Wednesday, August 8, 2012

How A Failed Alzheimer's Drug Illustrates The Drug Industry's Gambling Problem

by Michael Waldholz Forbes Contributor
On Monday, Pfizer, Johnson & Johnson, and Elan Pharmaceuticals announced that the Alzheimer’s drug they were developing together had yielded such a bad result in a clinical trial that they were stopping all further work on it – dashing hopes for the 5 million Americans suffering from Alzheimer’s and becoming the latest piece of evidence of the drug industry’s strange gambling problem.

Wall Street expectations weren’t high, with only 23% of investors expecting success after an earlier study also came up neutral, according to a poll by ISI Group analyst Mark Schoenebaum.

But some analysts were expecting at least a glimmer of hope, or, in the word’s of JPMorgan’s Christopher Schott on July 16, “not a black and white outcome.” But the outcome couldn’t have been starker.

There is “no path forward” for the intravenuous (or I.V.) version of bapineuzumab in mild-to-moderate Alzheimer’s disease, says Pfizer spokeswoman Victoria Davis. (Earlier-stage studies of a newer version that is injected under the skin are continuing.) “Representatives noted that the secondary endpoints results were not promising, and there were no positive signals from existing subset analyses thus far,” wrote Barclay’s analyst C. Anthony Butler in a note to his clients. “We interpret yesterday’s news as a definitive end to bapineuzumab, and any glimmer of hope is all but extinguished.” Pfizer does note that “These data, and the subgroup and biomarker analyses underway, will further inform our understanding of this complex disease and advance research in this field.”

What’s really surprising about bapineuzumab’s failure is that expectations were as high as they were. And what’s shocking is that Johnson & Johnson spent more than a $1 billion to invest in Elan and get one-quarter of the drug, and that Pfizer (or, rather, Wyeth, which Pfizer bought) and Elan chose to push the drug into broad clinical trials despite a single, uncomfortable fact, which was previously noted by Forbes contributors David Maris and Nate Sadeghi:

Bapineuzumab failed in earlier studies, and Wyeth and Elan decided to plunge forward anyway.

In an earlier, mid-stage study, the drug failed to help patients by a statistically significant margin. The drug didn’t work much at all in patients who carried a genetic mutation in a gene called APOE4, which increases the risk of Alzheimer’s and makes the disease worse. But it seemed to have a profound effect in patients who lacked the APOE4 mutation. The problem is that, in a clinical trial that does not meet its main goal, this kind of subgroup analysis (or, less charitably, data-dredging) is very likely to be due to chance.
his is not Monday morning quarterbacking. Plenty of people said this when the results were released in 2008. British Alzheimer’s researcher John Hardy, who came up with much of the basic science behind “bapi,” told Forbes at the time:

“Maybe, maybe, maybe [it works], but I am not convinced. I’m worried they are just looking at noise.”

Wyeth and Elan had plenty of options aside from just plunging ahead. They could have waited for new diagnostic tests to help figure out if the drug was working. They could have done a series of smaller studies to test whether the effects that were being seen were really there. Or, at the very least, they could have conducted late-stage studies only in patients who did not have the harmful APOE4 mutation. (Regulators could have been involved in the decision to test the drug on everybody; that doesn’t make the decision smarter.)

The logic behind going forward probably went something like this: Alzheimer’s is one of the world’s biggest health problems and any drug that can impact it would be simply huge. Even if bapi were not that effective, it could have generated $5 billion in annual sales, easy. It would be crazy not to try, right?

No. The crazy thing was trying. That logic would make perfect sense if it were not for the most difficult fact in drug development: the drug industry’s failure rate and the amount of research and development money spent for each success are both rising at an exponential rate. The number of new medicines invented per billion dollars spent is halved every 9 years.

Bernstein Research analyst Jack Scannell has dubbed this “Eroom’s Law,” not in a Wall Street research note but in Nature Reviews Drug Discovery. He’s also warned his clients that it may not be worth trying to fix drug companies with below-average R&D divisions. Even if you make them as good as the average company, R&D will still lose more money than it makes the way things are going.

Whatever a company’s scientists think the likelihood of a project failing is, it’s probably even worse than they think. With the current state of drug research, bapi wasn’t just a Hail Mary pass, but a Hail Mary pass thrown in gusty gale force wind at a randomly moving target.

What’s worse, these long-shot bets are getting more and more expensive. J&J is taking a $300 million to $400 million charge in the third quarter for discontinuing its share of the trials. Even at a scale of hundreds of millions of dollars, the money could be spent differently. Optimer Pharmaceuticals spent less than $200 million in R&D from the time it went public until the time its first drug, Dificid, was approved.

But bapi isn’t so bad in one regard. Increasingly companies are willing to make bets of billions of dollars almost completely blind, as clinical trials get larger. This goes for Eli Lilly’s Effient and AstraZeneca’s Brilinta, blood thinners that went through huge clinical trials and then failed to garner significant sales. It goes for all the Factor X blood thinners being developed by Pfizer, Bristol-Myers Squibb, J&J, and Bayer whose success will hinge on their bleeding side effects. Darapladib, a pill to fight artery inflammation being developed by GlaxoSmithKline that was one of the reasons for the company’s purchase of Human Genome Sciences, failed its main goals in an artery imaging study. But large studies of the compound went forward anyway.

Doing large studies of new medicines is a good thing – it’s the best way to truly establish both safety and efficacy. But that doesn’t mean, at a time when every drug company is struggling to improve its success rates, that we should be doing as many of these studies as possible. Sure, as former Pfizer research chief John LaMattina writes, drug development is very hard. This is a business where more than one in twenty projects fail. But that’s all the reason not to fly blind.

Even drugs that have built a strong track record in earlier trials fail in the late stages of development. There is no reason to push forward with medicines that have little chance for success.

Industry executives are probably thinking back to the success of drugs such as Plavix and statins like Lipitor and Zocor. But the big trials that proved the worth of those drugs came after the medicines were cleared and marketed, and there was a huge amount of data supporting doing big tests to prove the medicines extended lives. Certainly, sometimes doing a study to prove a drug’s benefit is worth it – but you should be able to prove that the odds of success are high in earlier, smaller, cheaper studies. Industry executives should think back to an earlier precedent: the move in cancer trials pushed by Genentech more than a decade ago to make mid-stage trials mirror approval studies, with control groups and patients who are randomly assigned either to get the drug or not. The result is fewer drugs progressing into large studies without evidence they work.

There is already a good example in Alzheimer’s, with a drug a lot like bapi. Some bapineuzumab skeptics don’t believe that the basic theory behind bapi, that plaques of a chemical are a cause of Alzheimer’s, are true. Amyloid believers think that the drugs just need to be tested earlier in the process. But it’s hard to know the truth, because in Alzheimer’s studies there is no way to even know if all the patients have Alzheimer’s or to measure whether a drug is removing amyloid plaques from their brains.

Genentech, now part of Roche, has come up with one elegant test. It’s testing its own antibody against amyloid, developed by its partner AC Immune, in a study in Colombia in a population of people who are sadly condemned by a rare genetic mutation to develop Alzheimer’s. “If with this trial we show that we’re removing or preventing a beta from accumulating in the brain, and the cognitive decline still happens, I’m going to wonder if maybe the amyloid beta hypothesis is true after all,” said Richard Scheller, executive vice president of research at Genentech. Some analysts think that Pfizer and J&J are keeping an under-the-skin version of bapi in trials like this to show a way to do studies earlier in Alzheimer’s.

Luckily, the strong trend in the industry is to insist, increasingly, on these kinds of proof of concept studies. It is the long-held research strategy at Novartis, and is very much what Elias Zerhouni, the head of research at Sanofi, has in mind when he talks about his plan to improve long-term success rates. If company executives can stick to this discipline, we’ll all benefit. Like good card players, good drug developers need to sit out many more hands than they play.


No comments:

Post a Comment