Medicine has always been personalized, in some sense. Physicians diagnose and treat illnesses according to not only symptoms but also the characteristics of the patient, like gender, medical history, and habits. Yet there are invisible differences between us. One woman's breast cancer may be more likely to spread than another's, even if they look the same; one patient's genetic makeup makes him develop dangerous bleeding after taking a medicine; a colon cancer treatment works in one person but not in another because their tumors are different. Personalized medicine focuses on those variations and aims to match each patient with the therapy most effective for her—not just for the average patient. To find out where it's headed, U.S. News talked to Edward Abrahams, executive director of the Personalized Medicine Coalition, whose members include governmental, academic, nonprofit, and for-profit entities. Here's an edited version of the conversation.
Cancer is where most of the progress in personalized medicine has occurred. Certain treatments are now offered only if a colon cancer patient's tumor has a mutated KRAS gene, for example. How's the progress in other areas of medicine?
There are now 37 products on the market that facilitate personalized therapy. Most of those are in oncology. But an increasing percentage are in other areas. There are products emerging in cardiology and in central nervous system disorders, in diseases such as autism, in diabetes, and so on. It would be very important, for example, if antidepressants were linked to particular diagnostic tests that would tell patients which would work for them. There is much work being done in that area, and it's likely to have an impact in the very near future. The Centers for Medicare and Medicaid Services recently decided that, except in clinical trials, it won't pay for a genetic test designed to determine the ideal dosage of the anticlotting drug warfarin, also called Coumadin. Isn't that a setback for personalized medicine?
We're not asking CMS or private insurers to pay for something until we're sure it works, and we know that we must generate more evidence to persuade CMS and others. But our assumption is that the trials are going to find [the test offers] very real benefits by avoiding [bleeding, a complication caused by overdoses of] Coumadin. But you've called trials for diagnostics potentially impractical.
To generate evidence for diagnostic tests that can have a low return on investment and low profit margins can be very difficult. They don't lend themselves to randomized clinical trials, which can be prohibitively expensive. This is a challenge for personalized medicine—to generate the necessary evidence. My suggestion is that we have variegated standards of evidence, including observational studies. What's the most significant scientific hurdle to progress in personalized medicine?
Biology is complex, and it doesn't always lend itself to easy answers. For example, there are multiple genetic bases for many illnesses—it's usually not as simple as the BRCA mutations in a subset of breast cancer cases. And genes interact with the environment to produce disease. Teasing out these complex origins, never mind prescribing a solution, can be difficult. Why would drugmakers want tests that reveal that some people won't be helped by their drugs?
That used to be a common view, but no longer. Drug companies have come to understand that putting safer and more efficacious drugs on the market serves their interests more than failed drugs, and they've embraced the principles of personalized medicine. There have been any number of drugs withdrawn from the market because of adverse events for a small minority of people. If you knew [who'd suffer side effects] in advance, that could be avoided. Are you worried that comparative effective research (CER), which compares how treatments stack up against one another, might slow progress in personalized medicine?
We're very supportive of CER because we need to know what works and for whom. The challenge is to make sure that individual variation is considered. We can't use old population data that show, for example, that the average person responds better to the blue pill than the red one. We need to see research showing whether the red pill might work better for some group of people. The government can save money with this approach because they'll get the treatment right the first time.