Explore Enrichment Factors
Background. Recently we completed a project and explored the enrichment factors that can be used to increase the ability of the clinical trial to demonstrate the treatment effect. The analyzed data is from a randomized, double blind, placebo-controlled Phase 3 clinical trial conducted in multinational centers.
Purposes and Assumptions. Using the analysis data from this study, a series of analyses on patients’ baseline measurements in the ITT population were performed with two purposes: 1) to identify a group of patients (enriching group) who might have greater benefit from the treatment; or 2) a large placebo responders. In terms of the disease characteristics, demographic factors, such as age, gender, and age-gender interaction were first assumed to be potential enriching factors. Meanwhile, the medical history, disease severity at baseline and concomitant drug usage as well as their interactions were also assumed to have influence on the drug effects.
Methods. Based on above assumptions, descriptive analyses on the baseline factors, such as age, gender, disease severity which is defined above certain behavior threshold as severe, medical and medicine history, biomarkers, etc., were conducted to better understand the distribution patterns of treatment effect. Following descriptive analyses, based on results from descriptive analyses, generalized linear regressions were performed to examine least square means change from baseline for potential enriching groups. Interactions of potential enriching factors were also examined. A variety of plots were drawn to demonstrate an intuitive association of drug effectiveness with potential enriching factors.
Results. The density plots illustrated that, the responses in treatment group was related to severity, and being consistent in male and female groups, while the correlation did not appear in the corresponding placebo arm. The box plot for distribution of means change from Baseline demonstrated the significant correlation of concomitant drug usage with Placebo responders. The highly statistically significant response for the patients with specific biomarker and non-concomitant drug usage was observed. Strong placebo effect for concomitant drug usage was observed. Analyses results also illustrated that baseline severity and the specific biomarker are two independent factors jointly correlated with placebo response. The response from the patients with the specific biomarker presented significant treatment effect among severe patients and strong placebo effect among non-biomarker patients. Furthermore, a significant interaction effect of treatment with severity was observed. In specific biomarker group, least square means change from baseline for the patients whose symptoms remain above predefined threshold value was significantly larger in treatment group compared to placebo group, while treatment effect was not significant among the corresponding patients with low severity.
Conclusion. Baseline severity is highly correlated to the treatment response. Concomitant usage is significantly correlated with placebo response. The group patients with specific biomarker and non-concomitant usage present an incredibly significant treatment effect.
The analyses results suggests that the specific biomarker, pre-defined threshold as severity and concomitant drug usage could be used to develop enrichment strategies in the new study propose to increase the ability of the clinical trial to demonstrate the treatment effect and potentially reduce the sample size for detecting treatment effects.
By Jeffery Zhang PhD.