The quantitative evaluation of retinal layers after resolution of subretinal fluid in acute central serous chorioretinopathy

Purpose To evaluate the average retinal layer thicknesses in eyes with unilateral acute central serous chorioretinopathy (CSC) (with subretinal fluid (SRF)) and after complete resolution of SRF in these eyes and to compare the results with those obtained in healthy eyes. Methods Fifty-four eyes of 27 patients with unilateral acute CSC (CSC in active phase) who had complete resolution of SRF and 25 eyes of 25 healthy control subjects enrolled in this retrospective study. The average thicknesses of the retinal layers were measured by segmentation analysis of optical coherence tomography at baseline and 6 months after complete resolution of SRF. Results The mean outer nuclear layer (ONL) thickness was significantly lower in eyes with CSC than in fellow eyes (p < 0.001). The mean ONL thickness was increased after resolution of SRF, but still low compared to unaffected fellow eye and the increment was not statistically significant (p > 0.05). There were significant strong inverse correlations between visual acuity and ONL thicknesses at baseline and 6 months after complete resolution of SRF (p < 0.001, r = − 0.810; p < 0.001, r = − 0.705, respectively). Conclusion ONL thickness was thinned in cases with acute CSC, and although there was some increment in ONL thickness 6 months after complete resolution of SRF, it was still thinner compared to unaffected fellow eyes.


Introduction
Central serous chorioretinopathy (CSC) is one of pachychoroid spectrum disorders which are characterised by increased choroidal thickness (diffuse or focal), dilation of outer choroidal vessels (pachyvessels), and attenuation and thinning of the choriocapillaris and the middle choroidal vessels overlying these pachyvessels. CSC is characterised by serous detachment of the neurosensory retina with dysfunction of the retinal pigment epithelium (RPE) and choroid. 1 In addition to the detachment of neurosensory retina, RPE detachments often occur in the disease. 2 The exact pathogenesis of CSC is not still fully understood, but choroidal abnormalities such as venous congestion, ischemia and/or inflammation lead to choroidal hyperpermeability, secondary RPE damage and serous detachment of neural retina. [3][4][5] Although CSC is self-limited in the majority of the patients, it often causes micropsia and metamorphopsia even if the patients have good visual acuity (VA). Moreover, especially in chronic and recurrent cases, permanent visual impairment may develop due to the persistent subretinal fluid (SRF) which results to RPE and photoreceptor damage. 6 Several studies have investigated the correlation between retinal morphological changes and visual outcomes in patients with CSC and reported that VA is correlated with outer nuclear layer (ONL) thickness. [7][8][9] Additionally, Eandi et al. 10 demonstrated an association between the lack of either external limiting membrane layer or photoreceptor bodies and outer segments beneath the fovea, and decreased VA in cases with resolved CSC.
Development of new software algorithms using spectraldomain optical coherence tomography (SD-OCT) allows manually or automated segmentation and thickness measurements of retinal layers. 11 This segmentation software is reliable to measure the retinal layers individually. 11,12 Prominent thinning of the fovea, particularly of ONL, in the cases with chronic CSC and changes of foveal photoreceptors including thinning of the ONL in eyes with acute and resolved CSC have been reported by previous studies. 7,9,[13][14][15] The aim of this study was to evaluate the average retinal layer thicknesses in eyes with unilateral acute CSC (CSC in active phase) at the time of diagnosis and after complete resolution of SRF and to compare these results with those of unaffected fellow eyes and healthy control eyes.

Study design and patient selection
The medical records of the patients with unilateral acute CSC, who had been referred the retina department of a tertiary eye hospital between January 2014 and May 2017, were consecutively recruited. The study protocol was approved by the Ethics Committee and the study was carried out in accordance with the Declaration of Helsinki. Each participant had a documented episode of acute CSC with serous retinal detachment in the macular region and complete resolution of SRF at follow up visits in one eye and no history of CSC diagnosed or suggested in the fellow eye by any history of visual loss, central scotoma or distortion. Patients who had presence of pigmentary changes on fundoscopy or fundus autofluorescence modifications suggestive of previous CSCR episodes were excluded from the study. Moreover, patients who first presented 20 days after the onset of symptoms were also excluded from the study. The diagnosis of acute CSC was made based on the presence of serous macular detachment on dilated fundus examination, a typical active angiographic leakage pattern (fokal dye leakage) on fundus fluorescein angiography (FFA, Heidelberg Engineering, Germany), SRF accumulation evident on SD-OCT (Heidelberg Engineering, Germany) and absence of drusen and/or choroidal neovascular membrane. Additionally, for the diagnosis of acute CSC, it was required that the SRF resolved within 6 months which was calculated from the onset of symptoms and did not show any recurrence for 6 months after resolution. Further, the complete resolved CSC was diagnosed based on the absence of SRF on SD-OCT and the absence of active leakage on FFA.
For this study, we included those eyes in which it was confirmed with OCT that serous retinal detachment associated with CSC affected the fovea at the initial visit, and those in which SRF resolved completely during the follow-up period without any intervention and/or treatment. Patients who had received any previous treatment including focal thermal laser photocoagulation, photodynamic therapy or intravitreal injection of anti-vascular endothelial growth factor medications for CSC or who had evidence of choroidal neovascularization, polypoidal choroidal vasculopathy, or other maculopathies including diabetic retinopathy, hypertensive retinopathy, age-related macular degeneration, pathological myopia on fundus examination, SD-OCT or FFA were excluded from the study. Chronic and recurrent cases or patients with earlier episodes of symptoms were also excluded.
Demographical data, the best corrected visual acuity (BCVA), Intraocular pressure (IOP) measurements, colour fundus photography, central macular thickness, macular optical coherence tomography scans, infrared reflectance images, fundus autofluorescence images and FFA findings were obtained from the patients' medical records.

Retinal layer segmentation and thickness measurement
First, spectralis automated segmentation software (Heidelberg Engineering), which automatically determines the internal and external boundaries of retinal layers, was activated in each case. Then, one experienced retinal specialist reviewed all B-scan frames for segmentation errors, which were described as misrepresentation of the representative lines in any part of at least one cross-sectional image. When necessary, manual adjustment was done starting from the innermost layer to the outer retinal layers. The average thicknesses of the retinal nerve fibre layer (RNFL), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), and ONL were measured in the 1-, 3-, and 6-mm circles of an Early Treatment Diabetic Retinopathy Study (ETDRS) macular map at baseline and 6 months after the complete resolution of SRF ( Figure 1). All measurements were performed by an experienced clinician, who was masked to the information about visual acuities of participants.

Statistical analysis
The study data were analyzed using the Statistical Package for Social Sciences (SPSS) version 22.0 for Windows (SPSS Inc., Chicago, IL). Descriptive data were presented as mean ± standard deviations, frequency distributions and percentages. The chi-square test was used in the analysis of categorical variables. Normal distribution of the variables was tested by analytical methods (Kolmogorov-Smirnov/ Shapiro-Wilk Test). Equality of variances was checked by the Levene test. The one-way analysis of variance, Welch analysis of variance, and Kruskal-Wallis tests were used to determine whether there was any significant difference between the three groups (affected eyes, fellow eyes and control eyes). Post hoc tests for pairwise comparisons were also done. Moreover, Pearson correlation tests were performed to investigate the correlations. The statistical significance was set at p < 0.05.

Results
Fifty-four eyes of 27 unilateral acute CSC patients (18 men and 9 women) with the mean age of 45.0 ± 7.7 years (range, 33-62 years) met the study inclusion criteria.
Twenty-five healthy control subjects (15 men and 10 women) with the mean age of 43.5 ± 8.1 years enrolled to the study. No statistically significant difference was observed between groups in terms of age and gender (p > 0.05, for each one). Demographic and some clinical characteristics of participants were illustrated in Table 1.
The mean BCVA at the baseline was 0.25 logMAR (range, 0.04-0.41 logMAR) for eyes with CSC and 0.00 logMAR for all of the fellow eyes and control eyes. The mean BCVA in the fellow eyes was better than that in the affected eye (p < 0.001). The mean BCVA at last visit was 0.16 logMAR (range, 0.00-0.31 logMAR) for eyes with CSC.
Comparison of average retinal layer thicknesses between the CSC and control groups at baseline was presented in Table 2. The mean ONL thickness was significantly lower in eyes with CSC than in fellow eyes (p < 0.001). Additionally, in eyes with CSC, the BCVA (LogMAR) at the baseline strongly correlated with ONL thickness (p < 0.001, r = −0.810) (Figure 2A).    The mean ONL thickness was 82.2 ± 19.8 μm at the time of the diagnosis and increased to 85.3 ± 20.8 μm 6 months after the complete resolution of SRF (Figure 3), but still low compared to uninvolved fellow eye. The BCVA (LogMAR) at the final visit strongly correlated with ONL thickness (p < 0.001, r = −0,705) ( Figure 2B).

Discussion
Microstructural changes in CSC have attracted much interest in recent years. And the improvements of technologies in the field of OCT has enabled reproducible, in-vivo, automatic and quantitative measurements of retinal layer thicknesses even in pathologic eyes. 16,17 In the present study of the OCT findings in unilateral resolved acute CSC, we found a decrement in the ONL thickness in the central 1-mm circle of the affected eyes. There might be possible explanations; first, the current understanding of the pathogenesis of CSC emphasises the role of the choroid, which is the main source of blood supply to the foveal avascular zone. And a great portion of retina in the foveal avascular zone is composed of the ONL. 18 Focal ischemia of the choriocapillaris may result in photoreceptor apoptosis leading to ONL thinning. Moreover, there is dual circulation and oxygen supply to the retina. The majority of the oxygen supplied to the retina comes from the choriocapillaris and diffuses into the outer retina, where it is consumed by the photoreceptors. There is either an oxygen tension minimum in the middle of the retina or an oxygen watershed between the inner and outer retina. [19][20][21] Therefore choroidal impairment leads to impaired nourishing of the outer retinal layers resulting thinning of ONL. Second, experimental animal models have shown that retinal detachment and reattachment can induce a variety of cellular changes that can decrease visual acuity. [22][23][24] Retinal detachment can cause photoreceptor cell loss, found as early as 1 to 3 days after retinal detachment. 25,26 Although the estimated onset of the CSC is based on the symptom perception of participant and is not completely definite, our results are consistent with experimental retinal detachment models in which the number of photoreceptor cells reduces within a few days, suggesting that ONL thinning may begin in the very early stage of CSC. 25,26 The baseline ONL thinning determined in our study can be explained by rapid affect of short-term retinal detachment. With reattachment, the retina can recover to some degree, but this recovery is accompanied by morphological abnormalities such as decreased number of cells in ONL. 27 Similarly in our study, the finding that the average ONL thickness increased 6 months after complete resolution of SRF, but still low compared to uninvolved fellow eye supporting that retina can recover to some degree. Third, Maruko et al. 28 reported that after reattachment, the ONL was thinner in eyes with CSC than in eyes with rhegmatogenous retinal detachment, although the duration of SRF did not differ between the two groups. This finding reveals that in CSC, not only detachment of neurosensorial retina from the RPE induces apoptosis of photoreceptors and also SRF, which is composed of leaking from the choroidal vascular system, may contribute to apoptosis. However, Goktas 29 investigated the correlation of SRF volume with choroidal thickness and macular volume in acute CSC and reported that SRF volume does not correlate with macular volume suggesting that formation of SRF is not solely responsible from changes in macular volume in acute CSC. Hata et al. 9 reported that ONL thickness starts to decrease with the occurrence of SRF, progresses until the resolution of SRF and showed no significant change for at least 6 months after complete resolution of SRF indicating that the loss of photoreceptors does not proceed after SRF resolution. However, Ersoz et al. 30 reported that decrease in ONL thickness starts with development of RPE alterations before SRF accumulation in pachychoroid pigment epitheliopathy. Similar to our results, Nakamura et al. 31 reported that the cone density remained lower after resolution of SRF in the eyes with CSC than in the normal eyes.
The ONL in affected eyes of patients with CSC was consistently thinner than in unaffected fellow eyes, despite the fact that unaffected fellow eyes may not always be completely normal; characteristic findings of CSC, such as hypoperfusion or hyperpermeability of the vessels in the choroid, can frequently be seen in both eyes. 32,33 For this reason we compared unaffected fellow eyes and healthy controls and we found that ONL thicknesses of eyes with CSC and unaffected fellow eyes were reduced compared to healthy control group, however the decrement was statistically significant only in eyes with CSC.
Previous OCT studies have shown that the thinning of ONL was related to poor visual outcomes. The ONL is composed of the nuclei of photoreceptors and its thinning reflects loss of photoreceptors. 34 According to a common assumption, ONL thinning in patients with CSC is a consequence of disturbed regeneration of photoreceptor outer segments due to the failure of phagocytosis. 7,9 Nakamura et al. 31 reported that the mean cone densities of eyes with CSC after SRF resolution were significantly decreased compared to the healthy control eyes. Similarly, a significant correlation of visual acuity and the ONL thickness was also observed in the current study at baseline and 6 months after complete resolution of SRF. Moreover, since the size and brightness of images determined by the number of stimulated cones, most of the patients reported micropsia and darkness after resolution of CSC despite good BCVA.
There were several limitations in our study, including the small number of cases, the short period of follow up, the retrospective nature of the study. Furthermore, only assessment of visual acuity is insufficient to assess macular function. Further studies are needed to investigate the relationship of foveal microstructural changes and retinal function.
In conclusion, ONL thickness was thinned in cases with acute CSC, and although there was some increase in ONL thickness 6 months after complete resolution of the SRF, it was still thinner compared to the unaffected fellow eyes. Additionally, ONL thickness was correlated with the BCVA in eyes with CSC before and after resolution of SRF.