0:00
Today, next, I'm going to introduce Tad George, who is the Senior VP of R&D at RareCyte.
0:07
Tad has had over 15 years of startup experience dedicated to creating scientific markers for novel instrumentation platforms that span basic research, drug discovery and clinical applications.
0:20
Prior to joining RareCyte, Tad has held similar positions at Biodesix Inc and DVS Sciences and was the Director of Biology at Amnis Corporation.
0:33
Tad completed his BA in Biochemistry at the University of Texas at Austin, PhD in immunology at UT Southwestern Medical Centre at Dallas, and post doctoral training at Immunex Corp in Seattle.
0:46
Right, Tad?
0:56
Thank you and thanks for coming.
1:00
Before I get started, as Mark mentioned, I'm always introducing new technology.
1:05
So please interrupt me with questions, anything that you come to mind.
1:11
So from a big picture, RareCyte is kind of a high tech engineering company that is very imaging bias and puts their energies towards all types of biopsies.
1:27
We saw instruments, reagents, consumables and services both for liquid biopsy, cell based liquid biopsy.
1:33
But also what I'm going to talk to you today is our high plex sort of translational tissue spatial tool called Orion.
1:42
And as most of you know, when you're analysing a tissue, it's made-up of lots of different cells.
1:51
They have different sort of functions.
1:54
They're also in different functional States and actually where they're located within the tissue as dramatic impact on patient health status.
2:04
So anyone that's doing sort of a, you know, spatial biology is kind of a buzzword, but it's basically you need to be, you're going to be using scanners or microscopes to analyse that complexity.
2:18
And traditional scanning technology has some fundamental challenges that get in the way of doing translational work at high plex.
2:27
What I mean by translational is usually I want to get through hundreds of samples with an imaging device.
2:34
Typically that means doing things with a single round, so like maybe single channel IHC or maybe a two or three parameter IF, but that's really not enough, you know, information content for the, you know, micro environment you're studying.
2:50
So the Orion was invented really to break that barriers, which allows you to do essentially 20 channel IHC level work in a single round for those translational studies.
3:02
That's what's truly unique about the Orion.
3:06
We call it spatial biology at lightspeed.
3:09
We'll talk about a lot of these features here as we go to the talk.
3:12
But really it's that single round high plex thing in imaging it gives you a 3 fundamental things.
3:20
One, it's super fast, so you can do hundreds of samples.
3:23
The other thing is the run costs are extremely low.
3:26
A lot of that is the reagent development that we did, we'll talk about.
3:31
And of course, because we're everything doing everything in a single round, the data quality is high because the tissue is intact.
3:37
The other thing that's really important, I'll kind of skip over the same section, H&E, whole specimen imaging, we'll talk about that.
3:44
Obviously we go to the talk, but it's really important also to have flexible, you know, panels line because you know you're going to be dressing different tissue types and indications.
3:54
But this really translates into really, you know, faster data generation really.
4:00
And we talked about Kaplan Meyer curves for clinical studies.
4:02
That's really just getting through those large cohort studies.
4:08
It's been applied pretty much to any tissue, any indication 20 channels in a single round.
4:13
We like I say, it can be whole slide tissues, it can be TMAs, it can be core needle biopsies, pretty much anything you can put on a regular microscope slide and label we can image it.
4:26
There's also, I encourage you to look at the Interactive Data sets on our website, because you can click on any of those links and see there's probably 18 or 20 different examples which show you some of the data quality there.
4:42
And I don't know, I don't have a pointer here, but if I could click on that, I could show you that.
4:46
But I, that's probably the most fun part of our website is to look at the different tissue types and stuff that we've done.
4:51
The data is fully quantitative.
4:53
So with immune imaging, of course you have to segment images into objects which allow you to get a data table that you can essentially do classification of different cell types which then allow you to do spatial biomarkers.
5:07
So fully quantitative and I'll give you a couple of sort of case studies of using the system in different ways.
5:14
The first one was actually a genus essentially had found empirically that two of their checkpoint inhibitors, if they were to administer it to colorectal cancer patients, it would dramatically shrink the tumours prior to surgical resection.
5:33
But a lot of time, you know, usually in the case where you're doing dual drug therapy, usually the, you know, pharma companies don't really understand what the mechanism of action is because lies are very exploratory.
5:45
So they turned really to the Orion to look at sort of immune recruitment using essentially a 12 Plex here IO bias Orion panel.
5:57
They enrolled 12 patients, pretreatment biopsy, post treatment surgical resection and if you look at you know one of the surgical resections for the patient these this is a very large colorectal cancer.
6:12
You know resection is probably 5 or 6 square centimetres in this portion of the colon.
6:18
This is normal colonic mucosa pan CK in yellow, CD3 T cells in white and you can see kind of its normal colonic mucosa with a little bit of T cell infiltration.
6:30
But as you move to the right, you see lots of T cells.
6:33
And if you pan further to the right, this is where the cancer is.
6:36
You can see highly proliferative cancer polyps being you know heavily invaded by those T cells.
6:42
In fact, across the top you can see that the immune system is pretty much killed off the tumour.
6:49
So of course you can visually see that dramatically.
6:51
But really we want, you know, spatial biomarkers.
6:54
So really we had a pathologist really draw essentially ROIs around the tumour lesion.
6:59
And we just reported the simplest, you know, spatial biomarker there is, which is density of the different cell types within the lesion.
7:07
And so here you see in blue pretreatment biopsy versus in gold post treatment surgical resection, you can see their drug is really causing recruitment of all the major immune cell types into the lesion, which is responsible for, you know, sort of, you know, reducing the tumour.
7:25
In fact, the pathologist that was leading this, I think the treatment, the between biopsy and surgical resection was about a month.
7:35
He came to his various, you know, he thinks they should do it for two or three months.
7:39
And I and he says, see, I told you, if we just let him go for a couple more months, it might have killed off the tumour without even doing the surgery right.
7:45
So this is just an example of in a prospective trial using the device to look at mechanism of action.
7:53
Of course, the major strength is being get through hundreds of samples.
7:57
So this is actually one of our customers at Harvard in also in the colorectal cancer space published this paper last year in Nature where they had a huge colorectal cancer cohort at the time they had 74 of the patients analysed, but they had an already clinical outcome data for these patients.
8:16
And what they did here is essentially process these resections with the 17 Plex Orion panel.
8:27
And then they actually, you know, did probably several thousand different 4 marker combinations from that 17 Plex, both in different locations and spaces and compared it to the gold standard, you know, it's prognostic test for CRC, which is the immunoscore.
8:48
And they found, I think about two or 300 potentially combinations of markers that were that actually outperformed the immunoscore in terms of prognostics, looking at Kaplan Meyer curve statistics, I don't actually know how large how they've expanded the study.
9:06
They continue to add more and more to that.
9:10
So the important thing about doing translational sort of spatial biology is you really need at least these 4 attributes.
9:20
First of all, you need high resolution, high quality data because that's the foundation for your quantitative data.
9:28
It's also really important to get complete spatial context.
9:32
You can kind of see, you know, down here, this is one of those surgical resections.
9:38
It was priced 6 or 7 square centimetres.
9:40
You can see these holes are like TMA core holes drilled out of the block.
9:45
If we were just analysing those spots, we would not have been able to look at all the different locations and sort of spatial biomarkers.
9:53
So complete spatial context is really critical as well.
9:58
You also need sufficient Plex like it's not sufficient to go through 200 samples and just looking at one biomarker.
10:06
You don't want 100 biomarkers on one sample.
10:08
You kind of need that sort of combination of, you know, 12 to 16 markers across hundreds of samples because that really allows you to, you know, derive, you know, actionable information based on statistics and which is necessary for that throughput, right.
10:25
So those are the constellation of factors really required.
10:28
And actually they were fed into the requirements that we got for in terms of inventing the Orion system, because we basically had, we actually had customers that were using some of our other traditional microscopes and doing cyclic immunofluorescence.
10:43
In fact, they came to us and said, oh, would you, could you please automate the cycling?
10:47
And we just asked him, what do you need?
10:48
And they said we need to get through hundreds of samples.
10:50
And he said, well, we need to do something different, right.
10:52
So that was the foundation for the Orion.
10:57
So in terms of, you know, building different panels, it's actually quite easy.
11:03
The building blocks, you know, based on our Orion reagent portfolio, which includes ArgoFluor conjugated antibodies, they're all IHC validated.
11:14
We have over 100 biomarkers that are available from our catalogue, including off the shelf panel, 16 for the human, 6 for the mouse.
11:23
But also we have conjugation kits and services for custom biomarkers because no matter how big your catalogue grows, there's always a particular biomarker or more that you need to integrate into your own panels.
11:38
It's super easy to label to mean conjugation chemistry, much like it’s done with flow cytometry that's been done for 50 years.
11:46
And then we also have a panel designer to configure custom panels for any application.
11:50
This is an example of our diabetes panel.
11:53
The other thing, they're very flexible.
11:54
All the reagents come as direct conjugated antibodies.
11:58
And even if you order a panel, the antibodies are in separate tubes.
12:02
So a lot of people like will order a panel.
12:04
Let's say this is the diabetes panel, maybe we take they don't care about B cells.
12:09
They say please take CD20 out, put in DC-LAMP.
12:12
It's very easy for us to customise that.
12:15
So in terms of sort of workflow for developing de Novo panels, this is basically what you do use the panel designer to select your biomarkers.
12:29
If there are no custom antibodies required, the panel designer will sign those biomarkers to channels and build an order.
12:36
When you order it, you just verify the performance of that on your tissue type of interest.
12:40
This is important because in production, we essentially validate all of our reagents by IHC with a controlled tissue in the production lab, which may be different from the tissue that you're using for your experimental samples.
12:56
So titration might be a little bit different.
12:58
So you basically take 2 serial sections.
13:01
One of them you stain everything at 200x, which is the recommended concentration, another one at 800x and you can look at those two samples and make individual titration decisions for each of those antibodies.
13:11
Usually the 200x works, but sometimes it's too bright.
13:15
So you might have one at 500x, one at 800x and then you just verify that on the third section usually takes a week and you can begin your study.
13:23
But if you have custom antibodies, then really the job here is to 1st validate those reagents in single plex on the Orion versus IHC.
13:33
So you first purchase the clone for the vendor and a controlled tissue.
13:38
Make sure that you can validate that clone and take controlled tissue by IHC.
13:44
Once you do that, you simply do the labelling and then validate that the pattern in IHC and IF look similar and then you insert that into your panel.
13:55
In our production lab it usually takes us about two weeks to do this.
14:00
We do about four custom conjugates at a time.
14:03
So it's actually pretty quick and easy to validate reagents in terms of once you have your panel, the actual sample testing workflow is quite nice.
14:14
We often will stain up to 24 samples a day.
14:17
The staining is done independently of the scanner.
14:19
So it's done and you know, you can image in parallel to scanning.
14:25
You can put, as I mentioned, super large or multiple specimens per slide.
14:31
It's a whole slide technology.
14:34
You also, what's actually quite convenient is after staining, you can either image it right away or you can bank them for up to six months without any signal loss.
14:44
So it's really good for like if you have multi site trials, we have different labs that are staining stuff.
14:50
You can actually send them to a centralised place and they can, you know, schedule the scanning so you don't have to scan at the same time you're staining or you know that way.
14:59
And then as I mentioned that this imaging is super fast.
15:02
So the 20 channels, it's the whole slide one scan, it takes a little over an hour per square centimetre.
15:07
So it's lightning which allows you to get through your large studies and then the output data is fully quantitative.
15:16
OK.
15:19
So in terms of the technology and validation, the question is how do we actually do what we do?
15:25
I think when we looked at trying to do high plex in a single round, there are a couple things that were requirements.
15:33
The first was that we couldn't use amplification schemes because a lot of those involve primary, secondary approaches and you'll have cross reactivity between your secondaries and your primaries, right.
15:46
That would limit the number of markers you could address simultaneously.
15:50
So we essentially said we need to go with antibodies directly conjugated to Fluors and there really wasn't, you know, a scope available that was sensitive enough to look at low abundance biomarkers with the right conjugates.
16:02
So we essentially and from a big picture built like a flow cytometer for tissue.
16:07
So that required high-powered, very specific laser excitation.
16:12
And because we're trying to image 20 channels in a single round, we wanted to also have narrow emission bands and we also needed super bright dyes.
16:21
So we basically screened about 300 small molecule dyes for brightness, photo stability and spectral spacing.
16:30
And we use that information to shop around that to a laser manufacturers who custom built a laser module a 9 laser module system for the Orion system.
16:41
The other thing that was really important about the labelling is we wanted to make sure that the labels would be very light touch.
16:50
Because if you're going to, you know, modify an antibody with a label, there's a risk that it will actually perturb its function.
17:00
And we really wanted to use something which was, you know, known to not do that.
17:05
So we basically use the original small molecule organic dyes that is traditional for flow cytometry.
17:12
It's very easy to do.
17:14
All you need to know how to use is a pipette and a spin column.
17:17
The degree of labelling is typically around 3 Fluors per antibody random labelling and we haven't yet had a case where we couldn't reproduce the IHC with the labelled antibodies.
17:29
The other thing that was really important because I've worked at other companies that have also built out reagents where stability is an issue where you know, for we actually screen various storage buffers and found one that allow the antibodies to be stable for at least five years.
17:49
So what's important about that for these for multi year studies that we have with our pharma partners will often have a single lot for the entire studies.
17:58
You don't have to worry about lot bridges.
18:00
The other thing it does from the manufacturing perspective, I mentioned the low run cost.
18:05
This was a big part of that because a lot of times if in manufacturing you have a large catalogue and many people like certain things aren't ordered, you have to discard things as their shelf life goes around and that goes enough driving the cost up for all the different antibodies that you buy.
18:21
For us, it's not really an issue because even if we have an esoteric antibody that's really important for one of our customers, we know as long as they order within five years, we're going to put it in the catalogue and we're not going to be discarding stuff a lot.
18:33
So that's part of the reason why the run costs are really low.
18:39
This is an example of validation like I mentioned, it's always versus IHC.
18:46
This is an example the SP267 SOX10 clone, again, we first thing is always established that you have a control tissue in the pattern that you expect.
18:56
Some of our customers skip this step.
18:58
They go straight to labelling and throw it in.
19:00
They don't get your results.
19:01
And I said, well, how did the IHC look and said we didn't check.
19:04
I said just do that first.
19:05
It's really easy to do.
19:06
And then it's just a matter of labelling it with the Fluor and then making the serial section showing that the pattern looks the same, right?
19:12
It's very straightforward.
19:14
Also of course with IF you can actually look, do a little bit of specificity.
19:19
This was one of our customers that, you know, downstream of a NanoString transcriptomics screen, they found several macrophage markers that they validated in protein by IHC.
19:31
One of them was IL4-I1, which they knew was in macrophages by IHC.
19:36
You can see it's very punctate.
19:37
But yeah, they know it's supposed to be in macrophages.
19:41
But of course with single channel IHC, you can't really tell if those are macrophages or not.
19:46
So you can do a specificity panel, in this case IL-4I1 with CD68, CD163.
19:51
And you can see that IL-4I1 is indeed, you know specifically staining, you know, within the cytoplasm of macrophages.
20:00
In terms of accuracy and precision for accuracy, you can see a couple of cases here where biomarker detection typically matches or outperforms IHC even for challenging biomarkers.
20:14
So if you look at say PD1 expression in the crypts of a colon, you can see these very faint sort of brown stains, very dim staining by IHC.
20:24
But we see it just fine with the ArgoFluor label conjugate.
20:29
In fact the IHC was actually more, we found more PD 1 positive cells, you know by the Orion technique.
20:36
And just to make sure they weren't false positive, we co staying with CD3 and found that they were on T cells.
20:42
So this is an example also for CPS for PDL1, but also one of our CROs actually validated a 16 Plex Orion panel head to but again, that's why we engineered the system to be highly sensitive and it's highly precise.
21:18
It's very repeatable.
21:19
Of course, if you have one instrument running the same thing over and over again, the CVs are less than 5%.
21:25
But even running a different instruments for different operators on different days, it's quite a reproducible system.
21:35
So I know we had like 25 minutes so, but you know, that's pretty much the Orion system.
21:44
So please ask any questions that you have based on what you've seen.
21:48
So thank you.