15th Congress of the European Society of Surgical Oncology (ESSO), 15–17 September 2010, Bordeaux, France

Interview with Dr Pat Whitworth (Vanderbilt University, Nashville, Tennessee, USA)

Oncotype DX classifier for breast cancer

Pat, you’ve just made the journey across the pond. This meeting here in Bordeaux, are you excited to be here?

Very much so. The Society of Surgical Oncology in the US has been a favourite meeting of mine for many years and the European Society of Surgical Oncology is something I’ve always wanted to come to so it’s a pleasure to be here.

Can I ask you how is it different from the American variety?

I don’t think that it is different, really, other than it’s European so it has a little bit longer heritage than the SSO, at least in terms of surgical treatment. But we are very much partners and we often see many of our colleagues from the European society in the US and we love to come here for the ESSO meeting as well.

And, of course, it’s wonderful to cross-fertilize between American and European techniques. Now we’re talking today about genomic classifiers, it’s a topic that’s very dear to your heart and specifically you’ve been working on the oncotype DX classifier. Can you tell me, first of all, why do you think this was needed?

Surgeons for years have seen our patients treated, we treat them surgically and then there’s a question of should they have systemic adjuvant chemotherapy. For many years all we’ve had have been the traditional clinical parameters which, given what we have known to date and from the studies from the NSABP and other organisations, led to treatment of most patients with chemotherapy. This was in spite of the fact that most patients don’t benefit, only a small fraction benefit. So we’ve always looked for a way to figure out which patients are those, which patients are the patients that have a benefit as opposed to the large number of patients we’re treating who don’t?

So looking at tumour stage, size, grade, things like that? Are you saying that there were errors made or it simply wasn’t possible to make the judgement?

That’s right. The NSABP, when they published the results of the B20 study many years ago, and that was node negative patients, ER positive, said, “We see this benefit and we cannot find a group of patients that doesn’t benefit.” So the National Cancer Institute had a consensus conference and, in fact, when that consensus conference was repeated the same recommendation came out and it was basically if the tumour is larger than 1cm, there’s enough benefit to say we should treat with systemic adjuvant chemotherapy, even though that benefit wasn’t huge, we couldn’t tell which patient benefits and which doesn’t.

Although on classical ways of looking at those patients, you could get it wrong so that there could be over-treatment or under-treatment.

That’s right. On the other end of the spectrum, many times in a long practice we’ve seen patients who had a 6mm or 7mm tumour who recurred years later. And to our dismay and the patients’ dismay and our shock, because given our traditional approach we wouldn’t have treated that patient, we’d say she has a great prognosis. But then again 90% is not 100%.

So how do gene arrays change all of this?

Gene arrays allow us to look at the biology of the individual tumour. It’s really seeing something that we couldn’t see otherwise. You can take two patients who look essentially identical, maybe it’s a 58 year old patient with a 13mm tumour, ER positive, node negative, and this patient has a 30% risk of distant metastases at ten years and this patient has a 4% risk. This patient has a huge benefit from systemic adjuvant chemotherapy and this patient has none but toxicity.

Could you tell me about the oncotype DX test – how is it tested, what does it consist of and what does it exactly do?

It came from the idea that if certain genes are expressed, that may lead to a patient having a better or a worse prognosis. And so the search started with 250 genes that were considered to be cancer related genes and looking at patients from three large studies, a total of 447 patients. A number of these 250 genes were tested and, in the end, sixteen were found to correlate with either a better or a worse outcome.

And you’ve got some reference genes in there too?

Yes, in addition to those sixteen genes there are five reference genes making a 21 gene recurrence score signature. So once it was developed, once it seemed that this was the right combination of genes and the algorithm used to generate a score that would either say low risk or high risk, it was time to test that. So if we took that score and tested it on those same 447 patients it would look great, obviously that’s where it came from, so the key is to test that in a group of independent patients and that was the NSABP B14 study. Tissues were saved and this was a very, very wise decision at the time, I don’t know if we would be funding that today but it was funded at the time so there were enough patients  in that study that had tissue archived. This is formal and fixed tissue where we could go back and say in this prospective randomised trial, what did the tissue show and now that we have twenty year outcomes, we can test, does this score really predict the outcomes?

So what part of the data?

The data from that study showed that the recurrent score, in a very powerful way, identifies patients with a good prognosis, as opposed to patients with a much worse prognosis. So it identifies in that B14 study patients who have, in the low risk group, about a 7% risk of distant failure at ten years and the confidence intervals still have that under 10% risk at ten years for distant metastasis, versus patients who have about a 30% risk at ten years. It turns out in subsequent studies looking at treatment benefit, prediction of outcome, those patients who had that 30% risk are the patients who benefit from chemotherapy.

And with this score you can actually classify a low, an intermediate or a high risk group of patients can you?

Yes, this is a blessing and a curse really. The intermediate group still leaves us with questions and reverting back to our traditional clinical parameters but fortunately we have just completed, in the US, a very large trial, about 11,000 patients, a prospective randomised trial looking at that intermediate risk group and those patients were randomised to chemotherapy or not. So in the near future we’re going to have a lot of information about that.

Which patients, specifically, are suitable for this test in terms of things like oestrogen receptor, nodal status and so on?

It started out with patients who were oestrogen receptor positive and it is still that way. So patients who are oestrogen receptor negative, if I send in a specimen and it turns out it’s oestrogen receptor negative on RTPCR, they’ll send me a report back and say this is not appropriate for testing. So oestrogen receptor positive, node negative in the beginning; we’ve learned now that this test applies as well to node positive patients, properly selected node positive patients. Patients who have a tumour generally bigger than 1cm but from 5-10mm, if the tumour is something other than low grade, so intermediate grade or high grade, 5-10mm, we’re testing those patients too.

So the patients who classically would have been in some doubt, you can now test and refine treatment decision making decisions?

Absolutely and that is, in the beginning, how most of us began using this test. We were on the fence in many cases, we knew the benefit from chemotherapy wasn’t huge and we used it as a way to lean one way or another off that fence. But we were also finding that our traditional decision making is completely contravened in some cases; cases where we thought we’re not on the fence, we know what to do, we’re now changing that.

In what proportion of cases, how many patients, typically in percentage terms, would you be sparing unnecessary chemotherapy and how many patients would you be treating who would have not been receiving treatment before?

If you just look at the NCCM guidelines or the Saint Gallon guidelines, it looks like you reclassify between 40% and 50% of the high risk patients, so it’s a big public health impact. But if you talk to medical oncologists, they’ll tell you, “I don’t just use guidelines, I look at the individual case,” and so we’ve done studies now where medical oncologists said, “My judgement is this patient doesn’t need chemotherapy,” or “My judgement is this patient needs chemotherapy,” and that’s the real decision. That decision is changed, we know from these studies, about 30% of the time.

You know a great deal then about prognosis, your score can tell what is likely to be the outcome in the long term for patients, what about predicting the effect of specific therapies though? Can you refine that sort of understanding?

To me that’s the most exciting thing. Predicting the benefit of endocrine therapy versus chemotherapy is something we’ve become very interested in because we’re doing a lot of neo-adjuvant treatment, converting patients who had a larger tumour that might have had to have had a mastectomy into patients who could have breast conservation. So if you know that this patient doesn’t respond to chemotherapy up front you can properly chose endocrine neo-adjuvant therapy and this score appears to tell us in the low risk patients they’re not low risk without adjuvant endocrine therapy; they could be high risk without endocrine therapy but we know these patients are low risk given endocrine therapy. So on the low risk range we know that’s a patient who is appropriate for neo-adjuvant endocrine therapy, or more appropriate, these aren’t things that have been locked down yet. We know that for patients who have a high score it would be much more appropriate to use neo-adjuvant chemotherapy.

Let me push you even a bit further on this. What about the choice between the type of endocrine therapy and the type of cytotoxic chemotherapy, is there any hope of understanding that too?

I think that is going to be part of what we get from the multi gene classifiers. We don’t get that presently. There are some tests out there that tell us whether tamoxifen might or might not be the proper endocrine therapy, that’s really more of a single gene analysis, it’s an analysis of the patient’s genetic make-up. Then there are tests that can help us decide between this chemotherapy and that chemotherapy, but that’s not part, presently, of the oncotype score or the MammaPrint score, both of those companies and many other universities etc are looking at multi gene classifiers or single gene classifiers that are going to tell us this is the better therapy for this patient.

It all sounds very exciting, could I get you to wrap this up by putting this in the clinical context? What should cancer doctors be making of the availability of this test, how much scope does it offer their patients?

In the beginning most clinicians are going to use this kind of information, the multi gene classifiers, to make a decision in a case where they’re on the fence. In fact, that’s what the Saint Gallon guidelines have suggested right now when there is uncertainty. But I think we are now seeing, perhaps a little more in the US, but in both places we are now using this to help us decide in cases where we thought we knew what to do - our traditional parameters said for certain we should go this way and we’re learning that many of these patients do not benefit from systemic adjuvant chemotherapy but many patients who we thought didn’t  need it will.

And the tests are easy to use?

In the beginning there is some learning to order properly those logistics but, yes, very easy and I’m sure those companies are making that easier.

Looking at overall population figures, do you see a clinical impact in terms of survival in the immediate future? We’ve had big strides forward in breast cancer from the introduction of modern therapies, but do you think testing like this could improve survival?

I do think it can, in fact we certainly are identifying patients who we wouldn’t have treated with systemic adjuvant chemotherapy who will really benefit from that. On the other side we are also eliminating systemic adjuvant chemotherapy in cases where it would not give us a benefit, in fact it would only result in toxicity and we just couldn’t tell who that was before.

And a very brief message for what you’d like doctors to remember from all of this?

It’s important for the medical oncologist and the surgeon to be working together because if the surgeon is on board with the medical oncologist they can order these tests much earlier in the process and that puts that information in the medical oncologists hands in a timely way so they can use it to make decisions about those patients.

Well this is all very exciting, Pat, but could you put this now into clinical context to just illustrate what busy doctors should be doing with this test?

Well we continue to use the ordinary clinical parameters to help us decide which patients should have testing. But presently with the recurrent score test, the oncotype DX test, patients who are oestrogen receptor positive and, in general, patients who are HER2 negative, if a patient is borderline with a FISH ratio of two or three then perhaps that patient should be tested. But patients with a very high ratio that are strongly HER2 positive are almost always going to have a high score, but patients who are ER positive we used to say if they’re node negative and now we know that this is applicable in patients who are node positive, at least certainly those who are post-menopausal and node positive.

I think it’s important for the surgeon and the medical oncologist to be working closely together so the surgeon knows which cases the medical oncologist would like to have this information. So when the patient arrives at the medical oncologist’s consultation they have information and they can make a decision in its timely way, the patients don’t suffer longer waiting to find out.

And in ballpark figures, just how much difference might this make overall in terms of cancer survival, which of course has improved a huge amount by introducing modern treatments?

In the first place we know that this changes the decision making overall in about 30% of cases. So the experienced established knowledgeable clinician is deciding that this should be the treatment, whether it’s chemotherapy or no chemotherapy. We know that the recurrent score changes that decision in about 30% of cases. In the majority it’s to avoid chemotherapy and use endocrine therapy, to properly select endocrine therapy, but in a significant number, more like 5-10% we’re identifying patients who would not be given systemic adjuvant chemotherapy who need it. So these are patients who would have had a bad outcome and we would have been very surprised about in the past.

And as we have strong evidence that therapies save lives, then that could mean quite a big extension of life for many patients.

And now we’re able to see a lot more clearly which therapy is the most effective therapy, whether it’s endocrine therapy for these patients in the low recurrent score range or systemic adjuvant chemotherapy in the patients in the high recurrent score range.

So if you were to summarise in just a very few words what doctors should take home from this new knowledge, what would that be?

Try to be on board with the team, so the medical oncologist and the surgeon know early on this patient should be tested as soon as that information is identified so that that information is available to make decisions later on.

Well Pat, it’s been great having you here with us in Bordeaux for the European Society of Surgical Oncology meeting. I wish you lots of luck with the rest of your exciting work and thank you very much for taking part in ecancer.tv.