Invited talks

I3 Using patient stratification to define genetics of disease

Abstract

Background Systemic lupus erythematosus (SLE) is a heterogeneous disease with unpredictable patterns of disease activity measured using mostly the SLEDAI. However, patients with similar SLEDAI scores may have different prognosis and molecular abnormalities. We reported the longitudinal stratification of SLE into 3 clusters based on correlation between gene expression and SLEDAI (1). Each cluster showed differences in molecular pathways involved, clinical manifestations, and how cell populations evolved with activity. In addition we asked ourselves if the genetic associations would differ with the new cluster stratification.

Methods For drug analysis we used two described sets of patients (1) selecting gene expression data of one visit/patient with active SLE (SLEDAI>5). Patient gene signatures were compared to drug derived gene signatures from CLUE database, giving a connectivity score. A negative score reflects inverse patterns between two signatures implying the drug may revert the disease-signature while a positive score would simulate disease. The magnitude of the score reflects the potential efficacy of the drug. Genetic data was performed in independent sets of individuals, focusing on the HLA.

Results Patient stratification based on drug connectivity scores revealed the same cluster structure described (correlation between neutrophil/lymphocyte ratio and SLEDAI dNLR p=1×10–7), implying that differential treatment depends on the cluster to which patients belong. Although drugs commonly used in SLE did not show the best scores, we found different values for each cluster suggesting that expression of target genes may provide insight in the prioritization of compounds.

We next constructed a model to classify patients using cluster information to inform on drug use and predict nephritis applied to 3 new longitudinal cohorts. A meta-analysis showed a significantly higher incidence of nephritis in patients classified to a neutrophil-driven cluster (2). In addition we observed differences in the genetic associations to disease in the HLA region depending on the clusters.

Conclusions Drug patterns reverting disease gene expression follow the cell-specificity of the disease clusters and provide a clinically useful model to treatment selection and nephritis. Clustering, at least in one case is also guided by the genetic contribution to disease.

Article metrics
Altmetric data not available for this article.
Dimensionsopen-url