P186 Beyond the snapshot: leveraging longitudinal c-reactive protein trends to predict number of biologic use in axial spondyloarthritis

Le résumé fourni par la source

Abstract Background/Aims Axial Spondyloarthritis (axSpA) is characterised by chronic inflammation of the axial skeleton. C-reactive protein (CRP), as a marker of inflammation, can be a predictor of response to targeted therapies. We aim to explore whether CRP fluctuations prior to biologic drug initiation could predict the number of targeted therapies required. Methods Electronic medical records of axSpA patients (n = 244) attending the Royal National Hospital for Rheumatic Diseases and starting targeted therapies (Anti-TNF, JAK inhibitor and IL-17 inhibitor) between 2016 and 2018 were retrospectively reviewed. Patients required a minimum of three CRP results over five years pre- and post- biologic initiation to be included on the analysis. We applied clustering methods to create five clusters of CRP measurements. These cluster models were then compared in terms of their prediction accuracy to find the best prediction model. Results Table-1 shows the clusters and the mean value of up to five CRP measurements. Different clusters represent different trends of CRP. The R-Squared value of the prediction model is 68%. This indicates that the CRP trends can be used to predict the number of biologic drugs patients will likely use. The analyses demonstrate that multiple CRP measurements add value to the prediction of response to initial targeted therapies. Table-1 also shows the number of drugs used by percentage of patients in the corresponding cluster. Patients in clusters 1 and 2 used three biologics, while cluster 3 patients are more likely to be responsive to first biologic. Patients in cluster 4 and 5 used two biologics and did not require to switch to a third biologic drug. Conclusion Our study shows the benefit of assessing multiple CRP measurements before biologic drug administration to predict how many times patients would likely switch biologic treatment. Based on our current analyses, we will further investigate if specific classes of targeted therapies can be used first line, stratified by the CRP cluster model to optimise drug survival. Disclosure M. Ho: None. A. Aziz: None. E.G. Ozpolat: None. C. Vasilakis: None. E.R. Gates: None. R. Sengupta: Consultancies; Pfizer, Abbvie, Biogen, BMS, Chugai, Lilly, Novartis, UCB. Honoraria; Pfizer, Abbvie, Lilly, Novartis, UCB. Grants/research support; UCB and Novartis - paid to institution. Other; support for attending meetings - Abbvie, Lilly, UCB, Novartis.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.