Participants
Twenty-seven soccer players (12 females, 15 males) from the Concordia University varsity teams volunteered to participate in this study and were assessed during the preseason (end of August and the beginning of September 2016). From these, a total of 18 players (11 females, 7 males) were available and reassessed at the end of the competitive playing season (mid-November 2016). All available players were invited to participate to maximize the sample size, and thus no a priori sample size calculation was made. The exclusion criteria included previous history of severe trauma or spinal fracture, previous spinal surgery, observable spinal abnormalities, as all of these can affect paraspinal muscle morphology and/or function. Pregnancy was also an exclusion criterion as undergoing a DEXA scan was a requirement of this study. The study was approved by the Research Ethical Committee of the Institution and by the Central Ethics Committee of the Quebec Minister of Health and Social Services. All players that participated in this study provided informed consent.
Procedures
A self-administrated questionnaire was used to collect information on players’ demographics and history of LBP during at the preseason. LBP was defined as pain localized between T12 and the gluteal fold with or without leg pain [16]; players were asked to answer “yes” or “no” to the presence of LBP during the past 3-months prior to the assessment. A visual Numerical Pain Scale (NRS) was used to assess the average LBP intensity (e.g. 10 point scale; 0 = no pain, 10 = worst pain possible). Players were also asked to indicate the LPB location (e.g. centered, right side, left side) and duration (in months) at both time points. Finally, players were questioned about their history of lower limb injury within the past 12-months and to provide the injured body part, if applicable. Similarly, at the end of the competitive season, players completed a related questionnaire asking about whether they experienced or suffered a lower limb injury during the season.
Ultrasound
LMM assessments were performed using a LOGIQ e ultrasound machine (GE Healthcare, Milwaukee, WI) with a 5-MHz curvilinear probe. The imaging parameters were kept consistent for all acquisitions (frequency: 5 MHz, gain: 60, depth: 8.0 cm). The reliability of ultrasound imaging to assess LMM size and thickness has been previously established (intra- and inter-rater reliability ICCs = 0.94–0.99 [17]. LMM thickness change measurement is also highly correlated to EMG activity (r = 0.79, p < 0.001) [18].
LMM measurements
Players were placed in a prone position, on a therapy table, with a pillow under their abdomen to minimize lumbar lordosis [17]. They were instructed to relax the paraspinal musculature, and the spinous process of L5 was palpated and marked on the skin with a pen prior to imaging. For the assessment of LMM CSA, acoustic coupling gel was applied to the skin and the ultrasound probe was placed longitudinally along the midline of the lumbar spine to confirm the location of the L5 level [18]. Then, the probe was rotated and placed transversally over the L5 spinous process for imaging. Transverse images at L5 level were obtained bilaterally to assess LMM CSA, except for athletes with larger muscles, where the left and right sides were imaged separately. A total of 3 images were captured and saved for each side. The L5 level was selected as the level of assessment based on a previous study in elite AFL players reporting that decreased LMM CSA and increased side-to-side asymmetry, at this level, was a predictor of lower limb injury [5].
LMM function (e.g. contraction) was then evaluated by obtaining thickness measurements at rest and during contraction via a contralateral arm lift. For the thickness measurement, the LMM was imaged in the parasagittal view, which allows for the visualization of the L5/S1 zygapophyseal joints. Players were instructed to relax, while 3 images of LMM thickness were captured bilaterally, at rest. Players were then instructed to perform a contralateral arm lift holding a handheld weight [based on players’ body weight 1) < 68.2 kg = 0.68 kg weight, 2) 68.2–90.9 kg = 0.9 kg weight, 3) > 90.9 kg = 1.36 kg weight] while raising the loaded arm 5 cm off the therapy table (shoulder was placed in 120° of abduction and elbow 90° of flexion), in order to induce a submaximal (~ 30%) LMM isometric contraction [17,18,19]. While performing this task, players were instructed to maintain the position for 3 s and hold their breath at the end of normal exhalation, in order to minimize the effect of respiration on the thickness measures. Each player first had a practice trial, followed by 3 repeated contralateral arm lifts on each side.
Similarly, LMM measurements were then obtained in the standing position. Players were asked to stand barefoot on the floor with their arms relaxed on each side [20]. To achieve a habitual standing posture, they were instructed to first march on a spot for few seconds and remain in the position where their feet landed [20]. LMM CSA and thickness measurements at rest were obtained using the same procedure as describe above. To contract the LMM in this position, players performed a contralateral arm lift with the shoulder placed in 90° of flexion, with complete elbow extension and wrist in a neutral position (palm facing down) [20]. The same handled weight as previously determined for the prone measurements was also used to perform this task. Players maintained the position for 3 s and first had a practice trial, followed by 3 repeated contralateral arm lifts on each side.
Images assessment
Ultrasound images were stored and analyzed offline using the OsiriX imaging software (OsiriXLiteVersion 9.0, Geneva, Switzerland). LMM CSA measurements were obtained by manually tracing the muscle borders on both sides, as showed in Fig. 1. The relative % asymmetry in LMM CSA between sides was assessed and calculated as follows: % relative asymmetry = [(larger side – smaller side)/larger side × 100]. The LMM thickness measurements (at rest and contracted) were obtained using linear measurements from the tip of the L5/S1 zygapophyseal joint to the inside edge of the superior muscle border (Fig. 2), in both the prone and standing positions. Each LMM measurement was obtained 3 times for each side, on 3 different images, and the average value was used for analysis. The following formula was used to assess the LMM contraction: thickness % change = [(thickness contraction – thickness rest)/thickness rest) × 100]. LMM EI was assessed using grayscale and standard histogram function (e.g. pixels expressed as a value between 0 (black) and 255 (white)) from the ImageJ software (National Institute of Health, USA, Version 1.49) [21]. Previous evidence confirmed that enhanced EI is indicative of a greater amount of intramuscular fat and connective tissue [22]. This measure was acquired by manually training the LMM region of interest (ROI), representing the CSA using the transverse ultrasound images obtained in the prone position, while avoiding the inclusion of surrounding bone or fascia. All LMM measurements were acquired by an experienced blinded researcher, with over 9 years of experience in spine imaging analysis. The rater also received prior training by a senior musculoskeletal ultrasound radiologist prior to the beginning of this study. The intra-rater reliability of the same rater for all LMM measurements (ICC3,1) was tested in a previous related study [23] and ranged between 0.96–0.99, 0.96–0.98 and 0.99 for the prone, standing and EI LMM measurements, respectively.
DEXA
A full body DEXA scan (Lunear Prodigy Advance, GE) was obtained for each player and performed by a certified medical imaging technologist. All players removed any metal and were required to wear loose-fitting clothing, to avoid interference with the scan. The following information was entered into the system computer software prior to imaging: Age, height, weight, and ethnicity. Players were instructed to lie down supine in the center of the scanner, with their arms slightly away from the body, thumbs pointing upwards, and legs slightly apart with their toes pointing upwards. Total lean mass, total bone mass, total fat mass, and total percent body fat were acquired and used in the analysis.
Statistical analysis
Means and standard deviations were calculated for players’ characteristics and body composition measurements. Paired t-tests were used to assess the difference in LMM characteristics between the right and left sides within male and female players, and analysis of variance (ANOVA) was used to assess the difference in LMM characteristics between male and female players. The associations between LMM characteristics, LBP and lower limb injury were initially examined using univariate linear regression. Height, weight, sex and total % body fat were then tested as possible covariates in multivariate analyses. These covariates were retained in the multivariable models only if they remained statistically significant (p < 0.05) or had a confounding effect (led to a ± 15% change in the beta coefficients of significant variables included in the multivariable model). Diagnostic plots (e.g. qq-plots and pp-plots) were used to evaluate the normality assumption. Finally, Pearson correlation and linear regression models were used to assess the relationship between LMM measurements of interest and body composition measurements. All analyses were performed with STATA (version 12.0, StataCorp, LP, College Station, Texas).