Endophthalmitis refers to the inflammation of the ocular cavities and their immediate adjacent structures without extension beyond the sclera, usually secondary to infection. Endogenous endophthalmitis (EE) results from haematogenous spread of microorganisms in patients with bacteraemia or fungaemia into the eye and represents 2–15% of endophthalmitis cases.1-3 The last study conducted on the incidence of bacterial endogenous endophthalmitis (BEE) in Singapore was published in 2000 by Wong et al.4
The study reported Klebsiella spp. as the most common organism causing BEE especially among patients with hepatobiliary infections. Since then, it has become a routine practice to refer patients with Klebsiella bacteraemia to the ophthalmology service for endophthalmitis screening, independent of visual symptoms. However, there has been little new Singapore data evaluating the incidence and profiles of BEE, especially that of Klebsiella endogenous endophthal-mitis in recent years.
We retrospectively reviewed the charts of all patients diagnosed with BEE in a tertiary centre in Singapore from 1 January 2014 to 31 December 2021. In this study, endophthalmitis was defined clinically by the presence of inflammation of the posterior segment with or without anterior chamber inflammation. We also collected the blood culture results positive for Klebsiella pneumoniae over the study period to review the incidence of EE among patients with Klebsiella bacteraemia.
We found 15 patients diagnosed with BEE at our tertiary centre during the study period. All patients had unilateral BEE. The patients’ ages ranged from 36 to 74 years, with the median being 57 years. There was a slight male preponderance seen in our cohort (60.0%, n=9). Eleven (73.3%) patients were Chinese, 3 (20.0%) were Malay and 1 (6.7%) was Indian.
Seven patients (46.7%) initially presented to the emergency department for evaluation of their visual symptom(s), while the remaining 8 patients were diagnosed and treated for BEE as inpatients All patients received intravenous antimicrobial therapy (Table 1). Twelve out of 15 patients (80.0%) had underlying diabetes mellitus.
Most of the patients (86.7%, n=13) presented with blurring of vision, with 3 of them reporting concomitant floaters (Table 1). One patient reported having floaters without blurring of vision. Presenting visual acuity (VA) ranged from 6/7.5 to perception of light with 7 patients (46.7%) having vision of hand movement only or worse. Eight patients (53.3%) had hypopyon. All patients had vitritis, with 5 of them (33.3%) having retina and/or choroidal abscess on presentation.
The number of intravitreal injections received by the group during their treatment period ranged from 1 to 6 injections, with both the mean and median being 3 injections. Each injection might constitute either a monotherapy or combination of either vancomycin, amikacin, and/or ceftazidime. Seven patients (46.7%) underwent vitrectomy (Table 1). Final VA ranged from 6/6 to no perception of light. One patient had an eye eviscerated, and 2 patients passed away from their systemic infection.
Liver abscess (n=4, 26.7%) and urinary tract infection (n=3, 20%) were the most frequent sources of systemic infection. The infective sources of bacteraemia in the rest of our patients are summarised in Table 1. One (6.7%) patient (Patient 10) had septicaemia without any localising foci of infection despite extensive investigations.
Nine (60.0%) patients had positive blood cultures with gram-positive bacteria being the predominant organism—4 patients with Streptococcus agalactiae (Group B) bacteraemia and 2 patients with methicillin-sensitive Staphylococcus aureus bacteraemia. Three patients (20.0%) had gram-negative bactaeremia and all 3 had K. pneumoniae detected in their blood cultures. Six patients (40.0%) had negative blood cultures. Of these, 4 patients (26.7%) had culture-proven sources of extra-ocular infection (Patients 3, 6, 9 and 13). The remaining 2 patients (13.3%; Patients 1 and 5) had their sources of infection established clinically and radiologically.
During the same study period, there were 2,014 patients with blood cultures positive for K. pneumoniae being treated in our centre. Three of these patients were diagnosed with endogenous endophthalmitis (Patients 12, 14 and 15). Patient 12 had his blood culture done in another centre and was subsequently transferred to our hospital for further treatment of his liver abscess. This brought the incidence of Klebsiella BEE in our centre to 0.1% (2 out of 2,014) over 8 years.
Diabetes mellitus is the most common underlying systemic risk factors among our patients having BEE, with liver abscess and urinary tract infection being the most frequent sources of systemic infection in our cohort, similar to the trend observed in earlier studies on BEE.2,4,5 The last Singapore study on BEE by Wong et al. found gram-negative bacteria, particularly Klebsiella spp., to be the predominant causative organism in their cohort.4 In comparison, despite having an all- Asian cohort, our study found that gram-positive bacteria accounted for 60.0% of positive blood cultures. Interestingly, all 3 patients with gram-negative bacteraemia in our study had K. pneumoniae in their blood cultures.
An overwhelming proportion of our patients were symptomatic at the point of presentation, with 14 out of 15 patients reported to have blurring of vision and/ or floaters. It was not possible to obtain history in one patient as she had altered mental status from underlying septic shock upon initial review. Of note, it is a routine practice in our centre to refer all patients with Klebsiella bacteraemia to the ophthalmology service for endophthalmitis screening. Considering the low incidence of Klebsiella EE among patients with Klebsiella bacteraemia, it may be worthwhile to reconsider the current workflow of routine Klebsiella EE screening to one that is based on visual symptoms. Patients who are unable to give reliable history should still be screened for endophthalmitis by an ophthalmo-logist.
In summary, BEE was exceedingly rare, averaging around 2 cases per year in our centre. Hepatobiliary and urinary tract infections are the most frequent infective sources of our patients with BEE, with diabetes being the most common underlying comorbidity. As most patients with BEE are symptomatic upon presentation, routine endophthalmitis screening is of low value and should be reserved for patients with symptoms and those who cannot provide reliable history.
We would like to thank Dr Marcus Tan CJ and Mr Hariz ZW Liew from the National University Hospital with data collection.
- Durand ML. Bacterial and Fungal Endophthalmitis. Clin Microbiol Rev 2017;30:597-613.
- Jackson TL, Paraskevopoulos T, Georgalas I. Systematic review of 342 cases of endogenous bacterial endophthalmitis. Surv Ophthalmol 2014;59:627-35.
- Okada AA, Johnson RP, Liles WC, et al. Endogenous bacterial endophthalmitis. Report of a ten-year retrospective study. Ophthalmology 1994;101:832-8.
- Wong JS, Chan TK, Lee HM, et al. Endogenous bacterial endophthalmitis: an east Asian experience and a reappraisal of a severe ocular affliction. Ophthalmology 2000;107:1483-91.
- Cho H, Shin YU, Siegel NH, et al. Endogenous Endophthalmitis in the American and Korean Population: An 8-year Retrospective Study. Ocul Immunol Inflamm 2018;26:496-503.