| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Division of Laboratory Sciences, National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, GA 30341-3724.
2
Pacific Biometrics Research Foundation, Seattle, WA
98119.
3
Core Laboratory for Clinical Studies, Washington
University School of Medicine, St. Louis, MO 63110.
4
David Hassemer (State Laboratory of Hygiene, Madison,
WI); Santica Marcovina (Northwest Lipid Research Laboratories,
Department of Medicine, University of Washington, Seattle, WA); Robert
Rej (Wadsworth Center for Laboratories and Research, New York State
Department of Health, Albany, NY); Thomas G. Cole (Core Laboratory for
Clinical Studies, Washington University School of Medicine, St. Louis,
MO); Elizabeth T. Leary (Pacific Biometrics Research Foundation,
Seattle, WA); Christa M. Boersma-Cobbaert (Lipid Reference Laboratory,
Department of Clinical Chemistry, University of Rotterdam, Rotterdam,
The Netherlands); Masakazu Nakamura (Osaka Medical Center for Cancer
and Cardiovascular Diseases, Osaka, Japan); Chris J. Packard (Institute
of Biochemistry, Glasgow Royal Infirmary, Glasgow, Scotland); David W.
Seccombe [Canadian Reference Laboratory (1996) Ltd., Vancouver,
Canada]; Ferruccio Ceriotti (Laboratorio Analisi Cliniche, H.S.
Raffaele, Milan, Italy). Inactive laboratories that also participated
in the study: John H. Eckfeldt (Department of Laboratory Medicine and
Pathology, University of Minnesota Hospital and Clinic, Minneapolis,
MN); Joan A. Waletzky (Department of Biochemistry, The Cleveland Clinic
Foundation, Cleveland, OH); Judith R. McNamara (Jean Mayer USDA Human
Nutrition Research Center on Aging at Tufts University, Boston, MA).
a Address correspondence to this author at: Centers for Disease Control and Prevention, Mailstop F25, 4770 Buford Hwy. NE, Atlanta, GA 30341-3724. Fax 770-488-4192; e-mail mmk1{at}cdc.gov
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
Methods: CRMLN evaluated candidate DCMs and selected one that uses 50-kDa dextran sulfate with magnesium ions as the precipitation reagent followed by measurement of cholesterol by the CDC reference method. After validating the method, we transferred it to all CRMLN laboratories and successfully standardized it using CDC frozen serum reference materials. CRMLN laboratories participate in monthly performance evaluations.
Results: CRMLN laboratories were able to meet a precision goal,
as indicated by SD, of 
0.03 mmol/L (1 mg/dL) 94.4% of the time. They
were able to meet a bias goal of 0.05 mmol/L (2 mg/dL) for HDL-C
<1.09 mmol/L (42 mg/dL) 97.3% of the time and a goal of 3% for
HDL-C 
1.09 mmol/L (42 mg/dL) 95.6% of the time. CRMLN is working to
further improve its performance by implementing a bias criterion of
0.03 mmol/L (1 mg/dL) for all HDL-C concentrations.
Conclusions: CRMLN selected, validated, standardized, and implemented a DCM for HDL-C that is accurate, robust, transferable, and practical. The DCM is being used to assist manufacturers in calibrating their products so that ultimately, clinical laboratories using the products will more accurately measure HDL-C.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
1.55 mmol/L (60 mg/dL) is
considered protective against CHD (2).
Because accurate and reproducible HDL-C measurements are essential for
effective application of NCEP treatment guidelines, NCEP established
analytical performance goals for HDL-C measurement
(4)(5). These goals were stated in terms of
total analytical error, which takes into account inaccuracy and
imprecision. The goal is total error 13%. However, for the purposes
of standardization, the CDC's Cholesterol Reference Method
Laboratory Network (CRMLN) encourages manufacturers to strive to meet
separate accuracy and precision goals. The 1998 performance goals
recommended by the NCEP that are consistent with the total error goal
are a bias vs the accuracy base 5% and a precision of CV 4% at
HDL-C 1.09 mmol/L (42 mg/dL) and SD 0.044 mmol/L (1.7 mg/dL) at
HDL-C <1.09 mmol/L (42 mg/dL) (5).
In 1990, CRMLN established a process by which manufacturers can establish traceability to the National Reference System for Cholesterol of their diagnostics products used to measure TC (Myers GL, Kimberly MM, Waymack PP, Smith SJ, Cooper GR, Sampson EJ, manuscript in preparation). Traceability is accomplished by performing a fresh-sample method comparison, which is based on NCCLS EP9-A protocol (6), between the diagnostic product and the reference method for cholesterol by analyzing a minimum of 40 fresh serum specimens. When a product meets the NCEP performance goals, the manufacturer is issued a dated Certificate of Traceability. NCEP's Working Group on Lipoprotein Measurement recommended that CRMLN provide a similar program for HDL-C (5). We report here on the process that CRMLN used to select, validate, standardize, and implement a designated comparison method (DCM) for HDL-C, which has been used since 1994 to evaluate the accuracy of diagnostic products for measuring HDL-C.
The CDC reference method for HDL-C has served as the reference point for the CDC-National Heart, Lung, and Blood Institute (NHLBI) Lipid Standardization Program (LSP) since 1981 (7). Because NCEP used studies in which laboratory measurements were standardized by the LSP to establish the medical decision points for assessing an individual's risk of developing CHD, NCEP recommended that the CDC reference method for HDL-C continue to serve as the accuracy base for HDL-C measurements (5).
In the CDC reference method for HDL-C, 5-mL serum specimens are ultracentrifuged to separate HDL, LDL, intermediate-density lipoprotein, and lipoprotein a (d >1.006 kg/L bottom fraction) from VLDL and chylomicrons (d <1.006 kg/L top fraction) to remove TG interference. After LDL, intermediate-density lipoprotein, and lipoprotein (a) in the bottom fraction are precipitated with a heparin-manganese (HM) reagent, HDL-C in the supernatant is measured directly with the CDC reference method for cholesterol (8)(9).
CRMLN carefully considered using the CDC reference method for HDL-C, but it was deemed impractical for accuracy transfer via method comparisons that require 40 or more fresh serum specimens. Primary issues were (a) the large specimen volume required (5 mL), and (b) the high cost in time, skilled labor, and equipment because of the involvement of ultracentrifugation. CRMLN therefore looked for a DCM for HDL-C that would address each of these issues and be a practical way for manufacturers to establish traceability to the accuracy base. The ideal DCM would be accurate, robust, transferable, and practical for use in the CRMLN. It would also be traceable to the CDC reference method for HDL-C by standardization and regular surveillance of CRMLN laboratories. These criteria should be met by applying precipitation directly to serum without prior ultracentrifugation if maximum TG concentrations were specified.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
phase i: selection of a dcm for hdl-c
Selection of candidate DCMs.
A task force,
consisting of representatives of the CDC and two CRMLN laboratories,
Pacific Biometrics Research Foundation (PBRF) and Washington University
School of Medicine (WUSM), was appointed to evaluate the candidate
methods. All methods were two-step procedures that involved removing
the apolipoprotein B-containing lipoproteins (chylomicrons,
VLDL, intermediate-density lipoprotein, LDL, and lipoprotein a) by
precipitation and measuring the remaining lipoprotein cholesterol,
HDL-C, in the supernatant using the CDC reference method for
cholesterol. The task force initially considered four precipitation
reagents: HM; 50-kDa dextran sulfate-magnesium (DS); phosphotungstic
acid (PTA); and polyethylene glycol. The task force eliminated
polyethylene glycol because of the difficulty of pipetting this viscous
reagent. It ranked the remaining reagents using nine criteria based on
the experience of the CRMLN laboratories with each reagent and a
literature review. As a result, the task force eliminated PTA because
of a lack of consistency in the literature of the specific reagent
concentrations and methodologies used and concern about the stability
of the assay, which is highly sensitive to reagent pH and magnesium
concentrations. There was not a specific PTA method that was readily
apparent as agreeing with the reference method for HDL-C. The large
dilution factor that is typical of both the polyethylene glycol and PTA
methods also contraindicates use of these reagents with the reference
method for cholesterol because samples with low HDL-C concentrations
are diluted below the analytical range.
Because accuracy was the most important criterion, the task force performed a study to assess accuracy of the two candidate methods (DS and HM) relative to the reference method for HDL-C (11). The same precipitation reagent was used in the reference method for HDL-C and the HM candidate DCM. The DS candidate DCM was modified from the method of Warnick et al. (12) by reducing the magnesium chloride (MgCl2) concentration by 30% to improve accuracy vs the reference method.
Comparison specimens.
The Transfusion Services Unit of Emory
University Clinic, Atlanta, GA, obtained informed consent and collected
blood from 20 fasting donors. The procedures followed for collection
were approved by CDC's institutional review board. Whole blood was
collected in glass containers without any additive and allowed to clot
for 1 h at room temperature before the serum was separated by
centrifugation. Within 8 h after collection, the serum was
dispensed by 2.3-mL aliquots into sterile vials and frozen at
-70 °C. HDL-C concentration were between 0.78 and 1.81 mmol/L (30
and 70 mg/dL); all TG concentrations were <2.26 mmol/L (200 mg/dL).
The fresh-frozen samples were shipped to the participating laboratories
on dry ice and stored at -70 °C.
Candidate method DS: dextran sulfate-magnesium chloride
method.
Dextran sulfate (Dextralip 50) was obtained from Genzyme.
Magnesium chloride
(MgCl2 · 6H2O) and
sodium azide (NaN3) were obtained from Sigma
Chemical. The DS reagent was prepared by mixing equal volumes of a
stock solution containing 20 g/L DS and 0.5 g/L
NaN3 and a stock solution containing 0.7 mol/L
MgCl2 and 0.5 g/L NaN3.
This working reagent contained 10 g/L DS and 0.35 mol/L
MgCl2. The solution was stored between 2 and
8 °C. The three task force laboratories used the same working
reagent solution prepared at PBRF and shipped to CDC and WUSM at
4 °C.
The samples and working reagent were equilibrated to room temperature and mixed at a ratio of 10 volumes specimen to 1 volume reagent. Final reagent concentrations were 0.0091 g/L DS and 0.032 mol/L MgCl2. The samples were then incubated at room temperature for 10–30 min and centrifuged for 30 min at 4 °C and 1500g (or equivalent "g-minutes"). Clear supernatants were analyzed using the reference method for cholesterol. Supernates that could not be analyzed immediately were stored at -20 °C for up to 72 h.
Candidate method HM: heparin-manganese method.
A 5000 USP
units/mL heparin solution, derived from porcine intestinal
mucosa, was obtained from Wyeth-Ayerst Laboratories and used as is. It
was stored at 4 °C. Manganese chloride (MnCl2)
was obtained from Fisher Chemical. All three task force laboratories
used the same working reagent solution prepared at CDC and shipped to
PBRF and WUSM at 4 °C.
The samples and working reagent were equilibrated to room temperature. To 2 mL of serum were sequentially added 80 µL of heparin reagent and 100 µL of 1 mol/L MnCl2. Final reagent concentrations were 183 USP units/mL heparin and 0.046 mol/L MnCl2. Samples were incubated in an ice water bath for 30 min and centrifuged for 30 min at 4 °C and 1500g (or equivalent g-minutes). Clear supernatants were analyzed using the reference method for cholesterol. Supernates that could not be analyzed immediately were stored at 4 °C for up to 72 h.
Test run composition.
The laboratories analyzed individual
samples by each of the candidate methods in the same analytical
run. They also analyzed the three CDC QC materials in every
analytical run. CDC analyzed all samples in quadruplicate; WUSM and
PBRF analyzed all samples in duplicate. In addition to the candidate
DCMs, CDC analyzed the samples using the reference method for HDL-C
(9). The throughput was three to four fresh-frozen samples
per analytical run in all three laboratories.
Statistics.
The number of samples required to have adequate
power to detect a difference between two methods depends on the CV of
the method being tested. The lower the CV, the fewer the samples
required. Preliminary studies performed with similar methods showed
that CRMLN laboratories had CVs between 2% and 3%. Assuming that a CV
of 2% could be achieved, the three participating laboratories would
have 80% statistical power to detect differences of 1.25% from the
reference method when each laboratory analyzed 21 samples in duplicate
by each of the methods.
phase ii: validation of ds method as the dcm
Frozen vs fresh serum comparison.
Eleven CRMLN laboratories
participated in this part of the study. Each of the participating
laboratories analyzed between 14 and 20 samples, for a total of 209
samples. Samples were analyzed in three to six analytical runs in each
laboratory. Only samples with TGs <2.26 mmol/L (200 mg/dL) were used.
The serum was divided into two aliquots. Within 8 h of collection, analysis of one aliquot by the DCM and freezing of the other aliquot at -70 °C was begun at the same time. Between 11 and 21 days, the frozen samples were thawed and analyzed by the DCM. The three CDC QC materials were analyzed in every analytical run. All analyses were performed in duplicate.
Statistics.
It was desirable to have adequate power to detect
a 1% difference between the two sample types. At the time of this
study, the CRMLN laboratories were required to maintain a CV of 3% for
the DCM. The CRMLN set a goal of obtaining 240 samples for the entire
study, with the sample load divided equally among the participating
laboratories. This sample number provided 95% power of detecting a 1%
difference between results for fresh and frozen sera. The laboratories
performed three to six test runs to minimize run-specific bias.
The data set was evaluated to determine whether any samples exceeded the 99% range for duplicates. If results from fresh or frozen samples were outside the 99% range for duplicates, based on the QC range limits for the HDL-C control for the laboratory, they were omitted from the data set. The data set was also evaluated for outliers. The bias and relative bias (frozen vs fresh) were determined for each sample. The mean, SD, and an outlier test statistic (four times the SD) were calculated for each type of bias. Samples with both bias and relative bias that exceeded the test statistic were eligible outliers.
Linearity.
Blood from two fasting donors was collected and
processed as described in phase I. The HDL-C concentration of the serum
from these two donors was 0.80 and 2.47 mmol/L (31.0 and 95.5 mg/dL);
both had TG concentrations <2.26 mmol/L (200 mg/dL.) The linearity
study was performed according to the NCCLS EP6-P protocol
(13). Briefly, this involves mixing the two serum samples in
proportions to give five equally distributed HDL-C concentrations,
analyzing each for HDL-C in quadruplicate and carrying out the EP6-P
statistical analysis.
Working reagent stability.
Each of five participating
laboratories collected three to six fresh sera. Separate aliquots of
each specimen and the CDC QC material were precipitated by reagent that
was 18 or 24 months old and by reagent that was less than 12 months
old. The resulting supernatants were analyzed in duplicate in the same
analytical run. The participating laboratories analyzed a total of 32
fresh serum samples in eight analytical runs over a 2-year period. The
total number of measurements for the 18- and 24-month-old reagents were
40 and 54, respectively. One CDC HDL-C QC material was included in
every analytical run.
phase iii: standardization of the dcm in the crmln
Reference materials used for standardization.
The materials
used for standardization of the CRMLN laboratories were prepared by CDC
(10) and had reference values (RVs) assigned by the CDC's
reference method for HDL-C. The RVs for these materials are the means
of 12 analytical runs in quadruplicate over a period of 2-3
months.
Standardization of CRMLN laboratories.
Previous studies with
TC have shown that accuracy is the most important variable in CRMLN's
ability to assess performance (14). Therefore, CRMLN
laboratories are required to meet more stringent performance goals than
those established by NCEP for clinical laboratories. At the time CRMLN
laboratories were initially standardized, NCEP performance goals for
clinical laboratories were bias vs the accuracy base 10% and
precision of CV 6% at HDL-C 1.09 mmol/L (42 mg/dL) and SD 0.064
mmol/L (2.5 mg/dL) at HDL-C <1.09 mmol/L (5). Initially,
CRMLN set as its goal to achieve accuracy and precision that was
one-half the NCEP goals for clinical laboratories.
To help minimize potential lot-to-lot variations in the precipitation reagent, all laboratories used the same lot of DS. The first step in the standardization process was a transfer challenge in April 1994. Each laboratory analyzed three pooled, frozen CDC reference materials in duplicate in four analytical runs, and their results were compared to the CDC RVs. In the next standardization step, the laboratories were sent a CDC-NHLBI LSP Part II challenge—four pooled, frozen CDC reference materials analyzed randomly in eight analytical runs (7). The three CDC QC materials were analyzed in every analytical run for both the transfer challenge and the LSP Part II.
phase iv: performance of the dcm in the crmln
Reference materials used for performance monitoring.
Standardization of the DCM in CRMLN laboratories was maintained through
monthly surveys. The laboratories were sent CDC frozen reference
materials for two monthly surveys per quarter and fresh-frozen
materials once per quarter. The RVs for the CDC reference materials
were obtained as described above; those for the fresh-frozen materials
were the means of four analytical runs in quadruplicate with the
reference method for HDL-C performed at CDC.
Performance monitoring schemes.
Three different schemes have
been used and three different performance criteria have been applied
since the surveys began. Between June 1994 and November 1995,
laboratories analyzed three samples in duplicate in four analytical
runs for each survey (scheme 1). The performance criteria were as
follows: for HDL-C 1.09 mmol/L (42 mg/dL), a bias of 0.05 mmol/L (2
mg/dL) vs CDC and a SD of 0.04 mmol/L (1.7 mg/dL); for HDL-C
>1.09 mmol/L (42 mg/dL), a bias of 5.0% vs CDC and a CV of 3.0%
(A criteria.) At this time, CRMLN had an unofficial goal for bias of
3.0% vs CDC for HDL-C >1.09 mmol/L (42 mg/dL; A' criteria).
Between December 1995 and August 1998, the monthly surveys contained
four samples that were analyzed in duplicate in two analytical runs
(scheme 2). During part of this time (December 1995 to July 1997), the
performance was evaluated with the A and A' criteria, as in scheme 1.
In August 1997, CRMLN tightened the performance criteria to a bias vs
CDC of 0.03 mmol/L (1 mg/dL) and SD 0.03 mmol/L (1 mg/dL) across
the concentration range (B criteria).
In response to the more stringent 1998 NCEP performance goal for clinical laboratories, CRMLN modified the survey scheme to provide greater statistical power to assess its performance. In September 1998, CRMLN changed the survey scheme so that the laboratories now analyze four samples in duplicate in four analytical runs.
QC materials.
CDC provides all CRMLN laboratories with QC
materials as described above. When old QC materials are replaced, each
CRMLN laboratory performs a minimum of 20 overlapping analytical runs
with the old and new material. CDC establishes network QC limits based
on the performance of all laboratories. CRMLN laboratories apply both
their own limits and the network limits to all analytical runs.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
. Paired SD, calculated from the method comparison results, as
well as SD and CV calculated from the HDL-C control are shown in Table 1
. The paired SD, an indicator of within-run precision, was calculated
as follows:
 |
|
Bias vs the reference method For HDL-C.
The mean percentage of
bias, the mean absolute percentage of bias, and regression statistics
for the comparison between the candidate methods performed in all three
task force laboratories and the reference method for HDL-C performed at
CDC are shown in Table 2
.
|
phase ii: validation of ds method as the dcm
Frozen vs fresh serum comparison.
Manufacturers are required
to analyze fresh samples (samples <8 h old) in the method comparisons
conducted with the CRMLN, but for practical reasons, CRMLN laboratories
use frozen samples. CRMLN undertook a study to compare results from
fresh and frozen serum (16).
Five samples (four from one laboratory) were omitted from the data set because they exceeded the 99% range for duplicates. Five outliers were also omitted, leaving 199 samples for the final analysis. The HDL-C concentration range for the samples was 0.28–2.46 mmol/L (11–95 mg/dL), with a mean concentration (± SD) of 1.33 ± 0.38 mmol/L (51.5 ± 14.6 mg/dL). The sample distribution was gaussian.
The mean CV for the 11 laboratories during this study was 1.2% (range, 0.49–2.2%) as determined from results of the HDL-C QC material. At the observed CV, there was 95% power to detect a difference of 0.5% with 199 samples.
Regression analysis of frozen sample results vs fresh sample results
yielded a linear relationship, with a correlation coefficient of 0.997,
a slope of 0.992, and an intercept of -0.0005 mmol/L (-0.021 mg/dL).
The bias was not dependent on HDL-C concentration (Fig. 1
). Both the bias and relative bias were analyzed by univariate
analysis. As a further test of concentration dependence, the sample set
was divided into approximately equal parts, one greater than and the
other less than the median concentration [1.32 mmol/L (51 mg/dL)].
These two sets of data were also subjected to univariate analysis. The
results of all the univariate analyses are shown in Table 3
. The univariate analysis showed a small but statistically
significant difference of 0.010 mmol/L (-0.4 mg/dL), or -0.81%,
between the frozen and fresh samples analyzed by the DCM.
|
|
Linearity.
The linearity study showed a deviation from the
estimated line of less than 0.018 mmol/L (0.68 mg/dL) with 95%
confidence over the analytical range of 0.80–2.46 mmol/L (31–95
mg/dL) HDL-C.
Working reagent stability.
Initially, the limit on the shelf
life of the working reagent was set at 12 months. Because preparation
of a new reagent batch requires expensive, time-consuming cross-over
studies between old and new batches, CRMLN evaluated the shelf life of
batches at 18 and 24 months.
The mean concentration difference between samples analyzed using working reagent 18 or 24 months old and working reagent <12 months old was 0.0026 mmol/L (0.1 mg/dL) HDL-C. The SD of the means was 0.0181 mmol/L (0.7 mg/dL) for the 18-month-old reagent and 0.0207 mmol/L (0.8 mg/dL) for the 24-month-old reagent. The Student t-test showed that there was no significant difference between the results. The P values were 0.35 for the 18-month reagent and 0.34 for the 24-month reagent; therefore, CRMLN extended the acceptable shelf life of the working reagent to 24 months.
phase iii: standardization of the dcm in the crmln
During the transfer challenge, all but two of the laboratories met
the 5% criterion for bias (Fig. 2
A) for all of the materials analyzed; all but three met the
3% criterion for CV. During the LSP Part II standardization, all of
the laboratories met the 5% criteria for bias (Fig. 2B
); all but
three met the 3% criterion for CV.
|
CRMLN laboratories that had a bias vs CDC of 5% and CVs 3% were
considered standardized and could begin providing reference services to
manufacturers. CRMLN completed standardization of the DCM for HDL-C by
July 1994.
phase iv: performance of the dcm in the crmln
Precision.
The ability of CRMLN laboratories to meet the
precision criteria are shown in terms of pass rates in Table 4
. The pass rates in Tables 4
and 6
are the percentage of events,
where an event is defined as performance of one laboratory on one
material in one survey, that meet the criteria for a particular scheme.
CRMLN laboratories have consistently met all precision criteria. Before
the B criteria were implemented, data collected between December 1995
and June 1997 were evaluated statistically and it was determined that
the SD had no concentration dependence across the range. Based on these
results, CRMLN decided that a single criterion for precision, as
indicated by a SD of 0.03 mmol/L (1 mg/dL), without regard for HDL-C
concentration is appropriate (B criteria).
|
|
Table 5
lists the QC results for each laboratory between July 1994 and
April 1997. These data clearly demonstrate the long-term precision for
CRMLN laboratories in performing the DCM. The SDs observed are well
below the 0.03 mmol/L (1 mg/dL) criterion currently applied to CRMLN
laboratories.
|
Accuracy vs the reference method for HDL-C.
The ability of
CRMLN laboratories to meet the accuracy criteria are shown in terms of
pass rates in Table 6
. The CRMLN laboratories had little difficulty meeting the
initial criterion of bias 5%. Between July 1994 and August 1998,
CRMLN laboratories also improved their accuracy (the pass rate for the
unofficial goal of bias 3% improved from 87.1% to 95.6%); however,
the ability of the CRMLN laboratories to meet new criterion of bias
0.03 mmol/L (1 mg/dL) has been difficult to document. CRMLN will
continue striving to meet this criterion and expects that the new
survey scheme (four samples analyzed in duplicate in four analytical
runs) implemented in September 1998 will improve its ability to
assess its members' accuracy.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
phase ii: validation of ds method as the dcm
CRMLN decided that the small but statistically significant
difference between the frozen and fresh samples analyzed by the DCM
would not greatly impact the CRMLN's services. We also believed that
the practical advantages of the DCM outweigh the disadvantage of this
minor difference. The need for improving the accuracy of clinical
methods was very great, and this small difference would not hinder
CRMLN's goal to improve the accuracy of HDL-C measurement by providing
a stable accuracy point for HDL-C.
phase iii: standardization of the dcm in the crmln
The results of the initial transfer challenge showed that the DS
method could be transferred easily and that it met CRMLN's performance
criteria. The standardization challenge showed that it could be
performed consistently and that most of the CRMLN laboratories were
also able to meet an informal bias goal of 3%.
phase iv: performance of the dcm in the crmln
Currently, CRMLN provides manufacturers with evaluation rather
than certification. In an evaluation, manufacturers can establish
traceability to the reference method for HDL-C, and the CRMLN does not
make pass or fail judgments about the product; CRMLN is providing a
reference point for manufacturers. CRMLN must be able to document that
its accuracy, as indicated by bias vs the reference method for HDL-C at
CDC, is 0.03 mmol/L (1 mg/dL) before it can certify that a
manufacturer's performance meets NCEP performance goals. CRMLN
believes that by offering a stable target for which manufacturers of
clinical diagnostic products can aim improves accuracy in clinical
laboratory measurement of HDL-C (17)(18)(19)(20)(21)(22)(23)(24).
|
Conclusions |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
By making the DCM available to manufacturers and clinical laboratories for formal and informal method comparisons, CRMLN has provided a method to impact the accuracy of HDL-C measurements. The DCM has been used successfully in a program allowing manufacturers to establish traceability of their methods to the HDL-C accuracy base (17)(18)(19)(20)(21)(22)(23)(24). The DCM has several practical advantages over the use of the reference method for HDL-C: the cost of running the DCM is lower, and it requires less than one-half the volume of serum (it requires 2 mL of serum rather than 5 mL).
Standardization of HDL-C measurement is especially relevant because homogeneous methods that do not require a sample pretreatment step are being introduced. These methods present new calibration challenges because they use different principles to quantify HDL-C. Accuracy depends on providing calibration with set points traceable to the accuracy base. CRMLN has been able to offer a valuable service by standardizing these new products as well as those manufacturers that produce traditional products.
Manufacturers of clinical diagnostic products can contact any CRMLN member to obtain the protocol for collecting and analyzing samples. A list of all current members can be found on the World Wide Web at www.aacc.org/standards/cholesterolinfo.html. All manufacturers who follow the protocol to establish traceability to the accuracy base are issued a dated "Document of Comparison" that is valid for 2 years. After 2 years, manufacturers are encouraged to repeat the process to maintain traceability. Manufacturers who have a current Document of Comparison from CRMLN are listed on the World Wide Web site listed above.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionConclusionsReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  M. S. Yee, D. V. Pavitt, T. Tan, S. Venkatesan, I. F. Godsland, W. Richmond, and D. G. Johnston Lipoprotein separation in a novel iodixanol density gradient, for composition, density, and phenotype analysis J. Lipid Res., June 1, 2008; 49(6): 1364 - 1371. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. L. Vega, M. F. Weiner, A. M. Lipton, K. von Bergmann, D. Lutjohann, C. Moore, and D. Svetlik Reduction in Levels of 24S-Hydroxycholesterol by Statin Treatment in Patients With Alzheimer Disease Arch Neurol, April 1, 2003; 60(4): 510 - 515. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S Houterman, W. Verschuren, C. Oomen, C. Boersma-Cobbaert, and D Kromhout Trends in total and high density lipoprotein cholesterol and their determinants in The Netherlands between 1993 and 1997 Int. J. Epidemiol., October 1, 2001; 30(5): 1063 - 1070. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. R. Warnick, M. Nauck, and N. Rifai Evolution of Methods for Measurement of HDL-Cholesterol: From Ultracentrifugation to Homogeneous Assays Clin. Chem., September 1, 2001; 47(9): 1579 - 1596. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. M. Muller Implementation of Reference Systems in Laboratory Medicine Clin. Chem., December 1, 2000; 46(12): 1907 - 1909. [Full Text] [PDF]
|
|
|
|
|
|
G. L. Myers, M. M. Kimberly, P. P. Waymack, S. J. Smith, G. R. Cooper, and E. J. Sampson A Reference Method Laboratory Network for Cholesterol: A Model for Standardization and Improvement of Clinical Laboratory Measurements Clin. Chem., November 1, 2000; 46(11): 1762 - 1772. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. M. Cobbaert and T. K. J. Luderer Total Error Evaluation of Roche Direct HDL-Cholesterol Reagent and Calibrator across 31 Lot Combinations: A 2-Year Experience Clin. Chem., January 1, 2000; 46(1): 133 - 134. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, RV = 0.70 mmol/L (27.2 mg/dL); 
,
RV = 1.30 mmol/L (50.3 mg/dL); •, RV = 2.05 mmol/L (79.0
mg/dL). (B), plot of percentage of bias of HDL-C
in CDC reference materials during standardization of DCM in CRMLN
laboratories, plotted by laboratory. 
, RV = 1.53 mmol/L (59.1 mg/dL).