Investigation of the Microbiome of the Cornea in Microbial Keratitis

Microbial keratitis (MK) is an ophthalmological emergency that can lead to sight threatening complications such as corneal scarring, perforation, endophthalmitis and ultimately blindness. The patient requires aggressive topical antimicrobial therapy and close treatment response monitoring, often including hospitalisation followed by frequent outpatient visits.

Improving outcomes depends on rapidly identifying the causative microorganism. Currently the likely causative microorganism is only isolated in around 30 to 40% of cases using traditional scraping methods and standard conventional diagnostic culture (CDC), with results typically taking up to 4 days to become available to the clinician. More sensitive methods such as microorganism targeted polymerase chain reaction (MTPCR) and metagenomic analysis increase the potential to detect microorganisms but may increase the likelihood of picking up commensal microorganisms, making it difficult for the treating clinician to interpret which of the isolated organisms is likely to be causative in MK.

One of the barriers to identifying the organisms has been the difficulty in collecting samples from the cornea. In 2015, a group led by Professor Kaye at the University of Liverpool developed a non- invasive corneal sampling methodology using a corneal impression membrane (CIM) made from polytetrafluoroethylene. This was shown to have a significantly higher overall isolation rate compared to conventional scraping methods. As it is minimally invasive, the CIM sampling method offers a unique opportunity to sample both the affected cornea and unaffected cornea of patients presenting with MK as well as the eyes of patients unaffected by MK. This will enable a much better understanding of the clinical significance of isolated organisms in MK.

Metagenomic next-generation sequencing (NGS) techniques enable the genomic analysis of all the microbes in a sample, giving a wealth of information regarding the presence and interaction of microorganisms as well as the presence or absence of antimicrobial resistance (AMR) genes. Previous studies attempting to characterise the ocular surface microbiome in health using metagenomics have demonstrated a diverse resident homeostatic microbiome, however, have been limited to swabs taken from the lid, conjunctiva and tears rather than the cornea. Variations in the microbiome have been associated with conditions such as dry eye syndrome, contact lens wear and infectious pathologies, including a pathological microbiome dominated by pseudomonas spp. in a small number of patients with bacterial keratitis. Despite the cornea being the surface affected by MK, there is no data regarding the healthy corneal microbiome because traditional sampling methods have been invasive.

Despite NGS generating personalised results in a matter of hours, its sensitivity to detect all organisms present without an understanding of their clinical significance is a barrier to its introduction into clinical practice. In this project, the CIM sampling method and NGS processing will be used to improve understanding of the causative organisms in MK. Furthermore, through sampling both the affected and unaffected eyes of MK patients and from those without MK, a diagnostic rule will be developed to identify and discount commensal organisms. The inclusion of different control groups will also provide information on the influence of contact lenses and eye drops on the microbiome, which have a role both on the development and treatment of MK. This will greatly facilitate current interpretation of corneal sample results and facilitate potential future use of NGS in routine ophthalmic clinical practice.

Aims and objectives

The overall aim of the project is to better define the pathogenic microorganisms in patients with MK through a better understanding of the corneal microbiome in health and disease.

This will be achieved through the following objectives.

1. Using NGS, analyse the corneal microbiome of the affected and unaffected eye of patients with and without MK and compare with simultaneous results from CDC and MTPCR.
2. Determine the microbiological spectrum of the cornea, ocular surface and contiguous structures, in patients with MK, healthy controls, contact lens wearers and eye drop users.

Study Design:

Design: Prospective observational diagnostic study and methods-comparison.

Study Duration: Three years.

Number and type of subjects:

151 patients with MK will be recruited. This is based on the following sample size calculation: Current standard diagnostic reference method (CDM) has a diagnostic accuracy of 40%. For clinical implementation of NGS, diagnostic accuracy will be required to increase to at least 80%. NGS can only increase the proportion of detected micro-organisms. To estimate a diagnostic rate of 80% with a 95% confidence interval (of width 15%) 137 participants are required. This calculation assumes that the smallest diagnostic group has a prevalence of 20%. To adjust for a 10% rate of possible loss of follow-up, we will recruit 151 patients.

We will also recruit 90 participants to four control groups:

1. 20 patients with no history of MK who use no eye drop medication
2. 20 patients with no history of MK who are contact-lens wearers
3. 20 patients who have no history of MK but are on eye drop treatment for glaucoma. This group has been included to assess for changes in the corneal microbiome that could be secondary to drop treatment.
4. 30 patients with keratoconus who are undergoing cross-linking will be recruited. These participants as part of the routine cross-linking procedure will have their corneal epithelium removed. This removed epithelium from an otherwise healthy corneal surface will allow for a direct comparison between the corneal microbiome characterised from the CIM and that characterised directly from the epithelium.

Clinical data collection

At presentation, patient demographics, risk factors (ocular surface disease, contact lens wear, previous MK), and treatment received in the past or using at presentation will be obtained from interview, together with ulcer characteristics (major and minor axes of the corneal ulcer measured using a continuous scale). A standardised data collection form will be utilised.

Best corrected visual acuity (BCVA), size, location and depth of the ulcer will be recorded using a slit lamp biomicroscope. Ulcer size (minor and major axes) and location (minimum distance from the limbus) will be measured. Ulcer depth will be measured on a nominal scale of 1-4 based on percentage of remaining corneal thickness underneath the ulcer.

Specimen Collection

For each participant, the following samples will be collected:

• Three CIMs from the affected and unaffected eyes of participants with MK, or from one eye of control participants.
• Swabs from conjunctiva of the participant's affected and unaffected eyes, upper eyelids and nose.

Three CIMs will be applied (PTFE Millipore culture insert, pore size 0.40micrometres) to the surface of the cornea for 5 seconds and one CIM to the participants inferior fornix conjunctiva for 5 seconds using sterile gloves. A topical anaesthetic (one drop of 0.5% proxymethacaine) will be instilled into the lower conjunctival fornix prior to application of the CIMs. The first CIM sample obtained from the participants cornea will be placed and transported in a bottle containing 0.5mL of BHI broth for culture. The CIMs obtained from the participants cornea and the inferior conjunctiva will be kept dry in a sterile tube. In addition to the collection of samples described, the corneal epithelium removed from patients undergoing corneal cross-linking will be placed in a sterile eppendorf and stored at -80 C. Swabs will be taken from the participant's conjunctiva, upper eyelids and anterior nares in order to characterize the surrounding microbiome and endogenous sources.

MK participant follow up

Participants presenting with clinically suspected MK will be followed up at 3 days, 7 days and 1 month following presentation, as per normal procedure for follow-up of MK. At each follow up appointment BCVA, size and depth of the ulcer will be recorded. All samples will be repeated at each follow up appointment from the participants affected eye only. These will include three corneal CIMS, conjunctival, upper eyelid and nasal swabs.

Processing of samples and identification of microorganisms:

1. CDC (established diagnostic standard): Blood, chocolate, and Sabouraud's dextrose agar plates and a 24-hour subculture of the BHI broth inoculated with the bacterial swabs and membranes will be examined for evidence of bacterial growth after 24 and 48 h incubation.
2. MTPCR:DNA will be extracted and quantified from the CIM and virology conjunctival swab. A multiplex microorganism targeted PCR (MTPCR) master mix, including primers for HSV-1, HSV2, acanthamoeba, 16S and 18S ribosomal DNA will be prepared and PCR performed using a real-time PCR instrument
3. NGS: Metagenomic testing will be performed using high-throughput platforms, such as the Illumina HiSeq 4000. Sequencing depth will be determined empirically. Genome alignments, filtering and pathogen calling will be performed using established pipelines. An analysis approach based on read classification will be applied for taxonomic classification. A subtractive analysis method based on removal of host DNA and healthy control eye metagenome will be applied.

Statistical analysis:

A comparison of isolation rates and identified bacteria obtained from CDC, MTPCR and NGS processing of MK corneal samples will be made. This will be visualised using a Venn Diagram.

For NGS data, we will work with the Centre of Genomic Research (CGR) to assign taxonomy labels and calculate relative abundances in each sample. For detection of AMR, software will be used to directly map reads to those in a comprehensive AMR database and report the AMR genes present.

Microorganisms identified in the eyes with MK will then be compared to the control fellow eye and other control groups and subtractive bioinformatics methodology applied to identify the most likely pathogenic organisms compared to those seen in the healthy corneal microbiome.

Comparisons of the relative abundance of microorganisms obtained from MK corneal samples over the participant follow up visits will be used to evaluate longitudinal changes in the corneal microbiome during treatment and resolution of MK.

A direct comparison between the relative abundance of microorganisms isolated from participants cornea, conjunctiva, eyelids and nose will be made to identify any possible endogenous sources of infection for MK. Overlaps of microorganisms between the sample sites will be visualised using a Venn diagram and non-parametric tests used to assess for differences between the sample sites.