The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Menzies Research Institute, Private Bag 23, Hobart, Tasmania 7000, Australia. E-mail: firstname.lastname@example.org
Objective: To describe the associations among BMI, knee cartilage morphology, and bone size in adults.
Research Methods and Procedures: A cross-sectional convenience sample of 372 male and female subjects (mean age, 45 years; range, 26 to 61 years) was studied. Knee articular cartilage defect score (0 to 4) and prevalence (defect score of ≥2), volume, and thickness, as well as bone surface area and/or volume, were determined at the patellar, tibial, and femoral sites using T1-weighted fat-saturation magnetic resonance imaging. Height, weight, BMI, and radiographic osteoarthritis were measured by standard protocols.
Results: In multivariate analysis in the whole group, BMI was significantly associated with knee cartilage defect scores (β: +0.016/kg/m2 to +0.083/kg/m2, all p < 0.05) and prevalence (odds ratio: 1.05 to 1.12/kg/m2, all p < 0.05 except for the lateral tibiofemoral compartment). In addition, BMI was negatively associated with patellar cartilage thickness only (β = −0.021 mm/kg/m2; p = 0.039) and was positively associated with tibial bone area (medial: β = +7.1 mm2/kg/m2, p = 0.001; lateral: β = +3.2 mm2/kg/m2, p = 0.037). Those who were obese also had higher knee cartilage defect severity and prevalence and larger medial tibial bone area but no significant change in cartilage volume or thickness compared with those of normal weight.
Discussion: This study suggests that knee cartilage defects and tibial bone enlargement are the main structural changes associated with increasing BMI particularly in women. Preventing these changes may prevent knee osteoarthritis in overweight and obese subjects.
Osteoarthritis (OA)1 is a common condition of multifactorial origin. There is controversy in the literature about how best to diagnose OA. Most current classification systems use a combination of pain and radiographic criteria (which include both joint space narrowing and osteophytosis). Joint space narrowing largely reflects cartilage loss, whereas osteophytosis reflects marginal bony joint expansion. It is uncertain which of these is the initial step. It is well established that BMI is most strongly linked to knee OA (both symptoms and radiographic change) (1,2). Radiographic knee OA is increased nearly 4-fold in obese women (2), and there is a dose—response relationship between body weight and knee OA (3). However, the underlying mechanism remains obscure. Being overweight is associated with increases in the amount of force across a weight-bearing joint (4,5,6) and cartilage turnover biomarkers (7,8). In addition, adipose tissue may produce changes in hormone and growth factors (9). These biomechanical, biochemical, and metabolic changes may lead to loss of articular cartilage and sclerosis of subchondral bone; however, there is limited direct evidence to support this. This is, in part, because of the limitations of conventional radiographic measurement, which provides only an approximation of articular cartilage with measurement of joint space narrowing and poor characterization of other tissue. In contrast, magnetic resonance imaging (MRI) can visualize joint structure directly and is recognized as a valid, accurate, and reproducible tool to measure articular cartilage defects (10,11,12,13,14), volume, thickness, and subchondral bone size (15,16,17,18,19,20). However, the results from early MRI studies on the association between knee cartilage and BMI are contradictory because they are generally from small samples (19,21,22). Cartilage defects (G. Jones, unpublished data) and bone size (23) may also be important in the pathogenesis of knee OA, but there is little information on the associations among BMI, knee cartilage defects, and knee subchondral bone size. The aim of this study, therefore, was to describe the associations among BMI, knee cartilage defects, volume, thickness, and bone size in a large convenience sample of adult men and women.
Research Methods and Procedures
The study was carried out in Southern Tasmania, primarily in the capital city of Hobart, from June 2000 to December 2001. It was approved by the Southern Tasmanian Health and Medical Human Research Ethics Committee, and all subjects provided informed written consent.
A convenience sample was used for this study. Subjects were selected from two sources. One-half of the subjects were the adult children of subjects who had a knee replacement performed for primary knee OA at any Hobart hospital in the years 1996 to 2000 (cases). This diagnosis was confirmed by reference to the medical records of the orthopedic surgeon and the original radiograph where possible. The other one-half were randomly selected controls. These were selected by computer-generated random numbers from the most recent version of the electoral roll (2000). Subjects from either group were excluded on the basis of contraindication to MRI (including metal sutures, presence of shrapnel, iron filing in eye, and claustrophobia). No women were on hormone replacement therapy at the time of the study.
Weight was measured to the nearest 0.1 kg (with shoes, socks, and bulky clothing removed) using a single pair of electronic scales (Seca Delta Model 707, Seca, Germany) that were calibrated using a known weight at the beginning of each clinic. Height was measured to the nearest 0.1 cm (with shoes and socks removed) using a stadiometer. BMI was calculated. Overweight was defined as a BMI ≥25≤30 kg/m2, whereas obesity was defined as a BMI >30 kg/m2. Data on type and location of adiposity were not collected. Objective measures of physical function included measurement of muscle strength by dynamometry at the lower limb (involving both legs simultaneously). The subject was instructed in each technique before testing, and each measure was performed twice. Repeatability estimates (Cronbach's α) were 0.91. The devices were calibrated by suspending known weights at regular intervals. Physical work capacity was also assessed by a bicycle ergometer (24). Subjects were asked to cycle at a constant 60 rpm for 3 minutes each at three successively increasing but submaximal workloads. Heart rate was recorded at 1-minute intervals at each workload using an electric heart rate monitor. Physical work capacity at 170 beats/min (PWC170) was assessed by linear regression, with extrapolation of the line of best fit to a heart rate of 170 beats/min. The PWC170 was not considered a technically adequate measure unless subjects had spent a minimum of 2 minutes at each workload and the pulse rate increased by at least 5 beats/min with increasing workloads. Repeatability was not assessed in our subjects but has previously been reported as an intraclass correlation coefficient (ICC) of 0.92 (24).
A standing anteroposterior semiflexed view of the right knee was performed in all subjects. Radiographs were assessed using the Altman atlas (25). Each of the following was assessed: medial joint space narrowing (0 to 3), lateral joint space narrowing (0 to 3), medial osteophytes (femoral and tibial combined; 0 to 3), and lateral osteophytes (femoral and tibial combined; 0 to 3). Each score was arrived at by consensus with two readers (G.J. and F.S.) simultaneously assessing the radiograph with immediate reference to the atlas. Reproducibility was assessed in 50 radiographs, 2 weeks apart, and yielded an ICC of 0.99 for osteophytes and 0.98 for joint space narrowing (20).
Knee Cartilage Volume and Thickness Measurement
An MRI scan of the right knee was performed. Knee cartilage volume and thickness were determined by means of image processing on an independent work station using the software program Osiris as previously described (15,16). Knees were imaged in the sagittal plane on a 1.5-T whole body magnetic resonance unit (Picker) with use of a commercial transmit-receive extremity coil. The following image sequence was used: a T1-weighted fat saturation 3-day gradient recall acquisition in the steady state; flip angle 55°; repetition time 58 ms; echo time 12 ms; field of view 16 cm; 60 partitions; 512 × 512 matrix; acquisition time, 11 minutes 56 seconds; one acquisition. Sagittal images were obtained at a partition thickness of 1.5 mm and an in-plane resolution of 0.31 × 0.31 (512 × 512 pixels). The image data were transferred to the workstation. The volumes of individual cartilage plates (medial tibial, lateral tibial, and patella) were isolated from the total volume by manually drawing disarticulation contours around the cartilage boundaries on a section-by-section basis. These data were resampled by means of bilinear and cubic interpolation (area of 312 × 312 μm and 1.5 mm thickness, continuous sections) for the final 3-day rendering. The volume of the particular cartilage plate was determined by summing all of the pertinent voxels within the resultant binary volume. Femoral cartilage volume was not assessed because we have found that two tibial sites and the patella site correlate strongly with this site. The patellar, medial, and lateral tibial cartilage thicknesses were measured using calipers on all sections, and the maximum thickness of any section was recorded independently. Using this method, we had high intra- and interobserver reproducibility. The coefficient of variation (CV) for cartilage volume measures was 2.1% for medial tibial, 2.2% for lateral tibial, and 2.6% for patella. The CV for cartilage thickness was 2.3% for the lateral tibia and 2.2% for the medial tibia (15).
Cartilage Defect Assessment
Cartilage defects were graded on the above MR images with a modification of a previous classification system (10,11,12) at medial tibial, medial femoral, lateral tibial, lateral femoral, and patellar sites as follows: grade 0, normal cartilage; grade 1, focal blistering and intracartilaginous low-signal intensity area with an intact surface; grade 2, irregularities on the surface or bottom and loss of thickness of <50%; grade 3, deep ulceration with loss of thickness of >50%; grade 4, full-thickness chondral wear with exposure of subchondral bone. We found that cartilage surface in some images was still regular, but cartilage adjacent to subchondral bone became irregular, so we included these changes in the classification system. A cartilage defect also had to be present in at least two consecutive slices. The cartilage was considered to be normal if the band of intermediate signal intensity had a uniform thickness. The cartilage defects were regraded 1 month later, and the average scores of cartilage defects at medial tibiofemoral (0 to 8), lateral tibiofemoral (0 to 8), patellar (0 to 4), and whole compartments (0 to 20) were used in the study. A prevalent cartilage defect was defined as a cartilage defect score of ≥2 at any site of that compartment. Intraobserver reliability (expressed as ICC) was 0.90 for the medial tibiofemoral compartment, 0.89 for the lateral tibiofemoral compartment, and 0.94 for the patellar compartment. Interobserver reliability was assessed in 50 MR images and yielded an ICC of 0.90 for the medial tibiofemoral compartment, 0.85 for the lateral tibiofemoral compartment, and 0.93 for the patellar compartment.
Knee Bone Size Measurement
Knee tibial plateau bone area and patellar bone volume were determined by means of image processing on an independent work station using the software program Osiris (University of Geneva) as previously described (15,16,20). To transform the images to the axial plane, the Analyze Software package developed by the Mayo Clinic was employed. Medial and lateral tibial plateau bone areas were determined by creating an isotropic volume from the three input images closest to the knee joint. The bone areas of the medial and lateral tibial plateau were directly measured from the reformatted axial images. Area of patellar bone was determined individually by manually drawing contours around the target patella boundaries on a slice-by-slice basis on sagittal views. The volume of the patella bone was determined by summing all of the pertinent voxels within the resultant binary volume. Total volume was calculated for the patellar bone because of its irregular shape, which made it difficult to identify a simpler, representative measure of patellar size. The CVs for these measures in our hands are 2.2% to 2.6% (15).
Correlation analysis was used to examine the association between BMI and cartilage defects, volume, or thickness in the whole group, as well as in men and women separately. Linear regression analysis was used to examine the associations between cartilage variables and BMI or BMI categories in the total sample and in men and women separately before and after adjustment for age, sex, case control status (whether subjects were offspring or controls), bone size, and/or radiographic OA (ROA). For the cartilage defect scores in individual compartments, the range and distribution were such that basic assumptions for applying linear regression did not hold. However, for the combined total score, the assumptions did hold. For the sake of comparability, we used linear regression to assess multivariable BMI effects, adjusting for other covariates, for all score and total score variables. This seemed preferable to using an ordinal logistic regression model for a pooled score outcome and linear regression for total score. Logistic regression analysis was used to examine the associations between prevalence of knee cartilage defects and BMI or BMI categories before and after adjustment for age, sex, case control status, bone size, and/or ROA. Unpaired Student's t tests were used to examine the differences in cartilage defects, volume, or thickness among normal, overweight, and obese subjects. A p value <0.05 (two-tailed) or a 95% confidence interval not including the null point was regarded as statistically significant. A 10% change in the coefficient for age after adjustment for a variable was accepted as providing evidence of a factor such as bone size or ROA acting as an intermediate variable. All figures present correlation coefficients, odds ratios, and/or p values adjusted for age. A linear relationship between age and the various measures provided the best fit for the data. All statistical analyses were performed on SPSS version 10.0 for Windows (SPSS, Chicago, IL).
A total of 372 subjects (214 women and 158 men) between 26 and 61 years of age (mean, 45 years) took part in this study. The mean BMI was 27.1 ± 5.1 kg/m2, with 42% categorized as overweight and 21% categorized as obese. Demographic and study factors are presented in Table 1. Subjects who were normal, overweight, and obese were similar in terms of height, past knee injury history, smoking, and ROA prevalence, but obese subjects were older and had higher muscle strength and physical work capacity than normal subjects.
Obese female but not male subjects had significantly higher knee cartilage defect scores than normal subjects (Figure 1; Table 2). Knee cartilage defects were also more common in obese women but not men (Figure 2; Table 2). BMI was significantly associated with knee cartilage defect scores in all compartments except for patellar and lateral tibiofemoral compartment in men before and after adjustment for age, sex, and case control status (Figure 3; Table 3), and these associations were consistently higher in women than in men (Figure 3). BMI was also significantly associated with prevalent knee cartilage defects in all compartments (Table 3), except for patellar and lateral tibiofemoral compartments in men and the lateral tibiofemoral compartment in women. In subjects without ROA, the associations between BMI and knee cartilage defects were similar to those in the whole sample (data not shown).
Table 2. . Differences in cartilage and bone measures among normal, overweight, and obese subjects
Adjusted for previous factors, bone size at that site if cartilage variables, and cartilage defect score at that compartment if bone size.
Further adjusted for ROA.
+0.033 (+0.012, +0.054)
+0.032 (+0.011, +0.053)
+0.030 (+0.009, +0.051)
Medial tibiofemoral defects
+0.037 (+0.021, +0.053)
+0.031 (+0.015, +0.046)
+0.028 (+0.013, +0.044)
Lateral tibiofemoral defects
+0.016 (+0, +0.033)
+0.012 (−0.004, +0.028)
+0.013 (−0.003, +0.028)
+0.083 (+0.045, +0.12)
+0.066 (+0.028, +0.104)
+0.062 (+0.025, +0.100)
Patellar cartilage thickness (mm)
-0.024 (−0.044, −0.003)
−0.024 (−0.044, −0.003)
-0.021 (−0.041, −0.001)
Medial tibial thickness (mm)
0 (−0.016, +0.016)
−0.005 (−0.021, +0.010)
−0.003 (−0.019, +0.012)
Lateral tibial thickness (mm)
−0.013 (−0.033, +0.007)
−0.010 (−0.030, +0.010)
−0.012 (−0.032, +0.008)
Patellar cartilage volume (μL)
−5.4 (−22.3, +11.5)
−8.6 (−24.7, +7.5)
−6.5 (−22.4, +9.4)
Medial tibial cartilage volume (μL)
+1.7 (−9.0, +12.4)
−6.7 (−16.7, +3.3)
−5.5 (−15.4, +4.4)
Lateral tibial cartilage volume(μL)
+2.0 (−9.6, +13.7)
−1.4 (−12.1, +9.4)
−2.2 (−12.8, +8.3)
Patellar bone volume (μL)
+31.4 (−22.2, +84.9)
+28.1 (−26.2, +82.4)
+23.9 (−30.4, +78.2)
Medial tibial bone area (mm2)
+8.2 (+4.1, +12.3)
+6.6 (+2.4, +10.8)
+6.4 (+2.2, +10.6)
Lateral tibial bone area (mm2)
+3.3 (+0.1, +6.5)
+2.5 (−0.6, +5.7)
+2.5 (−0.6, +5.6)
OR (95%CI) (per kg/m2)
OR (95%CI) (per kg/m2)
OR (95%CI) (per kg/m2)
1.08 (1.03 to 1.14)
1.08 (1.03 to 1.14)
1.08 (1.02 to 1.14)
Medial tibiofemoral defects
1.12 (1.05 to 1.18)
1.09 (1.03 to 1.16)
1.09 (1.02 to 1.16)
Lateral tibiofemoral defects
1.05 (0.99 to 1.11)
1.04 (0.98 to 1.10)
1.04 (0.98 to 1.10)
1.08 (1.03 to 1.13)
1.07 (1.01 to 1.12)
1.07 (1.01 to 1.13)
In the whole sample, BMI was negatively associated with patellar cartilage thickness, an association that remained significant (p = 0.039) after adjustment for confounders (Table 3). This association became nonsignificant when men and women were analyzed separately in multivariate analysis. No significant associations were observed between BMI and tibial cartilage thickness and between BMI and knee cartilage volume in regression analysis (Table 3). No significant associations were observed between BMI categories and knee cartilage thickness or patellar cartilage volume (Table 2). Tibial cartilage volumes were positively associated with BMI categories (Table 2), but these associations did not persist after adjustment for sex, because there were more women (69%) in the normal group.
BMI was positively associated with medial and lateral tibial plateau bone area (all p < 0.01), and the associations remained significant (medial: p = 0.001; lateral: p = 0.037) after adjustment (Figure 4; Table 3). Those who were obese or overweight had larger knee bone size than those who were normal weight (Table 2). After adjustment for sex, age, and case control status, this difference persisted at the medial tibial site (p = 0.006) but not at other sites, mainly because of more women in the normal group.
The associations between BMI and cartilage defect scores at tibiofemoral and whole compartments were decreased by 16% to 25% in the total sample, 23% to 44% in men, and 12% to 22% in women after adjustment for bone size at that site. There was little change after further adjustment for ROA (Table 3). The associations between BMI and tibial bone areas were decreased by 20% to 24% in the total sample, 16% to 20% in men, and 17% to 25% in women after adjustment for cartilage defect score at that compartment. There was little change after further adjustment for ROA (Table 3). If the analysis was restricted to those without OA, significant associations persisted for the medial and lateral tibial areas. If it was restricted to those <45 years of age, significant associations persisted for the lateral area but not the medial tibial area (p = 0.16).
This cross-sectional study documents consistent associations between BMI and knee cartilage defects and tibial bone size that were present whether BMI was considered as a continuous or categorical variable. These associations were stronger in women. In contrast, inconsistent associations were present for cartilage thickness, and no associations were present for cartilage volume, suggesting that increasing BMI induces cartilage defects and tibial bone enlargement but not cartilage loss.
While previous studies have reported that increasing BMI is positively associated with knee joint space narrowing in subjects with knee OA (2,26), there is no significant association between BMI and joint space area or width in normal knees (27). BMI was not associated with incident knee joint space narrowing (28) and the loss of knee cartilage volume in OA knees (19), and knee cartilage volume in overweight children did not differ significantly from that in normal children, either cross-sectionally or longitudinally (29). However, BMI has been inversely associated with tibial cartilage volume standardized for bone size (reflecting cartilage thickness) in men (21). These inconsistencies are likely caused by low sensitivity of radiographic assessment, differences in disease groups, generally small sample sizes, and differing age groups. In this sample, we found no association between BMI and cartilage volume in subjects who were largely free of ROA. In addition, only patellar cartilage thickness decreased significantly with increasing BMI in the whole sample, with no significant association in men or women. This suggests that cartilage volume may not mediate the effect of BMI on OA.
To our knowledge, the association between BMI and knee cartilage defects has not been previously reported. In this study, we found that knee cartilage defects had a higher prevalence and that BMI and obesity were consistently associated with knee cartilage defect severity and prevalence in all compartments. These associations were more marked in women and in patellar and medial tibiofemoral compartments. These results provide direct evidence that increasing BMI may induce cartilage defects, even in those who are free of ROA, and may explain why women are at higher risk of obesity-related OA. They may also be of clinical relevance at an individual level because of the large differences observed in this study. The reasons for the BMI-related alterations in knee cartilage defects remain obscure. They are most likely caused by a failure to adapt to increased loading as the overall force across the knee in a single-leg stance increases 2 to 3 lb for each 1-lb increase in body weight (1). In addition, they may be linked to expansion of joint surface area, because the two processes were linked (see below). The differences in association between BMI and cartilage volume, thickness, and defects are of note. From the perspective of a standing person, tibial thinning is largely vertical in nature, whereas defects are horizontal, thus reflecting differing pathologies. Indeed, thickness and defects were only weakly and inconsistently associated with each other in this sample, whereas volume was negatively associated with defect scores as would be expected (data not shown).
There is recent evidence that knee bone size may be an important factor in the development of knee OA. It is under strong genetic control (30) and is higher in the offspring of those with severe OA (23). It also is markedly increased in those with early radiographic osteophytosis (20) and increases with age (C. Ding, unpublished data). Some studies have suggested that bone size alters in response to increasing BMI. Using DXA, a Swedish group (31,32) reported, in both boys and girls, that bone area in the tibia diaphysis and femoral diaphysis correlated positively with BMI. So far, there are few studies that describe an alteration in articular bone size with increasing BMI. A study reported that BMI was not correlated significantly with tibial plateau width measured on radiographs (26). In contrast to this finding, we found that medial and lateral tibial bone area increased significantly with increasing BMI both as a continuous and categorical variable. The clinical relevance at an individual level is uncertain because of the large spread of results and relatively weak associations. However, our results suggest that a greater BMI can induce larger knee subchondral bone size, especially medial tibial bone area, or that knee subchondral bone may respond to higher loads by expansion of the joint surface area. This is consistent with a recent report that the knee adduction moment, which provides a major contribution to the 70% of total knee joint load passing through the medial tibiofemoral compartment during walking, correlates significantly with medial tibial bone area but not cartilage volume (33). Knee bone size mediates, in part, the associations between BMI and knee cartilage defects, because the associations between BMI and knee cartilage defect scores decreased in magnitude after adjustment for bone size and vice versa, suggesting that there may be a shared contribution to BMI-associated alteration in knee cartilage defects and tibial bone area. Like the cartilage defect associations, the associations between BMI and tibial bone areas were stronger in women. However, these associations changed little after further adjustment for ROA. This suggests that they are independent of ROA or may reflect the inaccuracy of radiographic assessment in early OA. Further studies will need larger number of subjects with a higher prevalence of and more severe OA. Alternatively, given the cross-sectional nature of this study, it is possible that overall body habitus is associated with bone size and that endomorphs may have larger bones. It is also possible that shared genetic factors may lead to both obesity and larger bones, although this seems less likely because a recent study concluded that the genetic effects for obesity and OA were largely independent (2).
This study has a number of limitations. First, the study was designed to look primarily at genetic mechanisms of knee OA and used a matched design. The matching was broken for this study, but adjustment for case control status did not alter the results. Indeed, whereas there was a reduction in power, the results otherwise did not differ if examined in offspring and controls separately. While the sample is a convenience sample, Miettinen (34) states that, for these associations to be generalizable to other populations, three key criteria need to be met, explicitly defined selection criteria, adequate sample size, and adequate distribution of study factors, all of which were met by this study. Second, measurement error may influence results. However, scoring of knee cartilage defects, volume, thickness, and bone size measurement was highly reproducible, suggesting that this is unlikely. Third, ROA assessment allows us to examine whether the changes are of relevance to ROA. Given the relative rarity and mild severity of ROA in our sample, these results may not be generalizable to those with more severe OA. Last, the study was cross-sectional in design and cannot comment on causal directions; thus, longitudinal data will be required to confirm these results.
In conclusion, this cross-sectional study suggests that knee cartilage defects and tibial bone enlargement are the main structural changes associated with increasing BMI. Preventing these changes may prevent knee OA in overweight and obese subjects.
This study was supported by the National Health and Medical Research Council of Australia, Masonic Centenary Medical Research Foundation. We thank the subjects and orthopedic surgeons who made this study possible. The role of C. Boon in coordinating the study is gratefully acknowledged. We also thank Martin Rush, who performed the MRI scans, and Kevin Morris for technical support.
Nonstandard abbreviations: OA, osteoarthritis; MRI, magnetic resonance imaging; ICC, intraclass correlation coefficient; CV, coefficient of variation; ROA, radiographic OA.