RESUMEN
In targeted therapy or immunotherapy, it is common that only a subpopulation of patients are sensitive to and thus may benefit from the therapy. In practice, based on pre-clinical data, it is often assumed that the sensitive subpopulation is known. Subsequent clinical trial data, however, often show that this assumptions is false. We propose a randomized, group sequential enrichment (GSE) design to evaluate an experimental treatment against a control. The GSE design starts by enrolling patients under broad eligibility criteria and then alters the entry criteria to restrict enrollment to treatment-sensitive patients based on accumulating data of short-term and long-term survival endpoints. The short-term endpoint is used to facilitate enrichment when a limited number of survival events are observed, while the final determination of treatment efficacy and sensitive subpopulation is based on the survival. The group sequential approach is adopted to implement the adaptive enrichment strategy. The proposed design simultaneously achieves two primary goals of precision medicine: to determine whether the experimental drug is superior to the control and to identify the target population that is sensitive to the treatment. A simulation study shows that the proposed design well controls the type I error rate and yields substantially higher power than the conventional group sequential design and existing two-stage enrichment design.