AACR 2016

A presurgery combination therapy may improve outcomes for women with Her2-positive breast cancer

Dr Angela DeMichele - Perelmen School of Medicine, University of Pennsylvania, Philadelphia, USA

When you think about the landscape of HER2 positive breast cancer it’s clear that over the last twenty years we’ve seen outcomes improve with the combination of chemotherapy and HER2 directed therapy. TDM1 and pertuzumab are new HER2 directed therapies that target HER2 in unique ways. We also know in the neoadjuvant setting, that is giving chemotherapy before surgery, that achieving a pathologic complete response, that is no tumour seen in the breast or the lymph nodes, at the time of surgery is an excellent surrogate for long-term clinical outcomes. However, not every patient achieves a pathologic complete response so the goal is to try to find new therapies that will bring many more women to the goal of achieving a pathologic complete response. The ISPY2 trial tested the ability of TDM1 and pertuzumab to do just that.

So let me tell you a little bit about this rather unique trial. The ISPY2 trial is a standing platform trial, it’s a phase II adaptively randomised trial with the goal of identifying drugs or new drug combinations to take to phase III. There are numerous simultaneously running experimental arms and we match therapies with different breast cancer subtypes based on hormone receptors and MammaPrint. The comparator arm to each of these different experimental arms is standard neoadjuvant therapy and our primary endpoint is pathologic complete response. Now, we use a technique called adaptive randomisation which minimises the number of patients that are needed to reach this goal of determining efficacy, and I’m going to show you in a few minutes just how that works. But when a drug or a combination reaches that goal of showing efficacy we call that graduation and that means reaching that threshold of the predictive probability of success of that new drug in a subsequent phase III trial.

So this is how adaptive randomisation works. If you start in the upper left corner with the box that says ‘New patient accrues’, a patient is enrolled in the trial and randomised to one of the concurrently running arms. As patients move through their treatment we continuously update how they are doing, all of their outcome data, including MRIs that they obtain throughout the course of the trial and their pathologic complete response data. This is then used to update longitudinal models that can tell us the predicted probabilities that one of the experimental arms is better than the control in phase III for each signature. So as we continually update this data we then examine whether we’ve hit one of our termination rules. If none of the subsets of patients receiving that drug is showing a benefit, that drug stops for futility. If any of the subsets of patients receiving that drug or combination has hit our threshold for graduation the drug will graduate. But since this is an ongoing process we typically will not have seen either of those things happen and we will continue to randomise patients to these arms. We can add new arms as our accrual allows and continually update probabilities as new patients come in, looking at whether various arms are better than control by subtype and then calculating the adaptive randomisation probabilities, meaning that the next patient who gets randomised will be randomised based on how well all the other patients have been doing in the trial. This design is known to be very efficient; it helps direct patients to the drugs that are working best and it minimises the overall number of patients that you need to find the answer.

So this is the CONSORT diagram for the TDM1 pertuzumab arm. Let me tell you about the schema for the trial. So women who were eligible for enrolment in this trial had invasive breast cancer that was at least 2.5cm in size. They underwent MRIs and biopsies to collect the important data needed for assessment and they had to have adequate organ function and performance status. They were then, in the HER2 positive group, adaptively randomised to either the control, which was paclitaxel plus trastuzumab, which I will call TH, to TDM1 and pertuzumab or to one of the other concurrently randomising HER2 positive arms. They received that therapy for twelve weeks and then went on, all of them, to receive standard doxorubicin and cyclophosphamide. You can see that the doses for the control therapy were paclitaxel at 80mg/m2 and trastuzumab at 4mg/kg load followed by 2mg/kg, both given weekly for twelve weeks. The experimental arm consisted of TDM1 given at 3.6mg/kg and pertuzumab given as an 840mg flat dose load followed by 420mg and these drugs were given every three weeks for four doses.

Here’s the CONSORT diagram for the TDM1 pertuzumab arm. Overall on the trail we screened 1,540 patients and randomised 878 who met our eligibility criteria. Of these, 249 were HER2 positive, 165 were randomised to one of the other concurrently randomised arms, leaving us with 52 patients who were randomised to TDM1 and pertuzumab and 32 patients who were randomised to control. One of those patients did not receive allocated intervention so we’ll report on 31 of those control patients today.

Here are the characteristics of our study population. Notably they are well matched for age, race, ethnicity, hormone receptor status and MRI tumour diameter at the time of diagnosis. I’ll just point out to you that approximately two-thirds of the women in these arms were hormone receptor positive.

So here are the results and let me just step back for a moment and make sure you’ll understand exactly how we report these. First, we provide an estimated likelihood of achieving a pathologic complete response for both the control arm and for the experimental arm. Because of the Bayesian predictive probabilities, these are simply estimates and it’s important to look at the entire distribution. We then look at whether this probability of success in the investigational arm is superior to that in the control arm and we look at the probability of success of this arm in a subsequent phase III trial. So if you look at the top line, for all two HER2 positive patients we estimated that complete pathological response rate in the control group as 22% and for the patients receiving TDM1 and pertuzumab as 52%. That led to a probability that TDM1 pertuzumab was superior to TH of 99.5% and a probability that it would be successful in a subsequent phase III trial of 94%.

If we then look separately at the hormone receptor subtypes we can see first for the hormone receptor negative patients we see that the estimated pathologic complete response rate in the control arm was 33%, that for the TDM1 pertuzumab arm was 64% with a probability that TDM1 pertuzumab was superior to TH of 98% and a probability that it would be successful in this subtype in a phase III trial of 90%. Then, finally, for hormone receptor positive HER2 positive patients we see that the estimated PCR rate was 17% in control and 46% in the TDM1 pertuzumab arm for a 99% probability that TDM1 pertuzumab was better than TH in this subgroup and the probability of success in phase III of 93%.

Now, just to briefly tell you about the adverse events. First we’re going to compare the two drugs in terms of just the time when they’re receiving the experimental agents. If you’ll remember back to the schema that was the first half of the trial before they received the AC chemotherapy. What you can see here is that there are actually very few grade 3 or 4, the most severe types of toxicities, in either group. We also saw that for all grade toxicities there was very little difference in myelosuppression, fatigue or diarrhoea. However, we did see a much greater rate of hypertension, neuropathy and alopecia among women who took TH compared to TDM1 and pertuzumab and I would note that the rate of alopecia was relatively low here because there was a lot of use of cold caps within our patient population. Then, as would be expected, the rate of ALT and AST elevation was seen in excess in the TDM1 pertuzumab arm but was low grade and not clinically significant.

Then when we look at all of the treatment together we see a very similar picture with very little grade 3/4 toxicity, a bit more in the way of neutropenia and thrombocytopenia of low grade in the TDM1 arm and again the excess in hypertension, neuropathy and alopecia seen in the TH arm.

So, in conclusion, TDM1 pertuzumab graduated from ISPY2 both in the overall HER2 positive patient population as well as in both hormone receptor subsets. It had a very different toxicity profile than that seen with the standard of care, TH. Neuropathy, hypertension and alopecia were much less common and I would argue that these are side effects of treatment that really matter to women, that can really affect their day to day function and we have quality of life data that was collected concurrently that’s in the midst of being evaluated.

Moreover, these results demonstrate the benefit of the ISPY2 adaptive design and standing platform. We can show that we can see this benefit in a relatively modest number of patients and, in fact, using a standard platform where we can continuously evaluate drugs enables us to get results out very quickly. In fact, the last patient just came off study within the past month. So we were able to get these results out very, very quickly.

So where does this leave us? There are numerous phase III trials that are going on now to help establish the role of TDM1 with or without pertuzumab in the neoadjuvant setting and we anxiously await those results. But we are encouraged by this and are looking forward to adding new combinations to TDM1 as a less toxic backbone, including immunotherapies and other very targeted therapies. Importantly I think this gives us an opportunity to start really thinking about tailoring therapy, particularly within the neoadjuvant setting, where we can see that reaching a pathologic complete response has true meaning for longer term outcomes and could help us to start to give up some of the more toxic chemotherapies like paclitaxel that we have used in the past.