We recently had the opportunity to speak with Mark Lloyd, Founder and EVP at Inspirata, which provides a leading clinical digital pathology image management platform. Digital pathology is a space we track closely, and this interview is the first in a series of key opinion leader discussions.
● While reimbursement is key for the long-term clinical success of digital pathology, there are many business cases to justify digital pathology (DP) adoption for clinical sites today. For example, digital pathology can raise the average level of care across a hospital system especially in remote settings, facilitate cost savings on FTEs, serve as a point of differentiation for reference labs, and allow hospitals to insource patients that would otherwise be unreachable.
● In the clinical setting, image / slide management software must function as a central hub that facilitates streamlined workflows with all other software and hardware the pathologist uses. Additionally, partnerships between these pieces (e.g., scanner, slide management, and AI tool companies) allow pathologists to adopt the best version of each technology. Though consolidation is expected overtime, partnerships are essential for technology access.
● Novel predictive and prognostic algorithms are key for driving forward digital pathology adoption. Diagnostic algorithms and workflow efficiency tools are valuable, however, as long as they target only tasks that pathologists can do manually, they will not be “killer apps” fueling broader DP adoption.
Mark, thanks for joining us. I want to start this discussion with an introduction and a brief overview of your experience with digital pathology (DP), and your company, Inspirata.
Hi, I'm the Executive Vice President and Founder of Inspirata. My background is in cancer research; I did my PhD looking at the ecology and evolution of cancer cells. When all my friends were out there studying snow leopards in Nepal and tomatoes in Argentina, I was at the cancer center, looking at digital pathology whole side images and trying to understand the interactions happening. While I was a principal investigator in that role, I was also the director of a shared resource facility that had all of the tools necessary for any crazy idea that came through the door, such that I could quantify their digital pathology problems and provide statistical significance for the hypotheses that they were testing. It was there that I realized that digital pathology had come a long way for research, but we were missing a massive opportunity for the large-scale clinical adoption of digital pathology.
So, I founded 2DP, which stood for Digital Pathology, Diagnostic Precision, because it was all about helping pathologists get that more rapid and accurate diagnosis. That transitioned into Inspirata so that we could provide DP workflow software to give pathologists all the tools they need to be able to render that diagnosis. Our goal was to enable pathologists to make a diagnosis as fast as possible by decreasing the overall turnaround per case, receiving images faster, eliminating couriers, and giving them all the pieces of information they need from the LIS, EMR, grossing systems, radiology all in one place. Our focus at Inspirata is to help pathologists utilize digital pathology in their day to day to make it a little bit easier.
Thank you for that intro. Despite the promise of digital pathology, infrastructure is primarily limited to some AMCs in the U.S., and select EU healthcare systems (e.g., UK, Netherlands); there's not widespread access. How would you assess the current state of digital pathology adoption in the clinic and why hasn't it been more widely adopted? What are some of the key barriers and challenges?
The market is still wide open. We are still an analog medical science and for a number of reasons, we still rely on glass slides. Storage space is too expensive, there aren't enough systems that are FDA cleared… These are all things that people talk about quite frequently. But really what it boils down to is the financial challenges. There are governments that certainly have invested in digital pathology. We've been extraordinarily successful in the UK, for example, in the last two years through our partnership with Fujifilm. But when we look at the digital pathology clinical market here in North America, we have to focus on who's paying for it: the payers.
The primary reason why we don't have this kind of large-scale clinical adoption here is because we don't have the CPT codes for the reimbursement from those payors, which is largely just a function of how we pay for health care in the U.S ... There are certainly benefits to digital pathology that go beyond that return on investment, which is why AMCs invest in digital pathology and why reference laboratories see it as a key differentiator to help them in the market… However, those are still edge cases. And so, to be able to get that large scale clinical adoption in the U.S., we need support from the financial system to prop it up.
To that point, there were recently 13 new class 3 CPT codes introduced for digital pathology that will facilitate tracking of DP test volumes. How do you expect these new CPT codes to impact the state of digital pathology going forward in the US?
It's the best possible case. We need a starting point, and that’s what the CPT codes provide … These are the earliest stages of getting larger reimbursements for not just the technical components, but also the professional components which are going to continue to drive the financial engine that I mentioned … This was a very, very important milestone for us, and I’m looking forward to building on it.
We will begin to have a strong foundation for acquiring these images, then the image management systems come on top of that. Then in order to help pathologists use those images, the AI and image analysis comes on top of that, as more tools are built. Those are ultimately the pillars that will stand up digital pathology, built on a strong foundation of CPT codes. So, I was very enthused to hear about the CPT codes.
Currently, the financial incentives are somewhat complex because there's a lot of components, as you already mentioned, that go into enabling a hospital or practice to conduct digital pathology analysis. They need the scanner, the image / slide management software, and the algorithms. And today, CPT codes are typically tied to a specific use case for a specific patient. What is the typical decision-making process for a hospital that is thinking about bringing in the scanner, the image / slide management software, the cloud storage, that they need to enable this workflow?
It always depends on the intended use, right? What are the use cases that hospitals are looking to solve for? I've got customers that are hospital networks across provinces of Canada, for example, where the Canadian government has said, it doesn't matter whether you live in Moose Jaw, or in downtown Toronto, you should be able to have the same diagnosis. We have a similar scenario in a network of hospitals in West Virginia, where any rural hospital that is part of the West Virginia University System, there's over 30 of them, ought to be able to have that level of expertise of a pathologist that is located in Morgantown. Those are some of the initial use cases and metrics that Chief Quality Officers give back to payer groups. It allows them to say this was our time to diagnosis, this was the level of sub-specialty that my patients received. Those are the things that drive value in how much they get reimbursed. And that's true today. Given the way quality metrics work for reimbursement, it will spill into the ways in which organizations think about paying for digital pathology.
Other organizations, by contrast, will focus on their expenses, such as for the couriers needed to get tissue samples from the Bronx to Brooklyn. If the hospital can eliminate those 17 people that are constantly running around, that is a cost saving solution for them.
Then there are other organizations that are looking to in-source patients. Hospitals are often constrained by the number of patients that can become a patient there. There are ways in which digital pathology can help them to receive patients from other locations, other parts of the world, to be able to in-source cases that they wouldn't be able to otherwise, and these become new revenue streams. There's lots of different value propositions of digital pathology that different kinds of organizations focus on, and that is essentially how they develop their return on investment and business justification today, and that will be augmented even more as CPT codes become a little bit clearer as to what exactly they can get reimbursed for.
Switching gears, a little bit, to the image / slide management solutions specifically, what do you see as the role of these types of software solutions within the clinical lab?
I'm super biased because this is my business; this is my mission for digital pathology, and I see this as the linchpin of success. Look at where a pathologist spends their time. They spend their time in their office with their microscope, with their cases, with their LIMS system; it’s all about their day-to-day workflow. I equate these workflow solutions, whether it's ours or any of our competitors, to being the operating system on the phone. The phones will keep changing - they'll get smaller, cheaper, faster... and I can pick and choose the apps on my phone, but I still need the operating system to be able to use it. That's how I see the image analysis solutions, as the platform on which tools can be added at the discretion of the organization.
Ultimately, the place a pathologist spends all their time is in the software platforms, so it’s up to the software vendors, like Inspirata and others, to make that workflow as easy for them as possible so that they reap the benefits of digital pathology. We need to be able to help pathologists sign out cases easier, prepare for tumor boards faster, or be able to get their cases more rapidly. Those are the existing pain points that can be addressed by a digital pathology platform. These platforms will house multiple image analysis solutions in one place… It's also a future proofing mechanism because the scanners will change, the AIs will change, and they should have that one piece of glue that is consistent for them: the image management system.
We've seen Inspirata take a partnership forward approach, with upwards of 25 partnerships that have been publicly announced to date. How important are partnerships for building out this digital pathology ecosystem, especially as compared to other options, such as internally building capabilities or acquisitions (which I'm sure will happen over time)?
You know, you've got three options: build, buy, or partner. Certainly, Inspirata has built; Inspirata acquired five companies including a company from Microsoft, a company from GE, and so on. But when it boils down to it, any organization must look at what is in their DNA and what they’re best at. I'll point to some of my partners. Leica, who is extraordinarily good at histology and laboratory equipment, but image management software has not been their strong suit… And so, we look to partner where we have deficiencies. We each understand where our strengths and weaknesses are and how to partner to bring our customers all of the pieces that they need to have a complete solution… The way that I see it, I want them to be able to select the best scanner for their use case… And then when you want to do something fancy with it, and you want to look to IHC quantification or malignancy detection, it's on me to be able to bring that into one solution so that it's easy for you. We've taken that partnership approach to try to meet the really diverse needs of the laboratory. Our most recent partnership, for example, is with SpIntellx, a spatial biology company. This allows us to to bring new translational research and clinical analysis tools inside Dynamyx. Using this approach, we can leverage their unbiased approach directly inside our workflows.
I appreciate the role of Inspirata as this hub, and you have all these spokes that you're able to partner with. Do you think there's inherent value to the workflow software company remaining a third-party to the scanners and the AI / algorithms, in that you're able to serve as the software hub and the hardware and apps to change?
I consider it a future proofing mechanism for our customers, so that when other aspects of it change, they know what is consistent. And it's not that our software isn't always changing, we're always adding features and function enhancements, but we're evolving at different paces with different intentions. The intent of scanning companies might be to get a higher resolution image or a smaller file or whatever. My intent is to present it to a pathologist in a more effective manner. I think that there are advantages to these things staying separate. When you look at other industries, what you'll start to see in radiology, or cardiology, for example, is as the industry matures, there will be mergers and acquisitions of the pieces that make sense together. That will happen, inevitably, in digital pathology as well. However, right now, what our customers have the opportunity to do is to buy best of breed. I think that that is a massive advantage for a nascent industry where it has to be the best for the pathologist to adopt it, right? They need to have that really positive experience right out of the gate and best of breed often facilitates that. We have scenarios where we sell a complete solution with partners … But our preference is for everybody to focus on what you're really good at and play nicely in the playground so that we get the customers what they want. An example is Explainable AI. It’s a great way build trust and transparency between pathology disease experts and computational pathology tools. We purposely took a departure in our partnership strategy from the more traditional black-box AI approaches to digital pathology such as deep learning to partner with SpIntellx. Together we can bring customers the workflows they want from Inspirata with SpIntellx’s proprietary Explainable AI. It is together that these solutions can best help our users.
What you said about the nascency of the industry pointing towards this resonates because I think it's something we've seen in other industries over time. However, there are players out there that are developing more comprehensive end-to-end offerings: Roche, Philips, 3DHistech as examples. How do you think the standalone players fit in an ecosystem that also has these end-to-end players?
We've got groups that are end-to-end: Roche, Leica, Philips are great examples. Omnyx, the company that we acquired, was a great example of that as well. Quite intentionally, they had their own scanner, their own software, their own image analysis, and they wanted to be end-to-end, maybe they made it proprietary for a period of time. That can work for a group of customers that really just wants that one solution. What we observe is that that works to a point. And then when they need to go outside of specific use cases, partnerships start. So, you can either start from a point where you have all those partnering options available to you already, or you go to your one vendor that has a platform that is very well defined, but maybe nothing beyond that. And that's a decision point for any customer. Do they want a solution that is end-to-end that locks them out? Or do they want that flexibility? There's pros and cons to all those options though.
Now we will switch gears to the analysis / AI components of digital pathology. Currently, analytical tools are largely focused on assisting pathologists with workflow efficiencies and streamlining primary diagnosis. Which opportunities within multiplex tissue analysis, deep H&E, predictive algorithms based on morphology, multi omics, etc. do you expect to be the most impactful? Which tools push digital pathology from “nice to have” to “must have”?
Some tools replicate what a pathologist does, but either do it more quantitatively or more reliably, and those are nice (e.g., your IHC scoring algorithms, malignancy detectors, mitotic event detectors, or fungal and TB detectors). But you won't have that “killer app”, that large scale, “must have” moment until you can do something that a pathologist cannot do themselves. That is, your prognostic and predictive algorithms first, where you can use those morphological biomarkers to indicate high risk for five-year progression free survival or benefit of one therapy or another. When we look at, for example, why radiology digitized, it was not to get rid of light boxes, X rays and film. It was because PET CT scanning was inherently digital. It became critical for understanding how cancer was growing. Then every other aspect of radiology digitized behind it.
So, when we look at those kinds of “must haves”, we've got to be able to go beyond what a pathologist can see subjectively. And that can mean a lot of things. It can mean multi-omics, multiplexing, or prognostic and predictive morphologies, but all of those things provide new information. And that new information that has clinical value is the home run.
We agree that a killer application that brings novel clinical utility to the pathologist will be one of the key drivers of digital pathology clinically. Once we have that killer application, what do you think clinical implementation looks like? How will those types of algorithms be deployed?
It depends on the company that brings it to market, because of course, they've got a couple of options. They can pursue a 510K for prognostic algorithm or a PMA for a predictive algorithm. Or they can go to market in different geographic locations outside of the United States. Ultimately, it really all boils down to the commercial strategies of those companies.
What matters to me is the deployment of these things for our patients, and how we get more and more of these tests out there.... And so, my intent is, let's build these algorithms, and let's make them available to as many people as possible…whether it's an esoteric lab test for any one organization that has a locally developed test, or a MammaPrint or Oncotype DX strategy. What we have going for us here is that we're talking about digital images. So, a lot of the barriers of working with patient samples are eliminated when we utilize digital pathology. We have a lot more flexibility than the assays that have come before us. If we leverage that flexibility appropriately, we ought to be able to deploy it more rapidly.
Looking ahead over a 5-10 year period, what new technologies, developments, and trends do you expect to shape the digital pathology landscape?
I think there's two things that are overlooked oftentimes. One is the involvement of the other “ology” imaging types, like radiology, cardiology, or vendor neutral archives. Our ability to build enterprise imaging solutions for an organization, so that they don't have the silos of pathology and radiology anymore, will be key. The exchange between these groups should be far, far more facile, given the ways in which these workflows can be developed and how a CIO can manage this kind of data over different departments… I think that that is something that is undervalued today, and it’s a place where Inspirata spends a lot of time and energy to build integrations to make enterprise imaging a little bit easier…
The other is going to be the way in which we acquire images. We won't necessarily have the glass slides forever. There are ways in which these technologies can assist us with more rapid frozen, rather than a traditional frozen. Intravital microscopy, for example, becomes really, really critical toward bringing the pathologist into the surgical suite, and allowing them to have more interaction with the surgeon, rather than it being a stepwise process… Getting the interactions of the pathologist outside of our basements and into the rest of the hospital systems will be the best thing for our patients. We as pathologists are the doctor’s doctor, and we can do a better job if we have more meaningful, real-time conversations with other doctors when they're caring for their patients.
Great. Thank you so much for your time today, Mark. We look forward to following Inspirata’ developments and the digital pathology ecosystem more broadly in the coming years.