Proceedings of the Institution of Mechanical

Engineers, Part H: Journal of Engineering in

Medicine
http://pih.sagepub.com/

Mechanical anisotropy of the human knee articular cartilage in compression

J S Jurvelin, M D Buschmann and E B Hunziker
Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine 2003 217: 215
DOI: 10.1243/095441103765212712

The online version of this article can be found at:

http://pih.sagepub.com/content/217/3/215

Published by:

http://www.sagepublications.com

On behalf of:

Institution of Mechanical Engineers

Additional services and information for Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in
Medicine can be found at:

Email Alerts: http://pih.sagepub.com/cgi/alerts

Subscriptions: http://pih.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://pih.sagepub.com/content/217/3/215.refs.html

>> Version of Record - Mar 1, 2003

What is This?

Downloaded from pih.sagepub.com at NATIONAL CHUNG HSING UNIV on March 31, 2014
 215

Mechanical anisotropy of the human knee articular

cartilage in compression

J S Jurvelin1,2*, M D Buschmann3 and E B Hunziker4

1Department of Applied Physics, University of Kuopio, Kuopio, Finland
2Department of Clinical Physiology and Nuclear Medicine, Kuopio University Hospital and University of Kuopio,
Kuopio, Finland
3Biomedical Engineering Institute, École Polytechnique, Montreal, Quebec, Canada
4 M E Muller Institute for Biomechanics, University of Bern, Bern, Switzerland

Abstract: Articular cartilage exhibits anisotropic mechanical properties when subjected to tension.
However, mechanical anisotropy of mature cartilage in compression is poorly known. In this study,
both con
ned and uncon
ned compression tests of cylindrical cartilage discs, taken from the adult
human patello-femoral groove and cut either perpendicular (normal disc ) or parallel (tangential disc )
to the articular surface, were utilized to determine possible anisotropy in Young’s modulus, E, aggre-
gate modulus, H , Poisson’s ratio, n and hydraulic permeability, k, of articular cartilage. The results
a
indicated that H was signi
cantly higher in the direction parallel to the articular surface as compared
a
with the direction perpendicular to the surface (H =1.237±0.486 MPa versus H =
a a
0.845±0.383 MPa, p=0.017, n =10). The values of Poisson’s ratio were similar, 0.158±0.148 for
normal discs compared with 0.180±0.046 for tangential discs. Analysis using the linear biphasic
model revealed that the decrease of permeability during the o
set compression of 0–20 per cent was
higher ( p=0.015, n=10) in normal (from 25.5×10Õ 15 to 1.8×10Õ 15 m4/ N s) than in tangential
(from 12.3×10Õ 15 to 1.3×10Õ 15 m4/ N s) discs. Based on the results, it is concluded that the mechan-
ical characteristics of adult femoral groove articular cartilage are anisotropic also during compression.
Anisotropy during compression may be essential for normal cartilage function. This property has to
be considered when developing advanced theoretical models for cartilage biomechanics.

Keywords: articular cartilage, cartilage mechanics, anisotropy, modulus, permeability

NOTATION sile modulus in the direction of collagen
brils, i.e. paral-

lel to the articular surface, while the axial compressive
E Young’s modulus modulus of the super
cial layer is lower than that of the
H aggregate modulus deeper tissue [4]. The di
erence in the cartilage mechan-
a ical properties during tension and compression has pro-
K hydraulic permeability
found inuences on the deformation characteristics of
n Poisson’s ratio the tissue under load [5]. It may also complicate the use
of homogeneous, isotropic models for characterization
of cartilage mechanics. New theoretical models that take
1 INTRODUCTION into account tension–compression non-linearity have
been developed to more accurately simulate cartilage
Articular cartilage is structurally inhomogeneous [1], behaviour under uncon
ned or indentation geometries
mechanically non-linear and anisotropic tissue [2]. [6–9].
Cartilage equilibrium sti
ness is typically 5–20 times Although much is known about the anisotropy of car-
higher in tension than in compression [3]. In particular, tilage mechanical properties under tension, anisotropy of
tissue in the super
cial cartilage zone shows a high ten- the mechanical properties during compression has been
less adequately studied. In a preliminary investigation
The MS was received on 4 March 2002 and was accepted after revision with immature bovine carpo-metacarpal cartilage, no
for publication on 23 January 2003.
* Corresponding author: Department of Applied Physics, University of signi
cant variations in the compressive moduli along
Kuopio, POB 1777, FIN-70211 Kuopio, Finland. three mutually perpendicular directions were found [10].
Downloaded from pih.sagepub.com at NATIONAL CHUNG HSING UNIV on March 31, 2014

H02002 © IMechE 2003 Proc. Instn Mech. Engrs Vol. 217 Part H: J. Engineering in Medicine
216 J S JURVELIN, M D BUSCHMANN AND E B HUNZIKER

It is possible that the mechanical anisotropy of articular

cartilage in compression is species-, site-, depth- and/or
age-dependent. Information about the material sym-
metry during compression would be advantageous since
joint loading generates highly complex stress states in
cartilage layers. These include local compressive stresses
both in the directions parallel and perpendicular to the
external load. In the present study, the compressive
characteristics of the human knee articular cartilage were
measured in two directions, i.e. perpendicular and
parallel to the articular surface.
It was hypothesized that during compression, mature
cartilage would reveal anisotropic mechanical properties
that could contribute to the deformation and water ow Fig. 1 (a) Anatomical site (shaded area) for harvesting articu-
characteristics of loaded tissue and, thus, be crucial for lar cartilage samples from the human patello-femoral
understanding normal cartilage biomechanical proper- groove. (b) Orientation of normal and tangential discs
ties. With the present approach, the aim was also to (diameter, 1.7 mm; thickness, 1 mm) (polarized light
gather the information that is essential in order to microscopic image)
achieve a realistic, anisotropic model of the mechanical
characteristics of articular cartilage conducted using a custom-built material testing device,
equipped with a high-resolution (0.1 ím) motion con-
troller (PM500, Newport Corporation, Irvine,
2. MATERIALS AND METHODS
California) and a load cell (Model 31, 1 kg range,
Sensotec Corporation, Ohio). For the con
ned com-
2.1 Sample extraction pression, the sample was inserted into a con
ning
chamber and a stainless steel porous
lter (Meyer
Fresh, normal-looking knee joint cartilage was derived
Sintermetal AG, Studen, Switzerland; ~5 pores, 50 per
from the bodies of adult humans (n=10, three females
cent porosity, ~15 mm peak-to-peak surface roughness)
and seven males, aged 23–50 years, typical post-mortem
time 12–24 h). Permission for the biopsies was granted was placed on the articular surface. In the mechanical
tests, a fully automated series of stress relaxation tests
by the Medical Ethics Commission, University of Bern,
(step 10 ím, velocity 1 ím/s) was repeated up to a 20
Switzerland. Osteochondral samples were punched from
per cent strain (Fig. 2a). A new step was automatically
the patello-femoral groove of the femora using a dermal
initiated when the load relaxation was less than
biopsy punch (diameter, 8 mm; Stiefel Laboratorium,
AG, Winterthur, Switzerland ). Subsequently, full- 0.10 g /min. For each step, the complete time–position–
load data were recorded. Aggregate modulus, H , and
thickness (3–6 mm) cartilage layer was removed from a
Young’s modulus, E, were determined from the con
ned
the underlying subchondral bone with a razor blade.
and uncon
ned tests respectively, using the equilibrium
From the 8 mm discs, two adjacent, cylindrical 1.7 mm
stress–strain data in the linear range ( Fig. 2b). Based on
diameter discs were punched perpendicular (normal
the linear, isotropic biphasic model [11], Poisson’s ratio
direction) and parallel (tangential direction) to the
articular surface ( Fig. 1). Tangential discs were cut in a was estimated by comparison of the equilibrium
random direction along the surface plane, i.e. the disc responses of con
ned and uncon
ned tests [5]. This
model was also curve
tted to experimental con
ned
orientation was not controlled with that of the split lines.
stress relaxation measurements in order to determine the
All discs were cut to a length of 1 mm using a custom-
hydraulic permeability as a function of o
set strain [11].
designed slicer attached to the stage of a light micro-
scope. When cutting the discs, care was taken to exclude
a similar thickness of super
cial and deep radial tissue
2.3 Statistical analysis
from the disc. Before mechanical testing, the samples
were equilibrated in phosphate-bu
ered saline (PBS ), Wilcoxon matched-pairs, signed-ranks test was used for
pH 7.4. the statistical comparison of the mechanical parameters
obtained from the normal and tangential discs.
2.2 Mechanical testing

All normal (n =10) and tangential (n=10) discs were 3 RESULTS

tested in con
ned compression [5]. In addition, mechan-
ical testing in uncon
ned compression was conducted Typical biphasic curve
ts for the con
ned compression
for
ve (out of ten) disc pairs. Mechanical tests were data of normal and tangential samples are presented in
Downloaded from pih.sagepub.com at NATIONAL CHUNG HSING UNIV on March 31, 2014

Proc. Instn Mech. Engrs Vol. 217 Part H: J. Engineering in Medicine H02002 © IMechE 2003
 HUMAN KNEE ARTICULAR CARTILAGE IN COMPRESSION 217

Fig. 2 (a) Typical load relaxation behaviour of human knee

articular cartilage (normal and tangential samples, see
Fig. 1) in stepwise con
ned compression test (step,
10 ím at 1 ím/s up to a 20 per cent strain). Fig. 3 Typical curve
ts of isotropic biphasic model for
(b) Equilibrium stress–strain response of the normal con
ned compression data of normal (a) and
sample during the uncon
ned and con
ned com- tangential (b) samples
pression tests. The slopes in the linear range yield
Young’s modulus, E , and aggregate modulus, H , for
s a Table 1 Aggregate modulus, H (n=10), Young’s modulus,
uncon
ned and con
ned compression tests respectively a
E (n= 5), Poisson’s ratio, n (n=5), of human knee
articular cartilage derived for cylindrical discs cut
Fig. 3. Tangential samples showed signi
cantly higher perpendicular (normal) or parallel (tangential) to the
equilibrium moduli than the normal samples under con- articular surface. Values are means±SD

ned compression (1.237±0.486 MPa versus 0.845± Direction H (MPa) E (MPa) n
a
0.383 MPa, p=0.017). The values of Poisson’s ratio
were similar, 0.158±0.148 and 0.180±0.046 for normal Normal ( d ) 0.845±0.383 0.581±0.168 0.158±0.148
Tangential (? ) 1.237±0.486* 0.854±0.348 0.180±0.046
and tangential discs respectively ( Table 1). The typical
variation of the permeability values, as a function of the * p=0.017 ( Wilcoxon matched-pairs, signed-ranks test).
o
set strain, is presented in Fig. 4. The mean decrease
of permeability during the compression of 0–20 per cent compression increased, the permeability showed a
was higher ( p=0.015) in normal (from 25.5×10Õ 15 decrease which was signi
cantly higher in the perpen-
to 1.8×10Õ15 m4/ N s) than in tangential (from dicular direction. In the present study, values of
12.3×10Õ 15 to 1.3×10Õ15 m4/ N s) samples ( Table 2). Poisson’s ratio were similar in both directions.
The mean value of 0.854±0.348 MPa for the com-
pressive Young’s modulus in the tangential direction is
4 DISCUSSION lower than the typical values (>5 MPa) of tensile
Young’s modulus in the same direction [3]. The com-
The mechanical properties of the adult human knee pressive sti
ness of articular cartilage is primarily related
articular cartilage in con
ned and uncon
ned com- to the matrix proteoglycan content [12], which was simi-
pression were characterized in two perpendicular direc- lar in both normal and tangential discs (both discs were
tions, i.e. perpendicular and parallel to the articular prepared from the same relative tissue depth). Since the
surface. The mechanical characteristics were anisotropic: collagen content was also equal in both samples, the
the equilibrium moduli were higher in the direction par- statistically signi
cant di
erence in the values of equi-
allel to the articular surface as compared with the direc- librium modulus has to be related to di
erences in
tion perpendicular to the surface. As the o
set the three-dimensional tissue organization of matrix
Downloaded from pih.sagepub.com at NATIONAL CHUNG HSING UNIV on March 31, 2014

H02002 © IMechE 2003 Proc. Instn Mech. Engrs Vol. 217 Part H: J. Engineering in Medicine
218 J S JURVELIN, M D BUSCHMANN AND E B HUNZIKER

water ow characteristics depend signi
cantly on the

strain state of the tissue and may be non-uniform
throughout the tissue in loaded cartilage.
Earlier measurements on the normal and tangential
permeability suggest a similar or even higher per-
meability in the tangential direction [16–18]. At the typi-
cal strains used for cartilage mechanical testing (15–20
per cent), no statistically di
erent values were detected
for permeability in the normal and tangential directions.
Instead, with smaller strains (5–15 per cent), cartilage
permeability in the normal direction was higher than in
the tangential direction. In any comparison of the pre-
sent results to permeability values derived in earlier
studies it is important to note the di
erences in the
measurement techniques. The current study used
matched, cylindrical samples in two perpendicular direc-
tions (normal and tangential discs), that were loaded in
con
ned geometry and the permeability was derived in
the direction of the axial load. In earlier studies,
Fig. 4 Hydraulic permeability of the human knee articular measurements were typically conducted by loading carti-
cartilage (normal and tangential samples) as a function lage discs, cut perpendicular to the articular surface, with
of the o
set strain. The tangential disc shows a higher a geometry that allowed water to ow out from the tissue
change in permeability as a function of strain only in a normal or tangential direction. Estimates for
the normal and tangential permeability were then based
Table 2 Permeability, K, versus compressive strain, e, of on the calculations utilizing Darcy’s law. Interestingly,
human knee articular cartilage derived for cylindrical Quinn et al. [19] modelled the network of glycosamino-
discs cut perpendicular (normal) or parallel (tangen- glycans in cartilage to have a random geometrical organ-
tial ) to the articular surface. An exponential function ization and predicted that, against an increasing o
set
K= K +A eÕe/t was
tted to the experimental data. strain, the permeability would decrease more rapidly in
0
Values are means±SD, n=10
the direction perpendicular to compression than in the
Direction K A t parallel direction. The possible di
erences in the present
0
and previous results on cartilage permeability suggest
Normal ( d ) 1.75±1.82 23.7±12.5 2.5±1.4
that there are strain-related, non-symmetrical confor-
Tangential (? ) 1.26±0.76 11.0±5.4* 3.0±1.7
mational changes occurring in the matrix pore structure
* p=0.017 ( Wilcoxon matched-pairs, signed-ranks test). during compression.
In the current study, the most widely accepted theor-
structures, especially that of collagen, between normal etical model was applied, i.e. the linear biphasic model
and tangential directions. Indeed, Khalsa and Eisenberg of articular cartilage [11] for the analysis of experimental
[13] demonstrated, using simultaneous measurements of con
ned compression measurements. To justify the use
axial and radial stresses during con
ned compression, of the model the authors attempted to satisfy the model
that the osmotic stress produced by proteoglycans assumptions as well as possible. The most super
cial and
cannot completely explain the equilibrium mechanical deepest tissue was excluded to obtain more homo-
response of articular cartilage but the collagen network geneous samples with an isotropically oriented collagen
also controls the equilibrium sti
ness. Earlier, Maroudas network. Also, the study was limited to the use of a low
[14] suggested that the collagen
brils also contribute ramp speed ( 1 ím/s) and small strain steps (10 ím).
to the hydraulic permeability of cartilage. Mow and However, the theory did not perfectly agree with the
Lai [15] empirically demonstrated that there is an experimental measurements ( Fig. 3). Obvious exper-
exponential decrease in permeability with increasing imental diculties are present when conducting con
ned
dilatation. In the present study, the strain-dependent compression testing of articular cartilage. Both with
di
erences in the hydraulic permeability are evidence normal and tangential samples, the large reduction of
that cartilage matrix is not a homogeneous gel possessing permeability against the increased o
set strain may be
isotropic mechanical characteristics during compression. partially artifactual due to possible changes in the bulg-
By idealizing the cartilage tissue to consist of a water- ing of tissue matrix into the pores of the compressing

lled porous matrix with a typical pore size of 2–4 nm platen [20, 21]. Also, as indicated by the similar stress
[14], the current results suggest that the pores are not response at 0–5 per cent strains in both con
ned and
spherical but instead have a form and/or orientation that uncon
ned compression tests ( Fig. 2b), a non-perfect
t
creates varying, anisotropic friction for water ow. The between the sample and chamber walls may contribute
Downloaded from pih.sagepub.com at NATIONAL CHUNG HSING UNIV on March 31, 2014

Proc. Instn Mech. Engrs Vol. 217 Part H: J. Engineering in Medicine H02002 © IMechE 2003
 HUMAN KNEE ARTICULAR CARTILAGE IN COMPRESSION 219

to the high permeability values measured at small strains. isotropic biphasic model for uncon
ned compression of
In contrast, at high o
set strains, signi
cant friction growth plate and chondroepiphysis. J. Biomech. Engng,
between the sample and chamber walls may be present 1998, 120, 491–496.
during the ramp stage when the sample experiences 7 Li, L. P., Soulhat, J., Buschmann, M. D. and Shirazi-Adl, A.
Nonlinear analysis of cartilage in uncon
ned ramp com-
maximal lateral expansion [5]. Nevertheless, using a con-
pression using a
bril reinforced poroelastic model. Clin.
stant sample thickness and diameter under highly stan- Biomechanics, 1999, 14, 673–682.
dardized testing procedures, it is most unlikely that these 8 Soulhat, J., Buschmann, M. D. and Shirazi-Adl, A. A
bril-
potential uncertainties related to experimental set-up can network-reinforced biphasic model of cartilage in uncon-
account for the observed variation in the mechanical
ned compression. J. Biomech. Engng, 1999, 121, 340–347.
characteristics of normal and tangential samples. 9 Soltz, M. A. and Ateshian, G. A. A conewise linear elasticity
In summary, this study characterizes the mechanical mixture model for the analysis of tension-compressionnon-
anisotropy of adult human knee articular cartilage in linearity in articular cartilage. J. Biomech. Engng, 2000,
compression. Signi
cant di
erences were noted in the 122, 576–586.
compressive sti
ness in the direction parallel and perpen- 10 Soltz, M. A., Palma, C., Barsoumian, S., Wang, C. C.-B.,
dicular to the articular surface; however, for both direc- Hung, C. T. and Ateshian, G. A. Multiaxial loading of
bovine articular cartilage in uncon
ned compression. Trans.
tions the equilibrium moduli during compression were
Orthop. Res. Soc., 1999, 24, 888.
lower than those reported during tension. The authors 11 Mow, V. C., Kuei, S. C., Lai, W. M. and Armstrong, C. G.
believe that the present study provides information Biphasic creep and stress relaxation of articular cartilage
which can help in the development of advanced theoreti- in compression. Theory and experiments. J. Biomech.
cal, anisotropic models of cartilage biomechanics. Engng, 1980, 102, 73–84.
12 Kempson, G. E., Muir, H., Swanson, S. A. V. and Freeman,
M. A. R. Correlations between the sti
ness and the chemi-
ACKNOWLEDGEMENTS cal constituents of cartilage on the human femoral head.
Biochim. Biophys. Acta, 1970, 215, 70–77.
This work was supported by grants from: the M. E. 13 Khalsa, P. S. and Eisenberg, S. R. Compressive behavior
of articular cartilage is not completely explained by proteo-
Mueller Foundation, Switzerland; the Swiss National
glycan osmotic pressure. J. Biomechanics, 1997, 30,
Science Foundation, Switzerland; Technology Devel-
589–594.
opment Centre ( TEKES), Finland (Nr 40714/01) and 14 Maroudas, A. Physico-chemical properties of articular car-
Kuopio University Hospital ( EVO 5173), Finland. tilage. In Adult Articular Cartilage (Ed. M. A. R. Freeman),
1973, pp. 131–170 (Grune & Stratton, New York).
15 Mow, V. C. and Lai, W. M. Recent developments in syn-
REFERENCES ovial joint biomechanics. Soc. Ind. Appl. Math. Rev., 1980,
22, 275–317.
1 Hunziker, E. B. Articular cartilage structure in humans 16 Maroudas, A. Fluid transport in cartilage. Ann. Rheumatic
and experimental animals. In Articular Cartilage Dis., 1975, 34(Suppl. 2), 77–81.
and Osteoarthritis (Eds K. E. Kuettner, R. Schleyerbach, 17 McCutchen, C. W. The frictional properties of animal
J. G. Peyron and V. C. Hascall), 1992, pp. 183–199 joints. Wear, 1962, 5, 1–17.
(Raven Press). 18 Mullholland, R., Millington, P. F. and Manners, J. Some
2 Mow, V. C., Ratcli
e, A. and Poole, A. R. Cartilage and aspects of functional behavior of cartilage. Ann. Rheumatic
diarthrodial joints as paradigms for hierarchical materials Dis., 1975, 34(Suppl. 2), 104–107.
and structures. Biomaterials, 1992, 13, 67–97. 19 Quinn, T. M., Dierickx, P. and Grodzinsky, A. J.
3 Akizuki, S., Mow, V. C., Muller, F., Pita, J. C., Howell, Glycosaminoglycan network geometry may contribute to
D. S. and Manicourt, D. H. Tensile properties of human anisotropic hydraulic permeability in cartilage under com-
knee joint cartilage: I. Inuence of ionic conditions, weight pression. J. Biomechanics, 2001, 34, 1483–1490.
bearing, and
brillation on the tensile modulus. J. Orthop. 20 Mansour, J. M. and Mow, V. C. The permeability of articu-
Res., 1986, 4, 379–392. lar cartilage under compressive strain and at high pressures.
4 Schinagl, R. M., Gurskis, D., Chen, A. C. and Sah, R. L. J. Bone Jt Surg. (Am), 1976, 58, 509–516.
Depth-dependent con
ned compression modulus of full- 21 Buschmann, M. D., Soulhat, J., Shirazi-Adl, A., Jurvelin,
thickness bovine articular cartilage. J. Orthop. Res., 1997, J. S. and Hunziker, E. B. Con
ned compression of articular
15, 499–506. cartilage: linearity in ramp and sinusoidal tests and the
5 Jurvelin, J. S., Buschmann, M. D. and Hunziker, E. B. importance of interdigitation and incomplete con
nement.
Optical and mechanical determination of Poisson’s ratio of J. Biomechanics, 1998, 31, 171–178.
adult bovine humeral articular cartilage. J. Biomechanics, 22 Buschmann, M.D. Numerical conversion of transient to
1997, 30, 235–241. harmonic response functions for linear viscoelastic mate-
6 Cohen, B., Lai, W. M. and Mow, V. C. A transversely rials. J. Biomechanics, 1997, 30, 197–202.

Downloaded from pih.sagepub.com at NATIONAL CHUNG HSING UNIV on March 31, 2014

H02002 © IMechE 2003 Proc. Instn Mech. Engrs Vol. 217 Part H: J. Engineering in Medicine