American College of Rheumatology Criteria 1990 (ACR 1990)

The ACR criteria from 1990 for SV were the first to combine different parameters into one set of classification criteria. The criteria were developed from a dataset with 1020 cases of vasculitis from North America (USA, Canada, and Mexico). Through identification of features typical for different types of vasculitis, criteria for seven different forms of systemic vasculitis (GPA, EGPA, GCA, TAK, IgA vasculitis, PAN and hypersensitivity vasculitis) [13•] were proposed. The criteria were widely used and facilitated clinical and epidemiological research in the field of vasculitis, however with time several shortcomings became obvious: i) the criteria lacked ANCA, though assays became widely available in the following years, ii) MPA was not included as a separate disease entity, iii) the underlying dataset only included cases from North America not accounting for geographical differences in phenotype distribution, in addition certain subtypes were more represented than others in the dataset, and iv) cases could be classified to different subtypes at the same time. Though not intended as such, the criteria were used for diagnosis in individual cases in routine clinical practice. The criteria performed poorly in an evaluation as diagnostic criteria [31]. The criteria had considerable impact on epidemiologic research in vasculitis.

Chapel Hill Consensus Conference (CHCC)

In 1992 a meeting of a group of experts in Chapel Hill yielded a consensus document [14•] defining different types of vasculitis and establishing a nomenclature. The experts emphasized that the goal of the CHCC was not to develop classification or diagnostic criteria rather give definitions and define a common nomenclature. Ten different types of vasculitides were defined according to histological and clinical criteria and grouped on predominantly affected vessel size. The system accounted for the presence of ANCAs. The meeting further established MPA as a disease entity separate from GPA and EGPA as well as the association of ANCA with these small vessel vasculitides. As cases earlier attributed to PAN now became MPA, the former became a very rare disease. In an update 2012 [1], the group of small vessel vasculitides was further divided into the immune-complex related- or pauci-immune ANCA-associated vasculitides. Regarding the AAV, the consensus suggested using a prefix indicating ANCA reactivity, the concept of surrogate criterion (cavitary lung lesion on imaging can be a surrogate criterion for granulomatous disease without available histologic examination) is discussed (as it was in 1994) in the publication but without detail or specific guidance. In 2000 a study by Sorensen et al. tried using the CHCC definitions with surrogate markers as diagnostic classification criteria but it was not successful [32]

European Medicines Agency (EMA) Algorithm 2007

The two available widely used systems, the ACR 1990 and the CHCC 1994 definitions performed differently when applied in parallel in same cohort [33], leading to overlapping as patients may be classified into different disease phenotypes or remain unclassified by one of these systems. To avoid this overlapping and to gain a consensus on how the ACR 1990 classification and CHCC 1994 definitions should be used in epidemiologic studies, a group of experts met in London in 2004 and agreed on a stepwise algorithm [34•]. The algorithm, which is widely known as Watts algorithm or EMA algorithm, integrated ACR 1990 and CHCC in a hierarchic way. The algorithm starts with ACR EGPA [35] as it had the highest sensitivity and specificity of the ACR 1990 criteria, then the following steps ACR 1990 GPA [36] and CHCC 1994 MPA are applied. The algorithm also introduced the combined use of ANCA and surrogate markers of vasculitis and/or granuloma. The EMA algorithm has been used in several epidemiological studies in the last 15 years [8, 10, 11] and showed minimum of overlapping or unclassified cases.

The Diagnostic and Classification Criteria in Vasculitis (DCVAS)

The evolution of imaging, widespread use of ANCA-testing and the improved knowledge and understanding of pathophysiology in vasculitis initiated a discussion on revision of classification of the primary vasculitides. The multinational observational DCVAS study was announced in 2010 [6] with the goal to develop and validate diagnostic criteria and improve and validate classification criteria. The project accumulated an impressive cohort and invested great efforts and applied advanced statistics to develop totally new criteria that are intended to be used independent of the prior ones. However, succeeding with the endeavour to even develop diagnostic criteria seems highly unlikely at this point [37]. The DCVAS final cohort included 6991 patients from 136 sites in 32 countries (Europe 59%, North America 21%, other regions 20%). Patients ≥ 18 years with a diagnosis within 2 years of GPA, EGPA, MPA, GCA, anti glomerular basement membrane disease (anti GBM), PACNS, IgA vasculitis, aortitis, other large vessel vasculitis, PAN (within 5 years) or TAK (within 5 years) or a condition mimicking vasculitis secondary to tumour, infection, or other inflammatory condition as diagnosed by the submitting physician, were included. First a final set of high standard cases for each subtype was identified and included into the study, in the next step a set of candidate items were identified and further refined by expert opinion and a data driven approach (exclusion of items with low prevalence and/or not clinically relevant). In the next step a regression model was used which yielded the final criteria that were in the next step validated in a validation data set. Comparators used in the derivation and validation of the criteria for each subtype were the other AAV or to a minor extent other small or medium vessel and not large vessel vasculitides with generally very different phenotype. All these efforts have culminated in the introduction of new classification criteria for the AAV [38••, 39••, 40••] in 2022 as well as for GCA [41] and TAK [42].

The 2022 ACR/EULAR 2022 Criteria for AAV

The new 2022 ACR/EULAR (in the following text designated as ACR/EULAR) criteria derived from the largest cohort ever established in systemic vasculitis. The cohort included mainly patients from western countries, however the recruited cases were much more international than those used for earlier criteria as the ACR1990. Unlike the ACR1990 criteria diagnosis of the submitting physician was reviewed by an expert panel, thereby minimizing investigator bias. The criteria account for the evolution of diagnostics and clinical practice by including ANCA and modern radiology. Another innovation is the weighting of items to reflect clinical importance of features.

The final criteria include a different number of weighted items for the subtypes with threshold scores to be classified. Some items can have negative impact on the score, as for example nasal symptoms with MPA, or the presence of eosinophilia with GPA and MPA.

The following is a short summary of main items in respective criteria, described in detail in Table 1:

Table 1 The 2022 ACR/EULAR classification for AAV criteria, modified from original publications [38••, 39••, 40••], prerequisite for classification is the diagnosis of small-medium vessel vasculitis established and vasculitis mimics excluded GPA

The GPA dataset consisted of 1537 cases (724 GPA and 813 comparators), 20% were used in the validation set. The final criteria consist out of 10 weighted items with a threshold score of 5 needed for classification. The authors report a sensitivity of 92.5% and specificity of 93.8% in the validation data set [39••].

MPA

The MPA dataset consisted of 1113 cases (291 MPA and 822 comparators), 50% were used in the validation set. The final criteria consist of 6 weighted items with a threshold score of 5 needed for classification. A sensitivity of 90.8% and specificity of 94.2% in the validation set is reported [40••].

EGPA

The EGPA dataset consisted of 1113 cases (226 EGPA and 887 comparators), 50% were used in the validation set. The final criteria consist of 7 weighted items with a threshold score of 6 needed for classification. The authors report 84.9% sensitivity and 99.1% specificity in the validation set [38••].

Evaluation of the ACR/EULAR Criteria

An important question with new classification criteria or disease definition is if they improve shortcomings of earlier criteria and if they exhibit better sensitivity and specificity. The 1990 ACR-criteria did not include MPA, but this was added with the CHCC as well as ANCA. The lack of guidance concerning ANCA as a surrogate marker could be improved by the EMA algorithm. ACR/EULAR introduces weighted items with positive and negative scores. In addition, ANCAs are introduced, with considerable weight. Further, new items such as interstitial lung disease (ILD) are introduced. ILD has been observed especially in patients with MPO-associated disease [43], often before the onset of vasculitis [44] and has therefore become more recognized in AAV.

As of January 2024, the new criteria have been evaluated in studies from South Korea, Turkey, Netherlands, Japan, Portugal, India, Ukraine, and Sweden (Table 2). Some reports are only available as conference abstracts. Below is a summary of main findings of some of these evaluations:

Table 2 Characteristics of selected cohorts classified by EMA and 2022 ACR/EULAR and agreement between the systems (if provided)South Korea

Three separate studies for the AAV-subtypes from Korea have investigated reclassification of AAV-cohorts into disease phenotypes using the ACR/EULAR. For EGPA [45] the concordance rate with EMA was 86.3%, 7 former EGPA cases met criteria for two subtypes with the new criteria and 3 cases could not be classified. The authors consider the exclusion of the fixed pulmonary infiltrate item from ACR1990 as reason why these 7 EGPA cases could not be classified the same way with ACR/EULAR. Concerning GPA only 74% of cases remained GPA with ACR/EULAR [46], most were reclassified as MPA, predominantly due to MPO-ANCA positivity. The authors also encountered cases with granuloma on biopsy, now being classified as MPA and suggest score modification of the granuloma item and other clear GPA surrogates in the new criteria. In a further study on MPA cases [47] classified using the EMA algorithm, a high rate of agreement was observed with the ACR/EULAR (96.6%). A small number of patients were classified to GPA and MPA due to positivity to both antibodies. Almost 50% of the MPA cohort exhibited ILD, a feature now having considerable weight towards MPA classification. However, the authors commented that the definition of ILD is broad and a consensus on when this item should be scored is needed. In summary the authors criticize the high score assigned to MPO-ANCA and the lack of consideration of MPA-specific histopathological findings in the new criteria.

Netherlands

Hospital records of 337 patients with AAV from a tertiary centre with a confirmed clinical diagnosis were used in this evaluation [48]. After exclusion of a relatively high number of cases due to unclear diagnosis or insufficient records, 264 (GPA = 183, MPA = 54, EGPA = 27) patients were reclassified with ACR/EULAR. Information on prior classification with established criteria is not provided. The study reports sensitivities of 88% for GPA, 94% for MPA and 74% for EGPA, 17 patients could not be classified by ACR/EULAR. ANCA-serology is provided for 261 cases with 59% PR3-positives, 31% MPO-positives and 9% ANCA-negatives. The researchers noted a large impact of serology on reclassification to GPA and MPA reflected in the finding that 94% of GPA patients were PR3 positive and 100% of the MPA patients MPO-positive in their results. The study therefore considers the new criteria to be step closer to a serology-based classification.

Turkey

A Turkish study [49] with 164 patients with AAV (clinical diagnosis: 77% GPA, 15% MPA, 8% EGPA) diagnosed between 2016 and 2022, recruited from two academic centres and reclassified by ACR-EULAR, describes a kappa for agreement with EMA classification of 0.79 for GPA, 0.7 for MPA and 0.82 for EGPA. There were two double-classified cases (MPA + GPA) and four former GPA patients are now classified to MPA due to positive anti-MPO. Only 7% of cases could not be classified (2 with granuloma on biopsy). Four patients, earlier classified as GPA, all with granuloma on biopsy, were now assigned MPA due to MPO positivity.

Japan

Two nationwide inception cohorts were used in this evaluation study [50]. All patients were classified using EMA algorithm and were then validated using the ACR/EULAR 2022 criteria. A total of 477 patients were included in the study. The majority of the cases were MPO-ANCA positive (85%), 9% PR3-ANCA positive and 6% were ANCA-negative. Using these two classifications, MPA was the dominant phenotype (EMA:57%, ACR/EULAR: 76%). Details on the composition of the cohort are given in Table 2. A high proportion of patients being either classified to EGPA or MPA by EMA retained this diagnosis with ACR/EULAR, in contrast, 41% of EMA GPA patients were instead classified to MPA. 82.1% of ACR/EULAR GPA patients were PR3-positive whereas 100% of the ACR/EULAR MPA patients exhibited MPO-positivity, reinforcing the weight of serology in ACR/EULAR. In this study, a small number (n = 4) of patients classified to MPA with ACR/EULAR exhibit granuloma on biopsy. A large proportion of patients that could not be classified by EMA, could be classified by ACR/EULAR, all of these patients were MPO-positive and a majority exhibited ILD, thus most were assigned to MPA. The new criteria might be more useful in Asian cohorts with dominating MPO-positivity and a high degree of ILD. On the other hand the authors also report difficulties in classifying ILD in MPO-positive cases, ILD has been shown to occur in EGPA and GPA in a Japanese cohort study but most cases will be classified to MPA with ACR/EULAR.

Other Evaluations

A Portuguese study, presented as meeting abstract including 152 patients with MPA and GPA also compared EMA and ACR/EULAR [51], reporting a shift in classification of 34.7% earlier EMA GPA-classified patients to either MPA or unclassifiable vasculitis. Even in this cohort there were a few patients with EMA GPA and granuloma on biopsy now classified to MPA using the new ACR/EULAR criteria. An Indian evaluation, presented as abstract [52] with patients from the Indian vasculitis registry, reports on 302 relatively young AAV patients with a mean age of 42.9 ± 14.7, classified by EMA, 2022 ACR/EULAR and ACR1990. Classification was then compared to clinicians diagnosis, considered gold standard. The new criteria demonstrated highest agreement in all categories. In ANCA negatives, predominant ocular or CNS GPA were missed by the new criteria. An Ukrainian study, presented as abstract reports better performance than ACR1990 and EMA in 42 patients with AAV, but 20% of patients can either not be classified or are assigned two diagnoses [53].

Sweden

Our evaluation of the new criteria employed a population-based cohort of 374 (47% female) validated cases with AAV [54], previously classified by the EMA algorithm. We compared classification outcome of ACR/EULAR to i) EMA and to a ii) strictly ANCA serology-based classification without any impact from clinical or histologic features. In general, the new ACR/EULAR criteria showed good agreement with EMA, with 96% for EGPA, 85% for GPA and 75% for MPA. We observed a low number of cases that could not be classified (3.5%) or are classified to two categories (1.1%). Fourteen percent of cases (69% MPA, 31%GPA) are assigned a different diagnosis with the new criteria due to ANCA-specificity. We further observed 4 cases exhibiting granuloma on biopsy but assigned to MPA. Regarding ANCA serology-based classification; very high agreement was observed with 98.9% for GPA and 84% for MPA, upon excluding cases with EGPA (n = 23), these numbers increased to 99.5% and 88.2% respectively (Fig. 1). There were 13 unclassifiable cases, 6 were ANCA-negative which corresponds to a third of all ANCA-negatives in the cohort. This might indicate difficulties classifying ANCA-negatives with ACR/EULAR. ANCAs are less frequent in EGPA, the predominant ANCA serotype MPO, presumably favouring classification to EGPA in certain cases is not even included in the EGPA category, the other serotype is included with negative three points thus making classification to EGPA more unlikely. Of 23 EGPA (EMA) in the Swedish cohort eight are MPO positive which does not impact classification at all, there are no PR3 positive patients, but assuming all the EGPA patients were indeed PR3 positive, all would qualify to be classified to EGPA as they all generate sufficient score with the other items. Of course, some of these patients would then even be classified as GPA, so many would shift to the double classified group. This means inclusion of ANCA in the EGPA category does not change the classification towards EGPA score in our patients at all.

Fig. 1

Classification outcome in Skåne-AAV cohort with permission from Rheumatology (Oxford) [54]

Challenges in Using the new ACR/EULAR 2022 Criteria

Based on available evaluations, there are some challenges when applying/using the new 2022 ACR/EULAR classification criteria of AAV; these are summarized as:

1.

Overlapping classification; cases are classified to more than one phenotype.

2.

Unclassifiable cases are described in all studies.

3.

Disagreement with current understanding of AAV histopathology; the CHCC defined granuloma as a key feature of GPA, but cases with granuloma on biopsy are classified to MPA in several of the cited studies. Granuloma generates low score in the GPA towards GPA classification, but it is not included in the MPA at all, thus granuloma can now be a feature of MPA given MPO positivity or other items classifying towards MPA are present. In this case a serotype feature (MPO-positivity) clearly outweighs a phenotypic feature as the weight of granuloma is low or not existing in the MPA category.

4.

Interstitial lung disease (ILD) is poorly defined in the ACR/EULAR criteria. It is not clear when the criteria of ILD are met. However, given the results from the Japanese study, the addition of ILD is important, as more patients can be classified in populations where ILD as a clinical manifestation of AAV is common.

5.

Challenges related to ANCA

a.

ANCA-negative cases: potentially ANCA-negative cases could be become more difficult to classify, which our own findings support, however more studies are needed in this subgroup for clarification.

b.

ANCA weight and test method: the inclusion of immunofluorescence in the criteria can, in our eyes, be problematic, as positive IIF is observed in several other diseases, we consider the use of ELISA based test to be more specific.

c.

A considerable number of patients are classified to another phenotype. This applies primarily to MPO-positive patients who will often be assigned MPA primarily due to MPO positivity. Features typical for non MPA phenotypes might, due to ANCAs weight, be “outscored” and classified to MPA.