Comparison of a novel chemiluminescent based algorithm to three algorithmic approaches for the laboratory diagnosis of Clostridium difficile infection

Clostridium difficile is a major cause of healthcare-associated diarrhoea. Many different approaches are available for the laboratory diagnosis of C. difficile infection (CDI). American [1] and European [2] guidelines recommend testing patients with a two-step algorithm including the detection of glutamate dehydrogenase (GDH) as a screening method followed, in case of positive result, by the detection of free toxins or their genes as a confirmation method. The DiaSorin Liaison^® C. difficile GDH and toxins A and B (DiaSorin, Saluggia, Italy) is a sandwich chemiluminescent immunoassay (CLIA) performed on a stool extract. Luminescent assays for C. difficile diagnosis have not been previously reported in the literature. This method may be more reliable than enzyme immunoassays (EIA), which are known to lack sensitivity for free toxin detection. The objective of the study was to evaluate the performances of the newly available chemiluminescence-based DiaSorin algorithm for the detection of GDH and free toxins A and B using the Liaison apparatus in a routine laboratory.

The DiaSorin algorithm was compared to three other algorithms: (1) a widely used EIA algorithm including, the C. Diff Quik Chek GDH^® followed by detection of free toxins A and B, TOX A/B Quik Chek^® test (Alere, Waltham, MA, USA), (2) the same EIA-based algorithm with toxigenic culture as a third step in case of negative toxin results; for this purpose, the stools were directly inoculated on a cycloserine-cefoxitin-amphotericin agar (bioMérieux, Marcy l’Etoile, France) incubated at 37 °C for 3 days in an anaerobic atmosphere and the isolates were then tested for toxins A and B using the TOX A/B Quik Chek^® (Alere), and (3) a very sensitive algorithm [3‐8] using the EIA ImmunoCard^® C. difficile GDH for screening followed by a loop-mediated isothermal amplification assay for tcdA toxin gene detection (illumigene ^® Meridian, Cincinnati, OH, USA). The manufacturer’s recommendations were followed. All algorithms included a preliminary screening step for the detection of GDH and then, if positive, the stool samples were tested for toxins or toxin genes. The specimens that tested positive for the toxins were considered positive. The specimens that tested negative for GDH or for toxins were considered negative. In our study, the aggregate criteria for a true positive or a true negative result were a positive or negative result for the four algorithms, respectively. If an algorithm result was different from the three others, the sample was considered discordant. The discordant samples were resolved using enriched toxigenic culture (ETC) considered as a reference method and performed at the French National Reference Laboratory for C. difficile (Paris, France): stool samples were inoculated in pre-reduced taurocholate-cycloserine-cefoxitin BHI broth incubated for 5 days at 37 °C under anaerobic conditions; the broth was then plated onto laboratory standard selective plates containing taurocholate, cycloserine and cefoxitin. Toxinogenicity of the strains was demonstrated using an in-house polymerase chain reaction (PCR) targeting tcdA and tcdB genes. After ETC, a stool sample was considered positive for toxigenic C. difficile if the toxin genes were detected.

Between June and September 2013, C. difficile testing was prospectively performed with the four different algorithms at the Bacteriology Laboratory of the Bordeaux University Hospital (Bordeaux, France) on diarrhoeal stools either upon a physician’s request or systematically inpatients with diarrhea after day 3 of hospitalization. As recommended in the literature [9] and manufacturers’ instructions, the followings samples were excluded: formed stools, bloody stools, stools submitted from a patient with a positive C. difficile test result in the preceding 7 days, stool samples which were received in the laboratory more than 48 h after emission, and samples from patients less than 2 years of age. A total of 468 stool samples were included in the study. Before resolution, the overall algorithm concordance was 94.7 % (443/468) with 2.8 % true positive results (13/468), 91.9 % true negative results (430/468), and 5.3 % (25/468) discordant results. The 25 discordant samples were subjected to ETC (Table 1). Among the discordant samples, the DiaSorin algorithm showed 13 true positive, 4 true negative, 4 false positive and 4 false negative results after resolution. The sensitivity, specificity, positive and negative predictive values of the DiaSorin Liaison were 86.7, 99.1, 86.7 and 98.1 % (Table 1), respectively. There was no significant difference for sensitivity and specificity between the DiaSorin and Meridian algorithm (90.0 %) using a molecular test (McNemar’s test, p = 0.72). Moreover, the DiaSorin algorithm exhibited a significantly higher sensitivity compared to that of Alere with toxigenic culture confirmation (60.0 %) and without confirmation (50.0 %). There was no significant difference in GDH results between the screening assays (CLIA and EIA): the overall concordance for GDH tests was 95.7 % with 15 % of positive results.

Since the algorithms were based on an initial GDH screening with no significantly different results observed, their performance was related to the part of the test confirming the toxins. As expected, the novel DiaSorin chemiluminescent algorithm has a high sensitivity indicating that this sandwich ELISA-like test with enhanced signaling indeed compensates for the lack of sensitivity usually observed with EIA-based assays. Surprisingly, the overall performance of the DiaSorin algorithm was as high as the Meridian algorithm, already known to be very accurate. Combining this two-step algorithm with a confirmation test was not required. Moreover, tests that detect the presence of free C. difficile toxins in feces are significantly associated with clinical outcomes [10]. A retrospective chart review was performed for all the positive results of the DiaSorin algorithm. There was a good correlation with clinical pictures (Table 2).

The limitation of the study was the small number of positive cases rendering the estimation of sensitivity with a large confidence interval. Given the low incidence of CDI in this population, a larger number of specimens should have been evaluated. Choosing the right diagnostic approach is a matter of test accuracy, turnaround time, and cost for routine use in a clinical laboratory. In conclusion there is a good performance of the DiaSorin assay in comparison to the three other approaches. This method is sufficient for the diagnosis of CDI and there is no need for a confirmation test because the sensitivity of the test is as high as the molecular method. This algorithm could replace the others.