Eosinophilic Granulomatosis With Polyangiitis (Churg-Strauss)

Transcription

1 ARTHRITIS & RHEUMATISM Vol. 65, No. 1, January 2013, pp DOI /art , American College of Rheumatology Eosinophilic Granulomatosis With Polyangiitis (Churg-Strauss) Clinical Characteristics and Long-Term Followup of the 383 Patients Enrolled in the French Vasculitis Study Group Cohort Cloé Comarmond, 1 Christian Pagnoux, 1 Mehdi Khellaf, 2 Jean-François Cordier, 3 Mohamed Hamidou, 4 Jean-François Viallard, 5 François Maurier, 6 Stéphane Jouneau, 7 Boris Bienvenu, 8 Xavier Puéchal, 9 Olivier Aumaître, 10 Guillaume Le Guenno, 11 Alain Le Quellec, 12 Ramiro Cevallos, 13 Olivier Fain, 14 Bertrand Godeau, 2 Raphaèle Seror, 15 Bertrand Dunogué, 1 Alfred Mahr, 1 Philippe Guilpain, 16 Pascal Cohen, 1 Achille Aouba, 1 Luc Mouthon, 1 and Loïc Guillevin, 1 for the French Vasculitis Study Group Objective. Earlier studies of eosinophilic granulomatosis with polyangiitis (Churg-Strauss) (EGPA), with limited patient numbers and followup durations, demonstrated that clinical presentation at diagnosis, but not outcome, differed according to antineutrophil cytoplasmic antibody (ANCA) status. This study was undertaken to describe the main characteristics of a larger patient cohort and their long-term outcomes. Methods. A retrospective study of EGPA patients in the French Vasculitis Study Group cohort who satisfied the American College of Rheumatology criteria and/or Chapel Hill definitions was conducted. Patient Dr. Pagnoux is recipient of fellowships from the Vasculitis Clinical Research Consortium and the Toronto General Hospital/ Toronto Western Hospital Foundation/University Health Network. 1 Cloé Comarmond, MD, Christian Pagnoux, MD, MPH, MSc (current address: Mount Sinai Hospital, University Health Network, Toronto, Ontario, Canada), Bertrand Dunogué, MD, Alfred Mahr, MD, MPH, PhD, Pascal Cohen, MD, Achille Aouba, MD, MPH, Luc Mouthon, MD, PhD, Loïc Guillevin, MD: Hôpital Cochin, AP-HP, and Université Paris Descartes, Paris 5, Paris, France; 2 Mehdi Khellaf, MD, Bertrand Godeau, MD, PhD: Centre Hospitalier Universitaire Henri-Mondor, AP-HP, and Université Paris-Est Créteil, Créteil, France; 3 Jean-François Cordier, MD: Hôpital Louis-Pradel, Hospices Civils de Lyon, and Université Claude-Bernard Lyon I, Lyon; 4 Mohamed Hamidou, MD, PhD: University Hospital Hôtel-Dieu, Nantes, France; 5 Jean-François Viallard, MD, PhD: Hôpital Haut-Lévêque and Université Victor Segalen Bordeaux 2, Centre Hospitalier Universitaire de Bordeaux, Bordeaux, France; 6 François Maurier, MD: Centre Hospitalier Régional de Metz, Metz, France; 7 Stéphane Jouneau, MD, PhD: Centre Hospitalier Universitaire de Rennes, Rennes, France; 8 Boris Bienvenu, MD, PhD: Division of Internal Medicine, Centre Hospitalier Universitaire de Caen, Côte de Nacre, Caen, France; 9 Xavier Puéchal, MD, PhD: Le Mans General Hospital, Le Mans, France; 10 Olivier Aumaître, MD, PhD: Centre Hospitalier Universitaire, Hôpital Gabriel-Montpied, Clermont-Ferrand, France; characteristics and outcomes were compared according to ANCA status and year of diagnosis. Results. We identified 383 patients diagnosed between 1957 and June 2009 (128 [33.4%] before 1997 or earlier) and followed up for a mean SD of months. At diagnosis, their mean SD age was years, and 91.1% had asthma (duration years). Main manifestations included peripheral neuropathy (51.4%); ear, nose, and throat (ENT) signs (48.0%); skin lesions (39.7%); lung infiltrates (38.6%); 11 Guillaume Le Guenno, MD, PhD: Centre Hospitalier Universitaire Estaing, Clermont-Ferrand, France; 12 Alain Le Quellec, MD, PhD: Hôpital Saint-Eloi, Centre Hospitalier Universitaire de Montpellier, Montpellier, France; 13 Ramiro Cevallos, MD, PhD: Groupe Hospitalier Saint-Vincent, Strasbourg, France; 14 Olivier Fain, MD: Centre Hospitalier Universitaire Jean-Verdier, Bondy, France; 15 Raphaèle Seror, MD, PhD: Hôpital Cochin, AP-HP, Université Paris Descartes, Paris 5, Paris, France, and Université Paris, Sud 11 and AP-HP Hôpital Bicêtre, Le Kremlin-Bicêtre, France; 16 Philippe Guilpain, MD, PhD: Hôpital Cochin, AP-HP, and Université Paris Descartes, Paris 5, Paris, France, and Université Montpellier 1 and Hôpital Saint-Eloi, Montpellier, France. Drs. Comarmond and Pagnoux contributed equally to this work. Dr. Pagnoux has received consulting fees, speaking fees, and/or honoraria from Hoffmann-La Roche and Schering-Plough (less than $10,000 each). Dr. Bienvenu has received speaking fees from Octapharma and Sanofi-Aventis (less than $10,000 each). Dr. Seror has received consulting fees, speaking fees, and/or honoraria from Hoffmann-La Roche, Pfizer, and GlaxoSmithKline (less than $10,000 each). Address correspondence to Christian Pagnoux, MD, MPH, MSc, Division of Rheumatology, Mount Sinai Hospital, 60 Murray Street S-2-220, Toronto, Ontario M5G 1X5, Canada. cpagnoux@mtsinai.on.ca. Submitted for publication January 14, 2012; accepted in revised form September 20,

2 EOSINOPHILIC GRANULOMATOSIS WITH POLYANGIITIS (CHURG-STRAUSS) 271 and cardiomyopathy (16.4%). Among the 348 patients tested at diagnosis for ANCA, the 108 ANCA-positive patients (31.0%) had significantly more frequent ENT manifestations, peripheral neuropathy, and/or renal involvement, but less frequent cardiac manifestations, than the ANCA-negative patients. Vasculitis relapses occurred in 35.2% of the ANCA-positive versus 22.5% of the ANCA-negative patients (P 0.01), and 5.6% versus 12.5%, respectively, died (P < 0.05). The 5-year relapsefree survival rate was 58.1% (95% confidence interval [95% CI] ) for ANCA-positive and 67.8% (95% CI ) for ANCA-negative patients (P 0.35). Multivariable analysis identified cardiomyopathy, older age, and diagnosis during or prior to 1996 as independent risk factors for death and lower eosinophil count at diagnosis as predictive of relapse. Conclusion. The characteristics and long-term outcomes of EGPA patients differ according to their ANCA status. Although EGPA relapses remain frequent, mortality has declined, at least since Eosinophilic granulomatosis with polyangiitis (Churg-Strauss) (EGPA) is one of the rarest but still potentially life-threatening systemic necrotizing vasculitides predominantly affecting small vessels (1 3). Since the first description of the disease in 1951 (4), several descriptive series have been published, but each included fewer than 120 patients and had limited followup (5 15). Whereas EGPA pathophysiology remains only partially understood, it has consistently been found that 30 40% of affected patients have antineutrophil cytoplasmic antibodies (ANCA), mainly ANCA with a perinuclear labeling pattern on immunofluorescence (IF), usually directed against myeloperoxidase (MPO) on enzyme-linked immunosorbent assay (ELISA) (13,16,17). Two of the most recent series, including one of 112 patients described by our group, concurrently demonstrated that ANCA-positive patients differed from their ANCA-negative counterparts in that ANCA-positive patients had a lower frequency of cardiac involvement but more frequent renal disease and peripheral nerve involvement (16,17). However, and despite EGPArelated cardiomyopathy having been shown by our group since 1996 to carry a poor prognosis (6,18,19), along with gastrointestinal, central nervous system (CNS), and renal involvement, respective prognoses for patient subsets based on ANCA status did not differ. One possible explanation was the relatively short mean followup durations of those studies, which did not exceed 3 years (mean SD months for the French Vasculitis Study Group [FVSG] analysis [17]). Therefore, we decided to conduct a new investigation of all EGPA patients with data entered into the FVSG cohort database since its initiation in 1983, including the 112 previously described patients (6,17,19), in order to further analyze their characteristics and determine whether longer-term outcomes differed according to ANCA status and year of diagnosis. We also attempted to identify potential predictors of vasculitis relapse by using a stringent definition of relapse and analyzing nonspecific and isolated asthma or allergic rhinitis flares separately. PATIENTS AND METHODS Patients. We conducted a systematic retrospective study of patients with EGPA according to the American College of Rheumatology (ACR) 1990 criteria (20) and/or Chapel Hill definition (1) who had data entered into the FVSG database before July (See Appendix A for members of the FVSG.) All patient data have been collected in the FVSG database since its inception in 1983, when it was declared to the Commission Nationale Informatique et Libertés. For patients who participated in prospective therapeutic trials, written informed consent for the prospective collection and further analysis of their data was obtained at enrollment. All other patients received oral and written information attesting to their unrestricted rights to ask for the deletion of their data. The diagnosis of EGPA was reassessed by the investigators (CC, CP, and LG), taking into account the entire followup period, so that those patients who were eventually diagnosed as having other systemic eosinophilic disorders were excluded (21). Patients with insufficient data (only 1 visit and/or no clinical details), concomitant human immunodeficiency virus and/or hepatitis C virus infection, cryoglobulinemia, or other systemic diseases were also excluded. Baseline measurements. For patients who had been included in clinical trials, data were extracted from the standardized record forms that have been used since 1983 for FVSG clinical trials. For other patients, medical charts were retrieved and reviewed for demographic characteristics and clinical, biologic, radiologic, and histologic findings. Among the recorded and analyzed clinical manifestations, the following definitions were applied. Fever was defined as a temperature of 38.5 C, weight loss as a loss of 3 kg over the 3 months preceding diagnosis, and surgical abdomen as severe gastrointestinal symptoms requiring intervention or at least consultation with a surgeon. Heart manifestations, including cardiomyopathy, were diagnosed based on clinical findings in combination with ultrasound and electrocardiogram results (thereby corresponding to the Five-Factors Score [FFS] definitions) (19). The biologic parameters studied included proteinuria ( 0.4 gm/24 hours), serum creatinine level, C-reactive protein (CRP) level at diagnosis, and ANCA status (considered to be positive when ANCA were detected by IF and/or ELISA in the serum at diagnosis or later during the disease course when they had initially tested negative). For patients who were diagnosed as having EGPA before the dissemination of ANCA testing, these tests were run on serum samples that were obtained at the time of diagnosis and frozen, when available. Biopsy findings were also recorded, as normal or as

3 272 COMARMOND ET AL showing evidence of vasculitis, granuloma, and/or eosinophilic infiltrates (or necrotizing glomerulonephritis for kidney biopsy specimens). Disease activity and scores. Disease activity was assessed at diagnosis using the original 1994 Birmingham Vasculitis Activity Score (BVAS) (22) and recorded in the database. The prognostic FFS, devised in 1996 (19) and based on 5 items (severe vasculitis-related gastrointestinal involvement, CNS involvement, and/or cardiomyopathy, serum creatinine level 140 moles/liter, and/or proteinuria 1 gm/24 hours) was recorded, as was the revised FFS for EGPA (severe vasculitis-related gastrointestinal involvement and/or cardiac involvement, serum creatinine level 150 moles/liter, age older than 65, and/or absence of ear, nose, and throat [ENT] symptoms) (23). All scores were obtained from the FVSG database or calculated retrospectively for those patients who were diagnosed before the scoring systems were created. Treatment procedures. Patients included in clinical trials were treated according to the scheduled therapeutic regimens (24 29). Others were treated with what was considered to be the standard of care when they were diagnosed. Drugs used to achieve remission were recorded, as were those given subsequently for maintenance or relapse, when applicable, and prednisone dose at last assessment, when it was available. Outcomes. Outcomes studied included relapses, termination of corticosteroid treatment, deaths and their causes, and cancers occurring during followup. Relapse was predefined as either 1) the recurrence or worsening of a clinical EGPA manifestation following a period of remission of 3 months, or 2) the need for the addition of, or a change in, immunosuppressive drugs and/or the reinstitution of prednisone or an increase in prednisone dosage of more than twice the previous dosage, resulting in a total dosage of 30 mg/day. An increase in the eosinophil count without any other clinical EGPA manifestations; or isolated asthma, sinusitis, or rhinitis exacerbations, with or without a concomitant increase in eosinophil count, were recorded when considered clinically significant by the patient s referring physician to justify any therapeutic adjustment but were not considered a failure, recurrent EGPA manifestations, or relapse per se, and were analyzed separately. Remission was defined as the absence of disease activity attributable to EGPA vasculitis manifestations (parenchymal lung disease; peripheral nerve involvement; or skin, cardiac, CNS, renal, and/or gastrointestinal signs) for 3 consecutive months, corresponding to a BVAS of 0. According to this definition of remission, patients could still be receiving a specified prednisone dose for isolated asthma or rhinitis. Statistical analysis. Patient characteristics are reported as the number and percentage for categorical variables and as the mean SD for continuous variables. Patient subsets were established based on ANCA status, sex, or year of diagnosis. For these subsets, quantitative variables were compared using Student s t-test, and categorical variables were compared using the chi-square test or Fisher s exact test, as appropriate. Patient survival was analyzed using life tables and the Kaplan-Meier method (30) and were compared using log rank tests (31). Survival was calculated from the time of EGPA diagnosis to death or the last followup visit. Relapse-free survival was calculated from the time of EGPA diagnosis to relapse, death, or end of followup. Main patient demographics (sex, age at diagnosis, year of diagnosis prior to/during 1996 or after 1996, i.e. after the FFS was devised, and prior inclusion in a therapeutic trial) and clinical, biologic, and therapeutic parameters ( 1 immunosuppressant in the initial therapeutic regimen or during followup) with P 0.20 in the univariable analysis were included in multivariable Cox proportional hazards models to identify independent predictors of death or relapse. The FFS and revised FFS were not included in the multivariable model, because their composite items had already been analyzed separately. We used a backward stepwise procedure to determine the parameters of the final model (32). Results are expressed as hazard ratios (HRs) and 95% confidence intervals (95% CIs). The standardized incidence ratio (SIR) of cancers observed during followup was calculated for comparison with the expected cancer rate in the general population of France, with corresponding sex and age (expected cancer incidence rates from 1975 to 2005 were obtained from sante.fr/surveillance/cancers/estimations_cancers/default.htm, and incidence rates for the years 1975 and 2005 were used for the periods prior to 1975 and since 2005, respectively). All statistical analyses were conducted using Stata software version 10 (StataCorp), except for the SIR calculation, which was computed with Pamcomp software version 1.41 (33). P values less than 0.05 were considered significant. RESULTS Patient identification and characteristics. After excluding 60 patients with concomitant infections, systemic disease, and/or insufficient recorded data, we identified 383 patients in the FVSG database who satisfied the inclusion criteria and were diagnosed as having EGPA between March 1957 and July No patient was excluded because of a revised diagnosis of another systemic eosinophilic disorder during followup. Among these selected patients, 210 (54.8%) had been included in prospective therapeutic trials (24 29), and 128 (33.4%) had been diagnosed during or prior to Their demographic characteristics and main clinical manifestations at diagnosis are shown in Table 1. The mean SD age at diagnosis was years, and neither sex was predominant (male:female ratio 1.08). Of the 383 patients, 376 (98.2%) were white, 4 (1%) were of African or Caribbean ancestry, and 3 (0.8%) were of Asian descent. At EGPA diagnosis, 349 (91.1%) of the patients had asthma, which had been present for a mean SD of years, and became manifest no more than 6 months before EGPA diagnosis in 13 (3.4%) of them. The remaining 34 patients had ENT signs and/or developed asthma within 6 months following EGPA diagnosis (n 6). The most common EGPA manifestations at diagnosis included weight loss (49.3%), mononeuritis multiplex (46.0%), nonerosive sinusitis/polyposis (41.8%), skin lesions (39.7%), and lung infiltrates (38.6%). Pleural effusion was reported in 34 (8.9%), among whom 14 (3.7%) also

4 EOSINOPHILIC GRANULOMATOSIS WITH POLYANGIITIS (CHURG-STRAUSS) 273 Table 1. Main clinical characteristics at diagnosis of the 383 patients with EGPA, according to ANCA status* Characteristic All (n 383) ANCA status unknown (n 35) ANCA-positive patients (n 108) ANCA-negative patients (n 240) P Sex, no. male/female 199/184 14/21 60/48 125/ Age at diagnosis, mean SD years Diagnosis in 1996 or before, no. (%) 128 (33.4) 33 (94.3) 27 (25.0) 68 (28.3) 0.52 History of allergy, no. (%) 104 (27.2) 9 (25.7) 31 (28.7) 64 (26.7) 0.69 Asthma, no. (%) 349 (91.1) 31 (88.6) 100 (92.6) 218 (90.8) 0.59 Duration of asthma at EGPA diagnosis, mean SD years Symptoms and manifestations, no. (%) Weight loss 189 (49.3) 25 (71.4) 62 (57.4) 102 (42.5) 0.01 Mean SD kg lost Fever 149 (38.9) 22 (62.9) 44 (40.7) 83 (34.6) 0.27 Myalgias 149 (38.9) 14 (40.0) 44 (40.7) 91 (37.9) 0.62 Arthralgias 114 (29.8) 13 (37.1) 37 (34.3) 64 (26.7) 0.15 ENT manifestations 184 (48.0) 14 (40.0) 64 (59.3) 106 (44.2) 0.01 Rhinitis 65 (17.0) 4 (11.4) 27 (25.0) 34 (14.2) 0.01 Sinusitis/polyposis 160 (41.8) 13 (37.1) 56 (51.9) 91 (37.9) 0.02 Lung manifestations 350 (91.4) 31 (88.6) 100 (92.6) 219 (91.3) 0.68 Chest pain 28 (7.3) 0 6 (5.6) 22 (9.2) 0.25 Lung nodule(s) 20 (5.2) 1 (2.9) 9 (8.3) 10 (4.2) 0.11 Lung infiltrates 148 (38.6) 10 (28.6) 44 (40.7) 94 (39.2) 0.78 Pleural effusion 34 (8.9) 1 (2.9) 5 (4.6) 28 (11.7) 0.04 Alveolar hemorrhage 16 (4.2) 1 (2.9) 8 (7.4) 7 (2.9) 0.06 Cutaneous manifestations 152 (39.7) 16 (45.7) 49 (45.4) 87 (36.3) 0.11 Purpura 86 (22.5) 8 (22.9) 31 (28.7) 47 (19.6) 0.06 Pseudo-urticarial rash, hives 38 (9.9) 1 (2.9) 13 (12.0) 24 (10.0) 0.57 Subcutaneous nodule(s) 37 (9.7) 8 (22.9) 9 (8.3) 20 (8.3) 0.99 Livedo reticularis 15 (3.9) 3 (8.6) 2 (1.9) 10 (4.2) 0.27 Gangrenous necrotic lesions 14 (3.7) 0 5 (4.6) 9 (3.8) 0.70 Neurologic symptoms 211 (55.1) 25 (71.4) 72 (66.7) 114 (47.5) 0.01 Peripheral neuropathy 197 (51.4) 23 (65.7) 68 (63.0) 106 (44.2) 0.01 Mononeuritis multiplex 176 (46.0) 23 (65.7) 59 (54.6) 94 (39.2) 0.01 CNS involvement 20 (5.2) 4 (11.4) 7 (6.5) 9 (3.8) 0.26 Cranial nerve involvement 12 (3.1) 3 (8.6) 3 (2.8) 6 (2.5) 0.88 Cardiovascular manifestations 105 (27.4) 12 (34.3) 20 (18.5) 73 (30.4) 0.02 Raynaud s phenomenon 6 (1.6) 0 2 (1.9) 4 (1.7) 0.90 Cardiomyopathy (FFS item) 63 (16.4) 8 (22.9) 9 (8.3) 46 (19.2) 0.01 Pericarditis 58 (15.1) 4 (11.4) 14 (13.0) 40 (16.7) 0.38 Deep venous thrombosis/pulmonary embolism 29 (7.6) 1 (2.9) 8 (7.4) 20 (8.3) 0.77 Gastrointestinal involvement 89 (23.2) 9 (25.7) 24 (22.2) 56 (23.3) 0.82 Abdominal pain 78 (20.4) 6 (17.1) 21 (19.4) 51 (21.3) 0.70 Surgical abdomen 22 (5.7) 1 (2.9) 3 (2.8) 18 (7.5) 0.09 FFS-defined manifestations 7 (1.8) 0 1 (0.9) 6 (2.5) 0.33 Eye involvement 25 (6.5) 2 (5.7) 9 (8.3) 14 (5.8) 0.39 Renal manifestations 83 (21.7) 15 (42.9) 29 (26.9) 39 (16.3) 0.02 Proteinuria 0.4 gm/24 hours 49 (12.8) 6 (17.1) 24 (22.2) 19 (7.9) 0.01 Creatinine 140 moles/liter (FFS item) 11 (4.3) 1 (4.3) 5 (6.2) 5 (3.3) 0.31 BVAS at diagnosis, mean SD FFS, no. (%) (76.0) 23 (65.7) 87 (80.6) 181 (75.4) (21.4) 11 (31.4) 18 (16.7) 53 (22.1) 2 10 (2.6) 1 (2.9) 3 (2.8) 6 (2.5) Revised FFS, no. (%) 0 98 (25.6) 10 (28.6) 35 (32.4) 53 (22.1) (31.3) 6 (17.1) 37 (34.3) 77 (32.1) (43.1) 19 (54.3) 36 (33.3) 110 (45.8) * EGPA eosinophilic granulomatosis with polyangiitis (Churg-Strauss); ENT ear, nose, and throat; CNS central nervous system; FFS Five-Factors Score; BVAS Birmingham Vasculitis Activity Score. For comparison of antineutrophil cytoplasmic antibody (ANCA) positive patients versus ANCA-negative patients. Serum creatinine levels at diagnosis were available for 254 patients (23 patients with unknown ANCA status, 81 ANCA-positive patients, and 150 ANCA-negative patients).

5 274 COMARMOND ET AL Table 2. Main biologic characteristics at diagnosis of the 383 patients with EGPA, according to ANCA status* Characteristic All (n 383) ANCA status unknown (n 35) ANCA-positive patients (n 108) ANCA-negative patients (n 240) Creatinine, moles/liter Creatinine, mg/dl Estimated MDRD GFR, ml/minute/1.73 m Estimated Cockcroft GFR, ml/minute/1.73 m Other laboratory findings CRP, mg/liter ESR 30 mm/first hour, no. (%) 189 (56.1) 19 (54.3) 60 (58.8) 110 (55.0) Eosinophils, /mm 3 7,569 6,428 8,139 6,150 7,833 6,462 7,329 6,643 Neutrophils, /mm 3 7,427 3,763 7,662 3,292 7,916 3,794 7,100 3,810 Lymphocytes, /mm 3 2,046 1,194 1,524 1,239 1, ,238 1,378# Platelets, /mm 3 361, , , , , , , ,561 * Except where indicated otherwise, values are the mean SD. EGPA eosinophilic granulomatosis with polyangiitis (Churg-Strauss); MDRD GFR Modification of Diet in Renal Disease glomerular filtration rate. Serum creatinine levels at diagnosis were available for 254 patients (23 patients with unknown antineutrophil cytoplasmic antibody [ANCA] status, 81 ANCA-positive patients, and 150 ANCA-negative patients). Data on the C-reactive protein (CPR) level were available for 160 patients (1 patient with unknown ANCA status, 97 ANCA-positive patients, and 62 ANCA-negative patients), and data on the erythrocyte sedimentation rate (ESR) were available for 337 patients (35 patients with unknown ANCA status, 102 ANCA-positive patients, and 200 ANCA-negative patients). P 0.04 versus ANCA-positive patients. Complete white blood cell and platelet counts were available for 254 patients (21 patients with unknown ANCA status, 87 ANCA-positive patients, and 146 ANCA-negative patients). # P 0.03 versus ANCA-positive patients. had associated cardiomyopathy. Alveolar hemorrhage was diagnosed in 16 (4.2%) of the patients, 6 of whom had both alveolar hemorrhage (massive in only 1) and renal manifestations. Heart, gastrointestinal tract, and kidney involvement were found in 27.4%, 23.2%, and 21.7% of the patients, respectively, but only 16.4% (cardiomyopathy), 1.8% (surgical complications), and 4.3% (creatinine level 140 moles/liter) of them met FFS definitions of severe manifestations. CNS manifestations were present in 20 patients (5.2%), including ischemic stroke in 8 (2.1%) and hemorrhagic stroke in 2 (0.5%). As shown in Table 2, at diagnosis, 50% of the patients had an inflammatory syndrome, and the mean SD blood eosinophil count was 7,569 6,428/ mm 3, with a maximum of 34,110/mm 3. Only 10 patients (2.6%), who were taking oral prednisone for their underlying asthma, had blood eosinophil counts of 500/mm 3 at diagnosis. Thirty-five patients (33 of whom were diagnosed during or prior to 1996), could not be tested for ANCA. Of the remaining 348 patients, 108 (31.0%) were ANCA positive; 72 of them had a perinuclear labeling pattern on IF and 68 had MPO specificity on ELISA. Eighteen of the 108 ANCA-positive patients had a cytoplasmic labeling pattern, 4 with proteinase 3 (PR3) specificity on ELISA. Three of the 108 ANCApositive patients had both perinuclear and cytoplasmic labeling patterns on IF, with 1 being both MPO-positive and PR3-positive on ELISA. The remaining 15 ANCApositive patients had atypical IF labeling patterns. None of the patients who were initially ANCA negative subsequently tested positive. Among the 180 patients who underwent 1 biopsy, 145 had histologic findings supporting a diagnosis of EGPA (vasculitis in 61.0%, eosinophil infiltrate in 61.3%, and granuloma in 33.3%). Among these 145 patients, ANCA results were available for 130, and 52 (40%) were ANCA positive (see Supplementary Table 1, available on the Arthritis & Rheumatism web site at , for more details on these patients). More precisely, 74 patients underwent nerve and muscle biopsies and 40 underwent muscle biopsies which supported an EGPA diagnosis in 45 patients, and 20 patients (biopsy sensitivity 61% and 50%), respectively. Nine patients had myocardial biopsies (sensitivity 89%), 13 had open lung biopsies (sensitivity 77%), 21 had transbronchial biopsies (sensitivity 52%), 23 had nasal or sinus mucosa biopsies (sensitivity 35%), and 85 had skin biopsies. The skin biopsies reached 88% sensitivity, but the main findings for the latter were nonspecific leukocytoclastic vasculitis, with eosinophilic infiltrates and granulomas, in only 36% and 10%, respectively, of those with positive skin biopsy findings. Fourteen patients had kidney biopsies, 12 of which showed pauci-immune

6 EOSINOPHILIC GRANULOMATOSIS WITH POLYANGIITIS (CHURG-STRAUSS) 275 necrotizing glomerulonephritis, and 2 of which showed segmental and focal glomerulonephritis. Patient characteristics at diagnosis, according to ANCA status, year of diagnosis, and sex. As shown in Table 1, ANCA-positive patients had significantly more frequent ENT manifestations, peripheral nerve involvement, and renal disease than ANCA-negative patients; renal disease remained relatively uncommon. Conversely, ANCA-negative patients had significantly more frequent cardiomyopathy. The mean BVAS and CRP level were significantly higher in ANCA-positive patients, but no other major biologic difference was found according to ANCA status (Table 2). Among the patients whose biopsy findings supported an EGPA diagnosis, histologic vasculitis features were more frequent in ANCA-positive patients (77.4% versus 48.8% in ANCA-negative patients, P 0.01), while granulomatous and eosinophilic infiltrates were seen in 42.1% and 63.2%, respectively, of the ANCA-positive patients versus 29.0% and 69.7%, respectively, of the ANCAnegative patients (P 0.32 and P 0.63, respectively). Patients diagnosed during or prior to 1996, as compared to those diagnosed after 1996, were younger at diagnosis (mean SD years versus years; P 0.04) and more frequently had constitutional symptoms, asthma (96.1% versus 88.6%; P 0.02), mononeuritis multiplex (68.0% versus 35.0%; P 0.01), cardiomyopathy (22.7% versus 13.3%; P 0.02), and kidney disease (31.3% versus 16.9%; P 0.01). Patients diagnosed during the two time periods had a comparable duration of asthma prior to EGPA diagnosis, FFS distribution, mean BVAS, and frequency of ANCA positivity among those with ANCA status available (28.4% versus 32.8%; P 0.51) (see Supplementary Tables 2 and 3, available on the Arthritis & Rheumatism web site at journal/ /(issn) ). Compared to male patients, female patients with EGPA were significantly younger at diagnosis, had a longer duration of asthma before EGPA diagnosis, and had less frequent renal abnormalities, although the frequencies of biopsy-proven glomerulonephritis were comparable for both sexes. Treatments. Remission-induction treatment included corticosteroids for all patients and a combination of corticosteroids and immunosuppressants for 214 patients (55.9%). Throughout their entire followup, 271 patients (70.8%) eventually received 1 immunosuppressant, including cyclophosphamide for 217 (56.7%), azathioprine for 98 (25.6%), methotrexate for 26 (6.8%), and rituximab for 3 (0.8%). Eighty-five of the 92 patients with an original FFS of 1 received 1 immunosuppressant, compared to 186 of the 291 patients with an FFS of 0. Fifty-four (50%) of the ANCA-positive patients received immunosuppressants as part of their induction therapy, and 76 (70.4%) received immunosuppressants during followup, which was comparable to the frequency of immunosuppressant use in the ANCA-negative group (56.3% during induction therapy and 70.4% during followup; P 0.28 and P 0.99, respectively). Disease outcomes. The mean SD duration of followup was months (median 50.5 months) and was comparable for ANCA-positive and ANCAnegative patients ( and months, respectively; P 0.88). The duration of followup exceeded 1 year from diagnosis for all of the patients except those who died early. Forty-five patients (11.7%) died, including 6 ANCA-positive patients, 30 ANCAnegative patients, and 9 patients with unknown ANCA status (crude mortality rate 5.6% versus 12.5% [P 0.05] and 25.7%, respectively). Deaths occurred a mean SD of months (median 21.4 months) after diagnosis and were attributed to cardiac events (myocardial infarction, cardiac insufficiency, or arrhythmia) in 14 patients (10 ANCA-negative patients), infections or cancers in 5 patients each, active vasculitis or respiratory events (severe asthma attacks and/or terminal chronic obstructive pulmonary disease) in 4 patients each, and miscellaneous causes in the remaining 13 patients (such as hemorrhage under anticoagulation in 2 patients or food allergy associated anaphylactic shock in 1 patient). Among the 14 fatal cardiac events, 8 occurred in patients with known EGPA-related cardiomyopathy, a median of 14 months after diagnosis (range months), and only 5 occurred within the first 6 months after diagnosis. Vasculitis relapse occurred in 97 patients (25.3%), including 38 ANCA-positive patients, 54 ANCAnegative patients, and 5 patients with unknown ANCA status (relapse rates 35.2% versus 22.5% [P 0.01] and 14.3%, respectively), while 72 additional patients (18.8%) experienced asthma flares, sinusitis, and/or increased eosinophilia, justifying adjustments to therapy (22 ANCA-positive patients and 46 ANCA-negative patients [20.4% versus 19.2%, respectively; P 0.79]). Among the 280 patients for whom data on prednisone use at their last visit was available, 44 (15.7%) were not taking prednisone (mean SD duration of corticosteroid use months), with no difference according to ANCA status. For the remaining 236 patients, prednisone had to be continued (started months earlier) with a mean daily dose of mg at the last visit (with 62 patients taking 5 mg/day, 94 taking 7.5 mg/day, and 62 taking 15

7 276 COMARMOND ET AL Figure 1. Kaplan-Meier survival curve (A) and relapse-free survival curve (B) for the 383 patients with eosinophilic granulomatosis with polyangiitis (Churg-Strauss), according to antineutrophil cytoplasmic antibody (ANCA) status at diagnosis. mg/day), which was comparable in the ANCA-positive and ANCA-negative subgroups. Five-year survival rates (Figure 1A) were 88.9% (95% CI ) for all patients, 94.9% (95% CI ) for the ANCA-positive patients, and 87.7% (95% CI ) for the ANCA-negative patients (P 0.09). At 10 years, the survival rates were 78.6% (95% CI ) for all patients, 88.8% (95% CI ) for the ANCA-positive patients, and 76.0% (95% CI ) for the ANCA-negative patients (P 0.02). Vasculitis relapse free survival rates (Figure 1B) at 5 years were 64.8% (95% CI ) for all patients, 58.1% (95% CI ) for the ANCApositive patients, and 67.8% (95% CI ) for the ANCA-negative patients (P 0.35). At 10 years, the relapse-free survival rates were 54.4% (95% CI ) for all patients, with no significant difference between ANCA subgroups (P 0.43). Predictors of relapse and death. Detailed results of univariable and multivariable analyses are shown in Table 3. When either the estimated Modification of Diet in Renal Disease glomerular filtration rate or serum creatinine level was entered into the model, the only independent predictors of death retained in the stepwise multivariable analysis were specific cardiomyopathy at diagnosis (HR 4.11 [95% CI ]), older age at diagnosis (HR 1.06 [95% CI ] for each additional year of age; HR 4.12 [95% CI ] for age 65 years), and EGPA diagnosed during or prior to 1996 (HR 3.20 [95% CI ]). The 5-year survival rate (Figures 2A and B) for patients with cardiomyopathy was 78.2% (95% CI ), compared to 91.6% (95% CI ) for those without (P 0.01), while it was 82.9% (95% CI ) for those diagnosed during or prior to 1996 versus 93.4% (95% CI ) for those diagnosed later (P 0.01). Both the original and revised FFS were predictive of death in univariable analyses (HR 3.84 [95% CI ] and 3.16 [95% CI, ], respectively, for FFS of 1 versus FFS of 0; P 0.01 for both). The only significant and independent predictor of vasculitis relapse was lower eosinophil count at diagnosis (available for 254 patients). The mean SD eosinophil count at diagnosis was 5,980 5,244/mm 3 in those who experienced a vasculitis relapse during followup versus 8,150 6,729/mm 3 in those who did not experience a relapse; P 0.02). The HR for eosinophil count 7,569/ mm 3 (the mean for all patients) was 1.83 (95% CI ). Five-year relapse-free survival rates were 72.8% (95% CI ) and 58.1% (95% CI ) in patients whose eosinophil counts at diagnosis were 7,569/mm 3 or 7,569/mm 3, respectively (Figure 2C). When eosinophil count was not entered into the model, ANCA positivity (HR 1.60 [95% CI ]; P 0.04) and cutaneous signs (HR 1.67 [95% CI ]; P 0.02) were the only significant independent predictors of relapse. When the 35 patients with unknown ANCA status were excluded, the same significant independent predictors of relapse or mortality were retained.

8 EOSINOPHILIC GRANULOMATOSIS WITH POLYANGIITIS (CHURG-STRAUSS) 277 Table 3. EGPA* Results of univariable and multivariable survival analyses for the risk of death or relapse in patients with Univariable analysis Multivariable analysis Parameter HR (95% CI) P HR (95% CI) P Predictors of death Age at diagnosis 1.05 ( ) ( ) 0.01 Sex 1.40 ( ) 0.28 Diagnosis in 1996 or before 2.40 ( ) ( ) 0.01 Inclusion in a therapeutic trial 0.89 ( ) 0.69 Asthma 2.61 ( ) ( ) 0.70 ENT manifestations 0.46 ( ) ( ) 0.15 Cutaneous manifestations 1.38 ( ) 0.30 Peripheral nerve involvement 1.09 ( ) 0.79 CNS involvement 2.61 ( ) ( ) 0.44 Cardiomyopathy (FFS item) 2.91 ( ) ( ) 0.01 Gastrointestinal involvement (FFS item) 0.96 ( ) 0.95 Serum creatinine, moles/liter 1.01 ( ) ( ) 0.05 MDRD GFR estimate, ml/minute/1.73 m ( ) ( ) 0.91 Eosinophil count, /mm ( ) 0.40 CRP, mg/liter 1.00 ( ) 0.37 ANCA positive 0.43 ( ) ( ) 0.15 Immunosuppressant as initial treatment 2.12 ( ) ( ) 0.36 Ever received an immunosuppressant 1.32 ( ) 0.77 Predictors of relapse Age at diagnosis 0.99 ( ) 0.32 Sex 1.41 ( ) ( ) 0.99 Diagnosis in 1996 or before 0.76 ( ) 0.23 Inclusion in a therapeutic trial 0.98 ( ) 0.92 Asthma 0.79 ( ) 0.49 ENT manifestations 1.04 ( ) 0.87 Cutaneous manifestations 1.59 ( ) ( ) 0.09 Peripheral nerve involvement 1.35 ( ) ( ) 0.27 CNS involvement 1.33 ( ) 0.53 Cardiomyopathy (FFS item) 1.16 ( ) 0.58 Gastrointestinal involvement (FFS item) 1.67 ( ) 0.47 Serum creatinine, moles/liter 1.00 ( ) 0.42 MDRD GFR estimate, ml/minute/1.73 m ( ) 0.41 Eosinophil count, /mm ( ) ( ) 0.01 CRP, mg/liter 1.00 ( ) 0.34 ANCA positive 1.67 ( ) ( ) 0.36 Immunosuppressant as initial treatment 0.86 ( ) 0.48 Ever received an immunosuppressant 1.51 ( ) ( ) 0.58 * EGPA eosinophilic granulomatosis with polyangiitis (Churg-Strauss); ENT ear, nose, and throat; CNS central nervous system; FFS Five-Factors Score; MDRD GFR Modification of Diet in Renal Disease glomerular filtration rate; CRP C-reactive protein. All parameters with a P value of 0.20 in the univariable analysis were included in the multivariable analysis. When year of diagnosis (during or prior to 1996 versus after 1996) was excluded from the model, the final parameters predictive of death were age at diagnosis (hazard ratio [HR] 1.06 [95% confidence interval (95% CI) ], P 0.01) and cardiomyopathy (HR 3.85 [95% CI ], P 0.01). When eosinophil count was excluded from the model, the final parameters predictive of relapse were antineutrophil cytoplasmic antibody (ANCA) positivity (HR 1.60 [95% CI ], P 0.04) and cutaneous signs (HR 1.67 [95% CI ], P 0.02). Cancers. During followup, 14 malignancies were diagnosed in 13 patients (2 ANCA-positive patients, 9 ANCA-negative patients, and 2 patients with unknown ANCA status; P 0.48), 5 of whom died. The cancer incidence did not differ significantly from that expected in the general population (SIR 1.15 [95% CI ]). There were 11 solid tumors (1 colon cancer, 2 basal cell skin carcinomas, 1 bladder cancer, 1 breast cancer, and the 5 fatal cancers [1 colon, 1 ovarian, 1 lung, and 1 gastric, and 1 glioblastoma in a man who also had prostate cancer]) and 3 hematologic neoplasms (1 B cell lymphoma, 1 polycythemia vera, and 1 myelodysplasia). Nine of these patients had received cyclophosphamide; 2 took immunosuppressants other than cyclophosphamide; and 2 had been treated with corticosteroids only.

9 278 COMARMOND ET AL Figure 2. A and B, Kaplan-Meier survival curves for the 383 patients with eosinophilic granulomatosis with polyangiitis (Churg-Strauss) (EGPA), according to the presence of cardiomyopathy at diagnosis (A) and the year of diagnosis (before or after December 31, 1996) (B). C, Relapse-free survival curve for the patients with EGPA, according to the eosinophil count at diagnosis. The 7,569/mm 3 threshold value corresponds to the mean eosinophil count at diagnosis in all 254 patients for whom data were available. DISCUSSION This analysis of a large cohort of EGPA patients with followup exceeding 5 years confirms the previously reported differences in clinical disease presentation based on ANCA status (16,17) and provides new information on outcomes and relapse predictors that may have therapeutic implications. Compared to ANCAnegative patients, ANCA-positive patients more frequently had clinical vasculitis manifestations, such as peripheral neuropathy or renal involvement, but less frequently had cardiomyopathy. Their mortality rate was lower, but the risk of relapse appeared to be slightly higher. Cardiomyopathy remained the main independent predictor of death in EGPA patients on multivariable analysis. The patients included in this study satisfied the ACR classification criteria (20) and/or the Chapel Hill nomenclature definition (1). No established diagnostic criteria for EGPA are currently available, and in the absence of clinical or histologic evidence of vasculitis, it may be difficult to differentiate between EGPA and other systemic eosinophilic disorders (34,21). In some ANCA-negative patients, an overlap between some forms of EGPA and nonclassified systemic eosinophilic disorders cannot be totally excluded (35,36). However, in practice, response to treatment is usually achieved better and faster in EGPA than in primary systemic eosinophilic disorders. Subsequent patient outcomes can also further help to confirm or, inversely, correct the initial diagnosis. A pathogenic role of ANCA has not been demonstrated in EGPA, and only a minority of EGPA patients are indeed ANCA positive. Hence, whether EGPA includes 2 phenotypic subsets based on ANCA status or whether ANCA positivity is only a surrogate marker of more diffuse vasculitis-induced damage remains to be determined (35,36). We found the same differences in clinical presentation based on ANCA status as previous investigators (16,17). Despite those clinical differences, outcomes did not differ according to ANCA status in those earlier studies (13,16,17); however, their followup durations and sample sizes were limited. Furthermore, a recurrent limitation of those previous EGPA investigations is the unclear, or at least inconsistent, definition of disease flares. Authentic vasculitis flares, like purpura or glomerulonephritis, should be analyzed separately from simple exacerbations of asthma, ENT symptoms, or increased eosinophil counts. In addition, asthma or rhinitis exacerbations can be underreported, depending on whether physicians consider them significant or not. Our results, based on more patients, with longer followup and a more stringent definition of relapse, suggest that ANCA positivity is associated with a higher risk of relapse, but a lower mortality rate (in univariable analysis). Recent results obtained in an Italian series of 38 patients were similar, with 46.7% of relapses occurring in ANCA-positive patients versus 8.7% in ANCA-negative patients (37). Eosinophilia at diagnosis did not differ according to ANCA status, but a lower eosinophil count was unexpectedly identified as the main predictive factor associated with relapse. Because eosinophilia fluctuates broadly and is corticosteroid sensitive and because the use of inhaled or systemic corticosteroids for asthma or allergic rhinitis/sinusitis at diagnosis was not systematically recorded, this finding will have to be further

10 EOSINOPHILIC GRANULOMATOSIS WITH POLYANGIITIS (CHURG-STRAUSS) 279 analyzed in other cohorts. When the eosinophil count was excluded from the model for relapse, ANCA positivity and cutaneous signs at diagnosis were the strongest predictors of subsequent relapses. ANCA status was also associated with death, with ANCA-positive patients having a lower mortality rate in univariable analysis. However, in multivariable analysis, cardiomyopathy remained the main poor-prognosis factor, as repeatedly reported over the past decades and included in the 1996 FFS (19). Other independent poor-prognosis parameters were older age at diagnosis, which was included in the revised FFS (23), and diagnosis during or prior to 1996, i.e., before the FFS was devised and used to optimize treatment and guide the need for cyclophosphamide in patients with more severe EGPA (19,28,29,38,39). Alternative or additional hypotheses for the better outcomes of those patients diagnosed after 1996 include advances in the management of cardiac insufficiency within the last decade and less frequent major organ involvement at diagnosis, although their FFS and BVAS did not differ from those of patients who were diagnosed earlier. In our study, most of the fatal cardiac events occurred a median of 14 months after EGPA diagnosis, underlining the potential late consequences of cardiac damage and the importance of thorough and serial cardiovascular assessments. In the Italian study by Sinico et al (16), 52% of the patients, mainly those who were ANCA positive, received an immunosuppressant as part of their initial therapeutic regimen. However, therapeutic regimens did not differ markedly according to ANCA status in our study. All of the patients in the present study received corticosteroids; half of them received both corticosteroids and an immunosuppressant at diagnosis, and two-thirds received both corticosteroids and an immunosuppressant at some point during the course of their disease. Importantly, most patients required long-term oral corticosteroids, with a mean daily prednisone dose of 12 mg, whereas 7.5 mg/day is the upper threshold defining remission in the European League Against Rheumatism recommendations for studies of systemic vasculitides (40). However, the majority of the patients were in clinical remission with regard to vasculitis manifestations and needed prednisone only to control asthma, rhinitis, or poorly specific arthralgias. While this study has several strengths, including its sample size, long followup duration, and the stringent definition of EGPA relapse, which allowed EGPA relapse to be distinguished from asthma exacerbations, it has some limitations. It was a partly retrospective multicenter study and included patients who participated in therapeutic trials at some time during the course of their vasculitis. We did not analyze in detail the impact of each treatment separately, because that would have yielded the same conclusions as those reported for trials in which the patients had participated (28,29,38). The sensitivity of ANCA testing methods may vary, and not all of our patients were tested centrally. In addition, though not observed in the present study, it has been reported that ANCA testing can be negative at diagnosis and become positive only later, during a flare (13). However, the ANCA frequencies and clinical differences at diagnosis according to ANCA status that were observed in the present study are consistent with those reported earlier based on other independent cohorts (16,17,37,41). Although asthma duration was similar in both ANCA status groups, we cannot exclude the possibility that a longer time to diagnosis may also have affected the survival of ANCA-negative patients, whose diagnosis can be more challenging. That patients diagnosed during or prior to 1996 had more frequent asthma and major organ involvement than those diagnosed subsequently might also suggest that physicians were then less likely to enroll patients with less overt EGPA manifestations in a study, especially before the routine use of ANCA testing. Finally, the definition of cardiomyopathy, for this study, as for the FFS, remained based on easily detectable clinical parameters (19). Abnormalities on echocardiography, magnetic resonance imaging, or autonomic nervous system evaluations are frequent in EGPA patients but of still unknown prognostic value (42 48). In summary, our analysis of a large group of well-defined EGPA patients with long followup durations revealed that global survival has improved over the last few decades and that, in addition to some clinical differences at diagnosis, outcomes also varied depending on ANCA status. Although ANCA-positive patients had a lower mortality rate, their risk of relapse appeared to be higher. If confirmed in other independent patient cohorts, those findings could suggest that treatment may have to be adapted according to ANCA status. However, the number of ANCA-positive EGPA patients remained relatively low, and other prognostic parameters appear to be more important, especially cardiomyopathy. Several issues in EGPA remain to be addressed, including the identification of more effective therapies so that the use of corticosteroids and conventional cytotoxic drugs can be limited or avoided. ACKNOWLEDGMENT The authors thank Ms Janet Jacobson for editorial assistance.

11 280 COMARMOND ET AL AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Drs. Comarmond, Pagnoux, and Guillevin had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design. Comarmond, Pagnoux, Cordier, Puéchal, Aumaître, Le Guenno, Aouba, Guillevin. Acquisition of data. Comarmond, Pagnoux, Khellaf, Cordier, Hamidou, Viallard, Maurier, Jouneau, Bienvenu, Puéchal, Aumaître, Le Guenno, Le Quellec, Cevallos, Fain, Godeau, Seror, Mahr, Cohen, Aouba, Mouthon, Guillevin. Analysis and interpretation of data. Comarmond, Pagnoux, Viallard, Puéchal, Aumaître, Le Guenno, Cevallos, Seror, Dunogué, Guilpain, Guillevin. REFERENCES 1. Jennette JC, Falk RJ, Andrassy K, Bacon PA, Churg J, Gross WL, et al. Nomenclature of systemic vasculitides: proposal of an international consensus conference. Arthritis Rheum 1994;37: Eustace JA, Nadasdy T, Choi M. Disease of the month: the Churg-Strauss syndrome. J Am Soc Nephrol 1999;10: Vaglio A, Casazza I, Grasselli C, Corradi D, Sinico RA, Buzio C. Churg-Strauss syndrome. Kidney Int 2009;76: Churg J, Strauss L. Allergic angiitis and periarteritis nodosa. Am J Pathol 1951;27: Pagnoux C, Guilpain P, Guillevin L. Churg-Strauss syndrome. Curr Opin Rheumatol 2007;19: Guillevin L, Cohen P, Gayraud M, Lhote F, Jarrousse B, Casassus P. Churg-Strauss syndrome: clinical study and long-term follow-up of 96 patients. Medicine (Baltimore) 1999;78: Abu-Shakra M, Smythe H, Lewtas J, Badley E, Weber D, Keystone E. Outcome of polyarteritis nodosa and Churg-Strauss syndrome: an analysis of twenty-five patients. Arthritis Rheum 1994;37: Chumbley LC, Harrison EG Jr, DeRemee RA. Allergic granulomatosis and angiitis (Churg-Strauss syndrome): report and analysis of 30 cases. Mayo Clin Proc 1977;52: Davis MD, Daoud MS, McEvoy MT, Su WP. Cutaneous manifestations of Churg-Strauss syndrome: a clinicopathologic correlation. J Am Acad Dermatol 1997;37: Della Rossa A, Baldini C, Tavoni A, Tognetti A, Neglia D, Sambuceti G, et al. Churg-Strauss syndrome: clinical and serological features of 19 patients from a single Italian centre. Rheumatology (Oxford) 2002;41: Gaskin G, Clutterbuck EJ, Pusey CD. Renal disease in the Churg-Strauss syndrome: diagnosis, management and outcome. Contrib Nephrol 1991;94: Haas C, Le Jeunne C, Choubrac P, Durand H, Hugues FC. Churg-Strauss syndrome: retrospective study of 20 cases. Bull Acad Natl Med 2001;185: In French. 13. Keogh KA, Specks U. Churg-Strauss syndrome: clinical presentation, antineutrophil cytoplasmic antibodies, and leukotriene receptor antagonists. Am J Med 2003;115: Reid AJ, Harrison BD, Watts RA, Watkin SW, McCann BG, Scott DG. Churg-Strauss syndrome in a district hospital. QJM 1998;91: Solans R, Bosch JA, Perez-Bocanegra C, Selva A, Huguet P, Alijotas J, et al. Churg-Strauss syndrome: outcome and long-term follow-up of 32 patients. Rheumatology (Oxford) 2001;40: Sinico RA, Di Toma L, Maggiore U, Bottero P, Radice A, Tosoni C, et al. Prevalence and clinical significance of antineutrophil cytoplasmic antibodies in Churg-Strauss syndrome. Arthritis Rheum 2005;52: Sable-Fourtassou R, Cohen P, Mahr A, Pagnoux C, Mouthon L, Jayne D, et al. Antineutrophil cytoplasmic antibodies and the Churg-Strauss syndrome. Ann Intern Med 2005;143: Guillevin L, Le Thi Huong D, Godeau P, Jais P, Wechsler B. Clinical findings and prognosis of polyarteritis nodosa and Churg- Strauss angiitis: a study in 165 patients. Br J Rheumatol 1988;27: Guillevin L, Lhote F, Gayraud M, Cohen P, Jarrousse B, Lortholary O, et al. Prognostic factors in polyarteritis nodosa and Churg-Strauss syndrome: a prospective study in 342 patients. Medicine (Baltimore) 1996;75: Masi AT, Hunder GG, Lie JT, Michel BA, Bloch DA, Arend WP, et al. The American College of Rheumatology 1990 criteria for the classification of Churg-Strauss syndrome (allergic granulomatosis and angiitis). Arthritis Rheum 1990;33: Roufosse F, Weller PF. Practical approach to the patient with hypereosinophilia. J Allergy Clin Immunol 2010;126: Luqmani RA, Bacon PA, Moots RJ, Janssen BA, Pall A, Emery P, et al. Birmingham Vasculitis Activity Score (BVAS) in systemic necrotizing vasculitis. QJM 1994;87: Guillevin L, Pagnoux C, Seror R, Mahr A, Mouthon L, Le Toumelin P, the French Vasculitis Study Group (FVSG). The Five-Factor Score revisited: assessment of prognoses of systemic necrotizing vasculitides based on the French Vasculitis Study Group (FVSG) cohort. Medicine (Baltimore) 2011;90: Guillevin L, Cevallos R, Durand-Gasselin B, Lhote F, Jarrousse B, Callard P. Treatment of glomerulonephritis in microscopic polyangiitis and Churg-Strauss syndrome. Indications of plasma exchanges: meta-analysis of 2 randomized studies on 140 patients, 32 with glomerulonephritis. Ann Med Interne (Paris) 1997;148: Guillevin L, Fain O, Lhote F, Jarrousse B, Le Thi Huong D, Bussel A, et al. Lack of superiority of steroids plus plasma exchange to steroids alone in the treatment of polyarteritis nodosa and Churg-Strauss syndrome: a prospective, randomized trial in 78 patients. Arthritis Rheum 1992;35: Guillevin L, Lhote F, Cohen P, Jarrousse B, Lortholary O, Genereau T, et al. Corticosteroids plus pulse cyclophosphamide and plasma exchanges versus corticosteroids plus pulse cyclophosphamide alone in the treatment of polyarteritis nodosa and Churg-Strauss syndrome patients with factors predicting poor prognosis: a prospective, randomized trial in sixty-two patients. Arthritis Rheum 1995;38: Guillevin L, Lhote F, Jarrousse B, Fain O. Treatment of polyarteritis nodosa and Churg-Strauss syndrome: a meta-analysis of 3 prospective controlled trials including 182 patients over 12 years. Ann Med Interne (Paris) 1992;143: Ribi C, Cohen P, Pagnoux C, Mahr A, Arene JP, Lauque D, et al, for the French Vasculitis Study Group. Treatment of Churg- Strauss syndrome without poor-prognosis factors: a multicenter, prospective, randomized, open-label study of seventy-two patients. Arthritis Rheum 2008;58: Cohen P, Pagnoux C, Mahr A, Arene JP, Mouthon L, Le Guern V, et al, for the French Vasculitis Study Group. Churg-Strauss syndrome with poor-prognosis factors: a prospective multicenter trial comparing glucocorticoids and six or twelve cyclophosphamide pulses in forty-eight patients. Arthritis Rheum 2007; 57: Kaplan E, Meier P. Non parametric estimation from incomplete observations. Am Stat Assoc 1958;87: Peto R, Pike MC. Conservatism of the approximation sigma (O-E)2-E in the logrank test for survival data or tumor incidence data. Biometrics 1973;29: Cox DR. Regression models and life-tables. J R Stat Soc Ser B 1972;34: Taeger D, Sun Y, Keil U, Straif K. A stand-alone Windows applications for computing exact person-years, standardized mortality ratios and confidence intervals in epidemiological studies. Epidemiology 2000;11:607 8.