- Email: [email protected]
No loss of cartilage volume over three years in patients with knee osteoarthritis as assessed by magnetic resonance imaging
Osteoarthritis and Cartilage (2002) 10, 929–937 © 2002 OsteoArthritis Research Society International. Published by Elsevier Science Ltd. All rights reserved. doi:10.1053/joca.2002.0849, available online at http://www.idealibrary.com on
1063–4584/02/$35.00/0
International Cartilage Repair Society
No loss of cartilage volume over three years in patients with knee osteoarthritis as assessed by magnetic resonance imaging S. J. Gandy*†**, P. A. Dieppe‡, M. C. Keen*, R. A. Maciewicz§, I. Watt¶ and J. C. Waterton\ *Department of Medical Physics & Bioengineering, Bristol General Hospital, Guinea St, Bristol BS1 6SY, U.K. †Rheumatology Unit, University of Bristol Division of Medicine, Bristol Royal Infirmary, BS2 8HW, U.K. ‡MRC Health Services Research Collaboration, Canynge Hall, Whiteladies Rd, Bristol, BS8 2PR, U.K. §Respiratory and Inflammation Research and \Enabling Science & Technology, AstraZeneca, Alderley Park, Macclesfield, SK10 4TG, U.K. ¶Department of Clinical Radiology, Bristol Royal Infirmary, Bristol, BS2 8HW, U.K. **Current address: Medical Physics Department, Ninewells Hospital, Dundee, DD1 9SY, U.K. Summary Objective: Magnetic resonance imaging (MRI) has the potential to provide accurate quantification of structural changes in joint disease, with sensitivity to change, as it can provide direct visualization of the cartilage and bone. In this study, we investigated whether knee cartilage volume, as assessed by MRI, is sensitive to change over time in patients with osteoarthritis (OA). Design: Sixteen patient volunteers (10 male, six female) with established OA of the knee were entered into the study and demographic data recorded. At baseline, 12 months and 37±2 months, patients underwent simple measures of disease severity, as well as extended weight-bearing AP knee X-rays. In addition the patient’s index knee was imaged using MR at 1.0T using a 3-D spoiled gradient–echo sequence with fat-suppression, repetition time 50 ms, echo time 11 ms, flip-angle 40°, sagittal slice thickness 1.56 mm and in-plane pixel resolution 0.55 mm. Manual image segmentation was performed on all knee cartilage compartments and the respective cartilage volumes determined. Results: Eleven of the original patients recruited completed the 3-year study. Radiographic features indicated that the majority had a spectrum of well-established OA at entry. The average decrease in medial tibiofemoral joint space width was 0.21±0.37mm (mean± S.D.). Comparison of MR images at baseline and 37±2 months indicated little evidence of cartilage lesion shape or size change in any of the compartments. There was no significant MRI volume change in any of the knee cartilage compartments over the course of 1 year. The change in total knee cartilage volume, as measured by MRI, was a loss of only 1.6%, or 0.36±1.3 ml (mean± S.D.), over the 3 years. Conclusions: The failure to identify loss of cartilage volume over 3 years in this cohort of patients with established knee OA using MRI challenges the face validity of this endpoint to assess structural changes in OA. © 2002 OsteoArthritis Research Society International. Published by Elsevier Science Ltd. All rights reserved. Key words: MRI, Cartilage, Osteoarthritis, Progression. Abbreviations: MR, magnetic resonance; MRI, magnetic resonance imaging; OA, osteoarthritis; 2-D, two-dimensional; 3-D, threedimensional; TR, repetition time; TE, echo time; HAQ, health assessment questionnaire; T1, longitudinal relaxation time; AP, anteriorposterior; JSN, joint space narrowing; JSW, joint space width; CoV, coefficient of variation; ROI, region of interest; ACR, American College of Rheumatology; K&L, Kellgren and Lawrence; BMI, body mass index; PFJ, patello–femoral joint; TFJ, tibio–femoral joint.
value1,2; and imaging techniques, notably X-radiography and MRI3,4. The knee joint usually is used to try to demonstrate such changes because it is a large joint that is easy to image and commonly is affected by OA. Current guidelines suggest that the gold standard for the assessment of structural change in the knee joint should be medial tibiofemoral joint space narrowing (JSN) measured from standardized X-ray films5,6. JSN is used as an indirect measure of cartilage loss, one of the main pathological features of OA. Other major changes also take place in the marginal and subchondral bone, which can be visualized on radiographs. Some authors have suggested that these should be measured in addition to JSN3, but this is not usually done. Unlike radiography, MRI can provide direct visualization of the hyaline cartilage (as well as the meniscus and bone) and has the potential to provide accurate quantification with sensitivity to change4,7. In the 1990s,
Introduction There is an increasing number of interventions that may slow the progression of osteoarthritis (OA)1. However, research has been hampered by the lack of accurate methods for the in vivo assessment of changes in articular cartilage. Techniques under investigation include: arthroscopy, an invasive procedure; molecular markers, which have been not yet led to any measurement of clinical
Received 22 February 2002; accepted 27 August 2002. Grant sponsors: Medical Research Council, AstraZeneca, United Bristol Healthcare NHS Trust Address correspondence to: Dr J. C. Waterton, AstraZeneca, Alderley Park, Macclesfield, Cheshire, SK10 4TG, U.K. Tel.: +44(0)1625-513633; Fax: +44(0)1625-512126; E-mail: [email protected]
929
930
S. J. Gandy et al.: Lack of cartilage volume change in knee OA
MR techniques were introduced which allowed quantification of cartilage volume in the component compartments of the knee joint4,8–15 and because of the face validity of such a measure in OA it has been suggested that this should become a technique of choice to assess OA in the future16–28. However, few data have been published to substantiate the concept that cartilage volume is sensitive to change over time in OA27,28. This is the subject of this communication, which reports changes in total cartilage volume in 11 patients with knee OA, observed over a 3-year period.
Method PATIENTS
Following local ethical committee approval, 16 informed and consenting patient volunteers were recruited. These patient volunteers had established OA of the knee and were under secondary care in a specialized Rheumatology Department. Criteria for entry to the study included: current use-related pain in the index knee to be studied; crepitus in that knee; age >40, radiographic evidence of OA; and willingness to participate in the study. Exclusion criteria included: severe joint damage, defined as Kellgren and Lawrence (K&L) grade 4 radiographic changes29; and being on the waiting list for knee surgery. Demographic data, e.g. age, sex, disease duration, pain severity, body mass index (BMI) and duration of morning stiffness, were recorded at baseline. Both at baseline and each subsequent visit two simple measures of disease severity were also recorded, Health Assessment Questionnaire (HAQ), and presence of any severe joint deformity. STUDY DESIGN
Each patient was imaged by MRI at baseline and at 4±2, 12 and 37±2 months. To assess the reproducibility of the technique, two substudies were performed: • Substudy 1 addressed the reproducibility of analysis of the same image data, either with the same segmenter or with different segmenters. One 3-D image dataset from each of six different patient volunteers was analysed twice by Segmenter A and once by Segmenter B. • Substudy 2 addressed the reproducibility of images collected during two different imaging sessions and analysed by the same segmenter. 3-D images were collected 2 months apart from each of five different patient volunteers and segmented once by Segmenter A. This provides a conservative assessment, since the measurement includes not only the ‘true’ reproducibility, but also any underlying disease progression. All MRI scans were performed during the middle of the day on each occasion. Knee X-rays were acquired at baseline, 12 months and 3 years (37±2 months). ANTHROPOMORPHIC PHANTOM
An anthropomorphic cartilage phantom of the femur was constructed by molding a sheet of extruded polyvinylchloride directly around the femoral condyles of a male skeleton. A 3 mm layer of stone plaster was then smoothed around the femoral condyles and a second mold was taken,
forming the outer layer of the phantom. The layers, together with three small spacers, were secured using silicone rubber adhesive, and the phantom was filled with a known volume of water, doped with paramagnetic ions in order to bring the longitudinal relaxation time T1 into the physiologic range.
IMAGING
The knee X-rays were extended weight-bearing AP views. MR imaging was performed on a 1.0T Siemens ‘Impact’ clinical scanner using a Siemens circularly polarized extremity coil, using a 3-D spoiled gradient–echo sequence with fat-suppression. The sequence parameters used were TR/TE=50/11 ms (40° flip-angle), generating 64 sagittal image slices each 1.56 mm thick. The field-of-view was 140 mm, and each image was acquired with 192 phase-encode steps and displayed on a 256×256 matrix to provide an in-plane pixel resolution of 0.55 mm. The imaging time was restricted to approximately 10 min in order to minimize patient movement artefact.
QUALITATIVE IMAGE ANALYSIS
The baseline and 3-year X-ray films were read by two observers working together (PAD and IW), and blind to date order of the films. The radiographs were read for K&L grade29 and for size of tibiofemoral osteophytes using the 0–3 scale and Altman atlas for standards30. Any other major abnormalities or apparent differences between the two films were also noted. In a similar manner, MR images were blinded to date order and reported by a consultant radiologist (IW). Qualitative assessment of images was undertaken describing the main sites for cartilage remodeling (cartilage thickness), bony remodeling (e.g. subchondral cysts or osteophytes) and accumulation of joint effusion. Any other major abnormalities or apparent differences between the two images were also noted.
QUANTITATIVE IMAGE ANALYSIS
Joint space narrowing (JSN) was measured at the midpoint of the medial tibiofemoral compartment using the graticule method as described by Lequesne31. The precision of the X-ray measurement in this centre has previously been established32–34. An image analysis package (‘Tosca’, IBM, Winchester, U.K.) was used to segment the MR images for all four hyaline cartilage compartments (femoral, medial tibial, lateral tibial and patellar). In previous studies10 on healthy joints, a semi-automatic image segmentation technique, in which the segmenter placed within the cartilage a seed point invoking a region-growing algorithm, was employed. However, pilot studies of segmentations in this OA population proved this semi-automated approach unreliable. Consequently fully manual segmentation procedure was employed35. Manual segmentation was performed by two medical physicists familiar with MR images of the knee (SJG: Segmenter A and MCK: Segmenter B) under the supervision of a musculoskeletal radiologist (IW). Any cartilage associated with osteophyte formation was excluded from the segmentations. In studies of the effect of image quality on reproducibility in osteo- and rheumatoid arthritis (SJG, MCK, unpublished), there was no trend for worse
Osteoarthritis and Cartilage Vol. 10, No. 12 image quality (sum of movement artefact or subjectively determined ease of segmentation) to correlate with an increased intra- or intersegmenter coefficient of variation. To avoid partial volume effects, only slices with a clearly defined articulating surface were segmented. To aid observer interpretation of partial volume effects, a 3-D model of the knee joint was provided for anatomical reference (‘Somso NS50’, Adam Rouilly, Sittingbourne U.K.). Femoral, patellar, medial tibial and lateral tibial cartilage segmentations were generated by enclosing all cartilage pixels within a region-of-interest for each 2-D image slice. Following careful checking and realignment of the ROI contour where necessary, the process was repeated for each image slice until a cartilage compartment volume had been fully segmented. The process of segmenting an entire knee (four compartment segmentations), with thorough checking, took approximately 4–5 h. The volume of each segmentation was obtained by finding the area of each 2-D ROI and multiplying by the slice thickness over all slices. The Tosca package uses a curve-fitting algorithm described by Akima36 in order to interpolate a contour defining each 2-D ROI. This spline interpolation uses only local information to determine the slope of a curve at each point, and it has been argued that such an interpolation is closer to a manually drawn curve than those produced by other splines. The area of each enclosed ROI was obtained by interpolation using this spline and then re-sampling the interpolated curve with 1000 equally spaced vertices to produce a dense polygon. The area of a polygon with n vertices (xi, yi) {i=0 . . . n−1} is then given (CJ Taylor, AD Brett, unpublished) by:
In addition to the quantitative volume calculation, 3-D volume images were reconstructed and displayed using the image-processing package ‘DX data explorer’ (IBM, Winchester, U.K.).
STATISTICAL ANALYSIS
All data are expressed as mean±standard deviation. Differences in JSW and volumes were evaluated using the two-sided Student’s t-test. P values of less than 0.05 were considered to be significant. Errors in measuring cartilage volumes may arise from many different experimental or physiological sources. Such errors can be assessed either through a single experiment with nested analysis of variance in which all possible sources of error are systematically varied, or through a series of experiments with pairwise comparison in which each possible source of error is varied in turn. The latter approach was adopted in this work. The reproducibility of each pairwise comparison was assessed from the test– retest coefficient of variation (CoV). For each subject, i, the CoV is the standard deviation, i, for the two measurements on that subject, divided by the mean volume, i, for the subject. The overall test–retest CoV for a group of N subjects is then
Use of the CoV is most appropriate when i is proportional to i. If i is independent of i, it is appropriate to
931 compare i values directly. In MRI studies in osteo- and rheumatoid arthritis (JCW, SJG, unpublished) it has been found that i increases with i (but with an approximately two-thirds, not linear, power relationship, consistent with the hypothesis that the errors in this case depend strongly on the surface area characteristics of the segmentation, since the surface area tends to increase with the two-thirds power of volume). Hence the CoV was employed. Point estimates of CoV are imprecise unless very large numbers of patients are sampled: in this case it is also helpful to quote the 95% confidence upper bound which can be derived37 from:
Results CLINICAL
Sixteen patients were recruited to the study (10 male, six female). They had a mean age of 63.4 years (range 52–70) and mean disease duration of 10.3 years (range 4–20). Only 11 were available for the final assessment at three years: two died, one lost his leg in trauma and two were lost to follow-up. This report concentrates on the 11 patients in whom full assessments were available at baseline and 3 years. There were five men and six women in this group, their mean age was 61.3 years and mean disease duration was 9.2 years. All fulfilled ACR clinical criteria for knee OA38. In addition all were over 40 years of age, had less than 30 min of morning stiffness, and had current knee pain and crepitus. Seven of the 11 patients had a HAQ score of >1.5 at entry to the study. There were no major changes in HAQ or deformities over the 3 years of the study. Table I shows some clinical and radiographic features of the 11 patients who completed the study. It is apparent that the majority had well-established radiographic changes of OA at entry. Four patients had K&L grade 2 and five patients had grade 3, although one set of radiographs was reported as ‘almost normal’. The mean medial joint space width at baseline of 4.9±1.3 mm suggests that the disease was not very advanced in this compartment in the group as a whole. Two patients had predominant lateral compartment disease, and one had predominant patello–femoral changes. There was little change over the 3 years (JSW at 3 years was 4.7±1.6 mm), although there was an average decrease (not statistically significant) in medial JSW of 0.21±0.37 mm, and four patients had JSN of 0.3 mm or more over the 3 years. GENERAL MRI OBSERVATIONS
All subjects gave analysable MRI data: no datasets were excluded from quantitative analysis. Cartilage volumes tended to be larger in males than in females. Segmentation was found to be particularly difficult when dealing with regions of very thin cartilage, where the ‘band structure’ was lost or altered, regions of ‘merged’ cartilage between two articulating surfaces (e.g. central regions of the tibiofemoral joint, and the base of the patellofemoral joint), regions of cartilage above and surrounding bony abnormalities, such as erosions, femoral cartilage at the base of the intercondylar notch, and also in other weight-bearing areas; and cartilage associated with (central) ‘mushroom’
932
S. J. Gandy et al.: Lack of cartilage volume change in knee OA Table I Clinical and radiographic features of patients who completed the 3 year study
Patient ID
04 13 21 24 26 28 29 34 35 41 44
Clinical (baseline)
Radiographic features (baseline)
Radiographic features (three years)
Age
Sex
BMI
K&L Grade
Medial JSW (mm)
Grade of osteophyte
JSW (mm)
Change in osteophyte
Change in JSW (mm)
Other features
67 52 71 63 70 56 65 62 54 62 52
F M M F F M F F F M M
30 28 31 29 23 28 32 33 33 22 23
3 2 1 2 2 1 3 3 3 3 2
2.1 6.0 6.0 5.7 4.4 5.3 5.9 5.5 3.1 4.1 6.0
2 2 1 1 2 1 2 2 1 1 2
1.5 5.6 6.0 5.5 4.6 5.0 5.8 5.5 2.1 3.9 6.3
0 0 0 0 0 0 0 −1 0 0 +1
−0.6 −0.4 0.0 −0.2 +0.2 −0.3 −0.1 0.0 −1.0 −0.2 +0.3
— — Enthesophytes PFJ disease Chondrocalcinosis Lateral disease Almost normal Lateral disease — — Attrition Lateral TFJ Chondrocalcinosis
PFJ: patello-femoral joint; TFJ: tibio-femoral joint.
Table II Coefficients of variation associated with repeated measurements of cartilage volume in each compartment
Same scan, same segmenter Same scan, different segmenter Different scan, different session, same segmenter
Substudy
N
Femoral
Patellar
Medial tibial
Lateral tibial
Total
1 1
6 6
2.0 (4.9) 7.8 (19)
5.5 (13) 5.9 (14)
3.3 (8.1) 14 (35)
2.8 (6.9) 5.7 (14)
1.6 (3.9) 4.7 (11)
2
5
2.0 (5.7)
5.3 (15)
4.6 (13)
6.7 (19)
1.8 (5.2)
In each case, the first figure is the rms CoV and the second (in parentheses) is the 95% confidence upper bound.
osteophytes. All image sets showed at least one area of focal cartilage loss, and were characterized by focal regions of cartilage thinning and/or loss in all compartments. This was particularly evident in the femoral and tibial cartilage. Other common features were peripheral osteophytes on the femur and patella, and mushroom osteophytes mainly on the femur. VALIDATION AND REPRODUCIBILITY
The segmented phantom volume was 20.1 ml compared with the actual volume of 20.5 ml (a 2.2% underestimation using MRI). Table II shows the CoV for three elements of reproducibility: intrasegmenter, intersegmenter, and scan– scan. In comparison with the first segmentation by segmenter A, total cartilage volume was 1.4% higher at the second segmentation of the same image by segmenter A, and 4.3% lower when the same image was segmented by segmenter B. These differences were not statistically significant. In the study of scan–scan variation, total cartilage volume was 0.3% lower in the second scan in comparison with the first scan. This tiny difference was not statistically significant. PROGRESSION OVER 37 MONTHS
Figure 1 shows the intercondyle MR images at 0 and 37 months of patient 35 who had joint space narrowing of 1 mm over the 3-year period. These images were representative of those obtained in this study. The radiologist’s
report on the MRI scans described cartilage lesions in virtually all of the 3-D segmented knee compartments examined. These can be clearly observed both in the original MRI images and in the reconstructed 3-D image of the MRI scans (see Fig. 2 for 3D reconstruction of MR images from patient 35 at 0 and 37 months). In this patient population lesions on the femur were most commonly seen at the anterior aspect of the medial condyle, the intercondylar notch and the middle aspect of the lateral condyle. Lesions were generally located centrally on the patella, and at various sites on both tibiae (more commonly on the lateral tibia). The radiologist’s report noted both increases and decreases in lesion size between baseline and 3 years. The radiologist’s report also noted that regions of cartilage thinning in one or more compartments were clearly visible on the 3-D MRI data of four patients. In contrast to these observations, regions of cartilage thickening over the 3 years were identified on the 3-D MRI data of two patients. The corresponding X-rays from these patients showed no JSW changes over time. In three patients there was no visible evidence of temporal change to the cartilage or joint-space width on X-ray or MRI. Figure 3 shows total cartilage volumes at 0, 1 and 3 years for all patients who participated in the study. Total cartilage volumes were 15.5±1.6 ml (range of 12.3 to 16.6 ml) for females and were 21.4±4.5 ml (range of 14.7 to 29.9 ml) for males. There was no significant change in total cartilage volume over the course of 3 years in this patient cohort. Nor was there any significant change to any
Osteoarthritis and Cartilage Vol. 10, No. 12
933
Fig. 1. Examples of magnetic resonance (MR) images. Lateral condyle region of patient 35 at baseline (left) and 37 months.
Fig. 2. Examples of segmentations. 3-D reconstructed MR image of cartilage for patient 35 at baseline (left) and 37 months (right). During this period, joint space width increased by 0.2 mm, while cartilage volume changes were −0.19 ml in the femur, +0.21 ml in the patella and −0.06 ml in the tibia.
of the individual knee cartilage compartments (femur, lateral tibia, medial tibia, patella) as shown in Table III. Volume changes (loss or gain) did not even occur at a rate comparable to the same segmenter/same scan CoV of 1.8%. MRI measured a mean loss of only 1.6% or 0.36±1.3 ml (mean± S.D.) in the knee over 3 years. Neither this change, nor the changes in individual compartments or at intermediate timepoints, were statistically significant
Fig. 3. Changes in cartilage volume. Total cartilage volume at 0, 1 and 3 years in all patients (males: open symbols; females: filled symbols). The symbol at the right represents ±1.8%, i.e. ±1 S.D. for intrasegmenter reproducibility (same segmenter, different scan35).
Although the average cartilage volume did not change, there were some large changes in individual patients. In five of the 11 patients, the change was more than 5.5% (increase or decrease), more than three standard deviations (i.e. three times the intraobserver CoV).
934
S. J. Gandy et al.: Lack of cartilage volume change in knee OA Table III Changes in cartilage volume in each compartment over three years Compartment
Femoral Lateral tibial Medial tibial Patellar Total
Volume (S.D.) at baseline/ml (N=16) 12.1 (2.9) 2.3 (0.5) 2.1 (0.6) 2.6 (1.2) 19.2 (4.7)
Mean (S.D.) percentage change from volume at baseline At 4.6±1.9 months (N=16) +0.2 (4.5) −0.4 (6.6) +2.4 (4.8) −0.4 (7.3) +0.3 (3.4)
P>0.2 P>0.2 P=0.05 P>0.2 P>0.2
Discussion Methods to quantify cartilage volume from MRI have been available for over 10 years8. A number of publications exist which show the method is highly reproducible and reflects cartilage volume measured directly from postoperative or cadaveric samples9,12,39. Our data show a very low CoV of the method, comparable to that reported from OA patients in other centres18–20,24. Given the reproducibility of our method, in this study we have >80% power to detect a net change (two-sided =0.05) of 1%/year for the total cartilage or the femoral compartment, 2%/year for the medial tibial compartment, or 3%/year in the lateral tibial or patellar compartment. We therefore have confidence in our ability to measure cartilage volume in the knee joint. Using this technique we have found that there was no significant loss of total cartilage volume in 11 patients with knee OA studied over a 3-year period. These data appear to challenge the face validity for the use of total cartilage volume to assess structural changes in OA. The study was performed in a typical clinical research setting using a readily available hospital scanner, a standard RF coil, and standard radiographic procedures in a hospital radiology department. The protocol has been implemented in a multicentre trial setting21. The design of any study of chronic disease progression using rapidly developing technology raises methodological issues, since the methods will almost inevitably appear outdated by the time the study is completed. The duration of the present study from inception to manuscript preparation was about 5 years, and in that time, significant developments in MRI methodology have occurred. The present study used a field strength of 1.0 T. We have shown that with this knee protocol in a multicentre study, that data collected at 1.0 T and 1.5 T are comparable21, but no doubt the instruments operating at 3 T, now in routine clinical use, will offer better sensitivity and resolution, and improved statistical power for the detection of tiny morphologic change. Improvements in gradient technology and pulse sequence design now allow shorter TR values, permitting higher resolution in a given scanning time20 and shorter TE values, providing better discrimination of cartilage from other tissues14. Improved image analysis approaches10, focusing for example only on weight-bearing areas or only on regions with established lesions, may provide improved sensitivity to disease progression. Examination of weight-bearing regions from coronal sections has been advocated14. Additionally, better understanding of the disease itself may allow better identification of patients at greatest risk of rapid cartilage loss, such as those with obesity. The patients included in this study were a highly selected group being recruited from a specialist clinic and undergoing secondary care for their knee OA. Such groups of
At 11.9±0.4 months (N=15) −0.1 (4.8) −0.1 (6.9) +2.9 (7.3) +0.8 (9.7) −0.2 (3.4)
P>0.2 P>0.2 P=0.06 P>0.2 P>0.2
At 37.2±2.2 months (N=11) −2.4 (7.3) +3.0 (8.4) −1.3 (12.0) −2.5 (14.3) −1.6 (6.0)
P>0.2 P>0.2 P>0.2 P=0.13 P>0.2
patients generally represent more advanced OA than that seen in the community and are arguably more likely to progress than unselected groups. However, we deliberately excluded those with the most advanced OA, and the radiographic data shown indicate that this group of patients did not have very severe structural changes at entry to the study. The demography of the group indicates that they were reasonably representative of other hospital based patient cohorts, although there were rather more men than would be expected. In general, these are the sorts of patients who are enrolled to many studies of progression or interventions for knee OA. There is no doubt that our group had OA; as they all fitted the ACR clinical criteria, and all had some radiographic changes; however, in view of the nature of the selection and the relatively small numbers we fully accept that this group may not be representative and the results are not necessarily generally applicable. The generally accepted means of assessing the presence of joint changes consistent with OA is through plain X-radiography, and in the case of the knee, the weightbearing AP view as used in this study. Recently it has been shown that the semi-flexed view is more sensitive to JSN than the standard film (with full knee extension) as used in this study40, but again, the views reported here are those that have been used in most other studies of knee OA. All our patients had osteophytes, four had clear narrowing of the medial joint space width at entry (<4.5 mm), two others had obvious lateral compartment disease and one patello– femoral compartment changes, suggesting that the majority had some cartilage loss at entry to the study. We therefore felt confident that the majority had cartilage pathology as well as bone changes (osteophytes) and the clinical as well as radiographic criteria for entry to an OA study30,40,41. Mean JSN at 3 years showed an overall loss of only 0.23 mm, which is consistent with an annual rate of change of only 0.08 mm/annum. This is less than the rate of change reported in most studies, which is nearer 0.2 mm/ annum3,42. Previous studies have suggested that cartilage loss with time may be faster in people who have already lost a significant amount of joint space width43, Table I suggests that this may be apparent in our cases as well, in that the two people who showed the greatest change in medial joint space width over 3 years each had relatively low joint space width at entry. Using the techniques of joint space width and radiographic views described, it has been calculated that the standard error of the measurement mean for medial joint space width is about 0.30 mm41. Four of our patients showed medial JSN of 0.3 mm or more over the 3 years, which could be regarded as significant progression of cartilage damage43, in addition, three of the 11 who were reviewed after 3 years had predominant lateral
Osteoarthritis and Cartilage Vol. 10, No. 12 compartment disease; in two of these cases we recorded an increase of medial compartment joint space width due to the ‘tilting’ of the knee associated with progression of lateral compartment disease. Therefore, although the mean rate of change was low, we believe that at least six of the 11 patients had conventional and significant radiographic evidence of total cartilage loss over the 3 years, which we would expect to be able to document with quantitative MRI techniques. All our patients had lesions (areas of thin or absent cartilage), which are never seen in normals. Thus, it appears that our patients had already embarked on a course of disease progression with considerable cartilage loss before entering the study. However, our data show small variations in total cartilage volumes over the 3-year study duration, with no individual patient showing much loss. There are several possible reasons for the lack of change in total cartilage volume of knee joint (measured from MRI) in the face of disease progression in OA. The most obvious explanation is that OA is a focal disease and cartilage change is usually concentrated on small areas of the joint subjected to maximal loading. Assessment of total cartilage volumes will dilute any change in these areas. A second explanation is offered by data from other studies using histology, MRI or arthroscopy, which have shown that some parts of the articular cartilage increase in volume due to excess hydration in the early phases of OA. It is quite possible that progression of OA in whole joints will result in thickening of cartilage in some areas and loss of cartilage volume in others resulting in no measurable change in total cartilage volume. Specifically, progression of relatively advance lesions in one compartment might be accompanied by earlier changes in swelling of the cartilage in another compartment. In support of this, Table III shows that while there was no net change in volume from baseline, there was a firm trend to increasing S.D. with increasing time, implying that some patients showed increased cartilage while others show a decrease. A third possible explanation is that loss of articular cartilage in the group of patients studied was occurring too slowly to be detected even after 3 years. However, the change in the radiographs and visual findings on MR suggest that cartilage lesions were progressing in many of the patients over the 3 years, but in spite of that no individual showed any significant change in total volumes measured from MRI. It should be noted, that the change in JSN may reflect not cartilage loss, but meniscal extrusion during weightbearing, as suggested by recent MRI studies44. One major difference in the way in which X-ray and MRI image the joint is that the former is done with weight bearing and the latter without. If one of the major causes for cartilage thinning in OA is increased deformability on weight-bearing then one might get a major discrepancy between volume measurements on MRI and JSN assessed on weight-bearing X-rays. However, conventional pathological studies of OA suggests that the major loss of structural elements of the cartilage occurs in the focal areas that are most affected.
Conclusions Given these data and the arguments in the literature reviewed above, we suggest that total cartilage volumes will not prove as useful as hoped in the assessments of structural change in OA. However, we do believe that MRI is the preferred technique for the quantification of cartilage
935 damage. Future quantitative work should concentrate on the assessments of focal volume areas, thickness mapping10,45 or quality change46 rather than the volume measurement of the whole joint.
Acknowledgments The authors would like to acknowledge Bristol University Rheumatology Unit staff, including Dr John Kirwan, Dr Margaret Byron, Dr Jon Tobias and Research Sisters Hilary Brownett, Kathryn Mitchell and Trudy Sheen for their help in patient recruitment and care; all the MRI Radiography staff at Bristol Royal Infirmary for imaging the patients; Dr Alan Brett at the University of Manchester (presently at ImageMetrics) for deriving and implementing the calculation of area from ROI, and Dr John Foster, presently at Western Royal Infirmary, Glasgow for early contribution to the work.
References 1. Felson D, Hochberg M, McAlindon T, Dieppe P, Minor M, Blair S, et al. Osteoarthritis: new insights—Part 2: Treatment approaches. Ann Int Med 2000;133: 726–37. 2. Felson D, Lawrence R, Dieppe P, Hirsch R, Helmick C, Jordan J, et al. Osteoarthritis: New Insights—Part 1: The disease and its risk factors. Ann Int Med 2000;133:635–46. 3. Mazzuca S, Brandt K, Katz B. Is conventional radiography suitable for evaluation of a disease-modifying drug in patients with knee osteoarthritis? Osteoarthritis Cart 1997;5:217–26. 4. Peterfy CG, Van Dijke CF, Janzen DL, Gluer CC, Namba R, Majumdar S, et al. Quantification of articular cartilage in the knee with pulsed saturation transfer subtraction and fat-suppressed MR imaging: Optimization and validation. Radiology 1994;192: 485–91. 5. Lequesne M, Brandt K, Bellamy N, Moskowitz R, Menkes J, Pelletier J, et al. Guidelines for testing slow-acting and disease modifying drugs in osteoarthritis. J Rheumatol 1994;21(suppl 41): 65–71. 6. GREES—osteoarthritis section. Recommendations for the registration of drugs used in the treatment of osteoarthritis. Ann Rheum Dis 1996;55:552–7. 7. Konig H, Sauter R, Deimling M, Vogt M. Cartilage disorders: Comparison of spin–echo, CHESS, and FLASH sequence MR images. Radiology 1987;164: 753–8. 8. Paul PK, Wang JZ, Mezrich RS, Rakhit A, Dunton AW, Furst D, et al. 3D-MRI: a novel approach to quantitation of articular cartilage. Arthritis Rheum 1990;33: S91. 9. Eckstein F, Sittek H, Milz S, Putz R, Reiser M. The morphology of articular cartilage assessed by magnetic resonance imaging. Surg Radiol Anat 1994;16: 429–38. 10. Waterton JC, Solloway S, Foster JE, Keen MC, Gandy S, Middleton BJ, et al. Diurnal variation in the articular cartilage of the knee in young adult humans. Magn Reson Med 2000;43:126–32. 11. Eckstein F, Schnier M, Haubner M, Priebsch J, Glaser C, Englmeier KH, et al. Accuracy of cartilage volume and thickness measurements with magnetic resonance imaging. Clin Ortho Rel Res 1998;352: 137–48.
936
S. J. Gandy et al.: Lack of cartilage volume change in knee OA
12. Peterfy CG, Van Dijke CF, Lu Y, Nguyen A, Connick TJ, Kneeland JB, et al. Quantification of the volume of articular cartilage in the metacarpophalangeal joints of the hand—accuracy and precision of 3-dimensional MR-imaging. Am J Roent 1995;165: 371–5. 13. Eckstein F, Reiser M, Englmeier K-H, Putz R. In vivo morphometry and functional analysis of human articular cartilage with quantitative MRI—from image to data, from data to theory. Anat Embryol 2001;203:147–73. 14. Hyhlik-Du¨rr A, Faber S, Burgkart R, Stammberger T, Maag K-P, Englmeier K-H, et al. Precision of tibial cartilage morphometry with a coronal waterexcitation MR sequence. Eur Radiol 2000;10: 297–303. 15. Hardy PA, Newmark R, Liu YM, Meier D, Norris S, Piraino DW, et al. The influence of the resolution and contrast on measuring the articular cartilage volume in MRI. Magn Reson Imag 2000;18:965–72. 16. Foster JE, Dieppe P, Maciewicz R, Taberner J, Watt I, Waterton JC. Quantification of cartilage volume and visualisation of defects in osteoarthritis using a clinical MR system. Arthritis Rheum 1996;39:S170. 17. Cicuttini F, Wluka AE, Stuckey SL. Tibial and femoral cartilage changes in knee arthritis. Ann Rheum Dis 2001;60:977–80. 18. Lynch JA, Zaim S, Zhao J, Stork A, Peterfy CG, Genant HK. Cartilage segmentation of 3D MRI scans of the osteoarthritic knee combining user knowledge and active contours. Proc Soc Photo-Optical Instrument Engineers 2000;3979:925–35. 19. Cicuttini F, Forbes A, Asbeutah A, Morris K, Stuckey S. Comparison and reproducibility of fast and conventional spoiled gradient-echo MR sequences in the determination of knee cartilage volume. J Orthop Res 2000;18:580–4. 20. Burgkart R, Glaser C, Hyhlik-Du¨rr, Englmeier K-H, Reiser M, Eckstein F. MRI-based assessment of cartilage loss in severe osteoarthritis. Arthritis Rheum 2001;44:2072–77. 21. Moots RJ, Morgan SR, Nash AFP, Waterton JC, Maciewicz RA, Farebrother JE, et al. MRI measurement of knee cartilage volume in a multicentre study. Rheumatology 2002;41(suppl 1): 123. 22. Wild M, Jacobi V, Windolf J. The quantitative detection of posttraumatic osteoarthrosis after fractures of the tibial plateau using MRI-based chondrovolumetry. Eur J Trauma 2000;26:15–24. 23. Glisson M, Forbes A, Morris K, Stuckey S, Cicuttini F. Comparison of X-rays and magnetic resonance imaging in the definition of tibiofemoral joint osteoarthritis. Radiography 2000;6:205–209. 24. Steines D, Cheng C, Wong A, Berger F, Napel S, Lang P. Segmentation of osteoarthritic femoral cartilage from MRI. In: Lemke HU, Inamura K, Doi K, Vannier MW, Farmer AG, Eds. Computer-assisted Radiology and Surgery. The Netherlands: Elsevier 2000:578– 83. 25. Boegard TL, Rudling O, Petersson IF, Jonsson K. MRI of the knee in chronic knee pain. A 2-year follow-up. Osteoarthritis Cart 2001;9:473–80. 26. Pessis EG, Drape J, Aryal X, Auleley G, Dougados M, Chevrot A. Quantification with MRI of chondral lesions of the medial tibiofemoral compartment: longitudinal study with MRI and arthroscopic followup. Radiology 1998;209P:414.
27. Peterfy C, White D, Zhao J, VanDijke C, Genant HK. Longitudinal measurement of knee cartilage volume in osteoarthritis. Arthritis Rheum 1998;41:S361. 28. Raynauld JP, Kauffman C, Godbout B, Beaudoin G, Berthiaume MJ, DeGuise J, et al. Knee osteoarthritis progression evaluated by magnetic resonance imaging and a novel quantification software tool. Osteoarthritis Cart 2000;8(b): S82. 29. Kellgren JH, Lawrence JS. Radiographic assessment of osteoarthritis. Ann Rheum Dis 1957;16:494–502. 30. Altman R, Hochberg M, Murphy WD, Wolfe F, Lequesne M. Atlas of individual radiographic features in osteoarthritis. Osteoarthritis Cart 1995;3(suppl A): 3–70. 31. Lequesne M. Quantitative measurements of joint space during progression of osteoarthritis: chondrometry. In: Kuettner K, Goldberg V, Eds. Osteoarthritic Disorders. Rosemont: American Academy of Orthopaedic Surgeons 1995:427–44. 32. Dieppe PA, Cushnaghan J, Young P, Kirwan J. Prediction of the progression of joint space narrowing in osteoarthritis of the knee by bone scintigraphy. Ann Rheum Dis 1993;52:557–63. 33. Dieppe PA, Cushaghan J, Shepstone L. The Bristol ‘OA 500’ study: progression of osteoarthritis (OA) over 3 years and the relationship between clinical and radiographic changes at the knee joint. Osteoarthritis Cart 1997;5:85–97. 34. Mazzuca SA, Brandt KD, Dieppe PA, Doherty M, Katz BP, Lane KA. Effect of alignment of the medial tibial plateau and X-ray beam on apparent progression of osteoarthritis in the standing anterioposterior knee radiograph. Arthritis Rheum 2001;44:1786–94. 35. Gandy SJ, Dieppe P, Watt I, Brett AD, Keen MC, Maciewicz RA, et al. Reproducibility of knee cartilage volume measurements in arthritis patients. Proc Soc Magn Reson Med Ninth Scientific Meeting and Exhibition, Glasgow, UK 2001; 2098. 36. Akima H. A new method of interpolation and smooth curve fitting based on local procedures. J Assoc Computing Machin 1974;17:589–602. 37. Johnson RA, Bhattacharyya GK. Statistics Principles and Methods, 4th ed. New York: Wiley 2001: 397–401. 38. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al. Development of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Committee of the American Rheumatism Association. Arthritis Rheum 1986;29: 1039–1049. 39. Eckstein F, Sittek H, Milz S, Schulte E, Kiefer B, Reiser M. The potential for magnetic resonance imaging (MRI) for quantifying articular cartilage thickness—a methodological study. Clin Biomech 1995;10: 434–40. 40. Dieppe P, Brandt K, Lohmander S, Felson D. Detecting and measuring modification in osteoarthritis: the need for standardized methodology. J Rheumatol 1995;22:201–3. 41. Buckland-Wright J, Macfarlane D, Williams S, Ward J. Accuracy and precision of joint space width measurements in standard and macroradiographs of osteoarthritic knees. Ann Rheum Dis 1995;54:872–80. 42. Mazzuca S. Plain radiographic evaluation of knee osteoarthritis. Curr Opini Rheumatol 1997;9:263–7.
Osteoarthritis and Cartilage Vol. 10, No. 12 43. Dougados M, Gueguen A, Nguyen M, Thiesce A, Listrat V, Jacob L, et al. Longitudinal radiographic evaluation of osteoarthritis of the knee. J Rheumatol 1992;19:378–84. 44. Gale DR, Chaisson CE, Totterman SM, Schwartz RK, Gale ME, Felson D. Meniscal subluxation: association with osteoarthritis and joint space narrowing. Osteoarthritis Cart 1999;7:526–532.
937 45. Eckstein F, Gavazzeni A, Sittek H, Haubner M, Lo¨sch A, Milz S, et al. Determination of knee joint cartilage thickness using three-dimensional magnetic resonance chondro-crassometry (3D MR-CCM). Magn Reson Med 1996;36:256–65. 46. Frank LR, Wong, EC, Buxton RB, Resnick D. Mapping the physiological parameters of articular cartilage with MRI. Topics in Magn Reson Imaging 1999;10:153–79.
1063–4584/02/$35.00/0
International Cartilage Repair Society
No loss of cartilage volume over three years in patients with knee osteoarthritis as assessed by magnetic resonance imaging S. J. Gandy*†**, P. A. Dieppe‡, M. C. Keen*, R. A. Maciewicz§, I. Watt¶ and J. C. Waterton\ *Department of Medical Physics & Bioengineering, Bristol General Hospital, Guinea St, Bristol BS1 6SY, U.K. †Rheumatology Unit, University of Bristol Division of Medicine, Bristol Royal Infirmary, BS2 8HW, U.K. ‡MRC Health Services Research Collaboration, Canynge Hall, Whiteladies Rd, Bristol, BS8 2PR, U.K. §Respiratory and Inflammation Research and \Enabling Science & Technology, AstraZeneca, Alderley Park, Macclesfield, SK10 4TG, U.K. ¶Department of Clinical Radiology, Bristol Royal Infirmary, Bristol, BS2 8HW, U.K. **Current address: Medical Physics Department, Ninewells Hospital, Dundee, DD1 9SY, U.K. Summary Objective: Magnetic resonance imaging (MRI) has the potential to provide accurate quantification of structural changes in joint disease, with sensitivity to change, as it can provide direct visualization of the cartilage and bone. In this study, we investigated whether knee cartilage volume, as assessed by MRI, is sensitive to change over time in patients with osteoarthritis (OA). Design: Sixteen patient volunteers (10 male, six female) with established OA of the knee were entered into the study and demographic data recorded. At baseline, 12 months and 37±2 months, patients underwent simple measures of disease severity, as well as extended weight-bearing AP knee X-rays. In addition the patient’s index knee was imaged using MR at 1.0T using a 3-D spoiled gradient–echo sequence with fat-suppression, repetition time 50 ms, echo time 11 ms, flip-angle 40°, sagittal slice thickness 1.56 mm and in-plane pixel resolution 0.55 mm. Manual image segmentation was performed on all knee cartilage compartments and the respective cartilage volumes determined. Results: Eleven of the original patients recruited completed the 3-year study. Radiographic features indicated that the majority had a spectrum of well-established OA at entry. The average decrease in medial tibiofemoral joint space width was 0.21±0.37mm (mean± S.D.). Comparison of MR images at baseline and 37±2 months indicated little evidence of cartilage lesion shape or size change in any of the compartments. There was no significant MRI volume change in any of the knee cartilage compartments over the course of 1 year. The change in total knee cartilage volume, as measured by MRI, was a loss of only 1.6%, or 0.36±1.3 ml (mean± S.D.), over the 3 years. Conclusions: The failure to identify loss of cartilage volume over 3 years in this cohort of patients with established knee OA using MRI challenges the face validity of this endpoint to assess structural changes in OA. © 2002 OsteoArthritis Research Society International. Published by Elsevier Science Ltd. All rights reserved. Key words: MRI, Cartilage, Osteoarthritis, Progression. Abbreviations: MR, magnetic resonance; MRI, magnetic resonance imaging; OA, osteoarthritis; 2-D, two-dimensional; 3-D, threedimensional; TR, repetition time; TE, echo time; HAQ, health assessment questionnaire; T1, longitudinal relaxation time; AP, anteriorposterior; JSN, joint space narrowing; JSW, joint space width; CoV, coefficient of variation; ROI, region of interest; ACR, American College of Rheumatology; K&L, Kellgren and Lawrence; BMI, body mass index; PFJ, patello–femoral joint; TFJ, tibio–femoral joint.
value1,2; and imaging techniques, notably X-radiography and MRI3,4. The knee joint usually is used to try to demonstrate such changes because it is a large joint that is easy to image and commonly is affected by OA. Current guidelines suggest that the gold standard for the assessment of structural change in the knee joint should be medial tibiofemoral joint space narrowing (JSN) measured from standardized X-ray films5,6. JSN is used as an indirect measure of cartilage loss, one of the main pathological features of OA. Other major changes also take place in the marginal and subchondral bone, which can be visualized on radiographs. Some authors have suggested that these should be measured in addition to JSN3, but this is not usually done. Unlike radiography, MRI can provide direct visualization of the hyaline cartilage (as well as the meniscus and bone) and has the potential to provide accurate quantification with sensitivity to change4,7. In the 1990s,
Introduction There is an increasing number of interventions that may slow the progression of osteoarthritis (OA)1. However, research has been hampered by the lack of accurate methods for the in vivo assessment of changes in articular cartilage. Techniques under investigation include: arthroscopy, an invasive procedure; molecular markers, which have been not yet led to any measurement of clinical
Received 22 February 2002; accepted 27 August 2002. Grant sponsors: Medical Research Council, AstraZeneca, United Bristol Healthcare NHS Trust Address correspondence to: Dr J. C. Waterton, AstraZeneca, Alderley Park, Macclesfield, Cheshire, SK10 4TG, U.K. Tel.: +44(0)1625-513633; Fax: +44(0)1625-512126; E-mail: [email protected]
929
930
S. J. Gandy et al.: Lack of cartilage volume change in knee OA
MR techniques were introduced which allowed quantification of cartilage volume in the component compartments of the knee joint4,8–15 and because of the face validity of such a measure in OA it has been suggested that this should become a technique of choice to assess OA in the future16–28. However, few data have been published to substantiate the concept that cartilage volume is sensitive to change over time in OA27,28. This is the subject of this communication, which reports changes in total cartilage volume in 11 patients with knee OA, observed over a 3-year period.
Method PATIENTS
Following local ethical committee approval, 16 informed and consenting patient volunteers were recruited. These patient volunteers had established OA of the knee and were under secondary care in a specialized Rheumatology Department. Criteria for entry to the study included: current use-related pain in the index knee to be studied; crepitus in that knee; age >40, radiographic evidence of OA; and willingness to participate in the study. Exclusion criteria included: severe joint damage, defined as Kellgren and Lawrence (K&L) grade 4 radiographic changes29; and being on the waiting list for knee surgery. Demographic data, e.g. age, sex, disease duration, pain severity, body mass index (BMI) and duration of morning stiffness, were recorded at baseline. Both at baseline and each subsequent visit two simple measures of disease severity were also recorded, Health Assessment Questionnaire (HAQ), and presence of any severe joint deformity. STUDY DESIGN
Each patient was imaged by MRI at baseline and at 4±2, 12 and 37±2 months. To assess the reproducibility of the technique, two substudies were performed: • Substudy 1 addressed the reproducibility of analysis of the same image data, either with the same segmenter or with different segmenters. One 3-D image dataset from each of six different patient volunteers was analysed twice by Segmenter A and once by Segmenter B. • Substudy 2 addressed the reproducibility of images collected during two different imaging sessions and analysed by the same segmenter. 3-D images were collected 2 months apart from each of five different patient volunteers and segmented once by Segmenter A. This provides a conservative assessment, since the measurement includes not only the ‘true’ reproducibility, but also any underlying disease progression. All MRI scans were performed during the middle of the day on each occasion. Knee X-rays were acquired at baseline, 12 months and 3 years (37±2 months). ANTHROPOMORPHIC PHANTOM
An anthropomorphic cartilage phantom of the femur was constructed by molding a sheet of extruded polyvinylchloride directly around the femoral condyles of a male skeleton. A 3 mm layer of stone plaster was then smoothed around the femoral condyles and a second mold was taken,
forming the outer layer of the phantom. The layers, together with three small spacers, were secured using silicone rubber adhesive, and the phantom was filled with a known volume of water, doped with paramagnetic ions in order to bring the longitudinal relaxation time T1 into the physiologic range.
IMAGING
The knee X-rays were extended weight-bearing AP views. MR imaging was performed on a 1.0T Siemens ‘Impact’ clinical scanner using a Siemens circularly polarized extremity coil, using a 3-D spoiled gradient–echo sequence with fat-suppression. The sequence parameters used were TR/TE=50/11 ms (40° flip-angle), generating 64 sagittal image slices each 1.56 mm thick. The field-of-view was 140 mm, and each image was acquired with 192 phase-encode steps and displayed on a 256×256 matrix to provide an in-plane pixel resolution of 0.55 mm. The imaging time was restricted to approximately 10 min in order to minimize patient movement artefact.
QUALITATIVE IMAGE ANALYSIS
The baseline and 3-year X-ray films were read by two observers working together (PAD and IW), and blind to date order of the films. The radiographs were read for K&L grade29 and for size of tibiofemoral osteophytes using the 0–3 scale and Altman atlas for standards30. Any other major abnormalities or apparent differences between the two films were also noted. In a similar manner, MR images were blinded to date order and reported by a consultant radiologist (IW). Qualitative assessment of images was undertaken describing the main sites for cartilage remodeling (cartilage thickness), bony remodeling (e.g. subchondral cysts or osteophytes) and accumulation of joint effusion. Any other major abnormalities or apparent differences between the two images were also noted.
QUANTITATIVE IMAGE ANALYSIS
Joint space narrowing (JSN) was measured at the midpoint of the medial tibiofemoral compartment using the graticule method as described by Lequesne31. The precision of the X-ray measurement in this centre has previously been established32–34. An image analysis package (‘Tosca’, IBM, Winchester, U.K.) was used to segment the MR images for all four hyaline cartilage compartments (femoral, medial tibial, lateral tibial and patellar). In previous studies10 on healthy joints, a semi-automatic image segmentation technique, in which the segmenter placed within the cartilage a seed point invoking a region-growing algorithm, was employed. However, pilot studies of segmentations in this OA population proved this semi-automated approach unreliable. Consequently fully manual segmentation procedure was employed35. Manual segmentation was performed by two medical physicists familiar with MR images of the knee (SJG: Segmenter A and MCK: Segmenter B) under the supervision of a musculoskeletal radiologist (IW). Any cartilage associated with osteophyte formation was excluded from the segmentations. In studies of the effect of image quality on reproducibility in osteo- and rheumatoid arthritis (SJG, MCK, unpublished), there was no trend for worse
Osteoarthritis and Cartilage Vol. 10, No. 12 image quality (sum of movement artefact or subjectively determined ease of segmentation) to correlate with an increased intra- or intersegmenter coefficient of variation. To avoid partial volume effects, only slices with a clearly defined articulating surface were segmented. To aid observer interpretation of partial volume effects, a 3-D model of the knee joint was provided for anatomical reference (‘Somso NS50’, Adam Rouilly, Sittingbourne U.K.). Femoral, patellar, medial tibial and lateral tibial cartilage segmentations were generated by enclosing all cartilage pixels within a region-of-interest for each 2-D image slice. Following careful checking and realignment of the ROI contour where necessary, the process was repeated for each image slice until a cartilage compartment volume had been fully segmented. The process of segmenting an entire knee (four compartment segmentations), with thorough checking, took approximately 4–5 h. The volume of each segmentation was obtained by finding the area of each 2-D ROI and multiplying by the slice thickness over all slices. The Tosca package uses a curve-fitting algorithm described by Akima36 in order to interpolate a contour defining each 2-D ROI. This spline interpolation uses only local information to determine the slope of a curve at each point, and it has been argued that such an interpolation is closer to a manually drawn curve than those produced by other splines. The area of each enclosed ROI was obtained by interpolation using this spline and then re-sampling the interpolated curve with 1000 equally spaced vertices to produce a dense polygon. The area of a polygon with n vertices (xi, yi) {i=0 . . . n−1} is then given (CJ Taylor, AD Brett, unpublished) by:
In addition to the quantitative volume calculation, 3-D volume images were reconstructed and displayed using the image-processing package ‘DX data explorer’ (IBM, Winchester, U.K.).
STATISTICAL ANALYSIS
All data are expressed as mean±standard deviation. Differences in JSW and volumes were evaluated using the two-sided Student’s t-test. P values of less than 0.05 were considered to be significant. Errors in measuring cartilage volumes may arise from many different experimental or physiological sources. Such errors can be assessed either through a single experiment with nested analysis of variance in which all possible sources of error are systematically varied, or through a series of experiments with pairwise comparison in which each possible source of error is varied in turn. The latter approach was adopted in this work. The reproducibility of each pairwise comparison was assessed from the test– retest coefficient of variation (CoV). For each subject, i, the CoV is the standard deviation, i, for the two measurements on that subject, divided by the mean volume, i, for the subject. The overall test–retest CoV for a group of N subjects is then
Use of the CoV is most appropriate when i is proportional to i. If i is independent of i, it is appropriate to
931 compare i values directly. In MRI studies in osteo- and rheumatoid arthritis (JCW, SJG, unpublished) it has been found that i increases with i (but with an approximately two-thirds, not linear, power relationship, consistent with the hypothesis that the errors in this case depend strongly on the surface area characteristics of the segmentation, since the surface area tends to increase with the two-thirds power of volume). Hence the CoV was employed. Point estimates of CoV are imprecise unless very large numbers of patients are sampled: in this case it is also helpful to quote the 95% confidence upper bound which can be derived37 from:
Results CLINICAL
Sixteen patients were recruited to the study (10 male, six female). They had a mean age of 63.4 years (range 52–70) and mean disease duration of 10.3 years (range 4–20). Only 11 were available for the final assessment at three years: two died, one lost his leg in trauma and two were lost to follow-up. This report concentrates on the 11 patients in whom full assessments were available at baseline and 3 years. There were five men and six women in this group, their mean age was 61.3 years and mean disease duration was 9.2 years. All fulfilled ACR clinical criteria for knee OA38. In addition all were over 40 years of age, had less than 30 min of morning stiffness, and had current knee pain and crepitus. Seven of the 11 patients had a HAQ score of >1.5 at entry to the study. There were no major changes in HAQ or deformities over the 3 years of the study. Table I shows some clinical and radiographic features of the 11 patients who completed the study. It is apparent that the majority had well-established radiographic changes of OA at entry. Four patients had K&L grade 2 and five patients had grade 3, although one set of radiographs was reported as ‘almost normal’. The mean medial joint space width at baseline of 4.9±1.3 mm suggests that the disease was not very advanced in this compartment in the group as a whole. Two patients had predominant lateral compartment disease, and one had predominant patello–femoral changes. There was little change over the 3 years (JSW at 3 years was 4.7±1.6 mm), although there was an average decrease (not statistically significant) in medial JSW of 0.21±0.37 mm, and four patients had JSN of 0.3 mm or more over the 3 years. GENERAL MRI OBSERVATIONS
All subjects gave analysable MRI data: no datasets were excluded from quantitative analysis. Cartilage volumes tended to be larger in males than in females. Segmentation was found to be particularly difficult when dealing with regions of very thin cartilage, where the ‘band structure’ was lost or altered, regions of ‘merged’ cartilage between two articulating surfaces (e.g. central regions of the tibiofemoral joint, and the base of the patellofemoral joint), regions of cartilage above and surrounding bony abnormalities, such as erosions, femoral cartilage at the base of the intercondylar notch, and also in other weight-bearing areas; and cartilage associated with (central) ‘mushroom’
932
S. J. Gandy et al.: Lack of cartilage volume change in knee OA Table I Clinical and radiographic features of patients who completed the 3 year study
Patient ID
04 13 21 24 26 28 29 34 35 41 44
Clinical (baseline)
Radiographic features (baseline)
Radiographic features (three years)
Age
Sex
BMI
K&L Grade
Medial JSW (mm)
Grade of osteophyte
JSW (mm)
Change in osteophyte
Change in JSW (mm)
Other features
67 52 71 63 70 56 65 62 54 62 52
F M M F F M F F F M M
30 28 31 29 23 28 32 33 33 22 23
3 2 1 2 2 1 3 3 3 3 2
2.1 6.0 6.0 5.7 4.4 5.3 5.9 5.5 3.1 4.1 6.0
2 2 1 1 2 1 2 2 1 1 2
1.5 5.6 6.0 5.5 4.6 5.0 5.8 5.5 2.1 3.9 6.3
0 0 0 0 0 0 0 −1 0 0 +1
−0.6 −0.4 0.0 −0.2 +0.2 −0.3 −0.1 0.0 −1.0 −0.2 +0.3
— — Enthesophytes PFJ disease Chondrocalcinosis Lateral disease Almost normal Lateral disease — — Attrition Lateral TFJ Chondrocalcinosis
PFJ: patello-femoral joint; TFJ: tibio-femoral joint.
Table II Coefficients of variation associated with repeated measurements of cartilage volume in each compartment
Same scan, same segmenter Same scan, different segmenter Different scan, different session, same segmenter
Substudy
N
Femoral
Patellar
Medial tibial
Lateral tibial
Total
1 1
6 6
2.0 (4.9) 7.8 (19)
5.5 (13) 5.9 (14)
3.3 (8.1) 14 (35)
2.8 (6.9) 5.7 (14)
1.6 (3.9) 4.7 (11)
2
5
2.0 (5.7)
5.3 (15)
4.6 (13)
6.7 (19)
1.8 (5.2)
In each case, the first figure is the rms CoV and the second (in parentheses) is the 95% confidence upper bound.
osteophytes. All image sets showed at least one area of focal cartilage loss, and were characterized by focal regions of cartilage thinning and/or loss in all compartments. This was particularly evident in the femoral and tibial cartilage. Other common features were peripheral osteophytes on the femur and patella, and mushroom osteophytes mainly on the femur. VALIDATION AND REPRODUCIBILITY
The segmented phantom volume was 20.1 ml compared with the actual volume of 20.5 ml (a 2.2% underestimation using MRI). Table II shows the CoV for three elements of reproducibility: intrasegmenter, intersegmenter, and scan– scan. In comparison with the first segmentation by segmenter A, total cartilage volume was 1.4% higher at the second segmentation of the same image by segmenter A, and 4.3% lower when the same image was segmented by segmenter B. These differences were not statistically significant. In the study of scan–scan variation, total cartilage volume was 0.3% lower in the second scan in comparison with the first scan. This tiny difference was not statistically significant. PROGRESSION OVER 37 MONTHS
Figure 1 shows the intercondyle MR images at 0 and 37 months of patient 35 who had joint space narrowing of 1 mm over the 3-year period. These images were representative of those obtained in this study. The radiologist’s
report on the MRI scans described cartilage lesions in virtually all of the 3-D segmented knee compartments examined. These can be clearly observed both in the original MRI images and in the reconstructed 3-D image of the MRI scans (see Fig. 2 for 3D reconstruction of MR images from patient 35 at 0 and 37 months). In this patient population lesions on the femur were most commonly seen at the anterior aspect of the medial condyle, the intercondylar notch and the middle aspect of the lateral condyle. Lesions were generally located centrally on the patella, and at various sites on both tibiae (more commonly on the lateral tibia). The radiologist’s report noted both increases and decreases in lesion size between baseline and 3 years. The radiologist’s report also noted that regions of cartilage thinning in one or more compartments were clearly visible on the 3-D MRI data of four patients. In contrast to these observations, regions of cartilage thickening over the 3 years were identified on the 3-D MRI data of two patients. The corresponding X-rays from these patients showed no JSW changes over time. In three patients there was no visible evidence of temporal change to the cartilage or joint-space width on X-ray or MRI. Figure 3 shows total cartilage volumes at 0, 1 and 3 years for all patients who participated in the study. Total cartilage volumes were 15.5±1.6 ml (range of 12.3 to 16.6 ml) for females and were 21.4±4.5 ml (range of 14.7 to 29.9 ml) for males. There was no significant change in total cartilage volume over the course of 3 years in this patient cohort. Nor was there any significant change to any
Osteoarthritis and Cartilage Vol. 10, No. 12
933
Fig. 1. Examples of magnetic resonance (MR) images. Lateral condyle region of patient 35 at baseline (left) and 37 months.
Fig. 2. Examples of segmentations. 3-D reconstructed MR image of cartilage for patient 35 at baseline (left) and 37 months (right). During this period, joint space width increased by 0.2 mm, while cartilage volume changes were −0.19 ml in the femur, +0.21 ml in the patella and −0.06 ml in the tibia.
of the individual knee cartilage compartments (femur, lateral tibia, medial tibia, patella) as shown in Table III. Volume changes (loss or gain) did not even occur at a rate comparable to the same segmenter/same scan CoV of 1.8%. MRI measured a mean loss of only 1.6% or 0.36±1.3 ml (mean± S.D.) in the knee over 3 years. Neither this change, nor the changes in individual compartments or at intermediate timepoints, were statistically significant
Fig. 3. Changes in cartilage volume. Total cartilage volume at 0, 1 and 3 years in all patients (males: open symbols; females: filled symbols). The symbol at the right represents ±1.8%, i.e. ±1 S.D. for intrasegmenter reproducibility (same segmenter, different scan35).
Although the average cartilage volume did not change, there were some large changes in individual patients. In five of the 11 patients, the change was more than 5.5% (increase or decrease), more than three standard deviations (i.e. three times the intraobserver CoV).
934
S. J. Gandy et al.: Lack of cartilage volume change in knee OA Table III Changes in cartilage volume in each compartment over three years Compartment
Femoral Lateral tibial Medial tibial Patellar Total
Volume (S.D.) at baseline/ml (N=16) 12.1 (2.9) 2.3 (0.5) 2.1 (0.6) 2.6 (1.2) 19.2 (4.7)
Mean (S.D.) percentage change from volume at baseline At 4.6±1.9 months (N=16) +0.2 (4.5) −0.4 (6.6) +2.4 (4.8) −0.4 (7.3) +0.3 (3.4)
P>0.2 P>0.2 P=0.05 P>0.2 P>0.2
Discussion Methods to quantify cartilage volume from MRI have been available for over 10 years8. A number of publications exist which show the method is highly reproducible and reflects cartilage volume measured directly from postoperative or cadaveric samples9,12,39. Our data show a very low CoV of the method, comparable to that reported from OA patients in other centres18–20,24. Given the reproducibility of our method, in this study we have >80% power to detect a net change (two-sided =0.05) of 1%/year for the total cartilage or the femoral compartment, 2%/year for the medial tibial compartment, or 3%/year in the lateral tibial or patellar compartment. We therefore have confidence in our ability to measure cartilage volume in the knee joint. Using this technique we have found that there was no significant loss of total cartilage volume in 11 patients with knee OA studied over a 3-year period. These data appear to challenge the face validity for the use of total cartilage volume to assess structural changes in OA. The study was performed in a typical clinical research setting using a readily available hospital scanner, a standard RF coil, and standard radiographic procedures in a hospital radiology department. The protocol has been implemented in a multicentre trial setting21. The design of any study of chronic disease progression using rapidly developing technology raises methodological issues, since the methods will almost inevitably appear outdated by the time the study is completed. The duration of the present study from inception to manuscript preparation was about 5 years, and in that time, significant developments in MRI methodology have occurred. The present study used a field strength of 1.0 T. We have shown that with this knee protocol in a multicentre study, that data collected at 1.0 T and 1.5 T are comparable21, but no doubt the instruments operating at 3 T, now in routine clinical use, will offer better sensitivity and resolution, and improved statistical power for the detection of tiny morphologic change. Improvements in gradient technology and pulse sequence design now allow shorter TR values, permitting higher resolution in a given scanning time20 and shorter TE values, providing better discrimination of cartilage from other tissues14. Improved image analysis approaches10, focusing for example only on weight-bearing areas or only on regions with established lesions, may provide improved sensitivity to disease progression. Examination of weight-bearing regions from coronal sections has been advocated14. Additionally, better understanding of the disease itself may allow better identification of patients at greatest risk of rapid cartilage loss, such as those with obesity. The patients included in this study were a highly selected group being recruited from a specialist clinic and undergoing secondary care for their knee OA. Such groups of
At 11.9±0.4 months (N=15) −0.1 (4.8) −0.1 (6.9) +2.9 (7.3) +0.8 (9.7) −0.2 (3.4)
P>0.2 P>0.2 P=0.06 P>0.2 P>0.2
At 37.2±2.2 months (N=11) −2.4 (7.3) +3.0 (8.4) −1.3 (12.0) −2.5 (14.3) −1.6 (6.0)
P>0.2 P>0.2 P>0.2 P=0.13 P>0.2
patients generally represent more advanced OA than that seen in the community and are arguably more likely to progress than unselected groups. However, we deliberately excluded those with the most advanced OA, and the radiographic data shown indicate that this group of patients did not have very severe structural changes at entry to the study. The demography of the group indicates that they were reasonably representative of other hospital based patient cohorts, although there were rather more men than would be expected. In general, these are the sorts of patients who are enrolled to many studies of progression or interventions for knee OA. There is no doubt that our group had OA; as they all fitted the ACR clinical criteria, and all had some radiographic changes; however, in view of the nature of the selection and the relatively small numbers we fully accept that this group may not be representative and the results are not necessarily generally applicable. The generally accepted means of assessing the presence of joint changes consistent with OA is through plain X-radiography, and in the case of the knee, the weightbearing AP view as used in this study. Recently it has been shown that the semi-flexed view is more sensitive to JSN than the standard film (with full knee extension) as used in this study40, but again, the views reported here are those that have been used in most other studies of knee OA. All our patients had osteophytes, four had clear narrowing of the medial joint space width at entry (<4.5 mm), two others had obvious lateral compartment disease and one patello– femoral compartment changes, suggesting that the majority had some cartilage loss at entry to the study. We therefore felt confident that the majority had cartilage pathology as well as bone changes (osteophytes) and the clinical as well as radiographic criteria for entry to an OA study30,40,41. Mean JSN at 3 years showed an overall loss of only 0.23 mm, which is consistent with an annual rate of change of only 0.08 mm/annum. This is less than the rate of change reported in most studies, which is nearer 0.2 mm/ annum3,42. Previous studies have suggested that cartilage loss with time may be faster in people who have already lost a significant amount of joint space width43, Table I suggests that this may be apparent in our cases as well, in that the two people who showed the greatest change in medial joint space width over 3 years each had relatively low joint space width at entry. Using the techniques of joint space width and radiographic views described, it has been calculated that the standard error of the measurement mean for medial joint space width is about 0.30 mm41. Four of our patients showed medial JSN of 0.3 mm or more over the 3 years, which could be regarded as significant progression of cartilage damage43, in addition, three of the 11 who were reviewed after 3 years had predominant lateral
Osteoarthritis and Cartilage Vol. 10, No. 12 compartment disease; in two of these cases we recorded an increase of medial compartment joint space width due to the ‘tilting’ of the knee associated with progression of lateral compartment disease. Therefore, although the mean rate of change was low, we believe that at least six of the 11 patients had conventional and significant radiographic evidence of total cartilage loss over the 3 years, which we would expect to be able to document with quantitative MRI techniques. All our patients had lesions (areas of thin or absent cartilage), which are never seen in normals. Thus, it appears that our patients had already embarked on a course of disease progression with considerable cartilage loss before entering the study. However, our data show small variations in total cartilage volumes over the 3-year study duration, with no individual patient showing much loss. There are several possible reasons for the lack of change in total cartilage volume of knee joint (measured from MRI) in the face of disease progression in OA. The most obvious explanation is that OA is a focal disease and cartilage change is usually concentrated on small areas of the joint subjected to maximal loading. Assessment of total cartilage volumes will dilute any change in these areas. A second explanation is offered by data from other studies using histology, MRI or arthroscopy, which have shown that some parts of the articular cartilage increase in volume due to excess hydration in the early phases of OA. It is quite possible that progression of OA in whole joints will result in thickening of cartilage in some areas and loss of cartilage volume in others resulting in no measurable change in total cartilage volume. Specifically, progression of relatively advance lesions in one compartment might be accompanied by earlier changes in swelling of the cartilage in another compartment. In support of this, Table III shows that while there was no net change in volume from baseline, there was a firm trend to increasing S.D. with increasing time, implying that some patients showed increased cartilage while others show a decrease. A third possible explanation is that loss of articular cartilage in the group of patients studied was occurring too slowly to be detected even after 3 years. However, the change in the radiographs and visual findings on MR suggest that cartilage lesions were progressing in many of the patients over the 3 years, but in spite of that no individual showed any significant change in total volumes measured from MRI. It should be noted, that the change in JSN may reflect not cartilage loss, but meniscal extrusion during weightbearing, as suggested by recent MRI studies44. One major difference in the way in which X-ray and MRI image the joint is that the former is done with weight bearing and the latter without. If one of the major causes for cartilage thinning in OA is increased deformability on weight-bearing then one might get a major discrepancy between volume measurements on MRI and JSN assessed on weight-bearing X-rays. However, conventional pathological studies of OA suggests that the major loss of structural elements of the cartilage occurs in the focal areas that are most affected.
Conclusions Given these data and the arguments in the literature reviewed above, we suggest that total cartilage volumes will not prove as useful as hoped in the assessments of structural change in OA. However, we do believe that MRI is the preferred technique for the quantification of cartilage
935 damage. Future quantitative work should concentrate on the assessments of focal volume areas, thickness mapping10,45 or quality change46 rather than the volume measurement of the whole joint.
Acknowledgments The authors would like to acknowledge Bristol University Rheumatology Unit staff, including Dr John Kirwan, Dr Margaret Byron, Dr Jon Tobias and Research Sisters Hilary Brownett, Kathryn Mitchell and Trudy Sheen for their help in patient recruitment and care; all the MRI Radiography staff at Bristol Royal Infirmary for imaging the patients; Dr Alan Brett at the University of Manchester (presently at ImageMetrics) for deriving and implementing the calculation of area from ROI, and Dr John Foster, presently at Western Royal Infirmary, Glasgow for early contribution to the work.
References 1. Felson D, Hochberg M, McAlindon T, Dieppe P, Minor M, Blair S, et al. Osteoarthritis: new insights—Part 2: Treatment approaches. Ann Int Med 2000;133: 726–37. 2. Felson D, Lawrence R, Dieppe P, Hirsch R, Helmick C, Jordan J, et al. Osteoarthritis: New Insights—Part 1: The disease and its risk factors. Ann Int Med 2000;133:635–46. 3. Mazzuca S, Brandt K, Katz B. Is conventional radiography suitable for evaluation of a disease-modifying drug in patients with knee osteoarthritis? Osteoarthritis Cart 1997;5:217–26. 4. Peterfy CG, Van Dijke CF, Janzen DL, Gluer CC, Namba R, Majumdar S, et al. Quantification of articular cartilage in the knee with pulsed saturation transfer subtraction and fat-suppressed MR imaging: Optimization and validation. Radiology 1994;192: 485–91. 5. Lequesne M, Brandt K, Bellamy N, Moskowitz R, Menkes J, Pelletier J, et al. Guidelines for testing slow-acting and disease modifying drugs in osteoarthritis. J Rheumatol 1994;21(suppl 41): 65–71. 6. GREES—osteoarthritis section. Recommendations for the registration of drugs used in the treatment of osteoarthritis. Ann Rheum Dis 1996;55:552–7. 7. Konig H, Sauter R, Deimling M, Vogt M. Cartilage disorders: Comparison of spin–echo, CHESS, and FLASH sequence MR images. Radiology 1987;164: 753–8. 8. Paul PK, Wang JZ, Mezrich RS, Rakhit A, Dunton AW, Furst D, et al. 3D-MRI: a novel approach to quantitation of articular cartilage. Arthritis Rheum 1990;33: S91. 9. Eckstein F, Sittek H, Milz S, Putz R, Reiser M. The morphology of articular cartilage assessed by magnetic resonance imaging. Surg Radiol Anat 1994;16: 429–38. 10. Waterton JC, Solloway S, Foster JE, Keen MC, Gandy S, Middleton BJ, et al. Diurnal variation in the articular cartilage of the knee in young adult humans. Magn Reson Med 2000;43:126–32. 11. Eckstein F, Schnier M, Haubner M, Priebsch J, Glaser C, Englmeier KH, et al. Accuracy of cartilage volume and thickness measurements with magnetic resonance imaging. Clin Ortho Rel Res 1998;352: 137–48.
936
S. J. Gandy et al.: Lack of cartilage volume change in knee OA
12. Peterfy CG, Van Dijke CF, Lu Y, Nguyen A, Connick TJ, Kneeland JB, et al. Quantification of the volume of articular cartilage in the metacarpophalangeal joints of the hand—accuracy and precision of 3-dimensional MR-imaging. Am J Roent 1995;165: 371–5. 13. Eckstein F, Reiser M, Englmeier K-H, Putz R. In vivo morphometry and functional analysis of human articular cartilage with quantitative MRI—from image to data, from data to theory. Anat Embryol 2001;203:147–73. 14. Hyhlik-Du¨rr A, Faber S, Burgkart R, Stammberger T, Maag K-P, Englmeier K-H, et al. Precision of tibial cartilage morphometry with a coronal waterexcitation MR sequence. Eur Radiol 2000;10: 297–303. 15. Hardy PA, Newmark R, Liu YM, Meier D, Norris S, Piraino DW, et al. The influence of the resolution and contrast on measuring the articular cartilage volume in MRI. Magn Reson Imag 2000;18:965–72. 16. Foster JE, Dieppe P, Maciewicz R, Taberner J, Watt I, Waterton JC. Quantification of cartilage volume and visualisation of defects in osteoarthritis using a clinical MR system. Arthritis Rheum 1996;39:S170. 17. Cicuttini F, Wluka AE, Stuckey SL. Tibial and femoral cartilage changes in knee arthritis. Ann Rheum Dis 2001;60:977–80. 18. Lynch JA, Zaim S, Zhao J, Stork A, Peterfy CG, Genant HK. Cartilage segmentation of 3D MRI scans of the osteoarthritic knee combining user knowledge and active contours. Proc Soc Photo-Optical Instrument Engineers 2000;3979:925–35. 19. Cicuttini F, Forbes A, Asbeutah A, Morris K, Stuckey S. Comparison and reproducibility of fast and conventional spoiled gradient-echo MR sequences in the determination of knee cartilage volume. J Orthop Res 2000;18:580–4. 20. Burgkart R, Glaser C, Hyhlik-Du¨rr, Englmeier K-H, Reiser M, Eckstein F. MRI-based assessment of cartilage loss in severe osteoarthritis. Arthritis Rheum 2001;44:2072–77. 21. Moots RJ, Morgan SR, Nash AFP, Waterton JC, Maciewicz RA, Farebrother JE, et al. MRI measurement of knee cartilage volume in a multicentre study. Rheumatology 2002;41(suppl 1): 123. 22. Wild M, Jacobi V, Windolf J. The quantitative detection of posttraumatic osteoarthrosis after fractures of the tibial plateau using MRI-based chondrovolumetry. Eur J Trauma 2000;26:15–24. 23. Glisson M, Forbes A, Morris K, Stuckey S, Cicuttini F. Comparison of X-rays and magnetic resonance imaging in the definition of tibiofemoral joint osteoarthritis. Radiography 2000;6:205–209. 24. Steines D, Cheng C, Wong A, Berger F, Napel S, Lang P. Segmentation of osteoarthritic femoral cartilage from MRI. In: Lemke HU, Inamura K, Doi K, Vannier MW, Farmer AG, Eds. Computer-assisted Radiology and Surgery. The Netherlands: Elsevier 2000:578– 83. 25. Boegard TL, Rudling O, Petersson IF, Jonsson K. MRI of the knee in chronic knee pain. A 2-year follow-up. Osteoarthritis Cart 2001;9:473–80. 26. Pessis EG, Drape J, Aryal X, Auleley G, Dougados M, Chevrot A. Quantification with MRI of chondral lesions of the medial tibiofemoral compartment: longitudinal study with MRI and arthroscopic followup. Radiology 1998;209P:414.
27. Peterfy C, White D, Zhao J, VanDijke C, Genant HK. Longitudinal measurement of knee cartilage volume in osteoarthritis. Arthritis Rheum 1998;41:S361. 28. Raynauld JP, Kauffman C, Godbout B, Beaudoin G, Berthiaume MJ, DeGuise J, et al. Knee osteoarthritis progression evaluated by magnetic resonance imaging and a novel quantification software tool. Osteoarthritis Cart 2000;8(b): S82. 29. Kellgren JH, Lawrence JS. Radiographic assessment of osteoarthritis. Ann Rheum Dis 1957;16:494–502. 30. Altman R, Hochberg M, Murphy WD, Wolfe F, Lequesne M. Atlas of individual radiographic features in osteoarthritis. Osteoarthritis Cart 1995;3(suppl A): 3–70. 31. Lequesne M. Quantitative measurements of joint space during progression of osteoarthritis: chondrometry. In: Kuettner K, Goldberg V, Eds. Osteoarthritic Disorders. Rosemont: American Academy of Orthopaedic Surgeons 1995:427–44. 32. Dieppe PA, Cushnaghan J, Young P, Kirwan J. Prediction of the progression of joint space narrowing in osteoarthritis of the knee by bone scintigraphy. Ann Rheum Dis 1993;52:557–63. 33. Dieppe PA, Cushaghan J, Shepstone L. The Bristol ‘OA 500’ study: progression of osteoarthritis (OA) over 3 years and the relationship between clinical and radiographic changes at the knee joint. Osteoarthritis Cart 1997;5:85–97. 34. Mazzuca SA, Brandt KD, Dieppe PA, Doherty M, Katz BP, Lane KA. Effect of alignment of the medial tibial plateau and X-ray beam on apparent progression of osteoarthritis in the standing anterioposterior knee radiograph. Arthritis Rheum 2001;44:1786–94. 35. Gandy SJ, Dieppe P, Watt I, Brett AD, Keen MC, Maciewicz RA, et al. Reproducibility of knee cartilage volume measurements in arthritis patients. Proc Soc Magn Reson Med Ninth Scientific Meeting and Exhibition, Glasgow, UK 2001; 2098. 36. Akima H. A new method of interpolation and smooth curve fitting based on local procedures. J Assoc Computing Machin 1974;17:589–602. 37. Johnson RA, Bhattacharyya GK. Statistics Principles and Methods, 4th ed. New York: Wiley 2001: 397–401. 38. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al. Development of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Committee of the American Rheumatism Association. Arthritis Rheum 1986;29: 1039–1049. 39. Eckstein F, Sittek H, Milz S, Schulte E, Kiefer B, Reiser M. The potential for magnetic resonance imaging (MRI) for quantifying articular cartilage thickness—a methodological study. Clin Biomech 1995;10: 434–40. 40. Dieppe P, Brandt K, Lohmander S, Felson D. Detecting and measuring modification in osteoarthritis: the need for standardized methodology. J Rheumatol 1995;22:201–3. 41. Buckland-Wright J, Macfarlane D, Williams S, Ward J. Accuracy and precision of joint space width measurements in standard and macroradiographs of osteoarthritic knees. Ann Rheum Dis 1995;54:872–80. 42. Mazzuca S. Plain radiographic evaluation of knee osteoarthritis. Curr Opini Rheumatol 1997;9:263–7.
Osteoarthritis and Cartilage Vol. 10, No. 12 43. Dougados M, Gueguen A, Nguyen M, Thiesce A, Listrat V, Jacob L, et al. Longitudinal radiographic evaluation of osteoarthritis of the knee. J Rheumatol 1992;19:378–84. 44. Gale DR, Chaisson CE, Totterman SM, Schwartz RK, Gale ME, Felson D. Meniscal subluxation: association with osteoarthritis and joint space narrowing. Osteoarthritis Cart 1999;7:526–532.
937 45. Eckstein F, Gavazzeni A, Sittek H, Haubner M, Lo¨sch A, Milz S, et al. Determination of knee joint cartilage thickness using three-dimensional magnetic resonance chondro-crassometry (3D MR-CCM). Magn Reson Med 1996;36:256–65. 46. Frank LR, Wong, EC, Buxton RB, Resnick D. Mapping the physiological parameters of articular cartilage with MRI. Topics in Magn Reson Imaging 1999;10:153–79.