P489. Exposure–response relationship during vedolizumab induction therapy in adults with ulcerative colitis
M. Rosario1, J. French2, N. Dirks2, A. Milton1, I. Fox1, M. Gastonguay2, 1Takeda Pharmaceuticals International Company, Clinical Pharmacology, Cambridge, United States, 2Metrum Research Group LLC, Biotechnology, Tariffville, United States
Vedolizumab (VDZ) 300 mg, currently in development for treating ulcerative colitis (UC), has demonstrated an apparent exposure-response relationship (ERR) in the GEMINI 1 study. However, confounding variables may impact VDZ pharmacokinetics (PK) and/or efficacy and thus the ERR. Analyses presented here evaluate the VDZ ERR and explore the potential for confounding variables from data obtained in the GEMINI 1 study (NCT00783718).
Exploratory plots of week 6 clinical response (decrease in complete Mayo score of ≥3 points and ≥30% from baseline, and a decrease in rectal bleeding subscore of ≥1 point or an absolute rectal bleeding subscore of ≤1 point) and clinical remission (complete Mayo score of ≤2 points and no subscore >1 point) versus individual predicted average VDZ concentration (Caverage), derived from the VDZ population PK model, were generated using data from the GEMINI 1 study. ERRs were assessed using a logistic regression model, which adjusted for potentially confounding baseline covariates (eg, prior anti-tumour necrosis factor alpha [TNFalpha] use, albumin, fecal calprotectin) and an interaction model, which assessed ERR dependence on the covariates. Adjusted and interaction models were compared using a likelihood ratio test.
Exploratory plot results suggest that increasing VDZ Caverage increased the probability of both clinical response and remission, with the ERRs appearing to plateau at high Caverage values. Estimated ERRs from the logistic regression model were positive for clinical response (P < 0.01) and remission (P < 0.01); however, the ERR slopes were less steep than the exploratory plot, suggesting the plots may not adequately characterize the effect of VDZ exposure on efficacy. Within groups defined by quartiles of baseline covariates between Caverage values of 35 to 84 mcg/mL (5th and 95th percentiles), average probability of clinical response and clinical remission increased by ∼20% and ∼15%, respectively. On average, patients with higher baseline albumin levels, lower baseline fecal calprotectin levels, and no prior anti-TNFalpha use had higher probability of clinical response whether receiving placebo or VDZ; however, the largest difference in probability of clinical response within baseline groups appears to be prior anti-TNFalpha use. No significant improvement of the interaction models over the adjusted models was observed for either end point (clinical response, P = 0.971; clinical remission, P = 0.227).
Exploratory ERR analysis demonstrated that increasing Caverage leads to higher probability of clinical remission and clinical response at week 6. While confounding variables can affect the VDZ ERR, it is not dependent on any covariates tested.