NEW YORK – Since the first US Food and Drug Administration approval for a whole-slide imaging slide scanner for digital pathology in 2017, the use of digital pathology in clinical laboratories for diagnostic use has grown significantly.
The onset of the COVID-19 pandemic three years later further pushed the field forward, as the new work-from-home paradigm and digital capabilities allowed pathologists to perform their duties away from the lab. As a result, the pandemic changed the conversation around digital pathology, according to David West, CEO of digital pathology software company Proscia, and many lab directors who were previously skeptical of digital pathology realized they needed to create a plan for implementing digital pathology solutions in their own facilities.
Those digital solutions are myriad, ranging from whole-slide imaging systems to review digital slides to artificial intelligence-based algorithms that can improve primary diagnosis by selecting areas of slides for pathologists to focus on.
But despite the growing trend, some labs have shied away from adopting digital pathology solutions, citing concerns around interoperability challenges, cost of implementation, and lagging reimbursement as primary barriers.
Early adopters like Memorial Sloan Kettering Cancer Center have been using digital slide scanners to retrospectively scan and archive slides for over a decade, but MSKCC needed an additional push to transition to digital pathology for primary diagnosis. Matthew Hanna, director of digital pathology informatics at MSKCC, said that while the use of digital pathology for primary diagnosis has "always been on our radar," the pandemic helped catalyze its adoption. MSKCC first internally validated its digital pathology solution for primary diagnosis as a laboratory-developed test in 2018 before submitting it to the New York State Department of Health for approval as a laboratory-developed test, which it received in 2020.
Pathologists at MSKCC were working at home during the pandemic, but the hospital didn't see a long-term drop-off in patients. It needed a way to "provide services and protect the pathologists," and digital pathology tools allowed pathologists to sign out their cases remotely, Hanna said.
Paige CEO Andy Moye said that the pandemic forced labs and pathologists to think differently. Many laboratories now recognize that they don't need to hire pathologists that live in the same location: With digital pathology solutions, labs can hire pathologists from anywhere in the world, providing one possible solution to the ever-declining pathologist workforce.
That dwindling workforce has incentivized some labs to jump on the digital pathology train, Hanna said. Digital solutions allow labs to maintain their work levels — sometimes even increasing the volume of samples they can review — and get subspecialist expertise quickly without having to ship a slide to another laboratory. Beyond digital slide scanning, the artificial intelligence-based models being developed for diagnostic use can also "help democratize knowledge" and give pathologists more confidence in their diagnoses, perhaps without the need for a consult, he added.
Bilal Ahmad, managing director of the pathology division at physician practice Spectrum Healthcare Partners, also noted that the decline in workforce compounded by the COVID-19 pandemic has had an impact on digital pathology growth. As many older pathologists retired early during the pandemic, it has been difficult to recruit high-quality pathologists, especially with fewer new pathologists entering the workforce. That has created a need to figure out how to balance existing pathologists' workloads and "fill the gap between the clinical need and the number of pathologists available," he said.
At Ahmad's practice, digital pathology is slowly being phased in for primary diagnosis; if everything works as intended, he believes his lab will be able to provide "pretty significant value for patients" by reducing the time to diagnosis, improving access to specialists, and decreasing the overall cost of diagnosis by using artificial intelligence tools to maximize efficiency.
Derek Welch, chief medical officer of anatomic pathology at the private pathology practice PathGroup, has already seen boosts in efficiency since implementing digital pathology for primary diagnostic pathology reviews across all subspecialties at his practice. The "growing shortage" of US pathologists "underscores the need to make pathologists more efficient individually, so that the patient population can be serviced by this decreasing number of practicing and experienced pathologists," he said via email. Most of the pathologists at his practice have proven more efficient when using a digital solution, some of them substantially so, he noted.
Regulatory approval
Since the approval of Philips' IntelliSite Pathology Solution six years ago, a small handful of other firms' digital pathology solutions have been approved by the FDA, including Leica Biosystems' Aperio system in 2019 and Hamamatsu's NanoZoomer S360MD instrument last year. The agency also approved the first AI-based pathology product, Paige Prostate, for marketing authorization in 2021.
Though the Philips scanner's approval did have an impact on digital pathology adoption, Hanna said, it's "not like scanners were flying off the shelves" suddenly. The approval served as a check box, but in his view, it only had a minor impact.
Digital pathology company PathAI's CEO Andy Beck added that while FDA approval has been helpful as a tailwind for adoption, plenty of labs are comfortable validating different technologies internally that don't go through the FDA regulatory process, so approval isn't essential for implementation.
FDA approval also comes with some problems, according to Victor Brodsky, an associate professor of pathology and immunology and associate medical director of information systems at Washington University School of Medicine in St. Louis. He has been scanning glass slides for archiving, educational purposes, research projects, and consultations with other pathologists within the hospital system. FDA-approved products for primary diagnosis are more expensive and have stricter configuration requirements, such as requiring the use of approved monitor models that may already be out of production or outdated by the time a laboratory decides to implement them, Brodsky said.
The need to stay within the confines of FDA approval "introduces a level of rigidity that ends up slowing progress" and forces labs to purchase older models of instruments that may be obsolete or less advanced, he said.
Jon Ritter, a professor of pathology and immunology at WUSTL and Brodsky's colleague, noted that the US insistence on FDA approval has been a "tremendous drag" on the adoption of digital pathology and cuts competition on the price of instrumentation. The FDA-approved scanners from Philips, Leica, and Hamamatsu are also high-throughput instruments that are unnecessary for many smaller hospitals and laboratories who could benefit from a lower-throughput and less expensive scanner.
Although not all institutions require the use of FDA-approved devices, those that do — or that have a strong institutional bias toward FDA approval, even if it's not technically required — are more limited in their options for digital pathology implementation, he said.
Interoperability struggles
However, regulatory approval often isn't the main hurdle for many laboratories looking to go digital. Instead, interoperability and the ability to integrate the technology into other clinical systems are key difficulties, particularly since pathology "is currently undergoing a digital evolution," Hanna said. Integrating new, updated digital pathology solutions with very old laboratory information systems and other technologies can be a tall order for many labs, particularly those without the resources of a large academic medical center or hospital system.
Some of the interoperability challenges are caused by the companies making the technology, Hanna noted. Some scanners produce images that can't be seen with other companies' viewers, which led MSKCC to build a vendor-agnostic viewer in house, and many of the artificial intelligence-based overlays that can be added to a digital pathology system are not interoperable or have few clinical applications.
The "seamless experience is definitely still lacking," and it will "take time for the market to mature" and meet the needs of clinicians, he said.
Although the ability to consult with pathologists across other hospitals and institutions is a major selling point for digital pathology implementation, Spectrum's Ahmad noted that his practice also has concerns about interoperability across organizations. His lab is hoping to work with a variety of hospitals and academic medical centers, but those institutions all have different information systems.
Ahmad emphasized that there is a "lack of understanding" of a lab's day-to-day workflow among vendors and many vendors offer solutions with "a lot of these bells and whistles that have been built in … [but that] have very limited value, at least for our organization."
Patrick Myles, CEO of pathology image sharing platform PathPresenter, said that there has been a movement among vendors toward "making everything work together," from scanners to software. "Most vendors recognize that hospitals want the best of everything," and those hospitals are relying on vendors to partner up from an interoperability standpoint. The field is entering an "era of consolidation" between scanner companies and software developers, he said, and his firm, for example, recently announced an integration deal with fellow digital pathology company Pramana that allows DICOM whole-slide images created by Pramana to be imported and viewed through PathPresenter's software.
As more large companies with broad installed instrument bases and existing enterprise imaging portfolios, such as Siemens Healthineers, enter the digital pathology market, they are looking to offer full solutions that include digital pathology, Myles said. Smaller pathology companies, meantime, want access to the global footprint and resources of large multinational corporations, which could encourage partnerships and strategic investments — and possibly acquisitions, like Fujifilm's purchase of Inspirata's digital pathology business at the end of 2022.
In just the past month, multiple digital pathology deals across large and small companies have been announced. Siemens recently inked a multiyear distribution agreement with Hamamatsu to distribute that firm's whole-slide scanners in the Americas and Europe, and last year it announced an agreement to combine its enterprise imaging offering with Proscia's Concentriq Dx platform for digital pathology.
Earlier this month, Agilent Technologies announced a similar distribution agreement with Proscia for its Concentriq Dx platform to review and analyze whole-slide images. Agilent also announced this month that it will make Hamamatsu's whole-slide scanners available for use with Agilent's digital pathology solutions.
Meantime, Danaher subsidiary Leica Biosystems announced this month that it has chosen Paige as its preferred provider of image management and viewing software, allowing customers using Leica's scanners to access Paige's full software suite. Roche has also inked multiple digital pathology deals in the past two years, including a partnership with Bristol Myers Squibb to develop digital pathology algorithms for clinical trials, an agreement to jointly develop an embedded image analysis workflow with Ibex Medical Analytics, and a development and distribution agreement with PathAI to provide access to PathAI's algorithms through Roche's cloud software.
Even tech giants outside the healthcare space are jumping into digital pathology — in January, Paige announced that Microsoft would invest in the firm to help it develop artificial intelligence-based diagnostics and partner with Paige on cloud-based delivery of Paige's platform.
And sequencing firms are also turning to digital pathology to enhance their tests for diagnosing cancers.
Cost of implementation
For many labs, cost is a major obstacle to implementing digital pathology, particularly as laboratory budgets continue to shrink. The significant upfront cost can be a deterrent for community health centers and smaller labs, and many scanner vendors don't offer flexible purchasing agreements, Hanna said. Outside of digital pathology, instrument and reagent rental models are popular, allowing labs to rent an instrument and pay per test or reagent, but as far as he knew, only one or two companies offer leasing price models for their digital pathology products.
The cost of implementation also goes beyond the cost of scanners and instruments. WUSTL's Brodsky noted that the large file size of whole-slide images "eventually results in large expenditures associated with data storage." Being able to reach the appropriate speed while panning and zooming through those large images can require upgrades to network equipment, application servers, and local workstations, while enabling an appropriate field of view comparable to the view from a microscope can necessitate upgrades to higher-resolution monitors.
The dedication of those IT resources for digital pathology implementation is often outside of the control of the pathology department, at least in a large hospital or health system, Brodsky said.
Paige's Moye also noted that the size of a lab matters when considering cost and implementation. In general, large for-profit reference laboratories that can invest in broad infrastructure have moved fastest in implementing digital pathology. Those labs are closely followed by academic medical centers that "want to be on the cutting edge of technology." In contrast, demand from community health systems has lagged, not due to lack of interest but because those hospitals "operate on a very thin margin" and aren't able to make the infrastructure investments or pay the upfront costs to implement digital pathology.
In Brodsky's view, the driving force that will make digital pathology a necessity for every lab is the benefit of image analysis algorithms in clinical workflows, such as algorithms that are able to help pathologists locate the areas with tumor tissue for diagnosis on a slide or separate out cases that will need a specialist consultation. Right now, there is a need for those algorithms to prove their worth and show that they're not negatively affecting the quality of a diagnosis. Once the accuracy and efficiency of those algorithms as diagnostic tools is proven, scanning slides digitally will be an essential step and will push labs to implement digital pathology solutions on a broader scale, he said.
But many labs may remain reluctant to spend money and resources on adopting digital pathology until those algorithms are clinically proven and receive FDA approval, he noted.
Reimbursement also plays a major role in the potential adoption of digital pathology, Brodsky added. Earlier this year, the American Medical Association created 13 new digital pathology Category III add-on codes to measure clinical staff work and service requirements associated with digitizing glass microscope slides for primary diagnosis. Those codes aren't yet tied to reimbursement but are instead intended to measure and audit the uptake of digital pathology solutions. In a couple of years, however, those codes may be priced and available for reimbursement.
Proscia's West noted that the addition of the CPT codes is laying the groundwork for reimbursement and acting as a tailwind for digital pathology adoption. The "pathways for reimbursement for this technology are being explored," and labs are "now under pressure to lay the infrastructure to take advantage of these technologies as they come down the pike." The creation of those codes gives laboratories more confidence that eventually, the technology will be utilized and reimbursed, West said.
Moye added that the CPT codes indicate that the government has recognized there is a value to digital pathology, but that it's "not sure what that value is just yet" and is looking to determine what coverage should be offered. However, he noted that while the Centers for Medicare and Medicaid Services may be inching closer to reimbursement, private payor coverage will continue to lag as insurers wait on CMS's verdict.
But the current lack of reimbursement is a problem for many labs considering going digital, Spectrum's Ahmad noted. Without reimbursement, there is a question of how his practice will be able to continue funding the implementation and use of digital pathology over the next several years.
US vs global adoption
The US has trailed behind many countries in its adoption of digital pathology, particularly those in Western Europe and the UK. According to WUSTL's Ritter, the main reason for this is the more fragmented nature of healthcare in the US. Other countries have united healthcare systems, like the UK's National Health Service, that are "large-scale operation[s]" that can quickly get all hospitals on board with new technologies.
But institutions in the US "can never find capital expenditure" for digital pathology infrastructure, he noted. Hospital systems also have "no history of investing big chunks of money into anatomic pathology," so the argument for digital pathology is difficult to get across to administrators and executives at many health systems.
PathAI's Beck agreed and noted that in the EU there has been more of a top-down push from governments to encourage digital pathology implementation. Government policy can serve as a lever to speed up adoption of new technologies, he said. As an example, earlier this month the UK NHS expanded access to Ibex Medical Analytics' artificial intelligence-based pathology products, making the firm's breast and prostate cancer tools available to 25 NHS trusts. And last year, two NHS Foundation Trusts in the UK signed a deal with Swedish firm Sectra allowing them to introduce Sectra's digital pathology platform to the local healthcare systems.
In 2020, the UK government also provided £50 million in funding to scale up digital pathology and imaging artificial intelligence.
However, Muhammad Aslam, a consultant pathologist and clinical director of North Wales managed clinical support services at Betsi Cadwaladr University Health Board, said that much of the push for digital pathology, at least in Wales, has come from the pathologists themselves. While there has been support from NHS Wales, the financial pressures on the health service are numerous. In addition, since individual hospitals can't procure instruments and equipment by themselves, the business case for digital pathology must be made to the NHS and approved before the country can go 100 percent digital, he said.
Aslam's health board has adopted digital pathology solutions for lymphoma panels, which it used for a proof of concept to prove the utility of digital pathology, and it uses Ibex's artificial intelligence-based Galen Prostate tool as a screening method for prostate cancer to ensure pathologists aren't missing anything. He said he hopes the business case to fully implement digital pathology will be approved by NHS Wales early next year, with national procurement beginning soon after.
He also noted that adoption even among the UK has varied by country as Northern Ireland is fully digital and Scotland is about 90 percent digital, while England and Wales are further behind.
Hopes and desires
While there has been significant progress made across the digital pathology landscape in the last few years, there remains tools that pathologists want. Brodsky said new features that he'd like to see include automated image data management using tiered storage, image annotation standards, convenient controllers for panning and zooming across slide images, and faster whole-slide scans.
PathGroup's Welch emphasized the need for mechanisms to selectively and permanently archive images of interest, such as complex tumors in unusual cases, while WUSTL's Ritter cited a need for built-in software that indicates whether a slide is adequate for review or needs to be rescanned. Aslam, whose institution is already using algorithms for prostate cancer, is keen to see artificial intelligence-based solutions for dermatopathology and gastrointestinal pathology that can at least be used as triage tools.
Despite the barriers to adoption, labs that have implemented digital pathology have seen a variety of positives. For Brodsky's lab, the benefits for clinical use in the short term are numerous: improved efficiency and turnaround time, the ability to avoid transporting glass slides, faster retrieval of slide images from previous cases, and the avoidance of recut costs for educational slide sets.
Although Paige's Moye said he believes the field isn't quite at its inflection point, the space is "moving in the right direction," with available tools providing a better experience for both patients and clinicians and improving patients' quality of life.
The technology is also clearly here to stay, Hanna said. It's "the next logical evolution of pathology."