You’ve been looking at gene signatures, particular gene signatures, and subtypes of breast cancer. What were you trying to do in this study looking also at the interaction of gene signatures and neoadjuvant therapy?

The study was originally designed to look at whether the addition of either carboplatin or bevacizumab to a standard neoadjuvant chemotherapy regimen would increase the pathologic complete response rate in patients with triple negative breast cancer. The results I presented last year said that it did but we know that triple negative breast cancer is comprised of many subtypes, in other words it’s not a monolithic entity. So the question was whether looking at some of those subtypes, and the most important being what’s called the basal-like subtype, whether the addition of carboplatin or bevacizumab would increase the pathologic complete response rate in those patients and also whether there was an interaction between subtype and the benefit in terms of pathologic complete response rate.

Now gene signatures that can give you a subtype are not necessarily routine, are they?

They aren’t routine; basically to get them we took tissue samples before we started chemotherapy in all of our patients and then we subjected these samples to RNA sequencing. We compared the result we got from the RNA sequencing to results that are present in something called the Cancer Genome Atlas which has a large number of breast cancer samples. By doing that we were able to categorise our cancers as being either basal-like or some other type which we call non-basal-like.

Now in the study what exactly did you do? Which patients did you look at and how did you conduct the study?

The eligible patients for this study were those with stage 2 or 3 triple negative breast cancer; so these were women with larger tumours or lymph node involvement. So these were all patients who, in a standard situation, would get chemotherapy. All of the patients received a standard chemotherapy regimen of weekly paclitaxel followed by dose dense AC and then they were randomised as to whether they would also get carboplatin given during the paclitaxel or bevacizumab given throughout the neoadjuvant chemotherapy.

And what were the primary objectives of this study?

The primary objective of the study was to see whether we would increase the percentage of patients who would achieve a pathologic complete response and that is defined as when the patient goes to surgery and the surgeon cuts out the area where the cancer was and samples the lymph nodes that there would be no invasive disease left in the breast and the lymph nodes.

What did you find in terms of which subtypes did you detect and their association with either bevacizumab addition or, indeed…?

Carboplatin.

Either bevacizumab addition or carboplatin addition.

When we designed the study we expected the dominant subtype to be this, what we call, basal-like breast cancer. We expected, based on retrospective analyses of triple negative breast cancer that about 70-80% of the patients would have basal-like cancers. Actually we were a little surprised that even a higher percentage, 87% of the cancers, typed out as basal-like cancers. So then we asked the question, alright, what impact does being basal-like have versus non-basal-like first on the overall pathologic complete response rate? And we found that there was no difference. 54% of our basal-like cancers had a pathologic complete response versus 52% of the non-basal-like cancers. So then we asked do we see an increase in the pathologic complete response rate with the addition of carboplatin or bevacizumab? And the answer is yes we did, we saw statistically significant increases in the pathologic complete response rate both with carboplatin and with bevacizumab. Then we asked the question, well, did subtype matter? There we’ve had a little bit of a difference. So for carboplatin the increase in the pathologic complete response rate was the same whether you had a basal-like cancer or one of the less common non-basal-like cancers. But for bevacizumab we saw the opposite, that is while there was an increase in the pathologic complete response rate with the addition of bevacizumab in basal-like cancers, the pathologic complete response rate actually fell with the addition of bevacizumab in the small number of non-basal-likes. So we saw what we call an interaction between subtype and bevacizumab benefit but we didn’t see an interaction between subtype and carboplatin benefit.

Now, that’s a sort of proof of principle because you wouldn’t necessarily be able to apply that in today’s therapy with bevacizumab no longer being a recommended therapy.

Right. For bevacizumab there’s still a lot of interesting data. There were a number of studies done with bevacizumab both in the pre-operative setting and in the post-operative setting. Many of them have demonstrated small benefits with bevacizumab but not enough to really make it exciting in terms of continuing to use it in breast cancer. So the question is if we could identify a group of patients who really don’t benefit from the addition of bevacizumab and if we were able to exclude those patients from our studies, would we see a much larger benefit in the rest of the patients with the addition of bevacizumab.

But what clinical messages do you think are coming out of this now? Because not only can you look for basal-like or non-basal-like, there are other factors that could be genetically detected, predispositions towards those things.

That’s certainly true and we actually have a large number of on-going correlative studies using the samples from our clinical trial. We’re going to ask questions about BRCA status, about this test called homologous recombination deficiency; we’re going to look at other signatures that may give us more information about what subgroup of patient’s benefits more and what subgroup may benefit less. For right now the take home message in regards to carboplatin, let’s say, because I know there has been a lot of interest in whether we should add carboplatin to neoadjuvant chemotherapy in patients with triple negative breast cancers. Our study and a study done by the Germans, both demonstrated higher pathologic complete response rates with the addition of carboplatin. We don’t yet have long term data that is we don’t know whether the increase in the disappearance of the cancer from the breast and the lymph nodes is going to translate into long-term outcome improvement. But I think the take home message in regards to carboplatin from our study is that we haven’t yet identified a group of patients who may get a greater benefit of carboplatin in the pre-operative setting but we also haven’t identified a group of patients who gets no benefit from carboplatin. So if you, as a clinician, were inclined to add carboplatin to neoadjuvant chemotherapy in triple negative patients, that’s OK to do. In other words there is no group of patients in whom we would say, “Well don’t do it in those patients.” There are other clinicians who are waiting, waiting for long-term outcomes before they decide what to do about carboplatin.

For bevacizumab, as you comment, it’s not available, it’s expensive, it has more toxicity. So for bevacizumab the take home message is can we look at some of the other studies that have been done with bevacizumab and ask the question can we identify a similar subgroup of patients who aren’t benefitting and do then the rest of the patients get a larger benefit?

What should doctors take away from this, very briefly, in terms of what sorts of gene signatures might be useful in the future?

If down the road we identify the fact that, yes, there is a subgroup of patients who don’t benefit from bevacizumab and another group that does then you could say is it worth trying to get one of these gene signatures on your patients before you start treatment? In other words, if the turnaround time for the test could be short enough. For carboplatin we don’t yet have a test identified that’s going to identify a group that gets greater or lesser benefit so there’s no additional testing that we would recommend at this time but hopefully in six months, a year or more we’ll have more information about that.

So the bottom line message coming out of your study in just a sentence or two, what would that be?

It would be if you’re inclined to use carboplatin in the neoadjuvant setting for triple negative breast cancer patients go ahead and use it. For bevacizumab wait and see.