Cytokine profiles of blood T lymphocytes in rheumatoid arthritis identify different subsets associated with responsiveness to biologic therapies

J Bystrom1, T Taher1, M Albogami1, S Alzabin3, S Kelly1, P Mangat2, RO Williams3, AS Jawad1, RA Mageed1. 1WHRI, Barts and the London, QMUL, London, UK, 2Department of Rheumatology, Royal Free Hospital, London, UK, 3Kennedy Institute of Rheumatology, Oxford University, Oxford, UK

Background

Biologic therapies have revolutionized treatment of patients with chronic inflammatory disease and cancer. However significant proportions of such patients do not respond to these agents. This hampers wider use of targeted therapies and personalising healthcare. Our studies indicate that non-responsiveness to biologic agents is due to different disease pathways causing pathology in patient subsets with the same disease.

Aims

To characterize blood T-lymphocyte subsets and their signature cytokines and to determine whether could predict the response of rheumatoid arthritis (RA) patients to biologic anti-TNFα agents.

Methods

Blood from 70 RA patients prior to and after 4 and 12 weeks of treatment were obtained and level of cytokines in plasma and produced by stimulated T-lymphocytes and monocytes in vitro quantified. Response to treatment was determined after 12 weeks of treatment by reduction of ≥1.2 DAS28 units. Highly enriched T-lymphocytes were stimulated with anti-CD3/CD28 antibodies and monocytes with lipopolysaccharide (LPS). Cytokines were quantified using a multiplex platform (MSD Technologies). Surface markers and intracellular cytokines levels in T-lymphocyte subsets were determined by FACS for measuring the frequency of T-cell subsets in blood. Statistical analysis was performed using Mann Whitney U test. Ethical approval was granted by the local committee of the Health Research Authority and all recruited patients gave informed consent to take part in the study.

Results

Prior to treatment, T-lymphocytes from responder patients produced 8.9 folds higher levels of TNFα and 1.9 folds GM-CSF compared with T-lymphocytes from non-responder patients (p=0.02). Intracellular staining revealed that both cytokines were co-produced by the same effector memory (CD45RO+) T-lymphocytes in responder RA patients. Unlike Th17 lymphocytes the TNFα/GM-CSF-producing T lymphocytes expressed no or low level of CD161. The frequency of these T-lymphocytes was not associated with age or gender. There was significant correlation between the level of GM-CSF produced by ex vivo T-lymphocytes and its plasma levels and both were significantly higher in responder compared with non-responder patients before treatment (plasma: 1.8 folds higher, p=0.04). The odds of RA patients responding to treatment with biologic anti-TNFα agents were 11.4 folds higher when their T-lymphocytes produced >1.5pg GM-CSF/1000 cells in culture while a plasma level of >3.5pg/mL increased the odds to 14.0 folds (1.535 to 127.7, p=0.0095).

Conclusion

A subset of effector memory T-lymphocytes that produces TNFα and GM-CSF is present at higher frequencies in the blood of RA patients that respond to treatment with biologic anti-TNFα agents before treatment than in non-responders. This is in contrast to non-responder patients who have higher frequencies of Th17 lymphocytes. Our data suggests that RA patients can be stratified into responder and non-responder patients to treatment with biologic anti-TNFα based on the level of blood T-lymphocyte subsets. Further, measurement of GM-CSF and IL-17 levels in plasma and culture supernatants of activated T-lymphocytes can be useful biomarkers for identifying responder/non-responder patients prior to treatment.