We have demonstrated a threefold to fourfold increased risk of primary hip and knee joint replacement for OA, when comparing the fourth quartile with the first quartile, for weight, BMI, FM, and percentage fat. The waist circumference and WHR were less strongly associated with the risk. When knee and hip replacements were examined separately, all adiposity measures persisted as risk factors for joint replacement at either anatomical site – with the exception of WHR, which was not significantly associated with hip replacement risk. Moreover, when comparing the strength of the associated risks between the adiposity measures and knee and hip joint replacement, all adiposity measures were stronger risk factors for knee replacement rather than hip joint replacement.
There are few previous studies examining the relationship between directly measured adipose mass or the distribution of adipose mass and the risk of joint replacement. Most studies have employed BMI as a measure of obesity, and have shown a consistently positive association between BMI and the risk of both knee replacements [13
] and hip replacements [13
] for OA – as has the present study. In addition to examining the association with BMI, it would be necessary to examine whether the pattern of fat distribution or body composition affects this risk [18
], in order to explore the mechanism that may explain the association between obesity and the increased risk of joint replacement due to severe OA. In a cohort of the Swedish general population (the Malmo Diet and Cancer study), Lohmander and colleagues found that all body mass measures were significant risk factors for knee and hip OA leading to joint replacement [20
]. Consistent with this study, we also showed that measures of adipose mass (FM and percentage fat) were associated with an increased risk of primary knee and hip joint replacement 10 to 15 years after their measurement. Whilst both measures of central adiposity (waist circumference and WHR) were associated with an increased risk of primary knee replacement, only waist circumference but not WHR was significantly associated with the risk of primary hip replacement. Moreover, for all measures of obesity, stronger evidence was observed for the knee than for the hip.
The Malmo Diet and Cancer study [20
] and our study are the only two prospective studies of which we have knowledge that have investigated the association of different adiposity measures (including direct measurement of percentage fat using bioelectrical impedance) with joint replacement. The findings of the Malmo Diet and Cancer study (n = 27,960) have been confirmed in our larger study (n = 39,023). Our study also included about 24% of participants who were Southern European migrants, whereas the Malmo Diet and Cancer study excluded participants with a lack of Swedish language skills, thus resulting in a more homogeneous population. This significantly strengthens the findings since similar results have been found in different populations, suggesting that the association is more likely to be causal since it is unlikely that both studies were subject to the same type of errors (chance, bias or confounding).
The mechanism for the associations between adiposity measures and the risk of primary knee and hip joint replacement is unclear, but may be due to both biomechanical and metabolic factors. The adipose mass, by virtue of its added body mass, contributes to an increased joint loading, which may increase the risk of OA progression and subsequent joint replacement performed for severe end-stage OA. This biomechanical hypothesis may be most evident at the knee – given the anatomical disadvantage that the knee joint lacks a stable bony configuration compared with the hip, whereby load is disproportionately distributed to the medial tibiofemoral compartment during dynamic tasks – and it has been shown that much of the effect of BMI on the severity of medial tibiofemoral OA was explained by varus malalignment [30
]. The association between fat distribution and the risk of knee and hip OA has been investigated in different study populations with inconsistent results. While some studies showed positive associations [31
], other studies showed no association [5
Nevertheless, metabolic factors are also likely to be important since we have shown that waist circumference and WHR, the surrogate measures of central adiposity and known risk factors for the metabolic syndrome [34
], were more strongly associated with the risk of knee replacement than hip replacement. Indeed adipose tissue, which was once thought to be a passive store of energy, is now considered an endocrine organ, releasing a multitude of factors, including cytokines such as TNFα and IL-6, as well as adipokines, such as leptin, adiponectin and resistin [35
]. Both TNFα and IL-6 have been implicated in cartilage destruction in OA [36
], while leptin is a key regulator of chondrocyte metabolism and plays an important role in the pathophysiology of OA [38
]. Such findings demonstrate the potential role of metabolic factors related to adiposity in the context of OA and, ultimately, of joint replacement.
The WHR is a surrogate measure of central adiposity that includes the visceral and abdominal subcutaneous depots. Recent data have shown some biological differences between intraabdominal visceral fat and peripheral subcutaneous fat [39
]. Visceral adipose tissue and its adipose-tissue resident macrophages produce more proinflamatory cytokines, like TNFα and IL-6, and less adiponectin [39
]. Leptin secretion is greater from subcutaneous than from visceral fat tissue [40
]. A limitation of the WHR is that it is not able to discern between the metabolically and physically different types of fat. In addition, the WHR becomes even less reliable in people who have both greater central and gynoid fat, and therefore may lead to an underestimation of observed associations. This underestimation may in part explain the lower sensitivity and weaker association of WHR with joint replacement than the other adiposity measures.
We had virtually complete follow-up in this prospective study as the identification of incident primary knee and hip replacement was done by record linkage to the NJRR, which has complete coverage of the cohort participants. While the recruitment of MCCS participants and data collection commenced in 1990 to 1994, the NJRR started joint replacement data collection in Victoria in 2001. We therefore do not have complete and reliable joint replacement data for the study population prior to 2001. Although we excluded those MCCS participants who reported a joint replacement prior to 1 January 2001 at the second follow-up, this information may be unreliable and is only known for 68% of the original cohort. As a result, some misclassification of joint replacement status may have occurred – although it is likely to have been nondifferential in relation to the adiposity measures, which may have underestimated the strength of any observed associations. The MCCS did not collect data on occupational activities such as bending and lifting, and thus we were unable to adjust for these factors in the analysis. Although total joint replacement is used as a proxy for severe symptomatic OA, the utilization of joint replacement in the treatment of OA may be influenced by a number of factors such as access to healthcare, physician bias, and patient-level factors, in addition to disease severity [41
]. We therefore adjusted for age, gender, country of birth, and highest level of education in the analysis to counter this issue.
A particular issue for bioelectric impedance analysis is the absence of a standard equation to estimate the fat-free mass. We chose a formula developed using subjects of similar ethnicity, age, and BMI distribution to the MCCS population [23
] that was validated using sound statistical techniques. There is evidence that body hydration, a status difficult to assess in large epidemiological studies, has a strong effect on the estimation of FM based on bioelectric impedance analysis [42
]. Any between-subject variability in hydration level in the current study would therefore have resulted in greater attenuation of the relationship between FM and the risk of primary joint replacement. Another concern is that the measurement error in the anthropometric variables would have underestimated the associations observed in the study, and this effect would be greatest for the bioelectric impedance analysis-based measures.