| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Department of Medicine, Laboratory of General Internal Medicine, University Medical Center St. Radboud, Nijmegen, The Netherlands;
aaddress correspondence to this author at: Laboratory of General Internal Medicine, University Medical Center Nijmegen, PO Box 9100, 6500 HB Nijmegen, The Netherlands; e-mail P.Demacker{at}AIG.UMCN.NL
Hyperbolic relationships can be expected when comparing data obtained by apolipoprotein(a) [apo(a)] isoprotein-sensitive assays against data obtained by an isoprotein-independent assay (x axis), and almost all polyclonal antibody-based apo(a) assays are isoprotein sensitive (1).
We have developed a dual polyclonal antibody ELISA for quantification of plasma apo(a) and attempted to standardize it with regard to its accuracy in nmol/L while calibrating on the IFCC PRM2 calibrator. We also followed the traditional approach and compared the results with other established methods that express the results in mass or molar units. We found that our assay gave linear relationships with two other assays, of which at least one was isoprotein-independent. We confirmed the apo(a) isoprotein dependency by testing the IFCC PRM-2 calibrator and material differing in apo(a) molecular weight. Our results suggest that polyclonal antibody-based apo(a) assays a priori are not isoprotein-dependent. The work also gives insight into the suitability and comparativity of three different calibrators with target values in molar or mass units. Here we discuss our experiences in the hope that they are helpful for others who also want to standardize apo(a) assays.
apo(a) is a glycoprotein bound to LDLs; a small amount may also circulate in free form (1)(2). There is no doubt that the standardization of total apo(a) measurements in routine laboratories world-wide needs improvement (1). The main problem in standardization is the specificity of the detecting antibody because of the repetitive character of the apo(a) kringles. Molar detection is possible only with monoclonal antibodies that recognize a nonrepetitive epitope on the apo(a) antigen. Polyclonal antibody-based assays are not expected to be accurate, but recalculation of the primary results with use of a common calibrator considerably decreased the interlaboratory CV at a certain concentration (1). We reasoned that a plasma pool as calibrator probably could assure accuracy over the whole pathophysiologic concentration range of apo(a). In the present study we tested and confirmed this presumption for two ELISA-based assays.
EDTA-containing plasma samples were aliquoted and stored frozen until being assayed. Descriptions of the development of the dual polyclonal apo(a) ELISA and details on the apo(a/a) Delfia and the homogeneous IRMA method are provided in the Data Supplement that accompanies the online version of this Technical Brief at http://www.clinchem.org/content/vol50/issue11/. In the IRMA, results are expressed primarily in U/L, where 1 U is 
0.7 mg. Also described in the online Data Supplement are the characteristics of the three calibrators used: the WHO-IFCC PRM-2; the International Reference Standard Human apo(a); and the Nijmegen plasma pool.
Because the apo(a) concentration is inversely correlated with the size of the apo(a) molecule (3), we determined apo(a) concentrations in samples containing a wide range of apo(a) concentrations to test the isoprotein-related bias of our ELISA.
With our dual-antibody ELISA, we were post hoc allowed to participate in the IFCC apo(a) Standardization Program. For this purpose, five lyophilized sera with different apo(a) isoproteins were analyzed, with the lyophilized PRM-2 used not only as calibrator but also as an unknown sample with a target value of 107 nmol/L [as determined by Marcovina et al. (4) with the reference ELISA method, which is insensitive to isoprotein variation]. The biases varied between –17 and +18 nmol/L (Table 1
). When corrected for the between-day CV of PRM-2 measured as unknown sample (8%), the biases for samples L1–L5 decreased to <7 nmol/L and imprecision improved considerably (Table 1
). The results show that our dual-antibody apo(a) assay is isoprotein-independent.
|
We measured apo(a) in 267 plasma samples with the homogeneous IRMA (x) method and with the newly developed ELISA (y), with their respective calibrators number 2 and 3, respectively, in mass units. The samples were selected to cover a wide range of apo(a) concentrations, including samples with concentrations greater than 500 mg/L, which were diluted fourfold. The following relationship was obtained: y = (0.80 ± 0.01)x; r = 0.98 (see Fig. 1 in the online Data Supplement). The relationship was linear over the whole curve, and the line was straight, ruling out any isoprotein dependency and any lack of linearity attributable to dilution. It is likely that the slight systematic bias between both assays can be corrected by use of a common calibrator, preferably a serum pool.
When we assayed 11 plasma samples with a wide range of apo(a) concentrations with the apo(a/a) Delfia (x) calibrated with the IFCC PRM-2 in nmol/L (and recalculated in mg/L with use of a mean mass number of 317 kDa), we obtained results that agreed well with the newly developed dual polyclonal apo(a) ELISA (y): y = (0.99 ± 0.04)x + (12 ± 15) mg/L; r = 0.99; n = 11 (see Fig. 2 in the online Data Supplement). These results suggest that the apo(a/a) Delfia is an isoprotein-independent assay as well. In addition, both calibrators used were equivalent.
In our ELISA, the IFCC PRM-2 calibrator had only a limited linear range: results were linear only up to 900 milliabsorbance units, corresponding to 0.09 nmol/L apo(a), vs 2200 milliabsorbance units for all other samples tested, including our calibrator (1.24 nmol/L). Consequently, samples with high apo(a) concentrations had to be tested in a higher dilution to assure linearity. The observed matrix effects in our ELISA for PRM-2 were not observed with various plasma samples.
Classic methods for apo(a) measurement involve the use of polyclonal rather than of monoclonal antibodies, and it was thought that interlaboratory standardization would be simple, similar to the assays for apolipoproteins A-1 or B, for which a common pool suffices to decrease interlaboratory CVs to acceptable values. Marcovina et al. (1) showed that standardization of apo(a) measurements was more complex. Using a common standard, they obtained an among-laboratory CV of only 2.8% for 22 different systems, but for 17 of those systems, they observed hyperbolic relationships between results obtained by the system under study vs a system that was known to be isoprotein independent, which means that results obtained for samples with apo(a) concentrations substantially higher or lower than the PRM-2 were significantly biased. We show here that when a proper calibrator is used, two dual polyclonal antibody assays gave results that were similar to those obtained with an assay in which detection is on a molar basis only. We wonder whether this observation can be generalized (for additional discussion, see the online Data Supplement).
We are not certain that the matrix effects we observed for PRM-2 were also present in the study by Marcovina et al. (1) of various apo(a) immunoassays, not only because of the time interval of 4 years but also because of the use of different methodologies (for additional discussion, see the online Data Supplement).
Acknowledgments
We thank Prof. Santica Marcovina, North West Lipid Research Clinic (Seattle, WA) for the gift of PRM-2 samples and for discussions, and Prof. Gerhard Kostner, Institute of Medical Biochemistry, University of Graz (Graz, Austria) for measuring apo(a) concentrations by the Delfia method.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
