Anatomical and Functional Recovery Kinetics After Epiretinal Membrane Removal

Introduction

Idiopathic epiretinal membrane (ERM) affects about 11% of the general population aged 49 years or older.¹ It is characterized by the growth of fibrocellular tissue on the internal limiting membrane (ILM), the basement membrane of the Muller cells.^2,3 Treatment of symptomatic ERM remains pars plana vitrectomy (PPV) with surgical peeling of the ERM which results in very good outcomes with 80% of patients improving by more than two lines of vision.⁴

Despite fairly universal surgical approaches, both anatomic and functional outcomes can be quite variable. Many studies have tried to find predictors of final outcomes, but they have focused predominantly on pre-operative findings. These previous studies have examined duration of symptoms, preoperative best corrected visual acuity (BCVA), pre-operative metamorphopsia, as well as optical coherence tomography (OCT) parameters, such as central macular thickness (CMT), ellipsoid zone (EZ) integrity and ganglion cell–inner plexiform layer thickness as potential prognosticators of both final BCVA and anatomic improvements.^4–27

To our knowledge, very few studies have reported on early (less than 30 days) visual and anatomical changes after ERM removal and no study has attempted to understand the kinetics of anatomical and functional changes in this early postoperative period. Hence, we herein present a retrospective study on early and late kinetics of spectral domain optical coherence tomography (SD-OCT) and functional findings after ERM removal without intentional ILM peeling and correlate these with outcomes.

Methods

Patient Selection

The medical records of patients with idiopathic ERM who underwent PPV with surgical removal of an idiopathic ERM between January 2012 and June 2017 under the care of a single surgeon (DGV) were reviewed. Inclusion criteria included patients who had performed SD-OCT before the surgery and after the 1st day, 1st week and 1st month. Patients with concurrent macular holes, macular disease, venous or arterial occlusion, diabetic retinopathy, retinal tear, retinal detachment, endophthalmitis or trauma were excluded. This study was conducted in accordance with the tenets of the Declaration of Helsinki and received approval from the Institutional Review Board of Massachusetts Eye and Ear Human Studies Committee (title: 1,351,674-2, protocol:1,351,674-2/(18-172H)) prior to its commencement. Patient consent was waived due to the retrospective nature of this study and Health Insurance Portability and Accountability Act (HIPPA) compliance was maintained throughout.

Surgical Procedure and Post Surgical Care

All surgical procedures were performed in a teaching hospital with surgical fellows under the supervision of a single surgeon (DGV) using standard 3 port vitrectomy (25g or 27g) and no intentional ILM peeling. No dye was used to investigate if ILM was inadvertently peeled with the ERM. Triamcinolone acetonide staining of ERM was used for the majority of these cases. Post operative care included routine antibiotic drops for a week as well as topical steroids and non-steroidal medications for 4–6 weeks post op.

Anatomical and Functional Outcomes

SD-OCT was performed using Cirrus HD-OCT (Carl Zeiss Meditec AG, Jena, Germany) or Spectralis SD-OCT (Heidelberg Engineering GmbH, Heidelberg, Germany) pre-operatively and post-operative on day 1, week 1, months 1, 3, 6, 12 and 24. ETDRS BCVA was measured on the respective follow-up dates and converted to the logarithm of the minimum angle of resolution (logMAR) for statistical analysis. Outcomes included anatomical changes based on CMT and functional changes based on BCVA. Proportional early CMT and BCVA changes form the pre-operative baseline values were also evaluated as a percentage of the final respective values at 24 months follow-up and as a percentage of the healthy fellow eye CMT and BCVA.

Statistics

Statistical analysis was performed using the statistical software STATA (College Station, TX, USA). Spearman correlation coefficient (r) was used to investigate strength and direction (positive or negative) of linear relationships between paired variables. Paired t-test analysis was performed with p values <0.05 considered as an indication of statistically significant results.

Results

Out of 200 patients (213 eyes) that underwent ERM removal between 2012 and 2017 in our clinic, seventy nine (79) eyes of 71 patients had 1st day and 1st week SD-OCT, 35 eyes were excluded due to concurrent macular hole (n= 20), diabetic retinopathy (7), branch venous occlusion (3), retinal tear (2), retinal detachment (1), or history of endophthalmitis (1). After these exclusions, there were 45 eyes of 42 patients that were included for analysis in this study, 22 male and 20 female, with a mean age of 73 ± 13.4 years and a mean follow-up of 20 months. Thirty eyes had at least 12 months follow up and 26 eyes had at least 24 months follow-up. In terms of lens status at the time when ERM was removed, 25 eyes were pseudophakic and 20 phakic. Of the 20 phakic eyes, 15 became pseudophakic during the follow-up period. Baseline characteristics are shown in Table 1.

Table 1 Baseline Characteristics

Mean CMT was 475 ± 129.7 preoperatively improving to 445 ± 110.3 at 1st day, 408 ± 106.4 at 1st week, 381 ±72.0 at 1st month, 361± 58.6 at 3 months, 365 ± 75.2 at 6 months, 356 ± 59.2 at 12 months and 346 ± 57.3 at 24 months post-operatively. Improvement in CMT was statistically significant after the 1st week (p=0.005), 1st month (p<0.001), 3 months (p=0.001), 6 months (p=0.001), 12 months (p < 0.001) and 24 months (p < 0.001) Proportional decrease of CMT compared to the baseline pre-operative values as a percentage of the final CMT reached at 24 months was 24% at 1st day, 54% at 1st week, 74% at 1st month, 88% at 3 months, 92% at 6 months and 95% at 12 months. When plotted, the improvement of CMT was found to be logarithmic with time, with an R² of 0.96 (Figure 1).

Figure 1 Proportional decrease of CMT values as a percentage of the final CMT at 24 months was found to be logarithmic with time.

Mean BCVA was 0.49 ± 0.43 (Snellen equivalent of 20/60) pre-operatively and 0.80 ± 0.56 (20/125) at 1st day, 0.40 ±0.24 (20/50) at 1st week, 0.33 ± 0.26 at 1st month (20/40), 0.26 ± 0.21 (20/36) at 3 months, 0.28 ± 0.25 at 6 months (20/38), 0.22 ± 0.20 (20/32) at 12 months and 0.25 ± 0.23 (20/35) at 24 months. BCVA was significantly worse at 1st day post-operatively (p=0.003), not significantly different from the pre-operative values at 1st week (p=0.085), significantly improved at 1st month (p=0.018), 6 months (p=0.006), 12 months (p=0.0001) and at 24 months (p=0.00082) post-operatively.

When plotted over time, the improvement of CMT and BCVA were found to be logarithmic (R² = 0.84 and R² = 0.96 respectively) suggesting that early (less than 30 days) post-operative structural and functional changes may be predictive of long-term outcomes (Figure 2).

Figure 2 Overlay of CMT and BCVA scatter-plots with time revealing a slight delay in the improvement of BCVA compared to CMT. The improvement of CMT and BCVA were found to be logarithmic with time (R2=0.84 and R2=0.96 respectively).

Preoperative BCVA revealed a weak positive correlation with BCVA at 12 and 24 months. (R²= 0.12, r=0.348 and R²=0.13, r=0.36 respectively with p<0.05) (Figure 3A). Likewise, there was only a weak positive correlation between preoperative CMT and CMT measurements at 12 and 24 months (R²= 0.19, r=0.44 and R²=0.16, r=0.40 respectively with p<0.05) (Figure 3B).

Figure 3 Correlation between baseline characteristics and final outcomes. (A) Correlation of preoperative BCVA to BCVA measured at 24 months (R2=0.13, r= 0.36); (B) Correlation of preoperative CMT to CMT measured at 24 months (R2= 0.16, r= 0.40).

When proportional decrease of CMT compared to the baseline pre-operative values was plotted as a percentage of the fellow normal eye CMT, the respective values were 14.7% at 1st day, 32% at 1st week, 50% at 1st month, 64% at 3 months, 63% at 6 months, 62% at 12 months and 68.9% at 24 months post-operatively. When plotted, the improvement of CMT was found to be logarithmic with time, with an R² of 0.94 (Figure 4).

Figure 4 Proportional decrease of CMT values as a percentage of the fellow normal eye CMT was found to be logarithmic with time.

Of note, although CMT decreased significantly after ERM removal, it never reached the CMT of the normal fellow eye, which is in accordance with previous studies.^28–30

When plotted as a percentage of the fellow normal eye CMT, first week proportional improvement in CMT from pre-operative baseline was found to be correlated with final CMT proportional decrease (R²= 0.72, r= 0.85, p<0.001) (Figure 5). Hence, it appears that early postoperative CMT could be predictive of final CMT, more so than preoperative baseline CMT. (r=0.846 vs r=0.405) Our study was found to be underpowered in detecting a similar relationship between early postoperative and final BCVA or between CMT and BCVA (sample size needed N= 66 and N=91 respectively).

Figure 5 Correlation of first week proportional improvement in CMT from pre-operative baseline to proportional decrease of final CMT, plotted as a percentage of the fellow normal eye CMT (R2 =0.72, r =0.85).

Of interest, when CMT and BCVA scatter-plots with time are over-layered, the logarithmic improvement in BCVA seems to follow the logarithmic decrease in CMT. The worsening in 6-month mean BCVA is attributed to four patients developing cataract and was eliminated by cataract extraction and IOL placement by the 12th month of our follow-up (Figure 2).

In the phakic cohort, mean pre-operative CMT was significantly higher than in the pseudophakic cohort (p<0.05). CMT of the two cohorts was not found to be significantly different at any of the follow-up time points. Similarly, proportional decrease of CMT as a percentage of the preoperative CMT was not found to be significantly different at any of the follow-up time points either. Of note, in our cohort of 45 eyes that underwent ERM removal without intentional ILM peeling, there were no recurrences noted at 24 months of follow-up.

Discussion

To our knowledge, this is the first study reporting on the early (less than 30 days) kinetic functional and anatomical changes after ERM removal as prognosticators of final outcomes. Our study reports on proportional decrease in CMT compared to pre-operative baseline values as a percentage of the final CMT, in order to account for any differences between the pre-operative baseline values.

Several studies have attempted to correlate BCVA and CMT values at various post-operative time-points mostly rendering poorly correlated anatomical and functional results.^6,8,10,11 Our early postoperative kinetics analysis revealed a logarithmic improvement in BCVA over time that appears to follow the logarithmic decrease in CMT. As described above, following ERM removal, the CMT reached 24% of the final CMT at 1st day, 54% at 1st week, 74% at 1st month, 88% at 3 months, 92% at 6 months and 95% at 12 months. The logarithmic improvement of CMT with time (R^{2 =} 0.84) suggest that early (less than 30 days) post-operative anatomical changes appear to be predictive of long-term outcomes. To date, only two studies have reported on early (less than 30 days) OCT changes after ERM removal with intentional ILM peeling; none reported on early CMT changes. The first study Clark et al³¹ reported on RNFL changes at 1 week post op, with 31% of patients having swelling of the arcuate nerve finer layer (SANFL). In the second study, Scupola et al³² analyzed the RNFL changes identified in 39 out of 46 eyes (84.8%) at 14 days post-op.

In terms of functional results, in our study following ERM removal, the logMAR VA decreased on average to 0.8 (from 0.49 pre-operatively) on the 1st day, and improved to 0.4 at 1st week, 0.33 at 1st month, 0.26 at 3 months, 0.28 at 6 months, 0.22 at 12 months and 0.25 at 24 months. The logarithmic improvement of BCVA with time (R^{2 =} 0.96) suggest that early (less than 30 days) post-operative functional changes appear to be predictive of long-term outcomes for our cohort of patients but not necessarily for an individual patient. These results are in accordance with those of Scupola etal's study that found rapid improvement of the BCVA during the 1st postoperative month.

Of the parameters previously studied as potential final BCVA prognosticators, baseline preoperative BCVA^{4–6,11,13–15,20,22,23} and baseline preoperative CMT^14,15,21 have been widely suggested to be well correlated with the final visual outcome. In this study, preoperative BCVA and CMT only revealed a weak positive correlation with final BCVA and CMT (R²= 0.13 and R²=0.16) (Figure 3A and B) while early postoperative CMT was predictive of final CMT, more so than preoperative baseline CMT (R²=0.72 vs R²=0.16). Pre-operative ellipsoid zone integrity (formerly called IS/OS junction) has also been proposed as a prognosticator for final BCVA by several studies^{9–11,13,17–19} while other studies found no significant correlation.^21,33,34 In accordance with our findings, Kim et al¹² among other factors suggested first postoperative month CMT as a good prognosticator of final BCVA at 12 months.¹² In terms of lens status, Pesin et al⁶ reported the presence of preoperative pseudophakia to be associated with significantly greater BCVA improvement at 6 months compared to baseline phakic eyes. Lastly, recent evidence suggests that macular ganglion cell–inner plexiform layer (GC-IPL) thickness may also be correlated with postoperative visual outcomes after ERM peeling,³⁵ yet with conflicting evidence on whether decreased GC-IPL thickness is positively^16,25,27 or negatively^24,26 correlated with BCVA improvement and final BCVA.

A limitation of this study apart from its retrospective nature is that it does not report on SD-OCT-based overall macular volume, as readings differ between the two SD-OCT devices that were used, Heidelberg and Cirrus. Another limitation of this study is that staining was not performed following each procedure to ascertain in how many cases the ILM was or was not spontaneously peeled along with the ERM.

Conclusion

Early postoperative kinetics analysis revealed a logarithmic improvement in BCVA over time that appears to follow the logarithmic decrease in CMT. More than 50% of the improvement occurs in the first week, 75% by the 1st month and almost 90% by the 3rd month.

Early (less than 30 days) post-operative anatomical and functional changes appear to be predictive of long-term outcomes. Incorporating first week post-ERM peeling SD-OCT in the clinical practice may help offer the patient valuable information about their final outcome.