- Email: [email protected]
Analytical evaluation of a new capillary electrophoresis method for carbohydrate-deficient transferrin measurement
Clinica Chimica Acta 382 (2007) 48 – 53 www.elsevier.com/locate/clinchim
Analytical evaluation of a new capillary electrophoresis method for carbohydrate-deficient transferrin measurement François Schellenberg a,⁎, Catherine Girre b , Bertrand Nalpas c , Arnaud Plat b , Antonio Tome b , Jean Christophe Pagès a a
CHRU de Tours, Pôle de Biologie, Laboratoire de Biochimie, Hôpital Trousseau, France b Addictology Unit, Hôpital Fernand Widal, Paris, France c INSERM U567, Hôpital Necker, Paris, France
Received 6 February 2007; received in revised form 16 March 2007; accepted 17 March 2007 Available online 28 March 2007
Abstract Background: Carbohydrate-deficient transferrin (CDT), the sum of a- and disialotransferrin, is considered the most efficient routine biological marker of alcohol abuse. In recent years, methods based on capillary zone electrophoresis (CZE) have been developed using specialized monocapillary systems. These are characterized by a high analytical detection level, counterbalanced by a poor productivity. We evaluated a new CZE method for CDT measurement on the Sebia Capillarys®, an eight-capillary system developed for routine serum capillary electrophoresis. Methods: Precision and possible biases due to abnormal (low or high) transferrin levels or lipemic samples were assessed. Exactitude and precision were tested by comparison with a HPLC procedure acknowledged to be the most reliable to date. The validity of the manufacturer's cutoff was checked by measuring CDT in a population comprising abstaining patients, moderate alcohol consumers and alcohol abusers. Lastly, the method was compared to the routine %CDT TIA and N Latex CDT methods. Results: The imprecision was 18.5% at the minimum detection level and decreased to 6.1% for high CDT values. No significant shift in the CDT results was observed in relation to abnormally low or high serum transferrin, or in lipemic samples. A high level of concordance was observed with the HPLC method used as reference. The results were strongly correlated with both other routine methods (r N 0.90). The diagnostic values were comparable to the literature data, even if differences in the studied populations make difficult a direct comparison of the results. Our data suggested that the cut-off could be raised from 1.3% to 1.4% to reduce the number of false positive values without loss of diagnostic efficiency. Conclusions: This Capillarys® method from Sebia showed good precision as compared to those published using other CZE methods. Capillarys® method correlated well with HPLC and two routine methods. However, we noticed significant bias at low CDT concentrations. Therefore, with the advantage of high throughput and full automation, these results indicate that the new method is a consistent alternative to the other methods proposed for routine CDT measurement. © 2007 Elsevier B.V. All rights reserved. Keywords: CDT; Alcohol; Capillary electrophoresis
1. Introduction
Abbreviations: AUD, Alcohol Use Disorders; AUDIT, Alcohol Use Disorders Identification Test; BAC, Blood Alcohol Concentration; CDT, Carbohydrate-Deficient Transferrin; CZE, Capillary Zone Electrophoresis; HPLC, High-Performance Liquid Chromatography; IEC, Ion Exchange Chromatography; IEF, Isoelectric Focusing; IF, Immunofixation; ROC, Receiver Operating Characteristics. ⁎ Corresponding author. Laboratoire de Biochimie, Hôpital Trousseau, CHRU de Tours, 37044 Tours, France. Tel.: +33 2 47 47 46 84; fax: +33 2 47 47 46 88. E-mail address: [email protected] (F. Schellenberg). 0009-8981/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.cca.2007.03.020
Carbohydrate-deficient transferrin (CDT) is made up of the transferrin glycoforms with two or less sialic acid residues on the polypeptide chain. CDT appears as the result of an inefficient glycosylation process resulting from the toxic effect of alcohol metabolism within hepatocytes. Other clinical conditions might result in an abnormal transferrin glycosylation profile, including rare heritable deficiency in the glycosylation enzymes and some polymorphism within the transferrin gene. It is now widely recognized as the most valuable marker of alcohol abuse that can
F. Schellenberg et al. / Clinica Chimica Acta 382 (2007) 48–53
be used for routine purposes [1]. It is used in the detection and follow-up of patients suffering from alcohol use disorders (AUD), and also as a marker of temperance in driving license restitution after suspension due to positive blood alcohol levels. Laborious and time-consuming methods have hindered CDT in its development as a routine use marker [2]. In recent years, capillary zone electrophoresis (CZE) based methods have been developed [3], and a fully automated technique based on this analytical principle was proposed by SEBIA (Evry, France). In this paper, we report the analytical evaluation of this new method and correlation with results provided by other well known procedures. 2. Materials and methods 2.1. Analytical methods 2.1.1. Capillary electrophoresis Capillarys® is a capillary zone electrophoresis system for serum analysis at high voltage (8200 V) in alkaline buffer (pH 8.8). The separation of charged molecules is based on their electrophoretic mobility according to pH and electroosmotic flow. Transferrin glycoforms are directly detected at 200 nm. The system can run seven samples simultaneously. Total running time is approximately 10 min. In this study, we used the 5.4.1 revision of the software, with the recently developed algorithms for peak integration, resulting in an improved separation between di- and trisialotransferrin, and a separate quantification of pentasialotransferrin. The measurements were done according to the manufacturer's instructions; samples are first automatically diluted with an iron saturation solution by the Capillarys®, then aspirated and submitted to electrophoresis. The peaks corresponding to penta-, tetra-, tri-, di- and asialotransferrin are quantified. CDT quantification includes di- and asialotransferrin. 2.1.2. Other CDT determinations CDT results obtained by CZE were confronted to those obtained by two other routine methods: The %CDT TIA (Bio-Rad, Hercules, CA, USA) method, an ion-exchange separation of transferrin glycoforms on disposable microcolumns, measures CDT by determination of di- and asialylated forms in the eluates by nephelometry or turbidimetry. In this study, total transferrin and CDT measurements were run using a Immage® 800 nephelometric system (Beckman Coulter, Brea, CA, USA). %CDT was calculated as the ratio of these two measurements. The N Latex CDT (Dade Behring, Marburg, Germany) assay is a direct method designed for the BN® systems, based on the direct nephelometric measurement of di- and asialotransferrin by specific antibodies. We used in this study a BN Prospec ® (Dade Behring, Marburg, Germany) nephelometric system for N Latex CDT and for total transferrin used in the calculation of %CDT.
2.2. Study design Precision of the technique was determined by running samples in duplicate for five consecutive days over 5 weeks. This was applied to two pools with low (0.6%) and high (4.8%) CDT values. We also evaluated CDT variations in nonalcoholic patients with extreme values of serum transferrin, ranging from 0.5 to 4.5 g/l. The interference of excess lipids was tested by analyzing 30 lipemic samples before and after delipidation by 30 min ultracentrifugation. In the absence of available standard material for CDT, the accuracy and the linearity of the CZE method were measured by a comparative analysis of CDT using the % CDT by HPLC method on a Variant™ analyzer (Bio-Rad, Hercules, CA, USA). This comparison was performed with 100 samples covering a wide range (0.69% to 21.36%) of CDT concentrations. A case-control study was carried out to verify the validity of the manufacturer's cut-off. One-hundred forty-seven (147) control subjects were included in this study. They were 80 subjects recruited in a clinical study with AUDIT score b7 and a daily alcohol intake b40 g, and 67 hospital patients without any history of alcohol, no suspicion of alcohol abuse and no biological sign (excluding CDT) of alcohol abuse. Two-hundred ten alcoholic (210) patients
49
were included with AUDIT score N10, corresponding to daily alcohol N50 g, or BAC N 0. The participation of these patients and controls in a clinical study was agreed by the ethical committee of Hôpital Fernand Widal (Paris, France). The CZE procedure was compared with %CDT TIA and N Latex CDT procedures in these samples. The statistical calculations were done using MedCalc 9.2 software (MedCalc Software, Mariakerke, Belgium). Pearson's correlation coefficient for non Gaussian distribution was used in the correlation studies.
3. Results and discussion CDT was discovered using isoelectric focusing/immunofixation (IEF/IF) [4]. This method was laborious and time consuming, although it allowed the quantification of CDT [5]. Later, methods were introduced based on separation of the transferrin glycoforms by ion exchange chromatography (IEC) followed by CDT measurement by radioimmunoassay [6], nephelometry [7], or turbidimetry [8]. The two latter semiautomated procedures allowed routine use of CDT in almost any laboratory. In the last 10 years, a number of procedures based on CZE methodology [9–13] and high-performance liquid chromatography (HPLC) [14] have been developed. CZE and HPLC differ from IEC in that they produce a graphic representation of all of the transferrin glycoforms. Through analysis of graphic's profile, the risk of false positive results due to genetic D variants is suppressed [15]. Up to now the productivity of the CZE methods was low, as they were developed on monocapillary systems. Recently, a fully automated method was developed on the eight-capillary Capillarys® system, which can run up to 40 determinations per hour. This system is presented by the manufacturer as having the same analytical features as specialized monocapillary systems, while providing the advantages of higher productivity and total automation. The analytical features of this method have been thoroughly studied in a recent study [16], which demonstrated the good analytical performances of the system, compared to single capillary electrophoresis. 3.1. Precision The daily (5 days/week) duplicate measurement of two sample pools over a 5-week period showed no drift toward lower or higher values during the study. The values of the low CDT pool were close to the detection limit of the system (m = 0.7%). The between-run precision was 18.5%. The mean value for the high CDT pool was 4.8% at a precision of 6.1%. At this CDT level, the asialotransferrin peak was small, and its area could not be evaluated in some runs. The precision of the high CDT pool improved to 4.2% (mean CDT value 4.4%) when excluding asialotransferrin from the CDT calculation and considering solely the disialotransferrin fraction. These results are comparable to those from previous studies [9,10,17], while better precision was obtained in other studies [18,19]. 3.2. Total transferrin Forty-eight samples from non-AUD subjects with serum transferrin ranging from 0.55 to 4.43 g/l were analysed. CDT
50
F. Schellenberg et al. / Clinica Chimica Acta 382 (2007) 48–53
Fig. 1. Bland–Altman plot of the difference between CDT measured by CZE Capillarys® method and Bio-Rad %CDT by HPLC.
values were distributed from 0.2% to 0.9%, and Passing Bablock regression analysis between CDT and total transferrin did not indicate any significant slope (%CDT = 0.046 (95%CI 0.000 to 0.103) ⁎ Trf + 0.359). In addition, no correlation was found between CDT and total transferrin.
(Pearson's correlation coefficient = 0.220, p = 0.134). So, despite the lack of precision of low CDT values demonstrated by the precision study, the low signal had no effect on the results obtained with low serum transferrin concentration.
Fig. 2. Scattergram of the measurement of CDT in 295 patients by the Capillarys® CZE method compared to the %CDT TIA and N Latex CDT methods.
F. Schellenberg et al. / Clinica Chimica Acta 382 (2007) 48–53
3.3. Lipemic samples Thirty lipemic samples were run before and after delipidation by ultracentrifugation (30 min at 45,000 g). CDT values were ranging from 0.3% to 9.1%. Mean values were, respectively, 1.26% (before) and 1.30% (after). The difference was not significant (Wilcoxon test), demonstrating the absence of influence of lipemia on the CZE measurement of CDT.
51
explanation of these discrepancies could be found by comparing the CZE and HPLC diagrams. Both methods were highly correlated (Pearson's correlation coefficient = 0.989). The equation of the Passing and Bablock regression was CZE = 0.9677 ⁎ HPLC − 0.29. The Cusum test for linearity was not significant (p N 0.10), indicating that both techniques were correlated following a linear mode. 3.5. Correlation with other routine methods
3.4. Accuracy and linearity Compared to other molecules, the absence of purified CDT standards makes difficult the evaluation of the analytical characteristics of the different methods. Consequently, new procedures have to be evaluated by comparison with an HPLC procedure, one of them being proposed as a reference method [20]. Here, we compared the CZE Capillarys® method to the Bio-Rad %CDT by HPLC, which has been demonstrated to be highly correlated to the candidate reference method [21]. Hundred samples from the laboratory were first measured with the Variant™ system, covering a CDT range from 0.69% to 21.36%, and then with the CZE method. The Bland and Altman plot is shown in Fig. 1. The mean difference for CZE method, with respect to the HPLC method, showed a negative bias of − 0.5%. When considering 1.96 SD as the limit for acceptance, the accepted positive difference was +0.8%, and the negative difference was − 1.9%. 3 values fell outside this limit, no
Three-hundred ten samples (310) were analyzed to establish a correlation with the %CDT TIA and N Latex CDT [22] methods. Ten samples were excluded because the Capillarys® was not able to identify and quantify the disialotransferrin peak (the detection limit is approximately 0.2%). Five other patients were genetic variants for transferrin (1 CD and 4 BC), and thus were rejected. The discrepant results were reanalyzed by the three methods and no difference was observed. Thus, the results of 295 subjects were compared (Fig. 2). The Capillarys® method was highly correlated with the IEC method (r = 0.953 with %CDT TIA) and with the direct nephelometric method (r = 0.949 with N Latex CDT). The higher CDT values observed with the %CDT TIA® method in patients with high trisialotransferrin (N 6%), resulted from partial co-elution of tri- and disialotransferrin [23]. The high number of results in this study may be the reason for the higher correlation with %CDT TIA, as compared to other studies [12]. However, as shown in Fig. 2, the
Fig. 3. Distribution of CDT values measured by the CZE Capillarys® method in 139 control subjects and 203 alcohol abusers.
52
F. Schellenberg et al. / Clinica Chimica Acta 382 (2007) 48–53
correlation was weak for subjects with CDT values below the cut-off. This heterogeneity might be the consequence of both imprecision in low values and methodological differences. 3.6. Case-control study Among the 357 subjects included, 147 controls subjects and 210 alcohol abusers, the Capillarys® CZE method was not able to measure CDT in 15 patients, 7 with a disialotransferrin peak lower than the integration threshold, 4 with high trisialotransferrin peak associated to low disialotransferrin, 3 genetic variants and 1 with a monoclonal peak. Thus results from 342 patients were included. They were 203 alcohol abusers (24 female, 179 male, aged 25–71 years, mean 47 ± 9 years), and 139 control subjects (26 female, 113 male, aged 18–75 years, mean 48 ± 7 years). Mean CDT was 0.9% (±0.4%) in the control group and 4.5% (± 5.0%) in the alcohol abuser group. The D'Agostino–Pearson test for normal distribution was performed to check the distribution of CDT values in this group. The test indicated that this distribution was outside normality acceptance limits (p = 0.002). As shown in Fig. 3, there is an overlap of the results in both groups for CDT values ranging from 0.4% to 2.1%. In this sample, the distribution of CDT values in the AUD group appears bimodal, as if this population was constituted of a group of “non-responder” subjects with a distribution of CDT values apparently similar to that of the control subjects, and a group of “responder” with high CDT values. The area under the receiver operating characteristics (ROC) curve was 0.89 (C.I. 0.85–0.92), as compared with values ranging from 0.66 [24] to 0.91 [25] in previous studies based on other CZE methods. When applying the cut-off proposed by the manufacturer (1.3%), 16 control patients had abnormally high values (Sp = 0.88). In the alcohol abuser group, the number of false negative values was 52 (Se = 0.74). Considering the lower prevalence of AUD in the general population compared to the study sample, and the possible consequences of falsely positively classifying patients, we considered that specificity should have a greater importance than sensitivity in the interpretation of the CDT results. So we calculated the diagnostic values using 1.4 as cut-off. This threshold increased the specificity to 0.91, while the sensitivity decreased to 0.71. This cut-off should be confirmed in other studies to avoid bias due to subject recruitment. Such bias may partly explain the discrepancies in the diagnostic values observed in other studies, with sensitivities ranging from 0.56 [26] to 0.88 [11] and specificities from 0.68 [24] to 0.99 [11]. Compared to specialized monocapillary instruments, the Capillarys® system allows the measurement of CDT with comparable precision, it is an automated process including sample pre-treatment, and has the high productivity of an eightcapillary system. Acknowledgements The authors are grateful to Mrs. Annie Détruit, Nadine Puche, and Geneviève Wattelet for their technical assistance. The
authors thank Sebia, Roche and Bio-Rad for providing the reagents used in this study. They thank Roche and Bio-Rad for the lending of the BN Prospec® and Variant™ systems. The case-control study was partly supported by INSERM grant “ATC Alcool” #ALC0205. References [1] Salaspuro M. Carbohydrate-deficient transferrin as compared to other markers of alcoholism: a systematic review. Alcohol 1999;19:261–71. [2] Bortolotti F, De Paoli G, Tagliaro F. Carbohydrate-deficient transferrin (CDT) as a marker of alcohol abuse: a critical review. J Chromatogr B 2006;841:96–109. [3] Wuyts B, Delanghe JR. The analysis of carbohydrate-deficient transferrin, marker of chronic alcoholism, using capillary electrophoresis. Clin Chem 2003;41:739–45. [4] Stibler H, Allgulander C, Borg S, et al. Abnormal microheterogeneity of transferrin in serum and cerebrospinal fluid in alcoholism. Acta Med Scand 1978;204:49–56. [5] Schellenberg F, Weill J. Serum desialotransferrin in the detection of alcohol abuse. Definition of a tf index. Drug Alcohol Depend 1987;19: 181–91. [6] Stibler H, Borg S, Joustra M. Micro anion exchange chromatography of carbohydrate-deficient transferrin in serum in relation to alcohol consumption (Swedish patent 8400587-5). Alcohol Clin Exp Res 1986;10:535–44. [7] Schellenberg F, Martin M, Caces E, et al. Nephelometric determination of carbohydrate deficient transferrin. Clin Chem 1996;42:551–7. [8] Viitala K, Lähdesmäki K, Niemelä O. Comparison of the Axis %CDT TIA and the CDTect method as laboratory tests of alcohol abuse. Clin Chem 1998;44:1209–15. [9] Wuyts B, Delanghe JR, Kasvosve I, et al. Determination of carbohydratedeficient transferrin using capillary zone electrophoresis. Clin Chem 2001;47:247–55. [10] Legros F, Nuyens V, Minet E, et al. Carbohydrate-deficient transferrin isoforms measured by capillary zone electrophoresis for detection of alcohol abuse. Clin Chem 2002;48:2177–86. [11] Prasad R, Stout RL, Coffin D, et al. Analysis of carbohydrate deficient transferrin by capillary electrophoresis. Electrophoresis 1997;18:1814–8. [12] Fermo I, Germagnoli L, Soldarini A, et al. Capillary zone electrophoresis for determination of carbohydrate-deficient transferrin in human serum. Electrophoresis 2004;25:469–75. [13] Lanz C, Kuhn M, Bortolotti F, et al. Evaluation and optimization of capillary zone electrophoresis with different dynamic capillary coatings for the determination of carbohydrate-deficient transferrin in human serum. J Chromatogr A 2002;979:43–57. [14] Jeppsson JO, Kristensson H, Fimiani C. Carbohydrate-deficient transferrin quantified by HPLC to determine heavy consumption of alcohol. Clin Chem 1993;39:2115–20. [15] Helander A, Eriksson G, Stibler H, et al. Interference of transferrin isoform types with carbohydrate-deficient transferrin quantification in the identification of alcohol abuse. Clin Chem 2001;47:1225–33. [16] Bortolotti F., De Paoli G., Pascali J.P., et al. Fully automated analysis of Carbohydrate-Deficient Transferrin (CDT) by using a multicapillary electrophoresis system. Clin Chim Acta in press, doi:10.1016/j.cca.2006.12.028. [17] Crivellente F, Fracasso G, Valentini R, et al. Improved method for carbohydrate-deficient transferrin determination in human serum by capillary zone electrophoresis. J Chromatogr B Biomed Sci Appl 2000;739:81–93. [18] Lanz C, Kuhn M, Deiss V, et al. Improved capillary electrophoresis method for the determination of carbohydrate-deficient transferrin in patient sera. Electrophoresis 2004;25:2309–18. [19] Joneli J, Lanz C, Thormann W. Capillary zone electrophoresis determination of carbohydrate-deficient transferrin using the new CEofix reagents under high-resolution conditions. J Chromatogr A 2006;1130: 272–80.
F. Schellenberg et al. / Clinica Chimica Acta 382 (2007) 48–53 [20] Helander A, Husa A, Jeppsson JO. Improved high-performance liquid chromatography method for carbohydrate-deficient transferrin in serum. Clin Chem 2003;49:1881–90. [21] Helander A, Bergström JP. Determination of carbohydrate-deficient transferrin in human serum using the Bio-Rad %CDT by HPLC test. Clin Chim Acta 2006;371:187–90. [22] Delanghe JR, Helander A, Wielders JPM, Pekelharing JM, Roth HJ, Schellenberg F, Born C, Yagmur E, Gentzer W, Althaus H. Clin Chem in press, doi:10.1373/clinchem.2006.08.4459. [23] Alden A, Ohlson S, Pahlsson P, et al. HPLC analysis of carbohydrate deficient transferrin isoforms isolated by the Axis-Shield %CDT method. Clin Chim Acta 2005;356:143–6.
53
[24] Daeppen JB, Anex F, Favrat B, et al. Carbohydrate-deficient transferrin measured by capillary zone electrophoresis and by turbidimetric immunoassay for identification of young heavy drinkers. Clin Chem 2005;51:1046–8. [25] Legros FJ, Nuyens V, Baudoux M, et al. Use of capillary zone electrophoresis for differentiating excessive from moderate alcohol consumption. Clin Chem 2003;49:440–9. [26] Wuyts B, Delanghe JR, Kasvosve I, et al. Carbohydrate-deficient transferrin and chronic alcohol ingestion in subjects with transferrin CDvariants. Clin Chem Lab Med 2001;39:937–43.
Analytical evaluation of a new capillary electrophoresis method for carbohydrate-deficient transferrin measurement François Schellenberg a,⁎, Catherine Girre b , Bertrand Nalpas c , Arnaud Plat b , Antonio Tome b , Jean Christophe Pagès a a
CHRU de Tours, Pôle de Biologie, Laboratoire de Biochimie, Hôpital Trousseau, France b Addictology Unit, Hôpital Fernand Widal, Paris, France c INSERM U567, Hôpital Necker, Paris, France
Received 6 February 2007; received in revised form 16 March 2007; accepted 17 March 2007 Available online 28 March 2007
Abstract Background: Carbohydrate-deficient transferrin (CDT), the sum of a- and disialotransferrin, is considered the most efficient routine biological marker of alcohol abuse. In recent years, methods based on capillary zone electrophoresis (CZE) have been developed using specialized monocapillary systems. These are characterized by a high analytical detection level, counterbalanced by a poor productivity. We evaluated a new CZE method for CDT measurement on the Sebia Capillarys®, an eight-capillary system developed for routine serum capillary electrophoresis. Methods: Precision and possible biases due to abnormal (low or high) transferrin levels or lipemic samples were assessed. Exactitude and precision were tested by comparison with a HPLC procedure acknowledged to be the most reliable to date. The validity of the manufacturer's cutoff was checked by measuring CDT in a population comprising abstaining patients, moderate alcohol consumers and alcohol abusers. Lastly, the method was compared to the routine %CDT TIA and N Latex CDT methods. Results: The imprecision was 18.5% at the minimum detection level and decreased to 6.1% for high CDT values. No significant shift in the CDT results was observed in relation to abnormally low or high serum transferrin, or in lipemic samples. A high level of concordance was observed with the HPLC method used as reference. The results were strongly correlated with both other routine methods (r N 0.90). The diagnostic values were comparable to the literature data, even if differences in the studied populations make difficult a direct comparison of the results. Our data suggested that the cut-off could be raised from 1.3% to 1.4% to reduce the number of false positive values without loss of diagnostic efficiency. Conclusions: This Capillarys® method from Sebia showed good precision as compared to those published using other CZE methods. Capillarys® method correlated well with HPLC and two routine methods. However, we noticed significant bias at low CDT concentrations. Therefore, with the advantage of high throughput and full automation, these results indicate that the new method is a consistent alternative to the other methods proposed for routine CDT measurement. © 2007 Elsevier B.V. All rights reserved. Keywords: CDT; Alcohol; Capillary electrophoresis
1. Introduction
Abbreviations: AUD, Alcohol Use Disorders; AUDIT, Alcohol Use Disorders Identification Test; BAC, Blood Alcohol Concentration; CDT, Carbohydrate-Deficient Transferrin; CZE, Capillary Zone Electrophoresis; HPLC, High-Performance Liquid Chromatography; IEC, Ion Exchange Chromatography; IEF, Isoelectric Focusing; IF, Immunofixation; ROC, Receiver Operating Characteristics. ⁎ Corresponding author. Laboratoire de Biochimie, Hôpital Trousseau, CHRU de Tours, 37044 Tours, France. Tel.: +33 2 47 47 46 84; fax: +33 2 47 47 46 88. E-mail address: [email protected] (F. Schellenberg). 0009-8981/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.cca.2007.03.020
Carbohydrate-deficient transferrin (CDT) is made up of the transferrin glycoforms with two or less sialic acid residues on the polypeptide chain. CDT appears as the result of an inefficient glycosylation process resulting from the toxic effect of alcohol metabolism within hepatocytes. Other clinical conditions might result in an abnormal transferrin glycosylation profile, including rare heritable deficiency in the glycosylation enzymes and some polymorphism within the transferrin gene. It is now widely recognized as the most valuable marker of alcohol abuse that can
F. Schellenberg et al. / Clinica Chimica Acta 382 (2007) 48–53
be used for routine purposes [1]. It is used in the detection and follow-up of patients suffering from alcohol use disorders (AUD), and also as a marker of temperance in driving license restitution after suspension due to positive blood alcohol levels. Laborious and time-consuming methods have hindered CDT in its development as a routine use marker [2]. In recent years, capillary zone electrophoresis (CZE) based methods have been developed [3], and a fully automated technique based on this analytical principle was proposed by SEBIA (Evry, France). In this paper, we report the analytical evaluation of this new method and correlation with results provided by other well known procedures. 2. Materials and methods 2.1. Analytical methods 2.1.1. Capillary electrophoresis Capillarys® is a capillary zone electrophoresis system for serum analysis at high voltage (8200 V) in alkaline buffer (pH 8.8). The separation of charged molecules is based on their electrophoretic mobility according to pH and electroosmotic flow. Transferrin glycoforms are directly detected at 200 nm. The system can run seven samples simultaneously. Total running time is approximately 10 min. In this study, we used the 5.4.1 revision of the software, with the recently developed algorithms for peak integration, resulting in an improved separation between di- and trisialotransferrin, and a separate quantification of pentasialotransferrin. The measurements were done according to the manufacturer's instructions; samples are first automatically diluted with an iron saturation solution by the Capillarys®, then aspirated and submitted to electrophoresis. The peaks corresponding to penta-, tetra-, tri-, di- and asialotransferrin are quantified. CDT quantification includes di- and asialotransferrin. 2.1.2. Other CDT determinations CDT results obtained by CZE were confronted to those obtained by two other routine methods: The %CDT TIA (Bio-Rad, Hercules, CA, USA) method, an ion-exchange separation of transferrin glycoforms on disposable microcolumns, measures CDT by determination of di- and asialylated forms in the eluates by nephelometry or turbidimetry. In this study, total transferrin and CDT measurements were run using a Immage® 800 nephelometric system (Beckman Coulter, Brea, CA, USA). %CDT was calculated as the ratio of these two measurements. The N Latex CDT (Dade Behring, Marburg, Germany) assay is a direct method designed for the BN® systems, based on the direct nephelometric measurement of di- and asialotransferrin by specific antibodies. We used in this study a BN Prospec ® (Dade Behring, Marburg, Germany) nephelometric system for N Latex CDT and for total transferrin used in the calculation of %CDT.
2.2. Study design Precision of the technique was determined by running samples in duplicate for five consecutive days over 5 weeks. This was applied to two pools with low (0.6%) and high (4.8%) CDT values. We also evaluated CDT variations in nonalcoholic patients with extreme values of serum transferrin, ranging from 0.5 to 4.5 g/l. The interference of excess lipids was tested by analyzing 30 lipemic samples before and after delipidation by 30 min ultracentrifugation. In the absence of available standard material for CDT, the accuracy and the linearity of the CZE method were measured by a comparative analysis of CDT using the % CDT by HPLC method on a Variant™ analyzer (Bio-Rad, Hercules, CA, USA). This comparison was performed with 100 samples covering a wide range (0.69% to 21.36%) of CDT concentrations. A case-control study was carried out to verify the validity of the manufacturer's cut-off. One-hundred forty-seven (147) control subjects were included in this study. They were 80 subjects recruited in a clinical study with AUDIT score b7 and a daily alcohol intake b40 g, and 67 hospital patients without any history of alcohol, no suspicion of alcohol abuse and no biological sign (excluding CDT) of alcohol abuse. Two-hundred ten alcoholic (210) patients
49
were included with AUDIT score N10, corresponding to daily alcohol N50 g, or BAC N 0. The participation of these patients and controls in a clinical study was agreed by the ethical committee of Hôpital Fernand Widal (Paris, France). The CZE procedure was compared with %CDT TIA and N Latex CDT procedures in these samples. The statistical calculations were done using MedCalc 9.2 software (MedCalc Software, Mariakerke, Belgium). Pearson's correlation coefficient for non Gaussian distribution was used in the correlation studies.
3. Results and discussion CDT was discovered using isoelectric focusing/immunofixation (IEF/IF) [4]. This method was laborious and time consuming, although it allowed the quantification of CDT [5]. Later, methods were introduced based on separation of the transferrin glycoforms by ion exchange chromatography (IEC) followed by CDT measurement by radioimmunoassay [6], nephelometry [7], or turbidimetry [8]. The two latter semiautomated procedures allowed routine use of CDT in almost any laboratory. In the last 10 years, a number of procedures based on CZE methodology [9–13] and high-performance liquid chromatography (HPLC) [14] have been developed. CZE and HPLC differ from IEC in that they produce a graphic representation of all of the transferrin glycoforms. Through analysis of graphic's profile, the risk of false positive results due to genetic D variants is suppressed [15]. Up to now the productivity of the CZE methods was low, as they were developed on monocapillary systems. Recently, a fully automated method was developed on the eight-capillary Capillarys® system, which can run up to 40 determinations per hour. This system is presented by the manufacturer as having the same analytical features as specialized monocapillary systems, while providing the advantages of higher productivity and total automation. The analytical features of this method have been thoroughly studied in a recent study [16], which demonstrated the good analytical performances of the system, compared to single capillary electrophoresis. 3.1. Precision The daily (5 days/week) duplicate measurement of two sample pools over a 5-week period showed no drift toward lower or higher values during the study. The values of the low CDT pool were close to the detection limit of the system (m = 0.7%). The between-run precision was 18.5%. The mean value for the high CDT pool was 4.8% at a precision of 6.1%. At this CDT level, the asialotransferrin peak was small, and its area could not be evaluated in some runs. The precision of the high CDT pool improved to 4.2% (mean CDT value 4.4%) when excluding asialotransferrin from the CDT calculation and considering solely the disialotransferrin fraction. These results are comparable to those from previous studies [9,10,17], while better precision was obtained in other studies [18,19]. 3.2. Total transferrin Forty-eight samples from non-AUD subjects with serum transferrin ranging from 0.55 to 4.43 g/l were analysed. CDT
50
F. Schellenberg et al. / Clinica Chimica Acta 382 (2007) 48–53
Fig. 1. Bland–Altman plot of the difference between CDT measured by CZE Capillarys® method and Bio-Rad %CDT by HPLC.
values were distributed from 0.2% to 0.9%, and Passing Bablock regression analysis between CDT and total transferrin did not indicate any significant slope (%CDT = 0.046 (95%CI 0.000 to 0.103) ⁎ Trf + 0.359). In addition, no correlation was found between CDT and total transferrin.
(Pearson's correlation coefficient = 0.220, p = 0.134). So, despite the lack of precision of low CDT values demonstrated by the precision study, the low signal had no effect on the results obtained with low serum transferrin concentration.
Fig. 2. Scattergram of the measurement of CDT in 295 patients by the Capillarys® CZE method compared to the %CDT TIA and N Latex CDT methods.
F. Schellenberg et al. / Clinica Chimica Acta 382 (2007) 48–53
3.3. Lipemic samples Thirty lipemic samples were run before and after delipidation by ultracentrifugation (30 min at 45,000 g). CDT values were ranging from 0.3% to 9.1%. Mean values were, respectively, 1.26% (before) and 1.30% (after). The difference was not significant (Wilcoxon test), demonstrating the absence of influence of lipemia on the CZE measurement of CDT.
51
explanation of these discrepancies could be found by comparing the CZE and HPLC diagrams. Both methods were highly correlated (Pearson's correlation coefficient = 0.989). The equation of the Passing and Bablock regression was CZE = 0.9677 ⁎ HPLC − 0.29. The Cusum test for linearity was not significant (p N 0.10), indicating that both techniques were correlated following a linear mode. 3.5. Correlation with other routine methods
3.4. Accuracy and linearity Compared to other molecules, the absence of purified CDT standards makes difficult the evaluation of the analytical characteristics of the different methods. Consequently, new procedures have to be evaluated by comparison with an HPLC procedure, one of them being proposed as a reference method [20]. Here, we compared the CZE Capillarys® method to the Bio-Rad %CDT by HPLC, which has been demonstrated to be highly correlated to the candidate reference method [21]. Hundred samples from the laboratory were first measured with the Variant™ system, covering a CDT range from 0.69% to 21.36%, and then with the CZE method. The Bland and Altman plot is shown in Fig. 1. The mean difference for CZE method, with respect to the HPLC method, showed a negative bias of − 0.5%. When considering 1.96 SD as the limit for acceptance, the accepted positive difference was +0.8%, and the negative difference was − 1.9%. 3 values fell outside this limit, no
Three-hundred ten samples (310) were analyzed to establish a correlation with the %CDT TIA and N Latex CDT [22] methods. Ten samples were excluded because the Capillarys® was not able to identify and quantify the disialotransferrin peak (the detection limit is approximately 0.2%). Five other patients were genetic variants for transferrin (1 CD and 4 BC), and thus were rejected. The discrepant results were reanalyzed by the three methods and no difference was observed. Thus, the results of 295 subjects were compared (Fig. 2). The Capillarys® method was highly correlated with the IEC method (r = 0.953 with %CDT TIA) and with the direct nephelometric method (r = 0.949 with N Latex CDT). The higher CDT values observed with the %CDT TIA® method in patients with high trisialotransferrin (N 6%), resulted from partial co-elution of tri- and disialotransferrin [23]. The high number of results in this study may be the reason for the higher correlation with %CDT TIA, as compared to other studies [12]. However, as shown in Fig. 2, the
Fig. 3. Distribution of CDT values measured by the CZE Capillarys® method in 139 control subjects and 203 alcohol abusers.
52
F. Schellenberg et al. / Clinica Chimica Acta 382 (2007) 48–53
correlation was weak for subjects with CDT values below the cut-off. This heterogeneity might be the consequence of both imprecision in low values and methodological differences. 3.6. Case-control study Among the 357 subjects included, 147 controls subjects and 210 alcohol abusers, the Capillarys® CZE method was not able to measure CDT in 15 patients, 7 with a disialotransferrin peak lower than the integration threshold, 4 with high trisialotransferrin peak associated to low disialotransferrin, 3 genetic variants and 1 with a monoclonal peak. Thus results from 342 patients were included. They were 203 alcohol abusers (24 female, 179 male, aged 25–71 years, mean 47 ± 9 years), and 139 control subjects (26 female, 113 male, aged 18–75 years, mean 48 ± 7 years). Mean CDT was 0.9% (±0.4%) in the control group and 4.5% (± 5.0%) in the alcohol abuser group. The D'Agostino–Pearson test for normal distribution was performed to check the distribution of CDT values in this group. The test indicated that this distribution was outside normality acceptance limits (p = 0.002). As shown in Fig. 3, there is an overlap of the results in both groups for CDT values ranging from 0.4% to 2.1%. In this sample, the distribution of CDT values in the AUD group appears bimodal, as if this population was constituted of a group of “non-responder” subjects with a distribution of CDT values apparently similar to that of the control subjects, and a group of “responder” with high CDT values. The area under the receiver operating characteristics (ROC) curve was 0.89 (C.I. 0.85–0.92), as compared with values ranging from 0.66 [24] to 0.91 [25] in previous studies based on other CZE methods. When applying the cut-off proposed by the manufacturer (1.3%), 16 control patients had abnormally high values (Sp = 0.88). In the alcohol abuser group, the number of false negative values was 52 (Se = 0.74). Considering the lower prevalence of AUD in the general population compared to the study sample, and the possible consequences of falsely positively classifying patients, we considered that specificity should have a greater importance than sensitivity in the interpretation of the CDT results. So we calculated the diagnostic values using 1.4 as cut-off. This threshold increased the specificity to 0.91, while the sensitivity decreased to 0.71. This cut-off should be confirmed in other studies to avoid bias due to subject recruitment. Such bias may partly explain the discrepancies in the diagnostic values observed in other studies, with sensitivities ranging from 0.56 [26] to 0.88 [11] and specificities from 0.68 [24] to 0.99 [11]. Compared to specialized monocapillary instruments, the Capillarys® system allows the measurement of CDT with comparable precision, it is an automated process including sample pre-treatment, and has the high productivity of an eightcapillary system. Acknowledgements The authors are grateful to Mrs. Annie Détruit, Nadine Puche, and Geneviève Wattelet for their technical assistance. The
authors thank Sebia, Roche and Bio-Rad for providing the reagents used in this study. They thank Roche and Bio-Rad for the lending of the BN Prospec® and Variant™ systems. The case-control study was partly supported by INSERM grant “ATC Alcool” #ALC0205. References [1] Salaspuro M. Carbohydrate-deficient transferrin as compared to other markers of alcoholism: a systematic review. Alcohol 1999;19:261–71. [2] Bortolotti F, De Paoli G, Tagliaro F. Carbohydrate-deficient transferrin (CDT) as a marker of alcohol abuse: a critical review. J Chromatogr B 2006;841:96–109. [3] Wuyts B, Delanghe JR. The analysis of carbohydrate-deficient transferrin, marker of chronic alcoholism, using capillary electrophoresis. Clin Chem 2003;41:739–45. [4] Stibler H, Allgulander C, Borg S, et al. Abnormal microheterogeneity of transferrin in serum and cerebrospinal fluid in alcoholism. Acta Med Scand 1978;204:49–56. [5] Schellenberg F, Weill J. Serum desialotransferrin in the detection of alcohol abuse. Definition of a tf index. Drug Alcohol Depend 1987;19: 181–91. [6] Stibler H, Borg S, Joustra M. Micro anion exchange chromatography of carbohydrate-deficient transferrin in serum in relation to alcohol consumption (Swedish patent 8400587-5). Alcohol Clin Exp Res 1986;10:535–44. [7] Schellenberg F, Martin M, Caces E, et al. Nephelometric determination of carbohydrate deficient transferrin. Clin Chem 1996;42:551–7. [8] Viitala K, Lähdesmäki K, Niemelä O. Comparison of the Axis %CDT TIA and the CDTect method as laboratory tests of alcohol abuse. Clin Chem 1998;44:1209–15. [9] Wuyts B, Delanghe JR, Kasvosve I, et al. Determination of carbohydratedeficient transferrin using capillary zone electrophoresis. Clin Chem 2001;47:247–55. [10] Legros F, Nuyens V, Minet E, et al. Carbohydrate-deficient transferrin isoforms measured by capillary zone electrophoresis for detection of alcohol abuse. Clin Chem 2002;48:2177–86. [11] Prasad R, Stout RL, Coffin D, et al. Analysis of carbohydrate deficient transferrin by capillary electrophoresis. Electrophoresis 1997;18:1814–8. [12] Fermo I, Germagnoli L, Soldarini A, et al. Capillary zone electrophoresis for determination of carbohydrate-deficient transferrin in human serum. Electrophoresis 2004;25:469–75. [13] Lanz C, Kuhn M, Bortolotti F, et al. Evaluation and optimization of capillary zone electrophoresis with different dynamic capillary coatings for the determination of carbohydrate-deficient transferrin in human serum. J Chromatogr A 2002;979:43–57. [14] Jeppsson JO, Kristensson H, Fimiani C. Carbohydrate-deficient transferrin quantified by HPLC to determine heavy consumption of alcohol. Clin Chem 1993;39:2115–20. [15] Helander A, Eriksson G, Stibler H, et al. Interference of transferrin isoform types with carbohydrate-deficient transferrin quantification in the identification of alcohol abuse. Clin Chem 2001;47:1225–33. [16] Bortolotti F., De Paoli G., Pascali J.P., et al. Fully automated analysis of Carbohydrate-Deficient Transferrin (CDT) by using a multicapillary electrophoresis system. Clin Chim Acta in press, doi:10.1016/j.cca.2006.12.028. [17] Crivellente F, Fracasso G, Valentini R, et al. Improved method for carbohydrate-deficient transferrin determination in human serum by capillary zone electrophoresis. J Chromatogr B Biomed Sci Appl 2000;739:81–93. [18] Lanz C, Kuhn M, Deiss V, et al. Improved capillary electrophoresis method for the determination of carbohydrate-deficient transferrin in patient sera. Electrophoresis 2004;25:2309–18. [19] Joneli J, Lanz C, Thormann W. Capillary zone electrophoresis determination of carbohydrate-deficient transferrin using the new CEofix reagents under high-resolution conditions. J Chromatogr A 2006;1130: 272–80.
F. Schellenberg et al. / Clinica Chimica Acta 382 (2007) 48–53 [20] Helander A, Husa A, Jeppsson JO. Improved high-performance liquid chromatography method for carbohydrate-deficient transferrin in serum. Clin Chem 2003;49:1881–90. [21] Helander A, Bergström JP. Determination of carbohydrate-deficient transferrin in human serum using the Bio-Rad %CDT by HPLC test. Clin Chim Acta 2006;371:187–90. [22] Delanghe JR, Helander A, Wielders JPM, Pekelharing JM, Roth HJ, Schellenberg F, Born C, Yagmur E, Gentzer W, Althaus H. Clin Chem in press, doi:10.1373/clinchem.2006.08.4459. [23] Alden A, Ohlson S, Pahlsson P, et al. HPLC analysis of carbohydrate deficient transferrin isoforms isolated by the Axis-Shield %CDT method. Clin Chim Acta 2005;356:143–6.
53
[24] Daeppen JB, Anex F, Favrat B, et al. Carbohydrate-deficient transferrin measured by capillary zone electrophoresis and by turbidimetric immunoassay for identification of young heavy drinkers. Clin Chem 2005;51:1046–8. [25] Legros FJ, Nuyens V, Baudoux M, et al. Use of capillary zone electrophoresis for differentiating excessive from moderate alcohol consumption. Clin Chem 2003;49:440–9. [26] Wuyts B, Delanghe JR, Kasvosve I, et al. Carbohydrate-deficient transferrin and chronic alcohol ingestion in subjects with transferrin CDvariants. Clin Chem Lab Med 2001;39:937–43.