At this virtual ASCO in 2020 we presented the long-term results of the MINDACT trial. Just to remind everyone, the MINDACT trial enrolled 6,693 patients and its primary aim was to identify those patients who needed adjuvant chemotherapy and those who could safely forego this treatment. So the way the trial was designed was that all patients had their risk assessed by our traditional clinical pathological features and we did that using a modified version of Adjuvant! Online and all patients had also their risk assessed using the 70-gene profile also called MammaPrint. One of three situations could occur: both tests would say that the risk of recurrence was low and in that case we would not propose chemotherapy, or both would say that the risk was high and in that case chemotherapy was proposed. For the discordant groups where one would say high and the other low there would be a randomisation and the randomisation would be between the methods of risk assessment that we would follow. This is a common misconception about the MINDACT trial – the randomisation is not about chemotherapy versus not but it’s rather follow the clinical risk or follow the genomic risk and then suggest or not chemotherapy according to the results.
MINDACT also included the different subtypes of breast cancer but the majority, more than 80%, were ER positive, HER2 negative, which we now know is the subtype where genomic testing can have an impact. The primary endpoint of MINDACT was related to this discordant group and MINDACT should be seen as a de-escalation study where the primary endpoint is not a comparison but it is a threshold. So we wanted to see if at five years the distant metastasis free survival was at least 92%, so the lower boundary of the confidence interval, for the group of patients classified as high clinical risk but MammaPrint low risk, randomised to follow MammaPrint and did not receive chemotherapy.
As the secondary endpoint we wanted to look at the potential impact of chemotherapy in the two groups of discordant risk assessment. It’s also useful and important to know that the clinical high risk by MINDACT was a really high risk. What do I mean by that? So 50% of these patients classify as clinical high risk, have at least one positive lymph node, between one and three. About one third had grade 3 and about 60% had tumours bigger than 2cm. So clearly a population of higher risk than, for example, the ones included in the TAILORx population, in the TAILORx trial, or where an evaluation according to the definition of clinical risk by MINDACT shows that in the intermediate recurrence score group we have about 74% of clinical low risk patients and only about 26% of clinical high risk. This is important for the interpretation and implementation in clinical practice.
The primary results were presented and published in 2016 and showed that the MINDACT had met its primary endpoint with a DMFS at five years of 94.7% with a confidence interval started at 92.5% and that we could, by following the genomic risk, reduce chemotherapy prescription in about 46% of patients. Secondary endpoint we saw a difference of 1.5% between chemotherapy versus not in that subgroup of patients clinical high, genomic low.
Now we present the results with 8.7 years’ median follow up and at this time we have more than 90% of patients followed for at least five years and more than 70% of patients followed by at least eight years. What do we see in the primary endpoint? A distant metastasis free survival at five years of 95.1% with a lower boundary of the confidence interval starting at 93.1%, so clearly above the threshold of 92% that I just explained, confirming that MINDACT met its primary endpoint and it’s a positive de-escalation trial.
We also continued to prove the clinical utility of MammaPrint because we see that the prognosis is different according to the classification by MammaPrint, so the group and prognosis. The two discordant groups have intermediate prognosis but still quite good with DMFS at five years above 90%. The group that has clinical high genomic high has the worst prognosis, even receiving chemotherapy.
Now, looking at the secondary endpoint, I told you that we saw a 1.5% difference at five years. Now with longer follow-up at five years the difference is 0.9%. At eight years we see more relapses, just as we would normally see with any breast cancer trial, and the difference at eight years is 2.6% overall in this group of clinical high genomic low.
Now, if we focus only on ER positive HER2 negative, clinically high risk MammaPrint low risk and we look at the potential benefit of chemotherapy by splitting according to age we see that in women above the age of 50, so mostly postmenopausal women, there is absolutely no benefit in providing chemotherapy with a 0.2% at eight years and even negative at five years. So we can safely forego chemotherapy in postmenopausal women even if classified as clinical high risk. However, in the group of younger, less than 50 years old, mostly premenopausal women this difference at five years is 2.6% and at eight years it’s 5% which we consider that it’s now a clinically relevant difference. How can we interpret this age difference? The same way as we discussed last year in San Antonio and together with the TAILORx results which are also similar, we believe that this is not a direct effect, a cytotoxic effect, of chemotherapy but it’s rather due to the ovarian function suppression induced by chemotherapy. Why do we say that? Because this is a late effect, we see this with longer follow-up and we know from the Oxford overview that the cytotoxic effect of chemotherapy is seen in the first five years. We now have this result very similarly in two very large trials. Can we prove that this is because of ovarian function suppression? We cannot, neither with MINDACT nor with TAILORx, and this has to be carefully discussed with the patient.
So, in conclusion, at 8.7 years median follow-up MINDACT continues to be a positive de-escalation study and the results are even stronger. At eight years the DMFS gain for chemotherapy administration in the clinical high genomic low group is 2.6% but when looking according to age this difference does not exist for postmenopausal women and so we have a fully preserved performance of MammaPrint in this group of patients and we can safely forego chemotherapy. In premenopausal women the difference is relevant and we need to discuss very carefully with patients if chemotherapy is needed, if ovarian function suppression can be given and a very careful discussion between risk and balance should take place.
I didn’t have time to show you or discuss with you the results but these results are the same for node negative and node positive up until three positive nodes. It’s also shown that for those who have a clinical low risk, having a genomic high risk does not add also any advantage in guiding treatment based decisions.
So I’d like to finish by thanking all the patients, all the investigators and all the enormous team that work both at EORTC, at the site levels and at the big headquarters for this long but very important trial.