THU0035 DOMINANCE OF DRUG TARGET MECHANISMS IN RHEUMATOID ARTHRITIS

Background Biological and Targeted Synthetic Disease-Modifying Anti-Rheumatic Drugs (b- and tsDMARDs) have been developed over the years for patients with rheumatoid arthritis (RA). They can be grouped into families of drugs according to their mechanisms of action. Here we focus specifically on anti-TNFs, anti-IL6s, anti-IL1s, T or B Cells inhibitors and JAK inhibitors. There is still a significant proportion of patients inadequately responding to RA treatments. Lacking predictive biomarkers of response or personalised medicine approaches to guide use of targeted therapies in RA patients, EULAR and ACR guidelines [1,2] both recommend to loop over available ts- and bDMARDs as long as the patient’s response is considered as inadequate. Meanwhile, for historical reasons and despite the similar clinical efficacy of these drugs, anti-TNFs have a large dominance in medical practice. Objectives Determine in what proportions the different mechanisms targeted by existing DMARDs families are dominant in RA patients synovial tissue. Methods Retrospective analysis of 7 private or public datasets of 300 rheumatoid arthritis patients, consisting in all cases of transcriptomic data from synovial biopsies. Three datasets come from the RheumaKit platform [3,4] (low-density microarrays or qPCR). Four other datasets are publicly available on Gene Expression Omnibus (GSE89408, GSE45867, GSE97165, GSE21537, produced on Illumina, KTH or Affymetrix platforms). Each mechanism of action is associated to a « drug target complex » (DTC), consisting of the genes coding for proteins directly targeted by the DMARDs of interest. For a given dataset, computations are made in four steps: 1) each patient is described by its different DTC values (an average of the expression values of the members of each DTC) 2) each patient is scored for each of its DTC value. Each of these DTCs are scored as a percentile of the corresponding DTC distribution over the full dataset from which a given patient’s data is extracted. This gives, for a single patient, as many percentiles as defined DTCs. 3) for each individual patient, a ranking of DTC values is then operated. This allows to overcome the fact that absolute numerical values of two different DTCs should not be compared. 4) A summary statistic is then computed over each dataset separately, to conclude which DTC is dominant in which proportion of the cases among this dataset. Results This analysis exhibits different dominance patterns for RA-related mechanisms of action in individual patients. Statistically, no unique mechanism is shown dominant in a majority of patients; On the four public datasets, where all DTCs of interest are available, averaged dominance proportions across datasets are : IL1∼15%, IL6∼20%, TNF∼11%, B Cells∼20%, JAK∼17%, T Cells∼7%. About 10% of the patients exhibit equivalent dominance patterns between multiple DTCs. On all seven datasets, this analysis also outputs weak or moderate correlations between the dominance levels of multiple DTCs. Conclusion These results highlight large variability in metabolic patterns underlying RA; such observation is consistent with the similar efficacy observed for b or tsDMARDs when evaluated in clinical trials. Herein we show that TNF-dependent pathways is dominant in only a relatively small proportion of RA patients, whereas non-TNF-dependent pathways (e.g. IL-6 or JAK-dependent pathways) are more dominant. This work pleads for a more balanced use of available treatments. In addition, these results support further investigations towards precision medicine-oriented approaches, namely based on biomarkers from synovium, in the treatment of RA. Data analyses included in this work have been financially supported by Sanofi Genzyme. References [1] Smolen J.S. et al., Ann. Rheum. Dis. 2017 [2] Singh J.A. et al., Arthritis Care & Res. 2015 [3] www.rheumakit.com : for founding work & detailed description, see Lauwerys B. et al., PloS ONE 2015 [4] Helleputte Th. et al., Ann. Rheum. Dis. 2016. Disclosure of Interests Jerome Paul Grant/research support from: The data analysis included in this work has been financially supported by Sanofi Genzyme., Pierre Gramme Shareholder of: Shareholder of DNAlytics., Grant/research support from: The data analysis included in this work has been financially supported by Sanofi Genzyme., Thibault Helleputte Shareholder of: Founder of DNAlytics., Grant/research support from: The data analysis included in this work has been financially supported by Sanofi Genzyme.