Predictive Association of Pre-Operative Defect Areas in the Outer Retinal Layers With Visual Acuity in Macular Hole Surgery

Abstract

Purpose: The purpose of this study was to develop methods to model the external limiting membrane (ELM) and ellipsoid zone (EZ) within the elevated cuff surrounding a macular hole (MH) to determine if the predicted size of the defect in these layers after virtual flattening was associated with the actual postoperative defect and best-corrected visual acuity (BCVA).

Methods: Patients were included who had undergone successful MH surgery. The defects in the ELM and EZ after virtual flattening were modeled using in-house software. Main outcomes were postoperative defects in ELM and EZ at 2 months and BCVA at 12 months.

Results: Fifty-eight patients were included. BCVA improved from 0.87 (0.31) logMAR pre-operatively to 0.26 (0.21) at 12 months (P < 0.001). For both the ELM and EZ, the predicted virtually flattened pre-operative defects were associated with the actual postoperative defects at 2 months (R2 = 0.33, P < 0.01 and R2 = 0.50, P < 0.01, respectively). There was a significant association of BCVA at 12 months (adjusted R2 = 0.85) with the pre-operative modeled area of the defect in the ELM (P < 0.01) and to a lesser extent with the defect in the EZ (P < 0.01) and base of the MH (P < 0.01).

Conclusions: Virtually flattening of the pre-operative defect in the ELM provides important predictive information of visual acuity. Incorporation of tools into commercially available optical coherence tomography (OCT) devices to facilitate such measurements would provide the clinician with important prognostic information.

Translational relevance: We have developed methodology that can potentially be used to predict the postoperative state of the outer retinal layers and the associated visual outcome in patients undergoing surgery for MH.

Figure 1.

Two different sections of a radial OCT scan in the 1:1 µm mode of a 51-year-old woman who presented with a 12 month history of a MH with an MLD of 363 µm. She underwent combined phaco-vitrectomy with ILM peeling and an inverted ILM flap and gas tamponade. BCVA at 12 months was 0.1 logMAR.

Figure 2.

(A to H) display the same two sections of the radial OCT scan shown in Figure 1 but in the 1:1 pixel mode with E and F demonstrating the marked ELM layers within the cuffs for each section. Of note is that the ELM in the right cuff in section E contains less ELM than in section F. G and H Show the marked residual EZ layers within the cuff, that are approximately equal in extent in both cuffs in both sections. (I to L) demonstrate the corresponding 2 sections 8 weeks postoperatively, with K showing a confluent ELM but a residual defect in the EZ, whereas L displays a confluent ELM and EZ.

Figure 3.

(A) Illustration of flattening process in a B-scan. Assume AB and CD (dotted lines) is the left and right side, respectively, of a measured feature (ELM and EZ) before flattening. The d_L and d_R are the distances between the end points B and C to the scan center, respectively. AB’ and C'D are after flattening from AB and CD by mapping the arc length of AB to the imaginary reference line of AD. The $d_L^{\text{'}}$ and $d_R^{\text{'}}$ are the unflattened distance between the end points B’ and C’ to the scan center. In general, after flattening, the gaps between B’ and C’ would be smaller than that between B and C. (B) Illustration of area estimation. Assume a radial scan volume has six B scans for simplicity. For each B scan i, we can estimate the gaps on the left and right to the scan center (middle of B scan), $d_L^i$ and$d_R^i$ as described above. An estimate of the enclosed area is made after converting the locations from polar to Cartesian coordinates.

Figure 4.

The correlation of BCVA at 12 months with the pre-operative defect areas in the ELM after virtual flattening was 80% (R² = 0.80, P < 0.01; n = 58).

Figure 5.

A ratio of the pre-operative defect area in the ELM after virtual flattening to before flattening >0.8 was associated with a significant drop in the BCVA at 1 year (n = 58).