In this study, we analyzed the incidence and risk factors of acute endophthalmitis after intravitreal anti-VEGF injections in patients with neovascular AMD by using the KNHIS health claims data and neovascular AMD registration codes from 2010 to 2019. Acute endophthalmitis after intravitreal anti-VEGF injection occurred very rarely (0.019%), and the risk of developing endophthalmitis was higher with aflibercept injections than with ranibizumab injections (IRR, 1.80; 95% CI, 1.16-2.79; p = 0.01).

Few studies have analyzed the rate of endophthalmitis after intravitreal anti-VEGF injections in patients with wet AMD. The incidence of endophthalmitis in our study differed in comparison with the findings reported in previous studies, which seems to be mainly related to differences in study design, patient characteristics, and the definition of endophthalmitis. In a 2-year single-center retrospective study, the incidence of clinically diagnosed endophthalmitis was reported to be 0.014%, which is consistent with our study [12]. However, another 4-year multicenter retrospective study found that the incidence of presumed infectious endophthalmitis was much higher (0.04%) [7]. Also, a single-center prospective study reported an incidence as high as 0.083% [6]. Similar to our study, these studies targeted patients who were clinically suspected of having infectious endophthalmitis after intravitreal anti-VEGF injection, received vitreous tap, and underwent intravitreal antibiotic injection or vitrectomy. However, unlike our study, these studies included patients with other retinal diseases, such as diabetic macular oedema, proliferative diabetic retinopathy, and retinal vein occlusion other than wet AMD, which may have affected the overall incidence of endophthalmitis. In particular, diabetic patients are vulnerable to infection due to innate immune suppression and blood circulatory dysfunction, which may affect the incidence of endophthalmitis after intravitreal injection [7,13].

The incidence of endophthalmitis according to the medications used also differed between studies. In this study, the incidence was slightly higher with intraocular injections of aflibercept (0.025% vs. 0.014%). In a study using US claims data, the incidence of antibiotic-associated endophthalmitis was higher with aflibercept than with ranibizumab (0.44% vs. 0.22%), and the risk of developing endophthalmitis was also higher with aflibercept (adjusted relative ratio, 2.04; 95% CI, 1.42-2.92; p < 0.05) [14]. However, some studies showed no statistically significant differences in the incidence of endophthalmitis by medication type [12,15,16].

The higher incidence of endophthalmitis after intravitreal aflibercept injection is difficult to explain. One possible explanation is that patients with severe sterile intraocular inflammation were also included in our analysis, which may have increased the incidence of acute endophthalmitis requiring invasive treatment. Aflibercept contains an Fc domain that can act as a proinflammatory factor and can cause sterile intraocular inflammation [17]. In patients with wet AMD, the Fc-gamma receptor IIa (FcγRIIa) is widely expressed throughout the retina [18]. Unlike ranibizumab, aflibercept, which has a higher molecular weight, does not permeate the retina through the intercellular space and is absorbed by neuronal cells and the retinal pigment epithelium [19]. Therefore, the Fc domain of aflibercept can bind to FcγRIIa, which is upregulated in wet AMD patients and ultimately induces inflammation [14,18].

Several potential factors are associated with endophthalmitis following intravitreal anti-VEGF injections. The most common cause may be bacterial or fungal contamination during the injection procedure. In a survey by Green-Simms et al. [20], 88% of US retinal specialists did not use a sterile drape, and only 33% wore sterile gloves. These practices may have increased the risk of endophthalmitis. However, antibiotic prophylaxis does not affect the incidence of endophthalmitis [21,22]. Environmental effects, such as septic injection techniques or contaminated aerosolized droplets, are another potential source of infection. An extremely low incidence of endophthalmitis (0.000%-0.008%) was reported when the procedure was performed in an operating room setting [23-26]. No difference was observed in the incidence rate of endophthalmitis following injections in the operating room and office (odds ratio, 3.06; 95% CI, 0.07-139.75), but the rate of development of culture-positive endophthalmitis was much higher following injection in the office (odds ratio, 21.52; 95% CI, 2.39-193.55) [27]. These results show that a sterile environment can contribute to reducing endophthalmitis after injection.

The incidence of endophthalmitis may also increase due to mishandling of the medication, i.e., problems with storage or preparation. A nationwide cohort study performed in France showed a higher risk of endophthalmitis with aflibercept than with prefilled ranibizumab (IRR, 1.80; 95% CI, 1.24-2.61). Antigenic particles derived from protein degradation or silicone oil droplets may form during rapid freezing, thawing, or storage outside the recommended temperature range of 2°C to 8°C [28]. In addition, during the process of transfer from a vial to a syringe, the medication may be exposed to the infection source. For patients treated with ranibizumab, prefilled syringes also showed a lower risk of endophthalmitis than nonprefilled syringes (IRR, 1.60; 95% CI, 1.14-2.25), which means that fewer manipulations and preparation under sterile environments are important to reduce postoperative endophthalmitis [4].

Our study had several limitations. First, the identification of cases with intravitreal anti-VEGF injections, endophthalmitis, and other comorbidities by using healthcare claims and KCD and KEDI codes might be inaccurate when compared with information obtained from medical charts. In particular, this method shows a greater potential for misclassification or underdetection-related bias. Second, the diagnosis of acute endophthalmitis was based on clinical suspicion rather than a strict bacteriological definition, and both culture-positive and culture-negative cases were included in the study. As a result, patients with severe sterile inflammation were also included in our analysis. Third, the lack of clinical information regarding factors such as medical history, visual acuity, and subtypes of neovascular AMD is another inherent limitation. Despite its low risk, endophthalmitis is a vision-threatening disease; therefore, an understanding of the visual prognosis of endophthalmitis after injection is important. Fourth, bevacizumab, which is less expensive than aflibercept or ranibizumab, is often intravitreally injected when the eligibility criteria for neovascular AMD are not met. However, since bevacizumab is an off-label drug, it is not covered by insurance and is excluded from the KNHIS database. Therefore, the incidence of acute endophthalmitis during intravitreal bevacizumab injections was not analyzed. Fifth, although the use of prefilled syringes may affect the incidence of endophthalmitis, it was not analyzed as a risk factor in this study. This is because the domestic approval date of the prefilled ranibizumab was June 2019, and prefilled aflibercept has not been marketed in Korea. In other words, prefilled syringe was used very limitedly within the study period, so it was not possible to compare the incidence of endophthalmitis between prefilled syringe and vials. In future research, analysis of syringe type should be considered as one of the risk factors for endophthalmitis.

Nevertheless, our study also had the following strengths. This is the largest retrospective claims data-based study of acute endophthalmitis after intravitreal anti-VEGF injection in patients with neovascular AMD in South Korea over a 10-year study period. Using this population-based database, we provided a reliable assessment of the incidence of acute endophthalmitis after intravitreal injection— an extremely rare complication—that was consistent with previous studies. We also demonstrated the safety of frequently performed procedures in neovascular AMD.

In this study, we demonstrated a generally low incidence of acute endophthalmitis after intravitreal anti-VEGF injections, consistent with previous studies. However, because neovascular AMD requires repeated intravitreal anti-VEGF injections, attention should be paid to the risk of endophthalmitis.