MATERIALS AND METHODS

Patients

A prospective study of suppurative keratitis was conducted in Ghana and southern India between June 1999 and May 2001. Patients were recruited at the eye unit of Korle Bu Teaching Hospital in Accra, Ghana, and also at two rural hospitals in Agogo (Ashanti region) and Bawku (Upper East region). In India, patients who presented with suppurative keratitis at two rural “taluk” hospitals in the Tiruchirapalli district of the southern state of Tamil Nadu were included in the study in addition to patients from the main study centre, Joseph Eye Hospital (JEH) base hospital.

All patients presenting with suspected suppurative keratitis were included in the study. Corneal ulceration was defined as loss of corneal epithelium with underlying stromal infiltrate and suppuration associated with signs of inflammation, with or without hypopyon.⁷ Patients with suspected or confirmed viral keratitis were excluded from the study. Patient consent was mandatory for inclusion in the study.

Clinical examination and laboratory investigation

Each patient was examined at the slit lamp; clinical features were noted and a drawing made for patient records. A corneal scrape was performed by an ophthalmologist using a sterile 21 gauge needle, or blunt Kimura spatula, following the instillation of local anaesthetic (amethocaine hydrochloride 0.5%, without preservative). In India, lignocaine (lidocaine) 4% was routinely administered, anaesthetic without preservative was sometimes used. If a patient was taking antibiotics at the time of presentation to the clinic treatment was stopped and investigations were delayed for 24 hours.

Corneal material obtained from scraping the ulcer was smeared onto two slides which were stained with Gram stain and lactophenol cotton blue mountant for microscopic examination. Material was inoculated directly onto 5% sheep’s blood agar, Sabouraud glucose agar, and into Sabouraud broth. In India, inhibitory mould agar (IMA) was substituted for Sabouraud glucose agar, as was routine practice of the laboratory. At the tertiary centres, if there was sufficient corneal material, additional culture media were inoculated (brain heart infusion (BHI) broth, chocolate agar, and cysteine tryptone agar). Blood agar plates, cystine tryptone agar, and BHI broth were incubated at 37°C. Sabouraud glucose agar plates, Sabouraud broth, and IMA plates were incubated at 27°C. A non-nutrient agar plate was inoculated if Acanthamoeba keratitis was suspected and an additional slide was prepared for microscopy.

Bacteria were further identified using routine biochemical identification tests and selective media. Identification of fungi was carried out in London and at the Mycology Reference Laboratory, Bristol, in addition to the overseas centres. Filamentous fungi were identified according to the macroscopic appearance of cultures on Sabouraud glucose, potato dextrose and cornmeal agars, and microscopic appearance of conidia and spore bearing structures. Yeasts were identified to species level using the germ tube test (incubation for 1.5 hours, at 37°C, in horse serum), Auxacolor and API kits (Bio Mérieux).

Microbial cultures were considered to be significant if growth of the same organism could be demonstrated on two or more solid phase cultures. Similarly, if there was semiconfluent growth at the site of inoculation or growth on one solid medium consistent with microscopy (that is, appropriate staining and morphology with Gram stain); or semiconfluent growth at the site of inoculation on one solid medium (if bacteria); or growth of the same organism on repeated scraping.^1,7,15,16 If, by microscopy, hyphae were observed in corneal tissue, but failed to grow in culture, the causative organism was reported as fungal.

RESULTS

A total of 1090 patients presenting with suppurative keratitis were enrolled in the study; 290 in Ghana and 800 in India. One patient from Ghana and one patient from India presented with bilateral infection. Fungi were identified as the dominant causative agent of infection (including mixed infections): 44% in southern India and 37.6% in Ghana. Bacteria were isolated from 29.3% of cases in south India and 13.8% in Ghana (Table 1). In each case of mixed infection a single bacterial species was associated with a single fungal species. This study revealed a higher percentage of “mixed” infections in India than Ghana: 5.5% versus 1.4%.

Seven patients presented with suppurative keratitis due to infection with Acanthamoeba species in south India. The first reported case of Acanthamoeba keratitis in Ghana was diagnosed during this study (BJO, accepted for publication). None of these patients were contact lens wearers.

In India 249 (31%) cases had no definitive laboratory diagnosis; Gram positive cocci were reported from the microscopy in 80% (200/249) of these cases; however, cultures were negative. It was not possible to determine the aetiological agent of 144 (50%) corneal ulcer cases in Ghana (Table 1). Microscopy and culture results were negative in 60% (87/144) of cases. Scanty bacteria were seen in 14% (24) of Gram stained smears. Of the remaining 33 cases which were microscopy negative, filamentous fungi were isolated in culture from 51% cases and bacteria from 49% cases but growth was not considered to be significant, using the criteria stated (see Materials and methods).

Pseudomonas species were the commonest bacterial isolates from corneal ulcers in Ghana (52.5%); P aeruginosa being the commonest reported species. Streptococci (20%) and staphylococci (10%) were commonly associated with corneal infection (Table 2). In India streptococci accounted for 46.8% of corneal ulcers followed by staphylococci (26.8%) and pseudomonads (14.9%). In both countries more than 80% of bacterial isolates were represented by these three genera.

Fusarium spp and Aspergillus spp were isolated from 61% of all fungal infections and comprised 83% of identified fungal isolates (Table 3). Although Fusarium species were the most prevalent fungal pathogens reported in Ghana, differences in the spectra of fungi were observed between the participating centres. In Accra (southern Ghana), 63% (27/43) of fungi isolated were Fusarium spp compared with 31.3% (15/48) at Bawku Hospital (northern Ghana), where a greater variety of fungi were isolated. Lasiodiplodia theobromae was the third commonest fungal isolate in Ghana. In India a greater number, and diversity, of dematiaceous moulds were cultured; Curvularia species were most frequently isolated.

DISCUSSION

Fungi were identified as the principal aetiological agent of corneal ulceration: 44% and 37.6% of cases in India and Ghana. Earlier studies in the same regions reported a similarly high proportion of fungal keratitis.^6,12 A review of the literature shows that there are distinct patterns of geographical variation in the aetiology of suppurative keratitis and considerable variation in the proportion due to fungi has been documented (Table 4). The proportion of corneal ulcers caused by filamentous fungi increases towards tropical latitudes (Fig 1). In more temperate climates, fungal ulcers are uncommon and are more frequently associated with Candida species than filamentous fungi.^21–24 Houang et al⁴³ reviewed the relation of fungal keratitis to climate concluding that, although a higher incidence of fungal keratitis could be expected in countries with similar annual rainfall and temperature range, this was not always so and was also dependent on the extent of urbanisation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Suppurative keratitis due to fungi as a proportion of total number of cases, by latitude.

Aspergillus spp and Fusarium spp are the most frequently reported fungal pathogens isolated from cases of fungal keratitis in the tropics.⁵ In both Ghana and south India the most commonly isolated fungal pathogens in the current series were Fusarium spp. Other studies in south India have reported Fusarium spp to be more common than Aspergillus species. Fusarium spp have also been found to be the principal fungal pathogen in Florida, Paraguay, Nigeria, Tanzania, Hong Kong, and Singapore (Table 4). Aspergillus species predominate in northern India, Nepal, and Bangladesh.^1,9,13,32,34 This phenomenon may be explained by differences in climate and the natural environment. A similar pattern was also observed in Ghana. Although Fusarium spp were the most commonly isolated fungi at all of the centres in Ghana, moulds with enteroblastic conidia adhering in dry chains—for example Aspergillus spp and Paecilomyces spp, were more frequently isolated from patients in the north of the country where the environment is drier and dustier, than in the more humid south. As observed by Khairallah²⁶ the high proportion of corneal infections caused by Aspergillus spp in drier climes may be due to the fact that spores of Aspergillus spp can tolerate hot, dry weather conditions. Aspergillus spp also predominate in more temperate latitudes.

A significant increase in the number of reported cases of suppurative keratitis was observed during the harvest period and windy seasons. However, the proportion of corneal ulcers due to fungi remained consistently high throughout the year. Other authors have made similar observations, noting an increase in cases of fungal keratitis during the dry, windy seasons compared with the wet, humid months of the year.^4,19 This trend is likely to be a direct consequence of increased agricultural activity before and immediately following the rains. Some studies have reported an increase during the hot and humid months.^8,35

The majority of filamentous fungi associated with corneal ulceration in the tropics are saprobic, thermophilic moulds that are found widely in this environment. These are ubiquitous in the soil and vegetation. Fusarium species are common plant pathogens, particularly of cereal crops or saprophytes of plant debris and are found in soil.⁴⁴ The aspergilli are ubiquitous and have been found almost everywhere on every conceivable type of substrate, including soil and decaying organic debris. Some of the less common isolates such as Bipolaris spp and Exserohilum spp are pathogens of grasses. Curvularia spp mostly occur on dead plant material.⁴⁴ Although injury by vegetable matter is considered to be predictive for fungal keratitis, in this study fungal corneal ulcers were more often preceded by dust or mud particles in the eye.

A shift in the predominant bacterial pathogens was observed when compared with earlier findings. In Ghana, >50% of bacterial isolates were Pseudomonas species compared with 27% of isolates in a previous study by Hagan et al.⁶ There is a paucity of information in the literature regarding the aetiology of bacterial corneal ulcers in sub-Saharan tropical Africa and, therefore, comparisons are only possible with similar geographical and climatic regions outside the continent of Africa. Pseudomonas species were identified as the commonest bacterial isolate in a study of 142 cases of suppurative keratitis in Bangladesh in a study by Dunlop et al and also by Williams et al, who found that 40% of bacterial isolates were Pseudomonas spp.^4,33 A predominance of Pseudomonas species has been reported in Hong Kong, Florida, and Paraguay.^19,25,43 In reports from Nepal and south India, Gram positive cocci have been reported as the primary cause of bacterial keratitis.^1,7,9 In this study, Gram positive cocci accounted for the majority (74%) of bacterial isolates in India, as found previously by Thomas et al.¹² The proportion of bacterial ulcers caused by Streptococcus spp increased from 18.5% (1986) to 46.8% in this study. Similar diagnostic criteria were used in the previous study and, therefore, the trend may be attributable to a genuine change of bacterial flora within the geographical area, as influenced by climate and environment.

Traditional diagnostic laboratory methods, including microscopy and culture may be negative despite a clear clinical presentation of suppurative keratitis. This was true in 50% and 31% of cases in Ghana and India, respectively. This may be due to difficulty in obtaining sufficient corneal material for conventional investigation. This applies particularly to large, late stage ulcers, because of the risk of perforation and, conversely, early stage, small ulcers from which little material is available. It is imperative that the quality and quantity of specimen is optimal for accurate laboratory diagnosis. Self administration of antibiotics by patients before seeking medical attention has been thought to affect the recovery of organisms in culture.⁴ It is possible culture positivity correlated with inappropriate antibiotic usage.

Although there were a high proportion of corneal ulcers in Ghana for which the aetiology was not determined, the number of proved fungal ulcers at each of the centres was greater than the number of proved bacterial ulcers. It is therefore not unreasonable to assume that this trend could be extrapolated.

It is usually not possible to determine the significance of bacteria observed by microscopy alone. Small numbers of Gram positive cocci may be contaminants from the lid margin. Conversely, the presence of fungal hyphae in corneal tissue is significant. In agreement with Dunlop et al the sensitivity, specificity, and predictive value of Gram stain microscopy is much higher for fungal ulcers than those caused by bacteria.⁴ In this study 95% of fungal infections could have been diagnosed based on the findings from microscopy alone. This is an important conclusion, since the majority of rural based clinics in areas where suppurative keratitis is a problem do not have culture facilities but may be able to perform simple microscopy. A wet mount preparation, using lactophenol cotton blue stain, was used as a supplementary stain in this study. Where scientific expertise, and/or resources, may be an issue, it is easier to observe fungal hyphae employing this method than using the potassium hydroxide (KOH) technique. Reports into the presence of fungi in the eyes of asymptomatic individuals, have shown that a wide variety of fungi may be transient in the conjunctival sac. This may be the case in as many as 37% of healthy eyes, thereby discounting the use of single culture.¹¹

In conclusion, it is imperative to know the “local” aetiology of keratitis in a particular region. Our comparative study demonstrates that there were differences between two countries with similar tropical climates at the same latitude. Equally, differences were observed within the same country. This is important information with regard to management, as many eye units do not have microscopy or culture facilities. Diagnosis is then dependent on clinical acumen and the treatment provided is, at best, empirical. It is clear that it is important to know the local aetiology, particularly if diagnosis is going to be reliant on clinical signs alone. Changing patterns of disease have also occurred in temperate, developed countries, as exemplified by the increase in keratitis due to Pseudomonas spp and Acanthamoeba spp in recent years. Awareness of changes in aetiology and antimicrobial resistance, when this information is available, are critical to managing keratitis cases.