Executive Interview with Dr Evan Keller

0:02
Good afternoon everyone and welcome back to another Oxford Global interview.

0:07
Today I will be speaking with Professor Evan Keller, a professor of Urology and Pathology at the University of Michigan, where he leads research into metastasis, the much human microenvironment, and explores the application of spatial and single cell analytics in urologic cancers.

0:25
So thank you, Evan, for joining us today.

0:29
And I guess to start off with, so I'm aware that you lead a single cell spatial analysis programme at the University of Michigan.

0:38
So could you maybe tell the audience a little bit about how your interest in spatial analytics emerged in the 1st place and the key gaps in urological Cancer Research that motivated you to pursue this field?

0:52
Yes.

0:52
Thanks for the opportunity to share some of our information.

0:56
So started off my interest, started off many years ago because of my interest in tumour microenvironment, even before it was called tumour microenvironment.

1:06
And I'm trying to understand how not only the cancer cells, but the cells surrounding those cancer cells interacted and worked and somehow influence the development and progression of cancer.

1:18
And so the ability maybe about 15 years ago or so to start doing single cell analysis came around in the commercial level.

1:27
And that was an exciting start to try try to really understand the mixture of cells within the tumour microenvironment and understand the gene expression within those individual cells.

1:38
From there, when spatial analysis came along, then we're able to answer more additional questions, not only what is the gene expression within the individual cells, but in the context of what their spatial relationships were.

1:52
This is now allowing us to look at the tumour microenvironment in unprecedented depth such that we can really explore cell to cell interactions, look at tumour niches and get a better understanding of how they may interact within the tumour cell itself, with the tumour cells and within the cells of the tumour microenvironment themselves.

2:13
In terms of gaps, we're really have a lot of gaps still in urology research.

2:19
My focus on prostate and renal cancer are my primary areas of research and really trying to understand progression, how they will the tumours will continue to grow and their response to therapies and development of resistance are all very important areas that we feel that spatial analysis can give us some insights into.

2:41
Again, through looking at cell cell interactions, cell expression and gene expression within those cells to be able to really get a clear understanding of how those genes may impact the overall progression and resistance of these cancers.

4:08
So in terms of gaps in urologic cancers, there are a lot of areas that we're trying to understand and that these include the progression of the cancer and biomarkers that can indicate the progression and biomarkers of resistance and experience to understand whether they will respond to certain therapies.

4:29
And so I think the spatial analysis can afford us a new way to look at these outcomes, including the cell to cell interactions and gene expression that may occur in different cell types that really will inform us on the ability to understand that resistance development or how the tumour itself is going to progress.

4:52
Perfect, thank you very much Evan.

4:54
And so one of your recent publications explores the role of androgen deprivation in a patient specific prostate cancer model.

5:04
Could you also tell the the audience a bit about your main insights and the results from this paper?

5:13
So we were trying to explore how androgen deprivation may affect the actual prostate in the tumour microenvironment and the gene expression within the prostate itself.

5:25
And this is really important as most men are treated with androgen deprivation therapy with for prostate cancer.

5:32
Eventually these men will go on to develop and what we call androgen independence, where the prostate cancer will start to regrow even in the absence of androgen.

5:42
And so we were trying to look in models to understand what this androgen deprivation may do in terms of the gene expression and the spatial relationships within the prostate itself.

5:53
And we found many areas and we found that the different areas of the prostate had different types of gene expression that were altered, some that were androgen dependent, but also some that were not dependent on androgen.

6:05
What this allowed us to really understand is that there may be different areas to target for androgen deprivation when we're trying to explore this as a therapy for prostate cancer.

6:16
And specific ones may be that we saw different types of cell types being developed, particularly prostate epithelial cells.

6:24
Those are the cells that actually make up the cancer itself, and also fibroblasts, which are a cell within the microenvironment that had developed different types of gene expression that we hadn't quite seen before.

6:35
And so now that we have this information, we'll be able to start looking at those genes to see if they could actually be targeted or somehow influenced so that they may be able to slow down the development of androgen independence and continued prostate cancer growth.

6:51
Great, thank you very much.

6:53
And I think if I remember correctly from our previous conversation, you mentioned that you were working mainly with in a 2D spatial setting, but maybe now beginning to look into like early 3D studies.

7:07
And I guess from there, what new biological questions do you hope that this 3D spatial mapping will help you answer?

7:16
Yeah.

7:16
So currently most of our studies have been at the 2D level.

7:20
And when we do that, that's a really thin section of the tumour itself.

7:24
It's not even a full cell thickness.

7:26
So what happens when we're looking at 2D is we're actually cutting through cells.

7:30
So we don't know the true cell to cell relationships above and below where we're actually evaluating.

7:36
So now we're starting to look at technologies that may allow us to look at 3D levels up to maybe 4 to 5-6 cells thick.

7:44
So there we'll be able to start appreciating more on what relationships there are between cells.

7:50
We can actually not look at cells from just side to side, but cells above and below any cells of interest.

7:56
This will give us a more complete and holistic view of this cell to cell relationships and interactions and the spatial relationships that are occurring within the tumours.

8:06
And this will also help us build networks of both gene expression and cell interactions that are occurring in the tissues.

8:13
As we get this more information and we relate this to outcomes within the cancers, we'll be able to start making better predictions of how the cells and the cancers are going to progress and respond to therapies.

8:27
Thank you.

8:28
And when you are analysing this complicated spatial data, in your opinion, what kind of features or signatures are the most informative biomarkers?

8:45
So it's interesting to look at what the biomarker outcomes may be.

8:48
One of the areas that we're so used to is just gene expression, typically RNA levels of genes or even protein levels.

8:56
We can actually look at proteins on these two and we can get a sense of their expression and whether they're up regulated or down regulated, and that gives us a signature.

9:07
We can combine different genes that are up and down regulated and proteins up and down regulated, and when we put those together, we get a signature for the outcome.

9:15
Now one of the areas though is when we typically do that, that's all mixed together and in a mishmash of cells.

9:22
Now we can look to see if there are spatial relationships that are informative.

9:26
And importantly, we also can not only just look at these expression of biomolecules, we can now look at cell to cell interactions.

9:34
In other words, we may be able to look at a tumour cell and look at some sort of immune cell and measure the distance from there.

9:41
And those distances have shown to be prognostic in different types of cancers.

9:45
So by doing this in a 2D and now a 3D approach will have a more clear representation of the distance, for example, again, between a tumour cell and an immune cell and be able to see if that actually can predict certain outcomes and responses to therapy.

10:02
Thank you.

10:03
And I guess it's also important to talk about the reproducibility across different assays.

10:13
And in your opinion, what do you think it will take to move spatial analytics from the research lab into like a reliable clinical tool?

10:26
So moving the spatial analysis into a clinical reliable assay is going to be very challenging.

10:34
We have to think of from different perspectives.

10:36
1 is that there is reproducible results not only within a lab, but between labs nationally and internationally.

10:43
Can we get the same types of results and outcomes?

10:46
That's going to be challenging.

10:48
And one of the areas that we need to kind of consider is what level of assays do we want to do?

10:54
In other words, right now for many spatial analysis methodologies, we're looking at 5010 thousand genes.

11:01
Clinically, that's probably not going to be very tractable.

11:04
The turn around time would be too long.

11:06
The inconsistency from assay to assay would probably be very challenging.

11:11
So we've probably considered these as discovery to help identify those genes and proteins that would be very useful.

11:18
And then we can narrow down to maybe 10 proteins and 10 genes that would be able to do on an assay.

11:24
Once we get down to that level, then we can start to maybe tackle the reproducibility by making sure we have very good reagents to be able to measure those particular genes and that they're consistent over and over again as we measure these types of assays over time within our own labs and among other labs nationally too, so that they really show that they're consistent and robust measurements.

11:51
Thank you.

11:52
And of some of these challenges that you mentioned, which one would you say is currently the biggest bottleneck if you have to pin it down to to once that's a challenge, they all are.

12:07
I can't say that there's one.

12:10
So I think that in terms of what do we really need to do again, is that just overall consistency?

12:16
So first identifying the right biomarkers and proteins and an RNAs that you want to be able to target, but then really having real consistency with one type of of assay.

12:28
Currently, there's multiple systems out there.

12:30
And so it'll be very challenging for people to all adopt just one particular system.

12:35
But if one assay and system gets proven and that would be one that would potentially be used as an approved assay over time.

12:44
And so really narrowing down that, that whenever you measure that one gene in any particular tissue that you're always going to get those same results spatially within those different tissues, that's going to be a very big challenge.

12:57
Thank you.

12:58
And are there any particular urologic cancer subtypes or clinical scenarios where you think spatial approaches will have a particularly high impact?

13:11
So in terms of areas that may have really high impact in neurologic cancers, one that I think can be very promising is the diagnosis of the grade of prostate cancer, for example.

13:22
So when we're talking about grade, we're talking about how severe that prostate cancer is.

13:27
Is it kind of mild one that doesn't look like it's going to spread or grow very fast versus a high grade one that looks very aggressive and may spread very rapidly.

13:36
And when we look at prostate cancers, we define what we call grade groups.

13:40
And grade Group 1 is one that doesn't look to be too aggressive.

13:44
And we typically may recommend to a patient well over time that this may not really progress very far.

13:50
And we might recommend actually just monitoring this every six months or so and not actually doing any curative surgery.

13:58
However, there's a likelihood of maybe 10 to 15%.

14:02
Even though it looked like grade Group One under the microscope, it may have actually, we may have missed an area in the prostate where there was a more advanced cancer.

14:10
And so then we might have missed an opportunity for cure because we thought it was just one area that we can monitor.

14:17
So by using spatial analysis, we may be able to get additional clues on these biopsies that suggest, oh, this is something that's above and beyond grade Group 1, even though we couldn't quite tell it.

14:29
But now we have this additional holistic information from the spatial analysis that's showing additional spatial relationships including gene expression and cell to cell relationships that suggest this actually was something more aggressive than I originally think that it was.

14:46
Thank you.

14:47
And I guess is the University of Michigan, are you guys partnering with any other academic groups at the moment?

15:00
Yes, actually.

15:01
So in terms of partnerships and spatial analysis on helping lead a programme where we're partnering with the Cancer Science Institute at the Singapore National University.

15:12
And so yeah, they are, we're looking at spatial analysis through for a variety of diseases, primarily lymphomas actually we're helping with because that seems to be a big challenging area that they're having some expertise with.

15:26
And so we're working on lymphomas where they are identifying tissues, they've send us tissues and then we do the spatial analysis here and then we have meetings to discuss the results.

15:35
So we are really having an international collaboration that our Rogal Cancer Centre is really trying to develop in terms of really getting a nice interaction with that particular Singapore Cancer Science Institute.

15:50
Nice, thank you.

15:51
And final question from me, Evan is looking to the future, how do you envision combining spatial analytics with other technologies to, you know, to generate a more integrated view of the tumour microenvironment?

16:09
So one of the challenges with spatial analysis is, particularly in cancer, is what we call tumour heterogeneity.

16:16
In other words, within a tumour, different parts of the tumour are a bit different from each other.

16:20
And so one of the big challenges is when we do spatial analysis, we're doing only a small fraction of the tumour.

16:26
So what we're looking at may not actually represent the whole tumour.

16:30
And so that is a concern.

16:33
So pairing this information that we can get with perhaps more holistic areas such as circulating free DNA where we're getting serum and then looking at the DNA in the blood may give us some other clues because that would DNA would be a summation of all the tissue, all of the cancer that maybe have broken down and released DNA.

16:56
So putting that together may give us a very nice integrative approach to give us a really complete picture of what's going on with those cancers.

17:07
Thank you, Professor Keller, and thank you for sharing your insights with us today and your important work.

17:15
And we hope to see you again, hopefully at another Oxford Global event in the future.

17:22
But yeah, otherwise, very much looking forward to what you to see what you do next in this space.

17:27
Thank you.

17:28
Thank you very much for the opportunity.

17:30
It's my pleasure.

17:31
Thank you.