Synergism between smoking and alcohol consumption with respect to serum gamma-glutamyltransferase

Authors


  • The funding sources played no role in study design or the collection, analysis, and interpretation of data.

  • Potential conflict of interest: Nothing to report.

Abstract

There is increasing evidence that serum levels of the liver enzyme gamma-glutamyltransferase (γ-GT) are an important predictor of incidence and mortality of various diseases. Apart from alcohol consumption, body mass index and smoking have been found to be associated with serum levels, but little is known about potential interactions of these factors. The aim of this study was to assess the individual and joint impact of alcohol consumption and smoking on levels of γ-GT, with particular attention to potential differences by sex. The study was based on data of 8465 subjects aged 50 to 74 years, obtained at baseline examination of the ESTHER study, a large population-based cohort study in Germany. Exposure–outcome relationships were assessed in women and men, adjusting for potential confounders by multiple regression. In both sexes, moderate to heavy alcohol consumption (100+ g/week) was associated with 1.7-fold increased odds of elevated γ-GT (>50 IU/L) in reference to nonsmoking alcohol abstainers, whereas smoking by itself was unrelated to γ-GT. However, when moderate to heavy alcohol consumption was present in combination with heavy smoking, the odds ratios (95% CI) increased to 2.9 (1.1–7.6) in women and to 3.8 (2.2–6.6) in men (test for interaction between alcohol consumption and smoking: Pfemales = 0.12, Pmales = 0.0017). Conclusion: Our results support the notion of a detrimental interaction between cigarette smoking and alcohol consumption as determinants of elevated serum γ-GT, especially in men. (HEPATOLOGY 2009.)

In recent years, evidence has accumulated implicating the liver enzyme gamma-glutamyltransferase (γ-GT) as an important predictor of all-cause mortality,1, 2 development of cancer,3 diabetes,4, 5 and cardiovascular disease.6–9 Few studies have described effects of smoking status on γ-GT, with increasing exposure to smoking being associated with higher serum concentrations, particularly in women.2, 10–12 In contrast, γ-GT sensitivity to alcohol appears to be higher in men, but the evidence base for this observation also is rather limited.12–14

Only one study has investigated potential interaction effects between alcohol and smoking in regards to γ-GT, describing an effect of smoking on γ-GT in all subjects but teetotallers, in other words, complete alcohol abstainers.15 The effect of smoking on γ-GT seemed to grow stronger with increasing alcohol consumption. Because of the substantial comorbidity of nicotine and alcohol use, further elucidation of their combined effects is of high clinical and public health relevance.

The aim of this study was to conduct a detailed analysis of the individual and joint impact of alcohol consumption and smoking on serum γ-GT levels, with particular emphasis on potential differences by sex.

Abbreviations

BMI, body mass index; γ-GT, gamma-glutamyltransferase; ln, natural logarithm.

Materials and Methods

Design, Setting, and Participants.

This analysis is based on data from the baseline examination of the ESTHER study,16–19 a statewide epidemiological cohort study specifically designed to address chronic disease outcomes in the elderly population in Saarland—a state located in the southwest of Germany. A total of 9953 unselected subjects of 50 to 74 years were recruited according to a standardized protocol by 466 general practitioners during health screening visits between July 2000 and December 2002. Participation was conditional on written informed consent, and the study protocol was approved by the ethics committee of the University of Heidelberg and the medical board of the State of Saarland.

Exposure Assessment.

Smoking and alcohol drinking exposure status were determined by standardized self-administered questionnaire items covering current and lifetime (at age 20, 30, 40, and 50 years) smoking and alcohol drinking behavior. Ever-smokers were defined as people reporting consumption of more than 100 cigarettes in their life or having smoked regularly at some point. Current smoking intensity in individuals smoking at recruitment was categorized as 1 to 10, 11 to 20, 21 to 30 and more than 30 cigarettes per day; here, the first and last two categories were combined to “light” and “heavy” smoking, respectively, for the purpose of multiple regression modeling. Current pipe or cigar smokers were excluded to avoid misclassification concerning current smoking intensity. Ex-smokers, who are a rather heterogeneous group with respect to intensity of former smoking and time since cessation, were treated as a separate exposure group (excluded from regression models), and never smokers represented the reference group in all analyses.

The amount of current alcohol consumption was calculated based on questions on average weekly intake of typical consumption units of beer (13.2 g alcohol per 330-mL bottle), wine (20 g alcohol per 250-mL glass), and liquors (6.4 g alcohol per 20-mL glass) in Germany. Drinking intensity was categorized in four groups (1–39, 40–99, 100–199, and 200+ g alcohol per week), combining the first and last two as “low” and “high” alcohol consumption, respectively, for the multiple regression. The selection of the cutoffs was motivated by the literature.10

Covariable exposure data were likewise assessed by questionnaire self-report, with the exception of the body mass index (BMI), which was based on physician-reported body weight and height information measured as part of the health screening examination. BMI was categorized in three levels: up to 25 (low-normal), 25 to 30 (overweight), and above 30 kg/m2 (obese).20

Laboratory Measurements.

Serum samples were obtained during the general practitioner visit at recruitment and were mailed to the study center, where they were stored at −80°C until analysis. Serum γ-GT was determined in blinded manner by an enzyme-kinetic photometric method21 as established in the central laboratory of the University of Heidelberg (assay kits manufactured by Beckman Coulter, Inc.). This assay measured the enzyme activity at 37°C. As a university hospital laboratory, the laboratory carrying out the assays is continuously subject to quality assurance by mandatory round robin tests controlled by the gauging office.

Statistical Analysis.

Exploratory analyses showed positive skewness of γ-GT concentrations, which could be reasonably normalized by natural logarithmic (ln) transformation (skewness = 0.92; kurtosis = 1.78). Thus, ln-transformed values were used in the analysis, whereas back-transformed (geometric) means and confidence intervals are presented in the tables.22 Associations between alcohol consumption and smoking and ln-transformed levels of γ-GT were assessed by multiple linear regression controlling for age, BMI, and coffee consumption. Coefficients estimated in these models can be expressed as the percentage of change in concentration of γ-GT.23 To improve interpretability of the results, we also examined high γ-GT concentration as a binary outcome, employing multiple logistic regression models. In this case, we adopted the definition of concentrations above 50 IU/L as elevated.10 Analyses using sex-specific cutoffs for elevated concentrations according to the γ-GT assay manufacturer's information (55 IU/L for males, 38 IU/L for females) were conducted to examine the robustness of our results. Furthermore, because it is widely accepted that an association between diabetes mellitus and serum γ-GT concentrations exists,4 and some relationship with cardiac insufficiency also has been discussed,24 we excluded subjects suffering from either ailment in additional sensitivity models.

Interactions of the effects of smoking and alcohol consumption, respectively, were first assessed through assessment of γ-GT concentrations by smoking × drinking category. Elevated γ-GT and ln-transformed serum levels were then modeled by logistic and linear regression, respectively, including dummy predictors for the individual smoking × drinking categories. The significance of interaction between the two exposures was analyzed in the linear regression models by coding both smoking and drinking categories as trends (0-1-2) and examining the significance of a product term of these two predictors. Statistical tests were done at a significance level of α = 0.05 (two-sided testing).

Results

Description of the Study Population.

Concentrations of serum γ-GT along with sufficient questionnaire information on smoking behavior (354 missing) and alcohol consumption (954 missing) were available for a total of N = 8470 individuals, excluding 156 current pipe or cigar smokers. Five individuals were removed from the analysis set because of outlying γ-GT concentrations in excess of 1000 IU/L, resulting in 8465 records eligible for the primary analyses.

Table 1 provides an overview of the distribution of the main characteristics considered in this analysis. Although the prevalence of current smoking was very similar in both sexes, many more men reported former smoking. Alcohol consumption was altogether substantially higher among the male participants. The altogether 1258 subjects excluded because of missing exposure information resembled the analysis population very closely in respect to the characteristics presented in Table 1 (data not shown).

Table 1. Characteristics of Study Cohort Contributing to the Analyses Presented in this Article
StratumTotalWomenMen
N(%)N(%)N(%)
  1. Data on body mass index and coffee consumption was missing for 155 subjects (69 women, 86 men) and 134 subjects (63 women, 71 men), respectively. Percentages are column-% of nonmissing values.

Total8465 4563 3902 
Age group      
 ≤50 years237(3)141(3)96(2)
 >50–55 years1506(18)857(19)649(17)
 >55–60 years1591(19)853(19)738(19)
 >60–65 years2378(28)1275(28)1103(28)
 >65–<70 years1470(17)772(17)698(18)
 ≥70 years1283(15)665(15)618(16)
Cigarette smoking      
 Never4239(50)3028(66)1211(31)
 1–10 cig./day345(4)208(5)137(4)
 11–20 cig./day619(7)332(7)287(7)
 21–30 cig./day238(3)99(2)139(4)
 >30 cig./day89(1)20(0)69(2)
 Formerly2935(35)876(19)2059(53)
Alcohol consumption      
 Abstinent2769(33)1974(43)795(20)
 1–39 g/w1559(18)994(22)565(14)
 40–99 g/w2135(25)1104(24)1031(26)
 100–199 g/w1279(15)368(8)911(23)
 ≥200 g/w723(9)123(3)600(15)
Body mass index      
 ≤25 kg/m22266(27)1437(32)829(22)
 >25–<30 kg/m23833(46)1841(41)1992(52)
 ≥30 kg/m22211(27)1216(27)995(26)
Coffee consumption      
 None358(4)159(4)199(5)
 <1 cup/day1286(15)587(13)699(18)
 1 cup/day3404(41)1932(43)1472(38)
 >1 cup/day3283(39)1822(40)1461(38)

More comprehensive descriptions of the ESTHER study population, including detailed information on cardiovascular risk factors and sociodemographics, have been published previously.16, 17 The prevalences of hypertension, hyperlipidemia, and diabetes were 42%, 40%, and 11% at enrollment, whereas 5% and 9% had a previous diagnosis of myocardial infarction and cardiac insufficiency, respectively.

Bivariate Analyses of γ-GT Concentrations.

Table 2 provides geometric means and associated confidence intervals for serum γ-GT concentrations in the different strata of variables of interest, separately for women and men. Apart from the anticipated overall sex difference, several sex-specific trends were apparent. In particular, γ-GT increased with higher intensities of cigarette smoking and alcohol consumption in both sexes. However, associations were weaker, and the association between alcohol consumption and γ-GT did not reach statistical significance in women. A decrease of γ-GT with age was seen in men only. No heterogeneity by sex was seen for the clear positive associations of BMI with serum γ-GT, and no clear patterns emerged for the associations with coffee consumption.

Table 2. Serum γ-GT Concentrations (IU/L) According to Levels of Various Predictor Variables
StratumWomenMen
Mean(95% CI)Mean(95% CI)
  1. Given are geometric means. P values refer to trend models sex-specifically predicting γ-GT from the respective variable (former smokers excluded from the smoking model).

Total27.3(26.8–27.9)41.1(40.1–42.1)
Age group    
 ≤50 years24.6(21.7–27.8)47.5(40.4–55.8)
 >50–55 years27.0(25.8–28.2)44.6(42.0–47.3)
 >55–60 years27.7(26.5–29.0)42.3(40.0–44.8)
 >60–65 years27.0(26.0–28.0)41.7(39.9–43.6)
 >65–<70 years27.6(26.4–29.0)38.9(36.9–41.1)
 ≥70 years28.2(26.7–29.8)36.7(34.8–38.8)
  P=0.075 P<0.0001
Cigarette smoking    
Never27.0(26.4–27.7)38.1(36.5–39.6)
 1–10 cig./day27.3(25.1–29.7)38.2(34.5–42.4)
 11–20 cig./day28.1(26.2–30.1)43.0(38.9–47.5)
 21–30 cig./day29.0(25.1–33.5)45.9(40.0–52.7)
 >30 cig./day36.6(27.1–49.4)52.0(42.8–63.1)
  P=0.030 P<0.0001
 Formerly27.8(26.6–29.1)42.2(40.9–43.6)
Alcohol consumption    
 Abstinent27.6(26.7–28.4)36.7(34.9–38.6)
 1–39 g/w26.6(25.5–27.6)37.7(35.6–39.9)
 40–99 g/w26.4(25.4–27.5)38.1(36.6–39.7)
 100–199 g/w29.1(27.1–31.3)43.1(41.1–45.2)
 ≥200 g/w33.7(29.3–38.8)54.7(51.1–58.5)
  P=0.13 P<0.0001
Body mass index    
 ≤25 kg/m223.5(22.8–24.3)36.6(34.7–38.7)
 >25–30 kg/m227.6(26.8–28.5)40.9(39.6–42.2)
 ≥30 kg/m232.1(30.9–33.4)46.0(43.8–48.2)
  P<0.0001 P<0.0001
Coffee consumption    
 None27.1(24.3–30.3)37.9(34.1–42.1)
 <1 cup/day28.3(26.7–29.9)41.7(39.4–44.3)
 1 cup/day27.4(26.6–28.3)42.5(40.9–44.2)
 >1 cup/day27.0(26.2–27.9)39.8(38.4–41.3)
  P=0.24 P=0.46

The geometric mean concentration of γ-GT was very close in women and men excluded because of missing exposure information and those included in the study (29.0 versus 27.3 IU/L and 41.9 versus 41.1 IU/L).

Joint Association of Smoking and Alcohol Consumption with Serum γ-GT.

To examine potential combined effects of intensity of current cigarette smoking and alcohol consumption on serum concentrations of γ-GT, we first tabulated crude geometric mean concentrations by smoking × drinking categories (Table 3). The geometric mean concentrations were consistent with a dose–response relationship between smoking intensity and γ-GT in all strata of weekly alcohol consumption, with the exception that concentrations generally appeared to be higher in self-reported ex-smokers than in subjects in the lightest smoking category. The increase of γ-GT with higher smoking intensity was very small in teetotalers and seemed to become more pronounced in higher alcohol strata. However, for combinations of higher levels of both smoking and alcohol, some of the confidence intervals became rather wide because of low numbers of participants in these categories.

Table 3. Serum γ-GT Concentrations (IU/L) According to Smoking × Drinking Categories Given as Geometric Means
Smoking CategoryAlcohol Consumption
Abstinent1–39 g/w40–99 g/w100–199 g/w≥200 g/w
NMean(95% CI)NMean(95% CI)NMean(95% CI)NMean(95% CI)NMean(95% CI)
Never160928.4(27.4–29.4)88728.5(27.2–29.8)107329.0(27.8–30.3)47434.5(32.3–36.9)19644.3(39.8–49.2)
1–10 cig./day11027.6(24.7–30.9)5427.6(23.2–32.8)10132.7(28.8–37.1)5036.2(29.9–43.9)3041.0(32.6–51.4)
11–20 cig./day23830.8(28.1–33.8)10930.7(26.6–35.3)15132.8(29.3–36.8)7142.3(34.6–51.8)5059.8(45.0–79.5)
21–30 cig./day8732.1(27.2–37.9)3330.0(25.2–35.6)4536.2(28.8–45.4)4747.2(37.4–59.5)2665.4(41.9–102.1)
>30 cig./day3335.4(28.7–43.7)743.0(22.0–84.2)1446.8(33.7–64.8)1950.9(35.1–73.7)1690.3(54.0–150.9)
Formerly69233.3(31.6–35.2)46933.6(31.6–35.8)75134.6(32.9–36.3)61840.7(38.3–43.1)40551.2(47.1–55.8)

Regression Modeling and Adjusted Analyses.

Table 4 provides results of logistic regression models predicting elevated γ-GT concentrations by cigarette smoking × alcohol consumption categories. In the absence of alcohol consumption, no effect of smoking was evident in these analyses, neither in women nor in men. The detrimental influence of high alcohol consumption (100 g/week or more), conversely, was seen in nonsmokers as well as smokers, but it was much more pronounced in participants smoking more than 20 cigarettes daily. These had 3.4-fold increased odds for having elevated γ-GT, as compared with only a 1.9-fold increase for strong drinkers among the non- respectively light smokers. Adjustment for the effects of the covariables sex, age, BMI, and coffee consumption did not substantially alter the results. Sex-stratified modeling suggested males to be somewhat more sensitive to alcohol consumption. Effect estimates from linear regression models of ln-transformed serum levels of γ-GT (Table 4) likewise were consistent with this. The statistical test for interaction between alcohol consumption and smoking in the linear regression was significant in the overall (P < 0.0001) and males-only model (P = 0.0017), but not in the females-only model (P = 0.12).

Table 4. Odds Ratios for Elevated γ-GT (>50 IU/L) and Percent Changes in Serum γ-GT (Δ%) Associated with Alcohol × Smoking Categories, Excluding Formerly Smoking Participants
AlcoholSmokingCrudeAdjustedΔ%adj(95% CI)1
OR(95% CI)OR(95% CI)1
  • Models included 3687 women and 1843 men. Numbers were reduced in adjusted models to 3586 respectively 1768, due to missing values in covariables.

  • 1

    Adjusted for sex, age, BMI, and caffeine consumption.

Overall model       
NoneNone110
 Light0.93(0.68–1.28)0.95(0.69–1.33)3.0(−4.2 to 10.7)
 Heavy1.03(0.64–1.66)0.98(0.60–1.60)4.0(−7.3 to 16.8)
LowNone0.89(0.74–1.07)0.88(0.73–1.07)−1.7(−5.6 to 2.5)
(1–99 g/week)Light0.97(0.73–1.30)1.04(0.77–1.40)5.1(−1.8 to 12.5)
 Heavy1.08(0.65–1.78)1.11(0.66–1.87)11.3(−2.1 to 26.5)
HighNone1.58(1.26–1.98)1.61(1.28–2.04)15.1(8.6 to 22.0)
(100+ g/week)Light1.78(1.27–2.49)1.89(1.33–2.69)39.7(27.4 to 53.1)
 Heavy3.30(2.17–4.98)3.43(2.21–5.33)66.4(46.9 to 88.6)
Females-only model       
NoneNone110
 Light1.01(0.68–1.50)1.07(0.71–1.61)5.0(−3.3 to 14.1)
 Heavy1.08(0.52–2.23)0.92(0.42–1.99)7.9(−7.4 to 25.8)
LowNone0.87(0.70–1.09)0.90(0.72–1.13)−1.8(−6.1 to 2.7)
(1–99 g/week)Light0.73(0.46–1.14)0.83(0.53–1.32)3.5(−4.9 to 12.7)
 Heavy0.35(0.08–1.46)0.36(0.09–1.52)6.2(−12.6 to 28.9)
HighNone1.48(1.05–2.09)1.68(1.18–2.39)15.7(6.9 to 25.2)
(100+ g/week)Light1.79(1.02–3.14)2.04(1.14–3.67)30.4(13.6 to 49.6)
 Heavy2.29(0.89–5.93)2.85(1.06–7.62)48.3(15.1 to 91.0)
Males-only model       
NoneNone110
 Light0.89(0.51–1.53)0.87(0.49–1.53)2.2(−12.0 to 18.9)
 Heavy1.10(0.57–2.13)1.08(0.55–2.13)2.9(−14.8 to 24.2)
LowNone0.97(0.68–1.39)0.95(0.65–1.37)3.0(−6.9 to 13.9)
(1–99 g/week)Light1.29(0.83–2.00)1.30(0.83–2.05)11.2(−2.0 to 26.2)
 Heavy1.58(0.86–2.91)1.62(0.85–3.07)18.8(−1.6 to 43.4)
HighNone1.80(1.25–2.59)1.75(1.21–2.55)20.0(8.0 to 33.4)
(100+ g/week)Light1.97(1.23–3.15)1.99(1.22–3.23)50.8(30.8 to 73.8)
 Heavy3.99(2.37–6.71)3.82(2.20–6.62)76.1(49.1 to 107.8)

Sensitivity Analyses.

The odds ratios obtained when defining elevated serum γ-GT based on sex-specific cutoffs differed very little from the main results (data not shown). When we excluded 596 subjects with known diabetes (diagnosed with the disease or taking pertinent drugs as reported by the physician) or 429 subjects with self-reported diagnosis of cardiac insufficiency from the linear regression models presented in Table 4 to examine the stability of our parameter estimates, these remained practically unchanged.

Discussion

In this large population-based epidemiological study, a detrimental interaction between intensity of smoking and alcohol consumption as determinants of serum γ-GT concentrations was observed. Because both risk factors are relatively common in the general population, preventive means should be intensified on all possible levels to encourage subjects to refrain from this combination, this way reducing the likelihood of various chronic diseases and mortality.

So far, empirical evidence regarding interaction of alcohol consumption and smoking with respect to γ-GT has been very limited. The current analyses extend suggestive findings relating to a mutual enhancement of smoking and alcohol effects on serum γ-GT reported by Whitehead et al. in 1996.15 In their large and well-conducted study, dose–response relationships of γ-GT with alcohol consumption and smoking amount were demonstrated. However, participants were all males, and neither a formal test of interaction was applied nor were details of trend estimates provided. In our study, we were able to assign clear significance to the interaction between the two exposures investigated. Statistical significance was restricted to males only, although the lack of significance among women might partly be attributable to the small number of women with higher levels of both alcohol consumption and cigarette smoking.

The estimated individual effects of smoking on serum γ-GT were very similar in men and women. This is in contrast to results from an earlier population-based study showing an association of smoking status with γ-GT only in females,11 but it is consistent with two more recent studies, in which the increases in odds for elevated γ-GT were quite similar in women and men, though the effects appeared not significant in women.12, 25 A main effect of smoking was not reflected in our logistic modeling of elevated γ-GT concentrations, whereas a strong enhancement of the detrimental effects of alcohol over increasing smoking intensity strata could be seen in both sexes in these models.

The slightly more pronounced impact of weekly alcohol consumption on γ-GT levels in males is generally consistent with estimates from previous pertinent studies.12–14 Likewise, the effects observed for the adjustment variable BMI were in line with the literature,2, 11, 12, 25 whereas caffeine consumption, although this was rather crudely assessed in our study, did not show the impact previously described.10, 26, 27

Although substantial literature exists on the neurobiological effects of alcohol and nicotine coconsumption, surprisingly little is known about interactions of these two agents on the metabolic level. Thus, for the time being, we can only hypothesize about the mechanisms leading to the interaction between alcohol drinking and smoking regarding serum γ-GT. It has previously been pointed out that elevated γ-GT levels may be induced by glutathione depletion and oxidative stress resulting from ethanol metabolism, that is, alcohol consumption.10 Likewise, the cardiovascular effects of tobacco smoking seem to be mediated at least to some extent by smoking-related oxidative stress.28 In a rat model, the two substances were reported to exert additive effects in their depletion of oxidative stress defense substances such as glutathione in various tissues.29 In conclusion, it appears altogether plausible that oxidative stress in the liver caused by consumption and metabolism of alcohol reaches some critical level much more easily in the presence of generalized oxidative stress caused by concurrent smoking.

Although the main focus of this study was on current smoking and drinking behavior, former smoking has been found to be associated with elevated γ-GT before,2, 10, 15 a trend also reflected in the current analysis. Even the excess of γ-GT in former as compared with light smokers agrees with previous work.15 Although this consistency of findings across studies is to some extent reassuring, the issue also points to a limitation of most alcohol-related and smoking-related studies because of potential misreporting by participants. To examine the extent of this problem in our study, we employed serum cotinine measurements in 849 former heavy smokers included in the current analysis. In 55 (6.5%) of these participants, the concentration of the nicotine metabolite exceeded 15 ng/mL, suggesting misreported abstinence. However, serum γ-GT levels were similar in the 794 confirmed quitters (results not shown), and the overall low number of misclassified subjects in this subgroup in relation to the overall study size suggests this issue to be of little concern for our analyses.

To investigate the impact of similar problems regarding alcohol consumption, we fitted linear regression models excluding current alcohol abstainers who reported consumption of 100+ g alcohol per week at age 40 or 50 (results not shown). This had virtually no consequence for the effect estimates in females, whereas the percentage changes of γ-GT associated with low and high alcohol consumption in the adjusted models for men indeed were shifted upward by approximately 7%–10%. Although this is consistent with some misreporting of alcohol abstinence in our male participants, the overall patterns described were unaffected by this phenomenon, and effect sizes would rather tend to be even somewhat larger in the absence of misreporting.

Because of recruitment taking place during voluntary health screening examinations, some selection bias might have occurred. However, previously published detailed comparisons with population registries and other epidemiological studies conducted in Germany have shown that the ESTHER cohort is representative of the study region's elderly population regarding both age and sex, and of the German general population in terms of cardiovascular risk factor distributions.16 This suggests at least no extensive selection to be present, and bias of this kind should not compromise the validity of the observed relationships, while affecting generalizability only modestly, if at all.

In conclusion, the epidemiological analyses presented in this paper suggest that smoking and alcohol consumption—at intensities that are common in the general population—interact with each other to enhance their elevating effects on serum γ-GT. Because of the high concurrence of these major avoidable risk factors, this finding has important public health and clinical relevance. In particular, it underlines the importance of the ubiquitous and tedious day-to-day task of encouraging patients smoking tobacco and drinking alcohol to tackle their harmful behaviors. The interaction of smoking and alcohol also warrants consideration in future efforts to investigate γ-GT as an emerging powerful predictor of morbidity and mortality.

Ancillary

Advertisement