Oral contraceptive pill use and the susceptibility to markers of exercise-induced muscle damage

A total of 18 females were divided into two groups: women taking a monophasic OCP (n = 9, 23.4 ± 2.4 years of age, 70.0 ± 9.9 kg and 1.70 ± 0.05 m) and eumenorrheic women who had never taken the OCP [(non-users) n = 9, 21.2 ± 1.5 years of age, 63.2 ± 5.6 kg and 1.65 ± 0.09 m]. All participants self-reported as being recreationally active (undertaking no more than 1 h of “moderate” physical activity per week) and did not take part in any structured resistance training. All procedures complied with the Declaration of Helsinki World Medical Association (2013) and ethical approval was obtained through the local ethics committee at Manchester Metropolitan University. All women reported regular menstrual cycles, documenting an average cycle length of 28 ± 1 days. On average the OCP users had been taking a combined monophasic OCP, with an ethinyl estradiol dosage between 20 and 30 µg for 3 ± 1 years. The types of OCP used within the current study are taken daily followed by a 7-day pill-free interval. Exclusion criteria included any resistance training undertaken in the last 6 months, occupation or lifestyle that required regular heavy lifting or carrying, any known muscle disorder, the use of dietary supplements (for example, multivitamins), and any musculoskeletal injury in the last 3 months. Further exclusion criteria included, previous use of any other forms of hormone-based contraception, continuous pill OCP or mini-OCP users, irregular menstrual cycles [where regular cycles were defined as 24–36 days (Cole et al. 2009; Landgren et al. 1980)] in the last 12 months, and pregnancy in the year preceding inclusion in the current study. All inclusion and exclusion criteria were determined through participant questionnaire prior to inclusion within this investigation.

Once selected, participants were asked to visit the laboratory on five different occasions over 9 days. Non-users were tested on the 14th day (self-reported) of the menstrual cycle to measure oestrogen levels at ovulation (Brown 1955; Cole et al. 2009). To reflect the timing of ovulation in the non-users and to try and avoid the secondary peak in oestrogen levels (Legro et al. 2008), OCP user’s oestrogen levels were tested on the 7th pill taking day (Elliott et al. 2005). The 7th pill taking day is equivalent to the 14th day of the overall pill cycle (7 pill-free days plus 7 pill taking days), therefore the measurement day will be referred to as the 14th day of the cycle from this point forward. To avoid an acute peak in oestrogen levels the OCP users were not tested within 2 h of taking their OCP (Sneader 2005). The design and several of the measurement techniques within the current study have been reported previously (Hicks et al. 2013). The testing sessions were as follows: (1) pre-damage; (2) damage (48 h post pre-damage); (3) 48 h; (4) 96 h and (5) 168 h post-damage (see Fig. 1). Pre-damage assessments consisted of stature and mass (anthropometric measures), patella tendon moment arm, 5–6 ml blood sample (for serum CK levels), dynamometer familiarisation (within pre-damage), morphological and mechanical measures of the patella tendon (tendon size and stiffness, respectively) and maximal voluntary isometric knee extension (MVC_KE) torque measurements at six angles [60°, 65°, 70°, 75°, 80° and 90° (full extension = 0°)]. Participants performed two practice maximal voluntary contractions, at two knee joint angles during the familiarisation session. Stature and mass were measured using a wall-mounted stadiometer (Harpenden, Holtain Crymych, UK) and digital scales (Seca model 873, Seca, Germany), respectively. The damage session was on the 14th day of the cycle (pill or menstrual) in OCP users and non-users. The damage session consisted of maximal voluntary eccentric knee contractions (MVE_KE), 5–6 ml venous blood sample (for serum CK and oestrogen levels), rating of muscle soreness and MVC_KE torque measurements. 48, 96 and 168 h testing session consisted of 5–6 ml blood sample (for serum CK levels), rating of muscle soreness, and MVC_KE torque measurements at six angles (60°, 65°, 70°, 75°, 80° and 90°).

All tests were carried out in the non-dominant leg. The non-dominant leg was defined as the leg that provided stability during movements such as kicking a ball. Participants were seated in an isokinetic dynamometer (Cybex Norm, Cybex International, NY, USA), with a hip angle of 85°. To reduce any extraneous movement during maximal efforts, participants were secured in a seated position using inextensible straps secured around the shoulders and hips. The isokinetic dynamometer axis of rotation was visually aligned with the knee joint’s centre of rotation. During pre-damage, the isokinetic dynamometer settings (chair position, distance from the dynamometer, dynamometer arm length) and anatomical zero (leg fully extended, knee angle 0°) were recorded to ensure accurate repeatability in the following sessions.

The method used to determine vastus lateralis (VL) cross-sectional area has been reported in detail previously (Hicks et al. 2013). In brief, in a supine position, VL anatomical cross-sectional area (VL_ACSA) was measured using a real-time B-mode ultrasound (AU5 Harmonic, Esaote Biomedica, Genoa, Italy). At 50% of VL muscle length echo-absorptive markers (Transpore surgical tape, Medisave UK Ltd, UK), were placed in parallel at intervals of 30 mm. The ultrasound probe (7.5 MHz linear array probe, 38 mm wide), was held perpendicular to the VL muscle in the axial plane, and moved steadily over the echo-absorptive markers from the lateral to the medial edge of the muscle. The images were recorded in real-time at 25 frames per second (Adobe Premier pro Version 6, Adobe Systems Software, Ireland). Using video editing software (Adobe Premier Elements, version 10), still images were acquired at each 30 mm interval. The shadows cast by the echo-absorptive markers allowed the neighbouring still images to be aligned, thus reconstructing the entire VL_ACSA in a single image (Adobe Photoshop Elements, version 10). Digitising software (ImageJ 1.45, National Institutes of Health, USA) was used to measure VL_ACSA.

At six different knee angles [60°, 65°, 70°, 75°, 80° and 90° (full extension = 0°)], participants were instructed to perform two MVC_KE lasting ~2 s at the plateau, with 90 s rest between contractions. Torque was presented, in real-time, on a Macintosh G4 computer (Apple Inc., Cupertino, CA, USA), via an A/D converter (Biopac Systems, Santa Barbara, CA, USA). Torque measurements were later analysed offline with the accompanying software (Acknowledge, version 3.9.2). The highest peak torque produced at each angle was taken as MVC_KE peak torque. The knee angle where the highest torque was produced was recorded as optimal knee angle. In a randomised order, MVC_KE torque measurements were repeated at all six knee angles 60 min post-eccentric exercise [to reduce any fatigue effect (Walsh et al. 2004)] and 48, 96 and 168 h post-eccentric exercise. MVC_KE torque loss was calculated from optimal knee angle identified at pre-damage.

A real-time B-mode ultrasound (AU5 Harmonic, Esaote Biomedica, Genoa, Italy) was used to measure patella tendon cross-sectional area and patella tendon length at a fixed 90° knee angle. The distance between the apex of the patella and the tibial tuberosity, marked using sagittal ultrasound images, was taken as patella tendon length. To measure patella tendon cross-sectional area, the ultrasound probe was placed in the transverse plane and images were captured at 25, 50, and 75% of patella tendon length. The images were later analysed offline using image analyses software, ImageJ (1.45, National Institutes of Health, USA) patella tendon cross-sectional area is presented as a mean of all three images (O’Brien et al. 2010).

The participants were seated in the isokinetic dynamometer, with the knee angle fixed at 90°, and were instructed to perform a ramped, isometric MVC_KE lasting ~5–6 s. Ramped MVC_KE torque and displacement of the patella tendon were synchronised using a 10-V square wave, signal generator. Patella tendon displacement was measured over two MVC_KE, once with the ultrasound probe positioned over the distal edge of the patella and on the second contraction over the tibial tuberosity (Onambélé et al. 2007), so that total displacement would be computed from the composite of proximal and distal patella motions (see below). Torque was presented on a Macintosh G4 computer (Apple Inc., Cupertino, CA, USA), via an A/D converter and subsequently analysed with the accompanying software (Acknowledge, Biopac Systems, Santa Barbara, CA, USA). To create an external marker on the ultrasound images, an echo-absorptive marker was placed on the skin. Using the marker to calculate displacement, the distance of the marker (shadow) from an anatomical reference point at the beginning of the contraction to the position of the shadow at the end of the contraction was calculated. A square wave, signal generator was used to synchronise the ultrasound images with the torque acquisition system. Images were captured at ~10% intervals of ramped MVC_KE torque (Onambélé et al. 2007). Total patella tendon displacement was calculated as displacement at the apex of the patella plus the displacement at the tibial tuberosity (Onambélé et al. 2007). Patella tendon forces were calculated as: (MVC_KE torque + antagonist co-activation torque)/patella tendon moment arm.

Patella tendon moment arm was measured at 90° (full extension = 0°) in the sagittal plane, from a dual-energy X-ray absorptiometry scan (frame 23.3 cm × 13.7 cm, Hologic Discovery, Vertec Scientific Ltd, UK), and subsequently analysed using a DICOM image assessment tool (OsiriX DICOM viewer, ver. 4.0, Pixemo, Switzerland). Patella tendon moment arm length was determined as the perpendicular distance from the centre of the patella tendon to the tibio–femoral contact point. Dual-energy X-ray absorptiometry scans have recently been shown to be a reliable and valid method of measuring patella tendon moment arm length when compared to the gold standard method (MRI) (Erskine et al. 2014).

For the calculation of tendon force, the calculation of antagonist co-activation torque is described below. The patella tendon force–elongation curve constructed from data analysed at every 10% MVC_KE, was then fitted with a second-order polynomial function forced through zero (Onambélé et al. 2007). The tangential slope at discreet sections of the curve, relative to MVC_KE force, was calculated by differentiating the curve at every 10% patella tendon force interval. In addition, to standardise the comparison of patella tendon stiffness at an absolute load, the slope of the tangential line, corresponding to the MVC_KE force of the weakest participant, was computed for each subject.

In order to compute patella tendon force, the co-activation of the biceps femoris during MVC_KE was measured. To determine co-activation of the biceps femoris during the ramped MVC_KE, electrodes (Ambu, Neuroline 720, Denmark) were placed in a bipolar configuration at 25% of biceps femoris muscle length (distal end = 0%). A reference electrode (Ambu, Blue Sensor, Denmark) was placed on the lateral tibial condyle. The raw electromyography signal was amplified (×2000) and filtered (through low and high pass filters of 10 and 500 Hz, respectively, notch at 50 Hz) with the sampling frequency set at 2000 Hz. Ramped MVC_KE torque and biceps femoris electromyography were recorded in real-time and synchronised using a (10-V) square wave signal generator. Participants performed two maximal voluntary isometric knee flexions (MVC_KF) at 90°. They were instructed to perform MVC_KF rapidly and as forcefully as possible and instructed to relax once a 2-s plateau had been attained (as observed on the dynamometer screen display). The root mean square of the biceps femoris electromyography signal was calculated 500 ms either side of instantaneous MVC_KF maximal torque from the contraction corresponding to the highest MVC_KF torque. Prior to contraction the baseline signal noise was calculated as the integral root mean square over 1 s and removed from the measured electromyography during MVC_KF and MVC_KE. At every 10% of ramped MVC_KE torque the absolute integral of the biceps femoris electromyography was taken over 250 ms. Co-activation torque was calculated as: (biceps femoris electromyography during ramped MVC_KE/biceps femoris electromyography during MVC_KF) × MVC_KF torque at 90° knee angle (Onambélé et al. 2007). This equation assumes that the biceps femoris is representative of the entire hamstring (Carolan and Cafarelli 1992) and that a linear relationship exists between biceps femoris electromyography and MVC_KF torque (Lippold 1952).

Patella tendon strain was calculated as a ratio of total patella tendon displacement and patella tendon length. Patella tendon stress was calculated as: patella tendon force (N)/patella tendon cross-sectional area (mm²).

Young’s modulus was calculated as: patella tendon stiffness × [patella tendon length (mm)/patella tendon cross-sectional area (mm²)].

The damaging protocol used within the study has been described in detail previously (Hicks et al. 2013). Following a warm-up of 10 isokinetic knee extensions and knee flexions, participants performed six sets of 12 MVE_KE. For the eccentric exercise, the knee extension range of motion was set at 20°–90° (0° = full extension). The eccentric phase of the contractions was performed at an isokinetic angular velocity of 30°/s (Jamurtas et al. 2005). The concentric phase was performed passively at an angular velocity of 60°/s. Two minutes rest was provided between each set. Participants remained seated in the isokinetic dynamometer throughout the entire exercise protocol, including rest periods. Visual feedback and verbal encouragement were continuously provided throughout the protocol. MVE_KE torque was recorded throughout each contraction and displayed via the torque acquisition system. For each set, peak MVE_KE torque was determined as the highest torque out of the 12 repetitions. Average peak MVE_KE torque was calculated as an average of peak MVE_KE across six sets.

Muscle soreness was measured using a visual analogue scale. The visual analogue scale consisted of a 100 mm line, with 0 mm labelled “No pain at all” and 100 mm labelled “Unbearable pain”. Seated in the isokinetic dynamometer, the leg was passively moved through a full range of motion at 30°/s. Participants were asked to mark a line perpendicular to the visual analogue scale to denote the level of pain they experienced during the passive movement. The visual analogue scale has been reported to be a reliable measure of muscle soreness [ICC > 0.96 (Bijur et al. 2001)].

Venous blood samples were taken to measure CK and oestradiol levels. A 21-gauge needle was inserted into the antecubital vein of the forearm using a 10 ml syringe. 5–6 ml of blood was drawn into a serum collection tube. The sample was left on a crushed ice bed for 60 min. The sample was then centrifuged at 4500 rpm at 0 °C for 10 min. Using a 200–1000 µl pipette (Eppendorf, Hamburg, Germany), the resulting serum sample was separated into three aliquots (~500 µl each) and stored in eppendorfs at −20 °C until CK and oestradiol analysis was performed.

Creatine kinase levels were measured using a standard colorimetry procedure, measuring at optical density 340 nm (BioTek ELx800 96 well Microplate Reader) and immediately analysed (Gen5, version 2.0). Each sample was run in duplicate-quadruplets using an EnzyChrom^TM CK Assay Kit (BioAssay Systems, Hayward, CA, USA, sensitivity 5 U/l, manufacturer intra-assay variability ≤5%). An average of 2–4 readings were taken. Creatine kinase activity is reported in two ways, firstly as absolute values and secondly, as the increase from pre to peak CK over the 168 h [(ΔCK_peak), i.e. the peak CK value − the pre-CK values].

Oestradiol was measured using a standard enzyme-linked immunosorbent assay (ELISA) procedure (Alpha Diagnostic International, San Antonio, USA). Absorbance was measured at optical density of 450 nm (BioTek ELx800 96 well Microplate Reader). The minimal oestradiol detection was ~10 pg/ml, the manufacturers intra-precision and inter-precision was 9.85 and 10.3%, respectively. To calculate oestrogen levels, a standard curve was plotted using the six standards against their absorbance. Using the mean absorbance of each sample, the concentration of the sample was read directly from the standard curve.

Statistical analyses were carried out using the statistical software package SPSS (v.19, Chicago, IL, USA) for Windows and Microsoft Excel. To ensure that the data were parametric, the Levene’s and Shapiro–Wilk’s tests were used to assess the variance and normality of the data. If parametric tests were violated, the equivalent non-parametric tests were used. For group differences in anthropometric measures, oestrogen levels, VL_ACSA, patella tendon properties, and ΔCK_peak independent T tests were used. A 2 × 5 mixed design analysis of variance [ANOVA, between factors: OCP use (2 levels) and time from EIMD (5 levels)] was used for muscle soreness, MVC_KE torque loss and CK levels. Wherever the assumption of sphericity was violated, the Greenhouse–Geisser correction was used. When a significant group effect was found a planned contrast, with LSD correction, was used to identify where the significant difference lay. Since an association was reported between age and MVC_KE torque loss, an analysis of covariance was used. For the variables being investigated as a potential determinant of EIMD, confidence levels and Cohen d were calculated. Significance was set at p ≤ 0.05. The data are presented as mean ± standard deviation.

OCP users were significantly older by 2.2 years (p = 0.010) and of an 11% greater mass (p = 0.047) than non-users. There was no significant difference in stature between OCP users and non-users (p = 0.061).

There was no significant difference in VL_ACSA between OCP users and non-users (21.9 ± 2.64 and 20.1 ± 3.40 cm², respectively, p = 0.224).

Patella tendon properties for OCP users and non-users are presented in Table 1. Patella tendon length, patella tendon cross-sectional area and patella tendon moment arm were not significantly different between OCP users and non-users (p ≥ 0.05). Furthermore, ramped MVC_KE torque and patella tendon force were not significantly different between OCP users and non-users (p ≥ 0.05). Patella tendon elongation was not significantly different between OCP users and non-users (7.20 ± 1.30, 8.16 ± 0.80 mm, respectively, p = 0.075). Tendon stiffness (Fig. 2) at MVC_KE was not significantly different between OCP users and non-users [p = 0.153, CI (−502, 88.0), d = 0.73]. Young’s modulus was not significantly different between OCP users and non-users [p = 0.213, CI (−190, 428), d = 0.38]. Patella tendon stiffness calculated at a standardised patella tendon force (2330 N) was not significantly different between OCP users and non-users (725 ± 219 and 565 ± 103 N/mm, respectively [p = 0.070, CI (−10.9, 331), d = 0.94]. Patella tendon Young’s modulus calculated at a standardised patella tendon force (2330 N) was not significantly different between OCP users and non-users [628 ± 219 and 536 ± 190 MPa, respectively (p = 0.175 CI (−111, 298), d = 0.453)].

There was no significant difference in the average peak MVE_KE torque of each set (six in total) between OCP users and non-users (p ≤ 0.05). There was no significant difference in the average peak MVE_KE torque (averaged over 6 sets) between OCP users and non-users (182 ± 33 and 167 ± 32 Nm, respectively, p = 0.093). Peak MVE_KE torque made relative to pre-MVC_KE torque was significantly higher in OCP users compared to non-users (108 ± 16.7 and 85.0 ± 11.9%, respectively, p = 0.002).

Pre-damage, optimal MVC_KE knee angle was not significantly different between OCP users and non-users (median, 75.0 ± 7.0°, and 75.0 ± 7.0°, respectively, p = 0.430). Post-EIMD both OCP users (80.0 ± 7.0°, p = 0.010) and non-users (median 80.0 ± 7.0°, p = 0.030) demonstrated a significant rightward shift in MVC_KE optimal angle. There was no significant difference in the magnitude of the rightward shift in MVC_KE optimal angle between OCP users and non-users (p = 0.311).

MVC_KE torque loss expressed as a percentage of pre-damage MVC_KE torque is illustrated in Fig. 3. A two-way repeated mixed ANOVA for MVC_KE torque loss post-EIMD, reported a significant main effect of time (p = 0.0004); however, no group effect (OCP users versus non-users p = 0.257) nor interaction (times versus group, p = 0.118) was reported. Pre-MVC_KE peak torque was not significantly different between OCP users and non-users (169 ± 24.3 and 194 ± 36.2 Nm, p = 0.055). Although a significant tendency was reported, peak MVC_KE torque loss made relative to pre-damage MVC_KE torque was not significantly different between OCP users compared to non-users (17 ± 7 and 28 ± 13%, respectively [p = 0.055, CI (−21.5, 10.5), d = 0.94)].

A significant correlation was observed between age and MVC_KE torque loss (r = 0.58, p = 0.011). Age was therefore considered as a covariate of MVC_KE torque loss. A subsequent ANCOVA revealed that no significant difference in MVC_KE torque loss between OCP users and non-users remained when age was accounted for (p = 0.179).

A two-way repeated mixed ANOVA for CK reported a significant main effect of time (p = 0.040); however, no significant group (OCP users versus non-users, p = 0.057) or interaction (time versus group, p = 0.576) effects were reported (Fig. 4). ΔCK_peak was significantly higher in OCP users compared to non-users over the 168 h [962 ± 968 and 386 ± 474 UL, respectively (p = 0.016, CI (−186, 1338), d = 0.76)].

A two-way mixed measures ANOVA for muscle soreness reported a significant main effect of time (p = 0.0004); however, there was no significant group difference (OCP users versus non-users, p = 0.598) or an interaction effect (p = 0.127). For both OCP users and non-users peak muscle soreness occurred 48 h post-EIMD. There was no significant difference in peak muscle soreness between OCP users and non-users (40 ± 3 mm and 41 ± 1 mm, respectively, p = 0.358).