| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Musculoskeletal Imaging |
1 From the Departments of Radiology (T.C.D., Y.L., H.J., S.M.) and Orthopaedic Surgery (M.D.R., S.M.), University of California at San Francisco, 185 Berry St, Suite 350, San Francisco, CA 94107-1739. Received June 20, 2003; revision requested August 29; final revision received December 19; accepted January 13, 2004. Supported by NIH grant R01 AR46905. Address correspondence to T.C.D. (e-mail: tcdunn@mrsc.ucsf.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Knee cartilage was evaluated in 55 subjects who were categorized with radiography as healthy (n = 7) or as having mild OA (n = 20) or severe OA (n = 28). Cartilage regions were determined with manual segmentation of an MR image acquired with spoiled gradients and fat suppression. The segmentation was applied to a map of T2 relaxation time and was analyzed in four knee cartilage compartments (ie, the medial and lateral tibia and femur). Differences between cartilage compartment T2 values and subject groups were analyzed with analysis of covariance. Correlations of cartilage T2 values with clinically reported symptoms and cartilage thickness and volume were examined. Cartilage T2 values were converted to z scores per voxel on the basis of normal population values in the same cartilage compartment to better interpret cartilage heterogeneity and variation from normal.
RESULTS: Healthy subjects had mean T2 values of 32.1–35.0 msec, while patients with mild and severe OA had mean T2 values of 34.4–41.0 msec. All cartilage compartments except the lateral tibia showed significant (P < .05) increases in T2 relaxation time between healthy and diseased knees; however, no significant difference was found between patients with mild and severe OA. Correlation of T2 values with clinical symptoms and cartilage morphology was found predominantly in medial compartments.
CONCLUSION: Femoral and medial tibial cartilage T2 values increase with the severity of OA.
© RSNA, 2004
Index terms: Cartilage, MR, 452.121411 • Knee, arthritis, 452.7 • Knee, MR, 452.121411 • Magnetic resonance (MR), tissue characterization, 452.121411
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Quantitative measures of T2 relaxation times may be useful in the characterization and long-term tracking of OA. Previous reports have demonstrated spatial variation of T2 relaxation times in cartilage explants (4), healthy subjects (5), and substantial changes with age (6), but we are unaware of reports that present T2 relaxation time variation with OA in humans. Cartilage tissue analysis has shown increased water content in tissue that has been degraded through OA processes (7) and decreased glycosaminoglycan concentration and proteoglycan size in diseased tissue (8,9). These findings support those of previous studies (10) that early cartilage degeneration due to collagen damage and changes in collagen content and arrangement will increase water mobility in the tissue, thus increasing T2 relaxation time. Thus, the purpose of our study was to evaluate differences in femoral and tibial cartilage T2 relaxation times in healthy subjects and patients with varying degrees of OA and to develop a mapping and display method based on calculating T2 relaxation time z scores for visual grading and assessment of cartilage heterogeneity in patients with OA.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Two radiologists (one with 10 years experience, the other with 20 years experience) worked in consensus and used the Kellgren-Lawrence scoring system (12,13) to grade OA with standing anteroposterior radiographs. The radiologists were blinded to patient information during evaluation. The Kellgren-Lawrence scoring system has five categories. A score of 0 (none) is given if no osteoarthritic features are present. A score of 1 (doubtful) indicates minute osteophytes of doubtful importance. A score of 2 (minimal) indicates definite osteophytes without reduction of the joint space. A score of 3 (moderate) is given when the joint space has diminished. A score of 4 (severe) is given for greatly reduced joint space and sclerosis of the subchondral bone.
In our study, subjects with a score of 0 were grouped in the healthy population, while those with a score of 1 or 2 were categorized as having mild OA. Subjects with a score of 3 or 4 were classified as having severe OA. Seven subjects were graded as healthy (mean age, 38 years; range, 22–70 years) and included three women (mean age, 34 years; range, 22–47 years) and four men (mean age, 41 years; range, 23–70 years). Twenty patients were classified as having mild OA (mean age, 63 years; range, 46–81 years) and included 13 women (mean age, 65 years; range, 50–78 years) and seven men (mean age, 56 years; range, 46–81 years). Twenty-eight patients were classified as having severe OA (mean age, 67 years; range, 43–88 years) and included 14 women (mean age, 70 years; range, 46–88 years) and 14 men (mean age, 64 years; range, 43–78 years).
We used the Western Ontario and McMaster University (WOMAC) osteoarthrosis index to determine function, quality of life, and joint pain (14). This arthrosis impact scale has been used in studies of patients with knee OA to determine both the effect of different therapies for treatment of pain and the relationship between disease severity and clinical importance of symptoms. Extensive validation of the WOMAC index has been performed with regard to the measures of function and quality of life in patients with OA (14).
An orthopedic surgeon (M.D.R.) interviewed all participants about the amount of knee pain and stiffness they were experiencing. Patients were also asked questions in an effort to assess their ability to complete everyday tasks such as ascending stairs, standing, walking, rising from bed, or going shopping. The responses were used to determine the overall WOMAC score on a scale that ranges from 0 (no pain or stiffness) to 500 (extreme pain, stiffness, and impaired function).
Imaging and Evaluation
Imaging was performed with a Signa 1.5-T echo-speed system (GE Medical Systems, Milwaukee, Wis) with a dual phased-array coil (USA Instruments, Cleveland, Ohio). A custom leg holder, with the knee in approximately 20° of flexion, was used to minimize motion and position the coil. A two-dimensional dual-echo spin-echo sequence was used (repetition time msec/echo times msec, 1,500/10 and 45; voxel size, 0.468 x 0.468 x 4 mm; examination time, 5 minutes 24 seconds; field of view, 12 cm; matrix, 256 x 256) to generate a sagittal T2 relaxation time map with custom software (IDL; Research Systems, Boulder, Colo), assuming a single exponential decay component. Spoiled gradient-echo images with fat suppression were acquired by using the spectral inversion at lipids, or SPECIAL, technique (15) (repetition time msec/echo time msec/inversion time msec, 30/3.3/8; flip angle, 40°; voxel size, 0.234 x 0.234 x 2 mm; examination time, 9 minutes 31 seconds, field of view, 12 cm; matrix, 512 x 512) along with the T2 relaxation time maps.
Because of the inhomogeneous reception of the surface coils, the signal intensity throughout the sections is inhomogeneous. Thus, all images are subjected to a preprocessing step with a three-dimensional low-pass filter-based correction algorithm that will normalize voxel intensity through the knee volume but will not affect the calculation of T2 values, as verified by the results of experiments that are not presented herein. In those experiments, the coefficient of variation was found to be less than 2% when comparing T2 values from regions of interest with and without the low-pass filter correction. The cartilage was manually segmented from the spoiled gradient-echo images by using custom graphic user interface programs that were developed with Matlab (Mathworks, Natick, Mass). Two research staff scientists at our institution, both with 2 years of experience, placed points on the image that were an average of 8–10 pixels (approximately 2 mm) apart and served as knots for a two-dimensional spline that defined the outline of the cartilage boundary, as shown in Figure 1. The boundary defined the outer limits of a binary mask, a matrix in which elements corresponding to cartilage have a value of one and otherwise have a value of zero. The mask was downsampled in two dimensions by a factor of two, which halved the number of in-plane pixels in the image in both dimensions. Also, the union of adjacent sections was created. In this way, the voxel sizes of the cartilage map and the T2 map were made equal. The mask was multiplied by the corresponding matrix element with the T2 map. Thickness measurements were made with custom software routines (Matlab; Mathworks) on sagittal cartilage images by using the diameters of the largest circle possible that could fit inside the cartilage mask with the center of the circles along the midline axis (16). The diameters ranged in size from 2 to 30 pixels (approximately 0.4–6.0 mm). The mean cartilage thickness for each section was the mean value of the circle diameters, while the thickness for each compartment was the average thickness of all sections. Cartilage volumes for the femur and tibia were computed from the segmented cartilage compartments. The analysis focused on the weight-bearing surfaces; therefore, patellar cartilage was not segmented for analysis.
|
Statistical Analysis
All statistical processing was performed with SAS version 8.2 software (SAS Institute, Cary, NC). First, we summarized and compared patient characteristics in the three groups. We used the analysis of variance to test for significant age differences and the Fisher exact test to compare the sex difference.
Several authors (T.C.D., Y.L., H.J., S.M.) analyzed in consensus the distribution of T2 values in each cartilage compartment of each subject. T2 values of more than 200 msec were considered outliers, occurred infrequently (usually <1% of voxels in the compartment), and were excluded from analysis because they were more than double the highest expected mean T2 value of healthy cartilage (5,17). The differences in mean T2 values for each disease group for each cartilage compartment were compared with the analysis of covariance. To account for the small differences in standard error of the estimated mean for each group, we used the weight of inverse variance of the estimated mean (1 divided by the squared standard error) for each patient and performed a weighted analysis of covariance.
Because age was significantly different among groups, we adjusted for this factor in our analysis of covariance models. Tukey-Kramer tests were used to compare pairwise differences in mean T2 values between groups. Residuals of the analysis of covariance models were also examined for normal distribution by using the Kolmogorov-Smirov test, and none were significantly different from normal values. Least-squares means and standard errors were used as summary statistics for each group.
The relations between mean T2 of cartilage with WOMAC scores, cartilage thickness, and cartilage volume were evaluated with the Spearman correlation coefficient. In addition to the relation of the T2 value to volume and thickness of each compartment, cartilage volume and thickness relations were evaluated for the opposing compartment. For example, the medial femoral cartilage is the opposing volume of the medial tibial cartilage.
Per-pixel and per-compartment z scores were generated by using the mean and standard deviation of T2 values from the healthy subjects in each compartment with the following equation:
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
Figure 3 shows cartilage viewed as a z score image. Each cartilage compartment is referenced to the normal values for the compartment in the conversion; areas of yellow indicate increases in T2 value.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The complex structure of cartilage provides its unique mechanical and imaging properties. The presence of a high concentration of proteoglycans that are immobilized in the collagen network of cartilage creates a large osmotic pressure that draws water into the tissue, which provides resistance to compressive loads (1). Previous descriptions of MR images of damaged cartilage have alluded to the apparent increase in T2 values because of an increase in the relative water content and a decrease of T2-shortening effects of the matrix on the water (19). Moreover, these effects could be seen at the earliest stages of degradation, before any thinning of the matrix occurred (19).
It is possible that the underlying cartilage changes that lead to a higher T2 value similarly occur early in the development of the disease and do not progress as much as mechanical degradation of the cartilage. Thus, defining disease groups on the basis of the Kellegren-Lawrence grading system, which is heavily influenced by bulk cartilage changes, may not be sensitive to cartilage changes that affect T2 relaxation time; hence, very little difference is seen between patients with mild OA and those with severe OA. Furthermore, the Kellgren-Lawrence grading system has some recognized limitations (2); in particular, an emphasis is placed on the presence of osteophytes, which we would not expect to alter the mean T2 value in the cartilage compartment.
Buckwalter and Mankin (20) have reviewed the mechanism of cartilage degradation in detail, and they describe the initial stage of OA changes to increase water mobility in the cartilage. The longer T2 values found with OA agree with this mechanism. Furthermore, Lusse et al (10) have demonstrated a positive correlation between increases in T2 relaxation time and cartilage water content in ex vivo cartilage from patients with OA who are undergoing total knee replacement surgery.
To our knowledge, previous studies of cartilage T2 values in patients with OA have not been performed; however, Mosher et al (6) have shown a significant increase in T2 values with aging. Unfortunately, in our study, age matching was not achieved between healthy subjects, patients with mild OA, and patients with severe OA. Thus, age-related changes in cartilage T2 values may confound OA-related changes in cartilage T2 values. We accounted for this possibility by using the analysis of covariance, with age as a confounding factor.
The changes in cartilage T2 values with OA severity are supported by significant negative correlation with cartilage compartment volume and thickness in the medial region. Thus, the cartilage degradation is reflected in higher T2 values. In our study, however, we evaluated two-dimensional thickness in the sagittal plane, and we did not correct for out-of-plane curvature of cartilage and bone. Clinical indicators of disease progression, as evaluated with WOMAC function and pain assessments, appear to correlate with increased cartilage T2 values, particularly in the medial compartments.
The correlation of bone marrow edema and pain or disease severity has been evaluated (11). In contrast to the present finding of a correlation of WOMAC function and pain scores with increased T2 relaxation times, bone marrow edema did not correlate with any of the WOMAC indexes.
Care needs to be taken when interpreting T2 values of articular cartilage. Goodwin and Dunn (21) have described how T2 values will vary with depth from the articular surface because of collagen fibril orientation to the constant magnetic induction field. Imaging of the femoral chondyles can be problematic, because the bulk curvature of the cartilage alters the depth-dependent fibril orientation. A comprehensive in vivo study, however, has shown that the magic angle effect may not be the major determinant of T2 heterogeneity in high-curvature articular cartilage (17). The effect of curvature on the T2 value in cartilage is an area of ongoing research; therefore, we recognize that the location of focal defects could alter the mean T2 value determined with our method and lead to greater uncertainty in the determination of T2 values. Further characterization of T2 values around cartilage lesions is an area of future work.
While we found significant differences between T2 values in healthy and OA cartilage in three of the four cartilage compartments analyzed, we did not find significant differences between the patients with mild OA and those with severe OA. We did note trends of increased T2 values in the severe OA group compared to the mild OA group in the medial compartments, which typically experience higher weight bearing and are more likely to develop OA (22); however, the number of patients in these groups may not have been large enough to allow us to detect a significant difference.
Previous investigators have used more than two echoes to determine the T2 value; however, we chose this technique because of a number of limiting factors. The current study was part of a longitudinal study (11); therefore, imaging duration was an important limitation, because multiple imaging series were performed, and the entire examination was limited to 1
hours. Furthermore, due to the variable and focal nature of OA, complete coverage of the knee compartment was a primary requirement. Finally, the original examinations were performed in 1999, when the MR imaging hardware was limited in gradient speed and power; thus, obtaining a T2-weighted image with more than two echoes would increase the imaging time by at least a factor of two if complete knee coverage was to be maintained. Over the course of the study, updates to the MR imaging hardware were made that allowed for faster acquisition of multiple echo time images, but changes were not made to the acquisition parameters to maintain consistency for all examinations. Had hardware and imaging duration not prevented the implementation of a sequence with multiple echo times for this imaging, we may have had better precision in the measured T2 values, and differences between healthy and OA cartilage may have differed in magnitude. The model developed by MacFall et al (23) supports the decrease in variance of the measured T2 relaxation time with an increased number of echoes.
It is proposed that viewing T2 values normalized to the expected normal values could aid in the assessment of tissue heterogeneity and tracking of OA progression. Further quantification of the variation of T2 values through the cartilage volume could enhance the information available to assess disease progression. For example, Mosher et al (6) and Dardzinski et al (5) have shown the utility of line profiles—viewing intensity values as a function of distance along a specific line—of T2 values from the subchondral region to the articulating surface. MR signal heterogeneity across the thickness of cartilage through the major histologic zones (tangential, transitional, and deep radial) has been shown previously (24). In the present study, we attempted to use a z score conversion per pixel to allow direct visualization of the T2 heterogeneity present in the cartilage by highlighting differences from T2 values of healthy cartilage.
Future work could lead to use of this technique for better visualization of variation through the cartilage thickness and in weight-bearing regions. Distinct depiction of the chondral layers, however, was difficult and inconsistent in the present study. This was likely due to partial volume effects from insufficient spatial resolution that obscured the chondral layers. As opposed to the study of Frank et al (25), no custom imaging hardware (ie, a local gradient coil) was used to enhance the signal-to-noise ratio to allow clear demonstration of the cartilage layering due to collagen fibril orientation. Analyzing T2 relaxation time in weight-bearing regions on the basis of margins of the meniscus could prove useful; however, this was not attempted in the present study because we were unable to define the meniscus in several patients with severe meniscal damage.
The z score images of T2 values based on normal values for each cartilage compartment may be a useful indicator of disease, particularly for focal areas of T2 changes. High z scores were observed both in patients with mild OA and in patients with severe OA in the medial and lateral femoral compartments, which are areas of maximal mechanical loading, indicating that T2 relaxation time changes at early stages of OA. Viewing a z score conversion of the T2 values could enable better differentiation from healthy tissue when tracking or diagnosing OA. Studies of the longitudinal progression of OA are recommended, and T2 values and z scores may provide a measure of temporal disease development.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: OA = osteoarthritis, WOMAC = Western Ontario and McMaster University
Author contributions: Guarantors of integrity of entire study, S.M., T.C.D.; study concepts and design, S.M., T.C.D.; literature research, T.C.D.; experimental studies, S.M.; data acquisition, S.M.; data analysis/interpretation, all authors; statistical analysis, Y.L., H.J.; manuscript preparation, T.C.D.; manuscript definition of intellectual content, S.M., T.C.D.; manuscript editing, T.C.D.; manuscript revision/review and final version approval, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. H. Welsch, T. C. Mamisch, S. E. Domayer, R. Dorotka, F. Kutscha-Lissberg, S. Marlovits, L. M. White, and S. Trattnig Cartilage T2 Assessment at 3-T MR Imaging: In Vivo Differentiation of Normal Hyaline Cartilage from Reparative Tissue after Two Cartilage Repair Procedures--Initial Experience Radiology, April 1, 2008; 247(1): 154 - 161. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Saupe, C. W. A. Pfirrmann, M. R. Schmid, T. Schertler, M. Manestar, and D. Weishaupt MR Imaging of Cartilage in Cadaveric Wrists: Comparison between Imaging at 1.5 and 3.0 T and Gross Pathologic Inspection Radiology, April 1, 2007; 243(1): 180 - 187. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. S. Bauer, S. J. Krause, C. J. Ross, R. Krug, J. Carballido-Gamio, E. Ozhinsky, S. Majumdar, and T. M. Link Volumetric Cartilage Measurements of Porcine Knee at 1.5-T and 3.0-T MR Imaging: Evaluation of Precision and Accuracy. Radiology, November 1, 2006; 241(2): 399 - 406. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. M. White, M. S. Sussman, M. Hurtig, L. Probyn, G. Tomlinson, and R. Kandel Cartilage T2 Assessment: Differentiation of Normal Hyaline Cartilage and Reparative Tissue after Arthroscopic Cartilage Repair in Equine Subjects. Radiology, November 1, 2006; 241(2): 407 - 414. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. G. Potter and L. F. Foo Magnetic Resonance Imaging of Articular Cartilage: Trauma, Degeneration, and Repair Am. J. Sports Med., April 1, 2006; 34(4): 661 - 677. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
