2.2. Robotic simulations

Fourteen (14) lower extremity cadaveric specimens from seven unique donors (age = 47.2 ± 8.9 years; mass = 876 ± 201 N; BMI = 29.6 ± 5.8) were acquired from an anatomical donations program (Anatomical Gifts Registry, Hanover, MD). Two additional specimens from one additional donor were excluded from data processing due to a damaged and non-functional ACL Methods of specimen preparation have been explicitly documented in previous literature (Bates et al., 2015; Boguszewski et al., 2011; Herfat et al., 2012b). Briefly, specimens were stored at −20 °C and thawed one day prior to testing. Specimens were resected of all soft tissue outside the knee joint capsule, leaving all intra-articular joint structures intact. A joint coordinate system was then defined using anatomical landmarks digitized with a coordinate measuring machine (Faro Digitizer F04L2, FARO Technologies, Inc., Lake Mary, FL) (Grood and Suntay, 1983). Custom mechanical fixtures then aligned and affixed the mechanical axes of the tibia with the primary axes of a 6-axis load cell (Theta Model; ATI Industrial Automation, Apex, NC) that was mounted on the end effector of a six-degree-of-freedom (6-DOF) robotic arm (KR210; KUKA Robotics Corp., Clinton Township, MI). The femur was then fixed to a static table. This setup allowed the robotic manipulator to articulate the tibia around the femur as it simulated the recorded, in vivo kinematics about the knee joint center point (Bates et al., 2015; Boguszewski et al., 2011; Herfat et al., 2012b). Prior to simulation, each specimen was articulated to 45° flexion and 3 mm microminiature differential variable resistance transducers (DVRT, LORD MicroStrain, Inc., Williston, VT) were implanted parallel to fiber alignment just superior to the tibial insertion site on the anteromedial bundle of the ACL as well as on the MCL (Beynnon et al.,1992; Levine et al., 2013). MCL insertion sites were distal to the femoral origin, midsubstance across the joint line, and proximal to the tibial insertion.

Initial limb orientation was different for each of the four simulated athletic tasks (male DVJ, male sidestep cutting, female DVJ, female sidestep cutting). For each task, initial limb position was matched to within 0.5° of the in vivo orientation that corresponded with initial ground contact for all three rotational DOFs. From this initial position, specimens were incrementally loaded in compression until a peak force of 2.0–2.5 bodyweights was achieved for DVJ simulations and 1.5–2.0 bodyweights was achieved for sidestep cutting simulations (Bates et al., 2013, 2015). Simulations were performed at room temperature and the joint was constantly hydrated with saline. All four motions were simulated on each specimen irrespective of specimen gender. Motions were mirrored to accommodate both right and left sides of each pair and identical kinematics pathways were articulated on each specimen. Prior to each simulation, specimens were run through 10 preconditioning cycles to minimize viscoelastic effects. After preconditioning, an additional 10 cycles were simulated where joint forces, joint torques, and ligament strains were recorded (Bates et al., 2015). Following the simulation of each in vivo recorded motion task, the robotic manipulator articulated each limb through isolated 4° rotations in each the internal, external, abduction, adduction, combined internal/abduction, and combined external/adduction DOFs. These limb articulations were performed relative to the initial contact orientation during landing for each motion task. This was done to assess ligament strain response relative to isolated knee rotations previously attributed to ACL injury risk (Hewett et al., 2005; Oh et al., 2012). After testing the knee was dissected down to the ligaments and the load cell was used to determine zero strain conditions based on the point of initial loading. In this isolated condition, where the ligament was the only structure transmitting force across the joint, the specimen was articulated back to initial contact orientation, compressed to an unloaded position, and slowly distracted until the force sensor first registered a constant distraction force. This location was determined to be the neutral strain position of the ligament, which allowed for the calculation of absolute ligament strain rather than strain relative to the DVRT insertion site. The remaining ligament was then resected and all simulations were repeated in a bone-only condition.