| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Department of Gastroenterology and2 Laboratory Medicine, University Hospital Gasthuisberg, KULeuven, Leuven, Belgium;3 Department of Gastroenterology and4 Immunology Laboratory, Ospedale Mauriziano Umberto I, Torino, Italy;5 INOVA Diagnostics Inc., San Diego, CA;6 Center for Statistics, Limburgs Universitair Centrum, Diepenbeek, Belgium;7 Harvard School of Public Health, Department of Biostatistics, Boston, MA;8 Department of Pathology, University of Iowa, Iowa City, IA
aaddress correspondence to this author at: Laboratory Medicine, University Hospital Gasthuisberg, KULeuven, 3000 Leuven, Belgium; fax 32-16-347931, e-mail Xavier.Bossuyt{at}uz.kuleuven.ac.be
Inflammatory bowel disease (IBD) represents a spectrum of disorders that affect the gastrointestinal tract (1). IBD includes two major entities, Crohn disease and ulcerative colitis (UC). Although the etiology of IBD is unknown at present, it is believed to be an immunologically mediated disease (2). Over the last 40 years, various (auto)antibodies have been described in IBD (3). Anti-Saccharomyces cerevisiae antibodies (4) and perinuclear anti-neutrophil cytoplasmic antibodies (pANCAs) (5) have relatively high prevalence in patients with Crohn disease and UC, respectively. Unlike ANCAs present in vasculitis (6) and in Wegener granulomatosis (7), the exact target antigen of UC-associated pANCAs has not been identified (5)(8).
As a consequence, immunofluorescence microscopy is the only widely available technique for screening for these antibodies. Commercially available substrates, however, are not standardized, and part of the discrepancy in results could be attributable to differences among the substrates/assays used, as reported recently (9). Moreover, because specific microscopic criteria to distinguish UC-associated pANCAs from pANCAs seen in vasculitis vary among laboratories, discrepant results could also be attributable to an investigator’s interpretation of the fluorescence pattern.
Despite these methodologic problems, it has been suggested that the determination of pANCAs in UC could serve as an adjunct to conventional tools in the diagnosis of IBD and could be used for better phenotypic classification of the disease. Therefore, pANCA analysis is widely performed in the context of laboratory evaluation of IBD.
The aim of this study was to assess the interassay and interobserver variability in the detection of UC-associated ANCAs.
Sera obtained from 50 patients with UC (23 females and 27 males; mean age, 40.7 years; range, 19–75 years), defined according to the Lennard-Jones criteria (10), were studied. The clinical data for this study population are shown in Table 1 of the Data Supplement that accompanies the online version of this Technical Brief at http://www.clinchem.org/content/vol50/issue8/. The same patient cohort was tested by all assays and by all laboratories, and serum from each patient originated from the same blood sampling.
Ethanol- and formalin-fixed human neutrophil substrates were used. Specimens with pANCA reactivity associated with vasculitis, as well as with UC, exhibit perinuclear staining in ethanol-fixed neutrophils. In formalin-treated neutrophils, however, the vasculitis-associated pANCA reactivity converts to a granular cytoplasmic (cANCA) pattern. Formalin fixes and prevents the artifactual migration of positively charged cytoplasmic proteins such as myeloperoxidase to the negatively charged nuclear membrane in ethanol-fixed neutrophils, thus leading to the apparent conversion of a vasculitis-associated pANCA pattern to a cANCA pattern. Recently, it has been suggested that, particularly for UC-associated pANCAs, use of both fixatives could be helpful (11).
For evaluation of the interassay variability, the patient cohort was tested by one experienced observer from one laboratory (Laboratory Medicine, University Hospital Gasthuisberg, Catholic University Leuven, Leuven, Belgium), using substrates obtained from four different commercial sources: INOVA Diagnostics, Immunoconcepts, Bio-Rad, and The Binding Site. Each assay was performed according to the manufacturer’s instructions. Manufacturers’ control samples were tested in each run. Sera were diluted in the manufacturer’s specified diluent and incubated for 30 min at room temperature. After a washing step, fluorescein isothiocyanate-labeled goat (INOVA, Immunoconcepts, Bio-Rad) or sheep (The Binding Site) anti-human IgG was added to each slide. The conjugates provided by INOVA, Immunoconcepts, and Bio-Rad were IgG specific, whereas the conjugate provided by The Binding Site was IgG heavy- and light-chain specific. After incubation for 30 min at room temperature, the slides were washed, covered with mounting medium, and covered with a coverslip. Slides were examined under ultraviolet light, and sera that had fluorescence on indirect immunofluorescence microscopy were carefully classified.
For evaluation of interobserver variability, we assayed sera from the same patient cohort as described above for the different ANCA patterns in four geographically distinct laboratories—INOVA Diagnostics (San Diego, CA); Immunopathology Laboratory, University of Iowa Hospitals and Clinics (Iowa City. IA); Immunology Laboratory, Ospedale Mauriziano Umberto I (Torino, Italy); and Laboratory Medicine, University Hospital Gasthuisberg (Leuven, Belgium)—with one company’s method (INOVA). All observers were experienced with reading the results indirect immunofluorescence microscopy. The microscopes used were a Leitz Ariston microscope (Iowa), an Olympus BX40 (Torino), a Nikon Labphot-2 (San Diego), and a Leitz Wetzlar Orthoplan (Leuven).
The prevalences of the ANCA patterns on ethanol-fixed substrate and/or on formalin-fixed substrate were calculated. The 
for nominal data as concordance between multiple rates was used to evaluate agreement (12): <0 indicates poor agreement; 0–0.2 indicates slight agreement; 0.2–0.4 indicates fair agreement; 0.4–0.6 indicates moderate agreement; 0.6–0.8 indicates substantial agreement; and 0.8–1 indicates almost perfect agreement (13).
In the intermethod study, ANCA was assayed with substrates from, respectively, The Binding Site, Bio-Rad, INOVA, and Immunoconcepts (Table 1A
). The prevalence of ANCAs varied for the respective ANCA patterns; i.e., p- and cANCA, pANCA on ethanol-fixed substrates, and pANCA on ethanol-fixed substrates combined with a negative finding on formalin-fixed substrate. Seven samples were identified to be pANCA-positive based on the ethanol-fixed substrate and cANCA-positive based on formalin-fixed substrates. This phenomenon was observed with three of the four substrates tested. However, the finding was not consistent among these assays, except for one sample in which the phenomenon was observed with two assays (INOVA and Immunoconcepts; see Table 1A
).
|
The values for agreement among the methods were 0–0.32 for p- and cANCA, 0.07–0.47 for pANCA on ethanol-fixed substrates, and 0.04–0.38 for pANCA on ethanol-fixed substrates combined with a negative finding on formalin-fixed substrate (see Table 2a in the online Data Supplement). Cytoplasmic staining (cANCA) was found in 1 of 50 with The Binding Site method and in 6 of 50 of the samples with the INOVA method. For all but one assay, a relatively high number of noninterpretable results [in which no (clear) pattern could be distinguished] was observed, ranging from 3 of 50 with the INOVA assay to 5 of 50 with the Immunoconcepts assay and 16 of 50 with The Binding Site assay.
For the interobserver study, the same assay was used by different readers. The prevalences of the respective patterns ANCA, pANCA (ethanol+), and pANCA (ethanol+/formalin–) are shown in Table 1B
and varied between 56% and 70%, between 42% and 64%, and between 40% and 64%. Five samples were reported to be pANCA-positive on the ethanol-fixed substrate and cANCA-positive on formalin-fixed substrates by three laboratories. For one sample, this finding was reported by all three laboratories, whereas for the remaining samples, this observation was unique for each laboratory (see Table 1B
). The values in the interobserver study were 0.36–0.65, 0.35–0.72, and 0.28–0.71 for p- and cANCA, pANCA on ethanol-fixed substrates, and pANCA on ethanol-fixed substrates combined with a negative finding on formalin-fixed substrate (see Table 2a in the online Data Supplement). cANCA was found by all observers: 3 of 50 samples in Torino, 6 of 50 in Leuven, 4 of 50 in INOVA, and 10 of 50 in Iowa. Except for Torino, noninterpretable results were observed in 3 of the 50 samples in Leuven, 7 of 50 in San Diego, and 6 of 50 in Iowa City.
The differences among the commercially available assays for UC-associated pANCA detection were remarkable. These results confirm earlier data of Sandborn et al. (9). We were unable to identify the exact nature of the differences observed. To clarify this issue, a more extensive study is needed in which only one variable is examined at a time (e.g., substrates from different manufacturers with the same conjugate and buffer). This, however, was beyond the scope of the present study.
Better agreement in UC-associated pANCA detection was found when we looked at the interobserver study, although there were differences in the interpretation of results. It has been suggested that differences among assays could be explained by the fact that different assays preferentially detect different antigens (9). Because low values were also seen in the present interobserver variability study, this suggestion is probably only a partial explanation. Immunofluorescence microscopy implies semiquantitative results. Determinations are thus dependent on the expertise of the technician, the variable quality of test reagents, and the equipment used.
cANCA staining patterns were found in this UC study population by all observers and by two of four assays. This observation suggests that sera from a subgroup of UC patients also reacts against a cytoplasmic antigen. This finding confirms previously reported data (5).
Despite the various observations of UC-associated pANCAs, these antibodies have been suggested to be clinically useful, and the controversies on such issues as their predictive value for pouchitis (14)(15)(16)(17) and the appearance of these antibodies in unaffected first-degree relatives of patients with UC (18)(19) may be partially explained by the large interassay and interobserver variability.
Strict guidelines for immunofluorescence detection of UC-associated pANCAs are needed to standardize this currently used technique until a solid-phase assay is available. Since 1993, an international cooperative study group has reported twice the development of a standardized methodology for the detection of cANCAs and pANCAs by indirect immunofluorescence and solid-phase assays (20)(21). These recommendations are widely accepted and serve as the gold standard for detection of Wegener granulomatosis and small-vessel vasculitis. This is not the case for UC-associated pANCAs. Recently, Terjung et al. (11) attempted to define reliable microscopic criteria by use of indirect immunofluorescence microscopy and confocal laser scanning microscopy as well as various fixatives to be able to distinguish, in particular, UC-associated pANCAs from other pANCAs. However, the criteria they defined have not yet been validated internationally. In our study, sera obtained from UC patients produced different patterns on ANCA substrates: pANCA, cANCA, noninterpretable, and negative results on ethanol-fixed substrates and negative or positive (cANCA pattern) results on formalin-fixed substrates.
A 50-kDa myeloid cell-specific nuclear envelope protein has been reported as the target antigen of UC-associated pANCAs. Reactivity to this new identified antigen was found in 92% of sera containing UC-associated pANCAs (22). Identification of the target antigen may not only lead to better understanding of the possible role of these antibodies in the immunopathogenesis of IBD, but may hopefully lead to the development of highly sensitive, specific, and reproducible assays.
Acknowledgments
This work was supported by a grant from the Fund for Scientific Research (Vlaanderen, Belgium) and by Grant MH59532 from the NIH. M. Pierik is aspirant, S. Vermeire is postdoctoral fellow, and X. Bossuyt is a senior clinical investigator of the Fund for Scientific Research–Vlaanderen. We acknowledge Jay McCabe (INOVA) for excellent technical support. We thank the following companies for proving the ANCA assays: INOVA Diagnostics, Bio-Rad (Hercules, CA), Immunoconcepts (Sacramento, CA), and The Binding Site (Birmingham, UK).
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  M. Papp, I. Altorjay, G. Lakos, J. Tumpek, S. Sipka, T. Dinya, K. Palatka, G. Veres, M. Udvardy, and P. L. Lakatos Evaluation of the Combined Application of Ethanol-Fixed and Formaldehyde-Fixed Neutrophil Substrates for Identifying Atypical Perinuclear Antineutrophil Cytoplasmic Antibodies in Inflammatory Bowel Disease Clin. Vaccine Immunol., April 1, 2009; 16(4): 464 - 470. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 X. Bossuyt Serologic Markers in Inflammatory Bowel Disease Clin. Chem., February 1, 2006; 52(2): 171 - 181. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
