Anwar Padhani: I'd like to start by thanking Silke and Aurelius for putting on this great meeting as usual. These are my disclosures. Now we've all heard this afternoon that mCRPC is a highly heterogeneous disease and there are multiple prognostic determinants. And of course, one of those is imaging. And within imaging, the distribution of disease is clearly important. It's volume, for example, and multiple other imaging factors are also prognostic. So for example, this is quite old data showing that if you have lymph nodal disease, you're going to survive for longer than if you have liver disease. And all of you are aware of this, but counting the number of deposits, and we shouldn't be counting, we should be voluming. But if you just count the number of deposits again, that is prognostic.

So, two deposits or less, you're going to live longer, 10 deposits or more, you're going to live less long. And of course, if you've got this from metastasis, then it's bad for you. And of course, that can be taken onto volume. And you can measure volume, as you've already heard by Michael Hoffman, by using PSMA PET, you can use whole-body MRI. And here is some data on whole-body MRI and how the volume of disease, so a step beyond just counting, is prognostic. And I thought I'd show you some PET PSMA data, because that it actually shows you the same thing but interestingly, now you see this gradation of how the volume is not binary, but in fact grades with the volume. So what this is telling us is that you need to switch early. Well, you need to switch early because of clonal resistance and therapy and resistance mechanisms like kicking in, because the earlier you treat the better is the likelihood of treatment response.

How good are you at predicting whether patients are progressing in a timely fashion using triple assessments? It turns out not very good. So it turns out that the most of you are using bone scans and CT scans. And I just wanted to say a few words about radiographic progression. Now all of you have seen today, beautiful slides where you see rPFS and OS are correlated. Whilst, I thought I'd show you a couple of slides where they're not correlated. And in fact, rPFS is not an approved endpoint. It's often a co-primary endpoint. It can be a secondary endpoint, but it's not a primary endpoint. And why is that? Because it isn't very good.

Why is that? Well, it turns out that bone scans, which I call bullshit scans, assess the osteoblastic activity to the presence of tumor cells. So you're not looking at the tumor cells, and ditto for CT scans, which I call shitty scans, right. Because again, you are not looking at the tumor. You will be looking at the tumor in the lymph node in a liver, but not in bone. Not in bone. Because in bone, what you're looking at is the osteoblastic response, again, to the presence of cancer. And it is not at all unusual to see patients progressing without seeing anything change on a CT scan or a bone scan.

And here's an example. So this man was on a clinical trial, pembrolizumab plus olaparib. And we've just heard that this trial failed, but you can see how on three separate occasions, we call this prospectively stable. The patient had unequivocal clinical progression. We were blinded to the PSA and you can see after unblinding, we did the whole body MRI and the patient's volume has clearly tripled. And of course, the PSA did report it, but of course, we were blinded to the PSA.

And in fact, this is not at all unusual. And your literature shows that. So unequivocal clinical progression, that is worsening the clinical state, not seen by a bone scan or a CT scan, with abiraterone, occurred in 40% in this particular trial. And the flip side also happens to be true, unequivocal imaging progression, with no clinical progression occurred in about 30%. So what are you guys doing sailing with a triple assessment when you're actually sailing blind.

Now you need better technology. And we've heard about the ability of Next Generation Imaging to predict the presence of microscopic disease. But in fact, you can predict micro-progressive disease also. And remember that imaging gives direct assessment of the tumor, where it is, and how it's responding to treatment.

So, Michael showed some nice spinning images with a PSMA PET. Well, you can do this with an MRI. And here is lymph nodal disease segmented out over several occasions and what you notice is in the castration resistance state, that you get new bone disease, so that arrow on the bottom right. And in fact, if you look at the lymph nodes, you can see discordant responses in the lymph nodes. So in the retroperitoneum, beautiful responses to Abiraterone, in the groin lymph nodes, there's progression. So this very precise delineation of what the tumor is doing and where it is doing is the advantage of something like a whole-body MRI.

Not only that, but you can also start to make comments on cellular habitats or viability, and you can color code this and you can spin it around if you want to. So in this example, what you notice is that after four cycles of docetaxel, there is therapy resistance in the right femur for example, and in the retroperitoneum. So cellular viability can also be mapped. So very precise delineation of what is happening where. Now in whole-body MRI, there is a system that was published in 2017 called the MET-RADS system, which defines what imaging criteria you should use for something called RAC five progression, highly likely please change the treatment. So we have defined this and we now know how reproducible it is between different observers. And you can see that in fact, it is highly reproducible between experts and non-experts.

This is now being tested currently in a clinical trial, which is in Milan. And look over here. So in Milan, they've been doing MRIs, whole body MRIs, CT scans of bone scans every three months. And you can see that we've now got 25 patients who have progressed. And when we know that the patient is progressing, is progressing either bone scans or CT scans or somewhere, in 80% of these patients you do not see anything on the bone scan, which is why I call it a bullshit scan, right? And in fact, this precisely mimics what we saw in the breast. In the breast, we saw the same thing. When we knew that the patient was progressing, progressing to an extent where you had to change the treatment, the bone scan did not report it reliably in 50% of patients.

So, the current ways of treatment switching isn't great. And here's an example of how the bone scan doesn't change, the CT scan doesn't change, but the whole body MRI is showing you progression highly likely, please change the treatment. But of course, we need to be very cognizant of the fact that just because you change a treatment does not mean that the patient will benefit. And this is the point that Maha Hussain referred to earlier yesterday. Just because you change your treatment does not mean that patients will benefit. And this was shown actually very nicely in this Lancet paper, where there was progression shown on a MRI scan for spinal cord compression. But ultimately when you follow these people over time, you did not delay clinical spinal cord compression by using a next generation technology early. An important lesson, I think. But of course, the interesting thing is in PSMA PET, it does appear that they may actually be a valid signature.

And this is from Sydney, Australia, and you can see how the presence of progression in the response assessment setting, is in fact predictive of overall survival. And this was confirmed in a paper last week where response assessment criteria were also assessed for PSMA-PET. And again, you can see how it's predicting for overall survival to lutetium PSMA. So it looks like these technologies might work, but we don't know for sure. And that speaks to where we are with the guidelines.

Now in 2017, a very important statement was made. And that said that the imaging schedule for metastatic castration-resistant prostate cancer should still use the triple assessment, but it had to be done regularly because of these discordant responses that I talked to you about. Now, this was extended a little bit by the ASCO guidelines, where it talked about next-generation imaging and a paucity of data. And that is true at the moment.

Now, when there's a high clinical suspicion that the patient is progressing and you don't see anything on a bone scan or CT scan, clearly you need to clarify the information and next-generation imaging is indicated. But in fact, the ASCO guidelines left open the door to say, if you started with a next-generation imaging, then you could follow with that. Now that's the point of debate.

So, what I'd like to do is leave you with two key conclusions. Number one, discordance using triple assessments is incredibly common, between 30 and 40%. So you need to be very vigilant about progression occurring and what you need is protocol-based triple assessments of disease status as a minimum standard. So you need to do it regularly. Otherwise, you're going to miss discordant responses. Number two, the management or survival advantages of using next-generation imaging in the mCRPC state have not yet been shown. There's some interesting data that's emerging but has not been shown. Just because a biomarker is good at prognosis does not mean that it will be good as a therapy assessment, a parameter, that needs to be tested. And therefore prospective clinical trials, with clinically meaningful endpoints still need to be done.

Thank you very much.