The pharma fates have had their nasty sport with AstraZeneca of late. Since February 2006, the British drug giant has had to bury four billion-dollar babies: Galida, its oral PPAR antagonist for diabetes; neuroprotector-wannabe NXY-059, for stroke; the anticoagulant Exanta, for thrombosis; and lipid buster AGI-1067, for atherosclerosis. Each one was a risky, push-the-envelope drug; each one bit the dust in Phase III. To paraphrase Oscar Wilde, to lose one late-stage drug may be regarded as a misfortune; to lose four looks like carelessness.

As these embarrassments were playing out in public, John Patterson, tagged to head development in December 2004, was behind the scenes, putting AZ's entire R&D process and products through the pharma equivalent of AA's famous Step 4—"a searching and fearless moral inventory." As he candidly explains in this interview, every team in every therapy area had to make the case for not killing their darlings. It cannot have been a happy time at storied Stanhope Gate.

—John Patterson, AstraZeneca

Over the last two years, AstraZeneca has had four late-stage pipeline failures. Is there a thread that runs through all of these or were they very individual problems?

We believe that they were all very different but, obviously, we have analyzed some of the issues and tried to get to the bottom of them. Anything in late-stage development went into development 10 years ago, so there's an awful lot of history buried in some of these things. And given that length of time, there's also a changing outside world in terms of the perception of risk and the willingness to look at a risk/benefit ratio in a different way. Certainly that's been true of the PPARs [peroxisome proliferator-activated receptors] for diabetes [Galida]—you see what's been happening with Avandia.

In another instance [Exanta], we have the late appearance of liver-function changes, which just simply weren't predicted from the Phase II data. [AGI-107] was an option that we in-licensed, and [NXY-059] was a stroke compound. It's a bit like some of these Alzheimer's and COPD drugs—until you do a definitive Phase III study, you just simply don't know whether you have a drug or not.

So they really were very different, but that hasn't stopped us from making some changes. We're never going to stop attrition, but the challenge is to reduce it and to make sure that those things that are predictable are dealt with quickly and early.