LIBRARY

Michigan State

University

 

 

PLACE iN RETURN BOX
to remove this checkout from your record.
TO AVOID FINE return on or before date due.

 

MTE DUE

AAA

MTE DUE

DATE DUE

 

"M I a 23;

 

091503:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1/” WM“

A STRESS-BASED DAMAGE CRITERION TO PREDICT
ARTICULAR JOINT INJURY FROM SUBFRACTURE INSULT
By

Patrick Joseph Atkinson

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

1998

ABSTRACT

A STRESS-BASED DAMAGE CRITERION TO PREDICT
ARTICULAR JOINT INJURY FROM SUBFRACTURE IN SULT

By

Patrick Joseph Atkinson

Subfracture trauma to articular joints is a routine occurrence in sports, falls, and
car accidents. In some cases, the trauma can produce microscopic fractures of bone and
cartilage. Approximately 25% of patients sustaining subfracture injuries will experience
chronic joint tissue degeneration within several years following the initial injury. Because
of the small size of the fractures, subfracture injuries are difficult to detect acutely and
patients are typically not treated. Unfortunately, there is no protocol for assessing the
injm'y risk associated with subfracture traumatic events. An established joint injury
criterion does exist, however, it is based on gross fracture of bone. The criterion is used
by the automotive industry and is based on impact experiments on human cadavers. The
criterion states that loads exceeding 10 kN will likely result in gross fracture of bone.
Unfortunately, this criterion does not address subfracture injuries. This has led to the
current investigation in which theoretical and experimental models were used to develop a
stress-based, subfracture injury criterion. The patellofemoral joint served as a
representative joint to develop the criterion. Numerous experimental and theoretical
investigations were conducted to determine the impact response of the knee to fracture
and subfracture level traumas. Impact loads were directed at the patella of 62 pair of
isolated human cadaver knees to simulate impact loading which can occur during realistic
trauma scenarios such as falls or car accidents. During such traumatic events, body
kinematics and the contacting impact surfaces can vary, thus, a range of impact interfaces
were used to impact the knee flexed at 60° , 90° , or 120°. Fracture level injuries typically
involved transverse and comminuted fractures of the patella. Subfracture impact

experiments at approximately 70% of the fracture load produced microscopic cartilage

fissures and occult subchondral bone microfractures. These injuries were documented via
histology and were observed for all flexion angles. It was also observed that higher impact
loads could be tolerated before fracture or subfracture level injuries were produced with
increased contact area over the knee. Increased contact area typically resulted from
experiments conducted with padded impact interfaces. A mathematical model was
developed to estimate the stresses in the bone and cartilage in an effort to understand the
injury mechanism of subfracture injuries. It was found that elevated shear stresses at the
cartilage surface were associated with impact induced fissures. Occult injuries at the
subchondral plate were associated with elevated shear and tensile stresses. The model
showed that increased contact area over the knee significantly reduced tensile stresses in
the subchondral plate. The degree to which the tensile stresses were reduced was
dependent on the magnitude of the impact load and large contact areas. The reduced
incidence of fracture and subfracture injuries was significantly associated with reductions
in the tensile stresses in the subchondral bone. This suggested that the load and contact
area data could be used as a gross estimation of the stresses in the bone. This concept was
expanded to develop an injury criterion based on peak impact load and the contact area on
the knee. This criterion is useful for predicting injury in the human cadaver knee,
however, experiments conducted by automotive companies typically utilize
anthropomorphic dummies for crash simulations. Because studies have suggested that the
load-area response of the dummy knee significantly differs fi'om the human knee, a
transformation protocol was developed to allow the load-area data from the cadaver
experiments to be transposed to the dummy knee. Thus, experimentally derived load-area
data points {Tom the dummy were compared to injurious load-area data points fiom the
cadaver. The load-area criterion can be applied in experimental or theoretical studies to
optimize the design of car instrument panels, sports padding, etc. Implementation of such
improvements may reduce the overall incidence of joint injuries and the associated

sequelae of chronic joint morbidity.

DEDICATION

I want to thank the Lord for the opportunities afforded me throughout my life. I want to
thank my beautiful wife; I have a deep love and respect for you and my successes are
rooted in your support. I want to thank my family: my parents, one of the greatest gifis
God has given me; my brothers and sisters for their loving encouragement.

iv

ACKNOWLEDGMENTS

I would like to acknowledge my major professor Dr. Roger Haut for taking a
chance on me five years ago. His mentoring, guidance, and encouragement have inspired
me throughout my time at Michigan State University.

I am also grateful to Dr. Nicholas Altiero for serving on my committee and playing
an important role in developing my understanding and appreciation of engineering.

I would like to acknowledge Dr. Dahsin Liu and Dr. Charles DeCamp for their
time and input through serving on my committee.

I would like to acknowledge Dr. Robert Soutas-Little for his time and energy in
helping with the analysis of inertial and stress wave effects presented in the concluding
chapter.

Finally, I would like to express my graditude to the many collaborators I have had
the pleasm'e of working with on my research. This includes faculty members at Michigan
State University, faculty fiom other universities, members of industry and many of my

fellow graduate students.

TABLE OF CONTENTS

LIST OF TABLES .......................................................................................................... x
LIST OF FIGURES ....................................................................................................... xii
CITATIONS OF PUBLISHED WORK CONDUCTED DURING GRADUATE
SCHOOL: DISSERTATION RESEARCH .................................................................... xv
CITAT IONS OF PUBLISHED WORK CONDUCTED DURING GRADUATE
SCHOOL. ADDITIONAL PUBLISHED RESEARCH ................................................. xvi
INTRODUCTION .......................................................................................................... 1
CHAPTER 1
DETECTION OF EXPERIMENTALLY PRODUCED OCCULT MICROCRACKS AT
THE BONE-CARTILAGE INTERFACE ........................................................................ 9
1.1 Abstract ..................................................................................................................... 9
1.2 Introduction ............................................................................................................. 10
1.3 Methods and Materials ............................................................................................. 1 1
1.3.1 Tissues ........................................................................................................... 11
1.3.2 Processing and decalcification .......................................................................... l 1
1.3.3 Slab-cut Method .............................................................................................. 12
1.3.4 Paraffin Tape-Transfer Method ........................................................................ 12
1.3.5 Paraffin Method using Methyl Salicylate .......................................................... 12
1.4 Results ..................................................................................................................... 13
1.5 Discussion ............................................................................................................... 14
1.6 References ............................................................................................................... 16
1.7 Figures .................................................................................................................... 20
CHAPTER 2
A METHOD TO INCREASE THE SENSITIVE RANGE OF PRESSURE SENSITIVE
FILM ............................................................................................................................ 24
2.1 Abstract ................................................................................................................... 24
2.2 Introduction ............................................................................................................. 25
2.3 Methods and Materials ............................................................................................. 26
2.3.1 Experimental Protocol ..................................................................................... 26
2.3.2 Theoretical Validation ...................................................................................... 28
2.4 Results ..................................................................................................................... 29
2.5 Discussion ............................................................................................................... 30
2.6 References ............................................................................................................... 32
2.7 Tables ...................................................................................................................... 33
2.8 Figures .................................................................................................................... 34

CHAPTER 3
A METHOD FOR DETERMINING REGIONS OF DIARTHRODIAL JOINT

CONTACT DURING DYNAMIC LOADING OF THE HUMAN KNEE .................... 36
3.1 Abstract ................................................................................................................... 36‘
3.2 Introduction ............................................................................................................. 36
3.3 Methods and Materials ............................................................................................. 37
3.4 Results ..................................................................................................................... 38
3.5 Discussion ............................................................................................................... 38
3.6 References ............................................................................................................... 39
3.7 Tables ...................................................................................................................... 40
3.8 Figures .................................................................................................................... 40
CHAPTER 4
SUBFRACTURE INSULT TO THE HUMAN CADAVER PATELLOFEMORAL
JOINT PRODUCES OCCULT INJURY ....................................................................... 42
4.1 Abstract ................................................................................................................... 42
4.2 Introduction ............................................................................................................. 43
4.3 Methods and Materials ............................................................................................. 45
4.4 Results ..................................................................................................................... 49
4.5 Discussion ............................................................................................................... 52
4.6 References ............................................................................................................... 58
4.7 Tables ...................................................................................................................... 61
4.8 Figures .................................................................................................................... 63
CHAPTER 5
PATELLOFEMORAL JOINT FRACTURE LOAD PREDICTION USING PHYSICAL
AND PATHOLOGICAL PARAMETERS ..................................................................... 67
5.1 Abstract ................................................................................................................... 67
5.2 Introduction ............................................................................................................. 68
5.3 Methods and Materials ............................................................................................. 69
5.3.1 Statistical Regression Analyses ........................................................................ 69
5.3.2 Parametric Study of Load and Contact Area .................................................... 71
5.4 Results ..................................................................................................................... 73
5.4.1 Regression Statistics ........................................................................................ 74
5.4.2 Parametric Study Findings ............................................................................... 74
5.5 Discussion ............................................................................................................... 75
5.6 References ............................................................................................................... 78
5.7 Tables ...................................................................................................................... 80
5.8 Figures .................................................................................................................... 81
CHAPTER 6

THE INFLUENCE OF IMPACT INTERFACE ON HUMAN KNEE INJURY:
IMPLICATIONS FOR INSTRUMENT PANEL DESIGN AND THE LOWER
EXT REMITY INJURY CRITERION ........................................................................... 84
6.1 Abstract ................................................................................................................... 84

vii

6.2 Introduction ............................................................................................................. 85

6.3 Methods and Materials ............................................................................................. 92
6.4 Results ..................................................................................................................... 99
6.5 Discussion ............................................................................................................. 102
6.6 References ............................................................................................................. 108
6.7 Tables .................................................................................................................... 1 14
6.8 Figures .................................................................................................................. 116
CHAPTER 7‘
HUMAN KNEE IMPACT INJURY THRESHOLDS AND LOCATION VARY WITH
FLEXION ANGLE AND IMPACT INTERFACE ...................................................... 124
7.1 Abstract ................................................................................................................. 124
7.2 Introduction ........................................................................................................... 125
7.3 Methods and Materials ........................................................................................... 130
7.3.1 Experimental Protocol ................................................................................... 130
7.3.2 Theoretical Modeling - Impact Interface Study .............................................. 133
7.3.3 Histology and Injury Determination 135
7.3.4 Statistics ........................................................................................................ 135
7.4 Results ................................................................................................................... 136
7.4.1 Subfracture Experiments ................................................................................ 136
7.4.2 Impact Interface Experiments ......................................................................... 139
7.5 Discussion ............................................................................................................. 142
7.4.1 Subfi'acture Experiments ................................................................................ 142
7.4.2 Impact Interface Experiments ......................................................................... 147
7.6 References ............................................................................................................. 150
7.7 Tables .................................................................................................................... 153
7.8 Figures .................................................................................................................. 157
CHAPTER 8
A PROPOSED MODEL TO TRANSFORM CADAVER KNEE INJURY CRITERIA
TO THE HYBRID III DUMMY ................................................................................ 172
8.1 Abstract ................................................................................................................. 172
8.2 Introduction ........................................................................................................... 173
8.3 Methods and Materials ........................................................................................... 176
8.3.1 Isolated Human and Dummy Knee Experimental Protocol .............................. 176
8.3.2 Transformation Protocol ................................................................................ 180
8.3.3 Anthropomorphic Dummy Experimental Protocol .......................................... 181
8.4 Results ................................................................................................................... 182
8.4.1 Isolated Knee Results .................................................................................... 182
8.4.2 Anthropomorphic Dummy Experimental Results ............................................ 183
8.5 Discussion ............................................................................................................. 183
8.6 References ............................................................................................................. 188
8.7 Tables .................................................................................................................... 191
8.8 Figures .................................................................................................................. 193

viii

CHAPTER 9
STUDY OVERVIEW AND ADDITIONAL CONSIDERATIONS FOR

COMPREHENSIVE JOINT INJURY PREVENTION ................................................ 205
9.1 Overview ............................................................................................................... 205
9.2 The Types of Subfracture Injury and their Probable Mechanism ............................. 206
9.3 Injury Prevention Strategies ................................................................................... 207
9.4 Appropriate Injury Thresholds I ............................................................................. 209

9.4.1 Inertial Effects ............................................................................................... 209

9.4.2 Stress Waves ................................................................................................. 212
9.5 Appropriate Injury Thresholds II ............................................................................ 214
9.6 The Use of Theoretical Models to Design Injury Prevention Stategies .................... 214
9.7 References ............................................................................................................. 215
9.8 Tables .................................................................................................................... 216
9.9 Figures .................................................................................................................. 217

ix

 

LIST OF TABLES

CHAPTER 2

Table l .......................................................................................................................... 33
Table 2: ......................................................................................................................... 33
CHAPTER 3

Table 1 .......................................................................................................................... 40
CHAPTER 4

Table 1 .......................................................................................................................... 61
Table 2 .......................................................................................................................... 62
CHAPTER 5

Table 1 .......................................................................................................................... 80
Table 2 .......................................................................................................................... 81
CHAPTER 6

Table 1 ........................................................................................................................ 114
Table 2 ........................................................................................................................ 1 14
Table 3 ........................................................................................................................ 1 14
Table 4 ........................................................................................................................ 1 15
Table 5 ........................................................................................................................ 115
Table 6 ........................................................................................................................ l 15
Table 7 ........................................................................................................................ 1 15
CHAPTER 7

Table l ........................................................................................................................ 153
Table 2 ........................................................................................................................ 154
Table 3 ........................................................................................................................ 154
Table 4 ........................................................................................................................ 154
Table 5 ........................................................................................................................ 155
Table 6 ........................................................................................................................ 155
Table 7 ........................................................................................................................ 156
Table 8 ........................................................................................................................ 156
Table 9 ........................................................................................................................ 157
Table 10 ...................................................................................................................... 15 7
CHAPTER 8

Table 1 ........................................................................................................................ 191
Table 2 ........................................................................................................................ 191
Table 3 ........................................................................................................................ 191
Table 4 ........................................................................................................................ 192

Table 5 ........................................................................................................................ 192

Table 6 ........................................................................................................................ 192
Table 7 ........................................................................................................................ 192
CHAPTER 9

Table l ........................................................................................................................ 216
Table 2 ........................................................................................................................ 216

xi

LIST OF FIGURES

CHAPTER 1

Figure 1 ......................................................................................................................... 20
Figure 2 ......................................................................................................................... 23
CHAPTER 2

Figure 1 ......................................................................................................................... 34
Figure 2 ......................................................................................................................... 35
CHAPTER 3

Figure 1 ......................................................................................................................... 40
Figure 2 ......................................................................................................................... 41
Figure 3 ......................................................................................................................... 41
CHAPTER 4

Figure 1 ......................................................................................................................... 63
Figure 2 ......................................................................................................................... 64
Figure 3 ......................................................................................................................... 65
Figure 4 ......................................................................................................................... 65
Figure 5 ......................................................................................................................... 66
Figure 6 ......................................................................................................................... 66
CHAPTER 5

Figure l ......................................................................................................................... 82
Figure 2 ......................................................................................................................... 82
Figure 3 ......................................................................................................................... 83
Figure 4 ......................................................................................................................... 83
CHAPTER 6

Figure 1 ....................................................................................................................... 1 16
Figure 2 ....................................................................................................................... 1 16
Figure 3 ....................................................................................................................... 117
Figure 4 ....................................................................................................................... 1 17
Figure 5 ....................................................................................................................... 1 18
Figure 6 ....................................................................................................................... 1 18
Figure 7 ....................................................................................................................... 118
Figure 8 ....................................................................................................................... 119
Figure 9 ....................................................................................................................... 1 19
Figure 10 ..................................................................................................................... 120
Figure 11 ..................................................................................................................... 120
Figure 12 ..................................................................................................................... 121
Figure 13 ..................................................................................................................... 121

xii

Figure 14 ..................................................................................................................... 122

Figure 15 ..................................................................................................................... 122
Figure 16 ..................................................................................................................... 123
Figure 17 ..................................................................................................................... 123
CHAPTER 7

Figure l ....................................................................................................................... 157
Figure 2 ....................................................................................................................... 158
Figure 3 ....................................................................................................................... 159
Figure 4 ....................................................................................................................... 160
Figure 5 ....................................................................................................................... 161
Figure 6 ....................................................................................................................... 161
Figure 7 ....................................................................................................................... 162
Figure 8 ....................................................................................................................... 162
Figure 9 ....................................................................................................................... 163
Figure 10 ..................................................................................................................... 164
Figure 11 ..................................................................................................................... 164
Figure 12 ..................................................................................................................... 165
Figure 13 ..................................................................................................................... 166
Figure 14 ..................................................................................................................... 167
Figure 15 ..................................................................................................................... 168
Figure 16 ..................................................................................................................... 169
Figure 17 ..................................................................................................................... 170
Figure 18 ..................................................................................................................... 170
Figure 19 ..................................................................................................................... 171
CHAPTER 8

Figure 1 ....................................................................................................................... 193
Figure 2 ....................................................................................................................... 193
Figure 3 ....................................................................................................................... 194
Figure 4 ....................................................................................................................... 194
Figure 5 ....................................................................................................................... 195
Figure 6 ....................................................................................................................... 195
Figure 7 ....................................................................................................................... 196
Figure 8 ....................................................................................................................... 198
Figure 9 ....................................................................................................................... 198
Figure 10 ..................................................................................................................... 199
Figure l l ..................................................................................................................... 199
Figure 12 ..................................................................................................................... 200
Figure 13 ..................................................................................................................... 200
Figure 14 ..................................................................................................................... 201
Figure 15 ..................................................................................................................... 201
Figure 16 ..................................................................................................................... 202
Figure 17 ..................................................................................................................... 202
Figure 18 ..................................................................................................................... 203

xiii

Figure 19 ..................................................................................................................... 203

Figure 20 ..................................................................................................................... 204
CHAPTER 9

Figure 1 ....................................................................................................................... 217
Figure 2 ....................................................................................................................... 218
Figure 3 ....................................................................................................................... 219
Figure 4 ....................................................................................................................... 220
Figure 5 ....................................................................................................................... 220
Figure 6 ....................................................................................................................... 221
Figure 7 ....................................................................................................................... 221
Figure 8 ....................................................................................................................... 222
Figure 9 ....................................................................................................................... 222
Figure 10 ..................................................................................................................... 223
Figure 1 l ..................................................................................................................... 223
Figure 12 ..................................................................................................................... 224
Figure 13 ..................................................................................................................... 224
Figure 14 ..................................................................................................................... 225
Figure 15 ..................................................................................................................... 226
Figure 16 ..................................................................................................................... 227
Figme 17 ..................................................................................................................... 227

xiv

CITATIONS OF PUBLISHED WORK CONDUCTED DURING GRADUATE
SCHOOL: DISSERTATION RESEARCH

Each chapter in this dissertation represents a complete paper. Following are the citations
for those papers.

CHAPTER 1
Atkinson PJ, Walsh JA, Haut RC: The human patella: a comparison of three preparation

methods. The Journal of Histotechnology In press.

CHAPTER 2
Atkinson PJ, Newberry WN, Atkinson TS, Haut RC: A method to increase the sensitive
range of pressure sensitive film. Journal of Biomechanics. In press

CHAPTER 3

Atkinson PJ, Haut RC: A method for determining regions of diarthrodial joint contact
during dynamic loading of the human knee. Transactions of the Orthopaedic Research
Society: 656, San Francisco, CA, February, 1997.

CHAPTER 4
Atkinson PJ, Haut RC: Subfracture insult to the human cadaver patellofemoral joint
produces occult Injury. Journal of Orthopaedic Research 13:936-944, 1995.

CHAPTER 5

Atkinson PJ, MacKenzie CM, Haut RC: Joint fracture load prediction using geometrical,
physical and pathological measures: the human knee. SAE International Congress: paper
980358, 1998.

CHAPTER 6

Atkinson PJ, Garcia JJ, Altiero NJ Haut RC: The influence of impact interface on human
knee injury: implications for instrument panel design and the lower extremity injury
criterion. 4lst Stapp Car Crash Conference, SAE: 167-180. paper 973327, 1997.

CHAPTER 7
Patrick Atkinson, Jose Garcia, Hasnin Ali', Nicholas Altiero', Roger Haut: Human Knee
Impact Injury Thresholds And Location Vary With Flexion Angle And Impact Interface.

8th Injm Prevention Through Biomechanics Symposium: 8:127-143, 1998.

CHAPTER 8
Patrick Atkinson, Nicholas Altiero, Roger Haut Chris Eusebi, Vivek Maripudi, Tim Hill,
Kiran Sambatur: A Proposed Model To Transform Cadaver Knee Injury Criteria To The

Hybrid Iii Dummy. 8th Injury Prevention Through Biomechanics Symposium: 8:157-172,
1998.

 

XV

CITATIONS OF PUBLISHED WORK CONDUCTED DURING GRADUATE
SCHOOL: ADDITIONAL PUBLISHED RESEARCH

Atkinson PJ, Oyen-Tiesma M, Decamp CE, Mackenzie CD, Haut RC: Mechanical
augmentation of the patellar tendon afier removal of its central third helps limit changes in
joint tissue. Journal of Orthopaedic Research Conditionally accepted.

 

Atkinson PJ, Atkinson TS, Lancaster RL, Amoczky SP, Haut RC, Weisbrode SE: In-Vivo
Breaking Strength Retention And Histological Effects of 1.3 mm PDS ORTHOSORB®
Absorbable Pins as Seen in Rabbits. Journal of Foot and Ankle Surgery 37(1):42-47, 1998

Patrick Atkinson, Nicholas Altiero, Roger Haut Chris Eusebi, Vivek Maripudi, Tim Hill,
Kiran Sambatur: Development of an injury criteria for human surrogates to address
current trends in knee-to-instrument panel injuries. 42nd Stapp Car Crfih Conference.
SAE: paper 98S-25, 1998.

Atkinson PJ, Haut RC: Insult to the human cadaver patellofemoral joint: effects of age on
fi'acture tolerance and occult injury. 39th Stapp Car Cra_s_h Conference, SAE: 281-294.
paper 952729, 1995.

Atkinson, PJ, Garcia JJ, Altiero NJ, Haut RC:Animal and human studies on injury
mechanisms during blunt insult to the knee. 7th In' Prevention Throu Biomechanics
Smpgsium, 1997, 7:53-64.

Atkinson, P, Newberry, W, Staton, T, Garcia, J, Altiero, N, Haut, R:Anima1 and human
studies on injury mechanisms during blunt insult to the knee. 6th Injury Prevention
Through Biomechanics Smppsium, 1996, 6:137-160.

Newberry, W, Atkinson, P, Li X, Decamp, C, Altiero, N, Haut, R: Tissue injuries resulting

from blunt impact on the knee. 5th Injgy Prevention Through Biomechanics Syr_np_osiurn,
1995,5:21-39

Haut R, Atkinson P, Garcia J, Altiero N: Impact model for the human patellofemoral joint.
1998 World Congress on Biomechanics. In press.

Haut R, Atkinson P, DeCamp C, MacKenzie D: Augmentation influences joint remodeling
afier removal of the central third of the patellar tendon. 1998 World Congress on
Biomechanics. In press.

Garcia JJ, Newberry WN, Atkinson PJ, Haut RC, Altiero NJ: Comparison of stresses in
human and rabbit knees as they pertain to injuries observed during impact experiments.
Advances in Bioengineering ASME:3 15-316, Sun River, OR, June, 1997.

xvi

Atkinson PJ, Decamp C, Kamps B, Hespenheide B, Zukosky D, Haut R: Healing response
of the patellar tendon afier removal of its central third and implantation of an
augmentation device. Transactions of the Orthopaedic Research Society : 751, Atlanta,
GA February, 1996.

 

Newberry, W, Shelp, D, Atkinson, P, Haut, R: The effect of temperature on the
mechanical response of intact and scarified articular cartilage. Advances in Bioengineering
ASME:73-74, Chicago, IL November, 1994.

xvii

INTRODUCTION

Knee injury is a considerable problem which currently represents a large economic
and societal cost. The knee joint is commonly injured in domestic (States 1970), sports
and car accidents. Studies of experienced and amateur football players reveal it is the
most injured body region for that sport. For the years 1979-1995, 10% of all injuries
associated with car accidents were to the knee. Only the head and face are injured more
frequently (Huelke 1991). The knee is commonly injured in car accidents (Huelke 1991)
during high speed contact with the instrument panel which can produce subfracture and
fiacture injuries (Crandall 1995, States 1970) leading to a significant financial burden for
acute care. In addition to acute injuries, knee trauma can lead to chronic joint problems
such as post-traumatic osteoarthritis which add additional costs and disability (MacKenzie
1988). Osteoarthritis (CA) is documented to be the most prevalent joint disease in the
world and has been strongly correlated to injm'y (Davis 1989). For example, independent
clinical studies (Nagel 1976, Volpin 1990, Chapchal 1978) document a 25% or greater
chance of developing OA after gross fracture of the knee as early as one year post-insult
(Chapchal 1978).

A well established criterion is currently used by the automotive industry to prevent
gross bone fi'acture in the human knee (Viano 1977). The criterion states that loads
exceeding 10 kN will likely result in gross fracture of bone. This criterion is used to
certify new car models in which crash simulations are conducted using anthropomorphic
dummies as human surrogates. However, long-term complications have also been

associated with subfracture joint trauma. OA has also been associated with subfracture,

repetitive loading such as occurs during sporting and occupation related activities. For
example, mail carriers, farmers, construction workers, and firefighters are significantly
predisposed to CA in the absence of fracture. It is hypothesized that the chronic joint
problems are a consequence of repetitive microfracture of cartilage and bone. This is
confirmed by Radin and coworkers who cyclically loaded the rabbit tibiofemoral joint
producing occult microfi’actures at the bone-cartilage interface which were associated with
the development of OA. Unfortunately, the small size and typically occult nature of occult
injuries precludes there acute detection and treatment (T omatsu 1992).

Chronic joint morbidity has also been documented following a single, subfracture
joint trauma (States 1970) which causes acute microfracture of cartilage and underlying
bone. In as early as 4 years, subfracture traumatic dislocations of the hip lead to CA in
24% of cases (Upadhyay 1983). In experimental studies, a single transarticular blow to
the knee at 45% of the energy required to cause fi'acture produces occult vertical clefis at
the bone—cartilage interface which have been associated with ensuing degenerative changes
(Donohue 1983, Thompson 1991). Clinical and experimental studies of human (Atkinson
1998) and animal (Newberry 1996, Armstrong 1985, Tomatsu 1992) joint tissues show
that the development of cartilage fissures and occult cracks are associated with chronic
pain and joint degeneration (Johnson-Nurse 1985, Newberry 1996). Occult bone
microfracture can appear in vivo as a ‘bone bruise’ on magnetic resonance images and is
associated with superficial degeneration of the overlying cartilage within 6 months (V ellet

1991).

It is hypothesized that the superficial and occult subfracture injuries noted above
are associated with elevated shear or tensile stresses (Atkinson 1998, Ateshian 1993,
Eberhardt 1991, Thompson 1991). In general, the magnitude and location of these stress
maxima are dependent, in part, on the radii of the contacting articular surfaces. For
example, contact between disparate radii results in elevated shear stresses at the surface
while contact between similar, larger radii result in elevated shear stresses at the bone-
cartilage interface (Atkinson 1998, Eberhardt 1991). The relative contact radii in the
patellofemoral joint vary with flexion, as evidenced by studies of patellofemoral contact
areas associated with passive joint loading. The largest contact areas are documented at
90° of joint flexion with reduced contact areas documented at at flexion angles of 120°
and 60° (Huberti 1984). This suggests that trauma at extreme flexion angles may produce
more superficial lesions due to the disparate radii which lead to the smaller areas of
contact. Altemately, increased intraarticular contact area reflects larger contact radii
suggesting deeper lesions would be more frequent at the bone-cartilage interface. The
flexion angle at which trauma occurs in car accidents, for example, can range from 70° to
120° at the point of impact with the instrument panel. The relatively large range of flexion
angles is attributed to the anthropometry of the individual (Dischinger 1994), the initial
distance between the knee and instrument panel and other factors associated with the car
interior. Injuries incurred during sports and domestic accidents can also occur at varied
flexion angles due to similar reasons. In addition to the influence of flexion angle, past
studies suggest that the composition of the impact interface will play an important role in

knee injury. Studies conducted with rigid impact interfaces and padded interfaces of

variable thickness and composition show a 65% reduction in patellar fractures with
padding (Melvin 1975, Powell 1975, Patrick 1965). Unfortunately, an investigation of
subfracture injuries was not conducted. While not measured, the padded interfaces
probably produced greater contact areas over the knee than the rigid interface
experiments. This shows that increased contact area can prevent fracture level injuries,
while it is unknown if it will prevent subfracture injuries.

In the automotive industry, assessments of knee joint injury is predicted via
instrumented anthropomorphic dummies. If impact load and contact area are important
parameters in predicting injury in the cadaver, then this concept would need to be applied
to the dummy knee in the crash environment. However, previous studies suggest that the
impact load and contact area response of the dummy knee is significantly different than the
cadaver knee (Hering 1977). Thus, a transformation of injurious load and area data in the
cadaver will be required for use with the dummy.

The aims of the current study were to investigate the response of the human knee
to subfi'acture transarticular loading with different interfaces at different flexion angles. It
was hypothesized that the flexion angle of the knee and the stiffness of the impact
interface would play an important role in the incidence of impact-induced subfracture
injuries of the knee. Specifically, the study hypotheses were: 1) extreme flexion angles
will be associated with more superficial lesions of articular cartilage while median flexion
angles will yield deeper, occult injuries; 2) increased contact over the knee will increase

the magnitude of the impact load that can be tolerated before injury occurs; 3) parameters

which are determined to be associated with injury in the cadaver can be transformed for
use with the dummy knee.

Below is an overview of the chapters contained in this dissertation:

Chapter 1 presents a comparison of three histological methods which can be used to
reliably prepare slides for microscopic analysis of subfiacture injuries. This study shows
that superior results can be achieved without preparation artifact via simple, alternative
processing methods.

Chapter 2 presents a new way to use different ranges of pressure sensitive film in
recording extraarticular and intraarticular contact areas and contact pressures associated
with the impact event. This method increases the sensitivity and accuracy of contact area
and contact pressure data.

Chapter 3 outlines a method to identify regions of diarthrodial joint contact on the
articular surfaces. This is particularly helpful in determining the correct placement of
boundary conditions for mathematical models.

Chapter 4 is the first known subfracture injury study conducted on human joints. Human
cadaver knees were impacted at 45% of the energy required to produce fracture with a
rigid irnpactor and the knee flexed 90°. Occult, horizontal microfractures were
documented at the bone-cartilage interface in four of seven subjects.

Chapter 5 examines which gross parameters describing test cadaver subjects (age, height,
mass, sex, joint pathology, joint geometry, contact area over the knee) might influence the
impact response of the knee. It was found that joint geometry and contact area

significantly affected the impact loads and bone and cartilage stresses.

Chapter 6 followed from the findings in the preceding chapter. Paired experiments were
conducted on cadaver knees in which consistent impact loads were delivered with the load
distributed over significantly different contact areas. Small contact areas produced gross
and microscopic injuries while large contact areas prevented injury.

Chapter 7 studied the impact response of the knee by conducting experiments with
different interfaces and with the knee flexed at a. range of flexion angles. It was found that
rigid interfaces produced small contact areas and a high incidence of injuries at all flexion
angles. These injuries were associated with elevated tensile and shear stresses in the
subchondral plate. Padding significantly reduced these stresses and the associated
incidence of injury.

Chapter 8 takes the findings fiom the cadaveric studies and applies them to the
anthropomorphic dummy commonly used as a human surrogate in car crash simulations.
Thus, the experimental findings of the current dissertation are applied in a pragmatic to
estimate the injury potential of a traumatic event. This will allow the design of
countermeasures which could prevent knee injuries.

Chapter 9 is a concluding chapter which provides an overview of the findings in the
dissertation such as injury classifications and probable mechanisms as well as potential
injury countermeasures. In addition, some analyses are presented for the first time in this
chapter: 1) inertial effects and stress waves are studied with regard to their injury
potential in the knee, 2) injury risks based on regression analyses of the experimental data,

and 3) modeling of the anthropomorphic dummy in simulated car crashes.

REFERENCES

l.

10.

11.

12.

Armstrong C, Mow V, Wirth C (1985) Biomechanics of impact-induced
microdamage to articular cartilage: a possible genesis for chondromalacia patella. In:
Amer Acad of Orth Surg Symp on Sports Med, ed by G Fenerman. pp70—84

Ateshian GA, Lai WM, Zhu WB, Mow VC (1994) An asymptotic solution for the
contact of two biphasic cartilage layers. J Biomechanics 27(1 1): 1347-1360

Atkinson, T.S., Haut, RC, and Altiero, NJ. (1998) An Investigation of Biphasic
Failure Criteria for Impact Induced Fissuring of Articular Cartilage. J of
Biomechanical Engineering (In Press)

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

Crandall J, Kuhlrnann T, Pilkey W (1995) Air and knee bolster restraint system:
laboratory sled tests with human cadavers and the Hybrid III dummy. J Trauma 38:
517-520

Davis M, Ettinger W, Neuhaus J, Cho S, Hauck W (1989) The Association of Knee
Injury and Obestiy with Unilateral and Bilateral Osteoarthritis of the Knee. American
Journal of Epidemiology 130(2): 278-288

Dischinger PC, Kems TJ, Kufera JA (1995) Lower extremity fracturesin motor
vehicle collisions: the role of gender and height. Accident Analysis and Prevention
27(4): 601-606

Donohue MJ, Buss D, Oegerna T, Thompson R (1983) The Effects of Indirect Blunt
Trauma on Adult Canine Articular Cartilage. The Journal of Bone and Joint Surgery
65-A(7): 948-957

Eberhardt, A., Keer, L., Lewis, J ., and Vithoontien, V. (1990) An Analytical model
of joint contact. Journal of Biomechanical Engineering 112:407-413

Hering W, Patrick L (1977) Response Comparisons of the Human Cadaver Knee and
a Part 572 Dummy Knee to Impacts by Crushable Materials. Twenty-First Stapp Car
Crash Conference 21: 1015-1053

Huberti H, Hayes W, Stone J, Shybut G (1984) Force ratios in the quadriceps tendon
and ligamentum patellae. Journal of Orthopaedic Research 2: 49-54

Johnson-Nurse C, Dandy D (1985) Fracture-Separation of Articular Cartilage in the
Adult Knee. The Journal of Bone and Joint Surgery 67-B(10: 42-43

 

l3. MacKenzie EJ, Siegel JH, Shapiro S, et.al. ( 1988) Functional Recovery and Medical
Costs of Trauma: An analysis by Type and Severity of Injury. Journal of Trauma 28:
281-298

14. Melvin J, Stalnaker R, Alem N, Benson J, Mohan D (1975) Impact Response and
Tolerance of the Lower Extremities. Stapp Car Crash Conf 19: 543-559

15. Nagel D, States J (1977) Dashboard and bumper knee-will arthritis develop? AAAM
proceedings 21: 272-278

16. Newberry W, Haut R (1996) The effects of subfracture impact loading on the
patellofemoral joint in a rabbit model. 40th Stapp Car Crash Conference 149-159

17. Patrick L, Kroell C, Mertz H (1965) Forces on tehehuman body in simulated Crashes.
Stapp Car Crash Conf 9: 237-259

18. Powell W, Ojala S, Advani S, Martin R (1975) Cadaver Femur Responses to
Longitudinal impacts. Stapp Car Crash Conf 19: 561-579

19. Radin B, Parker H, Pugh J, Steinberg R, Paul 1, Rose R (1973) Response of joints to
impact loading-III relationship between trabecular rnicrofractures and cartilage
degeneration. J Biomech 6:51-57

20. States J (1970) Traumatic Arthritis-- A Medical and Legal Dilemma. Ann Conf of the
Amer Assoc for Automotive Med 14: 21-28

21. Thompson RC, Oegema TR, Lewis JL, Wallace L (1991) Osteoarthritic changes afier
acute transarticular load. The Journal of Bone and Joint Surgery 73-A(7): 990-1001

22. Tomatsu T, Irnai N, Takeuchi N, Takahashi K, Kimura N (1992) Experimentally
produced fractures of articular cartilage and bone. J Bone Joint Surg 3:457-461, 1992

23. 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:
150-152

24. Vellet AD, Marks PH, Fowler PJ, Munro TO (1991) Occult post-traumatic
osteochondral lesions of the knee: prevalence, classification, and short-tenn sequelae
evaluated with MR imaging. Radiology 178(1):271-276

25. Viano D (1977) Considerations for a femur injury criterion. 2 1 st Stapp Car Crash
Conference 445-473

26. Volpin G, Dowd G, Stein H, Bentley G (1990) Degenerative arthritis after intra-
articular fractures of the knee. Long-term results. J Bone Joint Surg Br 72: 634-638

CHAPTER 1
DETECTION OF EXPERIMENTALLY PRODUCED OCCULT

MICROCRACKS AT THE BONE-CARTILAGE INTERFACE

Atkinson PJ 8”, Walsh JA 8‘, Haut RC 8‘“
&Orthopaedic Biomechanics Laboratories, College of Osteopathic Medicine
and the # Department of Materials Science and Mechanics, College of Engineering
Michigan State University, East Lansing, MI

ABSTRACT

We compared 3 histological preparation methods to detect experimentally
produced occult microcracks in decalcified human patellae: a paraffin tape-transfer
technique, a paraffin slab-cut method, and a paraffin method with methyl salicylate as the
clearing agent. Microcracks were observed at the bone-cartilage interface and were
oriented either parallel or perpendicular to the tidemark. Both types of microcracks were
documented with each preparation method. The slab-cut method was time consuming,
however, the section thickness allowed a detailed analysis of the architecture of
microcracks as they passed into the depth of the section. The methyl salicylate method
was efficient and produced thin, serial sections of good morphological detail, and minimal
cutting artifact. Reliable histological data were also derived from the tape-transfer

technique, however, this method was inconsistent. The methods summarized here for

processing decalcified human joint tissues provide a basis for future orthopaedic studies

investigating occult microfractures at the bone-cartilage interface.

INTRODUCTION

Clinical (Kern 1988, States 1970) and experimental (Donohue 1983, Newberry
1996, Newberry 1997, Radin 1978, Radin 1973, Shimizu 1993, Thompson 1991, Vener
1992) studies suggest that impact-induced trauma can initiate a chronic joint disease (i.e.
osteoarthrosis). The disease pathogenesis can be associated with an acute gross fracture
or occult microfi'acture involving bone and cartilage. The microcracks have been
documented as open cracks oriented perpendicular or parallel to the articular surface
occurring in areas of high intraarticular contact stress (Atkinson 1997). The microcracks
do not normally exist in untraumatized human joint tissues (Atkinson 1995). While gross
fi'actures can be documented via conventional techniques (i.e. radiographs, palpation,
etc.), occult microfractures are more difficult to diagnose. Currently, the sole diagnostic
technique for in viva detection of occult microfi'actures in humans and animals relies on
specific Magnetic Resonance Imaging (MRI) protocols (i.e. STIR). ‘Bone bruises’
associated with the microfracture can be detected with MRI (V ellet 1991), however, this
requires that a basal blood pressure exists to form the bone bruise. Unfortunately, this
technique is not suitable for in vitro studies. Histology is the only established technique in
documenting occult microfractures from experimental joint trauma studies (Atkinson
1995, Thompson 1991, Vener 1992). However, consistent and reliable methods are

needed. Recent studies (Thompson 1991, Vener 1992) base the interpretations on

10

whether the microfractures exist in serial sections. This study was designed to determine
reproducible preparation methods that yielded consistent sections of decalcified human
patella. The resulting sections were then analyzed to investigate experimentally induced
microfiactures.
METHODS AND MATERIALS
T_iss_ues

18 knee joints were harvested from human cadavers with no prior history of joint
trauma and/or evidence of advanced joint pathology. The femur and tibia were transected
and all musculature was removed. The knee joints were attached to a special fixture,
flexed 90° and impacted once at a subfracture load using a previously described protocol
(Atkinson 1995). The patellae were then dissected fiee.
Processing and Decalcification

The patellae were fixed in 10% neutral buffered formalin for 9-12 days. Formic
acid (20%) was used to decalcify the tissue. Decalcification solutions were changed every
3 days at which time a precipitate test was performed using ammonium oxalate (Lillie
1954). The dccalcification endpoint was determined with 2 successive negative precipitate
tests. Both fixation and decalcification were performed with continuous, moderate
agitation.

Following decalcification, samples were taken from each patellae (3-5 mm thick,
approximately 50 mm wide, 8 mm depth). 6 patellae each were randomly assigned to be
processed according to one of three methods: paraffin slab-cut method, paraffin tape-

transfer method, or a paraffin method using methyl salicylate. The processing schedules

ll

for each method have been previously described (Atkinson 1998). A synopsis of each

method follows.

 

fi

Slab-cut Method: Samples were dehydrated, infiltrated and embedded in paraffin.
The block was cut into 500um slabs. on an Isomet low speed saw (Buehler Ltd., Lake
Bluff, IL) using a diamond blade. The slabs were stained with safianin O-fast green and
coverslipped (Rosenburg 197 1).

Paraffin Tape-Transfer Method: Samples were dehydrated, infiltrated and
embedded in paraffin. Sections were cut at 8um on a rotary rrricrotome using a 150 mm
stainless steel knife. As the sections were cut, they were captured on 50.8 mm sealing
tape (3M Scotch 375, 3M St Paul MN) (Sterchi, BL. and Eurell, J.C. 1990), floated on a
water bath at 45°C, retrieved on slides, coated with Poly-l-Lysine (Sigma, St. Louis, MO),
and placed on a warming plate. The slides were covered with a strip of plastic wrap,
stacked, clamped together in a vise and maintained at 60°C overnight followed by an
additional 24 hr at 37°C. The slides were placed in xylene at room temperature overnight
to remove tape. The sections were stained with safranin O-fast green and coverslipped.

Paraffin Method using Methyl Salicylate: Samples were slowly dehydrated in a
series of increasing alcohols followed by three changes of methyl salicylate (Skinner
1986). Times for methyl salicylate ranged fi'om 3-6 hr with agitation. Infiltration
occurred in 3 changes of paraffin (6 hr each) at 60-65°C under vacuum (15-18 psi).
Samples were embedded in paraffin and cured at 4°C. A block sofiener was used in
addition to icing the block face for 10-15 minutes immediately prior to sectioning. A

rotary microtome was used to section the blocks at 6-9 um. The resulting sections were

12

floated in a 45-48°C water bath and retrieved on slides. No slide adhesive was used. The
slides were warmed in 2 steps: 53-55°C for 10 min and 62-75°C until the section was flat.
The sections were then dried overnight at 37°C, stained with safranin O-fast green and
coverslipped.

The resulting sections from each patellae were examined independently with light
microscopy by two of the authors (P.A., J .W.) for structural detail and then photographed.
RESULTS

Vertical (perpendicular to the articular surface) and horizontal (parallel to the
articular surface) occult microcracks were documented in approximately 50% of all
patellae using each method (Figure l, A-F). All cracks were ‘open’, involving a
separation of bone and/or cartilage. The horizontal cracks were observed with a
magnification of 40 power at the subchondral bone-calcified cartilage interface in one of
three areas: a delamination of the cartilage from the calcified cartilage, a delamination of
the calcified cartilage from the subchondral bone, or a crack confined to the subchondral
plate. The vertical cracks typically initiated in the trabecular bone near the subchondral
plate, and extended across the subchondral plate and the tidemark into the cartilage.
While these cracks did not communicate to the articular surface, they were visually
discernible on some tissues as a faint shadow when viewing the articular surface of the
patella prior to cutting. The appearance of the microcracks was consistent, regardless of
the processing method. With higher magnification (100-400X), a previously established
protocol (Frost 1960) was used to document the existence of small, intact cracks oriented

vertically in the calcified cartilage of some patellae. These cracks have been previously

13

described as normally occuring in human tissues, indicative of joint morbidity (Sokoloff
1993; Villanueva 1994). Therefore, these microcracks were considered a preexisitng
condition and not included for analysis in the current study.

Intact sections suitable for the detection of microcracks were procured with
varying levels of effort using the three methods. While serial sections were easily taken
with the slab-cut technique, it was very time consuming, requiring 3 to 5 hours to cut I
block. A top-lit stereomicroscope was found to be the best method of providing a detailed
analysis of the microcracks. Analysis with standard light microscopy was hindered by
reduced light transmission due to the thickness of the slab. However, because of the
section thickness, this method gave the added dimension of depth to the microcracks
which often communicated to both sides of the section (Figure 2). It was difficult to
obtain serial sections with the tape-transfer technique. Typically, 5 of 20 sections were
considered suitable for staining. The sections that were stained were of good quality.
Analysis with light microscopy at a range of magnifications (40-400X) showed the
cartilage, bone and tidemark to be free of preparation artifact. The methyl salicylate
method produced serial sections with considerable ease. Analysis with light microscopy
showed good cartilage, bone and cellular detail and the absence of preparation artifact.
While increased time was required for dehydration and clearing, the improved quality of
cutting was significant. Using a block softener and icing the block face before sectioning

for 10-15 min greatly enhanced the ease of cutting.

14

DISCUSSION

This study presents a comparison of 3 histologic preparation methods applied to
decalcified hm'nan patellae. It was designed to identify a reproducible method which
consistently produces high quality serial sections to allow the detection of occult
microfractures in the human patella. The detection of occult microfractures is currently
limited to histological methods for in vitra studies. This underscores the need to identify
suitable methods for continued study in this area.

Experimentally produced ‘open’ vertical and horizontal occult microfractures were
documented using all three methods. The architecture of the microcracks was visually
similar fiom all methods, and similar to past studies using a single method. For example,
animal and human experimental studies have shown horizontal (Armstrong 1985; Atkinson
1995, Radin 1978; Radin 1973; Zimmerman 1988) and vertical (Donohue 1983;
Thompson 1991; Vener 1992) occult microcracks at the bone-cartilage interface.

Additional studies have documented a variety of in viva (preexisting) occult
microcracking. While it is difficult to establish the exact cause of the microcracks, they
may be due to a single trauma (subfracture joint insult) (Thompson 1991), repetitive
microtrauma (Radin 1973), or a consequence of aging/joint morbidity (Villanueva 1994).
Studying the canine, Van Sickle and Evander (1997) document open, horizontal
microfractures at the bone-cartilage that are visually similar to the horizontal cracks in the
current study. Villanueva, et a1. (1994) retrieved typically degenerated femoral heads

during prosthetic surgery and discovered small, vertically oriented, ‘closed’ (intact)

15

microcracks in the subchondral bone and/or calcified cartilage. Sokoloff (1993)
documents similar closed microcracks confined to the calcified cartilage. The appearance
of the in viva, closed microcracks described in these studies differ from the experimentally
produced occult microfractures documented in the current study and the past studies
noted above. We hypothesize there may be two different explanations for this variation in
appearance. First, the mechanism of microfracture may be different. The larger, open
microfracture may be the result of a more acute injury associated with a high energy joint
insult. Altemately, the smaller, intact microfi'actures may be a chronic phenomenon (i.e.
associated with tissue remodeling). A second, related possibility is that the larger, open
microcracks may close as the tissue heals; eventually they may reduce in size taking on the
appearance of the smaller, closed microcracks.

In the current study, the presence of occult microcracks was verified a priori via
faint shadows observed on the articular surface of some patella prior to any cutting. If this
visualization is possible for in viva tissues, it may provide a helpful tool for clinical
detection of subfracture joint injuries, in addition to MRI (i.e. via arthroscopy).

The methods outlined in this study provides flexibility for studies investigating
occult microfractures when processing decalcified human joint tissues. For example, while
the slab-cut method was time consuming, it allowed a detailed analysis of the architecture
of microcracks as they passed into the depth of the section. This method also facilitated
taking serial sections free of cutting artifact. Altemately, the methyl salicylate method
should be used if thin sections are required. This method resulted in easy cutting,

consistent serial sections, good morphological detail, and a reduction of cutting artifact.

l6

Reliable histological data were also derived from the tape-transfer technique, however,

this methods was inconsistent. The methods summarized here may provide a basis for

future orthopaedic studies investigating disruptions at the bone-cartilage interface.

ACKNOWLEDGMENTS

This study was supported by a grant (R49/CCR503607) from the Centers for Disease

Control and Prevention. 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.

We are indebted to Mr. Rick Shumaker and the Michigan Tissue Bank for solicitation of

donors and retrieval of joints.

REFERENCES

1.

Armstrong, C., Mow, V., Wirth, C. 1985. Biomechanics of impact-induced
microdamage to articular cartilage: a possible genesis for chondromalacia patella.
American Academy of Orthopaedics Surgical Symposium on Sports Medicine, ed. G.
Fenerrnan, pp. 70-84.

Atkinson, P.J., Walsh, J. and Haut, RC. 1998. The human patella: a comparison of
three preparation methods. 1998. 21(2) Journal of Histotechnology (In Press)

Atkinson, P.J., Garcia, J.J., Altiero, NJ. and Haut, RC. 1997. The influence of impact
interface on human knee injury: implications for istrument panel designa and the lower
extremity criterion. 41 st Stapp Car Crash Conference. 41 :167-180.

Atkinson, RI. and Haut, RC. 1995. Subfi'acture insult to the human cadaver
patellofemoral joint produces occult injury. Journal of Orthopaedic Research 13, 936-
944.

Donohue, M.J., Buss, D., Oegema, T. and Thompson, R. 1983. The effects of indirect
blunt trauma on adult canine articular cartilage. The Journal of Bone and Joint Surgery
65-A(7), 948-957.

Frost, HM. 1960. Presence of microscopic cracks in viva in bone. Henry Ford
Hospital Medical Bulletin. 8:25-35.

Kern, D., Zlatkin, M. and Dalinka, M. 1988. Occupational and post-traumatic arthritis.
Radiologic Clinics of North America. 26: 1349-1358.

17

10.

ll.

12.

13.

14.

15.

l6.

l7.

18.

19.

Lillie, R. 1954. Histalagic T echnic and Practical Histachemistry. McGraw Hill, NY,
p. 424.

Newberry, W.N., Garcia, J .J . and Haut, RC. 1996. The effects of subfracture impact
loading on the patellofemoral joint in a rabbit model. 40th Stapp Car Crash Conference
SAE paper 962422, 149-159.

Newberry, W.N. and Haut, R. 1997. A single blunt impact to a rabbit patella-femoral
joint induces OA changes in cartilage and subchondral bone. Journal of Orthopaedic
Research 15, 450-455.

Radin, E., Ehrlich, M., Chemack, R., Abernethy, R, Paul, I. and Rose, R. 1978. Effect
of repetitive impulsive loading on the knee joints of rabbits. Clinical Orthopaedics and
Related Research 131 :288-293.

Radin, E., Parker, H., Pugh, J., Steinberg, R., Paul, I. and Rose, R. 1973. Response of
joints to impact loading-III relationship between trabecular microfiactures and
cartilage degeneration. Journal of Biomechanics 6:51-57.

Rosenburg, L. 1971. Chemical basis for the histological use of Safranin O in the study
of articular cartilage. Journal of Bone and Joint Surgery 53(A):69-82.

Shimizu, M., Tsuji, H., Matsui, H., Katoh, Y. and Sano, A. 1993. Morphometric
analysis of subchondral bone of the tibial condyle in osteoarthrosis. Clinical
Orthopaedics 293:229-239.

Skinner, RA. 1986. The value of methyl salicylate as clearing agent. Journal of
Histotechnology 9(1):27-28.

Sokoloff, L. 1993. Microcracks in the calcified layer of articular cartilage. Archives of
Pathology and Laboratory Medicine 117:191-195.

States, J. 1970. Traumatic arthritis--a medical and legal dilemma. Annual Conference
of the American Association for Automotive Medicine 14:21-28.

Sterchi, D.L., Eurell, J .C. 1990. Tape-transfer technique for large paraffin blocks.
Journal of Histotechnology 13(3):207-208.

Thompson, R.C., Oegema, T.R., Lewis, J.L. and Wallace, L. 1991. Osteoarthritic
changes after acute transarticular load. The Journal of Bone and Joint Surgery 73-
A(7), 990-1001.

20. Van Sickle, D. and Evander, S. 1997. The enigma of adult and immature articular

cartilage. Journal of Histotechnology 20(3):243-252.

18

21. Vellet, D., Marks, P., Fowler, P. and Munro, T. 1991. Occult posttraumatic
osteochondral lesions of the knee; prevalence, classification, and short-term sequelae
evaluated with MR imaging. Journal of Radiology 178:271-276.

22.Vener, M.J., Thompson, R.C., Lewis, J.L. and Oegema, TR. 1992. Subchondral
damage after acute transarticular loading - an invitro model of joint injury. Journal of
Orthopaedic Research. 10:759-765.

23. Villanueva, A.R., Longo, J .A. and Weiner G. 1994. Staining and histomorphometry of
microcracks in the human femoral head. Biotechnic and Histochemistry. 69(2):81-88.

24. Zimmerman, N.B., Smith, D.G., Pottenger, L.A., Cooperman, DR. 1988. Mechanical

disruption of human patellar cartilage by repetitive loading in vitro. Clinical
Orthopaedics and Related Research 229:302-307.

19

 

 

B

Figure 1: See subsequent page for caption.

20

 

Figure I: See subsequent page for caption.

21

 

 

F

Figure 1
Figure l: Photomicrographs of horizontal and vertical occult microcracks prepared by:
1) slab-cut method (A,B arrows point to occult injury; 40X), 2) paraffin tape-transfer
method (C,D; 40X), and 3) paraffin method using methyl salicylate (D,E; 40X). Note: a
rnicrocrack at the calcified cartilage-subchondral bone interface (A,C), and a rnicrocrack at
the tidemark (E).

22

 

Figure 2: Photomicrographs of both sides (A,B; 18X) of a section (slab-cut method)
showing a vertical, occult microcrack which passes through the section thickness. The
microcrack initiates in the trabecular bone and terminates in the middle zone of the
cartilage. The thickness of the section gave the added dimension of depth and allowed
meaningful data to be derived from both sides of the section. Note: the light area in A
(arrow) is where the microcrack exits the section on the reverse side (shown in B).

23

CHAPTER 2

A METHOD TO INCREASE THE SENSITIVE
RANGE OF PRESSURE SENSITIVE FILM

Atkinson PJ, Newberry WN, Atkinson TS, Haut RC
Orthopaedic Biomechanics Laboratories, College of Osteopathic Medicine
and the Department of Materials Science and Mechanics, College of Engineering
Michigan State University, East Lansing, MI

ABSTRACT

Pressure sensitive film is frequently used in biomechanics to document intra- and
extra-articular contact pressures. This often involves the contact of two surfaces of
varying curvature producing non-uniform pressure distributions. The purpose of this
study was to determine the feasibility of using multiple films in such experiments to yield
accurate pressure and contact area data. A composite arrangement of film was
dynamically loaded using cylindrical indenters of five radii. An analytical model of each
indentation was constructed to provide a standard for error analysis. The study showed
that several ranges of pressure sensitive film can be used simultaneously to accurately
transduce contact pressures arising from loading scenarios that produce contact pressure

gradients and contact pressures that involve suprathreshold loading of a given film range.

24

INTRODUCTION

Pressure sensitive film commonly used in biomechanics is constructed of a color
developing layer sandwiched by a polyethylene terephthaylate (PET) base and a
microcapsule layer. The film comes in several ranges (i.e. low range, 2-10 MPa; medium
range, 10-50 MPa; high range, 50-130 MPa) and has been extensively used to transduce
intra- and extra-articular dynamic (azAtkinson 1995, Newberry 1996, Haut 1995, Haut
1989, Vener 1992, Li 1995) and quasi-static loading (Huberti 1984, Fukubayashi 1980).
Because only a single range of film was used in these studies, contact area and/or pressure
data may have been inaccurate (Atkinson 1997, Vener 1992). For example, low range
film can yield accurate contact areas due to its sensitivity to low contact pressure.
However, pressure data may be erroneous if the experimental contact pressure field
exceeds the relatively low upper threshold of the film. Errors in contact area and pressure
measurements may also exist for films that have higher ranges of sensitivity. By
incorporating more than one range of film into experimental studies, more accurate
contact area and pressure data can be transduced.

The purpose of the current study was to determine the feasibility of using multiple
films simultaneously in a given dynamic experiment to yield accurate pressure data and
measures of contact area. Low, medium and high range pressure sensitive film coupons
were dynamically loaded in a composite arrangement (high, low and medium ranges) by

means of cylindrical indentations. Digital analysis of the film documented the mean

25

contact pressure and contact area. An analytical model of each indentation was
constructed to provide a standard for error analysis.
METHODS AND MATERIALS
Ermerimental Protocol

Idealized contact stress distributions were obtained experimentally by indenting
packets of pressure sensitive film coupons (described below) with cylindrical indenters
(Hale 1992) of five different radii (5.5, 12, 19, 66.5 and 94 mm). The film packets were
compressed between the cylinder and a thick plate to produce a contact pressure
distribution along the transverse axis. The length of the cylinder was much greater than
the contact width, so that the longitudinal variation in contact stress was negligible. The
cylinders and plate were machined from stainless steel and polished to a smoothed finish.
The packets were loaded via a 60 Hz haversine pulse in a materials testing machine (model
1331, retrofitted 8500 electronics, Instron, Canton, MA) to simulate pulses derived from
transarticular loading experiments (bzAtkinson 1995, Newberry 1997, Thompson 1991).
The indenters were attached to the machine actuator, and the plate was supported by a
load cell which was mounted on the fixed lower base of the machine. Load data was
sampled at 5 kHz on a storage oscilloscope. Multiple indentation experiments were
conducted for each indenter radius. A representative experiment was then analyzed per
methods described below.

The packets consisted of pressure sensitive film coupons (Fuji Film Prescale,
Itochu, Int., Montreal CA) of three ranges: dual sheet low (L W R270 10M), and single

sheet medium (M S R270 10M) and high (H S R270 10M). The stacked coupons were

26

inserted into a polyethylene packet to reduce shear loading artifact (Huberti 1984) and to
simulate how the film is commonly protected from moisture when used in hydrated
environments (articular joints). The overall thickness of the packet was 0.44 mm (high
and medium range film = 0.085 mm each, low range film ‘A’ and ‘C’ sheets = 0.08 rrrrn
each, two-ply polyethylene packet = 0.04 mm/ply). The packets were indented to achieve
a nominal peak contact pressure of 60 MPa to ensure that the high pressure film (lower
threshold of 50 MPa) reacted to the different loading scenarios.

The film was calibrated by conducting additional indentation tests on the stacked
film packets described above with a plane ended square indenter (1 cm square). A wide
range of loads was applied via a 60 Hz pulse. Pilot experiments showed wide variations in
pressure stains with regard to the stacking order, loading rate and whether the film
coupons were loaded separately or stacked. Therefore, the loading rate and film packet
construction were consistent for all tests in the current study. The coupons were stacked
in the following order: high, low and medium with the PET surfaces oriented in a
consistent direction. The peak applied load divided by the area of the steel indenter was
taken as the experimentally applied pressure. Following load application, the fihn was
removed fi'om the packet and digitized on a scanner (Scanmaker MRS-1200E6, Microtek,
Taiwan) using commercially available software (Photostyler, version 1.1A, Aldus Co.,
Seattle WA) at 150 dpi in 8-bit gray scale (0 = white, 255 = black) within four hours of
testing. The mean image density for each loading was determined using an image analysis

program (Image, version 1.6, NIH). A calibration table was made from the mean image

27

densities and experimentally applied pressure data for each range of film. Third order
polynomial curves were fit to these data (r2 2 0.98).

The pressure sensitive film coupons from the cylindrical indentation experiments
were analyzed separately (by film range) by taking an image slice several pixels thick along
the transverse axis of contact. Polynomial fits of each range of film were determined for a
given experiment (Figure 1). For example, a polynomial was fit to the low range film over
its usable range (2-10 MPa). Similar curves were made for the medium and high range
films over their respective ranges. The contact width was determined by calculating the
roots of the polynomial for a given range of film at the lower threshold of that film
(Mathernatica, version 3.0, Wolfiam Research, Champagne IL). The mean contact
pressure was determined for each range of film for a given experiment by integrating the
respective polynomial and dividing by the contact width. For instance, the mean pressure
predicted by the low range fihn was calculated as the area under its respective polynomial
curve. The area was bounded at each end by the intersection of the curve with the lower
boundary of the film (2 MPa). Finally, to incorporate data fiom all pressure ranges, a
‘composite’ curve was made by fitting a polynomial to the data points fi'om all pressure
ranges. For the composite curve, only the non-thresholded data points were used from the
low and medium range films. The mean contact pressure was also determined for the
composite curves.

Theoretical validation
A mathematical model of the cylindrical indentation was made as an independent

comparison for the experimental data (Hale 1992). A plane strain contact model

28

consisting of a rigid indenter, film packet, and rigid plate was constructed using a
commercial code (Abaqus, version 5.6-1) to predict the half-width contact pressures for
each experiment. The film packet was assumed to be a homogeneous solid and was
modeled using 80, 8-noded elements. The composite modulus of the packet, 22 MPa, was
determined by matching the contact area of a 12 mm radius pilot experiment then iterating
on the modulus to minimize the difference between the pressures predicted by the model
and those in the experiment. Indentations with the remaining indenter radii were simulated
by indenting the film packet layer until the contact width of the model matched that
predicted by the composite fihn. Poisson’s ratio was assumed to be zero as this parameter
had a minimal influence on the predicted contact pressures. The difference in the half-
width contact pressure profiles between the composite curves and the model were
quantified by calculating the Chi-squared difference.
RESULTS

Correlation coefficients for the polynomial fits of the experimental data were
typically high (mean r2 = 0.929 i 0.09, Figure 1). In all cases the composite film exhibited
the least error in mean contact pressure and peak pressure compared to the theoretical
solution (Table 1). The mean pressure and peak pressure recorded with the composite
film were within 5% and 6% of the theoretical, respectively. The high pressure film
documented mean and peak pressures that were 38% and 10% greater than theoretical,
respectively. Mean and peak pressures for medium film were less than the theoretical by
10% or less for all radii with the exception of 94.0 mm. For 94.0 mm radius indenter

these errors were approximately 25% and 45%, respectively. The higher errors for low

29

and medium film for the 94.0 mm radius are a result of the higher applied pressure for this
case. Mean and peak pressures for low film were less than theoretical by approximately
80% for all radii. The pressure distributions predicted by the theoretical model and those
documented experimentally using composite film were visually consistent for all radii, as
evidenced by the low chi-squared values (Table 1, Figure 2). The composite film typically
predicted a contact half width within less than 10% of the low pressure film (Table 2).
The medium pressure film typically underrepresented the contact half width predicted by
the low film by 15% to 22%, and the high pressure film underrepresented this contact half
width by 50% to 80%.
DISCUSSION

This study documented that several ranges of pressure sensitive film can be used to
accurately transduce contact pressures and areas in load scenarios which produce: 1)
contact pressure gradients, and 2) contact pressures which involve suprathreshold loading
of a given film’s pressure range. Such loading scenarios can occur during loading of
articular joints. For example, contact in the patellofemoral or tibiofemoral joints involves
two surfaces of varying curvature (i.e. the ‘curved’ femoral condyle contacting the flat
tibial plateau). This results in peak contact pressures at the center of contact, and
gradually reduces to zero pressure at the edge of contact (Hayes 1978). If the contact
pressures extend beyond a given film’s threshold limits, contact area and pressure data will
be compromised. In such cases, pilot experiments should be conducted to determine the
appropriate ranges of film needed to accurately transduce both contact area and pressure.

Pilot studies also showed that the image produced on the film can be influenced by the

30

stacking arrangement and therefore the calibration of stacked pressure sensitive film
should be performed in the desired stack sequence. Finally, due to the film’s rate
sensitivity (Haut, 1989), the film should be calibrated with a load pulse that approximates
the anticipated experimental conditions for which the film will be used.

While the composite curves best agree with the theoretically predicted contact
pressure distributions, small differences did exist, possibly due to our analysis technique.
Polynomial fits of the data were used for analysis, instead of the raw data. This protocol
was adopted to allow calculations to be performed (i.e. contact area, mean pressure)
without the influence of local discontinuities in the raw data. Further, the correlation
coefficients were typically high, suggesting that the polynomials adequately described the
raw data. In addition, the theoretical curves were obtained by modeling the layered packet
as a homogeneous body. It is possible that interactions between the film layers
contributed to differences between the model and experiments, however these interactions
were likely small given the fit of the model to the experimental data.

In the current study, data from all film ranges was used to produce a composite
curve that closely approximated the theoretical peak and mean pressure, and contact
pressure profile for a one-dimensional problem. The composite curve, because it contains
pressure data fi'om the low pressure fihn, provides an excellent measure of contact area.
This method could also be applied to two-dimensional pressure data, such as that arising
fiom contact between two articular surfaces. The pressure image fi'om a given film range
can be easily represented as a two-dimensional array of calibrated pressure data. The

arrays from different film ranges chould then be added, cell by cell, to yield the composite

31

pressure distribution. Pertinent data can then be deduced from the composite array of

data. The methods outlined here provide a new approach for accurate measures of

contact area and pressure during dynamic loading of a diarthrodial joint.

ACKNOWLEDGMENTS

This study was supported by a grant (R49/CCR503607) fiom the Centers for Disease

Control and Prevention. 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.

REFERENCES

l.

azAtkinson, PI. and Haut, RC. (1995). Subfiacture Insult to the Human Cadaver
Patellofemoral Joint Produces Occult Injury. Journal of Orthopaedic Research 13,
936-944.

. bzAtkinsorr, P. and Haut, R. (1995). Insult to the human cadaver patellofemoral joint:

effects of age on fi'acture tolerance and occult injury. 39th Stapp Car Crash
Conference 281-294.

Atkinson, R]. and Haut, RC. (1997). A Method for Determining Regions of
Diarthroidal Joint Contact During Dynamic Loading of the Human Knee. Proc. 44th
Annual Meeting Orthopaedic Research Society 656.

Fukubayashi, T. and Kurosawa, H. (1980). The Contact Area and Pressure
Distribution Pattern of the Knee. Acta Orthop. Scand. 51, 871-879.

Hale, IE. and Brown, TD. (1992). Contact stress gradient detection limits of
pressensor film. I of Biomechanical Engineering 114, 352-358.

Haut, R. (1989). Contact Pressures in the Patello-femoral Joint during Impact Loading
on the Human Flexed Knee. Journal of Orthopaedic Research 7, 272-280.

Haut, R., Ide, T. and DeCamp, C. (1995). Mechanical responses of the rabbit patello-
femoral joint to blunt impact. Journal of Biomechanical Engineering

Hayes, W., Swenson, L. and Schurman, D. (1978). Axisymmetric finite element
analysis of the lateral tibial plateau. Journal of Biomechanics 1 1, 21-33.

Li, X, Haut, R. and Altiero, N. (1995). An analytical model to study blunt impact
response of the rabbit P-F joint. Journal of Biomechanical Engineering 1 17, 485-491.

32

10. Newberry, W.N., Garcia, J .J . and Haut, RC. (1996). The effects of subfi'acture impact
loading on the patellofemoral joint in a rabbit model. 40th Stapp Car Crash Conference
SAE paper 962422, 149-159.

11. Newberry, W.N., Zukosky, D.K. and Haut, RC. (1997). Subfi'acture 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.

12. Thompson, R.C., Oegema, T.R., Lewis, J.L. and Wallace, L. (1991). Osteoarthritic
changes after acute transarticular load. The Journal of Bone and Joint Surgery 73-
A(7), 990-1001.

13. Vener, M., Thompson, R., Lewis, J. and Oegema, T. (1992). Subchondral Damage
after acute transarticular loading: an in vitro model of joint injury. Journal of
Orthopaedic Research 10, 759-765.

TABLES

Table 1. Mean pressure (over the half-width contact) and peak pressures @resented in
MPa as mean pressure / peak pressure) predicted by the theoretical model and
documented fi'om the composite, high, medium, and low pressure film data.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Radius Theoretical Chi- Composite High Medium Low

(mm) squared pressurefilrn pressurefilmpressurefilmpressmefilm
fit“

5.5 41.5 / 55.9 2.0 41.2 / 56.8 57.2/ 60.7 37.8 / 50.0 9.0/ 10.0
12.0 41.7 /55.0 0.2 41.6 / 55.5 55.5 / 58.2 37.9 / 50.0 9.2/ 10.0
19.0 43.0 / 55.9 8.0 44.3 / 59.2 56.2 / 59.4 40.6 / 50.0 9.2/ 10.0
66.5 47.4/ 63.6 32.3 52.6 / 71.5 67.6 / 76.3 38.8 / 50.0 8.8/ 10.0
94.0 59.4/ 90.0 6.3 61.9 / 90.5 79.9 / 95.0 44.3 / 50.0 10.0/ 10.0

“' Chi-squared fit: This represents a normalized measure of the summed difference

between the theoretical and composite film data.

Table 2. Contact half-width (mm) predicted by the theoretical model and documented by

low, medium, and high pressure film data.

 

 

 

 

 

 

 

 

 

 

 

Radius (mm) Composite High Medium Low
pressure film pressure film pressure fihn pressure film

5.5 0.82 0.24 0.75 0.88

12.0 0.82 0.19 0.72 0.92

19.0 0.72 0.28 0.65 0.80

66.5 1.42 0.46 1.30 1.41

94.0 3.94 2.26 3.76 4.26

 

 

33

 

FIGURES

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1W . 7 O Highrangedata
339,4 ~----Highrangepoiy
. I O Mediumrange
. ——Mediumrangepoiy
A [j Lowrangedata
a“: ---Lowrangefilmpoiy
g 70 [IA/f 3‘ —Compositefitpoiy
e .
3
i
O.
E

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

81910

3 4 5 6 7
Contact width (mm)

Figure 1. Plot of the calibrated pressure film data and the corresponding polynomial fits
(poly) used for data analysis. Contact half-width, mean and peak pressures were
calculated individually for the three ranges of film (low, medium and high) and for the
composite fit.

34

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

100
WAQ
ao \
E 70 \
2 V
E so \
a \\
3 so
E
~ 40
s \\
C
o 30
o \
20 —-modelprediction \\
1o —oompositefiimdata \\
o
o as 1 t5 2 25 3 as 4

Half-width contact distance (mm)

Figure 2. Plot. of the half-width contact pressure profile (r = 94 mm) from the composite
polynomial fit of the experimental data and the profile predicted by the mathematical
model. The film packet was compressed between a cylindrical indenter and a flat plate in

the mathematical model (inset).

35

CHAPTER 3

A METHOD FOR DETERMINING REGIONS OF DIARTHRODIAL JOINT
CONTACT DURING DYNAMIC LOADING OF THE HUMAN KNEE

Atkinson PJ, Haut RC
Department of Materials Science and Mechanics
Michigan State University, East Lansing, MI

ABSTRACT

Areas of joint contact are necessary to understand joint pathologies and joint
mechanics. In this study a method is described for contact area measurement fiom
dynamic loading.
INTRODUCTION

Accurate determination of diarthrodial joint contact is necessary for theoretical
modeling of joint contact mechanics. Stresses on articular cartilage and bone arise fiom
both static and dynamic loading. Accurate predictions of these stresses can aid in the
understanding of tissue modeling and remodeling, and the development of stress based
damage criteria for cartilage and bone. For example, recent studies propose stress based
criteria for diarthrodial joints to explain how high levels of stress in bone and cartilage may
cause osteoarthrosis (Atkinson 1995, Donahue 1983, Ateshian 1994) and chondromalacia
patellae (Goodfellow 1976, Minns 1979). A number of methods have been employed to

determine joint contact area under static loading (Ateshian 1994, Huberti 1984). Pressure

36

sensitive film has been shown suitable for use during dynamic loading of joints to quantify
both contact pressure and area (Atkinson 1995, Vener 1992). However, because the film
is designed to function within certain pressure ranges, joint contact area at pressures less
than the film’s minimum pressure threshold is not recorded. Further, it is unknown where
contact occurs on the articular surfaces. In this study, we investigated the effectiveness of
biological stains in producing a permanent record of dynamic joint contact. Water soluble
stains were eliminated as they could not differentiate contact areas in the hydrated joint.
The ideal stain would be sensitive to low contact pressures and react to loads over short
time durations.
METHODS

Sudan Black adipose stain was selected for this study after pilot investigations on
numerous other biological stains. A packet was prepared using medium Fuji pressensor
film (10-50 MPa, dynamically) encased in thin polyethylene film (Figure 1). Double sided
tape affixed to the outside of the packet acted as the Sudan stain carrier. The technique
was evaluated during blunt insult experiments on the 90° flexed human knee joint
(Atkinson 1995). The packet was inserted into the patellofemoral joint of 5 hmnan
cadavers with grossly healthy articular surfaces. The joints were impacted (Figure 2) with
a free flight inertial mass of 4.5 kg at an average energy of 27J. After insult the articular
surfaces were irrigated and photographed. The stained contact area was manually
outlined on a thin plastic sheet. The contact areas from the pressure film and plastic sheet

were measured using a previously documented computerized technique (Atkinson 1995).

37

RESULTS

The blunt insults on the human knee generated average contact loads of 5 kN in 7
ms. No bone fi'actures were generated in 4/5 experiments. The images of the contact
areas generated by the stain after irrigation had the same general shape as the contact area
recorded by the film. The stained area on the articular surface, however, was 12% larger
(Table 1) on average. The stained area on the patellar and femoral surfaces were visually
consistent (Figure 3). The variations in contact area between the stain and film methods
were small within specimens, but there were large variations between specimens.
DISCUSSION

We present a simple, quick, inexpensive, noninvasive technique for measurement
of contact areas in diarthrodial joints during impact loading. Sudan stain on a film packet
records articular joint contact in the same regions where the film records contact pressure.
The larger contact area produced by the stain suggests it reacts at pressures less than the
minimum threshold of the fihn (10 MPa). This suggests the stain yields a more accurate
contact area in addition to being able to locate contact area on articular surfaces.
Comparisons of this method with others will be difficult during dynamic loads because of
the biphasic nature of cartilage. A ‘low pressure’ film (2.5-10 MPa) would have yielded
more accurate contact area, however, the film would be saturated by pressures (15-20
MPa) resulting fi'om dynamic loading of the human patellofemoral joint (Atkinson 1995).
The Sudan or a similar stain could be used to provide important positioning information
for 3-D analytical models or, in conjuction with the pressure film data, indicate the most

appropriate orientation for 2-D analytical models. The marking also provides a means by

38

which areas of potential damage might be isolated for histologic analysis. In a clinical
setting the method may assist those investigating the influence of flexion and Q-angle on
knee joint contact areas for in vitro and possibly in vivo studies.

We hypothesize two possible mechanisms by which the Sudan technique may
function. The Sudan stain may be loosely impacted into the superficial layer of the
cartilage. Altemately, the Sudan stain may be reacting with small, freshly exposed areas of
fat in areas of joint contact - in essence reacting in areas of very subtle damage. Therefore,
this method may also highlight areas of existing damage to articular surfaces and play a
role in advanced diagnosis of a chronic disease.

ACKNOWLEDGMENTS
Supported by the Centers for Disease Control and Prevention (R49/CCR503607).
REFERENCES

1. Atkinson P, Haut R (1995) Insult to the human cadaver patellofemoral joint: effects of
age on fracture tolerance and occult injury. 39th Stapp Car Crash Conference 281-294

2. Donohue MJ, Buss D, Oegema T, Thompson R (1983) The Effects of Indirect Blunt
Trauma on Adult Canine Articular Cartilage. The Journal of Bone and Joint Surgery
65-A(7): 948-957

3. Ateshian G, Kwak S, Soslowsky L, Mow, V (1994) A stereophotographic method for
determining in situ contact areas in diarthrodial joints, and a comparison with other
methods. Journal of Biomechanics 27(1): 11 1-124

4. Goodfellow J, Hungerford D, Woods C (1976) Patellofemoral joint mechanics and
pathology: chondromalcia patellae. The Journal of Bone and Joint Surgery
583(3):291-99

5. Minns RJ, Birnie AJM, Abemethy PJ (1979) A stress analysis of the patella, and how
it relates to patellar articular cartilage lesions. Journal of Biomechanics 12: 699-71 1

6. Huberti H, Hayes W, Stone J, Shybut G (1984) Force ratios in the quadriceps tendon
and ligamentum patellae. Journal of Orthopaedic Research 2: 49-54

39

7. Vener MJ, Thompson RC, Lewis JL, Oegema TR: Subchondral damage after acute
transarticular loading: an in vitro model of joint injury. Journal of Orthopaedic
Research 10:759-765, 1992

T 2) data

Specimen Contact Area: Contact Area:

    
   

 

  

 

 

 

packet '

r-II-I-I-------------------

      

 

   

Figure l: A schematic of the prepared film packet.

40

packet

 

 

Figure 2: A side view of the knee showing placement of the fihn packet in the
patellofemoral joint. The arrow points in the direction of the impact force.

 

stained patella stained femur

 

 

 

   

 

 

 

 

 

Figure 3: A typical imprint from the pressure sensitive film (A). The fihn image
superposed on schematics of the patellar (B) and femoral (C) articular surfaces.

41

CHAPTER 4

SUBFRACTURE INSULT TO THE HUMAN CADAVER
PATELLOFEMORAL JOINT PRODUCES OCCULT INJURY

Patrick J. Atkinson and Roger C. Haut
Department of Materials Science and Mechanics

Orthopaedic Biomechanics Laboratories

Michigan State University

ABSTRACT

The current criterion used by the automotive industry for lower extremity injury is
based on visible bone fracture. Studies suggest, however, that chronic joint degeneration
may occur after subfracture impact loads on the knee. We hypothesized that subfracture
loading of the patellofemoral joint could result in previously undocumented microtrauma
in areas of high contact pressure. In the current study, seven human cadaver
patellofemoral joints were each impacted with successively greater energy until visible
fi'acture was noted. Transverse and comrrrinuted fractures of the patella were noted at 6.7
kN of load. Forty-five percent of that impact energy was then delivered to the
contralateral joint. Subfiacture loads of 5.2 kN resulted in no gross bone fracture in five
of seven specimens. Histological examination of the patellae revealed occult trauma in
four of seven subfracture specimens consisting of a horizontal split fracture in the
subchondral bone, at the tidemark, or at the calcified cartilage-subchondral bone interface.

The trauma appeared predominantly on the lateral facet, adjacent to, or directly beneath

42

preexisting fibrillation of the articular surface. Surface fibrillation was noted (in
histological sections of non-impacted control patellae, but occult damages were not
observed. While the mechanism of this occult trauma is unknown, similar damage has
been shown to occur from direct shear loading. Since these microcracks can potentiate a
disease process in the joint, this study may suggest that the current injury criterion, based

on bone fi'acture alone, is not sufficiently conservative.

INTRODUCTION

Traumatic injury is the third leading cause of death in the U.S., surpassed by heart
disease and cancer, and motor vehicle injuries rank second to cancer in total societal cost
(Rice 1989). While more recent statistics suggest a lowering of fatalities from motor
vehicle accidents (likely due to the mandated use of seatbelts and airbags), more cases of
less severe injury are being reported. A high frequency of lower extremity injuries, in
particular, are showing up even with seat belt restrained drivers (Dischinger 1992). The
currently accepted criterion for prediction of a lower extremity injury during a motor
vehicle accident is based on bone fracture alone. The injury criterion for knee contact
with the instrument panel is 10 kN, based on femur forces measured in an
anthropomorphic dummy (Nyquist 1985). This criterion is based on knee contact forces
fiom seated, human cadavers that cause fracture of the femur or a split fiacture of the
patella (Patrick 1965, Melvin 1975, Powell 1975).

An additional complication associated with lower extremity trauma is post-
traumatic osteoarthrosis (States 1970). Clinical studies have long associated mechanical

insult with the onset of disease (Davis 1989, Johnson-Nurse 1985, Kellgren 1958).

43

Animal studies have been conducted in order to establish a correlation between mechanical
insult and the onset of a progressive disease. Donahue et a1. (1983) reported changes in
the zone of calcified cartilage of the canine patella-femoral joint two weeks after
transarticular loading at 45% of the energy needed to cause visible fracture of bone. More
recent canine studies report vertically oriented microfractures in calcified cartilage prior to
bone fracture, without disruption of the articular surface (Donahue 1983, Thompson
1991, Vener 1992). These researchers suggest that changes in the calcified cartilage and
underlying bone form the basis for development of a diseased state in the joint. The
hypothesis is consistent with earlier studies of Radin and colleagues (197 3) who theorize
that rnicrodamage and subsequent stiffening of the underlying subchondral bone results in
abnormal stresses in the overlying cartilage, leading to a progressive degeneration of this
soft tissue layer. Shimizu et a1. (1993) studied the human tibial plateau and found
degradation in the tmderlying bone preceding changes in the overlying cartilage. A recent
clinical study of high energy knee injuries indicate changes in the subchondral bone
detected by magnetic resonance imaging that were associated with "overt cartilage loss or
defect in 48% of the patients evaluated 6 months after injury" (Vellet 1991).

Past studies on the mechanisms of human osteoarthrosis have attempted to identify
which factors play a role in initiating joint degeneration. It has been suggested that
microcracking at the cartilage-bone interface (Minnsl979) in the human joint may be due
to high shear stresses. A recent mathematical model of the impacted human
patellofemoral joint indicates large shear stresses along the zone of calcified cartilage on

the lateral patellar facet (Shelp 1994). A study on the impact response of human cartilage

44

biopsy samples to impact loading concluded that 25 MPa is a damage threshold of human
cartilage (Repo 1977). However, pressures averaging of approximately 25 MPa in the
human cadaver patellofemoral joint are only reached or exceeded at loads causing bone
fi‘acture (Haut 1989).

While clinical associations have been made between joint insult and subchondral
bone damage, there are no experimental data documenting such occult types of injury
fi'om impact on the hmnan knee. If the injury does exist prior to visible bone fi'acture, then
the current lower extremity injury criterion may not be a conservative enough measure for
the long term health of a diarthrodial joint under blunt insult.

The primary objectives of the current study was to conduct subfi'acture
experiments on the flexed human knee joint and examine joint tissues for occult injury.
Specifically, we hypothesized: 1) subfracture blunt insult to the human cadaver knee
would produce microcracks in the calcified cartilage and/or subchondral bone; 2)
subfiacture contact loads would be less than 10 kN; and 3) that microfi'actures would
occur at contact pressures less than 25 MPa.

METHODS AND MATERIALS

Blunt impact was delivered to isolated, flexed lmee joints of seven human cadavers
aged 46fi9 years of age (1 female, 6 males). Tissue from an additional four specimens
aged 40fi22 years (seven total patellae), was used as histological controls (Table l).
Tissues were procured by the Tissue Bank (see Acknowledgment) from donors with no
prior history of joint trauma and/or evidence of advanced joint pathology (areas of

denuded bone, advanced chondromalacia, etc.). Shortly after death, the patellofemoral

45

joint was examined through a medial incision. Joint preparations were excised
approximately 15 cm from the knee proper within 4 hours of death and stored at 4° C in
17mg/ml L-15 culture medium (Sigma Chemical, St. Louis, MO) with 85mg/ml
Gentamicin Sulfate (Gibco/BRL, Gaithersburg,MD) for approximately 3 days prior to use
in this study. Superficial tissues proximal to the femoral condyle were excised and the
femoral shaft was cleaned with alcohol. The femur was potted in a cylindrical steel sleeve
with room temperature curing epoxy, and was mounted to a rigid test fiame (Haut 1989).
Each specimen was maintained at a flexion angle of 90° prior to impact with a loosely
knotted tether tied to the quadriceps tendon. The knot was such that minimal load was
required to cause slippage, thus minimizing tensile loads in the tendon during impact. The
patella was brushed with India ink and photographed prior to impact to document any
baseline surface roughness or fibrillation.

A 4.5 kg impact mass was accelerated to a given pre-impact velocity, while aligned
on two steel guide rails and supported on four nylon blocks to minimize fiiction. A 15.3
cm diameter, 6061-T6 aluminum impact interface was mounted to the fi'ont of a load
transducer (Model 3173-2k, Lebow Products, Troy, MI, USA). While a rigid impact
interface does not model the energy absorbing characteristics of an instrument panel, this
surface provided baseline results that could be compared to previous studies (Haut 1989,
Powell 1975). The impact mass was selected based on a previous study (Haut 1989).
The resonant frequency of the load cell with the impact interface attached was 1333 Hz.
Pilot studies indicated that the load transducer and attached mass would not resonate

during impact on the human knee. Load data were recorded continuously on a personal

46

computer via a 16 bit A/D board (Model DAS 1600, Computer Boards, lnc., Mansfield,
MA, USA). Load data were inertially compensated (Beckman 1969) and sampled at
10,000 Hz.. The impacting mass was accelerated to a predetermined velocity by adjusting
pressure in a mass accelerator tank. Velocity of the mass was measured prior to impact
on the knee using two infiared optical sensors (Part No. OR518-ND, Digi-Key
Corporation, Thief River Falls, MN). These sensors are designed with a 5 rmn separation
between the emitter and detector. A tab connected to the impact mass crossed the 5 rrrrn
gap interrupting the first sensor and then the second. The individual sensors were
separated by 76.2 rrrrn to eliminate cross-talk. The test protocol consisted of sequentially
increasing the velocity of the impact mass, so as to progressively increase the impact
energy (3, 11, 22, 38, 61 J). The experiment was terminated when impact loading resulted
in a visible fi'acture of bone. Damage was assessed by visual inspection through lateral
and medial incisions in the joint, palpation and documented photographically. The
contralateral limb of each specimen was impacted with 45% of the energy that caused
bone fi'acture in the first limb.

A medium range (7-40 MPa) pressure sensitive film (Prescale, single sheet, Fuji
Film, Ltd., Tokyo, Japan) was inserted into the joint prior to each impact through the
medial and lateral incisions. The film was encased in two sheets of polyethylene film
approximately 0.05 mm thick to prevent exposure to body fluids and help reduce shear
loading artifacts (Huberti 1984). The strips of film were sufficiently long to pass through
the patellofemoral joint and cover the anterior surface of the patella to measure the

location and distribution of the contact load on the knee. This also minimized fiictional

47

forces between the rigid impactor and the knee. Prior to impact, one mm diameter holes
were drilled at the proximal and distal poles through the patella in an anterior to posterior
direction. After impact, needle probes were inserted through the holes to pierce the film
and help locate position of the film with respect to the patellar facets. l

The pressure sensitive film was calibrated in a servohydraulic testing machine
(Model 1330, Instron, Canton, MA, USA) using a previous methodology (Haut 1989).
The film showed the highest sensitivity over the range of approximately 8 to 30 MPa
(Haut 1991). Briefly, the film was encased in the polyethylene packet and loaded with a
100 ms load-controlled haversine pulse. The fihns were analyzed within one day using a
video scanner (Model MSF-3OOZ, MicroTek International Incorporated, Hsinchu, Taiwan,
ROC) with a sensitivity of 11.8 pixels per nnnz. The analysis of the exposed film was
performed with Image 1.44 public domain software (National Institutes of Health, ftp
Wayne@helix.nih. gov). The distribution of contact pressures and contact areas were
determined within the patella-femoral joint. The image was divided to analyze the medial
and lateral pressures and areas. A characteristic vertex present in most images
corresponded to the median patellar ridge separating the image into medial and lateral
facets (Figure 1). Technical difficulties during scanning the film from specimen H03
resulted in no pressure or area data from that specimen.

Following the mechanical test, or in the case of controls following harvest of the
tissues, patellae were placed in 10% buffered formalin for approximately seven days. The
tissue was decalcified in 20% formic acid for approximately 14 days. Due to the size of
the human patella, only one 2-3 mm block was excised between the pin holes and used in

this study (Figure 2). Blocks were cut medial to lateral across the entire patella. The

48

blocks were processed for routine paraffin embedding. Eight micron thick sections were
cut (33), stained with Safranin O-Fast Green and examined in light microscopy at 12-400
power.

The experimental data resulting from analysis of the load-time curve and the
pressure sensitive film were analyzed with a commercial statistics package (Excel 5.0,
Microsoft Corporation, Redmond, WA) to calculate means and standard deviations. The
peak contact load, average contact pressure, average medial pressure, average lateral
pressure, contact area, time to reach peak load and contact duration (pulse) were the
dependent variables. Both linear and second-order polynomial functions were used to
identify the fimctional dependence of each dependent variable with impact energy.
Correlation coefficients were computed for each fit.

RESULTS

A typical impact on the knee for the impacts leading to fracture resulted in a short
period of low contact load followed by a sharp rise to a peak in 5.3fi3.1 ms. The contact
load declined gradually yielding a total contact duration of 16.1rT6.2 ms. These data did
not vary significantly with input energy. A second order polynomial was used to correlate
peak contact force and pressure in successive impacts leading to fracture for impact

energies of 3, 11, 22, 38 and 61 J. Peak contact force increased with increasing energy
(r2=.82) delivered to the knee (Figure 3). .The average patellofemoral contact pressures

increased with increasing energy (r2=.85) delivered to the knee (Figure 4). The pressure
over the lateral facet was consistently higher than on the medial facet, but no statistical

differences were documented. Contact pressures were uniform (standard deviations less

49

than 2 MPa) over the facets (Figure l). The lateral contact area was greater than the
medial area.

At approximately 60125 J of energy and a contact force of 6.7121 kN
visible failure of bone was noted (Table 2). Four specimens suffered transverse fractures
of the patella, two specimens suffered comminuted fi'actures of the patella, and one had a

significant disruption of articular cartilage and underlying bone. The average

patellofemoral contact pressure and contact area was 15 .5il.6 MPa and 677:1:177 mmz,
respectively, in these fracture-producing experiments.

The average input energy for the subfracture experiments was 26fi10 J, which was
43.4% of the energy for the fracture-producing group. This level of input energy resulted
in a contact force of 5.2:t2.1 kN (Table 2). Gross evaluation of the specimens revealed no

visible bone fiacture in five of seven cases. The average contact pressure and contact area

were 11.8i5.5 MPa and 461i] 11 mmz, respectively for the sub-fracture experiments.
Gross evaluation of the fi'acture and subfracture knee joints prior to testing
revealed baseline pathology in the form of varying levels of chondromalacia in all patellae.
This generally included local softening, mild surface roughness and surface fibrillation of
the articular cartilage. No specimens exhibited denuded bone in the contact areas.
Analysis of the patellar histology from the subfracture experiments revealed occult
injuries in four of seven patellae. The injury was visible at 12 power and was observed in
the deep zones of the articular cartilage, the zones of calcified cartilage and the
subchondral bone. The traumas were observed as horizontal (parallel to the articulating

surface) occult cracks approximately one-third the width of the patella, located under the

50

lateral facet. The characteristic shape was that of an open slit. The cracks were identified
as two types. The first type involved a split confined to the subchondral bone (Figure 5)
and was observed in three specimens. The entire inner surfaces of the split were jagged.
The surface of the crack involved a thin layer of subchondral bone still firlly intact along
the calcified cartilage. In two of these specimens the occult traumas occurred adjacent to
areas of surface fibrillation (Figure 6). The fibrillation appeared as a rough, uneven
surface across the articular cartilage. The second type of trauma, observed in one
specimen, was a split at the subchondral bone and calcified cartilage interface. The edge
of the crack along the calcified cartilage was smooth, while the other edge along the
subchondral bone was jagged.

Analysis of histological slides fi'om the fractured specimens also revealed occult
trauma in four of seven patellae. The characteristic shape and size of these occult traumas
were similar in shape and size to the traumas observed in the subfiacture experiments.
The cracks were again classified as two types. The first type involved, as in the
subfiacture experiments, cracks confined to the subchondral bone. This type of trauma
was observed in two specimens. They were contralateral patellae from the subfracture
group with the same occult trauma. One of these fractured specimens (H07) had an occult
trauma adjacent to minor surface pathology. The other specimen (H08) suffered a
transverse fracture of the patella contiguous with the pin hole used to locate the film.
Each of these two specimens showed the same type of occult trauma in the contralateral
patellae after subfiacture loading. The second type of trauma involved a crack at the

tidemark and was noted in the other two specimens. Both edges of this crack appeared

51

smooth. Occult traumas occurred directly beneath surface fibrillation for both specimens.
One of these specimens (H03) had an occult split of the subchondral bone in the
contralateral subfiacture experiment. The other specimen did not appear to have an occult
trauma in the contralateral subfracture experiment.

Because the occult fiactures were observed after subfiacture and fi'acture
experiments on four specimens, we considered the possibility that these damages were not
impact induced, but rather were pre-existent in the joint. To help ensure that these occult
damages were experimentally induced, seven tmimpacted patellae from four subjects were
histologically processed. They were viewed in light microscopy at a range of powers
similar to the impacted patellae. Although baseline pathologies existed in the form of
articular surface fibrillation, multiplication of tidemark and loss of proteoglycan staining,
no horizontal occult traumas were observed. The cartilage, calcified cartilage and
subchondral bone appeared intact with no horizontal splits.

DISCUSSION

The primary objective of this study was to examine the human knee for occult
trauma following blunt insult which did not produce visible fracture of bone. A second
objective was a comparison of the peak fracture and subfracture contact loads with the
bone fiacture criterion of 10 kN. A final objective was to determine if microcracks
occurred at contact pressures less than 25 MPa.

Our histological findings revealed horizontal, occult fi'actures involving the
subchondral bone, calcified cartilage and articular cartilage at subfracture loads averaging

5.2 kN. This load was 22% less than that required to produce visible fracture of bone in

52

the contralateral limb. These subfracture loads were also 48% less than the current lower
extremity injury criterion for new automobile certification using anthropomorphic
dummies. The location of the occult traumas was predominantly on the lateral facet for
pressures less than 25 MPa. To our knowledge, this is the only study to document
controlled, experimentally induced, occult rrricrocracks at the cartilage-bone interface in
the human knee. Previous studies impacted the human cadaver knee with similar
experimental methods, but microtrauma of the joint tissues was not investigated.

A limitation of the current study was the location of histological sectioning.
Excision of the 3 rrnn block fiom the patella, between the pin holes, was the basis of a
mathematical model predicting maximal interface shear stresses on the lateral facet (Shelp
1994). While this sectioning method resulted in the observation of occult traumas in four
of seven cases in both fracture and subfiacture experiments, other sections may have
shown occult damage in the remaining 3 cases. In addition, it is unknown at which energy
or contact load occult damages actually occurred during the fiacture experiments. The
study was also limited to a rigid interface condition. While this type of interface was used
by Powell (1975), whose data largely serves as the basis for the current injury criterion,
automotive occupants commonly interact with more deformable structures. During this
more typical situation some portion of the contact load might be supported by structures
surrounding the patella, such as the femoral condyles (Haut 1989). Furthermore, it is
known that for more rigid impacts, femur loads in the current anthropomorphic dummy
typically exceed those measured in a human cadaver under similar impact conditions

(Hering 1977). Thus, while occult traumas have been reported in cadavers for loads less

53

than the current injury criterion, the direct applicability of these data to the automotive
setting with anthropomorphic dummies needs firrther clarification.

Comparison of the peak contact loads at fracture with the bone fracture criterion
reveals a difference of approximately 3 kN. Since this injury criterion was based on
studies using seated cadavers, those impact loads also resulted in inertial acceleration of
the specimen (Patrick 1965, Powell 1975, Melvin 1975). More energy is being used to
actually deform tissues in the current isolated joint experiments. On the other hand, a
previous study with older cadavers (72:11 years) utilized the same experimental setup as
the current study, and reported a fracture load of 8.5i3.0 kN that often resulted in split
fiacture of the femoral condyles (Haut 1989). While no absolute explanation can be given
for the lower fi'acture loads in the current study, one possible factor could be the influence
of small pin holes inserted into the patella to help locate the pressure sensitive film. These
holes may have produced stress concentrations and local weaknesses in the patella. Yet,
in only one case was fracture actually contiguous with a pin hole. Another possible
explanation may be related to the relatively good quality, thickness and overall distribution
of articular cartilage in the patellofemoral joint of the younger specimens. Haut (1989)
and most previous studies, use specimens 70 years and older. Though not documented in
all studies, with this advance in age a high degree of joint pathology would be expected,
possibly involving areas of denuded bone (Haut 1989). The absence of cartilage in the
joint can result in the transmission of high loads to the femur yielding higher shear stresses

in the femur to possibly cause split fiactures of the condyles. Altemately, the presence of

54

as

6
L'

13'

E
o.&

a soft layer of cartilage in the young joint can yield high localized shear stresses in the
patella that could yield more patellar fractures.

Tomatsu et al. (1992) recently reported horizontal split fractures comparable to the
current study during shear loading of the porcine femoral condyle. The horizontal
microcracks we observed closely resemble his IB and 11B injury classifications. That study
suggests the location of trauma is dependent on the mass and velocity of the impactor.
Tomatsu et al. conclude that high impact speeds and light masses tend to rupture the
cartilage, resulting in open lesions. Low impact speeds and high masses more often
generate occult damage in the underlying bone. As the speed is increased, combined
injury to the bone and cartilage occurs. The higher speeds used in our fracture group also
resulted in visible fractures of bone and gross disruption of the articular cartilage. It is
possible the occult traumas recorded in the fractured specimens may have also been
produced in one of the subfiacture impacts at a lower velocity, and therefore, lower
energy. The results of Tomatsu et al. suggest that the occult microfractures observed in
the current study could be largely due to our use of a relatively large mass and a
corresponding low velocity of impact. Since impact trauma during an automobile collision
can occur at speeds of 30 mph or more, it may be questioned whether these occult
traumas would be generated. Clinical studies of joints after high energy insult, however,
have suggested the possible presence of microfiactures at the tidemark or in subchondral
bone (V ellet 1991).

While many of these microcracks were confined to the subchondral bone, we also

Observed cracks at the tidemark. Earlier studies investigating mechanisms of human

55

osteoarthrosis suggest microdamage at the interface between hyaline and calcified
cartilage is due to excessive shear stresses (Meachim 1978, Minns 1979). Armstrong et
al. (1985) suggested that a separation of cartilage and bone following high energy loading
of the porcine patellofemoral joint is due to high shear stresses. A recent theoretical
model studied the contact between two biphasic cartilage layers and shows that sudden
loading results in high shear stress at the cartilage-bone interface (Ateshian 1994). Our
recent mathematical analysis of the human patellofemoral joint also indicates high levels of
shear stress are developed in bone underlying the lateral patellar facet during impact
loading (Haut 1994). This being the most common area of occult microfi'acture suggests
that excessiVe shear stresses may be the major factor causing impact injuries in the current
study. Interestingly, our studies indicate that these occult microfractures occur under and
adjacent to existing surface pathology. Since other studies indicate that bone seems to
remodel before histological changes occur in the overlying cartilage (Radin 1973, Shimizu
1993), perhaps these specimens were predisposed to microcracks because of a weakened
subchondral bone plate. On the other hand, a weakened and fibrillated layer of cartilage
could also expose the underlying subchondral bone to localized areas of high shear stress
during the blunt insult.

Repo and Finlay (1977) document visible damage to human articular cartilage and
death of chondrocytes at 25 MPa of contact pressure. Impact studies on the canine report
no damage to the surface tangential zone or calcified cartilage at pressures of 17 MPa
(Chin 1986). Thompson (1991) suggests damage to the canine patellar articular surface

occurs at pressures greater than 25 MPa. Vener et al. (1992) indicates 40 MPa is a

56

threshold pressure for damage in the canine phalangeal joint. In contrast, contact
pressures of 15-17 MPa resulted in gross fi'acture of bone in the current study. The
contrasting results fiom human and animal studies may be due to anatomical differences
between joints, or possible differences in material properties of various joint tissues.

Fuji pressure sensitive film was used in the above mentioned studies and a number
of others (Fukubayashi 1980, Haut 1989, Huberti 1984) to record intra-articular contact
pressures. The accuracy of the film in transducing pressure gradients is recently reported
to be 10 MPa/mm for the more commonly used dual layer type of film (Hale 1992).
Since, pressures in the current study predominantly ranged from 9 to 22 MPa for the
fi'acture and subfi'acture impacts, the film functioned in its most sensitive range. Dynamic
calibration of the pressure-sensitive film has been performed in the rabbit knee by the
product of contact pressure and patellofemoral contact area (Haut 1991). A similar
calculation in the current study was difficult because the human patellar facets are more
complex and largely inclined with respect to the line of the impacting force (Haut 1989).

Finally, recent studies have focused on identifying occult bone trauma and relating
it to clinical findings. "Radiographically occult injuries to the bone, otherwise referred to
as occult fiactures, bone bruises, or rnicrotrabecular trauma/fracture, may account for
patient pain" (Kapelov 1993), perhaps by impeding venous drainage through extraosseous
veins (Arnoldi 1991). Complaints of joint pain immediately following insult to a joint has
been reported in approximately 88% of patients eventually progressing to osteoarthrosis
(Chapchal 1978). Engebretsen et al. (1993) suggests injury to the anterior cruciate

ligament often results in "bone bruises" that can be seen on MRI and provide a basis for

57

joint pain and a subsequent arthrosis. It is tempting to speculate that split fractures of
subchondral bone might be associated with post-traumatic knee pain with a corresponding
appearance of a "bone bruise" by MRI.
ACKNOWLEDGMENT

This study was supported by a grant (R49/CCR503607) from the Centers for
Disease Control and Prevention. 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 Ms. Jane Walsh for help in the
interpretation of histological slides, Mr. Cliff Beckett for technical support, and Mr.
Richard Dalirnonte for his assistance with the experiments. We are also indebted to Mr.
Rick Shumaker and the Michigan Tissue Bank for solicitation of donors and retrieval of
joints.
REFERENCES
1. Armstrong C, Mow V, Wirth C: Biomechanics of impact-induced rnicrodamage to
articular cartilage: a possible genesis for chondromalacia patella. In: Amer Acad of Orth
Surg Symp on Sports Med, ed by G Fenerman. pp70-84, 1985

2. Amoldi C: Patellar Pain. Acta Orthop Scand 62:1-29, 1991

3. Ateshian GA, Lai WM, Zhu WB, Mow VC: An asymptotic solution for the contact of
two biphasic cartilage layers. J Biomechanics 27(1 1):1347-1360, 1994

4. Beckrnan DC, Palmer MF: Thoracic force-deflection studies in primates. Proc 3rd
Biomech Factors Con]: ASME, Ann Arbor, MI Paper 69-BHF-12: 1969

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

6. Chin M, Donahue J, Erdman A, Oegema T, Thompson R: Biomechanical analysis of an
adult canine patella under an indirect blunt trauma. Trans Orthop Res Soc 1 1:232, 1986

58

7. Davis M, Ettinger W, Neuhaus J, Cho S, Hauck W: The association of knee injury and
obesity with unilateral and bilateral osteoarthritis of the knee. American J Epidem
l30(2):278-288, 1989

8. Dischinger P, Cushing B, Kems T: Lower extremity fractures in motor vehicle
collisions: influence of direction of impact and seatbelt use. Proc 36th Conf AAAM
Portland, OR pp3l9-326, 1992

9. Donohue MJ, Buss D, Oegema T, Thompson R: The effects of indirect blunt trauma
on adult canine articular cartilage. J Bone Joint Surgery 65-A(7):948-957, 1983

10. Engebretsen L, Arendt E, Fritts HM: Osteochondral lesions and cruciate ligament
injuries. Acta Orthop Scand 64:434-436, 1993

11. Fukubayashi T, Kurosawa H: The contact area and pressure distribution pattern of the
knee. Acta Orthop Scand 51:871-879, 1980

12. Hale JE, Brown TD: Contact stress gradient detection limits of pressensor film. J
Biomech Eng 114:352-358, 1992

13. Haut R, Dalimonte R, Li X, Altiero N: Mechanical damage in the human P-F joint
after blunt impact loading. Proc Second World ConfBiomech 2:297, 1994

14. Haut R: Contact pressures in the patella-femoral joint during impact loading on the
human flexed knee. J Orthop Res 7:272-280, 1989

15. Haut RC, Ide TM, DeCamp CW: Irnpact-induced damage to articular cartilage of the
patella-femoral joint. First Injury Through Biomech Symp] 3-26, 1991

16. Hering WE, Patrick LM: Response comparisons of the human cadaver knee and a part
572 dummy knee to impacts by crushable materials. Twenty-First Stapp Car Crash Conf
21:1015-1053, 1977

17. Huberti H, Hayes W: Patellofemoral contact pressure. J Bone Joint Surg 66-A(5):715-
724, 1984

18. Johnson-Nurse C, Dandy D: Fracture-separation of articular cartilage in the adult
knee. J Bone Joint Surg 67-B( 10:42-43, 1985

19. Kapelov R, Teresi L, Bradley WG, Bucciarelli NR, Murakarni DM, Mullin WJ, Jordan
JE: Bone contusions of the knee; increased lesion detection with fast spin-echo MR

imaging with spectroscope fat saturation. Radiology 901-904, 1993

20. Kellgren J, Lawrence J: Osteoarthritis and disk degeneration in the urban population.
Annals Rheum Diseases 17:386-396, 1958

59

21. Meachim G, Bentley G: Horizontal splitting in patellar articular cartilage. Arthr Rheum
21 :669-674, 1978

22. Melvin J, Stalnaker R, Alem N, Benson J, Mohan D: Impact response and tolerance of
the lower extremities. Nineteenth Stapp Car Crash Conf 19:543-559, 1975

23. Minns R, Birnie A, Abemethy P: A stress analysis of the patella and how it relates to
patellar articular cartilage lesions. J Biomech 12:699-71 1, 1979

24. Nyquist G, King A: Lower extremities. Rev Biomech Impact Res Inj Auto Env 6: 163-
201, 1985

25. Patrick L, Kroell C, Mertz H: Forces on the human body in simulated Crashes. Ninth
Stapp Car Crash Conf9:237-259, 1965

26. Powell W, Ojala S, Advani S, Martin R: Cadaver femur responses to longitudinal
impacts. Nineteenth Stapp Car Crash Conf 19:561-579, 1975

27. Radin B, Parker H, Pugh J, Steinberg R, Paul 1, Rose R: Response of joints to impact
loading-III relationship between trabecular microfiactures and cartilage degeneration. J
Biomech 6:51-57, 1973

28. Repo R, Finlay J: Survival of articular cartilage after controlled impact. J Bone Joint
Surg 59-A(8): 1068-1076, 1977

29. Rice D, MacKenzie E, et al : Cost of Injury in the United States: A Report to
Congress 1989. Atlanta, GA, 1989

30. Shelp D, Newberry W, Li X, Dalimonte R, DeCamp C, Altiero N, Haut R: Tissue
damage resulting from blunt impact on the knee: animal and cadaver studies. 4th Inj
Prevention Through Biomech Symp 4:43-62, 1994

31. Shimizu M, Tsuji H, Matsui H, Katoh Y, Sano A: Morphometric analysis of
subchondral bone of the tibial condyle in osteoarthrosis. Clin Orthop 229-239, 1993

32. States J: Traumatic arthritis--a medical and legal dilemma. Ann Conf of the Amer
Assoc for Automotive Med 14:21-28, 1970

33. Sterchi DL, Eurell JC: A new method for preparation of undecalcified bone sections.
Stain Tech 64:201-205, 1989

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

35. To

incur

mm

ml“

M _iL L IF it .ngilbi giflglifiuiiuiic

35. Tomatsu T, Irnai N, Takeuchi N, Takahashi K, Kimura N: Experimentally produced
fractures of articular cartilage and bone. J Bone Joint Surg 3:45 7-46 1 , 1992

36. Vellet AD, Marks PH, Fowler PJ, Munro TG: Occult post-traumatic osteochondral
lesions of the knee: prevalence, classification, and short-term sequelae evaluated with MR
imaging. Radiology 178(1):271-276, 1991

37. Vener MJ, Thompson RC, Lewis JL, Oegema TR: Subchondral damage after acute
transarticular loading: an in vitro model of joint injury. J Orthop Res 10:759-765, 1992

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLES
Table 1. Donor profiles
Specimen # Sex Age Mass (kg) Height (m) Cause of Death
Test Specimens
H02 F 39 79.5 1.6 colon cancer
H03 M 39 44.5 1.7 renal failrn'e
H04 M 49 95.0 1.8 colon cancer
H05 M 52 81 .8 1.8 myocardial infarction
H06 M 59 88.6 1.8 myocardial infarction
H07 M 42 75.0 1.8 myocardial infarction
H08 M 34 68.2 1.8 myocardial infarction
Control Specimens

H13 F 27 71.3 1.7 suicide
H14 M 71 95.5 1.9 cardiac arrest
H15 M 42 - - -
H16 F 21 - - -

 

 

 

 

 

 

 

 

61

Table 2. Fracture and subfiacture producin experimental data.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Spec. Peak Impact Average Gross Visual Occult
contact Energ patella-femoral Damage Trauma
force y contact
(kN) (J) pressure (MPa)
Fracture producing experimental data.
H02 4.9 24 12.4 Transverse fiacture of No observable
the patella damage
H03 3.2 45 - Comminuted fiacture Split at
of the femur tidemark
H04 9.4 90 18.4 Gross disruption of Split at
cartilage tidemark
H05 7.7 64 15.2 Transverse fi'acture of No observable
the patella damage
H06 6.6 43 12.7 Transverse fiacture of No observable
the patella damage
H07 8.1 92 19.4 Comminuted fi'acture Split in
of the patella subchondral
bone
H08 6.8 65 21.3 Transverse fiacture of Split in
the patella subchondral
bone
Subfi'acture-producin experimental data.
H02 3.9 10 9.4 Transverse fracture of No observable
patella damage
H03 4.8 31 - Failure of lateral Split in
condyle subchondral
bone
H04 4.8 30 9.7 No gross visual No observable
damage damage
H05 5.6 27 12.7 No gross visual No observable
damage damage
H06 3.4 18 4.4 No gross visual Split at
damage calcified
cartilage/subch
ondral bone
H07 8.2 41 13.6 No gross visual Split in
damage subchondral
bone
H08 5.7 27 20.8 No gross visual Split in
damage subchondral
bone

 

62

 

FIGURES

 

Figure l: Successive film imprints of the patellofemoral joint resulting fi'om increasing
load delivered across the joint. The final imprint coincides with a fracture of the patella.
The arrow indicates a characteristic vertex which corresponded to the median ridge and
was used to separate the imprint into median and lateral halves.

63

 

Figure 2: A fi'actured patella with the superimposed image of the pressure distribution as
recorded by the pressure sensitive film (m=media1, l=lateral). The horizontal lines pass
through areas of high pressure adjacent to the fracture and indicate where sections were
taken for histology. The open arrows indicate two areas of impact induced trauma. The
solid arrow indicates the location of an occult split in the subchondral bone. The two
small dots represent the location of the 1 mm pin holes used for location of the pressure
sensitive fihn.

 

 

_L

l
l

 

l
o

i
l

i

I

o 1
. l

i

i

\

 

 

 

l
o

 

 

 

 

Peak Contact Force (RN)
0 -* N (to A 01 0') \1 on (D O

 

 

 

 

 

-- . ‘ F = -0.002E2 + 0.2482E + 0.6296
0 10 20 30 40 50 60 70
Energy (J)

 

Figure 3: Experimental impact load data from successive impacts leading to an observed
fiacture of bone. Impacts were delivered at increasing energy levels; the boxes indicate
the range of data for each energy level. Not all specimens required the entire impact series
to fiacture. The second-order polynomial fit of the data is shown.

 

 

    
 

 

 

 

25
a 20 -
n.
E
0 15 -
‘5
g 10 -
g _ -0.0065E2 + 0.6991E - 1.0181
5 _
O l 1 T 1 I i

0 10 20 30 40 50 60 70
Energy (J)

 

 

 

Figure 4: Experimental data from successive impacts leading to an observed fi'acture of
bone. The pressure has been averaged over the patellofemoral contact area. Impacts
were delivered at increasing energy levels; the boxes indicate the range of data for each
energy level. Not all specimens required the entire impact series to fracture. The second-
order polynomial fit of the data is shown.

65

 

Figure 5: Transverse section of the patellae (hematoxylin and eosin 40X) showing a
representative occult split fracture confined to the subchondral bone.

 

Figure 6: Transverse section (hematoxylin and eosin 25X) of the lateral patellar facet
showing proximity of surface pathology (fibrillation) to an occult trauma.

66

CHAPTER 5

PATELLOFEMORAL JOINT FRACTURE LOAD PREDICTION
USING PHYSICAL AND PATHOLOGICAL PARAMETERS

Patrick J. Atkinson, Charles M. Mackenzie“ and Roger C. Haut
Department of Materials Science and Mechanics, College of Engineering,
Orthopaedic Biomechanics Laboratories, College of Osteopathic Medicine
*College of Veterinary Medicine
Michigan State University

ABSTRACT

Lower extremity (knee) injury prediction resulting fiom impact trauma is currently
based on a bone fracture criterion derived fiom experiments on predominantly aged
cadavers. Subsequent experimental and theoretical studies indicate that more aged,
pathological specimens require higher, not lower, loads to initiate bone fiacture. This
suggests that a bone fracture criterion based solely on aged specimens may not be
representative of the current driving population. In the current study, we sought to
determine if cadaver age, physical size, sex, baseline joint pathology, or patellar geometry
correlated with fiacture load. An analysis was made of data fiom previous impact
experiments conducted on fifteen isolated cadaver knees using a consistent impact
protocol. The protocol consisted of sequentially increasing the impact energy with a rigid
interface until gross fiacture. Gross bone fractures occurred at loads of 6.9i2.0 kN (range
3.2 to 10.6 kN) using this protocol. Regression analyses revealed that fracture load was

predicted by only one parameter: patellar geometry. Altemately, we developed a 2-D

67

mathematical model of the human knee to explore parameters that might influence the
loads required to cause gross bone fracture. In support of our recent experimental studies
using rigid and padded impact interfaces, the model suggested that load intensity and it’s
distribution over the knee play a role in defining the fracture load as well as the site
(patella or femur) of patellofemoral joint injury.
INTRODUCTION

Numerous studies have described the in vivo and in vitro response of human and
animal joints to impact loading. Such data are necessary to understand joint mechanics
during dynamic loading and to establish injury criteria. In particular, whole human
cadaver studies were conducted in the 1960’s and 70’s to establish a lower extremity
injury criterion to address bone fracture injury sustained during loading through the knee
in automotive accidents (Patrick 1965, Melvin 1975, Powell 1975). The average age of
the specimens was 64 years, perhaps, due in part to the availability of the more aged
subject and because these early studies hypothesized that the more aged cadaver would
result in a more conservative benchmark for lower extremity injury. Indeed, Patrick, et
al., note: “The fact that all of the subjects used exceeded fifty years of age and that all
were ernbalmed should render the foregoing data somewhat conservative...” (Patrick
1965). However, more recent studies suggest that some aged cadavers require higher
loads to cause bone fracture (Atkinsonza 1995, Haut 1989). Isolated knee joints from a
group of ‘aged’ (72 yrs) versus ‘young’ (46 yrs) cadavers require 8.5 kN and 6.7 kN,
respectively, to initiate gross bone fracture. In fact, the aged group typically suffered split
femoral condylar fractures, while patellar fractures were more common in the younger

Specimens. The difference in injury modalities might be explained by the overall health of

the joints. The study involving aged cadavers documented advanced joint pathology in the
test subjects involving areas of denuded bone. This likely resulted in increased stress
transmission across the joint leading and may explain the femoral fractures (Atkinson:b
I995). The study involving the younger group of cadavers documented a complete and
healthy distribution of cartilage in the patellofemoral joint. This scenario likely results in
elevated stresses in the patella, and might explain the higher fiequency of patellar fractures
in younger cadavers. These differences suggest that age and/or the pathological state of
the joint might influence fracture patterns in this isolated joint model, and that a criterion
based solely on aged specimens may not be representative of the current driving
population.

In the current study, an analysis of data was performed of blunt insult experiments
on isolated human cadaver knees from a prior study. Additionally, specimen age, physical
sizes, patellar geometry and degree of joint pathology on fracture load was investigated
using regression methods. Because a previous study (Atkinsonza 1997) suggests that
contact area and peak load might influence knee joint fi'acture, theoretical parametric
analyses were performed on a finite element model of the patellofemoral joint in an effort
to more completely understand parameters that may influence fiacture of this joint.
METHODS AND MATERIALS
Statistical Regression Analyses

The current investigation was an analysis of a previous study (Atkinson:b 1995).
In that study, blunt impacts were delivered to isolated, flexed knee joints of fifteen
cadavers with the following physical characteristics: 12 males and 3 females aged

58.5:t14.7 years old, weighing 78.6;t21.9 kg and a height of 1.77:0.14 meters. The

69

specimens were procured by a tissue bank shortly after death fiom donors with no prior
llistory of joint trauma and presenting a wide range of joint health: from no gross evidence
of j oint disease to advanced joint pathology, including areas of denuded bone, advanced
chondromalacia, etc. The impact protocol has been previously described (Atkinson 1995,
Haut 1989). Briefly, the patellofemoral joint was examined through a medial incision.
Joint preparations were excised approximately 15 cm fi'om the knee proper. The femur
Was potted in a cylindrical steel sleeve and attached to a rigid fi'ame such that the long axis
0f tlle femur was oriented horizontally. Each specimen was maintained at a flexion angle
0f 90° prior to impact with a loosely knotted tether tied to the quadriceps tendon. A 4.9
kg inlpact mass was pneumatically accelerated to a given pre-impact velocity. A 15.3 cm
dialrleter, 6061-T6 aluminum impact interface was mounted to the front of a load
transducer (Model 3173-2k, Lebow Products, Troy, MI, USA) attached to the impact
11"£188. The resonant frequency of the load cell with the impact interface attached was
approximately 1300 Hz. Load data was inertially compensated (Atkinson 1997) and
recorded on a personal computer via a 16 bit A/D board at a sampling frequency of 20
1{1‘12 (Model DAS 1600, Computer Boards, Inc., Mansfield, MA, USA). Velocity of the
It138$ was measured prior to impact on the knee using two infi'ared optical sensors (Part
NO- OR518-ND, Digi-Key Corporation, Thief River Falls, MN). The test protocol, which
has been in used in numerous studies by our laboratory (Atkinsonza 1995, Atkinson:b
1 995, Haut 1989) consisted of sequentially increasing the velocity of the impact mass in 1
“3118 increments starting at 1 m/s, so as to progressively increase the impact energy (2.5,

1 O, 22, 38, 61, 88 J). The experiment was terminated when impact loading resulted in a

70

visible fiacture of bone. Damage was assessed by visual inspection through lateral and
medial incisions in the joint, palpation and documented photographically.

Each subject's patellar geometry was also quantified on the unimpacted,
confialateral knee using a previously described technique (Grelsarner 1994). Briefly, the
length of the bone and articular cartilage surface were measured along a mid-sagittal plane

(F i gave 1). The ratio of bone to cartilage length (the morphology ratio) was then

Computed.

Parametric Study of Load and Contact Area

 

To investigate some of the potential effects of load intensity and distribution in
determining stress states in various knee joint tissues, a parametric study was performed
using a plane strain finite element model of the patellofemoral joint (Garcia 1997). The
I11<><1el (Figure 2) geometry was taken from a transverse slice taken through a 90 degree
flexed human cadaver knee. Due to the short duration of impact loading in which there is
iIlsuriicienr time for fluid flow in the biphasic cartilage, the cartilage was considered to be
111 incompressible elastic isotropic material (Eberhardt 1990). A quasistatic analysis was
ac1<>pted due to the negligible effect of tensile stress wave transmission through the knee
jQi-Ilt for short duration loading (Anderson 1991). The articular cartilage, subchondral
ane, and trabecular bone were modeled with isopararnetric elements. Values for
?Omg’s modulus and Poisson’s ratio were taken from the literature (Atkinson:b 1997,

IVicente 1994, Repo 1977, Townsend 1975): 20 MPa and 0.49 for cartilage, 3750 MPa and
0 ~ 30 for subchondral bone, and 300 MPa and 0.30 for trabecular bone.
Pilot studies determined that 45 MPa of pressure applied to the anterior of the

Patella was sufficient to reach referenced bone failure stresses in the human knee joint

71

(Garcia 1997). Areas of load contact on the patella were 25, 40, 55, 65, 75, and 80% of
fine available contact area for the patella in this study (Figure 2). The contact problem was
solved assuming finite displacement and the iterative solutions were written to an output
file for post-processing. These solutions were reviewed to determine which load step
coincided with documented failure stresses of the subchondral plate and articular cartilage
Of flue patella and femoral condyles (Garcia 1997). The failure stresses for the subchondral
pl ate were taken to be 42 MPa and 78 MPa, respectively, in tension and shear. Both
Stl‘esses were considered because injury may result from high levels of either stress
(Atkinsonza 1997). Failure of the articular surface has been documented to be mediated
“"1081: significantly by high levels shear stress (Atkinson:b 1997). Thus, articular cartilage
failllre was based on a shear stress of 6 MPa. The failure load in the plane of the model
was calculated by integrating the applied boundary condition. To estimate the
Corresponding failure load on a whole lmee, the calculated failure load from the 2-D model
was multiplied by 1t*a 2 (a is the half-width of contact). This procedure was adopted
be‘-'-‘-£Iuse previous studies by our group have shown that blunt insult to the patella results in
an approximately circular area of contact.

To quantify the health of patellar bone and cartilage histological sections were
plTepared. Following the mechanical test, the patellae were placed in 10% buffered
fQTl‘Inalin for approximately seven days and decalcified in 20% formic acid for
aI>proximately 28 days. One 2-3 mm block was cut medial to lateral across the entire
Patella in an area of high contact pressure adjacent to cartilage and bone trauma
(Atkinsonza 1995)). The blocks were processed for routine paraffin embedding. Eight

micron thick sections were cut and stained with Safianin-O and examined under light

72

microscopy. Patellar sections were scored in a randomized, blinded fashion by one of the

autlrlors (C.M.) using a modified Mankin scale (Mankin 1971). The cartilage was scored

for: structure (possible score of 0-4), cellular appearance (0-6) and tidemark integrity (0-

1 ) and was modified to include a three point pathological measure of the subchondral

plate: 0=normal, l=moderate changes, 2=advanced pathological changes. The total

possible combined histopathological score was 13, where lower numbers indicate healthy

tissue.

Subject age, sex, mass, height, histopathological score, fracture load and

II'1<>I‘1'Jhology ratio were analyzed using linear regression (SigmaStat, Jandel Corp., San

1{afznel CA) to identify correlations in the data. Specifically, fracture load,

hist<>pathological score and morphology ratio were first analyzed by conducting separate
urlivariate regression analyses versus each remaining predictor variable (Table 2).
Mmtiple linear regression analyses were conducted by using all possible combinations of
ale predictor variables in stepwise fashion to predict fiacture load, histopathological score
a‘thi morphology ratio. Subject age, mass, height, histopathological score, fi'acture load
a‘l‘ltl morphology ratio data were then separated by sex and compared with the unpaired

S‘JJdent's t-test to determine if these data varied by sex. Statistical significance was set at
IXxoos.

RESULTS

Typical impacts causing fracture resulted in a short period of low contact load
lelowed by a sharp rise to a peak in 5:26 ms (Figure 3). The contact load declined
gradually yielding a total contact duration of 12.2:32 ms. All knee joints did not require

t1'le firll series of impact energies to fracture. Peak contact force increased with increasing

73

energy delivered to the knee. Visible bone fracture occurred for a range of energies (24-
92 J) and contact farces (3.2-10.6 kN) (Table 1). There were nine transverse fractures of
the patella, two comminuted fractures of the patella, two distal femoral fractures, one

sigtlificant disruption of articular cartilage, and one joint presented no gross visual
damage.

Re gxession Statistics

The regression analyses of various subject parameters revealed two significant

correlations using univariate regression and no significant correlations using multivariate

models. First, the fracture load was statistically correlated with the morphology ratio

(Table 2). As the morphology ratio decreased, the load required to cause gross fi'acture of
bone significantly increased (Figure 1). Second, the patellar histopathological score
Sigtlificantly increased with increasing subject mass. The morphology ratio, however, was
not predicted by any of the subject parameters. Analysis of the patellae revealed a median

l11<>rphology ratio of 1.30 (range 1.17-1.38) and a median histopathological score of 5
(range 2-8).

\P aramteric Study Findings

The theoretical parametric analysis revealed that increasing contact area increased

t11e load tolerance of the patella and femur. For small contact areas (<40%) the shear and
t§l'lsile failure stresses in the patella occurred before those in the femur. For larger contact
11Teas, femoral fracture was the limiting criterion. Failure of the patellar or femoral

CEar-tilage due to large shear stresses occurred at higher loads than for bone fractures for
file 90 degree flexed knee.

74

DI SCUSSION

In this study we hypothesized that joint pathology and other physical parameters

which describe impact test subjects would influence the load to cause fracture of bone

fi-om a blunt insult to the human knee. The study indicated that subject age, joint

patliology score, sex, height and mass did not correlate with the impact load required to
cause bone fracture in the isolated human knee using a rigid interface. Patellar geometry
(the morphology ratio) was, however, found to be a' predictor of fracture load. This
suggested that joint geometry may play a significant role in determining fi'acture load.
It is important to note that the fracture load data presented in the current study
Was thought to be conservative, and may not reflect the realistic impact event for several
reasons. Previous investigators (Patrick 1965, Melvin 1975, Powell 1975), directed
it'I'lIDact missiles at the 90 degree flexed knee of seated cadavers using both single and
Se(llrential impact protocols. They documented fiacture loads of approximately 10-17 kN,
Sigl'lificantly higher than those of the current study. Firstly, the sequential impact protocol
used in the current study was not ideal, as “this technique introduces the uncertainty of
I3“()gressive predarnage to the skeletal structure”(Powell 197 5). Repeated impacts likely
IFQSult in accumulated bone microdamage leading to a reduced load at gross fi'acture.
ItTlpact trauma at subfracture levels to the human knee is known to result in occult
t1‘1icrocracking in the subchondral plate of the patella (Atkinson:a 1995, Atkinson:b 1997).
These types of experimentally produced defects in the bone likely introduce stress
Q()ncentrations which may reduce the load necessary to initiate a gross bone fracture. On
the other hand, all specimens in this study were exposed to the same protocol. It seems

I‘easonable to assume that the accumulation of microdamage in the bone would be

75

consistent across all specimens, or perhaps, even more pronounced in the more fragile
specimens. Yet, age, pathology, etc. did not correlate with the final fracture load. A
second reason for the lowered fracture loads in the study versus earlier studies using the

whole cadaver was the use of isolated knee preparations. The impact loads in this case

were generated by deformation of the joint tissues. In an automotive crash event,

however, loads derived by contact with a knee bolster, for example, would be the result of
tissue/knee bolster deformation, as well as inertial motion of the occupant. Another
reason for the lower expected fracture loads in the current study was the use of a rigid
irnpact interface. This allowed us to make more unambiguous comparisons to past

StI-ICIies. The earlier studies with rigid interfaces largely serve as the basis for the current

mjul‘y criterion. Yet, automotive occupants commonly interact with more deformable
S‘3‘lllctures, such as the instrument panel. This type of interaction may result in injury to
me femoral condyles versus the patella, as described in the model analysis.

In this study we proposed the idea of a stress based injury criterion for the
patellofemoral joint. Fundamentally, the state of stress in a body is due to three main
13a~I‘ameters: geometry, material properties and boundary conditions. While the

I'rlorphology ratio is admittedly a very simple measure of patellar geometry, it is a step
tQV'vards quantifying the shape of the knee. Since the morphology ratio relies only on one
1a‘teral 2-D view, medial-lateral geometrical variations are not included. Yet, this simple
hiatomical ratio was found to significantly correlate with fracture load.

The material properties of the knee joint tissues from each cadaver was not

quantified in the current study. This is a difficult problem due to their spatial variation

76

within subjects and the inherent variability between subjects. The latter may be due, in
part, to confounding parameters not measured in this study, such as heredity, physical
demands during life, etc.

Finally, the influence of the impact boundary conditions on the state of stress in the
lcriee was studied with the 2-D human knee model. A uniform boundary condition of
valying contact width was applied in an iterative fashion to identify injurious levels of
Stress in the patella and femur. Clearly, the shape of the curves is a direct result of our
1'1‘3l<)deling protocol, and the curves might shift with variations in modeling protocol. For
e”Kall'lple, the failure stresses in the current study require additional study and further
SCI'llltiny as more experimental/theoretical data become available. Currently, the geometry
is 1inrited to a single knee at 90 degrees flexion with uniform loading. Different flexion

axlgles will change the boundary conditions by altering patellofemoral contact and contact
between the knee and impact interface. Further, a hyperbolic, versus a uniform load
I>rOfile would concentrate load at the center of the contact area, potentially resulting in an
aJ‘iered stress state for a given contact area. It is possible then that contact area and load
intensity might not be sufficient in accurately predicting tissue stresses. Knowledge of
Q<)Irtact pressure gradients, such as if the occupant contacts the corner of a heater vent,
glove box knob, etc., resulting in small areas of high contact pressure, would also be
needed to assess the state of stress in the interior of the joint. We believe the model
ultimately can have the utility to investigate different impact interfaces and load profiles,
tiftaking it a valuable tool to predict stresses in the human knee. Ultimately, it will be

l”lecessary to adjust the injury curves to be relevant to the crash environment. This will

involve a transformation of the isolated human cadaver knee data to the current dummy
knee, and then to the whole body setting in the crash environment. Simple methods, such
as the use of pressure sensitive film on the dummy knee might be used to record the
contact area and distribution of contact pressures over the dummy knee. These efforts
represent a first step towards a more comprehensive injury criterion for the human knee.
ACKNOWLEDGMENTS

This study was supported by a grant (R49/CCR503607) from the Centers for
Disease Control and Prevention. 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 Ms. Jane Walsh and Mr. Marc
Newmark for help in the histological slides, Mr. Cliff Beckett and Hasnin Ali for technical
support and Mr. Rick Shumaker and the Michigan Tissue Bank for solicitation of donors
and retrieval of joints.
REFERENCES

I. Anderson DD, Radin EL: Stress wave effects in a finite element analysis of an
impulsively loaded articular joint. Journal of Engineering in Medicine 205:27, 1991

2. Atkinson PJ, Haut RC: Subfiacture Insult to the Human Cadaver Patellofemoral Joint
Produces Occult Injury. Journal of Orthopaedic Research 13:936, 1995

3. Atkinson P, Garcia J, Altiero N, Haut R: The influence of impact interface on human
knee injury: implications for instrument panel design and the lower extremity injury
criterion. 41 st Stapp Car Crash Conference 167, 1997

4. Atkinson P, Haut R: Insult to the human cadaver patellofemoral joint: effects of age on
fracture tolerance and occult injury. 39th Stapp Car Crash Conference 281, 1995

5. Atkinson T, Haut R, Altiero N: An investigation of failure criteria for impact induced

fissuring of articular cartilage. Injury prevention through biomechanics symposium 85,
1997

78

6. Eberhardt AW, Keer LM, Lewis JL, Vithoontien V: An Analytical Model of Joint
Contact. Journal of Biomechanical Engineering 112:407, 1990

7. Garcia J, Newberry W, Atkinson P, Haut R, Altiero N: Comparison of stresses in
human and rabbit knees as they pertain to injuries observed during impact experiments.
ASME Bioengineering conference 315, 1997

8. Grelsarner R, Proctor C, Bazos A: Evaluation of patellar shape in the sagittal plane: A
clinical analysis. Amer J Sports Med 61, 1994

9. Haut R: Contact Pressures in the Patello-femoral Joint during Impact Loading on the
Human Flexed Knee. Journal of Orthopaedic Research 7:272, 1989

10. Mankin H, Dorfman H, Lippiello L, Zarins A: Biochemical and Metabolic
Abnormalities in Articular Cartilage from Osteo-arthritic human Hips. J of Bone and Joint
Surgery 13:523, 1971

ll. Melvin J, Stalnaker R, Alem N, Benson J, Mohan D: Impact Response and Tolerance
of the Lower Extremities. Stapp Car Crash Conf 19:543, 1975

12. Mente P, Lewis J: Elastic modulus of calcified cartilage is and order of magnitude less
than that of subchondral bone. Journal of Orthopaedic Research 12:637, 1994

13. Patrick L, Kroell C, Mertz H: Forces on the human body in simulated crashes. Stapp
Car Crash Conf 237, 1965

14. Powell W, Ojala S, Advani S, Martin R: Cadaver Femur Responses to Longitudinal
impacts. Stapp Car Crash Conf 19:561, 1975

15. Repo R, Finlay J: Survival of Articular Cartilage after Controlled Impact. The Journal
of Bone and Joint Surgery 59-A(8):1068, 1977

16. Townsend P, Raux P, Rose R, Miegel R, Radin E: The Distribution and Anisotropy of

the Stiffness of Cancellous Bone in the Human Patella. Journal of Biomechanics 8:363,
1975

79

TABLES

Table 1: Impact and sub 'ect parameters

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Specimen gross injury impactor fx load morphology histology sex age height mass
(fx=fracture) energy (J) (kN) ratio score (m) (kg)
H02 Transversefir 24 4.9 1.38 3.5 f 39 1.62 79.5
of the patella
H03 Comminuted 45 3.2 1.32 5.5 m 39 1.74 44.5
firof the
femur
H04 Gross 90 9.4 1.26 3.0 m 49 1.83 95.0
disruption of
cartilage
H05 Transversefir 64 7.7 1.17 5.0 m 52 1.87 81.8
of the patella
H06 Transversefir 43 6.6 1.23 6.5 m 59 1.87 88.6
of the patella
H07 Comnrinuted 92 8.1 1.26 4.5 m 42 1.83 75.0
fxof the
patella
H08 Transversefii 65 6.9 1.22 3.0 m 34 1.87 68.2
of the patella
H10 Nogross 89 10.6 1.24 5.5 m 61 1.83 95.4
visual
damage
H18 Femurshaft 63 5.6 1.36 6.5 m 70 1.83 113.6
fit
H20 Transversefx 62 8.9 1.29 2.5 m 72 1.87 40.9
of the patella
H21 Transversefir 38 6.0 1.29 2.5 f 62 1.36 77.5
of the patella
H23 Transversefii 59 6.9 1.23 2.0 m 73 1.92 50.9
of the patella
H25 Transversefir 52 8.9 1.18 7.0 m 85 1.76 68.2
of the patella
H26 Comminuted '33 5.2 1.30 8.0 f 70 1.76 120.5
fitof patella
H27 Transversefir 37 4.6 1.36 5.5 m 70 1.65 79.5
of the patella

 

80

 

Table 2: Linear regression results

 

 

 

 

 

 

 

Predicted Variable Predictor Variables Univariate Linear
Regression
I) Fracture Load 1) Morphology ratio 1) p=0.012
2) Sex 2) p=0.153
3) Height 3) p=0.208
4) Age 4) p=0.476
5) Histopathological score 5) p=0.505
6) Mass 6) p=0.962
2) Histopathological 1) Mass 1) p=0.030
Score 2) Age 2) p=0.269
3) Fracture load 3) p=0.505
4) Height 4) p=0.523
5) Morphology ratio 5) p=0.917
6 Sex 6) p=0.974
3) Morphology Ratio 1) Fracture load 1) p=0.04l
2) Sex 2) p=0.192
3) Height 3) p=0.275
4) Mass 4) p=0.421
5) Age 5) p=0.838
6) Histopathological score 6) p=0.917

 

81

 

FIGURES

 

     

 

 

 

Morp ology Ratio=alb
11 - I
A10 . Cartilage
z
5 9 -
'U
i .. -
a 7 '
3 6 '
E 5 . I
u_ y = - 20x + 32.5 I
4 ' r2 = 0.41
3 u u I l—l__'l
1.15 1.20 1.25 1.30 1.35 1.40
Morphology Ratio

 

 

 

FIGURE 1: Fracture load significantly decreased with increasing morphology ratio. The
morphology ratio is a measure of the proximal-distal patellar bone length on a mid-sagittal
plane divided by the length of the articular surface.

   
    
   
 
    

Available
0 I . c ‘ .

Area of contact
<-

Patella

  

“:1. ;' "' r I _ p . _
graphic. : '_£
[ti-he Tag. gag-Mass“? . g

. . '— {run 1 .
. ~ \ Femur 51.5%. ;

\,...- " ..\ 93’
r- \
l‘ki us'T IS./1/\/'\K/\)I

        

FIGURE 2: A 2-D contact model was developed from a transverse slice taken centrally
through a typical cadaver knee flexed at 90 degrees. The influence of load intensity and
contact area on injurious levels of stress was investigated by considering a range of
contact areas (A: small area of contact, B: large area of contact). The tissue failure
stresses are taken from Garcia, etal., 1997.

82

 

 

7000

6000

5000 -

4000 -

3000 .

ZQQOI-

A
V

2000 ~

1000 -

 

 

 

0 0.002 0.004 0.006 0.008 0.01 0.012
Time (s)

Figure 3: Typical load-time u'ace for blunt trauma to the isolated knee by a free flight
inertial mass.

 

 

 

 

18

Iriuy Cuve For Patellar
‘5 Stbchond'al Bone I
z

14

Load (KN)
or on 3 R;

&

 

%COntactArea

 

 

 

Figure 4: A plot of the combination of impact load intensity and contact area over the
knee which produce failure levels of stress. Combinations of contact area and load which
are below the curves would probably not result in injury. The horizontal axis is the area of
contact divided by the total available contact area (see Figure 2). For small contact areas
(<40%) the patellar bone is apparently more susceptible to injury, while the femoral
condylar bone is more susceptible to injury with larger areas of contact. The
patellofemoral articular cartilage does not appear to be the limiting injury factor for the
range of contact areas and loads studied.

CHAPTER 6

THE INFLUENCE OF IMPACT INTERFACE ON HUMAN
KNEE INJURY: IMPLICATIONS FOR INSTRUMENT PANEL
DESIGN AND THE LOWER EXTREMITY INJURY CRITERION

Patrick J. Atkinson, Jose J. Garcia, Nicholas J. Altiero, Roger C. Haut
Department of Materials Science and Mechanics, College of Engineering
Orthopaedic Biomechanics Laboratories, College of Osteopathic Medicine
Michigan State University

ABSTRACT

Injury to the lower extremity during an automotive crash is a significant problem.
While the introduction of safety features (i.e. seat belts, air bags) has significantly reduced
fatalities, lower extremity injury now occurs more frequently, probably for a variety of
reasons. Lower extremity trauma is currently based on a bone fracture criterion derived
from human cadaver impact experiments. These impact experiments, conducted in the
1960’s and 70’s, typically used a rigid impact interface to deliver a blunt insult to the 90°
flexed knee. The resulting criterion states that 10 kN is the maximmn load allowed at the
knee during an automotive crash when certifying new automobiles using anthropomorphic
dummies. However, clinical studies suggest that subfracture loading can cause
osteochondral rnicrodamage which can progress to a chronic and debilitating joint disease.
Additional studies suggest that the stiffness of the impact interface influences knee injury

modalities resulting from impact loading. In the current study, a single impact at 27 J of

energy with a rigid interface was delivered to one knee of isolated joint preparations of six
cadavers resulting in an average peak load of 5 kN. Contralateral knees were impacted
with a padded interface at an additional level of energy at approximately the same load.
All rigid impact experiments resulted in some form of injury to the patella including occult
microcracking and gross fracture of the patella. No injuries were detected in the knees
from the padded experiments. Math modeling of the patellae showed significantly reduced
tensile and shearstresses in the bone with padding. This study suggests that the current
lower extremity injury criterion, based solely on load, may not be sufficiently conservative.
Increasing contact area over the knee reduces stresses in the bone and prevents both gross
bone fi'acture and bone and cartilage rnicrodamage. Such data may be useful in future
instrument panel designs and might suggest revision to the current lower extremity injury
criterion.
INTRODUCTION

Three extensive field studies conducted in 1977 examined the frequency of injuries
associated with automotive accidents (Nagel 1977, Harternann 1977, Bourret 1977). The
studies found as many as 37% of automotive accident survivors sustained injuries
involving the lower extremity (foot, knee, thigh or hip). Injury typically involved gross
fiacture of bone and the knee accounted for approximately a third of these injuries. This
suggests that approximately one in ten automotive accidents resulted in knee injury two
decades ago. Knee injury typically resulted from high speed contact with the instrument
panel (Hartemann 1977, Bourret 1977). It was estimated in 1977 that 900,000 knee

injuries associated with motor vehicle accidents occurred per year in the United States.

Severe injury to articular joints, such as those incurred to the knee during automotive
accidents, are difficult to treat (Huelke 1982, Pattimore 1991, Hayashi 1996) and are
implicated in chronic joint morbidity and pain (Nagel 1977, Radin 1972, Upadhyay 1983,
Radin 1973, Radin 1978, Donohue 1983, Nyquist 1985, Thompson 1991, Vellet 1991,
Gardner 1994, Volpin 1990, Chapchal 1978, Amoldi 1991). In 1976 a federally regulated
injury criterion (FMVSS 208) was adopted to address lower extremity injury. The
criterion states that axial femoral loads cannot exceed 10 kN when certifying new
automobiles with anthropomorphic dummies at a crash velocity of 13.4 m/s (30 mph).
The criterion is based on a bone fi'acture threshold derived fi'om human cadaver impact
experiments (Patrick 1965, Melvin 1975, Powell 1975, Viano 1977, Viano 1980).
Analysis of more recent injury incidence studies conducted in the 1990’s shows
that the incidence of lower extremity injury has not decreased over the past twenty years
and has, in fact, increased in some analyses (Pattimore 1991, Siegal 1993, Morgan 1991,
Thomas 1995, Otto 1996). These results are attributed to the increased safety of
automobiles following the mandated use of seat belts and the fleetwide phase-in of air bags
into production vehicles from 1987-97. In 1995 it was documented that seat belts and air
bags are 42% and 15% effective, respectively, in preventing fatal injuries (Lund 1995,
Viano 1995). A decrease in the incidence of fatalities associated with severe crashes has
resulted in a higher incidence of non-life threatening injuries, such as those involving the
lower extremity. While seat belts and air bags are effective in protecting vital organs
(Mackay 1995) fiom injurious impact with the windshield, steering wheel, etc., the lower

extremity may be the sole site of unprotected contact between the occupant and the

86

automotive interior (Hartemann 1977, Bourret 1977, Pattimore 1990, Siegal 1993,
Morgan 1991, Thomas 1995, Otte 1996, Mackay 1995, Crandall 1995). Currently,
4,800,000 injuries are attributed to automotive accidents each year (King 1995) and
500,000 injuries are serious enough to require hospitalization. Estimates of the current
incidence rate of lower extremity injury are 30% (Morgan 1991), 37% (Pattimore 1991),
50% (Huelke 1982) and as high as 58% (Siegal 1993). This suggests between 150,000-
300,000 lower extremity injuries per year require treatment. Eight per cent of the total
treatment cost from motor vehicle accidents is due to the lower extremities (Malliaris
1982). In 1993, the overall cost associated with treating injured automotive occupants
was approximately $55,000 per occurrence (Siegal 1993), suggesting that lower extremity

injury may account for $2,200,000,000 annually.

In addition to gross, acute trauma, clinical studies suggest irreversible injury may
occur in the absence of bone fiacture (States 1986, MacKenzie 1988). Acute post-
traumatic pain and a subsequent disease (osteoarthrosis or osteoarthritis, known as CA)
are complications commonly associated with joint trauma (Nagel 1977, Radin 1972,
Upadhyay 1983, Radin 1973, Radin 1978, Donahue 1983, Nyquist 1985, Thompson
1991, Vellet 1991, Gardner 1994, Volpin 1990, Chapchal 1978, Amoldi 1991, States
1986). OA may account for a societal cost of $54,000,000,000 annually in both treatment
costs and lost work days. OA affects cartilage and subchondral bone of articulating joints
(Harnerman 1989). While the etiology of DA is not well understood, subfiacture
mechanical insult has long been associated with its development via controlled in viva

animal experiments (Radin 1972, Radin 1973, Radin 1978, Donahue 1983, Thomspon

87

1991). These studies suggest that bone rnicrodamage, and its subsequent healing, causes
sclerosis of the underlying subchondral bone inducing abnormally high stresses in the
overlying cartilage. This leads to progressive cartilage degeneration and loss of this tissue.

Clinical and post-mortem studies have been conducted on the human in an effort to
understand joint disease pathogenesis and progression. Shimizu et al. (1993) studied
biopsy tissues from the human tibial plateau and found degradation in the underlying bone
preceding changes in the overlying cartilage. A recent clinical study of high energy knee
injmies indicates changes in the underlying subchondral bone as detected by magnetic
resonance imaging (MRI). These injuries were associated with "overt cartilage loss or
defect in 48% of the patients evaluated 6 months after injury" (V ellet 1991).
"Radiographically occult injuries to the bone, otherwise referred to as occult fractures,
bone bruises, or microtrabecular trauma/ fracture, may account for patient pain" (Kapelov
1993), perhaps by impeding venous drainage through extraosseous veins (Amoldi 1991).
Additional clinical studies document a strong correlation between a history of joint trauma
and the development of osteoarthritis (Davis 1989, Kellgren 1958). Indeed, complaints of
joint pain immediately following insult to a joint has been reported in approximately 88%
of patients eventually progressing to osteoarthrosis (Chapchal 1978). Additionally,
Johnson-Nurse and Dandy (1985) document that 50% of osteoarthritic patients recall a

traumatic injury to the diseased joint in years prior to the development of symptoms.
Recent studies by our laboratory have shown that occult microcracks can be

produced near the cartilage-bone interface following blunt subfracture insult to the 90°

flexed human knee (Atkinson:a 1995, Atkinson:b 1995). The microcracking is evident at

approximately 70% of the load required for gross fracture of bone in the knee. These
microcracks may yield the so-called “bone bruises” in MR images. Because these injuries
occur below the gross fracture tolerance of bone, the current criterion used in the
automotive industry may not be sufficiently conservative to protect the long term health of
the joint. Two-dimensional mathematical models of the human knee suggest that these
occult microcracks may be the result of excessive shear stresses (Atkinson:b 1995).
Parametric studies performed on these models indicate the distributions of stress in
articular cartilage and bone depend on the input boundary loads and joint geometry
O-Iayashi 1996, Li 1995, Atkinson 1996, Newberry 1996). For example, distributing
impact loads over a greater area (i.e. padded interface) might limit injury to bone and
cartilage (Hayashi 1996). Theoretical models suggest shear stresses are reduced by more

than 50% with a padded versus rigid impact interface (Hayashi 1996, Newberry 1996).

While the field investigation data presented above show lower extremity injury to
be a significant problem, numerous experimental studies attempting to simulate contact
between the knee and instrument panel, have also been undertaken in an effort to identify
injury thresholds and mechanisms. The experimental studies are advantageous aver field
observations since tests can be conducted in a controlled, repeatable manner with the
ability to obtain detailed data. Altemately, field observations are made after the incident
has taken place. Assumptions must be made as to vehicle/occupant kinematics, contact
points, etc. Therefore, to quantify what is already known, a retrospective analysis was
conducted of four main experimental impact studies using whole cadavers (Patrick 1965,

Melvin 1975, Powell 1975, Viano 1980). A total of 68 limbs were tested. Typically, the

89

cadaver was seated with the knee at 90° flexion and an impact missile of variable mass
(10-50kg) was directed at the knee at various velocities (5-13.4 m/s). In 32 tests the
impact missile was faced with a flat, rigid plate. In the remaining 36 tests, the plate was
covered by pads of variable thickness and composition. The incidence of fracture to the
patella, femoral condyles, femoral shaft, femoral neck and pelvis are shown in Table 1.

The majority of fracture injuries occur in the knee with the rigid interface in which
the patella, femoral condyles or both suffered gross fracture. The localized distribution of
load on the patella has been implicated in the high incidence of knee injuries with the rigid
interface (Hayashi 1996, Powell 1975). The contact between the rigid impact interface
and the bony anterior surface of the patella likely resulted in a very localized distribution
of the impact load. The flesh thickness over the anterior patella is approximately 4.8 mm
(Hering 1977), offering little protection to a rigid impact. Altemately, padding likely
resulted in a distribution of the impact force over part or all of the patella and possibly the
femoral condyles (Hayashi 1996). This resulted in a significant reduction of knee injuries.
These studies suggest that the distribution of impact load over the knee is an important
factor in injury prevention.

The lower extremity injury criterion currently utilized in the automotive industry
(FMVSS 208) is based, in part, on experimental studies using a rigid interface. This
suggests that the criterion is a somewhat conservative predictor of lower extremity injury
since typical instrument panels are not rigid, thus offering some degree of protection by
distributing impact load over the knee. However, as mentioned above, recent field reports

document that lower extremity injury persists at or above past incidence rates, probably

for a variety of reasons. While increasing the amount of contact area over the anterior
surface of the knee decreases the incidence of injury, the criterion is based solely on peak
impact force and does not take into account how the force is distributed. Currently, it is
not possible to quantify the contact area required to prevent injury at loads less than 10
kN, based on an analysis of the whole cadaver experiments cited above, due to the
following:

1) While load data was measured in all of the past studies, the contact area over the knee
was not measured (i.e. with pressure sensitive film, etc.).

2) Intrastudy comparisons/contrasts of specific data may not be valid due to variations in
protocol (i.e. impact energy, pre-impact cadaver positioning).

3) Between-subject comparisons may not be valid due to the large variability documented
in fiacture load and energy with cadaver age, height and sex (Powell 1975.

4) Between-subject comparisons may not be valid due to anthropometric variations of the
anterior surface of the knee (size variations of 20% in a random sample of knees) (Hering
1977)

Anterior knee surface geometry is of particular interest because the contact of the
impact interface with this anatomy influences, in large part, the distribution of load over
the knee. For example, scaled shape differences can lead to larger or smaller contact
areas, whereas, local anatomical incongruities could lead to load distribution gradients in
the area of contact. Intersubject articular joint geometrical variation may be due to a
variety of effects including age, sex, height, weight and heredity (Grelsamer 1994). In

general, joint geometry probably reflects the demands placed on the joint during normal

91

activities such as ambulation, rising from a chair, etc. Patellar geometry is a significant
predictor of anterior knee pain (Grelsamer 1994) and may indicate a predisposition to
early joint pathology which may influence knee joint injury mechanisms (Atkinson:b
1995)

The objective of the current study is to investigate the relationship between load
and contact area while addressing the concerns noted above. The goal was to investigate
how different distributions of a given load would affect impact biomechanics and
macro/micro injuries to the knee. Therefore, paired experiments were conducted on
human cadaver knees with rigid and padded interfaces at consistent ‘within subject‘ impact
loads. Each patella was theoretically modeled to investigate whether the state of stress in
the bone and cartilage correlated with injury. A geometrical measure was also made of
each patella to investigate how individual joint geometry may have affected the theoretical
and experimental results. Subsequent gross and histological examinations were conducted
and injury to bone and cartilage were documented.

METHODS AND MATERIALS
Experimental Methods

Impact experiments were conducted on six pairs of isolated human knee joints
(65:12 years). The sample size was selected following a confidence interval analysis to
determine how many subjects would be required to yield sufficient statistical power. The
premartem age of the subjects was not of particular interest for the specific type of testing
performed in the current study. While other studies have documented age effects in the

properties of joint tissues, a previous study showed that cadaver age is not a significant

92

predictor of impact biomechanics or injury (gross or microdamage) to the human knee
(Atkinson:b 1995). Advanced joint pathology, however, may affect joint injury and
impact biomechanics. Therefore, gross joint inspection was used to select the sample pool
of test subjects. Tissues were procured by the Tissue Bank (see Acknowledgment) from
donors with no prior history of joint trauma and/or evidence of advanced joint pathology
(areas of denuded bone, advanced chondromalacia, etc.). Shortly after death, the
patellofemoral joint was examined through a medial incision. Joint preparations were
excised approximately 15 cm from the knee proper within 4 hours of death and stored at -
20° C until the day of testing. Superficial tissues proximal to the femoral condyle were
excised and the femoral shaft was cleaned with alcohol. The femur was potted in a
cylindrical steel sleeve with room temperature curing epoxy, and was mounted to a rigid
test frame (Haut 1989). Each specimen was maintained at a flexion angle of 90° prior to
impact with a loosely knotted tether tied to the quadriceps tendon. The knot was such
that minimal load was required to cause slippage, thus minimizing tensile loads in the
tendon during impact. One knee from each subject was impacted once (see below for
details) with a rigid interface at an impact energy of 27 J. This energy was selected
following a logistical regression analysis of a previous experimental study involving twenty
human cadavers (Atkinson:b 1995). The analysis suggested that 27 J of energy would
provide the greatest chance of producing an occult trauma while accepting a 50% chance
of gross bone fracture (Figure 1).

Following the rigid interface experiment on one knee of each subject, the

contralateral knee was impacted once with a padded interface (Hexcel, 1.35 MPa crush

93

strength) at an energy which resulted in approximately the same load. A greater impact
energy was required to achieve the same load as in the rigid interface experiments, due to
the deformable characteristics of the padding. Pilot impact experiments were conducted
on a knee analog to determine the relationship between energy and peak contact load for
the rigid and padded interfaces. The knee analog was constructed from a short length of
10 cm x 10 cm wood block covered by a 6.4 rmn thick rubber sheet (Figure 2).
Experiments with a rigid and padded interface using a 4.8 kg mass were conducted at a
range of velocities. This data was crossplotted to eliminate the influence of the knee
analog and to develop a functional relationship between the padded and rigid interface.
The increased impact energy required for padded interface experiments to yield the same
load as rigid experiments could then be determined (Figure 3).

The data is presented as a so-called ‘energy multiplier’ plotted versus the rigid
interface peak impact load. While the impact energy was consistent for all rigid interface
experiments, the impact load was variable for all cadavers. Therefore, following the rigid
experiment for each cadaver, the increased energy required to produce the same peak load
with the padded interface was determined by looking up the energy multiplier (Figure 3).
The rigid impact energy (27 J) was then multiplied by this factor and the contralateral knee
was impacted with the padded interface at that energy.

Impact was delivered to the knees via a 4.8 kg missile accelerated to the pre-
irnpact velocity required for each experiment. The missile was aligned on two steel guide
rails and supported on four nylon blocks to minimize fiiction. A 7 .5 x 8 cm, 6061-T6

aluminum impact interface was mounted to the fi'ont of a load transducer (Model 3173-2k,

Lebow Products, Troy, MI, USA). The impact mass was selected based on a previous
study (Haut 1989). The resonant frequency of the load cell with the impact interface
attached was 1333 Hz. Pilot studies indicated that the load transducer and attached mass
would not resonate during impact on the human knee. Load data was recorded
continuously on a personal computer via a 16 bit A/D board (Model DAS 1600,
Computer Boards, Inc., Mansfield, MA, USA). Load data were inertially compensated
(Atkinson 1995) and sampled at 20,000 Hz. The velocity of the mass was measured prior
to impact on the knee using two infrared optical sensors (Part No. OR518-ND, Digi-Key
Corporation, Thief River Falls, MN). These sensors were designed with a 5 rrrrn
separation between the emitter and detector. A tab connected to the impact mass crossed
the 5 mm gap interrupting the first sensor and then the second. The individual sensors
were separated by 76.2 mm to eliminate cross-talk.

A medium range (7-40 MPa) pressure sensitive film (Prescale, single sheet, Fuji
Film, Ltd., Tokyo, Japan) was inserted into the joint prior to each impact through medial
and lateral incisions of sufficient size to allow the film to pass through. The film was
encased in two sheets of polyethylene film approximately 0.05 mm thick to prevent
exposure to body fluids and help reduce shear loading artifacts (Huberti 1984). The strips
of film were sufficiently long to pass through the patellofemoral joint and cover the
anterior surface of the patella to measure the location and distribution of the contact
pressure in the patellofemoral joint and at the impactor-anterior patella interface. This
also minimized fiictional forces between the impactor and the knee. A method for

indicating regions of articular cartilage contact suitable for impact loading was also

95

employed (Atkinson 1997). This method sernipennanently colors the area of intraarticular
contact and allows documentation of the contact area to aid in locating the contact
pressure boundary condition for math modeling.

The pressure sensitive film was calibrated in a servohydraulic testing machine
(Model 1330, Instron, Canton, MA, USA) using a previous methodology (Haut 1989).
The film showed the highest sensitivity over the range of approximately 8 to 30 MPa.
Briefly, the film was encased in the polyethylene packet and loaded with a 40 ms time to
peak load-controlled haversine pulse. The films were analyzed within one day using a
video scanner (Model MSF-3OOZ, MicroTek International Incorporated, Hsinchu,
Taiwan, ROC) at 150 dpi. . The average pressure over that length was determined using
the public domain NIH image program, version 1.6 (developed at the US. National
Institutes of Health and available on the intemet at http://rsb.infa.nih.gav/nih-image/).
The distribution of contact pressures and contact areas were determined within the
patellofemoral joint.

Following the mechanical test the patellar geometry was quantified using the
method of Grelsamer, et al. (1994), in which the so-called ‘morphology ratio’ is
determined. The ratio is the quotient of the mid-sagittal length of the patellar bone
divided by the sagittal length of the articular surface (‘a’ and ‘b’, respectively, in figure 4).
The morphology ratios were used to categorize each patella into one of three categories:
Type I patellae, 1.2 S a/b .<_ 1.5; Type II patellae, a/b > 1.5, Type III patellae, a/b < 1.2.
Type I patellae are typical of 94% of the population (Grelsamer 1994).

The patellae were placed in 10% buffered formalin for 10-14 days. The tissue was

decalcified in 20% formic acid. Each patella was transected into three to five 2-3 mm

sample blocks, processed for routine paraffin embedding, slab-cut and stained with
Safi‘anin O-Fast Green. All sections were examined in light microscopy at 12-400 power
and photographed.

The experimental data resulting from analysis of the load-time curve and the
pressure sensitive film were tabulated and statistically analyzed. The time to peak load,
contact duration, peak contact load, average anterior patella contact pressure and area,
average patellofemoral contact pressure and patellofemoral contact area data fi'om rigid
interface and padded interface experiments were compared with paired t-tests (Si grnastat).
A regression analysis was conducted to determine the influence of cadaver age, sex,
height, weight and patellar geometry on peak load, average anterior patella contact
pressure and area and average patellofemoral contact pressure and patellofemoral contact
area.

Theoretical Modeling

The cartilage, trabecular and cortical bone geometry of the patellae from all rigid
and padded interface experiments were digitized into a commercial CAD software
(AutoCAD, version 12) (Figures 5,6). The geometry was taken from a transverse plane in
the area of highest patellofemoral contact pressure (Atkinson:b 1995). Twelve 2-D
mathematical models (AN SYS, version 5.3) were developed to analyze each patella along
that plane. The patella was meshed with 8 node elements and elastic isotropic behavior
was assumed for the analysis with the material properties in Table 2. The assumption of
elastic behavior is based on the work of Eberhardt, et a1 (1991). Cartilage and, to a lesser

extent bone, are commonly considered viscoleastic and are typically modeled with a rate

97

dependent, biphasic analysis (Setton 1990, Maw 1984). However, impact loading
(<200ms) results in negligible differences between the linear elastic and biphasic analyses.
The model was run assuming static loading since stress wave transmission through
articular joints causes little deviation from the quasistatic stress state (Anderson 1990,
Anderson 1991).

The patellofemoral pressure distribution from the subfracture impact experiment
was discretized and applied as a compressive elemental pressure to the retro-patellar
surface of the model. The pressure distribution was located by the intraarticular contact
method (Atkinson 1997). A one pixel slice of the pressure film image, consistent with the
2-D model, was discretized into 1 mm lengths. The average pressure over that length was
determined using the public domain NIH Image program, version 1.6 (developed at the
US. National Institutes of Health and available on the intemet at
http://rsb.info.nih.gov/nih-image/). The resulting pressure distribution was then applied to
the posterior surface of the patella. The anterior surface of the patella was fixed in the area
of contact with the impactor. Three areas were investigated: the cartilage surface, the
subchondral bone, and the trabecular bone. The maximum principal stress, maximum
shear stress and their respective locations were tabulated. The stresses from the rigid and
padded experiments were compared with a paired t-test (Sigmastat). A regression analysis
was conducted to determine the influence of cadaver age, sex, height, weight and patella

geometry on the maximum stresses predicted by the model.

98

RESULTS

Experimental Results

The typical load-time response was a symmetrical haversine. The time to peak
load (loading rate) and contact duration (pulse) for rigid experiments were 4.14i0.45 ms
and 11.33i3.0 ms, respectively. These data were significantly less than padded
experiments which had a time to peak load of 7.01:1:1.54 ms and a contact duration of
17.03:l:3.9 ms. A typical load time curve (H34) (Figure 7) shows similar peak loads with
different loading rates and contact durations. The impactor-patella average contact
pressure and contact area were different between rigid and padded interfaces. The rigid
interface contacted a small area in the center of the anterior patellar surface. The average
anterior patellar contact pressure and area were 182-$4.7 MPa and 247187 mmz. With
padding the entire patella was contacted resulting in significantly increased knee contact
area and reduced contact pressure. The anterior patellar contact area was approximately
2000 mm2 and the contact pressure was 1.4 MPa. There was no statistical difference
detected for peak contact load or patellofemoral contact pressure between rigid and
padded experiments. While the patellofemoral contact area tended to be larger with the
rigid interface, this difference was not significant (Table 3). The patellofemoral contact
was typically proximal and adjacent , to the femoral notch on the femur (Figure 8) and
centrally located on the patella (Figure 9). The contact area had a ‘kidney’ shaped
appearance. The dark areas in Figures 8 and 9 illustrate regions of intraarticular joint

contact.

Two rigid interface experiments resulted in gross transverse fracture of the patella
(H32, H33) (Figure 9, 10) and one resulted in gross transverse patella fi'acture and gross
fracture of the femoral condyles (H36). While two patellar fractures were full thickness
through the patella (H33, H36), the fracture in the remaining specimen (H32) was partial
thickness and emanated from the posterior surface. Analysis of the histological sections
from all rigid interface experiments revealed occult microcracking in four of six specimens
(H31, H32, H35, H36). No gross bone fracture or microdamage was detected in padded
experiments.

Categorization of the patellae by morphology ratio (Table 4) revealed that five of
six were of Type I and one was of Type II (H31). Therefore, data from H31 was
therefore not included in regression analyses including the morphology ratio. Regression
analyses of biomechanical impact data (Table 3) against cadaver, age, sex, height, and
weight were not statistically significant. Regression analyses of the data against patella
morphology ratio revealed a significant influence of decreasing peak contact force with an
increasing morphology ratio (Figure 13). In addition, the three fiacture specimens (H32,
H33, H36) exhibited a significantly (p<0.05, unpaired t-test) higher morphology ratio
(1.35fl.03) than specimens which did not fracture (H34, H35: morphology
ratio=1.268:l:0.02).

Theoretical Results

Stresses in the cartilage and trabecular bone were not statistically different with

rigid or padded interfaces. In the subchondral bone for both padded and rigid interface

experiments, the minimum principal stress field was compressive and minimal compared to

100

the maximum shear stress and maximum principal stress, which was tensile. The
maximum principal stress was tensile. Maximum principal (‘Normal’) and shearing
stresses occurred in the subchondral bone and were significantly reduced with padding
(Table 5). The stress distributions presented in Figures 14, 15 (H36) are typical stress
distributions. Light areas indicate regions of larger magnitude stresses (tensile: 55-145
MPa; shear >40-80 MPa) and dark areas represent lower magnitude stresses (<30 MPa).
The distributions are restricted to the posterior half of a transverse section through the
patella.

The high stresses documented in the subchondral bone in the rigid interface
experiments (Figure 16,a) were due to either global (Figure 16,b,c) of the patella or local
(Figure 17) bending of the subchondral plate. Maximum tensile and shear stresses were
documented on the lateral facet and at the median ridge. Compressive pressure fiom
contact with the femur that was medial or lateral with respect to the area of contact with
the rigid impactor on the anterior surface apparently caused the global bending
phenomenon. In all cases, global bending was maximal on the lateral facet and coincided
with the highest levels of stress in the subchondral bone. Global bending on the medial
facet was minimal and the state of stress was reduced in this region. Altemately, local
bending was observed in regions of high contact pressure. Due to the low compressive
stress field, the distribution of maximum tensile and shearing stresses appeared to be
consistent in each model of rigid interface experiments. Because the average maximum

tensile and shear stresses were both near the failure stresses (shear, oy=68 MPa; tension,

101

oy=l33 MPa, (Hayes 1991)), it was not clear if injury was related to tensile or shearing
stresses.

Padding supported the entire anterior surface of the patella, eliminating global
bending and reducing local bending of the subchondral plate. This resulted in an 80%
decrease in the normal stress and a 70% decrease in the maximum shear stress (Table 5,
Figure 15). With padding the magnitude of these stresses was significantly reduced and
more nearly approximated the magnitude of the minimum principal stress. Therefore, the
distribution of maximum tensile and shearing stresses was different for each padded
interface experiment.

DISCUSSION

The specific aims of the current study were to experimentally and theoretically
investigate the influence of padding on the impact response of the human knee. Padding
was experimentally and theoretically shown to prevent gross bone fi'acture and occult
rnicrodamage and to limit normal and shearing stresses in the subchondral bone plate of
the patella. Paired impact experiments were conducted with rigid and padded interfaces
to limit test variability. Both knees fi‘om a given cadaver were impacted at approximately
the same impact load. While the padding better reflects more realistic impact scenarios
(i.e. knee impacting the instnnnent panel), it also served to significantly change the
boundary conditions at a given impact load. The rigid impact interface only contacted a
small portion of the central anterior patella. Independent, outboard reactions fiom the
medial and lateral femoral condyles resulted in approximately three point global bending of

the patella. This may explain the elevated stresses in the subchondral plate for rigid

102

interface experiments. Altemately, padding provided greater support across the entire
anterior surface of the patella resulting in less bending and drastically reduced stresses in
the subchondral bone. This may help explain the genesis of microcracking in the
subchondral bone.

Hayashi and coworkers (1996) also conducted impact experiments on the 90°
flexed knee using different impact interfaces. In that study, a rigid interface and a variety
of Hexcel interfaces were used to conduct several experiments with each interface at a
fixed impact energy. A direct comparison to the current study may not be possible
because in a majority of the experiments the same interface was used on both limbs of the
test subjects. In general, however, Hayashi, et al. (1996), showed similar results to the
current study whereby padding can prevent transverse patellar fractures.

In the current study, normal and shearing stresses in the cartilage and trabecular
bone were not different between rigid and padded interface experiments, however, the
subchondral bone state of stress was significantly reduced with padding. Rigid interface
experiments resulted in gross bone fi'acture in two specimens and occult microcracking in
the subchondral bone was detected in three specimens. No gross fracture of bone or
occult rnicrodamage was detected in contralateral padded experiments. Further, because
the impact loads were not different in paired experiments, and the state of stress in the
cartilage and trabecular bone were consistent, these data may suggest that gross fracture
of bone initiates in the subchondral plate. This assumes that patellar fracture initiates near
the retropatellar surface. Viano and Stalnaker (1980) impacted whole, human cadavers

and documented patella fi'actures which appeared to emanate from the posterior surface.

103

This is supported by the results of our math model which predicts high stresses, both
gobally and locally, in the subchondral bone on the posterior surface.

The patella has been documented to predominantly fracture transversely in both
past investigations (Hayashi 1996, Melvin 1975, Patrick 1965, Powell 1975, Atkinson:a
1995, Atkinson:b 1995, Haut 1989) and in the current study. This is suggestive of a
tensile failure in which the tension is directed proximal-distal. Such tensile stresses are
predicted in a stress analysis of the whole patella conducted by Minns and coworkers
(1979). The model also predicts high shear stresses near the bone cartilage interface as
seen in the current study. These results suggest the subchondral bone may be acting like a
shell subjected to an externally applied compressive traction through contact with the
femur. Because the patellofemoral contact area is much wider medial-lateral than in the
transverse proximal-distal direction, the very stiff subchondral bone is essentially subjected
to a narrow medial to lateral loading in the middle of the patella. The model used in the
current study may be somewhat limited because the model geometry is taken from a
transverse plane and the model is analyzed with the assumption of plane strain.

Even though a consistent flexion angle and impact energy were used with all rigid
impact interface experiments, a range of peak contact loads and injuries were documented
in the current study. These differences may be related to subj ect-to-subject variability’s in
the strength of the bone and cartilage, the relative size of the knees and the relative
position of the patella to the femoral condyles during the impact event. Because the
premartem state of the subjects can involve extended illnesses, long periods of bedrest

may be expected to lead to some level of tissue atrophy. As such, the peak loads may be a

104

somewhat conservative measure of the current driving population (Viano 1980). While
admittedly unlikely, if a consistent knee stiffness was true for all subjects, the size of the
knee may then be expected to result in peak loads which scale with the size of the knee.
Even with the same knee stiffness and size, more subtle joint geometry differences may
affect joint impact response. Hence, the geometry of the patellae in the current study was
quantified with the so-called ‘morphology ratio’ in effort to determine if impact
biomechanics and the state of stress predicted by the math model would vary with patella
geometry in a random sample of six human subjects. The morphology ratio is easily found
as the ratio of two linear dimensions of patella bone and cartilage. While other
investigators have used more complex methods (i.e. topographical maps,
stereophotogrammetry) (Minns 1975, Mcleod 1977, Eckstein 1992, Ateshian 1994,
Blenkevoort I991, Ateshian 1991) to quantify joint geometry, the morphology ratio was
sufficiently fine to detect differences in peak load and the incidence of gross bone fracture.
Altemately, gross test subject parameters (i.e. height, weight, sex, age) did not
significantly correlate with impact biomechanics in the current study or in past
investigations (Powell 1975, Atkinson:b 1995). While considering which independent
variables might correlate best with impact biomechanics and injury, it is interesting to note
the predictive power of the logistical regression used in the current study. For isolated,
90° flexed human knee impact experiments at 27J of energy, the regression predicted a
high chance of occult microcracking while accepting a 50% chance of gross bone fi'acture.
Occult microcracking was observed in four of six specimens (67%) and gross bone

fracture in three of six (50%) specimens.

105

Past in vivo studies by our group have developed an experimental rabbit model of
post-traumatic osteoarthrosis (Atkinson 1996, Newberry 1996, Shelp 1994, Newberry
1995, Haut 1993, Newberry 1995, Newberry:b 1996). Subfracture impacts were
delivered at two levels. ‘Low’ severity impacts did not progress to osteoarthrosis, while
‘high’ severity resulted in higher tissue stresses and did progress to a chronic joint disease.
Theoretical models of these experiments help to define safe and harmful levels of joint
tissue stress leading to chronic problems (T able 6). While these data only represent two
levels, it may be assumed that the incipient state of stress necessary for pathogenesis of a
joint disease may lie somewhere between the low and high severity impact stresses. In the
current study, shear and normal stresses in the subchondral plate fi'om rigid interface
experiments (Table 7) were approximately 30% and 300%, respectively, greater than the
stresses shown to cause disease in the rabbit. While padding significantly limited these
stresses, they were still not reduced to safe (i.e. ‘low’ severity) levels established in our
rabbit model. This suggests a more optimized padding system might be utilized to further
reduce these stresses to ‘safe’ levels. This might be accomplished by allowing greater
intrusion into the padding system and making contact with the femoral condyles.

The results of the current study suggest the importance of load distribution over
the knee during impact which may have implications for future instrument panel designs
and the current lower extremity criterion. Two instrument panel design factors appear to
be important for occupant injury prevention: 1) the stiffness of the instrument panel should
be optimized to distribute the load over the knee (and possibly the femoral condyles) to

sufficiently reduce stress levels to prevent acute gross (i.e. bone fi'acture, ligamentous

106

injury) and micro- injuries (occult microcracking) and chronic joint pain and morbidity
(post-traumatic osteoarthrosis (Newberry 1995)), 2) local incongruities (i.e. plastic heater
vents) should be located in areas where knee-instrument panel contact is unlikely (States
1986). In an effort to validate the above suggestions, quantitative field studies of
instrument panel designs and associated lower extremity injury frequency are needed to
identify design features which aid occupant safety. Experimental methods of evaluating
instrument panels may also prove helpfirl in making additional, more controlled
comparisons. In 1969, Wilson developed a self-contained impact apparatus which could
be mounted in a vehicle and simulate knee impact into the instrument panel. A
hemispherical knee analog mounted on a pendulum impactor head was directed at the
instrument panel at variable velocities. A similar method might prove useful in future
studies.

The current lower extremity criterion used by the automotive industry is based
solely on peak impact load and does not take into account how the load is distributed over
the knee. The theoretical and experimental data from the current study suggest that bone
and cartilage failures are related to the state of stress in the tissue and not solely the
magnitude of the impact load on the knee. While it may be difficult to measure stresses in
bone and cartilage in simulated accidents using anthropomorphic dummies, a more indirect
approach may prove useful. From the current study it appears that anterior patellar
contact area relates to the damaging stresses in the subchondral plate. By instrurnenting
dummy knees with a transducer to measure contact area (i.e. pressure sensitive film), peak

forces on the knee may be indirectly correlated to tissue stresses. This data could be used

107

to predict the acute (i.e. gross and occult fracture of bone) and chronic (i.e. OA) response
of the typical knee joint to impact loading. Future studies will be needed to develop a
better understanding of the relationship between contact area, load and tissue stresses.
ACKNOWLEDGMENTS

This study was supported by a grant (R49/CCR503607) from the Centers for Disease
Control and Prevention. 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 appreciate the support of Hexcel] corporation for their technical and materials
support. The authors wish to gratefirlly acknowledge Ms. Jane Walsh for help in the
interpretation of histological slides and Mr. Cliff Beckett for technical support. We are
also indebted to Mr. Rick Shumaker and the Michigan Tissue Bank for solicitation of

donors and retrieval of joints.

REFERENCES

1. Anderson D, Brown T, Radin E (1991) Stress wave effects in a finite element analysis
of an impulsively loaded articular joint. MechE, Part H: J Eng Med 205: 27-34

2. Anderson DD, Brown TD, Yang KH, Radin EL (1990) A dynamic finite element
analysis of impulsive loading of the extension-splinted rabbit knee. Journal of
Biomechanical Engineering 112: 119-128

3. Amoldi C (1991) Patellar Pain. Acta Orthop Scand 62: 1-29

4. Ateshian G, Kwak S, Soslowsky L, Mow, V (1994) A stereophotographic method for
determining in situ contact areas in diarthrodial joints, and a comparison with other
methods. Journal of Biomechanics 27(1): 111-124

5. Ateshian GA, Soslowsky LJ, Mow VC (1991) Quantitation of articular surface

topography and cartilage thickness in knee joints using stereophotogrammetry. Journal
of Biomechanics 24(8): 761-776

108

10.

ll.

12.

l3.

14.

15.

l6.

17.

18.

Atkinson PJ, Haut RC (1995) Subfiacture Insult to the Human Cadaver Patellofemoral
Joint Produces Occult Injury. Journal of Orthopaedic Research 13: 936-944

Atkinson P, Haut R (1995) Insult to the human cadaver patellofemoral joint: effects of
age on fracture tolerance and occult injury. 39th Stapp Car Crash Conference 281-294

Atkinson P, Newberry W, Staton T, Garcia J, Altiero N, Haut R (1996) Animal and
human studies on injury mechanisms during blunt insult to the knee. 6th CDC Injury
Prevention Through Biomechanics 137-160

Atkinson P, Haut R (1997) A method for determining regions of diarthrodial joint
contact during dynamic loading of the human knee. Trans Orthop Res Soc 656

Blankevoort L, Kuiper JH, Huiskes R, Grootenboer HJ (1991) Articular contact in a
three dimensional model of the knee. J Biomechanics 24: 1019-1031

Bourret P, Corbelli S, Cavallero C (1977) Injury agents and impact mechanisms in
fiontal crashes about 350 in the field accidents. 21st Stapp Car Crash Conference 215-
258

Chapchal G (1978) Posttraumatic Osteoarthritis after Injury of the Knee and Hip Joint.
Reconstr Surg Traumat 16: 87-94

Crandall J, Kuhlmann T, Pilkey W (1995) Air and lmee bolster restraint system:
laboratory sled tests with human cadavers and the Hybrid III dummy. J Trauma 38:
517-520

Davis M, Ettinger W, Neuhaus J, Cho S, Hauck W (1989) The Association of Knee
Injury and Obestiy with Unilateral and Bilateral Osteoarthritis of the Knee. American
Journal of Epidemiology 130(2): 278-288

Donahue MJ, Buss D, Oegema T, Thompson R (1983) The Effects of Indirect Blunt
Trauma on Adult Canine Articular Cartilage. The Journal of Bone and Joint Surgery
65-A(7): 948-957

Eckstein F, Muller-Gerbl M, Putz R (1992) Distribution of subchondral bone density
and cartilage thickness in the human patella. Journal of Anatomy 180: 425-433

Gardner DL (1994) Problems and paradigms in joint pathology. J Anat 184: 465-476

Grelsarner R, Proctor C, Bazos A (1994) Evaluation of patellar shape in the sagittal
plane: A clinical analysis. Amer J Sports Med 22: 61 -66

109

l9.

Hamerrnan D (1989) The Biology of Osteoarthritis. New England Journal of Medicine
320(20): 1323-1330

20. Hartemann F, Thomas C, Henry C, Foret-Bruno J, Faverjon G, Tarriere C, Got C,

21.

22.

23.

24.

25.

26.

27.

28.

29.

Patel A (1977) Belted or not belted: The only difference between two matched
samples of 200 car occupants. 2lst Stapp Car Crash Conference 97-150

Haut R (1989) Contact Pressures in the Patello-femoral Joint during Impact Loading
on the Human Flexed Knee. Joumal of Orthopaedic Research 7: 272-280

Haut R, Shelp D, LI, X, Altiero N (1993) Studies on the impact response of articular
cartilage to blunt impact. Symposium Proceedings: Centers for Disease Control 3:
177-200

Hayashi S, Choi H, Levine R, Yang K, King A (1996) Experimental and analytical
study of knee fracture mechanisms in a frontal knee impact. 40th Stapp Car Crash
Conference 161-171

Hayes W (1991) Biomechanics of cortical and trabecular bone: implications for
assessment of fracture risk. In: Basic orthopaedic biomechanics, ed Mow and Hayes
93-142

Hering W, Patrick L (1977) Response Comparisons of the Human Cadaver Knee and
a Part 572 Dummy Knee to Impacts by Crushable Materials. Twenty-First Stapp Car
Crash Conference 21: 1015-1053

Huberti H, Hayes W, Stone J, Shybut G (1984) Force ratios in the quadriceps tendon
and ligamentum patellae. Journal of Orthopaedic Research 2: 49-54

Huelke D, O'Day J, States J (1982) Lower Extremity Injuries in Automobile Crashes.
Accid Anal & Prev 14: 95-106

Johnson-Nurse C, Dandy D (1985) Fracture-Separation of Articular Cartilage in the
Adult Knee. The Journal of Bone and Joint Surgery 67-B(10: 42-43

Kapelov R, Teresi L, Bradley W, Bucciarelli N, Murakami D, Mullin W, Jordan J
(1993) Bone Contusions of the knee;lncreased Lesion Detection with fast Spin-Echo
MR Imaging with Spectroscope Fat Saturation. Radiology 901-904

30. Kellgren J, Lawrence J (1958) Osteoarthritis and disk degeneration in the urban

31.

population. Annals of Rheumatory Diseases 17: 386-396

King A, Viano D, Mizeres N, States J (1995) Humanitarian benefits of cadaver
research on injury prevention. J Trauma 38: 564-569

110

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

45.

Li, X, Haut R, Altiero N (1995) An analytic model to study blunt impact response of
the rabbit P-F joint. Journal of Biomechanical Engineering

Lund A, Ferguson S (1995) Driver Fatalities in 1985-1993 Cars with Airbags. Journal
of Trauma 38: 469-475

Mackay G, Hill I (1995) The limitations of current seatbelts in Europe - same
population considerations. J Trauma 38: 533-537

MacKenzie EJ, Siegel JH, Shapiro S, eta]. (1988) Functional Recovery and Medical
Costs of Trauma: An analysis by Type and Severity of Injury. Journal of Trauma 28:
281-298

Malliaris A, Hitchcock R, Hedlund J (1982) A search for priorities in crash protection.
Proceedings of the 1982 SAE International Congress and Exposition 959-991

Mcleod W, Moschi A, Andrews J, Hughston J (1977) Tibial plateau topography. The
American Journal of Sports Medicine 5(1): 13-18

Melvin J, Stalnaker R, Alem N, Benson J, Mohan D (1975) Impact Response and
Tolerance of the Lower Extremities. Stapp Car Crash Conf 19: 543-559

Minns R, Bremble G, Campbell J (1975) The geometric properties of the human tibia.
Journal of Biomechanics 8: 253-255

Minns RJ, Birnie AJM, Abemethy PJ (1979) A stress analysis of the patella, and how
it relates to patellar articular cartilage lesions. J Biomechanics 12: 699-71 1

Morgan R, Eppinger R, Hennessey B (1991) Ankle joint injury mechanism for adults
in fiontal automitve impact. 35th Staoo Car Crash Conference 189-198

Mow V, Holmes M, Lai W (1984) Fluid transport and mechanical properties of
articular cartilage a review. Journal of Biomechanics 17: 377-394

Nagel D, States J (1977) Dashboard and bumper knee-will arthritis develop? AAAM
proceedings 21: 272-27 8

.Newberry W, Atkinson P, Li X, DeCamp C, Altiero N, Haut R (1995) Tissue injuries

resulting from blunt impact on the knee. 5th Injury prevention through biomechanics
symposium

Newberry W, Shelp D, Coles A, Haut R (1995) Chronic alterations in the
physiologically relevant properties of cartilage after mechanical insult. Proceedings
4lst Meeting of the Orthopaedic Research Society

111

46. azNewberry W, Haut R (1996) The effects of subfracture impact loading on the
patellofemoral joint in a rabbit model. 40th Stapp Car Crash Conference 149-159

47. szewberry WN, Zukosky DK, Walsh JC, Haut RC (1996) Blunt trauma to a joint
produces surface damage to cartilage with subsequent changes in subchondral bone.
ORS 749

48. Nyquist G, King A (1985) Lower extremities. Rev Biomech Impact Res Inj Auto Env
6: 163-201

49. Otte D (1996) Biomechanics of lower limb injuries of belted car drivers and the
influence of intrusion and accident severity. 40th Stapp Car Crash Conference 193-206

50. Patrick L, Kroell C, Mertz H (1965) Forces on tehehuman body in simulated Crashes.
Stapp Car Crash Conf 9: 237-259

51. Pattimore D, Ward E, Thomas P, Bradford M (1991) The nature and cause of lower
limb injuries in car crashes. 35th Stapp Car Crash Conference 177-188

52. Powell W, Ojala S, Advani S, Martin R (1975) Cadaver Femur Responses to
Longitudinal impacts. Stapp Car Crash Conf 19: 561-579

53. Radin E, Ehrlich M, Chemack R, Abemethy P, Paul 1, Rose R (1978) Effect of
repetitive impulsive loading on the knee joints of rabbits. Clinical Orthopaedics and
Related Research 131: 288-293

54. Radin B, Parker H, Pugh J, Steinberg R, Paul 1, Rose R (1973) Response of joints to
impact loading-III relationship between trabecular microfractures and cartilage
degeneration. Journal of Biomechanics 6: 51-57

55. Radin E, Paul 1, Rose R (1972) Role of Mechanical Factors in Pathogenesis of Primary
Osteoarthritis. The Lancet 3: 519-522

56. Setton L, Zhu W, Mow V (1990) BIphasic poroviscoelastic behavior of articular
cartilage. First World Congress of Biomechanics 17

57. Shelp D, Newberry W, LI, X, Dalimonte R, DeCamp C, Altiero N, Haut R (1994)
Tissue damage resulting from blunt impact on the knee: animal and cadaver studies.
4th Inj prevention through biomechanics symposium 4: 43-62

58. Shimizu M, Tsuji H, Matsui H, Katoh Y, Sana A (1993) Morphometric Analysis of
Subchondral Bone of the Tibial Condyle in Osteoarthrosis. Clin Orthop 229-239

112

59.

60.

61.

62.

63.

65.

66.

67.

68.

69.

70.

Siegal J, Mason-Gonzales S, Dischinger P, Cushing B, Read K, Robinson R, Smialek
J, Heatfield B, Hill W, Bents F, Jackson J, Livingston D, Clark C (1993) Safety belt
restraintsand compartment intrusions in fi'ontal and lateral motor vehicle crashes:
mechan

States J (1970) Traumatic Arthritis-- A Medical and Legal Dilemma. Ann Conf of the
Amer Assoc for Automotive Med 14: 21-28

States J (1986) Adult occupant injuries of the lower limb. In: Biomechanics and
medical aspects of lower limb injuries. Warrendale, PA: SAE 97-107

Thomas P, Charles J, Fay P (1995) Lower limb injuries-the effect of intrusion, crash
severity and the pedals on injury risk and injury type in frontal collisions. 39th Stapp
Car Crash Conference 265-280

Thompson RC, Oegema TR, Lewis JL, Wallace L (1991) Osteoarthritic changes after
acute transarticular load. The Journal of Bone and Joint Surgery 73-A(7): 990-1001

. 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: 150-
152

Vellet A, Marks P, Fowler P, Munro T (1991) Occult Posttraumatic Osteochrondral
Lesions of the Knee: Prevalence, Classification, and Short Term Sequelae Evaluated
with MR Imaging. Radiology 271-276

Viano D (1995) Restraint effectiveness, availability and use in fatal crashes:
implications to injury control. The Journal of Trauma 38:538-546

Viano D (1977) Considerations for a femur injury criterion. let Stapp Car Crash
Conference 445-473

Viano DC, Stalnaker RL (1980) Mechanisms of Femoral Fracture. Journal
Biomechanics 13: 701-715

Volpin G, Dowd G, Stein H, Bentley G (1990) Degenerative arthritis after intra-
articular fi’actures of the knee. Long-term results. J Bone Joint Surg Br 72: 634-638

Wilson W (1969) Evaluating knee-to-instrument panel impacts. Stapp Car Crash
Conference 13: 134-148

113

TABLES

Table 1: Lower extremity bone fracture injuries by site; data taken from past

experimental studies (Patrick 1965, Melvin 1975, Powell I 975, Viano I980)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Impact # of Patella Femoral F ernoral Femoral Pelvis Total
Interface tests Condyles Shaft Neck Injuries
Rigid 32 27 14 10 4 l 56 injuries
Interface
Incidence” 84% 44% 25% 13% 3%
Padded 36 7 1 1 7 0 5 30 injuries
Interface
Incidence“ 19% 30% 19% 0% 14%
& Incidence = # of injuries/# of tests
Table 2: Materialpropertiesfor math model
Tissue Modulus (MPa) Poisson Ratio Reference
Cartilage 20 0.49 Repo/ Atkinson
Cortical/Subchondral Bone 3750 0.3 Mente
Trabecular Bone 300 0.3 Townsend

 

 

 

 

 

 

Table 3: Paired (right-left) subject impact biomechanical data with rigid and padded
impact interfaces

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Inter 3'ng padded rigid padded pigid padded p'gid padded
face
:9
peak contact p-f contact p-f contact Knee Joint Injury Knee Joint
farce pressure area Injury
Spec (kN) (kN) (MP8) (MP8) (mmz) (mmz) Gross. Micro“ Gross Micro
H31 6103.4 6861.4 22.6 13.3 381.6 241.3 None HOMc None None
H32 4654.3 4788.2 11.4 6.4 217.4 99.4 TPFx HOMc None None
H33 3324.5 5605.3 11.3 14.9 301.6 161.7 TPFx None None None
H34 6845.6 6716.8 8.2 11.1 509.3 263.6 None None None None
H35 5839.1 7082.6 12.6 16.9 305.7 251.1 None H&VO None None
Mc

H36 3211.8 3860.6 14.8 5.9 260.7 260.9 TPFx, H&VO None None

FCFx Mc
aver 4996.4 5819.1 13.5 11.4 329.4 213.0 3TP 4HO 0 0
age Fx, Mo,

1 F C Fx 2 V 0

Mc

std 1513.4 1298.6 4.9 4.5 103.6 67.3
dev

 

 

* T P F x = Transverse patellar fracture, F C Fx = F femoral condyle fracture,
** H 0 Mc, V 0 Me = Horizontal or Vertical Occult Microcracking (Figure 6)

 

114

 

Table 4: Geo

H31

Er

_—

 

H32

 

H33

 

H34

 

H35

 

H36

 

 

 

. * See Figgre

 

Interface

3

Table 5: ME

pg

 

Tissue:

 

 

Stresses

:9

 

average

 

 

std dev

 

 

 

Table 4: Geometrical variation of test subjectjatellae

 

 

 

 

 

 

 

 

 

 

 

Specimen Morphology Ratio (a/b)* Type Classification *
H31 1.540 Type II

H32 1.315 Type I

H33 1.380 Type]

H34 1.277 Type I

H35 1.252 Type 1

H36 1.346 Type I

* See Figure 4 and methods section for explanation

 

 

Table 5: Maximum normal and shear stresses (MPajfrom the math models .....................

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Interface rigid padded rigid padded ngid padded
=>
Tissue: Cartilage Subchondral Bone Trabecular Bone
Stresses Norm Shear Norm Shear Norm Shear Norm Shear Norm Shear Norm Shear
:9

average 6.9 6.5 2.0 5.7 94.4 54.8 26.3 25.2 10.1 8.7 2.1 6.6
std dev 5.5 3.2 2.6 2.1 43.5 22.4 21.4 15.5 10.2 4.7 3.3 4.1
Table 6: Stresses from low and high severity impact

Tissue Impact Severity Max Shear Stress (MPa) Max Norm Stress (MPa)

Cartilage Low 6.0 6.3

Cartilage High 12.9 16.1

SC Bone Low 16.9 12.6

SC Bone High 41.8 28.9
Table 7: Maximum stresses from current study

Max Shear Stress (MPa) Max Norm Stress (MPa)

Tissue Rigid Padded Rigid Padded

Cartilage 6.5 5.7 6.9 2.0

SC Bone 54.8 25.2 94.3 25.2

 

 

 

 

 

 

 

115

 

FIGURES

 

Probability of Bone Fracture vs Impact Energy

 

 

 

 

 

 

100

 

Y ' ' so
ENERGY (J)

 

 

Figure 1

rubber Interface
\
1 w

"

    

 

 

 

 

 

 

  

Figure 2

116

 

“0"-'U""’_CZ ‘D‘ODI'I'I

 

F
i

 

4 6 8 10 I
PeakflContact From ngig lnterace (kN)?

 

Figure 3

Cartilage
0"
93
Bone

Figure 4: Side view of patella

117

 

Figure 5: Rigid interface model

'h
10"

I
’ t'” .7 n- "~
. «safe-5W
v. as ....§'::.
.¢ 1

I"?
DA
f 0

f1:
.1-
0 Q
8..
:00
1=II3

'1 .

I!
e

.Skbr
a.
3...!-
s

.1
9
9
c

v

‘e

 

Figure 6: Padded interface model

. padded

8

6
time (ms)

 

118

 

Figure 9: Region of patellofemoral contact on the patella. Arrows denote a transverse
fracture.

119

 

Figure 11: Combination occult trauma: horizontally (lefi) oriented occult trauma near a
vertically oriented occult microcrack (right) (40X).

120

 

Figure 12: Horizontally oriented occult microtrauma (40X).

Peak Contact Force (kN)

 

 

7

6

5

 

.25

y = -27.229x + 40.554
r2 = 0.79

1.3 1.35
Morphology Ratlo

 

Figure 13

121

 

 

 

   
   

Maximum
Principal Stress

Maximum
Shear Stress

 
  

Fig 141 ngi 0
Interface

 

 

Arrows point to regions of maximum stress.

 

     
 
 
  

Maximum
Principal Stress

   

Figure 15:
Padded Interface
Maximum Shear
Stress

 

 

 

   

Rigid impactor

Patellofemoral reaction

c) .Tens.

Figure 16: Global bending with a rigid interface

 

Trabecular bone

Subchondral platea

 
 
 
 

 

 

Local bending of
the subchondral
plate

 

 

 

Figure 17: Patellofemoral contact and local bending

123

CHAPTER 7

HUMAN KNEE IMPACT INJURY THRESHOLDS AND
LOCATION VARY WITH FLEXION ANGLE AND IMPACT INTERFACE.

Patrick Atkinson, Jose Garcia, Hasnin Ali', Nicholas Altiero’, Roger Haut
Orthopaedic Biomechanics Laboratories
*Department of Material Science and Mechanics
Michigan State University

ABSTRACT

Injury to the knee, thigh and hip is currently based on a bone fracture criterion
derived from human cadaver impact experiments. Typically a rigid impact interface is
used to deliver a blunt insult to the 90° flexed knee. 10 kN is the maximum load allowed
during an automotive crash using anthropomorphic dummies. More recent studies,
however, suggest that the impact interface and the knee flexion angle influence its
response to impact. In the current study, bilateral impacts were delivered to isolated knee
preparations of thirty-six cadavers. The cadavers were separated into two test groups. In
each test group, subgroups of 6 cadavers were assigned to flexion angles of 60, 90, or
120 degrees. One knee from each subject was impacted at increasing energy levels until
gross fracture of bone. In the first group of 18 cadavers, the contralateral knee was
impacted once with a rigid interface at a subfracture level. In the second group, the
contralateral knee was impacted once with a padded interface (Aluminum Hexcel) at

approximately the fracture load in the contralateral knee. The rigid interface experiments

124

showed that increasing flexion angle was associated with higher fracture loads. The
location and mode of injury varied with flexion angle. At 60° of flexion, there was an
avulsive failure of the patellar tendon from the distal pole of the patella, at 90° a
transverse patellar fracture located centrally on the patella, and at 120° transverse
fractures of the proximal pole of the patella, comminuted fractures of the femoral
condyles and tibial plateau fractures. Subfracture experiments typically showed occult
microcracks -at the bone-cartilage interface in regions showing gross fracture in
contralateral knees. Padding limited gross and microscopic injury at 90° and 120°
flexion, however, injuries still occurred for 60° of flexion. 3-D mathematical models of
the patella indicated high shear and tensile stresses in the contact regions corresponding
to regions of injury. This study showed that tissue stresses and injury thresholds vary
with flexion angle and impact interface.
INTRODUCTION

Lower extremity injury is a significant problem which can lead to considerable
hospital stays and extended periods of lost work days (MacKenzie 1988, Huelke 1982).
Lower extremity injuries resulting from automotive accidents are second in frequency
only to head injuries (States 1986). Within the lower extremity, the knee is the most
injured body region due to high speed contact with the instrument panel (Dischinger
1992). Such injurious contact can also occur in sporting activities and falls (States 1970).
The resulting injury can be at a fracture or subfi’acture level, however, subfracture
injuries are reported four times as frequently as gross fractures. Unfortunately, the small

size and typically occult presence of subfracture injuries can preclude their acute

125

detection and treatment (Vellet 1991, Kapelov 1993). Clinically, such injuries may be
treated as a simple contusion (T omatsu 1992) or missed entirely. Without proper
treatment, such injuries can lead to chronic joint pain and disease, such as post-traumatic
osteoarthrosis (Upadhyay 1983, States 1970, Donohue 1983, Thompson 1991).

Several parameters have been hypothesized to be important factors influencing the
potential for impact induced injuries in the knee. First, studies have documented a link
between the flexion angle of the knee and its tolerance to injury. In automotive accident
field studies, Dischinger, et al., (1995) found that shorter automotive occupants of both
sexes are more than twice as likely to injure the knee than taller occupants. This may be
due to the seat placement for shorter occupants which results in elevated knee flexion
angles which might reduce the injury tolerance of the knee. Viano, et al., (1978) used
formed interfaces to impact the human cadaver knee and tibia at flexion angles of 45°
and 90°. A consistent impact energy was used for all tests and only one impact
experiment was conducted for each knee. At 90°, the impact load was 7 kN and gross
injuries to the bones and ligaments occurred in all but one specimen. At 45°, the peak
load was significantly lower (3.65 kN) and there were no gross injuries. Haut (1989)
conducted similar experiments at flexion angles of 60°, 90°, and 110° but restricted the
impact to the patellofemoral joint. In addition, progressive experiments were conducted
on the same knee at increasing impact energies until gross fracture was noted. Similar to
Viano, et al., the fracture load was observed to increase with flexion angle.
Unfortunately, in both of these studies there was no investigation of subfracture injuries.

Another parameter which affects knee impact injury is the stiffness of the material

which contacts the knee. Extensive experiments have been conducted on seated, fi'esh

126

cadavers (Patrick 1965, Melvin 1975, Powell 1975, Viano 1980) in which impact loads
were directed at the 90° flexed knee with rigid and padded impactors. Rigid impactors
resulted in transverse fractures of the patella, split fractures of the femoral condyles, and
mid-femoral shaft fractures in 84%, 44%, and 25% of cases, respectively. Padding
reduced fracture to the patella by 65%, to the femoral condyles by 14%, and to the
femoral shaft by 6%. Again, however, there was no investigation of subfracture injuries
in these studies.

Contact between rigid impact interfaces and the bony anterior surface of the patella
likely result in a localized load distribution. The flesh thickness over the anterior patella
is approximately 4.8 mm (Hering 1977), offering little protection to a rigid impact.
Altemately, padding likely produces increased load distribution over part or all of the
patella and possibly the femoral condyles. The localized load distribution associated
with rigid interfaces has been implicated in the high incidence of injuries with the rigid
interface (Hayashi 1996, Powell 1975). This suggests that the boundary conditions on
both the anterior and posterior patella may be important in determining the injury
mechanism of patellar fracture. There is some experimental evidence that patellar
fractures probably initiate on the retropatellar surface (V iano 1980), however, the exact
mechanism of injury is unknown.

While the exact mechanism of fracture injury is unknown, the mechanism of
subfracture injuries has been the subject of theoretical and experimental studies. These
studies suggest that microfractures of bone and cartilage may be mediated by elevated
shear and/or tensile stresses. Theoretical analyses of generalized diarthrodial joint

contact (Eberhardt 1990, Ateshian 1994) document elevated shear stresses in the cartilage

127

 

at the osteochondral junction as a result of sudden joint loading. These studies
hypothesize that this might explain occult osteochondral lesions. Elevated shear stresses
at the cartilage surface have also been implicated in the development of cartilage surface
fissures. Atkinson, et al., (1998) conducted impact experiments on the tibial plateau
using rigid, spherical indentors of various radii. Smaller indentor radii produced a higher
fiequency of surface lesions which were associated with elevated shear stresses at the
cartilage surface. Larger radii created elevated shear stresses at the bone-cartilage
interface near the edge of contact suggesting that larger contact areas might be associated
with deeper lesions. Because the location and intensity of stresses vary with contact
radii, these data suggest that knee joint flexion will likely affect tissue stresses. This
results from the changing contact radii between the patella and femur which vary as a
function of the flexion angle. Studies of passive joint loading document maximal contact
area at 90° of flexion which decreases at 120° and 60° (Huberti 1984). This suggests that
extreme flexion angles may produce elevated shear stresses at the cartilage surface due to
the disparate contact radii producing the reduced intraarticular contact areas (Atkinson
1998). Altemately, the larger contact areas would indicate contact between larger
contact radii which suggest elevated stresses at the bone-cartilage interface.

Subfracture injuries have been documented in knee impact studies in canine and
porcine experimental models with the knee flexed 100°-110°. Saggital sectioning reveals
occult, horizontal and vertical splitting at the tidemark, and fissures of the articular
surface (Tomatsu 1992, Armstrong 1985, Thompson 1991, Donohue 1983). In the
canine, these injuries are in the patella and are located within the region of intraarticular

contact. For the canine at 100° of flexion, the retropatellar contact area is a narrow band

128

at the distal pole and extends medial-lateral across the patella. However, differences in
the articular geometry of the canine and human knee are such that the contact region in
the canine at 100° of flexion is different from the human at the same flexion angle. This
makes a direct application of the experimental animal data to the human difficult.

There have been a limited number of subfracture studies on the human. These
experiments were conducted at 90° of flexion (Atkinson 1995, Atkinson 1997). Unlike
the animal studies, the patellae were transversely sectioned for histological analysis.
Occult, horizontal injuries were documented in the region of intraarticular contact which
is located in the middle of the patella for the human. The observed absence of fissures or
occult, vertical occult microcracks, as seen in the animal studies, may have been due to
the plane of sectioning or the shape of the joint. For example, the contact region in the
canine which is associated with subfracture injuries (100°) occurs in the human at lower
flexion angles (360°) (Huberti 1984). It is unknown what effect low and high flexion
angles will have on the risk of subfracture injuries in the human knee.

In the current study, subfracture and fracture impact studies were conducted at
flexion angles of 60°, 90°, and 120°. Rigid and padded impact interfaces were used to
alter the load distribution over the knee. Detailed histological analyses were conducted
to identify the type and location of subfi'acture injuries. A theoretical model of the
patella was used to predict the stresses associated with these experiments. Specifically,
we hypothesized that: 1) impact loads would increase with flexion angle, and 2)
subfi'acture injuries at lower flexion angles would include cartilage fissures and occult,
microfiactures, similar to experiments in the canine, and 3) increased load distribution

would reduce the incidence of knee injury.

129

METHODS AND MATERIALS

Experimental protocpl

Blunt impact was delivered to isolated, flexed knee joints of 36 human cadavers. The
knee joints were randomly assigned to subfracture and fracture level experiments (Table
1). The specific test protocols are provided later. Shortly after death, the patellofemoral
joint was examined through a medial incision. Tissues were procured from donors with
no prior history of joint trauma and/or evidence of advanced joint pathology (areas of
denuded bone, advanced chondromalacia, etc.). Joint preparations were excised
approximately 15 cm from the knee proper and stored at -20° C until the day of testing.
The tissues were allowed to thaw at room temperature ovemight prior to testing.
Superficial tissues proximal to the femoral condyle were excised and the femoral shaft
was cleaned with alcohol. The femur was potted in a cylindrical steel sleeve with room
temperature curing epoxy, and was mounted horizontally to a rigid test frame (Figure 1).
Specimens were randomly assigned to be tested at flexion angles of 60°, 90° or 120°.
The flexion angle was maintained by constraining the tibia by means of a rigid collar
attached to the distal end of the tibia. A loosely knotted tether was attached to the
quadriceps tendon to position the patella against the femur. The knot was such that
minimal load was required to cause slippage, thus minimizing tensile loads in the tendon
during impact. The patella was brushed with India ink and photographed prior to impact
to document any baseline surface roughness or fibrillation.

A 4.8 kg free flight impact missile was accelerated to a given pre-impact velocity,
while aligned on two steel guide rails and supported on four nylon blocks to minimize

fiiction. A rigid (6061-T6 aluminum) or padded (478 psi crush strength aluminum

130

Hexcel) impact interface was mounted to the front of a load transducer attached to the
front of the missile (Model 3173-2k, Lebow Products, Troy, MI, USA). The impact
mass was selected based on previous studies (Atkinson 1995, Haut 1989). Pilot studies
indicated that the load transducer and attached mass would not resonate during impact on
the human knee. Load data were recorded continuously on a personal computer via a 16
bit A/D board (Model DAS 1600, Computer Boards, Inc., Mansfield, MA, USA). Load
data were inertially compensated (Atkinson 1997) and sampled at 20,000 Hz. The
impacting mass was accelerated to a predetermined velocity by adjusting the air pressure
in a pneumatic impact cannon. Velocity of the mass was measured prior to impact on the
knee using two infrared optical sensors (Part No. OR518-ND, Digi-Key Corporation,
Thief River Falls, MN). These sensors are designed with a 5 mm separation between the
emitter and detector. A metal tab connected to the impact mass crossed the 5 mm gap
interrupting the first sensor and then the second. The individual sensors were separated
by 76.2 mm to eliminate cross-talk.

All experiments started with a fracture level test on one knee of a cadaver at a given
flexion angle using the rigid interface. The protocol consisted of sequentially increasing
the velocity of the impact missile in separate experiments, so as to progressively increase
the impact energy (2, 10, 22, 38, 60, 86, 112 J). The experiment was terminated when
impact loading resulted in a visible fi'acture of bone. Damage was assessed by visual
inspection through lateral and medial incisions in the joint, palpation and documented
photo graphically. For the subfiacture experiments, the contralateral limb of each subject
was impacted once with the rigid interface at 45% of the impact energy that caused bone

fracture in the first limb. The specific level of 45% was taken from experimental studies

131

(Donohue 1983, Thompson 1991, Atkinson 1995, Atkinson 1997) which document acute
subfiacture injuries at this energy level. For the impact interface study, each
contralateral knee was impacted once with the padded interface at the fracture load for
that subject. Typically, an elevated impact energy was needed for the padded interface
experiments to produce the same load as the rigid interface experiments. The additional
energy was predicted using a previous methodology (Atkinson 1997).

The contact area and contact pressure associated with the impact event was recorded
with pressure sensitive film (Prescale, Itochu Int., Quebec, CA). The film was
sandwiched between two sheets of polyethylene film (0.05 mm thick/sheet) to prevent
exposure to body fluids and help reduce shear loading artifacts (Huberti 1994). The
strips of film were sufficiently long to pass through the patellofemoral joint and cover the
anterior surface of the patella to measure the distribution of the contact load on the knee.
This also minimized frictional forces between the rigid impactor and the knee. To
increase the sensitive range of the film, low (2-10 MPa), medium (10-50 MPa), and high
(50-139 MPa) range films were stacked to record the contact between the rigid impactor
and bony anterior surface of the patella. Low and medium film were stacked for use
inside the joint. This specific use of the film was based on pilot experiments conducted
to ensure the films would not threshold for typical loads. The calibration of the pressure
sensitive film was calibrated in a servohydraulic testing machine (Model 1330, Instron,
Canton, MA, USA) using a previous methodology (Atkinson 1995, Hale 1992).

The film was located on the patellar articular surface using a previous methodology
(Atkinson 1997). Briefly, a monolayer of gauze placed over the articular cartilage of the

patella and Sudan IV biological stain was dusted over the retropatellar surface (Figure 2).

132

The film packet was inserted into the joint and the impact experiment was conducted.
Afier the impact event, the film packet and gauze were removed and the stain was evenly
distributed over the retropatellar surface (Figure 2 A). The articular surface was irrigated
with physiological saline to reveal an image of the patellofemoral contact region (Figure
2 B,C) which correlated with the pressure sensitive film. Two reference points were
made on the cartilage with a paint pen adjacent to the area of intraarticular contact. The
contact region and reference points were transferred to a clear, flexible sheet and the
reference points were registered on the film. Following load application, the film was
removed from the packet and analyzed using established methods (Atkinson 1998) to
detennine anterior and posterior contact areas and pressures. The contact area was taken
from the low range and the average contact pressure was derived from analysis of all
stacked films to derive a composite pressure.
e ' ' -I I

To investigate the state of stress in the patellar bone and cartilage, a reference 3-D
finite element model was made of the patella. The model was based on an untraumatized
patella with a typical geometry (Grelsamer 1994). To identify a ‘typical’ patella the
'morphology ratio' was assessed for all patellae. The morphology ratio is defined as the
length of the bone divided by the length of the articular surface in a saggital plane. The
morphology ratio of the reference patella was 1.30 which represented the average
morphology ratio of the subjects in the current study and agrees with average patellae
from clinical studies (Grelsamer 1994). The geometry was assessed by taking
successive, high resolution transverse MRI slices of the patella. The patella was wrapped

in cellophane and fixtured in a plexiglass cage (Figure 3). The cage maintained a fixed

133

patellar position and located the identification points in the MR images (Figure 3A). The
identification points were located by means of a 2.5 cm thick plexiglass track suspended
directly over the patellae. Two mm diameter blind holes were drilled in the track directly
above the identification points on the patella and a hypodermic needle was used to fill the
blind holes with water. Upon imaging the water-filled holes were clearly visible and
were used to locate the specific slice and position on that slice of the identification points
(Figure ZB). The patellae were imaged transversely in a GE 1.5 T magnet (high
resolution extremity coil, no fat suppression, TE=61, TR=11, FA=30°, FOV=8cm x 8cm,
3 mm slices). The cartilage, subchondral bone and trabecular bone geometry and the
spatial coordinates of the identification points were digitized into a commercial finite
element code (Ansys, version 5.3) and a solid model was constructed (Figure 3C). The
material properties were adopted from previous studies (cartilage: E=20 MPa, v=0.49
(Atkinson 1998), subchondral bone: E=3750 MPa (Mente 1994), v=0.3, trabecular bone:
E=350 MPa, v=0.3 (Townsend 1975)). The anterior surface was constrained by
imposing a zero-displacement boundary condition in the normal elemental direction for
all elements making contact with the impactor. It was observed during the experiments
that the patella attempted to migrate proximally during the 60° experiments and, to a
lesser extent, at 90° (Viano 1978). To account for tensions at the distal pole of the
patella due to the patellar tendon, a zero displacement boundary condition which only
resisted tension was imposed in the proximal-distal direction for the bone along the
insertion of the tendon. The pressure sensitive film data from the patellofemoral joint
was discretized to correspond to the cartilage surface layer of elements. The contact

pressure was then averaged over each region representing an element face to allow a

134

direct mapping of the boundary conditions on to the cartilage. All 36 impact interface
experiments were mapped onto the reference model and stresses were tabulated at the:
cartilage surface, cartilage at the interface with the subchondral bone, subchondral plate,
and the trabecular bone in the neighborhood of the subchondral bone.
i olo a :4d n' u Dtrn'aur. ' ' ofracture r . . . .1! act In e .c- S .

Following the impact experiments, patellae were placed in 10% buffered formalin
followed by successive decalcification solutions (20% formic acid). Typically each
patella was divided into five equal segments by making saggital cuts along the entire
length of the patella. Areas of interest included the articular cartilage, the subchondral
bone and the trabecular bone (Figure 4). The samples were initially reviewed to
document any gross or microscopic injuries. In some cases evidence of fissures or occult
cracks were visible at this stage. Two total samples were taken for further histological
analysis: one from the medial half and one from the lateral half of the patella. The
samples were selected based on evidence of gross or subfracture level injuries. If no
injuries were detected in any of the samples, a representative sample was taken from the
mid-lateral and mid-medial halves for histological analysis. Samples were processed
using established methods (Atkinson 1998) and stained with Safianin O-Fast Green. The
sections were read in blinded fashion by two individuals (P.A., J.W.) via light
microscopy at 12-400 power. Gross and microscopic injuries were tabulated according
to a predefined injury classification scheme (Figure 5).
S . .

The load-time curve and the pressure sensitive film were analyzed to determine the

following parameters: anterior and posterior contact area and mean pressure, peak

135

contact load, the time to attain peak load and the total contact duration. The time to
reach peak load was defined as the time interval from when 50 N of load was first
attained after contact until peak load was attained. The contact duration was defined as
the time interval from when 50 N was attained after contact until the load pulse returned
to 50 N. The contact area, contact pressure, contact load, time to peak load, and contact
duration data associated with the subfracture experiments were analyzed with a
commercial statistics package (SigmaStat, San Rafael, CA). A one way ANOVA with
Student-Neuman-Keuls post-hoc testing was used to detect differences in these data with
regard to flexion angle. The frequency of gross fractures, cartilage fissures, and vertical
and horizontal occult injuries were documented at each flexion angle and compared with
the Chi-Square test.
RESULTS
c r x ' ent

Impacting the knee with the free-flight missile produced a load-time pulse with an
initial short period of low contact load followed by a sharp rise to the peak load. The
load then gradually returned to zero (Figure 6). The amplitude and duration of the load-
tirne traces varied with flexion angle. For example, the peak load significantly increased
with increasing flexion angle (Table 2). The load at 120° of flexion was approximately
double the load at 60° and 30% greater than at 90°. The time to peak load was
approximately 5 ms at 90° and 120° of flexion which was approximately half of the time
for the 60° experiments. The contact duration was significantly lower for the 90°

experiments versus 60° or 120°. The average subfi'acture impact energy was 18.7 and

136

16.5 J for the 60° and 90° flexion angles, respectively (Table 3). These energies were
approximately half of the average impact energy for the 120° flexion angle. The average
anterior contact area was greatest at 90° of flexion and was lower at 60° and 120° flexion
angles. The average anterior contact pressure significantly increased with increasing
flexion angle. The anterior contact pressure at 60° and 90° was approximately 20% and
40% of the average anterior contact pressure at 120°.

The patellofemoral contact areas were lower than the anterior contact areas at 60° and
90° (Table 4). In addition, for these flexion angles the patellofemoral contact pressures
were larger than the anterior contact pressures. The opposite was true at 120° of flexion;
on average, the patellofemoral contact area was larger and the contact pressure was lower
than on the anterior surface of the knee. In the patellofemoral joint, the maximum
contact area occurred at 90° of flexion and was lower at 120° and 60°. Similar to the
anterior surface, the patellofemoral contact pressure increased with increasing flexion
angle, although the increases were not as significant.

The intra-articular contact area moved proximally with increasing flexion angle
(Figures 7, 9-rigid only, 10). At 60° the region of retropatellar contact was a narrow
band extending across the distal pole of the patella. This resulted from contact on the
femoral groove, approximately a centimeter above the femoral notch. At 90° the
retropatellar contact moved centrally on the patella and appeared ‘kidney’ shaped. This
resulted from femoral contact around the upper rim of the femoral notch. At 120° the
retropatellar contact was divided into two distinct lobes at the proximal edge of the

articular surface and separated at the median ridge. This resulted fiom contact with both

137

femoral condyles on either side of the femoral notch and an absence of contact with the
femoral groove.

Analysis of the histological samples and sections revealed cartilage surface fissures
and occult injuries for all flexion angle groups (Table 5, Figure 8). Even though the
impact experiments were conducted at subfi'acture impact energies, three gross fractures
were observed for the 60° flexion angle and two 90° flexion angle experiments.
Subfracture injuries were detected in three locations: on the median ridge, on the lateral
facet, or the injury extended across the patella and was observed on the median ridge and
lateral facet. Cartilage fissures were most frequently detected on the lateral facet and
were more common in subjects impacted at 60° versus 120° and 90°, although these
differences were not significant. Vertical occult clefts were observed to typically extend
across the patella and were observed grossly as a faint shadow when viewing the articular
surface. Gross analysis of the samples and microscopic analysis of the histological
sections revealed that these microfractures extended from the deep zones of the cartilage,
across the subchondral plate and into the trabecular bone. The injuries did not
communicate posteriorly to the articular surface and terminated anteriorly in the
trabecular core, or, in some cases, in the thick, anterior layer of cortical bone. The
vertical clefis were equally represented on the medial and lateral halves of the patella and
were observed more fi'equently as flexion angle increased. Horizontal occult cracks were
documented as either a separation at the tidemark or between the calcified cartilage and
the subchondral plate. These injuries were most common on the median ridge and were
equally represented at all flexion angles. A final classification of avulsive injuries was

documented for four of six 60° experiments in which the cortical bone was separated

138

from the trabecular bone at the distal pole (Figure 8). If this injury was closed (occult or
did not communicate to the surface), then the injury was listed as an occult. If the
avulsion was opened it was classified as a gross fracture.

Impact ipterface gxpm‘ments

The padded interfaces produced a significantly greater amount of contact area over
the knee than the rigid interface for all flexion angles (Table 6). The peak loads for the
rigid and padded interfaces at 60° and 90° of flexion were not different (Table 7). For
the 120° flexion angle, the padded interface peak load was significantly greater than with
the rigid interface. For both interfaces, the peak load increased with increasing joint
flexion.

The region of extra-articular contact was similar between the interfaces (Figure 9).
As flexion angle increased, the region of extraarticular contact moved proximally. The
intraarticular regions of contact were similar regardless of interface (Figure 10) with one
exception. At 120° of flexion, the medial femoral condyle is in approximately the same
plane as the anterior surface of the patella. Thus, impact experiments conducted at this
flexion angle typically resulted in contact with both the patella and the femur. An
example of a padded impact at 120° is shown (Figure 11).

Gross injuries included transverse or comminuted patellar fractures, split fractures of
the femoral condyles (Figures 12, 13), and, in two cases, fracture of the tibial plateau.
The majority of the gross injuries were to the patella and the femoral condyles (Table 8).
All rigid interface experiments (n=18 for the three flexion angles) resulted in gross
fracture of at least one of the bones making up the knee joint. Padded interface

experiments resulted in only five gross bone fractures. This represents a 72% reduction

139

in the incidence of bone fracture at essentially the same load. Subfracture injuries were
documented in the patellae impacted with both interfaces and were detected at all flexion
angles. Cartilage fissures and vertical occult traumas were common (z83% frequency) at
60° and 90° of flexion for both interfaces. At 120°, padding reduced the number of
subfracture injuries; there were no cartilage fissures and only two subjects suffered occult
patellar injuries.

The mathematical model of the impact experiments showed that the maximum shear
and tensile stresses at the cartilage surface occurred in different regions (Figure 14). The
shear stresses (Table 9) at the cartilage surface were maximal in the region of
patellofemoral contact and coincided with regions of greatest contact pressure. The
maximum principal stresses at the cartilage surface were typically compressive within the
region of contact and tensile outside, and adjacent to, the periphery of patellofemoral
contact. The cartilage shear stresses at the interface (Table 9) were slightly larger than at
the surface and were typically located adjacent to the maximum surface shear stresses.
This coincided with large gradients in the surface stress distribution. The shear stresses
at the surface and the interface increased with increasing joint flexion. Similar to the
cartilage, the shear and tensile stresses in the trabecular bone were not different for the
different impact interfaces and tended to increase with increasing joint flexion.

The shear and tensile stress distributions in the subchondral plate were nearly
identical (Figure 15). This was due to a relatively small compressive stress field (< 20%
of the tensile stresses) which, based on a simple Mohr’s circle analysis, produced tensile
stresses which were approximately twice the maximum shear stress for each point in the

subchondral bone. Tabulation of the maximum stresses (Table 10) for the 60° subjects

140

reveals that the average maximum tensile stresses were only reduced 2% with padding
versus the rigid interface. The maximum shear stresses were actually 20% greater with
padding versus the rigid interface. At the 90° and 120° flexion angles, however, padding
reduced both the maximum tensile and shear stresses versus the contralateral experiments
conducted with the rigid interface. At 90° of flexion the maximum tensile and shear
stresses were reduced 24% and 11%, respectively. Padding produced the greatest
reduction in stresses at 120° of flexion; the maximum tensile and shear stresses were
reduced 58% and 42%, respectively. For a given interface, the maximum tensile and
shear stresses at the 60° and 90° flexion angles were similar (approximately 20 MPa in
shear and 40 MPa in tension). The stresses at 120° of flexion, however, were
significantly greater, approximately double the stress magnitudes at the lower flexion
angles.

Analysis of the principal directions for the subchondral plate revealed tensile stresses
acting in the plane of the plate most notably in the proximal-distal direction (Figure 16).
This occurred in the region of intra-articular contact and was true for all flexion angles.
These tensile stresses may explain the transverse patellar fi'actures that coincide with the
location and direction of the maximum tensile stresses in the subchondral plate (Figure
17). In addition, at 90° and 120° of flexion, there was medial-lateral directed tensile
field that also acted within the plane of the subchondral plate. This always occurred in
the center of the patella, near the distal pole. The addition of this local tensile stress
maxima with the tensile stresses acting in the proximal-distal direction might explain the

comminuted patellar fi'actures (Figure 18).

141

DISCUSSION
c x 'ment

In the current study we investigated the response of the human knee to subfracture
transarticular loading at flexion angles of 60, 90 and 120 degrees. We found that the
tolerance of the human knee to both gross and microscopic injury increases with flexion
angle. Such data may be useful in the clinical setting to aid in the detection and
treatment of subfiacture injuries. Further, the results of the current study suggest that
injury prevention protocols need to account for knee flexion angle to provide optimized
protection.

Trends in the experimental data were detected as a function of flexion angle. The
contact load, and intra- and extra-articular contact pressures increased with flexion angle
while the time to peak load and the contact duration generally decreased with increasing
flexion angle. The region of inu'aarticular contact moved proximally with increasing
flexion angle. Subfracture injuries were confined to the posterior side of the patella and
were documented within the region of patellofemoral contact. Superficial fissures of
cartilage and occult lesions at the bone cartilage-interface were observed for all flexion
angles. Superficial fissures were most common in subjects tested at 60° and 120°
degrees of flexion, while the number of subjects with occult lesions typically increased
with flexion angle. Occult injuries at 60° degrees also included avulsive separations of
the cortical bone fi'om the cancellous bone at the respective patellar or quadriceps tendon
insertions.

The subfi'acture injuries documented in the current study are visually similar to those

documented in clinical and experimental studies. Cartilage surface fissures have been

142

documented in human (Zimmerman 1988) and laprine (Atkinson 1998) cartilage
following either a single impact or cyclical loading via direct contact of the articular
surface with a rigid indentor. Clinically, fissures have been documented via arthroscopy
in the ankle and have been associated with chronic pain. Occult vertical clefis have been
documented in a chronic OA canine model (Donohue 1983, Thompson 1991) following a
single transarticular blow, similar to the current study. In addition, transarticular
subfracture impact loads delivered to laprine (N ewberry 1996) and porcine (Armstrong
1985, Tomatsu 1992) joints are documented to cause cartilage fissures or occult
horizontal cracks, respectively. Regardless of the injury, however, clinical (Johnson-
Nurse 1985, Vellet 1991) and experimental (Newberry 1996, Donohue 1983) studies
document that superficial and occult injuries are associated with chronic joint pain and
degeneration. The occult microfractures documented in the current study might be
associated with 'bone bruises' (V ellet 1991). Bone bruises are typically observed
adjacent to the osteochondral junction and radiate outward into the trabecular bone
(Kapelov 1993). The bruises can be detected via MRI and may represent ruptured
arteries (Vellet 1991) in the richly vascularized patella (Amoldi 1991). Subfracture
deformation and compression of the patella have been associated with patellar pain and
might be related to a decreased venous drainage and an increase in intraosseous pressure
(Amoldi 1991). While in vivo studies document that these occult lesions can appear
resolved after several months, early degenerative changes are documented at 6 months
post-insult in 50% of patients suffering bone bruises (Vellet 1991 ).

The different frequency of subjects suffering cartilage fissures for the various flexion

angle groups might be attributed to several factors. First, the reduced frequency of

143

fissures at 90° might be associated with decreased shear stresses at the surface due to an
increase in the contact area (Atkinson 1998). In addition, the cartilage is thickest in the
region of contact associated with these experiments. Theoretical studies suggest this
would reduce stresses at the articular surface (Eberhardt 1990). Altemately, the region of
intraarticular contact coincides with a central medial-lateral band of preexisting surface
fibrillation documented previously (Atkinson 1995). An experimental study (Silyn-
Roberts 1990) found that it was difficult to mechanically produce a second fissure in the
neighborhood of an existing fissure, suggesting that preexisting fibrillation would
preclude the development of additional fissures in such a region.

It is reasonable to question whether the subfracture injuries documented in the current
study were experimentally produced, an artifact of histological processing, or evidence of
a preexisting condition or trauma. This was addressed, in part, in a previous study in
which unimpacted, healthy patellae were histologically processed and no subfi'acture
injuries were detected (Atkinson 1995). In addition, the location of the injuries in the
current study coincided with the region of intraarticular contact. Because these regions
varied greatly with flexion angle, the existence of injuries was easily verified group-by-
group. Furthermore, brushing the articular surface with India ink before and after the
experiment allowed the visualization of experimentally produced fissures. Finally, the
existence of vertical, occult microcracks could be verified via faint shadows in the
cartilage which correlated with the histological data.

The experimental data from the current study reflect trends similar to previous
studies. Progressive transarticular impact experiments leading to fracture have

previously documented a trend of increasing fracture loads with increasing flexion angles

144

(Haut 1989), although subfracture experiments were not conducted in those studies.
Subfi'acture experiments similar to the current study have been conducted, however, only
at a flexion angle of 90° (Atkinson 1995, 1997). The peak load and retropatellar contact
pressure from that study deviates by only 10% from the current study. The retropatellar
contact area, however, is 30% greater in the current study. This discrepancy is probably
related to the film ranges used in the different studies. In the previous study, the lower
threshold of the film was 10 MPa versus 2 MPa in the current study. Hence, regions of
low contact pressure at the periphery of contact were probably not transduced with the
higher threshold film. Experimental (Huberti 1984) and theoretical studies of passive
patellofemoral loading document a similar trend with respect to the contact area, region
of contact on the retropatellar surface, and the magnitude of the contact pressure.
Perhaps not surprisingly the contact pressures documented in the current study are
greater (~2-4 times) than intraarticular pressures generated by tensioning the extensor
mechanism. This is probably related to the direction and relatively low magnitude of the
loads typically attributed to joint flexion during normal activities.

It should be noted, however, that the fiacture load data presented in the current study
are significantly different than previously documented from experiments conducted on
human cadavers. Melvin, et al., (1975) Patrick, et al., (1965) and Powell, et al.,
(l975)directed impact missiles at the 90 degree flexed knee of whole, seated cadavers
using both single and sequential impact protocols. They document fracture loads of
approximately 10-17 kN, significantly higher than those of the current study. The
increased loads are probably related to the inertial motion of the cadaver and the use of

single impact protocols (Powell 1975, Atkinson 1995). The repeated impact protocol

145

used in the current study likely resulted in accumulated bone rnicrodamage leading to a
reduced load at gross fracture. Note, however, that the subfracture experiments in the
current study were only impacted once and a range of microscopic injuries and some
gross fractures were documented. The subfracture impact energy was based on the
energy to cause fiacture in the pair—matched contralateral knee. If one assumes the
fiacture energy may have been attenuated, due to the reasons mentioned above, then the
fracture energy was probably conservative. However, subfracture injuries still resulted at
45% of that ‘conservative’ fracture energy. This suggests that the incipient threshold for
the initiation of subfracture injuries is probably less than 45% of the fracture energy.
Another reason for the lower than expected fracture loads in the current study was the
use of a rigid impact interface. This allowed us to make more unambiguous comparisons
to past studies which largely serve as the basis for the current injury criterion used by the
automotive industry (Viano 1977).

Finally, the current study suggests that a subfracture knee injury criterion which
accounts for the influence of flexion angle would provide the greatest level of protection,
both acutely and in the longterrn. This would likely produce a more conservative
criterion than currently exists. For example, knee injury associated with car accidents is
based on gross bone fracture from experiments conducted at a flexion angle of 90°.
However, the current study suggests that the knee is less tolerant to gross and subfi'acture
injuries at flexion angles less than 90°. In addition, subfracture injuries are more difficult
to detect acutely and are associated with chronic joint morbidity. States (1970) suggested
that lower extremity injury criteria should be based on subfracture injury thresholds.

This would require the mechanisms for subfracture injury to be established to allow the

146

design of preventative strategies in car interiors, protective sports equipment, and floor
designs for the elderly. Preventing such injuries would reduce the longterrn economic
and societal costs currently associated with knee trauma.

W

The current criterion, used by the automotive industry, for knee injury is based on
impact loads which produce gross bone fracture at a 90° flexion angle. However, studies
have suggested that the contact area over the knee may also be an important parameter in
the injury tolerance of the knee, and that at lower flexion angles the knee may be more
susceptible to injury. In the current study, paired impact experiments were conducted on
isolated human knees flexed 60°, 90°, or 120°.

Fracture producing experiments were conducted on one knee at a given load and the
contralateral knee was impacted at the same load with a padded interface. The fi'actures
typically occurred in the patella and were oriented transversely. On average, the padded
interface produced a 300% increase in contact area versus the rigid interface. Padding
also reduced the incidence of gross injury by 72%.

A theoretical stress analysis of the rigid and padded experiments showed that the
shear and tensile stresses in the cartilage and the shear stresses in the trabecular bone
were not different for the different interfaces. However, the tensile and shear stresses in
the subchondral bone were found to be reduced at the 90° and 120° flexion angles with
padding. The tensile and shear stresses were dramatically reduced at 120° of flexion
which might explain the absence of patellar fractures and the low frequency of occult
patellar injuries. For both impact interfaces, the stresses at 120° were approximately

double the stress magnitudes at the lower flexion angles. This might be explained by the

147

patellar motion at the higher versus lower flexion angles. At 120° of flexion, the patella
is driven into the femoral condyles and is essentially undergoes simple compression
(Haut 1989; Viano 1978). At the 60° and 90° flexion angles, the patella is both driven
into the condyles and undergoes a proximal migration, putting the patellar tendon in
tension. Thus, in the region of intraarticular contact, the patella undergoes compression,
however, at the distal pole, the cortical and trabecular bone are put in tension. These
differences between the lower and higher flexion angles might explain the different loads
and stresses documented for the different flexion angles. While both tensile and shear
stresses were reduced at 90° of flexion, only the reduction in tensile stresses was
significant. At this flexion angle, there was one patellar fracture and numerous vertical
occult injuries documented with the padded interface. The benefit of the padding may be
the significant reduction in tensile stresses which may have prevented the occult injuries
from opening into gross fractures. Padding did not appreciably alter the subchondral
bone stresses at the 60° flexion angle. This might explain the incidence of two gross
patellar fractures and the high frequency of subfracture injuries at this flexion angle.
This also suggests that this flexion angle will be difficult to protect with padding.

For all flexion angles and both interfaces, the tensile stresses were generally maximal
within the region of intraarticular contact and the maximum stresses acted within the
plate in a proximal-distal direction. These tensile stresses are important as there location
and direction are consistent with the observed transverse fractures. The origin of these
stresses may be explained by the shape of the intraarticular contact area, which is
essentially a wide band of contact across the patella (Figure 19). This is similar to

modeling the femoral condyle as a long, cylindrical indentor compressing the

148

retropatellar surface. This would subject the subchondral plate to local bending which
would develop large tensile stresses in the subchondral plate.

In addition to the transverse fractures, comminuted fractures were documented at the
90° and 120° flexion angles. These fractures typically appeared as a transverse fracture
with the addition of a proximal-distal oriented fracture emanating from the distal pole. It
should be noted that the distal fracture was not in the region of intraarticular contact.
The distal fracture may be explained by the contact areas and stress distributions
associated with elevated flexion angles. At 90° and 120° of flexion, the intraarticular
contact area was essentially two lobes of contact on the medial and lateral patellar facets;
this arose from contact with the medial and lateral femoral condyles. This essentially
puts the patella in three point bending and generated regions of tensile stress acting in the
medial-lateral direction. Altemately, at 60° of flexion, the patella contacted the femoral
groove which produced a narrow band of contact across the distal pole of the patella. For
this contact, only transverse fiactures were documented, again in the region of contact.
Analysis of the model revealed that the tensile stresses only acted in the proximal-distal
direction; there were no medial-lateral acting tensile stresses. Thus, the contact areas and
regions of contact associated with different flexion angles alters the state of stress in the
patella and the potential for injury.

The current study shows that the injury tolerance of the knee varies with flexion
angle and the degree of load distribution over the knee. These data may be helpful in
developing a more comprehensive injury criteria. Such a criterion might account for the
flexion angle and the manner in which the load is distributed over the knee. Injury

prevention strategies might also benefit from the stress analysis of the patella. For

149

example, subchondral plate tensile stresses were associated with gross fractures. Thus,

injury prevention strategies may be studied theoretically and use the reduction of such

stresses as a method to reduce the risk of injury to the knee.

ACKNOWLEDGMENTS

This study was supported by a grant (R49/CCR503607) from the Centers for Disease

Control and Prevention. 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 Ms. Jane Walsh for help in the

interpretation of histological slides and Mr. Cliff Beckett for technical support. We are

also indebted to Hexcel Corporation and Mr. John Porter for technical and materials

support. Finally, we wish to thank Mrs. Kathy Pearson and the Michigan Tissue Bank

for solicitation of donors and retrieval of joints.

REFERENCES

1. Armstrong C, Mow V, Wirth C (1985) Biomechanics of impact-induced
microdamage to articular cartilage: a possible genesis for chondromalacia patella. In:
Amer Acad of Orth Surg Symp on Sports Med, ed by G Fenerrnan. pp70-84

2. Amoldi C (1991) Patellar Pain. Acta Orthop Scand 62: 1-29

3. Ateshian, G., Kwak, S., Soslowsky, L., and Mow, V (1994) A stereophotographic
method for determining in situ contact areas in diarthrodial joints, and a comparison
with other methods. Journal of Biomechanics 27(1):111-124

4. Atkinson, T.S., Haut, R.C., and Altiero, NJ. (1998) An Investigation of Biphasic
Failure Criteria for Impact Induced Fissuring of Articular Cartilage. J of
Biomechanical Engineering (In Press)

5. Atkinson PJ, Haut RC (1995) Insult to the human cadaver patellofemoral joint:

effects of age on fracture tolerance and occult injury. 39th Stapp Car Crash
Conference: 28 1-294

150

Atkinson PJ, Haut RC (1995) Subfracture Insult to the Human Cadaver
Patellofemoral Joint Produces Occult Injury. Journal of Orthopaedic Research 13:
936-944

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. 4lst Stapp Car Crash Conference, SAE: 167-180. paper
973327

Atkinson PJ, MacKenzie C, Haut RC (1998) Joint fracture load prediction using
geometrical, physical and pathological measures: the human knee. SAE International
Congress: paper 980358

Dischinger PC, Kems TJ, Kufera JA (1995) Lower extremity fracturesin motor
vehicle collisions: the role of gender and height. Accident Analysis and Prevention
27(4): 601-606

10. Dischinger P, Cushing B, Kerns T (1992) Lower extremity fi'actures in motor vehicle

11.

12.

l3.

14.

15.

16.

17.

collisions: influence of direction of impact and seatbelt use. Proc 36th Conf AAAM
Portland, OR pp319-326

Donohue MJ, Buss D, Oegema T, Thompson R (1983) The Effects of Indirect Blunt
Trauma on Adult Canine Articular Cartilage. The Journal of Bone and Joint Surgery
65-A(7): 948-957

Eberhardt, A., Keer, L., Lewis, J., and Vithoontien, V. (1990) An Analytical model
of joint contact. Journal of Biomechanical Engineering 112:407-413

Eckstein F, Muller-Gerbl M, Putz R (1992) Distribution of subchondral bone density
and cartilage thickness in the human patella. Journal of Anatomy 180: 425-433

Grelsamer R, Proctor C, Bazos A (1994) Evaluation of patellar shape in the sagittal
plane: A clinical analysis. Amer J Sports Med 22: 61-66

Hale J E, Brown TD (1992) Contact stress gradient detection limits of pressensor film.
J Biomech Eng 114:352-358, 1992

Haut R (1989) Contact Pressures in the Patello-femoral Joint during Impact Loading
on the Human F lexed Knee. Journal of Orthopaedic Research 7: 272-280

Hayashi S, Choi H, Levine R, Yang K, King A (1996) Experimental and analytical

study of knee fracture mechanisms in a frontal knee impact. 40th Stapp Car Crash
Conference 161-171

151

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

Hering W, Patrick L (1977) Response Comparisons of the Human Cadaver Knee and
a Part 572 Dummy Knee to Impacts by Crushable Materials. Twenty-First Stapp Car
Crash Conference 21: 1015-1053

Huberti H, Hayes W, Stone J, Shybut G (1984) Force ratios in the quadriceps tendon
and ligamentum patellae. Journal of Orthopaedic Research 2: 49-54

Huelke D, O'Day J, States J (1982) Lower Extremity Injuries in Automobile Crashes.
Accid Anal & Prev 14: 95-106

Johnson-Nurse C, Dandy D (1985) Fracture-Separation of Articular Cartilage in the
Adult Knee. The Journal of Bone and Joint Surgery 67-B(10: 42-43

Kapelov R, Teresi L, Bradley W, Bucciarelli N, Murakami D, Mullin W, Jordan J
(1993) Bone Contusions of the knee;Increased Lesion Detection with fast Spin-Echo
MR 1m: wing with Spectroscope Fat Saturation. Radiology 901-904

MacKenzie EJ, Siegel JH, Shapiro S, eta]. (1988) Functional Recovery and Medical
Costs of Trauma: An analysis by Type and Severity of Injury. Journal of Trauma 28:
281-298

Melvin J, Stalnaker R, Alem N, Benson J, Mohan D (1975) Impact Response and
Tolerance of the Lower Extremities. Stapp Car Crash Conf 19: 543-559

Newberry W, Haut R (1996) The effects of subfracture impact loading on the
patellofemoral joint in a rabbit model. 40th Stapp Car Crash Conference 149-159

Patrick L, Kroell C, Mertz H (1965) Forces on tehehuman body in simulated Crashes.
Stapp Car Crash Conf 9: 23 7-259

Powell W, Ojala S, Advani S, Martin R (1975) Cadaver F emur Responses to
Longitudinal impacts. Stapp Car Crash Conf 19: 561-579

Silyn-roberts, H. and Broom, N. (1990) Fracture Behavior of Cartilage-On-Bone in
response to repeated impact loading. Connective Tissue Research 24:143-156

States J (1970) Traumatic Arthritis-- A Medical and Legal Dilemma. Ann Conf of the
Amer Assoc for Automotive Med 14: 21-28

States J (1986) Adult occupant injuries of the lower limb. In: Biomechanics and
medical aspects of lower limb injuries. Warrendale, PA: SAE 97-107

States J (1970) Traumatic arthritis--a medical and legal dilemma. Ann Conf of the
Amer Assoc for Automotive Med 14:21-28

152

32.

33.

34.

35.

36.

37.

38.

39.

40.

Thompson RC, Oegema TR, Lewis J L, Wallace L (1991) Osteoarthritic changes after
acute transarticular load. The Journal of Bone and Joint Surgery 73-A(7): 990-1001

Tomatsu T, Imai N, Takeuchi N, Takahashi K, Kimura N (1992) Experimentally
produced fractures of articular cartilage and bone. J Bone Joint Surg 3:457-461, 1992

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:
150-152

Vellet AD, Marks PH, Fowler PJ, Munro TO (1991) Occult post-traumatic
osteochondral lesions of the knee: prevalence, classification, and short-term sequelae
evaluated with MR imaging. Radiology 178(1):271-276

Vener MJ, Thompson RC, Lewis JL, Oegema TR (1992) Subchondral damage after
acute transarticular loading: an in vitro model of joint injury. J Orthop Res 10:7 59-
765, 1992

Viano D (1977) Considerations for a femur injury criterion. 2 1 st Stapp Car Crash
Conference 445-473

Viano DC, Stalnaker RL (1980) Mechanisms of Femoral Fracture. Journal
Biomechanics 13: 701-715

Viano, D., Culver, C., Haut, R., Melvin, J ., Bender, M., Culver, R., and Levine, R.
(1978)Bolster impacts to the knee and tibia of human cadavers and an
anthropomorphic dummy. Soc.Automotive Engineers :403-428

Zimmerman, N.B., Smith, D.G., Pottenger, L.A., and Cooperman, DR. (1988)
Mechanical disruption of human patellar cartilage by repetitive loading In Vitro..
Clinical Orthopaedics and Related Research (229):302-307

TABLES

Table 1: Experimental design

 

Flexion Subfracture lrnpact interface
Angle experiments experiments

 

 

 

 

60° n=6 n=6
90° =6 n=6
1 20° n=6 =6

 

 

 

 

153

Table 2: Load-time data

 

 

 

 

 

 

 

 

Knee Flexion Peak Load Time to peak Contact Duration (Pulse)
Angle
N ms ms
60 2394.5:8495" 9.5:] .1 18.7i-1.5
90 3848.0:15118 5.4:0.9# 12.6i3.2#*
120 5368.5il723.3 5.3i1.2# 16.53536

 

 

*Significantly different from 120° flexion angle
# Significantly different from 60° flexion angle

 

Table 3: Impact energies and anterior contact areas and pressures

 

 

 

 

 

 

 

Knee F lexion Impactor Anterior Contact Anterior Contact
Angle Energy Area Pressure
Degrees J mm2 MPa
60 18.7i11.5* 603.1:3012 50:13“
90 165:6.3“ 711.1:444.1 10.4i3.1*
120 32.9140 5079:2284 23.6i8.8

 

 

 

*Significantlydifferent from 120° flexion angle

 

 

 

 

 

 

Table 4: Patellofemoral contact areas and pressures
Knee Flexion Patellofemoral Contact Patellofemoral Contact
Angle Area Pressure
mm MPa
60 431 .4i307.9 94:45
90 651 .7i312.6 12.7i5.6
120 609.3i296.9 165:4.0

 

 

 

 

154

 

 

 

 

Table 5: Gross and microscopic injuries fiom impact with a rigid interface:
paired experiments at three flexion angles at fracture and subfiacture level
impact energies. Injury classifications (A-E) are defined in Figure 5.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fracture level experiments Contralateral subfracture
experiments
F lexion Subject A B C D E A B C D E
Angle

60 H72 1 0 1 1 0 0 0 1 1 0
60 H73 1 0 1 0 0 l 0 l 1 0
60 H74 1 0 1 l 0 1 0 0 1 0
60 H75 1 0 1 0 0 1 0 0 1 0
60 H76 1 1 O l 0 0 1 0 O 0
60 H77 1 1 1 1 0 0 1 0 l 0
Total 6 2 5 4 0 3 2 2 5 0

Injuries
90 H57 1 l 1 1 0 0 0 0 0 0
90 H58 1 0 l 0 0 0 0 0 l 0
90 H60 1 1 l l 0 0 0 l 1 0
90 H63 1 0 l 0 0 l 1 l 0 0
90 H64 1 O 1 0 0 1 0 1 0 0
90 H65 1 1 0 l 0 0 1 l l 0
Total 6 3 5 3 0 2 2 4 3 0

Injuries
120 H66 0 1 l 1 1 0 0 l l 0
120 H67 1 l l 0 0 0 0 l l 0
120 H68 1 0 0 0 0 0 0 0 l 0
120 H69 1 0 1 0 0 0 0 1 1 0
120 H70 0 0 l O l 0 l 0 O 0
120 H71 1 0 1 0 0 0 1 0 0 0
Total 4 2 5 l 2 0 2 3 4 0

Injuries
Grand Totals for all 16 7 15 8 2 5 6 9 12 0

flexion angles

 

 

 

 

 

 

 

 

 

 

Table 6: Anterior contact area on the cadaver knee

 

 

 

 

Flexion Angle Rigid Impact Interface Padded Impact Interface
60° 346:1:127 . l494i757“
90° 729i88# l614:l:667*
120° 703:1 14# 2990i626

 

 

 

‘Significantly different than 120° padded interface
#Significantly different than 60° rigid interface

 

 

 

155

Table 7: Peak loadiforjaired impact experiments on each cadaver knee (kN)

 

 

 

 

Flexion Angle Rigid Impact Interface Padded Impact Interface
60° 3.71:1.98 4.31:0.9
90° 5.04i1.4 5.47i1.83
120° 5.761138 8.28i-l.2*#

 

 

 

*Significantly different than 120° padded interface
#Significantly different than 60° rigid interface

 

 

 

 

Table 8: Gross and microscopic injuries fiom impact with rigid and padded
interfaces: paired experiments at three flexion angles. Injury classifications (A-F)
are defined in Figure 5.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Rigid Impact Interface Contralateral Padded Interface
Experiments Experiments
Flexion Subject A B C D E F A B C D E F
Angle

60 H72 1 0 1 1 0 0 0 0 0 0 0 0
60 H73 1 0 0 0 0 0 0 1 1 1 0 0
60 H74 1 1 0 l 0 0 l l 0 l 0 0
60 H75 1 0 l 1 0 0 0 0 l l 0 0
60 H76 1 0 l 0 0 0 l 0 0 0 0 0
60 H77 1 0 l 0 0 0 0 0 0 1 0 0
Total 6 l 4 3 0 0 2 2 2 4 0 0

Injuries
90 H57 1 0 l l 0 0 l 0 l 0 0 0
90 H58 1 0 l 0 0 0 0 0 l 1 0 0
90 H60 1 0 l 0 0 0 0 0 l 0 O 0
90 H63 1 0 l 0 0 0 0 0 l 0 0 0
90 H64 1 l 0 0 0 0 0 l l 0 0 0
90 H65 1 0 l 0 0 0 0 0 0 1 0 0
Total 6 1 5 l 0 0 l 1 5 2 0 0

Injuries
120 H66 1 0 0 l 0 0 0 l l 0 0 0
120 H67 0 0 0 l 1 0 0 0 0 0 0 0
120 H68 0 0 0 1 l 0 0 0 0 0 l 0
120 H69 0 0 0 0 0 l l l l 0 0 0
120 H70 1 1 1 0 l 0 0 0 0 0 l 0
120 H71 1 0 0 0 0 1 l 0 0 0 0 0
Total 3 l l 3 3 2 2 2 2 0 2 0

Injuries
Grand Totals for all 15 3 10 7 3 2 5 5 9 7 2 0

flexion angles

 

 

 

 

 

 

 

 

 

 

 

 

 

156

 

Table 9: Shear stresses in the cartilage at two sites: articular surface and at the tidemark

 

Table 10: Shear and tensile stresses in the subchondral bone

*Significantly different than 120°

 

FIGURES

    

Patella

  
 

 

v-J: Am,

Load Cell -

 

 

 

 

Flexion : Tibia
Angle
i Collar

///7///7//

 

 

Figure 1

157

 

Figure 2

158

 

 

 

 

 

Figure 3

159

 

  
    
  
 
 

 

Articular cartilage: thickness 3-5 mm l

 

 

 

Subchondral or cortical bone: thickness 1-2.5
mm

 

 

 

Fabecular or cancellous bone

 

 

POSTERIOR

    

 

 

 

 

trabecular or cancellous bone

 

  

 

Cortical bone: thickness 3-5 mm I

 

 

 

 

 

 

 

I Saggital section through the human patella l

 

Figure 4: A schematic depicting areas of interest for analysis to identify injuries to the
cartilage, subchondral bone, and trabecular bone.

 

4 5

Figure 5: Schematics of a saggital patellar section depicting the typical location of gross
and microscopic injuries for the 60°, 90°, and 120° flexion angles (1, 2, 3). Typical gross
injuries to the femur (4) and tibia (5) are also shown. The labels correspond to tabulated
injury data shown in Tables and . Injuries A-D are specific to the patella and injuries E
and F are specific to the femur and tibia, respectively: (A) gross patellar fracture, (B)
horizontally oriented occult microfracture, (C) vertically oriented occult microfracture,
(D) superficial cartilage fissure, (E) split or comminuted gross fracture of the femoral
condyles, (F) gross fi'acture of the tibial plateau (this injury is also termed a Type I ‘Salter’
fracture).

 

Load

 

 

 

Figure 6

161

Lateral
Medial

 

Figure 7

 

 

 

 

Figure 8: Schematics of patellar injuries documented via analysis of histological samples

and sections for subjects in which both knees were impacted with a rigid impact interface
in separate experiments. The top box represents fracture level experiments and the
bottom box represents the pair-matched contralateral experiments conducted at a
subfracture impact energy (45% of the fracture energy). The subject mumber is presented
next to each schematic. The schematic and injuries are defined in Figure 4 and 5.

162

Rigid Padded

1“ El“

    

   

120 degres

Figure 9: Extra-articular contact regions (between the impactor and the anterior patella)
for the rigid and padded interfaces for the three flexion angles.

163

Retro- Femoral
patella condyles

60 degrees:

   
   
 
  
 

  
  
 

 

.11!

aihi? W

90 degrees

120 degrees

Figure 10: Inna-articular contact regions superposed on the retropatellar and femoral
articular surfaces. The regions of intra-articular contact were identical for the impact
experiments conducted with rigid or padded impact inetrfaces except as noted in Figure
11.

 

Femur at 120 degrees:
padded

Figure 11: Experiments at the 120° flexion angle with the padding typically resulted in
direct contact between the padding and the medial femoral condyle. This occurs because
the medial condyle is in a similar plane as the anterior patellar surface for elevated
fflexion angles.

 

Transverse
patellar fracture

 

Comminuted
patellar fracture

“x.

    

Femur fracture

Figure 12: Predominant gross injuries to the patella and femur.

165

 

 

 

 

Figure 13: Schematics of patellar injuries documented via analysis of histological samples
and sections for subjects in which one knee was impacted with a rigid impact interface in a
fiacture level experiment. The contralateral knee was then impacted at the rigid interface-
fracture load with a padded interface. The top box represents fracture level-rigid interface
experiments and the bottom box represents the pair-matched contralateral experiments
conducted with the padded interface. The subject mumber is presented next to each
schematic. The schematic and injuries are defined in Figure 4 and 5.

166

 

Maximum
shear stress

 

1st principal stress

Figure 14: Arrows point to regions of maximum stress at the cartilage surface for a
typical experiment conducted at a 90° flexion angle. The outline depicts the region of
intraarticular contact.

167

   
 

Maximum
shear stress

 

lst principal stress

Figure 15: Arrows point to regions of maximum stress in the subchondral plate for a
typical experiment at a 90° flexion angle. The outline depicts the region of intraarticular
contact.

168

co n
attritiiur .. n.
n‘““?;vavolaa I c
I... ”W
“flux? §**"9"’ I ”l
Munich A garden: I an
it irate... a
it iv a
u“: g siege!!! (an
O

.- 4..
1“.l”‘f.'l'a' *‘4?’ I no I I $05

a! .II : O u
:...w 23‘s extras: .a $53"

   

.I

         

13:

if”

a.

0 .w‘..
0.-

Principal directions

cocooooooo-w
q
a

 

.1

 

x
r. i 4.: 'I I . o a no “no... Mm
was 00». [\‘vl‘c': 00‘..." ..
chest... \ t. . . . 3.3.3 a
a” w Ilsealblil‘lxa: . s h a “at“: . .mm
Jae '0‘." C “\‘t‘lvb “Q a “o. o
“|\”' (‘0‘. 3‘ a ‘ N % \‘k“~u avg
1. of»! on 11.00 s x as 2 .u.
is a . eve. .. s \ .. - C
if ’I In a \ \ ‘ ‘k‘
W In a o o In. x \ \ u.‘
v . - . . 3%.
an......H““s
a a u u H n so
u m u s s

Figure 16: A whole-field view of the principle directions in the subchondral plate and a
simplified schematic showing the dominant tensile stresses and their predominant
169

directions for the stress field shown in Figure 14.

 

Figure 17: An example of how the tensile stresses acting in the proximal-distal direction
(shown at left) are associated with a simple, transverse patellar fi'acture.

 

Figure 18: An example of a more complicated, comminuted fracture in which both the
proximal-distal and medial-lateral tensile stresses (shown at left) are associated with two
planes of fracture.

170

 

 

  

 

 

Femoral condyle

 

 
 
 
 

 

 

Cartilage

 

 

 

 

 

4 2,! F
SubchondralBone

 

 

 

 

Figure 19: A schematic of the retropatellar contact and a saggital section passed through
the patella (inset). An idealized representation of contact (along the saggital section
viewed in the direction of the arrow) is modeled as a cylindrical indentor compressing
the articular cartilage.

r
l

171

CHAPTER 8
A PROPOSED MODEL TO TRANSFORM CADAVER KNEE INJURY
CRITERIA TO THE HYBRID III DUMMY
Patrick Atkinson, Nicholas Altiero“, Roger Haut
Orthopaedic Biomechanics Laboratories
*Department of Material Science and Mechanics
Michigan State University

Chris Eusebi, Vivek Maripudi, Tim Hill, Kiran Sambatur
Breed Technologies, Sterling Heights, Michigan

ABSTRACT

A fracture threshold of 10 kN is currently used in the automotive industry to predict
gross fracture of bone to the knee, thigh and hip. However, recent studies show that load
alone is not sufficient to predict joint trauma. Experiments on isolated cadaver knees
show that increasing the contact area for a given contact force over the knee significantly
reduces the occurrence of gross and microscopic injuries at a given contact load over the
knee. These data, however, are not directly applicable to the Hybrid III knee because the
contact force-area relationships for the cadaver and dummy differ. In the current study
transformations were established between the cadaver injury data and the dummy knee.
Numerous experiments were conducted on the dummy to establish comparisons with the
cadaver data for the isolated knee. Mathematical models were developed to address
transformation of these data to the whole dummy. Sled tests were run, using an idealized
instrument panel, to show the utility of the new injury curves in the prediction of joint

trauma for depowered air bags, restrained versus unrestrained occupants, etc. This study

172

begins to show a strategy whereby injury data from isolated cadaver knee experiments can
be utilized to predict injurious conditions in the automobile crash environment using
anthropomorphic dummies.

INTRODUCTION

Traumatic lower limb injuries in frontal car crashes continues to be a significant
problem. Numerous field studies conducted in the last 20 years conclude that lower
extremity injuries occur in about one of every three car accidents (Bourret 197 7,
Hartemann 1977, Morgan 1991, Otte 1996, Pattimore 1991, Thomas 1995, Siegal 1993,
States 1986). In addition, these studies document that the knee is the most injured body
region within the lower extremity. The knee may either suffer fracture or subfracture
(contusions, lacerations, abrasions) injuries. The majority of knee injuries result from high
speed contact with the instrument panel or steering column. In an effort to prevent the
more severe, fracture level injuries, the National Highway Traffic and Safety
Administration adopted an injury criterion in 1976 (Viano 1977). The criterion states that
axial femoral loads cannot exceed 10 kN when certifying new automobiles with
anthr0pomorphic dummies at a crash velocity of 13.4 m/s (30 mph). The criterion is
based on a bone fi'acture threshold derived fiom hmnan cadaver impact experiments
(Patrick 1965, Melvin 1975, Powell 1975, Viano 1980). These impact experiments,
conducted in the 1960’s and 70’s, delivered impact loads to the 90° flexed knee. Gross
fiacture at the knee, femoral shaft, or hip occurred at loads ranging from 10-17 kN for
these tests conducted on whole cadavers. Analysis of the data from these experiments

suggests, however, that peak load may not be the sole parameter of interest. Of the 68

173

total knee impact experiments, 32 were conducted with a rigid interface and 38 were
conducted with padded interfaces of variable thickness and composition. In general, the
peak loads were similar, regardless of the impact interface. However, even at similar
loads, the padded interface experiments resulted in a 65% and 14% reduction in patellar
and femoral condyle fractures, respectively, versus the rigid interface experiments. One
explanation may be related to the increased extraarticular contact area over the patella and
possibly, the femoral condyles.

Several recent studies specifically address the relationship between peak load, contact
area over the knee, and the associated injury potential. In studies on the isolated human
cadaver knee, rigid and padded interfaces were used to deliver a single impact load to the
isolated knee joint (Atkinson 1997). The protocol consisted of impacting one knee with a
rigid interface and then impacting the contralateral knee at the same load with a padded
interface. All rigid interface experiments resulted in small contact areas over the anterior
surface of the patella. All of these experiments produced gross fractures of the patella or
subfracture injuries at the bone-cartilage interface. Altemately, padded impacts resulted in
significantly greater contact areas over the patella and no gross or microscopic injuries.
Theoretical models of the patella reveal that the shear and tensile stresses in the
subchondral plate were reduced 70% with padding. In another study, a stress-based
parametric analysis of contact area and impact load shows that greater impact loads can be
tolerated with greater contact area over the knee (azAtkinson 1998). These studies
underscore the association between elevated tissue stresses and tissue failure, rather than

simply peak load and bone fracture. It should be noted, however, that the magnitude of

174

the load is still an important variable. Indeed, the stresses in an arbitrary body of material
are related to the magnitude of the load as well as its distribution over the body.
However, stress is a derived quantity and cannot be experimentally quantified. But, based
on the above studies, it appears that contact area over the knee is associated with the
magnitude of tissue stresses and the incidence of injury. Thus, knowledge of the impact
load and the contact area over the knee might allow a rough estimation of the su'ess
magnitudes. This may provide a more accurate injury criterion than peak load alone.

To establish an injtn'y criterion based on peak load and contact area would require an
understanding of the loads and areas associated with impact trauma events. Currently,
impact loads are routinely measured in the automotive industry during experimental crash
simulations. Instrumented anthropomorphic drnnmies serve as hmnan surrogates and are
designed with a load cell in the femoral shaft. Thus, the load associated with knee-
instrurnent panel contact can easily be recorded. The contact area is not currently
measured, however, established methods (i.e. pressure sensitive film, instrumented
pressure mats) do exist to measure the contact area. The load-area relationship for the
dummy, though, may not be directly applicable to the human. While it is true that the
dummy is designed to imitate the human cadaveric response (for which the dummy is
based on), limited experimental data (Hering 1977, Nyquist 1974) suggest that the load-
area relationship for the dummy can be significantly different fiom the human. Further,
these differences may be accentuated based on the composition of the impact interface (i.e.

rigid, padded). The current study was designed to address these issues.

175

The overall study aim was to develop a transformation between the load-area
relationship between the dummy and the human. Impact experiments were conducted on
the isolated dmnmy and cadaver knee using different impact interfaces. The
transformation is based on impact energy and empirically derived load-area data. This
transformation is based on the fact that impact energy is one of the few parameters which
can be independently controlled a priori. Thus, while the load-area impact response of the
knee is clearly a function of its geometry and material properties, the response of the
dummy and cadaver knee was related based on consistent levels of impact energy.

In addition to the isolated knee tests, automotive crash simulations were also
conducted using whole anthropomorphic dmnmies with a variety of restraint scenarios.
Load-area data were recorded to provide an example of the data transformation from the
isolated cadaver knee to the automotive crash environment with the dummy.

METHODS AND MATERIALS
Ex eriment rot col

Impact experiments were conducted on the knee of a 50th percentile Hybrid III
dummy and a group of 16 pair of isolated cadaver knees with an age of 69115 years (12
male, 3 female, 1 unknown). The cadavers had an average height of 1.75i0.17m and a
mass of 70.4il 1.8. The cadaveric data comes from separate, parallel studies of impact
trauma. The specific test protocols were similar for the dummy and cadaver knees and are
provided later. For the cadaveric experiments, tissues were procured from donors with no
prior history of joint trauma and/or evidence of advanced joint pathology (areas of

denuded bone, advanced chondromalacia, etc.). Shortly after death, the patellofemoral

176

joint was examined through a medial incision. Joint preparations were excised
approximately 15 cm from the knee proper and stored at -20° C until the day of testing.
The tissues were allowed to thaw at room temperature overnight prior to testing.
Superficial tissues proximal to the femoral condyle were excised and the femoral shaft was
cleaned with alcohol. The femur was potted in a cylindrical steel sleeve with room
temperature curing epoxy, and was mounted horizontally to a rigid test fiame (Figure 1).
Specimens were tested at a 90° flexion angle. The flexion angle was maintained by
constraining the tibia by means of a rigid collar attached to the distal end of the tibia. A
loosely knotted tether was attached to the quadriceps tendon to position the patella
against the femur. The knot was such that minimal load was required to cause slippage,
thus minimizing tensile loads in the tendon during impact. The patella was brushed with
India ink and photographed prior to impact to document any baseline surface roughness or
fibrillation.

The isolated dummy knee was mounted to the testing fixture in a similar fashion as the
cadaver knees (Figure 2). The dummy knee component included the distal half of the
thigh and the entire lower leg without the foot.

To impart the impact force, a 4.8 kg impact missle was accelerated to a given pre-
irnpact velocity, while aligned on two steel guide rails and supported on four nylon blocks
to minimize fiiction. A rigid (606l-T6 aluminum), ‘stiffly’ padded (3.3 MPa crush
strength aluminum Hexcel), or ‘softly’ padded (1.65 MPa crush strength aluminum Hexcel
impact interface) impact interface (Figure 3) was mounted to the front of a load

transducer which was attached to the impact missile (Figure 1,2) (Model 3173-2k, Lebow

177

Products, Troy, MI, USA). In the dummy tests, the femur load cell (Figure 2) was used
as a redundant measure of the impact load. The impact mass was selected based on
previous studies (Atkinson 1997, Haut 1989). Pilot studies indicated that the load
transducer and attached mass would not resonate during impact on the human knee. Load
data were recorded continuously on a personal computer via a 16 bit A/D board (Model
DAS 1600, Computer Boards, Inc., Mansfield, MA, USA). Load data were inertially
compensated (Atkinson 1997) and sampled at 20,000 Hz.. The impacting mass was
accelerated to a predetermined velocity by adjusting the air pressure in a pneumatic impact
cannon. Velocity of the mass was measured prior to impact on the knee using two
infi'ared optical sensors (Part No. OR518-ND, Digi-Key Corporation, Thief River Falls,
MN). These sensors are designed with a 5 mm separation between the emitter and
detector. A metal tab connected to the impact mass crossed the 5 mm gap interrupting the
first sensor and then the second. The individual sensors were separated by 76.2 mm to
eliminate cross-talk.

The cadaver experiments started with a fracture level test on one knee of a cadaver
using the rigid interface. The protocol consisted of sequentially increasing the velocity of
the impact missle in separate experiments, so as to progressively increase the impact
energy (2, 10, 22, 38, 60, 86, 112 J). The experiment was terminated when impact
loading resulted in a visible fracture of bone. In four of the subjects, the contralateral knee
was subjected to a single impact with the rigid interface at 45% of the impact energy that
caused bone fiacture in the first limb (Donohue 1983, Thompson 1991, a:Atkinson 1995).

In the remaining 12 subjects, six contralateral knees were impacted once with the stiff

178

 

 

padding and six with the soft padding at the rigid interface fracture load for that subject.
Data from experiments with any of the interfaces in which a gross fracture of bone was
noted were excluded from further analysis in the current study.

Impact experiments on the dummy knee were conducted in a similar fashion as the
cadaver. A similar range of impact energies were used to impact the knee with the rigid
and padded interfaces. For the rigid interface, testing was terminated when the maximum
load for the dummy femur load cell was approached (13.3 kN). Testing with the padded
interface was terminated when successive experiments did not yield additional contact
area.

The anterior contact area associated with the impact event was recorded with pressure
sensitive film (Prescale, Itochu Int., Quebec, CA). The fihn was sandwiched between two
sheets of polyethylene film (0.05 mm thick/sheet) to prevent exposure to body fluids and
help reduce shear loading artifacts (Huberti). This also minimized fiictional forces
between the impactor and the knee. To optimize the measurement of contact area, low
range pressure sensitive film was used (range: 2-10 MPa). The pressure sensitive film was
calibrated in a servohydraulic testing machine (Model 1330, Instron, Canton, MA, USA)
using a previous methodology (Atkinson 1997). Following the impact experiment, the
film was removed fi'om the packet and analyzed using established methods (bzAtkinson
1998) to determine the contact area over the knee. Briefly, the film images were digitally
scanned at 150 dpi. A commercial computer code (SigmaScan) was then used to outline

the images to determine the contact area.

179

 

_T_r_ansformation protocol

The load-contact area-impact energy data were tabulated for all dmnmy-cadaver
experiments. These data were used to develop a transformation for the load-area
relationship between the cadaver and dummy for the impact interfaces used in this study.
The protocol essentially consisted of four steps:
1) Tabulate the load-area-energy data for the dummy knee for a given interface:
Example: Rigid interface data presented in Table 1 (note the arbitrary assignments of the
data to the given coordinate axes). Ideally, a linear or polynomial fit of the above data in
three space would define the dummy impact reponse for the rigid interface over the entire
range of the data. This would allow a direct transfonnation to the cadaver.
Unfortunately, regressions are restricted to functions of one variable. This problem was
overcome by first performing a linear regression on the load-area data (Figure 4).
2) Next, fit a surface to the area, load, energy data. This was performed by conducting an
iterative, nonlinear regression through these data via the Marquardt-Levenburg method
(SigmaStat). This requires an assumption of the functional form that will fit the data along
with initial value assumptions of the function’s constant values. The simplest surface is a
plane which can be written as z=(-ax-by+d)/c. The program iterates to identify the values
of the constants which will minimize the sum squared error (i.e. min 2. [energy data - (-ax-
by+d)/c]2) to produce the best fit planar fit of the data (Figure 5).

3) Now, a continuous array of area, load, energy data may be obtained by the following:

180

A) Write an incremental array of contact areas over a meaninful range determined by the
experimental data (Table 2).
B) Using the regression equation determined in 1) above, calculate the corresponding
loads for the given contact areas (Table 3).
C) Using the equation determined in 2), calculate the corresponding energy for the given
area-load data (Table 4).
4) Match energies between the cadaver and the dummy to transform the area-load data:
Example: A cadaver impact experiment with a rigid interface resulted in an area-load data
point of 236 mm2 and 992 N at an impact energy of 2.09 J. Using the table above, this
impact energy corresponds to an area-load data point in the dummy of 653 mm2 and a
load of 1742 N. This transformation shows that the dummy load-area data point is greater
than the cadaver (Figure 6).
This protocol (step 4) was repeated for the remaining cadaver data points in which the
rigid impact interface was used.

The entire procedure (steps 1-4) was repeated for the stiff padding and then the soft
padding.
Anthrommpgphic dummy tests

In addition to the isolated knee tests, car crash simulations were performed with a
whole dummy. These tests were conducted to illustrate the utility of the transformation
methodology to a realistic trauma environment. The tests were conducted on a HYGE
(HYdraulic Gas Energized) test system using a generic buck and a 50th percentile male

dummy. The dummy was placed in the front passenger seat (Figure 7 A,B) placed in the

181

mid-position. A 76.2 mm thick piece of aluminum Hexcel (1.65 MPa Hexcel) served as
the knee bolster (Figure 7 C). The surface of the bolster facing the dummy’s knees was
covered with a sheet of ‘low’ range (2-10 MPa) pressure sensitive film (Figure 7 D).
Similar to the isolated knee tests, the film was encased in a polyethylene packet secured
peripherally to the Hexcel. Two tests were run with the given parameters (Table 5). The
femur load cell and inner/outer knee clevis force transducers were sampled per SAE
specification 1211. The clevis transducers are designed to measure loads transnritted to
the knee via foot and/or tibial contact with the toepan and/or instrument panel. The
femoral load cell is designed to measure axial loads acting on the thigh. Thus, a portion of
the femoral load cell data arises from direct loading of the knee and the remainder
represents contact with components distal to the knee. The loads on the knee were
therefore taken as the femoral load cell data less the clevis load data. The associated
contact area was determined in the same manner as the isolated knee tests. These data
were then plotted on a transformed graph to estimate the cadaveric response.
RESULTS

The load-contact area response for the isolated cadaver and dummy knee were fit with
linear fimctions for a given impact interface (Table 6). In general, the linear fits of the
data yielded high correlation coefficients. For the cadaver, the rigid interface load-area
response was significantly steeper than with padding (Figure 8). There was no difference,
however, in slopes between soft or stiff padding. Analysis of the y-intercepts revealed that
the rigid interface curve originated near the origin. The padded interface curves, however,

intercepted the y-axis well above the origin.

182

For the dummy, the three impact interfaces resulted in three distinct load-area curves
(Figure 9). The slope of the curves increased with increasing impact interface stiffiress. In
addition, the rigid and 3.3 Hex curves appeared to originate from nearly the same point:
400 sq m and zero load. The 1.65 Hex curve originated from near the origin.

As in the example given in the Methods section, the load-area data in the dmnmy was
always greater than the cadaver load-area data for a given impact energy. This was true
for all impact interfaces. Partly because of this, it was necessary to forward extrapolate
some of the padded durmny data to allow transformation of some padded cadaver data
points. The transformations are shown for each of the interfaces (Figures 10-12). A
consistent x-y scale was used for all of the graphs to allow a direct comparison between
the graphs. A summary plot of all the transformed cadaver knee load-area data for the
rigid, 3.3 MPa Hexcel, and 1.65 MPa Hexcel interfaces to the dummy knee allows a
comparison of the different data associated with the different interfaces (Figure 13).
Anthrommorphic dummy tests

The femur/clevis loads, and associated contact areas for the two sled tests were similar
for both tests (Table 7). These load-area data points can be plotted with the data in Figure
13 to provide a comparison with the transformed cadaver data (Figure 14). The durmny
load-area data points from the sled tests fall on the 1.65 Hex curve and at a reduced load
and area than the majority of the transformed isolated tests.

DISCUSSION
The main objective of the current study was to develop a methodology to convert

impact induced load-area data for the cadaver knee to the dummy knee. Currently, the

183

prediction of knee injury in car crash simulations is assessed with an instrumented dummy.
The current injury criterion is based solely on a peak load associated with bone fracture in
the human cadaver. However, experimental and theoretical studies show that the injury
tolerance of the human knee to impact loading increases with increasing contact area over
the knee. Accounting for the load and area over the knee would be yield a more sensitive
injury criterion.

To develop a wide range of load-area data points, three impact interfaces were used to
impact 16 pair of isolated human cadaver knees and the Hybrid III isolated dummy knee.
Separate experiments conducted at increasing levels of energy to gradually increased the
load and contact area. The transformation of the cadaver load-area data to the dummy
was performed by comparing similar levels of impact energy. For a given impact interface,
the load and area in the dmnrny was always greater than the cadaver. Linear functions
provided a good fit of the load-area data for the cadaver and the dummy. The slope of the
load-area curves for the dummy were significantly different based on the impact interface.
The stiffer interfaces resulted in more elevated slepes. For the cadaver, the slope of the
rigid interface curve was greater than with padding, however, no distinction could be
made between the padded interface curves.

Analysis of the cadaver and dummy load-area regression lines highlights some
differences between the different interfaces and between the dummy and cadaver knees.
For example, the wide range of y-intercepts showed that a majority of the load-area curves
did not intersect the origin. For the padded cadaver experiments, the load-area curves

intersected the positive y-axis, well above the origin. This implies that a finite impact load

184

 

 

is associated with zero contact area. Clearly, this is not a realistic outcome; in general, it
would be expected that zero load should be associated with zero contact area. The rigid
interface curve did, however, pass very near the origin. One possible explanation for the
difference between the rigid and padded cadaver curves is probably related to the impact
energies used for the cadaver experiments. The rigid interface experiments started at very
low energies which were progressively increased. Thus, data representing low loads and
low contact areas were generated with the initial impact energies for a given cadaver. The
rigid interface experiments provided a good definition of the cadaver response near the
o: .z- . .: -:r‘ :lie expect-"d result (i.e. passing through the origin) occurred. Altemately, in
t’: - padded cadaver experiments, only one impact was delivered per cadaver at an elevated
impact er: .- gy. Thus, there was no experimental data at low loads and low contact areas.
Additional experiments are needed with the padded interfaces to elucidate the cadaver
response for low impact energies.

For the dummy, the load-area curves for all of the impact interfaces passed very near a
point approximately represented by a 400 nrrn2 contact area and zero load. This suggests
that a finite contact area would develop for zero load. Again this appears to be
counterintuitive. However, the shape of the anterior dummy knee might account for the
large contact areas at low loads. The dummy knee has a rounded profile but is essentially
flat in the medial-lateral direction (Figure 15). Thus, even low loads could be associated
with large contact areas due to the inherent shape of the dummy knee.

The utility of the transfonnation presented in the current study could be an assessment

of ‘safe’ pairs of load and area on the knee. Cadaver experiments in which injuries were

185

documented could be used to establish the load-area injury threshold for a given interface
stiffness. If a sufficiently diverse group of interfaces were used, a comprehensive injury
threshold could be developed. Based on the experimental cadaver data presented in the
current study in addition to injury producing experiments not presented in this study,
several fimdamental conclusions can be made.
1) Load and area data points are unique to a given interface. That is to say, a rigid
interface and a styrofoam interface would never yield an identical load and area, regardless
of impact energy. Clearly, knee geometry material properties and will vary between
cadavers and this will result in some data scatter. However, tests conducted on an
‘average’ cadaver of a given anthropometry will yield a similar response as the dummy:
increasing interface stiffness will yield load-area curves of increasing slope.
2) For a given impact interface stiffness, there exists a maximum load (and
associated area) which can be tolerated without an associated injury. While this may
be intuitively obvious, this idea would allow the data fiom the current study to be used in
conjunction with experiments which resulted in injury to develop a more comprehensive
criterion. To illustrate this idea, consider a load area curve for a given interface (Figure
16). After a threshold level of load and area has been attained, injury will occur. A
logistic regression superposed on the load area line would identify the statistical
probability of injury (Figure 17). Injury probability curves could then be added to identify
the injury risk associated with a given load and area (Figure 18).

Similar experiments as those mentioned in the previous example are needed on a wide

range of impact interfaces. The relative risk for each interface could again be determined

186

‘ 1;: 55

—-_.___-_
MA

A

 

 

by conducting logistical regressions along each load-area curve. For example, cadaveric
experiments conducted with several impact interfaces might produce injury and non-injury
data points (Figure 19). Note that as the impact interface stiffness reduces, the load and
area increases before injury occurs. The 50% injury risk has been arbitrarily chosen for
this example and is denoted by the arrows for each interface curve. If the 50% injury risk
points were connected, the entire field of load-area data points associated with this level of
injury risk could be interpolated (Figure 20). Thus, any load-area data point above the
50% curve would be associated with a high probability of injmy. Any data points below
this line, would be associated with a lower probability of injmy. Interestingly, this injury
criterion is independent of the impact interface. Thus, tests can be conducted without
requiring a material’s characterization of the impact interface prior to determining an
associated injury risk. Rather, simply the load and area are sufficient to identify the risk of
a given impact event.

Currently, the automotive injury criterion is based on gross bone fracture. However,
consideration of subfi'acture or microscopic injuries might be warranted to produce a more
conservative criterion. While fiacture level injuries of the patella, tibia, and femoral
condyles are a well documented impact induced injmy, numerous recent studies also
document microscopic bone and cartilage injuries (Armstrong 1985, a:Atkinson 1995,
bzAtkinson 1995, Atkinson 1997, Donohue 1983, Thompson 1991). These injuries may
be even more insidious than gross injuries because they are difficult to detect acutely and
are typically not treated. Further, these subfracture traumas have been associated with

chronic joint pain and disease (Upadhyay 1983, States 1970). Thus, preventing

187

..u - ‘_'.-_..g

I

 

 

subfiacture injuries will be an important consideration in new injury criteria. Incidentally,
there is theoretical evidence that preventing subfracture injuries would yield a
conservative criterion. This is based on the hypothesis that preventing subfracture injuries
will also prevent the fracture level injuries. This follows from the idea that fracture level
injuries begin as subfi'acture microfractures which can coalesce and form gross fractures if
additional energy is imparted to the joint. Future work is needed to develop a
comprehensive injury criterion which accounts for subfi'acture level injuries, different
impact interfaces, and the differences between the dummy and human cadaver knee.
ACKNOWLEDGMENTS
This study was supported by a grant (R49/CCR503607) from the Centers for Disease
Control and Prevention. 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 Mr. Cliff Beckett for technical support. We
are also indebted to Hexcel Corporation and Mr. John Porter for technical and materials
support. Finally, we wish to thank Mrs. Kathy Pearson and the Michigan Tissue Bank for
solicitation of donors and retrieval of joints.
REFERENCES
1. Armstrong C, Mow V, Wirth C (1985) Biomechanics of impact-induced rnicrodamage
to articular cartilage: a possible genesis for chondromalacia patella. In: Amer Acad of
Orth Surg Symp on Sports Med, ed by G Fenerman. pp70—84
2. a:Atkinson PJ, Haut RC (1995) Insult to the human cadaver patellofemoral joint:
effects of age on fracture tolerance and occult injury. 39th Stapp Car Crash
Conference: 281-294
3. bzAtkinson PJ, Haut RC (1995) Subfracture Insult to the Human Cadaver

Patellofemoral Joint Produces Occult Injury. Journal of Orthopaedic Research 13:
936-944

188

10.

ll.

12.

l3.

14.

15.

16.

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. 4lst Stapp Car Crash Conference, SAE: 167-180. paper
973327

a:Atkinson PJ, MacKenzie C, Haut RC (1998) Joint fiacture load prediction using
geometrical, physical and pathological measures: the human knee. SAE International
Congress: paper 980358

bzAtkinson PJ, Newberry WN, Atkinson TS, Haut (1998) A method to increase the
sensitive range of pressure sensitive fihn. Journal of Biomechanics. In review

Bourret P, Corbelli S, Cavallero C (1977) Injury agents and impact mechanisms in
about 350 frontal crashes in field accidents. 21 st Stapp Car Crash Conference 215-258

Donn’nue MJ, Buss D, Oegema T, Thompson R (1983) The Effects of Indirect Blunt
Trauma on Adult Canine Articular Cartilage. The Journal of Bone and Joint Surgery
65-A(7): 948-957

Hartemann F, Thomas C, Henry C, Foret-Bruno J, F averjon G, Tarriere C, Got C,
Patel A (1977) Belted or not belted: The only difference between two matched
samples of 200 car occupants. let Stapp Car Crash Conference 97-150

Haut R (1989) Contact Pressures in the Patello-femoral Joint during Impact Loading
on the Human Flexed Knee. Journal of Orthopaedic Research 7: 272-280

Hering W, Patrick L (1977) Response Comparisons of the Human Cadaver Knee and
a Part 572 Dummy Knee to Impacts by Crushable Materials. Twenty-First Stapp Car
Crash Conference 21: 1015-1053

Melvin J, Stalnaker R, Alem N, Benson J, Mohan D (1975) Impact Response and
Tolerance of the Lower Extremities. Stapp Car Crash Conf 19: 543-559

Morgan R, Eppinger R, Hennessey B (1991) Ankle joint injury mechanism for adults
in frontal automitve impact. 35th Stapp Car Crash Conference: 189-198

Nyquist GW ( 1974) Static force-penetration response of the human knee. 18th Stapp
Car Crash Proceedings. Paper 741189

Otte D (1996) Biomechanics of lower limb injuries of belted car drivers and the
influence of intrusion and accident severity. 40th Stapp Car Crash Conference 193-206

Patrick L, Kroell C, Mertz H (1965) Forces on tehehuman body in simulated Crashes.
9th Stapp Car Crash Conf: 237-259

189

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

Pattimore D, Ward E, Thomas P, Bradford M (1991) The nature and cause of lower
limb injuries in car crashes. 35th Stapp Car Crash Conference: 177-188

Powell W, Ojala S, Advani S, Martin R (1975) Cadaver Femur Responses to
Longitudinal impacts. Stapp Car Crash Conf 19: 561-579

Siegal J, Mason-Gonzales S, Dischinger P, Cushing B, Read K, Robinson R, Smialek
J, Heatfield B, Hill W, Bents F, Jackson J, Livingston D, Clark C (1993) Safety belt
restraintsand compartment intrusions in frontal and lateral motor vehicle crashes:
mechanisms of injuries, complications and acute care costs. The Journal of Trauma

34:736-759.

States J (1970) Traumatic Arthritis-- A Medical and Legal Dilemma. Ann Conf of the
Amer Assoc for Automotive Med 14: 21-28

States J (1986) Adult occupant injuries of the lower limb. In: Biomechanics and
medical aspects of lower limb injuries. Warrendale, PA: SAE 97-107

Thomas P, Charles J, Fay P (1995) Lower limb injuries-the effect of intrusion, crash
severity and the pedals on injury risk and injury type in fi'ontal collisions. 39th Stapp
Car Crash Conference: 265-280

Thompson RC, Oegema TR, Lewis JL, Wallace L (1991) Osteoarthritic changes after
acute transarticular load. The Journal of Bone and Joint Surgery 73-A(7): 990-1001

Upadhyay S, Moulton A, Srikrishnamurthy K (1983) An analysis of the late effects of
traumatic posterior dislocation of the hip without fiactures. J Bone Joint Surg 65: 150-
152

Viano D (1977) Considerations for a femur injury criterion. let Stapp Car Crash
Conference: 445-473

Viano DC, Stalnaker RL (1980) Mechanisms of Femoral Fracture. Journal of
Biomechanics 13: 701-715

190

TABLES

Table 1: Contact area-load-impact energy data fiom impacting the Hybrid III 5 0th male
oecyic details regarding the test protocol.

knee. See text for s

 

 

 

 

 

 

 

 

 

Contact Area (x) Impact Load (y) Impact Energy (2)
550 mm2 1 1 15 N 2 J
950mm2 4180N 9.1 J
1284 mm2 7500 N 20.9 J
1420 mm2 9960 N 35 J
1637 mm2 11850 N 43.3 J

 

 

Table 2: Array of incremental contact areas

 

Contact Area (x)

Impact Load 0!)

Impact Energy (2

 

651 mm2

 

653 mm

 

655 mm

2
2
2

 

657 mm

 

 

 

 

 

Table 3: Regression-derived load data points associated with the given contact areas.

 

 

 

 

 

Contact Area (x) Impact Load (y) Impact Energy (2)
651 mm2 1722 N
653 mm2 1742 N
655 mm2 1762 N
657 mm2 1782 N

 

 

 

 

 

191

I
I

 

 

Table 4: Associated impact energies for the given area-load data points. The load and
area data were inserted into an equation describing a best-fit plane describing the area-
load-energy experimental data.

 

 

 

 

 

 

 

Contact Area (x) Impact Load (y) Impact Energy (2)
651 mm2 1722 N 2.00 J
653 mm2 1742 N 2.09 J
655 mm2 1762 N 2.32 J
657 mm: 1782 N 2.53 J

 

 

 

Table 5: Experimental parameters for crash simulations using a HYGE sled and a
Hybrid III 50th male anthropomorphic dummy

 

 

 

 

 

 

Test Delta V Air Bag Seat Belt
3 14 mph dgrowered none
b 30 mph none 3-point

 

 

 

Table 6: Regression parameters describing the load-area relationship for the cadaver
and dummy knee using three impact inte aces (j variable sti

 

 

 

 

 

 

 

 

 

 

 

 

 

cadaver m b

rigid 6.50 130.40 0.63
3.3 Hex 2.15 1758.00 0.94
1.65 Hex 3.17 615.30 0.98
dummy in b r2

rigid 10.06 -4828.00 0.98
3.3 Hex 3.51 -1374.00 0.99
1.65 Hex 1.42 50.35 0.99

 

 

BSS .

Table 7: Load-area data from experimental crash simulations with a whole dummy.

 

 

 

 

 

 

 

 

Test Limb Contact area (mmz) Peak load (N)
a right 3945.1 5535
a left 4058.5 5665
b right 4891.0 5739
b left 3655.5 6174

 

 

 

192

 

 
 
 

Load Cell

Patella

 

    

   

 

 

337% ?» $11.13;?pr .12:
* * _ a _ femur.
.5. i

;

Flexion i Tibia
Angle
i : i Collar

Figure 1

Load cells

\

 

 

   

 

 

///////?//

l

 

 

193

 

7/////////

f

     
    
 

  

MW
“:90“on

fizofttz’tzttzr

      
 
     

  

           

Rigrd 3.3 Hex 1.65 Hex

 

 

 

 

 

Figure3
12000
3 10000—
2 8000-
45 6000-
3. 4000' y=10.061x-4828
g 20000“ . . R2=10.99
0 500 1000 1500 2000
Contactarea

 

 

 

Figure 4: A best-fit linear fit of the load-area data fiom the rigid interface experiments on
the dummy knee.

194

   

Impact Energy

Figure 5: A best-fit plane describing the experimental area-load-energy data points. The
plane was determined by iterating to identify the parameters which minmized the surn-
squared error. Please see the text for more details on this procedure.

 

1800
1600 -
1400 - Dummy
1200 -

1000 —
800 -
600 - Cadaver
400 -
200 -

o

 

Impact load

 

 

 

0 1 00 200 300 400 500 600 700
Contact area

 

 

 

Figure 6: An example of transforming one load-area data point fiom the cadaver to the
dummy

195

 

   
 

Knee Bolster _
Y }' r—'

\

  

r .00" .
is c1 fittest.

Figure 7: B

Figure 7: The pre-test experimental setup for HYGE sled crash simulations (A,B).

196

    

Knee

 

     
  
    
 

 

 

 

 

 

 

Bolster
Load Cell
Knee Clevis
Transducer
Toepan
Figure 7: C
Pressure

sensitive film

Polyethylene ~p.
packet

 

Side View of knee bolster
with attached film packet
Figure 7:D

Figure 7: A schematic of the durmny and the Hexcel knee bolster (C). A schematic
sideview of the knee bolster showing the pressure film packet.

197

 

 

 

 

 

 

 

 

 

 

7000 Rigid 1.65 Hex; ",3
6000 ° . ‘ ,xt’iv’i"
z 50(1) . (915556"; 3.3 Hex
4000 . . ”a"
3000 I; ’21:)”
1000 .5’.
0 500 1000 1500 2000 2500
Contact Area (sq mm)
Figure8
12000
10000-1
Cé: 8000 .. 3.3 Hex”
‘U 60001

 

 

 

 

500 1000 1600 2000 2500 3000 3500
Contact Area (sq m)

 

Figure 9

198

 

 

 

 

14000

12000

10000

8000

6000

Impact load (N)

4000

2000

 

- Rigid interface

 

. cadaver

 

- dummy

 

 

 

 

 

 

l l

0 2000 4000

1

6000

Contact area (sq mm)

8000

 

 

Figure 10

 

 

14000

12000

10000

8000

6000

Impact load (N)

4000

2000

 

3.3 MPa Hexcel

 

 

. cadaver
udummy

 

 

 

 

 

 

7

0 2000 4000

l

6000

Contact area (sq mm)

8000

 

 

Figure 1 1

199

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Contact area (sq m m)

12000
1.65 MPa Hexcel ,
10000 — /'
/
/
A /
5 8000 - /./
D /
8 6000 /’
‘5 o/
6
Q
g 4000 -
x cadaver
0 I l I
0 2000 4000 6000 8000
Contactarea (sq mm)
Figure12
Cadaver load-area data transformed to the
dummy
14000
,-
12000 - ' ' /
R'9_'° 3.3 MPa .
2 10000 " I, ’I” Hexcel //0
v , /
g 8000 - I / /.//
:: f l a/ 1.65 MPa
3 6000 - . ,/
at .’ Hexcel
E 4000 - ’
2000 — j
I
0 . . .
O 2000 4000 6000 8000

 

Figure 13

200

 

 

 

 

Cadaver load-area data transformed to the dummy with

sled test data

 

 

 

 

 

 

 

14000 I
12000 3 Rigid ”3’ 3.3 MPa
:2: 10000 — ," X" Hexcel / x"
1’ 8000 ~ I" " - / /
§ '1' 1’ /,/ / 1.65 MPa
g 6000 ‘ ,' / A“ ‘ Hexcel
_g_ 4000 4 I
2000 , . test a
- test b
0 I I I
0 2000 4000 6000

Contact area (sq mm)

 

8000

 

 

 

Figure l4

"5

8

. 3
Th1 gh I g
'8

2

Top View of Flexed
Dummy Knee

 

 

Tibia

Side View of Flexed
Dummy Knee

Figuew 15

201

 

 

 

 

Load

 

 

 

 

 

 

 

 

 

 
 
   
   

 

 

 

 

Area
Figure 16
Load :100% chance
of injury
0% chance 0 Injury
Of injury 0 0 No Injury
Area

 

 

Figure 17

202

 

 

 

 

 

 

 

 

 

 

 

 

 

Load 75% —100%
50% '
25% ,
0% -
‘ oO .
‘ Area
Figurel8
Rigid Stiff Padding
Load - :1
O ‘ .
. . - ." Soft Padding
I O .o

 

 

 

 

 

 

Figure 19

203

. a“ ..

 

 

Load

 

 

I / /

Higher probability of injury

 

 

 

 

I \50% chance

 

 

 

 

 

 

 

 

‘ . .
/ of Injury
Lower probability of injury
Area
Figure 20

204

mffi

CHAPTER 9

STUDY OVERVIEW AND ADDITIONAL CONSIDERATIONS FOR
COMPREHENSIVE JOINT INJURY PREVENTION

Overview

This concluding chapter will provide an overview of the findings presented in
chapters 2, 4, 5, 6, 7, and 8. Five main topics will be discussed:
1) The types of subfracture injuries and their probable mechanisms:
Injury data from all chapters will be used in concert with the stress analyses to identify
probable injury mechanisms.
2) Injury prevention strategies:
Extraarticular load distribution and joint flexion angles will be discussed.
3) Appropriate injury thresholds I: the influence of inertial loads and stress waves
on injury tolerance.
While inertial loads were briefly discussed in chapters 4, 5, 6, and 7, this topic will receive
additional attention in this chapter. In addition, the influence of stress waves on joint
injury will also be investigated. These topics are important when comparing impact data
from various experimental designs.
4) Appropriate injury thresholds II: logistic regressions:
A method will be suggested to incorporate the above mentioned issues into an injury risk
curve.
5) The use of theoretical models to design injury prevention strategies:
The implementation of the injtuy criteria developed in the current study into ‘whole-body’

theoretical models will be discussed. Such models could be used in car crash simulations

205

to predict the injury risk to the knee while also providing knowledge of whole body
kinematics. Thus, the injury risk to the knee, head, chest, etc. could be predicted from a
single model.

1) The types of subfracture injuries and their probable mechanisms:

Three main classifications of subfiacture injuries were identified in chapters 2, 4, 5,
and 7: vertical occult microfi'actures, horizontal occult nricrofractures, and cartilage
fissures. All injuries were detected in the region of contact and were associated with
elevated shear or tensile stresses. It is noted in passing that gross fractures were also
documented and were nearly always observed in the patella, although there were some
femoral condyle fractures. The most common patellar fiacture was transverse followed by
comminuted. The experimental data suggests that the occult tramnas may be the
precursor to gross fracture. This will be discussed in further detail below.

Vertical occult traumas were detected in regions of high contact pressure gradients
at the cartilage surface. This coincided with elevated tensile and shear stresses in the
subchondral plate, however, based on the direction of the injury and the principal stress
directions, the injury is most likely tensile in nature. Viewed on a saggital plane, the
vertical occult traumas traversed the subchondral plate and extended into the deep zones
of the cartilage but did not communicate with the articular surface. The trauma also
extended in the opposite direction into the trabecular bone. In some of the fracture
specimens it appeared that the gross fi'acture was simply the advanced growth of a vertical
occult trauma. This idea was supported by several of the subfracture histological
specimens in which vertical occult traumas were documented which had almost
communicated with the articular surface. These traumas had also extended in the opposite
direction to the anterior surface of the patella. While this injury was technically still
occult, an additional amount of energy imparted to the joint would have likely opened the

occult fracture into a transverse or comminuted patellar fiacture.

206

Horizontal occult injuries were documented in three categories: separation at the
tidemark, separation at the calcified cartilage-subchondral bone interface, or horizontal
cracks confined to the subchondral bone. The first two injuries were noted in regions of
elevated shear stress in the cartilage at the tidemark. These stress maxima occurred in
regions of high contact pressure gradients at the cartilage surface. Microcracks in the
subchondral plate were associated with elevated shear stresses in the subchondral plate. In
some of the histological data there was occult evidence of a gross fracture connected to a
horizontal occult injury. This suggests that horizontal as well as vertical injuries may be
precursors to gross fracture.

Cartilage fissures at the articular surface occurred in the center of contact which
coincided with regions of elevated shear stress. Tensile stresses were ruled out as a
potential mechanism because these maxima were located adjacent to and outside of the
region of contact, remote to any evidence of fissures. The fissures appeared in the shape
of a 'V‘: the articular surface was opened and the fissure extended into and terminated in
the middle zones of the cartilage. It should be noted that it does not appear likely that
cartilage fissures play a role in initiating gross fi'actures. The histological data shows
several specimens in which fissures were the sole injury. In some cases, however, a fissure
was located over a vertical occult trauma. An additional amount of energy could open up
the fissure to expose the vertical occult trauma and this would be termed a gross fracture.

However, it was the vertical occult trauma that actually initiated the injury to the bone.

2) Injury prevention strategies:

Chapters 5, 6, and 7 show that increased extraarticular contact area (via padding)
reduces the incidence of fiacture and subfiacture level injuries. Increased contact area
allowed higher loads to be tolerated without gross or microscopic injury. Math modeling
of the patella also showed that increased contact area was associated with significant

reductions in tensile stress in only the subchondral bone. This finding suggested two

207

 

conclusions: 1) vertical occult injuries probably initiated in the subchondral plate, and 2)
preventing the occult injuries will prevent the incidence of gross fracture.

While increasing the contact area reduced injury, there were limits on this injury
prevention strategy. There is a maximal contact area that can be achieved over the knee.
In cases in which the maximal contact area is achieved, there would be a great deal of
penetration into the impact interface. This may have negative implications on body
kinematics in car accidents. For example, a large penetration into a knee bolster, while
protecting the knee, may precipitate unwanted whole body kinematics. Deep penetration
into the bolster would move the entire body forward putting the head in potentially life-
threatening contact with the header or windshield. Thus, injury prevention in automobiles
must be viewed as a system in which body kinematics will change as a function of the car
interior.

Another kinematic parameter which affected injury potential in the current study
was the knee joint flexion angle. Significantly greater loads and energies were required to
produce injury at the higher versus lower flexion angles. This suggests that for scenarios
in which knee impact may occur (i.e. car accidents), the car interior should be designed
such that impact will occur at higher flexion angles. The experimental data showed that at
lower flexion angles (S 90°) significantly less load was required to produce fracture and
subfracture level injuries than at higher flexion angles (120°). This was because the patella
attempted to migrate proximally at lower flexion angles. This put the patellar tendon in
tension which caused an avulsive injury at the distal pole of the patella. This may explain
the lower fi'acture loads for such experiments as the distal portion of the patellar bone
tissue is loaded in tension. This can be contrasted to experiments at higher flexion angles
in which the patella was loaded in compression as it was driven into the femoral condyles.
In these experiments, the fracture load was nearly twice that for the lower flexion angles.
Unfortunately, the influence of flexion angle is not reflected in the current automotive

industry safety standards. These standards are based on loads measured with dummies

208

i...

 

with an injury tolerance referenced to cadaver experiments conducted solely at 90°. Thus,
the dummy cannot accurately predict injury at flexion angles other than 90°. At higher
flexion angles, the durmny will understate the tolerance to injury, while the reverse will
happen at lower flexion angles. It will be important for these considerations be included in

future dummy designs.

3) Appropriate injury thresholds I: the influence of inertial loads and stress waves
on injury tolerance.
Inertial Effects

A comparison of the data from the current study with past studies reveals a great
difference in fiacture load magnitudes. For example, at a 90° flexion angle, experiments
on whole, seated, fresh cadavers show fiacture at loads of 10-17 kN. In the current study,
however, experiments at the same flexion angle were conducted on isolated knees and
fiacture loads of 5-7 kN were recorded. These data vary by a factor of approximately
two. Three potential reasons for these differences are: inertial motion of the cadaver,
stress wave effects, and/or strain rate effects. The latter is discounted because the tests on
the isolated knees and the whole cadavers all occurred in a similar time fiame (about 15
ms). The other two candidates will be discussed below.

'Inertial loads' refers to those loads arising from the rigid body acceleration of a
mass where the force is the product of the mass and acceleration. In tests conducted on
rigidly constrained, isolated knees, the resulting acceleration is zero. Thus, the load that is
measured is simply due to the resistance of the joint tissues to undergo deformation.
Altemately, tests conducted on whole cadavers result in an acceleration of the body, in
addition to deformation of the tissues. Thus, injury in such experiments will only occur if
sufficient deformation of the joint tissue occurs. However, because some of the load is
used to inertially move the body, additional load would be needed to achieve an incipient

level of deformation at the knee. To study this problem, a lumped mass model was

209

‘1
l
l
1
j.
I.

 

 

developed describing the body as two masses: one representing the lower extremity and
the other the upper torso. The two body segments were connected by two discrete
elements: a spring and dashpot positioned in parallel to incorporate compliance of the
system. A pair of non-linear, coupled differential equations of motion were written to
describe the impact response of the system. To model the whole cadaver, seated
experiments, reasonable masses for a 50th percentile male were prescribed to the two
body segments (Figure 1, 2). A frictional force was modeled under the upper body to
simulate contact with the seat. An impulsive load was imparted to the 'knee'. The pulse
was described as a haversine load-time curve with a 6700 N peak and a 15 ms duration
(Atkinson 1995). To model the isolated knee experiments, the same impulsive load was
prescribed to the knee, however, the upper body mass was prescribed as infinite to
simulate a rigid fixation at the mid-thigh (Figure l, 2). The equations of motion were
solved numerically in MathCad and sample input and output data for the whole body and
isolated knee cases are provided (Figures 3, 4).

A sensitivity analysis was conducted to determine the influence of some of the
model parameters on the load in the spring and the inertial load associated with the lower
extremity (m*a(t)). Specifically, sensitivity analyses were conducted for the impulse time
duration (Figure 5), spring constant (Figure 6), and body segment mass (Figure 7)
parameters. For infinitesimal impulse time durations (1 us), the load in the spring for both
models was zero and the inertial force of ‘m’ (the mass of the lower extremity) was
identical to the peak applied impulsive force. This occurred because the mass m
experienced a very small displacement for short pulse durations. This is the result of its
inertial resistance to motion leading to an infinitesimal spring deformation and an
associated infinitesimal load in the spring. Equilibrium was maintained by the inertial force
in which the product of ‘m*a’ identically equaled the applied force. For all pulse durations
greater than zero, the spring force was always higher in the isolated case than in the whole

body case. In addition, the inertial force associated with m was essentially identical for all

210

.‘4-

EA.

 

impulse durations. More specifically, for time pulses in the range (0ms<t<7ms), the spring
force was nonzero. The sum of a spring force and the inertial load approximately equaled
the peak applied force. For longer pulse durations (7ms<t<35ms), the peak spring force
exceeded the peak applied force, with the greatest spring force occurring in the 15ms to
20ms range. With increasing pulse durations, the inertial force associated with m
gradually decreased to approximately 33% of the peak applied force. The elevated spring
forces arose fiom the spring’s resistance to the relative displacements of the two body
masses and, at the same instant, the load associated with inertially decelerating the mass
m. Similarly, the inertial load associated with m decreased at the longer timepoints (>20
ms) as the spring was allowed to deform thus allowing the acceleration of m to diminish at
increasing pulse times.

A sensitivity analysis of the spring constant of the knee again revealed that the
spring force was greater in the isolated knee model compared to the whole body model.
In addition, for the spring constant range (2e+5 N/m < k < 1e+8), the peak spring force in
the isolated model was greater than the peak applied force. The whole body model peak
spring force followed a similar trend as the isolated model, however, the peak forces were
less than the isolated model. For both models, the maximal spring force was documented
for spring constants of approximately 1e+6 N/m. This value is identical to experimental
measurements of the cadaveric knee spring constant. For larger values of k (k > 1e+6),
the inertial force of m was larger for the whole body. This occurred because the elevated
spring constant essentially makes the entire whole body system react and move as a rigid
body. Such rigid body motion is not possible for the isolated knee model, and thus, the
inertial loads are reduced.

A sensitivity of the amount of mass distal to the knee again showed that the forces
in the spring were always greater in the isolated knee model. For infinitesimal mass
values, the peak spring force is identical to the applied load and the inertial force is zero

(because Fina,“ = m’a and m->zero). As the mass is increased (1 kg < m < 20 kg), the

211

51

i1
11-.

 

spring force and inertial force are increased. This range encompasses 14.2 kg, the realistic
mass of the distal lower extremity for the 50th percentile male (Figure 1). For greater
mass values (> 20 kg), the spring force decreased and the inertial force steadily increased.
This is because there is less spring displacement as the mass value increases because the
mass ‘m’ is undergoing less displacement due to it’s increased inertial resistance to
motion.

When realistic values of the model parameters are used (pulse time=15ms, k=le+6
N/m, m=14.2 kg), the whole body model requires 30% more load to achieve the same
deformation as the isolated model. The load in the spring and the relative displacement of
the lumped masses can be thought of as the stress and strain, respectively, in the thigh.
Because tissue failure is considered to be mediated by elevated stresses (and/or strains),
the results of the lumped mass model show that the isolated leg is more prone to injmy for
a given load in comparison to the whole body model. Thus, a comparison of isolated and
whole body impact data should take inertial effects into account to make accurate
transformations between the two experimental protocols. .
Stress Waves

To determine if stress wave transmission could potentially affect the state of stress
in the knee joint (for the whole body and isolated lower extremity models shown in
Figures 1 and 2), an estimation was made of the speed of a stress wave in the thigh. For
such a case it can be shown that the speed is related to the modulus and density by the
following equation: wave velocity = (E/p)°‘5. For this simple analysis, the thigh was
modeled as a tube of cortical bone. Inserting appropriate values (E=3.75 GPa,
p=1.85"‘103 kg/m3) into the wave equation yields a velocity of 1424 m/s. Assuming a 20
cm femur length for the isolated case and 40 cm for the whole femur in the whole body
case allows the calculation of the time required for the wave to pass down the femur. In
the isolated case, the wave reaches the rigid constraint and returns to the knee joint in

0.28ms. This is far less than a realistic pulse duration of 15 ms discussed in the previous

212

 

 

section. Thus, the initial wave initiated by the first part of the applied load will return to

the knee as additional load is still being applied from the impact. Thus, wave trains would '

be constantly passing down the leg, reflecting off the wall, returning to the knee and
engaging in a complicated construction/extinction interaction with the applied load and
with prior wave trains. This phenomenon is even more complicated because the initial
wave trains initiated by the applied load are compressive. At a fixed node such as the rigid
constraint, the waves are reflected with the same polarity. That is the stress wave returns
to the knee as a compressive wave. However, as the wave reaches the fi'ee end at the
knee, the stress wave is reflected back up the thigh with a reversed polarity. Thus, the
initial trip a wave makes down the thigh is a dilatational wave, compressive in nature. On
the second pass, the wave is tensile. One final complication are local reflections at
material discontinuities such as the tidemark. At such discontinuities, some portion of the
stress will pass through the interface at a diminished magnitude; the remainder of the wave
will be reflected. Clearly, the overall effect of stress wave transmission is highly
complicated for the above mentioned reasons. In order to capture these effects, a simple
finite element model was constructed to model stress wave transmission in the isolated leg
and whole body models. Four noded elements were used to develop an axisyrnrnetric
model of the knee and thigh (Figm'e 8). The distal end was comprised of trabecular bone
(E=350 MPa, v=0.3), patellar and femoral subchondral plates (E=3750 MPa, v=0.3), and
a layer of articular cartilage (E=20 MPa, p=0.49) (Figure 9). The remaining proximal
portion of the fernm' was modeled with u'abecular bone. For the whole body model, a
layer of very soft elements were placed at the ‘hip’ to model the nonrigid interaction with
the seat back. Similar to the lumped mass model of the previous section, the applied
forcing function was a compressive haversine with a peak load of 6700 N and a pulse
duration of 15 ms. The quasistatic solution was also computed as a reference for the
dynamic stress magnitudes. The maximum and minimum principal stress histories were

acquired and the maximum shear stress history was determined fiom these data (tn... (t) =

213

11.,

 

 

(lo.(t)-o3(t)|)/2). These data were tabulated and compared between the whole body model
and isolated leg model and the quasistatic solution (Table 1). Peak stresses were
determined as well as ‘average’ stress values in which the stress history was averaged over
time.

Reviewing the stresses in the knee as function of time show the stress wave
moving through the knee and into the thigh (Figure 10). The shear stress magnitude
(which reflects the maxirmnn and minimum principal stress histories) and wave frequency
were greatest in the patellar subchondral bone for the isolated knee model (Figure 11).
Within a given model (isolated or whole body), the shear stress magnitude and wave
fiequency were greatest in the patellar subchondral bone versus the femoral subchondral
bone (Figure 12). This might explain, in part, why patellar injuries are more common than
femoral fractures. With regard to stress magnitudes and temporally averaged stresses, the
isolated model suffers approximately 20% greater stresses than the whole body model
(Table 1). This suggests that some additional load would be required in the whole body to
achieve the same stress state as the isolated lmee. Assuming a very simple proportional
relationship (although it is probably nonlinear), a 20% increase in load in the whole body
case would approach the isolated case. Regardless of the magnitude, as in the case with
inertial effects, a. comparison of isolated and whole body impact data should take stress

waves into account when comparing isolated knee and whole body data.

4) Appropriate injury thresholds II: logistical regressions:

To identify appropriate injury thresholds based on the data presented in the current
dissertation, a logistical regression was performed on load-area data from a series of rigid
interface cadaver tests (Figure 13) in which fracture and subfracture data are represented.
This allowed injury risks to be associated with specific load-area data points. The injtny
risk for the isolated leg (for which the experiments are based) were used to identify a load-

area data point associated with a 50% chance of fracture. This datapoint was then

214

1:"?

1.-

 

 

 

increased by 50% (30% for inertial effects and 20% for stress wave effects, see above) to
estimate the associated risk for the whole cadaver. These data points were then multiplied
by 70% to estimate the 50% injury risks for subfiacture trauma (Figure 14). The specific
level of 70% comes from a study in the human cadaver knee (Atkinson 1995) which
showed that subfracture injuries were commonly associated with subfiacture impacts at
70% of the contralateral fracture load. Tabulations of the fracture and subfiacture 50%

injury risk data were tabulated for the isolated leg and whole body cases (Table 2).

5) The use of theoretical models to design injury prevention strategies:

To take the data fiom the current study and apply it in a pragmatic setting, models
useful to injury prevention in car crashes were initiated. The isolated dummy lower
extremity’s interaction with aluminum hexcel (a simulated instrument panel) was modeled
as a first step in designing safer car interiors (Figure 15). This simple model was
constructed because corresponding data (presented in Chapter 8) already existed for
experimental validation. The thigh, hip, shin, and foot were modeled as rigid, structural
bodies. The vinyl skin of the dummy knee and the hexcel were modeled with finite
elements. The vinyl was modeled as a viscolelastic solid which obeyed the ‘Zener’
material model which is essentially a maxwell discrete elemental model in parallel with a
spring (Figure 16). The Hexcel was modeled as an elastoplastic solid with different
modulae assigned to each orthogonal material direction. For a given material direction
there were three values of the modulus: the initial slope (E1), the plateau (E2), and the
final steep slope (E3) as the hexcel bottoms out (Figure 17). The different modulae are
encountered at specific compaction factors which are defined as the current volume
divided by the initial volume. This results in nine modulae and six compaction factors
which were determined experimentally by compressing carefully machined samples of
hexcel material. The vinyl skin material parameters were taken from previous,

unpublished work on the dummy head and then the parameters were ‘tuned’ based on

215

 

 

 

baseline experimental validation (bulk modulus=0.82 GPa, long term shear
modulus=0.00074 GPa, short term shear modulus=0.00441 GPa, decay constant= 2.48 s'l
). By impacting the dummy knee with a free flight inertial mass at the same energy as
corresponding impact experiments, the model successfully predicted the peak load and the
load-time history within 3% (Figure 18). Future work with this model will include placing
the whole durmny in a simulated vehicle to conduct simulated car crashes. By
theoretically estimating the load and contact area, the injury risk to the knee can be
ascertained. Thus, variables such as instrument panel design and seat positioning can be
studied for their effect on lower extremity injury.

REFERENCES

1. Atkinson PJ, Haut RC (1995) Subfracture insult to the human cadaver patellofemoral
joint produces occult Injury. Journal of Orthopaedic Research 13:936-944, 1995.

216

'11-

M—_.——-._
LA . O
1-
.

 

 

TABLES

 

Table l: Stresses values (Pa) for the isolated leg and whole leg solved dynamically

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

and statically.
Max Princ Min Princ Max Shear Max Princ Min Princ Max Shear
Stress: Stress: Stress: Stress: Stress: Stress:
Patella Patella Patella Femur Femur Femur
Isolated leg peak 8.74E+06 -1.06E+07 8.82E+06 8.00E+06 -7.89E+06 7.84E+06
stress
Whole leg peak 9755-1-06 -9.67E+06 7.70E+06 7.20E+06 -7.15E+06 7.03E+06
stress
Peak isolated/ 90°/o 1 10% 1 15°/o 1 11°/o 1 10°/o 1 12°/o
peak whole lgl
Isolated leg 3.27E+06 -3.27E+06 3.27E+06 2.74E+06 -2.7SE+06 2.75E+06
average stress
Whole leg 2.68E+06 -2.68E+06 2.68E+06 2.00E+06 -2.02E+06 2.01E+06
average stress
Average isolated 121.8% 121.8% 121.8% 136.9% 136.1% 136.5%
leg/average
whole leg
Peak isolated 108.7% 130.2% 108.9% 127.8% 97.2% 109.2%
fig/static solution
Peak whole 121.2% 118.7% 95.1% 115.0% 88.1% 97.8%
Eistetic solution

 

 

Table 2: Loads (N) and contact areas (sq mm) associated with a 50% fracture or]

subfracture injury risk for different loading scenarios. The increase in contact area and

load due to whole body inertial effects and stress waves is given.

 

 

 

 

 

 

lnjury=lracture lnjury=sublracture
Load case Contact Impact Contact Impact
Area Load Area Load
Isolated leg 887.5 5856 566 4099
Whole body: increase: inertial effects 1213 7607 790 5325
Whole body: increase: inertial effects 1423 8784 941 6149
+ stress waves

 

 

 

 

 

 

217

 

FIGURES

  

Figure 1: Whole body, seated model (A) and the isolated lower extremity model (B) in
which an impact load is delivered at the knee. ‘M’ (43.4 kg) denotes the mass of the body
proximal to the knee and ‘m’ (14.2 kg) denotes the mass distal to the knee for a 50th
percentile individual. ‘k’ (1e+6 N/m, Atkinson 1995) denotes the spring constant of the
knee and ‘c’ (200 Ns/m) represents the viscous damping coefficient.

 

 

 

 

 

 

 

 

 

 

F(t) — - » va‘v M
m —E[—
k A Friction with seat

 

 

 

 

 

 

 

 

Figure 2: Lurnped mass models of Figure l: the whole body (A) and the isolated knee (B).

218

i1=0..2000

At 2: 0.00001 Whole Body Case
t.i :=i-At

01:.015
P 2:6700
F(t) :=Cb(o:— t)-P-sin l -7z-t
o:
m2=142 M :=43.4000 k := 1000000 c :=200 5 :=9.81 3L:=0.1

 

al(vl,x1,v2,x2,t) :=-l—-(F(t) - ko(xl _ x2) - c-(vl - v2))
m

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a2(vl,xl,v2,x2,t) 2=—1—- lit-(x1 — x2) + c-(vl - v2) — pM-g-Lz-
M M]
lVIo 0
‘10 __ 0
v2o " o
O
bxzo 4
“1+1 VIP” ‘1 (V1? "11’ fli'fli'tfi)“ Numerical integration:
x1i +1 .- xli+ v1 {At Eulerian-tangent method
v2H1 v21+ a2(v1i,xii,v2i,x2i,t‘)-m
bx21+l . . x2i+v21-At
(A)
1 ' .0?
F“) - — x 1 -x 2
l l
B tim e E tim e
In ertial A _
X l . /\ f0 rc c l = .- \\/—
/ massl‘accell
1 l
C tim e '
F tim e
" / S p rin g __,.. a
x 2 M/ ‘ f0 rc e = /\
k "' (x l - x 2 ) 1
l
D tim e G tim c

Figure 3: Sample input code (A, for MathCad) to solve the coupled, non-linear differential
equations of motion (shown in box) describing the whole body case. Sample output as a
fimction of time is shown in subfigures B-G: the forcing function (B), the displacements of
m (C) and M (D), the spring displacement (E), the inertial force of ‘m’ (F), and the force
produced in the spring (G).

219

i:=0.. 2000

A ==000001
ti :=i-At

0t 2:.015

P 2:6700

F(t) := Cb(oz— t.)-P-sin[ (i) 'fl-t]
o:

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

m :=14.2 M z=434000 k := 1000000 c 2:200 5 :=9.81 at 2:0
Ll(vl,xl,v2,x2,t) i=i-(F(t) — k-(xl — x2) - c'(v1- v2))
m
_ 1 v2
v1,x1,v2,x2,t) .=—~[k-(xl — x2) + c-(vl — v2) — )u-M-g-—]
M - Iv2l ,
v10 0
11° - 0
v20 0
0
.a04
“1+1 “1"“ “(flir’dr‘ar’at'td‘m Numerical integration:
”1+1 x11+v1tqm Eulerian -tangent method
v2H_1 v21+ a2(vl 131,, v2i,x2£,t‘)-At
391+: 201+ v2i-At J
A
I
‘ .03
F t I >— ._ h—n
( ) \ x l -x 2 —/
J
B tim e . l
E tim e
T
I
x l /\ Inertial A
- forccl =
1 massl‘accell
. l
C tim e ,
F tim e

 

 

j I

x 2 — — Putting: — /\\:

k*(xl-x2) __,....-’

 

 

 

 

 

 

D tim e G tim e
Figure 4: Sample input code (A, for MathCad) to solve the coupled, non-linear differential
equations of motion (shown in box) describing the isolated knee case. Sample output as a
function of time is shown in subfigures B-G: the forcing function (B), the displacements of
(C) and M (D, this displacement is essentially zero), the displacement of the spring (E),
the inertial force of ‘m’ (F), and the force produced in the spring (G).

220

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 Sensitlvlty codllcient plot 01 1‘0ch versus Impulse duration Isolated 7, +mng m.
!::I +hertial Iorceottheknee
Whole + spring lance
L\ +‘nenid force oi the knee
1m \‘
E
3000 j
\ I
a ...... b ..... N
6000 - ‘1
4000 E
N \x g
2000 ‘ \"I
0 .'
0 DADS 0.01 0.015 0.02 0&5 0.03 0&5 0.04
fig

 

 

ure 5: The inertial forces of m and the force developed in the spring for a range of
impulse durations for the whole body and isolated leg cases. The dotted line represents
the peak applied load at the knee.

 

Force

§§§§

 

12000

10000

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Isolatedknee +3W‘9'°'°°
WWMMhoee washelmeeeprhnconctmt +inenialbrcedknee
+em'n9br00
Wholebodyl +imrtielioroedknee
//4>-\ \\
................... J/// \\
.— 1
i

 

3..

100000

1000000

10000000

Logarithmic eprlng constant of the knee

 

 

Figure 6: The inertial forces of m and the force developed in the spring for a range of
spring constants for the whole body and isolated leg cases. The dotted line represents the
peak applied load at the knee.

221

 

Sensltlvly coefficient pbt of forces versus trace of the dhtel lower extremity

Force

 

0.01 0.1 l 10 100 1G!)
Logrlthrrlc mes of the distal lower extrerrity

 

 

 

Figure 7: The inertial forces of m and the force developed in the spring for a range of
distal lower extremity masses for the whole body and isolated leg cases. The dotted line
represents the peak applied load at the knee.

 

 

Whole body

 

 

Proximal Distal

    
   

 

     

Hip Thigh Knee

 

Isolated leg

 

 

 

 

Figure 8: Simplified schematics representing the finite element models used to study the
effects of stress waves on the impact response of the whole body and isolated knee
experiments.

222

Subchondral Bone

  
    

All proximal elements:
Trabecular bone <—

K

Figure 9: A schematic showing the different material assignements to model the knee.

Trabecular Bone

Min Princ Stress Time

i=0, Load
impulse started

t=0.00020 s

t=0.00040 s

 

t=0.00040 s

     

Figure 10: Thestresses in the knee and distal thigh (the region depicted in Figure 9)
showing the stresses moving up the femur with some reflection at the material
discontinuities.

223

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

9.E+06 1 - - Isohted case
—'Whole body case
8.E+06 fl '.\
7.E+06 [‘1 - :1
I I' 1 fl ' 'I
.. came a 1' .l ‘JI it
s. Ii '1 5 I i' n l'
g "1191‘ II"\ n"
5.8+“ f‘r. I . 'i i I] "i i
g N II'.’I I 'lill|l ml
4E+06 ”'1' I'll ililil “l I
' ii iii ll H In in
) I (It‘ll II I I‘ A 3”
3,506 l i‘ i T y ‘ “ '
ill" "' ”I" II "
2.506 A M A“ I i1]. ‘1 I. ._
i I I
v u I H. y .
1.E+00 -
it w l '
0.E+00 - J I v .
o 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
theme)

 

 

 

Figure 11: Temporal history of the stress wave magnitude (maximum shear) of a node
contained within the patellar subchondral bone for the isolated knee and whole body
models for the first 5ms of impact. Note that the stresses are similar for the first 1 ms.
For the ensuing time points, the isolated model experiences far more peaks and at higher

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

magnitudes.
9.5.004 ‘ ‘
0.!900 .
l 3
{£900 ‘ i‘l '
...... l l I”; !i .'
g i' I 5 I l” l
3 . 'l I .1 I‘ I ,I I 1 ii '
5.5.00 ‘ it | . It I ' [il
a : I : . ii i' 'l ‘ :l g:.
4.5+“ I ii I ‘ ' | I I ‘
' H. . l | 1 I i I I | A a ' ' L_':
3.5.0., Ii 4" II 'I' ,"u'i :' It'll *
l i ii l'l " l l‘ | rill ‘ i l
2.5000 l ‘H I % |“ ' i]
I ‘ 111'“ l 1 in“ N l I
1.5900 : Al . ' -
' Vii;
0.E+00 - v . r 4 ,
0 0.5 1 1.5 2 2.5 8 3.5 4 4.5 5
“(.0

 

 

 

Figure 12: Temporal history of the stress wave magnitude (maximum shear) of two nodes:
one contained within the patellar and one contained within the femoral subchondral bone
for the isolated knee model only for the first Sms of impact. Note that the stresses are
largely dissimilar similar for all time points. The patellar suffers more oscillations and the
stress amplitudes are typically greater than for the femur.

224

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

mm 50% Fracture inpry W.N.,
moon risk for the whole body —74—;
50% Fracture in'pry /

°°°° risk for the isolated leg I / \
2 70m . ‘ I /
3 - /
8 0000 0/ a
..l e . ° :
‘5 5°00 C
a , . I K \ 4
g m o I

3000 . .1 4 ‘

. . . rigid interface-no injury
zooo . " . _
. I - rigid interface-fracture
1000 e . A __
. Q. 0 D
o o no 400 sec no 10m 1200 1400 1000 race
Contact Area (sq nun)

 

 

Figure 13: 50% fracture injury risk points determined for a series of cadaver tests
performed with a rigid interface. The 50% injury risk for the isolated case was increased
1.5 times for the whole body case to account for inertial effects (30% increase) and stress
waves (20% increase). Please see section 3 above regarding these values.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

11M -.
l l l
10°00 50% Sibtracture /,/
in risk for the
w ‘h— 50% Stbfl'aclure _ 1Kka may I 5
w ._ in'pry risk for the /
i isolated leg \ - //
v m a I i
. / g
g soon 1 ° ;
-I Q . :
1'5 6000 v \
g 4000 . I [I l
_ e T l
r u \ - . . .4
m . . \ ' e rigid interface-no injury 3
“ a . . . ___,
m g ’ . l'lgld interface-fracture
won 3 . : o D .

o 0 200 400 am m 1m 1m 1400 1K!) mo

Contact Area (sq nun)

 

 

Figure 14: 50% subfi'acture injury risk points determined for a series of cadaver tests
performed with a rigid interface. The 50% risk load-area data points from Figure 13 were
multiplied by 70%, the amount of fracture load required to cause subfiacture injuries

(Atkinson 1995).

225

 

 

 

 

\. .: ' .
.'-.' pl... I'll-v;- I: “

Figure 15: A schematic of a solid model of the dummy lower extremity. The hip is fixed
and the hip, thigh, shin, and ankle are modeled as structural elements. The knee dummy
skin (9.7 mm thick) and hexcel impact interface are modeled with finite elements. The
vinyl material model is shown in Figure 16 and the hexcel material model is shown in
Figure 17. The problem with a Madymo-Parncrash coupled solution in which the
Madyrno software code accounted for the structural elements and the Pamcrash code
accounted for the finite element solution.

226

 

inf

Ginf ' Go B

Figure 16: Viscoelastic material model of the vinyl for the dummy knee.

 

Stress

 

 

 

 

CFl CF2 Strain
B

Figure 17: The material directions (A) and the material model for the hexcel. The
parameters (three modulae and compaction factors-CPI, CF2) were required for all three
material directions and were experimentally determined.

227

 

 

Experimental and theoretical prediction of the load-time history tor
an experiment on the Hybrid Ill dummy knee conducted at v=3.77
mls with a rigid lntertace at a 90 degree flexlon angle

 

8000
7000

 

 

 

 

——experirmnt

6000 l
5000 /// \ »—mm.
.000 / / \\
3000 / / \\
// \\
1000 ~ \\ In
0 ‘ \WL ‘4
time (ms)

 

 

 

 

 

 

 

 

Impact load (N)

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 18: A comparison of the load-time histories fiom an impact experiment on the
dummy and the load-time history predicted by the Madymo-Pamcrash model.

 

 

228

 

"ililillillllilif