Osteoarthritis (OA) is a chronic degenerative joint disease with increasing global prevalence, becoming a leading cause of disability among middle-aged and elderly populations (1). OA represents a complex, whole-joint pathology characterized by progressive degenerative and pathological changes across all articular tissues, including cartilage, subchondral bone, synovium, menisci, ligaments, and infrapatellar fat pad (IFP). Each of these tissues undergoes specific but interconnected degenerative processes that collectively promote disease progression (2, 3). With accelerating population aging and the obesity epidemic, the disease burden of OA is expected to further intensify, posing enormous challenges to healthcare systems and socioeconomic structures (4).

The infrapatellar fat pad (IFP), despite being frequently underrepresented in OA research, plays a significant role as an active local player in OA pathology. The IFP becomes inflamed and fibrotic, and undergoes biomechanical changes affecting OA progression (5). Furthermore, IFP functions as a local source of inflammatory cytokines, which promote joint inflammation and facilitate both peripheral and central sensitization, exacerbating pain in knee OA (6, 7). These inflammatory mediators contribute to cartilage degradation and synovial inflammation, further accelerating the degenerative processes within the joint (8).

OA is a multifactorial disease with several risk factors involved in its onset and progression. These include age, female gender, obesity, genetic predisposition, mechanical load, previous joint injuries, metabolic syndrome, diabetes, hyperlipidemia, and other metabolic disorders (9, 10). Epidemiological studies indicate that patients with Diabetes mellitus (DM) and Prediabetes mellitus (Pre-DM) have significantly higher risk of OA compared to the general population, an association that extends beyond the influence of common risk factors such as age and obesity (11, 12).

The relationship between metabolic disorders (particularly DM/Pre-DM) and OA represents a complex pathophysiological network with multiple interconnected mechanisms. Diabetes creates a systemic metabolic environment characterized by chronic low-grade inflammation, oxidative stress, insulin resistance, and accumulation of advanced glycation end products (AGEs) that directly impacts joint tissues (13). AGEs accumulation in articular cartilage increases its stiffness and brittleness, compromising its biomechanical properties and resistance to mechanical stress. Meanwhile, the chronic inflammatory state activates synoviocytes and chondrocytes to produce matrix-degrading enzymes, accelerating cartilage breakdown. Additionally, hyperglycemia-induced oxidative stress disrupts normal chondrocyte function and promotes cellular senescence, further impairing cartilage repair capacity.

In this complex metabolic-inflammatory environment, accurate assessment of body composition and fat distribution becomes critical, as these factors significantly influence both metabolic health and joint pathology. Traditional anthropometric measures like BMI have limitations in characterizing the metabolic impact of adiposity, as they fail to distinguish between different fat distribution patterns and their metabolic consequences (14, 15). This is particularly relevant in DM/Pre-DM patients, where visceral adiposity and altered fat metabolism play central roles in disease progression and complication development.

Recent studies emphasize that fat distribution patterns have a more critical impact on metabolic health than total fat mass (14). Visceral fat accumulation and dysfunction are considered core links connecting obesity, metabolic diseases, and systemic inflammation (15). The Visceral Fat Metabolism Score (METS-VF), as an emerging composite indicator, integrates anthropometric parameters and blood biochemical markers to accurately reflect visceral fat functional status (16). Compared to traditional visceral fat measurement methods, such as Computed Tomography (CT) or Magnetic Resonance Imaging (MRI), METS-VF offers advantages of being non-invasive, cost-effective, and easily applicable in clinical settings (17). Similarly, the BRI, based on principles of human geometry, evaluates body shape distribution characteristics through the relationship between waist circumference and height, and has been proven to reflect central obesity more accurately than Body mass index (BMI) (18). Previous research has demonstrated that Body Roundness Index (BRI) is closely associated with the risk of various metabolism-related diseases (19).

The interrelationship between METS-VF, BRI, and OA in the context of diabetes and prediabetes represents a critical conceptual framework for understanding metabolic OA pathogenesis. METS-VF captures the functional metabolic status of visceral fat, which serves as a primary source of pro-inflammatory adipokines and cytokines that can directly impact joint tissues. These inflammatory mediators, including TNF-α, IL-1β, and IL-6, propagate systemic low-grade inflammation and directly contribute to cartilage matrix degradation when reaching the joint environment (20). BRI, by characterizing body shape and central adiposity, reflects the mechanical loading component of OA risk while also serving as a proxy for metabolic dysfunction associated with central obesity. In patients with diabetes and prediabetes, these relationships are particularly pronounced, as hyperglycemia amplifies both the inflammatory effects of visceral adiposity (captured by METS-VF) and the metabolic consequences of altered body fat distribution (reflected by BRI).

The clinical significance of understanding these relationships extends beyond academic interest. Current clinical practice lacks effective tools for early identification of high-risk individuals for OA among DM/Pre-DM patients, severely limiting the implementation of targeted prevention and intervention strategies (21). By elucidating how METS-VF and BRI relate to OA risk in the diabetic population, we can potentially develop more effective risk stratification approaches that consider both metabolic and biomechanical components of OA pathogenesis. This would facilitate personalized preventive strategies that simultaneously address hyperglycemia, adiposity-related inflammation, and altered biomechanics, potentially interrupting the vicious cycle that accelerates joint degeneration in this high-risk population.

Despite the promising applications of METS-VF and BRI in assessing metabolic health and body shape distribution, their association with OA risk in DM/Pre-DM patients has not been systematically investigated. Considering that visceral fat dysfunction and altered body shape distribution may be key links connecting metabolic disorders with OA development (20), exploring the value of METS-VF and BRI in OA risk assessment among DM/Pre-DM patients has significant clinical implications. Furthermore, determining optimal cut-off values for these indicators and evaluating the additive effect of their combined application would help construct more precise risk prediction models (22). Research findings based on large representative populations can provide scientific evidence for developing OA prevention strategies targeted at DM/Pre-DM patients, promoting precision medicine practices (23).

Therefore, this study aims to investigate the association between METS-VF and BRI with OA risk in DM/Pre-DM patients based on nationally representative NHANES 1999–2020 data, evaluate their clinical application value as risk prediction tools, and further analyze the potential advantages of their combined application. These findings will address current research gaps, provide clinicians with practical risk assessment tools, and ultimately improve early identification and intervention strategies for OA in DM/Pre-DM patients.

This study is the first systematic investigation of the associations between METS-VF, BRI, and the risk of OA in patients with diabetes or prediabetes. Based on extensive representative data from the NHANES 1999-2020, we found that both METS-VF and BRI are independent predictors of OA risk in patients with diabetes or prediabetes, with both showing significant dose-response relationships. These findings not only expand the current understanding of the relationship between metabolic factors and OA but also provide new perspectives and tools for clinical risk assessment and early intervention.

Our research demonstrates that, after adjusting for multiple potential confounding factors, each unit increase in BRI was associated with a 15% increased risk of OA, while each unit increase in METS-VF was associated with a 133% increased risk. This strong independent association highlights the important role of fat distribution patterns and metabolic health status in OA pathogenesis. Niu et al.’s findings from the Framingham OA study are highly consistent with our results, reporting that central obesity was significantly associated with the risk of knee OA progression, independent of BMI (27). Furthermore, in a prospective study of the Southern Swedish Health cohort, Lohmander et al. found that waist circumference was positively associated with the risk of severe OA (requiring joint replacement surgery), and this association persisted after adjusting for BMI (28). Our study further extends these findings by specifically focusing on the particular high-risk population of patients with diabetes or prediabetes, a group that is especially important in clinical practice but has received less targeted attention in previous research. Unlike previous studies, we employed more comprehensive and clinically practical indices for assessing metabolic health and body shape distribution. BRI, as a geometric index based on waist circumference and height, more accurately reflects central obesity compared to BMI (29). Meanwhile, METS-VF, as an integrated score incorporating multiple anthropometric and metabolic parameters, may more comprehensively reflect visceral fat functional status than traditional single indicators like waist circumference or waist-to-hip ratio (17). This provides clinicians with more convenient and integrative risk assessment tools.

Through quartile analysis and RCS analysis, this study thoroughly investigated the dose-response relationship between METS-VF, BRI, and OA risk. The quartile analysis revealed a clear stepwise risk increase trend: compared to the lowest quartile, the highest quartile of BRI was associated with a 178% increased risk of OA, while the highest quartile of METS-VF was associated with a 302% increased risk. This significant dose-response relationship further strengthens the reliability and biological plausibility of their associations with OA risk. Notably, RCS analysis revealed a significant non-linear relationship between METS-VF and OA risk (P = 0.001), while BRI demonstrated a linear association (Pnon-linearity=0.088). This finding echoes the results of Collins et al., who identified complex non-linear interactions between metabolic syndrome components and knee OA (30). The concept of a “metabolic joint inflammation threshold” proposed by Schett et al. may provide a theoretical explanation, suggesting that metabolic inflammation might accelerate joint tissue damage beyond a specific threshold (31). As a comprehensive metabolic indicator, METS-VF may more sensitively capture this threshold effect, while BRI, primarily reflecting body shape distribution, may maintain a relatively linear relationship with OA risk. Rosa et al. observed similar non-linear patterns in a cross-sectional study on the relationship between metabolic syndrome and knee OA, speculating that this might be related to the differential effects of various metabolic disorder combinations on the joint metabolic microenvironment (32). Our findings not only align with this speculation but also provide more refined quantitative evidence, indicating that when assessing and managing OA risk in patients with diabetes or prediabetes, the non-linear impact of metabolic indicators must be considered.

The optimal cutoff values for BRI and METS-VF determined through ROC curve analysis in this study were 5.865 and 7.255, respectively, providing objective standards for identifying high-risk individuals in clinical practice. These cutoff values have important clinical translational value, helping healthcare professionals easily screen for high-risk patients who may require more aggressive intervention. Compared to the cutoff values for body fat distribution indicators determined by Dawson et al. in European populations, our results show certain differences (33). This discrepancy may reflect racial differences and the influence of specific disease backgrounds (diabetes/prediabetes). Cheon et al.’s research suggests that Asian populations may require specific body fat distribution indicator cutoff values to accurately assess metabolic risk (34), which aligns with our findings and emphasizes the necessity of validating these cutoff values in populations with different racial and disease backgrounds. Our study specifically determined the optimal cutoff values for METS-VF and BRI in the particular high-risk population of patients with diabetes or prediabetes, filling an important knowledge gap in this field and supporting precision medicine practices.

Surprisingly, our additive effect analysis showed mixed results when evaluating the combined application of BRI and METS-VF for OA risk prediction. When assessed by the conventional discrimination metric (AUC), the combined BRI and METS-VF model (AUC = 0.650) did not significantly improve the predictive ability for OA risk compared to BRI (AUC = 0.632) or METS-VF (AUC = 0.650) alone (P>0.05 for both comparisons by DeLong test).This is not entirely consistent with Berenbaum et al.’s hypothesis, which suggested that integrating multiple metabolic indicators should more accurately predict OA risk (35). However, further analysis using more sensitive measures of predictive improvement revealed a more nuanced pattern. Compared to BRI alone, the combined BRI+METS-VF model showed a significant improvement in risk classification, with a continuous NRI of 0.219 (95%CI: 0.116-0.322, P<0.0001) and an IDI of 0.015 (95%CI: 0.010-0.020, P<0.0001). This indicates that while overall discriminative ability was not significantly enhanced, the combined model did provide meaningful improvements in correctly classifying individuals’ risk levels when compared to using BRI alone. Karvonen-Gutierrez et al. observed similar phenomena, finding that despite similar AUCs for different obesity measurement methods, they might exhibit significant differences in risk classification capability (36). This underscores the importance of utilizing multiple statistical approaches when evaluating prediction models, rather than relying solely on discrimination metrics such as AUC. When comparing the combined model to METS-VF alone, the improvement was much more limited, with a small NRI of 0.003 (95%CI: 0.001-0.005, P = 0.004) and a non-significant IDI of 0.076 (95%CI: -0.026-0.174, P = 0.156). This suggests that METS-VF, as a comprehensive metabolic indicator, may already capture most of the information relevant to OA risk, with BRI providing minimal additional predictive value beyond what METS-VF already contributes. This finding aligns with Hart et al.’s research comparing different metabolic indicators, which found that comprehensive metabolic indicators generally have better predictive ability than single body shape indicators (37). These results suggest that the clinical value of combining BRI and METS-VF depends on the specific context and purpose. For practices currently using BRI alone, adding METS-VF assessment would provide significant improvement in risk classification. However, for those already using METS-VF, there would be limited benefit in additionally measuring BRI.

Our research found that patients with a combination of high BRI (≥5.865) and low METS-VF (<7.255) might have the highest risk of OA, although this association did not reach statistical significance. This finding differs somewhat from traditional views, which suggest that the synergistic effect of metabolic abnormalities and obesity should lead to the highest OA risk (38). Our results may reflect the so-called “metabolic phenotype diversity,” where different metabolic phenotypes might influence OA risk through various mechanisms. Zhuo et al. proposed that the special phenotypes of “metabolically healthy obesity” and “metabolically unhealthy non-obesity” might affect OA onset and progression (13). The high BRI and low METS-VF combination we observed may represent a unique metabolic phenotype, where abnormal body shape distribution (central obesity) coexists with relatively good metabolic parameters. Richmond et al.’s research supports this view, finding that the associations between certain metabolic indicators and OA might vary depending on body type characteristics (39). This suggests that when assessing OA risk, we should not simply assume an additive effect of metabolic abnormalities and obesity, but rather consider the specificity of different metabolic phenotypes.

Subgroup analysis results showed that the associations of BRI and METS-VF with OA risk remained consistent across different age, gender, race, and hypertension status subgroups, with no statistically significant interactions (P>0.05). This consistency enhances the reliability and universality of our main findings, indicating that the roles of METS-VF and BRI as predictors of OA risk are relatively robust and not significantly affected by major demographic characteristics. This result aligns with Felson et al.’s multicenter study findings, which showed that the impact of metabolic factors on OA risk was relatively stable across populations with different demographic characteristics (40). However, contrary to our findings, Sowers et al. reported that the influence of metabolic syndrome on OA risk might be greater in women than in men (41). Lee et al.’s study also found that the association between visceral fat and OA might be particularly strong in postmenopausal women (42). These differences may stem from variations in study design, sample size, OA definition methods, and adjusted confounding factors.

Nevertheless, this study has several limitations that need to be considered. Firstly, the cross-sectional design limits causal inference, making it impossible to determine whether elevated METS-VF and BRI are causes or consequences of OA, or whether both are influenced by common pathophysiological mechanisms; this requires further verification through prospective cohort studies. Secondly, OA diagnosis was based on self-reported physician diagnosis rather than clinical examination or imaging confirmation, which may lead to a certain degree of misclassification. Thirdly, although we adjusted for many known confounding factors, the possibility of residual confounding cannot be ruled out. For example, we lack information on joint injury history, occupation-related risk factors, and medication use (such as glucocorticoids), which might influence OA risk. Finally, we did not distinguish between OA in different joint sites (such as knee, hip, or hand OA), while OA at different sites might be affected by metabolic factors to varying degrees.

In conclusion, this study demonstrates that METS-VF and BRI are independently associated with OA in patients with diabetes or prediabetes, with both showing significant dose-response relationships. These findings emphasize the important roles of metabolic health and body shape distribution in their relationship with OA among patients with diabetes/prediabetes, providing new biomarkers for clinical assessment and potential targets for intervention. Future research should explore the associations of these indicators across different OA subtypes and validate their potential predictive value through prospective study designs.