Predicting late breast cancer recurrence with screening assays

Bookmark and Share
Published: 9 Dec 2017
Views: 1314
Prof Jack Cuzick - Wolfson Institute of Prevention Medicine, London, UK

Prof Jack Cuzick speaks with ecancer at SABCS 2017 about predicting late breast cancer recurrence with screening assays.

He describes how gene panels, such as the one discussed by Dr Jonathan Lancaster here, came to be following ATAC and TransATAC trials, and have incorporated a wider array of gene signatures which may indicate likelihood of relapse.

Prof Cuzick goes on to consider the future of screening in cancer prevention strategies patient survival.

For more on prediction of distant disease recurrence, watch our interview with Dr Ivana Sestak here.

The ATAC trial has been going for a long time, and we’ve published ten year follow-ups some years ago. One of the important aspects of that trial is that we set up a thing called TransATAC in which we were able to get the blocks from almost all the patients in the United Kingdom and a few from other countries and, as a consequence, it’s produced an incredible resource for looking at biomarker studies. We have most of the major genetic scores that we’ve actually evaluated on this. So it’s the only database in which we actually have direct comparisons of most of the major studies in terms of predicting prognosis based on tumour markers and genetic factors. It’s been an incredible resource and we’ve continued to mine that, focusing now more on what predicts late recurrence.

The panels that we’ve looked at give about the same predictive value for looking at who’s going to get a recurrence in the first five years. Our own simple one which we started with we called IHC4; it’s a simple immunohistochemical marker just looking at oestrogen receptor, progesterone receptor, HER2 positivity and Ki-67. It’s essentially a measure of cell proliferation. It works as well as almost all of the other ones – nothing works better than that in the first five years. They virtually all work the same because essentially it’s proliferation that’s the key driver in terms of recurrences in the first five years.

What’s been interesting is that in more recent work we’ve actually found that there are substantial differences across these different panels in terms of predicting recurrences in the second five years. Because breast cancer is a very long term disease it’s really important to be able to predict what the recurrence rate is going to be over the first ten years, not over the first five years. It’s one of the few solid cancers in which non-recurrence in the first five years is not an indicator of disease eradication. The recurrence rates in the second five years are almost as high as in the first five years. So knowing what’s going to happen in the second five years is very important, not only for deciding whether you need chemotherapy up front or whether you should consider extended endocrine therapy into the second five years.

Many of these things have been looking at that. In fact, we were trying to set up a pilot study which we’re struggling to make work to actually look more specifically at late recurrence treatments. Many of the studies have said after five years of endocrine treatment should we add two years or five years? In fact, if it’s tamoxifen there’s pretty clear evidence that extending treatment beyond five years is a good idea and it’s just a question of selecting which patients would benefit most. For aromatase inhibitors, because they’re more effective in the first five years, it may be that five years is enough. We’ve had studies of two years’ additional treatment versus five years’ additional treatment, which are the same, but they don’t address the fundamental question which is do you need any additional endocrine therapy after five years?
Our study was aimed at looking at if you’ve had just an AI and not tamoxifen should you stop at five years or continue for another three years? That study is struggling to recruit, so we may not be able to give you the answer to that, but I think it’s one of the fundamental questions now with AIs, what is the appropriate duration? There’s even some evidence, mostly based on the BIG 198 data which switched from an aromatase inhibitor to tamoxifen or vice versa at two years, that the patients that had two years of an AI followed by three years of tamoxifen did just as well as the patients that had five years of an AI. It raises the question, for low-risk patients, can we actually just treat them for a couple of years? Do we really need even five years of an AI? Those are important open questions.

Of the tests that we’ve looked at in ATAC there are two that have stood out as being particularly useful for predicting long-term recurrence. One is the NanoString test, which is called Prosigna, and the other one is the EPICLIN test, which is essentially now Myriad’s test. Both of those do seem to provide additional predictive value in the second five years, whereas many of the other ones show very little effect in terms of predicting late recurrence.

I was speaking with Jonathan Lancaster yesterday from Myriad and he was very enthusiastic about myRisk.

That’s a different thing. That’s looking at risk of developing breast cancer, so basically that’s an attempt to pull together genetic factors. We’ve been doing some work on that too. We have a thing that’s either called the IBIS model, or as it’s become more commonly known the Tyrer-Cuzick model for predicting who is at increased risk of breast cancer. Recently we’ve added SNP, or singular nucleotide polymorphism score, to that which is based on 88 genes which are much more prevalent than these ones in myRisk, but the risks individually are very small, like in the order of 5%-10% higher or lower risk. So individually they’re not important, but when you have a panel of 88 of them, it adds really very useful information for who is at risk of breast cancer. We think this is going to be important not only for identifying who is at high enough risk to consider preventive therapy with an aromatase inhibitor or tamoxifen, but also beginning to think about risk-adapted screening. It’s not a one size fits all for breast cancer screening. Some women need more screening, like every year, even MRI screening if they’re high-risk; some need a lot less screening, possibly even none. So if we can begin to identify who really is going to benefit from screening, give them more, and then also identify those for which breast cancer is not really a problem for them, they’re very low-risk, then we might be able to avoid screening. So there’s a lot of interest in using these risk models for who is at risk of developing breast cancer to guide screening and preventive strategies.

I suppose following that thread of risk for cancer development and then cancer recurrence and then just moving cancer into the lifetime, you get to your kind of annual check-up to see if there’s any change I guess.

Well you may not need it every year, that’s the whole issue. You certainly need a good check at baseline. In the UK annual check-ups are not so often done, but every two to three years, a mammogram – if it’s three years in the UK, it probably should be every two years for the average woman. Some, you’re going to be able to say, look, your risk is very low, you’ve got other things to worry about in terms of your health than breast cancer, which is a nice statement to be able to give, if we can give that reliably. That’s a very important and challenging area, you clearly don’t want to get it wrong and we will get it wrong sometimes as all these models are only models. We feel that the idea of equity is not that everyone should be screened the same but everyone should get the same benefits. That means women at higher risk should get screened more often and possibly with more aggressive methodologies like MRI and some at lower risk probably don’t need as much screening. The hope is we would be able to find a group that don’t need any screening. We’re not there yet, but we’re beginning to find of the order of 10%-20% of the population where the risk is low enough that five yearly screening might be enough.

Breast cancer is kind of leading the way but that’s going to be important for all the screening programmes. Colorectal cancer screening, if we can begin to say who is at high enough risk they need endoscopic screening maybe more regularly, who is not going to get colorectal cancer, a very low risk, and doesn’t really need much screening at all. It’s really an important area, we’re just beginning to delve into better risk assessment programmes. The reason that there’s interest is we can start doing something about it. If you start doing risk-adapted screening you need to be able to predict risk, so having a use of this information drives the desire to get better information.

And then hopefully it taps into cancer prevention strategies, looking at social causes.

Yes, absolutely. My vision is that we should think about renaming breast cancer screening as a breast cancer prevention programme. You go along, you get a risk assessment, you have a trained nurse who can talk to you about what your risks are, what you can do to lower your risk – dietary issues, exercise is important – and would be able to begin to tailor that more and more as we learn more about who is at risk of which particular kinds of cancers. I think it is clearly the way forward, and this idea that screening is the same for everybody and it’s a mass-production thing has got to change, we’ve got to begin to personalise that.