Knee osteoarthritis (OA) affects ∼6% of adults ages 30 years and older (1). There are no treatments available that would impede the course of this disease, which is generally characterized as progressive cartilage loss. Persons with knee OA are, on average, heavier than those without disease, and longitudinal studies have shown that being overweight increases the risk of developing disease (2–4). Since disease incidence is influenced by a person's weight, it is possible that being heavy also has an effect on the risk of disease progression in those who already have OA in which case weight loss might be an effective treatment to prevent structural deterioration.
However, the effect of body weight on disease progression has been less consistently demonstrated than its effect on disease incidence. Some longitudinal studies of patients with knee OA have shown that, compared with nonobese subjects, those who are obese have a higher risk of joint space loss, suggestive of cartilage loss, as visualized on radiographs (5–7), but these findings are not universally reported. Indeed, some prospective studies of knee OA have shown no association between being overweight and disease progression (8, 9). Indeed, in one study (10) among women with unilateral knee OA, the risk of developing new disease in the contralateral knee was affected by a woman's weight, but the risk of experiencing progression of disease in the index knee was not affected by weight.
There are several problems in the design of studies that have examined the relationship of body weight to disease progression, raising questions about the validity of their conclusions. First, prior studies have used radiographic views of knee limbs in a fully extended, standing position (or even non–weight-bearing radiographs) to evaluate disease progression (11, 12). In such studies, joint space loss over time would have been difficult to characterize reliably, given the tremendous variability of joint space width from examination to examination. In addition, the larger studies (5) were characterized by high rates of loss to followup, raising questions about whether those subjects with missing data had the same disease progression as those who remained in the study; indeed, if nonobese persons with knee OA were more likely to have knee replacements than obese persons with knee OA and were dropped from the analysis at followup (as was often the case), then obese persons would appear to have an increased risk of progression even if this was not the case.
Identifying a target population of subjects at high risk of knee OA progression for recruitment into clinical trials testing weight loss and for management in clinical practice is an important objective. The conflicting results of studies evaluating body weight and its relationship to progressive knee OA could be due to the inclusion of different types of patients in different studies. One important feature distinguishing patient groups may be their limb alignment. Limb malalignment (13–15) is a powerful predictor of disease progression that is defined as joint space loss. Since body weight is likely to affect knee OA through transarticular loading, it stands to reason that the alignment status of individuals may influence how weight affects the trajectory of knee OA.
Using data from a 30-month longitudinal study in which serial radiographs of fluoroscopically positioned knees were obtained and in which loss to followup was minimal, we examined the effect of weight on the trajectory of joint space loss in knee OA and tested whether this effect differed by varying levels of malalignment. Since moderate malalignment increases local stresses in cartilage, and since obesity would have a multiplier effect on the increase in stress, we postulated that the effect of weight on a subject's knee would be significant if the limb were moderately malaligned. We further speculated that there might be a category of limbs with malalignment sufficiently severe that progression would be likely even without the subject being overweight, and therefore, in this group, a person's weight would not contribute greatly to the risk of progression.
PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
Subjects were recruited to participate in a natural history study of symptomatic knee OA. All subjects in this study are a subset of subjects whose recruitment has been described in detail elsewhere (15). Briefly, subjects were recruited from 2 prospective studies of the quality of life of veterans (1 involving men and 1 involving women), from clinics in the Veterans Administration Boston Healthcare System and from advertisements in local newspapers. Potential participants were asked 2 questions: “Do you have pain, aching or stiffness in one or both knees on most days?” and “Has a doctor ever told you that you have knee arthritis?” For subjects who answered positively to both questions, we conducted a followup interview in which we asked about other types of arthritis that could cause knee symptoms. If no other forms of arthritis were identified, then the individual was eligible for recruitment.
To determine whether subjects had radiographic OA, they underwent a series of knee radiographic assessments. If the subject had a definite osteophyte on any view in the symptomatic knee, they were eligible for the study. Based on the presence of frequent knee symptoms and radiographic OA, all subjects met the American College of Rheumatology criteria for symptomatic knee OA (16). For the natural history study, we enrolled subjects who were interested in participating and who could walk with or without a cane.
The study included a baseline examination and followup examinations at 15 and 30 months. At each examination, a knee radiograph was obtained. In addition, subjects were weighed with their shoes off, using a balance beam scale, and height was assessed. At the first followup visit, long-limb films were obtained with a 14 × 51 cassette, using methods previously described (15). The baseline and followup examinations were approved by the Boston University Medical Center and the Veterans Administration Boston Healthcare System Institutional Review Boards.
Subjects underwent weight-bearing, posteroanterior (PA) radiography, using the protocol of Buckland-Wright et al (17). With the use of fluoroscopic positioning, we aligned the beam relative to the center of the knee. The knee was flexed so that the anterior and posterior lips of the medial tibial plateau were superimposed. The feet were rotated until the tibial spines were centered in the notch, and outlines of foot rotation were then made on foot maps so that, for subsequent films, the foot rotation was the same. Fluoroscopic positioning has been shown to produce more accurate assessments of the joint space, relative to nonfluoroscopic acquisition, and to improve reproducibility of joint space assessment (11, 12).
For evaluation of knee OA progression, we focused on the width of the joint space in the medial and lateral compartments, since these have been found to correlate with cartilage thickness (11). Films were read using the Osteoarthritis Research Society International Atlas (18), in which each view of the medial and lateral tibiofemoral joint space is graded on a semiquantitative scale from 0 (normal) to 3 (bone on bone). We defined progression of joint space narrowing in a knee compartment as an increase of ≥1 grade. One reader (DTF) read all films. Whereas all knee radiographs were read with the examiner unblinded to sequence, a subsample of knee radiographs was read with the examiner blinded to sequence in order to 1) test the reproducibility of progression measurement, and 2) evaluate possible bias in characterizing progression based on the fact that the sequence of the films was not masked. Intraobserver agreement (kappa) for reading progression in a blinded versus unblinded manner was 0.81 (P < 0.001), and disagreements between blinded and unblinded readings were in no particular direction (there was no greater tendency for unblinded readings to be read as showing progression). Knees were also read for Kellgren/Lawrence severity grade (19).
Determination of alignment.
Mechanical alignment, assessed at the first followup examination, was measured on a continuous scale, with results expressed in degrees. Interobserver agreement for reading alignment was high (intraclass correlation coefficient 0.97, P < 0.001). Since our focus was on neutrally aligned limbs compared with malaligned limbs, we combined data on varus/valgus alignment and characterized a limb based on the absolute value of alignment, creating 3 ordinal categories based roughly on tertiles of the distribution of alignment in our sample: 0–2°, 3–6°, and ≥7°.
Body mass index (BMI).
BMI was computed as the ratio of weight (in kg) divided by height (in square meters). Each person in our study had to be examined at both the baseline and first followup examinations, when alignment was assessed. To optimize accurate assessment of average BMI over the study, we used the average BMI at baseline and first followup as our measure of BMI for the purposes of analysis.
Subjects in our natural history study were eligible for this study if they 1) underwent assessment of weight and height, 2) had a fluoroscopically positioned PA radiograph at baseline, 3) underwent assessment of alignment at the first followup examination at 15 months, and 4) did not have grade 3 joint space narrowing or a total knee replacement in the knee at baseline. We compared the average BMI by alignment groups, using analysis of variance (ANOVA). We then assessed the association between BMI and the risk of progression of knee OA after adjusting for age, sex, and limb alignment, using logistic regression analyses and generalized estimating equations to adjust for the correlation between knees. We used the same approach to examine our results by sex and by different categories of alignment.
We then evaluated whether the relationship of BMI to the risk of knee OA progression varied across categories of alignment, using the odds ratio (OR). Specifically, for both sexes combined and then for each sex separately, we assessed the relative effect (the OR) of BMI on the risk of knee OA progression within each alignment category, adjusting for age. All P values reported are 2-tailed.
- Top of page
- PATIENTS AND METHODS
Of the 228 persons who had both a baseline radiographic assessment and an alignment measurement at the first followup and who did not have bilateral end-stage (grade 3) joint space narrowing at baseline, 227 had followup radiographs evaluated for progression. Four hundred three knees from the 228 subjects were considered eligible for progression at baseline. Of those, 9 knees were excluded (of which 2 were from the single person lost to followup and the other 7 were due to total knee replacements), leaving 227 persons with 394 knees considered eligible for progression. A description of the 227 subjects and the clinical characteristics of the knees is provided in Table 1. Given the Veterans Administration origin of the subjects, most were men. The mean age was 66 years. More than 20% of knees showed radiographic progression at some point over the followup period. Of 90 knees showing progression, 78 had Kellgren/Lawrence grade 2 or grade 3 OA severity at baseline, and most had joint space narrowing scores of 1 or 2 at baseline.
Table 1. Description of 227 subjects followed up for progression of knee osteoarthritis
|Age, mean ± SD years||66.4 ± 9.4|
|Mean ± SD BMI, kg/m2||30.6 ± 4.7|
|Kellgren/Lawrence grade|| |
| Median (interquartile range)||2 (1–3)|
| Distribution, no. knees in grade 0/1/2/3/4||54/100/103/128/9|
|No. (%) knees showing progression||90/394 (22.8)|
|Median alignment, degrees||2.1 (varus)|
|No. varus/neutral/valgus limbs*||263/28/103|
Among all subjects followed up, the BMI significantly affected the risk of radiographic progression. For each 2-unit increase in BMI, there was an 8% increase in the risk of progression (OR 1.08, 95% CI 1.00–1.16, P = 0.03). BMI was modestly correlated with alignment (Table 2), with higher levels of BMI associated with greater malalignment (P < 0.001 by ANOVA). Furthermore, alignment was strongly related to radiographic progression (Table 2), with the proportion of knees showing progression increasing from 9.2% of neutrally aligned limbs (0–2°) to 22.3% of limbs with 3–6° of malalignment to 48.7% of limbs with severe malalignment (≥7°). The association of alignment with progression was similar between men and women (data not shown).
Table 2. Body mass index (BMI) and risk of radiographic progression of osteoarthritis of the knee in relation to limb alignment
|BMI, mean ± SD kg/m2||29.3 ± 4.7||30.9 ± 4.7||31.7 ± 4.6|
|Knees with progression|| || || |
The effect of BMI on progression was different at different levels of alignment, with the risk being much greater for limbs with moderate malalignment (3–6°) (Figure 1). For example, in analyses combining both sexes, the risk of progression per 2-unit increase in BMI in neutrally aligned limbs was 1.00; this increased to an OR of 1.23 in the moderate malalignment group, but decreased to an OR of 0.93 in the severe malalignment group. For an increase of 5 BMI units in limbs with moderate malalignment, this translates into an OR of 1.69.
Figure 1. The effect of a 2-unit increase in body mass index on the risk of osteoarthritis progression by stratum of knee-limb alignment. OR = odds ratio; 95% CI = 95% confidence interval.
Download figure to PowerPoint
Similar results were obtained when we limited the analysis to knees that displayed progression after examination 2, at which time we assessed alignment (40 knees with progression). With regard to the risk of disease progression per 2-unit increase in BMI, in those subjects with neutral alignment the OR for progression was 0.98, for those with moderate malalignment the OR was 1.27, and for those with severe malalignment the OR was 0.93. Additional analyses that used slightly different cutoff levels for alignment, used BMI from examination 1 only, or adjusted for baseline Kellgren/Lawrence grade yielded similar findings.
Among the men, higher BMI did not increase the risk of radiographic progression in those with neutral limb alignment (OR per 2-unit increase in BMI 0.96, 95% CI 0.72–1.26) or in those with severe malalignment (OR 0.92, 95% CI 0.71–1.18). Only in men with moderate malalignment was there a modest effect of BMI on progression (OR 1.14, 95% CI 0.92–1.40). Similarly, among the women, higher BMI had no effect on the risk of knee OA progression in limbs that were neutrally aligned (OR 1.00, 95% CI 0.83–1.20) or severely malaligned (OR 0.96, 95% CI 0.68–1.36). In knees from limbs that were moderately malaligned, however, BMI had an effect on progression (OR 1.39, 95% CI 1.07–1.80).
- Top of page
- PATIENTS AND METHODS
In our examination of the effect of weight on progression of knee OA, we found that the risk of progression increased significantly with an increase in weight. This confirms the findings of some, but not all, previous natural history studies of knee OA. However, this effect of weight on progression was not present in all knees, but was limited to knees from limbs that were moderately malaligned. In neutrally aligned limbs on one end of the spectrum, and severely malaligned limbs on the other, body weight had no effect on risk of knee OA progression. These findings were apparent in both sexes, although, as has been suggested elsewhere (1), the effect was stronger in women than in men.
Why would increased body weight have no measurable effect on progression in knees from limbs with severe malalignment? Malalignment is the most potent risk factor for structural progression yet identified (13–15). In severe malalignment, local cartilage stresses during weight bearing may exceed thresholds for loss of cartilage matrix, and no other risk factor may be required for progression to occur. In studies (20) of large-amplitude compression of healthy cartilage explants, cyclic compressions producing 50% thinning of cartilage caused release of preformed proteoglycan and collagen from cartilage and led to cartilage structural damage and later cartilage swelling. In tissues that are damaged already, Sah and colleagues (20) speculated that “compressive strain and fluid flow would be particularly high” and that such strain and high fluid flow has great relevance to progression of OA. Since the knees in the present study were all affected by OA at baseline, partially damaged cartilage was almost certainly the substrate on which the local stresses of severe malalignment acted.
If increased body weight does not contribute to progression in severe malalignment, why did all knees in this group not show progression? We suggest that radiographs are an insensitive tool with which to evaluate progression and that magnetic resonance imaging will reveal more changes. Other explanations are possible, including resistance of some knees to progression despite excess loading, and lack of a sufficiently long followup. Since this is a small group with a small number of progressors, the result in this particular subset might not be the same in a larger or different group.
Next, why would knees in moderately malaligned limbs be most susceptible to the loading effects of high body weight? Based on our data, we suggest that loading in these limbs is not inevitably damaging to matrix and that additional loading factors accentuate the cartilage injury. Our findings suggest that the group with moderate malalignment may be the most sensitive to interventions that unload the joint.
Last, why would increased body weight have no measurable effect on progression in neutrally aligned limbs? In neutral alignment, the increased loading that accompanies increased body weight would be distributed across much of the joint surface, so that although local stress would diffusely increase, it would not exceed a threshold that causes injury. Even so, it seems likely that other, larger studies may show small effects of increased body weight on progression in neutrally aligned limbs.
There are several limitations to our study. First, alignment was assessed at the midpoint of followup and malalignment could, for some, have been a consequence of progression. We chose to look at alignment at this time point because it constitutes the best measure of average alignment during the followup, short of measuring alignment at all time points. We were interested in average malalignment, not necessarily the alignment at the beginning or the end of followup. Additional analyses focusing on progression only after the measurement of alignment showed similar results, although with fewer numbers.
Given the high BMI values among participants in our study, it is conceivable that our results regarding the effects of body weight on progression would not be generalizable to samples with a lower prevalence of obesity. We speculate that, if body weight were likely to have an effect in neutrally aligned and severely malaligned limbs, we would be most likely to detect this effect in a sample such as ours, in which there were persons of normal weight as well as severely obese subjects.
Although we used the Kellgren/Lawrence scale to measure global disease severity, the scale was not developed with fluoroscopically positioned films and has not been validated for films acquired in this manner. Moreover, subjects in our study reported a physician diagnosis of arthritis, and those with symptomatic OA who have not been diagnosed may have a different disease course. Finally, our sample was unusual among studies of patients with OA since the majority of the subjects were men. Even so, our results were similar between men and women.
Our results, in theory, could have been explained by disease severity and not malalignment or body weight: if neutrally aligned knees were nonosteoarthritic, then they would not progress anyway, and if severely malaligned knees had end-stage disease, then high body weight would not affect them. We found a range of disease severity in all malalignment strata, and progression occurred in all of these strata. Increased body weight, however, increased the risk of progression only in moderately malaligned limbs. Adjusting for Kellgren/Lawrence grade in our analyses did not affect the results.
In conclusion, although high body weight increases the risk of structural progression in knee OA, its effect on progression appears to be limited to knees in limbs that are moderately malaligned. Our findings, which need to be confirmed in other studies, suggest that prevention and treatment efforts for obesity and knee OA could be efficiently targeted to those subjects with moderate malalignment. These findings may have broad implications not just for the effect of body weight on OA, but for other risk factors that affect joint loading.