Precision medicine sounds like an inarguably good thing. It begins with the observation that individuals vary in their genetic makeup and that their diseases and responses to medications differ as a result. It then aims to find the right drug, for the right patient, at the right time, every time. The notion certainly has its supporters among medical experts. But for every one of them, there is another who thinks that efforts to achieve precision medicine are a waste of time and money. With a multimillion-dollar government-funded precision medicine initiative currently under way, debate is intensifying over whether this approach to treating disease can truly deliver on its promise to revolutionize health care.
Ask scientists who favor precision medicine for an example of what it might accomplish, and they are likely to tell you about ivacaftor, a new drug that has eased symptoms in a small and very specific subset of patients with cystic fibrosis. The disease stems from any of several defects in the protein that regulates the passage of salt molecules into and out of cells. One such defect prevents that protein from reaching the cell surface so that it can usher salt molecules back and forth. Ivacaftor corrects for this defect, which is caused by a handful of different genetic mutations and is responsible for roughly 5 percent of all cystic fibrosis cases. Genetic testing can reveal which individuals are eligible for this treatment.
The U.S. Food and Drug Administration fast-tracked development of ivacaftor a few years ago, and the drug been hailed ever since as the very essence of what of precision medicine is all about. Indeed, when President Barack Obama announced the launch of the government-funded precision medicine initiative in January 2015, he, too, sang ivacaftor’s praises: “In some patients with cystic fibrosis, this approach has reversed a disease once thought unstoppable.” Later the president declared that precision medicine “gives us one of the greatest opportunities for new medical breakthroughs that we have ever seen.”
But ask opponents for an example of why precision medicine is fatally flawed, and they, too, are likely to tell you about ivacaftor. The drug took decades to develop, costs $300,000 a year per patient, and is useless in the 95 percent of patients whose mutations are different from the ones that ivacaftor acts on.
Moreover, a recent study in the New England Journal of Medicine found that the extent to which ivacaftor helped its target patients was roughly equal to that of three far-lower-tech, universally applicable treatments: high-dose ibuprofen, aerosolized saline and the antibiotic azithromycin. “These latter innovations are part of many small-step improvements in [cystic fibrosis] management that have increased survival rates dramatically in the past two decades,” says Nigel Paneth, a pediatrician and epidemiologist at Michigan State University. “They cost a fraction of what the [high-tech] drugs cost, and they work for every patient.”
The same paradox applies to nearly every example of precision medicine you can find: clinicians viewed the use of a patient’s genotype to determine the right dose of the anticlotting medication warfarin as a godsend until some studies suggested that the approach did not work any better than dosing through old-fashioned clinical measures such as age, weight and gender. And the drug Gleevec was hailed as an emblem of targeted cancer therapy when it shrank tumors in a subset of leukemia patients with a very specific mutation in their tumors. But then a lot of those tumors developed new mutations that made them resistant to the drug, and when they did, the cancer returned. Gleevec bought patients time—a few months here, a year there—but it did not change the final outcome.
The debate over the merits of precision medicine has its roots in the Human Genome Project, the 13-year, $3-billion (in 1991 dollars) effort to sequence and map the full complement of human genes. Building on that work, scientists devised a shortcut for linking particular gene variants to specific diseases with as little sequencing as possible. That shortcut, known as GWAS, for genome-wide association studies, involved examining selected sites across the genome to see which ones differed consistently between individuals who suffered from a certain medical condition and individuals who did not. Hoping for a bonanza of new drug targets, pharmaceutical companies invested heavily in GWAS. But the approach proved poor at exposing the genetic roots of disease. Study after study turned up many clusters of gene variants, any one of which could predispose someone to a condition. In most cases, these variants nudged risk up or down only by a tiny sliver, if at all. The results cast a pall on the notion of studying genetic variation to develop targeted therapies on a large scale.
Proponents of precision medicine argue that the problem is not the notion of exploring genetic differences per se but the extremely limited scope of GWAS. Instead of looking for a few types of common gene variants that correlate with disease, they say, researchers need to examine the entire genome—all six billion nucleotides, the building blocks of DNA. And they need to superimpose those data on top of several other layers of information about everything from family history to the microbes that inhabit the body (the microbiome) and the chemical modifications to DNA that affect how active individual genes are (the epigenome). If they compared all the data, among as many individuals as possible, they would finally be able to pinpoint which constellation of forces drive which diseases, how best to identify those forces and how to devise treatments that target them.
The precision medicine initiative that President Obama announced last year aims to do exactly that. Its centerpiece is a million-person cohort, from whom data of every conceivable kind—including genome, microbiome, epigenome—will be collected and stored in one colossal database, where scientists can access it for an endless array of studies and analyses.
To understand how all these data are supposed to help scientists conquer humanity’s diseases, consider the example of warfarin. Knowing how fast or how thoroughly a person is apt to metabolize the drug should have made it easier to determine the best dose for that individual and should therefore have led to better outcomes. So why didn’t it? Might diet or other factors play a role? Scientists do not know, but with a million-person cohort, they think they might be able to find out. “I guarantee that there would be tens of thousands of them taking [warfarin],” says Francis Collins, director of the National Institutes of Health. “With that many subjects, you’ll be able to say, ‘Well, actually it does look like it helps this subset, but they happened to have a diet that was this form instead of that form.’” Furthermore, he notes, one would be able to see the subtleties of why and how a treatment works or does not work.
One thing supporters and detractors of the new initiative agree on is that the challenges of such an undertaking will be mammoth. It will require integrating terabytes of existing health data, spread across scores of databases whose content and quality will vary widely. And it will involve storing blood and tissue samples from one million people—no small feat, especially if those samples are collected at regular intervals. If it succeeds—if scientists find reliable predictors of disease in that mass of data and then devise ways to treat individual patients by targeting those predictors—doctors will still need to become fluent in this new language. Most physicians are not trained to make sense of existing genetic tests, and so far no one has come up with a good way to train them.
In theory, personalized medicine could work like Netflix and Amazon. They know every book and movie you have bought in the past few years, and armed with that information, they can predict what you are likely to purchase next. If your doctors had that kind of information at their fingertips—not about your purchase history but about how you live, where you work, what your genetic predispositions are, and which microbes are populating your skin and gut—then maybe cures would finally come like movie recommendations do.
But it seems fair to say it will be a very long time before science gets to the point where it can offer individually tailored treatment to the masses, if it ever does. The question is, Should it even try? Although precision medicine might make sense for people with certain conditions that are difficult and expensive to treat, such as autoimmune diseases, critics argue that on the whole, simpler approaches to treating disease are better because they cost less and benefit far more patients. “Let’s say we find a [targeted] drug that can lower risk of diabetes by two thirds,” Paneth says. “It would cost about $150,000 [a year per person] for that drug if we had it. A simple program focused on diet and exercise will do the same. Life span has increased by about a decade in the past 50 years. And none of that gain is related to DNA. It’s learning about smoking and diet and exercise. It’s old-fashioned stuff.”
In the end, this moon shot may make more sense as a research enterprise than a public health initiative. Scientists learn more every day about the distinct forces that interact to produce disease in individuals. It is natural and fitting that they should start putting that information together in a systematic way. But society should not expect such efforts to completely transform medicine any time soon.