William Check, PhD
The future fascinates humans. People of all stripes have formulated aphorisms about this elusive aspect of life. Albert Einstein said, “I never think of the future—it comes soon enough.” British novelist John Galsworthy felt differently. “If you don’t think about the future,” he said, “you cannot have one.” Perhaps the most widely known dictum on this topic is that of physicist Niels Bohr: “Prediction is very difficult, especially about the future.”
Despite the hazards of predicting the future, few of us can resist doing so. So it was not surprising that CAP TODAY’s request of several experts in molecular diagnostics to foretell where this field will be in the next five to 10 years was met with enthusiasm and thoughtful speculation. Some of the ideas are extensions of current trends, but that’s often how change comes about. As cyberpunk novelist William Gibson said, “The future is here. It’s just not widely distributed yet.” And all of the experts’ ideas took greater cost for granted, a trend that science fiction author John Sladek summed up: “The future, according to some scientists, will be much like the present only more expensive.”
Frederick L. Kiechle, MD, PhD, medical director of clinical pathology at Memorial Healthcare System, Hollywood, Fla., cites a trio that will re-shape the pathologist’s world. First is the spread of point-of-care applications in molecular diagnostics. “We need to miniaturize all the steps of PCR in a nanotechnology microfluidics setting,” he says. “I think there would be no end to the applications for such a device.”
Second is personalized medicine. “Molecular puts the ‘person’ in personalized,” Dr. Kiechle says. “I think it will grow faster than anyone can keep up with. Pathologists will have to learn about this or they will be left in the dust.” Nucleic acid sequencing is No. 3. “It has been slow in coming,” Dr. Kiechle says, “but the number of applications of sequencing will continue to increase.”
Lance R. Peterson, MD, says automation and molecular technology will transform diagnostic microbiology before the decade is out. “By 2020 the micro lab as we know it won’t exist,” says Dr. Peterson, who is director of microbiology and infectious diseases research in the clinical microbiology laboratory of the Department of Pathology and Laboratory Medicine, NorthShore University HealthSystem, and clinical professor, University of Chicago. “Clever robotics” are already moving into the clinical setting, he says, and adds, “Mass spectrometry will probably revolutionize bacterial identification.”
Like Dr. Peterson, Daniel H. Farkas, PhD, HCLD, vice president of clinical diagnostics and laboratory director at the Sequenom Center for Molecular Medicine, Grand Rapids, Mich., sees mass spectrometry as a rising star. “It will have a good run,” Dr. Farkas predicts. “As target numbers become larger, from a handful to hundreds, mass spec will shine in oncology and genetics, as well as microbiology.”
At the same time, he says, “Next-generation sequencing has a chance to blow everything away. We need economic and regulatory models, but it could change the way lab medicine is practiced.” Pathology could move from “wet work” to a primarily information-based discipline, somewhat akin to radiology.
Many of the key elements of future molecular diagnostic technology have already begun to enter practice, says John Bishop, CEO of Cepheid, who sees the transition from the present to the future as a continuum. “The single biggest reason why molecular grew up as a specialty is complexity of technology,” Bishop says. “Complexity is still there, but it has been made transparent to users.” One result: a decrease in the need for specially trained technologists. “Now we can basically train anyone in the lab to run molecular tests,” he says.
“Way in the future,” Bishop says, the simplicity of contemporary molecular tests will allow them to be placed in outpatient and walk-in care sites like CVS or WalMart. “Patients will be able to pick up their treatment at the same place they get a test result,” Bishop says. “This will be important in infectious disease and even more so in oncology.” In this context, he sees the key role for pathologists continuing to be interpretation and, with clinicians, making use of test information to formulate good patient care.
“Cost is a big issue,” Bishop admits. “However, our whole idea of cost is appropriately changing. We need to look at cost on a holistic basis in terms of managing overall patient care.”
Predictive tests that help to select therapy will be a rapidly growing segment of molecular diagnostics, says Brian Buxton, principal of Easton Associates LLC, a management consulting firm that works on strategy and business development for diagnostics and pharma companies. Noting “all the action in this space,” Buxton points out that predictive testing volume has increased by 25 percent in academic cancer centers and calls predictive or companion testing “one of the most exciting developments in our lifetime.” Here, too, reimbursement issues arise. Many oncologists are ordering tests the FDA hasn’t yet approved. Current method-based reimbursement rates don’t begin to cover the value these tests provide in selecting the right therapy, especially a targeted therapy for cancer, Buxton says. How will third-party payers reimburse such tests?
“Where do you draw the line between funding the science and charging the payers for value to the patient and total health costs?” Buxton asks. “In some cases, we are using the standard insurance system to fund new science, but in others, the current system is severely underpaying for the value these tests deliver.”
Dr. Kiechle says three steps need to be transformed for miniaturization of PCR to become a reality: purifying DNA, amplifying it, and measuring the presence or number of amplified copies. “We do MRSA admission screening here,” Dr. Kiechle says of Memorial Healthcare. “It takes a couple of hours once we get the sample. If we had POC testing in the nursing and admitting areas, they could take a nasal swab and get the answer immediately. Clinicians would quickly have information to treat or not, and to isolate or not.” The machine that comes closest to simplifying all three steps, in his view, is the Cepheid GeneXpert. “However,” he adds, “it is still bigger than a breadbox, so we have a ways to go.” Moreover, each GeneXpert cartridge is about $50, so using it can become expensive. “Lots of startups are working on this problem,” Dr. Kiechle says, pointing to a paper he had just seen on single-molecule denaturation mapping of DNA in nanofluidic channels (Reisner W, et al. Proc Natl Acad Sci USA. 2010; 107:13294–13299).
Personalized medicine, Dr. Kiechle notes, consists of tests that ask what mutations a person’s DNA contains that might affect drug metabolism or efficacy, so doses can be adjusted or effective drugs chosen. Warfarin is the chief example in the pharmacogenomics arena. A cytochrome assay that measures the activity of a P450 cytochrome, which converts tamoxifen to an active form, is recommended before tamoxifen treatment is initiated. In Dr. Kiechle’s opinion, there are three obstacles to wider adoption of warfarin pharmacogenomics: difficulty in demonstrating improved clinical outcomes; having only three FDA-recognized mutations (“not nearly adequate, especially for ethnic diversity”); and clinicians feeling they don’t have sufficient guidelines. Dr. Kiechle knows a pathologist who was prescribed Coumadin for atrial fibrillation, but the clinician treating him didn’t want to do the three FDA-approved molecular assays for mutations in Coumadin-metabolizing enzymes in the absence of guidelines that explain how to use the results. The pathologist insisted, and it turned out he has a VKOR mutation for which a 20 percent dose reduction is indicated.
“There is a lot of confusion about these tests,” Dr. Kiechle says. “It is not as hard to do as clinicians think. It is just that correlations and data accumulate faster than anyone can keep track of.”
Also in the personalized medicine realm are molecular tests for signal transduction molecules such as KRAS, BRAF, and EGFR to guide treatment of solid tumors. Pathologists would be well advised to become expert about signal transduction and therapies to disrupt or up-regulate these pathways, Dr. Kiechle says. “Our field is not just about red and blue slides but molecules. It always has been about molecules, but until FISH we were not looking at specific genes,” he says.
For the third innovation, DNA sequencing, Dr. Kiechle points to next-generation methods such as the 454 instrument Roche acquired. “These are very fast and focused on genome sequencing,” he says. “But they can be adapted to sequencing individual genes.” He cites the HLA system as an example of a group of genes in which mutations can lead to medical issues and that can be analyzed by new sequencing methods. Pyrosequencing, for example, can be used to look at smaller regions of DNA when a specific codon or region might be implicated, such as KRAS 12/13.
“I see that technology as being the answer to genes like KRAS where not all information on relevant mutations is available right now,” Dr. Kiechle says. “Using an assay that looks for a limited number of specific point mutations could miss one that is unique to that patient.” Dr. Kiechle notes that a Qiagen pyrosequencing instrument costing about $60,000 can do 24 samples in about 1.5 hours after extraction and amplification.
Under the term “clever robotics,” Dr. Peterson includes an automated system produced by the privately held Dutch company Kiestra that is being beta-tested in a Cambridge University hospital in the United Kingdom. This robot bar codes samples; stores streaked plates in an incubator in a defined place from which it can retrieve them; reads negative culture plates, including chromogenic plates, without supervision; and images plates at 24 hours with a high-resolution camera. Technologists view electronic images, reducing hands-on time. To some extent, this robot combines tasks that other robots already do. However, having everything in one system allows for coordination and for easily adding new functions. Kiestra says its robot reduces the need for workers by 50 percent to 70 percent, Dr. Peterson reports. Data are now emerging from the Cambridge experience. “They are making targets,” Dr. Peterson says, “but a month or two behind schedule.”
The advantages of mass spectrometry, according to Dr. Peterson, are that the turnaround time is short (two hours or less); it requires virtually no reagents; and it works on a pure colony with no DNA extraction and very little specimen preparation. The cost averages less than $2.50 per specimen. Demonstrations at the 2010 European Congress of Clinical Microbiology and Infectious Diseases showed an instrument that is small—tabletop size. But the instrument is expensive, Dr. Peterson says. Presentations at the European Congress reported that the method identifies organisms to the species level. “We’re going to try to seriously look at it in our lab,” Dr. Peterson says. That mass spec doesn’t work on viruses does not present a problem, he says, explaining, “Bacterial pathogens are overwhelming everything else.” For viruses, amplification will continue to be used.
Radical alterations in an existing fundamental technology, real-time PCR, will also alter the landscape of molecular diagnostics. “Molecular is no longer a special operation,” Bishop says. “It is clearly moving into the core lab and disseminating throughout the institution into the ICU or the ED or stat lab. It will be up to each hospital to determine where they want to employ it.” Bishop foresees molecular testing moving outside the hospital setting into outpatient surgicenters and physicians’ offices, particularly for infectious diseases and oncology.
Cepheid is following this strategy with its smallest GeneXpert model, GeneXpert-1, which Bishop says is no larger than a small briefcase. “First you need to move your technology to CLIA-waived status,” he says. “We are discussing with FDA what is required to get that.” Once GeneXpert-1 has CLIA-waived status, Cepheid plans to use it to determine Staphylococcus aureus colonization status of patients in outpatient surgical centers. “With the instrument there, you simply swab the nasal area and get a result in an hour,” Bishop says. Patients colonized with a pathogenic or virulent strain of S. aureus (not just MRSA) are at nine times higher risk for infection. “We are aiming to have the instrument in clinicals in 2011 and to have a CLIA-waived product by the end of the year,” Bishop says.
He tempers his forecast of molecular moving outside the hospital with a note of caution. “As this kind of shift occurs,” he says, “one can exaggerate and say that molecular will move entirely out of the molecular core lab. Certainly that will not be the case. There will be a continuum of tests, some done in the core lab and some done on a disseminated basis.”
It would also be a mistake to say that black-box-type instruments are going to make pathologists redundant. That’s because advanced instruments and human interpretation are complementary. “The knowledge base is ongoing,” Bishop says. “The pathologist’s job is to stay on top of that. As we achieve better understanding within the genome and better understanding of the physiological interactions of various markers, pathologists will have to keep up with these developments more than ever before so they can give accurate interpretations.”
Though some costs will rise, Bishop describes as “one of the biggest mistakes we’ve made” that labs have historically been looked at only as cost centers. “This has caused a silo mentality,” he says. “What we need to be doing, and I believe what will happen with initiatives on comparative efficacy, is to understand which new technologies are cost-effective.” New technology is more costly, yes, but if it provides clinically useful and more timely information, it will lead to more effective patient management and improve bed use. “It will allow us to prescribe the most appropriate antibiotics and anticancer therapy and avoid potentially debilitating side effects, so we can bring down the total cost of health care,” Bishop says.
Nucleic acid extraction, amplification, and detection will, of course, continue to be integral components of molecular diagnostics, Dr. Farkas says. “Next-generation sequencing is a variation on a theme. It may be used to detect, in a more elegant, more sophisticated informatics-based approach, parts of the genome associated with disease, drug responsiveness, and predispositions.” As costs come down over the next five years, Dr. Farkas suggests, it may be possible to sequence the entire genome of an individual for less than what it costs now to perform a one-gene diagnostic sequencing test. “How do we take advantage of that?” he asks. “I can imagine at age 18 or 21 an individual has his or her entire genome sequenced and put on a smart phone. They can then bring it to their doctor’s office, thereby providing a foundational element of a portable electronic medical record.” Having the genome available could add a lot more fluidity to diagnosis, in Dr. Farkas’ view. “One can easily imagine it happening. It would certainly be a paradigm shift in the IVD industry. All the wet lab work will have been done and, for any medical application that is genomics-related, and that’s a lot, the pathologist will work completely in silico.” This would make the laboratorian even more an interpreter of information.
Dr. Farkas sees mass spectrometry as having “clinically appropriate density” for future molecular tests that will target many more markers than typically examined today. When only a few targets are needed, up to six to eight, real-time PCR is a good modality. At the other end of the spectrum, DNA microarrays are very good at interrogating hundreds of thousands of targets. “But no medical test today requires that kind of density,” Dr. Farkas says. In his view, mass spec would be very good for tests such as cystic fibrosis or predisposition oncology or pharmacogenomics, with tens to hundreds of biological targets.
Mass spectrometry has already made inroads into clinical microbiology labs in Europe, says Markus Kostrzewa, PhD, director of molecular biology R&D for Bruker Daltonic in Bremen. “Currently we have about 60 systems in Germany, most in clinical labs,” he says. “We learned years ago that, aside from quality, three things are needed to make mass spec suitable for clinical settings.” First, software and usability had to be moved from the research level to the clinical level and sample preparation made simple. Second, “Biggest is not always best,” Dr. Kostrzewa says. An instrument that is small and easy to use is optimal. Third, Bruker had to build a reliable database to interpret results. “Doing these things made us successful much quicker than we anticipated,” Dr. Kostrzewa says.
In the U.S. market, awareness of mass spec applications for clinical microbiology was lagging behind the EU, but this is changing rapidly, says George Goedesky, Bruker’s executive director-NA for marketing/business development.
Predictive or companion testing, on the other hand, is “growing like crazy,” Buxton says. His explanation: “Oncologists are going well beyond established markers. They are reading articles that describe new markers” and putting them into practice. Buxton doesn’t see this as a black-and-white issue. “Critics of the FDA say it is moving too slowly. Oncologists see smaller studies that suggest that a therapeutic selection or predictive test is valuable, and they don’t think they should have to wait for FDA approval,” he says. Buxton was surprised in 2008, when the KRAS story broke and oncologists quickly started ordering it for colorectal cancer patients, that “they didn’t just order it for metastatic cancer, stage 3b or 4, where it would be reasonable to prescribe Erbitux.” They were also ordering it for stage 2 or 3a, to have the information in case the patient’s stage changed, he says.
He calls this situation “a bit unusual,” saying it goes beyond standard-of-care protocols. “Oncologists have a great interest in knowing the tumor’s characteristics,” Buxton says. “I think volume will continue to go up faster than FDA-approved drugs and biomarkers.” While some expect the current 25 percent-plus growth rate to continue, “often something comes in to depress the growth curve,” Buxton notes. “It might be that payers won’t pay for two or more targeted therapies or companion diagnostic tests simultaneously. But leaders in academia and industry, in the pharmaceuticals, diagnostics, and laboratory fields, and in the oncology community are working together to address the challenges of appropriate validation, use, and reimbursement for these tests, and working with both the CMS and the FDA,” he says. Buxton is optimistic that the regulatory and reimbursement environment of the future will be more favorable—“one that will allow this exciting area of medicine and science to continue its amazing progress.”
This discussion of the future of molecular diagnostics is perhaps best summed up with another technology-oriented aphorism, this one from computer scientist Alan Kay: “The best way to predict the future is to invent it.”
William Check is a medical writer in Wilmette, Ill.