Feature Story

Stopping rogue infectious agents in the nick of time

July 2003
Paul Karr

In most hospitals, using the best available technology, it could

take from two days to a week to recognize an outbreak of bacterial infection, positively identify the strain of pathogen and its history of resistance to antimicrobial agents, and then begin fighting back with the correct antibiotics.

But a new technique may be able to cut response time substantially if it proves to be reliable. And that’s just what pathologists and technologists at Johns Hopkins Hospital in Baltimore are hoping to find out.



The technique works by searching for special repeated chromosomal

DNA sequences in locations unique to pathogens and pathogenic strains, then typing and cataloging the distinctive pattern of each bacteria’s DNA fingerprint for comparison with close relatives and historically archived infections.

At Johns Hopkins, a medical technologist loads DNA extracted from organisms recovered from infected patients’ blood, respiratory, or stool specimens into a small tray of computer chips, then inserts the tray into an analyzing device. Within hours, reactions will have been performed to amplify certain segments of the DNA, and computer software will have typed the DNA of the segments, compared them with each other and a library of archived data, and picked them out of a database lineup as a specific strain of, for example, methicillin-resistant Staphylococcus aureus. The infection might be contained before it becomes a full-blown outbreak.

“The advantage to real-time molecular typing,” says Karen Carroll, MD, associate professor of pathology at Johns Hopkins University School of Medicine and principal investigator in the pilot program to test the typing product in Johns Hopkins Hospital, “is that you may have an answer [about strain type] the same day an organism is identified. Notification of the hospital’s infection control staff would ideally lead to interventions such as improvements in hand hygiene and isolation of patients, among other control strategies, 24 to 48 hours sooner. And sooner is better.”

Although clinical studies have not yet been published, the system Johns Hopkins is testing—unveiled by Houston-based Bacterial Barcodes in 1999 and improved in 2002 and again in 2003—may type pathogenic strains of DNA more quickly, compactly, and cost-effectively than other existing methods. It was developed by medical researchers at Baylor College of Medicine, at Texas biotechnology company Bacterial Barcodes, and at a California molecular microtechnology firm.

The typing with Bacterial Barcodes’ technology takes “hours instead of days, in some cases,” says Dr. Carroll, director of the hospital’s Division of Medical Microbiology. “In this situation, you’re trying to determine whether you have a transmission problem in the hospital and what you can do to intervene. Isolate the patients? Get health care workers to change their hand-washing procedures? Diagnostic labs can have a real impact on this intervention process by doing real-time typing, and Bacterial Barcodes’ kits may be able to help hospitals achieve that goal.”

Hospital-acquired infections are a serious and growing problem in American hospitals.

A Feb. 13 article in the New England Journal of Medicine (Burke JP. 2003;348:651-656) said incidents of nosocomial infection afflict five to 10 percent of all patients admitted to acute-care hospitals, causing 90,000 deaths and generating $4.5 to $5.7 billion in health care costs annually in the United States alone. The incidence of hospital-acquired infections appears to be steadily increasing, and for dangerous bloodstream infections and antibiotic-resistant Staphylococcus aureus infections, it has risen at alarming rates during the past two decades.

“Although rare, laboratories may recover organisms that are resistant to every antibiotic that is tested,” Dr. Carroll says. “We certainly don’t want those organisms to be disseminated throughout our hospital environments.”

The key to arresting such infections and outbreaks is to rapidly detect drug-resistant or variant strains of a pathogen. In a 1999 study conducted at Chicago’s Northwestern Memorial Hospital and reported in the American Journal of Clinical Pathology, the use of DNA typing cut the nosocomial infection rate at the hospital by 23 percent in two years, reducing patient infections by an estimated 270 cases per year and saving nearly $4.4 million in estimated health care costs during that period (Hacek DM, et al. 1999;111:647-654).

“Such an approach for managing nosocomial infections is technically possible, medically useful, and economically justified,” the authors of the report concluded.

“They analyzed outbreaks in a timely fashion, and that was the key difference,” explains James Versalovic, MD, PhD, director of microbiology laboratories at Texas Children’s Hospital in Houston and assistant professor of pathology at Baylor College of Medicine—and a co-founder of Bacterial Barcodes. “With DNA typing, they could perform the tracking much more rapidly and accurately.”

Traditionally, pulsed-field gel electrophoresis has been used to track rogue infectious agents. PFGE is a useful, if tedious, method of separating out long fragments of pathogenic DNA by administering staggered bursts of electrical currents to the gel from two different angles—a process that keeps the large molecules cohesive and intact enough to be readily identified. But manually preparing and obtaining results from gels means

it can take many hours to get reactions and up to a week to get complete results.

Bacterial Barcodes’ proprietary technique appears to be more efficient than PFGE and similar processes, thanks to several innovations. First among them was the repetitive sequence-based polymerase chain reaction technique, or rep-PCR, invented by Dr. Versalovic and co-founder James R. Lupski, MD, PhD, during the early 1990s. Dr. Lupski, professor of molecular and human genetics at Baylor College of Medicine, was examining bacterial and mammalian DNA in his Houston laboratory when he and Dr. Versalovic noticed that certain relatively short intergenic sequences of bases seemed to repeat themselves regularly, in high copy numbers, throughout the chromosomes of pathogens.

Using oligonucleotide primers with nicknames like Eric and Box, the polymerase chain reaction process could be applied to separate, mark, and amplify these unique repeated fragments of DNA; analyze them; and produce bar-code-like banding patterns in gels. The resulting graphs could then be compared with those of existing profiles of pathogenic strains.

“To our surprise, we found many copies of these conserved, interspersed, repetitive sequences in bacteria,” says Dr. Versalovic, who was then working in Dr. Lupski’s laboratory. “We had a sense that we had hit on something that might be useful in DNA typing of pathogens, and we

were very excited about that—although it did take some time to convince ourselves that what we had found was truly going to prove robust and useful.”

It did indeed. A 1999 literature review by the Wisconsin research firm Millennium Strategies reported that rep-PCR is not only the quickest but also the most cost-effective of the six most commonly used methods of typing pathogenic DNA.

Drs. Lupski and Versalovic’s microbial typing process was a time

improvement over PFGE, but still it demanded that the samples be prepared manually and results came 16 hours later. Seeking to automate and further quicken its process, Bacterial Barcodes began searching for a technological alliance—and found it in California’s Silicon Valley. In December 2001 the company incorporated a second innovation: Mountain View-based Caliper Technologies’ “lab-on-a-chip.”

The two-inch-by-two-inch chips play a critical role in speeding the analysis and making it more uniform. After the rep-PCR process is performed on the patient samples in a thermal cycler—a device that accelerates the heating and cooling cycles necessary to separate and amplify segments of pathogenic DNA—the PCR products, along with molecular weight and internal “markers,” are loaded into each lab-on-a-chip. The chips are placed into a proprietary analyzer and activated by electrical currents; the samples and fluorescent chemical reagents flow through tiny channels (each about the width of a human hair) in a sieving process that separates multiple fragments of the pathogenic DNA on the basis of their size.

A focused laser beam detects the results of these reactions within minutes and converts them into peak-profile graphs. Those data are then swiftly uploaded in digital format to a central database via the Internet, where the company’s specially designed software analyzes correlations and draws gel-like images, graphs, dendrograms, and scatter plots that can be used to visualize clonal relationships between the genetic fingerprints of the problem strain and those of other known strains.

The system implementing the lab chips cuts the total time needed for fragmentation and analysis of pathogenic DNA an additional 75 percent from that previously required by the manual rep-PCR process; its analysis equipment now takes up no more desk space than, say, a personal computer’s hard drive mini-tower.

Mimi Healy, PhD, who joined Bacterial Barcodes as chief scientific officer in February 2002, has been instrumental in the effort to develop analysis software. Thanks to her streamlining of the test protocols, results can now be obtained within three or four hours of sampling.

“PFGE is technically challenging and requires experienced operators,” says Dr. Healy. The agarose plugs and enzymatic reactions required for gel methods and the original, manual rep-PCR technique tend to cause greater variations in results, she adds. Bacterial Barcodes’ DiversiLab System, as the platform is known, has standardized kit reagents and protocols. “Detection is automated, simple, and reproducible, and the analysis is by computational methods—less subject to interpretation errors,” she says.

The system is also less expensive than other methods because labor costs are lower when using the automated laboratory chips.

The company is still working, however, to develop the rigorous interpretive standards that will enable clinical laboratory scientists to definitively say whether a problem strain is a clone of, related to, or unrelated to a known strain in an outbreak. PFGE technicians already use a hard-and-fast standard: If the gel band pattern is identical to that of a known organism, it is considered identical to the organism causing the outbreak; if there are fewer than three “band differences” between the mystery strain and a known strain, it is considered related and part of the outbreak; and if there are more than three differences in band size, it is considered unrelated to the known organism and the outbreak.

As clinical studies—most of them comparing data from the latest beta version of the automated rep-PCR technique with data derived from PFGE typing results—filter in from Johns Hopkins and other hospitals and reference labs, Bacterial Barcodes will develop stronger standards and a larger library of pathogen strain types, says Dr. Carroll of Johns Hopkins.

The company shipped its first test kit—which can be ordered for small or large laboratory setups—for Staphylococcus bacteria last January. Soon afterward came test kits for Enterococcus, Clostridium, Salmonella, Acinetobacter, and Listeria bacteria using the same principles and procedures; then, in May, kits for Streptococcus, Pseudomonas, Campylobacter, Shigella, Escherichia, Klebsiella, Serratia, and Stenotrophomonas bacteria as well as tests for two major fungal pathogens, Candida and Aspergillus.

Future versions of the DiversiLab kits may make it possible to load blood, urine, or other specimens directly into an integrated lab-on-a-chip, which would save additional time by automating the extraction, amplification, and DNA fragment separation steps in the process. Future releases of the software will speed analysis even more by going beyond the typing of pathogenic DNA to specific identification of the pathogenic strains involved in an infection.

The rep-PCR technique and kits may also be useful in such areas as food safety, air quality (to detect the presence of deleterious fungi), viticulture (to identify strains of bacteria or yeasts involved in beer and wine manufacturing), veterinary medicine, parasitology, and even bioterrorism. For example, the company has tested a kit that could be used to quickly type the DNA of mysterious spores found in an envelope as being—or not being—a strain of Bacillus anthracis.

“I believe this technology,” says Dr. Carroll, “when completely developed, has the potential to allow us to identify an organism to the strain level while simultaneously allowing the lab to tell whether the organism in question is part of the hospital clone—or a new strain introduced into that environment.”

Paul Karr is a writer in North Bergen, NJ.