Siddhartha Mukherjee: So we talked a little bit about the cancers arising. Dr. Schwartzentruber, tell me a little bit about do cancers ever go away on their own?

Doug Schwartzentruber: Some cancers do spontaneously regress. Probably some of the best examples are melanoma, a type of skin cancer that is uncommon, but can be very deadly because it can spread to other parts of the body, but we talk about a certain number of these cancers that have spread and we never identify the original melanoma in the first place and that is probably the best example for a spontaneous regression, which probably is mediated by our immune system. We assume that the immune system at least at that time point, was powerful enough to get that melanoma on the skin surface to go away, yet, it had already spread its cancer cells throughout the body and those then grow and the immune system at that time point is not able to handle it.

Lewis Cantley: So I think there are other examples. For example, we mentioned the heart. There actually are cases of rare heart cancers that occur prior to birth. They’re actually picked as defects in heart development and those very often spontaneously go away, so they’re probably driven by growth factors that are present at that time of development of the heart and in combination with mutations in fact, in cases of familial hematoma syndrome, but they almost invariably self resolve. You don’t have to treat them. NF1 is another hematoma syndrome where the same thing can happen in brain cancers, so that is probably maybe the immune system is also involved, but in this case it’s probably the stage of development in the environment at that stage.

Harold Varmus: But your question raises a much broader issue. It is true there are quite a number of very dramatic cases in which cancers seem to have gone away or arrested and regressed. Neuroblastoma for example, sometimes we see these in stage four, very advanced disease, but a bigger question that we really need to understand is why very early cancers sometimes progress and sometimes don’t and as our detection...

Siddhartha Mukherjee: Give us an example of this.

Dr. Harold Varmus: Well for example it’s very likely that early breast lesions that are called introductoral carcinomas called in situ lesions frequently will probably never progress and we need ways to be able to predict which will and which won’t progress. These are very common. Men who develop prostate cancer frequently would never have advanced disease, but they are found to have early stage prostate cancer. That presents an incredible clinical dilemma for people. There are many potential ways to treat it, almost all of them with severe or very significant side effects and we don’t have good ways to predict these. Very recently the Cancer Institute announced the so called low dose helical CT scan. It can pick up early lung cancers and can reduce mortality from lung cancer by about 20%, but a large number of the early lesions that are seen probably never would go on to being lethal cancer and being able to discriminate between those that will cause trouble and those that won’t make a huge difference in cancer therapy.

Siddhartha Mukherjee: And how might one go about discriminating between such lesions without knowing the future as it were in the present moment?

Harold Varmus: Well the simplistic way to think about that is and I'm not sure this is the way it will be worked out, is to be able to take just a few cells from those early lesions and examine them genetically or for other kinds of marks on the DNA that would predict whether or not this is some- this is a lesion which might or an early stage growth that might never be able to progress, but it is also possible that every early tumor of that kind has some probability of expanding and invading and growing to become a medical problem, so getting that right will obviously be crucial because it’s very difficult to say when you’ve diagnosed something that is an early stage tumor that it won’t progress. Take the example of colonoscopy. Many of us undergo colonoscopy. It’s the right thing to do. We find frequent so called adenomas. These are growths that haven’t yet- which cells don’t yet behave in this antisocial invasion fashion, but since it’s benign to take it away we remove the polyps, these early stage growths and we are probably in many cases preventing cancer, but probably not in all cases.

Deborah Schrag: This is enormously challenging from a societal perspective because what Harold is talking about is we have two problems. We have a problem of under diagnosis. We know that far too many patients with curable cancers if they are caught at earlier stage, colon cancer, breast cancer, cervical cancer, these cancers are still detected late when they are less amenable to curative treatments. That is still a problem in this country even though we have some techniques, not perfect ones, but some techniques to help identify these cancers early, but at the same time as we still have under diagnosis we have this new problem, which is exploding at the same time of over diagnosis. Now when we have an under diagnosis and an over diagnosis, particularly for the same cancer like breast cancer we create enormous confusion on the part of the public. It is tough to get these messages across. Not only the public, physicians, clinicians because we don’t have the techniques that Harold is talking about to figure out this is the one you have to pay attention to and we’ve got to deal with and this one we can let it sit and cancer, just the word cancer is still so terrifying that at the individual level people and physicians feel compelled to act and that is going to be a big challenge for us. 

Lewis Cantley: Yeah, I completely agree and I think particularly in the case of prostate cancer this is a huge problem. I think we all agree that we over treat prostate cancer in this country compared to what you see in Europe with similar outcomes and so but I think this is where the human or the Cancer Genome Project or at least better biomarkers for looking at mutations that we already know occur in these diseases I think hopefully within the next few years every single patient who is diagnosed with prostate cancer those biopsies will be characterized and then the patients followed, hopefully watchful waiting and we’ll ultimately get a correlation between what mutational events predict a rather dormant disease as opposed to a very aggressive disease.

Harold Varmus: I think the way to think about this is, is as a probability argument. That is we can make a risk assessment, but I think it’s important people understand that this is not a binary decision. It’s not as though lung cancer is definitely going to cause trouble and one early tumor is never going to cause trouble and that is where people quite understandably have difficulty even though cancer I think is a disorder that is less frightening than it used to be and we have been able to create an environment which we can have a very rational discussion about one’s odds, but since we’re not going to know the answer and there is going to be a certain level of probability here. It’s going to be difficult even if we get this additional information, which is already accumulating.

Lewis Cantley: So what are you happy with, the 1% odds of getting it, that means we’ll take it out or maybe 5% odds and you say no, leave it in. Different people are willing to take different risks.