Comparison of molecular diagnostic methods for the detection of Acanthamoeba spp. from clinical specimens submitted for keratitis

Acanthamoeba spp. are responsible for a significant annual number of keratitis (AK) cases leading to vision-threatening disease worldwide. Current methods rely on direct examination of specimens by microscopy and/or culture. The former lacks sensitivity and the latter suffers from a poor turnaround time. We undertook a comparison of all published molecular methods, evaluating performance characteristics such as analytical sensitivity, specificity, limit of detection (LOD), reproducibility, accuracy, and cost of test. The study population comprised 128 patients. Eligible specimens were tested prospectively between April 2007 and May 2010 by microscopy and/or culture. Eleven different specimen types were used including corneal scrapings (51.5%), corneal swab (17.9%), and contact lens material (10.9%). Results of 2 published gel-based polymerase chain reaction (PCR) and 2 published real-time quantitative (Q) PCR methods were compared in a blinded manner to direct microscopic examination and/or culture for the detection of Acanthamoeba in clinical specimens. QPCR (Riviere method) had the highest sensitivity at 89.3%, excellent accuracy using ROC analysis (AUC ∼0.90), lowest LOD down to 0.1 organism per microliter, and superior linear correlation with parasite density (R2 = 0.9965) when compared with microscopy, culture, and other molecular methods. Phylogenetic analysis using a sequence-based typing method revealed that clinical isolates in this population with AK were genetically distinct from granulomatous amebic encephalitis or environmental isolates. The QPCR method was more expensive ($14.80) than traditional methods such as culture ($2.50) or microscopy ($2.50). However, 13 culture- and microscopy-negative specimens were positive by QPCR during the study period, suggesting that detection using QPCR may result in reduced complications and health care costs associated with misdiagnosed AK.