| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Klinisch-Chemisches Laboratorium, Kantonsspital Basel, Switzerland.
2
Abteilung Klinische Chemie, Klinikum Universität
Freiburg, Germany.
3
Abteilung Klinische Chemie, Zentrum Innere Medizin,
Georg-August-Universität Göttingen, Germany.
4
Institut für Laboratoriumsdiagnostik,
Zentralkrankenhaus, Gauting, Germany.
5
Institut für Klinische Chemie, Klinikum Mannheim,
Germany.
6
Abteilung Klinische Chemie und Hämatologie,
Zentrum Geburtshilfe und Gynäkologie, Universität Bonn,
Germany.
7
Hôpital Universitaire Brugmann, Bruxelles,
Belgium.
8
Boehringer Mannheim GmbH, Mannheim, Germany.
a Address correspondence to this author at: Klinisch-Chemisches Laboratorium, Petersgraben 4, CH-4031 Basel, Switzerland. Fax 0041-61-265-4600; e-mail scholer{at}ubaclu.unibas.ch
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
a Digoxin, Boehringer Mannheim) was
evaluated in seven laboratories. It can be performed without sample
pretreatment with ready-to-use reagents on nondedicated analyzers in
combination with routine clinical chemistry. The studies revealed a
good analytical performance: lower limit of detection 0.12 µg/L (3 SD
from mean of blank); linearity up to 7.5 µg/L; median between-run CVs
8.1% (0.6 µg/L), 2.8% (1.5 µg/L), 1.9% (3 µg/L); mean
analytical recovery in control sera 98–102%; slopes from 0.97 to 1.09
and intercepts from -0.28 to 0.10 µg/L in comparison with four
immunoassays; and a high resistance to common interferents. The test
was more resistant to digoxin-like immunoreactive factor (DLIF)
interference than other methods, showing cross-reactivity only in some
intensive care patient samples. Among 192 patients in whom DLIF is
expected (e.g., pregnant women, patients with renal failure, newborns),
90% of results were 
0.26 µg/L digoxin. Cortisol showed no
cross-reactivity and digoxigenin had a low reactivity. An
interlaboratory survey revealed a good comparability of the Tina-quant
a test with the median of all methods (slope 0.99,
intercept -0.06 µg/L). An HPLC method for digoxin based on isocratic
separation of samples on an RP-18 column followed by detection by an
immunoassay yielded a reasonable comparability with the immunochemical
tests with noncritical samples. Divergent results of immunoassays
caused by DLIFs or different cross-reactivities with digoxin
metabolites or derivatives can be explained by the use of this HPLC
method.
Key Words: indexing terms: turbidimetry • immunoassay • methods comparison • liquid chromatography
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
Automation can help overcome some of the analytical problems
encountered in the determination of serum digoxin. The majority of
methods available are not fully automated and require a pretreatment
step that by itself represents a source of variation (6).
Here we report on the multicenter evaluation of an immunoturbidimetric
digoxin assay (Tina-quant® a Digoxin)
involving monoclonal antibodies that can be performed without any
pretreatment of samples. This assay, as well as other immunological
assays, has been compared with a recently developed HPLC method, which
is a useful tool to explain discrepancies in the results obtained by
different digoxin immunoassays.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
a Digoxin assay (Boehringer Mannheim,
Mannheim, Germany) involves a competitive technique, the TINIA
principle (turbidimetric inhibition
immunoassay). In the first reaction step, digoxin
in the sample and monoclonal anti-digoxin antibodies form soluble
immune complexes. In the second reaction step, the antibodies remaining
from the first reaction then form insoluble precipitates with digoxin
bound to latex particles, which are measured by an increase in
absorbance at 700 nm. The turbidimetric signal is hence inversely
proportional to the concentration of digoxin in the sample. The assay
allows the direct determination of digoxin in serum or plasma without
sample pretreatment on routine clinical chemistry analyzers such as
Boehringer Mannheim (BM)/Hitachi analysis systems. Only 15 µL of
sample is required on BM/Hitachi 717. The reagents are ready-to-use.
The test is performed by endpoint measurement with sample blank. A
six-point mode of calibration with ready-to-use calibrators is used.
The measuring temperature is 37 °C.
The HPLC method for digoxin is based on isocratic separation of samples
on an RP-18 column. The individual fractions are collected and tested
for digitalis glycosides by immunoassay. A typical example of the
separation is shown in Fig. 1
. Between-run CVs determined over 12 days range from 15% to
20%. The analytical recovery in the concentration range between 0.5
and 4 µg/L is ~65%. The measuring range extends from 0.39 to 3.9
µg/L. A brief description of the method follows:
|
Equipment.
Equipment included a pump solvent delivery
system [Varian (Palo Alto, CA) 9001], a variable UV-detector (Varian
2550), and an automated sample preparer (AASP; Varian).
Stationary phase.
The stationary phase was
RP-18 spheri 5 µm, 220 x 4.6 mm (Brownlee OD-224) with
precolumn RP-18 Newguard 7 µm, 15 x 3.22 mm (Brownlee G
18–013)
Mobile phase.
The mobile phase included
520 mL/L methanol (Baker, Phillipsburg, NJ), 30 mL/L ethanol p.a.
(Merck, Darmstadt, Germany), 10 mL/L 2-propanol p.a. (Merck), and 440
mL/L water. The flow rate was 0.6 mL/min; 1-min fractions (0.6 mL) were
collected.
Sample preparation.
Samples are prepared
with the aid of the semiautomated sample preparation system, AASP
(solid-phase extraction C2). The solid phase is conditioned with
methanol and water; sample volume is 0.6 mL; wash solution is 20 mL/L
acetonitrile in water; elution directly by the mobile phase; adjustment
for AASP preinjection purge 0, after-injection purge 40 (Millipore,
Bedford, MA); valve reset 5 min; purge solvent 500 mL/L methanol
(Baker), 500 mL/L water
Detection.
Direct analysis of HPLC
fractions was by enzyme-multiplied immunoassay (Emit Convenience Pack
Digoxin; Behring Diagnostics, Palo Alto, CA) on a Cobas®
Fara (Hoffmann-La Roche, Basel, Switzerland) analyzer.
Calibration.
The calibrators of the Emit
test were treated like the samples and used as reference.
evaluation of the tina-quant a digoxin test
The test was evaluated according to a standardized protocol in
seven clinical laboratories on different BM/Hitachi analysis systems
(one BM/Hitachi 704, five BM/Hitachi 717, one BM/Hitachi 911).
Different control sera as well as human sera were used in the studies
of assay evaluation.
The limit of detection was determined by two different approaches. First, 9 g/L NaCl solution was measured in six independent runs (21 replicates per run) and the detection limit was determined as 3 SD of the mean of these measurements. Second, 192 specimens from critical patient groups not receiving digoxin and suspected of containing DLIF were measured, and the data were analyzed by the calculation of nonparametric fractile limits. The linearity of the test was analyzed by dilution of a serum sample supplemented with digoxin with either analyte-free human serum or with isotonic saline according to a recently proposed protocol (7). The method was considered linear if the measured concentration deviated <5% from the expected concentration.
Precision studies were carried out according to two different protocols with control materials or human sera. Within-run precision (21 replicates per run) as described in the ECCLS guidelines (8) was determined in five laboratories, and between-run precision (two aliquots per sample on 20 days) in four laboratories. The NCCLS EP5-T protocol (9) for determination of within-run, between-run, and total precision was performed twice in one center on 20 days.
The recovery of assigned digoxin values was analyzed in three-level control sera from different manufacturers (controls 1, 2, 3, Precinorm® TDM, Boehringer Mannheim; and controls 4, 5, 6, Lyphochek® control materials, BioRad, Munich, Germany).
Interferences by endogenous metabolic products were tested according to Glick et al. (10) by supplementing aliquots of human pooled sera (0.49–0.64 µg/L digoxin) with different amounts of interfering agents. In addition, the effect of rheumatoid factors, serum proteins, and the anticoagulants heparin and EDTA was checked in different laboratories. The influence of 31 therapeutic drugs was investigated as previously described (11).
method comparison studies between hplc as reference method
and various immunoassays
Method comparison studies were performed in each laboratory with
routine specimens including hemolytic, icteric, lipemic, and
analyte-free samples. The Tina-quant a Digoxin test was
compared with routine methods established at the evaluation sites as
fluorescence polarization immunoassay (FPIA on TDx analyzer and FPIA
direct on AxSYMTM analyzer; Abbott Diagnostics,
Abbott Park, IL), fluorescence immunoassay (FIA on Stratus analyzer;
Baxter, McGaw Park, IL), Emit, RIA (Diagnostic Products Corp., Los
Angeles, CA), CEDIA® assay (CEDIA SC reagent on
BM/Hitachi 717 analyzer, Boehringer Mannheim), and two enzyme
immunoassays (EIA 1 on aca discrete analyzer; DuPont,
Wilmington, DE; EIA 2 Enzymun-Test Digoxin on Enzymun-Test Systems ES
300 and 600, Boehringer Mannheim). The HPLC method for digoxin
described above was carried out in addition to the routine methods.
Digitoxin was measured in one laboratory by using a FPIA on the TDx analyzer (Abbott Diagnostics). The tests were performed as described by the manufacturers. Regression analysis was carried out according to Passing and Bablok (12).
interlaboratory survey
Seven laboratories participated in the interlaboratory survey. Up
to 43 human sera (either from individual patients or pooled sera;
patients from the intensive care unit and samples supplemented with
digoxin or digoxin derivatives were included) with digoxin
concentrations covering the relevant measuring range and 10
analyte-free sera were shipped to the participants in frozen aliquots
on dry ice. Samples from patients undergoing hemodialysis were excluded
from the interlaboratory survey because they are known to contain
DLIFs, which exhibit a method-dependent cross-reactivity
(3)(13)(14). The concentration of
digoxin in the samples was determined by immunological routine methods
and by HPLC. All results of an individual sample determined by all
participants with a single method were calculated to obtain the overall
median (three or more participants) or the mean (two participants)
digoxin value. These values were compared with the median values
calculated from the results of all methods by the Passing–Bablok
regression analysis (12).
|
Results and Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
a digoxin
testa Digoxin assay
established in the traditional way by analyzing blank specimens was
0.12 µg/L. Alternatively, the limit of detection was determined in a
more practical approach by measuring apparent digoxin values in five
groups of individuals without digoxin treatment and suspected to
contain DLIFs (n = 192; pregnant women, newborns, patients from
the intensive care unit, patients with renal failure, and hemodialysis
patients). Concentrations from 0 to 0.86 µg/L were obtained, with a
median of 0.07 µg/L. Taking the 90th percentile as a statistical
measure to describe the detection limit of the new turbidimetric test,
we obtained 0.26 µg/L. This value is far below the lower limit of the
therapeutic range of digoxin. An upper limit of linearity of 7.5 µg/L was found even with reagents stored uncovered on the BM/Hitachi analyzer for 30 days (data not shown). Dilution of the highly concentrated sample with either digoxin-free human sera or physiological saline resulted in comparable results, thereby showing that the test is insensitive to matrix effects. The therapeutic range recommended for digoxin therapy is 0.8 to 2.0 µg/L. Thus, the observed range of linearity is more than sufficient for clinical purposes, and rerun of samples is limited.
The reagents are ready-to-use. They are stable for 60 days under
routine conditions (~10 °C, open reagent bottles on the analyzer).
The calibration curve can be used for at least 30 days. Calibration
curves obtained in 23 calibration runs showed an excellent
reproducibility (Fig. 2
).
|
The results of the precision studies are summarized in Table 1
. Comparable CVs for within-run and between-run precision were
obtained by using the two different evaluation protocols. Similar
within-run CVs were found with human sera and control materials. In the
therapeutic range, total precision (CV) ranged from 3.8% to 10.1%
(data not shown). Comparable CVs have been reported for other digoxin
methods (5)(15)(16)(17)(18).
|
The studies on the recovery of digoxin values assigned to different
control sera are summarized in Table 2
. The mean recovery for all control sera including one with a
rather low digoxin value (target value 0.64 µg/L) ranged from 98% to
102%.
|
No interferences occurred up to 690 mg/L bilirubin, 20 g/L hemoglobin,
and 1% of a 20% Intralipid® solution, which
corresponds to 9 g/L (10.3 mmol/L) triglycerides. With lipemic native
human sera, no interference by turbidity was observed up to 15.95 g/L
triglycerides (Fig. 3
).
|
Rheumatoid factors did not interfere with the Tina-quant
a Digoxin assay up to an activity of 100 kIU/L and total
protein up to 100 g/L. The comparison between serum and EDTA plasma as
well as Li-heparinate plasma revealed median deviations of -0.02
µg/L and -0.05 µg/L digoxin, respectively. Thus, both types of
plasma can be recommended as sample material.
Serum samples supplemented with 31 drugs in concentrations above the
therapeutic range including ascorbic acid had no influence on the
results of the Tina-quant a Digoxin assay (data not shown).
comparison of specificity
The sensitivity of different digoxin tests to cross-reactivity
with DLIF was tested with five groups of individuals (Table 3
). All assays studied exhibited only minor cross-reactivity in
sera from pregnant women. The Tina-quant a Digoxin test
was very resistant to DLIF interference in samples from newborns. In
contrast, a high number of results above the detection limit was
obtained with the FPIA and particularly with the CEDIA test, which
yielded nine values located within the therapeutic range. A high
frequency of occurrence of apparent digoxin values was found for all
three assays for patients from the intensive care unit, the measured
concentrations being the lowest for the Tina-quant a test.
For the FPIA and CEDIA, >20% of results were located within the
therapeutic range. The magnitude of DLIF interference in patients with
renal failure and in a special collective of hemodialysis patients was
analyzed with the Tina-quant a test and the direct FPIA.
Also in these patients the new turbidimetric test was more resistant to
DLIF interference than the comparison method. In accordance with
previous studies (5)(19), cross-reactivity
with the same specimens varied considerably between different assays.
|
The therapeutic range of digitoxin is ~10-fold that of digoxin. Thus, one must be aware of cross-reactivity of this drug in digoxin assays in countries such as Germany, where digitoxin is frequently prescribed. To assess this type of cross-reactivity, both drugs were determined in routine sera of 207 digitalis-treated patients in one laboratory. Of these, 24 samples contained digitoxin (range 5.7–32.0 µg/L) determined by FPIA. In all of these samples, positive digoxin values (range 0.8–2.4 µg/L) were obtained by the FPIA for digoxin determination. The new turbidimetric digoxin test exhibited a similar sensitivity to cross-reactivity with digitoxin (24 samples positive, range 0.8–2.0 µg/L). Digitoxin had been requested by the treating physician in only 11 of the 24 patients with positive digitoxin values. This investigation clearly demonstrates that further efforts must be undertaken to increase the specificity of digoxin tests against digitoxin.
A special case of cross-reactivity of digoxin assays with cortisol was reported from one laboratory. One patient under digitalis therapy who received additional treatment with hydrocortisone exhibited erroneously high digoxin values, up to 7.4 µg/L with one of four digoxin assays studied. A cross-reactivity with cortisol of 0.12% was given in the package insert of this assay, which clearly is too high if one considers that cortisol is administered in the 1000-fold concentration in comparison with digoxin. In this patient, digoxin values in the therapeutic range were obtained with the new turbidimetric assay (1.4 µg/L), FIA (1.3 µg/L), and EIA (1.9 µg/L).
Samples suspected of giving discrepant results were measured in one
laboratory with HPLC and five immunochemical methods (Table 4
). Similar results were found with all methods in the sample
supplemented with digoxin and in the sample from a digoxin-treated
patient. ß-Methyldigoxin is often administered in some countries.
Digoxin is formed from this compound in vivo. Reliable results were
obtained with HPLC in sample no. 3 (supplemented with
ß-methyldigoxin) and in sample no. 4 (originating from a
digitalis-treated patient), which contained a mixture of digoxin and
ß-methyldigoxin. In contrast, divergent results were found in this
sample with the immunochemical tests, which can be explained by
different cross-reactivities with ß-methyldigoxin. The largest
discrepancies ranging from values below the detection limit up to 1.05
µg/L were obtained with sample no. 5 from a hemodialysis patient not
receiving digoxin therapy. The HPLC method is particularly suited for
explanation of these erroneous results caused by the presence of DLIF.
|
Cross-reactivity of digoxin metabolites in commercially available
digoxin immunoassays has been compared with the pharmacological
activity of these substances in an interesting study by Miller et al.
(20). This prompted us to analyze a sample supplemented
with digoxigenin (no. 6), a digoxin metabolite that has ~10%
bioactivity compared with digoxin (21). Very high apparent
digoxin values were found with three immunoassays in this sample, which
by far exceeded the amount added (Table 4
). Digoxigenin as well as an
additional compound, which could not be identified, were detected by
HPLC, thereby indicating the presence of an impurity that seemed to
react unspecifically in some tests. In agreement with the
pharmacological activity of digoxigenin, low apparent digoxin
concentrations were found with the new turbidimetric test in sample no.
6. In addition, this test was most resistant to interference from the
unknown impurity.
method comparison studies
The results of the method comparison studies performed with fresh
routine specimens are summarized in Table 5
. The slopes were within a range of ±9% deviation except for
the comparison between the immunoturbidimetric test and FIA (slope
0.84), which had only been carried out in one laboratory. The
intercepts ranged from -0.28 to 0.10 µg/L digoxin. Comparable slopes
and intercepts between different digoxin assays were obtained in other
studies (17)(18).
|
Two representative examples of the method comparison studies are shown
graphically in Fig. 3
. Lipemic samples are marked in the comparison between the
Tina-quant a test and FPIA. Comparable results were
obtained even in samples with triglyceride values up to 16 g/L (18
mmol/L), thereby indicating that the turbidimetric test is not
interfered with by lipemia in native human sera.
Taking into account the different analytical procedures of HPLC and
immunoturbidimetry, the results obtained with HPLC in comparison with
the Tina-quant a test can be judged as agreeing well. No
major discrepancies between the two methods were found. Larger
deviations can be explained by the higher imprecision of the HPLC
method (CV 15–20%), the lower number of samples analyzed, and the
limited concentration range of samples.
interlaboratory survey
Digoxin concentrations were determined in up to 43 human sera in
seven different laboratories with the Tina-quant a Digoxin
test, HPLC, and the routine methods shown in Table 6
. The digoxin values obtained by the Tina-quant
a test compared well with the overall median of the other
tests. This indicates a good interlaboratory transferability of the new
immunoturbidimetric assay. Most of the other methods studied showed an
acceptable comparability with the median of all methods, with slopes
deviating less than ±10% and intercepts between ±0.1 µg/L digoxin.
Larger systematic deviations were obtained for the FIA and especially
for the RIA, which indicate differences in the standardization of these
assays.
|
The results obtained with HPLC were in reasonable agreement with those
obtained with immunochemical assays (Table 6
). The low correlation
coefficient of 0.8 and the high scatter can be explained by the
imprecision of the HPLC method and the small number of samples tested.
In addition, discrepancies may be caused by the presence of
ß-methyldigoxin and by inclusion of patient samples from the
intensive care unit that might contain DLIF. As shown previously in
Table 4
, digoxin derivatives or DLIF, which can be correctly detected
by HPLC, may give discrepant results in immunochemical assays.
Ten analyte-free samples not suspected to contain DLIF were included in the interlaboratory survey. No results above the detection limits declared for the different assays were obtained (data not shown).
routine application and practicability
The Tina-quant a Digoxin test can be performed
without sample pretreatment on nondedicated random access analyzers
used in routine clinical chemistry. This concept was judged positively
by the participants of the multicenter study who had been asked to rate
the practicability of the test. Handling of samples and of the
calibration, labor intensity, and integration into routine work were
evaluated particularly positively in comparison with the currently
performed routine method.
The majority of digoxin tests must be carried out on dedicated batch analyzers or manually. On routine clinical chemistry analyzers such as BM/Hitachi analysis systems, all tests necessary for a complete patient profile can be run out of one primary tube. This is especially attractive for emergency and for low-volume (e.g., pediatric) samples. Advantages in terms of the preanalytical phase and in terms of costs are obvious. Hazards and dangers of sample mix-up are reduced. In rare cases the new digoxin test may cause falsely increased results in a low number of other tests on BM/Hitachi analysis systems. This can be avoided by using evasion software.
Others have speculated that drug tests produce more erroneous results
if they are implemented on nondedicated analyzers
(22)(23). The analytical performance of the
immunoturbidimetric digoxin test described in this study does not
support this speculation. Precision and accuracy of the
Tina-quant a Digoxin assay are comparable with that of
other digoxin assays performed on nondedicated analyzers. In addition,
the new assay offers major advantages in terms of practicability and
integration into the laboratory organization.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Nonstandard abbreviations: DLIF, digoxin-like immunoreactive factor; FPIA, fluorescence polarization immunoassay; FIA, fluorescence immunoassay; and EIA, enzyme immunoassay.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  W. Steimer, C. Muller, and B. Eber Digoxin Assays: Frequent, Substantial, and Potentially Dangerous Interference by Spironolactone, Canrenone, and Other Steroids Clin. Chem., March 1, 2002; 48(3): 507 - 516. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Van Lente Methodology and Subspecialty Consolidation in the Clinical Laboratory Clin. Chem., January 1, 1997; 43(1): 11 - 12. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) native human sera.