Purpose
To study the prognostic factors that influence best-corrected visual acuity (BCVA) outcome in patients with secondary epiretinal membrane (ERM) after retinal detachment surgery.
Design
Retrospective case series.
Methods
Forty-two patients with ERM were divided into macula-on and macula-off groups based on the macular status before retinal detachment surgery and were studied using the same spectral-domain optical coherence tomography device. Several variables, including the integrity of the external limiting membrane (ELM), the status of the photoreceptor inner segment/outer segment (IS/OS) junction line, and central foveal thickness were evaluated in 17 treated and 25 untreated patients. Linear regression analysis was used to determine the best combination of all variables affecting BCVA.
Results
Final BCVA was significantly better in macula-on and macula-off eyes with intact ELMs and IS/OS junction lines (0.35 ± 0.18 logarithm of the minimal angle of resolution [logMAR] and 0.51 ± 0.17 logMAR, respectively) than in macula-off eyes with disrupted or absent ELMs and IS/OS junction lines (0.83 ± 0.17 logMAR and 1.04 ± 0.05 logMAR, respectively; P < .001, analysis of variance). Final BCVA also was better in the treated group than in the controls (0.55 ± 0.31 logMAR and 0.73 ± 0.26 logMAR, respectively; P = .05, t test). ELM and IS/OS junction line integrity were the main variables significantly affecting the final BCVA outcome (β = 0.42; P = .006, linear regression analysis). Disruption of the ELM and IS/OS junction line was observed in 21 of the 42 cases studied.
Conclusions
ERM secondary to retinal detachment surgery is accompanied by a high incidence (50%) of IS/OS junction line and ELM disruption. Among the variables studied, the condition of the IS/OS junction layer and the ELM are the main factors that predict final BCVA after ERM peeling.
Epiretinal membrane (ERM) formation may follow retinal detachment (RD) surgery, thus affecting visual function outcome. Pathogenesis includes proliferation of retinal pigment epithelium (RPE) cells, retinal glia, perivascular connective tissue, and hyalocytes. In accordance with the Gass classification, ERMs were characterized as either cellophane maculopathy (CM) or macular pucker (MP). ERMs develop in approximately 6% to 7% of eyes with primary rhegmatogenous RD after successful surgical repair with a scleral buckle (SB). A higher incidence of ERM formation (12.8%) has been found after pars plana vitrectomy (PPV), with one third of these patients undergoing reoperation for membrane peeling. Wilkins and associates first described structural distortion of the macula resulting from ERM using optical coherence tomography (OCT) and proposed an initial categorization. In subsequent years, OCT became more sensitive than clinical examination in detecting and monitoring ERMs.
Peeling of an ERM secondary to primary RD surgery contributes to anatomic restoration of the macula. However, visual acuity does not always improve proportionally as a result. Although most patients show an improvement in visual acuity after ERM removal, 12% do not, despite good anatomic restoration of the macula.
The purpose of this study was to use high-resolution OCT to investigate foveal microstructural changes, in particular changes in the status of the external limiting membrane (ELM) and inner segment/outer segment (IS/OS) junction line in the photoreceptor layer, in secondary ERMs after successful RD surgery and to evaluate visual outcomes after ERM peeling by PPV. A further aim was to determine possible prognostic factors regarding visual outcomes of ERM surgery using modern OCT technology.
Methods
The medical records of 451 consecutive RD patients who underwent surgery from July 2007 through December 2009 were evaluated retrospectively. Only cases with an anatomically successful primary operation were included in this retrospective study. RD cases treated with pneumatic retinopexy, patients with lens opacities graded above N03 or NC3 according to the Lens Opacities Classification System III at any point during the study, and eyes with concomitant diseases at presentation or during follow-up, such as diabetic retinopathy, age-related macular degeneration, various maculopathies, or uveitis, were excluded. Cases with best-corrected visual acuity (BCVA) worse than 20/200 after initial RD surgery also were excluded because a BCVA of at least 20/200 was required to achieve sufficient target fixation during examination with high-resolution spectral-domain (SD) OCT.
Finally, 389 patients met the inclusion criteria, each of whom had been operated on either with an SB or with standard PPV without dye in 2 different centers by different surgeons (P.G.T., G.P.T., A.C., V.G.). Routine follow-up after initial RD surgery revealed the formation of an ERM in 42 patients 1 to 3 months after surgery.
Spectral-Domain Optical Coherence Tomography Examination
ERM was diagnosed both clinically, using fundus examination after pupil dilation, and using SD-OCT examination. The same SD OCT device was used in all cases (Cirrus HD-OCT; Carl Zeiss Meditec, Inc, Dublin, California, USA). This SD OCT device provides noninvasive, cross-sectional imaging of the tissues using an 830-nm wavelength light source. It allows an ultra-high scan speed of 27 000 A scans per second, resulting in an axial resolution of 5 μm and a transverse resolution of 15 μm. A raster line protocol consisting of 5 parallel high-resolution scan lines was used for qualitative evaluation of the fovea. Additional horizontal and vertical crosshair scan lines were obtained (a minimum of 3 were obtained in each direction). The precise position of scan lines was displayed on high-definition fundus images captured simultaneously. The macular thickness protocol resulting in a macular cube (512 × 128 scans) was used for quantitative evaluation of macular thickness. Central foveal thickness (CFT) was determined as the value of the innermost central circle (1000 μm in diameter) in the macular thickness protocol. BCVA and CFT changes were assessed.
Photoreceptor Layer Evaluation
Retinal microstructure was evaluated in detail, with special emphasis placed on the integrity of the ELM and the photoreceptor IS/OS. The ELM and the photoreceptor IS/OS junction line were evaluated based on horizontal and vertical scans for 500 μm in either direction of the center of the fovea and were identified either as intact, disrupted, or completely absent. When the IS/OS junction line was continuous, it was characterized as intact. When it was interrupted by gaps shorter than 200 μm, it was characterized as disrupted. When the gaps were 200 μm or longer, it was characterized as absent. The status of the ELM and the IS/OS junction line were confirmed using a minimum of 2 horizontal or vertical crosshair scan lines. Evaluation was accomplished using the same masked OCT specialist (V.S.L.).
Group Formation
To evaluate final visual outcomes in relation to foveal microstructure, patients with secondary ERM (n = 42) initially were divided into 2 groups: the macula-on group, with an attached fovea before initial RD surgery, and the macula off group, with a detached fovea before initial RD surgery. These groups were divided further according to the ELM and the photoreceptor IS/OS junction layer status (intact, disrupted, or absent) as evaluated based on the SD OCT results ( Table 1 ).
| Macula Status (before RD Surgery) | Groups | Subgroups | Photoreceptor IS/OS Junction Layer Status (When ERM Was Diagnosed) | Duration of RD (Days), (Mean ± SD) | BCVA with the ERM (LogMAR), Mean ± SD | CFT with the ERM (μm), Mean ± SD | ERM Peeling (Yes:No) | Final BCVA (LogMAR), Mean ± SD | Final CFT (μm), Mean ± SD |
|---|---|---|---|---|---|---|---|---|---|
| On | 1 | 1A | Intact (n = 10) | 2.7 ± 1.4 | 0.55 ± 0.80 a | 270.90 ± 97.31 | 7:3 | 0.35 ± 0.18 b | 225.30 ± 30.11 |
| 1B | Disrupted (n = 0) | 0 | 0 | 0 | 0 | 0 | 0 | ||
| 1C | Absent (n = 0) | 0 | 0 | 0 | 0 | 0 | 0 | ||
| Off | 2 | 2A | Intact (n = 11) | 10.0 ± 6.6 | 0.48 ± 0.15 a | 303.18 ± 103.62 | 4:7 | 0.51 ± 0.17 b | 292.18 ± 30.70 |
| 2B | Disrupted (n = 16) | 30.3 ± 27.6 | 0.78 ± 0.16 | 362.06 ± 126.98 | 5:11 | 0.83 ± 0.17 | 349.00 ± 112.07 | ||
| 2C | Absent (n = 5) | 48.0 ± 21.7 | 1.12 ± 0.13 | 444.60 ± 52.90 | 1:4 | 1.04 ± 0.05 | 431.00 ± 42.20 | ||
| Average (n = 42) | 20.5 ± 23.9 | 0.69 ± 0.25 | 334.76 ± 117.87 | 17:25 | 0.66 ± 0.29 | 314.43 ± 106.32 | |||
| One-way ANOVA | P < .001 | P < .001 | P = .024 | P = .171 | P < .001 | P < .001 | |||
a Not significant ( P = 1, Bonferroni post hoc analysis).
A second operation for ERM peeling was proposed to all patients. Treated eyes underwent PPV (standard 3-port PPV without using dye) with membrane peeling and simultaneous lens extraction within 3 months after ERM was diagnosed (treated group). The remaining patients, who were reluctant to undergo a second operation, served as controls (untreated group).
Parameters Assessed
Visual acuity before and after RD surgery, duration of RD from the onset of symptoms until surgery (time elapsed) and macular status (macula on or macula off) were evaluated retrospectively. The extent of RD was measured in quarters of the retina. BCVA and CFT were measured before and after ERM removal. Foveal thickness, presence of subretinal fluid, IS/OS junction layer integrity, ELM status, and RPE changes were assessed using the same SD OCT device.
Follow-Up Protocol
Treated eyes with ERM peeling were examined after surgery on day 1, at 1 and 2 weeks and at 1, 3 and 6 months. Untreated eyes were examined in the first, third, and sixth months after the ERM was diagnosed. All patients included underwent a 6-month follow-up examination. Final BCVA and final foveal thickness were determined in the treated group as BCVA and CFT 6 months after ERM peeling and in the untreated (control) group as BCVA and CFT 6 months after the ERM was diagnosed. During follow-up, all patients underwent detailed ophthalmic examination including BCVA (logMAR), slit-lamp examination after pupil dilation, and SD OCT. For statistical analysis, BCVA was measured using a Snellen chart and was analyzed based on a logarithm of minimal angle of resolution (logMAR) scale. LogMAR values of 2.0 and 3.0 were assigned for counting fingers and hand movements, respectively.
All examinations performed adhered to the tenets of the Declaration of Helsinki and European legislation. The study was approved by the Institutional Scientific Board of “Attikon” University Hospital and “Henry Dunant” Hospital.
Statistical Analysis
Statistical analysis was performed using SPSS software version 13.0 (SPSS inc, Chicago, Illinois, USA). A number of factors (independent variables) related to the target variable assessed (i.e., final BCVA) were evaluated, for example, BCVA before RD surgery, time elapsed between RD appearance and surgery, macular status (macula on or off), RD extension, the procedure chosen (SB or PPV), ERM type (CM or MP), CFT after ERM formation, ELM and IS/OS junction status, and peeling (or not) of the ERM. We assumed that final BCVA was not affected by each factor independently, but rather by a combination of several factors. Therefore, linear regression analysis was conducted to determine the best linear combination of all independent variables assessed, as indicated. A hierarchical approach was chosen to correct multicollinearity. We examined whether the combination of these factors could predict final BCVA. Additionally, we evaluated the impact of each of the independent variables. Multifactorial analysis of variance (ANOVA), one-way ANOVA with Bonferroni post hoc analysis, and unpaired and paired t tests also were used as appropriate. Statistical significance was determined as P < .05.
Results
The study includes 42 cases. The mean age of the patients was 63.5 ± 6.4 years (mean ± standard deviation); 17 (40.5%) were male and 25 (59.5%) were female. When RD was diagnosed, the mean BCVA of the 42 patients in whom secondary ERM subsequently developed was 0.78 ± 0.57 logMAR (mean ± standard deviation); 27 (64%) were phakic, and the rest were pseudophakic. Time elapsed from the onset of symptoms until RD surgery was 20.5 ± 23.9 days; 10 RDs (10/42; 23.8%) were macula on and 32 (32/42; 76.2%) were macula off at the time of RD surgery; 2.6 ± 1.2 quarters of the retina were detached; 10 RDs (10/42; 23.8%) were operated on using an SB procedure and 32 (32/42; 76.2%) were operated on with PPV. Peripheral proliferative vitreoretinopathy (PVR) sparing the macula was observed in 5 cases (5/42; 11.9%). Any existing vitreoretinal traction was removed in all cases during initial PPV. Subretinal fluid was seen in 4 cases (4/42; 9.5%) subfoveally after primary SB, and this resolved spontaneously within 4 months. Submacular RPE atrophies also were noted in another 2 cases (2/42; 4.7%). When ERM first was observed after initial RD surgery, the mean BCVA was 0.69 ± 0.25 logMAR (mean ± standard deviation), and the CFT was 334.76 ± 117.87 μm (mean ± standard deviation; Table 1 ). The ERM had the characteristics of MP in 37 eyes (37/42; 88%) and of CM in 5 eyes (5/42; 12%). SD OCT revealed significant correlation between macular status (on or off) and IS/OS junction layer status (Pearson r = 0.503; P = .001). Indeed, all 21 eyes with disrupted or absent IS/OS junction layer had been diagnosed initially with macula-off RD (21/42; 50%). BCVA was significantly lower in the macula-off group than in the macula-on group (0.73 ± 0.26 logMAR vs 0.55 ± 0.08 logMAR, respectively; P = .002, t test).
During follow-up of the 42 patients, macula-on and macula-off cases with intact ELM and IS/OS junction line had significantly better BCVA than did eyes with disrupted or absent IS/OS junction lines ( P < .001; Table 1 ). Disruption or absence of the ELM and IS/OS junction lines were found only after macula-off RD with delayed initial repair. On the contrary, macula-on and macula-off RDs treated without delay were characterized by an intact IS/OS junction line ( Table 1 ).
There was significant difference in BCVA between cases with intact and disrupted or absent IS/OS junction lines ( P < .001, one-way ANOVA), as shown in Table 1 . However, the difference between macula-on and macula-off eyes with an intact IS/OS junction line was not statistically significant ( P = .2, Bonferroni post hoc analysis). CFT also was significantly higher in macula-off eyes with disrupted or absent IS/OS junction lines than in macula-on and macula-off eyes with an intact IS/OS junction layers ( P < .001, one-way ANOVA). No change in IS/OS junction layer status was noted during follow-up. In cases with IS/OS junction layer disruption, both the ELM and the IS/OS junction layer were involved. Eyes with an intact IS/OS junction layer also were characterized by an intact ELM. During follow-up, a positive correlation was observed between IS/OS junction layer integrity and BCVA.
Seventeen of the patients diagnosed with secondary ERM (17/42; 40%) were reoperated with PPV and membrane peeling (treated group). No significant complications resulting in reduction of BCVA or macular distortion were observed in treated eyes. In the remaining 25 patients (25/42; 60%) who served as controls (untreated group), the lens status remained unchanged during the 6-month follow-up period. A comparison of treated and control cases showed that BCVA improved significantly in treated eyes after surgery (from 0.70 ± 0.24 logMAR to 0.55 ± 0.31 logMAR, respectively; P < .001, paired t test), as shown in Table 2 . Mean final BCVA was better in treated eyes than in control eyes (0.55 ± 0.31 logMAR and 0.73 ± 0.26 logMAR, respectively; P = .0503, t test). Further analysis of the BCVA in the treated eyes revealed that, despite general improvement, the final BCVA remained unaltered in 5 eyes (BCVA ≤ 20/100). In all of these eyes, SD OCT showed disruption or absence of the ELM and the IS/OS junction layer ( Figure 1 ). On the contrary, BCVA decreased in the untreated controls during follow-up (from 0.68 ± 0.26 logMAR to 0.73 ± 0.26 logMAR; P = .003, paired t test). Additionally, CFT was reduced significantly in treated eyes (from 343.65 ± 124.77 μm to 291.82 ± 100.45 μm; P = .003, paired t test), whereas CFT remained unaltered in the untreated eyes ( Table 2 ). During follow-up, we observed that the status of IS/OS junction layer and ELM remained unaltered in both treated and untreated cases ( Figure 2 ).
