I'm Melissa McShane. I'm one of the breast medical oncologists at Fox Chase, and I'm gonna be covering the medical oncology aspects abstracts that were covered for ER positive HER2 negative early breast cancer at San Antonio. So I'm gonna start with the Ladier trial which looked at the oral, um, SED, uh, Gorgestri, and then I'm going to tackle artificial intelligence in the adjuvant setting. So please don't fall asleep. Um, uh, here we go. So, and I will bring it back to the case at the end. So, the Ladier breast cancer trial, um, this was looking at Goroestrin versus standard of care endocrine therapy as adjuvant treatment for ER positive HER2 negative early breast cancer. So, as many of us know, Geroestrin is a potent next-generation oral SED. It has full ER antagonists, and it's designed to drive deep and sustained, um, ER signaling. And, um, for the trainees in the room or those who are not aware, you know, aromatase inhibitors, they decrease estrogen levels. Uh, tamoxifen is a modulator and it, uh, modulates and blocks the estrogen receptor, but SEDS degrade it, and so they, they produce more potent ER, um, signaling blockade through this, um, downward pathway. And it also is able to overcome an ESR one mutation. So in the study design, um, there's a few things I want to point out about this study. So the key eligibility criteria, pre and postmenopausal women were included. Um, this is a higher-risk patient population. You had to be node positive, and if you were node negative, you had to have another high-risk feature. So if your tumor was over 1 centimeter, you had to have a grade 3, a T 67 over 20%, um, or a high score on a genomic assay, um, or you could be T4N0. So you had to be high risk. And if you were pre-menopausal, Um, and you received Gerardresin, you had to receive ovarian suppression, which is in, in line with the standard of care for all SERDS. So Falvestrin or other oral SEDs, um, the label does include that you have to be on ovarian suppression. So they're randomized 1 to 1 to GRresin or standard of care endocrine therapy, and this could be tamoxifen or an AI plus or minus ovarian suppression if you are premenopausal. So just some baseline demographics I wanted to point out, 60% of the patients were postmenopausal, uh, 40% were premenopausal, uh, 78% of the patients were node positive, and 80% of patients received chemotherapy. So again, just, um, showing that it's a higher risk patient population that we're included. So the primary endpoint was invasive disease-free survival. And if you can see, there's a significant benefit to gerdran. Um, the median follow-up was 32.3 months, and so, 92.4% of patients in the jadetroon versus 89.6, um, were free of invasive disease. Uh, the hazard ratio was 0.7. They also did this exploratory analysis and looking at it, um, compared to the standard of care endocrine based on tamoxifen or an AI. And as you can see, gerridestran, um, improved, was, uh, preferred in both when compared to AI or tamoxifen. Secondary endpoint was distant recurrence-free survival, um, and had a statistically significant hazard ratio of 0.69. So again, 32.3 months, um, 94.4% versus 92.1% in the standard of care arm. The overall survival is immature, um, but a positive trend is being observed. Whether that will continue to be observed, time will tell, uh, but overall survival testing will come at future analysis, but just to show you what the curves are looking at thus far. And I think it's important because this is a new medication potentially bringing it to the adjuvant setting to really understand what are the adverse events compared to our standard of care. And so, um. The, a big thing to note is that the adverse events leading to a treatment discontinuation was lower in the geodetroon. It was 5.3% of patients, so very low overall, but it was higher in the standard of care arm at 8.2%. Arthralgias were the most common, followed by hot flashes, which makes sense. Um, and if you look at the top right here, arthralgias, um, being the reason for discontinuation was lower in geodetri arm compared to standard care. Um, so 1.6% versus 3.7%. Um, another, uh, uh, adverse event that was more common in the geoderon arm was bradycardia. It was a low proportion overall, but of those 95% of those that had bradycardia were all grade one, which means that they were, um, asymptomatic. They didn't require a drug hold. They didn't require treatment. Um, they were just picked up on routine monitoring. There were no grade 3 or grade 4 events of bradycardia. And there were um an increase in grade 3 or grade 4 VTEs in the standard of care arm, which makes sense because tamoxifen was included in that. So, in conclusions from this trial, this is the first trial to demonstrate benefit with a novel endocrine therapy, uh, in early breast cancer since approval of the AIs in the 2000s. Uh, the median follow-up at 32.3 months already is showing an improved invasive disease-free survival compared to standard of care with the HAS ratio of 0.7, and the discontinuation rate of geodetriment was lower compared to standard of care, showing that it's very well tolerated. So considerations for this to think about premenopausal women do require ovarian suppression. That's not something to, you know, forget, um, and so, uh, there, but there are ongoing trials actually looking at omission of ovarian suppression in SEDS for premenopausal patients. That's not ready yet, but that might be coming. There is a substudy of Ladira looking at the combination of abemiiclib and, um. GRdestran, which will, um, which we'll talk about with the role of CDK46 inhibitors. Um, and should everyone get it, or should we just restrict it to this higher, uh, risk patient population. And then the role of CDK 46 inhibitors. So in this trial, patients were not allowed to be on a CDK 46 inhibitor. They could get a 12-week, um, they could have been on it for up to 12 weeks prior to randomization, but that was it. And so, our standard of care for this patient that we just talked about who's node positive would be to offer a CDK 46 inhibitor. And so how does that play a role with what we're going to be doing if this does get approved? And then the second half of my talk in biomarker testing and AI. Um, can we use that to determine who would benefit from an oral SERD? And then also, do we need to extend therapy? Do we need to go to 10 years of an oral SERD for high risk or 5 years enough because it's just a better ER, um, target and suppressor? So things to consider. So for our patient, she's postmenopausal, she's hormone positive, HER2 negative. She's a T2N1A. So we didn't provide the oncotype, but she may meet criteria for chemotherapy. Um, and after she gets that, she would also meet criteria for an adjuvant CDK 46, and she would also meet criteria for geodestrian. So which do you choose? Um, do you get them together off-label? Um, we have safety data for a lot of the other oral certs in the metastatic setting. Um, and like I said, there's an ongoing substudy looking at geridetri with the bemicycllib, or do you sequence them? There's a lot of sequencing trials going on right now with AI or tamoxifen that then go to oral cert later on, um, or do you do the AI and CDK 46 and then do the oral cert afterwards? Obviously, this is all uncharted territory, but this is how we're gonna be having to think about it when we're incorporating this into our patient population. So I don't have a good answer for her. Well, I do. It's not approved, so she wouldn't be getting it. But, um, give me like 6 to 12 months and then obviously we're really going to be having to have these discussions with our patients. So that was the easier part. Now I'm gonna go into our artificial intelligence in the adjuvant setting. Um, I am trying to go through this in a time friendly uh fashion, and, um, I've done a lot of research, so hopefully it works, um, but I wanna start with this schematic, um. And so I, I thought this was a good way to, to illustrate how AI models presented at San Antonio are structured. So they're integrating multiple data streams to support clinical decision making in early breast cancer. So think about it, if genomics tells us, uh, what the tumor is and pathology tells us what it looks like, uh, multimodal deep learning through AI tells us how it behaves. So the top are the key data inputs. Each of these capture a different dimension of the tumor biology. So you have the pathology, which is whole slide H&E imaging that allows the AI to analyze tumor architecture, stromal features, immune infiltration, spatial patterns that you can't see with the eye. Then you have the clinical data, which is your standard, like tumor grade, nodal status, the treatment received. It really, um, solidifies the, the real-world, um, aspect of the clinical medicine. And then you have the, um, the genomic data, so the molecular assays that, that look at the, the tumor on a more genomic level. And then what happens is that it goes into this, this, um, multimodal deep learning, um, AI algorithm that, that looks for patterns across, across it. So you're not feeding them pre-defined specifications, but allowing the AI to really use the data set to determine patterns and associate them with outcome. And so, um, then this, this, this then goes into the clinical decision making. So you have recurrence risk, um, you know, is this patient going to recur earlier or later? You have chemotherapy benefit. Do they, do they benefit from chemotherapy or not and extended endocrine therapy. So the goal of this is to really personalize treatment decisions, um, in our, in our clinics for breast cancer patients. I don't think you can do that. OK. So, um, this is the, these are the abstracts that I'll be covering during this, um, uh, uh, section, um, and the data sets that were used in multiple of these were included Taylor RX, NSABPB20, and, um, B42. Um, and so, let me get right into it. So this is by Doctor Sperano. This is in the general session, um, and this was looking at multimodal artificial intelligence, integrating image, clinical and molecular data for predicting early and late breast cancer recurrence in Taylor RX. I tried to simplify this as much as possible. So basically, the objective of this analysis was to develop a diagnostic test with better distant recurrence risk prognostication than the oncotype DX21 gene. So this was his schematic of how they did it, but basically, they utilize samples from Taylor RX to build this model. And so they used, um, it's called the multimodal ICM plus model, uh, but they use pathomic imaging, they used clinical covariates that are listed up there. And then they used uh molecular data. So this included a 42 gene signature, included genes from the 21 gene recurrence score, the BCI gene, and then also the Endoredict. And this was, um, developed and utilized through a partnership between ECOG, Akron, and Carris Life Sciences. And so this, um, I'm gonna talk a little bit about the model to better understand it, but it developed a continuous risk score, which was designated into high risk or low risk based on the model. So, um, How is this done? I think it was very hard to follow in the presentation, but there were two sets that were used, and I'm gonna explain a little bit of how it works. So you have the training five-fold cross-validation set. So this was, um, of the patients in Taylor X that were available with their imaging and their molecular data, they took 3, 2/3 of them and put them in. The training set, and they took another 1/3 for the validation set. So the training set is what is used to build the model. So it learns which features matter and which, which way to give to them. They don't know the outcomes, and they just look at the information that's given without the outcomes of what actually happened to the patients, or sorry, with the outcomes of what happened to the patients. So it learns what to associate features with, whether recurrence occurred, when it occurred. And then they had the holdout validation set. Now, this is completely excluded from the model development. So it's not used for feature selection or waiting or tuning. The final locked model that was made based on the training is then applied to this other set of patients to see without knowing their outcomes. And then you look at what they, what it told you they would do and see if the prediction predictions generated were accurate. So, this is a very simplistic view of what they found. So they use the C index, which is the concordance index, and this is to quantify performance. So what this means is that it reflects how well the model, the AI model, correctly ranked a patient's recurrence by risk over time. So if it comes out at 0.5, it's no better than chance. If it's at 0.6, it's moderate, but once it gets to 0.7 or above, it's pretty good discrimination. One is perfection, extraordinarily rare. Um, and so if you look here, you see that in the, in both the training and the holdout validation, the ICM plus model outperformed the oncotype in both overall distant recurrence and late distant recurrence. So it better predicted patients that were going to recur as compared to the as compared to the oncotype DX. Um, and so it outperformed them on all these sites. So what are the key take-home points? It improves recurrence prediction. Um, it integrates pathology images, clinical molecular data, and it performs the single modality tools that we currently use for prediction. Um, it outperformed Alcotype DX in regards to, um, recurrence risk, and, um, it just, it very well, um, uh, had a strong performance for late recurrence risk. So this addresses a major unmet need in hormone positive early-stage breast cancer, um, and will likely shape our future of how we, um, determine risk for these patients. So, the second abstract, um, this looked at lobular carcinoma. Um, the two things that it looked for was basically, can we assess lobular versus ductal with AI, um, and can we do it just as well as the pathologists do? And then also looking at the tumor immune biomarkers in Taylor X, can we better, uh, assign risk for these patients? Um, that may have low genomic scores that were seen in Taylor RX. And so, um, both centralized pathology review and the CDH1, which is the AI, um, review of classifying, which was lobular, consistently showed that there was a higher risk of recurrence for non-lobular carcinoma between years 5 to 15. And so the curves, as you can see in both of these, so then the left is, is the histology by the pathologist, and then the right is by the AI. And you can see the curves start to separate at 5 years. Um, and of note, in Taylor X, most patients only got 5 years of endocrine therapy. The discordances between pathology, histology, and AI were low, only 6.8%. Did they, were they discordant on what was lobular versus ductal. And then they also looked at this TE TME risk score, um, and found that in these lower genomic risk patients, right, they were less than 25, those that were lobular versus ductal, with this TME risk score for AI they were able to further stratify recurrence risks, especially, um, in both ductal and lobular, but even more so in lobular. So the whole conclusion from the pathologist, um, who gave this presentation was that within patients who have a low genomic risk score, like they're less than 25, you may find a subset of patients who, who based on TME biomarkers are actually not low risk and, um, may, may benefit from additional therapy, whether that be extended endocrine therapy. So key take-home points, lobular breast cancer carries higher late recurrence risk than non-lobular disease. An AI derived TME analysis provides independent risk stratification, um, in addition to oncotype. And then the findings could have implications for duration of endocrine therapy in lobular patients. Due to time, I'm gonna skip the next abstract, um, which looked at, um, chemotherapy benefit in NSABPB20. Um, and the reason we skip it is because, because of time, but also the take-home point is that they weren't really able to show, um, based on this data that, um, they saw numerically that there was, um, a benefit in this high-risk model that, um, patients who had a higher risk score, who got chemotherapy did better than those who had a high risk score and got tamoxifen. Um, but it wasn't statistically significant unless you're over the age of 50. So, um, but, so this is like, I'd say a hypothesis generating result versus an actual, um, practice changing result. And then the last abstract I'm going to go through is um a multimodal M3T model um that was looked at NSA was trained in NSABPB42 and validated in Taylor RX. And so what this looked at was distant late recurrence in hormone positive early breast cancer. Um, this trial, uh, B42 looked at hormone-positive breast cancer that had gotten 5 years of endocrine therapy, and the question was, do you benefit from an additional 5 years of endocrine therapy? The outcomes from that trial showed, yeah, by about 2 to 3%, you would benefit. Um, and so they wanted to better further investigate, can we really figure out which patients actually did benefit from endocrine therapy extended. And what they found is that if they put the algorithm in and they had this M3T algorithm, and it was high or low risk, which is most of these AI, um, algorithms that are presented here go into a continuous risk score that is either high or low. And they found in the high-risk patient population, there was an absolute benefit of, of, um, 4.09% of extended endocrine therapy compared to placebo. But what was really interesting is in when they broke it down to the node negative versus node positive. And what they found is that in node negative, no matter what your risk score was, you actually didn't benefit that much from extended endocrine therapy if you were node negative. But if you're node positive, you did benefit if your risk was high, but you didn't if your risk was low. So this data implicates that you could de-escalate treatment in node positive where you may have extended therapy if their M3T score is low, um, and then confirms that if they're node positive and their M3T score is high, they would benefit from extended endocrine therapy. Um, they also validated this model in, uh, Taylor RX, and they better predicted the risk of late distant recurrence. And so if you had a high risk score, um, you had a higher risk of a distant recurrence by about 6.35% compared to if your risk was, if your number was low. And it was also, um, an independent predictor of distant recurrence along with tumor size and grade when you looked at the multivariate analysis. So, the take-home points for this is that this model identified low-risk patients unlikely to obtain meaningful benefit from extended endocrine therapy, which is big for our, um, clinical decision-making, um, potential treatment guidance beyond standard clinical pathological features or factors, and then external validation and Taylorx confirmed independently, um, distant recurrence prognostication. So what do these collectively show? Um, AI consistently has been improves prognostic accuracy. It adds biologic insight beyond genomics alone. It also has strong signals for late recurrence prediction, chemotherapy benefit, and biologic tumor classification. And AI is definitely moving from more exploratory into decision support in our clinics, and it has a potential impact on CDK 46 inhibitor selection, extended endocrine therapy, chemo de-escalation. So, San Antonio firmly positions AI um as a complementary layer to genomics and enhances precision across prognosis, prediction, and long-term risk assessment. So how does it apply to our patient? Well, not yet, because it's not for prime time, but I would say for her, um, she meets criteria of two of the trials that were presented that the scores were, um, looked at. And so the N3T score could help determine if she would actually benefit from extended endocrine therapy, given that she was no positive, as long as she doesn't get geodetrin cause that wasn't included in that patient population. Um, and then the ICM plus model could also help maybe better predict her distant recurrence for us to have that conversation with her oncologist, um, and for closer monitoring. So, um, that was heavy, but, uh, I think that it applies to our place very nicely, and yeah. We'll move on to my next. I think you're up next. All right, good evening all. Uh, thanks for joining us. Uh, I'll have the privilege of talking about the surgical trials that were, uh, presented at San Antonio in relation to ER positive HER2 negative early breast cancer. I do have two disclosures. The first, um, isn't relevant, and the second is, yes, it's still about the axilla. Uh, the first trial that I'll present is the alternate trial, which is Alliance A011106, uh, which really, uh, looked at the use of neoadjuvant endocrine therapy. And we know that we have great paradigms that really guide our, uh, surgical management of the axilla after neoadjuvant chemotherapy. And that's much better defined than paradigms after the use of neoadjuvant endocrine therapy. Secondly, we have little expectation, as in the patient that was hormone receptor positive and node positive in this case of a nodal PCR after neoadjuvant endocrine therapy. So it provides some challenges in the management. And so what the alternate study was, was a phase 3 randomized trial enrolling postmenopausal stage 2 to 3 ER positive HER2 negative breast cancer patients. Uh, and they all received 24 weeks of either anastrozole, Fulvestrin, or a combination of both. They then went on to have surgical management per standard of care, which I'm unclear what that means, and you probably will be after you see the data as well. Uh, so baseline characteristics of the 933 patients, uh, who were enrolled in the trial, you can see that, uh, 3-quarters of them were T2. Uh, about 2/3 were node negative and really well represented, uh, proportion of lobular cancers, about 25% of them. Uh, when we look at the, um, the biopsy results again, about 1/3 were, were positive or, or atypical. Surgeons were required to, before beginning neoadjuvant endocrine therapy, determine whether or not the patient was a breast conserving surgery candidate or not, and then this 2x2 table shows you what ultimately the surgery was performed. So about 20% of the patients who were BCS candidates, even to start with, ultimately had a mastectomy, but nearly half of patients who were BCS who were not BCS candidates only could do a mastectomy, converted to BCS eligible, and underwent breast conserving surgery after that 24 weeks of neoadjuvant endocrine therapy. 10% of the clinically node positive downstage to either node negative or micrometastatic disease. And in the 382 sentinel node biopsy positive patients, 80% ultimately had only 1 to 2 positive sentinel nodes. And so we harken back to our surgical de-escalation trials with That number. 46% underwent axillary dissection, but only a half of those who had 3 or more positive sentinel nodes underwent axillary dissection, highlighting, I don't know what standard of care is if the surgeons here, uh, were not performing axillary dissection in, in, in some of those patients. Uh, axillary dissection is far more common in patients undergoing mastectomy. So, throw the kitchen sink at the, at the patients is what those surgeons, uh, were probably doing. 2/3, um, versus just 1/3, uh, of those who underwent breast conserving surgery. And axillary dissection was also much more common in patients who had a positive pre-neoadjuvant endocrine therapy biopsy, meaning that if the sentinel nodes were positive but wasn't known preoperatively, those patients were much less likely to undergo axillary dissection. What did we find in patients who underwent axillary dissection? Well, you can see in the patients with one positive sentinel node, uh, about 2/3, between 1/2 and 2/3, ultimately had no additional positive lymph nodes if they went on to axillary dissection. But if we look at the far right of this table, not an insignificant. The proportion of patients in both 1 and 2 or more positive sentinel lymph nodes had more lymph node disease at the time of axillary lymph node dissection, about 50 if you had 2 or more positive sentinel nodes. So the nodal burden was not insignificant in the study. So in a summary of alternate and really this analysis of the surgical outcomes, we, we still lack a standardized surgical approach to the exo after neoadjuvant endocrine therapy. I think these data help us, but You can see at least half of the patients with 2 or more positive sentinel nodes had a non-insignificant nodal burden, and these have implications for adjuvant therapy. So I think we have to be very careful just saying, you know, we can de-escalate in these folks. And obviously long-term follow-up is needed in those who underwent sentinel lymph node biopsy alone to really understand how this impacted their local regional recurrence. We'll move on to, uh, a study, Taxis, uh, which really does, um, dovetail well into the case in that this patient has a palpable lymph node, hormone receptor positive, where we don't expect a great response to neoadjuvant chemotherapy. And so we're left deciding how to manage the axilla, uh, if we do upfront surgery. So, what the Taxis trial asks is whether tailored axillary surgery or task. is inferior to axillary lymph node dissection in clinically node positive patients. And this is, uh, uh, an initial analysis of these data, really looking at the nodal disease burden in these patients. So, what is tailored axillary surgery? Tailored axillary surgery is a sentinel lymph node biopsy with removal of the clipped, biopsy-proven lymph node and any palpable disease in the axilla. And the rationale for tailored axillary surgery is that if we don't perform an axillary lymph node dissection, we're reducing the nodal burden to a degree where this is treatable by both radiation and systemic therapy. Uh, we are still awaiting the Alliance data, the A011202 trial, where these patients were all post neoadjuvant chemotherapy patients who had residual nodal disease and were randomized to axillary lymph node dissection or no further axillary surgery. Taxes, on the other hand, included both patients who had residual nodal disease after neoadjuvant chemotherapy, but also those undergoing upfront surgery. So our patient in the case would have fit the, uh, the inclusion criteria very well. And we completed accrual to taxes in in just this past December. Taxis, what we're calling Taxis 2.0, but it's really called NOAC, uh, is now. Opening at sites and it's enrolling only the upfront surgery candidates because I think this is really where we're likely to, uh, move the, the needle. 2300 patients have been screened to taxis with 1500 patients randomized, 60 sites in 13 countries. Fox Chase was the third in the US to participate in taxis, and they were randomized to either axillary lymph node dissection or no further dissection and comprehensive. Nodal radiation, and you can see 2/3 of the patients in tax underwent upfront surgery, and a third were those after neoadjuvant chemotherapy. To give you a sense of who the patients were, who were included in taxis, 80% hormone receptor positive, HER2 negative. Makes total sense. They undergo upfront surgery and they're if they're node positive, and they have residual nodal disease if they had neoadjuvant chemotherapy. Makes total sense. Uh, and 87% were clinically N1. When looking at the nodal disease in the Taxis trial patients, 60% had additional node positive or positive nodes rather at the time of axillary lymph node dissection, 55% in the neoadjuvant cohort and 64% in the upfront surgery cohort. 24%, so a quarter of patients had 4 or more additional positive nodes removed at the time of axillary lymph node dissection, and this is after removal of all palpable disease, the clipped node, and any blue or hot sentinel lymph nodes. When we look at nodal stage between the groups, you can see no axillary lymph node dissection on the left and axillary lymph node dissection on the right. Three-quarter of those who underwent no further axillary lymph node surgery were either, um, micrometastatic disease or pathologically N1, but only 50% of those Who underwent axillary lymph node dissection. So that's a difference of 25% of patients who were understaged surgically, meaning that we didn't know that they were N2 or N3 pathologically because we didn't remove the additional lymph nodes to prove that. And you can see the vast majority having macro metastatic disease. So, Taxis is the most progressive axillary surgery, surgery de-escalation trial, meaning the highest nodal burden that's ever been studied in axillary, uh, lymph node surgery de-escalation. 60% of patients had additional nodal disease and 25% were under-stage. So, there are likely implications here for some adjuvant therapy decisions. But at the interim analysis, there were no safety concerns, um, and this has been open for a number of years, uh, and we know that axillary recurrence is, uh, is an early, uh, recurrence. Long term outcomes obviously are going to be essential. This just closed, uh, as I said, in December, so we've got a couple of years until we have full long-term outcomes. Not so much applicable to our case, but, uh, an important study nonetheless, BOOG 1308. Uh, what we know is that most clinically node negative patients are going to ultimately have a negative sentinel lymph node biopsy, and we know that nodal status has taken a backseat to biology as it comes to many of the, the adjuvant therapies. This is one of multiple studies, sound and sema nautilus, etc. investigating or really questioning, uh, the utility of axillary surgery in these patients. Uh, it, they enrolled 1700 patients. They were clinically T1 to T2 N0. They all had a negative preoperative ultrasound, just like the other studies, undergoing breast-conserving surgery and whole breast irradiation, randomized as in the other studies to undergo routine sentinel node biopsy or omission of sentinel lymph node biopsy. Who were these patients? You'll not be surprised to see that they matched the, uh, the patients in the other studies in that they were largely postmenopausal hormone receptor positive patients, the mean age being 62, with 89% being over 50, 83% T1, 83% low grade or intermediate grade, and 87% hormone receptor positive. In the sentinel node biopsy cohort, 86% were ultimately pathologically node negative, with 6% having micrometastatic disease and only 8% of patients having macro metastases. But since this is a randomized study, then 8% of the patients who had no axillary surgery had untreated or unremoved macrometastatic disease in their axilla, and to no one's surprise, you can't tell a difference between the curves. When we look at the other outcomes of interest, again, no sentinel node biopsy on the left, sentinel node biopsy on the right. There's no difference between, uh, regional recurrence-free survival, uh, regional recurrence, which was 1% if, uh, axillary surgery was omitted, and there are 5 years of data in this study, and distant disease-free survival was also equivalent. So, emission was non-inferior to sentinel node biopsy. Interestingly, let's look at the adjuvant therapies. Half of patients in both groups had no additional therapy afterwards, so they were really undertreated, uh, in, in the vast majority of situations. And if we look also down at endocrine therapy, fewer than 50% of patients received adjuvant endocrine therapy in this study. Chemotherapy, there was no difference, and targeted therapy was the small amount of HER2 positive patients who were enrolled in this study. So the results of this study confirmed the non-inferiority in well-selected patients of sentinel lymph node biopsy omission. There was a low rate of macrometastatic disease when preoperative axillary ultrasound is negative, and this is an interesting study in that it questions whether or not endocrine therapy is actually mandatory, even when sentinel lymph node biopsy is omitted. Speaking of, uh, omission of axillary surgery, the secondary analysis of Ema was presented also at San Antonio this year. And to refresh your mind, similar schema to what we've just reviewed for the BOG study, but much larger study, 5000 patients randomized to sentinel node biopsy. Or no sentinel node biopsy in a 1 to 4 randomization. The 1 to 4 randomization was specifically for the second randomization, which we're going to talk about tonight, and just to refresh your mind, no difference between those groups, um, in invasive disease-free survival. What the authors were interested in are the patients who underwent sentinel node biopsy and had 1 to 3 macro metastases, which was 11% of the cohort. And these patients were then further randomized to either receive no additional axillary surgery or an axillary lymph node dissection. And you say, I thought we've already put this issue to bed. Well, Uh, let's review what the inclusion criteria were for all of our axillary surgery de-escalation trials, starting with Z11, 1 to 2 positive sentinel nodes, but 65% had macrometastatic disease. So a third of those patients had only micrometastatic disease. So, really a, a, a Favorable population. And if you look down the table, most of these 1 to 2 positive sentinel nodes and only Senneak had all macrometastatic disease. So what Ecima was asking is, in this higher risk population, can we still omit axillary lymph node dissection, and is it non-inferior? Looking at the groups, you can see that they're well balanced in terms of the number of positive sentinel lymph nodes, and 11% of the group who underwent completion dissection had 4 to 9 positive nodes in total once dissection was performed, and you say, Hey, uh, does that mean in the patients that had no sentinel node biopsy, that 10% of them had N2 disease? That's not true. Remember, this is 11% of the sentinel node biopsy cohort that had 1 to 3 macro metastatic macro metastases. So it's only about 1% of the overall cohort that had N2 disease. When we look at the, uh, the uh results in terms of invasive disease-free survival, we can see that the curves do, uh, do diverge at around 36 months. So we, when we look at the hazard ratio, it does cross the 95% confidence interval does cross one. So we can't say that these are statistically different. However, non-inferiority could not be shown. And I think that the surgeons in the audience look like this right now, but let's, let's look at what made up these invasive disease-free survival events. So, importantly, axillary recurrence happened in only one of the patients enrolled here. That was only 0.5% in sentinel lymph node biopsy. And all of the events that composed the invasive disease-free survival included things like distant relapse, 7% in the sentinel node biopsy group. 4% in the axillary lymph node dissection group, and there was no difference in um in death between the two cohorts. There were issues here as well. The planned recruitment number, especially of those who had 1 to 3 macro metastases, was not met. So this was underpowered. Uh, there was a bunch of treatment crossover. So, the different, there were differences between the intention to treat and the per protocol analysis, and there were significant differences in, in radiotherapy between the groups, uh, as well. So, this is the highest nodal burden of all the Z11 validation studies, as I've shown you. There was no significant difference that we were able to see in this analysis of invasive disease-free survival or overall survival, but we could not demonstrate non-inferiority in this study. 10-year outcomes are going to continue to be reported, so we'll all be on the lookout here for understanding whether or not um our de-escalation strategy applies to this um more high-risk group. I believe that's where I am. All right, good evening, everyone. Um, so I'm Rebecca Shulman. I will be giving the radiation oncology update. Um, there is one study that applies to our patient here, um. And this was the primary results for the HR positive HER2 negative cohort of TBCRC 053, or the PRAD study, which was a randomized trial of no low or high dose preoperative radiation with pembrolizumab and chemotherapy in node positive HER2 negative breast cancer. So the background here is that, um, we all know that radiation kills cancer cells, but radiation is an inflammatory event. There are a lot of immune properties of radiation and activates both innate and adaptive immune responses as this, this, uh, slide depicts. Um, and the question is whether or not we can take that to use that to our advantage. Can we exploit those immunogenic properties of radiation? Uh, radiation can synergize with immune checkpoint inhibitors. Um, so there are certainly pre-clinical data to support that, um, largely in, in triple negative breast cancer, but also in hormone receptor-positive breast cancer. Um, but clinically this has really been seen in other disease sites, not breast cancer. So the Pacific trial in locally advanced non-small cell lung cancer really showed a significant benefit to the addition of chemo, radiation, immunotherapy. Um, but what about in ER positive breast cancer? So this study really looked at trying to repurpose radiation for immune stimulation in node positive early stage breast cancer. So the standard, uh, of course, is usually doing radiation at the end. So patients get surgery, they get, uh, chemotherapy, and they get radiation afterwards. Um, certainly when patients decide, uh, to or are candidates for breast conservation. In surgery, they get a boost dose to the radiation to the lumpectomy site. Um, and the authors here, uh, thought that they could use the boost radiation, uh, at the beginning, um, prior to chemotherapy and surgery in a way of priming the immune system. Uh, in the MOC space, there, there are data to support using immune checkpoint inhibitors in HR positive HER2 early stage breast cancer. Recently reported the checkmate. 7FL and the keynote 756 studies showed that there was an increase in past ER with immune checkpoint inhibitors added to neoadjuvant chemotherapy. Uh, but what we also saw was that this wasn't true for all patients. Um, so what they found was that, um, the patients that really benefited from this combination were patients with high stromal tills, uh, and also high PDL1 expression. So the key question here is, does the addition of neoadjuvant radiation improve these response rates? Uh, here's the, the PRAD clinical trial. Uh, so again, they analyzed the high risk HR positive, uh, HER2 negative patients. Um, they had high risk tumor features, so grade 3 or high KI-67. Um, a significant number of patients had clinical N2N3 disease. Uh, patients either received no radiation, a low dose of radiation, so 3 gray times 3, or a higher dose of radiation, 8 gray times 3. All patients received pembrolizumab, neoadjuvant chemotherapy, and then had surgery. Um, they also collected, uh, tumor tissue 2 weeks after radiation. Uh, very importantly, is that they, uh, were, they ensured that they did not irradiate the lymph node. Um, they only targeted the primary breast tumor. Statistically, their primary endpoint was breast tumor T cell infiltration at the two-week biopsy measured as a rank-based immunoscore. They also had a co-primary endpoint of YPN 0 rate. Uh, they had secondary endpoints that composite PCR and RCB, um, as well as exploratory analysis, um, transcriptional signatures of response. Um, patient had disease characteristics, uh, so the majority of patients were T3, um, followed by T2, um, the, the majority were N1, 50%, followed by a quarter being N2 and a quarter being N3, uh, and the majority were grade 3, but they also did have 35% of patients that were grade two. here's really the, the result of the study, which was that neoadjuvant radiation, 24 gray only, not the, the lower dose radiation. led to increased T cell infiltration. So in the Pembroke chemo patients, it was 31% versus with the addition of radiation, it was 53%. And they found that T cell responders induced PDL1 expression. In terms of surgical outcomes, um, the focal radiation plus the immune checkpoint blockade led to nodal clearance in about a third of patients. Um, you can see the trends here in the YPN 0, the PCRs, um, and the RCB. There's a trend towards, um, increases and improvements in outcomes with the addition of radiation. They did also have exploratory analysis, predictors of PA CR in the patients that received radiation, uh, and they, what they found was that the PCR significantly correlated with non-luminelli type, uh, tumors and PDL1 expression. So the interpretation and conclusions here really were that 24 gray preoperative radiation with, with, uh, Pembrose significantly increases T cell infiltration and node positive higher risk, uh, hormone receptor positive HER2 negative early-stage breast cancer. There were encouraging rates of past CR in the radiation plus immune checkpoint blockade arm. Uh, this table here just high. Lights, uh, the several studies recently that have looked at this combination chemo immunotherapy plus minus radiation. Um, in this, in this study, the PRAD study, there were actually lower than predicted PCR rates. You can see there are the 25% with the addition of radiation versus the 6% with just chemo immuno, which was much lower than the chenote and checkmate study. Uh, the hypotheses there were that the, the PRAD study included more luminal A type tumors and patients with lower PDL1 expressions at baseline. So future directions, clinical considerations, uh, for those in the audience that treat, uh, malignancies other than breast cancer, you know that there's a trend towards, um, preoperative radiation, um, you know, certainly in GI malignancies or sarcomas, um. We see that you can deliver lower doses. There could be smaller radiation fields, um, and as a result, fewer long-term radiation side effects. Um, future trials examining this combination of pre-op radiation plus immune checkpoint blockade in this patient population are needed to clarify the disease. Control benefit, uh, but hopefully in the future, um, for select patients, there will be a neoadjuvant radiation plus immune checkpoint blockade combination that allows, uh, an increase in PDL1 expression and potentially a de-escalation of some of the systemic therapy or, uh, axillary surgery. Wow, that was amazing. I think we can try to take 1 or 2 questions before we move on. If anybody wants to be brave enough. Well, I have one. So, Rebecca, am I sending you a patient for neoadjuvant radiation with pembrolizumab? This is like brand new information. I would love that. But, um, but no, I think it's, it's current, it's still largely exploratory. Um, I do think that as, um, you know, patient selection improves for this very heterogeneous group of hormone receptor positive patients, that, uh, there will be select patients where we could potentially use this neoadjuvant approach, um, uh, to potentially, you know, like I said, de-escalate either some of the immune immunotherapy or the axillary surgery. I'm even wondering about the patient who comes in with the large breast mass, 56 centimeters that's not eligible for, you know, breast conservation therapy. Is this going to be an easier way for me to get them there, us to get them there? Yeah, I think that's a great point, and I thought that's what's interesting about the study is that the majority of patients were T3 tumors. Um, so that's, you know, large tumors that are getting, uh, preoperative radiation. Um, so, yes, certainly, I think, I think if you wait long enough and you combine with, with a good systemic agent, uh, we can see these tumors respond. Anyone else? OK. So then my question goes to Doctor Williams. So, you have a patient that gets neoadjuvant radiation therapy to a large 5 centimeter breast mass. How does that impact your surgery? That's certainly concerning for uh morbidity, for wound infection, all of these things. Um, interestingly, there was a poster at um at San Antonio looking at some preoperative radiotherapy for inflammatory breast cancer and in speaking with those and it, uh, being done at memorial. Speaking with those authors, they really don't have, uh, they haven't seen any surgical issues, uh, more so than they would otherwise, knowing that all of those patients are undergoing mastectomy, uh, not breast conservation. But, you know, I, I think we don't know until we try. Yeah, absolutely. I mean, Melissa, I'm thinking you have this patient with this really large breast mass who doesn't want chemotherapy, and we get to give them pembrolizumab with radiation, and then Doctor Williams takes the tumor out and their oncotype score is 8, and I'm like, I think we won the game. I think many of the patients on that trial did get chemotherapy, right, in new admin, so I don't think it completely fits, but maybe they could decline it after radiation and we'll, we'll go off trial, um, but yeah, I think it just allows more like stratification of patients that don't completely fit, you know, what we can currently do, and so it gives more options for those. I, I really like the idea of the bigger tumors that aren't going right to surgery. Maybe the AI modeling will help us with this. Maybe we'll see. Maybe next time. All right, thank you so much, guys.
Related Presenters
