Distinguishing Central Serous Chorioretinopathy From Neovascular Age-Related Macular Degeneration: A Prospective Study

Abstract

Purpose:

This article identifies clinical features that differentiate central serous chorioretinopathy (CSR) from neovascular age-related macular degeneration (nAMD) and uses this information to develop a diagnostic tool.

Methods:

A prospective observational study was conducted of patients with a new diagnosis of CSR, nAMD, or indeterminate presentation. All patients underwent clinical assessment, axial length measurement, enhanced-depth imaging–optical coherence tomography, and intravenous fluorescein angiography. A final consensus diagnosis was derived following review of these factors.

Results:

A total of 56 eyes of 56 patients were enrolled (CSR = 34; nAMD = 22). The subfoveal choroidal thickness was greater in the CSR group (421 ± 106 µm) than the nAMD group (219 ± 91 µm, P < .001). The following odds ratio of CSR reached statistical significance: age 70 and younger (72.00, 95% CI: 11.99-432.50), subfoveal choroidal thickness greater than or equal to 300 µm (33.92, 95% CI: 4.06-283.18), dome-shaped neurosensory detachment (13.24, 95% CI: 3.22-54.45), retinal pigment epithelial changes (0.31, 95% CI: 0.10-0.97), subretinal hyperreflective material (0.11, 95% CI: 0.03-0.42), and fibrovascular pigment epithelial detachment (0.05, 95% CI: 0.01-0.47). A stepwise CSR vs nAMD clinical decision-making algorithm is proposed.

Conclusions:

Choroidal thickness is increased in CSR when compared with nAMD. The presented odds ratios and the CSR vs nAMD clinical decision-making tool can be applied to distinguish CSR from nAMD.

Introduction

Central serous chorioretinopathy (CSR) and neovascular age-related macular degeneration (nAMD) are commonly encountered conditions with overlapping clinical presentations. ¹ Although they share common symptoms and clinical findings, they are distinct entities with different etiology, prognosis, and treatment. CSR tends to occur at a younger age with a mean age of 51 years, ² whereas AMD is more common with increasing age. ³

Recently the age range in patients with CSR has been shown to be broader than previously thought. ² Clinically, CSR is commonly characterized by a neurosensory retinal detachment with or without an associated pigmented epithelial detachment. Drusen and hemorrhage are lacking, whereas fluorescein angiography (FA) will often show a hot spot that may or may not demonstrate a smokestack configuration (Figure 1). CSR will often resolve spontaneously with preservation of vision. ⁴ Neovascularization from age-related macular degeneration may show different clinical features including drusen, retinal pigment epithelial (RPE) changes, fibrovascular pigment epithelial detachment (PED), and hemorrhage. FA most commonly shows either classic or occult neovascularization without a smokestack (Figure 2).

Figure 1.

Prototypical central serous chorioretinopathy. A 38-year-old man presented with metamorphopsia and a visual acuity of 20/25. (Top left) Fundoscopy revealed a dome-shaped neurosensory detachment. (Top right) Optical coherence tomography demonstrated a neurosensory retinal detachment and serous pigment epithelial detachment. (Bottom frames) There was an expansile dot pattern on angiography.

Figure 2.

Prototypical neovascular age-related macular degeneration. An 82-year-old man presented with blurry vision and 20/80 visual acuity. (Top left) There were clinically apparent retinal pigment epithelial changes and drusen. (Top right) Optical coherence tomography demonstrated subretinal hyperreflective material with subretinal fluid and (bottom frames) angiography confirmed classic choroidal neovascularization.

Choroidal imaging with enhanced-depth imaging–optical coherence tomography (EDI-OCT) is a clinical tool that may help differentiate CSR and nAMD. Patients with CSR commonly have increased choroidal thickness in affected as well as unaffected eyes. ^{5 -7} Eyes with nAMD do not demonstrate increased choroidal thickness. ^{8 -10}

Unfortunately, not all cases of CSR and nAMD fall within these prototypical categories or are easily identified on initial presentation (Figure 3). CSR can be associated with choroidal neovascular membrane (CNV), and nAMD can present with a dome-shaped neurosensory detachment (DSND) without hemorrhage or drusen. ^11,12 This variation in clinical appearance can make a specific diagnosis difficult, thereby leaving the clinician unsure what treatment is best.

Figure 3.

Indeterminate clinical presentation. A 71-year-old woman presented with a central blur and a visual acuity of 20/80. (Top left) There were no clinically apparent drusen. (Top right) Optical coherence tomography did demonstrate shallow subretinal fluid and some subtle drusen. (Bottom frames) Subsequent angiography demonstrated an occult choroidal neovascular membrane.

The purpose of this study is to identify the frequency of features associated with CSR and nAMD and to develop a diagnostic tool to assist the clinician in making a more definitive diagnosis for indeterminate cases.

Methods

Patients

Consecutive patients with a new diagnosis of CSR, nAMD, or indeterminate presentation were prospectively enrolled at a group retina practice in Edmonton, Canada, between June 2015 and November 2016. Patients with a previous diagnosis of CSR or nAMD and those unable to perform the required testing were excluded. Only 1 eye from each patient was enrolled.

Measurement and Interpretation of Parameters

All patients underwent a comprehensive ophthalmic evaluation supplemented with measurement of axial length, EDI-OCT, and intravenous FA (IVFA). Axial length measurement was obtained using a Zeiss IOL Master (Carl Zeiss Meditech). Standardized EDI-OCT (central 20° by 15°, 19 scans, 50 frames/scan) was performed with Heidelberg Spectralis (Heidelberg Engineering). IVFA was performed using an Optos 200Tx system. A consensus diagnosis was derived following a review of all available clinical and diagnostic information by 2 retina specialists and a retina fellow.

Presence or absence of drusen was determined using fundus photographs (as opposed to OCT). Gravity-dependent fluid was defined as asymmetric gravity-dependent retinal and RPE changes evidenced by OCT, color photographs, and/or autofluorescence imaging. RPE tears were identified using color photographs in conjunction with OCT and IVFA. EDI-OCT variables analyzed included intraretinal cysts, RPE atrophy, fibrovascular PED, serous PED, subfoveal choroidal thickness (SFCT), subretinal fluid, and subretinal hyperreflective material (SRHM). SFCT was measured by a single observer measuring from the border of the RPE to the inner scleral border using 1:1 μ images at the fovea. This technique has previously demonstrated high interobserver and intraobserver correlation. ^13,14 Choroidal thickness decreases with axial myopia. ¹⁵ To account for this, choroidal thickness was corrected for axial length based on previously reported correction factors. ¹⁶

The pattern and location of leakage were noted by IVFA. It was noted whether the leakage was focal or multifocal and whether it was juxtafoveal (< 200 µm), extrafoveal (200-1000 µm), or extramacular (> 1000 µm).

Statistics

A Microsoft Excel spreadsheet was used to compile the data, and SAS version 9.3 (SAS Institute Inc) was used for all data analyses. For univariate analyses, simple logistic regression was used to compare variables between groups. For continuous variables, thresholds based on distribution and a priori considerations were used to stratify the data and then calculate odds ratios. To create the decision tool, variables that demonstrated a statistically significant difference between groups were considered. Variables that most sharply divided the groups on univariate analyses were selected for the tool. This was then conducted again on the subgroup of variables remaining until the smallest possible number of residual eyes remained.

Results

Enrollment and Demographic Features

A total of 56 eyes of 56 patients were enrolled. CSR was diagnosed in 34 eyes (60.7%) and nAMD was diagnosed in 22 eyes (39.3%). CSR patients were younger (50.4 ± 11.4 years) than nAMD patients (74.4 ± 9.3 years; P = .0001). Although there were proportionally more men in the CSR (70.6%) than in the nAMD group (50.0%), this difference was not statistically significant (P = .124).

Clinical and Fundus Features

Clinical features results are summarized in Table 1. Patients in the CSR group had better mean visual acuity (Snellen equivalent 20/44) compared with the nAMD group (Snellen, 20/73; P = .032). DSND was more common in CSR (P < .001). RPE changes were more common in the nAMD group (P = .041). Hemorrhage and RPE tears were only present in the nAMD group (P = .027 and P = .210 respectively).

Table 1.

Demographic, Clinical, and Enhanced Depth Imaging—Optical Coherence Tomography Features.^a

		CSR	nAMD	P
Demographics	Mean ageb, y	50.4 (± 11.4)	74.4 (± 9.3)	< .01
	Male	70.6%	50.0%	.12
	Dome-shaped neurosensory detachmentb	67.7%	13.6%	< .01
Clinical features	Drusenb	0%	63.6%	< .01
	Gravity-dependent fluid	29.4%	9.0%	.07
	Hemorrhageb	0%	13.6%	.03
	RPE changesb	23.5%	50.0%	.04
	RPE tear	0%	4.6%	.21
	Visual acuityb, logMAR	0.35 (± 0.25)	0.56 (± 0.41)	.03
EDI-OCT features	SFCTb, µm	421 (± 106)	219 (± 91)	< .01
	SFCTb, corrected, µm	417 (± 161)	215 (± 135)	< .01
	Intraretinal cysts	11.1%	27.7%	.14
	PED	35.5%	59.0%	.08
	PED, serous	32.4%	18.2%	.24
	PED, fibrovascularb	2.9%	36.4%	< .01
	RPE atrophy	14.7%	27.3%	.25
	Subretinal fluid	94.1%	81.8%	.15
	Subretinal hyperreflective materialb	11.1%	54.2%	< .01

Abbreviations: CSR, central serous chorioretinopathy; EDI-OCT, enhanced-depth imaging–optical coherence tomography; logMAR, logarithm of the minimum angle of resolution; nAMD, neovascular age-related macular degeneration; PED, pigment epithelial detachment; RPE, retinal pigment epithelium; SFCT, subfoveal choroidal thickness.

^aDistribution of specified demographic, clinical, and EDI-OCT features both in the CSR and nAMD groups along with corresponding P values.

^b Values that reached statistical significance (P < .05).

Enhanced-Depth Imaging–Optical Coherence Tomography Features

Mean SFCT was thicker in eyes with CSR (421 ± 106 µm) when compared with nAMD (219 ± 91 µm, P < .001). This effect persisted when the SFCT was adjusted for axial length (CSR, 417 ± 161 µm; nAMD, 215 ± 135 µm; P < .001). Fibrovascular PED and SRHM were more common in the nAMD group (P = .001 for both variables). The other EDI-OCT features results are summarized in Table 1.

Angiography

An expansile dot pattern was seen only in CSR eyes (n = 30, 88.2%; P < .0001), and 46.6% of these (n = 14) had multiple expansile dots. Similarly, a smokestack pattern (n = 3, 8.8%), diffuse leakage (n = 3, 8.8%), and a PED pattern (n = 2, 5.9%) were observed only in CSR eyes. Classic CNV leakage was observed in 2.9% of CSR eyes, in comparison to 18.2% of nAMD eyes (P = .051). In contrast, occult CNV (n = 11, 50.0%; P < .001) and mixed CNV (n = 5; 22.7%; P = .004) leakage patterns were present only in the nAMD group.

Odds Ratios

Table 2 depicts the odds of an eye having CSR for each of the considered risk factors. The odds of CSR were higher for younger patients (age < 70 years), in eyes with thick choroid (≥ 300 µm), and DSND. In contrast, the odds of CSR were lower in eyes with RPE changes, SRHM, and fibrovascular PED. Odds ratios were not calculated for angiography features because they were considered part of the criterion standard in diagnosis. The odds ratios for drusen, hemorrhage, and RPE tears were undefined because these were present exclusively in the AMD group.

Table 2.

Odds Ratios of Central Serous Chorioretinopathy.^a

	Odds Ratio	95% CI
Age ≤ 70 yb	72.00	11.99-432.50
SFCT ≥ 300 µmb	33.92	4.06-283.18
SFCT, corrected ≥ 300 µmb	17.36	4.43-68.01
Dome-shaped neurosensory detachmentb	13.24	3.22-54.45
Gravity-dependent fluid	4.17	0.82-21.26
Subretinal fluid	3.56	0.59-21.36
Male sex	2.40	0.79-7.32
Visual acuity better than 20/100	2.30	0.46-11.44
PED, serous	2.15	0.59-7.90
RPE atrophy	0.46	0.12-1.75
PED	0.38	0.13-1.14
Intraretinal cysts	0.36	0.09-1.45
RPE changesb	0.31	0.10-0.97
Subretinal hyperreflective materialb	0.11	0.03-0.42
PED, fibrovascularb	0.05	0.01-0.47

Abbreviations: CSR, central serous chorioretinopathy, PED, pigment epithelial detachment; RPE, retinal pigment epithelium; SFCT, subfoveal choroidal thickness.

^aOdds ratios of CSR for the specified demographic, clinical, and enhanced-depth imaging–optical coherence tomography features in descending order. The corresponding 95% CI and P values are specified.

^b Features that reached statistical significance.

Clinical Decision-Making Tool

A clinical decision-making algorithm (CSR vs nAMD) is presented in Figure 4. When faced with CSR or nAMD patients, this tool provides a stepwise decision-making algorithm to assist with diagnosis. The tool incorporates features that demonstrate statistically significant differences between AMD and CSR eyes, allowing a clinician to have an ordered algorithm by which to sort patients into each diagnostic category. Using Optos features and patient demographics alone sorts approximately two-thirds of eyes, and the remainder can be sorted using OCT (when looking for PED) and FA (to find occult CNV). Of note, of the initially enrolled patients, 13 eyes were felt by the enrolling retina specialist to be of indeterminate presentation. When the clinical decision-making tool was applied to this subset, it accurately identified the final consensus diagnosis in all 13 eyes.

Figure 4.

Central serous chorioretinopathy (CSR) vs neovascular age-related macular degeneration (nAMD) clinical decision-making tool shows our proposed stepwise decision-making algorithm for diagnosis of nAMD or CSR. Retinal pigment epithelial (RPE) changes are based on fundoscopic appearance, pigment epithelial detachment (PED) is based on optical coherence tomography, and occult choroidal neovascular membrane (CNV) is based on fluorescein angiography. The P values for each decision variable and the number of identified patients for each diagnosis are specified in parentheses.

Conclusions

The purpose of this study was the development of a diagnostic tool to minimize misdiagnosis of CSR and nAMD in atypical cases and thereby reduce the risk of ineffective therapeutic intervention. Prototypical CSR or nAMD as depicted earlier (Figures 1 and 2) do not present a diagnostic dilemma. Appropriate diagnosis can be more difficult for atypical examples of these 2 disease entities (Figure 3) and has implications for treatment. Photodynamic therapy has been found to stop leakage and improve final visual acuity when compared with observation or intravitreal antivascular endothelial growth factor therapy for CSR. ⁴ Intravitreal antivascular endothelial growth factor therapy improves vision and reduces leakage more effectively than photodynamic therapy for the treatment of nAMD. ³

To our knowledge this is the first study to prospectively investigate distinguishing characteristics between CSR and nAMD. It is also the first study to confirm prospectively that SFCT determined by EDI-OCT is increased in CSR compared with nAMD. The presented clinical odds ratios (Table 2) and the CSR vs AMD clinical decision-making tool (Figure 4) should assist the clinician in appropriately diagnosing atypical CSR in a setting of possible AMD. The odds ratios provide quantitative value for considered diagnostic features. Clinicians can weigh and apply these accordingly in atypical presentations to reach a diagnosis. Of particular value are the features on either end of Table 2, with young age (≤ 70 years), thick choroid (SFCT ≥ 300 µm), and DSND strongly favoring CSR, whereas RPE changes, SRHM, and fibrovascular PED strongly favor nAMD. The clinical decision-making tool presents a simple stepwise decision algorithm starting with clinical factors and moving on to OCT and IVFA features if needed.

This study highlights the difficulty in arriving at a specific diagnosis when faced with an eye with indeterminate presentation and provides a potential solution. We hope that the clinical odds ratios and clinical decision-making tool will assist clinicians in making the correct diagnosis and thereby facilitate optimal treatment. Furthermore, the presented tools may be of value for ophthalmologists in training.

Limitations of our study included low prevalence of chronic CSR, absence of indocyanine green angiography (ICGA) testing, and absence of OCT angiography (OCT-A) results. This study recruited patients with newly diagnosed CSR, nAMD, or an intermediate presentation only, likely reducing the prevalence of chronic CSR in our study population. Chronic CSR is more likely to be associated with RPE changes or secondary CNV and therefore may be more challenging to distinguish from nAMD. ⁷

Our study included slightly more patients with CSR than nAMD. Potential factors contributing to this were new nAMD patients who required timely treatment, and prioritizing this left less attention and time for research recruitment. Further, CSR patients tended be younger and more willing and able to participate. Unfortunately, although ICGA and OCT-A are valuable tools in CSR, we were unable to include them in our study protocol because of availability and logistical barriers. ICGA can help demonstrate the extent of choroidal pathology and provide further clues regarding other differential diagnoses. OCT-A provides further information regarding choroidal perfusion abnormalities and assessment of choroidal neovascularization. Although we were unable to include these in our study, we felt we were able to reach an accurate consensus diagnosis, particularly because the majority of our cases were typical CSR or nAMD and did not represent a diagnostic dilemma. Accordingly, we felt we were able to achieve our purpose of comparing diagnostic features between the groups.

Next steps include prospective validation of the clinical decision-making tool. Specifically, distinguishing indeterminate cases of nAMD presents the greatest challenge. The algorithm performed well in this subset and warrants investigation in a prospective fashion.

We hope that the clinical odds ratios and the CSR vs nAMD clinical decision-making tool will assist the busy clinician in correctly identifying atypical CSR from nAMD before treatment.