| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Department of Clinical Chemistry, University Hospital Rotterdam, 3015 GD Rotterdam, The Netherlands.
2
Department of Clinical Chemistry, De Baronie Hospital,
4819 EV Breda, The Netherlands.
3
Department of Epidemiology and Biostatistics, Erasmus
University Rotterdam, 3015 GE Rotterdam, The Netherlands.
Departments of
4
Clinical Chemistry and
5
General Internal Medicine, University Hospital Nijmegen,
6500 HB Nijmegen, The Netherlands.
6
Department of Clinical Chemistry, Queen Beatrix
Hospital, 7101 BN Winterswijk, The Netherlands.
a Address correspondence to this author at: Department of Clinical Chemistry, De Baronie Hospital, Langendijk 75, 4819 EV Breda, The Netherlands. Fax 31 (0)76-5277043; e-mail cobbaert{at}worldonline.nl
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
Methods: The total error (TE) of HDL-cholesterol measurements was assessed in native human sera by 25 Dutch clinical chemistry laboratories. Concomitantly, the suitability of lyophilized, saccharose-containing CRMs (n = 11) for HDL-cholesterol was evaluated.
Results: In the precipitation method group, which included 25
laboratories and four methods, the mean (minimum–maximum) TE was
11.5% (2.7–25.2%), signifying that 18 of 25 laboratories satisfied
the TE goal of 
13% issued by the National Cholesterol Education
Program (NCEP). In the homogeneous HDL-cholesterol method group, which
included five laboratories, each performing two different methods, the
mean (minimum–maximum) TE was 9.5% (6.0–17.3%) for the Boehringer
assay and 15.7% (3.3–30.7%) for the Genzyme assay. For the
Boehringer homogeneous assay, one of five laboratories did not meet the
TE criterion, whereas for the Genzyme homogeneous assay, three of five
laboratories exceeded the 13% criterion. The biases on the
HDL-cholesterol values found by various precipitation methods were
highly variable in all CRMs, irrespective of the quality, whereas the
biases found by the homogeneous method from Boehringer were far less
than ±5% for the highest-quality CRMs (CRMs 4–6).
Conclusions: The NCEP goal was met by 24 of 35 laboratories assessed by use of native human sera. Selectively pooled, lyophilized CRMs that are cryoprotected with 200 g/L saccharose have ample potential for use in the standardization of homogeneous HDL-cholesterol methods. © 1999 American Association for Clinical Chemistry
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
13%, which is achieved, for example, if clinical
laboratories have a mean bias of no more than ±5% and perform with a
CV 4% at 
1.09 mmol/L. To date, clinical chemistry laboratories frequently have measured HDL-chol by chemical precipitation of the apolipoprotein B (apoB)-containing lipoproteins with either polyethylene glycol (PEG) 6000, dextran sulfate (DS)/MgCl2 or phosphotungstic acid (PTA)/MgCl2, followed by quantification of the cholesterol content in the HDL-containing supernate (6). Standardization of these HDL-chol assays is challenging because, in addition to matrix effects, there is documented variability between the commonly used precipitation reagents; these differences in precipitation efficiency and recovery become obvious especially in lipemic sera (7)(8)(9). Therefore, homogeneous HDL-chol assays that no longer require sample pretreatment have recently become attractive (10)(11)(12)(13)(14).
In The Netherlands, the accuracy of HDL-chol measurements has not been investigated thus far. Consequently, a pilot HDL-chol survey encompassing a representative sample of 25 (14%) Dutch clinical laboratories was conducted. The project was a joint project of both the Lipid Reference Laboratory (LRL) Rotterdam and the Dutch National External Quality Assurance Society, named the SKZL. The survey aimed at documenting the state-of-the-art accuracy, imprecision, and TE of the four most commonly used second-generation methods and of two recently introduced third-generation HDL-chol assays, using fresh human sera. Because periodical monitoring of the accuracy of HDL-chol measurements in the Dutch proficiency testing program awaits the availability of commutable reference materials, 11 candidate reference materials (CRMs) with different concentrations of HDL-chol and triglycerides were developed and investigated for their suitability. The HDL-chol Designated Comparison Method (DCM) and the HDL-chol Reference Method of the CDC were used as the reference methods (15)(16).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
In the 25 laboratories, assessment of accuracy and imprecision was performed using a standardized protocol that was essentially based upon the cholesterol protocol for certification of clinical laboratories developed by the CDC [Mulder and Cobbaert, submitted for publication; Refs. (15)(16)]. In brief, accuracy was assessed using a split-sample comparison with fresh human specimens and performing duplicate measurements during 3 consecutive days. Between assays, native specimens were stored at 4 °C. The overall imprecision of the field HDL-chol methods was calculated from the repeated measurements of the native sera. In parallel and in all three assays, the suitability (or lack thereof) of the HDL-chol CRMs was investigated. To this end, the HDL-chol CRMs were freshly reconstituted at each occasion, i.e., for each assay.
HDL-chol value assignment was done centrally by the LRL Rotterdam, using the CDC DCM for the native sera, and using the CDC Reference Method or a modification of that method for the CRMs.
native human sera
Because the traditional approaches for assessing accuracy are
complicated by matrix effects, the CDC strategy for transferring
accuracy from reference laboratories to clinical laboratories was used
(15)(16). To this end, 6 to 10 fresh,
normotriglyceridemic (triglycerides <2.26 mmol/L) specimens from
hospital workers, spanning the clinically relevant HDL-chol
range, were selected by each contributing laboratory (Mulder and
Cobbaert, submitted for publication). All sera were measured in
duplicate in a single assay during 3 consecutive days by each
participant. Each day a new calibration was performed. Between assays,
sera were stored at 4 °C. These field HDL-chol data were all
gathered within 72 h after sample collection. In addition, on the
day of sample collection, aliquots of each serum sample were stored at
-20 °C or lower for future value assignment by the LRL Rotterdam
(15)(16).
CRMs for HDL-chol
CRMs intended to closely mimic fresh patient sera were developed
by the SKZL. Pools were prepared as described previously
(17). In total, 11 serum pools were made, containing various
concentrations of HDL-chol, triglycerides, and cryo- and lyoprotectant
(saccharose). Each pool consisted of material originating from at least
150 different patients. Before individual patient sera were
frozen at or below -20 °C, either 50 or 200 g/L saccharose was
added to each individual serum aliquot that was intended for use in
future pool preparation. By the time that a few liters of serum
pool were gathered per stratum, individual patient specimens were
thawed and mixed or supplemented as follows.
CRMs 1, 2, and 3, with low, medium, and high HDL-chol concentrations, were prepared by mixing the appropriate HDL-chol concentrations. The final saccharose concentration in CRMs 1–3 was maintained at 50 g/L. CRMs 4, 5, and 6, with low, medium, and high HDL-chol concentrations, were prepared from the same mother serum pools as those used for preparing CRMs 1, 2, and 3, respectively. Before CRMs 4–6 were aliquoted and lyophilized, their final saccharose concentration was increased to 200 g/L with additional saccharose. The addition of saccharose caused volume expansion. To achieve nearly equal HDL concentrations in the reconstituted CRMs, the volumes before lyophilization were adjusted to 1 and 1.4 mL for specimens containing 50 and 200 g/L saccharose, respectively. In the end, upon reconstitution in 1 mL of distilled water, the final HDL-chol concentrations in CRMs 4, 5, and 6 were somewhat higher than in CRMs 1, 2, and 3, respectively.
CRMs 7, 8, and 9, with low, medium, and high HDL-chol concentrations, respectively, and a final saccharose concentration of 200 g/L, were prepared from randomly gathered serum aliquots that were initially frozen with 200 g/L saccharose. In contrast to the preparation of CRM pairs 1 and 4 and 3 and 6, the low and high HDL-chol concentrations in CRMs 7 and 9 were created by, respectively, diluting or concentrating the randomly gathered serum pool. Notably, total triglycerides were <2.6 mmol/L in CRMs 1–9. Finally, CRMs 10 and 11, with medium (4.85 mmol/L triglycerides, mainly VLDL) and high concentrations of triglycerides (15.8 mmol/L triglycerides, mainly chylomicrons), respectively, and a final saccharose concentration of 200 g/L, were prepared from appropriate serum aliquots that were initially frozen with 50 g/L saccharose. Sample preparation and subsequent lyophilization of all CRMs was performed as described previously (17).
sera used for the validation of the modified cdc reference
method procedure
A frozen serum pool from the CDC, i.e., AQ15, which is usually
used for internal and external quality-control assessment of the
HDL-chol Reference Method and the HDL-chol DCM, was used for verifying
the HDL-chol accuracy in the 2-mL procedure compared with the 5-mL
procedure. In addition, precision was checked with two freshly frozen
single donor sera from hospital workers. All pools were stored at
-80 °C, and none contained saccharose.
evaluation of lipoprotein integrity in the HDL-chol
CRMs
Reconstituted CRMs were first checked macroscopically for
turbidity. Lipoprotein integrity was then studied by means of density
gradient ultracentrifugation (UC) (17). To this end,
serum samples were stained with Coomassie Brilliant Blue before UC,
which yielded blue-colored lipoprotein bands within the density
gradient after UC. The HDL-chol CRMs were then characterized by
lipoprotein electrophoresis with the Paragon system (Beckman
Instruments), with staining performed with Sudan Black (17).
Finally, because the CRMs were targeted with the CDC Reference Method,
the infranatants obtained after UC at serum density were inspected
visually and judged on the basis of homogeneity, i.e., the absence or
presence of flakes.
description of the various HDL-chol methods
used in this study protocol
CDC DCM used for accuracy assessment of normotriglyceridemic human
sera.
HDL-chol value assignment in both the freshly
frozen human specimens and the CRMs was carried out by the LRL of the
Academic Hospital Rotterdam, The Netherlands. The LRL Rotterdam is a
permanent, international member of the CDC Cholesterol Reference Method
Laboratory Network (15)(16)(18). The accuracy base for
HDL-chol reportedly consists of the CDC DCM (for normotriglyceridemic
sera) and the CDC Reference Method (for hypertriglyceridemic sera)
(19).
The CDC DCM, a two-step procedure encompassing Mr 50 000 DS/MgCl2 precipitation of apoB-containing lipoproteins, and subsequent Abell-Kendall analysis (18) was used to determine HDL-chol concentrations in the frozen native specimens sent by each participating laboratory to the LRL. To this end, duplicate determinations were performed in a single assay.
CDC Reference Method and modified CDC Reference Method for value
assignment of the CRMs for HDL-chol.
The CDC Reference Method, a
three-step procedure consisting of an UC step at serum density,
heparin/MnCl2 precipitation of apoB-containing
lipoproteins, and subsequent Abell-Kendall analysis, was used for
targeting the CRMs. CRMs 1–3 were assigned values with the original
CDC Reference Method on the basis of quadruplicate determinations in
four independent assays.
In the presence of 200 g/L saccharose, no lipoprotein separation could be achieved in CRMs 4–11, using the original CDC HDL-chol Reference Method. Consequently, the CDC Reference Method was modified. In essence, the saccharose concentration was "diluted" to 80 g/L saccharose by pipetting 2.00 mL of CRM in a bell-top Quick-Seal ultracentrifugation tube (Beckman Instruments) and by overlayering with 3.00 mL of an electrolyte-albumin solution. The composition of the electrolyte-albumin solution was critical because it affected the HDL-chol recovery in the subsequent precipitation step, i.e., HDL-chol recoveries were only complete if the overlayering "diluent" contained physiological concentrations of sodium, potassium, calcium, magnesium, chloride, and protein. Bovine albumin (60 g/L) was used as a protein source. Other than this modification, all other steps and solutions used were similar to those of the CDC Reference Method procedure. Note that after UC, chylomicrons and/or VLDL were eliminated by cutting at a similar, fixed height as in the 5-mL procedure. In addition, subsequent recovery of the infranatant, which contained HDL and LDL, and the serum proteins was performed by aspiration, washing of the inner side of the tube, and dilution with 0.15 mol/L NaCl solution up to a final volume of 5.00 mL. CRMs 4–11 were assigned values with the modified CDC Reference Method, on the basis of quadruplicate determinations in four independent assays.
The accuracy and the precision of the 2-mL procedure was validated and compared with the analytical performance of the 5-mL procedure, using fresh frozen human sera and a frozen human serum control from the CDC.
Field HDL-chol methods.
Four types of commercial precipitation
methods were evaluated by the participating clinical laboratories. Five
of the participating laboratories investigated, in parallel with their
precipitation methods, two recently introduced homogeneous HDL-chol
assays.
The HDL-chol chemical precipitation methods were assessed as follows: (a) the PEG 6000 method was assessed at seven laboratories, using precipitation reagents obtained from Merck (cat. no. 807491; three laboratories) or from Instruchemie (cat. no. 2073; three laboratories), or reagents prepared in house (one laboratory) (20)(21); (b) the PTA/MgCl2 method was assessed at nine laboratories, using precipitation reagents obtained from Boehringer Mannheim (cat. no. 543004; eight laboratories) or from Merck (cat. no. 114210; one laboratory); (c) the DS (Mr 50 000)/MgCl2 method was assessed at six laboratories, using precipitation reagents obtained from Beckman, Johnson and Johnson, and Sigma; and (d) the PEG/DS/MgCl2 was assessed at three laboratories, using precipitation reagent obtained from Instruchemie (cat. no. 2258). In the PEG/DS/MgCl2 precipitation method, the DS was Mr 15 000 (22). In addition, the lyophilized version of the homogeneous HDL-chol method from Boehringer (direct HDL-chol reagent, cat. no. 1661426; calibrator f.a.s. HDL/LDL-c, cat. no. 1778501) and the lyophilized N-geneousTM HDL-chol method from Genzyme (direct HDL-chol reagent, cat. no. 2570; direct HDL-chol calibrator, cat. no. 2574) were evaluated.
In the homogeneous HDL-chol method from Boehringer, sulfated
-cyclodextrin and DS form, at pH 7 and in the presence of
MgCl2, water soluble complexes with LDL, VLDL,
and chylomicrons (reagent 1), which are not accessible to PEG-coupled
cholesterol esterase and cholesterol oxidase (reagent 2). In the
Genzyme homogeneous HDL-chol method, a polyanion and synthetic
polymer (reagent 1) together form complexes with
chylomicrons, VLDL, and LDL particles and prevent them from reacting
with the second reagent, which is a mixture of enzymes (cholesterol
esterase, cholesterol oxidase, and peroxidase), 4-aminoantipyrine,
detergent, and buffer. The HDL particles are disrupted by the
detergent, thereby releasing the cholesterol and cholesteryl esters.
The HDL concentration is then determined enzymatically, using a
two-point reaction. The homogeneous HDL-chol methods are
hereafter denoted as "-cyclodextrin sulfate/PEG-coupled enzyme"
and "polymer/polyanion" assays, respectively.
In the chemical precipitation methods, the precipitation steps for prior isolation of HDL were done manually, whereas the cholesterol in the supernates was measured on automated clinical chemistry analyzers. In the homogeneous HDL-chol methods, sample pretreatment was no longer required, and fully automated applications were used. Application was done according to the instructions of the manufacturers, and all assays were performed with calibrators that were included in the kits. Results were scored after several assays in which reproducible results were obtained.
statistics
In the method comparison study using native specimens, overall
analytical imprecision (CVa) was calculated from
measurement repeats, whereas mean laboratory biases were calculated
from the linear regression line fitted between the HDL-chol data
produced in the field (y-axis) and the data produced with
the HDL-chol DCM in the LRL (x-axis). TE was calculated as:
1.96 x CVa (%) + absolute mean bias (%).
For final acceptance or rejection of HDL-chol method performance, a TE
criterion of 13% was used (5).
To investigate the suitability of the CRMs for HDL-chol standardization
purposes or accuracy assessment, laboratory means and method group
means (± SD) were calculated using basic statistics. Laboratory means
of the CRMs were compared with the target values assigned by the LRL
with either the CDC Reference Method or the modified CDC Reference
Method for HDL-chol. A significance level of = 0.05 was used
throughout the study.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
); i.e., 24 of 25 laboratories met the former interim TE goal of
22%, whereas 18 of 25 laboratories satisfied the 1998 TE goal of
13% (Fig. 1A
). The bias criterion ±5% or less was met in 12 of 25
laboratories (Fig. 1B
), and the precision criterion of 4% was
reached in 20 of 25 laboratories (Fig. 1C
). The laboratories with
HDL-chol biases exceeding ±5% were distributed equally across
different precipitation methods; therefore, bias differences could not
be attributed to a specific manufacturer's reagents or to a specific
analyzer (Fig. 1B
).
|
In the homogeneous HDL-chol method group, the mean (minimum–maximum
range) percent TE, bias, and overall imprecision were 9.5%
(6.0–17.3%), 3.3% (-6.4% to 13.4%), and 1.9% (0.7–2.4%)
for the -cyclodextrin sulfate/PEG-coupled enzyme assay, and 15.7%
(3.3–30.7%), 1.66%(-13.1% to 34.2%), and 1.9% (1.2–2.8%) for
the polymer/polyanion assay. In the -cyclodextrin
sulfate/PEG-coupled enzyme assay group, one of five laboratories
did not meet the 1998 TE criterion, whereas in the polymer/polyanion
assay group, three of five laboratories exceeded the 13% TE criterion
because of excessive biases (Fig. 1
).
The results of the linear regression analyses and the analytical
performance, averaged for the participating laboratories per HDL-chol
method group, are presented in Table 1
. In general, the slopes and intercepts scatter around one and
zero, respectively. However, for the homogeneous polymer/polyanion
assay, the slopes and intercepts of the regression lines were
significantly different from one and zero, respectively (P
<0.05), both overall and for individual participating laboratories.
Overall imprecision was far below 4% in each individual laboratory
performing either of the two homogeneous HDL-chol methods. Conversely,
the imprecision criterion was not reached unanimously in individual
laboratories performing chemical precipitation methods, with the
exception of the three laboratories performing the
PEG/DS/MgCl2 precipitation method. The range of
interlaboratory biases, i.e., minimum–maximum range, was smallest and
closest to 0% in the PTA/MgCl2 method group
(-6.3% to 9.8%), and largest in the homogeneous polymer/polyanion
method group (-13.1% to 34.2%). TE was smallest for the PEG and
PTA/MgCl2 precipitation method groups, i.e., the
maximum TE was ~15% and was largest for the homogeneous
polymer/polyanion method group.
|
characterization of the HDL-chol CRMs
After reconstitution of the CRMs a slight turbidity was visible in
CRMs 4–8, whereas a stronger turbidity was observed in CRMs
1–3 and in CRMs 9–10. CRM 11 was found to be lipemic. The CRMs under
study, both with 50 and 200 g/L saccharose, exhibited well-defined -
and ß-lipoprotein bands, but faint or missing pre-ß-bands,
especially in CRMs 1–3. Notably, some precipitation appeared at the
application site as well as a smear of indiscriminate staining between
the application and the ß-region, being again much more pronounced in
CRMs 1–3 than in CRMs 4–10. As an example, lipoprotein patterns of
CRMs 1–3 and CRMs 4–6 on agarose gel electrophoresis are shown in
Fig. 2
, A
and B
, respectively. For comparison, a normotriglyceridemic
control was run in lanes 7 and 8 of Fig. 2A
.
|
In agreement with previous results (17), density-gradient UC (data not shown) revealed sharp and well-separated HDL2 and HDL3 bands in all CRMs. In addition, tiny flakes were observed at the interphase with the atmosphere after UC. Flakes were also observed in the CDC Reference Method: after UC at serum density minuscule flakes were observed in the infranatants in all CRMs. More flakes were noted in CRMs 1–3 compared with CRMs 4–11.
validation of the modified cdc HDL-chol reference
method procedure
The mean HDL-chol ± SD (CV) for AQ15, as measured in
duplicate in 20 separate assays, was 1.288 ± 0.029 mmol/L (2.3%)
when the 2-mL procedure was used. In comparison, the mean HDL-chol
± SD (CV) for AQ15, as measured in duplicate in 20 assays, was
1.292 ± 0.027 mmol/L (2.1%) when the regular 5-mL procedure was
used and 1.287 ± 0.009 (0.7%) when the CDC DCM was used.
Analogously, in fresh-frozen single donor sera measured using duplicate spins in three separate assays, the overall CVs obtained with the 2-mL procedure was 0.8% at 2.24 mmol/L and 1.9% at 1.25 mmol/L.
Moreover, the accuracy of the modified CDC UC 2-mL procedure for
HDL-chol in the presence of 80 g/L saccharose, i.e., the final
saccharose concentration in the modified 2-mL procedure, was evaluated.
Previously, the HDL-chol concentration in a saccharose-supplemented
serum pool was compared with the HDL-chol target value, as determined
with the original 5-mL procedure in an undiluted serum pool and with
the modified 2-mL procedure in the same, non-saccharose-supplemented,
diluted serum pool. Because weighing 80 g/L of saccharose into serum
pools diluted the pool 1.135-fold, the non-saccharose-supplemented
serum pool was, in case of the 2-mL procedure and before UC, diluted to
a similar extent. The diluent was saline solution (9 g NaCl/L). In each
condition, HDL-chol value assignments were done in quadruplicate in two
different UC runs. When the 5-mL UC procedure was used, the HDL-chol
target value was 1.219 ± 0.009 mmol/L (0.7%); in the 2-mL
procedure, the mean HDL-chol concentration was 1.214 ± 0.035
mmol/L (2.9%) in the non-saccharose-supplemented serum pool and
1.203 ± 0.021 (1.8%) mmol/L in the 80 g/L saccharose-containing
serum pool (not significant; P 0.05).
value assignment to the HDL-chol CRMs
The HDL-chol values assigned by the LRL Rotterdam to the SKZL CRMs
are presented in Table 2
. Each target value is the mean of quadruplicate analyses
measured in four different assays, being produced with the CDC
Reference Method in the case of CRMs 1–3 and with the modified
CDC Reference Method in the case of CRMs 4–11. CRMs were reconstituted
freshly and pooled before each UC run. Notably, SDs ranged between
0.043 and 0.166 mmol/L and CVs between 2.7% and 10% (Table 2
); i.e.,
SDs and CVs in processed CRMs were three- to fivefold higher than those
obtained in frozen sera (see above).
|
performance of HDL-chol CRMs 1–11 for accuracy
assessment of HDL-chol precipitation methods
The dispersion of the mean HDL-chol concentrations
obtained in CRMs 1–11 by each participating laboratory—the mean being
based on duplicate analyses repeated during 3 consecutive days—around
the target values of CRMs 1–11 is presented in Fig. 3
. The outer horizontal lines display the uncertainty on the
values assigned by the (modified) CDC Reference Method and represent
the mean ± 2 SD. From Fig. 3
it is obvious that there is a large
scatter of the mean HDL-chol values around the target values.
Nevertheless, most data fit within the mean plus or minus the measuring
error (expressed as SD) of the reference range established with the
(modified) CDC Reference Method. Within each method group,
interlaboratory variability was similar for CRMs 1–10; however, in the
case of the chylomicron-rich CRM 11 the interlaboratory dispersion was
more pronounced. Across method groups, the PEG 6000 and the
PTA/MgCl2 precipitation methods performed better
than the DS/MgCl2 method both in terms of
accuracy and in terms of interlaboratory differences (Table 2
).
Conversely, the DS/MgCl2 had the worst
performance in combination with the lyophilized, saccharose-containing
CRMs, especially in the CRM containing 200 g/L saccharose.
|
performance of HDL-chol CRMs 1–11 for accuracy
assessment of HDL-chol homogeneous methods
In the -cyclodextrin sulfate/PEG-coupled enzyme
assay, the dispersion of the mean HDL-chol concentrations measured in
CRMs 1–11 by each participating laboratory (four laboratories; one
laboratory failed to analyze the CRMs in combination with the
homogeneous HDL-chol methods) around the target values of CRMs 1–11
was very moderate, the mean from the participating laboratories being
well within the mean ± 2 SD confidence limits of the CDC
Reference Method (Fig. 3
). From Table 2
it is obvious that for CRMs
2–6 the mean HDL-chol values were within ± 1% of the assigned
values. In the diluted CRM 7, the mean HDL-chol values displayed a
positive bias for all four laboratories (overall mean bias, 5.4%),
whereas in the "concentrated" CRM 9 mean HDL-chol values were
negatively biased (overall mean bias, -7.2%). Even in
hypertriglyceridemic CRMs, i.e., CRMs 10 and 11, which contained 4.85
and 15.8 mmol/L triglycerides, respectively, biases were less than
±5%.
In the polymer/polyanion assay, a large spread in mean HDL-chol values
around the target values was observed (Fig. 3
and Table 2
),
irrespective of the CRM analyzed. These observations are in line with
those observed in native human sera.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
Excessive bias, exceeding ±5%, was found to be present in ~50% of
the clinical chemistry laboratories performing chemical precipitation
methods (Fig. 1B
). In addition, because mean biases ranged between
-14.2% and 11.5%, interlaboratory differences between HDL-chol
measurements were as large as 25%, the latter observation underscoring
the findings of Crook (9) in 32 lipid clinics in the United
Kingdom. Excess bias was also observed in three of five laboratories
that implemented the homogeneous HDL-chol methods. In the direct
-cyclodextrin sulfate/PEG-coupled enzyme method, the overall mean
bias of 3.3% was in line with previous findings of our group
(14). In the direct polymer/polyanion assay, apparent
improper reagent formulation and/or application rather than improper
value assignment of the calibrator explained the observed findings
(data not shown). Although the survey was based on a small group, it
seems that the NCEP precision criterion of 4% was amply met for the
homogeneous HDL-chol methods compared with the precipitation methods
(Fig. 1C
), probably because the direct HDL-chol assays make the
cumbersome manual pretreatment step for isolation of HDL redundant.
The major advantages of this bias survey are related to the fact that the targeting of the native sera was done in a CDC Network Laboratory, using the recommended HDL-chol accuracy bases (15)(16). In addition, firm conclusions could be drawn from this survey because biases were calculated from HDL-chol measurements performed in fresh and native human sera, and the analysis of sixhuman sera on 3 consecutive days per laboratory led to an acceptable uncertainty for the bias and TE estimates (Mulder and Cobbaert, submitted for publication).
Assuming that the observed analytical performance of the participating
clinical laboratories for HDL-chol is representative for The
Netherlands, we concluded that especially the accuracy, and to a lesser
extent the precision, of conventional HDL-chol precipitation methods
were insufficient to satisfy the 1998 NCEP TE goal in one-third of the
clinical chemistry laboratories (Fig. 1A
). Whereas precision of the
direct HDL-chol methods was excellent (Fig. 1C
), inaccurate results
were obtained in about one-half of the participating laboratories (Fig. 1B
). Consequently, the results from this pilot HDL-chol survey
underscore the urgent need for stable and commutable calibration
materials that emulate patient sera.
evaluation of HDL-chol CRMs
Because improved nationwide accuracy awaits the availability of
reference materials that are free of matrix effects, are sufficiently
stable to be shipped by overnight postage without freezing, and have
long-term stability (preferably at 4 °C), an attempt was made to
develop such CRMs for HDL-chol. To this end, either 50 or 200 g/L
saccharose was weighed into serum pools to create different degrees of
cryo- and lyoprotection upon freezing and subsequent lyophilization
(17). A final saccharose concentration of 200 g/L was found
to be a better protectant because macroscopic turbidity after
reconstitution of the lyophilized CRMs was significantly less in CRMs
4–9 compared with CRMs 1–3, the latter being cryoprotected with only
50 g/L of saccharose. However, reconstituted CRMs were slightly turbid
even in the presence of 200 g/L saccharose, pointing to some degree of
lipoprotein disintegration and/or apolipoprotein denaturation. The
observed turbidity in the CRMs could be established on the basis of the
absorbance of the reconstituted materials at 620 nm (data not shown),
the presence of nonmigrating material at the application site after
agarose gel electrophoresis (Fig. 2
), and the presence of tiny flakes
in the infranatants after UC at serum density or at the interphase with
the atmosphere after density gradient centrifugation (data not shown).
The minuscule flakes probably were denatured apoB that could no longer
be dissolved, which was supported by the faint or absent pre-ß bands
on agarose gel electrophoresis.
Subsequent value assignment of the CRMs with the CDC HDL-chol Reference
Method was hampered by the high saccharose concentration, as reflected
by the three- to fivefold increases in CVs for targeting CRMs (Table 2
)
compared with those of frozen single-donor sera and frozen
quality-control material (see Results). To target CRMs
4–11, which contained 200 g/L saccharose, the CDC HDL-chol Reference
Method had to be modified. In essence, the saccharose concentration in
CRMs 4–11 was diluted to 80 g/L before UC to enable lipoprotein
separation, which automatically adds a dilution factor of 2.5,
contributing greatly to the higher CVs. Because the accuracy of the
modified CDC HDL-chol Reference Method was within 0.5% of the target
value obtained with the original 5-mL procedure (mean ± SD of
1.288 ± 0.029 mmol/L for AQ15, using the 2-mL procedure, compared
with 1.292 ± 0.027 mmol/L, using the official 5-mL procedure) and
because value assignment of each CRM was based on 16 spins using either
the Reference Method or the Modified Reference Method, i.e., performing
quadruplicate spins in four different UC runs, "target" values for
HDL-chol can be considered reliable and accurate.
In the present pilot HDL-chol survey, clinical chemistry laboratories
analyzed freshly reconstituted CRMs in duplicate in three separate
assays. Per participant and per CRM measured HDL-chol concentrations
were averaged. Fig. 3
shows that these HDL-chol concentrations were
highly variable both within and between precipitation method groups.
Because the between-run CVs of the CRMs produced by individual
laboratories did not differ significantly from the between-run CVs
produced with internal, non-saccharose-based, quality-control material
(P 0.05; data not shown), it is unlikely that
pipetting errors caused by the high viscosity of the CRMs explain the
interlaboratory and intermethod differences. Probably, the high amount
of saccharose in the (manual) pretreatment step interfered with HDL
isolation. From the scattered HDL-chol results around the assigned
values (Fig. 3
), we concluded that saccharose-based reference materials
cannot be used for standardizing precipitation methods.
HDL-chol concentrations in the CRMs as measured by all four
laboratories that performed the -cyclodextrin sulfate/PEG-coupled
enzyme assay were similar to those obtained for fresh material, ranging
between 93% and 105%. Likely, the 101-fold dilution of the saccharose
concentration in the final reaction mixture (4 µL of CRM was
incubated with 300 µL of R1 and 100 µL of R2) did not interfere
with assay specificity and HDL-chol recovery. More specifically,
HDL-chol recoveries were almost ideal in CRMs 4–6, being 100% ± 1%.
Because the aforementioned CRMs 4–6 were pools of the highest quality,
these data support the validity of the targeting procedure in the CRMs
with the modified CDC Reference Method. However, for CRM 7, the mean
HDL-chol recovery was 105.4%, whereas for CRM 9, the mean HDL-chol
recovery was 92.8%. The positive bias obtained in CRM 7 and the
negative bias in CRM 9 can be linked to the dilution or concentration
of these pools, respectively, and point to the sensitivity of the
homogeneous method for large variations in protein concentration.
In CRMs 10 and 11, characterized by moderately increased concentrations
of triglycerides, HDL-chol recoveries were 96% and 99%, respectively,
illustrating the relative insensitivity of the Boehringer direct
HDL-chol method to hypertriglyceridemia.
In the polymer/polyanion method, the HDL-chol concentrations measured
in the CRMs could not be interpreted because an unacceptable,
method-related bias was already disclosed in native human sera during
the pilot HDL-chol survey. The bias issue of the direct
polymer/polyanion method from Genzyme currently is solved in the liquid
formulation of the reagent because the reagent formulation of the
liquid HDL-chol assay differs distinctly from that of the lyophilized
version in that there is no polymer in the first reagent, the magnesium
concentration has been reduced, an HDL-selective detergent has been
introduced in the second reagent, and the chromogens 4-aminoantipyrine
and DSBmT have been segregated into the first and second reagent,
respectively. The different modifications have been aimed successively
at preventing precipitation with alkaline wash solutions on analyzers
with reusable cuvettes; at improving assay specificity, especially
among hypertriglyceridemic specimens; at reducing the background
absorbance during the R1 phase; and at improving the stability of the
reagent blank. In addition, the assay has incorporated a true endpoint
reading that is independent of analyzer cycling time, and the
wavelength selection has been changed (600 nm/700 nm instead of 546
nm/660 nm for the main and subsidiary wavelengths, respectively) to
reduce hemoglobin interference. Consequently, the liquid version of the
Genzyme assay displays an overall mean bias of 
±5% against the
CDC HDL-chol DCM in normotriglyceridemic sera. Moreover, preliminary
data from our group reveal that the CRMs tested perform equally well
with the latest liquid version of the Genzyme direct HDL-chol method
and with the Boehringer direct HDL-chol method.
In conclusion, more effort must be made in The Netherlands to
standardize HDL-chol methods to the CDC Reference Method and/or to the
CDC DCM and to reduce method bias to less than ±5%. To this end, the
lyophilized, saccharose-containing, human sera investigated seem to
have ample potential for use in standardizing homogeneous HDL-chol
methods. More specifically, the selectively collected CRMs 4–6,
prepared from lipoprotein-stabilized serum pools and having a normal
protein matrix with the exception of the presence of 200 g/L
saccharose, seem to be the first choice candidates for future
standardization of homogeneous HDL-chol assays because the mean
HDL-chol concentrations obtained in CRMs 4–6 by the -cyclodextrin
sulfate/PEG-coupled enzyme method were within 1% of the assigned
values obtained by the modified CDC Reference Method. Nevertheless,
more work must be done to investigate whether the data can be
extrapolated to all homogeneous HDL-chol assays. Conversely, the
viscous CRMs 1–11 cannot be used for standardization of the HDL-chol
precipitation methods. However, it can be expected that virtually all
clinical laboratories will switch to direct HDL-chol methods in the
near future, abolishing the need for standardization of precipitation
methods.
|
Appendix 1 |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionAppendix 1References |
|---|
-cyclodextrin. Clin Chem 1995;41:717-723.
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  C. Cobbaert, C. Weykamp, H. Baadenhuijsen, A. Kuypers, J. Lindemans, and R. Jansen Selection, Preparation, and Characterization of Commutable Frozen Human Serum Pools as Potential Secondary Reference Materials for Lipid and Apolipoprotein Measurements: Study within the Framework of the Dutch Project "Calibration 2000" Clin. Chem., September 1, 2002; 48(9): 1526 - 1538. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. M. Weggemans, P. L. Zock, S. Meyboom, H. Funke, and M. B. Katan Apolipoprotein A4-1/2 polymorphism and response of serum lipids to dietary cholesterol in humans J. Lipid Res., October 1, 2000; 41(10): 1623 - 1628. [Abstract] [Full Text]
|
|
|
|
|
|
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 |
