American Journal of Epidemiology Copyright 2 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved Vol. 152, No. 8 Printed in U.S.A. Associated Factors of CDT and GGT Sillanaukee et al. Dose Response of Laboratory Markers to Alcohol Consumption in a General Population Pekka Sillanaukee, 1 3 Nuria Massot, 1 Pekka Jousilahti, 4,5 Erkki Vartiainen, 4 Jouko Sundvall, 6 Ulf Olsson, 7 Kari Poikolainen, 8,9 Maritta Pönniö, 1 John P. Allen, 1 and Hannu Alho 9,11 The dose response to alcohol use of carbohydrate-deficient transferrin (CDT), γ-glutamyltransferase (GGT), and their combination (γ-cdt) was studied in an age- and gender-stratified, random sample from Finland in 1997. A linear association with a threshold between alcohol consumption and the three markers was observed. was negatively associated with CDT and positively with GGT. was positively associated with GGT and γ-cdt. In conclusion, CDT appears to be an early phase marker of alcohol consumption. The combined marker, γ-cdt, was less associated with factors such as body mass index but more strongly correlated with alcohol consumption than were the two markers separately. Am J Epidemiol 2;152:747 51. alcohol drinking; gamma-glutamyltransferase; transferrin Among all the routine biochemical tests for alcohol abuse, serum γ-glutamyltransferase (EC.2.2, GGT), a widely used indicator of hepatobiliary disorders, and carbohydrate-deficient transferrin (CDT), the main plasma protein involved in the transport of the soluble iron, are the most accurate currently available (1 3). However, to incorporate laboratory markers into clinical practice, it is important to determine dose response, associated factors, and diseases influencing results in a general population. Previous reports have focused on the effect on the markers of varied levels of alcohol consumption among healthy Received for publication December 21, 1998, and accepted for publication December 3, 1999. Abbreviations: CDT, carbohydrate-deficient transferrin; γ-cdt, combination of carbohydrate-deficient transferrin and γ-glutamyltransferase; GGT, γ-glutamyltransferase. 1 Alcohol-related Diseases, Pharmacia & Upjohn Diagnostics AB, Uppsala, Sweden. 2 Medical School, University of Tampere, and Dept. of Clinical Chemistry, Tampere University Hospital, Tampere, Finland. 3 Dept. of Neuroscience, Karolinska Institute, Stockholm, Sweden. 4 Dept. of Epidemiology and Health Promotion, National Public Health Institute, Helsinki, Finland. 5 Tampere School of Public Health, Tampere, Finland. 6 Dept. of Biochemistry, National Public Health Institute, Helsinki, Finland. 7 Dept. of Biometry and Informatics, Swedish University of Agricultural Sciences, Uppsala, Sweden. 8 Järvenpää Addiction Hospital, Haarajoki, Finland. 9 Dept. of Mental Health and Alcohol Research, National Public Health Institute, Helsinki, Finland. 1 National Institute on Alcohol Abuse and Alcoholism, Rockville, MD. 11 Research Unit of Alcohol Diseases, University of Helsinki, Helsinki, Finland. Correspondence to Dr. Pekka Sillanaukee, Oy Finnish Immunotechnology, Ltd., Lenkkeilijänkatu 8, 33 52 Tampere, Finland (e-mail: Pekka.Sillanaukee@finnish-immunotech.com). subjects, but results have been inconclusive (4 6). It has been believed that CDT levels increase after alcohol consumption exceeding 6 g/day for at least 2 3 weeks, and GGT has been reported to rise after an alcohol consumption of 8 2 g/day after one or several weeks (7, 8). The population determinants have been studied more thoroughly for GGT than for CDT. False positive results for CDT can be caused by an inborn error of glycoprotein metabolism, pregnancy, genetic D-variant of transferrin, and severe liver diseases (1, 4, 9). On the other hand, increased GGT have been reported in response to most hepatobiliary disorders, obesity, diabetes, and hypertriglyceridemia (8, 1 13). Various authors have reported associations between GGT and age (14) as well as a weak positive association between this factor and CDT among men (4) and women (15). Other possible influences on CDT levels await explanation. The aim of this study was to determine the distribution of and the dose of alcohol required to increase CDT, GGT, and their mathematical combination (γ-cdt) (16). The association of body mass index, age, and smoking status with CDT, GGT, and γ-cdt was also evaluated. MATERIALS AND METHODS A large cross-sectional risk factor survey was carried out in five geographic areas of Finland at the beginning of 1997. In each study area, an age- and gender-stratified random sample was drawn from the population aged between 25 and 74 years. The participation rate was 71 percent among men and 76 percent among women. Complete data on CDT and GGT were available for 4,11 women and 4,14 men. Pregnant women (n 76), those who had used hormone treatment (n 197), and those presently using any hormone treatment (n 656) were excluded. Because of 747
748 Sillanaukee et al. TABLE 1. Distribution of CDT,* GGT,* and γ-cdt* among the study population, Finland, 1997 No. Mean Median Range Females Mean Median Range Mean Median Range 86 736 68 1,215 271 1 16.6 19.4 9.7 3.7 19.1 () 2. (7.1) 19.1 (7.2) 17.6 (6.7) 1 (5.) 18.2 19.1 17.6 1 15.1 5.3 7 6 6 34.7 5.2 1 1 19.1 (2) 2 (2) 27.9 (29.) 3 (5) 29.6 (2) 15.5 17. 2 2 488 317 294 1,584 131 6.7 5.3 8.7 () () () () () BMI* 188 1,283 955 414 28 2 1 8.6 5.8 2 (7.3) 2. (7.1) 17.6 () 16. () 14.7 () 2 18.8 1 15.2 14.3 42.6 7 5 38.8 2 8.5 15. 19.7 (1) 19.3 (1) 28.6 (5) 33.7 (3) 4 (36.) 16. 16. 25. 29. 195 188 1,583 329 274 5.2 6.9 7. 9.6 () () () () () Males 756 836 828 1,777 645 1 1 1 1 1 14.3 () 15. (6.8) 14.8 (6.9) 15.1 (7.3) 1 () 1 1 1 1 1 5 74.5 65.7 67.3 83.7 8.3 13.7 8.2 33.2 (26.7) 4 (4) 59.6 (13) 55. (12) 4 (4) 26. 3. 37. 34. 3. 286 88 424 1 1 1 () () 6.5 () 6.5 () () 4.3 7.2 7.7 BMI 67 1,29 1,961 631 146 29.9 1 8.4 7.5 17.5 (7.9) 1 (8.) 14.3 () 13.9 (5.6) 1 () 15.7 1 13.2 1 1 35. 83.7 7 65.2 4.3 6. 4.5 9.9 22.6 3 (26.5) 3 (85.7) 45. (49.8) 6 (66.5) 8 (186) 2 25. 3 4 48. 133 972 713 2,155 17.9 1 1 15.2 18.5 () () () 6.5 () 6.6 () * CDT, carbohydrate-deficient transferrin; GGT, γ-glutamyltransferase; γ-cdt, combination of CDT and GGT; BMI, body mass index. The range is the distance between the largest and smallest in the sample and is expressed as the value of the difference. Numbers in parentheses, standard deviation. 6.5 6.6 7.3 missing data on alcohol consumption, 4 men and 94 women were excluded. Thus, 2,988 females and 3,974 males were included in the present study. A questionnaire about health behavior and alcohol-related factors was sent to subjects in advance. Questionnaires were returned to the survey site during a medical examination. was calculated as the ratio of weight (kg) to height squared (m 2 ). Alcohol consumption from the preceding week and mean alcohol consumption during the past year were calculated and expressed as grams of pure ethanol per week. The frequency of alcohol intoxication per year was also evaluated. Serum CDT levels were analyzed by a double antibody assay (CDTect; Pharmacia & Upjohn Diagnostics, Uppsala, Sweden). Levels of GGT were measured by the kinetic method (Oy Medix Biochemica AB, Kauniainen, Finland) TABLE 2. Summary of significant correlations between the natural logarithm of CDT,* GGT,* and γ-cdt* and selected indicators of alcohol consumption and health behavior in males (n = 3,974) and females (n = 2,988) in a random sample drawn from a general population from Finland, 1997 Variable Males (n = 3,974) Females (n = 2,988) Alcohol consumption during the preceding week Alcohol consumption during the past year Frequency of alcohol intoxication 8 9 4.4.16 8.3 1.19.11.35.38.3.15.14.19.16 2 4.13.32.14.13.9 9.9.38 4.19.13.14.8 * CDT, carbohydrate-deficient transferrin; GGT, γ-glutamyltransferase; γ-cdt, combination of CDT and GGT. Associations at p <.1.
Associated Factors of CDT and GGT 749 on the recommendation of the European Committee for Clinical Laboratory Standards (ECCLS) (17). Univariate relations among the three markers, behavioral factors, and alcohol consumption indicators were evaluated by analysis of variance. The critical alcohol consumption dose was calculated by using piecewise regression models (18). The SAS procedures GLM and Nlin (SAS Institute, Inc., Cary, North Carolina) were used for the analyses, along with a custom-written SAS macro. Thresholds of the association between alcohol consumption and the three markers were estimated simultaneously with other parameters of the model (19). TABLE 3. Thresholds of the association between alcohol consumption and CDT,* GGT,* and γ-cdt,* Finland, 1997 Marker CDT GGT γ-cdt Threshold of alcohol consumption (g/week) Males 5 73.7 66.7 Females 1 6.3 4 * CDT, carbohydrate-deficient transferrin; GGT, γ-glutamyltransferase; γ-cdt, combination of CDT and GGT. Odds ratios and 95 percent confidence intervals were calculated for age, alcohol consumption, smoking status, and body mass index. RESULTS In females the mean for CDT decreased and GGT increased with age and body mass index. In males, CDT decreased with body mass index, while the mean for GGT and γ-cdt increased (table 1). In males, CDT and γ-cdt were most strongly associated with alcohol-related factors, while GGT was most strongly correlated with alcohol intake during the past year and body mass index (table 2). A positive association was also observed between age and GGT and γ-cdt. In females, only γ-cdt was ly correlated with both measures of alcohol intake. was negatively related to CDT and positively related to GGT levels (table 2). In both genders, body mass index was negatively associated with CDT and positively associated with GGT and γ-cdt. In males, CDT were elevated when alcohol consumption was er than 5 g/week, whereas the levels of GGT and γ-cdt increased with alcohol consumption exceeding 73.7 and 66.7 g/week, respectively. In females, the critical doses were 1, 6.3, and 4 g/week, respectively (table 3). TABLE 4. Odds ratios (ORs) and 95 confidence intervals (CIs) for elevated CDT* in males (2 units/, GGT* (8 units/, and their combination (γ-cdt) (7.2) by age, alcohol intake, cigarette smoking, and body mass index, unadjusted and adjusted for the other row variables, Finland, 1997 Covariate Unadjusted Adjusted (years),,,,, 5.2, 4.5,,, 3.9,, 2.2,,, 3.7, 3.2,,, 3.7,,,, 5.3, 4.7 Alcohol consumption during the past year (g/week) 1 7 71 14 141 21 211 28 281 35 351 42 >42 4. 3.9,,,,, 6.6, 6.9 5.7, 15.3 2.6,,,, 3.9, 6.9, 8. 5.5, 15.1 5.3 6.5 13.9,,,, 5.6, 8.5 3.7, 1 8.4, 2 3.7,,, 3.7, 5.5, 7.1, 5.3, 1 9.6,,, 3.9, 7.7 2.2, 7.6 5.6, 16.5 5.7 13.9,, 2.2,,, 9.3, 1 8.2, 23.9 1 1 11 2 >2, 3.2 2.6, 2.6,,,, 3.9 3.2,, 2.2,, 2.2, 3.7, 4.,,,,,.3.3,,.1,.1,.3,,,, 6..3,.3,,,.1.3,.1,.1,.3.1,.3,,,, 3.9,,.3,.3, * CDT, carbohydrate-deficient transferrin; GGT, γ-glutamyltransferase. The first group of each variable was considered the reference group with an odds ratio equal to.
75 Sillanaukee et al. TABLE 5. Odds ratios (ORs) and 95 confidence intervals (CIs) for elevated CDT* in females (26 units/, GGT* (5 units/, and their combination (γ-cdt) (7.2) by age, alcohol intake, cigarette smoking, and body mass index, unadjusted and adjusted for the other row variables, Finland, 1997 Covariate Unadjusted Adjusted (years).3,,.3,.1, 6.9 7.7, 5.3, 9.7, 1 4.3, 14.3 3.9 2.2, 4.3, 6.6, 5., 4.3, 2.2,,, 4. 5.,,, 7.5, 9.6, 4. 2.2, 6.6, 5.6, Alcohol consumption during the past year (g/week) 1 7 71 14 141 21 211 28 >28 7.4 5.6,,, 6.6 3.2, 16.6, 1 7.3 4.7,,, 4.8, 16.6, 1 2.1 37.1,,, 8.6 8.8, 46. 1, 12,, 3.2, 5.1, 14.3, 1 1 5.,, 7. 4., 26.9, 17.1 4. 17.7 23.7,,, 8. 7.3, 4 7.7, 8 1 1 11 2 >2,,, 7.6,,.3,, 2.2, 5., 1,,, 6.6,,,,,, 7.3.4,,.1,.,.1 4.7,,, 6.8, 9.6,, 2.2,,.3.4,,.1,.,.1,,, 5.1, 7.3,,,, * CDT, carbohydrate-deficient transferrin; GGT, γ-glutamyltransferase. The first group of each variable was considered the reference group with an odds ratio equal to. In both genders, increasing alcohol consumption increased the risk of having er of the three markers. γ-cdt produced the est odds ratios in different alcohol-drinking categories (tables 4 and 5). The odds ratios did not change much after adjustment for age, smoking, and body mass index. In both genders, the risk of elevated CDT and γ-cdt increased with increasing number of cigarettes smoked per day. The association between smoking and GGT was weaker and was not significant in females. In the unadjusted analyses, increased body mass index were associated with a er probability of having low levels of CDT and elevated levels of GGT, whereas γ- CDT was independent from body mass index. Adjusting for age, alcohol intake and smoking did not change these associations, which were similar in both men and women. In males the risk of having GGT and γ-cdt levels increased from the age group years. After adjustment for body mass index, alcohol intake, and smoking, the risk of a CDT level also increased with age. Females aged years had an increased risk for levels of the three markers, while those older than 55 years had an increased risk of having low CDT and GGT and γ-cdt levels. The adjustment for alcohol intake, body mass index, and smoking affected only the association between CDT and age for the group of women older than 55 years, which was not significant. DISCUSSION Distribution analyses for CDT and GGT confirmed that the cutoffs for CDT (2 units/liter for males and 26 units/liter for females) and for GGT (8 units/liter and 5 units/liter, respectively, as recommended by the European Committee for Clinical Laboratory Standards) (17) are appropriate for the general population of Finland. The previously suggested limit of 6.5 for γ-cdt for both genders was studied among 2, subjects (16) and found to be clearly lower than our 9 percentile limits. The original work was done according to previously recommended and methodology for GGT, whereas the present study followed those presented by the European Committee for Clinical Laboratory Standards. The GGT measurement with this new standard method is based on the addition of pyridoxal phosphate as a cofactor known to increase enzyme activity by approximately 1 percent. This probably accounts for the elevation of the cutoff value for γ-cdt from 6.5 to 7.2. A limit of 7.2 as demonstrated in the present study is recommended. In both genders, the strongest effect of age was seen on GGT, while γ-cdt was least influenced by this variable. It is striking that in females, the odds ratios for CDT decrease and those for GGT increase with age. Our results indicate that in both genders, γ-cdt is more strongly correlated with alcohol-related factors and less affected by other factors such as body mass index.
Associated Factors of CDT and GGT 751 Within the general population, the level of alcohol needed for the marker to exceed the reference depends on gender and on the particular marker. Among women, CDT increased with lower alcohol consumption (15 g/week) compared with GGT and γ-cdt, indicating its potential as an early phase marker. It is notable that in women, the alcohol consumption needed to produce a response in each of the three markers was much lower than in men. This phenomenon might be related to an earlier effect of alcohol abuse on the metabolism of females (2 22). In conclusion, our study shows that in a general population the increase of CDT, GGT, and γ-cdt can be observed with lower alcohol consumption than that reported in studies based on controlled drinking of individuals (1, 5, 6, 8). A significant association with age, body mass index, and smoking was observed for the three markers. The results presented here for γ-cdt, together with its of sensitivity and specificity among the general population (23), suggest the usefulness of this new combined marker in clinical practice. REFERENCES 1. Stibler H. Carbohydrate-deficient transferrin in serum: a new marker of potentially harmful alcohol consumption reviewed. Clin Chem 1992;37:219 37. 2. Anton RF, Sillanaukee P. The use of carbohydrate-deficient transferrin as an indicator of alcohol consumption during treatment and follow-up. Alcohol Clin Exp Res 1996;2:54 6. 3. Allen JP, Litten RZ, Anton RF, et al. Carbohydrate-deficient transferrin as measure of immoderate drinking: remaining issues. Alcohol Clin Exp Res 1994;18:799 812. 4. Stibler H, Borg S, Beckman G. Transferrin phenotype and level of carbohydrate-deficient transferrin in healthy individuals. Alcohol Clin Exp Res 1988;12:45 3. 5. Salmela KS, Laitinen K, Nyström M, et al. Carbohydratedeficient transferrin during 3 weeks heavy alcohol consumption. Alcohol Clin Exp Res 1994;18:228 3. 6. Storey EL, Mack U, Powell LW, et al. Use of chromatofocusing to detect a transferrin variant in serum of alcoholic subjects. Clin Chem 1985;31:1543 5. 7. Stibler H. Carbohydrate-deficient transferrin in serum: a new marker of potentially harmful alcohol consumption reviewed. Clin Chem 1991;37:229 37. 8. Sillanaukee P. Laboratory markers of alcohol abuse. Alcohol Alcohol 1996;31:613 16. 9. Bean P, Peter JB. Allelic D variants of transferrin in evaluation of alcohol abuse: differential diagnosis by isoelectric focusingimmunoblotting-laser densitometry. Clin Chem 1994;4: 278 83. 1. Meregalli M, Giacomini V, Lino S, et al. Carbohydratedeficient transferrin in alcohol and nonalcohol abusers with liver disease. Alcohol Clin Exp Res 1995;19:1525 7. 11. Whitfield JB, Pounder RE, Neale G, et al. Serum gammaglutamyltranspeptidase activity in liver disease. Gut 1972;13: 72 8. 12. Ellis G, Worthy E, Goldberg DM. Lack of value of serum gamma-glutamyltransferase in the diagnosis of hepatobiliary disease. Clin Biochem 1979;12:142 5. 13. Kristenson H, Trell E, Fex G, et al. Serum gamma-glutamyltransferase: statistical distribution in a middle-aged male population and evaluation of alcohol habits in individuals with elevated levels. Prev Med 198;9:18 19. 14. Van der Meulen EA, van Sittert NJ, de Koningh AG, et al. General approach to correction for bias in analytical performance in longitudinal studies illustrated by estimating the effect of age on gamma-glutamyltransferase activity. Clin Chem 1993; 39:1375 81. 15. Stauber RE, Vollmann H, Pesserl I, et al. Carbohydratedeficient transferrin in healthy women: relation to estrogens and iron status. Alcohol Clin Exp Res 1996;2:1114 17. 16. Sillanaukee P, Olsson U. Improved diagnostic classification of alcohol abusers by combining carbohydrate deficient transferrin and gamma-glutamyltransferase. Clin Chem (in press). 17. Leino A, Impivaara O, Irjala K, et al. Health-based reference intervals for ALAT, ASAT and GGT in serum, measured according to the recommendations of the European Committee for Clinical Laboratory Standards (ECCLS). Scand J Clin Lab Invest 1995;55:243 5. 18. Seber GAF, Wild CJ. Nonlinear regression. New York, NY: Wiley, 1989. 19. Friendly M. SAS macro programs for statistical graphics: DENSITY. 1996. (SAS macro posted at http://pascal.math. yourku.ca/scs/sssg/density.html). 2. Saunders JB, Davis M, Williams R. Do women develop alcoholic liver disease more readily than men? Br Med J 1981;282: 114 3. 21. Tuyns AL, Pequignot G. Greater risk of ascitic cirrhosis in females in relation to alcohol consumption. Int J Epidemiol 1984;13:53 7. 22. Lieber CS. Medical disorders of alcoholism. N Engl J Med 1995;333:158 65. 23. Sillanaukee P, Massot N, Jousilahti P, et al. Enhanced clinical utility of γ-cdt in a general population. Alcohol Clin Exp Res 2;24:122 6.