Aleksandra: [00:00:00] When doing digital pathology, shall you do it in incremental steps and check if it’s gonna work for you, or shall go all in and reap the benefits? All the aspects of digital pathology can give you. In ideal world, if it was not associated with a big cost you would go all in and reap the benefits as fast as you can.
But in the real world, you probably have to start small. There is a solution to that. There are companies you can partner with so that you can start small but take advantage of the full infrastructure. And in this episode, I’m gonna be showcasing a company like that, TriMetis Life Sciences.
Usually when I talk to digital pathology companies, they focus on one digital pathology component, maybe scanning, maybe image analysis, maybe image management system, maybe lab automation, maybe partnering with labs. This company is doing it all and is being an [00:01:00] external hub for every. Who wants to start doing digital pathology but does not have all the components.
It’s a really unique value proposition, and my guest, Leif Honda, is gonna talk about it in this episode.
Welcome to the podcast, my digital pathology trailblazers. Today my guest is Leif Honda. He’s the Chief Innovation Officer at TriMetis Life
Sciences. And TriMetis Life Sciences is an interesting company that I wanted to present on the podcast.
I let Leif introduce himself and the company. But just to give you an initial perspective, TriMetis used to be a biobank, and now they’re serving other biobanks with digital pathology and AI.
Leif, welcome to the podcast. How are you today?
Leif: Thank you. Thank you Aleks. well it’s nice to see you again.
Aleksandra: Great to see you as well. We start with, you tell our trailblazers, I call my people digital pathology trailblazers, and you belong to this club as well by being on this podcast and driving this [00:02:00] innovation. And you, especially because your chief innovation officer, not every company has that position. So go ahead and tell us about yourself.
Leif: Yeah, that’s correct. I’m the Chief Innovation Officer at Trimetis and yeah, that role is very interesting one, it’s one that came about over a period of years where I’ve been tasked with finding new innovations that’ll help our biobanking and future processes with supporting pharma and diagnostic companies.
Aleksandra: Who are you? What’s your background? How did you end up in Trimetis?
Leif: I have a molecular biology and economics degrees, and for years.
Aleksandra: That’s a rare combination Leif.
Leif: Yeah, it was actually difficult to get my school really didn’t support that, but I did get finally get to both degrees at the same time.
And that combination really comes from a place where I had been looking at, into medicine and I was going to go to medical school. And I found a very interesting crossover when I took an economics course and realized that these two are kind of ivory towers. They separate very early on and say, you study very hard in chemistry and biology the sciences.
And there you do [00:03:00] think about what costs and how to do things? And so, I was really intrigued by that and underpinned my future by trying to get these two degrees simultaneously and combine them. And then as molecular biology emerged as something more important, In sciences, it really showed where pretty inefficient in the medical system as well as in the research area.
And over a period of years, I’ve been working in life sciences and finance essentially and funding these companies. And then I was picked up by a gentleman who was running Olympus America and had retired for one day and decided that he wanted to use his money that accumulated into investing in companies.
Aleksandra: After one day of retirement.
Leif: Yeah, he’s that kinda guy he’s a great person, Sidney Brisky and he really taught me a lot about the business of life sciences and he, the first investment was in F T I R for your transfer infrared technology, which is used in teaching missiles and food stuffs and growing computer boards and things like that, and so we were very successful.
Won that business and eventually flipped that business, from there, [00:04:00] he reinvested in a company that was really in a subset of proteomics, which is called Peptidomics, and that’s where the genesis of all this kind of innovation came from is how do you use visible light and different wavelengths of light to identify new analytes, new targets, a new biomarker in disease.
We had invested in a very exceptional company outta Germany that had been around for 20 years. Really cutting edge. And when I got in there, I realized like we have a lot of scientists and we have no business whatsoever, so we had a problem to deal with. And so, he sent me in there to change that.
And it wasn’t about right sizing it or, downsizing or anything like that. It was about how do you get all this great research into the world, how do you put the marketing material around it, but also how do you get it to validate? And that’s where the impetus for me came from is I started saying these programs are exceptional.
And some of these programs are just emerging today, to be honest with you. Some of the Alzheimer’s programs, we were doing science was already done 20 years ago. Colorectal cancer, breast cancer certain markers in breast cancer, and even things that are emerging today, which are in diabetes drugs, [00:05:00] type two diabetes drugs.
Now our fashionable use for weight loss, we had discovered all that already, but we couldn’t prove it because we didn’t have the samples and the tissue to get it. And we were on a hospital campus in Germany, but we still couldn’t get those tissues. And so that’s where this journey began and I had I met.
Aleksandra: Getting the tissues, right?
Leif: Getting the tissues. Yep. So, working with the doctors, the phlebotomists, everyone involved to get tissues and images. Even when you’re doing diabetes, you need to have images. You’re looking at the kidneys. When you’re looking at cancer, you need to have images. And that just wasn’t mature yet. That was very difficult to get.
And that’s where, where the journey started as, how do we make this better? Cuz this can’t be just our problem, and it turns out it’s everybody’s problem that a lot of studies are delayed or waylaid because they don’t have samples to do discovery on or validation on or verification or what have you along that process.
So you think about why it takes 12 years to get a new diagnostic out there. Why does it take more than that to get a new therapy out there? And although we’re doing better with AI and different technologies now, we’re [00:06:00] still taking a long time to get new therapeutics. And the disease book is big, as as pathologists, it’s massive.
So that that was where we started and I, Said, I think we can solve for this. I think we need fix this and get forward in our own programs, accelerate this process.
Aleksandra: So I’m gonna ask you in a second about the biobanking workflow because this is not a standard digital pathology application or startup, but I wanna emphasize what you just said.
And I see a trend in recently in my podcast guest that they are a personification of some kind of bridge, bridging two disciplines. And you said that you had this company that had so much great research, but there was no business. And I don’t know if it’s in other disciplines as well, but somehow like the academia is a little bit like with an antagonistic attitude towards commercial business where to drive this thing forward, the money has to be behind it like you.
Invent whatever you want in the academia. If it doesn’t make it to [00:07:00] people and it doesn’t benefit, doesn’t provide value, money is never gonna follow. You can apply for millions of grants. That’s gonna be your only money you ever have gotten, and this research is gonna just be stuck in papers. So I’ve been promoting a lot of open source software, but now I also wanna, because digital pathology is the academia, is the practitioners, and it’s the vendors as well, because you guys as vendors, In which area of digital pathology you are those who are taking the hit from all the clients.
And your job is to actually incorporate this feedback as complaining about everything that’s not working perfectly and it’s not gonna be, and you take it and you make a viable business and you provide enough value so that people follow with their money. And I think this is something that should be emphasized more that we should work together.
And that’s why I love the startup stem from academia and then decide to become a business. So that’s one thing, and I very much salute you for that and the other thing is, tell us about biobanking, just like [00:08:00] a bird’s eye view about biobanking. You already mentioned, okay, research is slow because there are no samples.
So they’re biobanks with sample. What’s the workflow? Where do those samples come from? Who needs them in general? And then how do you assess those samples regarding quality?
Leif: Which do you want first, the first question, second first.
Aleksandra: Let’s do, where do they come from? Where do the samples come from, for in biobanks?
Leif: As you can imagine, for biobanking, the biobanking piece is really just the repository and inventory.
The samples that are collected, but they come from all over. And each culture, each society has a different way of thinking about the human body and the materials that come out of the human body. So in many cases, you want to have a consented sample, one that is ethically consented, you know it’s not gonna be misused, and that it’s fit for purpose.
And you have to go through this process, but the samples can come from anywhere because it’s along the, what we would call the patient continuum, right? the minute that it’s identified, even before that, understanding pre-disease into disease and then falling it, and so that involves a lot of different [00:09:00] practitioners.
It involves a pathologist, obviously, but before that it’s your general practitioner, then it’s maybe a subspecialist into, oncology and the pathologist has a look, but then, If it’s dramatic pathology, then it involves them. Then there’s the laboratory that has supporting with the testing. So you could have anywhere from just regular CBCs and normal bodily function tests that go to a LabCorp, for example.
Or you can have genomic testing that goes to foundation medicine or standard biomarker testing using IC immunochemistry know, so there’s so many different elements to it that are supporting this process include clinical trials in that too their clinical trial pathways that involve those doctors.
And so that’s what makes it the most comp one of the most complicated things and a lot of people ignore the fact that it’s really important, but it’s super complicated because you have to involve all these people and say this consent travels to this doctor, and who owns that material?
And is that material then allowed to be used in research? And what are the applications? Most of our ethics are around in biobanking around using the material, but [00:10:00] not making a commercial product element, so you can’t always just take cells from a cancer and regenerate those and then sell them for profit, you have to have a fit for purf approval.
So that’s one pathway, the other pathway is after they’ve been processed and the patient doesn’t need that material, then you can go for a waiver of consent, which is essentially saying this material’s going into the garbage, or it’s being stored for long-term use by a responsible party and then you can get those samples.
But those are then all put into different things, so bio fluids could be frozen into subsets of whole blood pellets and serum plasma, for example. We have whole bloods that are going fresh to locations within 20 hours so that they can be processed in the circulating tumor cells and identified in that we have the pathology blocks and slides that are going to other people to be stained.
And retest it and then we have a number of other frozen and other products that come from the human body, including any bodily fluid, essentially in any solid tissue. And what makes it difficult for a lot of things is that we know enough [00:11:00] about the solid tumor tissue to know that we need to know what happened before and after in order to say was there a change?
And that becomes difficult because that’s not standard of care necessary you don’t always get biopsy before you have surgery, for example. And they identify, pathologist looks at it and says, yes, you have cancer. Then they may hit it with a adjuvant therapy or some therapy and initial to shrink it and start attacking the cancer.
And then you have the resection, but you may not have a subsequent biopsy because they say we removed that tissue. And so that we’re really have to weave these parts together for pharma and to say, to get access for them to do these things. And so it’s a labor of love and there aren’t many people do it because it is so complicated.
It requires, sometimes it requires academia because of the way thes or like in the UK it’s a socialized system, which is great, but they buy things differently they have different pathways. Or in the United States we have privatized system, oncologists, groups that have nothing to do with the hospital and they’re completely disparate, and so they’re working together, but they really aren’t under one control.
So, it becomes very difficult from a legal perspective. [00:12:00] So that is the emphasis of the biobank is like, how do you get access to the tissue and then it’s stored in the biobanks. Once it’s in there, one of the problems is it’s in there if you don’t annotate it properly, if you don’t know what’s in there you don’t know what’s in there again.
And then, when it comes out, it has to be rechecked and so it may need to be retested completely and so it goes through that whole process again. And that’s where we, we had been working on this for years with protocols and SOPs and training people saying, here’s how you handle the sample.
Here’s how you consent a patient, here’s how you handle the sample, here’s how you assess the sample, here’s what data needs to go with that sample. And very meticulously orchestrating that. But that is very labor intensive and it’s not the priority of those people that are providing the care. Their first priority is the patient and then maybe subsequent to that, it’s maybe a clinical trial and then research. And so we’re the third person, if not the fourth person in line and we have to be very thorough.
In this process and what goes in, you may not know what’s in there after a while and over the years you kinda lose track of that.
And you also, as those materials age, they lose relevancy, right? So, you don’t [00:13:00] know if there’s a new marker out, for example, if it’s K R A S or E G F R or ALP or an immunotherapy, those markers are new, so they’re only in the new generation of tissues and you’re immediately.
Aleksandra: You’re newly expressed and newly discovered.
Leif: Yeah. So it’s already becoming older and less valuable to the researcher because it’s now, it doesn’t have the information. So that’s the where that inflection point, we have been waiting for years thinking about how can we do this better? We talked to all our clients and say, how do we do this better?
And a lot of them say, we don’t know how you do it. We’ll share our manual with you, we’ll share our SOPs with you and they’re like that’s just the way it’s always been done and that’s the most common answer. So in the Biobank, we have biobanks, we partner with biobanks. It’s a great way to get tissue that isn’t required to be fresh necessarily.
But again, it becomes this kinda nightmare scenario where you can have lots and lots of samples that are frozen in time information wise. And what we want to do as business people is we want to turn that inventory over. And that means working with all types of people, [00:14:00] academics, government, pharma, diagnostics, whatever the case may consortiums, we wanna turn that inventory over and a lot of them just host that inventory and it never comes out again because it’s so much effort to get it in there.
And then forget about the part which is researchers have access to it, they can do more with it and the more information.
Aleksandra: Yeah. It was collected for in the first place.
Leif: Yeah, exactly.
Aleksandra: Definitely not to have great repository that sits there because then the value depreciates with every week that it’s stored there.
So, I heard about you from a fellow podcaster from Denny Strang who had the People of Pathology podcast, and he always gives me hint, who’s a cool person to interview about digital pathology that he’s interviewing. So, I would love to hear the story from you, how you guys pivoted from being a biobank and being aware of this problem of the how to assess the tissues and how to check their quality to solving this problem with digital pathology and offering this as a solution to other biobanks?
Leif: Yeah. And we really appreciate Dennis [00:15:00] interviewing us, and it’s important to get the message out there. So in our process there are certain things when they get into the biobank or we get access to them, there are, there’s information in there that is from the pathologist, for example, that is in written form or a limb system or HL seven format where they basically have some data on those tissues.
But that data isn’t really necessarily explicit to research. It’s explicit to care. And so, there are features in there that we want to know in those tissues, and it starts with this.
So, if you’re building a new diagnostic in your companion diagnostic or a new biomarker, you have an idea from basic research. You’ve done some samples and you say, okay, I found this very interesting marker. Now I need to get more samples. Let’s go to a biobank and get those samples.
They come to us, and they say, hey, we want these samples with this biomarker on it. We say okay, we can find those tumors and we can find those formats if you want. And then they say, we want to make sure that there’s enough tumor purity tumor material in the block, for example, that allows us to say that we’re buying something that we can then test our assay.
So [00:16:00] what we would do is we, for years we have a pathologist on staff that pathologist, we regard very highly because they have to do a lot of work and it’s sometimes it’s beyond their daily work that they do for care, but they essentially have to look at every single slide and then they have to annotate what is the tumor purity on there.
How much percent is of that slide by surface area is tumor and how much is necrosis and how much is normal tissue? And this is something that is, for a pathologist, it goes to school for 12 years. It’s pretty low on their totem third experience, but it’s not the greatest use of time.
And they’re the first people to say, we would rather not do this. But for research, you want to know that. You want to know if there is enough DNA in that tissue or RNA to have a successful test.
Aleksandra: Yeah, it’s a mandatory information. Otherwise, if you don’t have that, your research can go to the garbage.
Leif: Go to pieces, right?
So, we would do that, have our pathologist look at the slide, annotate it, give information by based on surface area, then we send that block to a farmer, anyone who’s using research ready to use it. Then [00:17:00] they actually cut their own slide, which now the next layer down, they stain an H and E on it. And then they do their own analysis with their own pathologist.
So now we have competing opinions about the same thing. Now you’re doubling the work. Instead of trusting like this is a confirmed tissue with this percent tumor necrosis. They doubled the work and because of their skepticism around this tissue, and where they come from and how they’ve been obtained and their age, they do this anyway.
So that was a painful process for we could deliver these tissues that they wanted and then it could take three to six months for them to get their pathology pipeline into a place where they could analyze because as we all know, there’s only 1200 something pathologists, board certified pathologists serving us.
And then, if it’s a vet path thing, it’s even fewer. We basically wanted to remove that situation and for years have been pining about how do we remove this? Because if they have the sample for 30, 60, 90 days, we’re waiting for them to tell us it’s right or wrong. And then some of these tissues are rare enough where whether it’s a rare indication or it’s a rare access point, like I was saying, within the clinical care, or it’s got a [00:18:00] marker that is, 0.5% of all of those cases.
So, you don’t exactly have them as much as cancer is prevalent and horrible, it’s not exactly accessible in the way that you want. So we wanna know right away if they want that tissue or not. So that was the impetus for us when we started looking at AI and digital pathology, we said once you scan that image, we can then do things with AI to count tumor nuclei and look at percent tumor necrosis by surface area.
That is done very quickly. And then a pathologist can just say, I agree with that. I agree with what they said and for, I’m telling you something you already know, but for pathologist account tumor nuclei is a really mundane.
Aleksandra: Could be a break. No.
Leif: And it’s not within the human eyes. It’s not a, not an ideal situation, but computers do this stuff really well. So we raised the bar by saying, we’re not just gonna show you by surface area, we’re gonna show you by tumor nuclei, how much is in there. And what it does is it opens up these tissues that often would. Bypassed by percent surface area, and they would open up and say this actually has enough tumor nuclei to have reactivity to show that the assay [00:19:00] works or to test the assay.
And so what you find is in the United States, for example, they have, the food administration has grading of strawberries, for example. And John Wetzel, my co-founder, likes to give this example, is, there are strawberries that are perfect that you see in the supermarket. And you’re like, okay, so that is a perfect strawberry and I will eat that strawberry.
But the strawberries that go in the jam that go. Jelly that go into other things are not the most beautiful things and but they have utility and that’s what we found about the tissues is the most beautiful tumor tissues don’t always mean that they’re the best research tissue. It may be that this is a ROS one, which is very rare, 1% of the population and it’s usable.
It may be really dense and a really small on this by surface area, but it may have enough data there, enough utility there that we can make access that. So we’re opening up our biobanks even further now because we can say these tissues are actually usable. We can prove to you they’re usable. And we remove that step where we’re doubling up the effort to have them see, have them accept the tissue.
And what we’ve done is we dig with digital pathology; we can show the h and e [00:20:00] and a pathologist can zoom in and look at it our research can zoom in and look at it. They can also see our overlays and they can see the tumor nuclei; they can see the normal, and I can show you the video if you want.
Aleksandra: Yeah.
Leif: But basically.
Aleksandra: But just to give like a little framework, so you had a biobank and then you decided to introduce digital pathology, and the first level was just scanning your stained H and E so that the viewing of the specimen is reduced to viewing one and the same specimen by whichever pathologist wants to view it, right?
Leif: Yeah.
Aleksandra: Then the second level is, okay, let’s eliminate the manual work of annotations and use image analysis and artificial intelligence based image analysis to delineate the structures that the pathologist had to do manually previously, and then show this as well so that everybody who’s interesting can look at it.
And I remember from what you showed me before, it’s like a catalog, you can basically pick from a catalog.
Leif: Correct.
Aleksandra: Of the slides that you wanna use for your research with all the information that’s available [00:21:00] above the slide, right?
Leif: Yeah, that’s right. When you get into digital pathology there are companies that are doing the digital pathology supporting digital pathologists, but the value in the utility is that you give the image, gives you portability, right?
And that’s where the world has changed. It’s great if you can do digital pathology online and it enables pathologists from all around the world collaborate. That is absolutely fantastic and that’s the first step. And to me that’s like where there was Lotus Notes 1, 2, 3 and then Microsoft excel and was like, okay, so now we can put these Datas in data and spreadsheets and we can put this information.
In the next level is that portability? Now I can send that to somebody, and someone can read that. So that’s where we started to use image handling, distribute our network and expand our network of pathologists that can review these things and allow the risk end user, the researcher to see them.
The next level is the AI, it’s really starting to interrogate these samples and say what’s in these things and the beauty of the AI is that I can go back and add it later. So we’ve developed a system where a user can upload an image very easily it’s called Arch then you can choose which AI you wanna run on it. And so those AI [00:22:00] can be for research use only, or they can be clinically approved diagnostics or what have you.
And we expect more and more those come around. But for us, the first part was, let’s start with quality of the tissue, because that supports our business. The next level is starting to look at lymphocytes and starting to look at different markers, PD-l1, and starting to look at not just the h and e but look at the biomarkers that are in immunochemistry.
So, we’re building out those apps and adding those ourselves but as there are many authors out there that are working on these things and we actually created a, an enablement within our system so we can actually distribute those images. Images their AI’s have never seen before and have pathologists validate very quickly, is this working or not?
Do we agree that this AI is working? We also can introduce bad samples of poor quality this way very quickly. And in that process of validating a diagnostic, you not only have to have the statistical power in the perfect sample, you have to have it showing that it has efficacy in all kinds of samples.
So we add now the ability the portability of digital images to give them normals, to give them things that they would need to prove [00:23:00] statistically, in fact, not just that ai, but the diagnostic is actually accurate. Yes, we added AI and we were able to add any AI to our system the way we structured it in the cloud is that you can call those ai, and that’s where we commercially charge.
We don’t charge to upload the image or anything, but we do charge to run the AIs, and our business model is almost like cable television where you pay for the. You get the access to the cable and then you pay for content essentially and we hope to get lots and lots of ais in there that are validated, whether internally validated or, and what I mean by that is let’s say it’s a university or it’s the NCI using something internally to say, we wanna see if this is biomarkers really effectively being identified by the ai.
From there, it gets into multimodal, spectral imaging, three-dimensional imaging, and all sorts of things where we can then build on our platform to do those types of things. That’s a little futuristic that’s where we’re looking, but It is where we want to go.
In this point, we are able then now to automate these processes, so we actually have it where our system, you can tie your scanner into our [00:24:00] system, it’ll start to log all the information, and this is where the biobanking comes in. Essentially it starts the biobanking process where you have your consent recorded, you have your pathology reports recorded.
You have all your data in there you have the images in there. Now you have a dosier around this biobank material that you can then interrogate over and over again, but you can also distribute it anywhere you want and we feel that’s very powerful but then you can automate processes where if a company wants to batch those, and let’s say that the pathologist is an available 24 hours, 24 7, 365, you know the computers are, and they can run batches while you are sleeping.
The computers run the analysis then they show you the results and we can set all sorts of parameters around minimums and maximums and thresholds that say these samples failed our criteria. We would like a pathologist to review those thoroughly, open up the case, look at it, annotate it, and give their feedback.
These past, no problem, we still need a pathologist maybe to look at them, but it’s not gonna be four to 20 minutes a slide. It’s gonna be.
Aleksandra: Yeah, it’s gonna be instant approval rather than sorting [00:25:00] out what went wrong.
Leif: Yeah, so we furnished the data to them. The pathologist looks at it and said, I agree with that, they press a button, their signature is then tied to that, and it says, that is, I agree with that. I’ll come and it makes the throughput a lot faster, especially if you’re trying to test a lot of material. It makes that throughput a lot faster. Now we’re chipping away at that time that it takes to find samples we’re able to aggregate samples from all over the world in certain indications very quickly, and we’re chipping away at that time that takes 12 years to get a companion diagnostic or more for a therapeutic, right?
That’s where digital pathology really, we decided that we wouldn’t play in the area of trying to reproduce what pathologists do online, but we will enable, we will use AI to enable people to take out the mundane, the stuff that people don’t want to do, that computers do well.
And then just show people results and let them make an assessment of what happens and so we are able to then move that on to automation and notify people of batches and things like that. But also, again, we can use the internet of things to access different devices in the laboratory. And now those [00:26:00] things become of that workflow becomes completely automated where humans aren’t touching slides and bringing to the next person, here’s the pathologist, the pathologist then does things.
We actually have it so that it can really automate, and we have it, we’re to the point where the AI can circle the area, the region of interest on it, on an h and E, which is done manually now by a pathologist, right?
They hold up, they see the two area. They say that looks like the most tumor will circle that, and then we’ll hand it off to the laboratory to scrape those and put those in tubes for dissociation. Now we have it where we can print out a report either electronically or on a piece of paper that’s valid.
And essentially put that slide down and then they could scrape their material and put it in tubes and so that helps take the burden of manually circling things choosing areas where the tumor’s most dense. And a subsequent step to that is that we actually have robots that we can call that can punch those areas of highest tumor density.
Aleksandra: I wanna talk about the robots, but let’s.
Leif: And put them in the tubes.
Aleksandra: So, we said digitization is the super, very first step so that everybody can look at the same time we have AI [00:27:00] for different whatever you can use image analysis slash ai four. And you have infrastructure for AI suppliers, let’s say image analysis companies to plug in into your image management system, which is gonna be the third thing.
So, you guys like in a hub where people, regardless of their needs, can plug in with whatever they have, and you now mention internet of things and trying to remote in or connect with external devices and your robots.
Leif: I can show up video. This is essentially an ovarian tumor tissue that we run our AI’s on, and we put this technology using this application as Visiopharm based that we developed with them and.
Aleksandra: Feel free to talk about the partners that you have. We wanna know the names of people who are working with you because we wanna spread the word about everybody that can basically integrate with what you guys have.
Leif: Yeah, absolutely. So, we chose Visiopharm because they have 20 years’ experience in AI and we tested them against other systems about the how accurate the [00:28:00] ground truths are, how they make adjustments to different images and things like that and we felt that they were, that’s where we wanted to start.
Again, we can use AI from anywhere essentially, as long as we, our pathologists can validate the accuracy of them. And like you said, there’re gonna be things like, let’s say you have prostate samples and you want to just have the Gleason score. The Gleason score is not in the pathology report, but it’s in another report.
We can go back and run a and Gleason score these very quickly so that we know that they fit the criteria of that study, which we have to keep in mind. Basically, it runs through, and it looks, the green in this case is normal it’s stroma there’s actually, it’s not just showing green as if it’s not tumor.
There are actually calculations based on the cellular structure in the pathology there to say that is actually normal. The red in this case is necrosis you can see the sample’s actually pretty good one. A lot of necrosis and the purplish blue is really the tumor.
And what we were able to do is and so you can see here now we show the tumor grows densely in certain tumor tissues and I’m not gonna tell pathologist what I see here, essentially, Aleks, these outlines then become very useful.
And if you were doing micro dissection but what we were able to [00:29:00] do is we can verify in these tissues that these, in fact, are tumor nuclei in the pinkish purple ones, and then the bluish ones are essentially normal nuclei. So we count those and we’re able to show that this tumor has 46,509 nuclei, which is plenty for most genomic profiling tests, dna, rna, even good plenty for ic.
Is a good example of a tissue. I could show you a tissue that has a very small microdot of tumor but is actually in fact highly usable as a marker on it but it’s 55% tumor that’s pretty high. But now you see we did a heat map, so we know where the tumor is dense, and this is based on the physical distribution of tumor nuclei where it’s dense.
And so, then we’re able to really start to identify where we would punch this, whether we’re doing a tissue microarray or we’re doing something for DNA testing. And so, we then have a collaborator with.
Aleksandra: So, can exactly go to the high-density area marked in red and have your cores taken out from there.
Leif: That’s exactly right.
Aleksandra: The cores scraped or whatever your robots [00:30:00] are doing.
Leif: That’s right. And one of the reasons we chose Visiopharm is cause they essentially, we keep an XY coordinate of every nuclei. So we know if you punch one area because you feel that it has the highest density you still have, in this case, you have four other areas you could punch, later on.
But we know where all the nuclei so we can map those out and that’s what generates the position of the robot.
Aleksandra: So, you keep, you trace the use of your sample as well?
Leif: Yes. Yeah.
Aleksandra: So, you could basically service, let’s say we have five circles where it could be punched. You could service five clients with this one sample rather than selling the whole thing to somebody who’s just gonna use part of it.
Leif: Yeah. So we can, what we call syndicate this block so that there are more researchers doing more research, having access to, like I said, if this were a, this is a variant, but if it were a ROS one, for example, and you say we only have so many of those now, let’s.
Let’s try to make it so that more people have access to the research and part and parcel of that is if it’s internal, you can always go back and now you have other studies that you can do simultaneously. And being in the business for 20 years, essentially for acquiring samples, one of the [00:31:00] problems is that you don’t, we see these studies come around.
We have a view on the world with pharma and diagnostics, where we see the study come back around every so often. There’s a publication then they dust off those studies and say, we gotta go into that space and that is really inefficient from a development perspective, right? Is if you didn’t have the samples and you didn’t have some evidence and some progress in that study, they’ll table that study, new CEO comes in and he or she says, we’re gonna focus on these areas.
And then you table all the research and so there’s a lot of research that’s idle and a lot of programs that actually, come out years later to be essentially worthwhile. Every four years, we see them turn over and say we’re not gonna do that study anymore. We had such a hard time getting the samples, and that researcher may have four or five studies going simultaneously because they don’t have enough samples to prove and enough science done to prove to the next level that they should continue to pursue that and invest in that area.
So being able to go back to these tissues, reuse them, re interrogate them, now you always have the digital image, it goes back again. That original image is great because you always go back and interrogate that and say, what did I see there?
And that’s where [00:32:00] multiplexing and different layers of information start to build and then, You continue to build that, what I call a dosier about that, a file about that particular tissue that gives it real information over time. And if you’re doing safety talks or you’re doing something where you want to go back and say, I want to see what this tissue looked like pre-post or whatever, that becomes really, has a lot of utility.
And for years a lot of this stuff gets stored in basements or in biobanks and you never see them again. And then you say, Aleks, do you remember when we did this safety talk study? For this therapy or even something as simple as like silver or things that the FDA requires you to show and you’re like, no, I think John has that file.
Aleksandra: Yeah.
Leif: It’s really not efficient in that way so.
Aleksandra: That’s a big pain point in drug development research in general and now we have systems before we didn’t even have systems, and now we have some systems, now we need some awareness, we need to make aware.
Make people aware of that, you do need image management system for digital pathology. It’s now more a couple of folders worth of slides. It’s [00:33:00] gets out of control so quickly and if you wanna interrogate this data in any way, you need software for that.
Leif: Yeah. And we’ve come a long way, but this is the beginning of that, being able to stack this data and look at it ballistically.
And I think that’s, again, we’re chipping away at the inefficiencies and limitations that were just technologically, being on paper or being excel shared spreadsheets and being able to bring this to light and deliver it in a way that is meaningful for research and development, right?
Aleksandra: Definitely. And I think, in Digital Pathology World, there is no discussion if you need a scanner or not, right? If you wanna do it at scale, everybody’s gonna get the scanner, but still all the other things like maybe image analysis but maybe not. Maybe image management system, but maybe not. Maybe integration with something, but maybe not.
Leif: Yeah.
Aleksandra: It’s no more maybe not. You have to do it if you wanna reap all the benefits of what’s possible. Scanning is just the first step, and I think slowly we’re starting to realize that and like in any discipline, any area of biotech of life sciences first, everybody starts [00:34:00] working on their own in their little silos, and then slowly they start coming together.
And you guys are a super example of all those developments coming together and deploying it for others to benefit with business strategy in mind because, if money does not follow those developments, then they’re not gonna be distributed and they’re not gonna be used.
Leif: Yeah. They fail essentially and that’s, and I’ll comment you asked a question earlier and comment on funding and getting from research into development and stuff like that. We’re very passionate about samples. It’s not the only thing that limits research and development, but it definitely gates the progress that happens in there.
And so, when you talk about money and how much money it takes to do something and getting money to flow, you really need to have results and results come from, being able to do research like I said, I know lots of very smart researchers in pharma and diagnostics that just, I impress me every day what they do.
But at the same time, they can’t make progress and so why does the therapy cost billions of dollars? It costs billions of dollars. Cause they have to do a lot in between stage three and stage four [00:35:00] development that is not being done in the initial rounds to prove that it’s actually good.
And so, the cohorts are small in the beginning those areas are underfunded. That’s why you have things like in the United States, it’s the NCI, they do a lot of great work and they’re trying to do more with tech transfer to get that going and funded, but there’s a handoff there.
In academia, there’s actually tons and tons of good work being done, but it’s up to those tech transfer offices to notify people of what’s happening and a lot of times you’ll see that the statistical value that they have in their studies.
And their study designs are very weak and a lot of that it comes from, they weren’t able to access the patient to get those so if you have to prove it in a subset of 12, maybe that’s doable. But if you really have to start to prove it in somewhere where the statistical power becomes interesting, that doesn’t happen and they get caught up in this web. And so again, they raise more money, they try to raise more money on the proof that they haven’t, and then it becomes more of a, you know how interested it is the person who has the purse.
And take to the next level we hope that we can get more of this information out. Cause like I said, 20 years ago or longer, we had some of these therapeutics targets that are now [00:36:00] just emerging and for future generations, we hope we can do this a little bit better, a little faster. We’ve been behind a lot of novel diagnostics, alk, and different things like that.
Our samples have been supplied to those people that drive those innovations and immunotherapies, and we’re very proud of that. Taken a long time. John and Phil both lost their hair over this process so.
Aleksandra: You didn’t.
Leif: I didn’t, but someone have to lose it. But, so we try to empower this and our technology is made in the cloud so that people can access it so that people with one doctor, one PhD, one whomever, who’s starting to look at these, can upload their images essentially for free.
There’s no set up costs there’s no, get a scan, but we have partners that can scan those images. If they don’t have a scanner in house, we can subsidize those scanners so that they pay over time. So there’s not a huge lift to get those in, and we allow them to get that first part, which you pointed out was get the digital scan and then subsequent to that, it’s analysis.
For a regular person to do. What we’ve done up to this point is $99, which not everyone has access to a great [00:37:00] pathologist I live outside of Boston. There are so many biotechs, it’s amazing they can’t all hire the same pathologists cuz there are only so many pathologists. So we try to empower them.
We have, we do have partners with partnerships, with pathologists that can help with research that are board certified. They’re allowed to do research support, but we try to enable them as quickly as possible, as low a price as possible to do that. Now, if you’re a big company and you’re gonna do thousands of these, or you’re a genomic profiling company, this is where this automation comes in big because you’re waiting, let’s say a pathologist works an eight hour shift, you know what happens the other hours of the day and those tests accumulate and you really get reimbursed on the pathologist giving a diagnosis at the end of the day, or making a call on a test.
That’s where the value is. So we let the machines do all that and that’s where this robot comes in. And I’ll show you that right now. It’s basically the, we partner with a company called Isonet out of Italy and these gentlemen essentially built.
Yeah. Isonet has a really great technology called Galilee Thematic, and we joined forces with them to automate the process of punching the tissues. Now their original robot was for TMAs, which is great, but [00:38:00] we did it for DNA and an RNA extraction. And hopefully there are future tests that go this way too.
But you can see it can take different formats of FFP blocks. And we’ve already scanned the images upstream and run our AI and then delivered that through Visiopharms AI, which we put in the web. We punched, posted it on Amazon to the robot and the robot know. Where to punch those blocks. Exactly. That is so cool.
And then we actually have it so that you have certain ro other robots, it’ll pull that tray.
Aleksandra: Hamilton Robots.
Leif: Yeah, Hamilton Robots in this case, it’ll put it into the dissociation and those tubes are filled and already starting the process of getting the DNA ready to be sequence. So we’re, picking away at these workflows so that the human doesn’t have to handle them as much.
That doesn’t mean that people are gonna be let go from their jobs. It just means they can do more and.
Aleksandra: They can shift their energy and expertise into those places where the machine cannot help us, and I am starting to see this as a trend where the manual digital pathology workflow is being [00:39:00] shifted to the automated one because we couldn’t totally do it and we should do it.
Leif: Yeah.
Aleksandra: There is no reason to have, more overhead for digital pathology workflow than we had for a normal pathology lab.
So Leif, before we go and before we tell everyone where to find you online, if you can tell us, list all the people that can work with you, like who can do business with you, who do you do business with?
Leif: Yes.
Aleksandra: So, there are different parties that can be involved with you and how?
Leif: We built this using salesforce.com. So, we built it on that platform because that is, it’s a misnomer it’s not just for sales, it’s for, they used to rebrand as force because it handles all the data that comes with this.
So, your account down to your sub user accounts. So that allowed us to do certain things. One is that we’re HIPAA compliant, we also have national security agency level security on it. We are GDPR compliance, so we can service anyone in the world and there are people that are in countries with few pathologists that we can service those people to.
And so, we put it in the cloud and [00:40:00] then we moved the VisioPharm software from being hosted locally to be hosted in the cloud on Amazon Web Services. And that allows us to distribute and scale these services anywhere in the world. So, we can not only put it in a certain language, we can not only secure it for privacy, but we can actually distribute that information and localize information and scale up those servers.
So, if you had one day you had 10 samples and the next day you had 3000 samples, it knows scale up those servers so that it’s not like adding one pathologist, it’s like adding enough pathologist to push that in information through. So, what that means is we can service one person doing research that’s an academic or in a small biotech, we can service people that are in large biotech, medium biotech, any size we can service the, we have it on Gov cloud.
We can actually do it for the NCI, the NIH, and those institutions that require a government, pardon, security, system. We have it so that new molecular profiling companies, life sciences companies that have new platforms. Anyone we can do animal companies, we can do companies that are veterinary practices that are [00:41:00] with everyday pets, and we can do anything that comes with preclinical.
So, the way we built it was to be as encompassing and welcoming as possible, basically… And so, you can set up these accounts that can set security for you and your subordinates, or you and your peers. And I can log those things and I can show those logs on when people access them and really, it’s meant to be for, to expand our horizon.
Because one of the problems in we do in Biobanking is I have a great biobanking of white people and I don’t have representative populations throughout the world. We’re not just making drugs for one person; we’re making drugs for.
Aleksandra: Yeah.
Leif: Cancers that affect everyone.
Aleksandra: Known problem in healthcare in general and all the studies and all the research, we just do research on a certain sub population and mostly it’s a white male. And that’s not all the people in the world but.
Leif: No, there.
Aleksandra: No, there are some other people.
Leif: It’s true. There’re more women than men, right? So let’s be honest about it.
Aleksandra: So also, so okay. Just like to simplify it. Super simplify. What do you guys sell? I go what do I buy from you? Do I [00:42:00] buy the samples? Do I buy the data with samples and like samples with all the information? It’s not clear for me, like I can upload my samples. What do I then buy from you? The output of an AI algorithm, there are a couple of different things that I could give you money for, what do I buy from you?
Leif: Yeah. So, there’s three pillars of this. One is the, AI’s, right? You can buy AIS from us, run your own images, and those, you just pay a fee for those. You can buy the sample material too, that is blocks. If you need that research, you can buy the slides if you need those, you can buy Communistic chemistry services on those slides.
You can buy different things. So, we have a partner with a laboratory that allows us to stain and do all the molecular profiling that we want to do on those tissues so that when you get it, it’s got the characterization you want. and you’ve got the image. You can also buy the images and license those images so that you can do your own AI.
We partner with Visiopharm not only for, hosting what we wanna do, but they part we can allow them to install the authoring tools so that you can author your own ai, so you can buy those [00:43:00] images to start to do your own training. Ai, that’s all in what’s called a marketplace, and the marketplace is just transactional, right?
So, we can sell, buy, and sell anything through that market. Suppliers can put their images and their things in there for other people to buy, and buyers can come in and buy those things that support their research. Then once they buy The AI’s, they can buy storage space and further that. Then there’s a separate product called Lab Flow that they can buy the laboratory workflow that allows them to automate these steps.
To in their laboratory. So, it’s, as many hands free digital handling as possible and that comes as a user base. So, you buy user subscriptions.
Aleksandra: You guys have like everything?
Leif: Yeah. Unfortunately for us, we have so much experience with this that we knew we couldn’t just do one thing in order to achieve our goal.
We had to do a number of things. So, we built these three pillars, which is the AI, the marketplace, and then the automation and the automation.
You can buy the instrument through is net we can buy that instrument so you can, we can put in the automation and the robot so you can automate that tie into the lab flow, which ties into the marketplace system.
So, there’s a lot you can buy from us to say the [00:44:00] least. But what you find is when you talk to someone who’s doing digital pathology, the lights go on and they’re like, okay, so now that we’ve committed.
Yeah, you’ll need, it’s not that you just need one thing, you can compartment. And you can do incremental if you wanna reap all the benefits, which is when the cost justification is actually happening, where you can justify, you need more than just one step.
Yeah, totally possible to go incremental to check if it’s working for you when you decide to deploy and gain all the benefits then you have to go all in. That’s at least what I’m seeing in the industry and if you can partner with somebody who has experience with that and you can learn from their mistakes, that’s a big advantage.
Aleksandra: Cuz mistakes cost a lot in healthcare.
Leif: Yes, this is so true. We partner with limb systems. We have APIs to limb systems. We have partnerships with image handling systems that are already in place like Prosha and the, and folks like that where, they already have some handling in there.
They’re doing traditional pathology relief, but we’ve put in this infrastructure and the way we built it is [00:45:00] you can incrementally add as you go. You don’t have to buy the whole shooting match. And that was one of the problems. Even implementing printing systems in the United States, for example, was like, okay, buy this IBM machine and then 10 years later you can buy new IBM machine.
And it’s this we think things are moving so fast in the imaging side and the scanning side of things that you don’t wanna put a lot of money into that. So we have a replacement system where we can place those imaging handling systems and scale them up. And what we want to do is have a low barrier to entry so anyone can sign in and start to do this stuff.
Low fees so that lots of people can do it and then just have lots and lots of capabilities that then grow them into what we believe the future is, which is, people like PATH AI and p AI have diagnostics that are interesting, we can deploy those on this system if and when they’re ready.
So that’s nice. It’s an end-to-end system and hopefully I was able to articulate it relatively clearly for you.
Aleksandra: So, Leif, where do we find you online when people wanna go and buy stuff from you and engage in this digital pathology journey?
Where do they find you?
Leif: Yeah, so www trimetis, T R I M E T I S L S as in life sciences.com. [00:46:00] And then if you click on the arch system, a rch, it’s the acronym for abbreviation for accelerating research, changing healthcare. And click on there, create an account it only takes a couple minutes. you only have a SaaS agreement to get in there. And then depending on what you wanna do, if you wanna buy services or you just wanna peruse, it’s all there.
And then, that’s it. That’s all there is too.
Aleksandra: Okay. Thank you so much for joining us and for letting us know about you and I hope you have a great day.
Leif: Thank you. You as well. Appreciate the time.
Aleksandra: We try to pack a lot into this episode, and I know you might be a little overwhelmed with the breadth of services that TriMetis is offering, and with all the things that you actually have to do to be able to leverage the power of digital pathology, and it’s not insignificant.
You can start small, but to leverage the full power you would have to go all in. So be sure to check the website and reach out to them if you think there is an area of overlap where you could work together.
