In melanoma patients, resistance to BRAF inhibitors and the development of secondary tumors after treatment may be due to mutations in RAS, according to a study published in the January issue of the New England Journal of Medicine. In a related editorial in the same issue, the Wistar Institute's Ashani Weeraratna described BRAF inhibition as a good example of when a tailored treatment is necessary. She recommended that patients taking BRAF inhibitors be tested and their RAS status determined to reduce the risk of secondary tumors. Such testing could better define the population of patients who should be taking these drugs, she added.
The literature is full of such examples — findings that, were they implemented in the clinic, could help clinicians determine how best to treat their patients while avoiding adverse side effects or the risk of a drug being ineffective for a specific person. According to the Personalized Medicine Coalition, more such tests are getting to the clinic. In 2011, the coalition counted 72 prominent examples of personalized medicine-based treatments and diagnostics available in the clinic, compared to 13 such examples in 2006. In addition, the coalition says many drugs are now being developed with companion diagnostics in mind, and hospitals in the US are slowly starting to use tools like electronic medical records that will also help them handle the rush of genomic data they may soon be dealing with.
Despite the advances, however, many in the scientific community say the acceptance of personalized medicine in the clinic has been slow and painful, and there needs to be a shift in how everyone from the medical community, to the regulatory community, to the patients themselves think about personalized medicine.
"There's a profound chasm between the knowledge and the change of medical practice," says Eric Topol, professor of translational genomics at the Scripps Research Institute. "Part of it is, certainly, that there's always a need for more knowledge. That's what the naysayers will say — 'You need to know more before we do anything.' But in reality, we actually have plenty now to actualize important changes in medical practice, and that's where I think the opportunity exists and where we're not clicking on the potential."
The primary challenge facing the adoption of personalized healthcare is that "the systems that are in place that govern healthcare were developed over time for a one-size-fits-all, dry-and-narrow world. They do not send the right signals to industry providers or patients to move to a world of personalized medicine or targeted therapeutics," says PMC President Edward Abrahams. "Essentially, these decades-long patterns of intertwined and misaligned regulations have to be changed, and should be changed."
The problems creating a bottleneck in the implementation of personalized medicine fall into three categories, he says — education, regulation, and reimbursement. In terms of education, the problem is that the medical establishment is fairly set in its ways, and many clinicians are both unfamiliar with new genomics research, and reluctant to change the way they do things.
"We know that there's a knowledge gap in the physician community — 90 percent in large surveys of tens of thousands of doctors have said they don't feel comfortable using genomic data in managing their patients," says Topol, who has a new book on the subject called The Creative Destruction of Medicine. "So we've got a real problem, because among consumers that have been polled, 90 percent trust their doctors the most with their data. So we have a lack of comfort, and we have an increasing availability of this type of data and information, and it just sits there and goes nowhere."
Many doctors are unwilling to give up the "rituals of medicine" that they are comfortable with, he adds. For example, although it might be more useful for sequencing and genomic research to freeze tumor samples, surgeons and pathologists most often store tissue in formalin, which tends to make meaningful sequencing more difficult. "That's one of those things where it's like, 'We've done it this way for decades, and we're not going to change,'" Topol says.
Similarly, Sultan Meghji, a vice president at analytics development firm Appistry, says that it is important to find the "cultural critical mass to adoption" for personalized medicine before these scattered tests and targeted therapies can become part of a larger cohesive strategy for treating patients. "There are people out there who are very forward thinking and thinking about this in the right way, which is that this is a proven technology," Meghji says. "But just like with electronic medical record systems in hospitals and clinics over the past few years, you'll see some doctors who just refuse to use them."
Recently, Meghji recounts having dinner with a prominent oncologist who felt there was no proof that genetic diagnostics had any utility or value for the treatment of cancer. "That's really an incredible thing to hear from a leader in that field, especially because the science over the past two to three years has completely proven that, and there's no doubt," he adds. "But there are people who are from a different generation and there's a cultural lag that we're going to have to get through."
Although younger generations of doctors are increasingly learning about genomics and how to apply it in practice, Topol says there isn't time to wait for those doctors to effect widespread change in the clinic. "There's certainly a higher interest level among the digital natives who are in medicine, but not the digital immigrants," he says, referring to younger doctors' greater ease with new technologies and methods. "If we rely on that, however, we're in trouble."
Regulation and reimbursement
Overcoming cultural biases isn't the only challenge proponents of personalized medicine face, however. Regulations governing the translation of genomics research to the clinic, and the problem of figuring out who pays for these tests and drugs are also standing in the way of full implementation.
In terms of regulation, the current system of government approval for a new drug or diagnostic in the US is "at best, muddled," PMC's Abrahams says. While the US Food and Drug Administration is working on new guidelines and regulations for the approval of personalized medicines and diagnostics, there are still hurdles to overcome, especially with how clinical trials are designed, and how the utility of a drug or diagnostic is gauged — not by the number of people it helps, but by how much it helps a very targeted population of patients.
Appistry's Meghji calls the regulatory environment "archaic," adding that the rules are designed for "a universe where you have an assumed process about how you do a diagnostic" or develop a drug. And because the process for approving a new drug or diagnostic takes so long, many say it's not even worth it to go through the process only to risk possible rejection of a product. "So many biotech startups are trying to piggyback on something that's already out there or find alternative uses for something that has approval, because it's just such a disaster," Meghji says.
And the question of who would pay for a test were it administered is also important. Many biotech and pharmaceutical companies are busy trying to show insurance companies that not only are their genomics-based products effective, but that they can also save the payors money. "In reimbursement, there is again a challenge where we do not reimburse for value," Abrahams says. "It's more focused on quantity of performance, rather than the value of the product. We would like to see an environment where, if personalized medicine products are approved, they have an expedited consideration."
Topol adds that, at least in diagnostics, hospitals also have a role to play in keeping the cost of tests down. Hospitals and clinics that send out simple diagnostic testing to be done at large corporate labs for hundreds of dollars, instead of in-house in their own labs for less, send the impression that they're complacent and unwilling to be part of changing the system, he says.
"I think some [insurance companies] are covering more tests, but I don't see a forcing function because you have all these other blocks upstream before it even gets to them," Meghji adds. "There isn't a lot of pressure on them from their customer base to actually support this, so it's actually going very slowly."
The technical aspects
There are also still technical gaps that need to be bridged before genomics research discoveries can be applied to patients in the clinic. Electronic medical records are a necessary part of a personalized medicine future, Abrahams says. Without a reorganization of the clinical infrastructure, it would be impossible for doctors to keep track of their patients' data or the analyses of that data. But electronic medical records are more than just organizational tools, Abrahams adds. "The more electronic medical records we have, the more integrated that is into the system, the better the healthcare system will facilitate the understanding of patient variation, and more research, and, in turn, better products and better outcomes. So all of that is very, very important," he says.
Meghji says that, in addition to more research, what's also needed are less expensive sequencing technologies. Although the price of sequencing has dropped dramatically in the past few years as the speed has increased, it's still too slow and too expensive to be used as a healthcare tool. "The acquisition costs, the operation costs, the per-unit costs, and then the time to get the useful data out of them are still too high," Meghji says. "You're still talking six figures for even a small sequencer, and your per-run costs are still in the thousands of dollars. They need to be in the hundreds of dollars."
And the additional bioinformatics analysis needed to make sense of the data to create an actionable healthcare plan currently takes too much time. "One of the things we're seeing is that, because there's not a lot of bioinformatics talent out there — especially in the applied clinical sphere — that even though something may be well understood in the scientific perspective and validated, it takes a lot of effort for most organizations to productionalize that and make it useable in a commercial or clinical setting," Meghji adds.
On the bright side
Despite these problems, however, some say the progress in moving discoveries from the bench to the clinic is proceeding at the right pace. "We can see development, we can see progress, and this benefits patients and the healthcare system, and has the potential to change the way medicine is considered in the future," PMC's Abrahams says. "There [are] many successful examples of personalized medicine in the clinic." In particular, Abrahams points to Pfizer's new drug, Xalkori, which treats a subtype of non-small-cell lung cancer. The drug only works for 3 percent to 5 percent of non-small-cell lung cancer patients, but for those patients, it is very effective in treating their disease — in two studies of the drug, 50 percent and 61 percent of patients, respectively, saw their tumors shrink partially or completely. "This is an example that is linked to a diagnostic and is an example of a success of personalized medicine," Abrahams adds. "For a particular sub-population of patients, it has the ability to make a difference."
The Institute for Systems Biology's Leroy Hood — who coined the term "P4 Medicine" in 2003 to represent his ideal of predictive, preventive, personalized, and participatory medicine — says he disagrees with the idea that there is a holdup in the pace of personalized medicine reaching the clinic. "I think that's a narrow, doctrinaire view from someone who probably doesn't understand personalized medicine too well, and there are a lot of physicians that fall into this category," Hood says. "I think people are resistant to any kinds of changes, and what's very easy to do is to always have a pessimistic interpretation of anything."
Hood says personalized medicine is on the right 10- to 15-year trajectory. And, he says, 10 years into the future, he envisions that every patient will be surrounded by a virtual cloud of billions of data points, and that physicians will be able to use bioinformatics tools to distill all those data points into "simple hypotheses about health and disease."
Examples of personalized medicine successes abound, Hood says. For instance, there are nearly 200 examples of actionable gene variants being tested for, and the results of those tests are being used by clinicians and their patients when making healthcare decisions. Doctors already have tools that make blood a "window into assessing health and disease," he adds. Clinicians also have a number of tests to check for certain organ-specific proteins in blood to determine if a patient has a disease and to what stage it may have progressed. Doctors also have the ability to stratify various diseases, like the subtypes of different cancers, and tailor a patient's treatment regimen to that specific subtype.
Over the long term, researchers are learning how to re-engineer disease networks using various compounds and chemicals, and this knowledge is being used by drug companies to create new therapies, Hood says. Though there may not be many such drugs on the market today, an increasing number are in development.
More drugs are also being developed to have large effects on specific subgroups of patients, instead of smaller impacts on a more varied population. Herceptin is a good example of that, Hood says. The Genentech drug is effective in attacking breast cancer in HER2-positive patients — about 20 percent of all breast cancer patients — and this target population is identified by a companion diagnostic that accompanies the drug. "That's a classic example of personalized medicine," Hood says.
"We're already getting fundamental new insights into disease mechanisms and thinking in new ways about diagnosis and therapy," he adds. "Once you get your genome done, it's going to be re-analyzed each year for the new actionable variants, so your genome is an investment in your future health that will pay off year by year."
To say the promise of personalized medicine hasn't yet materialized seems wrong to Hood, who says he's unsure what the critics want. "Do they want us to get full-blown and everybody is being treated individually? That's not how any change in science occurs. It's an incremental thing that is cumulative over time," he adds.
The way forward
This is not to say that Hood is unaware of the challenges involved in making personalized medicine widespread. There are two barriers to the acceptance of P4 medicine, he says. One is technical and the other is social. Although he says the technical aspects are proceeding apace, with increasingly faster sequencing and more streamlined bioinformatics, the social barrier — "ethical considerations, considerations of data security, questions likes who owns the data and who can have access to the data, and so forth" — is more challenging to overcome.
"I argue that the data for every patient should belong to society and not the individual," Hood says. "It's society that has created the tools to enable the data to help with the individual's health and, even more important, it's critical that that data be available after anonymization for people to analyze and mine for the predictive medicine of the future. This is what's going to revolutionize medicine for your children and your grandchildren."
For companies like Appistry, part of a commitment to personalized medicine means not only creating diagnostic tests for clinicians to use to treat their patients, but also creating bioinformatics tools and analysis software to help researchers get the necessary information they need to treat their patients, without asking them to wade through raw sequencing data that they might not be familiar with.
"Everything I talk about in terms of taking an individual patient and doing a lot of work to get to very nuanced, very specific, recommendations that go into a doctor's hands or a patient's hands, that is not a 2015 or a 2020 kind of conversation — that's a right-now conversation," Appistry's Meghji says. He compares the adoption of genomics in medicine to the movement in the 1990s that saw more people taking charge of their own finances with at-home stock trading platforms, and the early adopters of premium technologies like Apple's iPhone. Just as there was an appetite 15 years ago for people to handle their own finances more closely, there will be a similar clamoring for people to take charge of their own health, he says. And just as five years ago, enthusiastic early adopters of Apple's iPhone spent more money than most people thought was prudent, there are early adopters of genomic technology now who find it's worth it to pay for a whole-genome sequence out of pocket.
Similarly, Scripps' Topol says the drive toward widespread personalized medicine will not come from healthcare professionals, but from consumers. "This consumer-driven health revolution — if they go to their doctors and say, 'What should I do now?' — this is going to force and accelerate this new medicine," he says. Topol likens the coming consumer-driven wave to the late 1990s when drug companies were allowed to market their products directly to consumers. "We learned … that that's a highly effective thing to change medicine, is that patients come in and say, 'I want this medicine.' And it works really well," Topol says. Because personalized medicine is outside of the medical establishment's comfort zone, he adds, it may be up to the consumer to drive the adoption.
Industry also has a role to play in this personalized medicine paradigm, he adds. Pharmaceutical companies are starting to realize that the old way of one drug working for all patients is coming to an end, and that the future is in designing clinical trials that test more drugs on smaller numbers of patients for a more targeted, and more effective, impact on patients. "Instead of having a drug that you can put in the water supply and have a whole planet take the drug, how about a whole new look where you do a trial on 100 people because they have this particular allele or particular way to identify them that there's going to be overwhelming benefit? And then you get the drug registered, where instead of getting 15 percent improvement on some major endpoint, you have 90 percent improvement," Topol says.
That's already starting to happen. Topol points to Vertex Pharmaceuticals' recently approved cystic fibrosis drug, Kalydeco, as an example. The company tested the drug only on the 3 percent to 4 percent of patients with a specific mutation called G551D, and it had a "eureka effect" on those patients, he says. Because of the early success shown in clinical trials, FDA fast-tracked the drug and approved it in just three months.
"That's the model of the future — a lot more drugs, very small trials, quick registration, and none of this billion-dollar development stuff. We've already seen the beginning of this, and there will be hundreds more if the industry wakes up to the opportunity," Topol adds. If clinical trials are strategically designed with a specific and small subpopulation of patients in mind, there may be fewer failures than in the current model of drug testing where one compound is tested on a very heterogeneous population of patients who may not all react to it the same way.
The participation of government agencies and research communities is also integral to the progress of personalized medicine, Topol says. Government agencies are learning that they need to change the way they oversee and regulate the development of new tests and drugs, while researchers continue to make strides in genomics research. "We have enough knowledge now that isn't being used, but it's only going to get exponentially better over time," Topol says. "When should we start having genomics in the clinic? It's never going to perfect, but we have plenty to work with right now."
Similarly, Hood says, "The important part is personal medicine is here. It's increasing in incremental fashion. And the future is very bright."