Comparison of Macular Choroidal Thickness Among Patients Older Than Age 65 With Early Atrophic Age-Related Macular Degeneration and Normals

Abstract

Purpose.: To compare macular choroidal thickness between patients older than 65 years with early atrophic age-related macular degeneration (AMD) and normals.

Methods.: This was a consecutive, cross-sectional observational study. Enhanced depth imaging spectral-domain optical coherence tomography using horizontal raster scanning at 12 locations throughout the macula was performed in one eye of consecutive patients presenting with large soft drusen alone, drusen with additional features of early AMD, or a normal fundus. Choroidal thickness was measured at 7 points for each raster scan in the central 3 mm of the macula (total 84 points per eye). In addition, a single subfoveolar measurement was obtained for each eye.

Results.: One hundred fifty eyes of 150 patients were included. There was no significant difference between mean refractive error for each diagnosis category via one-way ANOVA (P = 0.451). Mean macular choroidal thickness (CT) was 235 ± 49 μm (range, 125–334 μm; median 222 μm) for normals, 161 ± 39 μm (range, 89–260 μm; median = 158 μm) for the drusen group, and 115 ± 40 μm (range, 22–256 μm; median = 112 μm) for patients with AMD. Mean macular CT was significantly different via one-way ANOVA among all diagnosis categories (P < 0.001).

Conclusions.: The presence of features of early AMD without geographic atrophy and/or soft drusen alone is associated with decreased mean macular CT in vivo compared to that in patients with no chorioretinal pathology. Using enhanced depth imaging, measurement of a single subfoveolar choroidal thickness is highly correlated to mean central macular CT.

Introduction

Age-related macular degeneration (AMD) is a common chorioretinal degenerative condition and is the most common cause of severe visual loss in adults older than age 65 in North America and Europe. ^1–3 Several known risk factors are associated with the development of AMD, including soft drusen, ² genetic factors, ⁴ smoking, and antioxidant supplementation. ^5–7 The natural history of the disease is well known, and several clinical variants and ranges of disease severity are recognized. ^2,8 Neovascular AMD consists of pathologic choroidal angiogenesis, which can be successfully treated with anti-vascular endothelial growth factor (VEGF) agents. It can be induced in experimental models by creation of a defect in Bruch's membrane or by upregulation of subretinal ⁹ or retinal pigment epithelial (RPE) VEGF production, which is an ischemia response element in vivo. Despite a vast body of laboratory and clinical research regarding various aspects of AMD, the precise pathomechanism underlying disease development and progression remains incompletely understood. 

Histologically, the spectrum of AMD includes basal laminar and linear drusen, RPE atrophy, and choroidal neovascularization, ^10–13 which occurs commonly in a sub-RPE location. ^14,15 With the development of spectral-domain optical coherence tomography (SD-OCT), cross-sectional chorioretinal anatomy has become visible in high resolution for the first time in vivo, and many reproducible aspects of choroidal morphology can be visualized by commercially available SD-OCT devices. ¹⁶ With the subsequent development of enhanced depth imaging (EDI), ¹⁷ macular choroidal thickness (CT) and morphology are visible along their entire extent, from Bruch's membrane to the inner scleral border. This has led to the recent observation that CT appears inversely correlated with advanced age ^18–20 and axial length, ^19,21,22 with a very thin choroid present in age-related choroidal atrophy (ARCA) ²³ and high myopia. ¹⁹ In addition, there appears to be a high degree of interocular symmetry regarding macular CT. ²⁴  

Many recent reports have discussed aspects of the choroid in AMD as imaged with SD-OCT, particularly with respect to CT. ^25–28 While an anatomically thin choroid is not necessarily indicative of decreased blood flow, ²⁹ there is some evidence that decreased regional choroidal perfusion is associated with drusen ³⁰ and AMD, ³¹ which may be related to choroidal blood volume, ³¹ flow, ³² and AMD progression. ^32,33 We hypothesized that the pathogenesis of AMD involves an underlying chorioretinal phenotype that is predisposed to accelerated atrophy compared to normal eyes. The purpose of the present study was to compare the central macular CT among age-matched patients with no chorioretinal pathology, soft drusen alone, or soft drusen with additional anatomic evidence of early AMD. 

Methods

This consecutive, cross-sectional, observational case series conformed to the tenets set forth in the Declaration of Helsinki and was performed in accordance with the Health Insurance Portability and Accountability Act of 1996. All patients were seen and examined by an experienced vitreoretinal specialist at the Charles Retina Institute, Memphis, Tennessee. The study was approved by the institutional review board of the University of Tennessee, Memphis, Health Sciences Center. 

Patients

Consecutive patients were recruited at a regularly scheduled office visit for EDI SD-OCT examination after meeting the following inclusion criteria: 

1.  
   Age greater than 65 years;
2.  
   No previous eye disease other than refractive error, cataract extraction with intact posterior capsule, drusen, or early AMD;
3.  
   No history of uncontrolled diabetes or uncontrolled hypertension;
4.  
   Spherical equivalent refraction (autorefraction either of phakic patients or measured prior to cataract extraction) ± 2.00 diopters;
5.  
   No anti-VEGF therapy in fellow eye for at least 3 months prior to SD-OCT;
6.  
   Intraocular pressure between 11 and 19 mm Hg as measured by Goldmann applanation tonometry on the day of SD-OCT examination; and
7.  
   Willingness to undergo informed consent for study participation.

The spectrum of early AMD included patients with soft drusen alone, as well as those with additional features of early AMD such as pigmentary derangement and subretinal drusenoid deposits. Patients were divided into three groups based on imaging characteristics (Fig. 2). The drusen group had the presence of multiple macular soft drusen greater than 125 μm without evidence of hyperpigmentation or RPE atrophy on fundus photography, no evidence of outer retinal or RPE atrophy on SD-OCT, and no additional features of AMD. The AMD group was defined by the presence of macular soft drusen with RPE hyper- or hypopigmentation without geographic atrophy as seen by ophthalmoscopy and fundus photography. Patients with subretinal drusenoid deposits and focal areas of outer retinal atrophy on SD-OCT were considered part of the atrophic AMD group. Patients with Snellen best-corrected visual acuity (BCVA) less than 20/200 in the study eye or areas of definite macular geographic atrophy on fundus photography were excluded. Patients in the drusen and AMD groups had to have had an available fluorescein angiogram (IVFA) within the past 6 months, and patients with any evidence of neovascularization in the study eye or transmission RPE defect indicating geographic atrophy were excluded. Patients with a history of choroidal neovascularization in the fellow eye were not specifically excluded. Thus, rather than using a traditional classification system, ^2,5,6,8 we used a modification combining fundus photographic, angiographic, and SD-OCT criteria. Photographic and OCT features between groups are shown in the Table. Normals were recruited from patients presenting for dilated fundus examination with a unilateral disease process or no eye disease. 

Table

 

Photographic and Optical Coherence Tomography Characteristics by Group of Patients Undergoing Macular Choroidal Thickness Evaluation

Group	Placoid Drusen > 125 μm	CNV Fellow Eye	Subretinal Drusenoid Deposits	Pigmentary Abnormality	Mean Age, y*	Mean Macular Choroidal Thickness, μm†
Normal, n = 51	0	0	0	0	77 ± 6.5	235 ± 49
Drusen, n = 39	39	9	0	0	79 ± 5.6	161 ± 39‡
AMD, n = 60	60	18	42	60	80 ± 6.4	115 ± 40‡

 CNV, choroidal neovascularization treated no less than 3 months earlier in patient's fellow eye; AMD, drusen with additional features of early age-related macular degeneration without geographic atrophy or CNV.

*  No significant difference among group means via one-way ANOVA (P = 0.163).

†  Measurements obtained via manual caliper tracings at 84 points per eye in central 3 mm of macula.

‡  Significantly less (P < 0.001) than in normal group (Dunnett's method) and from overall mean (one-way ANOVA).

Imaging

The SD-OCT protocol is shown in Figure 1. All imaging was performed by a single experienced ophthalmic photographer. Fundus photography (seven-field standard, Visucam Pro; Carl Zeiss Meditec, Dublin, CA) was obtained for all patients without fundus photography within the 3 months prior to SD-OCT. Spectralis (Heidelberg Engineering, Heidelberg, Germany) 12-line radial raster scan pattern was performed using the Enhanced Depth Imaging (EDI) function with 20 scans averaged per raster. Autorefraction was performed on the day of each SD-OCT and entered into the Spectralis patient data entry window prior to each scan. Computer randomization with equal allocation ratio was used to determine which eye would be included for patients with bilateral AMD or drusen. Grayscale brightness scan (b-scan) images were viewed on the device console. Four scans were initially excluded due to low signal strength secondary to media opacity. The device caliper tool was used to perform all measurements. Initially, a single subfoveal CT measurement was obtained for the horizontal raster from the outer border of the RPE to the inner scleral border. In the context of a subfoveal drusenoid RPE elevation, Bruch's membrane was used as the initial point, if visible. If Bruch's membrane was not visible in the context of a drusenoid RPE elevation on greater than 50% of the rasters (more than six for a given scan), the scan was excluded. Six scans were initially excluded for this reason. The caliper was then used to sequentially measure 500-μm distances in both radial directions, and a new measurement was obtained at each location. Choroidal thickness was measured as a caliper tracing from the outer border of RPE or Bruch's membrane to the inner scleral border. This was performed for three radial positions in each direction for each raster, for a total of seven choroidal measurements in each raster, encompassing the central 3 mm of the macula (84 measurements per scan). All values were then combined and averaged for the recorded overall macular CT. 

Figure 1. 

 

Choroidal thickness imaging protocol. (A) Fundus photography demonstrating soft drusen without hyperpigmentation or significant retinal pigment epithelial atrophy. (B) Scanning laser ophthalmoscopic (SLO) image of the enhanced depth spectral-domain optical coherence tomographic analysis; 12-line raster pattern (green tracing) centered on the fovea. (C) Choroidal thickness measurement; green vertical caliper tracings spaced 500 μm apart encompassing 3 mm centered on the foveola; this phakic, 78-year-old emmetropic patient in the soft drusen subgroup had an average macular choroidal thickness of 146 μm.

Figure 2. 

 

Photographic and optical coherence tomography (OCT) categories of patients undergoing choroidal thickness analysis. (A) Fundus photography and corresponding spectral-domain OCT of a normal 79-year-old; mean macular choroidal thickness is 231 μm. (B) Images of typical patients in the soft drusen category demonstrating large soft macular drusen without pigment alterations; mean macular choroidal thickness is 136 μm. (C) Images of typical atrophic age-related macular degeneration patient; in addition to macular soft drusen, hyperpigmentation is present in the macula, and OCT demonstrates drusen as well as numerous subretinal drusenoid deposits; mean macular choroidal thickness is 92 μm.

Data Analysis

A single observer (EJS), initially masked to previous diagnoses, screened scans for image quality and the absence of exclusion criteria. Patients were assigned to a diagnosis group based on imaging characteristics (Table). Power calculation (α = 0.05, Σ = 2) was performed for CT to detect a 30-μm (twice the minimum estimated axial scan resolution) difference between the three groups and indicated a 90% chance of discovery of a significant difference between means with a sample size of 108 (36 for each group). We chose a target sample size of 150 subjects to improve clinical significance and ensure sufficient sample size for each diagnosis category. Two raters, initially blinded to any previous diagnoses (EJS and JCR), performed independent CT measurements for both mean macular CT and a single subfoveal CT in a random order from the study image database. Means were compared among all groups using one-way analysis of variance (ANOVA). Additionally, two-sample t-test was used to compare CT in patients with any features of AMD (the drusen and AMD group combined) and normals. Pairwise comparisons were performed using Student's t-test and Dunnett's method. Bivariate fit-line analysis was used to compare CT with both age and spherical equivalent in two separate instances across all patients. Single subfoveal CT was plotted against mean macular CT via bivariate fit-line analysis. Interobserver correlation was evaluated by Spearman rank correlation. 

Results

One hundred fifty eyes of 150 patients were included. Normals consisted of 28 patients with epiretinal membrane in the fellow eye, 7 patients with choroidal nevus in the fellow eye, 12 patients with acute posterior vitreous detachment in the fellow eye, and 4 patients referred for retinal evaluation but without any vitreoretinal pathology. One hundred forty-two patients were Caucasian, and 8 patients were African American. Mean age ± standard deviation was 79 ± 6.4 years. Mean overall spherical equivalent was 0.042 ± 0.021 diopters. There was no significant difference between mean refractive error for each diagnosis category via one-way ANOVA (P = 0.451). Mean age by diagnosis category was as follows: normal = 77 ± 6.5 years (n = 51); drusen = 79 ± 5.9 years (n = 39); atrophic AMD = 80 ± 6.4 (n = 60). There was no significant difference between mean age among diagnosis categories via one-way ANOVA (P = 0.163). Advanced age was significantly correlated with decreased mean macular CT across all patients (P = 0.003) via bivariate fit-line analysis. 

Mean macular CT was not significantly correlated with spherical equivalent (P = 0.331) across all patients. Statistical analysis of mean macular CT is demonstrated in Figure 3. Mean macular CT for normals was 235 ± 49 μm (range, 125–334 μm; median 222 μm). Mean CT for the drusen group was 161 ± 39 μm (range, 89–260 μm; median = 158 μm). Mean macular CT was 115 ± 40 μm (range, 22–256 μm; median = 112 μm) for patients with atrophic AMD. Mean macular CT for the combined group of patients with AMD features (combined drusen and AMD group) was 168 μm (n = 99; range, 22–260 μm; median = 130 μm). Mean macular CT was significantly different via one-way ANOVA among all diagnosis categories (P < 0.001). Pairwise comparisons (Dunnett's method, normals as controls) demonstrated a significantly thinner (P < 0.001 for all comparisons) mean macular CT among patients with atrophic AMD compared to both normals (mean difference 121 ± 23 μm) and the drusen group (mean difference 74 ± 18 μm). For the single subfoveal CT, the means were as follows: normal = 243 ± 50 μm, drusen = 164 ± 39 μm, atrophic AMD = 120 ± 40 μm. Mean single subfoveal CT was significantly different via one-way ANOVA among all diagnosis categories (P < 0.001). Single subfoveal CT was highly correlated with mean macular CT via bivariate fit-line analysis (P < 0.001). Twenty-seven patients had a history of CNV in the fellow eye (9 in the drusen group and 18 in the AMD group). Decreased mean macular CT was associated with a history of CNV in the fellow eye via logistic regression (P < 0.001) across all patients as well as across patients with AMD features (drusen + AMD group) alone. Mean macular CT was significantly less in patients with a history of CNV in their fellow eye, compared to those without, via two-sample t-test (P < 0.001). There was high interobserver correlation for both mean macular CT (Spearman ρ = 0.98; P < 0.001) and single subfoveal CT (Spearman ρ = 0.9; P < 0.001). 

Figure 3. 

 

Statistical analysis of macular choroidal thickness. (A) One-way ANOVA of macular choroidal thickness (μm) between diagnosis categories; green diamonds centered on mean, with diamond width proportional to sample size and height spanning one standard deviation from mean; a significantly lower mean is present in both the drusen and atrophic age-related macular degeneration (AMD) subgroups. (B) Two-sample t-test of normals compared to the combined drusen and AMD group; a significantly lower mean choroidal thickness is present in the AMD feature group (P < 0.001). (C) Two-sample t-test comparison of macular choroidal thickness between patients with and without a history of choroidal neovascularization (CNV) in their fellow eye; a significantly thinner mean was observed in patients with a history of CNV in their fellow eye (P < 0.001). (D) Bivariate correlation of mean macular choroidal thickness and a single subfoveal value obtained at the horizontal meridian; the values are highly positively correlated (P < 0.001).

Discussion

The results of the present study indicate that macular CT is significantly reduced in patients with early AMD compared to normal healthy adults. This agrees with some previous reports indicating choroidal thinning in vivo in AMD. ^26,28 Additional previous investigations have found mixed results with respect to CT in AMD ^34–36 or no correlation between early AMD and CT. ²⁷ This is likely due to CT measurements limited to specific meridians, lack of control for age or axial length, and the inclusion of a wide range of disease severity. In the present study, we focused on the early disease process and made every effort to control for these variables, thus decreasing the likelihood of these factors confounding the data and results. 

We found that applying EDI protocol to a 12-line radial raster pattern with reduced scan averaging (20 instead of 100) was very effective in both visualizing the choroid and obtaining information about multiple macular meridians with an efficient acquisition time. In contrast to measuring one raster position, this allowed for the manual quantification of average central macular CT. There appears to be a strong correlation between this relatively tedious method of obtaining measurements and a single, subfoveal measurement. This may be an important observation for the assessment of AMD patients, as it implies a useful, practical modality for manual quantification of CT for clinical evaluation with SD-OCT. While EDI is a proprietary function (Heidelberg Engineering), additional OCT devices have been described for obtaining CT measurements, ^20,35–37 and there appears to be a high interdevice reproducibility. ¹⁶ As CT appears to be consistently thinner with advanced age, ^18–21,23 the penetration of many OCT scan protocols is likely sufficient to obtain a relatively accurate subfoveal choroidal measurement in elderly adults. 

While it appears that a thin choroid is associated with AMD, the clinical importance of this finding, as well as potential factors influencing the presence or development of a thin choroid, remains obscure. Patients developing AMD may begin with a relatively thin choroidal phenotype, which during normal ARCA becomes particularly thin. One potential pathomechanism for progressive choroidal atrophy involves choriocapillaris dropout secondary to the development of drusen and RPE atrophy. Drusen may physically separate RPE cells from their choroidal blood supply, and the loss of RPE cells and photoreceptors may decrease regional demand for blood flow, resulting in choroidal atrophy. ^38,39 In the present investigation, patients with subretinal drusenoid deposits and pigmentary abnormalities without geographic atrophy had a thinner choroid than those with soft drusen alone, which would support this potential sequence of events. One recent study investigating ophthalmoscopic characteristics and CT found features of AMD such as subretinal drusenoid deposits associated with a thinner choroid. ²⁶ Additionally, we found a high correlation between a thin choroid and history of CNV in fellow eyes. While this may be flawed by the lack of an ability to evaluate morphologic changes in the eye with a history of CNV prior to the development of disease, there is a high interocular correlation in macular CT, ²⁴ and this may therefore support the notion that a thin choroid may be a risk factor for the development of CNV. Additionally, a particularly thin choroid has been recently suggested to be a risk factor for the development of CNV in pathologic myopia, ⁴⁰ and may be a poor prognostic factor for the resolution of CNV following anti-VEGF therapy in this context. ⁴¹  

Histopathologic studies have indicated multiple changes in choroidal morphology and vascular density associated with AMD. ^11,12,42 There is some histologic evidence that large choroidal vessels are reduced in density in AMD, ¹² and that the ratio of choriocapillaris to larger choroidal vessels is increased. This finding may support the possibility of primary choroidal atrophy or morphologic alteration in AMD, as primary RPE dropout would likely lead to reduced choriocapillaris density. ^38,39 An additional hypothesis for the increased ratio of choriocapillaris to large vessels involves a subclinical neovascular “neochoriocapillaris” resulting from regional ischemia secondary to primary dropout of RPE and pre-existing choriocapillaris. ⁴³ In fact, while most aspects of histologically excised CNV demonstrate nonspecific wound repair-like granulation tissue, ¹⁴ neovascular fronds display fenestrated endothelium, ⁴⁴ which is similar to what is seen in normal choriocapillaris. Subclinical intrachoroidal neovascularization remains a possibility, as both IVFA ⁴⁴ and SD-OCT require the presence of subretinal or sub-RPE fluid or hemorrhage to be clinically apparent. Intrachoroidal neovascularization has been described in nonhuman experimental CNV models ^45,46 including those not involving a break in Bruch's membrane. ^9,46,47 Theoretically, this may occur without visual or clinical consequences unless a break in Bruch's membrane or persistent VEGF elevation induces sub-RPE or subretinal CNV formation. This process could add to CT prior to the development of visual loss or typical angiographic or OCT evidence of CNV, and both alter CT measurements in preclinical CNV and serve as a clinical marker for neovascular disease development and/or progression. Although we did not detect this effect in the present study, there was significant overlap between the range of CT in each diagnosis category, and no longitudinal analysis. 

The present study remains limited by the predominantly Caucasian study population, single-center location, single-observer quantification, and additional potentially unrecognized confounding variables. In addition, the cross-sectional design fails to examine longitudinal changes in CT in patients with features of AMD, which may be important in progression of disease. The standard deviation of the CT analysis was at times larger than the mean difference, and significant overlap in CT range occurred across groups, which may limit the use of CT as a clinical tool. There may be additional unrecognized confounding variables, such as concomitant vascular disease and history of smoking. We did not exclude patients with a history of CNV in the fellow eye, whom some investigators would consider high-risk for advanced AMD. Additionally, we used strict inclusion criteria based on both IVFA and SD-OCT, which may exclude some patients who exhibit findings on the spectrum of AMD, such as those with nodular (hard) drusen alone. This was strictly an anatomic study, without a formal analysis of visual function, which may have an important relationship with CT and otherwise unexplained visual impairment. ²³ We conclude that patients with soft drusen or additional features of atrophic macular degeneration have reduced central macular CT compared to age-matched controls. While the risk of development of late or neovascular AMD imparted by a thin choroid remains to be elucidated, a thin choroid appears to be associated with a history of CNV in fellow eyes of atrophic AMD patients. Future studies concerning the pathogenesis of AMD should consider macular CT as a potentially important clinical variable. 

Acknowledgments

Presented at the Association for Research in Vision and Ophthalmology/International Society for Imaging in the Eye imaging conference, Seattle, Washington, 4 May, 2013. 

Disclosure: E.J. Sigler, None; J.C. Randolph, None