METHOD-DEPENDENT CHARACTERISTICS OF CARBOHYDRATE-DEFICIENT TRANSFERRIN MEASUREMENTS IN THE FOLLOW-UP OF ALCOHOLICS

Transcription

1 Alcohol & Alcoholism Vol. 39, No. 1, pp , 2004 doi: /alcalc/agh021, available online at METHOD-DEPENDENT CHARACTERISTICS OF CARBOHYDRATE-DEFICIENT TRANSFERRIN MEASUREMENTS IN THE FOLLOW-UP OF ALCOHOLICS PETRA ANTTILA 1, KIMMO JÄRVI 2, JAANA LATVALA 1 and ONNI NIEMELÄ 1,3,* Departments of 1 Clinical Chemistry and 2 Psychiatry, EP Central Hospital, Seinäjoki and 3 University of Tampere, Tampere, Finland (Received 4 August 2003; first review notified 18 September 2003; in revised form 20 October 2003; accepted 29 October 2003) Abstract Aims: There are only limited data comparing the diagnostic characteristics of carbohydrate-deficient transferrin (CDT) measurements in assays for excessive alcohol consumption under controlled conditions. Methods: We compared different CDT assays and the conventional laboratory markers of ethanol consumption, γ-glutamyl transferase (gamma-gt) aspartate aminotransferase (AST) and mean corpuscular volume (MCV) in the assessment and follow-up of 36 alcoholics (31 men, five women, mean age 44 years), who were admitted for detoxification. Detailed interviews to assess the amount of alcohol consumption were carried out for each patient. A hospital follow-up with supervised abstinence for 8 ± 4 days (range 5 19 days) was carried out for 17 patients. Controls were 30 apparently healthy individuals (22 men, eight women, mean age 49 years), who had no history of hazardous drinking. Results: At the time of admission, the %CDT method, which excludes the trisialotransferrin isoform from the measurement, yielded elevated values in 69% of the patients, compared to 61% for CDTect. The corresponding sensitivities for gamma-gt, AST and MCV were 61, 56 and 47%, respectively. The self-reported alcohol consumption for a period of 1 month prior to admission showed a stronger correlation with the %CDT results (r = 0.59, P = ) than with the CDTect results (r = 0.36, P = 0.04), GT (r = 0.40, P = 0.02) or AST (r = 0.35, P = 0.05). During follow-up with supervised abstinence the mean %CDT values were found to show a slower rate to normalization (mean 14 ± 4 days) than the CDT values measured with the CDTect method (mean 10 ± 5 days) (P < 0.05). Conclusions: The data indicate distinct differences and method-dependent rates of normalization in CDT assays, possibly reflecting different degrees of transferrin desialylation in the alcoholics. The present findings should be considered in studies on alcohol markers for monitoring abstinence. INTRODUCTION Sialic acid-deficient isoforms of transferrin are known to increase as a result of excessive alcohol consumption. Such isoforms have traditionally been referred to as CDT (carbohydrate-deficient transferrin, desialotransferrin), and their measurements have been widely used as diagnostic tools for detecting alcohol-related health problems. However, the interpretations of the assay results have suffered from a lack of uniform international standardization (Arndt, 2001; Helander et al., 2001; Conigrave et al., 2002; Helander, 2002; Tagliaro et al., 2002). In healthy people, the most common transferrin isoform is tetrasialotransferrin, whereas in alcoholics, various degrees of sialic acid-deficient isoforms are generated (Stibler, 1991; Allen et al., 1994; Helander et al., 2001). Usually, the definition of CDT has referred to those with 0 2 or 0 3 sialic acid side chains (Stibler, 1991; Allen et al., 1994; Arndt, 2001; Helander, 2002). Various methods with different reactivities to different transferrin isoforms are currently available. The diagnostic characteristics of such methods in the clinical assays for detecting alcohol misuse have, however, not been fully established (Scouller et al., 2000). The new %CDT assay, which excludes the trisialo isoform (the isoform with three sialic acid side chains) from the measurement, has recently been suggested to yield improved specificity as compared to the conventional CDTect method, which has previously been the most widely used CDT assay (Anton et al., 2001). * Author to whom correspondence should be addressed at: EP Central Hospital, Laboratory, FIN Seinäjoki, Finland. Tel.: ; Fax: ; onni.niemela@epshp.fi We designed this study to compare the characteristics of the different CDT methods in the detection and follow-up of alcoholics, who were admitted for detoxification. For comparison, the conventional alcohol markers (GT, AST, MCV) were also measured. The data suggest distinct differences between the different CDT assays in monitoring alcohol drinking and abstinence. MATERIALS AND METHODS Patients and control subjects We studied 36 heavy drinkers (31 men, 5 women, mean age 44 ± 10; range years), who were admitted for detoxification, and 30 apparently healthy control subjects (22 men, 8 women, mean age 49 ± 13; range years). All the alcoholic patients showed a well-documented history of continuous alcohol consumption or binge drinking, which consisted of 132 ± 71 (mean ± SD) g/day during the 4-week period prior to sampling. The documentation of alcohol misuse was based on detailed personal interviews using a time-line follow-back technique. The patients were asked how many drinks of alcohol (standard drink = 12 g of ethyl alcohol corresponding to one beer, one glass of table wine or 3 cl 40% proof spirit) they had consumed during the: (1) 24 h, (2) 1 week and (3) 4 weeks preceding admission. The mean duration of abstinence prior to sampling was 1.8 ± 2.0 (range 0 6) days. A follow-up with supervised abstinence during hospitalization for 8 ± 4 (range 5 19) days was carried out with 17 patients. Blood-alcohol concentrations were controlled by repeated analyses from breath air. The main clinical and laboratory characteristics of the study population are summarized in Table Alcohol & Alcoholism Vol. 39, No. 1 Medical Council on Alcohol 2004; all rights reserved

2 60 P. ANTTILA et al. Table 1. Clinical and laboratory characteristics of the study population Parameter Alcoholics Controls n Men/Women 31/5 22/8 Age (years) 44 ± ± 13 Ethanol consumption a (g/day) 132 ± 71**** 0.9 ± 1.7 CDT (%CDT method) (%) 4.9 ± 2.7**** 2.0 ± 0.4 CDT (CDTect method) (U/l) 29 ± 15**** 14 ± 5.6 GT (U/l) 185 ± 291**** 26 ± 11 AST (U/l) 65 ± 50**** 32 ± 28 MCV (fl) 96 ± 5.0** 92 ± 3.5 Bilirubin (µmol/l) 17 ± ± 5.5 Albumin (g/l) 44 ± ± 2.7 The values are presented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < a Mean daily consumption from the past four weeks prior to sampling. CDT, carbohydrate-deficient transferrin; GT, γ-glutamyl transferase; AST, aspartate aminotransferase; MCV, mean corpuscular volume. For reference values, see Methods section. The controls were either abstainers (n = 20; 16 men, four women, mean age 47 ± 9 years) or social drinkers (n = 10; six men, four women, mean age 55 ± 12 years) whose mean daily ethanol consumption did not exceed a mean of 15 g ethanol per day or 40 g on any single occasion. All serum samples were stored at 70 C until analysis. All participants in the study gave their informed consent and the study was carried out according to the provisions of the Declaration of Helsinki. CDT analyses The concentration of CDT was measured in serum samples by two different methods. First, a turbidimetric immunoassay after ion-exchange chromatography (%CDT; Axis-Shield, Oslo, Norway), which detects transferrin variants with 0 2 sialic acid residues, was used. In this assay, CDT concentration is expressed as a percentage of the total amount of serum transferrin. The measurements were carried out on Behring Nephelometer II (Dade Behring; Behring Diagnostics, Marburg, Germany) according to the instructions of the manufacturer s. Amounts exceeding 2.60% were considered high (Anton et al., 2001). In addition, CDT was measured by a competitive radioimmunoassay after micro-column separation (CDTect). In this assay, serum transferrin isoforms with 0 1 sialic acid residues (α- and monosialotransferrins) and minor amounts of isotransferrin with two sialic acid residues (disialotransferrin) are detected. The reference range is 0 20 U/l for men and 0 26 U/l for women (Anton et al., 2001; Helander, 2002). Other methods Serum γ-glutamyl tansferase (GT), aspartate aminotransferase (AST), alanine aminotransferase (ALT), albumin, bilirubin and mean corpuscular volume (MCV) of erythrocytes were measured by standard clinical chemical methods in an accredited (SFS-EN 45001, ISO/IEC Guide 25) laboratory of EP Central Hospital, Seinäjoki, Finland. The reference ranges for these parameters were as follows. GT, men <80 U/l, women <50 U/l; AST, men <50 U/l, women <35 U/l; ALT, men <50 U/l, women <35 U/l; albumin g/l; bilirubin 2 20 µmol/l; MCV fl. Table 2. Correlation between the different marker values at the time of admission and the amount of alcohol used prior to sampling EtOH consumed in EtOH consumed in EtOH consumed in Marker the past 1 day the past 1 week the past 1 month %CDT * 0.59*** CDTect * GT * AST * 0.35* MCV ** 0.53** *P < 0.05, **P < 0.01, ***P < See Table 1 for abbreviations. Calculations and statistical methods The data are expressed as means ± SD. The comparisons between the alcoholic and non-alcoholic groups were carried out using Student s t-test or the Mann Whitney test for parameters with skewed distributions of values. Comparisons of the analytical characteristics between the different markers were based on ROC-assays using Analyse-it software (Analyse-it Software, Leeds, UK) Calculations of marker normalization rates were based on observed values and related to the starting value and to the time point of the last drink. Correlations were calculated using the Pearson product moment correlation coefficients. A P-value less than 0.05 was considered statistically significant. RESULTS At the time of admission, the markers of ethanol consumption in 36 alcoholic patients with a mean ethanol consumption of 132 ± 71 g/day and a mean duration of abstinence of 1.8 ± 2.0 days were all significantly higher than those of the control population (Table 1). When the controls were further divided into subgroups of social drinkers (n = 20) and abstainers (n = 10), the mean %CDT (2.0 ± 0.4, 2.0 ± 0.4), CDTect (12 ± 4, 17 ± 7), GT (27 ± 11, 23 ± 10) or AST (32 ± 24, 26 ± 8) values were not found to be significantly different between these subgroups, respectively. Elevated marker values in the alcoholics at the time of admission were observed with the following frequencies: %CDT 69%, CDTect 61%, GT 61%, AST 56% and MCV 47%. Except for MCV, the differences in the analytical characteristics between the above marker pairs were not statistically significant. Serum bilirubin and albumin did not differ significantly between the alcoholic and control populations, suggesting a lack of significant liver dysfunction in the present sample of alcoholic patients (Table 1). The amount of alcohol consumed, as estimated by a time-line follow-back method from different time intervals prior to sampling, correlated most significantly with the %CDT and MCV results (Table 2). A follow-up with supervised abstinence during hospitalization for 8 ± 4 (range 5 19) days was completed for 17 patients. At the initiation of this follow-up, %CDT was elevated in 77%, CDTect in 71%, GT in 47%, AST in 47% and MCV in 53% of the patients. All markers except for MCV and AST, decreased significantly during abstinence, as reflected both in the incidence of abnormal values (Fig. 1) or marker concentrations (Table 3). Interestingly, during the follow-up

3 CDT IN MONITORING ALCOHOL ABUSE 61 Table 3. Changes in marker values during follow-up of alcoholics with supervised abstinence Observed time for Previously reported time for Marker Admission After 8 ± 4 days P-value of the paired t-test normalization (days) normalization %CDT 5.2 ± ± ± 4 ND CDTect 29 ± ± ± weeks GT 222 ± ± 307 NS 23 ± weeks AST 62 ± ± 44 NS 15 ± weeks MCV 96 ± ± ND 2 4 months ND, not determined. Fig. 1. The proportions of elevated values in various laboratory markers of alcohol consumption in the alcoholics at the time of admission (black) and after a mean of 8 ± 4 days of supervised abstinence (grey). CDT, carbohydrate-deficient transferrin; GT, γ-glutamyl transferase; AST, aspartate aminotransferase; MCV, mean corpuscular volume. CDT concentrations in repeated sampling were found to show different rates to normalization with the two methods (14 ± 4 days for %CDT and 10 ± 5 days for the CDTect) (P < 0.05). At the end of the follow-up, %CDT values remained elevated in 65% of the cases, whereas CDTect remained elevated only in 47% of the patients. The corresponding percentages for serum GT, AST and MCV were 35, 47 and 59%, respectively. Among the individual patients, %CDT gave a lower value for all patients after abstinence whereas with CDTect, GT and AST occult marker elevations were also seen (Figs 2 and 3). DISCUSSION The present data indicate that the CDT assays, which have different reactivities towards the different desialylated isoforms of transferrin, show different analytical characteristics in the assessment and follow-up of alcohol misusers. Although CDT measurements have been widely employed in alcohol screening programmes, the methods have been heterogeneous and the data on their clinical sensitivities have remained controversial, with reports ranging from <20 to 100% (Stibler, 1991; Nyström et al., 1992; Allen et al., 1994; Bean et al., 1997; Salaspuro, 1999; Scouller et al., 2000; Arndt, 2001; Helander et al., 2001; Sillanaukee et al., 2001; Conigrave et al., 2002). Therefore, there is an apparent need for further work on methodological standardization and method comparisons. Fig. 2. Individual CDT values at the beginning and end of the follow-up period in the alcoholic patients, as assessed with the two different methods. The follow-up period was 8 ± 4 days. A, at admission; B, at end of follow-up period. Most of the current knowledge on CDT measurements has been based on the use of the conventional CDTect method. The more recently introduced %CDT measurements have offered the benefit of expressing the data conveniently as percentages of total transferrin with similar cut-offs for both sexes. However, the first versions of the %CDT assays suffered from poor analytical sensitivities (Viitala et al., 1998). More recent comparisons between the new %CDT and CDTect methods have shown essentially similar sensitivities (Anton et al., 2001). Accordingly, the new %CDT method in the present relatively small material appears to yield even a slightly higher (although not statistically significant) sensitivity (69%) than the conventional CDTect method (61%)

4 62 P. ANTTILA et al. Fig. 3. The proportions of alcohol marker values showing increases (A) or decreases (B) after the follow-up period. For abbreviations and details, see legend to Figure 1. or the other conventional markers of ethanol consumption (Fig. 1) when analysed at the time of admission. It may be emphasized that, in assays for alcohol misuse, which has an extremely high prevalence (typically 10 20% in many Western countries), even a small improvement in the analytical performance should lead to a more correct identification of a large number of hazardous drinkers, which, in turn, is of crucial importance in the clinical assessment of such patients. However, as the current material is relatively small future studies for testing marker function in larger materials appear warranted. Interestingly, current data also show that the correlation between the marker results and the actual amount of ethanol consumed is stronger with the %CDT results than with the CDTect results, indicating that the measurements (which exclude the trisialofraction but include the entire disialofraction into the measurement) may reflect ethanolinduced changes in transferrin protein in a more specific manner. This may also be due to the fact that the CDTect method may respond to unspecific changes in transferrin protein metabolism, iron balance, or hormonal status (Niemelä et al., 1995; Sorvajärvi et al., 1996; De Feo et al., 1999; DiMartini et al., 2001; Whitfield et al., 2001). Such phenomena may also explain the previously reported poor sensitivities for CDTect assays in female patients (Anton and Moak, 1994; Löf et al., 1994; Reif et al., 2001; Anton et al., 2002). Preliminary analyses of serum sialic acid contents from these patients have indicated no significant relationship between the assay-dependent characteristics and sialic acid metabolism (data not shown). In this study we included both abstainers and social drinkers in our control material. No significant differences in CDT values emerged between these subgroups, suggesting a lack of correlation between the consumption of low doses of alcohol and CDT values. In the alcoholics, the presence of severe liver disease has previously been found to be a significant confounding variable in the interpretation of CDT results (Anttila et al., 2003). In this study the possible effect of alcohol-related liver disease on the findings could not be assessed by the method-of-choice, because liver biopsies were not done. It should be noted, however, that based on clinical examination and laboratory findings indicating that serum bilirubin and albumin levels were not significantly different between the alcoholics and controls, the former group should have been devoid of any significant liver dysfunction. The present follow-up data show a consistent decrease of the %CDT marker values during abstinence, which is also in line with the view that the %CDT results show a close association between the marker data and drinking habits. The discrepancies between GT, CDTect, AST and abstinence perhaps could be due to induction of liver pathology, which may be reflected in marker values also during abstinence. In particular, the CDTect method may be sensitive to changes occurring during liver pathology even in its early phase (Tsutsumi et al., 1994; Niemelä et al., 1995). The present findings indicating that CDT methods may also have kinetic differences in their rates of normalization should also be considered in further method comparisons and in monitoring abstinence in alcoholic patients. Acknowledgements This study was supported in part by a grant from the Finnish Foundation for Alcohol Studies. REFERENCES Allen, J. P., Litten, R. Z., Anton, R. F. and Cross, G. M. (1994) Carbohydrate-deficient transferrin is a measure of immoderate drinking: remaining issues. Alcoholism: Clinical and Experimental Research 18, Anton, R. F. and Moak, D. H. (1994) Carbohydrate-deficient transferrin and γ-glutamyltransferase as markers of heavy alcohol consumption: gender differences. Alcoholism: Clinical and Experimental Research 18, Anton, R. F., Dominick, C., Bigelow, M. and Westby, C. (2001) Comparison of Bio-Rad %CDT TIA and CDTect as laboratory markers of heavy alcohol use and their relationships with γ- glutamyltransferase. Clinical Chemistry 47, Anton, R. F., Lieber, C., Tabakoff, B. and the CDTect Study Group (2002) Carbohydrate-deficient transferrin and gammaglutamyltransferase for the detection and monitoring of alcohol use: results from a multisite study. Alcoholism: Clinical and Experimental Research 26, Anttila, P., Järvi, K., Latvala, J., Blake, J. E. and Niemelä, O. (2003) Diagnostic characteristics of different carbohydrate-deficient transferrin methods in the detection of problem drinking: effects of liver disease and alcohol consumption. Alcohol and Alcoholism 38, Arndt, T. (2001) Carbohydrate-deficient transferrin as a marker of chronic alcohol abuse: a critical review of preanalysis, analysis, and interpretation. Clinical Chemistry 47, Bean, P., Liegmann, K., Lovli, T., Westby, C. and Sundrehagen, E. (1997) Semiautomated procedures for evaluation of carbohydratedeficient transferrin in the diagnosis of alcohol abuse. Clinical Chemistry 43,

5 CDT IN MONITORING ALCOHOL ABUSE 63 Conigrave, K. M, Degenhardt, L. J., Whitfield, J. B., Saunders, J. B., Helander, A. and Tabakoff, B. (2002) CDT, GGT, and AST as markers of alcohol use: the WHO/ISBRA collaborative project. Alcoholism: Clinical and Experimental Research 26, De Feo, T. M., Fargion, S., Duca, L., Mattioli, M., Cappellini, M. D., Sampietro, M., Cesana, B. M. and Fiorelli, G. (1999) Carbohydratedeficient transferrin, a sensitive marker of chronic alcohol abuse, is highly influenced by body iron. Hepatology 29, DiMartini, A., Day, N., Lane, T., Beisler, A. T., Dew, M. A. and Anton, R. (2001) Carbohydrate deficient transferrin in abstaining patients with end-stage liver disease. Alcoholism: Clinical and Experimental Research 25, Helander, A. (2002) Multicentre validation study of instrument applications for %CDT, an immunoassay for quantification of carbohydrate-deficient transferrin in serum. Alcohol and Alcoholism 37, Helander, A., Eriksson, G., Stibler, H. and Jeppsson, J. O. (2001) Interference of transferrin isoform types with carbohydrate-deficient transferrin quantification in the identification of alcohol abuse. Clinical Chemistry 47, Löf, K., Seppä, K., Itälä, L., Koivula, T., Turpeinen, U. and Sillanaukee, P. (1994) Carbohydrate-deficient transferrin (CDT) as an alcohol marker among female heavy drinkers: a population-based study. Alcoholism: Clinical and Experimental Research 18, Niemelä, O., Sorvajärvi, K., Blake, J. E. and Israel, Y. (1995) Carbohydrate-deficient transferrin as a marker of alcohol abuse: relationship to alcohol consumption, severity of liver disease, and fibrogenesis. Alcoholism: Clinical and Experimental Research 19, Nyström, M., Peräsalo, J. and Salaspuro, M. (1992) Carbohydratedeficient transferrin (CDT) in serum as a possible indicator of heavy drinking in young university students. Alcoholism: Clinical and Experimental Research 16, Reif, A., Keller, H., Schneider, M., Kamolz, S., Schmidtke, A. and Fallgatter, A. J. (2001) Carbohydrate-deficient transferrin is elevated in catabolic female patients. Alcohol and Alcoholism 36, Salaspuro, M. (1999) Carbohydrate-deficient transferrin as compared to other markers of alcoholism: a systematic review. Alcohol 19, Scouller, K., Conigrave, K. M., Macaskill, P., Irwing, L. and Whitfield, J. B. (2000) Should we use carbohydrate-deficient transferrin instead of gammaglutamyltransferase for detecting problem drinkers? A systematic review and metaanalysis. Clinical Chemistry 46, Sillanaukee, P., Strid, N., Allen, J. P. and Litten, R. Z. (2001) Possible reasons why heavy drinking increases carbohydrate-deficient transferrin. Alcoholism: Clinical and Experimental Research 25, Sorvajärvi, K., Blake, J. E., Israel, Y. and Niemelä, O. (1996) Sensitivity and specificity of carbohydrate-deficient transferrin as a marker of alcohol abuse are significantly influenced by alterations in serum transferrin: comparison of two methods. Alcoholism: Clinical and Experimental Research 20, Stibler, H. (1991) Carbohydrate-deficient transferrin in serum: a new marker of potentially harmful alcohol consumption reviewed. Clinical Chemistry 37, Tagliaro, F., Bortolotti, F., Dorizzi, R. M. and Marigo, M. (2002) Caveates in carbohydrate-deficient transferrin determination. Clinical Chemistry 48, Tsutsumi, M., Wang, J. S. and Takada, A. (1994) Microheterogeneity of serum glycoproteins in alcoholics: is desialo-transferrin the marker of chronic alcohol drinking or alcoholic liver injury? Alcoholism: Clinical and Experimental Research 18, Viitala, K., Lähdesmäki, K. and Niemelä, O. (1998) Comparison of the Axis %CDT TIA and the CDTect method as laboratory tests of alcohol abuse. Clinical Chemistry 44, Whitfield, J. B., Zhu, G., Heath, A. C., Powell, L. W. and Martin, N. G. (2001) Effects of alcohol consumption on indices of iron stores and of iron stores on alcohol intake markers. Alcoholism: Clinical and Experimental Research 25,