Article Text

OP0209 Cluster Analysis to Explore Subclassification of Eosinophilic Granulomatosis with Polyangiitis (Churg–Strauss)
  1. T. Neumann1,
  2. F. Moosig2,
  3. A. Vaglio3,
  4. J. Zwerina4,
  5. P. Bremer2,
  6. A. Gioffredi3,
  7. J. Grosskreutz5,
  8. C. Kroegel6,
  9. F. Maritati3,
  10. M. Resche-Rigon7,
  11. A. Mahr8
  1. 1Department of Medicine III, Jena University Hospital Jena, Jena
  2. 2Department of Rheumatology, Klinikum Bad Bramstedt, Universitätsklinikum Schleswig-Holstein, Bad Bramstedt, Germany
  3. 3Department of Clinical Medicine and Nephrology, University Hospital of Parma, Parma, Italy
  4. 4Department of Internal Medicine 3, Institute of Clinical Immunology, University of Erlangen-Nuremberg, Erlangen
  5. 5Department of Neurology
  6. 6Department of Medicine I, Jena University Hospital Jena, Jena, Germany
  7. 7Department of Statistics, Saint-Louis Hospital, University Paris 7–René Diderot
  8. 8Department of Internal Medicine, Hospital Saint-Louis Dept. of Internal Medicine, Paris, France


Background Results from descriptive studies on eosinophilic granulomatosis with polyangiitis (EGPA) hint at distinct clinical subclasses, which might notably be determined by ANCA status.

Objectives This study explored the possibility of subclassifying EGPA using hierarchical cluster analysis.

Methods A standardized retrospective dataset from a cohort with clinical diagnoses of EGPA followed in 4 tertiary referral centers was used. Hierarchical cluster analysis using Ward’s method was performed based on the following 12 input variables assessed at diagnosis: constitutional symptoms, mucucutaneous, ophthalmological, ENT, cardiovascular, gastrointestinal, renal, and central nervous involvement, peripheral neuropathy, non-fixed lung infiltrates and ANCA positivity. The resulting clusters were described by their most prominent summary characteristics. In addition, the distribution of clinical variables according to ANCA status was compared with chi-square tests.

Results The analyzed dataset included 262 EGPA cases diagnosed 1984–2012. ANCA were detected in 30.9%. The cluster analysis produced 3 clusters of respectively 39 (cluster 1), 92 (cluster 2) and 131 subjects (cluster 3). They were characterized as follows: cluster 1 by renal involvement (84.6%) and high ANCA prevalence (92.3%); cluster 2 by virtually absent renal involvement (3.3%) and ANCA (4.3%); and cluster 3 by an intermediate phenotype with renal involvement (13%), presence of ANCA (31.3%) and more frequent cardiovascular (59.5% vs. 17.9% and 35.9% for clusters 1 and 2, respectively) and gastrointestinal involvement (42% vs. 15.4% and 12%). Stratification of the 11 clinical input variables by ANCA status showed that ANCA positivity was associated with renal disease (P < 0.0001), peripheral neuropathy (P = 0.005) and constitutional symptoms (P = 0.02).

Conclusions Although reinforcing the link between ANCA positivity and renal involvement, the cluster analysis does not suggest that EGPA is composed of clearly separated and mutually exclusive subclasses.

Disclosure of Interest None Declared

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.