The mean visual acuities are indicated by the red squares. P values indicate significance of visual acuity changes compared with before injection.
eTable 1. Baseline Characteristics
eTable 2. Presentation Characteristics and Visual Acuity Changes
eTable 3. Injection Characteristics
eTable 4. Management Details
Hahn P, Chung MM, Flynn HW, Huang SS, Kim JE, Mahmoud TH, Sadda SR, Dugel PU. Postmarketing Analysis of Aflibercept-Related Sterile Intraocular Inflammation. JAMA Ophthalmol. 2015;133(4):421-426. doi:10.1001/jamaophthalmol.2014.5650
Aflibercept-related sterile inflammation, an event that is poorly understood, has been the subject of ongoing postmarketing reports.
To analyze cases of aflibercept-related sterile inflammation reported to the American Society of Retina Specialists (ASRS) Therapeutic Surveillance Committee (TSC), an independent task force formed to monitor drug- and device-related safety events.
Design, Setting, and Participants
A retrospective review of 56 cases in 55 patients was performed of all cases of sterile inflammation after aflibercept injection that were voluntarily reported by 12 practices throughout the United States to the ASRS TSC from December 1, 2011, through February 12, 2014.
Main Outcomes and Measures
Cases of aflibercept-related sterile inflammation were analyzed for baseline and demographic information, presenting symptoms and findings, visual acuity changes, injection technique, and management details.
Among 56 reported cases of sterile inflammation, mean time to onset was 3.5 days (median, 2 days; range, 0-30 days). Most cases consisted of initial loss of vision and intraocular inflammation without prominent redness, severe pain, or hypopyon. Thirty-seven cases (66%) were treated with topical corticosteroids and/or observation alone. Mean time to resolution was 28.6 days (median, 28 days; range, 4-84 days). Although final vision was overall unchanged, some patients developed permanent vision loss, which may have resulted from inflammation and/or progression of the underlying disease. Age older than 80 years was associated with worse visual outcomes. No difference in visual outcome was detected in patients with sterile inflammation undergoing topical therapy alone vs invasive procedures (vitreous biopsy and/or intravitreal antibiotic administration and/or vitrectomy).
Conclusions and Relevance
With the largest number of reported cases of aflibercept-related sterile inflammation to our knowledge, this analysis suggests typical findings and an often favorable prognosis of this event. Analysis of real-world, postmarketing data has limitations, and these findings should be considered as hypothesis-generating assessments rather than a definitive reflection of this event or its typical course. Distinguishing sterile inflammation and infectious endophthalmitis at the time of presentation may often be difficult, and cautious evaluation and management of these patients are warranted. The ASRS TSC encourages active postmarketing surveillance by all physicians.
The American Society of Retina Specialists (ASRS) Therapeutic Surveillance Subcommittee (TSC) serves as an independent task force to objectively monitor drug- and device-related adverse events on behalf of the greater community of retina specialists and their patients. The ASRS TSC has maintained a history of prompt investigation of adverse events, including identification of a fungal endophthalmitis outbreak related to improperly compounded Brilliant Blue G in a single responsible pharmacy,1 which was subsequently closed, and comprehensive classification and analysis of postmarketing adverse events reported after ocriplasmin (Jetrea; ThromboGenics Inc) administration. Within the first 3 months of aflibercept use (Eylea; Regeneron Pharmaceuticals Inc) after approval by the Food and Drug Administration (FDA) in November 2011, a cluster of aflibercept-related sterile intraocular inflammation was similarly reported. The ASRS TSC received reports of 15 cases in those 3 months, surveyed the reporting physicians for detailed information surrounding these events, and reported findings of sterile inflammation not associated with any identifiable causative factors.2 Of note, 47% of those cases were successfully treated with a return to baseline vision with topical corticosteroids alone, demonstrating that invasive interventions, including vitreous biopsy and intravitreal antibiotic administration or vitrectomy, may be avoided in select circumstances. Subsequent cases of sterile inflammation after aflibercept have continued to be submitted to the TSC. This report serves to describe these events and to analyze these data for any associated factors, consistent with the mission of the TSC to investigate and inform the ophthalmology community regarding drug- and device-related safety events.
This study was deemed exempt from institutional review board review or oversight by the Research Subjects Review Board of the University of Rochester. The ASRS TSC responded to unsolicited, voluntary reports of all drug- and device-related adverse events, which were submitted online through a TSC reporting form on the ASRS website or through personal communication with TSC members. In response to reports of aflibercept-related sterile inflammation submitted from December 1, 2011, through February 12, 2014, the TSC requested detailed information surrounding these events, including injection and adverse event dates; eye treated; history of prior ranibizumab (Lucentis; Genentech Inc), bevacizumab (Avastin; Genentech Inc), and aflibercept treatment; presenting symptoms and signs; demographic information (age, sex, race, lens status, medical and ocular history, and disease treated); visual acuity (VA) at baseline, presentation, and resolution; time to resolution; injection parameters (dose, lot number, preparation used, quadrant injected, anesthesia protocol, and syringe and needle size and brand); and management details.
All cases were diagnosed as sterile intraocular inflammation at the discretion of the reporting physician. The VA was converted from Snellen to logMAR units for quantitative analysis. Counting fingers at “X” feet was assigned “X” divided by 200 Snellen acuity for conversion to logMAR units. Hand motion was assigned 3 logMAR. The Wilcoxon test of difference between medians was used to compare outcomes.
There were 56 cases in 55 patients of aflibercept-related intraocular sterile inflammation reported by 12 practices throughout the United States, with a range of 1 through 9 cases in each practice. Physicians anecdotally reported that these cases often arose in groups. Baseline and demographic information for these patients and eyes are reported in Table 1 and eTable 1 in the Supplement. Most eyes (45 [80%]) were treated with aflibercept for neovascular age-related macular degeneration, and all reported eyes had a history of prior treatment with anti–vascular endothelial growth factor agents. More left eyes (35 [63%]) were treated than right eyes (20 [36%]); one patient was treated with bilateral, same-day, same-lot injections with subsequent onset of bilateral, simultaneous sterile inflammation.
Of the 46 eyes for which symptoms at presentation were reported, the most common symptoms included blurry vision (n = 35 [76%]) and floaters (n = 21 [46%]) (eTable 2 in the Supplement). Any level of pain, ache, or a sore eye was specifically noted in 21 eyes (46%). Severe pain was noted in 1 eye (2%), and absence of pain was specifically noted in 7 eyes (15%). Reference to pain was not reported in 17 eyes (37%). Redness was noted by the patient in 6 eyes (13%).
Of the 19 eyes for which examination findings at presentation were reported, all were associated with intraocular inflammation. Fourteen eyes (74%) had anterior chamber inflammation and vitritis, 3 (16%) had anterior chamber inflammation only, and 1 (5%) had vitritis only. In 1 eye (5%), only uveitis without further identifiers was reported. Infrequent findings included corneal edema (n = 4 [21%]), hypopyon (n = 1 [5%]), and intraretinal hemorrhages in the superotemporal quadrant (n = 1 [5%]). Of note, conjunctival injection was not specified in any of the 19 cases; in 3 cases, absence of conjunctival injection was specifically noted.
Baseline (preinjection) VA was reported in 46 eyes with a median of 20/40 Snellen (mean, 0.46 logMAR or 20/58 Snellen) (Figure and eTable 2 in the Supplement). Mean time to onset of symptoms was 3.5 days (median, 2 days) after aflibercept injection, with a range of 0 through 30 days. At presentation, VA was reported in 19 eyes with a median of 20/400 Snellen (mean, 1.35 logMAR or 20/448 Snellen). The mean VA loss at presentation compared with baseline was significant at 20/160 Snellen (0.89 logMAR) (P < .001). After resolution at the discretion of the reporting physician, final VA was reported in 45 eyes with a median of 20/40 Snellen (mean, 0.48 logMAR or 20/60 Snellen). Compared with baseline, no significant mean change in VA was identified after resolution (P = .85).
Of the 45 eyes for which initial and final VA were reported, 27 (60%) regained equal or better vision compared with baseline VA, 10 (22%) lost 1 Snellen line, 5 (11%) lost 2 lines, 1 (2%) lost 3 lines, and 2 (4%) lost greater than 3 lines. Both patients who lost greater than 3 lines were recorded as having final counting fingers vision that had deteriorated from 20/70 and 20/200 baseline VA. The reporting physicians did not report whether this loss of vision was considered to be secondary to inflammation or from progression of underlying disease. In addition, one patient who was treated with intravitreal antibiotics initially improved to within one line of baseline VA after 3 months but developed a retinal detachment 8 months after the injection with subsequent counting fingers vision. Of these 45 patients, mean time to resolution, as reported in 35 eyes (78%), was 28.6 days (median, 28 days; range, 4-84 days). Recovery was still ongoing in 2 (4%) and unreported in 8 (18%).
Most eyes for whom the administered aflibercept dose was reported were given the on-label dosage of 2 mg in 0.05 mL (Table 2 and eTable 3 in the Supplement). Lot numbers for 53 cases were reported, implicating 18 different lots. Each lot was associated with a mean of 2.9 cases (median, 2.5 cases; range, 1-9 cases). Four lots were associated with events in multiple practices, including one lot associated with 9 reports in 3 practices.
Standard injection techniques were reported. All eyes with reported information were prepared with povidone-iodine. A balance number of eyes were injected in the superotemporal and inferotemporal quadrants. Most eyes received topical anesthesia with injection using a 30-gauge needle. All eyes (n = 45) were injected with a 1-mL syringe, and 34 (76%) were injected with a prepackaged 1-mL Luer-Lok syringe (Becton, Dickinson and Company). The details were not specified for 10 eyes (22%). Only one eye (2%) was specified to have been injected with a “Becton Dickinson TB syringe” that was not part of the aflibercept packaging.
Noninvasive treatments that consisted primarily of topical corticosteroids only were administered in 37 eyes (66%). Invasive treatments, primarily vitreous biopsy with administration of intravitreal antibiotics, were used in 17 eyes (30%) (Table 2 and eTable 4 in the Supplement). No difference in VA at the time of presentation (P = .96) was identifiable between those treated with and without invasive procedures.
Of the 15 eyes with intraocular biopsy samples, 13 tested negative for growth. In one eye, a simultaneous anterior chamber biopsy yielded “rare, broth-only Streptococcus, α-hemolytic” on day 1, but the vitreous sample test result remained negative. In another eye with a negative Gram stain result and without pain and injection, Staphylococcus (species not specified by the reporting physician) in the vitreous biopsy was interpreted as a contaminant.
Aflibercept rechallenge was reported in 3 eyes from 2 patients; none developed recurrent inflammation. Rechallenge was reported not to have been attempted in 15 eyes, and no information was provided in 38 eyes.
Subgroup analysis was performed to assess for any variables that affect visual outcomes (Table 3). Age older than 80 years was significantly associated with worse visual outcomes. Lens status, history and number of prior aflibercept injections, time to resolution, and management approach had no effect on visual outcomes.
We present results from an independent analysis by the ASRS TSC of 56 cases of presumed sterile inflammation after aflibercept injection. This analysis was based on voluntary reports submitted during 27 months to the TSC without specific solicitation to individual physicians. The TSC had previously reported an analysis of the first 15 of these cases, which occurred within the first 3 months after FDA approval of aflibercept.2 These 56 cases should not be considered comprehensive, and the TSC is aware of both published and currently unpublished series that were not reported to the TSC and therefore not included in this analysis. In one published report,3 20 cases were identified from a single practice with similar characteristics and outcomes to those presented in this report. In another report,4 2 cases of panuveitis arose more than 4 weeks after aflibercept injection.
The 56 cases reported herein, the largest group reported thus far to our knowledge, consist of sterile inflammation treated with topical therapy alone and presumed sterile inflammation with negative cultures after vitreous biopsy and/or intravitreal antibiotics and/or vitrectomy. This report included all cases submitted as sterile inflammation at the discretion of the reporting physician. It is possible that some of these cases represent culture-negative infectious endophthalmitis and not sterile inflammation. However, VA at presentation and after resolution were not different based on management approach, suggesting a similar disease process among these cases and corroborating a diagnosis of sterile endophthalmitis. Furthermore, no overall change in final VA was measured in either group, in contrast to the poor visual prognosis typically associated with infectious endophthalmitis. Comparison of VAs in this study was based on conversion from Snellen to logMAR units, which is an inherently flawed approach given that Snellen charts do not follow a logMAR progression and that nonreading VAs (hand motion and counting fingers) were arbitrarily assigned logMAR designations. Moreover, these real-life data were often inconsistent and incomplete, further confounding any such comparisons. These comparative evaluations of VAs should therefore be interpreted cautiously as hypothesis-generating and not definitive assessments. In addition, these cases were identified based on retrospective selection after resolution, identified as presumed sterile endophthalmitis given their typical courses, including favorable outcomes. As always, at the time of presentation, infectious endophthalmitis should remain at the top of any differential diagnosis.
Analysis of these cases suggests that there may be key differentiating features between aflibercept-related sterile inflammation and infectious endophthalmitis. In this report, the median time to presentation was 2 days after injection but can be as late as 30 days. Eye redness was reported in 13%, but no physician reported any conjunctival injection on examination. Although retinal physicians may be likely to underemphasize conjunctival injection, particularly because dilating eye drops may mitigate injection, individual communications between the TSC and reporting physicians suggest that many of these eyes were indeed white. Pain at any level was reported in 21 cases (46%), and severe pain was reported in only a single eye. Communications with reporting physicians anecdotally suggest that severe pain was not a predominant finding. Although the eye with severe pain was treated with intravitreal antibiotics and vitreous biopsy, only 7 (33%) of the 21 eyes with reported pain were treated with invasive procedures (vitreous biopsy and/or intravitreal antibiotics and/or vitrectomy), similar to the 30% of all 56 eyes in this series that were treated with invasive procedures. In these voluntary reports, pain was not objectively scaled, and we hypothesize that the pain reported was not uniformly the severe pain characteristic of infectious endophthalmitis, which likely would have prompted invasive procedures. Intraocular inflammation, sometimes severe, was generally present, but hypopyon was reported in only one case. A total of 37 patients (66%) were treated with topical treatment alone. Resolution occurred within a median of 28 days, generally with a return to baseline VA. Although older age was associated with worse visual outcome, a finding that may reflect diminished healing responses in elderly patients and warrants further investigation, no other variables, including management approach, affected visual outcomes. These characteristics are in contrast to infectious endophthalmitis, which classically has severe pain and redness as predominant symptoms, often with a more delayed presentation, slower recovery time, and poor visual prognosis despite invasive interventions. On the basis of the results from this and others’ reports,2,3 a diagnosis of sterile inflammation may be cautiously considered in cases of intraocular inflammation after intravitreal injection if presenting within a few days in an otherwise white eye without severe pain or hypopyon. In many cases, however, such as if the eye is not white and comfortable, differentiation between sterile inflammation and infectious endophthalmitis may not be straightforward. Given the paramount importance of prompt and aggressive treatment of infectious endophthalmitis, it is prudent to assume an infectious cause for cases in which it may be difficult to distinguish between a sterile and infectious source. Patients undergoing intravitreal injections should similarly be counseled to report new or worsening symptoms promptly.
In our analysis, no responsible factors could be implicated in the pathogenesis of these cases. Neovascular age-related macular degeneration was the most common disease treated, and the patients were correspondingly predominantly white, pseudophakic, and elderly, likely consistent with the patient population rather than indicative of any causative risk factors. Most cases had a history of prior aflibercept injections, and some cases were successfully rechallenged with aflibercept, suggesting that these cases may not arise from a patient-specific immunologic response. Injections were generally performed with standard techniques, including povidone-iodine before injection, topical or subconjunctival lidocaine, use of the syringe and needle included in the aflibercept packaging, and administration in the superotemporal or inferotemporal quadrants. Although some lots were associated with multiple cases in multiple practices, which may implicate a manufacturing concern, many lots were associated with a single report. One patient in this series experienced bilateral inflammation after bilateral, same-day, same-lot injections, but at least one case of unilateral sterile inflammation after bilateral, same-day, same-lot injections has come to the attention of the TSC (H.W.F., oral communication, August 26, 2013). This unilateral event suggests that the cause is likely not due to a single factor, including manufacturing contamination, patient-specific immune reactions, or physician-related procedural issues, which should have resulted in bilateral safety events, but rather supports a multifactorial or random cause.
Treatment appears to have been successful in certain cases with topical therapy alone. Increased awareness of this event within the ophthalmology community may reduce invasive treatments in select patients that may be associated with increased complications. Use of topical therapy in reported cases increased from 47% in the first 3 months of aflibercept use2 to 66% during 27 months, likely representing increased education and awareness of this event. However, there are no fail-safe distinguishing features of sterile vs infectious endophthalmitis, and clinical judgment of disease severity should determine treatment strategy.
On the basis of the 56 cases in this report, estimation of a rate of inflammation was not attempted because of the underreporting inevitably associated with voluntary, postmarketing analysis. Other limitations of this postmarketing analysis include absence of uniform diagnostic criteria: cases of sterile inflammation were diagnosed at the discretion of the physician without defined guidelines. In addition, the data set of postmarketing analyses is subject to the nature of voluntary reporting and ongoing experiences and is often incomplete and nonstandardized. Given these limitations inherent to any postmarketing analysis, this report should serve as a descriptive presentation to increase our aggregate understanding of aflibercept-related sterile inflammation. Findings should be considered as hypothesis-generating rather than definitive reflections of this event, its course, or its prognosis, and as always, rigorous clinical studies are warranted.
The ASRS TSC was formed to objectively monitor drug- and device-related adverse events on behalf of the greater ophthalmology community. Active postmarketing surveillance is critical to our understanding of an increasing number of available therapies. This report suggests typical features of aflibercept-related sterile inflammation and its prognosis in the largest such report to date, to our knowledge. As postmarketing monitoring continues, the ASRS TSC will continue to analyze ongoing reports of all drug- and device-related adverse events to learn more about the findings and pathogenesis associated with these events.
Submitted for Publication: August 26, 2014; final revision received October 28, 2014; accepted October 31, 2014.
Corresponding Author: Pravin U. Dugel, MD, Retinal Consultants of Arizona, 1101 E Missouri, Phoenix, AZ 85014 (firstname.lastname@example.org).
Published Online: January 15, 2015. doi:10.1001/jamaophthalmol.2014.5650.
Author Contributions: Dr Dugel had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Hahn, Chung, Flynn, Huang, Kim, Mahmoud, Dugel.
Acquisition, analysis, or interpretation of data: Hahn, Chung, Flynn, Huang, Kim, Mahmoud, Sadda, Dugel.
Drafting of manuscript: Hahn, Dugel.
Critical revision of the manuscript for important intellectual content: Hahn, Chung, Flynn, Huang, Kim, Mahmoud, Sadda, Dugel.
Statistical analysis: Hahn, Flynn, Huang, Kim, Mahmoud, Dugel.
Administrative, technical, or material support: Chung.
Study supervision: Sadda, Dugel.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Chung reported consulting for Glaxo Smith Kline and receiving grant funding from Canon, both unrelated to the submitted work. Dr Huang reported receiving consulting agreements unrelated to the submitted work with Bausch & Lomb, Sequenom, Second Sight, Notal Vision, Lumoptik Eye, and BioFirst. Dr Kim reported receiving clinical trial funding unrelated to the submitted work from Regeneron and Genentech and research support from Optos, serving on the advisory board for Allergan and ThromboGenics, and receiving funding from an unrestricted grant from the Research to Prevent Blindness Inc. Dr Mahmoud reported roles on advisory boards unrelated to the submitted work with Alimera, Allergan, and Alcon. Dr Sadda reported receiving consulting agreements unrelated to the submitted work with Allergan, Genentech, Carl Zeiss Meditec, and Optos; grants and pending grants from Carl Zeiss Meditec, Optovue, and Optos; and royalties from Topcon Medical Systems. Dr Dugel reported receiving consulting agreements with Regeneron, consulting agreements unrelated to the submitted work with Abbott, Alcon, Allergan, Arctic DX, Alimera Sciences, Acucela, Digisight, Genentech, LUX, Macusight, Neovista, ORA, Ophthotech, and ThromboGenics, and stock or stock options with Arctic DX, Macusight, Neovista, Ophthotech, and Digisight. No other disclosures were reported.
Additional Contributions: The American Society of Retina Specialists Therapeutic Surveillance Committee thanks the reporting physicians who shared their experiences for the benefit of the greater ophthalmology community.
Additional Information: The authors are members of the American Society of Retina Specialists Therapeutic Surveillance Committee.