Cataract Surgery Outcomes in Eyes With Primary Epiretinal Membrane

Key Points

Question  In the setting of cataract and primary epiretinal membrane (ERM), what is the magnitude of visual acuity improvement and incidence of cystoid macular edema after cataract surgery?

Findings  A clinical database study involving 812 ERM eyes and 159 184 reference eyes undergoing phacoemulsification cataract surgery showed an increase in visual acuity of 0.27 logMAR (approximately 3 Snellen lines) and an 8.6% incidence of cystoid macular edema in eyes with primary ERM.

Meaning  Cataract surgery in eyes with primary ERM, on average, can lead to an improvement in visual acuity but higher rates of cystoid macular edema and a lower postoperative visual acuity gain compared with cataract surgery in eyes without primary ERM.

Importance  Primary epiretinal membrane (ERM) is a common retinal disorder with a prevalence of 4% to 18.5%. Although ERM and cataracts commonly occur together, to our knowledge, no studies have investigated the outcome of cataract surgery alone in this setting.

Objective  To analyze the visual outcome and cystoid macular edema risk with cataract surgery in eyes with primary ERM.

Design, Setting, and Participants  In this retrospective clinical database study, data were collected from July 2003 to March 2015 from 8 locations in the United Kingdom. Cataract surgery data of 217 557 eyes were extracted from the electronic medical record of the UK National Health Service. After exclusion of 57 561 eyes with combined surgery, prior vitrectomy, copathology, and complications, 812 eyes with primary ERM and 159 184 reference eyes were analyzed.

Main Outcomes and Measures  We report on visual acuity (VA), the incidence of cystoid macular edema, and the need for ERM surgery.

Results  The mean (SD) age of patients in the ERM group was 73.7 (9.23) years, and 395 of 812 were men (46.8%). The mean (SD) age of patients in the reference group was 74.4 (12.19) years, and 65 265 of 159 184 were men (41%). Epiretinal membrane eyes assessed at 4 to 12 weeks postoperatively gained 0.27 (0.32) logMAR (approximately 3 Snellen lines), with 200 of 448 (44.6%) improving by 0.30 logMAR or more (≥3 Snellen lines) and 32 of 448 (7.1%) worsening by 0.30 logMAR or more. Reference eyes gained a mean (SD) of 0.44 (0.26) logMAR (approximately 4 Snellen lines), with 48 583 of 77 408 (62.8%) improving by 0.30 logMAR or more and 2125 of 77 408 (2.7%) worsening by 0.30 logMAR or more. Although all eyes with preoperative VA of 20/40 or less improved, only reference eyes with preoperative VA of more than 20/40 showed improvement. Cystoid macular edema developed in 57 of 663 ERM eyes (8.6%) (95% CI, 6.69-10.98) and 1731 of 125 435 reference eyes (1.38%) (95% CI, 1.32-1.45) (P < .001). Epiretinal membrane surgery was performed in 43 of 663 (6.5%) ERM eyes.

Conclusions and Relevance  On average, VA improved 0.27 logMAR (approximately 3 Snellen lines) in eyes with ERM. Eyes with ERM and VA of 20/40 or less showed more benefit after cataract surgery than those with better preoperative vision. However, compared with eyes without ERM, higher rates of cystoid macular edema and a lower postoperative VA gain were noted.

Primary epiretinal membrane (ERM) is a common disorder caused by the proliferation of glial tissue on the surface of the macula.^1,2 Although 1 autopsy-based study³ noted an incidence of 5.4%, its prevalence ranges from 4% to 18.5% in population-based studies.^2-9 Patients with ERM may experience vision loss, distortion, metamorphopsia, and micropsia.² However, up to 90% of patients may remain asymptomatic.^6,10 Age-related cataract is the leading cause of blindness worldwide for which surgical procedures are commonly performed.¹¹ Epiretinal membrane and cataract commonly occur together and the detection, grading, and assessment of the visual significance of each factor may be limited by the other.¹² Furthermore, there could be an underestimation of the prevalence and severity of ERM in patients with cataract.^2,10,12,13 Studies have demonstrated positive and equivalent benefit of combined (pars plana vitrectomy [PPV]/membrane peel [MP] plus cataract surgery) vs consecutive surgery (PPV/MP, then cataract surgery), with vision improvements from 0.2 to 0.3 logMAR (approximately 2-3 Snellen lines) postoperatively.^14-18 However, to our knowledge, no large studies have analyzed the visual outcome of cataract surgery alone in eyes with coexisting ERM.

In the United States, big data is advancing research and clinical practice in ophthalmology by allowing for large-scale examination of complex medical and surgical questions through the use of Medicare and the Intelligent Research in Sight registry.¹⁹ The UK National Health Service has provided a similar detailed cataract surgery data set developed by the UK Royal College of Ophthalmologists.²⁰ In this multicenter retrospective database study, we use electronic medical record data from the UK National Health Service to investigate the outcome of cataract surgery alone in the setting of ERM. We describe visual acuity (VA), the development of cystoid macular edema (CME), and the need for consecutive PPV/MP surgery in this cohort. We include a reference group undergoing cataract surgery in the absence of ERM for comparison.

Data sets from 8 UK National Health Service departments using the same electronic medical record system (Medisoft Ophthalmology; Medisoft Limited) were obtained, extracted, and pooled to a centralized database for analysis. A 2016 study by Chu et al²⁰ details standards of care at these clinic sites. A study period of 12 years (July 2003 to March 2015) was selected to provide a large cohort with sufficient follow-up information. Analyses began in April 2016. Fields extracted include sex, laterality, preoperative and postoperative VA, operative complications, diabetic status, presence of ERM, presence of CME, need for additional surgery, and time of follow-up. Race/ethnicity information was not extracted; however, based on the geographic location of the contributing centers, we conclude that the population studied is predominantly white. The lead clinician and Caldicott Guardian (who oversees data protection) at each center gave approval for the data extraction.²⁰ This study was conducted in accordance with the Declaration of Helsinki, the UK Data Protection Act, and National Institute for Health Research guidance on ethical approval. Patient consent was not required because this was a retrospective study using deidentified patient data.

Eyes were categorized on the basis of the presence or absence of ERM before cataract surgery. Eyes undergoing combined surgery and those with visually significant comorbidities or secondary ERMs, intraoperative complications (posterior capsular rupture with or without vitreous loss or dropped nuclear fragments), or missing preoperative data were excluded. Owing to the anonymized extraction of records, patients who underwent bilateral sequential surgical procedures during the study period had both eyes included, and data on individual eyes were treated as independent for analyses.

Our 2 main outcome variables were VA and the incidence of CME. Visual acuity was defined as the best value of uncorrected or corrected distance VA available at each time. Preoperative VA was that recorded closest to the date of cataract surgery, no more than 3 months prior. Follow-up was divided into 3 periods: 0 to 4 weeks, 4 to 12 weeks, and 12 to 24 weeks. As postoperative recovery is usually complete at 4 weeks, postoperative VA at 4 to 12 weeks was chosen as the primary visual outcome for this study (Table 1 and Table 2 include VA results at other follow-up intervals). In eyes that developed CME or worsening of ERM requiring vitrectomy surgery, VA was included as measured at a time, not related to the development or resolution of pathology. Visual acuity was measured in either Snellen fractions or logMAR. Snellen fractions were converted to logMAR for analysis during data extraction, with counting fingers, hand motions, light perception, and no light perception assessed as 2.10, 2.40, 2.70, and 3.00 logMAR, respectively, consistent with previous publications.^21-23

Postoperative CME was defined as a recorded diagnosis of CME within 90 days of the surgical procedure. For diabetic eyes, a newly recorded diagnosis of CME (or clinically significant macular edema) with a documented absence on the date of the preoperative examination was considered diagnostic. Although optical coherence tomography and/or fluorescein angiography were likely used for the diagnosis of CME, interpretations of these studies were not recorded consistently, preventing their analysis. Associated factors such as diabetes and prostaglandin analogue (PGA) use were analyzed. The record of perioperative nonsteroidal anti-inflammatory drug use was incomplete, limiting analysis.

Data were analyzed using multiple t test analyses with the Holm-Sidak method for comparing mean values, Fisher exact test for proportional differences, and multiple and logistic regression analyses. The time from cataract surgery to development of CME and PPV/MP were modeled using Kaplan-Meier survival curves, in which the failure events were the development of postoperative CME and PPV/MP surgery, respectively.

A total of 217 557 eyes underwent phacoemulsification cataract surgery with intraocular lens implantation. This included 2243 eyes with ERM (1%) and 215 314 without ERM (99%). The Figure shows the distribution of eyes and defines filtered copathology. Of the 2243 eyes with ERM, 1046 (46.8%) were excluded on the basis of combined surgery, 286 (12.8%) for prior PPV, 84 (3.8%) for significant copathology, and 15 (0.7%) for operative complications. This left 812 eyes with visually significant cataract and primary ERM that had undergone uncomplicated cataract surgery. Of the 215 314 eyes that underwent cataract surgery in the absence of ERM, 11 192 (5.2%) were excluded on the basis of combined surgery, 40 254 (18.7%) for significant copathology, 4566 (2.1%) for operative complications, and 118 (0.05%) for missing preoperative VA data. This left a reference group of 159 184 eyes having undergone uncomplicated cataract surgery in the absence of ERM.

Mean preoperative VA and demographic characteristics and are shown in Table 1 and Table 3. In the ERM group of 812 eyes, 404 (49.8%) were right eyes, and 408 (50.2%) were left eyes. The reference group of 159 184 eyes included 80 987 (50.9%) right eyes and 78 197 (49.1%) left eyes. The mean (SD) age of patients at the time of surgery was 73.7 (9.23) years (range, 23-96) in the ERM group and 74.4 (12.19) years (range, 19-99) in the reference group. There was no difference in the mean, nonsegregated, preoperative VA between the ERM and reference groups (mean [SD], 0.61 [0.44] logMAR [Snellen 20/80] vs 0.59 [0.49] logMAR [Snellen 20/80], respectively). However, percentages of diabetic eyes differed between groups as did the history of diabetic macular edema (Table 2).

Of the ERM group, 663 eyes (81.7%) were assessed during follow-up: 394 eyes (48.5%) at 0 to 4 weeks, 448 eyes (55.2%) at 4 to 12 weeks, and 273 eyes (33.6%) at 12 to 24 weeks. Of the reference group, 123 084 eyes (77.3%) were assessed during follow-up: 60 810 eyes (38.2%) at 0 to 4 weeks, 77 408 (48.6%) at 4 to 12 weeks, and 37 180 (23.4%) at 12 to 24 weeks. In the ERM group, preoperative VA did not differ between eyes assessed at follow-up and those lost to follow-up; however, eyes lost to follow-up had a lower incidence of diabetes. Within the ERM group, those lost to follow-up had a 22.15% (33 of 149) incidence of diabetes, and those who followed up had a 32.43% (215 of 663) incidence.

Table 1 and Table 2 detail changes in VA in the ERM and reference groups at all postoperative times. Among 448 eyes (55.2%) in the ERM group assessed at 4 to 12 weeks postoperatively, mean (SD) change in VA was a gain of 0.27 (0.32) logMAR (approximately 3 Snellen lines), with 200 (44.6%) improving by 0.30 logMAR or more (≥3 Snellen lines) and 32 (7.1%) worsening by 0.30 logMAR or more. Among 77 408 (48.6%) reference eyes assessed at 4 to 12 weeks postoperatively, mean (SD) change in VA was a gain of 0.44 (0.26) logMAR (approximately 4 Snellen lines), with 48 583 eyes (62.8%) improving by 0.30 logMAR or more and 2125 eyes (2.7%) worsening by 0.30 logMAR or more.

To control for unmatched differences in preoperative characteristics, multiple and logistic regression analyses were conducted. In both models, age, preoperative VA, presence of diabetes, presence of proliferative diabetic retinopathy, and history of diabetic macular edema were chosen as the predictor variables, in addition to the presence of ERM. The criterion variable (tested outcome) was mean postoperative VA at 4 to 12 weeks for the multiple regression analysis and the likelihood of improvement of VA by 0.3 logMAR or more for the logistic regression analysis. Both models had a P value less than .001, and all predictable variables except for diabetic status had a P value of less than .05. In the logistic regression analysis, the presence of ERM was associated with a decrease in the odds of achieving improvement of VA of 0.30 logMAR or more by a factor of 0.41 (95% CI, 0.238-0.692, P < .001) (Table 4).

To evaluate the benefits of cataract surgery as stratified by preoperative VA (better than 20/40 vs 20/40 or worse), a subgroup analysis was performed. Among 63 ERM eyes and 14 744 reference eyes with preoperative vision of more than 20/40 (<0.30 logMAR) postoperatively, vision worsened by 0.04 logMAR (<1 Snellen line) (95% CI, –0.018 to 0.098) in the ERM group and improved by 0.06 logMAR (<1 Snellen line) (95% CI, –0.063 to –0.057) in the reference group. Among 385 ERM eyes and 62 415 reference eyes with preoperative vision of 20/40 or less (≥0.30 logMAR), vision improved by 0.32 logMAR (approximately 3 Snellen lines) (95% CI, –0.355 to –0.286) in the ERM group and by 0.57 logMAR (approximately 6 Snellen lines) (95% CI, –0.572 to –0.568) in the reference group.

Cystoid macular edema developed in 57 eyes (8.6%) (95% CI, 6.69-10.98) in the ERM group and 1731 eyes (1.38%) (95% CI, 1.32-1.45) in the reference group (P < .001). The mean time from cataract surgery to the initial clinical detection of CME was 38.2 days (median, 29; range, 8-84) in the ERM group and 39.8 days (median, 34; range, 15.7-63.9) in the reference group.

We analyzed the incidence of CME in individuals with diabetes and PGA users (Table 5 and eFigure 1 in the Supplement). A higher proportion of diabetic eyes in the ERM group (20 [11.3%]) (95% CI, 7.43-16.81) developed CME than in the reference group (758 [3.3%]) (95% CI, 3.08-3.54) (P < .001). On stratification by retinopathy grade, there were increases in the incidence of CME with increasing grades of diabetic retinopathy, but a significant difference between groups was not detected. A higher proportion of ERM eyes with PGA use developed CME compared with reference eyes with PGA use (10.4% vs 1.7%, respectively). Excluding eyes with history of diabetes or PGA use, 30 eyes (7.14%) in the ERM group and 829 eyes (0.89%) in the reference group developed CME.

Forty-three eyes (6.5%) in the ERM group underwent PPV/MP surgery following cataract surgery (eFigure 2 in the Supplement). Median time from cataract surgery to PPV/MP was 28 weeks (mean, 85.9; range, 22-262). Of those undergoing PPV/MP, 11 eyes (25.6%) had 0.30 logMAR or more (approximately 3 Snellen lines) of VA loss, and 6 (13.9%) exhibited postoperative CME. Vision loss was a significant predictor for the need for PPV/MP with 6 of 35 (17.1%) (95% CI, 6.56-33.65) of those who lost vision (≥0.30 logMAR) requiring surgery vs 37 of 627 eyes (5.9%) (95% CI, 4.19-8.04) where vision did not worsen by 0.30 logMAR or more (P = .02).

This study examined the outcome of cataract surgery alone in patients with coexisting cataract and primary ERM. It was based on electronic medical record data of 217 577 eyes that underwent cataract surgery at 8 sites in the United Kingdom. In this cohort, we found that cataract surgery in eyes with ERM was associated with an improvement in VA of 0.27 logMAR (approximately 3 Snellen lines), substantial VA gain (≥0.30 logMAR) in 44.6%, and a substantial loss in 7.1%. Our data also suggest that compared with eyes without ERM, there are increased rates of CME and a decreased chance for substantial visual gain.

Although direct comparison with other studies is difficult owing to variations in reporting, the levels of improvement we observed after cataract surgery in the setting of ERM are comparable with data from studies of combined and consecutive PPV/MP and cataract surgery. Such studies have shown improvement of 0.20 to 0.34 logMAR (approximately 2-3 Snellen lines).^16,17,24 To evaluate the benefits of cataract surgery in eyes with good preoperative vision, we conducted a subgroup analysis of VA gain stratified by preoperative vision category. Eyes with ERM and mean preoperative VA better than 20/40 demonstrated no mean improvement in VA, while eyes with ERM and mean preoperative VA of 20/40 or worse exhibited a substantial improvement in VA of 0.32 logMAR (approximately 3 Snellen lines). Although a possible ceiling effect may have limited the potential for improvement in both the ERM and reference groups for eyes with good vision at baseline, these results suggest that eyes with ERM and VA of 20/40 or worse (≥0.30 logMAR) may be more likely to benefit from cataract surgery with improvement in VA.

Although 1 study did not show worsening of ERM after cataract surgery,²⁵ others have described progression in as many as 45% of cataract surgery patients over a 3-year period.^13,26,27 In our study, 7.1% of eyes with ERM (vs 2.7% of reference eyes) assessed at 4 to 12 weeks postoperatively demonstrated VA worsening by 0.30 logMAR or more. Nine of these 32 eyes (28.1%) had developed postoperative CME. Additionally, approximately 7% required consecutive PPV/MP surgery. Taken together, these findings indicate possible worsening of ERM in these eyes; however, other factors, including uncorrected refractive error or unresolved CME, could have contributed to this decrease in vision.

Cystoid macular edema is a known complication of cataract surgery, which leads to more postoperative visits and may affect final best-corrected VA.²⁸ The reported incidence of CME after uncomplicated cataract surgery in a normal eye is 1% to 2%.²⁹ A study by Henderson et al²⁸ demonstrated a 7% incidence of postoperative CME in eyes with ERM undergoing cataract surgery. In our study, the incidence of CME was 8.6% in eyes with ERM. Diabetic retinopathy is a risk factor for CME after cataract surgery,^20,28 with its risk being higher with increasing grades of retinopathy.²⁰ Our data indicate that the presence of ERM further increases the risk of postoperative CME with diabetes by more than 3-fold (from 3.3% to 11.3%). Although the incidence of CME was higher with increasing grades of diabetic retinopathy and ERM as compared with those without ERM, numbers of eyes in these subgroups were small, and these differences were not statistically significant. Mixed results exist on the association between PGA use and postoperative CME.^20,28,30,31 We found the risk of CME in eyes without ERM to be 1.79% with PGA use. However, the presence of coexisting ERM significantly increased this risk by more than 5-fold (10.4%).

Our study is limited by its nonrandomized, retrospective design; therefore, the improvement we observed in VA may be subject to selection bias as eyes with more severe ERMs may have been selected for combined cataract and PPV/MP surgery from the outset. This may also explain the small number of consecutive PPV/MP surgeries recorded in our study. Short follow-up time may have also influenced the detected worsening of ERM and the need for PPV/MP in the long term. The lack of raw optical coherence tomography/fluorescein angiography data makes it impossible to grade the severity or worsening of ERM, differentiate ERM-related macular thickening from CME, and elucidate swelling of the optic nerve. High loss to follow-up and unaccounted differences in ERM severity, diabetes prevalence, and history of diabetic macular edema may also have influenced the visual outcomes. As such, severity of ERM should remain a major consideration in determining whether primary cataract surgery would be appropriate for a given patient.^10,12 Qualitative changes in vision, including distortion, metamorphopsia, and micropsia, may also help to distinguish the influence of ERM from cataract and to determine which surgical approach would be most beneficial.²

As cataract surgery is commonly performed and ERM may occur together in a substantial proportion of patients,¹² our findings may well influence clinical practice. Patients older than 65 years often have greater access to cataract surgery than retina surgery, and access to a retina specialist may be limited to some individuals in certain geographic locations. Our findings may help the cataract surgeon determine whether a patient may benefit from cataract surgery primarily or if they should be referred to a retina specialist. They may also aid in the discussion of the risk for CME in this setting and expectations regarding visual outcome. Other strengths of our study include its use of “big data” from multiple sites, with analysis of prospectively collected structured data sets for more than 217 000 eyes, more than 800 having undergone cataract surgery in the setting of ERM. As such, our findings are more generalizable and avoid the potential biases inherent to small retrospective studies or studies at single institutions.

Examining the effect of cataract surgery on vision in the setting of ERM, we found a mean change in VA of 0.27 logMAR, a substantial visual gain in 44.6% of eyes, and a substantial loss in 7.1% of eyes. Our data indicate that individuals with ERM and VA worse than 20/40 may benefit more from cataract surgery than those with better preoperative VA. However, our data suggest that compared with eyes without ERM, the rates of CME increase from 5.24% to 9.66% with a 13.2% to 23% decreased chance for visual gain.

Corresponding Author: Ahmed B. Sallam, MD, PhD, FRCOphth, Jones Eye Institute, University of Arkansas for Medical Sciences, 4301 W Markham St, #523, Little Rock, AR 72207 (ahmedsallam11@yahoo.com).

Accepted for Publication: November 2, 2017.

Published Online: December 21, 2017. doi:10.1001/jamaophthalmol.2017.5849

Author Contributions: Drs Sallam and Gauldin had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Hardin, Soliman, Chu, Yang, Sallam.

Acquisition, analysis, or interpretation of data: Hardin, Gauldin, Soliman, Sallam.

Drafting of the manuscript: Hardin, Soliman, Sallam.

Critical revision of the manuscript for important intellectual content: Hardin, Gauldin, Soliman, Chu, Yang.

Statistical analysis: Gauldin, Sallam.

Administrative, technical, or material support: Hardin, Chu.

Study supervision: Yang, Sallam.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Meeting Presentation: This paper was presented at the American Academy of Ophthalmology Annual Meeting; October 16, 2016; Chicago, Illinois.

Additional Contributions: We thank Clare Bailey, MD, University of Bristol and Bristol Eye Hospital; Arijit Mitra, MD, Sandwell and West Birmingham Hospital National Health Service Trust; Atul Varma, MD, Midyorkshire Hospital National Health Service Trust; Martin Mckibbin, MD, Leeds Teaching Hospitals; Muhammed Irfan Tahir, MD, King Edward VII Hospital and Royal Berkshire National Health Service Foundation Trust; Nick Lee, MD, Hillingdon Hospital and Western Eye Hospital; and Peter Scanlon, MD, Gloucestershire Hospital National Health Service Trust; for data collection. This work is dedicated to the memory of Robert L. Johnston, MD, FRCOphth (1966-2016), who played a major role in designing this study and sadly passed away before the completion of the work. No compensation was received for their contributions.