Digestive Diseases and Sciences

, Volume 53, Issue 5, pp 1358–1363

Ethnicity and Nonalcoholic Fatty Liver Disease in an Obesity Clinic: The Impact of Triglycerides

Authors

    • Department of MedicineUniversity of Illinois at Chicago
    • Section of Digestive Diseases and Nutrition (MC 716)
  • Mrudula Kumar
    • Department of MedicineUniversity of Illinois at Chicago
  • Radhika Aggarwal
    • Department of MedicineUniversity of Illinois at Chicago
  • Reed Berger
    • Department of MedicineUniversity of Illinois at Chicago
  • Jennifer Layden-Almer
    • Department of MedicineUniversity of Illinois at Chicago
  • Neil Gupta
    • Department of MedicineUniversity of Illinois at Chicago
  • Scott J. Cotler
    • Department of MedicineUniversity of Illinois at Chicago
Original Paper

DOI: 10.1007/s10620-008-0234-x

Cite this article as:
Kallwitz, E.R., Kumar, M., Aggarwal, R. et al. Dig Dis Sci (2008) 53: 1358. doi:10.1007/s10620-008-0234-x

Abstract

Nonalcoholic fatty liver disease (NAFLD) is a growing problem that is associated with the metabolic syndrome. The goal of the present study was to evaluate for ethnic differences in NAFLD and clinical correlates of NAFLD. The study population consisted of 567 patients seen at an urban obesity clinic. Elevated aminotransferase levels were used as a surrogate marker for NAFLD. The prevalence of elevated aminotransferases was highest in Hispanics (39%), followed by Caucasians (28%), and African Americans (12%). In univariate analysis, elevated aminotransferase levels were associated with ethnicity (Hispanic > African American, P < 0.001, and Caucasian > African American P = 0.030), hypertriglyceridemia (P < 0.001), and male gender (P < 0.001). The pattern of results was confirmed in multivariate analysis, except that the differences between Caucasians and African Americans was no longer significant. In conclusion, in an obesity clinic population, elevated aminotransferase levels and hypertriglyceridemia were most common in Hispanics and least common in African Americans.

Keywords

Fatty liverEthnologyTriglyceridesObesity

Introduction

The Third National Health and Nutrition Examination Survey (NHANES III) estimated that 66% of the adult population in the United States (US) was overweight (body mass index—BMI ≥ 25 kg/m2) and 32% was obese (BMI ≥ 30 kg/m2) [1]. Obesity is a component of the metabolic syndrome, which also includes insulin resistance, atherogenic dyslipidemia, elevated blood pressure, and proinflammatory and prothrombotic states [2]. Recently, the prevalence of the metabolic syndrome in adults in the US was estimated at 22% [3]. NAFLD is an important health concern related to the metabolic syndrome with risk factors including obesity, type 2 diabetes mellitus, and hypertriglyceridemia [4]. In fact, NAFLD is often considered as the hepatic manifestation of the metabolic syndrome [5].

The NHANES III survey found that 5.4% of US adults had elevated aminotransferase levels that were not attributable to hepatitis B, C, alcohol, or hemochromatosis [6]. A majority of these cases were felt to represent NAFLD, based on the presence of features of the metabolic syndrome. NAFLD encompasses a range of histological changes including steatosis, steatohepatitis (NASH), fibrosis, and cirrhosis. In patients with NASH, the risk of developing cirrhosis over a period of 5 years might be as high as 15% [7].

Factors associated with more severe liver disease in patients with NAFLD include older age, obesity, increased waist-to-hip ratio, presence of insulin resistance, and hypertension [812]. Although earlier case series identified more women with advanced hepatic histology, population-based studies showed that the prevalence of NAFLD was higher in men [68, 11, 13]. In addition to gender differences, rates of NAFLD differ among ethnic groups. The lowest rates of NAFLD were found among African Americans [1419]. Particularly disturbing was the high rate of NAFLD observed in Hispanic youths [17]. Ethnic differences in NAFLD remain poorly understood.

The University of Illinois at Chicago has a busy Nutrition and Wellness Center, serving an ethnically diverse population of patients. The aim of the current study was to assess for ethnic differences in clinical features of NAFLD in an urban obesity clinic.

Methods

The protocol was approved by the Institutional Review Board at the University of Illinois at Chicago.

Subjects

Records from 1,042 consecutive patients who presented to the Nutrition and Wellness Center for management of obesity at the University of Illinois at Chicago from 2001 to 2005 were reviewed retrospectively. Patients were excluded for the following reasons: no recorded aminotransferase levels, BMI < 30 kg/m2, or a history or laboratory evidence of other causes of liver disease including excessive alcohol consumption, defined as more than an occasional drink on social occasions.

Data Collection

Demographic, historical, and laboratory data were extracted from the medical record during a single chart review. Ethnicity was self-reported. The presence or absence of diabetes, hypertension (HTN), alcohol consumption, and tobacco use were recorded. Other data collection included aminotransferase levels, random serum glucose, glycosylated hemoglobin, and lipid levels.

For the primary analysis, aminotransferase elevation was defined as alanine aminotransferase (ALT) >40 U/l or aspartate aminotransferase (AST) >37 U/l in males and ALT or AST >31 U/l in females, based on the definitions used by NHANES III [6]. A secondary analysis using an ALT >30 U/l in males and >19 U/l in females as abnormal was also performed, as these values were proposed as a more sensitive measure of liver disease in patients with hepatitis C [20]. Triglycerides levels >150 mg/dl were designated as elevated and high-density lipoprotein (HDL) levels <40 mg/dl in males and <50 mg/dl in females were considered abnormally low, consistent with the definition of atherogenic dyslipidemia [2]. Data on use of medications that can impact on NAFLD including agents used to treat diabetes, lipid-lowering agents, orlistat, vitamin E, and corticosteroids were also collected.

Statistical Analysis

Descriptive statistics were used to evaluate for normalcy of data and for erroneous entries. Normally and non-normally distributed continuous variables were analyzed using analysis of variance (ANOVA); when an overall significant difference existed, pairwise comparisons were conducted. Pearson’s chi-square testing was used to compare categorical variables. Both univariate and multivariate binary logistic regression was performed in which the dependent variable was elevated aminotransferase level. The primary independent variable was race; African American race was set up as the reference group for the purpose of the analysis. Independent variables that were statistically significant in the univariate analysis, and that appeared to be related to race were also entered in both the univariate and multivariate analysis. Only the final multivariate analyses are displayed in this manuscript. Data analysis was performed using SPSS (SPPS, Inc., version 12.0, Chicago, IL).

Results

Subjects

The study population consisted of 1,042 consecutive patients who presented to the obesity clinic between 2001 and 2005. Subjects were excluded for the following reasons: 342 due to incomplete laboratory data, 62 for a history of significant alcohol consumption, 29 for BMI <30 kg/m2, and 31 for evidence of other causes of liver disease. Given that the dataset included only 11 patients of Asian descent, this group was excluded as well. In total, 567 patients were included in the analysis.

The ethnic distribution of the study population was 73% African-American, 16% Hispanic, and 11% Caucasian. The subjects’ mean age was 42 ± 13 years and 88% were women. There were more men in the Caucasian group than in the other two ethnic groups (Table 1). Overall, the mean BMI was 46 ± 11 kg/m2 and African Americans had a higher BMI than Hispanics. Twenty-eight percent of subjects had diabetes, 23% had hypertriglyceridemia, 67% had low HDL levels, and 47% had hypertension. Hypertriglyceridemia and low HDL levels were least common in African Americans. Current tobacco use was present in 12% of subjects, with 22% having past exposure. Occasional alcohol use was identified in 28% of patients. The rate of medication use that could impact on NAFLD was similar among the three ethnic groups (Table 2).
Table 1

Demographic and clinical features by ethnic group

Variable

African American n = 414

Caucasian n = 60

Hispanic n = 93

P-value

Age (years)

42.3 ± 12.9

43.6 ± 11.7

40.5 ± 12.4

CA/AA, NS

AA/H, NS

Female n (%)

374 (90.3%)

43 (70%)

83 (89.2%)

CA/AA**

AA/H, NS

CA/H**

Body mass index (m/kg2)

46.8 ± 11.4

44.9 ± 11.6

42.6 ± 10.7

CA/AA, NS

AA/H**

Diabetes n (%)

118 (28.5%)

16 (26.7%)

24 (25.8%)

CA/AA, NS

AA/H, NS

CA/H, NS

Hypertriglyceridemia n (%)a

58 (15.8%)

16 (38.1%)

38 (46.9%)

CA/AA*

AA/H**

CA/H, NS

Low HDL n (%)a

232 (62.7%)

34 (79.1%)

66 (80.5%)

CA/AA**

AA/H**

CA/H, NS

HTN n (%)

215 (51.9%)

26 (43.3%)

23 (24.7%)

CA/AA, NS

AA/H**

CA/H*

P-values represent pairwise comparisons. * P ≤ 0.05 and ** P ≤ 0.01

aSerum triglyceride level was available for 489 patients and HDL was available for 495 patients

Table 2

Use of medications that can impact on NAFLD

Medication

Caucasian (%)

African American (%)

Hispanic (%)

Insulin

7.0

8.3

5.4

Metformin

14.0

13.2

15.2

Sulfonylurea

1.8

7.1

5.4

Statin

15.8

17.6

9.8

Fibrate

3.5

1.5

1.1

Orlistat

3.4

3.2

0

Vitamin E

5.2

1.5

4.3

Corticosteroid

1.8

6.4

4.3

No significant differences were noted in any case

Demographic Data and Aminotransferase Levels

Clinical and laboratory data are presented in Table 3, dichotomized by elevated or normal aminotransferase levels. There were significant ethnic differences in the prevalence of elevated aminotransferase levels. Hispanics had the highest rate of elevated levels (39%), followed by Caucasians (28%), and then African Americans (12%) (Fig. 1). Males were more likely than females to have elevated aminotransferase levels. The same pattern of ethnic differences was observed when females, who made up 88% of the study population, were analyzed separately. Although there was no difference in rates of diabetes between the two groups, patients with elevated aminotransferase levels were more likely to have higher glucose values (128 ± 65 mg/dl versus 111 ± 43 mg/dl, P = 0.004) and glycosylated hemoglobin levels (7.0 ± 1.5 versus 6.5 ± 1.5, P = 0.03). Hypertriglyceridemia was also associated with elevated aminotransferase levels. Subjects with elevated aminotransferase levels were significantly more likely to be taking vitamin E (P = 0.01), reflecting use of vitamin E for the treatment of NAFLD during the time period of the study. Occasional alcohol consumption and tobacco use were not significantly associated with aminotransferase elevation.
Table 3

Demographic and clinical features as a function of aminotransferase levels

Variable

Normal ALT/AST

Elevated ALT/ASTa

P value

Ethnicity (H/AA/C)

57/365/43

36/49/17

<0.001

Age (mean ± SD)

42.1 ± 12.9

42.5 ± 12.0

0.770

Gender (M/F)

44/421

24/78

<0.001

BMI (mean ± SD)

45.9 ± 11.3

45.8 ± 11.9

0.878

Diabetes (yes/no)

128/337

30/72

0.701

Hypertriglyceridemia (yes/no)

80/330

32/47

<0.001

Low HDL (yes/no)

276/138

56/25

0.666

Hypertension (yes/no)

215/250

49/53

0.741

M = male, F = female, H = Hispanic, C = Caucasian, AA = African American

aALT >40 U/l or AST >37 U/l in males and ALT or AST >31 U/l in females were considered elevated

https://static-content.springer.com/image/art%3A10.1007%2Fs10620-008-0234-x/MediaObjects/10620_2008_234_Fig1_HTML.gif
Fig. 1

Odds ratios and 95% confidence intervals in univariate analysis for elevated aminotransferase levels by ethnicity

In multivariate analysis, Hispanic ethnicity, male gender, and hypertriglyceridemia were significantly associated with elevated aminotransferase levels (Table 4). In multivariate analysis including only female patients, Hispanic ethnicity and hypertriglyceridemia remained associated with elevated aminotransferase levels (Table 4). In neither case did the higher rate of elevated aminotransferase levels observed in Caucasians compared to African Americans in univariate analysis reach statistical significance in multivariate analysis.
Table 4

Multivariate analysis

Variable

Β coefficient

SE

P value

Odds ratio

95% CI

All patients included

Hispanica

1.190

0.303

0.000

3.29

1.82–5.95

Caucasiana

0.168

0.451

0.709

1.18

0.49–2.86

Male gender

1.022

0.335

0.002

2.78

1.44–5.36

High TG

0.646

0.283

0.023

1.91

1.10–3.33

Female patients

Hispanica

1.117

0.338

0.001

3.06

1.58–5.92

Caucasiana

0.921

0.483

0.056

2.51

0.98–6.48

TG

0.820

0.317

0.010

2.27

1.22–4.23

B = beta, SE = standard error, CI = confidence interval, TG = triglyceride

aCompared to African Americans as the reference group

The pattern of results was slightly different when ALT values of >19 U/l in females and >30 U/l in males were considered elevated. In this analysis, Hispanics and Caucasians had a similar prevalence of elevated ALT levels (66% and 60%), while African Americans continued to have the lowest rate (36%) (P < 0.001). Moreover, the prevalence of elevated ALT levels was similar in females (42%) and males (51%).

Further analysis was performed comparing the 374 patients with class III obesity (BMI ≥40 kg/m2) to the 193 patients with class I (BMI 30–34.9 kg/m2) or class II (BMI 35–39.9 kg/m2) obesity. There were no significant differences in rates of aminotransferase elevation, hypertriglyceridemia, or low HDL between these groups. Compared to patients with class I or II obesity, a higher proportions of patients with class III obesity had diabetes (33% versus 18%) (P < 0.001) and hypertension (53% versus 35%) (P < 0.001).

Discussion

Patients with obesity are at risk for NAFLD [7]. The current study evaluated a large cohort of obese patients at an urban center, using elevated ALT and AST levels as a surrogate marker for NAFLD. Although a majority of patients had class II or III obesity, only 18% of cases overall had elevated aminotransferase levels by conventional criteria [6]. More specifically, 39% of Hispanics, 28% of Caucasians, and 12% of African Americans had elevated aminotransferases. The findings are consistent with previous reports showing that African Americans had the lowest rates of NAFLD [1419]. Similar to the present study, most of the prior data was based on serum markers of liver injury, although one study confirmed these differences on the basis of histology [18]. When more-stringent criteria were applied (ALT >30 U/l in males and >19 U/l in females), the prevalence of elevated ALT levels in the current study increased dramatically and differences in elevated ALT levels between men and women were no longer apparent. Further studies of NAFLD are needed to assess whether observed gender differences in rates of elevated aminotransferase levels reflect liver histology or represent an artifact of aminotransferase thresholds.

Data from the current study clearly show that Hispanics had the highest rate of aminotransferase elevation and that African Americans had the lowest rate. Reasons for ethnic differences in NAFLD have not been fully explained. However, in large epidemiologic studies, Hispanics had the highest rates of diabetes and hypertriglyceridemia [3, 21, 22], which are associated with NAFLD.

The low rate of elevated aminotransferase levels observed in African Americans is particularly interesting, given that African Americans have a similar prevalence of diabetes as Hispanics in some series [3, 21, 22]. One hypothesis proposed to explain this observation is that African Americans have less abdominal visceral fat at a given level of adiposity than Hispanics or Caucasians [23]. Fat labeling studies in a fasting state indicated that a majority of hepatic triglycerides originate as nonesterified fatty acids, which are a product of adipose tissue [24]. Differences in lipid metabolism also might play a role. The accumulation of triglycerides within hepatocytes is regarded as the first step in the pathogenesis of NAFLD and is related to insulin resistance [25]. A recent study found that elevated triglyceride levels, along with low HDL, central obesity, and increased fasting glucose levels, were the components of the metabolic syndrome most strongly associated with increased ALT levels [26]. In the present study, African Americans, who had the lowest prevalence of elevated aminotransferase levels, also had the lowest rate of hypertriglyceridemia. Moreover, differences in the rates of elevated aminotransferases observed in univariate analysis between African Americans and Caucasians were no longer apparent in multivariate analysis controlling for hypertriglyceridemia. In other studies, hepatic insulin resistance was found to correlate with triglyceride levels in Caucasians, but not in African Americans [27, 28]. The data raise the possibility that lower levels of triglycerides in African Americans with the metabolic syndrome might contribute to ethnic differences in NAFLD.

Another interesting observation involves apolipoprotein C-III, a lipoprotein that is strongly associated with serum triglycerides and inversely related to insulin levels [29]. In one report, increased apolipoprotein C-III levels were associated with insulin resistance in Caucasians and Hispanics, but not in African Americans [30], which could potentially contribute to the lower triglyceride and aminotransferase levels observed in African American patients.

Limitations of the current study include the retrospective design and lack of data regarding central obesity and insulin resistance. In addition, liver histology was not available. However, many previous studies have used aminotransferase levels as a surrogate marker for NAFLD [1417, 19]. Strengths of this study included the availability of clinical data from a large, ethnically diverse study population, ranging from stage I to stage III obesity.

In conclusion, only a small proportion of obese patients had elevated aminotransferase levels, including those with stage III obesity. The prevalence of elevated aminotransferase levels increased significantly from African Americans to Caucasians to Hispanics. High triglyceride levels differed by race, and were associated with elevated ALT levels in patients at risk for NAFLD.

Copyright information

© Springer Science+Business Media, LLC 2008