| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Department of Clinical Biochemistry, St. Bartholomew's and the Royal London School of Medicine & Dentistry, Turner Street, London E1 2AD, UK.
2
Department of Clinical Biochemistry, The London
Hospitals Trust, Whitechapel, London E1 1BB, UK.
a Author for correspondence. Fax (44) 71-377-1544; e-mail p.g.holownia{at}mds.qmw.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Key Words: indexing terms: diabetes mellitus • hemoglobin variants
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
"Glycohemoglobin" refers to the carbohydrate–protein linkage on the N terminus of the ß chains of hemoglobin, predominantly HbA in adults. HbA1c specifically refers to glucose addition and accounts for the majority of glycated species present (9). HbA1c is usually expressed as a percentage of the total hemoglobin (THb) and can be measured by ion-exchange chromatography (10)(11), affinity chromatography (12), electrophoresis (13), or immunoassay (14)(15)(16). These methods all demonstrate systematic differences between their reported values and thus do not give interconvertible results; rather, comparison of results determined by different methods requires both knowledge of the glycated species of hemoglobin measured and the use of appropriate calibrators and reference ranges. The presence of interfering substances will also produce different responses in these methods; e.g., the high proportion of circulating carbamylated hemoglobins in patients with chronic renal failure is responsible for positive interference in assessments of HbA1c by methods based on charge separation, whereas affinity chromatography and enzyme immunoassay are unaffected (17)(18). The presence of genetically determined hemoglobin variants within a population, most commonly HbS, HbC, and HbE, has also been shown to influence the analysis of HbA1c by some types of HPLC and immunoassays (19).
Several HPLC instruments dedicated solely to the measurement of %HbA1c are available commercially (20), offering the user all the convenience of automated analysis. However, the immunoassay techniques, particularly the light-scattering examples, also offer advantages, in that they can be adapted equally to automated clinical chemistry analyzers (16), thereby avoiding the cost of dedicated equipment, and to integrated unit-dose analytical devices, which that can be operated at the point of care (15)(21).
Here, we report the adaptation of a light-scattering immunoassay (Unimate®; Roche Diagnostic Products, Welwyn Garden City, UK) for HbA1c to a routine, automated clinical chemistry analyzer (Dimension®; Dade International, Dudingen, Switzerland) and its comparison with a colorimetric method for measuring THb. The immunoassay is based on the release of hemoglobin from erythrocytes, followed by enzymatic digestion with pepsin to release ß-N-terminal fragments of hemoglobin; these fragments are then bound to a fixed amount of monoclonal antibody, present in excess. The remaining (unbound) antibody molecules are subsequently reacted with a synthetic polymer (a polyvalent complex of the ß-N-terminal fragments) to form an agglutination, the amount of which can be determined by turbidimetry and is inversely related to the concentration of HbA1c. The amount of THb present is obtained from the same hemolysate by using the alkaline hematin method described in the Roche product information. In the absence of an internationally agreed upon reference material, the Roche method applies appropriate correction factors established from a previous evaluation to adjust the results of the light-scattering method to match the results of the HPLC method used by the recent Diabetes Control and Complication Trial (1).
The immunogen used to develop the monoclonal hybridoma cells is a synthetic peptide consisting of the six N-terminal amino acids identical to the human hemoglobin ß-chain and glycated on the N-terminal valine residue. As reported elsewhere, the first three amino acid residues, together with the glucose moiety of the ß-N-terminal hemoglobin fragment, constitutes the epitope for the monoclonal antibody, which has also been used in another immunoassay method (22). This offers the potential advantage of a decreased sensitivity to change in HbA1c recognition at the sixth amino acid residue (Glu), the site at which the most common point-mutations occur that give rise to such hemoglobinopathies as HbS (sickle cell) and HbC (23).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Effect of variations in the amounts of
HbA1c reagents.
Signal response at 540 nm
was greatest when we doubled all of the reagent volumes specified in
the Roche method instructions. Minor adjustments for diluent flush
volumes were insignificant; however, decreases of only 20% in the
volume of antibody–latex reagent used reduced the signal by 60%.
Reaction linearity and optimal wavelength.
We used a fixed-interval rate format to quantify HbA1c. The
wavelength recommended by Roche (540 nm) demonstrated a 120-s period of
linearity during the progress of the reaction; this measuring interval
gave satisfactory increases in signal at both extremes of the
HbA1c assay range (30–250 mA). We chose to
measure absorbance at 45 and 145 s after the agglutination had
been started by addition of reagent R3 (see below).
HbA1c calibration.
We
serially diluted the Unimate monomeric calibrator, to represent
HbA1c concentrations of 0.8–12.0 µmol/L. To construct
the calibration curve, we followed an iterative five-parameter logistic
curve-fitting procedure, as found on the Dimension, but using a
software package (Deltagraph®; Delta Point, Monterey, CA)
to calculate the initial coefficients off-line. A typical calibration
is shown in Fig. 1
.
|
Effect of incubation time and temperature.
A
minimum preincubation of sample with hematin reagent for 5 min at
37 °C was found to be necessary for absorbance stability across the
major peaks of the derivatized form of hemoglobin ("Alkaline Hematin
D-575").
THb calibration.
Dilution of the THb
calibrant (Technicon SETpointTM Calibrator; Bayer
Diagnostics, Tarrytown, NY) to span the range 3.0–9.1 mmol/L gave a
linear calibration curve for absorbance measured at the main (Soret)
band for hemoglobin at 405 nm. The wavelengths corresponding to minor
hemoglobin peaks (540 and 574 nm) also generated linear calibration
curves but with much shallower gradients. We therefore selected the
absorbance at 405 nm, which had the greatest signal change, as giving
the most suitable response.
%hba1c determinations
By the Dimension analyzer.
The Dimension analyzer was
obtained from Dade International. Reagents, supplied by Roche under the
product name Unimate HbA1c (cat. no. 0755656), were as
follows: R1, hematin reagent in 400 mmol/L potassium phosphate buffer,
pH 11.5, plus preservative; R2, 0.5 g/L suspension of latex particles,
coated with mouse monoclonal antibody against ß-N-terminal fragments
of HbA1c, in 50 mmol/L Hepes buffer, pH 8.1, containing 2
g/L bovine serum albumin and 150 mmol/L sodium chloride; R3, a
ß-N-terminal fragment polyvalent complex (0.5 mg/L) in 10 mmol/L
formate buffer, pH 3.0, also containing 1 g/L bovine serum albumin.
Hemolysis reagent consisted of porcine pepsin (>100 kU/L) in 20 mmol/L
citric acid. The HbA1c calibrator consisted of a monovalent
ß-chain-glycated HbA1c peptide (Roche; cat. no. 0755664,
stated value 25.6 µmol/L); the THb calibrator (Bayer) had a stated
value of 9.1 mmol/L.
The Dimension analyzer can make as many as three additions of reagents or mixtures of individual components plus an aliquot of sample into a cuvette manufactured on board the analyzer. In addition to using the normal printout of results, we captured on computer the raw absorbance ("filter"; see below) data from the defined photometric readings of each method, using a switching device into a Microsoft Excel® spreadsheet by means of an interface communications application (Versaterm Pro®; Synergy Software PCS, Reading, PA). Empty Dimension reagent containers ("flexes") were punctured to allow introduction of the appropriate reagents for the Dimension open-channel application. The flex configurations were 3.5 mL of R2 in wells 1–5 and 3.5 mL of R3 in well 6 for the HbA1c immunoassay, and 3.8 mL of distilled water in well 1 and 3.8 mL of R1 in wells 2–6 for the THb assay. The reaction for the HbA1c method was started by adding R3 after a 5-min preincubation of sample with R2 (antibody-coated latex); monochromatic readings of absorbance at 540 nm were then performed at 45 and 145 s from this point. The THb assay likewise required a 5-min preincubation of sample and reagent (R1), after which the reaction volume was diluted with distilled water; a single photometric reading (at 405 nm) was performed 1 min later.
By HPLC.
Dimension results were compared with an
automated HPLC analyzer dedicated to measurement of %HbA1c
(Daiichi HI Auto HA-8140; UK distributor: Biomen Diagnostics,
Finchampstead, UK). Glycated forms were separated by reversed-phase
ion-exchange chromatography. The desired peaks, defined by retention
time, were used to quantify %HbA1c by an on-board
calculation algorithm for determining the areas under the curves.
By enzyme immunoassay.
We also compared the Dimension
results with those of an enzyme immunoassay (Dako Diagnostics, Ely, UK)
in which HbA1c is captured and denatured on the surface of
a microtiter plate well. The amount of antigen is detected by using an
enzyme-labeled monoclonal antibody directed against the glycated
N-terminal sequence of the first 8 amino acid residues
(14). The results were related to calibrators of known
%HbA1c (assigned by HPLC).
samples
All blood samples were obtained from routine laboratory operation
and had been submitted for determination of HbA1c as
part of individual patient management. The samples were collected into
EDTA-containing tubes and stored at 4 °C until analysis, all work
being completed within 1 week of collection. Samples for the
comparisons between methods were drawn from six patient groups:
nondiabetics (n = 50), diabetics (n = 50), patients with
renal failure (n = 20; urea and creatinine concentrations of
16–33 mmol/L and 318-1150 µmol/L, respectively), patients with
hyperlipidemia (n = 10; triglycerides 3–12 mmol/L), patients
whose samples were icteric (n = 10; bilirubin >100 µmol/L), and
patients with identified hemoglobinopathies (n = 20; HbS and HbC,
10 each).
We also used three in-house control pools of whole blood, prepared to assess defined aspects of method performance. Each pool contained clinically low, medium, or high amounts of HbA1c. To ensure homogeneity of sample, we thoroughly mixed each pool before dividing it into aliquots and stored these frozen at -70 °C ready for use.
statistical analyses
An ANOVA procedure recommended by the National Committee on
Clinical Laboratory Standards (24) was used to estimate
individual aspects of imprecision such as total assay and within-assay
reproducibility, calibration stability, and on-board reagent stability
for both %HbA1c and the separate components THb and
HbA1c. Using Microsoft Excel 5.0, we constructed a
spreadsheet to perform the ANOVA calculations (24).
Chi-squared test was used to assess the significance of the calculated
reproducibilities relative to target values having been previously
defined as analytically correct or to identify the significance of
factors that contributed to the total imprecision.
Regression analysis performed by the method of Passing and Bablok (25) was used for the method comparison studies; 95% confidence limits for slope and intercept were calculated from the standard error of regression and used to determine concordance with the target values of 1.0 and 0.00, respectively (26). This procedure not only offers the advantage of the Deming regression (25), where errors in both x and y planes are taken into account, but also is independent of the distribution of errors in both planes and is not unduly influenced by the presence of extreme points or outliers; i.e., all measurement points have equally valid weighting in the estimation of the regression line.
The statistical package used was Astute (Diagnostic Development Unit, University of Leeds, Leeds, UK). For the other assessments of performance, simple linear regression was used to determine compliance of observed results within defined tolerance limits.
evaluation and validation
Imprecision.
Both within-assay and total assay
reproducibility were determined by the ANOVA procedure: a
one-run-per-day design in which the three in-house pools (low, medium,
and high %HbA1c) were analyzed in duplicate over a period
of 20 days with use of daily calibration. The within-assay
reproducibility was calculated from the average of each day's mean and
SD; the spreadsheet ANOVA calculation provided the total
reproducibility of the assay. An alternative estimate of assay
imprecision in the form of a sample precision profile was also
calculated from the observed variation (CV) in duplicate analyses of
the clinical samples (n = 160) used for the method comparison
studies.
Stabilities of calibration and on-board reagents.
Calibration stability was determined by reanalyzing each day's
"filter" data obtained from the imprecision data and calculating
values according to the calibration from Day 1. The total assay
reproducibility was determined by the recommended ANOVA procedure.
Stability of the reagents remaining on board the Dimension with
punctured flexes required by the Dimension open-channels format was
also assessed by the off-line analysis of the captured filter data
during the 20 days and calculated from the first day's calibration.
Accuracy.
Method comparisons were performed on 160
specimens of whole blood from the patient groups defined previously.
Duplicate analyses were performed with the method adapted to the
Dimension analyzer, singleton analysis with the HPLC and EIA methods
(calibrated where appropriate according to the respective
manufacturer's instructions). Quality-control samples recommended for
these methods were regularly assayed in the same runs with the
unknowns.
Sedimentation stability of particle reagent.
The Roche
literature recommends an initial mixing of the antibody-coated latex
reagent. Concerns about the effect of potential sedimentation of the
latex-antibody component over time when loaded onto the Dimension
reagent carousel led us to analyze aliquots of control materials (low,
medium, and high concentrations) in duplicate every hour for 1 day.
Freshly thawed aliquots of control material were used each time, and
the reagent was left undisturbed for the duration of the assessment.
Sample stability.
To assess sample deterioration during
storage at 4 °C, we chose at random 10 freshly collected samples and
analyzed them serially in duplicate with the Dimension over 17 days.
Influence of Schiff base and hematocrit.
To determine
whether any cross-reactivity ensued from alteration in the Schiff base,
we analyzed samples that had been incubated with various concentrations
of glucose. A single whole-blood sample with an %HbA1c
value within the routine reference range was divided into three
aliquots and centrifuged; the resulting plasma was put aside. The three
erythrocyte samples were then resuspended in phosphate-buffered saline,
20 mmol/L glucose solution, or 50 mmol/L glucose solution and incubated
at 37 °C for 4 h. After centrifugation (<2 min) to remove the
incubating solutions, we resuspended the erythrocytes in their own
plasma and analyzed them in duplicate for %HbA1c within 15
min after the removal of the incubating solutions.
To determine the effect of hematocrit, we centrifuged a single sample of whole blood for which %HbA1c and hematocrit were within the quoted reference ranges and separated the plasma from the cells. We then recombined the two components in various proportions (10% to 100% erythrocytes, by volume), to create samples with a range of hematocrit values, and analyzed each sample in duplicate.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
4.0% and 6.5%, respectively, for each of
the three control pools. Both values conform to target specifications
(27), being not significantly different from CV = 5%
as confirmed by chi-squared analysis (Table 1
). The relation between mean control pool results and duration
of storage demonstrated in all cases that the confidence limits
calculated from linear regression encompassed a slope of zero. The
observed scatter about the line of best fit was random and fell <5%
from each pool mean, thereby confirming the absence of any trends
during the 20-day period of assessment. Respective slopes and 95%
confidence limits for each pool were: low, 0.012 ± 0.029; medium,
0.011 ± 0.031; and high, 0.035 ± 0.039. Sample imprecision
calculated from a precision profile gave a mean CV of 4% across the
full analytical range of 2–15% HbA1c.
|
Calibration stability.
There was no significant
difference (chi-squared analysis) between the imprecision due to
calibration for total assay reproducibility (CV <4.8%) and the CV
target of 5% (Table 2
). The variation of mean results calculated from the Day 1
calibration during the 20-day assessment period also demonstrated a
zero slope within 95% confidence limits and minimal (<5%) random
scatter about the regression line. Respective slopes and ± 95%
confidence limits for each were: low, 0.002 ± 0.018; medium,
0.002 ± 0.028; and high, 0.0008 ± 0.042. The calibration
was stable for at least 20 working days or one calendar month.
|
Stability of on-board reagents.
Fig. 2
shows the plot of daily means for each pool during the 20-day
study, together with results of the linear regression analysis and 95%
confidence limits of the slope for each pool provided. The line of best
fit dropped below the limits of ± 5% variation about the means of the
controls from Day 1 at Day 12 for the medium and high pools. The low
pool showed a 10% drop after 3 days; after that, the mean values
stabilized. The Day 20 results for the low, medium, and high controls
were respectively 12%, 9%, and 0% lower than those on Day 1. Assay
reagents on board the Dimension, which were contained within punctured
wells, were stable for 1–2 days across the range of
%HbA1c.
|
Method comparison.
The comparison data are shown
graphically in the form of scattergrams with regression analysis
statistics calculated according to Passing and Bablok (see Fig. 3
). As evidenced by the concordance of observed slopes and
intercepts with respective target values of 1.00 and 0 (within 95%
confidence limits), the results from Dimension showed close overall
agreement with HPLC; the slope displayed a small but significant
rotational bias, 0.902 ± 0.04, but the intercept, -0.096 ±
0.25 was not significantly different from 0. The EIA demonstrated a
larger and significant negative bias in slope (0.863 ± 0.064) and
intercept (-0.825 ± -0.268) in comparison with the HPLC assay,
but closely agreed with the Dimension assay: slope, 1.024 ±
0.046, and intercept, 0.687 ± 0.21.
|
The Dimension and HPLC methods showed a positive bias as great as 2.5
times in results for patients with hemoglobinopathies, whereas the EIA
method appeared to give no change in %HbA1c in this
patient group in comparison with the HPLC method. A separate comparison
between all three methods that excluded the results from the patients
with renal failure showed essentially no difference in the slopes and
intercepts of the regressions from those shown in Fig. 3
; i.e., for
Dimension vs HPLC, Dimension vs EIA, and EIA vs HPLC, the respective
slopes were 0.89, 1.02, and 0.87, and the respective intercepts were
-0.01, 0.64, and -0.62. Closely similar results were also observed
when the patients with high concentrations of bilirubin or
triglycerides were omitted from the comparisons.
Sample stability.
As Fig. 4
shows, there was no significant change in results over the
17-day period of storage and analysis for any of the 10 samples
measured.
|
Influence of Schiff base effect and hematocrit.
The
samples incubated with glucose at either concentration demonstrated no
significant variation of results from that of the samples incubated in
phosphate-buffered saline: %HbA1c values were 5.6% in
buffer, 5.4% in 20 mmol/L glucose, and 5.8% in 50 mmol/L glucose.
There was also no change in %HbA1c values at hematocrits
ranging between 30% and 100%; i.e., mean %HbA1c = 4.6%
(range 4.5–4.7%). At 10–20% erythrocytes, however, the
%HbA1c measured increased to 5.0%.
Sedimentation stability of particle reagent.
Variation
in the mean hourly concentrations of R2 over time was assessed by
regression analysis. The results (Fig. 5
) demonstrate the absence of any significant change of slope
during the 1-day period of analysis, within calculated 95% confidence
limits.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The within-assay and total reproducibilities of the %HbA1c method, as assessed for imprecision (CV) on the basis of daily calibration, were 4.0% and 4.0–6.5% across the assay range—values comparable with those presented in many reports in the literature for other methods (11)(16)(20). The imprecision marginally improved to ~3.6% (within-assay reproducibility) and ~4.8% (total reproducibility) when the results were calculated on the basis of the initial day's calibration, an illustration of the stability of the reagents. This level of imprecision meets the requirements of a total reproducibility of CV 5% identified by the National Diabetes Data Group (27) and is similar to the goal specifications defined by other strategists (28)(29)(30). The reagents have been demonstrated to be stable during a whole day on board the Dimension without mixing, and the stability of samples kept at 4 °C was essentially unchanged throughout a normal working week. A small trend in decreasing %HbA1c was observed when open-well on-board reagents (punctured flexes) were used in the open-channels format for all three in-house control pools. It is unlikely that this small decrease would be clinically significant, however. The total imprecision data demonstrated that this trend is absent if fresh flexes are prepared daily. Given a future situation where flexes have already been sealed at manufacture after the dispensing of reagents, an initial agitation of the flex before introduction on board the Dimension should be sufficient to maintain stability.
There was close agreement between the Dimension and EIA methods for samples containing normal adult hemoglobin (n = 140) and spanning the range of clinical interest (3–15% HbA1c). A small negative bias in slope was seen when the Dimension method was compared with the HPLC assay. In keeping with the practice used by the manufacturer and other study methods (1), one can, for the reasons stated in the introduction, use an appropriate correction formula to adjust the final calculated result on the Dimension to account for this small negative bias. A small negative bias in the comparison of the established EIA method with HPLC was greater than that between Dimension and HPLC. The Dimension, therefore, generated results intermediate between these two methods, in the concentration range where the amount of scatter about the line of best fit was sufficient to ensure agreement with the EIA method within the calculated 95% confidence limits of slope and intercept.
The data on samples containing abnormal hemoglobins showed an apparent overestimation by the light-scattering method in comparison with the HPLC method, and suggested that the variants might be discriminated into two groups that give separate and distinctive positive biases. The HbS results obtained with the Dimension are about double (range 2.0–2.5 times) those of the HPLC method, whereas the HbC results lie almost within the upper boundary (range 1.0–1.3 times) of the scatter found in the comparison of normal hemoglobins. The EIA method clearly demonstrates even greater interference from both hemoglobin variant forms. The complete separation of the variant hemoglobin forms by the HPLC method used in this study permitted the direct calculation of results from peak areas, thus achieving a very high level of accuracy—comparable with that of a labor-intensive isoelectric focusing method reported to give the highest specificity (28). Other dedicated HPLC analyzers, because of incomplete separation, calculate HbA1c by comparison with calibrant chromatograms stored on the instrument (20). In the absence of a reference method or material, the Daiichi method is arguably one of the best automated HPLC methods and accurately quantifies %HbA1c and its variants. We therefore concluded that the Dimension method gave a better agreement with HPLC results and performance than did the established EIA method.
The monoclonal antibody used in the EIA (14) was raised against the amino acid sequence of the first eight N-terminal residues; the HbS and HbC variant changes at position 6 (of Glu to Val or Lys, respectively) would therefore interfere with antibody recognition. The situation with the antibody used in this study is more complex. After peptic digestion (cleavage of aromatic residues on the N-terminal sides), the ß-N-terminal fragment, which contains the epitope on the first 3 amino acids linked to the Amadori glucose, is a 14-residue peptide. The variant change at position 6 from a negatively charged and acidic amino acid (resembling HbA) to either a hydrophobic (e.g., HbS) or oppositely charged basic residue (e.g., HbC) will inevitably result in a significant change in the secondary structure of the fragment. The conformational presentation of the epitope to antibody will be altered, which may influence recognition. The observed positive bias between Dimension and HPLC methods leads to the conclusion that the variant change in amino acid may be principally due to enhanced affinity of the antibody–antigen interaction, thereby leaving less latex-bound antibody to agglutinate with the polyvalent antigen reagent. Other, less likely, possibilities may be that the glycated forms of the hemoglobin variants bind the monoclonal antibody nonspecifically under the conditions of the assay, or that the agglutination between unreacted antibody and polyvalent antigen is inhibited by some other mechanism. Studies are currently in progress with HbS, HbC, and other variant forms (such as HbD and HbE) in which the point mutations are much further removed from the N-terminal side; these studies use nondigested hemoglobin chains in which the N-terminal epitope has been exposed by unfolding of the ß-chain (the HbF variant does not contain ß-chains and so is not included).
A review of the literature on actual performance of HbA1c immunoassays in relation to glycated hemoglobin variants is limited to assessment (18) of the EIA method of Dako (14) and the light-scattering method of the DCA 2000TM (22), developed and manufactured by Bayer. The former does not recognize hemoglobin variants at all; the latter, however, shows results similar to those of affinity chromatography, which correctly estimated all variants except homozygous HbS and HbC. Alternative light-scattering immunoassays, using broadly the same principle as described here, claim to detect variants HbS, C, E, and D through the specificity of the polyclonal antibody, residing in the first four amino acids of the ß-chain N-terminus; one such method is part of an automated analyzer, the other is an integrated unit device (16). To the best of our knowledge, however, assessments of their performance in the literature have not included hemoglobin variants.
Some methods for quantifying HbA1c have been reported to give falsely high results in patients with chronic renal failure (31)(32); results in such patients are also high for measurements of fructosamine, then decrease after dialysis. The immunoassay for HbA1c described here is not affected by exposure of hemoglobin to higher concentrations of urea or other metabolites. Patients with conditions associated with a shortened lifespan of the erythrocyte (e.g., hemolytic anemia, homozygotes for HbS or HbC) will inevitably invalidate this and all other methods unless new reference ranges can be established (19). There was no evidence of other interferences in patients with lipemia or hyperbilirubinemia within the conditions assessed. The presence of the Schiff base intermediate and variations in hematocrit, as would be encountered under most clinical circumstances, did not significantly change any result.
In summary, this adaptation offers the user all the benefits provided by an automated analyzer, e.g., rapid sample throughput (up to ~150 tests/h), minimum operator intervention, and a small pretreatment step (hemolysis) performed off-line. The total imprecision of the method is enhanced by the stability of the reagents (reflected in the need to calibrate only about once a month). Improvement in method imprecision by use of stable reagents, thereby reducing the variation implicit in frequent calibration, has been observed for other light-scattering immunoassays (33)(34)(35). We conclude that the light-scattering immunoassay described here provides a rapid, precise, and accurate means for estimating the proportion of HbA1c, the first result being available within 8 min of sample insertion into the analyzer after the hemolysis pretreatment step. As adapted to the Dimension analyzer, the assay shows close agreement with an HPLC/ion-exchange method, which we consider the method currently available that is nearest to being an automated reference procedure. Because %HbA1c estimates are inappropriately high in the presence of some hemoglobinopathies, it will be important to recognize the presence of variants when reporting results.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  W. L. Roberts, B. K. De, D. Brown, C. M. Hanbury, J. D. Hoyer, W. G. John, T. L. Lambert, R. B. Lundell, C. Rohlfing, and R. R. Little Effects of Hemoglobin C and S Traits on Eight Glycohemoglobin Methods Clin. Chem., February 1, 2002; 48(2): 383 - 385. [Full Text] [PDF]
|
|
|
|
|
|
E. L. Frank, L. Moulton, R. R. Little, H.-M. Wiedmeyer, C. Rohlfing, and W. L. Roberts Effects of Hemoglobin C and S Traits on Seven Glycohemoglobin Methods Clin. Chem., June 1, 2000; 46(6): 864 - 867. [Full Text] [PDF]
|
|
|
|
|
|
W. L. Roberts, J. M. Chiasera, and K. M. Ward-Cook Glycohemoglobin Results in Samples with Hemoglobin C or S Trait: A Comparison of Four Test Systems Clin. Chem., June 1, 1999; 45(6): 906 - 909. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
), medium (
), and high (
) HbA1c
pools and number of days on-board the Dimension analyzer (reflecting
reagent stability).
), and (bottom) EIA
vs HPLC (
).
