Identifying distinct clinical phenotypes and outcomes in adult-onset IgA vasculitis using unsupervised clustering analysis.

Abstract

IgA vasculitis (IgAV) is a small-vessel vasculitis. The diversity of clinical presentations and outcomes requires the identification of groups with a distinct phenotype for management and follow-up. The French retrospective IGAVAS database (n = 260) includes adult-onset IgAV. Agglomerative hierarchical clustering was performed, using k-means and Ward's method, based on 7 parameters (sex, age, constitutional symptoms, skin necrosis, joint and digestive involvement, abnormal kidney function). These results were validated on an independent cohort of Slovenian patients (n = 208). Of 260 IgAv patients, 3 identified clusters were identified. Cluster 1 (n = 114, 44%) included a majority of men (68%), of young age (median 42 years), with gastrointestinal (100%) and joint (77%) involvement, with low rate of renal insufficiency (7%). Cluster 2 (n = 96, 37%) included 56% males, with a mean age of 49 years, without gastro-intestinal involvement, and 4% renal insufficiency. Cluster 3 (n = 50, 19%, 66% men) included older patients (mean age 70 years), with skin necrosis (52%), and abnormal kidney function (90%). Very similar clusters were found in the Slovenian cohort. Of the 144 renal biopsies available, cluster 3 had more tubulointerstitial damage, there was no significative difference of extracapillary proliferation. Regarding outcome, cluster 1 an 2 tended to have more relapses (20 and 25%), cluster 3 had more residual chronic renal failure (79%). Through this clustering study, validated in an independent cohort, three very distinct groups in adult-onset IgAV emerged with clear-cut presentations and outcomes. This approach enables a personalized evaluation to be proposed at baseline and during the follow-up.