Fulminant Type
Onset: Within 1–3 days after injection
Causative organisms: Highly virulent bacteria such as Streptococcus species.
Features: Rapid intraocular inflammation. Severe hypopyon and vitreous opacification, with the worst visual prognosis.
Post-injection endophthalmitis (PIE) is an intraocular infection that occurs as a complication of intravitreal injection (IVI). After intravitreal administration of anti-VEGF agents, steroids, or other drugs, pathogenic bacteria invade the eye, causing acute endophthalmitis.
Intravitreal injections are widely performed for the management of neovascular age-related macular degeneration (nAMD), diabetic retinopathy, retinal vein occlusion, and other conditions, and the number of procedures is increasing worldwide. The incidence is reported to be 0.028–0.056% (0.28–0.56 per 1000 injections)2), and PIE accounts for up to 8.5% of all endophthalmitis cases1).
Endophthalmitis is broadly classified into three types based on the mode of onset and the virulence of the causative organism.
Fulminant Type
Onset: Within 1–3 days after injection
Causative organisms: Highly virulent bacteria such as Streptococcus species.
Features: Rapid intraocular inflammation. Severe hypopyon and vitreous opacification, with the worst visual prognosis.
Acute type
Onset: Within 2–7 days after injection
Causative organisms: Mostly coagulase-negative staphylococci (e.g., S. epidermidis).
Features: The most common form. Early intervention improves visual prognosis.
Pseudo-endophthalmitis
Onset: Within 1 day to a few days after injection
Causative organisms: None (non-infectious)
Features: Crystals (e.g., triamcinolone) migrate into the anterior chamber, forming a white opacity. The absence of eye pain is a key distinguishing feature.
The incidence per injection is constant, and the cumulative risk increases proportionally with the number of injections. In the MARINA trial, the risk per injection was 0.05%, while in the VIEW trial, the cumulative 1-year risk was less than 1.0%6).
PIE often develops rapidly within a few days after injection, with the following characteristic symptoms.
Evaluation combines slit-lamp microscopy and fundus findings.
The culture positivity rate is high at approximately 94% 2), and identification of the causative organism contributes to treatment decisions.
Pseudo-endophthalmitis is a condition where crystals such as triamcinolone migrate into the anterior chamber, often without eye pain and with minimal visual loss 3). Infectious endophthalmitis is accompanied by eye pain, rapid visual loss, and vitreous opacity. When differentiation is difficult, it is safe to manage as infectious.
The distribution of causative organisms in post-intravitreal injection endophthalmitis is shown below.
| Causative bacteria | Approximate frequency |
|---|---|
| S. epidermidis (CoNS) | Approximately 59% |
| Other gram-positive bacteria | Around 30% |
| Gram-negative bacteria (e.g., M. morganii) | Rare |
Staphylococcus epidermidis (coagulase-negative staphylococcus) is the most common, accounting for approximately 59%2). Co-infection with Enterococcus faecalis and Morganella morganii has been reported as the first case worldwide, suggesting the involvement of multidrug-resistant bacteria1).
The main sources of PIE infection are considered to be the patient’s own conjunctival resident bacteria and oral droplets.
PIE should be diagnosed promptly based on clinical findings, and treatment should be started immediately. Do not delay treatment while waiting for test results.
The main examination methods used for diagnosis are listed below.
| Examination | Purpose | Notes |
|---|---|---|
| Slit-lamp microscopy | Check for hypopyon and flare | Essential, first choice |
| B-scan ultrasonography | Confirms vitreous opacity and retinal detachment | Essential when fundus is not visible |
| Anterior chamber and vitreous tap | Identifies causative organism | Collect samples before treatment |
Collect the following samples immediately before intravitreal antibiotic injection:
The collected samples are used for culture, Gram staining, and drug susceptibility testing.
The first-line treatment for PIE is intravitreal injection of two broad-spectrum antibiotics.
Standard doses and administration of antibiotics are shown below.
| Drug | Dose | Target bacteria |
|---|---|---|
| Vancomycin | 1 mg/0.1 mL | Gram-positive bacteria |
| Ceftazidime | 2–2.25 mg/0.1 mL | Gram-negative bacteria |
Intravitreal administration of vancomycin 1 mg plus ceftazidime 2–2.25 mg is considered standard treatment 1). Administer promptly after specimen collection. After culture results are available, switch to antibiotics targeting the causative organism.
If clinical improvement is not achieved within 48–72 hours after the initial injection, consider pars plana vitrectomy (PPV).
Kvopka et al. (2023) reported a case in which vitrectomy was performed after two intravitreal injections of vancomycin 1 mg + ceftazidime 2.25 mg failed to resolve hypopyon, and the BCVA at 12 weeks was 6/90 (0.12)1). The causative organisms were a co-infection of M. morganii and E. faecalis.
A large study using the IRIS Registry (1044 cases) showed no significant difference in final visual acuity between early vitrectomy and injection-only treatment5). This result supports the validity of a stepwise approach (first intravitreal injection, then PPV if no improvement).
It has been reported that the frequency of anti-VEGF injections decreases and the injection interval significantly lengthens after the onset of PIE (1.09 times/month before onset → 0.52 times/month after onset, p=.001)2). In 12% of cases, injections were not resumed after PIE2). The average time to resumption was 44±30 days2).
The incidence of endophthalmitis after vitrectomy is low, at less than 0.05%7).
In many cases, injections are resumed after PIE. One report found that the average time to resumption was 44±30 days2), and complete discontinuation occurred in 12% of cases. Consider resuming at an appropriate time while assessing the activity of the underlying disease.
PIE is an exogenous endophthalmitis caused by the entry of external microorganisms into the eye during intravitreal injection. The two most important routes of infection are as follows:
Morganella morganii is a multidrug-resistant Enterobacteriaceae Gram-negative rod, and its involvement in intraocular infection has been reported as a rare case worldwide1).
Pseudoendophthalmitis is not infectious; it occurs when crystalline particles of the injected drug (mainly triamcinolone acetonide) migrate into the anterior chamber3). If there is pre-existing disruption of the blood-retinal barrier (BRB) (e.g., increased vascular permeability associated with neovascularization), drug particles are thought to move more easily from the posterior chamber to the anterior chamber3). The reported rate of endophthalmitis after bevacizumab injection is approximately 0.066%, and after triamcinolone, 0.10–0.87%3).
A phenomenon in which exudative changes in nAMD temporarily stabilize or decrease after PIE has been observed in some cases2). It is speculated that post-inflammatory fibrosis and proliferation suppression may affect neovascular activity in nAMD, but the mechanism remains unclear2).
A retrospective study (2025) of 1044 cases using the US IRIS Registry (Intelligent Research in Sight) showed no significant difference in final visual acuity between the early vitrectomy group and the initial injection-only group5). This result supports the current stepwise approach of not performing vitrectomy uniformly for all PIE cases but rather responding to intravitreal injection response.
Binczyk et al. (2023) reported that anti-VEGF injection frequency significantly decreased after PIE in 17 eyes (1.09→0.52 IVI/month, p=.001), and nAMD activity stabilized in some cases2). Further prospective studies are needed on the impact of local inflammation after PIE on the pathology of nAMD.
Kvopka et al. (2023) reported the world’s first case of PIE caused by co-infection with Morganella morganii and Enterococcus faecalis1). This case demonstrates the difficulty of treating intraocular infections caused by multidrug-resistant gram-negative bacteria and highlights the importance of culture and antibiotic susceptibility testing.
The “tunneled technique” has been proposed to reduce the risk of reflux and infection by changing the injection needle insertion angle to create a non-linear needle track4). Comparative studies with standard techniques are still limited, and further evidence is needed for widespread clinical adoption.
Current evidence is insufficient to support that prophylactic topical antibiotics reduce PIE4). The mainstay of prevention is conjunctival sac disinfection with povidone-iodine, and routine use of antibiotic eye drops may carry the risk of promoting antibiotic resistance.