TOPICS > Science

Extended Interview: Dr. Lee Hartwell Discusses Cancer Biomarker Research

March 28, 2007 at 4:10 PM EDT

SUSAN DENTZER: Thanks so much for talking with us. I mentioned to you we interviewed this woman yesterday, Kathy Lilleman. Diagnosed in in 2004 with stage three ovarian cancer. Two large tumors removed from her ovaries. Complete hysterectomy. On and off chemotherapy since. In your view, what is she an example of?

LEE HARTWELL: Well it’s an example of detecting cancer at a late stage. And what is really impressive– perhaps the most important statistic we know about cancer, is that if you detect it at an early stage, then standard treatments usually surgery and radiation will cure the disease. But if the disease is detected late– it’s very– unusual to cure the disease.

SUSAN DENTZER: We have, just as of this morning, the new statistics announced by the American cancer society showing that cancer mortality again dropped last year as it has been dropping for a number of years. What should that tell us about our success in treating cancer so far?

LEE HARTWELL: Well the drop in mortality from cancer is primarily due to reduced smoking. So it’s prevention. And that’s the best of all worlds, if we can prevent cancer. Which involves identifying its causes.

But in the absence of that, and for many cancers we don’t really know what the causes are. Then I think the next best thing we can do is to detect the cancer early when it can be cured.

SUSAN DENTZER: You gave a speech at ASCO going on three years ago that caused quite a hubbub. If you could just Paraphrase for us briefly what you said at that point and what point you were making about where we need to put our resources going forward in cancer research.

LEE HARTWELL: Well the model for how to cure this disease has been to try to make drugs that will kill the cancer cells but not normal cells. And tens of billions of dollars are spent every year by pharmaceutical companies and a good deal by academic research trying to find drugs that will cure late stage cancer.

And to a large extent, that has been a failure. There are a few good drugs. We can count them on one hand. And hopefully there will be many more. But it has not solved the problem. The mortality from cancer is not, although it has decreased a small amount, not very different than it was 30 years ago.

And the probability that a person who gets cancer will die of their disease has changed modesty. It was about 50 percent in the ’70s. And it’s about 36 percent now, the chance that a person will die of their disease within five years. So we have a long ways to go. And the point I made at ASCO was that there is a new opportunity to really implement a program to detect cancers earlier when we know they’re curable.

Early detection with biomarkers

SUSAN DENTZER: Would you say as a cover story in Fortune proclaimed several years ago, that the war on cancer has been a failure?

LEE HARTWELL: I wouldn't say it's been a failure. This is just a very difficult disease. And what we have accomplished in cancer is a tremendous amount of insight into the nature of the disease. We have been slower than we'd like in converting that knowledge to effective therapies.

But that will come. And that knowledge investment is going to be extremely important and necessary for making advances in the disease. I think we're sort of at an inflection point now. Where we know enough about the disease that we're going to be more effective in all areas of managing the disease, prevent, early detection, and therapy.

SUSAN DENTZER: So to turn to the early detection piece. And the degree to which we need to make more advances in that area. One of the key things that we want to talk about, obviously today is biomarkers. What are biomarkers? What's your best definition for what biomarkers are?

LEE HARTWELL: Well a biomarker is something that is very strongly correlated with the disease or the disease outcome. That is it's a source of information. It's telling you what's going on with a reasonably good degree of certainty in the body.

The kind of biomarkers that we're looking for are proteins in our blood. That are useful biomarkers like cholesterol, HDL and LDL for cholesterol. And but there are a whole variety of biomarkers. I mean even imaging, where you look for things in the body the image is a is a biomarker of the disease.

SUSAN DENTZER: So biomarkers are not the disease per se, but they're an indication of something going on in the body that is accompanying the disease process?

LEE HARTWELL: They're something you can measure that tells you what's going on. Right. And we need them at all stages of disease management. You know, we need-- we need to determine who's at risk for disease. We need to determine when people for-- have disease. We need to determine what kind of disease they have. Whether they are responding to the therapy they're being given. And whether the disease recurs later. And all of these things require diagnostic information. That for most diseases we don't have now.

SUSAN DENTZER: Would you feel comfortable in saying a biomarker could be anything from a gene to a protein-- that a gene codes for, but the ones you are particularly interested in here are the proteins. Are you comfortable saying it that way?

LEE HARTWELL: Yeah, yeah a-- a biomarker can be a gene or a protein or a-- an image. And-- in fact in the cancer area, the thing that is so encouraging is that because of technology that was developed a decade or two ago, we are using genes as effective biomarkers for many of these processes of disease management.

And that's what tells us that biomarkers are gonna be very powerful in all diseases. Cancer is unique because there are DNA gene markers in cancer. But-- we think there'll be even more information in proteins.

Biomarkers we know already

SUSAN DENTZER: Let's-- take a pause here and just talk about the fact that we do have some existing biomarkers that people would readily recognize in the case of-- ovarian cancer, CA 125, PSA, etcetera. So there are some biomarkers that we already know of. And are somewhat familiar with. What are those?

LEE HARTWELL: Well as you say, PSA for prostate cancer and CA 125 for ovarian cancer are markers that are best used to look for-- therapeutic response and disease recurrence. They are not very effective as screen markers on a population basis for the occurrence of new disease.

Biomarkers have to be very sensitive and very specific-- in order to be effective-- to detect early stage disease in a-- in a healthy population. So what we think we're going to need is-- to discover many, many more biomarkers. And we think there are thousands out there. And we know there are thousands of proteins that are involved in the cancer process.

And-- that there will be panels of proteins-- tens to hundreds that will be informative about the disease process. Probably no-- no single one will be effective.

SUSAN DENTZER: You-- let's just take an example. Go back to the example of PSA. You said that's not effect at-- as a screening mechanism. Why? Briefly.

LEE HARTWELL: Well the problem is that the-- the incidence of disease is really quite small. I mean when you know that over time a significant fraction of people will die of cancer and of different cancers. But if you're screening a population at any one point in time, the frequency of ovarian cancer, let's say, is very, very low.

And-- and so if the test gives any false positives, that is it-- it looks like someone is positive when they're not-- then that incurs all sort of-- interventions. Expensive medical treatments. And possibly treatments that increase morbidity-- that shouldn't be performed.

And so the-- the biomarkers have to perform very well to be useful in population based screening. And the best of all worlds is if, you know, what I think eventually we'll have, is we'll know more about people's risk. So we won't be screening everybody for every disease.

We'll be screening people for the disease they're most at risk for. Then it doesn't have to perform quite as well. And-- the biomarker test will be an indication-- that there's a much greater probability of disease. It won't be perfect in itself. And that will be followed by more effective-- molecular imaging techniques.

Where imaging is used to actually look at bio-chemical processes. And proteins and biomarkers that are within the body. That will be ultimately the more effective diagnosis. But the-- the whole imaging techniques are-- are much more expensive. So if we can stratify people into increased probability of disease, it will be much more effective for the costs.

The genetics of cancer

SUSAN DENTZER: What I'd like to do now is just for the benefit of our viewers is do sort of a genetics 101 of cancer. What we now understand happened in-- in-- really any cancer in terms of the changes in the gene and leading down to the production of proteins, etcetera. So in the simplest possible terms, what do we understand now about the genetic basis of cancer?

LEE HARTWELL: Well what's-- unique about cancer is that-- it's-- it's a disease that involves changes in genes as it progresses. And so-- you know, most of the information that is available-- to the public talks about-- the inherited genes that can increase the risk for cancer.

And-- the ones that we know about account for only relatively small percentage of the risk of cancer. But-- but that just is about increased risk. Okay?

When a cancer actually starts in the body there are further genetic changes that occur progressively. And-- cancer cells have many, many genetic changes in them. We don't really know how many. But they're-- incurring those-- the damage to their DNA and their genes at an increased rate.

And that's the fundamental basis of the cancer process. Is changes in the genes during the course of the disease. And-- if it weren't for that, cancer wouldn't occur. Because-- single genetic change doesn't induce cancer. It takes a whole series. But-- as I say, those changes we can use them as diagnostic markers for the type of cancer and its presence and its response in various things.

SUSAN DENTZER: And as those gene changes take place, those are the gene changes that essentially prompt the cells to uncontrolled growth. Correct?


SUSAN DENTZER: Could you say that?

LEE HARTWELL: Yeah, yeah, what the-- we-- we know about a whole lot of processes. And that's what I say about the knowledge that's accumulated about cancer. So-- cells normally-- are not dividing in an adult body.

And cancer cells divide more. So they-- getting signals or making their own signals to divide. That's part of the change. Normal cells, if they're in a situation where they shouldn't be dividing, commit suicide. And cancer cells overcome that process and they don't anymore. So that's another genetic change.

Normal cells will divide a certain number of times and then stop. There's a built in aging to normal cells. Cancer cells overcome that and become immortal. So these are all processes that have to mutate from a normal cell to produce a cancer cell. And we know of at least a half a dozen such processes.

SUSAN DENTZER: And does all of these-- genetic changes take place, the proteins that these genes are coding for are changing as well, correct?

LEE HARTWELL: Yes it can either be that-- the protein that does the functions, so the gene is the information to make the protein. The protein goes out, does the function. Some changes produce an excess of protein over the normal.

It's the same protein, but it's just a lot more of it. Other-- those are sort of growth promoting genes. Other genes get inactivated. So that a protein that's normally produced is not produced anymore. And those are the genes that normally put the brakes on cells and keep them from dividing. And so they get lost then the cells can divide more.

And-- then a third type of change is where the proteins produced is actually a different protein. And has some abnormal function.

SUSAN DENTZER: So when we think about looking for proteins as biomarkers, which of those various types of proteins produced in different ways or not produced are we-- are we looking for?

LEE HARTWELL: Well I think primarily what we're looking for is the presence of proteins-- in the blood or other body fluids that are not normally there. Or present in more-- greater concentration than are normally there.

So-- that's not necessarily a result of just the fact that the cancer produces a certain kind of protein. But cancer cells are very different than normal cells in that-- they're dying more even though they're dividing. They're dying more and releasing their contents.

They have a blood supply that's leaky. The blood vessels are leaky and proteins get out that don't normally get out. There are enzymes that chew proteins into funny pieces. And then there are the mutations that occur that make proteins different than normal. So all of those are possibilities for-- for biomarkers.

SUSAN DENTZER: It sounds like we're talking about a-- a practically infinite number of biomarkers.

LEE HARTWELL: Yeah the-- the-- the human genome-- encodes about 25,000 genes. And so each of our cells has the capability of producing these 25,000 different genes. Which you would think might each produce one protein.

But in fact there's ways of rearranging the information in the gene normally so that the typical gene actually produces several different varieties of that protein. And you know, probably ten to 100 different varieties.

And then there are modifications that occur to the protein to change it's activity later. So in fact it's-- it's-- there are-- there are tens to hundreds of different forms of each of the 25,000 different proteins. So there is a tremendous amount of-- different-- functional-- entities that are produced by cells-- that could provide diagnostic information.

SUSAN DENTZER: So does that-- all that add up to tens of thousands of potential biomarkers? Or even more?

LEE HARTWELL: Oh yeah I think tens of thousands at least. And-- and very possibly more. The problem is that the methods for discovering biomarkers have not been effective. And so we have a very small number of protein biomarkers that are actually used for diagnostic purposes.

We think that's gonna change dramatically over the next-- decade. Because-- first we have this catalogue of information from the human genome as to what could be produced. And that's-- that's important. We also have all this functional information about cancer and other diseases. So we know what proteins to look for as an informative.

And then thirdly, the instrumentation that's needed to look for these proteins has become much more sophisticated and functional in the last few years. So-- we're just beginning to be at the process where we can start-- finding these thousands of pieces of information. So at the present time it's a little hard to project-- how much information is there. But I think it's likely to be enormous.