Cartilage and meniscus assessment using T1rho and T2 measurements in healthy subjects and patients with osteoarthritis

Osteoarthritis and Cartilage 18 (2010) 1408e1416

Cartilage and meniscus assessment using T1rho and T2 measurements in healthy subjects and patients with osteoarthritis Z.A. Zarins y, R.I. Bolbos y, J.B. Pialat yz, T.M. Link y, X. Li y, R.B. Souza yx, S. Majumdar y * y Musculoskeletal and Quantitative Imaging Research Group, Department of Radiology and Biomedical Imaging, University of California, San Francisco, CA, USA z INSERM Research Unit 831, University of Lyon and Hospices Civils de Lyon, Lyon, France x Department of Physical Therapy and Rehabilitation Science, University of California, San Francisco, CA, USA

a r t i c l e i n f o

s u m m a r y

Article history: Received 24 February 2010 Accepted 20 July 2010

Objective: The purpose of this study was to evaluate meniscal degeneration in healthy subjects and subjects with osteoarthritis (OA) using T1r and T2 measurements and to examine the interrelationship between cartilage and meniscus abnormalities. Methods: Quantitative assessment of cartilage and meniscus was performed using 3T Magnetic Resonance Imaging (MRI) with a T1r and T2 mapping technique in 19 controls and 44 OA patients. A sagittal T2-weighted fast spin echo (FSE) fat-saturated image was acquired for cartilage and meniscal WholeOrgan Magnetic Resonance Imaging Score (WORMS) assessment. Western Ontario and McMasters Universities Arthritis Index (WOMAC) scores were obtained to assess clinical symptoms. Results: The posterior horn of the medial meniscus (PHMED) had the highest incidence of degeneration. Stratifying subjects on the basis of PHMED grade revealed that the T1r and the T2 measurements of the PHMED and the medial tibial (MT) cartilage were higher in subjects having a meniscal tear (meniscal grade 2e4) compared to subjects with a meniscal grade of 0 or 1 (P < 0.05). While not statistically significant, there was a trend for T1r and T2 being higher in PHMED grade 1 compared to grade 0 (P ¼ 0.094, P ¼ 0.073 respectively). WOMAC scores had a stronger correlation with meniscus relaxation measures than cartilage measures. Conclusions: Magnetic Resonance (MR) T1r and T2 measurements provide a non-invasive means of detecting and quantifying the severity of meniscal degeneration. Meniscal damage has been implicated in OA progression and is correlated with cartilage degeneration. Early detection of meniscal damage represented by elevations in meniscal relaxation measures may identify subjects at increased risk for OA. Ó 2010 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.

Keywords: Osteoarthritis Cartilage Meniscus T1r T2 Magnetic Resonance Imaging

Introduction Osteoarthritis (OA) is a heterogeneous disease characterized by the gradual loss of hyaline cartilage and is a whole-joint disorder involving the interaction between articular cartilage, subchondral bone, synovium, and the meniscus among others. The meniscus functions in joint stability, joint lubrication, shock absorption, and helps to maintain the integrity of the articular cartilage1,2. Previous studies have demonstrated that there is a high incidence of meniscal tears in patients with symptomatic arthritis as well as asymptomatic controls and that meniscal tears are associated with an increase in cartilage loss3e9. Quantitative measurements of T1r and T2 relaxation times of cartilage provide a non-invasive means of detecting early biochemical * Address correspondence and reprint requests to: Sharmila Majumdar, University of California San Francisco, Campus Box 2520, QB3 Building, 2nd Floor, Suite 203, 1700 e 4th Street, San Francisco, CA 94158. E-mail address: [email protected] (S. Majumdar).

changes in cartilage degeneration prior to morphological or clinical changes. A decline in the proteoglycan content, increase in the tissue water content, and alteration in the collagen content and orientation occur in cartilage during early OA and can be quantified using T1r and T2 relaxation times10,11. In addition, cartilage T1r and T2 relaxation times are correlated with the severity of OA in vitro and in vivo12e16. To our knowledge no study has investigated the relationship between T1r and T2 relaxation measures with the meniscal grades as assessed by a modified Whole-Organ Magnetic Resonance Imaging Score (WORMS)17. An increase in the intrameniscal signal or intra-substance tear is commonly seen in asymptomatic subjects and represents an early stage in meniscus degeneration culminating in a meniscal tear most commonly seen in the posterior horn of the medial meniscus (PHMED)18e20. The purpose of our study was three-fold: (1) To evaluate T1r and T2 measurements of cartilage and meniscus with respect to morphological meniscal grade, (2) To examine the correlation of cartilage and meniscus T1r and T2 with clinical symptoms as assessed with the Western Ontario and

1063-4584/$ e see front matter Ó 2010 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.joca.2010.07.012

Z.A. Zarins et al. / Osteoarthritis and Cartilage 18 (2010) 1408e1416

McMasters Universities Arthritis Index (WOMAC) score, and (3) To examine the interrelationship between measures of cartilage and meniscus degeneration. Materials and methods Subjects The patient recruitment for the study was a combination of referral by UCSF orthopaedic surgeons and recruitment from general public. The inclusion criteria for OA patients were frequent clinical symptoms of OA (including pain, stiffness and dysfunction) and demonstration of typical signs of OA in radiographs. The controls had no history of diagnosed OA, clinical OA symptoms, previous knee injuries, or signs of OA on radiographs. Standard standing anteroposterior radiographs of the knee were obtained in all subjects at baseline to determine the KellgrenLawrence (KL) grade and OA severity21. The 63 subjects (29 men, 34 women) that participated in this cross-sectional study had a mean age of 51 13.6 years and a mean body mass index (BMI) of 26.2  5.3 kg/m2. Of these subjects, 19 were classified as controls (KL ¼ 0, mean age ¼ 39  10 years), 26 were classified as mild OA (KL score ¼ 1 or 2, mean age ¼ 52  10 years), and 18 were classified as severe OA patients (KL score ¼ 3 or 4, mean age ¼ 62  10 years). The subject characteristics are presented in Table I. All of the subjects included in the study were in good health as assessed by a medical history and physical examination and had no contraindications to Magnetic Resonance (MR) imaging. All subjects completed the WOMAC questionnaire to assess pain, stiffness, and function of the knee joint through a five-point scale22. The study was approved by the local Institutional Review Board and conducted in accordance with the Committee for Human Research. All subjects gave written informed consent prior to participation in the study. Magnetic Resonance Imaging (MRI) protocol MRI of the knee was performed using a 3T GE Excite Sigma MR Scanner (General Electric, Milwaukee, WI, USA) and an eight-channel phased-array knee coil (Invivo, Orlando, FL, USA). In the OA subjects, the knee with more severe findings on the radiographs was imaged. In control subjects, the dominant leg was imaged. Parallel imaging was

1409

performed with an array spatial sensitivity technique (ASSET) using acceleration factor (AF) ¼ 2. Cartilage volume and thickness were assessed using a high spatial resolution volumetric fat-suppressed spoiled-gradient-echo (SPGR) sequence (repetition time TR/ TE ¼ 15/6.7 ms, flip angle ¼ 12, field of view (FOV) ¼ 14 cm, matrix ¼ 512  512, in-plane spatial resolution ¼ 0.273  0.273 mm2, slice thickness ¼ 1 mm, BW ¼ 31.25 kHz, number of excitations ¼ 0.75, acquisition time ¼ 7 m 37 s). A T2-weighted fat-saturated FSE sequence (TR/TE ¼ 4300/51 ms, FOV ¼ 14 cm, matrix ¼ 512  256, slice thickness (ST) ¼ 2.5 mm, gap ¼ 0.5 mm) was used to determine the clinical WORMS measurements for the cartilage and the meniscus. Sagittal three-dimensional (3D) T1r and T2 sequences were used to assess the cartilage and meniscus sub compartments. The T1r was obtained using a spin-lock technique followed by SPGR acquisition using transient signals evolving towards steady-state with the following parameters: TR/TE ¼ 9.3/3.7 ms, time of recovery ¼ 1500 ms, FOV ¼ 14 cm, matrix ¼ 256  192, slice thickness ¼ 3 mm, BW ¼ 31.25 kHz, views per segment ¼ 48, time of spin-lock (TSL) ¼ 0/10/40/80 ms, FSL ¼ 500 Hz, acquisition time of approximately 13 min23. The T2 quantification was also based on a magnetization preparation sequence with 3D SPGR using transient signals evolving towards steady-state24. The T2 preparation pulses contained an MLEV train of nonselective composite 90x180y90x refocusing pulses. The imaging parameters were the same as the T1r quantification except for magnetization preparation TE ¼ 3.1/13.5/ 23.9/44.8 ms. The T2 quantification sequence covered the same region as T1r quantification, and had an acquisition time of 11 min. MR imaging analysis Morphological analysis The clinical assessment of cartilage, meniscus, and ligaments was performed using a sagittal T2-weighted FSE fat-saturated image by two experienced radiologists (TML with 20 and JBP with 9 years of experience). The radiologists were blinded to subject information and relaxation data, and performed separate readings, with a consensus in the case of disagreement. Meniscal scoring was graded using a modified WORMS score of the knee as follows: 0 ¼ no lesion, 1 ¼ intra-substance abnormality, 2 ¼ non-displaced tear, 3 ¼ displaced or complex tear without deformity, and 4 ¼ maceration of the meniscus17. Meniscal scores two

Table I Subject Characteristics stratified by KL grade WOMAC Age (years)

Height (cm)

Weight (kg)

BMI

KL grade

Controls (n ¼ 19) Mean Standard deviation (SD)

39.10 10.15

172.5 9.09

70.79 12.66

23.47 3.44

0.00 0.00

Mild OA (n ¼ 26) Mean SD P-value (vs controls)

52.27* 10.30 <0.001

170.04 9.06

75.28 13.49

26.00 4.01 0.200

1.46* 0.51

Severe OA (n ¼ 18) Mean SD P-value (vs controls) P-value (vs mild OA) ANOVA Prob > F

62.61*, y 10.33 <0.001 0.005 <0.001

166.58 8.48

80.89 16.67

3.47*, y 0.52

0.137

0.106

29.33*, y 6.78 0.001 0.071 0.002

Patients (mild and severe) Mean SD P-value (vs controls)

56.50* 11.42 <0.001

168.62 8.89 0.120

77.57 14.95 0.089

27.36* 5.50 0.006

Pain

Stiffness

Function

3.26 3.26

1.63 1.67

6.79 8.98

3.19 3.14 0.998

1.88 1.24 0.842

10.42 11.11 0.551

10.72*, y 4.48 <0.001 <0.001 <0.001

4.67*, y 1.64 <0.001 <0.001 <0.001

6.27* 5.26 0.025

3.03* 1.97 0.009

35.55*, y 14.15 <0.001 <0.001 <0.001 20.70* 17.53 0.002

Controls are defined as subjects having a KL score of 0, Mild OA are defined as subjects with a KL score of 1 or 2, and Severe OA are defined as subjects having a KL score of 3 or 4. * Significantly different than control (P < 0.05). y Significantly different than mild OA (P < 0.05).

1410

Z.A. Zarins et al. / Osteoarthritis and Cartilage 18 (2010) 1408e1416

through four were collapsed into one category to represent the group with a meniscal tear. The meniscal WORMS sum was calculated as a sum of the meniscal WORMS scores for every compartment. Ligaments were graded using a modified WORMS score of the knee as follows: 0 ¼ no lesion, 1 ¼ signal abnormality around the ligament, 2 ¼ signal abnormality within the ligament, 3 ¼ partial tear, 4 ¼ complete tear17. The ligaments that were graded included the following: anterior cruciate ligament (ACL), posterior cruciate ligament (PCL), medial collateral ligament (MCL), and lateral collateral ligament (LCL). Quantitative assessment All MR images were transferred to a Sun Workstation (Sun Microsystems, Palo Alto, CA) for data processing and quantification of cartilage volume, cartilage thickness, and T1r and T2 relaxation measurements of cartilage and meniscus. Cartilage sub compartments were segmented semi-automatically in high resolution SPGR images using the in-house software developed with Matlab (Mathworks, Natick, MA, USA) based on edge detection and Bezier splines. Cartilage segmentation was performed by investigators with extensive experience in performing cartilage segmentation. The cartilage compartments analyzed included the following: lateral femoral condyle (LFC), medial femoral condyle (MFC), lateral tibia (LT), medial tibia (MT), and the patella (P). Cartilage thickness and volume measurements were determined using the in-house software developed with Matlab. Previous studies have shown that variations in joint size have a greater effect on cartilage volume than cartilage thickness25. Therefore, cartilage volume was normalized by dividing the cartilage volume by the epicondylar distance to account for differences in joint size amongst subjects. The meniscus compartments were segmented using a T1r image with TSL ¼ 40 ms, rigidly registered to the SPGR image using the VTK CISG Registration Toolkit. The meniscus compartments analyzed included the following: anterior horn lateral meniscus (AHLAT), anterior horn medial meniscus (AHMED), posterior horn lateral meniscus (PHLAT), and PHMED. The meniscal body was excluded because of partial volume effects. The anatomical landmarks defined for meniscus segmentation have been previously described in detail26. Additional studies from our laboratory utilizing the same techniques and protocols for cartilage and meniscus quantification have reported good reproducibility and precision26,27. Specifically, the CVs for the reproducibility for the T1r are: 4.6% for the AHLAT, 3.3% for the PHLAT, 3.7% for the AHMED, and 4.9% for the PHMED27. The CVs for the T2 are: 8.91% for the AHLAT, 10.57% for the PHLAT, 7.44% for the AHMED, and 9.35% for the PHMED26. T1r and T2 maps were reconstructed by fitting the T1r and T2 images pixel by pixel using a LevenbergeMarquardt mono-exponential fitting algorithm developed in-house using the equations below:


SðTSLÞfexp TSL=T1r for T1rho SðTEÞfexpðTE=T2 Þfor T2 For cartilage, all four T1r or T2-weighted images were used to reconstruct the maps. For meniscus, only the first three T1r or T2weighted images were used. T1r-weighted images with TSL ¼ 80 ms and T2-weighted images with TE ¼ 44.8 ms had a very low SNR (<5) for meniscus due to short T1r and T2 in meniscus respectively, and therefore were not used during map reconstruction. The reconstructed T1r and T2 maps were then rigidly registered to the SPGR image using VTK CISG Registration Toolkit28. 3D cartilage and meniscus contours after segmentation were overlaid to T1r and T2 maps. Mean T1r and T2 values were calculated in defined regions. T1r and T2 measurements were then compared between subjects with a meniscal grade of 0, 1, and those with a meniscal tear (grades 2e4) (Fig. 1).

Statistical analysis All statistical analyses were conducted using JMP software, version 8.0 (SAS Institute, Cary NC). All of the analyses were conducted using two-tailed tests and a P-value of less than 0.05 was considered statistically significant. Analysis of variance (ANOVA) was used to assess the differences between the subject characteristics and relaxation measurements by KL grade and meniscal grade. Post-hoc comparisons were made with Tukey Honestly Significant Difference Test (HSD), using an a of 0.05. In cases where the ANOVA test was statistically significant, the P-values for the Post-hoc analysis are reported. We did not make statistical adjustments for weight or BMI because there was no statistical difference in weight or BMI between the PHMED grades (P ¼ 0.198 for weight, P ¼ 0.099 for BMI). Spearman correlation coefficient measurements were conducted to assess the correlations of relaxation measurements of cartilage and meniscus compartments with WOMAC scores as well as the correlations of meniscal WORMS sum with WOMAC scores. Results Subject characteristics The number of subjects having a meniscal tear in a location other than the PHMED was very low. Therefore, because we did not have the statistical power to analyze the other meniscal compartments, we focused our analysis on the PHMED. Table II shows the subject characteristics stratified by three groups of PHMED grade based on meniscal WORMS assessment: meniscal grade 0 (n ¼ 23), meniscal grade 1 (n ¼ 17), and meniscal grades 2e4 (n ¼ 23). Subjects with meniscal grades 2e4 had statistically higher WOMAC scores compared to subjects with a meniscal grade 0 (P ¼ 0.005, P ¼ 0.005, and P¼ < 0.001 for pain, stiffness, and function respectively). WOMAC scores for meniscal grade 1 were not statistically different than grade 0 (P ¼ 0.511, 0.252, and 0.169 for pain, stiffness, and function respectively) or grade 2e4 (P ¼ 0.148, 0.347, 0.220 for pain, stiffness, and function respectively). There were no statistically significant differences in height, weight, or BMI between subjects stratified by meniscal grade (P ¼ 0.210, P ¼ 0.198, P ¼ 0.099 for height, weight, and BMI respectively). Meniscal assessment Table III illustrates the incidence of meniscal grade by KL group and by meniscal sub compartment. The overall incidence of meniscal tears was 16% in controls and 57% in OA patients. Among OA patients, 42% of mild OA and 78% of severe OA patients had a meniscal tear. The incidence of a meniscal tear was the highest in the PHMED regardless of KL group. Ligament assessment Control subjects had no ligament abnormalities. Among OA subjects, 20% had a signal abnormality around (WORMS ¼ 1) or within (WORMS ¼ 2) a ligament (13.6% ACL, 6.82% PCL), 2.27% had a partial ACL degenerative tear (WORMS ¼ 3), and 4.55% had a complete degenerative ACL tear (WORMS ¼ 4). Cartilage morphometry measurements Table IV shows the cartilage thickness and normalized volume measurements respectively of the medial femoro-tibial cartilage compartment in subjects stratified by PHMED grade. The only cartilage

Z.A. Zarins et al. / Osteoarthritis and Cartilage 18 (2010) 1408e1416

1411

Fig. 1. MR images showing the PHMED in a control subject with a PHMED WORMS grade of 0 (A, B, C), a mild OA subject with a PHMED WORMS grade of 1 (D, E, F), and a mild OA subject with a PHMED WORMS grade of 3 (G, H, I). The differences in the FSE images (A, D, G), T1rho (B, E, H), and T2 (C, F, I) can be seen clearly between the subjects with different meniscal grades. The color bar indicates the gradation of relaxation measures.

compartment that was statistically different between the meniscal grades was the MT. The MT cartilage thickness was 17% lower and the normalized volume was 24% lower in the meniscal grade 2e4 group compared with grade 0 (P ¼ 0.027, P ¼ 0.024 respectively). Table II Subject characteristics stratified by posterior horn medial meniscal grade WOMAC Age Height (years) (cm) Meniscus grade 0 (n ¼ 23) Mean 42.9 171 SD 10.5 7.23 Meniscus grade 1 (n ¼ 17) Mean 51.4* 167 SD 11.9 10.8 P-value 0.070 (vs grade 0) Meniscus grades 2e4 (n ¼ 23) Mean 59.5* 170 SD 12.8 9.07 P-value <0.001 (vs grade 0) P-value 0.090 (vs grade 1) ANOVA Prob > F <0.001

0.210

Weight BMI (kg)

Pain

Stiffness Function

72.5 10.7

24.4 3.46

3.30 3.08

1.69 1.49

73.7 16.5

26.5 5.94

4.94 2.65 4.94 1.73 0.511 0.252

16.76 17.0 0.169

79.8 15.8

27.7 5.87

7.74* 3.48* 5.53 2.23 0.005 0.005

25.0* 18.1 <0.001

0.148 0.347

0.220

0.099 0.007 0.008

0.001

Stratifying the data by PHMED grade produced similar trends for T1r and T2 relaxation. Tables V and VI show the T1r and T2 relaxation times respectively, in subjects stratified by PHMED grade for cartilage and the meniscus. The MT was the only cartilage compartment that was statistically different between meniscal grades, the T1r and T2 of meniscal grade 2e4 being higher than grade 0 (P ¼ 0.017 for T1r, P ¼ 0.001 for T2) and grade 1 (P ¼ 0.027 for T1r, P < 0.001 for T2). However, MT cartilage T1r and T2 relaxation times were not statistically different between meniscal grade 0 and 1 (P ¼ 0.998 for T1r, P ¼ 0.873 for T2). With respect to meniscal compartments, the meniscal grade 2e4 group had higher T1r and T2 relaxation times in the PHLAT (P ¼ 0.043 for T1r, P ¼ 0.035 for T2) and the PHMED (P < 0.001 for T1r and T2) compared to grade 0. The PHMED was the only compartment that had statistically different T1r and T2 times between more than one meniscal grade. Specifically, the T1r and the T2 values in the PHMED grade 2e4 were higher than grade 0 (P < 0.001 for T1r and T2) and grade 1 (P < 0.001 for T1r and T2). Furthermore, there was a trend for T1r and T2 being higher in PHMED grade 1 compared to grade 0 (P ¼ 0.094 for T1r, P ¼ 0.073 for T2). Correlation between T1rho and T2 with WOMAC scores

0.198

Significantly different than meniscal grade 0 (P < 0.05). y Significantly different than meniscal grade 1 (P < 0.05).

*

7.88 9.68

T1rho and T2 relaxation measurements stratified by PHMED grade

Correlations of cartilage and meniscus T1r and T2 relaxation with clinical WOMAC scores are shown in Table VII. WOMAC scores were

1412

Z.A. Zarins et al. / Osteoarthritis and Cartilage 18 (2010) 1408e1416

(P < 0.001 for all measures). These correlations were similar to the correlations of PHMED T2 with WOMAC scores.

Table III Meniscal grade by collapsed meniscal grading scheme Grade 0

Grade 1

Grades 2e4 collapsed

Discussion

Lateral meniscus Anterior horn Control (n ¼ 19) Mild OA (n ¼ 26) Severe OA (n ¼ 18) All subjects, %(n ¼ 63)

19 (100%) 19 (73.97%) 10 (55.55%) 76.19

0 (0%) 3 (11.54%) 3 (16.67%) 9.52

0 (0%) 4 (15.38%) 5 (27.78%) 14.29

Posterior horn Control (n ¼ 19) Mild OA (n ¼ 26) Severe OA (n ¼ 18) All subjects, % (n ¼ 63)

19 (100%) 19 (73.08%) 10 (55.56%) 76.19

0 (0%) 1 (3.85%) 4 (22.22%) 7.94

0 (0%) 6 (23.08%) 4 (22.22%) 15.87

Medial meniscus Anterior horn Control (n ¼ 19) Mild OA (n ¼ 26) Severe OA (n ¼ 18) All subjects, % (n ¼ 63)

18 (94.74%) 25 (96.15%) 12 (66.67%) 87.30

0 (0%) 1 (3.85%) 3 (16.67%) 6.35

1 (5.26%) 0 (0%) 3 (16.67%) 6.35

Posterior horn Control (n ¼ 19) Mild OA (n ¼ 26) Severe OA (n ¼ 18) All subjects, % (n ¼ 63)

12 (63.16%) 9 (34.61%) 2 (11.11%) 36.51

5 (26.31%) 9 (34.61%) 3 (16.67%) 26.98

2 (10.53%) 8 (30.77%) 13 (72.22%) 36.51

Meniscal grade by collapsed meniscal grading scheme. The table illustrates the distribution of subjects by meniscal grading schemes and KL grade. Each column shows the number of subjects assigned to the meniscal grade. The percentage in parenthesis represents the percentage of subjects in each KL grade that have a given meniscal grade. Controls are defined as subjects having a KL score of 0, Mild OA are defined as subjects with a KL score of 1 or 2, and Severe OA are defined as subjects having a KL score of 3 or 4.

more strongly correlated with the meniscus than cartilage, with three of four meniscal compartments, having significant correlations with the total WOMAC scores compared with only two of the five cartilage compartments. The PHMED had the strongest correlation with WOMAC scores. While both T1r and T2 meniscal measurements in the PHMED, AHLAT, and the PHLAT, demonstrated significant correlations with WOMAC scores, T2 consistently had a higher correlation than the T1r. The medial cartilage compartment (MFC and MT) demonstrated significant correlations between T1r and WOMAC pain, stiffness, and total score. This same relationship applied with regard to cartilage T2 and WOMAC scores in MT, but not in MFC. Correlations of meniscal WORMS sum with WOMAC scores were as follows: 0.437 for pain, 0.483 for stiffness, and 0.578 for function

In this cross-sectional study quantitative MRI was used to evaluate T1r and T2 relaxation measurements in subjects with various grades of meniscal degeneration. Our results indicate that quantitative MRI measurements of T1r and T2 can not only distinguish between subjects with a normal meniscus and those with a meniscal tear, but it can also distinguish between subjects with increased intra-substance signal and those with a meniscal tear. Correlation data reveal that clinical WOMAC scores are more closely related to T1r and T2 of the meniscus than T1r and T2 of cartilage.

Clinical findings Our finding of a 16% incidence of meniscal tears in controls and 57% incidence in the OA patients is similar to previous reports but differs from other studies involving an older subject population (65 vs 51 years)3,4,7. Because the incidence of meniscal tears increases with age, the difference in age between the study participants may account for the discrepancy20,29. Consistent with the findings of others we report a higher prevalence of meniscal abnormalities in the PHMED regardless of KL grade (Table III)20,29e31. The medial meniscus has a decreased mobility, larger diameter, and smaller thickness relative to other meniscal locations and thus may have an increased susceptibility of damage20. Furthermore, the highest weight-bearing pressure is in the medial compartment. In the normal state 60e80% of the total intrinsic compressive load is transmitted across the knee in the medial compartment32. In addition to the increased meniscal damage in the medial compartment, like others, we also report a significantly decreased cartilage thickness and volume in the MT in subjects with a PHMED tear (Table IV)5,6,31. These findings are all consistent with previous reports demonstrating that the medial compartment is the site of the most structural changes during the early course of OA33. Consistent with previous reports, we report that the ACL was the ligament with the highest incidence of abnormality34. However, since only 4.55% of patients had a complete ACL degenerative tear the focus of this investigation was on meniscal abnormalities.

Table IV Cartilage morphology measures stratified by PHMED grade Cartilage volume (mm3)

Cartilage thickness (mm)

Meniscus grade 0 (n ¼ 23) Mean SD Meniscus grade 1 (n ¼ 17) Mean SD P-value (vs grade 0) Meniscus grades 2e4 (n ¼ 23) Mean SD P-value (vs grade 0) P-value (vs grade 1) ANOVA Prob > F

LFC

MFC

LT

MT

LFC

MFC

LT

MT

1.50 0.264

1.48 0.334

1.95 0.351

1.35 0.276

503 145

365 136

253 88.7

182 53

1.53 0.317

1.49 0.279

1.74 0.533

1.29 0.280 0.797

449 152

353 118

216 92.7

159 56 0.426

1.59 0.355

1.30 0.376

1.70 0.470

1.12* 0.291 0.027 0.174

488 195

289 128

205 91.9

138* 54 0.024 0.477

0.608

0.106

0.148

0.029

Significantly different than meniscal grade 0 (P < 0.05). y Significantly different than meniscal grade 1 (P < 0.05).

*

0.620

0.123

0.191

0.031

Z.A. Zarins et al. / Osteoarthritis and Cartilage 18 (2010) 1408e1416

1413

Table V Cartilage and Meniscus T1rho stratified by posterior horn medial meniscal grade Cartilage T1rho (ms)

Meniscus grade 0 (n ¼ 23) Mean SD Meniscus grade 1 (n ¼ 17) Mean SD P-value (vs grade 0)

MFC

LT

MT

AHLAT

PHLAT

AHMED

PHMED

39.87 4.18

41.48 5.39

35.36 4.08

34.86 4.20

13.47 2.79

12.50 2.26

12.42 2.11

13.94 2.00

40.48 3.86

42.14 3.85

36.07 4.10

34.78 3.45 0.998

14.14 3.56 0.862

14.36 4.54 0.176

13.13 2.65

17.16 3.72 0.094

44.15 4.63

37.37 5.84

39.34*, y 7.23 0.017 0.027

16.44* 4.85 0.036 0.187

14.87* 2.85 0.043 0.878

13.73 3.09

23.73*,y 6.96 <0.001 <0.001

0.008

0.037

0.042

Meniscus grades 2e4 (n ¼ 23) Mean 41.99 SD 4.10 P-value (vs grade 0) P-value (vs grade 1) ANOVA Prob > F * y

Meniscus T1rho (ms)

LFC

0.204

0.151

0.365

0.256

<0.001

Significantly different than meniscal grade 0 (P<0.05). Significantly different than meniscal grade 1 (P<0.05).

T1rho and T2 measurements of cartilage and meniscus Both cartilage and meniscus consist of a macromolecular framework of collagen fibers, proteoglycan, and water, albeit in different concentrations. Articular cartilage is primarily composed of type II collagen and 5e10% proteoglycan, while the meniscus is composed of 98% type I collagen and only 1% proteoglycan27,35. In hyaline cartilage, T1r relaxation is inversely related to proteoglycan content while T2 relaxation is related to cartilage collagen orientation and water content10,36e38. The mechanisms of T1r and T2 relaxation in the meniscus are not fully understood, and therefore our understanding of the biochemical mechanism of T1r and T2 is based upon on the findings in hyaline cartilage. Subjects with a PHMED tear had higher T1r and T2 in the PHMED compared to subjects without meniscal damage. Histological studies indicate that articular cartilage and meniscal degeneration results in alterations in the matrix including the deterioration and disorientation of the collagen network, decline in the proteoglycan content, an increase in water content, and an infiltration of synovial fluid into the damaged location30,39. The decline in proteoglycan content decreases the resiliency of the cartilage and meniscus while the disorganization of the collagen network compromises the ability of the tissues to withstand the compressive forces. A meniscal tear may also contribute to altering the biomechanical properties of the articular cartilage and meniscus, altering joint

stability, and impairing balance40,41. Together these biochemical and biomechanical alterations may account for the elevations in T1r and T2 seen in the MT cartilage and the posterior meniscus in the presence of a PHMED tear. Our findings indicate that T1r and T2 measurements can distinguish between some of the different meniscal grades. Specifically, T1r and T2 of the PHMED could not only distinguish between subjects with and without a meniscal tear but could also distinguish between subjects with an intra-substance meniscal degeneration and those with a meniscal tear. The magnitude of T1r and T2 was directly related to the severity of the meniscal degeneration. While not statistically significant, there was a trend showing higher T1r and T2 in subjects with an intra-substance meniscal degeneration compared to subjects with a normal meniscus. The results of this study have important implications for early detection and intervention. The intra-substance meniscal abnormality represents an early degenerative change in the meniscus that predisposes individuals to further damage or progression to a meniscal tear19,39,42. Furthermore, meniscal tears are significantly associated with an increased progression of OA5e9. Therefore, being able to distinguish between meniscal grades using T1r and T2 provides a non-invasive means of detecting and monitoring the degenerative changes in the meniscus. Furthermore, because relaxation times provide a continuous variable of tissue properties,

Table VI Cartilage and Meniscus T2 stratified by posterior horn medial meniscal grade Cartilage T2 (ms)

Meniscus grade 0 (n ¼ 23) Mean SD Meniscus grade 1 (n ¼ 17) Mean SD P-value (vs grade 0)

MFC

LT

MT

AHLAT

PHLAT

AHMED

PHMED

32.63 2.94

32.14 3.35

27.73 4.53

28.14 4.03

10.78 2.40

10.42 1.45

11.18 1.88

11.26 1.51

33.24 2.61

33.34 2.43

27.87 4.99

27.41 2.72 0.873

11.96 3.40

12.06* 2.77 0.048

11.81 1.69

13.51 2.53 0.073

32.98 6.04

30.06 4.81

33.08*, y 5.82 0.001 <0.001

12.85 3.91

12.03* 2.12 0.035 1.00

12.36 2.09

17.89*, y 4.51 <0.001 <0.001

Meniscus grades 2e4 (n ¼ 23) Mean 33.14 SD 2.76 P-value (vs grade 0) P-value (vs grade 1) ANOVA Prob > F * y

Meniscus T2 (ms)

LFC

0.751

0.667

Significantly different than meniscal grade 0 (P < 0.05). Significantly different than meniscal grade 1 (P < 0.05).

0.197

<0.001

0.122

0.019

0.125

<0.001

1414

Z.A. Zarins et al. / Osteoarthritis and Cartilage 18 (2010) 1408e1416

Table VII Correlations of T1rho and T2 relaxation times [ms] with WOMAC scores Pain T1r

Stiffness T2

Function

T1r

T2

T1r

Total T2

T1r

T2

Cartilage LFC P-value MFC P-value LT P-value MT P-value P P-value

0.123 0.338 0.302* 0.016 0.028 0.825 0.331* 0.008 0.021 0.868

0.130 0.308 0.227 0.074 0.149 0.245 0.387* 0.002 0.227 0.079

0.249* 0.049 0.371* 0.003 0.148 0.242 0.304* 0.015 0.108 0.403

0.089 0.486 0.234 0.065 0.252* 0.047 0.371* 0.003 0.164 0.205

0.231 0.068 0.298* 0.018 0.031 0.807 0.270* 0.033 0.098 0.449

0.135 0.289 0.127 0.322 0.212 0.096 0.415* <0.001 0.153 0.239

0.216 0.089 0.314* 0.012 0.028 0.817 0.294* 0.019 0.085 0.512

0.134 0.293 0.161 0.206 0.208 0.102 0.417* <0.001 0.174 0.180

Meniscus AHMED P-value PHMED P-value AHLAT P-value PHLAT P-value

0.341 0.793 0.503* <0.001 0.319* 0.014 0.419* <0.001

0.259* 0.042 0.526* <0.001 0.3956* 0.002 0.414* <0.001

0.011 0.905 0.403* 0.001 0.296* 0.023 0.361* 0.004

0.142 0.269 0.404* 0.001 0.352* 0.007 0.325* 0.010

0.022 0.864 0.502* <0.001 0.256 0.050 0.419* <0.001

0.174 0.176 0.550* <0.001 0.208* 0.048 0.351* 0.005

0.018 0.891 0.501* <0.001 0.268* 0.040 0.422* <0.001

0.186 0.148 0.533* <0.001 0.305* 0.020 0.364* 0.004

*

Correlation coefficient significant at P < 0.05.

they may be used to objectively diagnose early meniscal degeneration prior to a diagnosis based on categorical meniscal grading. Quantitative MRI offers an advantage over arthroscopy in regards to meniscal assessment19,43. While arthroscopy can only examine the surface areas of the meniscus, MR imaging can examine the entire meniscus30. Furthermore, the PHMED is difficult to visualize using arthroscopy, accounting for the decreased accuracy of detection of meniscal tears19,30,43. In contrast, MRI can be used to accurately detect meniscal tears in the medial compartment. Lastly, while arthroscopy can only detect the presence of a meniscal tear, quantitative MR imaging provides the unique ability to detect the presence of early meniscal degeneration19. Early detection may allow clinicians to intervene early to minimize the need for arthroscopic surgery. Previous studies demonstrate that nonoperative treatment of patients with meniscal degeneration is effective in improving function44. Interrelationship between cartilage and meniscus The elevation in the T1r and T2 seen in the MT cartilage in subjects with a PHMED tear is consistent with our finding of a corresponding decrease in cartilage thickness in this region (Table IV). Previous studies investigating the effect of meniscal tears on cartilage loss have reported similar results31. The MT cartilage may have the greatest cartilage loss because it lies adjacent to the area of meniscal damage and withstands the greatest weight-bearing load45. Our finding of cartilage damage and loss occurring in a compartment adjacent to a damaged meniscal compartment underscores the interrelationship between cartilage and the meniscus.

strongly correlated with the meniscus than cartilage, and there was a direct relationship between the strength of the correlation and the incidence of meniscal tears. These findings may be explained by the fact that articular cartilage is avascular and aneural while the outer one-third of the meniscus is vascularized, contains nerves, and nociceptive fibers46,47. Comparing the correlation results of the T1r data to the T2 data, we found that cartilage T1r had a stronger correlation with WOMAC scores, while meniscal T2 had a stronger correlation with WOMAC scores (Table VII). The finding that T1r is a more sensitive indicator of clinical cartilage parameters than T2 is consistent with previous reports showing that T1r is a more sensitive indicator of cartilage degeneration than T216. One of the factors that may account for this finding is that T1r has a more dynamic range and is more sensitive at detecting changes in chondral defects than T248. Furthermore, the higher proteoglycan concentration in articular cartilage may explain the stronger relation between T1r and cartilage, while the higher collagen concentration and low proteoglycan content in the meniscus may account for the stronger relationship between T2 and the meniscus. Together these findings indicate that T1r may be more suitable at measuring the dynamic changes in articular cartilage, while T2 may be more appropriate for measuring the dynamic changes in the meniscus associated with OA progression. WOMAC correlations revealed that the WOMAC pain scores had a stronger correlation with the T1r and the T2 of the PHMED than the meniscal WORMS sum. These results indicate that relaxation measures of the PHMED may provide a more accurate assessment of pain measures than morphological meniscal parameters.

Correlation of T1rho and T2 the clinical WOMAC scores

Limitations

To our knowledge this is the first study to investigate the correlation between clinical WOMAC scores and T1r and T2 in cartilage and meniscus sub compartments. Looking at a similar subject population as ours, Rauscher et al. reported that the T1r and T2 measurements of the medial, lateral, and both menisci were correlated with WOMAC scores26. However, meniscal sub compartments and cartilage measurement correlations were not reported. Our findings indicate that WOMAC scores were more

One of the limitations is that we were unable to measure meniscal extrusion and joint alignment because we did not acquire coronal images or long limb radiographs. Both of these factors are associated with cartilage loss and OA progression and therefore this may be a confounding factor6,32,49. Another limitation is the relatively small sample size and the cross-sectional design. The results of the current study are exploratory and longitudinal studies using a larger cohort are needed to confirm these findings.

Z.A. Zarins et al. / Osteoarthritis and Cartilage 18 (2010) 1408e1416

In conclusion, we have shown that quantitative MRI measures of T1r and T2 offer a non-invasive means of early detection of cartilage and meniscus degeneration. Most importantly, because relaxation measures provide a continuous variable of tissue properties, this information may be used to accurately and objectively diagnose early meniscal degeneration prior to a diagnosis made using a categorical grading system. Author contributions All coauthors made significant contributions to the design of the study, critical revision, and approval of the final version of the manuscript. Obtaining funding: Majumdar. Collection and assembly of data and technical support: Zarins, Bolbos, Pialat, Link, Li, Souza. Dr. Zarins takes responsibility for the integrity of the work as a whole. ([email protected]). Conflict of interest The authors have no conflict of interest to disclose in regard to this research or manuscript.

10.

11.

12.

13.

14.

15.

Acknowledgments The authors thank D Kuo, J Schooler, and J Zuo for their technical support, and T. Munoz and M. Guan for their help in recruiting and consenting patients for the study. This study was supported by NIH RO1 AR46905.

16.

17.

References 1. Radin EL, Lamotte FD, Maquet P. Role of the menisci in the distribution of stress in the knee. Clinic Orthop Res 1984;185:290e4. 2. Hunter DJ. Insights from imaging on the epidemiology and pathophysiology of osteoarthritis. Radiol Clin North Am 2009;47:539e51. 3. Englund M, Guermazi A, Gale D, Hunter D, Aliabadi P, Clancy M, et al. Incidental meniscal findings on knee MRI in middle-aged and elderly persons. N Engl J Med 2008;359: 1108e15. 4. Bhattacharyya T, Gale D, Dewire P, Totterman S, Gale E, McLaughlin S, et al. The clinical importance of meniscal tears demonstrated by magnetic resonance imaging in osteoarthritis of the knee. J Bone Joint Surg Am 2003;85:4e9. 5. Ding C, Martel-Pelletier J, Pelletier J-P, Abram F, Raynauld J-P, Cicuttini F, et al. Meniscal tear as an osteoarthritis risk factor in a largely non-osteoarthritic cohort: a cross-sectional study. J Rheumatol 2007;34:776e84. 6. Berthiaume M-J, Raynuld J-P, Martel-Pelletier J, Labonte F, Beaudoin G, Boch D, et al. Meniscal tear and extrusion are strongly associated with progression of symptomatic knee osteoarthritis as assessed by quantitative magnetic resonance imaging. Ann Rheum Dis 2005;64:556e63. 7. Englund M, Guermazi A, Roemer F, Aliabadi P, Yang M, Lewis CE, et al. Meniscal tear in knees without surgery and the development of radiographic osteoarthritis among middleaged and elderly persons. The Multicenter osteoarthritis study. Arthritis Rheum 2009;60:831e9. 8. Biswal S, Hastie T, Andriacchi T, Bergman G, Dillingham M, Lang P. Risk factors for progressive cartilage loss in the knee. A longitudinal magnetic resonance imaging study in forty-three patients. Arthritis Rheum 2002;46:2884e92. 9. Hunter DJ, Zhang Y, Niu JB, Tu X, Amin S, Clancy M, et al. The association of meniscal pathologic changes with cartilage loss

18.

19.

20.

21. 22.

23.

24.

25.

26.

1415

in symptomatic knee osteoarthritis. Arthritis Rheum 2006;54:795e801. Akella SV, Regatte RR, Gougoutas AJ, Borthakur AB, Shapiro EM, Kneeland JB, et al. Proteoglycan-induced changes in T1r-relaxation of articular cartilage at 4T. Magn Reson Med Sci 2001;46:419e23. Duvvuri U, Kudchodkar S, Reddy R, Leigh JS. T1r relaxation can assess longitudinal proteoglycan loss from articular cartilage in vitro. Osteoarthritis and Cartilage 2002;10:838e44. David-Vaudey E, Ghosh S, Ries M, Majumdar S. T2 relaxation time measurements in osteoarthritis. Magn Reson Imaging 2004;22:673e82. Regatte RR, Akella S, Lonner JH, Kneeland JB, Reddy R. T1r relaxation mapping in human osteoarthritis (OA) cartilage: comparison of T1r with T2. J Magn Reson Imaging 2006;23:547e53. Dunn TC, Lu Y, Jin H, Ries M, Majumdar S. T2 relaxation time of cartilage at MR imaging: comparison with severity of knee osteoarthritis. Radiology 2004;232:592e8. Regatte RR, Akella SVS, Wheaton AJ, Lech G, Borthakur A, Kneeland JB, et al. 3D-T1r-relaxation mapping of articular cartilage: in vivo assessment of early degenerative changes in symptomatic osteoarthritic subjects. Acad Radiol 2004;11:741e9. Li X, Ma B, Link TM, Castillo DD, Blumenkrantz G, Lozano J, et al. In vivo T1r and T2 mapping of articular cartilage in osteoarthritis of the knee using 3 T MRI. Osteoarthritis and Cartilage 2007;789e97. Peterfy CG, Guermazi A, Zaim S, Tirman PFJ, Miaux Y, White D, et al. Whole-organ magnetic resonance imaging score (WORMS) of the knee in osteoarthritis. Osteoarthritis and Cartilage 2004;12:177e90. LaPrade RF, Burnett QM, Veenstra MA, Hodgman CG. The prevalence of abnormal magnetic resonance imaging findings in asymptomatic knees. With correlation of magnetic resonance imaging to arthroscopic findings in symptomatic knees. Am J Sports Med 1994;22:739e45. Stoller DW, Martin C, Crues JV, Kaplan LK, Mink JH. Meniscal tears: pathologic correlation with MR imaging. Radiology 1987;163:731e5. Kornick J, Trefelner E, McCarthy S, Lange R, Lynch K, Jokl P. Meniscal abnormalities in the asymptomatic population at MR imaging. Radiology 1990;177:463e5. Kellgren JH, Lawrence JS. Radiological assessment of osteoarthritis. Ann Rheum Dis 1957;16:494e502. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol 1988;15:1833e40. Li X, Han ET, Busse RF, Majumdar S. In vivo T1r mapping in cartilage using 3D magnetization-prepared angle-modulated partitioned k-space spoiled gradient echo snapshots (3D MAPSS). Magn Reson Med 2008;59:298e307. Pai A, Li X, Majumdar S. A comparative study at 3 T of sequence dependence of T2 quantification in the knee. Magn Reson Imaging 2008;26:1215e20. Faber SC, Eckstein F, Lukasz S, Lukasz S, Muhlbauer R, Hohe J, et al. Gender differences in knee joint cartilage thickness, volume, and articular surface areas: assessment with quantitative three-dimensional MR imaging. Skeletal Radiol 2001;30:144e50. Rauscher I, Stahl R, Cheng J, Li X, Huber M, Luke A, et al. Meniscal measurements of T1r and T2 at MR imaging in healthy subjects and patients with osteoarthritis. Radiology 2008;249:591e600.

1416

Z.A. Zarins et al. / Osteoarthritis and Cartilage 18 (2010) 1408e1416

27. Bolbos RI, Link TM, Ma B, Majumdar S, Li X. T1r relaxation time of the meniscus and its relationship with T1r of adjacent cartilage in knees with acute ACL injuries at 3T. Osteoarthritis and Cartilage 2009;17:12e8. 28. Rueckert D, Sonoda LI, Hayes C, Hill DL, Leach MO, Hawkes DJ. Nonrigid registration using free-form deformations: application to breast MR images. IEEE Trans Med Imaging 1999;18:712e21. 29. Fukuta S, Masaki K, Korai F. Prevalence of abnormal findings in magnetic resonance images of asymptomatic knees. J Orthop Sci 2002;7:287e91. 30. Reicher MA, Hartzman S, Duckwiler GR, Bassett LW, Anderson LJ, Gold RH. Meniscal injuries: detection using MR imaging. Radiology 1986;159:753e7. 31. Crema MD, Guermazi A, Li L, Nogueira-Barbosa MH, Marra MD, Roemer FW, et al. The association of prevalent medial meniscal pathology with cartilage loss in the medial tibiofemoral compartment over a two year period. Osteoarthritis and Cartilage 2010;18:336e43. 32. Cerejo R, Dunlop DD, Cahue S, Chanin D, Song J, Sharma L. The influence of alignment on risk of knee osteoarthritis progression according to baseline stage of disease. Arthritis Rheum 2002;46:2632e6. 33. Raynauld JP, Martel-Pelletier J, Berthiaume MJ, Labonte F, Beaudoin G, de Guise JA, et al. Quantitative magnetic resonance imaging evaluation of knee osteoarthritis progression over two years and correlation with clinical symptoms and radiographic changes. Arthritis Rheum 2004;50:476e87. 34. Griffin LY, Agel J, Albohm MJ, Arendt EA, Dick RW, Garrett WE, et al. Noncontact anterior cruciate ligament injuries: risk factors and prevention strategies. J Am Acad Orthop Surg 2000;3:141e50. 35. Englund M. The role of the meniscus in osteoarthritis genesis. Rheum Dis Clin North Am 2008;34:573e9. 36. Duvvuri U, Reddy R, Patel SD, Kaufman JH, Kneeland JB, Leigh JS. T1r-relaxation in articular cartilage: effects of enzymatic degradation. Magn Reson Med 1997;6:863e7. 37. Xia Y. Magic-angle effect in magnetic resonance imaging of articular cartilage. Invest Radiol 2000;35:602e21.

38. Liess C, Lusse S, Karger N, Heller M, Gluer CC. Detection of changes in cartilage water content using MRI T2-mapping in vivo. Osteoarthritis and Cartilage 2002;10:907e13. 39. Hajek PC, Glys-Morin VM, Baker LL, Sartoris DJ, Haghighi P, Resnick D. The high signal intensity meniscus of the knee: magnetic resonance evaluation and in vivo correlation. Invest Radiol 1987;22:883e90. 40. Lange AK, Singh MA, Smith RM, Foroughi N, Baker MK, Shnier R, et al. Degenerative meniscus tears and mobility impairment in women with knee osteoarthritis. Osteoarthritis and Cartilage 2007;15:701e8. 41. Jones RS, Keene GCR, Learmonth DJA, Bickerstaff D, Nawana NS, Costi JJ, et al. Direct measurements of hoop strains in the intact and torn human medial meniscus. Clin Biomech 1996;11:295e300. 42. Noble J, Hamblen DL. The pathology of the degenerate meniscus lesion. J Bone Joint Surg 1975;57B:180e6. 43. Von Engelhardt LV, Schmitz A, Pennekamp PH, Schild HH, Wirtz DC, von Falkenhausen F. Diagnostics of degenerative meniscal tears at 3-Tesla MRI compared to arthroscopy as reference standard. Arch Orthop Trauma Surg 2008;128:451e6. 44. Rimington T, Mallik K, Evans D, Mroczek K, Reider B. A prospective study of the nonoperative treatment of degenerative meniscus tears. Orthopedics 2009; Aug;32:8. 45. Hsu RW, Himeno S, Coventry MB, Chao EYS. Normal axis alignment of the lower extremity and load-bearing distribution at the knee. Clin Orthop 1990;255:215e27. 46. Mine T, Kimura M, Sakka A. Innervation of nociceptors in the menisci of the knee joint: an immunohistochemical study. Arch Orthop Trauma Surg 2000;120:201e4. 47. Day B, Mackenzie WG, Shim SS, Leung G. The vascular and nerve supply of the human meniscus. Arthroscopy 1985;1:58e62. 48. Burnstein D, Gray M, Mosher T, Dardzinski B. Measures of molecular composition and structure in osteoarthritis. Radiol Clin North Am 2009;675e86. 49. Adams JG, McAlindon T, Dimasi M, Carey J, Eustace S. Contribution of meniscal extrusion and cartilage loss to joint space narrowing in osteoarthritis. Clin Radiol 1999;54:502e6.