POS0234 MACHINE LEARNING AND PROTEOMICS FOR PREDICTING ANTI-TNFα RESPONSE IN PSORIATIC ARTHRITIS: IDENTIFICATION OF DRUG MODULATED PROTEINS

Abstract

Background:

Psoriatic arthritis (PsA) presents challenges in treatment due to its clinical heterogeneity and variable patient responses. Reliable biomarkers of therapeutic response are essential for optimizing and personalizing management strategies. Proteomics offers a powerful tool to investigate the molecular mechanisms of PsA and identify treatment-modulated proteins.

Objectives:

1) To identify potential proteomic biomarkers of anti-TNFa response in peripheral blood mononuclear cells (PBMCs) of PsA patients using an advanced proteomic technique and machine learning, and 2) To explore the modulation of specific proteins by anti-TNF-α therapy after six months of treatment.

Methods:

This study was conducted with 71 PsA patients, categorized into responders and non-responders based on DAPSA reduction, with responders defined as those exhibiting a decrease of more than 50% in their DAPSA score after 6 months of anti-TNFαtreatment. A total of 384 proteins were analyzed in PBMCs using the Olink platform (384 Explore inflammation). A machine learning algorithm was applied to identify potential biomarkers of treatment response. Additionally, a longitudinal study was conducted on 20 PsA patients treated with anti-TNF-α therapy for six months. In this study, the levels of 384 proteins were measured using the Olink platform both at baseline and 6 months after the initiation of treatment, to assess the impact of therapy on protein expression levels.

Results:

Firtsly, a proteomic profile consisting of 8 proteins with significant differences between responders and non-responders was identified. Using machine learning algorithms, a model combining two of these proteins was developed, demonstrating strong discriminatory ability for non responders patients, achieving an AUC of 0.80 and an accuracy of 0.92. Secondly, 65 proteins were identified as differentially expressed after 6 months of anti-TNF-α treatment, with 57 proteins upregulated and 8 downregulated. Enrichment pathways analysis showed that upregulated proteins were enriched in B-cell-related pathways, while the downregulated proteins were associated with neutrophil pathways. The protein with the most significant changes at six months was CD200, which showed increased expression over this period. CD200 is primarily expressed by B cells and plays a role in neutrophil regulation, aligning with the observed results. In this context, we analyzed the modulation of lymphocyte and neutrophil levels with therapy. After six months of treatment, a significant increase in lymphocyte counts and a corresponding decrease in neutrophil counts were observed. Notably, when patients were stratified into anti-TNFα responders and non-responders, CD200 levels increased exclusively in responders.

Conclusion:

Our study shows the potential of proteomics and computational tools to identify biomarkers of anti-TNF-α response in PsA patients. We identified a machine learning model combining two inflammatory proteins showing excellent discriminatory ability. Additionally, significant changes in protein expression after six months of anti-TNF-α therapy were observed, with upregulated proteins linked to B-cell pathways and downregulated proteins associated with neutrophil pathways. These findings suggest that specific proteins modulated by treatment could serve as valuable biomarkers for predicting therapeutic outcomes and guiding personalized treatment strategies in PsA.

REFERENCES:

NIL.

Acknowledgements:

Project "PI22/00539", funded by Instituto de Salud Carlos III (ISCIII) and co-funded by the European Union, and project "PI-0243-2022" funded by the "Junta de Andalucia/Consejeria de Salud y Consumo".

Disclosure of Interests:

None declared. © The Authors 2025. This abstract is an open access article published in Annals of Rheumatic Diseases under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Neither EULAR nor the publisher make any representation as to the accuracy of the content. The authors are solely responsible for the content in their abstract including accuracy of the facts, statements, results, conclusion, citing resources etc.