| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
-Glutamyltransferase for Detecting Problem Drinkers? A Systematic Review and Metaanalysis
1
Drug and Alcohol and
2
Biochemistry Departments, Royal Prince Alfred Hospital, Missenden Rd., Camperdown, NSW 2050, Australia.
a Address correspondence to this author at: Drug and Alcohol Department, Royal Prince Alfred Hospital, Missenden Rd., Camperdown, NSW 2050, Australia. Fax 61-2-9515-8970; e-mail katec{at}med.usyd.edu.au
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-glutamyltransferase (GGT), and compared the accuracy of different
CDT assay methods. Methods: We performed a systematic review using summary ROC analysis of 110 studies prior to June 1998 on the use of CDT in the detection of alcohol dependence or hazardous/harmful alcohol use.
Results: We identified several potential sources of bias in studies. In studies examining CDT and GGT in the same subjects, subject characteristics were less likely to influence the comparison. In such paired studies, the original Pharmacia CDT assay was significantly more accurate than GGT, but the modified CDTect assay did not perform as well as the original and was not significantly better than GGT. The accuracy of the AXIS %CDT assay was statistically indistinguishable from modified CDTect. Several CDT assay methods appeared promising, in particular, liquid chromatography (chromatofocusing, HPLC, fast protein liquid chromatography) and isoelectric focusing, but there were insufficient paired studies from which to draw firm conclusions.
Conclusions: In studies published before June 1998, the results obtained with commercially available CDT assays were not significantly better than GGT as markers of excessive alcohol use in paired studies. Further high-quality studies comparing CDTect (modified) and other CDT assays with GGT in the same subjects are needed.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Nonetheless, there remain conflicting data on the accuracy of CDT.
Reported sensitivities range from <20% to 100%, with typical
specificities of 75–100% (3)(4)(5)(6). It is difficult to
compare results from different studies because of the variety of assay
techniques used to measure CDT, the different cutoff points used to
define an abnormal result and to define excessive alcohol consumption,
and the differences in populations studied (ranging from hospitalized
alcoholics with liver disease to healthy volunteers). It is also
difficult to summarize the benefit CDT offers over the conventional and
less expensive measurement of -glutamyltransferase (GGT).
Furthermore, it is not readily apparent which technique for the
measurement of CDT is the most accurate.
Comparing assay methods when they have been assessed in different studies is potentially misleading because apparent differences may be attributable to the study population, study design, and biases rather than to true differences between assays. Therefore, the best evidence for comparing assays comes from "paired" studies that evaluate two or more assays using the same sample and the same reference standard.
We report the results of a systematic review of studies on the diagnostic value of CDT as a marker of excessive alcohol consumption published before June 1998. Our aim is to assist clinicians and researchers to better evaluate the available data on factors associated with the accuracy of CDT, to evaluate the relative accuracy of classes of CDT assay, and to compare CDT with GGT as a single marker of excessive alcohol consumption. Particular attention is paid to studies where two or more tests were compared in the same group.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In addition, published abstracts from several alcohol-related conferences up to May 1998 were perused, including all meetings of the International Society for Biomedical Research on Alcoholism and relevant proceedings from the European and Japanese Societies for Biomedical Research on Alcoholism, the Research Society on Alcoholism, and the North Carolina Alcoholism Research Authority.
data extraction
Reports were evaluated by one of the authors (K.S.), using a
structured form developed by the authors with reference to published
guidelines for the evaluation of diagnostic tests (7)(8)(9)(10)(11)(12)(13).
Information was collected on population characteristics, study design,
and assay methods, as shown in Tables 1–3
. Where there was uncertainty
over interpretation or coding of published data, the report was
referred to one (K.M.C.) or more authors for resolution. Eight studies
(7.5%) were independently reviewed by a second author (K.M.C.), and
there was >95% agreement on all items coded, with 100% agreement on
data relating directly to test accuracy.
|
|
|
statistical analysis
A descriptive analysis of the 110 eligible studies (derived from
108 reports) was conducted to determine the scope and nature of the
available data. Where possible, the sensitivity and specificity were
computed for each assay in each study (105 of 144 sets of test
results).
For any diagnostic test, there is a trade-off between sensitivity and specificity, so that if a lower threshold is used to define an abnormal test, there will be more true cases detected (higher sensitivity) but more false positives (lower specificity). The reverse is true if a higher threshold is used. Metaanalytic techniques for diagnostic tests have been developed that take into account this trade-off (14). These techniques are an extension of the conventional ROC analyses. A summary ROC (SROC) curve is estimated using the log odds ratio as a global measure of test performance for each study. The log odds ratio incorporates both sensitivity and specificity and takes into account the trade-off between them.
For each assay in each study the log odds ratio (D) is
computed as:
 |
where sens is test sensitivity and spec is specificity (both with values in the range 0–1). An odds ratio is then obtained by exponentiating the log odds ratio. Higher odds ratios reflect higher test accuracy.
To form a SROC curve for an assay, the method developed by Moses et al.
(14) was used to combine the log
odds ratios (D) for that assay across the different studies.
These values were combined by fitting the linear regression model:
D = 
+ ßS. In this formula,
S is a proxy for test threshold and is computed for each
study as S = logit (sens) + logit (1 -
spec). The parameter ß is used to assess whether overall
test accuracy depends on test threshold (i.e., the cutoff chosen to
define an abnormal test). None of our analyses supported the inclusion
of S in the model; hence, overall test performance was
assumed to be constant across a range of test thresholds (a symmetric
SROC) and was defined by a constant log odds ratio ().
The SROC curve can be displayed graphically by plotting the predicted
sensitivity across a range of values of 1 - specificity
(14).
Covariates, such as study design and gender, were also considered for inclusion in the regression model. Backward elimination was used to assess whether these variables were associated with test accuracy. Indicator variables were used to describe categories of study design (case control vs prospective) and gender (all male vs all female vs mixed or unknown). Models were unweighted to provide estimates consistent with a random-effects model (14)(15).
After we derived summary data on test accuracy for CDT and GGT from all of the studies, our subsequent analyses focused on paired studies, i.e., studies that compared different assays or different confounding factors, such as gender and liver disease, in the same subject group. For studies comparing two tests in the same subjects, we computed indicators of test accuracy (D) for each of the pair of assays in each study. We used a paired t-test to determine whether there was a significant difference in test accuracy between the two assays performed in the same subject group. The difference in test accuracy between the two assays was expressed as the difference in the mean log odds ratios for these assays, which can be interpreted as a ratio of odds ratios. This is analogous to the manner in which polytomous logistic regression uses the ratio of odds ratios to compare estimates (16).
We used the same approach to investigate whether the performance of an
assay was associated with the cutoff chosen for the reference standard.
For this analysis, we used studies that provided test results for three
or more alcohol use groups (e.g., a study that assessed the ability of
CDT to differentiate those drinking 60–79 g/day from light drinkers,
and those drinking 
80 g/day from light drinkers). Paired
t-tests were also used to examine the effect of gender and
liver disease on test performance by assessing within-study differences
where results were stratified for these factors.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Forty-six relevant published abstracts from 17 conferences up to 1998 were examined, and of these 31 (67%) had subsequently been published as journal articles. Because insufficient data were presented, no abstracts were included in the metaanalysis.
systematic review of journal articles
Table 1
provides an overview of the information provided in the 110
eligible studies. The median sample size was 101 subjects. The mean
subject age in >50% of studies was in the 40s (interquartile range,
41–48 years). When we averaged the unweighted proportions for each
study, the mean percentage of male subjects was 70% and of females was
30%. The majority of studies were based in Western Europe,
Scandinavia, and/or the US and Canada (37%, 35%, and 16%,
respectively) with smaller numbers of studies involving Australia and
New Zealand, Japan, and/or other countries (6%, 5%, and 2%,
respectively).
A variety of recruitment sites were used, with 82% of studies including patients with known alcohol problems and 63% involving community or staff members. Of the 88 studies in which a single blood sample from inpatients was analyzed, 50 (57%) reported that all or most venipunctures were performed within 72 h of admission. In five studies (6%), all or most blood samples were taken later than this, whereas 32 (36%) did not report on timing of venipuncture. In the additional nine studies (9%) in which blood was taken serially, the results closest to the time of admission were used for the metaanalysis. Time since last drink was given in 43 of the 96 (45%) studies involving inpatients, and in only 2 (5%) of these were samples taken more than 1 week after the last drink.
study quality
The quality of reports was variable, with the method of subject
selection documented in only 35% of studies, 
50% of studies
documenting gender or age comparability between cases and controls,
<50% documenting criteria for diagnosing alcohol abuse or dependence,
and <10% providing a statement about blinding in relation to test and
diagnosis (Table 2
). Stratification of results by age was provided in only 11% of
studies.
Subjects were most often recruited because they were known a priori to fulfill certain consumption or diagnostic criteria, but in 32% of studies subjects were first recruited, and then a consumption threshold was subsequently used to define consumption categories. Selection of consecutive patients was done in only 22% of studies, whereas 56% used a case-control design (i.e., tests were evaluated in a diseased population and a separate control group), which has been shown to be the most important source of bias in studies of diagnostic tests (13). However, despite this concern, the current data revealed no significant influence of study type (case-control vs population sampling) on the odds ratio estimate.
Studies generally used as their reference standards either the presence of usual self-reported alcohol consumption above a threshold value (e.g., mean alcohol consumption >60 g/day in the recent month or week) or the diagnosis of an alcohol use disorder. Alcohol dependence was more often used as a diagnostic category than was hazardous/harmful drinking or abuse (66% of studies compared with 26%). Criteria for the diagnosis were stated in 42% of such studies, whereas the method used to ascertain whether subjects fulfilled the criteria (e.g., structured interview or medical records review) was stated in only 27%.
In 56 studies (51%), both CDT results and GGT results were presented for the same subjects. However, data enabling full cross-classification of results were presented in only two of these studies, so that it was not generally possible to determine whether subjects with increased GGT were more or less likely to have increased CDT.
comparison between different alcohol consumption thresholds
When alcohol consumption was used as a reference standard,
subjects were generally divided into two or three categories, usually
(a) abstainers and light or moderate drinkers, whose upper
limit of consumption was up to 60.0 g of ethanol/day (mean, 25
g/day); (b) hazardous or abusive drinkers, whose upper limit
of consumption ranged from 40 to 143 g/day (mean, 75 g/day); and
(c) heavy, harmful, or dependent drinkers, whose consumption
was higher than that of subjects in other categories.
To determine whether use of different consumption thresholds influenced test accuracy, the nine reports that used three consumption categories were examined. These studies, which provided test results for both a higher and lower threshold, were used to calculate SROC curves for both thresholds. Because the two thresholds were compatible with the same curve (t8 = 0.73; P = 0.48), the cutoff at or closest to 40 g of alcohol/day was chosen for further analyses.
influence of gender, age, and liver disease on cdt concentrations
On unpaired analysis, gender was not found to have a significant
effect on the performance of most CDT assays or of GGT. It should be
noted that studies generally used different test thresholds for men and
for women. The performance of only one CDT assay [HPLC or fast protein
liquid chromatography (FPLC)] was associated with gender
(P = 0.05).
Sixteen studies with a median sample size of 85 (interquartile range,
71–147) presented stratified CDTect
(modified)2
results for males and females. A paired t-test
revealed no statistically significant difference in test accuracy
between the genders (t15 = -0.65;
P = 0.53; Fig. 1
A).
|
Because there were only 12 studies that provided stratified results by age and these studies used different assay techniques, there were not sufficient data to perform paired comparisons for any one CDT assay by age category.
Five studies with a median sample size of 108 (interquartile range,
82–112) presented stratified CDTect (modified) results for subjects
with and without liver disease. CDTect performed significantly better
in patients without liver disease than in those patients with liver
disease (t4 = 4.25; P
= 0.01; Fig. 1B
).
performance of different assays and comparisons between
assays
Categorization of assay methods.
CDT assay methods were
divided into seven categories (Table 3
). Where an assay did not fit the major categories, the assay
type was recorded as "other" and the assay category that it most
closely resembled was noted. The accuracy of the less common assay
method was then compared with the major assay category using SROC
analysis. In all such cases there was no statistically significant
difference in performance, so the less common assay was grouped with
that major category for further analyses.
When we derived SROC curves for each CDT assay method and for GGT, none of the slopes of D on S were significantly different from zero, indicating that the SROC curves were defined by a constant odds ratio that could be used as an overall measure of test accuracy.
Unpaired comparisons.
The odds ratio from SROC
analysis for each CDT assay method and for GGT are presented in Table 3
. All of the CDT methods yielded higher summary odds ratios than GGT,
although for the
AXIS3
method and CDTect (modified), the confidence intervals
overlapped and touched, respectively, the confidence interval for GGT.
The odds ratios suggest that the FPLC/HPLC methods, chromatofocusing
(the latter was used in only two studies), CDTect
(original),4
and isoelectric focusing (IEF) methods have higher odds ratios
than GGT.
Paired comparisons.
Where there were five or more studies
providing stratified results, paired comparisons between assay types
were performed:
and Fig. 2A
). In contrast, in the 19 studies where CDTect (modified) and
GGT were assayed in the same subjects, the performance of CDTect
(modified) did not differ significantly from that of GGT (Table 4
and
Fig. 2B
). IEF methods appeared to perform somewhat better than GGT
(paired analysis with seven studies), but this difference was not
statistically significant (Table 4
and Fig. 2C
). Data were not
available to perform paired comparisons between HPLC/FPLC methods and
GGT.
and
Fig. 2D
).
).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In studies on the accuracy of biological markers of alcohol use, the reference standard generally involves self-report of alcohol consumption, symptoms of dependence, or related problems. Studies vary in the criteria used to determine problem drinking. Some misclassification of subjects is likely to occur, which is likely to cause an underestimation of test accuracy (27). However, identification of whether one test is better than another is not likely to be biased when the tests are compared within the same patient.
In studies where CDT and GGT were measured in the same subjects, the original CDTect method was shown to be significantly more accurate than GGT. In contrast, CDTect (modified), which is one of the commercially available methods for measuring CDT, offered no significant benefit over GGT. The accuracy of the other commercially available method, AXIS %CDT, did not differ significantly from that of CDTect (modified).
Other assay types showed considerable promise on unpaired comparison, in particular chromatofocusing and HPLC/FPLC methods. The odds ratio for HPLC/FPLC methods was 20-fold higher than that for GGT in unpaired comparisons, which translates into a sensitivity of 93% for HPLC/FPLC compared with 39% for GGT at test cutoffs that give a specificity of 90% for both. However, insufficient reports were available to compare chromatofocusing or HPLC/FPLC with GGT in the same subjects, so it remains possible that the difference in test accuracy has been exaggerated by differences in subject characteristics or study methods. IEF methods were significantly more accurate than CDTect (modified), but they were not demonstrated to be significantly more accurate than GGT on paired comparison.
Differences in the accuracy of CDT assay methods may result from differences in the ways assays treat the isotransferrins, particularly trisialotransferrin. This is the major difference between the original and modified Pharmacia methods (28) and illustrates the impact that variation in an assay method and hence in the analyte can have on test performance. In a similar way, immunoassays may differ in their clinical performance through measurement of different epitopes, of C- and N-terminal fragments, or of different subunits of a multimeric protein.
effect of gender and liver disease
In studies that reported separate results for CDTect (modified) in
males and females, gender did not influence test accuracy. These
studies generally used different test reference intervals for men and
for women. There were insufficient data to perform a paired analysis
for gender in other CDT assays, but it appears likely that the
establishment of different test thresholds adequately compensates for
differences in baseline CDT concentrations.
A major limitation of GGT is that it is affected by liver disease of any cause and by several medications. However, where CDTect (modified) results were stratified by liver disease, test performance was significantly reduced in the presence of liver disease. There were insufficient paired data to establish whether CDT is more accurate than GGT in this setting.
practical implications
Measurement of CDT by most methods (with the possible exception of
the AXIS method) costs at least twice as much as measurement of GGT,
and from the evidence of this study, routine use of commercial CDT
testing as a single test for alcohol consumption is not indicated.
CDTect (original) was significantly more accurate than CDTect
(modified) or GGT; however, practical constraints may limit its
clinical use. CDTect (original) was modified to form the new
commercially available test, CDTect (modified), because of buffer
instability. It remains a challenge to maintain the test performance of
the original assay in a form that is readily available and easily used.
Noncommercial CDT assay methods may be more accurate than the
commercial methods, but their use may stretch the resources of many
clinical laboratory facilities.
There may be situations where CDT testing by the commercially available assays is likely to be more accurate than GGT assay, e.g., in subjects using medications such as antiepileptics, where GGT would be more likely to produce false-positive results. Furthermore, there have been promising reports on the use of CDT in combination with GGT, where the combined tests have increased sensitivity without significant compromise in specificity (3)(18)(29)(30). CDT also shows promise in prospective monitoring of an individual’s drinking (31)(32)(33)(34)(35), but these aspects of CDT use were outside the scope of the current report.
In conclusion, there is a tendency to enthusiastically adopt new tests that show promising results. The findings of the current study emphasize the need for ongoing and careful assessment of the evidence and for high-quality research. The original Pharmacia CDT assay, which is not now commonly used, was the only CDT assay method shown to be significantly better than GGT on paired testing. CDT as measured by commercially available assay was not shown to be a significantly better test than GGT when performed on the same subjects. Although the use of noncommercial assay methods such as chromatofocusing and HPLC/FPLC may be more accurate than commercial methods, these may be feasible only in laboratories with a high volume of testing. Further high-quality studies using comparisons of tests in the same subjects are indicated.
|
Acknowledgments |
|---|
of Contents
(http://www.clinchem.org/content/vol46/issue12). |
Footnotes |
|---|
1 Nonstandard abbreviations: CDT, carbohydrate-deficient transferrin; GGT, -glutamyltransferase; SROC, summary receiver-operating characteristic; FPLC, fast protein liquid chromatography; and IEF, isoelectric focusing. 
2 Pharmacia CDTect assay, modified method, subsequently supplied by AXIS.
3 CDT isoforms as a proportion of total transferrin (%CDT), supplied by AXIS Biochemicals AFA, Oslo, Norway.
4 Pharmacia CDT assay (Pharmacia and Upjohn Diagnostics AB, Uppsala, Sweden), based on original method by Stibler et al. (25).
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-glutamyltransferase as markers of heavy alcohol consumption: gender differences. Alcohol Clin Exp Res 1994;18:747-754.[Web of Science][Medline]
[Order article via Infotrieve]
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  H. Tonnesen, P. R. Nielsen, J. B. Lauritzen, and A. M. Moller Smoking and alcohol intervention before surgery: evidence for best practice Br. J. Anaesth., March 1, 2009; 102(3): 297 - 306. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. P. Bergstrom and A. Helander Clinical Characteristics of Carbohydrate-Deficient Transferrin (%Disialotransferrin) Measured by HPLC: Sensitivity, Specificity, Gender Effects, and Relationship with other Alcohol Biomarkers Alcohol Alcohol., July 1, 2008; 43(4): 436 - 441. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. B. Whitfield, V. Dy, P. A.F. Madden, A. C. Heath, N. G. Martin, and G. W. Montgomery Measuring Carbohydrate-Deficient Transferrin by Direct Immunoassay: Factors Affecting Diagnostic Sensitivity for Excessive Alcohol Intake Clin. Chem., July 1, 2008; 54(7): 1158 - 1165. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. HIETALA, H. KOIVISTO, P. ANTTILA, and O. NIEMELA COMPARISON OF THE COMBINED MARKER GGT-CDT AND THE CONVENTIONAL LABORATORY MARKERS OF ALCOHOL ABUSE IN HEAVY DRINKERS, MODERATE DRINKERS AND ABSTAINERS Alcohol Alcohol., September 1, 2006; 41(5): 528 - 533. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Koch, G.-J. Meerkerk, J. O. M. Zaat, M. F. Ham, R. J. P. M. Scholten, and W. J. J. Assendelft ACCURACY OF CARBOHYDRATE-DEFICIENT TRANSFERRIN IN THE DETECTION OF EXCESSIVE ALCOHOL CONSUMPTION: A SYSTEMATIC REVIEW Alcohol Alcohol., March 1, 2004; 39(2): 75 - 85. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Anttila, K. Jarvi, J. Latvala, and O. Niemela METHOD-DEPENDENT CHARACTERISTICS OF CARBOHYDRATE-DEFICIENT TRANSFERRIN MEASUREMENTS IN THE FOLLOW-UP OF ALCOHOLICS Alcohol Alcohol., January 1, 2004; 39(1): 59 - 63. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Chen, K. M. Conigrave, P. Macaskill, J. B. Whitfield, and L. Irwig COMBINING CARBOHYDRATE-DEFICIENT TRANSFERRIN AND GAMMA-GLUTAMYLTRANSFERASE TO INCREASE DIAGNOSTIC ACCURACY FOR PROBLEM DRINKING Alcohol Alcohol., November 1, 2003; 38(6): 574 - 582. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Anttila, K. Jarvi, J. Latvala, J. E. Blake, and O. Niemela DIAGNOSTIC CHARACTERISTICS OF DIFFERENT CARBOHYDRATE-DEFICIENT TRANSFERRIN METHODS IN THE DETECTION OF PROBLEM DRINKING: EFFECTS OF LIVER DISEASE AND ALCOHOL CONSUMPTION Alcohol Alcohol., September 1, 2003; 38(5): 415 - 420. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. Schwarz, I. Domke, A. Helander, P. M. W. Janssens, J. van Pelt, B. Springer, M. Ackenheil, K. Bernhardt, G. Weigl, and M. Soyka MULTICENTRE EVALUATION OF A NEW ASSAY FOR DETERMINATION OF CARBOHYDRATE-DEFICIENT TRANSFERRIN Alcohol Alcohol., May 1, 2003; 38(3): 270 - 275. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. J. Legros, V. Nuyens, M. Baudoux, K. Zouaoui Boudjeltia, J.-L. Ruelle, J. Colicis, F. Cantraine, and J.-P. Henry Use of Capillary Zone Electrophoresis for Differentiating Excessive from Moderate Alcohol Consumption Clin. Chem., March 1, 2003; 49(3): 440 - 449. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. B. Whitfield Transferrin Isoform Analysis for the Diagnosis and Management of Hazardous or Dependent Drinking Clin. Chem., December 1, 2002; 48(12): 2095 - 2096. [Full Text] [PDF]
|
|
|
|
|
|
F. J. Legros, V. Nuyens, E. Minet, P. Emonts, K. Z. Boudjeltia, A. Courbe, J.-L. Ruelle, J. Colicis, F. de L'Escaille, and J.-P. Henry Carbohydrate-deficient Transferrin Isoforms Measured by Capillary Zone Electrophoresis for Detection of Alcohol Abuse Clin. Chem., December 1, 2002; 48(12): 2177 - 2186. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Arndt and J. Kropf Alcohol Abuse and Carbohydrate-deficient Transferrin Analysis: Are Screening and Confirmatory Analysis Required? Clin. Chem., November 1, 2002; 48(11): 2072 - 2074. [Full Text] [PDF]
|
|
|
|
|
|
F. Tagliaro, F. Bortolotti, R. M. Dorizzi, M. Marigo, J. R. Delanghe, B. Wuyts, and M. L. De Buyzere Caveats in Carbohydrate-deficient Transferrin Determination Drs. Delanghe, Wuyts, and De Buyzere respond: Clin. Chem., January 1, 2002; 48(1): 208 - 209. [Full Text] [PDF]
|
|
|
|
|
|
A. Reif, H. Keller, M. Schneider, S. Kamolz, A. Schmidtke, and A. J. Fallgatter CARBOHYDRATE-DEFICIENT TRANSFERRIN IS ELEVATED IN CATABOLIC FEMALE PATIENTS Alcohol Alcohol., November 1, 2001; 36(6): 603 - 607. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J. McQueen Overview of Evidence-based Medicine: Challenges for Evidence-based Laboratory Medicine Clin. Chem., August 1, 2001; 47(8): 1536 - 1546. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
