Discussion
Clinical outcomes in MK depend on a complex interplay of host factors, microorganism virulence factors and the MIC of an antimicrobial agent against the respective microbe. Previous corneal sampling methods have relied on specialist equipment and skills. The CIM does not require specialist skills or equipment and has a higher presumed pathogen isolation rate compared with traditional scraping methods in a UK population.13 14 The CIM therefore, lends itself to being used in settings where such skills and equipment are unavailable.
In this study, we successfully demonstrated that the CIM could be used to obtain corneal samples and identify a possible causative microorganism using both culture and targeted PCR techniques in 83% of patients presenting with presumed MK in Malawi. We also demonstrated that it was feasible to collect such samples in a rural setting where immediate access to a microbiology laboratory was not possible.
The average time between symptom onset and presentation was longer than typically seen in the UK with ulcers being much larger on presentation (average major ulcer diameter on presentation in this study was 4 mm vs 2.14 mm seen in an average UK population).13 This isolation rate was higher than the 65.5% isolation rate that we demonstrated using the same CIM and sampling technique in a UK population and this is likely due to the longer presentation times, less antimicrobial use and more advanced disease on presentation.14
Due to a lack of specialist equipment, corneal samples are not routinely carried out in Malawi and therefore, the microbiological profile of MK is unknown. The CIM has previously been shown to increase isolations of presumed pathogenic microorganisms in MK compared with traditional scraping methods in a UK population13 14 and therefore direct comparisons between the CIM and traditional scraping methods in this population were not made. It could be argued that as the microbiological profile of MK in Malawi is unknown, the possibility that the CIM may not have detected all microorganisms that traditional corneal scrapes could have detected cannot be ruled out. That said, this study was carried out in the winter months of Malawi where the climate is typically cool and dry. Gram-positive bacteria were the most common microorganisms isolated in this population and this is consistent with other studies that have been carried out in temperate climates using traditional corneal scraping methods.36
Several predisposing factors are implicated in MK which differ between countries. In this study, only 15.5% of patients had a history of previous trauma which is lower than other studies carried out in other sub-Saharan African countries.10 37 38 This low incidence of trauma, together with the cool and dry season that this study was undertaken in may account for the absence of fungal infections. In studies in sub-Saharan Africa (Ghana, Tanzania), that used similar fungal culture methods (Sabouraud glucose agar plates) to this study, fungal keratitis accounts for most positive culture cases and it is therefore assumed more fungal infections would be identified in the hot and rainy season.12 39 The CIM has successfully identified fungal infections previously using the same sampling methodology and therefore this would not account for the fungal isolation rate in this study.14 In the future, the microbiological profile of MK in Malawi should be studied over a longer time period to better understand seasonal variability of pathogen prevalence throughout the year.
While contact lens wear is the most common risk factor in industrialised countries, trauma and ocular surface disease are more significant in low-resource settings.40–43 In this study, contact lens wear was non-existent and there were higher rates of trichiasis and traditional medicine use compared with other sub-Saharan MK studies.10 44 45 Traditional medicine use is widely practised in Malawi and can lead to damage to the ocular surface and delays in presentations to hospitals and clinics.46 The different microorganism spectrum seen in this population with low rates of Pseudomonas sp infection and high rates of mixed infections compared with those seen in the UK are likely to reflect these different predisposing factors14
The prevalence of HIV infection (17%) was higher than that reported for the general population in Malawi (10%).47 This observation is suggestive of increased susceptibility to MK among HIV-positive individuals and is consistent with that found by Burton et al,10 in Tanzania. Like Burton et al,10 we found no association between the type of MK and HIV although previous studies have indicated that HSV-1 and fungal keratitis are more prevalent in HIV-positive patients.48 49 This study does emphasise the argument made by Burton et al10 for ophthalmologists offering MK patients referral to HIV testing services in such settings.
The percentage of CNS species and the numbers of samples obtained with more than one isolate in this study are similar to that seen using the CIM in a UK population and lead to the question of which of these are likely to be contributing to disease and which may either be contaminants or part of normal ocular flora in this population.14 The comparatively large surface area of the 12 mm diameter CIM used in this study compared with the median corneal ulcer size of 4×2.9 mm may favour the uptake of commensals from the patients’ unaffected cornea and use of a smaller diameter CIM may reduce the culture of these bacteria. The CIM is minimally invasive and therefore can be used to obtain corneal samples from healthy eyes. Further work, taking samples from asymptomatic or unaffected eyes of patients with MK may differentiate grades of pathogenic and non-genic microorganisms and inform future treatment strategies.
In this population, we demonstrated that the detection rate for HSV-1 was higher using the CIM than the conjunctival swab and this is in keeping with our previous work that demonstrated that the CIM is more likely to detect the presence of HSV-1 in suspected herpes simplex keratitis.15
The MIC is used to determine susceptibility criteria so that an appropriate antimicrobial treatment can be chosen. Ideally, the antimicrobial with the lowest MIC and an expectation to achieve a concentration above the MIC in the cornea should be selected. If the MIC of the used antimicrobial is below the first quartile concentration of the antimicrobial in the cornea, then it is likely that the microorganism is susceptible.24
The availability of antimicrobial eye drops in public health facilities in Malawi is erratic due to funding constraints. However, ciprofloxacin and gentamicin are the most available and these are the most prescribed antimicrobial agents for the treatment of MK. In this study, resistance was present to both agents in all bacterial groups other than S. aureus. Moxifloxacin offered the best coverage for both Gram-positive and Gram-negative isolates when susceptibility was determined using known antimicrobial first quartile concentrations whereas chloramphenicol offered the best coverage for all bacterial isolates when susceptibility was determined by EUCAST breakpoints. Although the numbers for each bacterial group are small and the results are limited to the dry season, these susceptibility data are the first of its kind to be reported on African keratitis isolates. Future research is needed to study antimicrobial resistance to topically available antimicrobials throughout the year in Malawi. It is also important to acknowledge the limitations of using in vitro sensitivity results based on systemic administration and more research is needed to study the bioavailability of antimicrobials reaching the cornea and aqueous and to establish ophthalmic breakpoints for topically applied antimicrobials.
In this study, we present the whole-genome sequences of seven Malawian S. aureus keratitis isolates. We observed a variety of virulence genes that have previously been associated with ocular S. aureus isolates or S. aureus keratitis.22 23 There are currently no studies looking at ocular S. aureus virulence factors in Africa to compare our data to and data concerning the virulence and pathogenesis of sub-Saharan Africa S. aureus clinical strains overall is limited. As the number of S. aureus keratitis isolates obtained in this study is small, and patients had multiple confounding host factors present, it is difficult to correlate the presence or absence of virulence factors with disease severity. Clinical outcome data were not collected as part of this study and therefore it is not possible to correlate the presence of virulence factors with clinical outcomes. Panton-Valentine leucocidin (PVL) gene expression was comparatively higher in the Malawian S. aureus keratitis isolates (3/7, 42.9%) than that seen in S. aureus keratitis isolates from the UK (9.5%).50 PVL is associated with poorer clinical outcomes in MK50 and is known to be high among clinical S. aureus strains in Africa (74% of methicillin-susceptible S. aureus isolates) compared with Europe (0.2%) and the USA (11.5%).51–54 Further investigation into the role of virulence factors in MK pathogenesis in Malawi is warranted.
In this study, we have demonstrated that the CIM sampling technique can feasibly be used to collect corneal samples from patients with suspected MK in a setting such as Malawi where access to specialist skills and equipment can be limited. Furthermore, we have demonstrated that corneal samples could feasibly be collected in rural settings where immediate access to a microbiology laboratory is not possible. Although this study is small in numbers and findings are limited to the dry season, it is the first study of its kind to characterise the microbiological profile of MK in Malawi. Results from this preliminary study support the use of the CIM in a larger study aiming to characterise the microbiological profile of MK in Malawi in a larger population over a longer time period and could also be feasibly used in other such settings to do the same. A better understanding of the causative microorganisms, microorganism susceptibilities to antimicrobials and microorganism virulence factors is greatly needed to improve patient outcomes and reduce corneal blindness in Malawi.