Endogenous Endophthalmitis

Endophthalmitis can be classified as exogenous – following either trauma or an intraocular procedure, or endogenous and hence hematogenous in origin.^1–3 While exogenous is far more common than endogenous endophthalmitis, the latter is much more likely to be associated with an underlying and potentially life-threatening systemic source of, or risk factor for, infection, including recent hospitalization or indwelling catheter or intravenous drug use (IVDU). Moreover, endogenous endophthalmitis occurs more commonly in those with one or more contributors to immunosuppression, such as early or advanced age, malignancy, diabetes mellitus (DM), or use of corticosteroids or non-corticosteroid immunosuppressive agents. Globally, bacterial infections account for most cases of endogenous endophthalmitis, with Gram-positive bacteria, specifically Staphylococcus and Streptococcus species (spp), occurring more often in most settings, and Gram-negative bacteria, particularly Klebsiella spp, encountered more often in Asia.⁴ Endogenous fungal endophthalmitis, most often due to Candida spp, is associated with use of either indwelling catheters or IVDU and, as such, occurs predominantly in the developed world – particularly the United States (US).^5,6 Endogenous endophthalmitis due to Aspergillus spp represents a far less common fungal cause, but is distinguished both by its frequent association with debilitating disease and by a tendency to localize under the retina.^7–9 Prompt diagnosis and aggressive treatment are essential for the successful management of endogenous endophthalmitis, and even then the prognosis is poor in many patients. Six original articles^10-15 and three letters^16-18 in this issue of Ocular Immunology & Inflammation address the causes, characteristics, management, and outcome of eyes with endogenous endophthalmitis.

Cho et al¹⁰ described the characteristics, treatment, and outcome of 128 eyes of 108 patients – 60 American and 48 Korean – with endogenous endophthalmitis seen at separate referral centers in Boston, USA and Seoul, South Korea (SC), respectively, between 2006 and 2013. The mean age was about 60 years in both cohorts, although ranged from 22–91 years. Men constituted slightly more than half of each subgroup. Time from onset of ocular signs or symptoms to diagnosis tended to be about 1 week, with most diagnoses occurring between 4 and 11 days after onset. Bilateral involvement occurred in seven American (11.7%) and 14 SK (29.2%) subjects. Most affected eyes (69; 53.9%) were treated with vitrectomy together with both systemic and intravitreal antimicrobial agents; just over one-third (39.8%) received systemic and intravitreal antimicrobial agents without vitrectomy; and a minority (6.3%) received intravitreal antimicrobial agents alone. Among predisposing factors, DM (22.4% vs 54.2%), cardiovascular disease (19.4% vs 22.9%), malignancy (16.4% vs 22.9%), and hepatitis/cirrhosis (11.9% vs 29.2%) were common in both American and Korean subjects, respectively. IVDU was common in the American cohort (43.3%), but was not seen in subjects from SK. An extraocular source for the infection was identified in 86.7% of American and 75.0% of SK subjects. Among Americans the most common sources included fungemia from either IVDU or an indwelling catheter (26.7%), followed by cellulitis (9.0%) and endocarditis (7.5%), whereas among SK subjects, liver abscesses were most common (33.3%), followed by pneumonia (6.6%), fungemia (6.3%), and cellulitis (4.9%). No liver abscesses were identified in American subjects. While aqueous cultures were performed in a minority of patients (61 samples in the US and 17 samples in SK), they were infrequently positive (9.8% and 5.9%, respectively). Vitreous cultures were most often positive in the US (50 of 61; 49.2%), but less so in SK (5 of 38; 13.2%). Blood culture were positive in both settings (23 of 62; 37.1% US; 45 of 55; 81.2% SK). Bacterial causes were common in both the US (46.3%) and SK (65.6%), but the type varied, with Gram-positive species dominating in the US (44.8%; Staphylococcus spp 25.4%; Streptococcus spp 17.9%) and Gram-negative species more often causative in Seoul (44.3%; Klebsiella spp 37.3%). Candida spp (34.3% in US vs 16.4% SK) were seen in both setting. Aspergillus spp were isolated in four (6.0%) of American subjects. Approximately 18% of subjects were culture-negative in both cohorts. While duration of follow-up was neither standardized nor well-documented in the study, vision at last visit was generally poor, and worse in eyes with bacterial as opposed to fungal endogenous endophthalmitis. This was particularly so for the more virulent bacterial infections due to Staphylococcus aureus, Streptococcus, and Klebsiella spp. In addition to causative organism, eyes that presented with worse vision also had worse outcomes. The authors observed no statistically significant influence of vitrectomy on final vision in the overall cohort, but did note a trend toward better vision in those eyes infected by the aforementioned highly virulent bacterial pathogens that underwent vitrectomy. The authors emphasized the importance of IVDU as a pre-disposing factor in the US, of hepatobiliary disease as a risk factor in Asia, and of underlying DM in both settings. Of note, DM is a recognized risk factor for Klebsiella infection,¹⁹ contributing to the increasing importance of this highly virulent pathogen as a cause of liver abscesses in patients with both DM and underlying hepatobiliary disease – both now common in Asia. DM also appears to be a risk factor for cytomegalovirus retinitis in non-human immunodeficiency virus (HIV) positive patients.^20–22

Silpa-archa et al¹¹ reported the characteristics of 42 eyes of 36 patients diagnosed with culture-positive endogenous endophthalmitis seen at a tertiary referral center in Bangkok, Thailand, between 2005 and 2015. Endogenous endophthalmitis constituted 16.5% of all endophthalmitis seen at this center for the period of the study. Nearly two-thirds (63.9%) were male and the median age was 58 years (range 19–77 years). Median time from onset of symptoms or signs to first treatment was 7 days (range 1–40 days). DM was the most common predisposing factor (30.6%). The inflammation was severe (2+ or greater) in 80.5%, associated with hypopyon formation in 61.0%, and bilateral in 14.0%. Median vision at presentation was light perception. Culture-based yield was highest from the vitreous (22 of 38; 57.9%), followed by body fluid/tissue (11 of 24; 45.8%), and then blood (11 of 30; 36.7%). Gram-negative isolates were the most common (20 of 41; 48.8%), particularly Klebsiella pneumonia (11 of 41; 26.8%) and Escherichia coli (4 of 41; 9.8%). Gram-positive isolates were nearly as common (18 of 41; 43.9%) and included Streptococcus spp (12 of 41; 29.3%), Staphylococcus spp (3 of 41; 7.3%), and Enterococcus spp (3 of 41; 7.3%). Fungus was cultured in three samples (7.3%), including two that grew Cryptococcus neoformans and one Aspergillus fumigatus. A source of infection was identified in 23 patients (63.9%), three of whom had two positive sources. The most common of sources included liver abscess (5 of 26; 19.2%) – four of which grew K. pneumonia and one for E. coli; and the urinary tract (5 of 26; 19.2%) – four of which had negative urine cultures and one of which grew E. coli. Among the 13 subjects with no identified extraocular source, the vast majority of whom were afebrile (12 of 123; 92.3%), three (23.1%) grew Enterococcus faecalis from the vitreous. All subjects received systemic antimicrobial agents; 35 of 41 involved eyes (85.4%) received intravitreal vancomycin and ceftazidime; two eyes infected by C. neoformans received intravitreal amphotericin B; and four eyes underwent urgent enucleation for impending panophthalmitis. Vitrectomy was performed in 24 of 41 eyes (58.5%). Median follow-up was 3 months (range 0–72 months) and complications occurred in more than half of all affected eyes (22 of 41; 53.7%), most commonly ocular hypertension (9 of 41; 22.9%) and rhegmatogenous retinal detachment (6 of 41; 14.6%). Among 39 eyes of 34 patients for whom 1-month follow-up was available, multivariate analysis showed that equal to or better than counting-fingers vision was associated with better vision and less severe inflammation (< 2+) at presentation. Overall, however, more than two-thirds of affected eyes had vision worse than counting-fingers at this 1-month follow-up. The authors emphasized the poor prognosis of endogenous endophthalmitis overall and confirmed the importance of DM and hepatobiliary disease as predisposing factors and of Gram-negative bacteria, including K. pneumonia and E. coli, in Asian populations.

Rishi et al¹² described E. faecalis endophthalmitis in 19 children 18 years of age or less and 18 adults seen at a referral center in Chennai, India, between 1995 and 2015. Among all 214 eyes of children with endophthalmitis seen over roughly half of this study period, 29.0% had endogenous endophthalmitis. A total of 17 post-traumatic and 2 endogenous (10.5%) E. faecalis endophthalmitis cases were seen in children over the full study period. Both children with endogenous E. faecalis endophthalmitis were febrile and in each diagnosis was based on vitreous culture since both blood and urine cultures were negative. Among the 18 eyes of adults with E. faecalis endophthalmitis, two (11.1%) were endogenous and both were based on vitreous culture. While long-term vision was not presented in all four cases, the authors noted that more than half of all eyes with E. faecalis endophthalmitis are left with on light perception vision.

Cherisch et al¹³ systematically reviewed the characteristics, treatment, and outcome of Listeria monocytogenes endophthalmitis, an entity first described by Ernest K. Goodner and Masao Okumoto from The Francis I. Proctor Foundation just over 50 years ago.²³ The Proctor patient was a 76-year-old retired plumber who presented with an acute hypertensive anterior uveitis; he raised goats and enjoyed drinking raw goat’s milk. The common finding of a pigmented or dark hypopyon was described by subsequent authors. The review summarized a total of 43 cases, 11 of which were previously unpublished. Of 37 where the source was either provided or identified, 33 (89.2%) were endogenous. Median age was 61 years (range 24–88 years), slightly more than half (57.1%) were male, and nearly half (47.5%) had one or more underlying factors contributing to relative immunosuppression, such as cancer, DM, or use of corticosteroids or non-corticosteroid immunosuppressive medications. The median time from presentation to diagnosis was 8 days (range 1–38 days) and the vast majority (84.4%) had 20/500 or worse vision in the affected eye. Among the 26 eyes were a hypopyon was reported (61.9%), the layered cells were noted to be either ‘tan’ or ‘dark’ in nine (34.6%). In approximately 90% of cases, the diagnosis was based on either culture of polymerase chain reaction-based testing of intraocular fluids. Combined use of multiple topical, intraocular, and systemic antibiotics was common. At final visit, approximately one-third of affected eyes had 20/63 vision or better and nearly 40% were 20/500 or worse. Such poor outcomes were two-fold more common in patients older than 65 years of age, although age as a predictor of outcome failed to achieve statistical significance. While the authors were unable to infer strong treatment recommendations for the case series reviewed, a recent review by Hof²⁴ recommended the combined use of intravenous ampicillin with intravitreal vancomycin as first line in culture-positive cases of endogenous Listeria endophthalmitis.

Ganesh and Priyanka¹⁴ described the characteristics, management, and outcome of 102 patients with unilateral ocular/periocular cysticercosis seen at a tertiary referral center in Chennai, India, between 2000 and 2015. The overall male:female ratio was 2.7:1. Orbital cysticercosis constituted 67.6% of the cohort. The cyst was located in the lid, conjunctiva, and subconjunctival space in one subject each. The most common symptoms in the orbital subgroup were restriction of extraocular movement (45.8%) and diplopia (27.8%), whereas the most prevalent signs were periocular swelling (13.9%) and proptosis (13.9%). In total, 30 subjects had intraocular cysticercosis – most often in the vitreous (17; 56.7%) or under the retina (11; 36.7%). Two (6.67%) subjects had cysts in the anterior chamber. All patients in the intraocular subgroup presented with decreased vision, and 40.0% had panuveitis. Subjects with orbital/adnexal cysticercosis were treated medically with oral albendazole (7.5 mg/kg twice daily) and oral prednisolone (1 mg/kg/day tapered) for 4–6 weeks. Two subjects were switched to praziquantel, 40 mg/kg. Adjunctive surgical removal of the cyst was performed in six subjects. All subjects with intraocular cysticercosis underwent surgical removal of the cyst. Clinical resolution with improved symptoms and vision was achieved in the vast majority of subjects. Both the authors’s review of the published literature and the most recent comprehensive review of the subject by Wender et al²⁵ support surgical removal as the preferred treatment approach for intraocular cystercercosis.

Dadia et al¹⁵ described a case of Salmonella typhi-associated unilateral endogenous endophthalmitis in a 28-year-old, immunocompetent man from Mumbai, India, who presented with redness and pain associated with a precipitous decline in vision to light perception without projection and an afferent pupillary defect following a history of enteric typhoidal fever requiring hospitalization 2 weeks prior. Findings consistent with the diagnosis of severe endogenous endophthalmitis included lid swelling, chemosis, corneal edema, hypopyon formation, and a dense panuveitis. Blood cultures grew S. typhi sensitive to vancomycin. The patient was treated with intravitreal vancomycin and dexamethasone, followed by intravitreal imipenem and colistin, but without improvement. A diagnostic and therapeutic vitrectomy was then performed and confirmed the presence of S. typhi in the vitreous. Within 2 weeks of presentation, the vision dropped to no light perception and the eye was eviscerated. The authors identified 13 additional, previously reported cases of endogenous Salmonella-associated endogenous endophthalmitis. The most noteworthy feature in this group was the high rate of infancy, with 7 of the 14 total patients less than 1 year of age. Other contributors to relative immunosuppression were also described, including one 55-year-old woman taking combined corticosteroid and non-corticosteroid immunosuppression for underlying rheumatoid arthritis, and a second patient with HIV infection. In total, 12 of the 13 patients for whom final vision was reported either lost or had no light perception vision in the affected eye. The authors emphasized the prevalence of typhoidal salmonellosis, particularly in the developing world, and the need to consider Salmonella-associated endogenous endophthalmitis in all patients who present with severe intraocular inflammation following enteric fever, and in infants with signs of endophthalmitis.

Amphornphruet et al¹⁶ described a 45-year-old woman from Boston, USA, who presented with multiple, elevated chorioretinitis lesion in each eye and with panuveitis with hypopyon formation and acute ocular hypertension in the left eye. Presenting vision was 20/20 in the right eye and 20/200 in the left eye. The patient had undergone craniotomy for resection of a brain tumor and placement of a ventriculoperitoneal shunt 10 years prior, but had no evidence of tumor recurrence or of HIV infection or other identified contributor to relative immunosuppression. She was treated with both intraocular amphotericin B and flucytosine followed by fluconazole with stablization. Vision at last visit was 20/800 on the right and hand motion on the left. The authors identified eight additional cases of endogenous Cryptococcus endophthalmitis in non-HIV infected patients and emphasized the shared presence of one or more chorioretinitis lesions, often large and elevated, and the finding of overlying vitreous inflammation in all but two affected eyes.

Shah et al¹⁷ described two non-immunocompromised adults from London, UK, with unilateral endogenous Meningococcus endophthalmitis in the setting of arthropathy. Neither had evidence of sepsis, meningismus, or fever often seen in patients with meningococcemia. Both presented with light perception vision and panuveitis associated with hypopyon formation, and one had acute ocular hypertension. In each case, erythrocyte sedimentation rates and C-reactive protein levels were markedly elevated, yet blood cultures were negative, and the diagnosis resulted from PCR-based identification of Neisseria meningitides ctrA gene DNA. Despite aggressive treatment with combined intravitreal and systemic antibiotics, vision either failed to improve or progressed to no light perception. The authors cited three additional reports of endogenous Meningococcus endophthalmitis with joint involvement and noted the generally poor outcome in these patients. They also noted the presence of fever, meningitis, a petechial or purpuric skin rash, and other signs of sepsis in a sizable proportion of patients with meningococcemia, but stressed the importance of considering the diagnosis of endogenous Meningococcus endophthalmitis in patients without these findings, but with a recent history of acute arthritis with or without fever – particularly the very young, the elderly, and those with non-age-related causes for immunosuppression. They also commented on the utility of commonly available PCR-based assays applied to both vitreous and joint fluid for diagnosis.

Ranjan et al¹⁸ described a 65-year-old woman from Uttar Pradesh, India, who developed endogenous S. aureus endophthalmitis associated with a pink, or blood-tinged, hypopyon. No systemic source was identified. The patient did well following treatment with combined intraocular and systemic antibiotics. The authors emphasized the importance of identifying contributors to immunosuppression, including extremes of age, cancer, DM, use of corticosteroid or non-corticosteroid immunosuppressive agents, and HIV infection. They also stressed the need to search for symptomatic or asymptomatic urinary tract or hepatobiliary infection. Other cited causes of pink hypopyon included Serratia marcescens and Klebsiella infection, whereas Mycobacterium tuberculosis, Streptococcus bovis and, as noted above, L. monocytogenes, were identified as infections associated with dark or pigmented hypopyon formation.

Together, these studies highlight the common causes and generally poor outcome in eyes with endogenous endophthalmitis. They also emphasize the importance of searching for a systemic source for the infection and, related to this, the frequent need for both intraocular and systemic antibiotics.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Additional information

Funding

Supported in part by The Pacific Vision Foundation (ETC), The San Francisco Retina Foundation (ETC), National Institute of Health (NIH) Center Core Grant P30EY014801 (Bethesda, Maryland) and Research to Prevent Blindness Unrestricted Grant (New York, New York).