We used data from the National Health and Nutrition Examination Survey (NHANES) collected between 1999 and 2012. This survey covers a representative sample from the United States and has been performed biannually by the National Center for Health Statistics since 1999 using questionnaire-based personal interviews, physical examinations, and laboratory tests. Among a population of 71,916, those 20 years of age or younger as well as those who lacked questionnaire-based data on CVD events; weight, height, and blood pressure (BP) measurements; and/or laboratory data including fasting glucose, triglycerides, and high-density lipoprotein were excluded. Ultimately, 25,270 adults were included in this study (Supplementary Fig. 1). Whole body DXA data were available between 1999 and 2005; therefore, 11,317 participants were included in the subgroup analysis for body composition.

WC was measured by placing a flexible tape around the uppermost lateral border of the ilium. BMI was defined as the weight (kg) divided by the square of the height (m). BP was measured three times after a minimum of 5 minutes of rest while the participant was in a sitting position, and the mean value of the three measurements was used in our analysis. Fasting blood glucose and lipid levels were measured using the enzymatic method. More detailed sample collection and test methods are described in the NHANES Laboratory Procedures Manual [12].

In the present study, overweight/obesity was defined as a BMI above 25 kg/m² based on the World Health Organization criteria [13]. Owing to lack of a standard definition for metabolically healthy status, the revised National Cholesterol Education Program–Adult Treatment Panel III criteria (NCEP–ATP III criteria) for metabolic syndrome, which are the most commonly used criteria in previous studies [14151617], were also used. Metabolic abnormality was defined as having two or more metabolic risk factors according to the revised NCEP–ATP III criteria, including impaired fasting glucose (i.e., a fasting glucose level >100 mg/dL or a diagnosis of DM), high BP (hypertensive BP >130 mm Hg and/or diastolic BP >85 mm Hg, or a diagnosis of hypertension), triglyceride level ≥150 mg/dL, and a high-density lipoprotein cholesterol level <40 mg/dL in men and <50 mg/dL in women. Owing to the collinearity between BMI and WC, the central obesity criterion was not used. Based on these aforementioned criteria, the participants were classified into the following groups: metabolically healthy normal weight (MHNW), MHO, metabolically unhealthy normal weight (MUNW) and metabolically unhealthy overweight/obese (MUO). We identified the existence of a CVD event if 1 or more of the following structured questions were answered in the affirmative:

Among the 25,270 study participants, 11,317 underwent a whole body DXA using a QDR 4500A fan beam X-ray bone densitometer (Hologic Inc., Marlborough, MA, USA). The total and regional body compositions as measured using the DXA scans were analyzed using the Hologic Discovery software version 12.1 (Hologic Inc.). The appendicular skeletal mass was defined as the sum of the total lean mass excluding bone mineral contents from the limbs, and the ASMI was defined as the appendicular skeletal mass divided by the square of the height (m). Sarcopenia was defined as an ASMI <7 kg/m² in men and <5.5kg/m² in women based on the revised European consensus on the definition and diagnosis of sarcopenia [18]. ABSI was calculated using the following equation: ABSI=WC/(BMI^2/3×height^1/2) [9].

The NHANES is a complex sample survey representing the United States population; the sample analysis was conducted using weighted values. Continuous variables are shown as mean values and 95% confidence intervals (CIs), while categorical variables are shown as the prevalence and percentage according to each obesity phenotype. Each variable was compared using 1-way analysis of variance and Pearson's chi-square test. We implemented Pearson's correlation coefficient to verify the correlation of ABSI with ASMI, BMI, and WC. A receiver operating characteristic (ROC) curve was used to analyze the correlation between ABSI and sarcopenia according to obesity phenotypes. A multivariate logistic regression analysis was performed to evaluate the odds ratio (OR) for CVD events according to the obesity phenotype. To determine the effects of muscle mass, groups were divided according to the presence of sarcopenia, and each group's OR for a CVD event was examined. Moreover, ABSI was divided into three categories, and the ORs for CVD events were calculated. Furthermore, changes in the ORs for CVD events in each obesity phenotype as a function of changes in ABSI were analyzed using the restricted cubic spline plots with 3 knots. The analysis was performed using R version 3.1.0 (R Foundation for Statistical Computing, Vienna, Austria; www.r-project.org) and IBM SPSS Statistics version 24.0 (IBM Co., Armonk, NY, USA). A P<0.05 indicated statistical significance.

The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Institutional Review Board of Hallym University Kangnam Sacred Heart Hospital (IRB No. HKS 2017-07-007). All the NHANES protocols were approved by the Research Ethics Review Board of the National Center for Health Statistics, US Centers for Disease Control and Prevention (NCHS IRB/ERB protocol numbers: 1999–2004, Protocol #98-12; 2005–2010, Protocol #2005-06; and 2011–2012, Protocol #2011-17). All participants volunteered and provided written informed consent prior to their enrollment. All participants' records were anonymized before being accessed by the authors, and all methods were performed in accordance with approved guidelines and regulations.

The baseline characteristics of the participants according to their obesity phenotypes are summarized in Table 1. Of the 25,270 participants assessed in this study, 5,176 had MHNW, 6,688 were MHO, 2,671 were MUNW, and 10,735 were MUO. Compared to metabolically healthy participants, those metallically unhealthy participants were more likely to be men and of older age; they also had higher BP, poorer lipid and glucose profiles, and a higher prevalence of CVD. Metabolically unhealthy participants had higher ABSIs than metabolically healthy counterparts. DXA data were available for 11,317 of the 25,270 subjects; participants who were obese had significantly higher ASMIs than those who were not. Moreover, participants with normal weight showed a higher prevalence of sarcopenia than obese participants did.

A total of 2,637 CVD events were reported. In the multivariate logistic regression model with adjustment for age, sex, ethnicity and smoking status, participants with MHO were at a moderate risk of developing CVD when compared with MHNW participants (OR, 1.543; 95% CI, 1.158 to 2.056), whereas participants with metabolic abnormalities were at high risk regardless of obesity status (OR in MUNW, 2.026; 95% CI, 1.539 to 2.667) (OR in MUO, 2.725; 95% CI, 2.186 to 3.398) (Table 2).

ABSI showed a moderate negative correlation with ASMI as well as a strong positive correlation with WC, regardless of obesity phenotype (Table 3). In contrast to ABSI, WC and BMI both showed a strong positive correlation with ASMI (overall correlation coefficient for WC, 0.631, P<0.001; overall correlation coefficient for BMI, 0.655, P<0.001). The area under the ROC curve (AUC) for ABSI showed significant accuracy for identifying sarcopenia (Table 4).

The multivariate logistic regression analysis showed that the participants with sarcopenia had higher risk for CVD than those without sarcopenia in the group with MHNW (OR, 2.69; 95% CI, 1.56 to 4.64). The obese participants with sarcopenia showed a significantly higher ORs of CVD than those without sarcopenia in both the metabolically healthy and unhealthy groups (Fig. 1). Fig. 1A shows each OR according to obesity phenotype and sarcopenia status compared to MHNW participants without sarcopenia as a reference. The information in Fig. 1B is based on MHNW participants with sarcopenia as a reference. MHO participants had a significantly higher risk for CVD than did MHNW participants without sarcopenia (OR in MHO without sarcopenia, 3.31; 95% CI, 1.94 to 5.64) (OR in MHO with sarcopenia, 8.59; 95% CI, 2.63 to 28.04) (Fig. 1A). However, MHO participants did not have a significantly higher risk for CVD compared to the MHNW participants with sarcopenia (OR in MHO without sarcopenia, 1.23; 95% CI, 0.77 to 1.96) (OR in MHO with sarcopenia, 3.19; 95% CI, 0.99 to 10.23) (Fig. 1B).

The multivariate logistic regression analysis showed that MHO participants who were within the second and third tertiles of ABSI had significantly higher ORs for CVDs, while those who were within the first tertile of ABSI had only a slightly higher risk for CVD, compared to the MHNW participants who were within the first ABSI tertile (Fig. 1C). However, MHO participants did not have a significantly higher risk of CVD compared to the MHNW participants within the third ABSI tertile (OR in MHO participants within the first tertile, 0.86; 95% CI, 0.56 to 1.32) (OR in MHO participants within the second tertile, 1.04; 95% CI, 0.71 to 1.54) (OR in MHO participants within the third tertile, 1.34; 95% CI, 0.98 to 1.84) (Fig. 1D). On restricted cubic spline regression analysis, ORs for CVD increased by ABSI in each obesity phenotype (Fig. 2).