Myofiber hypertrophy adaptations following 6 weeks of low-load resistance training with blood flow restriction in untrained males and females

Abstract

The effects of low-load resistance training with blood flow restriction (BFR) on hypertrophy of type I/II myofibers remains unclear, especially in females. The purpose of the present study is to examine changes in type I/II myofiber cross-sectional area (fCSA) and muscle CSA (mCSA) of the vastus lateralis (VL) from before (Pre) to after (Post) 6 wk of high-load resistance training (HL; n = 15, 8 females) and low-load resistance training with BFR (n = 16, 8 females). Mixed-effects models were used to analyze fCSA with group (HL, BFR), sex (M, F), fiber type (I, II), and time (Pre, Post) included as factors. mCSA increased from pre- to posttraining (P < 0.001, d = 0.91) and was greater in males compared with females (P < 0.001, d = 2.26). Type II fCSA increased pre- to post-HL (P < 0.05, d = 0.46) and was greater in males compared with females (P < 0.05, d = 0.78). There were no significant increases in fCSA pre- to post-BFR for either fiber type or sex. Cohen’s d, however, revealed moderate effect sizes in type I and II fCSA for males (d = 0.59 and 0.67), although this did not hold true for females (d = 0.29 and 0.34). Conversely, the increase in type II fCSA was greater for females than for males after HL. In conclusion, low-load resistance training with BFR may not promote myofiber hypertrophy to the level of HL resistance training, and similar responses were generally observed for males and females. In contrast, comparable effect sizes for mCSA and 1-repetition maximum (1RM) between groups suggest that BFR could play a role in a resistance training program.

NEW & NOTEWORTHY This is the first study, to our knowledge, to examine myofiber hypertrophy from low-load resistance training with blood flow restriction (BFR) in females. Although this type of training did not result in myofiber hypertrophy, there were comparable increases in muscle cross-sectional area compared with high-load resistance training. These findings possibly highlight that males and females respond in a similar manner to high-load resistance training and low-load resistance training with BFR.

INTRODUCTION

Resistance exercise training interventions are often implemented to promote neuromuscular adaptations, such as muscle hypertrophy and increased strength (1–5). However, older and clinical populations (6–9) or athletes in the early phases of rehabilitation (10–12) cannot safely and/or effectively perform resistance training exercises using high loads. Given the limitations in suitable exercise prescriptions, alternative resistance training methods are worthy of investigation and their effects on muscular size and strength should be explored.

Blood flow restriction (BFR) has gained traction in the scientific literature as being a viable exercise modality for promoting muscular adaptations (9, 13–24). The occlusion of blood flow during low-load resistance training contributes to greater hypertrophy than traditional low-load resistance training (25). Low-load resistance training with BFR is hypothesized to produce muscle hypertrophy through various mechanisms, which include, but are not limited to, metabolic stress (16) and increased higher-threshold motor unit recruitment (15). Greater heat shock protein responses and glycogen depletion in type I fibers compared with type II muscle fibers suggest that low-load resistance training with BFR preferentially stimulates type I fibers (26). For example, Bjørnsen et al. (21) reported a 12% increase in cross-sectional area (fCSA) of type I myofibers with no changes in type II myofibers from the vastus lateralis (VL) after 6.5 wk of lower body resistance training with ∼120 mmHg of BFR in elite powerlifters. The authors implemented four sets of front squats using 30% 1-repetition maximum (1RM) during weeks 1 and 3 (5 times/week, 10 BFR sessions total) of a 7-wk periodized powerlifting protocol. The authors do acknowledge, however, that the differences in fiber growth might be partially explained by the fact that elite powerlifters likely possess preferentially hypertrophied type II fibers from years of high-load training, limiting the growth of these fibers from a short-term intervention (27). Despite this limitation, whether or not BFR training elicits preferential hypertrophy of type I fibers remains equivocal, as some studies also report comparable hypertrophy of type I and II fibers (17) or no growth of either fiber type (20).

Previous results are conflicting regarding sex-related differences in fiber hypertrophy from traditional resistance training (28–33). Abou Sawan et al. (32) and Moesgaard et al. (33) both observed greater hypertrophy of type I fibers in males, with type II fibers exhibiting similar hypertrophy in males and females, after 8 wk of whole body resistance training (3 or 4 sets per exercise, 8–12 repetitions to failure). Other findings suggest that males demonstrate more growth of both fiber types (30), females display more growth of both fiber types (29), or training-induced muscle hypertrophy was similar between sexes (31). Surprisingly, there is a gap in our understanding regarding fiber hypertrophy in females in response to low-load resistance training with BFR (34). This is due to previous studies either examining males exclusively (17, 20) or not reporting sex-related differences because of a low number of female participants (21, 22).

Several studies have measured changes in fCSA following traditional resistance training using repeated-measures ANOVAs, which reduces the number of observations within an individual to a mean value per time point (29, 35–41). This approach can diminish sample size and also fails to incorporate intraindividual correlations in the fCSA measurements within and across time points. Furthermore, this approach does not allow for an unbalanced design, nor does it allow for different training responses for each individual. Therefore, the purpose of the present study is to examine changes in type I/II fCSA of the VL before and after 6 wk of high-load resistance training and low-load resistance training with BFR in untrained males and females. A mixed-effects model was utilized wherein myofiber data were nested within participants and a random slope was estimated for each participant’s hypertrophic response from training. Fixed effects of group, sex, fiber type, and time and interactions were included in the model to understand the interplay of these variables in the hypertrophic response of muscle fibers to resistance exercise training. Muscle hypertrophy is often times poorly defined and highly dependent on the measurement used to assess it. Therefore, single-fiber hypertrophy or myofiber hypertrophy is defined as an increase in fCSA as assessed by the muscle biopsy/histochemistry approach, whereas whole muscle hypertrophy is defined as an increase in mCSA as measured by ultrasound imagining and scan geometry. It is hypothesized that type I fibers will hypertrophy to a greater degree from low-load training with BFR but type II fiber hypertrophy will be comparable from high-load training and low-load training with BFR. Additionally, it is hypothesized that males will possess larger type II fibers than females but that type I fibers will be similar in size between sexes. Finally, we hypothesize that the level of myofiber hypertrophy will be similar between sexes for each fiber type and training method.

METHODS

Participants

Thirty recreationally active males and females were recruited for this study. Fifteen participants were randomly placed in either a high-load resistance training group (HL; n = 15, 8 female) or a low-load resistance training with BFR group (BFR; n = 15, 8 female) (Table 1). Exclusion criteria included any cardiovascular, metabolic, or musculoskeletal conditions. Individuals with a history of severe hip, knee, ankle, and/or other pathological conditions that could impair their motor control were not eligible. Participants included in the study were males and females between 18 and 30 yr old. Participants ranged from untrained to recreationally active as defined by having engaged in zero to six total structured resistance training sessions in the previous 6 mo. Each participant’s activity level was assessed during visit 1 with a health history questionnaire. Individuals who participated in recreational activities (e.g., jogging, intramural sports, etc.) were eligible to participate so long as they met the resistance training requirements. Throughout the study duration, participants were instructed to refrain from resistance training outside of the supervised sessions for this study. As indicated by the responses to the health history questionnaire, we are confident that participants of both sexes had minimal resistance training experience before the study. Furthermore, participants were instructed to refrain from resistance training outside of the supervised sessions for this study, and thus we are also confident that participants did not engage in any resistance training activities outside of the observed sessions that could impact the results of the study. The institutional review board for human subjects research at the participating institution approved this study, and it was conducted in accordance with the Declaration of Helsinki (IRB Study no. 00147374). This study was registered as a clinical trial before the recruitment of the first participant (June 25, 2021, NCT04938947). All participants read and signed an informed consent form and completed a preexercise health questionnaire.

Table 1.

Participant demographic information

	Count	Age, yr	Height, cm	Mass, kg
Males
HL	7	21.7 ± 1.9	181.5 ± 5.6	82.8 ± 18.5
BFR	7	21.4 ± 2.7	174.7 ± 9.2	73.5 ± 16.5
Females
HL	8	22.9 ± 4.3	164.9 ± 7.1	76.6 ± 14.5
BFR	8	21.3 ± 2.9	170.1 ± 7.9	67.7 ± 7.9

Data presented are means ± SD for males and females in either the high load (HL) or low load with blood flow restriction (BFR) resistance training group.

Research Design

Participants visited the laboratory twice before (Pre, visits 1 and 2) and after (Post, visits 3 and 4) a 6-wk (3 days/week) intervention that included leg extensions performed at either a high load or a low load with BFR (Fig. 1). Ultrasound imagining and muscle biopsy procedures were performed on the VL of the right leg, regardless of leg dominance. During visit 1, participants had ultrasound images taken of the VL and then performed a bilateral leg extension 1-repetition maximum (1RM) strength test. Within 2–7 days, participants returned to the laboratory for visit 2, when participants had a pretraining muscle biopsy performed. There is consistent evidence that suggests maximal lower body strength is not impaired by one or more phases of a female’s menstrual cycle (42–47). Nevertheless, to better account for possible strength fluctuations caused by the menstrual cycle, the start of the resistance training intervention for female participants was delayed 2–3 wk after visit 2 (∼8 wk, ∼2 menstrual cycle periods) to replicate hormonal conditions during pre- and posttraining 1RM tests. Menstrual cycles for female participants were assessed via a questionnaire in the prestudy screening process. Because of scheduling constraints, five females were unable to delay the start of their resistance training protocol and began within a week after visit 2. Within 3–7 days after visit 2, male participants in both groups began the 6-wk resistance training protocol. Within 2–7 days after concluding the final resistance training session, participants returned to the laboratory for visit 3, during which time participants had a posttraining muscle biopsy performed. Finally, within 3–7 days, participants returned to the laboratory for visit 4, when ultrasound images were taken followed by a 1RM strength test. The 1RM tests were performed before and after the pre- and posttraining biopsies, respectively. Thus, the 1RM tests did not provide a de facto training stimulus that could impact fCSA. The time periods between biopsy and 1RM testing visits were chosen to ensure adequate incision healing before the 1RM strength test.

Figure 1.

An overview of the study timeline. Participants in the high-load (HL) and blood flow restriction (BFR) groups completed experimental testing before (Pre) and after (Post) a 6-wk resistance training (RT) protocol. Dark arrows indicate what activities were performed during each visit. 1RM, 1-repetition maximum; AMRAP, as many repetitions as possible; F, females; M, males.

Resistance Training and Blood Flow Restriction

Participants in the high load (HL) and low load with blood flow restriction (BFR) groups completed three resistance training sessions per week for 6 wk (18 training sessions total), with each session consisting of three sets of bilateral leg extensions to failure. The training sessions took place throughout each week with the constraint that a maximum of two sessions could take place in consecutive days. Participants were instructed to refrain from performing any additional resistance training outside of these sessions so as not to confound the results of the study. During the first set of the first resistance training session, participants in the HL group performed repetitions with 80% of their 1RM. If participants achieved >12 repetitions or <8 on a given set, the load was increased or decreased ∼2.2 kg to ensure that failure was reached in the 8–12 repetition range. The third set of each resistance training session was used to determine the load for the first set of the next session.

Participants of the BFR group performed three sets to failure using 30% of 1RM (48). Because of the highly variable interindividual response to training with BFR, the load was held constant for all sets and training sessions for all members in the BFR group (49, 50). Participants in the BFR group performed leg extensions with 10-cm-wide BFR cuffs (SmartTools, Strongsville, OH) applied to the most proximal portion of both legs. The cuff pressure was set to 50% of each participant’s arterial occlusion pressure (AOP) as estimated from their thigh circumference, as measured at the most proximal region of the thigh (100–180 mmHg) (51, 52). The 50% AOP was used to minimize participant discomfort, as higher pressures have not been shown to increase muscle activity during leg extension training (52). The pressure was applied according to the instructions in the SmartTools User’s Manual (SmartTools). Briefly, with the participant in a standing position and both deflated cuffs positioned securely around the proximal thighs, one cuff was inflated gradually to the desired pressure. Once this pressure was achieved, an identical procedure was performed on the remaining cuff. The cuffs remained inflated until the conclusion of the final set of the training session.

For all resistance training sessions, failure was defined as the inability to complete another concentric muscle action through a complete range of knee extension as judged by an experienced member of the research team. Participants in both groups were instructed to perform the concentric phase of the repetition as fast as possible and to lower the weight under control during the eccentric portion. Two minutes of rest was utilized between sets for the HL group, and 1 min was utilized for the BFR group. Strong instruction and encouragement were provided throughout each resistance training session. Participants were encouraged to inform members of the research team of variables that could negatively impact training performance (e.g., poor sleep, stress, cold symptoms, premenstrual syndrome, etc.). Thus, training sessions were scheduled to minimize factors that would negatively impact performance. All training sessions were conducted under the direct supervision of the research staff. The training protocols for the HL (4, 39, 53) and BFR (53–55) groups were chosen because similar programs have demonstrated significant gains in muscular size in untrained participants.

1RM Testing

In visits 1 and 4 participants performed bilateral leg extension 1RM testing according to the guidelines established by the National Strength and Conditioning Association (2008). The leg extension 1RM test has demonstrated high test-retest reliability, even in untrained individuals (56). Specifically, participants performed a light warm-up set with 5–10 repetitions of estimated 1RM, followed by two or three heavier warm-up sets of 2–5 repetitions with increasing loads of 10–20% at each set. Participants then completed trials of 1 repetition with increasing loads (10–20%) until they were no longer able to complete a single repetition through a full range of motion. The highest load (kg) achieved during this sequence was deemed as the 1RM. Two to four minutes of rest was allowed between successive sets during 1RM testing.

Ultrasound Imaging

Ultrasound images of the right VL were collected during visits 1 and 4. Images were taken with a NextGen LOGIQ ultrasound console (GE Healthcare UK, Ltd., Chalfont, UK) with a multifrequency linear array transducer (model 12 L-RS: 5–13 MHz; 38.4-mm field of view). Images were collected in panoramic mode with a frequency of 10 Hz and depth held constant between time points within each participant. Before ultrasound imaging, participants rested ∼2–3 min supine on a cushioned examination table. Transverse panoramic images of the VL cross section were taken at 50% of the distance from the anterior superior iliac spine superior to the patella (57). A custom-made probe support composed of high-density foam padding was positioned perpendicular to the longitudinal axis of the thigh to ensure that the images captured muscle within the transverse plane. Ultrasound scans were performed with a sufficient amount of gel so as to ensure high image quality and low pressure on the skin. To calculate mCSA, the periphery of the VL was outlined with the polygon function in ImageJ (National Institutes of Health, Bethesda, MD, 1997–2014), with care taken to exclude the surrounding fascia. All ultrasound images were taken once, and all analyses were performed by a single experienced technician. Previous work, including from our laboratory, has determined that this method displays both high intra (58–62)- and inter (58, 63)-rater reliability. Additional evidence from cadaver (64) and magnetic resonance imaging (63) studies have supported ultrasound imaging as a valid and reliable tool for assessing mCSA.

Muscle Biopsies

During visits 2 and 3, a single muscle biopsy was taken from the VL at the midpoint of the thigh midway between the inguinal ligament and the patella on the right leg, with the percutaneous needle biopsy methods of Bergström (65). After careful cleaning of the sample site, a local anesthetic (2% lidocaine) was injected percutaneously and a small incision was made through the skin and deep fascia with a no. 11 scalpel. The sample was taken with a traditional Bergström needle (Pelomi Medicals, Abertslund, Denmark), utilizing the double chop and suction method (66). Samples were teased free of blood and surrounding connective tissue before being mounted to a cork block with tragacanth gum and insulated with a layer of optimum cutting temperature (OCT) medium. Samples were then frozen in liquid nitrogen-cooled 2-methylbutane and stored at −80°C until further analysis.

Muscle Fiber Analyses

The muscle biopsy samples were prepared for fiber type and fCSA analyses following previous methods (67). Samples were sectioned at 7 µm, mounted on positively charged histology slides, and stored at −80°C until staining. Sections were outlined by using a hydrophobic pen to retain solutions for incubation. Initially, phosphate-buffered saline (PBS) was applied for 10 min to rehydrate muscle sections. PBS was then removed, and a 3% peroxide solution was added to sections for 15 min. This solution was removed, and then slides were rinsed with PBS for three 5-min washes on a rocker. Sections were then incubated with TrueBlack Lipofuscin Autofluorescence Quencher solution (biotium, Fremont, CA) for 1 min, and slides were rinsed in PBS for three 5-min washes. Subsequently, the sections were blocked with 5% normal goat serum and 2.5% normal horse serum for 1 h and washed in PBS for 5 min. Next, the primary antibody solution was applied (1:20 Mandra, 1:20 of BA_D5, 9:20 and 9:20 5% normal horse serum, all diluted in PBS; all antibodies from Developmental Studies Hybridoma Bank, Iowa City, IA), and slides were incubated overnight at 4°C. The next day, slides were washed four times in PBS. A secondary antibody solution was applied for 1 h (1:100 goat anti-mouse IgG1 594 and 1:100 goat anti-mouse IgG2B 488 diluted in PBS). Slides were then washed four times in PBS and a DAPI fluorescent dye (1:10,000 DAPI, diluted in deionized water) was then applied for 15 min. Then, two 5-min PBS washes were applied to slides, a 1:1 PBS-glycerol solution was applied around the sections, and glass coverslips were applied thereafter. During this process, no distinction was made between type IIa and IIx muscle fibers. This choice was made because our imaging capabilities were limited to three detection filters (DAPI, FITC, TRITC) and previous work has demonstrated that the proportion of type IIx fibers in untrained individuals is scarce (68–70) in comparison to that found in elite-level athletes (71, 72). Moreover, work from our laboratory has found that distinguishing between type IIx and IIa myosin heavy chain isoforms does not further explain motor unit firing rates or action potential amplitudes in the VL (73–75). Overall, the physiological differences between type IIa and IIx fibers are minimal in comparison to the vast differences between type I and II fibers (76, 127). This staining method allows for the detection of type I fiber blue cell bodies and type II fiber black cell bodies (unlabeled). Standardized measurements of type I, type II, and mean fCSA were performed with an open-source software (MyoVision) (77) (Fig. 2). A pixel conversion ratio value of 0.493 pixels/µm was used to account for the size and bit depth of images, and a detection range of 200–12,000 µm² was used to ensure that artifacts and incorrectly orientated (oblong) fibers were removed.

Figure 2.

Images from MyoVision software used to calculate fiber cross-sectional area (fCSA) for type I and II muscle fibers of the vastus lateralis before (Pre) and after (Post) 6 wk of resistance training. Fibers shown are from representative male and female participants that either participated in low-load resistance training with blood flow restriction (BFR; A–D) or high-load resistance training (HL; E–H). Digital images were captured with a fluorescent microscope (Nikon Instruments, Melville, NY States) using a ×10 objective lens. This staining method allows for the detection of type I fiber blue cell bodies (detected by the FITC filter) and type II fiber black cell bodies (unlabeled). Dystrophin cytoplasmic protein is stained in bright green. The length of the white scale bar in A is 100 pixels at a pixel conversion ratio of 0.493 pixels/µm.

Statistical Analyses

Separate three-way mixed factorial ANOVA models [group (HL vs. BFR) × sex (male vs. female) × time (Pre vs. Post)] were applied to 1RM and mCSA. Mean type I and II fCSA for each participant were calculated for each pre- and posttraining biopsy. To better understand how changes in mCSA and strength correlate with changes in fCSA, linear regression analyses were performed on the percent changes of mCSA and 1RM with mean type I and II fCSA percent change scores computed for each participant. Estimated marginal mean comparisons were performed to follow up significant interaction terms in ANOVA models, with Bonferroni corrections applied for all possible pairwise comparisons. Cohen’s d effect sizes were calculated where appropriate. Statistical significance was set at 0.05.

To determine how differences in sex and resistance training method affected hypertrophy of type I and II muscle fibers, linear mixed-effects models were implemented (Fig. 3). After visual inspection of q-q plots and predicted scores versus residual plots, it was determined that the assumptions of residual normality and homoscedasticity were not met. To provide more defensible inferences, we employed nonparametric bootstrapping procedures using level 2 (participant) and level 1 (muscle fiber) resampling and 10,000 samples. Each bootstrapped sample was constrained such that the number of participants in each group reflected that of the original sample. To establish the most appropriate combination of fixed effects, four candidate models were fit to each bootstrapped sample and the Bayesian information criterion (BIC) values were compared (78). Each candidate model had a random intercept for each participant as well as a random slope for time. The candidate models were nested in their fixed effects in that the full model (M1) included a four-way group (level 2) × sex (level 2) × time (level 1) × type (level 1) interaction and all the possible combinations of lower-order three-way and two-way interactions as well as main effects; M2 had all possible three-way and two-way interactions and main effects; M3 had all possible two-way interactions and main effects; and M4 had only main effects. In 9,999 of 10,000 samples, M1 presented with the lowest BIC, and thus the coefficients for this model were analyzed with further bootstrapping procedures. For M1, pseudomarginal ($R_{\mathrm{m}}^2$) and conditional ($R_{\mathrm{c}}^2$) values were calculated, where $R_{\mathrm{m}}^2$ provides variance accounted for by fixed effects alone and $R_{\mathrm{c}}^2$ provides variance accounted for by both fixed and random effects (79). Bias-corrected and accelerated bootstrapped confidence intervals were calculated using the coxed package (80) to determine statistical significance for model coefficients (81). To follow up on significant interaction terms, estimated marginal means were bootstrapped to calculate confidence intervals for pairwise comparisons. Bootstrapped confidence intervals were set at 95% unless widened for multiple comparisons using the Bonferroni method. Results for mCSA and 1RM comparisons are reported as mean (SD). All statistical analyses and data visualizations were performed in R v.4.1.2 (R Core Team, 2021) using the RStudio environment (v.2022.02.0). All figures and plots were constructed with the ggplot2 package (82). Linear mixed-effects models were fitted with the lme4 package (83). ANOVA models were fitted with the afex package (84), and estimated marginal means were calculated with the emmeans package (85).

Figure 3.

Overview of the statistical analyses performed to understand how differences in sex and resistance training method affected hypertrophy of type I and II muscle fibers. In M1–M4, the asterisks in the interaction terms indicate that all lower-order interactions and main effects are also included in the model. BCA, bias corrected and accelerated; BFR, blood flow restriction group; BIC, Bayesian information criterion; CI, confidence interval; fCSA, fiber cross-sectional area; HL, high-load resistance training group; $R_{\mathrm{m}}^2$, marginal R²; $R_{\mathrm{c}}^2$, conditional R².

RESULTS

Three of the 30 participants’ VL ultrasound image files were corrupt, and one participant’s muscle samples exhibited freeze damage from storage. Thus, a total of 27 participants were included in the statistical analyses for mCSA (male = 12, female = 15; BFR = 13, HL = 14) and a total of 29 participants were included in the analyses for fCSA and muscle fiber type (male = 13, female = 16; BFR = 14, HL = 15).

Resistance Training Load, Repetitions, and Volume Load

Two participants in the BFR group chose to terminate their first BFR session after the first set because of discomfort and thus performed 52 sets as opposed to 54. The average weekly volume load (repetitions × load) was 757.4 (87.8) kg, whereas that for the HL group was 513.07 (75.5) kg. The load, repetitions, and volume load for each group, set, and week are presented in Fig. 4.

Figure 4.

Boxplots for load (A), repetitions (B), and volume load (C) presented for the first (red), second (green), and third (blue) sets for each week of the resistance training protocol. Data within each week represent all 3 resistance training sessions performed. Data are displayed for members of the high load group (HL; n = 15, 8 female) or the low load with blood flow restriction group (BFR; n = 15, 8 female); n represents the number of individuals/participants.

VL mCSA and Leg Extension 1RM

For VL mCSA, there were main effects for sex and time (P < 0.001) but not for group (P = 0.749). When collapsed across groups and times VL mCSA was greater in males compared with females, and when collapsed across sexes and groups VL mCSA was larger at posttraining compared with pretraining.

For 1RM, there was a significant three-way sex × group × time interaction (P = 0.027). For the HL group, post hoc comparisons revealed significant increases in 1RM from pre- to posttraining in males (P < 0.001) and females (P = 0.001). 1RM increases were also observed for males of the BFR group (P = 0.001). More detailed results regarding the effects of sex and resistance training on VL mCSA and leg extension 1RM are presented in Table 2.

Table 2.

Effects of sex and resistance training on VL mCSA and leg extension 1RM

	mCSA, cm2			1RM, kg
	Pre	Post*	Cohen’s d	Pre	Post	Cohen’s d
Males#
HL	28.9 ± 5.5	30.7 ± 7.6	0.78	86.5 ± 26.3	128.9 ± 21.1^$	2.42
BFR	29.8 ± 3.1	31.1 ± 4.4	0.61	92.0 ± 33.1	107.2 ± 32.9^	1.43
Females
HL	20.3 ± 3.6	21.9 ± 4.3	1.23	64.4 ± 23.6	82.8 ± 22.4^	2.39
BFR	18.8 ± 2.4	20.0 ± 2.3	1.71	60.1 ± 10.2	70.6 ± 6.9	1.49

Values are means ± SD and Cohen’s d effect sizes for vastus lateralis (VL) muscle cross-sectional area (mCSA) and leg extension one-repetition maximum (1RM). Data are from males and females in the high load (HL) and blood flow restriction (BFR) groups before (Pre) and after (Post) resistance training intervention. *Significant main effect for time (mCSA: Post > Pre); #significant main effect for sex (mCSA: male > female). ^Significant post hoc pairwise comparison between time points (1RM: males HL Post > males HL Pre, males BFR Post > males BFR Pre, females HL Post > females HL Pre); $significant post hoc pairwise comparison between sexes (1RM: HL Post males > HL Post females).

fCSA Models

A breakdown of fiber counts for each sex and group is presented in Table 3.

Table 3.

Breakdown of fiber counts before and after training

	Average Fiber Count				Total Fiber Count
	Type I		Type II		Type I		Type II
	Pre	Post	Pre	Post	Pre	Post	Pre	Post
Sex
Males	36.0 ± 18.6	24.2 ± 14.9	54.5 ± 30.3	47.5 ± 17.2	468	315	709	618
Females	49.8 ± 27.5	40.7 ± 18.3	62.7 ± 37.9	57.0 ± 37.0	796	651	1003	912
Group
HL	36.8 ± 20.9	24.5 ± 12.1	53.6 ± 27.4	44.0 ± 29.7	552	367	804	660
BFR	50.9 ± 26.8	42.8 ± 19.9	64.9 ± 40.7	62.1 ± 27.7	712	599	908	870

Values are means ± SD and total fiber counts for type I and II fibers obtained from vastus lateralis muscle samples before (Pre) and after (Post) resistance training. Fiber counts are grouped on the basis of sex or training group. BFR, blood flow restriction; HL, high load.

Of the candidate models chosen, the full model including a four-way group × sex × time × type interaction as well as all the lower-order interaction terms and main effects had the lowest BIC using 9,999 of 10,000 bootstrapped samples, and thus this model was chosen for further analyses. The means of the bootstrapped $R_{\mathrm{m}}^2$ and $R_{\mathrm{c}}^2$ were 0.17 and 0.59, respectively, suggesting that this model accounted for nearly 60% of the variability in fCSA. The bootstrapped coefficients revealed a significant three-way group × time × type interaction and a significant two-way sex × type interaction (Table 4). Estimated marginal means revealed that only type II fibers for the HL group exhibited hypertrophy from training when collapsed across sexes (Fig. 5). Additionally, males possessed larger type II fibers compared with females when collapsed across groups and time points (Fig. 6). Raincloud plots displaying raw fCSA pre- and posttraining data for each group and fiber type can be found in Fig. 5, whereas similar plots displaying the fCSA data for males and females for each fiber type are shown in Fig. 6. Cohen’s d values were calculated to indicate the hypertrophy effect size of each fiber type for each sex and group combination (Tables 5 and 6). Importantly, moderate effect sizes were present for type I (0.59) and II (0.67) fiber hypertrophy of males in the BFR group and type II fibers for females in the HL group (0.59).

Table 4.

Effects included in our bootstrapped mixed-effects model for fiber cross-sectional area

Coefficient	Mean	95% BCA CI
TimePost	93.3	[−740.9, 768.4]
GroupHL	507.9	[−648.3, 1,773.8]
SexM	199.9	[−903.9, 1,452.0]
TypeII	−20.1	[−232.3, 236.5]
TimePost.GroupHL	−26.5	[−846.4, 903.9]
TimePost.SexM	562.7	[−236.1, 1,450.3]
GroupHL.SexM	129.5	[−1,461.4, 1,608.4]
TimePost.TypeII	30.8	[−415.6, 281.9]
GroupHL.TypeII	−202.0	[−927.2, 305.7]
SexM.TypeII	607.2	[100.0, 1,755.0]*
TimePost.GroupHL.SexM	−117.4	[−1,549.4, 1,423.1]
TimePost.GroupHL.TypeII	1,047.0	[472.9, 1,943.4]*
TimePost.SexM.TypeII	−212.9	[−2,073.5, 718.0]
GroupHL.SexM.TypeII	41.0	[−1,110.5, 1,169.7]
TimePost.GroupHL.SexM.TypeII	−330.4	[−1,933.8, 1,545.6]
$R_{\mathrm{m}}^2$	0.17	[0.08, 0.34]
$R_{\mathrm{c}}^2$	0.59	[0.46, 0.68]

Values are means and bias-corrected and accelerated (BCA) confidence intervals (CIs) for each effect included in our bootstrapped mixed-effects model for fiber cross-sectional area. The level next to each factor included in a coefficient specifies the nonreference level. Periods indicate interactions between 2 or more factors (e.g., TimePost.GroupHL indicates the time × group interaction); bold type and * on CIs denote statistical significance at the 0.05 level. $R_{\mathrm{m}}^2$ and $R_{\mathrm{c}}^2$ are the marginal and conditional pseudo-R² values, respectively. The names of each factor include the name of each factor and the nonreference level (e.g., TimePost = post-resistance training vs. post-resistance training for the time factor; TypeII = type II fibers vs. type I fibers for the fiber type factor). The reference levels for the time, group, sex, and type factors were Pre, BFR, F, and type I, respectively. BFR, blood flow restriction; F, female; HL, high load; M, male.

Figure 5.

Boxplots and raincloud plots for fiber cross-sectional area (fCSA) data of male (M) and female (F) participants of the high load (HL, n = 15, 8 female; A and C) and low load with blood flow restriction (BFR, n = 14, 8 female; B and D) groups before (orange) and after (yellow) resistance training; n represents the number of individuals/participants. *Statistically significant pairwise comparison at the 0.05 confidence level as revealed by bias-corrected and accelerated bootstrapped confidence intervals of pairwise comparisons using 10,000 cases. fCSA of type II fibers significantly increased from pretraining to posttraining for members of the HL group when collapsed across sex.

Figure 6.

Boxplots and raincloud plots for type I (A) and type II (B) fiber cross-sectional area (fCSA) data of male (blue) and female (green) participants of the high load (HL; n = 15, 8 female) and low load with blood flow restriction (BFR, n = 14, 8 female) groups; n represents the number of individuals/participants. *Statistically significant pairwise comparison at the 0.05 confidence level as revealed by bias-corrected and accelerated bootstrapped confidence intervals of pairwise comparisons using 10,000 cases. fCSA of type II fibers was higher in males compared with females when collapsed across times and groups.

Table 5.

Vastus lateralis type I and II fiber cross-sectional area in HL and BFR groups before and after resistance training

	Type I fCSA, μm2			Type II fCSA, μm2
	Pre	Post	Cohen’s d	Pre	Post	Cohen’s d
Males
HL	3,918.3 ± 1,546.1	4,510.6 ± 2,405.5	0.29	4,489.1 ± 1,735.4	5,007.6 ± 1,653.4	0.31
BFR	3,403.5 ± 1,119.3	4,178.6 ± 1,497.5	0.59	3,765.9 ± 1,470.9	4,635.2 ± 1,120.3	0.67
Females
HL	3,631.6 ± 1,591.9	3,935.8 ± 1,720.2	0.18	3,356.0 ± 1,133.1	4,171.9 ± 1,597.7	0.59
BFR	2,968.3 ± 1,419.5	3,391.4 ± 1,486.2	0.29	2,601.0 ± 1,452.4	3,076.3 ± 1,310.4	0.34

Data are means ± SDs and Cohen's d effect sizes of vastus lateralis type I and II fiber cross-sectional area (fCSA). Data are from males and females in the high load (HL) and blood flow restriction (BFR) groups before (Pre) and after (Post) resistance training intervention.

Table 6.

Estimated marginal means, bias-corrected and accelerated confidence intervals for pairwise comparisons, and Cohen's d effect sizes

Contrast	Group	Type	Estimated Marginal Means	BCA CI	Cohen’s d
Post-Pre	BFR	I	377.5	(−221.0, 884.2)	0.36
		II	292.0	(−955.8, 997.9)	0.37
	HL	I	286.1	(−415.3, 1,176.9)	0.22
		II	1,093.2	(495.4, 1,801.6)*	0.46
Male-Female		I	512.5	(−380.0, 1471.5)	0.36
		II	953.2	(49.1, 1,990.3)*	0.78

Contrasts for fiber cross-sectional area were estimated using the estimated marginal means from the significant 3-way group × time × type interaction (Post-Pre) and the significant 2-way sex × type interaction (male-female). Bold type and * indicate significant contrasts. Confidence intervals (CIs) are adjusted for multiple pairwise comparisons using the Bonferroni method. For Post-Pre contrasts confidence widths of 98.75% are used, and for male-female contrasts 97.5% confidence widths are used. Cross-sectional areas are presented in μm². BCA, bias-corrected and accelerated; BFR, blood flow restriction; HL, high load.

Correlation between Changes in mCSA and 1RM and Changes in fCSA

There were no significant correlations between percent changes in mCSA and type I [r = 0.23, 95% confidence interval (CI) = [−0.17, 0.56], P = 0.26, df = 25] or II (r = −0.06, 95% CI = [−0.43, 0.33], P = 0.77, df = 25) fCSA. Furthermore, there were no significant correlations between percent changes in 1RM and type I (r = −0.11, 95% CI = [−0.46, 0.27], P = 0.58, df = 27) or II (r = 0.15, 95% CI = [−0.23, 0.49], P = 0.43, df = 27) fCSA (Fig. 7).

Figure 7.

Percent change values plotted for the vastus lateralis muscle cross-sectional area (mCSA) and one-repetition maximum strength (1RM) against percent changes in both type I and II fiber cross-sectional area (fCSA) following 6 wk of either high-load resistance training (HL, triangles) or low-load resistance training with blood flow restriction (BFR, circles). Data are shown for males (blue) and females (green). Regression lines are presented with 95% confidence intervals. Two additional data points are shown in C and D (HL, n = 15, 8 female; BFR, n = 14, 8 female) compared to A and B (HL, n = 14, 7 female; BFR, n = 13, 8 female); n represents the number of individuals/participants.

DISCUSSION

There is very limited information regarding potential sex-related differences in type I and II myofiber hypertrophy from resistance training with BFR. Our primary aim was to leverage two different training methods (traditional high load and low load with BFR) to determine whether differences in fiber type-specific hypertrophy occurred and to understand whether these differences were influenced by sex. The principal finding of the present study was that the type II fibers of the HL group exhibited significant myofiber hypertrophy, with no differences between males and females. Additionally, males possessed larger type II fibers than females, irrespective of training group or time point. Although not statistically significant, when separated by sex both type I and II fibers for males of the BFR group exhibited larger effect sizes for hypertrophy than those for females. Conversely, the effect size for type II fiber hypertrophy in the HL group was larger for females than for males. Finally, poor correlations between changes in mCSA and type I/II fCSA possibly illustrate the limitations of using the standard muscle biopsy/histochemistry technique to assess myofiber hypertrophy.

Linear mixed-effects models are becoming increasingly popular in the analysis of experimental data (86). This statistical approach is viable for human muscle physiology studies in which outcome variable(s) are represented from one or more samples of muscle fibers nested within biopsy samples from each participant (17, 20, 87, 88). There is a lack of agreement on the reporting of effect sizes from mixed-effects models for pairwise comparisons despite the utility and growing prevalence (89–91). Although Cohen’s d is relatively simple to understand and compute, its reliance on pooled standard deviations makes it a difficult metric to use alongside a mixed-effects model where standard deviations can be calculated at any level in the data hierarchy. Cohen’s d values here were calculated using standard deviations at the level of the individual muscle fibers (without reference to participants); however, this could also be done using standard deviations at the participant level (91).

The literature is mixed regarding changes in type I and II fCSA following high-load resistance training in males (32, 33, 37, 39, 92) and females (32, 33, 35). The results of the present study indicate that type II fibers of males and females displayed significant amounts of hypertrophy from high-load training, with no significant increases for type I fibers. Campos et al. (37) and Mitchell et al. (39) both reported significant increases in type I and II fCSA following a high-load training protocol consisting of multijoint and single-joint exercises in untrained males. The greater myofiber hypertrophy observed in these studies compared with those of the present study (type I: 13% and 17% vs. 15.1%; type II: 22% and 16% vs. 11.6%) likely relates to the higher volumes used in the protocols of Campos et al. (37) and Mitchell et al. (39) (240 and 90 sets, respectively, vs. 54 sets). Additionally, Abou Sawan et al. (32) observed an ∼20% increase in type I fCSA in males with no change in females and a larger relative increase in type II fCSA for females (27% vs. 17%) following high-load training. Our results mirror these findings, as type I fibers of males hypertrophied to a greater extent than those of females after high-load training (males: 15.1%, d = 0.29; females: 8.4%, d = 0.18); however, relative changes in type II fCSA in females (24.3%, d = 0.59) were larger compared with males (11.6%, d = 0.31). These findings indicate that type II muscle fibers of females have growth potential similar to or greater than males when adequate training volume and load are in place.

There is limited information regarding the response of type I and II fCSA in females following low-load resistance training with BFR because the majority of studies did not include or only enrolled a small number (<5) of females (17, 20–22). Low-load training with BFR resulted in no significant myofiber hypertrophy of either fiber type for males or females via the mixed-effects models. The effect sizes for both fiber types were lower in females (type I: d = 0.29, type II: d = 0.34) compared with males (type I: d = 0.59, type II: d = 0.67) (Table 5). The literature regarding BFR-related fiber-specific hypertrophy remains unclear, with studies showing no significant hypertrophy for either type II fibers (21, 22) or either fiber type (20). A previous study from Nielsen et al. (17) reported larger increases in type I (35%) and type II (37%) fCSA after 3 wk of unilateral leg extensions with 100 mmHg of BFR in untrained males in comparison to the present study’s findings (type I: 22.8%, type II: 23.1%). The greater number of sets to failure implemented by Nielsen et al. (17) compared with that of the present study (92 sets vs. 54 sets) is a possible contributing factor to differences in fCSA growth. In addition, Bjørnsen et al. (21) reported significant increases in type I fCSA (12%) with no changes in type II fCSA following 7 wk of a combination of high-load training and low-load training with BFR in national-level powerlifters (n = 17, 2 females). The differing results could be potentially explained by the highly trained powerlifters possessing already hypertrophied type II fibers that are less responsive to training from short-duration interventions (27). Although changes in fCSA from BFR training were not significant in the mixed-effects model, there were apparent increases when they were examined via effect sizes that tended to be higher in males. The relationship between training load and the hypertrophy of specific myofiber types is equivocal (93), with some studies suggesting that type II fibers preferentially hypertrophy from high-load training conditions (37, 94, 95) and others reporting similar hypertrophy of type II fibers regardless of training load (39, 41, 96). It cannot rule out the possibility that the lack of type II myofiber hypertrophy in the BFR group was caused by the low relative load of the training protocol (30% 1RM). The load utilized in this and similar resistance training studies (21, 22) could be insufficient to recruit the higher-threshold motor units comprised of type II fibers. Future studies are needed to further isolate potential sex-related responses in myofiber hypertrophy from low-load resistance training with BFR.

There is an abundance of studies that have examined sex-related differences in muscle fiber size (87, 97–102). Follow-up pairwise comparisons from the mixed-effects models revealed significantly larger fCSA for type II fibers in males compared with females (37% difference, d = 0.78). There was a small effect size indicating that males possess larger type I fibers than females (16% difference, d = 0.36), although the sex differences of type I fCSA did not reach statistical significance. There is agreement in the literature suggesting that males typically have larger type II fibers compared to females, with percent differences ranging from 18% to 85%. Differences in fiber typing methods probably contribute to this wide range of values, as some studies classified type IIa (97, 99–102), IIB (97, 99, 101), and/or IIx fibers (100, 102) as opposed to designating fibers as type II (87, 98) as was done in the present study. Although sex-related differences in type I fCSA were nonsignificant, the 16% difference is similar to those observed by Claflin et al. (87), Simoneau and Bouchard (97), and Staron et al. (99). It has been posited that single-fiber force and power production are similar between sexes when adjusted for fiber size (87, 103). Even when normalized to fiber size, however, type II fibers generate more power and higher forces than type I fibers (103). Thus, the larger type II fibers of males in the present study could partially account for sex-related differences in muscular strength and performance.

In addition to fCSA, previous studies have reported significant whole muscle hypertrophy from both high-load resistance training and low-load resistance training with BFR via measurements such as mCSA (104–108) or muscle volume (39, 109, 110). In the present study, there were significant main effects for time (d = 0.91) and sex (d = 2.26). The relative whole muscle hypertrophy for males in both the HL (5.5% increase) and BFR (4.2% increase) groups aligns with studies using similar protocols that report increases of ∼4.3% (3 sets to failure at 80% 1RM, 3 times/week, 12 wk) (107) and ∼4.5% (4 sets to failure at 15% MVC, 4 times/week, 4 wk, ∼230 mmHg) (111), respectively. In a similar fashion, the 6.4% increase in mCSA for the females of the BFR group agrees with the ∼7% increase observed by Ellefsen et al. (55), who had participants perform five sets to failure 2 times/wk for 12 wk using 30% 1RM and 100 mmHg of BFR. Alternatively, Hisaeda et al. (104) found a 3.6% increase in quadriceps mCSA in females from high-load leg extension training, which is less than that observed in the females of the present study (7.8% increase). This difference is likely partially due to the fact that participants performed eight or nine sets of four to six repetitions to failure, using only 90-s rest intervals between sets, which some report as insufficient recovery time when using such higher loads (112–114).

The degree to which various assessments of hypertrophy agree in tracking muscle growth longitudinally is of great interest (46, 47, 61, 115, 116). A noteworthy finding of the present study was that percent changes in both mCSA and 1RM associated poorly with percent changes in both type I/II fibers (r < 0.25). Similarly, Esmarck et al. (117) and Aagaard et al. (118) found higher relative increases in mean fCSA (∼22%, ∼16%, respectively) than mCSA (∼7%, ∼10%, respectively) following resistance training of the quadriceps. Additionally, Haun and colleagues (119) reported a 23% increase in fCSA from resistance training, whereas midthigh thickness only increased by 2%. A potential explanation for the disconnect between hypertrophy measured at the whole muscle and myofiber levels stems from the vast variability in intraindividual muscle fiber size. To illustrate, Horwath et al. (88) reported significant differences in type II fCSA between two nearby biopsy sites of the VL within the same participant and session. Moreover, the standard muscle biopsy/histochemistry technique is limited in sampling ∼50–150 fibers per sample, which pales in comparison to the ∼500,000 muscle fibers that are estimated to be contained in the VL. Although the mixed-effects models used in the present study attempt to account for intraparticipant variability, these limitations still must be acknowledged when using this technique to assess muscle hypertrophy. The results of this study and others suggest that this training strategy can increase mCSA and 1RM strength (55, 111, 120–122) despite the lack of significant myofiber hypertrophy from low-load training with BFR.

There are several limitations that could have affected the results of this study. The moderate (d > 0.5; Table 5) but nonsignificant effect sizes for type I/II fiber hypertrophy of males in the BFR group could be due to a relatively lower sample size at the levels of each sex in each group. For the BFR group, a cuff pressure was used that was set at a percentage of each participants’ AOP. This was done in an attempt to provide a relative BFR stimulus that would capture individual vascular physiology as opposed to applying a single pressure across all individuals. The estimates of AOP were calculated from thigh circumference measurements as conducted by Loenneke et al. (51, 52). Although this method is shown to provide a relative training response, more sophisticated models have emerged that suggest that additional variables including, but not limited to, tissue composition (123), arterial stiffness (124), and mean arterial pressure (125) also account for variability in AOP. It is plausible that improving upon the estimation of AOP could lead to a more stimulative applied pressure and, possibly, myofiber hypertrophy. Another limitation of the study is not discriminating type IIa and IIx myofibers. This was chosen a priori because 1) our analytical capabilities were limited to three detection filters (DAPI, FITC, TRITC) and 2) the presence of IIx myofibers is seldom seen from human VL biopsy sections, especially after resistance training interventions (109, 127). Additionally, work from our laboratory has shown that distinguishing between type IIx and IIa myosin heavy chain isoforms does not further explain motor unit firing rates or action potential amplitudes of the VL (73–75).

The results from the mixed-effects models indicate that only type II fibers demonstrated significant hypertrophy from high-load resistance training in males and females whereas there were no significant increases for either fiber type following low-load resistance training with BFR. The standard measures of effect size slightly contrast with the mixed-effects models, such as that there were moderate increases in type I and II fibers in males after low-load resistance training with BFR unlike in females. Furthermore, females may have undergone a greater magnitude of type II fiber hypertrophy after high-load resistance training in comparison to males. Drawbacks in standard effect size estimates, however, make interpretation difficult alongside mixed-effects models.

In conclusion, BFR in conjunction with low-load resistance training may not promote myofiber hypertrophy to the level of high-load resistance training with similar responses for males and females. In contrast, comparable effect sizes for mCSA and 1RM between groups suggest that BFR could play a valuable role in a structured resistance training program. It is important to note that these findings occurred in the early phase of adaptations to resistance training, during which time the capacity for muscular hypertrophy is relatively high (126). Additionally, we acknowledge that although the sample size at the level of the muscle fiber (level 1) is in the thousands, the participant sample sizes (level 2) are somewhat limited within the sex × group combinations (n < 10). Although nonparametric bootstrapping is a well-regarded approach for situations in which statistical assumptions are in question and small sample sizes are used, future research should attempt to replicate these experiments to instill confidence in the findings. Finally, further studies should be conducted to better understand sex- and training method-related differences in fiber type-specific hypertrophy in trained individuals and how BFR could be implemented in a whole resistance exercise program.

DATA AVAILABILITY

The data that support the findings of this study are available upon request.

GRANTS

This work was made possible by a Master’s Level Graduate Research Grant from the National Strength and Conditioning Association (Grant ID: 1002814). Additionally, this work was supported by an NIH-funded predoctoral fellowship to T. M. Reece (NIH T32HD007434).

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

T.M.R., B.G.V., P.M.G., M.D.R., and T.J.H. conceived and designed research; T.M.R., J.S.G., K.W.S., J.R.P., and P.M.G. performed experiments; T.M.R., J.S.G., M.J.S., and A.B.C. analyzed data; T.M.R., M.J.S., A.B.C., and T.J.H. interpreted results of experiments; T.M.R. prepared figures; T.M.R. and T.J.H. drafted manuscript; T.M.R., J.S.G., M.J.S., A.B.C., J.R.P., P.M.G., M.D.R., and T.J.H. edited and revised manuscript; T.M.R., J.S.G., M.J.S., A.B.C., K.W.S., J.R.P., B.G.V., P.M.G., M.D.R., and T.J.H. approved final version of manuscript.