Science Watch® Interview With Lewis Cantley, Harvard Medical School
Science Watch Newsletter Interview: November/December 2010
Page: < 1 | 2 >
How does that link to this observation that obesity, as well as diabetes, is associated with an increased risk of cancer?
If you’re obese, and particularly if you’re obese at the level that you’re getting or are going to get type 2 diabetes, you have high levels of insulin and elevated levels of insulin-like growth factor—IGF-1, specifically. Both insulin and IGF-1 activate PI3K. IGF-1 tends to do it throughout development and growth. Insulin does it throughout life.
All cancer cells have IGF-1 receptors. So this begins to suggest, and we have data ourselves on this, that in some cancers insulin or IGF1 or both are helping to activate PI3K. In those cases, you don’t need to mutate PI3K because the signal turning it on is there without the necessity of the mutation. And we think this is contributing to the growth of cancers and that it might explain this correlation between obesity and diabetes and a subset of common human cancers.
Can you show that you can inhibit cancer growth by inhibiting the PI3K pathway?
"Anyone who says we’re going to cure cancer in 10 years or even 30 isn’t in touch with reality."
Well, the prediction is that if you have PI3K activated in cancer, it could be because you’ve lost PTEN, or it could be because of activation of PI3K, or maybe high levels of insulin and IGF-1 turning it on. The consequence is to expect glucose to be taken up at a much higher rate, just as it is in fat and muscle. That is true in 80% of cancers. And we’ve shown that if we just introduce mutant PI3K into lung epithelial cells of a mouse, the mouse will come down with lung cancer, and the cancer will be taking up glucose at an increased rate. And if we give the mouse a drug to turn off PI3K, the cells will not be taking up glucose and the tumors will regress.
Recently, you’ve also taken a lead role in what people are calling personalized cancer therapy. Tell us the idea behind this and how it ties in with the PI3K work.
It’s humbling to look at what’s going on in a cancer cell. Take breast cancer, for instance. Probably no two women on Earth have the same breast cancer. So even though PI3K itself is perhaps the most frequent single mutation you can find in all breast cancer, it’s still the case that only about 20% of all breast cancers have that mutation. But then there are these other ways to activate it. You can lose PTEN. Maybe 15 to 20% of all breast cancers have lost PTEN. Three or four other mutations can activate this pathway. So the pathway is probably activated in the majority of cancers, but not always by the same mechanism. This is a random toss of the dice.
If you look at cancer cells from different patients, you’ll find that 20% have one mutation and 15% have another, and another 12% have a third, until you get down to the very rare events—one in a hundred have this, one in a thousand have that. But everybody doesn’t have just one event, or one mutation.
They have, on average, five relevant events—five mutations in tumor suppressor genes or oncogenes that cause the cancer. You start doing the math and you see that the probability that any two people have the exact same combination of events is very rare. So if you need a different drug for every set of events, you’ll need a million different treatments for breast cancer alone. It’s not that bad, though, because many events do fall in common pathways, like PI3K. If you just get a drug that hits the pathway, then you have a chance that it’s working. But by knowing what events or mutations the tumor has, you can know how best to treat it.
How many different pathways are you talking about?
That’s a good question. The two pathways we’re most focused on in cancer are PI3K and the MAP kinase pathway. We call these "pathways," but they’re really networks. Most growth factors activate both of those pathways. For example, epidermal growth factor activates both. Insulin activates PI3K but barely tickles the MAP kinase pathway. Others tend to go more toward the MAP kinase and less to PI3K.
Ras is a good example to think about. Ninety percent of pancreatic cancers are due to a point mutation in Ras, and maybe 30% of colorectal cancers. At least 40% of lung cancers have mutations in KRAS. We’ve known that for 20 years. Unfortunately no one has been able to make a drug that blocks the function of Ras. That’s been considered an impossible project, although I’m skeptical about whether it really is impossible. I think someday we’ll target Ras, although no one knows how to do it yet.
Ras activates the MAP kinase pathway and also the PI3K pathway. So you can’t drug Ras, but you might be able to drug the two major pathways it turns on. We published a paper a year and a half ago showing that in a mouse model of lung cancer, driven by a mutation in Ras, we can cure that mouse by combining a drug that inhibits PI3K with one that in inhibits the MAP kinase pathway. That had never been done before. By figuring out the logic of what Ras does, we were able to identify what drug combination we should be using. It’s not just a random, "let’s try everything possible" approach, but rather a specific approach to block the pathways that this mutation is activating.
Has that experiment led to trials in humans?
It launched two different collaborative initiatives—one between Merck and AstraZeneca and one between Novartis and GlaxoSmithKline—to combine PI3K inhibitors with MAP Kinase inhibitors in Phase I clinical trials. We don’t know whether that combination will be tolerable in humans. It may be too toxic, but we’ll find out.
Is there a link between the MAP kinase pathway and PI3K, or are they independent?
"...cancer cells work more like fat cells, and they revert to this type of metabolism, using PI3K to drive glucose uptake, but it diverts it not just into making fat but also protein and DNA."
One of the major things the MAP kinase pathway does is control entry into the cell cycle, but there’s some redundancy between what the PI3K and the MAP kinase pathways do. Both can control cell growth, although the PI3K pathway tends to be more important. They’re parallel pathways, but they can be activated by common events. Then they merge and do common things downstream. That’s why if we inhibit one of them, the other can take over and still drive cell growth and survival. You can think of it as a network with a lot of crosstalk and redundancy.
There are obviously other pathways also important in cancer. People are now targeting the NOTCH pathway and the WNT pathway, which are important in development. These other pathways tend to be more involved in cell differentiation. Mutations in those pathways can trap cells in pre-differentiated states. They’ll stay in some intermediate stage and so continue to divide.
The point is that there’s a finite number of these so-called pathways. There are five to ten that we focus on, and we believe that if we had five really good, targeted drugs, drugs that could hit each one of those pathways cleanly and without off-target toxicities, and if we could figure out how to use them in combinations, we could go a long way toward treating and curing cancers.
How do you see this progressing over the five years or so, in terms of real drugs for real patients?
Targeted therapies like this are not going to be simple. In the end there will be logic to making certain combinations of drugs based on the mutations we see, but we don’t understand human biology enough to be competent at this. So it’s going to be somewhat empirical. I suspect that in five years, everyone diagnosed with a tumor at a good hospital will at least have every exon in their tumor sequenced. We’ll look at the deletions and amplifications and point mutations, and then these mutations will dictate the logic of the therapy.
A given patient, for example, has the WNT pathway mutated and PI3K mutated, so let’s give a WNT antagonist and a PI3K inhibitor and send him home with these pills and see what happens. These will be guesses, though, and often wrong, so we’ll need response biomarkers to tell us whether the drugs are doing what they’re supposed to be doing.
An example is FDG-Pet imaging, which looks for enhanced glucose uptake in tumors. In mice treated with PI3K inhibitors or MAP kinase inhibitors, we’ve found a very good correlation, with either a single drug or a drug combination, that if the compound turned off glucose uptake in the tumor, the tumor would shrink. We can detect the glucose uptake within 48 hours; the shrinking takes a few weeks.
So we can tell our patients to take these pills and then come back in a week and we’ll measure glucose uptake or some other biomarker. If that has changed for the better, keep taking the pills. If it hasn’t changed, then the pills aren’t doing much good and we have to go with our second-best guess. That would be another therapy. I think that’s how we’re going to be doing it in the future.
Last question: How far can this go in making all cancer a treatable disease?
One way to pose this question is whether I think that, say, 30 years from now, there will still be a significant number of cancers we can’t do anything for. And the answer to that is yes. Anyone who says we’re going to cure cancer in 10 years or even 30 isn’t in touch with reality.
Page: < 1 | 2 >