Evaluation of Molecular Assays for Diagnosis of Amoebic Liver Abscess in India with Bayesian Latent Class Analysis.

Abstract

Amoebic liver abscess (ALA) is the most common extra-intestinal complication of Entamoeba histolytica, accounting for 50,000 deaths annually, and is endemic in South Asia. Diagnosis based on microscopic examination is insensitive, and serological assays are not discerning of current infections in endemic settings with high exposure. For a rapid and confirmatory laboratory diagnosis of ALA, the performance of a polymerase chain reaction (PCR), quantitative real time PCR (qPCR), digital droplet PCR (ddPCR), and a loop-mediated isothermal amplification (LAMP) assay that detects E. histolytica DNA in liver abscess pus, and a lectin antigen detection ELISA were evaluated against clinical diagnosis (based on predefined criteria) as the gold standard. Owing to the lack of a laboratory gold standard, a Bayesian latent class analysis approach was also used to determine sensitivity and specificity of these assays. In the latent class analysis, qPCR and ddPCR showed the highest sensitivity (98% and 98.1%) and specificity (both 96.6%), and although clinical diagnosis had a comparable sensitivity to qPCR and ddPCR (95.2%), poorer specificity (64.3%) was seen. Kappa agreement analysis showed that qPCR and ddPCR had a perfect agreement of 1 followed by an agreement of 0.76 (95% CI: 0.64-0.88) with PCR. Considering the performance characteristics and relative ease of setting up qPCR as well as the wide availability of qPCR equipment needed, this would be the most optimal assay for rapid, confirmatory, molecular diagnosis of ALA in the tertiary care laboratory setting in India, whereas further optimization of LAMP or antibody-based detection is required for use at smaller or secondary hospitals.