Background/Purpose: Because only 25–40% of the patients are ANCA+ and biopsy is not always taken or helpful, EGPA diagnosis can be challenging. Previous cohort analyses identified several covariates associated with poor outcomes and suggested differences between ANCA+ and ANCA– patients. We reanalyzed the FVSG cohort using different definitions of EGPA and cluster analysis to try to identify other disease subsets.

Methods: FVSG-cohort patients were grouped using the following 4 EGPA definitions: 1) all 383 patients whose EGPA diagnoses met ACR criteria or 1994 Chapel Hill nomenclature according to referral physicians; 2) only the 59 patients with biopsy-proven vasculitides; 3) the 138 with biopsy-proven vasculitis and/or ANCA+; 4) the 267 patients with biopsy-proven vasculitis and/or ANCA+ and/or surrogate manifestations (purpura, gangrene, alveolar hemorrhage, mononeuritis multiplex (MNM), renal disease (glomerulonephritis and/or hematuria)). Exploratory hierarchical cluster analyses were performed on the entire cohort and each subgroup to determine the number of clusters. The parameters, all binary, included in these cluster analysis were: skin manifestations, sinusitis, cardiomyopathy (CM), renal disease, MNM. In subsequent steps, we also included ANCA test results (missing for 35 patients excluded from these cluster analyses). Then K-means nonhierarchical cluster analysis examined the characteristics of the clusters.

Results: Table 1 shows characteristics of the different patient subgroups.

	Definition 2	Definition 3	Definition 4	All others/Undefined
n	59/383	138/383	267/383	116/383
Sex, (M/F, n)	30/29	74/64	141/126	58/58
Age at diagnosis (yrs)	51	51	51	49
Asthma	93%	94%	96%	80%
ENT manifestations	53%	54%	54%	34%
Lung infiltrate	37%	39%	42%	30%
Alveolar hemorrhage	5%	6%	6%	–
Skin manifestations	39%	41%	49%	16%
Purpura	31%	28%	32%	–
Mononeuritis multiplex	51%	53%	66%	–
CNS	5%	6%	6%	3%
Cardiomyopathy	9%	9%	19%	11%
Severe GI signs	7%	4%	6%	6%
Eye	7%	9%	8%	3%
Renal disease	25%	26%	28%	6%
BVAS at diagnosis	19.4 ± 6.8	21.10 ± 7.6	20.6 ± 7.9	12.0 ± 6.8
ANCA+ (ANCA status missing, n)	55% (6)	82% (6)	45% (26)	– (9)
Vasculitis relapse	36%	34%	30%	14%
Deaths	12%	7%	14%	6%

Cluster analysis of the entire cohort, not including ANCA, yielded 3 clusters:

• n=128, 89% with definite EGPA (definition 4), 29% ANCA+, 24% with renal disease, 66% with MNM, 27% with CM;
• n=160, 78% met definition 4, 38% ANCA+, 57% with MNM, 18% with CM, 16% with renal disease;
• n=95, only 29% met definition 4, 22% ANCA+, none with skin, renal, CM or MNM, meaning that 71% had undefined disease (others).

Without including ANCA and whichever the EGPA definition used, clusters differed mainly by their CM, MNM or skin-sign distributions. Including ANCA strongly influenced clustering results. Subgroup cluster analysis on the 241 patients with known ANCA status meeting definition 4, yielded 3 clusters:

• n=61, all ANCA+, with 41% ENT signs, 48% MNM, 25% renal disease, 7% CM, but no skin lesions, 31% vasculitis-relapse rate, 5% mortality;
• n=54, all ANCA–, with 89% MNM, 48% ENT signs, 22% CM, 22% renal disease, but no skin disease, 21% relapse rate, 13% mortality;
• n=126 with 37% ANCA+, 94% skin manifestations, 65% ENT signs, 60% MNM, 22% CM, 19% renal disease, 37% relapse rate, 16% mortality.

Conclusion: The results of this study suggest the need to establish rigorous EGPA definitions. Cluster analysis supported and strengthened findings from previous analyses, which showed the main subsets were identified by ANCA status.

---

« Back to 2013 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/eosinophilic-granulomatosis-with-polyangiitis-churg-strauss-syndrome-re-analysis-of-the-french-vasculitis-study-group-cohort-using-different-disease-definitions-and-cluster-analysis/