AN IMMEDIATE DECREASE IN SERUM-VEGF-LEVELS HELPS IN PREDICTING A MAJOR RESPONSE IN PATIENTS WITH ANKYLOSING SPONDYLITIS WHO WERE STARTED ON INFLIXIMAB: RESULTS FROM THE ASSERT BIOMARKER STUDY

Background: One of the major challenges in biomarker research is to find out which biomarkers can predict efficacy of expensive biological treatments in individual patients. We have shown previously that markers such as VEGF are responsive to infliximab treatment, but a formal evaluation of their clinical usefulness as predictors of response was lacking. Objectives: To investigate the potential of a panel of biomarkers in predicting a major? clinical response in AS patients under infliximab (INF) treatment. Methods: A randomly selected 80% sample of the ASSERT database was used. ASSERT was a 24-week RCT comparing INF (5mg/kg/6 weeks) with placebo, with a long term extension (LTE) on INF. In parallel, an extensive biomarker profile was collected in all participating patients at baseline and at week 2, 24, 48 and 52 of the RCT or LTE. The profile included serum levels of IFN-γ, TGF-β, OPG, receptor activator NFκB-ligand, IL-6, VEGF, COMP, CTX-I, NTX-I and BAP at all time points. Baseline and 2-week change data of all biomarkers were evaluated with regard to their association with 24-week major ASDAS response (decrease ≥2.0 units) as measure of clinical efficacy. Biomarkers were evaluated one by one as well as in combinations, together with 2-week ASDAS major response and 2-week CRP-response. All multivariate models were adjusted for age, sex and disease duration at baseline and treatment allocation. Results: 2-week decreases of IL-6 (p=0.006), NTX (p=0.026) and VEGF (p<0.001) were significantly associated with a major ASDAS response at 24 weeks. In the multivariate regression analysis a 2-week decrease in IL-6 (p=0.001) and VEGF (p<0.001) were significantly associated with the 24-week decrease in ASDAS. The effect of IL-6 disappeared after introduction of a 2-week change in CRP in the model. ROC-analysis revealed an optimal cut-off level of 145 pg/ml for the 2-week VEGF decrease (`VEGF response'). In the full logistic regression model (R-square:0.53) a 24-week ASDAS major response was best predicted by a 2-week ASDAS major response (OR: 13.9; p<0.001), a 2-week VEGF-response (OR: 3.9; p=0.001) and by treatment (OR: 3.9; p=0.042), after adjustment for age, sex and disease duration. Table 1 shows the translation of this model into response probabilities. Already two weeks after the first INF infusion, the presence of ASDAS major response and VEGF response preceded ASDAS major response at week 24 with 92% certainty. In the absence of both ASDAS major response and VEGF response after 2 weeks, the likelihood of a major clinical response after 24 weeks was only 17% for INF and 5% for placebo. Table 1. Prediction algorithm for a 24-week ASDAS major response TreatmentMajor responseVEGF responsen patientsProbability of major at 2wksat 2wksn patientsresponse at 24 wks PlaceboNoNo485% PlaceboNoYes614% INFNoNo7417% INFNoYes3042% INFYesNo1875% INFYesYes3092% Conclusions: Measuring VEGF levels in AS patients starting on INF therapy may help in predicting major clinical response. The predictive association between VEGF levels and ASDAS response on TNF-blocking drugs points to a potentially interesting role of angiogenesis in the pathogenesis of AS. Disclosure of Interest: X. Baraliakos: None Declared, R. Landewé: None Declared, B. Hsu Employee of: Centocor, D. Baker Employee of: Centocor, J. Braun Grant/Research support from: Centocor, D. van der Heijde Grant/Research support from: CentocorCitation: Annals of the Rheumatic Diseases, volume 70, supplement 3, year 2011, page 344Session: Spondylarthritis – treatment (Poster Presentations )