Activity Transcript

William J. Gradishar, MD: Hello, I'm Bill Gradishar, Chief of Hematology and Oncology and Professor of Oncology at Northwestern University Feinberg School of Medicine. Welcome to this program titled "Recent Updates in Hormone Receptor-Positive HER2-Negative Early Breast Cancer: A Deep Dive into Understanding the Data and Their Implications." 

Joining me today is Stephen Johnston, who is Professor of Breast Cancer Medicine and Head of Breast Oncology at the Royal Marsden and the Institute of Cancer Research in London. Welcome, Stephen. 

Stephen R.D. Johnston, MA, FRCP, PhD: Hi Bill. Thank you very much for asking me to join. 

Dr Gradishar: Today, we're going to spend some time discussing current data on the treatment of hormone receptor (HR)-positive, human epidermal growth factor receptor 2 (HER2)-negative early-stage breast cancer. It's important to recognize that although we've made significant advances in recent years clearly demonstrating that the use of endocrine therapy (ET) in this population reduces the risk of recurrence, we have many questions that remain open to further investigation. Those include identifying the patients who are at the highest risk of recurrence, because now we have some maneuvers that we can undertake that potentially could reduce the risk for those patients. But again, we don't want to apply those strategies to every patient. 

We also need to understand who are the patients that have primary endocrine resistance and may not benefit from our usual strategies of monotherapy with ET. Finally, we have to have an understanding of how we prevent or delay recurrences by adding newer therapies to our treatment regimens. Of course, as we think about what's been presented in the last few years, we know that there are new drugs that can be added to monotherapy, not unlike what we do in metastatic disease, at least in certain populations, that have further reduced the risk of disease recurrence. 

We know from several data sets that you can define in crude ways the risk of recurrence based on volume of disease. I often rely on the example of the Pan data in the New England Journal of Medicine. Looking at patients who'd received 5 years of tamoxifen, we can look at low-risk patients all the way up to high-risk patients based on the volume of disease, either node-negative smaller tumors all the way up to bigger tumors with a larger number of lymph nodes. As you follow those patients out over time, 5, 10, 15 and 20 years, we see an increasing fraction of patients even in the low-risk as well as in the high-risk group who will develop recurrence very late in the course of follow up. Obviously the higher the risk, the greater the probability that a patient will develop disease recurrence. So, understanding that and also potentially using other molecular tools, we may be able to refine our understanding of which patients need more therapy and those who can avoid therapy.

The other thing that we've come to appreciate in the last few years, and Stephen will be talking about this in a few minutes, is that with the addition of new therapies, many of which were developed first in the metastatic disease setting, the cyclin-dependent kinase (CDK) 4/6 inhibitors as an example, we now have those drugs entering into the clinic in the early-stage setting in high-risk patients, specifically abemaciclib. Furthermore, we also know that although poly(ADP-ribose) polymerase (PARP) inhibitors are often thought of as being most useful in patients perhaps with triple-negative disease where BRCA mutations have higher frequency, we know that in the universe of estrogen receptor (ER)-positive patients, there are more patients that harbor BRCA mutations that may be candidates for a PARP inhibitor to further reduce the risk of recurrence. These are but 2 examples where adding drugs to our typical endocrine strategy for reducing risk may further enhance the effect and eventually reduce the number of patients developing metastatic disease.

We now have guidelines that help us think about how we introduce or integrate these therapies into our adjuvant programs. If you think about abemaciclib, which will be discussed in a moment by Stephen, we have different versions of the recommendations. Some of them are based solely on clinical features such as the number of lymph nodes involved or whether or not there's a higher expression of Ki-67. We'll look at how these differences may influence what we do in the clinic. 

I want to sort of turn to Stephen because he's done a lot of work, particularly with the monarchE trial, and this led to the approval of abemaciclib in the adjuvant setting. I'd like to get his thoughts on how the guidelines may differ a little bit between organizations and perhaps how we integrate these into the clinic. How do we utilize these? 

Dr Johnston: Sure, thank you Bill. I mean, I think as you say, and we'll go into detail in a minute, the monarchE trial was designed to look at those high-risk patients. As you have clearly pointed out, the clinical pathological features of nodal burden, grade, and tumor size are 3 things that we know about our patients discussing our weekly tumor boards and allow us to rapidly identify those high-risk patients.

The addition of Ki-67, which was an exploratory aim in the trial, was to see whether that allowed us to look at biological resistance to ET, higher risk, more luminal B-type cancers, and whether they were at even greater risk over and above the clinical pathological features. 

We'll look at the data, the new data, as well as high and low Ki-67. But Bill, in the US, has the discrepancy between the FDA and the requirement for a high Ki-67 on top of the clinical pathological features vs American Society of Clinical Oncology (ASCO®) and National Comprehensive Cancer Network (NCCN)'s guidelines being more in line with the clinical trial with the straight high-risk clinical features, has that caused a lot of confusion? Do you think that's likely to change when the new data are looked at in detail? 

Dr Gradishar: It may. I mean, right now in the US, if something gets approved and the NCCN as an example, or ASCO®, adopts it into their guidelines, what we're often restricted by more than anything is insurance coverage. As long as we fit the criteria that ASCO® or NCCN approved, we're usually okay. I wouldn't be surprised, and obviously I'm not a spokesman for the FDA, that those recommendations may undergo some refinement over time.

I wanted to turn back a question to you though as well, and that is this whole notion of looking at Ki-67. I certainly understand the rationale for why it was done, but some of the expert breast pathologists will often question the reliability of scoring Ki-67. There's that gap between 5% and 30% where they're not as sure about what it really represents or how reliable the reading is. Do you find that to be a potential problem with these criteria? 

Dr Johnston: Yeah, I think it's one of the major issues. I mean, there are standardized guidelines on how to measure Ki-67, the assay, the antibody to use and so on. But you're absolutely right. The key issue is the scoring. Everyone can tell you a low Ki-67, less than 5% is a good prognosis. Everyone can tell you a high Ki-67 of 130, is a poor prognosis. It's that bit in the middle. In the trial it was all centrally analyzed, and the cutoff was 20%. But, if somebody's got high-risk features and a Ki-67 of 15%, perhaps somebody else's analysis of the tumor would give perhaps a 20% score. There is an element of subjectiveness by the pathologist. There was quantitative scoring by image analysis done within the trial. But in regular community practice, that's not going to happen. 

You get variability particularly around the cutoff. That's why elsewhere where the approval of the drug is regardless of Ki-67 based on the clinical pathological features, measuring Ki-67 is not a problem because it's not required for approval of the drug. But as we'll talk about, the Ki-67 I think was chosen initially by the FDA because it was the really high-risk patients. We'll look at the latest data. When they were balancing a new drug with side effects and whether or not there was a survival signal, perhaps they were being cautious in perhaps just going for the very, very high-risk patients first until the data matured. But as we'll see, the data are maturing very nicely now and that may tip the view. 

Dr Gradishar: Well, Stephen, you updated us at San Antonio just a few days ago about the updated data from monarchE. I think to anybody's eye, it offers reassurance. Would you mind going through that data? 

Dr Johnston: Sure. Great. Thank you. As we all know, adjuvant abemaciclib was approved a year or so ago based on the monarchE trial showing a significant improvement in both invasive disease-free and distant relapse-free survival in these high-risk patient populations. But the initial analysis median follow up was only 15 months. People thought that was really very early. Subsequent analysis at the time of approval of 27 months did show sustained benefit. But I think people were really waiting to see whether beyond the 2 years of treatment period, whether that benefit was sustained.

So, the new analysis was a pre-planned overall survival interim analysis defined to occur 2 years after the primary outcome data had been analyzed and presented. All patients are now off of abemaciclib. The median follow-up is now out to 42 months. The analysis now includes a definitive 4-year landmark analysis. There are now 835 events that contribute to this study as opposed to 320 events when the first data were announced. 

Just to remind everybody about the design of the trial, it had 2 cohorts of high-risk node positive patients. They were defined in cohort 1, which was 91% of the patients, as those with 4 or more nodes, or with 1 to 3 nodes, they had to have an additional factor, either grade 3 tumor or a large tumor size, T3, more than 5 centimeters. That has, as Bill has shown previously, been correlated with that 30% risk of recurrence on ET alone at 5 years. These are the high-risk patients. The vast majority in the trial, 95% had received chemo, either in the neoadjuvant or adjuvant setting. Despite standard of care, historical data suggests that these patients still have a risk of recurrence of up to 25%, 30% in the first 5 years. Therefore, we can often do something better if we select these patients and look at a new drug.

There was a second smaller cohort, an exploratory cohort, added later at the request of the regulators to look at those intermediate risk factors. So, 1 to 3 nodes, grade 1 or grade 2 tumors, T1 or T2. So not the big high-grade cancers, but the only factor was a high Ki-67. Now that's a smaller cohort and we don't have enough data in that analysis yet, and there were less than 9% of patients in the trial. But we did then measure Ki-67 in all of cohort 1, which we'll talk about in a second.

So, after standard of care treatment – surgery, chemo, radiotherapy – patients were randomized to an ET with or without abemaciclib for 2 years, the so-called treatment period, and then adjuvant ET to continue for between 3 and 8 years. The primary outcome was invasive disease-free survival (IDFS), key secondary endpoints, overall survival, distant relapse-free survival, and safety and it's important to remember that patients were staged at the outset to show that there was no evidence of metastatic disease.

The new Kaplan-Meier data curves that we presented at San Antonio last week show the updated analysis of this trial. The first thing to notice is the primary endpoint of IDFS. The Kaplan-Meier curves, which were separating early, continue to separate and appear to widen in terms of the absolute difference. Now at 4 years, we have an absolute difference of 6.4% between ET alone and ET plus abemaciclib. A hazard ratio of 0.664 suggests a 33.6% reduction in risk of developing an event, that measures the area between the curves. This 4-year analysis is 2 years after patients have finished treatment. So, there's no suggestion that off drug the curves are starting to come back together. If anything, they appear to be widening. This may indicate what we call a carryover effect, which we've seen with ETs before. That was the key headline message from the new data. It didn't really matter which subgroups of patients. There appears to be benefit in all the predefined subgroups.

When you look at distant relapse-free survival, which excludes local recurrence, it just talks about metastatic recurrences or death for any cause. Again, we see a similar separation of the curves and a degree of benefit that appears to be getting bigger as time goes by.

There was an analysis of overall survival because this was an interim overall survival analysis, but the overall survival data at this point are immature. That's because only 330 deaths have occurred. It's less than 6% of the patients in the population. In nearly any adjuvant study, you have to follow the patients for a long time to see what the impact of that early breast cancer treatment is on the overall survival. Having said that, numerically there are fewer deaths, 157 vs 173 with ET alone. The point estimate on the hazard ratio is now 0.929. Originally in the previous analyses it was hovering around the 1 mark. Now it is moving to suggest that there may start to be separation of those curves. I showed the analysis of patients who had a distant relapse event, which was death for any cause or evidence of metastatic disease. When we look at the number of patients that have relapsed with metastatic disease and are currently alive receiving treatment, there are half as many in the abemaciclib and ET arm compared to ET. Over time, that may well translate into an impact on overall survival. But we know in metastatic luminal breast cancer, patients can continue ET for a period of time.

When we did a step analysis or a piece-wise analysis of the hazard ratios, we could see that the hazard ratios were getting smaller year by year. This was an exploratory analysis, but another way of quantifying the fact that the curves appear to be separating.

Now the final comment I'll make, Bill, is about the Ki-67 data, because as I said, it was measured in the patients who were in cohort 1. It's important to see whether or not in that cohort there appears to be equal benefit or not in the high Ki-67 and the low Ki-67 groups. The analysis here shows that again, in the high Ki-67 patients, which were half of the cohort 1 patients, over 2000 patients, that they have the greatest risk of recurrence. Indeed, on ET alone, a quarter of those patients, 25%, have actually already relapsed at 4 years. That is significantly reduced with the addition of abemaciclib. It does appear that the addition of abemaciclib works very well in the high Ki-67 patients.

But if you look at the low Ki-67 group, we actually see a relatively similar degree of benefit. The curves are actually also separating such that the hazard ratio for low Ki-67 is 0.624. It's 0.618 for the high Ki-67. While Ki-67 is strongly prognostic and it does identify a higher-risk group on top of the clinical pathological features, for low Ki-67, their events are fewer because they've got slightly better prognosis. But there is still separation of the curves and benefits in the low Ki-67. So that is the more mature data. That will be interesting to see what the regulatory authority's view on that is.

There were no new safety signals. The safety profile of the drug is well known to clinicians. It's about early and prompt management of diarrhea, dose adjustments and dose holds, which were common -- 43% and 60% had an adjustment in dose or dose hold at some time. If you get the patient right on the drug in the first 2 or 3 months, find the dose that's right for them, that allows them to stay on study for the 2 years. By staying on study, that's why I think we then get the degree of benefit.

So, Bill, the new data I think are very reassuring that the benefit is not only persisting, but if anything, appears to be getting wider. I don't know what your reaction or thought to that was, whether you were surprised to see the curve separating or feel that is biologically consistent with what we might be achieving here. 

Dr Gradishar: Well, I was happily surprised. I mean, I think that data, particularly with what does look like a carryover effect and an expanding benefit with time is going to result in probably more patients going on the drug as they probably should. I think the other thing that many people were probably skeptical about is what you closed on. That is, if you manage the toxicity, you can keep patients on the trial. I don't know if you know off the top of your head, you probably do, what the fraction of patients who actually discontinued was, but it was actually relatively modest compared to some of the other adjuvant trials. 

Dr Johnston: It was 18% overall, but certainly discontinuations due to diarrhea, for example, was only 5%. Now that may be patients coming off the back of chemo, they're exhausted, they're tired, and new drug that gives them side effects here they decide it's just not for them. But I think the key education point is that if you take that time and effort in the first 3 months to work with the patient, maybe adjust the dose, then you're more likely to be able to keep them on study. The frequency of adverse events drops quite markedly after the first 2 to 3 months. 

Dr Gradishar: So, one of the other things that I touched on at the beginning, and I just want to get your comment before I forget I said it, is the other drug that's potentially useful in this population if they have a BRCA mutation is the PARP inhibitor olaparib. When you think about a high-risk ER-positive population with the data that you just presented as well as the data from the trial looking at olaparib, how do you think about using one over the other or perhaps both in sequence? 

Dr Johnston: Yeah, I mean, I get asked that a lot. In the Olympia trial, there were only 17% that were ER-positive. It was driven mainly by the triple negatives. But having said that, the ER-positive patients can be BRCA positive and the point estimate on the hazard ratio for benefit from olaparib and it has a survival signal as you know was moving in that right direction. You'd be hard pressed for a BRCA-mutated patient to deny them access to olaparib. Now, olaparib was a 1 year treatment. Abemaciclib is a 2. Both cause a degree of anemia and minor suppression. So, combinations are not an option at all. What I've said when asked this question is I would probably go for the olaparib first because that's 1 year of treatment and that's the BRCA, if the BRCA is really driving them. But if they're a high-risk node-positive patient with lots of nodes, it's very tempting to then give the abemaciclib in sequence after that. 

Having said that, in the trial, they came in within 3 months of starting their ET and there aren't any trials or studies of delayed abemaciclib. But, if you look at that ET alone curve, it's incremental rates of progression on a steady basis, and therefore it's hard to believe that it wouldn't have an impact if it came in a year or more later if they're still at risk. I think if you are faced with those rare scenarios of a high-risk BRCA ER-positive patient, sequential therapy would be the way forward rather than one or the other, but certainly not combination. 

Dr Gradishar: Yeah, and I agree. I think this is the kind of thing that as you pointed out when talking about toxicity with the patient right from the get-go and getting it right, you have to make patients who might be eligible for both drugs aware of the consideration that you may sequence these drugs over time, not sneak up on them after a year or 2 years of therapy and tell them they've got another drug to go. Because I think it is going to take a highly motivated patient, but it'll also be the patient that's at an elevated risk. 

Let's move on to just a brief discussion of some of the other trials that currently are out there that we're aware of from data that's been presented in the past. We know that the PALLAS trial and PENELOPE trial were previously presented. PALLAS being a big trial including patients with stage 2/3 disease having received palbociclib with ET. The duration of therapy was 2 years. In this trial, the discontinuation rate, since we were on that theme just a moment ago, was 42%. So, very different than what we're seeing in monarchE. But disappointingly what we're seeing is no benefit from the addition of palbociclib to ET in that population. IDFS and distant relapse-free survival were pretty much the same and the statisticians are pretty convinced that even with longer follow up, we're unlikely to see much of a benefit. 

Whereas PENELOPE-B a smaller trial with about 1,250 patients. These were patients who had residual disease after getting neoadjuvant chemotherapy. So, by virtue of having residual disease they're at a higher risk of developing a problem downstream. The same thing in this case. Patients received palbociclib plus ET or ET alone. Again, in this trial, even though numerically when we first looked at the data, there was some suggestion maybe the palbociclib would add something. At the end of the day when you're looking at 2, 3 and 4 year follow up with respect to disease-free survival, it's pretty much a wash. So, the palbociclib trials have not shown a benefit.

The one that's still hanging out there that we haven't seen yet is the NATALEE trial, another very large trial, about 5,000 patients, who receive ET with or without ribociclib, in this case for 3 years. We don't have data from this trial. We anticipate that will be coming, but we'll have to determine whether or not the drug makes any difference.

Which really leads to sort of the last piece that I'd like to get your opinion on. The metastatic disease setting is usually where we start developing drugs in the advanced disease setting. We saw at the outset a uniform positive effect with adding a CDK4/6 inhibitor to ET, really pushing the progression-free survival (PFS) out whether first- or second-line therapy. Maybe not so much, or maybe that's changing a bit as we look at survival data. But I'd like to get your take on why you think, at least in the adjuvant setting, why these outcomes of these trials, at least the palbociclib and abemaciclib, are so different with respect to improving outcome or not. 

Dr Johnston: Yeah, no, I've been thinking about it a lot since the preliminary data came out 2 years ago. There were really potentially 4 reasons that have been bandied around by everybody in the area. The first, as you say, was whether the high discontinuations were a problem. In the PALLAS trial there were very strict criteria on the development of neutropenia, which is more common with both palbociclib and ribociclib than with abemaciclib. If you had grade 3 or 4, you had to dose hold and reduce dose. If it occurred again, 1 more dose reduction. If it occurred again, then you had to come off the study. In clinical practice, we often will manage these patients very differently, keep them on study even with a very mild grade 3 neutropenia if they're well. But in the protocol they had to come off. So that was disappointing that the discontinuations were so high. But an analysis of the PALLAS team has looked at even those patients who stayed on drug compared to those who came off, there was no benefit. So, it's very hard to argue that it was the discontinuations that were the problem.

The second thought was around the population study because as you point out, Bill, they had some node-negative patients, about 17%. They had stage 2 diseases. It wasn't the high-risk node positive patients. Were they treating a lower risk group? Well, that was potentially an explanation, but even if they looked at the high-risk patients, the same patients that were in the monarchE trial, it was about 56% in PALLAS trial. Even in those high-risk patients there was no benefit.

The third thought comes to the potential differences between the drugs, either in the schedules of the drug or the potencies of the drug. In terms of the potencies, there is a greater potency against CDK4 for abemaciclib compared to palbociclib and ribociclib. There's a broader kinase inhibition. It inhibits some other kinases wider than just CDK 4 and 6. It could have other effects. But as you said quite rightly, in metastatic disease all 3 drugs work exactly the same. The same degree of improvement and the same hazard ratios for PFS in the first and second line. So, it's hard to say that the drugs are really very different in that respect.

Which brings us to the final issue about dosing. I really think this is the potential key reason why there are differences. Continuous dosing is what you do with abemaciclib. Intermittent dosing is what you do with palbociclib and ribociclib. In neoadjuvant studies where we've studied both drugs, palbociclib and abemaciclib before surgery, if you look at the proliferation marker Ki-67, it's switched off with the CDK inhibitor. But as soon as you come off the drug, and with palbociclib, that's what you do anyway. But we've done it with abemaciclib in patients before the surgery, then the Ki-67 in the primary tumor rises back up. You need to be on the drug to have the Ki-67 switched off. Now you could say, okay, in metastatic disease, it doesn't seem to make a difference. The tumors still shrink, the PFS benefit is the same. But that's a big amount of disease. You get stable disease where the tumors don't grow or change very much but you very rarely eradicate the disease. In early breast cancer, you are dealing with microscopic disease which you're trying to eliminate to cure and prevent the cancer from coming back. It could be that the kinetics of tiny, microscopic cells are very different in relation to continuous dosing vs intermittent dosing where you may or may not be eliminating all the cells if you have an on off approach. 

Now the answer to that will be NATALEE, because NATALEE is the same schedule. It's 3 weeks on, 1 week off. So if NATALEE is positive, it kind of debunks that theory about continuous dosing being the answer in early breast cancer. NATALEE is looking, however, at 3 years, not 2 years, so would longer be better? It's also looking at a lower dose, 400 as opposed to 600, so their discontinuations due to hematological toxicity may be less. So, we await with enormous interest the results of NATALEE which might help answer this conundrum of why there are differences.

Dr Gradishar: So let's sort of come to the closure of this and what I'd like to do is just offer my own thoughts for a moment based on much of the discussion we've had. I think the data that you presented at San Antonio is pretty compelling and it really positions abemaciclib for that population of patients who have higher risk as something that we should be considering adding to ET. Certainly, with the growing improvement, not static. The lines aren't coming together, they're actually moving further apart. It's a great argument for considering abemaciclib in that population. The PENELOPE and PALLAS trials, we may hear yet further data from these trials as the data matures, but as pointed out earlier, it's pretty convincing from the statistical standpoint that they're unlikely to see much of a difference between the treatment arms in those trials. 

The big one still sitting out there that we're waiting on, as you just discussed Stephen, is the NATALEE trial. That'll not only address the duration question, but the pharmacology, whether continuous therapy is important or not. With abemaciclib, it's continuous, palbociclib ribociclib, it's not. So, we'll see. That data is expected in the next year or so and we'll find if that influences our practice with the new data that emerges from those trials. So, Stephen, I want to get your final thoughts as well. What are you thinking about where we're going? 

Dr Johnston: Yes, I think coming back to how we started this conversation, Bill, it's about really big advances we've made in ER-positive breast cancer over the years. But in terms of a new therapy to add that overcomes this high-risk patient population that are relapsing early, this is, I think, a big breakthrough in terms of a new treatment that clearly works. The data are maturing very nicely, are clearly getting stronger, and therefore it's easy and straightforward to identify these patients based on high-risk, node-positive clinical pathological features. 

I think the Ki-67 data are also maturing very nicely that there's benefit regardless of whether you have a high proliferation or a low proliferation of Ki-67. We're doing a big translational piece in the trial to do a much more in-depth analysis of the gene expression profile, whole exome sequencing, plasma, DNA, everything that is sort of modern in breast cancer biology to deeply understand who benefits or not. But I think Ki-67 doesn't need to be a discriminator as to who does or doesn't benefit. You can choose the patients based on their characteristics of 4 or more nodes or of 1 to 3 nodes, high-grade or high tumor size. That is a reasonable number of patients in our practice. 

Dr Gradishar: Well, that's great and I want to thank you, Stephen. It's been a great discussion. I enjoyed participating in it. I want to thank all of you listening for participating in this activity. Please continue on to answer the questions that follow and complete the evaluation. Thank you very much. 

This transcript has been edited for style and clarity.