5.3 The Cost and Benefit of a Biomarker Paradigm
Background: What is the cost of failing to refute or accept a hypothesis at the right time? Regulatory approval for a new drug or device provides one measure for when the evidence in favor of a scientific hypothesis is sufficient to be accepted. However, some advocates of biomarker endpoints argue that the current regulatory threshold for acceptance is too high. At the other end of the spectrum, there is simply no well-accepted standard for failure—that is, when sufficiently negative evidence indicates that it is no longer ethical to continue expending further research resources or enrolling patients in studies. The shift from traditional drug development to biomarker-driven development further complicates this problem, since drugs can fail without necessarily demonstrating that the underlying biological theory—which is often conceptually linked to the biomarker—is wrong.
Aim: To provide a clear picture for how biomarker hypotheses evolve (or fail to evolve) in the face of accumulating evidence and the moral and economic burden associated with alternative thresholds for judging when a biomarker hypothesis should be considered decisively refuted or confirmed. Our analysis will provide recommendations for the thresholds of evidence that regulators and policymakers use to approve novel interventions or re-direct research funding streams.
Approach: Each of our four empirical projects will track the accumulation of evidence for and against a particular biomarker paradigm. For our disease domains with effective interventions (e.g., statins in cardiovascular disease), we will explore alternative thresholds for judging success, and can then consider the value of the “additional” trials that were used to support the drug’s regulatory approval. For the disease domains without successful interventions (e.g., amyloid-centric drugs in Alzheimer’s disease), we can similarly explore alternative thresholds for failure, considering the value of trials that continue to test a hypothesis after it should have been rejected. We will also determine whether we can estimate the total cost and benefit associated with a given biomarker research paradigm.