Hamstrings Stiffness and Landing Biomechanics Linked to Anterior Cruciate Ligament Loading

Participants

A total of 36 healthy individuals (18 men, 18 women; age = 23 ± 3 years, height = 1.8 ± 0.1 m, mass = 73.1 ± 16.6 kg) volunteered to participate. Participants had no history of ACL injury, lower extremity surgery, neurologic disorder, or lower extremity musculoskeletal injury within the 6 months before data collection and were physically active, involved in at least 20 minutes of physical activity 3 times per week. All participants provided written informed consent, and the study was approved by the University of North Carolina at Chapel Hill Biomedical Institutional Review Board.

Procedures

We collected data during a single testing session in which we assessed hamstrings stiffness and landing biomechanics in a counterbalanced order. Because hamstrings stiffness does not differ between limbs in healthy individuals,¹² all data were sampled from the right lower extremity only. Hamstrings maximal voluntary isometric contractions (MVICs) were performed to determine standardized loading conditions for the stiffness assessment. Participants were positioned prone with the right hip supported in 30° of flexion and the foot fixed to a loading device so the knee was maintained in 30° of flexion and the calcaneus was in contact with a load cell (model 41; Honeywell Sensotec, Columbus, OH) (Figure 1A). Participants performed a submaximal warm-up contraction, then a 5-second maximal knee-flexion effort during which we sampled load cell data.

View Figure

• Download as Powerpoint
• Download Figure

We assessed hamstrings stiffness by modeling the knee as a single-degree-of-freedom mass-spring system and observing the damping effect imposed by the hamstrings on oscillatory knee flexion and extension. We positioned participants identically as for the hamstrings MVICs except the foot was free to move (Figure 1B). A splint was secured over the plantar aspect of the foot and posterior shank to standardize ankle position and gastrocnemius length, and weights representing 45% MVIC were secured near the ankle. The investigator (J.T.B. or M.F.N.) positioned the participant's shank parallel to the floor, and the participant contracted the hamstrings isometrically to maintain this position. Within 5 seconds after this contraction, the investigator applied a downward manual perturbation to the calcaneus, extending the knee and initiating oscillatory knee flexion and extension. Participants were instructed not to intervene with the perturbation^7,13 and to attempt to keep the hamstrings active only to the level necessary to support the shank in the testing position. This oscillatory motion was recorded via the tangential acceleration of the shank segment obtained from an accelerometer (model 356A32; PCB Piezotronics, Inc, Depew, NY) attached to the splint on the posterior shank and foot. The time interval between the first 2 oscillatory peaks (t₁ and t₂) was used to calculate the damped frequency of oscillation (Figure 2). Next, we used this value to calculate stiffness via the equation k = 4π²mf², where k is stiffness, m is the system mass (shank and foot segment¹⁴ + 45% MVIC), and f is the damped frequency of oscillation. Participants performed 5 trials separated by 30-second rests to reduce the likelihood of fatigue. Given that stiffness increases as a function of mass,^15,16 we normalized stiffness values to participant mass before statistical analyses. This method has been used extensively to assess lower extremity muscle stiffness^7,8,12,16 and demonstrated excellent intrasession reliability in this investigation (intraclass correlation coefficient [2,1] = 0.82, standard error of the mean = 2.90 N/m·kg⁻¹).

View Figure

• Download as Powerpoint
• Download Figure

The musculotendinous unit is viscoelastic/loading-rate sensitive, so the magnitude of perturbation influences its stiffness. Therefore, fluctuations in the magnitude of the manual perturbation applied to the shank may have influenced the stiffness values. To address this potential limitation, we calculated the magnitude of perturbation as the product of the peak downward acceleration immediately after perturbation onset (Figure 2) and the system mass (shank and foot segment¹⁴ + 45% MVIC). The average perturbation magnitude was 56 ± 11 N, and the average coefficient of variation across trials (standard deviation / mean) was 12% ± 5%. Furthermore, the regression of normalized stiffness on perturbation magnitude (r² = 0.03) demonstrated that variance in stiffness attributable to variance in perturbation magnitude was negligible.

The landing task involved a double-legged jump landing from a 30-cm height located at 50% of the participant's height from 2 force plates (model 4060; Bertec Corporation, Columbus, OH). Participants were instructed to minimize upward displacement upon leaving the box, to land with each foot centered on a single force plate, and to perform a maximal vertical leap immediately after ground contact. Electromagnetic motion-capture sensors (miniBIRD 800; Ascension Technology Corporation, Burlington, VT) were placed on the pelvis over the sacrum, midthigh, and proximal anteromedial shank to measure lower extremity kinematics (Figure 3). Three-dimensional coordinate data and ground reaction forces were sampled during the landing task. Participants performed 3 practice trials to familiarize themselves with the task, then performed 5 recorded trials separated by 30-second rests to reduce the likelihood of fatigue. The foot segment was not included in the biomechanical model of the lower extremity due to the potential for substantial motion artifact in foot-segment kinematics near ground contact.¹⁷ Given that peak kinetic ACL-loading mechanisms and ACL injury likely occur shortly after initial ground contact,^18,19 these errors in ankle-joint motion can substantially influence knee-joint kinetics.¹⁷ Instead, we used a static foot contribution in the model in which angular acceleration of the foot segment was assumed to be zero. Authors^4,20–23 of previous investigations of ACL-loading mechanisms have used similar models. Unpublished data from our research group have indicated that excluding the foot segment has a negligible effect on knee-joint kinetics during this time interval (Appendix).

View Figure

• Download as Powerpoint
• Download Figure

Data Sampling and Reduction

Electromagnetic sensor coordinate data were sampled at 100 Hz, and force-plate, accelerometer, and load-cell data were sampled at 1000 Hz. Coordinate data were low-pass filtered at 10 Hz with a fourth-order Butterworth filter, time synchronized to the analog data, and resampled to 1000 Hz via linear interpolation. The world, segment, and force-plate axis systems were established so that positive values for the x-axis, y-axis, and z-axis were defined as forward/anterior, leftward/medial, and upward/superior, respectively. We created a segment linkage model of the lower extremity by digitizing the medial and lateral femoral epicondyles and malleoli and the left and right anterosuperior iliac spines. Locations of the knee- and ankle-joint centers were defined as the midpoints of the digitized medial and lateral femoral epicondyles and malleoli, respectively, whereas the hip-joint center was estimated as a function of the 3-dimensional distance between the digitized left and right anterosuperior iliac spines.²⁴ We calculated knee-joint angles as Euler angles (YXZ sequence) defined as motion of the shank reference frame relative to the thigh reference frame such that flexion, varus, and internal rotation represented positive values. Force-plate data were low-pass filtered at 50 Hz with a fourth-order Butterworth filter and combined with kinematic and anthropometric data to derive the net forces acting on the shank segment and internal knee moments via an inverse dynamics solution.²⁵ Load-cell and accelerometer data were low-pass filtered at 10 Hz with a fourth-order Butterworth filter. Biomechanical variables were derived via commercial motion-capture software (The Motion Monitor; Innovative Sports Training, Inc, Chicago, IL).

For MVICs, we calculated a consecutive 100-millisecond moving average (ie, all 100-millisecond intervals) from which we used the largest hamstrings force value to represent the MVIC. Peak kinematics (knee flexion and valgus angles) and kinetics (vertical and posterior ground reaction forces, anterior tibial shear force, and internal knee-extension and -varus moments) were identified during the loading phase of the landing, which was defined as the interval from initial ground contact (vertical ground reaction force > 10 N) to peak knee-flexion angle. Ground reaction forces and anterior tibial shear force were normalized to weight (N), and joint moments were normalized to the product of weight and height (N·m).⁴ We also identified knee-flexion angles at the instant of each peak kinetic variable.

Statistical Analyses

Mean values for each dependent variable were calculated across the 5 trials for each task. We arranged stiffness data into tertiles (n = 12 per tertile) and performed independent-samples t tests to compare stiffness and landing biomechanics between the highest and lowest tertiles (ie, high-stiffness and low-stiffness groups). The use of tertiles to define the high-stiffness and low-stiffness groups is arbitrary and not based on meaningful thresholds for stiffness. However, this grouping of participants eliminated individuals who overlapped the extremes of the distribution (ie, the middle tertile), creating 2 distinct groups with different mean stiffness values. All analyses evaluated 1-tailed hypotheses with the α level set a priori at equal to or less than .05. We used SPSS (version 19; IBM Corporation, Armonk, NY) for all analyses.

Kinematics

Peak knee-flexion (t₂₂ = 0.970, P = .17) and -valgus (t₂₂ = 0.971, P = .28) angles did not differ between the high-stiffness and low-stiffness groups. However, knee-flexion angle was greater in the high-stiffness group at the instants of peak internal knee-varus moment (t₂₂ = 2.224, P = .02), peak internal knee-extension moment (t₂₂ = 1.743, P = .048), and peak anterior tibial shear force (t₂₂ = 1.729, P = .049). We found no other group differences in kinematics. Values for each kinematic variable for the high-stiffness and low-stiffness groups are detailed in Figure 4.

View Figure

• Download as Powerpoint
• Download Figure

Kinetics

Peak internal knee-extension moment, internal knee-varus moment, and anterior tibial shear force occurred 77 ± 42 milliseconds, 72 ± 54 milliseconds, and 119 ± 74 milliseconds, respectively, after initial ground contact. Peak internal knee-varus moment was smaller in the high-stiffness than low-stiffness group (t₂₂ = 2.221, P = .02) (Figure 4B). Given that the net internal joint moment reflects the musculoskeletal response to external loading, these data indicate less knee-valgus loading (ie, a smaller internal knee-varus moment) in individuals with greater hamstrings stiffness. A statistical trend indicated that anterior tibial shear force was smaller in the high-stiffness than low-stiffness group (t₂₂ = 1.537, P = .07, observed power = 0.44) (Figure 4A). This difference represents a moderate-to-large effect size (0.63), and post hoc power analyses²⁶ indicated that 32 participants per group would have been necessary to provide an a priori power of 0.80. We found no other group differences or statistical trends (ie, P < .10 for kinetics). Values for each kinetic variable for the high- and low-stiffness groups are detailed in Figure 5.

View Figure

• Download as Powerpoint
• Download Figure

Limitations

The biomechanical data in this investigation were derived from a double-legged landing in a laboratory setting. Whereas similar tasks have been demonstrated to predict ACL injury risk,² we do not know if this task represents those during which ACL injury occurs or if the data are comparable with those of other more challenging dynamic tasks, such as single-legged landings and cutting tasks. Furthermore, the segment-linkage model of the lower extremity we used to calculate knee kinetics did not include the foot segment. We excluded the foot segment to avoid the kinematic noise caused by ground contact and the associated errors introduced to the calculations of knee-joint kinetics.¹⁷ Whereas this model produces kinetic values that are slightly larger than those resulting from models that include the foot segment, unpublished data from our research team indicated that these errors are small and likely had a negligible influence on our results (Appendix). These results may be specific to the task, however, because we calculated knee-joint kinetics soon after ground contact, when ankle-joint contributions to inverse dynamics were small. The participants also constituted a convenience sample of recreational athletes, and our definition of minimal physical activity was somewhat liberal; therefore, they may not represent individuals in whom ACL injury risk is greatest.⁵⁰ Last, the oscillatory technique for measuring musculotendinous stiffness reflects contributions from the knee-flexor musculature. However, using identical methods, Blackburn et al¹³ demonstrated that the contributions from the gastrocnemii were negligible. We attempted to standardize any contributions from the gastrocnemii by restricting sagittal-plane ankle motion via application of the rigid splint. Furthermore, the gastrocnemii can introduce ACL loading by producing posterior translation of the femur relative to the tibia,⁵¹ thus it is unlikely that they contributed to the “better” biomechanics profile in the high-stiffness group.

Summary and Clinical Application

Individuals with greater hamstrings stiffness displayed more favorable landing biomechanics for ACL loading and injury risk, as evidenced by smaller frontal-plane knee moments and a more-flexed knee at the instants of critical kinetic events. A trend in the data also suggested greater hamstrings stiffness is associated with less ACL loading attributable to anterior tibial shear force. These findings suggest a high level of hamstrings stiffness may limit ACL injury risk.

Whereas more research is certainly necessary to validate this notion objectively, increasing hamstrings stiffness may have negative consequences for hamstrings musculotendinous injury. Watsford et al⁵² demonstrated prospectively that individuals who sustained acute hamstrings strain injuries had greater hamstrings stiffness than an uninjured cohort. However, whereas the bilateral average hamstrings stiffness differed between groups, that of the injured limb was not different from the uninjured group. Although the implications of these seemingly contradictory findings are unclear, we suggest the existence of an optimal level of hamstrings stiffness at which the risks of ACL injury and musculotendinous injury may be minimized. Future research is necessary to evaluate the injury-specific influences of hamstrings stiffness.

Researchers should attempt to extrapolate these findings to more dynamic and challenging tasks that are more representative of scenarios during which ACL injury occurs and to the populations at heightened risk of ACL injury. Given that muscle stiffness can be enhanced via various training mechanisms, targeting this neuromuscular property may be beneficial for future ACL injury-prevention programs and for rehabilitation of patients with ACL deficiency or reconstruction. However, research is necessary to determine the most effective types of training to enhance hamstrings stiffness and the effects of enhancing hamstrings stiffness on ACL-loading mechanisms and ACL injury risk.