May 2011
Feature Story

William Check, PhD

When it comes to sophisticated instrumentation for clinical laboratories, industry’s slogan seems to be “No lab left behind.” In last month’s article on next-generation sequencing, we reported that NGS is providing benefits for diagnosis in clinical genetics and oncology, but not yet having an impact in infectious diseases (ID). However, ID labs have the prospect of a high-tech method all their own—mass spectrometry. And just as NGS comes in many formats suitable for different applications, two flavors of mass spec are in development, with distinct advantages that make each suited to particular applications.

The three companies in the mass spec marketplace are Bruker Daltonics, Ibis Biosciences, and bioMérieux, none of which yet has an FDA-approved instrument. Bruker’s MALDI-TOF Biotyper is being validated for patient diagnosis by a half-dozen laboratories in the United States that are capable of this rigorous process. (Many laboratories in Europe already use the mass spec made by Bruker, a German company.) It appears to perform at least as well as or better than conventional automated machines for bacterial identification. In published evaluation studies, a second mass spec instrument, the Ibis T5000, also performed as well as or better than automated instruments now in clinical microbiology labs for those organisms for which it was tested. Ibis, a subsidiary of the Abbott Diagnostics Group, has designed a newer model of its technology, called the PLEX-ID, and is now gathering clinical data on its performance. Ibis employs PCR/electro­spray ionization mass spec. The bioMérieux instrument, like Bruker’s, is MALDI-TOF mass spec. It is early days yet, and how well these instruments perform under laboratory conditions, and which instrument ends up being most appealing to laboratory directors, is still an open question.

What is not an open question is that laboratorians expect mass spec to confer substantial benefits in clinical diagnostics. Nathan A. Ledeboer, PhD, D(ABMM), who has validated the Biotyper, says it provides 10 percent greater sensitivity. “We are able to call 10 percent more organisms correct on first pass than standard biochemical systems,” says Dr. Ledeboer, assistant professor of pathology, Medical College of Wisconsin, and medical director of clinical microbiology and molecular diagnostics, Dynacare Laboratories and Froedtert Hospital, Milwaukee. “All the while we do our job faster and more accurately and with the potential to significantly impact patient care.”

Mass spec may not be suitable for all laboratories at the moment. “It is good for large labs,” Dr. Ledeboer says, “but the capital expense may be prohibitive for smaller labs that could be looking at an instrument simply for increased accuracy of identification. For now, lack of FDA approval will be a major inhibitor for bringing these new technologies into the clinical lab.”

At Children’s Healthcare of Atlanta, Robert C. Jerris, PhD, D (ABMM), also is validating the Biotyper for detection of bacterial pathogens, notably those found in cystic fibrosis patients. “It looks very good for organisms that can be grown in culture, but not so much for direct specimen testing,” Dr. Jerris, who is director of microbiology, says of the Biotyper. On the plus side, because mass spec is a broad-based method, it offers the opportunity to integrate bacterial testing, fungal testing (minus mycelial fungi), and mycobacteria into a one-step process. “What that does for the routine micro lab is to cut down on QC for all the different identification methods we use,” Dr. Jerris says. His prediction: “This technology will be in most clinical labs in some capacity somewhere down the road.”

Robin Patel, MD(CM), FRCP(C), D(ABMM), chair of the Division of Clinical Microbiology and consultant in the divisions of Clinical Microbiology and Infectious Diseases at the Mayo Clinic, also is validating the Bruker MALDI Biotyper. “Its accuracy, efficiency, and cost are excellent,” says Dr. Patel, professor of microbiology and medicine and director of the infectious diseases research laboratory, Mayo College of Medicine. “Those three attributes make it very favorable for use in clinical microbiology labs.” She adds, “It can replace many, but not all, conventional and automated biochemical methods we currently use for identification as well as some 16S ribosomal gene sequencing.”

With the Ibis instrument, too, broad-based identification is possible, says Charlotte Gaydos, MPH, DrPH. In acute respiratory tract infections, for example, “You don’t ask, ‘Is organism X there?’ You ask, ‘What’s there?’” says Dr. Gaydos, professor of medicine in the Division of Infectious Diseases, Johns Hopkins University. Dr. Gaydos has demonstrated the efficacy of the first Ibis instrument, the T5000, for rapid identification of viruses from nasopharyngeal aspirates in acute respiratory tract infections. “It is a very good assay for identifying viruses from respiratory samples,” she says. “Not only do you get identification and differentiation of influenza virus, you also get respiratory syncytial virus and other respiratory viruses that you might think are influenza but are actually something else.”

Donna M. Wolk, PhD, D(ABMM), and others demonstrated the efficacy of the T5000 for identifying and genotyping methicillin-resistant Staphylococcus aureus compared with traditional methods as part of the multicenter MRSA Genetic Profiling Project (Wolk DM, et al. J Clin Microbiol. 2009;47:3129–3137). Her team was the first to demonstrate performance and accuracy for identifying a broad range of microbes directly from blood culture bottles (Kaleta EJ, et al. J Clin Microbiol. 2011;49:345–353). While the Bruker MALDI Biotyper is less expensive than Ibis, “the breadth and depth of PCR-ESI/MS is greater with regard to its ability to identify genus, species, genotype, antimicrobial genes, and genetic pathogenicity determinants often without the need for agar-based culture,” says Dr. Wolk, associate professor of pathology, University of Arizona College of Medicine, and chief of the Division of Clinical and Molecular Microbiology, Arizona Health Sciences Center. “It does cost more, but it delivers a lot more,” she adds. She calls the Ibis methods “powerful discovery tools” but says their costs need to come down and the technology made capable of random access before clinical labs will be able to experience the full impact. (Abbott declined to provide the price in advance of regulatory approval.)

Frederick S. Nolte, PhD, D(ABMM), F(AAM), professor of pathology and laboratory medicine and vice chair of laboratory medicine, Medical University of South Carolina, says he is “excited” about mass spec for infectious diseases. “It seems that the method will have the most traction in the identification of microbes that have been isolated in culture, with the potential to completely eliminate the need for conventional biochemical tests.” That would reduce the time needed for and the cost of microbe identification, after an initial capital investment in a MALDI-TOF mass spec instrument, he says.

Dr. Nolte admits he is not a mass spec expert and does not have hands-on experience with either type of mass spec instrument. But he doesn’t foresee the PCR mass spec approach having much of an impact on diagnostics. While he acknowledges that PCR mass spec “is capable of massively multiplexed analysis and can simultaneously detect and identify a broad range of microorganisms in a given sample, including viruses, bacteria, and fungi,” he says, “It is very complicated technology and not likely to facilitate laboratory-developed tests even if the lab could afford the investment.”

Dr. Patel offers a somewhat different view of the relative merits of the two versions of mass spec. “The technologies are different,” she says. “While in theory both envision trying to do similar things, arguably the main aspect they have in common is that they both use the term ‘mass spec.’” For instance, the Ibis instrument measures nucleic acid, while the Bruker measures protein profiles. Because the Ibis has an upfront PCR step, it has greater sensitivity, so that testing may be done directly from patient specimens, while the Bruker generally requires testing to be done on isolates grown on a plate or in blood culture broth. Even so, Dr. Patel found that the Bruker Biotyper has “excellent” accuracy for the indications for which she tested it.

Once you have grown organisms, the turnaround time with MALDI-TOF mass spec is much faster than with PCR mass spec. However, because PCR mass spec can be performed directly on patient specimens, the time for growing organisms is eliminated. Dr. Patel would like to evaluate the PLEX-ID in her laboratory. “The applications are very different and possibly complementary,” she says.

In mass spectrometry in general, a complex sample is analyzed by the mass-to-charge ratio of its component parts. In traditional matrix-assisted laser desorption ionization–time of flight—or MALDI-TOF—mass spec, which is the basis of the Bruker and bioMérieux instruments, a sample is ionized and its components are separated in an electromagnetic field. In the Ibis instruments, electrospray ionization moves ions in solution into the gas phase, which is less destructive to nucleic acids. Time-of-flight analysis in an electromagnetic field then determines the mass-to-charge ratio, just as in MALDI-TOF.

Ibis’ products grew out of a contract from the Defense Advanced Research Projects Agency for broad-spectrum detection of biothreat agents—including bacteria, viruses, and fungi—from a variety of body fluids. Ibis’ early T5000 “was built with off-the-shelf components and integrated by us,” explains Mark Eshoo, PhD, director of new technology development at Ibis. After Abbott acquired Ibis, the current PLEX-ID system was built with exactly the same chemistry but designed and engineered for a clinical environment, Dr. Eshoo says. “It has features for more efficient sample processing and higher throughput,” he says, citing a dual-head sprayer as an example. “The original design had a single-head sprayer,” Dr. Eshoo says. “The dual-head sprayer enables us to run one head while the other is being cleaned. It literally doubles the number of samples going through the instrument in a period of time.”

Abbott and Ibis are sponsoring clinical trials of the PLEX-ID at major medical centers. They will have several posters this month at the American Society for Microbiology general meeting and two presentations at the European Congress of Clinical Microbiology and Infectious Diseases. “Accuracy of the PLEX-ID so far is exactly comparable to the T5000,” says David J. Ecker, PhD, division vice president of Ibis. “We haven’t seen anything different between the two systems.” This is not surprising, since much of the accuracy of the T5000 is determined by the PCR step, and the PLEX-ID uses the same primer sets.

One example of the accuracy of the T5000—and presumably the PLEX-ID—comes from Dr. Gaydos’ work on 454 nasopharyngeal aspirates from patients with acute respiratory tract infections gathered from across the U.S. during the 2007-2008 flu season (Chen KF, et al. J Virol Methods. 2011;173:60–66). At that time, the Ibis respiratory plate detected and differentiated flu A/B, coronavirus, bocavirus, parainfluenza 1-4, adenovirus A-F, RSV, and human metapneumovirus. (Bocavirus has since been dropped.) When compared with methods then in use in the Johns Hopkins virology laboratory, the sensitivity of the T5000 was 87.1 percent to 88.1 percent (samples were tested in two groups), and specificity was 92.3 percent to 96.9 percent. Sensitivity was lowest for parainfluenza. “The real strength of the T5000 is in its bioinformatics,” Dr. Gaydos says. “A positive is really positive.” Dr. Gaydos’ group is just finishing its first clinical trial with the PLEX-ID, for detection of H1N1 influenza. The method can differentiate the pandemic strain from the so-called seasonal strain of H1N1.

The T5000 is also good at identifying bacteria and yeast bloodstream pathogens from positive blood culture bottles, according to Dr. Wolk’s published data (J Clin Microbiol. 2011; 49:345–353). In this work, DNA was directly extracted from 234 BacT-Alert blood culture bottles, and results were compared with those obtained by standard methods. Concordance at the genus and species level was 98.7 percent and 96.6 percent, respectively. As well, the system identified mixtures of microbes in 29 blood culture bottles, including mixtures containing gram-positive and gram-negative organisms. Results were obtained in as few as five to six hours after a positive alarm from the automated blood culture system.

Dr. Wolk collaborated with a research group at the University of Geneva to compare the performance of the T5000 methods with the Bruker MALDI Biotyper for diagnosing bloodstream infections. The Biotyper performed on colony isolates from subcultures; the T5000 analyzed samples directly from blood culture bottles. “The two methods showed basically equivalent performance for accurate genus and species identification of bacteria and yeast compared to each other,” Dr. Wolk says. A report of the study, to be published in Clinical Chemistry, is in press.

Despite PCR mass spec’s advantages, Dr. Wolk says the T5000’s clinical utility is limited because it runs only in batch mode. “Batch mode is still a major impediment to incorporation [of PCR mass spec] into the clinical lab for identification of ­bacteria,” she says.

With the PLEX-ID, this obstacle is lessened in two ways. First, though the instrument still runs a 96-well plate, a plate can be run with as few as six to 12 samples. Second, the desalting step is uncoupled from electrospray ionization. After amplification, salts have to be removed from the DNA before it is ionized. In the PLEX-ID, one sample doesn’t have to wait for the next sample to be desalted to be analyzed. Turnaround time from loading the plate to first result is about 45 minutes, Dr. Eshoo says.

Dr. Ledeboer’s clinical laboratory has the distinction of being the first in the United States to go live with mass spec for microbiological diagnosis. He and colleagues have implemented mass spec for identifying yeast and bacteria from positive urine cultures. They have examined the Biotyper’s accuracy on many microbiological samples, and have four papers in the publication process. “Our verification studies consist of more than 1,000 isolates compared to sequencing and biochemistry,” Dr. Ledeboer says. “From the bacterial standpoint, MALDI-TOF mass spec can identify pretty much everything, including mycobacteria and yeast.” An area still under study is filamentous fungi. Dr. Ledeboer terms the results from a just-completed study of the Biotyper for positive blood cultures “nothing less than fantastic.” It identified organisms between 18 and 69 hours faster than standard methods, he says, adding, “You can have a genus and species identification within 30 minutes of the blood culture turning positive.”

With an estimated upfront purchase cost of about $200,000, and assuming a laboratory does 5,000 bacterial and yeast isolates per month, Dr. Ledeboer has calculated payback time for the Biotyper to be 14 to 18 months. From a labor standpoint, he says, there is not a major change in going from standard biochemical methods to the Biotyper. “You still have to inoculate plates, read plates, and pick colonies,” he says. “And you have to run susceptibility testing. The benefit is that you don’t have to do all the accessory or offline tests, such as catalase or Gram stain. One test and you’re done.”

Comparing MALDI and PCR/ elec­trospray mass spec, Dr. Ledeboer calls them “two absolutely different technologies.” PCR/electrospray is better suited to applications direct from specimens, such as virology and parasitology, where organisms are difficult or impossible to grow, he says. MALDI is currently most widely used from positive cultures.

At Children’s Healthcare of Atlanta, Dr. Jerris initially engaged with Bruker on a project to look for hard-to-identify organisms in cystic fibrosis, particularly Burkholderia cepacia (formerly Pseudomonas cepacia). “The Bruker Biotyper matched up quite well with gene sequencing in Dr. John LiPuma’s lab,” Dr. Jerris says, referring to the director of the Burkholderia cepacia Research Laboratory and Repository at the University of Michigan. Of 29 Burkholderia isolates, the Biotyper correctly identified genus and species in 24. Most important, all 29 were identified to the Burkholderia genus and could be referred for further workup as needed. At this point, Dr. Jerris has validated the Biotyper for 76 genera and species of bacteria relative to conventional methods. “We have just started a validation study with yeast isolates,” he says. “We will continue from there.”

In addition to establishing clinical validity of the Biotyper, Dr. Jerris did a lean and cost analysis, the results of which he will present at the ASM meeting. From a labor viewpoint, he says the Biotyper is not difficult for medical technologists to learn to run.

He notes some limitations to the instrument. For mycobacteria he had to modify the basic procedures to extract an adequate amount of material to obtain an identifiable protein profile. Other limitations relate to the database of protein profiles. “In many ways the database for MALDI-TOF exceeds what is in our phenotypic system,” Dr. Jerris says. “We now know that for some genera, such as Acinetobacter, there is a subgroup of overlapping species, which helps explain why our phenotypic analyses are sometimes confused. Whereas MALDI-TOF gives you a definitive species or a genomospecies designation. Another limitation is with viridans streptococci versus Streptococcus pneumoniae. So we just look to see if an isolate is bile soluble, rather than putting it on the instrument,” he says.

(Bruker carefully curates its proprietary database, says Markus Kostrzewa, PhD, director of molecular biology, R&D, at Bruker Daltonik GmbH in Bremen, Germany. “The quality control process is complex and involves the origin and identification of samples, MS measurement parameters, spectra quality as well as database consistency check,” he told CAP TODAY. “Each database reference has to be approved by two people of my department.”)

Dr. Jerris also sees a need for better integration of the mass spec with an instrument for susceptibility evaluation and with laboratory information systems. “Now, when you get information out of the MALDI-TOF instrument, you have to type it into a susceptibility identification system or directly into an LIS,” he says. “When susceptibility assessment gets together with MALDI-TOF, it will become a complete instrument.” Efforts are underway to accomplish this.

“This is a simple one-step process for identifying the vast majority of organisms you see routinely in the clinical laboratory,” Dr. Jerris says of MALDI-TOF mass spec. Once you have a colony, you pick it and place the sample on a steel grid, or target. Then you add a small drop of matrix solution and put the template directly into the instrument. The grid can be cleaned and reused. “[MALDI-TOF mass spec] is very cost-effective. It looks to me like this is going to be one of the true winners in the clinical micro lab,” Dr. Jerris says.

Dr. Patel plans to implement a Bruker MALDI Biotyper for clinical diagnostic use later this year. Her initial validation study compared the Biotyper to current methods for identifying gram-negative bacilli. “We selected 308 organisms we would see routinely in the lab plus 132 more esoteric organisms,” she says. “Common” organisms were those seen at least once per week, such as E. coli and Pseudomonas aeruginosa. In this evaluation, “The Bruker system compared very favorably to an automated identification system, the BD Phoenix,” Dr. Patel says (Saffert RT, et al. J Clin Microbiol. 2011;49:887–892). Among the common isolates, it performed equivalently to the compared method. “We found excellent species-level identification—93 percent—with both systems,” Dr. Patel says. “At the genus level the Biotyper identified 96 percent of isolates and the Phoenix 95 percent.” Among uncommon isolates the Biotyper outperformed the Phoenix at both species (56 percent versus 34 percent) and genus (85 percent versus 52 percent) levels.

Dr. Patel says the Biotyper was not perfect. She encountered a few database problems similar to those Dr. Jerris described. There were some situations in which it was challenged by very closely related organisms, such as E. coli and Shigella species. “We had a Shigella flexneri isolate in our evaluation that was misidentified as E. coli,” Dr. Patel says. The Bruker database evaluated did not contain Shigella species. “E. coli and Shigella species are likely not distinguishable by this technology,” Dr. Patel says, adding that users of the technology can deal with such issues provided they recognize they exist.

Dr. Patel has also evaluated the Biotyper for gram-positive cocci and blood culture isolates and will present those results at the ASM general meeting. Another group at Mayo Clinic, under the direction of Nancy L. Wengenack, PhD, has evaluated the Biotyper for yeast and found a high level of accuracy (Dhiman N, et al. J Clin Microbiol. 2011;49:1614–1616). They also determined that the Biotyper is rapid, with a hands-on time of 5.1 minutes per identification, and cost-effective, at $0.50 per sample.

Beyond pathogen identifications, Dr. Ledeboer sees MALDI-TOF mass spec as a step to broader automation in the clinical microbiology laboratory. “It’s going to allow us to better utilize full automation in the lab,” he says. “Automation offers great potential for labs to fill the void for retiring technologists.”

Dr. Jerris is literally positioning his Biotyper for broader use. “We have it housed between the core lab and the micro lab,” he says, “in anticipation of this becoming useful for both proteomics in the core or chem lab as well as identification for the micro lab.”

This is already happening in Germany where MALDI-TOF mass spec has been in clinical use longer. Says Bruker’s Dr. Kostrzewa, “Large multipurpose, centralized systems are being used in Germany, such as at the medical service of the German army in Munich.”