Resistance training with different repetition duration to failure: effect on hypertrophy, strength and muscle activation

Study design

In the present study, a repeated measures design was adopted. Given that the unilateral exercise model reduces inter-subject variability, it can serve to increase statistical power, as well as reduce the time and cost of a study (MacInnis et al., 2017). Volunteers performed two resistance training protocols with two RDs (2-s or 6-s RD protocol) for 14 weeks. The left and the right legs were randomly assigned and balanced for limb dominance to either one of the protocols. Pre and post-test measures included: CSA, MVIC and 1RM tests. To assess the lower limb dominance voluntaries were asked to answer the following question: “If you would shoot a ball on a target, which leg would you use to shoot the ball?”

In session 1, limb dominance was determined, volunteers were familiarized with all the procedures, and training protocols were assigned to each limb. In the next session, ultrasound images were recorded to determine rectus femoris and vastus lateralis CSA. The strength tests (MVIC and 1RM) were conducted in sessions 3 and 4 separated by at least 48 h. Next, subjects trained from sessions 5 to 39 for a total of 14 weeks and five training sessions per week. The training sessions were separated by at least 24 h. For each week, subjects trained their left or right either on days 1, 3, and 5, or on days 2 and 4 in an alternating way. Through this training schedule, a minimum of 48 h inter session rest was provided for each leg. In sessions 6 and 39 (for each protocol), the rectus femoris and vastus lateralis EMG amplitude were assessed through surface EMG while participants performed their respective training protocols. In session 40, separated between 72 and 120 h from the last training session, the post-test ultrasound measurements were conducted similar to session 2. Finally, in session 41, the MVIC and 1RM post-tests were executed for both lower limbs.

Participants and ethics

The sample size calculation was performed by using the software G.Power for Windows version 3.1.9.7 (Düsseldorf, Germany) and by following the guidelines proposed by Beck (2013), with a priori statistical power (1 - ß) = 0.80, effect size (f) = 0.57 and 5% significance level. Number of groups = 2 (protocols); number of measures = 2 (pre-post); correlation between measures = 0.90; sphericity = 1. For the sample size calculation, we used the absolute AST values of the vastus lateralis muscle from a previous study from our own laboratory, which was carried out with an experimental design similar to the present study (Lacerda et al., 2020).Ten males aged between 18 and 30 years (mean ±  SD: age = 23.1 ± 4.63 years; body height = 1.72 ± 0.07 m; body mass = 68.4 ± 9.46 kg; body fat percentage = 14.03 ± 6.56%) participated in this study.

The inclusion criteria for participation were: (1) no resistance training during the last six months; (2) no functional limitations that could influence the 1RM test or the training protocols; and (3) no use of pharmacological substances or ergogenics supplements, and no other modes of resistance exercise during the study period. Subjects were informed about the study aims, procedures, and risks prior to signing an informed consent form. The ethics committee of the Federal University of Minas Gerais approved this study (approval number: 79108117.5.0000.5149), which complied with the Declaration of Helsinki. Additionally, each subject was instructed not to engage in any physical activity immediately before the testing sessions and to maintain the same diet before each session.

Experimental Session 1(anthropometric measurements)

After receiving information about the goals and the purpose of the study and giving written consent, the volunteers answered the Physical Activity Readiness Questionnaire (PAR-Q). Next, they were submitted to an anamnesis examining possible limitations related to the study participation. In addition, body height, mass, and fat percentage (skinfold thickness) measurements were conducted. As a next step, volunteers were positioned on a seated knee extension machine (Master, Minas Gerais, Brazil) while maintaining a hip angle of 110° (angle between the backrest and the equipment seat). For measurement purposes, the lateral epicondyle of the femur was aligned with the rotational axis of the device and the pad of the device placed approximately three cm above the medial malleolus. These positions were registered for future replication in the subsequent tests and training sessions. All test sessions were held at the same time of the day for each volunteer.

Experimental sessions 2 and 40 (CSA - ultrasound measurements)

During these sessions, ultrasound images were recorded for the CSA analysis of the rectus femoris and vastus lateralis muscles. The acquisition procedure for the CSA images was conducted as described by Noorkoiv, Nosaka & Blazevich (2010) and Lacerda et al. (2020). Initially, volunteers remained in a dorsal decubitus position on a stretcher for 15 min. During this period, the anterior regions of both thighs were marked to identify the reference points for the ultrasound image acquisition. Next, the major trochanters and lateral epicondyles of the femurs were identified, and femur length was measured (Fig. 1A). From the proximal end of thigh, 40, 50, 60, and 70% of femur length were identified and marked on volunteer’s skin by using a tape measure and a pachymeter positioned parallel to the intercondylar line. Then, a line with a microporous adhesive tape was attached two cm from each percentage point on the thigh (Fig. 1B) to delimitate the probe guide area for the ultrasound image acquisition (Fig. 1C). Finally, the distances between the intercondylar line and each percentage point on the thighs were recorded for post-test replication. The procedures used to acquire images in the pre-test were the same for the post-test session (40^th session). The latter was started no earlier than 72–120 h following the last training session.

An ultrasound device (MindRay DC-7, Shenzhen, China) was used in an extended-field-of-view mode with a four cm linear transducer. The equipment was configured with 10 MHz frequency, an acquisition rate of 21 frames/s, a depth for the image capture ranging from 7.7 and 9.7 cm, and a gain between 50 and 64 dB. The settings were adjusted for each subject to produce the clearest images of the analyzed muscles. The same experienced examiner (∼120 h of training and 600 images acquired before of the study) conducted the acquisition of two images for each of the given femur percentage lengths (40, 50, 60, and 70%). For the acquisition procedure, the probe was placed transversely in parallel to the intercondylar line using a coupled guide on the subject’s thigh (probe guide) (Fig. 1C). This procedure was performed with constant speed (controlled by a metronome) and lasted between 12 and 15 s depending on the subject’s thigh circumference. Sixteen images per subject were obtained for the rectus femoris and vastus lateralis CSA analysis (8 pre-test + 8 post-test). Following the acquisition procedure, CSAs of each muscle scan were manually demarcated by a blinded examiner using specific software (OsiriX MD 6.0, Bernex, Switzerland) (Fig. 2). For the data analysis, the rectus femoris and vastus lateralis CSA mean values were calculated using two images acquired at each percentage of the femur length. Finally, based on the 40, 50, 60, and 70% length measurements, the sum of four CSAs of each analyzed muscle was calculated, generating a summary CSA value per muscle to avoid a possible misinterpretation based on one measurement site only (Noorkoiv, Nosaka & Blazevich, 2010). This value was used in the statistical data analysis.

Experimental sessions 3, 4 and 41 (strength tests)

The strength tests were conducted during the third session in order to familiarize the subjects with the procedures to be performed during the following session. After positioning the participants in the equipment, a familiarization MVIC test was conducted encompassing two attempts of 5 s in duration with knee flexion angles of 30° and 90° (knee extended = 0°). MVIC tests were conducted with both legs with 2-minute rest periods between each attempt (Lacerda et al., 2020). The testing order was randomized between legs. The same order was maintained during the post-test session. The highest force value registered for each attempt at knee flexion angles of 30° and 90° was used in further data analyses. During the MVIC test, a verbal command was given on which the subject exerted a maximum force against the fixed lever of the knee extensor machine. Visual feedback of the force trace was provided to the subject as well as verbal instruction from the examiners to achieve maximum strength. The load cell raw data (Tedea, Bavaria, Germany) were converted into digital data (Biovision, Wehrheim, Germany) and filtered through a 4^th-order Butterworth low-pass filter with a cutoff frequency of 10 Hz.

The 1RM test familiarization was performed 10 min after the completion of the MVIC test. Initially, according to procedures described by Lacerda et al. (2016) and Lacerda et al. (2020), subjects performed 10 repetitions without any weight on the equipment. The 1RM was determined in concentric mode within a maximum of 6 attempts with 5-minutes rest periods in between (Lacerda et al., 2016; Lacerda et al., 2020). In addition, a 5-minute rest period was given between the tests conducted with each of the lower limbs.

In session 4, the MVIC and 1RM tests of the familiarization session were repeated. These tests were also repeated in the 41supst experimental session with a rest interval of at least 48 h following the previous session 40 (ultrasound measurements). The data measured in sessions 4 and 41 were used for statistical analysis.

Experimental sessions 5 to 39 (training period)

After the initial testing period, the 14-week training commenced (35 training sessions). All participants completed 100% of the training sessions. The experimental protocols consisted of 3–4 sets at 50–60% 1RM with 3-minute rest periods in between. In the 2-s RD protocol, subjects completed each repetition in 2 s (1 s concentric, 1 s eccentric). In the 6-s RD protocol, subjects perform each repetition in 6 s (3 s concentric, 3 s eccentric). The protocols complied with recommendations for resistance training and muscle hypertrophy (ACSM, 2009). Previously, training protocols with similar concentric and eccentric durations were already investigated in our laboratory or in others’ (Lacerda et al., 2016; Lacerda et al., 2020; Sakamoto & Sinclair, 2012). For both protocols, all sets were executed until the subjects were unable to complete the concentric action within the required ROM (70°).

During the first two weeks, training sessions included 3 sets at 50% of 1RM. At third week (6^th training session), the load was increased to 60% of 1RM. From week 9 (20^th training session) until the end of the training period, one more set was added so the participants performed 4 sets at 60% of 1 RM. Given that any variation of the load characteristics in addition to the RD could possibly bias the training adaptations, the present study controlled the load configuration and progression.

Every two weeks, beginning in the third week (6^th training session), 1RM tests for both legs were re-assessed on a weekly basis before the first training session. A 10-minute rest period between the 1RM test and the start of the training session was provided. During these sessions, the 1RM test was conducted at the same day time as in the pre-test to standardize the circadian rhythm, which may possibly influence strength performance.

Experimental sessions 6 and 39 (2^nd and 35^th training sessions) (force and electromyography measurements)

The surface EMG procedure (Biovision, Wehrheim, Germany) followed the recommendations by Hermens et al. (2000). For the rectus femoris and vastus lateralis muscles, bipolar surface electrodes (Ag/AgCl - 3M-2223, Brazil) were aligned parallel to the muscle fiber orientation. Prior to the electrode placement above the muscle bellies, the skin areas were shaved, cleaned with alcohol using a cotton pad. The inter-electrode distance was 4 cm which each electrode to be placed 2 cm distant from the center of the muscle belly. The ground electrode was attached above the patella. After the electrode attachment, a silk paper was used to assess their positions as well as the patella and other relevant points on the skin. In addition, the subject’s two thighs were photographed with the electrodes positioned. These procedures were conducted in the 6^th session to map the electrode positions on the thigh and to verify high reproducibility in the post-test measurements (39^th session).

To measure the ROM and the muscle action durations during both protocols, the angular displacement was recorded using a potentiometer (aligned with volunteer’s knee-joint). For all training sessions, this device was coupled to the rotational axis of the knee extension device. The potentiometer raw data were converted into angular displacement data and filtered through a 4^th-order Butterworth low-pass filter with a cutoff frequency of 10 Hz. The duration of each muscle action was comprised of the time between the maximum (100° of knee flexion) and minimum (30° of knee flexion) angular positions. Thus, the duration of the concentric action corresponded to the period between the maximum and minimum angular positions. In turn, the duration of the eccentric action corresponded to the time between the minimum and maximum angular positions. Additionally, concentric/eccentric durations and the RDs were determined throughout the angular displacement time. This potentiometer data provided online information on a laptop screen for the subjects regarding the duration and ROM data of each muscle action throughout the training sessions and tests (Lacerda et al., 2016; Lacerda et al., 2020). Moreover, a metronome was used to help subjects maintain pre-established RDs.

All electromyographic, load cell, and potentiometer signals were synchronized and converted by an A/D board (Biovision, Wehrheim, Germany) with a sampling rate of 4,000 Hz. DasyLab software (Version. 11.0; Measurement Computing Corporation, Massachusetts, USA) was used to record and process the data. The methodological procedures to record force measurements were detailed in the strength tests section. The electromyographic data acquisition was amplified (factor 500) and filtered (4^th-order Butterworth band-pass filter of 20–500 Hz) to calculate the EMG amplitude as the root mean square. Before commencing each experimental session (6^th or 39^th), subjects were asked to perform a MVIC test for 5 s on the knee extension machine exercise at 60° knee flexion (controlled by the potentiometer). The highest force and EMG amplitude values in the MVIC test were used as a reference for the normalization of the subsequent measurements in the exercise protocols. The EMG amplitude during the MVIC was measured over a 1 s period from 500 ms before the MVIC peak force to 500 ms after (Piitulainen et al., 2013). The mean force and EMG amplitude of the concentric muscle actions for each 10° knee flexion area (100°–90°, 90°–80°, up to 40–30°) was calculated and normalized by the reference values from the normalization test. As a result, relatives force and EMG amplitude × knee-joint angle curves (normalized force and EMG-angle) were assessed. This procedure was performed for each protocol. For the acquisition of the force and EMG amplitude values during experimental sessions 6 and 39, participants performed 3 sets with 50% of the previous 1RM value in each protocol.

Statistical analyses

Statistical analysis was performed with SPSS for Windows version 20.0 (SPSS, Inc., Illinois, USA). The normal distribution was verified by the Shapiro–Wilk test. All data were expressed as mean ± SD. For the estimation of effect sizes, eta squared (η²) values are considered to reflect the magnitude of the differences (effect size) in each treatment with values ≤ 0.010 expressing a trivial effect; values between 0.010 and 0.059 expressing a small effect; values between 0.060 and 0.139 a moderate effect, and values ≥ 0.140 a large effect.

Initially, paired sample t-tests were used to test for differences between the training protocols in baseline values for the main variables analyzed (CSA, 1RM and MVIC), as well as for EMG amplitude and force values obtained during the normalization. The differences between the training protocols in the CSA scores were analyzed through a two-way repeated-measures analysis of variance test (ANOVA) for each muscle separately, having protocol (2-s RD or 6-s RD) and time (pre and post-test) as factors. In case of significant F-values a Bonferroni adjustment was used for comparison purposes. The intra-rater reliability was verified by the intraclass correlation coefficient (ICC_[3,1]). For the ICC calculations were conducted for both CSA measures (rectus femoris and vastus lateralis) and for both the test sessions (pre and post-test).

Similar to CSA measurements, a two-way repeated-measures ANOVA test (protocol × time) was applied for MVIC at 30° of knee flexion scores. However, the 1RM and MVIC at 90° of knee flexion values were significantly different at baseline. Therefore, the baseline values were considered as a covariate, and an analysis of covariance (ANCOVA) was implemented using a within-subject factors model. In addition, the ICC intersession values for 1RM and MVIC were obtained from measures during the third (familiarization) and fourth (pre-test) sessions. The familiarization and pre-test sessions were separated by at least 48 h.

Normalized EMG amplitude-angle relationships for the rectus femoris and the vastus lateralis muscles were established during the 6^th and 39^th sessions to compare EMG amplitude differences between the 2-s and the 6-s RD protocols. A three-way repeated measures ANOVA test (session × protocol × knee joint angle) was conducted to analyze the training effects in the normalized EMG amplitude for each muscle. Similar to EMG amplitude responses, a three-way repeated measures ANOVA test (session × protocol × knee joint angle) was used to compare normalized force–angle relationships in the 6^th and 39^th sessions. When necessary, a post hoc Bonferroni honest significant difference test was used to identify the differences reported in the ANOVA’s. Furthermore, the EMG amplitude and force values for each protocol obtained during the normalization test from experimental sessions 6 and 39 were compared by t-test. This procedure aimed to identify possible differences in measurements in both lower limbs of the same individual. Thus, the feasibility of comparing the EMG amplitude and force responses of the two training protocols should be established.

In addition, paired sample t-tests were used to compare the RDs, ROM and TUT mean values for all sets during training sessions between investigated protocols. Finally, given the number of repetitions for each protocol does not meet the precepts for a parametric analysis, Wilcoxon test was used to compare the values in this variable for both protocols. This data is presented as median (number repetitions per set) and interquartile interval values. The level of the error probability/statistical significance was set at p ≤ 0.05 for all statistical tests.

CSA

The intra-rater reliability values found in these sessions were 0.99 for both analyzed muscles. No significant interaction was observed between protocol and time for the rectus femoris CSA (F_1,9 = 1.889; p = 0.203; η² < 0.010 “trivial”). Also, no significant differences between groups were found (main protocol effect; F_1,9 = 0.001; p = 0.972; η² <  0.010 “trivial”). Both training protocols showed significant increases in rectus femoris CSA after training period (main time effect; F_1,9 = 64.353; p < 0.001; η² = 0.821 “large”) (Fig. 3A).

In addition, no significant interaction was observed between protocol and time for the vastus lateralis CSA (F_1,9 = 0.867; p < 0.376; η² < 0.010 “trivial”). No significant difference between the 2-s and the 6-s RD protocols were found (main protocol effect; F_1,9 = 0.009; p = 0.928; η² <  0.010 “trivial”). Both training protocols showed significant increases in vastus lateralis CSA after training period (main time effect; F_1,9 = 50.664; p <  0.001; η² = 0.814 “large”) (Fig. 3B).

1RM

The ICC intersession value for 1RM tests was 0.98. There were significant difference in 1RM values at baseline (Pre; t₉ = 2.688; p = 0.025). When baseline differences in 1RM values were taken into account (within-subject factors ANCOVA), no significant main protocol effect was found (F_1,9 = 0.412; p = 0.820; η² <  0.010 “trivial”) (Fig. 4).

MVIC at 30° of knee flexion

The ICC intersession value for MVIC test at 30° of knee flexion was 0.96. A significant interaction was observed between protocol and time for MVIC test at 30° of knee flexion measurements (F_1,9 = 6.576; p = 0.030; η² = 0.049 “small”), with higher values for the 6-s RD protocol. Furthermore, a significant main effect of time (F_1,9 = 7.581; p = 0.023; η² = 0,273 “large”) was detected. Conversely, no significant main effect of protocol was found (F_1,9 = 1.990; p = 0.192; η² = 0.101 “moderate”) (Fig. 5).

MVIC at 90° of knee flexion

The ICC intersession value for MVIC test at 90° of knee flexion was 0.94. There were significant differences at baseline (Pre; t₉ = 2.640; p = 0.027). When baseline differences in MVIC test at 90° of knee flexion values were taken into account (within-subject factors ANCOVA), no significant main protocol effect was found ( F_1,9 <  0.001; p = 0.996; η² <  0.010 “trivial”) (Fig. 6).

Normalized EMG amplitude-angle relationship

The EMG amplitude values obtained during the normalization tests showed no significant difference between the 2-s RD and the 6-s RD training protocols (rectus femoris - t ₁₉ = 0.701, p = 0.503) (vastus lateralis - t ₁₉ = 0.773, p = 0.455).

No significant interaction between time × protocol × knee joint angle was observed for the normalized EMG amplitude data of rectus femoris (F_6,54 = 2.025; p = 0.143; η² <  0.010 “trivial”) and vastus lateralis muscles (F_6,54 = 2.640; p = 0.106; η² <  0.010 “trivial”). However, significant differences in the rectus femoris EMG amplitude were observed between the 2-s RD and the 6-s RD training protocols in all knee-joint angles analyzed during 6^th and 39^th sessions (protocol × knee-joint angle interaction - F_6,54 = 66.554; p <  0.001; η² = 0.199 “large”) (Figs. 7A, 7B). The 2-s RD protocol resulted in larger normalized EMG amplitude scores in six of the seven knee-joint angles analyzed (100−90° to 50−40°). Conversely, the 6-s RD protocol provided significantly larger rectus femoris EMG amplitude in the last knee-joint angle (40–30°) only. The same results were verified for the vastus lateralis EMG amplitude (protocol ×knee joint angle interaction - F_6,54 = 51.007; p <  0.001; η² = 0.179 “large”) (Figs. 8A, 8B).

In addition, no significant interactions were detected between time × knee joint angle for rectus femoris (F_6,54 = 2.643; p = 0.070; η² <  0.010 “trivial”) and vastus lateralis muscles (F_6,54 = 0.414; p = 0.866; η² <  0.010 “trivial”). Also, no significant interactions were detected between time × protocol for rectus femoris (F_1,9 = 0.168; p = 0.692; η² <  0.001 “trivial”) and vastus lateralis muscles (F_1,9 = 0.014; p = 0.910; η² <  0.001 “trivial”). No significant main effect for the time factor (rectus femoris: F_1,9 = 0.143; p = 0.714; η² <  0.010 “trivial”) (vastus lateralis: F_1,9 = 2.434; p = 0.153; η² = 0.043 “small”) was detected. In contrast, significant main effects were found for the training protocol (rectus femoris: F_1,9 = 29.46; p <  0.001; η² = 0.202 “large”) (vastus lateralis: F_1,9 = 16.131; p = 0.003; η² = 0.225 “large”) and for knee-joint angle (rectus femoris: F_6,54 = 15.673; p < 0.001; η² = 0.111 “large”) (vastus lateralis: F_6,54 = 10.179; p >0.001; η² = 0.037 “small”).

Normalized force–angle relationship

The force values obtained during the normalization showed no significant difference between both training protocols analyzed (t₁₉ = 0.732; p = 0.478).

Significant differences in the normalized force–angle relationship were observed between the 2-s RD and the 6-s RD training protocols during 6^th and 39^th sessions (protocol × knee-joint angle × time - F_6,54 = 2.652; p = 0.025; η² = 0.033 “small”). In the 6^th session, the 2-s RD protocol exhibited significantly larger normalized force values in the first four knee-joint angles analyzed (100−90° to 70−60°). The same was true for the first three knee-joint angles (100−90° to 80−70°) in the 39^th experimental session. In contrast, for the 6-s RD protocol larger normalized force values were found in the last two knee-joint angles (50−40° 40−30°) during sessions 6 and 39. In addition, significant changes in the force ×angle relationships between training sessions were detected only for the 2-s RD protocol. These changes were related to a reduction in the force values from 70−60° until 40−30° of knee flexion (Figs. 9A, 9B).

Moreover, significant interactions were found between the time x knee joint angle (F_6,54 = 14.243; p <  0.001; η² = 0.011 “small”) and the protocol x knee-joint angle (F_1,9 = 296.591; p <  0.001; η² = 0.370 “large”). No significant interaction was observed for the interaction of time × protocol (F_1,9 = 2.483; p = 0.150; η² < 0.010 “trivial”). Finally, no significant main effect for time (F_1,9 = 1.807; p = 0.212; η² = 0.074 “moderate”) and for protocol (F_1,9 = 1.555; p = 0.703; η² <  0.010 “trivial”), but a significant main effect for knee-joint angle (F_6,54 = 7.453; p <  0.001; η² = 0.026 “small”) were identified.

RD, TUT, Number of repetitions and ROM

As expected, 6-s RD protocol showed longer average RD than 2-s RD protocol (2.04 ±  0.08 s; 5.98 ±  0.09 s, respectively; t₆₉ = 284.488, p <  0.001). In addition, larger TUT mean values were observed in the 6-s RD protocol (mean for all sets = 43.47 ±  10.92; 1^st set = 52.5 ±15.83, 2^nd set = 42.58 ±  12.25, last set (3^rd or 4^th) = 36.98 ±  10.89) as compared to the 2-s RD protocol (mean for all sets = 30.51 ±  7.52; 1^st set = 38.1 ±10.55, 2^nd set = 29.57 ±  7.98, last set (3^rd or 4^th) = 25.09 ±  7.17) (t ₆₉ = 15.951; p <  0.001). In regard to the number of repetitions, the Wilcoxon test showed significantly larger median values for the 2-s RD protocol (median for all sets = 14 [12–17]; 1^st set = 18 [21.25–17], 2^nd set = 14.5 [16.25–12.75], last set (3^rd or 4^th) = 12 [14–11]) as compared to the 6-s RD protocol (median for all sets = 7[6–8]; 1^st set = 9 [8–10], 2^nd set = 7 [6–8], last set (3^rd or 4^th) = 6 [5–7]) (U₆₉ = 7.294; p <  0.001). Last not least, but no significant differences were detected between the 2-s RD and the 6-s RD protocols in the ROM average values (2-s RD: 70.77 ± 0.79° -; 6-s RD: 70.99 ±  0.65° -; t ₆₉ = 1.903, p = 0.06).