Objective: Given that obesity is a risk factor for osteoarthritis (OA) of the knee, a study was undertaken to determine whether progressively higher body mass index (BMI) among obese women is associated with progressive increases in joint space narrowing (JSN).
Methods: Medial compartment JSN over 12 months in Lyon Schuss radiographs of 60 obese women (BMI 30.0–50.5 kg/m2) with radiographic and symptomatic OA was compared with that in 81 non-obese women (BMI <28 kg/m2) with normal radiographs and minimal or no symptoms of knee OA.
Results: Among the patients with OA, higher BMI tended to be associated with a higher Kellgren and Lawrence (KL) grade of OA severity. JSN in the non-obese controls was negligible, but in the 30 patients with KL grade 2 and KL grade 3 knees, mean (SD) JSN was 0.12 (0.31) mm and 0.32 (0.50) mm, respectively (p<0.005 and p<0.001). No association was seen between baseline BMI and 12-month JSN in patients with OA; indeed, the regression plot suggested a slight inverse relationship between the two.
Conclusions: In obese patients with OA, progressively higher BMI values were not accompanied by a progressively increasing rate of JSN. Joint loading was not evaluated, but it is possible that marked obesity limited the functional capacity of some subjects with OA, protecting their knees from loading. For investigators considering eligibility criteria for a trial of a structure-modifying OA drug, these data suggest that recruitment of patients with a BMI much higher than 30 kg/m2 will not enrich the sample of subjects who will have more rapid JSN than those with a BMI of only 30 kg/m2.
Statistics from Altmetric.com
Because the progression of structural joint damage in patients with osteoarthritis (OA) of the knee is relatively slow, those engaged in the design of placebo-controlled randomised clinical trials of structure-modifying drugs for this disease often seek to enrich the study cohort with subjects who have risk factors for progression of joint damage. In this way, they hope to maximise the chance of detecting a significant difference between the study drug and placebo and to reduce the length of the study, number of subjects required and cost (albeit at the risk of limiting the generalisability of the results).
The association between obesity and knee OA is well established.1 2 3 Consistent with previous reports, in a study that used standing anteroposterior (AP) radiographs of the knee, Reijman et al4 found that the odds ratio for progression of established radiographic knee OA over a mean follow-up period of 6.6 years among subjects with a body mass index (BMI) of >27.5 kg/m2 was 3.2 relative to subjects with knee OA who had a lower BMI. Within the subset of obese subjects, however, the authors did not attempt to assess the risk of OA progression in relation to increasing BMI. Dennison and Cooper5 cited several longitudinal studies which reported that obese patients with established radiographic OA were more likely to exhibit progression of their disease than non-obese patients.
Not all studies, however, have found obesity to be a risk factor for progression of established radiographic knee OA.6 7 Furthermore, conclusions about obesity as a risk factor for progression of knee OA have been based almost exclusively on analyses of conventional standing AP radiographs of the knee in extension, although the sensitivity to loss of joint space width (JSW, a commonly accepted surrogate for loss of articular cartilage in the tibiofemoral compartment) is notably poorer and variability in the rate of joint space narrowing (JSN) considerably greater in that view than in images obtained with protocols that more effectively standardise the radioanatomical position of the knee in flexion such as the Lyon Schuss (LS) view.8 9 10
The present study compared JSN over 12 months in knees of 60 obese patients with radiographic and symptomatic OA with that in 81 non-arthritic and non-obese control subjects. We examined whether the rate of JSN in the obese patients was affected by OA severity at baseline and whether a threshold existed for the effect of obesity on JSN or, alternatively, whether JSN increased progressively with increasing BMI.
The 141 study subjects were women aged at least 40 years without significant medical complaints enrolled in a multicentre 24-month observational study to evaluate the progression of OA. All of the patients with OA in the observational study were selected on the basis of the presence of risk factors that have been associated with progression of the disease: obesity, female sex, symptomatic and radiographic disease. Thus, all 60 patients with OA in the present study had definite radiographic OA at baseline (Kellgren and Lawrence (KL) grade 2–3 in a conventional standing AP view, with KL grade 0–3 in the contralateral knee); BMI values ranging from 30.0 to 50.5 kg/m2; and knee pain, aching or stiffness on most days of a month during the past year and/or had used medication for treatment of knee pain on most days of a month during that time. The remaining 81 subjects in the study were non-arthritic age- and sex-matched control subjects enrolled in the observational study. Each had a BMI <28 kg/m2; no evidence of radiographic OA in either knee (ie, KL grade 0 bilaterally); no or only infrequent knee pain, aching or stiffness during the past year; and no or only infrequent use of medication for treatment of knee pain during the past year.
Detailed characterisation of the subjects, extant risk factors for rapid progression of OA and the inclusion and exclusion criteria employed in the observational study have recently been reported elsewhere.11
The study knee in each patient with OA was defined as the more symptomatic knee. If pain scores were identical in both knees, the knee with more advanced radiographic changes of OA was designated the study knee. If pain scores and radiographic severity in both knees were identical, the knee in the subject’s dominant leg was chosen as the study knee. For the controls, the knee in the dominant leg was considered to be the study knee.
Scores on the function (C) subscale of the Western Ontario and McMaster’s Universities Osteoarthritis Index (WOMAC) at baseline and 12 months were used to assess changes in lower extremity functional impairment in relation to BMI. The possible range of scores on the 17-question subscale is 17–85.12
All radiographs were acquired without shoes. The screening film on which a determination of eligibility for the study was made on the basis of the KL grade of OA severity (see above) was a standing AP view of both knees in an extended position, imaged on a single cassette. Initial KL grading was performed by the clinical centre director. An experienced central reader (SAM), who was blinded to the grade assigned in the clinical centre, re-read each radiograph for standardisation of the KL classification. If the grade assigned in the clinical centre differed from that of the central reader, the difference was adjudicated by a third reader (KDB).
Intrareader reproducibility, determined using 30 radiographs exhibiting KL grades 0–3, showed an intraclass correlation coefficient of 0.91 and a kappa of 0.66. After ascertainment that the subject qualified for the study on the basis of the KL grade of the AP radiograph, an LS view of the study knee was obtained to assure that JSW of the medial tibiofemoral compartment was >2 mm. If this was not the case, the patient was deemed ineligible for the study. If the patient qualified for the study she was scheduled to return in 12 months for a follow-up evaluation and LS radiograph.
The posteroanterior (PA) view for the LS images was obtained using a SynaFlex frame (Synarc, San Francisco, California, USA) to position the feet reproducibly.13 Body weight was distributed equally between the two legs and the knees and thighs were pressed directly against the wall of the frame, the anterior wall of which was in contact with the Bucky or reclining tabletop of the radiographic unit. The patella was coplanar with the anterior aspect of the thigh and the tip of the great toe, fixing the angle of knee flexion at about 20°. A V-shaped angulation support on the base of the frame was used to fix the ipsilateral foot in 10° external rotation. The x ray beam was centred on the joint line of the knee and the beam angle was adjusted fluoroscopically so as to align (±1.5 mm) the anterior and posterior margins of the medial tibial plateau for each examination. Insofar as the LS view as described previously did not use a foot frame to help assure reproducible positioning, technically this represented a modified LS view.
Automated measurement of minimum JSW (mJSW)
mJSW of the medial tibiofemoral compartment was measured by an experienced observer using digitised image analysis software (Holy’s software-beta19, UCLB, Lyon, France) that permitted the contours of the joint space to be detected automatically with the help of an edge-based algorithm. Measurements were made automatically in a medial area of the compartment, the external limit of which was determined by the observer who took care to exclude marginal osteophytes. The computer then measured the minimum interbone distance in each medial area. The reproducibility of measurements of mJSW was determined by blinded remeasurement of 37 randomly selected LS radiographs by the same observer, who was unaware of the previous measurement. The SD of the mean difference between the two measurements was 0.06 mm and the coefficient of variation was 1.2%.
For scale variables, mean and SD are given. Groups were compared using the Student t test or one-way ANOVA. Slope, slope p value and coefficient of correlation between medial compartment JSN and baseline BMI, as well as correlation between WOMAC C score and baseline BMI and change in BMI, were calculated. 95% confidence intervals (95% CI) of the mean difference between groups and of correlation coefficients are reported.
BMI of patients with OA and controls
The mean (SD) BMI of the 60 obese patients with symptomatic and definite radiographic OA (KL grade 2 or 3 in the standing AP radiograph) was 37.2 (5.1) kg/m2, with a range of 30.0–50.5 kg/m2 (fig 1). In contrast, the mean (SD) BMI for the 81 control subjects was 23.5 (2.3) kg/m2, with a range of 17.8–27.9 kg/m2 (p<0.001 for the difference between patients and controls).
Table 1 shows the mean (SD) BMI values by KL grade as well as age. Among these obese patients with OA, a higher BMI tended to be associated with a higher KL grade. The mean difference between the mean (SD) baseline BMI of the 30 patients with KL grade 3 changes (38.4 (5.4) kg/m2) and that of the 30 with KL grade 2 radiographic changes (36.0 (4.5) kg/m2) was 2.4 kg/m2 (95% CI −5.0 to 0.12) and approached significance (p = 0.061).
JSN in relation to KL grade
Mean (SD) 12-month JSN in the study knee of the control subjects was negligible (0.02 (0.25) mm; p = 0.511). In contrast, in the 30 KL grade 2 and 30 KL grade 3 knees, mean (SD) JSN was 0.12 (0.31) mm and 0.32 (0.50) mm, respectively (p<0.005 and p<0.001 relative to that in the non-arthritic control subjects). The mean difference in JSN between OA knees with grade 3 changes and those with grade 2 changes was 0.20 mm (95% CI −0.42 to 0.01) and approached significance (p = 0.066). The age of the patients is shown in table 1. The mean difference in age between those with KL grade 2 and KL grade 3 was 2.6 years (95% CI −6.7 to 1.4); this difference was not significant (p = 0.191). The rate of JSN in the patients with OA was unrelated to age (r = 0.04, p = 0.747).
Lack of relationship between BMI and JSN
A regression plot showing the lack of association between baseline BMI and 12-month JSN in the patients with OA, all of whom had a baseline BMI ⩾30 kg/m2 (correlation coefficient (r) = −0.148 (95% CI −0.388 to 0.109, p = 0.256), is shown in fig 2A. Indeed, in these obese patients, the slope suggests a slight inverse relationship between BMI and JSN (ie, the higher the BMI, the more modest the rate of JSN). As noted above, no JSN was noted in the 81 non-obese controls over the 12-month interval. JSN was unrelated to baseline BMI (r = −0.005 (95% CI −0.224 to 0.213, p = 0.960; fig 2B).
Lack of relationship between BMI and WOMAC function
Because of the absence of evidence that higher grades of obesity were associated with progressive increases in JSN over the 12-month observation period, we considered the possibility that higher grades of obesity limited the functional ability of some of the patients with OA. Although we did not obtain a direct measure of lower extremity function during the study, we compared the baseline and 12-month WOMAC12 function scores for each subject.
Baseline scores of subjects with KL grade 2 and KL grade 3 on the WOMAC C subscale (functional impairment) were essentially identical (mean (SD) 22.4 (14.0) and 22.5 (14.5), respectively, possible range 17–85). WOMAC C scores at baseline were unrelated to the baseline BMI (r = 0.129, 95% CI −0.129 to 0.371, p = 0.325) and the 12-month change in WOMAC C score was unrelated to the 12-month change in BMI (r = 0.128, 95% CI −0.374 to 0.135, p = 0.340; fig 3). One patient shown in fig 3 had a very large leverage on the small positive slope and, if left out of the analysis, the slope decreased to 0.25 (p = 0.714) and the correlation fell to r = 0.05.
The principal finding in the present study is that, in obese patients with OA, we found no evidence that progressively higher BMI values were accompanied by a progressively increasing rate of JSN (fig 2). It should be noted that an interval as short as 12 months was sufficient to permit detection of significant JSN in KL grade 2–3 OA knees relative to knees of non-arthritic controls (p<0.001). Furthermore, the magnitude of JSN over 12 months in the OA knees in the present study (0.12 (0.31) mm for KL grade 2 knees and 0.32 (0.50) mm for KL grade 3 knees) was similar to the annualised rate of JSN in the placebo group in the randomised clinical trial of doxycycline (approximately 0.18 mm) in which subjects were, as in the present study, middle-aged obese women (mean (SD) BMI 36.5 (6.0) kg/m2 with KL grade 2–3 OA.14
The present study used a convenience sample of subjects who were enrolled in the ongoing observational study in which a BMI of >30 kg/m2 was used as an inclusion criterion, based on evidence that obesity is a risk factor for the progression of knee OA. It was anticipated that this criterion and a phenotype that included women with painful radiographically-confirmed knee OA would facilitate the identification of subjects with OA with rapid progression of JSN, a characteristic that might be particularly desirable in a clinical trial of a potential disease-modifying OA drug (DMOAD).
Although obesity is widely recognised as a risk factor for knee OA, we took advantage of the availability of the existence of this cohort of obese subjects with definite radiographic OA to ask a specific question that is highly relevant to investigators who design clinical trials of DMOADs—namely, among obese subjects who already have OA, does increasing obesity increase the progression of OA, as reflected by the rate of JSN?
We considered the possibility that their marked obesity limited the functional ability of some patients with high-grade obesity and that, as a consequence, they subjected their knees to less mechanical loading than non-obese individuals, accounting for the lack of a relationship between more rapid JSN and higher BMI levels. Although we did not use a direct measure of lower extremity function such as pedometers or accelerometers, the patients’ scores on the WOMAC C subscale (functional impairment) indicated, on average, only mild impairment of lower extremity function at baseline (mean approximately 22, possible range 17–85). As noted, baseline WOMAC C scores were unrelated to the BMI. Furthermore, there was no evidence that a change in BMI had any association with changes in the WOMAC score (fig 3). Thus, our failure to observe a progressive increase in JSN with progressively increasing BMI in these obese subjects could not be explained by a reduction in performance of activities that would have loaded their OA knee over the period of observation. However, a report by the patient of impairment in function (or lack of impairment) does not necessarily provide a true measure of what the patient actually does from day to day. It would be useful to obtain such information in a future study.
High-grade obesity is a serious disease. Individuals with a BMI of >39 kg/m2 are labelled as being “morbidly” obese. It is notable, therefore, that all the obese subjects with OA in this study considered themselves—and were considered by their physicians—to be “healthy” at the time of enrolment, except for problems associated with their knee OA. This possibly represents a source of bias in the study insofar as results with respect to JSN may differ in morbidly obese subjects in whom the metabolic effects of obesity and its consequences (eg, diabetes mellitus, hypertension, cardiovascular disease) are more apparent. However, such subjects are less likely to be enrolled in trials of DMOADs.
The chief limitations of this study are its relatively small sample size, limitation of the follow-up period to 12 months and lack of quantitation of loading of the knee over the period of observation. Continued follow-up of these subjects will permit us to ascertain whether the observations reported here persist. If they do, this will have obvious implications with respect to the inclusion criteria for studies of potential structure-modifying drugs for OA.
Because none of the patients with OA in this analysis had a BMI of <30 kg/m2, this study cannot address the question whether JSN in an OA knee is more rapid in obese subjects than in those with a lower BMI. As indicated above, several reports have found this to be the case However, in addition to showing that the LS view can detect significant JSN within only 12 months in middle-aged obese women with knee OA, for investigators who are contemplating eligibility criteria for a DMOAD trial the present data suggest that recruitment of patients with a BMI much greater than 30 kg/m2 will not enrich the sample in subjects who will have more rapid JSN than patients with a baseline BMI of only 30 kg/m2.
Funding Pfizer Global Research & Development
Competing interests M-PHLG is employed by Pfizer Inc. EV receives grant support from Pfizer. KDB provides consulting services to Pfizer. SAM receives grant support from and provides consulting services to Pfizer.
Ethics approval The study was conducted in compliance with the ethical principles derived from the Declaration of Helsinki and in compliance with local Institutional Review Board, informed consent regulations and International Conference on Harmonization Good Clinical Practices Guidelines.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.