Another lecture that I gave at the plenary session in the meeting yesterday was a session dealing with using genomic classifiers to predict local regional recurrence of the tumour. The background of this is that we developed several genomic classifiers, some of them are commercially available, but the main goal of this classifier, the one that we developed, is to predict the risk of distant recurrence and also to determine whether patients needed adjuvant chemotherapy or not. It was mostly developed in patients with oestrogen receptor positive, HER2 negative tumours.
In the process, though, we were able to identify that these predictors also predict the risk of local regional recurrence, in other words, recurrence in the breast chest wall or in the axillary lymph nodes. Because local regional recurrence in this kind of cancer is very tidily correlated so whatever factors increase distant recurrence, they also increase local regional recurrence. So, as a result, we were able to look at several datasets and determine that some of the genomic classifiers that we use for the prediction of distant recurrence also predict local regional recurrence.
Now, the question then you ask is does this have clinical implications, can we use this? There are two aspects of this: you potentially use it to perhaps change surgical approach but in reality genomic classifiers do not change surgical approach because surgery for breast cancer is determined usually by the extent of the disease. If there is a bigger tumour you have to do bigger surgery or you do mastectomy. So so far we haven’t been able to identify predictors where we can change our surgical approach, short of, of course, putting the patients through neoadjuvant chemotherapy and then shrinking the tumour and doing less surgery, as we discussed in the management of the axilla lecture. But the other aspect of that is that you perhaps can tailor radiotherapy to the breast or radiotherapy to the regional lymph nodes. There have been several datasets indicating that, indeed, these predictors of this kind of recurrence predict risk of local regional recurrence and a prospective trial now in the US is assessing this question. For example, patients with 1-3 positive nodes and a low 21-gene recurrence score on OncotypeDX are randomised to comprehensive regional nodal radiotherapy or not. So if this study eventually is conducted and shows that there is no benefit then we would again carve out a group of patients that don’t need radiotherapy based on the genomic classifier that puts them in the low risk.
The other question that we are asking is can we even eliminate breast radiotherapy after breast conserving surgery which, of course, is something we do routinely. Several studies also have shown that the risk of recurrence in the breast is actually less for patients that have low genomic profiling. However, most of the data we have is with radiotherapy because we have given radiotherapy in all these clinical trials then we take tumours and assess them with genomic classifiers. So, again as a result of that, several prospective studies have been designed to actually ask the question if we select patients based on genomic profiling as a low risk, can we observe them after breast conserving surgery and not do radiotherapy. So there have been some single arm trials doing that but also randomised trials. One of them is the one that we conducted in the US, our research group, the NRG Oncology, called DEBRA or the BR007 trial. This trial is a large trial comparing breast radiotherapy versus no breast radiotherapy for patients that get breast conservation and have a low 21-gene recurrence score, under 18. So that’s an important part because, again, if that shows that there is no difference we can avoid radiotherapy for these patients.
But all these genomic classifiers are prognostic of risk of local regional recurrence. In other words, they assess how big is the risk of local regional recurrence and if it’s low we say maybe we can avoid radiotherapy. But the holy grail, so to speak, is to find genomic classifiers that predict whether you benefit from radiotherapy or not and that’s a different question. Now we have some evidence from coupled data was presented at the San Antonio meeting this past month, a month ago, that there is a signature called the POLAR signature that in three studies and a meta-analysis of the three studies has shown actually to predict benefit from radiotherapy. So if you’re low risk on the signature you have very low risk whether you have radiotherapy or not. If you’re high risk on the signature you have low risk if you receive radiotherapy but you have very high risk when you don’t. So that’s a perfect biomarker because it tells us that not only do low risk patients have low risk by definition but also don’t benefit from radiotherapy. So it strengthens the argument of not giving radiation to the patients. Of course, this signature needs to be validated further in other analyses but it’s very promising and it was exciting to see the data in San Antonio.