The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Department of Family and Preventive Medicine, University of California San Diego, 9500 Gilman Drive, 0607, La Jolla, CA 92093. E-mail: email@example.com
Objective: To compare ethnic differences in visceral adipose tissue (VAT), assessed by computed tomography, and type 2 diabetes risk among 55- to 80-year-old Filipino, African-American, and white women without known cardiovascular disease.
Research Methods and Procedures: Subjects were participants in the Rancho Bernardo Study (n = 196), the Filipino Women's Health Study (n = 181), and the Health Assessment Study of African-American Women (n = 193). Glucose and anthropometric measurements were assessed between 1995 and 2002.
Results: African-American women had significantly higher age-adjusted BMI (29.7 kg/m2) and waist girth (88.1 cm) compared with Filipino (BMI, 25.5 kg/m2; waist girth, 81.9 cm) or white (BMI: 26.0 kg/m2; waist girth: 80.7 cm) women. However, VAT was significantly higher among Filipino (69.1 cm3) compared with white (62.3 cm3; p = 0.037) or African-American (57.5 cm3, p < 0.001) women. VAT correlated better with BMI (r = 0.69) and waist (r = 0.77) in whites, compared with Filipino (r = 0.42; r = 0.59) or African-American (r = 0.50; r = 0.56) women. Age-adjusted type 2 diabetes prevalence was significantly higher in Filipinas (32.1%) than in white (5.8%) or African-American (12.1%) women. Filipinas had higher type 2 diabetes risk compared with African Americans [adjusted odds ratio, 2.30; 95% confidence interval (CI), 1.09 to 4.86] or whites (adjusted odds ratio, 7.51; 95% CI, 2.51 to 22.5) after adjusting for age, VAT, exercise, education, and alcohol intake.
Discussion: VAT was highest among Filipinas despite similar BMI and waist circumference as whites. BMI and waist circumference were weaker estimates of VAT in Filipino and African-American women than in whites. Type 2 diabetes prevalence was highest among Filipino women at every level of VAT, but VAT did not explain their elevated type 2 diabetes risk.
Elevated type 2 diabetes prevalence has been reported among African Americans, Latinos, Native Americans, and Pacific Islanders compared with whites, and the excess has been attributed to ethnic differences in obesity (1). However, Filipinos, Japanese, and Chinese have higher type 2 diabetes prevalence than whites despite similar body size (2). The University of California San Diego (UCSD)1 Filipino Women's Health Study showed that 50- to 69-year-old Filipino women had over 3 times the prevalence of type 2 diabetes (33%) than white women (9%) of similar age, BMI, waist-to-hip ratio, and percentage body fat (by DXA) (3). Furthermore, truncal fat (by DXA) was independently associated with type 2 diabetes risk among white but not among Filipino women (3).
Central obesity is a known risk factor for type 2 diabetes and may induce insulin resistance (IR) by reducing insulin-mediated glucose disposal. Studies of Asian Indians suggest more visceral adipose tissue (VAT) compared with whites of similar BMI (4). However, African-American women have less VAT than white or Mexican-American women of similar BMI (5, 6). Ethnic differences in type 2 diabetes prevalence could reflect ethnic differences in VAT. However, estimates of central obesity including waist circumference, waist-to-hip ratio, and truncal fat measurements by DXA do not distinguish visceral from subcutaneous fat and may not adequately estimate visceral adiposity. Computed tomography (CT) provides a direct estimate of VAT and subcutaneous fat (7).
The objectives of this study were to compare correlates of obesity, including BMI, waist girth, waist-to-hip ratio, and truncal fat by DXA with visceral and subcutaneous fat among 55- to 80-year-old community-dwelling Filipino, African-American, and white women without known cardiovascular disease and to compare ethnic differences in the prevalence of type 2 diabetes after adjusting for VAT and other known risk factors.
Research Methods and Procedures
Filipino women were recruited between 1995 and 1999 to a cross-sectional study designed to estimate the prevalence of several chronic conditions, including osteoporosis, cardiovascular disease, hypertension, and type 2 diabetes (3). Population-based sampling was not possible because Filipinos were not identified separately in the 1990 census; consequently, a convenience sample was recruited from San Diego, CA, as described elsewhere (3). African-American women were participants of the UCSD Health Assessment Study of African-American Women, a cross-sectional study of osteoporosis, cardiovascular disease, hypertension, and type 2 diabetes conducted between 1995 and 1999. The comparison group of non-Hispanic white women (primarily of Northern European descent) were participants in the UCSD Rancho Bernardo Study, a San Diego community-based longitudinal study (8). From 1997 to 1999, ∼70% of surviving local, non-institutionalized members from the Rancho Bernardo cohort participated in a study of the same chronic diseases using the same research protocol. For comparison with the middle to upper middle class Rancho Bernardo cohort, an effort was made to recruit Filipino and African-American women of similar education and social class.
VAT was assessed between 2000 and 2002, using electron beam CT (EBCT) and the same protocol, clinical facility, and research staff (9). Filipino, African-American, and white postmenopausal women, who were then 55 to 80 years old, and without a history of myocardial infarction, angina pectoris, or coronary artery revascularization, were invited. Among age-eligible participants without known coronary heart disease (CHD), 79% of Filipinas, 78% of African-American women, and 82% of white women joined the EBCT study. All participants gave written informed consent, and the research protocol, consent forms, and data instruments were approved by the UCSD Human Research Protections Program.
Demographic characteristics, lifestyle characteristics (cigarette smoking, alcohol use, physical activity), physician-diagnosed diseases, hospitalizations, and surgeries were determined using structured questionnaires. A self-administered questionnaire was used to assess whether participants engaged in exercise or labor at least three times a week (for 30 minutes). Specific information on the frequency and duration of specific activities including but not limited to walking, jogging, dancing, playing golf, and gardening were collected but were not used for this analysis. Participants who were using prescription or non-prescription medications in the month before the clinic visit brought pills and prescriptions to the clinic to be verified and recorded by a nurse.
Height and weight were measured with participants wearing lightweight clothing without shoes. BMI (kilograms per meter squared) was computed as an estimate of obesity. Waist circumference was measured at the narrowest circumference. Hip circumference was measured at the largest circumference. Percentage of total body and truncal fat were determined by DXA (model QDR-2000 X-ray bone densitometers; Hologic, Inc., Waltham, MA).
A 75-gram oral glucose tolerance test was administered in the morning after a minimum 8-hour fast; blood samples were obtained by venipuncture at 0 and 2 hours. Plasma glucose was measured by the glucose-oxidase method, and insulin was determined by radioimmunoassay in a diabetes research laboratory. The homeostasis model assessment (HOMA) was used to estimate IR (10).
Type 2 diabetes was defined using the 1999 World Health Organization (WHO) criteria: fasting plasma glucose level ≥ 126 mg/dl, or 2-hour postchallenge glucose level ≥200 mg/dl, or a history of type 2 diabetes diagnosed by a physician, or treatment with an oral hypoglycemic agent or insulin (11). Lower waist and BMI cut-off points for Asians were applied to define large waist girth (Filipinas, >80 cm; white and African-American women, >88 cm) and overweight (Filipinas, BMI ≥ 23 kg/m2; white and African-American women, BMI ≥ 25 kg/m2) (12).
Visceral and subcutaneous fat were determined using an Imatron C-150 ultrafast CT scanner. A single-slice CT scan of the abdomen (at the umbilicus) was obtained to measure visceral and subcutaneous adipose tissue volume.
Data were analyzed using SAS version 8.2 (SAS, Cary, NC). Student's t tests, ANOVA, general linear models, and χ2 analysis were used for descriptive statistics. Pearson correlation coefficients were computed to assess the relationship between VAT and other anthropometric markers. Multivariable logistic regression was used to assess the relationship among ethnicity, VAT, and type 2 diabetes. Logistic regression models were constructed to compare diabetes risk among Filipinos vs. whites and Filipinos vs. African Americans, while adjusting for putative confounders including age, visceral adiposity, exercise, education, and alcohol intake. Multivariable regression was also performed separately for each ethnic group to assess the relationship between VAT and type 2 diabetes. A regression model that included ethnicity, VAT, and an ethnicity × VAT interaction term was also used to determine whether the relationship between VAT and type 2 diabetes differed by ethnicity. Statistical significance was designated at p < 0.05 and odds ratios (ORs) that excluded 1.
A total of 181 Filipino, 193 African-American, and 196 white women 55 to 80 years old completed the clinical evaluation between 1995 and 1999 and had an EBCT scan in 2000 to 2002. As shown in Table 1, Filipinas were significantly younger than whites and African Americans (p < 0.001), but mean age did not differ between whites and African Americans (p = 0.97). Filipinas were more likely to be college graduates than whites (p = 0.003), but college attainment did not differ between Filipino and African-American women (p = 0.13), or between whites and African Americans (p = 0.13). Filipinas were less likely to have ever smoked cigarettes or to drink alcohol (≥three drinks per week) compared with whites (p < 0.001) or African Americans (p < 0.001). Smoking history was similar among whites and African Americans (p = 0.96), but African-American women were significantly less likely to drink alcohol compared with white women (p < 0.001). Exercise (≥three times per week) did not differ between Filipinas and whites (p = 0.62) or African Americans (p = 0.083). However, exercise was significantly less frequent among African Americans compared with whites (p = 0.02).
Table 1. . Age-adjusted demographic, behavioral, and anthropometric characteristics among Filipino, white, and African-American women, 55- to 80-years-old, San Diego, 1995 to 2002
As shown in Table 1, African-American women had significantly larger age-adjusted BMI (29.7 kg/m2), waist girth (88.1 cm), and truncal fat (33.6% by DXA) compared with Filipinas or whites (p < 0.01); however, VAT was lowest for African Americans (57.5 cm3) and highest for Filipinas (69.1 cm3, p = 0.0004). Waist circumference, BMI, and truncal fat did not differ between Filipino (81.9 cm, 25.5 kg/m2, 31.4%) and white women (80.7 cm, 26.0 kg/m2, 30.5%); however, Filipinas had significantly more VAT compared with white women (69.1 vs. 62.3 cm3, p = 0.037). Although African Americans had significantly larger waist circumference, BMI, and truncal fat compared with whites, VAT did not differ between these two groups (p = 0.13). Waist-to-hip ratio was significantly lower for white women compared with Filipino or African-American women (p < 0.0001) and higher for Filipinas than African-American women (p = 0.001).
Subcutaneous fat was significantly higher among African-American women compared with Filipino or white women (p < 0.001) but did not differ between Filipinas and whites (p = 0.69). The ratio of visceral to subcutaneous fat was significantly higher among Filipinas compared with African Americans (p = 0.003) but did not differ between Filipinas and whites (p = 0.11) or between African Americans and whites (p = 0.16).
Filipino women had significantly higher age-adjusted fasting glucose levels compared with African-American women (p < 0.004), but fasting glucose levels did not differ between whites and Filipinas (p = 0.10) nor between whites and African Americans (p = 0.19). However, 2-hour postchallenge glucose levels were significantly higher among Filipinas compared with African Americans or whites (p < 0.0001) but did not differ between African-American and white women (p = 0.11). Age-adjusted fasting insulin, 2-hour postchallenge insulin, and HOMA-IR values did not differ by ethnicity.
Pearson correlation coefficients were computed to determine the relationship between VAT and anthropometric markers. VAT correlated positively with BMI, waist girth, waist-to-hip ratio, truncal fat (by DXA), subcutaneous fat, and visceral-to-subcutaneous fat ratio (p < 0.02). As shown in Table 2, Pearson correlation coefficients between VAT and BMI, waist, waist-to-hip ratio, truncal fat (by DXA), and subcutaneous fat were highest for whites. Correlation coefficients were highest for VAT and waist girth in each ethnic group; however, these values were higher in whites (r = 0.77) compared with Filipino (r = 0.59) or African-American (r = 0.56) women. VAT correlated reasonably well with BMI among whites, but correlation coefficients were lower among Filipino or African-American women.
Table 2. . Pearson correlation coefficients: CT-defined VAT and anthropometric markers among Filipino, African-American, and white women, 55- to 78-years-old, San Diego, 1995 to 2002
Filipino (n = 181)
White (n = 196)
African American (n = 193)
Truncal fat (DXA)
Subcutaneous fat (CT)
Visceral-to-subcutaneous fat ratio
Age-adjusted type 2 diabetes prevalence was 32.1% in Filipinas, 5.8% among whites, and 12.1% among African-American women. As shown in Figure 1, age-adjusted prevalence of type 2 diabetes was highest for Filipino women at every tertile of VAT. As expected, type 2 diabetes prevalence was higher for women in the highest VAT (≥74.9 cm3) tertile: whites, 9.8%, African Americans, 14.7%; and Filipino women, 46.6%. However, among women with the least VAT (≤46 cm3), type 2 diabetes was low (1.7%) among whites, 7.4% among African-American women, and as high as 22.8% among Filipino women (p < 0.001). Type 2 diabetes prevalence increased with increasing levels of VAT among whites but did not follow a linear trend among Filipino and African-American women.
Normal Weight and Waist Girth in Asians: WHO Criteria
WHO has recommended the use of lower BMI and waist cut-off points to define overweight and abdominal obesity among Asians (11). When these criteria were applied to define women with normal weight (Filipinas, BMI < 23 kg/m2; whites and African Americans, <25 kg/m2), as shown in Table 3, VAT was significantly higher among Filipinas (53.8 cm3) compared with white (43.8 cm3, p = 0.031) or African-American (37.9 cm3, p = 0.004) women. Among these normal-weight women, diabetes prevalence remained significantly higher among Filipinas (20.6%) compared with whites (2.1%) or African Americans (7.5%).
Table 3. . VAT among normal-weight or normal-waist women, by WHO criteria, among Filipino, white, and African-American women
Filipino, BMI < 23 kg/m2; white and African American, BMI < 25 kg/m2.
Filipino, waist <80 cm; white and African American, waist ≤88 cm.
Among women with normal-waist girth (Filipinas, ≤80 cm; whites and African Americans, ≤88 cm), VAT was significantly lower in African-American women but did not differ between Filipino and white women (Table 3). However, diabetes prevalence among these normal-waist women remained significantly higher for Filipinas (23.3%) compared with whites (3.3%) and African Americans (10.4%). As shown in Figure 2, VAT among normal-weight Filipinas was significantly higher at every level of waist girth compared with normal-weight white and African-American women (p < 0.005).
In multivariable analysis, compared with white women, Filipinas had a 7-fold higher risk of type 2 diabetes [adjusted OR, 7.51; 95% confidence interval (CI), 2.51 to 22.5], after adjusting for age, VAT, exercise, college education, and alcohol intake (Table 4). Similarly, compared with African-American women, Filipinas had a 2-fold higher risk of type 2 diabetes (adjusted OR, 2.30; 95% CI, 1.09 to 4.86), after adjusting for age, VAT, exercise, college education, and alcohol intake. However, in multivariable analysis limited to white and African-American women, ethnicity was not associated with the risk of type 2 diabetes (adjusted OR, 2.47; 95% CI, 0.84 to 7.23); VAT was the only covariate associated with type 2 diabetes (adjusted OR, 1.10; 95% CI, 1.03 to 1.17) in a model that adjusted for age, exercise, education, and alcohol intake. VAT was independently associated with type 2 diabetes risk in all ethnic groups, but ethnic differences in VAT did not fully account for the elevated type 2 diabetes risk among Filipino women.
Table 4. . Multivariable analysis: covariates associated with type 2 diabetes: Filipino vs. white and African-American women
0.81 to 1.35
0.83 to 1.31
2.51 to 22.5
1.09 to 4.86
1.05 to 1.15
1.06 to 1.18
Exercise (≥3 times/wk)
0.57 to 2.12
0.92 to 3.39
0.23 to 0.80
0.23 to 0.77
Alcohol (≥3 times/wk)
0.22 to 2.99
0.16 to 3.83
To determine whether the relationship of VAT to type 2 diabetes differed by ethnicity, an interaction term, ethnicity × VAT, was included in a multivariable regression model; however, the ethnicity × VAT interaction term was not associated with diabetes risk (p = 0.861), suggesting that the association between diabetes and visceral fat did not differ by ethnicity.
This is the first study, to our knowledge, to report excess VAT accumulation and prevalence of type 2 diabetes among Filipino women with similar BMI, waist girth, and truncal fat as white women. These observations of excess VAT among Filipinas persisted after lower BMI and waist girth cut-off points for Asians were applied to define women with normal weight and normal waist circumference. Hence, WHO recommendations for lower cut-off points to define overweight and large waist girth among Asians inadequately identifies excess VAT among Filipino-American women. Excess VAT is a known risk factor for type 2 diabetes; however, excess VAT in Filipinas did not explain their excess risk for type 2 diabetes compared with white or African-American women.
This study supports other observations that BMI, waist-to-hip ratio, and waist circumference are not comparable across different racial and ethnic groups. Similar to previous observations, our cohort of African Americans had the largest BMI, waist girth, and truncal fat (by DXA) but had the least VAT (5, 13). In our study, waist, BMI, waist-to-hip ratio, and truncal fat (by DXA) did not detect important ethnic differences in VAT demonstrated by CT. Similarly, African-American women defined as overweight by BMI (≥25 kg/m2) or having large waist girth (>88 cm) had significantly lower VAT compared with white women with similar BMI and waist circumference. Furthermore, BMI and waist girth correlated well with VAT for whites, but correlation coefficients were lower for African-American and Filipino women. Previous studies have reported lower visceral adiposity in African Americans but higher VAT volume among Asian Indians and Japanese Americans compared with whites of similar BMI (14, 15, 16). These observations coupled with the data reported here challenge the sensitivity of BMI or waist circumference as estimates of VAT except among whites of Northern European ancestry.
The mechanism for ethnic disparities in VAT accumulation is unclear. Excess subcutaneous fat accumulation and decreased VAT deposition have been observed among individuals of African descent, whether resident in Africa, the U.S., the Caribbean, South America, or Europe, but the mechanism for such fat distribution has not been elucidated. Similarly, the mechanism behind preferential deposition of VAT in selected ethnic groups, including Asian Indians, Mexican Americans, and Filipinos, remains unclear (16).
Previous studies have shown an adverse association between physical inactivity and VAT accumulation; however, exercise frequency was similar among Filipino and white women. Furthermore, multivariable analysis showed that physical activity was associated with VAT among white and African-American women but not among Filipinas.
In a prospective study of non-diabetic Japanese-American men, greater IR and reduced insulin secretion preceded VAT accumulation (17). Our cross-sectional design did not allow us to distinguish whether VAT accumulation preceded or followed metabolic abnormalities; however, neither fasting insulin nor IR (by HOMA-IR) differed by ethnicity. Other studies have suggested that IR is associated with alterations in fatty acid metabolism, and circulating free fatty acids inhibit the ability of insulin to stimulate muscle glucose uptake and to suppress hepatic glucose production (18). Furthermore, visceral fat may not be as important as subcutaneous fat in supplying free fatty acids to the liver (19). However, subcutaneous fat was not associated with type 2 diabetes in our cohort.
Ethnic differences in VAT could account for some of the ethnic disparities in type 2 diabetes prevalence, particularly in populations where general obesity is uncommon. No ethnic differences were observed in diabetes risk between African-American and white women after adjusting for VAT, age, exercise, alcohol intake, and education. However, the risk of type 2 diabetes was 7 times higher among Filipinas compared with whites and twice as high compared with African-American women after adjusting for known diabetes risk factors. Excess VAT did not explain the elevated prevalence of type 2 diabetes in these Filipino-American women. Type 2 diabetes prevalence was 23% among Filipinas in the lowest tertile of VAT, compared with just 1.7% among whites.
Excess VAT accumulation may result in changes in the production or action of adipocytokines, including adiponectin, resistin, leptin, and tumor necrosis factor-α. Filipinos may have lower levels of adipose-derived hormones, including adiponectin, which plays an important role in glucose homeostasis and insulin sensitivity and has been shown in prospective studies (20, 21, 22, 23) to be predictive of type 2 diabetes. Future studies should consider the etiologic relationship between adipocytokines and type 2 diabetes risk among Filipinos.
This study has several limitations. Obtaining a population-based sample of Filipino-Americans was not feasible when the study was initiated because 1990 census data reported Asian-American census collectively. However, our sample is representative of all Filipino women in the U.S. with regard to college attainment, which was used as a marker for socioeconomic status (24). Furthermore, African-American women were chosen to select for white-collar workers to minimize the effects of racial/ethnic differences in social class and access to health care that often confound studies of non-white women. The sample was limited to women without known cardiovascular disease or revascularization surgery in 2000 to 2002; consequently, women with longer duration of type 2 diabetes might have had more symptomatic CHD and would not have been represented. However, the proportion of women who were excluded from participating because of previously diagnosed cardiovascular disease did not differ by ethnicity (Filipinas 21%, white 23%) (9); therefore, participation bias resulting from differential diagnosis of prior CHD by ethnicity was unlikely.
In conclusion, Filipino women have excess VAT accumulation despite similar BMI and waist girth as white women. The risk of type 2 diabetes did not differ between African-American and white women after adjusting for VAT and other known risk factors; age-adjusted prevalence of type 2 diabetes was significantly higher in Filipinos than in white or African-American women; however, VAT did not explain the excess risk of type 2 diabetes among Filipino-American women. CT measurement of VAT volume is costly and inconvenient. Future studies should consider mechanisms mediating diabetes risk and methods to reduce VAT accumulation in ethnic groups that are not overweight even by Asian standards. Continued evaluation of fat depots other than VAT, including intermuscular adipose tissue, may enhance current knowledge regarding ethnic variations in fat accumulation and IR.
This work was supported by American Heart Association Grant 0070088Y, by NIH/National Institute of Diabetes and Digestive and Kidney Diseases Grant R03 DK60575, by NIH/National Institute on Aging Grant 5R01AG07181, and by NIH/National Institute of Diabetes and Digestive and Kidney Diseases Grant 5R01DK31801.
Nonstandard abbreviations: UCSD, University of California San Diego; IR, insulin resistance; VAT, visceral adipose tissue; CT, computed tomography; EBCT, electron beam CT; CHD, coronary heart disease; HOMA, homeostasis model assessment; WHO, World Health Organization; OR, odds ratio; CI, confidence interval.