Streptococcus pneumoniae endophthalmitis: clinical settings, antibiotic susceptibility, and visual outcomes

Streptococcus pneumoniae endophthalmitis is clinically more severe, more difficult to treat, and carry a higher risk of vision loss, evisceration, or enucleation. This study is to investigate the clinical settings, antibiotic susceptibility, and visual outcomes of S. pneumoniae endophthalmitis at a tertiary referral center in Taiwan. S. pneumoniae endophthalmitis was diagnosed in 38 eyes of 38 patients. The main clinical features were postcataract endophthalmitis (n = 13, 34%) and endophthalmitis associated with corneal ulcer (n = 12, 32%), trauma (n = 6, 16%), endogenous etiology (n = 4, 11%), trabeculectomy (n = 2, 5%), and pterygium excision-related scleral ulcer (n = 1, 3%). Presenting visual acuity ranged from counting fingers to no light perception. Pars plana vitrectomy with intravitreal antibiotics was performed in 17 eyes (39%) in primary or secondary treatments. S. pneumoniae isolates were susceptible to vancomycin (38/38, 100%), penicillin (37/38, 97%), ceftriaxone (37/38, 97%), cefuroxime (12/15, 80%), levofloxacin (13/15 ,87%), and moxifloxacin (15/17, 88%). Final visual acuity was better than 20/400 in 3 of 38 eyes (8%), 5/200 to hand motions in 3 eyes (8%), and light perception to no light perception in 32 eyes (84%). Ten eyes (26%) underwent evisceration or enucleation. Although S. pneumoniae isolates were susceptible to vancomycin, S. pneumoniae endophthalmitis had a very poor visual prognosis.


Results
Thirty-eight eyes of 38 patients with culture-proven endophthalmitis caused by S. pneumoniae infection were identified. Tables 1 and 2 illustrate the demographics, systemic illnesses, clinical features, management details, and visual outcomes of patients with S. pneumoniae endophthalmitis. In terms of sex, 25 of the patients (66%) were male and 13 patients (34%) were female. The median age was 61.3 ± 26.7 years (range, 2 to 92 years). The systemic illnesses of interest included diabetes mellitus (in nine patients) and primary hypertension (in nine patients). The median follow-up interval was 21 months (range, 1 to 85 months).
In the endogenous etiology group, endogenous causes included infective endocarditis and pneumonia, and two patients had unidentified primary sources of infection. One patient with scleral ulcer-related endophthalmitis underwent a complicated pterygium excision with subsequent scleral grafts. All patients with postcataract endophthalmitis developed acute-onset endophthalmitis (median, 3 days; range, 1 to 8 days). The two patients with trabeculectomy-associated etiologies developed endophthalmitis 2 years and 4 years posttrabeculectomy, respectively. The mean interval between trauma and diagnosis of endophthalmitis was 2 days (range, 1 to 4 days).
Severe blurring of vision and pain were symptoms observed for all 38 eyes (100%) and 35 of 38 eyes (92%), with mean durations of 2.2 and 2.0 days, respectively. Visual acuity (VA) test results were as follows: hand motions (HM; n = 14/38, 37%), light perception (LP; n = 12/38, 32%), no light perception (NLP; n = 7/38, 18%), counting fingers (4/38, 11%), and undetermined VA in one 2-year-old patient. Hypopyon was documented in 34 of 38 eyes (89%). Visualization in the posterior segment view was poor in all eyes due to severe anterior  Tables 3 and 4. One patient with traumatic endophthalmitis due to IOFB underwent removal of an anterior chamber IOFB and anterior chamber irrigation, and they were given antibiotics intravitreally. This patient was treated with a second course of anterior chamber irrigation with intravitreal antibiotics and eventually achieved 20/20 vision. Primary evisceration was performed on six eyes, and secondary evisceration or enucleation was performed on four eyes. In addition, intravitreal dexamethasone was administered to eight eyes (21%) during primary treatment. Statistical analysis. No significant difference in visual outcomes was observed between PPV and TAP groups (P = 0.99, Fischer's exact test). Because of the relatively small number of eyes and the nonrandomized nature of this retrospective study, statistical conclusions concerning the differences between the clinical settings encountered and the treatment administered in this series cannot be drawn.

Discussion
A comparison of previous and current studies is presented in Table 5 and includes clinical settings, the vancomycin sensitivity pattern, and final visual outcomes. The most common etiology in the current series was postoperative endophthalmitis; this finding is similar to those in studies conducted in the United States, Europe, Australia, and Thailand 5-10 . However, the most common cause in an Indian study was keratitis, followed by trauma and surgery 13 . In postoperative settings, the most common causes were cataract, trabeculectomy, and PK-related corneal ulcer. Because of the increasing trend toward the intravitreal administration of antivascular endothelial growth factor agents, streptococcus species such as viridans streptococci have become more common than S. pneumoniae as the causative organisms of intravitreal injection-related endophthalmitis 5 . S. pneumoniae is generally sensitive to vancomycin, penicillin, ceftriaxone, and fluoroquinolone. In Western countries, most related case series reported that it had susceptibility to vancomycin (100%) 4,5,7,9,10 . However, in India, some S. pneumoniae isolates showed resistance to vancomycin [12][13][14] . In this study and studies from Australia and Thailand, all S. pneumoniae isolates were sensitive to vancomycin (100%) 6,8 . Although fluoroquinolones, such as levofloxacin and moxifloxacin, are widely used in the treatment or prophylaxis of endophthalmitis, some S. pneumoniae isolates have been reported to still be resistant to these treatments 9,12 . In our study, some S. pneumoniae isolates were partially resistant to moxifloxacin or levofloxacin. Our study demonstrated that the MIC 90 of vancomycin for S. pneumoniae was 0.5 μg/mL. We propose that vancomycin be continued as the firstline intravitreal antibiotic in the treatment of S. pneumoniae endophthalmitis.
Studies on endophthalmitis caused by S. pneumoniae infection showed a higher rate of poor visual and anatomical outcomes in those with the disease. US case series 3,5,7,9 have found better visual outcomes than in Asian, European, and Austrian studies 6,8,10,12-14 . In the present study, only three eyes (3 out of 38, 8%) achieved vision of 20/400 or better. Patients with endophthalmitis caused by S. pneumoniae or P. aeruginosa infection have a   9 . In Spanish and Indian studies, higher rates of evisceration of eyes affected by S. pneumoniae endophthalmitis were identified 10,13 . Soriano et al. 10 reported an evisceration rate of 47% in 36 cases of S. pneumoniae endophthalmitis. However, our study demonstrated an evisceration or enucleation rate of 26% in 38 cases of S. pneumoniae endophthalmitis. Early detection and treatment are the best approaches to improving final visual outcomes in patients with endophthalmitis. Although some case series have shown no statistical difference between vitrectomy and TAP, early PPV still offers a higher likelihood of achieving a more favorable visual prognosis 6,9,10 . The need for a higher number of intravitreal injections was significantly associated with a poor final VA 7 . The rate of evisceration or enucleation in patients with endophthalmitis caused by S. pneumoniae did not decrease in recent studies 5,6,8,13,14 .  16 demonstrated that prevention of evisceration or enucleation in patients with endogenous bacterial panophthalmitis, no light perception, and scleral abscess can be achieved through multiple intravitreal and periocular injections of antibiotics and dexamethasone. No eye required evisceration or enucleation. Multiple intravitreal and periocular injections of antibiotics and dexamethasone could be an alternative to evisceration or enucleation in such cases. This study has several limitations. First, the study was retrospective. Second, a few subgroups, such as those with traumatic, trabeculectomy-related, and endogenous S. pneumoniae endophthalmitis, had a relatively small number of patients. This made statistical analysis difficult to perform. Third, only a few antibiotics and samples were subjected to antibiotic susceptibility testing. However, vancomycin is regularly tested by the laboratory department of our hospital. Because vancomycin is a routine antimicrobial agent used in treating Gram-positive bacterial endophthalmitis, our results still offer useful information. Despite these limitations, this study provided data on antibiotic susceptibility patterns and visual outcomes of endophthalmitis caused by S. pneumoniae.
In conclusion, the most common cause in S. pneumoniae endophthalmitis infection was postcataract status. Although all S. pneumoniae isolates were susceptible to vancomycin, patients with S. pneumoniae endophthalmitis typically have a very poor visual prognosis. Further studies are necessary to ensure better prevention and management of endophthalmitis caused by S. pneumoniae; this could include preventing he emergence of endophthalmitis and examining the effectiveness of prompt, aggressive medical and surgical interventions for the management of endophthalmitis.

Methods
The Institutional Review Board of Chang Gung Memorial Hospital in Taoyuan, Taiwan approved this retrospective study protocol and waived the need for written informed consent from the study patients. All clinical procedures were conducted according to the principles of the Declaration of Helsinki. The medical records of all patients treated for endophthalmitis caused by S. pneumoniae infection were reviewed at a tertiary referral center between April 2019 and January 2000. The inclusion criteria were as follows: (1) Positive S. pneumoniae cultures from intraocular fluid (aqueous humor and vitreous) were identified in patients with exogenous endophthalmitis, including after surgery and trauma; (2) For patients with endogenous endophthalmitis, their bodily fluids were cultured with growths of S. pneumoniae. (3) The patients received follow-ups for at least 3 months unless, based on their final best-corrected VA, they had NLP. Data collected and reviewed included demographic information, medical history, presenting signs and symptoms, duration of symptoms before diagnosis of endophthalmitis, intervals between event and diagnosis of endophthalmitis, culture sites, antibiotic sensitivities and resistance patterns, management administered, and final VA.
Intraocular fluids were collected and plated on blood agar (5% sheep), chocolate agar, fungus culture media, and thioglycolate broth. All microbiologic testing was performed at the Microbiology Department of Chang Gung Memorial Hospital, Taoyuan, Taiwan. Bacterial culture isolates were identified through conventional microbiological methods (2000-2013) and matrix-assisted laser desorption ionization time of flight mass spectrometry (MALDI-TOF-MS; 2014-2019). Conventional microbiological methods included Gram staining and biochemical tests. Gram staining revealed Gram-positive, lancet-shaped cocci. Positive reactions included optochin, bile solubility, inulin and alfa hemolysis, and negative reactions included catalase and oxidase. When the isolates were identified through MALDI-TOF MS (Bruker Daltonics, Germany), the microbial film was overlaid with 1 μL of 70% formic acid. After the sample had dried, the film was overlaid with 1 μL of the matrix solution (50% acetonitrile containing 1% a-cyano-4-hydroxycinnamic acid and 2.5% trifluoroacetic acid). After the sample had dried again, microorganism identification and data analysis were performed, for which the Bruker LT Microflex MALDI-TOF MS and Bruker Biotyper 3.0 system software programs were used. All analyses were performed using standard methods. An identification was considered successful at the species level when the score exceeded 2.0. The isolates were tested for susceptibility to various antibiotics using the Kirby-Bauer Disc diffusion method on Mueller Hinton blood agar. The MIC breakpoints were determined based on the 0.5 McFarland standards by using a microdilution broth and the E-test method. The Clinical and Laboratory Standards Institute (Wayne, PA) standards were used for interpretation and quality control for each corresponding year 17 . Because of the limitations inherent in a retrospective study, the antibiotic susceptibility testing of S. pneumoniae isolates was performed in some selected antibiotics, including vancomycin, penicillin, ceftriaxone, levofloxacin, and moxifloxacin. Some S. pneumoniae isolates were tested to determine the relevant MIC of antibiotics such as penicillin, ceftriaxone, cefuroxime, and vancomycin.
The treatment strategies were determined by the individual treating physicians and did not follow a standardized protocol. Before S. pneumoniae was cultured, intravitreal antibiotics included vancomycin with either ceftazidime or amikacin. After positive cultures of S. pneumoniae isolates and antibiotic susceptibility testing results were obtained, intravitreal vancomycin was administered. The doses of intravitreal agents were as follows: vancomycin, 1 mg/0.1 mL; amikacin, 0.25-0.4 mg/0.1 mL; ceftazidime, 2.25 mg/0.1 mL; and dexamethasone, 0.4 mg/0.1 mL. Tap with intravitreal antibiotics was treated as follows: (1) Corneal clarity was not suitable for PPV. (2) Patients with poor systemic condition were not suitable for PPV. (3) Patients refused PPV because of poor visual prognosis . Pars plana vitrectomy was performed as follows: (1) corneal clarity possible for PPV, and (2) poor response to initial tap treatment. Enucleation or evisceration was performed when the globe was not salvageable due to corneal or scleral necrotic laceration. Poor visual outcomes were defined by VA worse than 20/400, whereas favorable prognoses were defined by VA of 20/400 or better. Statistical analysis was performed using SPSS for Windows version 23 (SPSS Science, Chicago, IL, USA).