Abstract
Purpose.:
To determine whether there is a significant correlation between the integrity of the foveal microstructures and the best-corrected visual acuity (BCVA) after pars plana vitrectomy for epiretinal membrane (ERM) removal.
Methods.:
This was a retrospective, interventional case series. Forty-six eyes of 45 patients with an ERM underwent vitrectomy. The foveal area was examined by spectral-domain–optical coherence tomography (SD-OCT) preoperatively and postoperatively. The correlation between the length of the photoreceptor cone outer segment tips (COST) line defect, the inner segment/outer segment junction (IS/OS) line defect, the external limiting membrane (ELM) line defect, and the BCVA was determined.
Results.:
The length of the COST line defect was significantly correlated with the BCVA at postoperative 1, 3, 6, 9, and 12 months (P < 0.001 for all). Forward stepwise regression analyses showed that the postoperative BCVA was significantly correlated with the length of COST line defect (P < 0.001) but not with the IS/OS line and ELM line defects for up to 6 months. The preoperative length of the COST line defect was significantly correlated with the postoperative BCVA at 12 months (P = 0.005), but the lengths of the IS/OS line defect and ELM line defect were not. The factor that best predicted the postoperative BCVA was the length of the preoperative COST line defect (P = 0.04) but not the preoperative BCVA (P = 0.69).
Conclusions.:
The recovery of the foveal COST line defect is correlated with the BCVA after ERM surgery. The length of the preoperative COST line defect can predict the potential foveal function. (ClinicalTrials.gov number, NCT01549249.)
Of the original 283 eyes, 72 eyes with a secondary ERM, including 23 eyes with history of uveitis, 36 eyes with retinal tears, 6 eyes with diabetic retinopathy, 4 eyes with retinal vein occlusion, and 3 eyes with macroaneurysm, were excluded. Eyes with coexistence of asteroid hyalosis, familial exudative vitreoretinopathy, retinal schisis, pigmentary degeneration, and severe cataract (>3 Emery–Little classification) were also excluded. Fourteen eyes with high myopia with an axial length > 27.0 mm or refractive error greater than −8.0 diopters (D), 9 eyes with a pseudomacular hole, and 7 eyes with vitreomacular traction syndrome were excluded. Twenty-two eyes with history of vitreoretinal surgeries and 5 eyes with postoperative complications including retinal detachments were excluded. After exclusion of these 133 eyes, only those with more than 6 months of postoperative follow-up were included. In the end, 46 eyes of 45 patients (19 men, 26 women) met the study criteria for the data analyses.
All patients were diagnosed with an ERM by SD-OCT and had undergone surgery between November 2008 and September 2010 at the Kyorin Eye Center. The preoperative data analyzed were age, sex, right or left eye, axial length, symptom duration, BCVA measured with Landolt C chart, central foveal thickness (CFT), length of the COST line defect, and length of the IS/OS and external limiting membrane (ELM) line defects.
All surgeries were performed after the patients received a detailed explanation of the surgical and SD-OCT procedures. Informed consent was obtained from all patients, and the procedures adhered to the tenets of the Declaration of Helsinki. The study protocol was approved by the Institutional Review Committee of the Kyorin University School of Medicine, and all patients consented to our review of their medical records.
The surgery was performed by one of the three retina specialists (KH, MI, AH). After 2% lidocaine retrobulbar anesthesia, a standard three-port pars plana vitrectomy was performed to remove the ERM. Triamcinolone acetonide (MaQaid; Wakamoto Pharmaceutical Co., Ltd., Tokyo, Japan; or Kenacort-A; Bristol Pharmaceuticals KK, Tokyo, Japan) was injected intravitreally to make the vitreous gel and internal limiting membrane (ILM) more visible. This also allowed the surgeons to confirm that residual ERM and ILM were not present at the end of the vitrectomy.
Core vitrectomy was performed with the creation of a posterior vitreous detachment if it was not present; then the ERM and ILM were removed in all cases. The area of the ILM peeled was 2 to 3 optic disc diameters centered on the fovea. The lens was extracted from all patients > 55 years old, and all cataractous lenses were removed by phacoemulsification with an implantation of an intraocular lens. Anatomical success was defined as the complete removal of the ERM at the fovea as confirmed by slit-lamp biomicroscopy and SD-OCT. The main outcome measures were the postoperative BCVA and the condition of the foveal microstructures in the SD-OCT images at 12 months.
The patients had comprehensive ophthalmologic examinations before and 1, 3, 6, 9, and 12 months after the surgery. The examinations included measurements of the BCVA, binocular indirect ophthalmoscopy and noncontact lens slit-lamp biomicroscopy, and fundus photography. Spectral-domain–optical coherence tomography examinations were performed on all patients on the same day as the clinical examinations. The fovea was scanned, and the sections through the fovea were confirmed by the presence of the foveal bulge in the SD-OCT images. When a foveal bulge could not be identified, other morphological characteristics of the fovea, including a foveal depression, thinning of inner retinal layer in the SD-OCT images, and foveal avascular zone in the fundus images, were used as landmarks. The length of the COST line defect, IS/OS line defect, ELM line defect, and CFT were measured on the SD-OCT images.
A SD-OCT (OCT4000, Cirrus HD-OCT; Carl Zeiss Meditec, Inc., Dublin, CA) was used to record and analyze the images of the macular area. The entire macular area was scanned, and high-quality 6-mm scan images were obtained with the 5-line raster mode or high-definition 5-line raster mode. The length of the COST line defect, IS/OS line defect, and ELM line defect was measured with the software embedded in the SD-OCT system, and the lengths of the vertical and horizontal defects at the fovea were averaged. The CFT was defined as the maximum distance from the vitreoretinal surface to the RPE at the fovea. Two experienced investigators (YI, TR), who were masked to the patients' information including the postoperative period and the BCVA, measured each of the parameters on each SD-OCT image independently, and the average of the two investigators was used for the statistical analyses. Scans with a signal strength greater than 8/10 were examined, and a high-quality representative image was selected for the measurements. The axial length of the eye was measured with the Tomey OA1000 (Tomey Corp., Nagoya, Japan) preoperatively.
The decimal BCVA was converted to logarithm of minimal angle of resolution (logMAR) units for the statistical analyses. To compare two groups, Student's t-tests were used with continuous data. Multivariable analysis was also used to investigate the relationships between the BCVA and the preoperative or postoperative lengths of the COST line, IS/OS line, and ELM line defects at 1, 3, and 6 months. Simple linear regression analysis was used to determine the significance of correlations between the postoperative BCVA and preoperative or postoperative length of COST line defect. Forward stepwise regression analyses were performed to determine whether age, sex, symptom duration, axial length, CFT, and preoperative BCVA were significantly correlated with the BCVA at 12 months after the surgery.
The correlation between the pre- or postoperative BCVA and the length of the COST line defect at each time point was determined by simple linear regression analysis.
The length of COST line defect was significantly correlated with the BCVA at each postoperative time (
P < 0.001;
Fig. 4).
Forward stepwise regression analysis was performed to determine whether the pre- and postoperative BCVAs were correlated with and affected by the postoperative length of the COST line, IS/OS line, or ELM line defects as causative factors. Our analyses showed that the BCVA was significantly correlated with the length of COST line defect preoperatively and at 1, 3, and 6 months postoperatively (
R 2 value = 0.70, 0.65, 0.73, 0.71, respectively,
P < 0.001 for all times;
Table 2). After 9 months, the IS/OS and ELM lines were completely recovered in all cases, and we did not perform the multivariate statistical analysis.
Table 2. Forward Stepwise Regression Analysis Between SD-OCT Images and BCVA at 1, 3, and 6 Months Postoperatively
Table 2. Forward Stepwise Regression Analysis Between SD-OCT Images and BCVA at 1, 3, and 6 Months Postoperatively
| Before Operation | 1 Month | 3 Months | 6 Months |
SPRC | F Value | P Value | SPRC | F Value | P Value | SPRC | F Value | P Value | SPRC | F Value | P Value |
COST line defect | 0.83 | 23.7 | <0.001 | 0.77 | 43.9 | <0.001 | 0.97 | 105.4 | <0.001 | 0.9 | 60.8 | <0.001 |
IS/OS line defect | 0.41 | 2.10 | 0.15 | −0.06 | 0.28 | 0.60 | 0.69 | 0.34 | 0.56 | −0.57 | 0.72 | 0.40 |
ELM line defect | −0.53 | 4.2 | 0.09 | 0.15 | 2.21 | 0.14 | −0.9 | 0.58 | 0.45 | 0.56 | 0.67 | 0.42 |
The correlations between the postoperative BCVA and the preoperative length of the COST line defect, IS/OS line defect, and ELM line defect were also calculated. Forward stepwise regression analysis showed that the preoperative length of the COST line defect was significantly correlated with postoperative BCVA at 3, 6, 9, and 12 months (P = 0.018, 0.0005, 0.0005, 0.005, respectively), but not at 1 month (P = 0.055). The preoperative lengths of IS/OS line (P = 0.76, 0.89, 0.49, 0.09, 0.22, respectively) and ELM line defects (P = 0.43, 0.69, 0.16, 0.73, 0.89, respectively) were not significantly correlated with the BCVA at postoperative 1, 3, 6, 9, and 12 months.
Because the preoperative COST line defect was significantly correlated with the postoperative BCVA at 3, 6, 9, and 12 months by forward stepwise regression analysis, the correlation of the postoperative BCVA to the length of the COST line defect was evaluated at each time point by simple linear regression analysis. Our analyses showed that the preoperative length of the COST line defect was significantly correlated with the postoperative BCVA at postoperative 1, 3, 6, 9, and 12 months (
P = 0.047 at postoperative 1 month,
P < 0.001 at postoperative 3, 6, 9, and 12 months;
Fig. 5). The relationship between the preoperative length of the COST line defect and the postoperative BCVA at 12 months is given by the following equation:
where the BCVA is the postoperative BCVA in logMAR units at 12 months and
L is the preoperative length of the COST line defect in micrometers (95% confidence interval, 0.00006–0.00014). The estimation of the postoperative BCVA at 12 months from the preoperative length of COST line defect was significant (
F value = 25.9,
R 2 = 0.40;
P < 0.001).
Four hyperreflective outer retinal bands have been identified in images obtained by SD-OCT. The innermost (first) line is the ELM line, which consists of the zonular adherences between the photoreceptor inner segments and the Müller cell processes.
26 The second hyperreflective band posterior to the ELM line corresponds to the photoreceptor IS/OS line; and the third hyperreflective band posterior to the IS/OS line is the COST line, which is attributed to the scattering signals from the interface of the interdigitations between cone outer segments and the RPE microvillae.
26 The cone outer segments are shorter than the rod outer segments, and the RPE cell processes extend into the cone outer segment layer.
26 The fourth and deepest line represents the RPE layer.
Spaide and Curcio
27 demonstrated that the second band is aligned with the ellipsoid portion of the inner segments, and the third band corresponded to the ensheathment of the cone outer segments by the apical processes of the RPE in a structure known as the contact cylinder. These microstructures are vulnerable to various macular pathologies and can thus serve as hallmarks for the integrity of the photoreceptors.
Because the foveal microstructures were clearly observed in the SD-OCT images, we were able to measure the degree of integrity of the SD-OCT signals and calculate the correlation between the integrity of the photoreceptor microstructures and the postoperative BCVA. A reconstruction of the foveal ELM line in the early postoperative period is essential for the restoration of the foveal photoreceptor layer after macular hole surgery.
28
The integrity of the photoreceptor IS/OS line has been extensively studied, and it has been reported that an intact IS/OS line was significantly associated with better postoperative BCVA after successful removal of an ERM.
5,11,14–16 Shimozono and associates
29 reported that the diameter of the IS/OS line and the COST line defects at 1 month was significantly correlated with the BCVA at 6 months but the COST line defect at baseline was not. The incidence of the eyes with disrupted COST line or IS/OS line at postoperative 1 month increased compared with base line, but then decreased after 6 months. The early postoperative status of the COST line was critical for later recovery of the BCVA; however, it included the effects of surgical invasion and postoperative inflammation.
29 Thus, the early postoperative status may not be a good predictive factor.
Our results indicated that there was a continuous postoperative recovery from 1 to 12 months, and forward stepwise regression analysis showed that the preoperative length of the COST line defect was significantly correlated with the postoperative BCVA from 3 to 12 months. Simple linear regression analysis showed a significant correlation from 1 to 12 months.
Shimozono and associates
29 stated that the reason the preoperative BCVA was not significantly correlated with the postoperative BCVA at 6 months was that they did not exclude patients with moderate cataracts. Our study included patients who had undergone combined cataract surgery with vitrectomy, and the postoperative recovery of BCVA may have been influenced by the status of the cataract in the patients with combined cataract surgery (phacovitrectomy), although eyes with severe cataract were not included in the study. Because the BCVA at baseline may be affected by the status of the lens, it is reasonable that the status of the COST line was the most important predictive factor for the postoperative BCVA irrespective of lens status. The COST line at baseline was also strongly correlated with the BCVA at baseline, which may indicate that the status of the lens did not strongly affect the BCVA at baseline.
Our study demonstrated that the preoperative length of the COST line defect was significantly correlated with the postoperative BCVA at 12 months, but that the preoperative lengths of the IS/OS line defect and the ELM line defect were not. We recently reported that the length of the COST line defect was significantly correlated with BCVA at each postoperative period from 3 to 12 months after macular hole surgery.
25 However, the lengths of IS/OS line defect and the ELM line defect were not significantly correlated with the BCVA 6 months after macular hole surgery.
25 These findings indicate that the recovery of the COST line was more important for the postoperative BCVA than that of the IS/OS line and the ELM line after macular hole and ERM surgery.
The mean preoperative BCVA of our patients was 20/30, which was better than that of the average ERM candidate with BCVA more typically between 20/50 and 20/80. The recovery of the foveal COST line after successful macular hole surgery was correlated more strongly with eyes with postoperative BCVA better than 20/25. This may explain the stronger correlation between the BCVA and the COST line than with the IS/OS and ELM lines in our patients.
24
The metamorphopsia caused by an ERM has been suggested to be due to a displacement of photoreceptors induced by tangential traction by the ERM on the retina.
30,31 In a physiological analysis by binocular correspondence perimetry, the metamorphopsia was found to be due to photoreceptor displacements in eyes with an ERM.
32 The COST line is located closer to the RPE layer than the IS/OS line, and the COST line signals may be reduced by the photoreceptor displacement because of a tilting of the photoreceptor outer segments. Adaptive optics scanning laser ophthalmoscopy images of eyes with an ERM showed an abnormal cone mosaic pattern similar to microfolds of the foveal photoreceptor layer.
33 The location of the abnormal cone mosaic pattern corresponded to the sites of IS/OS line disruption.
33 The COST line may be disrupted by a similar mechanism. Decreased cone densities with altered cone mosaic patterns were detected by adaptive optics scanning laser ophthalmoscopy in eyes with a disrupted IS/OS line or COST line in patients with central serous chorioretinopathy.
21 Poor interdigitation of cone photoreceptor cells and the underlying RPE can result in a disruption of the COST line.
The CFT has been reported to be significantly correlated with the BCVA and to be a prognostic factor for the postoperative BCVA after ERM removal.
1,3 Our results showed that the preoperative CFT was not significantly correlated with the postoperative BCVA as previously reported.
5,11,15,34 Although these opposing results may be due to different study populations and different methods of evaluation, the CFT does not appear to be a major prognostic factor compared with the integrity of the COST line. We did not evaluate the presence of a dissociated optic nerve fiber layer (DONFL) appearance, which is frequently found in eyes after vitrectomy with ILM peeling for ERM of macular hole. However, the presence of DONFL has been reported to not preclude postoperative good visual recovery.
35 We also did not use microperimetry to evaluate the retinal sensitivity in the area of the COST line defect. This is relevant because it has been reported that the preoperative lengths of the IS/OS and ELM line defect were associated with the foveal sensitivity at 6 months after surgery in macular hole patients.
36
There are limitations to this study. This was a retrospective study without a control group. Also, the number of patients in each group was small, and the resolution of SD-OCT used may not have been sufficient to detect the COST line compared to high-speed ultrahigh-resolution OCT. Furthermore, even with the higher spatial resolution SD-OCT instruments, the foveal center can be missed, which results in lower-intensity images and evaluation of the wrong area of the fovea. Therefore, further studies with a larger sample size and higher-resolution SD-OCT are needed to confirm the results of this study.
In summary, our quantitative measurements of the photoreceptor COST line defects showed that the recovery of photoreceptor COST line was correlated with the postoperative BCVA after ERM surgery. We also found that the preoperative length of the COST line defect may predict the postoperative BCVA after successful ERM removal and the potential foveal function.