Comparison of the Diagnostic Accuracy of Three Real-Time PCR Assays for the Detection of Arcobacter butzleri in Human Stool Samples Targeting Different Genes in a Test Comparison without a Reference Standard

Potential etiological relevance for gastroenteric disorders including diarrhea has been assigned to . However, standard routine diagnostic algorithms for stool samples of patients with diarrhea are rarely adapted to the detection of this pathogen and so, is likely to go undetected unless it is specifically addressed, e.g., by applying pathogen-specific molecular diagnostic approaches. In the study presented here, we compared three real-time PCR assays targeting the genes , (both hybridization probe assays) and (fluorescence resonance energy transfer assay) of in a test comparison without a reference standard using a stool sample collection with a high pretest probability from the Ghanaian endemicity setting. Latent class analysis was applied with the PCR results obtained with a collection of 1495 stool samples showing no signs of PCR inhibition to assess the real-time PCR assays' diagnostic accuracy. Calculated sensitivity and specificity were 93.0% and 96.9% for the -PCR, 100% and 98.2% for the -PCR, as well as 12.7% and 99.8% for the -PCR, respectively. The calculated prevalence within the assessed Ghanaian population was 14.7%. As indicated by test results obtained with high-titer spiked samples, cross-reactions of the -assay and -assay with phylogenetically related species such as can occur but are less likely with phylogenetically more distant species like, e.g., . In conclusion, the -assay showed the most promising performance characteristics as the only assay with sensitivity >95%, albeit associated with a broad 95%-confidence interval. In addition, this assay showed still-acceptable specificity of >98% in spite of the known cross-reactivity with phylogenetically closely related species such as . If higher certainty is desired, the -assay with specificity close to 100% can be applied for confirmation testing with samples showing positive -PCR results. However, in case of a negative result in the -assay, this cannot reliably exclude the detection of in the -assay due to the -assay's very low sensitivity.