"Pharmacogenomics is a new field, but we intend to do all we can to use it to promote the development of medicines," said FDA commissioner Mark McClellan. "By providing practical guidance on how to turn the explosion of pharmacogenomic information into real evidence on new drugs, we are taking an important step toward that goal." Added the FDA's Janet Woodcock, "It gives us the possibility of only treating the people who stand to benefit, and weeding out the people who could have serious reactions."
Martin Madaus, president and CEO of Roche Diagnostics, enthusiastically embraces the role of releasing a drug with an accompanying diagnostic. "Pharmacogenomics has the potential to say a drug is 100 percent effective or 0 percent effective," he told delegates at a recent Frost & Sullivan conference. "It's a paradigm change, it's threatening — it's a huge opportunity ... I'm sure it will become the standard of care, and it will become reimbursed."
But the pharmaceutical industry is faced with a major fork in the road toward implementing pharmacogenomics. The dilemma is neatly summarized by Sir Tom McKillop, the CEO of AstraZeneca. In discussing the controversial drug Iressa, McKillop recently acknowledged: "There are 50 percent of patients with lung cancer who do not get any benefit from Iressa ... but then there is a small group, maybe 10 to 20 percent of patients [who] get an almost miraculous response, and their whole life can be transformed and extended for years."
The pharmaceutical industry is perceived to be clinging obstinately to the old blockbuster drug model, hoping that a handful of billion-dollar molecules will not only validate multibillion research budgets but also sustain high rates of annual growth. Few pharma executives will lose sleep over thousands of patients paying big bucks for ineffective drugs that amount to little more than a placebo.
But the pool of patients that register for clinical trials is not homogenous: Individuals carry millions of single nucleotide polymorphisms (SNPs) in their DNA. Some of these variations preordain the pathogenesis of disease, whereas others shape individual response to administered drugs. How marvelous it would be if McKillop's "miraculous" responders could be identified in advance, so that the remainder did not incur the needless expense and false hope of a drug that had no hope of working, given their body chemistry. More important still, what about identifying patients who might suffer adverse reactions, or even die, after taking approved drugs?
The introduction of 6-mercaptopurine (6MP) and other thiopurine drugs several decades ago dramatically improved the survival rates of children with acute lymphoblastic leukemia (ALL) from less than 5 percent to about 80 percent today. But about 1 in 300 people lack an enzyme called thiopurine methyltransferase (TPMT) that inactivates 6MP: Chemotherapy in these patients can cause — and has caused — dangerous accumulation of thiopurines, leading to potentially fatal hematological toxicity.
While some advocates have lobbied the FDA for mandatory genetic tests before prescribing 6MP, the medical establishment remains deeply suspicious, citing various problems, including cost (up to $300 per test), risk frequency, dosing levels, and physician education. A special FDA review panel this past summer declined to recommend mandatory testing, although the two clinical centers where the molecular pharmacogenetics of TPMT was unraveled — the Mayo Clinic and St. Jude Children's Research Hospital — do genotype all of their ALL patients as standard procedure prior to chemotherapy.
|FDA guidelines are a welcome first step in what promises to be a contentious debate.
In collaboration with Affymetrix, Roche Diagnostics has developed a cytochrome P450 DNA chip that will reveal genetic variations in a pair of genes that influence the metabolism of about 25 percent of currently prescribed drugs. But the FDA has told Roche to stop selling the test until it can evaluate more data. So far, Genentech's Herceptin is the only drug that is prescribed in conjunction with a diagnostic test, one that reveals which third of the 180,000 women diagnosed with breast cancer annually express the Her2 receptor and thus stand to benefit from the drug. Despite restricting the eligible patient population, Herceptin has been a major success, earning Genentech more than $200 million in the first half of 2003.
Many drug companies, locked into the traditional blockbuster model, will view the FDA's suggested guidelines (see page 1) with suspicion. They should not. The FDA requested voluntary submission of research information to help it learn as the field evolves, but pledged not to use such information for regulatory decisions.
But pharmacogenomics can help Big Pharma, and Gleevec shows why. Although chronic myelogenous leukemia accounts for fewer than 5,000 new cancer cases in the United States each year, the ability to stratify patients with the telltale Philadelphia chromosome during clinical trials enriched the pool of responsive patients, accentuated the drug's positive effects, and expedited approval. The drug earned more than $500 million globally in the first half of 2003 alone.
The issuance of draft FDA guidelines is a welcome first step in what promises to be a contentious debate to integrate pharmacogenomics into routine medical practice. The promise of genomic medicine assures us that countless more genetic associations with disease and drug toxicity will be uncovered in the next few years. It would be unconscionable not to use this information in the best interests of patient care.
Speaking of personalized medicine, Roche Diagnostics' Madaus said, "The train has left the station." And the patients are at the controls.
Kevin Davies, Ph.D.
PHOTO BY WEBB CHAPPELL