Correlations between intraretinal cystoid cavities and pre- and postoperative characteristics of eyes after closure of idiopathic macular hole

Intraretinal cystoid cavities have been detected at the edges of macular holes (MHs) but their clinical characteristics and their relationship to the MH variables have not been determined. We measured the areas of the intraretinal cystoid cavity in 111 eyes with MHs in the OCT images preoperatively. Our results showed that the intraretinal cystoid cavities were located in the Henle fiber layer-outer nuclear layer (HFL-ONL) complex in 106 eyes and in the inner nuclear layer (INL) in 89 eyes. All were resolved after the initial vitrectomy to close the MH. The mean area of the cystoid cavity was greater in the HFL-ONL complex (55.9 ± 42.7 × 103 μm2) than in the INL (9.1 ± 9.8 × 103 μm2; P < 0.001). The area of the cystoid cavities was significantly correlated with the basal MH size (r = 0.465,P < 0.001), the external limiting membrane height (r = 0.793, P < 0.001), and the maximum retinal thickness (r = 0.757, P < 0.001). The area of the cystoid cavities was significantly correlated with the preoperative best-corrected visual acuity (BCVA; r = 0.361, P < 0.001), but not with the postoperative BCVA or the integrity of any of the outer retinal microstructural bands. The presence of intraretinal cystoid cavities was related to some morphological characteristics, but not to the postoperative BCVA or the restoration of the outer retinal bands.


correlations between area of intraretinal cystoid cavity and morphology of functional variables.
The area of cystoid cavity was significantly correlated with the basal MH size (r = 0.465, P < 0.001), with the difference between the basal and minimum MH size (r = 0.505, P < 0.001), with the vertical height of the central external limiting membrane (ELM) above the retinal pigment epithelium (ELM height) (r = 0.793, P < 0.001), with the maximum retinal thickness (r = 0.757, P < 0.001), with the length of the ELM defect (r = 0.298, P = 0.003), with the length of the ellipsoid zone (EZ) defect (r = 0.312, P = 0.002), and with the length of the cone interdigitation zone (CIZ) defect (r = 0.499, P < 0.001; Table 2 www.nature.com/scientificreports www.nature.com/scientificreports/ The area of the cystoid cavity was significantly correlated with the preoperative BCVA (r = 0.361, P < 0.001) but not with the postoperative BCVA (P > 0.05; Fig. 3). The correlations between the summed areas of the cystoid cavities and the postoperative integrity score of the ELM, EZ, and CIZ were not significant ( Table 3). The postoperative BCVA was significantly correlated with the preoperative BCVA (r = 0.421, P < 0.001), the basal MH size (r = 0.443, P < 0.001), the minimum MH size (r = 0.494, P < 0.001), the length of the ELM defect (r = 0.387, P < 0.001), the length of the EZ defect (r = 0.331, P < 0.001), and the length of the CIZ defect (r = 0.211, P = 0.035) ( Table 4).
There were significant differences between the area of the cystoid cavities, the maximum retinal thickness, and the ELM height, and the MH stages (all, P < 0.001; Table 5). The area of the cystoid cavity and the ELM height were significantly larger in the Stage 3 MH group (Fig. 4).

Discussion
Our results showed that intraretinal cystoid cavities were present in 95.5% of the eyes with an idiopathic MH, and none was present in all of the eyes after the surgery. The area of intraretinal cystoid cavity was significantly correlated with the preoperative BCVA, the basal MH size, the ELM height, and the maximum retinal thickness, but not with the postoperative BCVA. The MH in the Stage 3 group had the largest intraretinal cystoid cavity area,     www.nature.com/scientificreports www.nature.com/scientificreports/ The presence of cystoid cavities in 95.5% of the eyes preoperatively agrees with an earlier study that reported that intraretinal cystoid cavities were observed in 92.5% (37 of 40) of the eyes with a MH 7 . Our results showed that the cystoid cavities were located in both the HFL-ONL complex and the INL, and the area of the cystoid cavity in the HFL-ONL complex was significantly larger than in the INL. These results are also consistent with the results of two earlier studies 13,14 .
There have been many reports describing the relationship between the BCVA and the preoperative morphology determined by OCT in eyes with MH, e.g. the basal and minimum MH size 8,9,15 , the ELM height 16 , and the maximum retinal thickness 10 . Our results showed that the area of the cystoid cavities was correlated with the preoperative BCVA. In addition, the area was positively and significantly correlated with the difference between the basal and minimum MH size, the ELM height, and the maximum retinal thickness. In addition, eyes with larger cystoid cavity areas had greater differences between the basal and minimum MH size, higher ELM height, and longer maximum retinal thickness. These findings indicate that the retina at the edge of the MH is more elevated in eyes with larger intraretinal cystoid cavity area, and this may be related to the poor preoperative BCVA. On the other hand, the area of the cystoid cavities was not significantly correlated with the postoperative BCVA.
Sugiura et al. reported that the area of the intraretinal cystoid cavities were significantly correlated with the degree of postoperative metamorphopsia but not with the postoperative BCVA 13 . In addition, the areas of cystoid cavities were not significantly correlated with the integrity of the postoperative outer retinal layers. The microstructures of the outer retinal layers, viz., the ELM, EZ, and CIZ, represent the integrity of the photoreceptors, and the postoperative integrity of these layers have been identified as predictive markers of the visual outcomes in retinochoroidal diseases [17][18][19] and MHs 11,12,20 . These findings suggest that the presence of the intraretinal cystoid cavities do not affect the postoperative BCVA and the restoration of any of the outer retinal layers.
In contrast to the area of the cystoid cavity, the basal and minimum MH size, and the lengths of the ELM, EZ, and CIZ defects were significantly correlated with the postoperative BCVA in our results. These results are   www.nature.com/scientificreports www.nature.com/scientificreports/ consistent with the findings of earlier reports 16,21 . Several studies have shown that the length of the outer retinal layer defect can be a predicting factor for the postoperative vision because they are related to the MH size. On the other hand, the difference from the basal MH size to minimum MH size and the maximum retinal thickness were strongly correlated with the area of intraretinal cystoid cavities but were not significantly correlated with the postoperative BCVA. Taken together, we suggest that it is more likely that improvement of the BCVA was due to the restoration of the outer retinal layers. The degree of restoration would be more affected by the MH size and the length of the defect than the area of the intraretinal cystoid cavities. This would then account for the lack of significant correlations between the area of the intraretinal cystoid cavities and the postoperative BCVA.
A case series of serial OCTs and morphing videos demonstrated that the VMT was the initiating event for the formation of the intraretinal cystoid cavities 22 . At the early stages of MH formation, the anteroposterior tractional forces on the fovea is caused by the VMT, and foveal cystoid spaces are frequently observed 23 . Simpson et al. reported that the retinal stretching induced by the VMT can indeed result in a reduction in interstitial fluid pressure, with subsequent influx of water 24 .
It is believed that the cystoid spaces will expand and separate the retinal layer along with the forming of MH, and finally the intraretinal cystoid cavities are observed around the edges of the MH 25,26 . Several studies have reported on the relationships between the VMT and basal MH size and outer retinal elevation 27,28 . The basal MH size and the outer retinal elevation is increased along with the progression of primary MH due to the VMT.
There is another explanation for the formation of the intraretinal cystoid cavities. The leaked fluids may be poorly absorbed in the elevated outer retina because the retina at the fluid cuff is detached from the RPE which is responsible for the fluid transport toward the choriocapillaris.
Previous study combining the immunohistochemistry with en face OCT images demonstrated the features of the cystoid cavities 29   www.nature.com/scientificreports www.nature.com/scientificreports/ Gass classified MHs by the extent of the vitreomacular adhesion: Stage 2, partial detachment of the posterior hyaloid; Stage 3, total detachment of the posterior hyaloid but still attached to the optic disc; and Stage 4, complete detachment of the posterior vitreous 1 . In our results, the Stage 3 eyes had significantly larger cystoid areas and higher ELM height than Stage 2 or Stage 4 MH eyes. There were 3 eyes in the Stage 2 group that had no intraretinal cyst in contrast to none in the Stage 3 group. This suggests that the intraretinal cystoid cavities around the MH are not formed at the early period after the formation of MH even if the edge of the MH has a VMT. Our results support the idea that the intraretinal cystoid cavities resulted from VMT as a consequence of the MH formation. The Stage 3 MH group had the largest area of the cystoid cavity which implies that the metabolic functions of Müller cells may remain altered after the release of the VMT and the formation of a full-thickness MH. On the other hand, the area of the cystoid cavity and the ELM height in Stage 4 MH group were smaller than those in Stage 3 eyes. Yun and associates examined the morphologic characteristics of eyes with a chronic MH 30 . They reported that all of the eyes with a chronic MH were at Stage 4, and intraretinal cystoid cavities were observed less frequently in eyes with a chronic MH compared to eyes at the acute Stage 4 30 . Our results showed that 2 eyes in the Stage 4 group had no intraretinal cystoid cavity and the area of the intraretinal cystoid cavities were smaller. These finding corroborate their report.
There are limitations in this study. First, this was a retrospective study, and we did not evaluate the longitudinal changes of the OCT parameters including the intraretinal cystoid cavities. It was not determined when the intraretinal cystoid cavities developed and how they progressed over time. Second, the area of intraretinal cystoid cavity was measured manually in the OCT B-scan images. We measured the mean values of the area on the perpendicular images with OCT radial scans and confirmed that the horizontal and vertical images had the same cystoid cavities. However, some studies have evaluated the cystic areas in the en face OCT images, and similar measurements in our patients would probably lead to a more precise assessment of the cystic areas 29,31 . Third, the follow-up period was relatively short. Fourth, metamorphopsia and its relationship between the morphology of the MH were not evaluated. Further prospective longitudinal studies with automated calculations of the morphological changes and an evaluation of metamorphopsia will be necessary to validate the relationship between the intraretinal cystoid cavities and the pre-and postoperative BCVAs.
In conclusion, we found that intraretinal cystoid cavities were present in 95.5% of the eyes with an idiopathic MH before vitrectomy, and all were resolved after the vitrectomy. The presence of intraretinal cystoid cavity is correlated with the preoperative vision but not the postoperative vision and the restoration of any outer retinal microstructures.

Methods ethics statement.
This was a retrospective, cross sectional, single center study, and the procedures used were approved by the Ethics Committee of the Nagoya University Hospital, Nagoya, Japan. The study was performed at the Nagoya University Hospital, and the procedures used conformed to the tenets of the Declaration of Helsinki. Written informed consent was obtained from all patients.

Subjects.
We reviewed the medical records of patients with a unilateral, idiopathic MH who had undergone successful MH closure by vitrectomy at the Nagoya University Hospital from January 2013 to June 2018. The inclusion criteria were the availability of high quality spectral domain-OCT (SD-OCT) images (Spectralis; Heidelberg Engineering, Heidelberg, Germany) and the presence of a Gass Stage 2, 3, or 4 idiopathic full-thickness MH as determined by OCT. All patients underwent comprehensive ophthalmic examinations including measurements of the BCVA, intraocular pressure, and axial length. In addition, the retinas were examined by slit-lamp biomicroscopy, fundus photography, and OCT before and 12 weeks after the vitrectomy. The Snellen visual acuities were converted to the logMAR units to create a linear scale of the BCVAs.
The exclusion criteria were previous or coexisting ocular conditions, e.g., mature cataracts, high myopia (axial length >27.0 mm); traumatic MH, recurrent MH, other retinal disease, and prior vitrectomy.
Analyses of OCT images. The preoperative images of the retina obtained by SD-OCT B scans through the center of the MH were analyzed. All images were adjusted to correct for the difference of the vertical and horizontal ratios to be the same proportion. The basal MH size, the minimum MH size, the maximum retinal thickness (Fig. 1A), the ELM height, and the angle of the fluid cuff were measured in the vertical and horizontal OCT images (Fig. 1B). The average of two measurements was used for the statistical analyses. The distance between each edge of the outer retina was defined as the defect length, and the lengths of the ELM, EZ, and CIZ defect were measured in the image in which the basal MH size was the largest (Fig. 1C).
The area of the intraretinal cystoid cavity was measured with the ImageJ software (National Institutes of Health, Bethesda, Maryland, USA; available at https://imagej.nih.gov/ij/). The analysts were two of the authors (K.G., T.I.) who were masked to the other clinical information. The analyst traced the outline of the intraretinal cystoid cavities using the ImageJ program by freehand drawing and summed all of the values in one cross-sectional image passing through the fovea (Fig. 1D). If no intraretinal cystoid cavities were observed, the area of the intraretinal cystoid cavity was designated as zero.
The OCT images obtained at 12 weeks postoperatively were used to determine the integrity of ELM, EZ, and CIZ for a 1,000-mm diameter area surrounding the fovea. The microstructures were graded on a 4-point scale: 1, line not visible; 2, line disrupted by 0.200 mm; 3, line disrupted by >200 mm; and 4, continuous line 17 .

Surgical techniques.
A standard three-port pars plana vitrectomy was performed using the Alcon Constellation 23-or 25-gauge (G) system (Alcon Laboratories, Inc., Fort Worth, TX, USA). Core vitrectomy was performed, and if necessary, a posterior vitreous detachment was created by suction with a vitrectomy probe www.nature.com/scientificreports www.nature.com/scientificreports/ around the optic nerve head. Then, the ILM was peeled from the retina circumferentially with ILM-peeling forceps in all cases. A peripheral vitrectomy with shaving of the vitreous base was performed, and fluid-air exchange was done followed by the injection of 20% sulfur hexafluoride or 12% perfluoropropane into the vitreous to tamponade the retina. The patients were instructed to maintain a prone position until a closure of the MH was confirmed by OCT.
Cataract surgery was performed on all 95 phakic eyes and a foldable acrylic IOL was implanted into the capsular bag.

Statistical analyses.
The values are presented as the means ± standard deviations. Pearson correlation coefficients were calculated to determine the significance of the associations between the areas of the intraretinal cystoid cavity and other variables. One-way analysis of variance (ANOVA) was used to evaluate the differences in the variables among the MH stages. All statistical analyses were performed using the Statistical Package for Social Sciences for Windows 21.0 (SPSS Inc, Chicago, Illinois, USA).