Do repeated intravitreous bevacizumab injections increase the risk for glaucoma surgery?
In this case-control study of 74 patients who underwent glaucoma surgery and 740 control participants, the adjusted rate ratio for glaucoma surgery was higher for patients who received 7 or more intravitreous bevacizumab injections per year than for those who received 3 or fewer intravitreous bevacizumab injections per year.
Patients who receive 7 or more intravitreous bevacizumab injections per year have an increased risk of requiring glaucoma surgery.
Intravitreous injections of anti–vascular endothelial growth factor (VEGF) agents are associated with a sustained increase in intraocular pressure. This sustained elevated intraocular pressure could lead to higher rates of glaucoma surgery to lower this pressure.
To determine the risk of glaucoma surgery following repeated intravitreous bevacizumab injections.
Design, Setting, Participants
This nested, case-control study acquired and analyzed data from large, population-based, linked health databases supported by the British Columbia Ministry of Health in Canada. Study participants included all patients with ophthalmic issues in British Columbia, such as those of the Provincial Retinal Diseases Treatment Program, who had received intravitreous bevacizumab injections for exudative age-related macular degeneration between January 1, 2009, and December 31, 2013. Cases were identified using glaucoma surgical codes for trabeculectomy, complicated trabeculectomy, glaucoma drainage device, and cycloablative procedure. For each case, 10 controls were identified and matched for age, preexisting glaucoma, calendar time, and follow-up time. The number of intravitreous bevacizumab injections received per year—3 or fewer, 4 to 6, or 7 or more—was determined for both cases and controls. Data analysis was performed from February 23, 2016, to November 14, 2016.
Main Outcomes and Measures
Risk of glaucoma surgery compared with the number of intravitreous bevacizumab injections per year in cases and controls. Rate ratios were adjusted for covariates (diabetes mellitus, myocardial infarction, stroke, and verteporfin use).
Seventy-four cases of glaucoma surgery and 740 controls were identified, with a mean (SD) age of 81.3 (8.4) years for cases and 81.4 (7.9) for controls. The case group had more males than the control group (38 [51.4%] vs 272 [36.8%]). The adjusted rate ratio of glaucoma surgery among those who received 7 or more injections per year was 2.48 (95% CI, 1.25-4.93). There was a 10.3% higher number of 7 or more injections among cases compared with controls. The adjusted rate ratio for those who received 4 to 6 injections per year compared with those who received 3 or fewer was 1.65% (95% CI, 0.84-3.23).
Conclusions and Relevance
Findings from this large, pharmacoepidemiologic study suggest that 7 or more intravitreous injections of bevacizumab annually is associated with a higher risk of glaucoma surgery and that 4 to 6 injections per year show a nonstatistically significant rate ratio in the same direction.
The advent of intravitreous anti–vascular endothelial growth factor (VEGF) injections to treat common causes of vision loss, such as exudative age-related macular degeneration (AMD),1-3 diabetic macular edema,4-6 and edema associated with retinal vein occlusion,7-10 has undoubtedly improved visual outcomes for many patients. Major prospective clinical trials investigating variable dosing regimens of intravitreous anti-VEGF injections for exudative AMD have revealed that some patients may require monthly injections during the first 1 to 2 years of treatment and that, on average, 10 injections are required in the first 2 years of treatment.11,12 Minimal adverse events were reported in those trials and some conflicting results exist, but subsequent studies suggest that repeated intravitreous anti-VEGF injections can lead to a sustained increase in intraocular pressure (IOP).13-22
In 2007, Jalil and colleagues13 reported a case of ocular hypertension with a marked rise in IOP 3 days after the patient’s fourth monthly intravitreous injection of bevacizumab for exudative AMD. In 2008, Bakri and colleagues14 reported a case series of 4 nonglaucomatous patients receiving intravitreous ranibizumab who developed a sustained increase in IOP. Similar studies have been published since those initial case series, including research suggesting that intravitreous bevacizumab injections cause sustained increases in IOP.15,16 Retrospective chart reviews support this association, reporting that approximately 6% to 9% of patients receiving intravitreous anti-VEGF injections experienced sustained IOP elevation.17-19 A similar value was observed in an exploratory analysis of the Diabetic Retinopathy Clinical Research Network randomized clinical trial.20 Many of those studies suggest that the risk of sustained IOP rise is positively associated with the number of injections that the patient has received.16,17,21 To date, there has not been a large epidemiologic study that has investigated the risk of a clear glaucoma outcome (ie, less prone to bias using a retrospective health database analysis), such as glaucoma surgery, following repeated intravitreous anti-VEGF injections.
Intravitreous injections of anti-VEGF agents may increase the risk of increased IOP13-22; however, the risk of developing moderate to advanced glaucoma requiring glaucoma surgery has been unclear. A single, small case series suggests that at least some patients with or without glaucoma at baseline may go on to require a trabeculectomy.22 Therefore, using large, population-based linked health databases, we conducted a pharmacoepidemiologic, nested, case-control study to examine the possible association of repeated intravitreous bevacizumab injections with subsequent glaucoma surgery.
Setting and Patient Population
This population-based, observational study used the linked databases supported by the British Columbia Ministry of Health. In British Columbia, Canada, patients with retinal diseases are entitled to receive anti-VEGF medication through the Provincial Retinal Diseases Treatment Program. The program records clinical ophthalmic data for all patients, including intravitreous anti-VEGF injections, in a comprehensive database that is subject to frequent quality control and is linkable to other Ministry of Health–supported databases. These databases have been used in many previous epidemiologic studies23-25 and include the Discharge Abstract Database,26 which captures all hospitalizations; the Medical Services Plan database,27 which contains all physician visits; and the Medical Services Plan prescription drugs database, which records all medications dispensed to patients.28 Data analysis was performed from February 23, 2016, to November 14, 2016. Ethics approval for this study was granted before commencement by the University of British Columbia’s Clinical Research Ethics Board, and the study design and execution adhered to the tenets of the Declaration of Helsinki.29 Patient informed consent was waived by the University of British Columbia’s Clinical Research Ethics Board because individual patients are anonymized in the databases.
Cohort and Case-Control Analysis
The cohort comprised patients who received their first intravitreous bevacizumab injection for exudative AMD between January 1, 2009, and December 31, 2013. Cases were defined as patients in the cohort who subsequently underwent glaucoma surgery. The date of glaucoma surgery was the index date. For each case, 10 controls (no surgical procedure code at any time of follow-up) were selected and matched by age, preexisting glaucoma, number of injections per year, calendar time, and follow-up time (matching protocol below). Glaucoma surgery cases were identified using these surgical procedural codes, which are part of the payment schedule created by the Medical Services Commission to reimburse services clinicians provide to beneficiaries of the Medical Services Plan: trabeculectomy (02187), complicated trabeculectomy (22187), glaucoma drainage device (22070), and cycloablative procedure (22185).
Calendar time and follow-up time matching was achieved by selecting controls who received their first bevacizumab injection in the same year and had the same duration of follow-up as the cases between injection and date of glaucoma surgery (index date). Calendar time matching decreases the likelihood that trends for a particular procedure that differ by calendar year (eg, introduction of new guidelines for a specific procedure) will affect cases and controls disproportionately. Matching by follow-up time balances between cases and controls the opportunity to undergo glaucoma surgery, controlling for time window bias.25 Cases and controls were matched for preexisting glaucoma to control for the increased probability of glaucoma surgery in patients with preexisting glaucoma.
The exposure investigated was the frequency of intravitreous bevacizumab injections in cases and controls. Thresholds of the number of injections received per year were 3 or fewer, 4 to 6, or 7 or more; these cutoffs were chosen to ensure a robust sample size in the reference category. Data were analyzed in the same manner but with the total number of intravitreous bevacizumab injections as the exposure and the thresholds as 6 or fewer, 7 to 11, or 12 or more (eTable in the Supplement).
Intravitreous Anti-VEGF Injections
Bevacizumab is the anti-VEGF medication provided by the Provincial Retinal Diseases Treatment Program to most patients in British Columbia with exudative AMD. In this study, bevacizumab was also the anti-VEGF medication used for all patients in the cohort. A licensed pharmacist at a compounding pharmacy, approved by the College of Pharmacists of British Columbia and the Provincial Retinal Diseases Treatment Program, compounds and dispenses bevacizumab. Commercially available vials of the anti-VEGF medication, which is supplied by the Ministry of Health, are used to fill ultrafine diabetic syringes (BD ultrafine II ½ CC syringes; BD Worldwide) containing 0.05 mL of the drug. The syringes are then transported and stored at 4°C until used by the vitreoretinal specialist.
The demographics of cases and controls were examined using descriptive statistics. For the case-control analysis, patients who received 3 or fewer injections per year (or ≤6 total injections in the supplementary analysis) served as the reference category. Controls were allowed to become future cases (ie, experience a future surgical procedure), allowing the odds ratio to closely approximate a rate ratio (RR) derived from a cohort study.30 Crude and adjusted RRs were approximated from a conditional logistic regression model. The RRs were adjusted for the following covariates: diabetes, myocardial infarction, stroke, and verteporfin use (Visudyne; Novartis and QLT PhotoTherapeutics). To avoid indiscriminant adjustment of covariates that may lead to overadjustment bias, causal diagrams of potential confounding variables were used to identify true confounding variables (ie, variables that appear to be associated with exposure and outcome) and adjustment of covariates at baseline.31,32
We identified the cases of 75 patients who had received intravitreous bevacizumab injections for exudative AMD. One patient also had a central retinal vein occlusion and was excluded from the study, leaving 74 cases for analysis. We identified 740 corresponding controls who were matched for age and preexisting glaucoma. The mean (SD) age was 81.3 (8.4) years for cases and was 81.4 (7.9) years for controls (Table 1). There were more males in the case group (38 [51.4%]) than in the control group (272 [36.8%]). The frequency of the potential confounding variables (ie, diabetes, myocardial infarction, stroke, and verteporfin use) was not different between the 2 groups.
The percentage of cases identified who received 7 or more intravitreous bevacizumab injections per year (44.6% [n = 33]) was greater than the percentage of controls who received 7 or more intravitreous bevacizumab injections per year (34.3% [n = 254]) (absolute increase, 10.3%; Table 2). Compared with the reference group (ie, those who received ≤3 intravitreous bevacizumab injections per year), an increased RR was observed for cases vs controls who received 7 or more intravitreous bevacizumab injections per year (adjusted RR, 2.48; 95% CI, 1.25-4.93). Compared with the reference group, there was no difference in the RR for cases vs controls who received 4 to 6 intravitreous bevacizumab injections per year (adjusted RR, 1.65; 95% CI, 0.84-3.23). These data show that receipt of 7 or more, but not 4 to 6, intravitreous bevacizumab injections per year is associated with an increased risk of subsequent glaucoma surgery. Similar results were obtained when the total number of injections were used as the exposure group (eTable in the Supplement).
Several case reports, retrospective studies, and post hoc exploratory analyses of prospective clinical trials13-22 have associated intravitreous injections of anti-VEGF agents with a sustained increase in IOP. Some of those studies suggest that the elevation in IOP becomes apparent only after repeated injections of anti-VEGF agents.16,17,19,21 In our study, we used large linked databases supported by the British Columbia Ministry of Health to examine the possible association of repeated intravitreous bevacizumab injections with glaucoma surgery. We observed an adjusted RR of 2.48 (95% CI, 1.25-4.93) for glaucoma surgery in patients who received 7 or more intravitreous bevacizumab injections per year. These data suggest that repeated intravitreous anti-VEGF injections could lead to a rise in IOP-lowering surgical procedures for glaucoma and a sustained increase in IOP.
The mechanism by which repeated intravitreous anti-VEGF injections may lead to sustained IOP elevation is not clear. Many hypotheses have been offered, including trabecular meshwork injury from rapid injections of high volumes of fluid,33 altered nitric oxide metabolism in the trabecular meshwork,34 a potentially toxic or inflammatory reaction after exposure to the biologic agent and/or vehicle, and mechanical blockade of the trabecular meshwork by protein aggregates or contaminant particles associated with packaging and injection techniques.35 Although our data do not help clarify the pathogenesis of this observation, they do suggest that the mechanism can be substantial enough to cause glaucoma that requires surgical intervention.
An important finding in this study is the increased risk of glaucoma surgery associated with 7 or more intravitreous injections of bevacizumab per year. This finding corroborates those of many studies that use sustained increased IOP as the outcome measure after repeated intravitreous anti-VEGF injections.16,17,19,21 Clinical trials investigating the efficacy of anti-VEGF in treating exudative AMD, such as MARINA (Minimally Classic/Occult Trial of the Anti-VEGF Antibody Ranibizumab in the Treatment of Neovascular Age-Related Macular Degeneration) and ANCHOR (Anti-VEGF Antibody for the Treatment of Predominantly Classic Choroidal Neovascularization of Age-Related Macular Degeneration), did not report a significant long-term effect of ranibizumab on IOP. However, a post hoc analysis of MARINA and ANCHOR did demonstrate that eyes treated with ranibizumab were more likely to have an IOP increase greater than 6 mm Hg and a preinjection IOP greater than 24 mm Hg on at least 2 consecutive visits.1-3 This post hoc analysis did not reveal statistically significant differences in glaucoma-related end points, such as use of IOP-lowering medications, laser trabeculoplasty, peripheral iridotomy, or trabeculectomy. Many differences exist between our study and the MARINA and ANCHOR trials, including the study design, use of ranibizumab in MARINA and ANCHOR, shorter duration of follow-up in MARINA and ANCHOR, and control group receiving sham injection, photodynamic therapy, and possibly steroid injections in MARINA and ANCHOR. Similarly, a separate ad hoc, exploratory analysis conducted by the Diabetic Retinopathy Clinical Research Network reported sustained IOP elevation with repeated ranibizumab injections for diabetic macular edema, suggesting that this effect is likely not limited to patients with exudative AMD.20
The present study was restricted to use of the VEGF inhibitor bevacizumab because this medication is supported by the Provincial Retinal Diseases Treatment Program and is widely used by most of its patients. Previous studies suggest that the putative effect of intravitreous anti-VEGF injections on IOP, and possibly on glaucoma, may not be produced by bevacizumab alone. Case studies have suggested that the effect of intravitreous anti-VEGF medications on sustained increased IOP is the result of both bevacizumab and ranibizumab.13-16,19 In a retrospective review by Choi and colleagues,18 sustained IOP elevation occurred after intravitreous injections of bevacizumab, ranibizumab, or pegaptanib sodium. In the case series by Skalicky and colleagues,22 all patients who went on to require a trabeculectomy had received ranibizumab. Larger studies are required to clarify any potential differences in effects of repeated intravitreous injections of different anti-VEGF agents on the need for glaucoma surgery.
Strengths and Limitations
Our study has many important strengths. First, its large sample size allowed for appropriate quantification of the relationship between the frequency (and total number; see eTable in the Supplement) of repeated intravitreous bevacizumab injections and surgical intervention for glaucoma. The use of glaucoma surgery as a hard outcome is a novel approach to addressing the potential association between glaucoma and intravitreous bevacizumab injections; previous large studies were limited to associations between repeated intravitreous anti-VEGF injections and sustained increases in IOP. Second, the study design involved the matching of preexisting glaucoma between cases and controls to decrease the possibility of preexisting glaucoma confounding our results, such as an increased baseline likelihood for glaucoma surgery between groups. Third, restricting the study to only patients who received intravitreal bevacizumab injections for the indication of exudative AMD was important because central retinal vein occlusion, a separate condition associated with glaucoma, may also be treated by intravitreous bevacizumab injections. Finally, the study demonstrates that a greater frequency and a greater total number of bevacizumab injections (eTable in the Supplement) are both associated with an increased risk for glaucoma surgery; this supports the robustness of our results.
One limitation of this study is that, although the databases we used are well linked and subject to rigorous quality control, they did not allow us to access details of individual disease severity and trajectory, including data such as changes in IOP and visual field loss. Similarly, making associations with prescription of glaucoma medications is difficult when prescribing practices are not reliably linked with bona fide glaucoma diagnoses. Second, this study was unable to address the issue of compounded vs prepared anti-VEGF medication because all anti-VEGF medication was compounded bevacizumab. Previous studies have shown the differences in protein concentration, aggregation, and particulate matter between compounded and prepared anti-VEGF medication.37,38 Finally, it is important to appreciate that, in general, causation cannot be conclusively established in case-control studies.
The results of this study are consistent with those of previous studies suggesting that repeated intravitreous injections of anti-VEGF medications can lead to a sustained increase in IOP. Our study expands on those studies by showing that this sustained IOP elevation may lead to a greater need for surgical intervention for glaucoma. This risk appears to be evident for patients who have received 7 or more intravitreous bevacizumab injections per year. Clinicians should be aware of the potential association of repeated, recent intravitreous anti-VEGF injections for diseases, such as exudative AMD, with subsequent need for glaucoma surgery.
Corresponding Author: Mahyar Etminan, PharmD, MSc, Department of Ophthalmology and Visual Sciences, Faculty of Medicine, University of British Columbia, Room 323-2550 Willow St, Vancouver, BC V5Z 3N9, Canada (firstname.lastname@example.org).
Accepted for Publication: December 12, 2016.
Published Online: March 16, 2017. doi:10.1001/jamaophthalmol.2017.0059
Author Contributions: Dr Etminan 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: Eadie, Carleton, Maberley, Mikelberg.
Acquisition, analysis, or interpretation of data: Eadie, Etminan, Carleton.
Drafting of the manuscript: Eadie, Maberley, Mikelberg.
Critical revision of the manuscript for important intellectual content: Eadie, Etminan, Carleton, Mikelberg.
Administrative, technical, or material support: Eadie, Maberley.
Study supervision: Etminan, Maberley, Mikelberg.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Etminan was a paid consultant in Mirena Propecia litigations. No other disclosures were reported.
et al; MARINA Study Group. Ranibizumab for neovascular age-related macular degeneration. N Engl J Med
. 2006;355(14):1419-1431.PubMedGoogle ScholarCrossref
et al; ANCHOR Study Group. Ranibizumab versus verteporfin for neovascular age-related macular degeneration. N Engl J Med
. 2006;355(14):1432-1444.PubMedGoogle ScholarCrossref
et al; VIEW1 and VIEW2 Study Groups. Intravitreal aflibercept (VEGF trap-eye) in wet age-related macular degeneration. Ophthalmology
. 2013;120(1):209-210.PubMedGoogle ScholarCrossref
et al; READ-2 Study Group. Primary end point (six months) results of the Ranibizumab for Edema of the Macula in Diabetes (READ-2) study. Ophthalmology
. 2009;116(11):2175-2181.e1.PubMedGoogle ScholarCrossref
et al; RISE and RIDE Research Group. Ranibizumab for diabetic macular edema: results from 2 phase III randomized trials: RISE and RIDE. Ophthalmology
. 2012;119(4):789-801.PubMedGoogle ScholarCrossref
et al; Diabetic Retinopathy Clinical Research Network. Aflibercept, bevacizumab, or ranibizumab for diabetic macular edema. N Engl J Med
. 2015;372(13):1193-1203.PubMedGoogle ScholarCrossref
et al; BRAVO Investigators. Ranibizumab for macular edema following branch retinal vein occlusion: six-month primary end point results of a phase III study. Ophthalmology
. 2010;117(6):1102-1112.e1.PubMedGoogle ScholarCrossref
et al; CRUISE Investigators. Ranibizumab for macula edema following central retinal vein occlusion: six-month primary end point results in a phase III study. Ophthalmology
. 2010;117(6):1124-1133.PubMedGoogle ScholarCrossref
et al. Intravitreal aflibercept for macular edema following branch retinal vein occlusion: 52-week results of the VIBRANT Study. Ophthalmology
. 2016;123(2):330-336.PubMedGoogle ScholarCrossref
et al. Vascular endothelial growth factor trap-eye for macular edema secondary to central retinal vein occlusion: six-month results of the phase 3 COPERNICUS study. Ophthalmology
. 2012;119(5):1024-1032.PubMedGoogle ScholarCrossref
et al. An optical coherence tomography-guided, variable dosing regimen with intravitreal ranibizumab (Lucentis) for neovascular age-related macular degeneration. Am J Ophthalmol
. 2007;143(4):566-583.PubMedGoogle ScholarCrossref
et al. A variable-dosing regimen with intravitreal ranibizumab for neovascular age-related macular degeneration: year 2 of the PrONTO Study. Am J Ophthalmol
. 2009;148(1):43-58.e1.PubMedGoogle ScholarCrossref
S. Intravitreal bevacizumab (Avastin) causing acute glaucoma: an unreported complication. Eye (Lond)
. 2007;21(12):1541.PubMedGoogle ScholarCrossref
DM. Persisent ocular hypertension following intravitreal ranibizumab. Graefes Arch Clin Exp Ophthalmol
. 2008;246(7):955-958.PubMedGoogle ScholarCrossref
N. Sustained elevation in intraocular pressure associated with intravitreal bevacizumab injections. Ophthalmic Surg Lasers Imaging
. 2009;40(3):293-295.PubMedGoogle ScholarCrossref
HR. Persistent ocular hypertension following intravitreal bevacizumab and ranibizumab injections. J Ocul Pharmacol Ther
. 2010;26(1):105-110.PubMedGoogle ScholarCrossref
MY. Sustained elevation of intraocular pressure after intravitreal injections of anti-VEGF agents. Br J Ophthalmol
. 2011;95(8):1111-1114.PubMedGoogle ScholarCrossref
et al. Sustained elevated intraocular pressures after intravitreal injection of bevacizumab, ranibizumab, and pegaptanib. Retina
. 2011;31(6):1028-1035.PubMedGoogle ScholarCrossref
et al. Clinical predictors of sustained intraocular pressure elevation due to intravitreal anti-vascular endothelial growth factor therapy. Retina
. 2013;33(1):179-187.PubMedGoogle ScholarCrossref
et al; Diabetic Retinopathy Clinical Research Network Investigators. Repeated intravitreous ranibizumab injections for diabetic macular edema and the risk of sustained elevation of intraocular pressure or the need for ocular hypotensive treatment. JAMA Ophthalmol
. 2015;133(5):589-597.PubMedGoogle ScholarCrossref
W. Long-term intraocular pressure changes after intravitreal injection of bevacizumab. Cutan Ocul Toxicol
. 2016;35(4):310-314.PubMedGoogle ScholarCrossref
A. Glaucoma filtration surgery following sustained elevation of intraocular pressure secondary to intravitreal anti-VEGF injections. Ophthalmic Surg Lasers Imaging
. 2012;43(4):328-334.PubMedGoogle ScholarCrossref
D. Oral fluoroquinolones and the risk of retinal detachment. JAMA
. 2012;307(13):1414-1419.PubMedGoogle ScholarCrossref
BC. Prescription drug dispensing profiles for one million children: a population-based analysis. Eur J Clin Pharmacol
. 2013;69(3):581-588.PubMedGoogle ScholarCrossref
BC. Risk of myocardial infarction and stroke with single or repeated doses of intravitreal bevacizumab in age-related macular degeneration. Am J Ophthalmol
. 2016;163:53-58.PubMedGoogle ScholarCrossref
L. Metformin and the risk of cancer: time-related biases in observational studies. Diabetes Care
. 2012;35(12):2665-2673.PubMedGoogle ScholarCrossref
D. Sampling strategies in nested case-control studies. Environ Health Perspect
. 1994;102(suppl 8):47-51.PubMedGoogle ScholarCrossref
RW. Overadjustment bias and unnecessary adjustment in epidemiologic studies. Epidemiology
. 2009;20(4):488-495.PubMedGoogle ScholarCrossref
KB. Predictors of sustained intraocular pressure elevation in eyes receiving intravitreal anti-vascular endothelial growth factor therapy. Am J Ophthalmol
. 2014;158(2):319-327.e2.PubMedGoogle ScholarCrossref
BM. High intraocular pressure following anti-vascular endothelial growth factor therapy: proposed pathophysiology due to altered nitric oxide metabolism. J Ocul Pharmacol Ther
. 2015;31(1):2-10.PubMedGoogle ScholarCrossref
MY. The link between intravitreal antivascular endothelial growth factor injections and glaucoma. Curr Opin Ophthalmol
. 2014;25(2):127-133.PubMedGoogle ScholarCrossref
et al. Eyes treated with monthly ranibizumab: a post hoc analysis of data from the MARINA and ANCHOR trials. Paper presented at: AAO Annual Meeting; October 15, 2010; Chicago, IL.
et al. Evaluation of compounded bevacizumab prepared for intravitreal injection. JAMA Ophthalmol
. 2015;133(1):32-39.PubMedGoogle ScholarCrossref
A. Acute intraocular inflammation following intravitreal injection of bevacizumab—a large cluster of cases. Acta Ophthalmol
. 2011;89(8):e664-e665.PubMedGoogle ScholarCrossref