NEW YORK (360Dx) – Roche's recent launch of a high-speed scanner for digital pathology marks a new step in the direction of more automated systems in a field with relatively slow adoption, according to the firm's executives and other industry sources.
With the development of the new Ventana DP 200 slide scanner, which is available to clinical labs outside the US and for research use in the US, Roche is "automating a piece of the pathology workflow in a robust way," Michael Rivers, vice president of digital pathology at Roche Tissue Diagnostics, said in an interview. Studies are in process with a view to getting clearance from the US Food and Drug Administration that would permit marketing the scanner for clinical use in the US in the future, he said.
Other parts of the in vitro diagnostic lab, such as clinical chemistry, are far more automated, but "digitization is on its way in pathology," said Rivers, who also sits on the board of the Digital Pathology Association, a nonprofit seeking to advancing the field of digital pathology.
He noted that Roche designed the new scanner to address some of the market challenges that have held back adoption not only of digital scanning but of the broader area of end-to-end digital pathology systems. These systems have begun to penetrate markets in which pathologists traditionally evaluate tissue biopsies on glass slides using microscopes.
Digital pathology companies see a lot of opportunity for adoption, Rivers said, adding that one of his firm's objectives is to "leverage its leadership and presence in the tissue lab and expand on it with these digital pathology tools."
Rivers noted that digital pathology technology is maturing, and manufacturers are making progress in resolving issues restraining adoption, and Roche's new scanner is an example of this.
The device has a scan speed of less than 49 seconds for a 15-by-15 millimeter scan area and applies a "unique" color profile to each image to ensure that the digital image closely matches colors observed under a microscope, Roche said. It easily integrates with image management services, include Roche's Virtuoso image and workflow management software, and is compatible with a Digital Imaging and Communications in Medicine (DICOM) Standard on which digital pathology firms have been collaborating.
Roche's "improved hardware" will help boost "the expansion of digital pathology," Eric Glassy, medical director of Affiliated Pathologists Medical Group, which specializes in anatomic, clinical, and digital pathology, said in an interview. "With the gradual move to using digital pathology for primary diagnosis, high-throughput scanners are key to successful implementation. Image quality is always paramount; pathologists expect the digital image to be equivalent to the glass slide."
In developing the new scanner, Roche "saw an opportunity for dramatic improvements" over both its own scanners and others on the market, Rivers said.
The firm's existing scanners — iScan Coreo for small- to medium-volume labs and iScan HT for high-volume labs — have been on the market for more than seven years, and they will continue to be the firm's go-to products in US clinical markets as the firm looks to achieve FDA clearance for its new scanner, Rivers said.
The DP 200 is suited for remote pathology, frozen sections, and high-value image analysis. Its design is not an iterative improvement, he said. The firm has focused on optimizing reliability and performance, reflected in image quality and scanning speed, he added.
As Roche executives look at digital pathology adoption, they see several drivers and barriers, Rivers said, adding, "One of the things that has held up the wide adoption of digital pathology within the clinical market is that technology maturity has been lacking, and scanning is just one part of that."
He noted that pathologists see a need for "excellent reproducible image quality" that ultimately gives them "something that is very similar to what they see using a microscope."
As labs consider digitizing their workflows, they are seeking solutions that enable them to handle precious clinical samples in a robust and reliable way, he said. "They need to be confident that they can get through workloads quickly and efficiently and without having to rescan and redo, and at worst recut and re-stain, tissue," he noted.
The benefit of enabling remote consultation is driving digital pathology adoption, he said. "There's an opportunity to support multiple hospitals in a system that may have a few labs," he noted. "Today, pathologists frequently travel between these labs, but remote pathology brings the image to the pathologist, and healthcare systems can easily and quickly get second opinions facilitating faster and more robust diagnoses."
Remote pathology will also prove valuable where pathologists are in short supply, he noted, adding that China, for example, has only 10 percent of the pathologists that it really needs.
Barriers to adoption
Among the barriers to adoption for digital pathology are system costs and complexity related to IT system integration, Rivers said.
Pathology images normally require one gigabyte of memory storage per image. Many labs that are looking to digitize pathology workflows will need to process between 100,000 and 200,000 slides per year, Rivers noted.
"This is how pathology really differs from radiology. People make that analogy all the time and there are a lot of similarities, but this piece is a big difference," said Rivers. "In pathology, the images are huge, and historically — as recently as a few years ago — costs had been prohibitive."
The move toward digital pathology began almost two decades ago and to date has achieved less than 15 percent market adoption, Abhi Ghopal, founder and CEO of OptraScan, a company that provides on-demand digital pathology services, said in an interview.
High costs of equipment acquisition have tended to deter small- and mid-size pathology groups from giving up microscopes, he said. More than 80 percent of pathologists in the US and 95 percent in emerging economies are still using optical microscopes, while "awaiting affordable, high-performance, software-driven solutions" that conform with lab workflows for clinical diagnostics, Ghopal said.
To address concerns of labs anxious about the cost of digitization, Roche's new scanner may be a good place to begin, Rivers said. Customers can scale up by adding several systems in parallel if needed, he noted. And Roche is introducing a pricing model that places less of an emphasis on capital expenditure upfront and more on payments received on a recurring basis, by employing an approach similar to that of software as a service.
An increasingly competitive market
Roche is among several firms vying for a piece of the digital pathology market. Others considered among the market leaders are Danaher's Leica Biosystems, Philips, 3DHistech, and Hamamatsu.
In December, Lecia Biosystems said it was among several companies to ink a deal with the UK government for the development of a digital pathology program leveraging artificial intelligence through the expanded use of the firm's Aperio Digital Pathology Solutions.
And in April 2017, through a de novo classification, Philips received FDA clearance to market the IntelliSite Pathology Solution for primary diagnostic use in the US, making it the first digital pathology solution cleared for primary diagnostic use in the US.
Meanwhile, OptraScan offers a subscription-based service for research use to pathologists who want access to digital pathology technology but don't want to pay the high instrument costs.
At a starting price of $1,000 per month, with additional fees for certain services, OptraScan provides digital pathology services based on a system that enables the acquisition of whole-slide images as well as viewing, sharing, on-demand analysis, and management of digital slides. In February, the firm said that the Indian government approved its whole-slide digital scanner, following a validation study.
In addition, the market has seen new entrants and existing companies developing diagnostic systems that integrate artificial intelligence and machine learning to complement pathologists' efforts to diagnose diseases.
Cernostics is developing an esophageal cancer classifier that it said can quantify multiple antigens per biopsy section and conduct objective reproducible extraction of quantitative biomarker, morphology, and spatial data by image analysis software.
In the academic setting, researchers at Emory and Northwestern universities have developed artificial intelligence software that analyzes data from tissue biopsies to predict the survival of patients diagnosed with glioma. They believe that their approach could help pathologists remove subjectivity and variability in doing glioma diagnostics.
"Machine-learning and deep-learning tools are going to carry digital pathology forward into the clinical space," Rivers said. The advent of machine learning is leading to new algorithms for image analysis and the discovery of biomarkers for pathology systems, he noted. Roche has "a significant team" in place and continues to invest in developing new techniques for use of artificial intelligence with machine learning, including deep learning, he added.
"We are using deep-learning tools right now to develop tools in the research space, and our focus is on bringing these solutions to the clinic," he said. "We have a lot of opportunities internally within Roche to experiment and develop some of these research tools."
Few of those tools have moved into the clinic to date, he noted, but Roche's breast panel algorithms are among those that have, and many more are on the horizon, creating an "opportunity to deliver new insights to the pathologist and generate real impact and medical value."
The cleared markers for breast cancer, running on its Benchmark system, are HER2, ER, PR, PI or Ki-67, and P53, he noted. "Those are five markers that are commonly used in breast cancer diagnosis," Rivers said. "The clearance involves the entire system, starting with the Ventana antibody assay running on a Benchmark staining platform." Slides are scanned on the Coreo scanner and transferred to Virtuoso image management software.
The system is cleared as a Class II 510k device and as support for a pathologist doing a diagnosis, Rivers said.
In looking at the overall digital pathology workflow, Roche's new scanner is just one, albeit important, component affecting function and adoption. "It's a key component for image acquisition, but it also integrates with enterprise software that supports the entire anatomic pathology workflow, including case management and assignment, image review, image analysis with different algorithms, and ultimately reporting either through our software or through the lab information system," Rivers said.
Many digital pathology firms are committed to being interoperable, Rivers said, and its new scanner produces a DICOM image, already a standard in radiology but one that digital pathology companies, including Philips and Leica Biosystems along with Roche, have been collaborating on. Roche's new scanner is its first to offer that format.
In terms of digital pathology product providers, Leica is most similar to Roche in the breadth of its digital pathology portfolio, and the firm has a strong presence in particular in pathology research, Rivers said.
Philips, he noted, is also making a big push into digital pathology, and several "niche players" develop and sell either hardware or software digital pathology products.
Through the Digital Pathology Association, he said, Roche works closely on standards development with other industry players. "We all recognize the tremendous potential for digital pathology, so the association is a mechanism to drive common interests, general awareness, and adoption," he said. He noted that among its tasks, the association's members work with the FDA to provide guidance and advice and to help it better understand some of the new digital pathology tools coming onto the market.
Ghopal said that some digital pathology companies "are trying to woo large hospitals" at the "tip of the pyramidal marketplace without giving much attention to [more than 70 percent] of the labs at the middle and bottom of that pyramid."
It's important to note that a microscopes costs less than $5,000, while a digital pathology solution can cost $300,000 and above, he noted. The need of the day, he said, is an "open, scalable, flexible, and affordable solution-as-a-service" that could transform "century-old optical microscopy to digital pathology for diagnostics, teleconsultation, research, and education."
In the coming years, it will be important to use affordable pathology-as-a-service systems to drive digital pathology adoption to "the masses," he said.
Rivers said that in the context of its new scanner launch, Roche aims to bring efficiency and medical value that justifies price, and price really depends on variables such as a customer's IT infrastructure and how many slides they are trying to archive and store.
In the future, digital pathology systems will also include more assays in which multiple biomarkers evaluate one slice of tissue and provide quantitation for the pathologist. "I think that's where digital pathology can make an incredible difference," Rivers said. "At some point, pathologists become limited in what they can evaluate on a slide, and with digital pathology and image analysis tools, we can do quantitation and take advantage of machine learning that will enable pathologists."