The vasculitides are a set of related disorders characterized by blood vessel inflammation leading to tissue or end-organ injury. The classification of vasculitis has been a challenging problem for decades . In 1990 the American College of Rheumatology (ACR) proposed criteria for the classification of primary vasculitides based on data comparing clinical features of patients with seven types of established vasculitis: giant cell arteritis (GCA), Takayasu arteritis (TA), Wegener's granulomatosis (WG), Churg–Strauss syndrome (CSS), polyarteritis nodosa (PAN), Henoch–Schönlein purpura (HSP) and hypersensitivity vasculitis (HSV) [2–8]. The sensitivities for classifying patients correctly ranged from 71·0% to 95·3% and the specificities ranged from 78·7% to 99·7%. The ACR criteria have been widely accepted, applied and beneficial for clinical investigation (the major purpose of the initiative).
Over the past 20 years, however, important limitations have been recognized in the ACR classification criteria for vasculitis which affect their validity with respect to clinical research and practice. Evolving diagnostic techniques (e.g. imaging) and better understanding of pathophysiology have contributed to a better distinction between conditions. ACR criteria were developed before the widespread use of anti-neutrophil cytoplasm antibody (ANCA) testing, which assists in the diagnosis of WG, microscopic polyangiitis (MPA) and CSS and helps to exclude a diagnosis of PAN [9–12]. Furthermore, no distinction was made between PAN and microscopic polyangiitis , even though the latter was described as a discrete entity in 1948 . Methodologically, each type of vasculitis was compared to a non-selective control group comprised of both small- and large-vessel vasculitis, and the submitting physician diagnosis was used as the gold standard. Finally, while the ACR criteria were not intended originally to be diagnostic tools, they have nevertheless been widely adopted as such; however, the criteria do not work well for diagnostic purposes  and this can lead to problems in clinical practice as well as in the conduct of clinical trials.
In 1994, the Chapel Hill Consensus Conference (CHCC) produced definitions for vasculitis including MPA, but these were not intended for either classification or diagnostic purposes . The definitions allow for the fact that relevant histological data are not always available and suggest the use of surrogate markers of vasculitis. Although the ACR criteria and CHCC definitions have been used in parallel, this approach results in an unacceptable degree of overlap, especially between MPA and PAN. To address this problem, an algorithm to classify AAV for epidemiological studies was developed that includes MPA and incorporates ANCA .
The shortcomings of the present systems for classification, diagnosis and definitions for the vasculitides affect both practising clinicians and researchers. For example, there are difficulties in deciding whether to use ACR classification criteria or CHCC definitions as inclusion criteria for studies. We have recently drafted a revision of the classification scheme based on (i) the traditional approach of classifying vasculitis by size of predominant vessel involved, (ii) diagnostic auto-antibodies (ANCA) and (iii) current understanding of pathogenesis . Vasculitis will probably continue to be divided into primary and secondary forms. Additional factors to be incorporated into a new classification system include aetiopathogenesis or the major salient features of condition, specific tissue biopsy data, ANCA testing, new forms of diagnostic imaging and any new biomarkers that may become available in future . A system incorporating new diagnostic approaches, updated knowledge of pathophysiology and patient-level data could result in novel groupings and a substantially revised set of classification and diagnostic criteria.
A major international effort is now under way to use data-driven methods  to develop both a revised single classification system for the vasculitides and a validated set of diagnostic criteria for the vasculitides in accordance with standards established by the ACR and the European League Against Rheumatism (EULAR); the study is named ‘Diagnosis and Classification of Vasculitis (DCVAS)’, and is supported by grants from the Vasculitis Foundation, the American College of Rheumatology (ACR) and the European League Against Rheumatism (EULAR). As a starting-point we are using existing classification terms and vessel size discrimination to identify patients with systemic vasculitis for study; we will focus on the most important forms of primary vasculitis for data collection to help avoid circularity. We plan to recruit patients with vasculitis and a comparator cohort with other autoimmune diseases/mimics of vasculitis and analyse the clinical, serological, pathological and radiological parameters used to make a diagnosis of vasculitis, and develop a multivariate analysis model of key factors which discriminate between conditions. As part of this process, we will create a series of vignettes based on the cases acquired to identify important discriminating variables that group patients into a specific type of vasculitis in the opinion of an expert panel. In addition, we will create criteria which distinguish patients with vasculitis from those with a similar presentation (context-specific diagnostic criteria).
Determining an appropriate gold standard is an inherent challenge to classifying disease syndromes. A gold standard derived from physician-based diagnosis is problematic, because data elements used to inform the gold standard may also be incorporated into the predictive model. The new initiative will avoid this inherent circularity by using real and hypothetical clinical vignettes, classified by consensus of an expert panel, to establish relevant cut-points and identify specific clustering of features to define the important components of each disease. Pathophysiologically based approaches will be made (e.g. using the presence of ANCA or immune complexes) to distinguish one form of vasculitis from another. As a consequence of this process, new categories of vasculitis may emerge. We will test the validity of these criteria in development and validation cohorts that will include patients with vasculitis and patients without vasculitis who present with similar clinical features to those with vasculitis. We anticipate recruiting > 2000 patients with primary systemic vasculitis and > 1500 patients with conditions that mimic vasculitis. Currently, we have commitment from more than 50 medical centres from Europe, North and South America, Australasia, China and Japan.
Recruitment for this new data-driven approach to classification and diagnosis in vasculitis has commenced. We urge as many centres as possible to join DCVAS. More information for the study is available at: http://clinicaltrials.gov/ct2/show/results/NCT01066208, and the study co-ordinator can be contacted at: firstname.lastname@example.org.