Clostridium difficile Testing: after 20 Years, Still Challenging

More than 20 years ago, as Clostridium difficile was being established as the cause of pseudomembranous colitis and antibiotic-associated diarrhea (AAD), many clinical laboratories were using or beginning to use cycloserine-cefoxitin-fructose agar, a selective medium developed by Lance George and colleagues (4) as an in vitro diagnostic aid for C. difficile disease. Isolates were presumptively identified by their colony and cellular morphologies, fluorescence, volatile fatty acid profiles, and other properties. In addition to detection by bacterial culture, the toxins of the organism were detected by tissue culture assay with human cell monolayers. Toxin identification was confirmed with C. difficile antitoxin (available from the Virginia Polytechnic Institute Anaerobe Laboratory) or antitoxin against Clostridium sordellii, which produces the cross-reacting toxins known as lethal toxin (LT) and hemorrhagic toxin. The tissue culture assay was not highly standardized, which resulted in variations in the most appropriate cell line to use, optimal fecal dilution, and interpretation of the cytotoxic or cell-rounding activity (e.g., does 25% cell rounding constitute a positive reaction?). These differences made it challenging to directly compare results among laboratories or even to obtain reproducible results within one laboratory. Twenty years later, tissue culture is still used as a diagnostic aid. Bacterial culture, for the most part, is not. In general, many clinical laboratories have moved to antibody-based tests (enzyme-linked immunosorbent assays [ELISAs] and lateral-flow or flowthrough devices) because the tests are more cost-effective and offer turnaround times of minutes to hours instead of the 2 to 3 days required for bacterial and tissue cultures. In addition, many of the antibody-based tests are simpler and easier to perform.

Now there are more than a dozen commercial antibody-based tests for C. difficile on the market. These tests target toxin A, a combination of toxins A and B, or the common antigen. The common antigen, which is glutamate dehydrogenase (GDH), was so named because studies in the mid-1980s showed it to be commonly produced and conserved by both toxigenic and nontoxigenic isolates. The antigen was actually identified as GDH based on its gene sequence, its homology with other GDH enzymes, and its enzyme activity. The first antibody-based C. difficile diagnostic on the market, the Culturette brand rapid latex test for C. difficile toxin A, turned out instead to be a latex agglutination test for the common antigen.

Because of the increased monitoring of patients for C. difficile, several important observations can be made about this organism's role in human disease and its effect on health care. This organism is now recognized as the primary cause of hospital-acquired colitis in patients who receive antibiotics, chemotherapeutics, or other drugs that alter their normal flora. In most industrialized countries, C. difficile is now the first organism suspected by health care personnel when a hospitalized patient develops diarrhea. C. difficile infection is a nosocomial disease that is spread primarily by the medical staff, and hospital epidemics are relatively common. Patients usually acquire the organism from the hospital, not from their own flora. In terms of cost and productivity, C. difficile is a major burden to our health care system. There are estimated to be 250,000 to 300,000 cases of C. difficile disease a year in U.S. hospitals, which translates into hundreds of millions of dollars for hospital care. A typical case results in 1 to 2 extra weeks of patient care costing roughly $10,000. This price assumes that the patient responds to treatment and does not relapse. Patients may respond to the simple discontinuation of the inciting agent, but many patients must be treated with metronidazole or vancomycin. Patients respond rapidly to this antimicrobial therapy, but since the therapy kills the normal bacterial flora, it can also cause the disease to recur. Once the colon has been injured, it seems to be more susceptible to reinfection. Relapses occur approximately 20% of the time, and some patients have a series of relapses, extending the illness.

Once C. difficile has been identified in an institution, efforts to minimize its transmission (enteric isolation, disinfection with hypochlorite solutions, etc.) may be implemented to minimize additional cases or outbreaks, further increasing costs. Unfortunately, the spores are almost impossible to eliminate from hospital wards, and some hospitals have experienced C. difficile outbreaks that continued for years. Fortunately, with increased awareness and easier access to new and improved in vitro diagnostic tests, physicians can identify and treat the disease more quickly and accurately than was possible even 5 years ago. New cases and relapses can be diagnosed earlier in the disease process, and hopefully, we are reducing the likelihood of outbreaks. In summary, increased testing and awareness have improved patient care and lowered medical costs.

All of the various types of in vitro tests for C. difficile (bacterial culture, tissue culture assay, and antibody-based tests for toxin or common antigen) are diagnostic aids for the physician. These tests are not stand-alone tests, and it is important that test results be considered in conjunction with the patient history. For example, up to 50% of infants acquire toxigenic C. difficile in the hospital nursery. In some instances, asymptomatic infants can have levels of toxin comparable to those seen in adult patients with pseudomembranous colitis. On rare occasions, C. difficile disease may occur in compromised infants (e.g., premature infants who are septic and who are treated with antibiotics), but only a few isolated incidents have been reported. In addition to asymptomatic infants, many hospitalized adults become asymptomatic carriers after receiving antibiotics. In fact, the percentage of patients who become carriers may be higher than the number of patients who develop serious diarrhea and colitis following infection (11). These patients usually have high numbers of the organism in their colons, but the amount of toxin in their stools can vary from very small to large. We do not understand why some people develop life-threatening colitis after becoming infected with toxigenic isolates while others do not, but in general the unaffected patients tend to be younger and in better health. The situation is further complicated because about 20% of patients who are positive for the common antigen of C. difficile carry a nontoxigenic strain of C. difficile. Nontoxigenic isolates are not responsible for the diarrhea because they do not produce toxins, but these patients should be monitored closely because they are susceptible to infection with a toxigenic strain and patients can harbor multiple strains at one time.

The epidemiology of an outbreak can be monitored only by bacterial culture, but this is seldom done because the procedure is time-consuming (72 h of incubation under anaerobic conditions) and expensive. Even when culture is done, the accuracy of results varies greatly between institutions because technologists often lack anaerobic-organism expertise and the enrichment methods and media are not standardized. Several typing methods have been published, and these provide information on the spread of outbreaks. The incidence of toxin-positive versus toxin-negative isolates in symptomatic patients is not well established. In our experience, about 75 to 80% of the C. difficile isolates from patients with AAD produce toxin (data on file, TechLab).

Although the tissue culture assay is time-consuming (48 h to rule a specimen negative) and tedious (specimens must be centrifuged and filtered before being tested), some laboratories use it because of its high sensitivity. The test detects picogram levels of C. difficile toxin (primarily toxin B), making it the most sensitive test available. Because of its high sensitivity, the tissue culture assay is often referred to as the “gold standard.” However, it is also the least-controlled test, and nonspecific reactions are common in some laboratories. The addition of too much fecal material to the tissue culture well can cause false-positive reactions. Current commercial tests recommend a final dilution of 1:40 to 1:50 (in the well) for assays, which minimizes the chance of nonspecific cell rounding, but some laboratories continue to test 1:10 or 1:20 dilutions. Specimens may also cause nonspecific cell rounding that is neutralized not only by the specific antitoxin but also by neutral serum. This happens only rarely, and laboratories do not run neutral serum as a control because this increases the cost and because normal serum is not included in any of the commercial assays. Most laboratories that perform tissue culture assays use commercially available human foreskin cell monolayers, but others use their in-house cell lines that include Chinese hamster ovary K-1 cells and MRC-5 lung fibroblasts. Each of these cell lines performs satisfactorily when conditions are carefully controlled and the technologist is experienced, but in the hands of an inexperienced technologist, in-house tissue culture assays can be very inaccurate.

Commercial tests for GDH (the common antigen) of C. difficile have been available for more than 10 years. The initial kit on the market was a latex agglutination test, but a lateral-flow device with a colored latex conjugate and flowthrough tests with enzyme conjugates have become available. GDH is an essential enzyme and is produced constitutively by all C. difficile isolates. Like bacterial culture, tests that detect the common antigen do not distinguish toxin-producing from nontoxigenic isolates. Even so, the common antigen has proven to be a good screening marker for C. difficile because the enzyme is produced in large amounts and can readily be detected in fecal specimens. The commercial GDH tests offer a turnaround time of 15 to 45 min, which is another reason the tests are used in many labs. Antiserum against C. difficile GDH cross-reacts with GDH from other anaerobes, including Clostridium sporogenes, Peptostreptococcus anaerobius, and proteolytic Clostridium botulinum. Some of the newer GDH tests have avoided this potential pitfall by using antibodies that react only with the C. difficile GDH. The greatest utility of the common antigen tests is their use as a screen to rule specimens negative and to select specimens for further testing. Because of the common presence of nontoxigenic strains in AAD patients, we believe that common antigen tests should be confirmed with a toxin test. By this approach, infections with toxigenic C. difficile can be confirmed, and exposure of patients to secondary infections following unnecessary treatment with metronidazole or vancomycin can be reduced.

ELISAs that detect toxin A began entering the marketplace more than 10 years ago. Clinical evaluations of these ELISAs show that most exhibit a sensitivity of >80% compared to that of the tissue culture assay, with some in the range of 85 to 95%. Although the sensitivity is lower than that of the tissue culture assay, its rapid turnaround time, ease of use, lower cost, and ability to batch specimens are among the reasons why laboratories choose an ELISA. Physicians appreciate the rapid results since C. difficile toxins act very quickly on the colons of susceptible individuals.

Until recently, ELISAs that detect both toxins A and B were deemed unnecessary. Toxin A had been shown to be a very potent enterotoxin that was capable of causing the diarrhea and mucosal damage associated with the disease. Therefore, it seemed logical to use toxin A as a diagnostic marker. In addition, only a single A⁻ B⁺ isolate, CCUG 8864, had been described when the first toxin A ELISAs appeared on the market (10, 18; S. P. Borriello, S. Hyde, S. V. Seddon, B. W. Wren, and A. B. Price, Abstr. XVth Int. Congr. Microb. Ecol. Dis., 1990). Interestingly, toxin B from the 8864 isolate was determined to be more enterotoxic than toxin B from typical A⁺ B⁺ isolates. At the time, the significance of this unusual toxin B was unclear, and there were doubts about whether such strains caused colitis.

By testing specimens in toxin A-specific ELISAs simultaneously with toxin A- and B-specific ELISAs, tissue culture, and PCR, we and others began to identify more A⁻ B⁺ isolates from patients with C. difficile disease. In these instances, no other enteropathogens or typical A⁺ B⁺ isolates were identified; only A⁻ B⁺ isolates were present. In the United States, the first specimens were from isolated cases. In Japan, the A⁻ B⁺ isolates identified by Naoki Kato and colleagues have primarily been associated with asymptomatic persons or associated only with mild diarrhea (7). However, in Winnipeg, Manitoba, Canada, there was an A⁻ B⁺ outbreak resulting in the deaths of two patients (2). In addition, a death attributable to an A⁻ B⁺ isolate was reported in the midwestern United States, and isolated cases as well as an outbreak were described in Europe (6, 9, 15). The actual incidence of A⁻ B⁺ isolates in the United States has not been established, but our results suggest that the strains are not common except when an outbreak occurs. Under those conditions, it is the most common isolate in the hospital experiencing the outbreak. In the British Isles, Jon Brazier (personal communication) estimated that ca. 5% of the toxigenic isolates were A⁻ B⁺ and that such isolates occurred in approximately 25% of the hospitals surveyed.

Based on recent data, several conclusions can be drawn about the clinical significance of A⁻ B⁺ C. difficile. First, A⁻ B⁺ isolates are capable of causing disease that can be severe or fatal if not promptly diagnosed and treated. Second, the incidence of A⁻ B⁺ isolates has not been studied in great detail, but the initial reports suggest that it is low. Third, like typical A⁻ B⁺ isolates, these isolates are capable of spreading and causing outbreaks in institutions. Fourth, and very importantly, these isolates do not react in toxin A-specific ELISAs. They can cause C. difficile disease but are not detected by toxin A-specific ELISAs. For this reason, we and others have recommended ELISAs that detect both toxins. Not only will the A- and B-specific tests detect these atypical isolates but the tests will also provide increased sensitivity for borderline specimens that contain only low levels of toxins A and B. The A⁻ B⁺ isolates produce the common antigen and will be detected by common commercial antigen tests and by the tissue culture assay.

The lack of reactivity of A⁻ B⁺ isolates in toxin A-specific ELISAs can be explained at the molecular level. A⁻ B⁺ isolates carry a truncated toxA gene (8, 13). Truncated toxA lacks the portion of the toxin (i.e., the immunodominant repeating units that comprise the binding portion) recognized by the antibodies used in the toxin A-specific ELISAs. In addition, there is a stop codon located near the 5′ end of the truncated gene. Thus, the lack of reactivity results from the complete absence of the toxin A peptide. Even if the truncated gene product were expressed, the peptide would still lack the portion of the toxin recognized by the antibodies.

These findings raise the question of how A⁻ B⁺ isolates cause disease if no functional toxin A is produced. The work of Aktories and Just suggests a possible mechanism. These researchers showed that both toxins A and B are monoglucosyl transferases that transfer a glucose moiety onto a specific amino acid located at a critical site on G proteins that control the cytoskeletal system (1). When glucosylated, the G protein no longer controls the polymerization of actin. As a result, cellular functions under the control of the cytoskeletal system fail, causing the cell-rounding effect and, eventually, cell death. Both toxin A and toxin B exhibit this activity and cause similar cell-rounding reactions, but toxin A binds mainly to cells surrounded by a glycocalyx. In typical A⁺ B⁺ strains, the G protein that is primarily targeted is rho. In A⁻ B⁺ isolates, however, the substrate specificity of toxin B is broader, resulting in glycosylation of other G proteins. In this regard, toxin B from A⁻ B⁺ isolates resembles toxin LT of C. sordellii, which cross-reacts with toxin B but which is more lethal (3, 19). Speculatively, the broader specificity may explain the increased toxicity of the aberrant toxin B, particularly its ability to damage the intestinal mucosa. Looking back to the original description of CCUG 8864 toxin B, we now recognize that the increased toxicity may be due to its broader substrate specificity. In summary, the ability of the aberrant toxin B to recognize and inactivate a broader range of G proteins may explain why A⁻ B⁺ isolates are capable of causing disease in the absence of a functional toxin A.

Most of the evidence surrounding atypical C. difficile isolates has focused on A⁻ B⁺ isolates. Interestingly, the same phenomenon has been reported for a C. sordellii isolate that produces LT, the toxin that cross-reacts with toxin B, but not HT, which is the toxin that cross-reacts with toxin A. There is now evidence of a clinical C. difficile isolate that produces an aberrant toxin B and a fully functional toxin A (12). In addition, an A⁺ B⁻ horse isolate has been reported. It is becoming apparent that various subgroups comprised of different toxin types exist, and they each appear to be capable of infecting and causing disease in animals and humans (16). From a clinical perspective, the current data indicate that these atypical C. difficile isolates can be effectively detected in fecal specimens with tests that detect both toxins or the common antigen. For this reason, we and others continue to support the use of ELISAs that detect both toxins, either as the primary test or as a follow-up test for specimens that are positive for the common antigen.

The newest additions to the line of C. difficile diagnostics are the rapid membrane tests that target toxin A. These are lateral-flow devices with colored conjugates or flowthrough formats that require multistep processing (such tests utilize peroxidase-tagged antibodies and a wash step followed by the addition of a substrate). The sample preparation for these tests requires centrifugation or filtration. Reports at recent American Society for Microbiology meetings indicate that these tests have a sensitivity in the range of 60 to 85% (depending on the test and clinical site) compared to the sensitivity of tissue culture. Because these are toxin A-specific tests, they do not detect A⁻ B⁺ isolates. Even so, they offer smaller laboratories the option of performing C. difficile testing in-house rather than sending specimens out for testing. In-house testing results in faster turnaround times with fresh specimens instead of results in 24 to-48 h with less-than-ideal specimens. Choosing between a rapid test for the common antigen or a rapid test for toxin A is difficult. The common antigen tests offer higher negative predictive values but, again, do not distinguish between the presence of toxigenic and nontoxigenic isolates in stool specimens. Rapid toxin A tests offer this advantage but give false-negative results because of decreased sensitivity.

The tissue damage to the intestinal mucosa by toxins A and B leads to a rapid influx of inflammatory cells. The inflammatory response plays a key role in how quickly the disease progresses to colitis and whether the disease develops into pseudomembranous colitis, which is life threatening if left untreated. Levels of fecal lactoferrin, which is released from the secondary granules of fecal leukocytes, and other inflammatory markers rise significantly in patients with advanced C. difficile disease compared with levels in patients with a milder case of the disease (17). Thus, the presence of elevated fecal lactoferrin may help to define the severity of the disease. Physicians may be treating patients who are not affected by the presence of C. difficile in their colons. The therapy might cause more harm than good. Tests that monitor the level of intestinal inflammation will provide valuable information to physicians, alerting them to the need for prompt therapy with metronidazole and vancomycin when there is inflammation in the presence of the organism.

In vitro diagnostic testing for C. difficile and its toxins continues to improve, resulting in better health care for the patient. For the clinical laboratory, the question of whether the lab should perform C. difficile testing has instead become a question of which test to use. Because of the increased awareness of this important pathogen, we now more fully understand the impact of the disease in terms of health care costs. With the progress that has been made with C. difficile diagnostic testing, it is easy to become complacent about the need to develop new tests. In reality, there are still quite a few pieces of the puzzle left to solve. We continue to believe strongly that tests which detect toxin should be used, either as the primary or confirmatory test. Tests that detect the common antigen are valuable, but a fairly large percentage of the positive specimens may contain nontoxigenic isolates. If such tests were coupled with tests for intestinal inflammation, there would be less need to test for the presence of toxins. It is important to remember that C. difficile is responsible for only 20% or less of AAD cases, and accurate testing helps to prevent unnecessary treatment while minimizing the overuse of antibiotics. Accurate testing becomes even more important as we begin to explore the association of C. difficile with other intestinal ailments. Recent reports at the 2002 Digestive Disease Week meeting described the presence of C. difficile and its toxins in a significant number of patients with inflammatory bowel disease. The significance of this association still is not known, but accurate tests can help to ensure that these patients receive the appropriate treatment.

Recent progress has led to the development of tests that recognize the aberrant toxins A and B of C. difficile. However, new results are showing that, in addition to toxins A and B, a low percentage of clinical isolates carry iota toxin (5, 14). This toxin is related to the iota toxin of Clostridium perfringens and Clostridium spiroforme, both of which are associated with AAD. Iota toxin is totally unrelated to toxins A and B of C. difficile, and any role for this toxin as part of the pathogenic repertoire of C. difficile is totally unknown at this stage. We do not know, for example, how often iota toxin is present in clinical isolates, whether it may contribute to tissue damage or diarrhea during the infection, or if it may be present in isolates that do not produce either toxin A or B. However, the fact that the toxin cross-reacts with the C. perfringens and C. spiroforme iota toxins, both of which are enterotoxins, raises our suspicions about a possible role of iota toxin in C. difficile disease.

The original name of the organism, Bacillus difficilis, was coined because of the difficulty that Hall and O'Toole encountered when isolating C. difficile from healthy newborn infants in the mid-1930s. Looking at the challenges that lie ahead for C. difficile testing and the interpretation of the test results by physicians, we believe that these scientists chose a most fitting name.