Conventional corneal cultures for infectious keratitis have long been plagued by low sensitivity1.With patients presenting on microbiologic therapy and fastidious organisms difficult to grow in microbiology laboratories, atypical and unexpected organisms can be missed2.
Metagenomic deep sequencing (MDS), both RNA sequencing (RNA-seq) and DNA sequencing (DNA-seq), carries potential for improved diagnostic sensitivity and accuracy3,4. The unbiased nature of total RNA/DNA sequencing allows for identification of almost any pathogen and includes the potential for pathogen discovery. However, these deep sequencing techniques have challenges with nucleic acid recovery, contamination from the environment, normal microbiota, and variable bioinformatic interpretive strategies5.
In the absence of a directly observed gold standard for the diagnosis of infectious keratitis, we used latent class analysis (LCA) to evaluate the sensitivity and specificity of conventional diagnostics tests, RNA-seq, and DNA-seq. Institutional Review Board approval was obtained. This study adhered to the tenets of the Declaration of Helsinki. Informed written consent was obtained from all patients.
Corneal scrapings were obtained for KOH wet mount, Gram stain and traditional cultures. For MDS, the affected cornea was swabbed with a sterile polyester tipped applicator (Puritan). A second swab of the inferior fornix of the unaffected, contralateral eye was obtained. Swabs were placed in DNA/RNA-Shield (Zymo Research) and shipped to the Proctor Foundation/UCSF on dry ice and stored at −80°C.
Conventional microbiologic testing and MDS were performed as previously reported4,6. Because the ocular surface is normally colonized, the taxa (at the species level) identified from the control contralateral conjunctiva were bioinformatically subtracted prior to final analysis. In cases where the suspected pathogen was common microbiota (e.g. Staphylococcus spp., Streptococcus spp.) a water control from the same sequencing run was used as background subtraction. The organism was identified as positive by MDS if it was known to cause ocular infection and if it represented the most abundant reads after background subtraction. All laboratory personnel were masked.
Sensitivity and specificity of different microbiologic diagnostic tests and latent gold standard prevalence were estimated using LCA. Uncertainty was estimated using bootstrap percentile 95% confidence intervals (1,000 simulations).
Baseline information from all participants and comparative diagnostic results are presented in Supplementary Table S1. KOH preparation or Gram stain was positive in 32 samples (70%). Of these, 23 were positive for fungus and 9 were positive for bacteria and 18 (56%) were taking antibacterial and/or antifungal drops. Traditional culture was positive in 24 (52%) patients.
MDS (combining RNA-seq and DNA-seq) was positive in 34 (74%) cases. Twenty ulcers were identified by MDS as fungal and 14 as bacterial. Nineteen of the 34 cases (56%) were on topical therapy. Of the 12 MDS negative cases, 6 (50%) were using antibacterial and/or antifungal drops. There were five cases (Patient numbers 14, 36, 43, 44, 45, Table S1) where conventional microbial diagnostics were negative and MDS was positive. There was one case where MDS was negative (Patient number 24, Table S1) and culture and Gram stain was positive for viridans streptococci. There was one case (Patient number 38, Table S1) where the culture results (fungus) was discrepant from the sequencing results (bacteria by RNA-seq and DNA-seq.) Gram stain and KOH prep were negative in this case. Six (13%) of the 46 ulcers were diagnostically negative by all tests.
An LCA comparing the performance of three diagnostic tests (combined KOH prep/Gram stain/culture, RNA-seq, and DNA-seq) is presented in Table 1. For this population of bacterial ulcers, the LCA estimated the highest performance from RNA-seq with an estimated sensitivity of 100% (95% CI: 79% to 100%) and an estimated specificity of 97% (95% CI: 84% to 100%). Similarly, for fungal ulcers, RNA-seq outperformed DNA-seq and conventional diagnostics with a sensitivity of 100% (95% CI: 86% to 100%) and a specificity of 100% (95% CI: 89% to 100%).
Table 1:
Latent Class Analysis Estimation of Diagnostic Sensitivity and Specificity of Three Microbiologic Diagnostic Tests.
| Test Type | Bacteriala | Fungalb | ||
|---|---|---|---|---|
| Sensitivity (95% CI) | Specificity (95% CI) | Sensitivity (95% CI) | Specificity (95% CI) | |
| KOH and Gm and Culturec | 61% (33%, 100%) | 94% (85%, 100%) | 95% (83%, 100%) | 81% (63%, 94%) |
| RNA-seq | 100 (79%, 100%) | 97% (84%, 100%) | 100% (86%, 100%) | 100% (89%, 100%) |
| DNA-seq | 62% (29%, 100%) | 97% (91%, 100%) | 65% (42%, 86%) | 100% (91%, 100%) |
Estimated latent class (true) prevalence 25% (12-39%)
Estimated latent class (true) prevalence 43% (28%-59%)
Gm= Gram stain. This test
An LCA comparing stains, culture. RNA-seq and DNA-seq separately is presented in Supplementary Table S2. LCAs considers each test as conditionally independent. In this study, the microbiology laboratory was not formally masked to the KOH and Gram stain results raising the possibility that culture determination was not strictly independent. Results of this four test LCA should be interpreted accordingly.
This masked prospective study of 46 corneal ulcers from the Aravind Eye Hospital in Madurai, India demonstrated that MDS may be used to identify pathogens in infectious keratitis. Compared to conventional diagnostics MDS outperforms even when the clinical setting (high burden of infection and high pre-test probability for fungal ulcers) favors traditional diagnostic tests.
It is also encouraging that MDS, in particular RNA-seq, performs well with identification of fungal organisms. This may indicate in parts of the world where fungal ulcers are cmore atypical, MDS may have higher diagnostic utility. While DNA-seq allows for easier sample handling, it lacks sensitivity compared to RNA-seq.
Over half of the ulcers (52%) were sampled after having been exposed to antimicrobial drops. The presence of prior topical treatment does not appear to limit one diagnostic modality more than another.
Performance evaluation of all diagnostic tests studied in this analysis could perform differently in other locations where the epidemiology of corneal ulcers is different and where the infectious load is less severe. Distinguishing a causative pathogen from a colonized and/or contaminated pathogen remains a challenge for every microbiologic test. The absence of viral and parasitic ulcers in this study limits analysis of MDS as pertains to other pathogens. As evidenced by some large LCA confidence intervals, this study is limited by sample size, in particular the small number of bacterial ulcers in this population.
In conclusion, no “gold” diagnostic test exists for the diagnosis of corneal infectious disease. Continuation of larger masked analyses of the performance of MDS as a diagnostic tool for corneal infections is required.
Supplementary Material
Acknowledgments
Funding:
Research reported in this manuscript was supported by the Peierls Foundation; the Research to Prevent Blindness Career Development Award (T.D.); the National Eye Institute of the National Institutes of Health under Award Number K08EY026986 (T.D.).
Footnotes
Meeting Presentation:
None
Financial Disclosures:
Gerami Seitzman is a consultant for Dompé Pharmaceuticals
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
REFERENCES:
- 1.McLeod SD, Kolahdouz-Isfahani A, Rostamian K, et al. The role of smears, cultures, and antibiotic sensitivity testing in the management of suspected infectious keratitis. Ophthalmology 1996;103(1):23–8. [DOI] [PubMed] [Google Scholar]
- 2.Seitzman GD, Hinterwirth A, Zhong L, Metagenomic Deep Sequencing for the Diagnosis of Corneal and External Disease Infections. Ophthalmology. 2019. December;126(12):1724–1726. [DOI] [PubMed] [Google Scholar]
- 3.Wilson MR, Sample HA, Zorn KC, et al. Clinical Metagenomic Sequencing for Diagnosis of Meningitis and Encephalitis. N Engl J Med. 2019;380(24):2327–2340. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Doan T, Wilson MR, Crawford ED, et al. Illuminating uveitis: metagenomic deep sequencing identifies common and rare pathogens. Genome Med 2016;8(1):90. Erratum in: Genome Med. 2016 Nov 22;8(1):123. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ung L, Bispo PJM, Doan T, et al. Clinical metagenomics for infectious corneal ulcers: Rags to riches? Ocul Surf. 2020. January;18(1):1–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Upadhyay MP, Karmacharya PC, Koirala S, et al. Epidemiologic characteristics, predisposing factors, and etiologic diagnosis of corneal ulceration in Nepal. Am J Ophthalmol. 1991. January 15;111(1):92–9 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.