First outcome results from MINDACT breast cancer clinical trial
Dr Martine Piccart - Jules Bordet Institute, Brussels, Belgium
I’m one of the three principal investigators of this very large clinical trial and in the room here we have Laura van 't Veer who is the leading scientist who discovered the 70 gene signature called MammaPrint. So today I have the big honour to present the primary analysis of this trial and the acronym stands for Microarray for Node Negative 1-3 Positive Node Disease May Avoid Chemotherapy.
Because advanced breast cancer is largely an incurable disease oncologists do prescribe to many women with early disease adjuvant medical therapies, the goal of which is to eradicate micrometastasis. Now, this task is a difficult one and it is associated with a number of risks that you see here, one of which is overtreatment. The biggest concern with overtreatment is with adjuvant chemotherapy and it will happen when the survival gain associated with adjuvant chemotherapy is small, in the range of 2-3%, which happens in good prognosis patients and is counterbalanced by the long-term severe risks of adjuvant chemotherapy which include secondary cancers, secondary leukaemia and congestive heart failure. Needless to say that the results are a high socioeconomic burden linked to adjuvant chemotherapy.
So in 2002 clinicians in Europe got very excited because… I think I missed one slide again; this is difficult to handle. Sorry.
So I just wanted to remind you that fortunately we have some tools helping us with adjuvant treatment decision making. One of these tools is a computerised tool developed by Dr Peter Ravdin in this country and it is probably one of the most objective ways to evaluate for an individual patient. On one hand the risk of relapse and death without adjuvant treatment and on the other hand the magnitude of benefit that would be associated with the prescription of adjuvant therapies.
Now, in 2002 in Europe we got quite excited because of the emergence of a new tool that appeared to better capture tumour biology, that was the result of fantastic work by laboratory scientists at the NKI in Amsterdam, a group led by Laura van 't Veer. So she discovered the 70 gene prognostic signature and you can see that the signature appears to do quite a good job in separating a group of women with very few deaths in a ten year follow up period from breast cancer without adjuvant chemotherapy from another group with a more substantial risk.
So we decided then that we should embark on the very difficult path of biomarker validation. At that point MammaPrint was a low level of evidence biomarker, it was developed on retrospective tumour material. The first thing we did was validate independently the signature in 202 chemotherapy untreated women with node negative breast cancer coming from hospitals outside of the Netherlands. When this was done successfully we moved on to verify the inter-laboratory reproducibility of the signature and then we opened the trial MINDACT in 2007. Because later on we were able to also show that the signature was quite robust also in women with 1-3 positive nodes an amendment was introduced and, as of 2009, women with positive nodes could enter the trial.
Essentially we got very confident that the genomic assay would outperform the clinical criteria by reducing the prescription of adjuvant chemotherapy without impairing patient outcome. So MINDACT is the only trial that is putting tumour biology against tumour anatomy with a few biological features added. Of course we needed to agree on the definition of what is a low clinical risk by Adjuvant! Online and a high clinical risk by Adjuvant! Online. We had a long discussion in the consortium, which included Europa Donna patient advocates, and came up to the conclusion that a 92% threshold for predicted ten year overall survival without adjuvant chemotherapy using Adjuvant! Online would be adequate. Why? This is because in such a good prognosis group if you would give chemotherapy you would probably improve survival by a few percent, 2-3% at the most, and that would be counter-balanced by the risk, as I explained before.
So this is the design of the trial. Women had first to consent to have part of their tumour snap frozen and sent to a central lab in Amsterdam for the MammaPrint test. At the same time the tumour would be evaluated in a traditional way locally in the pathology department. After surgery, assuming a successful MammaPrint test, women would have to consent to have their adjuvant treatment tailored to their clinical risk and their genomic risk. What does that mean? That if the woman would be low risk by both tests she would not be offered adjuvant chemotherapy but she could receive endocrine therapy. If the woman would be high risk by both tests chemotherapy would be strongly advised. But then if there would be a discordant risk evaluation there would be randomisation - yes or no adjuvant chemotherapy.
So you see that this trial was primarily a European trial conducted in 112 hospitals coming from seven cooperative groups and nine European countries. Here you see the characteristics of this large patient population, 6,693 women were enrolled, median age 55 years. One in five women had node positive disease and, interestingly, although the population was largely hormone receptor positive, 10% of the patients had HER2 positive tumours which are biologically aggressive.
After a median follow up of five years you can see that 3% of the women died, 10% experienced either a distant relapse, a local regional relapse, and quite a number experienced second primaries and 5.4% experienced distant relapses or death.
Let me show you the outcome of all the four groups. So first of all the concordant groups. The women who were low risk by both methods, the green curve, did extremely well. You see that their five year distant metastasis free survival is close to 98%. And these women didn’t get chemotherapy. In contrast the women high risk by both methods who received chemotherapy did somewhat worse. They had a 90.6% five year distant metastasis free survival. Now you see, of course, that the characteristics of these two groups were very different. The low risk patients were mostly node negative, small tumours with hormone receptors and the high risk patients had bigger tumours, they were node positive in 25% of the cases and there were even patients there with triple negative breast cancers.
Now let us move to the discordant groups, the very interesting discordant groups. They have an outcome in terms of distant metastasis free survival in between the concordant groups and you see there five year distant metastasis free survival rates which are pretty good, around 95%, just in between the two other groups. Now, very importantly, probably the most important slide of the entire presentation, I am explaining here the primary statistical test as written in the protocol. So the primary statistical test focussed on one of the two discordant groups. Women clinically high risk, genomic low risk randomised to receive no chemotherapy, meaning you trust the genomic assay; you don’t look at the clinical risk, you really trust the genomic assay. And, of course, what we see here is that we achieved in this group a five year distant metastasis free survival of 94.7%, quite good, and the 95% confidence interval, as you can see, does not include 92% and that was the definition of a successful trial.
Now, a difficult question. In these discordant patients does chemotherapy help? Well the trial was not powered to answer this question. What I’m showing you here is an intent to treat analysis which is the most robust from a statistical point of view. If you look first to the right the relatively small group of patients who are clinically low risk and genomically high risk, you see that the curves are almost superimposable, there are very few events and it’s very difficult to say if there is any chemotherapy effect. The same is true for the other critical group on the left – clinical high risk, genomic low risk. Again, the two curves, chemotherapy, no chemotherapy, are very close but a statistician will tell you if you look at the hazard ratio that it is not totally impossible that there is a very small chemotherapy benefit, a relative reduction in risk of 22%, which would translate into a very small absolute benefit that would not justify the risks of chemotherapy.
So, in conclusion, this trial has played a major educational role in Europe, mobilised hundreds of professionals, popularised the concept of biology driven treatment. It demonstrated that genomic scans provide important information in order to treat patients with early breast cancer in a more optimal way. It implemented the logistics to collect and freeze tumour materials in a quality controlled fashion and it’s very important for you to understand that MINDACT is associated with an invaluable biobank for future research because we have the full gene expression arrays on all these patients, not only the 70 genes.
MINDACT provides level 1a evidence of the clinical utility of MammaPrint. The clinically high low genomic patients which included about half of node positive patients had a five year distant metastasis free survival rate in excess of 94% whether randomised to adjuvant chemotherapy or not. I didn’t show you that but in the entire MINDACT population the trial confirmed our hypothesis that the genomic strategy would lead to a reduction in chemotherapy prescription as opposed to the clinical strategy. But the important message here is that among the clinical high risk patients the clinical use of MammaPrint is associated with almost a halving of the use of chemotherapy.
Last but not least I have to, of course, thank all the patients who accepted to be enrolled in this trial with a lot of enthusiasm, all my colleagues who worked on this trial, all the funding agencies, all the fellows who worked on the trial and the fantastic headquarter teams at the EORTC.