Mark Rubin: I'd like to thank Aurelius and Silke for inviting me to this meeting. And so, today I'm going to very briefly talk to you about the future of biomarkers in the context of pathologists who's joining this meeting. And I just want to say as a comment that it's really, I think, humbling to hear the hard work that's going on in order to take care of patients with advanced cancer or prostate cancer, and the challenges you face. So, hopefully, just a few disclosures here, hopefully by being really disciplined, and I'm not going to talk about research as much, I'm going to really focus on how we can have a road towards developing biomarkers. And maybe the next time we meet we'll have even a better plan to create more robust biomarkers.

But I think one of the starting points should be, at least from the context of pathology, is that we think very much about the rigorous development and the accuracy and predictive values of these biomarkers. And I just want to point out that I think there's been really nice discussions about biomarkers and some of the concerns we have. I think the predictive biomarker is the real challenges we've heard. Can we predict response to therapy? When would we actually use that test not to treat somebody? So, that's obviously quite important.

The areas that we're going to use to gather information have been talked about at this meeting. So, Germline Tumor DNA and RNA protein, as well as, as we just heard, very nice examples of how to collect a liquid biopsy. So, these are all going to be important areas. And imaging has been discussed, so I'm not going to discuss imaging at all. Obviously imaging is going to be a very important biomarker.

And Colin Pritchard I think made a very, very nice presentation yesterday and others have talked about the types and similar to Colin, just using very broad numbers, nothing that I would bank on, but just approximations of how common alterations are for example in DNA repair, microsatellite instability, and the fact that there are associated treatments.

But as we also have to remember there are a lot of missing pieces or a lot of gaps. So, how do we understand why patients are responding or not? And we will have new mechanisms to think about as we try to understand why patients are not responding. And just to give you an example of nucleotide excision repair, which is not really carefully understood in this paradigm we've been talking about with PARP inhibition for example.

So, what is needed next? And again, this is a pragmatic approach. So, as a pathologist, I think we're very aware of heterogeneity and I hope I give you a little feel for what I mean by heterogeneity. So, if we look at primary prostate cancer, and this is just an example of a primary prostate cancer where we have a cross-section, and I don't have a pointer, but you can see there's a T1 lesion and the different names of the lesions are divided. This a pathologist evaluating them. I think what's important is that there's a truncal mutation or alteration in the tempers 2 ERG gene seen in the lesion three, but in lesion one we don't see that. So, two lesions right next to each other have very different truncal alterations, and the one that doesn't have the tempers 2 alteration has a p53 alteration. So, this was already telling us about problems with biopsying samples and how the heterogeneity just inherent and prostate cancer makes it difficult to make an assessment based on where you have a biopsy to take. This case was part of a Stand Up 2 Cancer study and this case was contributed by University of Washington, Pete Nelson, and colleagues. And luckily we could start looking at the patient's metastatic sample and try to predict which is the lesion that would lead to metastatic progression.

We heard earlier today about different types of alterations that can occur in the context of treatment, and one of them is neuroendocrine or altered or aggressive state of prostate cancer. And this is relatively an indifference to AR therapy. And this is just an example where we can imagine a prostate cancer like this. So, we have adenocarcinoma, and transitioning to a small cell cancer. And at least in our mind's eye, this is what I think is very simple paradigm and this is where we started. But in reality, if we take a look at this slide that was prepared by one of my pathology research fellows, going from the top to bottom on the left, you can see that you go from an adenocarcinoma to something that's a small cell carcinoma.

And if I told you this was from four different patients, I think that would be something you could deal with. But the fact is this from the same patient. So, in the same individual is treated, we see a progression of a disease depending on where you sample. So, I think it's very important to recognize that we have a spectrum of disease that we're treating. It's not going to be black and white.

So, in the future, we're going to have various ways of looking at this. Our colleagues in other areas like in leukemia and other tumor types where it's a little bit easier, melanoma, to look at single-cell sequencing are using that. Here's just an example of an insight to approach where we can look at different regions of the same tumor and ask not only one or two biomarkers, but hundreds of biomarkers. And this is just an example of geospatial immune landscaping.

Clearly liquid biopsies are one way that we can get around the heterogeneity. Can we capture all the information as we just heard and put it together and figure out what's going on in the tumor? And this is a slide I showed from the last meeting, hasn't changed, and we wait patiently for the tests and I'm very optimistic that we're moving in the right direction. It's challenging. And as we saw, we really have to de-convolute the signal from a lot of information.

So, robust assays for prediction. So, we heard a lot about various assays and there's a little controversy and there has been controversy at which one works best, which is most predictive, which is prognostic. And I think what's important to I think remember is that we need to think about ways of designing studies to compare these biomarkers together and really be rigorous. And this is something that has been a tremendous challenge for the biomarker community.

We heard that it's not only a question of a good test, but it's how you understand the biology. So here, this is an example of a paper that was presented earlier in the meeting where we have a microsatellite instability test using impact from Memorial Sloan Kettering. So, it's an excellent test. It's a New York state approved test, very rigorous test, but nonetheless it only predicts 50% of the patients who may respond. And this has to do with biology and our lack of knowledge. So, we also have to recognize that the biomarkers that we're using can only do as well as as they can perform as we understand the biology.

So, being from Switzerland, I love the sound of cowbells. So I know I only have one minute, but I like that sound. So, here's the list I made with Pete Nelson of some promising tests and I'll just let you look at that for a moment. Obviously, on the top are MSI and DNA repair status, but there are other tests that we hope that will become real robust biomarkers.

So, finally. This is an image of Arthur Purdy Stout. So, Chuck, this is from your institution from Columbia Presbyterian. He started out as a surgeon and then the surgeon said, "we need some help to understand what we're actually taking out and what it looks like". And so, he became a pathologist. And so, I don't think it's so outlandish to think about today, how in precision oncology we can think about bringing pathologists into this type of meeting and actually creating an opportunity for us to understand the problems. And I mentioned that yesterday in my question and a little discussion with Johan, which is that we really need to find ways to bring pathology into this biomarker development. It's really important. I think that there's a lack of communication and that would be my main message is that I think we have a time that should be really exciting for young pathologists, molecular pathologists to get involved in this.

So, thank you very much.