All protocols were conducted in compliance with the ethical guidelines of the NHANES Institutional Review Board and the 1975 Declaration of Helsinki (revised 2000). Approval from the Institutional Review Board was secured from the National Center for Health Statistics Research Ethics Review Board. Written informed consent was obtained from all individual participants included in the study.

We used data from NHANES, a continuous, cross-sectional, and multistage probability study aimed at evaluating the health and nutritional conditions of the civilian, non-institutionalised demographic in the United States. Comprehensive protocols, alongside publicly accessible datasets, can be found at (http://www.cdc.gov/nchs/nhanes.htm). We analysed data from the 2005-2008 NHANES cycles. Among the 20,497 examined individuals, the following were sequentially excluded: pregnant individuals, those lacking data on LS BMD and the TBS or compatible dual-site data, and individuals for whom the basic PREVENT equation could not yield a valid 10-year CVD risk. Additionally, participants with missing waist circumference data were excluded. Consequently, the final study population consisted of 4,364 participants aged 30-79 years. The flowchart detailing the inclusion and exclusion of participants is shown in Figure 1.

The CKM syndrome stages were categorised in accordance with the guidelines presented in the 2023 American Heart Association Presidential Advisory^[4,5]. The overall health status of participants was divided into five distinct stages (0-4) based on the presence of specific risk factors and clinical conditions. Stage 0 was identified in individuals whose BMI ranged from 18.5 to 24.9 kg/m², with a waist circumference < 88 cm for female individuals and < 102 cm for male individuals, and those who did not meet the criteria for any higher stage. Stage 1 was identified in those who either had (a) a BMI ≥ 25 kg/m² or an increased waist circumference (≥ 88 cm for female individuals; ≥ 102 cm for male individuals); or (b) were diagnosed with prediabetes or were undergoing pharmacological treatment for hyperglycemia. Stage 2 was identified in individuals exhibiting at least one of the following conditions: (a) metabolic risk factors, such as fasting serum triglyceride level ≥ 135 mg/dL, hypertension, diabetes, or metabolic syndrome; (b) CKD, classified as moderate or high risk according to Kidney Disease: Improving Global Outcomes criteria^[11] [with an estimated glomerular filtration rate (eGFR) of 30-59 mL/min/1.73 m² or a urine albumin-to-creatinine ratio (UACR) of 30-299 mg/g], calculated using the race-free Chronic Kidney Disease Epidemiology Collaboration 2021 creatinine equation^[12]. Stage 3 was characterised by (a) very-high-risk CKD (with an eGFR < 30 mL/min/1.73 m² or a UACR ≥ 300 mg/g); or (b) a projected 10-year risk of atherosclerotic CVD ≥ 20%; utilising a condensed CKM risk protocol, we calculated the 10-year likelihood of CVD (the detailed formula is provided in Supplementary Table 1). Stage 4 was defined by the presence of self-reported, physician-diagnosed conditions, such as coronary heart disease, angina, myocardial infarction, heart failure, or stroke. For CKM syndrome staging, a hierarchical prioritization rule was applied: individuals meeting clinical criteria for multiple stages were consistently classified into the highest applicable stage to reflect maximum severity.

Bone quality and mass were assessed using three parameters, namely the TBS, FN BMD, and LS BMD. The TBS, a metric obtained from the fluctuations in grey-levels observed in DXA scans of the LS, was computed utilising TBS software (Med-Imap SA TBS Calculator, version 2.1.0.2). The FN BMD and LS BMD were assessed through DXA using Hologic QDR-4500A fan-beam densitometers. Additionally, binary bone outcomes were derived from the NHANES questionnaire: osteoporosis was operationalized as a self-reported physician diagnosis (item OSQ060: “Ever told had osteoporosis or brittle bones”), and prior fractures were identified based on affirmative responses to doctor-diagnosed fractures.

Covariates were selected based on biological plausibility and established associations with both CKM syndrome and bone health outcomes, encompassing demographic factors, biochemical markers, lifestyle and nutritional factors, and comorbidities/medication use. Additional covariates included age, sex, race, poverty-income ratio, marital status, education level, total calcium intake, vitamin D supplementation, physical activity level, alcohol consumption, history of rheumatoid arthritis, history of cancer or malignancy, history of glucocorticoid use, history of female hormone use, and current anti-osteoporosis drug use, as well as levels of biochemical markers such as serum 25-hydroxyvitamin D, serum phosphorus, serum calcium, serum alkaline phosphatase, and serum uric acid. Total calcium intake includes both dietary sources and supplements; Alcohol consumption was classified into four categories using NHANES standardized definitions: Never (lifetime intake < 12 drinks), Former (≥ 12 drinks in lifetime but none in the past 12 months), Current moderate (≤ 1 drink/day for women or ≤ 2 drinks/day for men, without binge drinking), and Current heavy (> 1 drink/day for women or > 2 drinks/day for men, or presence of binge drinking). Physical activity level was categorized using metabolic equivalent of task (MET-min/week) into four groups: No activity (0 MET-min/week), Insufficiently active (1-599), Sufficiently active (600-2,999), and Highly active (≥ 3,000). Current anti-osteoporosis medication use included bisphosphonates, selective estrogen receptor modulators, and denosumab.

Statistical analyses were conducted using EmpowerStats v4.2, EmpowerXYS 6.0, R v4.3.2, and Python. Continuous variables were presented as means with corresponding 95% confidence intervals (95%CI) or as medians with interquartile ranges, whereas categorical variables were expressed as absolute counts and percentages. Statistical significance was assessed using the Kruskal-Wallis rank-sum test for continuous variables and Fisher’s exact probability test for categorical variables with expected values less than 10. A significance threshold was set at P < 0.05. Three analytical models were developed: model 1 served as the unadjusted baseline; model 2 incorporated adjustments for age, sex, and race/ethnicity; and model 3 included comprehensive adjustments for all relevant covariates, including age, sex, race, poverty-income ratio, marital status, and education level, serum 25-hydroxyvitamin D, serum phosphorus, serum calcium, serum alkaline phosphatase, and uric acid, total calcium intake, vitamin D supplementation, alcohol consumption, physical activity level, history of glucocorticoid use, female hormone use, rheumatoid arthritis, cancer or malignancy and current anti-osteoporosis drug use. Adjusted marginal means of TBS, FN BMD, and LS BMD across CKM stages were estimated from Model 3 and visualized using dual-axis line graphs. Furthermore, subgroup analyses were conducted to evaluate the association between the CKM syndrome stage and bone health-related indicators (TBS, FN BMD, and LS BMD) across demographic and clinical subgroups. In these models, all covariates included in Model 3 were adjusted for, with the exception of the stratifying variable itself. Interaction terms were incorporated to evaluate the potential modifying effects of each subgroup variable on these associations. Additionally, the proportional odds assumption was tested using ordinal regression models. Piecewise linear regression with a breakpoint between CKM stages 2 and 3 was performed to examine differential associations between early (stages 0-2) and late (stages 3-4) CKM status and bone outcomes. This breakpoint was established a priori based on clinical relevance. It represents the fundamental pathophysiological watershed in the CKM taxonomy - transitioning from simple metabolic risk factors to overt target organ damage - allowing us to specifically evaluate bone parameters across this critical clinical threshold.

Among the 4,364 eligible participants, the proportions of those with CKM stages 0-4 were 12.3%, 33.5%, 41.0%, 4.8%, and 8.3%, respectively. Age increased sequentially from the group of participants with stage 0 to those with stage 3 (median ages: 42, 45, 52, and 74 years, respectively; P < 0.001), declined slightly in the group with stage 4 (64 years), and remained significantly higher in those with stages 0-2 (all P < 0.001). A male predominance was observed among the participants with advanced stages (P < 0.001); the proportion of male individuals increased from the group with stage 0 (40.90%) to those with stage 3 (63.51%), and declined slightly in the group with stage 4 (63.09%). Race differed across the groups with different CKM syndrome stages (P < 0.001). The Non-Hispanic White population remained the largest subgroup despite their proportion decreasing from 60.22% to 53.72%, whereas the proportion of non-Hispanic Blacks increased from 10.78% to 23.14%. Socioeconomic indicators deteriorated in more advanced CKM syndrome stages. The percentage of individuals who completed high school decreased from 79.93% to 63.91%, and the proportion with a poverty-income ratio > 3 declined from 55.21% to 30.38% (both P < 0.001). Consistent with higher CKM stages, serum 25-hydroxyvitamin D levels decreased (P < 0.001), whereas alkaline phosphatase and uric acid levels increased (both P < 0.001). Calcium levels showed no clinically relevant difference (P = 0.053), whereas phosphorus levels varied significantly but minimally (P < 0.001) based on CKM syndrome stage. Additionally, total calcium intake generally decreased at more advanced stages (P < 0.001). Alcohol consumption patterns (P < 0.001) and physical activity levels (P < 0.001) varied significantly across the CKM syndrome stages, with a higher proportion of physical inactivity observed in advanced stages. However, history of glucocorticoid use, rheumatoid arthritis, and cancer or malignancy was more prevalent among individuals with advanced CKM syndrome stages (all P < 0.001). In contrast, vitamin D supplementation (P = 0.178) and its dosage among users (P = 0.645), as well as current anti-osteoporosis drug use (P = 0.144), did not differ significantly across the groups [Table 1].

Multivariate linear regression analysis indicated that, in model 3, the CKM syndrome stage was inversely associated with the TBS. Compared with stage 0, the β coefficients (95%CI) for stages 1-4 were -0.03 (-0.05, -0.02), -0.06 (-0.07, -0.05), -0.05 (-0.07, -0.03), and -0.07 (-0.08, -0.05), respectively (all P < 0.001). In contrast, the CKM syndrome stage exhibited a positive and linear correlation with FN BMD and LS BMD. The β coefficients (95%CI) for stages 1-4 respectively were 0.07 (0.05, 0.08), 0.07 (0.06, 0.09), 0.08 (0.06, 0.10), and 0.07 (0.05, 0.09) for FN BMD and 0.05 (0.03, 0.06), 0.05 (0.04, 0.07), 0.09 (0.06, 0.11), and 0.08 (0.05, 0.10) for LS BMD (all P < 0.001).

Multivariate logistic regression analysis revealed that the CKM syndrome stage was not significantly associated with osteoporosis or prior fractures in the fully adjusted model (Model 3) [Table 2].

To further illustrate the divergent trajectories of bone quality and bone mass across the CKM syndrome stages, we calculated the adjusted marginal means of the TBS, FN BMD, and LS BMD based on the fully adjusted model (Model 3). As depicted in Figure 2 and Supplementary Table 2 of the additional file, a clear dissociation was observed. The adjusted marginal mean of the TBS demonstrated an overall downward trend, decreasing from 1.413 (95%CI: 1.402-1.425) in stage 0 to its lowest point of 1.348 (95%CI: 1.335-1.361) in stage 4. Conversely, the adjusted marginal means of both FN BMD and LS BMD exhibited an initial significant increase from stage 0 to stage 1, followed by a plateau or slight upward trajectory through the more advanced stages. These visual trends robustly corroborate the inverse association of advanced CKM syndrome with microarchitectural bone quality, despite the paradoxical preservation or increase in areal bone density.

To evaluate non-linear trends and potential threshold effects in the adjusted marginal means, a segmented regression analysis was performed with a predefined breakpoint between CKM syndrome stages 2 and 3. For the TBS, a significant and sharp decline was observed during the early-to-intermediate stages (stages 0-2), with a slope of -0.0288 (95%CI: -0.0347 to -0.0229; P < 0.0001). However, as the syndrome advanced to stages 3-4, the trajectory plateaued, and the slope was no longer significantly different from zero (slope = -0.0011; P = 0.7566). The change in slope (Δ slope) was statistically significant (0.0277; P < 0.0001). Regarding areal BMD, FN BMD exhibited a corresponding threshold effect, with its significant increase confined to the early stages before plateauing (Δ slope = -0.0344; P < 0.0001). In contrast, LS BMD demonstrated a more consistent and linear increase across all CKM syndrome stages, with no statistically significant change in slope (Δ slope = -0.0092; P = 0.1457) [Table 3 and Supplementary Figure 1].

Subgroup analyses consistently showed an inverse relationship between CKM syndrome stage and TBS. Interaction tests revealed no significant differences based on age, poverty-income ratio, education level, marital status, menopausal history, physical activity level, alcohol consumption, or history of glucocorticoid use, rheumatoid arthritis, or cancer or malignancy (all P-interaction > 0.05). However, the inverse association between TBS and higher CKM syndrome stages was significantly greater in male individuals than in female individuals (P-interaction = 0.0397). Additionally, significant interactions were observed for race (P-interaction = 0.0333) and current anti-osteoporosis drug use (P-interaction = 0.0010). The inverse association between CKM syndrome stage and TBS was prominent among non-users of anti-osteoporosis drugs but attenuated in users. For FN BMD and LS BMD, the positive associations with the CKM syndrome stage remained significant and directionally consistent across subgroups. Interaction tests showed no significant effect modification by sex, menopausal history, age, poverty-income ratio, education level, marital status, alcohol consumption, physical activity level, or history of glucocorticoid use, rheumatoid arthritis, cancer or malignancy (all P-interaction ≥ 0.05), with the exception of race for FN BMD (P-interaction = 0.0298). Although slight heterogeneity was observed in the race subgroups for FN BMD, the point estimates consistently demonstrated significant positive associations, with no evidence of trend reversal [Supplementary Tables 3-5 and Figures 3-5].

Several limitations of this study should be acknowledged. First, the cross-sectional design precludes the establishment of causal relationships between CKM syndrome stage and bone health outcomes. The observed associations should be interpreted as correlational rather than causal, and reverse causation cannot be entirely excluded. Second, the relatively low prevalence of osteoporosis and prior fractures in our study population limited the statistical power to detect significant associations with CKM syndrome stages. Specifically, the inclusion of a substantial proportion of younger adults (median age 51 years) naturally dilutes the incidence of these endpoint events. Furthermore, data regarding prior fractures were predominantly derived from self-reported questionnaires, which inevitably introduces recall and information bias, potentially further masking true clinical associations. Third, the absence of bone turnover markers, such as procollagen type 1 N-terminal propeptide and β-isomerised C-terminal telopeptide of type 1 collagen, restricted our ability to assess the dynamic processes of bone remodelling and to elucidate the underlying mechanisms linking CKM syndrome to compromised bone health. Fourth, the lack of longitudinal CKM trajectory data prevented us from evaluating how changes in CKM stage over time affect bone health, which would be more informative for clinical management. Fifth, the NHANES dataset does not include detailed information on certain potential confounders, such as sun exposure, specific dietary patterns, or family history of osteoporosis, which may have contributed to residual confounding despite our comprehensive adjustments. Sixth, the use of DXA-derived BMD may be influenced by degenerative spinal changes, aortic calcification, and abdominal adiposity, which are more prevalent in advanced CKM stages and may artificially elevate LS BMD measurements. Finally, while we observed a significant interaction between CKM stage and anti-osteoporosis medication use, the small number of medication users in our sample warrants cautious interpretation. Future multicentre prospective cohort studies integrating bone metabolic markers, advanced bone geometry assessment, and longer follow-up are needed to confirm our findings and to elucidate the causal mechanisms underlying the observed paradox.

In conclusion, this study is the first to reveal a paradoxical relationship in bone health associated with CKM syndrome, where advanced stages are associated with progressive decline in the TBS despite increases in the FN BMD and LS BMD. This dissociation, similar to the well-established bone paradox in T2DM, underscores that relying solely on BMD is insufficient for assessing fracture risk in this population. A comprehensive evaluation should incorporate the TBS to capture the full spectrum of bone fragility. For patients with early-stage CKM, particularly those with high fracture risk, we recommend concurrent TBS assessment during routine BMD testing to enable early detection of microarchitectural bone deficits and refine fracture risk stratification. Longitudinal studies are required to establish causality and inform personalised bone-health interventions for patients with CKM syndrome. Additional research is necessary to explore the pathophysiological mechanisms that contribute to these associations.