Genetically Informed Clinical Trials
Figure 18: Traditional cancer drug development is slow and leads too often to failures in Phase III trials, which are key components in drug approval.
Moreover, traditional Phase III trials are huge and costly. Phase II trials are either single-arm trials evaluating tumor response or small and randomized
and have a time-to-event endpoint. In both cases the patient population is assumed to be homogeneous, or the same, even though everyone agrees
that it is heterogeneous, or contains multiple subpopulations. The “Personalized Trial” approach, such as taken in I-SPY 2, recognizes the
heterogeneous nature of the disease and the possibility that different treatments are effective for different patients. The “trial” is really a phase II
drug screening process. Drugs enter the process, are evaluated, and move on. The figure shows 6 experimental drugs, but there could be more. And
they enter the trial at different times. The goal is to match experimental treatments with molecular subtypes of disease, or “biomarker signatures.”
Experimental arms are dropped early (red X) in Phase II if they fail to show efficacy in any subset of disease. In view of the many combinations of
treatments and biomarker signatures, for arms suggesting a benefit within a particular signature, that benefit is partially confirmed in Phase II. There
are numerous efficiencies in this process that speed drug development, including the simple device of having a common control arm (C). But the
major efficiency is enabling a Phase III trial that is an order of magnitude smaller than in the traditional approach because it focuses only on the
responding patient population.
follows patient response as a function of their genotype.
Early results from this trial suggest that this approach will be
successful at linking biomarker signatures to drug response.
Importantly, because of their design, adaptive trials can
reduce the number of patients that must be enrolled in order
to achieve statistical significance. A large Phase III study
may typically need to enroll 3,000 or more patients to obtain
American Association for Cancer Research
enough data to receive FDA approval, whereas only 300
patients may be required in an adaptive trial (see Figure 18,
p. 63). Utilizing fewer patients per trial is increasingly
important because not every targeted therapy will work for
every patient. We will continue to witness these advances as
modern clinical trial designs are adapted to incorporate
biomarkers (see Figure 12, p. 48). Validated biomarkers
have the potential to transform cancer research and