The Multi-Ethnic Study of Atherosclerosis (MESA) is a community-based longitudinal cohort study with 6,814 men and women aged 45 to 84 years at baseline, recruited from six United States communities. At this visit, it was comprised of 38% Caucasians, 28% African Americans, 22% Hispanics, and 12% Asians (primarily Chinese). The first examination began in July 2000, and the subsequent examinations were conducted regularly. The latest (6th) examination was from September 2016 to June 2018. Details of the study protocol and design have been published previously [13].

The components of the first examination included personal history, demographic data, socioeconomic status, medications and medical history, occupational and leisure-time physical activity, anthropometry, blood pressure measurement, phlebotomy, spot urine collection for microalbuminuria, and cardiac computed tomography (CT) for coronary calcification and magnetic resonance imaging scans for cardiac structure and function. Informed consent was obtained from all participants at each study clinic.

The abdominal body composition ancillary study consisted of a random subset of 1,975 MESA participants who underwent abdominal CT for measurement of aortic calcium during visit 2 and visit 3 examinations (2002 to 2005). These scans were then interrogated for different body composition measures to include visceral, subcutaneous, and intermuscular fat, as well as lean muscle mass in the abdomen (as described below). This data was used for cross-sectional analyses in the current study. Among the participants who underwent CT during second visit (visit 2), 595 underwent a follow-up CT scan during visit 4 (2005 to 2007), which was used for the longitudinal analyses in this study.

The MESA studies were approved by the Institutional Review Board of each study site and all participants provided written informed consent. The Institutional Review Board of Korea University Anam Hospital also approved this study (IRB number 2021AN0501).

Anthropometric measurements included height, body weight, WC, and hip circumference. BMI was calculated as weight (kg) divided by the square of the height (m²). WWI was calculated as WC (cm) divided by the square root of weight in kilograms (√kg).

In the MESA body composition ancillary study, abdominal slices at vertebral level L4/L5 from the abdominal CT were processed using Medical Image Processing, Analysis, and Visualization version 4.1.2 software (MIPAV, National Institutes of Health, Bethesda, MD, USA) that measured the areas of fat, muscle mass, and total tissue by a semiautomated method. The densities of all anatomical structures were measured and expressed in Hounsfield units (HU) for each tissue type. Fat tissue and skeletal muscle tissue were identified as being between −190 and −30 HU and between 0 and 100 HU, respectively [14]. Any densities other than these two HU ranges were labeled as undefined tissue type. Total abdominal fat area (TFA) was subdivided into subcutaneous fat area (SFA) and visceral fat area (VFA) according to the corresponding anatomical location. Muscle area was defined within their unique fascial planes as psoas, rectus abdominis, paraspinal muscle group, and oblique muscle group. Total abdominal muscle area (TMA) was subdivided into muscles of locomotion (psoas muscle) and muscles of stability (rectus abdominis, paraspinal muscle group, and oblique muscle group). Muscle density was the average HU within each muscle’s distinct fascial plane. Inter-rater and intra-rater reliability for total abdominal, subcutaneous, and visceral cavity areas were 0.99 for all measurements. Inter- and intra-rater reliability for all muscle groups ranged from 0.93 to 0.98.

Sociodemographic and lifestyle information including age, sex, race, current occupation, family income, insurance status, marital status, detailed history of cigarette smoking and alcohol consumption was obtained from standardized questionnaires. Current and past medical history including medications, treatments, and medical conditions was also obtained. Diabetes mellitus was defined as fasting glucose ≥126 mg/dL or use of glucose-lowering agents. Hypertension was defined as systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg or use of anti-hypertensive medications. Using the Typical Week Physical Activity Survey, participants self-reported their time spent in sedentary behavior and in physical activities.

Demographics, laboratory measures, and body composition measures using abdominal CT scans at baseline were presented as mean (standard deviation) or number (percentage). A Pearson correlation analysis was performed to investigate the linear relationship between WWI and both abdominal fat and muscle mass measures. The results are presented as correlation coefficients and corresponding 95% confidence intervals based on Fisher’s r-to-z transformation. To compare the distribution of WWI and BMI according to race, age and sex adjusted z-score for WWI and BMI were calculated by regressing WWI or BMI on age and sex, respectively and standardized residuals of the regression models. Density plot of age-sex or sex adjusted WWI z-score drawn by smooth curves that goes through the tops of the histogram bars were displayed according to race (Caucasian, Asian, African American, and Hispanic) and age (45-54, 55-64, 65-74, and 75-84 years). The difference in mean WWI by race and age was evaluated after adjusting for age and sex (separately) using analysis of covariance. Means of fat and muscle measures according to quartiles of WWI were compared by one-way analysis of variance and used Tukey’s method for post hoc analysis.

Five hundred and ninety-five participants underwent repeated abdominal CT during the 4th MESA examination. The mean interval between the 2nd and 4th examinations was 38.6±3.3 months. The longitudinal changes in the anthropometric index and abdominal fat and muscle area measures were assessed using the paired t-test. To evaluate the association between the change in anthropometric index and change in body composition measures, means of percentage changes in TFA or TMA were compared by quartiles of change in WWI or BMI using analysis of variance with adjustment for age.

Statistical significance was set at P<0.05. All statistical analyses were performed using the SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA) and R version 4.0.2 (R Foundation for Statistical Computing, Vienna, Austria).

Correlation analyses showed that abdominal fat area measures, including TFA, SFA, and VFA, positively correlated with WWI, whereas all abdominal muscle area measures negatively correlated with WWI, regardless of race/ethnicity (P<0.001 for all analyses) (Supplementary Table 2). Total abdominal muscle radiodensity negatively correlated with WWI, which indicated that the fat component within abdominal muscles was positively associated with WWI.

The association of WWI with abdominal fat and muscle area measures was examined by comparing the mean values and z-scores according to WWI quartiles (Table 2, Supplementary Table 3). The mean values of abdominal fat area z-scores showed increasing trends from the lowest quartile to the highest quartile of WWI, whereas those of muscle area z-scores showed decreasing trends with increasing quartiles of WWI. These results were consistent regardless of ethnicity (P for trends <0.001 for all ethnic groups), and age (Fig. 1). Fig. 2 is a representative figure showing the contrasting relationship between abdominal fat and muscle mass measures with WWI. In contrast, BMI or WC did not reflect the fat and muscle area measures differently. Not only fat area but also muscle area showed positive correlation with WC and BMI (Table 2).

Given that the trend for increasing WWI with increasing age, we examined the distribution of WWI according to age in all participants in the MESA cohort. Fig. 3A shows the distribution curves by age groups and reveals that WWI is positively associated with age. Compared to the 45 to 54 years age group, mean WWI was significantly higher with increasing age groups (P=0.001 for 55-64 years group and P<0.001 for 65-74 and 75-84 years groups). This relationship was consistent in each ethnic group (Supplementary Fig. 2); however, there were no significant associations between BMI and age.

On the other hand, distribution of WWI was similar between ethnic groups. For example, WWI of Asians did not significantly deviate from that of Caucasians and African Americans (P=0.512, and P=0.720, respectively) (Fig. 3B). However, BMI of Asians was significantly lower than that of other ethnic groups (P<0.001 for all).

We examined the longitudinal changes in CT-derived fat and muscle area measures and their associations with anthropometric index. Over 38.6 months, the mean TFA and VFA increased, but the TMA decreased in all participants (Supplementary Table 4). Mean SFA increased significantly only in men, not in women. During the same period, BMI did not change, but the WC and WWI significantly increased. Increase in WWI was commonly observed in all age groups, but the direction of changes in BMI and WC differed with age (Supplementary Table 5).

To examine whether longitudinal changes in WWI were associated with increasing fat mass and decreasing muscle area, we examined the percentage changes of TFA and TMA according to quartiles of changes in WWI (delta WWI) (Fig. 4). We found that as WWI increased, TFA increased (P<0.001) and TMA decreased, although changes in TMA were not statistically significant (P=0.963). In contrast, as BMI increased, both TFA and TMA increased (P<0.001 and P=0.056, respectively).