Risk prediction in rheumatoid arthritis

Student thesis: Doctoral ThesisDoctor of Philosophy

Abstract

As rheumatoid arthritis (RA) is a heterogeneous disease whose course and treatment response varies between patients, a stratified approach to its management is required. This thesis aimed to facilitate the risk prediction that underpins stratified medicine in RA. Its primary aim was to improve the knowledge of which clinical and genetic factors predict RA’s onset, disease course and treatment responses. Its secondary aim was to develop a prediction modelling framework that harnessed these factors to inform clinical care. There were five key findings.

Firstly, it demonstrated a significant inverse association between alcohol consumption and RA development when the evidence across published studies was pooled using meta-analytical techniques. This suggests alcohol may protect against RA. Secondly, it demonstrated that only HLA RA susceptibility variants associated with radiological progression in a clinical trial cohort of early, active RA patients. This suggests the non-HLA genetic architectures of RA susceptibility and severity may, at least partially, differ. Thirdly, it provided evidence that anti-citrullinated protein antibodies (ACPA) can identify patients with early, active RA that are most likely to benefit from combination treatments. Fourthly, it demonstrated that estimating an asymptomatic individual’s risk of RA is possible, through developing and validating a risk prediction model that uses computer simulation to improve upon the discriminative abilities of existing RA prediction models. Finally, it highlighted the importance of considering RA’s heterogeneity when assessing its predictive factors; alcohol’s likely protective effect was predominantly seen in ACPA-positive disease and genetic and environmental factors had different impacts on the risk of developing younger and older onset RA.

In conclusion this thesis has contributed to stratified medicine in RA by better characterising which predictive factors are relevant to its development, severity, treatment needs and responses and developing a risk prediction modelling framework that may be applicable to many aspects of stratified care.

Date of Award2014
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorCathryn Lewis (Supervisor) & Andrew Cope (Supervisor)

Cite this

'