| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1
Laboratory Medicine, Immunology Service, University Hospital Leuven, Herestraat 49, B-3000 Leuven, Belgium;
a author for
correspondence: fax 32-16-347010, e-mail Erna.Vanhoeyveld{at}uz.kuleuven.ac.be
Circulating immune complexes (CICs) are detectable in a variety of systemic disorders such as rheumatological and autoimmune diseases (1)(2)(3)(4); allergic diseases (5); viral, bacterial, and parasitic infections (1)(6)(7)(8); and malignancies (2)(9). Although detection of CICs is neither essential nor specific for any disease, it may provide useful clinical information regarding immunopathology, prognosis, and follow-up of rheumatic and autoimmune disorders.
A variety of tests that can detect CICs have been described (10)(11)(12)(13)(14)(15)(16)(17)(18). The assays that are used routinely in clinical laboratories are based on (a) precipitation of CICs by polyethylene glycol, (b) interaction of CICs with complement (C1q assay), (c) detection of CICs bound to C3 (Raji cell assay, conglutinin assay), and (d) binding of CICs to Fc-recognizing molecules (mRF assay). In a comparative WHO study (10), only a few tests were described as sufficiently sensitive and reproducible. This study revealed that the C1q solid-phase RIA (C1q-SP RIA) discriminated well between sera from healthy persons and those from patients. Several studies have indicated that this assay correlated better with disease activity than do other immune complex tests (8). The clinical activity of systemic lupus erythematosus correlated well with the concentration of CICs in the solid-phase RIA (19)(20). Subsequently, this assay has been used frequently in clinical investigations. The working group therefore concluded that the C1q-SP test was one of the most reliable tests. For these reasons, we used this test as the reference method.
Several ELISA kits have become commercially available over the last decade. However, their clinical usefulness is not clearly established. In the present study, we evaluated the concordance, sensitivity, and specificity of seven commercially available ELISA assays (four C1q or C1q-based binding assays and three C3d-based binding assays) by comparing their results with the results of the C1q-SP RIA.
In every assay, sera from 15 healthy donors were used as the negative control population. Patients sera with different positivities in the C1q-SP RIA were selected. Assays were performed in parallel, and sera were analyzed in duplicate. The C1q-SP RIA was performed as described previously (21). For interpretation of results, the mean of 15 control sera from healthy individuals was equalized to 100%. Results of patient sera were then expressed as the percentage of this mean and graded negative when less than the mean + 2 SD, equivocal when between the mean + 2 SD and the mean + 4 SD, and positive when more than the mean + 4 SD.
ELISA kits were obtained from Cogent Diagnostics, Bühlmann Laboratories AG, Quidel, and Scimedx Corporation. Serum CICs were measured according to the manufacturers’ instructions. Results were graded as positive or negative according to the scale provided with each kit. All assays used diluted serum (1:5 to 1:50), three incubation periods, horseradish peroxidase, absorbance readings at 405 or 450 nm and two to six calibrators. For coating, C1q was used for the Bühlmann and Quidel assays, and a monoclonal C1q antibody was used for the Cogent and Scimedx assays. Monoclonal C3d antibody was used in the Cogent and Scimedx assays, whereas the Raji cell replacement assay used monoclonal C3bi and C3d.g.
The results obtained with the C1q- and C3d-based ELISAs or the Raji cell replacement ELISA were compared with the results obtained with the C1q-SP RIA and with one another. A class comparison was done. Data in the same or adjacent classes were considered concordant. Samples with a difference of at least two classes were considered discrepant. For calculation of sensitivity and specificity, equivocal results were considered negative.
The comparisons between the various ELISAs and the RIA for the
detection of CICs are presented in Table 1
A.
|
Agreement between the C1q-based ELISAs and the C1q-SP RIA was 49%,
50%, 61.5%, and 61% for the Cogent, Quidel, Bühlmann, and
Scimedx assays, respectively. Equivocal results in the RIA and either
positive or negative results in the ELISA were found in 16%, 10%,
32%, and 12% of the samples for the Cogent, Quidel, Bühlmann,
and Scimedx assays, respectively. The percentage of samples that were
negative in the RIA and positive in the ELISA or positive in the RIA
and negative in the ELISA was 35% for the Cogent, 40% for the Quidel,
6.5% for the Bühlmann, and 27% for the Scimedx assay. When
concordant and one-class discordant samples were grouped, the present
study demonstrated the highest agreement between the Bühlmann C1q
assay and the C1q-SP RIA (94%). A possible explanation is that both
assays use C1q to detect CICs. Although the Quidel C1q assay also uses
C1q to detect CICs, concordance between this assay and the C1q-SP RIA
(Table 1
) and the Bühlmann assay (Table 1
) was only 60% and
68%, respectively. The agreement between the Cogent and Scimedx
C1q-based assays and the C1q RIA was moderate (65% and 73%,
respectively). In these ELISA assays, a monoclonal anti-C1q antibody is
used to capture CICs. As a consequence, these assays detect CICs that
already bear C1q. As might be expected, there was excellent agreement
(95%) between the Cogent maC1q-based assay and the Scimedx maC1q-based
assay (Table 1
).
Agreement between the C1q-SP RIA and the two C3d-based assays and the
Raji cell ELISA was 55%, 58%, and 63% for the Cogent, Quidel, and
Scimedx assays, respectively. In 23%, 14%, and 12% of the samples,
the RIA gave equivocal results, whereas the Cogent, Quidel, and Scimedx
assays gave either positive or negative results. The percentage of
samples that were RIA-negative and ELISA-positive or RIA-positive and
ELISA-negative was 22% for the Cogent, 28% for the Quidel, and 25%
for the Scimedx assay (Table 1
). Thus, the degree of concordance
between each of the C3d-based ELISAs and the C1q-SP RIA was moderate
and varied from 71% to 78%. In contrast, the agreement of the three
C3d-based assays with one another was >90% (Table 1
). In the
C3d-based ELISAs, a monoclonal aC3d antibody or a monoclonal aC3bi,
C3d.g antibody is used to detect CICs. In these cases, immune complexes
that contain C3d are detected. The concordance between the C3d-based
assays and the C1q-based assays, with the exception of the
Bühlmann C1q assay, was 81–91%.
When the C1q-SP RIA was considered the reference method, the
sensitivities of the C1q ELISAs from Cogent, Quidel, Bühlmann,
and Scimedx were 16%, 12%, 65%, and 29%, respectively. The
specificities of these assays were 97%, 98%, 93%, and 89%,
respectively (Table 1
). For the C3d-based ELISAs from Cogent,
Quidel, and Scimedx, the sensitivities were 41%, 48%, and 36%,
respectively, and the specificities were 90%, 91%, and 85%,
respectively (Table 1
).
Our observations illustrate that concordance is highest between assays that are based on the same principle. With the C1q-based assays, interference by anti-C1q, free C1q, or other complement components is possible. The anti-C3d-based assays can detect CICs that activate complement via the classical and the alternate pathways (18). Thus, various methods may detect different kinds of complexes. Therefore, the WHO study recommends use of at least two test systems with different specificities for antibody classes and complement fragments (10).
In conclusion, CIC assays are much simplified by modification to ELISA technology. The Bühlmann C1q assay may be a good ELISA alternative for the C1q-SP RIA. The maC1q assays and the maC3d assays agreed better with one another than with the C1q binding assay. This suggests that they measure a different variable. One of these assays, therefore, might be a good candidate as a second test system as recommended by the WHO. Their clinical importance, however, should be further studied.
Acknowledgments
We thank F. Van Acker for helpful discussions and A. Mewis, J. De Romagnoli, and J. L’Heureux for expert technical assistance. We acknowledge DPC (Humbeek-Grimbergen, Belgium), Forlab (Brussels, Belgium), Alphadia (Waver, Belgium), and Medigal (Villers-Poterie, Belgium) for providing the ELISA kits.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
