Dr. Wluka is the recipient of a National Health and Medical Research Council scholarship.
Comparison of tibial cartilage volume and radiologic grade of the tibiofemoral joint
Version of Record online: 28 FEB 2003
Copyright © 2003 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 48, Issue 3, pages 682–688, March 2003
How to Cite
Cicuttini, F. M., Wluka, A. E., Forbes, A. and Wolfe, R. (2003), Comparison of tibial cartilage volume and radiologic grade of the tibiofemoral joint. Arthritis & Rheumatism, 48: 682–688. doi: 10.1002/art.10840
- Issue online: 28 FEB 2003
- Version of Record online: 28 FEB 2003
- Manuscript Accepted: 2 DEC 2002
- Manuscript Received: 28 JUN 2002
- National Health and Medical Research Council of Australia
To compare tibial cartilage volume as measured by magnetic resonance imaging (MRI) with radiologic assessment of the tibiofemoral joint.
The MRI-determined tibial cartilage volume was compared with the radiologic grade of individual features of osteoarthritis (osteophytes and joint space narrowing [JSN]) in 252 subjects (mean ± SD age 60.2 ± 10 years, 62% female) who were participating in studies of knee cartilage.
JSN seen on both medial and lateral radiographs of the tibiofemoral joint was inversely associated with the respective tibial cartilage volume. This inverse relationship was strengthened with adjustment for age, sex, body mass index (BMI), and bone size. After adjustment for these confounders, for every increase in JSN grade (0–3), the medial tibial cartilage volume was reduced by 257 mm3 (95% confidence interval [95% CI] 193–321) and the lateral tibial cartilage volume by 396 mm3 (95% CI 283–509). The relationship between mean cartilage volume and radiologic grade of JSN was linear. Based on results in the subgroup of subjects with normal radiographic findings, we have proposed a model to estimate average “normal” cartilage volume in men and women for a given age, BMI, and bone size.
The results of this study demonstrate a strong negative, linear association between medial and lateral tibial cartilage volume and increasing grade of JSN. Using data from radiographically normal subjects, we have proposed a simple model for estimating “normal” cartilage volume. However, larger studies will be needed to confirm these findings and to determine whether they are valid in younger subjects.
Osteoarthritis (OA) is a common cause of disability in people age >65 years (1). Development of treatments for OA is limited by the lack of a noninvasive method that is reproducible and accurate in measuring early disease and disease progression. Change in width of the medial tibiofemoral joint space has been recommended, by expert consensus, as the primary measure of the effect of biologic treatments in OA (2). However, precise measurement of this variable is dependent on standardized radiographic techniques and could be biased when the presence of pain impairs the ability of the subject to position the knee (3).
There has been increasing interest in the use of magnetic resonance imaging (MRI) in the measurement of knee cartilage volume as a possible outcome measure in arthritis (4–7). Measurement of tibial cartilage has been shown to be a valid indicator of cartilage volume when MRI-determined cartilage volume is compared with findings of anatomic dissection and has been found to be reproducible, with coefficient of variations of <5% (4–7). This validity and reproducibility have been documented both in healthy subjects and in patients (8, 9).
There are a number of potential advantages of measuring articular cartilage volume rather than using radiologic assessment of joint space narrowing (JSN). This measurement provides a direct measure of joint cartilage, rather than an indirect measure via radiographs. Since the whole 3-dimensional (3-D) structure is examined, the potential problem of reselecting identical locations for followup, as occurs with knee radiography, is reduced. Small positional changes from one examination to the next may affect the reproducibility of JSN, particularly in longitudinal studies, and this should be minimized by measuring the entire 3-D structure (10). There is also evidence that mild-to-moderate joint space loss may reflect meniscal extrusion rather than joint cartilage erosion (11).
The work performed to date suggests that MRI will provide a very sensitive and reproducible method for measuring cartilage. We have previously shown a strong correlation between tibial and femoral cartilage findings in the medial and lateral tibiofemoral joint compartments (12). In all of the current clinical and epidemiologic work on knee OA, the knee joint is examined as being composed of 3 joints: the medial and lateral tibiofemoral joints and the patellofemoral joint. Any MRI cartilage measure needs to take this into account in order to be useful in assessment of disease stage. Since the femoral cartilage articulates with the 3 joints (the medial and lateral tibiofemoral joints and the patellofemoral joints), it is more difficult to clearly identify the relevant component of the femoral joint when assessing the medial and lateral tibiofemoral joints. In contrast, each of the tibial cartilages forms only part of 1 joint, and for this reason we have used the tibial cartilage in the present study. MRI of this area has the potential of being a useful and efficient method for assessing joint cartilage at the tibiofemoral joint. There has been information published to date on how tibial cartilage volume relates to radiologic grade of medial and lateral tibiofemoral OA, the latter of which has been used in most epidemiologic and clinical studies of OA. In this study, we compared tibial cartilage volume as measured by MRI with radiologic assessment of the tibiofemoral joint.
PATIENTS AND METHODS
Two hundred fifty-two subjects age >40 years who had radiographic and MRI assessment of the same knee within 1 month were recruited from among subjects participating in studies on knee cartilage at our department. The study was approved by the ethics committee of the Alfred Hospital. The subjects with knee OA were recruited by using a combined strategy including advertising through local newspapers and the Victorian branch of the Arthritis Foundation of Australia, as well as collaboration with general practitioners, specialist rheumatologists, and orthopedic surgeons. Healthy subjects were recruited through advertising in newspapers, through sporting clubs, and through the hospital staff association. Subjects were excluded if they had any form of arthritis other than OA (including evidence of chondrocalcinosis on plain radiographs or evidence of focal cartilage lesion on MRI suggesting a posttraumatic etiology), a contraindication to MRI (e.g., pacemaker, cerebral aneurysm clip, cochlear implant, presence of shrapnel in strategic locations, metal in the eye, or claustrophobia), hemiparesis of either lower limb, or planned total knee replacement.
Weight was measured to the nearest 0.1 kg (shoes and bulky clothing removed), using a single pair of electronic scales. Height was measured to the nearest 0.1 cm (shoes removed), using a stadiometer. Body mass index (BMI) (weight/height2, kg/m2) was calculated. MRI was performed on 1 knee of each subject. This was either the dominant knee (defined as the lower limb from which the subject steps off when walking) or the knee with less severe symptoms of OA.
Each subject had a weight-bearing radiograph performed on the same knee for which MRI was performed, taken with the knee in full extension according to recommended protocol (13). The knee radiographs were obtained with the patient standing, the weight equally distributed to both feet, and the back of the knee positioned as near as possible to the vertical cassette. The lower limb was rotated so that the tibial spines appeared centrally placed relative to the femoral notch, using fluoroscopy. The central ray of the x-ray beam was centered on the joint space and inclined downward to ensure that the medial tibial plateau was parallel to the x-ray beam.
All radiographs were scored independently by 2 trained observers who used a published atlas to classify disease in the tibiofemoral joint (13). The radiologic features of tibiofemoral OA were graded on a 4-point scale (0–3) for individual features of osteophytes and JSN, where 0 = no disease and 3 = most severe disease (13). In the case of disagreement between observers, the 2 observers together with a third independent observer reviewed the radiographs. Intraobserver reproducibility was 0.90 for osteophytes and 0.86 for JSN; interobserver reproducibility was 0.86 for osteophytes and 0.82 for JSN (kappa statistic).
All knees investigated in this study were imaged in the sagittal plane on the same 1.5T whole-body MR unit (Signa Advantage HiSpeed; GE Medical Systems, Milwaukee, WI), using a commercial transmit-receive extremity coil. The following sequence and parameters were used: a T1-weighted fat-suppressed 3-D gradient recall acquisition in the steady state; flip angle 55°; repetition time 58 msec; echo time 12 msec; field of view 16 cm; 60 partitions; 512 × 196 matrix; one acquisition time 11 minutes, 56 seconds. Sagittal images were obtained at a partition thickness of 1.5 mm and an in-plane resolution of 0.31 × 0.82 mm (512 × 196 pixels). Knee cartilage volume was determined by means of image processing on an independent work station, using the software program Osiris as previously described (7, 14). Two trained observers read each MRI. The coefficients of variation for total, medial, and lateral cartilage volume measures were 2.6%, 3.4%, and 2.0%, respectively (7, 14).
Medial and lateral tibial plateau areas were determined by creating an isotropic volume from the input images, which was reformatted in the axial plane. Areas were directly measured from these images. Coefficients of variation for the medial and lateral tibial plateau areas were 2.3% and 2.4%, respectively (14).
Linear regression was used to examine the relationship between radiologic features of medial and lateral tibiofemoral OA (JSN and osteophytes) and medial and lateral tibial cartilage volumes in univariate analyses. A multivariate model was used to adjust for the effects of age, sex, BMI, and bone size (as measured by tibial plateau area) on medial and lateral tibial cartilage volume. Results are presented as regression coefficients that represent differences in medial and lateral tibial cartilage volume per unit difference in radiologic grade, while other factors are held constant (i.e., controlled for). The assumption of a coefficient to represent a constant change in cartilage volume for each 1-unit change in radiologic grade was tested by adding to the model 2 terms that estimated departures from this “linear trend” across the 4 grades. The statistical significance of these 2 terms was assessed using F statistics.
Residuals from the regression of cartilage volume on age, sex, BMI, and respective plateau area represent the component of cartilage volume not explained by these 4 factors. We added to these residuals the mean cartilage volume of the respective joint compartment (1,788 mm3 for the medial compartment and 2,134 mm3 for the lateral compartment) and plotted these “adjusted” cartilage volumes against radiologic grade of JSN and osteophytes for the respective joint compartment. Subgroup analysis was performed on the subjects who were radiologically normal, defined as having no evidence of JSN or osteophytes in either the medial or lateral tibiofemoral joint.
The characteristics of the study population, including prevalence of radiologic features of tibiofemoral joint disease, are presented in Table 1. Univariate analyses showed that grade of JSN on both the medial and lateral tibiofemoral radiographs was inversely associated with the respective tibial cartilage volume. Figure 1 illustrates the relationship between radiologic measure of medial and lateral tibiofemoral OA and “standardized” medial and lateral tibial cartilage volume (calculated for each patient and adjusted for age, sex, BMI, and bone size). After adjustment for the above confounders, for every increase in JSN grade the medial tibial cartilage volume and the lateral tibial cartilage volume were significantly reduced, by 257 mm3 and 396 mm3, respectively. Although univariate analysis revealed no relationship between grade of osteophytes and cartilage volume, for every increase in grade of lateral tibiofemoral osteophytes the lateral tibial cartilage volume was significantly reduced by 255 mm3, after adjustment. There was a reduction of 77 mm3 in medial tibial cartilage volume for every increase in grade of medial tibiofemoral osteophytes, but this finding was only of borderline statistical significance. (Table 2).
|Age, years (n = 251)||60.2 ± 10|
|No. (%) female (n = 252)||157 (62)|
|Body mass index, kg/m2 (n = 251)||27.7 ± 5.3|
|No. (%) with medial tibiofemoral osteophyte score ≥1 (n = 248)||48 (19)|
|No. (%) with medial tibiofemoral JSN score ≥1 (n = 248)||100 (40)|
|No. (%) with lateral tibiofemoral osteophyte score ≥1 (n = 248)||67 (27)|
|No. (%) with lateral tibiofemoral JSN score ≥1 (n = 249)||47 (19)|
|Medial tibial cartilage volume, ml (n = 238)||1.79 ± 0.52|
|Lateral tibial cartilage volume, ml (n = 239)||2.13 ± 0.67|
|Univariate analysis, regression coefficient*||Multivariate analysis†|
|Regression coefficient*||95% confidence interval||P|
|Medial tibiofemoral JSN||−118||−257||(−321, −193)||<0.001|
|Medial tibiofemoral osteophytes||−81||−77||(−156, 3)||0.059|
|Lateral tibiofemoral JSN||−477||−396||(−509, −283)||<0.001|
|Lateral tibiofemoral osteophytes||−248||−255||(−336, −175)||<0.001|
The analyses presented in Table 2 assume that mean cartilage volume changes in a linear manner by the same amount with every increase in radiologic grade of JSN or osteophytes from grade 0 to grade 3. We tested this assumption against an alternative that did not assume that this relationship would be linear. For all 4 analyses presented in Table 2, there was no statistically significant improvement in the fit of our models to the observed data upon moving from the linear model to the general model. This suggests that the negative association between increasing grade of radiologic features of OA (JSN and osteophytes) is linear.
The mean medial and lateral tibial cartilage volumes and the effect of age, BMI, and bone size (measured by tibial plateau area) were examined in subjects whose joints were radiologically normal and who had no evidence of any features of OA of the tibiofemoral joint. The data, presented separately for male and female subjects, are shown in Table 3. The regression coefficients in Table 3 can be used to calculate the mean cartilage volume for healthy men and women of different ages, BMI, and bone sizes. For example, based on this model, men age 58 years, with a BMI of 26.2 kg/m2 and a medial tibial plateau area of 1,934 mm2 and lateral tibial plateau area of 1,390 mm2 (the mean values in this study population), would have an average medial cartilage volume of 2,440 mm3 (95% confidence interval [95% CI] 2,270–2,610) and lateral cartilage volume of 3,070 mm3 (95% CI 2,850–3,280), and women age 57 years, with a BMI of 25.9 kg/m2, a medial tibial plateau area of 1,648 mm2, and a lateral tibial plateau area of 1,048 mm2, would have an average medial cartilage volume of 1,610 mm3 (95% CI 1,540–1,680) and lateral cartilage volume of 2,020 mm3 (95% CI 1,920–2,110).
|Medial cartilage volume||Lateral cartilage volume|
|Regression coefficient†||95% confidence interval||Regression coefficient†||95% confidence interval|
|Male subjects (n = 36)|
|Mean cartilage volume (mm3)||2,440†||(2,270, 2,610)||3,070†||(2,850, 3,280)|
|Age (per year above 58 years)||−9||(−25, 6)||−21||(−37, −4)|
|BMI (per kg/m2 above 26.2 kg/m2)||−19||(−66, 29)||−60||(−116, −4)|
|Plateau area (per mm2 above 1,934 mm2 for medial and above 1,390 mm2 for lateral)||−0.08||(−0.66, 0.51)||0.41||(−0.53, 1.36)|
|Female subjects (n = 65)|
|Mean cartilage volume (mm3)||1,610†||(1,540, 1,680)||2,020†||(1,920, 2,110)|
|Age (per year above 57 years)||−2||(−14, 10)||−9||(−24, 6)|
|BMI (per kg/m2 above 25.9 kg/m2)||2||(−16, 20)||−0.2||(−23, 23)|
|Plateau area (per mm2 above 1,648 mm2 for medial and above 1,048 mm2 for lateral)||0.79||(0.37, 1.22)||0.35||(−0.36, 1.06)|
To obtain a prediction for an individual, the sampling variability quantified by these confidence intervals must be combined with the person-to-person natural variability in cartilage volume that is quantified by the mean squared error from the regression models. For an individual man with our “average” characteristics, the prediction interval is 1,450–3,430 mm3 for medial cartilage volume and 1,820–4,320 mm3 for lateral cartilage volume. For an individual woman with our “average” characteristics, the prediction intervals for medial cartilage volume and lateral cartilage volume are 1,040–2,180 mm3 and 1,280–2,750 mm3, respectively.
We have shown a significant, negative association between medial and lateral cartilage volume and radiologic grade of JSN. The medial tibial cartilage volume was reduced by 257 mm3 per increase in grade of JSN (0–3) and the lateral tibial cartilage volume was reduced by 396 mm3 per increase in grade, after adjustment for confounders. The relationship between cartilage volume and osteophyte grade was less clear. In the lateral compartment, the lateral tibial cartilage volume was reduced by 255 mm3 for every increase in grade of JSN. For every increase in osteophyte grade at the medial tibiofemoral joint there was a reduction of 77 mm3 in medial tibial cartilage volume, but this was of borderline statistical significance. Based on the subgroup of subjects whose radiologic findings were normal, we have also proposed a model to estimate average “normal” cartilage volume in men and women of a given age, BMI, and bone size.
There has been no published information on the relationship between the grade of the feature of OA, particularly JSN, and cartilage volume. Most of our knowledge of the epidemiology of OA at the tibiofemoral joint has been based on radiologic assessment of the joint. There is increasing interest in the use of cartilage volume in the assessment of structural change in joints. We have recently shown that knee cartilage volume is lost at the rate of 5% per year in patients with OA (15). Radiographic assessment of joints and measurement of joint cartilage volume are likely to have complementary roles in the investigation factors that affect the risk for OA. For example, although radiographic assessment is useful in defining the presence or absence of OA, it has major limitations in the examination of subjects with very early OA that might not have developed sufficiently to be detected radiographically. It would be expected that with an increase in radiologic grade of JSN there would be a corresponding reduction in articular cartilage volume, as we found. Although it has often been assumed that JSN may act as a proxy measure for articular cartilage, it has been recently shown that changes in JSN may relate to meniscal extrusion (10). Despite this, expert consensus suggests that JSN be used as a surrogate marker of articular cartilage volume, as an indicator of structural change in OA (2). The less consistent relationship between joint cartilage and osteophyte grade may be explained by the fact that osteophytes are measuring a different pathogenic feature of OA.
Our data suggest that there is a linear reduction in cartilage volume with increasing radiologic grade of JSN. This may be important in terms of understanding of the pathogenic process in OA and may have implications for recognition of early disease and prevention of disease progression. It may also help in the debate on whether the subgroup of subjects with radiologically more severe OA, for example, those with grade 2 JSN, may still benefit from chondroprotective measures. Our results indicate that patients with grade 2 JSN still have significant amounts of articular cartilage.
There has not been published information available regarding what might be normal cartilage volume. Based on the data from the 36 male and 65 female subjects with normal radiologic findings in this study and on the growing data on factors that affect cartilage volume, we have proposed a model to estimate expected cartilage volume in a healthy subject. We examined men and women separately since there is considerable data suggesting a significant sex effect on cartilage volume (7, 16). Bone size was included because it has been shown to relate to articular cartilage volume (17, 18) and BMI (19). Based on this model we calculated the predicted average medial and lateral tibial cartilage volume in radiologically normal men and women of a given age, BMI, and bone size. Although this model is based on relatively small numbers of radiologically normal subjects, the data are a starting point for interpreting cartilage volume results. It is important to note that the 95% confidence intervals from Table 3 do not give an idea of uncertainty when predicting for an individual man or woman, but rather the uncertainty when predicting the average for men and women of that age, BMI, and bone size. To obtain a prediction for an individual, the sampling variability quantified by these confidence intervals must be combined with the person-to-person natural variability in cartilage volume, which is quantified by the mean squared error from the regression models.
There are a number of potential limitations in using MRI for cartilage volume estimates. The accurate delineation of articular cartilage depends on high contrast relative to adjacent tissues. We therefore used a previously validated fat-suppressed gradient-echo sequence (4, 20). Furthermore, as has been recommended previously (6), in order to improve in-plane resolution we used a matrix of 512 × 196 pixels, resulting in an in-plane resolution of 0.31 × 0.82 mm (12). Some of our data, especially for grade 3 radiologic JSN and osteophytes, is based on small numbers of patients and will need to be confirmed using a larger sample. For the sake of this study, we defined “normal” subjects as radiologically normal with no evidence of osteophytes or JSN in either the medial or lateral tibiofemoral joints. However, some consensus will need to be reached about what constitutes a “normal” subject for such studies. Simply selecting a random sample of community-based subjects is likely to pose problems given that the prevalence of radiologic knee OA in this population would be expected to be high.
The role of MRI in evaluation of the knee joint is still in the early stages. Radiographic assessment of a joint and MRI examination of joint cartilage are likely to provide complementary information about joint structure in OA. MRI volumetric assessment may not be appropriate for monitoring specific cartilage lesions. However, there are a number of potential advantages of MRI over radiographic assessment of joints for studies of disease progression and in examination of joint structure in early OA or prior to development of radiologically detectable knee OA. MRI measures cartilage directly, rather than indirectly as with radiography. MRI measurement may be less subject to positional variation since the 3-D structure can be measured with thin image separations (6). There is also evidence that mild-to-moderate joint space loss may reflect meniscal extrusion rather than joint cartilage erosion (11). Direct measurement of joint cartilage using MRI avoids this problem.
The use of MRI in the management of OA raises questions of feasibility. Currently, MRI is an expensive imaging modality with limited accessibility. However, it is a very versatile tool and can be used to address a number of issues in knee OA. It allows noninvasive examination of articular and meniscal knee cartilage, internal knee structures, and bone changes. Each of these structures may contribute independently to the clinical picture of knee OA. However, comprehensive MRI of the knee to image each of these adequately to enable study would require ∼30 minutes of imaging time. Since the cost of MRI is proportional to imaging time, inclusion of all of these structures would more than triple the cost of a study such as ours. For some purposes, it may not be necessary to examine all of them. For example, if measurement of cartilage volume is found to be useful for assessing the stage of OA and for monitoring progression of disease either clinically or in clinical trials, more limited scanning sequences could be used. In the current study we used standard MRI sequences which take <10 minutes. Another potential limitation to the widespread use of MRI-determined cartilage volume in the evaluation of OA is that the measurement technique is not fully automated and therefore requires significant experience and training to achieve accurate and reproducible results.
We believe studies such as the present one, which examine the role of limited components of the knee such as tibial cartilage, will provide important information about the most efficient way of assessing the state of joints. This is analogous to the early work in the area of osteoporosis, which was aimed at identifying which components of bone mineral density were the most efficient in providing clinically useful information. It may be that in the future, when more is known about structural change in OA, the MRI may assist in the routine clinical assessment of the disease.
In summary, we have demonstrated a strong negative association between medial and lateral tibial cartilage volume and increasing grade of JSN. This relationship appears to be linear. We have developed a simple model for estimating average “normal” cartilage volume for men and women of a given age, BMI, and bone size. Larger studies will be needed to confirm these findings and to determine whether they are valid with regard to subjects under the age of 40 years.
The authors extend thanks to Ms Judy Hankin and Ms Judy Snaddon for coordinating the recruitment of participants for this study and special thanks to the study participants who made the work possible.