Pharma CPR

The trial ran for 5 years with nearly 3,000 patients—what the team called “a demographically diverse group”—but it was halted early after failing to show significant gains in all-cause mortality, the primary study endpoint. “When the plug was pulled, all we knew was the primary endpoint hadn’t been met,” says Bristow. “The company was running low on funds and needed to conserve cash. Later on, as more information came in, it was clear that the plug probably should not have been pulled.”

When the team published their disappointing results in the New England Journal of Medicine in 2001, they noted that there was a significant survival benefit for non-black patients, an observation that they attributed to genetic differences in either the renin-angiotensin system that regulates blood pressure and water balance or the beta-adrenergic pathway targeted by the beta blocker. The human genome project had just been completed and the promise of personalized medicine was still just that—a promise. As it turned out, however, the BEST trial was the first ever heart failure trial that had taken DNA samples of all the enrolled patients.

Even so, it wasn’t until August 2003 that the data-coordinating center in Palo Alto faxed Bristow the pharmacogenomic results, revealing that a polymorphism in the beta-adrenergic receptor played a role in the response to the drug. “I don’t know if it was that night or the next day,” Bristow says, but it didn’t take him long to launch Arca Biopharma with the hope of licensing bucindolol and getting it approved as a personalized therapeutic.

With a longtime champion of personalized medicine, Francis Collins, now at the helm of the National Institutes of Health, many expect the approach to finally take center stage in the quest for new, targeted therapeutics with greater efficacy and fewer side effects. “If Merck could have known in advance who would have cardiovascular problems from Vioxx,” explains Edward Abrahams, executive director of the Personalized Medicine Coalition in Washington, D.C., “they may not have had to withdraw it from the market.” For Abrahams, there is no such thing as a failed drug. “‘Failed’ just means it failed in a percent of the population.”

Several drug candidates in areas ranging from cancer to schizophrenia that fell short of becoming blockbuster therapeutics are now finding niche applications with the development of novel genetic tests. Take the case of methotrexate, a chemotherapy drug developed in the 1940s that is now limited to treating leukemia, as newer drugs have proven more effective against solid tumors. A recent laboratory study in EMBO Molecular Medicine (1:323–37, 2009) found that the drug can destroy cells that contain a mutant MSH2 gene, which is responsible for about 5% of bowel cancer cases, triggering targeted clinical trials for this novel application.

Some companion genetic tests are also having unexpected impacts on the way currently approved drugs are used. Iloperidone (Fanapt), from Vanda Pharmaceuticals, was approved by the Food and Drug Administration in May to treat schizophrenia after Novartis shelved it for nearly a decade. Although the drug was ultimately okayed for all schizophrenics when a clinical trial proved it was superior to another currently approved therapeutic, Vanda had been pursuing a companion diagnostic from the beginning to identify patients with the genetic profile that responded best, along with those who may experience a potentially fatal side effect, which causes the heart to beat too fast. The company says it plans to release that companion diagnostic in the future.

Of course, bringing drugs back from the dead is not necessarily easier than starting over again. First, there’s the expense: According to Rina Wolf, a consultant whose clients include Axial Biotech and Zeissen, developing companion diagnostics can cost between $15 million and $100 million, depending on the complexity of the test. Second, companies must contend with the stigma associated with a drug that has a bad track record, particularly when it has been pulled from the market for safety reasons. For instance, Novartis’s painkiller lumiracoxib (Prexige), which is a cox-2 inhibitor like Vioxx, has never been approved in the United States and was pulled from foreign markets in 2007 after causing liver damage. While some outlets reported in August that Novartis had developed a diagnostic test to go along with the drug, which would use undisclosed biomarkers to identify patients at risk for liver side effects, the company is keeping mum on any definitive plans to revive the drug.

Ellson Chen, a former Celera scientist who founded Vita Genomics in Taiwan, once referred to his company’s work as a drug-rescue program, but today he prefers to call it “repositioning.” “When we first contacted pharmaceutical companies,” Chen says, “they didn’t like the name ‘rescue’ because it means something went wrong and that gets investors nervous.”

Rescuing drugs through personalized medicine is fraught with scientific challenges as well. In the late 1990s, Ruth March, director of personalized medicine at AstraZeneca, was working on ximelagatran (Exanta), the first drug in a new class of blood thinners. During Phase 3 clinical trials, the team discovered that over the long term, Exanta could raise liver enzymes in a small number of patients, a red flag for causing liver damage. In 2000, the company decided to try to develop a test to screen those patients, but it had to collect genetic samples retrospectively because they had not been obtained during the original clinical trials. Over the next 5 years, March’s team was successful in developing and publishing a biomarker, but it only predicted 50% of patients with the adverse events. That wasn’t good enough. AstraZeneca finally abandoned the drug in 2006 after discovering further safety issues, but March says the biomarker, called HLA-DRB1*07, has proven useful in predicting similar adverse events with other drugs (Pharmacogenomics J, 8:186–95, 2008).

Today, the company has procedures in place to identify personalized medicine candidates early on in development to avoid running inappropriate—and costly—clinical trials. “We spent 5 years trying to get these things into the early pipeline,” March says, “Now, we’re trying to not have too many surprises.”

For March, the last big “surprise” will probably be Iressa. When the lung cancer drug was first in clinical trials 10 years ago, the drug only proved effective in 10 to 15 percent of patients in 2004, a borderline case for FDA approval at the time. It would probably not have been approved today. Iressa is now restricted to patients already using it, however, diagnostic tests are available to identify patients in the general population with an EGFR mutation making them responsive to the drug. AstraZeneca would not confirm whether it will seek reapproval of the drug.

Although such diagnostics hold promise for patients, some companies may still be quivering at the possibility that new tests could eat into their profits. Over the summer, the FDA announced a labeling change for the colorectal cancer drugs cetuximab (Erbitux) and panitumumab (Vectibix) to reflect the fact that they do not work for people with mutations in the KRAS gene. The diagnostic firm DxS currently sells a KRAS mutation test in Europe and is now seeking approval from the FDA.

And earlier this year, a study indicated that the bestselling antiplatelet therapy clopidogrel (Plavix) is ineffective in preventing heart attack in 30 percent of patients (NEJM 360: 354–62, 2009). This deficiency was linked to a single gene variant, and following publication the FDA initially requested a labeling change that would recommend patients get tested prior to using the medication. However, after discussions with the drug’s maker, Sanofi-Aventis, the FDA has apparently softened on that stance. Even so, with Plavix nearing the end of its exclusivity period and a new Eli Lilly drug prasugrel (Effient) now on the market as a more effective alternative, a diagnostic test that pinpoints people who thrive on Plavix may be just what the company needs to keep the drug alive.

In the case of bucindolol, Bristow was able to get seed funding and Series A that amounted to $16 million, launching his third biotech company. In 2006, Bristow and his collaborators published a reanalysis of the BEST trial using genetic data in the Proceedings of the National Academy of Sciences (103:11288–93, 2006). “For 50% of the population, this seems to be the drug of choice,” he says. Unfortunately, the FDA does not agree: Arca’s first submission was not accepted in June, owing to questions over the statistical analysis and the fact that some original records no longer exist at study sites used in 1999.

Bristow, for his part, seems unfazed, and says that the agency had not reviewed supplementary data submitted in May and the company is now in discussions about whether additional clinical trials are necessary. “You have to line up a lot of dominoes to make this thing work,” he says.