Agreement of anti-neutrophil cytoplasmic antibody measurements obtained from serum and plasma

Abstract

Serum and plasma are used interchangeably to measure anti-neutrophil cytoplasmic antibodies (ANCA), even though the release of ANCA target antigens during the preparation of serum could affect ANCA assays and cause discrepancies between the results obtained from serum and plasma. To what extent ANCA test results obtained from serum agree and correlate with results from plasma remains unknown. Therefore, a comprehensive comparison was performed using serum and plasma samples which were collected in 175 patients with active Wegener's granulomatosis at enrolment of a recent randomized trial. These paired serum and plasma samples were subjected to parallel ANCA testing by standard indirect immunofluoresence on ethanol-fixed neutrophils, a direct enzyme-linked immunoassay (ELISA) for proteinase 3 (PR3)-ANCA and myeloperoxidase (MPO)-ANCA, and two different capture ELISAs for PR3-ANCA. The concordance of categorical serum and plasma ANCA results was assessed using κ-coefficients. These were > 0·8 for all assays, indicating a very good concordance between positive and negative serum and plasma results. Spearman's correlation coefficients for serum and plasma PR3-ANCA values obtained by direct ELISA and both capture ELISAs were ≥ 0·95 (P < 0·0001). Our study shows that serum and plasma samples can be used interchangeably for measuring ANCA.

Introduction

Testing for anti-neutrophil cytoplasmic antibodies (ANCA) reacting with proteinase 3 (PR3) or myeloperoxidase (MPO) has become a routine part of the diagnostic evaluation of patients with suspected ANCA-associated vasculitis, such as Wegener's granulomatosis or microscopic polyangiitis. Currently, the principal methods used for ANCA detection are indirect immunofluorescence, antigen-specific direct enzyme-linked immunosorbent assay (ELISA) and antigen-specific capture ELISA, and despite ongoing multi-national efforts, these methodologies have not been standardized [1–4].

With few exceptions [5], routine ANCA testing is performed in serum because of historical reasons rather than a scientific rationale [6]. However, sometimes plasma is the only biospecimen available for study, and positive controls utilized as standards are usually obtained by plasmapheresis. Even though serum and plasma are used interchangeably to measure ANCA, the release of ANCA target antigens during the preparation of serum could affect ANCA assays and cause discrepancies between results obtained from serum and plasma, similar to assays for other factors affected by coagulation [7–10]. Finally, as functional and epitope mapping studies aimed at clarifying the pathogenic potential of ANCA move forward, it is important to know how data obtained using plasma samples relate to corresponding data obtained with serum, and vice versa. To our knowledge, no formal evaluation of ANCA detection methods performed in serum versus plasma has ever been published. Therefore, we performed this comprehensive comparison of ANCA test results obtained from simultaneously prepared serum and plasma samples.

Materials and methods

Serum and plasma samples

Samples used in this study were obtained from the Wegener's Granulomatosis Etanercept Trial (WGET), a multi-centre,randomized, placebo-controlled trial that evaluated etanercept for maintenance of remission in 180 patients with Wegener's granulomatosis [11]. During the trial, serum and plasma samples were collected simultaneously at entry (baseline), at 6 weeks, at 3 months and then every 3 months until the close of the study. For this study, only the matching baseline serum and plasma samples were used.

Whole blood (10 ml) was collected into evacuated blood-collecting tubes with lithium or sodium heparin (plasma), and without additives (serum). The former tubes were processed within 1 h at 4°C, and the latter were allowed to clot for at least 2 h at room temperature or for up to 24 h at 4°C. Samples were then centrifuged at 800 g for 10 min, and 1 ml aliquots of both plasma and serum were prepared, frozen and stored at − 80°C until analysed.

The WGET protocol was approved by the Institutional Review Board at each participating centre. Informed written consent was obtained from all participants. Details of the study design, patient characteristics and trial results have been published previously [11,12].

ANCA detection methods

Standard immunofluorescence was performed using ethanol-fixed neutrophils as described previously [13]. Samples were categorized as ‘C-ANCA positive’ if the characteristic centrally accentuated granular cytoplasmic staining pattern was detectable at a 1 : 8 dilution, as ‘P-ANCA positive’ if they caused a perinuclear or nuclear staining pattern. Fluorescence patterns not clearly identifiable as C-ANCA or P-ANCA were categorized as ‘indeterminate’, and the absence of any fluorescence as ‘negative’. Titre determinations were not performed for the purpose of this study.

Direct ELISAs for PR3-ANCA and MPO-ANCA were performed using commercially available kits (Scimedx, Corporation, Denville, NJ, USA) according to the manufacturer's instructions. A value > 5 EU/ml was considered positive for both assays. Details regarding the characterization of the target antigen used in these assays are not provided by the manufacturer.

Two validated capture ELISAs were also employed for PR3-ANCA detection. In the MCPR3-2 capture ELISA, a monoclonal anti-PR3 antibody (MCPR3-2) was used as the capturing antibody, mature-PR3 as the captured antigen, and a conjugated goat anti-human IgG as the detecting antibody [13–15]. A net absorbance ≥ 0·10 was considered positive, and the inter- and intra-assay coefficients of variation for this assay were 31 and 13%, respectively [14].

In the anti-c-myc capture ELISA, a recombinant c-myc tagged mature-PR3 was used as the antigen, which was captured onto plates coated with an anti-c-myc monoclonal antibody (Sigma P2241) as described recently [16]; conjugated goat anti-human IgG was used as the detecting antibody. A net absorbance ≥ 0·01 was considered positive, and the inter- and intra-assay coefficients of variation for this assay were 18·4 and 7·7%, respectively [16]. Sample dilutions of 1 : 20 were used in all solid phase assays.

Data analysis and statistical methods

Serum and plasma samples were tested in parallel by indirect immunofluorescence, and by direct and capture ELISAs. Agreement of categorical positive and negative ANCA test results obtained from serum and plasma was assessed using κ-coefficients (< 0·20, poor; 0·21–0·40 fair; 0·41–0·60 moderate; 0·61–0·80 good; 0·81–1·00 very good) and the McNemar's test. To determine the correlation of ANCA levels in serum and plasma, Spearman's correlation coefficient was calculated for each ELISA method. To assess further the agreement between ANCA measurements in serum and plasma, the mean difference was obtained and the limits of agreement were calculated as the mean difference ± 2 standard deviations (s.d.) [17]. The 95% confidence intervals (CI) were calculated when needed. P-values < 0·05 were considered statistically significant. Statistical analyses were performed using GraphPad Prism 4·0b for Macintosh (GraphPad Software, San Diego, CA, USA).

Results

Of the 180 patients who participated in the WGET, 175 pairs of serum and plasma samples (one pair per patient at baseline) were available to be tested in parallel by all ANCA detection methods. Table 1 shows that the concordance of ANCA test results obtained from serum and plasma was very good for all the methods. The majority of the discordant results occurred in patients who had very low ANCA levels, yielding weak positive results in either serum or plasma. These individual results are provided in the footnote to Table 1. Testing of serum yielded a few more positive results than plasma, but these differences reached statistical significance only for the MCPR3-2 capture ELISA.

Table 1.

Concordance of anti-neutrophil cytoplasmic antibodies (ANCA) detection methods in 175 pairs of serum and plasma samples.

	κ	Concordant (%)	Concordant positive (n)	Concordant negative (n)	Discordant (%) (95% CI)	Serum positive/ plasma negative (n)	Serum negative/ plasma positive (n)	P*
C-ANCA IIF**	0·83	92·6	119	42	7·4 (4·3–12·4)	7	1	0·077
P-ANCA IIF†	0·89	97·7	18	153	2·3 (0·7–6·0)	3	0	1·000
PR3-ANCA direct ELISA	0·89	95·4	122	45	4·6 (2·2–8·9)	7‡	1¶	0·077
MPO-ANCA direct ELISA	0·94	99·4	9	165	0·6 (0·0–3·5)	0	1§	1·000
MCPR3-2 capture ELISA	0·90	96·0	122	46	4·0 (1·8–8·2)	7††	0	0·023
Anti-c-myc capture ELISA	0·88	95·4	127	40	4·6 (2·2–8·9)	2‡‡	6§§	0·289

For both direct enzyme-linked immunosorbent assays (ELISAs), a value > 5 EU/ml was considered positive. For the MCPR3-2 and anti-c-myc capture ELISAs, a net absorbance ≥ 0·10 and ≥ 0·01 are considered positive, respectively. IIF: immunofluorescence; CI: confidence interval.

^*P-values were calculated using the McNemar's test based on 2 × 2 cross-tabulation of the results.

^**In six pairs there were indeterminate samples (in two, serum was negative while plasma indeterminate; in another two, serum was indeterminate while plasma positive; in one, serum was indeterminate while plasma negative; and in one both serum and plasma were indeterminate).

^†In one pair serum was serum positive while plasma was indeterminate.

^‡Serum results were 5·2, 5·6, 5·7, 6·2, 6·5, 6·6 and 9·9 EU/ml.

^¶Plasma result was 170·9 EU/ml.

^§Plasma result was 5·1 EU/ml.

^††Serum net absorbances were 0·11, 0·12, 0·23, 0·28, 0·30, 0·38 and 0·44.

^‡‡Serum net absorbances were 0·02 and 0·48.

^§§Plasma net absorbances were 0·03, 0·05, 0·06, 0·08, 0·28 and 0·99.

These analyses were repeated in only the 58 matched samples representing the lower tercile of the test results (based on the serum samples) obtained by PR3-ANCA direct ELISA, and MCPR3-2 and anti-c-myc capture ELISAs. The concordance was good for these three methods (data not shown).

There was a strong correlation between serum and plasma PR3-ANCA results obtained by direct ELISA (Fig. 1a, left panel), and the mean difference (s.d.) was 5·6 (37·6) EU/ml (Fig. 1a, right panel). Similarly, for MCPR3-2 and anti-c-myc capture ELISAs, the correlations between serum and plasma were strong (Fig. 1b,c, left panels). The mean differences (s.d.) of the net absorbances were 0·06 (0·17) for the MCPR3-2 capture ELISA and 0·15 (0·25) for the anti-c-myc capture ELISA (Fig. 1b,c, right panels).

Fig. 1.

Correlation (left panels) and Bland–Altman plots (right panels) of proteinase 3-anti-neutrophil cytoplasmic antibodies (PR3-ANCA) results obtained from serum and plasma by direct enzyme-linked immunosorbent assay (ELISA) (a), MCPR3-2 capture ELISA (b), and anti-c-myc capture ELISA (c).

Only 10 patients had positive MPO-ANCA results by direct ELISA. The correlation between serum and plasma values was strong, r = 0·99 (P < 0·0001).

Discussion

Our study shows that the concordance of positive and negative ANCA results obtained from serum and plasma for all principal methods of ANCA detection was very good, and that ANCA levels in serum and plasma measured by solid phase assays correlated well. These findings were derived from a large number of paired samples, which cover the entire range of ANCA levels, and from a patient population that represents both the clinical disease spectrum and the distribution of ANCA types in Wegener's granulomatosis. Furthermore, all assays were performed under ideal conditions.

For some biological markers, the choice of measuring them in serum or plasma has been the subject of debate [8–10]. For most circulating antibodies, measurements in serum or plasma are thought to yield equivalent results. However, in the case of ANCA, the release of ANCA target antigens from neutrophils occurring during in vitro coagulation [7] could affect ANCA assays and cause discrepancies between results obtained from serum and plasma. It has been speculated that during serum preparation, the release of enzymatically active PR3 could ‘artefactually’ decrease ANCA levels by proteolytic degradation or by in vitro immune-complex formation [14,18–20], making plasma the preferable alternative for ANCA testing. Indeed, when PR3 levels and PR3-ANCA immune-complexes were measured in serum and plasma samples prepared at different times after venipuncture, PR3 levels were found to increase over time during serum preparation, but not during plasma preparation [20]. However, this observation was based on experimental data obtained from a single patient with Wegener's granulomatosis [20].

Additionally, different methods of ANCA detection may be affected differently by these variables. In capture ELISAs, PR3-specific immune-complexes increase the background, reducing the net absorbance [14]. Consequently, because of the in vitro release of PR3 occurring during serum preparation, we expected the capture ELISA to detect fewer positives in serum compared to plasma, particularly at lower levels of ANCA. However, this prediction was not supported by our results. In fact, although not statistically significant, we found a tendency to detect a few more positives in serum than in plasma for most ANCA detection methods.

Significant limitations of our study are the small number of MPO-ANCA positive samples included in this patient cohort, and that only one MPO-ANCA detection method was used. Therefore, only limited conclusions can be drawn regarding MPO-ANCA from this study. In spite of these limitations, our study shows that serum and plasma samples can be used interchangeably for measuring ANCA. Our findings validate the prevailing use of serum for routine ANCA testing, and the use of plasma samples as standards for those assays.

Appendix

The WGET Research Group

Chairman's Office	The Johns Hopkins Vasculitis Center
	John H. Stone MD, MPH, Chairman
	Misty L. Uhlfelder MPH, Assistant Trial Coordinator
	Amanda M. Moore BS, Research Coordinator
Co-Chairman's Office	The Cleveland Clinic Foundation Center for Vasculitis Research and Care
	Gary S. Hoffman MD, Co-Chairman
Coordinating Center	The Johns Hopkins University Center for Clinical Trials
	Janet T. Holbrook PhD, MPH, Director
	Curtis L. Meinert PhD, Associate Director
	John Dodge, Systems Analyst
	Jessica Donithan, Research Coordinator
	Nancy Min PhD, Biostatistician
	Laurel Murrow MSc, Trial Coordinator (former)
	Jacki Smith, Research Data Assistant
	Andrea K. Tibbs BS, Trial Coordinator
	Mark Van Natta MHS, Biostatistician
Clinical Centers	The Beth Israel Medical Center, New York
	Robert Spiera MD
	Rosanne Berman MPH
	Sandy Enuha MPH
	Boston University
	Peter A. Merkel MD, MPH
	Rondi Gelbard BS
	Melynn Nuite RN
	Aileen Schiller MS
	The Cleveland Clinic Foundation
	Gary S. Hoffman MD, MS
	David Blumenthal MD
	Debora Bork MFA
	Tiffany Clark CNP
	Sonya L. Crook RN
	Leonard H. Calabrese DO
	Sharon Farkas
	Sudhakar Sridharan MD
	Kimberly Strom CNP
	William Wilke MD
	Duke University
	E. William St Clair MD
	Nancy B. Allen MD
	Karen Rodin RN
	Edna Scarlett
	Johns Hopkins University
	John H. Stone MD, MPH
	David B. Hellmann MD
	Amanda M. Moore BS
	Lourdes Pinachos RN, BSN
	Michael J. Regan MD, MRCP
	Misty L. Uhlfelder MPH
	The Mayo Clinic
	Ulrich Specks MD
	Kristin Bradt
	Kimberly Carlson
	Susan Fisher RN
	Boleyn Hammel
	Kathy Mieras
	Steven Ytterberg MD
	University of California, San Francisco
	John C. Davis MD, MPH
	Maureen Fitzpatrick MPH
	Ken Fye MD
	Steve Lund MSN, NP
	University of Michigan
	Joseph McCune MD
	Billie Jo Coomer BS
	Barbara Gilson RN
	Hilary Haftel MD
	Ana Morrel-Samuels BA
	Sandra Neckel RN
Resource Centers	The Johns Hopkins University Immune Diseases Laboratory
	Noel R. Rose MD, PhD
	C. Lynne Burek PhD
	Jobert Barin BS
	Monica Talor MS
Data and Safety Monitoring Board	Paul L. Canner PhD, Maryland Medical Research Institute
	Doyt L. Conn MD, Emory University (Safety Officer)
	Jack H. Klippe, MD, Arthritis Foundation (Chair)
	J. Richard Landis PhD, University of Pennsylvania
	Barbara White MD, University of Maryland