IMPAKT Breast Cancer Conference 2013

Breast cancer heterogeneity no barrier to predictive testing

Dr Michał Jarząb - Maria Skłodowska-Curie Institute of Oncology, Gliwice, Poland

Michał, thanks for joining us on ecancer.tv. You’re addressing the topic of heterogeneity, that seems to me a really difficult thing because on the one hand we’ve got all this new genetic knowledge coming out, on the other hand the more you find, the more complex it becomes. What do you make of this heterogeneity?

Well we understand that from the biologic point of view cancer is heterogeneous. However, when we sample the cancer tissues for many genomic tests we take only one sample usually and we miss the heterogeneity in that way, in genomic testing for example. And the issue couldn’t be missed if you would like to cancer target really. So probably taking one or more samples from tissue will be a good approach to understand that the cancer is a heterogeneous disease.

But you actually want to individualise therapy. So, bearing in mind this is a complex picture and you need to sample very carefully and perhaps more than once, what needs to be done to individualise therapy, especially at the beginning of the disease when the disease is first discovered?

It is the issue that when we are taking a core biopsy before, for example, chemotherapy, before the neoadjuvant chemo in breast cancer, taking one sample we are sampling a very small amount of tumour. The decision based on that may miss a number of relevant issues. When you’re taking three cores we first could see what the general outlook of this tumour is and the second issue is that we could understand the worst person of it because maybe the response is based on the whole tumour but maybe only the most resistant part of this is deciding on the resistance of the whole disease.

Could you tell me what you’ve been doing in your research that is beginning to clarify this issue?

In our study we are observing the patients treated by neoadjuvant chemo by many imaging studies like magnetic resonance and PET-CT. And before the chemo we are taking not only the single biopsy for genomic testing but we’re taking three cores for molecular analysis as well as routine cores for pathology. And in these cores we are analysing them by microarray test and we did this in 26 tumours, taking three cores for every, so that was 78 microarray profiles. And we analysed the data and what we found is that there is a heterogeneity, at least a substantial heterogeneity, but in three fourths of these tumours, 75% of these tumours, it is not very high so it doesn’t preclude the use of non-genomic signatures. So the first conclusion from our study is that generally, looking at the outlook of breast cancer, the number of genomic signatures could be reliably applied.

Have you found that by taking the three samples you can refine the treatment? Are there patients whom you would treat differently because you have three samples rather than just one?

Well that is the distant goal of our study. We analysed the number of genomic signatures to predict the prognosis in these patients and some signatures seem to be less heterogeneous than others. So some signatures and, for example, this was the case of the novel signatures derived from cell lines, derived by analysis of chemosensitivity of cell-lines, that’s the paper by Shen et al [1], these signatures seemed to be less heterogeneous than others so they could be potentially successfully applied to predict the treatment resistance or treatment sensitivity of breast cancer.

At the moment that’s theoretical, have you got any practical lessons that have come out if it yet?

No, not yet. We are analysing that and we are trying to assess what signature would be the best to test prospectively.  Obviously in our study we have collected more than 300 patients observed that way, we are looking at differences between patients that achieve pathological complete response and patients not responding but that’s not a matter of the presentation at the IMPAKT Conference here.

Indeed your talk has been headed in one of the descriptions of it ‘Heterogeneity no barrier to predictive testing’. How far down that path do you think you are so far?

We concluded that, first, in 75% of tumours that is no barrier at all because they are relatively homogeneous. In the remaining one fourth of tumours we see heterogeneity but we could probably overcome it by taking three cores and looking at this part of the profile which is homogeneous by taking three cores. Our study showed that sampling of breast tumour by core needle biopsy also for genomic testing is feasible. Patients didn’t feel discomfort during the procedure because obviously taking three cores more is lengthening the procedure slightly but this was not related to significant discomfort to patients.

And one quarter of them can then be separated from the rest.

Yes, yes.

What do you think doctors might consider doing, then, in the near future? Going to this three core approach?

Well I think that this is an approach which shall be tested in a prospective setting. We have to analyse more and more if doing that we are able to optimise the genomic testing. At the moment the number of data which we obtained done by one sample and the degree of heterogeneity shall be taken into account but I think this would be more in the research setting at the moment.

One of your findings, then, is a little bit reassuring, that heterogeneity is not as heterogeneous as perhaps you thought, yes.

Yes, that’s it.

What sort of message would you have for doctors finally, coming out of this then?

Certainly if we consider sampling of breast cancer by core needle biopsy, we couldn’t be sure that a single core is enough. When we see discordances in the results of the pathology we have to think that the cancer is heterogeneous and we have to take this into account.

Michał, thank you very much for joining us on ecancer.tv. 

Thank you very much. Thank you.

[1] Shen K, Song N, Kim Y, Tian C, Rice SD, Gabrin MJ, Symmans WF, Pusztai L, Lee J (2012) A systematic evaluation of multi-gene predictors for the pathological response of breast cancer patients to chemotherapy PLoS One 7 (11):e49529. doi: 10.1371/journal.pone.0049529 PMID: 23185353 

Shen K, Qi Y, Song N, Tian C, Rice SD, Gabrin MJ, Brower SL, Symmans WF, O'Shaughnessy JA, Holmes FA, Asmar L, Pusztai L (2012) Cell line derived multi-gene predictor of pathologic response to neoadjuvant chemotherapy in breast cancer: a validation study on US Oncology 02-103 clinical trial BMC Med Genomics Nov 16 5 51 doi: 10.1186/1755-8794-5-51 PMID: 23158478