P363 Are clinical decision support tools for vedolizumab and ustekinumab in biologic naïve patients with Crohn’s disease drug-specific?

Kim, J.E.(1);Won, H.(2);Kim, Y.H.(1);Kim, E.R.(1);Hong, S.N.(1);Chang, D.K.(1);

(1)Samsung medical center, Department of Medicine, Seoul, Korea- Republic Of;(2)Samsung Medical Center, Biomedical Statistics Center- Research Institute for Future Medicine, Seoul, Korea- Republic Of; Hojeong Won

Background

Clinical decision support tools (CDST) were developed to predict drug response with introduction of various biologics to Crohn’s disease (CD) treatment. This study aimed to determine whether CDST’s for vedolizumab (V-CDST) and ustekinumab (U-CDST) could be used as prognostic markers or drug-specific markers to predict response.

Methods

A hypothetical scenario was created with 872 CD patients who used first biologic at Samsung Medical Center between 1995 and 2020. V-CDST & U-CDST were calculated with clinical and laboratory data just before the first biologic was started. Cumulative Link Mixed Model (CLMM) test, weighted Kappa and plot, and Spearman’s correlation statistical techniques were used to check the degree of agreement and difference between the two tools.

Results

Twenty-five percent of biologic naïve CD patients could be categorized to different probability group using V-CDST and U-CDST. The difference between the two tools was significant [Two-sample Paired Ordinal Test with Cumulative Link Mixed Model (CLMM), p value < 0.001)]. Concordance between the two tools with a total of 654 people (75% of all patients) was evaluated to have the same probability (weighted Kappa: 0.47, 95% CI: 0.41-0.52).

Conclusion

V-CDST and U-CDST are similar and different. V-CDST and U-CDST are useful for choosing vedolizumab or ustekinumab in 25% of biologic-naïve CD patients in our hypothetical scenario.