A dissertation submitted to the. Division of Research and Advanced Studies of the University of Cincinnati

Transcription

1

2 Investigation of Anterior Cruciate Ligament and Medial Collateral Ligament Biomechanics during 6-Degree-of- Freedom, Robotically-Simulated Athletic Tasks A dissertation submitted to the Division of Research and Advanced Studies of the University of Cincinnati in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY (Ph.D.) in the Department of Biomedical Engineering of the College of Engineering & Applied Science 2014 by Nathaniel A. Bates B.S., University of Cincinnati, 2009 Committee Chair: Timothy E. Hewett, Ph.D. i

3 Abstract The anterior cruciate ligament (ACL) passively stabilizes the knee and plays a complex role in joint restraint during tibiofemoral articulations. ACL injuries are traumatic events that have short and long term consequences for affected athletes. Unfortunately, treatment through ACL reconstruction fails to completely restore native knee biomechanics or reduce the early onset of osteoarthritis following rupture. Therefore, the best treatment for ACL injuries may be to prevent their occurrence. To enhance ACL injury prevention, investigators must enhance the understanding of underlying, intra-articular mechanics that precede rupture. Use of robotic technology has allowed investigators to better examine native knee biomechanics during simulated clinical tests and gait. However, ACL injuries do not frequently occur during gait, but during athletic tasks that involve rapid deceleration or change in direction. The objective of these studies was to utilize in vivo recorded, three-dimensional kinematics to derive six-degree-of-freedom robotic simulations of athletic tasks that can assess native tibiofemoral mechanics in scenarios related to ACL injury. The created model successfully articulated cadaveric lower extremities though drop vertical jump and sidestep cutting tasks without specimen damage. The ACL serves as a secondary restraint to knee abduction and internal tibial rotation and, therefore, can be loaded through multiple rotational perturbations. Investigators dispute over which planes of motion contribute most significantly to ACL injury. The presented model found that combined knee rotations evoked the greatest ACL strains, but isolated knee abduction accounted for the majority of this loading. The model was then utilized to define how and why concomitant medial collateral ligament (MCL) injuries only occur in 30% of ACL ruptures, ii

4 despite the shared mechanism of abduction loading for both ligaments. It was observed that during controlled athletic tasks the MCL was generally less loaded and strained and, therefore, less exposed to injury risk than the ACL. Finally, ACL injuries are gender-specific events with higher incidence rates in female athletes. Mechanical assessment of sex-specific kinematic simulations of athletic tasks revealed that neither joint loads nor ligament strains exhibited increased injury risk in females. This therefore supported that, for the conditions simulated during these studies, non-contact ACL injuries may be black swan events, a product of unanticipated and abnormal joint loading generated from an unexpected loss of neuromuscular control. Clinically, the current investigations indicated that preventive measures should continue to focus on reduction of knee abduction in order to lower ACL injury incidence. Greater baseline loading within the ACL than the MCL during athletic tasks supported how ACL rupture occurs with limited concomitant MCL injuries. The absence of observed gender differences, relative to ACL protection, indicated gender-specific training and rehabilitation protocols should be unnecessary as structural loading during regulated athletic tasks is comparable. Findings from these investigations advance the understanding of intra-articular knee biomechanics and can be incorporated into efforts to prevent ACL injuries. Future considerations should focus on further development of subject-specific simulation models that address additional sources of joint perturbation as well as the application of present models to the evaluation and efficacious improvement of current repair and reconstruction methods. iii

5 iv

6 Acknowledgements This dissertation is the culmination of five years graduate work. It is a product of my own hard work and dedication, the support of my family and friends, as well as the generous opportunities that have been provided to me. I am most grateful for the experiences I have gained and for the people that have encouraged me along the way. First, I would first like to thank God for allowing me the opportunity to pursue this PhD and for giving me the ability to complete the degree. Through Him all things are possible, and I am most appreciative of all the blessings that He has afforded me. Secondly, I would like to thank my family, who has always firmly endorsed my pursuit of further education. To my parents, Tom and Terri, I would like to thank you for your love. Both of you have always stood behind me and directed me down the straight and narrow path in life. Without your devoted guidance and support, none of this would have been possible. You have pushed me to excel at everything, and the sense of pride you convey in my accomplishments has consistently made me feel both loved and encouraged. I could not have asked for better. I would also like to thank my grandfather Rudolph Kusbit whose profound interest and excitement for my studies was very uplifting throughout grad school. Pap, your continued generosity and financial support was very much appreciated and a significant aid to financing my education. Ginny and Zach, thank you for allowing me to be your annual guest over many Memorial Day weekends. Those trips provided a needed relief from work that I very much looked forward to each year. Russ, Betsy, Aunt Sharon, Uncle Jim, and Aunt Susan, thank you all for making those trips a memorable experience. Finally, Amanda, you are my greatest v

7 encourager and have supported me at every turn. Your unconditional love has been exceedingly generous and I am grateful to have you in my life. I love and thank all of you. I would like to especially thank my committee members, Dr. Tim Hewett, Dr. Jason Shearn, Dr. Marepalli Rao, and Dr. Michael Archdeacon. Tim, you provided me with the opportunity to pursue graduate studies, for which I am truly grateful. Your mentorship and friendship has provided me valuable guidance and instruction along the way. I appreciate everything that you have done to aid me in laying a strong and diverse foundation for my career. Jason, I can say much of the same in regards to your mentorship. It was your encouragement in my senior year that convinced me to pursue a PhD rather than an MS. It was also your willingness to collaborate with Tim that made this project possible. I appreciate the advice, insight, and direction that both of you provided over the past several years, all of which helped me to grow professionally, develop my skills, and prepare for my future career. I would also like to express my gratitude to everyone who impacted my time as a graduate student. Dr. Greg Myer, thank you for being an additional advisor to me throughout this process. Your knowledge and expertise from an area outside of my own has been valuable to my problem solving. Dr. Kevin Ford, thank you for helping me to get my feet underneath me in the early stages of my graduate education. Much of what you taught me provided the foundations and experience I needed to successfully complete projects later in my graduate career. Both of you provided projects that helped to diversify my experience and expand my thinking, for which I am grateful. I would also like to extend a special thanks to Rebecca Nesbitt and Dan Boguszewski. Rebecca and Dan, your assistance in operating the robotic system for biomechanics research was essential to the completion of these projects. Without each of your contributions, these studies would not have been happened. I very much enjoyed working with vi

8 both of you. I feel we made a good team and I immensely value and appreciate all of your contributions. I would also like to thank everyone I worked with from University of Cincinnati Biomedical Engineering, The Ohio State University Sports Health and Performance Institute, and Cincinnati Children s Hospital Medical Center Sports Biodynamics Center, including Dr. David Butler, Dr. Jeffrey Johnson, Linda Moeller, Safa Herfat, Sam Wordeman, April McPherson, Lynn Adams, Chris DiCesare, Kim Foss, Stacey Thomas, Chad Cherney, Aileen Cudia, Ben Rower, Dr. Steph Di Stasi, Mike McNally, Josh Hoffman, Chris Nagelli, Alisa Blazek and many others. It was a pleasure getting to know each of you. Finally, I would like to thank all of my friends. Tom Bek, Barret Daniels, Mo Efimba, Dan Burns, the hockey guys, and everyone else, your camaraderie helped to make grad school life entertaining. I am fortunate to have friends such as you and am glad for your companionships. vii

9 Table of Contents Abstract... ii Acknowledgements... v List of Tables... 3 List of Figures... 4 Chapter 1: Background, Rationale, and Objectives... 6 ACL Anatomy and Function... 6 ACL Injury and Reconstruction... 8 Rationale and Objectives... 9 Specific Aims and Hypotheses Chapter 2: Characteristics of Inpatient Anterior Cruciate Ligament Reconstructions and Concomitant Injuries from Abstract Introduction Methods Results Discussion Chapter 3: Anterior Cruciate Ligament Biomechanics During Robotic and Mechanical Simulations of Physiologic and Clinical Motion Tasks: A Systematic Review and Meta-Analysis Abstract Introduction Methods Results Discussion Chapter 4: A Novel Methodology for the Simulation of Athletic Tasks on Cadaveric Knee Joints with Respect to In Vivo Kinematics Abstract Introduction Methods Results Discussion Chapter 5: Prediction of Kinematic and Kinetic Performance in a Drop Vertical Jump with Individual Anthropometric Factors: Implications for Cadaveric Investigations Abstract

10 Introduction Methods Results Discussion Chapter 6: The Effect of Internal, Abduction, and Combined Tibial Rotations on Anterior Cruciate Ligament and Medial Collateral Ligament Biomechanics at Initial Contact During Simulated Jump Landing and Sidestep Cutting Tasks Abstract Introduction Methods Results Discussion Chapter 7: Relative Strain in Anterior Cruciate Ligament and Medial Collateral Ligament During Simulated Jump Landing and Sidestep Cutting Tasks: Implications for Injury Risk Abstract Introduction Materials and Methods Results Discussion Chapter 8: Gender Based Differences in Anterior Cruciate Ligament and Medial Collateral Ligament Biomechanics in Robotically Simulated Jump Landing and Sidestep Cutting Abstract Introduction Methods Results Discussion Chapter 9: Discussion and Conclusions Chapter 4 Synopsis Chapter 5 Synopsis Chapter 6 Synopsis Chapter 7 Synopsis Chapter 8 Synopsis Chapter 10: Future Perspectives and Recommendations Considerations Related to Further Model Development Considerations Related to ACL Injury Risk Factors Considerations Related to ACL Repair Efficacy Summary Bibliography

11 List of Tables Table 2.1 Table 2.2 Table 2.3 Table 2.4 Table 3.1 Table 5.1 Table 5.2 Table 5.3 Table 6.1 Table 6.2 Patient demographics calculated directly from the NIS data subset for all reported inpatient ACLR cases from National averages extrapolated from the NIS for all inpatient hospitalizations that involved ACLRs from Estimated national averages extrapolated from the NIS database for patients whose primary cause for impatient hospital admittance was ACL injury or ACLR Annual percentages of ACL classifications and concomitant injuries documented within the NIS inpatient ACLR population Forces and moments for the Native ACL and ACL-reconstructed knees at heelstrike, midstance, and toeoff. This table was adapted with permission from Boguszewski Depicts the r 2 values for linear relationships between the four independent variables and 12 kinematic dependent variables Depicts the r 2 values for linear relationships between the four independent variables and 12 kinetic dependent variables Depicts the slope of the best fit line between independent and dependent variables that were found to have both a significant linear relationship and a r 2 > Mean ACL ligament strains recorded for the intact knee and isolated ligament condition in response to rotational stimuli Mean MCL ligament strains recorded for the intact knee and isolated ligament condition in response to rotational stimuli Table 7.1 ACL and MCL biomechanics from uniaxial failure tensioning Table 7.2 Table 7.3 Table 8.1 Table 8.2 Peak forces and torques generated during simulations performed in the ACLisolated condition Peak forces and torques generated during simulations performed in the MCLisolated condition Peak forces and torques recorded at the tibiofemoral joint during robotically simulated DVJ tasks Peak forces and torques recorded at the tibiofemoral joint during robotically simulated sidestep cutting tasks

12 List of Figures Figure 1.1 Anterior view of the intra-articular anatomy of the human tibiofemoral joint... 6 Figure DOF robotic manipulator used to articulate cadaveric knees Figure 2.1 Incidence of inpatient ACLR and charges per patient across time Figure 2.2 Prevalence of knee injuries documented concomitantly with inpatient ACLR Figure 3.1 Flow chart of the systematic review process Figure 3.2 Mean ATT, ITR, and ligament force for the ACL-intact, ACLD, and ACLR conditions in response to a Lachman s test simulated with 134 N ATF Figure 3.3 Mean ATT, ITR, and ligament force for the ACL-intact, ACLD, and ACLR conditions in response to a Pivot-Shift test simulated with a combined 10 N*m abduction and 4-5 N*m internal torque Figure 3.4 Mean ATT and ligament force for the ACL-intact condition at two separate levels of Pivot-Shift loading Figure 3.5 Mean ATT, ITR, and ligament force for the ACL-intact conditions in response to a simulated 134 N ATF, 400 N quadriceps load, and a combined 400 N quadriceps with 200 N hamstrings load Figure 3.6 Mean ATT, ITR, and ligament force throughout non-weight-bearing, passive flexion for the ACL-intact, ACLD, and ACLR condition in response to simulated muscle loads of 400 N quadriceps Figure 4.1 Lower-extremity, cadaveric specimen affixed to the 6-DOF robotic manipulator and prepared for the simulation of athletic tasks at the knee joint Figure 4.2 Rotational knee joint kinematics recorded in vivo with the adjusted input for the robotic manipulator Figure 4.3 Translational knee joint kinematics recorded in vivo (solid line) with the adjusted input for the robotic manipulator Figure 4.4 Failures that resulted after raw in vivo kinematics Figure 4.5 Unfiltered internal knee torques and translational forces during each cycle of a 10-cycle male DVJ simulation on a single specimen Figure 4.6 Knee joint loading for all unique donors throughout a male DVJ simulation...75 Figure 5.1 Data plot of subject mass versus minimum knee flexion angle Figure 5.2 Data plot of subject mass versus knee flexion moment range Figure 6.1 Specimens in the intact knee, ACL-isolated, and MCL-isolated conditions Figure 6.2 Displays the mean change in ACL and MCL ligament strain generated in the intact knee by each rotational stimuli

13 Figure 6.3 Displays the mean change in ACL and MCL ligament strain generated in the isolated ligament condition by each rotational stimuli Figure 7.1 Sagittal view of a lower extremity specimen affixed to the robot manipulator and articulated through a male DVJ Figure 7.2 Frontal plane view of a specimen in the ACL-isolated and MCL-isolated condition Figure 7.3 An ACL specimen prepared for and following the completion of uniaxial failure loading Figure 7.4 Population average absolute strains for the ACL and MCL in the intact-knee and isolated-ligament condition throughout each simulated motion task Figure 7.5 Average peak ACL and MCL strains for the specimen population during each simulated task and compared to the uniaxial failure strain of the respective ligament Figure 7.6 Total knee force during landing phase of athletic tasks. Sum of the average translational components during each simulated task for the intact knee, ACLisolated knee, and MCL-isolated knee Figure 7.7 Total knee torque during landing phase of athletic tasks. Sum of the average rotational components during each simulated task for the intact knee, ACLisolated knee, and MCL-isolated knee Figure 8.1 Frontal and sagittal plane views of the arrangement of DVRTs implanted on a cadaveric specimen positioned at 45 flexion Figure 8.2 Population average force and torque measured at the tibiofemoral joint in response to simulated male and female DVJ and sidestep cut articulations Figure 8.3 Displays peak values for ACL and MCL strain in the male and female DVJ and sidestep cut simulations at four different time periods within landing phase Figure 9.1 A visual roadmap of this dissertation

14 Chapter 1 Background, Rationale, and Objectives ACL Anatomy and Function The anterior cruciate ligament (ACL) is a dense band of soft, fibrous connective tissue that originates from inside the femoral notch on the lateral femoral condyle and travels obliquely to its insertion at the medial tibial eminence (Figure 1.1). 2 Bundled ACL fibers typically range between mm in length with a midsubstance width of 7-12 mm. 3 The ligament ranges in cross-sectional area, being larger at the insertion sites (113 mm 2 & 136 mm 2 ) than in the midsubstance of the ligament (~40 mm 2 ), and exhibits a larger midsubstance cross-sectional surface area in males (44 mm 2 ) than females (36 mm 2 ). 4,5 There exist relatively few neural receptors within the ACL volume; however, the mechanoreceptors that are present within the ligament contribute significantly to the sense of joint position and proprioception. 6 Functionally, the ACL plays a critical role in knee joint stability. The ACL serves as the Figure 1.1: Anterior view of the intra-articular anatomy of the human tibiofemoral joint. primary passive restraint to anterior tibial translation (ATT) with respect to the femur, as it accounts for up to 86% of the force resistance in this direction. 7,8 The ligament is also a secondary resistor to knee abduction torque, 1,7 and there is conflicting data available as to how 6

15 the ligament contributes to the restraint of internal rotation torque. Data from some investigations suggest that the ACL is a secondary resistor to internal tibial torque, much as it is to knee abduction torque; 9-12 whereas, other investigations implicate that the ACL has a negligible or inconsistent influence on internal tibial rotation While the ACL is structurally a single band of tissue, it can functionally be divided into multiple bundles, the two most prominent being the anteromedial and posterolateral. 2,15 These bundles are named for their corresponding positions within the footprints of the tibial and femoral ACL insertion sites. It is generally accepted that these bundles work in a reciprocal fashion, where the anteromedial bundle tightens in knee flexion and loosens in knee extension and the posterolateral bundle loosens in flexion and tightens in extension. 3,15-17 ACL deficiency (ACLD), by injury or otherwise, creates instability at the knee that can lead to a multitude of consequences through joint degeneration. Specifically, the functional instability caused by ACLD can directly lead to multiple long-term complications, including contralateral injury, meniscus injury, failure of secondary stabilizers, and early onset osteoarthritis. 18 Following ACL disruption, the knee will experience greater ATT when anterior tibial force is applied and demonstrate greater internal tibial rotation during gait. 10,11,13,14,16,19-29 These conditions likely lead to the giving-way episodes and change in articular cartilage contact areas documented in ACLD patients Computational models have demonstrated that ACLD altered kinematics contribute to accelerated cartilage thinning and, thus, the early onset of osteoarthritis. 33 Such degradation corresponds with patient data as a majority of ACLD subjects exhibit osteoarthritis years post injury and these patients also report the effects of osteoarthritis typically between years earlier than healthy counterparts. 34 Also, the ACLD 7

16 condition can lead to the failure and degradation of other soft tissue structures around the knee as nearly one-third of ACLD patients received a meniscectomy within 15 years of injury. 35,36 ACL Injury and Reconstruction ACL injury is a traumatic event with both short and long-term impacts that can debilitate an athlete and devastate athletic careers. Immediate effects after injury can include loss of athletic participation, loss of scholarship, increased instability, knee pain, and altered dynamics. 37 In the United States, an estimated 250,00 ACL injuries are sustained every year. 38 These injuries are gender specific events as, when normalized to number of athletic exposures, female athletes experience ACL rupture at a rate 2-6 times greater than their male counterparts Additionally, up to 70% of ACL injuries are incurred in non-contact situations, while the final 30% occur as the result of a person to person or person to object collision. 39,42,43 Often, these non-contact ACL ruptures are preceded by one or several biomechanical factors that serve as predictors or riskfactors to the onset of injury. These factors include motion task performed, joint laxity, 44 knee valgus, 43,45-48 and lack of neuromuscular control. 45,46 Of these factors, both increased valgus torque at the knee and decreased neuromuscular control have been prospectively associated with athletes who went on to ACL injuries as compared to healthy controls. 46 Risk factors can lead to poor joint stability and high mechanical loads during athletic movements like landing, cutting, and pivoting. 45,49-52 Abnormal loading can be an important mechanism for non-contact injury, especially in cutting maneuvers where knee valgus moments are most sensitive. 50,53 Following an ACL injury, a few patient treatment options are available in hopes of restoring some aspects of the native knee stability. Currently, the preferred treatment method for ACL injuries is ACL reconstruction (ACLR) surgery. 54 Each year, approximately 127,000 ACLRs are performed in the United States, which account for 12.9% of all knee arthroscopies. 55 8

17 These procedures involve the surgical removal and replacement of the damaged ACL with either autograft or allograft material that are anchored in tunnels drilled into the femoral condyle and tibial plateau. 54,56,57 With a conservative estimate of $17,000 in surgical and rehabilitative costs per operation, annual medical costs of ACLRs reach into the billions of dollars. 58 With respect to the ACLD condition, ACLRs tend to partially restore antero-posterior knee stability and menisco-protective role of the ligament and ACLR patients exhibit slightly higher levels of activity, as assessed by Tenger-Score, post injury. 11,13,14,16,24,27,29,36,59,60 However, compared to the natural intact condition, ACLRs alter knee kinematics, knee kinetics, knee laxity, and relative ligamentous contributions to knee restraint during activities of daily living (ADL) and running. 1,20,61,62 These biomechanical differences between the ACL-intact and ACLR condition result in knee degradation that affects quality of life in approximately 75% of athletes who return to sport within 15 years post-surgery. 63,64 Though 85-95% of patients would report an initial restoration of their knee to normal conditions following ACLR, 65 there is evidence to suggest that ACLR does not reduce the long term prognosis of early-onset osteoarthritis relative to the ACLD condition. 36,66 Recent studies of long-term outcomes for ACLRs seem to indicate that between 50-85% of patients experienced at least the onset of osteoarthritis within years of surgery. 36,59,67-69 Rationale and Objectives Robotic simulations of motion have contributed to the understanding of ligamentous contributions during clinical Lachman s and pivot-shift tests and ADLs such as gait (Figure 1.2). 1,70-74 These robotic capabilities have allowed investigators to establish normal biomechanical properties of the intact ACL, which can be extrapolated into design parameters for ACLR grafts. 1,74 However, the vast majority of ACL injuries occur not during controlled 9

18 ADLs, but during athletic tasks that involve rapid deceleration or change in direction and subject the ligament to large mechanical loads. 42,43,45 As most athletes desire to return to sport following injury, it is important to investigate the functional parameters of the natural ACL during athletic tasks in order to optimize Figure 1.2: Frontal (A) and sagittal (B) views of the KUKA KR210 6-degree-of-freedom robotic manipulator that is used to articulated cadaveric joints at the University of Cincinnati biomechanics laboratory. ACLR design parameters for a return to sport without long-term knee degradation. Also, as up to 70% of ACL ruptures occur in noncontact situations, 39,42,43 robotic simulation of athletic tasks that are known to increase loads on the ACL, such as jump-landing and sidestep-cutting, will provide greater insight into how specific mechanical factors contribute to ACL strain. These investigations would ideally be conducted in vivo, but, due to the invasive and damaging methods of mechanical data collection, in vitro simulations of the recorded in vivo condition are necessary. The application of the proposed model to simulate in vivo athletic tasks may lead to novel and efficacious developments in orthopaedic injury prevention and repair. The overall purpose of my dissertation is to apply a novel method of in vitro motion simulation to examine how factors associated with ACL injury risk specifically contribute to knee ligament biomechanics. The improvement of ACL injury screening, injury prevention, and injury reconstruction necessitates an enhanced understanding of the underlying biomechanics associated with athletic tasks. To study ACL function during athletic tasks it is important to 10

19 establish a model that simulates in vivo joint motion on cadaveric specimens in a physiologic manner. A physiologic in vitro model can then be manipulated to directly determine the underlying contributors to increased ligament loads and strain. The resulting database of information can be incorporated into the design parameters of injury screening tools, injury prevention protocols, and injury reconstruction techniques with the intention of improving the efficacy of current methods. The Sports Medicine Biodynamics Laboratory at Cincinnati Children s Hospital has extensive experience in the capture and analysis of three-dimensional (3D), in vivo motion during the performance of athletic tasks. Their investigations have exhibited the ability to identify, diagnose, and address risk factors the lead to ACL injury in athletic populations. 46,75-77 Meanwhile, the biomechanics laboratory at the University of Cincinnati has experience in utilizing a robotic manipulator to simulate 6-DOF physiologic motions on cadaveric knees. Their investigations have demonstrated the ability to convert in vivo gait data into reproducible, kinematically-driven robotic simulations that record 3D joint forces and torques throughout a range of motion. 1,74,78,79 With the combined expertise of these facilities, we seek to implement a novel model to gain a better understanding of the dynamic functions of the ACL during simulated athletic tasks. Specific Aims and Hypotheses Specific Aim 1: Establish a method to simulate dynamic athletic tasks recorded in vivo on knee joint specimens in vitro. 6-DOF robotic manipulators have previously been used to simulate the ADL of gait on cadaveric knees and record the resultant 3D joint forces and torques. 74,79 However, during in vivo motion capture, gait cycles fail to produce the magnitude of loading that is recorded during athletic tasks Therefore, simulated gait cycles may not adequately represent the 11

20 biomechanics of athletic activities where ACL ruptures are more commonly reported. The objective of this study (Chapter 4) was to define techniques used with a 6-DOF robotic manipulator to create reproducible, physiologically-representative, in vitro simulations from in vivo skin-marker based kinematics recorded during athletic tasks. In vivo kinematics were recorded from a representative male and female subject (matched for height, mass, and level of athletic activity) during a drop vertical jump (DVJ) and sidestep cutting tasks using a 10-camera 3D motion capture system (Eagle cameras, Motion Analysis Corp, Santa Clara, CA) and skin-mounted markers. A systematic approach intended to reduce the impact of skin artifact errors was then applied to these recorded kinematics as they were prepared to be used as input to drive the robotic manipulator simulations. Up to the time of this study, no group had developed a method to convert in vivo kinematics recorded with skinmounted markers into physiologically-representative, robotically-driven, cadaveric joint simulations. Likewise, no group had developed a method to simulate athletic tasks on in vitro with respect to in vivo recorded kinematics. Specific Aim 2: Establish a method to normalize in vivo kinematics to individual cadaveric specimens. 6-DOF position-controlled robotic manipulators have been used to evaluate knee ligament biomechanics in cadaveric specimens during dynamic tasks such as gait. 73,74,79 However, biological variability in specimen geometry may artificially impact the magnitude of biomechanical forces reported by these devices. 1 Healthy, unmatched human subjects typically exhibit variation in anthropometric measures. Similarly, unmatched cadaveric specimens should exhibit variability in bony geometry and measures of anatomical size. In order to establish clinically relevant thresholds that identify significant differences, it is important to account for 12

21 the biological variability introduced by a population. For a kinematically-driven model of in vitro joint motion simulations derived from in vivo motion recordings, it is important to assess the potential impact of biologic variability on in vivo kinematics as well as in vitro forces and torques. The objective of this study (Chapter 5) was to examine live subject anthropometrics for significant linear relationships with in vivo kinematics and to examine cadaveric specimen anthropometrics for significant linear relationships with in vitro kinetics. The potential association of anthropometrics with kinematic and kinetic outcomes may suggest the need to incorporate specimen-specific normalization parameters when cadaveric joint simulations are conducted. To determine in vivo associations, 239 high school female athletes were measured anthropometrically and recorded for 3D knee kinematics during DVJ tasks. Peak values for the respective kinematic and kinetic outcome measures were then compared with the anthropometric data and used to determine the statistical significance of linear associations. Hypothesis 2.1: Anthropometric measures would not impact the magnitude of kinematic joint rotations observed between subjects Specific Aim 3: Determine the effect of abduction, internal, and combined tibial rotations on ACL biomechanics at initial contact during robotically simulated athletic tasks. During robotic simulations of walking gait, minor rotational perturbations have been shown to have minimal influence relative to the intra-articular mechanical demand on both the joint and ACL. 79 Conversely, when a passively flexed joint is rotationally perturbed to simulate clinical pivot-shift tests, significant increases in ACL load have been documented (Chapter 3). Similarly, mechanical impulse loading of cadaveric specimens has demonstrated that increased knee abduction and internal tibial rotations at the time of initial contact lead to greater peak ACL 13

22 strain In vivo, only knee abduction has been prospectively linked with ACL failure, as subjects who went on to injury were an average of 8 more perturbed at initial contact than their uninjured counterparts. 46 The purpose this study (Chapter 6) is to quantify how changes in individual and combined rotational DOFs contribute to ACL loading and strain during athletic tasks. The quantification of how rotational perturbations alter ligament mechanics should rank their relative importance to ACL injury risk; and thus, relative importance to prophylactic injury prevention interventions. 17 intact cadaveric lower extremity specimens from 11 unique subjects were positioned into the initial contact orientation for in vivo DVJ and sidestep-cutting. The start point orientation was then adjusted by ±4 abduction rotation, ±4 internal rotation, and ±4 combined abduction and internal rotation while ligament strain and joint forces were recorded. A ±4 rotational shift will create an 8 range for abduction angles at initial contact, which was previously reported as the mean difference observed between athletes who went on to ACL injury and healthy controls. 46 While no such range has been reported for internal rotation, the magnitude of internal rotation adjustments were chosen for consistency of comparison and to correspond with the increased ROM observed during gait following ACL injury. 20 Specimens proved unable to endure the loading conditions rendered from simulated motion tasks that began in the abducted, adducted, or combined rotation starting positions; therefore, rotational stimuli were only applied at the initial contact orientation. Specimens were instrumented with 3 mm differential variable resistance transducers (DVRT) on the ACL and MCL to record the relative ligament strains while joint forces were recorded by a 6-axis force sensor on the robot end effector. Following the completion of these simulations on each intact knee, the specimens were resected down into either an isolated ACL (N = 9) or MCL (N = 8) condition and the protocol was repeated. The 14

23 isolated ligament condition allowed for the identification of relative force/torque contributions throughout motion as well as the determination of the position where zero ligament load occurred for the ACL and MCL. A comparison of relative ligament strain with strain from the zero load position provided absolute strain values for both ligaments throughout each simulated task. Joint force and ligament strain differences were compared in order to determine how each rotational DOF contributed to structural loading. Hypothesis 3.1: Additional abduction will increase ACL and MCL strain relative to the normal condition. Hypothesis 3.2: Additional internal rotation will have minimal impact on ACL strain. Hypothesis 3.3: Combined abduction and internal rotation will increase ligament strain more than either individual rotation. Specific Aim 4: Determine the relative strain response of the ACL and MCL to robotically simulated athletic tasks. Knee abduction torque has been prospectively associated with increased risk of ACL injury, as athletes who went on to experience ACL rupture demonstrated greater peak knee abduction torque during DVJs than healthy controls. 46 As such, knee abduction torques are often credited as one of the underlying mechanisms that lead to ACL failure. Within the knee, the MCL functions as a significant ligamentous resistor to knee abduction. 87 However, despite the correlation of ACL injury with increased knee abduction torque, only approximately 30% of ACL ruptures involve concomitant MCL injury. 88,89 The overall goal of this study (Chapter 7) was to understand how athletic tasks load the knee joint in a manner that drives the ACL to failure without concomitant MCL injury in up to 70% of cases. 15

24 To conduct this study, cadaveric lower extremity specimens (N = 18) were instrumented with DVRTs on the ACL and MCL. A position-controlled 6-DOF robotic manipulator then simulated these specimens through 10 cycles of DVJ and sidestep-cutting tasks derived from in vivo knee kinematics. The DVRTs reported relative stain in both ligaments throughout motion. Following the initial simulations, additional tissue structures were then resected such that only either the ACL (N = 9) or MCL (N = 9) remained intact. Motion was re-simulated in the isolated ligament conditions in order to identify the relative ligament contributions to force resistance and to identify the position of zero ligament load. This allowed for the determination of absolute ligament strain throughout each simulated task as noted in Aim 3. Once robotic simulations were completed, the ACL and MCL were resected and subjected to uniaxial tensioning to failure. Uniaxial testing did not constitute physiologic loading, but was compared to the robotic motion strains in order to estimate the percentage of maximal strain experienced by the ACL and MCL during simulated athletic tasks. Hypothesis 4.1: The ACL will have a greater overall contribution to intact knee forces than the MCL. Hypothesis 4.2: The ACL will express greater relative peak strain during simulated landing and cutting than the MCL. Specific Aim 5: Determine gender-based differences between the robotic simulations of male and female athletic tasks. Gender-based differences in ACL injury rates and the prevalence of injury mechanisms have been well documented. When normalized for exposure, female athletes have demonstrated that they are at 2-6 times greater risk for ACL injury than their male counterparts Also, when performing athletic tasks that involve rapid deceleration or change in direction, female 16

25 athletes exhibit greater knee abduction, a trait that has been prospectively associated with ACL injury risk, than their male counterparts. 46,48,75,76,90-95 These gender-specific outcomes suggest that the underlying mechanics enacted within the knee during athletic tasks may also be genderspecific. The objective of this study (Chapter 8) was to identify how gender-specific, kinematically-driven simulations of like athletic tasks on identical cadaveric specimens lead to mechanical differences within the knee. In order to investigate potential gender-based differences, a 6-DOF robotic manipulator simulated cadaveric lower extremity specimens (N = 19, from 12 unique donors) through kinematics derived from in vivo 3D motion that was captured separately on a matched male and female athlete. Gender-specific cadaveric simulations were derived and performed for DVJ and sidestep-cutting tasks. Simulations were performed in weightbearing conditions that matched the single leg peak force of between times body weight reported during drop landings. 83 A 6- axis load cell mounted on the robot end-effector monitored specimens for knee joint forces and torques, while DVRT recorded strain in the ACL and MCL. The presence of significant differences in force, torque, and strain measures during gender-specific simulations could help identify the underlying mechanics that lead to higher rate of ACL injury in female athletes. Conversely, the absence of significant gender differences may indicate that controlled athletics tasks may not be strenuous enough to enact the mechanical pathways that lead to gender differences in injury rates. Hypothesis 5.1: The female simulations will generate loads associated with higher ACL injury risk than the male simulations. Hypothesis 5.2: The female motion simulations will exhibit greater ACL strain than the male motion simulation. 17

26 Chapter 2 Characteristics of Inpatient Anterior Cruciate Ligament Reconstructions and Concomitant Injuries from Nathaniel A. Bates, a.b.c April McPherson, a,c Marepalli Rao, a,d Gregory D. Myer, a,c,e,f,g Timothy E. Hewett a,b,c,e,h a Department of Biomedical Engineering, University of Cincinnati, Cincinnati, OH, USA b The Sports Health and Performance Institute, OSU Sports Medicine, The Ohio State University, Columbus, OH, USA c Sports Medicine Biodynamics Center, Cincinnati Children s Hospital Medical Center, Cincinnati, OH, USA d Department of Environmental Health-Genomics, University of Cincinnati, Cincinnati, OH, USA e Department of Pediatrics, College of Medicine, University of Cincinnati, Cincinnati, OH, USA f Department Orthopaedic Surgery, College of Medicine, University of Cincinnati, OH, USA g Athletic Training Division, School of Allied Medical Professions, The Ohio State University, Columbus, OH, USA h Departments of Physiology and Cell Biology, Orthopaedic Surgery, Family Medicine and Biomedical Engineering, The Ohio State University, Columbus, OH, USA This manuscript is currently in preparation for Knee Surgery, Sports Traumatology, Arthroscopy 18

27 ABSTRACT Purpose: The purpose of this epidemiologic study was to quantify the incidence, expense, and concomitant injuries for anterior cruciate ligament reconstruction (ACLR) procedures in the United States from 2003 to 2011 that required an inpatient stay. It was hypothesized that the relative reported rates of concomitant knee injuries would be greater with the MCL and menisci compared to all other concomitant knee injuries. Methods: The National Inpatient Sample from was retrospectively sampled using ICD-9-CM codes to identify patients who had ACLRs performed and extrapolated to a national average. Results: Between the years of , an average of 9,037 ± 1,728 inpatient hospitalization included ACLRs, of which 4,252 ± 1,824 were primarily due to the ACLR. Inpatient visits primarily due to ACLR involved an average hospitalization of 1.7 ± 0.2 days and cost $30,118 ± 9,066 per patient. Knee injuries that were commonly reported along with inpatient ACLRs included medial meniscus damage (18.13%), lateral meniscus damage (16.76%), collateral ligament repairs (12.33%), and medial collateral ligament strains (6.91%). Prevalence of meniscus injuries was consistent across years, but MCL related injuries increased over time. Conclusions: ACLR related inpatient hospitalizations account for approximately 7.12% of the total ACLRs performed annually in the United States. Inpatient ACLR procedures continue to decrease in frequency; however, the mean cost per patient increased. Meniscus and collateral ligament injuries were the most commonly reported concomitant knee injuries. INTRODUCTION Injuries to the anterior cruciate ligament (ACL) are highly prevalent, with an estimated 250,000 per year in the United States. 38 In the United States annually, there are 127,000 anterior cruciate ligament reconstructive (ACLR) procedures. 55 The estimated cost of per procedure and 19

28 rehabilitation is $17,000, 58 which results in approximately $2 billion per year spent on ACLRs. ACL injuries commonly occur in an athletic setting during tasks that involve rapid deceleration and/or change of direction. 42,43,45,96 As many as one in fifty female athletes sustain a knee injury per year. 97 ACL ruptures are especially prevalent in high school and college level female athletes as they are four to six times more likely to experience injury than their male counterparts. 42,48,98 The ACL is the primary passive restraint to anterior tibial translation in the knee, as it resists up to 87% of this force. 7 The ACL also serves as a secondary restrain to knee abduction and internal tibial rotations. 1 Athletes with poor neuromuscular control exhibit motion patterns consistent with the kinematics that would directly load ACL. 46,99 These patterns can place abnormal loads on the ACL that lead to rupture as 70% of ACL injuries occur in non-contact scenarios. 42,43 Specifically, increased knee abduction during landing phase of a drop vertical jump has been prospectively identified as a leading predictive risk factor for ACL injury. 46 This abduction load at the knee may lead to abnormal loading in the medial and lateral compartments of the tibiofemoral joint. Accordingly, the MCL, which is the primary passive resistor to knee abduction, 1,7,88 experienced concomitant failure in over 30% of ACL injuries. 88,89 Rupture of the ACL creates instability at the knee that can alter kinematics 61 and tibiofemoral contact. 20,100 The changes lead to abnormal loading of the knee and produce a longterm prognosis of increased risk for osteoarthritis patients. 34,100 To restore joint stability, ACLRs are the most common procedure performed in patients who wish to regain a physically active lifestyle. 61,62 Currently, the gold standard for ACLRs is an arthroscopic outpatient procedure that implants a bone-patellar-tendon-bone or hamstrings tendon autograft. 101 However, ACL ruptures are often accompanied by additional, concomitant, catastrophic knee injuries. 88,89,102 These cases may require multiple surgical procedures and necessitate inpatient hospital admittance. 20

29 Though multiple investigations have tried to quantify the incidence rate, 98 expenditure, 58 and reconstruction incidence 55 for ACL injuries, to our knowledge, the incidence of ACL tears that result in inpatient hospitalization have not been reported. The purpose of this epidemiologic study is to quantify the incidence, expense, and concomitant injuries for ACLR procedures in the United States from 2003 to 2011 that required an inpatient stay. It was hypothesized that the relative reported rates of concomitant knee injuries would be greater with the MCL and menisci compared to all other concomitant knee injuries. METHODS The source of the data utilized in this investigation was the Nationwide Inpatient Sample (NIS), Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality. The NIS was developed in 1988 to provide the largest all-payer inpatient care database in the United States. This database includes an annual survey of inpatient hospital visits accumulated from over 1500 medical centers across 45 US states that provides unidentified, patient-specific data relative to hospital admittance, diagnoses and procedures performed, patient expenditures, and patient demographics. The NIS provides a 20% stratified sample of United States community hospitals that can be extrapolated into national estimates. The present study analyzed the NIS database to determine the incidence of inpatient ACLR procedures from 2003 to The NIS database was analyzed on a patient-by-patient basis for the eight years included in this study. Patients were included in this study if their NIS data exhibited International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) or procedure codes were relevant to ACL injury. Specifically, patients were included in the present study if their NIS data exhibited the diagnosis code for an old disruption of anterior cruciate ligament 21

30 (717.83) or a sprain of cruciate ligament of the knee (844.2), or the procedure code for ACLR (81.45). All data was retrieved and analyzed using MATLAB code (version 2012a, The MathWorks, Inc., MA). Once patients with ACLR diagnoses and procedures were isolated, the remaining patient population was statistically evaluated using custom MATLAB code. Annual means and standard deviations were calculated for all continuous variables (age, length of stay, number of procedures, and total charge). Relative frequencies were calculated for all categorical variables (gender, and race). Additionally, a list of all additional ICD-9-CM and procedure codes that corresponded with ACL patients was compiled to evaluate injuries and operations that are frequently concomitant with ACLR. Due to the comprehensive nature of the coding database and multitude of corresponding codes reported for inpatient stays, additional codes were only incorporated if they occurred in greater than 1.00% of ACLR patients for a given year. Though the NIS provides a comprehensive database, it is not inclusive of all medical center data in the United States. Annually the NIS reports on approximately 8 million inpatient hospitalizations from a diversity of medical centers. HCUP estimates that approximately 40 million inpatient hospitalizations occur each year in the United States. Software available on the HCUP website ( was used to extrapolate national averages for the entire inpatient ACLR population based on the stratified NIS database. HCUP software was also utilized to estimate national averages for the subset of patients whose primary reason for inpatient hospitalization was ACLR. Analysis of variance (ANOVA) and Student s T-tests were used to determine differences between incidence percentages in categorical variables. All statistics were calculated in MATLAB using built-in functions. 22

31 RESULTS Annually, the NIS database reported an average of 1,344 ± 374 ACLR procedures that required inpatient hospital admittance, with 1.6 ± 1.4 annual deaths and an average cost, as calculated from the annual means, of $45,040 per patient (Table 2.1). With the exception of 2010, the annual number of inpatient ACLR procedures documented by the NIS declined between each year from However, average expenditures per patient increased annually across all years. NIS patients had a mean hospital admittance of 3.98 days, with an average of 2.95 procedures were performed. From the average number of procedures per patient increased by 0.72 and the average hospitalization increased by 2.73 days. Mean patient age, as calculated from annual NIS means, was 35.8 years. Inpatient ACLR procedures were more common among males (61.25 ± 1.69%) than females (38.75 ± 1.69%; P < 0.01). The breakdown of injuries by ethnicity is also displayed in Table 2.1. Extrapolation of the NIS data to a national average indicated an average of 9,037 ± 1,728 inpatient ACLR procedures per year (Table 2.2), which accounts for approximately 7.12 ± 1.36% of the estimated 127,000 ACLR performed annually in the United States. 55 The extrapolated national average was also gender specific (males = ± 1.11%, females = ± 1.17%, P < 0.01). ACLRs were significantly more prevalent between the ages of than in any other age group (54.70 ± 5.90%, P < 0.01). 23

32 Table 2.1: Patient demographics calculated directly from the NIS data subset for all reported inpatient ACLR cases from Procedures Deaths Age ± ± ± ± ± ± ± ± ± Length of Stay 2.61 ± ± ± ± ± ± ± ± ± 6.95 Operations per Patient 2.57 ± ± ± ± ± ± ± ± ± 3.01 Cost per Patient $27266 ± $25629 ± $32373 ± $36394 ± $38044 ± $44897 ± $55942 ± $72259 ± $72559 ± Gender Male Female 1197 (62.09%) 731 (37.91%) 1062 (59.66%) 718 (40.34%) 844 (60.50%) 551 (39.50%) Race Caucasian (74.22%) (70.81%) (69.91%) African (11.23%) (11.05%) (11.17%) Hispanic (9.65%) (12.27%) (10.98%) Asian (2.21%) (1.45%) (1.34%) Other (2.69%) (4.42%) (6.59%) *Length of stay is reported in days hospitalized. Age is reported in years. 767 (62.71%) 456 (37.29%) 619 (69.71%) 99 (11.15%) 123 (13.85%) 8 (0.90%) 39 (4.39%) 769 (63.76%) 437 (36.24%) 578 (66.21%) 121 (13.86%) 132 (15.12%) 9 (1.03%) 33 (3.78%) 709 (58.40%) 505 (41.60%) 623 (68.29%) 119 (13.16%) 101 (11.17%) 16 (1.77%) 45 (4.98%) 633 (62.18%) 385 (37.82%) 576 (67.05%) 125 (14.55%) 104 (12.11%) 19 (2.21%) 35 (4.08%) 657 (60.05%) 437 (39.95%) 581 (62.47%) 159 (17.10%) 119 (12.80%) 14 (1.51%) 57 (6.13%) 569 (61.92%) 350 (38.08%) 530 (64.48%) 124 (15.09%) 116 (14.11%) 17 (2.07%) 35 (4.25%) 24

33 Table 2.2: National averages extrapolated from the NIS for all inpatient hospitalizations that involved ACLRs from Procedures 12,051 11,486 9,638 8,640 8,404 8,560 7,761 7,972 6,827 Age Gender Male Female 1,731 (14.36%) 7,693 (63.84%) 2,019 (16.76%) 459 (3.81%) 82 (0.68%) 7,353 (61.02%) 4,499 (37.34%) 1,598 (13.91%) 7,159 (62.33%) 2,157 (18.78%) 430 (3.74%) 84 (0.73%) 6,748 (58.75%) 4,502 (39.20%) 1,651 (17.13%) 4,987 (51.74%) 2,192 (22.74%) 582 (6.04%) 150 (1.55%) 5,625 (58.36%) 3,793 (39.35%) 1,075 (12.44%) 4,882 (56.50%) 1,919 (22.21%) 686 (7.94%) 74 (0.85%) 5,280 (61.11%) 3,294 (38.12%) 946 (11.25%) 4,885 (58.12%) 1,803 (21.45%) 595 (7.08%) 143 (1.70%) 5,124 (60.97%) 3,189 (37.94%) 967 (11.30%) 4,474 (52.27%) 2,270 (26.51%) 689 (8.05%) 142 (1.66%) 5,061 (59.12%) 3,4,5,6 (40.38%) 799 (10.29%) 4,059 (52.30%) 1,968 (25.35%) 651 (8.39%) 174 (2.24%) 4,710 (60.69%) 2,928 (37.73%) 897 (11.25%) 3,863 (48.45%) 2,280 (28.60%) 698 (8.76%) 212 (2.66%) 4,706 (59.04%) 3,242 (40.67%) 636 (9.31%) 3,194 (46.78%) 2,192 (32.11%) 626 (9.17%) 149 (2.18%) 4,129 (60.48%) 2,632 (38.55%) 25

34 Extrapolation of the NIS data into a national average for patients whose primary cause hospitalization was ACLR indicated a mean of 4,252 ± 1,824 annual procedures at a cost of $30,118 ± 9,066 per patient (Table 2.3). Annually, inpatient ACLRs where the ACLR is the primary reason for hospitalization generated a mean national expense of $115,631,936 ± 26,288,717 and accounted for approximately 3.35 ± 1.44% of all ACLR procedures in the United States. 55 The incidence of patients primarily admitted for ACLRs decreased by 5,523 patients from whereas the cost per patient increased by $20,675 (Figure 2.1). Annual aggregate national patient charges for inpatient hospitalization primarily due to ACLR fluctuated between years, but decreased overall by $90,134,894 from Patients admitted primarily due to ACLR had an average length of stay of 1.7 ± 0.2 days, which was consistent between years. This population subset again demonstrated a male gender bias (58.09 ± 2.27%, annual range = %, P < 0.01) and was again more prevalent in the age range than any other group (61.00 ± 5.58%, annual range = %, P < 0.01). 26

35 Figure 2.1: Incidence of inpatient ACLR and charges per patient across time. The change in per patient cost (dotted line) is inverse to the changes in incidence for the whole inpatient ACLR population (solid line) as well as for ACLRs that were the primary cause of hospitalization (dashed line). Inpatient hospitalization due primarily to ACLR accounted for ± 11.59% of the whole inpatient ACLR population, but decreased from 62.05% in 2003 to 28.64% in Inpatient hospitalizations involving, but not primarily due to ACLRs, consistently accounted for approximately 5000 cases annually. 27

36 Table 2.3: Estimated national averages extrapolated from the NIS database for patients whose primary cause for inpatient hospital admittance was ACL injury or ACLR Procedures 7,478 6,632 5,018 4,062 3,893 3,516 3,067 2,646 1,955 Length of Stay Cost per Patient $23,501 $20,080 $21,392 $24,684 $25,465 $32,408 $39,642 $39,716 $44,176 National Charges Age Gender Male Female $175,506,012 $133,250,234 $107,457,208 $100,229,082 $99,437,564 $113,946,528 $120,865,906 $104,623,773 $85,371,118 1,285 (17.19%) 5,164 (69.05%) 864 (11.56%) 79 (17.19%) 0 (0.00%) 4,417 (59.07%) 2,885 (38.58%) 1,211 (18.26%) 4,379 (66.02%) 878 (13.24%) 92 (1.39%) 0 (0.00%) 3,716 (56.03%) 2,689 (40.54%) 1,298 (25.86%) 2,666 (53.12%) 836 (16.66%) 126 (2.52%) 0 (0.00%) 2,771 (55.22%) 2,048 (40.80%) *Length of stay is reported in days hospitalized. Age is reported in years. 731 (17.99%) 2,607 (64.17%) 599 (14.75%) 116 (2.85%) 0 (0.00%) 2,437 (59.98%) 1,586 (39.05%) 717 (18.42%) 2,531 (65.00%) 508 (13.06%) 97 (2.49%) 0 (0.00%) 2,405 (61.76%) 1,424 (36.57%) 605 (17.21%) 2,181 (62.03%) 634 (18.04%) 53 (1.50%) 0 (0.00%) 1,995 (56.75%) 1,483 (42.17%) 527 (17.19%) 1,812 (59.08%) 477 (15.56%) 129 (4.22%) 0 (0.00%) 1,736 (56.61%) 1,224 (39.90%) 603 (22.81%) 1,446 (54.64%) 460 (17.40%) 78 (2.94%) 0 (0.00%) 1,599 (60.45%) 1,022 (38.65%) 375 (19.20%) 1,093 (55.90%) 364 (18.64%) 77 (3.97%) 0 (0.00%) 1,113 (56.95%) 781 (39.96%) 28

37 According to the collected NIS subset, approximately ± 3.66% of ACLR inpatient admittance involved primary ACL ruptures, while ± 3.52% identified old disruptions of the ACL. The most commonly reported concomitant injury was close fracture of C1-C4 with unspecified spinal cord injury, as it appeared in ± 5.77% of cases (Table 2.4). The most commonly reported concomitant knee-related injury was a tear of the medial cartilage or meniscus of the knee (18.13 ± 1.47% of cases) followed by tear of lateral cartilage or meniscus of knee (16.76 ± 1.05% of cases). The difference in the prevalence of meniscus injuries between the medial and lateral sides was statistically significant (P = 0.04). Collateral ligament repairs were less frequent (P < 0.01), documented in ± 2.32% of cases. However, when combined with sprain of medial collateral ligament of the knee, collateral ligament damage significantly became the most commonly reported concomitant injury with inpatient ACLR procedures (19.24 ± 6.69% of cases), though there was no statistical difference compared to the incidence of medial meniscus injuries (P = 0.50). Additional unspecified knee repairs and knee arthroscopy were performed in ± 0.72% and ± 4.84% of ACLR cases, respectively. The incidence of meniscus injuries was consistent between years for both the medial (range %) and lateral (range %) sides (Figure 2.2). The combined incidence of collateral ligament repairs and strains increased by 18.43% between 2003 and Frequency of additional knee repairs was consistent between years (range %) while additional knee arthroscopies decreased by 16.77% from

38 Table 2.4: Annual percentages of ACL classifications and concomitant injuries documented within the NIS inpatient ACLR population. ICD-9 CM Description Sprain cruciate ligament of knee 61.39% 60.56% 65.16% 66.59% 61.73% 65.90% 70.52% 68.61% 69.43% Cruciate leg repair 88.64% 86.93% 79.65% 75.59% 73.83% 71.87% 69.46% 59.14% 55.11% Old disruption of anterior cruciate ligament 38.82% 39.61% 35.05% 33.58% 38.10% 34.34% 29.77% 31.67% 31.32% 806 Closed fracture of C1-C4 level with unspecified spinal cord injury 36.47% 37.64% 32.54% 31.06% 28.78% 26.57% 27.65% 21.20% 21.85% Unspecified essential hypertension 8.30% 7.93% 12.06% 14.60% 14.39% 17.66% 18.98% 20.47% 19.38% Tear of medial cartilage or meniscus of knee 20.58% 17.61% 19.09% 19.06% 15.94% 17.83% 18.40% 16.11% 18.51% Nondependent tobacco use disorder 7.44% 7.39% 8.43% 10.38% 14.39% 13.57% 13.68% 19.29% 17.01% Tear lateral cartilage or meniscus of knee 16.20% 16.74% 16.93% 16.79% 15.54% 17.66% 18.98% 15.83% 16.14% 8146 Collateral ligament repair 8.05% 11.05% 11.08% 13.30% 10.47% 13.57% 14.93% 13.47% 15.07% 8147 Other repair of the knee 14.82% 15.15% 13.31% 13.63% 14.64% 14.64% 15.61% 14.92% 14.32% 8026 Knee arthroscopy 29.04% 17.67% 15.54% 18.00% 16.11% 17.66% 20.13% 13.65% 12.27% Sprain of medial collateral ligament of knee >1% 8.10% 8.50% 10.79% >1% >1% 9.63% 11.74% 11.95% Nonspecific (abnormal) findings on radiological and other examination of abdominal area, including >1% >1% 8.22% 8.43% >1% >1% 10.69% 12.37% >1% retroperitoneum Closed fracture of upper end of tibia alone >1% >1% >1% 8.27% >1% >1% 10.60% >1% >1% Chondromalacia of patella 6.67% 7.17% 9.97% 8.76% >1% >1% >1% >1% >1% 30

39 Figure 2.2: Depiction of trends in the prevalence of additional knee procedures frequently documented with inpatient ACLRs. Prevalence is relative to inpatient ACLRs recorded in the NIS database. From the percentage of collateral ligament injuries increased in prevalence, whereas meniscus injuries and other repairs remained constant, and knee arthroscopies decreased. DISCUSSION The purpose of this epidemiologic study was to quantify the incidence, expense, and concomitant injuries for ACLR procedures in the United States from 2003 to 2011 that required an inpatient stay. It was found that inpatient ACLRs comprise a minority of ACLR procedures as they account for only approximately 7% of procedures performed annually. However, as conservative estimates place surgical costs for outpatient, autograft ACLR procedures at $4872, 103 as evidenced in the current dataset inpatient ACLRs are significantly more expensive. Between the years of , the average per patient charge of an inpatient stay that was primarily due to an ACLR was 618% greater than outpatient ACLRs. Previous studies have indicated that inpatient ACLRs have and average expense three times greater than the equivalent outpatient procedures. 104 The increased cost ratio in the present study is likely due to the 31

40 combination of multiple procedures that accompany the majority of present-day inpatient ACLR treatments as well as rising costs of hospitalization, bed space, nursing, and more. As hypothesized, the knee structures that were most commonly disrupted concomitantly with the ACL were the meniscus and MCL (which were both classified with multiple ICD-9 codes). 70% of sports-related ACL injuries occur in non-contact scenarios during rapid deceleration or change of direction movements. 42,43 Athletes with poor neuromuscular control have exhibited increased knee abduction during these athletic tasks. 75,99 Knee abduction has been shown to increase strain on the ACL 105,106 and has been prospectively associated with increased ACL injury risk. 46 These mechanics correlate well with the concomitant injuries recorded in the present study as the MCL is the primary ligamentous restraint to knee abduction rotation. 1,7,88 However, it is interesting to note that the rate of concomitant MCL injury for inpatient ACLRs reported in the present study was lower than MCL incidence reported for the overall population. 88,89,96 Similarly, meniscus injuries were less frequent in the present study than previously reported ACL-injury cohorts. 96 The current database identified concomitant injuries through ICD-9 codes. These codes indicate procedures performed; and therefore, track repairs. In many cases, the MCL can go unrepaired after injury and may not generate an ICD-9 code in these instances. As such it is possible that the incidence of concomitant MCL injuries presented in the NIS was lower than the actual rate of occurrence in the inpatient ACLR population. Prior to the mid 1990s, ACLR was an inpatient procedure with an average hospitalization of 2-3 days; however, surgical advancements allowed ACLRs to become less invasive and they have since shifted to outpatient operations. 104,107 In the present study, the incidence of inpatient ACLRs for all three cohorts consistently decreased between years from This was mostly due to a decrease in the number of inpatient hospitalizations due primarily to ACLR, as 32

41 the annual incidence of inpatient stays involving, but not primarily due to, ACLR was constantly between cases. Outpatient ACLRs have lead to reduced cost and enhanced patient satisfaction. 104 In 1995, inpatient ACLRs averaged approximately 300% of the cost of comparable outpatient procedures. As inpatient ACLRs become less frequent, this cost disparity has increased annually. That cost increase correlated with annual increases in number of procedures and length of hospitalization that were observed in the whole NIS cohort. These trends would seem indicate that the increasing cost of inpatient ACLRs may not be due to changing procedural costs, but due to increased severity in injuries that require inpatient hospitalization. However, in the cohort admitted primarily for ACLR, the length of hospitalization remained unchanged overtime, which implicates that rising costs in healthcare were also a driving force between the constantly increasing per patient charges. The significant presence of concomitant closed fractures of C1-C4 level with unspecified spinal cord injury in the current sample likely indicates that many inpatient ACLRs result from trauma rather than sports injuries. In athletic populations over 94% of ACL injuries occur as a result of sports injuries, but in non-athletic populations this drops to 75% as injuries from traffic and daily living accidents are more common. 96 Therefore, it is likely that a significant number of injuries documented in the present study occurred in traumatic events such as car accidents where large forces are applied across multiple areas of a patient s body and resulted in the need for multiple surgical procedures. The presented data would seem to support this theory as the mean charge and length of stay per patient for the whole NIS cohort were $14,161 and 1.03 days greater than the subset of patients admitted primarily for ACLR. Also, the subset admitted primarily for ACLR had significantly lower variability in charge and length of stay, which indicated greater consistency in treatment than was present in the whole NIS cohort. 33

42 Unfortunately, one of the limitations of the NIS dataset is that injury cause is not documented; and therefore, could not be reported. The ethnic distribution of ACLRs in the present study was representative of the national diversity reported by the United States Census Bureau for 2012 (Caucasian = 63.0%, Hispanic = 16.9%, African = 13.1%, Asian = 5.1%; The gender distribution reported by all three cohorts examined in the present study also compared favorably with previous data on knee injuries where males accounted for greater than 60% and females accounted for less than 40% of the population. 96,102 Therefore, inpatient ACLRs do not exhibit ethnicity bias and maintain the same gender bias as seen in the overall ACL-injury population. Similarly, the age bias seen in previous ACL-injured populations was maintained in the present study as over 50% of inpatient ACLRs were performed year old patients. 102 These statistics indicate that the cohort of inpatient ACLRs likely bears significant resemblance to the national ACL-injury patient population. A limitation of the NIS database is that it does not document the mechanism or cause of injury. Therefore, in the present study, it was not possible to definitively distinguish sportsinduced injuries from those caused by alternative sources of trauma such as traffic accidents. Similarly, the NIS does not itemize expenditures. As such it was not possible to partition out costs directly related to ACLR relative to those incurred from concomitant injuries. This inability to itemize expenses is likely what lead to the large standard deviations in annual treatment costs in the NIS database (Table 2.1). The authors acknowledge that a patient with concomitant spinal fractures, collateral ligament damage, and meniscus damage would most likely incur greater costs than an isolated ACLR. However, when isolated to inpatient stays where ACLR was the primary reason for hospitalization, the standard deviation for per patient was significantly 34

43 reduced. Therefore, cases where ACLR was the primary cause of hospitalization were more likely to be representative of sports related ACL-injuries. Although they have, and continue to, decrease in frequency, inpatient ACLRs still represent approximately 10% of the ACLRs performed annually in the United States. Inpatient operations represent significant financial burden and typically well exceed the reported costs of an average ACLR. The per-patient cost for inpatient ACLRs is rising, which may reflect inflation as well as improved surgical methods that necessitate hospitalization for only the most severe injuries. Demographics indicate that the inpatient cohort is likely representative of the overall ACLR population. However, relative to previously published ACLR cohorts, inpatient ACLRs exhibit differences in concomitant injury patterns. ACKNOWLEDGEMENTS This work was supported by NIH grants R01-AR049735, R01-AR055563, and R01- AR CONFLICT OF INTEREST There were no conflicts of interest in the preparation of this manuscript. 35

44 Chapter 3 Anterior Cruciate Ligament Biomechanics During Robotic and Mechanical Simulations of Physiologic and Clinical Motion Tasks: A Systematic Review and Meta-Analysis Nathaniel A. Bates a,b,c, Gregory D. Myer c,d,e,f, Jason T. Shearn a, Timothy E. Hewett a,b,c,d,g a Department of Biomedical Engineering, University of Cincinnati, Cincinnati, OH, USA b The Sports Health and Performance Institute, OSU Sports Medicine, The Ohio State University, Columbus, OH, USA c Sports Medicine Biodynamics Center, Cincinnati Children s Hospital Medical Center, Cincinnati, OH, USA d Department of Pediatrics, College of Medicine, University of Cincinnati, Cincinnati, OH, USA e Department Orthopaedic Surgery, College of Medicine, University of Cincinnati, OH, USA f Athletic Training Division, School of Allied Medical Professions, The Ohio State University, Columbus, OH, USA g Departments of Physiology and Cell Biology, Orthopaedic Surgery, Family Medicine and Biomedical Engineering, The Ohio State University, Columbus, OH, USA This manuscript is currently in preparation for submission to Journal of Biomechanics. 36

45 ABSTRACT In vitro joint simulations invasively study the biomechanical behaviors of the anterior cruciate ligament (ACL). These simulations aim to replicate physiologic conditions, but use multiple mechanisms to drive in vitro motions that may influence biomechanical outcomes. The objective of this review was to examine, summarize, and compare biomechanical evidence related to ACL function from in vitro simulations of knee motion. A systematic review was conducted (2004 to 2013) in Scopus, PubMed/Medline, and SPORTDiscus to identify peer-reviewed studies that reported kinematic and kinetic outcomes from in vitro simulations of physiologic or clinical tasks at the knee. Inclusion criteria for relevant studies were articles published in English that reported on whole-ligament ACL mechanics during the in vitro simulation of physiologic or clinical motions on cadaveric knees that were unaltered outside of the ACL-intact, deficient, and reconstructed conditions. A meta-analysis was performed to synthesize biomechanical differences between the ACL-intact and reconstructed conditions. 77 studies met our inclusion/exclusion criteria and were reviewed. Combined joint rotations have the greatest impact on ACL loads, but the magnitude by which individual kinematic degrees of freedom contribute to ligament loading during in vitro simulations is technique-dependent. Biomechanical data collected in prospective, longitudinal studies corresponds better with robotic-manipulator simulations than mechanical-impact simulations. Robotic simulation indicated that the ability to restore intact ACL mechanics with ACLRs was dependent on loading condition and degree of freedom examined. INTRODUCTION Worldwide it is estimated that over 2 million anterior cruciate ligament (ACL) injuries occur annually. 108 These injuries are devastating to athletic careers and expensive to repair and 37

46 rehabilitate, as conservative estimates place the cost of an ACL reconstruction (ACLR) surgery at $17,000 plus rehabilitation. 58 These surgeries exhibit short-term promise in the restoration of knee function as up to 86% percent of ACLR patients exhibit a negative pivot-shift score three years post-operative. 109 However, long-term outcomes are less desirable as up to 90% of ACLR patients continue to develop early onset osteoarthritis and knee degeneration within 20 years post-surgery. 66 In order to optimize preventative and reparative strategies for injured ACLs, it is essential to establish the underlying mechanics that contribute to excessive ligament loads and lead to failure. Approximately 65% of ACL ruptures occur in noncontact situations, which indicate that the injuries are likely influenced by poor neuromuscular control and mechanics, rather than an external impact force delivered directly to the knee joint. 110 Therefore, prophylactic training protocols are effective in the enhancement of neuromuscular control and reduction of the incidence of ACL injuries. 111 In order to design effective training protocols, the biomechanical contributors to ACL forces and strain must be identified. An expanse of in vivo research has been directed at the mechanisms associated with ACL failure and has identified that factors such as excessive knee valgus, asymmetry, and poor trunk position are associated with increased injury risk. 45,112,113 Despite their contributions, in vivo studies are limited in that direct, invasive measurements of ACL mechanics are unethical to perform on living subjects and the presence of sensors would interrupt native function. Unlike in vivo investigations, in vitro studies are able to use invasive techniques that directly evaluate ACL mechanics and responses to loads and stresses. In vitro studies have revealed the relative contributions of the ACL to anterior tibial force (ATF) force, 7 resistance to internal tibial torsion (ITT), 9 and muscular contributions to ACL strain. 114 Though valuable, 38

47 many of these in vitro investigations examined maximal, uniaxial loading, rather than complex multi-planar scenarios that are likely more physiologic, in vivo conditions. Functional tissue engineering principles indicate that the evaluation of ligament biomechanics within functional movement ranges that mimic in vivo activity will provide greater clinical relevance than the information obtained from non-physiologic testing methodologies. 115 Over the past 20 years, investigators have focused on in vitro approaches with methodologies designed to simulate in vivo loading conditions from motions of daily activities or clinical settings. 73,78, There presently exists fundamental differences amongst these in vitro methodologies as some protocols drive motion with robotic manipulators that constantly apply forces and actively control limb position, while other protocols drive motion with a singular impulse force and allow restraints to passively regulate limb position. Though all in vitro methods aim to correlate with in vivo physiologic conditions, differences in the mechanisms used to drive motion simulations could lead to disparities in biomechanical outcomes. The objective of this systematic review and meta-analysis was to synthesize the current data and compare robotic and mechanical methods of in vitro knee simulation. Specifically, we aimed to investigate the functional behavior of the ACL and ACLR and to analyze differences observed between methodologies. It was hypothesized that robotically-driven and mechanicalimpact knee simulations would demonstrate differential responses to rotational motions. It was further hypothesized that reconstruction of the ACL will restore anterior tibial translation (ATT) and ATF observed by the intact ACL, but will fail to restore kinetic and kinematic response in the other rotations (frontal and transverse) during simulated motions. METHODS A literature search related to methods of knee simulation was performed in the PubMed/MEDLINE, SPORTDiscus, and Scopus databases in May The systematic review 39

48 focus was to identify research articles published within the last decade ( ) that investigated in vitro ACL biomechanics through knee motion simulation. Search terms were limited to anterior cruciate ligament OR ACL and was further limited with robot, robotic, knee simulator, OR knee simulation. Additional articles were added through crossreferencing the identified studies. As this review focused on functional biomechanics, simulations were limited to physiologic (passive flexion, gait, and jump landing) or clinical (Lachman s and pivot shift test) knee motions. Non-physiologic simulations, such as uniaxial force or torque loading to joint failure, were excluded. Knee conditions included in this review were ACL-intact, ACLD, ACLR, and ACL-only. Inclusion was also limited to whole-ligament biomechanics; thus, any studies that investigated specimens with arthroplasty or individual bundle mechanics were excluded. In order to focus the review to ACL Figure 3.1: Flow chart of the systematic review process. biomechanical contributions 40

49 in a normal knee, data collected after the selective alteration of additional passive restraint structures within the knee (including but not limited to tibial osteotomy, posterior cruciate ligament resection, or meniscus resection) were excluded. In vivo simulations, simulations on joints other than the knee, computer models, computational models, papers without kinematic or kinetic dependent variables, methodology papers, review papers, and non-english articles were also excluded. The initial search compiled 621 published papers, which were then reduced to 77 papers by as documented in Figure 3.1. The included papers were divided into 3 classifications of robotic simulation (passive flexion, weight-bearing flexion, and kinematic reproduction) and one classification of mechanical-impact simulation. Methods of Robotic Simulation One method of robotic simulation has used a highly accurate and precise six-degree-offreedom (6-DOF) robotic manipulator in conjunction with a universal force sensor (UFS) to articulate a specimen through passive flexion with minimal loading. 117,118 Specimens were resected of soft tissue outside the knee joint, cemented into rigid fixtures, and affixed to the robotic end effector (tibia) and a static frame (femur). Local coordinate frames were identified by anatomical landmarks and were digitized relative to the robot s global position, which allowed for tibial articulation relative to the femur. Flexion was recreated in 1.0 increments while the robotic/ufs zeroed loads at each position. Simulations of clinical Lachman s test, through ATF, and pivot-shift tests, through combined abduction and internal torque, were executed at predetermined intervals. The robot recorded the initial path of passive flexion and was able to reproduce it with high precision. Thus, the same motion was applied to the same specimen for the ACL-intact, ACL-deficient (ACLD), and ACLR conditions, which allowed superposition to determine relative force contributions. 119 Since its inception at the University of Pittsburgh, this 41

50 methodology has been adopted at multiple research facilities including Harvard Medical Center, Wilhelms University, Kogakuin University, the Hospital for Special Surgery, and the United States Naval Academy. 10,27, An upright knee simulator (UKS) was developed at the University of Tubingen to simulate knee motion for weight-bearing flexion conditions. 116 Similar to the passive flexion methodology, this device simulated flexion at the knee, provided external tibial loads with a robotic/ufs, and used superposition to calculate ACL forces. Unlike passive flexion, the weightbearing flexion path was not controlled by the robotic/ufs. Rather, the proximal end of the potted femur was attached to a hip assembly with a vertically-oriented linear actuator and two rotational DOFs, while the potted tibia was attached to a vertically static ankle assembly that allowed for three rotational DOFs. Starting at 15º, the hip assembly actuator drove each specimen into knee flexion at a rate of 1º /second until 90º of flexion was reached. Linear actuators, attached to the quadriceps and hamstrings tendons via tension wires, simulated muscle forces that created constant weight-bearing forces of between N at the ankle. Similarly, the University of Waterloo developed a weight-bearing flexion simulator based on the vertical motion of a hip assembly and in vivo muscle forces. 120 This dynamic knee simulator (DKS) lacked a robotic/ufs and simulated flexion motion via regulated descent of its hip assembly, while simulated muscle forces dictated the remaining DOFs. Unlike the previous methods of robotic knee simulation, which simulated flexion motions relative to the geometry of each specimen, the University of Calgary developed a method of reproducing in vivo recorded kinematics. 73 Rigid markers directly implanted on bony structures recorded in vivo kinematics from ovine treadmill gait. The recorded limb was sacrificed, resected of soft tissues outside the joint capsule, and potted into a 6-DOF parallel robot. The limb was 42

51 digitized relative to the robot using global, tibial, and femoral coordinate frames. The in vivo gait kinematics were then used as positional input for the robot to articulate the tibia about the femoral coordinate system. This method was adapted to a serial robotic/ufs manipulator by the University of Cincinnati. 78 Further adaptation applied the mean ovine motion, as well as mean in vivo gait kinematics recorded from human subjects, onto cadaveric specimens. 74 Superposition again allowed investigators to quantify the biomechanical contributions of the ACL during kinematic-derived simulations. Mechanical-impact simulations of knee motion have been based on driving forces equivalent in magnitude to in vivo ground reaction forces through in vitro specimens and allowing each specimen to determine its own path of articulation. Investigators designed the jump-landing simulator to deliver a drop-weight impact force through the tibia of a lower extremity specimen resected of all soft tissues outside the knee joint capsule. 121 The proximal end of the femur and distal end of the tibia were potted into assemblies that simulated the hip and ankle joints as sagittal plane hinges. Limbs were flexed 10-40º prior to impact. Actuator cables were drilled into the approximate quadriceps and hamstrings insertion sites on the patella and tibia and simulated muscle forces that stabilized the joint during impact. Impact forces were designed to fit within the magnitude (2-4 * bodyweight) and rise time (~0.1 sec) of in vivo ground reaction forces during landing. Similarly, Withrow and colleagues developed the knee testing apparatus to simulate jump landings. 122 Specimens were resected of soft tissue down to the joint capsule and muscle tendons then potted at the proximal femur and distal tibia. These fixtures were adjustable in order to manipulate the rotational alignment of the knee in both the sagittal (15-25º flexion) and frontal planes (0-15º abduction/adduction) prior to impact testing. The tibial assembly was locked to prevent translational movement during impact. Cables were 43

52 connected between the quadriceps, hamstrings, and gastrocnemius tendons and independent tensioning mechanisms to mimic in vivo pre-landing muscle activations. Specimens were oriented vertically and impact was applied to the proximal end of the femur. Later iterations of the knee testing apparatus incorporated a torsional transformer that converted some of the vertical impact force into rotational torque, which simulated pivot landings. 12 Meta-analysis The passive flexion method of robotic simulation was selected for meta-analysis due to its prevalence and congruity between studies. To reduce confounding factors, force applications were limited to 134 N ATF in the simulated Lachman s test and 10 N*m abduction torque combined with 4-5 N*m internal rotation torque in the simulated pivot-shift test. Data and standard deviations at predetermined intervals (0, 15, 30, 45, 60, 90, 120 of flexion) were digitized and an average, weighted relative to the number of specimens in each qualified study, was determined. Standard deviations were used to calculate corresponding standard error of the means. This was repeated for the ACL-intact, ACLD, and ACLR conditions and the results were plotted (Figure 3.2 & 3.3). Two-sample t-tests (α = 0.05) determined the presence of statistical differences between each condition at each interval. ATT, internal tibial rotation (ITR), and ligament forces were tracked due to their consistently reported outcomes in passive flexion simulations. 29 studies were included in the Lachman s assessment and 25 studies were included in the pivot-shift assessment. This method of analysis was then adapted to assess differences in abduction magnitude during pivot-shift tests (N = 18 studies; Figure 3.4), simulated muscle forces during Lachman s tests (N = 5 studies; Figure 3.5), and ACL condition under simulated quadriceps force during Lachman s tests (N = 5 studies; Figure 3.6). 44

53 RESULTS Passive Flexion The ensemble mean mechanical responses of the ACL-intact, ACLD, and ACLR knee to robotically simulated Lachman s and pivot-shift tests throughout a range of passive flexion at constant mechanical loads are summarized in Figure 3.2 and ,13,14,16,21-29, In response to 134 N ATF throughout flexion, ACLRs reduced mean ATT (peak 9.7 mm) relative to ACLD (peak 18.2 mm), but failed to match intact ACLs (peak 8.1 mm). However, under combined torsional loading, the mean ATT for ACLRs only differed from intact ACLs at 15 flexion (5.9 vs. 4.6 mm). During ATF, mean ITR was restricted throughout flexion in ACLRs (peak 2.9 ) compared to intact ACLs (peak 5.8 ). Mean ITR between ACLRs and intact ACLs was only statistically different at 0 flexion (12.2 vs. 9.4 ) under combined torsional loads. No statistical differences were observed in ITR between ACL-intact (35 and 40 ), ACLD (39 and 41 ), and ACLR (38 and 41 ) conditions under 4 N*m of isolated ITT at 30 and 60 of knee flexion. 141 ACLRs restored the mean ligament force seen by the intact ACL at 0 flexion for both ATF (102 vs. 98 N) and combined torsional loads (82 vs. 76 N). However, at knee flexion, the mean ligament force for ACLRs exceeded magnitudes seen by the intact ACL. Isolated abduction/adduction torque of 10 N*m produced greater loads on intact ACLs (peak 40.9 N) than isolated internal/external torques of 4-5 N*m (peak 33.9 N) between 0 and 90 flexion. 10,142 A 3 N*m increase in abduction torque during pivot-shift loading significantly enhanced ensemble mean ATT by 1.6 mm at 15 knee flexion, 2.8 mm at 30, and 6.1 mm at 45 and ensemble mean in situ ACL force by 19 N at 0 flexion, 37 N at 15, 38 N at 30, and 29 N at 45 (Figure 3.4). 11,13-16,21,27,29,127, ,136,137,139, ATF of N increased in situ ACL force (mean peak N) more than isolated abduction moments of 5-10 N*m (peak N) 45

54 throughout flexion. 10,146,147 An isolated 200 N axial compressive force produced minimal ATT (2.7 to -0.4 mm throughout flexion); however, when combined with ATF, axial compression increased ATT relative to ATF alone. 127,148 At 30 flexion, 200 N of isolated axial compression generated slightly greater ACL forces than did 10 N*m of isolated valgus or 4 N*m of isolated internal torque; however, these forces were significantly lower than forces produced by combined torsional loading from Figure ,127,147 Figure 3.2: Displays the overall mean ATT, ITR, and ligament force for the ACL-intact, ACLD, and ACLR conditions in response to a Lachman s test simulated with 134 N applied ATF. A 134 N stimulus was selected for clinical relevance as this value represents the force generated by 30-pound test on a KT1000 device. (* indicates significant difference between ACL and ACLD, between ACLD and ACLR, and # between ACL and ACLR) 46

55 Figure 3.3: Displays the overall mean ATT, ITR, and ligament force for the ACL-intact, ACLD, and ACLR conditions in response to a Pivot-Shift test simulated with a combined 10 N*m abduction and 4-5 N*m internal torque. This loading condition was selected as it represented the most common Pivot-Shift scenario simulated in the non-weight-bearing passive flexion methodology. (* indicates significant difference between ACL and ACLD, between ACLD and ACLR, and # between ACL and ACLR) 47

56 Figure 3.4: Displays the mean ATT and ligament force for the ACL-intact condition at two separate levels of Pivot-Shift loading. The applied torques were 10 N*m abduction with 5 N*m internal and 7 N*m abduction with 5 N*m internal. (* indicates a significant difference was present) In the intact knee, a simulated 400 N quadriceps load produced ATT, ITR, and force on the ACL (Figure 3.5). 14,22-24,29 Relative to ATF, simulated quadriceps load produced less ATT throughout flexion, greater ITR from 0-60 flexion, and equivalent ligament force from 0-30 flexion. Simulated hamstrings co-contraction of 200 N reduced ATT and ITR relative to isolated quadriceps values and created a net posterior tibial shift when knee flexion exceeded Throughout flexion, abduction rotation and medial translation were greater under isolated quadriceps loading than under ATF, whereas the application of a hamstrings co-contraction decreased the magnitude of valgus rotations. 129 Abduction knee rotation and medial translation 48

57 under a quadriceps load were similar throughout flexion for an intact ACL and ACLR, but ACLD increased both values at 15 flexion. 29,129 The ensemble mean mechanical response of the ACL-intact, ACLD, and ACLR knee to simulated quadriceps forces throughout passive flexion are summarized in Figure ,9-12 Figure 3.5: Displays the mean ATT, ITR, and ligament force for the ACL-intact conditions in response to a simulated 134 N ATF, 400 N quadriceps load, and a combined 400 N quadriceps with 200 N hamstrings load. These loads were simulated in the non-weight-bearing passive flexion methodology. (* indicates significant difference between ATF and quadriceps loading, between ATF and combined loading, and # between quadriceps and combined loading) 49

58 Figure 3.6: Displays the mean ATT, ITR, and ligament force throughout non-weight-bearing, passive flexion for the ACL-intact, ACLD, and ACLR condition in response to simulated muscle loads of 400 N quadriceps. (* indicates significant difference between ACL and ACLD, between ACLD and ACLR, and # between ACL and ACLR) Weight-bearing Flexion Weight-bearing altered knee kinematics relative to zero load flexion simulations. In static positions of 15, 30, and 45 of flexion, the addition of 1*body weight increased ACL force by N, N, and 34.6 N N weight-bearing in the UKS increased ITR up to 16º, valgus rotation up to 2º, and medial, anterior, and proximal tibial translation up to 3mm compared with passive flexion. 150 In weight-bearing flexion, an ATF of 50 N increased peak ACL force (33 N to 55 N) from 15-55º of knee flexion. The application of ITT to weight-bearing 50

59 flexion across the same range did not affect ACL force relative to weight-bearing alone (peak 35 N vs. 33 N). 116 ITT of 5 N*m did increase ITR by a mean of 9º in both intact and ACLD specimens at flexion angles above 20º. 116,151 When compared to weight-bearing alone, the addition of ATF and ITT each increased ATT by only ~2 mm at all flexion angles in both the ACL and ACLD condition. 116,151 ATT increased from the ACL-intact to ACLD condition between 20-40º knee flexion in weight-bearing, between 15-65º flexion in weight-bearing with ATF, and between 15-30º flexion in weight-bearing with ITT. No ITR differences existed between ACL and ACLD specimens under any weight-bearing conditions. 151 Peak ACL strain for DKS weight-bearing was 4.3% with a peak rate of 120% /sec. 120 Peak quadriceps force was around 1000 N for DKS and UKS loading conditions. 116,120,152 ACLRs were only evaluated in the UKS. Compared to ACLDs, ACLRs decreased ATT by up to 2.4 mm in 50 N of weight-bearing, by 3.1 mm in weight-bearing plus 50 N ATF, and expressed no ITR change in weight-bearing plus 5 N*m ITT. 153 Kinematic Simulation Knee simulations driven by in vivo kinematics identified that porcine and ovine stifle joints are ACL-dependent structures during the stance phase of gait as they sustained peak ACL loads of up to 400 N. 78,154 Following hoof-strike, the mean ovine ACL load increased from 0-45º knee flexion and partial resection of the ACL decreased the ATF required to produce equivalent levels of ATT. 155 Compared to an intact specimen, the ACLD condition significantly reduced hoof-strike force in the anterior (44.3 vs N), medial (19.5 vs N), and compressive (70.8 vs N) directions and peak moments in flexion (9.3 vs. 0.6 N*m) and abduction (3.5 vs. 0.2 N*m) rotation during gait. Peak internal rotation moment during gait experienced no change between ACL-intact and ACLD knees. 1 The addition of up to 4 mm of ATT during 51

60 simulated gait in the porcine knee also increased peak anterior knee force in a linear fashion (~40 N /2 mm). 78 Force response to tibial displacement and rotation during cadaveric gait simulations indicated the ACL was a primary restraint to ATT (peak force 135 N) and a secondary restraint to medial tibial translation (peak force 47.8 N), knee flexion rotation (peak torque 12.4 N*m), and knee abduction/adduction rotation (peak torque 18.1 N*m). 74 Rotational perturbations of -0.5 to 0.5º applied to the initial position of cadaveric specimens had no effect on ACL-intact or ACL-only knee kinetics during gait. Conversely, anterior and compressive perturbations moving from -0.5 mm to 0.5 mm increased forces in the anterior, medial, and compressive directions. 79 In the porcine model, a bone-patellar-tendon-bone ACLR restored the anterior forces observed in the intact ACL, but did not match its rate of force loss. Relative force contributions during stance phase of gait were not consistent between the intact ACL and ACLR. The intact ACL was the primary contributor to anterior force, medial/lateral force, and flexion/extension moments and secondary contributor to compressive force and abduction/adduction moments, while the bone-patellar-tendon-bone ACLR was the primary contributor to all forces and moments except internal/external rotation. 1 The ACLR force contributions relative to ACL-intact and ACLD conditions are summarized in Table

61 Table 3.1: Forces and moments for the Native ACL (n = 11) and ACL-reconstructed (n = 6 each) knees at heelstrike (HS), midstance (MS), and toeoff (TO). Significant differences from the ACLintact condition are highlighted in bold (p < 0.05), with * indicating a significant difference from the intact ACL and indicating a significant increase from ACLD. This table was adapted with permission from Boguszewski Native ACL, ACL-D, and Grafted Knees - Forces (N ± SEM) and Moments (Nm ± SEM) Anterior Force Medial Force Compression Force Abduction Moment Flexion Moment Internal Moment ACL-D Native ACL Hybrid RTM BPTB HS -5.9 ± 1.9 * 44.3 ± ± ± ± 4.2 MS -5.0 ± 0.4 * 34.3 ± ± ± ± 4.0 TO -7.7 ± 1.0 * 34.3 ± ± ± 6.7 * 53.3 ± 2.5 * HS -0.3 ± 0.8 * 19.5 ± ± 5.7 * 44.7 ± 4.1 * 40.5 ± 4.3 * MS -2.1 ± 0.5 * 10.4 ± ± ± 2.8 * 22.6 ± 5.7 TO -1.8 ± 0.7 * 13.9 ± ± 5.4 * 33.6 ± 4.2 * 33.8 ± 33 * HS 33.6 ± 5.1 * 70.8 ± ± 15.7 * ± 6.7 * ± 6.9 * MS -0.6 ± ± ± 16.0 * 31.0 ± 8.0 * 29.0 ± 13.4 TO 33.3 ± 3.3 * 75.4 ± ± 15.8 * ± 11.3 * ± 8.5 * HS 0.2 ± 0.4 * 3.5 ± ± ± 0.5 * 7.6 ± 0.8 * MS -0.5 ± 0.1 * 1.3 ± ± ± ± 1.5 TO -0.8 ± 0.4 * 1.9 ± ± ± 0.6 * 5.8 ± 0.7 * HS 0.6 ± 0.3 * 9.3 ± ± ± ± 0.5 MS -0.1 ± 0.1 * 6.8 ± ± ± ± 0.7 TO 0.0 ± 0.2 * 7.4 ± ± ± ± 0.4 HS -0.2 ± ± ± ± ± 0.3 MS 0.1 ± ± ± ± ± 0.2 TO 0.0 ± ± ± ± ± 0.2 Mechanical Simulation The jump-landing simulator reproduced ACL ruptures in specimens loaded at 20º knee flexion with locked hip flexion, impact forces of approximately 1120 N, and quadriceps tendon force under 170 N. 121 Tests performed with large quadriceps load and increased impact force did not rupture the ACL, even in an abducted position. In ACL failures or loosening, the peak relative ACL strain was between % with a strain rate above 250% /sec. A positive correlation was identified between quadriceps pretension and static ACL strain, while a negative correlation was found between quadriceps pretension and the dynamic ACL strain that leads to rupture. 121,156 Conversely, with an impact force of 1400 N, the knee testing apparatus demonstrated that ACL strain directly correlated with increased quadriceps tension force. 86,122 Further, the application of hamstrings tension limited ACL strain by 70%, which was 53

62 predominantly due to a 1.2 mm reduction in ATT, though the added tension also influenced minor adjustments to valgus, flexion, and internal rotation. 157 The knee testing apparatus demonstrated that ACL strain was not dependent on impact force, but corresponded with valgus rotation as 10º of valgus increased ATT by 1.1 mm and peak relative ACL strain to 4.5%. 86,122 This represented a 30% increase over no valgus rotation. Once the torsional transformer incorporated ITT, it increased peak relative ACL strain (3.0% to 5.4%) and peak ACL strain rate (184% to 252%). 84 The addition of N*m of ITT during simulations increased ITR from 1.6º-11.6º in ACL-intact and from 2.5º-12.5º in ACLD specimens. ITT also increased ATT by 2.6 mm in ACL-intact specimens and 3.0 mm in ACLD knees. 12 When N*m ITTs were combined with 7º varus or valgus knee angles, peak relative ACL strain reached 7%. 85 Overall, It was reported that the ACL was a secondary contributor to ITT resistance and accounted for approximately 13% of ITT resistance during landing simulations. 12 DISCUSSION In vitro simulations of the knee have been used to evaluate how kinematics and kinetics contribute to ACL loading and injuries. The purpose of this review and meta-analysis was to compare robotic and mechanical methods of in vitro knee simulation in order to investigate the functional behavior of the ACL and ACLR and to analyze differences observed between these methodologies. The ACL is the primary restraint to ATT in the tibiofemoral joint, resisting up to 87% of the ATF. 7 ATF was the most influential loading condition in all simulation methodologies. This behavior was evidenced in the ensemble mean ATT and ligament force data compiled from passive flexion (Figure 3.2). When the intact ACL was resected, the largest kinematic change occurred in ATT during both the Lachman s and pivot-shift loading conditions (Figure 3.2 and 54

63 3.3). Also, the peak ligament forces generated from ATF and isolated quadriceps simulation were greater than the force from combined abduction and internal torsional loading. Throughout kinematic gait simulations, the largest force drops between the ACL and ACLD conditions occurred in the anterior direction and averaged above 40 N. 1 Further, from 0-30 flexion, the application of quadriceps force, which generates ATF due to its tendon insertion on the proximal anterior tibia, to passive flexion simulations exhibited increased ATT and equivalent ligament forces to those produced by ATF (Figure 3.5). These behaviors confirmed that robotic simulations maintained the integrity of the ACL as the primary restraint to ATT and ATF throughout motion. However, within mechanical impact simulations there was conflict over how ACL strain correlated with quadriceps force. The knee testing apparatus maintained the traditional convention of a direct relationship between increased quadriceps force and increased ACL strain, while the jump-landing simulator indicated an inverse relationship between quadriceps force and dynamic ACL strain. 86,121,122,156 The jump-landing simulator claimed that the compressive forces generated by simulated quadriceps contraction protected the ACL during motion more than the generated ATF strained the ligament. 55,58 This finding is in direct contrast with the passive flexion and weight-bearing flexion techniques that reported combined ATF and axial compressive loads increased ACL forces relative to either isolated loading condition. 116,127,148 The quadriceps forces applied in both impact testing devices and the weightbearing flexion simulations were in excess of 1000 N; therefore, dissimilarities between methods did not arise from a disparity in muscle force magnitude. 86,116, ,152,157 Robotic simulations demonstrated that abduction torque and rotation at the knee had greater impact on ACL forces and kinematics than internal torque. A slight increase in abduction torque throughout passive flexion increased ATT and in situ ACL force (Figure 3.4), which 55

64 exemplified its impact on ACL mechanics. Isolated abduction torque produced peak ACL forces that were ~20% greater than isolated internal torque. 10 The application of weight-bearing altered ITR angle more than any other kinematic variable with 8 times greater change than was observed in knee abduction under the same loads. 150 These magnitude differences indicate that knee abduction was more influential to ACL mechanics than internal rotation and are congruent with kinematic simulations that depicted the ACL was a secondary restraint to abduction torque during gait. 1 Hence, significant knee abduction torque, but not ITT, differences were observed between ACL-intact and ACLD specimens. 1 During weight-bearing flexion, ITT did not increase ACL force throughout flexion. 116 A lack of mechanical resistance to ITR was echoed in passive flexion simulations as ACLD did not increase ITR during weight-bearing and isolated ITT produced no ITR differences between ACL and ACLD specimens. 141,151 If the ACL were a functional restraint to ITT, then the intact ligament should have restricted the observed ITR relative to the ACLD condition. Mechanical impact testing exhibited different ACL mechanics in response to rotational stimuli than robotic simulation methods. Though the knee testing apparatus showed that additional knee valgus at the time of impact corresponds with increased ACL strain, it also reported that isolated ITT had a potentially greater influence on increasing ACL strain during landing. 86,122 The increase in ACL strain from additional valgus rotation at impact collaborates both with results from robotic simulations that indicate abduction is a significant antagonist to ACL forces and with literature that reported an 8 increase in knee abduction angle at initial contact of landing is associated with increased ACL injury risk. 46 However, the effect of ITT on ACL strain during mechanical simulation contrasts the findings of robotic simulations. Whereas ACL contributions to resist ITT could not be quantified in kinematic gait simulations, 1 the knee 56

65 testing apparatus found the ACL to be a secondary resistor to ITT, much as it is to knee abduction. 12 Peak ITR with respect to isolated ITT in the knee testing apparatus was less than the peak ITR for passive flexion, yet had a more profound impact on ACL mechanics. 10 All methods of simulation demonstrated that combined loading with coupled abduction and ITT had greater impact on ligament mechanics than either isolated condition. In mechanical testing, ACL strain was ~2% larger under combined loads than in either isolated condition ,122 In robotic testing, ACL forces from coupled loading were ~40 N larger than isolated abduction torque and ~50 N larger than isolated ITT at full extension. 10 These mechanical behaviors are in agreement with literature that supports valgus collapse at the knee, defined as the outward angulation of the distal segment of a bone or joint due to a pure abduction motion of the distal tibia relative to the femur or from transverse plane knee rotation motions, 158 to be a primary mechanism of ACL injury and injury risk prediction. 46,75 Therefore, though combined torsional loading had the greatest impact on ACL mechanics in all simulation methods, mechanical differences produced by isolated torques supported the primary hypothesis that rotational motions in the knee would elicit different responses in robotically-driven versus mechanical-impact knee simulators. Some of the biomechanical dissimilarities between robotic simulation and mechanical impact may have arisen from structural and motion-constraint limitations within each testing apparatus. Variability is naturally associated with human movement cycles as even passive flexion articulations of the knee exhibit pathway variance. 159,160 In robotic simulations highprecision robotic manipulators exclude the natural variability associated with human movement cycles, which predicates that structures are being abnormally loaded through constant and repetitive force application. Unlike robotic simulations, where limb positions were dictated by 57

66 either in vivo recorded kinematics or clinical exam procedures, motion pathways in mechanical impact simulations were not preset. 12,84-86,121,122,156,157 Instead, specimens uniquely reacted to each force impulse while pre-tensioned muscles and artificial hip and ankle joints constrained movement. However, during in vivo motion, individual muscle force contributions are in flux throughout a landing as subjects adapt to changes in geometry and ground reaction forces. 161,162 Conversely, muscle forces within the mechanical apparatus were constrained to either a constant level or lengthening determined by bone position. 12,84-86,121,122,156,157 The inability to match the dynamic nature of in vivo structures implies that mechanical simulation constraints may have been physiologically inaccurate. This concern of non-physiologic response was augmented by data that indicated peak knee flexion range of motion was ~6 during mechanical simulation, 85,122 whereas in vivo data from comparable-force landings has indicated that peak flexion range of motion exceeds An order of magnitude difference in knee flexion angle suggests that the mechanical testing apparatus enacted potentially non-physiologic pathways of force distribution as compared to in vivo landings. The genesis of these knee flexion differences may be that mechanical-impact simulations deliver singular impulse loads to each specimen; whereas, during in vivo landings, ground reaction forces propagate through the leg for the duration of stance phase. The instantaneous versus continuous application of force could greatly influence knee biomechanics. Also, for a methodology dependent on tissue structures to constrain motion pathways, much of the natural anatomy was resected including the iliotibial band, muscle mass, skin, and the ankle and hip joints. 12,84-86,121,122,156,157 Tissues were also resected in robotic simulations; however, in those models, the motion was prerecorded and constrained by the manipulator, not by soft-tissues within the model. 73, In robotic 58

67 simulations, positional control provided by the robotic manipulator represented the muscles and other resected tissues that would have constrained in vivo joint motion. Inconsistent restoration of ACL-intact kinematics across a wide spectrum of simulated, functional loading conditions partially rejected the hypothesis that ACLRs would restore ATT and ATF observed in the intact ACL, but fail to restore kinetics and kinematics in the other degrees of freedom. Though ATT increased at 15 flexion, ACLRs were able to restore ensemble mean ATT values to ACL-intact levels at 0, 30, 60, and 90 of passive flexion under combined torsional loading (Figure 3.3). However, under ATF, ACLRs reduced ATT relative to the ACLD condition, but did not restore ATT to the ACL-intact condition (Figure 3.2). Under constant quadriceps loading, ACLRs produced significantly lower ligament forces and greater ATT when compared to ACL-intact knees at flexion below 60 (Figure 3.5). Thus, the hypothesis was supported under combined torsional loading, but rejected under ATF loading. Mean ITR following ACLR was comparable to intact kinematics during combined torsional loading, but over constrained during ATF and quadriceps loading (Figures 3.2, 3.3, and 3.6). This again provided mixed support of the hypothesis based on the loading condition. Overconstrained knee kinematics were also present in simulated gait as ACLR grafts became primary loading restraints in two additional DOFs and exhibited larger medial forces compared to the intact ACL. 1 In the remaining 5-DOFs, ACLRs exhibited an inconsistent ability to restore ACLintact mechanics throughout a gait cycle. Compared to the intact knee, bone-patellar-tendonbone ACL grafts restored anterior force at heel strike and mid-stance, but expressed greater force at toe-off; matched the medial and compressive forces at mid-stance, but exhibited greater forces at heel strike and toe-off; restored the flexion and internal rotation moments of the intact knee at all points during gait; and only restored abduction moments at mid-stance. With the exception of 59

68 internal rotation, forces from bone-patellar-tendon-bone ACLRs were significantly increased in all DOFs at all points in the gait cycle relative to the ACLD condition. 1 Again, the inconsistency of these graft loading patterns both supported and rejected the hypothesis. It has been well documented that ACL injuries leads to early onset osteoarthritis, 63,66 and ACLRs do not appear to greatly improve the long term prospectus for knee injuries as 75% of athletes still complain of knee degradation affecting their quality of life less than 15 years post surgery. 63,64 It is possible that the inconsistent ability of the ACLR to restore intact kinematics across multiple loading scenarios, may alter the normal mechanical conditions across the articulating surfaces of the knee and lead to joint degeneration. Limitations to this systematic review include that it did not account for confounding variables that can impact the mechanical integrity of ACLRs. Variability in attributes such as anatomic versus non-anatomic tunnel placement, 164,165 double versus single bundle grafts, 29 graft materials, 1 number of tunnels, 27 graft fixation method, 11 graft tension, 124 and graft length 166 has been shown to alter the mechanical response of an ACLR and could alter the ensemble mean. Potentially confounding factors within ACLR grafts were not accounted for in the exclusionary criteria because opinions on optimal surgical technique varies between orthopaedic surgeons, which leads to graft variability in ACLR populations that should not be artificially controlled in a meta-analysis. An additional limitation to the evaluation of ACLRs in in vitro specimens is that the grafts are being evaluated for integrity at the time of surgery. In vivo, grafts are provided time to heal, experience bony ingrowth, and potentially restructure fiber orientation relative to mechanical environment prior to return to sport. Due to the nature of in vitro specimens, natural processes cannot be reproduced, which may alter graft response to mechanical stimuli

69 For passive flexion simulations, this investigation was limited to Lachman s tests of 134 N ATF, which corresponds to a 30-pound KT1000 test, and pivot-shift tests of 10 N*m valgus and 4-5 N*m internal torque. These loading magnitudes were restricted to constant values to improve comparability between results. Altered magnitudes in external loading could have confounded or biased the ensemble averages through increased variability. However, biomechanical tendencies in investigations with different loading magnitudes were often similar as ACLD increased ATT relative to ACL-intact knees throughout flexion under each Lachman s and pivot-shift loads; 10,15,141,144,147, ACLR restored ATT relative to ACLD under Lachman s loads, but not necessarily to the level of ACL-intact knees; 141,166,169,171 ACLR restored ITR relative to ACL-intact knees under rotational loads; 24,126,141 and ACLR graft forces were not consistent with ACL-intact ligament forces ,169,171 Therefore, normalized mechanical response of ACL, ACLD, and ACLR specimens generally remains constant across varied magnitudes of external loading, though absolute values may differ. Future studies should explore the development of more dynamic motion pathways through robotically simulated motions. ACL failures are most commonly associated with the jump-landing and side-step cutting activities in basketball and soccer, not gait or clinical motions presented in robotic simulations. 42,172 A combination of rigorous motion activities, such as the landing impact simulated in mechanical testing, combined with the precision of robotic manipulation should produce a wealth of biomechanical data that could be utilized to design more efficacious ACL injury prevention protocols and ACLR graft constructs. However, robotic systems require 6-DOF kinematic input to reproduce a motion and, though they are relatively accurate rotationally, 3D motion capture systems introduce large errors in the translational DOFs due to skin artefacts. 173,174 In order to appropriately simulate dynamic, in vivo activities, 61

70 investigators will either need to capture motion with bone-based markers, as was done in the development of gait simulations, 73,74 or address the kinematic errors generated by skin-based markers. CONCLUSIONS In vitro simulations of knee motion attributed ATT as the primary mechanical antagonist to the ACL and indicated that combined torsional loading has a greater biomechanical influence than uniaxial moments. Abduction rotation had a greater mechanical influence in robotic simulations, while ITR may have had a greater influence during mechanical-impact simulations. Both methodologies exhibited limitations, but greater concerns were raised with the ability of impact simulations to accurately recreate physiologic motions and in vivo data currently better supports the behaviors observed during robotically-driven simulations. Robotic simulations found that ACLRs constrain ACLD knees, but their ability to restore ACL-intact mechanics is dependent on the DOF observed and loading conditions applied. ACKNOWLEDGEMENTS This work was supported by NIH grants R01-AR049735, R01-AR055563, R01- AR and R The authors thank the clinical research staffs at The Ohio State University s Sports Health Performance Institute and the Cincinnati Children s Hospital Sports Medicine Biodynamics Center. 62

71 Chapter 4 A Novel Methodology for the Simulation of Athletic Tasks on Cadaveric Knee Joints with Respect to In Vivo Kinematics Nathaniel A. Bates, a,b,c Rebecca J. Nesbitt, a Jason T. Shearn, a Gregory D. Myer, b,d,e,f Timothy E. Hewett a,b,c,d,g a Department of Biomedical Engineering, University of Cincinnati, Cincinnati, OH, USA b Sports Medicine Biodynamics Center, Division of Sports Cincinnati Children s Hospital Medical Center, Cincinnati, OH, USA c The Sports Health and Performance Institute, The Ohio State University, Columbus, OH, USA d Department of Pediatrics, College of Medicine, University of Cincinnati, Cincinnati, OH, USA e Department of Orthopaedic Surgery, College of Medicine, University of Cincinnati, Cincinnati, OH, USA f The Micheli Center for Sports Injury Prevention, Boston, MA g Departments of Physiology and Cell Biology, Orthopaedic Surgery, Family Medicine, and Biomedical Engineering, The Ohio State University, Columbus, OH, USA This manuscript is currently in preparation for submission to Journal of Applied Biomechanics. 63

72 ABSTRACT Six degree of freedom (6-DOF) robotic manipulators have simulated clinical tests and gait on cadaveric knees to examine knee biomechanics. However, these activities do not necessarily emulate the kinematics and kinetics that lead to anterior cruciate ligament (ACL) rupture. The purpose of this study was to determine the techniques needed to derive reproducible, in vitro simulations from in vivo skin-marker kinematics recorded during simulated athletic tasks. Input of raw, in vivo, skin-marker-derived motion capture kinematics consistently resulted in specimen failure. The protocol described in this study developed an in-depth methodology to adapt in vivo kinematic recordings into 6-DOF knee motion simulations for drop vertical jumps and sidestep cutting. Our simulation method reliably and repeatably produced kinetics consistent with vertical ground reaction patterns while preserving specimen integrity. Athletic task simulation represents an advancement that allows investigators to examine ACL-intact and graft biomechanics during motions that generate greater kinetics, and the athletic tasks are more representative of documented cases of ligament rupture. Establishment of baseline functional mechanics within the knee joint during athletic tasks will serve to advance the prevention, repair and rehabilitation of ACL injuries. INTRODUCTION Six-degree-of-freedom (DOF) robotic manipulators have been used to simulate knee joint motions on cadaveric specimens for two decades. 118 These prior investigations have gleaned important biomechanical information on the function of the anterior cruciate ligament (ACL) and the relative performance of various reconstructions (ACLR) during simulated clinical motions, such as the Lachman s and pivot shift tests and walking gait cycles. 73,74,78,134,135 These clinical tests were created to assess normal ACL and ACLR integrity, but do not assess how the native 64

73 ACL or ACLR functions during activities of daily living (ADLs) including athletic activities. While gait is a physiologic activity, it does not provide insight into the potential injury mechanics or how ACLRs will perform in return to sport. ACL injuries are most commonly associated with tasks that involve rapid deceleration and/or change in direction, and these same activities will provide a significant challenge to the ACLR. 42 Therefore, to better understand the biomechanical precursors to ACL injury and assess the performance of ACLR grafts in athletic settings it would be valuable to simulate jump landing and sidestep cutting tasks on cadaveric specimens using six DOF robotic manipulators. In order to execute joint simulations, robotic manipulators require kinematic positional or kinetic force data inputs to define the 6-DOF of motion. Unfortunately, 3D kinematics collected in vivo from skin-marker-based motion capture are known to incorporate multiple sources of error It has been documented that, relative to the gold-standard of bone-pin-based motion capture, skin-based markers overestimate the range of motion (ROM) that occurs in the three rotational degrees of freedom of the knee during athletic tasks. 174 In respect to the translational DOFs, the standard errors associated with data collection can exceed the ROMs that actually occur within the knee joint. 173 Such large errors make translational kinematics unreliable and, unless bone pin markers were used during motion capture, they often go unreported. In relation to cadaveric simulations, the noted errors can lead to impingement of the bony structures and/or excessive distraction within the joint. When large enough these disassociation of the joint can result in structural damage. Therefore, in order to create physiologically representative in vitro simulations from in vivo motion data, investigators must account for these sources of error. 65

74 The purpose of this report is to define techniques used with a 6-DOF robotic manipulator to create reproducible, physiologically-representative, in vitro simulations derived from in vivo skin-marker-based kinematics recorded during athletic tasks. METHODS Specimen preparation Lower extremity cadaveric specimens were defined from the femoral head to the distal end of the tibia and were obtained from an anatomical donations program (Anatomical Gifts Registry, Hanover, MD). Specimen criteria were defined as no previous history of knee trauma, knee surgery, bone cancer, or ankle or shin implants. The limbs were kept frozen at -20 C until the day before testing, then removed from the freezer and allowed to thaw overnight. Specimens were dissected down to the joint capsule, leaving the collateral and cruciate ligaments and menisci intact. The femur was left intact while the tibia was cut transversely 15 cm inferior to the joint line and the fibula was cut inferior to the insertion site of the lateral collateral ligament. Anatomical landmarks were used to locate the joint coordinate system of the knee according to Grood-Suntay system. 178 The coordinate system was used to define the placement of mechanical fixtures on the tibia and femur of each specimen. On the tibia, a rod defining the medial/lateral axis was drilled through two points just distal of the LCL and MCL insertion sites and in line with the tibial spines. A pipe defining the longitudinal axis was then secured over the distal end of the bone and oriented perpendicular to the rod and parallel to the long axis of the bone. On the femur, a cylinder was placed over the long axis of the bone approximately 10 cm proximal to the joint line. 78 The tibial fixture was then attached to a 6-axis load cell (Theta Model Industrial Automation Load Cell, ATI Industrial Automation, Apex, NC) mounted on the end effector of a 6-DOF robotic manipulator (KR210; KUKA Robotics Corp., Clinton Township, MI). A digital 66

75 coordinate measuring machine (FARO Gauge; FARO Technologies, Lake Mary, FL) was used to determine the tibial coordinate system within the robot coordinate axes and define the knee joint center. 78,178 The femur was then secured to a rigid base and the robotic manipulator moved the tibia with respect to the static femur, Figure DOF kinematic inputs were then used to drive the robotic simulations. 3-dimensional (3D) motion capture Kinematic data were collected on a male (age = 24; height = 175 cm; mass = 68.8 Kg) and female subject (age = 25; height = 170 cm; mass = 64.4 Kg) who were matched for age, height, mass, and Figure 4.1: Depicts frontal and sagittal plane views of a lower-extremity, cadaveric specimen affixed to the 6-DOF robotic manipulator and prepared for the simulation of athletic tasks at the knee joint. athletic activity. These subjects were devoid of prior knee injuries and participated in athletics at the time of data collection. Informed consent was obtained from each subject prior to data collection and testing procedures were approved by the Cincinnati Children s Hospital IRB. Thirty-seven (37) skin-based, retroreflective markers were instrumented at anatomical landmarks on each subject in a modified Helen Hayes marker set. 179 Motion was sampled at 240 Hz with a 10-camera motion analysis system (Eagle cameras, Motion Analysis Corporation, Santa Rosa, CA). Ground reaction force (GRF) data was simultaneously sampled at 1200 Hz with dual, in-ground, multi-axis force platforms (AMTI, BP600900, Watertown, MA) such that each platform corresponded with a single leg

76 Participants performed three trials each of a drop vertical jump (DVJ) and sidestep cutting as previously described. 75,180 3D kinematics data were processed through Visual3D (version 4.0, C-Motion, Inc., Germantown, MD) with custom MATLAB code (version 2012b, The MathWorks, Inc., Natick, MA) using an established biomechanical model. 179 Marker trajectories were filtered through a fourth-order, low-pass, digital filter with a cutoff frequency of 6 Hz, while GRF data were filtered at a cutoff frequency of 100 Hz for kinematic and kinetic calculations. Kinematic curves for all three trials were averaged into an ensemble subject mean for each motion performed. 3D kinematics have demonstrated very high intra-session reliability. 179 Rotational kinematic input Scale factors were developed to restrict the overestimation of in vivo ROMs recorded for each rotational DOF. These scale factors were determined via comparison of 3D motion data that was concomitantly collected with bone-pins and skin-markers throughout the stance phase of sidestep cutting tasks. 174 Based on the literature, the bone-pin range of motion was 47.0% of the skin-marker range in internal/external rotation, 85.0% in flexion/extension, and 78.7% in abduction/adduction. The athletic tasks to be simulated via the 6-DOF robotic manipulator were of a similar level of rigor as those performed in the literature. Therefore, to reduce rotational overestimation from in vivo recordings, the listed scale factors were individually applied to each rotational DOF. These scaled rotational curves were then used as input to control the robotic manipulator (Figure 4.2). Translational kinematic input To reduce the influence of artifact errors on outcome measures, a single set of translational input was established for each motion task and utilized for all simulation conditions 68

77 Figure 4.2: Rotational knee joint kinematics recorded in vivo (solid line) with the adjusted input for the robotic manipulator (dashed line). All 3 rotational degrees of freedom from a male subject DVJ are represented. Time series was normalized to percent of landing phase. applied to that task. Translational input for the compression/distraction DOF were selected from in vivo recordings and were based on curve shape. For each athletic task, the male and female kinematics recorded in vivo were examined and the compression/distraction curve that best emulated the trajectory of the vertical ground reaction force curve was selected to represent this DOF. Prior to this investigation, a digital coordinate measuring machine was used to collect pilot data on the translational ROMs experienced by the tibia when cadaveric knees were passively flexed from The ROM of the origin of the tibial axis with respect to the origin of the femoral axis was 4 mm in the compression/distraction DOF. In order to articulate around, rather than through, the bony geometry of the knee, the translations recorded in vivo were condensed to satisfy this ROM. The anterior/posterior motion that corresponded with the selected compression/distraction curve was also used. Anterior/posterior input was scaled to exhibit 1.7 times the ROM of the compression/distraction curve. This factor was again established relative to the ROM relationships observed in the pilot passive flexion data. Medial/lateral translations 69

78 recorded in vivo consistently exhibited unrealistically large ROMs in excess of 1 cm. During passive flexion, the mean medial/lateral ROM was under 2 mm. Previous literature has demonstrated that small perturbations in the medial/lateral DOF during robotic knee simulations have a negligible effect on joint forces and torques. 79 Therefore, rather than input data degraded by skin artifact error, the medial/lateral DOF did not change from the initial limb alignment, Figure 4.3. Figure 4.3: Translational knee joint kinematics recorded in vivo (solid line) with the adjusted input for the robotic manipulator (dashed line). All 3 rotational degrees of freedom from a male subject DVJ are represented. Time series was normalized to percent of landing phase. Initial position The limb was initially oriented to the rotations specified by the scaled in vivo kinematics at the time of initial contact. Initial translational orientation was defined by articulating the specimen to a position that satisfied the conditions of near zero joint forces and torques, articular cartilage contact in both the medial and lateral compartments on the tibial plateau, and minimal 70

79 distance between the origins of the tibial and femoral coordinate axes. During in vivo testing, initial contact represented the time point when 10 N of force registered on the force platform. Therefore, the net external forces propagated to the knee joint should initially be negligible, which supports a zero force orientation. While this process determined a preliminary position for the specimen, muscle pre-activation makes it unlikely that the joint is actually unloaded at initial contact. However, when the joint was compressed relative to the force normalization described in the next section, compression was consequently incorporated at initial contact and presumably represented joint loading caused by muscle pre-activation. Force normalization In vitro joint forces and torques were measured with a 6-DOF force sensor mounted on the end effector of the robotic manipulator that was aligned with the vertical anatomical axis of the tibia for each specimen. This method has been readily applied in previous robotic simulation protocols. 74,78,181 Previous in vivo investigations have demonstrated that subjects generate between times bodyweight of vertical ground reaction force per leg when landing from a drop vertical jump with an initial height of 31 cm. 83 As approximately 90% of a person s mass is supported above the knee, initial joint compression during simulation was determined such that the peak compressive force attained within the specimen knee during simulation was matched to the in vivo condition of between times bodyweight. Initial compression for sidestep cutting was also established relative to peak in vivo forces of bodyweights. Simulation Duration Simulations of the DVJ and sidestep cut were restricted to the landing phase of their respective motions. Landing phase has been previously defined in the literature and consists of the period between initial contact with the force platforms and minimum height of the center of 71

80 mass. 83 For the present subjects, landing phase spanned 0.25 sec for the male and 0.23 sec for the female. These values correspond well with previous literature that indicates landing phase duration is approximately 0.25 sec during a DVJ. 83 The landing phase was isolated for simulation as this stage represents the timeframe where the ACL is most likely to experience injury and should be exposed to the most significant loading from ground reaction forces. Video analysis of ACL injuries supports this timeframe as footstrike was noted to immediately precede leg collapse in all observed noncontact injuries. 185 Reliability Once established, the presented methodology was applied to 19 limbs from 12 unique donors (age = 47.9 ± 7.0 years; mass = 832 ± 190 N). Three additional specimens were excluded due to specimen failure or non-functional ACLs. Specimens were preconditioned for 10 cycles, and then simulated through an additional 10 cycles where force/torque values were recorded by the force sensor. A mean of the 8 th and 9 th cycle from each 10-cycle test were used for statistical analysis to eliminate cycle effects. All data was time normalized to percentage of landing phase. Waveform reliability for the entire landing phase was calculated between specimens using the coefficient of multiple correlation (CMC). 186,187 Calculations were performed in MATLAB with custom code. RESULTS The use of the raw skin-marker-based, in vivo motion capture kinematics as positioncontrol input for robotic simulations of athletic tasks at the knee joint resulted in consistent specimen failure (Figure 4.4). Femoral fracture (Figure 4.4b) was the most common failure with raw kinematics as the femur could not support the large shear forces generated from improper tibial alignment during deep flexion. Following the described methodology, a robotic 72

81 manipulator simulated 6-DOF knee motion corresponding to DVJs and sidestep cutting with no damage to the specimen. Joint kinetics recorded by the force sensor across multiple motion cycles indicated that the athletic task simulations were consistently reproducible throughout the simulation (Figure 4.5). The specimen depicted in Figure 4.5 had a mass of 848 N and exhibited a maximal compressive force of 1726 N (2.03 * body weight) during simulation, which was within our design parameters. Peak internal joint torques attained at this level of loading were physiologically sustainable with N*m in knee extension, N*m in knee adduction, and 7.29 N*m in knee external rotation. Between-subject CMC reliability across the whole specimen population for the 3 translational DOFs (anterior/posterior CMC = 0.973, medial/lateral CMC = 0.990, compression/distraction CMC = 0.987) and 3 rotational DOFs (internal/external CMC = 0.934, flexion/extension CMC = 0.963, abduction/adduction CMC = 0.982) was excellent (Figure 4.6). Figure 4.4: Failures that resulted after raw in vivo kinematics were used as input to drive the robotic simulations. Tibia fracture from excessive joint compression (A), femoral fracture from excessive anterior translation (B), and lateral joint dislocation from overestimated rotations (C). 73

82 Figure 4.5: Unfiltered internal knee torques and translational forces during each cycle of a 10-cycle male DVJ simulation on a single specimen. After viscoelastic effects have been compensated, the forces and torques produced at the knee were highly reliable between cycles. 74

83 Figure 4.6: Knee joint loading for all 12 unique donors throughout the DVJ simulation that was derived from the male model. Blue lines represent individual subjects, while the red line and shaded area represent the population mean and standard deviation, respectively. Waveforms were highly repeatable between specimens as CMC values exceeded in all DOFs. 75

84 DISCUSSION The techniques defined in this investigation allowed a 6-DOF robotic manipulator to successfully simulate physiologically representative motions from skin-marker-based recorded in vivo athletic tasks with excellent reliability between specimens. Development of a methodology to simulate physiologic, in vivo athletic tasks offers investigators several advantages over current methods. Current robotic simulations are limited to clinical tests that do not represent physiologic motion or low intensity gait cycles, neither of which represent tasks related to ACL injury. The simulation of athletic tasks represents an advancement that will allow investigators to examine ligament biomechanics over a broader range of activities, with the potential to examine biomechanics during activities that are associated with non-contact ACL injuries. 42,172 The proposed methodology can be adjusted to simulated the biomechanical effects of kinematic risk factors that have been prospectively associated with ACL injury. 46 Also, as with previous methods, the current technique can be combined with ACLRs to evaluate the performance of graft type and surgical strategies during athletic tasks. 134,135,138 As one of the objectives of ACLR is to return athletes to sport, biomechanical data on graft performance in simulated athletic environments could prove instrumental for improving the tissue engineering, surgical, and rehabilitation design parameters. Functional tissue engineering has demonstrated that ACLR graft constructs do not need to replicate the maximal mechanical properties of the intact-acl. Rather, grafts only need to function at and provide a safety factor for physiologically relevant levels of mechanical loading that are seen during activates of daily living. 115 However, there exists a disparity between graft failure mechanics and those of activities of daily living. Similarly, there is also likely a disparity between ADLs and athletic tasks. Grafts that perform well during robotic simulations of clinical tasks and ADLs may not function as effectively during athletic tasks. Therefore, it is important to 76

85 understand ACL and ACLR biomechanics during athletic tasks, as return to sport is often a driving factor behind patients choosing ACL reconstruction. 188 The novel methodology presented in this report provides a critical bridge to understanding the required design criteria for the surgical technique and graft. The direct input of 6-DOF skin-marker-based in vivo kinematics to the robotic manipulator did not produce physiologic simulations of knee motion. The published literature documents that skin-marker based in vivo motion capture suffers from skin artifact errors due to the skin moving in relation to bone orientation When skin-marker kinematics were simulated using robotic, these errors produced large joint dislocations, femoral fractures, and tibial fractures. (Figure 4.4) The implementation of scale factors to reduce the ROM recorded in rotational DOFs reduced the disarticulation in both the medial and lateral compartments of the tibiofemoral joint. Similarly, the adjustment of translational inputs with respect to passive flexion constraints prevented further fractures of the tibia and femur. Prior to this modification, translational inputs were failing to articulate around the bony geometry of the joint and causing the bones to drive through one another to the point of fracture. Though physiologic translations would change for every unique motion simulation, the unification of translational input in the current methodology removes the impact of translational errors on our results. Perturbations in the medial/lateral DOF have minimal impact on the resulting biomechanics during robotic simulation of knee joints, but anterior/posterior and compression/distraction perturbations do influence joint forces and torques. 79 Therefore, the unreliable and inconsistent errors associated with skin-marker-based motion capture would make it difficult to discern the root of mechanical differences in robotically controlled simulations. Instead, a universal set of physiologically-derived translations applied across multiple 77

86 simulations of the same task will allow the present methodology to more accurately represent biomechanical differences related to rotational kinematic differences recorded between subjects during athletic tasks. Understanding the mechanical influence of rotational kinematics is important to ACL biomechanics as increased abduction and decreased flexion during landing have been associated with injury risk. 46 The biomechanical properties of cadaveric tissues are known to degrade with age In previous robotic studies that simulate lower intensity tasks such as clinical Lachman s test or gait cycles, specimen age has not affected the ability to perform reliable simulations. 74,134,135 However, the current procedures simulate athletic tasks with higher levels of intensity up to 2.5 times body weight. Specimens from elderly donors have often proven unable to sustain such high force magnitudes; and therefore, it is important to limit the specimen population during these simulations to younger and physically active donors. Tibiofemoral joint geometry varies between specimens. Accordingly, specimen specific normalization of input kinematics may be necessary to account for the unique anatomy of each specimen. Currently, the data that has been collected in limited pilot simulations does not support the need for specimen specific normalization, but this theory will be revisited once a greater database is established. CONCLUSION The current methodological report presents a critical step in bridging the gap between laboratory observations and in vivo performance. Through the establishment of baseline functional mechanics within the knee joint during athletic tasks the methodology presented will serve to advance the prevention, repair, and rehabilitation of ACL injuries. Moving forward this methodology will allow researchers to identify mechanical differences between ACL-intact, 78

87 ACL-deficient, and ACL-reconstructed specimens during physiologically-relevant athletic tasks where ACL function is critical to joint stability. Comparatively, current methods of simulation present limited clinical relevance to athletic settings. As athletes commonly experience osteoarthritis following ACL injury, the assessment of biomechanical differences between intact and reconstructed knees for athletic tasks may lead to surgical adjustments that compensate for deficiencies and improve the long term joint quality of life following ACLR. ACKNOWLEDGEMENTS This work was supported by the National Institutes of Health/NIAMS Grants #R01- AR049735, #R01-AR05563, #R01-AR and #R01-AR The authors would also like to acknowledge the support of the staff at the Sports Health and Performance Institute at The Ohio State University and the Sports Medicine Biodynamics Laboratory at Cincinnati Children s Hospital. CONFLICT OF INTEREST There were no conflicts of interest in the preparation of this manuscript. 79

88 Chapter 5 Prediction of Kinematic and Kinetic Performance in a Drop Vertical Jump with Individual Anthropometric Factors in Adolescent Female Athletes: Implications for Cadaveric Investigations Nathaniel A. Bates, a,b,c Gregory D. Myer, a,d,e,f Timothy E. Hewett, a,b,c,d,g a Cincinnati Children s Hospital Medical Center, Sports Medicine Biodynamics Center and Human Performance Laboratory, Cincinnati, OH, USA b University of Cincinnati, Department of Biomedical Engineering, Cincinnati, OH, USA c The Sports Health and Performance Institute, OSU Sports Medicine, The Ohio State University, Columbus, OH, USA d Department of Pediatrics, College of Medicine, University of Cincinnati, OH, USA e Department Orthopaedic Surgery, College of Medicine, University of Cincinnati, OH, USA f The Micheli Center for Sports Injury Prevention, Boston, MA g Departments of Physiology and Cell Biology, Orthopaedic Surgery, Family Medicine and Biomedical Engineering, The Ohio State University, Columbus, OH, USA This manuscript is currently submitted for publication to the Annals of Biomedical Engineering. 80

89 ABSTRACT Anterior cruciate ligament injuries are common, expensive to repair, and often debilitate athletic careers. Robotic manipulators have evaluated knee ligament biomechanics in cadaveric specimens, but face limitations such as accounting for variation in bony geometry between specimens that may influence dynamic motion pathways. This study examined individual anthropometric measures for significant linear relationships with in vivo kinematic and kinetic performance and determined their implications for robotic studies. Anthropometrics and 3D motion during a 31 cm drop vertical jump task were collected in high school female basketball players. Anthropometric measures demonstrated differential statistical significance in linear regression models relative to kinematic variables (P-range < ). However, none of the anthropometric relationships accounted for clinical variance or provided substantive univariate accuracy needed for clinical prediction algorithms (r 2 < 0.20). Mass and BMI demonstrated models that were significant (P < 0.05) and predictive (r 2 > 0.20) relative to peak flexion moment, peak adduction moment, flexion moment range, abduction moment range, and internal rotation moment range. The current findings indicate that anthropometric measures are less associated with kinematics than with kinetics. Relative to the robotic manipulation of cadaveric limbs, the results do not support the need to normalize kinematic rotations relative to specimen dimensions. INTRODUCTION In the United States an estimated 250,000 people sustain anterior cruciate ligament (ACL) injuries each year and over 125,000 ACL reconstructions are performed to repair these injuries. 55 Conservative cost estimates place the corresponding expense to surgically reconstruct and rehabilitate these injuries in excess of $1 billion annually. 58 Yet despite these costs, the long term outlook of ACL reconstruction is not encouraging. Greater than 50% of patients with ACL 81

90 reconstructions will experience early onset osteoarthritis within 10 years post-surgery. 66 In studies of both male and female soccer athletes years after ACL reconstructions, 75-84% of patients reported deterioration in knee quality of life, which corresponds with the predicted onset of osteoarthritis. 63,64 Furthermore, patients with ACL reconstructions are far more likely to suffer another ACL injury than healthy controls, as 13% of reconstructed athletes incur a secondary ACL injury, while only 1% of their healthy counterparts incur a primary ACL injury This rate was greater in an athlete-specific population that returned to sport following ACL reconstruction, as epidemiologic data within the first year of return identified secondary injury rate to be as high as 24%. 196 One rationale for these bleak post-reconstruction outcomes is that researchers lack detailed knowledge of the functional intra-articular biomechanics of the native ACL and ACL reconstructions during athletic activities. Recently, robotic studies have begun to quantify the biomechanical contributions of both the intact ACL and various ACL reconstructions. These studies utilized anterior tibial translation, internal tibial rotation, and valgus torque at the knee to simulate Lachman s and Pivot-Shift clinical exams. Through these investigations it was shown that though ACL grafts improve joint function compared to the ACL-deficient condition, grafts do not match the mechanical restraint properties demonstrated by the intact ACL. 24,29 Such disparity in biomechanical function between the reconstructed and intact conditions may be the source of joint degeneration and increased injury rates following ACL reconstruction. Due to these shortcomings in restoring biomechanics, longterm patient satisfaction, and injury rate disparity, ACL reconstructions are currently clinically inferior to intact conditions and encourage research related to the prevention of ligament injuries. One mechanism that has been used for the evaluation of ACL reconstruction mechanics is robotic-driven in vitro investigation. 73,74,78,146 Such studies allow researchers to garner valuable 82

91 mechanical data from cadaveric and animal models utilizing invasive methods that would be impossible to execute in vivo. While some in vitro models use simulated impact 86 or the path of least mechanical resistance 146 to articulate joints, others utilize kinematics recorded from 3D motion systems to define position-controlled joint articulations. 73,74 In animal models it would be feasible to record in vivo kinematics, sacrifice the limb, and then use the subject-specific kinematics as input to constrain the joint position. However, in human models, this practice is impossible. Therefore, the kinematic input applied to a cadaveric model must be derived from a secondary, living athlete. 3D kinematic reliability has been documented within and between subjects performing the same athletic task. 179 Between-subject kinematic reliability is lower than within-subject reliability; therefore, the introduction of kinematics recorded from one subject onto a cadaveric limb from as second subject may introduce errors in joint articulation. Due to biologic variability, it is unlikely that a cadaveric specimen and in vivo motion subject share identical anatomical geometry. Therefore, it would be useful to understand if differences in kinematic performance could be predicted relative to anthropometric properties such as height and mass. If these associations between basic anthropometric measures and kinematic performance were identified, then in vivo kinematics could be scaled relative to the size of each cadaveric specimen prior to their inclusion in simulation models. Any specimen-specific normalization applied to cadaveric simulations is likely decrease inter-specimen variability and strengthen the power of findings. The purpose of this study was to examine individual anthropometric measures for significant and clinically predictive linear relationships with kinematic and kinetic performance during a drop vertical jump (DVJ). The hypothesis tested was that anthropometric measures 83

92 would not impact the magnitude of kinematic joint rotations observed between subjects, but would impact kinetics. METHODS Participants in the current study consisted of a cohort of 239 middle and high school female basketball athletes (mass = 55.4 ± 13.2 kg, height = 1.60 ± 0.09 m, tibia length = 0.31 ± 0.03 m, BMI = 21.3 ± 3.9, age = 13.6 ± 1.6 years) from a prospective, longitudinal study. Female athletes were selected as the study population because they experience ACL injuries at 4 to 6 times the rate of their male counterparts. 46 Testing procedures were approved by the institutional review board and informed, written consent was obtained from the parent or legal guardian of each subject. Each subject also provided consent prior to participation. Participants were evaluated for anthropometric measures prior to motion testing. A stadiometer was used to measure height with subjects standing barefoot. A calibrated physician s scale was used to measure body mass again with subjects standing barefoot. Participants were also measured for shoe size as footwear for motion testing was provided to them. Subjects were instrumented with 43 retro-reflective markers for 3D biomechanical analysis. Markers were arranged in a modified Helen Hayes format that has been previously described. 83 Motion data was collected and sampled at 240 Hz with a 10 camera motion analysis system (Eagle cameras, Motion Analysis Corporation, Santa Rosa, CA). Ground reaction forces (GRF) were collected by dual, in-ground, multi-axis force platforms (BP600900, AMTI, Watertown, MA) and sampled at 1200 Hz. Prior to dynamic motion testing a static standing trial was collected for each subject to define body segments, dimensions, and neutral alignment. All joint angles were reported in reference to this neutral alignment. 84

93 Each participant performed three DVJ trials starting from a 31 cm box. 83,179 Motion was recorded for each trial and all successful trials from a subject were averaged into an individual mean. A trial was deemed successful if the subject left the initial box simultaneously with both feet and landed on the force platforms simultaneously with each foot entirely contained within separate plates. Trials that did not match these criteria were excluded and the subject average was taken from the remaining trials. Contact phase motion data were processed in Visual3D (version 4.0, C-Motion, Inc., Germantown, MD) with custom MATLAB code (version 2012a, The MathWorks, Inc, Natick, MA). Visual3D used the relative positions of retroreflective markers to define each body segment as a rigid body with length and volume, while an internal biomechanical model assigned segment mass based on a percentage of the subject s bodyweight. Data processing methods for segment definition as well as joint angle calculation followed previously specified conventions. 179 Only the first contact phase of the DVJ was analyzed. Contact phase was identified with GRF data and defined as the point of initial contact (IC) through toe-off. IC was defined as the first point where the ground reaction force exceeded 10 N, whereas toe-off was defined as the first point after IC where the ground reaction force was below 10 N. Marker trajectories and GRF data were processed through a fourth-order, low-pass, digital filter with a cutoff frequency of 12 Hz for kinetic and kinematic calculations. All moments were reported as external joint moments derived from the ground reaction forces created during contact with the force platforms. From the data collection process, four anthropometric (height, weight, body mass index (BMI), and tibia length) measures were selected as independent variables. These variables were selected for analysis as they represent easily identifiable anthropometrics that have previously been incorporated in multi-factorial ACL injury risk assessments. 77,197 Variables were also 85

94 selected relative to the anthropometric data that is typically included with cadaveric specimen procurement. Twelve kinematic (knee flexion, abduction, and internal angle at IC; maximum abduction, adduction, internal, external, flexion, and minimum flexion angle; and flexion, abduction, and internal rotation ranges of motion (ROM)) and 12 kinetic (knee flexion, abduction, and internal moment at IC; maximum abduction, adduction, internal, external, flexion, and minimum flexion moment; and flexion, abduction, and internal moment ROM) knee measures were then selected as dependent variables. Peak values and ROMs were selected as dependent variables for each rotational degree of freedom as metrics of this nature are associated with and used to predict ACL injury risk; and therefore, should be relevant to robotic simulations that wish to model ACL loading. 46,77 Each of the four independent variables was individually correlated with each of the 24 dependent variables, through single-factor linear regression models. An alpha level of 0.05 was used to judge statistical significance in all models. A 0.2 linear regression r 2 cutoff criterion was then used to determine if a significant model was correlated enough to be considered predictive. 46 Following univariate linear regression, these same criteria were used to evaluate multivariate linear regression models that correlate all four independent variables collectively against each of the 24 dependent variables. All statistics with were performed in MATLAB with the Statistical Toolbox and custom code and verified through SPSS (version 21, IBM Corp, Armonk, NY).. RESULTS Kinematics Linear model relationships between individual anthropometric independent variables and kinematic dependent variables were significant (P < 0.05) in 27 out of the 48 variable combinations (56%). The height, BMI, and tibia length variables produced the greatest number of significant models relative to kinematic variables at seven each. Of the 26 significant linear 86

95 models, no model reached the 0.2 linear regression r 2 cutoff criterion to be considered a predictive model (Table 5.1). The largest r 2 for kinematic models was mass correlated with minimum flexion angle (r 2 = 0.17; Figure 5.1). All but three of the significant linear models, which were all relative to maximum flexion angle, displayed an r 2 < Linear relationships were significant in all multivariate models apart from flexion angle at IC and flexion angle ROM. For all 24 dependent kinematic variables, the multivariate models universally exhibited larger r 2 values than the corresponding univariate models. However, the only model that reached the predictive cutoff criterion was for minimum flexion angle (r 2 = 0.20). Table 5.1: Depicts the r 2 values for linear relationships between the four independent variables (columns) and 12 kinematic dependent variables (rows). Dependent Variable ( ) Height Mass BMI Tibia Length Multivariate Flexion IC * 0.03* Abduction IC 0.06* * 0.11* Internal IC 0.04* 0.04* 0.02* 0.05* 0.07* Max Flexion Angle 0.06* 0.08* 0.06* 0.02* 0.09* Min Flexion Angle 0.12* 0.17* 0.12* 0.02* 0.20* Max Abduction Angle 0.09* * 0.16* Max Adduction Angle 0.06* * 0.05* 0.15* Max Internal Angle * 0.05* * Max External Angle * 0.03* * Flexion Angle ROM Abduction Angle ROM 0.06* * 0.08* Internal Angle ROM * * Indicates that the linear regression model was significant (P 0.05) 87

96 Figure 5.1: Data plot of subject mass versus minimum knee flexion angle. This plot exemplifies data where the factors exhibited a significant linear relationship, but a poor r 2 correlation. Kinetics Linear model relationships between individual anthropometric independent variables and kinetic dependent variables were significant (P < 0.05) in 39 out of 48 cases (81%). Height, mass, and BMI produced the greatest number of significant models relative to kinetic variables with 10 each. Of the 39 significant models, 16 exceeded the 0.2 linear regression r 2 cutoff criterion to be considered a clinically predictive model (Table 5.2). The predictive linear models were height, mass, BMI, and tibia length relative to maximum flexion moment; mass and BMI relative to maximum extension moment; height, mass, BMI, and tibia length relative to the range of flexion moment values; height, mass, and BMI relative to the range of abduction moment values; as well as height, mass, and BMI relative to the range of internal moment values. The 88

97 most predictive linear models were mass correlated with the knee flexion moment range and maximum flexion moment values (r 2 = 0.63 and 0.68, respectively; Figure 5.2). BMI correlated with knee flexion moment range and maximum flexion moment had r 2 > 0.44, whereas the remaining predictive models demonstrated r 2 values between 0.22 and The rates of change for all between independent and dependent variables for all predictive models are documented in Table 5.3. As with the kinematic dependent variables, multivariate regression models exhibited greater r 2 values than univariate models for all 24 kinetic variables. However, in all cases these increases were only marginally larger than the univariate model based on mass (r 2 difference 0.04). The multivariate models for maximum knee flexion moment, maximum knee extension moment, flexion moment range, abduction moment range, and internal moment range all exceeded the predictive r 2 threshold. Table 5.2: Depicts the r 2 values for linear relationships between the four independent variables (columns) and 12 kinetic dependent variables (rows). Dependent Variable (N*m) Height Mass BMI Tibia Length Multivariate Flexion IC 0.02* 0.07* 0.06* * Abduction IC Internal Max Flexion Moment 0.42* 0.63* 0.44* 0.23* 0.66* Max Extension Moment 0.12* 0.32* 0.27* 0.04* 0.34* Max Abduction Moment 0.09* 0.07* 0.03* 0.06* 0.10* Max Adduction Moment 0.07* 0.17* 0.15* 0.05* 0.18* Max Internal Moment 0.10* 0.17* 0.12* 0.05* 0.18* Max External Moment 0.08* 0.05* 0.02* 0.05* 0.09* Flexion Moment Range 0.42* 0.68* 0.48* 0.22* 0.69* Abduction Moment Range 0.27* 0.38* 0.26* 0.19* 0.41* Internal Moment Range 0.25* 0.27* 0.15* 0.14* 0.31* * Indicates that the linear regression model was significant (P 0.05) 89

98 Table 5.3: Depicts the slope of the best fit line (rate of change) between independent and dependent variables that were found to have both a significant linear relationship and an r 2 > Dependent Variable Independent Variable Rate of Change Max Flexion Moment Height N*m / cm Max Flexion Moment Mass N*m / kg Max Flexion Moment BMI N*m / BMI Max Flexion Moment Tibia Length N*m / cm Max Extension Moment Mass N*m / kg Max Extension Moment BMI N*m / BMI Flexion Moment Range Height N*m / cm Flexion Moment Range Mass N*m / kg Flexion Moment Range BMI N*m / BMI Flexion Moment Range Tibia Length N*m / cm Abduction Moment Range Height N*m / cm Abduction Moment Range Mass N*m / kg Abduction Moment Range BMI N*m / BMI Internal Moment Range Height N*m / cm Internal Moment Range Mass N*m / kg Figure 5.2: Data plot of subject mass versus knee flexion moment range. This plot exemplifies data where the factors exhibited a significant linear relationship and a moderately strong r 2 correlation. 90

99 DISCUSSION The objective of the current study was to examine anthropometric measures for significant and clinically-predictive linear relationships with kinematic and kinetic performance during a DVJ. It was believed that the identification of these linear correlations would provide a baseline by which to scale in vivo recorded kinematics relative to the geometry of each cadaveric specimen. Such specimen-specific considerations would optimize robotic simulation models through a likely reduction in variability. Though 65 of the individual linear models examined were found to represent significant relationships between their respective independent and dependent variables, poor r 2 values indicated that most of these associations did not account for enough of the variance to serve as clinically predictive models. Our lab has previously documented that in order for a variable to be considered predictive of kinematic and kinetic outcomes the coefficient of determination must account for greater than 20% of the error (r 2 0.2). 46 Congruent with this range of significance, Fleming et al 198 concluded that his isometer model could not accurately predict ACL reconstruction tension from an r 2 value of Of the 65 significant linear models found in the present study, 49 fell below the 0.2 threshold and therefore were not considered clinically predictive. No anthropometric variables were found to be clinically predictive of any kinematic factors. As anthropometric measures were not able to accurately predict changes in kinematics, it may be inappropriate to adjust rotational kinematic input relative to specimen size in cadaveric studies. Though 16 linear models exceeded the necessary threshold to be considered predictive, the presently reported r 2 values were below what is typically observed in the literature. Many documented clinical predictors of various knee behaviors account for greater than 70% of the 91

100 variance in the model. Prior reports predicted ACL injury with r 2 = 0.88, 46 while knee joint forces after total knee replacement have been predicted with r 2 between , 199 and lower leg motion in response to electrical stimulation has been predicted with r 2 between As none of the predictive linear models identified in this study exceeded r 2 = 0.70, they should be considered moderate to weak predictors. The presented models inability to account for variance in prediction may be a result of their single independent variable format. Previously developed ACL injury risk prediction nomograms rely upon five 77,197 and variables to derive their calculations with high sensitivity and specificity. Therefore, the weak r 2 values in the current study may indicate that models based on single independent variables may be too oversimplified to accurately predict kinematic and kinetic values. Similarly, the presented anthropometric multivariate regression models also failed to be significantly predictive of kinematic performance. In multivariate models used to predict ACL injury risk for adolescent athletes, additional intrinsic variables, such as hamstrings:quadriceps activation ratio, not based on anthropometric measures are incorporated to increase predictive accuracy. 77 Though hamstrings:quadriceps strength ratio might contribute significantly to kinematic prediction, as it does to ACL injury risk prediction, that data cannot be obtained from a cadaveric specimen and would be useless to the optimization of robotically-driven joint simulations. Unfortunately, the linear models that demonstrated both significant linear relationships and moderate predictive abilities failed to correspond with the kinematic and kinetic knee variables associated with ACL injury risk. Knee abduction angle and moments 46,75,197 and internal tibial rotation 9,47 have been associated with or used to predict likelihood of ACL injury. The moderately predictive models in this study were only relative to peak and minimum flexion moment, range of flexion moment values, range of abduction moment values, and range of 92

101 internal moment values. Sagittal plane moments were the most predictive models in the present study, but previous investigations have identified that sagittal plane moments alone do not threaten ACL integrity during the performance of athletic tasks. 202 As clinical investigations have not identified such variables as key contributors to ACL injury, the moderately predictive linear models identified in the current study do not offer much benefit for the optimization of robotically-driven cadaveric knee models relative to ACL biomechanics. Independent anthropometric variables were found to have more significant associations with kinetics than kinematics. The relative significance of these variables to kinetic prediction adheres to mechanical principles as moments are the product of force and distance. An extended tibia length should rapidly increase knee moments as it would move the times bodyweight GRFs incurred on a single leg when landing from a DVJ further away from the knee. 83 Similarly, increased subject mass will also increase the raw values of landing GRFs as they are relative to bodyweight. This finding echoes the weighted structure of the previously reported ACL injury risk prediction nomogram. 77,197 In that nomogram, tibial length has the highest weight and can account for up to 100 points of the total injury prediction score for each subject. A high tibial length score alone can place an athlete with a 0.75 probability of high knee moment. Pennation angles and lines-of-action for the major muscle groups surrounding the knee primarily correspond to the generation of flexion/extension torque and anterior/posterior loading at the joint. 114,122,203 As such, during neuromuscular-controlled in vivo tasks, the sagittal plane has been shown to experience a greater range of articulation and torque as it stabilizes against a greater proportion of ground reaction force than any other plane of motion at the knee. 204 Therefore, it is unsurprising that linear regression models developed from anthropometric 93

102 measures, especially mass, correlated best with peak sagittal plane torques. Larger mass should increase the mechanical demand at the knee. Unless there is a loss of neuromuscular control, load increases should mostly be countered through sagittal plane loading which leads to prevalent flexion/extension correlations. Correspondingly, the frontal and transverse planes, which support a lower proportion of knee loading, 204 were less correlated with changes in anthropometric variables. In conclusion, both univariate and multivariate linear models derived from simple anthropometric variables may not be sufficiently robust to be instituted as clinical predictors of kinematic performance during a DVJ. More complex, multivariate nomograms are necessary to accurately identify kinematic behaviors. Relative to cadaveric-based orthopaedic research, the current findings indicate that kinematics are not correlated with stature, and therefore provide no current evidence that input kinematic rotations should be normalized to specimen anthropometrics during simulations. ACKNOWLEDGEMENTS This work was supported by NIH grants R01-AR049735, R01-AR055563, and R01- AR The authors thank the entire Sports Medicine Biodynamics groups at Cincinnati Children s Hospital and The Ohio State University for their support. CONFLICT OF INTEREST STATEMENT There were no conflicts of interest to report in the preparation of this manuscript. 94

103 Chapter 6 The Effect of Internal, Abduction, and Combined Tibial Rotations on Anterior Cruciate Ligament and Medial Collateral Ligament Biomechanics at Initial Contact During Simulated Jump Landing and Sidestep Cutting Tasks Nathaniel A. Bates, a,b,c Rebecca J. Nesbitt, a Jason T. Shearn, a Gregory D. Myer, b,d,e,f Timothy E. Hewett a,b,c,d,g a Department of Biomedical Engineering, University of Cincinnati, Cincinnati, OH, USA b Sports Medicine Biodynamics Center, Division of Sports Cincinnati Children s Hospital Medical Center, Cincinnati, OH, USA c The Sports Health and Performance Institute, The Ohio State University, Columbus, OH, USA d Department of Pediatrics, College of Medicine, University of Cincinnati, Cincinnati, OH, USA e Department of Orthopaedic Surgery, College of Medicine, University of Cincinnati, Cincinnati, OH, USA f The Micheli Center for Sports Injury Prevention, Boston, MA g Departments of Physiology and Cell Biology, Orthopaedic Surgery, Family Medicine, and Biomedical Engineering, The Ohio State University, Columbus, OH, USA This manuscript is currently in preparation for submission to the Journal of Biomechanics. 95

104 ABSTRACT The ACL serves as a secondary restraint to multiple planes of rotation at the knee. However, literature on how these rotations contribute to ACL strain and ligament injury is conflicting. The purpose of this study was to apply robotically-controlled kinematic stimuli to joint orientations derived from in vivo recorded athletic activities determine their effects on knee ligament biomechanics. A 6-degree-of-freedom robotic manipulator was used to position 17 cadaveric specimens into orientations that mimic initial contact recorded from in vivo drop vertical jumps and sidestep cutting activities. 4 rotational perturbations were applied in the frontal, transverse, and combined planes as ACL and MCL strain was documented. Combined abduction/internal rotations produced the greatest ACL strains (absolute strain = 4.7%; strain change from neutral = 1.8%), while isolated abduction produced the greatest MCL strains (absolute = 2.4%; change = 1.8%). Internal rotations did not significantly alter strain in the ACL (absolute = 3.6%; change = 0.5%) or MCL (absolute = 0.7%; change = 0.1%) relative to the neutral limb orientation. Strain generated in the ACL was typically larger than the in the MCL for most of the applied rotational perturbations (P < 0.05). Therefore, reduction of knee valgus during athletic tasks should be a primary focus of ACL injury prevention programs as it is the greatest rotational contributor to ACL strain. Reduced strain response in the MCL versus the ACL contributes to the understanding of why only 30% of ACL ruptures exhibit concomitant MCL injuries. INTRODUCTION In the knee, the anterior cruciate ligament (ACL) serves as the primary soft tissue restraint to anterior tibial translation and as a secondary restraint to motion in additional degrees of freedom, while the medial collateral ligament (MCL) primarily resist knee abduction. 1,7,181 96

105 Following an ACL injury this motion restraint is lost and produces joint instability. 30,205 In order to return to sport, athletes often require ACL reconstruction surgery. With a conservative estimate of $17,000 per surgery and an estimated 127,000 ACL injures in the United States annually, there is a significant monetary investment in the repair and rehabilitation of ACL ruptures. 55,58 Despite these costs, the long-term prognosis following ACL reconstructions is early onset of osteoarthritis and degradation of knee quality of life within 15 years post operative. 63,64 Within the past 20 years, robotic methods of biomechanical joint articulation have allowed investigators to examine the underlying mechanical behaviors within a joint through the simulation of activities of daily living on cadaveric specimens. 1,73,74,181 These in vitro investigations have gathered data that would be unobtainable in vivo due to the invasive nature of biomechanical testing. Much of this work has evaluated and improved the efficaciousness of surgical methods and graft material selections used in ACL reconstructions with respect to native ACL biomechanics. 1,11,13,14,16,21,24,27,29,134,135,137 These simulations have also gathered data on the mechanical contributions of the intact ACL in response to Lachman s and pivot shift tests, 11,13,14,16,21,24,27,29,134,135,137 gait cycles, 74,181 and landing impact forces ,122,206,207 The accrued data has advanced the knowledge of ACL function and limitation as well as provided a baseline of comparison by which to evaluate the effectiveness of ACL injury treatments. One goal of in vitro knee simulations has been to identify the primary mechanical precursors to ACL injury. Unfortunately, data extracted from various methods of mechanical knee simulations has not always been congruent. Multiple simulation methods have documented that combined knee abduction and internal rotations place the greatest mechanical demand on the ACL. 85,106,208 However, simulations driven by robotic manipulators capable of force-torque and position control have demonstrated that knee abduction has greater impact on the mechanical 97

106 loading of the ACL than internal tibial rotation (Chapter 3); whereas, some simulations driven by pulse loads and torque transformers have indicated the opposite. 84,85 This disassociation may be caused by fundamental differences in simulation methods that lead to the varying degrees of physiological accuracy documented in Chapter 3. It is important to accurately identify the underlying contributors to ACL loading during physiologic tasks as this knowledge could be extrapolated to improve the efficacy of injury prevention and rehabilitation techniques. The purpose of this study was to apply robotically-controlled kinematic stimuli to joint orientations derived from in vivo recorded athletic activities in order to determine the effects of abduction, internal, and combined rotations on knee ligament biomechanics. It was hypothesized that additional rotation would increase ACL and MCL strain relative to the normal condition, that additional internal rotation would have minimal impact on ACL strain, and that combined abduction and internal rotation would increase ligament strain more than either individual contribution. METHOD Experimental Design 17 lower extremity cadaveric specimens from 11 unique subjects (age = 47.6 ± 7.3 years, mass = 829 ± 199 N) were acquired from and anatomical donations program (Anatomical Gifts Registry, Hanover, MD) and completed testing in this study. An additional four specimens were excluded due to specimen failure, pre-existing ACL damage, or a non-functional ACL. Specimens were randomized into either an ACL (N = 9, age = 47.3 ± 8.1 years, mass = 838 ± 216 N) or MCL group (N = 8, age = 47.5 ± 8.3 years, mass = 853 ± 197 N), ensuring that contralateral pairs were separated. Using a six-degree-of-freedom (6-DOF) robotic manipulator (KR210; KUKA Robotics Corp., Clinton Township, MI) mounted with a six-axis load cell 98

107 (Theta Model; ATI Industrial Automation, Apex, NC), the limbs were positioned in orientations associated with initial ground contact during athletic tasks. These orientations were derived from three-dimensional (3D), in vivo motion capture kinematics. From this the baseline orientation, each limb was articulated through rotational kinematics that have been associated with ACL injury risk or ACL deficiency (ACLD). Specimens were run through this simulation protocol in the intact condition and then in an isolated ACL or isolated MCL condition, dependent on group randomization. During testing the load cell recorded joint forces and torques while 3 mm microminiature differential variable strain transducers (DVRT, LORD MicroStrain Inc., Williston, VT) recorded ACL and MCL strain. Kinematic Model The acquisition of 3D motion data and its conversion into input for in vitro robotic simulations has been previously documented in Chapter Briefly, 3D motion capture was performed at 240 Hz by a 10-camera system (Eagle Cameras, Motion Analysis Corp, Santa Rosa, CA) on a matched male (age = 24 years; height = 175 cm; mass = 675 N) and female (age = 25; height = 170 cm; mass = 632 N) subject. This data was then filtered at 6 Hz and processed through an established biomechanical model in Visual3D (version 4.0, C-Motion, Inc., Germantown, MD). 179 It should be noted that the male and female subjects used for model development were respectively classified as low and high risk for ACL injury based on the peak knee abduction moment calculated in Visual3D from their drop vertical jump (DVJ) kinematics. 46,209 The resultant kinematics were mathematically adjusted to minimize the impact of skin-marker artifact errors during the execution of robotic simulations as described in Chapter 4. 99

108 Specimen Preparation An explicit account of specimen preparation procedures was previously documented in Chapter Specimen criteria were defined as no previous history of knee trauma, knee surgery, bone cancer, or ankle or shin implants. The limbs were kept frozen at -20 C and allowed to thaw the day prior to testing. The specimen was resected of all soft tissue down to the knee joint capsule, leaving the collateral and cruciate ligaments and menisci intact. Anatomical landmarks were marked and used to define the tibial joint coordinate system. 178 Using this system, custom biomechanical fixtures were affixed around the tibia, which was then rigidly mounted to the load cell on the robot end effector such that the tibia, load cell, and robot axes were all aligned. The tibial joint center point was digitized with a coordinate measuring machine (Faro Digitizer F04L2, FARO Technologies Inc., Lake Mary, FL) and all rotations, translations, forces, and torques were applied or recorded about this point, respectively. Mounted specimens were articulated to 45, as the ACL is likely minimally loaded in this position, 114 and DVRTs were implanted on the ACL and MCL using previously described techniques. 206,210 For the initial position of each simulated task, the in vitro limb orientation of all three rotational DOFs was verified to be within 0.5 of the initial contact limb orientation recorded in vivo. At this point the limbs were loaded incrementally and cycled through simulations until a peak force of bodyweights was attained for DVJ tasks. This force range represents the peak single leg vertical ground reaction force that is generated in vivo when landing from a DVJ. 83 Similarly, a peak force of bodyweights was attained for sidestep cut tasks. Robot Simulation All tests were performed at room temperature and the joint was consistently hydrated with saline. Initial contact orientation for four athletic tasks (male DVJ, male sidestep cut, female 100

109 DVJ, female sidestep cut) were simulated on each specimen in a randomized order. The initial position was used as a starting orientation to cycle a series of rotational adjustments (±4 isolated knee abduction, ±4 isolated internal tibial rotation, ±4 combined abduction and internal rotations). These kinematic adjustments were selected because they represent DOFs where the ACL resists knee motion and are associated with either ACL injury risk or ACLD. 1,20,46,86,197,208,211 A ±4 rotational shift will create an 8 range for knee abduction angles at initial contact, which has been prospectively reported as the mean difference between athletes who went on to ACL injury and healthy controls. 46 A ±4 change in internal rotation is representative of the additional tibial motion observed in ACL deficient subjects during gait. 20 Finally, valgus collapse, defined as the combination of knee abduction and internal rotation of the tibia, 158 is often cited in most analyses of non-contact ACL ruptures These rotational adjustments were cycled from initial contact orientation because pilot testing revealed that is where peak ACL strain is most likely to occur and injury synopses suggest that ACL ruptures occur immediately after initial contact in a position of limited flexion. 215 This description is best matched by our initial contact orientations, all of which expressed between of knee flexion. To minimize viscoelastic effects, 10 preconditioning cycles were simulated prior to the 10-cycle set where forces, torques, and strains were recorded. After all cycling was completed, the specimen was manually articulated to the initial position for the next task and the process was repeated. Once all simulations were performed on the intact knee specimen, it resected of all soft tissue and the distal portions of the femoral condyles such that the only load-bearing structure remaining in the knee was the ACL or MCL. The remaining ligament was dependent on specimen group randomization. All simulations were repeated in this isolated ligament condition (Figure 6.1). Following simulation, the joint was returned to initial contact orientation, 101

110 compressed to an unloaded position, and slowly distracted to identify the neutral strain position of the ligament. With the isolated ligament as the only intact load-bearing structure, neutral strain was identified when the force sensors first registered a constant distraction force. The remaining ligament was then resected and all simulations were repeated in a bone-only condition. Figure 6.1: Specimens displayed in the intact knee (A), ACL-isolated (B), and MCL-isolated conditions. The specimens displayed are right limbs positioned in the initial contact orientation for a male DVJ task. Data Analysis Having identified the neutral strain inflection point of the ACL and MCL, strains were reported as absolute values rather than changes relative to the DVRT insertion orientation. This eliminated a limitation typical of in vitro ligament strain assessments ,206 The 8 th and 9 th cycles of each 10 cycle set were used for analysis in order to eliminate cycle effects. All data 102

111 points were time normalized to percentage of landing phase for each task. A 2 x 3 ANOVA with a least significant difference post-hoc analysis was used to evaluate differences in ligament strain and joint loading between gender, ligament condition, or motion task and rotation type. Significance was determined at α < All statistical analyses were performed in SPSS (version 21, IBM Corp, Armonk, NY). RESULTS For ACLs in the intact knees, isolated external rotation consistently produced the least absolute ligament strain, followed by combined adduction/external, isolated internal, isolated adduction, isolated abduction, and finally combined abduction/internal rotation (Table 6.1). However, the differences between each increment were not statistically significant (P > 0.05). Combined abduction/internal rotation was the only adjustment that produced a statistically significant difference in absolute ACL strain, from the neutral limb alignment, for either motion task or gender (P < 0.05). External rotation was the only adjustment that resulted in a smaller mean ACL strain than the neutral limb alignment. This lead to statistically significant differences between absolute ACL strain generated from external rotation and abduction or combined abduction/internal rotations for both the DVJ and sidestep cut (P < 0.05). Other than external rotation, change in ACL strain relative to the neutral position exhibited no statistically significant differences between any two DOFs (P > 0.05; Figure 6.2 & 6.3). However, the mean change induced by internal rotation was on average 1.01% less than abduction rotation, which was 0.26% less than combined abduction/internal rotation. Absolute MCL strain in the intact knee relative to neutral position, isolated external, isolated internal, isolated adduction, combined abduction/internal, and combined adduction/external rotations were smaller than ACL strains for the same conditions (P < 0.05). 103

112 Within the MCL, isolated abduction and combined abduction/internal rotations produced significantly greater absolute strains than any other rotational degree of freedom (P < 0.05; Table 6.2). Similarly, the magnitude of change in MCL strain relative to the neutral position was significantly greater for isolated abduction and combined abduction/internal than for any other rotation (P < 0.05). Isolated adduction and combined adduction/external DOFs completely unloaded the MCL as they produced negative absolute strain values. When specimens were resected down to the MCL-isolated condition, the MCL remained unstrained throughout most rotational stimuli, with the exception of combined abduction/internal rotation. For both the DVJ and sidestep cut, no significant gender differences were noted in either the ACL or MCL for both the intact and isolated conditions (P > 0.37). Similarly, there were no significant differences between motion type for either the ACL or MCL (P > 0.16). There were no significant interactions between gender and rotational stimulus, motion task and rotational stimulus, or ligament condition and rotational stimulus (P > 0.05). 104

113 Table 6.1: Mean ACL ligament strains recorded for the intact knee and isolated ligament condition in response to rotational stimuli. Absolute strain is defined as a percentage of change in the DVRT sensor length relative to the zero strain length established for each ligament. Intact Male DVJ Female DVJ Male Cut Female Cut Neutral 3.1 f,* 3.4 * External 2.7 d,f,* 3.0 d,f,* 2.3 d,f 2.4 d,f Internal 3.7 * 4.0 * 3.2 * 3.7 * Abduction 4.6 b 5.0 b,* 4.1 b 4.7 b Adduction 3.8 * 4.2 * 3.7 * 3.9 * Isolated Ab/Int 4.9 a,b,g,* 5.1 a,b,g,* 4.2 b 4.8 a,b,* Add/Ext 3.1 f,* 3.5 f,* 3.0 * 3.1 * Neutral * 3.2 * External * Internal * Abduction Adduction 1.9 * 2.0 * 3.0 * 2.9 * Ab/Int Add/Ext 1.9 * 1.8 * 3.1 * 2.6 * a = significant difference from neutral, b = significant difference from external, c = significant difference from internal, d = significance difference from abduction, e = significant difference from adduction, f = significant difference from combined ab/int, g = significant difference from combined add/ext, * = significant difference between comparable MCL measure 105

114 Table 6.2: Mean MCL ligament strains recorded for the intact knee and isolated ligament condition in response to rotational stimuli. Absolute strain is defined as a percentage of change in the DVRT sensor length relative to the zero strain length established for each ligament. Intact Male DVJ Female DVJ Male Cut Female Cut Neutral 0.6 d,f,* 0.3 d,f,* 0.6 d,f 0.7 d,f External 0.6 d,f,* 0.3 d,f,* 0.6 d,f 0.9 d,f Internal 0.7 d,f,* 0.4 d,f,* 0.8 d,f,* 1.0 d,* Abduction 2.5 a,b,c,e,g, 2.1 a,b,c,e,g,* 2.5 a,b,c,e,g 2.5 a,b,c,e,g Adduction -0.2 d,f,* -0.6 d,f,* -0.0 d,f,* 0.0 d,f,* Ab/Int 2.4 a,b,c,e,g,* 2.0 a,b,c,e,g,* 2.4 a,b,c,e,g 2.3 a,b,e,g,* Add/Ext -0.5 d,f,* -0.8 d,f,* -0.5 d,f,* -0.3 d,f,* Isolated Neutral -1.8 e,g -1.9 e,g -1.1 e,g,* -2.6 e,g,* External -1.9 e,g -1.4 e,g -0.8 e,g e,g,* -2.6 Internal -1.7 e,g -1.1 e,g -0.6 e,g e,g,* -2.2 Abduction 0.0 e,g 0.0 e,g 0.8 e,g -0.7 e,g Adduction -5.2 a,b,c,d,f * -6.9 a,b,c,d,f,* -5.1 a,b,c,d,f,* a,b,c,d,f,* -6.1 Ab/Int 0.2 e,g 0.2 e,g 1.0 e,g -0.8 e.g Add/Ext -4.6 a,b,c,d,f * -6.3 a,b,c,d,f,* -4.9 a,b,c,d,f,* a,b,c,d,f,* -6.7 a = significant difference from internal, b = significant difference from external, c = significance difference from abduction, d = significant difference from adduction, e = significant difference from combined ab/int, f = significant difference from combined add/ext, * = significant difference between comparable ACL measure 106

115 Figure 6.2: Displays the mean change (relative to the neutral position) in ACL (top) and MCL (bottom) ligament strain generated in the intact knee by each rotational stimuli. For the ACL, the mean strain changes were -0.5% external, 0.5% internal, 1.5% abduction, 0.8% adduction, 1.8% combined abduction/internal, and 0.2% combined adduction/external. For the MCL, the mean strain changes were - 0.0% external, 0.1% internal, 1.8% abduction, -0.8% adduction, 1.8% combined abduction/internal, and - 1.0% combined adduction/external. * indicated significant differences between ligaments. 107

116 Figure 6.3: Displays the mean change (relative to the neutral position) in ACL (top) and MCL (bottom) ligament strain generated in the isolated ligament condition by each rotational stimuli. For the ACL, the mean strain changes were -0.3% external, 0.3% internal, 0.1% abduction, -0.1% adduction, 0.5% combined abduction/internal, and -0.5% combined adduction/external. For the MCL, the mean strain changes were -0.2% external, 0.5% internal, 2.0% abduction, -3.9% adduction, 2.2% combined abduction/internal, and -3.6% combined adduction/external. * indicated significant differences between ligaments. DISCUSSION The purpose of this study was to apply robotically-controlled kinematic stimuli to joint orientations derived from in vivo recorded athletic activities in order to determine the effects of abduction, internal, and combined rotations on knee ligament biomechanics. These rotations were selected because they have been identified as contributors to ACL strain, 84-86,106,206 motions secondarily resisted by the ACL, 1 and (in the case of knee abduction) predictors of ACL injury risk. 216 While most rotational stimuli altered the absolute ACL strain, only combined abduction/internal produced values that were statistically significant from the neutral limb position. This finding supports the hypotheses that combined abduction and internal rotation 108

117 would increase ligament strain more than either individual DOF and that isolated internal rotation would have a limited impact on ACL strain. In previous investigations, combined rotational stimuli have had greater impact on ACL strain than rotations in any isolated DOF. This has been documented in computational models, 106 impact testing, 84,85,206 and roboticmanipulator-driven articulations (as described in Chapter 3). Despite being the only stimulus to statistically deviate from neutral, abduction/internal rotation did not induce absolute strains or magnitudes of change that were statistically different from isolated internal rotation or abduction rotation. This absence of significant differences may be due to limitations of DVRT implantation. Though the gauges were barbed and sutured in place, the ACL is not a rigid structure and allows either end of the sensor some movement within the implantation site. With the relatively small changes that were documented in ACL strain (< 2.0%), increased variation could impede statistical significance. However, from a magnitude standpoint, the mean magnitude of change induced by combined abduction/internal rotations on ACL strain was 239% of internal rotation and 116% of abduction. None of the ACL strain values generated in the present study approached previously documented ligament failure strains. In drop landings simulated on cadaveric specimens, implanted DVRTs reported ACL failure at strains between % (mean 18.7%). 206 These injuries were induced immediately following contact in limbs that were positioned relative to in vivo initial contact orientations. Similarly, as presented in Chapter 4, specimens in the present study were also positioned relative to in vivo initial contact orientations; however, the largest mean ACL strain documented was 5.1%. This value would be expected to increase if the range of applied rotational stimulus were expanded. An 8 increase in abduction rotation at initial contact has been prospectively linked with likelihood of ACL injury. 216 To reproduce that 8 109

118 range without risk of specimen damage, the present investigation deviated by 4 on either side of neutral for each plane of rotation. While this effectively created linear changes in transverse plane rotations, such effect was not seen for frontal plane rotations. Rather, both the abduction and adduction rotational stimuli increased the magnitude of ACL strain relative to neutral orientation, though these changes were not statistically significant. Therefore, any frontal plane deviation from a neutral path throughout motion is likely to increase load on the ACL. Application of an expanded 8 stimulus on either side of the neutral orientation for each plane of rotation would likely increase relative magnitudes of change in ACL strain and identify additional statistical significance between the DOFs. The MCL serves as the primary ligamentous resistor to knee abduction rotation. 1 This was reflected in the present study, as isolated abduction and combined abduction/internal stimuli produced significantly greater MCL strains than all other DOFs. No other rotational stimuli expressed difference from one another or from the neutral orientation. Peak MCL strains during combined abduction/internal rotations were significantly smaller than ACL strains. MCL strains also trended toward smaller values in abduction. As a whole, when statistical differences were identified between ligaments in corresponding DOFs, MCL strains were smaller than the comparable ACL strains. Despite that both ligaments resist the knee abduction rotations that are commonly associated with ACL injury, 1,46,106 MCL failures only occur in approximately 30-40% of non-contact ACL injuries. 88,89,96 The generally reduced response of MCL strain to rotational stimuli as expressed in the current study may help explain why the MCL is often able to withstand loads that produce ACL failure. For most conditions, ligament strain for both the ACL and MCL decreased from the intact knee to isolated ligament conditions. This likely indicates that a portion of the strain in 110

119 each ligament is derived from interaction with the other intra-articular structures of the knee. With these structures removed, the ligaments are able to slacken as they travel a direct path between origin and insertion site. Secondly, the neutral orientation was selected relative to initial contact during in vivo athletic tasks as noted in Chapter 4. As such, the joint was in a compressed state during simulation. When the articulating surfaces that supported this compression were removed, the loads manifested in a reduced distance between the tibial and femoral insertion sites that allowed each ligament to slacken. For the MCL, this lead to negative strain that represented an unloaded condition throughout isolated ligament simulations. A limitation of the current investigation is that the applied stimuli are non-physiologic. Rarely in dynamic activity is rotation strictly confined to a single plane of motion. However, like clinical exams such as the Lachman s test, the current motions can still prove valuable in the assessment of functional ligament mechanics. As noted in Chapter 3, numerous investigators have used robotically-simulated Lachman s tests of pure anterior drawer to quantify ACL mechanics in the same fashion that current study evaluates rotational perturbations. The presented rotations could be applied to ACL reconstructed knees to identify their mechanical deficiencies from the native ACL at initial contact orientations. When simulated from initial contact orientations associated with in vivo athletic tasks, combined abduction/internal rotations had the greatest influence on ACL strain. Isolated abduction produced larger magnitudes of change than isolated internal rotation, though the differences were not statistically significant. Therefore, reduction of knee valgus during athletic tasks should continue to be a primary focus of ACL injury prevention programs. Similarly, abduction had the greatest influence on MCL strain. However, the MCL generally expressed less 111

120 absolute strain than the ACL in the presented orientations, which may be indicative of how and why ACL failure often occurs without concomitant MCL injury. ACKNOWLEDGEMENTS This work was supported by the National Institutes of Health/NIAMS Grants #R01- AR049735, #R01-AR05563, #R01-AR and #R01-AR The authors would also like to acknowledge the support of the staff at the Sports Health and Performance Institute at The Ohio State University and the Sports Medicine Biodynamics Laboratory at Cincinnati Children s Hospital. CONFLICT OF INTEREST There were no conflicts of interest in the preparation of this manuscript. 112

121 Chapter 7 Relative Strain in Anterior Cruciate Ligament and Medial Collateral Ligament During Simulated Jump Landing and Sidestep Cutting Tasks: Implications for Injury Risk Nathaniel A. Bates, a,b,c Rebecca J. Nesbitt, a Jason T. Shearn, a Gregory D. Myer, b,d,e,f Timothy E. Hewett a,b,c,d,g a Department of Biomedical Engineering, University of Cincinnati, Cincinnati, OH, USA b Sports Medicine Biodynamics Center, Division of Sports Cincinnati Children s Hospital Medical Center, Cincinnati, OH, USA c The Sports Health and Performance Institute, The Ohio State University, Columbus, OH, USA d Department of Pediatrics, College of Medicine, University of Cincinnati, Cincinnati, OH, USA e Department of Orthopaedic Surgery, College of Medicine, University of Cincinnati, Cincinnati, OH, USA f The Micheli Center for Sports Injury Prevention, Boston, MA g Departments of Physiology and Cell Biology, Orthopaedic Surgery, Family Medicine, and Biomedical Engineering, The Ohio State University, Columbus, OH, USA This manuscript is currently in preparation for submission to the American Journal of Sports Medicine. 113

122 ABSTRACT Background: The medial collateral (MCL) and anterior cruciate ligaments (ACL) are primary and secondary ligamentous restraints against knee abduction, respectively. Knee abduction is an ACL injury risk predictor and component of valgus collapse, which is often observed during ACL rupture. Despite this, MCL ruptures only occur concomitantly in 30-40% of ACL injuries. Hypothesis/Purpose: The purpose of this investigation was to understand how athletic tasks load the knee joint in a manner that drives the ACL toward failure without concomitant MCL failure. It was hypothesized that the ACL would provide greater overall contribution to intact knee forces than the MCL during simulated motion tasks. It was further hypothesized that the ACL would express greater relative peak strain than the MCL during simulated motion tasks. Study Design: Controlled laboratory study. Methods: A 6-degree-of-freedom robotic manipulator articulated 18 cadaveric knees through simulations of kinematics recorded from in vivo athletic tasks. Specimens were articulated in the intact-knee and isolated-ligament conditions. Following simulation, a servohydraulic uniaxial system tensioned each ACL and MCL to failure along their fiber orientations. Results: During a drop vertical jump, the ACL expressed greater peak strain (6.1% vs. 0.4%; P < 0.01) than the MCL. The isolated ACL also expressed greater peak anterior force (4.8% bodyweight vs. 0.3% bodyweight; P < 0.01), medial force (1.6% bodyweight vs. 0.4% bodyweight; P < 0.01), flexion torque (8.4 N*m vs. 0.4 N*m; P < 0.01), abduction torque (2.6 N*m vs. 0.3 N*m; P < 0.01), and adduction torque (0.5 N*m vs. 0.0 N*m; P = 0.03) than the isolated MCL. During tensioning, ACL specimens failed at 37.0% strain and 637 N, while MCLs failed at 17.6% and 776 N. 114

123 Conclusions: During controlled physiologic athletic tasks, the ACL provides greater contributions to knee restraint than the MCL, which is generally unstrained and minimally loaded. Clinical Relevance: An enhanced understanding of joint loading during in vivo tasks may provide insight that enhances the efficacy of injury prevention protocols. Current findings support that multiplanar loading during athletic tasks preferentially loads the ACL over the MCL, leaving it more susceptible to injury. Key Terms: anterior cruciate ligament injury, medial collateral ligament, cadaveric simulation, knee biomechanics, athletic tasks INTRODUCTION Annually, an estimated 2 million anterior cruciate ligament (ACL) injuries occur worldwide. 108 In the United States alone, an estimated 127,000 ACL reconstructions are performed annually in response to ACL injuries, with a cumulative cost that reaches into the billions of dollars. 55,58 The ACL is the primary stabilizer against anterior tibial translation and restrains up to 85% of the anterior force in the knee. The ligament also operates as a secondary restraint against additional degrees of freedom, such as abduction torque. 1,7,181 When the ACL ruptures, these restraints are lost and result in joint instability. ACL failures are devastating knee injuries that have both short and long term implications on athletic participation and activities of daily living. Specifically, the average return to sport time following ACL reconstruction is between 6-12 months and approximately 75% of reconstructed athletes report negative effects on their quality of life within 15 years post operative. 63,64,217 Repetitive study has demonstrated that up to 70% of ACL injuries are incurred in noncontact situations, while the final 30% are the result of a person to person or person to object 115

124 collision. 42,43,110 Often, non-contact ACL ruptures are preceded by one or several biomechanical factors that serve as predictors or risk-factors to the onset of injury. These factors include motion pattern, 42,43 joint laxity, 44 knee valgus, 43,46,105 and lack of neuromuscular control. 45,46 In particular, valgus rotation has been associated with ACL injuries and is a primary component used to diagnose ACL injury risk. 46,218 Risk factors can lead to poor control and high mechanical loads during athletic movements like landing, cutting, and pivoting. 45,50 Abnormal loading can be an important mechanism for non-contact injury, especially in cutting maneuvers where knee valgus moments are most sensitive. 202,219 Despite that knee valgus is correlated to ACL injury risk and the medial collateral ligament (MCL) serves as a primary knee restraint against valgus torque, 10,220 less than 30% of ACL injuries have concomitant MCL injury. 88,89 Few investigators have examined this lack of correlation between ACL and MCL injury. The overall goal of this study was to understand how athletic tasks load the knee joint in a manner that drives the ACL toward failure without concomitant MCL injury in up to 70% of cases. This was achieved through the robotic simulation of in vivo recorded knee kinematics on cadaveric lower extremity specimens. It was hypothesized that the ACL would have a greater overall contribution to intact knee forces than the MCL during simulated motion tasks. It was further hypothesized that the ACL would express greater relative peak strain than the MCL during simulated drop landing and sidestep cutting tasks. METHOD Experimental Design 18 Lower extremity human cadaveric specimens from 11 donors (age = 47.6 ± 7.3 years; mass = 829 ± 199 N) were procured from an anatomical donations program (Anatomy Gifts Registry, Inc., Hanover, MD) for inclusion in this study. These specimens were randomized into 116

125 ACL (N = 9, age = 47.3 ± 8.1 years, mass = 838 ± 216 N) and MCL (N = 9, age = 47.1 ± 7.8 years, mass = 853 ± 197 N) study groups. It was ensured that contralateral pairs were placed in opposite groups. A six-degree-of-freedom (6-DOF) robotic manipulator (KR210; KUKA Robotics Corp., Clinton Township, MI) with a six-axis load cell (Theta Model; ATI Industrial Automation, Apex, NC) mounted on the end effector articulated each limb through simulations of athletic tasks. The simulations were constrained by 6-DOF kinematics developed from in vivo three-dimensional (3D) motion capture recordings. Specimens were simulated through athletic task movement patterns in the intact joint condition and then in an isolated ACL or isolated MCL condition, dependent upon group randomization. In the isolated ligament conditions, the ACL or MCL were the only structures that supported load across the joint. Therefore, the isolated condition was used to quantify the individual mechanical contributions of each ligament to knee mechanics. Kinematic Model Chapter 4 documents the development of a model to perform in vitro simulations of athletic movement patterns based on in vivo recordings. Briefly, 3D motion data was collected on a matched male (age = 24 years; height = 175 cm; mass = 68.8 Kg) and female (age = 25; height = 170 cm; mass = 64.4 Kg) and processed into joint kinematics. 179 Based on their computed peak knee abduction moments, the male subject was classified as low-risk for ACL injury while the female was high-risk. 46,209 Mathematical factors were applied to these kinematics to reduce the confounding effects of skin-artifact errors and provide a 6-DOF input that would positioncontrol the robotic manipulator simulations. 117

126 Specimen Preparation Specimen criteria were defined as no previous history of knee trauma, knee surgery, bone cancer, or ankle or shin implants. The limbs were kept frozen at -20 C and allowed to thaw 24 hours prior to testing. Thawed specimens were resected of all soft tissue down to the joint capsule, leaving the collateral and cruciate ligaments and menisci intact. Anatomical landmarks were used to define the tibial joint coordinate system. 178 Relative to this coordinate system, custom biomechanical fixtures were affixed to the tibia, which was then rigidly mounted on the load cell such that the tibia, load cell, and robot end effector axes were all aligned. The tibial joint center point was digitized with a coordinate measuring machine (Faro Digitizer F04L2, FARO Technologies Inc., Lake Mary, FL) and all rotations, translations, forces, and torques were applied and recorded relative to this point. Mounted specimens were articulated to 45 where custom-barbed, 3 mm microminiature differential variable resistance transducers (DVRT, LORD MicroStrain Inc., Williston, VT) were implanted on the ACL and MCL. One DVRT was implanted on the anteromedial bundle of the ACL slightly proximal to the tibial insertion site. 206,210 MCL implantation sites were distal to the femoral origin, midsubstance across the joint space, and proximal to the tibial insertion, which matches previous protocol. 206 All DVRTs were implanted parallel to the fiber alignment of their respective ligaments. 45 was selected for DVRT implantation as previous robotic simulations and muscle-driven flexions of the joint indicate the ACL is unloaded around this position. 114 The initial limb position was different for each simulated task and was matched within 0.5 of the in vivo orientation recorded for all three rotational DOFs. From the initial position, limbs were incrementally loaded in compression and cycled through simulations until a peak force of bodyweights was achieved. This compressive force matched the magnitude vertical ground reaction forces recorded from a single 118

127 leg during DVJ tasks. 83 Similarly, a peak force of bodyweights was achieved for sidestep cutting tasks. A more detailed account of specimen preparation and initial positioning was previously documented in the literature as well as in Chapter Figure 7.1: Sagittal view of a lower extremity specimen affixed to the robot manipulator and articulated through a male DVJ. Simulations for the selected athletic tasks are cycled through landing phase, which is represented by the period from initial contact to maximum knee flexion as stated in Chapter Robotic Simulation All tests were performed at room temperature and the joint was consistently hydrated with saline. Irrespective of specimen gender, movement patterns from four athletic tasks (male DVJ, male sidestep cut, female DVJ, and female sidestep cut) were performed on each limb in a randomized order (Figure 1). Prior to each simulation, the specimen was cycled 10 times to precondition the joint and minimize viscoelastic effects. Preconditioning was followed by a second set of 10 cycles where forces, torques, and strains were recorded. After each motion was 119

128 completed, the specimen was manually articulated to the initial position for the following task and the process was repeated. Once all simulations were run in the intact condition, the specimen was resected of all soft tissues and the distal portions of the femoral condyles to achieve the isolated ACL or isolated MCL condition, respective to specimen grouping. In this isolated condition, the remaining ligament was the only structure transmitting force across the joint, which would quantify its respective contributions to force resistance during motion. All simulations were repeated in the isolated condition. Following simulation, the joint was returned to initial contact orientation, compressed to an unloaded position, and slowly distracted to identify the neutral strain position of the ligament. With the isolated ligament as the only intact load-bearing structure, neutral strain was identified when the force sensors first registered a constant distraction force. The remaining ligament was then resected and all simulations were repeated in a bone-only condition (Figure 2). Figure 7.2: Frontal plane view of a specimen in the ACL-isolated (A) and MCLisolated (B) condition. While the ACL may appear intact for the MCL-isolated condition, its femoral insertion was severed from the bone and bore no mechanical load. 120

129 Uniaxial Failure Loading After the completion of all robotic simulations for each specimen, the full ACL and MCL were removed in a bone-ligament-bone format. These specimens were individually secured into mechanical fixtures with bone cement and affixed to a uniaxial servohydraulic testing machine (Model 8501, Instron, Norwood, MA) with a 10 kn load cell. Each specimen was oriented such that the ligament fibers were aligned with the loading axis of the Instron (Figure 3). ACL specimens were oriented such that the anteromedial bundle was preferentially loaded, as DVRTs were implanted on this bundle during robotic simulation. Prior to failure testing, a small preload of 4 N was applied to each specimen after which they were cyclically preconditioned between 0% and 3% strain with a cycling rate of 1 Hz over 50 cycles. During failure testing the specimens were loaded linearly at a strain rate of 20% per second until failure was achieved, which corresponds with previous investigations Figure 7.3: An ACL specimen prepared for uniaxial failure testing (A), and following the completion of uniaxial failure testing (B). 121

130 Data Analysis All forces and torques were analyzed in the tibial reference frame based on the knee joint coordinate system. 178 The forces and torques recorded during the bone-only condition represent values generated from gravity and robot inertia; therefore, they were subtracted from the corresponding intact and isolated-ligament conditions. All forces and torques were smoothed through a Fourier transform with a 12 Hz frequency. Translational forces were normalized to percent bodyweight. The identification of a natural strain position allowed for the calculation of absolute ligament strain rather than strain relative to the insertion orientation of the DVRT. The 8 th and 9 th cycles of each 10 cycle set were used for data analysis to eliminate cycle effects. Forces were normalized to percent bodyweight for each cadaveric specimen. All data was normalized to percent of landing cycle, which began at the point of initial contact and ended when the minimum center of gravity was achieved. Univariate ANOVA was used to evaluate statistical significance (α < 0.05) between ligaments. Statistical analyses were performed in MATLAB (version 2012b, The MathWorks, Inc., Natick, MA) using built-in functions and verified in SPSS (version 21, IBM Corp, Armonk, NY). RESULTS Uniaxial Failure During uniaxial failure testing, MCL specimens expressed greater elongation to failure than ACL specimens (difference = 6.1 mm; P > 0.01; Table 1). However ACLs expressed greater strain at failure than MCLs (difference = 20%; P > 0.01). There were no statistically significant differences in ultimate load at failure between the ligaments. 122

131 Table 7.1: ACL and MCL biomechanics from uniaxial tensioning to failure. ACL MCL Ultimate Load (N) 637 ± ± 291 Ultimate Load (%BW) 79.2 ± ± 0.4 Length (mm) 21.8 ± 4.3* 79.7 ± 10.8* Elongation (mm) 7.9 ± 2.1* 14.0 ± 4.0* Failure Strain (%) 37.0 ± 0.1* 17.6 ± 0.1* * indicates significant difference between ACL and MCL Ligament Strain For the intact knee, the strain exhibited by the ACL throughout all simulated tasks was consistently greater than the strain exhibited by the MCL (Figure 4). In the isolated ligament condition, the ACL again consistently exhibited greater peak strain than the MCL. For both ligaments, strains were greater throughout the intact knee simulations than during isolated ligament simulations. Relative to their respective neutral strain positions, the ACL generally exhibited greater tension while the MCL generally exhibited greater compression throughout each simulated task. For the isolated ligament condition, the MCL was unstrained (strain < 0.0%) for all simulated tasks. During the intact-knee female DVJ simulation, the average peak ACL strain (6.1%) was greater than the average peak MCL strain (0.7%; P = 0.01; Figure 5). This was also true for the intact-knee female cut (3.8% vs. -0.3%; P = 0.03), male DVJ (6.1% vs. 0.1%; P = 0.01), and male cut (7.0% vs. 1.2%; P = 0.01) as well as for the isolated-ligament-knee male cut (6.3% vs. -0.6%; P = 0.04). 123

132 Figure 7.4: Population average absolute strains for the ACL (red) and MCL (blue) in the intact-knee (solid) and isolated-ligament (dashed) condition throughout each simulated motion task. Throughout the majority of simulated tasks, the MCL was unstrained, while the ACL expressed up to 6.3% strain. 124

133 Figure 7.5: Average peak ACL and MCL strains for the specimen population during each simulated task and compared to the uniaxial failure strain of the respective ligament. For the intact knee, peak ACL strains accounted for between 10.4% and 18.9% of the ACL uniaxial failure strain, while MCL strains accounted for between -1.6% and 12.2% of the MCL uniaxial failure strain. For the isolated ligament condition, peak ACL strains accounted for between 8.6% and 17.1% of failure, while peak MCL strains accounted for between -9.8% and -0.7% of failure. 125

134 Ligament Loading During both male and female simulated DVJs, ACL-isolated specimens expressed greater magnitudes of peak anterior and medial force than MCL-isolated specimens (P 0.01; Table 2). In both genders, the ACL also expressed greater peak flexion, abduction, and adduction torques (P < 0.05) during simulated DVJs. Neither the isolated ACL nor the isolated MCL expressed significant forces (peak < 0.5% * bodyweight) in the posterior, lateral, or compression DOFs for either DVJ model. Similarly, neither isolated ligament exhibited significant torques (peak < 0.5 N*m) in the external, internal, or adduction DOFs. For both the translational and rotational loading components, the ACL exhibited greater contributions than the MCL (Figure 6 & 7). Table 7.2: Peak forces and torques generated during simulations performed in the ACL-isolated condition. Forces and torques are directional. Posterior, lateral, distraction, external, extension, and abduction are represented by positive values. Force (%BW) Female Box Male Box Female Cut Male Cut Posterior Anterior -4.6* -5.0* * Lateral * Medial -1.8* -1.5* * Distraction Compression * -0.1 Torque (N*m) External Internal Extension * 1.5* Flexion -8.2* -8.6* * Abduction 2.9* 2.4* * Adduction -0.5* -0.4* * indicates significant difference between ACL and MCL 126

135 Figure 7.6: Total knee force during landing phase of athletic tasks. Sum of the average translational components during each simulated task for the intact knee, ACL-isolated knee, and MCL-isolated knee. As noted in previous simulations of the stance phase of gait, 181 the ACL and MCL offer limited contributions to joint restraint when compared to the total joint force. In three of four simulated tasks, the ACL had greater overall mechanical contributions than the MCL. Figure 7.7: Total knee torque during landing phase of athletic tasks. Sum of the average rotational components during each simulated task for the intact knee, ACL-isolated knee, and MCL-isolated knee. As seen in the translational components, the ACL had greater overall torsional contributions to knee joint restraint than the MCL. 127

136 There were fewer statistically significant differences between the isolated ACL and isolated MCL during sidestep cut simulations than during DVJ simulations (Table 3). Both the male and female sidestep cut models exhibited greater magnitudes of extension torque in the ACL-isolated condition than in the MCL-isolated condition (P < 0.05). The male sidestep cut model also exhibited greater peak loads in the ACL-isolated condition than the MCL-isolated condition for anterior force (P > 0.01), flexion torque (P > 0.01), and abduction torque (P > 0.01). These differences were not present in the female sidestep cut simulations. As with the DVJ models, neither the ACL-isolated or MCL-isolated specimens expressed significant peak magnitudes of posterior force, lateral force, compressive force, external torque, and internal torque during sidestep cut simulations. Table 7.3: Peak forces and torques generated during simulations performed in the MCL-isolated condition. Forces and torques are directional. Posterior, lateral, distraction, external, extension, and abduction are represented by positive values. Force (%BW) Female Box Male Box Female Cut Male Cut Posterior Anterior -0.4* -0.1* * Lateral * Medial -0.1* -0.1* Distraction Compression * 0.0 Torque (N*m) External Internal Extension * -0.3* Flexion -0.5* -0.2* * Abduction 0.6* 0.1* * Adduction 0.1* -0.1* * indicates significant difference between ACL and MCL 128

137 DISCUSSION Abduction torque at the knee is associated with increased ACL injury risk as well as increased MCL contributions to knee joint kinetics. 1,8,46 Video analyses of ACL injuries have reported that dynamic knee valgus collapse, which invokes knee abduction, 158 is the most common mechanism of ACL injury. 185,214,224 However, despite this correlation with a common loading mechanism, only 30-40% of ACL injuries experience concomitant MCL rupture. 88,89 Therefore, the purpose of this study was to understand how athletic tasks load the knee joint in a manner that drives the ACL toward failure without concomitant MCL failure. The hypothesis that the ACL would have a greater overall contribution to intact knee forces than the MCL during simulated motion tasks was supported. Irrespective of gender, data from the present study showed that DVJ and sidestep cutting tasks generated significantly greater peak loads in the ACL-isolated knee than the MCL-isolated knee. This indicated that the ACL contributed more to knee restraint than the MCL during physiologic athletic tasks. Accordingly, it would be expected that the ACL would continue to resist a greater proportion of joint loads than the MCL during an injury scenario. Previous literature from impact simulations confirmed the presence of a bias in the ratio of ACL:MCL loading during injury scenarios. 208 Such a loading disparity could lead to ACL failure without concomitant MCL rupture. The hypothesis that the ACL would undergo greater relative peak strain than the MCL during simulated DVJ and sidestep cutting tasks was also supported. Irrespective of gender, the current investigation found that during the simulation of athletic tasks in the intact knee, the ACL was continuously subjected to greater strain than the MCL. This is in agreement with previous investigations that have reported ACL:MCL strain ratios exceed 1.7 during simulated ground impact. 208 In the present study, the ACL:MCL strain ratios exhibited even greater disparity as the 129

138 mean ratio for DVJ simulations was 15.3 and for sidestep cut simulations was The increased ratios in the present study were the result of minimal strain being observed on the MCL throughout each simulation. In previous literature, loads were generated through an impulse force applied to the foot. 208 Specimen position was passively constrained through the ligaments within the joint as well as a series of constant-force cable-pulley systems attached to tendons surrounding the joint. Computational models demonstrate that muscle force activations are in flux throughout in vivo landing, constantly adjusting to actively restrain the knee. 162 In the present study, specimen position was actively constrained by the robotic manipulator and confined to the predetermined path generated from in vivo recorded kinematics. These differences in testing methodology could have lead to the magnitude differences observed in ACL:MCL strain ratios. Further, the impact system was designed to generate ACL rupture, 208 whereas the current system reproduced motions that were not damaging to the limbs as was documented in Chapter 4. ACL injures are often attributed to the poor neuromuscular control, which results in abnormal loading of passive knee restrains. 42,45,46 Therefore, it would be expected that ACL:MCL strain ratios would decrease during impact simulations, as they represent a loss of neuromuscular control and invoke passive structures like the MCL to constrain the knee. Conversely, the in vivo motions emulated by robotic simulation were representative of task that exhibited greater control through neuromuscular pathways. These conditions minimized mechanical demand on passive knee structures, which lead to large ACL:MCL strain ratios as the MCL strain approached 0.0%. Data from the present study corresponds with findings from Chapter 3 and 6. The isolated ACL contributed significant torque (> 1.0 N*m) to the flexion and abduction DOFs. The isolated ACL did not contribute significantly to internal torque resistance. Significant evidence, including 130

139 several biomechanical testing mechanisms documented in Chapter 3, has supported valgus collapse as a primary mechanism of noncontact ACL injury. 46,185,208,214,224 Yet, some investigators attribute ACL injury to increased internal tibial torsion. 9,84,85 Along with the data from Chapter 6 that demonstrated ACL strain values increased more from knee abduction rotations than from internal tibial rotation, force contributions from the DVJ and sidestep cut tasks simulated in the current study indicated that the ACL would be more susceptible to abduction torques from valgus collapse than from isolated internal tibial loads. In the present simulation model, the isolated ACL condition made significant contributions (force > 1.0% bodyweight) to the anterior and medial forces. This corresponds with previous literature that credits the ACL as a primary resistor to anterior tibial translation and a secondary resistor to medial tibial translation. 1,7,8,181 That the ACL followed accepted mechanical conventions of force resistance adds validity to the novel simulation method utilized in this investigation. As documented in chapter 4, simulations in the current study generated between bodyweights of compressive force at the knee in each specimen. This compressive load corresponded with the rise in total knee force beginning at approximately 20% of landing phase. The compressive force was protective to the ACL as it unloaded the ligament by 20% of landing phase and ACL loads did not reemerge till after 50% of landing phase when peak compression had subsided. This finding conflicts with other studies that have indicated tibiofemoral compression drives the femoral condyles along the slope of the tibial plateau to create anterior tibial translation and increase ACL strain. 9, However, those studies invoked average loads of kn that were 2-4 times the magnitude of compressive loads in the current study. Unlike those studies, the present investigation, precisely controlled the position of the tibia and 131

140 femur with the robotic manipulator. Therefore, compressive loads would not have shifted tibiofemoral orientation in the transverse plane, but would have reduced the distance between femoral and tibial insertions and, thus, mechanical demand on the ligaments. The 17.6% strain reported for MCL failure during uniaxial loading in the present study corresponded well with previous investigations where the MCL failed at 17.1% strain. 228 However, strain at ultimate load for the ACL has exhibited greater variation in the literature. For middle aged specimens, average failure strains during uniaxial loading have been reported at 15.0% 229 while impact testing has yielded ACL failure strains of % with a mean of 17.9%. 206 In the current study, uniaxial ACL failure occurred at 37.0% strain. Uniaxial testing was performed following completion of robotic simulations, which meant the specimens were failed after being subjected to 12 hours of intermittent mechanical simulations. As noted in the results, the MCL was not exposed to significant strain through the duration of these simulations. However, the ACL was consistently being strained throughout the duration of the testing protocol. As such, it is possible that the ACL was elongated during robotic simulation, which would lead to higher failure strains that previously reported, while the MCL remained consistent with previous reports. Also, in the present study, the ACL was preferentially loaded such that stress would be initiated in the anteromedial bundle. Initiation of stress across a single bundle rather than the whole ACL may have decreased the structural stiffness and permitted greater excursion prior to failure. These element of ligament fatigue and loading orientation may have also contributed to the reduced failure ACL loads observed in the current study (637 N) than in previous literature (1160 N). 223 A limitation of this study was that the individual contributions of the ACL and MCL to overall knee biomechanics were determined relative to the isolated ligament condition, where all 132

141 additional intra-articular structures of the knee were resected. As seen in Chapter 6, when specimens transitioned from the intact to isolated ligament condition, the magnitude of strain in the ACL and MCL decreased. This decline in strain magnitude indicated that ligament slack increased and load decreased as additional knee structures were resected. Therefore, interaction between the ACL and MCL and additional intra-articular knee structures likely influences the respective biomechanics of both ligaments. For this reason, comparison of the ACL-deficient or MCL-deficient knee mechanics to intact knee mechanics may provide a more precise depiction of the individual contributions from each ligament to intact knee restraint. CONCLUSIONS Data from this investigation demonstrated the relative contributions of the ACL and MCL to knee joint restraint during simulated in vivo athletic tasks. In these controlled athletic tasks, where no damage was inflicted on the specimen, the ACL exhibited significantly greater loading and strain than the MCL. In most of the simulated conditions, the MCL was unstrained. ACKNOWLEDGEMENTS This work was supported by the National Institutes of Health/NIAMS Grants #R01- AR049735, #R01-AR05563, #R01-AR and #R01-AR The authors would also like to acknowledge the support of the staff at the Sports Health and Performance Institute at The Ohio State University and the Sports Medicine Biodynamics Laboratory at Cincinnati Children s Hospital. CONFLICT OF INTEREST There were no conflicts of interest in the preparation of this manuscript. 133

142 Chapter 8 Gender Based Differences in Anterior Cruciate Ligament and Medial Collateral Ligament Biomechanics in Robotically Simulated Jump Landing and Sidestep Cutting Nathaniel A. Bates, a,b,c Rebecca J. Nesbitt, a Jason T. Shearn, a Gregory D. Myer, b,d,e,f Timothy E. Hewett a,b,c,d,g a Department of Biomedical Engineering, University of Cincinnati, Cincinnati, OH, USA b Sports Medicine Biodynamics Center, Division of Sports Cincinnati Children s Hospital Medical Center, Cincinnati, OH, USA c The Sports Health and Performance Institute, The Ohio State University, Columbus, OH, USA d Department of Pediatrics, College of Medicine, University of Cincinnati, Cincinnati, OH, USA e Department of Orthopaedic Surgery, College of Medicine, University of Cincinnati, Cincinnati, OH, USA f The Micheli Center for Sports Injury Prevention, Boston, MA g Departments of Physiology and Cell Biology, Orthopaedic Surgery, Family Medicine, and Biomedical Engineering, The Ohio State University, Columbus, OH, USA This manuscript is currently in preparation for submission to the American Journal of Sports Medicine. 134

143 ABSTRACT Background: ACL injury rates are 4-8 times greater in female athletes than their male counterparts. As female athletes are at increased risk of ACL injury compared to males, it is important to understand the underlying mechanics that contribute to this gender bias. Hypothesis/Purpose: The purpose of this investigation was to use a robotic manipulator to simulate male and female kinematics from athletic tasks on cadaveric specimens and identify gender-based mechanical differences relative to the ACL. It was hypothesized that simulations of female motion would generate higher loads associated with greater ACL injury risk than simulations of male motion. It was further hypothesized that female motion simulations would exhibit greater ACL strain than male motion simulations. Study Design: Controlled laboratory study. Methods: A 6-degree-of-freedom robotic manipulator articulated 19 cadaveric lower extremity specimens from 12 cadaveric donors through simulations of in vivo kinematics that were recorded from male and female sidestep cutting and drop vertical jump tasks. Results: Peak ACL strain during a simulated drop vertical jump was 6.27% and 6.61% for the female and male models, respectively (P = 0.86). For the sidestep cut peak ACL strain was 4.33% and 7.57% (P = 0.21). Female simulations exhibited lower peak lateral joint force during the drop vertical jump and lower peak anterior and lateral joint forces and external joint torque during the sidestep cut (P < 0.05). Conclusion: For the athletic tasks simulated, gender-based loading and strain differences were unlikely to have a significant bearing on the increased rate of ACL injures observed in female athletes. Additional perturbation is necessary to invoke the mechanical differences that lead to higher rates of ACL injury in female populations. The current investigation supports that, for the 135

144 conditions simulated, ACL injuries are black swan events where unexpected loss of neuromuscular control leads to ligament rupture. Clinical Relevance: The current investigation indicated that, during the performance of regulated athletic tasks, gender-specific movement patterns exhibit similar levels of ACL strain. As such the rehabilitation and prevention of ACL injuries should handled similarly across genders. Key Terms: anterior cruciate ligament strain, cadaveric simulation, robotic manipulator, gender bias, knee joint loading INTRODUCTION Over 127,000 anterior cruciate ligament (ACL) reconstructions are performed annually in the United States. 55 ACL failures are debilitating knee injuries that incur heavy surgical and rehabilitation costs, can be devastating to athletic careers, and provide a long term prospectus of osteoarthritis and decreased knee quality of life. 58,63,64 These injuries exhibit a gender bias as 1 in 50 to 70 female athletes sustain a traumatic knee injury each year, 97 and females are 4-8 times more likely to experience an ACL tear than their male counterparts. 42,48,98 The gender bias exhibited in ACL injuries may stem from reduced neuromuscular control as up to 70% of ACL ruptures occur in non-contact situations. 42,43,46 Rather than originate from a direct blow to the knee, these injuries occur as the joint is geometrically manipulated into positions that place abnormally large forces and torques on the ACL. 45,50 Females athletes tend to exhibit reduced neuromuscular control compared to their male counterparts, which has been associated with increased ACL injury risk. 75,90-94,230 Specifically, during athletic tasks that involve rapid deceleration and/or change of direction, poor neuromuscular control can lead to knee abduction coupled with internal rotation of the tibia, otherwise known as knee valgus. 75,94,95 136

145 Valgus torque has been identified as a precursor to ACL rupture and is used to predict athletes for relative injury risk. 46,77,197,211,218 As with neuromuscular control, female athletes demonstrate greater risk of ACL injury because they exhibit increased knee valgus compared to male counterparts during equivalent athletic tasks. 75,95, A multitude of investigations have demonstrated that the application of knee valgus increases the biomechanical demand on the ACL, 86,106,206 which was also documented in Chapter 6. Computerized simulations of the human knee have identified that peak ACL strain increases non-linearly with increasing valgus moments. 105 Similarly, the addition of knee abduction rotation to a mechanical simulation of jump landing in cadaveric specimens also increased peak ACL strain. 86 Further, robotic manipulators have simulated clinical tests on cadaveric knees and documented that the combined abduction and internal tibial rotations experienced during knee valgus increase ACL load and anterior tibial translation more than either rotation individually (Chapters 3 & 6). All of these scenarios indicate that knee valgus increases mechanical demand on the ACL and may make the ligament more susceptible to injury. As female athletes are at increased risk of ACL injury compared to males, it is important to understand the underlying mechanics that contribute to this gender bias. The purpose of this investigation was to use a robotic manipulator to simulate male and female kinematics from athletic tasks on cadaveric specimens and identify gender-based mechanical differences relative to the ACL. It was hypothesized that simulations of female motion would generate higher loads associated with greater ACL injury risk than simulations of male motion. It was further hypothesized that female motion simulations would exhibit greater ACL strain than male motion simulations. 137

146 METHOD Experimental Design Nineteen human cadaveric lower extremity limbs from 12 unique specimens (9 males; 3 females, age = 47.9 ± 7.0 years; mass = 832 ± 190 N) were obtained from an anatomical donations program (Anatomical Gifts Registry, Hanover, MD) and included in this study. Three additional specimens were excluded due to specimen failure and non-functional ACLs. The limbs were tested using a six-degree-of-freedom (6-DOF) robotic manipulator (KR210; KUKA Robotics Corp., Clinton Township, MI) mounted with a six-axis load cell (Theta Model; ATI Industrial Automation, Apex, NC). The robot simulated gender-specific drop vertical jump (DVJ) and sidestep cutting tasks derived from in vivo recorded kinematics while the force sensor recorded corresponding joint forces and torques as documented in Chapter 4. Specimens were simulated in intact and ACL-isolated conditions in order to quantify the whole joint mechanics and ACL contributions, respectively. Kinematic Model A model to develop gender-specific, 6-DOF input kinematics for robotic simulation from three-dimensional (3D), in vivo motion collected during athletic tasks has been previously described in Chapter 4. Briefly, a matched male (age = 24 years; height = 175 cm; mass = 675 N) and female (age = 25; height = 170 cm; mass = 632 N) subject performed three trials each of a drop vertical jump (DVJ) and sidestep cutting task while a 10-camera motion analysis system (Eagle Cameras, Motion Analysis Corp, Santa Rosa, CA) recorded the 3D positions of retroreflective markers placed at anatomical landmarks on the skin. 3D kinematics were processed from the positional data using custom code in MATLAB (version 2012b, The MathWorks, Inc., Nantick, MA) and an established biomechanical model in Visual3D (version 138

147 4.0, C-Motion, Inc., Germantown, MD). 179 Marker trajectories were filtered through a fourthorder, low-pass, digital filter with a cutoff frequency of 6 Hz and the resultant kinematics from all three trials of each task were averaged into a subject mean for each motion task. Literaturebased scale factors were then used to individually convert each DOF into kinematic input for a position-controlled robotic manipulator while reducing the confounding influence of skin-artifact errors. It should be noted that the male subject was computationally identified as low-risk for ACL injury, while the female was identified as high-risk based on their peak knee abduction moments calculated in Visual3D. 46,209 Specimen Preparation Specimen criteria were defined as no previous history of knee trauma, knee surgery, bone cancer, or ankle or shin implants. The limbs were kept frozen at -20 C and allowed to thaw 24 hours prior to testing. Thawed specimens were resected of all soft tissue down to the joint capsule, leaving the collateral and cruciate ligaments and menisci intact. Based on the tibial joint coordinate system, 178 the tibia was secured into custom fixtures then aligned with and rigidly affixed to the load cell and robot end effector axes. A coordinate measuring machine (Faro Digitizer F04L2, FARO Technologies Inc., Lake Mary, FL) digitized the tibial joint center point and all rotations and translations were applied while forces and torques were recorded about this point. The limb was articulated to 45 flexion, minimally loaded condition, where custombarbed, 3 mm microminiature differential variable resistance transducers (DVRT, LORD MicroStrain Inc, Williston, VT) were implanted on the anteromedial bundle of the ACL following previously described protocols (Figure 8.1). 206, was selected for DVRT insertion as previous human gait simulations have exhibited that the ACL is unloaded in this position. 114 DVRTs were also implanted on the MCL at midsubstance across the joint space, proximal to the 139

148 tibial insertion, and distal to the femoral insertion. For each task simulation, the knee was placed in a position that matched the initial contact orientation recorded in vivo to within 0.5 for all three rotations. From this position the limb was cycled several times to account for viscoelastic effects and establish the initial loading that would match the bodyweights peak vertical force previously reported from in vivo recordings of athletic tasks. 83 A more explicit account of how the specimens were prepared and oriented for testing can be found in Chapter 4 and in previous literature. 74 Figure 8.1: Frontal (A) and sagittal (B) plane views of the arrangement of DVRTs implanted on a cadaveric specimen positioned at 45 flexion. 140