Association between Gamma-Glutamyltransferase Level and Risk of Stroke: A Systematic Review and Meta-analysis of Prospective Studies

ARTICLE IN PRESS

Association between Gamma-Glutamyltransferase Level and Risk of Stroke: A Systematic Review and Meta-analysis of Prospective Studies Xiao-Wei Zhang, MD,*1 Min Li, MD,*1 Wen-Shang Hou, MD,* Kun Li, Jing-Ran Zhou, BM,† and Zhen-Yu Tang, PhD*

MD,*

Background: Stroke is often regulated by a number of modifiable and nonmodifiable risk factors. Recently, studies suggested high gamma-glutamyltransferase (GGT) level may be associated with stroke, but drew inconsistent conclusions. So, we conducted a meta-analysis to evaluate the relationship between GGT level and risk of stroke. Methods: We systematically searched PubMed, Embase, and Cochrane Library (updated to January 2015) for prospective cohort studies. Then, relative risk (RR) with 95% confidence interval (CI) was used to assess the association. Regression analyses, subgroup analyses, and sensitivity analyses were also performed. The Begg test, Egger test, and the trim-and-fill method were used to assess potential publication bias. Results: A total of 5707 cases and 926,497 participants in 10 prospective studies were included. Overall, high GGT level has a positive association with increased risk of stroke (RR = 1.28; 95% CI, 1.16-1.43). In the subgroup analyses, a positive association was consistently observed in each subgroup except in the women subgroup (RR = 1.45; 95% CI, .9-2.34) and a large number of stroke events subgroups (≥500) (RR = 1.25; 95% CI, .85-1.84). Heterogeneity was significantly reduced in the subgroup analysis by population characteristics. In the publication bias test, the resulting adjusted RR remained significant (RR = 1.10; 95% CI, 1.001.21) after using the trim-and-fill method. Conclusions: Our meta-analysis provides evidence that a high level of GGT is significantly associated with increased risk of stroke independently of alcohol intake. Gender and ethnicity variations may exist in the relationship between high GGT level and risk of stroke. Key Words: Gamma-glutamyltransferase—stroke—prospective cohort studies—meta-analysis. © 2015 National Stroke Association. Published by Elsevier Inc. All rights reserved.

Introduction From the *Department of Neurology, The Second Affiliated Hospital, Nanchang University, No. 1, Minde Road, Nanchang 330006, Jiangxi Province, China; and †Department of Hematology, The Second Affiliated Hospital, Nanchang University, No. 1, Minde Road, Nanchang 330006, Jiangxi Province, China. Received July 20, 2015; revision received August 7, 2015; accepted August 13, 2015. The authors declare that they have no conflict of interest. Address correspondence to Zhen-Yu Tang, Department of Neurology, The Second Affiliated Hospital, Nanchang University, No. 1, Minde Road, Nanchang 330006, Jiangxi Province, China. E-mail: [email protected]. 1 These authors contributed equally to this work. 1052-3057/$ - see front matter © 2015 National Stroke Association. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2015.08.015

Stroke is a cerebrovascular disease that has a serious impact on human health, and has high morbidity and mortality rates. Over the past 4 decades, the rate of stroke incidence has decreased 42% in high-income countries and increased over 100% in low- to middle-income countries globally.1 Even so, 6.8 million people suffer from stroke and approximately 795,000 people experience new or recurrent stroke in the United States each year.2 In China, stroke has become a leading cause of death and approximately 1.7 million people died from stroke in 2010.3 The risk of stroke is often regulated by a number of modifiable and nonmodifiable risk factors.4 Recently, several prospective studies have provided insights into the potential importance of high gamma-glutamyltransferase

Journal of Stroke and Cerebrovascular Diseases, Vol. ■■, No. ■■ (■■), 2015: pp ■■–■■

1

ARTICLE IN PRESS X.-W. ZHANG ET AL.

2

(GGT) level as a risk factor of the stroke, but have drawn inconsistent conclusions.5-16 GGT is a common biomarker of hepatocyte. The test of GGT has become an important routine for liver function17 and alcohol abuse.18 Although a previous meta-analysis19 has quantified the association of liver enzymes with cardiovascular disease (CVD) and stroke, it did not further investigate the relationship between GGT level and risk of stroke. Therefore, we conducted a meta-analysis to evaluate the correlation between GGT level and risk of stroke. If GGT level became an independent risk factor for stroke, it should be on high alert in clinical evaluation and primary prevention.

Method Literature Search We performed a systematic search of PubMed, Embase, and Cochrane Library (up to January 2015). The following key words were used in our search strategies: “gammaglutamyltransferase,” “gamma-GT,” “γ-GT,” “GGT,” “stroke,” “cerebrovascular accident,” “intracranial artery disease,” “brain ischemic,” “cerebrovascular disease,” “cerebrovascular disorders,” “cerebral infarct,” “ischemic stroke,” “intracranial hemorrhage,” “intracranial artery disease,” “longitudinal studies,” “cohort studies,” “prospective studies,” and “follow-up studies.” We restricted the search to human studies. There were no language restrictions. In addition, we searched for possible eligible studies in the references within the retrieved articles, as well as review articles and abstracts from recent conferences.

Study Selection Studies were included if they satisfied the following criteria: (1) The study patients had a community-based or population-based prospective cohort design; (2) the exposure had a high GGT level; and (3) reported quantitative estimates of relative risk (RR) and 95% confidence interval (CI) for risk of stroke along with elevated baseline level of GGT. Studies were excluded if (1) the study design did not include a cohort; (2) the study participants have a history of stroke or liver damage; and (3) the study did not provide all the data information that we needed.

Data Extraction and Quality Assessment Two authors (X.W.Z. and M.L.) extracted information independently and the disagreements were resolved via discussion and consensus. The following data were extracted from each included article: first author’s last name, year of publication, country where the study was performed, size of the cohort, gender proportion, age, followup time, stroke ascertainment, cutoff values or quartile of GGT level, adjusted RR, adjusted covariates, and study quality. When the same or similar patient cohort was included in several publications, only the most recent or

complete report was selected for analysis. Study quality was assessed based on the 9-star Newcastle–Ottawa Scale (NOS)20 using predefined criteria, namely, selection (population representativeness), comparability (adjustment for confounders), and ascertainment of outcome. The NOS assigns a maximum of 4 points for selection, 2 points for comparability (2 points were awarded for studies that reported estimates for the highest degree of adjustment defined above +++ and one point for ++), and 3 points for outcome. Nine points on the NOS reflect the highest study quality.

Statistical Analysis RR and 95% CI were chosen as the effect estimate to assess the association between GGT level and risk of stroke. Hazard ratios were assumed to approximate the same measure of RR. Data analysis used multivariate-adjusted outcome data (expressed as RRs and 95% CIs). Because different studies presented results on different scales, we converted these values in every study by using their natural logarithms and their corresponding 95% CIs. Heterogeneity among studies was examined by using chi-squarebased Q test and I2 test. When significant heterogeneity (P value < .10 and I2 > 50%) was detected, the pooled RR and 95% CI would be estimated in a random effects model. Otherwise, a fixed effects model was chosen. To evaluate the influence of alcohol intake on the overall results, metaregression analysis and stratified analysis were performed. To explore the possible sources of heterogeneity, we systematically performed subgroup analyses by ethnicity, gender, time of duration, study quality, number of stroke event, degree of adjustment, and sample state. To test the robustness of the association, we performed sensitivity analysis to investigate the influence of a single study on the overall risk estimate, and carried out by sequentially omitting 1 study at each turn with the metainf algorithm. Finally, we used the Begg test,21 the Egger test22 and visual inspection of a funnel plot to assess the potential publication bias. If any possible bias was found, the Duval and Tweedie nonparametric trim-and-fill method23 was performed to further assess the potential publication bias. For all tests, A P-value less than .05 was considered statistically significant. All statistical analyses were conducted with Stata 12.0 (StataCorp, College Station, TX).

Result Study Characteristics We retrieved 624 articles from PubMed, Embase, Science Citation Index (Web of Science), and Cochrane Library. Of these articles, 616 were excluded for one of the following reasons: (1) did not provide the GGT level as an outcome of interest; (2) was a duplicate article; (3) was not a human study; (4) was not a prospective cohort study;

ARTICLE IN PRESS HIGH GGT LEVEL COULD INCREASE THE RISK OF STROKE

3 5-7,9,11,12

Potentially relevant citations from PubMed, Embase, Web of science, and Cochrane Library (n = 624) Excluded on basis of title and/or abstract (n = 267)

Articles retrieved for detailed evaluation (n = 357)

Article excluded with reason: Duplicate articles (n =112) Not in human (n = 156) Review and case report (n = 7) Case-control study (n =4) No relevant exposure (n =32) No relevant outcome (n = 27)

Full-text articles included in meta-analysis (n = 14)

Same patient cohort study (n =6)

Articles included meta-analysis (n = 8)

Figure 1.

in

Six studies included men and women ; 3 studies included only men,10 and 1 study included only women.8 The ages of the subjects ranged from 25 to 79 years old. The follow-up duration ranged from 1 to 24 years. One study of sample state was from plasma12; the other 9 studies were from serum.5-11 As shown in Table 1, the ascertainment of high GGT level and stroke varied across studies. In addition, all studies provided risk estimates, overall quality of studies were good (range 6-9).

The Correlation between GGT Level and Risk of Stroke The RRs of stroke in relation to GGT level from individual studies and the combined RR are presented in Figure 2. In Figure 2, a total of 10 comparisons investigate the association between GGT level and risk of stroke.5-12 Pooling all 10 comparisons, participants with high GGT level experienced a significant increased risk for development of stroke (RR = .28; 95% CI, 1.16-1.43). There was significant heterogeneity among individual studies (P < .001; I2 = 74.5%).

Metaregression and Subgroup Analysis

Flow chart of article selection process for meta-analysis.

or (5) was the same patient cohort study. Finally, a total of 8 articles including results from 10 prospective studies (Fig 1) with 5707 cases of stroke and 926,497 participants were included in the meta-analysis. 5-12 The characteristics of the studies are presented in Table 1. Among 10 prospective studies, 8 were conducted primarily in Europe5,6,8-10,12 and 2 were from Asian countries.7,11

Figure 2. Forest plot of study specific estimates of relative risk of stroke associated with increased GGT level. The metaestimate in the forest plot is estimated from random effects metaanalyses. Abbreviation: CI, confidence interval; GGT, gamma-glutamyltransferase.

We explore the potential effect of alcohol intake on the overall outcomes of our study. The multivariate analysis based on alcohol consumption showed that alcohol intake has little influence on overall results (P = .151). The detailed results stratified by the characteristics of the study design are shown in Figure 3. In Figure 3, the estimated RR was 1.17 (95% CI, 1.05-1.31) after alcohol consumption was adjusted. The positive association between high GGT level and risk of stroke was

Author (Publication year)

RR (95% CI)

Jousilahti et al (M) (2000)

1.24 (1.03, 1.50)

Jousilahti et al (W) (2000)

1.33 (1.06, 1.65)

Bots et al (UK)

(2002)

1.24 (0.88, 1.75)

Bots et al (Fin)

(2002)

1.21 (0.89, 1.65)

Bots et al (The Net) (2002)

1.28 (1.02, 1.61)

Ebrahim et al

(2006)

1.05 (1.03, 1.08)

Fraser et al

(2007)

1.45 (0.90, 2.34)

Strasak et al (M)

(2008)

2.29 (1.51, 3.48)

Strasak et al (W)

(2008)

2.12 (1.37, 3.28)

Wannamethee et al (2008)

1.56 (1.20, 2.04)

Shimizu et al (M)

(2010)

1.11 (0.93, 1.32)

Shimizu et al (W)

(2010)

1.17 (0.99, 1.38)

Weikert et al

(2013)

1.20 (1.03, 1.40)

Overall (I-squared=74.5%,P=0.000)

1.28 (1.16, 1.43)

Z=4.67, P=0.000 0.3

Decreased risk

1

Increased risk

3.0

4

Table 1. Characteristics of prospective cohort studies evaluating associations between GGT and risk of stroke First author (year of publication)

Participants (% male)

Age range or mean (year)

Duration (year)

Jousilahti et al. (2000)5 Bots et al. (2002)6

Finland/ European United Kingdom/ European

14,874 (48)

25-64

7

181 (100)

45-59

11-15

Bots et al. (2002)6 Bots et al. (2002)6 Ebrahim et al. (2006)7

Finland/ European The Netherlands/ European Korea/Asian

200 (100)

42-60

4-9

429 (36)

≥55

1-4

30-64

15

Fraser et al. (2007)8

United Kingdom/ European

Strasak et al. (2008)9

Austria/ European

Wannamethee et al. (2008)10

United Kingdom/ European

6997 (100)

Shimizu et al. (2010)11

Japan/Asian

Weikert et al. (2013)12

Germany/ European

787,442 (84)

Stroke ascertainment National death registry and hospital discharge registry National mortality registry, hospital discharge records, self-report, and family report National mortality registry and FINMONICA stroke register Hospital discharge records

Cutoff values or quartile of GGT (U/L)

Adjusted RR (95% CI)

Adjustment for covariates

Study quality

M: 10.8, 16.4, 24.6, 64.2; W: 7, 9.9, 13.6, 33.9 17, 26, 41

M: 1.24 (1.03-1.50), W: 1.33 (1.06-1.65) 1.24 (.88-1.75)

Age, study year, smoking, TC, SBP, DBP, BMI Age, sex

8

15, 21, 36

1.21 (.89-1.65)

Age, sex

7

17, 22, 30

1.28 (1.02-1.61)

Age, sex

8

National statistical office and the death benefit record

≥81

Hemorrhagic: 2.02 (1.742.34), ischemic: 1.11 (.99-1.25)

8

Medical records and NHS central registry

...

1.45 (.90-2.34)

Age, sex, BMI, height, glucose, hypertension, ethanol consumption, smoking, physical activity, monthly pay, and area of residence Age, social class, physical activity, smoking, alcohol consumption, diabetes/insulin resistance, BMI, triglycerides, HDL-c, SBP Age, glucose, SBP, DBP, BMI, smoking, occupational status, triglycerides, and TC Age, social class, smoking, alcohol intake, physical activity, pre-existing evidence of undiagnosed CHD, BMI, SBP, cholesterol, blood glucose, and HDL-c Age, community, BMI, smoking, alcohol intake, TC, triglycerides, albumin, AST, ALT, SBP, antihypertensive medication use, and DM Age, sex, BMI, waist circumference, education, sports activity, smoking, alcohol intake, hypertension, diabetes, TC, HDL-c, CRP

2961 (0)

60-79

4.6

76,113 (43)

42

10.2

Death certificates and health authority

M: ≥28.6, W: ≥17.9

M: 2.29 (1.51-3.48), W: 2.12 (1.37-3.28)

40-59

24

NHS registry and death certificate

≥22

1.56 (1.20-2.04)

9752 (36)

40-69

18.1

Reports by local physicians, ambulance records, death certificates, public health nurses, and health volunteers

M: 15, 24, 45; W: 8, 11, 16

M: 1.11 (.93-1.32), W: 1.17 (.99-1.38)

27,548 (39)

35-65

6-10.4

Medical records and death certificates

M: 12.1, 20.9, 31.9, 65.5; W: 6.6, 11, 15.4, 33

1.20 (1.03-1.40)

6

ARTICLE IN PRESS

Country/ population

8

8

8

9

9

X.-W. ZHANG ET AL.

Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; BMI, body mass index; CHD, coronary heart disease; CI, confidence interval; CRP, C-reactive protein; DBP, diastolic blood pressure; DM, diabetes mellitus; GGT, gammaglutamyltransferase; HDL-c, high-density lipoprotein cholesterol; M, men; NHS, National Health Service; RR, relative risk; SBP, systolic blood pressure; TC, total cholesterol; W, women; . . ., null value.

ARTICLE IN PRESS HIGH GGT LEVEL COULD INCREASE THE RISK OF STROKE

Subgroup

P-value for heterogeneity

Population European Asian

nificantly reduced in the stratified analysis by population characteristics, indicating that the source of heterogeneity appeared to be contributed by ethnicity variations.

RR (95% CI)

0.083 0.377

1.37 (1.22, 1.53) 1.05 (1.03, 1.08)

0.000 0.396 -

1.26 (1.12, 1.42) 1.36 (1.14, 1.61) 1.45 (0.90, 2.34)

0.000 0.960

1.34 (1.12, 1.59) 1.25 (1.14, 1.37)

0.803 0.000

1.16 (1.06, 1.28) 1.37 (1.17, 1.60)

0.004 0.072

1.25 (0.85, 1.84) 1.28 (1.17, 1.41)

Degree of adjustment + 0.958 ++ 0.014 +++ 0.012

1.25 (1.06, 1.47) 1.58 (1.21, 2.08) 1.17 (1.05, 1.31)

Sample state Serum Plasma

1.30 (1.16, 1.46) 1.20 (1.03, 1.40)

.

Sex Both Men Women .

Duration ≥10 y 10 y .

Quality 9 9

5

Sensitivity Analysis To evaluate the impact of single studies on the combined results, we conducted sensitivity analysis by omitting 1 study at a time and reassessed the summary RR for the remaining studies. As shown in Figure 4, there was little influence in the quantitative pooled measure of RR or 95% CI when any study was omitted.

.

No. of events, n ≥500 500

Publication Bias The funnel plot is shown in Figure 5. In Figure 5, A, potential publication bias may exist through the Begg test (P = .077) and Egger test (P < .001). To address this we used the trim-and-fill method to correct the bias. We showed that if the publication bias was the only source of the funnel plot asymmetry, 7 more studies were needed to balance the funnel plot (Fig 5, B). The resulting adjusted RR was attenuated but remained significant (RR = 1.10; 95% CI, 1.00-1.21; P < .001) in the random effects model.

.

.

.

0.000 -

0.5 1

2.0

Figure 3. Prospective studies of GGT levels with risk of stroke, grouped according to several study characteristics. The summary estimate presented was calculated using a random effects model. Abbreviations: CI, confidence interval (bars); degree of adjustment: + , adjusted for age and/ or sex; ++, further adjustment for cardiovascular risk factors; +++, additional adjustment for alcohol consumption, other liver markers, or inflammatory markers; GGT, gamma-glutamyltransferase; no. of events, number of the stroke events in individual study; RR, relative risk.

Discussion Our meta-analysis with 10 prospective cohort studies involving 926,497 participants and 5707 stroke cases showed a significant positive relationship between high GGT level and risk of stroke. The risk of stroke was increased by 28% in people with high GGT level. In the present study, alcohol consumption is an important confounding factor. Evidence suggested that high baseline level of GGT was related to higher alcohol intake, and GGT was elevated in a high proportion of

consistently observed in each subgroup except in the women subgroup (RR = 1.45; 95% CI, .9-2.34) and in the large number of stroke events (≥500) subgroup (RR = 1.25; 95% CI, .85-1.84). In addition, the heterogeneity was sig-

Meta-analysis estimates, given named study is omitted Lower CI Limit Estimate Upper CI Limit Jousilahti(men) (2000) Jousilahti(women) (2000) Bots(UK) (2002) Bots(Fin) (2002) Bots(The Net) (2002)

Figure 4. Sensitivity analyses results of given named study omitted.

Ebrahim (2006) Fraser (2007) Strasak(men) (2008) Strasak(women) (2008) Wannamethee (2008) Shimizu(men) (2010) Shimizu(women) (2010) Weikert (2013) 1.13 1.16

1.28

1.43

1.47

ARTICLE IN PRESS X.-W. ZHANG ET AL.

6

A

Begg's funnel plot with pseudo 95% confidence limits 1

log[RR]

.5

0

-.5 0

.1

.2

.3

.2

.3

s.e. of: log[RR]

B

Filled funnel plot with pseudo 95% confidence limits 1

Figure 5. (A) Publication bias was assessed by Begg’ funnel plot analysis. Before the potential publication bias was corrected, the unadjusted RR was 1.28 (95% CI, 1.16-1.43) in the random effects model. (B) The trim-and-fill method estimated the presence of 7 potentially unreported studies. After these 7 potentially unpublished studies were filled, the resulting adjusted RR was 1.10 (95% CI, 1.001.21) in the random effects model. Abbreviations: CI, confidence interval; RR, relative risk.

theta, filled

.5

0

-.5

-1 0

.1 s.e. of: theta, filled

alcoholics.24,25 Freer and Statland26 proved that 3 doses of .75 g/kg on successive days were reported to produce a small increase in GGT. Given this, we conducted metaregression and subgroup analyses. The result of metaregression expressed that alcohol consumption did not significantly affect our overall results, and the result of the adjusted alcohol consumption subgroup analysis confirmed that high GGT level has a positive relationship with risk of stroke. In the subgroup analyses, we found the associations were still significant in different subgroups except in the women subgroup and the large number of stroke events (≥500) subgroup. No significant association in these 2 subgroups was found, maybe not only because the number of studies was small and the heterogeneity was significant, but also perhaps gender variation also exists in the

relationship between high GGT level and risk of stroke. Besides, the heterogeneity was significantly reduced in the subgroup stratified by population characteristics, which suggested that ethnicity variations might be the source of heterogeneity. Overall, gender and ethnicity variations might exist in the relationship between high GGT level and risk of stroke, but more data are needed to confirm these correlations. In addition, these available 10 studies of our metaanalysis reported the recruitment of participants from approximately general populations and met the stringent inclusion criteria, although the existence of prevalent diseases and confounding factors could not be completely ruled out in the study populations. In sensitivity analyses, no individual study substantially influenced the overall RR or 95% CI. In the publication bias test, the

ARTICLE IN PRESS HIGH GGT LEVEL COULD INCREASE THE RISK OF STROKE

resulting adjusted RR was not materially alerted after correcting for publication bias by the trim-and-fill method, suggesting the stability of our results. Overall, the results of our study were reliable, although there was significant heterogeneity among individual studies. From an epidemiological viewpoint, stroke is often explained by the exposure to various risk factors. Results from an international case–control research showed that 90% of the risks in stroke are explained by 10 potentially modifiable risk factors. 27 For the moment, the mechanisms underlying the increase of consequent stroke events in patients with high GGT level are still not completely understood. Results in atherosclerosis,28 through proinflammatory29-31 and pro-oxidant activity,32-34 may link the development of GGT level with the pathogenesis of stroke events. However, the potential limitations of this metaanalysis should be mentioned. First, potential publication bias may be existing in our study, even though we used the trim-and-fill method to correct the bias. Second, it would be interesting to determine whether the high GGT level–stroke association differed by stroke subtypes, but few data were available for a stratified analysis. Third, we did not examine the relationship between different GGT levels and the risk of stroke because there was insufficient information about the dose–response relationship in our studies.

Conclusion In conclusion, our meta-analysis provides evidence that high GGT level is significantly associated with increased risk of stroke. Gender and ethnicity variations may exist in the relationship between high GGT level and risk of stroke. Future studies, especially randomized controlled studies of agents that lower or prevent elevated GGT, should explore whether GGT is a potentially modifiable risk factor for stroke. Acknowledgments: Z.Y.T. and X.W.Z. conceived and designed the experiments and wrote the paper. M.L. and J.R.Z. analyzed the data. W.S.H. and K.L. performed the literature search and the data extraction. All authors saw and approved the final version of the manuscript.

References 1. Feigin VL, Lawes CM, Bennett DA, et al. Worldwide stroke incidence and early case fatality reported in 56 population-based studies: a systematic review. Lancet Neurol 2009;8:355-369. 2. Go AS, Mozaffarian D, Roger VL, et al. Heart disease and stroke statistics-2014 update: a report from the American Heart Association. Circulation 2014;129:e28-e292. 3. Yang G, Wang Y, Zeng Y, et al. Rapid health transition in China, 1990-2010: findings from the Global Burden of Disease Study 2010. Lancet 2013;381:1987-2015.

7 4. Allen CL, Bayraktutan U. Risk factors for ischaemic stroke. Int J Stroke 2008;3:105-116. 5. Jousilahti P, Rastenyte D, Tuomilehto J. Serum gammaglutamyl transferase, self-reported alcohol drinking, and the risk of stroke. Stroke 2000;31:1851-1855. 6. Bots ML, Salonen JT, Elwood PC, et al. Gammaglutamyltransferase and risk of stroke: the EUROSTROKE project. J Epidemiol Community Health 2002;56:i25-i29. 7. Ebrahim S, Sung J, Song YM, et al. Serum cholesterol, haemorrhagic stroke, ischaemic stroke, and myocardial infarction: Korean national health system prospective cohort study. BMJ 2006;333:22. 8. Fraser A, Harris R, Sattar N, et al. Gammaglutamyltransferase is associated with incident vascular events independently of alcohol intake: analysis of the British Women’s Heart and Health Study and MetaAnalysis. Arterioscler Thromb Vasc Biol 2007;27:2729-2735. 9. Strasak AM, Kelleher CC, Klenk J, 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. 10. 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. 11. Shimizu Y, Imano H, Ohira T, et al. Gammaglutamyltranspeptidase and incident stroke among Japanese men and women: the Circulatory Risk in Communities Study (CIRCS). Stroke 2010;41:385-388. 12. Weikert C, Drogan D, di Giuseppe R, et al. Liver enzymes and stroke risk in middle-aged German adults. Atherosclerosis 2013;228:508-514. 13. Higashiyama A, Wakabayashi I, Ono Y, et al. Association with serum gamma-glutamyltransferase levels and alcohol consumption on stroke and coronary artery disease: the Suita study. Stroke 2011;42:1764-1767. 14. Hozawa A, Okamura T, Kadowaki T, et al. Gammaglutamyltransferase predicts cardiovascular death among Japanese women. Atherosclerosis 2007;194:498-504. 15. Ulmer H, Brant LJ, Kelleher CC, et al. Gammaglutamyltransferase predicts mortality from both ischemic and hemorrhagic stroke in a cohort of 170,244 men and women. Stroke 2007;38:195. 16. Fujiyoshi A, Miura K, Hozawa A, et al. γ-Glutamyltransferase and mortality risk from heart disease and stroke in Japanese men and women: NIPPON DATA90. CVD Prevention and Control 2010;5:27-34. 17. Lum G, Gambino SR. Serum gamma-glutamyl transpeptidase activity as an indicator of disease of liver, pancreas, or bone. Clin Chem 1972;18:358-362. 18. Whitfield JB. Gamma glutamyl transferase. Crit Rev Clin Lab Sci 2001;38:263-355. 19. Kunutsor SK, Apekey TA, Khan H. Liver enzymes and risk of cardiovascular disease in the general population: a meta-analysis of prospective cohort studies. Atherosclerosis 2014;236:7-17. 20. Wells GA, Shea B, O’Connell D, et al. The NewcastleOttawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. 2000. Available at: http://www.medicine.mcgill.ca/rtamblyn/Readings/ The%20Newcastle%20-%20Scale%20for%20assessing %20the%20quality%20of%20nonrandomised%20studies%20 in%20meta-analyses.pdf. Accessed September 4, 2015. 21. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994;50:1088-1101.

ARTICLE IN PRESS X.-W. ZHANG ET AL.

8 22. Egger M, Davey Smith G, Schneider M, et al. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997;315:629-634. 23. Duval S, Tweedie R. Trim and fill: a simple funnel-plotbased method of testing and adjusting for publication bias in meta-analysis. Biometrics 2000;56:455-463. 24. Spencer-Peet J, Wood D, Glatt MM. Gamma-glutamyl transpeptidase in alcoholism. Lancet 1972;1:1122-1123. 25. Liangpunsakul S, Qi R, Crabb DW, et al. Relationship between alcohol drinking and aspartate aminotransferase:alanine aminotransferase (AST:ALT) ratio, mean corpuscular volume (MCV), gamma-glutamyl transpeptidase (GGT), and apolipoprotein A1 and B in the U.S. population. J Stud Alcohol Drugs 2010;71:249252. 26. Freer DE, Statland BE. The effects of ethanol (0.75 g/kg body weight) on the activities of selected enzymes in sera of healthy young adults: 1. Intermediate-term effects. Clin Chem 1977;23:830-834. 27. O’Donnell MJ, Xavier D, Liu L, et al. Risk factors for ischaemic and intracerebral haemorrhagic stroke in 22 countries (the INTERSTROKE study): a case-control study. Lancet 2010;376:112-123. 28. Franzini M, Corti A, Martinelli B, et al. Gammaglutamyltransferase activity in human atherosclerotic

29.

30.

31.

32.

33.

34.

plaques—biochemical similarities with the circulating enzyme. Atherosclerosis 2009;202:119-127. Anderson ME, Allison RD, Meister A. Interconversion of leukotrienes catalyzed by purified gamma-glutamyl transpeptidase: concomitant formation of leukotriene D4 and gamma-glutamyl amino acids. Proc Natl Acad Sci USA 1982;79:1088-1091. Singh J, Chander J, Singh S, et al. Gamma-glutamyl transpeptidase: a novel biochemical marker in inflammation. Biochem Pharmacol 1986;35:3753-3760. Bo S, Gambino R, Durazzo M, et al. Associations between gamma-glutamyl transferase, metabolic abnormalities and inflammation in healthy subjects from a population-based cohort: a possible implication for oxidative stress. World J Gastroenterol 2005;11:7109-7117. Dominici S, Paolicchi A, Lorenzini E, et al. Gammaglutamyltransferase dependent prooxidant reactions: a factor in multiple processes. Biofactors 2003;17:187-198. Emdin M, Pompella A, Paolicchi A. Gammaglutamyltransferase, atherosclerosis, and cardiovascular disease: triggering oxidative stress within the plaque. Circulation 2005;112:2078-2080. Gurbuzer N, Gozke E, Ayhan BZ. Gamma-glutamyl transferase levels in patients with acute ischemic stroke. Cardiovasc Psychiatry Neurol 2014;2014:170626.