Abstract
Objective:
While increasing evidence suggests that visceral adipose tissue (VAT) is a major underlying cause of metabolic syndrome (MetS), few studies have measured VAT at multiple time points in diverse populations. We hypothesized that VAT and insulin resistance differ by MetS status within body mass index (BMI) category in the IRAS Family Study, and further that baseline VAT and insulin resistance and increases over time are associated with incident MetS.
Methods:
We used generalized estimating equations for differences in body fat distribution and insulin resistance by MetS status. We used mixed effects logistic regression for the association of baseline and change in adiposity and insulin resistance with incident MetS across 5 years, adjusted for age, sex, race/ethnicity, and family correlation.
Results:
VAT and insulin sensitivity differed significantly by MetS status and BMI category at baseline. VAT and HOMA-IR at baseline (VAT OR=1.16 (1.12-2.31); HOMA OR=1.85 (1.32-2.58)) and increase over time (VAT OR=1.55 (1.22-1.98); HOMA OR=3.23 (2.20-4.73)) were associated with incident Mets independent of BMI category.
Conclusions:
Differing levels of VAT may be driving metabolic heterogeneity within BMI categories. Both overall and abdominal obesity (VAT) may play a role in the development of MetS. Increased VAT over time contributed additional risk.
Keywords: Visceral adipose tissue, obesity, metabolic syndrome, insulin resistance, African American, Mexican American, longitudinal
INTRODUCTION
The obesity epidemic has tremendous public health implications with regard to cardiometabolic risk;(1) however, interest in describing individuals with obesity who may not be at increased risk for these outcomes remains.(2) Specifically, significant variation exists in cardiometabolic risk factor distributions among individuals of similar body mass index (BMI), suggesting that an individuals’ cardiovascular disease (CVD) risk may depend jointly on body size and metabolic profile.(3, 4, 5, 6, 7) Individuals in BMI categories can be further characterized as either metabolically healthy or metabolically unhealthy,(2) resulting in more specific metabolic phenotypes that are intended to account for heterogeneity within BMI categories and more accurately describe individuals at risk for CVD.(3, 4, 5, 6, 7) Given that 30% or more of US individuals with obesity may be metabolically healthy,(7) there is specific interest in identifying those with obesity who are most at risk from adverse outcomes.
One of the most promising suggestions for distinguishing between metabolically healthy obesity (MHO) and metabolically unhealthy obesity (or metabolic syndrome (MetS) in obesity) involves utilizing the heterogeneity of risk associated with differing body fat distributions. Increasing evidence suggests that MetS status within obesity is distinguished by differences in visceral adipose tissue (VAT),(8, 9, 10, 11, 12) indicating that fat distribution may be a more specific predictor of the cardiometabolic dysfunction than overall excess adiposity. Despite this evidence, few studies have utilized longitudinal data to show that differences in VAT are associated with transition to MetS,(13, 14, 15, 16) and even fewer have the repeated measures needed to investigate the influence of changes in body fat composition over time.(17) Similar work has suggested that either differences in insulin resistance or other indicators of glucose control and response or inflammatory markers can be used as early indicators of who will progress to MetS from MHO.(11, 18, 19, 20) If any of these markers can be used to reliably discriminate between these phenotypes with regard to risk, then they could be targets for clinical risk assessment and intervention.
While the current estimate for obesity prevalence in the general US population is between 30-40%,(21) relative to US non-Hispanic whites (37.1), the prevalence of obesity is higher in US non-Hispanic blacks (48.5%) and Hispanics (42.6%).(21) These are therefore populations of particular interest for preventing the cardiometabolic risks of obesity, where understanding early markers of risk may have the greatest health impact. Unfortunately, most studies that have investigated the distinguishing characteristics of MHO and determinants of the transition to MetS have primarily studied Caucasian/Northern European populations.(1, 5)
In order to address these gaps, the goals of this study were twofold. The first goal was to assess whether body fat distribution and insulin resistance differ by MetS status within BMI categories in the African American and Mexican American participants of The Insulin Resistance Atherosclerosis Study (IRAS) Family Study. The second goal was to investigate the longitudinal association of body fat composition at baseline and change over time with transition to incident MetS.
METHODS
Study Population
The Insulin Resistance Atherosclerosis Study (IRAS) Family Study is an epidemiologic cohort study of men and women designed to investigate the genetics of insulin resistance and visceral adiposity.(22) Multi-generational families of African American and Mexican American background were enrolled using probands of the original IRAS study, plus those supplemented from the general population.(23) All studies were conducted using protocols approved by the Institutional Review Boards at each participating institution and all participants provided informed consent.
Briefly, three IRAS sites (Los Angeles, CA, San Antonio, TX, and San Luis Valley, CO) examined family members of African American or Mexican American ethnicity over a 2.5-year period from 2000-2002. Eligibility criteria for the cohort included: 1. African American or Mexican American ethnicity; 2. 18 years of age or older; 3. under 350 pounds for computed tomography (CT) examination; and 4. not having conditions that interfere with measurement of insulin resistance or any cardiometabolic risk factor.(24) The IRAS Family Study includes 3,716 observations on 1858 participants (585 African American and 1,273 Mexican American) across two visits. For the cross-sectional analysis we excluded those with prevalent diabetes (n=227), participants who were underweight (BMI<18.5kg/m2) (n=32), and participants who did not have body fat composition measured (n=157) for a total of 1475 participants (467 African American and 1009 Mexican American). There was some overlap between the excluded groups. In the longitudinal analysis, we included only those participants who returned for a second exam after 5 years where anthropometry, body composition, and MetS components were re-measured. For this analysis of incident events, we excluded 426 participants who did not have data from the second exam and 488 participants with prevalent MetS or T2D at baseline for a total of 944 (277 African American and 667 Mexican American).
Metabolic Phenotypes
Metabolic phenotypes were defined using a combination of the WHO-based BMI category cut-offs (normal weight <25.0, overweight 25.0-29.9, and obesity 30.0+kg/m2) and the consensus International Diabetes Federation (IDF) criteria for metabolic syndrome (MetS) (three or more of the following: Hypertension defined as either systolic/diastolic blood pressure ≥ 130/85 mmHg or documented antihypertensive use; high triglycerides defined as fasting triglyceride level ≥150 mg/dL; low high-density lipoprotein (HDL) cholesterol defined as HDL < 40 mg/dL in men and < 50 mg/dL in women; high fasting glucose defined as fasting glucose level ≥ 100 mg/dL; or high waist circumference defined as >88 cm in women and >102 cm in men):(25)
Healthy normal weight: BMI <25.0 kg/m2 and < 3 MetS criteria
MetS normal weight: BMI <25.0 kg/m2 and ≥ 3 MetS criteria
Healthy overweight: BMI 25.0- <30.0 kg/m2 and < 3 MetS criteria
MetS overweight: BMI 25.0- <30.0 kg/m2 and ≥ 3 MetS criteria
Healthy obesity (MHO): BMI ≥ 30.0 kg/m2 and < 3 MetS criteria
MetS obesity: BMI ≥ 30.0 kg/m2 and ≥ 3 MetS criteria
Height and weight were measured in duplicate to the nearest 0.5 cm and 0.1 kg, respectively. BMI was calculated as weight/height squared (kg/m2). Waist circumference (WC) was measured to nearest 0.1 cm at the level of the iliac crest at the end of normal respiration. The use of antihypertensive, lipid-lowering, and antidiabetic medications was assessed by self-report at the enrollment interview and at follow-up. Seated systolic/diastolic blood pressure was averaged over the last two of three measurements with a mercury manometer, after a five-minute rest by centrally trained technicians using identical equipment. Plasma triglycerides and HDL cholesterol were determined by enzymatic colorimetric assays using a Chemistry Analyzer Model ATAC 8000 (Elan Diagnostics, Smithfield, RI). Fasting plasma glucose was measured using the glucose oxidase technique on an automated autoanalyzer (YSI, Yellow Springs, OH). Insulin was measured by radioimmunoassay (Linco Research, St Charles, MO).
Body Fat Distribution
Computed tomography (CT) imaging for abdominal fat distribution was obtained under standardized protocol and scans were read centrally. Participants received a scout view of the abdomen and pelvis followed by three axial images all during suspended respiration. The 10-mm image obtained at the L4–L5 disc space was used for determination of VAT area and subcutaneous adipose tissue (SAT) area. All CT images were coded for pathology and image quality; poor-quality scans were excluded from analysis.(24) Total body fat was measured using dual-energy X-ray absorptiometry and a standardized protocol. VAT and SAT measurements were repeated using the same methodology at a five-year interval at Baseline and Visit 2, a five-year interval.
Markers of glucose homeostasis and inflammation
Insulin sensitivity (SI) and acute insulin response (AIR) were measured at baseline using the frequently sampled intravenous glucose tolerance test (FSIGT) with minimal model analysis.(26, 27) The FSIGT was not conducted at the follow-up visit and thus for longitudinal analysis we used HOMA-IR (Homeostasis model assessment) to evaluate insulin resistance at both visits using the formula: Fasting Plasma Insulin Level (uU/mL) x Fasting Plasma Glucose Level (mmol/L)/22.5.
Several inflammatory markers were also assayed from fasting plasma at baseline. Adiponectin concentration was quantified using radioimmunoassay (Linco Research, St Charles, MO). High sensitivity C-reactive protein (hs-CRP) was measured using ultrasensitive ELISA (Calbiochem, La Jolla, CA). Retinol Binding Protein 4 (RBP4) and Plasminogen Activator Inhibitor Type 1 (PAI-1) were measured using highly sensitive enzyme-linked immunosorbent assays (R&D systems, Minneapolis, MN).
Statistical Analysis
We described the baseline characteristics of the study population by VAT quartile using means and standard deviations for continuous variables and frequencies and percentages for categorical variables. We compared continuous characteristics cross-sectionally between metabolically healthy and unhealthy subgroups within each BMI category using an unadjusted generalized estimating equation (GEE) with an identity link that accounts for the correlation between family members using a sandwich estimator of variance with an exchangeable correlation structure. We used the GEE approach assuming a normal distribution adjusted for age, sex, and race/ethnicity to estimate adjusted means by metabolic phenotype and to examine whether metabolic phenotype was differentially associated with body fat composition, glucose homeostasis parameters, and inflammatory markers. Where noted, we transformed variables for analysis that did not conform to the normal distribution. These transformations were centrally determined by the IRAS Family Study Coordinating Center. Contrasts of the six metabolic phenotypes were made by using indicator variables. We formally tested for interaction of metabolic phenotype by age, sex, and race/ethnicity in all models by including interaction terms; none of these terms were statistically significant so the final models included main effects only in the full cohort. All cross-sectional analyses were performed using SAS 9.4 (SAS Institute, Cary NC).
For the analysis investigating the association between change in adiposity and insulin resistance with incident MetS, we used logistic regression with random effects to account for correlations within family and within individual over time. Variables were modeled continuously as difference in outcome per 1 standard deviation exposure. Models including SAT did not include BMI, and vice versa as we considered BMI to be a proxy measure for SAT. We used likelihood ratio tests to determine whether inclusion of variables significantly improved model fit. We specifically investigated the improvement of model fit from the inclusion of VAT to a model with BMI, and the inclusion of change variables to models with baseline values. We also investigated the association of VAT and HOMA-IR with each MetS component individually to determine whether a relationship with a single component explains the overall MetS associations. All longitudinal analyses were performed using Stata 14 (StataCorp, College Station TX).
For both the cross-sectional and longitudinal models we investigated sensitivity of the results by MetS definition, including 1. MetS defined as ≥2 components excluding waist circumference, and 2. MetS defined as ≥1 component to assess a truly healthy reference group.
RESULTS
Out of 1475 participants at baseline, 35% (n=514) were overweight, 35% (n=523) had obesity, and 16% (n=243) met the definition for MetS. Very few normal weight participants had MetS (n=8/438 or 1.8%), but the prevalence of MetS was higher in those met the criteria for overweight (12%) or obesity (33%). As seen in Table 1, participants with higher VAT at baseline were more likely to be older, male, and Mexican American. In addition, those with higher VAT were also more likely to have MetS components and cardiovascular disease risk factors, higher insulin resistance, and higher levels of inflammatory markers (Table 1).
Table 1.
Baseline characteristics (mean (standard deviation) or frequency (%)) of 1475 IRAS Family study participants by visceral adipose tissue quartile at baseline
| Visceral Adipose Tissue Quartile | ||||||
|---|---|---|---|---|---|---|
| Characteristic | 1 (<57.9 cm2) |
2 (57.9 to <92.0 cm2) |
3 (92.0 to <133.3 cm2) |
4 (≥133.3 cm2) |
p-value | |
|
|
|
|
|
|
|
|
| N | 368 | 370 | 369 | 369 | ||
| Visceral Adipose Tissue (cm2) | 38.7 (12.2) | 74.8 (9.71) | 111.3 (11.7) | 180.4 (43.0) | ||
| Age (years) | 32.1 (9.88) | 39.6 (12.6) | 43.7 (12.8) | 49.4 (13.0) | <0.001 | |
| Sex (N, % Female) | 253 (69) | 245 (66) | 189 (51) | 177 (48) | <0.001 | |
| Race/Ethnicity | ||||||
| (N, % Mexican American) | 201(55) | 266 (72) | 256 (69) | 286 (78) | <0.001 | |
| (N, % African American) | 167 (45) | 104 (28) | 113 (31) | 83 (22) | ||
| Systolic Blood Pressure (mmHg) | 109 (14.0) | 113 (13.5) | 120 (17.2) | 125 (17.4) | <0.001 | |
| Diastolic Blood Pressure (mmHg) | 71.1 (9.44) | 74.6 (8.73) | 77.2 (9.58) | 79.5 (9.74) | <0.001 | |
| Hypertension Medication (N, %) | 8 (2.17) | 26 (7.03) | 55 (14.9) | 93 (25.2) | <0.001 | |
| Total Cholesterol (mg/dL) | 161.8 (30.5) | 173 (32.7) | 185 (37.1) | 188 (38.9) | <0.001 | |
| HDL Cholesterol (mg/dL) | 51.3 (12.9) | 45.1 (12.3) | 43.7 (11.9) | 40.1 (12.0) | <0.001 | |
| Triglycerides (mg/dL) | 65.1 (39.67) | 93.6 (59.7) | 120 (77.6) | 144 (99.0) | <0.001 | |
| Lipid Medication (N, %) | 2 (0.54) | 6 (1.62) | 15 (4.07) | 25 (6.78) | <0.001 | |
| Fasting Glucose (mg/dL) | 89.0 (7.13) | 91.2 (7.68) | 95.5 (9.30) | 100 (10.1) | <0.001 | |
| BMI (kg/m2) | 24.0 (3.72) | 27.5 (4.51) | 30.4 (5.46) | 33.1 (5.93) | <0.001 | |
| Waist Circumference (cm) | 75.5 (7.82) | 85.1 (8.99) | 93.3 (10.4) | 102 (11.2) | <0.001 | |
| Total Adipose Tissue (%) | 30.1 (9.26) | 33.4 (8.49) | 33.3 (8.99) | 35.6 (8.67) | <0.001 | |
| Subcutaneous Adipose Tissue (cm2) | 225 (124) | 329 (142) | 376 (163) | 417 (156) | <0.001 | |
| Fasting insulin (uU/mL) | 9.68 (8.75) | 12.7 (9.46) | 15.8 (10.29) | 20.9 (11.5) | <0.001 | |
| HOMA-IR | 2.14 (1.99) | 2.91 (2.40) | 3.74 (2.57) | 5.27 (3.13) | <0.001 | |
| Acute Insulin Response (pmol/ml/min) | 778 (636) | 814 (668) | 874 (786) | 848 (756) | 0.033 | |
| Insulin Sensitivity Index (10−4min−1μU−1ml−1) | 3.04 (1.86) | 2.22 (1.63) | 1.66 (1.45) | 0.90 (0.75) | <0.001 | |
| C-Reactive Protein (ug/mL) | 1.86 (3.31) | 2.95 (3.83) | 4.05 (4.27) | 4.74 (5.11) | <0.001 | |
| Adiponectin (ug/mL) | 14.3 (6.84) | 12.2 (6.54) | 10.8 (5.73) | 10.3 (5.31) | <0.001 | |
| PAI-1 (ng/mL) | 17.1 (14.4) | 31.2 (33.5) | 42.9 (33.4) | 56.9 (40.9) | <0.001 | |
| RBP4 (ug/mL) | 25.7 (7.28) | 26.8 (7.83) | 28.4 (8.07) | 29.0 (7.56) | ||
P-values determined using a generalized estimating equation with a sandwich estimator for variance to account for the correlation between families.
VAT was significantly higher in individuals with MetS compared to metabolically healthy individuals (no MetS) within each BMI category and progressively higher with each BMI category in cross-sectional analysis adjusted for age, sex, and race/ethnicity (Figure 1 and Supplemental Table S1). SAT did not differ by MetS status (except for the normal weight group), but did differ significantly across the BMI categories (Figure 1B). BMI itself did differ by MetS status for the normal weight and overweight groups, but did not significantly differentiate MHO and obesity with MetS (Supplemental Table S1). HOMA-IR was significantly higher (and insulin sensitivity index significantly lower) in those with MetS compared to healthy individuals within each BMI category (Supplemental Table S1). Acute insulin response was significantly lower in the MetS group but only for those with obesity (Supplemental Table S1). Adiponectin was significantly lower for the group with MetS within each BMI category, with the inverse for PAI-1. RBP4 was higher in the MetS groups for those with overweight and obesity. CRP did not differ by MetS status, but did differ by BMI category (Supplemental Table S1).There was no evidence of interaction by age, sex, or race/ethnicity (p>0.05) for any of these factors. Estimates were similar when using different definitions of MetS, except that estimates for subcutaneous fat differed significantly by MetS status only when defining MetS as ≥1 component (Supplemental Table S2).
Figure 1.
Body fat distribution (means and 95% confidence intervals [CI]) by metabolic syndrome status and BMI category adjusted for age, sex, and race/ethnicity in 1,475 participants of the IRAS Family Study at baseline. All estimates derived from a generalized estimating equation using a sandwich estimator for variance to account for the correlation between family members. Open circles = metabolically healthy group. Black circles = group with metabolic syndrome. All estimates are significantly different at the P < 0.05 statistical level, except the following: For visceral adipose tissue: MetS normal weight/healthy overweight, MetS normal weight/MetS overweight, and MetS overweight/healthy obesity. For subcutaneous adipose tissue: healthy overweight/MetS overweight and healthy obesity/MetS obesity.
Longitudinal results
Among those with data at Visit 2 without prevalent MetS or type 2 diabetes at baseline, there were 125 incident cases of MetS. Both VAT at baseline and change across visits were significantly associated with incident MetS in a graded and monotonic fashion (Figure 2 and Table 2). Higher BMI category was also significantly associated with higher odds of MetS (Figure 2), but BMI change was not. Including both change in VAT and BMI at baseline significantly improved model fit, but change in BMI did not (likelihood ratio tests: VAT change p<0.001; BMI p<0.001; BMI change p=0.70). Similar relationships were seen for SAT, when adjusted for VAT, with higher SAT associated with higher odds of MetS (per standard deviation OR = 1.68 (1.25-2.67) with likelihood ratio test p < 0.001). Similar to the BMI results, a 1 standard deviation change in SAT was not significantly associated with incident MetS (OR = 0.99 (0.99-1.00); likelihood ratio test p = 0.33). Whether weight loss was reported as intentional or not did not significantly improve the model fit (p = 0.35). HOMA-IR at baseline and change over time were also significantly associated with incident MetS after adjustment for adiposity (VAT and SAT) (Figure 2 and Table 2), and significantly improved model fit (likelihood ratio test p = 0.003). Assuming that insulin resistance is on the pathway between adiposity and MetS, the addition of HOMA-IR variables to the model with demographics incompletely attenuated the estimates for VAT at baseline and change (2.89 (2.22-3.76) and 1.93 (1.56-2.40) to 1.91 (1.41-2.59) and 1.59 (1.26-2.02) respectively). There was no significant interaction between VAT and BMI (OR=0.93 (0.75-1.16) for continuous interaction term). No significant heterogeneity was found VAT for HOMA-IR by age, sex, or race/ethnicity (P>0.10).
Figure 2. Odds ratios and 95% confidence intervals for incident metabolic syndrome at 5 years by visceral adipose tissue (VAT) at baseline and change over time, baseline body mass index (BMI), and HOMA-IR at baseline and change over time in 938 IRAS Family Study participants.
* Mutual adjustment plus adjustment for age, sex, race/ethnicity, and family correlation structure. Baseline VAT by quartile (Q). Change in VAT: No change (less than a 30cm2 change between Visits 1 and 2), Decrease (participant lost 30cm2 or more between Visits), or Increase (participant gained 30cm2 or more between Visits). BMI at baseline (normal, overweight, and obesity by WHO category). Baseline HOMA-IR by quartile (Q). Change in HOMA-IR: No change (less than a 2.5 change), Decrease of at least 2.5, or Increase of least 2.5 between visits.
Table 2.
Odds ratios and 95% confidence intervals for the association of measures of adiposity and insulin resistance with incident metabolic syndrome in 944 IRAS Family Study participants
| Model | Incident Metabolic Syndrome | |
|---|---|---|
| OR | 95% CI | |
| Adjusted for age, sex, race/ethnicity | ||
| Visceral Adipose Tissue at Baseline (1SD = 49.31cm3) | 2.36 | 1.86-2.99 |
| Body Mass Index at Baseline (per 1SD = 5.90 kg/m2) | 2.39 | 1.92-2.97 |
| HOMA-IR at Baseline (per 1SD = 2.87) | 1.66 | 1.40-1.98 |
| Physical Activity | 0.98 | 0.84-1.15 |
| Adjusted for age, sex, race/ethnicity and baseline and change of each measure separately | ||
| Visceral Adipose Tissue | ||
| Visceral Adipose Tissue at Baseline (per 1SD = 49.3cm3) | 2.89 | 2.22-3.76 |
| Visceral Adipose Tissue Change (per 1SD=30.6 cm3) | 1.93 | 1.56-2.40 |
| Body Mass Index | ||
| Body Mass Index at Baseline (per 1SD = 5.90 kg/m2) | 2.57 | 2.05-3.22 |
| Body Mass Index Change (per 1SD = 2.77 kg/m2) | 1.18 | 1.09-1.28 |
| Insulin Resistance (HOMA-IR) | ||
| HOMA-IR at Baseline (per 1SD = 2.87) | 2.19 | 1.75-2.74 |
| HOMA-IR Change (per 1SD = 3.90) | 3.33 | 2.48-4.49 |
| Fully adjusted model: Age, sex, race/ethnicity, visceral adipose tissue (baseline and change), BMI (baseline), HOMA-IR (baseline and change) | ||
| Visceral Adipose Tissue at Baseline (per 1SD = 49.3cm3) | 1.61 | 1.12-2.31 |
| Visceral Adipose Tissue Change (per 1SD=30.6 cm3) | 1.55 | 1.22-1.98 |
| Body Mass Index at Baseline (per 1SD = 5.90 kg/m2) | 1.52 | 1.09-2.10 |
| HOMA-IR at Baseline (per 1SD = 2.87) | 1.85 | 1.32-2.58 |
| HOMA-IR Change (per 1SD = 3.90) | 3.23 | 2.20-4.73 |
| Fully adjusted model: Age, sex, race/ethnicity, visceral adipose tissue (baseline and change), subcutaneous adipose tissue (baseline), HOMA-IR (baseline and change) | ||
| Visceral Adipose Tissue at Baseline (per 1SD = 49.3cm3) | 1.69 | 1.22-2.34 |
| Visceral Adipose Tissue Change (per 1SD=30.6 cm3) | 1.57 | 1.24-1.99 |
| Subcutaneous Adipose Tissue at Baseline (per 1SD = 169 cm3) | 1.42 | 1.01-1.98 |
| HOMA-IR at Baseline (per 1SD = 2.87) | 1.82 | 1.32-2.52 |
| HOMA-IR Change (per 1SD = 3.90) | 3.04 | 2.10-4.41 |
A one standard deviation difference in SI and AIR at baseline were both inversely and significantly associated with incident MetS, even after adjustment for VAT and BMI (SI OR=0.68 (0.50-0.93) AIR OR=0.60 (0.47-0.78)). Estimates adjusted for subcutaneous fat were similar. Self-reported increased frequency of physical activity was not significantly associated with incident MetS in any model (Adjusted for age, sex, and race/ethnicity OR = 0.88 (0.75-1.04); Fully adjusted for VAT and BMI OR = 0.96 (0.80-1.16)). CRP at baseline was significantly associated with incident MetS (OR = 1.30 (1.05-1.61)), but this association was fully attenuated by adjustment for adiposity (OR = 0.92 (0.68-1.23)). Adiponectin and RBP4 at baseline were not significantly associated with incident MetS in either univariable or final multivariable models (OR=1.06 (0.85-1.33) for adiponectin and OR=1.18 (0.88-1.60) for RBP4 in the fully adjusted model).
Estimates were generally similar using different MetS definitions, except that BMI was not significantly associated with incident MetS when waist circumference was excluded from the definition (Supplemental Table S3). Similarly, Supplemental Table S4 shows that visceral adipose tissue and HOMA-IR are both associated with more than one MetS component separately.
DISCUSSION
In the IRAS Family Study, metabolic phenotypes were significantly associated with cross-sectional differences in VAT and insulin resistance. Specifically, the groups with MetS were distinguished from the metabolically healthy groups by higher VAT and insulin resistance, and lower insulin secretion. While SAT differed by BMI category, it did not differ by MetS status in the group with overweight or obesity, and inflammatory markers, with the exception of PAI-1 and adiponectin, did not consistently differ by BMI category or MetS status. Longitudinally, baseline levels and increases over time in VAT and insulin resistance were significantly associated with transition to incident MetS. Overall adiposity (SAT and BMI) at baseline, but not change over time were also associated with incident MetS in models including VAT, VAT change, insulin resistance, and insulin resistance change, but not as strongly as VAT and insulin resistance. The results for VAT and HOMA-IR were relatively insensitive to MetS definition.
Consistent with prior studies,(8, 9, 10, 11, 12) VAT provided more accurate discrimination of those with MetS compared to BMI, also in line with current knowledge about the specific repercussions of excess VAT, including higher likelihood of insulin resistance and cardiometabolic disease.(28, 29, 30) Bi et al. along with others have suggested that BMI as a marker of overall obesity inherently misclassifies individual risk, and that measuring VAT levels allows for more accurate classification of individuals at risk for adverse outcomes arising from excess adiposity.(10, 31) This is supported by studies showing strong associations between higher VAT and incident MetS.(13, 14, 15, 16) Longitudinal analysis from the Multi-Ethnic Study of Atherosclerosis bolster this conclusion showing that VAT, both at baseline and increases over time are more strongly associated with incident MetS than BMI, SAT, or waist circumference.(17) Our results build on these findings in a larger sample of repeated measures to support the premise that VAT may be a more specific indicator of cardiometabolic risk from obesity than SAT or BMI. An additional advantage to using VAT to classify metabolic phenotypes is the growing consensus that obesity and VAT accumulation specifically are originating causes of MetS components.(32) Other factors that have been reported to distinguish between metabolic phenotypes, such as insulin resistance and inflammation,(11, 20) indicate existing pathology along the pathway to cardiometabolic risk. Inflammation in particular may be later in the CVD pathway and so may not be as good at discriminating between metabolically unhealthy and healthy participants in the context of obesity or predicting transition to MetS. If inflammation comes only after cardiometabolic pathology has already reached the subclinical level, then these markers may not be sufficient to assess for primary prevention of cardiometabolic outcomes, since the damage may already need to have been done for these to be measureable. Therefore, VAT offers the rare opportunity to classify individuals into risk categories for primary prevention.
Insulin resistance and glucose control may be particularly susceptible to perturbations from obesity. While metabolic dysregulation is a consequence of obesity, it may be early on the pathway to cardiometabolic risk, so those with higher insulin resistance should be considered for prompt intervention and weight loss. Potentially aiding this early recognition of insulin resistance is the strong differentiation of these markers by MetS status, both from our findings and consistently in the literature.(12, 33) Specifically, we found similar results to those of Gaillard et al., who reported significantly higher insulin resistance for those with MetS in obesity compared to MHO in African American women with family history of type 2 diabetes.(33) Our findings also support the idea that insulin resistance is a partial mediator early in the pathway between excess adiposity and MetS. These results are not comparable to prior work in the IRAS Family Study Mexican American cohort due to differences in the analysis and definition of MetS;(34) However, our longitudinal analysis is consistent with prior results and further supports the importance of insulin resistance as an early indicator and contributor to the development of MetS.(16)
Several limitations should be considered when interpreting the study findings. First, the African American cohort of the IRAS Family Study comes from one clinic site which may limit generalizability to African Americans from other geographic areas. Second, this study may be underpowered to assess heterogeneity between age, sex, or race/ethnicity specifically for incident MetS. Third, the direct measurements of insulin sensitivity at baseline (FSIGT) were not repeated at Visit 2, limiting our ability to isolate the impact of these factors from change over time in adiposity to a surrogate measure - HOMA-IR. Finally, the IRAS Family Study did not measure gluteofemoral or leg fat mass, limiting investigation of subcutaneous fat to the abdominal area.(35)
The primary strength of the present study is that data were derived from a large sample of African American and Hispanic American participants that has been comprehensively phenotyped using direct measures of insulin resistance, inflammatory markers, and adipose tissue distribution.(24) Specifically, body fat distribution, insulin sensitivity, and insulin secretion were measured using gold standard methods. Of note, the IRAS Family Study is one of only a few known studies to describe fat distribution using sensitive and specific imaging techniques in African American or Mexican American individuals. This rare high quality data allows our study to contribute to the handful of prior articles that have characterized metabolic phenotypes in an African American or Mexican American cohort,(7, 33, 36, 37, 38), an understudied area in these groups. Even fewer studies have measured body fat composition at more than one time-point, especially in diverse populations.(17) These strengths make this study one of only a few that have investigated the association between change over time in VAT and SAT or insulin resistance with incident MetS, especially in a diverse study population.
CONCLUSION
This study provides critical information towards understanding the heterogeneity of cardiometabolic risk that has been observed among individuals with obesity, and the role that VAT plays in the development of cardiometabolic risk. Levels of VAT and insulin resistance, were significantly different by MetS status, suggesting that VAT level and markers of metabolic function may differentiate between metabolic phenotypes, particularly in the context of obesity. The strong association of VAT and VAT change with incident MetS further supports the idea that VAT may be a useful tool to establish which individuals will benefit most from weight loss as primary prevention for cardiometabolic disease. The ability to accurately discriminate between groups at different levels of risk from obesity would improve prevention and treatment of obesity both clinically and at the population level.
Supplementary Material
What is already known about this subject?
Higher visceral fat is more strongly and specifically associated with traditional cardiometabolic risk factors and outcomes than body mass index or subcutaneous fat.
Evidence against the concept of metabolically healthy obesity is growing, contradicting the premise that the condition is stable over time and confers lower cardiometabolic risk in the long term.
What are the new findings in your manuscript?
Differing levels of visceral adipose tissue may be driving metabolic heterogeneity within body mass index categories, helping explain differences between metabolically healthy and unhealthy obesity.
Both overall and abdominal obesity (visceral adipose tissue) may play a role in the development of metabolic syndrome in African American and Mexican American populations.
Increase in visceral adipose tissue and insulin resistance over time contributed to additional metabolic syndrome risk after accounting for baseline levels and body mass index.
How might your results change the direction of research or the focus of clinical practice?
Consideration of body fat distribution in addition to body mass index is warranted for cardiometabolic risk stratification.
Acknowledgements:
The authors are grateful to the participants, staff, and investigators of the IRAS Family Study.
Funding:
This work was supported by R01HL60944.
Footnotes
Competing Interests Statement: The authors have nothing to disclose.
REFERENCES
- 1.Stefan N, Kantartzis K, Machann J, Schick F, Thamer C, Rittig K, et al. Identification and characterization of metabolically benign obesity in humans. Archives of internal medicine 2008;168: 1609–1616. [DOI] [PubMed] [Google Scholar]
- 2.Karelis AD. Metabolically healthy but obese individuals. Lancet (London, England) 2008;372: 1281–1283. [DOI] [PubMed] [Google Scholar]
- 3.Wildman RP. Healthy obesity. Current opinion in clinical nutrition and metabolic care 2009;12: 438–443. [DOI] [PubMed] [Google Scholar]
- 4.Muller MJ, Bosy-Westphal A, Heller M. 'Functional' body composition: differentiating between benign and non-benign obesity. F1000 biology reports 2009;1: 75. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Velho S, Paccaud F, Waeber G, Vollenweider P, Marques-Vidal P. Metabolically healthy obesity: different prevalences using different criteria. Eur J Clin Nutr 2010;64: 1043–1051. [DOI] [PubMed] [Google Scholar]
- 6.Bluher M The distinction of metabolically 'healthy' from 'unhealthy' obese individuals. Current opinion in lipidology 2010;21: 38–43. [DOI] [PubMed] [Google Scholar]
- 7.Wildman RP, Muntner P, Reynolds K, McGinn AP, Rajpathak S, Wylie-Rosett J, et al. The obese without cardiometabolic risk factor clustering and the normal weight with cardiometabolic risk factor clustering: prevalence and correlates of 2 phenotypes among the US population (NHANES 1999-2004). Archives of internal medicine 2008;168: 1617–1624. [DOI] [PubMed] [Google Scholar]
- 8.Camhi SM, Katzmarzyk PT. Differences in body composition between metabolically healthy obese and metabolically abnormal obese adults. International journal of obesity (2005) 2014;38: 1142–1145. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hwang Y-C, Hayashi T, Fujimoto WY, Kahn SE, Leonetti DL, McNeely MJ, et al. Visceral abdominal fat accumulation predicts the conversion of metabolically healthy obese subjects to an unhealthy phenotype. International Journal of Obesity 2015;39: 1365–1370. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Bi X, Seabolt L, Shibao C, Buchowski M, Kang H, Keil CD, et al. DXA-measured visceral adipose tissue predicts impaired glucose tolerance and metabolic syndrome in obese Caucasian and African-American women. Eur J Clin Nutr 2015;69: 329–336. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Gonçalves CG, Glade MJ, Meguid MM. Metabolically healthy obese individuals: Key protective factors. Nutrition 2016;32: 14–20. [DOI] [PubMed] [Google Scholar]
- 12.Brochu M, Tchernof A, Dionne IJ, Sites CK, Eltabbakh GH, Sims EA, et al. What are the physical characteristics associated with a normal metabolic profile despite a high level of obesity in postmenopausal women? The Journal of clinical endocrinology and metabolism 2001;86: 1020–1025. [DOI] [PubMed] [Google Scholar]
- 13.Shah RV, Allison MA, Lima JA, Abbasi SA, Eisman A, Lai C, et al. Abdominal fat radiodensity, quantity and cardiometabolic risk: The Multi-Ethnic Study of Atherosclerosis. Nutrition, metabolism, and cardiovascular diseases : NMCD 2016;26: 114–122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Onat A, Uğur M, Can G, Yüksel H, Hergenç G. Visceral adipose tissue and body fat mass: predictive values for and role of gender in cardiometabolic risk among Turks. Nutrition 2010;26: 382–389. [DOI] [PubMed] [Google Scholar]
- 15.Lee YH, Park J, Min S, Kang O, Kwon H, Oh SW. Impact of Visceral Obesity on the Risk of Incident Metabolic Syndrome in Metabolically Healthy Normal Weight and Overweight Groups: A Longitudinal Cohort Study in Korea. Korean journal of family medicine 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Rosenquist KJ, Pedley A, Massaro JM, Therkelsen KE, Murabito JM, Hoffmann U, et al. Visceral and subcutaneous fat quality and cardiometabolic risk. JACC Cardiovasc Imaging 2013;6: 762–771. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Shah RV, Murthy VL, Abbasi SA, Blankstein R, Kwong RY, Goldfine AB, et al. Visceral adiposity and the risk of metabolic syndrome across body mass index: the MESA Study. JACC Cardiovascular imaging 2014;7: 1221–1235. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Karelis AD, Faraj M, Bastard JP, St-Pierre DH, Brochu M, Prud'homme D, et al. The metabolically healthy but obese individual presents a favorable inflammation profile. The Journal of clinical endocrinology and metabolism 2005;90: 4145–4150. [DOI] [PubMed] [Google Scholar]
- 19.Phillips CM, Perry IJ. Does inflammation determine metabolic health status in obese and nonobese adults? The Journal of clinical endocrinology and metabolism 2013;98: E1610–1619. [DOI] [PubMed] [Google Scholar]
- 20.Teixeira TF, Alves RD, Moreira AP, Peluzio Mdo C. Main characteristics of metabolically obese normal weight and metabolically healthy obese phenotypes. Nutrition reviews 2015;73: 175–190. [DOI] [PubMed] [Google Scholar]
- 21.Flegal KM, Kruszon-Moran D, Carroll MD, Fryar CD, Ogden CL. Trends in Obesity Among Adults in the United States, 2005 to 2014. Jama 2016;315: 2284–2291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Henkin L, Bergman RN, Bowden DW, Ellsworth DL, Haffner SM, Langefeld CD, et al. Genetic epidemiology of insulin resistance and visceral adiposity. The IRAS Family Study design and methods. Annals of epidemiology 2003;13: 211–217. [DOI] [PubMed] [Google Scholar]
- 23.Wagenknecht LE, Mayer EJ, Rewers M, Haffner S, Selby J, Borok GM, et al. The insulin resistance atherosclerosis study (IRAS) objectives, design, and recruitment results. Annals of epidemiology 1995;5: 464–472. [DOI] [PubMed] [Google Scholar]
- 24.Wagenknecht LE, Scherzinger AL, Stamm ER, Hanley AJ, Norris JM, Chen YD, et al. Correlates and heritability of nonalcoholic fatty liver disease in a minority cohort. Obesity (Silver Spring, Md) 2009;17: 1240–1246. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Alberti KG, Eckel RH, Grundy SM, Zimmet PZ, Cleeman JI, Donato KA, et al. Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation 2009;120: 1640–1645. [DOI] [PubMed] [Google Scholar]
- 26.Bergman RN, Finegood DT, Ader M. Assessment of insulin sensitivity in vivo. Endocrine reviews 1985;6: 45–86. [DOI] [PubMed] [Google Scholar]
- 27.Pacini G, Bergman RN. MINMOD: a computer program to calculate insulin sensitivity and pancreatic responsivity from the frequently sampled intravenous glucose tolerance test. Computer methods and programs in biomedicine 1986;23: 113–122. [DOI] [PubMed] [Google Scholar]
- 28.Ibrahim MM. Subcutaneous and visceral adipose tissue: structural and functional differences. Obes Rev 2010;11: 11–18. [DOI] [PubMed] [Google Scholar]
- 29.Matsuzawa Y, Shimomura I, Nakamura T, Keno Y, Kotani K, Tokunaga K. Pathophysiology and pathogenesis of visceral fat obesity. Obes Res 1995;3 Suppl 2: 187S–194S. [DOI] [PubMed] [Google Scholar]
- 30.Britton KA, Massaro JM, Murabito JM, Kreger BE, Hoffmann U, Fox CS. Body Fat Distribution, Incident Cardiovascular Disease, Cancer, and All-Cause Mortality. Journal of the American College of Cardiology 2013;62: 921–925. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Tomiyama AJ, Hunger JM, Nguyen-Cuu J, Wells C. Misclassification of cardiometabolic health when using body mass index categories in NHANES 2005-2012. International journal of obesity (2005) 2016;40: 883–886. [DOI] [PubMed] [Google Scholar]
- 32.Pi-Sunyer X The medical risks of obesity. Postgraduate medicine 2009;121: 21–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Gaillard TR, Schuster D, Osei K. Characterization of metabolically unhealthy overweight/obese African American women: significance of insulin-sensitive and insulin-resistant phenotypes. Journal of the National Medical Association 2012;104: 164–171. [DOI] [PubMed] [Google Scholar]
- 34.Samaropoulos XF, Hairston KG, Anderson A, Haffner SM, Lorenzo C, Montez M, et al. A metabolically healthy obese phenotype in hispanic participants in the IRAS family study. Obesity (Silver Spring, Md) 2013;21: 2303–2309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Stefan N, Schick F, Häring H-U. Causes, Characteristics, and Consequences of Metabolically Unhealthy Normal Weight in Humans. Cell Metabolism 2017;26: 292–300. [DOI] [PubMed] [Google Scholar]
- 36.Cherqaoui R, Kassim TA, Kwagyan J, Freeman C, Nunlee-Bland G, Ketete M, et al. The metabolically healthy but obese phenotype in African Americans. Journal of clinical hypertension (Greenwich, Conn) 2012;14: 92–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Doumatey AP, Bentley AR, Zhou J, Huang H, Adeyemo A, Rotimi CN. Paradoxical Hyperadiponectinemia is Associated With the Metabolically Healthy Obese (MHO) Phenotype in African Americans. Journal of endocrinology and metabolism 2012;2: 51–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Park YW, Zhu S, Palaniappan L, Heshka S, Carnethon MR, Heymsfield SB. The metabolic syndrome: prevalence and associated risk factor findings in the US population from the Third National Health and Nutrition Examination Survey, 1988-1994. Archives of internal medicine 2003;163: 427–436. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.