Liver Failure and Liver Disease
Elevated serum alanine aminotransferase activity and calculated risk of coronary heart disease in the United States†
Article first published online: 20 APR 2006
Copyright © 2006 American Association for the Study of Liver Diseases
Volume 43, Issue 5, pages 1145–1151, May 2006
How to Cite
Ioannou, G. N., Weiss, N. S., Boyko, E. J., Mozaffarian, D. and Lee, S. P. (2006), Elevated serum alanine aminotransferase activity and calculated risk of coronary heart disease in the United States. Hepatology, 43: 1145–1151. doi: 10.1002/hep.21171
Potential conflict of interest: Nothing to report.
- Issue published online: 20 APR 2006
- Article first published online: 20 APR 2006
- Manuscript Received: 25 OCT 2006
- Manuscript Accepted: 19 FEB 2006
- American College of Gastroenterology Junior Faculty Development Award
- Primary and Specialty Medical Care Service
- Research Enhancement Award Program
- Veterans Affairs Puget Sound Health Care System
In the United States, elevated serum alanine aminotransferase (ALT) activity in the absence of viral hepatitis or excessive alcohol consumption is most commonly attributed to nonalcoholic fatty liver disease (NAFLD). NAFLD is related to predictors of coronary heart disease (CHD) such as insulin resistance and central obesity. We examined the association between elevated serum ALT activity and the 10-year risk of CHD as estimated using the Framingham risk score (FRS). We performed a cross-sectional analysis comparing participants in the Third National Health and Nutrition Examination Survey with normal and elevated ALT activity (>43 IU/L), examining the mean levels of FRS. Among participants without viral hepatitis or excessive alcohol consumption, those with elevated ALT activity (n = 267) had a higher FRS than those with normal ALT activity (n = 7,259), both among men (mean difference in FRS 0.25, 95% CI 0.07-0.4; hazard ratio for CHD 1.28, 95% CI 1.07-1.5) and women (mean difference in FRS 0.76, 95% CI 0.4-1.1; hazard ratio for CHD 2.14, 95% CI 1.5-3.0). The ALT threshold for increased risk of CHD was higher in men (>43 IU/L) than in women (>30 IU/L). Elevated ALT activity was not associated with higher FRS among nonobese participants with viral hepatitis or excessive alcohol consumption. In conclusion, individuals with elevated serum ALT activity in the absence of viral hepatitis or excessive alcohol consumption, most of whom have NAFLD, have an increased calculated risk of CHD. This association is more prominent in women. (HEPATOLOGY 2006;43:1145–1151.)
Nonalcoholic fatty liver disease (NAFLD), a condition that may affect as much as 30% of the adult United States population,1 is strongly associated with many predictors of cardiovascular disease such as hypercholesterolemia, hypertriglyceridemia, insulin resistance, central obesity, and the metabolic syndrome.2–9 Given these strong associations, patients with NAFLD would be expected to have a high risk of coronary heart disease (CHD). However, the relationship between NAFLD and CHD in the United States is not well established. A recent study suggested that overall mortality was slightly greater among 420 persons diagnosed with NAFLD in Olmsted County between 1980 and 2000 than among the general population of Minnesota (standardized mortality ratio 1.34)10; unfortunately, specific data on excess cardiovascular mortality were not reported. A retrospective chart review of 98 patients with biopsy-proven NAFLD/nonalcoholic steatohepatitis from a tertiary referral center found 48 deaths over 8.3 years of follow-up, including 11 deaths attributed to cardiac causes11; however, there was no appropriate control group to be compared with this highly selected group of NAFLD patients. Patients discharged from Danish hospitals with a diagnosis of nonalcoholic or unspecified fatty liver without mention of alcohol use between 1977 and 1993 had a 2.6-fold greater mortality than the general population; however, this study was based solely on patients who were hospitalized and would thus be expected to have a higher mortality than the general population.12 In addition, it is questionable whether all the cases of “unspecified” fatty liver had NAFLD instead of alcoholic fatty liver disease.
Villanova et al.13 recently reported that the 10-year risk of coronary events of 52 patients with NAFLD attending the metabolic unit of the University Hospital of Bologna estimated using the Framingham risk score (FRS) was higher than the risk of 28 control patients without metabolic diseases. In addition, the NAFLD patients had lower flow-mediated vasodilation of the brachial artery in response to ischemia, suggesting the presence of endothelial dysfunction. Although the results of this study are interesting, they are based on a small sample of highly selected patients.
The magnitude of the excess risk of coronary events, if any, in persons with NAFLD in the general United States population is largely unknown. For the purposes of epidemiological studies, NAFLD can be defined by the presence of elevated serum aminotransferase activity—particularly alanine aminotransferase (ALT)—in the absence of other common causes of liver disease such as viral hepatitis and excessive alcohol consumption,3, 14 because the vast majority of patients meeting these criteria are proven to have NAFLD/nonalcoholic steatohepatitis when further investigated via liver biopsy.15–17 The goal of the present study was to determine whether in a large, representative sample of the United States population, persons with suspected NAFLD (defined by elevated serum ALT in the absence of viral hepatitis and excessive alcohol consumption) have an elevated 10-year risk of CHD as estimated using the FRS.
Subjects and Methods
Data were derived from the Third National Health and Nutrition Examination Survey (NHANES III), a cross-sectional study conducted by the National Center for Health Statistics between 1988 and 1994 to assess the health and nutritional status of the noninstitutionalized United States population.18 Data included laboratory investigations, physical examination findings, and structured questionnaires conducted in mobile examination centers at 89 locations throughout the United States.
Of 19,620 participants aged 18 years or older, we excluded 1,286 who reported a history of heart attack or congestive heart failure, 321 with a positive pregnancy test, 2,827 without available data on serum ALT level, 5,679 who fasted for fewer than 8 hours, 668 with missing information on at least 1 of the components of the FRS, 337 with missing information on hepatitis B or C status or alcohol consumption, and 121 with missing information on past history of heart attack or congestive heart failure (Fig. 1). This left 8,381 participants in the present analysis. Among these participants, 7,526 did not have viral hepatitis B (determined by presence of hepatitis B surface antigen) or C (determined by presence of hepatitis C viral RNA) or excessive alcohol consumption (defined as ≥1 drink/d for women and ≥2 drinks/d for men3, 19, 20), whereas 855 participants had at least 1 of these potential causes of chronic liver disease (Fig. 1).
The FRS was calculated separately for men and women using the coefficients of the proportional hazards model reported by Wilson et al.21 The score is computed based on each person's age, age squared (in women only), total cholesterol, high-density lipoprotein cholesterol, blood pressure, diabetes status, and smoking status (Appendix I). The higher the risk score, the greater the 10-year risk of developing CHD, which was defined as angina pectoris, myocardial infarction, or coronary death.
Total cholesterol was measured enzymatically. High-density lipoprotein cholesterol was measured after precipitation of other lipoproteins with a polyanion/divalent cation mixture. Three blood pressure readings were obtained in the mobile examination centers; we calculated the average of the second and third systolic and diastolic blood pressure readings. Patients were classified as diabetic if they had ever been told by a physician that they had diabetes mellitus or if their plasma glucose level was 126 mg/dL or higher after more than 8 hours of fasting. Smoking status was based on self-reported current smoking.
Definition of Elevated Serum ALT Activity.
Elevated ALT was defined as a level greater than 43 IU/L in both men and women in accordance with the NHANES III Plan and Operations Manual18 and previous studies.2, 22 To evaluate whether there was a threshold of ALT above which the FRS was elevated and to look for a dose–response relationship, we further categorized ALT into 0-25, 26-30, 31-43, and >43 IU/L.
After collection of venous blood, samples were immediately centrifuged. Specimens were then frozen and shipped weekly to a central laboratory, were they were initially stored at −20°C and then at −70°C. Serum ALT assay was performed with a Hitachi model 737 multichannel analyzer (Boehringer Mannheim Diagnostics, Indianapolis, IN) using the α-ketoglutarate reaction.
Determinants of the Association Between Elevated ALT Activity and FRS.
NAFLD is believed to be caused, to a large extent, by insulin resistance, obesity, and central fat distribution. These characteristics are also strong predictors of CHD. We evaluated whether any association between elevated ALT activity and elevated FRSs was due to the confounding effect of these characteristics by adjusting for them. Insulin resistance was calculated using the homeostasis model assessment (HOMA)23, 24 as: (fasting serum insulin [μU/mL] × fasting serum glucose [mmol/L])/22.5. Obesity was measured using body mass index (kg/m2). Central fat distribution was estimated using the waist-to-hip ratio.
Elevated ALT Activity in Persons With Viral Hepatitis or Excessive Alcohol Consumption.
Elevated serum ALT activity is a marker of hepatic inflammation. Therefore, it is possible that any association between elevated ALT activity and increased risk of CHD is related to hepatic inflammation and not to NAFLD or its predisposing conditions. To rule out this possibility, we performed a separate analysis limited to persons with viral hepatitis or excessive alcohol consumption who were not obese. (We excluded obese persons to exclude NAFLD as much as possible, because in a previous study we demonstrated that NAFLD may be responsible for a substantial proportion of elevated ALT activity, even in persons with viral hepatitis or excessive alcohol consumption22).
We used linear regression to determine the mean difference in FRS between persons with and without elevated ALT activity after adjusting for age. To determine the confounding effect of insulin resistance, obesity, and central fat distribution, we additionally adjusted for homeostasis model assessment insulin resistance, body mass index, and waist-to-hip ratio modeled as continuous variables.
We converted the FRS into 10-year risk of CHD for a 50-year-old man or woman with or without elevated ALT. This was done by first calculating the difference in FRS between such persons and the persons with “baseline” characteristics:
where FRS(baseline)is the FRS of persons with “baseline” or average characteristics and is given as 3.0975 for men and 9.92545 for women (as described by Wilson et al.21); FRS(x)is the FRS of a man or woman aged 50 years with either normal or elevated ALT. The FRS of 50-year-old men and women with normal ALT levels was the average FRS of all men and women aged 49-51 years in our dataset (3.18 for men and 10.38 for women). For men or women with elevated ALT levels, we used a FRS equal to the FRS of men or women with normal ALT plus the adjusted mean difference in FRS between men/women with elevated and normal ALT levels.
FRS(diff)was then used to calculate the expected 10 year risk of CHD using formula:
where Survival(baseline) is the proportion of persons with baseline characteristics without CHD at 10 years and is given as 0.90015 for men and 0.96246 for women (as described by Wilson et al.21).
To increase the efficiency of the sampling process, NHANES III employed a complex, multistage sampling design in which potential participants were identified around 89 locations in the United States. This induces a correlation structure among the observations that cannot be treated as arising from a simple random sample. In addition, because NHANES III involved increased rates of sampling for certain age and racial groups, sample weights are provided to reflect this and also to attempt to adjust for nonresponse bias (due to people refusing to participate) and noncoverage bias (due to people who do not live in households and, therefore, could not participate). We used STATA version 9.1 software (StataCorp, College Station, TX) in our analyses to account for both the sampling and the weighting processes.
Among 7,526 participants without viral hepatitis or excessive alcohol consumption, 267 had elevated serum ALT activity, corresponding to 3.2% of the United States population. This finding is similar to previously reported estimates of the prevalence of ALT levels above 43 IU/L among persons without known causes of chronic liver disease.2 Persons with elevated ALT levels were younger and more likely to be male than persons with normal ALT levels (Table 1). After adjustment for sex and age, persons with elevated ALT levels had higher total cholesterol, lower high-density lipoprotein cholesterol, higher systolic blood pressure, and higher diastolic blood pressure and were more likely to be diabetic compared with persons with normal ALT levels (Table 1). The prevalence of smoking was similar in the 2 groups.
|Variable||Normal ALT (n = 7,259)||Elevated ALT (n = 267)||Age- and Sex-Adjusted Mean Difference||Age- and Sex-Adjusted Odds Ratio|
|Age (yr), mean||42.3 (41.4–43.1)||37.1 (34.9–39.2)||NA||—|
|Total cholesterol (mg/dL), mean||200.6 (199–202)||207.3 (200–215)||11.7 (4.6–18.8)||—|
|HDL cholesterol (mg/dL), mean||49.9 (49–51)||42.8 (40–46)||−4.4 (−7.3 to−1.5)||—|
|Systolic blood pressure (mm Hg), mean||117.6 (117–118)||119.6 (116–123)||2.9 (−0.3 to 6.1)||—|
|Diastolic blood pressure (mm Hg), mean||72.0 (71.4–72.5)||79.4 (77–82)||6.4 (4.4–8.4)||—|
|Diabetes||3.8% (3.2–4.4)||8.4% (4.3–12.5)||—||3.9 (2.1–7.4)|
|Smoker||25.7% (24–27)||24.8% (15–35)||—||0.8 (0.5–1.4)|
|Men||46.5% (45–48)||74.0% (66–82)||—||NA|
After adjustment for age, the mean FRS was higher in persons with elevated ALT levels compared with persons with normal ALT levels, both among men and among women (Table 2). The adjusted mean difference in the FRS of men with elevated ALT levels compared with men with normal ALT levels was 0.25 (95% CI 0.07-0.4); among women this adjusted mean difference was 0.76 (95% CI 0.4-1.1). These adjusted mean differences in FRS correspond to an adjusted hazard ratio for CHD of 2.14 (95% CI 1.5-3.0) in women and 1.28 (95% CI 1.07-1.5) in men. The risk of developing CHD over 10 years in a 50-year-old man (as predicted by the Framingham equation21) was 10.8% if the ALT level was normal and 13.6% if the ALT level was elevated (see Table 2 legend for calculation). Among women, the risk of developing CHD over 10 years in a 50-year-old was 5.9% if the ALT level was normal and 12.1% if the ALT level was elevated.
|Age-Adjusted Mean FRS*||Age-Adjusted Mean Difference in FRS||Age-Adjusted Hazard Ratio† of 10-year CHD risk||Average Risk of Developing CHD over 10 Years in a 50-Year-Old Subject‡|
|Normal ALT (n = 7,259)||Elevated ALT (n = 267)||Normal ALT||Elevated ALT|
|Men||2.5 (2.4–2.6)||2.75 (2.6–2.9)||0.25 (0.07–0.4)||1.28 (1.07–1.5)||10.8%||13.6%|
|Women||8.7 (8.7–8.8)||9.5 (9.0–10.0)||0.76 (0.4–1.1)||2.14 (1.5–3.0)||5.9%||12.1%|
When ALT levels were further categorized into 0-25, 26-30, 31-43, and >43 IU/L, it appeared that there was a stepwise increase in FRS that was more prominent in women than in men (Table 3). FRS increased appreciably after age adjustment for ALT values above 30 IU/L in women and above 43 IU/L in men.
|Age-Adjusted Mean FRS*||Age-Adjusted Mean Difference in FRS|
|ALT 0–25 (n = 6,482)||ALT 26–30 (n = 401)||ALT 31–43 (n = 376)||ALT >43 (n = 267)||ALT 26–30 vs. ALT 0–25||ALT 31–43 vs. ALT 0–25||ALT >43 vs. ALT 0–25|
|Men||2.5 (2.4–2.6)||2.6 (2.4–2.7)||2.6 (2.4–2.7)||2.8 (2.7–3.0)||0.14 (0.03 to 0.2)||0.10 (–0.02 to 0.2)||0.28 (0.09 to 0.46)|
|Women||8.7 (8.7–8.8)||8.9 (8.3–9.5)||9.2 (8.7–9.7)||9.5 (9.0–10.0)||0.21 (–0.03 to 0.4)||0.53 (0.3 to 0.8)||0.78 (0.4–1.1)|
After further adjusting for insulin resistance, body mass index, and waist-to-hip ratio, the adjusted mean difference in FRS comparing persons with elevated ALT levels with those with normal ALT levels was much reduced and no longer statistically significant (0.13, 95% CI −0.05 to 0.3 in men; 0.32, 95% CI −0.02 to 0.7 in women).
In analyses limited to persons with viral hepatitis or excessive alcohol consumption who were not obese (n = 674, including 81 with elevated ALT levels), elevated ALT activity was not associated with a higher FRS in men (adjusted mean difference 0.40, 95% CI −0.7 to −0.1) or women (adjusted mean difference −0.11, 95% CI −1.1 to 0.9).
This study suggests that persons with elevated serum ALT levels (>43 IU/L) in the absence of viral hepatitis or excessive alcohol consumption have a higher risk of CHD (as calculated using the FRS) than persons with normal ALT levels. The relative increase in risk is much greater in women (hazard ratio 2.14, 95% CI 1.5-3.0) than in men (hazard ratio 1.28, 95% CI 1.07-1.5). Despite the fact that women have a lower absolute risk of CHD than men, the absolute increase in risk of CHD associated with elevated ALT activity was also higher in women than in men. Given that elevated serum ALT in the absence of viral hepatitis or excessive alcohol consumption is most commonly due to NAFLD in the United States, our results suggest that NAFLD is associated with an increased risk of CHD.
The threshold level of ALT above which there was appreciable increase in the calculated risk of CHD was higher in men (>43 IU/L) than in women (>30 IU/L). This supports the use of a higher threshold for men than for women to define “abnormal” ALT. However, the absolute values should be interpreted with caution, because there may have been some degradation of ALT activity as a result of freezing serum specimens in NHANES III.25
Given the known associations between NAFLD (or elevated ALT activity) and several predictors of CHD such as insulin resistance, obesity, hypercholesterolemia, hypertriglyceridemia, and the metabolic syndrome,2–9 it has been previously suggested that NAFLD is likely to be associated with increased risk of CHD.10, 26 This study demonstrates and quantifies this excess risk in the United States population, separately for men and women, and for different thresholds of ALT level.
The higher FRS in persons with elevated ALT activity is primarily due to the higher total cholesterol level, diastolic blood pressure, and diabetes prevalence and the lower high-density lipoprotein cholesterol level in persons with elevated ALT levels compared with persons with normal ALT levels (Table 1). It is interesting to speculate whether these differences can be explained by the factors that are suspected to cause NAFLD (i.e., insulin resistance, obesity, and central fat distribution) or whether there is any excess calculated risk of CHD related to NAFLD over and above what would be expected due to the higher prevalence of insulin resistance, obesity, and central fat distribution. When we adjusted for these characteristics, the association between elevated ALT activity and calculated risk of CHD was reduced and was no longer statistically significant. This finding suggests that most of the excess Framingham-predicted risk of CHD associated with elevated ALT activity may be explained by insulin resistance, obesity, and central fat distribution, particularly because our adjustments for insulin resistance (approximated by the homeostasis model assessment) and central fat distribution (approximated by the waist-to-hip ratio) are likely to be incomplete.
Among persons with viral hepatitis or excessive alcohol consumption who were not obese, elevated ALT activity was not associated with a high FRS. Therefore, the higher FRSs that we identified in association with elevated ALT values in persons without viral hepatitis or excessive alcohol consumption are unlikely to be related to hepatic necroinflammation, which is measured according to the ALT level. Interestingly, we found that among men with viral hepatitis or excessive alcohol consumption, an elevated ALT level actually appeared to be associated with lower FRS. This finding is consistent with previous findings that atherosclerosis is rare in patients with advanced liver disease27 and mortality from diabetes-related events is low.28
Our study is limited by the absence of liver biopsy confirmation of the underlying liver disease responsible for the ALT elevation. However, liver biopsy would be impossible to perform routinely in a population-based study. We used the FRS to predict the risk of future CHD rather than follow the participants in time to determine the actual occurrence of CHD. However, use of the FRS for predicting CHD is appropriate in free-living, middle-aged persons, such as the majority of persons in this NHANES study. In addition, earlier versions of the Framingham CHD equations were applied with reasonable accuracy to the NHANES I epidemiological follow-up study29 (which has a similar population composition to the NHANES III population used in the current study) as well as samples from Honolulu, Puerto Rico, Albany, Chicago, Los Angeles, Minneapolis, Tecumseh, and the Western Collaborative Group.30–32 The FRS does not consider current use of medications to lower cholesterol or blood pressure. However, repeat analyses after excluding persons who reported taking such medication showed almost identical results (data not shown). The FRS was derived21 in populations aged 30-74 years, whereas our study population was over 18 years of age. The FRS may predict CHD risk less accurately in persons 18-29 years old.
In conclusion, elevated ALT activity among persons without viral hepatitis or excessive alcohol consumption is associated with a high risk of CHD as determined by FRS. This association is more prominent in women than in men. Most of this excess calculated risk of CHD in persons with elevated ALT levels appears to be related to an increased prevalence of insulin resistance, obesity, and central fat distribution, all of which are thought to be predisposing conditions of NAFLD.
|Age squared (yr)||−0.00268|
|Total cholesterol (mg/dL)|
|HDL cholesterol (mg/dL)|
|Stage I hypertension||0.52168||0.26288|
|Stage II-IV hypertension||0.61859||0.46573|
|Baseline survival function at 10 years||0.90015||0.96246|
- 3The prevalence and etiology of elevated aminotransferase levels in the United States. Am J Gastroenterol 2003; 98: 960–967., , .Direct Link:
- 15Prospective evaluation of unexplained chronic liver transaminase abnormalities in asymptomatic and symptomatic patients. Am J Gastroenterol 1999; 94: 3010–3014., , , , .Direct Link:
- 18Plan and Operation of the Third National Health and Nutrition Examination Survey, 1988-1994. Hyatsville, MD: National Center for Health Statistics, 1994.
- 25The prevalence and predictors of elevated serum aminotransferase activity in the United States in 1999-2002. Am J Gastroenterol 2006; 101: 76–82., , .Direct Link:
- 32The Framingham Study applied to four other U.S. based epidemiological studies of cardiovascular disease (Section No. 31). Bethesda, MD: US Department of Health, Education, and Welfare, NIH, 1976: 76–1083., .