An anterior cruciate ligament injury does not affect the neuromuscular function of the non-injured leg except for dynamic balance and voluntary quadriceps activation

Table 1 shows the group characteristics of the ACL-deficient patients awaiting surgery (16 men, 16 women) and healthy volunteers (20 men, 20 women). Patient inclusion criteria were: age 18–30 years, unilateral ACL tear with/without partial meniscal resection, and time between ACL injury and testing <2 year. Patient exclusion criteria were: previous ACL reconstruction, history of a lower limb injury that required surgery, pregnancy, current or prior neurological conditions. Controls were between age 18–30 years and had no history of orthopaedic, cardiovascular, neurological, and cognitive impairments. Controls were recruited via ads on social media, where we specifically asked for active and sedentary persons. After recruitment, controls were subdivided into an active and less-active group based on the physical activity level (i.e. hours spent on sport per week). The ten most active men and women were allocated to the active group, and the ten least active men and women were allocated to the less-active group. We have also quantified the level of physical activity through the Tegner activity score [42]. Leg dominance was determined using the Waterloo Footedness Questionnaire [12].

As a warm-up, each participant started with 5 min of cycling on a bicycle ergometer. Next, maximal knee flexor and extensor strength, quadriceps force accuracy and variability, knee joint proprioception, voluntary quadriceps activation, static and dynamic balance, and single-leg hop distance were measured. Every participant performed every test with each leg randomized between legs.

Following strictly the manufacturer’s guidelines and our own previous protocols, we have measured isometric and dynamic (concentric and eccentric) quadriceps and hamstring MVCs on an isokinetic dynamometer (Biodex Medical Systems, Shirley, NY, USA) [7, 10, 11, 17, 27, 28]. Participants’ knee range of motion for the concentric and eccentric contractions was set between 0° (full knee extension) and 90° of knee flexion. After a thorough familiarization with the contraction conditions, participants performed three isometric MVCs at 65° of knee flexion [28], three eccentric MVCs at 60°/s, and six concentric MVCs each at 60, 120, and 180°/s. There was a 1-min pause between conditions. The order of quadriceps and hamstring contractions and the order of isometric and dynamic MVCs were alternated between participants. The peak torque value, normalized to body weight, was used in the statistical analysis.

Quadriceps activation was assessed with twitch interpolation and the central activation ratio (CAR) during isometric contractions [5, 31, 43, 44]. Participants were strapped to the seat of a custom-built dynamometer [46], with the hips and knees in 90° flexion and the arms folded in front of the chest. We have stimulated the quadriceps through two 10 × 14 cm aluminium foil electrodes, covered with water-soaked sponges (cathode: middle of rectus femoris, anode: distal 10 cm above patella), connected to a high-voltage stimulator (Digitimer DS7AH, Welwyn Garden City, UK) that discharged two pulses 10 ms apart (200-µs pulse, 100 Hz). We refer to the force evoked by a doublet as a twitch. The torque signal was amplified, sampled at 500 Hz (CED Power 1401 Plus; Cambridge Electronic Design, Cambridge, UK), visually inspected on a monitor, and recorded and offline-analysed by software (Spike 2, version 5.21). The protocol consisted of: 1. Three isometric quadriceps MVCs; 2. Maximal twitch torque determination during contractions at 10 % MVC (to remove slack); 3. Superimposed twitches at 30, 50, 75, and 100 % of MVC; 4. Two twitches at rest from which the higher of the two was classified as potentiated twitch.

At 10, 30, 50, and 75 % of MVC, we have computed a ratio as: (superimposed twitch/potentiated twitch) *100 %. The ratio for each contraction intensity was plotted against the respective force upon which the twitch was superimposed. A linear regression equation (y = ax + b) was then generated for each participant to determine the estimated maximal force and voluntary muscle activation (Fig. 1). The estimated maximal force was determined by calculating the intersection point with the x-axis, and voluntary activation was derived by determining the intersection point with the y-axis using the actual MVC torque [5]. The CAR was calculated as: MVC/(MVC + superimposed twitch) * 100 %.

Participants have matched the produced torque as steadily and accurately as possible with the target torque displayed as a horizontal line on the monitor set to 20 % of MVC for the isometric trials and to 40 Nm for the dynamic trials [27, 28]. After familiarization, participants performed three isometric trials at 65° of knee flexion (5-s duration) and four concentric and eccentric trials at 20°/s between 90° and 10° of knee flexion. The order of dynamic and isometric contractions was rotated between participants. Force accuracy and variability were computed in the final 3-s portion of the data for isometric trials and the middle 2-s portion for dynamic trials. Force accuracy was the absolute difference between the produced torque and the target torque. Force variability was the coefficient of variation (i.e. SD of the produced force divided by the mean force). Force accuracy and force variability were calculated for each data point, and the average across the trials was used in the statistical analysis.

Knee joint proprioception was measured, in a random order, at 15, 30, 45, and 60° of knee flexion using a joint repositioning task [27]. Knee joint proprioception was computed as the absolute difference between the actual leg position and the target position and was expressed in degrees.

Static balance was measured using the one-leg standing balance test, starting with eyes-open followed by eyes-closed condition [2]. The maximum score that participants could obtain was 60 s. The best score of the two trials was used in the statistical analysis.

The star excursion balance test (SEBT) was used to assess dynamic balance [19]. The normalized scores from the eight directions were averaged to create a composite score used for the statistical analysis. After 5-min of rest, the measurement continued with the other leg as the stance leg.

Participants performed the single-leg hop test for distance, allowing the use of the arms to accelerate [9]. The hop distance was measured from the toe at push-off to the heel where the participant landed. The maximal hop distance was used in the analysis. All participants provided written informed consent to the experimental procedures, which were approved by the medical ethics committee of the University Medical Center Groningen (ID 2012.362) and in accordance with the Declaration of Helsinki.

Data in the text and figures are presented as mean ± SD (SPSS version 22). Each variable was checked for normality. A one-way ANOVA was used to test for differences between groups in age, mass, height, BMI, and the amount of physical activity. Between-group differences in sex and Tegner activity score were tested using, respectively, a Chi-square and a Kruskal-Wallis test. A group (3) by leg (2) MANOVA was performed to test the between-leg differences in quadriceps MVCs (5 conditions), hamstring MVCs (5 conditions), voluntary quadriceps activation (5 conditions), force accuracy (3 conditions), force variability (3 conditions), proprioception (4 conditions), and static balance (2 conditions). Pillai’s Trace was used to determine between- and within-subject effects. A significant MANOVA was followed up by univariate ANOVAs. Dynamic balance and single-leg hop distance were analysed using a group by leg one-way ANOVA. In addition, we converted the outcome on every neuromuscular measure to a z-score. The z-scores were averaged per neuromuscular function (i.e. quadriceps MVCs, hamstring MVCs, voluntary quadriceps activation, force accuracy, force variability, proprioception, static balance, dynamic balance, and single-leg hop distance), and a mean z-score calculated across these nine functions was used to test the overall difference in neuromuscular function between legs. Significant F values from the ANOVA’s were subjected to a Tukey HSD post hoc pairwise comparison to determine the means that were different. The level of significance (α) was set at P < 0.05.

The sample size was based on a previous study reporting bilateral impairments in quadriceps strength and activation in ACL-deficient patients [31]. About 50 % more ACL-deficient patients were included compared to Lepley et al. [31], because our ACL patients would be less homogeneous with regard to the time since injury.

ACL-deficient patients were all recreational athletes, and 29 of 32 sustained a non-contact ACL injury, ruptured the ACL on the non-dominant side (N = 17), and reported relatively few knee complaints on a visual analogue scale (mean 28 ± 15, 0 no and 100 severe pain) [14]. The time between injury and testing was 208 ± 145 days (range 60–664 days) and between testing and surgery was 23 ± 17 days (range 2–62 days).

Table 1 shows the group characteristics. The groups did not differ in age, sex, mass, height, BMI, or leg dominance (all n.s.). Less-active controls had a lower Tegner score and a shorter duration of sport participation per week than ACL patients prior to injury and active controls (P < 0.01). In addition, these two variables were, respectively, 61 and 45 % lower for ACL patients after injury compared to active controls (P < 0.001).

Table 2 shows the static and dynamic quadriceps MVCs. The MANOVA showed a leg (F _5,65 = 8.4, P < 0.001) and a group by leg interaction effect (F _10,132 = 3.9, P < 0.001). Follow-up of univariate ANOVAs showed an interaction effect for all five MVC conditions (all P ≤ 0.018) caused by the greater between-leg differences in ACL patients than controls.

The MANOVA for hamstring MVCs showed a leg main effect (F _5,65 = 3.3, P = 0.010) and a group by leg interaction (F _10,132 = 2.5, P = 0.010). Follow-up by univariate ANOVAs showed an interaction effect for eccentric and isometric contractions (P ≤ 0.033) caused by the greater between-leg difference in ACL patients versus controls (Table 2).

Table 3 shows the voluntary quadriceps activation data. The MANOVA for quadriceps activation revealed a between-group difference (F _10,132 = 2.1, P = 0.028), a leg main effect (F _5,65 = 3.3, P = 0.011), and a group by leg interaction (F _10,132 = 3.1, P = 0.001). CAR in ACL patients was lower than in active controls (P = 0.002), and there was a greater between-leg difference in ACL patients versus controls for isometric MVCs and estimated maximal force.

MANOVAs did not show any statistical effects in quadriceps force accuracy and variability and knee joint proprioception (all n.s., Table 3).

Table 4 shows the multi-joint neuromuscular data. The MANOVA showed no effects for static balance (all n.s.). The ANOVA for dynamic balance revealed a group effect (F _2,69 = 9.0, P < 0.001) and group by leg interaction (F _2,69 = 6.0, P = 0.004). Dynamic balance in ACL patients was poorer compared with controls (P ≤ 0.004) and showed a greater between-leg difference in ACL patients and less-active controls than active controls. The ANOVA for single-leg hop distance showed a group by leg interaction (F _2,69 = 11.4, P < 0.001); between-leg differences were greater for ACL patients and less-active controls than active controls.

Figure 2 illustrates the group by leg interaction effect for overall neuromuscular function (F _2,69 = 7.0, P = 0.002) caused by better overall neuromuscular function in the non-injured leg (P < 0.05).