Alcohol and Alcoholism Advance Access originally published online on November 9, 2006
Alcohol and Alcoholism 2007 42(1):24-27; doi:10.1093/alcalc/agl090
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
ETHNICITY AND GAMMA-GLUTAMYLTRANSFERASE IN MEN AND WOMEN WITH ALCOHOL USE DISORDERS
1 Department of Medicine, School of Medicine and Biomedical Sciences NY 14260, USA
2 Research Institute on Addictions, University at Buffalo, State University of New York NY 14260, USA
3 Department of Biostatistics, School of Public Health and Health Professions, University at Buffalo, State University of New York NY 14260, USA
*Author to whom correspondence should be addressed at: Tel: +1 843 792-5226; Fax: +1 843 792-7353; E-mail: stewarsh{at}musc.edu
Received 2 March 2006; first review notified 13 June 2006; first review notified 28 August 2006; accepted 12 October 2006
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Aims: This study evaluated the associations of gender and ethnicity with GGT in a large sample of patients with DSM-IV alcohol abuse or dependence, as well as modification of alcohol's effects on GGT by gender and ethnicity. Methods: Subjects included 1691 African American, Mexican American, and non-Hispanic white individuals with DSM-IV alcohol dependence or abuse who participated in an alcoholism treatment trial. Detailed information on alcohol use was collected and GGT measured at baseline and at 3, 9, and 15 months post-baseline. Results: Changes in GGT occurring with changes in alcohol consumption were similar regardless of ethnicity. Although alcohol-associated changes were similar in these ethnic groups, African Americans had the highest average GGT at any given level of alcohol use. This ethnic pattern held for both sexes, with females having lower levels within each ethnic group. Drinking frequency had a slightly decreased association with GGT in females relative to males, but this effect was clinically unimportant. Conclusions: Gender and ethnic-specific cutoffs may be useful when screening for chronic heavy drinking, but the absolute increase in GGT occurring with relapse will be similar regardless of gender or ethnicity.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Gamma-Glutamyltransferase (GGT) has been used as a biomarker for chronically heavy alcohol consumption (Allen et al., 2003
). Validity studies have varied in populations studied, definitions of problem drinking, and criterion standards, but in general GGT has a sensitivity of 
45% at 90% specificity for detecting chronic heavy consumption (Conigrave et al., 2003). While some GGT elevation may occur with more modest consumption, several weeks of 
60–80 g of alcohol per day is probably required to cause abnormal levels, and normalization occurs over 4–8 weeks of abstinence (Whitfield, 2001). While clearly an imperfect marker, GGT does have some clinical utility as a means for detecting heavy drinking, but its greatest use may be as a marker of relapse to heavy drinking. When monitoring individuals with alcohol use disorders, a 30% increase from abstinent levels suggests that relapse has occurred (Anton et al., 2002). Because of limited test characteristics, GGT has been evaluated with other alcohol biomarkers such as carbohydrate-deficient transferrin (Salaspuro, 1999; Anton, 2001), and such approaches may lead to greater importance for GGT in screening for heavy drinking and for monitoring treatment outcomes (Litten et al., 1995; Huseby et al., 1997; Allen et al., 1999; Sillanaukee and Olsson, 2001; Chen et al., 2003).
GGT varies by gender and ethnicity among US adults, with males and African Americans having the highest average levels. In addition, analyses of survey data have shown that African Americans are more likely to have elevated GGT at higher levels of alcohol use (Stewart, 2002; Stranges et al., 2004). However, due to the cross-sectional designs, these studies were unable to assess if the effect of alcohol on GGT actually differed by ethnicity, or if other factors were responsible for ethnic differences. This is clinically relevant, since ethnicity may be an important consideration in determining optimal cutoffs for a positive screening result, and may affect the use of GGT as a relapse marker. This study evaluated the associations of gender and ethnicity with GGT in a large sample of patients with DSM-IV alcohol abuse or dependence, as well as modification of alcohol's effects on GGT by gender and ethnicity.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Subjects
Subjects included participants of Project MATCH, a large, multi-center randomized controlled trial that evaluated the relative effects of three counseling approaches in 1726 individuals (94% dependent, 6% abuse; Project MATCH Research Group, 1993
). Active treatment occurred for 12 weeks, with extended follow-up thereafter. This analysis included baseline measurements, as well as at 3, 9, and 15 month post-baseline data (corresponding to baseline, end-of-treatment, and 6 and 12 month post-treatment times in Project MATCH) (Project MATCH Research Group, 1997).
Dependent variable
GGT (U/l) was measured at each of the above-referenced time periods. Phlebotomy was performed as per the usual routine at each site on subjects participating in Project MATCH. GGT was measured locally and subsequently reported to the Coordinating Center. GGT was positively skewed, and was log transformed for analysis.
Independent variables
Alcohol consumption was measured using timeline follow-back methodology (Sobell and Sobell, 2000). This validated approach consists of presenting a calendar covering the time period of interest, which includes important dates in general and specifically for the subject. The subject is then instructed to estimate how many standard drinks were consumed on each day. Because alcohol use is difficult to describe with one variable, we modeled it in two ways, following what was done in the original Project MATCH design. The first variable was average number of drinks per drinking day (DDD) as a measure of drinking intensity. This was calculated as the total number of drinks divided by the total number of days on which any amount of alcohol was consumed. The second approach used percent drinking days (PDD) as a measure of drinking frequency. This was calculated as the percentage of all days on which any amount of drinking occurred. Detailed self-reported information on ethnicity was obtained. For this study, only those categorized as non-Hispanic white, African American, or Mexican American ethnicity were included, as there were too few subjects from other groups for reliable analysis. Results were also adjusted for age, gender, and study site.
Analysis
The primary aim of the analysis was to assess for ethnicity-associated modification of the relationship between GGT and alcohol use via an interaction term in the model, and subsequently assess main effects of ethnicity on GGT adjusting for alcohol consumption. Similar relationships between GGT and gender were also of interest. The relationship between log GGT and the alcohol use variables was modeled as an initial step, providing an average change in log GGT with changes in alcohol consumption for the entire study sample. Both PDD and DDD had significant non-linear relationships with log GGT, and all subsequent modeling included quadratic effects. Regression models were employed that described log GGT as a function of alcohol use (DDD or PDD), age, gender, study site, and ethnicity. The within-subject correlation was accounted for using a generalized covariance structure (Brown and Prescott, 1999). We also allowed the flexibility for the covariance structure to vary across ethnic groups. An alcohol use by ethnicity interaction was assessed to determine if the association of alcohol and GGT (i.e. the change in GGT expected with a given change in alcohol use) varied by ethnicity. This was not significant for DDD or PDD, and was not included in the final models. Interest then focused on main effects of ethnicity. A gender by alcohol use interaction was modeled to assess modification of the alcohol effect on GGT by gender. This was significant for PDD (P = 0.0003) for differing slopes, but not DDD, demonstrating a greater effect of PDD on GGT in males. To present the results in a clinically useful manner, expected GGT values (geometric means) were calculated by gender and ethnicity by back transforming log GGT values to the original units. This method provided a close estimate of median GGT.
Because alcohol use measures (DDD and PDD) were not normally distributed due to a floor effect from abstainers, we also assessed these measures as categorical variables (abstainers and drinking quartiles). Results were similar using this approach, and we report on analyses modeling alcohol use continuously as described above. To assess the influence of outliers, we also repeated analyses excluding the 10% of subjects with the highest absolute residuals. This did not result in any substantive changes.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Characteristics of the Project MATCH sample have been presented in detail (Project MATCH Research Group, 1997
). Characteristics of subjects included in this analysis are summarized in Table 1. The original study sample included 1726 subjects. Restricting the sample to Mexican American, African American, and non-Hispanic white participants yielded 1691 subjects (1381 non-Hispanic whites, 169 African Americans, and 141 Mexican Americans). Of a possible total of 6764 observations (a maximum of 4 per subject), 20% were missing GGT and/or alcohol data. The missing data was not strongly related to ethnicity (20% of Mexican American and non-Hispanic white, and 24% of African American observations were missing).
|
Modeling within-subject correlations was preferable to ignoring this correlation (P < 0.0001 for the null model likelihood ratio test), illustrating the importance of individual effects on GGT. As stated above, ethnicity did not modify the effect of alcohol on GGT (P-values for separate slopes were insignificant), but did exert important main effects on GGT. In the PDD analysis, log GGT was significantly lower in Mexican Americans [b = –0.472, 95% confidence interval (CI) = –0.708 to –0.237] and non-Hispanic whites (b = –0.538, 95% CI = –0.705 to –0.370) compared to African Americans. For example, for males drinking on 75% of all days, expected GGT (U/l) with 95% CI = was 103 (80–133) for African Americans, 64 (50–82) for Mexican Americans, and 60 (52–74) for non-Hispanic whites. A similar pattern held for females, although GGT values were lower compared to males in the same ethnic group. Results for the DDD analysis were similar. Log GGT was significantly lower in Mexican Americans (b = –0.418, 95% CI = –0.657 to –0.180) and non-Hispanic whites (b = –0.517, 95% CI = –0.688 to –0.347) compared to African Americans. Continuing with the same example, for males consuming an average of 10 DDD, expected GGT (U/l) was 71 (57–89) for African Americans, 47 (38–58) for Mexican Americans and 42 (36–50) for non-Hispanic whites. To illustrate ethnic and gender differences across a wider spectrum of alcohol involvement, predicted GGT values (transformed from the model-predicted log GGT) for ethnicity and gender-defined stratum are illustrated in Figures 1 and 2 for PDD and DDD, respectively. Mexican Americans and non-Hispanic whites were combined for this illustration, since these groups were not significantly different.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This analysis was mainly conducted to further explore the known association of GGT and ethnicity, with additional interest in gender effects. Results demonstrated that GGT varied by ethnicity among individuals with alcohol use disorders, with African Americans having higher GGT measurements relative to Mexican Americans and non-Hispanic whites. However, alcohol effects on GGT were similar regardless of ethnicity, indicating that the absolute change in GGT associated with a change in alcohol use did not vary by ethnicity. As expected, GGT levels were lower in females within each ethnic group.
Prior research has demonstrated the link between GGT and ethnicity, although the reasons for these differences have not been elucidated (Whitfield, 2001; Stranges et al., 2004). This analysis extends knowledge in this area by demonstrating that ethnic differences in GGT are likely independent of alcohol consumption, and shows that, at least among heavy drinkers, changes in alcohol use will have a similar effect on GGT regardless of ethnicity. Importantly, while our results suggest that the effect of alcohol on GGT does not vary by ethnicity, GGT itself does differ by ethnicity, presumably for non-alcohol related reasons. Additional research could focus on establishing ethnic-specific cutoffs for positive results when screening for chronically heavy drinking. However, this is a crude approach at best, as ethnicity should be considered a marker for differing prevalence of cultural, socioeconomic, or possibly genetic effects on GGT (Burchard et al., 2003). More productive work would focus on determining the causes of ethnic and even individual differences, as this may lead to more valid use of GGT as an alcohol biomarker.
Regarding gender, a lower level of GGT in females is well documented (Conigrave et al., 2002), but there is uncertainty about gender modification of the effect of alcohol on GGT. In one study sample, GGT was more responsive to alcohol consumption in males (Whitfield et al., 1978). Our finding that drinking frequency had a greater effect on GGT in males is consistent with that work. In this sample however the finding was clinically insignificant, as the effect was quantitative and small. GGT will likely respond to alcohol use similarly regardless of gender.
Strengths of this study included a large, ethnically diverse study sample, with repeated, detailed estimates of alcohol use and longitudinal measurement of GGT. As a result, we were able to account for subject effects on GGT, model alcohol-associated change in GGT, and estimate ethnic modification of the alcohol effect on GGT. Limitations included the observational nature of the analysis and reliance on self-reported alcohol consumption. While this may bias the estimated relationship between alcohol use and GGT, it is less likely that this bias would impact our estimates of ethnic factors. There is some evidence that African Americans may minimize involvement with other drugs (Del Boca and Darkes, 2003), and similar underreporting of alcohol consumption in this group may have contributed to their higher GGT values. However, even among lifetime abstainers, GGT appears to be higher in African Americans (Stranges et al., 2004). GGT has been shown to vary with other factors such as body-mass index and smoking. While our ethnic estimates may be partially confounded by such factors, cross-sectional evaluations suggest ethnic effects are largely independent of these (Stewart, 2002; Stranges et al., 2004). Alcohol and BMI also seem to have independent effects on GGT (Daeppen et al., 1998), so BMI is unlikely to alter the alcohol-related changes in GGT modeled in this study. An additional limitation is due to the missing data, which we treated as missing at random. The proportion of missing data was similar among the ethnic groups, making bias from this assumption less likely, and probably not of sufficient magnitude to change our conclusions. Finally, these results only apply to individuals with alcohol use disorders. Sensitivity of GGT to alcohol consumption would likely be diminished among populations with less severe alcohol involvement (Whitfield, 2001), and it is possible that ethnic effects are different in such groups.
GGT is one of several biomarkers for chronically heavy alcohol consumption, and its use either alone or in combination with other markers can improve the detection of or relapse to this pattern of drinking. For screening purposes, higher cutoffs may be appropriate for males and African Americans. If GGT is used to monitor alcohol consumption for an individual, absolute changes in GGT related to decreased drinking or relapse will be similar regardless of ethnicity and gender.
|
ACKNOWLEDGEMENTS |
|---|
This work was supported in part by a Career Development Award from NIAAA (K23AA14188).
|
FOOTNOTES |
|---|
Present address: Center for Drug and Alcohol Programs, Medical University of South Carolina, 67 President Street, PO Box 250861, Charleston, SC 29425, USA 
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Allen J. P., Sillanaukee P., Anton R. (1999) Contribution of carbohydrate deficient transferrin to gamma glutamyl transpeptidase in evaluating progress of patients in treatment for alcoholism. Alcoholism: Clinical and Experimental Research 23:115–120.[CrossRef][Web of Science][Medline]
Allen J. P., Sillanaukee P., Strid N., et al. (2003) Biomarkers of heavy drinking. In Assessing Alcohol Problems. A Guide for Clinicians and Researchers. 2nd edn (National Institutes of Health Publication No. 03-3745In Allen J. P. and Wilson V. B. (Eds.). , Bethesda, MD).
Anton R. F. (2001) Carbohydrate-deficient transferrin for detection and monitoring of sustained heavy drinking. What have we learned? Where do we go from here? Alcohol 25:185–188.[CrossRef][Web of Science][Medline]
Anton R. F., Lieber C., Tabakoff B. (2002) Carbohydrate-deficient transferrin and gamma-glutamyltransferase for the detection and monitoring of alcohol use: results from a multisite study. Alcoholism: Clinical and Experimental Research 26:1215–1222.[CrossRef][Web of Science][Medline]
Brown H. and Prescott R. (1999) Repeated measures data. In Applied Mixed Models in Medicine. (John Wiley and Sons, New York).
Burchard E. G., Ziv E., Coyle N., et al. (2003) The importance of race and ethnic background in biomedical research and clinical practice. New England Journal of Medicine 348:1170–1175.
Chen J., Conigrave K. M., Macaskill P., et al. (2003) Combining carbohydrate- deficient transferrin and gamma-glutamyltransferase to increase diagnostic accuracy for problem drinking. Alcohol and Alcoholism 38:574–582.
Conigrave K. M., Degenhardt L. J., Whitfield J. B., et al. (2002) CDT, GGT, and AST as markers of alcohol use: the WHO/ISBRA collaborative project. Alcoholism: Clinical and Experimental Research 26:332–339.[CrossRef][Web of Science][Medline]
Conigrave K. M., Davies P., Haber P., et al. (2003) Traditional markers of excessive alcohol use. Addiction 98:Suppl. 2, 31–43.
Daeppen J. B., Smith T. L., Schuckit M. A. (1998) Influence of age and body mass index on gamma-glutamyltransferase activity: a 15-year follow-up evaluation in a community sample. Alcoholism: Clinical and Experimental Research 22:941–944.[CrossRef][Web of Science][Medline]
Del Boca F. K. and Darkes J. (2003) The validity of self-reports of alcohol consumption: state of the science and challenges for research. Addiction 98:Suppl. 2, 1–12.[Web of Science][Medline]
Huseby N. E., Nilssen O., Kanitz R. D. (1997) Evaluation of two biological markers combined as a parameter of alcohol dependency. Alcohol and Alcoholism 32:731–737.
Litten R. Z., Allen J. P., Fertig J. B. (1995) Gamma-glutamyltranspeptidase and carbohydrate- deficient transferrin: alternative measures of excessive alcohol consumption. Alcoholism: Clinical and Experimental Research 19:1541–1546.[CrossRef][Web of Science][Medline]
Project MATCH Research Group. (1993) Project MATCH: Rationale and methods for a multisite clinical trial matching patients to alcoholism treatment. Alcoholism: Clinical and Experimental Research 17:1130–1145.[Web of Science][Medline]
Project MATCH Research Group. (1997) Matching alcoholism treatments to client heterogeneity: Project MATCH posttreatment drinking outcomes. Journal of Studies on Alcohol 58:7–29.[Web of Science][Medline]
Salaspuro M. (1999) Carbohydrate-deficient transferrin as compared to other markers of alcoholism: a systematic review. Alcohol 19:261–271.[CrossRef][Web of Science][Medline]
Sillanaukee P. and Olsson U. (2001) Improved diagnostic classification of alcohol abusers by combining carbohydrate-deficient transferrin and gamma-glutamyltransferase. Clinical Chemistry 47:681–685.
Sobell L. C. and Sobell M. B. (2000) Alcohol timeline followback. In Handbook of Psychiatric Measures. (American Psychiatric Association, Washington, DC).
Stewart S. H. (2002) Racial and ethnic differences in alcohol-associated aspartate aminotransferase and gamma-glutamyltransferase elevation. Archives of Internal Medicine 162:2236–2239.
Stranges S., Freudenheim J. L., Muti P., et al. (2004) Greater hepatic vulnerability after alcohol intake in African Americans compared with Caucasians: a population-based study. Journal of the National Medical Association 96:1185–1192.[Medline]
Whitfield J. B. (2001) Gamma glutamyl transferase. Critical Reviews in Clinical Laboratory Sciences 38:263–355.[CrossRef][Web of Science][Medline]
Whitfield J. B., Hensley W. J., Bryden D., et al. (1978) Effects of age and sex on biochemical responses to drinking habits. Medical Journal of Australia 2:629–632.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||