- Email: [email protected]
Gamma-glutamyltransferase, general and cause-specific mortality in 19,000 construction workers followed over 20 years
Research Article
Gamma-glutamyltransferase, general and cause-specific mortality in 19,000 construction workers followed over 20 years Lutz Philipp Breitling1,⇑, Heiner Claessen1, Christoph Drath2, Volker Arndt1, Hermann Brenner1 1
Division of Clinical Epidemiology and Aging Research, German Cancer Research Center, D-69120 Heidelberg, Germany; 2Workmen’s Compensation Board for Construction Workers, Occupational Health Service, D-71029 Böblingen, Germany
Background & Aims: Associations of serum gamma-glutamyltransferase levels with a variety of incident diseases and mortality have been suggested. The present study attempted to expand the body of evidence to especially relevant occupational cohorts in which exposure to established c-GT determinants may greatly differ from the general population. Methods: The study was based on occupational health examinations conducted from 1986 to 1992 in 19,090 German male workers from the construction industry, aged 25–64 years. Sociodemographics and other health-related information were collected during the exam. Vital status follow-up was conducted through 2008. Associations of baseline c-GT levels (measurements at 25 °C) with all-cause and cause-specific mortality were examined by Kaplan–Meier plotting and multiple adjusted Cox regression models. Results: A total of 2170 deaths occurred during 303,198 personyears of follow-up. The risk of death due to any cause was 2.5fold increased in subjects in the highest (P39 U/L) versus lowest (<11 U/L) c-GT quintile. To varying extents, elevated c-GT was associated with higher mortality due to cancer, circulatory, respiratory, and digestive causes, as well as accidents/poisoning. Conclusions: The findings in this cohort provide evidence for c-GT being associated with a broad range of causes of death, including less investigated outcomes. Some characteristics of the observed patterns need to be seen in the context of our cohort, featuring particularly high c-GT levels. Ó 2011 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.
Introduction The liver enzyme gamma-glutamyltransferase (c-GT) is an emergent risk marker for a variety of common diseases such as diabetes and cancer [3,18,19,28], as well as for overall and cause-specific mortality [1,5,16,21,22,31]. Apart from alcohol consumption as
Keywords: Epidemiological study; Prospective study; Liver function tests; Mortality. Received 19 January 2010; received in revised form 22 October 2010; accepted 6 December 2010; available online 21 January 2011 ⇑ Corresponding author. Address: Division C070 of Clinical Epidemiology and Aging Research, German Cancer Research Center, INF 581 (TP4), D-69120 Heidelberg, Germany. Tel.: +49 6221 421343; fax: +49 6221 421302. E-mail address: [email protected] (L.P. Breitling).
the most classical determinant of individual c-GT levels [8], a number of potentially modifiable predictors of c-GT have been identified in epidemiological studies, including body mass index (BMI), coffee consumption, and smoking [1,3,4,20,24,25,28,35]. In occupational settings, c-GT is assessed as part of routine health examinations both due to its role not only as a marker of alcohol abuse, but also to the increased risk of hepatotoxic occupational exposures in certain professions [1,17 and references therein]. Construction workers are of particular interest in this regard, because of the frequent accumulation of both life style (alcohol consumption, smoking) and occupational risk factors, including an elevated potential for frequent exposure e.g. to solvents used in paints and adhesives, heavy metals, vinyl chloride, and other hepatotoxic substances [10,17]. We have found c-GT to be a strong predictor of mortality in a cohort of construction workers in Germany [5]. However, this analysis was restricted to all-cause mortality due to limitations in both the size of the cohort and the length of follow-up. After substantial enlargement of the cohort and accumulation of additional years of follow-up, here we provide a detailed analysis of both all-cause and cause-specific mortality for major causes of death.
Materials and methods Design, setting, and participants The analyses presented were based on a prospective cohort study of disability and mortality in employees of the construction industry in the south of Germany [1,5,6]. In brief, all employees in Germany are entitled and invited to routine occupational health examinations free of charge. Male subjects, participating in such an examination (>75% of those invited during the recruitment period) at any of eight health centres of the Workmen’s Compensation Board for construction workers in Württemberg from August 1986 to December 1992, were included in this study. Whereas participation in such exams is non-mandatory for most occupation groups according to German occupational safety laws, data obtained in such exams are to be collected and analyzed. The proportion of invitees participating in the exam thus directly reflects the participation rate for the present study. The study protocol was approved by the ethics committees of the Universities of Heidelberg and Ulm, and by the Baden-Württemberg state ministry of social affairs. Data collection Occupational health examinations were conducted by experienced occupational health physicians and included detailed standardised questionnaires regarding occupational and life style factors, physical, and functional exams; and standard
Journal of Hepatology 2011 vol. 55 j 594–601
JOURNAL OF HEPATOLOGY laboratory measurements. Collected data of relevance to the present analysis included sociodemographics, body weight and height, status and intensity of smoking and alcohol consumption; and prevalent morbidity coded according to the International Classification of Diseases revision 9 (ICD-9). Serum concentrations of c-GT were measured centrally using a Hitachi 705/717 instrument, which measured at 25 °C with an upper reference limit for males of 28 U/L (dividing this value by a temperature conversion factor of 0.57 [29], one would obtain a corresponding reference limit of 49 U/L at 37 °C). Follow-up information on vital status was obtained in several waves, starting in 1992, from regional offices of the German pension fund. For the present analysis, follow-up was updated from January 2006 to March 2008. For deceased subjects, the main cause of death was identified from death certificates obtained from public health authorities, recoded to ICD-9 if necessary, and assigned to one of the following categories: cancer (ICD-9 140–208), circulatory (ICD-9 390–459), respiratory (ICD-9 460–519), digestive (ICD-9 520–579), or accidents/poisoning (ICD-9 800–999). Coding of cause of death was independently carried out by two trained investigators and any discrepancy was resolved in consensus after additional review. Deceased subjects with missing information on the cause of death were excluded from cause-specific analyses. Statistical analysis The study population was first characterised regarding sociodemographics (age, nationality, occupational group), serum c-GT, and important co-variables potentially confounding the c-GT-mortality relationship. Furthermore, aspartate (AST) and alanine transaminases (ALT) and their ratio were reported as markers of liver damage (measured on the same instrument as c-GT, with upper normal limits of 18 and 22 U/L, respectively), and additional laboratory markers assessed as part of the routine were included as indicators of metabolic syndrome-related alterations (blood glucose, triglycerides, and cholesterol; dichotomized according to assay specifications at 100, 150, and 220 mg/dl, respectively). The association of quintiles of c-GT with all-cause and cause-specific mortality were examined by Kaplan–Meier plots. After graphical check of the proportionality of hazards assumption, separate Cox regression models for all-cause and each individual cause-specific mortality were fit adjusting either only for age at baseline (<35, 35–44, 45–54, 55–64 years), or for age, nationality (German, Italian, Turkish, Yugoslavian, any other), occupational group (bricklayer, carpenter, painter, plasterer, plumber, unskilled worker), BMI (<25, 25–29.9, P30 kg/m2) [34], prevalent diseases (diabetes [ICD-9 250], ischemic heart disease [ICD-9 410–414], hypertension [ICD-9 401–405], and smoking (never, formerly,<20, 20, >20 cigarettes per day, or smoking including tobacco products other than cigarettes) [’’full model’’], or additionally for the presence of elevated blood glucose, elevated triglycerides, and elevated cholesterol. Missing values in co-variables were dealt with by multiple imputation employing the Monte Carlo Markov Chain method implemented in SAS 9.2 (PROC MI in combination with PROC MIANALYZE) [23]. This method assumes that the variables in the dataset originate from a multivariate normal distribution and creates multiple datasets with the missing values filled with random number according to this multivariate normal. Thereafter, model parameters, such as hazard ratios are estimated and combined across the multiple imputed datasets. The procedure is assumed to be fairly robust across deviations from normality, e.g. extendable to categorical variables [23]. Importantly, estimates in complete cases analyses were very similar to the results presented here (not shown). To further quantify discriminatory performance of c-GT, area under the curve (AUC) values were estimated in logistic regression models for the various endpoints, including only co-variables or also log(c-GT) as predictors in complete cases analyses, using the same adjustment set as in the full model. Miscellaneous models explored the adjusted associations when restricting digestive mortality to deaths due to chronic liver disease/cirrhosis (ICD-9 571), and when excluding deaths from hepatocellular carcinoma (ICD-9 155) from cancer mortality models. Sensitivity analyses included the comparison with additional Cox regression models further adjusting for alcohol consumption intensity (none, occasional, 1–30, 31–60, 61–90, or P91 g per day). Furthermore, the impact of excluding subjects with prevalent disease (diabetes, ischemic heart disease, hypertension [as defined above], cerebrovascular disease [ICD-9 430– 438], cancer [ICD-9 140–208]), or of excluding subjects with c-GT beyond the 99th percentile (i.e. in excess of 260 U/L), or of stratifying the baseline hazard by year of recruitment was examined. The dose–response relationship was explored by modeling c-GT using cubic spline functions with knots at the 5th, 25th, 75th, and 95th percentile as suggested in the pertinent literature [14]. Statistical analyses were carried out using SAS 9.2 for Windows (SAS Institute, Cary, NC, USA), tests being two-sided with a = 0.05 throughout.
Table 1. Baseline characteristics of 19,090a male construction workers in Germany 1986–1992.
Baseline characteristics Age group 25 to 34 years 35 to 44 years 45 to 54 years 55 to 64 years Nationality German Italian Turkish Yugoslavian Other Occupation Bricklayer Carpenter Painter Plasterer Plumber Unskilled worker Body mass index <25 kg/m2 25 to <30 kg/m2 ≥30 kg/m2 Prevalent disease Diabetes mellitus Ischemic heart disease Hypertension Cirrhosisc Dysglycemia and lipidemia Glucose <100 mg/dl Glucose ≥100 mg/dl Triglycerides <150 mg/dl Triglycerides ≥150 mg/dl Cholesterol <220 mg/dl Cholesterol ≥220 mg/dl Smoking Never Formerly <20 cig./day 20 cig./day >20 cig./day Other tobacco Alcohol consumption None Occasional 1-30 g/day 31-60 g/day 61-90 g/day ≥91 g/day b Serum transaminases AST ≤18 U/L AST >18 U/L ALT ≤22 U/L ALT >22 U/L AST/ALT ratio ≤2 AST/ALT ratio >2 Serum γ-GTb <11 U/L 11-<15 U/L 15-<22 U/L 22-<39 U/L ≥39 U/L a b c
Serum γ-GT (U/L)b Median (IQR)
Na
%
5,749 4,008 6,311 3,022
30 21 33 16
15 20 20 20
(11-25) (13-37) (13-37) (13-33)
14,149 1,318 1,215 1,761 590
74 7 6 9 3
19 18 12 18 17
(13-35) (12-29) (9-17) (12-31) (11-30)
6,104 2,583 2,888 1,937 2,725 2,853
32 14 15 10 14 15
19 17 18 19 18 17
(12-35) (12-30) (12-34) (12-35) (12-31) (12-31)
6,689 9,094 2,933
36 49 16
14 20 26
(10-24) (13-34) (17-46)
932 315 4,163 22
5 2 22 0.1
27 24 27 75
(17-53) (16-41) (16-51) (37-207)
12,150 6,930 10,070 9,009 9,017 10,068
64 36 53 47 47 53
17 22 15 24 15 23
(11-29) (14-40) (11-24) (15-43) (10-24) (15-41)
3,927 2,900 2,718 3,730 2,085 242
25 19 17 24 13 2
17 20 17 18 21 20
(11-28) (13-37) (11-32) (12-33) (13-40) (13-34)
1,723 6,296 1,758 3,403 1,667 1,710
10 38 11 21 10 10
13 16 17 22 28 38
(9-18) (11-25) (12-28) (14-39) (17-52) (21-76)
16,917 2,085 15,077 3,980 18,811 110
89 11 79 21 99 1
17 68 16 42 18 24.5
(12-27) (36-130) (11-25) (25-79) (12-33) (12-198)
3,280 3,674 4,181 4,081 3,874
17 19 22 21 20
9 12 17 28 63.5
(8-10) (11-13) (16-19) (24-32) (48-100)
Values not always adding up to 19,090 due to missing values. Measured at 25 °C. ICD-9 571.2, 571.5, and 571.6.
Journal of Hepatology 2011 vol. 55 j 594–601
595
Research Article All-cause mortality
Cancer mortality (ICD-9 140-208) 1.00 Proportion of survivors
Proportion of survivors
1.00 0.95 0.90 0.85 0.80 0.75
0.95
0.90
0.85
0.70 0
5
10 Years
15
20
0
Circulatory mortality (ICD-9 390-459)
15
20
Respiratory mortality (ICD-9 460-519) Proportion of survivors
0.95
0.90
0.99 0.98 0.97 0.96 0.95
0.85 0
5
10 Years
15
20
0
Digestive mortality (ICD-9 520-579)
5
10 Years
15
20
Accidents/poisoning (ICD-9 800-999) 1.00 Proportion of survivors
1.00 Proportion of survivors
10 Years
1.00
1.00 Proportion of survivors
5
0.99 0.98 0.97 0.96
0.99 0.98 0.97 0.96 0.95
0.95 0
5
10 Years
15
20
0
5
10 Years
15
20
Fig. 1. Kaplan–Meier curves showing the association of c-GT quintiles (at 25 °C) with all-cause and cause-specific mortality. Line types by quintile: 1 = solid black (<11 U/L); 2 = dashed gray (11–<15 U/L); 3 = solid green (15–<22 U/L); 4 = dashed blue (22–<39 U/L); 5 = solid red (P39 U/L). For respiratory and digestive mortality, quintiles 1–3 combined are shown in black, whereas 4th and 5th quintile are gray and green.
Results Description of the study population Serum c-GT values and vital status follow-up information were available for 19,090 (95.8%) of 19,930 initially included individuals. Baseline characteristics are presented in Table 1. The majority 596
of participants were German, and bricklayers constituted the largest occupational group. Median (interquartile range [IQR]) age at baseline equaled 44 (32–52) years. Both alcohol consumption and smoking intensity were high, with 20% of participants reporting consumption of more than 60 gr alcohol per day and 37% smoking 20 or more cigarettes per day. The median (IQR; range) c-GT was 18 (12–33; 1–2252) U/L (median [IQR] levels converted to
Journal of Hepatology 2011 vol. 55 j 594–601
JOURNAL OF HEPATOLOGY Table 2. Association of serum c-GT concentrations (at 25 °C) with mortality in German male workers.a Estimates from Cox regression models separately fit for the various mortality types.
Mortality type γ-GT (U/L)
Deaths/ Person-time (years)
b
HR
(95% CI)
HRc
(95% CI)
HRd
(95% CI)
HRe
(95% CI)
All-cause 1 ref.
1 ref.
1 ref.
<11
192/52,016
1 ref.
11-<15
298/58,748
1.23
(1.03-1.48)
1.26
(1.05-1.51)
1.28
(1.06-1.53)
1.27
(1.05-1.52)
15-<22 22-<39
39/67,086 495/ 65,006
1.28 1.55
(1.08-1.52) (1.31-1.83)
1.26 1.49
(1.06-1.50) (1.26-1.78)
1.30 1.55
(1.09-1.55) (1.30-1.84)
1.27 1.49
(1.06-1.51) (1.25-1.78)
≥39 per log-unit
794/60,342 2,170/303,198
2.84 1.67
(2.43-3.33) (1.59-1.74)
2.53 1.58
(2.14-2.98) (1.50-1.65)
2.64 1.60
(2.22-3.12) (1.53-1.68)
2.43 1.55
(2.04-2.89) (1.47-1.63)
Cancer <11
1 ref.
1 ref.
1 ref.
74/51,635
1 ref.
11-<15
114/58,250
1.21
(0.90-1.62)
1.19
(0.89-1.60)
1.22
(0.91-1.64)
1.21
15-<22
149/66,438
1.23
(0.93-1.63)
1.19
(0.90-1.58)
1.24
(0.93-1.65)
1.22
(0.91-1.62)
22-<39 ≥39 per log-unit
173/64,190 255/59,441 765/299,954
1.36 2.30 1.51
(1.03-1.78) (1.78-2.98) (1.39-1.63)
1.29 2.09 1.45
(0.98-1.71) (1.60-2.74) (1.33-1.57)
1.36 2.21 1.47
(1.02-1.81) (1.67-2.91) (1.35-1.60)
1.32 2.06 1.43
(0.99-1.76) (1.55-2.73) (1.31-1.56)
<11
39/51,635
1 ref.
11-<15
55/58,250
1.09
(0.73-1.65)
1.07
(0.71-1.62)
1.06
(0.70-1.60)
1.07
(0.71-1.62)
15-<22
95/66,438
1.46
(1.01-2.12)
1.29
(0.89-1.89)
1.27
(0.87-1.86)
1.27
(0.87-1.86)
22-<39
137/64,190
1.99
(1.39-2.84)
1.67
(1.16-2.41)
1.64
(1.13-2.37)
1.61
(1.11-2.35)
≥39 per log-unit
181/59,441 507/299,954
3.08 1.62
(2.18-4.36) (1.47-1.78)
2.20 1.41
(1.53-3.16) (1.27-1.56)
2.15 1.40
(1.49-3.12) (1.26-1.56)
2.02 1.35
(1.39-2.94) (1.21-1.50)
(0.90-1.63)
Circulatory
Respiratory
1 ref.
1 ref.
1 ref.
f
1 ref.
1 ref.
1 ref.
<22
49/176,323
1 ref.
22-<39
22/64,190
1.01
(0.61-1.66)
0.91
(0.54-1.51)
0.98
(0.58-1.64)
0.85
(0.51-1.43)
≥39
39/59,441
2.12
(1.39-3.24)
1.78
(1.14-2.78)
2.00
(1.27-3.16)
1.55
(0.97-2.48)
110/299,954
1.38
(1.10-1.71)
1.25
(0.99-1.58)
1.33
(1.05-1.68)
1.14
(0.89-1.47)
<22 22-<39
15/176,323 20/64,190
3.37
(1.73-6.60)
3.24
(1.65-6.38)
3.46
(1.75-6.82)
3.22
(1.62-6.38)
≥39
122/59,441
22.7
(13.3-38.9)
20.2
(11.7-35.1)
22.1
(12.6-38.6)
19.1
(10.8-34.0)
per log-unit
157/299,954
4.28
(3.77-4.86)
4.14
(3.62-4.74)
4.34
(3.77-4.99)
4.19
(3.63-4.84)
per log-unit f
Digestive
1 ref.
1 ref.
1 ref.
1 ref.
Accident/poisoning 1 ref.
1 ref.
1 ref.
<11
18/51,635
1 ref.
11-<15 15-<22
28/58,250 35/66,438
1.33 1.40
(0.73-2.40) (0.79-2.48)
1.36 1.45
(0.75-2.47) (0.81-2.59)
1.37 1.46
(0.75-2.49) (0.81-2.61)
1.38 1.48
(0.76-2.50) (0.83-2.66)
22-<39 ≥39
39/64,190 59/59,441
1.57 2.55
(0.89-2.75) (1.50-4.33)
1.60 2.44
(0.90-2.85) (1.40-4.26)
1.60 2.40
(0.89-2.87) (1.35-4.26)
1.64 2.41
(0.91-2.96) (1.35-4.32)
per log-unit
179/299,954
1.47
(1.25-1.72)
1.41
(1.19-1.66)
1.39
(1.17-1.65)
1.38
(1.16-1.66)
a
Models based on 19,090 (all-cause mortality) or 18,756 subjects (cause-specific analyses; excluding 334 deceased subjects for whom information on main cause of death was not available). b Adjusted for age. c +Nationality, occupation, DM, IHD, HT, BMI, smoking. d As ‘‘c’’ + elevated blood glucose, triglycerides, cholesterol. e As ‘‘c’’ + alcohol. f Quintiles 1–3 and nationalities other than German collapsed due to sparse events.
37 °C: 32 [21–58] U/L), 30% of subjects featuring elevated c-GT according to assay specifications (>28 U/L). The quintile cutoffs equaled 11, 15, 22, and 39 U/L. Bivariate patterns of association
between c-GT and baseline characteristics included a clear increase with higher body mass index and a monotonous positive association with alcohol consumption (Table 1).
Journal of Hepatology 2011 vol. 55 j 594–601
597
Research Article All-cause mortality
Hazard ratio (95 % CI)
4 3 2 1 0 0
20
40
60 80 γ-GT (U/L)
100
120
100
120
100
120
Cancer mortality Hazard ratio (95 % CI)
4 3 2 1 0 0
20
40
60 80 γ-GT (U/L)
Circulatory mortality
Hazard ratio (95 % CI)
4 3 2 1 0 0
20
40
60 80 γ-GT (U/L)
Fig. 2. Dose–response relationship between serum c-GT, all-cause, cancer and cardiovascular mortality. Depicted are cubic spline hazard ratio estimates (solid lines) and 95% confidence intervals (dotted lines) in reference to 14 U/L (at 25 °C), with knots at 8, 12, 33, and 101 U/L, adjusted for age, occupational group, nationality, BMI and smoking (complete cases analysis). Horizontal line: reference value of a hazard ratio = 1.
c-GT and mortality Among a total of 303,198 person-years in the analysis set (median follow-up time: 17.0 years), 2170 deaths occurred. Information on the cause of death could be obtained for 1836 (84.6%)
598
of these, allowing classification of 765 deaths as due to cancer, 507 as due to circulatory, 110 as due to respiratory, and 157 as due to digestive causes, as well as 179 as due to accidents/poisoning. Kaplan–Meier curves of all-cause and cause-specific survival experience are shown in Fig. 1. Monotonous risk increases across increasing c-GT quintiles were apparent for all-cause, cancer and circulatory mortality, whereas the survival experiences in the lower quintiles were less distinct for the respiratory, digestive, and accidents/poisoning outcomes. The log–log survival plots revealed no relevant deviations from the proportionality of hazards (Supplementary Fig. 1). Results from Cox regression models are presented in Table 2. For all-cause mortality, a statistically significant association with c-GT was observed already from the 2nd quintile on, and a fully adjusted hazard ratio (HR) of 2.5 was observed in the highest versus lowest quintile. The most frequent causes of death, cancer and circulatory disorders, showed fairly similar patterns of association, although the confidence intervals excluded the Null effect of no association only for the higher quintiles. As for the other mortality types, estimates were hardly affected by additional adjustment for blood alterations generally associated with metabolic syndrome. When excluding 41 deaths due to hepatocellular carcinoma from the adjusted cancer mortality models, the associations hardly changed (HRP39vs.<11U/L = 1.92 [1.46–2.53]; HRlog(c-GT) = 1.36 [1.25–1.49]). Respiratory disease and accidents/poisoning mortality appeared elevated in the 5th quintile. The three lowest quintiles combined served as reference category in the analyses for respiratory and digestive disease mortality, since deaths due to these causes were scarce in the bottom quintiles (eight for respiratory, one for digestive causes). Deaths due to disorders of the digestive system were substantively more likely in the highest quintile, despite the broader reference category. Indeed, compared to the 1st quintile, the unadjusted hazard ratio for the 5th quintile would have been 106 (95% confidence interval: 14.8–758). The associations appeared stronger when analyzing only deaths due to chronic liver disease/cirrhosis (HRP39vs.<22U/L = 36.5 [16.8–79.4]; HRlog(c-GT) = 4.70 [4.05–5.46]). The AUCs of models also including c-GT showed significantly better discrimination than age-only models for all endpoints, and were significantly better than for the full models (including only the other co-variables) for all endpoints but respiratory and accident/poisoning mortality (details not shown). For instance, the full model for all-cause mortality had an AUC = 0.789, dropping to 0.771 without c-GT (p <0.0001). Corresponding values for digestive mortality were 0.902 and 0.763 (p <0.0001). Cubic spline estimates of the dose–response relationship between c-GT and the major endpoints are shown in Fig. 2 and suggested that the growth of c-GT effects might almost level off beyond around 70 U/L. The associations of the adjustment variables with mortality were generally as expected (not shown). Age showed a very strong positive association with all causes of death except digestive and accidents/poisoning mortality, though there was still a pronounced and significant effect also for these categories. In comparison to Germans, all-cause mortality was lower in Italian, Turkish, and Yugoslavian construction workers. Unskilled workers were the only group with significantly higher mortality than the reference group (bricklayers, the most common profession), showing a 1.3-fold increase of all-cause mortality and a 1.5and 2.0-fold increase in circulatory and accidents/poisoning mortality, respectively. Smoking was consistently associated with
Journal of Hepatology 2011 vol. 55 j 594–601
JOURNAL OF HEPATOLOGY elevated mortality across categories, the notable exception being death due to accidents/poisoning. Sensitivity analyses The observed associations appeared overall robust in the various sensitivity analyses (not shown). When including alcohol consumption as an additional predictor in the adjusted models, there were no substantial changes in estimated hazard ratios (Table 2; alcohol itself was statistically significant only when assessing allcause or cancer mortality). Similarly, the associations remained stable when excluding subjects with pre-existing disease, or those with c-GT in excess of the 99th percentile. There also was no relevant change in estimates when allowing the baseline hazard of the Cox models to vary according to the year of recruitment.
Discussion The present study allowed an accurate investigation of serum
c-GT as a determinant of cause-specific mortality in a large cohort of construction industry employees, who feature an elevated potential for occupational as well as behavioral/lifestyleassociated exposures. Clear positive and independent associations between c-GT and mortality were observed across all causes of death investigated, with monotonous dose–response relationships seemingly present for most types of mortality, extending previous results pertaining to general mortality [1,5]. In particular for all-cause, cancer and cardiovascular mortality, a number of previous studies have presented detailed estimates of associations with c-GT. Most such studies were conducted in the general population, in which elevated c-GT levels tend to be much less prevalent, e.g. 13% in NHANES III [21] as compared to almost 30% in the present study. Published effect estimates thus often are not directly comparable to ours. However, the evidence for a positive association seems to be consistent across studies, and small variations in effect sizes reported might reflect different exposure codings and ranges in the various investigations [16,21,26,32]. The same applies to altogether even more consistent recent estimates for cancer mortality or incidence [16,21,28], although the published evidence for these endpoints remains somewhat sparser. Most knowledge to date has been accumulated in studies of cardiovascular mortality and incident disease. A meta-analytical estimate of 1.34-fold increase in risk per log-unit fits our findings very well [13], and the results from individual studies again may feature slightly lower or stronger estimates due to the issues discussed above [21,22,26,32,33]. Results for other cause-specific mortality types have been investigated or reported more rarely. In the general population, no association with death from respiratory causes or injury/poisoning was found in one large study [32], whereas c-GT may predict the risk of injuries in specific occupational settings [30]. We cannot fully explain the association with respiratory disease for the time being. However, with 58 of 110 respiratory deaths being due to chronic bronchitis, asthma, or chronic airway obstruction, the relationship of c-GT with smoking [4] and the possibility of residual confounding should be mentioned. The findings in our study with respect to deaths due to accidents/poisoning may at least partially be explained by the professional background of our cohort, in line with previous reports of elevated accident
mortality in this cohort [2]. The strong association we observed for digestive mortality was mainly driven by liver cirrhosis (128 of the 155 pertinent deaths) and consistent with the surprisingly few epidemiological studies of c-GT as a predictor of hard hepatic endpoints [16,21]. A limitation of our study was the restriction to male participants, which was imposed by the specific occupational setting, but kept us from contributing to the elucidation of similar patterns of associations in women, e.g. [15,27]. Furthermore, exposure assessments for the present analyses pertained to only a single point in time, whereas the benefit of including multiple c-GT measurements has been demonstrated in this context [26]. Exploring this issue in the present study by restricting analyses to the first 7.5 years of follow-up in additional sensitivity analyses in general further increased the strength of the associations (not shown). On the other side, the availability of data on alcohol consumption was an important asset of the present work. The stability of effect estimates when adjusting for this variable supports an independent association of c-GT in our cohort, but also further increases confidence in evidence consistent with our results and reported by others in the absence of alcohol consumption information, e.g. [22,28]. Furthermore, the mode of recruitment allowed us to examine a rather homogeneous and high-risk cohort. Whereas the findings, therefore, cannot be immediately generalised in the general population, this design allowed exceptionally reliable investigations of the c-GT associations in higher concentration ranges. In addition, occupational cohorts are of special relevance in that they are frequently exposed to regular exams, readily allowing not only the application of c-GT for screening in a routine setting, but also potentially providing an entry point for preventive interventions to a substantial number of subjects. Self-selection of participants in the occupational health exams analyzed in the present study – which might have lead to an under-representation of subjects with particularly high mortality and risky health behaviors [2] – similarly forbids uncritical generalizability to construction industry employees, but fully preserves both the internal validity of association patterns observed and their applicability to those employees actually accessible by this kind of routine screening. The mechanisms leading to the observed associations are clearly different for the various causes of mortality, and altogether only partially understood. For deaths due to disorders of the digestive system, which were mostly due to cirrhosis, the results are consistent with c-GT indicating serious liver damage. The fact that, as in the case of accidents/poisoning mortality, c-GT rather than alcohol consumption was a significant predictor in models including both variables might indicate a more accurate assessment of alcohol consumption by c-GT rather than self-report. A similar pattern and interpretation has been reported before [30]. For non-alcoholic liver pathology, c-GT obviously would be expected to better capture extent of liver damage and thus to be more predictive than alcohol consumption, even though the latter would be assumed to exacerbate the course of disease. In contrast, associations with cardiovascular disease appear largely unrelated to alcohol consumption, as demonstrated impressively in a large study of Japanese women featuring exceptionally high rates of never drinking [15]. Furthermore, laboratory evidence at the level of the atherosclerotic plaque and a detrimental association of c-GT with oxidative stress and progression of vascular damage have been put forward
Journal of Hepatology 2011 vol. 55 j 594–601
599
Research Article [11,12]. Oxidative stress similarly appears a reasonable explanation for associations of c-GT with malignant disease [9]. These have been described for several organ sites [27,28], and in the present study were hardly affected by exclusion of deaths due to hepatocellular carcinoma, which is particularly closely related to chronic liver disease. Large higher-resolution epidemiological as well as laboratory studies will be required to fully elucidate the distinct causal pathways that render c-GT such a robust and independent predictor of incident disease and mortality, especially given its well known associations with a wide range of risk factors for both cardiovascular disease and cancer [3,4,19]. The metabolic syndrome, for which incidence can be predicted by c-GT [19], might be of particular relevance with respect to cardiovascular endpoints. The pertinent associations in the current study, however, were adjusted for a variety of metabolic syndrome-related variables like BMI and prevalent diabetes, and rather independent from blood glucose and lipids. Note, however, that the occupational health setting might not have allowed strict adherence to fasting blood sampling, which could partially explain the high frequency of dysglycemia/lipidemia and the lack of impact of adjusting for these variables. The present report suggests that an individual’s serum c-GT concentration might allow risk stratification with respect to death due to a large variety of causes. Corroborating recent analyses identifying the same marker as an important predictor of allcause and cause-specific occupational disability [7], our findings imply that c-GT measurements in screening settings could justifiably serve as arguments for and arguments in individual risk factor modification. Largely independent from only partially understood causal relationships between this apparently powerful predictor and adverse health outcomes, occupational health professionals and other clinicians might one day use these results as an additional tool to help motivate subjects with high c-GT to attempt and maintain lifestyle changes by pointing to the strong association of this screening parameter with serious disorders for which a fair number of modifiable risk factors are known and preventive action is possible albeit not easy.
Conflict of interest The authors who have taken part in this study declared that they do not have anything to disclose regarding funding or conflict of interest with respect to this manuscript.
Financial support Funding of this cohort study by the Association of the Workmen’s Compensation Board for Construction Workers, Germany, and the German and Baden Württemberg Pension Fund is gratefully acknowledged. The funding sources had no role in study design, collection, analysis and interpretation of data, or in the decision to publish. Acknowledgments The authors are grateful to Elisabeth Bonner (Saarland Cancer Registry) and Claudia El Idrissi-Lamghari (German Cancer Research Center) for assistance with coding of causes of death.
600
Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhep.2010.12.029. References [1] Arndt V, Brenner H, Rothenbacher D, Zschenderlein B, Fraisse E, Fliedner TM. Elevated liver enzyme activity in construction workers: prevalence and impact on early retirement and all-cause mortality. Int Arch Occup Environ Health 1998;71:405–412. [2] Arndt V, Rothenbacher D, Daniel U, Zschenderlein B, Schuberth S, Brenner H. All-cause and cause specific mortality in a cohort of 20,000 construction workers; results from a 10 year follow up. Occup Environ Med 2004;61:419–425. [3] Bidel S, Silventoinen K, Hu G, Lee DH, Kaprio J, Tuomilehto J. Coffee consumption, serum gamma-glutamyltransferase and risk of type II diabetes. Eur J Clin Nutr 2008;62:178–185. [4] Breitling LP, Raum E, Muller H, Rothenbacher D, Brenner H. Synergism between smoking and alcohol consumption with respect to serum gammaglutamyltransferase. Hepatology 2009;49:802–808. [5] Brenner H, Rothenbacher D, Arndt V, Schuberth S, Fraisse E, Fliedner TM. Distribution, determinants, and prognostic value of gamma-glutamyltransferase for all-cause mortality in a cohort of construction workers from southern Germany. Prev Med 1997;26:305–310. [6] Claessen H, Arndt V, Drath C, Brenner H. Overweight, obesity and risk of work disability – a cohort study of construction workers in Germany. Occup Environ Med 2009;66:402–409. [7] Claessen H, Brenner H, Drath C, Arndt V. Gamma-glutamyltransferase and disability pension – a cohort study of construction workers in Germany. Hepatology 2010;51:482–490. [8] Conigrave KM, Davies P, Haber P, Whitfield JB. Traditional markers of excessive alcohol use. Addiction 2003;98:31–43. [9] Corti A, Duarte TL, Giommarelli C, De Tata V, Paolicchi A, Jones GD, et al. Membrane gamma-glutamyl transferase activity promotes iron-dependent oxidative DNA damage in melanoma cells. Mutat Res 2009;669:112–121. [10] Dossing M, Skinhoj P. Occupational liver injury. Present state of knowledge and future perspective. Int Arch Occup Environ Health 1985;56: 1–21. [11] Emdin M, Pompella A, Paolicchi A. Gamma-glutamyltransferase, atherosclerosis, and cardiovascular disease: triggering oxidative stress within the plaque. Circulation 2005;112:2078–2080. [12] Franzini M, Corti A, Martinelli B, Del Corso A, Emdin M, Parenti GF, et al. Gamma-glutamyltransferase activity in human atherosclerotic plaques – biochemical similarities with the circulating enzyme. Atherosclerosis 2009;202:119–127. [13] Fraser A, Harris R, Sattar N, Ebrahim S, Smith GD, Lawlor DA. Gammaglutamyltransferase is associated with incident vascular events independently of alcohol intake: analysis of the British Women’s Heart and Health Study and Meta-Analysis. Arterioscler Thromb Vasc Biol 2007;27: 2729–2735. [14] Heinzl H, Kaider A. Gaining more flexibility in Cox proportional hazards regression models with cubic spline functions. Comput Methods Programs Biomed 1997;54:201–208. [15] Hozawa A, Okamura T, Kadowaki T, Murakami Y, Nakamura K, Hayakawa T, et al. Gamma-glutamyltransferase predicts cardiovascular death among Japanese women. Atherosclerosis 2007;194:498–504. [16] Kazemi-Shirazi L, Endler G, Winkler S, Schickbauer T, Wagner O, Marsik C. Gamma glutamyltransferase and long-term survival: is it just the liver? Clin Chem 2007;53:940–946. [17] Leikin JB, Davis A, Klodd DA, Thunder T, Kelafant GA, Paquette DL, et al. Selected topics related to occupational exposures. Part IV. Occupational liver disease. Dis Mon 2000;46:295–310. [18] Meisinger C, Doring A, Schneider A, Lowel H. Serum gamma-glutamyltransferase is a predictor of incident coronary events in apparently healthy men from the general population. Atherosclerosis 2006;189: 297–302. [19] Meisinger C, Lowel H, Heier M, Schneider A, Thorand B. Serum gammaglutamyltransferase and risk of type 2 diabetes mellitus in men and women from the general population. J Intern Med 2005;258:527–535. [20] Poikolainen K, Vartiainen E. Determinants of gamma-glutamyltransferase: positive interaction with alcohol and body mass index, negative association with coffee. Am J Epidemiol 1997;146:1019–1024.
Journal of Hepatology 2011 vol. 55 j 594–601
JOURNAL OF HEPATOLOGY [21] Ruhl CE, Everhart JE. Elevated serum alanine aminotransferase and gammaglutamyltransferase and mortality in the United States population. Gastroenterology 2009;136:477–485. [22] Ruttmann E, Brant LJ, Concin H, Diem G, Rapp K, Ulmer H. Gammaglutamyltransferase as a risk factor for cardiovascular disease mortality: an epidemiological investigation in a cohort of 163,944 Austrian adults. Circulation 2005;112:2130–2137. [23] SAS Institute. Statistical analysis software, release 9.2. Cary, NC: SAS Institute; 2008. [24] Sillanaukee P, Massot N, Jousilahti P, Vartiainen E, Sundvall J, Olsson U, et al. Dose response of laboratory markers to alcohol consumption in a general population. Am J Epidemiol 2000;152:747–751. [25] Steffensen FH, Sorensen HT, Brock A, Vilstrup H, Lauritzen T. Alcohol consumption and serum liver-derived enzymes in a Danish population aged 30–50 years. Int J Epidemiol 1997;26:92–99. [26] Strasak AM, Kelleher CC, Klenk J, Brant LJ, Ruttmann E, Rapp K, et al. Longitudinal 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. [27] Strasak AM, Pfeiffer RM, Klenk J, Hilbe W, Oberaigner W, Gregory M, et al. Prospective study of the association of gamma-glutamyltransferase with cancer incidence in women. Int J Cancer 2008;123:1902–1906. [28] Strasak AM, Rapp K, Brant LJ, Hilbe W, Gregory M, Oberaigner W, et al. Association of gamma-glutamyltransferase and risk of cancer incidence in men: a prospective study. Cancer Res 2008;68:3970–3977.
[29] Thomas L, Müller M, Schumann G, Weidemann G, Klein G, Lunau S, et al. Consensus of DGKL and VDGH for interim reference intervals on enzymes in serum. J Lab Med 2005;29:301–308. [30] Tobari H, Yamagishi K, Noda H, Tanigawa T, Iso H. Body mass index and serum gamma-glutamyltransferase level as risk factors for injuries related to professional horse racing: a prospective study. J Occup Health 2009;51: 323–331. [31] Ulmer H, Kelleher C, Diem G, Concin H. Why Eve is not Adam: prospective follow-up in 149,650 women and men of cholesterol and other risk factors related to cardiovascular and all-cause mortality. J Womens Health (Larchmt) 2004;13:41–53. [32] Wannamethee G, Ebrahim S, Shaper AG. Gamma-glutamyltransferase: determinants and association with mortality from ischemic heart disease and all causes. Am J Epidemiol 1995;142:699–708. [33] Wannamethee SG, Lennon L, Shaper AG. The value of gamma-glutamyltransferase in cardiovascular risk prediction in men without diagnosed cardiovascular disease or diabetes. Atherosclerosis 2008;201:168–175. [34] World Health Organization. Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Consultation (TRS 854). Geneva: World Health Organization; 1995. [35] Yamada Y, Noborisaka Y, Suzuki H, Ishizaki M, Yamada S. Alcohol consumption, serum gamma-glutamyltransferase levels, and coronary risk factors in a middle-aged occupational population. J Occup Health 2003;45: 293–299.
Journal of Hepatology 2011 vol. 55 j 594–601
601
Gamma-glutamyltransferase, general and cause-specific mortality in 19,000 construction workers followed over 20 years Lutz Philipp Breitling1,⇑, Heiner Claessen1, Christoph Drath2, Volker Arndt1, Hermann Brenner1 1
Division of Clinical Epidemiology and Aging Research, German Cancer Research Center, D-69120 Heidelberg, Germany; 2Workmen’s Compensation Board for Construction Workers, Occupational Health Service, D-71029 Böblingen, Germany
Background & Aims: Associations of serum gamma-glutamyltransferase levels with a variety of incident diseases and mortality have been suggested. The present study attempted to expand the body of evidence to especially relevant occupational cohorts in which exposure to established c-GT determinants may greatly differ from the general population. Methods: The study was based on occupational health examinations conducted from 1986 to 1992 in 19,090 German male workers from the construction industry, aged 25–64 years. Sociodemographics and other health-related information were collected during the exam. Vital status follow-up was conducted through 2008. Associations of baseline c-GT levels (measurements at 25 °C) with all-cause and cause-specific mortality were examined by Kaplan–Meier plotting and multiple adjusted Cox regression models. Results: A total of 2170 deaths occurred during 303,198 personyears of follow-up. The risk of death due to any cause was 2.5fold increased in subjects in the highest (P39 U/L) versus lowest (<11 U/L) c-GT quintile. To varying extents, elevated c-GT was associated with higher mortality due to cancer, circulatory, respiratory, and digestive causes, as well as accidents/poisoning. Conclusions: The findings in this cohort provide evidence for c-GT being associated with a broad range of causes of death, including less investigated outcomes. Some characteristics of the observed patterns need to be seen in the context of our cohort, featuring particularly high c-GT levels. Ó 2011 European Association for the Study of the Liver. Published by Elsevier B.V. All rights reserved.
Introduction The liver enzyme gamma-glutamyltransferase (c-GT) is an emergent risk marker for a variety of common diseases such as diabetes and cancer [3,18,19,28], as well as for overall and cause-specific mortality [1,5,16,21,22,31]. Apart from alcohol consumption as
Keywords: Epidemiological study; Prospective study; Liver function tests; Mortality. Received 19 January 2010; received in revised form 22 October 2010; accepted 6 December 2010; available online 21 January 2011 ⇑ Corresponding author. Address: Division C070 of Clinical Epidemiology and Aging Research, German Cancer Research Center, INF 581 (TP4), D-69120 Heidelberg, Germany. Tel.: +49 6221 421343; fax: +49 6221 421302. E-mail address: [email protected] (L.P. Breitling).
the most classical determinant of individual c-GT levels [8], a number of potentially modifiable predictors of c-GT have been identified in epidemiological studies, including body mass index (BMI), coffee consumption, and smoking [1,3,4,20,24,25,28,35]. In occupational settings, c-GT is assessed as part of routine health examinations both due to its role not only as a marker of alcohol abuse, but also to the increased risk of hepatotoxic occupational exposures in certain professions [1,17 and references therein]. Construction workers are of particular interest in this regard, because of the frequent accumulation of both life style (alcohol consumption, smoking) and occupational risk factors, including an elevated potential for frequent exposure e.g. to solvents used in paints and adhesives, heavy metals, vinyl chloride, and other hepatotoxic substances [10,17]. We have found c-GT to be a strong predictor of mortality in a cohort of construction workers in Germany [5]. However, this analysis was restricted to all-cause mortality due to limitations in both the size of the cohort and the length of follow-up. After substantial enlargement of the cohort and accumulation of additional years of follow-up, here we provide a detailed analysis of both all-cause and cause-specific mortality for major causes of death.
Materials and methods Design, setting, and participants The analyses presented were based on a prospective cohort study of disability and mortality in employees of the construction industry in the south of Germany [1,5,6]. In brief, all employees in Germany are entitled and invited to routine occupational health examinations free of charge. Male subjects, participating in such an examination (>75% of those invited during the recruitment period) at any of eight health centres of the Workmen’s Compensation Board for construction workers in Württemberg from August 1986 to December 1992, were included in this study. Whereas participation in such exams is non-mandatory for most occupation groups according to German occupational safety laws, data obtained in such exams are to be collected and analyzed. The proportion of invitees participating in the exam thus directly reflects the participation rate for the present study. The study protocol was approved by the ethics committees of the Universities of Heidelberg and Ulm, and by the Baden-Württemberg state ministry of social affairs. Data collection Occupational health examinations were conducted by experienced occupational health physicians and included detailed standardised questionnaires regarding occupational and life style factors, physical, and functional exams; and standard
Journal of Hepatology 2011 vol. 55 j 594–601
JOURNAL OF HEPATOLOGY laboratory measurements. Collected data of relevance to the present analysis included sociodemographics, body weight and height, status and intensity of smoking and alcohol consumption; and prevalent morbidity coded according to the International Classification of Diseases revision 9 (ICD-9). Serum concentrations of c-GT were measured centrally using a Hitachi 705/717 instrument, which measured at 25 °C with an upper reference limit for males of 28 U/L (dividing this value by a temperature conversion factor of 0.57 [29], one would obtain a corresponding reference limit of 49 U/L at 37 °C). Follow-up information on vital status was obtained in several waves, starting in 1992, from regional offices of the German pension fund. For the present analysis, follow-up was updated from January 2006 to March 2008. For deceased subjects, the main cause of death was identified from death certificates obtained from public health authorities, recoded to ICD-9 if necessary, and assigned to one of the following categories: cancer (ICD-9 140–208), circulatory (ICD-9 390–459), respiratory (ICD-9 460–519), digestive (ICD-9 520–579), or accidents/poisoning (ICD-9 800–999). Coding of cause of death was independently carried out by two trained investigators and any discrepancy was resolved in consensus after additional review. Deceased subjects with missing information on the cause of death were excluded from cause-specific analyses. Statistical analysis The study population was first characterised regarding sociodemographics (age, nationality, occupational group), serum c-GT, and important co-variables potentially confounding the c-GT-mortality relationship. Furthermore, aspartate (AST) and alanine transaminases (ALT) and their ratio were reported as markers of liver damage (measured on the same instrument as c-GT, with upper normal limits of 18 and 22 U/L, respectively), and additional laboratory markers assessed as part of the routine were included as indicators of metabolic syndrome-related alterations (blood glucose, triglycerides, and cholesterol; dichotomized according to assay specifications at 100, 150, and 220 mg/dl, respectively). The association of quintiles of c-GT with all-cause and cause-specific mortality were examined by Kaplan–Meier plots. After graphical check of the proportionality of hazards assumption, separate Cox regression models for all-cause and each individual cause-specific mortality were fit adjusting either only for age at baseline (<35, 35–44, 45–54, 55–64 years), or for age, nationality (German, Italian, Turkish, Yugoslavian, any other), occupational group (bricklayer, carpenter, painter, plasterer, plumber, unskilled worker), BMI (<25, 25–29.9, P30 kg/m2) [34], prevalent diseases (diabetes [ICD-9 250], ischemic heart disease [ICD-9 410–414], hypertension [ICD-9 401–405], and smoking (never, formerly,<20, 20, >20 cigarettes per day, or smoking including tobacco products other than cigarettes) [’’full model’’], or additionally for the presence of elevated blood glucose, elevated triglycerides, and elevated cholesterol. Missing values in co-variables were dealt with by multiple imputation employing the Monte Carlo Markov Chain method implemented in SAS 9.2 (PROC MI in combination with PROC MIANALYZE) [23]. This method assumes that the variables in the dataset originate from a multivariate normal distribution and creates multiple datasets with the missing values filled with random number according to this multivariate normal. Thereafter, model parameters, such as hazard ratios are estimated and combined across the multiple imputed datasets. The procedure is assumed to be fairly robust across deviations from normality, e.g. extendable to categorical variables [23]. Importantly, estimates in complete cases analyses were very similar to the results presented here (not shown). To further quantify discriminatory performance of c-GT, area under the curve (AUC) values were estimated in logistic regression models for the various endpoints, including only co-variables or also log(c-GT) as predictors in complete cases analyses, using the same adjustment set as in the full model. Miscellaneous models explored the adjusted associations when restricting digestive mortality to deaths due to chronic liver disease/cirrhosis (ICD-9 571), and when excluding deaths from hepatocellular carcinoma (ICD-9 155) from cancer mortality models. Sensitivity analyses included the comparison with additional Cox regression models further adjusting for alcohol consumption intensity (none, occasional, 1–30, 31–60, 61–90, or P91 g per day). Furthermore, the impact of excluding subjects with prevalent disease (diabetes, ischemic heart disease, hypertension [as defined above], cerebrovascular disease [ICD-9 430– 438], cancer [ICD-9 140–208]), or of excluding subjects with c-GT beyond the 99th percentile (i.e. in excess of 260 U/L), or of stratifying the baseline hazard by year of recruitment was examined. The dose–response relationship was explored by modeling c-GT using cubic spline functions with knots at the 5th, 25th, 75th, and 95th percentile as suggested in the pertinent literature [14]. Statistical analyses were carried out using SAS 9.2 for Windows (SAS Institute, Cary, NC, USA), tests being two-sided with a = 0.05 throughout.
Table 1. Baseline characteristics of 19,090a male construction workers in Germany 1986–1992.
Baseline characteristics Age group 25 to 34 years 35 to 44 years 45 to 54 years 55 to 64 years Nationality German Italian Turkish Yugoslavian Other Occupation Bricklayer Carpenter Painter Plasterer Plumber Unskilled worker Body mass index <25 kg/m2 25 to <30 kg/m2 ≥30 kg/m2 Prevalent disease Diabetes mellitus Ischemic heart disease Hypertension Cirrhosisc Dysglycemia and lipidemia Glucose <100 mg/dl Glucose ≥100 mg/dl Triglycerides <150 mg/dl Triglycerides ≥150 mg/dl Cholesterol <220 mg/dl Cholesterol ≥220 mg/dl Smoking Never Formerly <20 cig./day 20 cig./day >20 cig./day Other tobacco Alcohol consumption None Occasional 1-30 g/day 31-60 g/day 61-90 g/day ≥91 g/day b Serum transaminases AST ≤18 U/L AST >18 U/L ALT ≤22 U/L ALT >22 U/L AST/ALT ratio ≤2 AST/ALT ratio >2 Serum γ-GTb <11 U/L 11-<15 U/L 15-<22 U/L 22-<39 U/L ≥39 U/L a b c
Serum γ-GT (U/L)b Median (IQR)
Na
%
5,749 4,008 6,311 3,022
30 21 33 16
15 20 20 20
(11-25) (13-37) (13-37) (13-33)
14,149 1,318 1,215 1,761 590
74 7 6 9 3
19 18 12 18 17
(13-35) (12-29) (9-17) (12-31) (11-30)
6,104 2,583 2,888 1,937 2,725 2,853
32 14 15 10 14 15
19 17 18 19 18 17
(12-35) (12-30) (12-34) (12-35) (12-31) (12-31)
6,689 9,094 2,933
36 49 16
14 20 26
(10-24) (13-34) (17-46)
932 315 4,163 22
5 2 22 0.1
27 24 27 75
(17-53) (16-41) (16-51) (37-207)
12,150 6,930 10,070 9,009 9,017 10,068
64 36 53 47 47 53
17 22 15 24 15 23
(11-29) (14-40) (11-24) (15-43) (10-24) (15-41)
3,927 2,900 2,718 3,730 2,085 242
25 19 17 24 13 2
17 20 17 18 21 20
(11-28) (13-37) (11-32) (12-33) (13-40) (13-34)
1,723 6,296 1,758 3,403 1,667 1,710
10 38 11 21 10 10
13 16 17 22 28 38
(9-18) (11-25) (12-28) (14-39) (17-52) (21-76)
16,917 2,085 15,077 3,980 18,811 110
89 11 79 21 99 1
17 68 16 42 18 24.5
(12-27) (36-130) (11-25) (25-79) (12-33) (12-198)
3,280 3,674 4,181 4,081 3,874
17 19 22 21 20
9 12 17 28 63.5
(8-10) (11-13) (16-19) (24-32) (48-100)
Values not always adding up to 19,090 due to missing values. Measured at 25 °C. ICD-9 571.2, 571.5, and 571.6.
Journal of Hepatology 2011 vol. 55 j 594–601
595
Research Article All-cause mortality
Cancer mortality (ICD-9 140-208) 1.00 Proportion of survivors
Proportion of survivors
1.00 0.95 0.90 0.85 0.80 0.75
0.95
0.90
0.85
0.70 0
5
10 Years
15
20
0
Circulatory mortality (ICD-9 390-459)
15
20
Respiratory mortality (ICD-9 460-519) Proportion of survivors
0.95
0.90
0.99 0.98 0.97 0.96 0.95
0.85 0
5
10 Years
15
20
0
Digestive mortality (ICD-9 520-579)
5
10 Years
15
20
Accidents/poisoning (ICD-9 800-999) 1.00 Proportion of survivors
1.00 Proportion of survivors
10 Years
1.00
1.00 Proportion of survivors
5
0.99 0.98 0.97 0.96
0.99 0.98 0.97 0.96 0.95
0.95 0
5
10 Years
15
20
0
5
10 Years
15
20
Fig. 1. Kaplan–Meier curves showing the association of c-GT quintiles (at 25 °C) with all-cause and cause-specific mortality. Line types by quintile: 1 = solid black (<11 U/L); 2 = dashed gray (11–<15 U/L); 3 = solid green (15–<22 U/L); 4 = dashed blue (22–<39 U/L); 5 = solid red (P39 U/L). For respiratory and digestive mortality, quintiles 1–3 combined are shown in black, whereas 4th and 5th quintile are gray and green.
Results Description of the study population Serum c-GT values and vital status follow-up information were available for 19,090 (95.8%) of 19,930 initially included individuals. Baseline characteristics are presented in Table 1. The majority 596
of participants were German, and bricklayers constituted the largest occupational group. Median (interquartile range [IQR]) age at baseline equaled 44 (32–52) years. Both alcohol consumption and smoking intensity were high, with 20% of participants reporting consumption of more than 60 gr alcohol per day and 37% smoking 20 or more cigarettes per day. The median (IQR; range) c-GT was 18 (12–33; 1–2252) U/L (median [IQR] levels converted to
Journal of Hepatology 2011 vol. 55 j 594–601
JOURNAL OF HEPATOLOGY Table 2. Association of serum c-GT concentrations (at 25 °C) with mortality in German male workers.a Estimates from Cox regression models separately fit for the various mortality types.
Mortality type γ-GT (U/L)
Deaths/ Person-time (years)
b
HR
(95% CI)
HRc
(95% CI)
HRd
(95% CI)
HRe
(95% CI)
All-cause 1 ref.
1 ref.
1 ref.
<11
192/52,016
1 ref.
11-<15
298/58,748
1.23
(1.03-1.48)
1.26
(1.05-1.51)
1.28
(1.06-1.53)
1.27
(1.05-1.52)
15-<22 22-<39
39/67,086 495/ 65,006
1.28 1.55
(1.08-1.52) (1.31-1.83)
1.26 1.49
(1.06-1.50) (1.26-1.78)
1.30 1.55
(1.09-1.55) (1.30-1.84)
1.27 1.49
(1.06-1.51) (1.25-1.78)
≥39 per log-unit
794/60,342 2,170/303,198
2.84 1.67
(2.43-3.33) (1.59-1.74)
2.53 1.58
(2.14-2.98) (1.50-1.65)
2.64 1.60
(2.22-3.12) (1.53-1.68)
2.43 1.55
(2.04-2.89) (1.47-1.63)
Cancer <11
1 ref.
1 ref.
1 ref.
74/51,635
1 ref.
11-<15
114/58,250
1.21
(0.90-1.62)
1.19
(0.89-1.60)
1.22
(0.91-1.64)
1.21
15-<22
149/66,438
1.23
(0.93-1.63)
1.19
(0.90-1.58)
1.24
(0.93-1.65)
1.22
(0.91-1.62)
22-<39 ≥39 per log-unit
173/64,190 255/59,441 765/299,954
1.36 2.30 1.51
(1.03-1.78) (1.78-2.98) (1.39-1.63)
1.29 2.09 1.45
(0.98-1.71) (1.60-2.74) (1.33-1.57)
1.36 2.21 1.47
(1.02-1.81) (1.67-2.91) (1.35-1.60)
1.32 2.06 1.43
(0.99-1.76) (1.55-2.73) (1.31-1.56)
<11
39/51,635
1 ref.
11-<15
55/58,250
1.09
(0.73-1.65)
1.07
(0.71-1.62)
1.06
(0.70-1.60)
1.07
(0.71-1.62)
15-<22
95/66,438
1.46
(1.01-2.12)
1.29
(0.89-1.89)
1.27
(0.87-1.86)
1.27
(0.87-1.86)
22-<39
137/64,190
1.99
(1.39-2.84)
1.67
(1.16-2.41)
1.64
(1.13-2.37)
1.61
(1.11-2.35)
≥39 per log-unit
181/59,441 507/299,954
3.08 1.62
(2.18-4.36) (1.47-1.78)
2.20 1.41
(1.53-3.16) (1.27-1.56)
2.15 1.40
(1.49-3.12) (1.26-1.56)
2.02 1.35
(1.39-2.94) (1.21-1.50)
(0.90-1.63)
Circulatory
Respiratory
1 ref.
1 ref.
1 ref.
f
1 ref.
1 ref.
1 ref.
<22
49/176,323
1 ref.
22-<39
22/64,190
1.01
(0.61-1.66)
0.91
(0.54-1.51)
0.98
(0.58-1.64)
0.85
(0.51-1.43)
≥39
39/59,441
2.12
(1.39-3.24)
1.78
(1.14-2.78)
2.00
(1.27-3.16)
1.55
(0.97-2.48)
110/299,954
1.38
(1.10-1.71)
1.25
(0.99-1.58)
1.33
(1.05-1.68)
1.14
(0.89-1.47)
<22 22-<39
15/176,323 20/64,190
3.37
(1.73-6.60)
3.24
(1.65-6.38)
3.46
(1.75-6.82)
3.22
(1.62-6.38)
≥39
122/59,441
22.7
(13.3-38.9)
20.2
(11.7-35.1)
22.1
(12.6-38.6)
19.1
(10.8-34.0)
per log-unit
157/299,954
4.28
(3.77-4.86)
4.14
(3.62-4.74)
4.34
(3.77-4.99)
4.19
(3.63-4.84)
per log-unit f
Digestive
1 ref.
1 ref.
1 ref.
1 ref.
Accident/poisoning 1 ref.
1 ref.
1 ref.
<11
18/51,635
1 ref.
11-<15 15-<22
28/58,250 35/66,438
1.33 1.40
(0.73-2.40) (0.79-2.48)
1.36 1.45
(0.75-2.47) (0.81-2.59)
1.37 1.46
(0.75-2.49) (0.81-2.61)
1.38 1.48
(0.76-2.50) (0.83-2.66)
22-<39 ≥39
39/64,190 59/59,441
1.57 2.55
(0.89-2.75) (1.50-4.33)
1.60 2.44
(0.90-2.85) (1.40-4.26)
1.60 2.40
(0.89-2.87) (1.35-4.26)
1.64 2.41
(0.91-2.96) (1.35-4.32)
per log-unit
179/299,954
1.47
(1.25-1.72)
1.41
(1.19-1.66)
1.39
(1.17-1.65)
1.38
(1.16-1.66)
a
Models based on 19,090 (all-cause mortality) or 18,756 subjects (cause-specific analyses; excluding 334 deceased subjects for whom information on main cause of death was not available). b Adjusted for age. c +Nationality, occupation, DM, IHD, HT, BMI, smoking. d As ‘‘c’’ + elevated blood glucose, triglycerides, cholesterol. e As ‘‘c’’ + alcohol. f Quintiles 1–3 and nationalities other than German collapsed due to sparse events.
37 °C: 32 [21–58] U/L), 30% of subjects featuring elevated c-GT according to assay specifications (>28 U/L). The quintile cutoffs equaled 11, 15, 22, and 39 U/L. Bivariate patterns of association
between c-GT and baseline characteristics included a clear increase with higher body mass index and a monotonous positive association with alcohol consumption (Table 1).
Journal of Hepatology 2011 vol. 55 j 594–601
597
Research Article All-cause mortality
Hazard ratio (95 % CI)
4 3 2 1 0 0
20
40
60 80 γ-GT (U/L)
100
120
100
120
100
120
Cancer mortality Hazard ratio (95 % CI)
4 3 2 1 0 0
20
40
60 80 γ-GT (U/L)
Circulatory mortality
Hazard ratio (95 % CI)
4 3 2 1 0 0
20
40
60 80 γ-GT (U/L)
Fig. 2. Dose–response relationship between serum c-GT, all-cause, cancer and cardiovascular mortality. Depicted are cubic spline hazard ratio estimates (solid lines) and 95% confidence intervals (dotted lines) in reference to 14 U/L (at 25 °C), with knots at 8, 12, 33, and 101 U/L, adjusted for age, occupational group, nationality, BMI and smoking (complete cases analysis). Horizontal line: reference value of a hazard ratio = 1.
c-GT and mortality Among a total of 303,198 person-years in the analysis set (median follow-up time: 17.0 years), 2170 deaths occurred. Information on the cause of death could be obtained for 1836 (84.6%)
598
of these, allowing classification of 765 deaths as due to cancer, 507 as due to circulatory, 110 as due to respiratory, and 157 as due to digestive causes, as well as 179 as due to accidents/poisoning. Kaplan–Meier curves of all-cause and cause-specific survival experience are shown in Fig. 1. Monotonous risk increases across increasing c-GT quintiles were apparent for all-cause, cancer and circulatory mortality, whereas the survival experiences in the lower quintiles were less distinct for the respiratory, digestive, and accidents/poisoning outcomes. The log–log survival plots revealed no relevant deviations from the proportionality of hazards (Supplementary Fig. 1). Results from Cox regression models are presented in Table 2. For all-cause mortality, a statistically significant association with c-GT was observed already from the 2nd quintile on, and a fully adjusted hazard ratio (HR) of 2.5 was observed in the highest versus lowest quintile. The most frequent causes of death, cancer and circulatory disorders, showed fairly similar patterns of association, although the confidence intervals excluded the Null effect of no association only for the higher quintiles. As for the other mortality types, estimates were hardly affected by additional adjustment for blood alterations generally associated with metabolic syndrome. When excluding 41 deaths due to hepatocellular carcinoma from the adjusted cancer mortality models, the associations hardly changed (HRP39vs.<11U/L = 1.92 [1.46–2.53]; HRlog(c-GT) = 1.36 [1.25–1.49]). Respiratory disease and accidents/poisoning mortality appeared elevated in the 5th quintile. The three lowest quintiles combined served as reference category in the analyses for respiratory and digestive disease mortality, since deaths due to these causes were scarce in the bottom quintiles (eight for respiratory, one for digestive causes). Deaths due to disorders of the digestive system were substantively more likely in the highest quintile, despite the broader reference category. Indeed, compared to the 1st quintile, the unadjusted hazard ratio for the 5th quintile would have been 106 (95% confidence interval: 14.8–758). The associations appeared stronger when analyzing only deaths due to chronic liver disease/cirrhosis (HRP39vs.<22U/L = 36.5 [16.8–79.4]; HRlog(c-GT) = 4.70 [4.05–5.46]). The AUCs of models also including c-GT showed significantly better discrimination than age-only models for all endpoints, and were significantly better than for the full models (including only the other co-variables) for all endpoints but respiratory and accident/poisoning mortality (details not shown). For instance, the full model for all-cause mortality had an AUC = 0.789, dropping to 0.771 without c-GT (p <0.0001). Corresponding values for digestive mortality were 0.902 and 0.763 (p <0.0001). Cubic spline estimates of the dose–response relationship between c-GT and the major endpoints are shown in Fig. 2 and suggested that the growth of c-GT effects might almost level off beyond around 70 U/L. The associations of the adjustment variables with mortality were generally as expected (not shown). Age showed a very strong positive association with all causes of death except digestive and accidents/poisoning mortality, though there was still a pronounced and significant effect also for these categories. In comparison to Germans, all-cause mortality was lower in Italian, Turkish, and Yugoslavian construction workers. Unskilled workers were the only group with significantly higher mortality than the reference group (bricklayers, the most common profession), showing a 1.3-fold increase of all-cause mortality and a 1.5and 2.0-fold increase in circulatory and accidents/poisoning mortality, respectively. Smoking was consistently associated with
Journal of Hepatology 2011 vol. 55 j 594–601
JOURNAL OF HEPATOLOGY elevated mortality across categories, the notable exception being death due to accidents/poisoning. Sensitivity analyses The observed associations appeared overall robust in the various sensitivity analyses (not shown). When including alcohol consumption as an additional predictor in the adjusted models, there were no substantial changes in estimated hazard ratios (Table 2; alcohol itself was statistically significant only when assessing allcause or cancer mortality). Similarly, the associations remained stable when excluding subjects with pre-existing disease, or those with c-GT in excess of the 99th percentile. There also was no relevant change in estimates when allowing the baseline hazard of the Cox models to vary according to the year of recruitment.
Discussion The present study allowed an accurate investigation of serum
c-GT as a determinant of cause-specific mortality in a large cohort of construction industry employees, who feature an elevated potential for occupational as well as behavioral/lifestyleassociated exposures. Clear positive and independent associations between c-GT and mortality were observed across all causes of death investigated, with monotonous dose–response relationships seemingly present for most types of mortality, extending previous results pertaining to general mortality [1,5]. In particular for all-cause, cancer and cardiovascular mortality, a number of previous studies have presented detailed estimates of associations with c-GT. Most such studies were conducted in the general population, in which elevated c-GT levels tend to be much less prevalent, e.g. 13% in NHANES III [21] as compared to almost 30% in the present study. Published effect estimates thus often are not directly comparable to ours. However, the evidence for a positive association seems to be consistent across studies, and small variations in effect sizes reported might reflect different exposure codings and ranges in the various investigations [16,21,26,32]. The same applies to altogether even more consistent recent estimates for cancer mortality or incidence [16,21,28], although the published evidence for these endpoints remains somewhat sparser. Most knowledge to date has been accumulated in studies of cardiovascular mortality and incident disease. A meta-analytical estimate of 1.34-fold increase in risk per log-unit fits our findings very well [13], and the results from individual studies again may feature slightly lower or stronger estimates due to the issues discussed above [21,22,26,32,33]. Results for other cause-specific mortality types have been investigated or reported more rarely. In the general population, no association with death from respiratory causes or injury/poisoning was found in one large study [32], whereas c-GT may predict the risk of injuries in specific occupational settings [30]. We cannot fully explain the association with respiratory disease for the time being. However, with 58 of 110 respiratory deaths being due to chronic bronchitis, asthma, or chronic airway obstruction, the relationship of c-GT with smoking [4] and the possibility of residual confounding should be mentioned. The findings in our study with respect to deaths due to accidents/poisoning may at least partially be explained by the professional background of our cohort, in line with previous reports of elevated accident
mortality in this cohort [2]. The strong association we observed for digestive mortality was mainly driven by liver cirrhosis (128 of the 155 pertinent deaths) and consistent with the surprisingly few epidemiological studies of c-GT as a predictor of hard hepatic endpoints [16,21]. A limitation of our study was the restriction to male participants, which was imposed by the specific occupational setting, but kept us from contributing to the elucidation of similar patterns of associations in women, e.g. [15,27]. Furthermore, exposure assessments for the present analyses pertained to only a single point in time, whereas the benefit of including multiple c-GT measurements has been demonstrated in this context [26]. Exploring this issue in the present study by restricting analyses to the first 7.5 years of follow-up in additional sensitivity analyses in general further increased the strength of the associations (not shown). On the other side, the availability of data on alcohol consumption was an important asset of the present work. The stability of effect estimates when adjusting for this variable supports an independent association of c-GT in our cohort, but also further increases confidence in evidence consistent with our results and reported by others in the absence of alcohol consumption information, e.g. [22,28]. Furthermore, the mode of recruitment allowed us to examine a rather homogeneous and high-risk cohort. Whereas the findings, therefore, cannot be immediately generalised in the general population, this design allowed exceptionally reliable investigations of the c-GT associations in higher concentration ranges. In addition, occupational cohorts are of special relevance in that they are frequently exposed to regular exams, readily allowing not only the application of c-GT for screening in a routine setting, but also potentially providing an entry point for preventive interventions to a substantial number of subjects. Self-selection of participants in the occupational health exams analyzed in the present study – which might have lead to an under-representation of subjects with particularly high mortality and risky health behaviors [2] – similarly forbids uncritical generalizability to construction industry employees, but fully preserves both the internal validity of association patterns observed and their applicability to those employees actually accessible by this kind of routine screening. The mechanisms leading to the observed associations are clearly different for the various causes of mortality, and altogether only partially understood. For deaths due to disorders of the digestive system, which were mostly due to cirrhosis, the results are consistent with c-GT indicating serious liver damage. The fact that, as in the case of accidents/poisoning mortality, c-GT rather than alcohol consumption was a significant predictor in models including both variables might indicate a more accurate assessment of alcohol consumption by c-GT rather than self-report. A similar pattern and interpretation has been reported before [30]. For non-alcoholic liver pathology, c-GT obviously would be expected to better capture extent of liver damage and thus to be more predictive than alcohol consumption, even though the latter would be assumed to exacerbate the course of disease. In contrast, associations with cardiovascular disease appear largely unrelated to alcohol consumption, as demonstrated impressively in a large study of Japanese women featuring exceptionally high rates of never drinking [15]. Furthermore, laboratory evidence at the level of the atherosclerotic plaque and a detrimental association of c-GT with oxidative stress and progression of vascular damage have been put forward
Journal of Hepatology 2011 vol. 55 j 594–601
599
Research Article [11,12]. Oxidative stress similarly appears a reasonable explanation for associations of c-GT with malignant disease [9]. These have been described for several organ sites [27,28], and in the present study were hardly affected by exclusion of deaths due to hepatocellular carcinoma, which is particularly closely related to chronic liver disease. Large higher-resolution epidemiological as well as laboratory studies will be required to fully elucidate the distinct causal pathways that render c-GT such a robust and independent predictor of incident disease and mortality, especially given its well known associations with a wide range of risk factors for both cardiovascular disease and cancer [3,4,19]. The metabolic syndrome, for which incidence can be predicted by c-GT [19], might be of particular relevance with respect to cardiovascular endpoints. The pertinent associations in the current study, however, were adjusted for a variety of metabolic syndrome-related variables like BMI and prevalent diabetes, and rather independent from blood glucose and lipids. Note, however, that the occupational health setting might not have allowed strict adherence to fasting blood sampling, which could partially explain the high frequency of dysglycemia/lipidemia and the lack of impact of adjusting for these variables. The present report suggests that an individual’s serum c-GT concentration might allow risk stratification with respect to death due to a large variety of causes. Corroborating recent analyses identifying the same marker as an important predictor of allcause and cause-specific occupational disability [7], our findings imply that c-GT measurements in screening settings could justifiably serve as arguments for and arguments in individual risk factor modification. Largely independent from only partially understood causal relationships between this apparently powerful predictor and adverse health outcomes, occupational health professionals and other clinicians might one day use these results as an additional tool to help motivate subjects with high c-GT to attempt and maintain lifestyle changes by pointing to the strong association of this screening parameter with serious disorders for which a fair number of modifiable risk factors are known and preventive action is possible albeit not easy.
Conflict of interest The authors who have taken part in this study declared that they do not have anything to disclose regarding funding or conflict of interest with respect to this manuscript.
Financial support Funding of this cohort study by the Association of the Workmen’s Compensation Board for Construction Workers, Germany, and the German and Baden Württemberg Pension Fund is gratefully acknowledged. The funding sources had no role in study design, collection, analysis and interpretation of data, or in the decision to publish. Acknowledgments The authors are grateful to Elisabeth Bonner (Saarland Cancer Registry) and Claudia El Idrissi-Lamghari (German Cancer Research Center) for assistance with coding of causes of death.
600
Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jhep.2010.12.029. References [1] Arndt V, Brenner H, Rothenbacher D, Zschenderlein B, Fraisse E, Fliedner TM. Elevated liver enzyme activity in construction workers: prevalence and impact on early retirement and all-cause mortality. Int Arch Occup Environ Health 1998;71:405–412. [2] Arndt V, Rothenbacher D, Daniel U, Zschenderlein B, Schuberth S, Brenner H. All-cause and cause specific mortality in a cohort of 20,000 construction workers; results from a 10 year follow up. Occup Environ Med 2004;61:419–425. [3] Bidel S, Silventoinen K, Hu G, Lee DH, Kaprio J, Tuomilehto J. Coffee consumption, serum gamma-glutamyltransferase and risk of type II diabetes. Eur J Clin Nutr 2008;62:178–185. [4] Breitling LP, Raum E, Muller H, Rothenbacher D, Brenner H. Synergism between smoking and alcohol consumption with respect to serum gammaglutamyltransferase. Hepatology 2009;49:802–808. [5] Brenner H, Rothenbacher D, Arndt V, Schuberth S, Fraisse E, Fliedner TM. Distribution, determinants, and prognostic value of gamma-glutamyltransferase for all-cause mortality in a cohort of construction workers from southern Germany. Prev Med 1997;26:305–310. [6] Claessen H, Arndt V, Drath C, Brenner H. Overweight, obesity and risk of work disability – a cohort study of construction workers in Germany. Occup Environ Med 2009;66:402–409. [7] Claessen H, Brenner H, Drath C, Arndt V. Gamma-glutamyltransferase and disability pension – a cohort study of construction workers in Germany. Hepatology 2010;51:482–490. [8] Conigrave KM, Davies P, Haber P, Whitfield JB. Traditional markers of excessive alcohol use. Addiction 2003;98:31–43. [9] Corti A, Duarte TL, Giommarelli C, De Tata V, Paolicchi A, Jones GD, et al. Membrane gamma-glutamyl transferase activity promotes iron-dependent oxidative DNA damage in melanoma cells. Mutat Res 2009;669:112–121. [10] Dossing M, Skinhoj P. Occupational liver injury. Present state of knowledge and future perspective. Int Arch Occup Environ Health 1985;56: 1–21. [11] Emdin M, Pompella A, Paolicchi A. Gamma-glutamyltransferase, atherosclerosis, and cardiovascular disease: triggering oxidative stress within the plaque. Circulation 2005;112:2078–2080. [12] Franzini M, Corti A, Martinelli B, Del Corso A, Emdin M, Parenti GF, et al. Gamma-glutamyltransferase activity in human atherosclerotic plaques – biochemical similarities with the circulating enzyme. Atherosclerosis 2009;202:119–127. [13] Fraser A, Harris R, Sattar N, Ebrahim S, Smith GD, Lawlor DA. Gammaglutamyltransferase is associated with incident vascular events independently of alcohol intake: analysis of the British Women’s Heart and Health Study and Meta-Analysis. Arterioscler Thromb Vasc Biol 2007;27: 2729–2735. [14] Heinzl H, Kaider A. Gaining more flexibility in Cox proportional hazards regression models with cubic spline functions. Comput Methods Programs Biomed 1997;54:201–208. [15] Hozawa A, Okamura T, Kadowaki T, Murakami Y, Nakamura K, Hayakawa T, et al. Gamma-glutamyltransferase predicts cardiovascular death among Japanese women. Atherosclerosis 2007;194:498–504. [16] Kazemi-Shirazi L, Endler G, Winkler S, Schickbauer T, Wagner O, Marsik C. Gamma glutamyltransferase and long-term survival: is it just the liver? Clin Chem 2007;53:940–946. [17] Leikin JB, Davis A, Klodd DA, Thunder T, Kelafant GA, Paquette DL, et al. Selected topics related to occupational exposures. Part IV. Occupational liver disease. Dis Mon 2000;46:295–310. [18] Meisinger C, Doring A, Schneider A, Lowel H. Serum gamma-glutamyltransferase is a predictor of incident coronary events in apparently healthy men from the general population. Atherosclerosis 2006;189: 297–302. [19] Meisinger C, Lowel H, Heier M, Schneider A, Thorand B. Serum gammaglutamyltransferase and risk of type 2 diabetes mellitus in men and women from the general population. J Intern Med 2005;258:527–535. [20] Poikolainen K, Vartiainen E. Determinants of gamma-glutamyltransferase: positive interaction with alcohol and body mass index, negative association with coffee. Am J Epidemiol 1997;146:1019–1024.
Journal of Hepatology 2011 vol. 55 j 594–601
JOURNAL OF HEPATOLOGY [21] Ruhl CE, Everhart JE. Elevated serum alanine aminotransferase and gammaglutamyltransferase and mortality in the United States population. Gastroenterology 2009;136:477–485. [22] Ruttmann E, Brant LJ, Concin H, Diem G, Rapp K, Ulmer H. Gammaglutamyltransferase as a risk factor for cardiovascular disease mortality: an epidemiological investigation in a cohort of 163,944 Austrian adults. Circulation 2005;112:2130–2137. [23] SAS Institute. Statistical analysis software, release 9.2. Cary, NC: SAS Institute; 2008. [24] Sillanaukee P, Massot N, Jousilahti P, Vartiainen E, Sundvall J, Olsson U, et al. Dose response of laboratory markers to alcohol consumption in a general population. Am J Epidemiol 2000;152:747–751. [25] Steffensen FH, Sorensen HT, Brock A, Vilstrup H, Lauritzen T. Alcohol consumption and serum liver-derived enzymes in a Danish population aged 30–50 years. Int J Epidemiol 1997;26:92–99. [26] Strasak AM, Kelleher CC, Klenk J, Brant LJ, Ruttmann E, Rapp K, et al. Longitudinal 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. [27] Strasak AM, Pfeiffer RM, Klenk J, Hilbe W, Oberaigner W, Gregory M, et al. Prospective study of the association of gamma-glutamyltransferase with cancer incidence in women. Int J Cancer 2008;123:1902–1906. [28] Strasak AM, Rapp K, Brant LJ, Hilbe W, Gregory M, Oberaigner W, et al. Association of gamma-glutamyltransferase and risk of cancer incidence in men: a prospective study. Cancer Res 2008;68:3970–3977.
[29] Thomas L, Müller M, Schumann G, Weidemann G, Klein G, Lunau S, et al. Consensus of DGKL and VDGH for interim reference intervals on enzymes in serum. J Lab Med 2005;29:301–308. [30] Tobari H, Yamagishi K, Noda H, Tanigawa T, Iso H. Body mass index and serum gamma-glutamyltransferase level as risk factors for injuries related to professional horse racing: a prospective study. J Occup Health 2009;51: 323–331. [31] Ulmer H, Kelleher C, Diem G, Concin H. Why Eve is not Adam: prospective follow-up in 149,650 women and men of cholesterol and other risk factors related to cardiovascular and all-cause mortality. J Womens Health (Larchmt) 2004;13:41–53. [32] Wannamethee G, Ebrahim S, Shaper AG. Gamma-glutamyltransferase: determinants and association with mortality from ischemic heart disease and all causes. Am J Epidemiol 1995;142:699–708. [33] Wannamethee SG, Lennon L, Shaper AG. The value of gamma-glutamyltransferase in cardiovascular risk prediction in men without diagnosed cardiovascular disease or diabetes. Atherosclerosis 2008;201:168–175. [34] World Health Organization. Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Consultation (TRS 854). Geneva: World Health Organization; 1995. [35] Yamada Y, Noborisaka Y, Suzuki H, Ishizaki M, Yamada S. Alcohol consumption, serum gamma-glutamyltransferase levels, and coronary risk factors in a middle-aged occupational population. J Occup Health 2003;45: 293–299.
Journal of Hepatology 2011 vol. 55 j 594–601
601