Abstract
Abdominal aortic calcification (AAC) serves as a reliable predictor of future cardiovascular incidents. This study investigated the association between relative fat mass (RFM) and AAC in US adults. A total of 3079 adults aged ≥40 years from the National Health and Nutrition Examination Survey 2013 to 2014 were included in the cross-sectional study. AAC was evaluated using a semi-quantitative scoring system known as AAC-24, with an AAC score >6 considered severe AAC (SAAC). Logistic regression analyses and restricted cubic splines were applied. Subgroup analyses were also conducted. After adjusting for confounding factors, a clear link was established between RFM and AAC score (β = −0.048, 95% CI: −0.078 to −0.017, P < .01) and between RFM and SAAC (OR = 0.96, 95% CI: 0.92 to 1.00, P < .05). Further restricted cubic spline analysis indicated a negative correlation between RFM and both AAC score and SAAC. In this cross-sectional study, an inverse association was observed between RFM and both AAC score and severe AAC, with a threshold of approximately 48 to 50. Below this threshold, lower RFM was associated with a sharply higher probability of calcification; above it, higher RFM corresponded to a gradually lower probability. Due to the cross-sectional design, causality cannot be inferred. Further prospective cohort studies are needed to validate these findings.
Keywords: abdominal aortic calcification, cross-sectional study, NHANES, relative fat mass
1. Introduction
Vascular calcification (VC), characterized by the ectopic deposition of calcium phosphate crystals in the arterial wall, is a significant pathological process associated with adverse cardiovascular outcomes.[1,2] Atherosclerotic plaques already develop in childhood, and their extent increases with age. Although coronary artery calcification has been extensively studied, atherosclerosis develops at a younger age in the aorta.[3] Notably, epidemiological studies have shown that abdominal aortic calcification (AAC) is associated with cardiovascular disease and cardiovascular mortality,[4,5] myocardial infarction,[6] stroke, and incident coronary heart disease.[7] Despite its clinical significance, no effective interventions currently exist to treat AAC. Therefore, identifying accessible and robust predictors for AAC risk stratification is of paramount importance. Early identification of high-risk individuals could enable timely preventive strategies, yet simple and reliable clinical indicators remain lacking.
Multiple studies have confirmed that obesity serves as an independent determinant for various diseases, including type 2 diabetes, cardiovascular diseases, and inflammatory bowel diseases.[8–10] The body mass index (BMI) is frequently used as a metric to quantify obesity levels. However, recent studies have emphasized the limitations of BMI in distinguishing muscle from adipose tissue distribution, potentially leading to significant inaccuracies in the estimation of body fat content.[11] Given the inherent limitations of BMI, researchers have identified an innovative metric, relative fat mass (RFM), which demonstrates superior predictive efficacy for estimating total body fat percentage in both sexes and significantly reduces misclassification rates in obesity assessment.[12] RFM is calculated using the formula: 64 − (20 × height/waist circumference) + 12 × sex coefficient, where the sex coefficient is assigned as 0 for males and 1 for females.[13] Nevertheless, evidence regarding the association between RFM and AAC remains limited.
Despite its advantages over traditional metrics, the relationship between RFM and AAC remains largely unexplored. Given that both adiposity distribution and vascular calcification (VC) share common pathophysiological pathways – including inflammation, oxidative stress, and metabolic dysregulation – RFM may offer unique insights into AAC risk beyond conventional measures. We therefore hypothesized that RFM is inversely associated with AAC, and that this association may follow a nonlinear pattern with a specific threshold.
Accordingly, this study aimed to investigate the association between RFM and AAC using data from the National Health and Nutrition Examination Survey (NHANES), with the goal of providing a novel, easily accessible indicator for AAC risk prediction from the perspective of body composition assessment.
2. Methods
2.1. Survey description
NHANES is conducted by the National Center for Health Statistics and aims to construct nationally representative health data of the American population. The survey employs a complex, multi-stage, stratified probability sampling and cross-sectional design to collect data. Information from approximately 5000 participants is gathered annually. The survey began sampling in 1999 and releases biennial data. The design of NHANES research has been approved by the National Center for Health Statistics Research Ethics Review Board, and the study has obtained informed consent from all participants. NHANES survey data, detailed survey operation manuals, informed consent forms, and manuals for each sampling cycle are publicly available on the NHANES website.
2.2. Study population
The survey spanned a 2-year period (2013–2014) within a single survey cycle. Eligibility criteria for participants included individuals aged 40 years or older, who had no reported radiation exposure in the previous 7 days. Seven thousand thirty-five participants were excluded due to missing AAC data, and an additional 61 participants were excluded due to missing RFM data. Consequently, 3079 participants remained and were included in the analysis (Fig. 1).
Figure 1.
Flow chart of participant selection. NHANES = National Health and Nutrition Examination Survey.
2.3. Assessment of RFM
The exposure variable in this study was RFM, which was used to assess body fat content. This metric incorporates waist circumference, height, and gender into its calculation. RFM was calculated using the formula: 64 − (20 × height (cm)/waist circumference (cm)) + (12 × sex), where sex is assigned a value of 1 for females and 0 for males. At the Mobile Examination Center, healthcare professionals use specialized instruments to measure participants’ height, weight, and waist circumference, with waist circumference measured to an accuracy of 0.1 cm.
2.4. AAC
The severity of AAC was assessed using the Kauppila scoring system via dual-energy X-ray absorptiometry. The AAC score quantifies the extent of calcification, with higher scores indicating more severe calcification. In this assessment, the abdominal aortic wall was assigned a score ranging from 0 to 6 based on the degree of calcium deposition, resulting in a cumulative AAC total score ranging from 0 to 24. An AAC score >6 was typically defined as severe AAC.
2.5. Covariates
To reduce potential confounding bias in the analysis, we selected several covariates based on previous relevant studies and clinical significance. These covariates included gender, age, race, education level, poverty-to-income ratio, smoking status, drinking status, creatinine, uric acid, calcium, phosphorus, vitamin D, diabetes, coronary heart disease, marital status, physical activity level, total energy intake, and hypertension. Race was categorized into 5 groups: Mexican American, other Hispanic, non-Hispanic White, non-Hispanic Black, and other race. Education level was stratified into 3 categories: below high school, completed high school, and above high school. Smoking status was defined as never, former, or current. Marital status was divided into 2 groups: married/living with partner, and widowed/divorced/separated/never married. Physical activity level was stratified into 4 groups: both moderate and vigorous, inactive, moderate only, and vigorous only. Drinking status, diabetes, coronary heart disease, and hypertension were all classified into 2 groups: yes or no.
2.6. Statistical analysis
All analyses were conducted in accordance with NHANES analytic guidelines, incorporating appropriate survey weights, stratification, and primary sampling unit variables. Specifically, the dietary 2-day sample weight WTDR2D was used as the probability weight, SDMVSTRA as the stratum variable, and SDMVPSU as the primary sampling unit. First, the baseline characteristics of the final participants were presented according to RFM levels in the descriptive analysis. Categorical variables were reported as frequencies (percentages), while continuous variables were reported as mean (standard error). The comparison of continuous and categorical variables across groups was performed using the t test and chi-squared test, respectively. For the primary exposure variable (RFM) and outcome variable (AAC), we applied listwise deletion, retaining only participants with complete data for both variables. For missing values in other covariates, we used multiple imputation. Second, weighted multivariate linear regression was performed with AAC scores as a continuous variable, along with weighted logistic regression using severe AAC as a dichotomous variable, in order to examine the relationship between RFM and AAC. Model 1 was unadjusted. Model 2 adjusted for gender, age, and race. Model 3 further adjusted for education level, poverty-to-income ratio, smoking status, drinking status, creatinine, uric acid, calcium, phosphorus, vitamin D, diabetes, coronary heart disease, marital status, physical activity, total energy intake, and hypertension on the basis of model 2. For covariate selection, we conducted a collinearity analysis and included only variables with a variance inflation factor <5. In addition, the potential nonlinear relationship between RFM and AAC score as well as severe abdominal aortic calcification (SAAC) was explored using restricted cubic splines. To validate the RFM threshold identified by RCS analysis for its association with AAC, we performed piecewise regression analysis using the “segmented” package in R. A piecewise linear regression was fitted for the continuous outcome (AAC score), and a piecewise logistic regression for the binary outcome (SAAC), both with the RCS-derived breakpoint pre-specified. Davies’ test was used to test the threshold effect, and the stability of the breakpoint was assessed by profile likelihood confidence intervals. All models were adjusted for the same set of covariates (age, gender, etc). Subgroup analyses were conducted to examine the possible influence of stratification variables on the associations of RFM with AAC and with SAAC. All statistical analyses were performed using R software (version 4.4.2). A 2-sided P value < .05 was considered statistically significant.
3. Results
3.1. Characteristics of participants
Table 1 compares the baseline demographic, lifestyle, clinical, and laboratory characteristics between the included and excluded participants in the NHANES 2013 to 2014 dataset. A total of 10,175 participants were initially eligible for this study, of whom 3079 were included in the final analysis, and 7096 were excluded due to missing data on key variables. No significant difference in gender distribution was observed between the included and excluded groups (P = .426). However, included participants were significantly older and exhibited notable differences in race composition, educational background, and socioeconomic status. Differences in lifestyle habits, serum biochemical indicators, and chronic disease prevalence were also identified between the 2 groups.
Table 1.
Baseline characteristics of the included and excluded groups in NHANES 2013 to 2014.
| [ALL] | Included group | Excluded group | P overall | |
|---|---|---|---|---|
| N = 10175 | N = 3079 | N = 7096 | ||
| Gender | .426 | |||
| Male | 5003 (49.2%) | 1495 (48.6%) | 3508 (49.4%) | |
| Female | 5172 (50.8%) | 1584 (51.4%) | 3588 (50.6%) | |
| Age | 31.5 (24.4) | 58.6 (12.0) | 19.7 (18.3) | .000 |
| Race | <.001 | |||
| Mexican American | 1730 (17.0%) | 403 (13.1%) | 1327 (18.7%) | |
| Other Hispanic | 960 (9.43%) | 295 (9.58%) | 665 (9.37%) | |
| Non-Hispanic | 3674 (36.1%) | 1347 (43.7%) | 2327 (32.8%) | |
| Non-Hispanic Black | 2267 (22.3%) | 611 (19.8%) | 1656 (23.3%) | |
| Other Race | 1544 (15.2%) | 423 (13.7%) | 1121 (15.8%) | |
| Education level: | <.001 | |||
| <High school | 455 (17.9%) | 291 (20.8%) | 164 (14.2%) | |
| High school | 791 (31.0%) | 419 (30.0%) | 372 (32.3%) | |
| >High school | 1303 (51.1%) | 686 (49.1%) | 617 (53.5%) | |
| PIR | 2.25 (1.63) | 2.69 (1.65) | 2.06 (1.59) | <.001 |
| Smoking status: | <.001 | |||
| Never | 3532 (57.8%) | 1653 (53.7%) | 1879 (62.0%) | |
| Former | 1347 (22.0%) | 855 (27.8%) | 492 (16.2%) | |
| Current | 1232 (20.2%) | 570 (18.5%) | 662 (21.8%) | |
| Drinking status: | .008 | |||
| Yes | 3790 (70.0%) | 2071 (71.6%) | 1719 (68.2%) | |
| No | 1623 (30.0%) | 822 (28.4%) | 801 (31.8%) | |
| Serum creatinine (umol/L) | 77.8 (43.1) | 83.4 (46.5) | 73.2 (39.4) | <.001 |
| Serum uric acid (umol/L) | 318 (83.5) | 324 (82.7) | 313 (83.8) | <.001 |
| Serum calcium (mg/dL) | 9.49 (0.37) | 9.45 (0.37) | 9.51 (0.37) | <.001 |
| Serum hosphorus (mg/dL) | 3.93 (0.65) | 3.79 (0.57) | 4.04 (0.69) | <.001 |
| Vitamin D (nmol/L) | 64.6 (25.6) | 70.5 (29.5) | 61.4 (22.5) | <.001 |
| Diabetes: | <.001 | |||
| Yes | 984 (9.67%) | 681 (22.1%) | 303 (4.27%) | |
| No | 9191 (90.3%) | 2398 (77.9%) | 6793 (95.7%) | |
| Coronary heart disease: | <.001 | |||
| Yes | 232 (4.03%) | 161 (5.24%) | 1 (2.65%) | |
| No | 5519 (96.0%) | 2910 (94.8%) | 2609 (97.4%) | |
| Marital status: | <.001 | |||
| Married/living with partner | 3382 (58.7%) | 1966 (63.9%) | 1416 (52.7%) | |
| Widowed/divorced/separated/never married | 2384 (41.3%) | 1112 (36.1%) | 1272 (47.3%) | |
| Physical activity level: | <.001 | |||
| Inactive | 6844 (72.2%) | 2001 (65.0%) | 4843 (75.6%) | |
| Moderate | 1468 (15.5%) | 568 (18.4%) | 900 (14.1%) | |
| Vigorous | 332 (3.50%) | 146 (4.74%) | 186 (2.90%) | |
| Both moderate and vigorous | 840 (8.86%) | 364 (11.8%) | 476 (7.43%) | |
| Total energy intake (Kcal) | 1904 (804) | 1962 (773) | 1873 (818) | <.001 |
| Hypertension: | <.001 | |||
| Yes | 2174 (33.7%) | 1449 (47.1%) | 725 (21.4%) | |
| No | 4285 (66.3%) | 1627 (52.9%) | 2658 (78.6%) |
The baseline characteristics of the final enrolled 3079 participants were further analyzed based on RFM quartile stratification (Table 2). The mean age of all participants was 58.6 ± 12.0 years. The mean AAC score for the overall population was 1.62 ± 3.48, and 277 participants (9.00%) were diagnosed with severe AAC. The mean RFM was 48.08 ± 8.10, with quartile ranges categorized as follows: Q1 (≤41.62), Q2 (41.62–47.31), Q3 (47.31–55.23), Q4 (≥55.23). The analysis revealed that those in the highest RFM quartile tended to be older, predominantly female, and exhibited a higher prevalence of hypertension, lower engagement in physical activity, and a higher proportion of nonsmokers. They also showed elevated serum phosphorus levels and higher total energy intake. Conversely, this group had a lower prevalence of diabetes and coronary heart disease, as well as lower educational attainment, lower proportions of married individuals and alcohol consumers, and lower socioeconomic status. Additionally, these participants demonstrated lower levels of serum creatinine, uric acid, and vitamin D.
Table 2.
Baseline characteristics of participants from NHANES 2013 to 2014 grouped by RFM levels.
| [ALL] | Q1 | Q2 | Q3 | Q4 | P overall | |
|---|---|---|---|---|---|---|
| N = 3079 | N = 769 | N = 770 | N = 770 | N = 770 | ||
| Gender | .000 | |||||
| Female | 1584 (51.4%) | 14 (1.82%) | 134 (17.4%) | 666 (86.5%) | 770 (100%) | |
| Male | 1495 (48.6%) | 755 (98.2%) | 636 (82.6%) | 104 (13.5%) | 0 (0.00%) | |
| Age | 58.6 (12.0) | 56.8 (11.6) | 59.3 (12.0) | 58.7 (12.3) | 59.5 (11.7) | <.001 |
| Race | <.001 | |||||
| Mexican American | 403 (13.1%) | 74 (9.62%) | 117 (15.2%) | 88 (11.4%) | 124 (16.1%) | |
| Non-Hispanic Black | 611 (19.8%) | 179 (23.3%) | 134 (17.4%) | 141 (18.3%) | 157 (20.4%) | |
| Non-Hispanic White | 1347 (43.7%) | 295 (38.4%) | 371 (48.2%) | 341 (44.3%) | 340 (44.2%) | |
| Other Hispanic | 295 (9.58%) | 67 (8.71%) | 69 (8.96%) | 72 (9.35%) | 87 (11.3%) | |
| Other race | 423 (13.7%) | 154 (20.0%) | 79 (10.3%) | 128 (16.6%) | 62 (8.05%) | |
| Education level: | 0.003 | |||||
| <High school | 710 (23.1%) | 170 (22.1%) | 189 (24.5%) | 154 (20.0%) | 197 (25.6%) | |
| >High school | 1681 (54.6%) | 436 (56.8%) | 404 (52.5%) | 460 (59.8%) | 381 (49.5%) | |
| High school | 686 (22.3%) | 162 (21.1%) | 177 (23.0%) | 155 (20.2%) | 192 (24.9%) | |
| PIR | 2.69 (1.65) | 2.88 (1.70) | 2.75 (1.67) | 2.79 (1.66) | 2.35 (1.53) | <.001 |
| Smoking status | <.001 | |||||
| Never | 1653 (53.7%) | 360 (46.9%) | 363 (47.1%) | 471 (61.2%) | 459 (59.6%) | |
| Former | 855 (27.8%) | 221 (28.8%) | 277 (36.0%) | 178 (23.1%) | 179 (23.2%) | |
| Current | 570 (18.5%) | 187 (24.3%) | 130 (16.9%) | 121 (15.7%) | 132 (17.1%) | |
| Drinking status | <.001 | |||||
| No | 822 (28.4%) | 109 (15.5%) | 139 (19.1%) | 249 (34.1%) | 325 (44.5%) | |
| Yes | 2071 (71.6%) | 596 (84.5%) | 587 (80.9%) | 482 (65.9%) | 406 (55.5%) | |
| Serum creatinine (µmol/L) | 83.4 (46.5) | 93.0 (43.8) | 92.2 (64.8) | 74.7 (32.8) | 73.9 (33.5) | <.001 |
| Serum uric acid (µmol/L) | 324 (82.7) | 340 (75.4) | 344 (85.7) | 295 (83.0) | 318 (77.4) | <.001 |
| Serum calcium (mg/dL) | 9.45 (0.37) | 9.46 (0.35) | 9.43 (0.34) | 9.46 (0.36) | 9.46 (0.41) | .364 |
| Serum phosphorus (mg/dL) | 3.79 (0.57) | 3.67 (0.54) | 3.72 (0.61) | 3.95 (0.56) | 3.84 (0.54) | <.001 |
| Vitamin D (nmol/L) | 70.5 (29.5) | 68.2 (26.0) | 68.1 (26.4) | 75.3 (32.4) | 70.2 (32.0) | <.001 |
| Diabetes | <.001 | |||||
| No | 2398 (77.9%) | 648 (84.3%) | 572 (74.3%) | 646 (83.9%) | 532 (69.1%) | |
| Yes | 681 (22.1%) | 121 (15.7%) | 198 (25.7%) | 124 (16.1%) | 238 (30.9%) | |
| Coronary heart disease | .022 | |||||
| Yes | 161 (5.24%) | 33 (4.30%) | 56 (7.29%) | 40 (5.22%) | 32 (4.17%) | |
| No | 2910 (94.8%) | 735 (95.7%) | 712 (92.7%) | 727 (94.8%) | 736 (95.8%) | |
| Marital status | <.001 | |||||
| Married/living with partner | 1966 (63.9%) | 546 (71.0%) | 552 (71.7%) | 469 (60.9%) | 399 (51.9%) | |
| Widowed/divorced/separated/never married | 1112 (36.1%) | 223 (29.0%) | 218 (28.3%) | 301 (39.1%) | 370 (48.1%) | |
| Physical activity level: | <.001 | |||||
| Both moderate and vigorous | 364 (11.8%) | 123 (16.0%) | 108 (14.0%) | 78 (10.1%) | 55 (7.14%) | |
| Inactive | 2001 (65.0%) | 462 (60.1%) | 458 (59.5%) | 521 (67.7%) | 560 (72.7%) | |
| Moderate | 568 (18.4%) | 118 (15.3%) | 154 (20.0%) | 155 (20.1%) | 141 (18.3%) | |
| Vigorous | 146 (4.74%) | 66 (8.58%) | 50 (6.49%) | 16 (2.08%) | 14 (1.82%) | |
| Total energy intake (Kcal) | 1962 (773) | 2289 (859) | 2090 (763) | 1809 (688) | 1683 (626) | <.001 |
| Hypertension: | <.001 | |||||
| Yes | 1449 (47.1%) | 269 (35.0%) | 363 (47.1%) | 347 (45.1%) | 470 (61.1%) | |
| No | 1627 (52.9%) | 499 (65.0%) | 407 (52.9%) | 422 (54.9%) | 299 (38.9%) | |
| AAC | 1.62 (3.48) | 1.54 (3.29) | 1.69 (3.54) | 1.53 (3.29) | 1.74 (3.79) | .555 |
| SAAC: | .779 | |||||
| AAC | 2802 (91.0%) | 700 (91.0%) | 705 (91.6%) | 703 (91.3%) | 694 (90.1%) | |
| SAAC | 277 (9.00%) | 69 (8.97%) | 65 (8.44%) | 67 (8.70%) | 76 (9.87%) |
AAC = abdominal aortic calcification; PIR = poverty-to-income ratio; SAAC = severe abdominal aortic calcification.
3.2. Association between RFM and AAC score and SAAC
Table 3 presents the association between RFM and AAC score. The associations were not statistically significant in either the unadjusted or minimally adjusted models. In the fully adjusted model (Model 3), RFM as a continuous variable showed a statistically significant but weak inverse association with AAC score (β = –0.048, 95% CI: –0.078 to –0.017). Given that the AAC score ranges from 0 to 24, this effect size is clinically modest. When RFM was analyzed by quartiles, a decreasing trend in AAC score was observed across higher RFM quartiles (Q4 vs Q1: β = –0.773, 95% CI: –1.157 to –0.239; P for trend < .01). However, the absolute difference between extreme quartiles remains small relative to the full score range.
Table 3.
The linear regression analysis of the association between RFM and AAC score, adjusted for demographic, lifestyle, and clinical covariates among participants.
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| β (95% CI) | P | β (95% CI) | P | β (95% CI) | P | |
| RFM | 0.008 (−0.007 to 0.023) | .30 | −0.017 (−0.042 to 0.007) | .17 | −0.048 (−0.078 to −0.017) | <.01 |
| Q1 | Ref | Ref | Ref | |||
| Q2 | 0.150 (−0.198 to 0.498) | .40 | −0.223 (−0.554 to 0.108) | .19 | −0.334 (−0.753 to −0.016) | .04 |
| Q3 | −0.010 (−0.358 to 0.339) | .96 | −0.491 (−0.981 to −0.001) | .05 | −0.385 (−1.325 to −0.222) | <.01 |
| Q4 | 0.197 (−0.152 to 0.545) | .27 | −0.379 (−0.921 to 0.163) | .17 | −0.773 (−1.517 to −0.239) | <.01 |
| P for trend | .44 | .21 | <.01 |
AAC = abdominal aortic calcification; β = regression coefficient; CI = confidence interval; RFM = relative fat mass.
Table 4 shows the association between RFM and severe AAC. Similarly, no significant associations were found in the unadjusted or minimally adjusted models. In the fully adjusted logistic regression model, each unit increase in RFM was associated with a borderline significant reduction in the odds of SAAC (OR = 0.962, 95% CI: 0.924–1.004; P = .040), with the upper confidence limit just reaching 1.004. Compared with the lowest RFM quartile (Q1), participants in higher quartiles showed a trend toward lower SAAC risk (P for trend = .040). However, the individual OR for Q4 (0.475, 95% CI: 0.212–1.080) was not statistically significant (P = .072), and the confidence interval included 1.0. Therefore, these findings are exploratory and warrant confirmation in future studies.
Table 4.
The logistic regression analysis of the association between RFM and SAAC, adjusted for demographic, lifestyle, and clinical covariates among participants.
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| OR (95% CI) | P | OR (95% CI) | P | OR (95% CI) | P | |
| RFM | 1.012 (0.992–1.019) | .302 | 0.979 (0.954–1.010) | .302 | 0.962 (0.924–1.004) | .040 |
| Q1 | Ref | Ref | Ref | |||
| Q2 | 0.936 (0.664–1.331) | .712 | 0.628 (0.432–0.944) | .024 | 0.544 (0.341–0.855) | .009 |
| Q3 | 0.974 (0.683–1.377) | .851 | 0.551 (0.301–1.008) | 0.055 | 0.534 (0.263–1.102) | .079 |
| Q4 | 1.105 (0.792–1.570) | .547 | 0.635 (0.325–1.245) | 0.194 | 0.475 (0.212–1.080) | .072 |
| P for trend | .515 | .187 | .036 | |||
CI = confidence interval; OR = odds ratio; RFM = relative fat mass; SAAC = severe abdominal aortic calcification.
Furthermore, restricted cubic spline analysis confirmed a nonlinear negative correlation between RFM and both AAC score and severe AAC (Figs. 2 and 3).
Figure 2.
The nonlinear negative correlation between RFM and AAC score among participants from NHANES 2013 to 2014. AAC = abdominal aortic calcification; NHANES = National Health and Nutrition Examination Survey; RFM = relative fat mass.
Figure 3.
The nonlinear negative correlation between RFM and SAAC among participants from NHANES 2013 to 2014. NHANES = National Health and Nutrition Examination Survey; RFM = relative fat mass; SAAC = severe abdominal aortic calcification.
3.3. Threshold validation
Davies’ test confirmed a significant threshold effect for both outcomes; for continuous AAC, P = .040; for SAAC, P = .035. The piecewise linear regression estimated an optimal breakpoint at RFM = 48.344 (95% CI: 46.376–50.313). Below this threshold, each 1-unit increase in RFM was associated with a significant increase in AAC (β = 0.017, 95% CI: 0.000–0.034, P = .048). Above the threshold, the association reversed to a significantly negative relationship (β = −0.451, 95% CI: −0.943 to −0.042, P = .040). The piecewise logistic regression yielded a highly consistent breakpoint at RFM = 48.311 (95% CI: 46.626–49.995). For SAAC, the odds ratio per 1-unit increase in RFM was 1.017 (95% CI: 1.011–1.034, P = .045) below the threshold and decreased significantly to 0.549 (95% CI: 0.242–0.991, P = .035) above the threshold (Table 5).
Table 5.
Threshold and segment-specific effects of RFM on AAC and SAAC.
| Outcome | Breakpoint (95% CI) | Below threshold | P | Above threshold | P |
|---|---|---|---|---|---|
| AAC | 48.344 (46.376–50.313) | β = 0.017 (0.000–0.034) | .048 | β = −0.451 (−0.943 to 0.042) | .040 |
| SAAC | 48.311 (46.626–49.995) | OR = 1..017 (1.011–1.034) | .045 | OR = 0..549 (0.242 to 0.991) | .035 |
AAC = abdominal aortic calcification; CI = confidence interval; OR = odds ratio; SAAC = severe abdominal aortic calcification.
These consistent findings across 2 outcome definitions support the robustness of the threshold effect. We further illustrated this relationship in the piecewise linear regression plot (Figs. 4 and 5).
Figure 4.
Piecewise linear regression of RFM versus AAC score, NHANES 2013 to 2014. The estimated breakpoint was at RFM = 48.344. The blue line shows predicted AAC scores, gray dots represent individual data, and the red dashed line marks the breakpoint. AAC = abdominal aortic calcification; NHANES = National Health and Nutrition Examination Survey; RFM = relative fat mass.
Figure 5.
Piecewise logistic regression of RFM versus the predicted probability of SAAC, NHANES 2013 to 2014. The estimated breakpoint was at RFM = 48.311. The blue line shows the predicted probability of SAAC, and the red dashed line marks the breakpoint. NHANES = National Health and Nutrition Examination Survey; RFM = relative fat mass; SAAC = severe abdominal aortic calcification.
3.4. Subgroup analyses
Figures 6 and 7 present the results of our investigation into the stability of the relationship between RFM and AAC. To examine this relationship, the data were categorized by age, gender, race, hypertension, diabetes, coronary heart disease, smoking status, and drinking status. After conducting subgroup analyses, no significant interactions were observed between RFM and these categorized variables (all P for interaction > .05), indicating that the relationship remained stable. Notably, as shown in Figure 7, a more pronounced beneficial trend associated with higher RFM levels was observed among participants who were younger than 60 years, of non-Hispanic White ethnicity, and had never smoked.
Figure 6.
Forest plot of subgroup analyses for the association between RFM and AAC score among participants from NHANES 2013 to 2014. AAC = abdominal aortic calcification; NHANES = National Health and Nutrition Examination Survey; RFM = relative fat mass.
Figure 7.
Forest plot for subgroup analysis of RFM and SAAC among participants from NHANES 2013 to 2014. NHANES = National Health and Nutrition Examination Survey; RFM = relative fat mass; SAAC = severe abdominal aortic calcification.
4. Discussion
To the best of our knowledge, this study may be the first to explore the relationship between RFM and AAC. Our results revealed a nonlinear negative correlation between RFM and both AAC and severe AAC, with a critical threshold identified at approximately 48 to 50. Below this threshold, a decrease in RFM was associated with a sharp increase in calcification risk; above the threshold, an increase in RFM corresponded to a gradual reduction in risk, suggesting a protective role. However, several important caveats must be emphasized. The effect sizes were weak (continuous RFM: β = –0.048 on a 0–24 scale; OR for SAAC per unit increase in RFM = 0.96, 95% CI: 0.92–1.00), and some findings were borderline significant (e.g., the OR for Q4 vs Q1 in SAAC was 0.475, but its 95% CI crossed 1 and the P value was .072). Therefore, these results should be considered exploratory and hypothesis-generating rather than conclusive.
While obesity is a well-established risk factor for many cardiovascular diseases, its specific role in AAC remains unclear, with previous studies reporting conflicting evidence. One study suggested that BMI showed a significant inverse association with AAC score and length after adjustment for potential arteriosclerosis risk factors.[14] Similarly, Ebad ur Rahman et al found that increasing BMI was inversely associated with AAC, and there was no statistically significant association between total body and trunk fat percentages and AAC.[15] However, another study indicated higher ABSI was closely associated with a higher risk of AAC.[16] Recent research has pointed out that a higher cardiometabolic index typically indicates more severe visceral fat accumulation and metabolic dysregulation, and has also demonstrated a positive correlation between cardiometabolic index and AAC.[17,18] In addition, studies showed that WWI was positively associated with AAC scores and exhibited a nonlinear relationship with SAAC.[19,20]
Our study, utilizing the NHANES database, is the first to reveal a nonlinear negative association between RFM and both AAC and severe AAC, which aligns with the findings reported by Jiawei Peng et al. The observed inverse association cannot be interpreted causally, and our study was not designed to test mechanisms. Nevertheless, several biologically plausible pathways may explain why lower RFM (i.e., lower fat mass) could be linked to a higher AAC risk.
The inverse association between low RFM and higher AAC risk can be speculatively explained by several mechanisms. First, adipokine imbalance: low RFM leads to adiponectin deficiency and elevated pro-inflammatory cytokines, triggering chronic inflammation and activating calcification pathways; meanwhile, reduced leptin diminishes vascular repair and protection.[21,22] Second, metabolic–inflammatory dysregulation: low RFM is accompanied by insulin resistance and dyslipidemia, promoting calcification; under low-adiposity conditions, anti-inflammatory capacity declines, exacerbating vascular inflammation and calcification.[23,24] Third, bone–vascular axis disturbance: low RFM depletes vitamin D stores and reduces the OPG/RANKL ratio, disrupting calcium–phosphorus metabolism and promoting VC.[25,26] Fourth, low RFM may reflect underlying frailty or subclinical disease (e.g., chronic inflammation, age-related decline), which themselves directly promote AAC.[27–29]
The nonlinear shape of the association (threshold near 48–50) might reflect a saturation of protective fat functions: below the threshold, fat-related protective mechanisms are deficient, leading to a steep increase in risk; within a moderate range, protection reaches a plateau; and at very high RFM, obesity-related metabolic disturbances could attenuate the protective effect. However, this threshold was not formally validated (e.g., by segmented regression or bootstrap) and therefore requires external validation in independent cohorts.
This study has several strengths. Firstly, by establishing a clinical reference threshold for RFM using restricted cubic spline analysis, we provide a practical tool for identifying individuals at elevated risk for AAC (characterized by RFM below 48–50). This finding also supplies a theoretical rationale for preventive strategies targeting body composition optimization to mitigate VC, underscoring the translational potential of our work. Secondly, this investigation is methodologically novel, representing the first effort to explore the RFM-AAC relationship using the nationally representative NHANES 2013 to 2014 dataset, thereby addressing an important knowledge gap. Additionally, the application of diverse statistical techniques – such as multivariable logistic and linear regression, restricted cubic splines, subgroup analyses, and interaction tests – ensures robust validation of the observed inverse associations, significantly enhancing the credibility of our conclusions.
Several limitations warrant discussion. First, the cross-sectional design precludes any causal inference. We cannot determine whether low RFM increases AAC risk or whether subclinical AAC leads to weight loss. Reverse causality remains possible. Second, a large number of participants (n = 7035, 71.6% of the eligible sample) were excluded due to missing AAC data. Third, the effect sizes were weak (e.g., a 1-unit increase in RFM reduced AAC score by only 0.048 points on a 0–24 scale, and reduced the odds of SAAC by only 4%). Even the difference between the highest and lowest RFM quartiles corresponded to a modest absolute change. Thus, the clinical relevance of these associations is unclear and should not be overemphasized. Fourth, despite adjustment for numerous confounders, residual or unmeasured confounding (e.g., medication use, dietary patterns) cannot be excluded. Fifth, the NHANES sample is limited to the U.S. population; whether our findings can be generalized to other racial/ethnic groups or countries remains unknown.
Prospective cohort studies are needed to establish temporality and examine whether low RFM predicts incident AAC. Mendelian randomization could help clarify causality. External validation of the suggested RFM threshold in diverse populations is essential. Mechanistic studies (in vitro or animal models) are required to test the proposed biological pathways.
In conclusion, this study found a weak, nonlinear, inverse association between RFM and AAC, with a suggestive threshold around 48 to 50. Given the cross-sectional design, modest effect sizes, borderline significance, and other limitations, these findings are exploratory and require confirmation in future studies.
Acknowledgments
The author acknowledges the participants and investigators of the NHANES.
Author contributions
Conceptualization: Lingxiao Fang.
Data curation: Lingxiao Fang.
Formal analysis: Lingxiao Fang.
Funding acquisition: Lingxiao Fang.
Investigation: Lingxiao Fang.
Methodology: Lingxiao Fang.
Project administration: Lingxiao Fang.
Resources: Lingxiao Fang.
Software: Lingxiao Fang.
Supervision: Lingxiao Fang.
Validation: Lingxiao Fang.
Visualization: Lingxiao Fang.
Writing – original draft: Lingxiao Fang.
Writing – review & editing: Lingxiao Fang.
Abbreviations:
- AAC
- abdominal aortic calcification
- BMI
- body mass index
- NHANES
- National Health and Nutrition Examination Survey
- RFM
- relative fat mass
- SAAC
- severe abdominal aortic calcification
- VC
- vascular calcification
The participants provided their written informed consent to participate in this study.
The studies involving humans were approved by NCHS Research Ethics Review Board.
The author has no funding and conflicts of interest to disclose.
The datasets generated during and/or analyzed during the current study are publicly available.
How to cite this article: Fang L. Cross-sectional association between relative fat mass and abdominal aortic calcification among US adults: The NHANES 2013 to 2014. Medicine 2026;105:28(e49727).
All claims expressed in this article are solely those of the author and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors, and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
References
- [1].Dabravolski SA, Popov MA, Maltseva ON, Pavshintsev VV, Utkina AS, Orekhov AN. Exploring macrophage contribution to vascular calcification in atherosclerosis. IUBMB Life. 2025;77:e70064. [DOI] [PubMed] [Google Scholar]
- [2].Bardeesi ASA, Gao J, Zhang K, et al. A novel role of cellular interactions in vascular calcification. J Transl Med. 2017;15:95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Garcia-Garcia HM, Sanz-Sanchez J, Pinilla-Echeverri N, Blanco PJ, Bourantas C, Alfonso F. Advances in coronary imaging of atherosclerotic plaques. EuroIntervention. 2025;21:e778–95. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Hoffmann U, Massaro JM, D’Agostino RB, Sr, Kathiresan S, Fox CS, O’Donnell CJ. Cardiovascular event prediction and risk reclassification by coronary, aortic, and valvular calcification in the Framingham Heart Study. J Am Heart Assoc. 2016;5:e003144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].van der Meer IM, Bots ML, Hofman A, del Sol AI, van der Kuip DA, Witteman JC. Predictive value of noninvasive measures of atherosclerosis for incident myocardial infarction: the Rotterdam Study. Circulation. 2004;109:1089–94. [DOI] [PubMed] [Google Scholar]
- [6].Bolland MJ, Wang TK, van Pelt NC, et al. Abdominal aortic calcification on vertebral morphometry images predicts incident myocardial infarction. J Bone Miner Res. 2010;25:505–12. [DOI] [PubMed] [Google Scholar]
- [7].Chen HC, Wang WT, Hsi CN, et al. Abdominal aortic calcification score can predict future coronary artery disease in hemodialysis patients: a 5-year prospective cohort study. BMC Nephrol. 2018;19:313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Perdomo CM, Cohen RV, Sumithran P, Clément K, Frühbeck G. Contemporary medical, device, and surgical therapies for obesity in adults. Lancet. 2023;401:1116–30. [DOI] [PubMed] [Google Scholar]
- [9].Powell-Wiley TM, Poirier P, Burke LE, et al. Obesity and cardiovascular disease: a scientific statement from the American Heart Association. Circulation. 2021;143:e984–e1010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Kreuter R, Wankell M, Ahlenstiel G, Hebbard L. The role of obesity in inflammatory bowel disease. Biochim Biophys Acta Mol Basis Dis. 2019;1865:63–72. [DOI] [PubMed] [Google Scholar]
- [11].Romero-Corral A, Somers VK, Sierra-Johnson J, et al. Accuracy of body mass index in diagnosing obesity in the adult general population. Int J Obes (Lond). 2008;32:959–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Woolcott OO, Bergman RN. Relative fat mass (RFM) as a new estimator of whole-body fat percentage: a cross-sectional study in American adult individuals. Sci Rep. 2018;8:10980. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Kobo O, Leiba R, Avizohar O, Karban A. Relative fat mass (RFM) as abdominal obesity criterion for metabolic syndrome. Eur J Intern Med. 2019;63:e9–e11. [DOI] [PubMed] [Google Scholar]
- [14].Togashi Y, Miyashita D, Tsuno T, et al. Abdominal aortic calcification is associated with Fibrosis-4 index and low body mass index in type 2 diabetes patients: a retrospective cross-sectional study. J Diabetes Investig. 2022;13:1861–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Rahman EU, Chobufo MD, Farah F, et al. Prevalence and risk factors for the development of abdominal aortic calcification among the US population: NHANES study. Arch Med Sci Atheroscler Dis. 2021;6:e95–e101. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Li W, Wang Z, Li M, Xie J, Gong J, Liu N. Association between a body shape index and abdominal aortic calcification in general population: a cross-sectional study. Front Cardiovasc Med. 2022;9:1091390. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Peng J, Wu J, Luo X, et al. The cross-sectional association between cardiometabolic index and abdominal aortic calcification in U.S. adults: evidence from NHANES 2013-2014. Front Nutr. 2025;12:1537795. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Ren M, Shen Y, Yuan Y, Han Z, Zhong Y, Liu S. Association between cardiometabolic index and abdominal aortic calcification in US adults from the NHANES. Sci Rep. 2025;15:17831. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Qin Z, Du D, Li Y, et al. The association between weight-adjusted-waist index and abdominal aortic calcification in adults aged ≥ 40 years: results from NHANES 2013-2014. Sci Rep. 2022;12:20354. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Xie F, Xiao Y, Li X, Wu Y. Association between the weight-adjusted-waist index and abdominal aortic calcification in United States adults: results from the national health and nutrition examination survey 2013-2014. Front Cardiovasc Med. 2022;9:948194. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Luo L, Liu M. Adiponectin: friend or foe in obesity and inflammation. Med Rev (2021). 2022;2:349–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Zhao S, Kusminski CM, Scherer PE. Adiponectin, leptin and cardiovascular disorders. Circ Res. 2021;128:136–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].Tadiotto MC, Corazza PRP, Menezes Junior FJ, et al. Lower adiponectin is associated with higher anthropometry and insulin resistance but not with low cardiorespiratory fitness in adolescents. J Endocrinol Invest. 2024;47:307–14. [DOI] [PubMed] [Google Scholar]
- [24].Flockhart M, Tischer D, Nilsson LC, et al. Reduced glucose tolerance and insulin sensitivity after prolonged exercise in endurance athletes. Acta Physiol (Oxf). 2023;238:e13972. [DOI] [PubMed] [Google Scholar]
- [25].Khandelwal A, Ahmed KA, Dikshit P. Vitamin D and vitamin K: synergistic roles and emerging evidence for combined supplementation. J Midlife Health. 2025;16:505–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].D’Elia S, Bottino R, Carbone A, et al. Modulation of cardiometabolic risk by Vitamin D and K2: simple supplementation or real drug? Uncovering the pharmacological properties. Int J Mol Sci . 2025;27:298. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Wang Y, Su Y, Luo C, Jiang W, Yang H, Qin X. The association between the aggregate index of systemic inflammation and abdominal aortic calcification: a national health and nutrition examination survey. J Cardiothorac Surg. 2025;20:367. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Li C, Xian X, Zhao X, et al. Genetic association between inflammatory factors and abdominal aortic calcification: insights from a genome-wide association study. Int J Cardiol. 2025;440:133746. [DOI] [PubMed] [Google Scholar]
- [29].Szulc P, Lewis JR, Chapurlat R. Accelerated bone loss in older men with severe abdominal aortic calcification: the prospective MINOS study. J Clin Endocrinol Metab. 2023;109:e32–9. [DOI] [PubMed] [Google Scholar]