Read the Full Video Transcript

Laura Esserman: So what's in a name? There is, of course, inadvertent harm from the designation and the label of cancer. We all know that now cancer is a collection of heterogeneous diseases, which we did not know way back when. Now, Webster's definition is, when you look it up, this is what patients think. It's "A blight, a disease that, left untreated, will kill you." So if it doesn't do that, is that the right term for it? You have to think about that.

In cancer, like in prostate cancer, everyone was ecstatic because initially the localized cancer as a percentage of total cancer incidence went up, big success. We're finding those early-stage cancers. But the problem is, of course, that was because we were finding more low-grade cancers, and screening. That's really what has led to the increase in cancer incidence in breast cancer. We have a lower fraction of indolent cancer than you do, but it's still a problem, right? The fast-growing cancers aren't found by screenings. The intermediate types C are probably well done, help with screening, but B and C or A B probably won't come to clinical attention and are potentially harmed.

So how do you know who those people are? We have evidence that suggests the ultra-low, which are advocates. But anyway, so pre-screening, maybe 10 to 20% of all cancers. Post-screening, no, maybe 25, 30, there's controversy about how many those are. It's a problem only if it's not recognized at the time of diagnosis and treated accordingly, and people have to believe that it exists. But it's very hard when the word cancer is in it. But it's also a problem that these cancers are in the denominator for your high-risk cancers because it dilutes your ability to say what treatments work. So it becomes an opportunity to improve your approach for screening, prevention, and treatment by knowing who's at risk for what type of cancer, even who should be considered for biopsy.

So more than that, you need something more objective. You can't just say, "Oh, I think they exist." So I partnered up with Laura van't Veer, a molecular biologist who came up with the 70 gene test, which is MammoPrint. They looked at five years. Again, slower-growing cancers tend to recur late, but in the first five years, they tend to recur less frequently. You can see that these are the metastatic events in the first five years. This was their low and high, and that's with the old threshold. That is something that would tell you whether there was an opportunity to do less or an opportunity to do more.

So what we did was we said, "Okay, in the Netherlands, they actually don't overtreat. They set a threshold beyond so that there would be no metastatic events." We set a threshold on the MammoPrint on that same test and said, "Okay, here is an ultralow-risk threshold. Do those things exist and can we classify them?" Then in fact, we did. We actually went to the Stockholm trial. Again, you have to have someone with a 20-year follow-up. In breast cancer, that's poly-prostate. That's what you need, these endocrine tumors, node-negative cohort, treated with Tamoxifen versus not and said, "If we designated this ultralow and we actually profiled all of these tumors." Because they have 99% follow-up and 97% of the samples, we were able to do this. It was like a Herculean effort, amazing.

So what happened is there you have on the green is the ultralow. You can see the blue is the low, but not ultralow that over time they still have risk. There you have a really low. So having a quantitative molecular test of what is not cancer could be super, super helpful. You certainly don't want to find the precursor. So the DCIS of that, that 100% is not cancer, right? This was published in JAMA Oncology.

So then we use something called risk partitioning. So looking at classification regression trees, we can pick out what are the most important features for defining outcomes. What happens is it selects the ultralow as the first most important thing. So if you can get those people out of the denominator, then guess what? Actually size matters, and there you can see it. So that green, it's really very promising. Now, this was done in people with node-negative disease. We have a group that we're looking at with high risk.

But from a clinical standpoint, you think about it, if you have an indolent tumor, if it's yes, that means the metastatic risk is extremely low. That means it's not a harbinger of distant disease. That means the initial treatment is safe and you don't have to freak people out. You can avoid treating 90% of people. Those 10% of people who come back, you can do a simple excision, you're not going to put them in harm's way. On the other hand, if local recurrences are a harbinger of bad disease, like on the bottom, then you're going to be more aggressive and treat them upfront. This matters in the way you think about things.

But then even beyond that, what is the biology for treating ultralow-risk cancers? Are they different? Is it different? So we had those 44,000 arrays, and we found that there are different features. If we actually harmonize these and look according to the way the hallmarks of cancer, you can see that the green, this green thing here is the ultralow. This looks different from the rest of those. What's really interesting is that if you then look at normal, it looks very much like normal. So maybe there's a way of saying this is like normal.

I think what's important about this is that we could look actually at where does normal and malignant begin? You can look across the TCGA, so we looked at breast and we looked at prostate, turns out to be a very important... So this is actually the prostate group. So you can see the blue is the ultralow. It's not perfect. But again, that's a super high-risk group of patients. So there may be molecular. So if we had common features that were farmed for thyroid, prostate, breast, and some screening lung cancers, this could be something that could unify us and help every field say what today should not be called cancer.

So when someone says, "Oh good, only one in seven are overdiagnosed." Well, in breast cancer, that's still 70,000 people getting treatments that they probably don't need. So I would say as a group to try and look to the science to say, how can we define some of these cancers that might cross all of these cancers to redefine today for this cancer with what we know?