Postintravitreal Injection Endophthalmitis: Incidence, Characteristics, Management, and Outcome

Purpose Postintravitreal injection (IVI) endophthalmitis is a rare but devastating complication. Herein, we report the incidence ,and clinical and microbiological characteristics, as well as the visual outcome, in IVIs endophthalmitis in two medical centers. Methods All patients undergoing intravitreal injections between 1/2018 and 12/2019 in two large medical centers were analyzed for post-IVI endophthalmitis. Results Of the total of 51,356 IVIs performed, 23 cases of post-IVI endophthalmitis were diagnosed, yielding an overall incidence of 0.045%. The median interval from IVI to symptoms onset was 2 days (IQR: 1–5). Cultures were positive in 56% of the cases (100% Gram-positive bacteria and 76% coagulase-negative staphylococcus). Parameters associated with higher culture-positive rates included samples taken during vitrectomy, WBC on vitreous smear, the number of IVIs in the 12 months prior to presentation, and the time interval from last IVI to diagnostic sampling. At 6- and 12-month follow-up, the median change in VA (logMAR) was −1.10 (IQR: (−1.32)–(−0.40)) and −1.02 (IQR: (−1.10)–(−0.30)), respectively. Younger age and better BCVA at presentation were associated with better VA outcome, while positive culture result and systemic steroids treatment were each associated with the worse visual outcome. We found no difference in visual outcomes between PPV and TAI as a primary procedure. Conclusion Post-IVI endophthalmitis is a rare complication, and most patients do not regain their initial VA. Certain parameters (clinical, microbiological, and therapeutic) may help anticipate the outcome and guide decision making regarding diagnosis and treatment.


Introduction
Te use of intravitreal injections (IVIs) of steroidal and vascular endothelial growth factor (VEGF) antagonists has dramatically increased due to the widening spectrum of indications for its use, such as age-related macular degeneration (AMD), diabetic macular edema (DME), proliferative diabetic retinopathy (PDR), macular edema following retinal vein occlusion (RVO), and choroidal neovascularization (CNV) related to other diseases.Although highly efective, IVIs are not without risk, with infectious endophthalmitis being one of the more fearsome complications due to its poor prognosis.Tere is no consensus regarding the preinjection management to lower the risk of endophthalmitis.Te American Academy of Ophthalmology recommends the application of a topical anesthetic, application of 5% or 10% povidone-iodine drops and/or periocular povidone-iodine eyelid preparation, insertion of a sterile speculum to separate the lids, and reapplication of povidone-iodine immediately over the injection site prior to injection.According to the literature, the incidence of endophthalmitis following a single IVI is as low as 0.056%; however, many of the indications for IVIs require multiple injections over a long period of time, ultimately leading to a higher overall cumulative risk [1].
Te purpose of this study was to report the incidence, clinical and microbiological characteristics, visual outcome, and disease activity in post-IVI endophthalmitis and evaluate the relation of clinical and therapeutic factors to culture results and visual outcome.In addition, it was aimed to examine the efect of endophthalmitis on disease activity in the nAMD subgroup of post-IVI endophthalmitis patients.

Methods
Tis was a retrospective, case series of patients diagnosed with post-IVI endophthalmitis between January 2018 and December 2019 in two large medical centers.Te Institutional Review Board approved the study protocol and waived patients' consent for it is a retrospective study.Te research adhered to the tenets of the Declaration of Helsinki.

Inclusion and Exclusion
Criteria.All eyes diagnosed with infectious endophthalmitis within 6 weeks of IVI of any agent with at least 12-month follow-up were included.Post-IVI endophthalmitis was defned as any case in which clinical suspicion was high enough to warrant surgical intervention with the means of vitreous tap and intravitreal antibiotics injection ("tap and inject," TAI) and/or PPV.Exclusion criteria included patients treated for infammatory endophthalmitis without TAI and/or PPV, endogenous endophthalmitis, intraocular surgery, or trauma more recent than last IVI, as well as cases with the follow-up period shorter than 12 months.

Injection Technique.
Te standardized method for IVI in both medical centers is as follows.All injections were performed at a designated clinic at a supine position.Sterile drape and lid speculum were used.Prior to injection, topical anesthetic drops were instilled followed by topical 5% povidone-iodine ophthalmic solution.Injection was performed with a 30-gauge needle, 3.5-4.0mm from the limbus.Eye quadrant for needle insertion was chosen individually by the physician.A single dose of postinjection topical antibiotics was routinely instilled immediately postinjection.No additional prophylactic antibiotics are recommended at home.
Bevascizumab was dispensed in a syringe prepared by a compounding pharmacy, and ranibizumab was dispensed in a preflled syringe.Afibercept and triamcinolone were dispensed in a vial which was extracted to a syringe by the injecting physician immediately prior to IVI.

Management of Endophthalmitis.
In both medical centers, all eyes with presumed infectious endophthalmitis underwent TAI or early PPV based on physician decision according to the clinical manifestation.Vitreous tap was conducted using a 25-or 27-gauge needle in order to aspirate vitreous and subsequently inject intravitreal antibiotics.Based on the physician decision and according to the clinical manifestation, in some cases, immediate PPV was conducted and IVI of antibiotics was performed.IVI of antibiotics in TAI or PPV includes vancomycin (1 mg/0.1 ml) and ceftazidime (2 mg/0.1 ml).In cases of penicillin allergy, amikacin (400 mg/0.1 ml) was injected as a substitute for ceftazidime.Culture results and sensitivities guided subsequent intravitreal antibiotics.Topical antibiotics drops were given to all the patients hourly around the clock and then tapered down based on clinical improvement.Based on the physician decision, fortifed vancomycin (25 mg/ml) and fortifed ceftazidime (50 mg/ml) or moxifoxacin hydrochloride (0.5%) were given.Cycloplegia (topical atropine sulfate 0.5% drops or cyclopentolate 2% drops) was given in all cases.Topical steroids (dexamethasone 0.1% or prednisolone acetate 1%) were also given in all cases.Systemic steroids were added according to physician discretion, given daily (0.5-1 mg/kg/day) and tapered gradually according to clinical response over a 6-8 week period.Patients were evaluated daily, and upon clinical improvement, topical treatment was tapered down and follow-up intervals were extended.

Data Collected.
Te yearly number of IVIs was determined using billing codes.Medical records were reviewed to identify patients treated for endophthalmitis within 6 weeks after IVI during the study period.Collected data included demographics, underlying diagnosis indicating IVI treatment, treatment history (agent injected, number of injections at 12 months previous to endophthalmitis, and date of last IVI), clinical fndings on slit-lamp biomicroscopy at presentation of endophthalmitis (corneal edema, anterior chamber cells, anterior chamber fbrin, hypopyon, posterior synechiae, vitritis, preretinal exudates, and intraretinal hemorrhages), initial procedure (TAI or PPV), secondary procedure (PPV), topical antibiotics drops treatment, systemic steroids treatment, white blood cells (WBCs) on direct smear, culture results, last recorded best corrected visual acuity (BCVA) before IVI ("baseline BCVA"), and BCVA at 2 Journal of Ophthalmology presentation and at 6-and 12-month follow-up.BCVA was measured using Snellen charts and subsequently converted into logMAR values for statistical analysis.Time intervals between last IVI and symptoms onset, between IVI and the frst procedure, and between symptoms onset and the frst procedure were recorded.
2.5.Outcomes.Primary outcome was BCVA at 6 and 12 months following treatment as well as the change in BCVA at these time points in comparison to the baseline BCVA and BCVA at presentation.Secondary outcomes were the incidence of post-IVI endophthalmitis, BCVA, and clinical fndings on presentation, as well as microbiological characteristics.
In addition, we evaluated the association of several demographic, clinical, laboratory, and management factors to visual outcomes including return to the baseline BCVA, any improvement in BCVA from presentation, and relatively low BCVA (defned as BCVA of 6/60 and lower).

Statistical Analysis.
Categorical variables were summarized as the frequency and percentage.Continuous variables were evaluated for normal distribution using histogram and reported as the median and the interquartile range (IQR).Te chi-square test and Fishers' exact test were applied to compare proportions between categorical variables.Continuous variables were compared between categories using the Kruskal-Wallis test and the Mann-Whitney test.Te Spearman correlation coefcient was used to study the association between continuous variables.Te Wilcoxon test was used to compare continuous variables between 2 time points.All statistical tests were two sided, and P < 0.05 was considered statistically signifcant.Data were analyzed using IBM SPSS statistical software version 25.0 (Armonk, NY: IBM Corp).

Endophthalmitis Rates.
During the study period, a total of 51,356 IVIs were performed in both medical centers.Te injections rate of the diferent medications is detailed in Table 1.

Clinical Presentation.
Te most common presenting symptom was decrease in vision in 20 (87%) patients, and the most common clinical signs were anterior chamber cells and vitritis seen in all patients (100%).Te incidence of presenting signs and symptoms is presented in Table 2. Te median interval from IVI to symptoms onset was 2 days (IQR: 1-5), from symptoms onset to the frst procedure was 1 day (IQR: 0-3), and from IVI to the frst procedure was 4 days (IQR: 2-7) (Table 2).

Laboratory Results
. Direct smear of the vitreous sample (from TAI or PPV) showed white blood cells (WBCs) in 12 (52%) patients, of which 11/12 (92%) eventually had positive culture results (positive predictive value of 92%).WBC on direct smear were associated with a higher rate of positive culture result (91.7% vs. 18.2%,P � 0.001).

Journal of Ophthalmology
A higher culture-positive rate was found in cases with longer time interval from IVI to any frst procedure (median 7, IQR: 4-10 vs. median 3.5, IQR: 2-4, P � 0.007).Other parameters analyzed were not associated with a higher culture-positive rate (P > 0.05).Tese include patients' age, injected agent, presenting signs or symptoms, BCVA at presentation or its decrease from baseline, time interval from IVI to symptoms onset, and time interval from symptoms onset to the frst procedure.

Visual Outcome.
Median BCVA at the baseline, at presentation, and at 6-and 12-month follow-up is detailed in Table 3. 20 (87%) patients presented with vision loss in comparison to the baseline BCVA.BCVA improvement from presentation was documented in 18 (78%) and 17 (74%) patients at 6-and 12-month follow-up, respectively.8 (35%) and 9 (39%) patients return to their baseline vision (within one line from the baseline BCVA) at 6-and 12-month follow-up, respectively.BCVA distribution along the study period is shown in Figure 1.
Tere was no diference in fnal VA between the two primary procedures (early PPV or TAI) (P > 0.999).Analysis of baseline characteristics revealed no statistically signifcant diference between the two groups (early PPV vs. TAI), including demographics, disease characteristics, presenting signs and symptoms, laboratory fndings, systemic steroids use and time intervals between injection, symptoms onset, and primary procedure.
Patients who received systemic steroids treatment were found to have a lower rate of improvement in VA along the follow-up (54.6% vs. 100% P � 0.035).Analysis of baseline characteristics revealed no statistically signifcant diference between the two groups (systemic steroids treatment vs. no systemic steroids treatment), including demographics, disease characteristics, presenting signs and symptoms, laboratory fndings, initial procedure (TAI vs. early PPV) and time intervals between injection, symptoms onset, and primary procedure.

Discussion
In this retrospective, case-series study, we report the incidence rate, clinical and microbiological characteristics, management, and outcome of post-IVI endophthalmitis in two medical centers, as well as evaluate the correlation between the presenting signs and symptoms, culture results, management, and visual outcome.

4.1.
Incidence.In our study, the overall incidence of post-IVI endophthalmitis was 0.045% which is comparable to the incidence reported in the literature [1].Te incidence of endophthalmitis after steroidal agents' injection was signifcantly higher in comparison to anti-VEGF agents (Table 1) as previously described [4].
Bevacizumab and afibercept each showed signifcantly higher rates of endophthalmitis compared to ranibizumab (Table 1).Tis fnding is possibly due to the use of preflled syringe of ranibizumab.Mun et al. reported a signifcant association of bevacizumab with endophthalmitis compared to nonbevacizumab drugs, combining ranibizumab and afibercept [17].A similar fnding was reported by Xu et al. in a study of 258, 357 anti-VEGF IVIs [18].Tis fnding might be related to the division process into individual aliquots in bevacizumab vials.A higher incidence of post-afibercept endophthalmitis compared to ranibizumab was described before by Kiss et al. [19].

Presentation.
Te most common presenting symptom was decrease in vision, followed by pain and redness, as described before in the literature [14].Te median interval between IVI and symptoms onset was 2 (IQR 1-5), which is comparable to the incidences reported in the literature [18,20].

Laboratory Results.
WBCs on direct smear of vitreous sample were associated with a higher rate of positive culture results.Tis fnding may assist in early prognostication as positive culture results were shown to have worse visual outcome [4,[11][12][13].Te overall culture-positive rate (57%) was comparable to the rate described in the literature with a higher culture-positive rate among samples taken during vitrectomy (75%) versus vitreous tap (36%) [7,8].Tis fnding might represent a higher yield of a vitreous sample taken by vitrectomy in comparison to vitreous tap.It was suggested to be a result of selection bias of cases that are more severe clinically, with higher bacterial load in the vitreous taken to early vitrectomy [18].On the other hand, in our series, the early PPV and TAI subgroups did not difer in several presenting signs and symptoms as detailed.In 3 cases, the vitreous sample volume was insufcient for laboratory evaluation.Although this fnding represents real life results, it might as well bias our results.
In one case, there was a positive culture of Globicatella sanguinis which grew as a coinfection with Staphylococcus simulans.In this case, visual acuity deteriorated from 6/10 preinjection to hand motion 12 months following the event of endophthalmitis.To our knowledge, we report the frst post-IVI endophthalmitis caused by Globicatella sanguinis, an uncommon cause of human infection that afects the bloodstream, urinary tract, and central nervous system [21][22][23][24].Only one case of Globicatella sanguinis endophthalmitis was described in the literature of a 9-year-old healthy boy presented with corneal abscess and endophthalmitis [25].Globicatella sanguinis is a catalase-negative, nonhemolytic, Gram-positive coccus which was frst described in 1992 by Collins and colleagues [26].Tis microorganism has been noted as a colonizer of the skin, which might explain the coinfection with Staphylococcus simulans, a member of the CoNS group, known to colonize the skin and mucous membranes [24,27].We believe that the low incidence described in the literature is due to relatively recent identifcation of Globicatella sanguinis as a specifc pathogen, as well as advanced laboratory tests required for its identifcation, as standard phenotypic techniques are usually insufcient [25].
Te time interval from IVI to the frst procedure was signifcantly associated with higher culture-positive rates.Tis interval may allow a longer time for bacterial growth and thus a higher bacterial load, resulting in a higher rate of positive culture results.
In a series of 65 cases of endophthalmitis post-IVI, Dossarps et al. reported no signifcant diference in the median number of IVIs before endophthalmitis between culture-positive and culture-negative cases [28].Similarly, we found that accumulative IVIs number at 12 months before endophthalmitis was not associated with a higher culture-positive rate.
Evidence regarding the correlation between presenting signs and symptoms to culture results in endophthalmitis is inconsistent.While Chirag et al. found no signifcant correlation, Collins et al. reported a correlation between several parameters of the clinical presentation and microbiologic culture results [26,29].In our study, presenting signs and symptoms, BCVA at presentation or its decrease from the baseline, were not associated with a higher rate of positive culture results.Tese fndings might be due to the narrow spectrum of culture results in our study (100% were Gram-positive bacteria and 76% were CoNS positive).Alternatively, it might be due to the relatively small number of endophthalmitis cases in our study.

Visual Outcome.
Te rate of vision improvement (within one line from the baseline BCVA) compared to presentation was 78% and 74% at 6-and 12-month followup, respectively.35% and 39% of the patients return to their baseline vision at the 6-and 12-month follow-up, respectively.Tese rates are comparable to those described in other studies [18].
Of the demographic parameters evaluated, younger age was associated with better visual outcome.In a recent study by Xu et al. of 40 patients with post anti-VEGF IVI endophthalmitis, younger age was associated with better visual outcome as well [18].
Of the presenting symptoms, we found that better visual acuity at presentation was associated with the better visual outcome, while Xu et al. reported that BCVA at presentation did not correlate to BCVA at the 6-month follow-up [18].
A positive culture result was signifcantly associated with the worse visual outcome, a fnding that is inconsistent in previous studies and seems to be afected by the specifc pathogens among the culture-positive subgroup [1,18,28,30].
We found no diference in visual outcomes between PPV and TAI as a primary procedure.In contrast to endophthalmitis following cataract extraction surgery, there are no guidelines for the treatment of post-IVI endophthalmitis [31].Based on the retrospective analyses published, there seems to be no signifcant diference in the outcome between patients treated frst by TAI or PPV and our study supports this fnding [32].
We found worse visual outcome in patients who received systemic steroids treatment.Tere is no consensus in the literature regarding the use of systemic corticosteroids in endophthalmitis.A retrospective trial by Robbins et al. of 133 eyes with endophthalmitis (23 post-IVI) found a higher   Journal of Ophthalmology rate of vision improvement of 3 lines or more among the 25% of patients treated with systemic steroids [33].Tey reported that oral steroids use was associated with culturepositive endophthalmitis, hypotony, conjunctival hyperemia, and anterior chamber fbrin on examination.We found a 48% rate of oral steroids treatment, and our analysis revealed no diference in demographic characteristics, disease characteristics, presenting signs and symptoms, culture results, primary procedure, and measured time intervals between patients treated with systemic steroids and patients that were not.
Our study has several limitations.First, its retrospective nature weakens the conclusions, and the relatively small number of endophthalmitis cases did not allow a multivariable analysis.Although the retrospective nature of this study allows an underestimation of the endophthalmitis rate, we believe the possibility of missing reports is low as our centers are of the largest in the country and the communication between retina teams in a small country like ours is fuent.Second, although all patients were managed by one retina team, the chosen primary procedure and the use of systemic steroids were chosen according to physicians' decision and not by defned protocols, as the literature lacks such protocols, which may result in selection bias.

Conclusion
We report an overall endophthalmitis rate of 0.045% after IVI of anti-VEGF or corticosteroids agents in two large medical centers.Tis rate was signifcantly higher in steroidal agents (vs.anti-VEGF agents) and in bevacizumab or afibercept (vs.ranibizumab).Younger age and better VA at presentation were associated with the better visual outcome, while positive culture result and systemic steroids treatment were each associated with the worse visual outcome.Te chosen initial procedure by TAI versus PPV was not associated with a diferent visual outcome.
Our fndings may help anticipate outcome and guide decision making regarding diagnosis and treatment of postinjection endophthalmitis.Further prospective trials are warranted in order to establish management guidelines in post-IVI endophthalmitis.

Table 1 :
Intravitreal injections rate and the endophthalmitis rate in two medical centers during the study period.
*( * ) P � 0.025 for steroidal agents in comparison to anti-VEGF agents; P � 0.037 for bevacizumab in comparison to ranibizumab; P �

Table 3 :
Best corrected visual acuity (logMAR) in post-IVI endophthalmitis along the study period.