Round 1: After progression on a CDK4/6 inhibitor, should an actionable mutation determine treatment selection in PIK3CA wild-type disease?

The debate opens with a patient with PIK3CA wild-type disease progressing on an aromatase inhibitor plus a CDK4/6 inhibitor. The panelists debate whether a newly identified actionable mutation should direct the next treatment or instead inform an individualized sequencing decision that also considers the overall evidence, clinical context, and patient factors. They then examine whether randomized evidence in PIK3CA wild-type disease supports considering a broader PAM-pathway–targeted approach at this post-progression decision point.

To hear arguments related to pathway inhibition and regimen intensity, see Round 2.

To hear arguments related to route of administration and patient preference, see Round 3.

To view the full series, click here.

Seth Wander, MD, PhD: So, round 1: After progression on a CDK4/6 inhibitor, should an actionable mutation determine treatment selection in PIK3CA wild-type disease? So this is another way of, I think, asking the question, should we sequence any more in the era of geda [gedatolisib]? You could argue, based on the data from the VIKTORIA-1 study that we're about to see, that geda looks quite good regardless of PIK3CA status and regardless of ESR1 status, because that included ESR1 and wild-type. So in the near future, should we actually be sequencing as much as we have in the past, looking for some of these things? 

Here's a patient that we should think about. A patient with PIK3CA wild-type, ER-positive/HER2-negative advanced disease, does okay on an AI and a CDK inhibitor for about a year and a half. So she's more than a year, but she's less than that sort of 2+ year median that you would see on the frontline CDK trials. She's still a candidate for endocrine-based therapy, no visceral crisis. So what do we do both, sort of, with and without an actionable mutation? Okay, I'll give it to VK first. 

VK Gadi, MD, PhD: All right. Thanks so much, Seth. That was a nice overview. I want everybody to do a little exercise with me. If you can reach into your backpacks, your swag bag, your purses, and take out the map of the city of New York. Can you do that? No one actually has that, right? Now, can you take out your phone and open up Google Maps or whatever and show me the traffic pattern outside? Yeah, everybody can do that. So when we think about genomics and cancer in general, we are very good at laying out the map. We can all do that. We can say, "This is the mutations, these are the alterations." What we don't have is a functional map of those changes. And so, we're left in the situation where we know where the roads are laid out, but we don't know where the traffic is, if that makes sense. Does that make sense? So this is the challenge in oncology. 

Now, there are people who are developing models, myself included and others, where we functionalize cancers. We grow patient-derived organoids, we do all these fancy things, and it's remarkable when we do that work. Yeah, we know this mutation exists, but yet it doesn't seem to have any consequence on the growth of the cancer, because we can inhibit it and we don't see something. Oh, but this other gene that's not even mutated, it's apparently super critical. So we're at this transition point where we're stuck between these static understandings of cancers and these functional understandings of cancers. And until that time comes where we're fully understanding, it just undermines that we should put all our faith in these mutations. 

So that's just kind of an opening volley for you to deal with, which is we call these things actionable mutations and they're defined by their actionability because we have a molecule that hits it. But does that actionable have anything to do with the cancer you're dealing with? And that's the knowledge gap we have. So I say this because we know in particular with PIK3CA, it's a truncal event. We've all been talking about that 40% of patients at time of diagnosis, up to 40% may have an alteration in this pathway as defined by mutation. But many of those patients do just great and we've never even targeted or even had the tools or the means to target it. Now we do. And the question is, how often is it an actual driver of the disease? 

This is a little bit different than what you see with the ESR1, which I do believe we should still test for, because that is not a mutation that is there at baseline. But for the appearance of that mutation, these cancers were previously responding to our endocrine therapies that we were offering them. And now we have these really powerful SERDs. It's not that fulvestrant is a bad SERD, it's actually a good SERD, we just can't give enough of it to a human being to get the job done. So we've always been targeting this in a meaningful way and it has actually led to improvements. So just keep that in mind as we think about it. 

And that's the part about understanding what actionability means. The other part is none of these things happen in every clone of the cancer immediately, right? So these are things that over time we're selecting for. It's evolution of the fittest in some sense. And at some point, those minor populations become the dominant sources of growth in the tumor. And we don't know when that's happening, so we don't know when we should switch one horse that's effective at controlling all the disease to focus on a new target while giving up gains in the other cells potentially. So we have to retain what we've been doing effectively everywhere else, while then also damaging the new areas of growth. And so, sometimes it becomes a picking game that becomes hard. I've not provided any data to point. This is just general thinking. 

So now let's look at some of the data, which is clinical trials. Now, clinical trials are easy to beat up. They're the best path forward we have for asking nice, clean questions, and they inform what could be happening in patients that we see in front of us. But keep in mind, the vast majority of the patients we see in clinic don't actually even meet eligibility criteria for many of the studies. So there's a misalignment with what we generate from idealized clinical trial populations and then the folks we really see in front of us. And so, this is another kind of phenomena that makes it hard, especially in the PIK3CA space where so many of the drugs have toxicities related to things like hyperglycemia. It's funny to me that in the first-line setting, I've yet to be able to find a patient that met the A1c criteria for getting inavolisib in the setting of a known PIK3CA inhibitor, PIK3CA mutation. 

And then lastly, treatment sequencing should be individualized. As I just said, everybody's patterns of growth, et cetera, we can put our thinking caps on and kind of see what's going on with a patient. Single areas of metastatic progression are things we can kind of pick and choose and target with other forms of therapy, while retaining our overall practice of kind of systemic therapy without having to necessarily dump everything in favor of a single site. That's just an example of individualized, although that patient probably qualified for the trial that led to the registration of some of these things. So we may do different things than what we see in the clinical trial practices. 

So this is just like a clarion to say, "Yeah, let's keep still looking at the patient in front of us, make good decisions for that patient based on the tempo of their disease, the characteristics of how they're feeling. If they're feeling well, do we need to jump?" Maybe you're tracking a ctDNA number and it's going up, but not dramatically so. So maybe you have a little time to stay on the same thing and let a patient take a vacation or whatever she's trying to do beforehand. So anyway, these are all kind of general themes that I think about when I'm thinking about sequencing therapies. 

So this is now looking at the PIK3CA wild-type cohort and the efficacy in the VIKTORIA-1 clinical trial. Again, we're going to see more of the details of this trial in the subsequent slide, but essentially this is a study with 3 arms, a fulvestrant alone arm in patients who had progressed on a prior line of therapy, often involving a CDK4/6 inhibitor. There's a doublet with gedatolisib and palbociclib, sorry, and fulvestrant. And then a triplet with gedatolisib, fulvestrant, and palbociclib. And you're looking at the response rates clearly better than what you get with fulvestrant alone. And then also, the duration of response is better, but also the PFS. So just checking off all the boxes, this is clear evidence that in this group of patients who are wild-type for PIK3CA, this could become an option. I'm smirking up here. Can everybody see me smirk? Because in the trial, it was not actually PIK3CA wild-type. It was PIK3CA not detected. And that's a very subtle distinction, but it's an important distinction. 

So did I go forward? Yes. So some other treatment sequencing considerations, as we discussed on briefly, disease tempo matters. How long were they on the prior line of therapy? In other words, is it this but for? Did something happen that triggered that event that's most likely to have led to that progression or were they on that therapy for a really long time, and again, but for this emergence? Then you can stamp on that thing and kind of get control as opposed to it was always there the whole time. So these are things I think about. And then treatment fit, we just touched briefly on A1c. That's making it very difficult in some of the patients who have PIK3CA mutations to select therapies. But now with gedatolisib in particular, we'll go through that in a little bit more detail in subsequent slides, that's a really well-tolerated drug from that perspective in patients who may have complications for that. 

There are other toxicities. I think notably with these pathways, we worry about things like mucositis and skin toxicity. I'll fully disclose, I did not have this particular study at my institution, but I did have the phase 1 trial and I enrolled quite a number of patients. And the pattern of that toxicity is interesting. It does happen, but it extinguishes itself. It's more like what you see with EGFR inhibitors in the old days where you just kind of had to get them through a few cycles and they would tolerate it. And then obviously we've talked a little bit about mode of administration. We're going to talk about that a little bit more. And then lastly, this idea of the PAM pathway. So that stands for PIK3CA/AKT/mTOR. And we're going to see a map here in just a bit showing the resiliency in this pathway that allows cancers to get very clever and work around any single target we might hit. So this is more like a comprehensive strategy that may inhibit resistance in several pathways at the same time. Good luck, sir. 

Paolo Tarantino, MD, PhD: Let me put on my boxing gloves here and tell you a story. A few years ago, I was in residency and fellowship in Italy, in Milan, and we had this phase 1 trial, this phase 1 trial testing a combination of cisplatin and gedatolisib for metastatic triple-negative breast cancer. Now, it sounds a bit crazy, because now we kind of think of this drug, gedatolisib in nature, positive disease, but we had a few patients that had incredible responses, long-lasting responses for the subtype with the poorest outcomes in breast oncology, triple-negative disease. Eventually it didn't turn out to lead to a phase 3 trial and approval, mostly because we were not able to identify those patients. We didn't have a biomarker. We didn't have something telling us these patients with metastatic triple-negative breast cancer will have amazing responses to platinum and gedatolisib. 

At that time, I also had a professor who was called kind of like me, Paolo Casali. He was an expert in sarcoma management. And I remember he was showing us all of these slides of patients with sarcomas, sometimes of the limbs, and showing all of these algorithms, where based on these sarcomas, he was discussing with the patients if it was worse to remove a limb or not to do an extensive surgery with some lower risk of recurrence or less extensive. And one thing that he was conveying us at the time was that patients respond differently. You can show them the same scan, the same percentages, and they have different ideas about if it's worth it to do a larger surgery or not. And that's because we're not treating tumors. We're treating patients with tumors. 

And so, I do think when it comes to breast oncology, we're so much luckier than Paolo Casali is. Paolo Tarantino is so much luckier because he has all of these biomarkers that help us in 2 different ways. The first way is that whenever you sequence the genome of a tumor, and maybe I should switch slide, whenever you sequence the geno of a tumor, you understand the underlying biology of the tumor. First of all, you understand what is going to be the prognosis of the patient, if that sounds like a nasty tumor based on the mutations or a more gentle one. And then you understand what's going on in response to treatment. Why did you give the patient letrozole and ribociclib and after 18 months the tumor progressed? You can decide to close your eyes and just give some treatment or actually try to understand what's going on there. As Seth was mentioning, if the tumor found its way around the estrogen pathway to an ESR1 mutation or develop some additional mutation, some that can be targeted or not, but I think we want to understand that to understand how to counsel the patients. 

But then the other part is actually being able to offer options to our patients. So you can think of this as the map of New York, or you can think of it as the map of Times Square and all of these are big shiny options that the patients have. But if you turn all of the options down, all of the lights out, Times Square is not Times Square anymore. And so, I do feel after a patient progressed on first-line endocrine therapy and CDK, if you do the attack on the ESR1 mutation, you know that that patient has been for a while on first line, maybe the patient could get a third monotherapy and have a brilliant performance with very minimal side effects, or you can boost that further with targeted treatment. 

PIK3CA mutations tend to be clonal, but actually AKT1 and P10 can emerge and you can target them with capivasertib. BRCA mutations, either germinal, somatic, are actionable and can emerge over time. ERB2 mutations, I think we don't talk about them enough, but so many lobular tumors especially have ERB2 mutations, and we have NCCN endorsement for tucatinib/trastuzumab, neratinib/trastuzumab/fulvestrant, so many options. I imagine taking them all away, and you only have the ones at the bottom, which can be very effective. Gedatolisib, we've seen that the PFS reaches up almost to 10 months, which is amazing, but you do want to have any option. You want to lay down the options for the patients and discuss, "Would you want to try a third monotherapy, or would you want to try a triplet or a doublet or instead go toward the ERB2 mutations and see if a TKI could help?" This is the only moment where you can do that. You cannot do that in sixth line after the patient having received ADCs, chemotherapy, and all that. 

And so, I think you don't want to miss that opportunity to really tailor. And I do think that's important, because we have once again, third monotherapy that in the right population can reach up to 8 months PFS. Patients that have done well in the first-line with mutations, you have imlunestrant/abemaciclib just approved. Another effective combo for patients with ESR1 mutations. In the future we may have giredestrant/everolimus as well with nice performance also, especially in ESR1 mutations. All of these options are out if you don't check for ESR1 mutations. And so, I do think, do we even need to do genomic sequencing? We need to discuss, but personally I feel I want Times Square with all the lights out. I want my options when I see the patient. I want to be able to tailor, and the only way to tailor is to sequence the tumor. 

Gadi: So after each round, do we go into our corner and get [inaudible]. 

Wander: Yeah, that's right. That's right. 

Gadi: The bruising and everything. 

Wander: All right. That was a good first round. Very good. It's going to get more heated. I mean, I appreciate all these points. I mean, I think this is a little bit of an academic argument, but the point remains that, especially for our friends in the community setting, all of the speed with which we're seeing these new drugs developed in breast cancer is happening all across the board in other areas. And it is an important debate to think about newer drugs like geda that clearly have activity across the board, regardless of PIK3CA, regardless of ESR1, versus increasing degrees of personalization. There's a clear split. Just think about PIK3CA. You have these newer next-generation mutant-specific PI3K inhibitors versus now the PAM-pathway inhibitor. And there's going to be advantages and disadvantages to both approaches in terms of how you administer it, in terms of the toxicity profile, and in terms of the biologic approach to resistance that could occur. 

The more specific you are, potentially the less toxic, maybe the easier to administer, but the more bypass pathways you might have even within the same node. The broader you go, the more you risk a little bit more toxicity, but you're cutting off the escape mechanisms for the tumor downstream in the case of PIK3CA, where we know P10 is a resistance driver. So I think there are translational research arguments to be made. There are patient-focused sort of toxicity and practicality arguments to be made.