Carbohydrate-deficient transferrin isoforms measured by capillary zone electrophoresis for detection of alcohol abuse.
Normal human transferrin (Tf), produced mainly in hepatocytes, occurs in several isoforms as a result of differences in glycosylation (6). Tf has two asparagine-linked N-glycosylation sites, both usually occupied by oligosaccharides (6). These N-glycan chains are composed of N-acetylglucosamine, mannose, galactose, and sialic acid (7,8). The two N-glycan chains of Tf show bi-, tri-, and tetraantennary branchings, each terminating with a negatively charged sialic acid residue (7). Tf is theoretically a group of isoforms with zero to eight sialic acid residues per molecule (9). The isoelectric points of these isoforms range from 5.2 to 5.9 (9), and their molecular masses vary from 75.37 to 79.61 kDa (10).
Although normal human serum contains high concentrations (70-80%) of tetrasialo-Tf and low concentrations of disialo- and trisialo-Tf, individuals with high alcohol intake display higher amounts of desialylated forms, i.e., di-, mono-, and asialo-Tf, the whole being known as CDT (11). In teetotalers, CDT accounts for <2.5-3% of total Tf (8,11,12). Alcohol consumption of 50-80 g/day has been shown to increase serum CDT above these values (8,11).
The major advantage of CDT compared with earlier laboratory tests is reported to be that it provides the highest specificity for alcohol exposure (13); however, in a review of 110 studies before June 1998, Scouller et al. (14) showed that results obtained with available CDT assays submitted to metaanalysis were not significantly better than [gamma]GT as indicators of excessive alcohol consumption when both were used in paired studies. Reported sensitivities ranged from <20% to 100%, with specificities varying from 75% to 100% (14). Such discrepancies have been attributed to the differences in populations studied, e.g., hospitalized alcoholics with liver disease vs healthy volunteers; to the various CDT assay methods; and to the different cutoff points used to define excessive alcohol consumption. To avoid the numerous biases of clinical studies on the biochemical diagnosis of alcoholism, Scouller et al. (14) recommended paired trials evaluating two or more assays with use of the same sample and the same reference standard.
In 1999, Arndt (15) asked the question: would asialo-Tf represent a specific marker of alcohol abuse? Assays performed by capillary electrophoresis (16) or HPLC (17) identified this isoform in alcoholics but not in teetotalers. Whether the best method is the en bloc separation of CDT desialylated Tf isoforms obtained by ion-exchange chromatography combined with a turbidimetric immunoassay (%CDT) or the separation of individual sialic acid-deficient Tf isoforms is a critical issue.
The present report deals with a capillary zone electrophoresis (CZE) method for the separation and detection of Tf isoforms. In this first report, we evaluated the potential diagnostic accuracy of this test by examining two highly contrasting groups, abstainers and individuals who chronically drank excessive amounts alcohol.
Materials and Methods
CDT was detected by anion-exchange chromatography/ immunoassay using the Axis-Shield %CDT reagent set (Axis-Shield). CZE was performed with a CEofix CDT reagent set (Analis) on a Beckman Coulter P/ACE System 5500 (Beckman Coulter) equipped with a ultraviolet detector and an interference filter at 214 run. Uncoated fused-silica capillaries (50-Am i.d.; length, 57 cm) were obtained from Analis. Anti-human Tf and anti-human C-reactive protein (CRP) polyclonal antibodies were purchased from Dako. Neuraminidase was obtained from Beckman Coulter. Reagent sets for [gamma]GT, AST, and ALT were provided by Beckman Coulter. CRP concentrations in alcohol abusers were measured by an immunoturbidimetric method (Beckman Coulter). The CRP concentration of teetotalers was determined by ultrasensitive kinetic nephelometry (Beckman Coulter). MCV was determined with a Cell-Dyn 4000 (Abbott). Colorimetric reactions were measured on a Synchron LX20 (Beckman Coulter). Immunoturbidimetric assays were performed on a Immage immunonephelometer (Beckman Coulter). ROC analyses were performed using Analyze-it, Ver. 1.6 (Analyze-it Software).
Between January and October 2001, we continuously enrolled 111 individuals known to be alcohol heavy consumers and 50 teetotalers. The classification of individuals as alcohol abusers or teetotalers was based on the Alcohol Use Disorders Identification Test (AUDIT) questionnaire (18) and self-reported alcohol habits.
Healthy adult teetotalers were volunteers recruited by the medical staff of the CHU Andre V6sale general hospital. One-half of these patients (n = 26) were postmenopausal women participating in a study approved by the medical council. They presented with no hypertension, diabetes, cardiovascular problems, or osteoporosis. The other volunteers were abstainers for philosophical reasons. Their anamnesis and laboratory tests did not reveal any chronic disease.
Eighty-nine patients admitted to the Hospital University Psychiatric Center Vincent Van Gogh and 22 inpatients of the Department of Gastroenterology of the CHU Andre V6sale, all entering a withdrawal program, were included in the study. Patients presenting an AUDIT score >11 were registered as alcohol abusers [36 females and 75 males; mean (SD) age, 37 (18) years]. Their mean daily ethanol intake during the last month before blood sampling was 201 (58) g, with consumption ranging from 80 to 400 g. Teetotalers [35 females and 15 males; mean age, 35 (15) years] had an AUDIT score of 0 and reported complete alcohol abstinence.
All patients were informed and agreed with the aims and modalities of the study, through overall anamnesis and one blood drawing aimed at overall screening of biological markers.
Clinicians and analysts worked double-blind. Two gastroenterologists (P.E. and J.-P.H.) selected the patients to be included in the study. Analysts were unaware of the selection, and gastroenterologists were not informed of the analyst results. Discussions comparing both types of data began in October 2001.
Blood samples were collected by venipuncture in Vacutainer serum tubes. Serum was obtained by centrifugation within 3 h after blood sampling and was stored at -30 [degrees]C. All samples were analyzed within the week after sampling. Enzyme markers ([gamma]GT, AST, and ALT) and CRP were analyzed according to IFCC methods.
The ratio of desialylated (0-2 sialic acid residues/molecule) isoforms to total Tf (%CDT) was determined according to the manufacturer's instructions for the %CDT assay (19). Serum Tf was saturated with F[e.sup.3+] before the desialylated Tf isoforms were separated on an anion-exchange chromatographic minicolumn. CDT and total Tf were measured by immunoturbidimetry with the same anti-Tf antibody.
Reagents from the CEofix CDT reagent set for P/ACE 5000 series (20) were used with modifications of the method recommended by the manufacturer. Serum Tf was saturated with iron by the addition of 50 [micro]L of 1 g/L ferric chloride to 50 [micro]L of serum. The capillary was first coated under pressure with a solution of polycation dissolved in 33 g/L malic acid, pH 4.8. This coating was followed by a 1.3-min rinse under pressure (20 psi) with a Tris-borate buffer, pH 8.5, containing a polyanion. The capillary was rinsed for 0.5 min under low pressure (0.5 psi) with the same buffer. After a 3-s low-pressure injection of 0.1 g/L sodium dodecyl sulfate, the iron-saturated sample was eluted by application of low pressure for 2 s. Borate buffer was then injected for 1 s. Separation of the Tf isoforms was performed over 7 min in the borate buffer at a constant voltage of 28 kV at 40 [degrees]C. This step was followed by a 1.5-min rinse with 0.2 mol/L NaOH. The reliability of the capillary was finally tested by a 1.5-min voltage increase reaching an intensity of 100 [micro]A. The same procedure was repeated with the next sample. Detection was by absorbance at 214 nm.
The peaks presumed to represent the different Tf isoforms were quantified as a percentage of the total Tf content, in terms of valley-to-valley area under the curve (AUC). Results were recorded on an electropherogram after treatment by integration software (Beckman Coulter). The detection limit for asialo- and monosialo-Tf was 0.03% of total Tf. CDT isoform concentrations were based on the ratio of the asialo- and disialo-Tf peak areas to total isotransferrins. The AUC for the percentage of trisialo-Tf was also calculated, and its relevance as a desialylated isoform was tested. The migration times (MTs) of the isoforms were registered and compared.
IDENTIFICATION OF Tf ISOFORMS
Anti-human Tf rabbit antiserum was added to a final 1:3 dilution after a first CZE analysis run of the undiluted serum. The electropherograms obtained before and after immunosubtraction (21) were compared. Anti-human CRP was injected into the capillary under pressure for 3 s before injection of the serum. The electropherograms were then compared.
[FIGURE 1 OMITTED]
After iron saturation by use of ferric chloride, neuraminidase was added to 100 [micro]L of serum to obtain a concentration of 100 U/L, and the mixture was incubated at 37[degrees]C for 24 h. Aliquots were removed after 2.5, 6, and 24 h and submitted to CZE analysis to monitor progress of the reaction. Immunoprecipitation with anti-human Tf antiserum was then performed.
PERFORMANCE OF THE ASSAY
The CVs for the percentage of each isoform and the MTs were calculated to assess the within- and between-run precision. The within-run precision for the asialo-, di-, tri-, and tetrasialo-Tf isoforms was determined by assaying the same serum 10 times consecutively. The between-run variation was assessed by analyzing samples from alcohol abusers and teetotalers once a day for 10 days.
Results are expressed as the mean and SD. SigmaStat[R] software (Jandle Scientific) was used. The statistical significance of the differences observed between the clinical groups for CZE-isolated Tf isoforms and %CDT was estimated by the Student t-test. The asymmetry of the range of biomarker values ([gamma]GT, AST, and ALT) was normalized by use of a natural logarithmic scale. Sample values were compared by use of the nonparametric Wilcoxon test. Geometric means and SDs were calculated, leading to asymmetric 95% confidence intervals. We investigated the correlations between the anion-exchange chromatography-immunoturbidimetry method and capillary electrophoresis by comparing the %CDT measured by the Axis assay with the percentages of asialo, disialo-, and (asialo- + disialo-To isoforms obtained by CZE, using the Pearson correlation test. Correlations between the two methods of CDT determination and [gamma]GT, AST, ALT, MCV, and AUDIT score were obtained by the same method.
The diagnostic accuracies of asialo-, disialo-, and trisialo-Tf; the sum (asialo- + disialo-TO; the sum (asialo- + disialo- + trisialo-TO; the sum (disialo- + trisialo-TO; and %CDT were assessed by calculation of the areas under the ROC curves (ROC areas) (22).
The enzyme activities and the MCVs of the teetotalers were within the reference intervals (Table 1). We observed significantly increased mean serum MCV, [gamma]GT, and ALT in individuals with high chronic alcohol consumption compared with those who abstained. Mean CRP was higher in alcohol abusers than in teetotalers (35 [+ or -] 53 vs 2 [+ or -] 1 mg/L; P <0.001). At cutoffs corresponding to the laboratory's published reference limits, the sensitivities of the currently used alcohol biomarkers were [gamma]GT > ALT > AST > MCV, and specificities were AST > ALT > MCV > [gamma]GT (Table 1).
Tf migrating in the [beta]-region was visualized in 7-min electropherograms by immunosubtraction with a polyclonal anti-human Tf serum (see Supplemental File 1, available with the online version of this article at http:// www.clinchem.org/content/vol48/issuel2/). Seven peaks migrating between 5.55 and 6.2 min were observed in the serum of an alcohol abuser (Fig. 1). These peaks were numbered PO to P6. A predominant peak, P4, was observed at ~6 min. Three, and occasionally four, peaks migrated earlier. Their MTs averaged 5.55, 5.68, 5.77, and 5.84 min, respectively (Table 2). Two isoforms migrating later than the predominant peak at 6 min were also visualized, with MTs of 6.1 and 6.2 min. Fig. 1 shows the absence of the early PO (MT, 5.55 min) and P1 (MT, 5.68 min) peaks in a teetotaler.
IMMUNOLOGIC RESOLUTION OF Tf ISOFORMS
Addition of anti-Tf polyclonal antibody suppressed all peaks of the two individuals shown in Fig. 1, with remnants of the P1 (MT, 5.68 min) and, occasionally, the P4 predominant peaks remaining. Anti-CRP serum partly immunosubtracted peak P1 of alcohol abusers (see Supplemental File 2, available with the online version of this article at http://www.clinchem.org/content/vol48/ issue12/).
ENZYMATIC TREATMENT OF Tf
During treatment with 100 U/L neuraminidase, the P1/ CRP, P3, P4, P5, and P6 peaks in the alcohol abuser's serum had disappeared after 2.5 h (Fig. 2A). The size of the P2 peak progressively decreased during 6 h of enzymatic treatment. A new, earlier peak (Pa; MT, 5.6 min) appeared after 2.5 h and progressively decreased up to 24 h. Another new peak (Pb; MT, 5.4 min) appeared and increased continuously from 2.5 to 24 h. An even earlier peak (Pc; MT, 5.3 min) appeared and increased from 6 to 24 h. The PO peak of alcohol abusers was not modified during the whole enzymatic treatment. Remaining peaks were immunoprecipitated by anti-Tf (Fig. 2A). Identical results were obtained with sera of 10 alcohol abusers.
[FIGURE 2 OMITTED]
Similar kinetics were observed during enzymatic treatment of the serum of a teetotaler (Fig. 213). Peaks P6 to P2 disappeared after 2.5 h of treatment with neuraminidase. Peak Pa appeared after 2.5 h, and slightly decreased thereafter. Peaks Pb and Pc appeared at 2.5 h and increased in size with time of incubation. Identical results were obtained with samples from seven teetotalers. The remaining peaks were immunosubtracted by anti-Tf (Fig. 213).
MOBILITY OF THE Tf ISOFORMS
When we combined the individual MTs of the different isoforms in the two populations, the CVs averaged 1% or lower. The MTs of each isoform were statistically identical in the two populations (P >0.05; Table 2). The mean MT of each peak was statistically different from the others (P <0.05).
CZE QUANTITATIVE DATA
Whereas peak PO was found in 102 of 111 alcohol abusers, it was not observed in any of the teetotalers (Table 3). Peak P2 was five times higher in sera from alcohol abusers than from teetotalers. The sum of peaks PO + P2 averaged 3% in alcohol abusers, and was significantly decreased by sixfold in teetotalers. The relative percentage (AUC) of peak P3 was similar for alcohol abusers and teetotalers. Peak P4 was higher in teetotalers than in heavy consumers. The relative percentages (AUC) of P5 and P6 were similar in the two groups (Table 3).
PRECISION OF THE CZE METHOD
The percentages of asialo-Tf varied between 0.03 and 2.3 in alcohol abusers. The mean between-run CVs for the "low" (<1% of the AUC for total To and "high" (>1%) asialo-Tf were 7.4% and 4.6%, respectively. The between-run CVs were also <8% for disialo-Tf and for (asialo- + disialo-To. The within-run CVs were 0.3% for tetrasialo-Tf, 2.5% for disialo-Tf, and 4.5% for asialo-Tf.
The MT within-run analytical imprecision of a sample analyzed 10 times consecutively did not exceed 0.3% in both groups. When analyzed consecutively over 10 days, CVs were [less than or equal to] 1%. The MTs were statistically identical during the whole treatment with neuraminidase, and their CVs did not exceed 2%.
DETERMINATION OF CDT BY ANION-EXCHANGE CHROMATOGRAPHY-IMMUNOTURBIDIMETRY
The mean percentage of CDT in teetotalers was below the thresholds of 2.6% or 3%. It was significantly (P <0.0001) increased in alcohol abusers (5.3% [+ or -] 3% vs 2.4% [+ or -] 0.4%; P <0.001). Results from %CDT were higher (5.3% vs 0.5%; P <0.0001) than those obtained by CZE (asialo- + disialoTf).
On the basis of the presence or absence of asialo-Tf as a marker of chronic alcohol abuse, the sensitivity of diagnosis was 0.92 at a specificity of 1.0 (Fig. 3 and Table 4). No apparent differences in the sensitivity and specificity of asialo-Tf were observed between males and females or between the two withdrawal centers.
For the other CZE isoforms, cutoffs were selected that emphasized specificity. For (asialo- + disialo-Tf), the specificity and sensitivity were lower, but they were higher than those for %CDT (Fig. 4 and Table 4). The sensitivity and specificity for trisialo-Tf were very low, approximating the ROC analysis discrimination limit (Fig. 3). The 95% confidence interval of the ROC area for disialo-Tf was inferior to that of asialo-Tf, but the addition of trisialo-Tf dramatically decreased the performance of CZE detection of CDT [see Supplemental File 3 (available with the online version of this article at http://www. clinchem.org/content/vol48/issuel2/) and Table 4]. A %CDT cutoff of 3% (12) gave a specificity of 0.92 and a sensitivity of 0.79, whereas the 2.6% threshold recommended by the manufacturer gave a specificity of 0.69 and a sensitivity of 0.83 (Table 4). When we applied the 2.8% cutoff of Helander et al. (19), the specificity was 0.83 and the sensitivity was 0.80.
CORRELATIONS BETWEEN CZE DETERMINATION OF CDT AND OTHER ASSAYS
We found correlations between the percentages (AUC) of asialo-Tf, disialo-Tf, and (asialo- + disialo-To measured by CZE in alcohol abusers (Table 5). We also found good correlations between %CDT values determined by %CDT and the relative percentages (AUC) of asialo-Tf, disialo-Tf, and (asialo- + disialo-To. We found no correlation between the relative percentages (AUC) of asialo-Tf, disialoTf, and (asialo- + disialo-To measured by CZE and the currently used biomarkers [gamma]GT, AST, ALT, and MCV, or the AUDIT score (Table 5).
CZE uses narrow-bore capillaries to perform high-efficiency separation of both large and small molecules. Separation occurs on the basis of electric charge and molecular mass and is facilitated by the use of high voltages, which may generate electroosmotic and electrophoretic flow of buffer solutions and ionic species, respectively, within the capillary (23). The latest migrating Tf isoform will present the most negative electric charges attributable to sialic acid residues.
[FIGURE 4 OMITTED]
CZE has been particularly effective for the resolution of protein glycoforms (24, 25). Because of the negative electric charges conferred to Tf by the terminal sialic acid residues of the glycan chains (9) and the different molecular masses of the isoforms (10), CZE represents a good candidate for resolution of these glycoprotein isoforms (16,20,26-30). Results are presented as printed charts, as currently recommended for CDT identification and quantification (12).
Tf is found within the [beta]-globulin fraction in serum protein capillary electrophoresis (23). Undiluted Tf offers a multitude of isoforms (6), depending on the iron supply, because molecules are iron-free or loaded with one or two [Fe.sup.3+] ions. Iron saturation of serum Tf has been performed before injection of the sample into the capillary to reduce the number of Tf isoforms occurring in serum (24), This represents the sole step performed outside the CZE instrument.
The relative percentages of Tf isoforms reported by other authors (8,12) were <1% each for asialo- and monosialo-Tf, <2.5% for disialo-Tf, 4.5-9% for trisialo-Tf, 70-80% for tetrasialo-Tf, 12-18% for pentasialo-Tf, and 1-3% for hexasialo-Tf. Similar values were found in the present study.
Peaks obtained inside double-coated capillaries were certified to be Tf isoforms by immunosubtraction (16, 20) with a polyclonal anti-Tf serum (Fig. 1 and Supplemental File 1). The Beckman Coulter software allowed us to focus inside this region (Figs. 1 and 2) by determining the MTs and the relative AUC percentages of the immunoprecipitated peaks (Tables 2 and 3). It is known that tetrasialo-Tf is the predominant (70-80%) isoform (5, 8,11,12, 30) and that it should correspond to peak P4 (Figs. 1 and 2). Peaks P5 and P6 would represent more sialylated isoforms, probably pentasialo- and hexasialo-Tf. The similarities between the percentages that we obtained for peaks P0, P1, P2, and P3 and the data in the literature also allowed us to associate these peaks with the asialo to trisialo forms, respectively. Heptasialo- and octasialo-Tfs were not visualized.
Addition of anti-CRP polyclonal serum led to partial immunosuppression of the P1, presumed monosialo-Tf, isoform of alcohol abusers (see Supplemental Files 2). In our experimental conditions, CRP comigrated with one monosialylated Tf isoform within peak P1. Immunoprecipitation of the CRP in peak P1 might be related to the increased CRP measured in the serum of most heavy alcohol consumers. On the other hand, the absence of the same peak in healthy teetotalers correlated with a low CRP concentration and implied a monosialo-Tf concentration <0.03%. This peak percentage was not included in CDT measurements, which were limited to asialo- and disialo-Tf.
Neuraminidase successively removes terminal sialic acid residues, leading to a shift from higher to less sialylated isoforms (16, 20, 30). During hydrolysis of the sialic acid residues by neuraminidase, PO was not modified for 24 h, although it was immunosubtractable by anti-human Tf polyclonal serum (Fig. 2A). An isoform with no sialic acids would not be affected by neuraminidase treatment when initially present in serum and would be immunosubtractable. PO is thus likely to constitute an asialylated isoform. Treatment with N-glycosidase would confirm whether PO is aglycosylated, as suggested by others (9,12).
During treatment with neuraminidase, two early-eluting isoforms, Pb and Pc, both of which were immunosubtractable by anti-Tf (Fig. 2), appeared, whereas latereluting sialylated forms disappeared. These results outline the scarcity of negative charges (sialic acid residues) in these forms resulting from treatment with neuraminidase. Treatment with N-glycosidase would confirm the N-glycan chain content of those forms. For the chemical structures of the disialo- to pentasialo-Tf isoforms, we referred to their currently accepted configuration (9,12). This implies that trisialo- to hexasialo-Tf isoforms contain two N-glycans, whereas only one chain occurs in disialo-Tf.
The presence of serum asialo-Tf induced by alcohol abuse has been reported in the literature. Our results confirmed this finding in the sera of alcohol abusers analyzed by CZE (16, 20). Recent reports mentioned the presence of several asialo-Tf isoforms detected by two-dimensional gel electrophoresis (31) and electrospray mass spectrometry (32). Our conclusions concerning Tf isoforms obtained by neuraminidase treatment are based on the generation of one monosialylated and three asialylated isoforms, one being found in the serum of alcohol abusers and two others being generated by enzymatic treatment with neuraminidase (Fig. 2). Successive enzymatic treatments with neuraminidase and N-glycosidase should highlight our present hypothesis. Dose-response studies of enzymatic treatment should also be performed (see Supplemental File 4, which accompanies the online version of this article at http://www.clinchem.org/ content/vo148 /issuel2/).
The similarities of the kinetics of the enzymatic treatments in both groups reinforced the chemical identities of the sialylated Tf isoforms in adults who abused or abstained from ethanol. Corresponding isoforms in both groups had the same MTs, which indicated identical charges and masses. The sole differences were the relative percentages of desialylated (disialo- and asialo-Tf) forms between the two populations. Asialo- and monosialo-Tf were not found in the serum of any teetotaler. This confirms the impaired sialylation of Tf induced by alcohol abuse, which probably is attributable to the alteration of glycosyltransferases, particularly sialyltransferase (33-35).
Our use of both clinical and laboratory data in classifying the populations was aimed at minimizing the influence of extraneous factors and biases (14). The major bias of the present study was the exclusion of alcohol drinkers who consumed <50 g of ethanol/day and presented an AUDIT score <11. We found no significant correlation between AUDIT scores and the numerical data for asialo-Tf, disialo-Tf, and (asialo- + disialo-Tf), based on the Pearson correlation (Table 5). However, our selection of teetotalers on the basis of an AUDIT score of 0 and alcohol abusers on the basis of AUDIT scores >11 and on self-reported alcohol habits fit rather well with a diagnosis based on the absence or presence of asialo-Tf (Table 4). We found no correlations between CZE results and MCV, [gamma]GT, AST, or ALT, the currently used biological markers of alcohol abuse (Table 5), as shown previously (36).
The most relevant points concerning the sensitivities and specificities of the various assays used in the present study are summarized in Tables 1 and 4. The highest sensitivities occurred for asialo- and disialo-Tf, 2.6% CDT, and (asialo- + disialo-To, all of which were >0.8. The absence of asialo-Tf yielded a specificity of 1.0, with a sensitivity of 0.92. Specificities of ~0.9 were obtained for AST, ALT, MCV, disialo-Tf, (asialo- + disialo-To, and 3% CDT, whereas [gamma]GT had a specificity of ~0.8.
The debate on whether to include or exclude trisialo-Tf from CDT has caused considerable confusion regarding the clinical use of CDT (22, 37-41). The absence of a significant increase in trisialo-Tf, as measured by CZE, after chronic alcohol abuse (Table 2) demonstrated that this isoform might not be useful for the diagnosis of alcoholism. In addition, ROC curves indicated that CZE measurements of trisialo-Tf, (asialo- + disialo- + trisialoTf), and (disialo- + trisialo-To had poor sensitivity, inferior to that of 3% CDT (Table 4). Inclusion of trisialo-Tf in CDT concentrations measured by CZE led to a specificity of -0.6, similar to that for the 2.6% CDT cutoff value (Table 4).
Our data on asialo-Tf confirm previous observations (16,17). Asialo-Tf has been found in the serum of alcohol abusers but not in the serum of teetotalers. Use of asialo-Tf as a clearly defined analyte, rather than the analyte group CDT, as hypothesized previously (15), improved the diagnostic accuracy of laboratory diagnosis in our study. Indeed, asialo-Tf showed the highest sensitivity and specificity when compared with the other (or combinations of other) sialic acid-deficient Tf isoforms. Thus, our data (Figs. 3 and 4 and Table 4) and data from the literature (15-17, 32) provide evidence for the possible development of a very specific test based on CZE.
The present study focused strictly on two completely contrasting groups, namely teetotalers and excessive consumers of alcohol. It is unlikely that a specificity of 1 and a sensitivity of 0.92 will be obtained in the clinical application of the CZE method, when individuals who consume moderate amounts of alcohol and chronic alcohol abusers will be the groups to be differentiated. A second study, dealing with alcohol abusers and moderate drinkers, will verify the usefulness of the analytes of the present study.
This work was supported by a grant from the Intercommunale de Sante Publique du Pays de Charleroi, which involves several Hospital University Centers of Charleroi County, including the CHU Andre Vesale. We particularly acknowledge Drs. Francois Charon, Eric Fontaine, and Dominique Schoefs (all from the University Psychiatric Hospital Vincent Van Gogh), and Dr. Catherine Gr6goir (CHU Andre Vesale) for ensuring the clinical follow-up of alcohol abusers and teetotalers. We thank Prof. Francis Cantraine (Department of Computer Science, University de Bruxelles, School of Medicine) for decisive help in statistical interpretation of the results. We are indebted to Jacques Janssens for training in the art of capillary electrophoresis. We are grateful to Nadya Sioiki for technical assistance with CZE; we also thank Mireille Roels for skillful assistance regarding immunoturbidimetric CDT and Liliane Kukolja for registration of patients and follow-up of files.
(1.) Conigrave KM, Saunders JB, Reznik RB, Withfield JB. Prediction of alcohol-related harm by laboratory test results. Clin Chem 1993; 39:2266-70.
(2.) Rosman AS, Lieber CS. Diagnostic utility of laboratory tests in alcoholic liver disease. Clin Chem 1994;40:1641-51.
(3.) Reynaud M, Schellenberg F, Loiseux-Meunier MN, Schwan R, Maradeix B, Planche F, et al. Objective diagnosis of alcohol abuse: compared values of CDT, [gamma]-glutamyl transferase (GGT), and mean corpuscular volume (MCV). Alcohol Clin Exp Res 2000;24: 1414-9.
(4.) Menninger JA, Baron AE, Conigrave KM, Whitfield JB, Saunders JB, Helander A, et al. Platelet adenyl cyclase activity as a trait marker of alcohol dependence. WHO/ISBRA Collaborative Study Investigators. International Society for Biomedical Research on alcoholism. Alcohol Clin Exp Res 2000;24:810-21.
(5.) Arndt T. Carbohydrate-deficient transferrin as a marker of chronic alcohol abuse: a critical review of preanalysis, analysis, and interpretation. Clin Chem 2001;47:13-27.
(6.) de Jong H, van Eijk H. Microheterogeneity of human serum transferrin: a biological phenomenon studied by isoelectric focusing in immobilized pH gradients. Electrophoresis 1988;9:589-98.
(7.) Spik G, Debruyne V, Montreuil J, van Halbock H, Vliegenhart JFG. Primary structure of two sialylated glycans from human serotransferrin. FEBS Lett 1985;183:65-9.
(8.) de Jong G, van Dijk JP, van Eijk HG. The biology of transferrin. Clin Chim Acta 1990;190:1-46.
(9.) Landberg E, Pahlsson P, Lundblad A, Ametrop A, Jeppsson J-A. Carbohydrate composition of serum transferrin isoforms from patients with high alcohol consumption. Biochem Biophys Res Commun 1995;210:267-74.
(10.) Peter J, Unverzagt C, Engel W-D, Renauer D, Seidel C, H6sel W. Identification of carbohydrate deficient transferrin forms by MALDITOF mass spectrometry and lectin ELISA. Biochim Biophys Acta 1998;1380:93-101.
(11.) Stibler H. Carbohydrate-deficient transferrin in serum: a new marker of potentially harmful alcohol consumption reviewed. Clin Chem 1991;37:2029-37.
(12.) Helander A, Eriksson G, Stibler H, Jeppsson J-0. Interference of transferrin isoform types with carbohydrate-deficient transferrin quantification in the identification of alcohol abuse. Clin Chem 2001;47:1225-33.
(13.) Meerkerk GJ, Njoo KH, Bongers IM, Trienekens P, van Oers JA. The specificity of the CDT assay in general practice: the influence of common chronic diseases and medication on the serum CDT concentration. Alcohol Clin Exp Res 1998;22:908-13.
(14.) Scouller K, Conigrave KM, Macaskill P, Irwig L, Withfield JB. Should we use carbohydrate-deficient transferrin instead of y- glutamyltransferase for detecting problem drinkers? A systematic review and metaanalysis. Clin Chem 2000;46:1894-902.
(15.) Arndt T. Carbohydrate-deficient transferrin (CDT)-should this be replaced by asialo-Fe2transferrin and thus standardized? [Abstract]. Alcohol Alcohol 1999;34:447.
(16.) Trout AL, Prasad R, Coffin D, DiMartini A, Lane T, Blessum C, et al. Direct capillary electrophoresis detection of carbohydrate-deficient transferrin in neat serum. Electrophoresis 2000;21:237683.
(17.) Jeppsson J-0, Kristensson H, Fimiani C. Carbohydrate-deficient transferrin quantified by HPLC to determine heavy consumption of alcohol. Clin Chem 1993;39:2115-20.
(18.) Saunders JB, Aasland OJ, Babor TF, de la Fuente JR, Grant M. Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO Collaborative Project on Early Detection of Persons with Harmful Alcohol Consumption-II. Addiction 1993;88:791804.
(19.) Helander A, Fors M, Zakrisson B. Study of Axis-Shield %CDT immunoassay for quantification of carbohydrate-deficient transferrin (CDT) in serum. Alcohol Alcohol 2001;36:406-12.
(20.) Wuyts B, Delanghe JR, Kasvosve I, Wauters A, Neels H, Janssens J. Determination of carbohydrate-deficient transferrin using capillary zone electrophoresis. Clin Chem 2001;47:247-55.
(21.) Katzman JA, Clark R, Sanders E, Landers JP. Prospective study of serum protein capillary zone electrophoresis and immunotyping of monoclonal proteins by immunosubtraction. Am J Clin Pathol 1998;110:503-9.
(22.) Arndt T, Korzec A, Bar M, Kropf J. Further arguments against including trisialo-Fe2transferrin in carbohydrate-deficient transferrin (CDT): a study on male alcoholics and hazardous drinkers. Med Sci Monit 2002;8:411-8.
(23.) Blessum C, Jeppsson JO, Aguzzi F, Bernon H, Bienvenu J. Capillary electrophoresis: principles and practice in clinical laboratory. Ann Biol Clin (Paris) 1999;57:643-57.
(24.) Landers JP, Oda RP, Madden BJ, Spelsberg T. High-performance capillary electrophoresis of glycoproteins: the use of modifiers of electroosmotic flow for analysis of microheterogeneity. Anal Biochem 1992;205:115-24.
(25.) Oda RP, Landers JP. High-resolution glycoprotein analysis using capillary electrophoresis. Mol Biotechnol 1996;5:165-70.
(26.) Prasad R, Stout RL, Coffin D, Smith J. Analysis of carbohydrate deficient transferrin by capillary zone electrophoresis. Electrophoresis 1997;18:1814-8.
(27.) Oda RP, Prasad R, Stout RL, Coffin D, Patton WP, Kraft DL, et al. Capillary electrophoresis-based separation of transferrin sialoforms in patients with carbohydrate-deficient glycoprotein syndrome. Electrophoresis 1997;18:1819-26.
(28.) Tagliaro F, Crivellente F, Manetto G, Puppi I, Deyl Z, Marigo M. Optimized determination of carbohydrate-deficient transferrin isoforms in serum by capillary zone electrophoresis. Electrophoresis 1998;19:3033-9.
(29.) Crivellente F, Fracasso G, Valentini R, Manetto G, Riviera AP, Tagliaro F. Improved method for carbohydrate-deficient transferrin determination in human serum by capillary zone electrophoresis. J Chromatogr 2000;739:81-93.
(30.) Beisler AT, Kelly RH, Landers JP. Circumventing complement C3 interference in the analysis of carbohydrate-deficient transferrin in fresh serum. Anal Biochem 2000;285:143-50.
(31.) Henry H, Froehlich F, Perrer R, Tissot J-D, Eilers-Messerli B, Lavanchy D, et al. Microheterogeneity of serum glycoproteins in patients with chronic alcohol abuse compared with carbohydrate deficient glycoprotein syndrome type I. Clin Chem 1999;45: 1408-13.
(32.) Bergen HR, Lacey JM, O'Brien JF, Naylor S. Online single-step analysis of blood proteins: the transferrin story. Anal Biochem 2001;296:122-9.
(33.) Stibler H, Borg S. Glycoprotein glycosyltransferase activities in serum in alcohol-abusing patients and healthy controls. Scand J Clin Lab Invest 1991;51:43-51.
(34.) Xin Y, Lasker JM, Lieber CS. Serum carbohydrate-deficient transferrin: mechanism of increase after chronic alcohol intake. Hepatology 1995;22:1462-8.
(35.) Lakshman MR, Rao MN, Marmillot P. Alcohol and molecular regulation of protein regulation and function. Alcohol 1999;19: 239-47.
(36.) Bell H, Tallaksen CM, Try K, Haug E. Carbohydrate-deficient transferrin and other markers of high alcohol consumption: a study of 502 patients admitted consecutively to a medical department. Alcohol Clin Exp Res 1994;18:1103-8.
(37.) Vittala K, Lahdesmaki K, Niemela 0. Comparison of the Axis %CDT TIA and the CDTect method as laboratory tests of alcohol abuse. Clin Chem 1998;44:1209-15.
(38.) Lipkowski M, Dibbelt I, Seyfarth M. Is there an analytical advantage from including trisialo transferrin into the fraction of carbohydrate-deficient transferrin? Lessons from a comparison of two commercial turbidimetric immunoassays with the carbohydrate-deficient transferrin determination by high performance liquid chromatography. Clin Biochem 2000;33:635-41.
(39.) Korzec A, Arndt T, Bar M, Koetler MWJ. Trisialo-Fe2transferrin does not improve the diagnostic accuracy of carbohydrate-deficient transferrin as a marker of chronic excessive alcohol intake. J Lab Med 2001;25:407-10.
(40.) Tagliaro F, Bortolotti F, Dorizzi RM, Marigo M. Caveats in carbohydrate- deficient transferrin determination [Letter]. Clin Chem 2002;48:208.
(41.) Delanghe JR, Wuyts B, de Bruyzere ML. Reply to Tagliaro et al. Clin Chem 2002;48:208-9.
FRANZ J. LEGROS,  * VINCENT NUYENS,  EDDY MINET,  PHILIPPE EMONTS,  KARIM ZOUAOUI BOUDJELTIA,  ANNE COURBE,  JEAN-LUC RUELLE,  JACQUES COLICIS,  FRANCOIS DE LTSCAILLE,  and JEAN-POL HENRY 
[1.] Laboratory of Experimental Medicine, Universite Libre de Bruxelles and Centre Hospitalier Universitaire Andre Vesale, 706, route de Gozee, B6110 Montigny-le-Tilleul, Belgium.
 University Department of Gastroenterology and
 Laboratory of Clinical Biology, Centre Hospitalier Universitaire Andre Vesale, 706, route de Gozee, B6110 Montigny-le-Tilleul, Belgium.
 R&D Laboratory, Analis SA, 14, rue Dewez, B5000 Namur, Belgium.
 Nonstandard abbreviations: yGT, y-glutamyltransferase; AST, aspartate aminotransferase; ALT, alanine aminotransferase; MCV, mean corpuscular volume; CDT, carbohydrate-deficient transferrin; Tf, transferrin; TIA, turbidimetric immunoassay; CZE, capillary zone electrophoresis; CRP, C-reactive protein; AUDIT, Alcohol Use Disorders Identification Test; AUC, area under the curve; and MT, migration time.
* Address correspondence to this author at: Laboratory of Experimental Medicine, CHU Andre V6sale, 706, route de Goz6e, 6110 Montigny-le-Tilleul, Belgium. Fax 32-71-92-47-10; e-mail email@example.com.
Received May 18, 2002; accepted September 26, 2002.
Table 1. Current markers of alcohol abuse. (a) MCV, [gamma] GT, AST, [micro]L (100) (b) U/L (50) U/L (37) AA (c) (n = 111) 99 (6) 92 (110) 51 (93) 95% Cl 81-115 10-352 12-55 TT (n = 50) 91 (4) (d) 25 (18) (e) 22 (9) (f) 95% Cl 86-106 11-36 11-41 Sensitivity 0.3 0.64 0.47 95% Cl 0.28-0.32 0.60-0.68 0.43-0.51 Specificity 0.92 0.86 0.95 95% Cl 0.90-0.94 0.83-0.89 0.94-0.96 ALT, U/L (43) AA (c) (n = 111) 47 (51) 95% Cl 9-69 TT (n = 50) 9 (4) (e) 95% Cl 4-19 Sensitivity 0.53 95% Cl 0.49-0.57 Specificity 0.93 95% Cl 0.91-0.95 (a) Mean (SD) and 95% confidence intervals of values measured in alcohol abusers and teetotalers. (b) Cutoff values correspond to the upper limits of the reference intervals used in the Laboratory of Clinical Chemistry, CHU Andree Veesale. (c) AA, alcohol abusers; CI, confidence interval; TT, teetotalers. (d) P <0.05. (e) P <0.0001. (f) P <0.005. Table 2. Mean (SD) MTs of the Tf isoforms in the two populations. MT, min P0 P1 P2 Healthy teetotalers ND (a) ND 5.77 (0.05) CV, % 0.8 Alcohol abusers 5.55 (0.06) 5.68 (0.06) 5.77 (0.06) CV, % 1.1 1.1 MT, min P3 P4 P5 Healthy teetotalers 5.83 (0.06) 5.96 (0.07) 6.09 (0.03) CV, % 1.1 1.2 0.5 Alcohol abusers 5.84 (0.07) 5.97 (0.08) 6.11 (0.02) CV, % 1.3 1.3 0.3 MT, min P6 Healthy teetotalers 6.19 (0.03) CV, % 0.5 Alcohol abusers 6.21 (0.02) CV, % 0.3 (a) ND, not detected. Table 3. Mean (SD) relative percentages (% total AUC) of the Tf isoforms. % total AUC P0 P1 P2 Healthy teetotalers 0 0 0.5 (0.2) 95% Cl (a) 0.45-0.55 Alcohol abusers 0.5 (0.4) ND 2.5 (2.3) (b) 95% Cl 0.4-0.61 2.1-2.9 % total AUC P0 + P2 P3 P4 Healthy teetotalers 0.5 (0.2) 4.9 (0.9) 79 (2) 95% Cl (a) 0.45-0.55 4.6-5.2 78.5-79.5 Alcohol abusers 3.0 (2.8) (b) 4.8 (2.2) (c) 77 (4) (b) 95% Cl 2.5-3.5 4.4-5.2 76.3-77.7 % total AUC P5 P6 Healthy teetotalers 12.8 (1) 2.2 (0.4) 95% Cl (a) 12.5-13 2.1-2.3 Alcohol abusers 12.2 (1) (c) 2.2 (0.3) (c) 95% Cl 12-12.4 2.15-2.25 (a) Cl, confidence interval; ND, not detected. (b) P <0.0001. (c) P >0.05. Table 4. Mean areas under ROC curves, 95% confidence intervals, sensitivities, and specificities at the optimal total Tf cutoffs (%). ROC area 95% Cl (a) Cutoff, % Asialo-Tf 0.96 0.93-0.99 0 Disialo-Tf 0.89 0.84-0.94 0.7 Trisialo-Tf 0.58 0.49-0.68 4.2 Asialo- + disialo-Tf 0.94 0.91-0.98 0.7 Asialo- + disialo- + trisialo-Tf 0.81 0.66-0.82 5.3 Disialo- + trisialo-Tf 0.78 0.63-0.79 5.3 % CDT 0.87 0.81-0.92 2.6 2.8 3 Sensitivity Specificity Asialo-Tf 0.92 1 Disialo-Tf 0.79 0.94 Trisialo-Tf 0.62 0.53 Asialo- + disialo-Tf 0.84 0.94 Asialo- + disialo- + trisialo-Tf 0.68 0.69 Disialo- + trisialo-Tf 0.67 0.74 % CDT 0.83 0.69 0.80 0.83 0.79 0.92 (a) Cl, confidence interval. Table 5. Pearson correlation coefficients ([r.sup.2]) and statistical significance (P) between alcohol abuse markers obtained from 111 alcohol abusers. [r.sup.2] (P) CE2 (a) CE 0 + 2 CDT MCV CE0 0.72 (0.0001) 0.95 (0.0001) 0.82 (0.0001) -0.038(0.78) CE2 0.90 (0.0001) 0.91 (0.0001) 0.11 (0.4) CE 0 + 2 0.62 (0.0001) 0.06 (0.62) [gamma] GT AST [r.sup.2] (P) [gamma] GT AST ALT AUDIT CE0 -0.12 (0.38) 0.04 (0.78) -0.08 ((0.58) 0.20 (0.32) CE2 -0.13 (0.33) -0.004 (0.98) -0.003 (0.83) 0.17 (0.45) CE 0 + 2 -0.10 (0.4) 0.02 (0.85) -0.05 (0.75) 0.17 (0.45) [gamma] GT 0.66 (0.0001) 0.38 (0.004) AST 0.64 (0.0001) (a) CE 0, asialo-Tf; CE 2, oligosialo-Tfs.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Proteomics and Protein Markers|
|Author:||Legros, Franz J.; Nuyens, Vincent; Minet, Eddy; Emonts, Philippe; Boudjeltia, Karim Zouaoui; Courbe,|
|Date:||Dec 1, 2002|
|Previous Article:||Detection of specific antinuclear reactivities in patients with negative anti-nuclear antibody immunofluorescence screening tests.|
|Next Article:||Structural diversity of cancer-related and non-cancer-related prostate-specific antigen.|