Choroidal nevi are the most common benign intraocular tumors, reported in large population studies to have anywhere between a 4% to 8% prevalence rate in the Caucasian population.1–4 These tumors are commonly diagnosed as incidental findings on fundoscopic examination, as almost all nevi are asymptomatic.5 Lesions are classically well-defined, round or oval, flat, pigmented or nonpigmented tumors that can be located anywhere in the fundus. The cumulative incidence of transformation to choroidal melanoma is 0.78% by the age of 80 years.6–11
Choroidal melanoma occurs in six to seven per 1 million white individuals and primarily affects fair-skinned patients in the fifth or sixth decade of life.6–11 These lesions are usually dome-shaped subretinal masses with variable pigment. The data from the various clinical studies, most notably the Collaborative Ocular Melanoma Study, have reported that mortality in these patients correlates with tumor thickness and diameter.12 However, about 30% to 50% of patients with melanomas will eventually have metastatic disease despite treatment with enucleation and brachytherapy. The mortality rate is much higher for a larger melanoma, suggesting that patients would benefit from early diagnosis.
Differentiation between small choroidal melanomas and choroidal nevi can be difficult. The diagnosis in patients with a small pigmented choroidal lesion is currently reliant on three imaging modalities: fundus photography, optical coherence tomography (OCT), and ultrasonography. Serial photographs of the lesion are used to monitor changes in size and appearance, whereas OCT can detect newfound intralesional details such as cavities, granularity, and an abnormal choriocapillaris.13–17 Unfortunately, even with the aid of these devices, delay in the diagnosis of choroidal melanomas still occurs and is associated with increased morbidity and mortality.4
Spectral-domain OCT angiography (SD-OCTA) is emerging as a novel imaging tool that can acquire volumetric angiographic information in three dimensions in a matter of seconds. The images are high-resolution and noninvasive and provide in vivo information about the vascular flow at a fixed point in time, providing structural and functional information in one scan without using intravenous dye. The ability to specifically image the choriocapillaris and choroidal vascular layers also makes OCTA appealing for evaluating pathology affecting these regions. In this report, we apply this technology to patients with choroidal nevi to correlate images obtained from color fundus photography with images from en face OCTA. We specifically aimed to evaluate the size of choroidal nevi, and (1) determine the inter-device variation in the size of choroidal nevi, and (2) determine the intra- and inter-rater reliability of nevi measurements using OCTA compared with fundus photographs. We also used en face OCTA to image the retinal microvasculature overlying nevi and evaluate any changes in the vascular plexi.
Patients and Methods
This retrospective review was approved by the Stanford University Institutional Review Board and included patients with choroidal nevi whose nevi had been imaged with both an OCTA and a fundus photography device. All OCTA images were performed using a SD-OCTA device (Carl Zeiss Meditec, Dublin, CA). This system had a scan speed of 67,000 kHz using a light source centered on 840 nm and a bandwidth of 45 nm. The tissue axial resolution was 5 μm and the transverse resolution was 15 μm. Scans were obtained of a 3 mm × 3 mm frame size. There were 350 x 350 A-scans per cube, and images had four-times oversampling. For patients whose nevi were larger than 3 mm in diameter, a 6 mm × 6 mm frame size was also obtained. Multiple images were obtained and repeated at the same location on the retina to reduce motion artifacts.
The automated segmentation algorithm available within the device software was used to locate the choriocapillaris and choroidal layers for the purpose of nevi measurement. Scans for which the automated segmentation could not accurately segment these layers were excluded. Deep and superficial retinal plexi were also assessed for microvascular abnormalities overlying the nevi. The boundaries of these layers were defined by the inner plexiform layer (IPL) (inferior boundary of the superficial layer) and the IPL/inner nuclear layer (INL) and INL/outer plexiform layer (OPL) borders (encompassed the deep layer). Images where automated segmentation varied from the above definitions were not assessed for abnormalities in these layers.
Fundus photography of each nevus was performed using either the Zeiss FF450 Plus fundus camera (Carl Zeiss Meditec, Dublin, CA), the Optos 200TX fundus camera (Optos, Dunfermline, United Kingdom), or in some cases, both devices. The shape of the nevus was characterized on the coregistered OCT B-scan as either plateau (scleral distension only), dome (retinal distension only), or almond (both retinal and scleral distention) (Figures 1A and 1D). Nevi were also defined as pigmented or nonpigmented by their appearance on fundus photography (Figures 1C and 1F).
Characterization of two choroidal nevi (A–C, D–F) on optical coherence tomography angiography (OCTA) and fundus photography. (A) Coregistered OCT B-scan showing an almond-shaped nevus. (B) 6 mm × 6 mm OCTA at the level of the choroid showing an area of reduced vascular density through the nevus shown in A. (C) Corresponding color fundus photo (Zeiss FF450) of the nevus shown in A and B. Both pigmented and nonpigmented regions are identified in this mixed nevus. (D) Coregistered OCT B-scan showing a dome-shaped nevus. (E) 3 mm × 3 mm OCTA at the level of the choroid showing an area of increased vascular density through the nevus shown in D. Increased flow (highlighted in red) corresponding to the choroidal nevus can also be seen in D. (F) Corresponding color fundus photo (Zeiss) of the nevus shown in D and E. A uniformly pigmented lesion can be identified.
This study investigated the inter-device variation in the size of choroidal nevi. Measurements of nevi were compared between OCTA images and each of two fundus photography devices (Zeiss and Optos). Subanalysis was also performed comparing all three devices. Nevus dimensions in the automatically segmented choriocapillaris and choroidal layers of the OCTA were assessed, and the layer in which the nevus was best visualized was used for OCTA measurement purposes. The greatest longitudinal dimension (GLD) of the nevi were determined using the built-in software caliper tool of each device (Figure 2). The maximum perpendicular dimension (PD) was measured using the same tool. Measurements were performed by two different assessors for OCTA and fundus camera devices, both of whom were masked to the measurements recorded from the alternative device. Two assessors (one for each modality) repeated the measurements on two occasions (on two separate days). These data were used to determine intra- and inter-rater reliability (repeatability and reproducibility, respectively). The ratio of GLD to PD (GLP/PD) as measured by each device was calculated for each nevus. Statistical analysis was performed using the SPSS version 20.0 (IBM, Armonk, NY). Ratios were compared using Wilcoxon signed-ranks and Friedman nonparametric tests. Intra- and inter-rater reliability was assessed using the intraclass correlation coefficient (ICC). P values of less than .05 were considered statistically significant.
Size of two choroidal nevi (A–C, D–F) using optical coherence tomography angiography (OCTA) and fundus photography. (A) Coregistered OCT B-scan showing a plateau-shaped nevus. Increased vascular flow can be seen through the choriocapillaris and choroid on cross-section. (B) 3 mm × 3 mm OCTA at the level of the choriocapillaris showing an area of increased vascular density through the nevus shown in A. Nevus measurements including greatest longitudinal dimension (GLD) and maximum perpendicular dimension (PD) (blue calipers). (C) Corresponding color fundus photo (Optos) of the pigmented nevus shown in A and B. GLD and PD (white calipers). The nevus borders are less well-defined on fundus photograph compared with OCTA. (D) Coregistered OCT B-scan showing a plateau-shaped nevus. Patchy areas of increased vascular flow can be seen through the choroid on cross-section. (E) 3 mm × 3 mm OCTA at the level of the choroid showing an area of increased vascular density through the nevus shown in D. GLD and PD (blue calipers). (F) Corresponding color fundus photo (Zeiss FF450) of the pigmented nevus shown in D and E. GLD and PD (white calipers). The nevus borders are less well-defined on fundus photograph compared with OCTA.
Measurements were performed for six pigmented, five nonpigmented, and three mixed nevi. Eleven of the 14 nevi were of plateau configuration. Nevi were best visualized on en face OCTA images in the choriocapillaris for eight patients and in the choroid for six patients. The characteristics and size of each nevus are presented in Table 1. The configuration of four of these nevi are described in Figures 1 and 2.
Characteristics and Nevus Size Measured on Three Devices
The absolute difference in GLD between the OCTA device and Zeiss FF450 fundus camera was not significantly different (mean difference −226.75; standard error of measurement = 270.66; P = .430). The absolute difference could not be compared against the Optos fundus camera due to the difference in unit measurement (microns versus pixels). Thus, the GLD/PD ratio of each nevus was used as a surrogate measure and compared between all three devices. Wilcoxon signed-ranks tests found no significant difference between the GLD/PD ratio of either of the device pairs (Zeiss FF450 fundus camera versus OCTA: Z = −1.96, P = .052; Optos fundus camera versus OCTA: Z = −1.6, P = .110; Optos fundus camera versus Zeiss FF450: Z = −0.135, P = .893). Of the 14 patients included, complete measurements from all three devices were available for five patients. Friedman test showed no significant difference between the GLD/PD ratio as measured by all three devices (Chi-square = 2.8; 2 df; P = .247).
Intra- and inter-rater reliability coefficients from ICCs are presented in Table 2. ICC coefficients were high for repeated measures and different investigators on all devices, suggesting good repeatability and reproducibility. The ICC for repeated OCTA measurements (0.920) was similar to both fundus photography devices (Zeiss FF450, 0.983; Optos, 0.991). Two nevi were poorly differentiated (Table 1; No. 5 and 11) in both choriocapillaris and choroidal en face OCTA images. The ICC for repeated OCTA measures with these two nevi excluded was 0.987 (95% CI, 0.958–0.996). The ICC for reproduced OCTA measurements (0.772) was high, although marginally lower than both fundus photography devices (Zeiss FF450, 0.966; Optos, 0.972).
Intra- and Inter-Rater Reliability Coefficients for Readings Performed on Two Separate Days
The deep and superficial retinal plexi were assessed in five patients with adequate segmentation. There were no changes in the appearance of the vasculature in these plexi overlying the nevi compared with the surrounding vascular architecture (Figure 3).
Optical coherence tomography angiography (OCTA) images of the patient shown in Figures 2A–2C. A 3 mm × 3 mm OCTA of the superficial (A) and deep (C) capillary plexi overlying the nevus. The superficial and deep plexi appear unremarkable. Motion artifact can be seen affecting a small region on the inferior part of the images. Coregistered OCT B-scans highlighting the level of the superficial (B) and deep (D) retinal OCTA (purple dashed lines).
This study investigates the clinical utility of en face OCTA in the evaluation of patients with choroidal nevi. OCTA has been shown to be advantageous in distinguishing between melanoma and nevi based on structural microvascular changes at the fovea.18 In particular, increased central macular thickness, decreased vascular density, and increased foveal avascular zone (FAZ) size are documented hallmarks of melanoma when compared with the patient's contralateral eye. Conversely, benign choroidal nevi have not shown distinct foveal changes on OCTA. FAZ size was not measured in the current study as the commercial software used did not have this feature available. However, the use of OCTA to evaluate nevus size and microvasculature directly overlying the nevus has not been previously documented in the literature. The measurement of nevi is currently limited to serial photographs, OCT and/or ultrasonography.
This study showed that choroidal nevi were consistently and reliably measurable on en face OCTA images of the choroid and/or choriocapillaris. Measurements with the investigational OCTA prototype and each of the two methods of color fundus photography were not statistically different based on GLD/PD. In addition, intra- and inter-rater reliability of OCTA measurements were also similar to that of the two fundus photography devices suggesting that OCTA may be an effective tool in the monitoring of choroidal nevi size.
Nevus morphology was characterized on the coregistered OCT B-scan accompanying each en face OCTA, with the majority of nevi causing no retinal distention (plateau shape). Increased vascular flow through the nevus on cross-sectional B-scan was also appreciated. In order to evaluate the size of the nevus, en face OCTA images at the level of the choriocapillaris and choroid were used. In general, the majority of nevi could be visualized through both layers, although generally borders were better preserved in one of the two layers. However, some could only be visualized in one of these layers, and two nevi were poorly differentiated in both choriocapillaris and choroidal en face OCTA images. The significance of nevus depth, extending into either the choriocapillaris, choroid, or both is unknown and warrants further investigation. This study suggests that longitudinal analysis of these nevi using OCTA could be useful in investigating risk for progression to melanoma based on nevus depth. No significant changes in microvasculature of the deep and superficial retinal plexi overlying the nevi was appreciated.
Compared with fundus photography, measurements performed on OCTA were less challenging in several respects. In the majority of cases, the outer border of the nevus was better defined on OCTA than on fundus photography, where borders were generally blurred. The ability to correlate the en face OCTA image with cross-sectional B-scan also allowed for confirmation of the nevus borders with flow through the nevus. However, OCTA was not helpful for the minority of nevi that could not be adequately visualized in either the choriocapillaris or choroid segmented layers. It is unclear why this effect was observed.
This study had several limitations that are important to address. This study was of low power due to small sample size and the use of two different fundus photography devices in addition to OCTA. Detection of additional differences or subtleties in measurements would require a larger study. In addition, the use of different imaging modalities limited the ability to directly compare nevus dimensions between each modality. Color fundus images showed a larger field of view, but the resolution of choroidal nevi edges was consequently more difficult to appreciate and measure. The Optos fundus camera's native caliper tool did not allow conversion of measurements in pixels to units of microns. Finally, the Zeiss FF450 fundus camera did not allow the user to zoom into lesions while performing the measurements, making it difficult to appreciate the edges of the nevus. Nevertheless, our findings of comparability between GLD/PD ratios between all devices, coupled with good intra- and inter-rater reliability for each device, show that these measurements of choroidal nevus dimensions are consistent within each device, including OCTA. Future studies may include a prospective study with longitudinal follow-up, if OCTA is to be used as a screening tool to monitor nevi size over time. With a higher-powered study, it may also be possible to perform subtype analysis investigating pigmented and nonpigmented nevi and the depth of involvement of the nevi on OCTA.
This is the first study to evaluate the appearance of choroidal nevi using en face OCTA imaging. We show preliminary data that OCTA does not significantly differ from standard fundus photography devices in measuring nevi diameter. Intra- and inter-rater reliability of nevi measurements on OCTA was also high. In addition, we were able to look at the superficial and deep retinal plexi to illustrate that there are no structural changes overlying nevi. These data suggest that OCTA imaging is no better than the current gold standard imaging for monitoring benign choroidal nevi over time.
- Ganley JP, Comstock GW. Benign nevi and malignant melanomas of the choroid. Am J Ophthalmol. 1973;76(1):19–25. doi:10.1016/0002-9394(73)90003-2 [CrossRef]
- Li HK, Shields CL, Mashayekhi A, et al. Giant choroidal nevus: Clinical features and natural course in 322 cases. Ophthalmology. 2010;117(2):324–333. doi:10.1016/j.ophtha.2009.07.006 [CrossRef]
- Shields CL, Furuta M, Mashayekhi A, et al. Clinical spectrum of choroidal nevi based on age at presentation in 3422 consecutive eyes. Ophthalmology. 2008;115(3):546–552.e2. doi:10.1016/j.ophtha.2007.07.009 [CrossRef]
- Sumich P, Mitchell P, Wang JJ. Choroidal nevi in a white population: The Blue Mountains Eye Study. Arch Ophthalmol. 1998;116(5):645–650. doi:10.1001/archopht.116.5.645 [CrossRef]
- Shields CL, Furuta M, Mashayekhi A, et al. Visual acuity in 3422 consecutive eyes with choroidal nevus. Arch Ophthalmol. 2007;125(11):1501–1507. doi:10.1001/archopht.125.11.1501 [CrossRef]
- Diener-West M, Hawkins BS, Markowitz JA, Schachat AP. A review of mortality from choroidal melanoma. II. A meta-analysis of 5-year mortality rates following enucleation, 1966 to 1988. Arch Ophthalmol. 1992;110(2):245–250. doi:10.1001/archopht.1992.01080140101036 [CrossRef]
- Murray TG, Sobrin L. The case for observational management of suspected small choroidal melanoma. Arch Ophthalmol. 2006;124(9):1342–1344. doi:10.1001/archopht.124.9.1342 [CrossRef]
- Shields CL, Shields JA, Kiratli H, De Potter P, Cater JR. Risk factors for growth and metastasis of small choroidal melanocytic lesions. Ophthalmology. 1995;102(9):1351–1361. doi:10.1016/S0161-6420(95)30864-0 [CrossRef]
- Shields CL, Cater JC, Shields JA, Singh AD, Santos MC, Carvalho C. Combination of clinical factors predictive of growth of small choroidal melanocytic tumors. Arch Ophthalmol. 2000;118(3):360–364. doi:10.1001/archopht.118.3.360 [CrossRef]
- Shields JA. Treating some small melanocytic choroidal lesions without waiting for growth. Arch Ophthalmol. 2006;124(9):1344–1346. doi:10.1001/archopht.124.9.1344 [CrossRef]
- Singh AD, Kalyani P, Topham A. Estimating the risk of malignant transformation of a choroidal nevus. Ophthalmology. 2005;112(10):1784–1789. doi:10.1016/j.ophtha.2005.06.011 [CrossRef]
- No authors listed. Factors predictive of growth and treatment of small choroidal melanoma: COMS report no. 5. The Collaborative Ocular Melanoma Study Group. Arch Ophthalmol. 1997;115(12):1537–1544. doi:10.1001/archopht.1997.01100160707007 [CrossRef]
- Espinoza G, Rosenblatt B, Harbour JW. Optical coherence tomography in the evaluation of retinal changes associated with suspicious choroidal melanocytic tumors. Am J Ophthalmol. 2004;137(1):90–95. doi:10.1016/S0002-9394(03)00868-7 [CrossRef]
- Francis JH, Pang CE, Abramson DH, et al. Swept-source optical coherence tomography features of choroidal nevi. Am J Ophthalmol. 2015;159(1):169–176. doi:10.1016/j.ajo.2014.10.011 [CrossRef]
- Muscat S, Parks S, Kemp E, Keating D. Secondary retinal changes associated with choroidal naevi and melanomas documented by optical coherence tomography. Br J Ophthalmol. 2004;88(1):120–124. doi:10.1136/bjo.88.1.120 [CrossRef]
- Shields CL, Mashayekhi A, Materin MA, et al. Optical coherence tomography of choroidal nevus in 120 patients. Retina. 2005;25(3):243–252. doi:10.1097/00006982-200504000-00001 [CrossRef]
- Spaide RF, Koizumi H, Pozzoni MC. Enhanced depth imaging spectral-domain optical coherence tomography. Am J Ophthalmol. 2008;146(4);496–500. doi:10.1016/j.ajo.2008.05.032 [CrossRef]
- Valverde-Megias A, Say EAT, Ferenczy SR, Shields CL. Differential macular features on optical coherence tomography angiography in eyes with choroidal nevus and melanoma. Retina. 2017;37(4):731–740. doi:10.1097/IAE.0000000000001233 [CrossRef]
Characteristics and Nevus Size Measured on Three Devices
|OCTA (μm)||FF450 (μm)||Optos (Pixels)|
Intra- and Inter-Rater Reliability Coefficients for Readings Performed on Two Separate Days
|Intra-Rater Reliability||Inter-Rater Reliability|
|Device||ICC Coefficient||95% CI||ICC Coefficient||95% CI|
|Zeiss FF450 Fundus Camera||0.983||0.933–0.996||0.966||0.859–0.992|
|Optos Fundus Camera||0.991||0.971–0.997||0.972||0.905–0.992|