March 2012
Feature Story

Anne Paxton

The 2012 ski season has brought stunning survival stories. At least a few snowboarders and skiers have managed to live through off-trail avalanches by deploying a backpack airbag that allowed them to tumble down a slope and come to a stop, almost magically, on the surface of the snow rather than beneath it. All due to a paradoxical phenomenon called granular convection, which causes objects with a larger surface area to rise to the top of a shifting mass, even when they’re heavier.

Visualizing granular convection might be a good way to imagine the shape in which anatomic pathology information technology will emerge as it is buffeted by an exploding digital universe. Figuratively speaking, how will the shifting trends shake out amid the avalanche? What will rise, perhaps surprisingly, to the top? Experts in the field, gazing at the lightning-fast advances in computing speed, scope, and connectivity, suggest the answer might be anything but obvious. But it could well include systems that integrate a wide range of diagnostic disciplines—even some from outside pathology—under one IT roof.

“Things are changing rapidly in diagnostics,” says Curt Johnson, chief operating officer of Orchard Software, which makes the CP/AP system Orchard Pathology. “We have molecular, clinical, and AP modalities all coming together, and as more digital pathology comes in, it will change the way information systems deal with pathology.”

Orchard Pathology, designed for the complexities of integrated diagnostics, he says, is available as a standalone AP system or as a complete system in which clinical, AP, and molecular information share the same database. The company’s approach, Johnson explains, has been to develop systems that would benefit the pathologist but still capture data in a way that it can be integrated with other systems, and continue to be user-friendly for all involved.

“In the past, pathology systems were all about the report, which was done in a very sophisticated word document information system. And that’s still critically important. But today we have a need for discrete data—codified information sorted into fields that are minable in other systems, rather than big blobs of text.” This need for discrete data is changing the way pathologists, histologists, and cytologists are interacting with the system, Johnson says. Another cause of change, he believes, is the integration of IT in health care, driven partly by regulation, partly by high-tech incentives.

“I learned a decade ago that discrete data was going to be king,” says Michael D. Glant, MD, Orchard medical director. During his work designing AP systems for labs that did international clinical trials, and for a reference lab that did 200,000 Pap tests annually, “we had all kinds of challenges with IT. And once you start having to transfer data into clinical trials, you know what you have to have. You know that unless you have discrete data, you’re not going to pass that data along in a very robust way, whether to the state board of health or to clinicians getting your report. In the commercial and even academic world, you didn’t tackle that problem in most instances.”

Orchard is trying this year to expand the abilities of imaging to come to the forefront. “Even though we think it will be a while before we get whole-slide imaging on pathologists’ desktops, a lot of pathologists already have basic still imaging and use it for surveillance and quality control within labs. The whole workflow is going to start changing, both for using digital data and for tracking the workflow through the lab,” says Dr. Glant. The company is developing a full imaging suite to add to Orchard Pathology, as a way for pathologists to become information managers without having to spend $300,000 on a whole-slide system. “It’s a sophisticated IT system that will let the pathologist at a community hospital have all the tools that you have at a medical center.”

Dr. Glant sees little use for a standalone molecular system. “From where we sit, the future lab is an integrated diagnostic lab, so we aren’t going to build a molecular system that isn’t integrated into AP and clinical; it’s all one system for us.” There is one problem with molecular, he adds: You don’t know what’s going to come out tomorrow. “You have to have a very flexible system with rules that can be easily managed and modified to take a molecular test that now perhaps has two or three steps added.”

One of his big complaints in using legacy and most established LISs, he says, is “once you want to make a change, it takes an arm and a leg to get it done.” That is why Orchard has focused, despite having a structured set of data and standardized files, on being flexible in the ability to add workflow, change how data are presented and how they’re channeled, and still make everyone use the same vocabulary. The advancing use of molecular testing, Johnson adds, will increase the role of the pathologist as an educator to primary physicians on what the reports mean. “You’re going to see a need for more video capture on reports, and even something like YouTube reports with video consults to the primary specialist from the pathologist in the next three to five years.”

Over the same period, he expects that the most advanced pathology labs will be doing more image analysis processes such as spectral analysis. For example, “I’ve recently seen some vendors demonstrate very sophisticated repeat testing of immunohistochemical stains where they take tissues and stain them with multiple specific markers to almost create fingerprints.” But the bulk of pathologists will wait for the high-cost research and instrumentation to come down into their price range so they can use them in their everyday work. “There will probably be opportunities for those pathologists to at least get involved in digital processes, so that’s going to be the next revolution, because you can now get consumer cameras that are better than the expensive scientific cameras of 10 years ago, and scanners are starting to be introduced that are $30,000 rather than $100,000.” But for most pathologists, in his opinion, there won’t be real growth of digital imaging for a decade or more.

Many pathologists are still out of tune with informatics, Dr. Glant notes. “That’s a big challenge, so one change we see in the future is that as pathologists become more educators and coordinators, they will also have to educate themselves to get more involved in informatics, and we hope as a company to help that education process along.”

“If you look at everything that’s happened in the clinical lab around robotics, automation, workflow efficiencies, specimen management, bar-coding, and discrete data, AP labs in general have lagged behind, as they’ve traditionally been more descriptive and unstructured,” says Pawan Singh, senior product manager, Sunquest Information Systems, marketer of Sunquest PowerPath and Sunquest CoPathPlus. “Our solutions have taken many of the best practices—for example, Sunquest specimen management and tracking—that we’ve had in the clinical laboratory for many years and enabled them in the AP LIS space.”

Sunquest’s focus is on the integration of all modalities—clinical, anatomic, and molecular—within the four walls of the lab, but also extending beyond those four walls, Singh says. “Our solutions include the outreach communities that are bringing samples into the lab from outreach clinics, physician offices, as well as for real-time analytics and performance indicators for operations personnel in the lab. And there are patient safety considerations at the point of care in specimen extraction all the way through laboratory accessioning and reporting out to health care agencies. We think of the laboratory as the hub with all these different spokes, and we manage the hub as well as the spokes.”

How will digital fit into the picture? “I would equate digital pathology with another instrument in the lab workflow,” Singh says. “AP workflow is fairly manual with a lot of steps.… And digital pathology, specifically whole-slide, imaging is adding a step in the process for making images out of slides and then sending to the pathologist. So the workflow doesn’t change that much, but by going digital you open the door for new opportunities. You can now have geographic separation between the pathologist and the laboratory, and you can do different kinds of collaboration.”

Sunquest has merged molecular diagnostics into the existing AP and CP LIS, Singh says. “In most hospitals, AP and CP departments have come up as independent practices and now the lines are getting blurred. The last thing we want to do is build a separate molecular system. Most IT departments and clinical departments don’t want to create another information silo.”

Singh worked in radiology for more than 10 years where he says there are many stark differences as well as similarities. “If you go back 30 years ago, MRI procedures did not exist. The radiology space started with X-rays, moved onto CT scans, then MRIs. Similarly in pathology, different modalities in the specialty have emerged independently over time and continue to evolve. But if you look at radiology today, they are able to inter-operate in a much more consistent manner than pathology. They can exchange information in a much more seamless way, and the reason is the adoption of standards like DICOM. Even within a single vendor in radiology, there could be very different IT underpinnings and different databases for the different modalities. But once you get to a common standard, then these underlying system differences don’t matter.”

Radiologists talk about overlaying data from different sources to digitally draw conclusions that are diagnostically meaningful. Could something like that happen in pathology? Once information is digitized, Singh points out, computer-aided computations can be applied. “The field of pathology involves a lot of pattern matching, knowing the difference between normal and abnormal cells, between cancerous and non-cancerous. Computers can do pattern matching really well, and especially for the more straightforward and routine procedures, you could mine data and derive conclusions in an automated fashion.” This could free up the pathologists to do much more advanced diagnosis and provide more consultative care. “That’s our mission,” Singh says, “to enable the pathologists to be much more successful in their practices.”

He is unsure about the concept of “overlaying” and its applicability in pathology. “Even in the radiology space it is something that is quite tricky. But you can imagine if there’s a mammogram and you have a digitized tumor image, you can visualize those two pieces of information together. And we already do some of that today for prostate or gastrointestinal cases where one of the important pieces of information is markings to indicate the anatomic region from which a tumor was extracted. Overlaying of stains is another possibility.”

No matter how dazzling the advances in IT, Singh says, it’s important to remember that the fundamental practices in pathology are not changing. “We’re still dealing with blocks and slides, and the idea of safe and effective specimen management is not going away. It’s actually more important to slide quality because when you start digitizing these slides, even the smallest artifact gets amplified and makes for a very difficult diagnosis.”

“There’s definitely starting to be a blurring of the lines between what has classically been done in AP and what is now happening in molecular,” says John David Nolen, MD, PhD, MSPH, director of laboratory strategy for Cerner Corp., which markets CoPathPlus and Millennium PathNet for the AP market. “Specimens are getting smaller, volumes going up, the tools are getting much more precise, and everybody’s getting coordinated and talking to each other. You’re seeing much more required interaction—synergy, if you will—between tests that happen in the AP space in pathology and tests in the molecular space.”

Pathologists are having to confront this change. “You have to be able to say, what does the molecular show, what does the clinical show, what does the cytogenetics show? Because the clinician is saying, ‘I want it all in a single report.’”

That’s one level of integration. The other is behind the scenes where the workflow is happening, Dr. Nolen explains. “You’ve got one specimen coming into the lab but it has to make several stops. You have to potentially split it, aliquot it, go from AP over to molecular over to the general lab, and all that specimen movement has to be tracked.” That workflow is making the discipline boundaries look ever more artificial, he believes. “I think we’ll look back and say, why do we have any of these distinctions between AP, CP, and molecular? It will all be together as one field of diagnostics.”

He suspects radiology will be part of the package. There was a time when diagnostic tools in pathology may have included inserting a finger through a piece of liver tissue to decide whether it was fibrotic, Dr. Nolen says. “Then we started cutting it up, looking at different tissue under a microscope, adding special stains to break the level of detail out. And now we’re at the molecular level.” Meanwhile, the size of that tissue keeps shrinking. “Our friends in radiology are getting really good at getting small biopsies from very hard-to-reach places, and they’ll give you this small core of tissue and say, here’s what I got you, you need to tell me the world.”

At the same time, radiology tools are becoming much more powerful. “It’s no longer just the broad brush, yes-you-have-a-broken-bone, but I can look at a nuclear stain and see what kind of odd molecular signature you’ve got because of some radioactive compound I gave you that I can trace. So the fields of radiology and pathology are converging at that lower level.” How much could different types of pathology data interact? Dr. Nolen points out that the world of digital pathology is expanding the possibilities.

“The first level was, ‘Can we get the camera controlled in a precise manner, scan an entire slide, and show it on the screen?’ The next level was ‘Can we look at the image of a slide and do calculations for the pathologist?’ Now you can have a more robust algorithm, to score a slide as a one-plus, a two-plus, or a three-plus, maybe an edge detection algorithm. But beyond that, you have people almost doing flow cytometry on a slide, where they are putting multiple markers on it and measuring various parameters.”

As for overlaying different data types, that might also be in the cards. A research team at Sunnybrook Hospital in Toronto has created giant digital slides of entire specimens, a full cross section of prostate tissue and a full slide of all the breast tissue. “They are building a massive tumor bank of specimens. They take many, many layers and they’re able to overlay the digital pathology with the radiology at the grossing bench and radiology before surgery. They’ve figured out a way to maximize the data coming in, because when pathologists get a surgical specimen, it’s like a blob; you’ve lost that architecture by its being on a slide. But on the radiology table you can really see where this thing was, and if you have to go back and get the precise therapeutic location, you can.”

Pathology has a significant volume of data, information, and knowledge built on morphology, Dr. Nolen notes. “You can’t just get rid of that overnight. I view the real convergence between radiology and pathology as starting to take place in five to 10 years, and it will be happening at the AP level because both fields are highly driven by morphology.” But he doesn’t underestimate the difficulties that might thwart that prediction. “Remember, people for a long time were wrongly forecasting with digital pathology that there would be no more glass. And here we are dealing with FDA Class III devices and people are sort of into digital pathology. But everything’s still glass.”

Dr. Nolen is involved with DICOM and HL7 as part of the effort to write common standards. “The challenge we’ve got with these interfaces is that devices in the AP lab are getting smarter every day and they’re starting to ask more sophisticated questions than just ‘give me an order and I’ll give you results.’ How was this specimen treated, how thin were the sections cut, how long did it take in the stainer bath? I need to know so I can do advanced analysis on it. Our job at the LIS is to be that single source of truth. We have the goal of trying to make all of the data portable between health care centers, so one of our pushes is to make some structured data concepts in AP that you can truly move between organizations, to convey the entire digital version of a record.”

He is not confident about the prospects of pathologists becoming liberated from paper, however. “All of our Cerner solutions right now fully embrace a paperless environment, but what I think holds people back is not so much the solution; it’s the workflow, or maybe the end user. Because people still use the word ‘report,’ when they’re talking about something that may be digital. The human brain is still trying to put it into an 8˝-by-11 inch piece of paper.” He considers PDFs to be not much of a step up from paper. “The beauty is that you can lock down formatting, arranging the report the way you want it to look, and your comment at the top of the page is not going to be at the bottom of the page. But you may have also locked down your content. If the lab wants a PDF, you’ve lost full integration at that point, because the electronic health record may want discrete values.”

Wally Soufi, CEO of NovoPath, points to the growth in connectivity and data exchange between the pathology laboratory and the various parts of the health care spectrum as the trend that will dominate patient care in years to come. Many NovoPath clients offer clinical and molecular tests today, he says, but it’s the data exchange that brings the test results together, “not necessarily the merging of these disciplines.”

“Ultimately, you want to centralize patient care information from all providers, not just diagnosticians,” Soufi says. “EMR systems will soon do that, and NovoPath, the only AP system to meet ONC certification as an EMR module, is ready.”

Soufi sees pathologists as specialists who do things in unique ways to suit their evolving service needs. NovoPath has focused on serving the AP IT marketplace since 1999, offering solutions for reference and hospital-based laboratories to manage workflow, QA, connectivity, billing, and reporting—in fact, all aspects of the AP lab, Soufi says.

“Pathologists need to use all the testing methodologies and protocols that are available. I don’t really see AP becoming like clinical testing such as in chemistry, where you’re just running a machine and producing a result. There are attempts today to automate or have computer-aided diagnosis in looking at slides, but they are a ways off, I think. Molecular tests may share similarities with clinical ones, but they produce results that need interpretation. They still need human evaluation. Until there is a computer or system out there that you can just say, ‘Here’s a slide, give me a result,’ you’re going to need that extra piece, that human workflow.”

NovoPath has dabbled with the idea of computer-aided diagnosis, which has been around for some time for screening of gynecologic pathology slides. But Soufi thinks more elaborate use of computer-aided diagnosis is farther down the road. For now, NovoPath is specializing in the integration of multiple disciplines and multiple specialties of pathology into one report to provide a diagnosis to a clinician. For example, in a hematopathology scenario, he says, where a person might be diagnosed with a type of leukemia, “you’d want to incorporate cytogenetic, FISH testing, flow cy-tom-etry, bone marrow aspirate, bone marrow biopsy, and clinical chemistry results into one integrated diagnostic report with discrete searchable data. The result is a comprehensive diagnosis that enables the clinician to provide the best patient care possible, and that’s a powerful tool our clients have at their disposal today.”

Stephen G. Ruby MD, MBA, found a range of possibilities three years ago when he was set to open an independent pathology practice that became 4Path, in Justice, Ill. He had a blank slate on which to devise a cost-effective IT solution, and he became an early client of LigoLab.

“You wouldn’t know it looking at the laboratory, but we’re functionally pretty paperless,” he says, noting that the practice has handled thousands of specimens from the outset, with zero transcription costs, using Dragon Naturally Speaking voice-recognition software. One obstacle has been that physician clients of his outpatient-geared practice still prefer paper. “So we have multi-part requisitions that come with each specimen, but once it hits the lab, we accession it into the system, and from there we don’t generate additional paper until it comes down to printing the report.” Even the requisitions and photocopies of insurance cards are scanned, bar-coded, digitized, and attached to the electronic reports available for users.

AP’s origins as essentially words rather than numbers have driven it to become more image-based, “and that took a little more horsepower than the old clinical lab systems,” Dr. Ruby points out. Of course, he adds, those old LISs from the 1970s and 1980s probably had less memory and less horsepower than an iPhone application does today.

But as molecular and flow cytometry data continue to merge with AP, a strange thing is happening, in his view. “We constantly say AP is becoming more like clinical lab IT, but actually clinical is becoming a little more like anatomic. With the graphical items and imaging that can occur, we can now do things in the clinical area that we couldn’t necessarily do before. So the issue is that AP and clinical are converging onto a common pathway of both numeric and morphologic data combined into a single report.”

Starting an IT system from scratch has its advantages, Dr. Ruby says. “Some of the old LISs are so large and unwieldy with old database structures and old file structures that it’s difficult to import the data into new systems. And the old legacy systems had very limited datasets because of storage capabilities.” His system has to interface with a few physician practices’ in-office EMRs, though no hospital EMRs as of yet, and he agrees that AP reports often don’t display very well. “A lot of times, the EMR designers have not geared their file structures to image displays in AP systems—not out of maliciousness, just short-sightedness, because they may not have realized this kind of data was out there.”

Digital pathology, such as Aperio is offering already in Europe, may be a ways off in the United States. “But one thing I did from the first report on is have a CCD camera on our microscopes. Every report we generate has an image of that patient slide on the report. So in many respects it’s a scaled-down version of a full-blown image analysis system.” LigoLab, he says, has also done interfacing to be able to bring together its LIS and digital pathology. “And I’m sure a lot of companies will be looking to try to integrate those as that kind of imaging becomes more prevalent. An imaging system integrated with an AP system is in many ways not a lot different in concept from an old chemistry or hematology analyzer interfacing with a clinical LIS,” Dr. Ruby says. “The point is, it’s a machine that is generating data or images and integrating those into an LIS.”

Basically, diagnostic pathology and diagnostic radiology are very similar, he believes. “We’re both image-based. To over-simplify, radiologists generate images by shooting electron beams through people, and we do it by taking pieces out of people and shooting light through slides.” But radiology has taken it further, in a way that many pathologists would like to see their specialty emulate. “Right now radiologists are able to overlay CTs with PETs with MRIs, all these different modalities to look at a lung lesion, for example, in comparison with each other. But radiology images are simpler and more often predominantly black and white than are pathology images, so pathology imaging hasn’t gotten to that point yet.”

“But I think people will be doing this in the future, say, by analyzing slides utilizing different antibodies to analyze different components within the cell, overlaying different things you couldn’t necessarily put on an individual slide. The tools aren’t there for us to be able to easily take a keratin stain and overlap it with an ER and a PR and a proliferation marker, and so on. But it’s one area, when we talk about imaging, that could be of enormous benefit to people.” To coin a term, he says, it might be called a “bio-IT” system or solution.

In the 1930s, pathology started off with handwritten notes, then progressed through typing and word processing, jumping from ASCII characters to greater manipulation of images with Apple, Dr. Ruby points out. “Now we’re on much more powerful computers, but fundamentally that report hasn’t changed over the years. People need to say, how can my report better serve my clients, the physicians? And that could be by integrating different pieces of information, by more quantitative analysis of morphologic images, by embedding differential diagnoses, or by amalgamating data with radiologic and other data and images.” The tools are available now, he says, but their full capacity isn’t being used. “We have a lot of potential for doing new and great things with them.”