“I've always found in our industry that our capacity to envision our future in a different state typically parallels our naivet”
In November 1998, Nancy Pelton was diagnosed with HER2-positive breast cancer, an aggressive form that strikes one in four women with the disease. A mutation in the HER2 gene in certain women causes excess production of a protein that fuels the proliferation of tumor cells. At the time, the 46-year-old tax accountant and mother of two from Bakersfield, California, had read about a new breast cancer drug from Genentech called Herceptin, but her doctor told her she would hopefully never need it because the drug was approved only for patients with a recurrence of the disease.
Instead, surgeons removed a small tumor from Pelton’s breast. She underwent four rounds of chemotherapy followed by seven weeks of radiation therapy and was then treated with the drug Taxol once every three weeks. By August 1999, Pelton was done with her treatments and walked away with a clean bill of health.
But this was not to last. By the summer of 2001, she began to feel pain in her armpits. Doctors discovered that the disease had returned. They removed 13 cancerous lymph nodes and gave her fresh rounds of chemotherapy and radiation. In February 2002, doctors added a new, more targeted weapon to the fight, Herceptin, the drug Pelton had heard about four years earlier. Unlike traditional chemotherapy drugs that indiscriminately kill fast-dividing cells in the body, Herceptin works by inhibiting the overproduction of a protein that fuels the tumor growth in HER2-positive breast cancer patients. More than five years later, she remains on the drug, going once a week for an hour-long intravenous infusion of Herceptin. Her breast cancer has remained in check.
“I don’t think the public knows enough about the types of cancer. It’s still too generic,” says Pelton. “If someone I know has breast cancer, or a friend comes to me about someone they know with breast cancer, I’ll say, ‘Okay, are they estrogen-positive or are they HER2-positive?’ They’ll say, ‘I don’t know. What’s that?’”
Indeed, as Pelton knows all too well, diseases can target different patients, and in different ways. Lucky for her and others, one-size-fits-all drugs and treatments are giving way to more tailored therapies. And Herceptin, in many ways, has come to exemplify this approach dubbed personalized medicine. That’s because it was one of the first drugs to require the use of diagnostics, in this case to determine if a patient was HER2-positive. Its use represents a new direction in therapeutics, which are increasingly targeted to interact with specific genes or molecular pathways involved in a disease. Now approaching its 10th anniversary, the drug last year won regulatory approval as a treatment for patients with early-stage HER2-positive breast cancer. The move has the potential for changing the course of treatment for the estimated 50,000 women diagnosed each year in the United States with HER2-positive breast cancer.
Born out of a chance meeting at the Denver airport between two researchers in 1986, Herceptin was an unlikely success for South San Francisco, California-based Genentech. A humanized monoclonal antibody, it was developed against a backdrop of spectacular industry failures around antibody technologies, which can target a specific biological process involved in a disease. It also faced great skepticism from within Genentech’s ranks because no such drugs had yet been approved and many people questioned whether the technology would ever lead to an approved product.
Late in the development of the drug, a big problem arose. With the company in the midst of a late-stage clinical trial, it became clear that the only hope of winning regulatory approval for Herceptin would require finding a partner to develop and commercialize a companion diagnostic similar to the home-brew assay the company was using in its trials. The diagnostic test would enable doctors to determine whether a breast cancer patient was over-expressing the HER2 gene. Turned down by a Roche subsidiary that had no interest in developing a diagnostic for what it saw as too small a market, a team from Genentech found themselves sitting over a meal of codfish in the dead of a Copenhagen winter negotiating with a Danish company a deal that was announced in 1998.
Since then, Herceptin has helped transform Genentech’s approach to drug development. Now a $1.2-billion-a-year product and growing, its success makes a persuasive case that personalized medicine is a promising development for Big Pharma. Targeting therapies to smaller populations of patients with a particular strain of a disease will not necessarily create niche markets that are too small for the industry to find profitable. What’s more, the emergence of personalized medicine reflects a movement away from defining diseases by their symptoms and locations within the body, and toward understanding them through their underlying genetic causes.
The notion of personalized medicine is seductive. A logical extension of the molecularization of science, proponents say personalized medicine will increase the efficiency of drug development, cut wasteful spending on therapies that don’t work for certain patients, and deliver more effective treatments to patients.
But the adoption of personalized medicine on a large scale faces significant barriers. The science of identifying appropriate biomarkers—detectable substances that might indicate the presence of disease—for diagnostics can be evasive. The business models to align pharmaceutical and diagnostic companies’ interests remain undefined. Moreover, regulatory framework and business models are not yet in place to allow companies to recoup their sizable R&D investment in products designed to reach smaller and smaller populations within a subset of a disease.
“I’ve always found in our industry that our capacity to envision our future in a different state typically parallels our naiveté about how difficult it is to get there,” says Paul Keckley, a health economist and executive director of the Deloitte Center for Health Solutions, a Washington, D.C.-based think tank that is part of Deloitte & Touche USA. “I don’t envision that the payment system is going to turn around overnight for personalized medicine. It’s going to require a substantial policy shift.”
Genentech’s Chief Medical Officer Hal Barron calls personalized medicine “an evolution, not a revolution.” In the 1940s and ’50s, pneumonia, he notes, was used to describe a wide range of lung infections. Doctors generally prescribed penicillin, which was effective about 30 percent of the time. But as diagnostic technologies improved, physicians learned to distinguish among variants of what turned out to be a very heterogeneous disease. Gram-negative bacteria caused some pneumonias, gram-positive bacteria caused others, and some were caused by viruses. Today, doctors can successfully treat most pneumonia by selecting the drug that matches the cause of the disease.
It is in this context that Barron argues that personalized medicine is not all that new. Instead, it’s an outgrowth of improved technology that allows clinical diagnosis to be enhanced through an understanding of the biology of disease. Today’s advances in basic science and rising healthcare costs create a perfect opportunity to take personalized medicine to the next level, believes Barron, who is also Genentech’s senior vice president of development. The mapping of the human genome has provided a deeper understanding of biology. Adding to interest in personalized medicine is the fact that drug industry R&D costs are rising, while the number of new approved molecules is falling, an indication that traditional drug development processes have become less efficient.
On a recent weekday, Barron flips a switch on a display projector in a conference room at Genentech’s headquarters to illustrate a point. The chart that pops up on screen is a familiar one within the industry. One set of bars shows a steady rise in pharmaceutical R&D spending. There is also an overlay trend line with a sweeping downward arch representing the decline in the number of new drugs approved by the U.S. Food and Drug Administration. For every drug that wins approval, he says, a dozen others fail in clinical development. The high cost of drug development—estimated as much as $1 billion per drug—is really the cost of failure: One success must not only pay for itself, it must also fund 12 dry holes.
Barron is the unemotional scientist when he switches to talk of personalized medicine, seeing it as a possible cure for what ails the industry as it promises to make clinical trials more efficient. “If you really want to make a significant dent in the cost of development, one way is to improve the success rate,” he says.
Herceptin remains the only example in the company’s roster of a product married to a diagnostic. But its success has spurred a big change at Genentech. The company now seeks to develop companion diagnostics for all of its drugs. One of the 2006 corporate goals set by Genentech’s leadership was for drug development groups to create a diagnostics plan for each new product. The idea was to use pre-clinical research to identify biomarkers.
Even if a companion diagnostic is not developed for a drug, using biomarkers has several benefits. It can accelerate drug development by helping weed out drugs that will be unsuccessful before large, expensive late-stage trials are launched. It can also help to refine the clinical trial population so that a drug is tested on people most likely to benefit. The approach can rescue a valuable targeted therapy from failure and control the size and cost of clinical trials.
Herceptin itself is a case in point. Genentech says if the drug’s late-stage clinical trial had used a general breast cancer population rather than HER2-positive patients, it would have had to enroll 2,000 patients over 10 years to show a statistically significant benefit. By narrowing the focus to a smaller population more likely to benefit, the trial included just 400 patients over 18 months. As a result, the drug was on the market helping to save lives a lot faster.
But finding biomarkers that actually tell researchers what they need to know can be tricky. A biomarker selected in preclinical research can turn out to be useless in the clinic. And tissue samples may be difficult to obtain for diagnostic use, which raises the challenge of finding blood tests or using imaging to obtain the desired information. Diagnostics has become part of Genentech’s “daily bread and butter,” says Sara Kenkare-Mitra, a senior director and head of Genentech’s development sciences. “Today, I don’t even look at diagnostics as something add-on. It’s woven into everything we do.”
But while Genentech may be culturally ahead of some of its big pharma brethren in its focus on biomarkers, it still has much to figure out beyond the science. While Kenkare-Mitra says that Genentech is working closely with a number of diagnostic companies, there are still “healthy tensions.” The biggest hurdle, she says, is the upfront payment diagnostic companies still demand as these companies try to determine the risk of developing a diagnostic to be paired with a drug that could be years away from approval—and may never win it at all. “We’re still working on these with a number of different companies to figure out what’s the best way to do this,” she says.
The problem, according to Ed Abrahams, executive director of the Personalized Medicine Coalition, a Washington, D.C.-based industry advocacy group, is that pharmaceutical companies are accustomed to expensive development cycles that extend 10 years or more. Diagnostic companies, meanwhile, have shallower pockets and seek to bring products to market in two years or less. “The developers of therapies aren’t so interested or willing to share the value with the diagnostic company, so there is some tension between them,” says Abrahams. “Those business models don’t comport very easily. That’s an issue.”
Such business issues are important sticking points. Diagnostic companies working alone have developed molecular diagnostics to determine the appropriateness or dosage of a therapy. But momentum for pairing diagnostics and therapeutics is building. Regulators worried about drug safety and governments and insurance companies concerned about rising healthcare costs want to see more drugs tied to diagnostics. This way, they hope, it will be clearer whether costly treatments are the right ones for the right patients. To do so will require closer relationships between drug and diagnostic makers.
A recent deal between pharmaceutical giant Pfizer and Monogram Biosciences is a sign that pharma companies and diagnostic outfits are forging more ties. Monogram is a small South San Francisco, California-based diagnostic maker that developed an assay to determine which of two co-receptors HIV uses to enter the CD4 cells of the immune system in an infected patient. Such information is important because some classes of HIV drugs block entry into the CXCR4 receptor, while Pfizer’s Selzentry, which the company began selling in September, is the first of a new class of drugs that block entry of the virus into the CCR5 receptor. The Monogram test identifies which patients can benefit from the Pfizer drug.
Pfizer, which began working with Monogram during the early clinical development of Selzentry, realized during late-stage development of the drug that its success depended in part on the availability of the Monogram assay. In May 2006, Pfizer struck a deal with Monogram, including a $25-million investment and a collaboration to make the test available globally.
“Frequently, you are going to find breakthroughs in smaller companies that don’t have the cash resources to be full partners to the pharma companies. There’s an imbalance between who can do what,” says William Young, CEO of Monogram. “You don’t know if there’s going to be a big market or a little market because of the risk of failure of the drug. Pharma will have to subsidize the development of these assays and potential commercialization for those reasons.”
But difficult relationships among pharmaceutical companies and diagnostic concerns may not be the biggest obstacle. More troublesome, perhaps, may be working out the regulatory questions of how to approve new molecular diagnostics, particularly when they are coupled with a therapeutic. Currently, the FDA has separate processes for approving diagnostics and drugs, with the one for drugs typically a lot more tedious. But complex molecular diagnostics will need to demonstrate their value and clinical utility in relation to a pharmaceutical treatment. The journey to approval may end up as trying as the ones for drugs that must undergo a set of clinical trials to prove their benefit.
An even bigger challenge, according to diagnostic makers, could be figuring out how to properly reimburse medical providers like doctors and hospitals who use the molecular diagnostics. The present system is based on paying for steps in the diagnostic processes—akin to charging for an automobile part by part—and not based on the value a patient receives.
“Many of the payers we have talked to and also some of the regulators have said, ‘Quite frankly, this field just got out in front of us and moved faster than anybody ever believed that it would,’” says Kim Popovitz, president of Genomic Health, a Redwood City, California-based maker of diagnostics.
In January 2004, Genomic Health introduced Oncotype DX, an assay that takes genetic information from tumor samples to determine whether patients with estrogen receptor-positive breast cancer that hasn’t spread would benefit from chemotherapy. The test can gauge both patients’ risk of breast cancer recurrence and the likelihood that they would respond to chemotherapy. About half of the women screened by the test are considered low-risk and not in need of chemotherapy. The tests, which assign a score based on the activity of 21 genes, cost $3,500, but can save an estimated $20,000 on the care of patients that would not benefit from chemotherapy.
The problem is, no insurance companies covered this new type of diagnostic. So Genomic Health had to decide whether to apply for coverage for the various steps in the Oncotype DX process that were already eligible for coverage or to make a case for new coverage for the whole package based on the distinct value of its assay. The test must be repeated three times for each of 21 genes. But the company felt that using existing billing guidelines that cover a single gene test and multiplying the steps to equal the process used in its test would still shortchange it. Genomic Health believed such an approach to reimbursement would fall far short of the value of the diagnostic and not allow it to recoup the more than $100 million it had invested in the development and launch of the test.
In 2004, Genomic Health finally got what it wanted. It took an experienced managed care team, several published studies in prestigious journals, and evidence which demonstrated that physician treatment decisions were changing because of the diagnostic 25 to 30 percent of the time. More than three and a half years after the product’s launch, the company now has acceptance from payers representing more than half of the covered population in the United States.
“We were maybe a little naive and stubborn,” says Popovitz, who argues that to develop these tests requires investment similar to a new drug and, as a result, deserves margins to those enjoyed by drugs. “Everybody said, ‘You’re crazy. You are never going to get CMS and payers to accept a $3,500 price for a diagnostic assay.’ We sat back and said, ‘We have to,’ because the future of the whole space and the industry is predicated upon the world accepting that these tools will deliver enough clinical value.”
Despite the demonstrated promise, many still remain skeptical about the future of personalized medicine. While Oncotype DX offers a striking example of the economic benefit a diagnostic can offer, economists say personalized medicine has yet to definitively show that it can deliver the financial payoff proponents claim awaits. As more high-priced diagnostics are introduced, skeptics say it’s unclear whether they will reduce wasteful spending on one-size-fits-all therapeutics that actually don’t benefit certain patients. The new approach might just end up shifting costs now going to drugs to diagnostics, say those who remain unsure that personalized medicine will be able to deliver savings.
It’s certainly true that for personalized medicine to take hold, the payers—both government and healthcare insurers—will have to sign on. And before insurers are willing to cover the cost of new diagnostics, they will need to be convinced not only of their clinical benefit but also of their economic utility. “There have actually been very few economic studies,” says Kathryn Phillips, professor of health economics and health services research at the University of California, San Francisco’s School of Pharmacy. She has reviewed the literature and says outcome studies on Herceptin and HER2, along with studies on gene expression profiling for breast cancer, are ambiguous. “What I tell people all the time, whether they are industry people or payers, is we really need to build the evidence base regarding diagnostics,” she says.
Though Genentech can point to Herceptin as a financial success, some suggest that it may prove to be the exception unless insurance payments for molecular diagnostics are resolved. The Deloitte center’s Keckley concedes that Herceptin has been able to find a broad enough population of patients to make it economically viable. But as targeted therapeutics move further down the line toward personal customization, smaller market segments will make it more difficult to recover the R&D expense invested in companion diagnostics, he says. The government may end up having to help subsidize R&D costs in some cases, he concludes.
But despite the many obstacles, personalized medicine continues to advance. The FDA’s approval in September of a new genetic test that will help physicians determine how a patient may respond to the widely prescribed blood-thinning drug warfarin, used to prevent potentially fatal clots in blood vessels, is another step forward. One-third of patients receiving warfarin metabolize it differently than expected and experience a higher risk of bleeding. Some of the unexpected response to the drug has been tied to the presence in patients of two gene variants.
“We are moving from one-size-fits-all trial and error to ‘who is the patient and what does the patient bring to the equation’ so therapies can be targeted and health outcomes maximized” says the Personalized Medicine Coalition’s Abrahams. “That’s where the science is pushing. That’s the excitement coming out of the clinical labs. This is not controversial. What is controversial is how fast this is going to happen and how to facilitate it happening.”
In the end, how quickly personalized medicine’s day comes depends on how long it takes makers of diagnostics, their pharmaceutical partners, and the payers for their products to agree on a new calculus. In any case, the result will have to be a system that factors in both the benefits to patients and the savings created by more effective therapies with fewer side effects, when determining the value of these tests.