Dr. Grochow: So anyway, as you heard, Dr. Sausville's motto is 'Molecules to Man,' and CTEP's motto is, 'Theory to Therapy.' That is to say we get an agent that's finished its clinical testing, its clinical toxicology, and we're the ones who have to turn it into a real treatment.

So one question is, in terms of the molecular models issues, how is molecular development in cancer changing to address these issues that have been raised by agents that are directed at new targets. You've been touching on that a little bit throughout the morning. The first area that you heard about was actually the agent selection. Initally, agent selection was with mouse xenograph models and so on, but they more or less didn't mimic any real human tumors.

Under the new model, we'll be looking at what are commonly called now credentialed targets and molecular models that exhibit those targets -- tumors that are diagnosed in a new way based on what we know about what's going on and how that makes a cancer cell a cancer cell.

A credentialed target is -- I'll go over that a little bit more later -- but a target that really is essential in the cancer cell's being a cancer is a well-credentialed target. The best example of that -- I'll have a little bit later -- is in chronic myelogic leukemia

In terms of priority, the best drugs are the ones that killed the biggest fraction of cells in the preview test. That was a great agent.

Now, we ought to be inhibiting the growth of cells, because if you can take a patient who's got metastatic disease but only has three 1-centimeter, asymptomatic tumors in their lung, and you can make that tumor stop growing -- even if the patient has to take something every day for the rest of their lives, if you can make it stop growing at that point when they're asymptomatic you've suddenly turned a lethal disease into something that's more like diabetes or hypertension.

It's changed the way we're going to have to do trial design. Previous trial designs were imperical - we just kept giving bigger doses until you couldn't give bigger doses without threatening people's lives. Whereas now they're going to be hypothesis driven, driven on -- as you heard about from some of the models Ed was talking about - on being able to show that the agent really has affected its target in a human being the way it affected the target usefully in a mouse.

And in terms of what dose point you're going to use, it used to be just use the endpoint of toxicity, and now we're going to be looking at molecular effect. There are all kinds of extra complications to doing clinical trials when you're looking at molecular effect: When do you measure that; How often do you measure it; What do you measure? And in terms of deciding what the endpoint is, there are traditional trials where we were trying to look at patients, we did very ordinary doctor kinds of things -- we did histories and physical exams repeatedly while the patient was receiving the investigational drug. We did standard clinical laboratory testing and standard X-ray kinds of imaging. All of that's still going to have to be done in the new model, but in addition, we're going to have to be looking at very complex issues -- whether that's investigational imaging agents, whether that's biopsies looking for specific endpoints or collecting peripheral blood and extracting tumor cells out of the blood to see if they've been affected.