Abstract
Objective. The aim of this review was to systematically examine the evidence for an association between measures of obesity [weight and body mass index (BMI)] and body composition (fat mass and fat-free mass) and patellar cartilage, assessed using magnetic resonance imaging.
Methods. Three electronic databases (MEDLINE, EMBASE, and CINAHL) were searched up to April 2016 using full text and MeSH terms to identify studies examining the associations between obesity and body composition, and patellar cartilage. Two independent reviewers extracted the data and assessed the methodological quality of included studies.
Results. Seventeen studies were included: 5 cross-sectional, 10 cohort studies measuring outcomes at 2 timepoints, and 2 longitudinal studies assessing outcome only at the timepoint. Eleven studies were of high or moderate quality. In asymptomatic middle-aged adults, elevated body weight and BMI were systematically associated with worse patellofemoral cartilage scores. There was more consistent evidence for patellar cartilage defects than patellar cartilage volume, particularly in women. Increased BMI was also consistently associated with increased cartilage loss in longitudinal studies, although not all attained statistical significance.
Conclusion. There is a need for more high-quality research to confirm these findings and to better explain the relative contributions of metabolic and biomechanical factors to the initiation of patellofemoral osteoarthritis, to devise effective strategies to manage this common and disabling condition.
Patellofemoral osteoarthritis (OA) is associated with greater disability and contributes more to knee pain than tibiofemoral involvement1,2,3. In knee OA, affecting the whole joint, articular cartilage loss is often used to measure disease severity and progression. The knee’s 2 joints, the tibiofemoral and patellofemoral, behave differently. The mechanical and biochemical properties of their articular cartilage differ, with patellar cartilage undergoing greater in vivo deformation with loading than tibiofemoral cartilage4. Further, risk factors for incident and progressive patellofemoral and tibiofemoral OA differ5. Knee injury is more closely associated with incident tibiofemoral than patellofemoral OA6. Although greater quadriceps strength protects against patellofemoral cartilage loss, it does not influence tibiofemoral OA progression7. Thus, it is important to consider the different compartments of the knee individually.
Measures of obesity, weight, body mass index (BMI), and body composition are recognized modifiable risk factors for tibiofemoral OA8. When body composition, which differentiates between fat mass and fat-free mass, is considered, a distinct effect of fat mass and fat-free mass has been shown9. Fat mass independent of fat-free mass is associated with a detrimental effect on cartilage volume, but not defects9. In contrast, fat-free mass independent of fat mass is positively associated with cartilage volume9. The effect on cartilage volume has been shown to be partially attributable to an independent effect of leptin10, suggesting a metabolic component10. Thus, obesity influences the risk of tibiofemoral OA by both biomechanical and systemic factors9,10.
Though obesity is a recognized risk factor for patellofemoral OA6,11, it is unclear whether it affects patellar cartilage, and if so, whether by biomechanical or systemic mechanisms. It is important to systematically analyze how obesity affects patellar cartilage prior to onset or during progression of OA. Magnetic resonance imaging (MRI), which visualizes all joint tissues and identifies early changes, can be used to study pathogenesis. Cartilage defects represent local focal cartilage abnormalities, which predict accelerated cartilage loss12. They are graded using a semiquantitative system where grade 0 is normal cartilage and grade 4 is a cartilage defect extending from the joint surface to subchondral bone13. The amount of patellar cartilage (cartilage volume) is associated with patellofemoral joint space narrowing and radiographic severity of patellofemoral OA14. Both cartilage defects and reduced cartilage volume have been independently related to increased risk of arthroplasty13. Determining these relationships can improve our understanding of the pathogenesis of patellofemoral OA, leading to development of more effective management strategies. Thus, the aim of our review was to systematically examine the evidence for an association between measures of obesity, weight, BMI, and body composition (fat mass and fat-free mass) and patellar cartilage assessed using MRI.
MATERIALS AND METHODS
The systematic review was performed according to the 2009 Preferred Reporting Items for Systematic reviews and Meta-Analyses guidelines15.
Selection criteria
Studies that reported the association between obesity measures and patellar cartilage in adults aged ≥ 18 years in the general population, participants with or without knee pain or knee OA, or participants with or without overweight/obesity were considered for inclusion. Studies that evaluated patellar cartilage using MRI and related this to measures of obesity were also included. Studies were excluded if the results were unavailable as an original research article (conference reports, case studies, review articles, or images). Studies were excluded if they dealt with participants’ post-knee arthroscopy, osteotomy, allograft, chondrocyte implantation, or meniscectomy or anterior cruciate ligament reconstruction, or if the underlying pathology was not OA, e.g., rheumatoid arthritis, gout, and malignancy. This was a systematic review of published articles and no ethical approval was needed.
Data sources and search strategy
Three electronic databases (MEDLINE, EMBASE, and CINAHL) were searched up to April 2016 using full text and MeSH terms to identify articles examining obesity or body composition, including “body weight,” “body weights and measures,” “obesity” and “adipose tissue,” “body composition,” “body mass index,” “weight,” “fat mass,” and “muscle.” To identify patellar cartilage, “patella” and “patellofemoral joint” were used. All terms were included as full text, with truncation used to identify variations in terminology. Reference lists of published articles were examined to identify additional sources. Searches were limited to human studies, published in English. Database search strategies are listed in Appendix 1 (available from the authors on request). Figure 1 shows the search results and study selection.
Selection of studies
Two authors (SMH and MT) independently reviewed records for eligibility by title, abstract, and then full text in a 3-stage determination method. Any disagreement was resolved by discussion with another author (AEW).
Risk of bias assessment
Two independent reviewers (SMH and LC) assessed the internal validity and risk of bias for each study using the US National Heart, Lung, and Blood Institute quality assessment tool for observational studies, which includes 14 criteria16. Each criterion is rated as “yes,” “no,” “cannot determine,” “not applicable,” or “not reported.” Overall judgment of bias risk is rated as low, moderate, or high according to the provided guidance16. This tool has been used in assessing internal validity and risk of bias in systematic reviews of several diseases17,18,19.
Data extraction
Three authors (SMH, MT, and YW) independently extracted data and tabulated them. These were cross-checked by another author (AEW). The data were extracted on (1) study characteristics: study design, year, country, number of participants, proportion of women, mean age of participants, years of followup, and OA status; (2) measures of obesity and body composition; (3) assessment techniques of structural change(s) in the patellofemoral joint; and (4) study results. The articles were presented according to measure of obesity, and then study design. A cohort study was considered the strongest study design because it potentially provides a higher grade of evidence than a case-control or cross-sectional study20.
RESULTS
Study selection
The database search identified 1446 studies (400 MEDLINE, 1001 EMBASE, and 45 CINAHL; Figure 1). After removal of duplicates, 1081 studies were screened. Based on title and abstract, 1009 articles were excluded because their outcomes were patellar tendon rupture, patella dislocation, or patellar instability. The search retrieved studies in which BMI or obesity was included as a confounder, but not as an exposure. From the remaining 72 full-text articles, 17 articles met inclusion criteria. Screening of the reference lists of included articles did not identify any additional articles.
Characteristics of included studies
Seventeen studies examined the relationship between measures of obesity, body composition, and patellar cartilage (Table 1)12,21–30,31,32,33,34,35,36. Of these, 5 were cross-sectional21,22,23,24,25, 2 related current patella cartilage to both obesity measured at the time of imaging and also change in obesity over the 10 years prior to imaging26,27, and 10 cohort studies examined the relationship between measures of obesity and change in patellar cartilage over time12,28,29,30,31,32,33,34,35,36. Thus, 7 studies reported the association between obesity and cartilage volume measured at the same time21,22,23,24,25,26,27, 2 assessed whether change in obesity over the preceding decade was associated with current patellar cartilage26,27, and 10 examined whether measures of obesity affected change in patellar cartilage over time12,28,29,30,31,32,33,34,35,36.
Of the 17 studies, 12 were performed in Australia12,21,23,26,27,28,30,31,32,33,35,36, 4 in the United States24,25,29,34, and 1 in Turkey22 (Table 1). Most participants were recruited from the community12,22,23,25,31,33,34,36, with the remainder recruited from existing cohorts (Melbourne Collaborative Cohort Study27,35, Osteoarthritis Initiative29, Geelong Osteoporosis Study26, Tasmanian Older Adult Cohort30). One study examined participants who were in an ongoing study evaluating the effects of exercise on OA24. Three studies examined adult children of knee replacement recipients for primary knee OA matched to controls from the electoral roll21,28,32.
The participants’ ages ranged from 24–82 years. The proportion of women ranged from 0% to 100% (median 60%). Nine studies included participants with neither clinical nor radiographic patellofemoral OA12,23,25,26,27,29,33,35,36 with the remainder including participants both with and without radiographic or clinical patellofemoral OA21,28,30,32. Only 3 studies consisted predominantly of participants with knee OA24,31,34. One study did not mention the OA status of participants22 (Table 1).
All 17 studies assessed measures of patellar cartilage, including patellar cartilage volume, thickness, defects/semiquantitative measures (whole-organ magnetic resonance imaging score), and quality. Three studies examined patellar cartilage volume and presence of patellar cartilage defects23,26,27, 3 examined cartilage volume only28,33,35, 1 examined the presence of cartilage defects and cartilage thickness21, 3 examined the presence of patellar cartilage defects only22,29,34 (including 2 that measured patellar cartilage defects semiquantitively29,34), 3 examined change in patellar cartilage defects12,30,32, 4 examined change in patellar cartilage volume12,28,31,36, and 2 examined cartilage quality by measuring either transverse relaxation (T2) time24 or cartilage strain25.
Obesity was measured using body weight and BMI, and body composition, fat-free mass, and fat mass using bioimpedance analysis. Nine studies used BMI only as the measure of obesity21,22,24,25,28,29,31,32,33, 4 studies used BMI and body weight12,23,26,36, 2 studies used BMI, weight, fat mass, and fat-free mass27,35.
Bias and methodological quality assessment
Table 2 provides details of the risk of bias and quality assessment. Six studies were judged to be at high risk of bias21,22,23,24,25,32, 6 were judged as moderate12,26,28,29,30,36, and 5 were low risk27,31,33,34,35. For most of the studies, the power calculation was not shown. However, apart from 2 studies25,33, others had large numbers of samples (≥ 100 participants). Many studies did not report the frequency of measurement of exposure. The reviewers were not blinded to the study (authors, title, and source). The rate of initial agreement between the 2 reviewers was 98.3%. Differences in scoring between reviewers were evaluated and resolved by consensus. Where the 2 reviewers could not achieve consensus through discussion, a third reviewer (AEW) adjudicated.
Body weight and patellar cartilage
Three cross-sectional studies23,26,27, including 2 that also examined change in weight over the previous decade26,27, assessed the relationship between body weight and patellar cartilage (Table 3).
Body weight and patellar cartilage defects: All cross-sectional analyses found that higher weight was significantly associated with greater prevalence of patellar cartilage defects ranging from 4% to 9%23,26,27. The 2 studies that examined whether change in weight over the previous decade related to patellar cartilage defects found inconsistent results26,27. Gunardi, et al found no significant relationship between increasing weight and the prevalence of patellar cartilage defects in women26. In contrast, Teichtahl, et al identified a significant positive relationship between increasing weight and prevalence of patellar cartilage defect in women, but not in men27.
Body weight and patellar cartilage volume: Three studies23,26,27 assessed the relationship between current weight and patellar cartilage volume, with 1 showing reduced cartilage volume associated with greater weight in women only27. Of the 2 studies that also assessed the association between change in weight over the past decade and cartilage volume26,27, 1 study that examined only women found that increased weight was significantly associated with reduced cartilage volume27. The only longitudinal study35 to examine the relationship between baseline weight and subsequent patellar cartilage volume loss found no significant relationship. Similarly, the only longitudinal study36 to examine the relationship between weight change and change in patellar cartilage volume found no significant association.
BMI and patellar cartilage
Eight cross-sectional analyses21,22,23,24,25,26,27,33, including 2 from longitudinal studies that examined change in BMI over the previous years26,27, and 6 longitudinal studies28,29,31,33,34,35 assessed the relationship between BMI and patellar cartilage (Table 4).
BMI and patellar cartilage defects: All 5 cross-sectional analyses found that BMI was associated with increased odds of patellar cartilage defects, ranging from 9% to 29%21,22,23,26,27. One study showed that change in BMI over the past decade was not associated with cartilage defects26,29, whereas another found a significant association in women27. Four cohort studies found that higher baseline BMI was not associated with the progression of cartilage defects12,30,32,34. In the Osteoarthritis Initiative, a 5% increase in BMI over time was associated with progression of patellar cartilage defects29.
BMI and patellar cartilage volume: Although 3 cross-sectional analyses showed that increased BMI was associated with reduced cartilage volume in women only23,26,27, 1 study found no association between BMI and rate of patellar cartilage volume change23. Change in BMI over previous years was not associated with patellar cartilage volume change27. Three cohort studies found that higher baseline BMI was associated with increased patellar cartilage volume loss28,31,35. However, in 1 study, the association applied only to those in the top BMI tertile (cutoff points not specified)28.
BMI and patellar cartilage quality: Two studies examined the relationship between BMI and cartilage quality24,25. One showed that higher BMI was associated with higher T2 values, signifying structural changes of patellofemoral OA24. The other found that higher BMI was not associated with cartilage strain25.
Body composition and patellar cartilage
Two studies examined the relationship between body composition and patellar cartilage (Table 5)27,35.
Fat mass and patellar cartilage defects: One study examined the relationship between fat mass and defects, and found that increased fat mass was associated with increased patellar cartilage defects among men only27. Change in fat mass over the past decade was not associated with increased prevalence of cartilage defects27.
Fat mass and patellar cartilage volume: Cartilage volume was not significantly related to fat mass or change in fat mass over the past decade27. Higher fat mass was associated with increased cartilage loss in women, but not men35. In the longitudinal study, change in fat mass was not significantly related to change in patellar cartilage volume35.
Fat-free mass and patellar cartilage defects: The presence of patellar cartilage defects was associated with fat-free mass and change in fat-free mass over the preceding decade27. In cross-sectional analysis, higher fat-free mass was associated with increased prevalence of cartilage defects in women only27. An increase in fat-free mass over the previous decade was associated with a higher prevalence of cartilage defects in men and women27.
Fat-free mass and patellar cartilage volume: One study examined the relationship between patellar cartilage volume and fat-free mass27. There was no significant association between fat-free mass and patellar cartilage volume at the time of the MRI. There was also no change in fat-free mass over the previous decade significantly associated with current patellar cartilage volume27.
Summary of the evidence synthesis
There was consistent cross-sectional evidence for a positive association between weight or BMI, and patellar cartilage defects21,22,23,26,27. Evidence for the relationship between prior change in weight or BMI and the prevalence of cartilage defects was varied26,27. The 4 studies that examined the relationship between BMI and subsequent change in patellar cartilage defects showed no significant relationship12,30,32,34. There was a consistent direction of a detrimental effect of measures of obesity and patellar cartilage volume in cross-sectional analyses, although 2 found significant results23,26,27. The 4 studies examining whether BMI was associated with increased cartilage volume loss showed results in the same direction28,31,33,35. However, only 3 studies showed a significant relationship, 1 involving both sexes and at low risk of bias31, 1 involving both sexes but only in participants in the top BMI tertile and at moderate risk of bias28, and 1 involving only women35. Two studies examined the association between body composition and cartilage defect/volume27,35; thus, no definitive conclusion could be drawn relating to body composition. When the 4 studies at high risk of bias were excluded from the summary of the evidence, the conclusions did not change21,22,23,32.
DISCUSSION
Our systematic review examined the relationship between obesity and the patellar cartilage across the spectrum of OA, from healthy to preclinical and then to symptomatic and radiographic disease. In asymptomatic middle-aged adults, a consistent detrimental influence of elevated weight and BMI on patellar cartilage was seen in all significant studies. Results from the few available cohort studies showed no effect of obesity on the progression of cartilage defects, but a tendency toward increased cartilage loss, with half the studies showing statistically significant results. No studies found a statistically significant beneficial effect of obesity on patellar cartilage.
The relationship between obesity and patellar cartilage change seemed stronger in women. A number of factors may contribute to this observation. First, women have relatively higher fat mass compared to men37. Increased fat mass is associated with higher levels of inflammatory cytokines (e.g., C-reactive protein, interleukin 6), which are detrimental to knee structure, cartilage in particular10,38. The sex difference may therefore relate to differences in the metabolic milieu. Second, the biomechanics of the patellofemoral joint differ in men and women, affecting loading and risk of OA39. Cadaveric female knees showed greater change in contact pressures to varying vastus medialis load at knee flexion angles compared with male knees, suggesting sex differences in patellofemoral contact areas and pressures39. Third, studies may have had limited power to detect an association in men since OA is more common in women and participants were mainly women. In studies where results were found to be significant in women but not men, the point estimates of effect were in the same direction for men as women, but did not reach statistical significance. However, there were differences, such as a significant association between fat mass and cartilage defects seen in men, but not in women27.
Despite increasing interest in a metabolic mechanism for the relationship between obesity and tibiofemoral OA10,40,41, only 2 studies examined the relationship between body composition and patellar cartilage27,35. While they suggested a detrimental relationship between fat mass and patellar cartilage in women and men27,35, 1 found a detrimental association of fat-free mass with cartilage defects in women27. This may suggest a stronger involvement of mechanical factors at the patellofemoral joint than at the tibiofemoral joint. At the tibiofemoral joint, a consistent detrimental relationship between fat mass and cartilage volume and defects was seen, with a beneficial relationship of fat-free mass and cartilage volume8. This finding at the patellofemoral joint in only 1 study requires further verification27.
Both metabolic and biomechanical factors are likely to contribute to the relationship between obesity and patellofemoral OA. Metabolic factors exemplified by leptin, which is raised in obesity, have been associated with reduced patellar cartilage volume, independent of BMI10. Increased loading by obesity may also affect patellar cartilage and its biomechanical properties42. The relative contributions of metabolic and biomechanical mechanisms to the initiation and progression of patellofemoral OA have not been examined and require further work.
Our review was limited with few high-quality and cohort studies. It was not until the late 1990s that MRI began to be widely used to investigate knee structure, albeit with little emphasis on the patellofemoral joint. Most of the performed studies were cross-sectional, thus limiting the level of evidence able to be extracted. The few longitudinal studies had relatively brief followup periods (2 yrs on average)12,30,31,32,33,35,36, which might be inadequate to detect patellar cartilage changes. Previous studies may have examined the relationship between patellar cartilage and obesity, but used an overall measure such as cartilage strain25 or T2 relaxation time24, and therefore did not identify separate relationships with cartilage volume and defects. Further, the magnitude of change in obesity measures was not large in the cohort studies. The differences in participants’ characteristics such as mean age and body weight/BMI might explain the inconsistent results presented in our review. It is possible that the effect of obesity on patellar cartilage may differ according to age and the severity of OA in the joint. However, given the limited number of longitudinal studies that used a variety of measures, and few studies in those with OA, the existing data are restricted in their capacity to identify this possibility. Results were similar in studies examining those with and without OA. Patellofemoral degeneration associated with malalignment or patellar incongruity has been proposed to influence patellar cartilage and may be a confounding factor. Measures of incongruity have not been accounted for, which may explain differences between study results. Nevertheless, there is no consensus regarding which should be included in analyses or their determinants43. Incongruity may lie on the pathway between obesity and patellofemoral OA.
We did not perform a metaanalysis for several reasons. First, patellar cartilage was measured using a variety of measures, each identifying a complementary construct or dimension of cartilage. Second, even where the same outcome was used, results were presented differently and adjusted for different cofactors, possibly introducing publication bias. Thus, we performed a systematic review, including data from studies where the effect of obesity was not necessarily the primary outcome.
Our systematic review identified some evidence for a detrimental association between obesity and patellar cartilage. Specifically, evidence was consistent for a relationship between greater weight and BMI and prevalence of patellar cartilage defects, particularly in women. Evidence was suggestive for an association between BMI and patellar cartilage volume loss. No clear conclusion could be drawn for the association between body composition and patellar cartilage. Therefore, more high-quality research is required to confirm these findings and to better understand the relative contributions of metabolic and biomechanical factors to the pathogenesis of patellofemoral OA, so that effective strategies to manage this common and disabling condition can be devised.
Footnotes
SMH is the recipient of the Arthritis Foundation of Australia Heald Fellowship, YW is the recipient of a National Health and Medical Research Council (NHMRC) Career Development Fellowship (level 1, #1065464), DMU is the recipient of an NHMRC Career Development Fellowship (level 1, #1011975), and AEW is the recipient of an NHMRC Career Development Fellowship (level 2, #1063574).
- Accepted for publication March 22, 2017.