Antivascular Endothelial Growth Factor Treatment Might Be Continued Even After Vitrectomy for Endophthalmitis in Neovascular Age-Related Macular Degeneration

Abstract

Purpose:

This article studies visual outcome and frequency of antivascular endothelial growth factor (anti-VEGF) injections continued in patients with neovascular age-related macular degeneration (AMD) who had an earlier vitrectomy for postinjection endophthalmitis.

Methods:

A retrospective interventional study was conducted reviewing our database for patients with a diagnosis of endophthalmitis in the course of anti-VEGF injections. Endophthalmitis diagnosis was made on clinical examination of pain, rapid decrease in visual acuity (VA), conjunctival hyperemia, hypopyon, and vitritis. In all eyes, core vitrectomy with intravitreal antibiotics was performed. Spectral-domain optical coherence tomography was performed monthly before and after surgery during follow-up. Anti-VEGF injections were continued after surgery in all cases.

Results:

Eight eyes with postinjection endophthalmitis were included. Mean VA immediately before endophthalmitis was 20/50 Snellen with a mean of 19 intravitreal anti-VEGF injections (P = .45). At time of endophthalmitis diagnosis, mean VA was 20/1000 (range, 20/2000-20/200). Mean time from injection to when the patient noted first symptoms was 4.3 days (range, 1-8 days). Mean time from first symptoms to surgery was 12 hours (range, 2.5-26 hours). Final mean VA at the end of follow-up (range, 12-84 months) did not statistically differ from VA at the visit immediately before endophthalmitis diagnosis (P = .69). Mean frequency of injections after vitrectomy did not significantly differ from the presurgical course of treatment (P =.97).

Conclusions:

Anti-VEGF treatment might be continued after vitrectomy for endophthalmitis and results in satisfactory anatomical and visual outcome. Surgery did not influence the frequency of anti-VEGF injections for neovascular AMD.

Introduction

Intravitreal injections are probably the most frequently performed ophthalmic procedures today. They have become the gold standard in the treatment of neovascular age-related macular degeneration (AMD), diabetic macular edema, vein occlusions, and other conditions. Endophthalmitis is one of the most serious complications of intravitreal injections. It has been reported that all antivascular endothelial growth factor (anti-VEGF) agents may have a relatively similar frequency of postinjection endophthalmitis, ¹ which has been reported to be 0.049% to 0.056%. ² Anti-VEGF treatment differs from other intraocular procedures because patients sometimes receive the injections dozens of times during their lifetime. It has been reported that the per-patient likelihood of developing endophthalmitis is 1% per 2 years. ³ The standards for prophylaxis have changed in recent years: povidone-iodine seems to be the most crucial, whereas antibiotics are now usually avoided. ^4,5

As with every infection, timing of treatment is crucial. However, there are no guidelines as to whether, in a specific case of postinjection endophthalmitis, a “tap and inject” procedure or vitrectomy should be performed. The Endophthalmitis Vitrectomy Study, which is the only evidence-based recommendation to date, was published more than 20 years ago (at a time when anti-VEGF treatment was not available) and presented data regarding endophthalmitis after cataract surgery. ⁶ The study concluded that vitrectomy should be performed only in patients with light perception. After 3 months, 41% of patients had visual acuity (VA) greater than 20/40, but 5% ultimately had no light perception.

Patients treated with anti-VEGF have a cumulatively higher risk of endophthalmitis than those treated with other ocular procedures. Additionally, in AMD patients, clear media are required as quickly as possible to evaluate the status of the fovea and to decide on reinjection.

Most papers focus on early results of vitrectomy and on methods of prevention of endophthalmitis. However, patients with choroidal neovascularization (CNV) in AMD have a chronic disease that does not resolve after vitrectomy for endophthalmitis.

Thus, the aim of this study is to present long-term results of anti-VEGF treatment performed after treating endophthalmitis with early vitrectomy.

Methods

We reviewed our electronic database (2004-2016) to identify patients in whom vitrectomy for endophthalmitis, in the course of neovascular AMD treated with anti-VEGF, was performed. Diagnosis of endophthalmitis was made on clinical examination of the following parameters: pain, rapid decrease in VA, conjunctival hyperemia, hypopyon, and vitritis. During 2004-2016 most patients were treated either with a treat-and-extend or an as-needed injection regimen. In neither course were patients scheduled for a routine postinjection follow-up examination the first days after the injection. As a routine, all our patients were informed to contact the clinic immediately if they observed decrease of vision, pain, and redness, and the phone number was available 24 hours per day, 7 days per week. Collected data included monthly VA checked with Early Treatment Diabetic Retinopathy Study (ETDRS) charts, number of days from causal injection, number of injections prior to endophthalmitis, elapsed time in hours from the patient’s first call reporting alarming symptoms of the injected eye, culture results, and number of injections after vitrectomy. Additionally, VA was checked 1 day and 1 week after surgery. Visual acuities were converted from ETDRS charts to logarithm of the minimum angle of resolution (logMAR) equivalents before analysis of visual acuities was performed. Central retinal thickness (CRT) was measured monthly and additionally 1 week after vitrectomy.

Spectral-domain optical coherence tomography (SD-OCT, Spectralis, Heidelberg Engineering) was performed monthly, both before and after surgery. Each month a single 7-mm scan and a 6 × 6-mm 3-dimensional scan were taken. The follow-up mode was used in all cases. Anti-VEGF injections were performed on an as-needed basis before and after surgery.

Injection Procedure

As a standard of care, we inform all treated patients about the risk of endophthalmitis. All patients are informed that if their eyes become red and painful and their vision decreases, they should call the clinic immediately. All injections were performed in the operating theater at our clinic. The eyelids were always washed once with alcohol and twice with povidone-iodine; the face was covered with a sterile cloth, and an eyelid speculum was placed. The conjunctival sac was washed with povidone-iodine for 30 seconds. Then, an anti-VEGF injection was administered. Until 2015, postoperative antibiotics were routinely used. Since that time, no antibiotics have been prescribed. The surgeon wore a face mask and sterile gloves. The patient and operating theater staff were discouraged from speaking during the injection procedure.

Surgical Technique

In 8 cases, core vitrectomy with intravitreal antibiotics infusion was performed. Three-port pars plana vitrectomy was performed. Samples were taken from the vitreous before opening the infusion line. Antibiotics were infused for approximately 10 minutes after the central part of the vitreous was removed. After surgery, patients received intravenous antibiotics for 7 days and topical antibiotics and dexamethasone 5 times daily for 3 weeks. Statistical analysis was performed with SigmaStat 3.5 for Windows.

Results

Eight eyes of 6 patients (mean age, 72 years; 3 men and 3 women) were identified from 13 002 injections (1 per 1625 injections or 0.01% per patient) during the years 2004-2016. In 3 additional patients who were not included in the current study, we performed tap and inject; all were confirmed to be sterile. All patients were continued on anti-VEGF afterward. One of the patients experienced endophthalmitis in both eyes and in 1 eye twice. Vitrectomy was performed in all 8 cases. Mean VA at the initial CNV diagnosis was 20/35 Snellen in these cases. Mean VA changed to 20/50 Snellen after a mean of 19 intravitreal bevacizumab injections (P = .45), which were performed before endophthalmitis occurred. At the first diagnosis of endophthalmitis, mean VA was 20/2000 Snellen. All the patients phoned and then arrived in the clinic after they experienced pain or visual loss, with an average mean time of 4.3 days after the injection (range, 1-8 days). The mean time from first symptoms noted by the patient to the time of surgery was 12 hours (range, 2.5-26 hours). The mean time from hospital admission to surgery was 3 hours (range, 1-7 hours). Mean VA was 20/50 Snellen 1 month after surgery. Final mean VA was 20/63 Snellen 1 year after vitrectomy and 20/50 Snellen at the end of follow-up (range, 12-84 months) (Figure 1). Final VA was not different when compared to VA at the last visit before endophthalmitis diagnosis (P = .2).

Figure 1.

Changes in visual acuity. BCVA indicates best-corrected visual acuity; logMAR, logarithm of the minimum angle of resolution.

One patient was pseudophakic at the time of occurrence of endophthalmitis. In other patients’ eyes there was no visually significant cataract. In 7 eyes cataract surgery was performed at 10-24 months after vitrectomy.

The mean CRT was 382 µm at presentation and decreased to 340 µm at the last measurement immediately before the anti-VEGF injection that led to endophthalmitis. CRT remained stable 1 week after vitrectomy for endophthalmitis but then gradually increased to reach a mean of 454 µm during the last follow-up (Figure 2). The mean final CRT was higher than the mean CRT immediately prior to the injection before endophthalmitis (P = .02) (Figures 2 -5).

Figure 2.

Changes in central retinal thickness (CRT).

Figure 3.

A 74-year-old woman with choroidal neovascularization in the setting of age-related macular degeneration. (A) Spectral-domain optical coherence tomography (SD-OCT) at presentation, before treatment was initiated. Visual acuity (VA) is 20/400 Snellen (1.3 logarithm of the minimum angle of resolution [logMAR]). (B) SD-OCT after 11 bevacizumab injections. VA is 20/40 Snellen (0.3 logMAR). (C) SD-OCT 1 month after vitrectomy for endophthalmitis. Endophthalmitis occurred 3 days after the last injection. Elapsed time from first symptoms noted by the patient (blurry vision) and vitrectomy was 24 hours, 4 hours since hospital admission. After the next 4 injections of aflibercept, the patient developed a second case of endophthalmitis in the same eye, treated again with vitrectomy. (D) SD-OCT 1 year and 10 injections after last vitrectomy. VA is 20/80 Snellen (0.6 logMAR).

Figure 4.

A 66-year-old man with choroidal neovascularization (CNV) in the course of age-related macular degeneration. (A) Spectral-domain optical coherence tomography (SD-OCT) at presentation, before treatment was initiated. Visual acuity (VA) is 20/20 Snellen (0 logarithm of the minimum angle of resolution [logMAR]). Paracentral scan is presented because CNV was not visible in the fovea. (B) SD-OCT after 6 bevacizumab injections. VA is 20/33 Snellen (0.22 logMAR). (C) SD-OCT 1 month after vitrectomy for endophthalmitis. Endophthalmitis occurred 5 days after the last injection. Elapsed time from first symptoms noted by the patient (pain and blurred vision) and vitrectomy was 10 hours. (D) SD-OCT 11 months and 7 antivascular endothelial growth factor injections after last vitrectomy. VA is 20/20 Snellen (0 logMAR). Only minor intraretinal fluid is visible near the optic nerve.

Figure 5.

(A) First spectral-domain optical coherence tomography (SD-OCT) scan taken in April 2006. The patient had already been treated for 2 years with pegaptanib sodium and bevacizumab (15 injections). Visual acuity (VA) is 20/25 Snellen (0.09 logarithm of the minimum angle of resolution [logMAR]). (B) SD-OCT after 40 bevacizumab injections. VA is 20/33 Snellen (0.22 logMAR). (C) SD-OCT 1 month after vitrectomy for endophthalmitis. The posterior hyaloid is visible. Endophthalmitis occurred 4 days after the last injection. Time from first symptoms noted by the patient (pain) and vitrectomy was 24 hours, 6 hours from hospital admission. (D) SD-OCT 24 months and 16 bevacizumab injections after last vitrectomy. The posterior hyaloid is visible. VA is 20/32 Snellen (0.22 logMAR).

Bacterial endophthalmitis was confirmed in 5 cases (Staphylococcus epidermidis in 3 eyes, Staphylococcus aureus in 1 eye, and Streptococcus sp in 1 eye).

In 1 patient the endophthalmitis occurred after the first injection, and in another patient after 4 monthly injections and then again in the same eye after 11 injections. In 6 other eyes endophthalmitis was noted after at least 1 year of anti-VEGF treatment. Those eyes were treated as needed (patients 1-3) or on demand (patients 4-6), and the mean interval between injections was 52 days. The injection course was analyzed for each patient and the frequency of needed anti-VEGF injections after endophthalmitis seemed unchanged compared to the frequency before endophthalmitis (P = .97) (Table 1). The mean interval between injections after diagnosis of endophthalmitis was 52 days.

Table 1.

Detailed Data on Number of Injections of Antivascular Endothelial Growth Factor (VEGF) for All Patients.

	Baseline Visual Acuity at Initiation of Treatment (Snellen)	Visual Acuity at Day of Last Injection Before Endophthalmitis	Final Visual Acuity	Culture Results	No. Injections Before Endophthalmitis	Mean No. Days Between Injections Before Endophthalmitis	Type of Anti-VEGF Before Endophthalmitis	Mean No. Days Between Injections After Endophthalmitis	Type of Anti-VEGF After Endophthalmitis
Patient 1	20/20	20/32	20/20	Streptococcus sp	6	104	Bevacizumab	107	Aflibercept
Patient 2	20/25	20/400	20/200	Staphylococcus epidermidis	30	40	Bevacizumab	33	Bevacizumab, aflibercept
Patient 3	20/400	20/63	20/80	S. epidermidis	4	31	Bevacizumab	36	Bevacizumab, aflibercept
Patient 3, second endophthalmitis	20/400	20/40	20/80	Culture negative	7	36	Bevacizumab	68	Aflibercept
Patient 3, second eye	20/32	20/32	20/32	S. epidermidis	24	56	Bevacizumab	53	Bevacizumab, aflibercept
Patient 4	20/25	20/32	20/40	Culture negative	40	51	Pegaptanib sodium, ranibizumab, bevacizumab	69	Bevacizumab, aflibercept
Patient 5	20/20	20/40	20/200	Staphylococcus aureus	1	0	Bevacizumab	52	Bevacizumab, aflibercept
Patient 6	20/125	20/63	20/400	Culture negative	31	46	Bevacizumab	56	Aflibercept

Conclusions

In this study we confirm that anti-VEGF injections might or even should be performed in the long term after diagnosis of endophthalmitis. Regular injections enable us to maintain stable VA. Moreover, the frequency of injections does not differ before and after vitrectomy (P = .97).

Endophthalmitis is a rare but serious adverse event of ophthalmic surgery. In the literature, bacterial endophthalmitis is confirmed in only approximately 60% of presumed cases, ⁷ compared with the 45% reported in our study. The most frequent causative organisms are Staphylococcus and Streptococcus.

Vitreoretinal techniques have progressed greatly since the publication of the Endophthalmitis Vitrectomy Study. ⁶ With the growing safety profile of vitrectomy and lack of clinical trial guidelines for postinjection endophthalmitis, it is not surprising that surgeons and patients may choose vitrectomy in cases with hand motion vision or better. The Endophthalmitis Vitrectomy Study suggested performing vitrectomy only if VA was light perception following cataract surgery. Five percent of patients in the study had no light perception at the last follow-up. Additionally, it has recently been reported that patients with endophthalmitis following intravitreal injections have an increased likelihood of having a final VA of counting fingers or less when compared with postcataract endophthalmitis. ⁸ One of the differences between the cited and present study is that the cited mean time from first symptoms to the first examination was reported to be 4.5 days, whereas in the present study we operated within a mean time of 12 hours from first symptoms. In the authors’ opinion, time is crucial in obtaining good final results, and prompt vitrectomy in our case series is probably associated with better visual outcome. The recently published Auckland Endophthalmitis Study reported at least no inferiority of vitrectomy to the tap and inject procedure. The mean final VA was reported to be 6/31, and further treatment was discussed in that study. ⁹ Visual acuity in the present series was 20/50 Snellen the day of the last injection before endophthalmitis, 20/50 Snellen 1 month after surgery, and was stable for at least 1 year. This result may suggest that early vitrectomy is a reasonable option in postinjection endophthalmitis.

Another recent study reports on 20 cases of endophthalmitis following anti-VEGF treatment. Final VA ranged from no light perception to 20/25, but the study did not focus on further anti-VEGF treatment. ¹⁰ We present results of at least 12 months’ follow-up after treatment for endophthalmitis (Figures 3 -5). The first injection after diagnosis of endophthalmitis was performed 1 month after surgery. Subsequent injections were performed on an as-needed basis, which enabled stabilization of VA.

It was previously substantiated that vitrectomy should not increase the clearance rate of anti-VEGF drugs ¹¹ ; thus there seems to be no reason for delaying vitrectomy and waiting for light perception if VA drops to less than 20/400 Snellen. Similarly, in our study the mean interval between injections did not differ between the time before (52 days) and after endophthalmitis (52.4 days) (P = .97). It must be considered that aflibercept is a longer-acting drug than bevacizumab. However, most patients with neovascular AMD do not require as frequent treatment over time. Those 2 factors together possibly influenced the fact that the frequency of injections did not significantly change.

Interestingly, in the literature several case reports demonstrated resolution of subretinal fluid neovascular AMD after treatment of endophthalmitis. Some authors contributed it to the upregulation of complement factor H, which might have a protective effect on AMD. ¹² Kokame and colleagues reported that 5 of 7 of their patients did not have further anti-VEGF treatment after treatment of endophthalmitis. However, it is difficult to estimate why those patients did not receive treatment because subretinal fluid is visible on images in the mentioned manuscript (Figure 2, case 3) of patients who apparently did not receive injections. It might be that the patients abandoned injections for psychological reasons, even if this was not mentioned in the study. The authors also presented cases with development of atrophy. ¹³ Although complement factor H has some protective effect, the generalization that performing vitrectomy stabilizes neovascular AMD seems to be an exaggeration. Further studies are still required regarding which patients might stabilize and which might not after vitrectomy. In the present study all patients required further treatment in a long-term observation period to maintain VA.

Patient 3 (Table 1) is worth mentioning. This diabetic woman experienced endophthalmitis 3 times in a 3-year period. The patient first had endophthalmitis in her left eye that was treated with pars plana vitrectomy and vancomycin. Her vision recovered to 20/40, and the endophthalmitis was completely resolved. Streptococcus epidermidis was confirmed. Four months later she again developed endophthalmitis in the same eye. Again vitrectomy with vancomycin was performed, but no specimens were identified from the vitreous. About 2 years later she developed another endophthalmitis in her better right eye. Again, S. epidermidis was confirmed, and the patient was treated with vitrectomy and antibiotics. Anti-VEGF treatment has continued with satisfactory visual results.

The limitations of this study include its retrospective nature and small sample size. A randomized clinical trial is needed to compare tap and injection with modern vitrectomy for postinjection endophthalmitis.

In conclusion, counseling of the patient regarding the possible side effects of intravitreal injections is crucial, as is early treatment. Vitrectomy may be considered as a treatment of postinjection endophthalmitis. Anti-VEGF injections should continue after the event to control wet macular degeneration and enable good functional results.