Abstract
Background: Limited data suggest that the effects of abdominal subcutaneous adipose tissue (SAT) on cardiovascular disease risk may depend on accompanying amounts of abdominal visceral adipose tissue (VAT).
Objective: The objective was to examine whether abdominal VAT area modifies the effects of abdominal SAT area on subclinical atherosclerosis and cardiometabolic risk factors in both whites and African Americans.
Design: Computed tomographic measures of abdominal SAT and VAT were examined in relation to carotid intima-media thickness (cIMT) and cardiometabolic risk factor levels in 500 African American and white women in midlife. A VAT × SAT interaction term was evaluated.
Results: The mean (±SD) age of the sample was 51.0 ± 2.9 y, and 37% were African American. Higher amounts of SAT and VAT were associated with higher cIMT, blood pressure, homeostasis model assessment insulin resistance index (HOMA-IR), and concentrations of glucose, triglycerides, and insulin and with lower concentrations of HDL cholesterol. However, in African Americans, but not in whites, higher amounts of VAT significantly attenuated associations between higher amounts of SAT and higher insulin concentrations (P for interaction = 0.032) and HOMA-IR (P for interaction = 0.011) and reversed associations with cIMT (P for interaction = 0.005) and glucose (P for interaction = 0.044).
Conclusions: These results suggest that in midlife African American but not white women, adverse associations between abdominal SAT and cardiometabolic risk factors are attenuated and, in the case of subclinical atherosclerosis, are reversed as VAT amounts increase. Given that African American women suffer disproportionately from obesity and cardiovascular disease, further research into the role of this effect modification on obesity-associated vascular disease in African American women is warranted.
INTRODUCTION
Visceral adipose tissue (VAT) accumulation is clearly associated with increased risk of type 2 diabetes and cardiovascular disease and is positively associated with cardiometabolic risk factors such as blood pressure, lipids, and measures of glucose homeostasis. However, there has been much debate regarding the role of subcutaneous adipose tissue (SAT). It is unclear if SAT is risk enhancing or, rather, protective for cardiovascular disease. A recent study in middle-aged men and women found that higher amounts of abdominal SAT are associated with lower levels of subclinical atherosclerosis (1), which echoes results of earlier studies of SAT's effects on cardiovascular variables (2–7). However, other studies, including the Framingham Heart Study, have shown that the strength and direction of association between abdominal SAT and risk factors or cardiovascular disease are nearly identical to those for abdominal VAT (8–11). A recent study by Demerath et al (12) may help to explain these discrepant findings. In assessing the relation between abdominal SAT and the prevalence of the metabolic syndrome among a wide age range of non-Hispanic whites, a highly significant interaction was found between abdominal SAT and VAT (12). At low amounts of VAT, more SAT was associated with a higher prevalence of metabolic syndrome, whereas at high amounts of VAT, more SAT was associated with a lower prevalence of the metabolic syndrome (12). These results suggest that the discrepant findings in the literature may be due to differences in VAT amounts between participants in the previous studies.
Few studies have examined whether the effects of SAT on cardiometabolic risk factors differ according to the accompanying amount of VAT (3, 12, 13), and to our knowledge none have examined whether this potential effect is seen on measures of atherosclerosis. In addition, despite the fact that African American women have higher amounts of abdominal SAT and lower amounts of abdominal VAT compared with white women (14–16), and that African Americans have a lower prevalence of dyslipidemia (17, 18), potential differences in these relations between African Americans and whites have not yet been examined. Therefore, the purpose of the present analyses was 2-fold: 1) to examine possible effect modification between abdominal SAT and VAT on levels of subclinical atherosclerosis and cardiometabolic risk factors in midlife women participating in the Study of Women's Health Across the Nation (SWAN) Heart ancillary study and 2) to examine whether effect modification is observed in both African American and white women.
SUBJECTS AND METHODS
The current study included participants from the SWAN. The SWAN is a multicenter, multiethnic, longitudinal study designed to characterize the biological and psychosocial changes that occur during the menopausal transition in a community-based sample. Details of the study design and recruitment have been previously published (19). Briefly, SWAN is being conducted at 7 sites: Boston, MA; Chicago, IL; the Detroit area, MI; Los Angeles, CA; Newark, NJ; Pittsburgh, PA; and Oakland, CA. A total of 3302 women aged 42–52 y were enrolled from 1996 to 1997.
The current data were derived from the SWAN Heart Study, an ancillary study to SWAN, which was designed to characterize the natural history of subclinical atherosclerosis during the menopausal transition. SWAN Heart was conducted at the Pittsburgh and Chicago SWAN sites and was initiated ≈4 y after the SWAN baseline enrollment, with enrollment into SWAN Heart occurring across calendar years 2000–2005. To be eligible for SWAN Heart, participants had to have undergone a carotid ultrasound scan at a previous SWAN visit or, if they had not, were required to meet the following criteria: have at least one intact ovary and an intact uterus, could not have evidence of clinical atherosclerosis (myocardial infarction, angina, intermittent claudication, cerebral ischemia, or revascularization), and could not report use of menopausal hormone therapy (MHT) within the prior 3 mo or current use of antihypertensive or diabetes medications. For those individuals who had undergone carotid scanning before the initiation of SWAN Heart, there were 2 women who experienced clinical atherosclerosis, 79 who initiated antihypertensive medications, 75 women who initiated MHT, and 5 who initiated diabetes medications in the intervening time period. A total of 608 women (259 from the University of Pittsburgh and 349 from Rush University in Chicago) were enrolled in SWAN Heart. For these analyses, the 2 women who experienced clinical atherosclerosis were excluded, whereas those initiating antihypertensive or diabetes medications were retained. Of the remaining 606 women, 49 were excluded because of incomplete VAT or SAT measurements, 56 were excluded because of missing data on cardiometabolic outcomes, and one woman was excluded because of underweight [body mass index (BMI; in kg/m2) <18.5], which left 500 women for analysis.
The institutional review boards of the participating institutions approved this study, and all women signed informed consent at all SWAN and SWAN Heart visits.
Carotid artery intima-media thickness measurement
Carotid artery intima-media thickness (cIMT) was assessed by using a Toshiba SSA-270A scanner (Toshiba American Medical Systems, Tustin, CA) at the Pittsburgh site and a Hewlett-Packard 5500 scanner (Hewlett-Packard, Andover, MA) at the Chicago site. The duplex scanners for the Chicago and Pittsburgh sites were comparable in image quality. B-mode images were obtained from the following 4 locations in both the left and right carotid arteries: the near and far walls of the distal common carotid artery, 1 cm proximal to the carotid bulb; the far walls of the carotid bulb, from the point where the near and far walls are no longer parallel and extending down to the flow divider; and the internal carotid artery from the flow divider to 1 cm distal from the flow divider. cIMT measurements were performed by electronically tracing the lumen-intima interface and the media-adventitia interface across a 1-cm segment of each of these locations. A computer-assisted measurement for each pixel over this area was generated for a total of ≈140 data points in each location. The average reading in each segment was calculated, and then these readings were averaged across all segments to obtain the cIMT value used in analysis. All readings were conducted at the University of Pittsburgh Ultrasound Research Laboratory. An intraclass correlation of 0.98 was yielded on replicate readings of 20 cIMT scans.
Abdominal VAT and SAT measurement
Abdominal adipose tissue area was measured by using an electron beam computed tomographic scan, as described elsewhere (11). Briefly, a 6-mm transverse image was obtained between L4 and L5 with a c-150 Ultrafast CT Scanner (GE Imatron, San Francisco, CA). Scans were read by a single reader at the University of Pittsburgh. A pixel range of −30 to −190 Hounsfield units was used to define fat. The area of adipose tissue was defined by using image analysis (AccuImage Diagnostics, South San Francisco, CA). A region-of-interest line was drawn at the interior of abdominal musculature along the fascial plane. Fat within this area was considered to be visceral fat area. Subcutaneous fat was calculated as the difference between the whole image and visceral fat area. Interobserver reliability was determined by repeat reads on 10 scans, with intraclass coefficient values of 0.97 and 0.94 for total and visceral fat area, respectively.
Questionnaire data
Race, current smoking habits, and educational status were obtained from a self-reported questionnaire. Women were also asked about their menstrual bleeding patterns in the 12 mo before recruitment, which were divided into the following categories: 1) premenopausal (menstrual periods in the past 3 mo with no cycle irregularity in the past 12 mo), 2) early perimenopausal (menstrual periods in the past 3 mo with some cycle irregularity in the past 12 mo), 3) late perimenopausal (>3 and <12 mo amenorrhea), 4) postmenopausal (≥12 mo amenorrhea), and 5) indeterminate (use of MHT in the previous 12 mo, before natural cessation of menses). Menopausal status categories were collapsed into 3 categories for data analysis as pre- or early perimenopause, late peri- or postmenopause or surgical menopause, and indeterminate. MHT use was ascertained by self-reported use of birth control pills, estrogen pills, estrogen injection or patch, combination estrogen and progestin, or progestin pills. As noted earlier, among the limited number of women at the Pittsburgh site who were enrolled into SWAN Heart at the SWAN baseline visit, 75 initiated MHT use between their initial SWAN Heart visit and their cIMT measurement.
Cardiometabolic risk factors
Blood biochemical assays
Fasting blood samples were assayed at Medical Research Laboratories (Lexington, KY), which is certified by the National Heart, Lung, and Blood Institute, Centers for Disease Control and Prevention Part II program, as previously described (12). Serum total cholesterol, HDL cholesterol, and triglycerides were measured directly; and LDL cholesterol was calculated by using the Friedewald equation (13), excluding women with concentrations of triglycerides ≥400 mg/dL. The homeostasis model assessment insulin resistance index (HOMA-IR) was calculated from fasting insulin and glucose as (fasting insulin × fasting glucose)/22.5 (14).
Physical measures
Blood pressure was measured in the right arm with the participant seated, after ≥5 min of rest. Two sequential blood pressure values were obtained and averaged. Height and weight were measured in participants while wearing light clothing and without shoes. BMI was calculated as weight in kilograms divided by height in meters squared. Waist circumference was measured with the participant wearing nonrestrictive undergarments, at the level of the natural waist, defined as the narrowest part of the torso as seen from the anterior aspect. For cases in which waist narrowing was difficult to determine, the measure was taken at the smallest horizontal circumference in the area between the ribs and the iliac crest.
Statistical methods
We compared demographic data, cardiometabolic risk factors, and cIMT levels across tertiles of VAT, and then again across tertiles of SAT, by using the Cochran-Armitage test for trend for categorical variables and a test of linear trend for continuous variables. We used regression models to assess linear trend across tertiles of SAT by treating a variable composed of the median SAT value in each tertile as a continuous variable to account for unequal spacing across tertiles. Linear trend across VAT tertiles was assessed in the same way. We used linear regression to assess the associations between VAT and cardiometabolic risk factors and IMT, with adjustment for age, race-ethnicity, menopausal status, smoking status, and study site; we then assessed the associations between SAT and these factors. To enable direct comparison of the magnitude of β coefficients associated with VAT compared with SAT, units of SD were used for VAT and SAT such that resulting β estimates represented the increase in the outcome of interest per SD increment in VAT or SAT. Model residuals were plotted for all linear regression models for assessment of normality. Model residuals were not normally distributed for glucose, insulin, and HOMA-IR; and these were log transformed for regression analyses. We then used linear regression to model the interaction between VAT and SAT for each cardiometabolic risk factor and cIMT, with adjustment for the covariates listed above. Independent variables included continuous main effects for VAT and SAT measures and a multiplicative VAT × SAT interaction term. The correlation between VAT and SAT was only 0.59, which reduced multicollinearity concerns. However, to confirm this, variance inflation factors were examined for all models and also indicated no concerns related to multicollinearity. Additional models were used that included a multiplicative VAT × SAT × race interaction term to assess consistency of effect modification across the 2 race-ethnic groups. In cases in which interaction terms were statistically significant, to enable illustration of the nature of the interaction, partial Pearson's correlation coefficients between SAT and outcomes were calculated within race-specific tertiles of VAT and again within race-specific quintiles of VAT. We performed additional analyses excluding women taking MHT or with surgical menopause (n = 67). Significance was set at P < 0.05 for main effects and P < 0.10 for interaction terms. All statistical analyses were carried out by using SAS 9.2 (SAS Institute Inc, Cary, NC).
RESULTS
The mean (±SD) age of the population was 51.0 ± 2.9 y, and 37% were African American. Characteristics of the study population stratified by abdominal VAT and then SAT are presented in Tables 1 and 2. Those with higher VAT amounts were older and more likely to be late peri- or postmenopausal, and those with higher VAT or SAT had adverse cardiometabolic risk factor levels compared with those with lower VAT or SAT areas. In addition, those with higher SAT amounts were more likely to be African American, whereas no discernible pattern in race-ethnicity was observed across VAT area.
TABLE 1.
Characteristics of the study sample by abdominal visceral adipose tissue (VAT) tertiles1
| VAT tertile |
||||
| ≤84.2 cm2 (n = 166) | 84.3–142.0 cm2 (n = 167) | ≥142.1 cm2 (n = 167) | P for trend | |
| Age (y) | 50.5 ± 2.72 | 51.0 ± 3.0 | 51.6 ± 3.0 | <0.001 |
| Black [% (n)] | 34.3 (57) | 43.1 (72) | 34.1 (57) | 0.966 |
| Education [% (n)] | ||||
| High school degree or less | 16.2 (26) | 15.2 (25) | 13.9 (22) | |
| Post high school/some college | 28.0 (45) | 28.1 (46) | 37.3 (59) | 0.208 |
| College degree or higher | 55.9 (90) | 56.7 (93) | 48.7 (77) | 0.964 |
| Current smokers [% (n)] | 16.3 (27) | 19.8 (33) | 12.6 (21) | 0.359 |
| Menopause status [% (n)] | ||||
| Premenopausal or early perimenopausal | 62.1 (103) | 49.7 (83) | 48.5 (81) | |
| Late perimenopausal, postmenopausal, or surgery | 33.1 (55) | 46.1 (77) | 46.1 (77) | 0.013 |
| Indeterminate | 4.8 (8) | 4.2 (7) | 5.4 (9) | 0.482 |
| Menopausal hormone therapy users [% (n)] | 13.3 (22) | 12.6 (21) | 10.8 (18) | 0.490 |
| Systolic blood pressure (mm Hg) | 112.8 ± 14.4 | 120.7 ± 15.2 | 124.4 ± 18.4 | <0.001 |
| HDL cholesterol (mg/dL) | 64.3 ± 14.6 | 57.6 ± 14.1 | 51.2 ± 11.8 | <0.001 |
| Triglycerides (mg/dL) | 83.0 (68.0–97.0)3 | 102.0 (71.0–131.0) | 132.0 (98.0–190.0) | <0.001 |
| Glucose (mg/dL) | 84.0 (79.0–88.0) | 88.0 (84.0–95.0) | 95.0 (87.0–103.0) | <0.001 |
| Insulin (μIU/mL) | 7.1 (5.7–8.4) | 9.1 (7.1–13.1) | 13.5 (9.5–19.2) | <0.001 |
| HOMA-IR | 1.4 (1.2–1.8) | 2.0 (1.5–3.0) | 3.1 (2.1–4.8) | <0.001 |
| BMI (kg/m2) | 24.5 ± 3.4 | 28.2 ± 4.2 | 34.8 ± 6.1 | <0.001 |
| Waist circumference (cm) | 77.0 ± 7.9 | 86.8 ± 7.9 | 102.8 ± 12.5 | <0.001 |
| SAT (cm2) | 229.1 ± 113.0 | 324.1 ± 116.5 | 452.7 ± 138.2 | <0.001 |
| cIMT (mm) | 0.643 ± 0.073 | 0.677 ± 0.095 | 0.690 ± 0.108 | <0.001 |
HOMA-IR, homeostasis model assessment insulin resistance index; SAT, subcutaneous adipose tissue; cIMT, carotid intima-media thickness.
Mean ± SD (all such values).
Median; interquartile range in parentheses (all such values).
TABLE 2.
Characteristics of the study sample by abdominal subcutaneous adipose tissue (SAT) tertiles1
| SAT tertile |
||||
| ≤250.8 cm2 (n = 166) | 250.9–391.1 cm2 (n = 167) | ≥391.2 cm2 (n = 167) | P for trend | |
| Age (y) | 50.7 ± 2.82 | 51.0 ± 3.0 | 51.3 ± 2.9 | 0.060 |
| Black [% (n)] | 28.3 (47) | 39.5 (66) | 43.7 (73) | 0.004 |
| Education [% (n)] | ||||
| High school degree or less | 17.4 (28) | 9.9 (16) | 18.0 (29) | |
| Post high school or some college | 23.6 (38) | 29.2 (47) | 40.4 (65) | 0.164 |
| College degree or higher | 59.0 (95) | 60.9 (98) | 41.6 (67) | 0.256 |
| Current smokers [% (n)] | 18.1 (30) | 21.0 (35) | 9.6 (16) | 0.035 |
| Menopause status [% (n)] | ||||
| Premenopausal or early perimenopausal | 57.2 (95) | 52.1 (87) | 50.9 (85) | |
| Late perimenopausal, postmenopausal, or surgery | 37.4 (62) | 43.1 (72) | 44.9 (75) | 0.186 |
| Indeterminate | 5.4 (9) | 4.8 (8) | 4.2 (7) | 0.793 |
| Menopausal hormone therapy users [% (n)] | 13.3 (22) | 13.2 (22) | 10.2 (17) | 0.391 |
| Systolic blood pressure (mm Hg) | 112.3 ± 14.0 | 120.9 ± 16.9 | 124.7 ± 16.9 | <0.001 |
| HDL cholesterol (mg/dL) | 62.5 ± 17.3 | 57.4 ± 12.6 | 53.1 ± 11.7 | <0.001 |
| Triglycerides (mg/dL) | 87.0 (68.0–111.0)3 | 98.0 (75.0–137.0) | 116.0 (83.0–156.0) | <0.001 |
| Glucose (mg/dL) | 86.0 (81.0–92.0) | 87.0 (82.0–95.0) | 93.0 (86.0–99.0) | <0.001 |
| Insulin (μIU/mL) | 7.4 (5.9–9.7) | 9.1 (6.8–12.9) | 12.6 (8.4–19.2) | <0.001 |
| HOMA-IR | 1.5 (1.2–2.0) | 1.8 (1.0–4.8) | 2.8 (1.8–4.8) | <0.001 |
| BMI (kg/m2) | 23.9 ± 3.0 | 28.3 ± 3.8 | 35.3 ± 5.6 | <0.001 |
| Waist circumference (cm) | 77.6 ± 8.7 | 87.4 ± 9.4 | 101.6 ± 12.9 | <0.001 |
| VAT (cm2) | 81.9 ± 46.8 | 115.8 ± 54.9 | 168.7 ± 61.4 | <0.001 |
| cIMT (mm) | 0.642 ± 0.076 | 0.677 ± 0.098 | 0.692 ± 0.103 | <0.001 |
HOMA-IR, homeostasis model assessment insulin resistance index; VAT, visceral adipose tissue; cIMT, carotid intima-media thickness.
Mean ± SD (all such values).
Median; interquartile range in parentheses (all such values).
When modeled separately, higher VAT area was associated with adverse cardiometabolic risk factor levels and higher cIMT, in both whites and African Americans, whereas higher SAT area was associated with adverse cardiometabolic risk factor levels in both whites and African Americans but with higher cIMT only in whites (Table 3). However, effect modification between SAT and VAT was observed for associations with cIMT and HDL cholesterol (Table 4; P value for interaction = 0.033 for both) in analyses in which the race-ethnic groups were pooled. For HDL cholesterol, the negative association observed between SAT and HDL concentrations at lower VAT areas was attenuated as the amount of VAT increased; this finding was supported by the positive sign of the interaction term as well as adjusted correlation coefficients between SAT and HDL across tertiles of VAT (adjusted ρ = −0.10 in the first VAT tertile and 0.02 in the third VAT quartile).
TABLE 3.
Multivariable-adjusted β estimates representing increase in carotid intima-media thickness (cIMT) and cardiometabolic risk factors per SD increment in abdominal visceral adipose tissue (VAT) or subcutaneous adipose tissue (SAT) in all women and in women stratified by race1
| VAT as abdominal adiposity exposure |
SAT as abdominal adiposity exposure |
|||||||||||
| All |
Whites |
African Americans |
All |
Whites |
African Americans |
|||||||
| β (SE) | P value | β (SE) | P value | β (SE) | P value | β (SE) | P value | β (SE) | P value | β (SE) | P value | |
| cIMT (mm) | 0.020 (0.004) | <0.001 | 0.019 (0.005) | <0.001 | 0.020 (0.009) | 0.022 | 0.014 (0.004) | 0.001 | 0.018 (0.005) | 0.001 | 0.008 (0.007) | 0.28 |
| Systolic blood pressure (mm Hg) | 4.1 (0.7) | <0.001 | 4.4 (0.7) | <0.001 | 3.1 (1.6) | 0.05 | 4.2 (0.7) | <0.001 | 4.9 (0.8) | <0.001 | 2.9 (1.3) | 0.022 |
| HDL cholesterol (mg/dL) | −6.2 (0.6) | <0.001 | −6.3 (0.7) | <0.001 | −5.6 (1.4) | <0.001 | −4.2 (0.7) | <0.001 | −4.8 (0.8) | <0.001 | −3.3 (1.1) | 0.004 |
| Log triglycerides (mg/dL) | 0.23 (0.02) | <0.001 | 0.23 (0.02) | <0.001 | 0.21 (0.04) | <0.001 | 0.13 (0.02) | <0.001 | 0.16 (0.03) | <0.001 | 0.09 (0.03) | 0.007 |
| Log glucose (mg/dL) | 0.06 (0.01) | <0.001 | 0.06 (0.01) | <0.001 | 0.06 (0.02) | <0.001 | 0.04 (0.01) | <0.001 | 0.05 (0.01) | <0.001 | 0.03 (0.01) | 0.042 |
| Log insulin (μIU/mL) | 0.28 (0.02) | <0.001 | 0.25 (0.02) | <0.001 | 0.36 (0.05) | <0.001 | 0.21 (0.02) | <0.001 | 0.19 (0.03) | <0.001 | 0.24 (0.04) | <0.001 |
| Log HOMA-IR | 0.34 (0.02) | <0.001 | 0.32 (0.02) | <0.001 | 0.42 (0.05) | <0.001 | 0.24 (0.03) | <0.001 | 0.23 (0.03) | <0.001 | 0.27 (0.05) | <0.001 |
VAT and SAT are considered in separate models. HOMA-IR, homeostasis model assessment insulin resistance index. Estimates were adjusted for age, race, menopause status, menopausal hormone use, educational level, smoking status, and study site. β Estimates are per SD of VAT (65.2 cm2) and SAT (153.3 cm2).
TABLE 4.
Multivariable-adjusted P values for visceral adipose tissue (VAT) × subcutaneous adipose tissue (SAT) interactions for associations with carotid intima-media thickness (cIMT) and cardiometabolic risk factor outcomes in all women and in women stratified by race-ethnicity1
| VAT × SAT interaction terms |
|||||||
| All women |
Whites |
African Americans |
|||||
| Dependent variables | β (SE) | P value | β (SE) | P value | β (SE) | P value | VAT × SAT × race interaction term P value |
| cIMT (mm) | −0.009 (0.004) | 0.033 | −0.005 (0.005) | 0.26 | −0.026 (0.009) | 0.005 | 0.0692 |
| Systolic blood pressure (mm Hg) | −0.23 (0.67) | 0.73 | −0.3 (0.7) | 0.72 | −0.7 (1.7) | 0.67 | 0.84 |
| HDL cholesterol (mg/dL) | 1.34 (0.63) | 0.033 | 0.8 (0.7) | 0.25 | 2.6 (1.5) | 0.084 | 0.16 |
| Log triglycerides (mg/dL) | −0.03 (0.02) | 0.10 | −0.2 (3.4) | 0.96 | −8.8 (5.1) | 0.087 | 0.41 |
| Log glucose (mg/dL) | 0.002 (0.007) | 0.73 | 0.01 (0.01) | 0.13 | −0.04 (0.02) | 0.044 | 0.0282 |
| Log insulin (μIU/mL) | −0.02 (0.02) | 0.33 | −0.004 (0.02) | 0.85 | −0.12 (0.05) | 0.032 | 0.0652 |
| Log HOMA-IR | −0.02 (0.02) | 0.46 | 0.008 (0.03) | 0.76 | −0.15 (0.06) | 0.011 | 0.0242 |
HOMA-IR, homeostasis model assessment insulin resistance index. Estimates were adjusted for age, race, menopause status, menopausal hormone use, educational level, smoking status, study site, and main effects.
Significant interaction term (P < 0.10).
For cIMT, modification of SAT effects by VAT was limited to African American women, as indicated by a significant P value for the VAT × SAT × race interaction term (P = 0.069). Similarly, in African Americans but not whites, significant effect modification of SAT by VAT was found for SAT associations with glucose, insulin, and HOMA-IR (Table 4). Similar results were suggested for HDL cholesterol and triglycerides, although the nonsignificant VAT × SAT × race interaction terms for these variables suggested no race difference in the pattern of effect modification of SAT by VAT. The negative sign associated with the interaction terms in African Americans suggests that the positive association between SAT and cIMT and glucose homeostasis outcomes in Table 2 was attenuated, or even reversed, as VAT area increased. To better show this, we present partial Pearson's correlation coefficients, adjusted for important covariates, between SAT and cIMT, fasting glucose, and HOMA-IR across tertiles of VAT in Table 5. Although power was insufficient within many tertiles to obtain statistical significance for individual correlation coefficients, the pattern of association between SAT and outcomes as VAT amounts increased can be seen. In African Americans only, in the lowest VAT tertile higher SAT was associated with higher cIMT and log glucose concentrations (adjusted ρ = 0.28 and 0.29, respectively), whereas in the highest VAT tertile higher SAT amounts were associated with lower cIMT and log glucose concentrations (adjusted ρ = −0.11 and −0.08, respectively). Higher SAT area was associated with higher HOMA-IR values in all 3 VAT tertiles in both whites and African Americans. However, this association was considerably less pronounced in the middle and highest VAT tertiles relative to that in the lowest VAT tertile in African Americans. Results were similar when quintiles rather than tertiles of VAT were used. In fact, SAT was negatively associated with HOMA-IR in the highest VAT quintile (adjusted ρ = −0.32), suggesting that similar to the cIMT and glucose findings, SAT associations with HOMA-IR may actually be reversed at high VAT amounts. Similar patterns were found for insulin.
TABLE 5.
Partial Pearson's correlation coefficients (ρ) between abdominal subcutaneous adipose tissue and carotid intima-media thickness (cIMT) and markers of glucose homeostasis by visceral adipose tissue (VAT) tertiles1
| Whites |
African Americans |
|||||
| n | ρ | 95% CI | n | ρ | 95% CI | |
| cIMT | ||||||
| First VAT tertile | 103 | 0.15 | −0.05, 0.34 | 104 | 0.28 | 0.00, 0.52 |
| Second VAT tertile | 103 | 0.15 | −0.05, 0.34 | 104 | −0.15 | −0.41, 0.12 |
| Third VAT tertile | 98 | 0.04 | −0.17, 0.24 | 100 | −0.11 | −0.38, 0.18 |
| Log glucose | ||||||
| First VAT tertile | 104 | −0.12 | −0.31, 0.08 | 57 | 0.29 | 0.01, 0.52 |
| Second VAT tertile | 104 | 0.24 | 0.05, 0.42 | 60 | 0.00 | −0.26, 0.27 |
| Third VAT tertile | 100 | 0.13 | −0.07, 0.33 | 58 | −0.08 | −0.35, 0.20 |
| Log HOMA-IR | ||||||
| First VAT tertile | 56 | 0.22 | 0.02, 0.40 | 57 | 0.28 | 0.01, 0.52 |
| Second VAT tertile | 59 | 0.07 | −0.13, 0.26 | 60 | 0.10 | −0.17, 0.36 |
| Third VAT tertile | 55 | 0.20 | 0.00, 0.39 | 58 | 0.06 | −0.21, 0.33 |
HOMA-IR, homeostasis model assessment insulin resistance index. Values were adjusted for age, menopause status, menopausal hormone use, educational level, smoking status, and study site. VAT tertiles were race-specific as follows—whites: 21.6–82.4, 82.5–143.7, and 143.8–382.8 cm2; blacks: 27.5–86.5, 86.6–135.6, and 135.7–336.5 cm2.
Analyses similar to those depicted in Table 5 were carried out to assess the interaction in the opposite framework: that of SAT modifying the effects of VAT. For this, we assessed the association of VAT with cIMT, stratified by SAT tertiles. In African Americans, the adverse associations between VAT and cIMT progressively lessened across tertiles of SAT. When MHT users and women with surgical menopause were excluded from analyses (n = 67), results were similar (data not shown). Similarly, when diabetes medication users (n = 5) and antihypertensive medication users (n = 79) were excluded, results were similar.
DISCUSSION
In this study, higher abdominal SAT amounts were associated with a greater burden of subclinical atherosclerosis and with a worse cardiometabolic risk factor burden in midlife women. However, in African American women, adverse associations between SAT and subclinical atherosclerosis and measures of glucose homeostasis were attenuated and perhaps even reversed when in the presence of higher VAT amounts, such that at the highest VAT amounts, higher SAT amounts were associated with lower cIMT and fasting glucose.
Few previous studies have assessed potential interaction between SAT and VAT. In the study by Demerath et al (12), at lower amounts of VAT, the probability of having the metabolic syndrome increased across increasing SAT tertiles, whereas at higher amounts of VAT, the probability of having the metabolic syndrome decreased across increasing SAT tertiles in men (P value for interaction = 0.0008). No effect modification was seen among women (12). However, the sample was restricted to non-Hispanic whites, and in the current study effect modification was limited to African American women. Similar potential effect modification was seen in relation to lipid concentrations in men and lipid concentrations and cardiovascular disease in women in the Framingham Heart Study (13). In contrast to our current results for glucose concentrations, in the Framingham Heart Study the prevalence of impaired fasting glucose was higher in those with higher SAT in both the low and high VAT groups (13). However, the Framingham Heart Study includes primarily white participants, whereas our glucose results were limited to African Americans.
The mechanism that may underlie the potential beneficial effects of SAT in the presence of high amounts of VAT is not clear. As adipose tissue depots, both VAT and SAT serve the dual function of storing energy in the form of triglycerides and releasing that energy in the form of free fatty acids. SAT has been shown to display less insulin- and adrenergic-related lipolytic activity and to release fewer inflammatory adipokines and appears to have greater adiponectin mRNA expression compared with VAT (20–23). Therefore, although SAT is present in larger quantity than is VAT, in individuals with excess adipose tissue preferential storage of energy in SAT compared with VAT is likely less pathogenic than preferential storage in VAT. In patients with lipodystrophy, a reduction in SAT results in visceral fat accumulation, hepatic steatosis, insulin resistance, and increased coronary atherosclerosis (24), giving rise to the hypothesis that once the storage capacity of SAT has been exceeded, triglycerides are stored in VAT and nonadipose tissues, such as the liver and muscle. Ectopic fat may have greater lipotoxic effects and has been shown to be associated with adverse cardiometabolic risk factor levels and atherosclerosis (25–28). Therefore, in obese individuals with high amounts of both VAT and SAT, SAT may act as a “metabolic sink,” sequestering circulating triglycerides with minimal release of inflammatory cytokines and with greater resistance to lipolytic signals than VAT. In these individuals with high amounts of VAT, although SAT does have lipolytic and adipokine effects compared with nonadipose tissue, the lesser lipolytic and adipokine effects in SAT compared with VAT may result in a more favorable cardiovascular risk profile when fat is stored preferentially in SAT rather than in VAT. Alternatively, in normal-weight individuals in whom amounts of both VAT and SAT are low, the effects of the lipolytic and adipokine activity by SAT on cardiometabolic risk factor levels are more evident. Our primary results are consistent with this paradigm. Furthermore, in our secondary analyses exploring this interaction from the standpoint of SAT modifying adverse effects of VAT, the adverse effects of VAT lessened with increasing amounts of SAT.
We found that the modification of SAT associations by accompanying VAT amounts was limited to African American women. Due to sample size, this is a preliminary observation that requires replication in additional studies. Should it be replicated, why this is seen in African American women but not in white women remains unclear. Abdominal SAT is partioned into 2 distinct depots split by the fascial plane. Limited prior work suggests that the deep abdominal SAT layer is more strongly associated with adverse cardiometabolic risk factors and insulin resistance than is the superficial SAT layer (10). Therefore, the confinement of beneficial SAT effects to African American women as seen here may result from race-ethnic differences in the ratio of deep to superficial abdominal SAT. The limited literature in this area suggests that African Americans have a smaller deep:superficial SAT ratio than do whites (29, 30). Further research in this area is warranted.
The results of the current study must be viewed within the context of its limitations. These analyses are preliminary and require confirmation in a larger, more definitive study. Our results can be generalized to midlife women only. The relation between abdominal SAT and VAT may be different among younger women. We could not distinguish superficial abdominal SAT from deep abdominal SAT and did not have measures of peripheral SAT. Abdominal SAT may be correlated with peripheral SAT and therefore may have acted as a marker for greater peripheral SAT in the current study. Alternatively, peripheral SAT may affect cardiometabolic risk factors differently than abdominal SAT. For women who had not had a cIMT scan before SWAN Heart ancillary study enrollment, antihypertensive and diabetes medications were reasons for exclusion. The use of diabetes medications was infrequent, and therefore the exclusion of those women likely did not substantially bias the results. The prevalence of antihypertensive medication use among women at the Chicago and Pittsburgh sites during the recruitment period of the ancillary study was ≈25%, in contrast to the 15.8% prevalence (n = 79) in the current analyses. However, results were similar when antihypertensive medication users were excluded, suggesting that these individuals were not driving the observed effect modification.
This study also had considerable strengths. We formally assessed effect modification of SAT by VAT and vice versa, rather than just the independent effects of SAT and VAT. Furthermore, representation of both African American and white women in SWAN Heart allowed for evaluation of the presence of effect modification in both whites and African Americans, separately. Finally, this assessment was performed in midlife women. Given that midlife is a time when cardiovascular disease risk increases in women, this is a critical time period for assessing the potential beneficial cardiometabolic and atherosclerotic effects of SAT.
In conclusion, these results in midlife women suggest that higher amounts of abdominal SAT are associated with adverse cardiometabolic risk factor levels and a greater burden of subclinical atherosclerosis when accompanied by low amounts of abdominal VAT. However, these results suggest that in African American women, as VAT amounts increase the adverse associations of abdominal SAT with subclinical atherosclerosis and insulin resistance may be attenuated or even reversed, such that higher amounts of abdominal SAT are associated with a lower burden of subclinical atheroclerosis and lesser insulin resistance in the presence of high VAT. Given that African American women suffer disproportionately from obesity and cardiovascular disease, larger, more definitive studies confirming these findings and further research into the role of this effect modification on obesity-associated vascular disease in African Americans are needed.
Acknowledgments
We thank the study staff at each site and all the women who participated in SWAN.
Clinical centers: University of Michigan, Ann Arbor [MaryFran Sowers, principal investigator (PI)]; Massachusetts General Hospital, Boston, MA (Joel Finkelstein, PI, 1999–present; Robert Neer, PI, 1994–1999); Rush University, Rush University Medical Center, Chicago, IL (Howard Kravitz, PI, 2009–present; Lynda Powell, PI, 1994–2009); University of California, Davis/Kaiser, Davis, CA (Ellen Gold, PI); University of California, Los Angeles, CA (Gail Greendale, PI); Albert Einstein College of Medicine, Bronx, NY (Rachel Wildman, PI, 2010–present; Nanette Santoro, PI, 2004–2010); University of Medicine and Dentistry–New Jersey Medical School, Newark, NJ (Gerson Weiss, PI, 1994–2004); and the University of Pittsburgh, Pittsburgh, PA (Karen Matthews, PI). NIH Program Office: National Institute on Aging, Bethesda, MD (Sherry Sherman, 1994–present; Marcia Ory, 1994–2001); National Institute of Nursing Research, Bethesda, MD (Program Officers). Central laboratory: University of Michigan, Ann Arbor, MI [Daniel McConnell (Central Ligand Assay Satellite Services)]. Coordinating center: University of Pittsburgh, Pittsburgh, PA (Kim Sutton-Tyrrell, PI, 2001–present); New England Research Institutes, Watertown, MA (Sonja McKinlay, PI, 1995–2001). Steering committee: Susan Johnson, current chair; Chris Gallagher, former chair.
The authors’ responsibilities were as follows—RPW: conceived the current hypothesis tested, wrote the initial draft of the manuscript, had full access to all of the data in the study, and takes responsibility for the integrity of the data and the accuracy of the data analysis; KAM and KS-T: conceived, designed, and conducted the SWAN Heart ancillary study; RPW, IJ, and SREK: analyzed data; and all authors: interpreted data and critically revised the manuscript over successive drafts. The Charles J and Margaret Roberts Trust, which supported the Chicago site of the SWAN Heart Study, was not involved in any aspect of the study, including the design, implementation, analysis, or interpretation of the data. There were no conflicts of interest to declare.
REFERENCES
- 1.Narumi H, Yoshida K, Hashimoto N, et al. Increased subcutaneous fat accumulation has a protective role against subclinical atherosclerosis in asymptomatic subjects undergoing general health screening. Int J Cardiol 2009;135:150–5 [DOI] [PubMed] [Google Scholar]
- 2.Yim JE, Heshka S, Albu JB, Heymsfield S, Gallagher D. Femoral-gluteal subcutaneous and intermuscular adipose tissues have independent and opposing relationships with CVD risk. J Appl Physiol 2008;104:700–7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Van Pelt RE, Jankowski CM, Gozansky WS, Schwartz RS, Kohrt WM. Lower-body adiposity and metabolic protection in postmenopausal women. J Clin Endocrinol Metab 2005;90:4573–8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Meininger G, Hadigan C, Rietschel P, Grinspoon S. Body-composition measurements as predictors of glucose and insulin abnormalities in HIV-positive men. Am J Clin Nutr 2002;76:460–5 [DOI] [PubMed] [Google Scholar]
- 5.Ferreira I, Snijder MB, Twisk JW, et al. Central fat mass versus peripheral fat and lean mass: opposite (adverse versus favorable) associations with arterial stiffness? The Amsterdam Growth and Health Longitudinal Study. J Clin Endocrinol Metab 2004;89:2632–9 [DOI] [PubMed] [Google Scholar]
- 6.Snijder MB, Visser M, Dekker JM, et al. Low subcutaneous thigh fat is a risk factor for unfavourable glucose and lipid levels, independently of high abdominal fat. The Health ABC Study. Diabetologia 2005;48:301–8 [DOI] [PubMed] [Google Scholar]
- 7.Williams MJ, Hunter GR, Kekes-Szabo T, Snyder S, Treuth MS. Regional fat distribution in women and risk of cardiovascular disease. Am J Clin Nutr 1997;65:855–60 [DOI] [PubMed] [Google Scholar]
- 8.Fox CS, Massaro JM, Hoffmann U, et al. Abdominal visceral and subcutaneous adipose tissue compartments: association with metabolic risk factors in the Framingham Heart Study. Circulation 2007;116:39–48 [DOI] [PubMed] [Google Scholar]
- 9.Porter SA, Massaro JM, Hoffmann U, Vasan RS, O'Donnel CJ, Fox CS. Abdominal subcutaneous adipose tissue: a protective fat depot? Diabetes Care 2009;32:1068–75 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kelley DE, Thaete FL, Troost F, Huwe T, Goodpaster BH. Subdivisions of subcutaneous abdominal adipose tissue and insulin resistance. Am J Physiol Endocrinol Metab 2000;278:E941–8 [DOI] [PubMed] [Google Scholar]
- 11.Vega GL, Adams-Huet B, Grundy SM. Influence of body fat content and distribution on variation in metabolic risk. J Clin Endocrinol Metab 2006;91:4459–66 [DOI] [PubMed] [Google Scholar]
- 12.Demerath EW, Reed D, Rogers N, et al. Visceral adiposity and its anatomical distribution as predictors of the metabolic syndrome and cardiometabolic risk factor levels. Am J Clin Nutr 2008;88:1263–71 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Pou KM, Massaro JM, Hoffmann U, et al. Patterns of abdominal fat distribution: the Framingham Heart Study. Diabetes Care 2009;32:481–5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Carroll JF, Chiapa AL, Rodriquez M, et al. Visceral fat, waist circumference, and BMI: impact of race/ethnicity. Obesity (Silver Spring) 2008;16:600–7 [DOI] [PubMed] [Google Scholar]
- 15.Kanaley JA, Giannopoulou I, Tillapaugh-Fay G, Nappi JS, Ploutz-Snyder LL. Racial differences in subcutaneous and visceral fat distribution in postmenopausal black and white women. Metabolism 2003;52:186–91 [DOI] [PubMed] [Google Scholar]
- 16.Conway JM, Yanovski SZ, Avila NA, Hubbard VS. Visceral adipose tissue differences in black and white women. Am J Clin Nutr 1995;61:765–71 [DOI] [PubMed] [Google Scholar]
- 17.Ford ES, Giles WH, Dietz WH. Prevalence of the metabolic syndrome among US adults: findings from the third National Health and Nutrition Examination Survey. JAMA 2002;287:356–9 [DOI] [PubMed] [Google Scholar]
- 18.Chen W, Bao W, Begum S, Elkasabany A, Srinivasan SR, Berenson GS. Age-related patterns of the clustering of cardiovascular risk variables of syndrome X from childhood to young adulthood in a population made up of black and white subjects: the Bogalusa Heart Study. Diabetes 2000;49:1042–8 [DOI] [PubMed] [Google Scholar]
- 19.Sowers MF, Crawford S, Sternfeld B, et al. Design, survey sampling and recruitment methods of SWAN: a multi-cetner, multi-ethnic community-based cohort study of women and the menopausal transition Lobos R, Marcus R, Kelsey JL. Menopause: biology and pathobiology. San Diego, CA: Academic Press, 2000:175–88 [Google Scholar]
- 20.Jensen MD. Lipolysis: contribution from regional fat. Annu Rev Nutr 1997;17:127–39 [DOI] [PubMed] [Google Scholar]
- 21.Fain JN, Madan AK, Hiler ML, Cheema P, Bahouth SW. Comparison of the release of adipokines by adipose tissue, adipose tissue matrix, and adipocytes from visceral and subcutaneous abdominal adipose tissues of obese humans. Endocrinology 2004;145:2273–82 [DOI] [PubMed] [Google Scholar]
- 22.Bruun JM, Lihn AS, Pedersen SB, Richelsen B. Monocyte chemoattractant protein-1 release is higher in visceral than subcutaneous human adipose tissue (AT): implication of macrophages resident in the AT. J Clin Endocrinol Metab 2005;90:2282–9 [DOI] [PubMed] [Google Scholar]
- 23.Fisher FM, McTernan PG, Valsamakis G, et al. Differences in adiponectin protein expression: effect of fat depots and type 2 diabetic status. Horm Metab Res 2002;34:650–4 [DOI] [PubMed] [Google Scholar]
- 24.Hegele RA. Premature atherosclerosis associated with monogenic insulin resistance. Circulation 2001;103:2225–9 [DOI] [PubMed] [Google Scholar]
- 25.Hamaguchi M, Kojima T, Takeda N, et al. The metabolic syndrome as a predictor of nonalcoholic fatty liver disease. Ann Intern Med 2005;143:722–8 [DOI] [PubMed] [Google Scholar]
- 26.Hanley AJ, Williams K, Festa A, Wagenknecht LE, D'Agostino RB, Jr, Haffner SM. Liver markers and development of the metabolic syndrome: the insulin resistance atherosclerosis study. Diabetes 2005;54:3140–7 [DOI] [PubMed] [Google Scholar]
- 27.Volzke H, Robinson DM, Kleine V, et al. Hepatic steatosis is associated with an increased risk of carotid atherosclerosis. World J Gastroenterol 2005;11:1848–53 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Brea A, Mosquera D, Martin E, Arizti A, Cordero JL, Ros E. Nonalcoholic fatty liver disease is associated with carotid atherosclerosis: a case-control study. Arterioscler Thromb Vasc Biol 2005;25:1045–50 [DOI] [PubMed] [Google Scholar]
- 29.Lovejoy JC, Smith SR, Rood JC. Comparison of regional fat distribution and health risk factors in middle-aged white and African American women: the Healthy Transitions Study. Obes Res 2001;9:10–6 [DOI] [PubMed] [Google Scholar]
- 30.Goedecke JH, Levitt NS, Lambert EV, et al. Differential effects of abdominal adipose tissue distribution on insulin sensitivity in black and white South African women. Obesity (Silver Spring) 2009;17:1506–12 [DOI] [PubMed] [Google Scholar]