LB1630 Modular analysis of the psoriasis transcriptome identifies biomarkers of clinical response to distinct classes of biologics and elucidates drug and disease endotypes across skin and blood

Biologics have transformed psoriasis treatment and improved outcomes for many patients. However, clinical response to these drugs is heterogeneous and reproducible predictive biomarkers are lacking. We aimed to identify gene expression changes induced by treatment with adalimumab (ADA; TNF inhibitor) and ustekinumab (UST; IL-12/23 inhibitor) and expression signatures associated with clinical response to these drugs. Bulk mRNA-Seq was performed on lesional (LS) and non-lesional (NL) skin biopsies and blood derived from 83 psoriasis patients (41 for each drug in skin; 40 ADA and 43 UST in blood) at baseline, week 1 and week 12 (718 total samples). Clinical response was defined as percentage reduction in psoriasis area and severity index (PASI) at week 12 compared to baseline. Weighted gene correlation network analysis (WGCNA) was applied to the skin and blood samples separately, identifying 34 co-expressed gene modules in skin and 26 in blood. The first principal component of each module was correlated with demographics, PASI and clinical response. In LS skin, an oxidative phosphorylation module exhibited ADA-specific downregulation at week 1 and a cell cycle module exhibited UST-specific downregulation at week 1. The eigengenes of 9 modules at baseline in LS skin were significantly (FDR <= 0.05) associated with clinical response to ADA; there were 6 such modules associated with response to UST. Importantly, these treatment and response-associated expression signatures were replicated in an independent validation cohort. Module-PASI associations were widespread in LS skin; strikingly, 17 modules in NL skin and 19 modules in blood were associated with PASI. In summary, our findings identify novel pathways modulated by biologic treatment, identify the first replicable transcriptomic biomarkers of response in distinct tissue compartments, and highlight the effect of PASI on the NL skin and blood transcriptomes.