Keywords
cartilage - MRI - quantitative - semiquantitative
Osteoarthritis (OA) is the most common joint disease in the United States, and its
prevalence is rising due to an aging population and increasing rates of obesity.[1]
[2] Irreversible and progressive degradation of the articular cartilage remains the
fundamental feature of OA pathophysiology. Conventional radiography is considered
the reference standard for imaging of OA; however, joint space narrowing (JSN) on
radiography provides an indirect measure of cartilage loss and is not sensitive to
progression of the disease.[3]
[4]
[5] Radiography-based JSN is nonetheless commonly used as the imaging outcome measure
to establish the effectiveness of disease-modifying osteoarthritis drugs (DMOADs).[6] The use of radiographic JSN as an outcome measure may partly explain why attempts
at developing DMOADs and behavioral therapy for OA have proven unsuccessful, despite
promising preclinical research.[7]
Regulatory agencies including the U.S. Food and Drug Administration (FDA) are increasingly
recommending imaging beyond radiography to assess early onset of abnormalities in
OA.[8] The multiplanar, multiparametric capabilities of magnetic resonance imaging (MRI)
and its excellent soft tissue contrast allow unparalleled evaluation of all joint
structures including cartilage. Hence MRI-based outcome measures are ideal for assessment
of cartilage degradation in OA. In addition to the routine clinical MRI, advanced
research techniques have been developed to assess the biochemical composition of cartilage
in the earliest stages of OA. These include relaxometry measurements (T2, T2*, and
T1ρ mapping), sodium imaging, delayed gadolinium-enhanced MRI of cartilage (dGEMRIC),
glycosaminoglycan specific chemical exchange saturation transfer (gagCEST), and diffusion
tensor imaging (DTI). These compositional MRI techniques serve as quantitative, reproducible,
and objective end points for OA research that will likely be introduced to clinical
radiology practice in the near future.
In this narrative review, we describe the standard clinical imaging of cartilage in
OA. We discuss semiquantitative scoring systems for the assessment of cartilage that
serve as important outcome measures in research. Lastly, we highlight advanced compositional
MRI techniques that allow the detection of early articular cartilage degradation.
Cartilage Microarchitecture
Cartilage Microarchitecture
Understanding the imaging of chondral degeneration in OA requires an understanding
of the basic ultrastructure of the articular cartilage. It is composed of primarily
fluid (70–80%) and extracellular matrix (ECM), both of which are essential for its
normal function.[9] The ECM is a network of collagen fibrils and proteoglycan molecules, with the proteoglycan
consisting of negatively charged glycosaminoglycan (GAGs) attached to the protein
core.[9] The negative charge attracts and holds water within articular cartilage while cations
such as sodium (Na+) counter the negative charge of GAGs. The redistribution of water within the ECM
provides the known biomechanical properties of cartilage, that is, its ability to
deal with tensile and compressive loads.[10] In OA, proteoglycan loss and disorganization and/or loss of the collagen fiber network
lead to impaired ability of the articular cartilage to deal with these loads and results
in progressive, irreversible breakdown.[11]
Clinical MRI of Cartilage
Clinical MRI of Cartilage
An in-plane resolution of 0.3 mm resolves the earliest stage of morphological cartilage
degeneration, that is, fraying of the articular surface.[12] Optimal evaluation of cartilage morphology on standard clinical MRI, therefore,
requires high signal-to-noise ratio (SNR) and a high spatial resolution, both of which
are advantages of higher field strength magnets (≥ 1.5 T, with dedicated extremity
coils). A recent systemic review and meta-analysis comparing 1.5-T and 3-T MRI for
detection of morphological cartilage lesions found both field strength magnets to
offer high diagnostic accuracy; however, the 3-T MRI had greater accuracy than the
1.5 T.[13] In 2017, the FDA approved the first 7-T MRI system for clinical diagnostic imaging
of the extremities. A comparison of routine clinical knee MRI performed at 3 T and
7 T found diagnostic confidence of radiologists for cartilage defects to be higher
with 7 T.[14]
In addition to adequate SNR and spatial resolution, detection of cartilage pathology
requires optimal cartilage–synovial fluid contrast. The International Cartilage Repair
Society (ICRS) protocol for imaging of cartilage includes two-dimensional (2D) fast
spin-echo (FSE) or turbo spin-echo (TSE) pulse sequences to obtain fat-suppressed
proton-density-weighted, T2-weighted, or intermediate-weighted images.[15] These sequences provide excellent tissue contrast allowing detection of cartilage
lesions with high accuracy; however, they require acquisition in multiple planes.
[Fig. 1] shows fat-suppressed proton-density images of the patellofemoral compartment cartilage
in a healthy volunteer ([Fig. 1a]) and in a patient with advanced osteoarthritis ([Fig. 1b]). Isotropic sequences (3D FSE or TSE) obviate the need for multiplanar acquisition,
greatly reducing acquisition time.[16] Isotropic imaging suffers from blurring and lower tissue contrast compared with
2D sequences; however, the diagnostic accuracy of isotropic FSE for cartilage morphology
was shown to be similar to 2D FSE at 3 T.[17]
Fig. 1 Sagittal fat-suppressed proton-density images of the patellofemoral compartment cartilage
in (a) a healthy volunteer and (b) a patient with advanced osteoarthritis (OA). The healthy volunteer has normal patellar
and femoral trochlear cartilage thickness and signal density. The patient with OA
demonstrates full-thickness cartilage loss overlying the patella (yellow arrow) and
overlying the femoral trochlea (red arrow), with associated subchondral bone marrow
edema.
Cartilage-sensitive techniques based on gradient spin-echo (GRE) such as 3D spoiled
gradient recalled echo produce images with cartilage signal more intense than the
surrounding tissues, which renders them insensitive to subtle cartilage lesions and
of limited utility in clinical imaging.[18] These techniques were used successfully for quantitative assessment of cartilage
thickness and volume in research studies.[19]
[20]
[21]
Semiquantitative Assessment of Cartilage
Semiquantitative Assessment of Cartilage
Semiquantitative scoring systems for cartilage including the Outerbridge (1961)[22] and Noyes and Stabler (1989)[23] classifications were originally developed for grading the appearance and quantity
of cartilage via direct evaluation during surgery. The ICRS classification, a 9-point
scale, succeeded these initial scoring systems[24] and provided a more comprehensive and detailed assessment of articular cartilage
pathology. All of these classifications have been adapted for assessment of joint
cartilage on MRI,[25]
[26]
[27] primarily for research.
As the use of MRI for OA was researched and increased, dedicated MRI-based semiquantitative
classifications for whole-organ assessment of joints, most commonly the knee joint,
were developed and validated. In the next section, we briefly describe a select few
of these systems that are most likely to be encountered in the published literature.
Whole-Organ Magnetic Resonance Imaging Score
Whole-Organ Magnetic Resonance Imaging Score
The Osteoarthritis Initiative (OAI) and the Multicenter Osteoarthritis Study (MOST)
are two of the largest multicenter and longitudinal studies of OA that included MRI
acquisition in addition to expansive clinical data for the study of OA.[28]
[29]
[30]
[31]
[32]
[33] The Whole-Organ Magnetic Resonance Imaging Score (WORMS) is the most commonly used
scoring system in knee OA research and was used as an outcome measure in the OAI and
MOST. The WORMS assesses 14 features in the knee joint that include articular cartilage
integrity, subchondral bone marrow abnormality, cruciate ligament, and meniscal integrity
among other features.[34] Cartilage is graded on an 8-point scale in 14 subregions subdivided by anatomical
landmarks. Cartilage grades in the different subregions are frequently summed to provide
composite or global cartilage scores ([Fig. 2]).
Fig. 2 Whole-organ magnetic resonance imaging score (WORMS) descriptions: 0 = normal thickness
and signal; 1 = normal thickness but increased signal on T2-weighted images (not used
in this study); 2.0 = partial-thickness focal defect < 1 cm in greatest width; 2.5 = full-thickness
focal defect < 1 cm in greatest width; 3 = multiple areas of partial-thickness defects < 75%
of region or a single partial-thickness defect wider than 1 cm but < 75% of the region;
4 = diffuse (> 75% of the region) partial-thickness loss; 5 = multiple areas of full-thickness
loss < 75% of the region or a single full-thickness lesion wider than 1 cm but < 75%
of the region; 6 = diffuse (> 75% of the region) full-thickness loss. Boston-Leeds
Osteoarthritis Knee Score (BLOKS) descriptions: Size of any cartilage loss (including
partial- and full-thickness loss) as a percentage of surface area as related to the
size of each individual region: 0: none; 1: < 10% of region of cartilage surface area;
2: 10 to 75% of region of cartilage surface area; 3: > 75% of region of cartilage
surface area; and percentage full-thickness cartilage loss of the region: 0: none; 1: < 10% of region
of cartilage surface area; 2: 10 to 75% of region of cartilage surface area; 3: > 75%
of region of cartilage surface area. Reproduced with permission from Lynch et al.[36]
Boston-Leeds Osteoarthritis Knee Score
Boston-Leeds Osteoarthritis Knee Score
Concerns about the responsiveness of WORMS and the validity of summation of subregional
WORMS cartilage measurements led to the development of the Boston-Leeds Osteoarthritis
Knee Score (BLOKS).[35] The BLOKS evaluates cartilage in nine subregions of the knee. BLOKS I cartilage
score, the more commonly used of the two-part cartilage scoring component of BLOKS,
assesses cartilage on a 4-point scale. It assigns separate scores for (1) the areal
extent of any cartilage loss in each subregion, and (2) the percentage of subregion
surface area that has a full-thickness loss ([Fig. 2]).
MRI Osteoarthritis Knee Score
MRI Osteoarthritis Knee Score
Both WORMS and BLOKS have limitations, highlighted in a two-part study comparing these
methods.[36]
[37] As a result, the MRI Osteoarthritis Knee Score (MOAKS) was derived from both the
BLOKS and WORMS to improve whole-organ assessment of the knee.[38] MOAKS grades cartilage in the same 14 subregions of the knee as are graded in WORMS,
but it uses the grading scale used in the BLOKS “cartilage I” score ([Fig. 1]).
Knee Osteoarthritis Scoring System
Knee Osteoarthritis Scoring System
The Knee Osteoarthritis Scoring System[39] is another whole-organ grading system focused on the knee that grades cartilage
in nine subregions. It assigns separate 4-point scores for the depth of the cartilage
and osseous components of an osteochondral defect. It also assigns a separate 4-point
score for the surface extent of an osteochondral defect estimated by its maximal diameter.
A focal cartilaginous defect is well defined with an acute angle between the defect
and surrounding cartilage. A diffuse defect has an obtuse angle between the normal
and thinned cartilage.
Use of Semiquantitative Assessment Methods in Research
Use of Semiquantitative Assessment Methods in Research
These classification schemes have been used extensively as outcome measures in research
including in large multicenter trials such as the OAI and MOST. The following are
select examples of studies of OA risk factors using semiquantitative assessment.
Osteoarthritis and Physical Activity
Osteoarthritis and Physical Activity
The impact of physical activity on OA remains a controversial topic. Although some
studies found exercise to be beneficial,[40]
[41] studies of patients from the OAI contradicted these findings. In particular, these
studies of OAI patients reported that individuals who have risk factors for OA may
suffer cartilage degeneration with high-intensity physical activity.[29]
[42]
[43] Even in asymptomatic individuals enrolled in the OAI, cartilage lesions were more
common and more severe in the highly active subjects compared with the less active
subjects.[29]
[44] In patients who have knee abnormalities at baseline, walking > 10,000 steps per
day was associated with higher cartilage defect scores.[45] Physical activity involving frequent knee bending was also implicated in a higher
prevalence of knee cartilage lesions and increased the progression of these lesions,
particularly in the patellofemoral compartment.[33]
Obesity
In obese adults, knee cartilage defects are associated with physical disability.[46] In the OAI cohort, obesity was associated with a higher prevalence and severity
of knee cartilage lesions as well as with increased cartilage lesion progression over
3 years.[32]
[47] High body mass index was also associated with rapid tibiofemoral cartilage loss
in patients enrolled in MOST who had or were at risk for OA.[48] Weight loss may help prevent development/progression of lesions and improve quality
of life.[49]
Injuries
In the OAI cohort, individuals with anterior cruciate ligament (ACL) abnormalities
had a greater prevalence of cartilage lesions that were also more severe compared
with individuals with a normal ACL.[30] Meniscal tears were also found to be associated with poor tibiofemoral cartilage
scores, even in patients without OA.[50] Meniscal root tears are particularly implicated.[51]
[52] The presence of meniscal extrusion is also associated with the prevalence and severity
of cartilage damage.[53]
[54]
Alignment
Knee malalignment in either the valgus or varus direction affects the distribution
of the load across the joint.[55] A large study examined 5,053 knees from the MOST and 5,953 knees from the OAI cohort
using either WORMS or BLOKS.[56] This study found valgus malalignment, particularly > 3 degrees, to be associated
with an increased risk of cartilage defect progression in the lateral tibiofemoral
compartment. In the MOST cohort, varus malalignment has, in contrast, been associated
with incident cartilage damage in the medial compartment.[57]
Compositional MRI Techniques for Assessment of Cartilage
Compositional MRI Techniques for Assessment of Cartilage
Primarily used in research, these techniques allow detection of the earliest changes
of cartilage degeneration in the ECM, well before the morphological cartilage defects
are apparent. Thus, compositional imaging sequences have the potential to serve as
quantitative imaging biomarkers of OA.
Higher field strength magnets (3 T and 7 T) are particularly useful for compositional
imaging, even more so than for clinical imaging of cartilage. These MR units afford
higher SNR with resultant greater spatial resolution and shorter MRI acquisition times.[58] Additionally, many biologically relevant nuclei in addition to 1H, such as sodium and phosphorous, occur in relatively low concentrations and warrant
high field strength magnets to allow signal acquisition. Among numerous challenges,
impediments to routine use of higher field strength MRI include increasing inhomogeneity,
changes in relaxation times,[59] increased sensitivity to susceptibility effects (decreased T2*),
[60] and increase in chemical shift artifact in the frequency-encode direction.[61]
T2 Mapping and T2* Mapping
T2 Mapping and T2* Mapping
T2 mapping was obtained as part of the knee MRI acquisition protocol in the OAI and
was the most widely studied of all compositional imaging techniques.[62] T2 measurements reflect dephasing in the transverse plane after application of a radiofrequency
(RF) pulse. These measurements were found to be associated with cartilage water content
and reflect an indirect measure of the ECM collagen content.[63] Laminar analysis of cartilage found these measurements to be higher in the superficial
layers of cartilage than in the deep layers.[64] Higher T2 values were shown to predict the development of cartilage lesions.[65] At our institutions, a few select surgeons request T2 mapping as part of the preoperative MRI to identify problem areas in the cartilage
before performing arthroscopy ([Fig. 3]). T2 mapping can discriminate between repaired knee cartilage and adjacent healthy cartilage,[66]
[67]
[68] and it may be particularly helpful in assessing the maturation of reparative cartilage.[69]
Fig. 3 T2 mapping was performed as part of the clinical preoperative MRI protocol per the
request of the referring orthopaedic surgeon. (a) Sagittal T2 color map is shown with the scale provided on the left; red represents
highest T2 measurements; dark blue reflects the lowest T2 values. Focal high T2 values
are seen within the central tibiotalar cartilage (yellow arrow) and with the subchondral
talar dome (red arrow). (b) Sagittal fat-suppressed proton-density images in the same patient demonstrate that
these findings correspond to full-thickness tibiotalar cartilage loss and subchondral
bone marrow edema, respectively.
T2* mapping measures transverse-plane dephasing using multiecho GRE techniques. These
sequences have a shorter acquisition time but are also more vulnerable to local field
inhomogeneity.[70]
[71] Both T2 and T2* are affected by the magic angle effect; that is, the values increase as the angle
between collagen fibers and B0 approaches 55 degrees.
T1ρ Mapping
T1ρ imaging is more challenging to acquire than T2 mapping and therefore only performed at a few select academic institutions. The imaging
is difficult to acquire due to B0 and B1 inhomogeneity, specialized RF pulse sequence requirements, and long acquisition times
that may result in high specific absorption rates (SARs). The SNR gain at 7 T has
been used, however, to show the feasibility of acquiring high-resolution T1ρ images
(0.2 mm2 in-plane resolution) in reasonable acquisition times (< 30 minutes) and within SAR
constraints.[72] T1ρ assesses the spin-lattice (T1) relaxation in the rotating frame[73] and is thought to reflect the proteoglycan content of the ECM.[74] T1ρ values are higher in patients with OA compared with healthy subjects.[75] T1ρ was also shown to predict morphological chondral wear.[76]
Ultrashort Echo Time and Zero Echo Time Imaging
Ultrashort Echo Time and Zero Echo Time Imaging
Like cortical bone, tendons, and menisci, the deep calcified part of cartilage contains
a high fraction of components with “ultrashort” transverse relaxation times. This
essentially equates to a post-RF pulse signal decay rate that is too rapid to allow
signal acquisition.[77] Ultrashort echo time (UTE) and zero echo time use specialized acquisition and reconstruction
techniques to capture these ultrashort components before signal decay. Although the
application of techniques for imaging of cartilage is not common, UTE was shown to
delineate the calcified deepest cartilage layer[78] and used to evaluate the integrity of this layer in osteochondral allografts.[79] UTE also enables T2 and T2* mapping of tissues with a high fraction of ultrashort components.[80]
Delayed Gadolinium-Enhanced MRI of Cartilage
Delayed Gadolinium-Enhanced MRI of Cartilage
The dGEMRIC MRI is performed after intravenous injection of a gadolinium-based contrast
with subsequent joint exercise and substantial time delay to allow diffusion of the
contrast into the joint. Gadopentetate dimeglumine (Gd-DTPA2−), the MRI contrast, is an anion and repelled by the negatively charged GAGs, allowing
this technique to map GAG content within the cartilage. Damaged cartilage with low
GAG content will accumulate more Gd-DTPA2− and therefore have a shorter T1 relaxation time. This technique was used in research to study a variety of topics
including cartilage repair tissue,[81]
[82] effects of tibial osteotomy on cartilage,[83] inflammatory arthritis,[84] and the effects of chronic joint unloading.[85] The need for intravenous contrast is the main drawback of dGEMRIC; however, it does
allow an indirect MR arthrogram to be obtained during the delay between injection
and acquisition of dGEMRIC T1 imaging. This may particularly be helpful in the morphological evaluation of the
acetabular or glenoid labrum.
Sodium (23Na) Imaging
In contrast to Gd-DTPA2−, sodium (23Na+) is a naturally occurring cation (albeit in very low concentrations) that is attracted
to and counteracts the negatively charged GAGs in the cartilage ECM. The distribution
of 23Na+ can hence also be used to map the cartilage GAG content, with cartilage degeneration
resulting in a lower concentration of 23Na+ ions.[86] Unsurprisingly, 23Na imaging correlates well with dGEMRIC.[87] The low concentrations of 23Na+ in cartilage, however, make it difficult to elicit signal during MRI, resulting in
noisy images and long acquisition times. The SNR gain at 7-T MRI is particularly useful
for 23Na imaging.[88] Because the Larmor frequency of 23Na+ differs from 1H, specialized transmit-receive coils are also required to perform sodium imaging.[89] As with T2 mapping, sodium imaging can discriminate between cartilage repair tissue and healthy
cartilage, with lower sodium signal intensity in repair tissue compared with healthy
cartilage reflecting a diminished GAG content[90]
[91] ([Fig. 4]).
Fig. 4 Axial T2-weighted 7-T MR image (left panel) of the left knee, demonstrating a displaced
osteochondral fragment reattachment (arrowhead) at the medial facet of the patella.
Synovial fluid is seen within a full-thickness fissure (arrow). On the conventional
23Na map (middle panel), hyperintense signal is seen from synovial fluid within the
fissure (arrow). On the 23Na-IR map (right panel), there is suppression of signal from synovial fluid within
the fissure. The calibration phantoms containing 300 mM, 250 mM, 200 mM, and 150 mM
are seen at the anterior aspect of the knee. The sodium images represent concentration
maps with colored bars indicating range of [Na+] in mM (red = 600 mM; blue = 0 mM). Reproduced with permission from Chang et al.[91]
Diffusion Tensor Imaging
The cartilage ultrastructure consists of a highly organized network of collagen that
results in anisotropic diffusion of water. In cartilage, DTI can assess both proteoglycan
content through mean apparent diffusion coefficient (ADC) and collagen microarchitecture
through fractional anisotropy (FA). Both mean ADC and FA values were found to be able
to discriminate cartilage in OA patients from healthy cartilage, with FA having higher
specificity.[92] DTI was found to have high accuracy for detecting cartilage damage as well as for
grading cartilage damage.[93]
GAG Chemical Exchange Saturation Transfer Imaging
GAG Chemical Exchange Saturation Transfer Imaging
Water exists in two states within cartilage, either bound to macromolecules or in
the free water state. Water protons bound to macromolecules have unique RF frequency
that can be saturated using off-resonance RF pulses. The bound water pool then interacts
with the free water pool resulting in partial saturation of the free water pool. This
effect can be measured to estimate local macromolecule content.
With gagCEST, off-resonance RF saturation pulses are designed specifically to saturate
exchangeable protons residing on the hydroxyl groups of cartilage GAGs ([Fig. 5]). This technique correlates well with 23Na+ imaging, and like 23Na+ imaging, it is optimally performed at ultrahigh field strength (7 T) magnets.[94]
Fig. 5 Representative T1rho maps of cartilage in the (a) lateral and (b) medial compartments. T1rho maps of menisci in the (c, d) lateral and (e, f) medial compartments, respectively, obtained from a doubtful/minimal osteoarthritis
patient. The color bars on the right show the T1rho values ranges, respectively. The
different bar scale ranges can display the respective T1rho values distribution of
cartilage and meniscus more effectively. Reproduced with permission from Wang et al.[95]
Conclusion
OA is the most prevalent joint disease in the United States with a tremendous socioeconomic
burden. With efforts to develop a DMOAD for OA proving unsuccessful to date, it is
clearly evident that imaging beyond radiography is needed for both clinical diagnoses
of OA and for use as an outcome measure in OA research. MRI provides an unparalleled
assessment of articular cartilage and has aptly been incorporated into the major clinical
studies of OA including the OAI and the MOST. For the purpose of quantifying data
from such trials, MRI-based semiquantitative grading systems for OA have been developed.
The most widely used of these classification schemes include the WORMS and the BLOKS,
with the MOAKS representing a hybrid of both these systems. In addition to the morphological
evaluation, advanced MRI techniques have been developed to assess the biochemical
composition of cartilage. These include relaxometry measurements (T2, T2*, and T1ρ
mapping), sodium imaging, dGEMRIC, gagCEST, and DTI. These techniques have the potential
to serve both as imaging biomarkers for OA and as quantitative, reproducible, and
objective end points for OA research.
