Serum infliximab concentrations and efficacy outcomes at week 8 (time for induction efficacy end points for both ACT-1 and ACT-2), week 30 (time for maintenance end points for both ACT-1 and ACT-2), and week 54 (additional time for maintenance end points for only ACT-1) were the primary focus of these analyses. The prognostic value of earlier EPZ015666 purchase infliximab concentrations on

subsequent efficacy outcomes also was evaluated. Patient characteristics and serum infliximab concentration data were summarized using descriptive statistics. The correlation between serum infliximab concentrations at different time points was assessed using the Pearson correlation coefficient. Serum infliximab concentration data were compared between patients with and without the specified efficacy outcomes using a 2-sided Wilcoxon–Mann–Whitney, 2-sample, rank-sum test. Serum infliximab concentrations also were categorized into quartiles, and the trend of the proportion of patients with clinical outcomes across the quartiles was evaluated using the 1-sided Cochrane–Armitage trend test. Comparison of the

proportions of patients with a given efficacy outcome across serum infliximab concentration quartiles or across a given categoric variable was performed using the Fisher exact test, and the Kruskal–Wallis Crizotinib clinical trial test was used to compare continuous variables across quartile groups. The association between serum infliximab concentration (log-transformed) and clinical outcomes was evaluated further by multivariable logistic regression modeling. The effects of covariates (body weight, age, albumin, C-reactive protein level, Mayo score, sex, ATI status, and the use of immunosuppressive agents and corticosteroids) were assessed by logistic regression analyses. A backward elimination approach using a significance level of .05 for a covariate as a requirement for continued inclusion next in the model was adopted. Receiver operating characteristic

(ROC) curve analysis was used to identify infliximab concentration thresholds associated with efficacy during induction and maintenance. Optimal thresholds were chosen using the Youden17 index, which maximizes the sum of the specificity and sensitivity of the ROC curve. All authors had access to the study data and reviewed and approved the final manuscript. The baseline characteristics of patients who participated in ACT-1 and ACT-2 have been detailed.2 A summary of characteristics for patients who were randomized to infliximab treatment in both studies is provided in Supplementary Table 1. The distribution of serum infliximab concentrations observed at each visit through week 30 in patients receiving infliximab 5 or 10 mg/kg is shown in Supplementary Figure 2. When assessed by clinical response status (using total Mayo score) at week 8, serum infliximab concentrations over time were higher among patients with clinical response than among patients without response, as illustrated for both dose regimens in Figure 2.