Multimodal Imaging Evaluations of Focal Choroidal Excavations in Eyes with Central Serous Chorioretinopathy

Purpose. To investigate the prevalence and characteristics of focal choroidal excavation (FCE) concurrent with central serous chorioretinopathy (CSC) using multimodal imaging. Methods. This was a retrospective single-institution study. Clinical features and multimodal imaging findings were analyzed in eyes with CSC and FCEs, using imaging methods including optical coherence tomography (OCT), OCT angiography (OCTA), fluorescein angiography (FA), indocyanine green angiography (ICGA), fundus autofluorescence (FAF), and multispectral imaging. Results. Seventeen patients (4.8%) with 21 FCEs (19 eyes) were found among 351 consecutive Chinese patients with CSC. Chronic CSC represented 47.1% of those cases. Window defects in 12 lesions identified through FA and hypoautofluorescence in 13 lesions identified through FAF revealed retinal pigment epithelial attenuation. Choroidal hemodynamic disturbances characterized by localized filling defects at the excavation and circumferential hyperperfusion were validated by both ICGA and OCTA, which were similar to the angiographic features of normal chronic CSC. The hyperreflective tissue beneath FCE, observed on B-scan OCT, presented as intensive choroidal flow signals on OCTA. Conclusions. FCE is not uncommon in patients with CSC. Multimodal imaging suggested that the aberrant choroidal circulation might be a contribution factor for leakage from the dysfunctional retinal pigment epithelium at the area of excavation.


Introduction
Focal choroidal excavation (FCE) was first described by Jampol et al. in 2006, as an unusual choroidal excavation in the fovea with an intact overlying retina, detected by time-domain optical coherence tomography (OCT) [1]. Subsequent application of spectral-domain OCT (SD-OCT) has elucidated the clinical and more detailed morphological characteristics of FCE. These studies indicate that FCE is a choroidal abnormality without scleral involvement (absent of posterior staphyloma or scleral ectasia), which appears in two patterns: nonconforming and conforming [2][3][4][5][6][7][8][9][10][11]. Recent studies using swept-source OCT used deep wave imaging to better visualize the choroid and sclera in the area of FCE [12][13][14]. However, the etiology and clinical implications of FCE remain debatable. Most people consider FCE a congenital choroidal malformation, but a few acquired cases have been reported [3,9,[15][16][17][18]. Initially, FCE was considered a relatively stationary choroidal lesion and an incidental finding in patients with either no symptoms or a slight visual disturbance. Later studies have increasingly suggested a close association between FCE and multiple vision-threatening choroidal vascular diseases, including central serous chorioretinopathy (CSC), choroidal neovascularization (CNV), or idiopathic polypoidal choroidal vasculopathy (PCV) [2-11, 13, 14].
Several studies reported that the incidence of CSC in patients with FCE ranged from 4.8% to 24.4%, far higher than incidence of CSC in normal people [3, 7-10, 12, 19, 20]. And the incidence of FCE in patients with CSC was reported to range from 2.8% to 7.8% [4,5,14]. Choroidal vascular disturbances in eyes with FCE were observed in indocyanine green angiography (ICGA) images, such as focal choroidal filling defect, venous dilation, and choroidal hyperpermeability, which were also typical findings in CSC [10,19,21]. Despite a plausible close relationship, the role of FCE in the occurrence of CSC is still unconfirmed. Suzuki et al. postulated that choroidal circulatory disruption and 2 Journal of Ophthalmology atrophic RPE at the FCE lesion might be related to CSC complicated by FCE [5].
Advances in OCT technique always further our knowledge about FCE. Optical coherence tomography angiography (OCTA) is a new, noninvasive, dyeless, depth-resolved technique, which enables visualization of the choriocapillaris and retinal microvasculature by combining vascular flow visualization using motion contrast and structural information [22,23]. Hence, we supposed that OCTA could provide additional information about circulation in eyes with FCE concurrent with CSC. The aim of this study was to evaluate the prevalence, clinical, and multimodal imaging features of FCE concurrent with CSC in Chinese patients, which may therefore contribute to better understanding the underlying pathophysiology of the disease. The possible correlation between FCE and the pathogenesis of CSC has been studied with particular interest.

Materials and Methods
The Ethics Committee at the Second Affiliated Hospital of the Medical School of Zhejiang University approved this study, which was conducted in accordance with the tenets of the Declaration of Helsinki. We performed a retrospective review of 351 consecutive Chinese patients (mainland of China) with CSC who visited the eye center at the Second Affiliated Hospital of the Medical School of Zhejiang University between January 2014 and December 2015. Patients with other macular abnormalities such as CNV and angioid streaks and history of eye trauma, infection, inflammation, tumor, scleral staphyloma, or vitrectomy were excluded from the study. All study participants underwent comprehensive ocular examinations, including autorefractometry, best-corrected visual acuity (BCVA) measurement with a Snellen chart, slit-lamp biomicroscopy, and intraocular pressure measurement. Fundus examinations included indirect ophthalmoscopy, fundus photography (FP) (TRC-NW8F; Topcon Corp., Tokyo, Japan), fluorescein angiography (FA), ICGA, and fundus autofluorescence (FAF) imaging (HRA Spectralis; Heidelberg Engineering, Heidelberg, Germany). The SD-OCT images were obtained using Cirrus HD-OCT with enhanced depth imaging (EDI) OCT (Carl Zeiss Meditec, Dublin, CA, USA) or RTVue XR Avanti AngioVue OCT (Optovue, Inc., Fremont, CA, USA). Several patients underwent OCTA (OptoVue RTVue XR Avanti AngioVue; Optovue, Inc., Fremont, CA, USA) and multispectral imaging (MSI) examination (RHA; Annidis, Ottawa, Canada).
FCE was defined as a macular excavated area in the choroid, along the retinal pigment epithelium (RPE)/Bruch membrane complex line on OCT scans [14]. FCEs were classified as the "nonconforming type" if there was separation between the photoreceptor tips and the RPE and the "conforming type" if there was no separation [3,13]. If a nonconforming FCE converted to the conforming type after the resolution of subretinal fluid (SRF), it was defined as conforming type for statistical analyses. FCEs were also classified as foveal or extrafoveal depending on whether the foveal center was located within the excavated area [3].
Three-dimensional (3D) en face imaging at the RPE level was constructed to observe the stereo shape of FCE and avoid the possibility of missing multiple lesions.
The choroidal thickness and greatest width and depth of FCE were manually measured with the built-in caliper tool on SD-OCT. Subfoveal choroidal thickness (SFChT) was defined as the distance from the outer border of the RPE to the inner scleral surface. When FCE was located within the fovea center, SFChT was measured from the uninvolved adjacent RPE layer. Choroidal thickness under the excavation was measured from the outer border of the lowest tip of the excavated RPE layer to the inner scleral surface [3,14].
CSC was diagnosed on the basis of medical history, serous retinal detachment (SRD) detected through the fundus examination and OCT, and characteristic appearance detected through FA. Eyes showing only 1 or few specific angiographic leakage points at the RPE level were classified as classic CSC. Eyes with broad areas of granular hyperfluorescence detected through FA associated with indistinct areas of leakage were classified as chronic CSC [3].
We used a newly developed SD-OCT device (OptoVue RTVue XR Avanti AngioVue; Optovue, Inc., Fremont, CA, USA) to obtain split-spectrum amplitude decorrelation angiography (SSADA) images. SSADA is a clinically feasible, commercialized technique that can visualize capillary blood flow, but without determination of the flow direction [22]. The angiographic features at the FCE lesion and surrounding area were analyzed and compared with FA and ICGA images.
Multispectral imaging (MSI) is an improved technique using different light wavelengths from light-emitting diodes, ranging from 550 (green) to 850 nm (infrared), to examine the layers of the retina and choroid progressively [24]. MSI was performed in 4 patients to give a preliminary comparison with other fundus examinations.
The measured visual acuity was converted to a logarithm of the minimum angle of resolution (LogMAR) for statistical analysis. Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS 16.0; SPSS Inc., Chicago, IL, USA). Wilcoxon Signed Ranks test and Mann-Whitney test were used to compare continuous variables. Fisher's exact test was used to compare the categorical variables. Differences were considered statistically significant when values were less than 0.05.

Demographics and Clinical
Features. Seventeen patients (4.8%) with FCEs were found in 351 consecutive Chinese patients with CSC, including 11 males and 6 females (ratio of 1.83 : 1). Twenty-one FCEs were identified in 19 eyes (11 right/8 left). Two patients (11.8%) had bilateral FCE lesions, and 2 patients (11.8%) had 2 distinct lesions in the same eye. Seventeen eyes with 18 FCEs were concurrent with CSC, while the other 2 eyes with 3 FCEs were asymptomatic. Patients' demographic and clinical features are summarized in Table 1. The mean patient age was 44.3 ± 10.6 years (range, 21-65 years). The best-corrected visual acuities (BCVA) ranged from 20/100 to 20/20. The mean refractive error (spherical  Jan, 2015   1 and 2). FAF reflected the RPE status and indicated decreased RPE activity in most FCE lesions. FCEs were discernable in the MSI images ranging from 620 nm to 850 nm (available in 7 eyes). They appeared as well-defined black (3 eyes) or blurred grayish lesions (4 eyes), some of which were surrounded by dilated choroidal vessels (Figures 2 and 3). The Dec, 2015  shape of excavation in MSI images seemed different from that in the en face structural OCT images, but similar to that in ICGA images (Figure 2).

Spectral-Domain Optical Coherence Tomography Findings.
The 3D and en face OCT scans allowed us to detect FCEs in the macular area with no omitting and to clearly visualize their morphology. Overall, 12 FCEs (57.1%) were located in the foveal area and the remaining 9 FCEs (42.9%) in the extrafoveal area. The 3D reconstructed en face images of the RPE level showed that the excavation shape varied from small and cone-like to broad irregular excavation. We noticed umbilicate excavations with small centered bulges in 3 eyes, which appeared as two conterminous excavations in the transverse B-scan OCT (Figure 1

Optical Coherence Tomography Angiography Findings.
OCTA images were available in 11 eyes with 13 FCEs (9 patients). The superficial and deep retinal plexus seemed almost undisturbed (Figure 3). FCE appeared as a dark flow signal-absent area, surrounded by hyperperfused areas in the default choriocapillaris slab of OCTA, and as a hyporeflective area with circumferential hyperreflective RPE in the corresponding en face structural OCT, whether the FCE was conforming or not (Figures 1, 2, and 3). In some chronic cases, it was difficult to distinguish FCE from the various focal hypoperfused lesions on the angiographic image without the assistance of B-scan OCT (Figure 4). Interestingly, the hyperreflective tissue beneath FCE in B-scan OCT presented as a plaque of intensive flow signals in OCTA and as a hyperreflective plaque in the corresponding en face structural OCT (Figure 2).
A predominant advantage of OCTA was the visualization of choroidal neovascularization. One 65-year-old woman presented with a two-year history of chronic CSC and a persistent SRD. A distinct and continuous choroidal neovascular network superior to a FCE lesion was first noticed through OCTA at the level of the inner choroid [25]. It appeared similar to a discontinuous branching vascular network in the early-phase ICGA image (Figure 4) [25].
We noticed that FCE could be accurately localized according to the default en face structural OCT image of the outer retina slab, based on the distinct positional relationship between FCE and projection artifacts of retinal vessels. All 13 FCEs were located within the fovea-centered 3 mm × 3 mm square area, including 8 FCEs (61.5%) within the central 1 mm × 1 mm area ( Figure 5).
FCEs were more chiseled and discernable on ICGA images. Fourteen lesions in 12 eyes were hypofluorescent from early to late phase. Late-phase cloud-like hyperfluorescence secondary to choroidal vascular hyperpermeability was observed around FCE lesions in all eyes. Obvious choroidal vascular dilation was found in 5 eyes. In 2 patients with a lengthy history of chronic CSC, it was not easy to distinguish FCE lesions from other hypofluorescent lesions in a large area of mottled fluorescence.

Clinical
Course. The mean follow-up duration was 12.6 ± 7.4 months. Patients' chief complaints seemed not very different from normal CSC patients, including blurred vision, micropsia, or metamorphopsia. There were 5 eyes with conforming FCEs and 14 eyes with nonconforming FCEs at the first visit. In the 5 eyes with conforming FCEs, one patient presented with metamorphopsia and was diagnosed as classic CSC by FA and subsequent emerging of SRF (patient 1). Two other eyes had a history of resolved CSC (patient 14) and a recurrent CSC with SRD distant from FCE (patient 6), respectively. The remaining two eyes were detected by routine bilateral OCT examination.
SRD including the area of FCEs existed in all 14 eyes with nonconforming FCEs. With the resolution of SRD, FCEs converted from nonconforming to conforming in 7 eyes. Three patients underwent fundus laser photocoagulation and two patients underwent half-dose photodynamic therapy (PDT). FCEs in three patients converted from nonconforming to conforming and FCEs in the other two remained nonconforming. The treatment appeared to have a limited effect on the rate of conforming FCE, compared with the rate in the patients without treatment ( = 1.000).
Visual acuity of most patients was restored with the resolving of SRF and the restoration of the ellipsoid zone (EZ), but one patient's vision remained 20/30 as a result of a significantly thinning EZ. At the end of follow-up, the visual acuity of 12 eyes with conforming FCEs was significantly better than that of 7 eyes with nonconforming FCEs ( = 0.022). During the entire follow-up period, no new FCEs formed and no obvious change of the existing FCEs was observed.

Discussion
To our knowledge, our study may be the largest case series published to date on FCE concurrent with CSC patients. In our single-institution study, the prevalence of FCEs in CSC patients was 4.8%. The rate was lower than the 7.8% reported by Ellabban et al. and the 6.0% by Luk et al., but higher than the 2.8% by Suzuki et al. [4,5,14]. The different rates are possibly, at least in part, due to the small sample size and also related to the OCT scanning protocol used to detect the FCE. 3D scanning protocol proved to be more sensitive in detecting FCE compared with line scanning [7]. The mean patient age was 44.3 ± 10.6 years, which was consistent with the reported age of FCE presence in patients 40 to 50 years old, and also conformed to the average age of CSC onset (45 to 51 years) [10,20]. Although there was no sex predilection of FCE in previous studies, we found approximately twice as many males as females among our FCE/CSC patients (11 male/6 female). This ratio was surprisingly close to the previously reported male : female ratio of 2 : 1-2.5 : 1 in FCE/CSC patients [3][4][5]14]. It is reasonable that the obvious male preponderance was due to the intrinsic sex predilection of CSC (male : female ratios range from 2.7 : 1 to 7 : 1) [20]. All our patients were Chinese, which was consistent with the reported Asian preponderance (90%) of FCEs [3]. As for refraction, myopia was observed in 61.1% eyes. Whether moderate myopia is a specific character of FCE is controversial; at least it is not present in some FCE/CNV patients [2,11,13,14]. Therefore, the racial/ethnic, sex, and refraction predisposition still require further multicenter/race validation.
The different pathogenesis of the conforming and nonconforming types of FCE and the mechanism of their mutual conversion are still disputable. We observed two cases of conforming to nonconforming conversion due to the onset of CSC, rather than the self-progression of FCE. One presented with nonconforming FCE concurrent with CSC, whose OCT image, taken in another hospital 2 years ago, showed a conforming FCE at the same site (patient 9), as did patient 1 described in Figure 1. Interestingly, the lines of the external limiting membrane (ELM) and EZ at the detached area were always below the normal counterparts in patient 1, as if they were restricted by an uncertain force. Margolis et al. hypothesized that conforming FCE could progress to nonconforming FCE as stress on the outer retina resulted in separation of the photoreceptors tips from the apical surface of RPE [2]. The progression was much like CSC, and it was hard to make a distinction, as obvious leakage points could not be observed in many CSC cases.
Nonconforming to conforming conversion is more common, as seen in the resolution of CSC and CNV [3,6,7].
Lim et al. postulated that continued weakening of the choroid with age may lead to slow enlargement of the excavation and conversion from nonconforming FCE into conforming FCE [12]. It is plausible that the mutual conversion of the two FCE types is mostly dependent on the prognosis of complicated choroidal diseases, rather than the spontaneous progression of FCEs. In our study, FCEs in two eyes remained nonconforming despite the SRD regression, which had a small hyporeflective space and hyperreflective material between the photoreceptor tips and the underlying RPE, respectively (Figure 3). A possible reason for the former could be preexisting nonconforming FCE before CSC progression, or dysfunction of the pumping action resulting from RPE atrophy at the FCE lesion [3][4][5]. The hyperreflective material in the latter case was supposed to be unabsorbed subretinal deposits. Decreased elasticity and tension made the outer retina fail to attach to the deeply outpouching RPE layer.
It is well accepted that there is a significantly attenuating choroid beneath the excavation [14,21]. Nevertheless, whether the uninvolved choroid is thickened or not is controversial. Margolis [20,28]. Interestingly, bilateral FCEs were observed in patient 7 with thinner-than-average choroidal thickness (153 and 196 m) and typical choroidal hemodynamic features of FCE. We propose that the choroidal thickness may not be the key factor in the pathogenesis of FCE, but the disturbed choroidal circulation may be.
In our study, SFChT of eyes with hyperreflective tissue beneath the excavation was significantly less than that of eyes without hyperreflective tissue. It seemed that the existence of hyperreflective tissue would result in more striking RPE retraction and choroidal thinning. Ellabban et al. suspected that hyperreflective tissue represented focal scarring in the choroidal connective tissue, whereby subsequent scar contraction results in RPE retraction and FCE formation [14]. It was unexpected that hyperreflective tissue beneath the FCE presented as intensive flow signals in the choroid segment of OCTA. The nonvascular nature of the hyperreflective tissue was seemingly not supported by the OCTA findings. We speculated the hyperreflective tissue to be fibrovascular malformation in the choroid, which might be formed by abnormal vascularization in embryo period, or by subtle or subclinical choroidal inflammation [18]. These findings may be involved in the pathogenesis of FCE.
Multimodal angiographies furthered our knowledge on the circulation around the excavation. FA/ICGA imaged the flow of blood plasma or staining of vessel walls and tissue, independent of cell movement, while OCTA measured the blood flow dependent on intravascular cell movements. Therefore, these three techniques provided complementary circulation information [22]. Based on FA manifestation, chronic CSC represented 47.1% of FCE/CSC cases, which seemed remarkably higher than that reported in normal CSC patients (5%) [29]. The abnormal high proportion was also reported by Ellabban [3,14]. In our study, FCEs in 88.2% of patients were located within or adjacent to the area of fluorescein leakage, which demonstrated the close association between FCEs and RPE dysfunctional area. Window defects identified through FA in 57.1% lesions and hypoautofluorescence identified through FAF in 76.5% lesions revealed varying degrees of RPE attenuation and dysfunction. Suzuki et al. speculated choroidal ischemic changes might cause RPE atrophy at the FCE [5].
Localized filling defects that appeared at the excavation on ICGA seemed to be related to the obviously thinning choroid tissue beneath FCE and choroidal ischemic changes [4]. Choroidal vascular hyperpermeability around the FCE lesions revealed irregular hyperperfusion status of the circumferential choroid. The aberrant choroidal circulation was validated in more detail by OCTA as a localized flow signalabsent area with circumferential hyperperfused area in the choriocapillaris slab. Hemodynamic abnormalities of FCE on OCTA were highly consistent and colocated with those on ICGA, which suggested association with the formation of FCE [4,10]. Obata et al. suggested CSC shared highly similar characteristics with FCE in ICGA [10]. A recent study suggested foci of ischemia surrounded by reactive choroidal hyperperfusion were an important factor in the pathogenesis of normal chronic CSC, through a comparison among OCTA, ICGA, and FA [22]. The multimodal imaging findings were quite similar to our findings of FCE. Combined with the unexpected high proportion of chronic CSC in FCE/CSC patients, we supposed that the typical choroidal hemodynamic disturbances would also contribute to the leakage at the FCE site and the conversion from conforming to nonconforming FCE.
Our study was mainly limited by the small number of patients and the retrospective and cross-sectional data acquisition. We had data for only a few patients' FCE status before the onset of CSC. To further improve our knowledge of the progression of, and vascular changes in, FCE, prospective, longitudinal, and multicenter studies are required to monitor FCE during its natural course and when it is concurrent with CSC. A large-scale epidemiology study in normal asymptomatic people is needed to explicate whether the prevalence of FCE in CSC patients is equivalent to or higher than that in normal people. The fields of view on OCTA were considerably smaller than those currently used in ICGA or SD-OCT [22]; therefore, information outside the fields was missed.

Conclusions
In conclusion, FCE was a phonotype on OCT and diagnosed only by OCT. Although sex predilection of FCE was still not explicit, we found the male : female ratio seemed to be approximately 2 : 1 in FCE patients concurrent with CSC. The hyperreflective tissue beneath FCE might partly have a vascular nature based on the OCTA findings. Multimodal imaging depicted that the choroidal hemodynamic disturbances in the area of excavation were characterized by focal hypoperfusion with circumferential choroidal hyperperfusion, which were accompanied with disturbed RPE function. The similar angiographic features shared by both FCE and chronic CSC, the unusual high proportion of chronic CSC in FCE/CSC patients, and the close topographic relationship between FCE and leakage area all led us to speculate that choroidal thinning and aberrant choroidal circulation at the area of excavation might be causes for the leakage from the dysfunctional RPE, and thus for the conversion of FCE from conforming type to nonconforming type. OCTA showed significant textural changes of the choriocapillaris flow pattern in FCE/CSC cases, and further quantitative examination of choroidal blood flow would expand our knowledge on FCE.