DOI:10.2214/AJR.05.0099
AJR 2005; 185:899-914
© American Roentgen Ray Society
MRI of Articular Cartilage: Revisiting Current Status and Future Directions
Michael P. Recht1,
Douglas W. Goodwin2,
Carl S. Winalski3 and
Lawrence M. White4
1 Cleveland Clinic Foundation, 9500 Euclid Ave., A21, Cleveland, OH 44195.
2 Dartmouth-Hitchcock Medical Center, One Medical Center Dr., Lebanon, NH
03756.
3 Cartilage Repair Center and Department of Radiology, Brigham and Women's
Hospital, 75 Francis St., Boston, MA 02115.
4 Mount Sinai Hospital, 600 University Ave., #563, Toronto, ON M5G 1X5,
Canada.
Received January 19, 2005;
accepted after revision April 6, 2005.
Address correspondence to M. P. Recht
(rechtm{at}ccf.org).
Abstract
OBJECTIVE. The purpose of this article is to review the current
understanding of the MRI appearance of articular cartilage and its
relationship to the microscopic and macroscopic structure of articular
cartilage, the optimal pulse sequences to be used in imaging, the appearance
of both degenerative and traumatic chondral lesions, the appearance of the
most common cartilage repair procedures, and future directions and
developments in cartilage imaging.
CONCLUSION. Articular cartilage plays an essential role in the
function of the diarthrodial joints of the body but is frequently the target
of degeneration or traumatic injury. The recent development of several
surgical procedures that hold the promise of forming repair tissue that is
hyaline or hyalinelike cartilage has increased the need for accurate,
noninvasive assessment of both native articular cartilage and postoperative
repair tissue. MRI is the optimal noninvasive method for assessment of
articular cartilage.
Introduction
MRI of articular cartilage has attracted intense interest and been the
subject of numerous research studies over the past several years. There are
several reasons: the essential role articular cartilage plays in the function
of the diarthrodial joints of the body, the high prevalence of degeneration
and traumatic injury of articular cartilage, and the recent development of new
surgical procedures that hold the promise of forming repair tissue that is
hyaline or hyalinelike cartilage. In 1994, a review article in AJR,
"MR Imaging of Articular Cartilage: Current Status and Future
Directions," summarized early experience with MRI of articular cartilage
by stating that "although the role of MRI in the evaluation of articular
cartilage remains undefined, this imaging method has great promise and may
emerge as an effective technique for detecting even the early stages of
chondral abnormalities"
[1]. Over the past 10 years,
studies have shown that although initial understanding of the MRI appearance
of articular cartilage may have been limited and the pulse sequences used to
evaluate articular cartilage suboptimal, MRI is indeed the optimal diagnostic
method for evaluation of articular cartilage. MDCT, with its superior
resolution, may play a role in the future imaging of articular cartilage, but
that role has yet to be defined. MRI plays a significant role not only in
diagnosis of chondral lesions but also in determination of the appropriate
surgical or pharmacologic treatment and evaluation of such treatment. This
article discusses the current understanding of the MRI appearance of articular
cartilage and its relationship to the microscopic and macroscopic structure of
articular cartilage, the optimal pulse sequences to be used, the appearance of
both degenerative and traumatic chondral lesions, the appearance of the most
common cartilage repair procedures, and future directions and developments of
cartilage imaging.
MRI Appearance of Normal Articular Cartilage
Layers or laminae of varying signal intensity are the characteristic
feature of MR images of normal articular cartilage (Figs.
1A and
1B). When images are acquired
with the articular surface perpendicular to the main magnetic field
(B0), a higher-signal-intensity transitional layer separates the
low-signal-intensity surface from a low-signal-intensity deep layer adjacent
to the subchondral bone. These layers reflect the continuous variation in T2
values across the thickness of the tissue
[2]. T1, diffusion, and proton
density have only minimal influence on tissue contrast
[3-7].
The minimum T2 relaxation time in articular cartilage is
short—approximately 10 msec. As a result, T2 is the determinant of
tissue contrast even on T1-weighted and proton density-weighted images
[2,
5,
8]. On short-TE images acquired
with gradient-echo sequences or projection reconstruction techniques, the
influence of T2 relaxation can be minimized, producing cartilage images with
uniform signal intensity
[9-11].
Truncation effects, however, may cause an appearance similar to the layers
seen on longer-TE images [6,
12]. Magnetization transfer
effects may also alter the laminar appearance on multiecho fast spin-echo
images by decreasing signal intensity
[13].
View larger version (88K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1A —In 65-year-old woman, spin-echo images (TR/TE, 1,000/20) of
femoral condyle fragment imaged at 7 T with articular surface perpendicular to
(A) and parallel with (B) main magnetic field. (Reprinted with
permission from [21])
Higher-signal-intensity transition layer (arrow) separates
lower-signal-intensity radial layer from low-signal-intensity surface.
|
|
View larger version (83K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 1B —In 65-year-old woman, spin-echo images (TR/TE, 1,000/20) of
femoral condyle fragment imaged at 7 T with articular surface perpendicular to
(A) and parallel with (B) main magnetic field. (Reprinted with
permission from [21]) When
imaged after rotation of sample by 90°, pattern of layering changes shows
influence of magic-angle effect. Orientation effect is evident at all levels
of sample, including transitional layer (arrow).
|
|
View larger version (60K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2A —Osteochondral fragment from 56-year-old man. (Reprinted with
permission from [32])
Spin-echo image (TR/TE, 1,000/20) of femoral condyle fragment imaged at 7 T
shows low signal intensity (arrows) in regions where cartilage matrix
is aligned with main magnetic field (B0).
|
|
View larger version (121K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 2B —Osteochondral fragment from 56-year-old man. (Reprinted with
permission from [32]) Sample
photographed after fracture sectioning. In regions where cartilage matrix is
aligned with B0, signal intensity is low (long arrow).
Striations on MR image appear to reflect fibrous-appearing structure revealed
in fractured sample (short arrows).
|
|
Although cartilage appears relatively uniform on gross inspection, the
orientation of collagen and the concentration of water, chondrocytes, and
proteoglycans varies across the thickness of the tissue
[14,
15]. The presence of layers in
cartilage seen on MR images suggests a correlation with histologic zonal
organization. Rubenstein et al.
[16] established a link
between tissue structure and the MR image by demonstrating T2 anisotropy
within cartilage. In their study, the pattern of layering seen on MR images
varied as cartilage orientation relative to B0 changed, proving
that the influence of structural elements within cartilage was responsible for
the presence of layering.
Within highly structured tissues such as cartilage, the restriction of
water mobility and the resulting enhancement of dipole interactions shorten T2
relaxation. In such tissues, however, T2 relaxation is dependent not only on
the distance between nuclei but also on orientation relative to B0.
As the internuclear vector is angled away from B0, T2 lengthens,
reaching a maximum effect at approximately 54.7°
[17-19].
This influence of orientation is commonly referred to as the magic-angle
effect.
Investigations into the influence of orientation on T2 and cartilage layers
led to speculation that regional differences in the orientation of collagen
fibrils could explain the T2 heterogeneity (layers) and anisotropy
(orientation dependence) [2,
16]. Microscopy studies have
shown that collagen fibrils are oriented in parallel arrays perpendicular to
the subchondral bone in the deepest level of cartilage and horizontal to the
surface superficially. A transitional zone of randomly oriented fibrils
separates these two regions or zones
[15]. Such an organization
would explain the short T2 and anisotropy in the deep layer and surface on MR
images of cartilage and the longer T2 and apparent lack of anisotropy in the
transitional layer on MR images.
The assertion that regional variations in collagen fibril orientation could
determine the T2 heterogeneity of cartilage is compelling, yet no direct
correlation between the histologic and MRI findings has established such a
link. Moreover, there appear to be several difficulties with this explanation.
First, the layers seen on MR images are not well defined and instead represent
gradual changes in the bell-shaped T2 profile of cartilage
[2]. Accordingly, the thickness
of layers is determined by the TE: On images acquired with a longer TE, the
width of the higher-signal-intensity transitional layer decreases
[4]. Second, as Grunder et al.
[20] reasoned, the predicted
influence of collagen fibrils on water should be too weak to explain the
strong effects observed on orientation. Finally, anisotropy has been shown in
all layers of cartilage [21]
(Figs. 1A and
1B). The structural elements
that determine T2 must therefore be present across the entire thickness of the
tissue.
An alternative theory explaining the relationship between cartilage
morphology and the MRI appearance of cartilage becomes apparent if one
considers the structure of cartilage at the macroscopic rather than
microscopic level. Collagen fibrils, proteoglycans, and the other constituents
of the extracellular matrix of cartilage are organized into a larger
continuous macroscopic structure that radiates from the subchondral bone and
curves into the plane of the articular surface (Figs.
2A and
2B). Fracture sectioning,
typically used to prepare cartilage for scanning electron microscopy, causes
cartilage to separate along lines of least resistance, thereby reflecting the
tissue organization in the plane of the fracture
[22]. The curved structure
revealed by the fracture is variably referred to in the microscopy literature
as fibers, layers, or leaves
[21-25].
Differences in the description of this macroscopic structure likely reflect
interspecies variation, differences in fracture technique, and unavoidable
artifacts [22].
Given the presence of a continuous structure radiating from the subchondral
bone and arcing into the plane of the joint surface, the magic-angle effect
provides an excellent explanation for T2 heterogeneity and anisotropy in
cartilage [21,
26,
27]. As the macroscopic
structure of cartilage curves away from B0, T2 gradually increases,
reaching a peak in the middle of the transitional layer where tissue alignment
is 55°. As the tissue continues to curve into the plane of the articular
surface, T2 shortens. Fibrous columnlike structures seen on the surface of
sectioned samples of fractures appear to correlate with striations within
low-signal-intensity MRI layers
[26,
27] (Figs.
2A and
2B), suggesting that MRI is
also capable of revealing macroscopic structure at an even more detailed
level.
The apparent lack of anisotropy reported in the transitional layer on a
number of studies is predictable if one considers the complexity of cartilage
structure and the limitations of MRI. When a sample of cartilage is tilted
55°, only portions of the curved cartilage will then be oriented parallel
to B0. Given the relatively thick slices used for MRI, T2
measurements of a curved tissue oriented at an angle will include a mixed
population of tissue orientations and resulting T2 values. The failure to show
anisotropy should therefore not be considered proof that it does not
exist.
Early investigations of cartilage histology suggested that the structure of
cartilage is joint-specific
[28]. Split lines, created in
articular surfaces by puncturing the cartilage with a round pin, are arranged
in patterns that are characteristic for individual joints
[28,
29]. Later investigations
established that the shape of the underlying matrix determines split-line
orientation [23]. Split lines
are, in effect, the most superficial aspect of a fracture plane. Surface
split-line patterns therefore reflect the organization of the entire
extracellular matrix of the cartilage of a particular joint. The
predictability of split-line patterns suggests that variations in matrix
structure are also predictable. Joint-specific and reproducible variations in
the shape of the tibial plateau matrix, suggesting a characteristic structure
or architecture, have been reported
[24].
Because of the strong influence of matrix structure on T2 contrast, the
appearance of articular cartilage on MR images should also be characteristic
or joint-specific. This specificity has been shown at the tibial plateau in a
study documenting characteristic and reproducible variations in T2 and signal
intensity that correlate well with an equally characteristic matrix
architecture [30].
Regional variations in the MRI appearance of cartilage should also be
expected in other joints. Unfortunately, this variability in cartilage
structure complicates efforts to develop standardized methods of monitoring
cartilage injury, degeneration, and repair. In a recent study, for example, a
failure to account for variability in cartilage structure likely explains the
observation of orientation effects on T2 measurements that were substantially
less than expected [31,
32]. On the other hand, the
strong influence of structure on MR image contrast indicates that MRI is
ideally suited to the evaluation of tissue integrity. Although current
clinical imaging sequences cannot reliably show early changes of cartilage
anatomy and composition, new techniques (which are discussed later in this
article) hold promise for being able to detect such changes.
Clinical MRI: Pulse Sequences
Critical factors that affect the MRI evaluation of articular cartilage
include image spatial resolution, image signal-to-noise ratio (SNR), and the
choice of image acquisition protocol. The advent and clinical validation of
pulse sequences optimized for the MRI assessment of articular cartilage have
provided the opportunity to confidently assess the status of articular
cartilage along with other articular structures on routine clinical
imaging.
A number of different clinically available pulse sequences can be used in
MRI of cartilage, with each technique taking advantage of differing contrast
characteristics of cartilage and adjacent tissue and affecting overall image
spatial resolution and image SNR. The most widely used and accurate of these
cartilage-specific sequences include spoiled gradient-recalled echo (SPGR) and
fast spin-echo imaging.
Three-dimensional T1-weighted SPGR acquisitions provide high-resolution
contiguous thin-slice images in shorter scan times than can conventional
spin-echo techniques. Fat-suppressed SPGR images show high contrast between
bright cartilage and relatively dark fluid, bone, fat, and muscle. The adjunct
use of fat suppression with this and other cartilage-imaging pulse sequences
has been advocated as a means of increasing the dynamic range of signal
intensity throughout the image, potentially allowing for better detection of
subtle signal-intensity alterations
[33,
34]. With 3D SPGR imaging,
cartilage abnormalities are seen as morphologic abnormalities of cartilage
contour. Fat-saturated 3D SPGR imaging has been shown to be more accurate than
standard spin-echo MRI in the detection of cartilage defects in the knee, with
sensitivities of as high as 93%
[35-39].
Limitations of the technique include relatively long acquisition times
compared with those of fast spin-echo imaging sequences. Possible techniques
to decrease imaging times include half-Fourier techniques and fat suppression
using spatial-spectral excitation pulses
[40,
41]. An additional potential
disadvantage of gradient-recalled echo imaging is its relative sensitivity to
susceptibility artifacts and intra-voxel dephasing, which may compromise image
quality in the setting of postoperative surgical debris or hardware in
patients who underwent prior arthroscopic surgery or prior cartilage
reparative procedures.
View larger version (102K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3A —MR images of 61-year-old man with knee pain. Fat-saturated 3D
spoiled gradient-recalled echo image (TR/TE, 50/11; flip angle, 30°)
(A) and fast spin-echo T2-weighted image (TR/TE, 5,334/91; echo-train
length, 4) (B) show chondral flap (arrows) involving medial
femoral condyle.
|
|
View larger version (174K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 3B —MR images of 61-year-old man with knee pain. Fat-saturated 3D
spoiled gradient-recalled echo image (TR/TE, 50/11; flip angle, 30°)
(A) and fast spin-echo T2-weighted image (TR/TE, 5,334/91; echo-train
length, 4) (B) show chondral flap (arrows) involving medial
femoral condyle.
|
|
Fast spin-echo imaging techniques allow high-resolution images to be
acquired in a relatively short time. Fast spin-echo imaging of cartilage
benefits, in addition, from inherent magnetization transfer effects within
normal cartilage, potentially increasing the relative conspicuity of cartilage
abnormalities [14,
42,
43]. Both intermediate- and
T2-weighted fast spin-echo imaging sequences, with and without fat
suppression, have been advocated in the assessment of articular cartilage
integrity
[44-48].
With such imaging, cartilage appears intermediate in signal intensity and
joint fluid appears bright in signal intensity. Cartilage abnormalities can be
seen as areas of morphologic contour defects or as regions of relatively
increased intrasubstance-cartilage signal intensity
[49,
50] most likely reflective of
increased intracartilaginous free water and collagenous ultrastructure
disruption [51,
52]. Similar to the results
for investigations of SPGR imaging, fast spin-echo imaging has been shown to
have sensitivities of as high as 94% in the detection of arthroscopically
documented cartilage disease
[46]. Of clinical importance,
fast spin-echo imaging techniques have been shown, in addition, to be valuable
pulse sequences in the diagnostic evaluation of other intraarticular
structures including menisci, ligaments, and subchondral bone
[53-55].
Disadvantages of typical 2D fast spin-echo imaging acquisitions include
limited through-plane resolution relative to 3D SPGR imaging and the potential
for image-blurring artifacts on short-TE-weighted acquisitions as a result of
high-spatial-frequency-encoding echoes late in the fast spin-echo echo train.
Practical problems related to fast spin-echo blurring, however, may be
minimized through appropriate minimization of interecho spacing and the use of
short echo-train lengths.
MR arthrography has been advocated particularly for the evaluation of
cartilage repair procedures because of the ability of contrast material to
outline the surface of and enter defects within repair tissue. Few studies
have been performed on the use of MR arthrography in cartilage repair
procedures, but indirect MR arthrography has been found to better define the
surface of repair tissue and differentiate between intact repair tissue and
delaminated tissue in patients after autologous chondrocyte implantation (ACI)
[56,
57].
Numerous variations of these imaging techniques exist, with imaging
parameters and acquisition sequences varying between centers on the basis of
multiple factors, including user preferences and the potential strengths and
weaknesses of available vendor hardware platforms. In general, practical MRI
protocols for evaluation of hyaline articular cartilage include at least one
cartilage-sensitive sequence obtained in the sagittal plane. Sagittal
acquisitions allow imaging assessment of cartilage in an orientation generally
perpendicular to the majority of weight-bearing cartilage in the knee.
Supplemental acquisitions or reformatting of 3D data sets in the axial and
coronal planes may help to optimize evaluation of patellar cartilage and
central weight-bearing aspects of the femoral condyles and tibial plateaus,
respectively.
View larger version (111K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4A —40-year-old woman with acute knee injury. Sagittal
fat-saturated 3D spoiled gradient-recalled echo image (TR/TE, 22/9; flip
angle, 45°) (A) and transaxial fast spin-echo image (TR/TE,
3,000/87; echo-train length, 5) (B) show chondral fracture
(arrows) with mild displacement of chondral fragment.
|
|
View larger version (93K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 4B —40-year-old woman with acute knee injury. Sagittal
fat-saturated 3D spoiled gradient-recalled echo image (TR/TE, 22/9; flip
angle, 45°) (A) and transaxial fast spin-echo image (TR/TE,
3,000/87; echo-train length, 5) (B) show chondral fracture
(arrows) with mild displacement of chondral fragment.
|
|
Clinical MRI: Cartilage Lesions
Articular cartilage lesions may be categorized as degenerative or traumatic
in cause. Early degenerative disease may be seen on MRI as early alterations
in cartilage contour morphology (fibrillation, surface irregularity) (Figs.
3A and
3B); changes in cartilage
thickness, including cartilage thinning or thickening, which may be an early
feature predating cartilage volume loss; or intrachondral alterations in
signal intensity potentially related to premorphologic intrasubstance collagen
degeneration and increased free-water content. Advanced degenerative chondral
lesions typically manifest on MRI as multiple areas of cartilage thinning of
varying depth and size, usually seen on opposing surfaces of an articulation.
Cartilage defects typically illustrate obtuse margins and may be associated
with corresponding subchondral regions of increased T2-weighted signal
reflective of subchondral edema or cysts or a low signal intensity reflective
of subchondral fibrosis or trabecular sclerosis. Other associated MRI findings
of degenerative cartilage disease include central and marginal articular
osteophytes, joint effusion, and synovitis.
In contrast, traumatic chondral lesions generally manifest on routine
clinical MRI as solitary focal cartilage defects with acutely angled margins
(Figs. 4A and
4B). Traumatic chondral
injuries are typically the result of shearing, rotational, or tangential
impaction forces and often result in high-grade partial- or full-thickness
cartilage tears or in osteochondral injuries of cartilage and the underlying
subchondral bone. Linear cartilage clefts or fissures may also be seen, and
they can extend for variable depths within the articular cartilage and may
result in chondral flap lesions or delamination injuries. Associated
alterations in subchondral marrow signal, including bone bruising, bone edema,
or subchondral fracture, may be helpful signs in delineating areas of
overlying cartilage injury. The finding of a focal signal change in the
subchondral bone marrow should encourage careful evaluation of possible
overlying hyaline articular cartilage injury or disease. In general,
delaminating injuries, superficial flap tears, and surface fibrillation are
the most difficult lesions to visualize and assess accurately with MRI. When a
traumatic cartilage lesion is identified, its description should include the
location, size, and depth of the lesion and the presence or absence of
associated chondral fragments. Traumatic cartilage fragments may remain in
situ, become partially detached, or become loose and displace into the joint
space. As a result, recognition of a traumatic chondral defect should prompt
careful inspection of the joint for a displaced intraarticular chondral
body.
Cartilage Repair Procedures
Articular cartilage itself has limited capability for repair. However,
several surgical procedures have been developed to treat cartilage defects.
The most common of these procedures can be grouped into two major categories:
local stimulation and autologous transplantation of cartilage. Few data exist
about how frequently each of these procedures is performed. Because the
instrumentation for local stimulation is minimal and the procedure can be
performed during arthroscopy, this technique is often performed as the primary
repair. Autologous transplantation generally requires exacting surgical
techniques and usually is performed by surgeons specializing in it.
View larger version (148K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5A —MR images after microfracture in 32-year-old male
professional basketball player. Fat-saturated T2-weighted fast spin-echo
coronal image (TR/TE, 3,250/90; echo-train length, 5) 2 months after
microfracture (A) and fat-saturated T2-weighted fast spin-echo image
(3,000/85; echo-train length, 5) 7 months after microfracture (B) show
that repair tissue (arrows) is greater at 7 months than at 2 months,
with congruent articular surface.
|
|
View larger version (141K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 5B —MR images after microfracture in 32-year-old male
professional basketball player. Fat-saturated T2-weighted fast spin-echo
coronal image (TR/TE, 3,250/90; echo-train length, 5) 2 months after
microfracture (A) and fat-saturated T2-weighted fast spin-echo image
(3,000/85; echo-train length, 5) 7 months after microfracture (B) show
that repair tissue (arrows) is greater at 7 months than at 2 months,
with congruent articular surface.
|
|
Local Stimulation for Cartilage Repair
Several techniques fall into the category of local stimulation, with the
three most commonly performed being abrasion arthroplasty, microfracture, and
subchondral drilling. All local stimulation techniques rely on formation of a
fibrin clot within the defect that is created by bleeding from the penetration
of the subchondral bone underlying the chondral lesion. The fibrin clot
contains pluripotent stem cells that differentiate and remodel, leading to the
formation of repair tissue. The repair tissue has been reported to be
fibrocartilage or a hybrid of fibrocartilage and hyaline cartilage
[58]. Although local
stimulation techniques are generally recommended for lesions smaller than 4
cm2, they have been used to treat defects as large as 10
cm2.
Local stimulation techniques differ in the manner in which the subchondral
bone is violated. In abrasion arthroplasty, a burr is used to penetrate the
bone [59], whereas in
subchondral drilling, a drill is used to perform this function
[60]. With microfracture, a
pick or awl is used to make multiple penetrations into the subchondral bone to
about 4 mm in depth and approximately 3-4 mm apart
[61]. Microfracture has been
advocated over drilling because less heat and necrosis are thought to
occur.
In the first few months after microfracture, MRI shows repair tissue of
intermediate signal intensity that is typically thinner than the adjacent
native articular cartilage
[56]. Over time, the amount of
repair tissue increases, with the optimal result being 100% defect fill with a
congruent articular surface and repair tissue of signal intensity similar to
that of native articular cartilage (Figs.
5A and
5B). It is common to see an
edemalike signal change within the subchondral bone after the procedure,
though this change usually resolves over several months
[56]. Failure and
complications of microfracture are shown as poor fill of repair tissue and
chondral fissures and flaps.
Autologous Transplantation of Cartilage
There are two techniques of autologous transplantation of cartilage:
autologous osteochondral transplantation (AOT) and ACI. Both have attracted a
great deal of interest because of their potential to form hyaline or
hyalinelike repair tissue.
AOT
AOT involves the use of cylindric osteochondral plugs of various sizes to
fill chondral defects. The osteochondral plugs are harvested from minimally
weight-bearing areas of the joint, typically the periphery of the femoral
condyles at the level of the patellofemoral joint. The plugs vary in diameter
from 2.7 to 8.5 mm and in length from 10 to 25 mm
[58,
62]. A number of instruments
have been developed for performing AOT, leading to several names for this
procedure based on the specific instrument used: MosaicPlasty (Acufex, Smith
& Nephew), osteochondral autograft transfer system (OATS, Arthrex), soft
delivery system (SDS, Sulzermedica), and COR system (Mitek)
[63-70].
The chondral defect being repaired is débrided down to viable
subchondral bone, and the osteochondral plugs are transplanted into the defect
site. The orientation, position, and number of osteochondral plugs are
important determinants of the outcome of the procedure. The goal of the
procedure is to create a congruent cartilage surface, and the plugs therefore
need to be placed perpendicular to the surface. If this placement is not
achieved, one edge of the transplant will be raised and one recessed in
relation to the native articular cartilage. Both "proud" and
recessed plugs have worse outcomes than flush plugs, most likely secondary to
abnormal mechanical stress
[70]. By using various sizes
of plugs, one can often increase defect fill to 90% or even 100%
[58]. The number of plugs that
can be used, however, is limited by the availability of donor sites and the
need to limit morbidity at the donor sites. Spaces between plugs fill with a
fibrocartilaginous grout, stimulated by abrasion arthroplasty or sharp
curettage of the base of the defect. There is often an incongruent bone-bone
interface even though there is a congruent cartilage-cartilage interface,
because the plugs typically come from a region of the knee joint where the
articular cartilage is thinner than at the recipient site.
View larger version (168K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 6 —37-year-old man 2 years after autologous osteochondral
transplantation of medial femoral condyle. T2-weighted fast spin-echo coronal
image (TR/TE, 6,192/132; echo-train length, 7) shows relatively congruent
cartilage surface (arrows).
|
|
Indications for autologous autograft transplantation include focal chondral
or osteochondral defects, most commonly of the knee but also of other joint
surfaces such as the talar dome, humeral capitellum, and femoral head, which
have a diameter of 1-4 cm2
[58,
62]. Larger lesions have been
treated, but lesions larger than 8 cm2 are contraindicated for
treatment because of the limits on the amount of donor cartilage that can be
harvested. Contraindications include diffuse cartilage abnormalities such as
those seen in inflammatory arthritis, in advanced osteoarthritis, and after
septic arthritis and an age greater than 50 years
[58]. Histologic evaluation of
autologous osteochondral transplants has shown viable graft hyaline cartilage,
with the interstices between graft plugs filled with fibrocartilagelike repair
tissue [58,
67,
69].
View larger version (149K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 7 —17-year-old boy 3 months after autologous osteochondral
transplantation. Fat-saturated 3D spoiled gradient-recalled echo sagittal
image (TR/TE, 50/11; flip angle, 40°) shows subsidence of osteochondral
plugs (arrows), with resultant incongruent articular surface.
|
|
Theoretic advantages of osteochondral transplantation include the ability
to provide a hyaline cartilage repair tissue, the ability to perform the
technique with one procedure, the possibility of bone-to-bone healing of the
grafts, and the ability to add bone to an osteochondral lesion such as
osteochondritis dissecans. Potential disadvantages include the need to use a
portion of the articular surface, though not a major weight-bearing surface,
as a donor site; the creation of an irregular bone-cartilage interface and
possible resultant irregular "tidemark;" and the technical
difficulty of the procedure, especially in the accurate placement of the
plugs.
MRI after AOT is accurate in evaluating several features of the procedure,
including graft positioning, surface congruity, graft incorporation, and donor
site morbidity. After the procedure, there should be a congruent
cartilage-cartilage interface (Fig.
6). MR images provide excellent visualization of the repair tissue
surface and its relationship to adjacent native articular cartilage. When
surface irregularity or incongruity is present, MRI can show the reasons for
such problems, such as improper positioning of the graft, graft subsidence
(Fig. 7), or gross graft
motion such as graft displacement or rotation. Surface incongruities, when
present, may be seen to decrease over time on serial MRI studies, presumably
because of filling of the defects by the fibrocartilaginous grout
[62].
The signal characteristics of the cartilage cap on the osteochondral plugs
typically follow those of adjacent native articular cartilage. The
fibrocartilaginous grout in the spaces between plugs may show increased T2
signal, compared with that of native articular cartilage, on fast spin-echo
images and decreased signal intensity, compared with that of adjacent native
articular cartilage, on 3D fat-suppressed SPGR sequences.
Animal studies after AOT have shown that graft revascularization begins as
early as 6-14 weeks after the procedure
[71,
72]. Signal-intensity changes
seen on MRI are believed to parallel this process, with the grafts showing a
normal, fatlike marrow signal at 2 weeks. By 4-6 weeks, however, an edemalike
signal change is seen within the marrow of the graft and in the surrounding
bone marrow, with intense enhancement of the graft after IV contrast
administration [73]. The
perigraft abnormal signal is believed to represent fibrovascular reparative
and inflammatory reactive tissue. Between 6 and 9 months, the graft returns to
fatty marrow and enhancement decreases; these changes are thought to be
secondary to incorporation of the graft. Edemalike signal changes in the
perigraft tissues also tend to decrease by 1 year after the procedure, though
persistence of some edemalike signal change up to 2 years after the procedure
has been shown in asymptomatic grafts that appear solidly incorporated on
arthroscopy [62]. However, a
large, persistent perigraft edemalike signal change or formation of cystlike
regions are worrisome indicators of poor graft incorporation.
Although the osteochondral plugs are taken from a relatively
non-weight-bearing area of the joint, donor site morbidity has been reported
in approximately 3% of patients
[58]. After harvesting of the
osteochondral plugs, the donor site may be left empty or may be filled with
material from the recipient site. Over time, the donor site fills with
cancellous bone and fibrocartilagelike material
[62]. In the early
postoperative period, the donor sites show low T1 and increased T2 signal
intensity, compared with the signal intensity of adjacent fatty marrow, with a
defect in the overlying articular cartilage and edemalike signal change in the
adjacent bone marrow. Approximately 6-9 months after the procedure, the donor
sites return to a fatty, marrowlike signal intensity and the overlying
articular cartilage defect fills with fibrocartilagelike repair tissue.
ACI
ACI is a cell-based surgical treatment for deep articular cartilage defects
[74]. It has been recommended
for defects from 2 to 12 cm2 and is performed in two stages. The
first stage is an arthroscopic assessment of the cartilage defect and the
harvesting of a small amount of articular cartilage from a relatively
non-weight-bearing site within the knee, usually the intercondylar notch or
the medial margin of the trochlea
[74,
75]. The cells within this
cartilage biopsy specimen are removed from the extracellular matrix by
enzymatic digestion and then cultured for several weeks until approximately 12
million cells are available for implantation
[76]. In the second stage, an
open arthrotomy, the repair site is prepared by débridement of the
defect by removal of the calcified cartilage layer and any loose cartilage
fragments from the margins of the defect. Penetration of the subchondral bone
plate is avoided. Periosteum of the same dimensions as the defect is harvested
from the tibia or femur. The periosteum is sewn over the prepared defect with
the cambium layer facing the bone. Fibrin glue is used to seal the margins of
the defect. The cultured cells are then injected beneath the periosteal
cover.
The repair site matures through several stages
[76]. During the first 6
weeks, or the proliferative phase, the implanted cells multiply and fill the
defect with a very soft tissue. Between weeks 7 and 26, or the transition
phase, the extracellular matrix expands and stiffens as collagen and
proteoglycans are produced. In the final, or remodeling, phase, the
extracellular matrix further matures until the stiffness of the repair tissue
becomes similar to that of the adjacent hyaline articular cartilage.
Histologic examination of biopsy samples obtained from repair sites showed a
hyalinelike repair in 75-80% of cases at a mean follow-up of 4.5 years after
ACI [77,
78].
The normal appearance of the repair site and of the underlying bone after
ACI changes as the repair tissue matures
[56,
57,
79]. During the first month,
the tissue within the repair site appears of intermediate signal intensity on
unenhanced T1-weighted and proton density-weighted images and of bright signal
intensity on T2-weighted and other fluid-sensitive sequences—only
slightly different from joint fluid. Images from IV (indirect) arthrograms
usually show bright enhancement of the immature repair tissue. During the
first several weeks, the periosteal cover may be identified as a separate
layer on the surface of the repair site that may extend above the expected
level of the articular surface. Over time, the signal intensity of the repair
tissue becomes less like fluid and more like native hyaline articular
cartilage. After about 12-18 months, the signal intensity of the repair tissue
stabilizes. The signal intensity of the mature repair tissue varies and may be
quite heterogeneous or only subtly different from adjacent nonoperated
articular cartilage (Figs. 8A,
8B,
8C,
8D, and
8E). Most often, the
appearance of the repair tissue is different from that of native articular
cartilage [80]. At this time,
the implications of the different signal patterns in the repair tissues are
not known.
View larger version (146K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8A —52-year-old man with failed abrasion arthroplasty of medial
trochlea who was treated by autologous chondrocyte implantation (ACI). Six
years after surgery, he was asymptomatic and underwent imaging with IV
(indirect) MRI arthrography. Photograph obtained near end of surgery shows
anterior margin of intercondylar notch (arrow) and 25-mm-long x
22-mm-wide ACI site (arrowheads) covering nearly entire medial
trochlear facet.
|
|
View larger version (207K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8B —52-year-old man with failed abrasion arthroplasty of medial
trochlea who was treated by autologous chondrocyte implantation (ACI). Six
years after surgery, he was asymptomatic and underwent imaging with IV
(indirect) MRI arthrography. Sagittal proton density-weighted fast spin-echo
image (TR/TE, 2,400/37; echo-train length, 8) of knee shows complete fill of
trochlear ACI site (arrowheads) by repair tissue, which appears
slightly darker than native articular cartilage. Levels of articular surface
and subchondral bone plate are slightly above those of adjacent, nonoperated
regions.
|
|
View larger version (197K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8C —52-year-old man with failed abrasion arthroplasty of medial
trochlea who was treated by autologous chondrocyte implantation (ACI). Six
years after surgery, he was asymptomatic and underwent imaging with IV
(indirect) MRI arthrography. Sagittal proton density-weighted fast spin-echo
fat-saturated image (2,900/25; echo-train length, 8) of knee shows normal
signal in bone marrow subjacent to ACI site (arrowheads).
|
|
View larger version (186K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8D —52-year-old man with failed abrasion arthroplasty of medial
trochlea who was treated by autologous chondrocyte implantation (ACI). Six
years after surgery, he was asymptomatic and underwent imaging with IV
(indirect) MRI arthrography. Transaxial proton density-weighted fast spin-echo
image (3,625/30; echo-train length, 12) of knee shows ACI site
(arrowheads) on medial trochlear facet filled with
low-signal-intensity repair tissue.
|
|
View larger version (193K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 8E —52-year-old man with failed abrasion arthroplasty of medial
trochlea who was treated by autologous chondrocyte implantation (ACI). Six
years after surgery, he was asymptomatic and underwent imaging with IV
(indirect) MRI arthrography. Oblique coronal T1-weighted spin-echo
fat-saturated image (650/12) of knee obtained in plane orthogonal to trochlear
ACI site shows area of native articular cartilage thinning (arrow)
medial to ACI repair site (arrowheads). Region of native cartilage
thinning is poorly shown on transaxial image (D) because of
partial-volume artifact.
|
|
The repair tissue should restore the contour of the articular surface and
fill the cartilage defect to the same level as that of the adjacent articular
cartilage, regardless of the depth of the original defect. At the margins of
the defect, the repair tissue usually integrates with the adjacent cartilage.
This ACI-cartilage interface usually appears as a dark band, or it may be
indiscernible. However, during the first few postoperative months, portions of
the normal ACI-cartilage interface may appear similar to fluid and simulate a
fissure. The fluidlike line in a healthy, immature interface is oriented
orthogonally to the articular surface and, unlike a cartilage flap, does not
extend between the bone and repair tissue. With time, the normal fluidlike
interface matures to become indiscernible or dark, whereas a true fissure
persists, with formation of a cartilage flap or subchondral cyst.
The subchondral bone plate beneath the repair site usually appears slightly
irregular in contour and remains unchanged over time. The bone marrow usually
shows intense edemalike signal deep beneath the repair site, often to the
level of the physeal scar, during the initial postoperative period. The marrow
signal intensity begins to return to normal over the next few months and
usually appears nearly normal by 1 year. A thin line of mild edemalike signal
may normally remain beneath the subchondral bone plate subjacent to the ACI
site indefinitely.
Postoperative complications after ACI are related either to the open
arthrotomy or specifically to the ACI graft. Although occurrences in either of
these categories may require arthroscopic surgery, they usually they do not
indicate treatment failure
[75,
77,
78]. Most ACI-related
complications are associated with the periosteal cover. Failure of the ACI
graft through degeneration, fissuring, or separation of the repair tissue from
the bone (delamination) is much less common
[75,
77,
78].
The formation of intraarticular adhesions causing knee stiffness is the
most common arthrotomy-related complication, occurring in about 5% of patients
treated with ACI [75].
Adhesions usually present early, in the first several postoperative months.
The diagnosis is most often made without the need for imaging, but the signs
and symptoms may be nonspecific. In such cases, MRI may be useful in excluding
ACI graft failure and in identifying the thickened fibrous bands of the
adhesions. On MR images, postoperative adhesions appear as thickening of the
joint capsule or as focal bands of tissue within the infrapatellar fat pad.
The signal intensity of the adhesions is usually lower than that of fat on
T1-weighted and proton density-weighted images but brighter than that of fat
on proton density-weighted images with fat saturation. These bands often
extend over the articular cartilage or the ACI repair-tissue surface.
Periosteal complications include fibrous overgrowth of the periosteal cover
(periosteal hypertrophy) and separation and detachment of a hypertrophic
periosteal cover from the underlying hyaline repair (periosteal delamination).
Periosteal hypertrophy is commonly seen at histologic examination and
arthroscopy [75,
77] and may be symptomatic in
up to 20% of patients [75]. On
MRI, periosteal hypertrophy appears as thickening of the repair tissue, with
protrusion of the tissue above the expected level of the articular contour
[57] (Figs.
9A,
9B, and
9C). The hypertrophic tissue
may extend over the surface of the adjacent articular cartilage (overlapping
type) like a pannus. Or, when the ACI site abuts the intercondylar notch, the
hypertrophic tissue may grow into the notch and interfere with the anterior
cruciate ligament, similar to the focal fibrosis or "cyclops"
lesion seen after anterior cruciate ligament reconstruction
[57,
76]. The treatment of
symptomatic periosteal hypertrophy is usually by arthroscopic chondroplasty
with removal of the overgrown fibrous tissue
[75].
View larger version (166K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 9A —22-year-old woman with catching sensation in knee 6 months
after autologous chondrocyte implantation (ACI) surgery for 24-mm-long x
19-mm-wide osteochondral defect of medial femoral condyle. Sagittal proton
density-weighted fast spin-echo image (TR/TE, 2,900/38; echo-train length, 8)
of knee from IV (indirect) MR arthrogram shows prominent periosteal
hypertrophy. Surface of ACI site (arrowheads) is above level of
native articular cartilage, best seen at junction between ACI and native
cartilage (arrow).
|
|
View larger version (163K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 9B —22-year-old woman with catching sensation in knee 6 months
after autologous chondrocyte implantation (ACI) surgery for 24-mm-long x
19-mm-wide osteochondral defect of medial femoral condyle. Sagittal proton
density-weighted fast spin-echo fat-saturated image (2,900/25; echo-train
length, 8) of knee from same examination as A shows mild edemalike
signal in marrow beneath ACI site (arrowheads).
|
|
View larger version (144K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 9C —22-year-old woman with catching sensation in knee 6 months
after autologous chondrocyte implantation (ACI) surgery for 24-mm-long x
19-mm-wide osteochondral defect of medial femoral condyle. Image of ACI site
from knee arthroscopic surgery performed 19 days after A shows
prominent mound of fibrous, periosteal overgrowth (arrow) and
junction between ACI and native articular cartilage (arrowheads).
Fibrous periosteal tissue was débrided, revealing firm, intact repair
tissue underneath.
|
|
Periosteal delamination may result in symptoms of catching or locking of
the knee. Arthroscopic removal of the periosteal cover is usually all that is
required, and most often the underlying repair tissue remains intact. The
detached fibrous tissue is usually hypertrophic and lying over the repair site
and adjacent articular cartilage and most often appears as a bulge in the
articular contour on MR images.
Failure of the ACI may occur from poor bone integration, poor repair-tissue
quality (soft graft), graft detachment (delamination), or degeneration of the
repair tissue [75]. Clinical
failures after ACI have been reported in 6-13% of patients
[75,
77,
81] Degeneration of the ACI
tissue most often presents as pain, years after surgery. On MRI, degeneration
of the ACI repair site appears similar to damage of native articular
cartilage. Focal full- or partial-thickness defects, fissures, and tissue
flaps may be present. The underlying bone often shows intense edemalike marrow
signal or cyst formation
[82].
Delamination of the ACI usually occurs in the first 6 months of the
postoperative course [76].
Delamination may involve all of the graft or only a portion of the graft. When
a partial delamination occurs, it usually involves the edge of the ACI where
the repair tissue meets the adjacent native articular cartilage
[76]. Symptoms are most
commonly the sudden onset of painful catching or knee locking
[76]. The delaminated tissue
may become displaced, leaving an empty defect, or remain in situ within the
repair site, producing a tissue flap. The MRI appearances of delamination
reflect the location of the separated repair tissue
[57]. When the delamination is
displaced, the ACI site is empty and fluid-filled. The displaced repair tissue
is often found as an intraarticular loose body elsewhere in the joint. On MRI,
delamination in situ appears similar to a cartilage flap. A thin, fluidlike
signal line is usually seen at the base of the ACI site between the repair
tissue and the subjacent subchondral bone plate. Most often, this abnormal
linear signal can be seen to connect with the joint space on one or more
images. In the first few weeks after surgery, and even in the first 3-4
months, it can be difficult to differentiate the fluidlike signal of the
immature ACI repair tissue from joint fluid at the base of the graft from a
delamination. MRI arthrography, either direct or indirect, can help because
the repair tissue is usually darker than fluid on these acquisitions. However,
in some instances, the repair tissue and joint fluid may appear similar even
with MRI arthrography.
Assessment of the bone marrow beneath the ACI is an important part of the
MRI evaluation. As with other cartilage repair procedures, an edemalike marrow
signal subjacent to the repair tissue is a normal finding in the early
postoperative period but persistence or reappearance of abnormal marrow signal
may indicate problems with the repair
[57,
62,
82]. The precise time frame
for normalization of the marrow signal after ACI has not been established.
However, persistence of abnormal marrow signal beyond the first postoperative
year is worrisome and may indicate postoperative complications such as
periosteal hypertrophy or poor integration of the repair tissue. Subchondral
cysts may develop beneath ACI sites, with poor integration of the repair
tissue to adjacent cartilage or with degeneration of the repair tissue
[82].
View larger version (96K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 10A —Transaxial MR images of patellofemoral joint of patient with
knee pain. (Reprinted with permission from
[90]) Driven equilibrium
Fourier transform (DEFT) image with fat saturation (TR/TE, 400/15) clearly
shows deep cartilage fissure with surface irregularity (arrow).
|
|
View larger version (57K):
[in this window]
[in a new window]
[as a PowerPoint slide]
|
Fig. 10B —Transaxial MR images of patellofemoral joint of patient with
knee pain. (Reprinted with permission from
[90]) On spoiled
gradient-recalled echo fat-saturated image (50/15; flip angle, 30°)
obtained at same location, abnormality is more difficult to visualize
(arrow).
|
|
New Imaging Techniques
Recent medical and surgical advances in the treatment of osteoarthritis and
other cartilage abnormalities require improved noninvasive methods to evaluate
articular cartilage. Much research is under way to increase the potential of
MRI as a noninvasive method of cartilage analysis, with efforts focused on
both qualitative and quantitative analyses. Technologic advances in high-field
imaging, pulse sequence development, and the application of MRI contrast
agents are the foundation of this work.
Imaging articular cartilage is a challenge. It is thin, with a maximum
thickness of about 4 mm, and usually has curved surfaces. These factors lead
to partial-volume averaging effects, which may reduce the sensitivity of MRI
for measurements of cartilage thickness and volume and for the detection of
thin fissures, cartilage flaps, and shallow defects. To obtain high-quality MR
images, one must maintain a balance between high spatial resolution and
adequate SNR. Any increase in the SNR that can be obtained affords the
potential to increase the spatial resolution and thus to improve MR image
sensitivity for small abnormalities and early disease in articular cartilage.
For example, a new MRI system that provides a twofold improvement in SNR
should be able to provide a twofold increase in spatial resolution at the same
SNR as the old system.
Because SNR is linearly proportional to magnetic field strength, imaging at
higher magnetic field strengths should allow greater spatial resolution with
equal imaging time and SNR
[83]. If all other factors
were to remain equal, a change from a 1.5-T system to a 3-T system, for
instance, should provide a twofold increase in SNR at the same spatial
resolution and imaging time, a twofold increase in spatial resolution at the
same SNR and imaging time, or a fourfold reduction in imaging time at the same
SNR and spatial resolution. However, at higher field strengths, the T1
relaxation times increase, particularly for cartilage. Also, the increase in
frequency difference between marrow fat and cartilage water at 3 T versus
lower field strength reduces the benefits obtained from an increase in magnet
field strength for non-fat-suppressed methods. Longer T1 relaxation times
require longer-pulse TRs, resulting in an increase in image acquisition time
or a change in the flip angle of the excitation pulse to maintain the same
image contrast at higher field strength
[83]. In addition, for
non-fat-suppressed images, the greater separation of fat and water frequencies
will lead to greater chemical shift artifacts at the cartilage-bone interface
because of marrow fat. Although increasing the bandwidth of the image
acquisition may reduce this artifact, the SNR of the acquisition decreases.
Despite these challenges, high-field imaging shows great promise as an overall
improvement in articular cartilage assessment by MRI. Currently, high-field
MRI systems are not as widely available and do not have the same range of
radiofrequency coils as are produced for 1.5-T systems. As high-field MRI
availability and equipment improve, they will likely become the preferred
systems for articular cartilage imaging.
New pulse sequences and image acquisition methods under development for
articular cartilage have focused on image contrast and spatial resolution that
would improve both quantitative cartilage analyses and the clinical diagnosis
of articular cartilage abnormalities
[84]. A major emphasis of
articular cartilage research has been the measurement of articular cartilage
thickness and volume as biomarkers for disease progression or treatment
response in osteoarthritis
[85-88].
To perform these analyses, it is desirable to use computer algorithms to
identify, or "segment," articular cartilage tissue in an automated
manner. This goal requires high image contrast between cartilage and the
surrounding tissues, including bone, bone marrow, joint fluid, menisci, joint
capsule, ligaments, and intraarticular fat. To date, no one acquisition
sequence has proven ideal, although the majority of publications on volume and
thickness measurements have used the 3D T1-weighted SPGR sequence
[84,
86]. Although this sequence
shows good cartilage-water and cartilage-fat image contrast, the
cartilage-meniscus and cartilage-capsule image contrast has been relatively
poor [89]. A number of other
acquisition techniques have been proposed to improve articular cartilage image
contrast. These methods include steady-state free precession (fast imaging
employing steady-state acquisition [FIESTA], true fast imaging with
steady-state free precession [true-FISP], balanced fast field echo) and its
variant, fluctuating equilibrium MRI (FEMR); multi-echo techniques such as
dual excitation in the steady state (DESS); driven equilibrium techniques such
as driven equilibrium Fourier transform (DEFT) and fast recovery fast spin
echo; echo-planar techniques such as 3D echo-planar imaging with fat
suppression and 3D DEFT; and 3D fast spin-echo methods
[90-96]
(Figs. 10A and
10B). Automated and
semiautomated image-processing approaches that combine two image
acquisitions—for example, subtraction of image acquisitions with and
without binomial pulse saturation transfer—have shown the ability to
isolate articular cartilage automatically for volume and thickness
me
Top
Abstract
Introduction
