FDA's Genomic Screening Approval Blurs Research and Practice

Vinay Prasad, MD, MPH


June 28, 2018

If I were to ask you, of all the patients you see in your clinic who are tested for a gene or mutation and then receive a drug that targets that gene or mutation, what percentage of these patients benefit from those drugs? What is your guess: 9%, 20%, or 40%?

Many of us would be surprised to know that the answer to that question is probably close to 9%. That is the answer we came to in a recent paper in JAMA Oncology.[1] Of all the people who present with metastatic cancer as of 2018, we estimate that 9% of them will have a mutation for which we have a drug that targets that mutation and is US Food and Drug Administration (FDA)-approved for that mutation and that tumor type. This includes the anaplastic lymphoma kinase (ALK) gene and the epidermal growth factor receptor (EGFR) gene for lung cancer. It also includes BCR-ABL for imatinib in chronic myeloid leukemia. But what is most sobering about this number is that many of us may have guessed that the answer is 20% or 40%, judging by the rhetoric around this topic.

Our analysis also found that if you graph the percentage of patients in this country who benefit from genome-targeted therapy over time, it looks like a straight line trending upward at about 0.5% per year. So, a few more patients benefit in 2018 than 2017, which is a few more than 2016, and a few more than 2015, et cetera. But it doesn't appear to be taking off; it doesn't appear to be an exponential growth curve.

Why does this matter? During the course of the 2018 American Society of Clinical Oncology (ASCO) annual meeting, we have debated the role of precision oncology in cancer medicine. I fully believe there is a role for this, but what we may be conveying to patients may overemphasize or exaggerate the role. This is the simple way I like to think about it: We have some precision or genome-driven oncology therapies that are ready for practice, and those should be a part of our practices. And we have a bunch of therapies that are still research questions.

There are fundamental differences between practice and research. The characteristic of practice is that [we use] interventions where we have proof that they help our patients, and thus we feel comfortable deploying them in the real world. They tend to be paid for by commercial payers, such as insurance companies, Medicare, or Medicaid, or sometimes patients, with their copays. We don't generally keep track of overall outcomes; we treat our patients. We know what happens to them, but we don't publish detailed statistics from our own practices.

Let's contrast that with research. With research, we are talking about drugs that may target certain mutations, but we don't know for sure that they will benefit our patients. We may believe they will and may have promising evidence that they might, but we don't know for sure. Research is typically paid for by commercial sponsors, the companies that make the drugs or the companies that sell the genetic tests, or the National Institutes of Health. Finally, in research, we keep meticulous track of the clinical outcomes. We want to know what happens to patients. We want to stop ineffective therapies early, and we want to rapidly publicize the results of therapies that do work. Those are the fundamental differences between practice and research.

Much of the debate about precision oncology is around at what point [a new test or a new drug should] move from research to practice. Proponents have argued that it should move very rapidly with a very low bar, but I believe that some of the [therapies] that are currently moving into practice are rightly classified as research. For example, this year the Centers for Medicare & Medicaid Services (CMS) has announced they will pay for F1CDx, Foundation Medicine's 324-gene panel for every patient with advanced solid tumors, for a one-time test.[2] I believe this is largely still research, although some of the results of that test are for mutations for which we have FDA-approved drugs. Many times, we get a FoundationOne report (the genomic assay for solid tumor cancers) and it is difficult to put it into practice. We are given recommendations for many drugs that are off-label, targeted drugs, and we really don't know whether that mutation and that tumor type will respond to that drug.

I would have preferred to see more research on this test, supported by commercial sponsors, such as Foundation Medicine, until there was reliable proof of benefit. I have argued[3] that we should conduct a randomized controlled trial, and I believe there is an easy one to do. Enroll 2000 people who come to our clinics, having exhausted proven therapies. Randomly assign 1000 to undergo Foundation Medicine or other genomic profiling, and then match them with whatever therapies [may be appropriate]; you can put them [in a clinical] trial or prescribe off-label use of drugs, as is a common practice in this country. The other 1000 patients are treated as we have always treated patients in oncology, looking at published phase 2 studies to find evidence of older drugs that may work in those tumor types, or using our best judgment with the knowledge that we have—the conventional treatment approach.

On the basis of some of the studies that have been published by other groups, I have even powered this study to detect a hazard ratio of 0.88, and I published that power calculation in the Annals of Oncology.[3] Thus, what CMS should have done instead of paying for this test broadly was to run the trial I propose. Instead of paying for the test in 500,000 people in the next year, pay for the test in 2000 people and, in about a year's time, we can answer whether or not there is a survival benefit from this precision oncology strategy.

Many people disagree with me, and [it is] easy to caricature my position [to say that] I disagree with all genomic therapies. That's wrong. I use every one of the FDA-approved therapies. Sometimes I wish I had better data than single response rates and uncontrolled studies. I wish I had more robust data or replication data, but I nevertheless use those proven therapies. But for many of these drugs that we are bringing into our practices on the basis of the results of this test, we don't have that information; we don't know whether that mutation in that cancer has the same benefit. I believe that is rightfully considered research, and commercial funders and insurers rightly should not pay for those just yet, until more evidence is behind them.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.