BFR-20 protocols until failure and completed the FF-20_WM protocol without any apparent signs of subjective fatigue. The mean 1-RM of subjects’ right leg was 45.4 ± 8.9 kg. The mean number of repetitions before failure for the first set of BFR-20 was 94 ± 24 repetitions in 160 ± 65 s, which decreased to 35 ± 16 repetitions in 58 ± 34 s on the fourth set (P < 0.001). For FF-50, the number of repetitions decreased from 27 ± 8 in 64 ± 27 s for the first set to 15 ± 7 in 58 ± 34 s on the fourth set (P < 0.001). During FF-20wm, subjects repeated the number of repetitions from BFR-20 in 152 ± 84 s in the first set and 51 ± 36 s in the fourth set.

Myocellular swelling/edema in exercised quadriceps (R₂ values).

Baseline R₂ values were 21.3 ±  1.2 s, with no significant differences between the legs assigned to BFR or FF exercises. All exercise protocols induced edema/pH decreases, as reflected by significant decreases in R₂ values compared with baseline (P < 0.001) in quadriceps muscle after the first exercise set (Figs. 2 and 3). BFR-20 exercise induced significantly larger decreases in R₂ than both FF-20_WM and FF-50 exercise (P < 0.01). After the first set of BFR-20 exercise, R₂ values decreased progressively to −0.7 ± 0.2 s over the subsequent sets of BFR-20 exercise (P < 0.01), including after the fourth set with cuff release. In contrast, no changes in R₂ values were observed in subsequent sets after the first set during free-flow exercise (FF-20_WM and FF-50 exercise protocol). Large spatial variation in activation was observed, especially after the FF-50 protocol (Fig. 2). R₂ values for either leg did not return to baseline levels within the scanning session (20 min post-FF-50 exercise protocol completion, P < 0.001).

Fig. 2.

Relative changes in R₂ after each set for the 3 exercise protocols in 1 subject. All 3 exercise protocols, low-load (20% 1-repetition maximum) blood flow-restricted exercise (BFR-20), free-flow work-matched (FF-20_WM), and free-flow medium load (FF-50), induced decreases in R₂ indicative of intracellular muscle swelling and pH changes. The largest decreases in R₂ were noted using the BFR exercise protocol. Rectus femoris demonstrated greater decreases using the FF-50 exercise protocol. Likewise, the adductor longus muscle demonstrated larger changes in response to FF-50 exercise but not BFR-20 and FF-20_WM exercise, where subjects lifted substantially lower loads.

Fig. 3.

Changes in R₂ values relative to baseline and after each exercise set for the 3 exercise protocols. A: changes in exercised quadriceps mean muscle R₂ for all subjects are presented for the 3 exercise protocols at repeated baseline, immediately after each of the 4 exercise sets (time elapsed: ∼5 s for $R_2^{\prime }$ and ∼30 s R₂ measures), and 3 min after exercise cessation. In the low-load (20% 1-repetition maximum) blood flow-restricted exercise (BFR-20) protocol, cuff release is after set 4 exercise stop; thereafter, MRI measures for set 4 are acquired. All protocols induced significant changes in R₂ values, which did not return to baseline within the given timespan. After 2 sets of BFR-20 exercise, R₂ continued to decrease, whereas R₂ stabilized during free-flow exercise [free-flow work-matched (FF-20_WM) and free-flow 50% 1-repetition maximum (1-RM) (FF-50)], indicating a plateauing in cell swelling/acidity in the latter exercise conditions. B: relative changes in $R_2^{\prime }$ values are presented for the same time points. BFR-20 exercise resulted in an increase in $R_2^{\prime }$ for all exercise sets, indicating markedly reduced blood oxygenation. The opposite was observed for FF-50 exercise, where $R_2^{\prime }$ decreased, indicating an increase in oxygenation. Although the mean and median $R_2^{\prime }$ values declined under baseline during FF-20_WM, the decrease was not significant. *Significant differences between measured values from BFR-20 and both free-flow protocols.

Blood oxygenation of exercised quadriceps ($R_2^{\prime }$ values).

Baseline $R_2^{\prime }$ values were 15.8 ± 3.5 s, with no significant differences between the leg to perform BFR or FF exercise. Blood oxygenation decreased during BFR-20 exercise, as reflected by an increase in $R_2^{\prime }$ values (P < 0.01), which was 13.6% above baseline on average after sets 1 to 3 (Fig. 3). Once the cuff was released after the fourth set, a decrease in $R_2^{\prime }$ was observed, indicating postexercise hyperemia (Fig. 3–4). In contrast, the FF-50 protocol induced an increase in blood oxygenation (P = 0.02) during exercise, with $R_2^{\prime }$ being reduced 10.4% below baseline when averaged after each set, whereas no changes in $R_2^{\prime }$ were observed during the FF-20_WM protocol. After cessation of exercise (and cuff release in the case of BFR-20), $R_2^{\prime }$ values were reduced following all exercise protocols (P < 0.001) (Fig. 4).

Fig. 4.

Mean change in $R_2^{\prime }$ for all subjects at baseline, immediately after each set, and 2 min after exercise. For free-flow (FF) exercise protocols, $R_2^{\prime }$ decreased from baseline between successive exercise sets, indicating increasing oxygenation. During the blood flow-restricted exercise using a load of 20% of 1-repetition maximum (BFR-20), $R_2^{\prime }$ exceeded baseline levels between exercise sets while the cuff remained inflated, indicating lower oxygenation. Following cuff release after set 4, $R_2^{\prime }$ declined in the BFR trained leg to reach values below post-FF exercise levels.

Macrovascular muscle blood flow.

All three exercise protocols led to increased arterial and venous blood flow in the exercising quadriceps (P < 0.001). During BFR-20 exercise (i.e., cuff inflation), arterial blood flow between sets was reduced by 52% and venous flow by 68% compared with mean flow during the free-flow protocols (Fig. 5). However, during each consecutive set of BFR-20 exercise, arterial and venous flow increased slightly at a mean rate of 23 ± 9 and 35 ± 13 mL/min per set, respectively, until cuff release (P = 0.03). Upon cuff release immediately after the fourth exercise set, venous blood flow increased 2.5-fold from 111 ± 114 mL/min after the third set to 278 ± 139 mL/min 2 min post-fourth set and cuff release (P < 0.001). Likewise, arterial blood flow increased 1.8-fold from 123 ± 73 mL/min to 220 ± 108 mL/min (P < 0.001). After 4 min of rest, arterial and venous blood flow had returned to baseline levels.

Fig. 5.

Arterial (A) and venous (B) flow in trained quadriceps at baseline after each set and 5 min postexercise for each of the 3 exercise protocols. During cuff inflation (i.e., pre-set 1 to post-set 4), both arterial and venous flow were lower in low-load (20% 1-repetition maximum) blood flow-restricted exercise (BFR-20) compared with free-flow work-matched (FF-20_WM) and free-flow 50% 1-RM (FF-50) protocols (P > 0.01). Arterial and venous blood flow were elevated with the heavier FF-50 protocol than the low-load FF-20_WM protocol (P < 0.01). After the 4th set of BFR-20 exercise (and cuff release), the arterial and venous blood flow exceeded the free-flow exercise protocols. For each time point: **significant differences (P < 0.05) between all 3 exercise protocols; *flow data for the BFR-20 protocol differs significantly from the free-flow exercises. Error bars represent ± SE.

Cross-sectional area of exercised quadriceps.

All exercise protocols gave rise to an acute increase in cross-sectional area (CSA) of the quadriceps muscle already after one set of exercise (P < 0.001; Fig. 6). CSA did not change significantly during the subsequent sets (sets 2–4) of the FF-20_WM and FF-50 exercise protocols. CSA continued to increase, however, by 1.3 ± 0.5% of baseline CSA per set for BFR-20 exercise after sets 2 and 3. The relative increase in CSA with BFR-20 exercise went from 7 ± 7% after the third set to 12 ± 8% after the fourth set with cuff release (P < 0.001). Increases in CSA after the fourth set were 7 ± 5% and 8 ± 5% for FF-50 and FF-20_WM, respectively, which were lower than BFR-20 (P < 0.05).

Fig. 6.

Mean change in cross-sectional area (CSA) of quadriceps at baseline, after each set of exercise, and 5 min postexercise. For all exercise protocols, the quadriceps CSA increased from baseline to the 1st set. Thereafter, there were no significant changes in CSA throughout the following sets, except for low-load (20% 1-repetition maximum) blood flow-restricted exercise (BFR-20), where CSA increased in response to cuff release following set 4 (values depicted at post-set 4). CSA of the quadriceps remained significantly increased compared with baseline after 1 min of rest for all exercise protocols; however, CSA in the BFR-20 leg maintained a larger CSA compared with the free-flow protocols. *Significant differences in CSA between BFR-20 protocol and the free-flow exercises. P < 0.05. Error bars represent ± SE.

Diffusion.

The acute changes in MD values were similar across the three exercise protocols. Specifically, MD for the exercised quadriceps increased by 0.20 mm²/s in response to BFR-20 exercise from a baseline value of 1.3 mm²/s (P < 0.01; Fig. 7). Likewise, quadriceps’ MD increased by 0.22 mm²/s after FF-20_WM exercise and by 0.20 mm²/s after FF-50 exercise from a baseline value of 1.23 mm²/s (P < 0.01). At 20 min after completing the last exercise protocol, MD was not significantly different from baseline in any muscle group.

Fig. 7.

Mean diffusivity (A and B) and fractional anisotropy (C and D) in the quadriceps and hamstring muscles of the blood flow-restricted (BFR) leg and free-flow (FF) leg at successive timepoints. Diffusion measures covering all muscles were taken at baseline, 1.5 min after each exercise protocol, and again after having rested for 20 min after the completion of all exercise protocols. After low-load (20% 1-repetition maximum) blood flow-restricted exercise (BFR-20), mean diffusivity (MD) of the quadriceps of the BFR leg increased from baseline (P < 0.05). At the same time point, a decrease in the MD of the hamstring muscles of the BFR leg and both muscle groups of the resting free-flow leg was observed (P < 0.001). After free-flow work-matched (FF-20_WM) and free-flow 50% 1-RM (FF-50) exercise, diffusion was elevated in the quadriceps of the exercised leg compared with baseline (P < 0.001). In the postexercise periods, MD was lower than baseline in the quadriceps of the BFR leg (P < 0.01), but not in the other muscle groups. After the BFR-20 exercise and cuff release, fractional anisotropy (FA) values increased in the hamstring muscles and the muscle groups of the opposing free-flow leg (P < 0.01). However, mean FA values in exercised quadriceps muscle (C) were not changed significantly by BFR-20 exercise. The free-flow exercise protocols (FF-20_WM and FF-50) were followed by lower FA values than the preceding BFR-20 time point and baseline (both P < 0.01). In the resting periods after the exercise protocols, FA values were elevated in the exercised quadriceps muscles of the BFR leg (P < 0.01) but not significantly different in any other muscle group. Values are presented as group means ± SE. *Data points significantly different from baseline; †significant differences between the FF and BFR leg (P < 0.05).

After completing BFR-20 exercise (including cuff release), MD values decreased on average by 0.2 mm²/s from baseline in all nonexercised muscles of both legs (P < 0.001; Fig. 7). This included the hamstring muscles of the cuffed BFR leg (P < 0.01) and the hamstring and quadriceps muscles of the noncuffed FF leg (P < 0.01). The same effect was not evident in nonexercised muscles after the FF-20_WM or FF-50 protocols.

Baseline diffusion along the muscle fiber direction (λ₁) was higher than diffusion radial to the fiber (λ_rad) for all muscle groups (P < 0.01; Fig. 8). Increases in diffusion after exercise did not alter the ratio between radial diffusion along the fiber as reflected by constant FA values. Although FA values remained unchanged in the exercised muscle after the BFR-20 protocol, FA values increased in the nonexercised muscles (P < 0.01). After free-flow exercise (FF-20_WM and FF-50), FA values decreased from the preceding post-BFR-20 time point (P < 0.01), although they were not reduced relative to baseline. The decrease in MD values of nonexercised muscles after the BFR-20 protocol (with cuff release) was more prevalent in the radial direction than along the direction of the fiber, as reflected by the respective increases in FA (P < 0.01; Figs. 7 and 8). After 20 min of rest following the completion of the last exercise protocol, FA values were elevated in the quadriceps muscles of the BFR leg (P = 0.04) but not in the hamstring muscle groups or quadriceps of the FF leg (Fig. 7). Mean subject SNR values for ROIs were 6.5 for the images with the b value of 200 s/mm² and 4.3 for 700 s/mm².

Fig. 8.

Diffusion along the direction of the fiber (λ₁; A) and radial (λ_rad; B) to the direction of the fiber. Diffusion along the muscle fiber direction (λ₁) was much higher than diffusion radial to the fiber (λ_rad) for all muscle groups (P < 0.001), as indicated by resting fractional anisotropy (FA) values >0.4 (FA = 0 is isotropic and FA = 1 is unidirectional diffusion). Increases in muscle mean diffusivity (MD) measured in both directions after exercise (regardless of protocol) were relatively similar along the direction of the fibers and radial to the fibers, which is the basis for the similar FA values postexercise (P < 0.01). However, after low-load (20% 1-repetition maximum) blood flow-restricted exercise (BFR-20), the decrease in MD in the nonexercised muscle was relatively larger for radial diffusion than for diffusion along the fiber; hence, a significant increase in FA values (P < 0.01).

Correlation analysis.

In the analysis of correlations between acute changes in measured parameters (R₂, $R_2^{\prime }$, CSA, FA, and MD) from exercise start until 5 min postexercise, only two correlations were found by using the criteria of an adjusted R² value >0.2 and a P value of <0.05 (corrected for 24 comparisons). Changes in FA correlated with changes in MD, with the best fit being ΔFA = −0.5 ΔMD + 5.5% (R² = 0.3, P = 0.03). Venous flow and arterial flow in the quadriceps were correlated as expressed by F_V = 1.1 F_A + 1.2 mL/min for free-flow and BFR exercise protocols (R² = 0.75, P = 0.02). Another two correlations were found to be specific to the exercise protocol. For BFR exercise, changes in R₂ values and CSA of the quadriceps were correlated ΔR₂ = −0.6ΔCSA + 12.4% (R² = 0.36, P = 0.03), but not for the free-flow exercise protocols. In contrast, the changes in R₂ values were weakly correlated with changes in MD values ΔR₂ = −0.2ΔMD − 10.5% (R² = 0.25, P = 0.02) for free-flow exercise (FF-20_WM and FF-50 protocols).

DISCUSSION

In this study, we examined acute changes in blood flow, diffusion, and the R₂ and $R_2^{\prime }$ relaxation times in human skeletal muscle in response to blood flow-restricted (BFR) exercise and compared them with conventional free-flow (FF) exercise. The different BFR and FF exercise protocols elicited highly differing changes in various MRI parameters, demonstrating that the physiological response to acute knee extensor exercise is dependent on both the magnitude of the load applied and the presence/absence of transient peripheral ischemia (BFR vs. FF). Marked changes were observed for most of the examined physiological variables after the first exercise set, whereas only minor or no changes were observed during the three following sets, indicating that a new steady state is rapidly established during exercise. As an exception, muscle CSA and R₂ continued to change from the first to the fourth set during the BFR-20 protocol.

BFR exercise elicited greater relative changes in R₂ and CSA, indicating a higher degree of cell swelling, as well as more marked elevations in $R_2^{\prime }$, indicating a higher degree of peripheral hypoxia, than free-flow exercise whether matched for work performed (FF-20_WM) or using heavier exercise loads (FF-50). Specifically, the relative increase in quadriceps CSA with BFR exercise almost doubled from post-third set to post-cuff release immediately after the fourth set. R₂ decreased progressively during the course of the BFR-20 exercise protocol, including upon cuff release, in contrast to the steady state observed across sets of free-flow exercises. Muscle blood flow and blood oxygenation were the parameters demonstrating the clearest differential response between the three exercise protocols. As expected, blood flow and tissue oxygenation were markedly reduced by BFR, while the cuff was inflated and quickly increased upon cuff release. In contrast, $R_2^{\prime }$ values were lowest and blood flow (both venous and arterial) was highest when performing FF-50 exercise, indicating higher tissue oxygenation than baseline. Acute changes in diffusion, MD, on the other hand, were similar across the BFR-20, FF-20_WM, and FF-50 protocols. Remarkably, BFR exercise caused a strong decrease in the movement of muscle water, MD, in nonexercised muscles of the same (cuffed) leg and the muscles of the opposing leg. Diffusion radial to the fiber orientation decreased relatively more than diffusion along the fiber, resulting in a more unidirectional diffusion, as quantified by increased FA values. This coincided with a decrease in CSA in the contralateral quadriceps, which means a reduction in axial area. Despite several parameters being vascular in nature, the observed changes in measured values appeared to occur quite independent of each other. Thus, only few associations emerged when exercise-related changes in various parameters were compared. First, arterial and venous blood flow were strongly correlated, as expected. Second, the observed increase in muscle MD was related to the concurrent decrease in R₂, hence strongly suggesting that increased diffusion is associated with myocellular swelling.

Although the acute physiological response to low-load, high-load, and BFR exercise is not fully elucidated, previous physiological study reports correspond well to the present findings. When activated under high and low loads, the quadriceps have been reported to quickly regain resting levels of oxygenation between successive sets of free-flow exercise as opposed to BFR exercise, where blood oxygenation remained low throughout the entire exercise protocol (6, 12, 30). When comparing similar BFR-20, FF-20_WM, and free-flow 80% 1-RM (FF-80) knee extensor exercise protocols, Ilet et al. (30) observed that tissue oxygenation (measured with NIRS) decreased during exercise with FF-80 and BFR-20 and, to a lesser degree, during FF-20_WM. They also observed that tissue oxygenation recovered to baseline between sets of FF-80 was slightly lower than baseline between sets of FF-20_WM and remained markedly reduced throughout the BFR protocol (30). Likewise, in the present study, blood oxygenation was estimated to be high between sets during free-flow exercise, whereas it was reduced throughout the BFR-20 protocol (Fig. 4). Using the conversion of Δ$R_2^{\prime }$ to the deoxyhemoglobin fraction (100% HbO₂) presented by Elder et al. (14), the fraction would be estimated as 39% during BFR-20 exercise. Notably, upon cuff release after set 4, BFR exercise led to a rapid increase in blood oxygenation, which exceeded that of the free-flow protocols.

It remains unclear how the long-term (i.e., weeks to months) muscular responses of BFR, low-load, or heavy-load resistance training differ mechanistically from one another. Heavy-load strength exercise has been shown to increase muscle strength, muscle mass, rapid force capacity [rate of force development (RFD)], and muscular power in both young and old individuals (1, 9). Likewise, BFR induces similar muscle growth, improved strength, and rapid force development despite shorter training duration and lighter workloads (34, 44, 61). This contradicts the argument of high mechanical muscle tension being the primary stimulus for hypertrophy and indicates that factors associated with metabolic stress and/or transient tissue ischemia may be equally important for increasing muscle mass (47, 49, 60). BFR has been shown to amplify the build-up of metabolites (47, 56, 60), which contributes to an increased osmotic pressure gradient between the extracellular and intracellular compartments that increase water flux into the cell. Increased cell swelling and elevated blood lactate due to blood flow-restricted exercise has previously been reported (30, 37, 68). Cell expansion is argued to affect intrinsic volume sensors in the cell-stimulating anabolic signaling cascades (37, 47, 55), which at least in part may explain the marked hypertrophy shown to occur following short periods of BFR resistance training (25, 34, 43, 62). The larger acute gains in muscle CSA and lower R₂ values observed during and immediately after BFR-20 exercise compared with free-flow exercise (FF-20_WM, FF-50) further indicate a higher degree of cell swelling and acidification with BFR exercise. Likewise, this study reports signs of higher metabolic fatigue in the form of more extensive hypoxic conditions, as indicated by higher $R_2^{\prime }$ values, during BFR exercise than the free flow protocols.

Conventional low-load strength exercise similar to the FF-20_WM protocol (i.e., sets not performed to contraction failure) has previously been shown to increase muscle mass and strength, albeit to a much lesser degree than heavy-load strength exercise (27). It is first when low-load strength exercise is performed to voluntary failure that increases in muscle mass and strength comparable with those induced by heavy-load strength exercise have been demonstrated (2, 16, 41). Low-load (FF-20_WM) and higher-load (FF-50) free-flow exercise did not demonstrate differential changes in R₂ or muscle CSA. However, blood flow was higher and oxygenation lower during free-flow exercise performed to failure (FF-50) than not to failure (FF-20_WM), i.e., in the latter case without fatigue. Finally, BFR exercise has been reported to induce positive exercise effects in muscle tissues of the body other than the trained limb (40). In the present study, we observed a restrictive response to BFR exercise in the nonexercised muscles, including all thigh muscles in the resting contralateral limb as well as the antagonist hamstring muscles of the cuffed leg, where the radial rate of diffusion was reduced by ∼25%. What, if any, role this effect may play for the adaptation to BFR exercise in nonexercised (“remote”) muscles remains unknown.

Changes in blood flow in the quadriceps measured in the present study were in the range of previous observations where exercise-induced increases in blood flow of 500–1000% have been reported (26, 52) and postexercise values returning to 20−100% above baseline by 3 min (8). Free-flow exercise macrovascular flow values in this study are coherent with previous peak flow measures of 250 mL·100 g⁻¹·min⁻¹ during knee extension exercise (4, 31) although the present study’s method is not tissue volume specific. Comparable (8–10%) acute increases in CSA (or measures of muscle thickness) have been reported previously (58) and have also been reported to be more pronounced with BFR than free-flow resistance exercise (16). Interestingly, when exercise bouts were repeated over several weeks, the observed differences between the acute transient increases in muscle thickness between BFR and free-flow exercise conditions diminished (16).

Reflecting acute alterations in diffusion, changes in FA and MD have been documented after exercise, with the greatest decrease measured shortly after exercise cessation (17, 19, 24, 67). Handgrip exercise, for example, induced an immediate 17% increase in the MD of the muscles of the forearm (17), and intense jumping induced a 7−9% increase in MD of calf muscles as well as an ≈30% decrease in FA values from 0.3 (24). In the latter, the FA and MD values recovered only partially by 6 h, which these authors attribute to inflammation and minute injuries of muscle fibers and muscle-cell membranes (24). In contrast, the present MD and FA values obtained in exercised muscles were found to return to baseline levels within 20 min following exercise cessation, with MD values even continuing to descend below baseline levels following the BFR protocol. Furthermore, fractional anisotropy (FA) was not found to be significantly different from baseline values, indicating that the movement of intramuscular water increased isotropically along and radial to the fiber direction.

Methodological considerations.

A number of potential methodological limitations may be listed for the present study. The FF-50 and FF-20_WM exercise protocols were performed in the same leg, separated by 12–15 min of rest. Consequently, the observed acute changes following FF-50 exercise may have included residual effects from the preceding FF-20_WM protocol. Furthermore, a weighted boot and limited range of knee joint motion were employed in the present study, which may alter the specific pattern of muscle recruitment and fatigue onset compared with standardized knee extensor exercise performed in an exercise machine, which may reduce the practical implications of the present observations. Second, several processes initiated by exercise can affect the parameters measured and, therefore, can be difficult to disentangle. For example, small R₂ changes 3 days after FF and BFR exercise have been reported (∆R₂ almost equal to −2 s), which the authors attributed to inflammation and microfocal myofiber damage (59). Likewise, FA and MD values have also been reported to be affected by other short-term effects such as temperature changes (42) and long-term effects such as post-exercise inflammation and muscle damage up to several days after exercise (24), although these reported changes are much less than the acute response observed in this study. Several technical considerations may also be mentioned for the obtained MRI data. In the present analysis R₂, $R_2^{\prime }$, and MD were assumed to be monoexponential processes. The R₂ relaxation both at rest and after exercise has been concluded to be well described by a single exponential decay function, suggesting that extravascular fluid of normal muscle is in relatively fast exchange with the larger intracellular fluid compartment, resulting in similar relaxation rates (10, 50). Multiexponential models to determine R₂ have, however, been found to provide significantly better determination, most notably by identifying a fraction of muscle water (8–15%) with fast relaxation (R₂ > 0.2 ms), which is believed to represent water in an environment bound to macromolecules (57). Calculation of MD can similarly be affected by assuming that diffusion attenuates MRI signal in a monoexponential relationship to the b value. Diffusion MRI is better utilized by also identifying the pseudo perfusion, fraction of free water, and fraction of slow-moving water in tissue where this study instead reduces contributions from the first two by using b values of 200–700 s/mm² (46, 54). The resting quadriceps muscle MD values of 1.23 mm²/s were lower than previous study reports of 1.44 ± 0.10 × 10⁻³ mm²/s (51) in thigh muscle and similar muscle groups such as the forearms 1.45 ± 0.09 × 10⁻³ mm²/s. However, when calculating MD using b values of a range similar (200, 250, 300, 350, 400, and 450 s/mm²) to the aforementioned studies, a mean MD value for all subjects of 1.43 ± 0.12 × 10⁻³ mm²/s was obtained. When investigating the effects of exercise, one has to keep in mind that the muscle perfusion and blood volume will increase, introducing changes to b_0–200 intensities. However, this influence can be eliminated by using a multiexponential fit of the tensor, excluding b values <150 s/mm² entirely, to avoid that the calculated diffusion is affected by pseudo diffusion. Consequently, in the present study, the reference b = 0 s/mm² images were not used in the fits performed to determine MD. Finally, the inclusion of the higher b values to the DTI acquisitions greatly reduces SNR (via longer echo times), which can lead to an overestimation of FA and, to a lesser extent, MD. This is a plausible explanation for the baseline muscle FA value of ∼0.5, which is higher than literature values ranging from 0.2 to 0.4 (46). A possible consequence of this would be that increases in SNR from reduced R₂ values could give reductions in FA and affect MD values.

Future perspectives.

Because the anabolic effects of BFR, load application, or fatigue development in exercise are still not fully understood, the use of multimodal MRI to assess various physiological variables with high temporal resolution during ongoing exercise represents a promising tool in decoding important mechanisms of adaptation while also allowing for the optimization of exercise. An improved physiological understanding of BRF exercise is likely to benefit patients, as this exercise modality requires less time per exercise bout and exposes joints, ligaments, and muscles to reduced compressive and shear forces while stimulating muscle growth and strength similarly to traditional heavy-load resistance exercise. The MRI sequences applied in the present study are clinically accessible and can be interleaved with other measures to further investigate other physiological parameters of interest such as oxygen tension (23) or intramuscular lactate concentrations. With the advent of PET/MRI, multimodal classification of muscle tissues or combinations that elucidate reactions to exercise in several tissues become possible (21, 22, 33). Furthermore, the present findings embody an interesting potential in the development of strategies to counteract muscle wasting in both healthy and clinical populations.

Conclusion.

Based on multimodal MRI, the present study was able to differentiate the acute exercise-induced response of selected physiological factors associated with skeletal muscle hypertrophy. As a novel observation, marked differences in these parameters were found to exist between blood flow-restricted (BFR) and unrestricted, i.e., free-flow (FF), exercise conditions. The present acute measurements of muscle blood flow, diffusion, and R₂ and $R_2^{\prime }$ relaxation times indicate that BFR exercise induces a higher degree of cell swelling/acidification and reduced oxygenation in the exercised muscle under conditions of restricted blood flow compared with unrestricted free-flow exercise, which contribute to explain the myogenic potential of low-load blood flow restricted muscle exercise in human skeletal muscle.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

B.T.H., S.K.H., U.L., U.F., P.A., H.B.W.L., and C.S. conceived and designed research; B.T.H., S.K.H., U.L., and C.S. performed experiments; B.T.H., S.K.H., and U.L. analyzed data; B.T.H., S.K.H., J.L.N., U.F., P.A., H.B.W.L., and C.S. interpreted results of experiments; B.T.H. prepared figures; B.T.H., S.K.H., U.L., and C.S. drafted manuscript; B.T.H., S.K.H., U.L., J.L.N., U.F., P.A., H.B.W.L., and C.S. edited and revised manuscript; B.T.H., S.K.H., U.L., J.L.N., U.F., P.A., H.B.W.L., and C.S. approved final version of manuscript.