University of Southern California Dissertations and Theses

Investigation of preclinical testing methods for total ankle replacements

(USC Thesis Other)

Investigation of preclinical testing methods for total ankle replacements

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Investigation of Preclinical Testing Methods for Total Ankle Replacements

by

Nathan C. Ho

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Biomedical Engineering)

May 2020

ii
Dedication
This dissertation is dedicated to my parents, Thomas Wong and Wai Chi Ho. You have always
pushed me to be better and to never be satisfied with adequacy. You have always led by example
and have shown me how to persevere and work through all hardship. Without your love and
support, I could not have accomplished this goal.

iii
ABSTRACT

A clear understanding of the loads applied to any natural joint and the resulting kinematics is
essential to the development and appropriate evaluation of its candidate prosthetic replacement
designs and materials. Excellent long-term success of current total hip replacement designs, 97%
survival at 25 years, can be attributed to decades of pre-clinical testing, failure analysis of retrieved
implants, and numerous clinical outcome studies. In contrast, the complexities of the ankle joint are
still unresolved, leading to poor outcomes in total ankle replacements. Short-term studies have found
total ankle replacements to have a 90% survivorship within the first 5 years, but sharply declining to
65% for mid- to long-term survivorship. Due to this harsh decrease in survivorship, universal
adoption of total ankle replacements has been greatly impeded.
The foundation of the disappointing performance of total ankle replacements is rooted in the
inconsistent nature of the preclinical testing of these implants. Unlike current hip replacement testing
standards, which took decades to develop and benefited from extensive failure analysis of retrieval
and clinical studies, for total ankle replacements, only recently a standard has been published by
International Standards of Organization (ISO) and no standards are established by American Society
for Testing and Materials (ASTM). The lack of standards has led to inconsistency among the
protocols by different investigators. For example, in certain testing protocols, critical degrees of
freedom affecting the natural response of the cadaver have been constrained, affecting the results.
Some of the discrepancies stem from a lack of understanding of the biomechanics of the ankle joint,
for example, the inappropriate assumption that the ankle behaves as a simple hinge joint.
Oversimplified models of the kinematics have been used for biomechanical and wear testing.
Specifically, with the lack of any standard, wear tests have been performed without a consensus on
the magnitudes of rotations and translations. Moreover, the majority of preclinical wear tests have
been conducted using displacement and angular rotation control due to ease of implementation.
However, more recently testing via force and torque control have been introduced with the argument
that these produce more physiologically relevant kinematics. With a clear lack of consistency and
numerous methodological variations, the lack of advancement in the designs of total ankle
replacements should come as no surprise.
The overarching purpose of the proposed project was to provide and contribute to the
establishment of the physiologically relevant testing methodology designed to predict in vivo
performance of total ankle replacements. The following Specific Aims were designed to approach
this objective.
1. Establish a model using cadaveric specimens to determine the 6 DOF motions of the natural
tibia, talus, and calcaneus under simulated physiologic gait.
2. Use the cadaveric model to evaluate the 6 DOF motions of total ankle replacement
components.
3. Use explanted total ankle replacement polyethylene components to assess the modes and
locations of damage during in vivo use.
4. Use the cadaveric model to establish 6 DOF motions of the natural talus and the artificial
talar component as a function of fusion of the subtalar joint.
5. Compare the 6 DOF motions of the natural ankle joint resulting from the proposed ISO
22622 force control testing standard to the proposed ISO 22622 displacement control testing
standard.

iv
To achieve these objectives, the following studies were performed. For Specific Aim 1, a gait
analysis experiment was performed using motion tracker cameras synced with force plate data. This
data was analyzed at four distinct instances: Heel strike, maximum weight acceptance, mid-stance,
and push off. This data was then used to correlate shank angle, ankle angle, ground reaction force
angle, and ground reaction force magnitude observed during gait. The results from the gait
experiment were then applied to three pairs of cadaveric lower limbs. Optical motion tracking was
used to measure the 6DOF motions of the tibia, talus, and calcaneus under simulated loading. One
side of each pair was tested in its natural state with all ligaments and soft tissue intact, while the
contralateral side was tested with implantation of a commercially marketed total ankle replacement
design. Through these results, Specific Aim 2 was achieved by comparing the kinematics of a TAR
components to those of a natural ankle.
For Specific Aim 3, fourteen explanted TARs were analyzed. The modes and location of
damage illustrated how the explant was articulating in vivo. By comparing these damage maps with
the results in Specific Aim 2, the testing method developed in Specific Aim 1 was supported.
Following this, a subtalar fusion was simulated using screws on both the natural and TAR
specimens used for Specific Aim 2. Using the model developed in Specific Aim 1, the same loading
methodologies were repeated. Specific Aim 4 was then achieved by analyzing the differences in
displacement of the specimens in the unfused and fused conditions.
For Specific Aim 5, an additional set of 12 cadaveric lower limbs were tested under the proposed
ISO 22622 force profile. From these experiments, the resulting displacements were recorded and
compared to the proposed displacement profiles.
Collectively, these studies contributed to understanding the complete kinematics of the ankle,
leading to a more comprehensive preclinical testing protocol for the evaluation of total ankle
replacements, resulting in improved long-term survivorship of total ankle replacements.

v
Table of Contents
1. Background ........................................................................................................................................... 1
1.1 Background and Historical Perspective on Ankle Biomechanics ................................................. 1
1.1.1 Ankle Anatomy ............................................................................................................................ 1
1.1.2 Indications for Ankle Surgery ...................................................................................................... 2
1.1.3 Surgical Interventions .................................................................................................................. 3
1.1.4 History of Total Ankle Replacement Design ............................................................................... 6
1.1.5 Clinical Outcome ......................................................................................................................... 9
1.2 In Vivo Biomechanics ..................................................................................................................... 11
1.2.1 Design ........................................................................................................................................ 11
1.2.2 Limitations ................................................................................................................................. 14
1.2.3 Applications ............................................................................................................................... 15
1.3 In Vitro Biomechanics .................................................................................................................... 16
1.3.1 Design ........................................................................................................................................ 16
1.3.2 Limitations ................................................................................................................................. 18
1.3.3 Applications ............................................................................................................................... 19
1.4 Wear Testing ................................................................................................................................... 21
1.4.1 Design ........................................................................................................................................ 21
1.4.2 Assumptions ............................................................................................................................... 23
1.4.3 Applications ............................................................................................................................... 24
1.4.3.1 Accelerated Wear Test ........................................................................................................ 24
1.4.3.2 Retrieval Analysis ............................................................................................................... 26
1.5 Applying Cadaveric and Wear Testing for Total Ankle Replacement Biomechanics ............. 28
1.5.1 Osteoarthritis in the Ankle ......................................................................................................... 28
1.5.2 Clinical and Societal Impact of Osteoarthritis ........................................................................... 28
1.5.3 Overview of Total Ankle Replacement ..................................................................................... 29
1.5.4 Benefits and Applicability of Cadaveric and Wear Testing for Total Ankle Replacements ..... 31
1.6 Limitations of Cadaveric and Wear Testing Methods ................................................................ 32
1.6.1 Limitations of Cadaveric Testing in Total Ankle Replacements ............................................... 32
1.6.2 Limitation of Wear Testing in Total Ankle Replacements ........................................................ 34
1.7 The Need for Improved Methods for Total Ankle Replacement Testing .................................. 35
1.8 Rationale for the Proposed Studies ............................................................................................... 37
1.9 Purpose and Aim of Proposed Studies .......................................................................................... 39
2. Materials and Methods ........................................................................................................................ 41
2.1 Overview .......................................................................................................................................... 41

vi
2.2 Gait Analysis .................................................................................................................................... 42
2.2.1 Experiment Design ..................................................................................................................... 42
2.3 Biomechanical Simulation Using Cadaveric Specimens .............................................................. 45
2.3.1 Specimen Preparation ................................................................................................................ 46
2.3.1.1 Radiographs ........................................................................................................................ 46
2.3.1.2 Surgical Procedures ............................................................................................................. 47
2.3.1.3 Flag placement .................................................................................................................... 49
2.3.2 Experimental Design .................................................................................................................. 50
2.3.2.1 Apparatus Design ................................................................................................................ 50
2.3.2.2 Loading Profiles .................................................................................................................. 54
2.3.2.3 Testing Parameters .............................................................................................................. 57
2.3.3 Data Acquisition ........................................................................................................................ 58
2.3.3.1 Motion Tracker ................................................................................................................... 58
2.3.3.2 Digitization .......................................................................................................................... 58
2.3.3.3 Strain Gauge ........................................................................................................................ 60
2.3.3.4 Rotation and Displacement Calculations ............................................................................ 62
2.3.4 Data Analysis ............................................................................................................................. 63
2.3.4.1 Reduction ............................................................................................................................ 63
2.3.4.2 Statistics .............................................................................................................................. 63
2.4 Retrieval Damage Analysis ............................................................................................................ 64
2.4.1 Damage Features ........................................................................................................................ 64
2.4.2 Semi-Quantitative Grading Scale ............................................................................................... 66
2.4.3 Data Analysis ............................................................................................................................. 66
2.5 Force versus Displacement Profiles ............................................................................................... 67
2.5.1 Specimen Preparation ................................................................................................................ 69
2.5.1.1 Dissection ............................................................................................................................ 69
2.5.2 Experimental Design .................................................................................................................. 71
2.5.2.1 Apparatus Design ................................................................................................................ 71
2.5.2.2 Loading Conditions ............................................................................................................. 73
2.5.3 Data Acquisition ........................................................................................................................ 82
2.5.3.1 Motion Tracker ................................................................................................................... 82
2.5.3.2 Digitization .......................................................................................................................... 83
2.5.3.3 Load cell .............................................................................................................................. 84
2.5.4 Data Analysis ............................................................................................................................. 84
3. Results ................................................................................................................................................. 85

vii
3.1 Overview .......................................................................................................................................... 85
3.2 Gait Analysis .................................................................................................................................... 86
3.2.1 Shank Angle ............................................................................................................................... 87
3.2.2 Ankle Angle ............................................................................................................................... 88
3.2.3 Ground Reaction Force Angle ................................................................................................... 89
3.3 Biomechanical Simulation of Intact Cadaveric Ankles Versus Implanted Artificial TAR ..... 91
3.3.1 Direct Anterior-Posterior Loading ............................................................................................. 92
3.3.2 Constrained Cyclic Loading ....................................................................................................... 95
3.3.3 Unconstrained Cyclic Loading ................................................................................................... 99
3.4 Biomechanical Simulation of Cadaveric Specimens with Subtalar Fusions ........................... 103
3.5 Patterns of Damage on Retrieved TAR Polyethylene Inserts ................................................... 106
3.6 Biomechanical Simulation of Force Control Wear Testing in a Cadaveric Model ................ 109
3.6.1 Application of ISO Load Profile Commands ........................................................................... 110
3.6.2 Axial Force Application ........................................................................................................... 111
3.6.3 Flexion Application .................................................................................................................. 112
3.6.4 AP Force Application ............................................................................................................... 113
3.6.5 Internal and External Torque Application ............................................................................... 114
3.6.6 Resulting Measured Displacements and Rotations .................................................................. 115
3.6.7 Measured Anterior Posterior Displacement Curve .................................................................. 116
3.6.8 Measured Internal and External Rotation of the Tibia ............................................................. 118
4. Discussion ......................................................................................................................................... 119
4.1 Overview ........................................................................................................................................ 119
4.2 Gait Analysis .................................................................................................................................. 120
4.3 Biomechanical Simulation of Intact Cadaveric Ankles Versus Implanted Artificial TAR ... 122
4.3.1 Direct Anterior Loading vs Constrained vs Unconstrained ..................................................... 122
4.3.2 Natural vs Total Ankle Replacement During Unconstrained Loading .................................... 125
4.4 Subtalar Fusion of Natural vs Total Ankle Replacement Specimens ...................................... 128
4.5 Patterns of Damage on Retrieved TAR Polyethylene Inserts ................................................... 131
4.6 Biomechanical Simulation of Force Control Wear Testing in a Cadaveric Model ................ 134
4.6.2 Axial Force Application ........................................................................................................... 136
4.6.3 Flexion Application .................................................................................................................. 136
4.6.4 Applied Anterior-Posterior Load Profile ................................................................................. 138
4.6.5 Internal and External Torque Application ............................................................................... 140
4.6.6 Measured AP Displacement Curve vs ISO AP Displacement Curve ...................................... 140
4.5.7 Measured Internal and External Angular Rotations ................................................................. 144

viii
5. Conclusion ........................................................................................................................................ 145
Appendix ................................................................................................................................................... 149
A. MATLAB Code ............................................................................................................................... 149
I. Rotation Transformation Code .......................................................................................... 149
II. Cyclic Loading Analysis ...................................................................................................... 151
III. ISO Force-Displacement Analysis ..................................................................................... 157
IV. ISO Force-Displacement Average Plots ............................................................................ 160
B. Schematic of ISO Force-Displacement Apparatus ...................................................................... 163
C. Biomechanical Comparison of Fixation Stability Using a Lisfranc Plate vs Transarticular
Screws ................................................................................................................................................... 164
D. The Objective Measurement of Brace-Use Adherence in the Treatment of Idiopathic Clubfoot
171
E. Biomechanical Comparison of Fixation Devices for First Metatarsocuneiform Joint
Arthrodesis .......................................................................................................................................... 179
F. Preclinical Biomechanical Testing Models for the Tibiotalar Joint and Its Replacements: A
Systematic Review ............................................................................................................................... 186
G. Systematic Review of Unsystematic Total Ankle Replacement Wear Evaluations .................. 191
H. Damage Patterns in Polyethylene Fixed Bearings of Retrieved Total Ankle Replacements
(Accepted) ............................................................................................................................................ 202
I. Increasing Loads and Diminishing Returns: A Biomechanical Study of Direct Vertebral
Rotation ................................................................................................................................................ 216
J. Is Load Control Necessary to Produce AP Displacement and Axial Rotation in Wear Testing of
TAR? .................................................................................................................................................... 224
References ................................................................................................................................................. 242

1
1. Background

1.1 Background and Historical Perspective on Ankle Biomechanics
Across the world over 100 million people suffer from osteoarthritis, including over 22
million people in North America alone [25, 84]. Woolf et al have predicted that by 2020
osteoarthritis will become one of the most prevalent causes of disability worldwide [159]. The
consequence of living with osteoarthritis results in limited movement at the affected joint, which
in turn limits activities and daily functions. Although only 6% to 15% of all cases of
osteoarthritis involve the ankle joint, it has been found that the severity of pain and loss of
function associated with this condition, is as detrimental to the patient as osteoarthritis at the hip
[25, 40, 51, 115]. While conservative approaches to manage arthritic pain are generally
recommended, invasive options such as arthrodesis (fusion) and arthroplasty (joint replacement)
are often resorted to. Historically, for ankle arthritis specifically, arthrodesis has been the
standard for treatment; however, with recent advancements, total ankle replacements have
become more common. Yet, total ankle replacement success remains controversial due to poor
mid and long-term outcomes which can be attributed in large part, to a lack of an established pre-
clinical testing protocol.
1.1.1 Ankle Anatomy
The ankle is a complex joint of the lower extremity and is necessary for function of daily
life activities. The ankle joint consists of four bones, namely the calcaneus, talus, fibula, and
tibia. Collectively, the distal portions of the tibia and fibula, and the superior portion of the
calcaneus form a synovial capsule containing the talus. A synovial joint encapsulates articulating
bones with synovial fluid which primarily functions to lubricate and provide shock absorption to

2
protect the joint. While the hip joint is a ball and socket and the knee joint is a hinge, the ankle
joint is often assumed to be primarily a hinge; however, its incongruent anatomy allows for the
foot to rotate and function through various ranges of motion including: dorsiflexion (upward)
and plantarflexion (downward), internal and external rotation (left and right), and inversion
(supination) and eversion (pronation). These six movements allow the ankle to articulate in four
degrees of freedom consisting of two primary movements: dorsiflexion/plantarflexion,
interior/exterior rotation and two secondary motions: inversion/eversion and anterior/posterior
displacement.
1.1.2 Indications for Ankle Surgery
The main indication for surgery is end stage arthritis. The two main causes of arthritis are
inflammatory and non-inflammatory conditions. Other causes include idiopathic arthritis,
neuropathic arthritis, osteonecrosis, hemophilic arthritis, septic arthritis, and gout. Inflammatory
conditions consist of rheumatoid arthritis, which is a chronic autoimmune disease resulting in
inflammation of the affected joints. Non-inflammatory conditions consist of primary
osteoarthritis, which is wear and tear degenerative joint disease usually age-related, or post-
traumatic osteoarthritis, which is a result of overuse and injury which is common in young
athletes.
Primary osteoarthritis occurs naturally during aging, when the healthy lining of the
cartilage deteriorates from repeated use, leaving neighboring bones to articulate directly against
each other resulting in pain and swelling of the joint. While there is a high rate of primary
osteoarthritis, the main indication of ankle surgery is secondary osteoarthritis, namely, post-
traumatic osteoarthritis [4, 23, 25, 44]. Post-traumatic osteoarthritis may be induced from a

3
previous injury or physical trauma. During this injury, the damaged joint cartilage may partially
or improperly heal, leading to accelerated degeneration of the joint and ultimately, osteoarthritis.
1.1.3 Surgical Interventions
Ankle arthritis can be managed conservatively through physiotherapy, rest, icing, etc.
However, if conservative measures are unsuccessful, surgery is an end-stage treatment and
patients may become candidates for ankle arthrodesis (fusion), which is the most established
treatment. Yet, due to the high number of long-term complications and altered biomechanical
functions, other forms of joint preservation treatments, such as total ankle arthrodesis
(replacement), have been pursued. While the first generation of implant designs resulted in
unacceptably high complication rates, current designs hold more promise. Yet, due to its relative
infancy, long-term survivorship of these implants remains unknown.
In 1882 the first ankle arthrodesis or fusion, was performed in Austria by Dr. Edward
Albert [55, 142]. This surgery was achieved by excising the articular cartilage between the talus,
fibula, and tibia, and immobilizing the lower leg in a cast [55, 142]. The first compressive
arthrodesis, where minimization of shearing strains and gap prevention between the cut bone
surfaces is achieved through compression, was performed in 1951 in Wrightington by Dr. John
Charnley, innovator of low friction arthroplasty of the hip [28]. Depending on the severity and
location of pain, ankle fusion may consist of a single, double, or triple fusion [137]. Successful
fusion allows for compression of the two bones of the injured joint, to combine or “fuse” into a
single unit. By eliminating movement in the affected joint, patient pain may subside. Although
fusions successfully reduce pain, this method comes at the cost of range of motion. With this
restricted motion, abnormal gait patterns in patients treated with fusions tends to follow. In order
to compensate for this reduced motion at the ankle, neighboring joints, such as the hip or knee,

4
must work past their natural capability, which in turn, leads to accelerated degeneration and
development of osteoarthritis in these adjacent joints [68, 73, 91, 155].
An alternative to fusion is a total replacement of the ankle joint. The first total ankle
replacement was introduced in the 1970s; however, limited knowledge of ankle kinematics led to
flawed designs and consequently high failure rates. Additionally, poor manufacturing and limited
instrumentation for implantation, severely reduced any chance of successful implantation. The
combination of inferior implant material and sparse knowledge in surgical technique led to the
rapid abandonment of this procedure.
However, with the advancement of total joint replacements in recent years, total ankle
replacement has become a viable alternative to fusion once more [60, 151]. Data from 1998 to
2012 has shown up to a seven-fold increase in total ankle replacement utilization with 61.9% due
to arthritis [60, 139]. Studies have reported comparable reductions in pain for TAR and fusion
with the additional benefits of preserved joint mobility, improved gait, and improved
biomechanics of the ankle. Unfortunately, limited mid to long-term data is available as older
TAR designs often failed in vivo. Additionally, these studies compared older second-generation
total ankle replacements against archaic ankle fusion methods.
Current total ankle replacements consist of two designs: fixed bearing and mobile
bearing. Fixed bearing designs are comprised of two components: a metal tibial component with
a locked polyethylene bearing and a metal talar component. Conversely, mobile bearing designs
include three components: a metal tibial component, a mobile or floating polyethylene insert (flat
superior surface and concave inferior surface), and a metal talar component. Although fixed
bearing implants more accurately represent the natural ankle anatomy, transverse-plane rotations
(internal-external rotations) are more restricted compared to the mobile bearing design, causing

5
higher rotational stresses at the bone-implant interface [33, 114, 120]. This in turn may
contribute and lead to aseptic loosening, one of the main causes of total ankle replacement failure
[52, 80]. Theoretically, implementation of a mobile polyethylene insert allows for anterior-
posterior translation and axial rotation on the superior surface while still allowing dorsiflexion
and plantarflexion on the inferior surface. By allowing more degrees of freedom, a higher
conformity on both the superior and inferior surfaces is possible. This higher conformity allows
for maximum bearing contact area which is intended to reduce the contact stresses and lower the
generation of wear debris [113]. Furthermore, this additional degree of freedom is meant to
improve patient gait kinematics. Although the potential benefits pertaining to the reduction in
wear debris remain to be tested, investigators have reported that the clinical benefits of a mobile
design may not be as significant as intended. Specifically, Queen et al reported no differences in
gait mechanics or patient-reported pain at one year following surgery, between the two implant
designs [120].
While total ankle replacements may have potential benefits over arthrodesis, patient
selection remains critical for implant success. Many factors including age, weight, and health are
critical in determining if patients are suitable for total ankle replacements. The age of the patient
is often considered to determine the patient’s bone quality. While a patient that suffers from
osteoporosis and poor bone quality would not be a suitable candidate, patients younger than 50
have a 1.45 times higher chance of revision and a 2.65 times chance of failure [61]. Perhaps
intuitively, patient weight must also be considered. In general, obese patients (BMI >30kg/m
2
)
are typically excluded from total ankle replacement consideration, as the forces at the ankle can
easily exceed 4.5 times bodyweight. These increased forces are believed to increase the risk of
aseptic loosening leading to the early failure of the device. However, Barg et al. reported that in

6
123 total ankle replacements in obese patients, the survival rate at six years was 93%, with no
observed trend on obesity’s influence on the rate of aseptic loosening [11]. It should be noted, in
general total ankle survival rate at five years often exceeds 90% but declines steeply in ten years.
While age and weight factor into patient selection for total ankle replacement, the pivotal
indicator lies in the patient’s activity level. High impact activities are advised against in order to
avoid implant loosening, polyethylene wear, and periprosthetic fracture in the operated ankle
[19, 154]. Ideal candidates for total ankle replacement success would be older, non-obese, and
low activity level patients.
1.1.4 History of Total Ankle Replacement Design
Due to the continued dissatisfaction of poor mobility associated with arthrodesis, total
ankle replacements were developed. Total ankle replacements were initially introduced in 1973,
when Lord and Marotte performed the first total ankle replacement surgery. Their initial total
ankle replacement was an inverted hip replacement implanted within the tibial shaft while
completely removing the talus and implanting a cemented acetabular cup in the calcaneus[45,
88]. Expectedly, this did not produce a desired effect, with 12 failures and only seven of the 25
total procedures considered satisfactory. By eliminating the entire talus and reducing the ankle to
a ball and sphere, the ankle was left with three unnatural rotations about the calcaneus,
dramatically changing the natural biomechanics of the ankle. This abysmal outcome spurred
further technological development in the field and created the first generation of total ankle
replacements later in the 1970s.
The first generation of devices for total ankle replacements mostly consisted of a two-
component design which included a polyethylene tibial component in conjunction with a metal
talar component. These implants used cement to promote fixation and were generally designed to

7
be highly constrained, relying on the implant design for tibiotalar stability. However, it was later
discovered that overly constrained designs experienced high impact forces which eventually led
to loosening and failure. To produce a less constrained implant, the Richard Smith total ankle
replacement (Dow Corning, Arlington, TN, USA) reverted back to a ball and socket design.
However, an inverted hip replacement was not used and a specifically designed tibial socket was
implemented along with a convex sphere which was fitted to the superior surface of the talus.
While this implant attempted to address the over-constrained design flaws of other implants of its
time, the Richard Smith implant again failed. In the USA, the Irvine TAR tried to develop an
implant which complemented the natural anatomy of the talus, but this failed due to high
ligament stress and misalignment. Another notable design was created by Dr. Buechel and Dr.
Pappas which used ultrahigh molecular weight polyethylene for a cylindrical talar component
and a cobalt chromium tibial component. While this implant design shared the same fate of all
other first-generation devices, these devices heavily influenced future designs. In general, the
first generation of total ankle replacements were highly constrained, consisted of two
components (metal tibial component and polyethylene talus component), and used bone cement
to achieve proper fixation.
Following the disappointing outcomes of the first generation of total ankle replacements,
the second generation began to develop in the 1980s. Three main designs dominated this era: The
Agility by Depuy, Buechel-Pappas by Dr. Buechel and Dr. Pappas, and the Scandinavian total
ankle replacement by STAR which was later acquired by Stryker. The second generation of total
ankle replacements, transitioned to a metal component on both the tibia and talus, with a
polyethylene insert between the two components, reminiscent of a total knee replacement as
opposed to total hip replacement. This design change may have occurred due to the realization of

8
the similarities between the knee joint and ankle joint. This component change brought about the
development of fixed and mobile design implants. While both designs consist of the previously
mentioned components (metal tibial component, metal talus component, and polyethylene
insert), the fixed design rigidly immobilized the polyethylene insert to the tibial component.
Conversely, the mobile design allowed the polyethylene insert to float uninhibited between the
two metal components. While the mobile design does not resemble the natural ankle anatomy,
the additional degree of freedom it provides, from the superior articulating surface of the insert,
may reduce implant fracture. However, as a consequence of the additional components, large
bone resections became a common complication with these implants.
The second generation of total ankle replacements transitioned towards more anatomical
designs, with the intent of replicating kinematics seen in a natural ankle. This allowed for a semi-
or non-constrained design which tried to address the difficulties with fixation seen in the initial
generation of constrained implants. Initially, these implants were fixed with
polymethylmethacrylate (PMMA). Gradually porous fixation coatings were added to promote
stronger osseointegration without the need of PMMA. This movement away from the use of
PMMA stemmed from the realization that bone cement leads to osteolysis and in turn loosening.
While many of the issues with the first generation of total ankle replacements (highly constrained
and cemented) were addressed within the second generation, fixation and wear debris remained
problematic. Overall the second generation of total ankle replacements introduced fixed and
mobile designs, became more anatomically inclined, and implemented the use of porous fixation
coatings.
Over the following three decades, iterations of the three key second generation implants
have led to the current third generation designs. In North America today, there are seven primary

9
designs based on these implants: Agility, InBone I/II, Infinity, Eclipse, Salto Talaris, Zimmer
Trabecular, and STAR. Of these implants, all are fixed component design, (where the
polyethylene insert is rigidly attached to the metal tray) except for the STAR which is a mobile
design. Mobile designs are more commonly used in Europe, due to the strict FDA regulations in
America. However, STAR and the previously mentioned total ankle replacements are all FDA
approved to be implanted without bone cement. These designs have drastically improved from
the second-generation designs by eliminating the use of cement, reducing the amount of bone
resection with the development of implantation guides, and creating more accurate anatomical
designs. Although patient outcomes have improved from generation to generation, the
survivorship of these newer implant designs are still unsatisfactory with a 66% survivorship at 15
years [53]. Continued analysis of the outcomes and failures along with the establishment of a
proper pre-clinical testing method is vital to improving these implants further.
1.1.5 Clinical Outcome
Regardless of the advantages and disadvantages of the varying designs, a universal
problem plagues all total ankle replacements; poor survivorship. Due to the relative infancy of
current third generation TAR, very little survivorship data beyond 15 years has been published.
However, one study by Frigg et al found the STAR implant to have a 55% survival rate at 19
years [48]. Most survivorship studies published, report a survival rate of 88%-98% survival rate
at five years and survival rate of 57%-83% at ten years [18, 24, 30, 39, 48, 53, 72, 122, 127, 145,
161]. While these studies have encompassed several designs, a clear degradation of survivorship
can be seen from the five to ten-year mark. Interestingly, Roukis et al, studied the survival rate of
1
st
, 2
nd
, and 3
rd
generation total ankle replacements. As expected, the survival rates at five years
increased from 88% to 93% and the survival rates at ten-years increased from 76% to 83% from

10
the first generation to the third generation of implants [127]. While there is clearly still a problem
with long-term survival, the incremental changes that have been made over the past few
generations of total ankle replacement design have been beneficial to the overall progression of
this implant.
Total ankle fusions and total ankle replacements have both successfully helped patients
return to mobile pain free activities. However, total ankle replacements have been observed to
provide better range of motion, resulting in a more symmetrical gait cadence, and more normal
vertical ground reaction force patterns [47, 57, 116]. By providing better outcomes in
temporospatial parameters, TAR may prevent further degradation of neighboring joints which is
often associated with fusions of any joint. While TAR has the potential to become a more viable
treatment than fusion, TAR surgeries are often plagued with high revision rates [56, 85].
However, Benich et al found after a 3-year follow up, total ankle replacements provided a better
improvement of comfort and function compared to fusion, suggesting that this increase in quality
of life, may justify the additional surgery. This improvement was only observed with newer
generations of total ankle replacements [14]. Although outcomes for total ankle replacement are
steadily improving, more advancements are clearly needed for TARs to reach the clinical success
of total hip replacements and total knee replacements.
The success of total joint replacements such as THR and TKR is the result of decades of
design and development guided by many iterations driven by extensive research. In addition to in
vitro testing, these comprehensive investigations have included preclinical in vitro testing as a
function of fixation methods (cementing techniques, implant shape etc.), joint wear simulations
as a function of bearing surface materials, along with the analysis of retrievals (implants
removed from patients). From the literature it can be observed that there is a scarcity of research

11
in all facets of the previously mentioned testing modalities responsible for the success of other
total joint replacements. Accordingly, improvements to the current preclinical testing standards
are essential to help identify the success and shortcomings of future designs of total ankle
replacements.
1.2 In Vivo Biomechanics
1.2.1 Design
The fundamental goal of arthroplasty is to enable patients to return to healthy routine
activities. While many novel and established treatments may successfully reduce pain, they often
disrupt and alter the body’s natural biomechanical movements. As the body adapts, new
complications often stem from these changes, leading to subsequent problems. Therefore,
restoration of a patient’s natural biomechanics is often as important as pain reduction for a
successful outcome. Current total ankle replacements have allowed improvements to both pain
and the primary range of motion for the ankle; however, many of these solutions continue to fail
long-term. This further points to the need for a solution that comprehensively restores the injured
joint to its natural conditions and all degrees of freedom.
In order to address the biomechanical problems involved in a diseased or injured joint, a
thorough knowledge and understanding of the biomechanics of a healthy joint are vital.
Specifically, knowledge of joint kinematics and joint forces during different activities must be
identified and quantified. This requires application of three-dimensional measurements of
kinematics, followed by calculation of joint forces or, alternatively, direct measurement of joint
forces using instrumented implants or sensors.
The first attempt to estimate joint forces involved the hip joint was purely based on
theoretical calculations using free body diagrams. Specifically, in 1947 Inman estimated hip joint

12
reaction forces during single legged stance by including bodyweight and resistance provided by
the hip abductor muscle [71]. It was not until 1976 when major advances were made by John
Paul in Strathclyde University [112]. A pioneer in the field of orthopaedic biomechanics, Paul
established the fundamental methods to estimate joint forces using external measurements.
Specifically, he used force plates equipped with six degree-of-freedom sensors to measure
ground reaction forces of a subject, during different activities. In conjunction, two cameras
(sagittal and lateral views) were used to capture 70 frames during the activity [112]. From these
signals, Paul was able to calculate the accelerations of each joint, in turn calculating the forces
transmitted through each joint throughout the body. The basis of current optical motion capture
systems in conjunction with force plates have been designed from Paul’s work to properly
understand the control and dynamics of joints during various objectives [26, 49, 93].
In an attempt to record and verify in vivo joint force calculations from Paul’s methods,
Bergmann et al have compiled a database of data from patients with instrumented hips, knees,
and other implants [54]. Bergmann et al. greatly contributed to the understanding of in vivo joint
forces during various everyday tasks including walking, stair climbing, and standing. While data
from instrumented implants is invaluable, due to the intensive methods to reproduce these
results, very few other studies can replicate these experimental methods and instead just apply
the basics introduced by Paul.
Optical motion capture systems involve multiple cameras which track a patient’s
movements based on markers strategically placed on bony landmarks. These markers allow the
systems to calculate the patient’s movements (kinematics) while performing goal-oriented tasks.
Markers may be active or passive. Active markers use a pulsing light emitting diode to allow the
camera system to track the patient. Conversely, passive markers rely solely on retroreflective

13
material to reflect light. Although active markers provide signals with less noise, the system is
cumbersome and requires patients to wear wires and electronics; whereas passive markers allow
unrestricted movement.
While the optical motion tracker records the patient’s kinematics, force plates can be used
in conjunction to record the applied ground reaction forces (kinetics) in all three planes of
movement. From these kinetics, researchers and scientist can begin to understand how a subject
interacts with the ground in order to achieve their goal-oriented task. With the ground reaction
force data, inverse dynamics can be used to calculate the individual joint reaction forces. By
understanding the ankle, knee, and hip joint’s individual contribution, we can understand more
clearly how a subject is achieving specific tasks. This data allows the visualization of how a
patient in pain compensates with neighboring joints to achieve their intended task. Additionally,
if the patient was seen before surgical intervention, their recovery progress back to normal
biomechanics can be tracked and recorded. Although inverse dynamics allow joint reaction
forces to be calculated fairly accurately, this method is just an estimation. To acquire direct joint
reaction forces, instrumented implants or pressure sensors must be used within the joint space.
Another tool used to acquire in vivo data is electromyography (EMG). EMG consists of
surface probes which record muscle contraction signals. These signals show which muscles are
being recruited to achieve the desired task. Although this method is beneficial in indicating
which muscles are contracting, the usefulness of the data is limited. In addition to being noisy
and requiring heavy processing, EMG signals are highly subjective to the patient, along with the
placement of the surface probes. For example, an EMG signal may indicate the quadriceps being
recruited at a higher rate than the opposing hamstrings, but the amount of recruitment may only
be quantified relative to each muscle. Along with this limitation, the EMG signal is highly reliant

14
on the surface probes having sufficient contact and being correctly placed parallel to the muscle
fibers.
1.2.2 Limitations
Using optical motion tracking, force plates, and electromyography signals, a deep
understanding of how healthy and injured patients achieve certain tasks can be studied. With this
information, scientist can evaluate how well implants are performing in vivo, along with how
well patients are recovering.
While the use of optical motion tracking systems is valid for many fields of research,
there are several limitations and assumptions. The most important limitation is the use of skin
mounted markers. Although retroreflective markers are placed on bony landmarks, movement
may cause the skin to shift and distort the data. This inaccuracy is compounded when attempting
to record the kinematics of specific bones, such as the talus, which is not directly adjacent to the
skin. Perhaps the most accurate measurement system is the use of intracortical bone screws
placed into specific bones to mount reflective markers rigidly attached to each bone. These
markers allow continuous three-dimensional motion tracking of the rigid bones they are attached
to. Unfortunately, this method is highly invasive and carries the risk of discomfort, infection, or
other clinical complications. Therefore, it has limited practicality, and has only been used in
small sample sizes with limited characteristics [33, 113, 114].
Along with the limitation of being mounted to the skin, the movement of these markers
must be recorded with a plethora of cameras. These cameras must be mounted in specific
orientations in the room, which constrains all data collections to constrained settings. While this
is not an issue with simple gait studies, this problem arises in other activities that are unable to be
presented in small inhibited rooms.

15
1.2.3 Applications
Using motion capture and force plates, the field of biomechanics has steadily begun to
analyze and understand how the body accomplishes desired tasks [99-102]. From the knowledge
acquired from studying healthy patients, in vivo studies can be expanded to help injured subjects.
Movements of injured subjects can be studied and compared to healthy subjects to determine
suitable treatment for these patients. This analysis may allow for modification of the injured
subject’s movements through strengthening of lacking muscles, or through changes in their
mechanical leverages rather than resorting to invasive surgeries. Along with helping injured
patients, motion analysis can also benefit healthy patients who are at risk of injury. Subjects can
be asked to perform specific movements, from which risk of injury can be assessed. This can
allow coaches or doctors to adapt the subject’s actions for injury prevention.
However, when surgical intervention is unavoidable, in vivo studies continue to provide
beneficial information. In vivo studies of patients who have undergone surgical procedures can
evaluate the biomechanical changes that occur by quantifying how a patient’s range of motion is
restored along with how closely their kinematics resemble a healthy subject’s kinematics [14, 56,
85, 120]. This data may give researchers insight on how a patient’s body utilizes an implant, and
how it compensates for the difference in the natural anatomy. From here, modifications to the
implant can be implemented to help improve the safety and efficacy of the implant in vivo.
Along with motion capture and force plates, many other tools can be used for in vivo
research. For instance, brace compliance is a crucial factor in successful treatment for diseases
such as multiple sclerosis and clubfoot [1, 121, 133, 147]. These studies implemented pressure
sensors along with temperature sensors to objectively record brace wear. This allowed
researchers to quantify the precise number of hours the patients were compliant with wearing the

16
brace along with quantifying the minimum number of hours needed to prevent relapse or
reoccurrence of the unwanted symptoms.
In addition to pressure sensors and temperature sensors, inertial measurement units
(IMUs) are also being used for in vivo studies. IMUs contain a combination of accelerometers,
magnetometers, and gyroscopes which work in conjunction to measure the kinematics of a
subject. By placing IMUs on specific limbs, IMUs provide similar data to motion tracker
analysis, but allows for movement to be recorded outside, without the physical constraints of a
laboratory. While IMUs are helpful, they cannot provide force data and often require reliable
connection to a home module to successfully record data.
Using a combination of tools, in vivo studies provide valuable data and insight on how
healthy and injured subjects achieve tasks. Although useful to understanding how the body
interacts with the environment, there are many questions that remain unanswered solely from in
vivo research. Therefore, additional approaches such as in vitro cadaver studies are essential for a
more granular view of what occurs within the body.
1.3 In Vitro Biomechanics
1.3.1 Design
While in vivo studies are vital to testing and evaluating implants, many factors prevent
this method from providing a comprehensive evaluation. The main disadvantage with in vivo
studies pertains to the inability to directly track specific bones within the body. While optical
motion trackers can be attached directly to the cortical bone, this method is highly invasive and
not feasible for living patients. Specifically, percutaneous fixation would introduce a high risk of
infection. Alternatively, in vitro research using cadavers is a useful method when invasive
techniques must be implemented.

17
The methods applied during in vitro studies, parallel in vivo studies with their use of force
plates and optical motion trackers. Although both methods use optical motion trackers, cadaver
studies can place motion tracker flags directly in bones via bone screws. Radiographs are often
used to confirm bone screw placement, and any ambiguity of bone movement seen with surface
markers during in vivo studies, can be reduced. Similar to in vivo testing, active and passive
markers can also be used for in vitro testing. However, unlike in vivo testing, it is simple to use
wired active markers on cadavers. Since cadavers are loaded in an apparatus, there are many
stable places to mount the required wires and modules associated with active markers. This
provides a more stable and stronger signal than passive markers which rely solely on reflective
material.
Custom apparatuses are commonly manufactured to properly and sufficiently apply loads
and moments to the cadaver. By applying physiological loads to cadavers, replication of the
natural joint kinematics measured from in vivo studies can be attempted in vitro. Similar to in
vivo experiments, load cells can be used to obtain the force data seen by the specimen. For in
vitro studies, smaller more targeted load cells may be used to increase precision. Load cells
contain strain gauges that can calculate the amount of force being applied to the load cell based
off the amount of deformation the strain gauge experiences. Depending on the complexity of the
load cell, forces can be recorded in 3D space. This force data can then be analyzed in
conjunction with the motions of the bones recorded from the optical motion tracker and used to
evaluate implant performance.
Another strength of in vitro testing is the ability to use synthetic materials that have the
same material properties as bone. When implant research does not need to take into account
ligaments or tissues, model bones with similar material property as natural bones can be used.

18
Using synthetic bones allows for uniform testing conditions, which is not possible with cadaver
bones due to each specimen’s unique anatomy. Synthetic specimens can be produced to have
uniform material properties, but can also be produced to have the same fractures or deformities,
eliminating variation among specimens due to bone size, morphology, cortical thickness and
strength. These models have been validated and are well tested and widely used when
appropriate [34, 35, 42, 43, 132, 134-136]. However, for joints which rely heavily on soft tissue
support, cadaver studies are more appropriate.
1.3.2 Limitations
Although in vitro cadaver studies are a vital step for implant success, there are many
limitations. The first limitation of cadaver studies is the specimen quality. The majority of
cadavers tested are donated from elderly subjects. Accordingly, the bones often have low bone
density making them somewhat fragile. In addition to poor bone quality, the lack of active
muscle and tendon forces also inhibit the cadavers from being able to withstand full in vivo
forces. Consequently, in vitro studies using cadavers must apply lower loads than would
normally be seen in a living patient. It is generally accepted that a cadaver specimen will not
sustain the same magnitude of forces as an in vivo subject. This limits the validity of the actual
kinematic data from these tests. Additionally, at lower rates of loading, ligaments provide the
majority of stabilization, while at higher loading bone shape provides the majority of
stabilization [157]. Without access to expensive specialized freeze clamps, ligament activation is
difficult. This may alter the results of the specimen’s reaction to loading.

19
1.3.3 Applications
“The following experiments were designed to test in vitro whether the load-bearing capacity of a
prosthesis cemented into bone accorded with claims made from clinical observations.”
– Dr. J Charnley.
Synonymous with forging the orthopaedic implant field, Dr. Charnley was one of the first
orthopaedic surgeons to successfully treat arthritis of the hip. His work included studying the
frictional properties of a healthy non-arthritic hip in order to reproduce this in an artificial hip.
Another notable contribution was the implementation of PMMA to attach hip implants to their
respective bony surfaces in order to achieve immediate and lasting fixation which had previously
been a common source of failure. Through innovation and abundant in vitro research, Dr.
Charnley was able to achieve these monumental achievements in hip replacements. Dr.
Charnley’s success has helped establish the significance of thorough in vitro testing for all
successful implants. By accurately recreating and applying in vivo conditions to implants,
researchers may gain insight on an implant’s performance within a patient.
The main objective of in vitro testing of implants is to ensure that the biomechanical
performance of the replacement joint, is comparable to the natural anatomical joint. These tests
are often implemented on human cadavers which are tested in its natural state to establish a
baseline of the specimen’s kinematic range of motion. From here, the injury can be recreated in
the specimen and the implant can be placed within the cadaver. While true ingrowth of the
implant within the specimen will not occur, the specimen will still give valuable insight into the
implant’s efficacy at restoring range of motion.
Forma&ed: Keep lines together

20
In addition to quantifying and evaluating implant kinematic performance, other major
topics of research is the quantification of short term fixation [64, 77]. For example, while joint
replacements are expected to survive for the lifetime of the patient, fracture fixation devices need
only stabilize the fracture until the bone heals. These studies employ similar methods of motion
trackers, load cells, and custom testing apparatuses. However, the restriction of movement is the
goal. Unlike joint replacement implants, these plate and screw implants strive to stabilize bones
to provide proper fixation which is vital for proper healing. Specimens are often tested intact and
unmarred to retrieve an initial baseline of fixation that the cadaver naturally possess. Following
intact testing, the injury being studied can be inflicted on the cadaver. Creating an injury may be
as simple as cutting a ligament with a scalpel or as complex as producing unique bone fractures.
The specimens are once again tested, but in an injured state. From here the fixation device can be
implanted in the specimen by a doctor or resident familiar with the surgery. The specimen can
then be tested, and the fixation capabilities of the device being investigated, can be analyzed.
Although complete ligamentous fracture healing cannot be replicated through cadavers, the
stability and quality of bone fixation can be evaluated with the use of motion trackers.
Another heavily studied area of in vitro biomechanical research, is risk of fracture and
implant micromotion within the bone. Small wire sensors, configured in specific orientations,
called strain gauges can be strategically placed to measure the microstrain experienced on the
bone’s surface during loading. Strain gauges can be used to correlate the maximum amount of
strain with bone quality, implant type, and many other factors. These studies can then help assess
the risk of bone fracture due to different variables. Along with fracture risk studies, micromotion
studies are often performed. Micromotion studies use displacement variable resistance
transducers (DVRTs), linear variable differential transducers (LVDTs), or displacement

21
transducers to measure the movement of the implant within the bone. DVRTs consists of a small
metal rod that changes the sensor’s electrical resistance when displaced. This small amount of
resistance change allows the researcher to calculate the actual micromotion displacement of the
implant within the bone.
Many of these in vitro studies contribute to the overall success of implant performance.
By investigating the behaviors of implants in vitro, data driven improvements can be developed
leading to higher success rates in vivo and fewer revision rates.
1.4 Wear Testing
1.4.1 Design
Wear tests are often performed in order to assess the intended lifespan of implanted total
joint replacements. These tests are performed for millions of cycles with artificial lubrication and
standardized load and motion profiles to simulate in vivo conditions. To quantify the amount of
wear that may potentially occur in vivo after years of use, several methods are used.
Wear is quantified using several techniques. The most common method implemented to
quantify wear is gravimetrical analysis. For this test, precision scales are used to measure the
mass of the implant before and after the test is performed. Another popular method used to
quantify wear is through volumetric analysis. This technique utilizes coordinate measuring
machines or Talyrond machines which may use a laser or contact stylus to scan the implant [16,
89, 150]. The machine is programmed to take discreet points and reassemble them to create
three-dimensional point cloud models of the implant. The implant is scanned before and after the
cyclic wear test to create a wear map, showing and quantifying sections of high wear. These
wear maps are then used to evaluate and critique the articulations between varying components.

22
The required number of simulations depends on various factors such as the type of device
and anatomical location. Types of devices could range from fixation plates, which should allow
very little movements, to joint replacements that are designed for millions of articulations.
Anatomical location can vary the amount of forces seen on the implant. For example, a clavicle
fixation plate should see low forces and little rotation, while a hip experiences much higher
forces and a high magnitude of rotations daily.
To perform a wear test, in vivo kinematics and kinetics are applied to the implant for
several millions of cycles. These implants are submerged in a lubricant such as water or bovine
fluid to properly replicate synovial fluid that is naturally seen in vivo. The type of lubricant used
can significantly alter the results. It is also known that the rate of wear can be affected by the
way in which the load and displacement parameters are controlled. Specifically, a test may be
conducted by controlling the force, allowing the peak displacements to change as the friction
between the materials changes over the duration of the test. Alternatively, the test could be
implemented under displacement control, restricting any changes in displacement as the friction
changes over the duration of the test. Two types of wear control tests can be performed: Force
control and Displacement control. Both methods are similar in that flexion and extension are
prescribed in angular displacement control and axial loads are prescribed in force control. Where
the two methods differ lies in the simulation of the anterior-posterior translations and internal-
external rotations [123].
Depending on the joint being studied, displacement control may produce less wear since
the established parameters are repeated regardless of force and may result in a decrease in
contact stresses due to wear, deformation and creep. Conversely, force control may produce
higher wear rates since the movements rely solely on the geometry of the implant’s design and

23
can increase the parameters needed to achieve the proper forces when the environment and
materials adapts, therefore increasing the wear. Displacement controlled wear testing is highly
suitable for symmetrical joints such as the hip where supporting stabilization is not needed. On
the other hand, for irregularly shaped joints such as the knee and ankle, force-controlled wear
testing may be more beneficial [123, 124]. However, there is very little evidence showing a
universally agreed upon testing method for either.
To illustrate this fact, there are two ISO standards established for wear testing of the knee
ISO 14243-1 for force control testing and ISO 14243-3 for displacement control testing. In 2010,
Sutton et al reported testing knee cadavers following the force control standard and found all the
produced kinematics to widely vary from the displacement control standard [146]. Clearly there
is much ambiguity due to a lack of research in the proper method of testing these more complex
shaped joints. More research must be performed for the society to come to a clear consensus on
the proper methodology to test knee and ankle joint replacement implants.
1.4.2 Assumptions
Similar to other testing techniques, several assumptions and limitations are made during
wear testing. The first limitation of this test is the ability to properly apply the proper joint
kinematics to the joint replacement. If inaccurate kinematics are applied to the implant, the wear
location and quantity can be misrepresented. While gravimetrical analysis is able to quantify the
amount of wear, it is unable to report the location of the wear. By determining the sites of high
wear with wear maps, more information can be analyzed on the kinematics of the implant. While
volumetric analysis has the ability to produce these wear maps, assumptions of the unmarred
implant must be made. Unless the implant was scanned before implantation, it must be assumed
the implant conformed to manufacturing specifications.

24
1.4.3 Applications
1.4.3.1 Accelerated Wear Test

Wear fatigue tests have become an essential step in bringing a product to market. In
order to successfully show that an implant will survive once implanted within a patient, the
principles mentioned above must be implemented to the best of the company’s abilities.
However, performing a wear test does not guarantee that the product will survive in the patient.
This is the current predicament that total ankle replacements are experiencing. Although
manufacturers are obtaining promising outcomes from their wear tests, conflicting results have
been observed once implanted in patients.
These widely conflicting outcomes clearly show the importance of properly validating
implant design through biomechanical and wear tests. As Sutton et al. have illustrated, wear
results can be skewed and manipulated based on whether a displacement or load profile is
applied. Furthermore, within displacement style testing, the results of a wear study can be
heavily influenced by the prescribed motion applied. Saikko et al. reported that total hip
replacements depend more on multidirectional motions and displacements than specific types of
load profiles for replicating realistic wear mechanisms in vitro [128]. This group showed that
changing the implemented cross path motion increased wear by a magnitude of up to two-fold.
Their findings may lend support to the pronounced variance seen in patients with individualized
non-uniform walking patterns, where some patients may apply greater cross path motion
resulting in higher rates of wear leading to a higher chance of failure. However, as previously
mentioned, Sutton et al compared the displacements of a knee cadaver as a result of force control
testing, and found the displacements to be up to 2.5 times greater than the prescribed movements
in displacement control [146]. Following this logic, force control wear test of the knee would

25
cause greater wear due to the higher multidirectional cross path motion. This has sparked wide
debate within the community on the proper methodology of testing total knee replacements.
Naturally these studies highlight the same complications that must be solved for testing total
ankle replacements. Due to the complex nature of the ankle, specifically the talus, there is very
little knowledge on the kinematics of the talus. As a direct result of this lack of information a
plethora of inadequate wear studies have been performed, leading to continual long-term failure
of total ankle replacements.
For hip and knee joint replacements, the influence of crossing-path motion on UHMWPE
has been heavily investigated [15, 41, 83, 128]. These studies have shown that with minimal
increases in crossing-path motion, polyethylene wear may increase substantially. For total knee
replacements, Laurent et al. found up to a 124% increase in wear when comparing the minute
differences (3.7° difference) in cross-path motion between the medial and lateral condyles. In
addition to crossing-path motion, consideration of the metal alloy and metal surface finish used
for the implant, must be taken into account when attempting to reduce wear. While it is well
established that metal-on-metal implants are able to successfully withstand wear in
multidirectional motion, metal-on-poly implants fail miserably under the same conditions [83].
However, it has been shown for total knee replacements, that metal surface roughness has the
greatest influence on wear at the tibial tray over both crossing-path motion and metal alloy
combinations [15]. This group reported that polishing the metal surface of the tibial tray was far
more effective than reducing the crossing-path motion or changing the metal alloy type. These
studies were critical to establishing the effects of crossing-path motion, metal alloy, and metal
surface finish for total hip and total knee replacements. However, there is a deep scarcity of
research exploring the effects of these three factors in total ankle replacements. While the

26
knowledge pertaining to metal alloys and metal surface finishes can be heavily borrowed from
total hip and total knee replacements, the greatest difference between the implants is the natural
kinematics of each joint. Therefore, the biggest strides in improving total ankle replacements
may come from learning how to properly reproduce the natural kinematics of the talus.
1.4.3.2 Retrieval Analysis

From these in vivo failures, the second component of wear testing can be assessed. The
analysis of wear that has occurred in vivo is invaluable to the progression of joint replacement
success. Implants that have been taken out of patients (retrievals), can illustrate its functionality
under genuine physiological loading. Historically, retrieval inspection of failed total hip
replacements played a critical role in incremental improvements to the understanding of the
modes of failure. Through careful examination of retrieved total hip replacements, invaluable
information on the creation of in vivo wear and its minimization through improved materials and
implant design led to high survivorship and longer lasting implants [97].
Similarly, many of the incremental improvements seen from sequential generations of
TARs can be attributed to retrieval analysis. For the first generation of TARs, retrievals revealed
that the overly constrained designs resulted in high impact forces which led to loosening and
failure. This knowledge led to second generation TARs to adopt a more anatomical shape along
with the introduction of a fixed and mobile design to reduce impact and rotational forces.
However, these implants were quite large and required high bone resections. Analysis of these
retrievals showed weak osseointegration, due to the subpar bone density and strength of the
tibia’s distal diaphysis compared to its epiphysis [69]. From these findings, the current
generation of TARs aimed to reduce impact forces and rotational torques along with achieving
better osseointegration, by creating more anatomically designed implants which require reduced

27
bone resections. Analysis of retrieved TARs over the past half century have helped slightly
improve performance; however, TARs continue to suffer from abysmal survivorship [31, 48, 53].
Further analysis on current TAR retrievals must be performed to achieve survivorship that rivals
the success of total hip replacements.
Additionally, while reduction in pain and restoration of a patient’s mobility are high
indicators of a joint replacement’s success, it has been observed from hip and knee replacements
that potential problems may still persist, and additional analysis must be performed. For
example, while total hip replacements allowed patients to return to daily activities, the contact
mechanics and occurrence of edge loading in modular metal-on-polyethylene created increased
peak contact stresses leading to accelerated failure which was unobservable purely through gait
kinematics [110, 111]. Another prominent example highlighting the value of observing the
specific joint kinematics through retrieval analysis, is the discovery of the effect of cross-path
motion on total hip replacement survivorship. While cross-path motion was unobservable from a
patient’s everyday activities, high wear rates of conventional polyethylene led to high failure
rates [95, 129]. Similarly, while studies have shown successful restoration of ankle ROM,
potential problems, unobservable through in vivo and in vitro experiments, may still be present.
While retrieval analysis of TARs would significantly contribute valuable insight into the
in vivo failure mechanisms of TARs, very few studies have been published on this subject. This
may be due to the low utilization of TARs which has resulted in an extremely limited database of
retrievals. Following the history of total hip replacements, it is clear that wear test across stages
of total ankle replacement development must be improved. The proper profile between force
control and displacement control must be thoroughly investigated. In addition to this, retrieval
analysis of the wear from implants that have failed must be analyzed carefully. These wear

28
studies in conjunction with in vivo and in vitro wear studies will greatly benefit and progress the
field of total ankle replacements.
1.5 Applying Cadaveric and Wear Testing for Total Ankle Replacement
Biomechanics

1.5.1 Osteoarthritis in the Ankle
Osteoarthritis occurs in any joint space, with the deterioration of the joint’s natural
cartilage. This lack of cushioning may lead to pain and inflammation of the joint. While primary
osteoarthritis is the most common indication for total hip and total knee arthroplasty, post-
traumatic arthritis is the most common indication for ankle arthrodesis [90, 131]. Post-traumatic
arthritis occurs from previous ankle fractures or recurrent ligamentous instability [131].
Studies have reported healthy patients to be able to reach an ankle range of motion
(ROM) of up to 32° in the sagittal plane (~ 10° Dorsiflexion, 20° Plantarflexion) [22, 50, 67,
160]. These patients also experience a maximum vertical ground reaction force (vGRF) of 1.2
times body weight (BW) and a walking velocity of 1.28 ± 0.09 m/s [47, 67, 138]. Expectedly,
with the onset of osteoarthritis, there is a clear degradation in the ankle ROM and kinetics. Flavin
et al found injured patients to have a sagittal ROM of 15.6° ± 3.8° pre-operatively, along with a
much slower velocity of 0.74 ± 0.28 m/s, almost half of a healthy patient [47]. Along with a loss
of range of motion, a patient’s speed and cadence is also negatively impacted. These problems
can lead to further debilitation if not treated correctly. This reduction in ROM results in a
variance of a patient’s natural kinematics which may lead to further injury in surrounding joints.
1.5.2 Clinical and Societal Impact of Osteoarthritis
Across the world over 100 million people suffer from osteoarthritis, including over 22
million people in North America alone [25, 84]. Woolf et al have predicted that by 2020

29
osteoarthritis will become one of the most prevalent causes of disability worldwide [159]. The
consequence of living with osteoarthritis results in limited movement at the affected joint, which
in turn limits activities and daily functions. Although only 6% to 15% of all cases of
osteoarthritis involve the ankle joint, it has been found that the severity of pain and loss of
function associated with this condition, is as detrimental to the patient as osteoarthritis at the hip
[25, 40, 51, 115]. Glazebrook et al along with Agel et al have reported patients who suffered
from end-stage ankle arthrosis, had physical function and pain scores two standard deviations
below the general population scores, leaving patients with difficulty accomplishing tasks such as
walking 100 yards or ascending one flight of stairs [4, 51]. Historically, physicians have helped
patients eliminate this pain through arthrodesis or fusion of the joint. Arthrodesis abolishes any
movements between these bones, in turn reducing the debilitating pain. However, this method
leads the patient to suffer from further degeneration in neighboring joints and great loss of range
of motion. As total ankle replacements (TAR) have slowly improved over the past few decades,
TARs have become a more popular method of treatment [60, 151].
1.5.3 Overview of Total Ankle Replacement
Current total ankle replacements consist of two designs: fixed bearing and mobile
bearing. Fixed bearing designs are comprised of two components: a metal tibial component with
a fixed inferior polyethylene bearing and a metal talar component. Conversely, mobile bearing
designs include three components: a metal tibial component, a mobile polyethylene insert (flat
superior surface and concave inferior surface), and a metal talar component. Although fixed
bearing implants more accurately represent normal ankle anatomy, transverse-plane rotations
(internal and external rotations) are restricted causing higher rotational stresses at the bone-
implant interface [33, 114, 120]. This in turn may lead to aseptic loosening, one of the main

30
causes of TAR failure [52, 80]. Theoretically, implementation of a mobile polyethylene insert
allows for anteroposterior translation and axial rotation on the superior surface, leading to higher
conformity on both the superior and inferior surfaces. This higher conformity allows for
maximum bearing contact area which potentially reduces wear rates and improves ankle
kinematics [113]. Although mobile bearing TAR may have potential wear benefits over fixed
bearing TAR, Queen et al found no statistical or clinical differences in gait mechanics or pain
between the two implant designs.
Recent improvements in total ankle replacements have led to an increase in popularity
allowing for standard treatment of ankles to shift from arthrodesis to arthroplasty. A recent study
by Sabour et al showed an increase of 31% in total ankle replacements performed from 2007
(14%) to 2013 (45%). Specifically, for patients suffering from post-traumatic osteoarthritis,
surgery rate was 12 times higher in 2013, for patients with primary osteoarthritis 4.93 times, and
for patients with rheumatoid arthritis 3.12 times.
Upon intervention with a total ankle replacement, range of motion and cadence was
improved to 19.2° ± 6.0° and 0.98 ± 0.13 m/s respectively [47]. Similarly, Queen et al reported
patients experiencing a conservation of ROM following TAR surgery, while also observing an
improvement in walking speed from 0.86 ± 0.04 m/s to 1.13 ± 0.03 m/s after 1-year post
operation [119]. In addition to reporting an increase in ROM and walking speed, Robin et al also
recorded a significant increase in peak vGRF from 1.25 ± 0.04 times BW to 1.31 ± 0.04 times
BW and 1.18 ± 0.04 times BW to 1.27 ± 0.04 times BW for the first and second half of the
stance phase respectively [119]. These studies along with many others have indicated that
multiple parameters contributing to ankle biomechanical health, improve with the use of TAR
following injury.

31
Although there has been an increased trend of total ankle replacement utilization, several
studies have noted the abysmal survivorship of these implants. These studies have found total
ankle replacements to have a 90% survivorship within the first 5 years, but have found a sharp
drop off from midterm to long term survival rate (65%) [56, 59, 78, 82, 122]. While the current
generation of total ankle replacements have high initial success, the sharp decline in long term
survivorship impedes TARs from becoming the universal standard of treatment for ankle
osteoarthritis.
1.5.4 Benefits and Applicability of Cadaveric and Wear Testing for Total Ankle Replacements
Cadaveric testing is commonly used to explore and improve implant performance prior to
clinical use. With the use of cadavers, the functionality of different total ankle replacement
designs can be verified and tested against ankle kinematics. In vivo load profiles and kinematics
can be applied to cadavers to accurately test these implants. Another benefit of in vitro testing is
the ability to invasively place motion trackers into specific bones with no repercussions. These
motion trackers enable kinematic data of the talus to be recorded and analyzed, which is difficult
to do with skin mounted markers placed on human subjects.
Wear testing can also be used to help predict the long-term success of new total ankle
replacement designs and materials. By applying millions of cycles of motion and loads to these
implants, the long-term survivorship can be explored. Researchers can predict how the device
will behave after being implanted in patients for multiple years without causing harm or failure
within patients. This will allow for any unforeseen problems to be addressed before implantation
in patients, in turn helping reduce revision surgeries of failed total ankle replacement implants.

32
1.6 Limitations of Cadaveric and Wear Testing Methods
While cadaveric and wear tests are essential aspects of preclinical testing, there are
inherent limitations to these tests. These limitations can cause poor testing methods to be adopted
when exploring total ankle replacements. These inadequate methods can lead to inaccurate
results which may explain the current pitiable long-term survivorship of total ankle replacements
in patients.
1.6.1 Limitations of Cadaveric Testing in Total Ankle Replacements
In addition to the general limitations pertaining to in vitro cadaver testing, such as poor
specimen quality, custom testing apparatuses and low testing loads, ankle cadaveric testing
introduces several other hindrances. While the hip joint includes the interaction of two bones
(pelvis and femur) and the knee includes the interaction of four bones (femur, tibia, fibula,
patella), the ankle consists of the interaction of four bones (tibia, fibula, talus, and calcaneus).
These bones allow the hip joint 3 degrees of freedom, (flexion and extension, abduction and
adduction, and internal and external rotation) while the ankle joint has 4 degrees of freedom
(dorsiflexion/plantarflexion, interior/exterior rotation, inversion/eversion, anterior/posterior
displacement). While each additional bone contributes to increasing the degrees of freedom in
vivo, the relatively large number of bones severely increases the complexity of properly testing
the ankle joint. With fewer degrees of freedom, simulating in vivo hip kinematics and kinetics is
relatively straight forward. Conversely, the ankle joint requires a complex balancing of plates
and cables to achieve temporary static equilibrium during testing. Many studies have avoided
this stability problem by fusing specific bones in the ankle joint or locking the testing platform,
allowing for a load to be applied to the joint without slippage [12, 68, 104]. However, by

33
introducing these stabilizing parameters, additional forces are applied to the ankle joint, in turn
skewing the produced kinematics.
In addition to the number of bones involved with the ankle joint, the bone morphology of
the talus itself, contributes to the instability predicament.

Figure 1. A. Superior view, B. Inferior view, C. Lateral view and D. Medial view of the talus.

34
Unlike the symmetrical spherical surfaces of the hip, the talus has an irregular shape and contains
no congruencies. This asymmetry makes it difficult to predict which surfaces allow rotations and
which allow translations, in turn making it difficult to accurately replicate in vivo kinematics
with different implant designs.
The final significant limitation specific to testing cadaveric ankles stems from the high
incidence of ankle sprain and injury. It is estimated that in the United States approximately
23,000 ankle injuries occur every day, with ankle sprains being the most common injury in
sports [125]. Although many of these injuries occur before the age of 35, there is up to a 30%
chance of residual symptoms [125]. Along with chronic symptoms, as patients get older, range of
motion is decreased with a reduction in every day activity. These natural factors make it difficult
to obtain cadaver specimens that are in pristine condition. Although cadaver specimens can be
screened with some exclusion criteria, the chance of having unmarred ankle specimens is very
low.
1.6.2 Limitation of Wear Testing in Total Ankle Replacements
While wear tests provide great insight into implant survivorship, there are still many
limitations specific to TAR wear tests. The key limitation with wear test for TARs is the lack of
an established testing protocol. Of the few wear studies published, there is a large variation in
loading and motion parameters, making the results difficult to directly and meaningfully
compare. Another important limitation of current TAR wear test is the ambiguity of applying
testing parameters with force control or displacement control. Due to the high reliance on
ligament support in the ankle, force control may produce higher displacements as the ligaments
stretch over time. For hip and knee joint replacements, studies have shown that with minimal

35
increases in crossing-path displacement, polyethylene wear may increase substantially, thus the
accuracy of replicating natural kinematics must be prioritized. [15, 41, 83, 128].
1.7 The Need for Improved Methods for Total Ankle Replacement Testing
Although potential positive outcomes have been shown using TAR, clinical outcome
studies have also highlighted some drawbacks that have impeded TAR from becoming the
universal solution to end-stage arthritis in the ankle. Specifically, it has been reported that
infection and aseptic loosening are the most common failure mechanism for TAR [52, 66, 80,
103, 141].
Historically, in hip and knee replacements, polyethylene wear debris produced at the joint
initiated a cascade of biomechanical events leading to osteolysis and aseptic loosening [70, 141].
In response to this important issue, highly wear-resistant polyethylenes were developed by
McKellop et al and others in the early 1990s, and have been used since the late 1990s [96]. The
polyethylene was developed through a process of highly crosslinking polyethylene by gamma
radiation or a chemical treatment, while importantly also preventing oxidation with time. Today
nearly all hip replacements take advantage of highly crosslinked polyethylene which has proven
to produce substantially less wear debris than previous polyethylene [92, 95]. However, while
highly crosslinked polyethylene decreases the production of wear debris, the fatigue strength of
the treated material has been found to be drastically reduced.
Due to the nature of higher joint forces in the knee, there has been major concern
regarding the fatigue strength of the polyethylene inserts used for these joint replacements. This
has prevented the same type of treated crosslinked polyethylene, which may produce lower
polyethylene wear rates in THR, to not be adopted [46, 81, 106, 126]. Instead, polyethylene has
been modified for increasing the wear resistant properties and preventing oxidation in vivo using

36
additives such as Vitamin E [20, 108]. An even higher degree of concern can be raised for TAR.
Specifically, the polyethylene insert is subjected to even higher stresses than hip and knee joint
replacements. This higher stress can be attributed to the higher joint forces observed in the ankle,
along with the fact that the physical size of the insert is smaller, due to space limitation. [5, 36,
58].
The overwhelming success of total hip arthroplasty today is a direct consequence of
major development efforts that involved a vast number of joint simulator wear tests to evaluate
the wear performance of various types of polyethylene and other potential materials. These wear
simulators in turn relied upon extensive studies of the kinematics and joint reaction loads
associated with hip joints. In stark contrast, only a handful of joint simulator wear tests have
been published to evaluate total ankle replacements. The biomechanics of the ankle joint are
arguably more complex than the hip or even the knee joint, emphasizing an even greater need for
in-depth analysis of ankle kinematics and joint reaction loads. By obtaining a solid understanding
of ankle kinematics and joint reaction loads, an established protocol can be created to
consistently and reliably conduct preclinical evaluation studies.
Currently there is no established preclinical ISO or ASTM testing standard for total ankle
replacements. A ISO standard draft for wear testing is under development (ISO 22622);
however, this outline is based on two previous ankle wear test which used varying kinematic
load and displacement parameters [123, 141]. Additionally, the sources of these testing
parameters were convoluted and gathered from multiple mathematical models or a collection of
incomplete data sets. A dedicated ASTM ankle wear testing standard does not exist, but the need
of wear testing is mentioned briefly in ASTM F2665 without providing specific test parameters.
The only current test standards for ankles pertains to ankle prostheses, ISO 22675:2016 and

37
ASTM F2665. The in vivo flexion measurements that are recommended in ASTM F2665 are 0°,
±10°, ±15°; however, no ankle literature is cited. While the ASTM standard does reference
studies that were used to find contact pressures at various flexion angles, these studies pertain to
the knee and wrist joints. Further, the loads that are suggested by ISO 22675:2016 for cadaveric
testing are broad and lack the clarity available for other joints. Recommended loads range from
722 N up to 5,345 N depending on the experimental setup. In addition, the suggested loads far
exceed cadaveric limitations. It is widely accepted that cadaveric specimens lack the opposing
muscle and tendon forces to allow it to sustain in vivo loads for in vitro testing. Without
establishing ISO or ASTM standardized testing protocols derived from well designed ankle
kinematics and load studies, future designs and modifications of total ankle replacements cannot
be properly evaluated. This in turn will continue to hinder the success of total ankle replacements
implanted in vivo.
1.8 Rationale for the Proposed Studies

As the ankle is encapsulated between the tibia, fibula, and calcaneus, in vivo assessment
of the talus is challenging. Some studies have used video fluoroscopy or motion tracking through
bone screws; however, both methods are invasive and may place the subject at risk of infection
or worse. A safer method to study the ankle is through cadaver models. However, there is no
consensus on how to properly test these specimens. Of the existing literature, many have either
constrained the experimental test setup or over simplified the test setup. These methods can
introduce additional forces or change the way the forces are applied to the specimen, in turn
skewing the results of the study. An accurate cadaver model is needed to provide accurate in
vivo kinematics of the ankle to allow noninvasive study of the ankle complex

38
Currently, a reliable cadaveric method does not exist, so the in vivo kinematics of TARs
are unknown. With the establishment of a cadaveric testing model, researchers can better analyze
how total ankle replacements are functioning within the body. This can help identify why
implants are failing within the body even though they have shown to improve gait and the
functional well-being of patients. An alternative method to analyze how implants are working in
vivo, is the study of retrievals. Retrievals can give insight into how the implants were articulating
within the body. This can be vital in validating not only the design improvements, but also the
test methods used to create those advancements.
This cadaveric model will allow researchers to evaluate their design improvements of total ankle
replacements more quickly and accurately before implanting them in patients. In addition to
evaluating how the TAR will work within the body, the biomechanical effects of common
symptoms seen in patients with TAR can be explored. For example, patients with TAR often
have fusions performed in the neighboring joints. As TARs are not originally designed to
function with fusions, the effects of a subtalar fusion on its performance can be explored.
Through an established cadaver testing model, in vivo kinematics of an ankle joint can be
reliably obtained. However, a point of contention remains on how these parameters should be
implemented for a TAR wear test. Historically, little difference between force and displacement
control wear test was seen for total hip replacements. However, due to the larger ranges of
motion and high reliance on ligament stability in the knee joint, wear tests for total knee
replacements are more complicated. It has been suggested that due to this difference, load control
should be used for total knee replacements as this will provide more accurate predictions of TKR
wear. For ankle joint replacements, both force and displacement control methods have been
proposed in the ISO 22622 wear test. TAR rely on both surrounding ligaments and its

39
articulating surfaces for stability; therefore, load control tests may be desirable in that regard. On
the other hand, displacement control wear tests are easier and more reproducible. However, it is
unknown whether the displacements recommended by the ISO standard are comparable to the
proposed force control ISO standard.
1.9 Purpose and Aim of Proposed Studies
The overarching purpose of these projects is to provide and contribute to the
establishment of the physiologically relevant testing methodology designed to predict in vivo
performance of total ankle replacements and in turn improving survivorship. The following
Specific Aims are designed to accomplish this objective.
1. Establish a model using cadaveric specimens to determine the 6 DOF motions of the
natural tibia, talus, and calcaneus under simulating physiologic gait.

2. Use the cadaveric model to evaluate the 6 DOF motions of total ankle replacement
components.

3. Use explanted total ankle replacement polyethylene components to assess the modes and
locations of damage during in vivo use.
4. Use the cadaveric model to establish 6 DOF motions of the natural talus and the artificial
talus component as a function of fusion of the subtalar joint.
5. Compare the 6 DOF motions of the natural ankle joint resulting from the proposed ISO
22622 force control testing standard to the proposed ISO 22622 displacement control testing
standard.

To achieve these objectives, the following studies were performed. For Specific Aim 1, a gait
analysis experiment was performed using motion tracker cameras synced with force plate data.
This data was analyzed at four distinct instances: heel strike, maximum weight acceptance, mid-
stance, and push off. This data was then used to correlate shank angle, ankle angle, ground
reaction force angle, and ground reaction force magnitude observed during gait. The results from
the gait experiment, were then applied to three cadaveric pairs where bone motion trackers were

40
implanted, allowing for 6 DOF kinematics to be recorded. One side of each pair was tested in its
natural state while the contralateral side was tested with an implant. Through these results,
Specific Aim 2 was achieved by comparing the kinematics of a TAR to those of a natural ankle.
For Specific Aim 3, fourteen explants or retrievals were analyzed. The modes and
location of damage illustrated how the explant was moving in vivo. By comparing these damage
maps with the results in Specific Aim 2, the testing method developed in Specific Aim 1 can be
supported.
Following this, an additional fusion was performed on both the natural and TAR
specimens. Using the model developed in Specific Aim 1, the same loading methodologies were
repeated. Specific Aim 4 was then achieved by analyzing the differences in displacement of the
specimens in the unfused and fused conditions.
For Specific Aim 5, an additional set of 12 cadavers were tested under the proposed ISO
22622 force profile. From these experiments, the resulting displacements were recorded and
compared to the proposed displacement profiles.
Collectively, these studies contributed to understanding the complete kinematics of the
ankle, leading to a more comprehensive preclinical testing protocol for the evaluation of total
ankle replacements, resulting in improved long-term survivorship of total ankle replacements.

41
2. Materials and Methods

2.1 Overview
To begin the project, an in vivo gait analysis was conducted. The purpose of this
experiment was to determine the shank angle, ankle angle, ground reaction force angle, and
ground reaction force magnitude during gait. While the entire stance phase of gait was recorded,
four points of interest on the ground reaction force curve were identified to represent the stance
phase of gait. The results from this gait analysis were then implemented as input parameters in
an in vitro cadaveric experiment, to establish an accurate cadaveric model to determine the 6
DOF motion of the tibia, talus, and calcaneus. With the establishment of this testing model, the
kinematics of an intact ankle joint and the kinematics of an artificial total ankle replacement
implanted in the cadaveric lower limb were compared. Following completion of this stage, the
effect of a subtalar fusion on both an intact ankle joint and a total ankle joint replacement were
compared.
The subsequent portion describes the method of analysis of 14 retrievals, of the same
implant design used in the cadaveric model, explanted after varying years of in vivo service.
Specifically, these retrievals were visually analyzed and scored for different types of damage in
order to assess their behavior in vivo. The damage modes and locations from these implants were
compared to the kinematics observed from the cadaveric model presented earlier.
The final stage of this project investigated the differences between the force-controlled
profile and displacement-controlled profile in the ISO standard proposed for total ankle joint
replacement wear (ISO 22622). To observe the differences in displacement, the force-controlled
profile was applied to a set of 12 intact cadaveric lower limb specimens. A custom designed

42
apparatus applied the specified forces and displacements by manipulating or controlling motion
of the intact talus and the tibia with all ligaments intact. The design of this apparatuses and
description of the forces are described in detail.
2.2 Gait Analysis
Joint replacement implant success is measured by its ability to reduce or eliminate pain,
restoration of range of motion, along with its survival rate within the patient. To achieve these
goals, it is vital for implants to accurately recreate the natural biomechanics of the joint. During
development, it is critical for implants to be subjected to preclinical testing to actively check that
these standards are achieved. Preclinical tests should implement physiologically accurate
biomechanical loads and kinematics to properly replicate in vivo settings. Due to the anatomical
location of the talus, direct tracking of the talus within an intact limb is cumbersome and difficult
with typical skin mounted motion tracking. As a result, the number of published studies
describing all 6DOF of the talus are sparse. Regardless, an extensive literature search was
performed to simultaneously identify in vivo shank angle, ankle angle, ground reaction force
angle, and ground reaction force magnitude during gait. Expectedly, this search yielded no peer
reviewed articles discussing this correlation. Although some historical gait studies have been
cited as references for input profiles, none comprehensively reported all four factors of interest
together [32, 118, 144]. Therefore, in order to acquire a comprehensive dataset, an in vivo gait
analysis experiment was conducted.
2.2.1 Experiment Design
Eight healthy subjects (3 females, 5 males) volunteered to participate in accordance with
the Institutional Review Board. The mean weight of all eight subjects was 783 N. All subjects
were healthy and had no previous major injuries of the ankle. The subjects performed a slow gait

43
while shod and unshod. Following these tasks, the subjects then performed a self-selected fast
gait while shod and unshod. All four tasks were performed five times for a total of 20 total trials
per subject. Each subject practiced each movement to ensure natural gait across the force plates.
Any trial the subject was deemed to lengthen or shorten their stride length, in order to strike the
force plates, was excluded and performed again. Additionally, any trial where the foot landed
between both force plates was excluded.
Kinematics and reaction forces were collected simultaneously during all tasks. Two-
dimensional sagittal plane kinematics were captured using a high-speed video camera at 60 Hz
(Panasonic, Secaucus, NY). The camera was placed on a level tripod approximately two feet
above the floor, perpendicular to the subject’s walking path. Reaction forces in three dimensions
were quantified using a single force plate. (0.4 x 0.6 m
2
, 1200 Hz, Kistler, Amherts, NY, USA).
MATLAB (R2017a; Natick, MA,USA) was used to sync the collected force and motion
capture data. The instant the subject’s heel made contact with the force plate (heel strike) was
used as a reference point for syncing the data (t = 0 s). Following successful syncing, force
vectors were overlaid on the motion capture data to allow for visual representation of the
direction and magnitude of the force vector. From here, Kinovea (https://www.kinovea.org) an
open-access video analysis software, was used to measure the shank angle, ground reaction force
angle, and ankle angle for four distinct instances during the stance phase: a) Heel strike (HS), b)
Maximum Weight Acceptance (MWA), c) Mid-Stance (MS), d) Push-off (PO). HS was chosen,
as it is the initial instance in time the body makes contact with the ground. MWA along with PO
were chosen, as these correlate with two of the maximum peak forces seen during the stance
phase. Finally, MS was analyzed since this is the moment the subject transitions from producing
a braking force to producing a propulsion force by translating the center of mass over the center

44
of pressure. Together, all four phases were measured to create a quasi-static representation of
gait.

Figure 2. From left to right (i) Heel strike, (ii) Maximum Weight Acceptance, (iii) Mid-Stance, and (iv) Push-Off with a ground
reaction force vector overlay.
Figure 2 shows a sample of the complete measurements that were made at each instance
(HS, MWA, MS, and PO). The shank angle was measured with respect to vertical, with the
vertex placed at a bony landmark (medial or lateral malleoli) of the measured leg to the medial or
lateral condyle of the knee (red). The ankle angle was measured as a relative angle between the
shank and the foot (green). Finally, the ground reaction force angle was measured along the
overlaid ground reaction vector (blue) with respect to vertical. Angles lying to the left of vertical
were denoted as negative and any angles to the right of vertical was denoted as positive. A
neutral ankle in an unflexed position (90° between the foot and shank) was denoted as 0°. Ankle
angles were recorded as positive for dorsiflexion (angle less than 0°) and negative for
plantarflexion (angle greater than 0°). The corresponding force data was first imported and
normalized by each subject’s individual body weight then filtered in MATLAB using a fifth
order Butterworth filter. A sample recorded force time curve showing each measured instance is
shown below (Figure 3).

45

Figure 3. Example measured force time curve with points of interest identified.
2.3 Biomechanical Simulation Using Cadaveric Specimens
The next experiment involved cadaveric simulation using six lower limb cadaveric
extremities, four instances during the stance phase of gait were simulated: Heel strike (HS),
Maximum Weight Acceptance (MWA), Mid-stance (MS), and Push off (PO). To accurately
replicate loading during each condition, the testing parameters were taken from the previously
performed gait analysis experiment. After measuring the kinematics and forces of each trial, an
average was taken to represent the loading parameters. By loading a cadaveric specimen to in
vivo parameters, accurate in vivo kinematics of the tibia, talus, and calcaneus could be replicated
and recorded, through in vitro biomechanical testing.

46
2.3.1 Specimen Preparation
2.3.1.1 Radiographs
Plain radiographs were taken to document the initial conditions of all specimens. Medial-
lateral and anterior-posterior radiographs of the specimen in a neutral position were also
documented. In addition, a maximum dorsiflexion and a maximum plantarflexion radiograph
was captured. Heavy lead cylinders were used to ensure the ankle stayed in maximum
dorsiflexion and maximum plantarflexion during each radiograph. For the specimens that
received total ankle replacements, radiographs were taken after the surgery to view the initial
implantation conditions. Additional radiographs were taken to ensure that flag holders were
placed securely in the correct bones of interest. Following testing, all specimens were again
radiographed from the sagittal plane and frontal plane.

Figure 4. Left radiograph shows maximum dorsiflexion. Right radiograph shows maximum plantarflexion.
Each specimen was also scanned using a Hologic 2000 bone densitometer (Hologic, Inc.,
Waltham, MA) to determine the bone mineral density (BMD) in g/cm
2
of the region of interest
(cuneiform, talus, fibula, and tibia).

47
2.3.1.2 Surgical Procedures
Three matched pairs of fresh-frozen cadaveric lower legs were obtained for this portion
of the studies. Each specimen was thawed to room temperature (~20°C) 24 hours in advance
before preparation. Medial-lateral and anterior-posterior radiographs were taken of each
individual cadaver in the following three positions: neutral, maximum dorsiflexion and
maximum plantarflexion. Motion tracker flag holders were then placed in the calcaneus, talus,
and tibia. The positioning of the holders was confirmed again with both medial-lateral and
anterior-posterior radiographs. After rigidly fixing the holders, small ¼-inch K-wire pins were
percutaneously positioned with the cadaver in neutral flexion, to serve as place holders for
motion tracker digitization. The placement of these pins was once again verified with
radiographs.

Figure 5. AP and ML radiographs confirming correct placement of flag holders and digitizing K-wire pins.
Of each of the three matched pairs, one side was chosen for implantation of an artificial
ankle joint replacement while the other was left intact. A thoroughly experienced orthopaedic
surgeon familiar with this clinical procedure implanted the total ankle replacements, using

48
instruments designed by the manufacturer for clinical application. The ankle replacement that
was used was the Infinity total ankle replacement (Wright Medical, Warsaw IN). Prior to
implantation, each specimen receiving the Infinity was radiographed in an anterior-posterior and
a medial-lateral view, in order to select an approximate size for the talar and tibial components.
However, the final sizing was verified during surgery to ensure proper seating of each
component. To begin the surgery, an anterior incision was made to expose the distal tibia and
talus. Using the medial gutter of the tibia as a reference point, a distal and a proximal 3.2 mm
Steinmann pin were installed in order to attach an alignment frame to the proximal face of the
tibia. The alignment frame was vital to creating proper targeting for bone resection in the sagittal
plane (flexion/extension) and coronal plane (varus/valgus). Once alignment was confirmed
through fluoroscopy, the alignment frame was removed and replaced with a resection guide. An
oscillating bone saw was then used with the resection guide to create the tibial and talar bone
resections. The guides were then removed, and any excess bone was further cleaned out with
osteotomes or chisels. Once completed, a trial size of the previously chosen tibial tray was
inserted to check for proper anterior/posterior and medial/lateral fitting. Similarly, a talar dome
trial was inserted to check for proper sizing. After ensuring proper fitting, the actual tibial and
talar components were implanted. Then, depending on the spacing between the two metal
components, a polyethylene thickness was chosen. Finally, the properly sized polyethylene insert
was placed and locked into the tibial tray using a plunger tool. Each total ankle replacement was
implanted by the same experienced orthopaedic surgeon.
Following completion of all tests, described in section 2.3.2, subtalar fusions were
performed on all specimens. Three to four-inch cannulated lag screws were used depending on
the specimen’s anatomical dimensions. To prevent rotation about the fusion screw’s axis, two

49
screws were implanted through the calcaneus into the talus. One screw was targeted through the
calcaneus and into the dome of the talus, while the second screw was guided to traverse through
the calcaneus and into the head or most anterior end of the talus. Kirschner wires were initially
placed to serve as a guide for the drill bit and fusion screw. Radiographs were used to confirm
correct placement of the Kirschner wires. From here, a driver with a hollow drill bit was threaded
over the Kirschner wire to create a pilot hole that followed the same trajectory, confirmed
through fluoroscopy. Finally, the lag screws were fed over the Kirschner guide wires and placed
through the calcaneus and into the talus. Fluoroscopy was once again used to ensure proper
placement of these fusion screws.

Figure 6. Radiograph showing fusion of the natural ankle (left) and total ankle replacement (right).
2.3.1.3 Flag placement
Motion of the calcaneus, talus, and tibia was recorded throughout testing using an
Optotrack Certus motion tracking system (Northern Digital, Inc., Waterloo, ON, Canada). This
system implemented rigid bodies (flags) which were attached to specific bones of interest via
bone screws. The four bones identified to contribute to ankle range of motion are the: calcaneus,

50
talus, tibia, and fibula. The tibiotalar joint or the interface between the talus, tibia, and fibula is
largely responsible for dorsiflexion and plantarflexion (flexion/extension in the sagittal plane).
The subtalar joint or the interface between the calcaneus and the talus is generally responsible for
inversion eversion or rotation medially/laterally within the frontal plane [21, 27, 86]. These bone
screws were comprised of a threaded shaft and a tapped cylinder that attaches to the flags.
Placement of the bone screws were verified using an anterior-posterior radiograph and a medial-
lateral radiograph.
After rigidly fixing the holders, small ¼ inch K-wire pins were placed through the skin
and into the bone, with the foot naturally positioned and unflexed, to serve as place holders for
motion tracker digitization. A total of 12 pins were placed, one on the medial side, one on the
lateral side, and two on the posterior side for the calcaneus, talus, and tibia/fibula construct. The
tibia and fibula were assumed to act dependently and as a single unit (shank). The lateral and
medial pins were placed on the same leveled transverse plane. One of the posterior pins was
placed in this same plane and the final pin was placed a centimeter directly above the first
posterior pin, in the same sagittal plane. The placement of these pins was once again verified
with radiographs.
2.3.2 Experimental Design

2.3.2.1 Apparatus Design
To begin the experiment, all specimens were thawed to room temperature and prepared
following the specimen preparation previously outlined. The superior end of all specimens was
rigidly fixed to the load frame (MTS Eden Prairie, MN) using a metal five-inch ring. Within the
ring, the tibia was fixed with 12 Steinman pins. A laser was illuminated along the tibial shaft to
ensure vertical alignment along the axis of the actuator. Once proper alignment was achieved, the

51
specimen was angled at the appropriate parameters which correlated with the instance in gait
being applied. To position the shank in the correct angle while allowing the force application to
remain concentric to the load cell, the ring holding the specimen was attached to a linear bearing.
Attached to the carriage of the linear bearing was the machined apparatus which could be locked
in the appropriate angle.
To appropriately test the ankle, no constraints were placed through the ankle joints, but
instead medial, lateral, anterior (for HS and MWA), and posterior (for MS and PO) cables were
used to help simulate the surrounding muscles that typically stabilize the ankle in vivo. Steel
cables were used in series with turnbuckles for each simulated muscle. The turnbuckles allowed
for lengthening or tightening to allow for the adjustment of the ankle angle to achieve the correct
angle during testing. For the anterior/posterior cable, an additional uniaxial load cell (Interface
Force, Scottsdale AZ, 250 lb/1000 N) was added for measurement of the assisting load. The
cables were attached to a custom plate made of acrylic and metal. This plate contained four
blocks that could travel and be locked medially and laterally to lock the foot to the plate during
testing. Additional screws were screwed through the posterior blocks into the calcaneus to
prevent any unwarranted heel rise and ensure a rigid union of the plate and heel interface.
Furthermore, the plate was lined with metal on the edges to prevent any bending of the plate that
may have occurred during loading.

52

Figure 7. Anterior view of heel strike position with stabilizing medial, lateral, and anterior cables.

Inferior to the plate was an attached hemisphere which articulated in a cup that was
attached to a load cell (Futek MTA500) which measured axial force and bending moments in
two perpendicular planes. These two planes were placed in line with the sagittal plane and the
frontal plane. This load cell was affixed superiorly to two perpendicular linear bearing tables.
These tables allowed for translation or displacement in the transverse plane.
Lateral Cable Medial Cable
Anterior Cable

53

Figure 8. Lateral view of the complete test setup for the Unconstrained Cyclic Loading.
The rationale behind the use of a hemisphere and cup, replicating a ball and socket joint
was to ensure that the load transmitted through this joint would be vertical without any resistance
to horizontal forces since the lower part of the apparatus was free to translate in the transverse
plane. That is, if an imbalance of horizontal forces was present, the system would not be in
equilibrium, thus creating a horizontal sliding motion. By eliminating any horizontal forces, the
apparatus was able to ensure vertical ground reaction force in the cadaver through a known point
of application that is the center of the ball joint. This loading mechanism was applied only in the
“Unconstrained Cyclic Loading,” described below.

Linear Bearing
Angler
Ring
Steel
Cables
Custom Plate
Uniaxial Loadcell
Ball & Cup
Loadcell
Perpendicular
Linear Bearings

54
For this experiment, three varying loading scenarios were implemented:
(i) Direct Anterior-Posterior Loading – A linear actuator was attached superiorly the
rail in the sagittal plane and translated in displacement control.
(ii) Constrained Cyclic Loading – Two linear actuators were attached to the rail in the
sagittal plane and the transverse plane. These actuators were affixed and did not
allow any movement in their respective planes.
(iii) Unconstrained Cyclic Loading – No actuators were attached, and the load cell was
free to slide in any direction along the transverse plane.
From previous literature along with our gait experiment, the ground reaction force was
observed to have a maximum magnitude of about 1.2 x BW. However, it is commonly accepted
that cadavers cannot sustain the same magnitudes of load experienced in vivo; therefore, a load
of 300 N was chosen, which was consistent with previous literature [9, 10, 94, 107, 130, 148,
152].
2.3.2.2 Loading Profiles

Although the previously discussed direct anterior-posterior loading and constrained
cyclic loading methods have drawbacks, it was included in the testing to facilitate comparison of
results to previous studies. At the beginning and end of each test, the specimen was positioned
vertically, then manually rotated through its full range of motion with and without an applied
heel load. From here the specimen was placed in the proper orientation for the position being
tested (HS, MWA, MS, PO). For each position, three stages of testing were implemented:
(I) Direct Anterior-Posterior Loading

The specimen was preloaded to an initial axial load of 150 N, with both the medial-lateral
and anterior-posterior locking actuators attached. After the initial preload was achieved, the

55
anterior-posterior linear actuator was used to translate the platform posteriorly, which was rigidly
attached to the foot. The anterior-posterior actuator was controlled in displacement control,
moving at 0.5 mm increments, until a full 3 mm was achieved. Following the 3 mm of posterior
translation, the specimen was held statically in this position for 3 seconds, to allow the motion
capture system to sufficiently record any movement of the talus. From this stage, the anterior-
posterior actuator was moved back to neutral, again at 0.5 mm increments. Upon returning to the
neutral position, the foot was allowed to rest for 3 seconds to re-establish a zero position, and
account for any changes that may have occurred in the talus. The anterior-posterior actuator was
then used to translate the foot anteriorly 3 mm in 0.5 mm increments. Again, this position was
held for 3 seconds to ensure proper optical motion readings. Finally, the platform was returned to
0 mm of displacement, placing the foot back in the initial preloaded position.
(II) Constrained Cyclic Loading

Following the Direct Anterior-Posterior Loading test, the medial-lateral and anterior-
posterior actuators were used to lock the platform rigidly in place preventing any movement. The
location of the platform relative to the foot was determined by the center of pressure specific to
the instance of gait being tested. For heel strike and maximum weight acceptance, the platform
was locked under the heel, for mid-stance the platform was locked towards the mid-foot, and for
push off the platform was locked inferior to the ball of the foot. Once the foot was in the correct
position, an initial preload of 150 N was again applied to the specimen. Throughout the process
of reaching this 150 N preload, the ankle angle was continuously monitored and adjusted with
opposing forces, provided by the metal wires and turnbuckles. These forces were provided by
medial, lateral, anterior (for HS and MWA), and posterior (for MS and PO) cables. Once the
preload was achieved, a cyclic axial load was applied at 0.2 Hz until 300 N of compression was

56
reached. To ensure the correct amount of force was applied as a function of time, overcoming the
viscoelastic responses of the bone and soft tissues, the axial load was applied in force control.
Once the 300 N of force was achieved, 3 cycles of loading at that level were recorded before
unloading the specimen.
(III) Unconstrained Cyclic Loading

After the Constrained Cyclic Loading test was completed, the medial-lateral and anterior-
posterior tables were removed. With these linear actuators removed, the load cell was
unconstrained and free to slide along the two perpendicular linear bearings in the transverse
plane. By removing any constraints and implementing a ball and socket, the resulting reaction
force was required to be vertical if the specimen was to be balanced. By eliminating any
horizontal forces and ensuring the reaction force be vertical, the ability to control the angle the
ground reaction force acted upon the cadaver exists. From here, the same methods described in
the Constrained Cyclic Loading test were followed to reach the 150 N preload. Throughout the
process of reaching this 150 N preload, the ankle angle was continuously monitored and adjusted
with opposing forces, provided by metal wires and turnbuckles. These forces were again
provided by medial, lateral, anterior (for HS and MWA), and posterior (for MS and PO) cables.
Once the specimen was properly preloaded in the correct angles, a cyclic axial load was applied
at 0.2 Hz until 300 N of compression was reached. Once the 300 N of force was achieved, 3
cycles of loading at that level were recorded before unloading the specimen.

57

Figure 9 An AP and ML view of the specimen set up in Heel strike.
2.3.2.3 Testing Parameters

The testing parameters for the Constrained and Unconstrained Cyclic Loading were
derived from the gait experiment previously performed with eight subjects. From the data, the
following table summarizes the shank angle, ankle angle, and ground reaction force angle used
as a basis for each of the four simulated instances of the stance phase of gait (heel strike,
maximum weight acceptance, mid-stance, and push-off).
Table 1. Testing parameters that were used during testing for each position.
Testing Parameters
HS MWA MS PO
Shank Angle -20° 6° 8° 15°
Ankle Angle -5° -5° 2° 7°
GRF Angle 0° 0° 0° 0°

58
2.3.3 Data Acquisition

2.3.3.1 Motion Tracker

An Optotrack Certus motion tracking system (Northern Digital, Inc., Waterloo, ON,
Canada) was used to record the motion of the tibia, talus, and calcaneus. The motion capture
system has an accuracy of 0.1mm, and a resolution of 0.01mm. Motion flags, equipped with 3
non-collinear light-emitting diode (LED) markers, pulsed at 3000 Hz. A minimum of 3 LED
markers were required to obtain rotations (Rx, Ry, Rz) along with translations (x, y, z), while a
minimum of 1 marker was needed for pure translations. These motion flags were attached via
bone screws to the flag holders placed in the talus, tibia, and calcaneus during the specimen
preparation phase. An additional flag was mounted on the load frame to establish a fixed local
coordinate axis and to minimize the effects of any vibrations of the machine.
A Position Sensor was used in conjunction with the motion tracker flags to track the
infrared lights emitted by the active markers embedded in the motion tracker flags. The Position
Sensor is 1126mm in length, 200mm in width, and 161mm in height in its horizontal
configuration. It contains three cameras at specific angles which allows for a total of 20 m
3
measurement volume. This Position sensor can be mounted vertically or horizontally depending
on the application. For the present experiment, the position sensor was mounted vertically. The
default global coordinate system is located at the middle camera sensor, but can be transformed
to a local coordinate system with the use of a motion tracker flag.
2.3.3.2 Digitization

The four points that were previously marked with inserted ¼ inch Kirschner wire pins
during the specimen preparation phase (Section 2.3.2) were used during digitization of the

59
motion tracker flags. During digitization, imaginary points were created and tracked by the
Position Sensor. These points were used to transform the individual coordinate system origin of
the motion tracker from the flag’s face to the middle of the respective bone it was rigidly
attached to. The medial and lateral points were digitized to generate a medial-lateral vector that
was bisected to create the translated origin. Following this digitization, a posterior digitized point
helped create the anterior-posterior axis by creating a vector to the previously created origin.
Finally, a second posterior point was digitized a centimeter above the initial posterior point
which allowed the motion tracker to create the sagittal plane. Using these vectors and the right-
hand rule, the remainder of the coordinate axis was created in the middle of the bone of interest.
An additional flag was mounted to the load frame and was used as the local coordinate
system origin. For this transformation from the global coordinate system, originally located in
the middle sensor of the Position Sensor, five imaginary points are digitized. A left, right, front,
and back point of a square’s vertices were used as reference points. The middle of all four points
was used to create the origin, while the left and right points were used to create the medial-lateral
axis. The origin and back points were then used to create the second axis which was crossed into
the medial-lateral axis to create the third perpendicular axis. The final Top digitized point was
used to create the sagittal plane through the origin. The accuracy and precision of this machine
has been independently validated. This system has been used previously in our biomechanics
laboratory for spine biomechanics studies as well as other foot and ankle biomechanics studies.

60

Figure 10 Ankle model showing the coordinate system used from digitization
2.3.3.3 Strain Gauge

In order to study the load transfer of the implant to the bone, a rosette strain gauge was
placed adjacent to the distal tibia. The rosette strain gauges consist of three 45° strain gauges to
measure principal strains and maximum shear strains. Each strain gauge consists of coiled metal
wires which change in resistance as they are stretched or compressed. This fluctuation of
resistance in conjunction with a Wheatstone bridge, allows for small changes in strain to be
recorded. From these measurements, the maximum stresses experienced at the specific strain
gauge location can be calculated using the stress-strain relationship. This relationship can be
represented by multiplying the strain by the system’s Young’s modulus:
𝜎 =𝐸∗𝜀
In this equation 𝜎 represents stress, E represents the materials specific Young’s modulus, and 𝜀
represents the measured strain. A group of three strain gauges assembled into specific
orientations at 45° increments create strain gauge rosettes (0°, 45°, and 90°). These arrangements
are called rosette strain gauges and allows for the principal axes where maximum and minimum
stresses occur to be calculated using the following equations:

61
𝜎
!,#
=
𝐸
2
∗
'
𝜀
$
+𝜀
%
1−𝜈
±
1
1+𝜈
-
(𝜀
$
−𝜀
%
)
&
+(2𝜀
&
−𝜀
$
−𝜀
%
)
&
0

𝜃
!,#
=
1
2
tan
'$
5
2𝜀
&
−𝜀
$
−𝜀
%
𝜀
$
−𝜀
%
6

Again, 𝜎 represents stress, E represents the materials specific Young’s modulus, 𝜀
$
represents
the horizontal (0°) measured strain, 𝜀
&
represents the 45° measured strain, 𝜀
%
represents the
vertical (90°) measured strain, and finally 𝜈 represents Poisson’s ratio.

Figure 11. Rosette strain gauge with a strain gauge at 0°, 45°, and 90°.
Generally, strain gauges are made to have an unstretched standard resistance of 120W or
350 W. The selection of strain gauge type depends on the application, but both are suitable for
most stress-analysis experiments. However, for transducers and high precision machines, 350 W
gauges are often used due to the ability to reduce heat generation, decrease lead wire effects, and
improve signal-to-noise ratios.
A single 120 W rosette gauge was glued to the anterior surface of the tibia 1 cm above the
distal transverse bone resection cut, performed during the total ankle replacement surgery.
Similarly, another 120 W rosette was glued onto approximately the same location of the
contralateral side (intact specimen). The axial strain gauge (90° from horizontal) of the Rosette
was placed in line with the tibial shaft, while the transverse strain gauge (0° from horizontal) was
placed parallel to the transverse cut. For reproducibility, the transverse strain gauge was oriented

62
towards the medial side of the ankle. Following proper placement and soldering, the incision to
access the tibia was sutured closed.

Figure 12 The picture on the left shows an AP view of a strain gauge that was placed on the anterior side of the tibia
approximately 2 cm above the distal end of the tibia. The picture on the right shows the specimen after the incision was sutured
closed.
2.3.3.4 Rotation and Displacement Calculations
The talus rotation and translation were taken with respect to the tibia in all four instances
of the stance phase. In order to properly measure talus movement in the tibial reference frame, all
exported recorded data was expressed with rotation matrixes and displacement in the X, Y, and Z
axis. From these raw data points, the following transformations were made to calculate the talus
motions with respect to the tibial reference frame:
𝑅
()*)+
=
8
𝑅
,,
𝑅
,-
𝑅
,.
𝑅
-,
𝑅
--
𝑅
-.
𝑅
.,
𝑅
.-
𝑅
..
9

𝑢
/
=
<
𝑥
𝑦
𝑧
@
()*)+
−
<
𝑥
𝑦
𝑧
@
(+012

63
𝑢
(+012
= 𝑇𝑎𝑙𝑢𝑠 𝑚𝑜𝑡𝑖𝑜𝑛 𝑤𝑖𝑡ℎ 𝑟𝑒𝑠𝑝𝑒𝑐𝑡 𝑡𝑜 𝑇𝑖𝑏𝑖𝑎 𝑟𝑒𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑓𝑟𝑎𝑚𝑒
𝑢
(+012
= 𝑅
()*)+
(
∗𝑢
/

𝑢
(+012
=
8
𝑅
,,
𝑅
,-
𝑅
,.
𝑅
-,
𝑅
--
𝑅
-.
𝑅
.,
𝑅
.-
𝑅
..
9
(
∗
R

<
𝑥
𝑦
𝑧
@
()*)+
−
<
𝑥
𝑦
𝑧
@
(+012
S

Specifically, the anterior/posterior displacement and dorsiflexion/plantarflexion of the talus with
respect to the tibia were calculated. Additional calculations for the internal/external rotation of
the tibia with respect to the local coordinate system, and the inversion/eversion of the calcaneus
with respect to the talus were performed. These calculations were done for both the intact and
implanted specimens.
2.3.4 Data Analysis
2.3.4.1 Reduction

A custom written MATLAB program was used to calculate the anterior/posterior
displacement and dorsiflexion/plantarflexion of the talus with respect to the tibia as well as
inversion/eversion. The program imported the optical motion tracker data and parsed out the
relevant data columns. The data was then checked to remove any drop outs of signal or unneeded
data. A fifth order Butterworth filter with a sampling frequency of 100 Hz and a cutoff frequency
of 0.2 Hz was applied to the data to reduce noise. From here the AP displacement that occurred
from the 150 N of preload to 300 N of compression was calculated.
2.3.4.2 Statistics

Matched pair experimental design was used in this experiment. A total of three pairs of
lower legs were used (N = 6). Rotational motion between the tibia and talus along with anterior-

64
posterior displacement between the two bones were calculated. The inversion/ eversion between
the talus and calcaneus was also calculated. In addition, the 3-Dimensional motion between all
points of interest were calculated from 150 N of preload to 300 N of total cyclic loading. These
motion calculations were repeated for the same specimens under intact, implanted total ankle
replacements, and fused conditions. Differences were calculated for each motion between
contralateral sides of each pair as follows: (Intact vs TAR), that is motion of each intact leg with
was compared to the motion of the matched contralateral side with the total ankle replacement
implanted. In addition, differences were calculated for each motion as follows: (Intact – Fused)
and (TAR – Fused). Paired t tests were used to compare differences in means of each motion for
all of the pairs listed above. All data analysis was performed using SPSS Statistics Version 17.0
(IBM, Houston Texas).
2.4 Retrieval Damage Analysis

In this experiment, three researchers independently examined 14 retrieved TAR
polyethylene inserts for burnishing, scratching, mechanical damage, pitting, and embedded
particles. Specifically, the aim was to correlate clinical damage and observations with in vivo
kinematics of the components.
2.4.1 Damage Features

Two IRB-approved retrieval databases were queried for total ankle replacement (TAR)
devices for which patient information was known and the polymer bearing was available for
analysis. From a collection of 70 TAR retrievals available to our collaborators, all polyethylene
inserts from implants with fixed bearing designs that were sterilized with ethylene oxide (EtO)
were selected. Fourteen retrieved polyethylene inserts designed by Wright Medical Technologies

65
(Memphis, TN) were included; twelve were InBone® prosthesis and two were Infinity®
prosthesis inserts. Each articulating surface was analyzed and scored on a scale of 0-4, using a
Keyence Microscope (VHX-2000, Chicago IL).

Figure 13. Example map used to rate damage for each retrieval.
Within each quadrant, five features of damage were rated: burnishing, scratching,
mechanical damage, pitting, and embedded particles, defined as follows:

(i) Burnishing: Damage that has “polished” the surface and is identified with a shiny hue
and absence of original machining marks.
(ii) Scratching: Damage that is characterized with a rougher matte surface and scratches.
(iii) Mechanical Damage: Markings with deep gouges and grooves or deformation. This
excluded damage caused during implantation and removal.
(iv) Pitting: Small divots or circular indentations, most likely attributed to 3
rd
body
particle.
1 2
3
4

66
(v) Embedded Particles: 3
rd
body particles that are embedded within the articulating
surface.
2.4.2 Semi-Quantitative Grading Scale

A semi-quantitative grading scale was implemented and ranged from zero to four, with
zero indicating no apparent damage and four indicating a high degree of damage. This method
was derived from the Hood scoring protocol [65] and has been well established for modern hip
and knee joint replacement components [37, 109].

0: Area is free of respective damage
1: 0 < 25% of area is covered by respective damage
2: 25% < 50% of area is covered by respective damage
3: 50% < 75% of area is covered by respective damage
4: 75% ≤ 100% of area is covered by respective damage
2.4.3 Data Analysis

Three investigators independently scored each polyethylene insert using the method
outlined above. Interobserver reliability was assessed by comparing the average differences in
score along with the minimum, maximum, and mode of the differences.
All data analysis was performed using SPSS Statistics Version 17.0 (IBM, Houston
Texas). A Wilcoxon signed-rank test (nonparametric, paired) was run to compare the anterior,
quadrants 1 and 2, against posterior, quadrants 3 and 4, for average burnishing and scratching
damage. An additional Wilcoxon signed-rank test was run to compare the medial, quadrants 1
and 3, and lateral, quadrants 2 and 4, aspects for average burnishing and scratching damage.

67
Finally, to assess the correlation between duration in vivo and damage, Spearman correlation
coefficient (nonparametric) was calculated.
2.5 Force versus Displacement Profiles

Joint replacement wear tests provide a method for researchers to reliably predict the
longevity of a joint replacement implanted in a patient, by placing the implant under simulated in
vivo loads and kinematics for millions of cycles and assessing wear of the component and any
possible damage. However, for accurate predictions, the correct parameters must be applied to
the replacement. ASTM and ISO standards provide a standardized set of input parameters which
allows different investigators to implement and provide data that can be compared and assessed.
Currently, no ASTM standard exists for TARs. An ISO standard (ISO 22622) draft for TARs has
been under development and was recently finalized. Similar to total hip replacements (THR) and
total knee replacements (TKR), two protocols are allowed: a displacement control or force
control protocol has been proposed. Historically, displacement-controlled testing has provided a
highly reliable method in testing both THR and TKR and has been validated using retrieval
studies as well as clinical follow up. However, recently force-controlled wear tests for TKR have
been advocated on the basis that knees are highly reliant on ligaments for stability. Therefore, the
nature and magnitudes of displacements during a force-controlled test will be substantially
different. As well, the interactions between the bearing surfaces will be different, especially at
ends of the range of motion. These factors are hypothesized to substantially affect resultant wear.
For example, Sutton et al. has showed an increase in displacement and rotations in a knee
cadaveric model under a force-controlled test; however, a change in the amount or nature of wear
from a force-controlled TKR wear test is yet to be proven.

68
As the ankle is reliant on both ligament and articulating geometry for stability, depending
on the magnitude of loads of gait, results from a force-controlled or displacement-controlled
wear test may differ. As indicated, ISO 22622 has proposed both a force and displacement-
controlled wear standard; however, no justification on the equivalence of both methods has been
provided. Moreover, no substantial research on the equivalence of the two testing methods has
been performed. Both methods specify the same force profile for axial load to be applied to the
tibial component, and angular rotation profile for dorsiflexion-plantarflexion to be applied to the
talar component. However, the displacement-controlled method specifies a displacement profile
for anterior-posterior displacement, and angular rotation profiles for internal-external rotation,
both to be applied to the tibial component. In contrast, the force-controlled method specifies a
force profile for anterior-posterior loads and a torque profile for internal-external torque, both to
be applied to the tibial component. The force-controlled method requires application of a
restraint system, since the implant components alone have no inherent resistance to loads or
torques.
As both ISO profiles are designed to test implant components directly, when the load
actuator is programmed to apply the ISO specified forces to a cadaveric talus and tibia, with
ligaments intact, the measured force profiles may differ due to the soft tissue laxity and inherent
variability of cadavers. Therefore, to observe the difference between the command profile and
the measured output profile, loads were recorded using a 6DOF load cell and plotted in a single
figure. The command profile programmed into the servohydraulic load frame for all specimens
was plotted in blue. The measured loads (that were applied to each specimen) were plotted in
gray. The average of the measured forces from all 12 specimens were plotted in red. From the
applied load profiles, the resulting AP displacement and internal-external rotation of the talus

69
were recorded using a motion tracking system. These measured displacements were then
compared to the displacements specified by the ISO 22622 displacement-controlled standard.
These were again illustrated on the same figure with each individual specimen’s displacements
plotted in gray, the average plotted in red, and the ISO 22622 displacement-controlled standard
in blue.
2.5.1 Specimen Preparation
2.5.1.1 Dissection

Six pairs of fresh-frozen cadaveric lower limbs cut midshaft were obtained from a
licensed willed body tissue company, Science Care. The specimens had all extraneous fat and
tissues removed, while keeping all ligaments around the ankle complex intact. Due to the
importance that the interosseus ligament between the tibia and fibula plays in overall ankle
syndesmosis stability, care was taken to not disturb this ligament [149]. Since the talus was
directly manipulated, the Achilles tendon was removed as it would not be needed to help achieve
and maintain the desired angles. Additionally, all bones except for the tibia, fibula, talus,
calcaneus, and navicular were removed from the cadaver. As more distal bones did not directly
affect the ankle articulations they were removed. Once the extraneous tissues and bones were
removed, a single two-inch wood screw was inserted at the proximal end through the fibula into
the tibia to ensure the interosseus ligament was not disturbed. Moreover, this allowed better
integration in the bonding cement used during potting. From here two 3-inch fully threaded
screws were placed through the entire talus from the posterior surface, past the neck, and into the
head, without piercing through the head of the talus. These screws extended about an inch from
the most posterior edge of the calcaneus. To ensure proper screw placement, radiographs were
taken to guide the angle of approach along with the distance the screw was inserted.

70

Figure 14. Left: Dissected specimen with screws inserted from the posterior aspect. Right: Radiograph confirming screw
placement through talus.
2.5.1.2 Potting
Following completion of dissection and screw placement, the proximal end of the
cadaver was potted in bonding cement (Bondo 3M, Maplewood, MN). Upon successful
completion of this step, bone cement was poured in a square dam which surrounded the entire
cadaver and captured the heads of the implanted screws. This created a “halo” of bone cement
which provided direct access to manipulate the talus into dorsiflexion or plantarflexion during
testing. Throughout this process, the specimen was fixed in a neutral position, with the tibia
vertical and the talus neither plantarflexed nor dorsiflexed as described in ISO 22622. The
neutral position was determined using an alignment similar to the method described by Leardini
et al [86]. Radiographs were again taken to confirm the neutral alignment prior to the creation of
the halo.

71

Figure 15. Specimen with bonding cement on the proximal end and a bone cement halo encapsulating the posterior screws.

2.5.2 Experimental Design

2.5.2.1 Apparatus Design
Specimens were loaded using an 8DOF load frame (858 Bionix, MTS Systems, Eden
Prairie, MN). A superior axial actuator applied both the axial force load and the torsional torque
in force and torque control respectively (Figure 16A and 16B). Attached to the axial actuator a
linear actuator applied the anterior-posterior (AP) force (Figure 16C). Following the ISO 22622
standard, the AP force was applied to the tibial component; therefore, the linear actuator acted
upon the tibia and fibula construct. However, since this actuator operated in displacement
control, a spring (Model 72487, Century Spring Corp) with a rate of 23.292 N/mm was applied
in series with the actuator. This allowed for conversion of linear displacement into force using
the relationship:
𝐹 = 𝑘∗𝑥

72
With k being the spring rate and x being displacement. An additional 6DOF load cell (Mini 58,
ATI Industrial Automation, Apex, NC) was attached to monitor and record the applied loads.
Correspondingly, a restraining AP force was applied through two pairs of compression springs,
each with a rate of 16.987 N/mm, giving an effective 33.97 N/mm. Two springs were placed on
the anterior side while the other two were placed on the posterior side. All were in a relaxed state
at neutral when no AP force was applied; however, depending on if an anterior or posterior force
was applied, compression of the opposing spring alternated. While ISO 22622 suggests a 20
N/mm restraint spring directly at the tibial component, due to the restraint system in this
experiment not being placed directly at the joint replacement interface, a stiffer spring was
deemed appropriate.
To apply the final parameter of dorsiflexion and plantarflexion an additional linear
actuator was used (Figure 16D). This actuator was attached to a custom apparatus which
translated its linear displacements into rotational displacements at a rate of 1 mm per 0.8 degree
(Figure 16E). These linkages were crimped around a 5/8
th
inch diameter metal bar which was
securely fixed into a custom 3D printed plate made of tough PLA filament (Tough PLA 202300,
Ultimaker, Utrecht, Netherlands). On the superior surface of the plate, four hardened steel bars
were attached at each corner. These bars served as support and guide pillars for the tray that held
the bone cement halo through eight set screws. On each pillar, two round flange linear ball
bearings (Model 6483K52, McMaster-Carr, Santa Fe Springs, CA) were mounted on the inferior
and superior aspect of the tray, to ensure frictionless vertical gliding of the bone cement halo
during loading. To guarantee the applied rotation occurred at the natural axis of rotation of the
talus, this apparatus was mounted to a frictionless carriage which was allowed to slide on a linear

73
rail. Together, these components effectively rotated the entire 3D printed plate with the attached
bone halo which in turn directly manipulated the dorsiflexion and plantarflexion of the talus.

Figure 16. Load frame setup showing the applied loads: (A) Axial Load, (B) Torsional Load, (C) AP Load, (D) Linear
displacement translated to (E) Dorsiflexion-Plantarflexion
2.5.2.2 Loading Conditions

As indicated in Section 2.5, the ISO 22622 standard was finalized late 2019. The present
project was initiated prior to this. Therefore, the protocol was based on parameters outlined in
the draft proposal that was available at the time.
Two methods are suggested by the proposed ISO standard: a displacement-controlled
method or a force-controlled method. Both methods specify the same force profile for axial load
to be applied to the tibial component, and angular rotation profile for dorsiflexion-plantarflexion

74
to be applied to the talar component. However, the displacement-controlled method specifies a
displacement profile for anterior-posterior displacement, and angular rotation profiles for
internal-external rotation, both to be applied to the tibial component. In contrast, the force-
controlled method specifies a force profile for anterior-posterior loads and a torque profile for
internal-external torque, both to be applied to the tibial component. However, due to the physical
limitations of cadavers and prevention of soft tissue damage, only 50% of the axial force, AP
force and tibial torque magnitudes were applied [146]. These were all implemented at a
frequency of 1 Hz ± 0.1Hz. A summary of all of the applied forces derived from ISO 22622 is
shown below (Figure 17).

Figure 17. Comprehensive force profile that was applied to the cadavers.
(I) Axial Force Command

A peak axial load of 2,365.7 N was suggested in the ISO profile. Since this was reduced
by half, 1,183 N was applied along the axial axis of the tibia. The first peak at around 10% of the
gait cycle represented maximum weight acceptance when the body is slowing itself down by
-20
0
20
40
60
80
100
-200
0
200
400
600
800
1000
1200
1400
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
Angle (°) and Moment (Nm)
Force (N) - Axial and AP
% Gait Cycle
Applied Force Protocol
Axial Force (N) AP Force (N) Plantar/Dorsiflex-ion angle (°) I/E Moment (Nm)

75
applying a braking force. The valley at 15% correlated with midstance where the body is
transitioning from a braking to a propelling phase. It is at this instance in time where the center
of mass moves from posterior to anterior of the contact force. The greatest peak occurred 40% in
to the gait cycle at about 1,183N. This correlated with push off and when the body produced the
greatest reaction force to propel itself forward. From here the specimen entered the swing phase
at 60% of the gait cycle with a load of 100 N.

Figure 18. Axial load applied to specimens.
(II) Flexion Command

A range of 15° plantarflexion to 15° dorsiflexion was applied directly to the talus.
However, because a linear actuator was used to apply the rotations, the amount of displacement
the actuator was required to move was calculated based on the 1 mm to 0.84° ratio. To achieve
the desired rotations the actuator was programmed to reach a maximum of 17.8 mm. Expectedly,
the ankle was programmed to plantarflex in the initial phase of the gait cycle. Following this
initial plantarflexion, the ankle was rotated into dorsiflexion until it hit a maximum of 15° or
0
200
400
600
800
1000
1200
1400
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
Force (N)
Percent Gait Cycle
Axial Load

76
17.8 mm at 50% of the gait cycle. From here, the ankle entered the pre-swing phase and the talus
was plantarflexed to the maximum 15°. Finally, the swing phase was entered, and the talus
dorsiflexed as seen in gait analysis studies [6, 27].

Figure 19. Applied dorsiflexion and plantarflexion in displacement control.

(III) Anterior-Posterior Load Command

The AP force was applied perpendicularly to both the tibial axis and the
dorsiflexion/plantarflexion axis. Following ISO 22622, the AP force was applied to the tibial
component or for this experiment, the tibia itself. A maximum load of 129N and a minimum load
of -62N was applied to the tibia. Due to the constraints of the linear actuator used, a compression
spring with a constant of 23.292 N/mm was used to convert displacement to force. An additional
pair of opposing springs were added to create the AP motion restraint system. Each spring had a
spring constant of 16.987 N/mm creating an effective 33.974 N/mm spring constant. A diagram
from the medial-lateral view of the spring setup is illustrated below.
-20
-15
-10
-5
0
5
10
15
20
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
Actuator Displacement (mm)
Percent Gait Cycle
Applied Dorsiflexion/Plantarflexion

77

Figure 20. Schematic of AP displacement conversion and AP spring restraint system.
To calculate the AP displacement required to hit the required loads, the following
calculation were performed.
Knowns:
𝑘
$
= 𝐴𝑃 𝐶𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛 𝑆𝑝𝑟𝑖𝑛𝑔 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝑘
&
= 𝐴𝑃 𝑅𝑒𝑠𝑡𝑟𝑎𝑖𝑛𝑡 𝑆𝑝𝑟𝑖𝑛𝑔 𝐶𝑜𝑛𝑠𝑡𝑎𝑛𝑡
𝑥
$
= 𝐴𝑃 𝐶𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛 𝑆𝑝𝑟𝑖𝑛𝑔 𝐷𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡
𝑥
&
= 𝐴𝑃 𝑅𝑒𝑠𝑡𝑟𝑎𝑖𝑛𝑡 𝑆𝑝𝑟𝑖𝑛𝑔 𝐷𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡
𝑥
%
= 𝐴𝑃 𝐷𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡
𝐸𝑞 1: 𝑥
$
+ 𝑥
&
= 𝑥
%

𝐸𝑞 2: 𝐹 = 𝑘
$
𝑥
$

𝐸𝑞 3: 𝐹 = 𝑘
&
𝑥
&

Rearranging Eq 1.
𝑥
&
= 𝑥
%
− 𝑥
$

78
Equating Eq 2 and Eq 3.
𝐹 = 𝑘
$
𝑥
$
= 𝑘
&
𝑥
&

Substituting Rearranged Eq 1
𝑘
$
𝑥
$
= 𝑘
&
(𝑥
%
− 𝑥
$
)
𝑘
$
𝑥
$
= 𝑘
&
𝑥
%
− 𝑘
&
𝑥
$

(𝑘
$
+𝑘
&
)𝑥
$
= 𝑘
&
𝑥
%

𝑥
$
=
𝑘
&
(𝑘
$
+𝑘
&
)
𝑥
%

𝑥
&
=
𝑘
$
(𝑘
$
+𝑘
&
)
𝑥
%

Substituting this back into Eq 2 or Eq 3
𝐹 = 𝑘
$
𝑥
$

𝐹 = 𝑘
$
𝑘
&
(𝑘
$
+𝑘
&
)
𝑥
%

Therefore
𝐴𝑃 𝐷𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡 = 𝑥
%
=
𝐹
𝑘
$
𝑘
&
(𝑘
$
+𝑘
&
)

79
Using the relationship derived above, the linear actuator was programmed to travel a maximum
of 13mm and a minimum of -6.3mm to achieve 129 N and -62N. However, due to the natural
laxity of the cadaver and general slop of the apparatuses, the full desired AP load was not
reached. Consequently, this profile was increased by 50% to allow the AP force to reach the
desired loads. This created an updated max displacement of 19.62mm and -9.47mm. The full
displacement profile is shown below. Positive displacement correlates with a force pushing the
tibia anteriorly. Conversely, a negative displacement correlates with a force pushing the tibia
posteriorly.

Figure 21. Applied AP displacement to reach desired AP force.

-15.00
-10.00
-5.00
0.00
5.00
10.00
15.00
20.00
25.00
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
Displacement (mm)
Percent Gait Cycle
Applied AP Displacement

80
(IV) Internal/External Torque Command
Using the same actuator as the Axial Force actuator a tibial torque was applied. In order
to keep the magnitude of torque proportional to the other forces, half of the torque described in
the ISO standard was applied. Positive torque indicated internal rotation which quickly
transitioned to negative torque resulting in external rotation. Whether the specimen was a left or
right ankle was taken into account by applying the inverted torque profile.

Figure 22. Internal and External moments applied to the tibia.

-3.5
-3
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
5
9
13
17
21
25
29
33
37
41
45
49
53
57
61
65
69
73
77
81
85
89
93
97
Torque (Nm)
Percent Gait Cycle
Tibial Rotational Torque

81
(V) ISO 22622 AP Displacement Profile
The resulting AP displacement from the four force profiles presented above were
compared to the ISO 22622 AP displacement command profile. This profile was taken from
Smyth et al in their 2017 study, who derived this profile from Conti et al [32, 144]. The profile
from Conti et al. is the maximum displacement measured from a subject in their cohort of ten
patients with unilateral Agility total ankle replacements. Since Conti et al. only presented the
stance phase (60% of gait cycle), the remaining portion, from 60%-100% of the gait cycle was
extrapolated by Smyth et al.

Figure 23. ISO Displacement Command Profile

-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
0 20 40 60 80 100 AP Displacement (mm)
Percent of Gait Cycle
ISO AP Displacement Command Profile

82
(VI) ISO 22622 Internal/External Rotation Profile
The resulting angular rotation from the four force profiles presented above were
compared to the ISO 22622 rotation command profile. Similar to the AP displacement profile,
this profile was taken from Smyth et al in their 2017 study; however, the profile for
internal/external rotation was taken from three studies [13, 105, 140]. The rotations of the talus
were recorded and compared to this rotation command profile.

Figure 24. ISO Internal and External Rotation Command Profile
2.5.3 Data Acquisition

2.5.3.1 Motion Tracker

An Optotrack Certus motion tracking system (Northern Digital, Inc., Waterloo, ON,
Canada) was used to record the motion of the tibia, talus, and calcaneus. The motion capture
system has an accuracy of 0.1mm, and a resolution of 0.01mm. Motion flags, equipped with 3
non-collinear light-emitting diode (LED) markers, pulse at 3000 Hz. A minimum of 3 LED
markers are required to obtain rotations (Rx, Ry, Rz) along with translations (x, y, z), while a
minimum of 1 marker is needed for pure translations. These motion flags were attached via bone
-10
-8
-6
-4
-2
0
2
4
0 20 40 60 80 100 120
Rotation (Degrees)
Percent of Gait Cycle
ISO Internal/Extrenal Rotation Command Profile

83
screws to the flag holders placed in the talus, tibia, and calcaneus during the specimen
preparation phase. An additional flag was mounted on the load frame to establish a fixed local
coordinate axis and to minimize the effects of any vibrations of the hydraulic machine.
A Position Sensor was used in conjunction with the motion tracker flags to track the
infrared lights emitted by the active markers embedded in the motion tracker flags. The Position
Sensor is 1126mm in length, 200mm in width, and 161mm in height in its horizontal
configuration. It contains three cameras at specific angles which allows for a total of 20 m
3
measurement volume. This Position sensor can be mounted vertically or horizontally depending
on the application. The default global coordinate system is located at the middle camera sensor,
but can be transformed to a local coordinate system with the use of a motion tracker flag.
2.5.3.2 Digitization

Four points were again used for digitization. During digitization, imaginary points were
created and tracked by the Position Sensor. These points were used to transform the individual
coordinate system origin of the motion tracker from the face of the flag to the middle of the
respective bone it is rigidly attached to. The medial and lateral points were digitized to generate a
medial-lateral vector that was bisected to create the translated origin. Following this digitization,
a posterior digitized point helped create the anterior-posterior axis by creating a vector to the
previously created origin. Finally, a second posterior point was digitized a centimeter above the
initial posterior point which allowed the motion tracker to create the sagittal plane. Using these
vectors and the right-hand rule, the remainder of the coordinate axis was created in the middle of
the bone of interest.

84
An additional flag was mounted to the load frame and was used as the local coordinate
system origin. To transform the global coordinate system, which was originally located in the
middle sensor of the Position Sensor, five imaginary points were digitized. The left, right, front,
and back vertices of a square were used as reference points. The middle of all four points was
used to create the origin, while the left and right points were used to create the medial-lateral
axis. The origin and back points were then used to create the second axis which was crossed into
the medial-lateral axis to create the third perpendicular axis. The final Top digitized point was
used to create the sagittal plane through the origin. The accuracy and precision of this machine
has been independently validated. This system has been used previously in our biomechanics
laboratory for spine biomechanics studies as well as other foot and ankle biomechanics studies.
2.5.3.3 Load cell

A six degree of freedom load cell (Mini 58, ATI Industrial Automation, Apex, NC) was
used to measure (1) axial load, (2) anterior-posterior force, and (3) torsional torque. This load
cell has a capacity of 6800 N of axial load, 120 Nm of torsional torque, and 2800 N of transverse
forces. Custom adapters were created from metal blocks to attach the load cell to the axial
actuator and the ring holding the specimen.
2.5.4 Data Analysis

The ATI six degree of freedom load cell was used to capture all force data including,
axial force, AP force, and torsional torque. The motion tracking system was used to capture the
dorsiflexion-plantarflexion command applied to the talus. All the measured forces recorded were
plotted and analyzed against the applied command signals. Since the forces were applied to
cadaveric specimens, a higher degree of variability was expected. Additionally, since the

85
actuator applying the AP force was converted from a linear displacement actuator in series with a
spring, it was important to check the actual applied loads. Similarly, since the dorsiflexion or
plantarflexion was not directly controlled, it was also important to check the measured applied
angular displacement. The resulting AP displacement from the load control ISO standard was
compared to the suggested AP parameters in the displacement ISO standard. In addition, the
observed tibial rotations from the torque control ISO standard were compared against the
suggested tibial rotations from the rotation control ISO standard.
3. Results

3.1 Overview
In this section, the results of the five sets of experiments conducted during the course of
this project are presented. The results of the first experiment are shown as three clustered
boxplots demonstrating in vivo gait analysis measurements of shank angle, ankle angle, ground
reaction force angle, and ground reaction force magnitude during gait. The results were then
implemented as input parameters in an in vitro cadaveric simulation. Scatter plots and mean
values for AP displacement and inversion-eversion rotations are presented for each instance of
interest during gait under three loading modalities: (1) Direct AP Loading, (2) Constrained
Cyclic Loading, and (3) Unconstrained Cyclic Loading. These scatter plots are presented in
pairs, contrasting natural specimens with all ligaments and soft tissues intact to the contralateral
side with implanted artificial TARs.
Following this section, the results of the explanted (retrieval) analysis of total ankle
replacement polyethylene inserts are presented. Specimens were examined for: burnishing,
scratching, mechanical damage, pitting, and embedded particles, using a semi-quantitative

86
grading system to assess the severity of damage. Then, the mean damage severity scores are
compared by anatomical location and contrasted against the tibiotalar kinematics found in the
cadaveric simulation. Next, the kinematics of the natural and artificial TAR, after performing
subtalar fusion of the calcaneus and talus, are presented, comparing unfused and fused states for
all instances of gait.
The final section presents the results from the experiment in which the ISO 22622 force-
controlled protocol was applied to the tibiotalar joint of cadaveric lower limbs with natural
ligaments intact. The ISO standard was developed as a guide for joint wear simulation testing,
which involves direct manipulation of artificial total ankle replacement components.
First, the results compare the ISO command profile programmed for each of the input
parameters, to the corresponding measured profile for each specimen throughout the dynamic
gait simulation. Next, the resulting AP displacement and internal-external rotation of the talus
with respect to the tibia are presented and compared to the corresponding values specified by the
ISO 22622 displacement-controlled standard.
3.2 Gait Analysis

Each subject completed five trials each of the four specified gait types: slow gait shod,
slow gait barefoot, fast gait shod, fast gait barefoot. For each task, four discreet moments of
interest were analyzed: Heel strike (HS), Maximum Weight Acceptance (MWA), Mid-stance
(MS), Push-off (PO).
The angles of the shank and ground reaction force were taken with respect to the vertical.
Any angle lying to the left of the vertical was considered negative while any angle lying to the
right of the vertical was considered positive. The vertex of the shank angle was taken at the

87
malleolus and extended parallel to the shank. The ground reaction force angle was measured
along the overlaid ground reaction force vector. The ankle angle was taken as a relative angle
between the shank and the foot.
3.2.1 Shank Angle

Shank angle ranged from about -22° to 28° throughout the stance phase of gait. As
expected, the shank angle began as a negative angle (left of the vertical), or anteriorly to the
center of mass as the subject traversed to the right of the room. This position aided in bracing the
body and slowing down the moment of inertia of the body (heel strike and maximum weight
acceptance). As the subject transitioned from braking to propulsion, the center of mass slowly
shifted over the shank and the center of pressure (mid-stance). Following successful braking, the
center of mass continued to translate over the center of pressure and continued into propulsion as
the shank maneuvered into the push-off position, or a positive angle (right of the vertical) with
the shank posterior to the center of mass. From Figure 25, it can be seen that fast shod had the
greatest variance in data measurements, possibly suggesting that as subjects traverse more
quickly, the body has more momentum and greater horizontal forces are needed to brake and
resist the center of mass.

88

Figure 25. Boxplot showing shank angle during the four points of interest. Vertical is represented as 0, Left of Vertical is
Negative, Right of Vertical is Positive.
3.2.2 Ankle Angle

From Figure 26, the boxplots follow the general movement an ankle experiences during
the stance phase. As the subject’s heel struck the ground this force caused a small ankle moment
which resulted in slight plantarflexion of the ankle (angle greater than 90°). From here the
maximum braking force was applied to the heel causing the ankle moment to increase, in turn
increasing the plantarflexion of the ankle. The force was then reduced, during midstance
allowing for the ankle to return to neutral and transition to dorsiflexion. This dorsiflexion
continued to increase as the center of mass translated anteriorly to the center of pressure and
transitioned into push off allowing the body to propel forward.

89

Figure 26. Boxplot showing ankle angle during the four points of interest. 0 represents neutral, Negative represents
plantarflexion, Positive represents dorsiflexion.
3.2.3 Ground Reaction Force Angle

Ground reaction forces describe the interaction of the body with the ground. Due to a
high variance of individualized cadence, a larger range of ground reaction force directions were
recorded during heel strike. This can be attributed to subjects striking the ground with different
aspects of their foot, whether it be the heel-, mid-, or forefoot. However, once initial contact was
established, the range of angles for the ground reaction force was very tight. As expected, during
maximum weight acceptance, the ground reaction force was directed posteriorly due to the shank
slowing down the center of mass. From here the ground reaction force became nearly vertical
(zero degrees) as the center of mass was mostly over the center of pressure. Following the

90
behavior of the shank, the ground reaction force shifted anteriorly (positive angle) during push
off as the center of mass translated anteriorly to the center of pressure.

Figure 27. Boxplot showing shank angle during the four points of interest. Vertical is represented as 0, Negative values represent
braking forces, Positive forces represent propelling forces.

91
The following tables present a summary of the shank angle, ankle angle and ground
reaction force angle, at each measured instance.

Table 2. Gait parameters found from gait experiment. (N = 8)
Gait Parameters
HS MWA MS PO
Shank Angle -19° -3.5° 6.5° 24.5°
Ankle Angle -5° -5° 2° 7°
GRF Angle -1.5° -10° -1.5° 9°

Positive shank angles represent angles to the right of vertical axis while negative angles represent angles to the
left of vertical axis. Positive ankle angles represent dorsiflexion while negative angles represent plantarflexion.
Positive GRF angle represent angles to the right of the vertical axis (propelling force) while negative angles
represent angles to the left of the vertical axis (braking force).

3.3 Biomechanical Simulation of Intact Cadaveric Ankles Versus Implanted
Artificial TAR

While Table 2 showed the acquired angles from the in vivo data collections, the applied
angles were adjusted to create an absolute vertical ground reaction force. Due to the complexity
and lack of repeatability of applying an angled ground reaction force, a ball and socket joint was
implemented into the experimental design to force the applied ground reaction force to be a
known absolute vertical. To account for the ground reaction force being vertical, the following
table shows the relative angles that were applied during the experimental setup.

92
Table 3. Testing parameters that were used during testing for each position.
Testing Parameters
HS MWA MS PO
Shank Angle -20° 6° 8° 15°
Ankle Angle -5° -5° 2° 7°
GRF Angle 0° 0° 0° 0°

3.3.1 Direct Anterior-Posterior Loading

This method used a linear actuator to directly apply a 3mm anterior displacement, return
back to neutral, and then apply a 3mm posterior displacement. Each position was held for a few
seconds to allow for stabilization of the signal. For both the natural and total ankle replacement
minimal AP displacement was observed. All the specimens experienced less than one millimeter
of displacement except for one occurrence of -1.01mm at maximum weight acceptance for
specimen 2L and six millimeters during heel strike for specimen 3R.

93
Table 4. This table shows the AP Displacement of the talus with respect to the tibia during direct anterior-posterior loading.

AP Displacement for Direct AP Loading (mm)

Implant Heel Strike
Max Weight
Acceptance
Mid-Stance Push Off
Specimen 1
(Right)
N 0.68 -0.58 -0.33 0.40
Specimen 2
(Left)
N -0.28 -1.01 0.15 -0.43
Specimen 3
(Left)
N 0.24 -0.32 0.16 -0.17
Specimen 1
(Left)
Y 0.32 0.17 -0.38 -0.37
Specimen 2
(Right)
Y -0.56 0.39 -0.41 -0.43
Specimen 3
(Right)
Y 6.00 0.14 0.13 0.25

Figure 28. Scatter plot illustrating AP displacements during Direct AP Loading for (a) Natural (left) and (b) TAR (Right)

94
Similarly, little to no rotation was recorded during this direct anterior-posterior loading. This
should be expected as the axial load was held nearly constant and the force in the AP direction
would not significantly affect a rotation in the frontal plane.

Table 5. This table shows the inversion and eversion rotation of the talus with respect to the tibia during direct anterior-posterior
loading.

Inversion and Eversion Rotation for Direct AP Loading (degrees)

Implant Heel Strike
Max Weight
Acceptance
Mid-Stance Push Off
Specimen 1
(Right)
N -0.4 0.24 0.12 0.17
Specimen 2
(Left)
N 1.74 -0.26 -0.11 -0.38
Specimen 3
(Left)
N -0.54 0.12 0.14 -0.09
Specimen 1
(Left)
Y -0.25 0.12 -0.2 0.09
Specimen 2
(Right)
Y -0.16 0.11 -0.11 -0.38
Specimen 3
(Right)
Y -0.84 -0.11 -0.08 -0.07

95

Figure 29. Scatter plot showing inversion and eversion rotations during Direct AP Loading for (a) Natural (left) and (b) TAR
(Right)
3.3.2 Constrained Cyclic Loading

For the constrained cyclic loading, the load cell was constrained in the transverse plane.
Once the specimen was placed in the correct position, a 150N preload was achieved. From here
motion tracking and load cell data acquisition was initiated. Upon initiation, the load frame
applied a sinusoidal force control signal at 0.2Hz until 300N was achieved. The specimen was
cycled at 300N for three cycles to allow multiple readings. The average of the readings for AP
displacement and inversion-eversion rotation are summarized in the tables below (Table 6, Table
7). Additionally, two plots are illustrated below for each kinematic movement of interest. Each
plot has the four instances of interest noted on the x-axis. On the y-axis, the displacement in
millimeters and rotation in degrees are labeled for AP displacement and inversion-eversion
respectively. Of each pair, the scatter plot on the left shows the motion tracked for the natural
talus. Conversely, the scatter plot on the right shows the motion tracked for the total ankle

96
replacement. For displacement, negative values represent anterior movement, while positive
values represent posterior movement of the talus. For rotations in the frontal plane, negative
values indicate eversion and positive values indicate inversion. All specimens are labelled by
their pair number.
The AP displacement recorded for the natural ankles during constrained testing were all
under 2mm. These displacements appeared to follow an upward trend from anterior to posterior
which was the expected trajectory of the talus during the stance phase of gait. Conversely, the
specimens with a TAR, appeared to have a negative trend, positive trend, and neutral trend for
the AP displacement of the talus. While seemingly random in trajectory, these components
experienced a greater magnitude of displacement than their contralateral natural side.
Table 6. This table shows the AP Displacement of the talus with respect to the tibia during constrained cyclic loading.

AP Displacement for Constrained Cyclic Loading (mm)

Implant Heel Strike
Max Weight
Acceptance
Mid-Stance Push Off
Specimen 1
(Right)
N -1.59 -1.37 0.39 0.94
Specimen 2
(Left)
N -0.43 -0.86 -0.17 0.44
Specimen 3
(Left)
N -0.25 0.47 0.16 0.38
Specimen 1
(Left)
Y -0.27 -0.46 -0.32 -0.33
Specimen 2
(Right)
Y -3.09 -0.44 0.31 2.98
Specimen 3
(Right)
Y 3.01 -0.3 0.16 -1.4
*Negative displacements represent talus anterior movement, while positive displacement represents talus posterior
movement.

97

Figure 30. Scatter plot showing AP displacement during Constrained Loading for (a) Natural (left) and (b) TAR (Right)
During the stance phase of gait, the ankle was expected to have an initial inversion and
transition to predominately eversion for the majority of the stance phase [21]. As seen from the
figure for the natural talus, a general parabolic shape from inversion, to eversion, and back to
inversion was plotted. While the majority of inversion and eversion does occur at the subtalar
joint, the tibiotalar joint also does contribute to this rotation [21]. This subtalar fusion could have
potentially affected the inversion and eversion range of motion as the TAR specimens do not
show a clear trend between all three specimens.

98
Table 7. This table shows the inversion and eversion rotation of the talus with respect to the tibia during constrained cyclic
loading.

Inversion and Eversion Rotation for Constrained Cyclic Loading
(degrees)

Implant Heel Strike
Max Weight
Acceptance
Mid-Stance Push Off
Specimen 1
(Right)
N 0.81 0.92 0.07 0.12
Specimen 2
(Left)
N -2.28 -2.72 -0.42 -1.43
Specimen 3
(Left)
N 0.65 0.34 -0.08 0.26
Specimen 1
(Left)
Y -0.23 0.68 -0.39 -0.53
Specimen 2
(Right)
Y 0.21 0.47 1.04 1.17
Specimen 3
(Right)
Y 0.59 0.2 -0.08 0.39

Figure 31. Scatter plot showing inversion and eversion during Constrained Loading for (a) Natural (left) and (b) TAR (Right)

99
3.3.3 Unconstrained Cyclic Loading

Following constrained cyclic loading, the load cell was released and allowed to move
freely in the transverse plane for unconstrained cyclic loading. Once again, the specimen was
maneuvered to the correct position for the moment of gait being tested. From here a 150N
preload was achieved in small incremental increases in load. During each step-up, the angle of
the ankle was carefully monitored. If any deviation occurred from the desired testing angle,
turnbuckles were adjusted to correct the angles. From here the motion tracking system and load
cell data acquisition system were initiated. Upon initiation, the load frame applied a sinusoidal
force control signal at 0.2Hz until 300N was achieved. The specimen was cycled at 300N for
three cycles to allow multiple readings. The average of the readings for AP displacement and
inversion-eversion rotation are summarized in the tables below (Table 8, Table 9). Two pairs of
plots are illustrated below for each kinematic movement of interest. These plots contain the four
instances in time that were applied, on the x-axis. On the y-axis, the displacement in millimeters
and rotation in degrees are labeled for AP displacement and inversion-eversion respectively. For
each pair, the scatter plot on the left shows the motion tracked for the natural talus. Conversely,
the scatter plot on the right shows the motion tracked for the total ankle replacement. For
displacement, negative values represent anterior movement, while positive values represent
posterior movement of the talus. For rotations in the frontal plane, negative values indicate
eversion and positive values indicate inversion. All specimens are labelled by their pair number.
The specimens with the natural talus displayed a very clear trend of anterior to posterior
displacement between the first two specimens. Additionally, these specimens displaced a much
wider range than previous load methodologies. The TAR group of specimens also showed a clear

100
trend between all three specimens. All specimens had an anterior to posterior positive trend from
heel strike to mid-stance then experienced a drop back to anterior displacement for push off.

Table 8. This table shows the AP Displacement of the talus with respect to the tibia during unconstrained cyclic loading.

AP Displacement for Unconstrained Cyclic Loading (mm)

Implant Heel Strike
Max Weight
Acceptance
Mid-Stance Push Off
Specimen 1
(Right)
N -1.26 -0.74 -0.19 1.46
Specimen 2
(Left)
N -1.5 -0.82 -0.28 0.97
Specimen 3
(Left)
N -0.48 0.34 -0.67 -0.35
Specimen 1
(Left)
Y -0.63 -0.29 0.54 0.26
Specimen 2
(Right)
Y -0.89 -0.79 0.27 -1.96
Specimen 3
(Right)
Y -0.71 -0.24 -0.26 -1.9
*Negative displacements represent talus anterior movement, while positive displacement represents talus posterior
movement.

101

Figure 32. Scatter plot showing AP displacement during Unconstrained Loading for (a) Natural (left) and (b) TAR (Right)
Unconstrained cyclic loading produced consistent trends for all three natural specimens.
These specimens observed a negative trend from inversion to eversion as expected. In general,
the TAR specimens also experienced the same trend. Two of the three had a negative trend while
the last specimen rotated less than half a degree of rotation throughout the experiment.

102
Table 9. This table shows the inversion and eversion rotation of the talus with respect to the tibia during unconstrained cyclic
loading.

Inversion and Eversion rotation for Unconstrained Cyclic Loading
(degrees)

Implant Heel Strike
Max Weight
Acceptance
Mid-Stance Push Off
Specimen 1
(Right)
N 0.46 0.51 0.1 0.74
Specimen 2
(Left)
N 2.06 -1.36 -0.93 -1.6
Specimen 3
(Left)
N 1.27 0.48 0.79 -0.21
Specimen 1
(Left)
Y 1.07 0.45 -0.28 -1.55
Specimen 2
(Right)
Y 0.22 -0.29 0.81 2.24
Specimen 3
(Right)
Y 0.49 0.11 0.37 0.19

103

Figure 33. Scatter plot showing inversion and eversion during Unconstrained Loading for (a) Natural (left) and (b) TAR (Right)

3.4 Biomechanical Simulation of Cadaveric Specimens with Subtalar Fusions

For this experiment, a subtalar fusion was performed by an experienced orthopaedic
surgeon who also implanted the total ankle replacements, for all specimens (n=6). Two
cannulated lag screws commonly used in these surgeries were implanted through the calcaneus
and into the talus to fuse the two bones into a single unit. Following these surgeries, the same
testing methodologies (Direct AP Loading, Constrained Cyclic Loading, Unconstrained Cyclic
Loading) used in the previous biomechanical simulation was implemented. While all three
loading methodologies were applied, the following graphs only present the data for the
unconstrained testing method. The AP displacement for the natural talus had a high variability
between specimens after a fusion. While the differences were very apparent in heel-strike, the
readings were more consistent for maximum weight acceptance and mid-stance. However,

104
between the three specimens no apparent trend was shown. For the specimens with total ankle
replacements, the magnitudes were also quite high with high variability.

Figure 34. Scatter plot showing AP Displacement after Fusion during Unconstrained Loading for (a) Natural (left) and (b) TAR
(Right)
The pair of graphs below plot the inversion and eversion of the natural and total ankle
replacement specimens after fusion. As expected, the rotations between the calcaneus and talus
are quite small due to the two screws fixing them together. Overall, for the specimens with total
ankle replacements, there was little to no movement. However, the third specimen measured a
substantial magnitude of locations, reaching over four degrees at push off.

105

Figure 35. Scatter plot showing inversion and eversion after fusion during Unconstrained Loading for (a) Natural (left) and (b)
TAR (Right)

From the strain gauge data, the maximum and minimum stresses were analyzed. A paired t-test
analysis was performed to compare the natural ankle with a subtalar fusion to the TAR
specimens with a subtalar fusion. None were found to be statistically significant.

Table 10. P-Values from paired t-test analysis between fused natural and fused TAR specimens.
HS MWA MS PO
Natural vs TAR p = 0.73 p = 0.51 p = 0.29 p = 0.9

106
3.5 Patterns of Damage on Retrieved TAR Polyethylene Inserts

Images of all 14 retrievals were captured using a Keyence Microscope. These images of
the articulating surfaces highlighted five features of damage: burnishing, scratching, mechanical
damage, pitting, and embedded particles. An example of each type of damage is displayed in the
following figure (Figure 36).

Figure 36. From left to right examples of: (i) Burnishing, (ii) Scratches, (iii) Mechanical Damage, (iv) Pitting, (v) Embedded
Particles
Using the semi-quantitative method described in the methods, each observer
independently graded all 14 specimens. These ratings were then averaged within each respective
quadrant (Table 11).

Table 11. Scores for each type of damage in each quadrant, averaged among the three investigators, with standard deviations in
parentheses.
Quadrant 1 Quadrant 2 Quadrant 3 Quadrant 4
Burnishing 2.3 (1.5) 2.3 (1.4) 2.2 (1.5) 2.3 (1.5)
Scratches 2.1 (1.2) 2.1 (1.1) 2.3 (1.2) 2.3 (1.2)
Mechanical
Damage
(Gouges, cuts)
0.9 (0.7) 0.6 (0.5) 0.8 (0.9) 0.7 (0.6)
Pitting 1.4 (1.1) 1.3 (1.0) 1.5 (1.0) 1.65 (1.0)
Embedded
Particles
0.4 (0.6) 0. 5 (0.7) 0.3 (0. 6) 0.4 (0.5)

107

Each specimen was graded in quadrants to provide a precise location of damage as well
as its extent. Afterwards, statistical analysis was performed to compare different anatomical
locations. Specifically, the following quadrants were combined: quadrant 1 & 3, medial or lateral
(determined by left or right implantation side), quadrant 2 & 4, medial or lateral (determined by
left or right implantation side), quadrant 1 & 2, anterior, and quadrant 3 & 4, posterior. No major
differences were found between the medial and lateral aspects for both burnishing (p = 0.96) and
scratching (p =0.33) damage. The same test failed to establish significance for burnishing
between the anterior and posterior aspects (p = 0.51). However, scratching was significantly
higher on the posterior aspect compared to the anterior aspect (p = 0.01) of the polyethylene
inserts.
There was a high correlation between the in vivo duration of the implants and burnishing
on both the anterior and posterior aspects of the polyethylene inserts(p=0.01). (Table 12) In
contrast, there was a negative correlation between the in vivo duration and scratching on both the
anterior and posterior aspects; however, these correlations were not statistically significant.
(Table 12)

Table 12. Results of a correlation test between in vivo duration and burnishing and scratching.

Burnishing Scratching
Anterior Posterior Anterior Posterior
In Vivo
Correlation
Coefficient
0.67 (p=0.01) 0.68 (p=0.01) -0.32 (p=0.27) -0.40 (p=0.16)

Deleted: ¶

108
To measure the interobserver error among all three independent observers, the absolute
error was calculated for anterior burnishing and scratching, as well as posterior burnishing and
scratching, (Figure 2) summing the two anterior quadrants, quadrants 1 and 2, and two posterior
quadrants, quadrants 3 and 4. No statistically significant differences between any two observers
were found for burnishing or scratching in any location (p>0.25 for all pair-wise nonparametric
comparisons).

Figure 37. Boxplot presenting error between each observers' ratings for burnishing and scratching damage.

109
3.6 Biomechanical Simulation of Force Control Wear Testing in a Cadaveric Model

The purpose of this experiment was to compare the two methods of applying the input
parameters for wear testing suggested in the ISO 22622. The standard allows either force-
controlled or displacement-controlled testing parameters. As discussed in the Methods section, it
is ambiguous if both would produce the same results. This is particularly unclear for the ankle
joint as it relies on both surrounding ligaments and articulating surfaces for stability. To
investigate the differences, the force-controlled protocol proposed in the ISO standard was
applied to 12 cadaveric specimens and the resulting displacements and rotations of the talus with
respect to the tibia were recorded through optical motion tracking. These recorded displacements
values were then compared to the displacement-controlled protocol proposed in ISO 22622.
All specimens were prepared by removing all extraneous tissues while keeping the
critical ligaments intact. The specimens were then potted, so they could be placed in custom load
frame adapters. Additional preparation included placing two screws through the talus from the
posterior aspect to allow for direct manipulation of this bone. A halo of bone cement was used to
encapsulate these screws which allowed the load frame to control the dorsiflexion and
plantarflexion of the talus.
According the ISO Standard 22622, which provides the recommended testing parameters
for the evaulation of total ankle replacement implant bearing surface wear, four motions and/or
loads must be controlled. Specifically, the standard specifies that the axial load should be applied
through force control, without restricting the axial displacement. Sagittal rotation, or
dorsiflexion/plantarflexion, should be applied by displacement control, as opposed to force (or
moment) control. The remaining two motions, anterior-posterior translation (AP, or sagittal
motion) and internal/external rotation, may be controlled by either force or displacement control.

110
In the present experiment, axial force control, angular plantar/dorsiflexion control, AP force
control, and internal/external moment control were applied. From this, the resulting AP
displacements of the talus and internal/external rotations of the talus were measured, to compare
to the propsed ISO alternative displacment-controlled profiles.
3.6.1 Application of ISO Load Profile Commands

The axial load, AP force, intenal/external moment load profiles were extracted from ISO
22622 to determine the force-controlled parameters for the servohydraulic load frame. Similarly,
the dorsiflexion/plantar flexion rotation parameters were also extracted from the standard to
determine the displacement-controlled parameters. Collectively, these four profiles defined the
testing conditions for the experiment.
As the ISO standard is meant for the testing of metal and plastic implants alone, the
profiles had to be scaled accordingly, to protect the cadaveric specimens. Therefore, the peak
loads were reduced to approximately 50% of the magnitudes specified in the standard. This
reduced the maximum axial load to 1,183 N, the maximum AP force to 129 N, and the maximum
moment about the tibial shaft (internal-external rotation) to 2.9 Nm. Since the dorsiflexion and
plantarflexion profile was applied in displacement control for both the load and displacement
profiles, these parameters remained unchanged.
Each specimen was taken through four complete cycles of the load profile. The first cycle
was used to precondition the specimens. The data for the remaining three cycles were considered
the experimental data, taking the averages of all three for analysis. Throughout the test, all of the
applied loads were recorded by a 6 degree of freedom load cell (ATI Mini58, ATI-IA, Apex,
NC) positioned superiorly to the specimen.

111
The loads and/or displacements specified by the standard were applied to quantify the
natural displacements of the tibia, talus, and calcaneus to those stipulated by the standard under
displacement control. In particular, the naturally occuring AP translation of the talus with
respect to the tibia, under force-controlled condtions was of interest. The graphs presented below
overlay (1) the recorded loads of all 12 specimens in gray, (2) mean of all 12 specimens in red
along with maximum and minimum bars at 10% increments, and (3) programmed load in blue
crosses.
3.6.2 Axial Force Application

The applied axial force initiated just below 200 N of compression, with the first peak
occurring at approximately 10% of the gait cycle coinciding with maximum weight acceptance.
From here the load decreased as midstance was simulated. The load then quickly increased to
about 1200 N as the ankle reached the push-off state, where maximum loading occurred.
Following this, the ankle entered the pre-swing phase where the axial load was reduced, as
expected. At 60% the stance phase terminated, and the swing phase begun, where the minimum
axial load was approximately 100 N, which remained constant until 100% completion of the gait
cycle.

112

Figure 38. The axial force applied to each specimen is graphed in gray, the average of all the specimens is shown in red, and the
programmed command is shown in blue crosses.

3.6.3 Flexion Application

Dorsiflexion and plantarflexion were applied through a displacement-controlled actuator.
This actuator manipulated the talus directly, through the posteriorly implanted screws, as
described in the Methods section. In a natural gait cycle, as the heel made initial contact with the
ground, the ankle rotated into plantarflexion. This was accurately described with a negative
rotation during the first 5% of the gait cycle in the present experiment. Following the initial heel
strike, the ankle entered dorsiflexion to allow the center of mass of the subject to translate
forward. This rotation occurred until approximately 50% of the gait cycle or just after peak
loading. Afterward, the pre-swing phase was entered, and the ankle began to enter plantarflexion
which is indicated by the negative trend from 50% to 65%. During the swing phase, the ankle

113
began to shift back to dorsiflexion, as accurately implemented from 65% to 100% of the gait
cycle.

Figure 39. The dorsiflexion and plantarflexion applied to each specimen is graphed in gray, the average of all the specimens is
shown in red, and the programmed command is shown in blue crosses.

3.6.4 AP Force Application

The anterior-posterior (AP) force was applied directly to the tibia, as described by ISO
22622. A positive force represented an anterior force while a negative force represented a
posterior force. As described in the Methods section, a spring with a constant of 23.3 N/mm was
used to convert a displacement-controlled actuator program commands to control the motion
with force, as opposed to displacements.
The AP force initiated by pushing the tibia anteriorly, which correlated with the anatomic
dorsiflexion of the talus. As the talus typically dorsiflexes following some degree of initial
Dorsiflexion
Plantarflexion
Dorsiflexion and Plantarflexion

114
plantarflexion, this movement is consistent with observations from gait studies. As seen from the
axial load command, following the peak axial load at about 40%, the ankle was expected to
plantarflex, represented by a shift from an applied anterior force to posterior force. At 60%, the
swing phase is initiated, and also as expected, there were minimal AP forces on the talus.

Figure 40. The AP force applied to each specimen is graphed in gray, the average of all the specimens is shown in red, and the
programmed command is shown in blue crosses.
3.6.5 Internal and External Torque Application

The internal-external rotation moment was applied directly to the shaft of the tibia. The
original command was designed for a left lower extremity. Therefore, for right foot specimens,
the command was inverted. For analysis, the results were reversed to allow for a more uniform
curve alignment. At the beginning of a gait cycle, the tibia was initially inverted. This was
implemented with an internal or positive moment for the initial 15%. Subsequently, an external
negative moment was applied until approximately 45% of the gait cycle. As the pre-swing phase

115
was entered, the tibia began to internally rotate, which was implemented through a reduced
negative moment up until the initiation of the swing phase (60% of gait cycle).

Figure 41. The internal and external moment applied to each specimen is graphed in gray, the average of all the specimens is
shown in red, and the programmed command is shown in blue crosses.

3.6.6 Resulting Measured Displacements and Rotations

The four testing parameters discussed above, were applied to determine the natural
displacement of the talus, relative to the tiba. Specifically, the axial load, AP force, and
intenal/external moment load profiles detailed above, defined the force-controlled parameters for
the servohydraulic load frame. Similarly, the dorsiflexion/plantarflexion rotation parameters
defined the displacement-controlled rotations in the sagittal plane. Collectively, these four
parameters defined the testing conditions. Using these applied command profiles, the resulting

116
motion of the tibia, talus, and calcaneus were captured by the motion tracker, and the
displacements of the individual bones were calculated. Specifically, the AP displacement and
axial rotation of the talus with respect to the tibia was of interest. Similar to the command profile
plots, the mean motion of all 12 specimens were plotted in gray, for comparison to their
corresponding suggested displacement-controlled command profiles in ISO 22622, represented
with blue crosses. Additionally, boxplots at 10% increments were overlaid in these plots.

3.6.7 Measured Anterior Posterior Displacement Curve

The measured displacement of the talus was consistent with the trend observed during
natural gait. Specifically, during the initial heel strike as plantarflexion occurs, the talus should
displace anteriorly (negative direction). From 10% to about 45% of the loading cycle, the talus
moved posteriorly (positive direction), which was expected, as the ankle is in dorsiflexion in this
phase of natural gait. Subsequently, plantarflexion occurs during the pre-swing phase, which was
illustrated by a negative displacement trend seen from 45% to about 60% of the loading cycle.
Finally, during the swing phase of natural gait, the talus reverts back to dorsiflexion, which
corresponded to the observed posterior displacement at the end of the gait cycle. At 50% of the
suggested load, the mean resulting AP displacement spanned a maximum of 1.3 mm of anterior
displacement and 0.5 mm of posterior displacement. The maximum recorded anterior
displacement of all specimens was 3.3 mm and the maximum posterior displacement, 3.0 mm.
According to the originally proposed ISO 22622, the maximum anterior displacement should
reach 3.17 mm, while the maximum posterior displacement should be considerably lower than
that observed, at 0.88 mm (Figure 42). However, the finalized ISO 22622 provided a modified
AP displacement profile (Figure 43).

117

Figure 42. The AP displacement recorded from each specimen is graphed in gray, the average of all the specimens is shown in
red, the ISO 22622 proposed displacement command is shown in blue crosses, and the boxplot at 10% intervals are shown.

Figure 43. The AP displacement measured from each specimen with the Finalized ISO 22622 displacement profile command
shown in blue.

118
3.6.8 Measured Internal and External Rotation of the Tibia

Similar to the AP displacement of the talus, the rotation of the talus was recorded using a
motion tracking system. To achieve the required 50% initial internal torque, the tibia was rotated
on average 0.77°. However, the rotation command profile has the specimen rotate 2.45°. To
achieve the maximum external torque, the tibia was rotated on average -3.2°. The rotation
command profile has a rotation of 8.02° to reach maximum external rotation.

Figure 44. The tibial rotation recorded from each specimen is graphed in gray, the average of all the specimens is shown in red,
the ISO 22622 proposed rotation command is shown in blue crosses, and the boxplot at 10% intervals are shown.
From these command profiles, the resulting motion of the tibia, talus, and calcaneus were
captured. Specifically, the AP displacement and axial rotation of the talus with respect to the
tibia was of interest. Similar to the command profile plots, the mean motion of all 12 specimens
were plotted in gray and the suggested displacement command profile in ISO 22622 was

119
represented with blue crosses. Additionally, boxplots at 10% increments were overlaid these
plots.
4. Discussion

4.1 Overview
Preclinical testing such as in vitro testing and wear testing of novel implants is a vital step
in the process of development and evaluation of successful joint replacements. Through accurate
preclinical testing methods, the in vivo performance and long-term longevity of iterations and
advances of novel designs can be properly evaluated. Moreover, different models and designs
can be scientifically compared to each other. However, due to the inherently unknown traits of
novel implants, the proper in vitro and wear testing methodologies for these implants are
unspecified and must be carefully investigated and established.
As a relatively new joint replacement, total ankle replacements (TAR) suffer from this
lack of an established preclinical testing methodology. In the limited number of TAR wear
studies published, many used widely accepted parameters for ankle flexion and axial load;
however, there is no consensus on the amount of anterior-posterior displacement that occurs in
the ankle [2, 13, 17, 117, 123, 141]. Many of these wear studies gathered their testing parameters
from previous gait analysis studies with incomplete data sets [32, 118, 144]. While gait studies
do provide valuable insight for ankle flexion and axial load, parameters such as AP displacement
cannot be directly observed without invasive means. Therefore, in vitro cadaver models can
provide invaluable insight into the in vivo kinematics of the ankle joint unable to be captured
through gait studies. However, as with wear studies, there is a clear lack of an established in vivo
testing methods to be used when investigating ankle cadavers.

120
The experiments performed in the present project contributed to the establishment of
proper testing preclinical testing protocols by addressing all aspects of an implant’s lifecycle.
Specifically, a comprehensive study of pre-clinical (in vivo cadaver test), clinical (retrieval), and
wear (ISO force-controlled test protocol) studies were performed, as each provide unique
insights in ascertaining appropriate preclinical testing parameters and in turn key improvements
to implant success. The following section discusses the analysis from each experiment
performed. The specific aims of the present project were to:
1. Establish a model using cadaveric specimens to determine the 6 DOF motions of the
natural tibia, talus, and calcaneus under simulating physiologic gait.
2. Use this cadaveric model to evaluate 6 DOF motions of total ankle replacement
components.
3. Use explanted total ankle replacement polyethylene components to assess the modes and
locations of damage during in vivo use.
4. Use the cadaveric model to establish 6 DOF motions of the natural talus and the artificial
talar component as a function of fusion of the subtalar joint.
5. Compare the 6 DOF motions of the natural ankle joint resulting from the proposed ISO
22622 force-controlled testing standard to the proposed ISO 22622 displacement-
controlled testing standard.
4.2 Gait Analysis

Although many gait analysis studies have been published, to our knowledge none have
reported a collection of shank angle, ankle angle, GRF angle, and GRF magnitude. The initial
gait analysis experiment aimed to correlate all four within a single study. Four different tasks
were performed by eight subjects: a slow and fast gait while shod or barefoot.

121
The behavior of the shank was predictable throughout the stance phase, independent of
the variability between each task. However, it can be seen that the speed of cadence affected the
shank angle greater than if the subject was shod or barefoot. With a slower speed, subjects were
able to control their gait more predictability and had more reproducible results, indicated by
tighter boxplots. Conversely, when subjects were asked to walk at a self-selected faster speed, a
wider band of shank angles were produced. This was seen at all four instances of gait that were
of interest.
The ankle angles observed in the present experiment agreed with commonly published
literature of ankle angles throughout the stance phase of gait [21, 74, 86, 160]. There seemed to
be no difference in any of the factors included in this experiment. The ranges at each analyzed
position were wide, but this may be due to a variety of factors such as natural dorsiflexion and
plantarflexion mobility.
The ground reaction force angle also demonstrated the predicted behavior. As the subject
made initial contact with the ground, a braking force was applied, which slowly transformed into
a propelling force as the center of mass translated over the center of pressure. Interestingly, after
a wide array of ground reaction force angles were seen at heel strike, all other instances of the
gait cycle had little variance. This broad range of angles can potentially be accredited to the fact
that individual subjects interact with the ground differently. There are many subjects whom make
initial ground contact with their heel, mid-foot, or fore-foot. This can highly impact the angle of
the ground reaction force at initial contact. However, once initial contact was made, the
following ground reaction forces at maximum weight acceptance, mid-stance, and push off were
all very similar with no differences greater than five degrees.

122
The averages of these different tasks and subjects, were used as input parameters for an in
vitro cadaver test. However, while the parameters for this test had been obtained, the method of
application to a cadaveric specimen remained unclear. The following section discusses the
method of applying the parameters obtained from this gait experiment to cadaveric specimens.
4.3 Biomechanical Simulation of Intact Cadaveric Ankles Versus Implanted
Artificial TAR

The majority of published cadaveric ankle biomechanical tests often constrain either parts
of the ankle during biomechanical loading or parts of the test setup, to compensate for the natural
instability of a cadaveric ankle. While this allows for successful loading for the purpose of their
study, these constraints may affect the natural articulations, resulting in altered kinematics and
skewed data. Moreover, by locking certain ranges of motion or parts of the test setup, the
orientation the ground reaction force acts upon the system becomes unknown. While seemingly
inconsequential, this may introduce additional reaction forces, making the applied force
magnitude and direction unknown. This may then result in researchers drawing unsound
conclusions of the tibiotalar joint and its physiological response to the selected loads.

4.3.1 Direct Anterior Loading vs Constrained vs Unconstrained

In the present experiment, the effects of different loading setups were explored. The first
loading protocol (direct AP force) applied an anterior displacement load for 3 mm posterior to
neutral, followed by 3 mm anterior to neutral. The second (constrained) and third
(unconstrained) loading protocol applied a cyclic axial load from a 150 N preload to 300 N of
peak load. The difference between the constrained and unconstrained experimental setups, was
the use of locking tables to inhibit any movement, in the transverse plane, of the platform. The

123
constrained setup used the locking tables, while the unconstrained setup removed these locking
tables to allow uninhibited movement of the platform within the transverse plane. It was
hypothesized that the use of locking tables, created additional reaction forces which were
introduced to the platform; thereby, creating a resulting ground reaction force of unknown
magnitude and direction. Conversely, by removing these locking tables, the ground reaction
force was setup to be absolutely vertical if the specimen was static and immobile on the
frictionless rails. This method allowed for control of the angle the ground reaction force was
applied to the specimen.
From the scatter plot for direct anterior loading, minimal displacement was observed.
While a three millimeters posterior and anterior displacement was applied to the specimen, under
two millimeters of motion was applied to the talus. More importantly, the direction of the
recorded displacements did not follow the expected physiological kinematics observed in vivo. It
can be observed that direct anterior loading does not provide any beneficial results and should
not be used as a testing method to produce in vivo tibiotalar kinematics.
Conversely, the scatter plots which compared the displacements of the natural talus
during the constrained and unconstrained setups, illustrated the expected physiological anterior-
posterior displacement. However, when studying the trends of each specimen as it was placed
through the four instances of interest (heel strike, maximum weight acceptance, midstance, and
push-off), only the unconstrained setup showed the expected in vivo behavior. As the specimen
initiated contact with the ground during heel strike and maximum weight acceptance, the ankle
was predominately plantarflexed and the ground reaction force acted as a braking force directed
posteriorly. These conditions suggested anterior movement of the talus with respect to the tibia
which was observed by the unconstrained setup. Moreover, during midstance and push-off, the

124
opposite conditions were applied, and the ankle underwent high dorsiflexion, with a ground
reaction force directed anteriorly. As expected, and observed during the unconstrained setup, this
applied force displaced the talus initially towards the anterior aspect and then to the posterior
aspect during these four instances of gait. While the constrained setup generally showed a similar
trend that represented the above behavior for the natural ankle, it was not as consistent and
prominent as the trends produced during the unconstrained setup. Additionally, for the specimens
with total ankle replacements, the unconstrained test setup produced a clear universal trend,
while the constrained test setup had one specimen displace posteriorly to anteriorly, a second
specimen moved anteriorly to posteriorly, and the third and final specimen remained neutral
throughout each loading scenario. It should be observed from these scatter plots, that to
consistently test these specimens, the reaction force should be accounted for through the
unconstrained test setup.

Figure 45 Natural talus kinematics resulting from all three test methods

125
4.3.2 Natural vs Total Ankle Replacement During Unconstrained Loading

The unconstrained cyclic loading setup produced the most consistent results in the natural
talus. Therefore, it was used to evaluate the kinematics of one total ankle replacement (TAR)
design (Infinity, Wright Medical). The scatter plots analyzed below, compared the natural talus
to the TAR during unconstrained cyclic loading (Figure 46). It can be observed that during heel
strike, maximum weight acceptance, and mid-stance the total ankle replacement correctly
simulated the measured natural movement of the tibiotalar joint. However, during push off there
was a sharp decline and a high anterior displacement where the opposite was expected.
Interestingly, Conti et al in 2006 showed similar behavior in a ten-patient in vivo video
fluoroscopy dynamic gait study [32] (Figure 47).

126

Figure 46. Unconstrained testing results comparing the Natural talus to the TAR

Figure 47. Plots adapted from “Kinematic analysis of the Agility Total Ankle During Gait,” Conti et al. 2006 showing trends
similar to our cadaver model from a video fluoroscopy study of dynamic gait.

127
Conti et al. observed that during the stance phase of gait, the contact point of the tibia on
the talus was posterior to the midline of the talus and transitioned anterior to the midline of the
talus. The movements recorded by Conti et al. align with the anterior to posterior displacements
of the talus observed in this cadaveric model. Similarly, for the specimens with TARs, Conti et
al. recorded an initial rise of posterior displacement of the talus, followed by a drop back to
anterior displacement. This behavior closely matched the kinematics produced, for both the
natural and artificial TAR, from the cadaveric model during the unconstrained cyclic loading
method developed in this project.
This contrasting behavior may be attributed to a few factors. The first factor may be due
to the intricacies and difficulties of TAR surgery. While experienced, the orthopaedic surgeon
may have had a difficult time fully seating the talar component over the apex of the natural talar
dome. This could cause the implant to be placed too anterior, causing the observed abnormal
behavior when the tibial component articulates around the posterior surface of the talar dome
rather than the anterior surface during push-off (Figure 48 Left). Although a different implant
design was used, Komistek et al. saw similar behavior in vivo. Komistek et al. found that in ten
patients, the implanted ankles had contact points on the more posterior aspect of the talus which
was possibly related to surgical technique [79]. Another reason for this abnormal behavior, could
be explained if the implant design inhibited dorsiflexion of the ankle. By constraining the
dorsiflexion of the ankle, the implant will again interact with the posterior side of the talar dome
during push-off causing an anterior displacement, when a posterior displacement is expected
(Figure 48 Right). As there is very limited clinical research to support this theory, this can only
be a theoretical explanation.

128

Figure 48. Left: Potential malalignment of the talar dome being seated too far anteriorly. Right: Illustration of talar dome
potentially being pushed on the posterior side with the specimen in max dorsiflexion.
4.4 Subtalar Fusion of Natural vs Total Ankle Replacement Specimens

Subtalar fusions are often performed on patients with TAR; therefore, the biomechanical
effect is of interest [62, 75, 143]. Following completion of the previous cadaver experiment,
subtalar fusions were performed on all six specimens. The top three peaks for each instance of
gait analyzed was plotted for all the specimens. The graphs below were presented in the results;
however, the unfused and fused graphs are now shown side-by-side to allow for ease of
comparison (Figure 49). For the natural talus, there was a much higher magnitude of AP
displacement. This was generally consistent with Witkoski et al’s findings that hindfoot fusion
increased the magnitude of AP displacement [158]. This increase in displacement could cause an
overuse of the neighboring joints resulting in the accelerated degeneration of these joints often
associated with fusions. In addition to an increase in AP displacement magnitudes, the resulting

129
kinematic trends following fusion appeared altered and random. This suggested that the direction
and trajectories of the fusion screws could potentially have a more consequential effect than just
increasing the magnitude of displacement.

Figure 49. AP displacement for the natural talus before fusion (left) and after fusion (right).

The scatter plot for AP displacement of the unfused and fused TAR specimens is shown
below (Figure 50). These plots have also been previously presented in the results, but not
simultaneously as a pair. Similar to the natural specimens, the magnitude of displacement
increased with fusion. Interestingly, the increased displacement only occurred in the anterior
direction. This may be explained by Kim et al.’s findings that following a subtalar fusion, the
talus was anteriorly displaced after surgery [76]. This physical anterior displacement of the talus

130
could be a factor into why more anterior displacement was seen in the fused group. Another
point of interest is that overall, the trends remained the same for the TAR group, but differed
wildly for the natural specimens. This may be explained by the physical constraints of the
artificial ankle replacement. Due to the high conformity of the implants, the trajectory of the
displacement is predefined and just increased in magnitude. Conversely for the natural
specimens which were free to perform its natural articulations, different trends may be produced
in addition to the increased AP displacement magnitude.

Figure 50. AP displacement for the replaced ankle before fusion (left) and after fusion (right).

Through the strain gauges placed at the distal tibia, the stresses produced by the natural
joint and the TAR could be studied. The highest amount of tensile strains observed was 1,000
microstrain. The highest amount of compressive strains was 1,500 microstrain. From a paired t-

131
test analysis it was observed that there were statistically no differences between the natural and
specimens with TAR. This may indicate that there was no strain shielding that occurred from
these implants. Consequently, this may indicate that the tibia or specifically the medial malleolus
may be at risk of fracture from the tibial tray.
4.5 Patterns of Damage on Retrieved TAR Polyethylene Inserts

Another key component in improving implant designs is the study of explants.
Specifically, these retrieved implants provide insight into how these components were
articulating and functioning within the body during everyday life. Typically, implants like total
hip replacements can be scanned using a machine such as a coordinate measuring machine and
compared to a theoretical sphere to indicate areas of wear. However, as a result of the intricate
surfaces of total ankle replacements (TAR), a simple shape cannot be fit to the replacement.
Instead, each unique TAR would need to be scanned before being implanted within the patient to
give a basis of which to compare the specific retrieval to. Since this would be a great logistical
challenge, retrieval studies of TAR mainly consist of damage ratings rather than quantitative
wear calculations. However, these ratings can provide valuable insight into how TARs are
functioning within the body and clues to why they are failing.
Major modes of failure of total ankle replacements have been identified as loosening,
polyethylene wear, malalignment, and abnormal kinematics [29, 38, 52, 80]. These failures are
closely related to non-uniform distribution of contact stresses. Previous retrieval studies have
examined the oxidation of polyethylene inserts, wear particles produced from polyethylene,
polyethylene fracture case studies, and damage modes [3, 7, 38, 156]. The purpose of the
present experiment was to build upon the previous knowledge base by analyzing the locations of
each type of damage against articulation of the polyethylene inserts. Based on the natural

132
kinematics of an ankle, we hypothesized that the damage would be uniformly distributed along
the anterior-posterior (AP) axis of the articulating surfaces. Only fixed bearing implants were
selected to reveal the worst-case scenario of damage by non-uniform contact stresses, since
mobile bearing inserts hypothetically minimize such damage at the risk of having two articular
surfaces, both of which may produce wear debris. Furthermore, the implants for the present
experiment were selected to have a consistent sterilization method in order to provide
comparable results.
In all quadrants, burnishing and scratching were the most prevalent damage modes.
There was also a presence of light mechanical damage, pitting, and a minute amount of
embedded metal particles. The majority of the specimens in this cohort failed due to loosening.
This multifactorial failure mode resulting from surgeon, patient, or implant factors may be
attributed to wear, device design, or biomechanics of implantation, all of which may be directly
related to a non-uniform distribution of contact stresses. This was consistent with our finding of
significantly higher amounts of scratching on the posterior aspect of these polyethylene inserts
when compared to the anterior aspect (p = 0.01). While burnishing was not statistically
significant between the anterior and posterior aspects, there was a high correlation between
burnishing of both the anterior and posterior aspects against the in vivo duration of the implant
(anterior: correlation coefficient 0.67, p=0.01, posterior: correlation coefficient 0.68, p=0.01).
Conversely, scratching did not produce a correlation with the in vivo duration of the implant
(anterior: correlation coefficient -0.32, p=0.27, posterior: correlation coefficient -0.40, p=0.16).
This finding indicated that the amount of burnishing damage increased as a function of follow-up
time. This finding is consistent with analysis of damage in total hip replacements [98] and total
knee replacements [8].
Deleted: posterior

133
It has been established that during gait as the ankle transitions from plantarflexion to
dorsiflexion the talus should move anteriorly to posteriorly [32, 86, 87] (Figure 51). From this
kinematics, the polyethylene insert would be expected to experience uniform damage throughout
its articulating surface. However, a higher incidence of scratching was observed on the posterior
aspect of the polyethylene inserts. This may indicate that, while total ankle replacements have
been shown to restore the range of motion during gait, the anterior to posterior length that the
talus must span throughout gait is being constrained [120, 153].

Figure 51. Illustration showing expected talus displacement in natural ankle from Heel Strike to Push Off.
This non-uniform distribution may be a contributing factor in the high frequency of loosening
and other failure mechanisms associated with total ankle replacements. In a wear study of TAR,
Affatato et al. found a similar trend of a high incidence of localized linear wear [3]. However,
they observed the wear to be highest on the anterior aspect of their specimens. This may be due
to a different TAR design used in their study. Specifically, the implant used in Affatato et al.’s
study was a mobile bearing design which has additional degrees of freedom to articulate.

134
Overall, our findings cannot be solely attributed to the implant design. As with all
orthopaedic implants, the general three pillars of success are: implant design, surgical variables,
and patient characteristics. The present experiment had limitations. Specifically, the performance
of these implants was not solely contingent on implant design, but also on proper alignment of
both components, and each patient’s ligamentous laxity and overall ankle stability. However, due
to the limited availability of clinical data, we were unable to analyze these additional factors.
Moreover, the sample size was small. Nevertheless, some correlations emerged, possibly related
to kinematics of the talar component.
Despite improvements in total ankle replacements, long-term outcome remains poor. The
present experiment of retrieved polyethylene inserts indicated that constraints built into the
design for anterior to posterior translation of the talus may contribute to an increase of localized
posterior damage. This localization of damage supports the theory that a non-uniform
distribution of stresses occurs in vivo, leading to possible edge loading and the eventual failure of
this type of joint replacement. Moreover, these damage maps agreed with the results of the
cadaver model previously presented. As seen in the unconstrained testing of the TAR specimens,
a higher incidence of anterior displacement of the talus occurred throughout each gait phase.
This would indicate a higher amount of localized posterior damage as seen in these retrievals.
While this is a small sample size and only one specific model of a TAR was studied, this
retrieval experiment provides some support for the necessity of an unconstrained testing model.
4.6 Biomechanical Simulation of Force Control Wear Testing in a Cadaveric Model

For the evaluation of the bearing surface wear of total knee replacement implants, Sutton
et al observed a wide discrepancy between the displacements resulting from the published ISO
force-controlled protocol, to the published displacement-controlled parameters [146].

135
Specifically, when the investigators tested the specimens under force-controlled conditions and
measured the resulting freely occurring linear and angular displacements, the reported tibial
internal/external rotations were more than three times larger than the prescribed displacement-
controlled rotations. Similarly, the freely occurring AP transitions resulted in an inverse
waveform from the prescribed ISO AP displacement [146].
For the evaluation of total ankle joint replacement bearing surface wear, both force and
displacement control methods have been proposed in the ISO 22622. However, it is unknown
whether the displacements recommended by the ISO standard are comparable to the proposed
force control ISO standard. The purpose of this experiment was to evaluate the differences and
similarities between the two testing profiles.
4.6.1 Determination of ISO Load Profile Commands
ISO standard 22622 describes recommended four motions and/or load parameters for the
evaluation of total ankle replacement implant bearing surface wear. In particular, the standard
specifies that the axial load should be applied through force control, without restricting the axial
displacement. Sagittal rotation, or dorsiflexion/plantarflexion, should be applied by displacement
control, as opposed to force (or moment) control. The remaining two motions, internal/external
rotation, and anterior-posterior translation (AP motion), may be controlled by either force or
displacement control. The standard provides representative force and displacement profiles for
both the internal/external rotation or moment, as well as AP translation or force. In the present
experiment, axial force control, angular dorsiflexion/plantarflexion control, AP force control, and
internal/external moment control were applied in order to quantify the resulting AP
displacements and internal/external rotations of the talus. These resulting motions were
compared to the proposed ISO alternative displacement-controlled profiles.

136
As the ISO standard is meant for the testing of metal and plastic implants alone, the
profiles had to be scaled accordingly, to protect the cadaveric specimens. Therefore, the peak
loads were reduced to approximately 50% of the magnitudes specified in the standard. This
reduced the maximum axial load to 1,183 N, the maximum AP force to 129 N, and the maximum
moment about the tibial shaft (internal-external rotation) to 2.9 Nm. Since the dorsiflexion and
plantarflexion profile was applied in displacement control for both the load and displacement
profiles, these parameters remained unchanged.
4.6.2 Axial Force Application

Overall, the axial load curve matched the implemented force command. The small
variability observed can be attributed to the natural differences between specimens. Additionally,
the small deviations at the curves can also be attributed to a force control feedback system and
the slight gain adjustments needed, based on the natural properties of each individual cadaver.
Interestingly, the maximum load of 1,183N was reached for all specimens without any serious
detriment to the specimens. One specimen had the navicular separate from the head of the talus
during the initial cycle; however, this did not seem to affect the readings for the remaining three
cycles.
4.6.3 Flexion Application

The dorsiflexion and plantarflexion motions were captured through a motion tracking
system rather than the load cell due to this parameter being implemented in displacement control.
Since the flexion profile was controlled in displacement control, there was less of a phase lag
between the command and specimens. The flexion recorded created the same plantarflexion and

137
dorsiflexion peaks in time; however, the magnitudes were a bit lower than the command. This
may be due to the physical range of motion of these specimens.
One point of difference between the recorded flexion profile and applied command
profile occurred during 20% to 30% of gait. Again, it was important to consider the entire system
and the complete set of load parameters when trying to identify the source of these
dissimilarities. It has been reported that at lower loads, ankle stability is dependent on ligaments;
however, at higher loads such as physiological loads, the articular geometry is the primary
stabilizer [148, 157]. Specifically, at these higher loads that occur from mid-stance to push off,
the bones are locked into a forced position to create a solid rigid lever of support in order to
propel the subject forward [27]. It can be reasoned that due to the sharp increase in axial load
from 20% to 30% of gait along with the highest peak of AP load, the specimen was relying on
the tibiotalar joint for stabilization which was achieved through a brief locked forced position
which was illustrated by the plateau during this time segment. Moreover, it can be reinforced that
these specimens were rigid by analyzing the torque produced in the sagittal plane during this
period.

138

Figure 52. Resulting dorsiflexion/plantarflexion moment from all specimens in gray. A steep rise in the ankle moment from 20-
35% of the gait can be observed.
While this moment was not directly measured at the ankle, it provides a good indication that the
loads appeared to create an increase in moments which would only occur if the ankle was acting
as a rigid lever as described by Chan et al [27].
4.6.4 Applied Anterior-Posterior Load Profile

For the majority of the anterior-posterior (AP) load curve, the measured AP force
behaved similarly to the programmed command. While the AP command profile had a maximum
of 128.5N, the mean AP load curve was observed to reach a maximum 149.5N. The differences
in AP loads can again be attributed to the variance in specimen flexibility. While a spring was
used to convert a displacement actuator to implement force, there was no feedback to adjust the
amount of force the actuator actually applied. Therefore, the load observed was reliant on the
stiffness of the specimen and its resistance to the force produced by the actuator and spring. The
phase lag seen can also be attributed to the gain of the system along with the spring.

139
An apparent difference between the observed AP load and the command load is the
plateau produced between 45% to 50% of the gait cycle. This plateau can be reasoned by
analyzing the entire experimental setup. An isolated anterior force would cause the specimen to
translate forward and contact the spring restraints. These restraints would in turn create a natural
plantarflexion of the ankle.

Figure 53. The AP force on the superior aspect will translate the entire system. This will cause the spring restraint system on the
inferior aspect to resist the system and create a natural plantarflexion motion.

Conversely a posterior force would create a dorsiflexion moment. During the time period where
the plateau occurs, the applied anterior force is declining and transitioning to a posterior force
resulting in a dorsiflexion moment. However, at this segment of the gait cycle, the specimen is

140
moving from its highest dorsiflexion into the plantarflexion associated with pre-swing phase.
This dorsiflexion created from the posterior force is being counteracted by the plantarflexion
command creating an instance when the AP force remains constant.
4.6.5 Internal and External Torque Application

The mean of the recorded internal and external moments followed the general trend of the
programmed torque command. The wide variability in the stiffness of the specimens created non-
uniform curves throughout the 12 specimens; however, the mean followed the torque command.
Due to the variability, the instances when the peak external moments occurred, did not align for
all 12 specimens which gave a mean that appeared lower in magnitude. Therefore, the max
external torques were identified for each curve individually and gave a mean of 3.2 ± 0.4Nm.
This was greater than the maximum programmed 2.9Nm of external torque.
4.6.6 Measured AP Displacement Curve vs ISO AP Displacement Curve

The main purpose of this experiment was to analyze the resulting AP displacement curve
that resulted from the force commands previously discussed, to the ISO AP displacement curve.
At 50% of the prescribed loads, the AP displacement curve of the talus, did not resemble the AP
displacement curve described in ISO 22622. While the mean recorded AP displacement of the
talus did not match the displacement parameters in magnitude or shape, the resulting curve
makes physiological sense. In the beginning of the gait cycle during heel strike, the ankle
plantarflexed which created the anterior (negative) displacement measured. Subsequently, during
maximum weight acceptance, midstance, and push off, the talus dorsiflexed indicated posterior
(positive) translation which was also observed. Upon completion of these sections of the stance

141
phase, the axial load began to decrease, and the ankle began to plantarflex which can be seen on
both the axial command graph (40%) and dorsiflexion and plantarflexion command graph (45%).
Both of these instances indicated that the talus began to move towards its anterior aspect which
was what was observed on the AP displacement graph. These trends matched the displacements
shown for the natural talus during unconstrained testing, in the in vitro cadaver model previously
discussed.

142

Figure 54. Comparison of AP displacement trend from cadaver model to resulting AP displacement from ISO force protocol

While the trends do match, the discrepancy in magnitude may be due to the fact that the bones
distal to the navicular were removed. By removing these bones, the arch of the foot that helps

143
stabilize and transfer loads has been compromised. Without the cuneiform and metatarsals,
during loading when the talus was seated firmly against the concave aspect of the navicular,
there was no support for the navicular [27]. This may have biomechanically changed the loading
and created less load transfer resulting in lower magnitudes of displacement.
In contrast, the ISO AP displacement protocol proposed may be an oversimplification.
While it correctly calls for a posterior displacement of the tibia (plantarflexion of ankle) in the
beginning of the gait cycle which transitions to an anterior displacement of the tibia (dorsiflexion
of ankle), where these instances occur appear incorrect. Specifically, the synchronization of the
peak anterior displacement, appears to occur too late within the gait cycle at 60%. Peak anterior
displacement correlates with peak dorsiflexion; however, at 60% when the swing phase should
be entered, the ankle should be at max plantarflexion or have a negative displacement. On the
other hand, for the AP force command, the peak anterior force or dorsiflexion appears to occur
prematurely. This discrepancy in synchronization of all parameters (Flexion, AP Force, and AP
Displacement), may be attributed to the fact that all three were parameters were aggregated from
different studies, resulting in all three action be executed out of phase. While seemingly
inconsequential, the minute differences could greatly affect the cross-path motion seen at the
articulating surface. As discussed in the background section, it was observed in THR that even
minute increases in displacements on the micron level, can drastically affect the resulting wear.
In order to effectively and properly execute these wear tests, the discrepancies in phase must be
sorted out.
Upon further investigation, the official ISO 22622 protocol was published within recent
months. While the majority of the standard remained unchanged, the ISO AP displacement
profile was modified. The graph presented in the results (Figure 43) showed the measured

144
cadaveric specimen AP displacement data, with the revised AP displacement profile. The new
parameters appear to resolve many of the questions previously raised from the proposed ISO
22622 draft. The waveform of the recorded data is very similar to the published ISO profile and
still did not exceed the AP displacement suggested in the ISO 22622 standard. However, it must
be remembered that this experiment applied the ISO Force profile command at 50% of the
maximum magnitudes. Further research must be conducted to investigate if testing at the full
100% will change the magnitudes observed or not.
4.5.7 Measured Internal and External Angular Rotations

The resulting internal and external rotations vary widely. However, the trend of the mean
rotations was similar to the suggested rotations of ISO 22622. A positive rotation should create
an internal moment which was seen for both the motion tracking data along with the internal
displacement ISO command. Likewise, as the gait cycle progressed, the tibia externally rotated,
which was agreed upon between the recorded data and the ISO protocol. As this experiment only
used 50% of the prescribed torque and the magnitudes of recorded displacement are lower, at
100% the rotation magnitudes may match.
This data may suggest that for tibial rotation, load or displacement control may
effectively produce the same kinematics. Conversely, for AP displacement of the talus, the
question remains whether displacement control or load control provide equivalent kinematics. As
Sutton et al. has shown for knees, force control may be more suitable due to the high reliance on
ligament stability for the knee. Interestingly, the ankle is unique in that it is highly stabilized by
ligaments at low loads, but by bone articulating surfaces at higher loads. While force control did
produce physiologically correct displacement trends, the benefits of force control may not
outweigh its complexities. At low loads, when the ankle relies on ligament stability,

145
displacements may be small and well within the viscoelastic range of the ligaments.
Furthermore, at higher loads, the ankle will rely on its articular surfaces for stability and
effectively become a rigid lever. In this configuration, the amount of displacement produced by
the force parameters should presumably be similar to the displacement control parameters, as
there will be less reliance on ligaments for stability.

5. Conclusion

As the incidence of osteoarthritis has steadily increased, orthopaedic surgeons have
begun to adopt total ankle replacements (TAR) as an alternative to fusion. While TAR seemingly
eliminates pain and restores joint range of motion, an underlying problem with poor survivorship
plagues these implants. Although advancements have been made in design and materials in
recent decades, poor mid to long-term survivorship continue to hinder the universal adoption of
TARs. A contributing factor in this problem resides in the fact that researchers do not possess the
proper tools to accurately predict the functionality and survivorship of these implants in vivo.
This collection of projects aimed to help with the development of these tools, to eventually
improve the long-term survivorship of TARs.
A comprehensive literature search showed a paucity of cadaveric simulations researching
the kinematics of the tibiotalar joint. Of the few existing studies, there were two main methods of
simulation: dynamic and static. Dynamic simulations have been overly complex, rendering them
difficult to repeat and limiting the applicability for parametric studies, such as comparisons of
implant designs. On the other hand, static simulations have applied a fraction of physiological
loads with no clear consensus on loading conditions and magnitudes (200 N-750 N) [63].

146
Specific Aim 1 aimed to address the absence of an optimized cadaveric simulation that
can be used to study the tibiotalar joint. This test established a quasi-static model in four
instances of interest identified from the peaks of a typical ground reaction curve: heel strike,
maximum weight acceptance, mid-stance, and push off. It was concluded from this experiment
that it is vital to account for the angle of application of the ground reaction force relative to the
ankle and shank parameters. In this experiment, this was achieved with the implementation of an
unconstrained cyclic loading where the cadaveric specimen was allowed free translation within
the transverse plane. With the development of this cadaveric model in Specific Aim 1, the
kinematics of a natural tibiotalar ankle joint was established.
For Specific Aim 2, using the established model from Specific Aim 1, the kinematics of
an artificial TAR was recorded and analyzed. It was observed for this specific TAR design, that
the full span of AP displacement was inhibited during gait. This restriction can cause an uneven
distribution of stresses during gait and eventually lead to edge loading, high wear, and
mechanical damage of the polyethylene surface. This is highly likely to be relevant in many
other TAR designs as loosening either by mechanical degradation, poor fixation interfaces, or
reaction to poly wear debris over time is highly prevalent in TAR patients.
For Specific Aim 3, a group of 14 Infinity TAR explants were obtained and analyzed.
The damage modes and locations showed a non-uniform distribution of damage, consistent with
the behavior seen in Specific Aim 2. Specifically, a statistically higher amount of damage on the
posterior aspect which aligned with the uneven distribution of stress seen in the cadaveric model.
This localization of damage supports the theory that a non-uniform distribution of stresses occurs
in vivo, leading to possible edge loading and the eventual failure of this type of joint
replacement. Interestingly, with the provided patient information, it was known that the majority

147
of these replacements failed due to loosening as expected from the uneven distribution of stresses
caused by a lack of uniform damage. With these findings, the optimized cadaveric model
discussed in Specific Aim 1 was supported.
In addition to serving as a tool for preclinical testing of novel or modified TAR designs,
the cadaveric model can be used to evaluate the effects of surgeries that are often performed in
unison with TAR. One such common surgery is subtalar fusions. Following subtalar fusion, an
increase in AP displacements for both the natural and artificial TAR were observed. This could
possibly provide some insight on why patients with natural intact ankles which undergo subtalar
fusions, may suffer from accelerated degradation in adjacent joints.
Specific Aim 5 addressed the kinematics of the tibiotalar joint under force-controlled
wear study and compared it to the input parameters specified by the ISO displacement profile.
The resultant AP displacements under force-control were similar to the finalized ISO 22622
displacement profile in both magnitude and pattern. However, further studies are required to
determine whether the contact stress distribution is affected in such a way as to produce different
wear patterns in a force-controlled test.
In the present experiment, the measured AP displacements of the talus relative to the tibia
had a very different profile compared to the draft proposal ISO 22622. A revised published
version of the ISO 22622, included a modified AP displacement profile, more closely matching
the results presented. While the trends and magnitude matched, if the full 100% of the ISO force
profile is applied, the magnitudes may or may not exceed the ISO AP displacement profile.
Collectively, this project has contributed to the establishment of a physiologically
relevant testing methodology designed to accurately predict in vivo functionality. Additionally,

148
the proper approach on how to apply these parameters to predict longevity has been investigated.
By contributing to the development of tools which allow scientist to predict how a total ankle
replacement will work in the body and for how long, preclinical testing can be improved and in
turn result in excellent long-term success rates for patients with total ankle replacements.

149
Appendix
A. MATLAB Code

I. Rotation Transformation Code

% Flags have been placed in Tibia, Talus, Calcaneus
% X direction is ML, Y direction is Axial, Z direction is AP
% Positive Z is Posterior, Negative Z is Anterior
%
% Run this code when the Talus motion with respect to Tibia or Calcaneus
% motion wrt to Talus needs to be calculated.

% This code can be used to calculate the Cal file from raw 6D data file.
% Raw data file must be exported with the Rotational Matrix and will be
% found in the 6D.xls file.
%
% COLUMNS MAY CHANGE BASE ON FILE/EXPERIMENT
% Made by Nathan Ho
% Last edit: 12/17/2018
clear;
clc;
close all;

%Import data
[DataFile,DataPath] = uigetfile('*.csv','Select 6D File that you want to
analyze');
Datafilename = strcat(DataPath,DataFile);
SixDMotion = xlsread(Datafilename);

%Extract Tibia Rotation Matrix.
TibiaRxx=SixDMotion(:,15);
TibiaRxy=SixDMotion(:,16);
TibiaRxz=SixDMotion(:,17);
TibiaRyx=SixDMotion(:,18);
TibiaRyy=SixDMotion(:,19);
TibiaRyz=SixDMotion(:,20);
TibiaRzx=SixDMotion(:,21);
TibiaRzy=SixDMotion(:,22);
TibiaRzz=SixDMotion(:,23);

%Extract Talus Rotation Matrix.
TalusRxx=SixDMotion(:,41);
TalusRxy=SixDMotion(:,42);
TalusRxz=SixDMotion(:,43);
TalusRyx=SixDMotion(:,44);
TalusRyy=SixDMotion(:,45);
TalusRyz=SixDMotion(:,46);
TalusRzx=SixDMotion(:,47);
TalusRzy=SixDMotion(:,48);
TalusRzz=SixDMotion(:,49);

%Extract Calcaneus Rotation Matrix.
CalRxx=SixDMotion(:,54);

150
CalRxy=SixDMotion(:,55);
CalRxz=SixDMotion(:,56);
CalRyx=SixDMotion(:,57);
CalRyy=SixDMotion(:,58);
CalRyz=SixDMotion(:,59);
CalRzx=SixDMotion(:,60);
CalRzy=SixDMotion(:,61);
CalRzz=SixDMotion(:,62);

%Extract Tibia and Talus Displacement Matrix [X, Y, Z]
TibiaDisp=[SixDMotion(:,24:26)];
TalusDisp=[SixDMotion(:,50:52)];
CalDisp=[SixDMotion(:,63:65)];

%%%%%%%% Calculate Euler Angle Rotation About X axis %%%%%%%%%
TibiaRotx = atan2(TibiaRzy,TibiaRzz);
TalusRotx = atan2(TalusRzy,TalusRzz);
CalRotx = atan2(CalRzy, CalRzz);

%%%%%%%% Calculate Euler Angle Rotation About Y axis %%%%%%%%%
TibiaRoty = atan2(-TibiaRzx,(TibiaRzy.^2+TibiaRzz.^2));
TalusRoty = atan2(-TalusRzx,(TalusRzy.^2+TalusRzz.^2));
CalRoty = atan2(-CalRzx,(CalRzy.^2+CalRzz.^2));

%%%%%%%% Calculate Euler Angle Rotation About Y axis %%%%%%%%%
TibiaRotz = atan2(TibiaRyx, TibiaRxx);
TalusRotz = atan2(TalusRyx, TalusRxx);
CalRotz = atan2(CalRyx, CalRxx);

%Compiling into one Rotation matrix for Tibia and one for Talus [Rx,Ry,Rz]
TibiaRot = [TibiaRotx, TibiaRoty, TibiaRotz];
TalusRot = [TalusRotx, TalusRoty, TalusRotz];
CalRot = [CalRotx, CalRoty, CalRotz];

%Calculate Talus Diplacement wrt to Tibia reference frame
for i = 1:length(TibiaRxx)
R = [TibiaRxx(i), TibiaRxy(i), TibiaRxz(i);...
TibiaRyx(i), TibiaRyy(i), TibiaRyz(i);...
TibiaRzx(i), TibiaRzy(i), TibiaRzz(i)];
u = TalusDisp(i,:)-TibiaDisp(i,:);
Tal2TibDisp(i,:) = R.'*u.';
end

%Calculate Talus Rotation wrt to Tibia reference frame
for i = 1:length(TibiaRxx)
R = [TibiaRxx(i), TibiaRxy(i), TibiaRxz(i);...
TibiaRyx(i), TibiaRyy(i), TibiaRyz(i);...
TibiaRzx(i), TibiaRzy(i), TibiaRzz(i)];
u = TalusRot(i,:)-TibiaRot(i,:);
Tal2TibRot(i,:) = R.'*u.';
end

%Calculate Calcaneus Rotation wrt to Talus reference frame (Inver/Ever)
for i = 1:length(TibiaRxx)
R = [TalusRxx(i), TalusRxy(i), TalusRxz(i);...

151
TalusRyx(i), TalusRyy(i), TalusRyz(i);...
TalusRzx(i), TalusRzy(i), TalusRzz(i)];
u = CalRot(i,:)-TalusRot(i,:);
Cal2TalRot(i,:) = R.'*u.';
end

%Convert from Radians to Degrees
Tal2TibRot = Tal2TibRot*180/pi;
Cal2TalRot = Cal2TalRot*180/pi;

CalcOutput = [Cal2TalRot, Tal2TibRot, Tal2TibDisp];
Output = strcat(DataPath,'CalcOutput.xls');
xlswrite(Output,CalcOutput); %Outputs to same folder as original data.

II. Cyclic Loading Analysis

%%% Code to analyze Talus motion in the AP direction %%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Flags have been placed in Tibia, Talus, Calcaneus
% X direction is ML, Y direction is Axial, Z direction is AP
% Positive Z is Posterior, Negative Z is Anterior
%
% This program will take import in the CalcOutput File that is created in
% Transform_Angles.m
%
% Made by Nathan Ho
% Last edit: 12/17/2018

clear;
clc;
close all;

%Import Data for all 4 positions

%%%%%%%%%%%%%%%%%%%%%%Heel Strike%%%%%%%%%%%%%%%%%%%%%
[HSDataFile,HSDataPath] = uigetfile('*.xls','Select Heel Strike CalcOutput
File');
HSDatafilename = strcat(HSDataPath,HSDataFile);
HSTalusMotion = xlsread(HSDatafilename);
AP_HS=HSTalusMotion(:,9); %9th column is Z Component Displacement
(Anterior/Posterior)
ML_HS=HSTalusMotion(:,7); %7th column is X Component Displacement
(Medial/Lateral)
Int_HS=HSTalusMotion(:,5); %5th column is Rotation about Y (Internal/External
Rotation
Inv_HS=HSTalusMotion(:,3); %3rd column is Rotation about Z of Calc wrt to Tib
(Inversion/Eversion)

AP_HS=AP_HS(~isnan(AP_HS)); %Identifies which rows are NaN and deletes
ML_HS=ML_HS(~isnan(ML_HS));
Int_HS=Int_HS(~isnan(Int_HS));
Inv_HS=Inv_HS(~isnan(Inv_HS));

152
%%%%%%%%%%%%%%%%%%%%%Maximum Weight Acceptance%%%%%%%%%%%%%%%%%%%%%
[MWADataFile,MWADataPath] = uigetfile('*.xls','Select Maximum Weight
Acceptance CalcOutPut File');
MWADatafilename = strcat(MWADataPath,MWADataFile);
MWATalusMotion = xlsread(MWADatafilename);

AP_MWA=MWATalusMotion(:,9); %9th column is Z Component Displacement
(Anterior/Posterior)
ML_MWA=MWATalusMotion(:,7); %7th column is X Component Displacement
(Medial/Lateral)
Int_MWA=MWATalusMotion(:,5); %5th column is Rotation about Y
(Internal/External Rotation
Inv_MWA=MWATalusMotion(:,3); %3rd column is Rotation about Z of Calc wrt to
Tib (Inversion/Eversion)

AP_MWA=AP_MWA(~isnan(AP_MWA)); %Identifies which rows are NaN and deletes
ML_MWA=ML_MWA(~isnan(ML_MWA));
Int_MWA=Int_MWA(~isnan(Int_MWA));
Inv_MWA=Inv_MWA(~isnan(Inv_MWA));

%%%%%%%%%%%%%%%%%%%%%%Midstance%%%%%%%%%%%%%%%%%%%%%
[MSDataFile,MSDataPath] = uigetfile('*.xls','Select Midstance Calc Data
File');
MSDatafilename = strcat(MSDataPath,MSDataFile);
MSTalusMotion = xlsread(MSDatafilename);

AP_MS=MSTalusMotion(:,9); %9th column is Z Component Displacement
(Anterior/Posterior)
ML_MS=MSTalusMotion(:,7); %7th column is X Component Displacement
(Medial/Lateral)
Int_MS=MSTalusMotion(:,5); %5th column is Rotation about Y (Internal/External
Rotation
Inv_MS=MSTalusMotion(:,3); %3rd column is Rotation about Z of Calc wrt to Tib
(Inversion/Eversion)

AP_MS=AP_MS(~isnan(AP_MS)); %Identifies which rows are NaN and deletes
ML_MS=ML_MS(~isnan(ML_MS));
Int_MS=Int_MS(~isnan(Int_MS));
Inv_MS=Inv_MS(~isnan(Inv_MS));

%%%%%%%%%%%%%%%%%%%%%%Pushoff%%%%%%%%%%%%%%%%%%%%%
[PODataFile,PODataPath] = uigetfile('*.xls','Select Push Off Calc Data
File');
PODatafilename = strcat(PODataPath,PODataFile);
POTalusMotion = xlsread(PODatafilename);

AP_PO=POTalusMotion(:,9); %9th column is Z Component Displacement
(Anterior/Posterior)
ML_PO=POTalusMotion(:,7); %7th column is X Component Displacement
(Medial/Lateral)
Int_PO=POTalusMotion(:,5); %5th column is Rotation about Y (Internal/External
Rotation
Inv_PO=POTalusMotion(:,3); %3rd column is Rotation about Z of Calc wrt to Tib
(Inversion/Eversion)

AP_PO=AP_PO(~isnan(AP_PO)); %Identifies which rows are NaN and deletes
ML_PO=ML_PO(~isnan(ML_PO));

153
Int_PO=Int_PO(~isnan(Int_PO));
Inv_PO=Inv_PO(~isnan(Inv_PO));

%Filter and massage data/graphs.
%Sampling Frequency = 100 Hz for MTS
Fs = 100;
%Cutoff Frequency chosen at 0.2 since sin wave was applied at 0.2 Hz
Fc = 0.2;

%Filter data to minimize noise
[b,a] = butter(5,Fc/(Fs/2));

%Filter each signal using Buttersworth filter made above
AP_HS_Filt = filtfilt(b, a, AP_HS);
AP_MWA_Filt = filtfilt(b, a, AP_MWA);
AP_MS_Filt = filtfilt(b, a, AP_MS);
AP_PO_Filt = filtfilt(b, a, AP_PO);

ML_HS_Filt = filtfilt(b, a, ML_HS);
ML_MWA_Filt = filtfilt(b, a, ML_MWA);
ML_MS_Filt = filtfilt(b, a, ML_MS);
ML_PO_Filt = filtfilt(b, a, ML_PO);

Int_HS_Filt = filtfilt(b, a, Int_HS);
Int_MWA_Filt = filtfilt(b, a, Int_MWA);
Int_MS_Filt = filtfilt(b, a, Int_MS);
Int_PO_Filt = filtfilt(b, a, Int_PO);

Inv_HS_Filt = filtfilt(b, a, Inv_HS);
Inv_MWA_Filt = filtfilt(b, a, Inv_MWA);
Inv_MS_Filt = filtfilt(b, a, Inv_MS);
Inv_PO_Filt = filtfilt(b, a, Inv_PO);

%%%%%%%%%%%%%%Figure out linear slopes of general trendline%%%%%%%%%%

%Combine data into cells, so for loops can be used. Cells are used rather
%than a matrix, since cells allow you to combine different sized vectors.
%They are denoted with '{}' signs.

HS_Data = {AP_HS_Filt, ML_HS_Filt, Int_HS_Filt, Inv_HS_Filt};
MWA_Data = {AP_MWA_Filt, ML_MWA_Filt, Int_MWA_Filt, Inv_MWA_Filt};
MS_Data = {AP_MS_Filt, ML_MS_Filt, Int_MS_Filt, Inv_MS_Filt};
PO_Data = {AP_PO_Filt, ML_PO_Filt, Int_PO_Filt, Inv_PO_Filt};

%Go through each column and use a linear polyfit and take the slope which
%is stored as c(1)
for i = 1:4
x = [1:length(HS_Data{1,i})]';
c = polyfit(x, HS_Data{1,i},1);
if c(1) > 0
HS_signs(i) = 1;
else
HS_signs(i) = -1;

154
end
end

for i = 1:4
x = [1:length(MWA_Data{1,i})]';
c = polyfit(x, MWA_Data{1,i},1);
if c(1) > 0
MWA_signs(i) = 1;
else
MWA_signs(i) = -1;
end
end

for i = 1:4
x = [1:length(MS_Data{1,i})]';
c = polyfit(x, MS_Data{1,i},1);
if c(1) > 0
MS_signs(i) = 1;
else
MS_signs(i) = -1;
end
end

for i = 1:4
x = [1:length(PO_Data{1,i})]';
c = polyfit(x, PO_Data{1,i},1);
if c(1) > 0
PO_signs(i) = 1;
else
PO_signs(i) = -1;
end
end
%%%%%%%%%%%%%%%%Graph AP plots on one subplot%%%%%%%%%%%%%%%%%%
%%%%%%% Use these plots to do a quick check that there is no noise or
%%%%%%% extraneous signals at the end of the trial that could result from
%%%%%%% not turning off the motion tracker after loading has finished.

figure
subplot(1,4,1);
plot(AP_HS_Filt);
title('AP Talus Motion during Heel Strike');
ylim([mean(AP_HS_Filt)-1 mean(AP_HS_Filt)+1])

subplot(1,4,2);
plot(AP_MWA_Filt);
title('AP Talus Motion during Maximum Weight Acceptance');
ylim([mean(AP_MWA_Filt)-1 mean(AP_MWA_Filt)+1])

subplot(1,4,3);
plot(AP_MS_Filt);
title('AP Talus Motion during MidStance');
ylim([mean(AP_MS_Filt)-1 mean(AP_MS_Filt)+1])

155
subplot(1,4,4);
plot(AP_PO_Filt);
title('AP Talus Motion during Push Off');
ylim([mean(AP_PO_Filt)-1 mean(AP_PO_Filt)+1])

%%%%%%%%%%%%%%%Plot ML plots on one subplot%%%%%%%%%%%%%%%%%
figure
subplot(1,4,1);
plot(ML_HS_Filt);
title('ML Talus Motion during Heel Strike');
ylim([mean(ML_HS_Filt)-1 mean(ML_HS_Filt)+1])

subplot(1,4,2);
plot(ML_MWA_Filt);
title('ML Talus Motion during Maximum Weight Acceptance');
ylim([mean(ML_MWA_Filt)-1 mean(ML_MWA_Filt)+1])

subplot(1,4,3);
plot(ML_MS_Filt);
title('ML Talus Motion during MidStance');
ylim([mean(ML_MS_Filt)-1 mean(ML_MS_Filt)+1])

subplot(1,4,4);
plot(ML_PO_Filt);
title('ML Talus Motion during Push Off');
ylim([mean(ML_PO_Filt)-1 mean(ML_PO_Filt)+1])

%%%%%%%%%%%%%%%Plot Int plots on one subplot%%%%%%%%%%%%%%%%%
figure
subplot(1,4,1);
plot(Int_HS_Filt);
title('Int Talus Motion during Heel Strike');
ylim([mean(Int_HS_Filt)-1 mean(Int_HS_Filt)+1])

subplot(1,4,2);
plot(Int_MWA_Filt);
title('Int Talus Motion during Maximum Weight Acceptance');
ylim([mean(Int_MWA_Filt)-1 mean(Int_MWA_Filt)+1])

subplot(1,4,3);
plot(Int_MS_Filt);
title('Int Talus Motion during MidStance');
ylim([mean(Int_MS_Filt)-1 mean(Int_MS_Filt)+1])

subplot(1,4,4);
plot(Int_PO_Filt);
title('Int Talus Motion during Push Off');
ylim([mean(Int_PO_Filt)-1 mean(Int_PO_Filt)+1])

156
%%%%%%%%%%%%%%%Plot Inv plots on one subplot%%%%%%%%%%%%%%%%%
figure
subplot(1,4,1);
plot(Inv_HS_Filt);
title('Inv Talus Motion during Heel Strike');
ylim([mean(Inv_HS_Filt)-1 mean(Inv_HS_Filt)+1])

subplot(1,4,2);
plot(Inv_MWA_Filt);
title('Inv Talus Motion during Maximum Weight Acceptance');
ylim([mean(Inv_MWA_Filt)-1 mean(Inv_MWA_Filt)+1])

subplot(1,4,3);
plot(Inv_MS_Filt);
title('Inv Talus Motion during MidStance');
ylim([mean(Inv_MS_Filt)-1 mean(Inv_MS_Filt)+1])

subplot(1,4,4);
plot(Inv_PO_Filt);
title('Inv Talus Motion during Push Off');
ylim([mean(Inv_PO_Filt)-1 mean(Inv_PO_Filt)+1])

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%Calculate AP Displacement from 150N to 300N
APdispHS=HS_signs(1)*(max(AP_HS_Filt)-min(AP_HS_Filt));
APdispMWA=MWA_signs(1)*(max(AP_MWA_Filt)-min(AP_MWA_Filt));
APdispMS=MS_signs(1)*(max(AP_MS_Filt)-min(AP_MS_Filt));
APdispPO=PO_signs(1)*(max(AP_PO_Filt)-min(AP_PO_Filt));

%Calculate ML Displacement from 150N to 300N
MLdispHS=HS_signs(2)*(max(ML_HS_Filt)-min(ML_HS_Filt));
MLdispMWA=MWA_signs(2)*(max(ML_MWA_Filt)-min(ML_MWA_Filt));
MLdispMS=MS_signs(2)*(max(ML_MS_Filt)-min(ML_MS_Filt));
MLdispPO=PO_signs(2)*(max(ML_PO_Filt)-min(ML_PO_Filt));

%Calculate Internal/External Rotation from 150N to 300N
IntdispHS=HS_signs(3)*(max(Int_HS_Filt)-min(Int_HS_Filt));
IntdispMWA=MWA_signs(3)*(max(Int_MWA_Filt)-min(Int_MWA_Filt));
IntdispMS=MS_signs(3)*(max(Int_MS_Filt)-min(Int_MS_Filt));
IntdispPO=PO_signs(3)*(max(Int_PO_Filt)-min(Int_PO_Filt));

%Calculate Inversion/Eversion Rotation from 150N to 300N
InvdispHS=HS_signs(4)*(max(Inv_HS_Filt)-min(Inv_HS_Filt));
InvdispMWA=MWA_signs(4)*(max(Inv_MWA_Filt)-min(Inv_MWA_Filt));
InvdispMS=MS_signs(4)*(max(Inv_MS_Filt)-min(Inv_MS_Filt));
InvdispPO=PO_signs(4)*(max(Inv_PO_Filt)-min(Inv_PO_Filt));

%%
AP_Displacement = table(APdispHS, APdispMWA, APdispMS, APdispPO)
ML_Displacement = table(MLdispHS, MLdispMWA, MLdispMS, MLdispPO)
Int_Rotation = table(IntdispHS, IntdispMWA, IntdispMS, IntdispPO)
Inv_Rotation = table(InvdispHS, InvdispMWA, InvdispMS, InvdispPO)

157
%Extra added on 1/5/2020 to get 3 top peaks for graphs for Defense.
% Top3_AP_HS=HS_signs(1)*(maxk(findpeaks(HS_signs(1)*AP_HS_Filt),3)-
min(HS_signs(1)*AP_HS_Filt));
% Top3_AP_MWA=MWA_signs(1)*(maxk(findpeaks(MWA_signs(1)*AP_MWA_Filt),3)-
min(MWA_signs(1)*AP_MWA_Filt));
% Top3_AP_MS=MS_signs(1)*(maxk(findpeaks(MS_signs(1)*AP_MS_Filt),3)-
min(MS_signs(1)*AP_MS_Filt));
% Top3_AP_PO=PO_signs(1)*(maxk(findpeaks(PO_signs(1)*AP_PO_Filt),3)-
min(PO_signs(1)*AP_PO_Filt));
%
% table(Top3_AP_HS, Top3_AP_MWA, Top3_AP_MS, Top3_AP_PO)

III. ISO Force-Displacement Analysis

%% Code to analyze Talus motion in the AP direction %%%

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Flags have been placed in Tibia, Talus, Calcaneus
% X direction is ML, Y direction is Axial, Z direction is AP
% Positive Z is Posterior, Negative Z is Anterior
%
% This program will take import in the CalcOutput File that is created in
% Transform_Angles.m
%
% Made by Nathan Ho
% Last edit: 1/14/2020

clear;
clc;
close all;

%Import Cal File from Optotrak & MTS File

[MTDataFile,MTDataPath] = uigetfile('*.xls','Select Motion Tracker _cal
File');
MTDatafilename = strcat(MTDataPath,MTDataFile);
MTTalusMotion = xlsread(MTDatafilename);

MTTime=MTTalusMotion(:,1); %Time from MT clock
AP=MTTalusMotion(:,7); %7th column is Z Component Displacement
(Anterior/Posterior)
ML=MTTalusMotion(:,5); %5th column is X Component Displacement
(Medial/Lateral)
Rot=MTTalusMotion(:,3); %3rd column is Rotation about Y (Internal/External
Rotation
Inver=MTTalusMotion(:,10); %3rd column is Rotation about Z of Calc wrt to Tib
(Inversion/Eversion)

%Finds NaNs (missing data) and takes median of surrounding 10 points and
%fills.
AP = fillmissing(AP,'movmedian',10);
ML = fillmissing(ML,'movmedian',10);
Rot = fillmissing(Rot,'movmedian',10);
Inver = fillmissing(Inver,'movmedian',10);

158

%Filter and massage data/graphs.
%Sampling Frequency = 100 Hz for MTS
Fs = 100;
%Cutoff Frequency chosen at 0.2 since sin wave was applied at 0.2 Hz
Fc = 0.2;

%Filter data to minimize noise
[b,a] = butter(5,Fc/(Fs/2));

%Filter each signal using Buttersworth filter made above
AP_Filt = filtfilt(b, a, AP);
ML_Filt = filtfilt(b, a, ML);
Rot_Filt = filtfilt(b, a, Rot);
Inver_Filt = filtfilt(b, a, Inver);

%Center plots to zero
AP_Filt = AP_Filt- mean(AP_Filt(1:1000));
Rot_Filt = Rot_Filt- mean(Rot_Filt(1:1000));
Rot_Filt = Rot_Filt*180/pi();

%Find starting point of testing after MTS goes to initial positions. Look
%from the beginning to about 20 seconds in. Since motion tracker is
%obtaining data at 100Hz, end index is 5000.
[APpks,APlocs]=findpeaks(AP_Filt(1:3000));
[Rotpks,Rotlocs]=findpeaks(Rot_Filt(1:3000));

%Find Max index of initial starting peak to begin plots at.
[APM,API] = max(APpks);
APstart = APlocs(API);

[RotM,RotI] = max(Rotpks);
Rotstart = Rotlocs(RotI);

%Separate each cycle. In order to sync everything, choosing the index of
%the first peak 'max(locs)'. We know that the cycle takes 100s to complete
%so 10000 data points should be recorded for each cycle.
APCycle1 = AP_Filt(APstart:APstart+9999);
APCycle2 = AP_Filt(APstart+10000:APstart+19999);
APCycle3 = AP_Filt(APstart+20000:APstart+29999);
APCycle4 = AP_Filt(APstart+30000:APstart+39999);

RotCycle1 = Rot_Filt(Rotstart:Rotstart+9999);
RotCycle2 = Rot_Filt(Rotstart+10000:Rotstart+19999);
RotCycle3 = Rot_Filt(Rotstart+20000:Rotstart+29999);
RotCycle4 = Rot_Filt(Rotstart+30000:Rotstart+39999);

APCycleComb = [APCycle2, APCycle3, APCycle4];
RotCycleComb = [RotCycle2, RotCycle3, RotCycle4];

APAvg_Cycle = mean(APCycleComb,2);
RotAvg_Cycle = mean(RotCycleComb,2);

159

%Display Max AP and Rotations
APMax = max(APAvg_Cycle)- min(APAvg_Cycle)
RotMax = max(RotAvg_Cycle)- min(RotAvg_Cycle)

%%
%Plot and store data in Specimen Data file

%Plot all four cycles
time = (0:.01:99.99);
figure; plot(MTTime,AP_Filt);
title('AP Displacement During Full Test'), xlabel('Time (s)'),
ylabel('Displacement (mm)');
saveas(figure(1), [MTDataPath 'AP_Filt.jpg'])

figure; plot(MTTime, Rot_Filt);
title('Rotation During Full Test'), xlabel('Time (s)'), ylabel('Rotation
(Degrees)');
saveas(figure(2), [MTDataPath 'Rot_Filt.jpg'])

%Plot last 3 cycles in gray and average of all 3 in red
figure;
plot(time, APCycle2, 'Color', [0.75,0.75,0.75]); hold on
plot(time, APCycle3, 'Color', [0.75,0.75,0.75]);
plot(time, APCycle4, 'Color', [0.75,0.75,0.75]);
plot(time, APAvg_Cycle, 'r');
title('Average AP Displacement'), xlabel('Time (s)'), ylabel('Displacement
(mm)');
saveas(figure(3), [MTDataPath 'Avg_AP.jpg'])

time = (0:.01:99.99);
figure;
plot(time, RotCycle2, 'Color', [0.75,0.75,0.75]); hold on
plot(time, RotCycle3, 'Color', [0.75,0.75,0.75]);
plot(time, RotCycle4, 'Color', [0.75,0.75,0.75]);
plot(time, RotAvg_Cycle, 'r');
title('Average Rot Displacement'), xlabel('Time (s)'), ylabel('Rotation
(Degrees)');
saveas(figure(4), [MTDataPath 'Avg_Rot.jpg'])

%%
%Import in the 6D file to get DF/PF angles of Talus
clc
clear
close all

[TalusDataFile,TalusDataPath] = uigetfile('*.xls','Select Motion Tracker 6D
File');
TalusDatafilename = strcat(TalusDataPath,TalusDataFile);
Talus = xlsread(TalusDatafilename);

dfpf = Talus(:,25);
dfpf = dfpf*180/pi();
dfpf = dfpf - mean(dfpf(1:1000));

160

Fs = 100;
%Cutoff Frequency chosen at 0.2 since sin wave was applied at 0.2 Hz
Fc = 0.2;

%Filter data to minimize noise
[b,a] = butter(5,Fc/(Fs/2));

[dfpfpks,dfpflocs]=findpeaks(dfpf(1:3000));
dfpf=dfpf(~isnan(dfpf));
df_Filt = filtfilt(b, a, dfpf);

%Find Max index of initial starting peak to begin plots at.
[dfM,dfI] = max(dfpfpks);
dfstart = dfpflocs(dfI);

dfCycle1 = df_Filt(dfstart:dfstart+9999);
dfCycle2 = df_Filt(dfstart+10000:dfstart+19999);
dfCycle3 = df_Filt(dfstart+20000:dfstart+29999);
dfCycle4 = df_Filt(dfstart+30000:dfstart+39999);

dfCycleComb = [dfCycle2, dfCycle3, dfCycle4];
dfAvg_Cycle = mean(dfCycleComb,2);

time = (0:.01:99.99);
figure;
plot(time, dfCycle2, 'Color', [0.75,0.75,0.75]); hold on
plot(time, dfCycle3, 'Color', [0.75,0.75,0.75]);
plot(time, dfCycle4, 'Color', [0.75,0.75,0.75]);
plot(time, dfAvg_Cycle, 'r');
title('Average Rot Displacement'), xlabel('Time (s)'), ylabel('Rotation
(Degrees)');

IV. ISO Force-Displacement Average Plots

%Code to plot the average of all 3 trials of the 12 specimens and the
%average of ALL trials on a single plot

%This program will take a excel sheet with the averages of all 12 specimens
%that has been calculated previously from FvD_Data_Analysis.m
%
%
% Made by Nathan Ho
% Last edit: 1/14/2020
clear;
clc;
close all;

%Import datafile with averages of all 3 cycles for each specimen.

[MTDataFile,MTDataPath] = uigetfile('*.xls','Select Motion Tracker _cal
File');
MTDatafilename = strcat(MTDataPath,MTDataFile);
CombinedSpecimenAP = xlsread(MTDatafilename);

161

%Plot all specimens in gray and have average in red.
%For Motion tracker data: time = (0:.01:99.99)';
%For MTS Data: time = (0:.009958:99.9900042)';
time = (0:.01:99.99)';
%time = (0:.009958:99.9900042)';

% %Only do this for Command Plots
% for i = 1:1:12
% CombinedSpecimenAP(:,i) = CombinedSpecimenAP(:,i) -
mean(CombinedSpecimenAP(1:20,i));
% end

%Plot all columns(all specimens)
figure;
hold on
for i = 1:1:12

plot(time, -CombinedSpecimenAP(:,i),'Color',[0.75, 0.75, 0.75]);

end
AvgAPALL = mean(-CombinedSpecimenAP, 2);
plot(time, AvgAPALL,'r');

%Store max and min rotation of rotation averages
rotationlim=zeros([12 2]);
for i =1:1:12
rotationlim(i,:) = [min(CombinedSpecimenAP(:,i)),
max(CombinedSpecimenAP(:,i))];
end

%Get readings for all specimens at 0%, 10%, 20%, 30%....100%
for i = 1:1:9
%For MT: j = i*1000+1;
%For MTS: j = i*0.1*length(CombinedSpecimenAP);
%j = i*1000+1;
j = round(i*0.1*length(CombinedSpecimenAP));
range(i,:) = [CombinedSpecimenAP(j,:)];
end
%Since we want the range of specimen data by instance in gait we need to
%transpose the matrix.
range = range';

%%
%Run this section if you want to produce an Error bar overlay
%Since errorbar cmd needs an error input for each index, we need to include
%a "zero" error for the other other instances we aren't making error bars

neg = zeros(10042,1);
pos = zeros(10042,1);
%Fill in positive and negative error for 10%, 20%, 30%, ... 100%
for i = 1:1:9
%j = i*1000+1;
j = round(i*0.1*length(CombinedSpecimenAP));

162
neg(j)= min(range(:,i))-mean(range(:,i)); %Need to subtract out the mean
since the error adds directly to the mean
pos(j)= max(range(:,i))-mean(range(:,i));
end
hold on; errorbar(time,AvgAPALL,neg, pos);

%%
%Run this section if you want to produce a Boxplot overlay
boxplot(range,
'positions',[10,20,30,40,50,60,70,80,90],'colors','k','PlotStyle','compact','
Jitter',0);

%For some reason, the boxplot code relabels the X label, so need to reset
%this.
set(gca,'XTickLabel', {' '})
delete(findall(gcf, 'type', 'text'));
set(gca, 'xticklabelmode', 'auto', 'xtickmode', 'auto');
ylim([-12 8])
xlim([0 100])

163
B. Schematic of ISO Force-Displacement Apparatus

164
C. Biomechanical Comparison of Fixation Stability Using a Lisfranc Plate vs
Transarticular Screws

165

166

167

168

169

170

171
D. The Objective Measurement of Brace-Use Adherence in the Treatment of Idiopathic
Clubfoot

172

173

174

175

176

177

178

179
E. Biomechanical Comparison of Fixation Devices for First Metatarsocuneiform Joint
Arthrodesis

180

181

182

183

184

185

186
F. Preclinical Biomechanical Testing Models for the Tibiotalar Joint and Its
Replacements: A Systematic Review

187

188

189

190

191
G. Systematic Review of Unsystematic Total Ankle Replacement Wear Evaluations

192

193

194

195

196

197

198

199

200

201

202
H. Damage Patterns in Polyethylene Fixed Bearings of Retrieved Total Ankle
Replacements (Accepted)

DAMAGE PATTERNS IN POLYETHYLENE FIXED BEARINGS OF RETRIEVED
TOTAL ANKLE REPLACEMENTS
NATHAN C. HO, M.S.
A
, SANG-HYUN PARK, PH.D.
A
, PATRICIA CAMPBELL, PH.D.
A
, DOUGLASS W. VAN
CITTERS, PH.D.
B
, EDWARD EBRAMZADEH, PH.D.
A
, SOPHIA SANGIORGIO, PH.D.
A

AFFILIATIONS:
A. The J. Vernon Luck, Sr., M.D. Orthopaedic Research Center
Orthopaedic Institute for Children 403 W. Adams Blvd.
Los Angeles, CA 90007
Nathan Ho, M.S., Graduate Student, Technician Email: nho618@gmail.com
Edward Ebramzadeh, Ph.D., Director of J. Vernon Luck, Sr., M.D. Orthopaedic Research Center Phone:
(213) 742-1378
Email: Edward.Ebramzadeh@ucla.edu
ssangiorgio@mednet.ucla.edu
B. Thayer School of Engineering at Dartmouth
14 Engineering Drive Hanover, NH 03755
Douglas W. Van Citters, Ph.D. Phone: (603) 646-6406
CORRESPONDING AUTHOR INFORMATION:
Edward Ebramzadeh Ph.D. Orthopaedic Institute for Children
403 W. Adams Blvd.
Los Angeles, CA 90007
Phone: (213) 742-1378
Email: Edward.Ebramzadeh@ucla.edu
Declaration of Interest: The Authors have nothing to disclose and have no conflicts of interests. IRB
approval was granted in the obtaining of the retrieval specimens.
Key Points
• • • •
Semi-quantitative grading of burnishing, scratching, mechanical damage, pitting, and particles Correlate
the observed damage patterns with the in vivo kinematics of the ankle

203
Results indicate greater posterior aspect damage when compared to the anterior aspect
May indicate that the anterior to posterior motion of the talus throughout gait is being constrained

Abstract

INTRODUCTION: Poor long-term outcomes continue to hinder the universal adoption of total ankle
replacements (TAR) for end stage arthritis. Implants retrieved after clinical service may provide valuable
information on the in-vivo interaction between the bearing surfaces. In the present study, polyethylene
inserts of TARs retrieved at revision surgery were analyzed for burnishing, scratching, mechanical
damage, pitting, and embedded particles. The damages were studied to assess motion and interactions
between the tibial and talar components in vivo.
METHODS: Fourteen retrieved polyethylene inserts from a fixed bearing total ankle replacement design
currently in clinical use were analyzed. Duration of time in vivo was between 11.5 months and 120.1
months. Three investigators independently graded each articular surface in quadrants for five features of
damage: burnishing, scratching, mechanical damage, pitting, and embedded particles.
RESULTS: No correlation was found for burnishing between the anterior and posterior aspects (p =
0.51); however, scratching and pitting were significantly higher on the posterior aspect compared to the
anterior aspect (p < 0.02). There was a high correlation between burnishing and in vivo duration of the
implant (anterior: R=0.67, p=0.01, posterior: R=0.68, p=0.01).
Conclusion: The higher concentration of posterior damage on these polyethylene inserts suggested that
the design restricted the talus from displacing its full anterior-posterior span during gait. Additionally,
surgeon implantation technique could contribute to a restricted articulation of the implant. The resulting
higher stresses in the posterior articular surfaces may have been a factor in loosening and other modes of
failure of the retrieved implants. The results may be useful in guiding next generation designs.
Keywords: Retrieval, Polyethylene Damage, Total Ankle Replacement

204

Introduction
Over the last few decades, there have been many incremental advances in the wear resistance and
fixation of total ankle replacements (TAR), including improvements in material, instrumentation, and
implant design. These modifications have helped more accurately reproduce in vivo kinematics
(dorsiflexion-plantarflexion, internal-external rotation, inversion-eversion, and anterior-posterior
displacement), increase longevity, and reduce malalignment. Collectively, these changes have contributed
to an improved short-to-midterm (five to ten years) outcome. However, poor long-term outcome
continues to hinder universal adoption of total ankle replacements for end-stage arthritis [7, 13, 15].
Failures have been attributed to aseptic loosening, bone fracture, polyethylene fracture, and wear [5, 12,
27].
Due to the complexity of ankle kinematics and difficulty of directly observing the talus, pre-
clinical, and forensic studies are pertinent to the advancement of total ankle replacements. Human gait
studies have been performed to study total ankle replacement kinematics. A significant limitation is the
use of skin mounted markers to record the kinematics of the talus and other bones not directly adjacent to
the skin. To address this, some studies have used intracortical bone screws to rigidly attach motion
trackers to the talus percutaneously. Unfortunately, this method is highly invasive and carries the risk of
discomfort, infection, or other clinical complications [2, 21, 22]. Other studies have used fluoroscopy or
dual fluoroscopy; however, these techniques may have low sampling rates, narrow fields of capture and
high rates of radiographic exposure [11, 26, 30]. Cadaver studies have been explored to provide important
information regarding ankle kinematics; however, there is no established ASTM or ISO testing protocol,
and this has resulted in a high variability of testing methods and kinematic results [16]. Collectively, these
studies contribute to the overall understanding of total ankle replacement kinematics; however, retrieval
studies have the potential to play a vital role in further understanding the functionality of total ankle

205
replacements in vivo. Ultimately, retrieval studies are also necessary to validate the findings obtained
from in vitro joint wear simulation studies.
The purpose of the present study was to analyze a group of 14 TAR retrieved polyethylene inserts
to determine patterns of mechanical damage. We assessed the inserts for magnitude and location of
burnishing, scratching, mechanical damage, pitting, and embedding of particles, with the goal of
correlating clinical damage with in vivo kinematics. Based on the natural kinematics of an ankle, we
hypothesized that the damage would be uniform along the anterior-posterior (AP) axis of the articulating
surfaces.
Material and Methods
The study was based on characterization and quantification of damage and location of
polyethylene inserts of TARs retrieved during revision surgery. The study was conducted at (Blinded for
Review). Fourteen retrieved polyethylene inserts designed by Wright Medical Technologies (Memphis,
TN) were available for the present study. Twelve were InBone® prosthesis and two were Infinity®
prosthesis inserts. Both designs included fixed bearing polyethylene inserts that were sterilized with
ethylene oxide (EtO). Duration of time in vivo was between 11.5 months and 120.1 months (median 46.7
months). Reasons for explantation were reported as: loosening in six, subsidence in two, reaction to metal
debris in one, periprosthetic fracture in one, pain in one, and other reasons in three. These implants were
previously analyzed by Currier et al, who investigated the effects of mobile versus fixed bearing implant
design, and methods of polyethylene sterilization on clinical damage, wear, and oxidation [10].
Each articulating surface was analyzed using a Keyence Microscope (VHX-2000, Chicago IL).
The bearing surface area of each polyethylene insert was divided into four quadrants (Figure 1). Within
each quadrant, five features of damage were rated: burnishing, scratching, mechanical damage, pitting,
and embedded particles, defined as follows:
(vi) Burnishing: Damage that has “polished” the surface and is identified with a shiny hue and
absence of original machining marks.
(vii) Scratching: Damage that is characterized with a rougher matte surface and scratches.

206
(viii) Mechanical Damage: Markings with deep gouges and grooves or deformation. This excluded
damage caused during implantation and removal.
(ix) Pitting: Small divots or circular indentations, most likely attributed to 3
rd
body particle.
(x) Embedded Particles: 3
rd
body particles that are embedded within the articulating surface.

Figure 55. Diagram of quadrants used during grading. In each quadrant, the damage scores ranged from 0 to 4. For the purpose
of comparing anatomical locations, anterior quadrants 1 and 2 were summed for a potential score of 0 to 8. Likewise, posterior
quadrants 3 and 4 were summed for a potential score of 0 to 8.
A semi-quantitative grading scale was implemented and ranged from zero to four, with zero
indicating no apparent damage and four indicating a high degree of damage. This method was derived
from the Hood scoring protocol [17] and has been well established for modern hip and knee joint
replacement components [9, 24].
0: Area is free of respective damage
1: 0 < 25% of area is covered by respective damage
2: 25% < 50% of area is covered by respective damage
3: 50% < 75% of area is covered by respective damage
4: 75% ≤ 100% of area is covered by respective damage

207
Images of all 14 retrievals were captured using a Keyence Microscope. These images of the
articulating surfaces highlighted five features of damage: burnishing, scratching, mechanical damage,
pitting, and embedded particles (Figure 2). An example of each type of damage is displayed in the
following figure.

Figure 56. Burnishing: Damage that has "polished" the surface and is identified with a shiny hue and absence of original
machining marks. ii) Scratching: Damage that is characterized with a rougher matte surface and scratches, identified as fine
lines iii) Mechanical Damage: Markings with deep gouges and grooves or deformation. This excluded damage caused during
implantation and removal. iv) Pitting: Small divots or circular indentations (localized pits), most likely attributed to 3
rd
body
particle. v) Embedded Particles: 3
rd
body particles that are embedded within the articulating surface.
Each specimen was graded in quadrants to provide a precise location of damage as well as its
extent. Afterwards, statistical analysis was performed to compare different anatomical locations.
Specifically, the following quadrants were combined: quadrant 1 & 3, medial or lateral (determined by
left or right implantation side), quadrant 2 & 4, medial or lateral (determined by left or right implantation
side), quadrant 1 & 2, anterior, and quadrant 3 & 4, posterior.
For the analysis of interobserver variability, three investigators independently scored each
polyethylene insert using the method outlined above. Interobserver reliability was assessed by comparing
the average of the absolute differences in scores along with the minimum, maximum, and mode of the
absolute differences between each pair of observers. Boxplots were used to display the differences in
measurements of damage by the three observers. Interobserver error analysis was calculated separately for
anterior burnishing and scratching (summing the two anterior quadrants, quadrants 1 and 2), as well as
posterior burnishing and scratching (summing the two posterior quadrants 3 and 4). Wilcoxon signed
ranked test was used to compare the absolute errors in measurement among the investigators.
All data analysis was performed using SPSS Statistics Version 17.0 (IBM, Houston Texas). A
Wilcoxon signed-rank test (nonparametric, paired) was run to compare the anterior, quadrants 1 and 2,

208
against posterior, quadrants 3 and 4, for average scores of all damage modes. An additional Wilcoxon
signed-rank test was run to compare the medial, quadrants 1 and 3, and lateral, quadrants 2 and 4, aspects
for average burnishing and scratching damage. Finally, to assess the correlation between duration in vivo
and damage, Spearman correlation coefficient (nonparametric) was calculated.
Results
Comparing results among the three observers, no statistically significant differences between any
two observers were found for burnishing or scratching in anterior or posterior aspects (p > 0.25) (Figure
3). Specifically, the median value of the absolute error between any two observers was 3 or lower, out of
a maximum error of 8 (anterior summed quadrants 1 and 2, posterior summed quadrants 3 and 4).

Figure 57. The top, middle, and bottom horizontal lines of each box represent the 75th percentile, the 50th percentile (or
median), and 25th percentile of the absolute value of interobserver error. The whiskers extend to 1.5 times the height of the box,

209
encapsulating 95% of the data if normally distributed. The median value of the absolute error between any two observers was 3
or lower, out of a maximum error of 8. That is, the anterior scores were the sum of quadrant 1 and quadrant 2, each with a
score 1 to 4, for a maximum total of 8. Likewise, the posterior scores were the sum of quadrant 3 and 4.
The scores of the different types of damage observed in each quadrant are presented in a table
(Table 1), as the averages among the three observers. Since each damage mode was analyzed with respect
to anatomical areas (anterior vs posterior), the summed damage scores are presented in a separate table
(Table 2). No major differences were found between the medial and lateral aspects for both burnishing (p
= 0.96) and scratching (p = 0.33) damage, and none of the differences were statistically significant. The
same test failed to establish significance for burnishing between the anterior and posterior aspects (p =
0.51). However, the incidence of scratching was significantly higher on the posterior aspect compared to
the anterior aspect (p = 0.01) of the polyethylene inserts. Additionally, the incidence of pitting was
significantly higher on the posterior aspect compared to the anterior aspect (p = 0.02).
Table 1. Average score and standard deviation of all specimens in this cohort from all three
investigators.
Quadrant 1 Quadrant 2 Quadrant 3 Quadrant 4
Burnishing 2.3 (1.5) 2.3 (1.4) 2.2 (1.5) 2.3 (1.5)
Scratching 2.1 (1.2) 2.1 (1.1) 2.3 (1.2) 2.3 (1.2)
Mechanical
Damage
(Gouges, cuts)
0.9 (0.7) 0.6 (0.5) 0.8 (0.9) 0.7 (0.6)
Pitting 1.4 (1.1) 1.3 (1.0) 1.5 (1.0) 1.65 (1.0)
Embedded Particles 0.4 (0.6) 0. 5 (0.7) 0.3 (0. 6) 0.4 (0.5)

Table 2. Average score, standard deviation, and p-value for the anterior (sum of quadrant 1 and 2) and
posterior (sum of quadrant 3 and 4) aspects.

Anterior
(Q1 + Q2)
Posterior
(Q3 + Q4)
P – value
Burnishing 4.6 (2.4) 4.5 (2.4) 0.51

210
Scratching 4.1 (1.8) 4.6 (1.9) 0.01
Mechanical Damage
(Gouges, cuts)
1.5 (0.6) 1.5 (0.9) 0.93
Pitting 2.7 (1.8) 3.1 (1.7) 0.02
Embedded Particles 0.8 (1.1) 0.7 (1.0) 0.93

There was a high correlation between the in vivo duration of the implants and burnishing on both
the anterior and posterior aspects of the polyethylene inserts (p = 0.01). In contrast, there was a negative
correlation between the in vivo duration and scratching on both the anterior and posterior aspects;
however, these correlations were not statistically significant (Table 3).
Table 3. Correlation analysis between in vivo duration and each of the five types of damage.
Damage Type Anterior Posterior
Burnishing 0.67 (p=0.01) 0.68 (p=0.01)
Scratching -0.32 (p=0.27) -0.40 (p=0.16)
Mechanical
Damage
0.44 (p=0.11) -0.01 (p=0.97)
Pitting -0.78 (p=0.001) -0.59 (p=0.03)
Embedded
Particles
-0.33 (p=0.25) -0.59 (p=0.03)

Discussion

Major modes of failure of total ankle replacements have been identified as loosening,
polyethylene wear, malalignment, and abnormal kinematics [6, 10, 14, 18]. These failures may be closely
related to non-uniform distribution of contact stresses. Previous retrieval studies have examined the
oxidation of polyethylene inserts, wear particles produced from polyethylene, polyethylene fracture case
studies, and damage modes [1, 3, 10, 29]. The purpose of the present study was to build upon the
previous knowledge base by analyzing the locations of each type of damage against articulation of the
polyethylene inserts. Based on the natural kinematics of an ankle, we hypothesized that the damage would

211
be uniformly distributed along the anterior-posterior (AP) axis of the articulating surfaces. Only fixed
bearing implants were selected to reveal the worst-case scenario of damage by non-uniform contact
stresses, since mobile bearing inserts hypothetically minimize such damage at the risk of having two
articular surfaces, both of which may produce wear debris. Furthermore, the implants discussed in the
present study were selected to have a consistent sterilization method in order to provide comparable
results.
In all quadrants, burnishing and scratching were the most prevalent damage modes. There was
also a presence of light mechanical damage, pitting, and a minute amount of embedded metal particles.
The majority of the specimens in this study failed due to loosening. This multifactorial failure mode
resulting from surgeon, patient, or implant factors may be attributed to wear, device design, or
biomechanics of implantation, all of which may be directly related to a non-uniform distribution of
contact stresses. This was consistent with the observed differences between anatomical areas of these
polyethylene inserts. While burnishing was not statistically significant between the anterior and posterior
aspects, there was a high correlation between burnishing of both the anterior and posterior aspects against
the in vivo duration of the implant (anterior: correlation coefficient 0.67, p = 0.01, posterior: correlation
coefficient 0.68, p = 0.01). Conversely, scratching did not produce a correlation with the in vivo duration
of the implant (anterior: correlation coefficient -0.32, p = 0.27, posterior: correlation coefficient -0.40, p =
0.16). These findings indicated that the amount of burnishing damage increased as a function of follow-up
time, consistent with analysis of damage in total hip replacements [23] and total knee replacements [4].
The negative correlation between scratching damage and in vivo time, although not statistically
significant, was consistent with high degrees of polishing of polyethylene in hip and knee replacements
once sufficient conformity between the metal and polyethylene has been established.
It has been established that during gait as the ankle transitions from plantarflexion to dorsiflexion
the talus should move anteriorly to posteriorly [8, 19, 20] (Figure 4). From these kinematics, the
polyethylene insert would be expected to experience uniform damage throughout its articulating surface.

212
However, a higher incidence of scratching was observed on the posterior aspect of the polyethylene
inserts. This may indicate that, while total ankle replacements have been shown to restore the range of
motion during gait, the anterior to posterior length that the talus must span throughout gait is being
constrained [25, 28] (Figure 4). While it is possible for the observed damage on the posterior aspect to
have been caused by third body particles generated during the surgical approach, the appearance of the
damage was consistent with scratching and burnishing rather than mechanical damage.

Figure 58. Illustration showing expected talus displacement in natural ankle from Heel Strike to Push Off.
This non-uniform distribution may be a contributing factor in the high frequency of loosening and
other failure mechanisms associated with total ankle replacements. In a wear study of TAR, Affatato et al.
found a similar trend of a high incidence of localized linear wear [1]. However, they found the wear to be
highest on the anterior aspect of their specimens. This may be due to a different TAR design used in their
study. Specifically, the implant used in Affatato et al.’s study was a mobile bearing design which has
additional degrees of freedom to articulate.
The present study had limitations. Specifically, the performance of these implants was not solely
contingent on implant design, but also on proper alignment of both components, and each patient’s
ligamentous laxity and overall ankle stability. However, due to the limited availability of clinical data or
radiographs, we were unable to analyze these additional factors. Moreover, the sample size was small,
and the specimens were obtained from multiple surgeons and hospitals. Although some correlations

213
emerged that could possibly have been related to the kinematics of the talar component, the clinical or
biomechanical relevance of these correlations are unknown. Specifically, statistically significant
differences were found in scratching and pitting between the anterior and posterior aspects of the
retrievals; however, the magnitude of the differences in scores were relatively small. Finally, all of the
damage scores were semi-quantitative, introducing inherent uncertainty into the ratings.
Conclusion
The present study of retrieved polyethylene inserts indicated that, for this particular non-
cemented fixed bearing design, constraints for anterior to posterior translation of the talus may have
contributed to the increased localized posterior damage observed. This localization of damage would
support the theory that a non-uniform distribution of stresses could occur in vivo, leading to possible edge
loading and the eventual failure often observed with this type of joint replacement.

214
References
1. Affatato S, Taddei P, Leardini A, Giannini S, Spinelli M, Viceconti M. Wear behaviour in total
ankle replacement: a comparison between an in vitro simulation and retrieved prostheses. Clin
Biomech (Bristol, Avon). 2009;24:661-669.
2. Ardnt AW, P.; Liu, A.; Nester, C.; Stacoff, A.; Jones, R.; Lundgren, P.; Lundberg, A. Intrinsic
Foot Kinematics Measured In Vivo During the Stance Phase of Slow Running. Journal of
Biomechanics. 2007;40:7.
3. Assal M, Al-Shaikh R, Reiber BH, Hansen ST. Fracture of the polyethylene component in an
ankle arthroplasty: a case report. Foot Ankle Int. 2003;24:901-903.
4. Atwood SA, Currier JH, Mayor MB, Collier JP, Van Citters DW, Kennedy FE. Clinical wear
measurement on low contact stress rotating platform knee bearings. J Arthroplasty. 2008;23:431-
440.
5. Berkowitz MJ, Clare MP, Walling AK, Sanders R. Salvage of failed total ankle arthroplasty with
fusion using structural allograft and internal fixation. Foot Ankle Int. 2011;32:S493-502.
6. Clough T, Bodo K, Majeed H, Davenport J, Karski M. Survivorship and long-term outcome of a
consecutive series of 200 Scandinavian Total Ankle Replacement (STAR) implants. Bone Joint J.
2019;101-B:47-54.
7. Cody EA, Bejarano-Pineda L, Lachman JR, Taylor MA, Gausden EB, DeOrio JK, Easley ME,
Nunley JA, 2nd. Risk Factors for Failure of Total Ankle Arthroplasty With a Minimum Five
Years of Follow-up. Foot Ankle Int. 2019;40:249-258.
8. Conti S, Lalonde KA, Martin R. Kinematic analysis of the agility total ankle during gait. Foot
Ankle Int. 2006;27:980-984.
9. Currier BH, Currier JH, Franklin KJ, Mayor MB, Reinitz SD, Van Citters DW. Comparison of
Wear and Oxidation in Retrieved Conventional and Highly Cross-Linked UHMWPE Tibial
Inserts. J Arthroplasty. 2015;30:2349-2353.
10. Currier BH, Hecht PJ, Nunley JA, 2nd, Mayor MB, Currier JH, Van Citters DW. Analysis of
Failed Ankle Arthroplasty Components. Foot Ankle Int. 2019;40:131-138.
11. de Asla RJ, Wan L, Rubash HE, Li G. Six DOF in vivo kinematics of the ankle joint complex:
Application of a combined dual-orthogonal fluoroscopic and magnetic resonance imaging
technique. J Orthop Res. 2006;24:1019-1027.
12. Deleu PA, Devos Bevernage B, Maldague P, Gombault V, Leemrijse T. Arthrodesis After Failed
Total Ankle Replacement. Foot Ankle Int. 2014;35:549-557.
13. Frigg A, Germann U, Huber M, Horisberger M. Survival of the Scandinavian total ankle
replacement (STAR): results of ten to nineteen years follow-up. Int Orthop. 2017;41:2075-2082.
14. Glazebrook MA, Arsenault K, Dunbar M. Evidence-based classification of complications in total
ankle arthroplasty. Foot Ankle Int. 2009;30:945-949.
15. Gougoulias N, Khanna A, Maffulli N. How successful are current ankle replacements?: a
systematic review of the literature. Clin Orthop Relat Res. 2010;468:199-208.
16. Ho NC, Ebramzadeh E, Sangiorgio SN. Preclinical biomechanical testing models for the
tibiotalar joint and its replacements: A systematic review. Foot Ankle Surg. 2018.
17. Hood RW, Wright TM, Burstein AH. Retrieval analysis of total knee prostheses: a method and its
application to 48 total condylar prostheses. J Biomed Mater Res. 1983;17:829-842.
18. Krause FG, Windolf M, Bora B, Penner MJ, Wing KJ, Younger AS. Impact of complications in
total ankle replacement and ankle arthrodesis analyzed with a validated outcome measurement. J
Bone Joint Surg Am. 2011;93:830-839.
19. Leardini A, O'Connor JJ, Giannini S. Biomechanics of the natural, arthritic, and replaced human
ankle joint. J Foot Ankle Res. 2014;7:8.

215
20. List R, Foresti M, Gerber H, Goldhahn J, Rippstein P, Stussi E. Three-dimensional kinematics of
an unconstrained ankle arthroplasty: a preliminary in vivo videofluoroscopic feasibility study.
Foot Ankle Int. 2012;33:883-892.
21. Lundberg A, Goldie I, Kalin B, Selvik G. Kinematics of the ankle/foot complex: plantarflexion
and dorsiflexion. Foot Ankle. 1989;9:194-200.
22. Lundgren P, Nester C, Liu A, Arndt A, Jones R, Stacoff A, Wolf P, Lundberg A. Invasive in vivo
measurement of rear-, mid- and forefoot motion during walking. Gait Posture. 2008;28:93-100.
23. McKellop HA, Campbell P, Park SH, Schmalzried TP, Grigoris P, Amstutz HC, Sarmiento A.
The origin of submicron polyethylene wear debris in total hip arthroplasty. Clin Orthop Relat
Res. 1995:3-20.
24. Park SH, Lu Z, Hastings RS, Campbell PA, Ebramzadeh E. Five Hundred Fifty-five Retrieved
Metal-on-metal Hip Replacements of a Single Design Show a Wide Range of Wear, Surface
Features, and Histopathologic Reactions. Clin Orthop Relat Res. 2018;476:261-278.
25. Queen RM, Franck CT, Schmitt D, Adams SB. Are There Differences in Gait Mechanics in
Patients With A Fixed Versus Mobile Bearing Total Ankle Arthroplasty? A Randomized Trial.
Clin Orthop Relat Res. 2017;475:2599-2606.
26. Roach KE, Wang B, Kapron AL, Fiorentino NM, Saltzman CL, Bo Foreman K, Anderson AE. In
Vivo Kinematics of the Tibiotalar and Subtalar Joints in Asymptomatic Subjects: A High-Speed
Dual Fluoroscopy Study. J Biomech Eng. 2016;138.
27. Smyth A, Fisher J, Suner S, Brockett C. Influence of kinematics on the wear of a total ankle
replacement. J Biomech. 2017;53:105-110.
28. Valderrabano V, Nigg BM, von Tscharner V, Stefanyshyn DJ, Goepfert B, Hintermann B. Gait
analysis in ankle osteoarthritis and total ankle replacement. Clin Biomech (Bristol, Avon).
2007;22:894-904.
29. Vaupel Z, Baker EA, Baker KC, Kurdziel MD, Fortin PT. Analysis of retrieved agility total ankle
arthroplasty systems. Foot Ankle Int. 2009;30:815-823.
30. Wang B, Roach KE, Kapron AL, Fiorentino NM, Saltzman CL, Singer M, Anderson AE.
Accuracy and feasibility of high-speed dual fluoroscopy and model-based tracking to measure in
vivo ankle arthrokinematics. Gait Posture. 2015;41:888-893.

216
I. Increasing Loads and Diminishing Returns: A Biomechanical Study of Direct Vertebral
Rotation

217

218

219

220

221

222

223

224
J. Is Load Control Necessary to Produce AP Displacement and Axial Rotation in
Wear Testing of TAR?

IS LOAD CONTROL NECESSARY TO PRODUCE AP DISPLACEMENT
AND AXIAL ROTATION IN WEAR TESTING OF TAR?

NATHAN C. HO, PH.D., COLIN MCCARTY, B.S., SANG-HYUN PARK, PH.D., JOAN R. WILLIAMS. M.D.,
NEDA FARBOD M.D. , EDWARD EBRAMZADEH PH.D., SOPHIA SANGIORGIO PH.D.

CONTACT INFORMATION
Orthopaedic Institute for Children
403 W. Adams Blvd.
Los Angeles, CA 90007

EEbramzadeh@mednet.ucla.edu
ssangiorgio@mednet.ucla.edu

PHONE: (213) 742-1440
FAX : (213) 742-1365

225
Abstract
Background
A recent standard, published by the International Standard Organization, specifies two
options for joint wear simulator evaluation of total ankle replacements (TAR): load-controlled
testing and displacement-controlled testing. The purpose of the present study was to apply the
load-controlled testing parameters to cadaveric specimens and to quantify and compare the
observed sagittal translations and axial rotations to those specified under the displacement-
controlled option.
Methods
Twelve cadaveric specimens were stripped of extraneous tissues, while keeping the
surrounding ankle ligaments intact. To protect the cadaveric specimens from structural failure,
all loads were applied at 50% of the specified load profile while plantarflexion-dorsiflexion
rotation was applied as specified. A halo was used to produce plantarflexion and dorsiflexion of
the talus through a set of screws, while a baseplate resisted axial loads. The axial force and axial
torque were applied to the proximal end of the tibia and fibula under force and torque feedback
control respectively. An anterior-posterior force was applied to the proximal end of the tibia.
Plantarflexion and dorsiflexion were effectively applied using rotation control.
Results
There was variation among the specimens in the magnitudes of anterior-posterior
displacements with peaks ranging from XXX to XXX. Likewise, there was variation among the
specimens in magnitude for axial rotation, with peaks ranging from XXX to XXX. However, the
mean magnitudes of AP displacement and axial rotation did not exceed those specified by ISO
22622.
Conclusion

226
Testing under load-controlled conditions has been advocated to produce physiological AP
displacements and axial rotations of the tibiotalar joint. However, in the present study, load control of a
cadaveric ankle did not produce substantially different magnitudes of AP displacements and axial
rotations than those specified by the displacement-controlled option defined in ISO 22622.
Introduction
Over the past few decades, total joint arthroplasty has revolutionized treatment of end-
stage arthritis. While highly successful in hip and knee joints, only short-term outcome has been
successful in ankle joints, with 90% survival at 5 years dropping to 65% at midterm to long-term
survival
1-5
. Evaluation of wear using in vitro joint simulators has been established as a necessary
pre-clinical test to predict the in vivo clinical performance of total joint replacements. A thorough
understanding of the kinematics of the joint is a prerequisite to design a physiologically accurate
test.
Previous studies of polyethylene wear of total hip arthroplasty acetabular components
have found the type of motion between articulating surfaces has a substantial effect on the
magnitude and type of wear damage. Specifically, higher degrees of crossing path motion lead to
higher magnitudes of polyethylene wear in metal on polyethylene bearing surfaces
6; 7
. Therefore,
it is important for the evaluation of polyethylene wear in total ankle replacements to produce
physiologically accurate relative motions between the tibial and talar components. Typically,
total ankle replacement wear simulators have controlled axial loading, plantarflexion and
dorsiflexion, anterior and posterior (AP) displacement, and axial rotation. Since wear simulators
for total ankle replacements are not widely available, investigators have used total knee
replacement (TKR) wear simulators for studying total ankle replacement wear since the degrees
of freedom are analogous for axial load, flexion extension, AP displacement, and axial rotation
8
.

227
Traditionally, for both TKRs and TARs, AP displacement was applied under
displacement control and axial rotation was applied under rotation control (angular rotation).
More recently, an AP force profile has been applied, allowing the components to translate based
on the constraints of the implant as well as simulated physiological constraints. Similarly, an
axial torque profile has been applied, allowing the components to rotate based on the constraints
of the implant and simulated physiological constraints
9; 10
.
Recently, an ISO standard for TAR wear evaluation was published (ISO 22622)
11
. This
standard is similar to ISO 14243-3
12
for TKR wear evaluation in that both standards allow a
displacement-controlled method or a force-controlled method. Both methods specify the same
force profile for axial load to be applied to the tibial component, and angular rotation profile for
dorsiflexion-plantarflexion to be applied to the talar component. However, the displacement-
controlled method specifies a displacement profile for AP displacement, and angular rotation
profiles for internal-external rotation, both to be applied to the tibial component. In contrast, the
force-controlled method specifies a force profile for anterior-posterior loads and a torque profile
for internal-external torque, both to be applied to the tibial component.
For TKRs, Sutton et al. applied the loads resulting linear and angular displacements, the
reported tibial internal-external rotations were more than three times larger than the prescribed
displacement-controlled rotations. Additionally, the resulting AP transitions produced an inverse
waveform from the prescribed ISO AP displacement profile
13
.
Due to the lack of validation studies, it is unclear whether the load control parameters
specified by ISO 22622 for TARs are appropriate. The purpose of the present study was to apply
the load control parameters, specified by ISO 22622, to cadaveric ankles with intact ligaments,

228
and to compare the observed motions at the tibiotalar joint to motions specified by the same ISO
standard under the displacement-controlled option.
Methods
Specimen Preparation
Six pairs of fresh-frozen cadaveric lower limbs cut midshaft of the shank were obtained
from a licensed willed body tissue institute, Science Care (Science Care, Phoenix, AZ). The
specimens had all extraneous fat and tissues removed, while keeping all ligaments around the
ankle complex intact. Care was taken to minimize the disruption to the interosseus ligament
between the retained tibia and fibula. The Achilles tendon was removed to allow direct access to
the talus. Additionally, the hard and soft tissues distal to the navicular were removed, leaving
only the tibia, fibula, talus, calcaneus, navicular, and their surrounding connective ligaments. A
single 5cm long, 4mm diameter screw was used to fix the proximal end of the fibula to the tibia
specified under the load-controlled protocol of ISO 14243-1 for TKR to cadaveric knees, with
collateral and cruciate ligaments intact. They observed a wide discrepancy between the
displacements resulting from the published ISO force-controlled protocol, to the published
displacement-controlled parameters
13
. Specifically, when the investigators tested the specimens
under force-controlled conditions and measured the
to protect the interosseus ligament. Two 80mm long, 7mm diameter screws were placed
through the talus from the posterior surface, past the neck, and into the head, without piercing
through the anterior surface. These screws extended from the most posterior edge of the
calcaneus. To ensure proper screw placement, fluoroscopy was used to guide the angle of
approach along with the depth of insertion.

229

Figure XXX. Left: Dissected specimen with screws inserted from the posterior aspect. Right: Radiograph confirming screw
placement through talus.
Potting
The proximal ends of the tibia and fibula were then potted in low temperature bonding
resin (Bondo 3M, Maplewood, MN). Bone cement was poured into a square shaped brace which
surrounded the entire specimen and captured the heads of the posterior screws that were
previously inserted into the talus. This created a “halo” of bone cement which allowed
controlling the position of the talus using the posterior end of the screws. The neutral position
was determined using an alignment similar to the method described by Leardini et al.
14
.
Radiographs were taken to confirm the neutral alignment prior to the creation of the halo.

230

Figure 59. Specimen with bonding cement on the proximal end and a bone cement halo encapsulating the posterior screws.
Apparatus Design
With the halo attached, each specimen was loaded using a custom apparatus mounted into
an 8DOF load frame (858 Bionix, MTS Systems, Eden Prairie, MN). The proximal end of the
specimen, potted into cement, was rigidly attached to the axial-torsional actuator, on the superior
aspect of the load frame. The distal end of the specimen (calcaneus) rested on a baseplate, 3D
printed from tough PLA filament (Tough PLA 202300, Ultimaker, Utrecht, Netherlands). The
halo was mounted into a tray which was in turn attached to the baseplate using four hardened
steel bars which allowed free vertical translation of the halo relative to the base plate, while
restricting rotation and horizontal translations. On each pillar, two round flange linear ball
bearings (Model 6483K52, McMaster-Carr, Santa Fe Springs, CA) ensured low-friction vertical
gliding of the tray. The halo was thus used to produce plantarflexion and dorsiflexion of the talus
through the screws, while the baseplate resisted axial loads. To allow plantarflexion and
dorsiflexion to occur at the natural axis of rotation of the talus, the baseplate and its attachments
were allowed free AP translation by a low-friction linear bearing mounted on the load frame.

231
The axial force and axial torque were applied to the proximal end of the tibia and fibula
under force and torque feedback control respectively (Figure XXX). Following the ISO 22622
standard, an AP force was applied to the tibia; specifically, the force was applied to the proximal
end of the specimen. Plantarflexion and dorsiflexion were effectively applied using rotation
control (in degrees).

Figure 60. Load frame setup showing the applied loads: (A) Axial Load, (B) Torsional Load, (C) AP Load, (D) Linear
displacement translated to (E) Dorsiflexion-Plantarflexion
Loading Conditions
Due to the physical limitations of cadavers and to prevent soft tissue damage, the axial
force, AP force and tibial torque magnitudes were reduced to 50% of the respective magnitudes
specified by ISO 22622. These were all implemented at a frequency of 1 Hz ± 0.1Hz. A
summary of all of the forces applied in the present study is shown below (Figure XXX).

232

Figure 61. Comprehensive force profile that was applied to the cadavers.
Data Acquisition
An Optotrack Certus motion tracking system (Northern Digital, Inc., Waterloo, ON,
Canada) was used to record the motions of the tibia, talus, and calcaneus. The motion capture
system has an accuracy of 0.1mm, and a resolution of 0.01mm. Motion flags, equipped with 3
non-collinear light-emitting diode (LED) markers, pulse at 3000 Hz. A minimum of 3 LED
markers are required to obtain rotations (Rx, Ry, Rz) along with translations (x, y, z), while a
minimum of 1 marker is needed for pure translations. These motion flags were attached via bone
screws to the flag holders placed in the talus, tibia, and calcaneus during the specimen
preparation phase. An additional flag was mounted on the load frame to establish a fixed local
coordinate axis and to minimize the effects of any vibrations of the hydraulic machine.
An ATI 6 DOF load cell (Mini 58, ATI Industrial Automation, Apex, NC) was used to
capture all force data including, axial force, AP force, and axial torque. The motion tracking
system was used to capture all actual translations and rotations. All the measured forces recorded
were plotted and analyzed against the applied command signals. Similarly, since the dorsiflexion
-20
0
20
40
60
80
100
-200
0
200
400
600
800
1000
1200
1400
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
Angle (°) and Moment (Nm)
Force (N) - Axial and AP
% Gait Cycle
Applied Force Protocol
Axial Force (N) AP Force (N) Plantar/Dorsiflex-ion angle (°) I/E Moment (Nm)

233
or plantarflexion was not directly controlled, it was also important to check the measured applied
angular displacement. The resulting AP displacement from the applied loads was compared to
those specified in the ISO displacement control standard. Similarly, the resulting tibial rotations
from the applied axial torque was compared to those specified in the ISO displacement standard.
Results
Divergence of Actual Loads and Motions from Applied Command Profiles
The actual applied loads and motions were compared to the respective programmed load
or motion profile based on ISO 22622. The graphs presented below overlay (1) the recorded
loads for each of the 12 specimens in gray, (2) the mean of all 12 specimens in red along with
maximum and minimum bars at 10% increments of gait, and (3) programmed load or motion
profile in blue crosses.

234

235

Under axial load, the specimens responded to the load command profile without
substantial deviation and without structrual failure (Figure XXXA). Specifically, the specimens
reached the peak load at 38%, of the gait cycle, consistent with the axial load command profile.
In contrast, the specimens did not follow the dorsiflexion-plantarflexion command profile
closely. Specifically, while the dorsiflexion command increased steadily between 10% to 50% of
the gait cycle, the actual dorsiflexion of the specimens increased later, after 30% of the gait cycle
(Figure XXXB). Similarly, the specimens responded to the AP load command profile with some
deviation. Specifically, the specimens peaked slightly earlier than the command profile and
reached the valley later than the command profile (Figure XXXC). Finally, the specimens
responded to the axial torque command profile with some deviation. Specifically, the specimens
demonstrated somewhat of a phase delay with respect to the command profile (Figure XXXD).
Observed Outcome: Motions of the Talus
A wide range of AP displacement of the talus was observed among the specimens. While
on average, the displacements were within approximately a millimeter of the specified ISO

236
profile, the peaks and valleys were as much as 2mm apart among the specimens (Figure XXXA).
Among the specimens, the talus moved from 2.8mm posteriorly to 3.2mm anteriorly.

A wide range of axial rotation of the talus was also observed among the specimens. On
average, the peak and valleys of the rotation were lower than those specified by the ISO profile
(Figure XXXB). However, the peaks and valleys of some of the individual speimens exceeded
the ISO profile (Figure XXXB). Overall, among the specimens the observed rotations ranged
from -11 to 4.1 degrees.

237

Discussion
Few studies have been conducted to evaluate the wear of total ankle replacements (TAR)
using joint simulators. Among these studies, a wide range of methods have been used. In
particular, there is no consensus on the appropriate magnitude of AP displacement or axial
rotation of the talus
8
. Moreover, only two studies validated the results by comparing the
simulator-worn implants to retrieved TARs
15; 16
. Despite the lack of consensus, the International
Standard Organization has recently finalized a standard (ISO 22622), specifying displacement
and loading parameters for wear-testing of TARs
11
.
ISO 22622 provides two options for wear tests: displacement control and load control
11
.
The intention of the load control test is to allow the specimen to traverse through AP
displacement and axial rotation in response to the applied AP force and axial torque,
respectively. However, the load control option requires implementation of a constraint system to
mimic ligament restraints, since the implant alone does not provide sufficient resistance to

238
horizontal forces. Proponents of load control wear testing have argued that relative motion of
implant components occur more physiologically under load control and resulting contact stresses
between the components may also be affected.
Observed AP Displacement Under Load Control vs ISO 22622 AP Displacement
We compared the observed motions of the ankle, that resulted from applying the load
control option in ISO, to the AP displacement and axial rotation profiles specified by the
displacement control option of the ISO. At 50% of the specified loads, the observed AP
displacement curve of the talus was similar to the published ISO AP displacement profile.
Moreover, the resulting displacement magnitudes did not exceed the AP displacements specified
in the ISO 22622 standard. The resulting curve followed the expected physiological movements
and is consistent with a typical gait cycle. Specifically, at the initiation of the gait cycle, when
heel-strike occurred (0%-10%), the ankle was plantarflexed which created the measured anterior
(negative) displacement. Subsequently, during 10%-45%, the ankle transitioned from maximum
weight acceptance to push off which was correlated with high dorsiflexion. This dorsiflexion was
represented by a posterior (positive) translation which was also observed. Following this section,
the ankle moved into heel off and the pre-swing phase which was characterized by ankle
plantarflexion. This would suggest anterior (negative) translation, which is the behavior that was
observed.
Observed IE Rotation Under Torque Control vs ISO 22622 IE Rotation
The resulting internal and external rotations varied widely among the specimens.
However, the trend of the mean rotations was similar to the specified rotations in ISO 22622. An
internal moment was expected to create a positive rotation which was consistent with both the
motion tracking data collected from the specimens and with the internal rotation ISO command.
Similarly, as the gait cycle progressed, an external moment was applied and was expected to

239
result in a negative rotation of the tibia. Both the specimen recorded data and the ISO rotation
protocol agreed on these rotations. Moreover, from the applied torque profile, the tibia did not
exceed the rotations suggested in the ISO 22622 internal-external rotation angles.
This data may suggest for AP displacement and tibial rotations, load or displacement
control may effectively produce the same kinematics. Conversely, Sutton et al. reported for
cadaveric knees, force control may be more suitable due to the high reliance on ligament stability
for the knee
13
. Interestingly, the ankle is unique in that it is stabilized by both ligaments and
articulating bone surfaces for stabilization
17; 18
. At low loads, displacements may be small and
well within the viscoelastic range of the ligaments providing stabilization. However, at high
loads, the ankle effectively becomes a rigid lever relying on its articulating structure for stability.
This may provide an explanation on why higher loads and torques do not necessarily produce
higher displacements in the ankle, as seen in knees. While force control produced
physiologically correct displacement trends, the benefits of force control may not outweigh the
complexity of conducting the experiment using the ISO load control option.
Limitations
The present study had several limitations. First, due to the limitations of structural
strength of cadaveric feet, the applied force profile command was at 50% of the magnitudes
specified by ISO 22622. However, the plantarflexion-dorsiflexion was applied under rotation
control as specified by ISO 22622. Further research must be conducted to investigate if testing
at the full 100% will change the observed AP and axial rotational displacements. While the
cadaver specimens successfully underwent the axial load profile at 50% of the magnitudes
specified by ISO these specimens did not precisely follow the dorsiflexion-plantarflexion
rotation profile, AP load profile, and IE torque profile perhaps due to ligament constraints and
various joint impingements. For example, at lower loads, ankle stability may be dependent on

240
ligaments; however, at higher loads, the articular geometry is the primary stabilizer
18; 19
.
Specifically, at these higher loads that occur from mid-stance to push off, the bones are locked
into a forced position to create a solid rigid lever of support in order to propel the subject
forward
17
. Therefore, under the sharp increase in axial load from 20% to 30% of gait along with
the highest peak of AP load, the specimen may have been relying on the tibiotalar joint for
stabilization. The differences in AP loads can be attributed to the variance in specimen flexibility. The
AP force applied to the tibia was not feedback controlled; therefore, the load observed was reliant on the
stiffness of the specimen and its resistance to the force produced by the displacement-controlled actuator
and spring mechanism. Finally, without the cuneiform and metatarsals, the specimens may have been
biomechanically compromised, altering the response.
Conclusions
In conclusion, the results of the present study indicated that load control to produce AP
displacement and axial rotations does not produce substantially different magnitudes of displacements
than those specified by ISO 22622 using the displacement-controlled option. Since joint wear simulators
previously used for TAR wear evaluation have operated under displacement control for AP motion and
axial rotations, the additional expenses and difficulties associated with the implementation of load control
for TAR wear testing may not be justified. Moreover, the implementation of load control for these
motions, while arguably more physiologically accurate, may introduce uncertainties in the resulting
motions which may lead to lack of reproducibility when comparing results for different designs or from
different investigators.

241

1. Haddad SL, Coetzee JC, Estok R, et al. 2007. Intermediate and long-term outcomes of total ankle
arthroplasty and ankle arthrodesis. A systematic review of the literature. J Bone Joint Surg Am
89:1899-1905.
2. Harston A, Lazarides AL, Adams SB, Jr., et al. 2017. Midterm Outcomes of a Fixed-Bearing Total
Ankle Arthroplasty With Deformity Analysis. Foot Ankle Int 38:1295-1300.
3. Koivu H, Kohonen I, Mattila K, et al. 2017. Long-term Results of Scandinavian Total Ankle
Replacement. Foot Ankle Int 38:723-731.
4. LaMothe J, Seaworth CM, Do HT, et al. 2016. Analysis of Total Ankle Arthroplasty Survival in the
United States Using Multiple State Databases. Foot Ankle Spec 9:336-341.
5. Raikin SM, Sandrowski K, Kane JM, et al. 2017. Midterm Outcome of the Agility Total Ankle
Arthroplasty. Foot Ankle Int 38:662-670.
6. Endo M, Tipper JL, Barton DC, et al. 2002. Comparison of wear, wear debris and functional
biological activity of moderately crosslinked and non-crosslinked polyethylenes in hip
prostheses. Proc Inst Mech Eng H 216:111-122.
7. Saikko V, Calonius O. 2002. Slide track analysis of the relative motion between femoral head and
acetabular cup in walking and in hip simulators. J Biomech 35:455-464.
8. Mujukian A, Ho NC, Day MJ, et al. 2020. A Systematic Review of Unsystematic Total Ankle
Replacement Wear Evaluations. JBJS Rev 8:e0091.
9. DesJardins JD, Banks SA, Benson LC, et al. 2007. A direct comparison of patient and force-
controlled simulator total knee replacement kinematics. J Biomech 40:3458-3466.
10. Reinders J, von Stillfried F, Altan E, et al. 2015. Force-controlled dynamic wear testing of total
ankle replacements. Acta Biomater 12:332-340.
11. ISO. 2019. Implants for Surgery- Wear of Total Ankle-Joint Prostheses - Loading and
Displacement Parameters for wear-testing machines with load or displacement control and
corresponding environmental conditions for test (22622). Geneva, Switzerland: International
Organization for Standardization
12. ISO. 2014. Implants for surgery- Wear of Total knee-joint prostheses - Part 3: Loading and
Displacement parameters for wear-testing machines with displacement control and
corresponding environmental conditions for test (14243-3). Geneva, Switzerland: International
Organization for Standardization.
13. Sutton LG, Werner FW, Haider H, et al. 2010. In vitro response of the natural cadaver knee to
the loading profiles specified in a standard for knee implant wear testing. J Biomech 43:2203-
2207.
14. Leardini A, O'Connor JJ, Giannini S. 2014. Biomechanics of the natural, arthritic, and replaced
human ankle joint. J Foot Ankle Res 7:8.
15. Affatato S, Taddei P, Leardini A, et al. 2009. Wear behaviour in total ankle replacement: a
comparison between an in vitro simulation and retrieved prostheses. Clin Biomech (Bristol,
Avon) 24:661-669.
16. Stratton-Powell A RA, Tipper JL, Fisher J, Brockett C. . 2017. Surface Characterization of Total
Ankle Replacement: Retrieved vs In-Vitro Simulation.
17. Chan CW, Rudins A. 1994. Foot biomechanics during walking and running. Mayo Clin Proc
69:448-461.
18. Watanabe K, Kitaoka HB, Berglund LJ, et al. 2012. The role of ankle ligaments and articular
geometry in stabilizing the ankle. Clin Biomech (Bristol, Avon) 27:189-195.
19. Tochigi Y, Rudert MJ, Brown TD, et al. 2005. The effect of accuracy of implantation on range of
movement of the Scandinavian Total Ankle Replacement. J Bone Joint Surg Br 87:736-740.

242

References
1. Abel MF. Brace Success as Related to Curve Type, Compliance, and Maturity in Adolescents
with Idiopathic Scoliosis: Commentary on an article by Rachel M. Thompson, MD, et al.: "Brace
Success Is Related to Curve Type in Patients with Adolescent Idiopathic Scoliosis". J Bone Joint
Surg Am. 2017;99:e59.
2. Affatato S, Leardini A, Leardini W, Giannini S, Viceconti M. Meniscal wear at a three-
component total ankle prosthesis by a knee joint simulator. J Biomech. 2007;40:1871-1876.
3. Affatato S, Taddei P, Leardini A, Giannini S, Spinelli M, Viceconti M. Wear behaviour in total
ankle replacement: a comparison between an in vitro simulation and retrieved prostheses. Clin
Biomech (Bristol, Avon). 2009;24:661-669.
4. Agel J, Coetzee JC, Sangeorzan BJ, Roberts MM, Hansen ST, Jr. Functional limitations of
patients with end-stage ankle arthrosis. Foot Ankle Int. 2005;26:537-539.
5. Anderson FC, Pandy MG. Static and dynamic optimization solutions for gait are practically
equivalent. J Biomech. 2001;34:153-161.
6. Ardnt AW, P; Liu, A; Nester, C; Stacoff, A; Jones, R; Lundgren, P, Lundberg, A,. Intrinsic Foot
Kinematics Measured In Vivo During the Stance Phase of Slow Running. Journal of
Biomechanics. 2006;40:7.
7. Assal M, Al-Shaikh R, Reiber BH, Hansen ST. Fracture of the polyethylene component in an
ankle arthroplasty: a case report. Foot Ankle Int. 2003;24:901-903.
8. Atwood SA, Currier JH, Mayor MB, Collier JP, Van Citters DW, Kennedy FE. Clinical wear
measurement on low contact stress rotating platform knee bearings. J Arthroplasty. 2008;23:431-
440.
9. Aubin PM, Whittaker, E.C., Ledoux, W.R. Foot Bone Kinematics at Half and Three Quarters
Body Weight: A Robotic Cadaveric Simulation of Stance Phase. 15th International Conference
on Advanced Robotics (ICAR). New York, NY: IEEE; 2011:5.
10. Bahr R, Pena F, Shine J, Lew WD, Engebretsen L. Ligament force and joint motion in the intact
ankle: a cadaveric study. Knee Surg Sports Traumatol Arthrosc. 1998;6:115-121.
11. Barg A, Knupp M, Anderson AE, Hintermann B. Total ankle replacement in obese patients:
component stability, weight change, and functional outcome in 118 consecutive patients. Foot
Ankle Int. 2011;32:925-932.
12. Baxter JR, Demetracopoulos CA, Prado MP, Tharmviboonsri T, Deland JT. Lateral column
lengthening corrects hindfoot valgus in a cadaveric flatfoot model. Foot Ankle Int. 2015;36:705-
709.
13. Bell CJ, Fisher J. Simulation of polyethylene wear in ankle joint prostheses. J Biomed Mater Res
B Appl Biomater. 2007;81:162-167.
14. Benich MR, Ledoux WR, Orendurff MS, Shofer JB, Hansen ST, Davitt J, Anderson JG, Bohay
D, Coetzee JC, Maskill J, Brage M, Houghton M, Sangeorzan BJ. Comparison of Treatment
Outcomes of Arthrodesis and Two Generations of Ankle Replacement Implants. J Bone Joint
Surg Am. 2017;99:1792-1800.
15. Billi F, Sangiorgio SN, Aust S, Ebramzadeh E. Material and surface factors influencing backside
fretting wear in total knee replacement tibial components. J Biomech. 2010;43:1310-1315.
16. Bills BJ, Racasan R, Underwood RJ, Cann P, Skinner J, Hart AJ, Jiang X, Blunt L. Volumetric
Wear Assessment of Retrieved Metal-on-Metal Hip Prostheses and the Impact of Measurement
Uncertainty. Wear. 2012;274-275:212-219.
17. Bischoff JE, Fryman JC, Parcell J, Orozco Villasenor DA. Influence of crosslinking on the wear
performance of polyethylene within total ankle arthroplasty. Foot Ankle Int. 2015;36:369-376.

243
18. Bonnin M, Gaudot F, Laurent JR, Ellis S, Colombier JA, Judet T. The Salto total ankle
arthroplasty: survivorship and analysis of failures at 7 to 11 years. Clin Orthop Relat Res.
2011;469:225-236.
19. Bonnin MP, Laurent JR, Casillas M. Ankle function and sports activity after total ankle
arthroplasty. Foot Ankle Int. 2009;30:933-944.
20. Bracco P, Oral E. Vitamin E-stabilized UHMWPE for total joint implants: a review. Clin Orthop
Relat Res. 2011;469:2286-2293.
21. Brockett CL, Chapman GJ. Biomechanics of the ankle. Orthop Trauma. 2016;30:232-238.
22. Brodsky JW, Coleman SC, Smith S, Polo FE, Tenenbaum S. Hindfoot motion following STAR
total ankle arthroplasty: a multisegment foot model gait study. Foot Ankle Int. 2013;34:1479-
1485.
23. Brown TD, Johnston RC, Saltzman CL, Marsh JL, Buckwalter JA. Posttraumatic osteoarthritis: a
first estimate of incidence, prevalence, and burden of disease. J Orthop Trauma. 2006;20:739-
744.
24. Brunner S, Barg A, Knupp M, Zwicky L, Kapron AL, Valderrabano V, Hintermann B. The
Scandinavian total ankle replacement: long-term, eleven to fifteen-year, survivorship analysis of
the prosthesis in seventy-two consecutive patients. J Bone Joint Surg Am. 2013;95:711-718.
25. Buckwalter JA, Saltzman C, Brown T. The impact of osteoarthritis: implications for research.
Clin Orthop Relat Res. 2004:S6-15.
26. Cappozzo A, Catani F, Croce UD, Leardini A. Position and orientation in space of bones during
movement: anatomical frame definition and determination. Clin Biomech (Bristol, Avon).
1995;10:171-178.
27. Chan CW, Rudins A. Foot biomechanics during walking and running. Mayo Clin Proc.
1994;69:448-461.
28. Charnley J. Compression arthrodesis of the ankle and shoulder. J Bone Joint Surg Br.
1951;33B:180-191.
29. Clough T, Bodo K, Majeed H, Davenport J, Karski M. Survivorship and long-term outcome of a
consecutive series of 200 Scandinavian Total Ankle Replacement (STAR) implants. Bone Joint J.
2019;101-B:47-54.
30. Cody EA, Bejarano-Pineda L, Lachman JR, Taylor MA, Gausden EB, DeOrio JK, Easley ME,
Nunley JA, 2nd. Risk Factors for Failure of Total Ankle Arthroplasty with a Minimum Five
Years of Follow-up. Foot Ankle Int. 2018:1071100718806474.
31. Cody EA, Bejarano-Pineda L, Lachman JR, Taylor MA, Gausden EB, DeOrio JK, Easley ME,
Nunley JA, 2nd. Risk Factors for Failure of Total Ankle Arthroplasty With a Minimum Five
Years of Follow-up. Foot Ankle Int. 2019;40:249-258.
32. Conti S, Lalonde KA, Martin R. Kinematic analysis of the agility total ankle during gait. Foot
Ankle Int. 2006;27:980-984.
33. Cracchiolo A, 3rd, Deorio JK. Design features of current total ankle replacements: implants and
instrumentation. J Am Acad Orthop Surg. 2008;16:530-540.
34. Cristofolini L, Teutonico AS, Monti L, Cappello A, Toni A. Comparative in vitro study on the
long term performance of cemented hip stems: validation of a protocol to discriminate between
"good" and "bad" designs. J Biomech. 2003;36:1603-1615.
35. Cristofolini L, Viceconti M, Cappello A, Toni A. Mechanical validation of whole bone composite
femur models. J Biomech. 1996;29:525-535.
36. Crowninshield RD, Brand RA. A physiologically based criterion of muscle force prediction in
locomotion. J Biomech. 1981;14:793-801.
37. Currier BH, Currier JH, Franklin KJ, Mayor MB, Reinitz SD, Van Citters DW. Comparison of
Wear and Oxidation in Retrieved Conventional and Highly Cross-Linked UHMWPE Tibial
Inserts. J Arthroplasty. 2015;30:2349-2353.
38. Currier BH, Hecht PJ, Nunley JA, 2nd, Mayor MB, Currier JH, Van Citters DW. Analysis of
Failed Ankle Arthroplasty Components. Foot Ankle Int. 2019;40:131-138.

244
39. Di Iorio A, Viste A, Fessy MH, Besse JL. The AES total ankle arthroplasty analysis of failures
and survivorship at ten years. Int Orthop. 2017;41:2525-2533.
40. Dunn JE, Link CL, Felson DT, Crincoli MG, Keysor JJ, McKinlay JB. Prevalence of foot and
ankle conditions in a multiethnic community sample of older adults. Am J Epidemiol.
2004;159:491-498.
41. Ebramzadeh E, Billi F, Sangiorgio SN, Mattes S, Schmoelz W, Dorr L. Simulation of fretting
wear at orthopaedic implant interfaces. J Biomech Eng. 2005;127:357-363.
42. Ebramzadeh E, Sangiorgio SN, Longjohn DB, Buhari CF, Dorr LD. Initial stability of cemented
femoral stems as a function of surface finish, collar, and stem size. J Bone Joint Surg Am.
2004;86-A:106-115.
43. Ebramzadeh E, Sangiorgio SN, Longjohn DB, Buhari CF, Morrison BJ, Dorr LD. Effects of total
hip arthroplasty cemented femoral stem surface finish, collar and cement thickness on load
transfer to the femur. J Appl Biomater Biomech. 2003;1:76-83.
44. Egloff C, Hugle T, Valderrabano V. Biomechanics and pathomechanisms of osteoarthritis. Swiss
Med Wkly. 2012;142:w13583.
45. Eloesser L. Implantation of Joints. Cal State J Med. 1913;11:485-491.
46. Fisher J, McEwen HM, Tipper JL, Galvin AL, Ingram J, Kamali A, Stone MH, Ingham E. Wear,
debris, and biologic activity of cross-linked polyethylene in the knee: benefits and potential
concerns. Clin Orthop Relat Res. 2004:114-119.
47. Flavin R, Coleman SC, Tenenbaum S, Brodsky JW. Comparison of gait after total ankle
arthroplasty and ankle arthrodesis. Foot Ankle Int. 2013;34:1340-1348.
48. Frigg A, Germann U, Huber M, Horisberger M. Survival of the Scandinavian total ankle
replacement (STAR): results of ten to nineteen years follow-up. Int Orthop. 2017;41:2075-2082.
49. Frigo C, Rabuffetti M, Kerrigan DC, Deming LC, Pedotti A. Functionally oriented and clinically
feasible quantitative gait analysis method. Med Biol Eng Comput. 1998;36:179-185.
50. Fuentes-Sanz A, Moya-Angeler J, Lopez-Oliva F, Forriol F. Clinical outcome and gait analysis of
ankle arthrodesis. Foot Ankle Int. 2012;33:819-827.
51. Glazebrook M, Daniels T, Younger A, Foote CJ, Penner M, Wing K, Lau J, Leighton R, Dunbar
M. Comparison of health-related quality of life between patients with end-stage ankle and hip
arthrosis. J Bone Joint Surg Am. 2008;90:499-505.
52. Glazebrook MA, Arsenault K, Dunbar M. Evidence-based classification of complications in total
ankle arthroplasty. Foot Ankle Int. 2009;30:945-949.
53. Gougoulias N, Khanna A, Maffulli N. How successful are current ankle replacements?: a
systematic review of the literature. Clin Orthop Relat Res. 2010;468:199-208.
54. Graichen F, Arnold R, Rohlmann A, Bergmann G. Implantable 9-channel telemetry system for in
vivo load measurements with orthopedic implants. IEEE Trans Biomed Eng. 2007;54:253-261.
55. Grass R, Rammelt S, Biewener A, Zwipp H. Arthrodesis of the ankle joint. Clin Podiatr Med
Surg. 2004;21:161-178.
56. Haddad SL, Coetzee JC, Estok R, Fahrbach K, Banel D, Nalysnyk L. Intermediate and long-term
outcomes of total ankle arthroplasty and ankle arthrodesis. A systematic review of the literature. J
Bone Joint Surg Am. 2007;89:1899-1905.
57. Hahn ME, Wright ES, Segal AD, Orendurff MS, Ledoux WR, Sangeorzan BJ. Comparative gait
analysis of ankle arthrodesis and arthroplasty: initial findings of a prospective study. Foot Ankle
Int. 2012;33:282-289.
58. Hardt DE. Determining Muscle Forces in the Leg During Normal Human Walking-An
Application and Evaluation of Optimization Methods. J Biomech Eng. 1977;100:7.
59. Harston A, Lazarides AL, Adams SB, Jr., DeOrio JK, Easley ME, Nunley JA, 2nd. Midterm
Outcomes of a Fixed-Bearing Total Ankle Arthroplasty With Deformity Analysis. Foot Ankle Int.
2017;38:1295-1300.

245
60. Heckmann N, Bradley A, Sivasundaram L, Alluri RK, Tan EW. Effect of Insurance on Rates of
Total Ankle Arthroplasty Versus Arthrodesis for Tibiotalar Osteoarthritis. Foot Ankle Int.
2017;38:133-139.
61. Henricson A, Skoog A, Carlsson A. The Swedish Ankle Arthroplasty Register: An analysis of
531 arthroplasties between 1993 and 2005. Acta Orthop. 2007;78:569-574.
62. Hintermann B, Valderrabano V. Total ankle replacement. Foot Ankle Clin. 2003;8:375-405.
63. Ho NC, Ebramzadeh E, Sangiorgio SN. Preclinical biomechanical testing models for the
tibiotalar joint and its replacements: A systematic review. Foot Ankle Surg. 2018.
64. Ho NC, Sangiorgio SN, Cassinelli S, Shymon S, Fleming J, Agrawal V, Ebramzadeh E, Harris
TG. Biomechanical comparison of fixation stability using a Lisfranc plate versus transarticular
screws. Foot Ankle Surg. 2017.
65. Hood RW, Wright TM, Burstein AH. Retrieval analysis of total knee prostheses: a method and its
application to 48 total condylar prostheses. J Biomed Mater Res. 1983;17:829-842.
66. Hopgood P, Kumar R, Wood PL. Ankle arthrodesis for failed total ankle replacement. J Bone
Joint Surg Br. 2006;88:1032-1038.
67. Hunt AE, Smith RM, Torode M, Keenan AM. Inter-segment foot motion and ground reaction
forces over the stance phase of walking. Clin Biomech (Bristol, Avon). 2001;16:592-600.
68. Hutchinson IDB, J.R.; Gilbert, S. et al. How Do Hindfoot Fusions Affect Ankle Biomechanics: A
Cadaver Model. Clinical Orthopaedics and Related Research. 2016;474:8.
69. Hvid I, Rasmussen O, Jensen NC, Nielsen S. Trabecular bone strength profiles at the ankle joint.
Clin Orthop Relat Res. 1985:306-312.
70. Ingham E, Fisher J. Biological reactions to wear debris in total joint replacement. Proc Inst Mech
Eng H. 2000;214:21-37.
71. Inman VT. Functional aspects of the abductor muscles of the hip. J Bone Joint Surg Am.
1947;29:607-619.
72. Jastifer JR, Coughlin MJ. Long-term follow-up of mobile bearing total ankle arthroplasty in the
United States. Foot Ankle Int. 2015;36:143-150.
73. Joveniaux P, Harisboure A, Ohl X, Dehoux E. Long-term results of in situ subtalar arthrodesis.
Int Orthop. 2010;34:1199-1205.
74. Khazzam M, Long JT, Marks RM, Harris GF. Preoperative gait characterization of patients with
ankle arthrosis. Gait Posture. 2006;24:85-93.
75. Kim BS, Choi WJ, Kim YS, Lee JW. Total ankle replacement in moderate to severe varus
deformity of the ankle. J Bone Joint Surg Br. 2009;91:1183-1190.
76. Kim BS, Knupp M, Zwicky L, Lee JW, Hintermann B. Total ankle replacement in association
with hindfoot fusion: Outcome and complications. J Bone Joint Surg Br. 2010;92:1540-1547.
77. Knutsen AR, Fleming JF, Ebramzadeh E, Ho NC, Warganich T, Harris TG, Sangiorgio SN.
Biomechanical Comparison of Fixation Devices for First Metatarsocuneiform Joint Arthrodesis.
Foot Ankle Spec. 2017;10:322-328.
78. Koivu H, Kohonen I, Mattila K, Loyttyniemi E, Tiusanen H. Long-term Results of Scandinavian
Total Ankle Replacement. Foot Ankle Int. 2017;38:723-731.
79. Komistek RD, Stiehl JB, Buechel FF, Northcut EJ, Hajner ME. A determination of ankle
kinematics using fluoroscopy. Foot Ankle Int. 2000;21:343-350.
80. Krause FG, Windolf M, Bora B, Penner MJ, Wing KJ, Younger AS. Impact of complications in
total ankle replacement and ankle arthrodesis analyzed with a validated outcome measurement. J
Bone Joint Surg Am. 2011;93:830-839.
81. Lachiewicz PF, Geyer MR. The use of highly cross-linked polyethylene in total knee arthroplasty.
J Am Acad Orthop Surg. 2011;19:143-151.
82. LaMothe J, Seaworth CM, Do HT, Kunas GC, Ellis SJ. Analysis of Total Ankle Arthroplasty
Survival in the United States Using Multiple State Databases. Foot Ankle Spec. 2016;9:336-341.
83. Laurent MPJTSY, J.Q.; Blanchard, C.R.; Crowninshield, R.D. In vitro lateral versus medial wear
of a knee prosthesis. Wear. 2003;255:6.

246
84. Lawrence RC, Felson DT, Helmick CG, Arnold LM, Choi H, Deyo RA, Gabriel S, Hirsch R,
Hochberg MC, Hunder GG, Jordan JM, Katz JN, Kremers HM, Wolfe F, National Arthritis Data
W. Estimates of the prevalence of arthritis and other rheumatic conditions in the United States.
Part II. Arthritis Rheum. 2008;58:26-35.
85. Lawton CD, Butler BA, Dekker RG, 2nd, Prescott A, Kadakia AR. Total ankle arthroplasty
versus ankle arthrodesis-a comparison of outcomes over the last decade. J Orthop Surg Res.
2017;12:76.
86. Leardini A, O'Connor JJ, Giannini S. Biomechanics of the natural, arthritic, and replaced human
ankle joint. J Foot Ankle Res. 2014;7:8.
87. List R, Foresti M, Gerber H, Goldhahn J, Rippstein P, Stussi E. Three-dimensional kinematics of
an unconstrained ankle arthroplasty: a preliminary in vivo videofluoroscopic feasibility study.
Foot Ankle Int. 2012;33:883-892.
88. Lord G, Marotte JH. [Total ankle prosthesis. Technic and 1st results. Apropos of 12 cases]. Rev
Chir Orthop Reparatrice Appar Mot. 1973;59:139-151.
89. Lu Z, McKellop HA. Accuracy of methods for calculating volumetric wear from coordinate
measuring machine data of retrieved metal-on-metal hip joint implants. Proc Inst Mech Eng H.
2014;228:237-249.
90. Lubbeke A, Salvo D, Stern R, Hoffmeyer P, Holzer N, Assal M. Risk factors for post-traumatic
osteoarthritis of the ankle: an eighteen year follow-up study. Int Orthop. 2012;36:1403-1410.
91. Mann RA, Beaman DN, Horton GA. Isolated subtalar arthrodesis. Foot Ankle Int. 1998;19:511-
519.
92. Manning DW, Chiang PP, Martell JM, Galante JO, Harris WH. In vivo comparative wear study
of traditional and highly cross-linked polyethylene in total hip arthroplasty. J Arthroplasty.
2005;20:880-886.
93. Mathiyakom W, McNitt-Gray JL, Requejo P, Costa K. Modifying center of mass trajectory
during sit-to-stand tasks redistributes the mechanical demand across the lower extremity joints.
Clin Biomech (Bristol, Avon). 2005;20:105-111.
94. McInnes KA, Younger AS, Oxland TR. Initial instability in total ankle replacement: a cadaveric
biomechanical investigation of the STAR and agility prostheses. J Bone Joint Surg Am.
2014;96:e147.
95. McKellop H, Shen FW, DiMaio W, Lancaster JG. Wear of gamma-crosslinked polyethylene
acetabular cups against roughened femoral balls. Clin Orthop Relat Res. 1999:73-82.
96. McKellop H, Shen FW, Lu B, Campbell P, Salovey R. Development of an extremely wear-
resistant ultra high molecular weight polyethylene for total hip replacements. J Orthop Res.
1999;17:157-167.
97. McKellop HA. The lexicon of polyethylene wear in artificial joints. Biomaterials. 2007;28:5049-
5057.
98. McKellop HA, Campbell P, Park SH, Schmalzried TP, Grigoris P, Amstutz HC, Sarmiento A.
The origin of submicron polyethylene wear debris in total hip arthroplasty. Clin Orthop Relat
Res. 1995:3-20.
99. McNitt-Gray JL. Kinetics of the lower extremities during drop landings from three heights. J
Biomech. 1993;26:1037-1046.
100. McNitt-Gray JL. Subject specific coordination of two- and one-joint muscles during landings
suggests multiple control criteria. Motor Control. 2000;4:84-88; discussion 97-116.
101. McNitt-Gray JL, Hester DM, Mathiyakom W, Munkasy BA. Mechanical demand and multijoint
control during landing depend on orientation of the body segments relative to the reaction force. J
Biomech. 2001;34:1471-1482.
102. McNitt-Gray JL, Munaretto J, Zaferiou A, Requejo PS, Flashner H. Regulation of reaction forces
during the golf swing. Sports Biomech. 2013;12:121-131.
103. Michael JM, Golshani A, Gargac S, Goswami T. Biomechanics of the ankle joint and clinical
outcomes of total ankle replacement. J Mech Behav Biomed Mater. 2008;1:276-294.

247
104. Michelson JD, Helgemo SL, Jr. Kinematics of the axially loaded ankle. Foot Ankle Int.
1995;16:577-582.
105. Moseley L, Smith R, Hunt A, Gant R. Three-dimensional kinematics of the rearfoot during the
stance phase of walking in normal young adult males. Clin Biomech (Bristol, Avon). 1996;11:39-
45.
106. Muratoglu OK, Mark A, Vittetoe DA, Harris WH, Rubash HE. Polyethylene damage in total
knees and use of highly crosslinked polyethylene. J Bone Joint Surg Am. 2003;85-A Suppl 1:S7-
S13.
107. Nester CJ, Liu, A.M., Ward, E., Howard, D., Cocheba, J., Derrick, T., Patterson, P. In Vitro
Study of Foot Kinematics using a Dynamic Walking Cadaver Model. Journal of Biomechanics.
2007;40:10.
108. Oral E, Christensen SD, Malhi AS, Wannomae KK, Muratoglu OK. Wear resistance and
mechanical properties of highly cross-linked, ultrahigh-molecular weight polyethylene doped
with vitamin E. J Arthroplasty. 2006;21:580-591.
109. Park SH, Lu Z, Hastings RS, Campbell PA, Ebramzadeh E. Five Hundred Fifty-five Retrieved
Metal-on-metal Hip Replacements of a Single Design Show a Wide Range of Wear, Surface
Features, and Histopathologic Reactions. Clin Orthop Relat Res. 2018;476:261-278.
110. Partridge S, Tipper JL, Al-Hajjar M, Isaac GH, Fisher J, Williams S. Evaluation of a new
methodology to simulate damage and wear of polyethylene hip replacements subjected to edge
loading in hip simulator testing. J Biomed Mater Res B Appl Biomater. 2018;106:1456-1462.
111. Patil S, Bergula A, Chen PC, Colwell CW, Jr., D'Lima DD. Polyethylene wear and acetabular
component orientation. J Bone Joint Surg Am. 2003;85-A Suppl 4:56-63.
112. Paul JP. Force actions transmitted by joints in the human body. Proc R Soc Lond B Biol Sci.
1976;192:163-172.
113. Penner MJ, Butterwick D. Fixed Versus Mobile Bearings in Total Ankle Arthroplasty. In: Roukis
TS, Berlet GC, Bibbo C, Hyer CF, Penner MJ, Wünschel M, Prissel MA, ed. Primary and
Revision Total Ankle Replacement: Evidence-Based Surgical Management. Cham: Springer
International Publishing; 2016:25-30.
114. Pfeiff C. The Scandinavian total ankle replacement (STAR). Orthop Nurs. 2006;25:30-33; quiz
34-35.
115. Picavet HS, Schouten JS. Musculoskeletal pain in the Netherlands: prevalences, consequences
and risk groups, the DMC(3)-study. Pain. 2003;102:167-178.
116. Piriou P, Culpan P, Mullins M, Cardon JN, Pozzi D, Judet T. Ankle replacement versus
arthrodesis: a comparative gait analysis study. Foot Ankle Int. 2008;29:3-9.
117. Postak PR, M; Greenwald, A; . Evaluation of the S.T.A.R. Total Ankle Replacement: An
Evolution in Design. American Academy of Orthopaedic Surgeons. San Francisco; 2008.
118. Procter P, Paul JP. Ankle joint biomechanics. J Biomech. 1982;15:627-634.
119. Queen RM, De Biassio JC, Butler RJ, DeOrio JK, Easley ME, Nunley JA. J. Leonard Goldner
Award 2011: changes in pain, function, and gait mechanics two years following total ankle
arthroplasty performed with two modern fixed-bearing prostheses. Foot Ankle Int. 2012;33:535-
542.
120. Queen RM, Franck CT, Schmitt D, Adams SB. Are There Differences in Gait Mechanics in
Patients With A Fixed Versus Mobile Bearing Total Ankle Arthroplasty? A Randomized Trial.
Clin Orthop Relat Res. 2017;475:2599-2606.
121. Rahman T, Bowen JR, Takemitsu M, Scott C. The association between brace compliance and
outcome for patients with idiopathic scoliosis. J Pediatr Orthop. 2005;25:420-422.
122. Raikin SM, Sandrowski K, Kane JM, Beck D, Winters BS. Midterm Outcome of the Agility
Total Ankle Arthroplasty. Foot Ankle Int. 2017;38:662-670.
123. Reinders J, von Stillfried F, Altan E, Sonntag R, Heitzmann DW, Kretzer JP. Force-controlled
dynamic wear testing of total ankle replacements. Acta Biomater. 2015;12:332-340.

248
124. Reinders J, von Stillfried F, Altan E, Sonntag R, Heitzmann DWW, Kretzer JP. Force-controlled
dynamic wear testing of total ankle replacements. Acta Biomater. 2015;12:332-340.
125. Renström PAFH, Lynch SA. Ankle ligament injuries. Revista Brasileira de Medicina do Esporte.
1998;4:71-80.
126. Ries MD. Highly cross-linked polyethylene: the debate is over--in opposition. J Arthroplasty.
2005;20:59-62.
127. Roukis TSB, Annette F.P. Primary and Revision Total Ankle Replacement: Springer Internation
Publishing; 2016.
128. Saikko V, Calonius O. Slide track analysis of the relative motion between femoral head and
acetabular cup in walking and in hip simulators. J Biomech. 2002;35:455-464.
129. Saikko VO, Paavolainen PO, Slatis P. Wear of the polyethylene acetabular cup. Metallic and
ceramic heads compared in a hip simulator. Acta Orthop Scand. 1993;64:391-402.
130. Salb KN, Wido DM, Stewart TE, DiAngelo DJ. Development of a Robotic Assembly for
Analyzing the Instantaneous Axis of Rotation of the Foot Ankle Complex. Appl Bionics Biomech.
2016;2016:5985137.
131. Saltzman CL, Salamon ML, Blanchard GM, Huff T, Hayes A, Buckwalter JA, Amendola A.
Epidemiology of ankle arthritis: report of a consecutive series of 639 patients from a tertiary
orthopaedic center. Iowa Orthop J. 2005;25:44-46.
132. Sangiorgio SN, Ebramzadeh E, Longjohn DB, Dorr LD. Effects of dorsal flanges on fixation of a
cemented total hip replacement femoral stem. J Bone Joint Surg Am. 2004;86-A:813-820.
133. Sangiorgio SN, Ho NC, Morgan RD, Ebramzadeh E, Zionts LE. The Objective Measurement of
Brace-Use Adherence in the Treatment of Idiopathic Clubfoot. J Bone Joint Surg Am.
2016;98:1598-1605.
134. Sangiorgio SN, Longjohn DB, Dorr LD, Ebramzadeh E. Challenges in relating experimental hip
implant fixation predictions to clinical observations. J Biomech. 2011;44:235-243.
135. Sangiorgio SN, Longjohn DB, Dorr LD, Ebramzadeh E. The influence of proximal stem
geometry and surface finish on the fixation of a double-tapered cemented femoral stem. J
Biomech. 2011;44:22-27.
136. Sangiorgio SN, Longjohn DB, Lee JL, Alexander JD, Dorr LD, Ebramzadeh E. Simulation of
extreme loads on the proximal femur for implant fixation assessment. J Appl Biomater Biomech.
2008;6:72-80.
137. Savory KM, Wulker N, Stukenborg C, Alfke D. Biomechanics of the hindfoot joints in response
to degenerative hindfoot arthrodeses. Clin Biomech (Bristol, Avon). 1998;13:62-70.
138. Sharkey NA, Hamel AJ. A dynamic cadaver model of the stance phase of gait: performance
characteristics and kinetic validation. Clin Biomech (Bristol, Avon). 1998;13:420-433.
139. Singh JA, Ramachandran R. Time trends in total ankle arthroplasty in the USA: a study of the
National Inpatient Sample. Clin Rheumatol. 2016;35:239-245.
140. Smith R, Rattanaprasert U, O'Dwyer N. Coordination of the ankle joint complex during walking.
Hum Mov Sci. 2001;20:447-460.
141. Smyth A, Fisher J, Suner S, Brockett C. Influence of kinematics on the wear of a total ankle
replacement. J Biomech. 2017;53:105-110.
142. Soren A, Waugh TR. The historical evolution of arthrodesis of the foot. Int Orthop. 1980;4:3-11.
143. Spirt AA, Assal M, Hansen ST, Jr. Complications and failure after total ankle arthroplasty. J
Bone Joint Surg Am. 2004;86-A:1172-1178.
144. Stauffer RN, Chao EY, Brewster RC. Force and motion analysis of the normal, diseased, and
prosthetic ankle joint. Clin Orthop Relat Res. 1977:189-196.
145. Stewart MG, Green CL, Adams SB, Jr., DeOrio JK, Easley ME, Nunley JA, 2nd. Midterm
Results of the Salto Talaris Total Ankle Arthroplasty. Foot Ankle Int. 2017;38:1215-1221.
146. Sutton LG, Werner FW, Haider H, Hamblin T, Clabeaux JJ. In vitro response of the natural
cadaver knee to the loading profiles specified in a standard for knee implant wear testing. J
Biomech. 2010;43:2203-2207.

249
147. Takemitsu M, Bowen JR, Rahman T, Glutting JJ, Scott CB. Compliance monitoring of brace
treatment for patients with idiopathic scoliosis. Spine (Phila Pa 1976). 2004;29:2070-2074;
discussion 2074.
148. Tochigi Y, Rudert MJ, Brown TD, McIff TE, Saltzman CL. The effect of accuracy of
implantation on range of movement of the Scandinavian Total Ankle Replacement. J Bone Joint
Surg Br. 2005;87:736-740.
149. Tochigi Y, Takahashi K, Yamagata M, Tamaki T. Influence of the interosseous talocalcaneal
ligament injury on stability of the ankle-subtalar joint complex--a cadaveric experimental study.
Foot Ankle Int. 2000;21:486-491.
150. Underwood R, Fowell M, Sayles R. The Development of a Standard Method for Assessing Wear
of Explanted Metal on Metal Hip Joints. ASTM International. 2013:130-145.
151. Usuelli FG, Maccario C. Pearls and Pitfalls for a Surgeon New to Ankle Replacements. Foot
Ankle Clin. 2017;22:477-489.
152. Valderrabano V, Hintermann B, Nigg BM, Stefanyshyn D, Stergiou P. Kinematic changes after
fusion and total replacement of the ankle: part 2: Movement transfer. Foot Ankle Int.
2003;24:888-896.
153. Valderrabano V, Nigg BM, von Tscharner V, Stefanyshyn DJ, Goepfert B, Hintermann B. Gait
analysis in ankle osteoarthritis and total ankle replacement. Clin Biomech (Bristol, Avon).
2007;22:894-904.
154. Valderrabano V, Pagenstert G, Horisberger M, Knupp M, Hintermann B. Sports and recreation
activity of ankle arthritis patients before and after total ankle replacement. Am J Sports Med.
2006;34:993-999.
155. van Engelen SJ, Wajer QE, van der Plaat LW, Doets HC, van Dijk CN, Houdijk H. Metabolic
cost and mechanical work during walking after tibiotalar arthrodesis and the influence of
footwear. Clin Biomech (Bristol, Avon). 2010;25:809-815.
156. Vaupel Z, Baker EA, Baker KC, Kurdziel MD, Fortin PT. Analysis of retrieved agility total ankle
arthroplasty systems. Foot Ankle Int. 2009;30:815-823.
157. Watanabe K, Kitaoka HB, Berglund LJ, Zhao KD, Kaufman KR, An KN. The role of ankle
ligaments and articular geometry in stabilizing the ankle. Clin Biomech (Bristol, Avon).
2012;27:189-195.
158. Witkowski GP, Bilkhu SK, Siskosky MJ, Gates ME, Demetropoulos CK, Fortin PT. Effects of
hindfoot constraint on syndesmotic displacement. Foot Ankle Int. 2009;30:367-374.
159. Woolf AD, Pfleger B. Burden of major musculoskeletal conditions. Bull World Health Organ.
2003;81:646-656.
160. Wu WL, Huang PJ, Lin CJ, Chen WY, Huang KF, Cheng YM. Lower extremity kinematics and
kinetics during level walking and stair climbing in subjects with triple arthrodesis or subtalar
fusion. Gait Posture. 2005;21:263-270.
161. Zaidi R, Cro S, Gurusamy K, Siva N, Macgregor A, Henricson A, Goldberg A. The outcome of
total ankle replacement: a systematic review and meta-analysis. The bone & joint journal.
2013;95-B:1500-1507.

Abstract (if available)

Abstract A clear understanding of the loads applied to any natural joint and the resulting kinematics is essential to the development and appropriate evaluation of its candidate prosthetic replacement designs and materials. Excellent long-term success of current total hip replacement designs, 97% survival at 25 years, can be attributed to decades of pre-clinical testing, failure analysis of retrieved implants, and numerous clinical outcome studies. In contrast, the complexities of the ankle joint are still unresolved, leading to poor outcomes in total ankle replacements. Short-term studies have found total ankle replacements to have a 90% survivorship within the first 5 years, but sharply declining to 65% for mid- to long-term survivorship. Due to this harsh decrease in survivorship, universal adoption of total ankle replacements has been greatly impeded. ❧ The foundation of the disappointing performance of total ankle replacements is rooted in the inconsistent nature of the preclinical testing of these implants. Unlike current hip replacement testing standards, which took decades to develop and benefited from extensive failure analysis of retrieval and clinical studies, for total ankle replacements, only recently a standard has been published by International Standards of Organization (ISO) and no standards are established by American Society for Testing and Materials (ASTM). The lack of standards has led to inconsistency among the protocols by different investigators. For example, in certain testing protocols, critical degrees of freedom affecting the natural response of the cadaver have been constrained, affecting the results. Some of the discrepancies stem from a lack of understanding of the biomechanics of the ankle joint, for example, the inappropriate assumption that the ankle behaves as a simple hinge joint. Oversimplified models of the kinematics have been used for biomechanical and wear testing. Specifically, with the lack of any standard, wear tests have been performed without a consensus on the magnitudes of rotations and translations. Moreover, the majority of preclinical wear tests have been conducted using displacement and angular rotation control due to ease of implementation. However, more recently testing via force and torque control have been introduced with the argument that these produce more physiologically relevant kinematics. With a clear lack of consistency and numerous methodological variations, the lack of advancement in the designs of total ankle replacements should come as no surprise. ❧ The overarching purpose of the proposed project was to provide and contribute to the establishment of the physiologically relevant testing methodology designed to predict in vivo performance of total ankle replacements. The following Specific Aims were designed to approach this objective. 1. Establish a model using cadaveric specimens to determine the 6 DOF motions of the natural tibia, talus, and calcaneus under simulated physiologic gait. 2. Use the cadaveric model to evaluate the 6 DOF motions of total ankle replacement components. 3. Use explanted total ankle replacement polyethylene components to assess the modes and locations of damage during in vivo use. 4. Use the cadaveric model to establish 6 DOF motions of the natural talus and the artificial talar component as a function of fusion of the subtalar joint. 5. Compare the 6 DOF motions of the natural ankle joint resulting from the proposed ISO 22622 force control testing standard to the proposed ISO 22622 displacement control testing standard. ❧ To achieve these objectives, the following studies were performed. For Specific Aim 1, a gait analysis experiment was performed using motion tracker cameras synced with force plate data. This data was analyzed at four distinct instances: Heel strike, maximum weight acceptance, mid-stance, and push off. This data was then used to correlate shank angle, ankle angle, ground reaction force angle, and ground reaction force magnitude observed during gait. The results from the gait experiment were then applied to three pairs of cadaveric lower limbs. Optical motion tracking was used to measure the 6DOF motions of the tibia, talus, and calcaneus under simulated loading. One side of each pair was tested in its natural state with all ligaments and soft tissue intact, while the contralateral side was tested with implantation of a commercially marketed total ankle replacement design. Through these results, Specific Aim 2 was achieved by comparing the kinematics of a TAR components to those of a natural ankle. ❧ For Specific Aim 3, fourteen explanted TARs were analyzed. The modes and location of damage illustrated how the explant was articulating in vivo. By comparing these damage maps with the results in Specific Aim 2, the testing method developed in Specific Aim 1 was supported. ❧ Following this, a subtalar fusion was simulated using screws on both the natural and TAR specimens used for Specific Aim 2. Using the model developed in Specific Aim 1, the same loading methodologies were repeated. Specific Aim 4 was then achieved by analyzing the differences in displacement of the specimens in the unfused and fused conditions. ❧ For Specific Aim 5, an additional set of 12 cadaveric lower limbs were tested under the proposed ISO 22622 force profile. From these experiments, the resulting displacements were recorded and compared to the proposed displacement profiles. ❧ Collectively, these studies contributed to understanding the complete kinematics of the ankle, leading to a more comprehensive preclinical testing protocol for the evaluation of total ankle replacements, resulting in improved long-term survivorship of total ankle replacements.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Control and dynamics of turning tasks with different rotation and translation requirements

PDF

Wear of metal-on-metal artificial discs for the lumbar spine

PDF

Regulation of linear and angular impulse generation: implications for athletic performance

PDF

Using nonlinear feedback control to model human landing mechanics

PDF

Understanding reactive balance control strategies in non-disabled and post-stroke gait

PDF

Identifying injury risk, improving performance, and facilitating learning using an integrated biomechanics informatics system (IBIS)

PDF

Computationally efficient design of optimal strategies for passive and semiactive damping devices in smart structures

PDF

Optimal clipped linear strategies for controllable damping

PDF

Pathomechanics of femoroacetabular impingement syndrome: utilizing subject-specific modeling approaches to investigate the influence of hip joint morphology and neuromuscular control

PDF

Design and development of ultrasonic array transducers for specialized applications

PDF

Trunk control during dynamic balance: effects of cognitive dual-task interference and a history of recurrent low back pain

Asset Metadata

Creator Ho, Nathan Chi Wai (author)

Core Title Investigation of preclinical testing methods for total ankle replacements

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Biomedical Engineering

Publication Date 05/15/2020

Defense Date 01/28/2020

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag biomechanics,implants,OAI-PMH Harvest,orthopaedic,total ankle replacements

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Ebramzadeh, Edward (committee chair), McNitt-Gray, Jill L. (committee chair), Sangiorgio, Sophia (committee chair), Masri, Sami F. (committee member), Yen, Jesse (committee member)

Creator Email honathan@usc.edu,nho618@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-308079

Unique identifier UC11663421

Identifier etd-HoNathanCh-8533.pdf (filename),usctheses-c89-308079 (legacy record id)

Legacy Identifier etd-HoNathanCh-8533.pdf

Dmrecord 308079

Document Type Dissertation

Rights Ho, Nathan Chi Wai

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA