Management Patterns and Outcomes for Intravitreal Injection–Related Endophthalmitis

Abstract

Purpose:

This work describes the clinical management and outcomes in cases of presumed infectious endophthalmitis following intravitreal injection at a tertiary academic medical center.

Methods:

A retrospective review took place of eyes that presented to the Duke Eye Center over a 9-year period and were diagnosed with intravitreal injection–related endophthalmitis. Clinical presentation, management, microbiologic yield, visual outcomes, and complications were abstracted from medical records.

Results:

Of 23 eyes diagnosed with postinjection endophthalmitis, 52.2% underwent anterior chamber tap (33.3% of which first underwent dry needle vitreous tap), 47.8% underwent needle vitreous tap, 17.4% underwent neither, and none underwent pars plana vitrectomy (PPV) for initial management. Subsequent PPV was performed in 6 eyes (26.1%). Mean visual acuity (VA) improved by 50 Early Treatment Diabetic Retinopathy Study letters at 6 months. Eyes that underwent initial anterior chamber tap had worse presenting VA than those that did not (P = .01). Eyes undergoing subsequent PPV had worse VA at presentation (P = .02) and at 6 months (P < .001). Eyes presenting with VA of hand motion (20/8000) or worse were more likely to undergo subsequent PPV (P = .02).

Conclusions:

Eyes with intravitreal injection–related endophthalmitis presenting with VA of hand motion or worse were more likely to undergo subsequent PPV. Future studies with larger cohorts may reveal whether earlier vitrectomy should be considered in these patients.

Introduction

Intravitreal injections allow for targeted drug delivery with reduced systemic toxicity in the management of a number of retinal conditions. ^1,2 The use of intravitreal antivascular endothelial growth factor (anti-VEGF) agents to manage common conditions such as neovascular age-related macular degeneration, diabetic macular edema, and retinal vein occlusion continues to grow. ^{3 –5} With the increasing number of injections performed, it is important to be cognizant of the associated risks.

Endophthalmitis is a serious vision-threatening complication that can occur following an intravitreal injection. ¹ A comprehensive review examining nearly 15 000 intravitreal injections reported an incidence of 0.2% per injection, ⁶ while larger, more recent meta-analyses have reported rates ranging from 0.038% to 0.065% per injection. ^3,7 Multiple injections over a prolonged period of time, however, can result in a cumulative rate of infection of up to 1%. ⁴ Endophthalmitis risk has been reported to differ among medication classes; antivirals (ganciclovir, cidofovir, and fomivirsen) carry an infection rate of 0.1% per injection, while corticosteroids such as triamcinolone carry an infection rate of 0.6%. ⁶ Studies have not shown a significant difference in infectious endophthalmitis rates among anti-VEGF agents (pegaptanib, bevacizumab, ranibizumab, aflibercept, and brolucizumab). ^{8 -13} Intravitreal injection of aflibercept has been linked to sterile intraocular inflammation in rare cases, at a rate of approximately 0.05%. ^14,15 Brolucizumab has also been linked to sterile intraocular inflammation. ¹⁶

Although infectious endophthalmitis and sterile intraocular inflammation may present with similar symptoms, cases of infectious endophthalmitis often present within 24 to 48 hours, exhibit a pronounced decline in visual acuity (VA), and display cell, flare, fibrin, and hypopyon during clinical examination. ¹⁷ Postinjection infectious endophthalmitis is relatively rare in clinical practice; however, prompt diagnosis and treatment may prevent severe vision loss in suspected cases.

In culture-proven cases of endophthalmitis following intravitreal injection, coagulase-negative Staphylococcus species are most commonly isolated, followed by Streptococcus species. ^1,3 Compared with endophthalmitis after cataract surgery or vitrectomy, cases related to intravitreal injection were more often associated with oral and respiratory flora. ⁷ Among these, endophthalmitis caused by the Streptococcus viridans species may progress more rapidly and result in worse visual outcomes. ^3,18,19

While nearly all cases of presumed infectious endophthalmitis following intravitreal injection are managed with broad-spectrum intravitreal antimicrobials, such as vancomycin and ceftazidime, precise treatment guidelines have not been well established. ^2,19 As such, diagnostic and management decisions are often extrapolated from results of the Endophthalmitis Vitrectomy Study (EVS), which examined only cases of endophthalmitis following cataract surgery or secondary intraocular lens placement.

Overall, the EVS found no difference in visual outcomes following treatment with systemic antibiotics, and that eyes with presenting VA of light perception or worse attained better visual outcomes if managed with initial pars plana vitrectomy (PPV) compared with those that solely underwent needle vitreous tap with intravitreal injection of antimicrobials. ²⁰ However, assessment of these interventions for the management of postinjection endophthalmitis may be warranted as differences in patient characteristics and microbiologic causes may lead to different treatment responses. Some studies have suggested that needle vitreous tap and injection of antimicrobials alone may successfully treat most cases of postinjection endophthalmitis, thus reserving PPV for only those with persistent infection or progressive clinical decline. ^18,21

The present study examined cases of intravitreal injection–related endophthalmitis evaluated at a tertiary academic medical center over a 9-year period to assess diagnostic and management patterns, complications, and visual outcomes.

Methods

Patient Identification and Data Collection

A retrospective medical-record review of consecutive eyes diagnosed with intravitreal injection–related endophthalmitis at the Duke Eye Center between January 1, 2009, and January 1, 2018, was conducted using the Duke Enterprise Data Unified Content Explorer. Patients who had the diagnosis of endophthalmitis made by an experienced retina specialist at the time of presentation and had at least 6 months of documented follow-up were included. Medical records were abstracted, and data related to ocular history, clinical presentation, management, microbiologic yield, visual outcomes, and complications were collected.

Statistical Analysis

Statistical analyses were performed using XLSTAT (Addinsoft). Means of continuous variables were compared using 2-tailed t tests; categorical variables were compared using χ² tests. Binary logistic regression analysis was used to characterize associations between binary dependent variables and continuous independent variables. In all analyses, an α of .05 was used to define statistical significance.

Results

Twenty-three eyes were diagnosed with presumed infectious endophthalmitis after intravitreal injection and met the inclusion criteria. Of these 23 eyes, 43.5% were men, and mean age was 74 years (range, 53-89 years). Administered intravitreal agents included aflibercept (60.9%) (Eylea, Regeneron), ranibizumab (17.4%) (Lucentis, Genentech), bevacizumab (13.0%) (Avastin, Genentech), and preservative-free triamcinolone acetate (8.7%) (Triesence, Alcon). The ocular conditions treated with these intravitreal injections included neovascular age-related macular degeneration (65.2%), diabetic macular edema (17.4%), central retinal vein occlusion (8.7%), branch retinal vein occlusion (4.3%), and cystoid macular edema due to uveitis or other causes (4.3%). Additional baseline and presenting clinical characteristics are detailed in Table 1.

Table 1.

Baseline Characteristics of Eyes Diagnosed With Presumed Infectious Endophthalmitis Following Intravitreal Injection (n = 23).

Characteristic	Value
Right eye, %	52.2
Male sex, %	43.5
Average age (range), y	74 (53-89)
Presenting VA	20/1145
Presenting characteristics, n (%)
Subjective blurred vision	22 (95.7)
Conjunctival hyperemia	10 (43.5)
Ocular pain	17 (73.9)
IOP >21 mm Hg	3 (13)
IOP <9 mm Hg	1 (4.3)
Hypopyon	15 (65.2)
Anterior chamber fibrin	11 (47.8)
Vitreous haze/no view	14 (60.9)

Abbreviations: IOP, intraocular pressure; VA, visual acuity.

Mean duration from intravitreal injection to presentation with presumed infectious endophthalmitis was 4.6 days (range, 1-19 days). The mean time from presentation to initial treatment was 0.26 days (range, 0-2 days). In the 2 cases that began treatment after 2 days, the initial presentation was not indicative of infectious endophthalmitis; however, as symptoms worsened, intravitreal antibiotics were administered. For initial management, 12 eyes (52.2%) underwent anterior chamber tap (of which 4 eyes had first undergone a dry needle vitreous tap), 11 eyes (47.8%) underwent needle vitreous tap, and 4 eyes (17.4%) underwent neither diagnostic procedure (3 of these eyes received intravitreal antimicrobials without diagnostic sampling). All needle vitreous taps were performed using a 25-gauge needle. No eyes underwent initial PPV.

Subsequent PPV after initial intervention was performed in 6 of 23 eyes (26.1%) after a median follow-up of 55.5 days (mean, 64.5 days). Of the 6 eyes, 2 (33.3%) had undergone a dry needle vitreous tap followed by anterior chamber tap, 3 (50.0%) had undergone anterior chamber tap only, and 1 (16.7%) had undergone needle vitreous tap only. Indications for subsequent PPV included recurrent inflammation in 3 of 6 cases (50.0%), retinal detachment (RD) in 2 cases (33.3%), and epiretinal membrane in 1 case (16.7%). None of these eyes underwent additional intravitreal anti-VEGF therapy between presentation and PPV. Anti-VEGF therapy was resumed in 10 of 23 eyes (43.5%) after an average of 98.4 days from presentation.

One of 13 Gram stains that was performed showed rare gram-positive cocci; the remainder did not reveal any organisms. Overall, positive culture results were found in 7 of 23 cases (30%). Of the 2 aqueous samples that yielded positive culture findings, 1 was positive for coagulase-negative Staphylococcus, and the other was positive for Stenotrophomonas maltophilia. Three of the 4 culture-positive needle vitreous samples yielded coagulase-negative Staphylococcus, and the fourth yielded Propionibacterium acnes. Mechanical vitreous biopsies were obtained in 2 of the 6 eyes that underwent subsequent PPV after initial treatment; one of these yielded no growth, whereas the other was positive for Streptococcus viridans. The eye that was culture positive for S viridans underwent PPV 10 days after the initial anterior chamber tap; the patient’s VA worsened from 20/8000 at presentation to 20/16 000 at 6 months.

Twenty-two of 23 eyes (95.7%) received intravitreal antimicrobials. Of these 22 eyes, all 22 (100%) were treated with intravitreal vancomycin, 21 (95.5%) with ceftazidime, 2 (9.1%) with amikacin, 1 (4.5%) with voriconazole, and 1 (4.5%) with amphotericin B. The 1 of 23 eyes (4.3%) that did not receive intravitreal antimicrobials on presentation was initially treated with topical steroids only; this eye had previously undergone intravitreal anti-VEGF therapy with ranibizumab.

Mean presenting corrected VA was 20/1145 (logMAR = 1.758), and mean VA at 6 months was 20/113 (logMAR = 0.754), with an average improvement of 50 Early Treatment Diabetic Retinopathy Study letters (change in logMAR = –1.004). For culture-positive cases, mean VA was 20/1761 (logMAR = 1.945) at presentation and 20/142 (logMAR = 0.851) at 6 months. For culture-negative cases, mean VA was 20/949 (logMAR = 1.676) at presentation and 20/100 (logMAR = 0.701) at 6 months; there was no significant difference in mean VA at presentation (P = .57) or at 6 months (P = .69) between culture-positive and culture-negative cases. Of the 7 culture-positive cases, 5 (71.4%) had undergone an intravitreal injection with aflibercept prior to endophthalmitis. Of the 16 culture-negative cases, 9 (56.3%) had undergone an intravitreal injection with aflibercept prior to endophthalmitis.

Eyes that underwent anterior chamber tap on presentation had significantly worse presenting VA than those that did not undergo anterior chamber tap (20/3508 vs 20/338, P = .01). Eyes that underwent subsequent PPV also had significantly worse presenting VA (20/7129 vs 20/601, P = .02) as well as worse VA at 6 months (20/1002 vs 20/55, P < .001) compared with eyes that did not undergo subsequent PPV. Presenting VA of hand motion at 1 foot (30 cm) (20/8000) or worse was significantly associated with undergoing subsequent PPV (P = .02).

Two of 23 eyes (8.7%) developed an RD within 6 months of the last diagnostic procedure. Both of these eyes had a dry needle vitreous tap followed by anterior chamber tap and were culture negative. In one case, VA at presentation with endophthalmitis was 20/8000 with a macula-on RD developing 20 days later. It was successfully repaired both with scleral buckle and PPV, resulting in improved VA of 20/80 1 month postoperatively. In the second case, VA at presentation with endophthalmitis was also 20/8000, and a macula-on RD developed 10 days later with associated choroidal detachment. The RD was repaired with scleral buckle and PPV; however, VA remained poor at light perception (20/16  000).

Conclusions

While the EVS provides guidelines for diagnosing and managing endophthalmitis following cataract surgery and secondary intraocular lens placement, the results may not be generalizable to other causes of endophthalmitis, such as those that occur after intravitreal injection. The EVS found that eyes presenting with VA of light perception or worse achieved better visual outcomes after undergoing initial PPV compared with those undergoing initial vitreous tap and injection. ²⁰ Studies examining postinjection endophthalmitis, however, have shown that many cases can be successfully managed with initial vitreous tap and antibiotic injection alone, reserving PPV for cases with progressive clinical decline or persistent infection. ^18,21 Chen et al reported no statistically significant difference in visual outcomes between eyes undergoing needle vitreous tap and eyes undergoing PPV, although those that had undergone vitreous tap trended toward smaller changes in VA from baseline (4 vs 19 Early Treatment Diabetic Retinopathy Study letter difference). ²²

In our cohort, no eyes underwent initial PPV, and overall mean VA improved from 20/1145 at presentation to 20/113 at 6 months. However, the eyes that underwent subsequent PPV after initial treatment had worse VA not only at presentation (20/7129) but also at 6 months (20/1002). We found that a presenting VA of hand motion (20/8000) or worse was significantly associated with undergoing subsequent PPV. Similarly, we previously have reported that eyes with endophthalmitis due to various causes were also more likely to undergo initial PPV if the presenting VA was either counting fingers or hand motion. ²³ Particularly in the setting of improved surgical instrumentation and techniques since the time of the EVS, and with additional outcomes data, it is possible that PPV may be considered as an initial intervention for eyes with endophthalmitis following intravitreal injection and VA of hand motion or worse.

In our cohort, 30% of cases were culture positive, which was consistent with the range of 30% to 60% described in previous reports. ^{22,24 -27} Although culture-positive and culture-negative cases in our study did not significantly differ in VA at presentation or at 6 months, Xu et al have found in a larger cohort that culture-positive cases are more likely to have worse VA. ²⁷

As in other studies, ^3,27,28 the most commonly isolated organism in our cohort was coagulase-negative Staphylococcus, with other cases growing S viridans, S maltophilia, and P acnes. Other studies have suggested that S viridans–associated endophthalmitis may be due to aerosol contamination from oral or respiratory flora during intravitreal injection and is often associated with a more severe clinical course. ^7,27 One case in our cohort that was positive for S viridans experienced a decline in VA from presentation (20/8000) to 6 months (20/16000) despite undergoing subsequent PPV much sooner than the other 5 eyes that also underwent subsequent PPV (10 days vs 75.4 days). This finding is consistent with other reports in the literature suggesting endophthalmitis secondary to S viridans may progress more quickly and lead to poorer visual outcomes. ^3,18,19 As such, it may be reasonable to consider PPV as an early and even initial intervention in such cases, although further study is warranted.

In the setting of low microbiologic yield, the utility of anterior chamber and needle vitreous taps in guiding clinical management is of particular interest. While the sensitivity and specificity of anterior chamber tap are low compared with needle vitreous tap, ^29,30 aqueous cultures may aid in identifying organisms that were undetected in vitreous samples in some cases. ²⁹ Of the 4 eyes that underwent anterior chamber tap after initial dry needle vitreous tap in our study, 2 aqueous samples yielded cultures positive for coagulase-negative Staphylococcus and S maltophilia. Both cases had an improvement in VA at 6 months without a change in clinical management, so it is possible that the anterior chamber taps were not necessary in guiding clinical management.

The value of aqueous and vitreous sampling continues to be debated. Patel et al found that in cases of postsurgical endophthalmitis, culture-positive cases had poorer final VA and a higher rate of subsequent RD compared with culture-negative cases, but there were no changes in clinical management based on positive culture results, suggesting that vitreous cultures may have a greater value in determining prognosis than in clinical management. ³¹ In our cohort, the eye that yielded S viridans on subsequent PPV and mechanical vitreous biopsy underwent a second intravitreal injection of antibiotics and had poor final VA. While further study is needed, culture results may be more helpful in such cases where clinical signs indicate that additional treatment may be necessary and visual outcomes may be poor.

In addition to the EVS, other reports in the literature have recommended initial PPV in eyes with endophthalmitis that present with poor VA. ^{32 -36} One study examining postinjection cases and other causes found that eyes with presenting VA of hand motion or worse experienced greater visual improvement than eyes with VA of counting fingers after undergoing PPV for endophthalmitis within 72 hours of presentation. ³² In another cohort of 8 postinjection cases, most of which presented with a VA of hand motion, all patients experienced visual improvement by the final follow-up after undergoing PPV and silicone oil tamponade within 24 hours of the initial vitreous tap and injection. ³³

While eyes undergoing initial vitreous tap and injection may require a secondary intervention in the form of PPV or another injection, additional intervention after initial or early PPV has not been widely reported. ^{34 -36} Artunay and colleagues reported that early PPV within 48 hours after initial vitreous tap and injection of antimicrobials was beneficial for visual outcomes in their cohort, even in cases where VA had not yet declined to light perception, which was the threshold described in the EVS. ³⁵ Xu et al further added that, because of higher culture yield, initial PPV may be considered when identification of the causative organism would greatly aid clinical management, such as in the setting of suspected atypical or multidrug-resistant organisms. ²⁷ In the context of our finding that patients presenting with endophthalmitis and poor VA of hand motion or worse were more likely to undergo PPV, these studies support the consideration of earlier PPV in some patients, although future comparative studies would be necessary to better understand the role and outcomes of initial PPV in the management of such cases.

During the COVID-19 pandemic, it would be useful to continue to assess the rates of presumed infectious endophthalmitis in eyes receiving intravitreal injections since both physicians and patients wear masks. One study found similar rates of endophthalmitis when physicians wore a face mask compared with a “no-talking” scenario, although no cases secondary to oral flora were reported in the mask group. ³⁷ Recent studies examining face mask use by patients during simulated intravitreal injections have reported increased air flow toward the eyes through the superior edge of the mask, potentially increasing the risk of endophthalmitis associated with oral flora. ^38,39 Taping the superior edge of a properly worn mask, however, has been shown to reduce airflow, particulate counts, and colony-forming units in the periocular area. ^{39 -41} While wearing a taped mask may reduce infection risk, additional studies are warranted to investigate rates of presumed infectious endophthalmitis when the physician and patient are both wearing properly fitting masks.

Limitations of this study include its retrospective nature and relatively smaller sample size. In addition, because no eyes in our cohort underwent PPV as the initial treatment, we were unable to directly compare outcomes between those who underwent initial PPV and those who underwent initial vitreous tap as was done in the EVS. Future prospective studies with larger cohorts may further elucidate the role of initial PPV at various VAs in the treatment of postinjection endophthalmitis.

We present the clinical characteristics and visual outcomes in a cohort of eyes diagnosed with presumed infectious endophthalmitis following intravitreal injection over a 9-year period. In the absence of clear treatment guidelines beyond the results of the EVS, our findings add to the literature by demonstrating that eyes with VA of hand motion at 1 foot (30 cm) (20/8000) or worse were more likely to undergo subsequent PPV. Given these findings, further investigation is needed to determine whether earlier vitrectomy may improve outcomes in these patients.