This study is a three-arm randomized placebo-controlled trial. This study was approved by the Human Research Ethics Committee of the Universidade Estadual Paulista (Protocol number: 1.389.046/2016) and registered at ClinicalTrials.gov (ID: NCT02918916). Prior to the start of the trial, participants received oral and written instructions regarding the procedures and objectives of the study and signed an informed consent form agreeing to participate in the study.

The study was conducted at Universidade Estadual Paulista. Participants were recruited from the Laboratory of Physical Therapy in Sports database via online media, telephone, or text message. Participants included in this study met the following inclusion criteria: male aged between 18 and 30 years old and considered healthy (by self-report). Individuals who reported history of muscle injury in the lower limbs or spine in the previous six months were excluded from the study. Participants were instructed to maintain their daily diet routine and to continue with their daily exercise routine, without change in training volume or intensity. All participants were instructed to ingest 30 g of carbohydrate-based gel supplement followed by 180 ml of water 20 min before the baseline assessment or training to standardize the energy reserve.

The sample size calculation was based on data from a previous study24 that used the same PBM device for a three-arm parallel study to detect a 10% difference between groups on maximal voluntary isometric contraction (MVIC), with standard deviation (SD) of 28.71 Nm, with significance set at 0.05 and power at 0.8. A total sample of 11 per group was required.

A stratified randomized procedure was performed based on the concept of responders and non-responders to adaptation capacity.25 The training adaptation capacity was recorded before and after the first six-week of the exercise training program. The adaptation capacity was calculated using the equation AC=0.5FP+0.5CP, where AC is adaptation capacity, FP is the functional parameter (i.e. strength test) and CP is the clinical parameter (i.e. perception of recovery). Participants were first stratified into three strata (tertiles) based on their adaptation capacity, then participants from each stratum were allocated into one of the three groups by block randomization. To ensure allocation concealment, randomization procedure was performed by an independent researcher who was not involved in the recruitment of participants and outcome assessment. The complete randomization process was described in the published protocol.23

To prevent the influence of the primary neuromuscular adaptation to exercise on the results, all participants had to participate in a period of six weeks of exercise training. This training period aimed to induce neuromuscular adaptation and equalize the level of physical activity among the participants. After this period, the participants were randomly allocated, according to their adaptation capacity, into one of the three groups that comprise the study: active PBM group (AG), placebo PBM group (PG) and control group (CG). Participants in each of the groups performed another six-week of exercise training with the adjusted loads and received the intervention they were allocated to between sprint and squats training.

The training program was adapted from an individualized combined training described by Marques et al.26 The training consisted of a combination of sprints and explosive squats performed twice per week for six-weeks (48 h between sessions). A detailed description of the training is presented in Table 1.

Recovery strategies were applied between sprint and squats training due to the preliminary hypothesis of possible association of ergogenic and protective effects of PBM in the same session.

Active PBM was applied using a MR4 LaserShower 50 4D emitter (Multi Radiance Medical, USA), to six sites of the quadriceps (two centrally – rectus femoris and vastus intermedius; two laterally – vastus lateralis; two medially – vastus medialis).24 The decision to apply the intervention at these sites was made because this muscle group is most directly involved during squats and sprints. The dosage (30 J per site; 180 J per muscle) was based on findings of Antonialli et al.24 More details are provided in Table 2.

Table 2.

Photobiomodulation parameters.

Number of lasers	4 super-pulsed infrared	The application was performed in direct contact with the skin, bilaterally, one member at a time with six sites simultaneously.
Wavelength (nm)	905
Frequency (Hz)	250
Peak power (W) – each	12.5
Average mean optical output (mW) - each	0.03125
Power density (mW/cm²) – each	0.07
Dose (J) – each	0.07125
Spot size (cm²) – each	0.44
Number of red LEDs	4 red
Wavelength (nm)	640
Frequency (Hz)	2
Average mean optical output (mW) - each	15
Power density (mW/cm²) – each	16.66
Dose (J) – each	3.42
Spot size (cm²) – each	0.9
Number of infrared LEDs	4 infrared
Wavelength (nm)	875	Magnetic field (mT)	35
Frequency (Hz)	16	Irradiation time per site (s)	228
Average mean optical output (mW) - each	17.5	Total dose applied (J)	180
Power density (mW/cm²) – each	19.44	Aperture of device (cm²)	20
Dose (J) – each	3.99
Spot size (cm²) – each	0.9

The same procedures were applied to the placebo group; however, the PBM probe was disabled. The therapist responsible for the application of active and placebo PBM was not aware of the equipment settings which was programmed by an independent researcher.

The control group participants performed passive recovery (remained seated) supervised by an independent researcher while the other groups were receiving the recovery strategies. Participants from all groups remained seated to receive the recovery strategies.

The MVIC was considered the primary outcome. Secondary outcomes were other functional and clinical outcomes and VEGF, as described below. The same assessor was responsible for the outcome assessment throughout the study. Outcome assessment was completed over two sessions separated by 24 h. The first session consisted of sprint and squat jump tests. In the second session, strength and power tests and MVIC measurements were performed. An interval of 15 min between tests was used to limit effects of fatigue on subsequent test. This rest is based on time to recovery of each test described in the literature.27–31

Maximal voluntary isometric contraction (MVIC): Participants were positioned with the dominant leg on the isokinetic dynamometer (Biodex System 4 Pro, New York, USA). Prior to the MVIC, a warm-up consisting of 10 repetitions of knee flexion-extension at 180°/s throughout the range of motion was performed. The MVIC was determined by the highest maximal isometric torque over three contractions of five seconds at 60° of knee flexion. The three contractions were separated by a 2-minute rest interval.20

Strength and power tests: Tests were performed using a guided bar.26 Participants were required to perform the squat with continuous movement until their thighs reached the horizontal plane, and immediately to perform the opposite movement at maximum speed without losing feet contact with the ground. For the strength test, the warm-up consisted of 10 repetitions at 40–60% of perceived maximum. A progression from warm-up load was performed with increment of 10 kg. Two minutes rest interval between sets were provided. The last acceptable repetition with the highest possible load was considered as 1-RM. For the power test, the warm-up consisted of five repetitions with the bar. To determine the load moved at 1 m/s the tests were conducted gradually with increments of 10 kg for each set and two attempts for each load. An interval of 3 min was given between attempts. The highest measure with each load was recorded. Information about the distance held eccentrically and concentric speed of each repetition were recorded by a linear velocity transducer (T-Force System Ergotech, Murcia, Spain) and analyzed by its specific software (Isocontrol Dinámico, 3.6, Spain).

Squat jump test: Participants performed three squat jumps on a platform without countermovements (Multisprint, Hidrofit, Brazil). A rest interval of 30 s was considered between jumps. The best trial was used for data analysis.32

Sprint test: Participants performed three maximum sprints exceeding 30 m with rest intervals of three minutes. Performance times at 10, 20, and 30 m were recorded by phothocells (Multisprint, Hidrofit, Brazil). The average of the two best measures was used for data analysis.26

Muscle soreness: Values of muscle soreness in the lower limbs were obtained by a 0–10 points visual analog scale (VAS),20 with ‘zero’ corresponding to absence of soreness and ‘10’ to the maximum soreness a participant felt could be tolerated.33

Pain threshold: Pressure algometry was applied at five specific sites of the quadriceps of the dominant leg.34 Participants were instructed to indicate when the pressure sensation became discomfort. The pressure did not exceed 2.55 kgf.35

Psychological questionnaire: Participants filled out a questionnaire about readiness for exercise, fatigue, vigor, and sleepiness. The information about pain was excluded due to use of a pain VAS.36

Vascular endothelial growth factor (VEGF): 10 ml of blood was collected 24 h before the training and 24 h after the last training session. The plasma from this sample was stored at −80 °C for later analysis. The plasma concentration of VEGF was analyzed using the ELISA method (enzyme-linked immunosorbent assays) following the manufacturer's instructions (R&D Systems, Minneapolis, MN, USA).37

The statistical analysis was conducted using the software SPSS (version 22; SPSS Inc, Chicago, Illinois, USA). Overall, data were normally distributed as tested by Shapiro-Wilk test and visually confirmed by Q-QPlots. Goodness-of-fit was tested by Akaike's Information Criterion (AIC) and therefore the General Mixed Model (GMM) with time and group as fixed factors, subjects as random factor and AR(1) matrix was performed for all outcomes. All tests were performed assuming a significance level of p < 0.05, including Bonferroni post-hoc when main effects were found. Results are expressed as mean difference estimates and standard deviation (SD) with 95% confidence intervals (CI) for group comparisons. Participants who dropped out of the study were excluded from the analysis.