Comparative Evaluation of Real-Time Screening PCR Assays for Giardia duodenalis and of Assays Discriminating the Assemblages A and B

Abstract

Due to superior sensitivity compared to traditional microscopy, real-time PCR has been well established for the diagnosis of Giardia duodenalis in human stool samples. In this study, screening real-time PCRs for different target genes of G. duodenalis, i.e., the 18S rRNA gene, the gdh (glutamate dehydrogenase) gene and the bg (beta-giardin) gene, were comparatively assessed next to various real-time PCR assays for the discrimination of the assemblages A and B of G. duodenalis targeting the bg gene with and without locked nucleic acid–containing probes as well as the tpi (triose phosphate isomerase) gene. The screening PCRs were assessed by including 872 non-preselected samples with a high pre-test probability for G. duodenalis in the statistical analysis, while 53 G. duodenalis-positive samples as indicated by at least two screening PCRs were finally included in the assessment of the assemblage-specific PCRs. For the screening PCRs, sensitivity estimated with latent class analysis (LCA) ranged from 17.5% to 100%, specificity from 92.3% to 100% with an accuracy-adjusted prevalence of 7.2% for G. duodenalis within the non-preselected sample collection. In detail, sensitivity and specificity were 100% and 100% for the 18S rRNA gene-specific assay, 17.5% and 92.3% for the gdh gene-specific assay, and 31.7% and 100% for the bg gene-specific assay, respectively. Agreement kappa was slight with only 15.5%. For the assemblage-specific PCRs, estimated sensitivity ranged from 82.1% to 100%, specificity from 84.0% to 100% with nearly perfect agreement kappa of 90.1% for assemblage A and yet substantial agreement of 74.8% for assemblage B. In detail for assemblage A, sensitivity and specificity were 100% and 100% for the bg gene-specific assay without locked nucleic acids (LNA) as well as 100% and 97.8% for both the bg gene-specific assay with LNA and the tri gene-specific assay, respectively. For assemblage B, sensitivity and specificity were 100% and 100% for the bg gene-specific assay without LNA, 96.4% and 84.0% for the bg gene-specific assay with LNA, and 82.1% and 100% for the tri gene-specific assay, respectively. Within the assessed sample collection, the observed proportion comprised 15.1% G. duodenalis assemblage A, 52.8% G. duodenalis assemblage B and 32.1% non-resolved assemblages. Only little differences were observed regarding the cycle threshold (Ct) values when comparing the assays. In conclusion, best diagnostic accuracy was shown for an 18S rRNA gene-specific screening assay for G. duodenalis and for a differentiation assay discriminating the G. duodenalis assemblages A and B by targeting the bg gene with probes not containing locked nucleic acids. By adding additional highly specific competitor assays for confirmation testing, diagnostic specificity can be further increased on the cost of sensitivity if optimized specificity is desired.