Skip to main content
NCBI home page
Rheumatology (Oxford, England) logoLink to Rheumatology (Oxford, England)
. 2011 Oct 28;51(1):100–109. doi: 10.1093/rheumatology/ker280

Value of ANCA measurements during remission to predict a relapse of ANCA-associated vasculitis—a meta-analysis

Gunnar Tomasson 1,✉, Peter C Grayson 1, Alfred D Mahr 2, Michael LaValley 3, Peter A Merkel 1
PMCID: PMC3276294  PMID: 22039267

Abstract

Objective. The value of repeated ANCA measurements among patients with an established diagnosis of ANCA-associated vasculitis (AAV) remains controversial. The aim of this study was to explore whether either of the two distinct patterns of ANCA values during remission, a rise in ANCA or persistently positive ANCA, predicted future relapse.

Methods. MEDLINE and EMBASE searches were performed. Studies with at least 10 subjects with AAV from which both sensitivity and specificity of a rise in ANCA and/or persistent ANCA for future disease relapse could be calculated were included. Likelihood ratios were calculated for each study and pooled to arrive at summary estimates. I2-values were calculated as a measure of heterogeneity and meta-regression was used to explore sources of heterogeneity.

Results. Nine articles on a rise in ANCA and nine articles on persistent ANCA were included. The summary estimates for positive likelihood ratio (LR+) and negative likelihood ratio (LR−) of a rise in ANCA during remission on subsequent relapse of disease were 2.84 (95% CI 1.65, 4.90) and 0.49 (95% CI 0.27, 0.87), respectively. The summary estimates for LR+ and LR− of persistent ANCA during remission for subsequent disease relapse were 1.97 (95% CI 1.43, 2.70) and 0.73 (95% CI 0.50, 1.06), respectively. There was substantial between-study heterogeneity, which was partially explained by the frequency of ANCA measurements.

Conclusion. Among patients with AAV, a rise in or persistence of ANCA during remission is only modestly predictive of future disease relapse. There is limited use to serial ANCA measurements during disease remission to guide treatment decisions for individual patients with AAV.

Keywords: vasculitis, anti-neutrophil cytoplasmic antibodies, biomarker

Introduction

ANCAs have an undisputed role in establishing the diagnosis of granulomatosis with polyangiitis (Wegener's; GPA) and microscopic polyangiitis (MPA); these diseases, along with Churg–Strauss syndrome (CSS), are often collectively referred to as ANCA-associated vasculitis (AAV) [1–5]. There are data supporting a role for ANCA in the pathogenesis of vasculitis [6] and ANCA has been advocated as a biomarker of disease activity in AAV [7]. Initial reports that serial ANCA measurements were measures of disease activity and valuable predictors of relapses of vasculitis [8, 9] were followed by additional studies that did not support a role for ANCA in patient management [10, 11]. With multiple studies arriving at conflicting conclusions on an important clinical question, a meta-analysis of the published literature could provide estimates for the value of serial ANCA measurements for predicting relapse, while taking into account the heterogeneity between individual studies. Such an analysis could also identify patient subgroups for which serial ANCA testing might be particularly useful and provide possible explanations as to why studies arrive at different conclusions.

Our objective was to specifically examine the usefulness of two distinct properties of ANCA to guide management in AAV: (i) whether a rise in ANCA during remission is predictive for future relapse; and (ii) whether those subjects with persistently positive ANCA during remission are more likely to have relapse of disease compared with those that become ANCA negative after remission is achieved.

Patients and methods

Literature search

We searched MEDLINE and EMBASE without language restrictions from inception through the first week of August 2009 for articles focusing on ANCA measurements for prediction of relapse AAV. Our search was based on various combinations of the following indexed medical subject heading (MeSH) terms: (i) Antibodies, Antineutrophil Cytoplasmic; (ii) Sensitivity and Specificity; (iii) Recurrence; (iv) ROC curves; (v) Vasculitis; and (vi) Longitudinal studies and related non-indexed text words (full search strategy available upon request). In addition, reference lists of retrieved studies and review articles were reviewed, as were abstracts from the recent (2007–09) annual meetings of the American College of Rheumatology, the European League Against Rheumatism and the American Society of Nephrology.

Study selection

One investigator (G.T.) reviewed the study titles and/or abstracts and selected reports for full-text review. Two investigators (G.T. and P.C.G.) reviewed the studies for final determination of inclusion or exclusion, except for two studies in French, which were reviewed by one investigator (A.D.M.), and their content discussed. In the case of a disagreement, a consensus was reached through discussion and any remaining disagreement was adjudicated by a third investigator (P.A.M.). Authors of the primary studies were contacted if clarification or only minimal data elements were needed for a study to meet the criteria for inclusion.

Data from articles were included in the meta-analysis if they met the following criteria: serial ANCA measurements were performed during remission of vasculitis on at least 10 patients and data could be extracted to calculate both sensitivity and specificity for a rise in or persistence of ANCA for future relapse. Articles were excluded from the analysis if the ANCA rise was only concurrent with relapse of disease, if only the number of relapses but not the number of patients with and without relapse during follow-up could be extracted or if ANCA levels were used to define remission or relapse. After a final set of included studies was identified, the papers were reviewed with respect to the possibility of double-counting subjects who might be included in more than one paper.

Data extraction

Data were extracted by two investigators (G.T. and P.C.G.) from the selected studies using a predefined extraction form that included information on type of vasculitis, duration of follow-up, definitions of remission and relapse, frequency of ANCA measurements, laboratory method of ANCA measurements (IF or ELISA), total number of patients who had serial ANCA testing performed during remission, total number of patients who experienced relapse of disease, number of patients who had a rise in or persistence of ANCA before relapse and number of patients who had a rise in or persistence of ANCA not followed by a relapse of disease. Patients described as having renal-limited vasculitis and/or an ANCA-positive necrotizing vasculitis other than GPA or CSS were considered as having MPA.

Statistical analysis

Positive likelihood ratio (LR+) and negative likelihood ratio (LR−) were calculated for each study and pooled separately using random-effects models. The standard errors for LRs were calculated as previously described [12] and 0.5 added to all counts for those studies that contained a zero count for true positives, false positives, true negatives or false negatives. Summary estimates for sensitivities and specificities were calculated using a bivariate model accounting for the correlation between sensitivity and specificity [13, 14]. To check for consistency, LRs were also back-calculated from the summary estimates of sensitivity and specificity obtained from the bivariate model [14]. Hierarchical summary receiver operator curves (HSROCs) were generated as previously described [15] to acount for the possible use of different thresholds of rise in and/or persistence of ANCA used by individual studies. All summary estimates are expressed with 95% CIs.

To evaluate for heterogeneity and publication bias, the arcsine transformation [16] of the data was used. This effect size is calculated as the difference in arcsine of the proportion of subjects with a rise in ANCA among those who had relapse and those without relapse. The I2-value was calculated and expressed with 95% CI [17]. To identify sources of heterogeneity, meta-regression was performed with the following independent variables: (i) method of ANCA measurement (IF vs ELISA); (ii) type of antibody (cANCA/anti-PR3 vs pANCA/anti-MPO); a variable was created for each study for the proportion of patients positive for cANCA/anti-PR3; (iii) whether or not ANCA measurements were conducted according to a predefined protocol; (iv) frequency of ANCA measurements; (v) duration of follow-up; and (vi) year of publication. To quantify how individual independent variables in meta-regression affect between-study heterogeneity, we calculated τ2 values from the intercept-only models and the models containing the independent variable of interest [18]. A funnel plot for publication bias was constructed and an Egger test for bias was performed [19]. All statistical analyses were done using SAS version 9.1 (SAS Inc., Cary NC, USA). To generate forest plots, the rmeta package for the R-project statistical software (www.R-project.org) was used.

Results

Literature search and included studies

Our search strategy identified 1429 titles/abstracts and 41 articles were selected for full-text review (Fig. 1; [7–11, 20–55]). Fifteen primary studies on the predictability of ANCA on relapse of disease were included, nine with a total of 503 patients for a rise in ANCA [8, 10, 11, 50–55] and nine with a total of 430 patients on the predictability of persistently positive ANCA [30–33, 36, 39, 50, 51, 55] (Table 1). We found no evidence of overlapping information reported between the included studies.

Fig. 1.

Fig. 1

Open in a new tab

Study flow diagram.

Table 1.

Characteristics of included studies

Study Subjects, n Subjects with relapse during follow-up (%) Mean follow-up time, months Type of vasculitis Type of antibody ANCA method Frequency of ANCA measurements Definition of rise in ANCA Definition of persistent positive ANCA Definition of relapse
Cohen Tervaert et al.a [8] 35 12 (34) 16 GPA: 35 cANCA/PR3: 35 IF Every 1 month (per protocol) 4-fold rise in titre NA Clinical
MPA: 0 pANCA/MPO: 0
Gaskin et al.b [36] 70 15 (21) 36.2 GPA: 34 cANCA/PR3: 48 IF Every 3 months (average) NA Positive ANCA at all or all but one visit Clinical
MPA: 26 CSS: 10 pANCA/MPO: 12 unclassified: 10
Pettersson et al.a,b [50] 17 6 (35) 36 GPA: 11 cANCA/PR3: 11 IF Every 2 monthsg (average) Conversion from negative to positive Not reaching negative Clinical
MPA: 6 pANCA/MPO: 6
Kerr et al.a [10] 68 33 (59) 11 GPA: 68 cANCA/PR3: 68 IF Every 2 monthsg (average) 4-fold rise in titre NA Clinical
MPA: 0 pANCA/MPO: 0
Stegeman et al.b [39] 54 21 (39) 12–42d (range) GPA: 54 cANCA/PR3: 54 ELISA Not provided NA Not reaching negative Clinical
MPA: 0 pANCA/MPO: 0
Jayne et al.a,b [51] 60 23 (38) 12 GPA: 45 cANCA/PR3: 39 ELISA Every 1 month (per protocol) Transition from negative to positive or 30% rise in binding Not reaching negative Clinical
MPA: 15 pANCA/MPO: 13 Unclassified: 8
Kyndt et al.b [30] 43 23 (53) 22c GPA: 21 cANCA/PR3: 19 IF Every 2 monthsg (average) NA Not negative, not decreasing Clinical
MPA: 22 pANCA/MPO: 24
Boomsma et al.a [52] 100 37 (37) 36c GPA: 100 cANCA/PR3: 85 ELISA Every 2 months (per protocol) 75% increase in binding NA Clinical
MPA: 0 pANCA/MPO: 15
Girard et al.b [33] 43 19 (44) Not provided GPA: 43 cANCA/PR3: 39-43f IF Every 3.5 monthsh (average) NA Not reaching negative Clinical
MPA: 0 pANCA/MPO: 0-4f
Nowack et al.b [31] 18 6 (33) 21 GPA: 14 cANCA/PR3: 14 IF Not provided NA Not defined Increase in BVAS by 2
MPA: 4 pANCA/MPO: 4
Han et al.a [53] 48 16 (33) 46 Not provided cANCA/PR3: 27 ELISA Every 2.3 months (average) 4-fold rise in titre NA BVAS from 0 to 2
pANCA/MPO: 21
Sanders et al.b [32] 87 53 (61) 24 (minimum) GPA: 68 cANCA/PR3: 87 ELISA Every 6 months (per protocol) NA Not defined Clinical
MPA: 19 pANCA/MPO: 0
Finkielman et al.a [11] 101 46 (46) 34 GPA: 101 cANCA/PR3: 101 ELISA Every 3 months (per protocol) 100% increase NA BVAS/GPA
MPA: 0 pANCA/MPO: 0
Damoiseaux et al.a [54] 46 23 (50)e 12 GPA: 46 cANCA/PR3: 46 ELISA Every 3 months (per protocol) 200% increase in binding NA Clinical
MPA: 0 pANCA/MPO: 0
Terrier et al.a,b [55] 38b28a 11 (29) 54

  • GPA: 15

  • MPA: 18

  • CSS: 5

  • cANCA/PR3: 0

  • pANCA/MPO: 28

 ELISA Not provided Transition from negative to positive Not reaching negative Clinical
Open in a new tab

Overview of the included studies (aincluded for meta-analysis on the effect of a rise in ANCA; bincluded for meta-analysis on the effect of a persistent ANCA; cmedian; dmean value not provided; e23 patients with relapse compared with 23 controls without relapse; fexact number could not be extracted, range of possible values is provided; gmeasurements were obtained every 1–3 months, assumed to be every 2 months on average; hmeasurements were obtained every 3–4 months, assumed to be every 3.5 months on average). NA: not applicable, CSS: Churg–Strauss syndrome.

Predictability of a rise in ANCA titre on future relapse of disease

The summary LR+ and LR− for future relapse were modest, but statistically significant (Fig. 2). A rise in ANCA level was associated with an LR+ of 2.84 (95% CI 1.65, 4.90) of a future relapse and absence of a rise in ANCA was associated with an LR− of 0.49 (95% CI 0.27, 0.87). HSROCs were consistent with individual studies using different diagnostic thresholds for defining a rise in ANCA on the underlying receiver operator characteristic (ROC) curve (Fig. 3). Estimated sensitivity and specificity of a rise in ANCA as a predictor of relapse were 0.56 (95% CI 0.33, 0.79) and 0.82 (95% CI 0.75, 0.90), respectively. Back-calculated LR+ and LR− from the summary estimates for sensitivity and specificity, were similar to those obtained from the main analysis, with values of 3.16 (95% CI 0.99, 5.33) and 0.53 (95% CI 0.24, 0.83), respectively.

Fig. 2.

Fig. 2

Open in a new tab

Forest plots. Forest plots of the LR+ and LR− of the nine included studies for the predictability of a rise (A) and of persistence (B) in ANCA on future relapse of disease along with summary estimate (♦) from a random-effects analysis.

Fig. 3.

Fig. 3

Open in a new tab

Summary receiver operator curves. HSROC for a rise in ANCA (A) and for persistent ANCA (B) as predictor of relapse. Open circles represent individual studies and the size of the circles represents the inverse variance (precision). Closed circles represent the summary estimate for sensitivity and specificity. The curves are a product of statistical modelling across the whole spectrum of sensitivity and specificity (broken line), but should only be interpreted at the range where they are supported by data (solid line).

The predictability of persistently positive ANCA titres for relapse of disease

The summary LR+ and LR− for future relapse were small and of borderline statistical significance (Fig. 2). Persistent ANCA during follow-up was associated with an LR+ of 1.93 (95% CI 1.41, 2.66) of a future relapse and absence of persistently positive ANCA was associated with an LR− of 0.74 (95% CI 0.51, 1.08). Estimated sensitivity and specificity of persistent ANCA as a predictor of relapse were 0.38 (95% CI 0.23, 0.52) and 0.78 (95% CI 0.71, 0.85), respectively (Fig. 3). The back-calculated LRs from the summary estimates for sensitivity and specificity were 1.73 (95% CI 0.81, 2.65) and 0.80 (95% CI 0.59, 1.00) for LR+ and LR−, respectively.

Subgroup analysis, heterogeneity and publication bias

Among the studies on a rise in ANCA there was high level of heterogeneity, which was statistically significant with an I2-value of 0.90 (95% CI 0.83, 0.94). Heterogeneity among the studies on persistence of ANCA was moderate and statistically significant with an I2-value of 0.68 (95% CI 0.35, 0.84). For a rise in ANCA, frequency of serial ANCA measurements during remission explained a moderate proportion of the between-study heterogeneity and is consistent with more frequent measurements during remission, resulting in better predictability for future relapse, with LR+ of 4.34 for measurements every month and 1.44 for measurements every 3 months (P = 0.12) and LR− of 0.19 and 1.11 for measurements every 1 and 3 months, respectively (P = 0.03) (Table 2, supplementary figure 1, available as supplementary data at Rheumatology Online). With respect to the type of ANCA, a rise in pANCA/MPO was estimated to predict the future relapse of disease better with LR+ of 10.0 vs 1.4 for cANCA/PR3 (P = 0.01) and LR− of 0.42 vs 0.52 (P = 0.81) for pANCA/MPO and cANCA/PR3, respectively. This finding was driven mainly by one study and did not explain between-study heterogeneity. Similar findings were not observed for studies on persistent ANCA and other study-level variables neither identified other potential subgroups where ANCA measurements might be of more value, nor explained the between-study heterogeneity.

Table 2.

Subgroups and heterogeneity

LR+
 LR−
 Heterogeneity (τ2)
Meta-regression models Subgroups P-value Subgroups P-value
A. Rise in ANCA
Main model 2.84 (95% CI 1.65, 4.90) 0.48 (95% CI 0.27, 0.86) 0.1315
    Laboratory method ELISA: 2.97 IF: 2.63 0.83 ELISA: 0.63 IF: 0.22 0.10 0.1027
    Type of antibody cANCA/PR3: 1.35 pANCA/MPO: 10.03 0.01 cANCA/PR3: 0.52 pANCA/MPO: 0.42 0.81 0.1234
    Predefined interval between measurements YES: 2.84 NO: 2.84 0.99 YES: 0.39 NO: 0.60 0.45 0.1251
    Frequency of measurementsa Every 1 month: 4.43b Every 3 months: 1.44b 0.12 Every 1 month: 0.19b Every 3 months: 1.11b 0.03 0.09109 (main model τ2 was 0.1558 for the eight studies excluding Terrier et al. [55])
    Duration of follow-up – – 0.1301
    Year of publication – – 0.1247
B. Persistent ANCA
Main model 1.97 (95% CI 1.43, 2.70) 0.73 (95% CI 0.50, 1.06) 0.04685
    Laboratory method ELISA: 1.84 IF: 2.13 0.66 ELISA: 0.97 IF: 0.57 0.11 0.03313
    Type of antibody cANCA/PR3: 2.12 pANCA/MPO: 1.45 0.60 cANCA/PR3: 0.75 pANCA/MPO: 0.72 0.96 0.04469
    Type of antibody cANCA/PR3: 2.12 pANCA/MPO: 1.45 0.60 cANCA/PR3: 0.75 pANCA/MPO: 0.72 0.96 0.04469
    Frequency of measurements Every 1 month: Every 3 months: Every 1 month Every 3 months Data are missing
    Duration of follow-up – – Data are missing
    Year of publication – – 0.04205
Open in a new tab

aAnalysis based on eight out of nine studies, as Terrier et al. [55] did not provide data on frequency of ANCA measurements during remission. bEffects of subgroups rely on the assumption that there is linear relationship between time between measurements and log of the LR+ and LR−. The table shows calculated positive and negative likelihood ratios (LR+ and LR−) from the main meta-analysis, subgroup effect and heterogeneity for (A) a rise in ANCA and (B) persistent ANCA. The results from the main meta-analysis (intercept-only model) are expressed, as well as results from five regression models where independent variables representing subgroups are tested. Statistical significance of the difference between subgroups is expressed by P-values. Heterogeneity is based on the difference in arcsine transformed outcome and expressed as τ2. It is evident that heterogeneity is not reduced in the meta-regression models compared with the intercept-only models.

For studies on a rise in ANCA, asymmetry in the funnel plot was present by visual inspection (supplementary figure 2, available as supplementary data at Rheumatology Online), consistent with the presence of publication bias; however, the Egger test did not reject the null hypothesis of no publication bias (P = 0.36). For studies on persistently positive ANCA, the funnel plot was symmetric by visual inspection, with no suggestion of publication bias, and the Egger test results were null (P = 0.96).

Discussion

The utility of serial ANCA measurements among patients with an established diagnosis of AAV to assess disease activity or predict disease relapse has received considerable attention, but remains highly controversial. The published data on serial ANCA testing is heterogeneous with multiple study-level variables possibly affecting the interpretation of results, including types of ANCA testing assays used, time intervals between measurements, definitions of rises in ANCA titres and inclusion of subgroups of patients with different disease manifestations.

In this analysis, we chose to address two specific aspects of serial ANCA testing on future disease relapse: (i) the predictability of a rise in ANCA titres, and (ii) the persistence of ANCA. We have carefully accounted for as many confounding variables as possible given the available data. We found that both a rise in and persistence of ANCA during remission of vasculitis only modestly predict subsequent relapse of disease and that routine ANCA testing does not alter clinical risk estimates enough to substantially affect clinical practice. For example, given that a rise in ANCA has a summary LR+ of 2.84 and a summary LR− of 0.46, if the pre-test risk of relapse is estimated at 40%, then the post-test risk for subsequent relapse is 69% if there is a rise in ANCA, and 23% if there is no rise in ANCA (additional examples shown in supplementary Table 1, available as supplementary data at Rheumatology Online). Furthermore, the summary risk estimates for a rise in ANCA should not be interpreted as being associated with immediate relapse since studies were included where a rise in ANCA sometimes preceded relapse by more than 1 year.

There are several strengths to our approach to this meta-analysis. Multiple studies have reported on the association between ANCA values and disease activity in AAV, including elevation in ANCA at the time of relapse [7, 20, 24, 26, 54]. The present analysis focused on serial ANCA measurements to predict future relapse of disease and thus has direct implications to current clinical practice. We also provide data on a pattern in serial ANCA testing that has not been a primary focus of many previous publications, i.e. persistently positive ANCA. Most original studies included for this latter analysis had another primary aim. In this situation, when the meta-analysis outcome is tangential to the main finding of the included studies, it is less likely that summary estimates are affected by biases in primary studies rising from low methodological quality. Our approach thus provides insight into the controversial topic of the value of ANCA to guide therapy in AAV, with results from meta-regression offering some explanation of why individual studies might differ. The identification of subtypes and frequency of ANCA measurements where testing might be of more value provides guidance for future research on the value of serial ANCA testing.

Our approach has some limitations to consider. With only a few studies in each meta-analysis, we had limited opportunity to explore heterogeneity. Additionally, it is possible that in the original studies the ANCA values were spuriously associated with future relapse due to closer surveillance of patients with ANCA rises or persistently positive ANCA, or led to decreased risk of relapse if any treatment was given based on the ANCA results. Approximately half of the included studies measured ANCA at fixed intervals during follow-up in a protocolized fashion; those studies are likely less susceptible to bias, but we did not find that their results differed compared with the studies in which measurement of ANCA during follow-up was not protocolized. Furthermore, we did not formally address the quality of individual studies. Although methods for quality assessment of studies on diagnostic tests have been proposed [56], we felt these criteria would not have allowed us to rank the included studies. We pooled LR+ and LR− in our main analyses, a common practice in diagnostic test meta-analysis [57], which has been recently warned against, as values within the CIs for LRs can correspond to impossible values of sensitivity and specificity (e.g. below 0 or above 1) [58]. This is why we also back-calculated LRs from the pooled sensitivity and specificity and found that the two methods gave consistent results.

The between-study heterogeneity merits further consideration. For both study questions, most of the overall variability rose from between-study heterogeneity with I2-values close to 1. However, the random-effects analysis incorporated the heterogeneity by making wider CIs; therefore, pooling of results to arrive at summary estimates is justifiable. We had low power to detect the effects of subgroups of patients and assays or to explore this heterogeneity; nonetheless, we did find that frequent ANCA measurements better predicted disease relapse than less frequent ANCA measurements. However, even the LRs estimated for monthly measurements of ANCA were not so different from 1.0 to provide strong evidence for or against future disease relapse. The frequency of ANCA measurements during remission was the only study-level factor that explained a moderately large proportion of the between-study heterogeneity. Our findings that pANCA/MPO might predict the future relapse of disease better than a rise in cANCA/PR3 should be interpreted with caution. Although this difference was statistically significant and of potential clinical importance for the LR+, this finding was largely determined by one study on MPO antibody levels as a predictor of future flares of vasculitis [55] and did not explain between-study heterogeneity. Current ELISA tests rely on distinct sources of antigens, coating materials and conjugates, and substantial variability exists among different ELISA assays [59, 60]. The between-study heterogeneity limits the generalizability of the studies and underscores the need for adopting international standards for ANCA testing. Other study-level factors tested (assay type, duration of follow-up, protocolized vs non-protocolized ANCA measurements and publication year) did not explain the observed heterogeneity or suggest that predictability of ANCA varied across subgroups. In addition, the effect of several other study-level factors, such as severity of flare and organs involved (renal vs non-renal), could not be tested, as this information was lacking from the primary reports. Different definitions for remission and relapse by the individual studies are likely to contribute to the between-study heterogeneity. We could not identify subgroups among the included studies with similar definitions of these disease states or account for this heterogeneity in other ways. The contribution to heterogeneity of the use of different definitions of a rise in ANCA by the individual studies was not assessed since different thresholds for ANCA rise were accounted for by the methods used to calculate summary estimates.

In conclusion, although drawn from a heterogeneous literature, we believe the summary estimates presented here provide the best estimate for the true value of serial ANCA testing for prediction of relapses in AAV. We found that both a rise in ANCA and persistently positive ANCA are significantly associated with disease relapse—although the performance of serial ANCA testing does not appear to be sufficient to rely on isolation for management decisions. It is possible that serial ANCA testing could contribute to a prediction rule that takes into account clinical features and future biomarkers. Our analyses are informative in several additional areas, and the better performances for predicting relapses of ANCA testing among pANCA/MPO-positive patients or more frequent (monthly) ANCA titre measurements are promising and should be the focus of future studies. Conversely, newer methods of ANCA measurements do not appear to perform better with respect to prediction of future relapse compared with older methods relying on IF.

graphic file with name ker280b1.jpg

Supplementary data

Supplementary data are available at Rheumatology Online.

Supplementary Data
supp_51_1_100__index.html (1.1KB, html)

Acknowledgements

We are grateful to Jan Damoiseaux, Laboratory of Clinical Immunology, Maastricht University Medical Center, for providing explanations and additional data elements regarding one of the included primary studies in this meta-analysis.

Funding: Supported in part by The National Institute of Arthritis and Musculoskeletal and Skin Diseases Multidisciplinary Clinical Research Center grant P60 AR047785.

Disclosure statement: The authors have declared no conflicts of interest.

References

  • 1.Choi HK, Liu S, Merkel PA, Colditz GA, Niles JL. Diagnostic performance of antineutrophil cytoplasmic antibody tests for idiopathic vasculitides: metaanalysis with a focus on antimyeloperoxidase antibodies. J Rheumatol. 2001;28:1584–90. [PubMed] [Google Scholar]
  • 2.Finkielman JD, Lee AS, Hummel AM, et al. ANCA are detectable in nearly all patients with active severe Wegener's granulomatosis. Am J Med. 2007;120:643e9–14. doi: 10.1016/j.amjmed.2006.08.016. [DOI] [PubMed] [Google Scholar]
  • 3.Hagen EC, Daha MR, Hermans J, et al. Diagnostic value of standardized assays for anti-neutrophil cytoplasmic antibodies in idiopathic systemic vasculitis. EC/BCR project for ANCA assay standardization. Kidney Int. 1998;53:743–53. doi: 10.1046/j.1523-1755.1998.00807.x. [DOI] [PubMed] [Google Scholar]
  • 4.Hoffman GS, Specks U. Antineutrophil cytoplasmic antibodies. Arthritis Rheum. 1998;41:1521–37. doi: 10.1002/1529-0131(199809)41:9<1521::AID-ART2>3.0.CO;2-A. [DOI] [PubMed] [Google Scholar]
  • 5.Merkel PA, Polisson RP, Chang Y, Skates SJ, Niles JL. Prevalence of antineutrophil cytoplasmic antibodies in a large inception cohort of patients with connective tissue disease. Ann Intern Med. 1997;126:866–73. doi: 10.7326/0003-4819-126-11-199706010-00003. [DOI] [PubMed] [Google Scholar]
  • 6.Xiao H, Heeringa P, Hu P, et al. Antineutrophil cytoplasmic autoantibodies specific for myeloperoxidase cause glomerulonephritis and vasculitis in mice. J Clin Invest. 2002;110:955–63. doi: 10.1172/JCI15918. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Specks U, Wheatley CL, McDonald TJ, Rohrbach MS, DeRemee RA. Anticytoplasmic autoantibodies in the diagnosis and follow-up of Wegener's granulomatosis. Mayo Clin Proc. 1989;64:28–36. doi: 10.1016/s0025-6196(12)65300-2. [DOI] [PubMed] [Google Scholar]
  • 8.Cohen-Tervaert JW, van der Woude FJ, Fauci AS, et al. Association between active Wegener's granulomatosis and anticytoplasmic antibodies. Arch Intern Med. 1989;149:2461–5. doi: 10.1001/archinte.149.11.2461. [DOI] [PubMed] [Google Scholar]
  • 9.Cohen-Tervaert JW, Huitema MG, Hene RJ, et al. Prevention of relapses in Wegener's granulomatosis by treatment based on antineutrophil cytoplasmic antibody titre. Lancet. 1990;336:709–11. doi: 10.1016/0140-6736(90)92205-v. [DOI] [PubMed] [Google Scholar]
  • 10.Kerr GS, Fleisher TA, Hallahan CW, Leavitt RY, Fauci AS, Hoffman GS. Limited prognostic value of changes in antineutrophil cytoplasmic antibody titer in patients with Wegener's granulomatosis. Arthritis Rheum. 1993;36:365–71. doi: 10.1002/art.1780360312. [DOI] [PubMed] [Google Scholar]
  • 11.Finkielman JD, Merkel PA, Schroeder D, et al. Antiproteinase 3 antineutrophil cytoplasmic antibodies and disease activity in Wegener granulomatosis. Ann Intern Med. 2007;147:611–9. doi: 10.7326/0003-4819-147-9-200711060-00005. [DOI] [PubMed] [Google Scholar]
  • 12.Simel DL, Samsa GP, Matchar DB. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. J Clin Epidemiol. 1991;44:763–70. doi: 10.1016/0895-4356(91)90128-v. [DOI] [PubMed] [Google Scholar]
  • 13.Reitsma JB, Glas AS, Rutjes AW, Scholten RJ, Bossuyt PM, Zwinderman AH. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol. 2005;58:982–90. doi: 10.1016/j.jclinepi.2005.02.022. [DOI] [PubMed] [Google Scholar]
  • 14.Simel DL, Bossuyt PM. Differences between univariate and bivariate models for summarizing diagnostic accuracy may not be large. J Clin Epidemiol. 2009;62:1292–300. doi: 10.1016/j.jclinepi.2009.02.007. [DOI] [PubMed] [Google Scholar]
  • 15.Macaskill P. Empirical Bayes estimates generated in a hierarchical summary ROC analysis agreed closely with those of a full Bayesian analysis. J Clin Epidemiol. 2004;57:925–32. doi: 10.1016/j.jclinepi.2003.12.019. [DOI] [PubMed] [Google Scholar]
  • 16.Rucker G, Schwarzer G, Carpenter J. Arcsine test for publication bias in meta-analyses with binary outcomes. Stat Med. 2008;27:746–63. doi: 10.1002/sim.2971. [DOI] [PubMed] [Google Scholar]
  • 17.Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. Br Med J. 2003;327:557–60. doi: 10.1136/bmj.327.7414.557. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Thompson SG, Sharp SJ. Explaining heterogeneity in meta-analysis: a comparison of methods. Stat Med. 1999;18:2693–708. doi: 10.1002/(sici)1097-0258(19991030)18:20<2693::aid-sim235>3.0.co;2-v. [DOI] [PubMed] [Google Scholar]
  • 19.Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. Br Med J. 1997;315:629–34. doi: 10.1136/bmj.315.7109.629. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Nolle B, Specks U, Ludemann J, Rohrbach MS, DeRemee RA, Gross WL. Anticytoplasmic autoantibodies: their immunodiagnostic value in Wegener granulomatosis. Ann Intern Med. 1989;111:28–40. doi: 10.7326/0003-4819-111-1-28. [DOI] [PubMed] [Google Scholar]
  • 21.Hachulla E, Hatron PY, Brouillard M, Cesbron JY, Reumaux D, Devulder B. Sensitivity and specificity of the antineutrophil cytoplasmic antibody (c-ANCA) in systemic vasculitis. Rev Med Interne. 1994;15:381–6. doi: 10.1016/s0248-8663(05)81452-3. [DOI] [PubMed] [Google Scholar]
  • 22.Ara J, Mirapeix E, Rodriguez R, Saurina A, Darnell A. Relationship between ANCA and disease activity in small vessel vasculitis patients with anti-MPO ANCA. Nephrol Dial Transplant. 1999;14:1667–72. doi: 10.1093/ndt/14.7.1667. [DOI] [PubMed] [Google Scholar]
  • 23.Boomsma MM, Damoiseaux JG, Stegeman CA, et al. Image analysis: a novel approach for the quantification of antineutrophil cytoplasmic antibody levels in patients with Wegener's granulomatosis. J Immunol Methods. 2003;274:27–35. doi: 10.1016/s0022-1759(02)00273-9. [DOI] [PubMed] [Google Scholar]
  • 24.Lurati-Ruiz F, Spertini F. Predictive value of antineutrophil cytoplasmic antibodies in small-vessel vasculitis. J Rheumatol. 2005;32:2167–72. [PubMed] [Google Scholar]
  • 25.Egner W, Chapel HM. Titration of antibodies against neutrophil cytoplasmic antigens is useful in monitoring disease activity in systemic vasculitides. Clin Exp Immunol. 1990;82:244–9. doi: 10.1111/j.1365-2249.1990.tb05434.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Halma C, Daha MR, Schrama E, Hermans J, van Es LA, van der Woude FJ. Value of anti-neutrophil cytoplasmic autoantibodies and other laboratory parameters in follow-up of vasculitis. Scand J Rheumatol. 1990;19:392–7. doi: 10.3109/03009749009097627. [DOI] [PubMed] [Google Scholar]
  • 27.Chan TM, Frampton G, Jayne DR, Perry GJ, Lockwood CM, Cameron JS. Clinical significance of anti-endothelial cell antibodies in systemic vasculitis: a longitudinal study comparing anti-endothelial cell antibodies and anti-neutrophil cytoplasm antibodies. Am J Kidney Dis. 1993;22:387–92. doi: 10.1016/s0272-6386(12)70140-3. [DOI] [PubMed] [Google Scholar]
  • 28.De'Oliviera J, Gaskin G, Dash A, Rees AJ, Pusey CD. Relationship between disease activity and ANCA level by ELISA in the long-term management of vasculitis. Adv Exp Med Biol. 1993;336:303–4. doi: 10.1007/978-1-4757-9182-2_47. [DOI] [PubMed] [Google Scholar]
  • 29.De'Oliviera J, Gaskin G, Dash A, Rees AJ, Pusey CD. Relationship between disease activity and anti-neutrophil cytoplasmic antibody concentration in long-term management of systemic vasculitis. Am J Kidney Dis. 1995;25:380–9. doi: 10.1016/0272-6386(95)90098-5. [DOI] [PubMed] [Google Scholar]
  • 30.Kyndt X, Reumaux D, Bridoux F, et al. Serial measurements of antineutrophil cytoplasmic autoantibodies in patients with systemic vasculitis. Am J Med. 1999;106:527–33. doi: 10.1016/s0002-9343(99)00064-9. [DOI] [PubMed] [Google Scholar]
  • 31.Nowack R, Grab I, Flores-Suarez LF, Schnulle P, Yard B, van der Woude FJ. ANCA titres, even of IgG subclasses, and soluble CD14 fail to predict relapses in patients with ANCA-associated vasculitis. Nephrol Dial Transplant. 2001;16:1631–7. doi: 10.1093/ndt/16.8.1631. [DOI] [PubMed] [Google Scholar]
  • 32.Sanders JS, Huitma MG, Kallenberg CG, Stegeman CA. Prediction of relapses in PR3-ANCA-associated vasculitis by assessing responses of ANCA titres to treatment. Rheumatology. 2006;45:724–9. doi: 10.1093/rheumatology/kei272. [DOI] [PubMed] [Google Scholar]
  • 33.Girard T, Mahr A, Noel LH, et al. Are antineutrophil cytoplasmic antibodies a marker predictive of relapse in Wegener's granulomatosis? A prospective study. Rheumatology. 2001;40:147–51. doi: 10.1093/rheumatology/40.2.147. [DOI] [PubMed] [Google Scholar]
  • 34.van der Woude FJ, Rasmussen N, Lobatto S, et al. Autoantibodies against neutrophils and monocytes: tool for diagnosis and marker of disease activity in Wegener's granulomatosis. Lancet. 1985;1:425–9. doi: 10.1016/s0140-6736(85)91147-x. [DOI] [PubMed] [Google Scholar]
  • 35.Esnault VL, Jayne DR, Weetman AP, Lockwood CM. IgG subclass distribution and relative functional affinity of anti-myeloperoxidase antibodies in systemic vasculitis at presentation and during follow-up. Immunology. 1991;74:714–8. [PMC free article] [PubMed] [Google Scholar]
  • 36.Gaskin G, Savage CO, Ryan JJ, et al. Anti-neutrophil cytoplasmic antibodies and disease activity during long-term follow-up of 70 patients with systemic vasculitis. Nephrol Dial Transplant. 1991;6:689–94. doi: 10.1093/ndt/6.10.689. [DOI] [PubMed] [Google Scholar]
  • 37.Bergelin A, Subra JF, Carrere F, et al. [Role of the titration of anti-neutrophil cytoplasmic antibodies in therapeutic follow-up of patients with systemic vasculitis] Rev Med Interne. 1993;14:1024. doi: 10.1016/s0248-8663(05)80141-9. [DOI] [PubMed] [Google Scholar]
  • 38.Gordon M, Luqmani RA, Adu D, et al. Relapses in patients with a systemic vasculitis. Q J Med. 1993;86:779–89. [PubMed] [Google Scholar]
  • 39.Stegeman CA, Tervaert JW, Sluiter WJ, Manson WL, de Jong PE, Kallenberg CG. Association of chronic nasal carriage of Staphylococcus aureus and higher relapse rates in Wegener granulomatosis. Ann Intern Med. 1994;120:12–7. doi: 10.7326/0003-4819-120-1-199401010-00003. [DOI] [PubMed] [Google Scholar]
  • 40.Cohen P, Guillevin L, Baril L, Lhote F, Noel LH, Lesavre P. Persistence of antineutrophil cytoplasmic antibodies (ANCA) in asymptomatic patients with systemic polyarteritis nodosa or Churg-Strauss syndrome: follow-up of 53 patients. Clin Exp Rheumatol. 1995;13:193–8. [PubMed] [Google Scholar]
  • 41.Davenport A, Lock RJ, Wallington T. Clinical significance of the serial measurement of autoantibodies to neutrophil cytoplasm using a standard indirect immunofluorescence test. Am J Nephrol. 1995;15:201–7. doi: 10.1159/000168833. [DOI] [PubMed] [Google Scholar]
  • 42.Ara J, Mirapeix E, Rodriguez R, Pascual J, Alvarez L, Darnell A. [Usefulness of anti-neutrophil cytoplasmic antibodies, anti-proteinase 3 and anti-myeloperoxidase in management of small vessel vasculitis] Med Clin. 1998;111:536–8. [PubMed] [Google Scholar]
  • 43.Arranz O, Ara J, Rodriguez R, et al. Comparison of anti-PR3 capture and anti-PR3 direct ELISA for detection of antineutrophil cytoplasmic antibodies (ANCA) in long-term clinical follow-up of PR3-ANCA-associated vasculitis patients. Clin Nephrol. 2001;56:295–301. [PubMed] [Google Scholar]
  • 44.Solans R, Bosch JA, Perez-Bocanegra C, et al. Churg-Strauss syndrome: outcome and long-term follow-up of 32 patients. Rheumatology. 2001;40:763–71. doi: 10.1093/rheumatology/40.7.763. [DOI] [PubMed] [Google Scholar]
  • 45.Gisslen K, Wieslander J, Westberg G, Herlitz H. Relationship between anti-neutrophil cytoplasmic antibody determined with conventional binding and the capture assay, and long-term clinical course in vasculitis. J Intern Med. 2002;251:129–35. doi: 10.1046/j.1365-2796.2002.00939.x. [DOI] [PubMed] [Google Scholar]
  • 46.Keogh KA, Specks U. Churg-Strauss syndrome: clinical presentation, antineutrophil cytoplasmic antibodies, and leukotriene receptor antagonists. Am J Med. 2003;115:284–90. doi: 10.1016/s0002-9343(03)00359-0. [DOI] [PubMed] [Google Scholar]
  • 47.Segelmark M, Phillips BD, Hogan SL, Falk RJ, Jennette JC. Monitoring proteinase 3 antineutrophil cytoplasmic antibodies for detection of relapses in small vessel vasculitis. Clin Diagn Lab Immunol. 2003;10:769–74. doi: 10.1128/CDLI.10.5.769-774.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Takwoingi YM, Dempster JH. Wegener's granulomatosis: an analysis of 33 patients seen over a 10-year period. Clin Otolaryngol Allied Sci. 2003;28:187–94. doi: 10.1046/j.1365-2273.2003.00683.x. [DOI] [PubMed] [Google Scholar]
  • 49.Slot MC, Tervaert JW, Boomsma MM, Stegeman CA. Positive classic antineutrophil cytoplasmic antibody (C-ANCA) titer at switch to azathioprine therapy associated with relapse in proteinase 3-related vasculitis. Arthritis Rheum. 2004;51:269–73. doi: 10.1002/art.20234. [DOI] [PubMed] [Google Scholar]
  • 50.Pettersson E, Heigl Z. Antineutrophil cytoplasmic antibody (cANCA and pANCA) titers in relation to disease activity in patients with necrotizing vasculitis: a longitudinal study. Clin Nephrol. 1992;37:219–28. [PubMed] [Google Scholar]
  • 51.Jayne DR, Gaskin G, Pusey CD, Lockwood CM. ANCA and predicting relapse in systemic vasculitis. Q J Med. 1995;88:127–33. [PubMed] [Google Scholar]
  • 52.Boomsma MM, Stegeman CA, van der Leij MJ, et al. Prediction of relapses in Wegener's granulomatosis by measurement of antineutrophil cytoplasmic antibody levels: a prospective study. Arthritis Rheum. 2000;43:2025–33. doi: 10.1002/1529-0131(200009)43:9<2025::AID-ANR13>3.0.CO;2-O. [DOI] [PubMed] [Google Scholar]
  • 53.Han WK, Choi HK, Roth RM, McCluskey RT, Niles JL. Serial ANCA titers: useful tool for prevention of relapses in ANCA-associated vasculitis. Kidney Int. 2003;63:1079–85. doi: 10.1046/j.1523-1755.2003.00821.x. [DOI] [PubMed] [Google Scholar]
  • 54.Damoiseaux J, Dahnrich C, Rosemann A, et al. A novel enzyme-linked immunosorbent assay using a mixture of human native and recombinant proteinase-3 significantly improves the diagnostic potential for antineutrophil cytoplasmic antibody-associated vasculitis. Ann Rheum Dis. 2009;68:228–33. doi: 10.1136/ard.2007.086579. [DOI] [PubMed] [Google Scholar]
  • 55.Terrier B, Saadoun D, Sene D, et al. Antimyeloperoxidase antibodies are a useful marker of disease activity in antineutrophil cytoplasmic antibody-associated vasculitides. Ann Rheum Dis. 2009;68:1564–71. doi: 10.1136/ard.2008.094714. [DOI] [PubMed] [Google Scholar]
  • 56.Whiting P, Rutjes AW, Reitsma JB, Bossuyt PM, Kleijnen J. The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol. 2003;3:25. doi: 10.1186/1471-2288-3-25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 57.Deeks JJ. Systematic reviews in health care: systematic reviews of evaluations of diagnostic and screening tests. Br Med J. 2001;323:157–62. doi: 10.1136/bmj.323.7305.157. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Zwinderman AH, Bossuyt PM. We should not pool diagnostic likelihood ratios in systematic reviews. Stat Med. 2008;27:687–97. doi: 10.1002/sim.2992. [DOI] [PubMed] [Google Scholar]
  • 59.Pollock W, Jovanovich S, Savige J. Antineutrophil cytoplasmic antibody (ANCA) testing of routine sera varies in different laboratories but concordance is greater for cytoplasmic fluorescence (C-ANCA) and myeloperoxidase specificity (MPO-ANCA) J Immunol Methods. 2009;347:19–23. doi: 10.1016/j.jim.2009.05.008. [DOI] [PubMed] [Google Scholar]
  • 60.Wang G, Csernok E, de Groot K, Gross WL. Comparison of eight commercial kits for quantitation of antineutrophil cytoplasmic antibodies (ANCA) J Immunol Methods. 1997;208:203–11. doi: 10.1016/s0022-1759(97)00154-3. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data
supp_51_1_100__index.html (1.1KB, html)
supp_ker280_Supplementary_figures.docx (1.6MB, docx)
supp_ker280_Supplementary_table.docx (27.5KB, docx)

Articles from Rheumatology (Oxford, England) are provided here courtesy of Oxford University Press

RESOURCES