Article Text

Download PDFPDF
Extended report
Revisiting the classification of clinical phenotypes of anti-neutrophil cytoplasmic antibody-associated vasculitis: a cluster analysis
  1. Alfred Mahr1,
  2. Sandrine Katsahian2,
  3. Hugo Varet2,
  4. Loïc Guillevin3,
  5. E Christiaan Hagen4,
  6. Peter Höglund5,
  7. Peter A Merkel6,
  8. Christian Pagnoux3,
  9. Niels Rasmussen7,
  10. Kerstin Westman8,
  11. David R W Jayne9,
  12. for the French Vasculitis Study Group (FVSG) and the European Vasculitis Society (EUVAS)
  1. 1Department of Internal Medicine, Hospital Saint-Louis, Paris, France
  2. 2Clinical Epidemiology and Biostatistics, Hospital Saint-Louis, Paris, France
  3. 3Department of Internal Medicine, Hospital Cochin, Paris, France
  4. 4Department of Internal Medicine, Meander Medical Center, Amersfoort, The Netherlands
  5. 5Department of Clinical Pharmacology, Department of Laboratory Medicine, Lund University, Lund, Sweden
  6. 6Division of Rheumatology, University of Pennsylvania, Philadelphia, PA, USA
  7. 7Otolaryngology, Rigshospitalet, Copenhagen, Denmark
  8. 8Nephrology and Transplantation, Skåne University Hospital Malmö, Lund University, Malmö, Sweden
  9. 9Department of Nephrology, Addenbrooke's Hospital, Cambridge, UK
  1. Correspondence to Dr Alfred Mahr, Department of Internal Medicine, Hospital Saint-Louis, University Paris 7-Paris Diderot, Assistance Publique–Hôpitaux de Paris, 1, Avenue Claude-Vellefaux, 75475 Paris Cedex 10, France; alfred.mahr{at}


Background Granulomatosis with polyangiitis (Wegener's) (GPA) and microscopic polyangiitis (MPA) are subgroups of anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) defined historically by clinical and histological features. GPA and MPA are heterogeneous entities with overlapping phenotypes. To identify novel subgroupings, cluster analysis was used to explore the phenotypic spectrum of AAV.

Methods This study used a dataset of patients newly diagnosed as having GPA and MPA enrolled in five clinical trials. One cluster model included nine clinical baseline variables as input variables, and a second cluster model additionally included ANCA specificities. The clustering process involved multiple correspondence analyses followed by hierarchical ascendant cluster analysis. The clinical relevance of the generated clusters was analysed by their summary characteristics and outcomes.

Results The analyses involved data for 673 subjects: 396 (59%) with GPA and 277 (41%) with MPA. Both cluster models resulted in five partially redundant clusters of subjects, and the model including ANCA resulted in more pertinent separations. These clusters were named ‘renal AAV with proteinase 3 (PR3)-ANCA’ (40% of subjects), ‘renal AAV without PR3-ANCA’ (32%) and ‘non-renal AAV’ (12%), ‘cardiovascular AAV’ (9%) and ‘gastrointestinal AAV’ (7%). The five clusters had distinct death and relapse rates. On the basis of 4 variables, 651 subjects (97%) could be accurately allocated to 1 of the 5 classes.

Conclusions This analysis suggests that AAV encompasses five classes associated with different outcomes. As compared with the traditional GPA–MPA separation, this classification system may better reflect the phenotypic spectrum of AAV.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.