Recent studies [29–32] have extensively examined the relationship between obesity and osteoarthritis. However, the evidence in the available studies is limited by showing observational correlations, and the results may be influenced by confounding factors. This study aimed to elucidate the causal effects of obesity on hip osteoarthritis. In this study, we used MR to investigate the association between obesity and the risk of hip osteoarthritis based on existing GWASs. Our results suggest that genetically predicted obesity is significantly associated with an increased risk of hip osteoarthritis. The risk of hip osteoarthritis increased by 96% for each one standard deviation increase in BMI.
In our study, statistical significance was assessed using the two-sample MR method, with summary statistics from the largest study on BMI (n = 339,224) and hip osteoarthritis (up to n = 2,396 hip osteoarthritis cases and 9,593 controls). Our results show a causal relationship between obesity and hip osteoarthritis. To check the reliability of our findings and minimize potential pleiotropic effects, we performed a series of sensitivity analyses. The results were robust when sensitivity analyses were performed using MR-PRESSO, MR‒Egger regression intercept, and leave-one-out analysis. We searched in Phenoscanner and the GWAS catalog and confirmed no evidence of pleiotropy, which suggests that our results are reasonable. Although we require the F value of the instrumental variable to be greater than 10 to eliminate weak instrumental variables, to avoid potential weak instrumental variables from affecting the results, we conducted MR.RAPS analysis, and the results were not affected. Our findings are also similar to previous results [33]. The MR Steiger test further showed that there was no evidence of reverse causality in our study. Our findings are also similar to previous findings. However, in the previous MR study, they used GWASs from mixed populations, which is a very large limitation, as they describe in the discussion. Therefore, it is necessary for us to perform MR analysis in a pure European population. Another limitation is the high heterogeneity of their studies and the substantial horizontal pleiotropy (P < 0.001) by using the MR-presso test. In addition, they did not remove IVs that exhibited reverse causality. Of course, our study is also different from the previous study. Our findings were specific to osteoarthritis of the hip, not the knee. We consider specific joint sites to be more specific than general osteoarthritis. In summary, we believe that our study provides a valuable contribution.
Osteoarthritis is an irreversible chronic disease characterized by two main pathological features: articular cartilage damage and subchondral sclerosis, which involve the entire joint, including cartilage degeneration, bone remodeling, osteophyte formation, and synovial inflammation, resulting in joint pain, stiffness, and loss of normal function. Inflammation of the synovium results in excessive production of synovial fluid, swelling of the joints, and inhibition and chronic disuse of the muscles that connect the joints, eventually leading to muscle wasting and disability. Pain is the main symptom and the main driver of clinical decision-making and health service use (3). Risk factors for osteoarthritis include fixed (e.g., age, sex) and modifiable (e.g., overweight or obesity, physical activity) factors, at least in principle. To date, the link between being overweight or obese and osteoarthritis has been consistently demonstrated in osteoarthritis of the knee [34]. However, data on hip joints, obesity, and osteoarthritis have been inconsistent. In a population survey from the United States [35], obesity was not associated with unilateral hip osteoarthritis. Similarly, in a prospective study of the Finnish population [36], BMI was not a predictor of risk factors for hip osteoarthritis. Similarly, in a prospective study of the Finnish population, BMI was not a predictor of risk factors for hip osteoarthritis. In contrast, the study by Holliday et al. [37] showed that each SD increase in BMI was associated with a 65% increased risk of hip osteoarthritis (P < 0.001). A prospective cohort study of more than 120,000 people found that only higher BMI and older age were associated with an increased risk of osteoarthritis [38]. In particular, women with the highest BMI had twice the risk of hip replacement surgery than women with the lowest BMI. A meta-analysis based on 14 studies [11] showed that BMI was significantly positively associated with hip osteoarthritis risk. Thus, there is increasing evidence of deleterious effects of obesity on hip osteoarthritis.
The mechanism of action of obesity on hip osteoarthritis remains unclear. At present, it is believed that obesity mainly causes hip osteoarthritis as a result of weight-bearing and nonweight-bearing aspects. The cause of the onset and progression of hip osteoarthritis may be abnormal mechanical loading due to weight gain in weight-bearing joints. The main function of articular cartilage is to provide the joint with a smooth surface with a low coefficient of friction and to facilitate the transmission of loads during joint motion [39]. When the cartilage load exceeds the critical value, articular cartilage will be irreversibly damaged. At first, articular cartilage swelling and proteoglycan loss are significantly increased, hypertrophic chondrocytes are activated, and then chondrocyte necrosis and cartilage thinning occur gradually [40]. In addition, supraphysiological loading can lead to increased proinflammatory cytokines in articular chondrocytes and synovial cells, thereby promoting a vicious cycle of synovial joint pathological changes. Several studies [41–43] have shown no increase in subchondral sclerosis with increasing BMI, confirming that the increase in symptoms in obese patients cannot be explained by more structural joint damage. Therefore, obesity may be involved in the pathogenesis of hip osteoarthritis not only by increasing mechanical load but also via local inflammation. Schelbergen et al. [44] showed that the activated macrophage-associated alarmin S100A/S100A9 induces overexpression and activation of matrix metalloproteinases, leading to cartilage matrix remodeling and osteophyte formation. Then, abnormal production and secretion of adipokines (including leptin, adiponectin, resistin, etc.) are also thought to play an important role in the association between obesity and systemic inflammation. For example, the consumption of a high-fat diet led to a significant increase in leptin, which was positively associated with OA-related cartilage damage, osteophytes, and increased infrapatellar fat pad size [45]. To date, the specific mechanism of action of obesity-induced hip osteoarthritis remains unclear.
Our study has several major strengths. The main advantage is the MR design, which estimates the causal effect of obesity on hip osteoarthritis without interference from residual confounding or reverse causality. Regarding causal inference from observational studies, no matter how well designed the epidemiological study is and how accurate the measurements are, potential, unmeasurable confounders cannot be eliminated. We performed MR.steiger filtering to exclude all SNPs that primarily affected hip osteoarthritis but not BMI. We also performed the MR.RAPS method, which can provide robust inferences for our MR analysis with many weak IVs. Finally, we also used various sensitivity analysis methods to ensure the robustness of the results.
Of course, our MR study had several limitations. First, the available data we use are summary-level statistics, not individual-level statistics. Second, our MR analysis is based on individuals of European ancestry; therefore, validation for individuals of other ancestries is needed. Finally, exposure and outcome studies used in two-sample MR analyses should not involve overlapping participants. We were unable to estimate the degree of overlap in the studies. To avoid this, we used GWAS data from different Consortiums while using powerful tools to minimize overlap (e.g., F-statistics much greater than 10).