Population difference and disease status affect the association between genetic variants and gene expression.

Abstract

Reply. We thank the authors of the letter for their thoughtful comments in response to our article. We agree that centralized endoscopic reading represents an important advance in inflammatory bowel disease (IBD) clinical trials. Across studies in the IBD space in the past decade, the sample sizes used to achieve statistical significance in late phase clinical trials have been decreasing. As the authors of the letter have pointed out, this trend could be the result of the decrease in placebo response rates in IBD trials over this same time period. Although this trend could be attributable to the use of central reading to identify patients with active disease for study inclusion, other factors could be involved. For example, in the trial leading to the approval of vedolizumab for ulcerative colitis, clinical remission rates were 16.9% in the active treatment group and 5.4% in the placebo treatment group. In the trial leading to the approval of vedolizumab for Crohn’s disease, clinical remission rates were 14.5% in the active treatment group and 6.8% in the placebo group. In UNITI-1, which led to the approval of ustekinumab for Crohn’s disease, clinical remission rates were 20.9% in the active treatment group and 7.3% in the placebo treatment group. A notable feature of all of these studies was the absence of central endoscopic reading as the basis for study entry. In the first 2 trials, 41% and 57.8% of patients, respectively, had failed 1 anti-tumor necrosis factor agent; in UNITI-1, 100% of patients had primary nonresponse to anti-tumor necrosis factor agents. In addition, across the 3 studies, patients were required to exhibit evidence of active disease by C-reactive protein, fecal calprotectin or a locally read sigmoidoscopy or colonoscopy. In a study examining the effects of central reading on mesalamine ulcerative colitis remission, retrospective application of central endoscopy resulted in only a 3.7% decrease in the placebo response rate, from 16.3% to 12.6%. These findings collectively suggest that the decreasing placebo rates in recent IBD clinical trials could at least be attributed, in part, to the inclusion of patients who were refractory to biologic therapies or who fulfilled more stringent criteria for active disease rather than use of central reading per se. Furthermore, screen failure rates are as much a concern for the conduct of a clinical trial as the decrease in the number of patients needed to be randomized to demonstrate a treatment effect. Although actual numbers are not published, experience suggests that the screen failure rates in clinical trials now exceed 50% in ulcerative colitis and approach 70% in Crohn’s disease. Although setting high bars for trial entry might decrease the sample sizes required for statistical significance, this strategy is counterproductive if suitable patients aremore difficult to find, resulting in the exclusion of patients who could otherwise participate in these trials. Finally, the absolute effect of more stringent criteria for study entry depends on the actual effects of these interventions on both the placebo and active treatment groups. One would be less excited by a decrease in placebo response rate if it were accompanied by a decrease in the active control group response or diminishing treatment effect. As pointed out by the authors of the letter, reductions in absolute drug efficacy rates in parallel with decreasing placebo rates are of concern. We applaud the efforts of the authors of the letter to improve the efficiency of IBD clinical trials and thank them for their suggestions. However, many factors need to be considered before any single solution to the problem of enrollment in IBD clinical trials can be identified.