Novel DNA methylome biomarkers associated with adalimumab response in rheumatoid arthritis patients.
DNA methylation
adalimumab
machine learning
rheumatoid arthritis
therapy response
Journal
Frontiers in immunology
ISSN: 1664-3224
Titre abrégé: Front Immunol
Pays: Switzerland
ID NLM: 101560960
Informations de publication
Date de publication:
2023
2023
Historique:
received:
27
09
2023
accepted:
04
12
2023
medline:
8
1
2024
pubmed:
8
1
2024
entrez:
8
1
2024
Statut:
epublish
Résumé
Rheumatoid arthritis (RA) patients are currently treated with biological agents mostly aimed at cytokine blockade, such as tumor necrosis factor-alpha (TNFα). Currently, there are no biomarkers to predict therapy response to these agents. Here, we aimed to predict response to adalimumab (ADA) treatment in RA patients using DNA methylation in peripheral blood (PBL). DNA methylation profiling on whole peripheral blood from 92 RA patients before the start of ADA treatment was determined using Illumina HumanMethylationEPIC BeadChip array. After 6 months, treatment response was assessed according to the European Alliance of Associations for Rheumatology (EULAR) criteria for disease activity. Patients were classified as responders (Disease Activity Score in 28 Joints (DAS28) < 3.2 or decrease of 1.2 points) or as non-responders (DAS28 > 5.1 or decrease of less than 0.6 points). Machine learning models were built through stability-selected gradient boosting to predict response prior to ADA treatment with predictor DNA methylation markers. Of the 94 RA patients, we classified 49 and 43 patients as responders and non-responders, respectively. We were capable of differentiating responders from non-responders with a high performance (area under the curve (AUC) 0.76) using a panel of 27 CpGs. These classifier CpGs are annotated to genes involved in immunological and pathophysiological pathways related to RA such as T-cell signaling, B-cell pathology, and angiogenesis. Our findings indicate that the DNA methylome of PBL provides discriminative capabilities in discerning responders and non-responders to ADA treatment and may therefore serve as a tool for therapy prediction.
Sections du résumé
Background and aims
UNASSIGNED
Rheumatoid arthritis (RA) patients are currently treated with biological agents mostly aimed at cytokine blockade, such as tumor necrosis factor-alpha (TNFα). Currently, there are no biomarkers to predict therapy response to these agents. Here, we aimed to predict response to adalimumab (ADA) treatment in RA patients using DNA methylation in peripheral blood (PBL).
Methods
UNASSIGNED
DNA methylation profiling on whole peripheral blood from 92 RA patients before the start of ADA treatment was determined using Illumina HumanMethylationEPIC BeadChip array. After 6 months, treatment response was assessed according to the European Alliance of Associations for Rheumatology (EULAR) criteria for disease activity. Patients were classified as responders (Disease Activity Score in 28 Joints (DAS28) < 3.2 or decrease of 1.2 points) or as non-responders (DAS28 > 5.1 or decrease of less than 0.6 points). Machine learning models were built through stability-selected gradient boosting to predict response prior to ADA treatment with predictor DNA methylation markers.
Results
UNASSIGNED
Of the 94 RA patients, we classified 49 and 43 patients as responders and non-responders, respectively. We were capable of differentiating responders from non-responders with a high performance (area under the curve (AUC) 0.76) using a panel of 27 CpGs. These classifier CpGs are annotated to genes involved in immunological and pathophysiological pathways related to RA such as T-cell signaling, B-cell pathology, and angiogenesis.
Conclusion
UNASSIGNED
Our findings indicate that the DNA methylome of PBL provides discriminative capabilities in discerning responders and non-responders to ADA treatment and may therefore serve as a tool for therapy prediction.
Identifiants
pubmed: 38187379
doi: 10.3389/fimmu.2023.1303231
pmc: PMC10771853
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
1303231Informations de copyright
Copyright © 2023 Hageman, Mol, Atiqi, Joustra, Sengul, Henneman, Visman, Hakvoort, Nurmohamed, Wolbink, Levin, Li Yim, D’Haens and de Jonge.
Déclaration de conflit d'intérêts
Author EL was employed by the company Horaizon BV. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.