C. diff Diagnosis versus Detection: Why Tests Remain Ambiguous

Jan. 21, 2019

Clostridioides difficileClostridioides difficile is the most common cause of infectious diarrhea in healthcare settings, and data from the CDC’s Emerging Infections Program surveillance system in 2011 estimates that it caused nearly half a million infections and 29,000 deaths within 30 days of diagnosis. There are a multitude of tests available for diagnosis of Clostridioides difficile infection (CDI)–detecting C. difficile-specific nucleic acid, enzyme, and/or toxins, in various combinations and algorithms–and this has led to significant confusion regarding clinical interpretation and the distinction between colonization and true infection.

The Infectious Diseases Society of America (IDSA) recently published updated clinical practice guidelines on CDI including recommendations for testing. These recommendations state that the preferred population for CDI testing includes patients with unexplained and new-onset diarrhea consisting of ≥3 unformed stools in 24 hours. For institutions in which there are no preagreed institutional criteria for patient stool submission, the best-performing method determined by positive and negative predictive values was a stool toxin test as part of a 2- or 3-step algorithm rather than a nucleic acid amplification test (NAAT) alone. They cite 2 common methods: 1) use of glutamate dehydrogenase plus toxin assays arbitrated by nucleic acid amplification testing (NAAT) or 2) NAAT plus toxin assay. This recommendation, however, is rated as “weak” with a “low quality of evidence.” The panel points out that in fact the most sensitive method of diagnosis is a NAAT alone or a multistep algorithm, which should be used when there are preagreed institutional criteria for stool submission.

These recommendations reflect the ongoing lack of consensus regarding optimal strategies for diagnosis of CDI. Recommendations vary further when one considers those from outside the U.S.: European guidelines prioritize toxin detection and place less emphasis on NAAT or multistep algorithms.
 

Diagnostic Strategies and Limitations

 
Toxin detection and culture: Historically, the laboratory gold standard has been toxigenic culture where C. difficile is cultured from stool and isolates are tested for their ability to produce toxin; stool filtrates can also be directly tested for toxin via a cell cytotoxicity assay (CCNA) as an alternative reference method. These methods take several days and are thus unsuitable for routine laboratory testing. A large UK-based study compared toxigenic culture with cytotoxicity testing on more than 12,000 specimens and correlated the results with clinical data. While they found that positive cytotoxicity assay results correlated with increased mortality, positive toxigenic cultures with negative toxicity assays did not, thus implying that the detection of toxin was critical for clinical diagnosis of CDI. Toxin detection by qualitative enzyme immunoassays (EIAs) used to be a mainstay of diagnosis, however, these assays have significant limitations in sensitivity compared to toxigenic culture (52-75%), although they do have higher specificity (96-98%) in comparison to culture. There are a variety of available commercial laboratory options for CDI testing that are well-described in recent reviews.
 
Glutamate Dehydrogenase detection: Glutamate dehydrogenase (GDH) immunoassays and other molecular tests evolved in order to address the poor sensitivity among toxin EIAs. GDH immunoassays detect the highly conserved metabolic enzyme present in all C. difficile isolates. However, this antigen is present in both toxigenic and nontoxigenic strains of C. difficile and, accordingly, GDH testing can only be a screening step in a 2- or 3-step algorithm before a more specific toxin test and/or molecular test for toxin gene detection.
 
NAAT: Although NAATs have been used in research settings since the early 1990s, the first US Food and Drug Administration approved platform was not available until 2009. At least 12 commercial platforms are available currently that detect gene targets including tcdA (toxin A gene), tcdB (toxin B gene), and 16S ribosomal RNA (rRNA). The assays are more sensitive than toxin EIA and possibly GDH EIAs, but less sensitive than toxigenic culture.

Algorithm-based Multistep Testing and Ultrasensitive Toxin Detection: The complexity of the CDI testing world is further confounded by conflicting data on the best algorithmic approach to diagnosis. Scientists in a single-center prospective cohort study compared the need for treatment and natural history of patients who were toxin-EIA-positive/PCR positive (131 patients) with toxin-EIA-negative/PCR positive patients (162) and toxin-EIA-negative/PCR negative patients (1123). They found that toxin-positive/PCR-positive patients had more diarrhea and CDI-related complications, whereas the toxin-negative/PCR positive and toxin-negative/PCR-negative patient groups had similar rates of gastrointestinal complications as compared to each other (7.6% vs. 0% vs. 0.3%; p <0.001). There were 11 CDI-related deaths among the toxin-positive/PCR-positive group, one death in the toxin-negative/PCR-positive cohort, and no deaths among the toxin-negative/PCR-negative group. The researchers thus concluded that toxin testing alone may be sufficient for CDI diagnosis and use of NAAT tests alone may lead to overdiagnosis and overtreatment.  Nevertheless, NAAT testing and identification of asymptomatic carriage is relevant for infection control and epidemiology purposes. 

Contrarily, another research group reported that the absence of toxin in stool may not be predictive of CDI severity and thus NAAT-positive, EIA-negative results are still clinically meaningful.  The researchers recommended that NAAT should be used as the primary diagnostic method for CDI, although they did not specify a preferred diagnostic algorithmic approach. In this study 296 patients were enrolled with 143 classified as true CDIs based on multiple different methods. They found no difference in toxin EIA positivity between patients with mild vs. severe disease (49% vs. 58%, p = 0.31). Toxin EIA detection, however, is limited by a relatively non-sensitive limit of detection. The best performing analytical limits of detection for EIAs range between 0.8-2.5 ng/ml whereas for cell-based cytotoxicity assays toxin concentrations were calculated in some scenarios to be as low as 30 pg/ml. More recent studies have therefore queried whether ultrasensitive toxin detection may in fact better differentiate colonization from true infection with the hypothesis that colonization would have lower levels of toxin.

This hypothesis was tested in a recent single-center prospective study that investigated the performance of an ultrasensitive assay for detection and quantification of C. difficile toxins using single molecule array (Simoa) technology capable of an analytical cutoff of approximately 1pg/ml and clinical cutoff of approximately 20 pg/ml. Researchers compared toxin concentrations in CDI NAAT-positive patients (n=122) defined as those having ≥3 unformed stools during the 24 hours before stool collection or persistent diarrhea in the clinical notes versus asymptomatic carriers who were NAAT positive (n=44) but who had no report of diarrhea during the 48 hours prior to stool collection. The scientists were surprised to find that stool toxin A and B concentration could not distinguish a patient with CDI from an asymptomatic carrier. Median toxin A, toxin B, and toxin A+B concentration, and NAAT cycle threshold (Ct) values in the two groups were in fact similar.  Frequencies of toxin A+B concentrations ≥ 20pg/ml (clinical cutoff) were comparable between CDI-NAAT+ (65%) vs. carrier-NAAT+ (77%) groups. They did note, however, that if they defined the CDI and carrier groups by not only NAAT positivity but also toxin positivity (where toxin A + B ≥ 20 pg/ml), then there were significant differences in median toxin concentrations (median toxin A, B, and A+B concentrations) and Ct values. Accordingly, while the study observed that patients with very low levels of toxin still could have significant diarrhea consistent with CDI and conversely asymptomatic patients could have significant amounts of toxin detected, above a cutoff threshold for toxin asymptomatic patients did have lower concentrations of toxin detected.

In summary, ultrasensitive toxin detection does not seem to be the holy grail answer to how to more effectively diagnose CDI and distinguish disease from colonization. The surprising results from this study raise the central question of why patients with high levels of toxin in their stool can be asymptomatic and contrarily patients with very low levels of toxin may be symptomatic. Some experts hypothesize that host factors, such as antitoxin antibodies, may explain why patients can be asymptomatic despite the presence of C. difficile toxins in their stool. It may be that a test which detects such antibodies or other host biomarkers, in addition to pathogen detection, may be needed for improved CDI diagnosis. While we eagerly await further research into CDI diagnostics, we remain in a familiar area of clinical medicine where tests provide supportive but not definitive evidence of a diagnosis, and we must be aware of their inherent limitations.

The above represent the views of the author and does not necessarily reflect the opinion of the American Society for Microbiology.

Author: Rose Lee

Rose Lee
Rose Lee, MD, is an infectious diseases and microbiology fellow at Beth Israel Deaconess Medical Center and Boston Children's Hospital.  She received her MD from Northwestern Feinberg School of Medicine and is a postdoctoral fellow in the laboratory of James Collins working on novel diagnostic technologies for infectious diseases.