College of American Pathologists
Printable Version

  Welcoming resistance tests,
  old and new


cap today

May 2004
Cover Story

William Check, PhD


When VCRs were introduced, they generated tremendous excitement, to the point that motion picture companies feared video cassettes would ruin the movie business. But movie rentals have become a booming business at the same time that attendance at movie theaters continues to climb. Perhaps this scenario offers us a view of the future of molecular testing for antibiotic resistance in bacteria.

PCR-based tests for antibiotic resistance are generating excitement because of their clear advantages. Performed directly on patient specimens or on isolates grown in blood culture bottles or colonies from plates, molecular tests are rapid and in some cases more accurate, making it possible to more effectively manage in-hospital infections and reduce expense. Yet any proposals that they might extensively replace traditional culture methods for antibiotic susceptibility testing must face difficult realities.

"The problem with detecting methicillin-resistant Staphylococcus aureus [MRSA] by conventional culture methods has been turnaround time," says Franklin Cockerill III, MD, Ann and Leo Markin professor of microbiology and medicine, director of bacteriology, and chair of microbiology at the Mayo Clinic and Mayo College of Medicine. "For vancomycin-resistant enterococci [VRE], there are problems with both turnaround time and sensitivity." At Mayo, 48 hours or more are frequently required to get results of MRSA testing to health care providers. Even longer time periods—often 72 hours or more—are necessary to detect carriers for VRE. And that is with 24/7 operation.

"If you are screening for MRSA or VRE and the laboratory doesn’t tell you a patient is a carrier for two, three, or more days, you can imagine the problems with infection control," Dr. Cockerill says. To address such problems, his group has developed, in collaboration with Roche Diagnostics, rapid real-time PCR tests for VRE and MRSA screening.

"In the last couple of years, we have seen the development of real-time PCR tests, the first extremely rapid assays, for detecting both bacterial pathogens and antibiotic resistance," says Lance Peterson, MD, director of microbiology and infectious disease research and a consultant in microbiology and infectious diseases at Evanston (Ill.) Northwestern Healthcare, where PCR assays for MRSA and VRE have been developed. "If you are doing these tests continually, from the time a specimen reaches the laboratory it is possible to get out a result in about two hours," says Dr. Peterson, who is also professor of pathology and medicine at Northwestern Universitys Feinberg School of Medicine, Chicago.

With conventional antibiotic susceptibility testing, or AST, patients with infections suspected to be caused by MRSA may be put into isolation for three to four days while awaiting a test result. With a same-day result provided by real-time PCR, the patient can be removed from isolation after one day or even kept out of isolation altogether. Dr. Peterson points out that the daily cost of isolation is usually more than the cost of the molecular test.

Rapid tests for MRSA and VRE may make it feasible for more laboratories to implement active surveillance for these organisms, which has been done to great effect in some Western European countries for several years and which was advocated in the recent guidelines of the Society for Healthcare Epidemiology of America (Muto CA, et al. Infect Control Hosp Epidemiol. 2003;24:362-386).

Because of the great interest in using molecular methods for AST, Karen Weck, MD, organized a workshop on this topic at the 2003 meeting of the Association for Molecular Pathology. "I wanted to give our members a chance to look at the state of molecular testing and its applications and to have useful debate and discussion," says Dr. Weck, assistant professor of pathology and assistant director of the molecular diagnostics laboratory at the University of Pittsburgh Medical Center. She adds, "It’s clear that there is a greater cost and greater length of stay [LOS] in patients who are infected with resistant organisms, specifically MRSA and VRE. So it’s a logical leap of faith that identifying these organisms faster could have an impact on hospital costs and LOS." But studies to prove these benefits have not yet been done thoroughly, she notes. "To translate faster results into action is going to be a challenge not only for clinical laboratorians but for the clinicians using these results," she says.

Gary Procop, MD, section head of clinical microbiology at the Cleveland Clinic Foundation, has also developed molecular assays for antibiotic resistance. To justify the cost of these more expensive assays, he agrees, it will be necessary to show that they change clinicians’ actions. He has planned studies to investigate this, but he’s had difficulty performing them. "When clinicians have a test like PCR that seems to be intuitively advanced, they just want to implement it," he has found. "They don’t want to do long economic studies."

One of the few studies of this kind was carried out at Evanston Northwestern. According to Richard Thomson Jr., PhD, one of Dr. Peterson’s collaborators, the data showed that the rapid result of methicillin susceptibility testing on S. aureus by real-time PCR was looked at and acted upon by physicians in conjunction with the antibiotic pharmacist more rapidly than results of conventional AST. "Doing real-time PCR rather than conventional testing reduced the time to significant clinical action," says Dr. Thomson, director of microbiology at Evanston Northwestern and professor of pathology at Northwestern University’s Feinberg School of Medicine.

For most clinically significant bacteria, molecular AST is currently not feasible because of the complex resistance mechanisms many pathogens generate. Fred Tenover, PhD, associate director for laboratory science in the Division of Health Care Quality Promotion at the Centers for Disease Control and Prevention, acknowledges the molecular success stories with MRSA and VRE. "Dr. Peterson and his colleagues at Northwestern have done some very nicely designed studies showing that you can use a PCR assay on material from a rectal swab to find out in a fairly rapid way which patients might be colonized with VRE," he says. Dr. Tenover cites multidrug-resistant Mycobacterium tuberculosis as another instance in which molecular AST is useful. With these few exceptions, however, phenotypic methods of AST still predominate. Dr. Tenover’s perspective on why this is so comes from years of writing the chapter on molecular detection methods for resistance in the American Society for Microbiology’s Manual of Clinical Microbiology. That chapter now lists about four pages of PCR assays that are potentially available to screen clinical samples for "a whole host" of resistance genes. Unfortunately, Dr. Tenover says, "the practicality of that has never been realized, partly, I believe, because bacteria are always generating new mechanisms of resistance."

A further benefit of conventional AST is that it reveals not only whether an isolate is resistant to a specific agent, but also which other agents it is susceptible to. "How do you make a molecular test that gives you the same information as a Kirby-Bauer or MIC test?" asked Frederick Nolte, PhD, D(ABMM), F(AAM), professor of pathology and laboratory medicine at Emory University School of Medicine, Atlanta, speaking at the AMP workshop. "It is hard for me to see how molecular methods will entirely replace culture because of the critical role of culture in susceptibility testing."

For the foreseeable future, then, clinical microbiology laboratory directors will be deciding not whether to replace conventional AST wholesale with molecular assays, but selectively choosing which molecular assays to bring on board and how to incorporate these two complementary methods into optimal protocols. In many cases a rapid molecular test will be used for an initial result while slower culture-based methods are perking in the background, just as many people see a first-run movie in the theater, then enjoy it later on DVD at home.

At the AMP workshop, Dr. Nolte set the stage for talking about molecular resistance tests by discussing conventional methods for detecting MRSA and VRE, particularly with regard to turnaround time. In July 2003, among 137 positive blood cultures for S. aureus from 40 patients at Emory Medical Laboratories, the average transport time to the laboratory was 2.5 hours, and the average time to positive results using an automated instrument was 18.7 hours. "This is a window we are not going to be able to touch with current molecular methods," Dr. Nolte said. Molecular tests for S. aureus are not now sensitive enough for direct testing of blood; the number of organisms in a bacteremia can be low, less than one organism/mL in adults.

After the presence of gram-positive cocci in clusters is verified and S. aureus is identified, susceptibility can be done in any one of several ways:

  • Disk diffusion or MIC.
  • Automated broth dilution methods.
  • Penicillin binding protein 2a (pbp2a) rapid latex agglutination test on isolated colonies.
  • Peptide nucleic acid (PNA) FISH probe on fixed smears to identify S. aureus followed by susceptibility testing by either molecular or conventional methods.
  • PCR on isolated colonies or directly on a positive blood culture.

A conservative estimate for detecting MRSA by conventional methods reveals a turnaround time of 38 to 62 hours, compared with 24 to 44 hours by PCR. For VRE, conventional identification and antibiotic susceptibility testing take 40 to 60 hours, while real-time PCR directly from stool or perirectal swab takes two to four hours. Dr. Nolte emphasized the work that would be involved in doing recommended surveillance for these two organisms by conventional methods. "Imagine what your practice might look like if these policies were instituted and you were processing hundreds of cultures each day," he said.

At the AMP workshop, Dr. Procop presented data on real-time PCR for detecting MRSA, and PNA FISH for detecting S. aureus. One complication is that coagulase-negative staphylococci can also carry the mecA gene and these organisms (for example, S. epidermidis) are present as normal flora contaminants in samples from surgical wound sites and are often present in the nares of hospitalized patients. Therefore, any assay for MRSA must identify the species as well as the resistance gene. Dr. Procop examined a real-time assay for LightCycler that incorporates primers and probes recommended by Reischl, targeting mecA and a 442-base segment unique to S. aureus (Reischl U, et al.J Clin Microbiol. 2000; 38: 2429-2433). Using this assay, Dr. Procop and his colleagues looked at 100 consecutive samples of gram-positive cocci in clusters from BacT/Alert blood culture bottles. In the 34 S. aureus isolates, 24 of which were methicillin-resistant by oxacillin susceptibility testing, the real-time PCR assay had 100 percent concordance with standard AST and was 18 to 36 hours faster (Shrestha NK, et al. J Clin Microbiol. 2002; 40: 2659-2661). Accuracy for coagulase-negative staphylococci was much lower and not deemed adequate for clinical use.

At the request of the institution’s surgeons, the SA442 assay was implemented for the preoperative detection of S. aureus colonization. Nasal swabs and real-time PCR were used for selecting patients for mupirocin treatment. It was found to be acceptably accurate and cost-effective (Shrestha NK, et al. Infect Control Hosp Epidemiol. 2003; 24: 327-333).

PNA FISH, which takes 2.5 hours to perform, has also been used successfully for detecting S. aureus on smears made directly from positive blood culture bottles growing gram-positive cocci in clusters (Jansen GJ, et al. J Clin Microbiol. 2000; 38: 814-817; Chapin K, et al. J Clin Microbiol. 2003; 41: 4324-4327). Dr. Procop took part in a collaborative evaluation of this assay that achieved 100 percent sensitivity and 99 percent specificity (Oliveira K, et al. J Clin Microbiol. 2003; 41: 889-891). "The advantage of PNA FISH is that you don’t have to do mecA gene determination on every positive blood culture that demonstrates gram-positive cocci in clusters," Dr. Procop says. "You can reserve mecA testing for the S. aureus-positive specimens which you know are clinically significant."

A similar real-time PCR assay has been developed jointly by the microbiology and molecular diagnostics laboratories at Evanston Northwestern, where it is in use. It was developed under the guidance of Kathy Mangold, PhD, research scientist in the molecular diagnostics laboratory at ENH, research assistant professor in pathology at Northwestern University’s Feinberg School of Medicine, and a colleague of Drs. Peterson and Thomson. It uses probes to the mecA gene and to the specific S. aureus version of the femA gene complex. Initial evaluation showed less than 100 percent sensitivity, due to a problem with reagents getting inside the cell. "A 10-minute enzyme digestion with achromopeptidase solved that problem," Dr. Mangold says. They also run blood specimens over a Gentra Systems Generation Capture column to remove interfering heme.

Samples they test are broth from positive blood culture bottles and colonies growing from clinical specimens rather than tissues tested directly, because of the risk of contamination by mixed flora in direct samples. When the assay is tested on nasal swabs, Dr. Mangold notes, occasional cases occur in which a patient has methicillin-resistant S. epidermidis (MRSE) and methicillin-sensitive S. aureus (MSSA), which the test can call MRSA. femA-positive, mecA-positive samples are now streaked on plates for verification.

To validate the assay clinically, it was compared with conventional AST on 51 colonies grown from blood cultures and 130 colonies grown from wound or sputum cultures. "There were only two disagreements," Dr. Thomson says. "In both those cases, further testing showed that the molecular test was correct and the conventional test was false-negative." He adds, "According to the conventional test result, both those patients would have been taken off vancomycin." The assay has been validated also on positive blood culture broths containing gram-positive cocci growing in clusters.

The next questions were how much earlier the real-time PCR result comes out and whether having a more rapid result is of clinical value. "We wanted to see if the result was acted on by clinicians," Dr. Thomson says. "If we told them it was methicillin-resistant, and the patient was not on the right drug, did they change it? And how long did it take?"

As expected, real-time PCR was much faster. From the time the blood culture instrument gave a positive signal until the methicillin susceptibility report was available was 42.5 hours by conventional AST and 24.9 hours doing the molecular test on colonies from the blood culture. If the molecular test was done directly on the blood culture bottle, the susceptibility result was available in 12.2 hours.

Their data also showed that clinicians acted on results from both modes of testing. An appropriate antibiotic change-from vancomycin to nafcillin-was made in the 14 cases in which the isolate was found to be methicillin—sensitive by conventional AST and in the 17 cases found to be methicillin—sensitive by PCR. In two cases a MRSA was being treated with nafcillin, which was switched to vancomycin. (PCR results were reported as "generated by molecular method," showing that this method is accepted.)

In addition, the Evanston Northwestern group measured the time from when the methicillin report went out to when an order for antibiotic change was put through. With the conventional approach, the result went into the computer as usual and immediately became available to the clinician. It was also reported in pharmacy, where the antibiotic pharmacist periodically picked up the reports and compared them with the charts to see that the appropriate antibiotic was being used. It took 15.7 hours from the time a report went into the computer until the pharmacy received a request for change from the physician.

With real-time PCR testing, the group added what Dr. Thomson calls "kind of a wrinkle." When the mecA result came back, instead of just putting it into the computer and having it print in the pharmacy office, they also called the antibiotic pharmacist to alert her to the report. "We assumed that if we hadn’t called, the time to when the antibiotic was changed would have been comparable with PCR and conventional testing," Dr. Thomson says. So this was a measure not of the type of test but of the phone call. With this new step, the time from when the pharmacy was called to when the clinician’s request arrived at pharmacy dropped substantially, to 1.9 hours.

Molecular testing for MRSA is now done once daily on weekdays. "As the number of tests goes up, we can do it more often," Dr. Thomson says.

"A whole program of surveillance and enhanced infection control is made possible by rapid testing," Dr. Peterson points out. At Evanston Northwestern, this program encompasses real-time PCR testing of nasal swabs for S. aureus on patients going into elective orthopedic surgery to have devices implanted. Patients who have a positive nasal swab get a molecular test for high-level mupirocin resistance before mupirocin decolonization. Nasal swabs for MRSA are also done on adult and infant ICU patients, and those with MRSA are put into isolation.

Dr. Peterson says, "We have good data from the neonatal ICU showing that screening reduces morbidity and mortality and pays for itself about three times over. Our numbers indicate that for every dollar spent on testing and enhanced infection control, you reduce nonreimbursable costs—which are most interesting to administration—by $3 to $5."

In addition to in-lab—developed molecular assays for MRSA, the first rapid real-time molecular assay was cleared by the Food and Drug Administration in March. It is marketed by Infectio Diagnostics, is done directly on nasal swabs, and gives a result in less than one hour. "This test is really focused on infection control applications," Dr. Tenover says. "It helps make decisions about which patients might be put into isolation immediately on admittance to the hospital or about the reverse decision—who you might take out of isolation if they are not colonized." It is performed on Cepheid’s SmartCycler and differentiates between S. aureus and coagulase-negative staphylococci.

Says Mayo’s Dr. Cockerill, "Infectio’s test is appealing because it allows you to do a direct test in a very short time." It is based on simultaneously detecting the mecA gene and the unique point where it inserts into the S. aureus chromosome, he explains. This approach potentially solves the problem of determining whether the mecA gene is in a S. aureus cell versus a coagulase-negative staphylococcus cell. "Their sensitivity looks good and the assay has performed well in clinical trials," Dr. Cockerill says.

However, he adds one note of caution. "A potential drawback is that not all strains of MRSA may be identified by this method. This is because an increasing number of different insertions of mecA into the S. aureus chromosome are being described, which may not be covered by the test. Also, there may exist other MRSA strains yet undiscovered for which the mecA insertion into the chromosome is also different." Additional experience with this test will ensure that it provides a comprehensive approach toward detecting all MRSA strains, he says.

Dr. Cockerill’s research group is also working to develop an ultrarapid assay for direct detection of MRSA.

James Versalovic, MD, PhD, director of microbiology laboratories at Texas Children’s Hospital and assistant professor of pathology at Baylor College of Medicine, Houston, is using an unusual molecular approach for investigating nosocomial infections, including MRSA: conventional PCR combined with microfluidics-based separation of DNA fragments on a small chip. The DNA separation chip, marketed by Agilent, is "much easier to use than pulsed-field gel electrophoresis for typing of bacterial pathogens in suspected outbreaks," Dr. Versalovic says. Resistant organisms can be distinguished by comparisons of DNA fragment patterns, which are "quite distinct," Dr. Versalovic says. Resistance may be confirmed by conventional culture-based methods.

Despite the excellent performance of molecular assays for MRSA, they may not be for all laboratories. Asked what method he has chosen at Emory, Dr. Nolte says, "We have just made the switch to the rapid latex agglutination test for pbp2a on colonies of S. aureus from all body sites." He expects it to save at least 24 hours for MRSA and to be a tossup compared with PCR. "Studies say they give you the same answer and latex agglutination is easier to do, probably cheaper, and easily accomplished in a traditional microbiology laboratory," he says. The main limitation is that you can’t do it directly on a positive blood culture bottle or on patient specimens. It requires isolated colonies, whereas mecA PCR can be done directly on positive blood cultures. On the other hand, PCR is expensive and labor-intensive.

At the Providence Portland (Ore.) Medical Center, a molecular assay has not yet been adopted, says Ronald Dworkin, MD, medical director of infectious diseases and of the molecular diagnostics laboratory. His routine microbiology laboratory is doing the pbp2a test for MRSA. "We do it from culture plates because it needs a substantial inoculum," he says. "Our molecular laboratory is considering whether to proceed using PCR. We would like a more rapid answer than waiting 24 hours. During that time, we can use vancomycin for S. aureus that we suspect might be methicillin-resistant. On the other hand, we are trying to reduce use of empiric vancomycin because of its effect on hospital flora and to reduce selection of VRE."

Dr. Dworkin notes the advantages of PCR: "Theoretically it would be more sensitive and maybe could be done directly off wound swabs." However, in practice the hospital laboratory doesn’t have trained technologists to do PCR assays around the clock. "To do that would change how we do molecular testing," he says. "We pretty much would have to mainstream it into our routine microbiology laboratory. You need to set up the PCR assay continuously to take advantage of its shorter turnaround time. Doing it twice a day wouldn’t be any faster than doing the antigen test."

At the University of California, Los Angeles, molecular detection of MRSA is done only for research projects, says Janet Hindler, MS, senior specialist in clinical microbiology in the Department of Pathology and Laboratory Medicine. "Where a laboratory is doing PCR for other pathogens, such as viruses, then it is easy to set up PCR for mecA," she says. "But you generally wouldn’t set it up for just one test. It wouldn’t be practical."

Turning to molecular assays for vancomycin-resistant enterococci, it has been shown that active surveillance can reduce or eliminate the transmission of VRE within an institution (Ostrowsky BE, et al. N Engl J Med. 2001; 344:1427-1433). Dr. Cockerill’s group developed a real-time PCR assay for VRE that can enhance surveillance: It shortens time to results by about 24 hours when performed on colonies. When done directly on stool samples or perirectal swabs, it allows same-day TAT. "Doing the assay directly on stool is pretty challenging, particularly with respect to inhibition of the PCR reaction," Dr. Cockerill says. Using STAR buffer and extracting DNA before the amplification reaction reduces inhibition to about one percent, he reports. Extraction and real-time PCR take 3.5 hours.

Dr. Cockerill’s group compared the accuracy of its molecular assay done directly on swabs to the standard VRE enterococcosel agar plate. Compared with plates containing 8 µg/mL of vancomycin, sensitivity was 100 percent and specificity 95 percent. When discordant results were adjudicated by other tests, 71 percent of "false-positive" PCR results had another positive result. Dr. Cockerill raises the question of whether the molecular test is the new gold standard. Other studies have found that the standard plate method for VRE misses about half of VRE colonization cases, Dr. Cockerill says (Palladino S, et al. J Clin Microbiol. 2003; 41: 2483-2486). "We are confident that what we are finding is real and that the plate method is just not sensitive enough," he concludes. (His work is scheduled for publication in the June 2004 issue of Journal of Clinical Microbiology.)

While the rapid molecular assay for VRE is ideal for universal surveillance, at Mayo it is still restricted to high-risk areas, such as ICU and organ transplantation, Dr. Cockerill says. "Is that the best approach?" he asks. "We don’t know. If we want to get our VRE colonization rates down to one percent, maybe we need to do broad-based entry screening as they do in Europe. But who is going to pay for it? And is it cost-effective?"

Dr. Peterson and his coworkers have also developed a PCR assay for VRE done directly on perianal or perirectal swabs that is more sensitive for detecting VRE colonization than standard culture methods (Paule SM, et al. J Clin Microbiol. 2003; 41: 4805-4807). "We showed that if you use PCR for screening admissions that were previously positive for VRE, you can very accurately tell who is and is not colonized and who needs to stay in isolation. Active surveillance was effective in reducing VRE bacteremia rates," he says (Price CS, et al. Clin Infect Dis. 2003; 3: 921-928).

In further work with the molecular assay, Dr. Peterson and colleagues showed that increased screening for VRE picked up all patients carrying it on admission, reduced the number of patients infected, shortened length of stay, and reduced overall hospital costs for each hospital admission by about $500. Among three models—screening patients on high-risk units, screening patients with impaired kidney function, or screening everyone who had been in a hospital or nursing home in the prior two years—the last option, in which about 30 percent of admissions were screened, was overall most cost-effective. "This was because you captured everybody with VRE," Dr. Peterson says. (This work has been accepted by Infection Control and Hospital Epidemiology.)

Putting molecular assays for resistance into context, Dr. Procop says: "I think that rapid methods are extremely useful to provide important information on single genes like mecA for MRSA and vanA/ vanB for VRE. After that it gets much more difficult." He predicts it will be a while before molecular tests can replace phenotypic testing, because of complexity and cost. "Right now when we do a susceptibility panel on a tray, we get answers from multiple drugs. Each of those would have to be a separate molecular test, each of which is likely to be more expensive than the entire panel," Dr. Procop says. These issues may be answered in the future through multiplex PCR reactions and limited gene chip microarrays, he speculates.

For the CDC’s Dr. Tenover, complex resistance mechanisms explain why molecular tests have not—and probably will not—become predominant. "It was thought for many years that one of the major modes of resistance to aminoglycosides in Pseudomonas aeruginosa was a permeability barrier," he says. "It turns out that Pseudomonas has several dozen efflux pumps to keep antibiotics from reaching their site of action." Genes for efflux pumps are difficult to assay at the molecular level because you would not be looking for a mutation. Efflux genes are always present but may be in either a constitutive or induced state. "In the early ’80s we talked a lot about multiplex PCR reactions to detect most common resistance genes, but in fact I think we were rather naive at that time," Dr. Tenover says.

Dr. Peterson weighs in with a visionary point of view: "I think there will be over the next eight to 10 years a whole host of tests to help with detection of multidrug resistance and to help distinguish patients with and without infections when the clinical picture looks the same. These tests will be key to managing drug resistance."

William Check is a medical writer in Wilmette, Ill.