 

 

 

 

am.”

3 5
‘ . .

n... . .. m .
. t: .
.ﬂf
3.? r...
t:

W2

q. a
: «Ta. 3

:93...

 

I: . I.
v ¢
urn Jr

.x‘nﬂi

..
N. *H- n ,
Launww.
.35.“. , u

f. .
1.. .
a .F. :

\ .
4'12 .hr

(«9”
‘ , 3r
.tri.

V

«m

 

This is to certify that the

dissertation entitled

CORRELATIONS OF STRESS AND STRAIN WITH

ALERATIONS IN CARTILAGE AND UNDERLYING
SUBCHONDRAL BONE FOLLOWING AN IMPACT IN AN
IN VIVO ANIMAL AND AN IN VITRO EXPLANT MODEL

presented by

Benjamin James Ewers 111

has been accepted towards fulﬁllment
of the requirements for

Ph.D. Engineering Mechanics

degree in

 

 

75 Major professor

Date 03/19/01
/ /

MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771

 

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE ‘ DATE DUE DATE DUE
Augggagugz

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6101 c-jCIRC/DateDuepes-pts

 

 

CORRELATIONS OF STRESS AND STRAIN WITH ALTERATIONS IN
CARTILAGE AND UNDERLYING SUBCHONDRAL BONE FOLLOWING AN
INIPACT IN AN IN VIVO ANIMAL AND AN IN VITRO EXPLANT MODEL

By

Benjamin James Ewers III

A DISSERTATION
Submitted to
Michigan State University
In partial fulﬁllment of the requirements
For the degree of
DOCTOR OF PHILOSOPHY
Department of Materials Science and Mechanics

2001

ABSTRACT
CORRELATIONS OF STRESS AND STRAIN WITH ALTERATIONS IN
CARTILAGE AND UNDERLYING SUBCHONDRAL BONE FOLLOWING AN
IMPACT IN AN IN VIVO ANIMAL AND AN IN VITRO EXPLANT MODEL
By

Benjamin James Ewers III

Osteoarthritis (0A) is a degradative disease affecting both the articular cartilage
and the underlying bone, with the ﬁnal stages being loss of cartilage leading to bone on
bone contact and pain. OA may cost society $54 billion annually in treatment and lost
workdays. Excessive or abnormal loading has been implicated in initiating this joint
degradation. Early signs of OA include ﬁssuring, softening and increased permeability of
the articular cartilage, chondrocyte death, and increased thickening of the underlying
subchondral bone.

To better understand the mechanisms of mechanically induced changes to joint
tissues our laboratory has developed an in vivo animal model that involves impacting the
patellofemoral joint of Giant Flemish rabbits. At 12 months post-impact the retropatellar
cartilage is ﬁssured and softened and the subchondral plate is thickened. A computational
model of the rabbit patella suggests that impact-induced shear stresses were associated
with acute ﬁssuring and chronic softening of the cartilage and thickening of the
underlying bone. To better control mechanical loading, studies on cultured explanted
cartilage have shown that a single severe impact can result in acute matrix damage
(ﬁssures) and chondrocyte death. Impact interfaces, impact orientation, and impact
duration are all important parameters that could inﬂuence the state of stress and strain in

joint tissues and therefore have an effect on both acute and chronic damages to the

tissues. Impact modeling of articular cartilage needs to take into account the well
documented non-linear stiffening and time-dependent effect of cartilage.

The overall aims of this dissertation were threefold First, to use the established
animal model to examine the inﬂuence of impact interface, impact direction, and impact
duration on chronic alterations in joint tissues. Second, use a cartilage explant system to
examine the inﬂuence of impact duration on matrix damage and cell death, and attempt to
correlate the state of stress and strain in the tissue and cells with these damages. Third,
determine a method to estimate the non-linear viscoelastic material properties of cartilage
ﬁ'om in situ indentation testing for the modeling of impact experiments.

This study found that distributing the impact load resulted in shear stresses being
decreased in the bone, but not the cartilage. As predicted there were chronic alterations in
the cartilage, but no thickening of the bone. Impacting the joint centrally compared to
slightly medial resulted in shear stresses being increased in the bone and decreased in the
cartilage. As expected there were no chronic changes in the cartilage, however, there was
also no thickening of the bone. An increase in normal bone stresses in the central impacts
could have helped protect the bone. Impacting at a higher rate of loading resulted in more
damage compared to a lower rate of loading in both the animal and explant models.
Modeling of the explant suggested that depth-dependent material properties are needed
and matrix damage was associated with high shear stresses while cell death corresponded
to high cell strains. Finally, in situ indentation testing was used to estimate the hyper-
viscoelastic material properties of cartilage, which were able to predict experimental
unconﬁned compression tests on cartilage. Future studies will be able to use these

material properties of cartilage for more appropriate modeling of impact scenarios.

DEDICATION

I would like to thank my soul mate and wife who has supported me throughout our life
together. I have endless love and respect for you, and all of my successes are because of
you. I would also like to thank my parents and sister for all of their support throughout
my entire life.

iv

ACKNOWLEDGMENTS

I would like to acknowledge my major professor Dr. Roger Haut for giving me
support, guidance and many opportunities throughout my time here at Michigan State
University.

I am extremely grateful to Dr. Michael Orth, Dr. Thomas Pence, and Dr. John
McGrath for serving on my committee and playing important roles in my education.

I would also like to thank Dr. Loic Dejardin, Dr. Steven Arnoczky, and Rhonda
Clark, for allowing me to play an important role in research outside my dissertation
topics.

Finally, I would like to express my gratitude to the many fellow research
assistants that I had the pleasure of working with: William Newberry, Dana Dvoracek-
Driksna, Richard Banglmaier, Brian Weaver, Vijay Jayaraman, Jill Krueger, Dr Patrick '

Atkinson, and Dr. Theresa Atkinson.

TABLE OF CONTENTS

LIST OF TABLES ................................................................................. ix
LIST OF FIGURES ................................................................................. x
CITATIONS OF PUBLISHED WORK CONDUCTED DURING GRADUATE
SCHOOL: DISSERTATION RESEARCH ..................................................... xii
CITATIONS OF PUBLISHED WORK CONDUCTED DURING GRADUATE
SCHOLL: ADDITIONAL PUBLISHED RESEARCH ..................................... xiii
INTRODUCTION .................................................................................. 1
CHAPTER 1

LONG TERM CHANGES IN RABBIT RETROPATELLAR CARTILAGE AND
SUBCHONDRAL BONE AFTER BLUNT IMPACT LOADING OF THE

PATELLOFEMORAL JOINT ..................................................................... 9
Abstract ............................................................................................. 9
Introduction ...................................................................................... 10
Methods and Materials ......................................................................... 13
Results ........................................................................................... 16
Discussion ....................................................................................... 19
References ...................................................................................... 25
Tables ............................................................................................ 29
Figures ........................................................................................... 32

CHAPTER 2

ALTERATIONS IN THE MECHANICAL PROPERTIES OF BONE UNDERLYING

ARTICULAR CARTILAGE IN A TRAUMATIZED JOINT ............................... 34
Introduction ...................................................................................... 34
Methods and Materials ......................................................................... 34
Results ........................................................................................... 35
Discussion ....................................................................................... 36
References ...................................................................................... 37
Figures ........................................................................................... 48

CHAPTER 3 .

CHRONIC SOFTENING OF CARTILAGE WITHOUT THICKENING OF

UNDERLYING BONE IN A JOINT TRAUMA MODEL ................................... 39
Abstract ............................................................................................ 39
Introduction ...................................................................................... 40
Methods and Materials ......................................................................... 42
Results ........................................................................................... 44

Discussion ....................................................................................... 45

References ...................................................................................... 49

Tables ............................................................................................ 5 1

Figures ............................................................................................ 52
CHAPTER 4

IIVIPACT SHEAR STRESSES IN AN ANIMAL JOINT PREDICT S CHRONIC
SOFTENING OF CARTILAGE BUT NOT THICKENING OF UNDERLYING BONE

......................................................................................................... 54
Abstract ............................................................................................ 54
Introduction ...................................................................................... 55
Methods and Materials ......................................................................... 56
Results ........................................................................................... 61
Discussion ....................................................................................... 63
References ...................................................................................... 67
Tables ............................................................................................ 70
Figures ........................................................................................... 72

CHAPTER 5

THE EFFECT OF LOADING RATE ON THE DEGREE OF ACUTE INJURY AND

CHRONIC CONDITIONS IN THE KNEE AFTER BLUNT IMPACT ................... 76
Abstract ............................................................................................ 76
Introduction ...................................................................................... 77
Methods and Materials ......................................................................... 81
Results ........................................................................................... 87
Discussion ....................................................................................... 92
References ...................................................................................... 97
Tables ......................................................................................... 101
Figures .......................................................................................... 107

CHAPTER 6

POLYSULPHATED GLYCOSAMINOGLYCAN TREATMENTS CAN MITIGATE
DECREASES IN STIFFNESS OF ARTICULAR CARTILAGE IN A TRAUMATIZED

ANIMAL JOINT ................................................................................. 113
Abstract ........................................................................................... 113
Introduction .................................................................................... 1 14
Methods and Materials ........................................................................ 116
Results .......................................................................................... 1 18
Discussion ...................................................................................... 120
References ..................................................................................... 123
Tables ........................................................................................... 126
Figures .......................................................................................... 127

vii

CHAPTER 7
THE EXTENT OF MATRD( DAMAGE AND CHONDROCYTE DEATH IN
MECHANICALLY TRAUMATIZED ARTICULAR CARTILAGE EXPLANT S

DEPENDS ON RATE OF LOADING ........................................................ 130
Abstract .......................................................................................... 130
Introduction .................................................................................... 131
Methods and Materials ........................................................................ 133
Results .......................................................................................... 135
Discussion ...................................................................................... 138
References ..................................................................................... 142
Tables ........................................................................................... 145
Figures ......................................................................................... 146

CHAPTER 8

THEORETICAL MODELS OF CARTILAGE EXPLANT S AND CORRELATIONS
WITH EXPERIMENTAL MATRIX DAMAGE AND CELL DEATH AFTER

IMPACT ING LOADS ........................................................................... 151
Abstract ........................................................................................... 151
Introduction .................................................................................... 152
Methods and Materials ........................................................................ 154
Results .......................................................................................... 158
Discussion ...................................................................................... 161
References ..................................................................................... 166
Tables ........................................................................................... 168
Figures .......................................................................................... 170

CHAPTER 9

DETERMINATION OF IN SITU MATERIAL PROPERTIES OF CARTILAGE
BASED ON A NON-LINEAR VISCOELASTIC MODEL FOR IMPACT LOADING

........................................................................................................ 176
Abstract .......................................................................................... 176
Introduction .................................................................................... 177
Methods and Materials ........................................................................ 179
Results .......................................................................................... 185
Discussion ...................................................................................... 187
References ..................................................................................... 190
Figures ........................................................................................... 193

APPENDD( A

A] Variables to deﬁne for user-deﬁned isotropic hyperelastic subroutine ............... 197

A2 UHYPER for a “generalized” Fung exponential energy function .................... 198

A3 UHYPER for the expanded, expanded exponential energy function .................. 199

APPENDIX B - Inﬁnitesimal elastic constants related to the strain-energy coefficients

8.] Pure Shear .................................................................................... 200

3.2 Uniform triaxial loading .................................................................... 200

LIST OF TABLES

CHAPTER 1

Table 1 .............................................................................................. 29
Table 2 ............................................................................................... 30
Table 3 .............................................................................................. 31
CHAPTER 3

Table l .............................................................................................. 51
Table 2 .............................................................................................. 51
CHAPTER 4

Table l .............................................................................................. 7O
Table2.....................- ......................................................................... 70
Table 3 .............................................................................................. 71
Table 4 ............................................................................................... 71
CHAPTER 5

Table 1 .............................................................................................. 101
Table 2 ............................................................................................. 102
Table 3 ............................................................................................. 103
Table 4 ............................................................................................. 104
Table 5 ............................................................................................. 105
Table 6 ............................................................................................. 106
CHAPTER 6

Table 1 ............................................................................................. 126
Table 2 ............................................................................................. 126
CHAPTER 7

Table l ............................................................................................. 145
CHAPTER 8

Table l ............................................................................................. 168
Table 2 ............................................................................................. 169
Table 3 .............................. ‘ ............................................................... 169

ix

LIST OF FIGURES

CHAPTER 1

Figure l .............................................................................................. 32
Figure 2 .............................................................................................. 32
Figure 3 .............................................................................................. 33
CHAPTER 2

Figure l ............................................................................................. 38
CHAPTER 3

Figure l .............................................................................................. 52
Figure 2 .............................................................................................. 52
Figure 3 ............................................................................................. 53
CHAPTER 4

Figure 1 .............................................................................................. 72
Figure 2 .............................................................................................. 73
Figure 3 ............. z ................................................................................ 74
Figure 4 .............................................................................................. 75
CHAPTER 5

Figure 1 ............................................................................................ 107
Figure 2 ..................................................................... ’ ....................... 107
Figure 3 ............................................................................................ 108
Figure 4 ............................................. . ............................................... 108
Figure 5 ............................................................................................ 109
Figure 6 ............................................................................................ 109
Figure 7 ............................................................................................ 110
Figure 8 ............................................................................................ 110
Figure 9 ............................................................................................ 1 11
Figure 10 ............................................................ 112
CHAPTER 6

Figure l ..................................................... ‘ ....................................... 127
Figure 2 ............................................................................................ 128
Figure 3 ............................. ' ............................................................... 1 29
CHAPTER 7

Figure l ............................................................................................ 146
Figure 2 ............................................................................................ 147
Figure 3 ............................................................................................ 148
Figure 4 ................................................................................. I .......... 149
Figure 5 ............................................................................................ 150

CHAPTER 8

Figure 1 ............................................................................................ 170
Figure 2 ............................................................................... ~ ............. 171
Figure 3 ............................................................................................ 172
Figure 4 ............................................................................................ 173
Figure 5 ............................................................................................ 174
Figure 6 ............................................................................................ 175
CHAPTER 9

Figure l ............................................................................................ 193
Figure 2 ............................................................................................ 193
Figure 3 ............................................................................................ 194
Figure 4 ............................................................................................ 194
Figure 5 ............................................................................................ 195
Figure 6 ............................................................................................ 195
Figure 7 ............................................................................................ 196
Figure 8 ............................................................................................ 196

CITATIONS OF PUBLISHED WORK CONDUCTED DURING GRADUATE
SCHOOL: DISSERTATION RESEARCH

CHAPTER 1

Ewers BJ, Weaver BT, Sevensma ET, Haut RC: Long term changes in rabbit retropatellar
cartilage and subchondral bone after blunt impact loading of the patello femoral joint. In
Preparation

CHAPTER 2

Ewers BJ, Newberry WN, Garcia JJ, Haut RC: Alterations in the mechanical properties
of bone underlying articular cartilage in a traumatized joint. Proceedings 44" Meeting of
the Orthopaedic Research Society, 1998.

CHAPTER 3
Ewers BJ, Newberry WN, Haut RC: Chronic softening of cartilage without thickening of
underlying bone in a joint trauma model. Journal of Biomechanics, 33: 1689-1694, 2000.

CHAPTER 4

Ewers BJ, Weaver BT, Newberry WN, Haut RC: Impact shear stresses in an animal joint
predicts chronic softening of cartilage but not thickening of underlying bone. Journal of
Biomechanical Engineering, In Review.

CHAPTER 5

Ewers BJ, Jayaraman VM, Banglmaier RF, Haut RC: The effect of loading rate on the
degree of acute damage and chronic conditions in the knee after blunt impact. Stapp Car
Crash Journal, 44:299-313, 2000.

CHAPTER 6
Ewers BJ, Haut RC: Polysulphated glycosaminoglycan treatments can mitigate decreases

in stifﬁress of articular cartilage in a traumatized animal joint. Journal of Orthopaedic
Research, 18:756-761, 2000.

CHAPTER 7

Ewers BJ, Dvoracek-Driksna D, Orth MW, Haut RC: The extent of matrix damage and
chondrocyte death in mechanically traumatized articular cartilage explants depends on
rate of loading. Journal of Orthopaedic Research, In Print.

CHAPTER 8

Ewers BJ, Krueger JA, Dvoracek-Driksna D, +Orth MW, Haut RC: Theoretical models
of cartilage explants and correlations with experimental matrix damage and cell death
after impacting loads. Annals of Biomedical Engineering, In Review.

CIT ATIONS OF PUBLISHED WORK CONDUCTED DURING GRADUATE
SCHOLL: ADDITIONAL PUBLISHED RESEARCH

Atkinson TS, Ewers BJ, Haut RC: The tensile and stress relaxation responses of human
patellar tendon varies with specimen cross-sectional area. Journal of Biomechanics, 32:
907-914, 1999.

Pruyn ML, Ewers BJ, Telewski FW: Thigmomorphogenesis: Changes in the morphology
and mechanical properties of two Populus hybrids. Tree Physiologv, 20:535-540, 2000.

Dejardin LM, Arnoczky SP, Ewers BJ, Haut RC, Clarke RB: Tissue engineered rotator
cuff tendon using swine small intestinal submucosa: Histological and mechanical
evaluation in dogs. Journal of Sports Medicine, In Print.

Atkinson PJ, Ewers BJ, Haut RC: Injuries to the patello femoral joint resulting from
transarticular loading are inﬂuenced by impactor mass and velocity. Journal of
Biomechaincal Engineering, In Print.

Fiechtner J, Ewers BJ, Mackenzie CD, Haut RC: Prevention of lapine post-traumatic
osteoarthritis with polysulphated glycosaminoglycans (PSGAG). Proceedings 62""
Meeting of the American Collage of Rheumatologt, 1998.

Ewers BJ, Orth MW, Haut RC: Urinary markers of post-traumatic 0A and effects of
PSGAG treatment after insult. Proceedings 45"I Meeting of the Orthopaedic Research
Society, 1999.

Ewers BJ, Garcia JJ, Altiero NJ, Haut RC: Modeling of the rabbit’s patello-femoral joint
for studies of acute blunt injury to the knee. Advances in Bioengineering ASME, 1999.

Atkinson PJ, Ewers BJ, Haut RC: Is impact energy the most important parameter
affecting transarticular loading of the patellofemoral joint? Proceedings 46'” Meeting of
the Orthopaedic Research Society, 2000.

Dejardin LM, Arnoczky SP, Ewers BJ, Haut RC: Tissue engineered rotator cuff tendon
using swine small intestinal submucosa histological and mechanical evaluation in dogs.
Proceedings 460' Meeting of the Orthopaedic Research Society, 2000.

Ewers BJ, Dvoracek-Driksna D, Orth MW, Altiero NJ, Haut RC: Matrix damage and
chondrocyte death in articular cartilage depends upon loading rate. Proceedings 46"
Meeting of the Orthopaedic Research Society, 2000.

Krueger J, Dvoracek-Driksna D, Ewers BJ, Haut RC, Orth MW: Effects of pretreatment
with glucosamine on mechanically traumatized cartilage explants, Proceedings 47"
Meeting of the Orthopaedic Research Society, 2001.

Ewers BJ, Krueger J, Dvoracek-Driksna D, Orth MW, Haut RC: Chondrocyte viability
decreases over 24 hours post-impact in a mechanically traumatized cartilage explant.
Proceedings 4 7" Meeting of the Orthopaedic Research Society, 2001.

Ewers BJ, Newberry WN, Garcia JJ, Orth MW, Haut RC: Early diagnosis, possible
treatment, and etiology of osteoarthritis due to blunt impact of the rabbit patello femoral
joint. 8‘ll Injury Prevention T hrough Biomechanics Symposium, 1998.

Ewers BJ, Dvoracek-Driksna D, Orth MW, Haut RC: Urinary collagen crosslinks and

lasting effects of polysulphated glycosaminoglycans in a traumatized animal joint. 9”I
Injury Prevention Through Biomechanics Symposium, 1999.

xiv

INTRODUCTION

An articular joint, such as the knee, is a union of bones which are covered by a
thin layer of articular cartilage. Articular cartilage consists of a ﬂuid phase composed of
water and a solid phase composed of chondrocytes and a dense network of collagen ﬁbers
interspersed with proteoglycans. It is this ultrastructure that gives cartilage its unique
material properties. Cartilage is responsible for providing a near frictionless interface
between the bones and the transmission and distribution of large loads. Osteoarthritis
(0A) is a disease affecting both the articular cartilage and the underlying bone.
Clinically, 0A is diagnosed radiograpically by joint Space narrowing which is indicative
of the joint developing full thickness loss of cartilage. At this later stage of the disease,
there is bone on bone contact that leads to severe pain and little intervention is possible
apart from joint replacement. OA may cost society $54 billion annually in treatment and
lost workdays (Hamerman, 1989).

Excessive or abnormal loading has been lmown to cause chronic alterations in
cartilage and the underlying bone indicative of early stages of this disease. Early signs of
OA include ﬁssuring, softening and increased permeability of the articular cartilage,
chondrocyte death, and increased thickening of the underlying subchondral bone plate
(Brandt et al., 1986; Kim et al., 2000). Chondrocyte death might play a signiﬁcant role in
the development of OA (Blanco et al., 1998). In support of this Simon et al. (1976) using
an animal model documented that chondrocyte death leads to chronic degradation of
cartilage. Another proposed mechanism is that the thickening of the underlying bone
causes abnormally high stresses to develop in the overlying cartilage which leads to its

further softening and eventual loss (Radin et al., 1990). While the etiology is not well

understood, mechanical trauma has clinically been implicated in initiating this joint
degradation (Chapchal, 1978; Felson, 1990).

Joint trauma routinely occurs during automobile crashes, sporting activities, and
domestic and occupational accidents. To better understand the mechanisms of
mechanically induced changes to joint tissues our laboratory has developed an in vivo
animal model (Newberry et aL, 1998a). Brieﬂy, this involves impacting the right
patellofemoral joint of Giant Flemish rabbits using a rigid impact interface directed
slightly medial with a subfracture blunt impact. At 12 months post-impact the
retropatellar cartilage is ﬁssured and softened compared to an unimmcted control
population (Newberry et al., 1998a). The impacted limb also has a thickened subchondral
plate compared to the unimpacted controls (Newberry et al., 1998a). Using a
computational model of the rabbit patella with boundary conditions obtained from
pressure sensitive ﬁlm, high impact-induced shear stresses were associated with acute
ﬁssuring and chronic softening of the retropatellar cartilage (Newberry et al., 1998b).
This is in support of a previous study that found high shear stress was the best predictor
of acute cartilage ﬁssuring (Atkinson et al., 1998). Likewise, high shear and tensile
stresses in the underlying subchondral bone corresponded with the chronic thickening of
this tissue (Newberry et al., 1998b).

To better control loading and to examine possible mechanisms of cartilage
damage, studies have been performed on cultured explanted cartilage tissue. These
studies mechanically load the explanted tissue in simple loading conﬁgurations, such as

unconﬁned compression. Previous studies have shown that a single severe impact to the

cartilage can result in acute matrix damage (ﬁssures and increased water content) and
chondrocyte death (Jeffrey et al., 1995; Repo and Finlay, 1977; Torzilli et al., 1999).

Impact interfaces, impact orientation, and impact duration are all important
parameters that could inﬂuence the state of stress and strain in joint tissues and therefore
have an effect on both acute and chronic damages to the tissues. Impact modeling of
articular cartilage nwds to take into account the non-linear stiffening effect of cartilage
(Oloyede et aL, 1992; Atkinson et al., 1998). We currently use an elastic modulus of 20
MPa for cartilage under impact loading (Newberry et al., 1998b). This elevated elastic
modulus was based on matching experimental displacements of cartilage under impact
loading (Atkinson et al., 1998). While there is no ﬂuid ﬂow in cartilage under one second
of loading (Donzelli et al., 1999), Ewers et al., (2000) has shown that rates of loading
under one second can affect the amount of deformation of cartilage. To fully model the
apparent fast relaxation characteristics of cartilage under an impact, a model that
incorporates viscoelasticity is needed (Suh and Bai, 1998).

The overall aims of the current study were threefold. First, to use the established
animal model to examine the inﬂuence of impact interface, impact direction, and impact
duration on chronic alterations in joint tissues, and to correlate impact-induced stresses
with these changes. Second, to use a cartilage explant system to examine the inﬂuence of
impact duration on matrix damage and cell death, and attempt to correlate the state of
stress and strain in the tissue and cells with these damages. Third, determine a method to
estimate the non-linear viscoelastic material properties of cartilage from in situ

indentation testing for the modeling of impact experiments. Future studies will be able to

use these material properties of cartilage for more appropriate modeling of impact

scenarios.

 

Chapter 1 documents changes in retropatellar cartilage and subchondral bone in
our established animal model out to 36 months post-impact. This study shows that this
animal model does not develop an end-stage disease by 36 months, but that there are still
progressive alterations going on in the joint tissues, such as increased thickening of
subchondral bone, and a thinning of the articular cartilage.

Chapter 2 uses computational models to examine the inﬂuence of soﬁened
articular cartilage on the state of stress in the underlying bone. This study also examines
the inﬂuence of the material properties of the subchondral bone on the state of stress in
the overlying cartilage. This study shows that in our animal model there does not seem to
be any mechanical interaction between retropatellar cartilage and underlying subchondral
bone.

Chapter 3 examines the effects of using a deformable impact interface to change
the load distribution on chronic alterations in joint tissues in our animal model. This
study ﬁnds that a deformable interface reduced the impact-induced stresses in the
underlying bone, mitigating any chronic thickening out to 12 months post-impact.
However, the impact-induced stresses in the cartilage are still elevated leading to a
chronic softening and ﬁssuring of retropatellar cartilage.

Chapter 4 examines the effects of impacting the patellofemoral joint with a central
oriented impact load. This study documents that there were no chronic alterations in the

joint tissues out to 12 months post-impact. The impact-induced shear stresses are reduced

in the articular cartilage, but increased in the subchondral bone. However, there is also an
increase in the compressive mean stress, which likely protected the underlying
subchondral bone.

Chapter 5 examines the effect of impact duration on the acute injuries in cadaver
knee impacts and chronic alterations in our animal model. This study indicates that a
higher rate of loading results in more ﬁ'acture and nonﬁacture injuries in the human
cadaver impacts than low rates of loading. In support of this, there is more chronic
damage to both the cartilage and underlying bone in our animal model with the high
versus low rate of loading.

Chapter 6 examines the inﬂuence of a possible chondroprotective drug
(polysulphated glycosaminoglycans) on joint tissues using our animal model. This study
suggests that the drug did not have any effect on thickening of the underlying
subchondral bone. However, the drug mitigated the usual decrease in cartilage moduli
post-impact, suggesting a possible chondroprotective effect.

Chapter 7 examines the inﬂuence of loading rate on matrix damage and
chondrocyte death using explanted cartilage plugs that were loaded in unconﬁned
compression. This study indicates that a high versus low rate of loading results in
signiﬁcantly more matrix damage. In contrast, the low versus high rate of loading results
in signiﬁcantly more chondrocyte death.

Chapter 8 examines different computational models of cartilage and attempted to
correlate high shear stresses with matrix damage and high cell strains with cell death.
This study suggests that the isotropic model with depth-dependent material properties

best associated areas of high stress with matrix damage and predicted the experimental

lateral expansion proﬁles documented in the recent literature. The study also suggestes
that the extent of cell strain, and thus cell death in each layer, was most inﬂuenced by the
shape of the cell, not the depth-dependent material properties of the matrix.

Chapter 9 describes a method to estimate the non-linear viscoelastic material
prOperties of cartilage using in situ indentation testing. This study indicates that the short-
time relaxation (less than 1 second) can be modeled by a linear viscoelastic response.
This study also shows that the four material constants of the strain energy function can be
approximated from the inﬁnitesimal elastic constants, and the unrelaxed response of a
spherical indentor indenting down to 50 % of the cartilage thickness. The method was
validated using bovine cartilage in unconﬁned compression. The study suggests that the
material properties determined from the above procedure can predict the experimental

loadwithin l7 %.

REFERENCES

l.

Atkinson, TS, RC Haut, and NJ Altiero. (1998) Impact-induced ﬁssuring of articular
cartilage: an investigation of failure criteria J Biomech Eng. 120: 1 8 1-187.

Blanco FJ, Guitian R, Vazquez-Martul E, de Toro FJ, Galdo F (1998) Osteoarthritis
chondrocytes die by apoptosis. A possible pathway for osteoarthritis pathology.
Arthritis Rhuem 41 :284-289.

Brandt KD, Mankin HJ, Shulman LE (1986) Workshop on etiopathogenesis of
osteoarthrits. J Rhuematol 13:1 126-1 160.

Chapchal, G. (1978) Posttraumatic Osteoarthritis after Injury of the Knee and Hip
Joint. Reconstructive Surgery T raumatology 16, pp. 87-94.

Donzelli, PS, RL Spilker, GA Ateshian, and VC Mow. (1999) Contact analysis of
biphasic transversely isotropic cartilage layers and correlations with tissue failure. J
Biomech. 32:1037-1047. 4

Ewers, BJ, D Dvoracek-Driskna, MW Orth, and RC Haut. (2000) The extent of
matrix damage and chondrocyte death in mechanically traumatized articular cartilage
explants depends on rate of loading. J Orthop Res. in press.

Felson DT. (1990) The epidemiology of knee osteoarthritis: results from the
Framingham osteoarthristis study. Sem Arth and Rheum, 20:42-50.

Hamerman, D. (1989) The biology of osteoarthritis, New Eng Journal of Med.
320:1323-1330.

Jeffrey JE, Gregory DW, Aspden RM (1995) Matrix damage and chondrocyte
viability following a single impact load on articular cartilage. Arch Biochem Biophys
322:87-96.

10. Kim, HA, YJ Lee, SC Seong, KW Choe, YW Song (2000) Apoptotic chondrocyte

death in human osteoarthritis. J Rheumatol. 27(2):455-462.

ll. Newberry WN, Mackenzie CD, Haut RC (1998a) Blunt impact causes changes in

bone and cartilage in a regularly exercised animal model. J Orthop Res 16:348-354.

12. Newberry WN, Garcia JJ, Mackenzie CD, Decamp C, Haut RC (1998b) Analyses of

acute mechanical insult in an animal model of post-traumatic osteoarthrosis. J
Biomech Eng, 120:704-09.

13. Oloyede A, Flachsmann R, Broom N (1992) The dramatic inﬂuence of loading

velocity on the compressive response of articular cartilage. Conn Tiss Res, 27:211-
224.

l4. Radin EL, Burr DB, Fyhrie D, Brown TD, Boyd RD (1990) Characteristics Of joint
loading as it applies to osteoarthrosis. In: Mow, VC, Ratcliffe, A., Woo, S.L-Y.
(Eds), Biomechanics of Diarthrodial Joints. Springer-Verlag, New York, pp. 437-
451.

15. Repo RU, Finlay JB (1977) Survival of articular cartilage after controlled impact. J
Bone Joint Surg [Am] 59: 1068-1076.

16. Simon WH, Richardson S, Herman W, Parsons JR, Lane I (1976) Long-term effects
of chondrocyte death on rabbit articular cartilage in vivo. J Bone Joint Surg [Am]
58:517-526.

17. Sub, JK and S Bai (1998) Finite element formulation of biphasic poroviscoelastic
model for articular cartilage. J Biomech Eng 120: 195-201.

18. Torzilli PA, Grigiene R, Borrelli Jr. J, Helfet DL (1999) Effect of impact load on
articular cartilage: Cell metabolism and viability, and matrix water content. J
Biomech Eng 121 :433-441.

CHAPTER 1

LONG TERM CHANGES IN RABBIT RETROPATELLAR CARTILAGE AND
SUBCHONDRAL BONE AFTER BLUNT IMPACT LOADING OF THE
PATELLOFEMORAL JOINT

Benjamin J Ewers, Brian T Weaver, Eric T Sevensma, and Roger C Haut

ABSTRACT

Animal models of 0A are useful for the study of early changes in the joint tissues
that may eventually lead to an end stage disease. Our laboratory has developed a joint
trauma model using a single blunt impact to the patellofemoral joint of rabbits and has
documented softening of retropatellar cartilage and thickening of underlying bone out to
12 months post-trauma. The objective of this study was to examine the changes in joint
tissues with this animal model out to 36 months post-impact. Forty-nine Flemish Giant
rabbits were impacted on the right patello femoral joint and sacriﬁced at one of ﬁve times:
4.5, 7.5, 12, 24, and 36 months post-impact. The mechanical properties of cartilage and
thickness of subchondral bone were documented. At 36 months the animals had not
developed an end stage disease. A reduction in the compressive modulus of traumatized
retropatellar cartilage, based on indentation testing, occurred at 4.5 months and remained
constant to 36 months. The thickness of the cartilage, however, decreased from baseline
over time resulting in overall structural mechanical stiffening of the retropatellar
cartilage. Furthermore, the traumatized cartilage had a signiﬁcant increase ﬁom baseline
in its ﬂuid permeability over time based on data from mechanical tests. In addition, the

impacted patellae had progressive changes with a signiﬁcant increase in thickness of the

subchondral bone over time from baseline. In conclusion, since there is a need to more
fully understand the mechanisms of a post-traumatic 0A, or a process of trauma with
healing, this animal model may be of interest. While not developing into an end stage
disease by 36 months, progressive changes were still occurring, so the model may be
useful for the study of the mechanisms leading to a possible post-traumatic disease, early
diagnostic protocols, and disease mitigating interventions. On the other hand, the model
may also be one of trauma with a subsequent healing response over time.
INTRODUCTION

Osteoarthritis (0A) could cost society approximately 54 billion dollars a year
(Yelin, 1998). Epidemiology studies of human 0A have advanced our understanding of
this disease; however, they do not provide much data on the mechanisms and temporal
progression of joint degeneration. This is particularly true in the early stages of the
disease. Experimentally, animal models provide a tool to study early changes in joint
tissues that may eventually lead to an end stage disease. Many animal models of 0A have
been described using various species, various means of inducing joint degradation, and
resulting in varying degrees of end stage disease. These models have included
spontaneous disease (Okabe, 1989; Alexander, 1992), chemically induced disease models
(Coulais et aL, 1984; Van der Kraan et al., 1989), and mechanical or surgical models
(Pond and Nuki, 1973; Ehrlich et al., 1975). Hallmark indicators of an end stage disease
are the total loss of cartilage and exposure of bone, sclerosis of the underlying
subchondral bone, and formation of osteophytes around the joint periphery (Brandt et al.,

1986)

10

One surgical model, menisectomy in canines, has shown decreases in the tensile
modulus of femoral surface-zone articular cartilage 12 weeks post-surgery (Elliot et al.,
1999). This is signiﬁcant because diseased human cartilage has decreased material
stiffness (or modulus) in compression, tension, shear loading, and enhanced swelling
(Akizuki et aL, 1987; Armstrong et aL, 1982; Maroudas and Venn, 1977; Setton et al.,
1999). A more widely used surgical model is transection of the anterior cruciate ligament
(ACL) in canines, which has been shown to result in a decreased compressive modulus,
an increased ﬂuid permeability, and a slight thickening of articular cartilage on the
medial tibial plateau three months post-surgery compared to controls (Setton et al., 1994).
Along with these mechanical alterations observed in cartilage, ACL transection has been
shown to increase subchondral bone thickness on the medial tibial plateau over time
(Dedrick et aL, 1993). While this model appears to be self-limiting at two years, it
develops an end-stage disease at 54 months (Brandt et al., 1991). A mechanical model
using repetitive impulse loading of the rabbit tibiofemoral joint has shown an increase in
tetracycline labeling and bone formation in the underlying subchondral plate before
cartilage degradation (Radin et al., 1984). Nine weeks of repetitive loading, followed by
24 weeks of cage rest, leads to a complete loss of cartilage on the weightbearing surface
of the tibia and the formation of osteophytes (Radin et al., 1990).

‘ Previous knee injuries lead to an increased risk for the subsequent development of
OA (Felson, 1990). A recent analysis of automobile accidents has shown that while safety
features signiﬁcantly reduce fatalities, lower extremity injuries are now more frequent
(Otte, 1996; Pattirnore et aL, 1991). The current criterion for certifying new automobiles

is based on a bone fracture in human cadavers. For the knee-femur-hip complex the

11

femoral loads cannot exceed 10 RN in a simulated crash. However, a majority of injuries
to the knee are non-fracture, such as contusions and lacerations (Atkinson et al., 1997).
Clinically, such non-fracture injuries can initiate diseases such as OA (Upadhyay et a1.
1983; Chapchal, 1978). However, even major joint injuries with bone ﬁ‘acture take 2-5
years for development of disease symptoms. Less severe injuries may take 10 or more
years for degenerative changes to be diagnosed (Wright, 1990).

A number of research animal models have been developed to study the
pathogenesis of post-traumatic 0A An impact model using a single transarticular load to
the canine patellofemoral joint has shown surface lesions on retropatellar cartilage and
step-off fractures across underlying calciﬁed cartilage (Thompson et al., 1991). One year
after impact these ﬁ'actures heal, and the density of cartilage proteoglycans appears
normal (Thompson et al., 1993). The disease process looks to be self-limiting, but the
model may need to be carried out over a longer time frame as in the ACL transection
model. One difﬁculty, however, in such studies is the cost of maintaining these large
animals over long periods of time (Brandt et al., 1991).

Our laboratory has been developing a small animal model, Giant Flemish rabbits,
that involve less research costs than the canine model. A single blow to the
patellofemoral joint leads to histological changes and decreased stifﬁress of retropatellar
cartilage, along with thickening of the underlying subchondral bone (Newberry et al.,
1998). To date post-traumatic changes in the joint have been documented for only 12
months. The current study was conducted to document the more long-term effects of a
single blunt insult to the patellofemoral joint on retropatellar cartilage and its underlying

subchondral bone using this animal model.

12

METHODS AND MATERIALS

Forty-nine mature Flemish Giant rabbits (4.8 i 0.9 kg, 6-8 months) were used in
the study. Random checks in prior pilot studies displayed closed epiphysis for animals
weighing 5 kg at 6 — 8 months of age (Newberry et al., 1998). The blunt impact
experiments have been described in previous studies (Newberry et al., 1998). Brieﬂy, a
1.33 kg mass with a ﬂat 1 in. diameter aluminum impact interface was dropped ﬁom 0.46
In onto the right hind limb and impacting the medial aspect of the joint, which was ﬂexed
approximately 120° (Figure 1). During the impact the animals were maintained at a
surgical plane of anesthesia using Isoﬂurane and oxygen (2% Isoﬂurane). The mass was
stopped electronically after the first impact, preventing multiple impacts. A load
transducer (Sensotec, Columbus OH: model 31/1432, 500 lb capacity) recorded the
impact load at a 10 kHz sampling rate.

After impact, all animals received one injection of Butorphenol for early post-
surgical pain. After a ﬁve-day period of rest, a daily exercise program was initiated for all
animals. This program consisted of ten minutes of exercise, ﬁve days a week, on a
treadmill running at 0.3 mph (Oyen-Tiesma et al., 1998). The exercise schedule lasted the
entire length of the study. A certiﬁed veterinary technician was devoted to this animal
colony throughout the duration of the study (J A). The animals were housed in individual
cages (48” x 24” x 19”) when nOt being exercised. This study was approved by the MSU
All-University Committee on Animal Use and Care.

Eight animals were sacriﬁced at each of the following timepoints: 0, 4.5, 7.5, 12,
and 36 months post-impact. Nine animals were sacriﬁced twenty-four months after

impact. Immediately after sacriﬁce the patellae were excised for mechanical indentation

l3

tests on the retropatellar cartilage, as described previously (Ewers et al., 2000a). The
patellae were immersed in a room-temperature phosphate buffered saline bath during the
indentation tests. The tests were performed with a custom-built instrument that used a
computer controlled stepper motor (Physik Instruments, .Waldbronn, Germany : model
M-168.30) to indent the cartilage. A ﬁxture, which held the patella with its retropatellar
surface facing the indenter, allowed X-Y-Z movement to position the site of indentation.
A camera mount (Bogen, Ramsey, NJ) attached to the base of the ﬁxture allowed rotation
of the patella to insure that the indentation was performed normal to the cartilage surface.
A 1 mm diameter ﬂat, non-porous probe was pressed into the cartilage 0.1 mm in 30 ms
and maintained for- 150 seconds. The resistive loads were measured (Data Instruments,
Acton MA : model JP - 25, 25 1b capacity), ampliﬁed, and collected at 1000 Hz for the
ﬁrst second and 20 Hz thereafter. After the tissue was allowed to recover for 5 minutes,
the test was repeated with a 1.5 mm diameter, ﬂat non-porous probe in the same location.
The probe was then replaced with a needle that was slowly pressed into the cartilage to
determine the thickness of cartilage at that location. The above procedure was performed
at two sites adjacent to surface lesions that are typically produced on the lateral
retropatellar facet during blunt impact to the patellofemoral joint at this severity of
impact (Newberry et al., 1998).

Mechanical data was analyzed using a biphasic (poroelastic) model having a
transversely isotropic (TI) solid structure (Garcia, 1998). Brieﬂy, the model assumed
isotropy in the plane of the cartilage and Poisson’s ratio in the plane (V12) was set equal to
zero. The remaining four elastic parameters and the two permeability measures, namely:

a thickness direction modulus (E33), an in-plane modulus (E;;), a shear modulus (G13), a

14

Poisson’s ratio (v13), 3 thickness direction permeability (kg), and an in-plane permeability
(k1), were then computed using the Garcia curve ﬁtting algorithm (Garcia, 1998).

After mechanical tests on each patella, they were stained with India ink to
highlight surface lesions and photographed for permanent documentation. The patellae
were then placed in 10% buffered formalin for a week, and decalciﬁed in 20% formic
acid for another week. Tissue blocks were cut medial to lateral across the patellae in the
area known to produce high contact pressures during blunt impact to the patello femoral
joint at this ﬂexion angle (Haut et al., 1995). Tissue blocks were processed in parafﬁn
and six sequential sections, 8 [.1 thick, were prepared for analysis. The sections were
stained with Saﬁ'anin O-Fast Green and examined under light microscopy at 12-400
power. The thickness of the subchondral bone plate underlying retropatellar cartilage was
measured at 25x for all six sections with a calibrated eye-piece at the center of each facet
and the mid-line of the patella by three independent investigators, using established
methods (Newberry et al., 1998). To quantify the degree of cartilage matrix damage, the
gross photographic images were digitally scanned (ScanJet 6300C, Hewlett Packard,
Singapore) and the total length of the surface lesions was measured using image software
(Sigma Scan, SPSS Inc., Chicago, IL). The number and average depth of these surface
ﬁssures were also measured at 40x with the calibrated eye-piece for three sections for
each patella by a single investigator (ES).

Signed Rank tests were used for comparisons between impacted and contralateral
unimpacted limbs for the number of surface lesions and the average lesion depth. Paired
t-tests were used to compare impacted to unimpacted contralateral limbs for all other

parameters. ANOVAs with SNK post-hoc tests were used to determine statistical

15

differences in parameters between timepoints. ANOVAs with SNK post-hoe tests were
also used to determine differences in subchondral bone thickness at each location
between readers. A Spearman Rank Order correlation was performed to examine changes
over time for the number of ﬁssures and their average depth. A Pearson Product Moment
correlation was used for the remaining parameters. Statistical signiﬁcance was set at p <
0.05 for all tests.

RESULTS

Daily observations during treadmill exercise by the certiﬁed veterinary technician
indicated that the rabbits did not favor the impacted limb. The 24 and 36 months post-
impact groups had 4 and 5 animals, respectively, develop ulcerative pododermatitis (sore
hocks). Sore hocks, which is treated with disinfecting ointments, is a common condition
found in Giant Flemish rabbits associated with age, weight, and/or living quarter
conditions. The 24 and 36 months post-impact groups had 6 and 4 animals, respectively,
require antibiotics (Baytril, 5.0 mg/kg enroﬂoxacin) for an average of two weeks, for the
treatment of various complications caused from chronic pasteurellosis. This is common
for non-patho gen free animals.

Gross visual examination of the impacted joints revealed surface lesions on all
impacted patellae typically occurring along the center ridge of the patella on the lateral
facet? (Figure 2A-B). Surface lesions on the contralateral unimpacted retropatellar
cartilage were documented for three, two, and two patellae at 12, 24, and 36 months,
respectively. At 36 months, three impacted patellae had osteophytes develop on the
medial or lateral sides, while osteophytes were not documented on any unimpacted

patellae (Figure 2C).

16

The results of mechanical indentation tests revealed that there were no signiﬁcant
diﬁ‘erences between limbs in any mechanical parameter at time zero. Furthermore, there
were no time-dependent changes from baseline in any mechanical parameter for the
unimpacted sides. Signiﬁcant reductions in the E1; and E33 moduli were, however,
documented at all timepoints other than zero for the impacted retropatellar cartilage
compared to the contralateral, unimpacted side (Table 1). Interestingly, these reductions
occurred early (4.5 months) and remained constant at approximately 35 and 28% of
contralateral values for E11 and E33, respectively. There was a signiﬁcant reduction in E“
compared to baseline values at all timepoints except 7.5 months. The reduction in
cartilage moduli was accompanied by a signiﬁcant increase in the k; and k3
permeabilities for the traumatized cartilage versus the contralateral unimpacted side at all
timepoints except zero. While the increase in the k; permeability remained constant over
time, approximately 88% of the contralateral values, the k; permeability signiﬁcantly
increased with time at 0.18 m4/Ns x 10'” per month. The k1 permeability was also
signiﬁcame increased at 36 months compared to baseline values.

An analysis of the histological sections revealed an average variance of 1.0% for
subchondral bone thickness between sections, so an average value was calculated at each
location for each patella for each observer. There was no signiﬁcant differences between
the three independent observers in measured subchondral bone thickness at any location,
with an average variance of 2.3% between observers. Therefore, an average value was
calculated for each patellae at each location (medial, central, lateral) from the three
independent readers and used in subsequent statistical analysis. At time zero and 4.5

months post-impact there were no signiﬁcant differences in the thickness of the

17

subchondral plate for the impacted patellae compared to the unimpacted limb under any
region (Table 2). At 7.5 months the thickness of the subchondral bone plate under the
central region and the medial facet were signiﬁcantly greater on the impacted patellae
compared to those areas on the contralateral unimpacted limbs. At 12, 24, and 36 months
the thickness of the subchondral bone under the central, the medial, and the lateral
regions were signiﬁcantly greater for the impacted patellae compared to those locations
on the contralateral unimpacted limbs. There were also signiﬁcant increases in
subchondral plate thickness with time for the impacted patellae at all locations, and for
the contralateral unimpacted patella in the medial and lateral facets. For the unimpacted
patellae the thickness of the subchondral bone increased from 0.002 and 0.001 mm per
month for the lateral and medial regions, respectively. For the impacted patellae,
however, the thickness of the subchondral bone plate increased 0.013, 0.007, and 0.007
mm per month for the central, lateral, and medial regions, respectively. These differences
in the rate of the subchondral bone thickening were statistically signiﬁcant at each
location. Finally, there was a signiﬁcant increase in subchondral bone thickness at 36
months compared to baseline values at each location.

From the biomechanical testing protocol the thickness of retropatellar cartilage
could be recorded over time. The traumatized retropatellar cartilage on the lateral facet
was thicker than that of the contralateral unimpacted cartilage and baseline cartilage at
4.5 months post-trauma (Table 3). This thickening was followed by signiﬁcant decreases
in thickness for the traumatized cartilage compared to the contralateral unimpacted sides
of 15 and 48% at 24 and 36 mOnths, respectively. While there was no signiﬁcant time-

dependent change over baseline in articular cartilage thickness for the unimpacted limb,

18

the thickness of the traumatized retropatellar cartilage signiﬁcantly decreased from the
baseline level with time. Cartilage thickness decreased at 36 months compared to the
baseline cartilage thickness.

Gross photographs of the retropatellar cartilage surface were studied and revealed
signiﬁcantly greater ﬁssure lengths on impacted patellae versus contralateral unimpacted
patellae at all timepoints post-trauma (Table 3). There was, however, no signiﬁcant time-
dependent change over baseline in the total length of surface ﬁssures on retropatellar
cartilage for impacted or unimpacted limbs.

Analysis of the histological sections revealed lesions in traumatized cartilage at
all timepoints (Figure 3). There was a signiﬁcant increase in average surface ﬁssure depth
and number of surface lesions for the impacted retropatellar cartilage compared to the
contralateral unimpacted cartilage at 4.5, 7.5, 12, 24, and 36 months post-impact (Table
3). The average ﬁssure depth and number of ﬁssures for the unimpacted patellae were not
signiﬁcantly different than zero at any timepoint. There were no time-dependent changes
over baseline in the average depth of the surface lesions or the number of surface fissures
for either the impacted or unimpacted limb.

DISCUSSION

The current experimental study was undertaken to document the long-term effects
of a single blunt insult to the rabbit patellofemoral joint on retropatellar cartilage and its
underlying subchondral bone plate. While, the study showed no evidence of full thickness
loss of cartilage, except in one animal at 24 months, there was a signiﬁcant thinning.
Furthermore, there was a progressive increase in the permeability of retropatellar

cartilage and the thickness of underlying subchondral bone plate with time post-trauma

l9

versus baseline. The elastic moduli of the cartilage, however, did not signiﬁcantly
decrease over time. Since the cartilage was shown to thin, these effects indicate that the
cartilage became structurally stiffer with time post-trauma. This could have resulted in an
altered stress state in the underlying subchondral bone, leading to its observed
remodeling. This suggestion, however, would be in contrast to a recent mathematically
simpliﬁed analysis of the rabbit patellofemoral joint which showed that a change in
material properties of cartilage would not likely alter the state of stress in the underlying
bone, and vise versa (Ewers et aL, 1998). Furthermore, recent studies with this animal
model suggest that damage to the retropatellar cartilage and the underlying subchondral
plate may be initiated by acute overstressing, and that chronic softening of the
retropatellar cartilage occurs independent of subchondral bone thickening (Ewers et al.,
2000b). That study, however, was limited to 12 months post-impact. Thickening of
subchondral bone underlying cartilage may be needed more for the progression to an end
stage disease (Dedrick et al., 1993).

Our results compare well to earlier studies with this animal model in which the
mechanical analysis of the cartilage was performed with an elastic versus a biphasic
model. Newberry et a1. (1998) showed a 42 and 25% reduction in the instantaneous
modulus (GU) and relaxed modulus (GR), respectively, of traumatized cartilage compared
to its contralateral unimpacted Mb at 3, 6, and 12 months post-trauma In the TI biphasic
model analysis, changes in GU and GR are reﬂected largely by changes in E“ and E33,
respectively (Garcia, 1998). In the current study B“ of the impacted cartilage was
reduced compared to the unimpacted side by approximately 35% at 4.5, 7.5, and 12

months, respectively. E3; was reduced by approximately 26% on the impacted versus

20

contralateral sides at these same timepoints. Newberry et al. (1998) also documented a
37, 29 and 72% increase in the thickness of underlying subchondral bone on the impacted
patellae versus the contralateral unimpacted side at the center region for 3, 6, and 12
months post-trauma, respectively. This is similar to the current study which showed 15
and 46% increases in the thickness of subchondral bone underlying impacted cartilage
versus the unimpacted side at 7.5 and 12 months, respectively, at the center region of the
patella

The mechanical results from the biphasic analysis of retropatellar cartilage in the
current study also compare well to other animal models of OA. Setton et a1. (1994) using
a canine ACL-transection model, documented a 25% decrease in the aggregate and shear
moduli of tibial cartilage and a 48% increase in its permeability and a thickening of
cartilage at “cover ” sites three months post-surgery. In the current study we
documented; 28% reductions in the elastic moduli of traumatized retropatellar cartilage, a
77% increase in its permeability, and a 15% increase in its thickness versus the
contralateral unimpacted cartilage at 4.5 months post-trauma

The reductions in the elastic moduli and increases in tissue permeability of
retropatellar cartilage with trauma may reﬂect changes in its microstructure. The
increased permeability of the tissue may reﬂect damage to the matrix that would likely
increase tissue hydration. In fact, Setton et a1. (1994) has documented a positive
correlation between tissue permeability and its water content. This is signiﬁcant since
swelling or tissue edema has been documented in osteoarthritic human cartilage
(Maroudas et al., 1977). In the current study, the E33 modulus is largely determined by

the equilibrium response of the tissue (Garcia, 1998). Studies have shown that the

21

equilibrium response of cartilage is largely controlled by the content and integrity of
tissue proteoglycans (Jurvelin et al, 1988). The in-plane modulus (Eu) has been shown
to largely depend on the “instantaneous” response of cartilage. A study indicated that this
characterization of cartilage may be reﬂective of the integrity of collagen network in the
tissue (Mizrahi et al., 1986).

As previously stated above, the current impact model indicate early mechanical
changes in the retropatellar cartilage similar to those that have been documented in the
canine ACL-transection model (Setton et al., 1994). The increase in subchondral bone
thickness is also similar. Dedrick et a1. (1993) documented a trend for an increased tibial
subchondral bone thickening on the ACL-transected side compared to the contralateral
side at 18 and 54 months post-surgery, along with a statistical increase in thickness from
three to 54 months post-surgery. At 24 to 36 months, the current study showed that in this
impact model the joints did not develop an end stage disease. This is similar, however, to
the ACL-transection model that showed only minimal degenerative changes in articular
cartilage at 23 months post-surgery (Marshall and Olsson, 1971). It took 54 months post-
surgery for this animal model to lead to a full loss of cartilage thickness (Brandt et al.,
1991). This may suggest that we have to take the animals out ﬁrrther post-impact to
develop an end stage disease. On the other hand, this model may be one of impact trauma
and a subsequent healing response without development of an end stage disease.

Mazieres et al. (1987) using a similar impact model with New Zealand rabbits,
have documented full thickness loss of cartilage at six months post-impact. A possible
explanation for a difference in that study versus the current study is the use of a different

breed of rabbit, and a higher impact energy (10 J versus 6 J in our model). Previous

22

studies with our animal model have shown that a high severity energy impact leads to a
softening of the retropatellar cartilage and thickening of the underlying subchondral bone,
while lower energies (and loads) do not result in any changes in joint tissues (Newberry
et al., 1998b). Therefore, while causing some changes to the patellofemoral joint tissues,
our impact energy (load) may not be as severe as the impacts in the Mazieres study so as
to cause rapid joint degradation. Future studies with our model would be needed to see if
an even more severe cases of blunt trauma, short of a gross fracture, would result in an
accelerated disease process. In an impact model using canines, Thompson et a1. (1991)
documented multiple extensive fractures though the zone of calciﬁed cartilage and the
subchondral bone. These ﬁactures heal one year post-impact and the density of cartilage
proteoglycans appears normal. While the endpoint of the Thompson et a1. (1991) level of
joint trauma is yet unknown, these data may suggest that rapid progression to an end
stage disease may only occur for a select range of joint impact intensities. If conﬁrmed in
future studies, this might suggest that the time frame for progressive disease may
signiﬁcantly depend on the intensity of acute trauma.

A limitation of the current study was the lack of a control population. Since for
the contralateral unimpacted limb no parameter except subchondral bone thickness
changed with time versus baseline, the chosen protocol may be justiﬁed given the
expense of maintaining these animals over such a long time frame. Furthermore, the
unimpacted side at time zero was not statistically different than the unimpacted limb at
any timepoint for any parameter, except subchondral bone thickness. At the current
severity of impact, primarily because of the life expectancy of the rabbits using the

current experimental protocol, this model may not be appropriate to study end GA. More

23

severe insults to the patellofemoral joint need to be examined to determine if this would
lead to a more progressive disease model

In conclusion, since there is the need to more ﬁrlly study joint injury after a blunt
trauma, the current small animal model may be of interest. While the current model did
not develop an end stage disease at 36 months post-impact, progressive changes are still
occurring in the retropatellar cartilage and the underlying subchondral bone. This model
may then be quite useful for the study of some early diagnostic protocols and/or the
development of drugs to mitigate or limit early stages of disease post-trauma (Ewers and
Haut, 2000). On the other hand, these data may also be indicative of a trauma model with
subsequent healing Over time. The study of the mechanisms of acute damage to tissues
causing progressive degenerative changes, even with or without healing may also have
utility in the design of “safer” impact interfaces in automobiles.
ACKNOWLEDGEMENTS

This study was supported by a grant for the Centers for Disease Control and
Prevention (R49/CCR503607). Its contents are solely the responsibility of the authors and
do not necessarily represent the ofﬁcial views of the Centers for Disease Control and
Prevention. The authors wish to gratefully acknowledge William Newberry for assistance
in the experimental impact protocol, Jane Walsh for her preparation of the histological
sections, Jean Atkinson for her care of the animals and Cliff Beckett for technical

assistance in the experimental phase of these studies.

24

REFERENCES

l.

Akizuki S, Mow VC, Muler F, Pita JC, Howell DS. (1987) Tensile properties of
human knee joint cartilage. II. Correlations between weight bearing and tissue
pathology and the kinetics of swelling. J Orthop Res, 5:173-186.

Alexander JW. (1992) The pathogenesis of canine hip dysplasia Vet Clin North
America: Small Animal Practice, 22:503-1 1.

Armstrong CG, Mow VC. (1982) Variations in the intrinsic mechanical properties of
human articular cartilage with age, degeneration, and water content. J Bone Joint
Surg [Am], 64:88-94.

Atkinson PJ, Garcia JJ, Altiero NJ, Haut RC. (1997) The inﬂuence of impact
interface on human knee injury: implications for instrument panel design and the
lower extremity injury criterion. 41" Stapp Car Crash Conference, 167-180.

Brandt Kd, Mankin HJ, Shulman LE. (1986) Woprkshop on etiopathogenesis of
osteoarthritis. J Rheum, 13:1126-60.

Brandt KD, Myers SL, Burr D, Albrecht M. (1991) Osteoarthritic changes in canine
articular cartilage, subchondral bone, and synovium ﬁfty-four months after
transection of the anterior cruciate ligament. Arth and Rheum, 34: 1560-70.

Chapchal G. (1978) Posttraumatic osteoarthritis after injury of the knee and hip joint.
Reconstr Surg, 16:87-94.

. Coulais Y, Marcelon G, Cros J, Guiraud R. (1984) An experimental model of

osteoarthritis. II: biochemical study of collagen and proteoglycans. Pathol Biol,
32:23-8.

Dedrick DK, Goldstein SA, Brandt KD, O’Connor BL, Goulet RW, Albrecht M.
(1993) A longitudinal study of subchondral plate and trabecular bone in cruciate-
deﬁcient dogs with osteoarthritis followed up for 54 months. Arth and Rheum,
36:1460-67.

10. Ehrlich MG, Mankin JH, Hones H, Grossman A, Crispen C, Ancona D. (1975)

Biochemical conﬁrmation of an experimental osteoarthritis model. J Bone Joint Surg,
57 :392—6.

11. Elliot DM, Guilak F, Vail TP, Wang JY, Setton LA. (1999) Tensile properties of

articular cartilage are altered by meniscectomy in a canine model of osteoarthritis. J
Orthop Res,17(4):503-8.

25

12. Ewers BJ, Newberry WN, Garcia JJ, Haut RC. (1998) Alterations in the mechanical
properties of bone underlying articular cartilage in a traumatized joint. Proc 44th
Annual Meeting Orthopaedic Research Society, 459.

13. Ewers BJ, Jayaraman VM, Banglmaier RF, Haut RC. (20003) The effect of loading
rate on the degree of acute injury and chronic conditions in the knee after blunt
impact. STAPP Car Crash Conference, 44:299-314.

14. Ewers BJ, Newberry WN, Haut RC. (2000b) Chronic softening of cartilage without
thickening of underlying bone in a joint trauma model J Biomech, 33:1689-1694.

15. Ewers BJ, Haut RC. (2000) Polysulphated glycosaminoglycan treatments can
mitigate decreases in stiffness of articular cartilage without thickening of underlying
bone in a joint trauma model. J Orth Res, 18:756-761.

16. Felson DT. (1990) The epidemiology of knee osteoarthritis: results from the
Framingham osteoarthristis study. Sem Arth and Rheum, 20:42-50.

17. Garcia JJ. (1998) A transversely isotropic hypo-elastic biphasic model of articular
cartilage under impact loading. Michigan State University, Ph.D. dissertation.

18. Haut RC, Ide TM, De Camp CE. (1995) Mechanical responses of the rabbit patello-
femoral joint to blunt impact. J Biomech Eng, 117:402-08.

l9. Jurvelin J, Saamanen A-M, Arokoski J, Helminen HJ, Kiviranta I, Tammi M. (1988)
Biomechanical properties of the canine knee articular cartilage as related to matrix
proteoglycans and collagen. Eng Med, 17:157-62.

20. Maroudas A, Venn M. (1977) Chemical composition and swelling of normal and
osteoarthrotic femoral head cartilage. II. Swelling. Ann Rheum Dis, 36:399-406.

21. Marshall JL, Olsson S-E. (1971) Instability of the knee: a longterm experimental
study in dogs. J Bone Joint Surg [Am], 53: 1561-70.

22. Mazieres B, Blanckaert A, Thiechart M. (1987) Experimental post-contusive
osteoarthritis of the knee: quantitative microscopic study of the patella and the
femoral condyles. J Rheum, 14:119-21.

23. Mizrahi J, Maroudas A, Lanir Y, Ziv I, Webber TJ. (1986) The “instantaneous”
deformation of cartilage: effects of collagen ﬁber orientation and osmotic stress.
Biorheology, 23:311-30.

24. Newberry WN, Mackenzie CD, Haut RC. (1998) Blunt impact causes changes in
bone and cartilage in a regularly exercised animal model. J Orth Res, 16:348-54.

26

25. Newberry WN, Garcia JJ, Mackenzie CD, Decamp C, Haut RC. (1998) Analyses of
acute mechanical insult in an animal model of post-traumatic osteoarthrosis. J
Biomech Eng, 120:704~09.

26. Okabe T. (1989) Experimental studies on the spontaneous osteoarthritis in C57 black
mice. J Tokyo Med, 47:546-57.

27. Otte D. (1996) Biomechanics of lower limb in'uries of belted car drivers and the
inﬂuence of intrusion and accident severity. 4 Stapp Car Crash Conference, 193-
206.

28. Oyen-Tiesma M, Atkinson J, Haut RC. (1998) A method for promoting regular rabbit
exercise in orthopaedics research. Contemp Topics in Lab Animal Sci, 37:77-80.

29. Pattimore D, Ward E, Thomas P, Bradford M. (1991) The nature and cause of lower
limb injmies in car crashes. 3.5"l Stapp Car Crash Conference, 177-188.

30. Pond MH, Nuki G. (1973) Experimentally induced osteoarthritis in the dog. Ann
Rheum Dis, 32:387—8.

31. Radin EL, Martin RB, Burr DB, Caterson B, Boyd RD, Goodwin C. (1984) Effects of
mechanical loading on the tissues of the rabbit knee. J Orth Res, 2:221-34.

32. Radin EL, Burr DB, Fyhrie D, Brown TD, Boyd RD. (1990) Characteristics of joint
loading as it applies to osteoarthrosis. In: Mow VC, Ratcliffe A, Woo S L-Y, eds.
Biomechanics of diarthrodial joints. Vol 1. New York: Springer-Verlag, pp. 437-51.

33. Setton LA, Mow VC, Muller FJ, Pita JC, Howell DS. (1994) Mechanical properties
of canine articular cartilage are signiﬁcantly altered following transection of the
anterior cruciate ligament. J Orth Res, 12:451-63.

34. Setton LA, Elliott DM, Mow VC. (1999) Altered mechanics of cartilage with
osteoarthritis: human osteoarthritis and an experimental model of joint degeneration.
Osteoarthritis Cart, 7(1):2-14.

35. Thompson RC Jr., Oegema TR Jr., Lewis JL, Wallace L. (1991) Osteoarthritic
changes after acute transarticular load. J Bone Joint Surg, 73-A(7):990-1001.

36. Thompson RC Jr., Vener MJ, Grifﬁths HJ, Lewis JL, Oegema TR Jr., Wallace L.
(1993) Scanning electron-microscopic and magnetic resonance-imaging studies of
injuries to the patellofemoral joint after acute transartricular loading. J Bone Joint
Surg, 75-A(5):704-13.

37. Upadhyay S, Moulton A, Srikrishnamurthy K (1983) An analysis of the late effects

of traumatic posterior dislocation of the hip without fractures. J Bone Joint Surg, 65-
B(2):150-52.

27

38. Van der Kraan PM, Vitters EL, van den Berg WB. (1989) Development of
osteoarthritis models in mice by ‘mechanical’ and ‘metobolical’ alterations in the
knee joints. Arthritis Rheum, 32:S107.

39. Wright V. (1990) Post-traumatic osteoarthritis — a medico-legal mineﬁeld. J Rheum,
29:474-78.

40. Yelin E. (1998) The economics of osteoarthritis. In: Brandt KD,Doherty M,

Lohmander LS, eds. Osteoarthritis. New York: Oxford University Press Inc., pp. 23—
30.

28

TABLES

Table 1: Mechanical data for the retropatellar articular cartilage were analyzed using a
biphasic model having a transversely isotropic solid structure, with isotropy in the plane

of cartilage (1) (Avg i SD).

 

 

 

 

 

 

 

Time LilDb E11 E33 013 ‘013 K1 K3
Point (MPa) (MPa) (MPa) (ml/N310 (ml/N810
M________;L’L___”L_
0 ‘ Imp 6.41:1: 1.21 : 0.19 : 0.09 : 4.86 : 1.87 :
1.52 0.27 0.03 0.03 2.60 0.42
Unimp 6.67 : 1.16 : 0.21 : 0.10 : 4.75 : 2.09 :
2.62 0.37 0.08 0.03 1.72 1.52
4.5 ‘ Imp 4.40 : 0.85 : 0.22 : 0.10 : 7.45 : 5.92 :
232*” 025* 0.05 0.03 3.43 3.75*
Unimp 6.30: 1.15: 0.16 : 0.10: 5.54 : 2.70:
1.83 0.30 0.08 0.02 4.79 1.86
7.5 ‘ Imp 4.93 :1: 0.95 : 0.20 : 0.10 : 5.76 : 3.43 :
1.76* 028* 0.06 0.03 388* 186*
Unimp 6.49: 1.34: 0.18 : 0.11 : 4.31: 1.77:
1.61 0.50 0.05 0.05 2.86 0.91
12 ‘ Imp 2.93 : 0.78 : 0.25 : 0.10 : 6.70 : 4.20 :
094*“ 023* 0.14 0.03 262* 1.92*
Unimp 5.96 : 1.14 : 0.21 : 0.11 : 4.43 : 2.55 :
2.42 0.25 0.14 0.03 2.49 1.93
24‘5 Imp 4.21 : 1.00: 0.17: 0.13: 8.02: 3.19:
123*“ 030* 0.03 0.05 435* 139*
Unimp 6.81: 1.41: 0.19: 0.13: 4.13: 1.77:
1.59 0.20 0.03 0.04 2.15 0.79
36 ' Imp 3.97 : 0.96 : 0.20 : 0.10 : 12.31 : 2.57 :
162*“ 0.21* 0.07 0.02 511*” 1.44+
Unimp 6.01 : 1.23 : 0.17 : 0.11 : 5.94 : 1.42 :
1.83 0.22 0.05 0.03 2.67 0.32

Imp — impacted limb

Unimp — Imimpacted limb

a-n=8

b—n=9

Signiﬁcant increase in k; over time for the impacted cartilage (Pearson Product Moment

Correlation, p <0.05).

+ - signiﬁcantly different frOm contralateral unimpacted limb (paired t-test, p < 0.05)
# - signiﬁcantly different from baseline (time 0) (ANOVA, SNK post-hoe, p <0.05)

29

Table 2: Subchondral bone thickness (mm) was measured at 25x with a calibrated eye-
piece at the central region of the patella and midline of the lateral and medial facets (Avg

 

 

 

 

 

 

 

: SD).
M
(months)
0' Imp 0.54 : 0.26 0.78 : 0.44 0.54 : 0.24
Unimp 0.50 : 0.16 0.74 : 0.26 0.47 : 0.08
4.5 ‘ Imp 0.59 : 0.08 0.77 : 0.13 0.49 : 0.10
Unimp 0.54 : 0.07 0.73 : 0.11 0.51 : 0.07
7.5 ' Imp 0.59 : 0.10 0.81 : 018* 0.60 : 009*
Unimp 0.57 : 0.07 0.69 : 0.19 0.49 : 0.06
12 ' Imp 0.74 : 015* 1.15 : 0.34+ 0.69 : 018*
Unimp 0.60 : 0.12 0.79 : 0.32 0.49 : 0.11
24" Imp 0.71 : 012* 1.16 : 023* 0.67 : 013*
Unimp 0.59 : 0.09 0.68 :1: 0.13 0.54 : 0.07
36 : Imp 0.82 : 017+ “ 1.23 : 0.27+ ” 0.80 : 009* ‘*
Unimp 0.59 : 0.11 0.67 : 0.21 0.58 : 0.14

Imp — impacted limb
Unimp — unimpacted limb

a-n=8
b-n=9

Signiﬁcant increase in thickness at all three locations over time for the impacted patellae

(Pearson Product Moment Correlation, p <0.05).
Signiﬁcant increase in thickness at the medial and lateral locations over time for the

unimpacted patellae (Pearson Product Moment Correlation, p < 0.05).

+ - signiﬁcantly different from contralateral unimpacted limb (paired t-test, p < 0.05)
# - signiﬁcantly different from baseline (time 0) (AN OVA, SNK post-hoe, p <0.05)

Table 3: The cartilage thickness was measured at the sites of biomechanical testing on
the lateral facet (Avg :1: SD). Gross photographs of the patellae after staining with India
ink were used to determine total ﬁssure length (Avg i SD). The histological sections
were used to determine average ﬁssure depth (Avg :1: SD) and total number of ﬁssures
(median (range)).

 

 

 

 

 

Time Limb Cartilage Total Fissure Average Number of
Point Thickness Length (mm) Fissure Fissures (#)
(mm) E23194!
Unimp 0.57 : 0.06 2.3 : 3.5 10 : 24 0 (0-1)
4.5 ' Imp 0.70 : 0.07 *” 11.2 : 5.8 * 33 : 15 * 4 (2-5) *
Unimp 0.61 :006 3.1 :3.6 7: 18 0 (0-1)
7.5a Imp 0.51 :007 13.1 :63 + 30: 12* 4(1-5) *
Unimp 0.55 : 0.07 4.0 :1: 3.8 12 : 15 0 (0-1)
12' Imp 0.56:0.09 25.7:88 * 37: 14 * 7 (4-12) *
Unimp 0.54 : 0.07 4.4 :1: 4.3 19 : 22 2 (0-8)
24" Imp 0.52:0.10* 14.8 :98 + 31: 10 * 4(0-8) *
Unimp 0.61 : 0.06 3.2 :1: 2.7 6 : 17 o (0-3)
36' Imp 0.31 :0.13*” 17.6:9.6* 41 :8 * 4(0-8) *
Unimp 0.60 : 0.05 4.5 : 3.7 13 : 25 o (0-5)

Imp - impacted limb

Unimp - unimpacted limb

a-n=8

b—n=9

Signiﬁcant decrease in cartilage thickness over time for the impacted patellae (Pearson
Product Moment Correlation, P <0.05).

+ — signiﬁcantly different ﬁ'om contralateral unimpacted limb (paired t-test, p < 0.05)

* - signiﬁcantly different from contralateral unimpacted limb (Signed Rank test, p <0.05)
# - signiﬁcantly different from baseline (time 0) (ANOVA, SNK post-hoe, p <0.05)

31

FIGURES

 

\ \ '-\~ )
“ _\:\\~ V

1,”),
.- ' a

\ i" a/7 /’

 

 

 

 

 

 

 

 

 

 

 

Figure 1: Impact experiments were performed by dropping a rigid mass onto the
patellofemoral joint'with 6.0 J of energy. The impacting mass was stopped electronically
to prevent multiple impacts.

 

(A) (B) (C)

Figure 2: Gross visual examination of the impacted joints after application of India ink
revealed small surface ﬁssures on all impacted patellae at time zero typically along the
center ridge of the patella on the lateral facet (A). By 12 months post-impact more lesions
had developed (B).At 36 months post-impact osteophytes (9) developed on the medial
or lateral sides of three impacted patellae (C).

32

 

Figure 3: The histological sections revealed small acute surface ﬁssures for the impacted
patellae at time zero (A). At 12 months post-impact the lesions were more severe (B). At
36 months the lesions were similar in histological appearance compared to those at 12
months (C).

33

CHAPTER 2

ALTERATIONS IN THE NIECHANICAL PROPERTIES OF BONE
UNDERLYING ARTICULAR CARTILAGE IN A TRAUMATIZED JOINT

Benjamin J Ewers, William W Newberry, Jose J Garcia, and Roger C Haut

INTRODUCTION

Studies of the pathogenesis of osteoarthrosis (OA) have suggested that changes
occur in both the underlying bone and articular cartilage. Changes in the mechanical
properties of the subchondral bone have been thought to contribute to altered stresses in
the overlying articular cartilage causing degradation of this soft material. Earlier studies
have indicated both increases (Radin et aL, 1973) and decreases (Serink et al., 1977) in
subchondral bone stiffness. Our laboratory has been developing a small animal model of
post-traumatic osteoarthrosis (Haut et al., 1995; Newberry and Haut, 1996). The studies
indicate a decrease in the modulus of articular cartilage, and an increase in thickness of
the underlying bone post-impact. The objectives of this current study were to determine
the mechanical properties of the underlying bone as the disease progresses, and study the
role of potential mechanical-based interactions between bone and overlying articular
cartilage post-insult using mathematical analyses.
METHODS

The patello-femoral joints of 32 mature Flemish rabbits (4.7i 0.8 kg, 6-8 months
of age) were impacted by a rigid, gravity-dropped impactor as described previously (Haut

et al, 1995). The animals were sacriﬁced at 1, 3, 6, and 12 months, and the patellae were

34

excised The integrity of the articular cartilage was examined along the lateral facet by
indentation stress relaxation tests in a room temperature PBS bath (pH 7.2). The test
consisted of indenting a 1 mm ﬂat, non-porous probe 0.1 mm into the cartilage and
holding it for 150 seconds while recording the resulting load The shear modulus of the
articular cartilage was determined based on load at 80 ms using an elastic analysis (Hayes
et al., 1972). Sixteen pairs of patellae were histologically processed to determine the
thickness of the underlying bone, using a calibrated eye-piece at 25x. The remaining
patellae were stored at -20° C. Before mechanical testing of the underlying bone, the
patellae were thawed and retro-patellar cartilage was removed under a dissecting
microscope. Using the same probe as for cartilage tests, the subchondral plate was
indented 0.04 mm and held for 150 seconds at six locations along the medial and lateral
facets. An elastic modulus was computed using load at 50 ms based on the solution for an
indentor on an elastic half-space (Timoshenko and Goodier, 1970). Paired t-tests were
used to compare mechanical properties between impacted and non-impacted sides. The
mathematical model of the rabbit patello-femoral joint was based on a 2-D plane strain
contact analysis (ABAQUS). The geometry of the model was obtained by ﬁxing in situ
and sectioning a representative rabbit patello-femoral joint. Poisson’s ratio for the
cartilage and bone were assumed to be 0.49 and 0.3, respectively. The elastic modulus for
the bOne and articular cartilage was based upon the experimental data.
RESULTS

The elastic modulus for the unimpacted underlying bone was 230: 30 MPa. No
change was recorded on the impacted side at 1 month. At 3 months the impacted side

signiﬁcantly decreased to 77% of the unimpacted value (p<.05). The properties of the

35

bone returned to contralateral levels at 6 and 12 months (Fig. 1). The elastic modulus of
the unimpacted articular cartilage was 3.30 i036 MPa. At 1 month post-insult no
change from the unimpacted value was recorded At 3 months the impacted side
signiﬁcantly decreased to 66% of the unimpacted level (p<.05). Unlike the bone, 42 and
33% reduction was observed at 6 and 12 months respectively (Fig. 1). The thickness of
underlying bone on the impacted side was signiﬁcantly greater, at all time points,
compared to, the contralateral unimpacted side. The mathematical model showed a 3%
decrease in maximum shear stress at the cartilage surface when the elastic modulus of the
bone was reduced by 23% at 3 months. The model also showed a 12% and 15% decrease
in maximum shear stress and maximum tensile stress, respectively, in the underlying
bone when the elastic modulus of the articular cartilage was reduced by 42%.
DISCUSSION

A current hypothesis for the progression of 0A is that changes in the underlying
bone cause increased stress in the articular cartilage which leads to degradation of this
soft joint tissue (Radin et al., 1973). The degradation and subsequent softening of
cartilage, in turn, leads to additional increased stress in underlying bone causing sclerosis
and stiffening which further drives deterioration of the overlying cartilage. Our study
showed only a subtle change in bone underlying traumatized and degrading articular
cartilage 3 months after blunt impact, while the subchondral plate thickened signiﬁcantly
at all time points. Our mathematical analyses suggested only a minor decrease would
occur in the stresses within articular cartilage with the documented decrease in modulus
of underlying bone. Such a decrease in stress is not likely to cause softening of articular

cartilage from disuse. Furthermore, the documented softening of overlying cartilage

36

would, as suggested by the model, result in lowered stresses in the underlying bone. This
would not explain thickening of the subchondral plate. These data, and the corresponding
model analyses, suggest that degradation of articular cartilage in this traumatized joint
was independent of mechanical changes in the underlying bone. This implies that
degradation of cartilage was not likely due to increased stress as a result of stiffening in
the subchondral plate. The documented softening of cartilage was more likely due to
wear and tear of the compromised tissue, and the subchondral bone thickening was a
result of healing after microdamage produced by the initial insult. These conclusions are
supported by other studies in which cartilage degrades without thickening of the
subchondral plate (Dedrick et al., 1993). Our studies suggest that interventions in this
disease process; surgical repair, pharmacological treatments, or speciﬁc physical
therapies should focus on repair of the traumatized articular cartilage rather than
thickening of underlying subchondral plate.
ACKNOWLEDGMENTS
This work was supported by the Centers for Disease Control and Prevention
(R49/CCR503607).
REFREENCES
1. Radin EL, Parker HG, Pugh JW, Steinberg RS, Paul IL, Rose RM (1973) Response to
impact loading - III: Relationship between trabecular micro fractures and cartilage

degeneration. J Biomechanics 6:51-57.

2. Serink MT, Nachemson A, Hansson G (1977) The effect of impact loading on rabbit
knee joints. Acta. Orthop. Scand. 48:250-262.

3. Haut RC, Ide TM, De Camp CE (1995) Mechanical responses of the rabbit patello-
femoral joint to blunt impact. J Biomech Eng, 117:402-08.

4. Newberry and Haut (1996) Sac. Phys. Reg. Bio. Med. 16th annual meeting 32-33.

37

5. Hayes WC, Keer LM, Herrmann G, Mockros LF (1972) A mathematical analysis for
indentation tests of articular cartilage. J Biomechanics 5:541-551.

6. Timoshenko and Goodier (1970) Theory of Elasticity pp:380-409.

7. Dedrick DK, Goldstein SA, Brandt KD, O’Connor BL, Goulet RW, Albrecht M
(1993) A longitudinal study of subchondral plate and trabecular bone in cruciate-

deﬁcient dogs with osteoarthritis followed up for 54 months. Arth and Rheum,
36:1460-67.

 

 

 

 

 

FIGURE
3 -- I Bone - Modulus
* ICartilage - Modulus

B 2.5 -- El Bone - Thickness a.
1:3: I
0.

.§ *

g. 1
8

(B

Q

E

 

 

 

 

 

 

 

 

3 Month 6

Figure l: Ratios of mechanical data and subchondral bone thickness at each time point. 11
= 4 for bone modulus and thickness, 11 = 8 for cartilage modulus. * indicates signiﬁcant
difference between impacted and contralateral unimpacted sides using paired t-tests

(p<.05).

38

CHAPTER 3

CHRONIC SOFTENING OF CARTILAGE WITHOUT THICKENING OF
UNDERLYING BONE IN A JOINT TRAUMA MODEL

Ewers BJ, Newberry WN, and Haut RC

ABSTRACT

We have recently developed a trauma model to study degradation of the rabbit
patello-femoral joint. Our current working hypothesis is that alterations in retropatellar
cartilage and underlying bone in our model are initiated independently by acute
overstresses developed in each tissue during blunt insult to the joint, and that the
processes of chronic degradation in each tissue are not related in a mechanical sense. The
current study was conducted in an attempt to help validate our hypothesis by impacting
the patello-femoral joint with a padded interface. Based upon earlier human cadaver
experiments, we believe this would reduce the acute overstresses in patellar bone while
the stresses developed in the overlying retropatellar cartilage would be sufﬁcient enough
to initiate a chronic softening of the tissue. Twenty-four animals received an impact to
the patello-femoral joint and were sacriﬁced at either 0, 4.5, or 12 months post-insult.
Three acute animals were impacted to develop a simpliﬁed computational model to
estimate the stresses in joint tissues. The study showed there was a signiﬁcant softening
of the retropatellar cartilage at 4.5 and 12 months post-trauma, compared to unimpacted

controls. However, no thickening of the underlying subchondral bone was documented at

39

any timepoint. This was consistent with a reduction of stress in the bone compared to
earlier studies, which document thickened subchondral bone post-insult at the same
applied impact load In conclusion, this study helped validate our hypothesis by
documenting chronic softening of cartilage without remodeling of the underlying
subchondral bone. Furthermore, this study, along with our earlier studies, suggest that
impact load alone, which is currently used by the automobile industry to certify new
automobiles, is not a good predictor of chronic injuries to a diarthrodial joint, and that
simply the addition of padding to impact interfaces may not be adequate to protect
occupants from chronic injuries.
INTRODUCTION

We have recently developed a trauma model to study degradation of the rabbit
patello-femoral joint following a single, blunt impact. In various studies we have
documented chronic alterations in retropatellar cartilage, as well as the underlying
subchondral bone, out to one year post inSult (Newberry et al., 1997; Newberry et al.,
1998a). In one study with this joint trauma model we documented a signiﬁcant decrease
in the modulus of retropatellar cartilage three months post-insult with a signiﬁcant
increase in the thickness of underlying subchondral bone at 12 months (N ewberry et al.,
1998a). These data suggest that alterations in retropatellar cartilage may have occurred
prior to alterations in the underlying bone. The results are in contrast to those suggested
by Radin et a1. (1978) where signiﬁcant changes have been documented in underlying
bone before histological alterations have been noted in the overlying cartilage using a
repetitive trauma protocol on the rabbit tibial-femoral joint. These studies and others

(Radin et al., 1990) have lead to the commonly accepted hypothesis that overloading a

40

diarthrodial joint may ﬁrstly result in damage to bone underlying cartilage, followed by a
stiffening of the bone with a subsequent increase in the state of stress in the overlying
cartilage resulting in its degradation over time. Studies with other models of joint injury,
such as transection of the anterior cruciate ligament, have shown degradation of
overlying cartilage before signiﬁcant alterations are noted in the histology of underlying
bone (Dedrick et al., 1993). Since bone changes are typically seen in the clinical setting,
Burr (1998) suggested that alterations of bone underlying cartilage may occur late and be
required to bring forth an end stage disease in the joint. Recently Ewers et a1. ( 1998),
using a simpliﬁed mathematical model of the rabbit patello-femoral joint, suggested that
signiﬁcant alterations in the mechanical properties of bone underlying cartilage in our
joint trauma model would not inﬂuence the state of stress in the overlying cartilage. Nor
will softening of the overlying cartilage, as a result of a severe blunt trauma, alter the
state of stress in the underlying bone and initiate a remodeling process. That study has
been used to establish the basis for our current working hypothesis that alterations in
retropatellar cartilage and underlying bone in our joint trauma model are initiated
independently by acute overstresses developed in each tissue during blunt insult to the
joint, and that the processes of chronic degradation in each tissue are not related in a
mechanical sense (Li et al., 1995; Newberry et al., 1998b).

The current study was conducted in an attempt to help partially validate the above
hypothesis by impacting the rabbit’s patello-femoral joint in such a way that acute
overstresses in patellar bone would be minimized, while the stresses developed in the
overlying retropatellar cartilage would be sufﬁcient enough to initiate a chronic softening

of the tissue. The idea was based on a study by Atkinson et a1. (1997) using the isolated

41

human cadaver joint. That study suggests that by padding the impact interface loads can
be distributed over a large area on the anterior surface of the patella to keep stresses in
bone signiﬁcantly low so as not to generate acute damage or initiate a chronic remodeling
process of bone underlying retropatellar cartilage. On the other hand, by using an impact
intensity sufficient enough to cause cartilage damage, the study further suggests that
acute stresses in the overlying retropatellar cartilage could still be sufﬁcient to induce
injury and lead to a chronic degradation of this soft tissue.
METHODS AND MATERIALS

Thirty-ﬁve mature Flemish Giant rabbits (4.6 :1: 0.8 kg, 6-8 months of age) were
used in the study. The project was approved by the Michigan State University All-
University Committee on Animal Use and Care. Twenty-four animals received a single
impact to the right patella-femoral joint. Eight animals were sacriﬁced at each of three
times post-insult: zero months, 4.5 months, and twelve months. Earlier studies with this
animal model show no signiﬁcant changes in the biomechanical properties of
retI'Opatellar cartilage or the thickness of underlying bone over time in control animals
(Newberry et al. 1998a). Therefore a single, non-impacted control group of eight animals
was used in the study. Three animals (acute animals) were sacriﬁced prior to impact for
determination of the input boundary conditions for computational models.

These impact experiments have been described in an earlier study (Haut et al.,
1995). In this study, however, a deformable impact interface (Hexcel, 3.76 MPa crush
strength) was used to distribute the impacting load over the patella, versus a rigid
interface used in earlier studies by our laboratory. Brieﬂy, with the animal supine and

maintained under anesthesia (2% isoﬂurane), the right hind limb was ﬂexed 120° and a

42

1.33 kg mass was dropped onto the patello-femoral joint from 0.51 m, yielding 6.6 J of
impact energy (Figure 1). The load-time history of the impact was recorded at 10 kHz.
The peak contact load, time to peak, and total contact duration were documented in each
experiment. Starting ﬁve days after impact the animals were regularly exercised, as
previously described (Oyen-Tiesma et al, 1998).

Immediately after sacriﬁce, patellae were excised and the cartilage surface wiped
with India ink to highlight defects on the retTOpatellar surface. Using previous
documented biomechanical tests (Newberry et al., 1998a), the stiffness of the
retropatellar cartilage was determined based on a shear modulus calculation at 50 ms
(instantaneous shear modulus) and 150 s (relaxed shear modulus) using an elastic
analysis (Hayes et al, 1972). Following mechanical testing the patellae were
histologically processed and the thickness of the underlying subchondral plate was
measured at 25x with a calibrated eye-piece at the center of each facet and at the mid-line
of the patella (Newberry et al., 1998a).

To estimate the acute state of stress in the patella at impact, a two-dimensional,
plane-strain, ﬁnite element model of the patella has been developed (N ewberry et al.,
1998b). Brieﬂy, the geometry was taken from a slab cut of a representative patella, where
the subchondral bone plate was distinguished from the cancellous bone. The retropatellar
cartilage was treated as an incompressible elastic material with an elastic modulus and
Poisson’s ratio of 20 MPa and 0.49, respectively (Atkinson et al., 1998). Large
deformations were allowed in the analysis. Due to the characteristics of the rabbit patella,
with high-density cancellous bone, the patellar bone was assumed to have an equivalent

elastic modulus of 350 MPa and a Poisson’s ratio of 0.3, as previously computed for the

43

entire patella (Ewers et al., 1998). The boundary conditions on the patella were derived
from pressure sensitive ﬁlm (Prescale, single sheet, medium range, Fuji Film Ltd,
Tokyo, Japan) inserted into the patello-femoral joint of the three acute animals, as
previously described (Newberry et al., 1998b). The anterior surface of the patella was
ﬁxed in the area where the impact interface contacted the joint. The experimental
retropatellar contact pressure distribution, medial to lateral across the region of highest
contact pressure, was applied to the model of the patella (ANSYS 5.5). In the cartilage
layer maximum shear stresses at the surface and at the interface with subchondral bone
were reported. Maximum shear and peak principal stress distributions were also
documented in the underlying subchondral plate.

An analysis of variance with S-N-K post-hoe tests was used to evaluate
differences in moduli and subchondral bone thicknesses for unimpacted and impacted
limbs between groups (Glantz, 1996). Paired t-tests were used for comparisons between
the impacted and the contralateral, unimpacted limbs. Errors in the mean of data were
reported as one standard deviation. Statistical differences were indicated for p < 0.05.
RESULTS

Qualitative examination after impact loading and daily observations during
treadmill exercise by a veterinary technician (J .A.) indicated that the rabbits did not favor
the impacted limb. Gross visual examination of the impacted joints revealed surface
ﬁssures typically along the center ridge of the patella on the lateral facet in 22 of 24
cases. NO surface lesions were documented on the retropatellar cartilage of the
unimpacted joints or that of controls. Contact loads on the patella for the three acute

animals increased to a maximum of 636 :1: 80 N in 5.4 i 0.8 ms with a total contact

duration of 13.4 i 1.0 ms. This was not signiﬁcantly different from the chronic animals
which increased to a maximum of 620 :1: 50 N in 5.5 i 1.0 ms with a total contact
duration of 13.6 i 0.9 ms (Figure 2).

The instantaneous and relaxed moduli of the retropatellar cartilage were
signiﬁcantly less on the impacted side compared to the unimpacted limb and those of
control animals at 4.5 and 12 months post-impact (Table 1). The subchondral bone
underlying the retropatellar cartilage was thicker at the central site than at the medial and
lateral sites for each group. There was no statistical thickening of the subchondral bone
for the impacted limb compared to either the unimpacted, contralateral limb or control
animals at any time post-impact (Table 2).

The computational model showed a maximum shear stress in the retropatellar
cartilage of 6.1 i 0.9 MPa and 8.6 i 2.0 MPa at the surface and interface, respectively
(Figure 3). These maximum shear stresses were located at the center of the lateral facet
near the impact-induced lesions. The maximum shear and tensile stress in the underlying
bone were 15.6 :1: 2.9 MPa and 7.7 i 1.9 MPa, respectively, and also located under the
lateral facet of the patella near the area of surface lesions.

DISCUSSION

, The objective of the current study was to partially validate our working
hypothesis that alterations in retropatellar cartilage and underlying bone are initiated
independently by acute overstresses developed during blunt impact to the patello-femoral
joint in this animal model. Furthermore, that the documented softening of retropatellar
cartilage and thickening of the underlying subchondral bone occur independently post-

trauma. In support of our hypothesis, there was a signiﬁcant softening of traumatized

45

cartilage compared to controls at 4.5 and 12 months post-trauma, and with no thickening
of subchondral bone at any timepoint in this study.

The shear moduli and subchondral bone thicknesses of the control population
used in this study were statistically similar to previous control animals using this model
(Newberry et aL, 1998a). The degree of chronic softening in the retropatellar cartilage
post-trauma was similar to other studies using this animal model. Newberry et a1. ( 1998a)
document 40 % and 30 % decreases in GU for traumatized cartilage compared to controls
at 6 and 12 months post-trauma, respectively. This was consistent with 41 % and 26 %
decreases observed at 4.5 and 12 months post-trauma in the current study. Furthermore,
Newberry et al (1998a) document 19 % and 28 % decreases in G; for impacted cartilage
compared to unimpacted contralateral patellae at 6 and 12 months post-trauma. The
current study documented 20 % and 32 % decreases in Gp, at 4.5 and 12 months,
respectively. In contrast to the current study, Newberry et al. (1998a) document a
thickening of the subchondral bone at one month post-trauma compared to the
unimpacted contralateral patella, and a signiﬁcant thickening at 12 months compared to
unimpacted controls. NO such thickening was observed with padded impact interfaces in
the current study.

The chronic observations of the current study could be explained by the acute
stresses generated in these joint tissues during blunt impact. The acute stresses generated
in the retropatellar cartilage of the current study were consistent with high severity
impacts which have previously resulted in chronic softening of traumatized cartilage
(Figure 2) (Newberry et al., 1998b). On the other hand, there was a 35 % and 52 %

decrease, respectively in maximum shear and tensile stress in the subchondral bone

46

compared to the previous high severity impacts using a rigid impact interface, which
produced chronic subchondral bone thickening (Newberry et al., 1998b). Previous studies
have suggested that surface lesions on the retropatellar cartilage occur for impacts that
generate a shear stress of 5.5 MPa (Atkinson et al., 1998). This compared well with the
current study that documented surface lesions for an average maximum surface shear
stress of 6.1 MPa in computational models using padded impact interfaces.

Our trauma model suggests that damage and subsequent alteration of joint tissues
may be due to acute impact induced stresses that are generated during a single
overloading of the joint. Therefore, cartilage softening and underlying bone thickening
may occur independently, which would be in contrast to the currently accepted
hypothesis due to Radin et a1. (1990). In contrary to the current study, Radin et a1. (1978)
used low level repetitive overloadings that may have resulted in fatigue damage to the
bone with a secondary effect of cartilage degeneration. The differences between a single
overloading and repetitive overloadings may result in different pathways of joint
degeneration. Additional studies will be needed to support this suggestion.

There were some limitations in the computational model used for the current
study. It was assumed that the patella was a 2-D object, and that the retropatellar cartilage
was isotropic, homogeneous and linearly elastic. This allowed direct comparisons to be
made to previous published studies. In addition because of the large medial to lateral
variation of contact pressure versus the proximal to distal variation, a plane strain
analysis would likely be appropriate. It has also been suggested that the instantaneous
response of biphasic cartilage is equivalent to that of an incompressible elastic material

(Armstrong et al., 1984; Ateshian et al., 1994). It has been shown that the shear stress in a

47

cartilage layer under impact loading are the same for an incompressible elastic and
biphasic material at time zero (Garcia et al., 1998). However since tensile stresses in the
solid phase of articular cartilage might be responsrhle for acute damage to the matrix,
there is the need for future biphasic analyses in order to decompose the stresses in the
impacted tissue.

In conclusion, the current study helped to validate our current working hypothesis
on the chronic alterations which have been documented in the patello-femoral joint of our
model following blunt trauma In addition this study, along with a previous study
(Newberry et al., 1998b), suggests that load alone is not a good predictor of chronic
injury to a diarthrodial joint. This is in contrast to a federally regulated lower extremity
injury criterion that states that axial loads in the patella-femur complex can not exceed 10
kN for certiﬁcation of new automobiles using anthropomorphic dummies. Atkinson et a1.
(1997), using isolated human cadavers, prevented microcracks in bone underlying
retropatellar cartilage when using a particular padded impact interface. That study,
however, did not address damage to articular cartilage. Assuming damage to articular
cartilage can be correlated with diseases such as post-traumatic osteoarthritis, there may
be a need to incorporate a method to better predict damage to joint cartilage in the
certiﬁcation of new automobiles. The current investigation suggests that simply the
addition of padding to impact interfaces may not be adequate, and that optimized
interfaces may be required
ACKNOWLEDGEMENTS

This study was supported by a grant for the Centers for Disease Control and

Prevention (R49/CCR503607). Its contents are solely the responsibility of the authors and

48

do not necessarily represent the ofﬁcial views of the Centers for Disease Control and
Prevention. The authors wish to gratefully acknowledge Jane Walsh for her preparation
of the histological sections, Jean Atkinson for her care of the animals and Cliff Beckett
for technical assistance in the experimental phase of these studies.

REFERENCES

1. Armstrong, C.G., Lai, W.M., Mow, V.C., (1984) An analysis of the unconﬁned
compression of articular cartilage. Journal of Biomechanical Engineering 106, 165-
173.

2. Ateshian, G.A., Lai, W.M., Zhu, W.B., Mow, V.C., (1994) An asymptotic solution for
the contact of two biphasic carilage layers. Journal of Biomechanics 27, 1347-1360.

3. Atkinson, P.J., Garcia, J.J., Altiero, N.J., Haut, RC, (1997) The inﬂuence of impact
interface on human knee injury: Implications for instrument panel design and the
lower extremity injury criterion. Society of Automobile Engineers, Paper No. 973327.

4. Atkinson, T.S., Haut, R.C., Altiero, N.J., ( 1998) Impact induced ﬁssuring of articular
cartilage: An investigation of failure criteria Journal of Biomechanical Engineering
120,181-187.

5. Burr, D.B., (1998) Subchondral bone. In: Brandt, KD., Doherty, M., Lohmander, L.S.
(Eds), Osteoarthritis. Oxford University Press Inc., New York, pp. 144-156.

6. Dedrick, D.K., Goldstein, S.A, Brandt, KD., O’Conner, B.L., Goulet, RW., Albrecht,
M., (1993) A longitudinal study of subchondral plate and trabecular bone in cruciate-
deﬁcient dogs with osteoarthritis followed up for 54 months. Arthritis and
Rheumatism 36, 1460-1467.

7. Ewers, B.J., Newberry, W.N., Garcia, J.J., Haut, RC, (1998) Alterations in the
mechanical properties of bone underlying articular cartilage in a traumatized joint.
Transactions of the 44“ Annual Meeting of Orthopaedic Research Society, New
Orleans, LA.

8. Garcia, 1]., Altiero, N.J., Haut, RC, (1998) An approach for the stress analysis of
transversely isotropic biphasic cartilage under impact load. Journal of Biomechanical
Engineering 120, 608-613.

9. Glantz, SA, (1996) A better approach to multiple comparison testing. In:

Wonsiewicz, M., McCurdy, P. (Eds), Primer of Biostatistics McGraw-Hill, New
York, pp. 92-97.

49

10. Haut, R.C., Ide, T.M., Decamp, CE, (1995) Mechanical responses of therabbit
patello-femoral joint to blunt impact. Journal of Biomechanical Engineering 117,
402-408.

11. Hayes, W.C., Keer, LM, Herrmann, G., Mockros, LR, (1972) A mathematical
analysis for indentation tests of articular cartilage. Journal of Biomechanics 5, 541-
551.

12. Li, X., Haut, RC., Altiero, N.J., (1995) An analytical model to study blunt impact
response of the rabbit P-F joint. Journal of Biomechanical Engineering 117, 485-
491.

13. Newberry, W.N., Zukosky, D.K., Haut, R.C., (1997) Subfracture insult to a knee joint
causes alterations in the bone and in the functional stiffness of overlying cartilage.
Journal of Orthopaedic Research 15, 450-455.

14. Newberry, W.N., MacKenzie, C.D., Haut, R.C., (1998a) Blunt impact causes changes
in bone and in cartilage in a regularly exercised animal model Journal of
Orthopaedic Research 16, 348-354.

15. Newberry, W.N., Garcia, J.J., Mackenzie, C.D., Decamp, C.E., Haut, RC, (1998b)
Analysis of acute mechanical insult in an animal model of post-traumatic
osteoarthritis. Journal of Biomechanical Engineering 120, 704-709.

16. Oyen-Tiesma, M, Atkinson, J., Haut, RC, (1998) A method for promoting regular
rabbit exercise in orthopaedics research. Contemporary Topics in Laboratory Animal
Science 37, 77- 80.

17. Radin, E.L., Ehrlich, M.G., Chernack, R., Abemethy, R, Paul, I.L., Rose, R.M.,
(1978) Effect of repetitive impulsive loading on the knee joints of rabbits. Clinical
Orthopaedics 131, 288-293.

18. Radin, E.L., Burr, D.B., Fyhrie, D., Brown, T.D., Boyd, RD, (1990) Characteristics
of joint loading as it applies to osteoarthrosis. In: Mow, V.C., Ratcliffe, A., W00, S.L-
Y. (Eds), Biomechanics of Diarthrodial Joints. Springer-Verlag, New York, pp. 437-
451.

50

TABLES

Table 1: Measured instantaneous (GU) and relaxed (GR) moduli of retropatellar cartilage
(MPa) from animals subjected to a single blunt impact with a padded interface and in the
unimpacted, control animals (mean i SD) (n = 8).

 

 

 

 

 

Time point Side GU (MPa) GR(MPa)
Control 1.00 : 0.25 0.29 : 0.06
Time-zero Impacted 1.11 i 0.46 0.29 i 0.08
Unimpacted 1.03 : 0.48 0.27 : 0.09
4.5 months Impacted 0.59 : 0.10 9" 0.23 : 0.03 ‘"
Unimpacted 0.88 : 0.21 0.29 : 0.09
12 months Impacted 0.74 : 0.24 "1" 0.23 : 0.05 ‘1"
Unimpacted 1.08 : 0.29 0.34 : 0.03

 

‘ Signiﬁcantly different than unimpacted side by a paired students’ t-test.
b Signiﬁcantly different than controls by an analysis of variance.

Table 2: Measured subchondral bone thickness (mm) in animals subjected to a single

impact with padded interface and in the unimpacted control animals (mean :1: SD) (n=8).
Subchondral bone thickness is greater in the center compared to the lateral or medial

 

 

 

 

 

locations.
Time int ﬁe Location
Medial Central Lateral
Control 0.39 :l: 0.04 0.59 :l: 0.03 0.33 :l: 0.05
Time-zero Impacted 0.36 i 0.09 0.55 i 0.10 0.34 :l: 0.07
Unimpacted 0.39 i 0.11 0.56 :1: 0.22 0.35 i 0.09
4.5 months Impacted 0.33 :l: 0.01 0.56 i 0.07 0.39 d: 0.01
Unimpacted 0.32 i 0.03 0.55 i 0.09 0.39 :l: 0.01
12 months Impacted 0.45 :1: 0.09 0.67 i 0.15 0.49 i 0.10
Unimpacted 0.45 :l: 0.08 0.64 :I: 0.16 0.52 i 0.16

 

51

FIGURES

 

 

 

 

 

 

 

 

 

 

Figure 1: Impact experiments were performed by dropping a mass with a padded impact
interface (A) (3.76 MPa crush strength — Hexcel) onto the patellofemoral joint with 6.6 J
of energy.

      

 

— Rigid
—— Padded

 

 

 

 

        

 

I l I I

0 2‘4 6 810121416
Time(ms)

I

Figure 2: Representative force time plots for experiments with a rigid and padded impact
interface. Notice the slightly longer time to peak with the padded interface but a similar
contact dmation. This representative rigid impact data is taken from a previous study

(Newberry et al., 1998b).

52

Scale

 

D 1: ~2.8MPa
El 2: 2.8~5.6MPa
I 3: 5.6~8.4MPa
I 4: 8.4~11.2 MPa
I 5: 11.2 MPa~

 

 

 

 

DA: ~ 6.6 MPa

EB: 6.6 ~13.2MPa
IC: 13.2 ~19.8MPa
ID: 19.8 ~ 26.4 MPa
IE: 26.4 MPa~

 

 

 

 

(B)

Figure 3: A representative shear stress distribution for padded (A) and rigid (B) impact
interfaces. The cartilage is on the top with the subchondral plate underneath. Pressures
were applied to the retropatellar surface and the anterior surface of the patella was ﬁxed
in the area of contact (A). Notice that the shear stresses in the cartilage were similar for
the two impact interfaces, but that there was a signiﬁcant reduction in shear stresses in
subchondral bone underlying the retropatellar cartilage for the padded interface
experiment. The representative shear stress distribution for the rigid impact interface is
based upon a previous study (Newberry et al., 1998b).

53

CHAPTER 4

MACT SHEAR STRESSES IN AN ANIMAL JOINT PREDICTS CHRONIC
SOFTENING OF CARTILAGE BUT NOT THICKENING OF UNDERLYING
BONE

Benjamin J. Ewers, Brian T. Weaver, William N. Newberry, and Roger C. Haut

ABSTRACT

Our laboratory has recently developed a joint trauma model, using rabbits, which
exhibits chronic softening of retropatellar cartilage and thickening of underlying
subchondral bone. We hypothesize that these alterations are initiated by acute shear
stresses generated in each tissue during impact loading. With the use of a Finite Element
Method (FEM), an earlier analytical study suggests shear stresses in the cartilage can be
signiﬁcantly reduced by reorientation of the impact load more centrally on the joint. The
current study suggested that these types of impacts indicated reduced shear stresses and
tensile strains in retropatellar cartilage with enhanced shear stresses in the underlying
bone. In contrast to earlier studies we documented no degradation of retropatellar
cartilage or remodeling of the underlying bone. The study supports our contention that
degradation of cartilage may be related to the intensity of generated shear stress, but

contradicts the role of shear stress alone in the remodeling of underlying bone.

54

INTRODUCTION

We have recently developed a trauma model to study degradation of the rabbit
patellofemoral joint following a single, high severity blunt impact on the medial aspect of
the anterior patella. In various studies we have documented acute superﬁcial lesions on
the retropatellar cartilage and chronic changes in cartilage, as well as the underlying
subchondral bone, out to one year post-insult (Newberry et aL, 1997; Newberry et al.,
1998a). We document a signiﬁcant decrease in stiﬂness of retropatellar cartilage
beginning at three months post-insult with a signiﬁcant increase in the thickness of
underlying subchondral bone at 12 months (Newberry et al., 1998a). In a recent study
with this joint trauma model we lowered the severity of the medial-oriented blunt impact
and did not document any signiﬁcant changes in these tissues post-insult (Newberry et
al., 1998b). We hypothesize that changes in cartilage and underlying bone are correlated
with the intensity of the blunt insult and the corresponding state of acute stress generated
in these tissues.

In contrast to the above, Radin et al. (1990) have hypothesized that repetitive
loading to a diarthrodial joint may ﬁrst result in damage to bone underlying cartilage,
followed by a stiffening of the bone with a subsequent increase in the state of stress in the
overlying cartilage resulting in its degradation over time. Recently, however, Ewers et al.
(1998) using a simpliﬁed mathematical model of the rabbit patellofemoral joint, suggest
that signiﬁcant alterations in the mechanical properties of bone underlying cartilage in
our joint trauma model would probably not inﬂuence the state of stress in the overlying
cartilage. Nor would a softening of the overlying cartilage, as a result of a severe blunt

trauma, alter the state of stress in the underlying bone and initiate its remodeling. That

55

study has been used to establish the basis for our current working hypothesis that
alterations in retropatellar cartilage and underlying bone in our joint trauma model are
initiated independently by acute overstresses developed in each tissue during blunt insult
to the joint, and that the processes of chronic degradation in each tissue are not related in
a mechanical sense (Newberry et al, 1998b; Li et al., 1995). In a recent study supporting
this hypothesis, we used a padded impact interface. This allowed us to lower the stresses
in the subchondral bone, while keeping the same high level of stresses in the cartilage.
Interestingly, at one-year post-insult we still documented a softening of retropatellar
cartilage without a thickening of the underlying subchondral bone (Ewers et al., 2000).

The current study was conducted in an attempt to further validate our working
hypothesis by impacting the rabbit’s patello femoral joint in such a way that acute
overstresses in retropatellar cartilage would be minimized, while the stresses developed
in the underlying subchondral bone would be sufﬁcient enough to initiate its chronic
thickening. The idea was based on an earlier study by Li et al. (1995), using a
mathematical model of the rabbit patellofemoral joint. That study suggests that a single
central-oriented impact would result in reduced tensile strains in the retropatellar
cartilage, which might mitigate its acute damage. It also suggests that the shear stresses
in the underlying subchondral bone would be greater than that for a single medial-
oriented impact, so as to cause" its chronic thickening. Our question was whether the
overlying cartilage would degrade within one year post-impact.
METHODS

A total of twenty-nine mature Flemish Giant rabbits (4.8 :1: 0.6 kg, 6-8 months of

age) was used in this investigation. The study was approved by the Michigan State

56

University All-University Committee on Animal Use and Care. Twenty-four animals
were randomly selected for the chronic aspect of the study (chronic animals). Eight
animals received a single blunt impact to the right patello femoral joint with a medial-
oriented impact, while eight animals received a blunt impact to the patello femoral joint
that was central-oriented An additional eight animals were not impacted, and served as
controls. The chronic animals were all sacriﬁced at 12 months post-impact. Five
additional animals (acute animals) were sacriﬁced prior to impact for the determination
of boundary conditions in computational models.

The medial-oriented blunt impact protocol has been described in earlier studies
(Newberry et al., 1998a; Haut et al., 1995). Brieﬂy, the animals to be subjected to a blunt
trauma were maintained under anesthesia (2% Isoﬂurane) with the right hind limb ﬂexed
approximately 120° (Figure 1A) with the animal supine. The medial-oriented impact
required the femur to be positioned so that the impact load was aligned along the medial
side of the anterior patella and directed towards the lateral patellar facet (Figure 13),
whereas the central-oriented impact was aligned along the medial-lateral center of the
patellofemoral joint (Figure 1C). These orientations were veriﬁed in pilot studies using
radiographic analyses. Impact was administered by dropping a rigid 1.33 kg mass onto
the patellofemoral joint from a height of 0.46 m (6.0 J) [2]. The impact mass was arrested
electronically to prevent multiple blows to the ﬂexed limb. A load transducer (model
31/1432, 500 lb. capacity, Sensotec, Columbus, OH) was attached behind a one-inch
diameter, ﬂat aluminum interface to measure impact loads. The load-time history of the
impact was recorded at 10 kHz. The peak contact load, time to peak, and total contact

duration were documented for each experiment. The animals were given two weeks of

57

pre-impact exercise consisting of ten minutes of daily exercise on a treadmill running at
0.3 mph (Oyen-Tiesma et al., 1998). Blunt impact loading was followed by a ﬁve-day
period of rest, and then the daily exercise program was resumed for the duration of the
study. When not exercising, the animals were housed individually in cages
(48"x24"xl9").

Immediately after sacriﬁce, patellae ﬁ'om impacted and unimpacted limbs were
excised. The patellae were then placed in a bath of room temperature phosphate buffered
saline (pH 7.2) for mechanical tests in a servo-controlled hydraulic material testing
machine (model 1331, retroﬁtted 8500+ electronics, Instron, Canton, MA). Structural
integrity of the retropatellar cartilage was determined using indentation stress relaxation
tests at two sites on the lateral facet near the central ridge (Newberry et al., 1998a). The
stiffnesses of retropatellar cartilage from impacted, non-impacted, and control joints were
compared by a calculation of the shear modulus from an elastic layer (Hayes et al., 1972),
based on the load at 80 ms (instantaneous shear modulus - GU) and at 150 s (relaxed shear
modulus - GR). The ren'opatellar cartilage was wiped with India ink to highlight surface
lesions and then photographed

Patellae were placed in 10% buffered formalin for seven days, and decalciﬁed in
20% formic acid for another seven days. Only four of the central-oriented impacted
patellae were randomly selected and histologically processed Tissue blocks were cut
transversely across the patella in the area of highest patello femoral contact pressure,
based on the acute experiments described below. The blocks were then embedded with
parafﬁn according to an established protocol (Newberry et al., 1998a). Six sections, eight

microns thick, were stained with Saﬁ'anin O-Fast Green and examined in light

58

microscopy at 12-400 power. The thickness of the subchondral bone plate was measured
at 25x with a calibrated eye-piece at the center of each facet and near the mid-line of the
patella by a single investigator (BE), using an established protocol (Newberry et al.,
1998a).

To estimate the impact-induced state of stress in the patella, a two-dimensional
non-contact ﬁnite element model of the patella was developed consisting of 1537 3-
noded plane strain elements (Figure 2A) (Ansys 5.5, SAS IP Inc., PA, USA). A
representative model of the patellae was taken from a 0.5 mm transverse slab cut of a
patellae. Brieﬂy, the cross section of the patella was photographed and then used to form
a digital image of the geometry by tracing the boundary of the cartilage and subchondral
and trabecular bone. The layer of cartilage was assumed isotropic with an elastic
modulus and Poisson's ratio of 20 MPa and 0.49, respectively (Newberry et al., 1998b;
Atkinson et al., 1998). The instantaneous (impact) response Of biphasic cartilage is
equivalent to that of an incompressible elastic material (Armstron et al., 1984; Ateshian
et aL, 1994). Large deformations were allowed in the analysis. The bone had a Poisson's
ratio of 0.3 and an elastic modulus of 3750 MPa and 300 MPa for the subchondral
(Mente and Lewis, 1994) and trabecular bone (Townsend et al., 197 5), respectively.

The determination of boundary conditions on the patella for the computational
model has previously been described (Newberry et aL, 1998b). Brieﬂy, ﬁve acute animals
were sacriﬁced in order to document the posterior and anterior pressure distributions on
the patella during the impact experiments. Both a medial and central-oriented impact was
conducted on each animal. Prior to impact, medial and lateral incisions were made in the

joint capsule to insert pressure sensitive ﬁlm (Prescale, single sheet, medium range, Fuji

59

Film, Ltd, Tokyo, Japan) into the patellofemoral joint. The ﬁlm was encased in a packet
of sterile polyethylene to protect it from joint ﬂuids (Haut et al., 1995). In order to locate
the pressure distribution on the patella, two holes were drilled through the center of the
patella, anterior to posterior, at two locations away from the area of contact (one
proximal, one distal) (Newberry et al., 1998b). Prior to impact, small needles were
placed into these holes. The needles were cut so they extended approximately 1 mm
beyond the anterior and posterior surfaces. The ﬁlm was placed in the joint and wrapped
around the anterior surface of the patella. The joint was impacted The needles produced
marks (holes) that were easily visible on the ﬁlm. The pressure ﬁlm was dynamically
calibrated in a servohydraulic testing machine (Haut et al., 1995). The ﬁlm was scanned
at 150 dpi (~59 pixels per m) using a video scanner (Microtek, model MRS-1200136).
The stained imprints were converted to pressure and post-processed with PC-based image
analysis software (Image 1.6, NIH). The distributions of contact pressure and contact area
were documented ﬁ'om the patello femoral pressure imprints for each acute animal
experiment. The anterior area of contact was located on the patella, and this area was
ﬁxed in space in the computational model to prevent rigid body motion. Patellofemoral
pressure proﬁles were discretized in the area of the largest contact pressures, and located
on a single representative patellar cross sectional geometry for each experiment (Figure
2A). A pilot study conducted with multiple impacts documented that there was less than
an 8 % difference in contact area and average contact pressure, respectively.

The maximum shear stress and tensile strain were calculated along the
retropatellar articular cartilage surface on the lateral facet where lesions are historically

generated with this model (Figure 2B). Maximum shear stresses, principal stresses and

mean stresses were determined in the underlying subchondral bone in the center region of
the patella where chronic thickening is typically observed (Figure 2C).

An analysis of variance with Bonferroni post-hoe tests were used to evaluate
differences in the mechanical properties of retropatellar cartilage and the subchondral
bone thickness differences between controls and unimpacted joints and between controls
and impacted joints. Paired t-tests were used for comparisons of mechanical data from
cartilage and subchondral thickness data between impacted and unimpacted sides of each
animal. Experimental data in this manuscript has been reported as mean :1: one standard
deviation, and statistical signiﬁcance was indicated for p<0.05.

RESULTS

Peak load, rise time, and duration of impact were similar for the acute and chronic
medial-oriented impacts and for the acute and chronic central-oriented impacts (Table 1).
The magnitude of the patellofemoral contact pressures varied across the area of contact,
typically with the most intense pressure located in the middle of the lateral and medial
facets (Figure 3A & 38). Quantitative analysis of the pressure ﬁlm from the acute
impacts indicated that there was no signiﬁcant difference in mean or peak pressures
between the central-oriented and the medial-oriented impact groups (Table 1). The
patello femoral contact area slightly increased in the central-oriented versus medial-
oriented acute groups, but again the result was not statistically different.

No surface lesions were documented on the retropatellar cartilage for the chronic
central-oriented impacted animals. On the other hand, there were Stu-face lesions on the
retropatellar cartilage in all of the chronic medial-oriented impacted patellae. These

lesions were typically on the lateral facet near the central ridge. Indentation tests on

61

retropatellar cartilage of the chronic animals showed no statistical differences in Gr or GR
between controls and the unimpacted limbs ﬁom the medial- or central-oriented groups
(Table 2). There was a signiﬁcant increase in GR from the unimpacted limb in the central-
oriented group compared to the unimpacted limb of the medial-oriented group. There was
no signiﬁcant difference in the central-oriented group in either GU or GR between the
impacted and the contralateral unimpacted patellae. Conversely, there was a signiﬁcant
decrease in GU and GR on the impacted compared to the contralateral unimpacted patellae
in the medial-oriented group. There was also a signiﬁcant decrease in both GU and GR
ﬁom the impacted limb of the medial-oriented group compared to the impacted limb of
the central-oriented and the unimpacted control groups.

The computational models revealed that the maximum tensile strain on the
surface retropatellar cartilage was signiﬁcantly reduced from 0.44 i 0.06 to 0.34 :1: 0.01
in medial-oriented versus central-oriented acute impacts. The maximum shear stress on
the surface of the retropatellar cartilage was also signiﬁcantly decreased ﬁom 6.50 i 0.94
MPa to 5.19 i 0.20 MPa for the medial-oriented and central-oriented impacts,
respectively. The contour plots of the maximum shear stress in the cartilage illustrate how
these stresses increase along the lateral facet for the high severity medial-oriented impact
group and are more constant and lower for the central-oriented impact group (Figure 4).

Quantitative analysis post-impact revealed that there was no statistical difference
in the subchondral bone thickness between the unimpacted limbs in the medial- and
central-oriented groups and controls at any location (Table 3). There was no signiﬁcant
difference in subchondral bone thickness at any location in the central-oriented group

between the impacted and the contralateral unimpacted patellae. In contrast, at the medial

62

and central locations there was a signiﬁcant increase in subchondral bone thickness in the
impacted versus contralateral unimpacted limb in the medial-oriented group. There was
also a signiﬁcant increase in subchondral bone thickness at the central location for the
impacted patellae in the medial-oriented group compared to controls.

As expected the computational models predicted a slight increase in average
maximum shear stresses in the central region of the subchondral bone for the central-
oriented versus medial-oriented impacts (Table 4). In the central region of the
subchondral bone there was a signiﬁcant increase in the average normal stresses for the
central-oriented compared to the medial-oriented impacts (Figure 4).

DISCUSSION

A The objective of this study was to validate our current working hypothesis that
alterations in retropatellar cartilage and underlying bone are initiated independently by
acute overstresses that are developed during a blunt impact to the patello femoral joint in
our animal model. Based on a previous study by Li et a1. (1995), we hypothesized that a
central-oriented patellar impact on our animal model would result in lower tensile strains
and stresses in the retropatellar cartilage compared to a medial-oriented impact, thus
mitigating the typically observed surface ﬁssures and chronic soﬁening of this tissue
(Newberry et al., 1998a). On the other hand, the central-oriented impact would result in a
higher state of shear stress in the underlying subchondral bone, compared to a medial-
oriented impact, to cause a remodeling and subsequent thickening of this tissue post-
trauma.

By impacting the joint centrally our computational model did indicate a reduction

in tensile strains and maximum shear stresses in the retropatellar cartilage compared to a

63

medial-oriented impact. Subsequently, we did not observe any surface ﬁssures or chronic
softening of the cartilage in the chronic central-oriented impacted animals, whereas there
was surface ﬁssuring and chronic soﬁening of retropatellar cartilage following a medial-
oriented impact. The computational models indicated an increase in maximum shear
stresses in the subchondral bone for the central-oriented versus medial-oriented impacts.
However, while there was signiﬁcant thickening of the subchondral bone in the chronic
medial-oriented impacted animals, there was no thickening of the subchondral bone for
the chronic central-oriented impacts.

One possible explanation for this result is that even though the shear stresses were
similar for the medial and central-oriented impact conﬁgurations, the central-oriented
impacts produced signiﬁcantly greater mean compressive stresses compared to the
medial-oriented impacts. Microdamage may form a basis for remodeling of bone (Burr et
a1, 1984; Mori and Burr, 1984). Impact induced microcracks may coalesce and be the
precursors of gross ﬁactures in bone (Atkinson et al., 1997; Vashishth et al., 1996).
Therefore, since bone behaves similar to a brittle engineering material, a ﬁ'acture criterion
based on a Coulomb-Mohr model has been proposed by others (Keyak and Rossi, 2000;
Plank et al., 1998; Shigley, 1977). In this model high mean stresses can effectively help
protect brittle materials ﬁom shear induced damage. Interestingly, we documented a
signiﬁcant increase in the average mean stresses in the subchondral bone for the central-
oriented impact group compared to the medial-oriented group (Table 4). Therefore, this
might be indicative of a Coulomb—Mohr scenario. The combination of the shear and

mean stresses in the medial-oriented impact group may have exceeded any threshold that

might exist for bone damage. This combination in the central-oriented group possibly lies
below this threshold

The current study suggested that since the chronic central-oriented experiments
did not result in surface lesions or degradation of G; and GR post-impact, the impact-
induced shear stresses were “safe”. The computational model indicated an average
maximum shear stress of 5.2 :1: 0.2 MPa. This is in support of a previous study that
proposed a shear stress of 5.45 MPa is required for impact induced ﬁssuring of articular
cartilage on the rabbit tibial plateau (Atkinson et al., 1998).

The chronic changes documented in the retropatellar cartilage and underlying
subchondral bone in the medial-oriented impacts in the current study are similar to
previous studies with this animal model Newberry et al. (1998a) at 12 months post-
impact documented a signiﬁcant 43 and 28 % decrease in GU and GR, respectively, for
the traumatized cartilage compared to the contralateral unimpacted limb. This is similar
to the 38 and 27 % reduction in GU and GR, respectively, for the impacted versus the
contralateral unimpacted cartilage in the current study. Newberry et al. (1998a)
documented at 12 months post-impact a signiﬁcant 72 % increase in subchondral bone
thickness at the central location of the patellae for the traumatized limb compared to the
contralateral unimpacted limb. This is also similar to the current study that documented a
74 % increase in subchondral bOne thickness at the central location of the traumatized
patellae compared to the contralateral unimpacted limb.

There were a number of limitations in our current computational model. We
assumed the patella was a 2-D object, and that retropatellar cartilage was isotropic,

homogeneous and linearly elastic. The model was selected so that the computational

65

aspects of the current study could be correlated with earlier studies by our laboratory
(Newberry et al., 1998b; Li et al, 1995; Ewers et al, 2000). Because of large medial to
lateral variation of contact pressure versus the proximal to distal variation, a plane strain
analysis can be justiﬁed Since the instantaneous response of biphasic cartilage has been
shown equivalent to that of an incompressible elastic material, an elastic model was
deemed appropriate for this study (Armstrong et al., 1984; Ateshian et al., 1994). On the
other hand, earlier studies have suggested that articular cartilage in uniaxial compression
under rapid loads responds linearly to only approximately 30% strain (Garcia et al.,
2000). The effects of a nonlinear stress-strain response on the distribution of stress in our
impact joint model are currently under investigation (Garcia et al., 2000). Another
limitation is that an elastic model does not allow for the decomposition of stress into its
ﬂuid and solid phases. The analysis of solid stresses might provide vital information for
ﬁssuring and subsequent degradation of the cartilage.

In conclusion, this study suggested that shear stress alone might be a good
predictor of cartilage damage; however, this is not the case for the subchondral bone. A
slight relationship may exist between high shear stresses and mean stresses, which might
be indicative of a Coulomb-Mohr scenario. If, as suggested by others, microdamage is the
basis for such a response, then the proposed model may .be useful as a conservative
indicator of gross bone ﬁacture. Gross fracture, based on total load applied to the joint, is
currently used for the federally regulated lower extremity injury criterion in the
automobile setting (Atkinson et al., 1997). The current study would suggest that the state

of bone stress, and the probability for fracture, might also depend on the orientation of

66

the impacting load on the knee. This issue needs future study using a human knee joint

model

ACKNOWLEDGMENTS

This study was supported by a grant from the Centers for Disease Control and

Prevention (R49/CCR503607). Its contents are solely the responsibility of the authors

and do not necessarily represent the ofﬁcial views of the Centers for Disease Control and

Prevention. The authors wish to gratefully acknowledge Jane Walsh for preparation of

the histological slides, Jean Atkinson for care of the animals, and Charles DeCamp,

D.V.M. for veterinary consultation.

REFERENCES

1.

Armstrong, C. G., Lai, W. M. and Mow, V. C., (1984) An Analysis of the Unconﬁned
Compression of Articular Cartilage. Journal of Biomecanical Engineering 106, pp.
165-173.

Ateshian, G. A., Lai, W. M., Zhu, W. B. and Mow, V. C., (1994) An Asymptotic
Solution for the Contact of Two Biphasic Cartilage Layers. Journal of Biomechanics
27, pp. 1347-1360.

. Atkinson, T. S., Haut, R. C., and Altiero, N. J., (1998) Impact Induced Fissuring of

Articular Cartilage: An Investigation of Failure Criteria Journal of Biomechanical
Engineering 120, pp. 181-187.

Atkinson, P. J., Garcia, J. J., Altiero, N. J., and Haut, R. C., (1997) The Inﬂuence of
Impact Interface on Human Knee Injury: Implications for Instrument Panel Design
and the Lower Extremity Injury Criterion. Proceedings, 41" Stapp Car Crash
Conference pp. 167-180.

Burr, D. B., Martin, B. R., Schafﬂer, M. B., and Radin, E. L., (1984) Bone
Remodeling in Response to In Vivo Fatigue Microdamage. Journal of Biomechanics
18, pp. 189-200.

Ewers, B.J., Newberry, W.N., Garcia, JJ., and Haut, R.C., (1998) Alterations in the

Mechanical Properties of Bone Underlying Articular Cartilage in a Traumatized Joint.
Proceedings, 44th Annual Meeting Orthopaedic Research Society pp. 45 9.

67

7. Ewers, B. J., Newberry, W. N., and Haut, R. C., (2000) Chronic Softening of
Cartilage Without Thickening of Underlying Bone in a Joint Trauma Model. Journal
of Biomechanics, In Print.

8. Garcia, J. J., Altiero, N. J., and Haut, R C., (2000) Estimation of in Situ Elastic
Properties of Biphasic Cartilage Based on a Transversely IsotrOpic Hypo-Elastic
Model Journal of Biomechanical Engineering 122, pp. 1-8.

9. Haut R. C., Ide T. M., Decamp C. E., (1995) Mechanical Responses of the Rabbit
Patello-Femoral Joint to Blunt Impact. Journal of Biomechanical Engineering 117,
pp. 402-408.

10. Hayes, W. C., Keer, I. M, Herrmann, and Mockros, I. E., (1972) A Mathematical
Analysis for Indentation Tests of Articular Cartilage. Journal of Biomechanics 5, pp.
541-551.

11. Keaveny, T. M., Wachtel, E. R, Ford, C. M., and Hayes, W. C., (1994) Differences
Between the Tensile and Compressive Strengths of Bovine Tibial Trabecular Bone
Depend on Modulus. Journal of Biomechanics 27, pp. 1137-1146.

12. Keyak, J. H., and Rossi, S. A, (2000) Prediction of Femoral Fracture Load Using
Finite Element Models: An Examination of Stress- and Strain-Based Failure
Theories. Journal of Biomechanics 33, pp. 209-214.

13. Li, X., Haut, R C., and Altiero, N. J., (1995) An Analytical Model to Study Blunt
Impact Response of the Rabbit P-F Joint. Journal of Biomechanical Engineering 117,
pp. 485-491.

14. Mente, P. L., and Lewis, J. L., (1994) Elastic Modulus of Calciﬁed Cartilage is an
Order of Magnitude Less than that of Subchondral Bone. Journal of Orthopaedic
Research 12, pp. 637-647.

15. Mori, S., and Burr, D. B., (1993) Increased Intracortical Remodeling Following
Fatigue Damage. Bone 14, pp. 103-109.

16. Newberry, W. N., Zukosky, D. K., and Haut, R. C., (1997) Subfracture Insult to a
Knee Joint Causes Alterations in the Bone and in the Functional Stiffness of
Overlying Cartilage. Journal of Orthopaedic Research 15, pp. 450-455.

17. Newberry, W. N., Mackenzie, C. D., and Haut, R C., (1998a) Blunt Impact Causes
Changes in Bone and Cartilage in a Regularly Exercised Animal Model. Journal of
Orthopaedic Research 16, pp. 348-354.

18.Newberly, W. N., Garcia J. J., Mackenzie, C. D., Decamp, C., and Haut, R. C.,

(1998b) Analyses of Acute Mechanical Insult in an Animal Model of Post-Traumatic
Osteoarthrosis. Journal of Biomechanical Engineering 120, pp. 704-709.

68

19. Oyen-Tiesma, M., Atkinson, J., and Haut, R.C., (1998) A Method for Promoting
Regular Rabbit Exercise in Orthopaedics Research Contemporary Topics in
Laboratory Animal Science 37, pp.77-80.

20. Plank, G. R, Kleinberger, M., and Eppinger, R. H., (1998) Analytical Investigation of
Driver Thoracic Response to Out of Position Airbag Deployment. Proceedings, 42""
Stapp Car Crash Conference PA, pp. 317-329.

21.Radin, E. L., Burr, D. B., Fyhrie, D., Brown, T. D., and Boyd, R D., (1990)
“Characteristics of Joint Loading as it Applies to Osteoarthrosis. Biomechanics of
Diarthrodial Joints, Springer-Verlag, New York, pp. 437-451.

22. Shigley, J. 13., (1977) Mechanical Engineering Design, 3" Edition, McGraw-Hill,
New York, pp. 173-175.

23. Townsend, P. R, Raux, P., Rose, R M., Miegel, R. E., and Radin, E. L., (1975) The
Distribution and Anisotropy of the Stiffness of Cancellous Bone in the Human
Patella Journal of Biomechanics 8, pp. 363-367.

24. Vashishth, D., Behiri, J. C., and Bonﬁeld, W., (1996) Crack Growth Resistance in

Cortical Bone: Concept of Microcrack Toughening. Journal of Biomechanics 30, pp.
763-769.

69

TABLES

Table 1: Impact characteristics for chronic and acute central-, and acute medial-oriented
impacts (mean :1: SD). The peak pressure, mean pressure and contact area are from the
patello femoral contact pressure distributions.

 

 

Peak Rise Peak Mean
Group Load Time Duration Pressure Pressure Area
(N) (1118) (m8) (MPa) (MPa) (mmz)
Aerie‘
Medial 575 :1: 4.3 i 11.7 i 42.4 i 23.8 i 22.2 i-
32 0.9 1.0 1.9 2.5 1.0
Central 575 :1: 4.0 i 10.8 :t 40.2 i 25.1 :1: 24.3 i
70 0.3 0.6 2.2 2.9 2.5
Chronic"
. 630 i 4.2 i 10.5 i
Medral 94 0.5 1 .8 ---.. ----- ---..
Central 604 i 4.5 d: 10.8 :t ---- ---- ----
38 0.5 3.0
‘ N = 5
b N 8

Table 2: Measured indentation response of cartilage in control and chronic animals at 12
months post impact (MPa, mean :1: SD).

Central‘I Medials Con trol‘
_ Impacted Unimpacted Impacted Unimpacted
Gu 1.23 i- 0.29 1.28 :1: 0.19 0.55 i 0.14” 0.88 i 0.36
Gr 0.43 :1: 0.12 0.43 :1: 0.08” 0.19 m 0.07‘+ 0.26 r 0.07
‘ N = '8
:N = 7
# Signiﬁcantly less than Unimpacted group, by paired t-test m < 0. 05).
OSigniﬁcantly greater than Unimpacted Medial group, by ANO VA Bonferroni t-test @ <
. 05).
+ Signiﬁcantly less than Control and Impacted Central group, by ANO VA Bonferroni t-
test @ < 0. 05).

 

1.06 i 0.35
0.35 :1: 0.11

 

70

Table 3: Thickness of the subchondral plate of the patella in chronic and control animals
12 months post impact (mm, mean i SD).

 

 

 

Impact Location
Group Medial Central Lateral
M 1' 1' Impacted 0.74 r 0.23‘ 1.10 :1: 038‘” 0.60 i 0.20

Unimpacted 0.49 :1: 0.13 0.63 :1: 0.26 0.51 :t 0.21
Impacted 0.48 :1: 0.06 0.69 i 0.12 0.46 :t 0.08

 

 

1)
central Unimpacted 0.40 i 0.08 0.69 i 0.08 0.42 i 0.08
Control" 0.52 r 0.14 0.72 r 0.14 0.52 :1: 0.12
‘N=8
bN=4

. Significantly thicker than Unimpacted group, by paired t-test 02 < 0. 05).
“ Significantly thicker than Control group, by ANOVA Bonferroni t-test @ < 0.05).

Table 4: Average maximum shear stresses and average mean stresses in the central
region of the subchondral bone of the Finite Element Model for central- and medial-

oriented impact groups (MPa, mean :1: SD).

FEM model Central Medial

Tm Gnoml Irnax 0mm]
10.51 i 3.37 -30.82 :1: 6.02 9.29 :1: 3.32 -21.43 :1: 8.71

#
1
2 9.58 :1: 2.85 -26.37 :1: 5.82 8.81 i 2.33 -19.45 :t 4.32
3 8.95 :1: 1.26 -26.16 :1: 1.53 8.50 i 1.58 -24.22 :1: 1.63
4 14.24 :1: 4.76 -29.01 :1: 10.16 6.03 :1: 1.91 -12.95 i 1.45
5 8.95 :1: 1.26 -26.17 :1: 1.54 6.74 :1: 2.07 -15.58 i 1.59

MEAN 10.45 ' -27.71 7.87 -18.73
SD 2.21 2.12 1.41 4.51

. Significantly greater than Medial group, by Unpaired t-test @ < 0.05).

FIGURES

 

 

 

 

 

 

 

 

 

 

 

 

 

[F In — ' ’1
\N K
-_ - J -1.
(A) (B) (C)

Figure 1. Impact experiments were performed by dropping a rigid mass onto the
patellofemoral joint with 6.0 J of energy (A). medial-oriented impacts required the femur
to be positioned so that the impact load was aligned along the medial side of the anterior
patella and directed towards the lateral patellar facet (B), while the Central-oriented
impacts were aligned along the medial-lateral center of the patello femoral joint (C).

72

Posterior (patellofemoral)

Cartilage
(B)
Lateral

 

Bone
(C)
Anterior

(A)

Figure 2. A two-dimensional non-contact ﬁnite element model (A) of the patella
consisting of 1537 3-noded plane strain elements. The lateral facet on the retropatellar
articular cartilage (B) surface where lesions were generated in acute experiments was
used for documenting the impact induced stresses. The central one third of the
subchondral bone (C) where chronic thickening is typically observed was used to
document the impact induced stresses.

73

 

(A) (B)

O Central-Oriented
A Medial-Oriented

 

451
1€404!
d35j
E30,»
2251
an]
215.
lI-10i
51
ol .

medial lateral
(C)

Figure 3. Contact pressure distributions for medial-oriented (A) and central-oriented (B)
impacts, as measured by pressure sensitive ﬁlm. The graph (C) represents the digital
processed mean contact pressures of the medial-oriented (A) and central-oriented (0)
impacts.

Medial-Oriented Central-Oriented

Cartilage

‘OWVOMhUN-‘O

2.222

4.444

6.667

8.889

11.111 Bone
13.333

15.556

17.773

20

 

IIIIIIDUU IIIIIIDDD

Figure 4. Acute impact-induced maximum shear stress (MPa) distributions in the lateral

facet of retropatellar cartilage and central region of underlying subchondral bone medial-
oriented and central-oriented impacts.

75

CHAPTER 5

THE EFFECT OF LOADING RATE ON THE DEGREE OF ACUTE INJURY
AND CHRONIC CONDITIONS IN THE KNEE AFTER BLUNT INIPACT

Ewers BJ, Jayaraman VM, Banglmaier RF, Haut RC

ABSTRACT

Lower extremity injmies due to automobile accidents are often overlooked, but
can have a profound societal cost. Knee injuries, for example, account for approximately
10% of the total injuries. Fracture of the knee is not only an acute issue but may also have
chronic, or long term, consequences. The criterion currently used for evaluation of knee
injuries in new automobiles, however, is based on experimental impact data from the 70’s
using seated human cadavers. These studies involved various padded and rigid impact
interfaces that slightly alter the duration of contact. Based on these data and a simple
mathematical model of the femur, it appears ﬁ'acture tolerance increases as contact
duration shortens. In contrast, more recent studies have shown mitigation of gross
fractures of the knee itself using padded interfaces. The use of padded interfaces,
however, result in coincidental changes in contact duration and knee contact area.
Therefore, it is difficult to extract the direct effect of loading rate on fracture tolerance of
the knee. The object of the current study was to isolate the effect of loading rate alone on
fracture tolerance of the human knee joint. Paired experiments were conducted on eight

pairs of isolated cadaver knees impacted with a rigid interface to approximately 5 kN at a

76

high (5 ms to peak) or low (50 ms to peak) rate of loading. Gross fracture and occult
microﬁactures of the knee joint were documented A second part of the study examined
some chronic effects of loading rate on “subfracture” injuries in an animal. Thirty-four
rabbits were subjected to a “subfracture” knee load at the same rates as used in the human
studies. Alterations in the mechanical properties of retropatellar cartilage and thickening
of subchondral bone were documented out to one year post “subfracture” trauma to the
joint.

The current study documented an opposite effect than that expected based on 70’s
experiments with seated cadavers. There was an increase in the number of gross fractures
and occult micro ﬁ'actures in high versus low rate of loading experiments. A similar effect
was also seen in the “subﬁacture” chronic animal experiments, which showed relatively
more degradative change in the mechanical properties of cartilage following high versus
low rate of loading experiments. There was also a signiﬁcant increase in subchondral
bone thickening underlying cartilage and increased ﬁssuring of cartilage in high versus
low rate of loading experiments. The current study suggests a relative decrease in
tolerance of the knee at high versus low rates of loading in acute experiments with human
cadavers and in the chronic setting with animals. Therefore, it would appear that rate of
knee loading may be an important issue in establishing a ﬁrture injury criterion for the
knee itself.

INTRODUCTION

Lower extremity injuries due to automobile accidents are often overlooked, but

can have a profound societal cost. One study documented that 40 % of total annual

trauma treatment charges from automobile collisions were for lower extremity injuries

77

(MacKenzie et al., 1988). Additionally, it has been estimated that approximately $21.5
billion is spent annually on treatment, rehabilitation, and lost workdays associated with
lower extremity injuries (Miller et a1, 1995). A previous study, analyzing a survey of
accident statistics (The National Accident Sampling System), documents that injuries to
the knee account for approximately 10 % of total annual injmies (Atkinson et al., 1998).
Knee injuries in an automobile accident typically occur when the knee contacts the
instrument panel, or a front knee bolster. Gross fracture of the femur or patella may occur
in severe load cases, however, the majority of knee injuries are non-fracture, such as
contusions, abrasions, and lacerations (Atkinson et al., 1998). Fracture of the knee is not
only an acute issue but may initiate chronic problems, such as a secondary post-traumatic
osteoarthritis (OA) (Chapchal, 1978; States, 1986; Wright, 1990). Less severe damage to
joint tissues, such as ﬁbrillation of articular cartilage or occult microcracks at the
cartilage-bone interface can also occur without radiographic evidence of bone ﬁacture
(Pritsch et al., 1984). Subfracture injuries, such as occult microcracks (bone bruises),
have been documented in recent years with improvements in magnetic resonance
imaging. Often cartilage overlying certain types of bone bruises is ﬁssured, softened, or
indicates an occult fracture (Johnson et al., 1998). It has be suggested that these occult
microcracks of bone are precursors to a gross ﬁacture (Atkinson et al., 1997).

The cm'rent criterion used for evaluation of knee injuries in new automobiles is
based on axial load in the femur. This criterion was largely established from experimental
impact data on seated human cadavers that caused gross fracture of the patella, femur, or
hip (Powell et al., 1975; Melvin et aL, 1975; Patrick et al., 1965; Viano and Stalnaker,

1980). These hallmark studies from the 70’s were performed with rigid, slightly padded

78

and heavily padded impact interfaces. The rigid interfaces were shown to generate more
patellar fractures than did the padded interfaces. The padded interface experiments
generally indicated a slightly increased contact duration of 19.9 i 9.8 ms versus 3.9 :1: 0.7
ms for the rigid interface experiments. Atkinson et al. (1997), using impacts to isolated
cadaver knees, have also shown that using a padded interface causes increased contact
area on the knee compared to the use of a rigid interface. These increased contact areas
were shown to result in a reduction of the impact-induced stresses in joint tissues,
resulting in relatively less joint damage than in comparative load experiments with a rigid
interface. Hayashi et al. (1996) documented similar results. These data would suggest
impacts with padded interfaces yield longer contact durations on the knee and increase its
fracture tolerance. Using a ﬁnite element model of the human femur under impact loads
Viano and Khalil (1976) suggest that fracture tolerance is constant for pulse durations
longer than approximately 20 ms, whereas the ﬁacture tolerance of the femur is increased
as pulse duration is shortened The model predicts a doubling of the fracture load as the
pulse duration decreases from 20 ms to approximately 2 ms. The experimental data from
the rigid and padded cadaver impacts from the 70’s was also ﬁt to the model of Viano
and Khalil (1976). King et a1. (1973) uses this result and states “In the case of short-
duration impacts, this femur load criterion is much too conservative because the ultimate
strength of the femur bone increases markedly due to the effect of strain rate sensitivity
of bone on fracture tolerance”. So, the accepted notion is that the fracture tolerance of the
knee complex also increases for short versus long duration loads. This suggests that
impacts on the knee with padded interfaces, which generally produce longer contact

durations, would result in a decrease in the ﬁacture tolerance of the knee rather than an

79

increase as suggested by Atkinson et a1 (1997). This being true would indicate that the
rate of loading effects on tissue strength in short duration pulses must have more of an
effect on knee ﬁacture tolerance than does the reduction of impact stress in joint tissues
due to an increase in contact area In fact, studies on isolated specimens of cortical bone
have documented signiﬁcant increases in the tensile and compressive strength of cortical
bone with increasing strain rate (McElhaney, 1966; Wright and Hayes, 1976). However
this dependency of strength on strain rate appears relatively weak, with only a 20 — 50 %
increase of strength with an order of magnitude increase in strain rate. And therefore any
difference in the strength of the knee for the contact duration of 2 to 20 ms would not
explain the 70’s data as presented by Viano (1977), unless the patella-femur-hip complex
is more rate sensitive than that. represented in experimental data ﬁom isolated bone
coupons.

It is not currently possible, however, to extract the direct effect of loading rate on
impact to the knee joint from the previous 70’s studies, since padded interfaces result in
coincidental changes in contact duration and contact area. The object of the current study
was to isolate the effect of loading rate alone on tolerance of the intact human knee joint
to blunt impact.

Paired experiments were conducted on eight pairs of isolated human cadaver
knees, which were impacted with a rigid interface at a high rate of loading (5 ms to peak)
or a lower rate of loading (50 ms to peak). Gross fractures and occult microfractures were
documented after loads of approximately 5 kN were applied to the knee specimens. In a
second part of the study some of the chronic effects of rate-of-loading on “subfracture”

injmy to the knee were investigated in an animal Thirty-four rabbits were subjected to

80

loads on the knee with a rigid interface at the same high or low rate of loading used in the
human studies. Alterations in the mechanical properties of retropatellar cartilage and
thickness of underlying subchondral bone were documented out to one year post
“subfracture” trauma.

METHODS

HUMAN IMPACT S - ACUTE STUDY

Paired impact experiments were conducted on eight pairs of isolated human knee
joints (aged 74.8 :1: 13.3 years). The limbs were obtained via a willed-bodied program
(see Acknowledgment), and stored at -20°C until testing. The limbs were thawed at 27°C
for 24 hours. The tibia and femur were sectioned 15 cm proximal and distal to the knee.
Superﬁcial tissues proximal to the femoral condyles were excised, the femoral shaft was
cleansed with 70% alcohol, and potted in a cylindrical aluminum sleeve with room
temperature curing epoxy using a previously documented protocol (Atkinson and Haut,
1995)

One knee from each subject pair was randomly selected for a single
patellofemoral joint impact at a high rate of loading. This protocol followed previous
studies by our laboratory, and results in peak loads being applied in approximately 5 ms
(Atkinson et al, 1997). The potted femur and sleeve were mounted horizontally to a rigid
test ﬁxture with a ﬂexion angle of 90° by a loosely tied tether to the quadriceps tendon
and the test ﬁxture (Figure 1). In order to measure patellofemoral joint and anterior
patella contact pressures and areas, pressure sensitive ﬁlm was used (Prescale, Fuji Film
Ltd, Tokyo, Japan). On the anterior side of the patella, low range (0-10 MPa), medium

range (10-50 MPa), and high range (50-126 MPa) pressure ﬁlms were stacked together.

81

In the patellofemoral joint, high pressure ﬁlm was not used, because previous studies
have indicated that pressures within the joint do not exceed 50 MPa (Atkinson et al.,
1997). The ﬁlm was encased in a polyethylene packet to prevent exposure to body ﬂuids
and reduce shear loading artifacts. The packet was inserted into the joint prior to each
impact through medial and lateral incisions, and wrapped around the anterior surface of
the patella (Figure 1). The joint was impacted with a rigid interface at an energy of 27 J.
This was based on a logistic regression analysis from a previous study that suggests 27 J
would provide the greatest risk of producing an occult trauma, with a 50% chance of
generating a gross ﬁacture (Atkinson et al., 1997; Atkinson and Haut, 1995). Impact was
delivered to the patellofemoral joint via a 4.5 kg horizontal impact sled accelerated along
steel rails to the required pre-impact velocity by a pneumatic cannon. A 5.7 x 5 .7 cm,
6061-T6 aluminum impact interface was mounted to the front of a load transducer
(Model 3173-2k, Lebow Products, Troy, MI). Load data were collected at 20 kHz
continuously via a 16 bit A/D board (Model DAS 1600, Computer Boards Inc.,
Mansﬁeld, MA), and inertially compensated with an accelerometer (Model #72624a-
2000, Endevco, San Juan Capistrano, CA). The velocity of the impact sled was measured
prior to the impact event using two infrared optical sensors separated by 76.2 mm (Part
No. ORSl8-ND, Digi-Key Corporation, Thief River Falls, MN). The energy input into
the bee was further computed by integration of the force-deﬂection curve. The
deﬂection data was determined by double integration of the deceleration curve from the
impact mass.

Following the high rate of loading impact, the opposite limb underwent a single

patellofemoral joint impact using a low rate of loading. The potted femur and sleeve were

82

vertically mounted to the base of a servo-hydraulic material test machine (Model 1331,
Instron Corp., Canton, MA), with the joint ﬂexion angle maintained at 90°. The same
impact interface was mounted to the front of a load transducer (Model #10101a—2500,
Instron Corp). Prior to application of load, pressure ﬁlm was inserted into the
patello femoral joint, as described above. Impact was delivered to the patello femoral joint
via the actuator of the test machine under computer control (Flaps 5+, Instron Corp). The
impact pulse was a haversine waveform designed to generate the required peak load
within 50 ms. The desired peak load was based on the maximum load generated in the
contralateral limb during the high rate of loading experiment. Load data were recorded
continuously at 1000 Hz. Energy input into the bee was computed by integration of the
load-displacement curve, based on hydraulic actuator motion.

The individual pressure ﬁlms were scanned on a ﬂatbed scanner (ScanJet 6300C,
Hewlett Packard, Singapore) using commercial software (Photo Styler, Aldus
Corporation, Seattle, WA). The scanning resolution was 101 dpi, which is equivalent to
25 pixels per square millimeter. The images from the pressure sensitive ﬁlm were
represented digitally as gray scale values using public domain software (NIH, ver. 1.6).
The gray scale values were then converted to pressure using a previously established
methodology (Atkinson et al., 1998). Brieﬂy, calibration tests on low and medium
stacked ﬁlm were performed using a haversine waveform to generate peak loads within
50 ms with the servohydraulic-testing machine. Past studies have shown that ﬁlm
calibrated at this rate can accurately represent loads with a time to peak of 5 ms (Haut et
al., 1995). Peak loads ranged from 135 N to 6600 N, in increments of approximately 150

N. The calibration ﬁlms were scanned, as stated above, and calibration curves were

83

created relating the grayscale values to pressure. The calibration curves also determined
the range for which each ﬁlm was usable under these relatively high rates of loading
(Figure 2). The ranges were 3 - 10 MPa and 10 - 49 MPa for the low and medium ﬁlms,
respectively. The same calibration procedure was performed for the low, medium, and
high pressure ﬁlm stacks. A macro program (Microsoft Excel 97) was written to
eliminate sections of the digitized pressure ﬁlm images that were not in the useable
range. The images from the pressure ﬁlms were combined to obtain the average contact
pressure and area on the anterior surface of the patella and within the patello femoral joint
(Figure 3). i

The patellae were grossly examined for evidence of gross bone fracture, and then
placed in 10 % buffered formalin for 10-14 days. The tissue was decalciﬁed in 20 %
formic acid Each patella was cut into a 2-3 mm sample block on the center of the medial
and lateral facets, processed for routine parafﬁn embedding, slab-cut and stained with
Safranin O-Fast Green. All sections were examined in light microscopy at 12-400 power
and photographed. Injuries were classiﬁed as to the number of gross ﬁ'actures,
histological fractures (slight fractures through the cartilage and into the underlying bone),
and the number of occult microcracks or splits in the zone of calciﬁed cartilage that do
not propagate to the cartilage surface.

' Paired t-tests were performed to evaluate the following differences in impact
characteristics for high versus low rate experiments: peak load, time to peak load, contact
duration, input energy, retropatellar contact area and pressure, and contact area on the
anterior surface of the patella. McNemar’s test was used to examine the difference in the

number of gross ﬁactures between high and low rate of loading experiments. Signed

Rank tests were used to examine the differences in the number of histological fractures
and occult microﬁactures between the high and low rate of loading experiments.
Signiﬁcant statistical differences were reported for p < 0.05.

ANIMAL MODEL. - CHRONIC STUDY

Thirty-four mature Flemish Giant rabbits (4.9 :t 1.2 kg; 6-8 months old) were
used Sixteen animals received a blunt insult to the right patellofemoral joint with a high
rate of loading (~5 ms to peak). The impact protocol has been described in earlier studies
(Newberry et al., 1998). Brieﬂy, a 1.33 kg mass with a ﬂat 1 in. diameter impact surface
was dropped from 0.46 m onto the right ﬂexed (120°) hind limb. During the impact the
animals were maintained at a surgical plane of anesthesia using Isoﬂurane and oxygen
(2% Isoﬂurane). The impactor was stopped electronically to prevent multiple impacts. A
load transducer (Sensotec, Columbus OH: model 31/1432, 500 1b capacity) recorded the
impact load at 10 kHz. The remaining eighteen animals received a blunt insult to the right
patellofemoral joint with a low rate of loading (~50 ms to peak). These impacts were
performed as described above, but with a servo-hydraulic testing machine (Model 1331,
Instron Corp) using a single haversine waveform to generate loads comparable to the
average load generated in the high rate of loading experiments.

Alter trauma, all animals received one injection of Butorphenol for post-surgical
pain, and were carefully monitored by the veterinary technician. After a ﬁve day period
of rest following the impact, a daily exercise program was initiated for all animals. This
program consisted of ten minutes of exercise, ﬁve days a week, on a treadmill running at

0.3 mph (Oyen-Tiesma et al., 1998). The exercise schedule lasted the entire length of the

85

study. This study was approved by the MSU All-Univesity Committee on Animal Use
and Care.

Eight animals from each of the high and low rate of loading groups were
sacriﬁced at 4.5 months post-impact. The remaining animals were sacriﬁced at 12
months. Immediately after sacriﬁce, the patellae were excised for mechanical indentation
tests on the retropatellar cartilage, as previously described (Ewers et al., 2000). Brieﬂy,
the patellae were immersed in a room-tempeature phosphate buffeed saline bath during
the indentation tests. These tests were performed with a custom built instrument that used
a compute opeated steppe motor (Physik Instruments, Waldbronn, Gemany: model M-
168.30) to indent the cartilage. A 1 mm diamete ﬂat, non-porous probe was pressed into
the cartilage 0.1 mm in 30 ms and maintained for 150 seconds. The resistive loads wee
measured (Data Instruments, Acton MA: model JP - 25, 25 lb capacity), ampliﬁed, and
collected at 1000 Hz for the ﬁrst second, and 20 Hz theeafte. Afte the tissue was
allowed to recove for 5 minutes, the test was repeated with a 1.5 mm diamete, ﬂat non-
porous probe at the same location. The probe was then replaced with a needle, which was
slowly pressed into the cartilage to determine tissue thicbess at that location. The above
procedure was performed at two sites on the lateral retropatellar facet adjacent to impact-
induced surface lesions, if present.

. The analysis of the mechanical data from the retropatellar cartilage was
performed using a biphasic model having a transversely isotropic (TI) solid structure. A
technique has been developed to determine the mechanical constants using the
indentation relaxation data (Garcia, 1998). Brieﬂy, the model assumed isotropy in the

plane of the cartilage and the Poisson’s ratio in this plane (v12) was set equal to zeo. The

86

remaining four independent mechanical propeties and two permeabilities, namely: a
thicbess direction modulus (E33), an in-plane modulus (Eu), a shear modulus (613), a
Poisson’s ratio (v13), a thicbess direction permeability (kg), and an in-plane permeability
(k1), wee then computed using the Garcia algorithm.

Afte the mechanical tests, each patella was stained with India ink to highlight
surface lesions and then photographed The patellae wee then placed in 10% buffeed
formalin for a week, and decalciﬁed in 20% formic acid for anothe week. Tissue blocks
wee cut medial to lateal across the patellae in the area bown to produce high contact
pressure during impact to the joint (Haut et al., 1995). The tissue blocks wee processed
in parafﬁn, and six sections, 8 11 thick, wee cut medial to lateal. The sections wee
stained with Saﬁ'anin O-Fast Green and examined unde light microscopy at 12-400
powe. The thicbess of the subchondral bone plate undelying retropatellar cartilage was
measured at 25x with a calibrated eye-piece at the cente of each facet and mid-line of the
patella by a single investigator (BE), using existing methods (Newberry et al., 1998).

To help assess cartilage matrix damage, the gross photographic images wee
scanned and the total length of surface ﬁssuring was measured for each patella using '
image software (Sigma Scan, SPSS Inc., Chicago, IL). Using the histological sections,
the numbe and aveage depth of the surface ﬁssures wee also measured at 40x with a
calibrated eye-piece by one investigator (ES) (Figure 4).

RESULTS
HUMANIMPACTS - ACUTE STUDY
The typical load-time response for the high rate of loading expeiments was a

nearly symmetrical haversine, while the low rate of loading expeiments resulted in a

87

skewed havesine (Figure 5). Thee was no signiﬁcant diffeence in peak loads between
the high rate of loading (4.6 :1: 1.0 kN) and the low rate of loading (4.5 i 1.2 kN)
expeiments (Table l). The high rate of loading expeiments produced an aveage time to
peak of 4.9 :l: 0.9 ms, while the low rate of loading expeiments resulted in a time to peak
of 54.2 :1: 2.6 ms. The total contact duration for the high rate of loading experiments was
10.3 :1: 2.6 ms, while the low rate of loading experiments produced a contact duration of
218.8 :1: 23.4 ms. While not statistically signiﬁcant, the input enegy for the high rate of
loading experiments (26.2 i 9.6 Nm) had a tendency to be greate than the low rate of
loading experiments (15.8 i 8.6 Nm) (Figure 6).

The retropatellar pressure ﬁlm indicated that the region of contact typically had a
“kidney” shaped appearance for the high and low rate of loading experiments. Thee wee
no signiﬁcant diffeences in the retropatellar contact area between the high at 749.6 i
185.8 mm2 and the low rate of loading experiment at 587.3 i 224.1 mm2 (Table 2).
Retropatellar contact area ﬁ'om two high rate of loading expeiments had to be removed
from the study because of contamination from unbown sources. The retropatellar
contact pressures wee also similar for experiments at the high rate having a contact
pressure of 13.7 i 3.3 MPa, while the low rate pressure was 14.4 i 2.7 MPa. The
pressure distribution on the anterior surface of the patella for the high and low rate of
loading experiments was typically circular. Thee was also no signiﬁcant diffeence in
the area of contact between the high (740.1 :1: 176.5 m2) and the low (586.9 :1539
m2) rate of loading expeiments.

Six pairs of specimens wee histologically processed (Table 3). As predicted from

earlie studies, the high rate of loading expeiments produced transvese patellar fractures

88

in 50 % of the expeiments (Figure 7). On the othe hand, the low rate of loading
experiments resulted in gross ﬁacture of the patella in only 17 % of the cases. The
characte of the gross fractures for high and low rate loading was similar. The high rate of
loading resulted in approximately twice the numbe of histological fractures than the low
rate of loading expeiments. The occult mieocracks wee typically horizontal (Figure 8).
The high rate of loading impacts also produced approximately 50 % more microcracks
than the low rate of loading expeiments, for the same impact load intensity.
ANIMAL MODEL - CHRONIC STUDY

Examination afte impact and daily obsevations by a certiﬁed veteinary
technician (JA) assigned only to this project indicated no noticeable joint effusions, and
the rabbits did not appear to favor their impacted limbs. A typical load-time response for
the high rate of loading expeiments was a nearly symmetrical haversine, while the low
rate of loading experiments resulted in a skewed havesine. The aveage peak impact
loads of 590 :I: 45 N and 630 :1: 94 N for the low and high rate of loading, respectively,
wee not statistically diffeent. Howeve, as expected, thee wee signiﬁcant increases in
time-to-peak and contact duration for the low rate of loading compared to the high rate of
loading experiments. The low rate of loading experiments resulted in a time-to-peak of 60
i 8 ms and a contact duration of 257 d: 26 ms. In contrast, the high rate of loading
experiments produced a time-to-peak of 4.2 :1: 0.5 ms and a contact duration of 10.5 i 1.8
ms.

The data for one animal in the low rate of loading group at 12 months was
removed from the study because of excessive pathology in both limbs. Gross visual

examination of the remaining patellae in the study, afte wiping with India ink, indicated

89

surface ﬁssures on the retropatellar cartilage in 13 of 17 low rate of loading impacts and
in all high rate of loading impacts. These lesions typically ran proximal to distal on the
lateal facet near the central ridge of the patella.

The high rate of loading experiments resulted in greate changes in the
mechanical propeties of the traumatized retropatellar cartilage compared to the low rate
of loading experiments at both timepoints. Thee was no signiﬁcant diffeence in any
mechanical parameters between unimpacted limbs of the high and low rate of loading
experiments at both 4.5 and 12 months post-trauma. At 4.5 months afte a high rate of
loading on the patellofemoral joint, statistically signiﬁcant reductions of 39% and 29% in
E“ and E33, respectively, wee documented for the traumatized cartilage compared to the
unimpacted limb. Thee was also signiﬁcant increases of 57% and 119% in
pemeabilities k; and k3, respectively, for the traumatized versus contralateral,
unimpacted cartilage (Table 4). The low rate of loading experiments resulted in 32% and
21% reductions in E1 1 and E33, respectively, for the traumatized cartilage compared to the
contralateal unimpacted limb, and a 48% increase in permeability k1. At 12 months post-
trauma, the high rate of loading expeiments resulted in 51% and 32 % reductions of E1 1
and E33, respectively, for the traumatized cartilage compared to the unimpacted
contralateal limb. A 51% and 65% increase in k1 and k3, respectively, was obtained in
the traumatized vesus the unimpacted cartilage at this timepoint. Howeve, the low rate
of loading resulted in only 25% and 22% decreases in EU and E33, respectively, for the
impacted cartilage compared to the unimpacted limb, and no signiﬁcant diffeence in

pemeability at 12 months past impact.

90

The high rate of loading expeiments caused more thickening of the undelying
subchondral bone compared to the low rate of loading expeiments. At 4.5 months, the
high rate impacts resulted in 27%, 28%, and 18% increases in subchondral bone
thicbess under the lateal, central, and medial sites on the patella, respectively, for the
impacted patellae compared to the contralateal unimpacted limbs (Table 5). The low rate
experiments resulted in only 10% and 20% increases in subchondral bone thicbess at the
lateal and central sites, respectively, for the impacted patellae compared to the
unimpacted limbs, while thee was no signiﬁcant increase at the medial site. These types
of results wee also documented at 12 months post-trauma The high rate of loading
expeiments again resulted in 29%, 60%, and 44% increases in subchondral bone
thicbess for the impacted patella compared to the unimpacted limb at the lateal, central,
and medial sites, respectively. The low rate impacts resulted in only 9% and 24%
increases in subchondral bone thicbess at the lateal and central sites, respectively, with
no signiﬁcant increase at the medial site (Figure 9). Importantly, at the lateal and medial
sites the subchondral bone of the impacted patellae from the high rate of loading
expeiments was signiﬁcantly thicke than that of the impacted patellae from the low rate
of loading expeiments.

The histological sections showed that the typical depth of surface ﬁssures on
traumatized cartilage was approXimately 30 - 40% of the cartilage thicbess (Figure 10).
The aveage ﬁssure depth for the impacted cartilage was signiﬁcantly greate than any
ﬁssures appearing on the unimpacted limbs (Table 6). Thee was a signiﬁcant increase in
the numbe of ﬁssures for the traumatized cartilage compared to the contralateal

unimpacted sides. At 12 months, the high rate of loading experiments produced

91

signiﬁcantly more surface ﬁssures on retropatellar cartilage than in the low rate of
loading expeiments. The high rate of loading also resulted in greate length of total
surface ﬁssuring on retropatellar cartilage than in the low rate of loading expeiments at
12 months post-trauma In effect, the high rate of loading expeiments resulted in more
surface ﬁssuring of the retropatellar cartilage than produced in the low rate of loading
experiments.
DISCUSSION AND CONCLUSIONS

The criterion for evaluation of bee injuries in new automobiles is based on
experimental impact data from the 70’s using seated human cadavers. These studies used
various padded impact intefaces that only slightly alteed contact durations. Howeve,
Viano and Khalil (1976) used a FEM model of the human femur to suggest that its
fracture toleance would ineease for shortened contact durations. Viano (1977) has also
shown that the experimental data from the 70’s whole cadave expeiments ﬁt this model.
On the othe hand, Atkinson et al. (1997), document that by using a padded inteface the
contact duration increases and gross ﬁ'acture can be mitigated, suggesting that fracture
toleance of the bee increases with increasing contact durations. Howeve, the use of
padded intefaces results in coincidental changes in the contact duration and contact area.
Therefore, it is not cm'rently possible to extract the direct effect of loading rate on
fracture toleance of the be. The object of the current study was to isolate the effect of
loading rate alone on toleance of the bee joint to blunt trauma.

The current study documented an opposite rate of loading effect than that
expected based on a model of the femur alone (Viano and Khalil, 1976). As the

experimental time to peak load was decreased from 50 ms to 5 ms, thee was an increase

92

in the frequency of gross fractures, indicating a decrease in fracture toleance of the be
as contact duration was shortened This result likely suggests that the model analysis of
the femur alone, presented by Viano and Khalil (1976), may not be suitable for studying
fracture of the patella. The mechanism of injury is signiﬁcantly diffeent and because of
the way soft tissue and the bones make up the be, the rate sensitivity may be radically
diffeent from that of the femur. Theefore, a bee load toleance criterion should
probably not be based on the curve presented in Viano (1977). Thee was also a doubling
of the numbe of histologically obseved fractures, and a 50% increase in occult
microcracks for the high versus low rate of loading experiments. Histological fractures
and occult microcracks have been suggested to be precursors of gross fractures, and
theefore these data support the documented increase in the numbe of gross fractures
geneated in the high rate of loading expeiments (Atkinson et al., 1997). Occult
microcracks of subchondral bone have also been implicated in so-called “subﬁ'acture”
injuries to the bee (Johnson et al., 1998).

The animal was used in this study to examine some effects of the rate of loading
in a chronic expeiment. Groups of rabbits wee exposed to loads with the same rate of
loading used in the human cadave impacts and examined 4.5 and 12 months post-
trauma. Thee was a tendency for increased softening and increased permeability of the
traumatized versus unimpacted retropatellar cartilage in the high rate of loading
expeiments compared to the low rate of loading expeiments. Thee was also
signiﬁcantly more ﬁssuring on the surface of this cartilage in the high versus low rate of
loading expeiments. Furthemore, thee was signiﬁcantly more subchondral bone

thickening in traumatized patellae for the high versus low rate of loading experiments.

93

This thickening has been suggested due to a remodeling of the bone as a result of
microﬁactures (Radin et al, 1990). These data could then be in agreement with the
results from the human impacts that documented more occult microcracks in the high
vesus low rate of loading experiments, howeve microcracks pe se wee not seen in the
animal patellae.

One possibility for the reduction in injuries at low versus high rates could be due
to rigid body motion. The low rate of loading might have allowed sliding of the patella,
which could result in large contact areas. The pressure sensitive ﬁlm, howeve, showed
no increase in contact areas for the low versus high rate of loading experiments. This may
be due, howeve, to the fact that the ﬁlm was encased in a polyethylene packet, which
was meant to mitigate or reduce shear loading artifacts on the ﬁlm. Since the force-
deﬂection curve was stiffe for the high vesus the low speed expeiments, thee is still a
strong possibility that more of the available impact enegy on the bee went into patellar
deformation in the high rate of loading experiments resulting in a high incidence of gross
and microfractures of the patella. Howeve, in orde to conﬁrm this conclusion, high
spwd ﬁlms would be required Atkinson et al. (1997) has suggested that acute impact-
induced stresses are responsible for injury. Even if thee was some rigid body movement
of the patella, the retropatellar contact pressures and anterior contact areas wee the same
for the high and low rate of loading experiments. This would suggest that impact-induced
stresses in the joint tissues wee the same for both rates of loading, and theefore the two
rates of loading should result in the same ﬁ'equency of injury.

Anothe possibility may be the rate-sensitive or viscoelastic nature of bone. A

load applied ove a short time period would cause less tissue strain, and theefore would

94

be less damaging for a strain failure criterion than the same load applied ove a longe
time frame, assuming the strain at failure does not depend on time. Furthemore, previous
studies have documented with isolated bone coupons that the ultimate strength of bone is
increased with increasing strain rate (McElhaney, 1966; Wright and Hayes, 1976). This
would not seem to explain our results, because impact loads wee the same in our high
and low speed tests and we documented more fractures in the high rate test. Theefore,
our results do not seem to be explained based on the documented strain rate sensitivity of
bone strength alone. On the othe hand, the mechanical behavior of isolated bone
specimens may be diffeent than bones in a whole joint model.

A third possibility may be proposed assuming bone behaves similar to othe
engineeing materials. An ineease in strain rate typically tends to promote brittleness in
engineeing materials (Felbeck and Atkins, 1984). This would result in more enegy
going into crack propagation to develop a vesus plastic zone at the crack tip for high
compared to low rate of loading experiments. The increase in crack propagation could
lead to more connected cracks resulting in a greate numbe of microcracks. This process
would result in a greate ﬁequency of gross fractures in the high compared to the low rate
of loading expeiments.

A forth possibility is based on the complex microstructure of bone. Since bone is
composed of solid and ﬂuid phases, it has been modeled as poroelastic (Cowin, 1999).
Experimentally, whole bones have shown a degree of hydraulic stiffening (Ochoa et al.,
1997). The same effect has been surmised in a ﬂuid-saturated model of cortical bone
(Zeng et al., 1994). Hydraulic stiffening of cortical bone is likely to occur unde impact

loading (Zhang et al., 1998). It has been documented that the pore ﬂuid pressure

95

relaxation time is approximately 5 ms, and that for a uniaxial impact the ﬂuid pressure
differential in the bone can approach 13% of the applied boundary stress (Zhange et al.,
1998). These intenal compressive ﬂuid pressures must be balanced by tensile stresses in
the solid phase of the bone. We hypothesize that these increased tensile stresses in the
solid phase could be responsible for the geneation of occult microcracks, and eventual
gross fracture of bone unde an impact load This notion would support our obsevation
of a greate number of microscopic and gross ﬁactures in high versus low rate of loading
experiments.

A similar poroelastic hypothesis has been described for articular cartilage. Garcia
et al. (1998) suggested that increases in tensile stress in the solid phase of cartilage may
be responsible for impact-induced ﬁssures. Indeed, with a simple geometry and a
poroelastic analysis, Garcia et a1. (1998) documented the largest tensile stresses in the
solid phase of cartilage to be at the surface, whee ﬁssures have been documented
following impact experiments (Newberry et a1, 1998). The current study documented
more ﬁssures in the high versus low rate of loading expeiments. This may be the result
of the high rate of loading causing large ﬂuid pressures and correspondingly highe
tensile stresses in the solid matrix of the retropatellar cartilage. A previous study that
traumatized isolated cartilage explants supports this hypothesis by documenting more
surface ﬁssures with a high vesus low rate of loading (Ewers et al., 2000).

One implication of these new data is that previous impact data from our
laboratory with a high rate of loading maybe too conservative with respect to normal
instrument panel loading scenarios (Atkinson et al., 1997). The current study would

suggest a relative decrease in the toleance of the be for high versus low rates of

96

loading acutely using the cadave and chronically in an animal. Thee is still, howeve, a
need for a better understanding of the behavior of cartilage and bone as poroelastic
materials at these relatively high rates of loading. While it has been suggested that the
impact-induced stresses in joint tissues cause gross fractures, based on the current study it
seems that the rate of loading may be anothe important issue in establishing an injury
criteion for the bee joint.

ACKNOLWEDGMENTS

This study was supported by a grant for the Centers for Disease Control and

Prevention (R49/CCR503607). Its contents are solely the responsibility of the authors and

do not necessarily represent the ofﬁcial views of the Centers for Disease Control and

Prevention The authors wish to gratefully acbowledge Jane Walsh for preparation of

the histological slides, Jean Atkinson for care of the animals, Eric Sevensma for help

reading the slides, and Cliff Beckett for technical assistance in this study. We also thank

Mr. RS. Wade, Director of the Anatomical Sevices Division / State Anatomy Board,

University of Maryland for help procuring human test specimens.

REFERENCES

l. Atkinson, RI and Haut, RC. (1995) Subﬁ'acture Insult to the Human Cadave
Patellofemoral Joint Produces Occult Injury. Journal of Orthopaedic Research 13(6),
pp. 936-944.

2. Atkinson, P.J., Garcia, J.J., Altieo, N.J., and Haut, RC. (1997) The Inﬂuence of
Impact Inteface on Human Knee Injury: Implications for Instrument Panel Design
and the Lowe Extremity Injury Criterion. Society of Automotive Engineers 973327,
pp. 167-180.

3. Atkinson, P.J., Newberry, W.N., Atkinson, T.S., and Haut, RC. (1998a) A Method to

Increase the Sensitve Range of Pressure Sensitive Film. Journal of Biomechanics
31(9), pp. 855-859.

97

10.

11.

12.

13.

14.

15.

Atkinson, P.J., Atkinson, T.S., Haut, RC., Eusebi, C., Maripudi, V., Hill, T., and
Sambatur, K. (1998b) Development of Injury Criteria for Human Surrogates to
Address Current Trends in Knee-to Instrument Panel Injuries. Society of Automotive
Engineers 983146, pp. 13-31.

Chapchal, G. (1978) Posttraumatic Osteoarthritis afte Injury of the Knee and Hip
Joint. Reconstructive Surgery T raumatology 16, pp. 87-94.

Cowin, S. (1999) Bone Poroelasticity. Journal of Biomechanics. 32(3), pp. 217-238.

Ewes, 8.]. and Haut, RC. (2000)Polysu1phated Glycosaminoglycan Treatments Can
Mitigate Decreases in Stiffness of Articular Cartilage in a Traumatized Animal Joint.
Journal of Orthopaedic Research, 18:756-761.

Ewes, B., Dvoracek-Driksna, D., Orth, M., Altieo, N., and Haut, R. (2000) Matrix
Damage and Chondrocyte Death in Articular Cartilage Depends Upon Loading Rate.
46th Annual Meeting of Orthopaedic Research Society.

Felbeck, DK, and Atkins, A.G. (1984) Strength and Fracture of Engineeing Solids.
Prentice-Hall, pp. 327-330.

Garcia, JJ. (1998) A Transversely Isotropic Hypo-Elastic Biphasic Model of
Articular Cartilage Unde Impact Loading. PhD Dissertation. Michigan State
Univeisity.

Haut, RC. and Atkinson, R]. (1995) Insult to the Human Cadave Patellofemoral
Joint: Effects of Age on Fracture Toleance and Occult Injury. Society of Automotive
Engineers, 952729, pp. 281-294.

Haut, R.C., Ide, T.M., and DeCamp, CE. (1995) Mechanical Responses of the Rabbit
Patello-Femoral Joint to Blunt Impact. Journal of Biomechanical Engineering 117(4),
pp. 402-408.

Hayashi, S., Choi, H., Levine, R, Yang, K, and King, A. (1996) Experimental and
Analytical Study of Knee Fracture Mechanisms in a Frontal Knee Impact. 40th Stapp
Car Crash Conference, pp. 161-171.

Johnson, D.L., Urban, W.P., Caborn, D.N.M., Vanarthos, W.J., and Carlson, CS.
(1998) Articular Cartilage Changes Seen With Magnetic Resonance Imaging-
Detected Bone Bruises Associated With Acute Anterior Cruciate Ligament Rupture.
American Journal of Sports Medicine 26(3), pp. 409-414.

King, II, Fan, W.RS., and Vargovick, RI. (1973) Femur Load Injury Criteia--A
Realistic Approach 1 7th Stapp Car Crash Conference, pp. 509-525.

98

l6.MacKenzie, E.J., Siegel, J.H., Shapiro, S., et.al. (1988) Functional Recovery and
Medical Costs of Trauma: An Analysis by Type and Severity of Injury. Journal of
Trauma 28, pp. 281-298.

l7.McE1haney, J. (1966) Dynamic Response of Bone and Muscle Tissue. Journal of
Applied Physiolog 21, pp. 1231-1236.

18. Melvin, J., Stalnake, R, Alem, N., Benson, J., and Mohan, D. (1975) Impact
Response and Toleance of the Lowe Extremities. 19Ill Stapp Car Crash Conference,
pp. 543-559.

19. Mille, T., Martin, P., and Crandell, JR (1995) Cost of Lowe Limb Injuries in
Highway Crashes. Proceedings of the International Conference on Pelvic and Lower
Extremity Injuries, pp. 47-57.

20. Newbery, W.N., MacKenzie, C., and Haut, RC. (1998) Blunt Impact Causes
Changes in Bone and Cartilage in a Regularly Execised Animal Model. Journal of
Orthopaedic Research, 16(3), pp. 348-354.

21. Ochoa, J.A., Sandes, A.P., Kiesle, T.W., Heck, D.A., Toombs, J.P., Brandt, KD.,
and Hillberry, RM. (1997) In Vivo Obsevations of Hydraulic Stiffening in the
Canine Femoral Head. Journal of Biomechanical Engineering, 119(1), pp. 103-108.

22. Oyen-Tiesma, M, Atkinson, J., and Haut, RC. (1998) A Method for Promoting
Regular Rabbit Execise in Orthopaedics Research. Contemporary Topics in
Laboratory Animal Science, 37, pp. 77-80.

23. Patrick, L., Kroell, C., and Mertz, H. (1965) Forces on the Human Body in Simulated
Crashes. 9" Stapp Car Crash Conference, pp. 237-259.

24. Powell, W., Ojala, S., Advani, S., and Martin, R. (1975) Cadave Femur Responses to
Longitudinal Impacts. 19'” Stapp Car Crash Conference, pp. 561-579.

25. Pritsch, M., Comba, D., Frank, G., andHoroszowski, H. (1984) Articular Cartilage
Fractures of the Knee. Journal of Sports Medicine, 24(4), pp. 299-302.

26.Radin, E. L., Burr, D. B., Fyhrie, D., Brown, T. D., and Boyd, R. D. (1990)
Characteristics of Joint Loading as It Applies to Osteoarthrosis. In: Biomechanics of
Diarthrodial Joints. Springe-Velag, pp.437-451.

27. States, J. (1986) Adult Occupant Injmies of the Lowe Limb. In: Biomechanics
andMedical Aspects of Lowe Limb Injuries. Warrendale, pp.97-107.

28. Viano, DC. and Khalil, T.B. (1976) Investigation of Impact Response and Fracture of
a Human Femur by Finite Element Modeling. 20’” Stapp Car Crash Conference, pp.
412.

29. Viano, DC. (1977) Consideations for a Femur Injury Criterion. 21" Stapp Car Crash
Conference, pp. 445-473.

30. Viano, DC. and Stalnake, RL. (1980) Mechanisms of Femoral Fracture. Journal of
Biomechanics, 13(8), pp. 701-715.

31. Wright, T. and Hayes, W. (1976) Tensile Testing of Bone Ove a Wide Range of
Strain Rates; Effects of Strain Rate, Microstructure and Density. Medical and
Biological Engineering, pp. 671-680.

32. Wright, V. (1990) Post-Traumatic Osteoarthritis — A Medico-Legal Mineﬁeld. British
Journal of Rheumatology, 29, pp.474-47 8.

33. long, Y., Cowin, S.C., and Weinbaum, S. (1994) A Fibe Matrix Model for Fluid
Flow and Streaming Potentials in the Canaliculi of an Osteon. Annals of Biomedical
Engineering, 22(3), pp. 280-292.

34. Zhang, D., Weinbaum, S., and Cowin, SC. (1998) Estimates of the Peak Pressures in
Bone Pore Wate. Journal of Biomechanical Engineering, 120(6), pp. 697-703.

100

TABLES

8 Table 1: Impact Data for High and Low Rate of Loading on Paired Subjects

 

Time To Peak Contact Peak Force (kN) Input Enegy
Force (ms) Duration (N m)
(1118)

Rate of High“ Low High“ Low High Low High Low
Loading

Specimen
I.D.

 

99.370 4.63 50.0 10.76 226 3.49 3.42 21.71 4.87
99535 5.53 52.0 10.96 224 5.35 5.80 # 13.91
99-571 533' 55.0 10.46 228 5.11 5.05 44.95 20.90
99-729 5.83 57.0 15.54 228 3.23 2.58 28.04 9.76
99-750 4.48 56.0 9.46 228 4.65 3.92 28.44 17.23
99-775 3.34 55.0 8.12 229 5.25 5.36 22.03 21.73
mm 4.43 49.0 10.4 165 3.66 3.90 24.48 7.00
00-540 2.64 66.0 6.52 222 6.11 5.97 13.82 30.67

 

Average 4.9 54.2 10.3 218. 4.6 4.5 26.2 15.8
8 .

 

 

 

Std Dev. 0.9 2.6 2.6 23.4 1.0 1.2 9.6 8.6

 

 

 

 

* — signiﬁcantly diffeent ﬁ'om impacted limb in low rate of loading group (t-test, p <
0.05)
# - Outlie

101

Table 2: Contact Area and Pressure Data for High and Low Rate of Loading on Paired

 

 

 

 

 

 

 

Subjects
m
Contact Area (mmz) Pressure (MPa) Area (mmz)
Loading
Specimen
I.D.
99-370 734.3 N/A 9.8 N/A’ 869.5 N/A
99-535 478.8 443.9 19.3 17.4 1051.2 523.1
99-571 745.1 895.8 12.1 13.1 515.2 816.9
99-729 # 377.3 15.2 12.0 749.1 531.8
99-750 478.8 623.9 12.6 10.5 563.3 462.9
99-775 967.8 605.3 17.6 17.6 713.4 569.5
00-524 637.1 317.9 11.2 14.0 834.1 420.4
00-540 # 846.7 11.8 15.9 625.5 783.4
Aveage 749.6 587.3 13.7 14.4 740.1 586.9
Std Dev. 185.8 224.1 3.3 2.7 176.5 153.9
# - Outlie

N/A — Pressure ﬁlm data lost

102

 

Table 3: Frequency of Injury Following High and Low Rate of Loading Expeiments on

 

 

Paired Subjects
Gross Fractures Histological Occult Microcracks
Fractures
Rate of High Low High Low High Low
Loading
Specimen
I.D.
99-370 1 0 2 0 2 0
99-535 0 O 1 1 0 0
99-571 0 0 0 0 2 1
99-729 1 0 4 2 4 4
99-750 0 0 0 0 2 1
99-775 1 l 2 l 3 3
Total 3 1 9 4 13 9
Range (0-4) (0-2) (0-4) (0-4)
A

 

 

 

 

 

 

103

Table 4: Mechanical Properties of Retropatellar Articular Cartilage Post Trauma (Avg 1'

 

 

 

 

 

 

 

 

 

 

 

 

 

81)).
Time 1.11111) E11 E33 G13 013 In k3
Point MPa MPa MPa m‘/Ns10'” m4/Ns10'l5
4.5
Months
Low' Imp 3.641 0.811 0.191 0.121 7.6612.89* 2.421124
1.49* 0.17* 0.05 0.02
Unimp 5.341 1.031 0.171 0.121 51912.84 25411.44
2.19 0.18 0.02 0.05
I-Iigh‘ Imp 3.821 0.821 0.221 0.101 7.5913.71* 5.921375”
1.30* 023* 0.05 0.03
Unimp 6.301 1.151 0.161 0.101 48313.14 2.70:1:1.86
1.83 0.30 0.08 0.02
12
Months
Low" Imp 4.151 0.981 0.211 0.111 6.1512.86 4.3213.48
221* 044* 0.04 0.03 -
Unimp 5.561 1.261 0.211 0.131 4.39:1:1.82 25811.59
2.18 0.36 0.07 0.04
High‘ Imp 2.931 0.781 0.251 0.101 6.7012.62* 4.2011.92*
0.94* 023* 0.14 0.03
Unimp 5.961 1.141 0.211 0.111 4.4312.49 2.5511.93
2.42 0.25 0.14 0.03

Imp 4 impacted limb, Unimp - unimpacted limb

+ - signiﬁcantly diffeent ﬁ'om contralateal unimpacted limb (paired t-test, p < 0.05)

# — signiﬁcantly diffeent ﬁ'om impacted limb in low rate of loading group (t-test, p <
0.0)

104

Table 5: Subchondral Bone Thicbess (mm) at the Cente of the Patella and the Midline
of the Lateal and Medial Facets (Avg :1: SD).

 

 

 

 

Time Point Limb Lateal Cente Medial
Low . Imp 0.44 1 0.08* 0.67 1 007* 0.45 1 0.11
Unimp 0.40 1 0.08 0.56 1 0.14 0.41 1 0.08
High: Imp 0.38 1 0.09* 0.74 1 016* 0.39 1 0.10*
Unimp 0.30 1 0.05 0.53 1 0.17 0.32 1 0.07
12 Months
Low" Imp 0.63 1 0.15* 0.97 1 024* 0.64 1 0.14
Unimp 0.58 1 0.11 0.78 1 0.18 0.58 1 0.21
High‘ Imp 0.85 1 020* “ 1.33 1 048* 0.85 1 023* ’*
Unimp 0.66 1 0.21 0.83 1 0.36 0.59 1 0.13

 

 

 

 

 

 

 

Imp — impacted limb
Unimp — unimpacted limb
+ — signiﬁcantly diffeent from contralateal unimpacted limb (paired t-test, p < 0.05)

# — signiﬁcantly diffeent from impacted limb in low rate of loading group (t-test, p <
0.0)

105

Table 6: Total Fissure Length (Avg i SD) ﬁom Gross Photographs of the Rabbit

Patellae, and the Average Fissure Depth (Avg i SD) and the Numbe of F issures (median
(range)) from Histological Sections.

 

 

 

m
Length (mm) Depth (%) Fissures (#)
111111011111:
Low ‘ Imp 9.2 1 5.2 + 37 114 * 2.5 (0-6) *
Unimp ‘ 2.4 1 2.8 5 1 12 , 0 (0-2)
High‘ Imp 11.2 1 5.8 * 36 1 24 * 4 (2-5) *
Unimp 3.11 3.6 7 118 0 (0-1)
12 Months
Low" Imp 11218.3 34118* 1(0-10)
Unimp 6.3 1 5.5 9 1 12 1 (0-5)
High‘ Imp 25.7188 *” 42115 * 7(4-12) *’
Unimp 4.4 1 4.3 23 1 20 2 (0-8)

 

 

 

 

 

 

 

Imp - impacted limb

Unimp — unimpacted limb

+ — signiﬁcantly different from contralateal unimpacted limb (paired t-test, p < 0.05)

* — signiﬁcantly diffeent from contralateal unimpacted limb (Signed Rank test, p <
0.05)

# — signiﬁcantly diffeent from impacted limb in slow group (t-test, p < 0.05)

$ - signiﬁcantly diffeent from impacted limb in slow group (Rank Sum test, p < 0.05)

106

FIGURES
. Rigid
P0111118 Femur Patella Irnpactor

Sleeve /
111
ll

 

 

 

 

 

Pressure Film
Tibia Locations

Figure 1: A 5.7 x 5.7 cm, 606l-T‘6 aluminum rigid interface with a mass of 4.5 kg was
accelerated in a horizontal impact sled by a pneumatic cannon. The rigid interface
impacted the joint. The pressure sensitive ﬁlm was inserted into the joint to capture
patellofemoral joint contact pressure and area, as well as the contact pressure and area on
the anterior surface of the patella. High, medium, and low pressure ﬁlm were stacked in
the patellofemoral joint. The anterior surface of the patella had stacked low and medium
pressure ﬁlm.

 

1

8

 

88688388

Contact Pressure (MPa)

 

Contact Width (mm)

Figure 2: This chart indicates the useable pressure ranges for the low, medium, and high
pressure ﬁlm. The low pressure ﬁlm had a useable range of 3MPa to lOMPa. The
useable range for the medium pressure ﬁlm was between lOMPa and 49MPa. The high
pressure ﬁlm was not used since pressures ﬁom past experiments never exceeded 49MPa
(reproduced from reference 22).

107

10 MPa < Press. < 49 MPa

 

3 MPa<Press. < 10 MPa

1:

 

Figure 3: The digitized images of the medium and low pressure ﬁlms were combined to
obtain the contact pressure distribution and contact area.

 

Figure 4: The number and average depth of the surface ﬁssures was measured at 40x
with a calibrated eye-piece by one investigator using the histological sections of the
patellae.

108

 

..§
”Fwd

 

 

 

 

 

0 ' i i : .
i 0 50 100 150 200
I rim (In) [

 

 

Figure 5: This chart shows typical load versus time responses for the high and low rate
of loading experiments. Both tests showed a nearly symmetrical haversine. The peak
loads were not signiﬁcantly different between the tests. The contact durations for each of
these tests were signiﬁcantly different.

 

 

 

 

 

 

 

Figure 6: Typical load-displacement curves for the high and low rate of loading
experiments.

109

 

Subchondral
__ Bone

 

Figure 7: This shows a gross ﬁ'acture of the patella that was readily observable on the
retropatellar surface.

 

Figure 8: This histological slide of the patella shows a horizontal microcrack near the
zone of calciﬁed cartilage between hyaline cartilage from the underlying subchondral
bone.

110

 

(B)

Figure 9: The traumatized rabbit patellae in the high rate of loading experiments (A)
resulted in a signiﬁcant increase in subchondral bone thickness compared to the impacted
patellae in the low rate of loading experiments (B) under the midline of the medial and
lateral facets, with the same trend apparent in the center of the patellae.

111

 

Figure 10: Representative histological sections of the rabbit retropatellar cartilage for
unimpacted (A), impacted at a high rate of loading (B), and impacted at a low rate of
loading (C). Notice the average ﬁssure depth to be between 30 and 40% of the cartilage
thickness, and deeper ﬁssures tended to exist following high versus low rate of loading
experiments at 12 months post trauma.

112

CHAPTER 6

POLYSULPHATED GLYCOSAMINOGLYCAN TREATMENTS CAN
MITIGATE DECREASES IN STIFFNESS OF ARTICULAR CARTILAGE IN A
TRAUMATIZED ANIMAL JOINT

Ewers BJ and Haut RC

ABSTRACT

A single, blunt impact to the rabbit patello-femoral joint has been shown to cause
a decrease in the stiffness of retropatellar cartilage and an increase in the thickness of
underlying bone. Polysulphated glycosaminoglycan (PSGAG) treatments, on the other
hand, have been shown to inhibit the degradation of articular cartilage, and possibly
increase synthesis of collagen and glycosaminoglycans in experimental studies on
diseased joints. The aim of the current study was to examine the effect of early PSGAG
treatments on cartilage using an in vivo post-trauma animal model. The study used 24
Flemish Giant rabbits in three groups: control, impacted, and impacted with treatment.
Treatment consisted of intramuscular injections the day of insult and every four days
thereafter for six weeks. At 30 weeks post-trauma mechanical tests were performed on
the retropatellar cartilage to determine its mechanical stiffness. The patellae were also
grossly evaluated for surface lesions on the retropatellar cartilage, and histologically
processed to measure the thickness of subchondral bone. The group receiving no
treatment had a statistically signiﬁcant decrease in stiffness (modulus) for the impacted

versus contralateral, unimpacted cartilage and versus the control group. The degradation

113

in mechanical stiffness, however, was not observed in the group receiving treatment.
Thee was also a signiﬁcant increase in the underlying thickness of the subchondral plate
on the impacted patellae compared to the contralateral, unimpacted sides for both the
treated and untreated groups. In conclusion, the PSGAG treatments mitigated a decrease
in mechanical stiffness (modulus) of retropatellar articular cartilage at 30 weeks post-
trauma. The mechanism by which the mechanical stiffness of the cartilage was preserved
is currently unknown.
INTRODUCTION

A single, severe blunt insult to the rabbit patello-femoral joint has been shown to
cause changes in retropatellar cartilage and underlying subchondral bone (Newberry et
al., 1998). This includes a decrease in stiffness of the retropatellar articular cartilage,
starting at three months post-trauma versus unimpacted controls. Studies with human
cartilage have also shown that degenerated tissue has a decreased stiffness via indentation
testing (Kempson et al., 1971) and a decrease in equilibrium modulus has been correlated
with this degeneration (Armstrong and Mow, 1982). The thickness of the subchondral
plate underlying retropatellar cartilage is greater than that of the contralateral, unimpacted
patella. Such early changes have been hypothesized leading to further softening of the
articular cartilage, erosion, and eventually osteoarthrosis (0A) in the joint (Radin et al.,
1984). Since there can often be a single known incident, or trauma, which could initiate
this disease process a contraceptive treatment, such as a “chondroprotective” drug, may
prove to be effective in slowing or mitigating changes in joint tissues post-trauma.

The etiology of 0A is bown to involve an increase in extracellular matrix

turnover, which eventually leads to a net loss of proteoglycans and collagen from the

114

articular cartilage. These changes result in degradation of cartilage and ultimately
degeneration of the joint (Bayliss, 1992). Polysulfated glycosaminoglycans (PSGAG)
have been used clinically in Europe and in the veterinary community for the treatment of
diseased joints. PSGAG have been shown to decrease the rate of synthesis of certain
degradative enzymes in these degenerating joints (Howell et al., 1986; May at al., 1988).
In addition to inhibiting degradation, in vitro experiments have shown that PSGAG can
stimulate a net increase in collagen and glycosaminoglycan synthesis in tissue explants
(Glade, 1990). While animal models of OA have been used to study PSGAG treatments
(Fubinieta1., 1993; Gaustad and Larsen, 1995; Golding etal., 1983: Howell et al., 1986),
the effectiveness of PSGAG treatment has not been examined in a post-trauma model.
Mazieres et a1. (1993) has shown a positive effect of the anti-inﬂammatory drug
diacerhein on the histological appearance of traumatized patellae with a post-contusion
model of CA using New Zealand rabbits. Our model uses Giant Flemish rabbits along
with a post-trauma exercise protocol and documents the mechanical stifﬁless of
traumatized cartilage over time. The mechanical properties of articular cartilage have
been suggested to be a better indicator of its ability to function as a weight bearing
material than its histological appearance (Armstrong and Mow, 1982).

The aim of the current study was to evaluate whether early, intramuscular
injections of PSGAG would be effective in mitigating the previously recorded softening
of retropatellar articular cartilage following a severe, blunt trauma to the patello-femoral
joint in a rabbit model. We hypothesized that these early treatments would help preserve
the mechanical stiffness (modulus) of the retropatellar articular cartilage, even in the

presence of impact generated surface lesions.

115

METHODS AND MATERIALS

A total of twenty-four mature Flemish Giant rabbits (5.0 i’ 0.5 kg, 6-8 nronths of
age) were used in the study. Sixteen animals received a blunt insult to the right patello-
femoral joint. Eight animals served as a control population, and did not receive an insult.
The impact protocol has been described in earlier studies (Newberry et al., 1998). Brieﬂy,
a 1.33 kg mass with a ﬂat 1 in. diameter impact surface was dropped from 0.46 m onto
the right hind limb, which was ﬂexed approximately 120°. During the impact the animals
were maintained at a surgical plane of anesthesia using Isoﬂurane and oxygen (2%
Isoﬂurane). The impactor was stopped electronically to prevent multiple impacts. A load
transducer (Sensotec, Columbus OH : model 31/1432. 500 lb capacity) recorded the
impact load at 10 kHz.

After trauma, all animals received one injection of Butorphenol for post-surgical
pain. Eight animals were treated with the veterinary drug Adequan'!b 100 mg/ml (Luitpold
Pharmaceuticals), which is a semi-synthetic glycosaminoglycan containing primarily
chondoitin sulphate with three to four esters per disaccharide unit and a molecular weight
of 3000 to 15000 daltons. The treatment consisted of 2 mg/kg injections every four days
for six weeks beginning on the day of the insult. The remaining eight impacted animals
received no treatment.

h After a ﬁve day period of rest following the impact, a daily exercise program was
initiated for all animals, including the non-impacted control group. Previous studies have
shown that with execise the decrease in stiffness of cartilage and increase in thickness of
subchondral bone is accelerated in time (Newberry et al., 1997; Newberry et al., 1998).

This program consisted of ten minutes of exercise, ﬁve days a week, on a treadmill

116

running at 0.3 mph (Oyen-Tiesma et al, 1998). The exercise schedule lasted the entire
length of the study. The animals were housed in individual cages (48” x 24” x 19”) when
not being exercised. This study was approved by the MSU All-University Committee on
Animal Use and Care.

All twenty-four animals were sacriﬁced after thirty weeks. Immediately after
sacriﬁce, the patellae were excised for mechanical indentation tests on the retropatellar
cartilage. The patellae were immersed in a room-temperature phosphate buffered saline
bath during the indentation tests. These tests were performed with a custom built
instrument that used a conrputer operated stepper motor (Physik Instruments, Waldbronn,
Germany : model M-l68.30) to indent the cartilage. A ﬁxture, which held the patella with
its retropatellar surface facing the indenter, allowed X-Y-Z movement to position the site
of indentation. A camera mount (Bogen, Italy) attached to the base of the ﬁxture allowed
for rotation of the patellae, to insure that the indentation was normal to the cartilage
surface. A 1 mm diameter ﬂat, non-porous probe was pressed into the cartilage 0.1 mm in
30 ms and maintained for 150 seconds. The resistive loads were measured (Data
Instruments, Acton MA : model JP - 25, 25 lb capacity), ampliﬁed, and collected at 1000
Hz for the ﬁrst second and 20 Hz thereafter. After the tissue was allowed to recover for 5
minutes, the probe was replaced with a needle that was slowly pressed into the cartilage
to determine the thickness of cartilage at that location. The above procedure was
performed at two sites approximately one indentor diameter away from the surface
lesions on the lateral retropatellar facet (Figure l).

The mechanical data from the indentation tests on the cartilage were analyzed

using an elastic analysis due to Hayes et al. (1972). This analysis assumed an inﬁnite

117

elastic laye (cartilage) bonded to a rigid half spaCe (bone). It allowed the determination
of two shear moduli for the cartilage. These were based on the load at 30 milliseconds (an
instantaneous shear modulus — GU) and at 150 seconds (a relaxed shear modulus — GR).

After the mechanical tests, each patella was stained with India ink to highlight
surface lesions and photographed. The patellae were then placed in 10% buffered
fornmlin for a week, and decalciﬁed in 20% formic acid for another week. Tissue blocks
were cut medial to lateral across the patellae in the area known to produce high contact
pressure during impact to the joint (Newberry et al., 1998). The tissue blocks were
processed in parafﬁn, and a minimum of six sections, 8 11 thick, were cut medial to
lateral. The sections were stained with Safranin O-Fast Green and examined under light
microscopy at 12-400 power. The thickness of the subchondral bone plate underlying
retropatellar cartilage was measured at 25x with a calibrated eye-piece at the center of
each facet and mid-line of the patella by a single investigator (BE), using existing
methods (20) (Figure 2).

A mixed design two-factor ANOVA with S-N-K post-hoe testing was used to
evaluate the differences in shear modulus (stiffness) and subchondral bone thickness
between limbs and groups. Limb was the repeated factor with group being the
independent factor. Signiﬁcant statistical differences were reported for p < 0.05.
RESULTS

Examination alter impact and ensuing daily observations by a veterinary
technician (J.A.) indicated no noticeable joint effusions, and the rabbits did not appear to
favor their impacted limbs. Gross visual examinations of the patellae after wiping with

India ink indicated surface ﬁssures on the retropatellar articular cartilage in 15 of 16

118

impact cases. These lesions typically ran proximal to distal on the lateral facet, near the
central ridge of the patella (Figure l). PSGAG treatments did not have an effect on the
gross appearance of these surface lesions. The impacted patellae in the treated group still
had typical impact-induced surface lesions (Figure 3 ). No surface lesions were seen on
the retropatellar surfaces of the unimpacted, contralateral or control patellae.

The mechanical data was lost for one animal in the impacted group. There was a
statistically signiﬁcant decrease in GU of the retropatellar cartilage on the impacted
patella versus the controls at this 30 week timepoint (Table 1). There were no signiﬁcant
differences in subchondral bone thickness between the impacted limb from the non-
treated or PSGAG-treated groups compared to controls at any location on the patella. In
contrast, G; of the impacted cartilage from the PSGAG-treated group was not different
from that of controls or the unimpacted, contralateral limb of the impacted treated or non-
treated groups. The relaxed modulus GR from the impacted (non-treated) group was not
different from controls. On the other hand, this measure of tissue stiffness was
signiﬁcantly less than that of its contralateral, unimpacted limb. This signiﬁcant
difference in GR between limbs was not present in the PSGAG-treated group.

From the histological sections of the patellae no signiﬁcant differences were
measured in the subchondral plate thicknesses between left and right limbs at any
location on the patella in the control group (Table 2). In contrast, there was a signiﬁcant
increase in thickness of the subchondral plate underlying retropatellar cartilage at the
medial and central locations in the impacted versus the contralateral, unimpacted patellae.
Animals receiving PSGAG treatments after impact also indicated signiﬁcant increases in

the thickness of the subchondral plate on impacted versus contralateral, unimpacted

119

patellae at these same two locations. There was also a trend for subchondral plate
thickening in impacted versus unimpacted patellae at the lateral location.
DISCUSSION

The results supported our hypothesis that the stiffness (modulus) of retropatellar
cartilage would not degrade signiﬁcantly, post-trauma for animals injected early with
PSGAG. After six weeks of treatment immediately following insult the 39 % decrease in
stiffness for impacted versus unimpacted, control retropatellar cartilage was mitigated in
the model. On the other hand, early PSGAG treatment did not appear to have an effect on
the gross appearance of impact-induced surface lesions on the retropatellar cartilage. The
PSGAG treatments also did not mitigate the thickening of the subchondral plate
underlying retropatellar cartilage that had previously been measured versus the
contralateral unimpacted patella in this post-trauma animal model.

Our results compare well with earlier studies using this model. Newberry et al.
(1998), observed a statistically signiﬁcant (40 %) reduction in the instantaneous modulus
(GU) of impacted retropatellar cartilage versus unimpacted, controls at six months post-
trauma. Similar to the previous study (Newberry et al., 1998), we also documented a
signiﬁcant decrease in the relaxed modulus (GR) of the impacted cartilage compared to
the contralateral unimpacted limb, but not compared to controls. The appearance of the
impact-induced surface lesions on the retropatellar cartilage was also similar to earlier
studies (Newberry et al., 1998). Finally, we observed a 23 % increase in the thickness of
the subchondral plate at the central location for the impacted limbs compared to
contralateral, unimpacted limbs. Newberry et a1. (1998) documents a 29 % increase at six

months post-trauma.

120

The PSGAG treatment did not grossly appear to heal the mechanically induced
surface lesions in our trauma model. This result was consistent with others who have
shown no effect of PSGAG treatment on healing cartilage defects (Barr et al., 1994;
Trotter et al., 1989). In our trauma model statistical changes in thickness of underlying
bone versus controls is not documented until 12 months after trauma. A longer duration
study would be needed to document the effect of PSGAG treatments on the underlying
bone in a more chronic setting. Armstrong et al.(1994) observed that subchondral bone
thickening was mitigated or slowed after regular treatment of the joint with hyaluronan
(HA) in a sheep meniscectomy model. This effect was believed to be an indirect result of
the treatments by an improvement in the quality of the overlying cartilage. In our model,
Ewers et al.(1998), recently showed that slight changes in the overlying cartilage
properties theoretically would have no signiﬁcant affect on the underlying bone stresses
which might cause subchondral bone remodeling. In contrast, the effect might be
different for the sheep tibial-femoral joint because of a different geometry. On the other
hand, the remodeling of bone underlying impacted joint cartilage in our model may be
due to the acute overstressing generated during blunt impact. PSGAG treatments may not
signiﬁcantly affect this remodeling post-trauma. These issues need further study.

Irrrpaction of the patello-femoral joint in the current study may have overstressed
retropatellar cartilage and damaged the extracellular matrix leading to an increased
hydration of the tissue (Donahue et al., 1983). The damaged matrix and the increased
hydration of the cartilage may be the cause for the decrease in the modulus of the tissue
post-trauma. Previous research, by others, suggests that the equilibrium response of

cartilage in an indentation test (approximated by GR in the current study) is controlled by

121

the content and integrity of tissue proteoglycans (Jurvelin et al., 1988). Furthermore, they
suggested that GU may be more reﬂective of the content and integrity of collagen in the
tissue (Mizrahi ct al., 1986). Yet, these investigators also suggest that tissue
proteoglycans can play a role in determining the structural integrity of the collagen
network by affecting its pretension. The unrelaxed and relaxed shear moduli are
“apparent” material properties that are derived from modeling the cartilage as a single-
phase, elastic layer. The unrelaxed modulus is related to tissue permeability, fluid
pressure, and matrix properties. The impact may have caused the tissue to become more
permeable causing the decrease in GU. A possible mechanism for the action of the
PSGAG treatments on traumatized cartilage might involve its ability to alter the content
of proteoglycans post-insult by either increased synthesis or inhibited degradation. For
example, Glade et al.(l990) showed increased synthesis of proteoglycan by chondrocytes
in culture when exposed to PSGAGs. Harman et al.(1987), also showed less depletion of
tissue proteoglycans with PSGAG treatments versus no treatments in a canine
meniscectomy model. This increase in proteoglycans could cause the permeability to be
lowered, resulting in a restoration of GU.

Another potential effect of the PSGAG treatments may have been to help return
normal loading to the impacted limb, mitigating a small increase in the stiffness of
cartilage on the unimpacted contralateral limb that has been documented to become
statistically signiﬁcant at one year post-trauma in our model (Newberry et al., 1998). On
the other hand, our studies have not been able to actually quantify a favoring of the
impacted limb in our model. The possibility of complementary effects of PSGAG

treatment and altered joint loading, or with or without post-trauma exercise will also need

122

to be investigated in future studies. Another limitation of the current study was the

absence of histochcmical and histological analyses of the treated and non-treated

retropatellar cartilage.

In conclusion, this investigation suggested positive effects of early PSGAG
treatments on mechanical stiffness (modulus) of retropatellar cartilage following severe
blunt trauma to the rabbit patello-femoral joint. The applicability of these results to the
human scenario will need to be explored in clinical studies.

ACKNOWLEDGMENTS

This study was supported by a grant for the Centers for disease Control and
Prevention (R49/CCR503607). Its contents are solely the responsibility of the authors and
do not necessarily represent the ofﬁcial views of the Centers for Disease Control and
Prevention. The authors wish to gratefully acknowledge Jane Walsh for her preparation
of the histological sections, Jean Atkinson for her care of the animals and Cliff Beckett
for technical assistance.

REFERENCES

1. Armstrong CG, Mow VC: Variations in the intrinsic mechanical properties of human

articular cartilage with age, degeneration, and water content. J Bone Joint Surg [Am]
64:88-94, 1982.

2. Armstrong S, Read R, Ghosh P: The effects of intraarticular hyaluronan on cartilage
and subchondral bone changes in an ovine model of early osteoarthritis. J Rheumatol
21:680-688, 1994.

2. Barr AR, Duance VC, Wotton SF, Waterman AE: Inﬂuence of intra-articular sodium

hyaluronate and polysulphated glycosaminoglycans on the biochemical composition
of equine articular surface repair tissue. Equine VetJ 26:40-42, 1994.
4. Bayliss MT: Metabolism of animal and human osteoarthritic cartilage. In: Articular

Cartilage and Osteoarthritis, lst ed, pp 487-500. Ed by KE Kuettner, R Schleyerbach,
JG Peyron, VC Hascall. New York, Raven Press, 1992.

123

. Donahue JM, Buss D, Oegema TR Jr, Thompson RC Jr: The eﬁects of indirect blunt

trauma on adult canine articular cartilage. J Bone Joint Swg [Am] 65:948-957. 1983.

. Ewers BJ, Newbeny WN, Garcia JJ, Haut RC: Alterations in the mechanical

properties of bone underlying articular cartilage in a traumatized joint. 44th ann.
meeting ORS 459, 1998.

. Fubini SL, Boatwright CE, Todhunter RJ, Lust G: Effect of intramuscularly

administered polysulphated glycosaminoglycan on articular cartilage from equine
joints injected with methylprednisolone acetate. Am J Vet Res 54: 1359-1365, 1993.

. Gaustad G, Larsen S: Comparision of polysulpated glycosaminoglycan and sodium

hyaluronate with placebo in treatment of traumatic arthritis in horses. Equine Vet J
27:356-362, 1995.

9. Glade MJ: Polysulphated glycosaminoglycan accelerates net synthesis of collagen and

10.

ll.

12.

13.

14.

15.

16.

glycosaminoglycans by arthritic equine cartilage tissues and chondrocytes. Am J Vet
Res 51:779-785. 1990.

Golding J, Ghosh P: Drugs for osteoarthrosis II. The effect of glycosaminoglycan
polysulfate ester (Arteparon) on proteoglycan aggregation and loss from articular
cartilage of immobilized rabbit knee joints. Curr Ther Res 34:67-80, 1983.

Hannan N, Ghosh P, Bellenger C, Taylor T: Articular cartilage damage produced by
meniscectomy in the canine. J Orthop Res 5:47-59, 1987.

Hayes WC, Keer LM, Herrmann G, Mockros LF: A mathematical analysis for
indentation tests of articular cartilage. J Biomech 5:541-551, 1972.

Howell DS, Carreno MR, Pelletier JP, Muniz OE: Articular cartilage breakdown in a
lapine model of osteoarthritis: action of glycoaminoglycan polysulfate ester (GAGPS)
on proteoglycan enzyme activity, hexuronate, and cell counts. Clin Orthop 213:69-76,
1986.

Jurvelin J, Séiimanen A-M, Arokoski J, Helminen HJ, Kiviranta I, Tammi M:
Biomechanical properties of the canine knee articular cartilage as related to matrix
proteoglycans and collagen. Eng Med 17: 157-162, 1988.

Kempson GE, Spivey CJ, Swanson SAV, Freeman MAR: Patterns of cartilage
stiffness on nomral and degenerate human femoral heads. J Biomech 4:597-609,
1971.

May SA, Hooke RE, Lees P: The effect of various drugs used in the treatment of

equine degenerative joint disease on equine stromelysin (proteoglycanase). Br J
Pharmacol 93:28], 1988.

124

17. Mazieres L, Berdah M, Thiechart, Viguier G: Etude de la diacetylrheine sur un
modéle post-contusif d’arthrose expérimentale chez le lapin. Rev Rhum [Ed Fr]
60:77S-818, 1993.

18. Mizrahi J, Maroudas A, Lanir Y, Ziv I Webber TJ: The “instantaneous” deformation
of cartilage: effects of collagen ﬁber orientation and osmotic stress. Biorheologv
23:311-330, 1986.

19. Newben'y WN, Zukosky DK, Haut RC: Subfracture insult to a knee joint causes
alterations in the bone and in the functional stiffness of overlying cartilage. J Orthop
Res 15:450-455, 1997.

20. Newberry WN, MacKenzie CD, Haut RC: Blunt impact causes changes in bone and
in cartilage in a regularly exercised animal model. J Orthop Res 16:348-354, 1998.

21. Oyen-Tiesma M, Atkinson J, Haut RC: A method for promoting regular rabbit
exercise in orthopaedics research. Contemporary Topics in Laboratory Animal
Science 37:77- 80, 1998.

22. Radin EL, Martin RB , Burr DB, Caterson B, Boyd RD, Goodwin C: Effects of
mechanical loading on the tissues of the rabbit knee. J Orthop Res 2:221-234, 1984.

23. Trotter GW, Yovich JV, McIlwraith CW, Norridin RW: Effects of intramuscular

polysulphated glycosaminglycan on chemical and physical defects in equine articular
cartilage. Can J Vet Res 43:224-230, 1989.

125

TABLES

Table l: Instantaneous (unrelaxed) shear modulus (GU) and relaxed shear modulus (GR)
of the articular cartilage for each group (Ave i SD).

 

 

 

Group 01; GR
(MPa) (MPa)
mm

Left 0.949 i 0.138 0.278 at 0.062
NO PSGAG Impacted 0570 i 0.156 * 3 ‘ 0.226 i 0.062 "
Unimpacted 1.081 i 0.276 0.323 i 0.092

PSGAG Impacted 0.940 is 0.180 0.297 1: 0.066
Unimpacted 0.910 1- 0.253 0.311 i 0.077

 

CONT - control group (n=8)
NO PSGAG - no treatment (n=7)
PSGAG - 6 weeks of treatment (n=8)

+ - Signiﬁcantly different from unimpacted side

5 - Signiﬁcantly different from controls
* - Signiﬁcantly different from impacted side on PSGAG group

Table 2: Subchondral plate thickness at three locations for each group (Ave i SD, n=8).

 

 

 

WWW
mm mm mm
mm
Left 0.41 i 0.08 0.62 i 0.20 0.46 i: 0.12
NO PSGAG Impacted 0.48 i 0.09 + 0.64 i 0.09 + 0.43 i 0.08
Unimpacted 0.38 i 0.06 0.52 i 0.12 0.38 i 0.06
18 PSGAG Impacted 0.54 i 0.14 + 0.84 i 0.19 + 0.50 i 0.10
Unimpacted 0.41 i 0.12 0.52 i 0.15 0.33 3: 0.06

 

 

CONT - control group
NO PSGAG - no treatment
PSGAG - 6 weeks of treatment

+ - Signiﬁcantly different from unimpacted side

126

FIGURES

Proximal

Medial

 

Drstal

Figure 1: Impact ﬁssures were found on the lateral side of the central ridge between the
patellar facets (ID). Indentation tests were conducted at two locations along the lateral
facet (o) (x7.5).

127

 

Figure 2: The subchondral bone thickness was measured at three locations: midline of
the patella, and at the midline of the medial and lateral facets. The thickness was
measured with a calibrated eyepiece between the tidemark and trabecular bone (->).

128

 

Figure 3: No differences were grossly observed in the histological appearance of lesions
in the retropatellar cartilage of the impacted (A) and impacted with PSGAG treatment (B)

groups.

129

CHAPTER 7

THE EXTENT OF MATRIX DAMAGE AND CHONDROCYTE DEATH IN
NIECHANICALLY TRAUMATIZED ARTICULAR CARTILAGE EXPLANTS
DEPENDS 0N RATE OF LOADING

Ewers BJ, *Dvoraeek-Driksna D, +Orth MW, and Haut RC

SUMMARY

Mechanical loads can lead to matrix damage and chondrocyte death in articular
cartilage. This damage has been implicated in the pathogenesis of secondary
osteoarthritis. Studies on cartilage explants with the attachment of underlying bone at
high rates of loading have documented cell death adjacent to surface lesions. On the other
hand, studies involving explants removed from bone at low rates of loading suggest no
clear spatial association between cell death and matrix damage. The current study
hypothesized that the observed differences in the distribution of cell death in these studies
are attributed to the rate of loading. Ninety bovine cartilage explants were cultured for
two days. Sixty explants were loaded in unconﬁned compression to 40 MPa in either a
fast rate of loading experiment '(~900 MPa/s) or a low rate of loading experiment (40
MPa/s). The remaining thirty explants served as a control population. All explants were
cultured for four days after loading. Matrix damage was assessed by measuring the total
length and average depth of surface lesions and the release of glycosaminoglycans to the
culture media. Explants were sectioned and stained with calcein and ethidium bromide

homodimer to document the number of live and dead cells. Greater matrix damage was

130

documented in explants subjected to a high rate of loading, compared to explants exposed
to a low rate of loading. The high rate of loading experiments resulted in cell death
adjacent to ﬁssures, whereas more dead cells were observed in the low rate of loading
experiments and a more diffuse distribution of dead cells was observed away from the
ﬁssures. In conchrsion, this study indicated that the rate of loading can signiﬁcantly affect
the degree of matrix damage, the distribution of dead cells, and the amount of cell death
in unconﬁned compression experiments on explants of articular cartilage.
INTRODUCTION

Excessive levels of mechanical load can generate matrix damage and chondrocyte
death in articular cartilage (Jeffrey et aL, 1995). Matrix damage has been documented
following severe blunt impacts to a joint using an animal model of osteoarthrosis (OA)
(Newberry et al., 1998), and clinically it is seen in the early stages of OA (3).
Chondrocyte death, 0 leading to a reduction in tissue cellularity or chondrocyte
malfunction, has been suggested to facilitate development of OA (Blanco et al, 1998;
Buckwalter, 1995). Death of chondrocytes has also been shown to result in long-term
degradation of cartilage in viva (Simon et al., 1976).

Studies using cartilage explants with the attachment of underlying bone have
documented matrix damage and cell death following blunt insult at high rates of loading.
Oyen-Tiesma et aL (1999), using a single insult of 53 MPa in 250 ms (212 MPa/s) onto
an explant of cartilage, documented that cell death correlated with the degree of
mechanical disruption to the matrix. The study observed that cell death occurred adjacent
to surface ﬁssures, and the investigators did not observe cell death in specimens without

visible damage to the matrix. Another study, using a single insult of load ranging from 30

131

to 50 MP3 with a time to peak ranging from 40 to 100 ms (350 — 1250 MPa/s), showed
cell death adjacent to ﬁssures in 12 of 15 specimens (Repo and Finlay, 1977).

Other studies, using cartilage explants without underlying bone and for low rates
of loading have also documented matrix damage and cell death. Torzilli et al. (1999),
using a single impact of 20 MPa in 571 ms (35 MPa/s), documented surface ﬁbrillation
into the middle zone of cartilage directly under the indentor. Cell death was also localized
under the indentor, but extended throughout the thickness of the explant. Quinn et a1.
(1998) saw no clear spatial association between cracks in the matrix and cell viability.
That study used a very low rate of loading that compressed explants to 50 % strain in 15
seconds, and the load was held for one hour.

The distribution of cell death with respect to visible matrix damage is inconsistent
between these two types of studies. Interestingly, these studies exhibit two important
differences that may help explain the contrasting results: high rate of loading with the
attachment of bone, and a lower rate of loading on explants without bone. The constraint
of the underlying bone has been suggested to have a protective eﬁect on the matrix of
cartilage during mechanical loading (Jefﬁ'ey et al., 1995). Articular cartilage exhibits a
ﬂow-dependent viscoelastic behavior due to its biphasic composition (Mow et al., 1980).
Studies have suggested that at high rates, loads on this biphasic tissue are mainly
supported by the ﬂuid phase, while for lower rate experiments or at equilibrium, the loads
are supported by the solid phase (Ateshian and Wang, 1995 ; Mow et al., 1980).

Our previous study using unconﬁned compression of cartilage explants
documented an increase in matrix damage and cell death with increasing load for two

different loading rates (Ewers et al, 2000). The study also suggested that at a severe lead

132

level, the higher rate of loading (~900 MPa/s) produced more matrix damage compared
to a lower rate of loading (40 MPa/s). Furthermore, it grossly appeared that the low rate
of loading experiments resulted in more cell death than the high rate of loading
experiments especially at an applied pressure of 40 MPa.

The aim of the current study was to investigate the effect of loading rate on the
extent of matrix damage and cell death in severely traumatized explants of articular
cartilage. Based upon the rate dependent nature of cartilage and earlier studies (Chen et
al, 1999; Ewers et al., 2000), we hypothesized that the extent of matrix damage would be
higher in high versus low rate of loading experiments, while cell death would be greater-
in low versus high rate experiments.

METHODS AND MATERIALS

Bovine forelegs were obtained within three hours of slaughter (18-24 months of
age). The legs were cut just above the knee joint leaving the metacarpal joints intact. The
legs were rinsed with distilled water, skinned and then rinsed again prior to opening the
knee joint under a laminar ﬂow hood A 6.35 mm punch with a smooth edge was used to
make 90 plugs, approximately 15 from the metacarpal surfaces of each limb. The plugs
were separated ﬁ'om the underlying bone using a scalpel. The specimens were placed in a
petri dish containing Dulbecco’s Modiﬁed Eagle’s Medium (DMEM): F12 (Gibco, USA
#12500-039). All explants were washed three times (10 minutes each wash) in DMEM:
F12. Approximately 60 mg of cartilage (2 explants) was placed randomly in wells. Each
well contained one ml of DMEM: F12 supplemented with 50 rig/ml ascorbic acid, 20%
fetal bovine serum, additional amino acids, and antibiotics (penicillian 100 units/ml,

streptomycin 1.0 rig/ml, amphotericin B 0.25 ug/ml) (Rosselot et al., 1992). The explants

133

were allowed to equilibrate in this media for two days in a humidity-controlled incubator
(37° c, 7.2% coz, NuAire, Plymouth, MN).
Alter equilibration, thirty explants (15 wells) were randomly assigned to a non-
impact control group. The remaining specimens were assigned to either a high rate (45
milliseconds to peak, ~900 MPa/s, n=30) or low rate (1 second to peak, 40 MPa/s, n=30)
impact test group. The thickness of each explant was measured using a digital micrometer
(Digimatic, Mitutoyo Corp., Japan). Each specimen was loaded to a peak load of 1247 N
(~ 40 MPa) using a single haversine load-time pulse. The specimens were placed between
two highly polished stainless steel plates in an unconfined compression experiment using
a servo-hydraulic machine (Instron, model 1331, Canton, MA). During this entire process
sterility was maintained. Peak load, time to peak load, and maximum displacement were
recorded in each experiment. After loading, the explants were washed in media three
times (ten minutes each wash) before placing them in pre-assigned wells. One ml of
media was added to each well. Media were collected from each well and replaced daily
for four days post-test.
One day post-test ﬁve explants from each group were randomly chosen for the
cell viability study. Two 1 mm slices were cut from the center of each explant using a
specialized cutting tool The sections were stained with a kit containing calcein and
ethidium bromide homodimer (Live/Dead & Viability/Cytotoxicity, Molecular Probes,
Oregon). All specimens were viewed in a ﬂorescence microscope (Leica DM LB
(frequency: 50-60 Hz), Lecia Mikroskopie und Systeme GmgH, Wetlzar, Germany). Full
thickness digital images (100x) of the explant were taken at the center of impact over a

2.5 mm length for each slice (Spot Digital Camera, Diagnostic Instruments Inc.). Cell

134

viability was quantiﬁed by manually counting the dead (red) and viable (green) cells for
each explant using image software (Sigma Scan, SPSS Inc., Chicago, IL). The data from
the two slices were averaged for each explant. The above procedure was repeated on
another ﬁve explants from each group four days post-test. The remaining explants, not
used for the cell viability study, were stained with India ink to highlight surface ﬁssures.
Surface photographs were made of each explant. Two ﬁrll width sections of each explant
were used to estimate the depth of these surface lesions. The total length of the surface
ﬁssuring was measured in the photographs, and their average depth through the thickness
was documented using image software (Sigma Scan, SPSS Inc., Chicago, IL).

Following the last collection from each well, the media were quantitatively
analyzed for nitric oxide and glycosaminoglycans. Nitric oxide was determined as
previously described and documented as the concentration of nitrite in the media per well
per day (Orth et al., 1999). The content of glycosaminoglycan in each well was
determined using the DMB assay (Chandrasekhar et al., 1987). Chondroitin sulphate was
used as a standard

The data for lesion length and depth and chondrocyte viability was subjected to an
ANOVA with SNK post-hoe tests. Nitric oxide and glycosaminoglycan concentrations
were analyzed every day post-test with an ANOVA and SNK post-hoe tests. T-tests were
used at each strain level to compare axial stress generated for high and low loading rate
experiments. Statistical signiﬁcance was indicated at p < 0.05.

RESULTS
The peak load generated in the high rate of loading group was not signiﬁcantly

different compared to that in the low rate of loading group (1245 i 13 N). The time to

135

peak was 43.0 :t 2.3 ms (~930 MPa/s) and 1.00 i 0.01 seconds (40 MPa/s) for the high
and low rate of loading groups, respectively. The maximum strain produced in specimens
from the high rate group was signiﬁcantly decreased relative to the low rate group (0.409
i 0.074 versus 0.476 :1: 0.088). The stress generated at strains ranging from 0.1 to 0.5 in
the high rate of loading experiments was signiﬁcantly greater than that generated in the
low rate of loading experiments (Figure 1).

Gross observations after loading indicated surface ﬁssures existed across the top
surface for all loaded explants (Figure 2). No ﬁssures were observed on the sides or
bottoms of any loaded or control explants. Two unimpacted, control explants had one
superﬁcial lesion each. There was a statistically signiﬁcant greater total length of
ﬁssuring, and average depth of the ﬁssuring for both test groups compared to controls, as
expected (Table l). Importantly, the specimens loaded at the high rate had a signiﬁcantly
greater total length and average depth of surface lesions than measured on explants
loaded at the low rate.

Cell viability was not signiﬁcantly different in any group between one and four
days post-test, so the data were pooled. No differences were seen in the total number of
cells between groups (Table 1). Since at the edge (approximately 5 % of the diameter) of
each specimen there was a small region of cell death due to the punch (22), this region
was not included in the digital analysis. While there was a signiﬁcant increase in the
number of dead cells and percentage of dead versus live cells for both test groups
compared to controls, in explants subjected to a low rate of loading the number of dead
cells and the percentage of dead to live cells was signiﬁcantly higher than in explants

subjected to a high rate of loading. Qualitatively it was observed that the distribution of

136

dead cells ﬁom the high loading rate experiments was typically near the surface ﬁssures
(Figure 3). In contrast, following the low rate of loading experiments a more diffuse
distribution of dead cells was present throughout the explant. However, in this study, only
total number of dead cells within the deﬁned area of ﬁrll thickness to 2.5 mm slices was
quantiﬁed using the image analysis software. Differences in distribution of dead cells
among different locations within the deﬁned area were not quantiﬁed. Our cell viability
assay does not distinguish between apoptosis and necrosis. However, in a preliminary
study (unpublished observation) we observed all cell death occurring by three hours,
which suggests that the majority of cells are dying by necrosis, not apoptosis.

One day after loading the explants there was a signiﬁcant increase of NO in
specimens from the high rate of loading group compared to controls (Figure 4). This was
followed by an increase of NO in specimens ﬁom both test groups compared to controls
at two days. The amount of NO returned to control levels in both test groups at day three.
N0 NO was measured in the media of any group after four days. Glycosaminoglycan
released to the media increased for both traumatized groups compared to controls each
day post-test (Figure 5). A signiﬁcantly greater amount of glycosaminoglycan was
released to the media from those explants subjected to a high rate of loading (96 :l: 6
rig/ml) than from those exposed to a low rate of loading (74 :l: 6 rig/ml) after one day in
culture. After two days, however, the amount of glycosaminoglycan released to the media
from explants subjected to the low rate of loading (85 i 7 rig/ml) was greater than that
released from specimens exposed to the high rate of loading (75 :l: 4 jig/m1). At days three

and four there were no signiﬁcant differences in the amount of glycosaminoglycan

released to the media between the test groups of explants.

137

DISCUSSION

Our hypothesis was supported in the current study. We documented greater matrix
damage in the explants subjected to a high rate of loading compared to explants exposed
to a low rate of loading. The high rate of loading experiments resulted in cell death
adjacent to ﬁssures, whereas the low rate of loading experiments produced a more diffuse
distribution of cell death away from the ﬁssures. Interestingly, we found that the low rate
of loading experiments produced more cell death compared to the high rate of loading
experiments at a given applied pressure.

The applied pressure in the cmrent study was chosen based upon a previous study
that documented signiﬁcant matrix damage and cell death with both the high and low
rates of loading (Ewers et al., 2000). Furthermore, the high rate of loading (~900 MPa/s)
and the low rate of loading (40 MPa/s) are representative of the previous two sets of
studies examining matrix damage and cell death. The current study was consistent with
this previous study in that a low rate of loading grossly resulted in more cell death
compared to a high rate of loading. The earlier study also suggested that a higher rate of
loading would result in more matrix damage versus a lower rate of loading.

The results of the current study compare well to previous studies of others. Chen
et al. (1999) using a repetitive loading protocol on cartilage explants, documented
increased water content, indicative of matrix damage, compared to controls for an impact
event with a peak stress of 2.5 MPa for a 30 MPa/s or greater maximum stress rate. On
the other hand, a smoothly rising load did not result in increased water content until the
peak stress was 10 MPa for a ‘30 MPa/s or greater maximum stress rate. In the current

study the high rate of loading resulted in cell death largely adjacent to ﬁssures, which has

138

also been documented in previous studies using higher rates of loading (Oyen-Tiesma et
al, 1999; Repo and Finlay, 1977). The low rate of loading experiment produced a more
diffuse distribution of cell death, which is similar to previous studies that documented no
clear spatial association between cell death and matrix ﬁssures with lower rates of
loading (Quinn et al., 1998; Torzilli et al., 1999). Loening et al. (1999), however, using a
low rate of loading (32 MPa/s) on cartilage explants documented chondrocyte apoptosis
before detectable mechanical damage as measured by glycosaminoglycan (GAG) release,
water content, and altered material properties.

The increase in NO released into the media after loading at one and two days
post-test was similar to previous studies that documented a marginally signiﬁcant
increase in NO one day after severe compression (Loening et al., 1999). However, the
concentrations of NO were very low relative to cytokine—stimulated NO production
(unpublished observation). Thus the relevance of the released NO during loading needs to
be clariﬁed. The current study documented an increase in GAGs released into the media
for both rate of loading experiments compared to controls out to four days post-trauma.
This result was similar to a previous study that documented an increase in proteoglycan
release into the media out to six days post-loading (Quinn et al., 1998). Based upon a
previous study, the content of proteoglycan in bovine articular cartilage is approximately
32 rig/mg (Homandberg et al., 1992). During two days of equilibration the DMB assay
indicated an approximate 10% loss of GAG. On average the loaded explants lost
approximately 12% more GAG after four days, while the control explants lost
approximately 6%. The greater release of GAGs for the high rate of loading experiments

compared to the low rate of loading group at one day post-test may be the result of more

139

matrix damage, allowing for the release of GAGs to the media On the other hand, at two
days post-test the low rate of loading experiments produced a higher level of GAG
released to the media than for the high rate of loading group. This result might have been
caused by the higher cell death for the low rate of loading experiments resulting in a cell-
mediated degradation of proteoglycans after two days in culture. It is unclear if this loss
of GAG represents cell-mediated degradation or a direct loss due to matrix damage (Kurz
et al, 2000). At this time the mechanism of GAG release into the media is unknown and
would require fruther investigation. .

The observed differences between the high and low rate of loading experiments
may be due to the ﬂow-dependent viscoelastic behavior of articular cartilage. Since
cartilage is biphasic, faster rates of loading result in more of the load being supported by
the ﬂuid phase, causing an increase in tissue ﬂuid pressure. These large ﬂuid pressures
result in correspondingly larger tensile stresses in the solid matrix, which may be
responsible for producing gross ﬁssuring of the articular cartilage (Garcia et al., 2000).
However, a contact analysis of transversely isotropic, biphasic cartilage showed no
changes in contact area out to one second after application of joint load, suggesting a
minimal rate of loading effect over this time frame (Donzelli et al., 1999). Conversely, a
non-linear, biphasic model of cartilage may result in an increased percentage of the load
being carried by the ﬂuid at large deformations (Garcia et al., 2000). Cartilage, when
modeled as a non-linear, biphasic material, may result in signiﬁcant differences in the
stress ﬁeld for high and low rate of loading experiments, which would help explain the
ﬁndings of the current study. A further complexity in the analysis of the stress ﬁeld

results from the ﬁssures that are generated during the loading of these explants. Fissures

140

would cause large stress concentrations, dramatically altering the stress ﬁeld. Thus, it is
clear that the stress ﬁeld is not uniform through these explants, and further computational
modeling is needed to describe the stress causing surface lesions and cell death.

One limitation of the current study was that the cartilage was removed from the
underlying bone before loading. Bone is known to constrain the cartilage, and this has
been suggested to have a protective effect on the cartilage matrix during loading (Jeffrey
et al., 1995). Thus, a layer of underlying bone would likely alter the state of stress in the
cartilage, and therefore change the stress state around ﬁssures to alter the propagation of
these lesions and the distribution of dead cells. It was interesting to note that in the
current study, even without the constraint of the underlying bone, visible matrix damage
was limited to the surface of the explant. Furthermore, the intent of this study was to
explain a contradiction in the literature dealing with the inﬂuence of loading rate on
matrix damage and cell death for explanted cartilage tissue. This study does not address
the response of an in vivo joint to blunt loading. However, clinical studies do indicate
that surface lesions and dead cells are characteristics of degenerative joint diseases such
as OA (Brandt et al., 1986; Simon et al., 1976). Another limitation of this study was that
it examined only a severe load level. The rate of loading effects may not be as apparent at
lower levels of loading, but future studies are warranted at less severe levels of load. A
ﬁnal limitation is that the equilibration and possible subsequent swelling may have
resulted in the tissue becoming more sensitive to mechanical loading. The increase in
water content would lead to relatively more tensile stresses developed in the solid phase

during impact that could enhance the extent of matrix damage (Garcia et al., 2000).

141

However, this effect needs further study, as does the effect of a loss of lateral constraint
in the tissue explants.

In conclusion, this study indicated that the rate of loading can signiﬁcantly affect
the degree of matrix damage, the distribution of dead cells, and the amount of cell death
in unconﬁned compression experiments on explants of articular cartilage. The role played
by rate of loading on a joint and the pathogenesis of a chronic disease such as OA will
need further investigation
ACKNOWLEDGMENT

This study was supported by a grant for the Centers for disease Control and
Prevention (R49/CCR503607). Its contents are solely the responsibility of the authors and
do not necessarily represent the ofﬁcial views of the Centers for Disease Control and
Prevention. The authors wish to gratefully acknowledge Cliff Beckett for technical
assistance.

REFERENCES

1. Ateshian GA, Wang H (1995) A theoretical solution for the frictionless rolling
contact of cylindrical biphasic articular cartilage layers. J Biomech 28:1341-1355.

2. Blanco FJ, Guitian R, Vazquez-Martul E, de T oro FJ, Galdo F (1998) Osteoarthritis
chondrocytes die by apoptosis. A possible pathway for osteoarthritis pathology.
Arthritis Rhuem 41 :284-289.

3. Brandt KD, Mankin HJ, Shulman LE (1986) Workshop on etiopathogenesis of
osteoarthrits. J Rhuematol 13:1 126-1 160.

4. Buckwalter JA (1995) Osteoarthritis and articular cartilage use, disuse, and abuse:
experimental studies. J Rhuematol Suppl 43:13-15.

5. Chandrasekhar S, esterman MA, Hoffman HA (1987) Microdetermination of

proteoglycans and glycosaminoglycans in the presence of guanidine hydrochloride.
Anal Biochem 161 :103-108.

142

6. Chen C-T, Burton-Wurster N, Lust G, Bank RA, Tekoppele JM (1999) Compositional
and metabolic changes in damaged cartilage are peak-stress, stress-rate, and loading
duration dependent. J Orthop Res 17:870-879.

7. Donzelli PS, Spilker RL, Ateshian GA, Mow VC (1999) Contact analysis of biphasic
transversely isotropic cartilage layers and correlations with tissue failure. J Biomech
32:1037-1047.

8. Ewers BJ, Dvoracek-Driksna D, Orth MW, Haut RC (2000) Matrix damage and
chondrocyte death in articular cartilage depends upon loading rate. Transactions 46"'
ORS (abstract).

9. Garcia JJ, Altiero NJ, Haut RC (2000) An approach for the stress analysis of
transversely isotropic biphasic cartilage under impact load. J Biomech Eng, 12211-8.

10. Homandberg GA, Meyers R, Xie D (1992) Fibronectin fragments cause chondrolysis
of bovine articular cartilage slices in culture. J Bio Chem 267(6):3597-3604.

11. Jeffrey JE, Gregory DW, Aspden RM (1995) Matrix damage and chondrocyte
viability following a single impact load on articular cartilage. Arch Biochem Biophys
322:87-96.

12. Kurz B, Jin M, Patwari P, Lark MW, Grodzinsky AJ (2000) Biosynthetic response
and mechanical properties of articular cartilage after injurious compression: The
importance of strain rate. Transactions 46“ ORS (abstract).

13. Loening AM, Levenston ME, James IE, Nuttal ME, Hung I-IK, Gowen M,
Grodzinsky AJ, Lark MW (1999) Injurious compression of bovine articular cartilage
induces chondrocyte apoptosis before detectable mechanical damage. Transactions
45“ ORS (abstract).

14. Mow VC, Kuei SC, Lai WM, Armstrong CG (1980) Biphasic creep and stress
relaxation of articular cartilage in compression: Theory and experiments. J Biomech
Eng 102273-84.

15 . Newberry WN, Mackenzie CD, Haut RC (1998) Blunt impact causes changes in bone
and cartilage in a regularly exercised animal model. J Orthop Res 16:348-354.

16. Orth MW, Fenton JI, Chlebek-Brown KA (1999) Biochemical characterization of
cartilage degradation in embryonic chick tibial explant cultures. Poultry Science
78:1596-1600.

l7. Oyen-Tiesma M, Deloria LB, Meglitsch T, Oegema TR, Lewis JL (1999) Cell death
alter cartilage impact depends on matrix damage. Transactions 45" ORS (abstract).

143

18.

19.

20.

21.

22.

23.

Quinn TM, Grodzinsky AJ, Hunziker EB, Sandy JD (1998) Effects of injurious
compression on matrix turnover around individual cells in calf articular cartilage
explants. J Orthop Res 16:490-499.

Repo RU, Finlay JB (1977) Survival of articular cartilage after controlled impact. J
Bone Joint Surg [Am] 59: 1068-1076.

Rosselot G, Reginato AM, Leach RM (1992) Development of growth hormone an
insulin-like growth factor I on cultured post-embryonic growth plate chondrocytes. In
Vitro Cell Dev Biol 28Az235-244.

Simon WH, Richardson S, Herman W, ParsOns JR, Lane J (1976) Long-term effects
of chondrocyte death on rabbit articular cartilage in vivo. J Bone Joint Surg [Am]
58:517-526.

Tew SR, Kwan APL, Hann A, Thompson BM, Archer CW (2000) The reactions of
articular cartilage to experimental wounding: Role of apoptosis. Arth Rheum 43:215-
225.

Torzilli PA, Grigiene R, Borrelli Jr. J, Helfet DL (1999) Effect of impact load on

articular cartilage: Cell metabolism and viability, and matrix water content. J
Biomech Eng 121:433-441.

144

TABLES

Table 1: Quantiﬁcation of matrix damage and chondrocyte viability for control
(Control), high rate of loading experiments (High), and low rate of loading

 

 

 

experiments (Low) (Avg :1: SD).
Control High Low
Manganese ‘
Total ﬁssure length (mm) 0.2 i 0.5 62.4 a 15.7 ‘ 52.1 a 11.4 ‘*
Average ﬁssure depth (%) 1.5 :l: 4.4 34.5 i 10.5 i 23.0 :t 8.2 N
Chairmen: viah' ility b
Total cells (#) 163 i 59 177 :l: 45 172 j: 55
Deadcells(#) 17:10 83:1:29’ 122:35"
Deadcells(%) 921:6 47:12’ 66:1: 13““
a - n = 18
b-n=10

* - signiﬁcantly greater versus control group by AN OVA.
+ - signiﬁcantly different versus high rate of loading experiments by AN OVA.

145

FIGURES

 

 

 

 

50 -
+Low

4o - +High'
it?
n.
g 30 -
3
g 20 '1
a

10 -

O ’l r r l T j
0 0.1 0.2 0.3 0.4 0.5 0.6
Strain

Figure 1: The average stress-strain responses of cartilage in unconﬁned compression for
high (I) and low (0) rate of loading experiments (Ave :l: SD). There was a signiﬁcant
increase in the stress generated in the high rate of loading experiments compared to the
low rate of loading experiments for compressive strains levels ranging from 0.1 to 0.5 (p
<0.05).

146

 

Figure 2: The surfaces of the cartilage explants were wiped with India ink to highlight
surface ﬁssures. Typical specimens are shown from the high rate of loading group (A)
and ﬁ‘om the low rate of loading group (B).

147

   

(A) (B)

Figure 3: In the stained sections of the explants it was impossible to distinguish the dead
cells in the grayscale image. The image was digitally altered to make the background
gray, the live cells appear as light gray points and the dead cells appear as black points.
The sections are oriented with the surface to the left. The cartilage surface and fissure
have been outlined in these sections, since they are difficult to distinguish. The high rate
of loading experiments typically showed cell death adjacent to the surface ﬁssure (A).
The low rate of loading experiments typically indicated a greater range of cell death with
dead cells beyond the ﬁssure (B).

148

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VIII Controll

g Slow
.3 I Fast
0
P.
x
O
.2
E-
z

1 2 3 4

Days Post-Trauma

Figure 4: The average amount of nitric oxide (NO) released into the media per well each
day post-test for control (Control), high rate of loading experiments (High), and low rate
of loading experiments (Low) (Ave 1 SD, n=12). The (+) symbol indicates a statistically
signiﬁcant increase in NO release compared to controls at that day post-test (p <0.05).

149

_h
N
O

r

43:

+ l—‘ ; [1 Control
‘ Slow

—L
O
O
I
+

    

m
0
r

.A
O
r

Glycosaminoglycans (uglml)
N O)
O O

 

o
l

Days Post-Trauma

Figure 5: The average glycosaminoglycan released into the media each day post-test
control (Control), high rate of loading experiments (High), and low rate of loading
experiments (Low) (Ave :t SD, n=12). The (+) symbol indicates a statistically signiﬁcant
increase in glycosaminoglycan release compared to controls at that day post-test (p
<0.05). The (# ) symbol indicates a statistically signiﬁcant difference between high rate
of loading experiments and low rate of loading experiments (p <0.05).

150

CHAPTER 8

THEORETICAL MODELS OF CARTEAGE EXPLANTS AND
CORRELATIONS WITH EXPERIIVIENTAL MATRIX DAMAGE AND CELL
DEATH AFTER IMPACTING LOADS

Ewers BJ, Krueger JA, +Dvoracek-Driksna D, +Orth MW, Haut RC

ABSTRACT

Excessive mechanical loading of articular cartilage can lead to matrix damage and
chondrocyte death. The relationship between this damage and secondary osteoarthritis is
still unknown. In vitro studies of explants have suggested that matrix damage and cell
death correlate with the state of stress and strain produced by applied boundary tractions.
Recently, however, confocal microscopy has shown that local matrix strains differ from
the strains produced in adjacent cells under physiological loading. The current study
documents mechanically induced matrix damage and cell death in explants, and then
attempts to correlate the mechanical “damage with the state of stress and strain in the
matrix and embedded cells. Thirty bovine cartilage explants were equilibrated for two
days. Ten served as controls and the remaining randomly underwent unconﬁned
compression to 30 MPa at either a high («667 MPa/s) or low rate of loading (30 MPa/s).
Matrix damage and cell viability were documented for each group. The experimentally
determined displacements for both the high and low rate experiments were then applied
to isotropic and transversely isotropic computational models of the explant, with or

without depth-dependent, matrix material properties. The amount of displacement on a

151

cell was then linearly interpolated and applied to a computational model of the cell. The
isotropic model with depth-dependent material properties best associated areas of high
stress with matrix damage and predicted the lateral expansion proﬁles documented in the
recent literature. The current study also suggested that the extent of cell strain in each
layer was most inﬂuenced by the shape of the cell, not the depth-dependent material
properties of the matrix.

INTRODUCTION

Osteoarthritis (OA) is a disease affecting diarthrodial joints and is characterized
by full thickness loss of articular cartilage resulting in bone-on—bone contact and joint
pain. OA may cost society $54 billion annually in treatment and lost workdays
(Hamermann, 1989). Excessive and abnormal mechanical loading of cartilage has been
implicated in the pathogenesis of OA Clinically, matrix damage of cartilage, such as
surface ﬁssuring, and chondrocyte death have been documented in osteoarthritic cartilage
(Kim et al., 2000). However the role of acute injuries, such as those given above, in the
mechanism of a post-traumatic, secondary 0A are not well understood.

In vitro studies, using isolated cartilage explants, have attempted to correlate
excessive mechanical loading with cell death and matrix damage. In hallmark studies
Repo and Finlay(l977), using tissue explants on bone and subjected to a single impact of
25 MPa pressure, document cell death localized around ﬁssures in the superﬁcial layer of
the cartilage. Torzilli et al. (1999) also documents cell death in the superﬁcial zone of
cartilage explants without bone at lower applied pressure (10 MPa). Chen et al. (1999)
showed that matrix damage occurs in the same area under repetitive loading with peak

boundary pressures of 2.5 MPa. Ewers et al. (2000) document in cartilage explants

152

without underlying bone that the rate of loading can affect the distribution of cell death.
They found that a low rate of loading (40 MPa/s) causes cell death deeper in the explant
compared to that occurring in a high rate of loading (800 MPa/s) experiment to the same
peak load Matrix damage, on the other hand, is primarily located in the superﬁcial layer
of the explant for both rates of loading.

The above in vitro studies have been used in attempts to correlate the state of
stress and strain in a loaded cartilage explant with the locations and degree of matrix
damage and cell death, by considering only the levels of applied boundary traction.
Clearly, comparing studies when some explant studies have underlying bone and others
do not is difﬁcult. It is evident that with bone attached to the explant there is a
distribution of tissue stress through its depth as one approaches the attached bone.
Boundary tractions are also not necessarily representative of the state of tissue stress and
strain through the explant thickness in the case of no underlying bone, primarily because
the material properties of cartilage are known to be depth-dependent (Roth and Mow,
1980; Schingal et al., 1997). This factor may play a signiﬁcant role in the non-uniform
distribution of matrix damage and cell death documented in previous experimental, in
vitro studies.

The state of stress and strain in the explant matrix may also be quite different than
that in adjacent cells. Guilak et al. (1995) have experimentally shown with confocal
microscopy that local matrix strains, in fact, are different than strains produced in
adjacent cells under physiological levels of loading. Guliak and Mow (2000) have
subsequently generated a computational model of cartilage cells embedded in a cartilage

matrix. Their study suggests that large differences in the elastic properties of the cell and

153

adjacent matrix result in nearly double the strains in the cell versus the adjacent matrix.
The previous study, however, did not speciﬁcally examine the effect of depth-dependent
changes in the material properties of the matrix.

The object of the current study was to extend the modeling efforts of Guilak and
Mow (2000) in the analysis of an impact event where cell death occurs, and attempt to
correlate the distribution of matrix damage and cell death with the state of stress and
strain in the explant matrix and the embedded cells. The current study will also begin to
address some of the effects of depth-dependent material properties of cartilage. Tissue
shear stress and tensile strains have previously been correlated with acute ﬁssuring of
cartilage (Atkinson et al., 1998). Large cell strains have also been hypothesized to cause
signiﬁcant leakage of ﬂuorescent indicators across cell membranes in preliminary studies
(Guliak et al., 1995). This suggests that membrane damage and the possibility of cell
death may occur from excessive cell strain (Guliak et al., 1995). We hypothesize that
matrix damage will correlate with high tissue stress, based on previous studies from our
laboratory, and that cell death will be associated with high cell strain. For both
hypotheses, we also believe that the depth-dependent material properties of the explant
matrix should be considered in theoretical models.
METHODS
I 1 l . 1 I 1' E l E 1

One pair of bovine forelegs were obtained ﬁom a local abattoir within six hours
of slaughter, and skinned prior to opening the knee joint under a laminar ﬂow hood A 6-
mm diameter biopsy punch was used to make 30 plugs from the metacarpal surfaces. The

plugs of cartilage were removed from the underlying bone with a scalpel. The specimens

154

were placed in Dulbecco’s Modiﬁed Eagle’s Medium (DMEM):F12 (Gibco, #12500-039,
and allowed to equilibrate for two days in a humidity controlled incubator (37° C, 7.2%
CO;, NuAire, Plymouth, MN). Ten explants were randomly assigned to a non-impact
control group. The remaining explants were loaded to 850 N (~30 MPa) using a high rate,
computer controlled haversine pulse (50 ms to peak, n=10) or a low rate pulse (1 s to
peak, n=10). The specimens were loaded between two highly polished stainless steel
plates in unconﬁned compression using a servo-hydraulic testing machine (Instron,
model 1331, Canton, MA). After the impact loads were applied, 2 explants
(approximately 50 mg of cartilage) were placed in each well with 1 ml of media. Cell
viability was determined 24 hours post-impact using established techniques. Brieﬂy,
three 0.5 mm thick slices per explant were stained with Calcein AM and Ethidium
Homodimer (Live/Dead & Viability/Cytotoxicity, Molecular Probes, Oregon). Specimens
were viewed in a ﬂorescence microscope (Lecia DM LB, Lecia Mikroskopie und
Systeme GmgH, Germany), and full thickness digital images were made of the center 3-
mm of each slice (Spot Digital Camera, Diagnostic Instruments Inc.). Each full thickness
slice was divided into three layers: superﬁcial (top 20%), middle (50%), and deep
(bottom 30%). A blinded observer (MK) quantiﬁed cell viability by manually counting
live and dead cells in each zone using image software (Sigma Scan, SPSS Inc., IL). The
percentage of dead cells in each zone was calculated (# dead cells / # total cells).
ANOVA with SNK post-hoe tests were used to determine statistical differences in cell
death in each zone for the two rates of loading, and versus controls. Data were
documented as the mean i one standard deviation. Statistical signiﬁcance was indicated

at p<0.05.

155

QamprrtatienalMadsl

The geometry of the computational model (Abaqus 5.8, Hibbit, Karlsson &
Sorenson Inc., RI) consisted of an axisymmetric 6-mm diameter disk of cartilage one mm
thick, which corresponds to the average thickness of cartilage disks taken ﬁom bovine
metacarpal joints (Ewers et al., 2000). Since it was assumed that the stainless steel
surfaces in the impact experiments resulted in near frictionless interfaces with the
cartilage explant, the boundary conditions for the model corresponded to a simple,
unconﬁned compression test. The bottom nodes were ﬁxed in the axial (thickness)
direction The top nodes on the model were vertically displaced with the average
deformation recorded in either the high or the low rate of loading experiment. The mesh
consisted of 300 eight-noded poroelastic elements (CAX8P). The model was divided into
the same three layers as used for the cell viability experiments. For the non-layered
model, each layer was given the same material properties. A biphasic analysis was used
for the layered and non-layered models with either an isotropic (I) or transversely
isotropic (TI) solid skeleton (Table 1). Under unconﬁned compression the axial direction
was in compression, while the radial direction was in tension. The TI model allowed for
an approximation of the differences in the compressive and tensile moduli of the cartilage
explant, as given in the literature (Table 1). The axial modulus (E33) was determined from
Schingal et al. (1997) by averaging the conﬁned compression modulus in each layer.
Poisson’s ratios were assumed zero, per Cohen et al. (1993) in each layer. The radial
modulus (Eu) was taken from Roth and Mow (1980) by averaging the tensile moduli
parallel and perpendicular to superﬁcial split-lines for each zone. The remaining material

property (out-of-plane shear modulus - 013) was assumed to behave in an isotropic

156

manner, ie: G; = E33 / 2(1+v13). The permeability of the cartilage (k) was assumed
isotropic, as given by Cohen et al. (1993). The fraction of solid volume ((0’) was also
taken ﬁ'om Cohen et al. (1993). For the isotropic model the radial and axial elastic moduli
were averaged for each layer. The maximum shear stresses and the tensile strains were
averaged in each layer of the matrix.

To approximate the state of strain in cells, an axisymmetric elastic microscopic
cell model (Abaqus 5.8) was derived from the macroscopic cartilage model (Figure l).
The cell model combined the geometry and assumed material properties of both the cell
and the pericellular matrix, per Guliak (2000). In the superﬁcial layer, the geometry of
the chondrocyte was assumed to be elliptical with a height of 5 um and a width of 15 um,
as experimentally measured by Guilak et al.(1995). In the middle and deep zones, a
spherical shape was assumed with a diameter of 15 um (Guliak et al., 1995). The
pericellular matrix was assumed to be 2.5 11m thick (Guliak and Mow, 2000; Lee et al.,
1997). Both the cell and the pericellular matrix were assumed isotropic with an elastic
modulus of 9.5x10“ MPa and 1.5x10'3 MPa, respectively, and nearly incompressible (v =
0.499) (Jones et al., 1999; Jones et al, 1997). The material properties of the matrix in the
cell model were the same as those used in the macroscopic explant model. The mesh for
the cell model consisted of 1550 and 1034 eight-noded axisymmetric elements (CAX8)
for the elliptical and spherical cell shapes, respectively. The displacements between the
nodes from one element along the axis of symmetry in the center of each layer of the
macroscopic model were linearly interpolated for use as a distributed displacement

boundary condition in the cell model. The average tensile (£1) and average maximum

shear (YMAx) strains were determined in cells for each layer.

157

RESULTS

Unconﬁned compression of the cartilage explants resulted in experimental impact
loads of 848 :t 11 N and 838 :l: 21 N for the low and high rate of loading tests,
respectively. These data were not statistically different (p > 0.05). The time to peak load
was 1.00 i 0.05 s and 45.0 i 2.5 ms for the low and high rate of loading experiments,
respectively (p < 0.05). Analysis of the load-displacement data from the experiments
indicated that the low rate of loading experiments resulted in slightly more axial
deformation of the explant than in the high rate of loading experiments, 0.32 i 0.04 mm
and 0.29 i 0.04 mm, respectively. The axial deformations, however, were not
signiﬁcantly different (p > 0.05). Matrix damage was observed in all specimens (both rate
of loading experiments) as minor lesions in the superﬁcial zone with no observable radial
variation (Figure 2).

Cell death was produced in all loaded specimens. Cell death did not have any
observable radial variation in the loaded explants. The total percentage of dead cells
following the low rate of loading experiments was signiﬁcantly greater than that
produced in the high rate of loading experiments, 51 i 16% versus 33 i 16%,
respectively. Both cases resulted in signiﬁcantly more dead cells than in controls (0.63 i
0.85%). For the high and low rate of loading experiments there were signiﬁcantly more
dead cells in the superﬁcial layer of the explant than in the deeper zones (Figure 3). In the
superﬁcial and middle zones a signiﬁcantly higher percentage of dead cells existed in the

low versus the high rate of loading experiments (Figure 4).

158

Using the average experimental deformations generated in the low and high rate
of loading experiments, the computational explant models indicated slightly larger shear
stresses and tensile strains in the matrix for the low rate of loading than for the high rate
of loading models (Table 2). As expected, the non-layered computational models showed
no radial or depth-dependent differences in maximum shear stresses within the matrix
(Figure 5 A). While the layered computational models showed only a minor radial-
dependent distribution of shear stress, they did indicate depth-dependencies in maximum
shear stress (Figure 5B). The layered models indicated a 5 and a 2.3 fold increase in the
maximum shear stress in the superﬁcial layer compared to the deep zone, for the isotropic
and transversely isotropic models, respectively (Table 2). This region of high shear stress
in the layered models corresponded with the region of experimentally produced matrix
damage in the superﬁcial layer of the explants.

The non-layered computational models showed no radial or depth-dependent
differences in tensile strain in the matrix (Figure 6A). The layered models, however,
indicated both radial and depth-dependent differences in matrix tensile strains. The
isotropic layered model did not show signiﬁcant radial-dependent changes in matrix
tensile strains in the superﬁcial and middle layers, but there was a signiﬁcant increase in
tensile strain in the deep layer near the edge of the model (Figure 6B). Yet, the
experiments did not produce any matrix damage in this region of the explant. On average,
the isotropic layered model indicated a 33% increase in matrix tensile strain through the
depth of the explant. The transversely isotropic model showed large radial-dependent
variations in tensile strains in all layers, with the largest strain near the edge of the

explant (Figure 6C). On average, the transversely isotropic model indicated little

159

variation in matrix tensile strain through the depth of the explant (Table 2). Overall, the
variations in maximum tensile strain in the explant models did not correspond well with
the locations of matrix damage in the experiments.

The cell models resulted in larger strains in the cells than in the local matrix. As
in the macroscopic explant models, due to more defamation in the low compared to high
rate of loading models, the cell models indicated slightly larger tensile and shear cell
strains in the low rate of loading than in the high rate of loading experiments (Table 3).
The cell models indicated that the strains were minimum in the center of the cells and
maximum near the periphery, with a concentric distribution that varied radially by less
than 10%. The computational cell models from the non-layered explant models indicated
similar average tensile strains between layers (3% variance). The cell models for the
layered, isotropic explant model indicated approximately a 24% lower tensile strain in the
superﬁcial layer than in the deep layer. This was in contrast to more cell death in the
superﬁcial than in the deep zone. The layered transversely isotropic cell model also
showed minor variations in tensile strains between layers (10%), and the results were also
in contrast to the observed distribution of dead cells in the corresponding experiments.

The computational cell models from the non-layered explant models indicated
approximately 34 and 27% increases in average shear strain for the isotropic and
transversely isotropic models, respectively, in the superﬁcial layer compared to the
middle and deep zones. These results corresponded better with the experiments where
signiﬁcantly more dead cells were observed in the superﬁcial layer than in the deeper

layers. The layered models also indicated approximately 20 and 15% increases in average

160

cellular shear strain in the superﬁcial layer compared to the deeper layers for the isotropic
and transversely isotropic models, respectively.

The non-layered models also exhibited large cellular shear strains in the
superﬁcial layer compared to the deeper zones. This result indicated that the large cellular
strains generated in the superﬁcial layer may be due to the cell shape, not necessarily the
depth-dependent material properties of the matrix. A small parametric study was
performed by inserting a spheroidal shape cell into the superﬁcial layer of the cartilage
for the transversely isotropic, layered cell model. When the superﬁcial cells were
modeled as spheroids, rather than ellipsoids the average tensile strains were only slightly
decreased from 0.36 to 0.35. On the other hand, the average cellular shear strains were
more dramatically reduced from 0.62 to 0.52 for the ellipsoid versus the spheroidal cell
models, respectively.

DISCUSSION

The object of the current study was to correlate the distribution of matrix damage
and cell death with the state of stress and strain in the explant matrix and the embedded
cells. We hypothesized that matrix damage would correlate with high tissue stress and
that cell death would be associated with high cell strain. For both hypotheses, we also
theorized that the depth-dependent material properties of the explant matrix would have
to beconsidered in the theoretical models. Both matrix damage and extensive cell death
were documented in the superﬁcial layer following impact in this study. As hypothesized,
a theoretical model with depth-dependent material properties was found to have high
shear stresses in the superﬁcial zone, which corresponds to the documented location of

matrix damage. On the other hand, depth-dependent material properties were not found to

161

produce high cell strains corresponding to regions of cell death High cell strain related
more to cell shape than to depth dependent changes in material properties.

Jurvelin et al. (1997), using an optical measuring system, noted that under
unconﬁned compression the superﬁcial cartilage layer underwent a smaller lateral
expansion than the deep region immediately after ramp loading. After ramping down to
10% compression in 3 seconds, they documented 72% less lateral expansion in the
superﬁcial layer compared to the deep zone. In order to correctly approximate this depth-
dependent lateral expansion with a theoretical model, there was a need to include the
depth-dependent material properties of the cartilage. The current study indicated that the
isotropic, layered model did a better job than the TI layered model of predicting the
distribution of lateral expansion in the deep versus the superﬁcial layers. However, the
isotropic layered model exhibited only a 34% decrease of lateral expansion in the
superﬁcial layer compared to the deep zone. The discrepancy between the earlier
experimental data by Jurvelin et al. (1997) and the current theoretical model may be due
to the timeframe used for loading the explant. Li et al. (2000) recently developed a ﬁbril
reinforced, poroelastic model of cartilage that took into account depth-dependent material
properties of the tissue. They modeled unconﬁned compression and applied ramp
compression in 5 seconds. The study noted that the percentage decrease in lateral
expansion in the superﬁcial region versus the deep zone changed from 58, 61, and 78%,
in 5, 20, and 40 seconds, respectively. Therefore, following Li’s (2000) observations, it is
likely that a loading time of less than 5 seconds, like the 1 second and 50 milliseconds
used in the current study, would result in less than 58% variation in lateral expansion

with depth.

162

The current study also suggested that an important parameter inﬂuencing the
extent of cell strain was the shape of the cell. Ellipsoidal versus spherical cells in the
superﬁcial zone resulted in increased shear strain, which corresponded with the notion of
cell death. Previously, Baer and Setton (2000) examined the inﬂuence of cell geometry
on the micromechanical environment of intervertebral disc cells. In one part of the study,
they modeled the matrix as transversely isotropic and applied a lateral load resulting in
constant nominal strains for both ellipsoidal and spheroidal cells. They also noticed that
the cell strains were decreased when the cell was modeled as ellipsoidal versus spherical.
However, the contrasts between our ﬁndings and those of Baer and Setton (2000) may be
attributed to modeling differences. First, the major to minor cell axis ratio was 3:1 for
ellipsodial cartilage cells in the current study; whereas, Baer and Setton (2000)used a
ratio of 10:1 for the ellipsoidal intervertebral disc cells. The current study also loaded the
cells in the axial direction under large nominal strains (30%), while Baer and Setton
(2000) loaded the cells in the lateral direction under small nominal strains (5%).

In the current study, cellular strains were approximately twice as large as the local
nominal matrix strain, which supports the ﬁndings of previous embedded cell models.
Guilak and Mow (2000) determined that when the elastic modulus of a cell is two or
more orders of magnitude less than that of the surrounding matrix, the peak strains in the
cell are also two times greater than the local matrix strain. Experimentally Guliak et
al.(1995) recorded a 27—37% increase in cell strain versus local matrix strain, in contrast
to the 74-121% in the current model. The Guilak et a1 study, however, was performed
with a physiological loading, which results in a 15% surface-surface strain. In the current

study, the model was subjected to approximately 30% strain for high and low rates of

163

loading. This suggested that the material properties of cells used in the current model
may be inadequate for large deformations. A possible solution to this problem may be to
incorporate a stiffening effect into the material properties of the cells for large strains and
high strain rates.

The biphasic model in the current study was not able to account for differences in
compression for high and low rate of loading experiments. While the deformations
recorded between the two rate of loading experiments were not statistically signiﬁcant in
this study, previous studies with larger sample sizes have shown signiﬁcantly more
deformation in the low versus high rate of loading experiments (Ewewrs et al., 2000).
The permeabilities. used in the models were taken ﬁom the literature and are based on
ﬁtting long duration relaxation experiments, unlike the very short times exhibited in
impact tests. This is supported by a previous study that examined contact analysis for
biphasic materials (Donzelli et al., 1999). The study showed no change in contact area for
times to one second after applying a load. To fully model the apparent fast relaxation
characteristics of cartilage under an impact, a model that incorporates a viscoelastic solid
phase may be needed (Suh and Bai, 1998).

A limitation of the current computational models was that actual experimental
loads could not be applied This was due to the low modulus values used for the tissue
layers, based on the literature. Atkinson et aL (1998‘), using an isotropic elastic model for
the cartilage to model an impact event, suggested that the elastic modulus of cartilage
needs to be approximately 20 times that determined in small deformation experiments.
Previous studies have already attempted to model a nonlinear stiffening effect for the

impact response of cartilage (Garcia et al., 2000; Li et al., 2000). More studies will be

164

needed to incorporate the nonlinear, viscoelastic effects into cartilage models for the
impact event. The material properties used in the current study also led to strains in cells
that exceeded the limits of inﬁnitesimal theory. In order to remedy this problem, there '
will also be a need to determine some of the nonlinear material properties of cells under
impact load scenarios. Although the current study was not able to estimate the absolute
values of cellular strain in different layers of cartilage under impact, it has provided
information on some of the factors that will be needed in models to more accurately
represent cellular strains in various layers of the tissue.

Since matrix and cell stresses and strains were not uniform throughout the
modeled tissue, boundary tractions may not be the best indicator of matrix damage or cell
death. The current study showed that areas of high matrix stress correspond with areas of
matrix damage by incorporating depth-dependent material properties for the cartilage.
The study also indicated that depth-dependent, isotropic models could better predict
experimentally observed variations of lateral expansion from unconﬁned compression
experiments than a transversely isotropic model. Cell shape was also determined to be an
important parameter when examining cellular strains under impact loading.

In summary, the current study has helped conﬁrm the need to take into account
depth dependent variations in cellular shapes and material properties to understand better
the mechanisms of cell death and matrix damage of cartilage from blunt impact loading.
Future studies must also consider models of structures such as the cell membrane and its
cytoskeleton to more completely understand the mechanism of cell death from blunt

impacting loads.

165

ACKNOWLEDGEMENTS

This study was supported by a grant for the Centers of Disease Control and
Prevention (R49/CCR503607). Its contents are solely the responsibility of the authors and
do not necessarily represent the ofﬁcial views of the Centers of Disease Control and
Prevention. The authors with to gratefully acknowledge Cliff Beckett and Masaya
Kitagawa for technical assistance.
REFERENCES

l. Atkinson, TS, RC Haut, and NJ Altiero. (1998) Impact-induced ﬁssuring of articular
cartilage: an investigation of failure criteria J Biomech Eng. 120: 18 1-187.

2. Baer, AE and LA Setton (2000) The micromechanical environment of intervertebral
disc cells: effects of matrix anisotropy and cell geometry predicted by a linear model. J
Biomech Eng. 122:245-251.

3. Chen C-T, N Burton-Wurster, G Lust, RA Bank, and JM Tekoppele. (1999)
Compositional and metabolic changes in damaged cartilage are peak-stress, stress-rate,
and loading-duration dependent. J Orthop Res. 17:870-879.

4. Cohen, B, TR Gardner, and GA Ateshian. (1993) The inﬂuence of transverse isotrop
on cartilage indentation behavior: a study of the human humeral head. Transactions 3
ORS (abstract).

5. Donzelli, PS, RL Spilker, GA Ateshian, and VC Mow. (1999) Contact analysis of
biphasic transversely isotropic cartilage layers and correlations with tissue failure. J
Biomech. 32: 1037-1047.

6. Ewers, BJ, D Dvoracek-Driskna, MW Orth, and RC Haut. (2000) The extent of matrix
damage and chondrocyte death in mechanically traumatized articular cartilage
explants depends on rate of loading. J Orthop Res. in press.

7. Flores, RH and MC Hochberg. (1998) “Deﬁnition and classiﬁcation of osteoarthritis.”
In: Osteoarthritis, edited by KD Brandt, M Doherty, and LS Lohmander. New York:
Oxford University Press, ppl-12.

8. Garcia, JJ, NJ Altiero, and RC Haut. (2000) Estimation of in situ elastic properties of

biphasic cartilage based on a transversely isotropic hypo-elastic model. J Biomech Eng.
122:1-8.

166

9. Guilak F and VC Mow. (2000) The mechanical environment of the chondrocyte: a
biphasic ﬁnite element model of cell-matrix interactions in articular cartilage. J Biomech.
33:1663-1673.

10. Guilak F, A Ratcliffe, and VC Mow. (1995) Chondrocyte deformation and local
tissue strain in articular cartilage: a confocal microscopy study. J Orthop Res, 13:410-
421.

11.Hamerman, D. (1989) The Biology of osteoarthritis, New Eng Journal of Med
320:1323-1330.

12. Jones, WR, GM Lee, SS Kelley, and F Guilak. (1999) Viscoelastic properties of
chondrocytes form normal and osteoarthritic human cartilage. Transactions 45"l
ORS.

13. Jones, WR, HP Ting-Beall, GM Lee, SS Kelley, RM Hochmuth et al, (1997)
Mechanical properties of human chondrocytes and chondrons from normal and
osteoarthritic cartilage. Transactions 43"! ORS (Abstract).

14. Jurvelin, JS, MD Buschmann, and EB Hunziker. (1997) Optical and mechanical
determination of Poisson’s ratio of adult bovine humeral articular cartilage. J
‘ Biomech. 30:235-241.

15. Kim, HA, YJ Lee, SC Seong, KW Choe, and YW Song. (2000) Apoptotic
chondrocyte death in human osteoarthritis. J Rheumatol. 27(2):455-462.

16. Lee, Gm, TA Paul, M Slabaugh, and SS Kelle . (1997) The pericellular matrix is
enlarged in osteoarthritic cartilage. Transactions 43 ORS (abstract).

17. Li, LP, MD Buschmann, and A Shirazi-Adl. (2000) A ﬁbril reinforced
nonhomogeneous poroelastic model for articular cartilage: inhomogeneous response in
unconﬁned compression. J Biomech. 33: 1533-1541.

18. Repo RU and JB Finlay. (1977) Survival of articular cartilage after controlled impact.
J Bone Joint Surg [AM]. 59:1068-1078.

l9. Roth, V and VC Mow, (1980) The intrinsic tensile behavior of the matrix of bovine
articular cartilage and it variation with age. J Bone Joint Surg. 62-A:l 102-1117, 1980.

20. Schinagl, RM, D Gurskis, AC Chen, and RL Sah. (1997) Depth-dependent conﬁned
compression modulus of hill-thickness bovine articular cartilage. J Orthop Res. 15:499-
506.

21. Sub, JK and S Bai. (1998) Finite element formulation of biphasic poroviscoelastic
model for articular cartilage. J Biomech Eng. 120: 195-201.

167

22. Torzilli, PA, R Grigiene, J Borrelli, Jr., and DL Helfet. (1999) Effect of impact load
on articular cartilage: cell metabolism and viabiltiy, and matrix water content. J Biomech
Eng. 121:433-441.

TABLES

Table 1: Matrix biphasic material properties for the non-layered isotropic (I) and
transversely isotropic (TI) and layered transversely isotropic (TI) models.

 

 

 

 

Material Layered Non-Layered
property TI

Superﬁcial Middle Deep 1‘ rl‘
E33 (MPa) 0.16 5 0.60 t 1.78 5 0.69 0.46
Err(MPa) 28.0” 8.8 t 1.8 t 0.69 5.80
v12 0.0 0.0 0.0 0.0 0.0
w, 0.0 0.0 0.0 0.0 0.0
G13(MPa) 0.08 0.30 0.89 0.345 0.37
K (x10-‘5m4/Ns) 5.1’ 5.1 ‘ 5.1‘ 3.0 5.1
o8 0.25 ‘ 0.25 ‘ 0.25 ‘ 0.25 0.25

 

 

* — Cohen et al. (1993)
§ — Schingal et al. (1997)

# — Roth and Mow (1980)

168

Table 2: Average maximum shear stress (1: MAX) (MPa) and tensile strain (81) for each
region and rate of loading for the non-layered and layered isotropic (I) and transversely
isotropic (T I) models.

 

*

 

 

 

 

 

 

 

Rate of Non-Layered Layered
Loading Region I TI I TI
T MAX 31 T MAX 31 T MAX 81 T MAX 31
Low Superﬁcial 0.33 0.16 1.08 0.16 2.61 0.14 1.26 0.21
Middle 0.33 0.16 1.08 0.16 1.12 0.16 0.82 0.20
Deep 0.33 0.16 1.08 0.16 0.55 0.21 0.55 0.22
High Superﬁcial 0.30 0.145 0.98 0.145 2.37 0.13 1.12 0.18
Middle 0.30 ' 0.145 0.98 0.145 0.99 0.14 0.72 0.17
Deep 0.30 0.145 0.98 0.145 0.49 0.19 0.47 0.19

Table 3: Average maximum tensile strains (81) and maximum shear strain (7 MAX) in

the cell for each region and rate of loading for the non-layered and layered isotropic (I)
and transversely isotropic (TI) models.

 

 

 

 

 

 

169

 

 

Rate of Non-Layered Layered
Loading Region I TI I TI
81 MAX 81 MAX 81 MAX 81 MAX
Low Superﬁcial 0.35 0.68 0.30 0.59 0.28 0.61 0.43 0.71
Middle 0.34 0.51 0.31 0.47 0.33 0.50 0.40 0.58
Deep 0.34 0.51 0.31 0.47 0.38 0.54 0.45 0.62
High Superﬁcial 0.32 0.62 0.27 0.54 0.26 0.54 0.36 0.62
Middle 0.31 0.46 0.28 0.42 0.30 0.44 0.33 0.50
De_ep 0.31 0.46 0.28 0.42 0.33 0.45 0.37 0.53

 

 

FIGURES

lllllllll

   

Figure 1: A biphasic axisymmetric ﬁnite element model was developed for a cartilage
explant under unconﬁned compression From this solution a linear interpolation of the
displacements was used as the boundary condition for a 100 x 100 um model of an
axisymmetric cell embedded in the matrix. Cell models were constructed in the
superﬁcial, middle, and deep layers of the explant.

170

 

(b)

Figure 2: Matrix damage was always observed as small surface ﬁssures (a), extending
only into the superﬁcial layer (b).

171

 

Figure 3: From stained sections dead cells appeared red and live cells appeared green. In
these grayscale pictures, the dead cells are dark while the live cells appear bright. Cell
death was observed throughout the thickness of the cartilage, however, the superﬁcial
layer (S) showed the largest percentage of cell death, followed by the middle zone (M).
Minor degrees of cell death also appeared in the deep zone (D).

172

 

.h
o

 

 

I Low
DHigh

(O
QQQQ

Average % of dead cells
N (A) -> 0| 0) \1 cc
0. O. Q 0

_L
on

 

 

 

 

 

 

v7

Superficial Middle I Deep
Cartilage Layer

 

 

Figure 4: There was signiﬁcantly more cell death in the low versus high rate of loading
experiments (0 ). There was also signiﬁcantly more cell death in the middle zones for the

low compared to high rate of loading experiments (0 ). Minimal cell death occurred in the
deep zone for both rate of loading experiments.

173

 

 

 

""' I

 

 

 

LLLAAAHA‘AALL

 

 

LLAAK‘LALLAIA

(b)

 

h Str

(MPa)
2.30
1.85
1.55
1.25
0.95
0.65
0.35

DUE-II

 

 

 

Figure 5: There were no radial or depth-dependent distributions of maximum shear stress
in the isotropic non-layered explant model (A). On the other hand, the isotropic layered
model indicated depth-dependent variations in maximum shear stresses, with the largest
stresses appearing in the superﬁcial layer (B).

174

 

 

Center

 

 

Center

 

 

0.26
0.22
0.18
0.14
0.10

 

 

DUE-II

 

Figure 6: This ﬁgure shows the deformed explant. There were no radial or depth-
dependent variations in tensile strain for the isotopic non-layered model (A). In contrast,
the isotropic, layered model exhibited large tensile strains in the deep region and smaller
lateral expansion in the superﬁcial layer (B). The transversely isotropic, layered model
also indicated similar levels of average tensile strain in the superﬁcial and deep layers

(C).

175

CHAPTER 9

DETERMINATION OF IN SITU MATERIAL PROPERTIES OF CARTILAGE
BASED ON A NON-LINEAR VISCOELASTIC MODEL FOR INIPACT
LOADING

Benjamin James Ewers

ABSTRACT

A method was developed to determine the nonlinear isotropic viscoelastic
properties of articular cartilage from in situ indentation testing for modeling of articular
cartilage under impact loading scenarios. The inﬁnitesimal viscoelastic properties were
determined by ﬁtting force-time curves from a non-porous ﬂat-ended indentor
experiment for loads times less than one second. A commercial ﬁnite element code was
employed to solve the problem of a viscoelastic layer indented by a ﬂat-ended indentor.
The components of a ﬁrst order Prony series were then determined. The non-linear
isotropic material properties were estimated by ﬁtting force-displacement curves obtained
from large deformation indentation relaxation tests on cartilage using a non-porous
spherical indentor. The curve consisted of peak forces obtained at the initial stage of
' rapid indentation versus the associated displacements. A commercial ﬁnite element code
was employed to solve the problem of a hyperelastic layer indented by a rigid sphere. The
components of the strain energy function could then be determined. Near

incompressibility was assumed to model the rapid loading. The method was applied to

176

indentation data from bovine cartilage to document the non-linear viscoelastic material
properties for impact loading. The method was validated using experimental data from
unconﬁned compression of bovine cartilage.

INTRODUCTION

Osteoarthritis (01A) is a disease affecting diarthrodial joints and is characterized
by full thickness loss of articular cartilage resulting in bone-on-bone contact and joint
pain. Clinically, matrix damage of cartilage, such as surface ﬁssuring, and chondrocyte
death have been documented in osteoarthritic cartilage (Kim et al., 2000). Excessive and
abnormal mechanical loading of cartilage has been implicated in the pathogenesis of OA.
However the role excessive or abnormal mechanical loading plays in the mechanism of a
post-traumatic, secondary OA are not well understood.

Repo and Finlay (1977), using tissue explants on bone and subjected to a single
impact, document cell death localized around ﬁssures in the superﬁcial layer of the
cartilage. Torzilli et aL (1999) also document cell death and ﬁssuring of cartilage
following a single impact in explants without the attachment of underlying bone. In vitro
studies have also been used in attempts to correlate the state of stress and strain in loaded
cartilage with the locations and degree of matrix damage and cell death. Atkinson et al.
(1998) using impacts on isolated rabbit tibial plateaus suggested that maximum shear
stress is the best predictor of cartilage ﬁssuring. Large matrix strains corresponding to
large cell strains have been hypothesized to cause signiﬁcant leakage of ﬂorescent
indicators across cell membranes in preliminary studies (Guliak et al., 1995). This
suggests that membrane damage and the possibility of cell death may occur from

excessive cell strain (Guliak et al., 1995). Recently, Ewers et al. (2001) modeling an

177

explant under impact loading suggest that shear stress correlates with matrix damage and
large cell strains correlates with areas of extensive cell death.

However, it is difﬁcult to use the inﬁnitesimal material properties of cartilage for
modeling impact scenarios because of the large deformations. Atkinson et al. (1998),
using an isotropic elastic model for the cartilage to model an impact event, suggested that
the elastic modulus of cartilage needs to be approximately 20 times that determined in
small deformation experiments. This is in support of a previous study that documented
stiffening effects of cartilage due to high strain rates during impact experiments (Oloyede
et al., 1992). Therefore, there is a need to take into account the non-linear stiffening
effect of cartilage for modeling impact experiments.

Cartilage is composed of a ﬂuid and solid phase and as such, has been modeled as
a biphasic material with ﬂuid ﬂow being responsible for the time-dependent creep and
stress-relaxation behavior (Mow et al., 1980). However, the instantaneous response of
biphasic cartilage is equivalent to that of an incompressible elastic material (Armstrong et
al., 1984; Ateshian et al., 1994). In addition, the shear stresses in a cartilage layer under
impact loading are the same for an incompressible elastic and biphasic material at time
zero (Garcia et al., 1998). A previous study that examined contact analysis for biphasic
materials, shows no change in contact area for times out to one second alter applying a
load, suggesting little or no ﬂuid ﬂow (Donzelli et al., 1999). Under one second of
loading, however, there is extensive relaxation of cartilage from in situ indentation tests
(Newberry et al., 1988). Ewers et al. (2000) also supports this result and show that the
rate of loading under one second can affect the deformation and possibly the state of

stress in unconﬁned compression experiments on bovine cartilage. They found that a low

178

rate of loading (40 MPa/ 1 second) caused more deformation and cell death deeper in the
explant compared to a high rate of loading (40 MPa/ 50 milliseconds) experiment to the
same peak load. On the other hand, matrix damage was more extensive with the higher
rate of loading suggesting larger shear stresses were developed assuming high tissue
shear stresses are responsible for matrix damage. To fully model the apparent fast
relaxation characteristics of cartilage under impact loading, a model that incorporates a
viscoelastic solid phase may be needed (Suh and Bai, 1998).

The object of the current study was to model cartilage for impact loading as a
non-linear viscoelastic material and determine the material properties ﬁom in-situ
indentation testing. This will allow for improved modeling of cartilage under impact
loading and assist with understanding the mechanisms of matrix damage and cell death
due to excessive mechanical loading.

METHODS AND MATERIALS
W
The inﬁnitesimal shear modulus (G) was assumed to be viscoelastic and the

reduced relaxation function (gR(t)) was modeled using a Prony series:

 

GR") =1-§Gf<1—e%)

8120) =
Go i=1

As a ﬁrst approximation we assumed only one term:

gr(t)=-G—R-(Q=l-GP(1-e%)
Go

179

Go is the instantaneous shear modulus, so there are only two parameters to determine the

viscoelastic Prony series: GP and t, where GP corresponds to the amount of relaxation,

while 1 determines the shape of the relaxation curve.

In a hyperelastic formulation the Cauchy stress is:

2 BW

'1 J ' ackm 1’"
Where Fik is the deformation gradient, Ck“, is the right Cauchy-Green deformation tensor,
J is the determinant of the deformation gradient, and W is a strain-energy ﬁmction. The

appropriate stress-strain relationship in terms of derivatives of the strain-energy function

with respect to the strain invariants for an arbitrary isotropic compressible material is:

aw 2 aw aw 26W
—5,.j+— —+
at,

=2J _ ..___
J at, 1312 ’1 Jar2

0'1] BikBlg'
Where Bij is the left Cauchy-Green deformation tensor, 1), 12, and 13 are the strain
invariants of the left Cauchy-Green deformation tensor. We chose a strain energy

function that has been used to model the non-linear stiffening of large arteries (Simon et

al., 1998). This is a “generalized” Fung strain-energy function of the form:

W=§C0(e¢ —1)

¢ = C,(1‘1-3)+C2(i2 —3)+K(J— 1)2

180

The bar above the strain invariants indicates they are the deviatoric strain invariants:
" _ —2 / 3 — _ —4 / 3
ﬁ—J h Q—J 5

This strain-energy ﬁrnction can be incorporated into a commercial ﬁnite element code
using a user-deﬁned isotropic hyperelastic function, UHYPER (ABAQUS, version 5 .8;
Hibbit, Karlsson, & Sorensen, Inc.). This involves taking derivatives of the strain-energy
function with respect to 11,12, and J (A.l-A.2). This strain-energy function is stress free in
the undeformed conﬁguration. Therefore there are four independent co efficients. The two
inﬁnitesimal elastic constants (Go, v) were expressed in terms of the four coefﬁcients of
the strain energy function, by a theoretical pure shear test and a uniform triaxial test (B.l-
B.2). Since we were interested in impact loading the cartilage was assumed to be nearly
incompressible (v=0.49).

In a pilot study using bovine cartilage it was determined that the strain-energy
function stated above did not give a good ﬁt to the experimental data, The exponential
form had stiffened too rapidly for the experimental deformations. Therefore, the
exponential term in the strain-energy function was expanded and the higher order terms

dropped:

e¢=1+¢+é¢2=w

The strain energy function was put in the form:

1
W=—'C
2 0V

181

Using the expanded version of the strain energy function in UHYPER (A3) ﬂattened the
load-displacement curve at the higher strains giving a better ﬁt of the pilot experimental
data (Figure 1).

To fully model the cartilage as a non-linear viscoelastic material we need to
determine six material parameters: four coefﬁcients of the strain-energy function, and
two coefﬁcients in the Prony series.

Emma

Indentation relaxation experiments were performed on cartilage from a fresh
bovine metacarpal joint. The joint was opened and placed in phosphate buffered saline
(pH 7.2) and mounted in a ﬁxture with the cartilage surface facing an indentor probe that
was attached to a stepper micrometer (M-168.30, PI Physik Instrumente, Auburn, MA).

The following procedure was performed at each of four sites. First a plane-ended
non-porous cylindrical indentor of 1.0 mm diameter was pressed 0.1 mm into the
cartilage at an average velocity of 2.5 mm/s and maintained for 150 seconds. For two of
the sites, after a period of ﬁve minutes, a spherical non-porous indentor was pressed 0.05
mm into the cartilage at 2.5 mm/s and maintained for 150 seconds. This procedure for the
spherical indentor was subsequently repeated while increasing the indentation depth to
0.1, 0.15, 0.2, 0.225, and 0.25 mm with a wait period of ﬁve minutes between tests. For
the other two sites the above procedure with the spherical indentor was done in reverse
order: ﬁrst 0.25 m, then 0.225, 0.2, 0.15, 0.1, and ﬁnally the 0.05 mm indentation
depth. In all of the above tests the resistive loads were measured (Data Instruments,

Acton MA: model JP - 25, 25 lb capacity). ampliﬁed, and collected at 1000 Hz for the

182

ﬁrst second, and 20 Hz thereafter. Finally tissue thickness was measured at three
locations around each testing site by pressing a needle into the cartilage to determine
tissue thickness at that location. The thickness at the testing site was assumed to be an
average of the three measured thicknesses.

A spherical shape indentor was used to help avoid damage to the tissue at the
large deformations. However, to help determine if tissue damage was occurring at the
large deformations, a second indentation test was performed to examine the difference in
peak loads between repeated tests. Finally to examine the effect of repeated testing on a
single site the ﬁlth set of indentation experiments were performed with the spherical
indentor at various locations. First the spherical indentor was pressed into the cartilage
followed by a ﬁve-minute wait period, then the tissue thickness was determined at that
site. The indentor was moved to the next site and the second indentation depth
experiment was performed. This procedure was repeated for all indentation depths.

From the peak resistive load of the ﬂat-ended indentor experiments the
“instantaneous” shear modulus (G0) was determined based on an elastic analysis at the

four sites and averaged (Hayes et al., 1972):

P
8awrt‘

 

GO:

Where P is the peak load, a is the indentor radius (0.5 mm), w is the indentation depth
(0.1 mm) and K is a geometric factor depending on the indentor radius and thickness of
the elastic layer. The force-time plots of the four ﬂat-ended indentor experiments were
averaged and plotted over one second. The relaxation that occurred by one second was

due to the viscoelastic nature of the cartilage because we are assuming no ﬂuid ﬂow

183

during this time frame. A computational model was constructed for a linear viscoelastic
layer bonded to a rigid half space and indented by a rigid cylindrical ﬂat-ended probe.
The geometry consisted of a large, axisymmetric cylindrical disk of cartilage with a
thickness determined ﬁ'om the average experimental cartilage thickness. The bottom
nodes were ﬁxed in space while the nodes under the indentor were displaced 0.1 mm in
50 ms and maintained for 1 second The mesh consisted of 394 eight-noded quadrilateral
elements (CAX8). The inﬁnitesimal elastic modulus was based on the average
experimental Go and the assumption of v = 0.49. The two viscoelastic parameters (GP, 1)
were varied until the computational load data ﬁt the experimental relaxation data.

From the ﬁve experiments performed with the spherical indentor an average curve
of peak load versus normalized displacement (strain) was determined directly under the
center of the indentor. A computational model was constructed of a hyperelastic layer
bonded to a rigid half space indented by a rigid spherical probe. The geometry consisted
of a large, axisymmetric cylindrical disk of cartilage with a thickness determined from
the average experimental cartilage thickness. The bottom nodes were ﬁxed in space while
the spherical rigid indentor was displaced to 0.3 mm (50 % strain) while predicting the
resistive loads. The mesh consisted of 440 eight-noded quadrilateral elements (CAX8). In
order to ﬁt the experimental force-displacement curves, two coefﬁcients (Co and C1) were
adjusted in this model The other two coefﬁcients could then be found in terms of the
inﬁnitesimal elastic constants (Go, v) using equations (B-.1-B.2).

In order to validate this method of determining non-linear viscoelastic material
properties of bovine cartilage unconﬁned compression was modeled at two rates of

loading. A 6.0 mm diameter punch with a smooth edge was used to make 32 plugs from

184

the metacarpal surfaces of bovine forelegs. The plugs were separated from the underlying
bone using a scalpel. The specimens were assigned to either a high rate (50 milliseconds
to peak, 600 MPa/s, n=16) or low rate (1 second to peak, 30 MPa/s, n=l6) test group. The
thickness of each explant was measured using a digital micrometer (Digimatic, Mitutoyo
Corp., Japan). Each specimen was loaded to a peak load of 840 N (~ 30 MPa) using a
single haversine load-time pulse. The specimens were placed between two highly
polished stainless steel plates in an unconﬁned compression experiment using a servo-
hydraulic machine (Instron, model 1331, Canton, MA). Peak load, time to peak load, and
maximum displacement were recorded in each experiment. A computational model was
constructed assuming a hyper-viscoelastic layer. The geometry consisted of a large,
axisymmetric cylindrical disk of cartilage with a thickness determined from the average
experimental cartilage thickness. The material properties were those determined using the
above procedures. The bottom nodes were ﬁxed in the axial direction, but allowed to
move radially. The t0p nodes were displaced downward. For the high rate of loading
model the top nodes were displaced to the average displacement from the experiments in
50 ms, while the low rate of loading model had the average experimental displacements
applied in 1 second. The mesh consisted of 300 eight-noded quadrilateral elements
(CAX8). The theoretical load was then compared to the average experimental load for
each rate of loading experiment.
RESULTS

The average cartilage thickness was 0.60 i 0.05 mm for these experiments. The
elastic inﬁnitesimal analysis of the four indentation relaxation experiments with the ﬂat-

ended indentor resulted in G0 = 3.2 i 0.4 MPa. The load relaxed from 50 milliseconds to

185

1 second in these relaxation tests approximately 56 %. As a ﬁrst approximation GP = 0.5
and 1: was varied ﬁom 0.4 to 0.08 (Figure 2). As expected, as ‘C decreased the relaxation
of load was faster. However, for GP = 0.5 there was not enough relaxation at one second.
In the second attempt GP 5 0.6, and the computational model ﬁt the peak load and load at
one second (Figure 3). However, using a one term Prony series the relaxation curve was
not ﬁt well. Since we were interested in using these material properties for impact
loading, we were mostly interested in ﬁtting early relaxation part of the curve. Therefore
the best ﬁt for the early relaxation (< 100 milliseconds) was GP = 0.625 and 1: = 0.08.
Future studies will need to use a two term Prony series to better model the relaxation
behavior of cartilage for the entire one second of relaxation.

The experimental data from the spherical indentation tests indicate that there was
no difference in peak load between the ﬁrst and second tests at displacements between
0.15 - 0.25 mm (Figure 4). This indicated that the tissue was not damaged at 50 % strain
with the spherical indentor. The experimental data for the four sites of testing showed no
differences in the peak load versus strain curve whether the test sequence was varied
starting at the small or large displacement tests (Figure 5). Furthermore, when testing
each location once on the opposite plateau there was no difference in the peak load versus
strain curve compared to the repeated indentation sites.

When attempting to ﬁt the experimental peak load versus strain data with the
hyperelastic function, it was not ﬁtting the curve correctly at the smaller strains. It can be
shown by linearly extrapolating the ﬂat-ended indentation relaxation data backward that
there is an approximate 30 % reduction in load between time zero and 50 milliseconds

(Figure 6). This translated into a true instantaneous shear modulus of 4.2 MPa. When

186

using G0 = 4.2 MPa the hyperelastic energy-function could now ﬁt the experimental peak
load versus strain data reasonably well (Figure 7). I determined that the one set of strain-
energy coefﬁcients that best ﬁt the experimental data over the entire strain range was: Co
= 6.0, C1= 0.68, C; = 0.02, K = 35.25.

Analysis of the experimental unconﬁned compression tests on the bovine cartilage
explants indicated that the high rate of loading resulted in an average peak load of 803 i
42 N and an average peak strain of 0.46 i 0.06. On the other hand, the low rate of loading
experiments resulted in an average peak load of 857 i 36 N and an average peak strain
of 0.56 i 0.12. Using the average strains as input boundary conditions in the model along
with the four coefﬁcients of the strain-energy function determined above, a hyperelastic
model of unconﬁned compression did a reasonable job predicting the peak load for the
high rate of loading experiments, but drastically over estimated the peak load for the low
rate experiments (Figure 8). In contrast, when the cartilage was modeled as a hyper-
viscoelastic material, with the estimated material properties determined above the
predicted peak load was 670 and 706 N for the high and low rate of loading experiments,
respectively. While the hyper-viscoelastic model slightly under predicts the experimental
peak load for both rate of loading experiments (~17 %), the same trend is followed.
Experimentally, there was a 6.3 % decrease in average peak load for the high versus low
rate of loading experiments. The computational model predicted a decrease of 5.1 %. The
high rate of loading model, with less deformation compared to the low rate of loading
model, predicted a tissue shear stress of 5.65 MPa versus 5.0 MPa for the low rate of
loading explant model.

DISCUSSION

187

The object of the current study was to develop a non-linear viscoelastic model of
cartilage for impact loading, and determine material properties from in-situ indentation
tests. The current study indicated that all six material properties for a hyper-viscoelastic
model could be determined from in situ indentation testing. The estimated material
properties could then be used to predict responses of cartilage in unconﬁned compression
test performed at two rates of loading.

Other investigators have modeled bovine cartilage as a biphasic material with a
viscoelastic solid phase. Suh and DiSilvestro (1999) used a simpliﬁed relaxation function

with a discrete spectrum to model the viscoelastic solid phase and determined rs and “CL,

where Is is the short-term relaxation time and It, is the long time constant. For short times
it can be assumed that the relaxation of the Suh and DiSilvestro (1999) model is only due
to relaxation of the solid phase (no ﬂuid ﬂow), so that Is with I from the current study
can be compared. The previous study documented rs = 0.02 3 versus 1: = 0.08 s of the
current study.

In the current study, the unconﬁned compression of bovine cartilage explants
indicated, as expected, that there was more tissue deformation for the low versus high
rate of loading experiments. Excessive deformation has been suggested to cause cell
death by possibly damaging the cell membrane (Guliak et al., 1995). Interestingly, the
hyper-viscoelastic model in the current study found that while there is more deformation
in the low rate of loading experiments, the high rate of loading impacts would result in
increased shear stresses. These high shear stresses have been implicated in acute ﬁssuring
of articular cartilage due to impact loading (Newberry et al, 1998; Atkinson et al., 1998).

The hyper-viscoelastic modeling of the current study is in support of a previous study that

188

 

documented increased ﬁssuring of cartilage explants under unconﬁned compression at a
high versus low rate of loading (Ewers et al., 2001).

There were a number of limitations in the current study. First, the hyper-
viscoelastic material properties for cartilage determined using this method are only
appropriate for loading under one second, where there is no ﬂuid ﬂow, such as impact
scenarios. For loading times greater than this, there is the need to model the cartilage as a
biphasic material. A biphasic model has the advantage of being able to separate the solid
and ﬂuid stresses in the cartilage. This may provide important since solid tensile stresses
have been implicated in cartilage ﬁssuring (Garcia et al., 1998). The relaxation of the
cartilage could not be ﬁtted precisely over the entire one second with a single term in the
Prony series thus future studies will need to incorporate at least one additional term like
previous investigators (Suh and Bai, 1998). This study did not take into account the
known depth-dependent material properties of cartilage (Schingal et al., 1997). These
depth-dependent material properties may need to be taken into account to predicate
experimental deformations and matrix damage (Ewers et al., 2001). For instance, in the
current study the homogenous computational model predicts a constant shear stress, but
experimentally matrix damage is initiated in the superﬁcial layer. Likewise, the current
homogenous model predicts a constant lateral expansion, but experimentally the lateral
expansion of cartilage under unconﬁned compression is not constant. Future studies can
use the current method to determine the hyper-viscoelastic material properties in different
layers of cartilage. Finally, there are many sets of material properties that could estimate
the experimental response of cartilage under indentation testing. Therefore the material

properties determined with this method are not unique. There is a need to incorporate an

189

optimization scheme to get the material properties that “best” ﬁt the experimental data, so
other laboratories can predict the same material properties using these techniques.

In conclusion, the hyper-viscoelastic material properties of cartilage were
determined from in situ indentation testing and used to predict other impact loading
scenarios. Previous studies have used an elevated elastic modulus to model impact
scenarios to get appropriate displacements. Modeling the cartilage as hyperelastic allows
for experimental loads to be used as boundary conditions and the appropriate
displacements can be obtained. The use of a hyperelastic model is critical because the
state of tissue stress will vary between a model using an elevated elastic modulus and a
model incorporating a hyperelastic material for contact impact modeling. Since cartilage
has fast relaxation behavior that can not be accounted for by ﬂuid ﬂow, a viscoelastic '
model is imperative for modeling impact scenarios. The current study allows for the
determination of these viscoelastic prOperties, so that the effect of different rates of
loading on cartilage stress and strain can be investigated with computational models.
Therefore, this study will help with modeling impact experiments to be able to better
understand mechanisms behind mechanically induced matrix damage and cell death that
could possibly lead to a chronic degradative disease such as OA. Future studies will need
to examine the depth-dependent material properties of cartilage.

REFERENCES

1. Armstrong, C. G., Lai, W. M. and Mow, V. C., (1984) An Analysis of the Unconﬁned
Compression of Articular Cartilage. Journal of Biomechanical Engineering 106, pp.
165-173.

2. Ateshian, G. A., Lai, W. M., Zhu, W. B. and Mow, V. C., (1994) An Asymptotic

Solution for the Contact of Two Biphasic Cartilage Layers. Journal of Biomechanics
27, pp. 1347-1360.

190

 

10.

ll.

12.

13.

14.

Atkinson, TS, RC Haut, and NJ Altiero. (1998) Impact-induced ﬁssuring of articular
cartilage: an investigation of failure criteria. J Biomech Eng. 120:181-187.

Donzelli, PS, RL Spilker, GA Ateshian, and VC Mow. (1999) Contact analysis of
biphasic transversely isotropic cartilage layers and correlations with tissue failure. J
Biomech. 32:1037-1047.

Ewers, BJ, D Dvoracek-Driskna, MW Orth, and RC Haut. (2000) The extent of
matrix damage and chondrocyte death in mechanically traumatized articular cartilage
explants depends on rate of loading. J Orthop Res. in press.

Ewers BJ, Dvoracek-Driskna D, MW Orth MW, Haut RC (2001) Theoretical models
of cartilage explants and correlations with experimental matrix damage and cell death
after impacting loads. Annal of Biomedical Eng in review.

Garcia JJ, Altiero NJ, Haut RC (1998) An approach for the stress analysis of
transversely isotropic biphasic cartilage under impact load J Biomech Eng 120:608-
613.

. Guilak F, A Ratcliffe, and VC Mow. (1995) Chondrocyte deformation and local

tissue strain in articular cartilage: a confocal microscopy study. J Orthop Res, 13:410-
421.

Hayes, W. C., Keer, I. M., Herrmann, and Mockros, I. E., (1972) A Mathematical
Analysis for Indentation Tests of Articular Cartilage. Journal of Biomechanics 5, pp.
541-551.

Kim, HA, YJ Lee, SC Seong, KW Choe, and YW Song. (2000) Apoptotic
chondrocyte death in human osteoarthritis. J Rheumatol. 27(2):455-462.

Mow VC, Kuei SC, Lai WM, Armstrong CG (1980) Biphasic creep and stress
relaxation of articular cartilage in compression: Theory and experiments. J Biomech
Eng 102:73-84.

Newberry, W. N., Mackenzie, C. D., and Haut, R. C., (1998) Blunt Impact Causes
Changes in Bone and Cartilage in a Regularly Exercised Animal Model. Journal of
Orthopaedic Research 16, pp. 348-354.

Oloyede A, Flaachsmann R, Broom ND (1992) The dramatic inﬂuence of loading
velocity on the compressive response of cartilage. Connective Tissue Research
27:211-224.

Repo RU and JB Finlay. (1977) Survival of articular cartilage alter controlled impact.
J Bone Joint Surg [AM]. 59:1068-1078.

191

15. Schinagl, RM, D Gurskis, AC Chen, and RL Sah. (1997) Depth-dependent conﬁned
compression modulus of full-thickness bovine articular cartilage. J Orthop Res
15:499-506.

16. Suh, JK and S Bai. (1998) Finite element formulation of biphasic poroviscoelastic
model for articular cartilage. J Biomech Eng. 120: 195-201.

17. Suh JK, DiSilvestro MR (1999) Biphasic poroviscoelastic behavior of hydrated
biological soft tisue. J Applied Mech 66:528-535.

18. Torzilli, PA, R Grigiene, J Borrelli, Jr., and DL Helfet. (1999) Effect of impact load

on articular cartilage: cell metabolism and viabiltiy, and matrix water content. J
Biomech Eng. 121:433-441.

192

 

 

 

 

 

 

—Average
25 ‘ —Average - SD
A 20 _ ——-Average 4» SD
5 + Exponential
13 15 7
8 + Exponential - Expanded
.1
10 t
/
5 .4
o - --- ’ a . . 4 .
O 0.1 0.2 0.3 0.4 0.5

Straln

Figure 1: Experimental load-strain curves for indentation of a rigid sphere into bovine
articular cartilage (Average :1: one standard deviation). Gray lines show the computational
models using the same coefﬁcients of the strain-energy function. Notice that for strains
less than 30 % the models predict the same load thereafter the expanded form of the
strain-energy function ﬂattens out. The expanded form had a similar shape as the

experimental data.

 

4.5 "

 

‘ —Average - SD

 

—Average
—Average + SD

+G=0.0

+ G=0.5, 1 =0.4
+G=O.5. r =02
+ G=0.5, 1 =0.08

 

 

 

 

 

 

 

 

 

o r r 4 A

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Time (s)

‘1

Figure 2: First attempt to model the relaxation of bovine cartilage. Notice there was not

enough relaxation. Decreasing 1: caused the load to relax faster.

193

 

4,5 - -—Average '

  

 

 

 

 

 

 

 

 

 

 

 

 

4 4 —Average + SD
, —-Average - SD
3.5 - (:5 +G=o.6, r=o.1
3 r l 1“- +G=0.65,r=0.15
2 \ +G=o.625, r =0.08
v 2.5 - , . +G=0625 1:02
13 ‘ . , .
i 2- \__._. ..__-
\ '\ ‘ - __ . ﬂ - _...
1.5 - i ewe—n
1 1'
'1
0.5 .
0 ‘ I T T l l

0 0.1. 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Tlme(s)

Figure 3: Final curve ﬁtting of the relaxation experiments. With only a one term Prony
series it is hard to get a good ﬁt of the relaxation. Because impacts are generally under
100 ms, we took the best ﬁtting curve ﬁom that region (GP = 0.625, 1: = 0.08).

 

 

6 T f l T l

6 10 14 18 22 26
Test 1 (N)

Figure 4: Peak load for repeated indentation tests with the spherical indentor for
displacements greater than 0.15 mm. The line represents zero error. Notice no trend for
the second test to be above or below the line suggesting that there is no tissue damage
due to the ﬁrst indentation test.

194

 

25‘

 

 
 
  
   

 

 

 

 

 

+ Site 1 .
20 i + Site 4
+ Site 2
g 15 - +Site 3
'8 0 Site 5
3 10
5
O r
O 0.1 0.2 S . 0.3 0.4 0.5
train

Figure 5: The experimental peak load versus strain curves for all the testing sites. Sites 1
through 4 were all repeated measures. Sites 1 and 4 went from small to large
displacements, while sites 2 and 3 went from large to small displacements. Site 5 had
only one indentation test per location to examine the inﬂuence of repeated tests on the
load-displacement response of cartilage. Notice how site 5 falls within the experimental
data from the repeated sites indicating that it was reasonable to conduct repeated tests on
the cartilage using a spherical indentor.

 

 

 

 

 

O
- . ’
2 i "
1 J
o ‘1 r n T f 7
O 20 4O 60 80 100

Time (ms)

Figure 6: By linearly extrapolating to determine an instantaneous load, there is an
approximate 30 % reduction in load from 0 to 50 ms.

195

 

 

 

 

 

 

30 -
—Average
25 ' —Average - SD /
——Average + SD
20 - + bb1a /
2 + bb2a /‘_ I
" + bb3a //
v 15 - /
g + bb4a
_1
1 - , ,,
0 / /
5 T /
o '1 m if r i r 1
0 0.1 0.2 0.3 0.4 0.5
Strain

Figure 7: The average experimental peak load versus strain curve along with the
computational model with four possible sets of strain-energy coefﬁcients. This assumed
that the instantaneous shear modulus was 1.3*Go based on the amount of relaxation in 50
milliseconds. The one set that best ﬁt the experimental data over the entire strain range
was bb2a (Co = 6.0, C1= 0.68, C; = 0.02, K = 35.25).

 

1800 - DExperiments

1600 ‘ I Hyperelastic Model
2 1400
1200 ‘
1000 7
800 ‘
600 ‘
400 1
200 7
o 1
High Rate of Low Rate of

Loading Loading

  

l

 

 

l Hyper-viscoelastic Model

 

Peak Load (

 

 

 

 

 

 

Figure 8: Experimental peak load from the unconﬁned compression experiments and the
peak load prediction from the hyperelastic and hyper-viscoelastic models. Deformation
was the input boundary condition. Notice that the hyper-viscoelastic slightly under
predicted the peak experimental load.

196

APPENDIX A

A1 Variables to deﬁne for user-deﬁned isotropic hyperelastic subroutine
V ' l 'v

B11 - ﬁrst deviatoric strain invariant

B12 — second deviatoric train invariant

AJ - determinant

V ' 1 n o ﬁn

U — strain energy function

(111(1) = 3%;1 (113(1) = 33%7123 J
U11(2) = 3%].2 UI3(2) = 33%ijaJ
(111(3) = 3%, U1 3‘3) = 83%291181
U12(1)=32%i12 U13<4>=3’%,-lajz
(112(2) = 82%? U1 3(5) = a3%1’28J2
(112(3) = £3203]2 (”3(6) = 83%J 3

(112(4) = 82%71 372

(112(5) =32Ua- a,
1

(”2(6) = 32%1-213.’

197

A.2 UHYPER for the “generalized” exponential Fung strain-energy equation

*USER SUBROUTINE
SUBROUTINE UHYPER<BIl,BIZ,AJ,U,UIl,UIZ,UIB,TEMP,NOEL,CM)
IMPLICIT DOUBLE PRECISION (A-H,O-Z)
CHARACTER*8 CM
PARAMETER (ZERO=0.0D0)

DIMENSION UI1(3),U12(6),UI3(6)

C0=21.0

C1=0.l85

C2=0.005476

CK=9.65
PHI=C1*(BIl-3.0)+C2*(BIZ-3.0)+CK*(AJ-l)**2
U=0.S*C0*(EXP(PHI)-l.)

DUDPHI=0.5*C0*EXP(PHI)

DU2DPHIZ=DUDPHI

DU3DPHI3=DUDPHI

UIl(1)=DUDPHI*C1

UI1(2)=DUDPHI*C2

UIll3)=DUDPHI*2.*CK*(AJ-1)
UIZ(1)=DU2DPHI2*C1*C1

UI2(2)=D02DPHI2*C2*C2
UIZ(3)=DU2DPHIZ*2.*CK*(1.+2.*CK*(AJ-l.)**2)
UIZ(4)=DU2DPHIZ*C1*C2
UI2(S)=DU2DPHIZ*C1*2.*CK*(AJ-1.)
UIZ(6)=DU2DPHIZ*C2*2.*CK*(AJ-1.)
UI3(1)=DU3DPHI3*C1*C1*2.*CK*(AJ-1.)
UI3(2)=DU3DPHI3*C2*C2*2.*CK*(AJ-1.)
U13(3)=DU3DPRI3*C1*C2*2.*CK*(AJ-1.)
UI3(4)=DU3DPHI3*C1*2.*CK*(1.+2.*CK*(AJ-l.)**2)
UI3(5)=DU3DPHI3*C2*2.*CK*(1.+2.*CK*(AJ-l.)**2)
U13(6)=DU3DPHI3*CK**2*(AJ-l)*(12.+8.*CK*(AJ—l)**2)
RETURN

END

198

A.3 UHYPER for the expanded exponential strain-energy equation

*USER SUBROUTINE

SUBROUTINE UHYPER(BIl,BIZ,AJ,U,UIl,UIZ,UI3,TEMP,NOEL,CM)

IMPLICIT DOUBLE PRECISION (A-H,O-Z)

CHARACTER*8 CM

PARAMETER (ZERO=0.0D0)

DIMENSION UIl(3),UI2(6),UIB(6)

C0=6.0

C1=0.68

C2=0.02

CK=35.25

D1=(BIl-3.0)

D2=(BIZ-3.0)

D3=(AJ-1.0)

TEMP=1+C1*D1+C2*D2+CK*D3**2+0.5*C1**2*D1**2+0.5*C2**2*D2**2

l +0.5*CK**2*D3**4+C1*C2*D1*D2+C1*CK*D1*D3**2+C2*CK*D2*D3**2

U=0.5*C0*(TEMP)

UI1(1)=0.5*C0*C1*(1+C1*D1+C2*D2+CK*D3**2)

UIl(2)=0.5*C0*C2*(1+C2*D2+C1*D1+CK*D3**2)

UI1(3)=C0*CK*(D3+CK*D3**3+C1*D1*D3+C2*D2*D3)

UI2(1)=0.5*C0*C1**2

UI2(2)=0.5*C0*C2**2

U12(3)=C0*CK*(l+3*CK*D3**2+C1*D1+C2*D2)

UIZ(4)=0.5*C0*C1*C2

UIZ(5)=C0*C1*CK*D3

UIZ(6)=C0*C2*CK*D3

UI3(1)=0.0

UI3(2)=0.0

U13(3)=0.0

UI3(4)=C0*C1*CK

UI3(5)=C0*C2*CK

UI3(6)=6*C0*CK**2*D3

RETURN

END

199

APPENDIX B

Relationships between the inﬁnitesimal elastic constants (Go, v) and the coefﬁcients
of the strain-energy function.

The following two independent theoretical problems were used to relate the
inﬁnitesimal elastic constants to the coefﬁcients of the strain-energy function. For rapid
loading it was assumed that the cartilage was nearly incompressible (v = 0.49).

B.1 Pure shear. The inﬁnitesimal shear modulus is related to the derivatives of the strain

energy function by the following well known function

8W (W
at, 612

 

00:2

evaluated in the undeformed conﬁguration (Ir=3, 12:3, 13=l). Using the strain energy

function in the current study yields the equation:

Go = CO(C1+ C2)

B.2. Uniform triaxial loading assuming equal stretches.

0:01:02503 4:41:427-43

Then the strain invariants are I 1 = 32.2, 12 = 325’, 134.6, and the equation for the stresses:

0W 2 8W W
= -2i3— +42—
ar, lat, an

200

The derivative of the stress with respect to stretch evaluated in the undeformed

conﬁguration is equal to three times the bulk modulus (B):

do
— = 3B
“’4 i=1

Evaluated with the “generalized” Fung strain energy function gives:
B = C0(K—C1—2C2)

Where B is the bulk modulus and is related to the shear modulus Go and v by:

B = MOO”) = C0(K—C1—2C2)
3(1—2v)

 

201

 

 

ll