May 2010
Feature Story

William Check, PhD

For most of human history, monetary systems were based on a gold standard of some kind, first circulation of gold coins, more recently a fixed exchange rate between gold bullion and the U.S. dollar. All the experts knew this was the best method. Suddenly, in 1971, the U.S. government ended direct convertibility of the dollar to gold, because it wasn’t working. A shock reverberated throughout global markets, but now a more effective arrangement is in place.

Microbiology laboratories have experienced analogous changes in the accepted gold standard for detecting toxigenic Clostridium difficile in fecal samples. In fact, three putative gold standards have been adopted and abandoned over the past 15 years. While toxigenic culture has always been acknowledged as the most accurate method for detecting pathogenic C. difficile, it is slow and cumbersome. Almost no one does it. So cell cytotoxin assay (CTA, also known as cell culture neutralization assay, or CCNA) was anointed as the surrogate gold standard. A few years ago, data showed that CTA lacks sensitivity, so testing for glutamate dehydrogenase (GDH, so-called common antigen) became the new test of choice. Now numerous rigorous evaluations show that GDH, too, lacks adequate sensitivity.

The misperception of GDH tests rested on their evaluation in terms of cytotoxin assay. “A relatively insensitive assay compared to a relatively insensitive assay looks good,” says Lance R. Peterson, MD, director of microbiology and infectious diseases research, Department of Pathology and Laboratory Medicine, Evanston (Ill.) Hospital, and associate epidemiologist, NorthShore University HealthSystem. “Compared to CTA,” he adds, “GDH is 90 percent sensitive, but both are only 55 percent to 85 percent sensitive compared to toxigenic culture.”

In practice, most laboratories do not use any of these methods. According to a 2009 CAP Survey, 95 percent rely on inexpensive, rapid enzyme immunoassays for C. difficile toxins A and B. No one would mistake these EIAs for a gold standard, but just how deficient they are has only been fully appreciated in the past few years. Recent evaluations have found sensitivity of EIA for C. difficile to be well under the target of 90 percent and often as low as 50 percent. “Microbiology labs need to get back to saying 50 percent sensitivity is not OK,” Dr. Peterson says.

Indeed, these have been dark days for C. difficile detection. As it is written in Genesis, “And the earth was without form, and void; and darkness was upon the face of the deep.” In the past few years, however, a light has shown forth across these troubled waters. It is the beacon of real-time PCR, which is gaining recognition as the long-awaited solution for accurate detection of toxigenic C. difficile. Just how much interest real-time PCR for C. difficile has generated was evident at the 2009 meeting of the Association for Molecular Pathology, at which the frequency of talks on real-time PCR for C. difficile was only slightly lower than the incidence of this pathogen among hospitalized patients.

Talks on assays for C. difficile, including PCR, will also be thick on the ground at the 2010 American Society for Microbiology meeting, says Ellen Jo Baron, PhD, D(ABMM), director of medical affairs for Cepheid and director of the clinical virology laboratory and associate director of microbiology, Stanford University Medical School, who reviewed abstracts. “There are more reports on C. difficile detection than I’ve ever seen,” Dr. Baron says. “I think people are looking for the best way to identify this pathogen.” It’s not only the quantity but also the quality that is higher this year, Dr. Baron says. “Last year many abstracts on C. difficile assays used CTA, which is only 70 percent sensitive, as the comparator. During the last year many papers showed the need to use toxigenic culture as the reference method, which is reflected in this year’s abstracts.”

Many comparative studies, published and unpublished, have evaluated the accuracy of real-time PCR for C. difficile. All come to similar conclusions. According to the authors of one study, which evaluated nine EIAs for toxins A and B, an EIA for GDH, and a commercial real-time PCR: “The optimum rapid single test was PCR for toxin B gene, as this had the highest negative predictive value” (Eastwood K, et al. J Clin Microbiol. 2009;47:3211–3217).

“We have done extensive evaluations of PCR for C. difficile in stool,” says Richard B. Thomson, PhD, medical microbiologist and director of microbiology laboratories, Evanston Hospital, a colleague of Dr. Peterson’s. “We had to see whether combinations that are less expensive than PCR, like GDH reflexed to a second test, gave the same sensitivity.” Some of this work has been published (Peterson LR, et al. Clin Infect Dis. 2007;45:1152–1160); a broader comparison was reported in March at the Fifth Decennial International Conference on Healthcare-Associated Infections. “We are now convinced that PCR is by far the most sensitive test,” Dr. Thomson says. In recent years, he notes, the focus has swung from specificity to sensitivity. Now it’s time to swing back. “To my mind, the end of the story will be an inflammatory mediator, maybe a marker of necrosis of colonic epithelium, to tell us who’s sick,” Dr. Thomson says. “That will close the circle with these very sensitive molecular tests.”

Brian Currie, MD, director of infection control and vice president and medical director for research, Montefiore Medical Center, Bronx, NY, says PCR will “revolutionize” the way patients infected with C. difficile are managed and how the pathogen is dealt with from an infection control perspective. “One of the big obstacles to adopting PCR is cost,” he notes. How to address this obstacle? “We need to look beyond the lab budget across the hospital budget.” Dr. Currie did a study of stool samples from 520 patients with diarrhea, in which EIA missed 53 of 116 positive samples. This finding “horrified” him, he says, until he realized it didn’t matter. “Clinicians have so little faith in EIA that the 53 missed cases were managed as though they were positive,” he says. Almost 90 percent of the 445 negative patients remained in isolation and 40 percent were treated. In theory, Dr. Currie says, eliminating unnecessary use of antibiotics and personal protective equipment can more than offset the added cost of PCR.

Frederick S. Nolte, PhD, D(ABMM), vice chair of laboratory medicine and director of clinical laboratories, Medical University of South Carolina, put it more bluntly in his AMP presentation: “You’re doing substandard diagnostics if you’re not doing PCR [for C. difficile].” He admitted PCR is more expensive, but added, “I think it’s worth it.”

Superiority of PCR for C. difficile has broad implications, says Ferric C. Fang, MD, director of the clinical microbiology laboratory, Harborview Medical Center, Seattle. “Molecular technology is really the moving front of new diagnostics in the microbiology lab,” Dr. Fang says.

Results of what may be the first large head-to-head comparison of the three commercial real-time PCR assays for C. difficile were presented in April at the 26th Annual Clinical Virology Symposium by Arthur E. Crist Jr., PhD, technical director/ clinical microbiologist, Department of Laboratory Services, Division of Clinical Microbiology, York (Pa.) Hospital. Using toxigenic culture plus chart review as a composite gold standard, Dr. Crist and his co­workers compared EIA, CCNA, GDH, and the three PCR assays—BD GeneOhm Cdiff Assay, Cepheid Xpert C. difficile Assay, and Gen-Probe Prodesse ProGastro Cd Assay. The most sensitive was the Cepheid assay (94.4 percent); the most specific was the BD assay (99.7 percent). However, the difference in sensitivity and specificity among the three PCR methods was not statistically significant.

While knowing the accuracy of the commercial PCR assays is helpful, Dr. Crist does not believe this is the most useful information from his work. “Even though the three PCRs may vary somewhat in sensitivity, all were better than CCNA,” he emphasized, adding, “Not too long ago CCNA was the gold standard at my hospital.” He switched in 2004 to a two-step algorithm with GDH and CCNA. He later added EIA to make a three-step algorithm. “All of these approaches are not as sensitive as PCR,” Dr. Crist says. “Short of performing toxigenic culture, which is impractical for most clinical laboratories, PCR needs to replace CCNA as a standalone or confirmatory test.”

Sensitivity and specificity are not the only factors in choosing a PCR platform. “I’m inclined to go to PCR,” Dr. Crist says. “But first we want to do a cost analysis.” Cost of equipment and reagents, hands-on time, turnaround time, batching versus random access, and ease of use are all considerations. A lab’s capital budget, test volume, and expertise with molecular technology will determine the importance of each.

Should all labs now adopt a PCR assay for C. difficile? Karen C. Carroll, MD, director of the Division of Medical Microbiology, Johns Hopkins Hospital, told CAP TODAY, “If you are a classical microbiologist trained in aseptic technique, you can learn how to use any of these kit-based [PCR] assays.” During her career Dr. Carroll has worked through all the changes in assays for C. difficile, and recently she evaluated two real-time PCRs. She adopted the BD GeneOhm platform. “You don’t need to be a molecular expert to use these assays, especially the Cepheid GeneXpert,” she says. “I don’t think that is a legitimate excuse anymore. These assays are FDA-cleared, and their accuracy is well substantiated in the literature, even outside corporate-sponsored trials.” The availability of these platforms in labs is more of an impediment than the performance of tests from a technical standpoint, Dr. Carroll says. All three platforms have multiple assays, she notes, adding, “MRSA testing in particular has become very important.”

If EIA is used, Dr. Peterson says, the lab also has to set up for CTA, which has low sensitivity, or send samples to a reference lab for molecular. “Why not do it yourself?” There are three molecular choices, “at least one of which is very simple to use,” he says, and adds: “It is hard to make a practice argument against molecular,” especially since, in his view, more accurate diagnosis and treatment could lower mortality from C. difficile below the current one percent figure.

The accuracy of any testing method for C. difficile can be enhanced by testing only appropriate specimens. In a presentation at last year’s AMP meeting, Dr. Carroll said, “Testing should be limited to patients with clinical suspicion and true diarrhea,” defined as three or more loose stools (those that take the shape of a container) in a 24-hour period. Many hospitalized patients are colonized with C. difficile but do not have diarrhea. “This is particularly important,” she says, “because PCR detects the toxin gene, not the toxin itself.” A low prevalence of disease in the tested population reduces the positive predictive value of even a highly specific test. Like many others, Dr. Carroll has implemented rejection criteria so that she tests only patients with a high pre-test probability of having clinical infection. Others agree with this approach, though Drs. Peterson and Fang point out that testing may occasionally be indicated in patients without diarrhea under special circumstances, such as in patients with ileus who have a rare and dangerous form of the disease known as toxic megacolon.

Dr. Peterson has been emphatic about testing only patients who meet the clinical criteria for C. difficile illness, Dr. Crist points out. “At this point our data support that premise,” Dr. Crist says of the chart review for one-fourth of the patients in his study. “In that subset there were no positive PCR results in patients who did not fulfill clinical criteria. All PCR-positive patients were clinically ill, not just colonized, even if they were culture negative.” Predictive value can be increased by testing only patients who meet clinical criteria for true infection, he adds.

Although C. difficile infection, or CDI, has been recognized as the cause of antibiotic-associated pseudomembrane colitis for more than 30 years (Bartlett JG, et al. Gastroenterology. 1978;75:778–782), awareness of CDI rose when surveillance data showed that its incidence and mortality in hospitalized patients had dramatically increased around 2001 (McDonald LC, et al. Emerg Infect Dis. 2006;12:409–415; Redelings MD, et al. Emerg Infect Dis. 2007;13:1417–1419). Complications of severe CDI include sepsis, shock, megacolon (in which there is no stool production), perforation, and, rarely, a fulminant course leading to colectomy or even death (Dallal RD, et al. Ann Surg. 2002;253:363–372). After rising for several years, “There is now a suggestion that the CDI epidemic may be leveling off,” said Dale Gerding, MD, associate chief of staff for research and development, Hines (Ill.) VA Hospital, in his AMP talk, “Epidemiology, Pathogenesis, and Treatment of Clostridium difficile Infection.”

That we are all carrying C. difficile in our stools is a common misperception, Dr. Gerding said. That we have it cryptically and it “pops out” when we take an antibiotic is not true. Nor is it true that it is spread by health care workers who harbor it in their stools. In reality, hospitalized patients are treated with antibiotics and then exposed to C. difficile in the hospital setting, and a minor fraction of patients develop CDI. Probably three to five inpatients are colonized asymptomatically for every one who gets CDI, Dr. Gerding said.

Coincident with the overall CDI epidemic, an epidemic of a hypervirulent strain of C. difficile, typically identified by its ribotype, 027, has occurred. C. difficile 027 causes 30 percent to 40 percent of CDI cases in North America. Its higher pathogenicity may be due to greater production of toxin or increased resistance to fluoroquin­olone antibiotics, or both. As do many toxigenic strains of C. difficile, the 027 strain produces two toxins, called A and B, of which the B toxin has been shown to be necessary and sufficient for virulence (Lyras D, et al. Nature. 2009;458:1176–1179). In addition, the hypervirulent strain produces a third toxin, binary toxin, for which the contribution to virulence has not been fully elucidated. All three commercial PCR assays detect the gene for toxin B. “We expect that 027 won’t remain the dominant strain,” Dr. Gerding said. “These things cycle.”

Treatment of the first episode of CDI is unchanged in 30 years, according to Dr. Gerding: Simply stopping offending antibiotics is effective in 25 percent of patients. If antibiotics are continued, the success of CDI treatment is less likely. Patients can be treated with metronidazole or vancomycin; both are equally effective for mild disease (90 percent to 97 percent clinical cure), while the efficacy of metronidazole is greatly reduced in severe disease (Zar FA, et al. Clin Infect Dis. 2007;45:302–307). Fidaxomicin, a new antibiotic now finishing phase three trials, is equivalent to vancomycin for initial response but has a significantly lower recurrence rate (13.3 percent versus 24 percent).

In her AMP presentation, Dr. Carroll reviewed recent evaluations of toxin A/B EIAs, including one in which none of six assays met acceptability criteria—90 percent sensitivity and a false-positive rate of less than or equal to three percent (Planche T, et al. Lancet Infect Dis. 2008;8:777–784). With evidence accumulating for the insensitivity of EIAs, Dr. Carroll told CAP TODAY, in 2004 she decided, “We’re not using EIA anymore. We were getting calls in the lab about patients who had classic signs and symptoms of C. difficile infection but were negative on EIA.” To aggravate matters, clinicians were following the standard practice of ordering “stools for C. diff × 3,” automatically submitting three stool samples for EIA. The three results could be negative or positive in any sequence. “This was very confusing for clinicians,” Dr. Carroll says. “If you have a bad test and you repeat it, it’s still a bad test. We abandoned EIA.” Dr. Peterson and a colleague conducted a similar analysis (Peterson LR, Robiscek A. Ann Intern Med. 2009;151: 176–179). Mayo Clinic researchers, too, have shown the nonutility of repeat testing (Aich­inger E, et al. J Clin Microbiol. 2008;46:3795–3797).

(Doing three stool samples for C. difficile is based on a 1995 paper by John Bartlett, MD, who played a major part in the recognition of C. difficile as the agent of antibiotic-associated colitis [Manabe Y, et al. Ann Intern Med. 1995;123:835–840]. Dr. Carroll once asked Dr. Bartlett, who until recently was chief of infectious diseases at Hopkins, about that work. “He said it was never his intention to make that a standard of care,” Dr. Carroll says. “He was trying to point out that EIA is not very sensitive.”)

After jettisoning EIA, Dr. Carroll adopted cytotoxin assay, which she used until mid-2005. It turned out to be very labor-intensive, so she adopted an EIA for GDH as an initial screen to reduce the number of samples for CTA. While sensitivity of the GDH assay in early evaluations was above 90 percent, in recent studies it has been in the 80 percent to 85 percent range. “We did a look back at our procedure when we evaluated the PCR methods,” Dr. Carroll says, “and for reasons that I don’t understand, the sensitivity of the GDH assay had dropped to the mid-80s.”

Dr. Carroll says GDH was “a nice bridge between traditional methods and before commercial PCR became available. Now that we have molecular methods, for most labs this is really the way to go.” She showed an evaluation of one lab-developed real-time PCR for C. difficile in which the assay was 86 percent sensitive compared with toxigenic culture; in the same study sensitivity was 48 percent for three EIAs and 76 percent for a commercial GDH assay (Sloan LM, et al. J Clin Microbiol. 2008;46:1996–2001). Dr. Carroll’s evaluation of the BD GeneOhm Cdiff platform found 83.6 percent sensitivity and 98.2 percent specificity relative to toxigenic culture (figures for CTA were 67.2 percent and 99.1 percent, respectively) (Stamper PD, et al. J Clin Microbiol. 2009;47:373–378). Two other studies found sensitivities of 88.5 percent and 93.9 percent for the BD GeneOhm assay (Barbut F, et al. J Clin Microbiol. 2009;47:1276–1277; Eastwood K, et al. J Clin Microbiol. 2009;47:3211–3217). “You have to be careful when you compare these data,” Dr. Carroll cautioned. “Ours was the lowest. Probably 88 percent to 94 percent is a more representative range.”

For the Gen-Probe Prodesse ProGastro Cd Assay, which uses the BioMérieux NucliSens EasyMag extractor and is done on the SmartCycler, Dr. Carroll’s group found 77.3 percent sensitivity and 99.2 percent specificity (Stamper PD, et al. J Clin Microbiol. 2009;47:3846–3850). The package insert for the Prodesse assay cites a sensitivity of 91.7 percent, but that is using CTA as the reference.

Dr. Carroll showed unpublished data from the U.S. beta trial of the GeneXpert in which sensitivity was 95.9 percent and specificity 96.6 percent. Very similar results were obtained in a study conducted at seven Kaiser Permanente sites—94.4 percent sensitivity and 96.3 percent specificity (Novak-Weekley SM, et al. J Clin Microbiol. 2010;48:889–893). Both trials used toxigenic culture as the reference. Across all studies, GeneXpert consistently has the highest sensitivity and the lowest specificity of the three PCRs, which agrees with what Dr. Crist found by direct comparison.

In the Kaiser Permanente study, 13 specimens were positive by PCR and negative by culture. When these discrepancies were adjudicated, a striking finding fell out: Seven of the 13 were positive by at least two other methods (toxin EIA, GDH, or CCNA) and thus were probably true positive. In addition to leaving PCR with a corrected sensitivity and specificity of 95.1 percent and 99.4 percent, these results show that PCR is the first method that is so sensitive that it can give the lie to toxigenic culture. Perhaps the oldest gold standard for C. difficile testing is about to fall.

Comparing turnaround times, Dr. Carroll showed 45 minutes for GeneXpert, with two-minute specimen processing; 75 to 90 minutes for BD GeneOhm; and three hours for Gen-Probe Prodesse. Cost-per-test is highest for the Cepheid platform. Dr. Carroll has a four-slot GeneXpert. “It’s not very practical for a high-throughput lab,” she says. “Larger, more-automated versions are available, but they are very expensive.” Dr. Carroll adopted the BD GeneOhm assay. “We participated in the clinical trial, so we were familiar with it,” she explains. “And we already had several SmartCyclers.” Before going live, Dr. Carroll sent an announcement on the hospital Web site that is a model of clear communication (http://apps.pathology.jhu.edu/blogs/pathology/2010-ushers-in-changes-to-clostridium-difficile-testing).

To understand why experienced laboratory professionals give greater credence to studies like the one by Novak-Weekley and colleagues at Kaiser Permanente that use toxigenic culture as the reference method, it helps to appreciate the painstaking work that lies behind the phrase “toxigenic culture.” In an interview with CAP TODAY, Dr. Baron elaborated on the method used in the Kaiser study, which Cepheid sponsored. Dr. Crist’s study, as well as those of Drs. Peterson and Thomson and Dr. Carroll’s evaluations of the BD Gene­Ohm and Gen-Probe Prodesse platforms, used the same techniques. For toxigenic culture, stool samples are treated to enrich for spores by killing off vegetative cells with alcohol or with “heat shock” (heating the stool suspension to 80°C), after which the treated suspension is inoculated onto cycloserine-cefoxitin-fructose agar (CCFA) and incubated anaerobically for two to five days. Colonies are verified to be C. difficile by standard methods (gas-liquid chromatography or biochemical tests), and confirmed isolates are grown in anaerobic chopped meat-glucose broth for two days, then tested for toxigenicity by CTA. In all three studies stool samples were also inoculated onto enriched Brucella agar if no colonies grew on CCFA. (This description also explains why toxigenic culture is used only in a research context.)

Using 674 stool samples submitted for C. difficile testing, and considering any positive culture and chart review as the gold standard, Dr. Peterson’s group reported at the Decennial meeting that the sensitivity of four EIAs ranged from 40 percent to 50 percent. Sensitivity was 81 percent and 88 percent for two GDH assays and 54 percent for CTA. “It appears none of these tests should be considered as reliable assays in the diagnosis of Cdiff infection,” the investigators concluded. On the same samples, their in-house PCR assay had a sensitivity of 74 percent, compared with 93.8 percent for the BD GeneOhm. Both assays gave results between two and four hours, while culture took from five to 10 days. They concluded: “PCR assays are fast, reliable, and accurate and can be very useful for patient management and infection control practice.”

“We will change to PCR,” Dr. Thomson told CAP TODAY. “It is just a matter of pricing and which test we choose”—BD GeneOhm or Cepheid. Dr. Peterson adds: “Gene­Xpert effectively provides point-of-service testing. It would allow us to do this test at each of our four hospitals.” Like everyone else, they are struggling with Gene­Xpert’s high cost.

Dr. Crist’s study was performed on 400 loose or watery stool samples. As in other contemporary work, sensitivity for toxin A/B EIA was quite low, 55.6 percent. For CTA it was also low, 75 percent. Among PCRs, sensitivity was highest for the Gene­Xpert, 94.4 percent, while specificity was highest for BD GeneOhm, 99.7 percent, which had a sensitivity of 87.5 percent. Sensitivity was lower for the Gen-Probe Prodesse platform, 83.3 percent. Now Dr. Crist must choose. He already has SmartCyclers for MRSA and flu testing, so capital equipment cost is a factor. “I like the GeneXpert technology, and that assay has the least hands-on time and is easiest to do,” Dr. Crist says. “But it is the most expensive. For me to do the GeneXpert, I either have to buy the equipment or do reagent-rental.” He would need a 16-compartment unit, which is in the ballpark of $160,000, he says, adding that he’s buying very little capital equipment at this time. Of the rental option, he says, “Reagent cost for that assay is high.” Dr. Crist is leaning toward the BD Gene­Ohm, “mainly because I have the instruments and we batch almost everything.”

Like Dr. Carroll, Dr. Fang of Seattle’s Harborview Medical Center has accommodated to the many changes in C. difficile testing during his career. “For years experts said the cytotoxin assay was the gold standard,” he says. “But the fact is that one can see a patient with C. difficile and illness and a negative cytotoxin assay.” In Dr. Fang’s experience, CTA is only about 70 percent sensitive. “When labs across the country switched over to EIA,” Dr. Fang says, “that aggravated the problem. Cell cytotoxicity was replaced by an even less sensitive assay. Conventional EIA misses half of cases. This created a catastrophic situation in clinical microbiology.”

Dr. Fang developed a PCR assay for the C. difficile toxin B gene and found it to be superior to the two-step toxin/GDH EIA that he had been using (Larson AM, et al. J Clin Microbiol. 2010;48:124–130). “We saw a major deficiency in the GDH assay,” he says. “It was not as sensitive as we believed. It may miss 15 percent to 20 percent of positive specimens. We learned that the best approach is to do PCR on every specimen.” That is what he now recommends.

Dr. Fang has replaced his lab-developed PCR with the Cepheid assay. “With our PCR, we had to do manual extraction from stool using magnetic beads, which is time-consuming,” he says. GeneXpert uses a sonication step to break spores. Dr. Fang finds the GeneXpert assay rapid and convenient. “Because it is easy for techs, we can provide lab support seven days a week and several times per day,” he says. “What I don’t like is that it’s rather costly. But we did a cost-benefit analysis, and we believe the short turnaround time, convenience, and extra cases detected outweigh the disadvantage of high reagent costs.” Using the four-cartridge unit and doing about 250 assays per month, Dr. Fang’s lab has a reagent-rental agreement with a cost of $42 per assay.

At BD GeneOhm’s AMP workshop, Dr. Nolte described how he implemented the changeover to PCR for C. difficile. He had evaluated BD GeneOhm’s PCR and found it clearly superior to both GDH and toxin EIA, the lab’s existing method. In the study, samples with discrepant results on PCR and toxin EIA were sent to a reference lab for toxigenic culture when possible. Of 20 discrepant samples, sufficient volume was available for 11; all were resolved in favor of PCR.

In May 2009, Dr. Nolte sent a memorandum around the hospital saying the BD GeneOhm would become the standard assay for C. difficile. The switch from EIA to PCR, especially moving the test from microbiology to molecular pathology, required changes in lab operation. “The molecular pathology lab had to move to weekend and holiday coverage,” Dr. Nolte said. Also, new rules were initiated. Samples are now limited to specimens from patients with at least three liquid stools per day for one or two days, only one test per patient per week is allowed, and requests for tests of cure are denied. However, exceptions can be made in appropriate clinical scenarios with approval of the laboratory directer. Before the changeover, Dr. Nolte spoke to the GI, infection control, infectious disease, and nursing services. “Early involvement of stakeholders made changes of test method and utilization easy,” he said. “I actually got applause when I spoke to nursing about the changes.” When physicians order “C. diff × 3,” nurses are the ones who collect the specimens. While some money was saved by reducing the number of tests, switching to PCR still increased overall cost.

Dr. Nolte told CAP TODAY that, when he gave a talk on this topic recently at a meeting of laboratorians from community hospitals, all of those who were testing for C. difficile were still doing EIA. “They are convinced they can’t switch to PCR,” he says. “They believe it is too expensive and too difficult. It is a paradigm shift for them to do PCR, and they are having a hard time making it.” In his view, sample preparation with real-time PCR is simple and “within anybody’s capability.” To reduce cost, some labs may use GDH as a screening test and PCR only for confirmation. “Why put an insensitive screening test in front of a very sensitive confirmatory test?” a puzzled Dr. Nolte asks. “It makes no sense.”

Timothy S. Uphoff, PhD, D(ABMG), MT(ASCP)CM, section head of the molecular pathology laboratory at Marshfield (Wis.) Labs, chose the Gen-Probe Prodesse PCR for C. difficile detection. In a 60-sample comparison, Dr. Uphoff says, the three real-time PCRs were “essentially equal.” Dr. Uphoff’s laboratory had been using a two-tier GDH/toxin EIA protocol for C. difficile. Forty of the 60 samples were negative by this method; three of these were positive on all three PCRs.

Dr. Uphoff says one reason for adopting the Gen-Probe Prodesse PCR was that the lab had experience with other Prodesse assays. Gen-Probe Prodesse also had 100-fold higher analytic sensitivity (when testing two different C. difficile clinical strains) and a lower inhibition rate than the other two PCRs. Marshfield is using the GeneXpert on weekends, after hours, and for hospital admissions, Dr. Uphoff says, explaining that the GeneXpert’s moderate complexity allows them to use it in the 24/7 lab where they don’t have molecular specialists. “So we don’t need to keep the molecular lab open all the time.” The need for on-demand MRSA testing drove the purchase of GeneXpert, Dr. Uphoff says. It required a large capital outlay and a costly service contract. “We’ve struggled with that,” he says. “We can’t justify those costs if we only run a small number of tests per year. We are trying to maximize our utilization of that major expense.”

At New York’s North Shore-Long Island Jewish Health System Laboratories, which does about 45,000 C. difficile assays per year, Christine C. Ginocchio, PhD, senior director of the Division of Infectious Disease Diagnostics, has adopted the BD GeneOhm platform and automated it on a Hamilton instrument. Few labs can take technology that far. However, Dr. Ginocchio says, “I don’t believe there is any reason why any size lab can’t move to molecular testing. I don’t think small labs can use the argument that ‘The technology is too difficult’ anymore.” She points to several molecular platforms that are already available or in development, some of which are single-cartridge units—GeneXpert, HandyLab, Idaho Technology, Nanosphere, and Verigene among them. “That’s what the field is going toward,” Dr. Ginocchio says, “new technology that will make molecular very simple. And it’s all due to real-time PCR.”