Matthew Smith: Our next speaker is Gert Attard. And Gert will comment on the role of biology in informing management in this disease state. Gert.

Gert Attard: I'm now regretting not slipping in the photo of our lovely trip to Davos. For next time. So thank you [inaudible] for asking me to discuss this topic. These are my disclosures. The question the panelists were asked to discuss that relates to the subject of my talk is that, outside a clinical trial will the information of tumor genomic profiling, primary tumor or metastatic disease influence your first line treatment of mHSPC? And I thought this was a straightforward negative answer, but this is APCCC and the panel are astute in identifying controversy. There may be discussion later. I will explore three messages through my talk. The first is yes, there're ongoing phase three trials for mHSPC patients where molecular selection is required. Two, I believe, or I'll explore the role for prognostic tests for patients treated with ADT plus ARSI, or I think ARPI has been chosen as the preferred term for APCCC and then the discussion on the requirement for selecting patients for docetaxel using predictive biomarkers.

So I'm all for molecular testing for patients who are being considered for trials. And here's a few examples, many others planned I'm sure I've missed a couple, but I am not convinced to have level one evidence to use a molecular test to decide treatment. I think that's something we will discuss APCCC 2025, 2026. So, moving on to the second message, we had consensus earlier for the addition of abiraterone for two years in the STAMPEDE high risk nonmetastatic patients. And as Nick implied in one of the comments, we were not expecting such a large magnitude of benefit. And when you look across the disease settings for abiraterone, and I'm using abiraterone here, because that's where currently we have the evidence across the spectrum.

Chris will discuss AR antagonists. I think we see the same for AR antagonists as well. A real consistent magnitude of effect. If you have patients, what high risk of dying from prostate cancer? The M1 categorization here is as by latitude criteria. I think we'll move on, going forward to using CHAARTED volume, I think that's more helpful. But really the message is consistent that ARPIs are showing a consistent effect. However, I'm not convinced all patients will benefit from additional treatment. So, increasingly we're going to see ADT plus ARPI as the backbone, and then do all patients need an additional treatment? And when we look at 80 follow up data from STAMPEDE, 30 to 40% of patients are in remission. This is going to be a greater proportion in the low risk or low burden patients. And potentially that proportion will increase further with a combination of radiotherapy plus ARSI and we'll see data from piece one in the future. So, I think we need prognostic tests and we need them to identify who has bad disease and needs additional treatment.

There's also small proportion of patients who do not progress on ADT alone, and we all have them in our clinic 10 years on, 12 years on started ADT, presented with metastatic disease. I think it's going to be much harder to implement a biomarker to de-intensify NHT, at start not administer NHT. I think SPOP is a potential interesting gene. This is data from Emanuel and Neeraj, looking at outcome in a retrospective institutional analysis on ADT alone. I think it merits further exploration, but certainly not for implementation to clinical practice. So, working with Nick Noel, many other people in the room, Chris, Felix, we have been working on developing prognostic markers, leveraging the STAMPEDE trial data set we're trying to align so that across many trials we're using similar tests, we've explored Ki67, which is a relatively cheap test, two pounds, clearly highly prognostic.

We're looking at copy number burden. We're looking at whole transcript terms and here is data will present more on, I hope Atesmo to further support use of transcript terms in main treatment ADT and ARSI. Message three, I think in contrast to ARSI for docetaxel, we do not see a consistent effect across disease states. Again, I suspect, or I know Chris will speak about this further for metastatic disease, but it was really emphasized when looking at the high risk non-metastatic data where we see no evidence of an improvement in survival for additional of docetaxel, either in the STAMPEDE dataset, nor in the GETUG-12, which Nick and Karim have presented over the past two to three years. This really contrast with the data we generated for two years of abiraterone. So I think there is a role for identifying patients who derive greatest benefit for docetaxel, not only because of that inconsistent to fade, but also there's a cost with docetaxel is an analysis published last year from 515 patients directly randomized between docetaxel or abiraterone and the STAMPEDE trial.

And I guess we expect that the quality of life would be worse in the first six to nine months during treatment, but it appears that quality of life never improves long term to the level reported by the patients treated with abiraterone. Now this caveat in that statement, but I guess you'll recognize what I'm implying. So my view is that the field needs a predicted biomarker for tailor docetaxel use. We'll discuss triple therapy throughout this panel. I think that should be incorporated with the prognostic biomarker to identify those patients who need it most.

We have, I think, compelling hypothesis generating data from the CHAARTED trial in I guess a small number of patients that is 160 subjected to whole transcriptomic profiling and collaboration veracite. And then looking at PAM50 gene signatures, you'll see from the left hand side that in metastatic patients, we do not see luminal A about 2% of patients have luminal A. So the majority are basal or luminal B, and they split about equally. And there's a suggestion in this analysis that the benefit with docetaxel is restricted to the luminal B patients, as you see in the KM plots here.

So the matter for debate in this context in my mind is what evidence do we need to change practice, to implement molecular tests for better selection of these two treatments?So we can initially aim for data from large independent sets from prospective trials is a number of both academic trials and pharma. We have endeavored to use the same clinical assay, and I think we need to work really hard to do that. And luckily most of the involved people are on the table here and in the room today, we need a pre declared SAP and there'll be much debate over how much can we trust an analysis plan that we've written well after completing accrual and in this case after reporting the primary result, but the best we can achieve at this point is to pre-define and pre-specify and pre-declare that SAP and state the magnitude and direction of docetaxel effect we expect in those molecular subgroups. And then following on from that implementing trials, will we prospectively select patients based on that test.

Of course the timelines for that are going to run into several years. So I look forward to returning here 2023, 2024, and discussing predicted biomarkers for docetaxel. Finally, similarly for ARPI we're aiming to develop prognostic models in our independent sets, train in one test in another, using the same clinical assay. And then we can debate whether we will need a trial to prospectively select patients. So thank you for your attention.