Effect of Total Knee Replacement Conformity on Articular Constraint in Mechanical Testing

Transcription

1 Clemson University TigerPrints All Theses Theses Effect of Total Knee Replacement Conformity on Articular Constraint in Mechanical Testing Madeline Bebler Clemson University, Follow this and additional works at: Recommended Citation Bebler, Madeline, "Effect of Total Knee Replacement Conformity on Articular Constraint in Mechanical Testing" (2018). All Theses This Thesis is brought to you for free and open access by the Theses at TigerPrints. It has been accepted for inclusion in All Theses by an authorized administrator of TigerPrints. For more information, please contact

2 EFFECT OF TOTAL KNEE REPLACEMENT CONFORMITY ON ARTICULAR CONSTRAINT IN MECHANICAL TESTING A Thesis Presented to the Graduate School of Clemson University In Partial Fulfillment of the Requirements for the Degree Masters of Science Bioengineering by Madeline Bebler August 2018 Accepted by: Dr. Melinda Harman, Committee Chair Dr. John DesJardins Dr. Hai Yao

3 ABSTRACT Total knee replacement (TKR) survival rates depend heavily on the kinematic performance of the knee. Various factors contribute to implant failure, with the most common being aseptic loosening, infection, and instability. Instability is defined as abnormal or excessive displacement and can have many surgical or technical causes, and surgeons adjust for concerns of instability by choosing certain designs over others based on preference and experience. The designs of interest for this study include cruciate retaining and posterior stabilized, fixed bearing and mobile bearing, and symmetric and asymmetric. These designs vary in conformity, which is defined by the radii of the articular surfaces and exists on a spectrum, unlike the other categorical TKR design classifications. Conformity index (CI) is defined as the ratio of the femoral component radii and the polyethylene insert radii in the sagittal plane, with more similar radii predicted to provide more articular constraint. The broad objective of this thesis was to understand the role of TKR conformity for providing articular constraint to resist mechanical loads and ultimately affect joint stability. The specific aim of this thesis was to measure CI and articular constraint for a broad spectrum of TKR designs to test the hypothesis that increased CI increases overall articular constraint in AP translation and IE rotation. The results from mechanical testing were used to develop a predictive model of TKR behavior. Twenty total TKRs were selected from an IRB approved registry and tested under ASTM F1223 standard for anterior-posterior translation and internal-external rotation at 0, 30, 60, and 90 of flexion. After testing, the hysteresis loops produced were used to ii

4 classify four movement characteristics to describe the overall behavior of each implant at each angle. CI was measured using reverse engineering software at the approximate center of the medial and lateral condyle on the femoral component at all four angles, and the polyethylene insert radii was measured in the medial and lateral plateau. CI broadly decreased within individual TKR designs, contradicting the perception of a device as overall high or low conforming. Due to differences in movement characteristics of CI > 0.7 and CI < 0.7, the observations were divided into high and low conforming groups. The conclusions of this study indicate the importance of selecting design parameters to induce constraint in measurable ways. TKR conformity varied widely among the contemporary TKR designs, and provided noticeable articular constraint in certain designs. Increased CI was not found to increase overall articular constraint for all design groups, but was found to increase articular constraint in AP translation and IE rotation for some design groups. The predictive model was found to correlate certain design parameters with increased articular constraint, but further work includes improving the specificity of the model. Overall, CI added articular constraint in AP translation and IE rotation for many common TKR designs of the last twenty years. iii

5 ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Melinda Harman, who has continued to advise me beyond the lab and shaped not only the work I produce, but also the engineer I want to become. I also gratefully acknowledge the effort and guidance from my committee members, Dr. John DesJardins and Dr. Hai Yao, as without the foundation from their classwork this research would not be possible. This work would not be possible without the contributions of the students involved in Clemson University Retrieval of Explants Program and Registry of Orthopaedics, including Sean Flannery, Lucy Young, Haley Leslie, and Nicole Meillinger. The Academic Success Center Course Support leadership has offered me the flexibility I needed to balance coursework, research, and my assistantship, and this thesis would not have been possible without them. I would finally like to acknowledge the support and encouragement provided to me by the ReMED lab. I would have not accomplished this project without the many voices of encouragement, guidance, and expertise. iv

6 TABLE OF CONTENTS TITLE PAGE... i ABSTRACT...ii ACKNOWLEDGMENTS... iii LIST OF TABLES... vii LIST OF FIGURES... viii LIST OF EQUATIONS...ix LIST OF ACRONYMS... x CHAPTER I. INTRODUCTION... 1 TKR Design Parameters... 2 Conformity... 4 Purpose of Study... 6 II. METHODS AND MATERIALS... 7 TKR Selection... 7 Conformity Index... 7 Articular Constraint Data Processing Statistical Analysis III. RESULTS CI and Flexion Angle CI and TKR Designs Statistical Analysis Case Studies Predictive Model Page v

7 Table of Contents (Continued)...Page IV. DISCUSSION CI and Flexion Angle CI and TKR Designs Case Studies Predictive Model Conclusion REFERENCES APPENDICES A: TESTING PROCEDURES B: ADDITIONAL FIGURES AND TABLES C: MOVEMENT GRAPHS vi

8 LIST OF TABLES Table Page 3.1 Categorical division of 20 implants Statistical results of movement characteristics comparison Comparison of Implant A and Implant B across all movement characteristics Comparison of Implant K and Implant L across all movement characteristics Comparison of Implant D and Implant E across all movement characteristics Predictive General Linear Model vii

9 LIST OF FIGURES Figure Page 2.1 Femoral and polyethylene radii Mechanical testing set up Movement characteristics on movement graphs CI values for each design with respect to flexion angle AP F/D as compared to CI AP F/D as compared to CI with respect to HC and LC AP F/D as compared to CI with a polynomial model AP F/D characteristic by design group AP Free Motion characteristic by design group IE Peak to Peak characteristic by design group IE Free Motion characteristic by design group Implant D-1 representing CR designs Implant C representing PS designs Implant G in fixed and mobile orientation Implant E representing CR-F-A designs Behavior of Implant A and Implant B Behavior of Implant K and Implant L Behavior of Implant D and Implant E viii

10 LIST OF EQUATIONS Equation Page 2.1 CI AP F/D AP Free Motion IE Peak to Peak IE Free Motion ix

11 LIST OF ACRONYMS Phrase Total Knee Replacement Cruciate Retaining Posterior Cruciate Ligament Posterior Stabilized Fixed Bearing Mobile Bearing Symmetrical Asymmetrical Anterior-Posterior Internal-External Conformity Index Anterior Posterior Force/Displacement High Conforming Low Conforming Acronym TKR CR PCL PS F M S A AP IE CI AP F/D HC LC x

12 CHAPTER ONE INTRODUCTION Total knee replacement (TKR) is a common procedure in the United States, with 1.52% of individuals having at least one TKR in 2010 (Maradit Kremers et al., 2015). TKRs are used primarily for treatment of end-stage knee osteoarthritis or rheumatoid arthritis (Dieppe et al., 1999; Losina et al., 2012). This number is expected to grow in the future due to the aging population, obesity, and other reasons, and the prevalence is predicted to reach 7.4 million individuals in the US alone by 2030 (Losina et al., 2012; Maradit Kremers et al., 2015). These procedures for the most part are highly successful, with over 90% of cases surviving longer than ten years, but various factors can contribute to implant failure, with the most common being aseptic loosening, infection, and instability (Le et al., 2014; Schroer et al., 2013; Sharkey et al., 2002; Sharkey et al., 2014). The reasons for revision have changed within the past five to ten years as polyethylene wear has become less common (Sharkey et al., 2014). Simultaneously, articular geometry was changed to introduce larger contact areas and provide for more constraint, but instability continues to be a common cause of implant failure (Sharkey et al., 2014). Instability is defined as abnormal or excessive displacement of the TKR (Rodriguez-Merchan, 2011). Instability has been found to act in several patterns, including global instability, or laxity through the range of motion, and local instability, which may only occur during mid-flexion, typically 30 to 60 of flexion (Firestone and 1

13 Eberle, 2006; Yercan et al., 2005a). Causes of instability can include inadequate soft tissue balance, component wear, improper component sizing and malposition (Abdel et al., 2014; Hamilton et al., 2014; Lombardi et al., 2014). There are already technologies in use to address surgical causes of instability, including computer-assisted navigation systems (Stulberg et al., 2002). Additionally, surgeons can adjust for concerns of instability by choosing certain designs over others based on preference and experience (Sharkey et al., 2014; Comfort et al., 2014; Conditt et al., 2004). However, there are also methods that TKR manufacturers can add stability through design parameters. TKR Design Parameters In the normal healthy knee, stability is provided by the primary ligaments and muscles crossing the knee joint. However, with progressive arthritic degeneration (Harman et al., 1998) and surgical resection during TKR, those structures are less able to provide joint stability after TKR. Common TKR design parameters are used to provide a continuum of constraint (Lombardi et al., 2014). Cruciate retaining (CR) designs allow the surgeon to retain the posterior cruciate ligament (PCL) following surgery, relying on the ligament to provide support, and some claim, to offer stability, greater range of motion, and improved proprioception (Comfort et al., 2014; Conditt et al., 2004). However, for posterior stabilized (PS) designs, the surgeon removes PCL and the patient relies on a polyethylene post and cam mechanism to control the movement and rotation of the femur relative to the tibia. In addition to translation of the knee joint, axial rotation is also typically addressed by soft tissue restraints in a healthy knee. In order to accommodate this 2

14 rotation, there are two methods commonly used. Fixed bearing (F) designs allow the polyethylene insert to be placed into the tibial tray and remain immobile, and are relatively flat to allow the femoral component rotational freedom. Mobile bearing (M) inserts, however, can freely move in the tibial tray, and are theorized to improve function and longevity by reducing polyethylene wear (Breeman et al., 2013; Graves et al., 2014; Smith et al., 2010). However, these theorized benefits have not definitively translated into lower instability instance and higher range of motion. In a mega regression of 41 studies involving F and M devices, there were no clinically relevant differences in revision rate or range of motion of the devices (van der Voort et al., 2013). Several other studies that used bilateral patients with one M and one F design also saw no difference in either patient preference or in range of motion (Beard et al., 2007; Hasegawa et al., 2009; Y. H. Kim et al., 2007). A final binary design parameter is the use of symmetrical (S) or asymmetrical (A) components. Polyethylene inserts can incorporate different radii to increase tibiofemoral conformity and restrict translation of either the medial or lateral compartment. Such medial/lateral asymmetry is intended to improve kinematic behavior, in theory more closely approximating normal knee kinematics (Banks and Hodge, 2004; Fan et al., 2010; Schmidt et al., 2003; van Duren et al., 2007). In part, this is accomplished by restricting unstable femoral translation and providing condylar offset in flexion (Ginsel et al., 2009). One example from industry of A design has been shown to provide more flexion and internal tibial rotation as compared to other similar S TKRs (Ginsel et al., 2009). This benefit in kinematics from the articular constraint may also render the PCL redundant, as 3

15 Watanabe et al., found no statistical difference in ROM through flexion angles between two populations that received an A CR TKR, one with PCL intact and one without (Watanabe et al., 2013). However, not all studies demonstrated these A components to approximate normal knee kinematics at all flexion angles, as van Duren et al found the medial pivot similar to normal knee kinematics only at low flexion angles, while Moonot et al found similar kinematic patterns in deep flexion (Moonot et al., 2009; van Duren et al., 2007). Conformity In addition to design parameters, inherent stability for TKR is provided by the articular geometry rendering conformity between the articular surfaces of the polyethylene insert and femoral component. Such inherent stability is commonly demonstrated using mechanical testing to measure articular constraint as the resistance to anterior-posterior (AP) translation and internal-external (IE) rotation (Haider and Walker, 2005; Moran et al., 2008; P. Walker and Sathasivam, 2000). Altering conformity can have conflicting effects on TKR behavior, as increased conformity can increase tibiofemoral contact area and decrease the resulting contact pressure, but may overly restrict the kinematics (Sathasivam and Walker, 1999). Low conforming TKR tend to give more displacement and rotation when under similar load conditions as high conforming TKR, and high conforming TKR can be self-stabilizing without soft tissue restraint (Luger et al., 1997). However, historically high conforming polyethylene inserts can also suffer from higher rates of pitting and delamination from the sterilization technique used on the polyethylene (Bartel et al., 1986; Wimmer et al., 2012). As 4

16 polyethylene processing has improved, the overall rates of polyethylene wear causing revision have decreased, and concerns of delamination have decreased accordingly (Sharkey et al., 2014). Unlike categorical TKR design classifications (e.g. CR versus PS, F versus M), conformity inherently exists on a spectrum defined by the radii of the articular surfaces. Specifically, conformity index (CI) is defined as the ratio of the femoral component radii to the polyethylene insert radii in the sagittal plane (Haider and Walker, 2005; Pascau et al., 2009; P. S. Walker et al., 2014). TKR designs with high conformity have more similar femoral and polyethylene insert radii, with a CI approaching one. Modern TKR designs have multiple radii, such that CI varies with flexion angle. Therefore, categorizing TKR designs as high CI or low CI depends on the flexion angle of interest. Typically, TKR designs are categorized based on the CI measured at 0 flexion, which would fail to address mid flexion instability reported clinically (Abdel et al., 2014). The rationale for categorizations of high or low conformity is not well described, as different estimates are given in various studies. Low conforming TKR CI has been defined in two studies to range from 0.25 to 0.77, while high conforming TKR CI values range from 0.45 to 0.86 (D lima et al., 2001; Luger et al., 1997). A third finite element approach found the fluid pressure peaks of a TKR changed only when CI was greater than 0.96 (Pascau et al., 2009). The wide range of CIs for high and low conformity lead to overlapping CI values between the groups. 5

17 Purpose of Study The broad objective of this thesis is to understand the role of TKR conformity for providing articular constraint and resisting mechanical loads to ultimately affect joint stability. The aim of this thesis is to measure CI and articular constraint for a broad spectrum of TKR designs commonly used over the past two decades. The TKR designs will be tested at 0, 30, 60 and 90, and the results used to test the hypothesis that increased CI increases overall articular constraint in AP translation and IE rotation. The results from mechanical testing will be used to create a predictive model of TKR behavior from of the designs evaluated. Additionally, three case studies of design changes will be assessed to determine if there are measurable differences in constraint for incremental changes in articular geometry. The first case study will consider the effect of increasing conformity of the polyethylene insert with use of the same femoral component. It is hypothesized that the increase in conformity will lead to an increase in articular constraint. The second case study will consider a CR and a PS TKR from the same manufacturer and assess the effect of adding a post-cam mechanism in a PS design without altering the articular geometry of the original CR design. The third case study considers the effect of asymmetry using a femoral component from one manufacturer that could be used with either a S or A polyethylene insert, such that each combination had different CI. Specifically, the lateral condyle of the A insert was similar in CI to the S insert, and the medial condyle was much less conforming. 6

18 CHAPTER TWO METHODS AND MATERIALS TKR Selection A joint replacement archive established with institutional review board approval (Clemson University Retrieval of Explants Program and Registry in Orthopaedics, IRB ) was queried to identify TKR designs having a broad spectrum of design parameters. In particular, designs with polyethylene inserts representing incremental changes in conformity were targeted. The following inclusion criteria were defined to aid the query: included a tibial baseplate, polyethylene insert, and femoral component; had functional duration of 6 months or less; did not have visible wear on any component; was medium sized based on manufacturer sizing and had articular features classified as cruciate-retaining (CR) or posterior-stabilized (PS), symmetric or asymmetric, and fixedbearing (F) or mobile-bearing (M). Sixteen TKR designs meeting those criteria were included and grouped according to their articular features (Table B.1). Overall, there were 11 CR designs and 5 PS designs. Within CR designs, three were F-A, two were M- S, and the remaining six were F-S. Among the PS designs, none were A, two were M-S, and the remaining four were F-S. Conformity Index CI was measured using reverse engineering software (Materialise 3-Matic Research, version 11.0, Materialise, Belgium, Germany) applied to 3D models captured with a laser scanner (Next Engine, Santa Monica, California). The measurements were taken in the sagittal plane, which was found using a perpendicular plane of the base of 7

19 each component. From the bisecting plane, parallel planes were created in the approximate center of either the femoral condyles or the polyethylene tibial plateaus (Figure 2.1). The femoral components radii were measured from the approximate center of their lateral and medial condyle at 0, 30, 60 and 90. This generated four radii values for the medial condyle and four radii values for lateral condyle of each femoral component. The femoral radii were measured by creating an arc from three points on the outer edge of the implant: the point of intersection of the angle of flexion, and the points of intersections of ±15 from the angle of flexion. Polyethylene insert radii were measured on the approximate center of the lateral and medial plateau. Insert radii were calculated by creating an arc that included points on the anterior lip, the center, and the posterior lip to generate one radii for each of the medial and lateral articular surfaces. For a given TKR, CI was calculated at the four flexion angles (0, 30, 60, 90 ) as the ratio of the femoral radii to tibial radii (Equation 2.1) for both the medial and lateral components. For S components, average CI was calculated as the mean of the medial and lateral CI. Each flexion angle was considered an independent sample. Full tables of radii and conformity can be found in Appendix B, Tables B.2-4. Equation 2.1: CI. X refers to the angle of interest. 8

20 A B Figure 2.1: Femoral and polyethylene radii. A) The femoral component was measured by an arc ±15 at 0, 30, 60, and 90. The changes in color indicate the different arc measured along that segment of the femoral condyle. B) The tibial component was measured by an arc from lip to lip. The three dots indicate the points used to create a line. 9

21 Articular Constraint Articular constraint was measured at four flexion positions using mechanical testing conforming to ASTM Standard F1223 Antero-Posterior Draw Test and Varus- Valgus Test (Appendix A, Section 1) (ASTM Standard F1223, 2014). Four of the implants were tested in both mobile and fixed configurations. Thus, there were 20 total tested implants at four flexion positions, leading to 80 independent samples. Four flexion angles were chosen to represent the extension and flexion angles commonly used during TKR surgery to evaluate passive stability and zones of mid-flexion instability between 30 to 70 degrees (Yercan et al., 2005b). Each TKR design was secured to a test frame (Instron 8864, Instron Corp, Norwood, Massachusetts) using custom fixtures and tested at 0, 30, 60 and 90, with 0 representing extension and 90 representing flexion. In the AP translation test, the relative position of the femoral and tibial components were tracked by a motion capture system (Polaris, NDI Medical, Ontario, Canada), with reflective markers placed on a global origin, the femoral component and tibial component (Figure 2.2). Components were centered into a neutral starting position by allowing translational degrees of freedom to be free under 712 N of axial load (simulating one bodyweight). AP translation was actuated using a pneumatic pump to apply incremental loads to the tibial component, ranging from 117N to 471N, depending on the constraint of the TKR being tested. This deviates from ASTM F1223 for the Antero-Posterior Draw test; which only calls for a maximum anterior and posterior position under 10 N. This deviation was necessary to demonstrate the difference in magnitude of constraint, as most devices were able to surpass 10 N before becoming constrained. IE rotation was actuated 10

22 using the hydraulic actuator and recorded using the test software (Wave Matrix, Instron Corp, Norwood, Massachussets). Testing was halted when the femoral component reached the rim of the polyethylene insert for both AP translation and IE rotation, as detected visually. M TKR were tested first by allowing free rotation of the polyethylene insert, followed by rigid fixation to limit the insert motion so that it functioned essentially as a F design. Figure 2.2: Mechanical testing setup. The reflective markers on the femoral component, tibial and polyethylene component, and a global reference are shown by the silver balls. The pneumatic pump was used to provide AP pressure leading to translation. Data Processing Data recorded during mechanical testing (axial load, AP translation, axial rotation) were used to generate movement graphs following the procedure given in Appendix B, Section 2. AP translation of the relative tibial and femoral motion was recorded for each incremental AP load using snapshots taken by the motion capture system after the incremental load was exerted. IE rotation captured by the test software was parsed to include every fifth data point. Data from each TKR design was used to 11

23 define four movement characteristics: AP Force/Displacement, IE Peak to Peak, and IE and AP Free Motion defined below (Figure 2.3). Figure 2.3: Movement characteristics on movement graphs. A) AP movement characteristics were AP F/D (red) and AP Free Motion (black). B) IE movement characteristics were IE Peak to Peak (red) and IE Free Motion (black). Anterior Posterior Movement Characteristics AP motion was compared to articular constraint using two movement characteristics: AP Force/Displacement (F/D) to describe resistance to translation throughout the full range, and AP Free Motion to describe the ability to translate with low applied force around the neutral position. The AP F/D characteristic was found using the movement of the tibial component relative to the femoral component and was calculated 12

24 using the most anterior and posterior positions (Equation 2.2). A high AP F/D characteristic indicates higher articular constraint. The AP Free Motion characteristic was any motion that included the neutral position that required less than 5 N/mm for movement between two snapshots (Equation 2.3). This 5 N/mm threshold was defined to distinguish the areas of excessive displacement which would indicate instability. It was also sensitive enough to exclude regions exhibiting stick-slip behavior associated with some friction at the articular surface. Lower AP Free Motion indicated higher articular constraint. AP Free Motion was expressed as a percentage of total motion by dividing by twice the sum of the maximum anterior displacement and the maximum posterior displacement to account for a full hysteresis loop. Equation 2.2: AP F/D. The Max(Anterior Force) and Max(Posterior Force) is the magnitude of the maximum force exerted in the respective direction. The Max(Anterior Displacement) and Max(Posterior Displacement) are the magnitude of maximum displacement in their respective directions. Equation 2.3: AP Free Motion. Position A and Position B need to have opposite signs (cross the neutral point). If the AP Motion AB is less than 5 N/mm, then the displacement between A and B is added to the AP Free Motion total. AP Free Motion is then divided by twice the sum of the range of displacement. If the AP Motion AB is less than 5 N/mm, the adjacent points to A and B were also tested, ie A to C, to encompass the full range of free motion. 13

25 Internal External Movement Characteristics IE motion was compared to articular constraint using two movement characteristics; IE Peak to Peak to describe resistance to rotation throughout the full range, and IE Free Motion describing the ability to rotate with low applied torque around the neutral position. IE Peak to Peak was the difference between the most negative and most positive torque, with higher IE Peak to Peak indicating higher articular constraint (Equation 2.4). IE Free Motion, similar to AP Free Motion described above, is any full degree of rotation that included the neutral position that required under 0.1 Nm of torque to rotate (Equation 2.5). 0.1 Nm was chosen as the threshold for similar reasons as given above for AP Free Motion. IE Free Motion was expressed as a percentage of total motion by dividing by twice the sum of maximum internal rotation and the maximum external rotation to account for a full hysteresis loop. Equation 2.4: IE Peak to Peak. The magnitude of the maximum internal torque and maximum external torque are summed. Equation 2.5: IE Free Motion. Degree A and Degree B are the closest data points to an integer degree, and need to have opposite signs (cross the neutral point). If the IE Motion AB is less than 0.1 Nm, then the displacement between A and B is added to the IE Free Motion total. IE Free Motion was then divided by twice the range of rotation. If the IE Motion AB is less than 0.1 Nm, the adjacent points to A and B were also tested, ie A to C, to encompass the full range of free motion. 14

26 Statistical Analysis The relationship between the independent variable CI, and the dependent variable of movement characteristics (AP F/D, AP Free Motion, IE Peak to Peak, IE Free Motion) were compared using non-parametric statistical analysis due to the lack of normality. Movement characteristics were analyzed with either Mann-Whitney or Paired Wilcoxon tests. Groups with distinct independence were compared using the Mann-Whitney ranked sum test, while F/M comparisons were tested using Paired Wilcoxon due to the inherent pairing in using the same device in both F and M conditions. The results were also modeled using a general linear model with backward elimination of non-significant terms using statistical software (Minitab, Version 18.1, State College, Pennsylvania). Backwards elimination was chosen in attempts to improve the specificity of the model. 15

27 CHAPTER THREE RESULTS CI and Flexion Angle CI ranged broadly within individual TKR designs, contradicting the visual perception of conformity at 0 flexion that is commonly used to classify designs as high or low conforming (Appendix B, Table B.2-B.4). As flexion angle increased from 0 to 90, CI decreased on average 0.41 ± 0.17 within a given design (Figure 3.1). CI values ranged from to 1.13 across TKR designs, but two distinct patterns of behavior in AP translation emerged. CI values above CI = 0.7 visually had a more linear response to increases in conformity in AP F/D, as opposed to the lower, more scattered response of CI values below 0.7 (Figure 3.2, Figure 3.3). Additionally, a polynomial model with degree of two was fit for CI and AP F/D responses in total (Figure 3.4). IE movement characteristics were not correlated to conformity index (IE Peak to Peak, R 2 = 0.06, IE Free Motion, R 2 = 0.00). To account for the difference in AP behavior, CI values were analyzed as high conforming (CI > 0.7) or low conforming (CI 0.7). The other movement characteristics with respect to CI are in Appendix B, Figure B.1-3. Asymmetrical designs were categorized according to their higher CI. 16

28 Figure 3.1: CI values for each design with respect to flexion angle. For all designs, CI decreased with flexion angle from 0 to 90. In one case, CI remained constant for 0 and 30 (Implant H). In several cases, CI increased from 60 to 90 (Implant A, B, C, K, L). Figure 3.2: AP F/D as compared to CI. The black line indicates CI = 0.7. Below 0.7, the response of AP F/D to increases in CI is scattered, while above 0.7, the response is more linear to increases in CI. 17

29 Figure 3.3: AP F/D as compared to CI with respect to HC and LC. The CI values were divided according to high and low conformity in order to show the differences in response. Figure 3.4: AP F/D as compared to CI with a polynomial model. A polynomial model was fitted to the data in addition to the linear models in Figure

30 After dividing the implants based on high and low conformity, it is clear that the design groups are not evenly distributed (Table 3.1). CR devices made up 52 of the 80 observations (65%), with the majority of those devices categorized as CR-F-S-LC (24 observations, 30% of all observations). There were fewer M devices for both CR and PS designs as compared to the F designs. A designs were the smallest represented group, containing only three implants (15% of all observations). Table 3.1: Categorical division of 20 implants (16 implants, 4 tested in two orientations). CR/PS F/M S/A HC/LC Count CR F S HC 8 LC 24 A HC 8 LC 4 M S HC 5 LC 3 PS F S HC 4 LC 16 M S HC 4 LC 4 An overall comparison of each movement characteristic for the different design groups revealed a wide range in amount of articular constraint (Figure 3.5-8). AP F/D had the largest range, from N/mm to N/mm. AP Free Motion was right skewed, with 70% showing no free motion. IE Peak to Peak also had a large range from 0.58 Nm to Nm required to move the devices through IE rotation. Most devices had some IE Free Motion, ranging from 0.0% to 95% free motion. The movement graphs of each TKR 19

31 can be found in Appendix C. The full table of movement characteristic data can be found in Appendix B, Table B.5. Figure 3.5: AP F/D characteristic by design group. X marks the mean of each category, and all individual data are shown. Figure 3.6: AP Free Motion characteristic by design group. X marks the mean of each category, and all individual data are shown. 20

32 Figure 3.7: IE Peak to Peak characteristic by design group. X marks the mean of each category, and all individual data are shown. Figure 3.7: IE Free Motion characteristic by design group. X marks the mean of each category, and all individual data are shown in circles. 21

33 CI and TKR Designs Several patterns of behavior emerged from the movement graphs. CR-F-S designs, many of which had a HC CI at 0 with LC CIs at all other angles, had even, symmetric AP hysteresis loops with smaller AP F/D as CI decreased (Figure 3.9). The IE hysteresis loops showed a similar pattern, with increases in torque as the device moved further in the internally. As CI decreased, IE Peak to Peak decreased. The angle range would decrease for some implants and remain constant for other implants. AP Translation IE Rotation Figure 3.9: Implant D-1 representing CR designs. At 0, the implant was HC, and was LC for all other angles. PS-F-S designs in AP translation had much steeper AP F/D than CR-F-S designs (Figure 3.10). Most PS designs were LC at every angle, but would get numerically lower with flexion angle. The implants were able to freely move around the neutral position when the post-cam mechanism was not engaged, but the force necessary for movement increased once there was post-cam engagement. In IE rotation, engagement between the post and intracondylar region on the femoral component led to increased torque at the extreme IE rotation range, as opposed to the constraint provided by the increased 22

34 curvature on the insert lip. The femoral component was able to rotate about the neutral position without interacting with the post, but torque increased when the post-cam mechanism was engaged. AP Translation IE Rotation Figure 3.10: Implant C, representing PS designs. The implant was LC for all angles. M designs had similar AP hysteresis loops to their fixed counterparts, but different IE rotation. The AP hysteresis loop for M designs were similar in range of magnitude and AP F/D to their F orientation. Instead of minimal torque around the neutral position and increased torque magnitudes at the extreme IE rotation range, torque was consistent throughout the entire range (Figure 3.11). The low torque magnitude was consistent with the polyethylene insert articulating with the metal tibial baseplate, not the femoral component, thus leading to a constant force exerted on the device and minimal changes with respect to CI. 23

35 Fixed AP Translation IE Rotation Mobile Figure 3119: Implant G, in fixed and mobile orientation. The AP behavior of the devices is similar in range and magnitude of force, but the IE behavior has a smaller IE Peak to Peak and larger IE Free Motion. A designs had very distinct IE movement patterns compared to their S counterparts (Figure 3.12). The AP translation hysteresis curves, while steep, were similar to the AP F/D and AP Free Motion of the S designs (Figure 3.10). However, the IE rotation curves were visually different, as the internal and external rotation each had a different behavior. Even as the CIs are both considered LC, there was a noticeable difference in the IE behavior on the lateral (high LC) and medial (low LC) sides. The less conforming side behaves like a PS design, only reaching maximum torque at the extreme, 24

36 while the more conforming side shows increases in torque through the neutral position on the lateral compartment. Measurements of articular constraint decreased with decreases in CI for A designs overall. AP Translation IE Rotation Figure 3.12: Implant E representing CR-F-A designs. At 0 and 30, the implant was HC, and was LC for 60 and 90. Statistical Analysis Table 3.2: Statistical results of movement characteristic comparison. Any box with a p value shown was statistically significant under p<0.05. Anything not shown had a higher p value than 0.05 Under Test, MW refers to Mann Whitney and PW refers to Paired Wilcoxon. For PW, there was only one group (N1) and the difference in two testing situations was compared. Group Parameter N1, N2 CR-F- S CR-M- S Test AP F/D [N/mm] AP Free Motion [- ] IE Peak to Peak [Nm] IE Free Motion[-] HC vs. LC 8, 24 MW HC vs. LC 5, 3 MW PS-F-S HC vs. LC 4, 16 MW PS-M- S HC vs. LC 4, 4 MW CR-S F vs. M 8 PW

37 Group Parameter N1, N2 Test AP F/D [N/mm] AP Free Motion [- ] IE Peak to Peak [Nm] IE Free Motion[-] PS-S F vs. M 8 PW M- HC-S M-LC- S F-HC- S F-LC- S CR vs. PS 5, 4 MW CR vs. PS 3, 4 MW CR vs. PS 8, 4 MW CR vs. PS 24, 16 MW TKRs with increased CI had higher articular constraint for certain designs. CR-F- S designs had a significant increase in all movement characteristics of articular constraint when HC and LC were compared, but all other design comparisons were not statistically significant (Table 3.2). Among PS-LC designs, there was a trend for less articular constraint in AP, but more articular constraint in IE. Based on those data, the post and cam mechanism was the dominant cause of articular constraint in AP translation and IE rotation, reducing the effect of conformity at the articular surface. Among the M designs, CI was not a factor affecting AP or IE articular constraint. Among comparisons of CR and PS designs, CR-HC designs had higher IE articular constraint than PS-HC designs, but not AP articular constraint. However, the reverse is true; PS-LC designs had higher AP articular constraint than CR-LC designs. F and M designs had similar AP movement characteristics, but statistically significant 26

38 differences in IE movement characteristics, regardless of CI. Conformity did not appear to play a role for M designs, and only effected CR-F designs as discussed. Because of the differences in movement behavior, the A designs were excluded from the statistical analysis and model design, as the defined movement characteristics did not properly address their articular constraint during mechanical testing. However, the A designs had behavior consistent with HC designs in AP translation, indicating that high CI dominated the movement behavior. In IE rotation, the movement behavior was different from S designs, and as such the IE characteristics do not accurately reflect the constraint produced. IE Peak to Peak does not distinguish between internal and external torque. A designs had higher torque on their higher conforming side as compared to the lower conforming side. This distinction would falsely inflate the measure of constraint for the LC compartment, while decreasing the effect of the HC compartment in an A design. Case Studies In the first case study, two different polyethylene inserts from the same manufacturer were tested using the same femoral component, such that differences in movement behavior were solely attributed to the polyethylene insert. While one was more conforming with a CI nearly 14% greater at each angle, both designs were consider LC (below 0.7). The behavior of the devices in AP translation resulted in similar AP F/D, but the differences in IE rotation resulted in differences in movement characteristics and range of motion (Table 3.3, Figure 3.13). The higher LC design had a much smaller range of motion as flexion angle increased, but decreases in IE Peak to Peak. 27

39 Table 3.3: Comparison of Implant A (lower LC) and Implant B (higher LC) across all movement characteristics. These demonstrate the effect of incremental change in CI. CI [-] AP F/D [N/mm] AP Free Motion [%] IE Peak to Peak [Nm] IE Free Motion [%] Angle A B A B A B A B A B AP Translation IE Rotation 30 28

40 60 AP Translation IE Rotation 90 Figure 3.13: Behavior of Implant A (lower LC) and Implant B (higher LC). The AP movement has a similar AP F/D, while the IE movement is different for both IE Peak to Peak and IE Free Motion. Despite the low torque experienced by Implant B at 90 in IE movement, the device would exceed the insert lip at further rotation. In the second case study, the behavior of the PS (Implant L) and CR (Implant K) designs were similar at the low angles, but different at high angles for AP translation (Table 3.4, Figure 3.14). The CI values were classified as LC for all angles for the PS and CR implants, and they both were F-S. The CR design had lower CI for all flexion angles. The CR design had movement characteristics consistent with less articular conformity in AP translation, with lower AP F/D and higher AP Free Motion. At 30, the AP F/D was very similar between the two designs, and yielded a similar pattern of motion. This 29

41 similarity only appeared at 30, as all other angles had differences in their movement characteristics. At 90, for example, the PS design had an AP F/D almost five times larger than the CR design, implying it would require five times the force to achieve the same AP translation. In IE rotation, the PS design s range of motion (±5 ) was greatly reduced compared to the CR design, which ranged from ± Additionally, greater torque was required to move the PS design, resulting in larger IE Peak to Peak, and for most angles, negligible IE Free Motion compared to the CR design. Table 3.4: Comparison of Implant K (CR-F-S) and Implant L (PS-F-S) across all movement characteristics. CI [-] AP F/D AP Free IE Peak to IE Free Motion [%] [N/mm] Motion [%] Peak [Nm] Angle K L K L K L K L K L AP Translation IE Rotation 30

42 30 AP Translation IE Rotation Figure 3.14: Behavior of Implant K (CR-F-S) and Implant L (PS-F-S). Both implants were LC for all degrees of flexion and came from the same manufacturer. In the third case study, the S (Implant D) and A (Implant E) design had similar patterns of movement and similar movement characteristics in AP translation (Table 3.5, Figure 3.15). The AP F/D was similar for both designs throughout every angle. The S design had a higher AP Free Motion percentage than the A design, despite having very 31

43 similar patterns of movement. In IE rotation, however, the S design had lower articular constraint as compared to the A design. The patterns of movement between the two are different, with the A design seeing larger free motion continuing into the internal condyle. This was also the less conforming condyle, and showed less constraint. The IE Peak to Peak were similar between the two designs, but the S design had a smaller range of motion. These differences were more apparent at higher flexion angles. These findings are consistent with the A design rationale (Banks et al., 1997), with higher CI in a single compartment providing AP stability from 0º to 90º without restricting the magnitude of IE rotation. Table 3.5: Comparison of Implant D (CR-F-S) and Implant E (CR-F-A) movement characteristics. Implant E CI were the larger of the medial and lateral CI for each angle. CI [-] AP F/D [N/mm] AP Free Motion [%] IE Peak to Peak [Nm] IE Free Motion [%] Angle D E D E D E D E D E AP Translation IE Rotation 32

44 30 AP Translation IE Rotation Figure 3.15: Behavior of Implant D (CR-F-S) and Implant E (CR-F-A). The AP translation of the implants were similar in AP F/D and AP Free Motion, but Implant E was less conforming in IE rotation, with a larger range of motion and larger IE Free Motion. 33

45 Predictive Model Table 3.6: Predictive General Linear Model. Any entry with a dash indicates that the variable in question did not have a statistically significant effect (α = 0.05) in predicting the movement characteristic. The model was created following the equation: Characteristic = Underlying Response + Factor CR/PS +FactorF/M + FactorHC/LC. Characteristic Underlying Response CR/PS F/M HC/LC R-sq Adj. AP F/D [N/mm] / / % AP Free Motion / % [-] IE Peak to Peak / % [Nm] IE Free Motion [-] / % The predictive general linear model reveals that AP and IE articular constraint are dependent on different design parameters (Table 3.6). AP movement characteristics depend primarily on high and low conformity, as a CR-HC design would have a predicted AP F/D of N/mm, while a CR-LC design would have a predicted AP F/D of N/mm. PS designs offered more articular constraint than the CR designs when looking at AP F/D. A PS-HC design would have N/mm AP F/D predicted, as compared to N/mm in a CR-HC design. F and M designs had no significant effect on AP characteristics, but were the dominant effect on IE characteristics, as a F design IE Peak to Peak prediction is Nm, but the M design s prediction is 4.51 Nm. The elimination of non-statistically significant components did not alter the overall fit of the model, as the R-adjusted did not change by more than 2%. A designs were not considered due to their small sample size and distinct movement patterns. 34

46 CHAPTER FOUR DISCUSSION The broad objective of this thesis was to understand the role of TKR conformity for providing articular constraint and resisting mechanical loads to ultimately affect joint stability. The aim of this thesis was to measure CI and articular constraint for a broad spectrum of TKR designs commonly used over the past two decades. The TKR designs were tested at 0, 30, 60 and 90, and the results used to test the hypothesis that increased CI increases overall articular constraint in AP translation and IE rotation. The results from mechanical testing were used to create a predictive model of TKR behavior. Three case studies of designs with incremental changes in articular geometry were also evaluated in detail. CI increased articular constraint for design groups without other mechanisms of articular constraint, and was significant as a constraint for AP translation. Based on comparisons with other studies defining TKR stability (Haider and Walker, 2005; Heim et al., 2001; Sathasivam and Walker, 1999), the measured movement characteristics in this thesis are reasonable representations of articular constraint in modern TKR designs. Using movement characteristics defined in this thesis on raw data, Haider and Walker reported AP F/D ranging from 25 N/mm to 50 N/mm and IE Peak to Peak ranging from 4 Nm to 16 Nm for A TKR, PS TKR, and CR TKR designs (Haider and Walker, 2005). Raw data from mechanical testing of CR-M TKRs from Heim et al reported AP F/D ranging from 37 N/mm to 200 N/mm, and IE Peak to Peak from 11 Nm to 22 Nm (Heim et al., 2001). Sathasivam and Walker also reported 35

47 raw data from mechanical testing of a low and high CI TKRs that had N/mm AP F/D and 20 Nm IE Peak to Peak (Sathasivam and Walker, 1997). CI and Flexion Angle Using reverse engineering software applied to virtual 3D models, this thesis successfully measured CI at different flexion angles (0º-90º) for a wide spectrum of TKR designs. In general, CI decreased from 0º to 90º, with most TKR designs presenting a broad range of CI across that flexion range. The decrease in CI from extension to flexion, and associated changes in articular conformity as measured in this thesis, may contribute to reported mid flexion (30-60 ) instability experienced by some TKR patients (Yercan et al., 2005b). Additionally, the changes in CI may explain, in part, why stability perceived during intra-operative evaluations at full extension and 90 flexion may not indicate stability in mid flexion positions (Yercan et al., 2005a). Two methods of design regarding midflexion CI are the single radii and multi radii approach. The multi radii approach is the traditional one, which aims to create an anatomic joint through a large femoral radius at 0 that reduces posteriorly (Iwaki et al., 2000). The single radii approach however, seeks to create consistent tension in soft tissue restraints through the full range of motion (Stoddard et al., 2013). Jo, et al. found an increase in mean total stability at 30 for a single radii femoral component as compared to a multi radii, J- shaped femoral component in vivo (Jo et al., 2014). Research has also shown single radius designs more stable than a multi radii design when rising from a chair and stair climbing in vivo (Banks et al., 1997; Wang et al., 2006). The single radii component s larger radii throughout the range of flexion contributed to a higher CI, likely leading to 36

48 more articular constraint and stability than the J-shaped femoral component, which had a much smaller femoral radii articulating with the polyethylene insert in mid flexion. For one of the J-shaped implants (Implant G, Figure 3.9), 0 and 30 had an 11% decrease (1 to 0.89), but the 60 CI was nearly 35% less than that of the 30 (0.89 to 0.58). The smaller femoral radii in mid flexion meant the implant went abruptly from very high conforming to less conforming between 30 and 60, which decreased articular constraint. CI and TKR Designs In cases where conformity provides the dominant constraint role, i.e. CR devices, high conformity leads to higher articular constraint. In Figure 3.7, CR-F-S devices had higher AP F/D and higher IE Peak to Peak at 0 (HC) as compared to the other angles (LC). These results support those reported by Willing and Kim, 2011; in that high conformity in CR designs can lead to over-constraint, particularly in AP translation (Willing and Kim, 2011). While Willing and Kim negated the constraint concern by creating an A design, the same could be accomplished with decreasing CI. In PS designs however, the dominant constraint factor primarily comes from the post and cam mechanism, reducing the effect of decreased conformity. In Figure 3.8, the PS-F-S devices had high AP F/D and high IE Peak to Peak despite being LC. Banks et al. has reported that a post and cam mechanism in PS designs engages at approximately 40 flexion, which would affect high flexion degrees (60, 90 ) where CI is lower (Banks et al., 1997). This mechanism may explain the additional constraint found in PS-LC designs as compared to their CR-LC counterparts in the current study. The PS-LC designs were 37