Arthritis & Rheumatism, Volume 62,
November 2010 Abstract Supplement

Abstracts of the American College of
Rheumatology/Association of Rheumatology Health Professionals
Annual Scientific Meeting
Atlanta, Georgia November 6-11, 2010.

Cluster Analysis and Clinical Phenotypes of ANCA-Associated Vasculitis.

Mahr3,  Alfred, Katsahian1,  Sandrine, Guillevin4,  Loïc, Hagen5,  E. Christiaan, Hoglund2,  Peter, Merkel9,  Peter A., Pagnoux4,  Christian

Clinical Epidemiology and Biostatistics, Hospital Saint-Louis, Paris, France
Clinical Pharmacology, Department of Laboratory Medicine, Lund University, Lund, Sweden
Internal Medicine, Hospital Cochin, Paris, Paris, France
Internal Medicine, Hospital Cochin, Paris, France
Internal Medicine, Meander Medical Center, Amersfoort, Netherlands
Nephrology and Transplantation, Skane University Hospital Malmö, Lund University, Sweden
Nephrology, Addenbrooke's Hospital, Cambridge, United Kingdom
Otolaryngology, Rigshospitalet, Copenhagen, Denmark
Rheumatology and Clinical Epidemiology Unit, Boston University School of Medicine, Boston, MA


Wegener's granulomatosis (WG) and microscopic polyangiitis (MPA) have widely overlapping clinical, serological, and histological features and are increasingly considered phenotypic variants of a single entity termed "ANCA-associated vasculitis (AAV)". Applying current definitions to distinguish between WG and MPA is sometimes problematic, and physicians' and researchers' disease classifications are heterogeneous. In addition, clinical practice and observational study findings suggest that WG and MPA clinical forms may not reflect the full range of AAV expressions. This study applied cluster analysis of a large combined cohort to explore the clinical phenotyping spectrum of AAV.


A large dataset of patients with newly diagnosed WG and MPA originally enrolled in 5 clinical trials on AAV was analyzed. Eleven baseline variables, including 9 clinical variables (age at diagnosis, sex, ear/nose/throat, ocular, lung, renal, neurological, skin and cardiovascular manifestations), PR3-ANCA, and MPO-ANCA, served as primary variables for the cluster analysis. The clinical relevance of the generated clusters was analyzed by describing their summary characteristics and their patient outcomes.


The dataset comprised 673 subjects, diagnosed as follows: 396 (59%) WG and 281 (41%) MPA. Over a mean follow-up of 4.4 years, 136 (20%) patients died. The analyses yielded 4 distinct phenotypic classes: 87 (13%) non-renal AAV, 291 (43%) renal AAV with PR3-ANCA, 231 (34%) renal AAV without PR3-ANCA, and 64 (10%) AAV with cardiovascular involvement. The distribution of WG-diagnosed patients into those 4 respective classes was 95%, 79%, 18% and 64%. Those 4 classes represented highly distinct mortality rates (P < 0.0001, log-rank test) (Figure) and frequencies of patients relapsing >=1 times (range: 20–53%; P < 0.0001, c2 test). Using 3 of the 9 variables (presence or absence of renal involvement, cardiovascular involvement, and PR3-ANCA), 651 of the 673 subjects (97%) were accurately allocated to 1 of the 4 classes.


Our cluster analysis results suggest that AAV encompasses 4 phenotypic classes: non-renal AAV, renal AAV with PR3-ANCA, renal AAV without PR3-ANCA and AAV with cardiovascular involvement, associated with different outcomes. Compared to the traditional WG–MPA separation, this new classification might better reflect the phenotypic groupings within the AAV spectrum and more accurately stratify patients into homogeneous disease groups for clinical, epidemiological and basic research.

To cite this abstract, please use the following information:
Mahr, Alfred, Katsahian, Sandrine, Guillevin, Loïc, Hagen, E. Christiaan, Hoglund, Peter, Merkel, Peter A., et al; Cluster Analysis and Clinical Phenotypes of ANCA-Associated Vasculitis. [abstract]. Arthritis Rheum 2010;62 Suppl 10 :2014
DOI: 10.1002/art.29779

Abstract Supplement

Meeting Menu