Amgen Looks to Biomarkers to Boost Its Batting Average in Developing New Drugs
Amgen had an epiphany of sorts a little more than four years ago. The world’s largest biotech company decided clinical trials had become too much of a crapshoot, and it needed to shake up its way of developing new drugs. This really isn’t unique to Amgen, it’s more like an industry emergency. Drug companies spent $44.5 billion on research and development last year, yet last year they managed to produce the fewest number of new medicines in 24 years, a big reason why companies are scrambling for a better way.
One of Amgen’s strategies is to gather samples from patients in early-stage clinical trials, and study them for biomarkers, or signature proteins, that can offer clues about whether the drug’s working (or not). Amgen now has a team of 240 people doing work in this area, which it calls medical sciences. Company scientists in Seattle with expertise in clinical immunology and computational biology are teaming up with molecular biology experts in Cambridge, MA, and people with other skills in the San Francisco Bay Area, as well as at headquarters in Thousand Oaks, CA.
Their job is to run a gauntlet of questions about drugs in the early phases of development, says Scott Patterson, Amgen’s executive director of medical sciences. Is the drug hitting its intended target on human cells? What affect is it really having on the biological pathway envisioned in the lab? Can the samples offer insight into the best possible dose? Perhaps most importantly, can the samples show Amgen how to stratify patients based on who’s likely to benefit from the drug, and who isn’t?
“This gives us greater confidence in our drugs as they move through the pipeline,” Patterson says. “You’ve got to have confidence to make the investments to take them through development.”
The biomarker analysis is being applied across the company’s pipeline, but the most visible example of how this is playing out for Amgen is with the colorectal cancer drug Vectibix. It was first approved in the U.S. in September 2006 and is designed to block the same protein on cells as ImClone Systems’ Erbitux, the pioneering colorectal cancer drug that was approved in February 2004.
Amgen paid a fortune—$2.2 billion—in 2006 to acquire Abgenix to get full control over Vectibix. Yet the drug has not emerged as the kind of $2 billion-a-year commercial success originally predicted by analysts. That’s partly because a study in March 2007 was halted when it showed Vectibix raised the risk of death in a clinical trial when used in tandem with Genentech’s Avastin, another colorectal cancer drug that works differently.
Instead of letting Vectibix fade into a treatment of last resort, Amgen went back to what it learned from tumor samples collected from earlier trials. When the company filtered patients based on whether they had a normal gene called KRAS, or a mutated form of the gene that amplifies tumor growth, they found a striking result. About 17 percent of patients with normal KRAS genes had tumors shrink after they took the drug, while none of those with mutated KRAS genes saw that benefit, according to research published in March in the Journal of Clinical Oncology. About 43 percent of patients have the mutated KRAS, and suddenly looked like they should avoid the drug, which is high-priced, and causes a significant rash as a side effect.
European regulators, who initially balked at the company’s application when the tumor shrinkage rate was about half that good when spread across the whole population, agreed with the company’s analysis. The EU approved the medicine in December for patients with the normal KRAS genes. U.K-based DsX makes a test that can determine a patient’s KRAS status, Patterson says.
The FDA hasn’t yet accepted the same subpopulation analysis in the U.S., preferring to wait for results from more trials, Patterson says. But the evidence is mounting for KRAS, particularly since Vectibix’s competitor, Erbitux, was shown to offer a similar type of benefit for patients with normal KRAS genes, according to a presentation in June at the American Society of Clinical Oncology.
What does it mean for Amgen as a company? Like a lot of people, I wonder if the company is really keen on finding out when its drug doesn’t work for more than 40 percent of patients, and that it therefore shouldn’t be given to them. It’s possible that by narrowing the pool of customers, but raising the overall odds of success among those who do take it, Amgen could sell even more of the drug. After all, Genentech’s breast cancer treatment, Herceptin, which is approved for about one-fourth of breast cancer patients with a certain gene mutation, still generated $1.3 billion in U.S. sales last year.
“We want our drug given to the patients who are most likely to benefit,” Patterson says.
The biomarker project has yielded some other valuable information. One small-molecule drug in development, which Patterson didn’t name, was supposed to block two different receptors on cells. An early-stage human study went fine, with no adverse events. However, when the team looked at biomarkers in samples, it found that the drug completely blocked one receptor as planned, but only partially blocked another—and was likely to fail in further clinical trials. It saved at least a year’s work and several million dollars, Patterson says. “It was a clean kill decision,” he says.
The method has also worked in the opposite direction, by pumping new life into a drug that researchers thought was probably a dud. A drug for metabolic diseases was thought to require twice-daily dosing, at possibly more than 100 milligrams, making it inconvenient for patients and costly to manufacture, Patterson says. After studying samples, company scientists found it was blocking its target enzyme much more efficiently than expected, and could be dosed once-daily at about 1 milligram, which is what is being tested in a Phase II trial, Patterson says.
The effort may not solve all of Amgen’s well-publicized woes with its top-selling anemia drugs, Aranesp and Epogen. But if nothing else, as Patterson says, the biomarker teams in Seattle and Cambridge will help give the corporate bosses some more confidence that the company’s R&D resources aren’t being thrown around in quite as big a crapshoot as in the past.