Ultrastructural MR imaging techniques of the knee articular cartilage: problems for routine clinical application
- 179 Downloads
The high incidence of cartilage lesions together with new surgical treatment techniques have necessitated the development of noninvasive cartilage evaluation techniques. Although arthroscopy has been the standard for cartilage evaluation, MR imaging has emerged as the imaging method of choice, allowing morphological evaluation of cartilage and cartilage repair tissue, as well as evaluation of its biochemical content. This article deals with current ultrastructural MR imaging techniques for cartilage evaluation, indicating the advantages as well as the drawbacks for routine clinical application.
KeywordsCartilage MR imaging Knee Osteoarthritis
Injuries and degenerative changes in the articular cartilage are a significant cause of morbidity and diminished quality of life, with osteoarthritis ranking second only to cardiovascular disease as a cause of work-related disability . Because of this high incidence of cartilage damage and thanks to new surgical cartilage repair procedures, the evaluation of articular cartilage has gained increasing attention in the scientific community. Although arthroscopy has been the gold standard for diagnosing and monitoring cartilage damage and repair, it is less than optimal for several reasons. It is invasive and expensive. Moreover, it only allows visual inspection of the cartilage surface, enforcing additional evaluation (e.g., physically probing the cartilage) to find hidden defects within the midsubstance of the tissue.
Although several imaging methods exist for cartilage evaluation, MR imaging of cartilage has been shown to be accurate for the detection of chondral abnormalities [2, 3, 4, 5, 6, 7, 8] and in evaluating cartilage-repair procedures [9, 10, 11].
Apart from the evaluation of morphological defects, certain MR imaging contrast mechanisms have shown to reveal detailed information regarding the biochemical content (ultrastructure) of cartilage.
This article deals with the current available ultrastructural MR imaging techniques related to their biochemical target, indicating the advantages as well as the drawbacks for routine clinical application. For a good understanding of these imaging techniques a brief discussion on cartilage anatomy is given.
Structure of articular cartilage
Scanning electron microscopy performed after freeze-fracture sectioning shows the three-dimensional structure of the collagen network [16, 17] and has put the aforementioned light microscopy model into new perspective. Matrix collagen within cartilage is organized in continuous leaf-like structures that radiate from the subchondral interface in a perpendicular orientation and then curve into the horizontal orientation at the articular surface (Fig. 2). Moreover, within the leafs the orientation of individual collagen fibers appears random in all regions. No qualitative difference can be made in the orientation of collagen fibers in the different layers as determined by the light microscopy model . This scanning electron microscopy cartilage model supports a complex three-dimensional organization of the collagen network which can explain the different layers seen on MR images (see later in the discussion on differences in T2 decay and the magic-angle effect).
The overall functional integrity of cartilage is determined not only by the solid/fluid volume fraction, collagen composition, molecular structure and organization, and GAG composition, but also on their interdependent effects. Although the exact pathophysiological mechanisms for cartilage damage still need to be determined, one can intuitively understand that each matrix component is involved and has determinative effects on the others.
Ultrastructural MR imaging techniques for articular cartilage
Overview of different techniques for ultrastructural cartilage imaging and their relationship with the major constituents of the cartilage matrix. dGEMRIC delayed gadolinium-enhanced magnetic resonance imaging of cartilage
23Na spectroscopic imaging
Magnetization transfer mapping
Proton density (ρ) mapping
Diffusion imaging and fractional anisotropy
The first group of techniques are related to the proteoglycan (GAG) content of articular cartilage and comprise delayed gadolinium-enhanced MR imaging of cartilage (dGEMRIC), T1ρ imaging, and sodium MR spectroscopy.
In practice, cartilage T1 is high in normal cartilage and low in GAG-depleted, osteoarthritic cartilage. The T1 value is determined using a multi-inversion recovery turbo-spin-echo imaging series, followed by curve fitting the T1 images to generate T1 maps. Color or gray-scale post-processing of these T1 maps is widely used to render the images de visu interpretable (Fig. 4). The value of this technique and the possible clinical applications have been repeatedly emphasized in the literature [21, 22, 23, 24].
The observed differences in calculated T1 values between different research groups (ranging from 300 to 580 ms for normal cartilage and 210 to 360 ms for diseased cartilage), however, deserve some attention [20, 22, 23, 26]. An explanation for these differences is not readily available. A possible cause could be the use of different field strengths (which is not the case in above-mentioned studies, all performed at 1.5 T): an increase in T1 values is noted with increasing field strength .
A second issue concerns the “relaxivity” of Gd-DTPA2−: relaxivity is defined as the change in relaxation rate of contrast agent. Studies revealed the relaxation rate of Gd-DTPA2− to be temperature dependent as well as field-strength dependent. When not corrected, this change in relaxivity introduces errors in the T1 relaxation rate especially at lower field strengths . Thirdly, the mathematical algorithm used to calculate T1 may account for observed differences. Fourthly, some protocol issues for practical clinical application of this technique have been described : a double dose of Gd-DTPA2− should be injected intravenously, followed by an immediate active-joint exercise period. T1 relaxation measurements should be performed 2–3 h post-injection. Color mapping of T1 calculated images is appropriate to discern variations in T1 across the tissue. Incoherencies in the applications of these issues introduces T1 differences.
As stated previously, T1 of cartilage in the presence of Gd-DTPA2− is a good indicator of relative FCD. If information on the absolute FCD (and thus GAG concentration) is desirable, an electrochemical model must be used . The knowledge of absolute FCD, however, seems of relative importance in practical clinical applications, as it can be compared with known normal T1 relaxation rates.
T1ρ imaging is a new technique especially in the field of cartilage imaging: T1ρ imaging means spin-lattice relaxation in the rotating frame and is characterized by the time constant that defines the magnetic relaxation of spins under the influence of a radio-frequency field (T1ρ). Changes in T1ρ were observed in chemically proteoglycan-depleted cartilage plugs but not in collagenase-treated tissue [29, 30]. Recently, the application of this technique in a 1.5-T MR imaging system was described . A strong correlation exists between proteoglycan loss and increase in T1ρ relaxation time, not noted with collagen loss, making it a sensitive and specific method. T1ρ imaging can therefore be used to map the proteoglycan distribution in cartilage.
T1ρ imaging seems to provide a convenient method to evaluate the slow-motion components that cause relaxation. These slow-motion processes seem to play a larger role in cartilage T1ρ compared with that of muscle . The exact physical mechanisms responsible for this type of dispersion behavior are not well known, so further research needs to be done to determine and separate these mechanisms.
T1ρ imaging can be obtained by preparing the magnetization with a spin-locking prepulse followed by a standard fast-spin-echo (FSE) sequence . Although the clinical usefulness of this technique is not yet established well in literature, T1ρ imaging could serve as an attractive alternative to the dGEMRIC method in the future. In comparison with the dGEMRIC method, T1ρ imaging does not require intravenous injection of contrast agent, no joint exercise nor a time-delay period of 2 h before imaging.
Na-23 spectroscopic imaging
Sodium MRI is a recently described new technique for cartilage imaging. Its utility is based on the ability of sodium imaging to depict regions of proteoglycan depletion . According to the same principle as described for dGEMRIC imaging, 23Na atoms are associated with the FCD present in GAG. Because of the inherent positive charged 23Na, there is a direct relationship between local 23Na concentration and FCD. Some spatial variation in 23Na concentration seems present within normal cartilage and in cartilage samples.
Sodium imaging has been shown to be sensitive to small changes in proteoglycan concentration . There is a substantial reduction (~50%) in image intensity in enzymatic GAG-degraded regions compared with the nondegraded regions ; however, despite these promising results, the application of this technique in routine clinical cartilage imaging seems precluded since it requires special hardware modifications (e.g., preamplifier for sodium frequency, double-tuned coils), not available on routine clinical MR imaging systems.
The second group of techniques are related to the collagen content of cartilage and comprise T2-mapping, magnetization transfer (MT) imaging, and magic-angle imaging.
T2 mapping of articular cartilage is a widely discussed technique in the literature, demonstrating its value in the evaluation of early (ultrastructural) cartilage damage [34, 35, 36, 37, 38, 39, 40, 41, 42]. There is, however, no consensus on the ultrastructural components determining the T2 value of cartilage: this makes T2 mapping of cartilage a sensitive although non-specific technique. The vast majority of authors agree on the major contribution of collagen orientation (determined by the magic-angle effect) and collagen concentration to the T2 value [34, 35, 38, 39, 40]. Some authors indicate a significant additional contribution from proteoglycan and water content [36, 37, 41]. All authors, however, agree that ultrastructural cartilage damage is associated with an increase in T2 relaxation time, which is important for practical clinical applications.
As already mentioned, the magic-angle effect is a major determinant in the T2 behavior of cartilage, explained by the anisotropic motion of water molecules parallel to the direction of collagen fibers. Spin-spin coupling (T2) is mediated by the inter-spin distance and by the respective magnetic fields of each proton, which are characterized by the term (3cos2θ-1). θ is the angle between the main magnetic field (B0) and a vector through adjacent protons . In isotropic tissues, where water is freely moving, this angular dependence averages to zero. With the angular anisotropy of water molecules along the collagen fibers, however, a net angular dependence remains, leading to anisotropic T2. With this anisotropy, spin-spin coupling is minimized (i.e., T2 is maximal) at the angles where (3cos2θ-1)=0. This corresponds to θ=54.7°, 125.3°, 234.7°, or 305.3°, with θ=54.7° called the magic angle (θm). Applied to cartilage, with its highly organized collagen matrix, T2 is heterogeneous, anisotropic, and dependent on the tissue depth : T2 of cartilage is therefore a local parameter, associated with the local structure of the tissue and orientation dependent. This is nicely demonstrated by Mosher et al. [45, 46, 47].
A second issue, apart from the magic-angle effect, that should be accounted for is the multi-exponential T2 behavior of articular cartilage. A study by Mlynarik et al.  suggested a bi(multi)exponential T2 behavior, with a short T2 component in the first 10 ms. These ultra-short T2 values are not resolved with most clinically used techniques and thus accounting for erroneously high values .
Resulting T2 values as presented by different authors. The cartilage was arbitrarily divided in a deep and superficial portion, from which mean T2 values were calculated in order to demonstrate spatial differences in T2 relaxation times
Aging appears to be associated with an asymptomatic increase in T2 relaxation times in the “transitional zone” . This finding, together with the described angular T2 increase, should be considered in the differentiation between normal and pathological T2 increase when interpreting T2 maps. Three different patterns of pathological T2 were suggested by Mosher et al. : (a) focal increased T2 confined to the radial zone; (b) heterogeneously elevated T2 extending to the articular surface; and (c) a focal cartilage tear with associated change in the spatial T2 distribution.
Magnetization transfer imaging
The MT pulse can be added to every MR imaging sequence . In practice, gradient-echo techniques are most often used. Two data sets are often required, one with MT saturation on and one with it off, followed by image subtraction. As opposed to the signal drop in normal cartilage, a region with disruption of the collagen framework is supposed to show increased signal due to a lower MT effect. Consequently, the subtracted image of this pathological region appears dark. Major disadvantages of this subtraction tool is patient motion between the two acquisitions leading to subtraction artifacts.
Despite the high specificity of the MT imaging technique for collagen matrix damage, Hohe et al.  stressed the important inter-individual as well as the intra-individual differences in MT rate. This may preclude its use in clinical applications.
Moreover, the two-proton pool model (i.e., free and bound pool) might be too simplified for cartilage MT imaging . A four-site exchange system (comprising bulk water, collagen related water, collagen fibrillar water, and proteoglycan-related water) is proposed as an accurate model for cartilage relaxation and inter-spin group coupling. The actual clinical MT imaging only evaluates the collagen–bulk water subsystem, despite that the other coupling subsystems may be more relevant in early osteoarthritis. The authors state that new MT imaging techniques should be developed to focus on the other coupling subsystems to detect molecular abnormalities in the clinical setting of early osteoarthritis.
Magic-angle imaging, wherein the angular dependence of T2 is measured, provides a specific indication of collagen ultrastructure [18, 34, 44, 50, 56, 57]. This technique uses the magic-angle effect by placing the normal axis of the cartilage approximately 55° with respect to B0. The difficulty in getting the exact angle, as patients cannot be turned in the magnet, precludes its use clinically, although it is likely to become a practical method for in vitro applications.
The third group of techniques are related to the interstitial water content of cartilage and comprise proton-density (PD) mapping and diffusion imaging.
Mapping the spin density of articular cartilage is directly related to the interstitial water content. It is obtained as a by-product from either T1 or T2 measurements. The water content can be quantified by ensuring negligible T1/T2/diffusion and calibrating to a known water phantom. The use of phantoms is mandatory since proton-density (PD) measurements are always relative . Additional factors need to be concerned, precluding the practical use of PD mapping: PD measurements should be corrected for focal solid matrix density; secondly, inherent relaxation processes affect the calculated PD, especially fast T2 decay . Moreover, PD mapping reflects only the MRI detectable water and thus obscuring other water fractions which inevitably play a role in the disease process. Further research seems necessary to evaluate this highly sensitive but practical challenging technique.
Self-diffusion of water provides a possible strategy for measuring water content because diffusivity is very sensitive to hydration . The ability of water (and other solutes) to move through cartilage is important because cartilage is an avascular tissue. The apparent diffusion coefficient (ADC) increases with proteoglycan and collagen depletion [60, 61] and has been related to degenerated human cartilage ex vivo . To obtain a calculated ADC map, in which the signal intensity is proportional to the diffusivity, a series of images is acquired with different applied diffusion gradients (termed by the b-value). The diffusion gradients will induce a net phase change for moving spins depending on the distance they moved, resulting in signal loss.
The steric hindrance of the extracellular matrix (volume fraction, composition) is an important determinant for the diffusivity, explaining the ADC increase in diseased cartilage. The specific relationship, however, has yet to be determined . An extended application of ADC mapping is diffusion tensor (i.e., fractional anisotropy) mapping, which gives an estimate of diffusion anisotropy. In normal cartilage, diffusion seems isotropic . Cartilage matrix damage probably alters the diffusion anisotropic in analogy to the experiments performed by Quinn et al. .
The described ultrastructural MRI techniques provide a probe to give information on cartilage which is inaccessible with any other non-invasive modality. Articular cartilage is a unique tissue from an MRI point of view: it consists out of a highly ordered collagen network, responsible for its T2 orientational dependence, and charged hydrophilic GAGs responsible for its T1 relaxation behavior in the presence of a charged MRI contrast agent.
For the evaluation of articular cartilage, specific methods should be used: three methods permit a very specific measure of the cartilage biochemical state: PD mapping directly measures water content, magic-angle imaging is a specific measure of collagen ultrastructure and charge-based methods (either dGEMRIC or sodium MRI) provide a specific measure of the GAG distribution.
For practical clinical applications, the whole picture seems less attractive: out of the specific techniques, only dGEMRIC can be used when keeping in mind above-mentioned drawbacks and major disagreements in literature concerning resulting T1 values. For the evaluation of the collagen component, the less specific technique of T2 mapping is used, and although a correlation is found between MT rate and collagen damage, it still remains unclear whether this technique has sufficient sensitivity and specificity to be used in clinical cartilage imaging. The evaluation of the water component poses even more problems, as the clinical use of diffusion imaging seems precluded because of limited resolution and motion artifacts.
Concerning clinical imaging of articular cartilage the overall long acquisition schemes (i.e., patient motion), the 2-h interval needed for dGEMRIC and the need for image post-processing preclude a widespread use in daily clinical practice. Suggested solutions might be the use of higher field systems (3 T); however, susceptibility artifacts, inevitably present in each patient after surgical cartilage repair procedures, are enforced in higher field systems and thus possibly disabling correct image interpretation. Further work is necessary to deal with this issue. Another approach might be to limit the field of view to a specific cartilage area, enabling the shortage of acquisition time or the use of small surface coils to improve the signal-to-noise ratio.
Although these conclusions seem disenchanting for clinical cartilage imaging, further refinement of existing methods and development of new techniques may in the event lead to imaging protocols ready for daily use able to reveal both the morphology and biochemical content of cartilage lesions and cartilage repair tissue.
- 1.Cooper C (1995) Occupational activity and the risk of osteoarthritis. J Rheumatol 43:10–12Google Scholar
- 4.Bredella MA, Tirman PF, Peterfy CG, Zarlingo M, Feller JF, Bost FW, Belzer JP, Wischer TK, Genant HK (1999) Accuracy of T2-weighted fast spin-echo MR imaging with fat saturation in detecting cartilage defects in the knee: comparison with arthroscopy in 130 patients. AJR 172:1073–1080PubMedGoogle Scholar
- 7.Sonin AH, Pensy RA, Mulligan ME, Hatem S (2002) Grading articular cartilage of the knee using fast spin-echo proton density-weighted MR imaging without fat suppression. AJR 179:1159–1166Google Scholar
- 10.Alparslan L, Minas T, Winalski CS (2001) Magnetic resonance imaging of autologous chondrocyte implantation. Semin Ultrasound CT MR 22:341–351Google Scholar
- 12.Buckwalter JA, Mankin HJ (1997) Articular cartilage. I. Tissue design and chondrocyte matrix interactions. J Bone Joint Surg Am 79:600–611Google Scholar
- 13.Buckwalter JA, Mow VC (1992) Cartilage repair in osteoarthritis. In: Moskowitz RW, Howell DS, Goldberg VM, Mankin HJ (eds) Osteoarthritis. Saunders, Philadelphia, pp 71–107Google Scholar
- 14.Mankin HJ, Brandt KD (1992) Biochemistry and metabolism of articular cartilage in osteoarthritis. In: Moskowitz RW, Howell DS, Goldberg VM, Mankin HJ (eds) Osteoarthritis. Saunders, Philadelphia, pp 109–154Google Scholar
- 15.Benninghoff A (1925) Form und Bau der Gelenkknorpel in ihren Beziehungen zur Funktion. Anat Entwicklungsgesch 76:43Google Scholar
- 26.Stevens K, Berger F, Yoshioka H, Steines D, Genovese M, Lang P et al. (2001) Contrast-enhanced MRI measurements of GAG concentrations in articular cartilage of knees with early osteoarthritis. Proc 87th Scientific Assembly and Annual Meeting of the Radiological Society of North America, RSNA 2001. Radiology 221(P):288Google Scholar
- 45.Mosher TJ, Smith H, Dardzinski BJ, Schmithorst VJ, Smith MB (2001) MR imaging and T2 mapping of femoral cartilage: in vivo determination of the magic angle effect. AJR 177:665–669Google Scholar
- 46.Mlynarik V (2002) Magic angle effect in articular cartilage. AJR 178:1287–1288Google Scholar
- 47.Goodwin DW, Dunn JF (2002) MR imaging and T2 mapping of femoral cartilage. AJR 178:1569–1570Google Scholar