| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Department of Medicine and Public Health, Unit of Forensic Medicine, University of Verona, University Hospital (Policlinico), Verona, Italy;2 Department of Clinical Laboratory, Hospital of S. Bonifacio, Verona, Italy;3 Laboratory of Clinical Biochemistry, Hospital of Bolzano, Bolzano, Italy
aaddress correspondence to this author at: Department of Medicine and Public Health, Unit of Forensic Medicine, University of Verona, University Hospital (Policlinico), Piazzale L.A. Scuro, 37134 Verona, Italy; fax 39-045-8027623, e-mail franco.tagliaro{at}univr.it
Most assays for carbohydrate-deficient transferrin (CDT) (1)(2)(3) use cartridge extraction of CDT isoforms followed by immunoassay (4)(5). In 2001 the US Food and Drug Administration cleared the %CDT turbidimetric immunoassay (TIA; Axis Shield Plc) for detection of sustained and harmful alcohol use. Another method, based on anion-exchange HPLC separation of the CDT isoforms with direct detection at 460 nm [selective for the iron–transferrin (Tf) complex], was first developed by Jeppsson et al. (6) and adopted with minor changes by several authors, who reported advantages over immunoassays in terms of analytical selectivity, accuracy, and precision (7)(8)(9)(10). Recently released commercial reagents for HPLC analysis (Recipe®) offer advantages in interlaboratory analytical standardization.
Capillary zone electrophoresis (CZE) with dynamically coated capillaries and direct ultraviolet detection at 200 nm (11)(12)(13)(14)(15)(16)(17) has also been successfully applied to CDT determination in human serum (2), and a CZE method with a proprietary capillary coating is commercially available (CEofix® CDT; Analis).
Validation studies for each of these techniques have been reported in the literature, but no direct comparison studies have been published. We compared 3 commercial methods, based on TIA, CZE, and HPLC, focusing on their application to the diagnosis of chronic alcohol abuse in the forensic environment and in other areas requiring high diagnostic reliability and objectivity.
CDT immunoassays were performed with a commercial reagent set (%CDT TIA) from Axis Shield on a BN2 nephelometer (Dade Behring), according to the manufacturer’s instructions. The procedure included iron saturation followed by CDT extraction onto anion-exchange disposable cartridges and a TIA.
CZE was performed on a P/ACE MDQ capillary electropherograph (Beckman Coulter) with uncoated fused-silica capillaries [50 cm (total length) x 50 µm (i.d.)] obtained from Polymicro Technologies. The analytical method was based on commercial reagents (CEofix CDT; cat. no. 10-004760) providing a dynamic double coating of the capillary (with polycations and polyanions). Before injection the sample was saturated with a ready-to-use ferric solution provided by the manufacturer. Injection was carried out by pressure application (0.5 psi for 15 s). Separation occurred at 30 °C under an applied voltage of 28 kV, and detection was at a wavelength of 200 nm.
We performed analysis with a gradient HPLC system equipped with an ultraviolet-visible detector (Shimadzu Europe) set at a wavelength of 460 nm. The separation procedure was a modified version of the method reported by Jeppsson et al. (6). The analytical anion-exchange column [65 x 4.6 mm (i.d.)], the precolumn, and the mobile phase were provided in a commercial reagent set, ClinRep® CDT in Serum (Recipe). Sample preparation required iron saturation (with ferric solution) and lipoprotein precipitation with CaCl2.
For both CZE and HPLC, CDT isoform quantification (%CDT) was based on calculation of the percentage ratio of the (asialo-Tf + disialo-Tf) peak area to the sum of the peak areas for all isotransferrins from asialo-Tf to pentasialo-Tf. TIA results are reported as the CDT isoforms as a percentage of total serum Tf.
The cutoff between alcohol nonabusers and alcohol abusers was established at the 97.5th percentile from a group of 100 healthy blood donors, including social drinkers, who had no laboratory evidence of alcohol abuse or any alcohol-related disease and had a declared daily alcohol intake 
50 g. The resulting %CDT cutoff values were 2.6% for the TIA (which corresponded to the manufacturer’s recommendation), 1.8% for CZE, and 1.9% for HPLC.
The specimens analyzed in the present study were from 99 individuals undergoing mandatory control for certification to obtain a new driving license after theirs had been confiscated for driving under the influence of alcohol. Serum samples were stored frozen at –20 °C and analyzed within 3 days from collection.
We used parametric and nonparametric statistical methods (least-squares method, Mann–Whitney U-test, and Wilcoxon rank test) for evaluation of quantitative data.
Both instrumental techniques (HPLC and CZE) measured the CDT isoforms directly after very simple and standardized serum pretreatment, whereas TIA required ferric saturation and a delicate ion-exchange extraction on disposable cartridges, without any form of internal standardization to monitor recovery. Both HPLC and CZE provided excellent separation of Tf glycoforms (resolution factor between disialo-Tf and trisialo-Tf: HPLC, 1.50; CZE, 1.40) and lower limits of detection at least comparable to TIA. In terms of limit of quantification (lowest CDT concentration measurable with a relative SD 20%), the limits of quantification were 0.4% CDT for CZE and 0.3% CDT for HPLC. The imprecision of the methods (expressed as day-to-day relative SD) was 6.7% for TIA, 3.0% for CZE, and 2.9% for HPLC for samples near the respective cutoffs and 3.9% for TIA, 1.2% for CZE, and 1.2% for HPLC for positive samples (i.e., CDT concentrations nearly double the respective cutoffs). Because its detection wavelength (460 nm) corresponded to the maximum absorbance of the ferric iron, HPLC was more selective than CZE, which used ultraviolet absorbance at 200 nm. The CZE method was more rapid, requiring simpler sample preparation and a shorter overall analysis time (10 min for CZE; 36 min for HPLC), including separation, cleaning, and reconditioning of the system. Both methods provide graphic output showing the quality of the separations, and the output can be stored for further documentation. TIA includes a blind separation/extraction step, the quality of which relies mostly on manual/visual operation without any form of documentation, and an immunoassay using an anti-human Tf antiserum lacking any selectivity for the different CDT glycoforms. Thus, the overall specificity of the assay relies on the preliminary cartridge separation step.
The mean (SD) %CDT (range) in the 99 participants were 3.04 (1.47)% (1.3%–8.17%) with the TIA, 2.03 (2.07)% (0.54%–14.16%) with CZE, and 2.25 (2.27)% (0.45%–16.63%) with HPLC. The mean differences between results obtained with the TIA and both separation methods (TIA vs CZE, 1.01%; TIA vs HPLC, 0.79%), evaluated with the Mann–Whitney U-test, were highly significant (P <0.001), whereas the smaller difference between HPLC and CZE (0.22%) was not significant.
The overestimation of CDT concentrations in the TIA could tentatively be ascribed to interference from non-CDT Tf isoforms (trisialo- and/or tetrasialo-Tf), which may elute from the extraction column together with the CDT glycoforms (asialo- and disialo-Tf), as recently reported by Aldén et al. (18), or to nonspecific interferences of unknown origin.
On the other hand, the small difference between HPLC and CZE reflects the higher sensitivity of the former technique in the identification of asialo-Tf. This minor isoform of CDT, which is present in low concentrations, was detected by HPLC in 14 cases, whereas it was not detected by CZE. When this component of CDT was excluded from the calculation, the difference between the 2 techniques was negligible (0.06%).
We studied the correlation between the different methods, after excluding the results for the 2 patients with the highest measured CDT concentrations (12.7% and 16.6% by HPLC, 13.5% and 14.2% by CZE, and 8.17% and 10.1% by TIA), which were outside the range of values obtained for the other cases, to avoid any artifactual increase in the correlation coefficient. The least-squares method indicated a highly significant correlation (P <0.001) between HPLC and CZE (r = 0.9849; n = 97; Fig. 1A
). We also found statistically significant correlations between CZE and TIA (r = 0.7848; n = 95; P <0.001; Fig. 1B
) and HPLC and TIA (r = 0.8024; n = 95; P <0.001; Fig. 1C
), after excluding 2 outliers (0.81% and 0.44% by CZE, 1.02% and 0.73% by HPLC, and 6.36% and 8.88% by TIA). Nonparametric analysis (Wilcoxon rank test) also indicated a significant correlation for the 3 methods.
|
We performed a qualitative comparison of the results in terms of "positive-negative" findings, as often is required for forensic purposes, for the cutoffs described above (TIA, 2.6%; HPLC, 1.9%; CZE, 1.8%). Using these cutoffs, we classified the 99 samples as follows: TIA, 51 positives (35 concordant with HPLC, 32 with CZE) and 48 negatives (44 concordant with HPLC, 44 with CZE); HPLC, 40 positives, (36 concordant with CZE, 34 with TIA) and 59 negatives (59 concordant with CE, 44 with TIA); CZE, 36 positives (36 concordant with HPLC, 33 with TIA) and 63 negatives (59 concordant with HPLC, 45 with TIA).
We found excellent qualification concordance between HPLC and CZE, with only 4 discordant samples, which had values that were near the respective cutoffs. Comparison of TIA with HPLC yielded 16 false positives among 51 positives (31%) and 4 false negatives among 48 negatives (8%), and comparison of TIA with CZE yielded 19 false positives among 51 positives (37%) and 4 false negatives among 48 negatives (8%).
In addition, the TIA method was also susceptible to interference leading to false results in cases of D variants of Tf, which are present in 0.5%–1% of the population in our region. All of these variants were easily identified by either HPLC or CZE.
These data confirm the need for a systematic confirmation of TIA results with an independent technique, such as HPLC or CZE, which has already been documented (19). The nonnegligible occurrence of false negatives, however, casts doubt on the usefulness of the TIA for CDT to support a sensitive and specific diagnosis of alcohol abuse, particularly in forensic or administrative environments. The announced future commercial availability of multicapillary CZE instruments, presumably able to outperform the productivity of current CZE and HPLC systems, could make CZE an ideal screening method for CDT.
Acknowledgments
The present study was co-funded by Ministero dell’Istruzione, dell’Università, e della Ricerca (Research Grant PRIN 2003 2003060354) and by "Donazione Amleto Loro e Miria Cherubini".
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
