@article{716,
  keywords = {acanthamoeba keratitis, diagnostic accuracy, fungal keratitis, molecular diagnostics},
  author = {Binh Cao and Chi Vu and Cung Le and Tuan Anh and Dong Ngoc and Hiep Xuan and Thu Tran and Dionna Wittberg and Krisianne Aromin and Cindi Chen and Thuy Doan and Jeremy Keenan},
  title = {Diagnostic Accuracy of Clinic-Based Loop-Mediated Isothermal Amplification and Quantitative PCR for Fungal and Acanthamoeba Keratitis.},
  abstract = {<p><b>PURPOSE: </b>To evaluate the diagnostic accuracy of loop-mediated isothermal amplification (LAMP) and quantitative polymerase chain reaction (qPCR) for fungal and acanthamoeba keratitis when performed on simple clinic-based platforms.</p><p><b>METHODS: </b>A total of 177 corneal scrapings were collected from 174 patients ≥18 years old diagnosed with presumed infectious keratitis at Vietnam National Eye Hospital. Smear was performed per routine by the microbiology laboratory. Clinic-based LAMP and qPCR testing for acanthamoeba and fungal targets was performed by a trained technician. The diagnostic accuracy of acanthamoeba LAMP, acanthamoeba qPCR, and panfungal qPCR was assessed relative to smear.</p><p><b>RESULTS: </b>Of 177 cornea swabs tested with LAMP and qPCR for acanthamoeba, 5 were smear-positive for acanthamoeba. Sensitivity was 40% (95% CI, 12%-77%) for acanthamoeba LAMP and 100% (95% CI, 57%-100%) for acanthamoeba qPCR, and specificities were 100% (95% CI, 98%-100%) and 99% (95% CI, 96%-100%), respectively. Of 177 swabs tested with panfungal qPCR, 61 had a smear positive for fungus. Panfungal qPCR had a sensitivity of 84% (95% CI, 73%-91%) and specificity of 72% (95% CI, 64%-80%).</p><p><b>CONCLUSIONS: </b>A clinic-based qPCR was highly sensitive and specific for diagnosis of acanthamoeba keratitis but had poorer diagnostic performance for fungal keratitis. The LAMP protocol used in this study requires optimizations to improve sensitivity and reliability.</p>},
  year = {2025},
  journal = {Cornea},
  month = {08/2025},
  issn = {1536-4798},
  doi = {10.1097/ICO.0000000000003944},
  language = {eng},
}