The analytic power of microarays
William Check, PhD
Each year the new vintage of Bordeaux wine is auctioned, with wholesalers typically predicting a great future. Sometimes a demurring voice is heard. Reputations—and large amounts of money—are staked on an outcome that will not be known for years. A somewhat similar situation surrounds the future of the newest molecular technology, DNA sequencing, in the diagnostic laboratory.
Many observers predict that sequencing will play a major role in clinical microbiology laboratories, such as in detecting resistance mutations in viruses and bacteria, especially as more-automated instruments become available. Others remain to be convinced. Meanwhile, vendors invest substantial resources in a technology that is being judged alike by laboratorians, clinicians, and market forces, and whose value remains to
Eric Shulse, director of molecular diagnostics at Applied Biosystems, Foster City, Calif., is bullish on DNA sequencing. "We have seen, over the last 10 years, a rapid acceleration in technology development as well as application of that technology to various genomes, both human and disease-associated organisms," Shulse says. "We believe there will be a major change in the way medicine is actually practiced in our lifetime." Sequencing will play an important role in that change, Shulse predicts, with the only question being "whether it plays the major role or not."
Michael Wilson, MD, director of the Department of Pathology and Laboratory Services at Denver Health Medical Center, agrees that molecular diagnostics will play an important role in the clinical microbiology laboratory. However, he cautions against rampant expectations. "As each new generation of molecular tests has come out, starting in the 1980s, there have been widespread predictions that this technology will revolutionize the field and supplant traditional diagnostic methods," Dr. Wilson says. "But these predictions have not come true." Molecular techniques already adopted, such as nucleic acid amplification, have their niches. But those niches are limited and often consist of detection or quantitation of a pathogen for which no satisfactory test previously existed. "No technology has ever completely replaced whatever has existed before," Dr. Wilson says. DNA sequencing, too, will most likely have to find its place within this larger picture.
Whatever contribution sequencing will eventually make to infectious disease diagnostics, several factors suggest that now is the time for that contribution to begin. Current microbial identification systems work by phenotypic methods, using biochemical panels. Yi-Wei Tang, MD, PhD, medical director of the molecular infectious diseases laboratory and assistant professor of medicine and pathology at Vanderbilt University, says: "For the majority of samples, an experienced medical technologist can say what the organism is. But with some organisms, phenotypic methods have problems."
Dr. Tang cites these examples:
- Many bacteria, including Enterobacteriaceae, exhibit
- Sometimes a patient who has bacteremia or sepsis is already on antibiotics; it is still important to know the identity of the infectious agent, but it is very hard to isolate the organism from the patient's blood due to the antibiotics.
- Many new microbes are causing problems and often we don't know their biochemical reactions. Dr. Tang points to Bordetella holmesii and Treponema whipellii, the latter of which is still not cultivatable in routine conditions.
- Some microorganisms, such as Acinetobacter sp, are negative on most common biochemical reactions, both enzymes and sugars.
- For Mycobacterium tuberculosis (Mtb), the phenotypic method works, but physicians cannot wait weeks or months for a result.
More generally, sequencing will make an impact with esoteric organisms where the delay to definitive identification is significant, says David Persing, MD, PhD, vice president of diagnostics development, Corixa Corp., and medical director of the Infectious Disease Research Institute, Seattle. A disproportionate amount of time is required for identification of five to 10 percent of clinically important organisms. "Many laboratories throw in the towel rather than making special media and devoting the necessary technologist time to identifying these organisms," Dr. Persing says, "even though there may be some clinical benefit in doing so."
Sequencing is being introduced into microbial diagnosis not only because of the stimulus from the demand side but also because of the rising availability of sequencing technology. David Hillyard, MD, director of molecular pathology, ARUP, Salt Lake City, observes that sequenced-based assays for hemoglobinopathies are used in the human genetics laboratory and fragment analysis for clonality often accompanies the workup of cancers. In the infectious disease arena, HIV resistance testing is already set up in a number of mid-size and large reference laboratories, and several are using sequencing for hepatitis C genotyping. As a result, Dr. Hillyard says, "There is enough of a critical mass in many institutions that people are establishing laboratory units that get inspected and certified and function as clinical sequencing centers. Having that expertise and equipment in place opens the door to other applications."
He also notes that the spinoff of more sophisticated technology from the human genome project has made DNA sequencing increasingly easier to perform. And over the last five years, the price of doing sequencing has dropped five- to 10-fold, Dr. Hillyard estimates, so that its cost is now "fairly reasonable." Adding these two advances together, he concludes, "I don't see these as tests that are going to be cost-prohibitive just because they are sophisticated."Among actual or potential applications of sequencing in the clinical microbiology laboratory, HIV genotyping to detect antiretroviral resistance mutations was one of the first to become part of the standard of care. While the largest reference laboratories do HIV genotyping by in-house assays, two kits have been developed: Visible Genetics' TruGene HIV-1 and Applied Biosystems' ViroSeq. Both are for research use only, although FDA 510(k) approval is being sought.
Some large hospital and reference laboratories have developed in-house sequencing assays for antimicrobial resistance genes, such as mecA in methicillin-resistant Staphylococcus aureus and vancomycin resistance in Enterococcus. Among viruses, assays for hepatitis B virus resistance to lamivudine and famciclovir are done by both in-house and kit-based assays, and tests for cytomegalovirus drug resistance to ganciclovir and cidofovir are done by in-house methods.
Dr. Hillyard describes a more extensive use of sequencing: faster identification of mycobacterial species. "Up until very recently we have been able to identify basically a group of five or six common mycobacteria by probe hybridization," Dr. Hillyard says. "Atypical mycobacterial species not identified by hybridization had to go through traditional methods—growth on various media, biochemical tests, and HPLC, which are very time-consuming."
But a form of DNA sequencing has been introduced in the past year that has provided a powerful tool for rapidly identifying organisms that are difficult to identify quickly. It is sequencing of the 16S ribosomal RNA (rRNA) gene. "The impact of that technology, at least in our laboratory, has been felt the most for mycobacteria," Dr. Hillyard says. With older methods, it could take four to six weeks to identify some mycobacterial species. "Now," Dr. Hillyard says, "as soon as the specimen can be grown and we know it is some sort of mycobacterium, DNA is prepared and submitted for sequencing. And that sequence result is able to identify the vast majority of important mycobacterial species.
"In our laboratory," Dr. Hillyard says, "turnaround is on the order of a couple of days, once we have growth of the organism. That's a tremendous improvement."
He doesn't see the economics of mycobacterial identification changed substantially by the new method. "Identification of M. tuberculosis by traditional methods is very expensive for any laboratory and the cost is often not recovered by the laboratory," he points out. "So I don't see the cost equation changing radically. The time to identification, however, is dramatically improved."
Both in-house and commercial kits using 16S rRNA gene sequencing are being developed to identify gram-negative and other eubacteria, especially fastidious organisms and anaerobes, says Stephen Dumler, MD, associate professor of pathology and director of the division of medical microbiology at Johns Hopkins University. "This method is good for the identification of bacteria that are not amenable to identification through the usual metabolic profiles, because they grow slowly, for instance, or are relatively metabolically inert," Dr. Dumler says.
In 16S rRNA sequencing, standard means are used to amplify the gene encoding the 16S rRNA subunit, which comprises about 1,500 base pairs. Then the amplified DNA is sequenced. "This gene has conserved regions that are nearly identical among all bacterial species," Dr. Dumler explains, "plus intervening hypervariable regions that are very different." Primers for amplification are based on conserved regions, while a bacterial isolate can be quickly identified by comparing the sequence of its variable region(s) to sequences of known species in a database. Although 16S rRNA gene sequencing has been used as a research tool for many years, its application to clinical microbiology is rather new.
Some laboratories have devised in-house assays for 16S rRNA gene sequencing. A commercial system, called MicrosSeq, is available from Applied Biosystems for research use only. MicrosSeq includes a database of more than 1,000 prototype bacterial species; users buy a software subscription and a package for microbial identification. Applied Biosystems hasn't decided whether to commit the resources to get MicrosSeq approved for clinical use, Shulse says.
Besides mycobacterial species, 16S rRNA gene sequencing can help identify such slow-growing or inert organisms as actinomyces, nocardia, and suspected Bartonella, Dr. Dumler says. Dr. Tang adds Bordetella holmesii and Acinetobacter sp as unusual bacteria whose identification is facilitated by the method.
One problem with this method is expense—to sequence the entire rRNA gene costs about $120 just for materials, Dr. Tang says. He and Dr. Persing found that sequencing about the first 500 base pairs—the most polymorphic region of the gene, which can be amplified in a single reaction—provides a match good enough to initiate clinical treatment and at a reduced cost.
Using the MicrosSeq kit with an Applied Biosystems instrument and database, Drs. Persing and Tang evaluated abbreviated 16S rRNA gene sequencing for identifying organisms, both gram-negative and gram-positive, that could not be identified by routine laboratory methods (J Clin Microbiol. 1998;36 (12):3674-3679. J Clin Microbiol. 2000;38(4):1676-1678). At the time, they were both at the Mayo Clinic, where difficult specimens were sent to a referral area of the clinical microbiology laboratory for more esoteric testing. Identification based on sequencing achieved a high level of agreement with identification by special phenotypic tests used in the referral area of the laboratory.
"This is an area in which 16S rRNA sequence technology will have its first impact," Dr. Persing says. One isolate was a new species of Bordetella, Bordetella holmesii, which, Dr. Persing suggests, could be an emerging pathogen. Other isolates were Aeromonas spp.
In some cases, it seemed as though the sequence was more accurate than phenotypic tests, Dr. Persing observes. This raises a question: When taxonomy is unclear, which do you believe—sequence or biochemical reactions? "We will need to set up criteria for which is the preferred method," he concludes.
In a separate evaluation, Drs. Persing and colleagues at the Mayo Clinic amplified 16S rRNA sequences directly from blood culture bottles showing growth. In this study, organisms could be identified by conventional methods, but, Dr. Persing says, "We wanted to see if we could detect organisms at the first sign of growth in blood culture. 16S amplification and sequencing worked to detect all of the organisms identified by conventional methods," he says. "The theoretical advantage of sequencing was one to four days saved in the time to get a full identification."
Dr. Tang cites specific instances from the referral study in which the 16S rRNA sequence proved useful. Pathogenesis and resistance patterns differ among Corynebacteria, he notes, but it is difficult to identify these organisms to the species level by phenotypic methods. To complicate matters, C. diphtheriae is less common now, while C. jeikium, a new Corynebacterium species that causes bacteremia and endocarditis, is becoming more common. Sequencing differentiated these species in less than two days for about $60, Dr. Tang says; biochemical methods could be more expensive and take even longer.
In another case, a patient with endocarditis came to the hospital on many antibiotics. But it was not clear which was effective. The patient's blood did not grow out an organism, so the 16S rRNA gene was amplified from the patient's blood. "The bacteria were dead, but their corpses were still floating around in the blood," Dr. Tang says. The pathogen turned out to be a standard viridans streptococcus, so the patient could be left on a conventional regimen.
"We have used this sequencing system in more than 300 cases," Dr. Tang says. "In some instances we worked with pure colonies, in others with direct sterile specimens, such as cerebrospinal fluid, blood, heart valve, or synovial fluid. We found the method to be a quick, accurate, and sensitive technique for bacterial identification where conventional methods didn't work very well or not at all."
Looking forward months or years, an even more impressive story will be the identification of fungus and yeast isolates, Dr. Hillyard says. By traditional methods, these are slow and hard to do. "Doing those by sequencing will be revolutionary," he says.
The fungal identification database is growing rapidly, Dr. Persing adds. "One of the best ways to identify fungi now is by sequence," he agrees. "There is still controversy in this area," he acknowledges. "But, like any field when you have a major change in the way things are done, it takes time to catch up."
Three companies market commercial sequencing systems: Applied Biosystems, Visible Genetics, and Beckman Coulter.
Applied Biosystems has provided sequencing equipment to the research market for almost 15 years, Shulse says. Its entry into molecular diagnostics occurred almost 10 years ago. About five years ago the company worked with the AIDS Clinical Trials Group to develop the first HIV genotyping system. Shulse says, "Probably all major reference laboratories in the U.S., as well as Virco and ViroLogic, have standardized [their in-house HIV genotyping] assays on our sequencing platforms."
The initial HIV genotyping assay developed by Applied Biosystems with the AIDS trials group formed the basis for ViroSeq, which Shulse calls "our sequencing assay of most interest right now. Because the virus mutates so rapidly, other technologies have not been effective in determining resistance," he adds. "Some array technologies have focused in this area, but they have not been successful." Sequencing is currently the only tool that works for detecting resistance mutations in HIV. Applied Biosystems has submitted for 510(k) approval the ViroSeq HIV-1 genotyping system, which will consist of the reagent kit plus interpretive software. Shulse says the interpretive HIV sequencing software "will make it very easy for users to review mutations and report out specific sequences. And the software will indicate resistance profiles by antiretroviral drug."
Included in the ViroSeq system is a license to use PCR on an Applied Biosystems sequencing platform. "We have now expanded our access to PCR for diagnostics," Shulse says, "so all our diagnostic kits will carry a license for PCR with them." That acquisition of additional rights has allowed a broader molecular diagnostics initiative consisting of a joint venture between Applied Biosystems and Celera Genomics, which will be called Celera Diagnostics, and which is still in the formative stages. Applications could include HCV genotyping, as well as noninfectious uses such as detection of cystic fibrosis and fragile X syndrome and HLA reagents.
Initially, sequencing depended on separation of DNA fragments with slab gel electrophoresis. But this technique requires considerable manual operation, making it more time-consuming: Gels have to be made just before each separation, and a technologist must be trained to prepare samples and load each lane individually. Moreover, Shulse says, "Every time you make a gel you are running a chemistry experiment—that polymerization may or may not work. There has been a lot of effort to make that less of a problem, but that is still where most problems occur with slab gel sequencing."
Because of this drawback, Applied Biosystems' 377 slab gel sequencer, which has been the workhorse in DNA sequencing, is being discontinued and replaced by capillary electrophoresis platforms, which are inherently more automatable. With capillary electrophoresis, amplified samples are placed into trays and the machine automatically loads the samples into capillaries.
Applied Biosystems has marketed a single-channel capillary instrument, the 310 Genetic Analyzer, for five to six years. But it does not have the capacity needed for large clinical laboratories. The 3700 DNA Analyzer, with 96 channels, "was really the tool that finished the sequencing of the human genome," Shulse says. But it is too expensive for most clinical laboratories.
Applied Biosystems' most recent instrument, and the one that appears to be most suited to clinical laboratories, is the 16-channel 3100 Genetic Analyzer. "The nice thing about that instrument for laboratories used to automation is that it has many automated features," Shulse says. With the 3100, a technologist can load amplified samples into a tray, push a button, and walk away. Each ViroSeq sample uses six lanes and the 3100 finishes a run in about two hours, so it runs 24 amplified patient samples with quality controls in 24 hours.
Both the capillaries and the polymer used for separation are pre-manufactured and quality controlled. Capillaries have to be replaced every 100 or so runs, so a 3100 can run full time for about four days before capillaries have to be changed.
Dr. Hillyard reports a positive initial experience with the model 3100 capillary electrophoresis sequencing system compared to his long-term experience with the 377 slab gel system. "It is a step forward in several ways," he says. He likes the increased potential throughput, based partly on shortening the run time from six to two hours. Even with only 16 lanes, he calls throughput "remarkable." Because runs are shorter, it is possible to do many different kinds of sequence analysis. "So you can offer a fairly broad menu of sequencing tests and still have lowered turnaround time," Dr. Hillyard notes.
He also cites the 3100's automated features. "Sequencing capillaries can be loaded robotically out of microtiter plates," he says. "Beyond that," he adds, "we see strikingly improved quality of sequence in the first assays we have looked at with capillary electrophoresis compared to conventional slab gel electrophoresis." He finds that the quality of the HIV sequence is high enough to detect mixed infection.Visible Genetics, of Suwanee, Ga., has developed a gel cassette-based electrophoresis sequencing system called OpenGene, which is "the first sequencing system designed for use in the clinical laboratory," says Stephen Day, PhD, director of medical affairs. It involves "no cumbersome gel casting," he says. Rather, the system includes premade, preassembled glass cassettes that clamp horizontally into the drawer of an instrument called a Toaster. Premade cartridges of polyacrylamide fit into an injector. Using a handle on the injector, the technologist squeezes the polyacrylamide directly into the cassettes. Using UV light, the Toaster polymerizes the gel in about five minutes.
Visible Genetics' first application of OpenGene, HIV genotyping for detecting antiretroviral resistance mutations, was submitted to the FDA in September 2000. The FDA has completed its review and found the product to meet its requirements for clearance, subject to ordinary course labeling and other routine issues, Dr. Day says. The product has also received regulatory approval for clinical use in Canada, France, and Argentina.
This integrated system includes the TruGene HIV-1 Genotyping Kit (the chemistry), a sequencer and the gel Toaster (the hardware), and a computer worksystem with GeneObjects software. GeneObjects merges, aligns, and reports sequences base by base. It generates an HIV-1 mutation pattern and tells what that pattern means using GuideLines Rules, an algorithmic rule set. Interpretive rules are set by an international panel of academic experts who meet at least semiannually to review HIV resistance data. TruGene purchasers will get rules upgrades free.
"What makes our algorithmic rule system unique," Dr. Day says, "is that it includes correlations with in vitro phenotypic resistance data and it considers data from in vivo virologic response studies. Where those contradict, it always defers to the in vivo virologic response."
The Visible Genetics software uses the GuideLines Rules to generate the interpretive TruGene HIV-1 Resistance Report. "This differentiates our system from all other commercially manufactured HIV-1 genotyping systems, none of which have software that provides an interpretation of the mutational patterns," Dr. Day says. In addition to listing the detected drug resistance mutations and giving an interpretation, the TruGene report provides the clinician the rule and the evidence basis ranking for each rule used in making the interpretation for each patient sample, he adds.
The OpenGene system has a footprint less than one square foot, which Dr. Day says is much less than capillary electrophoresis systems.
Visible Genetics offers purchasers training on site and at its facility in Atlanta, furnishing a proficiency and certification panel for each individual technologist. "That is something that is going to be required by FDA," Dr. Day predicts.
At Beckman Coulter, says Jeff Chapman, PhD, strategic marketing manager for the CEQ 2000 XL DNA Analysis System, the focus is on both DNA sequencing, either of genomes or of infectious agents; and genotyping analysis, to identify organisms or to associate specific genes with disease states. "In other companies' instruments, this is done with different packages of software," Dr. Chapman says. "In the CEQ, we integrate all elements into a core software package and build in simple interfaces to each application." This architecture requires less expertise, he says, and "moves analysis from the hands of the few into the hands of the many."
To address sample preparation and analysis and information management—key components of the genetic analysis process—Beckman Coulter offers several systems, including the CEQ 2000 XL DNA Analysis System, the P/ACE MDQ DNA Analysis System, and the Biomek 2000 and Biomek FX automated sample preparation systems. Although these product lines are for research use only, there is interest in qualifying them for future clinical use. Because of the modular nature of the CEQ 2000 XL, which makes it possible to add platforms as capacity increases, "Pretty well any laboratory can get into sequencing and genotyping at the cost of HPLC," Dr. Chapman says. At about $85,000 for the base unit, the CEQ 2000 is priced comparably to other analytical equipment that a standard laboratory would purchase, he says.
Will DNA sequencing turn out to be a classic vintage among molecular diagnostic technologies? There seems to be agreement that sequencing will not displace conventional culture-based phenotypic methods of speciation, which usually take only 24 to 48 hours and are significantly cheaper than molecular methods. "The simple inexpensive technology already used for most organisms will continue to be used," Dr. Hillyard says.
"Even though not every laboratory will be equipped to do this technology," he adds, "it will impact many laboratories." Those that can't do sequencing themselves will be able to send samples to a reference laboratory when sequencing is called for and get identification more quickly and more precisely than in the past.
"Sequencing has become a standard technology at large high-complexity reference laboratories," Shulse says. "Whether it will penetrate into community hospitals is a question that I think will be impacted by the economics of what makes sense. Also, it still requires special expertise. You can't just take any clinical laboratory technologist and tell them to read the manual and run the test. It is more complex than a viral load test. But laboratories that have adopted viral load technology certainly should be able to use this technology."
William Check is a freelance medical writer in Wilmette, Ill.