Can you give an example?

If you take a molecule into man, and it has a very short half-life and can't be used once or twice a day, it may well not be a successful medicine in the long run. If the only way you can get it to twice a day is to give huge doses, and if it's an anticoagulant, then your peak plasma levels will be so high that you cause bleeding problems in order to get the person anticoagulated. But something that has a 12-hour half-life can be given once a day with a nice smooth curve.

Or say you have a molecule that's first in class. It's exciting and interesting, but it has a little toxicology issue and may not be tolerated. But you take it forward, you end up in long-term tox, and halfway through your clinical program you have to stop because of major preclinical active issues.

Back to reduction in time: Where are you actually able to shave that time?

Every single piece of the process, from soup to nuts. We have target time intervals for everything. For instance, in the watershed between Phase II and Phase III. It's not uncommon for a Phase II program to stop dosing patients. Then you have a period of time where you get the results in and analyze them. Then you have your internal management decision making process. Then you say, "Yeah, that looks good. We'll go into Phase III." Then you design your Phase III programs. Then you manufacture the material for Phase III. Then you go talk to FDA and EMEA. And suddenly, you've lost 12 to 18 months. But if you start by planning for success, you can actually knock huge amounts of time off that.

It does mean going at risk. You have to spend money on manufacturing extra material. You design studies you may never do. But if you've got smart ways of looking at your safety and efficacy early in your development program, then you may well be in a position where you can take that calculated risk in Phase II.

We established with our oncology group something called Mission Impossible, to get the first Phase I dosing within two days of the IND. That means you've got to line up your investigators.

We've made lots of little changes, like having standard protocols. We have a small number of units that we are going to do a lot of Phase I oncology trials with. We've got an overall agreement, so we don't spend six months with the lawyers haggling over the agreement. We've managed to start a recent study within two days of getting the IRB approval.

We've heard a lot about molecular imaging and new imaging techniques speeding up the process. Are you using these as well?

Absolutely, especially in areas like CNS and oncology, where you can very quickly get evidence of efficacy, sometimes even in Phase I. An example would be if you could image amyloid or a particular receptor in the brain. If in the Phase I studies you can show that the drug's getting to the right place or changing the amyloid, you know that you may have a chance of changing the disease, which otherwise you might take years of subjective assessments and clinical trials to achieve.

With this kind of technology—and with biomarkers and genomic markers—we can make go/no-go decisions much quicker than we did previously. If we can measure a marker of success, we can see whether or not we're achieving that level at the maximum tolerated dose in Phase I in an oncology setting. And if not, we stop.

So it really is happening in Phase I.

Six of our last seven oncology products all had decisions made in Phase I based on biomarkers or imaging.