Quantitative magnetic resonance imaging evaluation of knee osteoarthritis progression over two years and correlation with clinical symptoms and radiologic changes




To evaluate the change in osteoarthritic (OA) knee cartilage volume over a two-year period with the use of magnetic resonance imaging (MRI) and to correlate the MRI changes with radiologic changes.


Thirty-two patients with symptomatic knee OA underwent MRI of the knee at baseline and at 6, 12, 18, and 24 months. Loss of cartilage volumes were computed and contrasted with changes in clinical variables for OA and with standardized semiflexed knee radiographs at baseline at 1 and 2 years.


Progression of cartilage loss at all followup points was statistically significant (P < 0.0001), with a mean ± SD of 3.8 ± 5.1% for global cartilage loss and 4.3 ± 6.5% for medial compartment cartilage loss at 6 months, 3.6 ± 5.1% and 4.2 ± 7.5% at 12 months, and 6.1 ± 7.2% and 7.6 ± 8.6% at 24 months. Discriminant function analysis identified 2 groups of patients, those who progressed slowly (<2% of global cartilage loss; n = 21) and those who progressed rapidly (>15% of global cartilage loss; n = 11) over the 2 years of study. At baseline, there was a greater proportion of women (P = 0.001), a lower range of motion (P = 0.01), a greater circumference and higher level of pain (P = 0.05) and stiffness in the study knee, and a higher body mass index in the fast progressor group compared with the slow progressor group. No statistical correlation between loss of cartilage volume and radiographic changes was seen.


Quantitative MRI can measure the progression of knee OA precisely and can help to identify patients with rapidly progressing disease. These findings indicate that MRI could be helpful in assessing the effects of treatment with structure-modifying agents in OA.

In people over the age of 60 years, osteoarthritis (OA) is a common cause of disability. Assessment of cartilage damage is important for monitoring disease progression and evaluating therapeutic response in OA. For many years, clinical studies of drug interventions in symptomatic knee OA have focused specifically on clinical parameters, such as pain and joint function, without assessing the effect of treatment on structural changes caused by the disease or the role of treatment in preventing cartilage degradation. Serial radiographs of affected joints appear to be a logical means of documenting the progression of OA over time (1), provided that a validated, reliable, and easily reproducible technique is used (2–5).

Improvements in the standardization and interpretation of radiographs have produced better measurements of joint space width (JSW) and the progression of joint space narrowing (6). It has been suggested that this progression is such that a minimum followup of at least 2 years, using large numbers of patients, is necessary to establish the effect of pharmacologic interventions on OA progression (7–10). The validity of measurements of JSW in knee OA has been demonstrated by correlations between the medial tibiofemoral JSW and the sum of the thickness of both the femoral and the tibial articular cartilage (11). The use of arthroscopy to assess cartilage appears to be reliable and sensitive to change at 1 year (12, 13). However, only the cartilage surface can be evaluated, and the method is both semiquantitative and, most importantly, invasive, which is problematic for multicenter studies.

Magnetic resonance imaging (MRI) allows for the precise visualization of joint structures, such as cartilage, bone, synovium, ligaments, and menisci, as well as pathologic changes thereof. Recent advances in MRI technology have led to significant improvements in spatial resolution and contrast, enabling researchers to evaluate anatomic damage of all these joint structures in the axial, coronal, and sagittal planes (14–24). Recently, we and other investigators have developed a system for quantifying cartilage volume using MRI acquisitions combined with sophisticated software (25–41); however, to date, few studies have used MRI technology to evaluate cartilage volume changes over time (42–46). Some clinical parameters, such as age, weight, body mass index (BMI), trauma, and baseline radiologic JSW, have already been identified as predictors of rapid radiologic progression of knee OA (47, 48). Such risk factors for disease progression have not been assessed in terms of cartilage volume loss on MRI as an outcome for OA progression.

The objective of this study was to evaluate a cohort of patients with moderately symptomatic knee OA longitudinally, in a typical rheumatology practice setting, by following changes in cartilage volume over time, as determined by MRI. Moreover, we sought to identify risk factors for greater disease progression and to contrast these changes with data collected from the already validated clinical and radiologic assessment tools for OA.


Patient selection.

Forty patients were recruited from the outpatient Rheumatology Clinic at the Centre Hospitalier de l'Université de Montréal, Hôpital Notre-Dame. Rheumatologists at the Arthritis Division of Notre-Dame Hospital provided the patients. Male and female patients were eligible for the study if they were between the ages of 40 and 80 years, fulfilled the American College of Rheumatology (ACR) criteria for knee OA (49), and had symptomatic disease requiring medical treatment in the form of acetaminophen, traditional nonsteroidal antiinflammatory drugs (NSAIDs), or selective cyclooxygenase 2 (COX-2) inhibitors. Eligible patients were required to have radiologic evidence of OA of the affected knee on a radiograph obtained within 6 months of the start of the study. A severity grade of 2 or 3 on the Kellgren/Lawrence scale (50) for joint space narrowing, osteophytes, or sclerosis, was also required. In addition, a minimum JSW of 2–4 mm in the medial compartment was also required, as measured with a ruler on knee radiographs obtained with the patient in a standing position with the knee fully extended (9).

JSW was measured on the extended-knee radiographic views obtained for screening purposes at the time of recruitment. If eligible for the study, a baseline radiograph was obtained with the patient's knee in the standardized semiflexed view to be compared with films obtained during followup. On the fully extended–knee views at baseline, 20 patients had a narrower medial compartment, and 14 did not. However, in patients who had evaluable semiflexed-knee views at baseline and 2 years, 25 had a narrower medial compartment compared with the lateral compartment at baseline, and 9 did not. None of the patients had OA that affected only the lateral compartment.

Patients with chondrocalcinosis were excluded from the study. Patients were also excluded if they had isolated patellofemoral OA, if they had knee OA that was secondary to other conditions (including inflammation, sepsis, metabolic abnormalities, and trauma), if they had an acute or chronic infection (including tuberculosis), and if they had any contraindications for the use of MRI. Additional exclusion factors consisted of a history (past or present) of gastrointestinal ulceration, and radiologic grade 4 OA on the Kellgren/Lawrence scale. Patients with severe (class IV) functional disability, candidates for imminent knee joint surgery, and patients with a contralateral total joint replacement were also excluded.

In patients in whom both knees were symptomatic, the most symptomatic knee was selected for the investigation. Patients were permitted to receive simple analgesics, NSAIDs, or intraarticular steroid injections, and analgesic regimens could be changed according to the preference of the patient's rheumatologist and according to the patient's clinical course. There was no medication washout period prior to the clinical evaluation. Medication regimens and any changes in them were closely monitored and recorded. The use of indomethacin was not permitted because of its potential to promote degeneration of OA cartilage (13, 51–53). Likewise, the use of glucosamine sulfate was not permitted because of its potential to affect OA progression (54).

This study was approved by the ethics committee of the Notre-Dame Hospital. Patients gave their informed consent for study.

Clinical evaluation.

Patients underwent clinical evaluation at baseline and every 6 months thereafter. Patients were first evaluated on the basis of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). This measure, developed by Bellamy et al (55), is a tri-dimensional self-administered questionnaire about pain (5 items), stiffness (2 items), and physical function (17 items). It has been fully validated in French Canadian patients and established to be reliable (56). In addition, physicians made a subjective judgment of the patient's disease activity (physician's global assessment) based on the patient's symptoms, functional capacity, findings of a physical examination, and, if applicable, laboratory findings, using a visual analog scale (VAS; 0 = best and 100 = worse). The patients themselves also used a VAS to make a global assessment of their condition (0 = very good and 100 = very bad) and to rate the pain they were having that day (0 = no pain and 100 = most severe pain). The range (flexion minus extension) of motion (ROM) of the affected knee was measured with a long-arm goniometer, and the amount of time required to walk a distance of 50 feet “as fast as possible” was measured with a stop-watch and reported in seconds. Flares of knee pain and effusions were also recorded during and between patient visits. The evaluators were unaware of the results of previous radiologic or MRI data. Patients continued taking concomitant medications throughout their clinical evaluation period.

Knee radiographs.

Progression of joint space narrowing in the study knee over 2 years was evaluated at the narrowest point in the medial tibiofemoral compartment, according to the published protocol (7). This protocol permitted standardization of radiographs by positioning the knee in a semiflexed position, aided by fluoroscopy, and attaching a metal sphere to the head of the fibula to correct the effect of radiographic magnification. Plain extended-knee radiographs were only used for the screening step of this study, as described above. If the patient was eligible for study, a baseline radiograph in the standardized semiflexed view was obtained; this would be contrasted with the followup radiographs using the same standardized semiflexed views.

The films were digitized using a Lumiscan 200 laser film digitizer (Lumisys, Sunnyvale, CA). Prior to digitization, all films were bar coded to ensure that on digitization, the computer database linked patient/visit data to the JSW measurement obtained from each radiograph of a given patient. Minimum JSW was measured in the medial compartment on each of the radiographs using an automated computerized method of measurement (57). Manual intervention was occasionally required in instances where the radiographic quality of the film prevented the implementation of the automatic JSW measurement software. This intervention ensured reliable JSW measurement (58). The coefficient of variation for the JSW measurement was 1% for test–retest repeat radiographs of the knee in the semiflexed position (7). JSW data at 2 years were available for 34 patients. However, both MRI and JSW data at 2 years were available for 31 patients.

Knee MRI.

Each patient underwent high-resolution, 3-dimensional (3-D) MRI at baseline and at 6, 12, 18, and 24 months using the Magnetom Vision 1.5T machine with a dedicated knee coil commercially available from Siemens (Erlangen, Germany), as previously described (36). These examinations are optimized 3-D, fast imaging with steady-state precession acquisitions with fat suppression. All parameters were set to produce images with the highest cartilage-contrast, resolution, and signal-to-noise ratio within a reasonable acquisition time: repetition time 42 msec, echo time 7 msec, flip angle 20°, bandwidth 98 Hz/pixel, matrix size 410 × 512 pixels. The sagittal field of view was set to 160 mm and was rectangular whenever possible. About 80–110, 1.0-mm–thick partitions yielded a volume data set with an effective voxel size of 0.31 × 0.39 × 1.0 mm3. A strict positioning and immobilizing protocol was used to reduce movement during image acquisition. Two small, high-precision, geometric phantoms, shaped as cylinders, were positioned on the internal and posterior aspect of the knee to monitor patient movement. The total time for patient positioning and MR acquisition ranged from 24 to 31 minutes. The patients were able to tolerate this procedure without any significant problem.

MRI processing.

Cartilage thickness and knee joint volume were measured by 2 trained and blinded readers using a specially developed computer program (Cartiscope; ArthroVision, Inc., Montreal, Quebec, Canada) running on a Windows NT/9x workstation, as previously described (36). Cartilage thickness was defined as the Euclidean distance between the bone–cartilage interface delineated by the baseline image and the cartilage–surrounding tissue interface. It was computed in 3-D space for each point of the bone surface along a direction perpendicular to the bone surface. Each thickness value measured can be expressed as a standardized map representation and is designated as a thickness map. Hence, a typical data set provides ∼60,000 measurement points for the femur and ∼40,000 for the tibia. Such a thickness map can be represented as either gray-scale or pseudo color-coded images on a conventional screen or can be remapped with associated 3-D geometry using computer graphing techniques. Global or subregional volumes are evaluated directly from offset-maps as the sum of the elementary volumes. An elementary volume is defined as the volume between the bone–cartilage interface offset-map and its corresponding cartilage–synovium offset-map.

Registration of different data sets for followup evaluation.

Because a patient's knee cannot reliably be placed in the MR machine in precisely the same position (orientation and flexion) at each visit, a registration procedure was required to provide a similar standardized view of the tibia and femur for followup evaluation over time. Baseline bone–cartilage interfaces were used as a reference, under the assumption that they would show less geometric variation over time than would the cartilage–synovium interface. The 3-D tibial and femoral bone–cartilage interface from the first visit could be automatically registered in the new volume data set by using a 3-D image–edge extraction algorithm and a least-squared distance minimization technique.

As was done for the bone–cartilage interfaces, the cartilage–synovium interfaces for the femur and tibia obtained in the first data set were positioned into the new image set to be used as initial estimates for the delineation of cartilage–synovium interfaces. The segmentation was then performed, as it was for the first data set, with the semiautomatic method under reader supervision and with corrections when needed.

Finally, difference-maps between data sets from 2 different visits can also be computed and displayed. This registration procedure was previously used to demonstrate the high test–retest reliability and intrareader variability of our procedure (36). The reproducibility for knee cartilage volume measurements was further analyzed using the coefficient of variation of 4 intraobserver, interscan, and image acquisition in 1 session values for 8 healthy subjects (ages 25–35 years) and 8 subjects with knee OA (ages 55–65 years; Kellgren/Lawrence OA grade 2 or 3) using the root mean square (RMS) coefficient of variation percentage (CV%), as previously suggested (59).

Difference-maps between baseline and acquisitions at 6, 12, 18, and 24 months were blinded for the time point that was being assessed. The change in knee cartilage volume was obtained by subtracting the volume at followup from the initial volume. The percentage change was calculated as follows:

equation image

Statistical analysis.

All of the data (clinical, radiological, and laboratory) were systematically entered into a computerized database using a blinded double-entry procedure. Descriptive statistics for patient characteristics were tabulated. The primary outcome variable was the change in cartilage volume over time for the entire knee (global) and for each of the knee compartments (medial or lateral femoral condyle and tibial plateau), respectively. The changes in cartilage volume between baseline and months 6, 12, 18, and 24 were estimated blindly for our longitudinal cohort, with a 1-sample t-test for each time point. The cartilage volume losses are presented as percentage losses compared with baseline.

A linear discriminant analysis was used to identify slow or fast progressors based on cartilage volume loss over 2 years. These subgroups were further analyzed to contrast their baseline demographic and clinical and radiologic features; a nonparametric Wilcoxon 2-sample test or chi-square test was done to assess the statistical relevance. Finally, the relationship between cartilage volume loss and JSW loss was explored at 2 years in 31 patients, using the Spearman correlation test. All statistical analyses were done using Mathlab software packages (version 13; Mathworks, Natick, MA). All tests were 2-sided, and P values ≤0.05 were considered statistically significant. Analyses were done without corrections for multiple comparisons.


Patient characteristics.

Forty patients were enrolled in this longitudinal and observational study. Four patients were lost to followup early in the study (2 patients died; 2 patients withdrew consent), and 1 or more MRI acquisitions were missing for 4 other patients and they were therefore excluded from the MRI analyses and demographic summaries. Thus, 32 patients had all MRI acquisitions over the 2 years, 34 patients had all radiographic acquisitions, and 31 patients had both. In terms of demographic and disease characteristics, the study cohort was generally similar at baseline to a general population of OA patients. Their mean ± SD age was 62.9 ± 8.2 years, 74% of them were women, their mean ± SD weight was 84.1 ± 15.1 kg, most of them were taking analgesics or NSAIDs, and most of them had disease activity scores in the mid-range (Table 1).

Table 1. Baseline characteristics of the 32 patients with osteoarthritis of the knee who completed the 2-year followup*
  • *

    Except where indicated otherwise, values are the mean ± SD. NSAIDs = nonsteroidal antiinflammatory drugs; VAS = visual analog scale (0–100 mm); SF-36 = Short Form 36; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index (VAS version); JSW = joint space width.

Age, years62.9 ± 8.2
Sex, % of patients 
Education, years11.4 ± 3.6
Duration of knee osteoarthritis, years8.9 ± 7.2
Weight, kg84.1 ± 15.1
Kellgren/Lawrence grade, % of patients 
 Grade 265
 Grade 335
Concomitant medications, % of patients 
50-foot walking time, seconds11.4 ± 3.9
Range of movement, degrees125.9 ± 11.1
Physician's global assessment, by VAS59.8 ± 3.1
Patient's global assessment, by VAS54.5 ± 3.7
Patient's assessment of pain, by VAS48.2 ± 5.0
SF-36 physical component scale37.1 ± 1.6
WOMAC Osteoarthritis Index 
 Pain59.2 ± 3.9
 Stiffness45.7 ± 4.7
 Function60.3 ± 3.9
 Total56.9 ± 3.9
Minimum JSW, mm, by Buckland-Wright method4.07 ± 1.01

Precision of the cartilage volume measurements.

Patient positioning and intrareader performance were to be precise, as shown by RMS CV% for the repeated measures of 2.2%, 1.6%, and 2.6% for the global, medial compartment, and lateral compartment cartilage volume, respectively (Table 2). These findings were very similar to data published by other research teams (21, 60).

Table 2. Coefficient of variation for measurements of knee cartilage volume in 8 healthy subjects and 8 OA patients*
Knee cartilage volume assessmentRMS CV%
Relative volume, %Absolute volume, mm3
  • *

    The osteoarthritis (OA) patients had Kellgren/Lawrence scores of 2–3. Values are the root mean square (RMS) coefficient of variation percentage (CV%) (intraobserver, interscan, image acquisition in 1 session) of the relative and absolute knee cartilage volumes.

Healthy subjects  
 Medial compartment1.248
 Lateral compartment3.1133
OA patients  
 Medial compartment1.676
 Lateral compartment2.678

Cartilage volume changes over time.

Already at 6 months, there were statistically significant losses in global (mean ± SD −3.8 ± 5.1%), medial compartment (–4.3 ± 6.5%) and lateral compartment (–3.3 ± 4.9%) cartilage volumes compared with baseline (P < 0.0001 for each comparison) (Table 3). Moreover, a statistically significant trend toward a continuous loss of cartilage volume was seen from month 6 to month 24 (P < 0.0001 for each comparison). At 24 months, there were statistically significant losses in the global (–6.1 ± 7.2%), medial compartment (–7.6 ± 8.6%) and lateral compartment (–5.0 ± 7.0%) cartilage volumes compared with baseline (P < 0.0001 for each comparison).

Table 3. Change (loss) in cartilage volume in the entire knee and the medial and lateral compartments over the 2-year followup in 32 patients with osteoarthritis*
Knee cartilage volume assessmentAssessment
6 months12 months18 months24 months
  • *

    Values are the mean ± SD. Cartilage volume loss at all time points was found to be statistically significant compared with baseline (P < 0.0001 by t-test).

% change from baseline    
 Global−3.8 ± 5.1−3.6 ± 5.1−4.4 ± 6.3−6.1 ± 7.2
 Medial compartment−4.3 ± 6.5−4.2 ± 7.5−5.6 ± 8.4−7.6 ± 8.6
 Lateral compartment−3.3 ± 4.9−3.1 ± 4.5−3.4 ± 5.4−5.0 ± 7.0
Amount of cartilage lost, mm3    
 Global−489.7 ± 664−472.7 ± 658−565.9 ± 814−795.7 ± 940
 Medial compartment−267.9 ± 409−260.6 ± 471−346.1 ± 528−473.6 ± 538
 Lateral compartment−221.8 ± 333−212.0 ± 301−219.8 ± 366−322.1 ± 475

Upon further evaluation using linear discriminant function analysis, 2 subgroups of disease progression were identified (Figures 1A–C). A subgroup of 11 patients clearly demonstrated a faster progression of global cartilage volume loss, showing a 6.2 ± 0.6% loss at 6 months, 6.4 ± 0.7% at 12 months, 8.1 ± 1.2% at 18 months, and 15.2 ± 0.8% at 2 years (Figure 1A). All of these changes were statistically significant (P < 0.0001). Almost no progression in cartilage volume loss was demonstrated in the remaining 21 patients (0.8 ± 0.3%, 0.6 ± 0.6%, 0.9 ± 0.5%, and 0.5 ± 0.4% loss of global cartilage volume at 6, 12, 18, and 24 months, respectively). The fast progressors had even greater relative cartilage volume loss for the medial compartment, with losses of 6.1 ± 0.8% at 6 months, 7.3 ± 1.1% at 12 months, 10.1 ± 1.2% at 18 months, and 17.9 ± 1.8% at 2 years (Figure 1B). All of these changes were also statistically significant (P < 0.0001). Similar to the findings for global cartilage volume loss, the slow progressors had very little progression of cartilage volume loss in the medial compartment. The fast progressors also had cartilage volume losses, although to a lesser degree, in the lateral compartment, with losses of 7.2 ± 0.6% at 6 months, 6.1 ± 0.7% at 12 months, 7.5 ± 1.2% at 18 months, and 12.9 ± 1.8% at 2 years (P < 0.0001) (Figure 1C).

Figure 1.

Changes in cartilage volume in the slow versus the fast progressors at different time points over the 2-year followup period. A, Global knee cartilage. B, Medial compartment cartilage. C, Lateral compartment cartilage. Values are the mean loss of cartilage volume (%). In the fast progressors, the loss of cartilage volume at all time points versus baseline was statistically significantly different (∗ = P < 0.0001 by t-test), regardless of the area evaluated.

These changes are further illustrated in Figure 2, with MRI maps showing changes in femoral cartilage thickness in a fast and a slow progressor. The global cartilage thickness map of the knee of a patient with fast disease progression from baseline shows clear changes at 2 years. The global cartilage thickness map of the knee of a patient with slow disease progression, however, shows almost no progression at 2 years.

Figure 2.

Changes in femoral cartilage thickness at 6, 12, 18, and 24 months in a patient with fast progression and a patient with slow progression of osteoarthritis of the knee, as demonstrated by magnetic resonance imaging. Red areas indicate greater loss of cartilage thickness, as shown in the fast progressor. Almost no progression is seen over the same time period in the slow progressor.

Baseline characteristics of fast versus slow progressors.

The baseline characteristics and clinical information for the fast and slow progressors are shown in Table 4. The fast progressor group had more women (73% versus 48%; P < 0.001), a higher BMI (32.2 ± 1.4 kg/cm2 versus 29.2 ± 0.9; P = 0.09), a reduced range of motion of the study knee (113.9 ± 10.1 degrees versus 131 ± 1.7; P < 0.01), and a greater circumference of the study knee, a possible surrogate for the presence of joint effusion compared with the slow progressor group (40.4 ± 0.07 cm versus 38.9 ± 0.6; P = 0.10). In addition, most of the relevant clinical assessments of OA were higher in the fast progressor group, as determined by the WOMAC questionnaire and the patient's and physician's global assessments. This group experienced more pain (49.9 versus 34.0; P = 0.05), stiffness (63.5 versus 43.2; P = 0.06), and loss of function, although to a lesser degree (43.4 versus 32.9; P = 0.18). Moreover, the patient's and physician's global assessments of the disease were worse at baseline in the group with fast progression of OA. Additional analyses of the entire cohort of 32 patients with regard to interactions between age, sex, weight, BMI, pain, stiffness, and use of NSAIDs revealed no significant results (data not shown).

Table 4. Characteristics and clinical data at baseline in patients with slow and fast progression of knee OA*
VariableOA progression groupP
Slow (n = 21)Fast (n = 11)
  • *

    Statistical analyses contrasting the group with slow progression of osteoarthritis (OA) and the group with fast progression of OA were performed with the nonparametric Wilcoxon 2-sample test. Values are the mean ± SD. VAS = visual analog scale (0–100 mm); SF-36 = Short Form 36; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index (VAS version).

 Age63.9 ± 1.764.8 ± 2.20.68
 Female, %48730.001
 Height, meters1.68 ± 0.021.65 ± 0.020.35
 Weight, kg82.1 ± 2.887.8 ± 4.10.46
 Body mass index, kg/cm229.2 ± 0.932.2 ± 1.40.09
 50-foot walking time, seconds10.2 ± 0.411.3 ± 0.80.37
 Study knee circumference, cm38.9 ± 0.640.4 ± 0.70.10
 Study knee range of motion, degrees131 ± 1.7113.9 ± 10.10.01
Clinical data   
 Patient's global assessment, by VAS37.5 ± 4.448.1 ± 7.00.08
 Physician's global assessment, by VAS35.1 ± 3.445.5 ± 5.20.10
 SF-36 global2.71 ± 0.052.6 ± 0.030.35
 WOMAC Osteoarthritis Index   
  Pain34.0 ± 4.449.9 ± 6.10.05
  Stiffness43.2 ± 5.963.5 ± 7.40.06
  Function32.9 ± 4.743.4 ± 6.80.18
  Total34.0 ± 4.746.4 ± 6.20.13

Comparison of changes in cartilage volume versus clinical data over time.

Evaluation of the clinical course in all 32 OA patients (Table 5) demonstrated no significant correlation between changes in cartilage volume and changes in clinical variables, such as the patient's and physician's global assessments, the 3 dimensions of the WOMAC (pain, stiffness, and function), and the physical components of the Short Form 36 health survey. Correlation values of r < 0.2 and P > 0.25 were found for all of the variables as compared with cartilage volume loss (data not shown).

Table 5. Spearman's correlation coefficients for loss of cartilage volume and changes in clinical parameters over time*
VariableCorrelation coefficient
6 months12 months18 months24 months
  • *

    VAS = visual analog scale (0–100 mm); SF-36 = Short Form 36; WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index (VAS version).

Patient and physician assessments, by VAS    
 Patient's global assessment−0.243−0.315−0.212−0.103
 Patient's assessment of pain0.193−0.015−0.037−0.122
 Physician's global assessment−0.068−0.349−0.331−0.316
 Physical function0.1920.120−0.245−0.341
 General health0.077−0.043−0.113−0.152
WOMAC Osteoarthritis Index    

Cartilage volume versus JSW.

Comparison of the cartilage volume in the medial compartment and the narrowest JSW obtained by radiography at baseline in 31 knee OA patients (Figure 3), revealed that some level of correlation exists between these 2 measurements (r = 0.46, P < 0.007). However, at 2 years, there was a striking difference in progression, as demonstrated by the global cartilage volume versus the JSW (Figure 4A) and the medial cartilage volume versus the JSW (Figure 4B). While 27 of the 31 patients demonstrated a loss of medial cartilage over 2 years by MRI, ∼50% of the patients with a JSW measurement at both baseline and year 2 showed a decrease in the minimum JSW. However, the mean JSW change among the 19 patients with primary medial disease and an osteophyte present at baseline was −0.13 mm, which illustrates the importance of considering this subgroup when measuring medial progression by JSW. Importantly, at 2 years, only 1 patient demonstrated disease progression according to the JSW measure showed an increase in cartilage volume by MRI.

Figure 3.

Scatter plot contrasting the baseline minimum joint space width, as measured by standardized radiography, with the baseline medial compartment cartilage volume, as measured by magnetic resonance imaging (MRI), in 31 patients with osteoarthritis of the knee. The correlation coefficient for the 2 baseline measurements was r = 0.46, P < 0.007.

Figure 4.

Scatter plots contrasting changes in A, the global cartilage volume and B, the medial cartilage volume, as measured by magnetic resonance imaging (MRI), versus changes in the joint space width, as measured by standardized radiography of the medial compartment, over 2 years in 31 patients with osteoarthritis of the knee. No correlation between the changes in cartilage volumes and changes in the joint space width was found.


Very few studies have examined changes in cartilage volume over time in a population of patients with static OA of the knee. In this longitudinal study of 32 patients with symptomatic knee OA evaluated with the use of MRI sets based on fat-suppressed, gradient-echo sequences, we demonstrated a significant loss in global cartilage volume of 6.1% at 2 years of followup (P < 0.0001). Moreover, such changes were clearly demonstrated statistically as early as 6 months after the start of study and had further increased at 18 and 24 months, reflecting a progression in the loss of cartilage volume over time. The rate of progression in our patients is similar to that identified by another group of investigators who specifically examined changes in tibial cartilage volumes in a younger population of patients with knee OA (42). In our study, the loss of cartilage volume in the medial compartment of the knee were even more striking and reflected a more rapid disease progression in this region. This may be explained in part by the higher weight-bearing pressure in this compartment or by baseline varus–valgus misalignment (61).

The changes seen over time exceed the variability of our quantitative MRI assessment, which has excellent reproducibility in knees affected by OA, with a coefficient of variation ranging from 1.8% to 2.5%, depending on the compartment being evaluated, and high interreader reliability, as previously demonstrated (36). The lack of linear progression of cartilage volume loss at 12 months is intriguing. Possible causes include cartilage edema (which may be encountered in the normal progression of OA), the potential presence of more-active disease at study entry (patients selected for study might have been more symptomatic initially), and/or the small number of study subjects (patient or instrument variability may have affected the results).

The identification of 2 subgroups of patients, those with fast and those with slow progression of cartilage loss, is quite striking and reflects the heterogeneity of the OA patient population. Although we selected patients based on the same inclusion and exclusion criteria, as dictated by most clinical trials in OA, there were still differences in disease progression among this population of patients with primary knee OA. Such heterogeneity has been described in previous clinical trials (62, 63), and our study reinforces those findings. Based on the findings in our OA cohort, fast disease progression, as assessed by quantitative MRI, may be associated with specific clinical variables at the time of study entry: being female, having a high body mass index, having a higher level of pain and stiffness, and having reduced joint mobility. These predictors make sense clinically, have previously been identified in major epidemiologic studies, and are sometimes in clinical trials to specifically select patients with higher levels of OA progression. These variables may identify a patient whose disease is likely to progress dramatically over time and in whom therapeutic intervention is even more important in order to prevent joint damage. It is not known, however, whether this population is at such a high risk of rapid and aggressive disease progression that therapeutic interventions may not be effective. Since this was an observational study, such questions remain unanswered and await further exploration.

We are currently conducting another study evaluating 20 normal individuals (10 between the ages of 25 and 35 years and 10 between the ages of 55 and 65 years) in whom knee cartilage volume was assessed at baseline and 1 year, has shown no loss of cartilage volume over time, even in older subjects (Raynauld JP, et al: unpublished observations). This finding suggests that the progression seen in our knee OA cohort is probably disease-related and not age-related. The slow progression of OA remains an enigma. Slow disease progression is unlikely to simply reflect the normal aging process, since these patients experienced pain and loss of function and met the ACR criteria for knee OA. Indeed, the entire cohort of patients demonstrated loss of cartilage volume compared with normal subjects. It is possible that the slow progressors or nonprogressors may constitute a subgroup of patients whose knee OA may be in a quiescent phase. This obviously brings to light the importance of gathering more longitudinal information on the natural course of knee OA.

Measurement of changes in the JSW at its minimum on standardized radiographs of the knee over time as a surrogate for knee cartilage evaluation is considered by many to be the best available method for evaluating the anatomic progression of OA. There was little correlation between baseline values of JSW in the medial compartment and concomitant values for global cartilage volume (r = 0.28, P not significant) (data not shown), with a slightly better correlation between the baseline JSW and cartilage volume in the medial compartment (r = 0.46, P < 0.007). One difference between these methods is that JSW is measured on radiographs of the knee during weight-bearing, which quantifies the thickness of the tissue under compression, whereas MRIs of the knee are performed with the patient in a supine position. In this position, the knee joint is not under load, and the cartilage is not compressed and is in a relaxed state. Additional research should be conducted to determine further, more subtle differences between these methods.

The presence of a wider joint space was related to the presence of a higher cartilage volume, thus demonstrating the validity of both approaches to the assessment of cartilage damage. Our data showing that ∼50% of the patients with reductions in the JSW at 1 and 2 years had no loss of cartilage volume could be explained by the presence of a subgroup of patients who did not have primary medial compartment disease or who had an osteophyte at baseline, or by the presence of a subgroup of patients in whom the OA was in a quiescent phase. The radiologic findings contrast with the MRI findings, which showed a loss of cartilage volume over time in the vast majority of patients. There are many possible explanations for the discrepant results of these 2 imaging techniques. Each method provides a different measure of the status of the articular cartilage. Our results also showed that the methods are not strongly related to each other. Since 9 of the patients did not have primary medial compartment disease at baseline (based on the semiflexed radiographic view), the potential for identifying OA progression by measuring JSW in the medial compartment was attenuated because disease is not yet committed to the medial compartment. In view of the small size of our patient cohort, it would be appropriate to wait for the results of larger studies (currently under way) before making more-definitive statements. Nevertheless, our study provides an initial comparison between MRI assessments of cartilage volume loss and standardized radiologic methods for determining the progression of knee OA, a comparison seldom done in the past.

As mentioned above, there are obviously some limitations to this study. Caution is always advised when extrapolating results obtained in relatively small cohorts of subjects. However, our cohort is representative of the average patient population with typical knee OA seen at a rheumatology clinic. We also believe that our study is of great interest as a “first look” into the results of future studies of larger cohorts. Since this study used noninvasive techniques, caution is advised with regard to clinical interpretation and external validity of these results in the absence of other, invasive evaluation methods, such as arthroscopy.

Even though no medication for slowing the disease progression is yet available, it is still unclear to what extent such medication may affect the structural progression of OA and whether this effect would translate clinically. It is also not clear which subpopulation might benefit the most from treatment with disease-modifying OA drugs. The obvious bias would be in favor of treating patients who have rapidly progressing disease, since these patients are more likely to require surgical intervention.

Our study considered changes in the global and compartmental cartilage volumes. However, the potential of quantitative MRI was not fully utilized. Since OA is a local disease that does not affect all areas of the cartilage at the same rate, smaller areas of the knee cartilage, for example the central portion of the medial femoral condyle and its corresponding area of the tibial plateau, may show even greater relative changes in proportion to other areas of cartilage. We deliberately chose to assess greater areas of cartilage in order to avoid the problem of selecting a priori an area of greater relative change that may differ substantially from patient to patient. Finally, there was no washout period for the analgesic medications (NSAIDs, etc.) prior to the clinical evaluation, which may partly explain the apparent lack of correlation between the clinical and the cartilage volume findings. A washout period for analgesic/antiinflammatory medication is now a standard procedure requested by the regulatory agencies for studies assessing drugs that may affect OA progression clinically as well as structurally (64).

In conclusion, our study demonstrated the feasibility of long-term longitudinal followup of changes in cartilage volume over time in patients with OA. Loss of cartilage volume loss in the knee, as demonstrated by quantitative MRI, was demonstrated as early as 6 months in our small, yet typical, OA cohort, which would imply that this imaging approach is more sensitive to change than is standardized radiography. Clinical variables may help to identify subgroups of patients who are at risk of more rapid disease progression, which may affect the selection of patients for clinical trials. The use of quantitative MRI to measure cartilage degradation should, hopefully, reduce the number of patients needed for clinical trials evaluating disease-modifying OA drugs, as well as improve patient retention in such trials by reducing the length and overall cost of the trials.


We would like to thank Rupert Ward (King's College, London, UK) for measuring the JSW on our radiographs. We would also like to thank Raymonde Grégoire and France Frenette for providing outstanding patient support and Lorraine Doré for assistance in manuscript preparation.