Abstract
Introduction
High-resolution optical coherence tomography (OCT) allows for the visualization, subclassification, and monitoring of full-thickness macular holes (FTMHs).1,2 To date, baseline OCT scans of eyes of patients with FTMH have been extensively investigated using linear modeling to identify predictors of surgical outcomes. 3 OCT parameters, such as minimum linear diameter (MLD), base diameter, arm length, and various indices calculated based on ratios of these linear measurements, have been shown to correlate with pre- and postoperative visual acuity as well as rates of hole closure.4–9 Area measurements of the hole have also displayed predictive potential; however, most studies obtained measurements from only a single cross-section of the OCT scan, which carries significant variability in their predictive values.10–12
Volumetric analysis is a novel way to assess FTMH parameters and features, which differs from traditional OCT analysis in that it uses the entire OCT volume rather than single-line or cross-sectional data. Performing volumetric analysis allows better quantification of OCT parameters such as intraretinal cyst (IRC) volume and FTMH volume to prognosticate outcomes.
Few prior studies have performed volumetric analysis of macular holes. Those that have assessed FTMH volume and the associated ratios for outcome prediction have yielded variable results, as the highly asymmetric nature of macular holes poses many complications in its volumetric computation.4,11,13 IRCs observed around the FTMH are hypothesized to be related to vitreopapillary adhesion formation as well as intra- and extracellular fluid accumulation.14,15 Recently, linear IRC parameters (such as area and height) have been analyzed for correlations with existing parameters and surgical outcomes16–19; however, volumetric measurements and IRC analyses have yet to be done. Similarly, Barak et al found pre–macular hole foveae to be significantly more asymmetric than normal foveae, 20 but no prior research has attempted to analyze FTMH symmetry as a prognostic predictor.
Given the anatomic advantages of OCT volumetric analysis in the characterization of FTMH, our study aimed to use in-depth segmentation of the full-volume OCT scans of the eyes of patients with FTMH to analyze the impact of cystoid volume and macular hole symmetry on visual and surgical outcomes. These parameters may be useful to inform perioperative planning and prognosis for patients.
Methods
This was a multicenter, retrospective cohort study of eyes of patients with FTMH who underwent pars plana vitrectomy (PPV) with internal limiting membrane (ILM) peeling and gas tamponade (eg, sulfur hexafluoride) at 2 vitreoretinal surgical referral centers, the Vanderbilt Eye Institute and the Retina Group of Washington, from 2017 to 2022. Due to the time-intensive nature of manual segmentation of the OCT scans, which was performed by 3 graders (as described in more detail below), only the 35 most recent patients who met the study inclusion criteria from each center were analyzed in this pilot study. The study was approved by the Vanderbilt University Medical Center Institutional Review Board and adhered to the principles of the Declaration of Helsinki.
Inclusion criteria were as follows: (1) patients with FTMH must have undergone a PPV surgery coded with current procedural terminology code 67042; and (2) the patient’s pre- and postoperative (6-month follow-up) OCT imaging scans had to be obtained using a Spectralis SD-OCT system (Heidelberg Engineering, Inc.). In addition, patients with diabetes were included in the study as long as they did not have macular edema, severe nonproliferative diabetic retinopathy, or proliferative diabetic retinopathy. Patients were excluded if they had (1) prior ocular or retinal surgeries (other than uncomplicated cataract extraction), (2) prior intravitreal injections, (3) a history of other macular conditions, ocular trauma, severe glaucoma, or optic neuropathy, or (4) high myopia (defined as an axial length of >26 mm).
Preoperative OCT scans were performed within 2 months before surgery. Each OCT volume, consisting of 60 B-scans over the same field-of-view centered on the macular hole, was manually segmented by 3 graders to demarcate the retinal pigment epithelium (RPE), IRCs, and the macular hole. Of the 60 B-scans per volume, 11–13 B-scans encompassed the central 1500 µm, which would include the macular hole and most IRCs. MLD was determined by manually measuring the narrowest diameter in the mid-retina, roughly parallel to the RPE, on the central horizontal raster B-scan. Volume was extrapolated by assuming that the areas adjacent to the B-scan held the same areas. Thus, IRC volume was calculated as the cumulative area of all segmented cystic spaces across all B-scans. FTMH symmetry was calculated by mapping several ellipses onto the en-face projection of the macular hole volume. Figure 1A and 1B show the segmentation process and the generated IRC areas and FTMH areas, which were summed to provide a volume. Ellipses were mapped at 3 levels: the inner level (ILM), the base level (RPE), and the waist level (narrowest diameter of the hole). The ILM was set as the line connecting the apex of the retina on either side of the FTMH on the horizontal raster B-scan. The ellipse’s major and minor diameters were divided to create a symmetry ratio at each level. An example of these calculations is given in Figure 1C, which shows the cross-sectional image of an FTMH mapped by an ellipse, along with the major and minor radius. These values (or their equivalent diameters) were divided to calculate the macular hole symmetry. The mean of the 3 ratios was also calculated.
A visual representation of the optical coherence tomography B-scan segmentation and analysis processes. (A) Segmentation of a B-scan of a full-thickness macular hole (FTMH). Red dashed lines divide the inner, waist, and base levels of the hole. (B) Volumetric representation of the FTMH areas (purple) and intraretinal cyst areas (light blue), which are summed to calculate the volume. (C) Cross-sectional images of an FTMH (yellow) at the 3 levels, with overlaying ellipses (red dashed lines) and the major radius (blue) and minor radius (green).
In addition, preoperative clinical features were collected, including baseline lens status, duration of symptoms, and best-corrected visual acuity (BCVA). Duration of symptoms at baseline was classified as follows: <3 months, 4–6 months, or >6 months. Additional collected OCT features included the FTMH stage as well as the presence of complete posterior vitreous detachment (PVD), vitreomacular traction, and epiretinal membranes. FTMH stage was defined according to standard biomicroscopic Gass staging, in which stage 1 is defined as an impending hole (not included in this study), stage 2 as an MLD <400 µm, stage 3 as an MLD ≥400 µm without a PVD, and stage 4 as an MLD ≥400 µm with a PVD. 21
For each patient, a standard 25-gauge PPV with ILM peeling and gas tamponade was performed by (or supervised by) a vitreoretinal surgeon who had more than 5 years of experience with the procedure. Postoperative 6-month BCVA and FTMH closure rates were recorded. ILM samples were stained using a vital membrane tissue stain, such as indocyanine green or brilliant blue. If an epiretinal membrane was present, it was peeled along with the ILM. The ILM was removed from the fovea and parafovea, but the final extent of the peel and whether stain was reapplied to verify complete removal of the ILM was determined at the discretion of the surgeon. No ILM flaps were utilized in any of the cases.
Linear regression models, adjusted for patients’ baseline lens status, duration of symptoms, and MLD, were used to analyze the impact of cystoid volume and hole symmetry ratios on preoperative BCVA, preoperative FTMH size, and lines of improvement after surgery. BCVA was measured with either pinhole correction or a manifest refraction. Since our study had a limited number of failed closures postsurgery, Wilcoxon rank-sum and Pearson’s chi-square tests were used to compare FTMH closure outcomes. For correlation analyses using linear regression models, beta coefficients with 95% confidence intervals (95% CIs) were reported, along with their P values. A 2-sided P value <.05 was considered statistically significant. Analyses were conducted using R Statistical Software (version 4.3.1; R Core Team 2023).
Results
Sixty-nine eyes of 69 patients with FTMH were included in this study. The mean (±SD) age was 69 ± 8.0 years, and 40 patients (58%) identified as female. The majority of eyes (67%) were phakic, while the remainder (33%) were pseudophakic; the lens status, treated as a covariate, was not significant during these analyses. An epiretinal membrane was present in 34 eyes (49%). The majority of patients (n = 48, 70%) had symptoms for <3 months, 11 patients (16%) had symptoms for 4–6 months, and 10 patients (14%) had symptoms for >6 months. In analyses comparing patients with a symptom duration of <3 months with patients in the other duration groups, we found that there were no statistically significant differences in either IRC volume or macular hole symmetry calculations. A summary of the baseline characteristics can be found in Table 1.
Baseline Characteristics of Patients With Full-Thickness Macular Hole (69 Eyes; 69 Patients).
Abbreviations: BCVA, best-corrected visual acuity; ERM, epiretinal membrane; IRC, intraretinal cystoid.
Six months after surgery.
Small = minimum linear distance (MLD) <250 µm; medium = MLD between 250 and 400 µm; large = MLD ≥400 µm.
Macular Hole Characteristics
In this cohort at baseline, the MLD was a mean (±SD) 361 ± 170 µm (range 56–698 µm). By the original Gass classification, 59.4% of the holes were stage 2 (<400 µm, without PVD), 24.6% were stage 3 (≥400 µm, without PVD), and 15.9% were stage 4 (≥400 µm, with PVD). The mean (±SD) preoperative logMAR BCVA was 0.69 ± 0.44, corresponding to 20/98 Snellen equivalent. At 6 months after surgery, the mean (±SD) logMAR BCVA was 0.39 ± 0.23, corresponding to 20/49 Snellen equivalent. To better understand the magnitude of the visual change after surgery, the number of lines of improvement was calculated. The mean (±SD) number of lines of improvement was 3.1 ± 4.7. The percentage of holes that closed after a single surgery was 96%.
Intraretinal Cystic Volume Correlation Analysis
The IRC volume varied widely among eyes. Overall, the mean (±SD) IRC volume was 0.118 ± 0.119 mm3, the median was 0.072 mm3, and the range was 0.00392 mm3 to 0.673 mm3. For stage 2 FTMH (small holes), the mean IRC volume was 0.087 mm3; for stage 3 FTMH (medium holes), the mean IRC volume was 0.147 mm3; and for stage 4 FTMH (large holes), the mean IRC volume was 0.188 mm3.
In regression analyses adjusted for baseline lens status, greater IRC volume was significantly associated with worse preoperative BCVA (beta = 0.0013 [95% CI, 0.0005 to 0.0021]; P = .003), larger FTMH MLD (beta = 0.714 [95% CI, 0.423 to 1.005]; P < .001), and greater BCVA improvement at 6 months postsurgery (beta = 0.013 [95% CI, 0.003 to 0.023]; P = .015). Additional adjustment for baseline MLD in the model weakened the correlation between IRC volume and preoperative BCVA (beta = 0.0006 [95% CI, −0.0003 to 0.0016]; P = .17) and the correlation between IRC volume and BCVA improvement at 6 months postsurgery (beta = 0.0048 [95% CI, −0.0065 to 0.016]; P = .40). The positive relationship between IRC volume and FTMH size (as represented by the MLD) is highlighted in Figure 2. IRC volume was not associated with the FTMH closure rate after surgery (P = .64). Figures 3A–D highlight a large FTMH and small FTMH and the central horizontal raster with B-scan, which was manually segmented.
Relationship between the preoperative intraretinal cystoid (IRC) volume and macular hole size (minimum linear distance [MLD]), determined in regression analyses adjusted for baseline lens status. The slope of the linear regression line indicates a positive association between these measures.
Examples of optical coherence tomography (OCT) B-scan images of a large full-thickness macular hole (FTMH) and small FTMH. (A) A large FTMH with (B) large intraretinal cystoid (IRC) spaces is highlighted. (C) A small FTMH with (D) small IRC volume is highlighted. OCT images in (B) and (D) were manually segmented by 3 graders. The purple delineates the retinal tissue, the teal delineates IRC spaces, and the dark purple delineates the macular hole itself.
Macular Hole Symmetry Correlation Analysis
Based on the OCT volumetric segmentation, analysis of FTMH symmetry was conducted at 3 levels (inner, base, and waist levels) and the mean (±SD) ratios for the 3 levels were calculated. The inner level ellipse ratio was 1.26 ± 0.30, the base level ellipse ratio was 1.22 ± 0.26, and the waist level ellipse ratio was 1.24 ± 0.29. For all 3 levels, the ellipse ratio was 1.24 ± 0.24. These ratios may be interpreted to indicate that mean values closest to 1 represent greater FTMH symmetry and mean values >1 represent less FTMH symmetry. Examples of a small, asymmetric hole and a large, symmetric hole are shown in Figure 4.
Examples of optical coherence tomography B-scan images of an asymmetric and a symmetric full-thickness macular hole (FTMH). (A) A small FTMH that is asymmetric is shown, with ellipse cross-sections at the level of the (B) internal limiting membrane (ILM), (C) waist, and (D) retinal pigment epithelium (RPE). (E) A large FTMH that is symmetric is shown, with ellipse cross-sections at the level of the (F) ILM, (G) waist, and (H) RPE. The blue line in each cross-section is the major axis, and the green line is the minor axis. The red dashed line shows the ellipse that was mapped onto the FTMH cross-section, which is displayed in yellow.
In correlation analyses adjusted for baseline lens status, greater FTMH asymmetry was associated with smaller MLD (beta = −194 [95% CI, −360 to −28.6]; P = .02), both at the inner level (beta = −156 [95% CI, −293 to −19.3]; P = .03) and at the waist level (beta = −191 [95% CI, −329 to −53.5]; P = .007). However, FTMH symmetry and MLD were not significantly correlated at the base level (P = .36), which is the level of the RPE.
In correlation analyses adjusted for baseline lens status and MLD, FTMH symmetry was not significantly associated with the preoperative BCVA (beta = 0.07 [95% CI, −0.356 to −0.496]; P = .74), nor was it associated with BCVA improvement at 6 months after surgery (beta = −0.68 [95% CI, −7.74 to −6.38; P = .85) or hole closure rates after surgery (P = .19).
Conclusions
Despite its increasing popularity in retinal diagnosis and therapy surveillance, OCT has not been completely harnessed to employ volumetric analysis to assess retinal features; current clinical practice is limited to retinal thickness measurements and the use of single cross-sectional OCT scans for measurements. Volumetric analysis is particularly underutilized in the analysis of FTMH, in which most physicians typically utilize the MLD of the single horizontal B-scan of the OCT. 1 This process neglects the remaining frames, which may contribute additional information about the scan, particularly as it relates to the 3-dimensional structure and characteristics of the retina and its pathology. By using in-depth segmentation of the entire OCT volume, this analysis revealed valuable insights related to the importance of IRC volume and macular hole symmetry in guiding macular hole prognosis. Larger IRC volume was associated with larger holes and worse preoperative BCVA. These holes with larger IRC volumes were also the most likely to have greater BCVA improvement after surgery. When using the volume to better understand the importance of symmetry, the analysis showed smaller holes were more asymmetric compared to larger holes. These findings point to the importance of considering both IRC volume and macular hole symmetry when discussing visual prognosis with patients and when planning for surgery.
Other studies have attempted to use the presence of IRCs to predict FTMH outcomes. Goto et al studied the characteristics of IRC volume of the horizontal and vertical raster and found that 95.5% of macular holes contained IRCs. 18 In their cohort, greater IRC area was associated with greater baseline macular hole size, but it was not associated with postoperative visual acuity. 18 An investigation by Tsuboi et al using volumetric analysis found increased inner layer IRC volume of FTMHs to be associated with decreased postsurgery visual acuity. 22 This seeming discrepancy with our findings is likely due to our decision to analyze correlations of the IRC volume with a change in BCVA rather than with the final BCVA. Being able to counsel the patient on how much visual improvement they may expect to experience after surgery is a useful parameter in clinical decision-making and may be more clinically relevant to the patient. In our study, greater IRC volume was associated with worse preoperative BCVA and greater BCVA improvement after surgery. Two important comments must be made regarding this association. First, the floor effect, where a better preoperative BCVA has decreased potential to improve due to fewer lines of improvement being available until the patient has reached a BCVA of 20/20 Snellen equivalent. Second, the MLD was also significantly associated with worse preoperative BCVA and greater BCVA improvement after surgery. When MLD was included in the model, it became the variable with the strongest association. This supports already published findings indicating that MLD plays an important role in predicting FTMH outcomes.23,24
Using volumetric analysis to analyze macular holes was previously attempted in a study by Geng et al. 11 They found a positive association between postsurgery visual acuity and the area ratio factor, which was defined as the body surface area divided by the bottom base area. They did not measure the IRC volume or symmetry of the FTMHs. Their conclusions parallel the previously described diameter hole index, which is the equivalent 2-dimensional ratio.7,10
This current analysis is unique in utilizing volumetric analysis to describe clinical factors associated with the symmetry of macular holes. Chen et al performed automated segmentation to describe FTMH symmetry and found that the MLD, determined only from the horizontal raster B-scan, may differ in the minimum and maximum dimensions by an average of 55 µm and up to 144 µm. 13 Such a difference could potentially reclassify an FTMH from a small hole to a large hole (thereby passing over the medium size) according to the International Vitreomacular Traction Study classification criteria. This further justifies the importance of volumetric analysis in the evaluation of macular holes, rather than cross-sectional measurements.
There are several theories for the presence of IRCs in FTMHs, including the RPE contact loss theory and damage to Müller cells.13,16,22 The contact loss theory posits that the separation of the RPE from the neurosensory retina alters the RPE’s normal function of fluid transport from the subretinal and intraretinal spaces. Müller cells are the main support cells of the retina and maintain retinal dehydration through ion transport. These theories support our finding that increased IRC volume was associated with worse preoperative BCVA.
FTMH symmetry may be affected by multiple factors. Ostensibly, the asymmetry of the macular hole stems from uneven anterior-posterior forces transferred by the posterior hyaloid. That is, if the posterior hyaloid is still attached at 1 edge of the macular hole but has been released from the other, it is reasonable to expect some asymmetry. However, this theory was rebutted by a cleverly designed study comparing the drawbridge elevation of asymmetric holes over time. 25 Those investigators instead introduced a theory that implicates Müller cells in the pathogenesis of FTMHs. Thus, the main cause of FTMH symmetry could be attributed to Müller cell traction. Other causes of asymmetry might be attributed to retina-RPE adhesions and scarring. For these reasons, the volumetric sum of IRCs and symmetry may be surrogate measures for Müller cell capacity. Therefore, IRCs and symmetry may be useful FTMH biomarkers.
Because FTMH asymmetry is associated with smaller FTMHs, we would expect more asymmetric holes to have higher rates of FTMH closure. This was not the case in our study. As previously mentioned, one reason could be that the sample size was limited in our study, which hindered our ability to draw conclusions regarding a finding such as FTMH closure, which only occurred in 4% of cases.
Duration of symptoms was included as a covariate in our analysis since longer symptom duration has been found to be associated with decreased FTMH closure rates and worsened visual acuity. 26 Our analysis did not show a statistically significant correlation between IRC volume or hole asymmetry and duration of symptoms. A definitive conclusion could not be made, as the majority of patients (70%) had symptoms for <3 months, which does not allow for a large enough sample in the other duration groups.
This study has several limitations. Due to the low incidence of surgical failure in our study sample, a larger sample size is needed to reveal if there are any meaningful relationships between IRC volume or symmetry and closure rates. Additionally, the correlation between IRC volume and macular hole size could also confound this association. Automated segmentation through machine learning may aid in developing a larger data set with which to draw more subtle findings. It helps alleviate the most time-consuming aspect of volumetric analysis in our study, the manual segmentation of each OCT scan. Automated segmentation may also improve the accuracy of the method when compared to human-directed segmentation, and would allow us to leverage a larger data set to glean insights. 13 Another limitation is the 2-month allowance between OCT imaging and surgical intervention. For some holes, this period could have allowed for evolution or changes to the hole that may affect outcomes. Fortunately, these cases comprised a small portion of the data set.
In this pilot study, multivariate analysis of volumetric measurements of IRCs and FTMH symmetry revealed important insights regarding the characteristics of FTMHs, particularly with regard to correlations with the preoperative IRC volume. As our results indicate, greater IRC volume was associated with greater FTMH size, worse preoperative BCVA, and greater BCVA improvement after surgery. Furthermore, greater FTMH asymmetry was associated with smaller hole sizes. Both of these associations were weakened once MLD was incorporated into the model, suggesting that MLD remains the most important OCT measurement in prognosticating FTMH outcomes. Future studies will build on these findings to analyze larger data sets for further volumetric insights on macular holes.
Footnotes
Authors’ Note
Presented at the annual meeting of the American Society of Retina Specialists, July 17–20, 2024, Stockholm, Sweden, and the annual meeting of the Retina Society, September 11–15, 2024, Lisbon, Portugal. Dr. Dinh and Howard Zhang contributed equally to this work.
Ethical Approval
Approval for the study was obtained from the Vanderbilt University Medical Center Institutional Review Board. All research adhered to the tenets of the Declaration of Helsinki. The collection and evaluation of all protected patient health information was performed in a Health Insurance Portability and Accountability Act–compliant manner.
Statement of Informed Consent
In this study of human subjects, the need for informed consent was waived.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported in part by funding from the Macula Society/Retinal Research Foundation grant.
