Background/Purpose: In axial spondyloarthritis (axSpA), the clinical benefits of TNF inhibition (TNFi) are well documented although, by design, most studies report average benefits in groups of patients.  Inevitably, a subset of patients will not respond to therapy.  Identifying characteristics of these patients is important as it may inform the use of TNFi in clinical practice.  The aim of the current study was to identify factors that predict satisfactory treatment response, in a nationwide register of patients commencing TNFi.

Methods: The British Society for Rheumatology Biologics Register for Ankylosing Spondylitis (BSRBR-AS) recruited biologics-naïve patients with axSpA from 83 secondary care rheumatology centres across Great Britain between 2012 and 2017.  Clinical data was collected from patients’ medical records, and additional data was collected via postal questionnaires.  Treatment response was determined at the first eligible follow-up and was defined as moving from high/very high Ankylosing Spondylitis Disease Activity Scale score (ASDAS≥2.1) to moderate/inactive disease activity (ASDAS< 2.1). Factors associated with treatment response were assessed using logistic regression.  Thereafter, a forward stepwise logistic regression model was used to identify which group of factors best predicted treatment response.  Analysis was conducted on the June 2017 BSRBR-AS dataset.

Results: 249 participants were eligible for the current analysis; 69% were male, with median age 47yrs (inter-quartile range: 36-56).  96% met the ASAS imaging criteria, of which 67% had ankylosing spondylitis.

Median follow-up was 14wks, at which point 35% were classified as treatment responders.  For every 1 unit increase in disease activity (BASDAI) there was a 29% decrease in the odds of response (odds ratio 0.71; 95%CI 0.60-0.85).  A similar effect was seen with increasingly poor function (BASFI: 0.70; 0.61-0.81).  Other factors associated with response on univariable analysis were wide-ranging, including clinical, socioeconomic and patient-reported factors, see Table 1.

Only four independent predictors of response were identified.  Patients in full-time employment and with high education were more likely to respond, as were those with better mental health scores.  Increasing comorbidities was associated with poor response.  The final model demonstrated a good level of fit with positive (PPV) and negative predictive values (NPV) of 63% and 77% respectively.

Conclusion: Four variables, none of them disease specific, identified axSpA patients commenced on biologic therapy, who were unlikely to have responded, four months later.  Other patients may need additional therapeutic approaches and additional support (e.g. support so that they do not lose their job) to achieve optimal outcomes.

ACR table

« Back to 2019 ACR/ARP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/predicting-response-to-biologic-therapy-in-patients-with-axial-spondyloarthritis-axspa/