P631 Development and validation of a clinical scoring tool for predicting treatment outcomes with vedolizumab in patients with ulcerative colitis

• Published in: Journal of Crohn's and colitis 2019-01, Vol.13 (Supplement_1), p.S433-S433
• Main Authors: Dulai, P S, Singh, S, Vande Casteele, N, Meserve, J, Winters, A, Chablaney, S, Aniwan, S, Shashi, P, Kochhar, G, Weiss, A, Koliani-Pace, J L, Gao, Y, Boland, B S, Chang, J T, Faleck, D, Hirten, R, Ungaro, R, Lukin, D, Sultan, K, Hudesman, D, Chang, S, Bohm, M, Varma, S, Fischer, M, Shmidt, E, Swaminath, A, Gupta, N, Rosario, M, Jairath, V, Guizzetti, L, Feagan, B G, Siegel, C A, Shen, B, Kane, S, Loftus Jr, E V, Sandborn, W J, Sands, B E, Colombel, J-F, Lasch, K, Cao, C
• Format: Article
• Language: English
• ISSN: 1873-9946; 1876-4479
• DOI: 10.1093/ecco-jcc/jjy222.755

Abstract Background We created and validated a clinical decision support tool (CDST) for vedolizumab (VDZ) therapy in active ulcerative colitis (UC). Methods To identify factors associated with corticosteroid-free remission (CSFREM; full Mayo score ≤2, no sub-score >1), logistic regression analyses were run on data from the GEMINI 1 VDZ trial for UC (derivation set; n = 620) and used to develop a CDST. Correlations between VDZ exposure, onset of action, and efficacy across predicted-probability groups were explored, and the CDST was externally validated in an observational cohort of VDZ-treated UC patients (validation set; n = 199). Results Factors independently associated with CSFREM were absence of previous tumour necrosis factor antagonist exposure (+3 points), disease duration ≥2 years (+3 points), baseline endoscopic activity (moderate vs. severe) (+2 points), and baseline albumin concentration (+0.65 points per g/l). Patients were stratified into low (≤26 points), intermediate (>26 to ≤32 points), or high (>32 points) probability of response groups. The higher probability group more rapidly achieved symptom activity reductions and attained higher rates of CSFREM (p < 0.001). In the validation set, a 26-point cut-off value showed high sensitivity (93%) for identifying non-responders. A statistically significant linear relationship was observed between VDZ exposure, probability groups, and efficacy in the derivation set (p < 0.001). In the validation set, only the low–intermediate probability group benefited from VDZ interval shortening for lack of response (p = 0.02). Conclusions We developed and externally validated a CDST with good discriminative performance for predicting CSFREM with VDZ in UC patients. Pending further validation, this tool could be a helpful aid in identifying patients who would benefit from VDZ interval shortening due to insufficient response. (GEMINI 1: NCT00783718).