fax: +49 3834 866684.
Steatohepatitis/Metabolic Liver Disease
Article first published online: 19 JUN 2009
Copyright © 2009 American Association for the Study of Liver Diseases
Volume 50, Issue 5, pages 1403–1411, November 2009
How to Cite
Haring, R., Wallaschofski, H., Nauck, M., Dörr, M., Baumeister, S. E. and Völzke, H. (2009), Ultrasonographic hepatic steatosis increases prediction of mortality risk from elevated serum gamma-glutamyl transpeptidase levels. Hepatology, 50: 1403–1411. doi: 10.1002/hep.23135
Potential conflict of interest: Nothing to report.
SHIP is part of the Community Medicine Net (http://www.medizin.uni-greifswald.de/cm) of the University of Greifswald
- Issue published online: 29 OCT 2009
- Article first published online: 19 JUN 2009
- Accepted manuscript online: 19 JUN 2009 12:00AM EST
- Manuscript Accepted: 15 JUN 2009
- Manuscript Received: 19 FEB 2009
- German Federal Ministry of Education and Research (BMBF). Grant Number: 01ZZ0403
- Ministry for Education, Research, and Cultural Affairs
- Ministry for Social Affairs of the Federal State of Mecklenburg, West Pomerania
- Competence Network Diabetes mellitus of Germany Federal Ministry of Education and Research
- Novo Nordisc
The aim of the present study was to investigate the association of serum gamma-glutamyltransferase (GGT) levels with all-cause mortality and to assess the impact of ultrasonographic findings of hepatic hyperechogenicity in that association. We used data from 4,160 subjects (2,044 men and 2,116 women) recruited for the population-based Study of Health in Pomerania (SHIP) without baseline hepatitis B and C infections or liver cirrhosis. GGT was divided into age- and sex-dependent quintiles to calculate overall and sex-specific crude incidence mortality rates. Hepatic steatosis was defined by elevated GGT levels (>80%) and the presence of hyperechogenic liver ultrasound. We used multiple-adjusted Cox proportional hazards regression models, first, to assess the direct effect of GGT on all-cause mortality, second, to stratify according to the ultrasonographic finding, and third, to investigate potential mediating effects of cardiometabolic risk factors. During 29,810 person-years (7.3 years, median) of follow-up, 307 individuals (7.5%) died, resulting in a death rate of 0.86 deaths per 1000 person-years. Elevated GGT levels were associated with increased risk of mortality in men (hazard ratio [HR] 1.49; 95% confidence interval [CI], 1.08–2.05), but not in women (HR 1.30; 95% CI, 0.80–2.12). This association was even stronger in men with hepatic steatosis (HR 1.98; 95% CI, 1.21–3.27). Cause-specific mortality analysis by cardiovascular disease deaths confirmed the sex-specific association. Adjustment for cardiometabolic risk factors did not affect the estimates. Conclusion: In the case of increased GGT levels, liver ultrasound should be performed, not only for diagnosis, but also for further risk stratification. (HEPATOLOGY 2009.)
Serum gamma-glutamyltransferase (GGT) has been widely used as an index of liver dysfunction and a marker of alcohol intake. But despite its clinical use, several recent epidemiological studies cast light on elevated GGT levels as an independent predictor of morbidity and mortality. Among 7,613 middle-aged British men followed for 11.5 years, elevated GGT levels were independently associated with a significant increase in all-cause and cardiovascular disease (CVD) mortality.1 Two prospective population-based studies in Austrian adults also reported an association of elevated GGT levels with increased risk of CVD mortality.2, 3 However, pathways linking GGT to mortality risk are not fully understood and several mediating variables have been proposed. For example, GGT was found to be positively associated with metabolic syndrome,4, 5 type 2 diabetes,6, 7 hypertension,8 and stroke.9, 10 Hepatic steatosis or fatty liver, as a common finding in the general population,11 has also been discussed as a possible indirect mechanism to explain the GGT/mortality association.12, 13 Given the links between GGT, hypertension, carotid atherosclerosis, type 2 diabetes,14–16 features of the metabolic syndrome,17 and hepatic steatosis,18 we assumed that these meditational variables might, at least in part, account for the relation of GGT with mortality.
The aim of the present study was two-fold. First, to investigate the direct effects of GGT levels on all-cause mortality in a population-based sample of 4160 subjects (2044 men and 2,116 women). And second, to assess the impact of ultrasonographic finding of hepatic hyperechogenicity in that association, as well as potential mediating effects of cardiometabolic variables.
Patients and Methods
The Study of Health in Pomerania (SHIP) is a population-based cohort study in West Pomerania, a region in northeastern Germany. Details on the SHIP study design and data have been published.19, 20 In brief, from the 212,157 inhabitants living in the area, a representative sample of 7008 subjects aged between 20 and 79 years was selected using population registries where all German inhabitants are registered. Only individuals with German citizenship and main residency in the study area were included. The net sample (without migrated or deceased persons) comprised 6,267 eligible subjects, whereof 4,310 finally participated (response proportion 68.8%). All participants gave written informed consent. The study conformed to the ethical guidelines of the 1975 Declaration of Helsinki as reflected in an a priori approval by the local Ethics Committee of the University of Greifswald.
Of the 4,310 SHIP participants, 54 had an uncertain diagnosis of liver echography, 14 tested positive for hepatitis B surface antigen (HBsAg), 20 for hepatitis C virus antibody (anti-HCV), and 11 had a known history of liver cirrhosis. Furthermore, 51 participants lacked measured GGT. In all, 150 participants were excluded from further analysis. This resulted in a study population of 4,160 subjects (2,044 men and 2,116 women) who were available for the present analysis.
Baseline data included socioeconomic characteristics, behavioral risk factors, individual medical history, as well as sonographic, laboratory, and somatometric examinations. Behavioral risk factors and socioeconomic characteristics were assessed by computer-aided personal interviews and included sex, age, educational level (<10, =10, or >10 years at school), civil status (living alone versus married or cohabiting), and “equalized” household income (in Euros). Because income is a household-level variable, the “equivalence” criteria allow for economies of scale at the household level. We used the commonly adopted procedure of the Luxembourg Income Study to divide the household income by the square root of the number of household members, thus assuming an equivalence parameter of 0.5.21 Subjects who participated in physical training during summer or winter for at least 1 hour a week were classified as being physically active. Alcohol consumption was evaluated as beverage-specific alcohol consumption (beer, wine, and distilled spirits) on the last weekend and last weekday preceding the examination, and the mean daily alcohol consumption was calculated using beverage-specific pure ethanol volume proportions.22 Medical history included a recall of physician's diagnoses of diabetes mellitus, stroke, and myocardial infarction. Peripheral artery disease was defined according the four stages proposed by Rose et al.23 Definition of heart failure was based on recommendations of the New York Heart Association,24 and angina pectoris on the Rose Angina Questionnaire.23 Comorbid health status was measured by the Functional Comorbidity Index, which is a summary measure of comorbid diseases selected and weighted according to their association with physical functioning.25
Sonographic examinations were performed by trained physicians who were unaware of the participant's clinical and laboratory characteristics. Liver ultrasound was performed using a 5 MHz transducer and a high-resolution instrument (Vingmed VST Gateway, Santa Clara, CA). A hyperechogenic liver ultrasound was defined as the presence of an ultrasound pattern of a bright liver, with evident density differences between hepatic and renal parenchyma.26 In SHIP, ultrasound examinations and readings underlie strict quality standards.27–30 For the laboratory examinations, nonfasting blood samples were drawn from the cubital vein in the supine position. The laboratory takes part quarterly in the official national German external proficiency testing programs. In addition, internal quality controls were analyzed daily. Markers of HBsAg and anti-HCV were determined by enzyme-linked immunosorbent assays (AxSym HBSAG and AxSym HCV, Abbott, Abbott Park, IL). High-sensitivity C-reactive protein (hs-CRP) was determined immunologically on a Behring Nephelometer II with commercially available reagents from Dade Behring (Eschborn, Germany). The assay was performed according to the manufacturer's recommendations with a test sensitivity of 0.2 mg/L. Serum GGT levels were measured photometrically (Hitachi 717; Roche Diagnostics, Mannheim, Germany), and categorized by sex-specific quintiles, expressed as μmol/L × s, which corresponds to (μmol/L × s) × 60 = IU/L. Assays were performed according to the manufacturers' recommendations. Hepatic steatosis was defined by elevated GGT levels (>80%) together with hyperechogenic liver ultrasound. Waist circumference (WC) was measured to the nearest 0.1 cm using an inelastic tape midway between the lower rib margin and the iliac crest in the horizontal plane, with the subject standing comfortably with weight distributed evenly on both feet. Information on vital status was acquired at regular intervals from the time of enrollment into the study through August 31, 2007. The mean duration of follow-up was 7.2 years (25th, 6.6; 75th, 8.0). Subjects were censored at the time of death, based on death certificates requested from the local health authority of the place of death. The number of months between baseline examination and censoring was used as follow-up length. Causes of deaths were coded by a certified nosologist according to the International Classification of Diseases, 10th revision (ICD-10). Additionally, two internists (H.W. and M.D.) independently validated the underlying cause of death and performed a joint reading in cases of disagreement. A third internist (H.V.) finally decided in cases of still existing disagreement.
Data on quantitative characteristics are expressed as mean with standard deviation. Data on qualitative characteristics are expressed as percent values. A χ2 test (qualitative data) or Mann-Whitney U test (quantitative data) were performed for intergroup comparisons by sex. GGT was divided by age- (in decades) and sex-specific quintiles (according cutoffs provided in Supporting Table 1) to calculate overall and sex-specific crude mortality incidence rates (per 1000 person-years). To assess the association between GGT and mortality we used Cox proportional hazards regression models. Three successive regression models were estimated accounting for different covariates: the first only included GGT quintiles as predictors, the second added age (in decades), and the third additionally adjusted for WC, alcohol consumption, physical activity, educational level, civil status, equalized income, and Functional Comorbidity Index. Next, the association of GGT with mortality was stratified by ultrasonographic finding of hepatic hyperechogenicity, assessing interaction by a product term of continuous GGT and bivariate liver ultrasound. To test for trend across quintiles of GGT, we estimated Cox models in which the categorical GGT variable appeared as a continuous predictor. Kaplan-Meier survival curves were used to illustrate the association of hepatic steatosis and mortality, with differences tested by a log-rank test. To compare the predictability of GGT with liver ultrasound we estimated area under receiver operating characteristic curves (AUCs area) from full-adjusted Cox models and tested their difference. Sensitivity analyses were performed excluding heavy drinkers (>60 g/day for men and >30 g/day for women), or nondrinkers (<10 g/day for men and women). Finally, models testing for mediation estimated the association of GGT with mortality in multivariate Cox models, controlling for the intermediate variable each in a separate model. We analyzed mediating effects of diabetes, a summative CVD score31 comprising angina pectoris, peripheral artery disease, heart failure, stroke, myocardial infarction, and hs-CRP. We used both graphical and hypotheses testing methods for examining the proportional hazard assumption.32 Schoenfeld's tests as well as visual inspection of smoothed estimates of the hazard ratio (HR) against time confirmed the proportional hazard assumption for all variables. HRs were calculated with a 95% confidence interval (95% CI). Two-sided probability values <0.05 were considered statistically significant. This article was written in accordance with the STROBE statement, giving guidelines for reporting of observational studies.33 All statistical analyses were performed using Stata 9 (Stata, College Station, TX).
Baseline characteristics stratified by sex are presented in Table 1. Compared to women, men were older, less often living alone, had higher income, and fewer years of schooling. Mean baseline levels of GGT and the prevalence of hyperechogenic liver ultrasound were significantly higher among men than women (Table 1). The overall prevalence of hepatic steatosis was 10%; 12.3% in men and 7.8% in women. Differences between men and women were also found in WC, alcohol consumption, hs-CRP, and prevalence of peripheral artery disease, heart failure, stroke, and myocardial infarction (Table 1).
|Characteristic||Men (N = 2044)||Women (N = 2116)||P Value*|
|Age, years||50.56 (16.6)||48.74 (16.2)||<0.001|
|Civil status, living alone||19.7||29.3||<0.001|
|Equalized household income, Euros||1003.6 (495.7)||931.8 (456.9)||<0.001|
|Educational level, years of schooling|
|Serum gamma-glutamyl transpeptidase, μmol/L s||0.80 (1.69)||0.37 (0.51)||<0.001|
|Hyperechogenic ultrasound pattern||38.1||22.2||<0.001|
|Waist circumference, cm||95.6 (11.7)||83.2 (13.1)||<0.001|
|Alcohol consumption, g/day||19.6 (23.0)||5.3 (9.0)||<0.001|
|High-sensitivity C-reactive protein, mg/L||2.79 (6.31)||2.99 (4.43)||<0.001|
|Functional Comorbidity Index||1.2 (1.3)||1.3 (1.4)||0.668|
|Peripheral artery disease||1.0||1.8||0.016|
During 29,810 person-years (7.3 years, median) of follow-up, 307 individuals (7.5%) died, resulting in a death rate of 0.86 deaths per 1000 person-years. Overall and sex-specific crude mortality incidence rates by quintiles of GGT are presented in Table 2. Overall rates tended to increase over GGT quintiles, whereas sex-specific rates showed differences only for subjects in the upper quintiles (Table 2). Crude, age-, and multiple-adjusted HRs are presented in Table 3. In men, Cox proportional hazard regression models revealed a positive association between elevated GGT levels and mortality in the crude (HR 1.52; 95% CI, 1.01–2.26) and age-adjusted model (HR 1.58; 95% CI, 1.06–2.36), but not in the multiple-adjusted model (HR 1.46; 95% CI, 0.94–2.05). We detected no evidence of trend over quintiles of GGT. Cox models with the highest GGT quintile versus all lower revealed positive associations between elevated GGT levels and mortality throughout all models (Table 3). In women, we detected no association between elevated GGT levels and mortality in any applied model, whereas a significant trend across GGT quintiles in crude and age-adjusted models was found (Table 3).
|GGT Levels in Quintiles (μmol/L s)|
|1. Quintile||2. Quintile||3. Quintile||4. Quintile||5. Quintile|
|Men and women|
|Number of deaths (crude incidence rate per 1000 person-years)||58 (8.9)||59 (9.7)||51 (8.9)||59 (10.2)||80 (14.0)|
|Number of deaths||41 (12.4)||43 (14.6)||39 (13.9)||35 (12.2)||57 (20.5)|
|Number of deaths||17 (4.9)||16 (5.1)||12 (4.1)||24 (8.2)||23 (7.8)|
|GGT||HR (95% CI) for All-Cause Mortality|
|2. Quintile||1.06 (0.69; 1.63)||1.06 (0.69; 1.62)||1.06 (0.68; 1.64)||1.04 (0.53; 2.06)||0.92 (0.47; 1.83)||0.97 (0.48; 1.95)|
|3. Quintile||1.02 (0.66; 1.58)||1.00 (0.65; 1.55)||1.00 (0.64; 1.57)||0.83 (0.40; 1.74)||0.82 (0.39; 1.71)||0.68 (0.32; 1.47)|
|4. Quintile||0.89 (0.57; 1.40)||0.91 (0.58; 1.42)||0.87 (0.54; 1.41)||1.67 (0.90; 3.10)||1.45 (0.78; 2.70)||1.22 (0.63; 2.36)|
|5. Quintile||1.52 (1.01; 2.26)*||1.58 (1.06; 2.36)*||1.46 (0.94; 2.24)||1.60 (0.86; 3.00)||1.62 (0.87; 3.04)||1.27 (0.65; 2.46)|
|5. vs. 1.-4. Quintile||1.52 (1.13; 2.06)*||1.59 (1.18; 2.16)*||1.49 (1.08; 2.05)*||1.42 (0.89; 2.28)||1.54 (0.96; 2.47)||1.30 (0.80; 2.12)|
|P for Trend||0.115||0.072||0.200||0.045||0.044||0.296|
We further ran Cox models for the association of GGT and mortality, stratified by ultrasonographic finding of hepatic hyperechogenicity (Table 4). The results indicate that elevated GGT was strongly associated with an increased risk of mortality in men with a hyperechogenic liver ultrasound (HR 3.01; 95% CI, 1.28–3.64). Furthermore, among men with a hyperechogenic liver, P for trend statistics revealed a significant trend for increased mortality risk across GGT quintiles. Additionally, Fig. 1 shows that subjects with hepatic steatosis had significantly shorter survival times compared to subjects without hepatic steatosis (log-rank test, P < 0.001). Hyperechogenic liver alone was not a predictor for mortality risk (age-adjusted model: men: HR, 0.92; 95% CI, 0.70; 1.21; women: HR, 1.22; 95% CI, 0.80; 1.84). Full adjusted models confirmed the missing association (men: HR, 0.77; 95% CI, 0.57; 1.04; women: HR, 0.98; 95% CI, 0.63; 1.53). Analyzing predictability, we found that GGT has the best predictive ability in terms of mortality risk with AUCs area of 0.774, compared to 0.609 for liver ultrasound. Test of differences in AUCs area revealed a P-value of <0.001. We further investigated the mediating effect of diabetes, CVD, and hs-CRP in the association of GGT with all-cause mortality (Table 5). Comparing models with and without the intermediate variables revealed that the addition of diabetes, CVD, or hs-CRP to the models predicting all-cause mortality by elevated GGT levels affected the GGT estimates only slightly.
|GGT in Quintiles||HR (95%-CI) for All-Cause Mortality†|
|N||Normal Ultrasound Pattern (N = 1,265)||N||Hyperechogenic Ultrasound Pattern (N = 779)||N||Normal Ultrasound Pattern (N = 1,646)||N||Hyperechogenic Ultrasound Pattern (N = 470)|
|1. Quintile||373||1.00 (ref.)||67||1.00 (ref.)||447||1.00 (ref.)||34||1.00 (ref.)|
|2. Quintile||304||0.82 (0.48; 1.40)||108||2.17 (0.89; 5.30)||364||1.14 (0.47; 2.76)||39||0.70 (0.22; 2.26)|
|3. Quintile||234||0.94 (0.55; 1.62)||141||1.60 (0.64; 3.97)||314||1.08 (0.42; 2.80)||103||0.34 (0.10; 1.26)|
|4. Quintile||205||0.96 (0.53; 1.74)||211||1.44 (0.57; 3.64)||289||1.71 (0.71; 4.09)||130||0.76 (0.27; 2.15)|
|5. Quintile||149||1.10 (0.63; 1.93)||252||3.10 (1.28; 7.50)*||232||1.52 (0.61; 3.78)||164||0.82 (0.27; 2.46)|
|5. vs. 1.–4. Quintile||1.18 (0.74; 1.91)||1.98 (1.21; 3.27)*||1.27 (0.61; 2.65)||1.25 (0.59; 2.67)|
|P for trend||0.658||0.046||0.224||0.973|
|GGT||HR (95% CI) for All-Cause Mortality|
|(Ref.: 1.Quintile)||Multiple-Adjusted†||+ Diabetes||+ CVD Score||+ hs-CRP||Multiple-Adjusted†||+ Diabetes||+ CVD Score||+ hs-CRP|
|2. Quintile||1.06 (0.68; 1.64)||1.06 (0.68; 1.64)||0.99 (0.63; 1.55)||1.05 (0.68; 1.64)||0.97 (0.48; 1.95)||0.97 (0.48; 1.95)||0.93 (0.46; 1.91)||0.98 (0.47; 2.04)|
|3. Quintile||1.00 (0.64; 1.57)||1.00 (0.64; 1.57)||0.99 (0.63; 1.54)||0.96 (0.60; 1.52)||0.68 (0.32; 1.47)||0.65 (0.30; 1.42)||0.70 (0.32; 1.50)||0.70 (0.47; 1.57)|
|4. Quintile||0.87 (0.54; 1.41)||0.87 (0.54; 1.41)||0.85 (0.52; 1.37)||0.85 (0.52; 1.39)||1.22 (0.63; 2.36)||1.14 (0.58; 2.24)||1.24 (0.64; 2.40)||1.28 (0.64; 2.55)|
|5. Quintile||1.46 (0.94; 2.24)||1.46 (0.94; 2.24)||1.39 (0.90; 2.15)||1.37 (0.87; 2.14)||1.27 (0.65; 2.46)||1.20 (0.61; 2.36)||1.27 (0.65; 2.49)||1.26 (0.62; 2.56)|
|5. vs. 1.–4. Quintile||1.49 (1.08; 2.05)*||1.49 (1.08; 2.05)*||1.46 (1.06; 2.02)*||1.43 (1.02; 2.00)*||1.30 (0.80; 2.12)||1.27 (0.78; 2.08)||1.30 (0.79; 2.15)||1.27 (0.76; 2.11)|
|P for Trend||0.200||0.201||0.243||0.341||0.296||0.415||0.277||0.302|
Cause-specific mortality analysis revealed a more than 2-fold increased risk of CVD mortality from elevated GGT levels for both sex in crude and age-adjusted models, but only for men in the multiple-adjusted model (Table 6). Among these, we also detected a positive trend across quintiles of GGT, whereas Cox models with the highest GGT quintile versus all lower confirmed the aforementioned results only for men (Table 6). Stratified by ultrasonographic finding, elevated GGT was strongly associated with an increased risk of CVD mortality in men with hepatic steatosis (HR 6.22; 95% CI, 1.22–31.62), with a significant trend of increased mortality risk across GGT quintiles (P = 0.022). In women, further stratification by liver ultrasound did not improve risk prediction considerably (Table 7).
|GGT||HR (95% CI) for CVD Mortality|
|(Ref.: 1. Quintile)||Crude||Age-Adjusted||Multiple-Adjusted†||Crude||Age-Adjusted||Multiple-Adjusted†|
|2. Quintile||1.52 (0.68; 3.38)||1.47 (0.66; 3.28)||1.67 (0.72; 3.88)||1.83 (0.44; 7.65)||1.54 (0.37; 6.46)||1.52 (0.36; 6.43)|
|3. Quintile||1.62 (0.73; 3.61)||1.59 (0.71; 3.54)||1.86 (0.81; 4.30)||1.94 (0.46; 8.11)||1.84 (0.44; 7.71)||1.22 (0.28; 5.33)|
|4. Quintile||1.46 (0.65; 3.29)||1.42 (0.63; 3.19)||1.48 (0.61; 3.60)||3.52 (0.95; 12.99)||2.92 (0.79; 10.79)||1.85 (0.47; 7.21)|
|5. Quintile||2.44 (1.15; 5.15)*||2.58 (1.22; 5.45)*||2.80 (1.24; 6.31)*||3.90 (1.08; 14.19)*||3.78 (1.04; 13.73)*||2.34 (0.61; 8.96)|
|5. vs. 1.–4. Quintile||1.75 (1.06; 2.87)*||1.89 (1.14; 3.10)*||1.86 (1.10; 3.14)*||1.93 (0.91; 4.08)||2.07 (0.98; 4.36)||1.64 (0.74; 3.66)|
|P for Trend||0.032||0.022||0.027||0.013||0.014||0.166|
|GGT in Quintiles||HR (95%-CI) for All-Cause Mortality†|
|N||Normal Ultrasound Pattern (N = 1,265)||N||Hyperechogenic Ultrasound Pattern (N = 779)||N||Normal Ultrasound Pattern (N = 1,646)||N||Hyperechogenic Ultrasound Pattern (N = 470)|
|1. Quintile||373||1.00 (ref.)||67||1.00 (ref.)||447||1.00 (ref.)||34||1.00 (ref.)|
|2. Quintile||304||1.57 (0.59; 4.22)||108||2.14 (0.39; 11.81)||364||2.02 (0.18; 22.74)||39||0.93 (0.14; 6.60)|
|3. Quintile||234||1.65 (0.59; 4.65)||141||3.45 (0.68; 17.43)||314||2.88 (0.28; 29.76)||103||0.42 (0.05; 3.59)|
|4. Quintile||205||1.37 (0.43; 4.39)||211||2.92 (0.57; 15.03)||289||3.94 (0.40; 38.67)||130||0.81 (0.13; 5.03)|
|5. Quintile||149||2.25 (0.80; 6.31)||252||6.22 (1.22; 31.62)*||232||7.31 (0.84; 63.22)||164||0.98 (0.11; 8.84)|
|5. vs. 1.–4. Quintile||1.63 (0.75; 3.53)||2.41 (1.05; 5.55)*||3.13 (1.02; 9.67)*||1.41 (0.32; 6.22)|
|P for trend||0.193||0.022||0.030||0.991|
Sensitivity analysis with study population stratified by sex and riskful alcohol consumption (<10/>10 g/day in women and <20/>20 g/day in men), revealed no subgroup differences concerning risk of mortality from elevated GGT among both men (riskful alcohol consumption, HR, 1.62, 95% CI, 0.97–2.33; no riskful alcohol consumption, HR, 1.83, 95% CI, 1.06–3.19) and women (riskful alcohol consumption, HR, 1.46, 95% CI, 0.83–2.56; no riskful alcohol consumption, HR, 1.68, 95% CI, 0.97–3.12). Additional sensitivity analyses excluded subjects due to their alcohol consumption habits. Excluding heavy drinkers did not change the estimates at all. Conversely, excluding nondrinkers increased the risk of mortality from hepatic steatosis, but only for males (data not shown).
The present study revealed a direct association of elevated GGT levels with all-cause and CVD mortality in men. Furthermore, this association was even stronger among those with a hyperechogenic liver ultrasound. To our knowledge this is the first population-based cohort study to further stratify the association between elevated GGT levels and mortality according to the ultrasonographic finding of hepatic hyperechogenicity. Our finding of a sex-specific pattern is in agreement with previous prospective studies reporting that women had lower overall GGT levels,34 and showed weaker associations of elevated GGT levels with mortality than men.3 Possible explanations account for the positive correlation of longitudinal GGT change with an increase in CVD risk factors like body mass index, total serum cholesterol, high-density lipoprotein (HDL) cholesterol, and number of cigarettes per day in men.35 This explanation is supported by our finding of a sex-specific positive association of GGT with CVD mortality. Although cause-specific estimates for risk of mortality from CVD showed much wider confidence intervals compared to all cause analysis, our findings suggest sex-specific interactions among CVD risk factors, GGT, and mortality. However, due to the small number of CVD deaths in our sample, analyses should be replicated in larger samples with more events.
Men with high GGT levels have higher mortality, partly because of the association between GGT and other risk factors, and partly because GGT itself might act as an independent predictor of mortality. The latter is supported by studies linking elevated GGT levels directly to increased risk of CVD and all-cause mortality.1–3, 36, 37 Parallel evidence for indirect effects emerges from studies suggesting that higher GGT levels are associated with CVD risk factors, including metabolic syndrome, diabetes, hypertension, and dyslipidemia.4–10 GGT levels also correlate positively with novel cardiovascular risk factors, such as CRP,38 fibrinogen, F2-isoprostanes,39 and inversely with antioxidant levels.40 Thus, high GGT levels are suggested as an early marker of oxidative stress and inflammation, explaining various associations in the pathway of GGT-disease-mortality. However, considering GGT as a multifunctional protein,37 mechanisms that explain the contribution and underlying factors of GGT to CVD and mortality have not been fully elucidated.
When we investigated the potential role of hyperechogenic liver ultrasound in the pathway from GGT to mortality, we found that men with hepatic steatosis exposed significantly shorter survival times compared to subjects without hepatic steatosis. Although increased liver enzymes are insensitive at detecting hepatic steatosis,41 the sensitivity of ultrasound also varies between 49% and 94%.42 In population-based cohorts, ultrasonographic evidence of hepatic steatosis is found in almost 20% of normal weight persons and 80% of overweight persons.43 Thus, we determined hepatic steatosis by using both laboratory markers as well as ultrasound. Doing so, risk of mortality from hepatic steatosis was higher compared to assessment of elevated GGT levels only, suggesting that in case of increased GGT levels, liver ultrasound should be performed—not only for diagnosis, but also for further risk specification. Furthermore, we compared the predictability of GGT and liver ultrasound and found that GGT is the better predictor for mortality risk compared to liver ultrasound. This finding supports our conclusion that in the need for risk assessment, GGT has the best predictive ability.
Finally, we analyzed cardiometabolic variables as mediators in the pathway of GGT-mortality. As development of diabetes is predicted by high GGT levels,7, 34 even if in its reference range, diagnosis of diabetes may mediate the revealed association between GGT and mortality. But inclusion of diabetes in a model that predicted all-cause mortality by elevated GGT levels did not affect the estimates, as well as inclusion of a summative CVD score. Despite previous studies suggesting a direct association of serum GGT with CRP,38 and growing evidence that hs-CRP acts as an independent mortality risk factor,44, 45 our mediation analyses with the inclusion of hs-CRP only modestly changed the estimates. Sensitivity analysis considering alcohol consumption also lacked an impact on the pathway from GGT to mortality, suggesting that elevated GGT levels acts as an independent risk factor for mortality.
Our study has several strengths and potential limitations that should be considered. Major strengths are the population-based design, ultrasound examinations performed by standardized protocol and trained medical staff, and the routinely documented alcohol consumption. Despite the fact that liver biopsy is the best diagnostic tool for confirming diagnosis of hepatic steatosis, the inability to perform this in a large-scale population-based study like SHIP arises from logistic concerns.
In summary, the present study reports a direct association of elevated GGT levels and ultrasonographic evidence of hepatic steatosis with all-cause mortality in men. When we assessed the potential role of hyperechogenic liver ultrasound in the pathway from GGT to mortality, we found increased risk of mortality in men with hepatic steatosis. Our findings strongly suggest that in case of increased GGT levels, liver ultrasound should be performed, not only for diagnosis, but also for further risk stratification. Although our findings need to be confirmed in other populations, longitudinal monitoring of GGT changes may be beneficial, because many subjects will experience death before they are able to develop liver disease, and especially because GGT is a low-cost and widely used laboratory measurement.
- 1Gamma-glutamyltransferase: determinants and association with mortality from ischemic heart disease and all causes. Am J Epidemiol 1995; 142: 699–708., , .
- 2Gamma-glutamyltransferase as a risk factor for cardiovascular disease mortality: an epidemiological investigation in a cohort of 163,944 Austrian adults. Circulation 2005; 112: 2130–2137., , , , , .
- 3Longitudinal change in serum gamma-glutamyltransferase and cardiovascular disease mortality: a prospective population-based study in 76,113 Austrian adults. Arterioscler Thromb Vasc Biol 2008; 28: 1857–1865., , , , , , et al.
- 4Serum gamma-glutamyltransferase and risk of metabolic syndrome and type 2 diabetes in middle-aged Japanese men. Diabetes Care 2004; 27: 1427–1432., , .
- 5Gamma-glutamyl transpeptidase and the metabolic syndrome. J Intern Med 2000; 248: 230–238., , , , , .
- 6Three-year increase of gamma-glutamyltransferase level and development of type 2 diabetes in middle-aged men and women: the D.E.S.I.R. cohort. Diabetologia 2006; 49: 2599–2603., , , , .
- 7Gamma-glutamyltransferase is a predictor of incident diabetes and hypertension: the Coronary Artery Risk Development in Young Adults (CARDIA) Study. Clin Chem 2003; 49: 1358–1366., , , , , , et al.
- 8Serum gamma-glutamyl transferase level in predicting hypertension among male drinkers. J Hum Hypertens 1994; 8: 445–449., , , , , , et al.
- 9Gamma-glutamyltransferase and risk of stroke: the EUROSTROKE project. J Epidemiol Commun Health 2002; 56(Suppl 1): i25–i29., , , , , , et al.
- 10Serum gamma-glutamyl transferase, self-reported alcohol drinking, and the risk of stroke. Stroke 2000; 31: 1851–1855., , .
- 11The epidemiology of nonalcoholic fatty liver disease in adults. J Clin Gastroenterol 2006; 40(Suppl 1): S5–S10..
- 12Effect of body mass and other factors on serum liver enzyme levels in men attending for well population screening. Ann Clin Biochem 1989; 26(Pt 5): 393–400., .
- 13Fat distribution and gamma-glutamyl transferase in relation to serum lipids and blood pressure in 38-year old Dutch males. Eur J Clin Nutr 1989; 43: 809–818., , , .
- 14Nonalcoholic steatohepatitis and the metabolic syndrome. Am J Med Sci 2005; 330: 326–335..
- 15Hepatic steatosis is associated with an increased risk of carotid atherosclerosis. World J Gastroenterol 2005; 11: 1848–1853., , , , , , et al.
- 16Menopausal status and hepatic steatosis in a general female population. Gut 2007; 56: 594–595., , , , , , et al.
- 17Nonalcoholic fatty liver disease and the metabolic syndrome. Minerva Cardioangiol 2006; 54: 229–239., .
- 18Non-alcoholic fatty liver disease and increased risk of cardiovascular disease. Atherosclerosis 2007; 191: 235–240., .
- 19Extended recruitment efforts minimize attrition but not necessarily bias. J Clin Epidemiol 2009; 62: 252–260., , , , , , et al.
- 20Study of Health In Pomerania (SHIP): a health examination survey in an east German region: objectives and design. Sozial Praventivmed 2001; 46: 186–194., , , , , , et al.
- 21The relationship of income inequality to mortality: does the choice of indicator matter? Soc Sci Med 1997; 45: 1121–1127., .
- 22Laboratory markers carbohydrate-deficient transferrin, gamma-glutamyltransferase, and mean corpuscular volume are not useful as screening tools for high-risk drinking in the general population: results from the Study of Health in Pomerania (SHIP). Alcohol Clin Exp Res 2004; 28: 931–940., , , .
- 23Cardiovascular survey methods. Geneva: World Health Organization; 2004., , , .
- 24NYHA. Nomenclature and Criteria for Diagnosis of Diseases of the Heart and Great Vessels. 9th ed. Boston: Little Brown; 1994: 253–255.
- 25The development of a comorbidity index with physical function as the outcome. J Clin Epidemiol 2005; 58: 595–602., , , .
- 26Prevalence of and risk factors for hepatic steatosis in Northern Italy. Ann Intern Med 2000; 132: 112–117., , , , , , et al.
- 27The association of thyroid function with cardiac mass and left ventricular hypertrophy. J Clin Endocrinol Metab 2005; 90: 673–677., , , , , , et al.
- 28Methods for quality assurance of medical examination in epidemiological field studies: the “Study of Health in Pomerania” (SHIP).] Gesundheitswesen (Bundesverband der Arzte des Offentlichen Gesundheitsdienstes, Germany) 2000; 62: 234–243., , , , , , et al. [
- 29Thyroid function and carotid wall thickness. J Clin Endocrinol Metab 2004; 89: 2145–2149., , , , , , et al.
- 30Risk factors for goiter in a previously iodine-deficient region. Exp Clin Endocrinol Diabetes 2005; 113: 507–515., , , , , , et al.
- 31A score for predicting risk of death from cardiovascular disease in adults with raised blood pressure, based on individual patient data from randomised controlled trials. BMJ (Clin Res) 2001; 323: 75–81., , , , , .
- 32Applied Survival Analysis: Regression Modeling of Time to Event Data. Hoboken, NJ: John Wiley & Sons; 2008., .
- 33The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 2007; 370: 1453–1457., , , , , .
- 34gamma-Glutamyltransferase, obesity, and the risk of type 2 diabetes: observational cohort study among 20,158 middle-aged men and women. J Clin Endocrinol Metab 2004; 89: 5410–5414., , , , .
- 35Seven-year longitudinal population study of change in gamma-glutamyltransferase: the Tromso Study. Am J Epidemiol 1994; 139: 787–792., .
- 36Distribution, determinants, and prognostic value of gamma-glutamyltransferase for all-cause mortality in a cohort of construction workers from southern Germany. Prevent Med 1997; 26: 305–310., , , , , .
- 37Gamma glutamyl transferase and metabolic syndrome, cardiovascular disease, and mortality risk: the Framingham Heart Study. Arterioscler Thromb Vasc Biol 2007; 27: 127–133., , , , , , et al.
- 38Association between serum gamma-glutamyltransferase and C-reactive protein. Atherosclerosis 2005; 178: 327–330.,
- 39Is serum gamma glutamyltransferase a marker of oxidative stress? Free Radic Res 2004; 38: 535–539., ,
- 40Is serum gamma-glutamyltransferase inversely associated with serum antioxidants as a marker of oxidative stress? Free Radic Biol Med 2004; 37: 1018–1023., , , , , .
- 41Molecular mediators of hepatic steatosis and liver injury. J Clin Invest 2004; 114: 147–152., .
- 42The role of ultrasound in the diagnosis of hepatic steatosis in morbidly obese patients. Obes Surg 2004; 14: 635–637., , , , , , et al.
- 43The spectrum of liver disease in the general population: lesson from the Dionysos study. J Hepatol 2001; 35: 531–537., .
- 44Serum C-reactive protein levels can be used to predict future ischemic stroke and mortality in Japanese men from the general population. Atherosclerosis 2009; 204: 234–238., , , , , , et al.
- 45C-reactive protein and all-cause mortality in a large hospital-based cohort. Clin Chem 2008; 54: 343–349., , , , , , et al.
Additional Supporting Information may be found in the online version of this article.
|HEP_23135_sm_SuppTable1.doc||44K||Supporting Table 1: Age and sex-dependent quintiles of baseline serum gamma-glutamyl transpeptidase (GGT) levels in μmol/L.|
Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.