희귀질환 등 소집단 대상 임상시험에서의 연구대상자 수에 대한 고찰

Abstract

Since clinical trials for rare disease subjects have a small number of patients, it is difficult to obtain sufficient size of subjects to test their effects. In addition, the number of samples calculated based on the effect size, significance level, and power is larger than the total number of patients, so the application of the clinical trial itself may not be possible. Therefore, there is a need to study for clinical trial in small population such as recommendation guidelines and designs. In this study, we compared the main issues of each country related to orphan drugs, and reviewed preferentially in the guidelines for issues and precedent studies in clinical trials for small population. Next, a total of three methods was explored to calculate the optimal number of samples. We studied general method that use significance level and power, Assurance, and Decision theoretic method. Finally, we used the results of clinical trials of Parkinson's disease, which is one of the rare diseases, to present the optimal sample size calculated by each method. In previous studies, the design methods that can be applied to small population clinical trials include Adaptive Design, Sequential designs, Boundaries design, Within-patient designs(Repeated measures, Crossover trial, N-of-1 trial), Enriched enrollment, and Factor design. Based on the Parkinson 's disease, a total of 154 persons were calculated by general method. With Assurance method, the sample size was calculated to be 190 people. Also we calculated 67 with Decision theoretic method, which is the smallest among 3 methods. In addition, it was confirmed that the optimal number of samples was determined according to the parameters of prior information and the total population size N to clinical trials, using the decision theoretic method. In the result, Assurance method can not be an effective alternative because more sample size is required than the general method. Decision theoretic method can test the same effect size with much fewer samples than the previous two methods. But uncertainty and distribution of initial parameters may not be appropriate, careful consideration is required to apply it in actual clinical trials.