Fundus autofluorescence and optical coherence tomography findings in choroidal melanocytic lesions.

PURPOSE: To establish the characteristics of secondary retinal and retinal pigment epithelial (RPE) changes associated with the presence of choroidal melanoma and choroidal nevus as documented by optical coherence tomography (OCT) and fundus autofluorescence (FAF).
MATERIALS AND METHODS: PubMed review of major English publications examining the correlation between clinical characteristics of choroidal melanoma and nevus with OCT and FAF findings.
RESULTS: The intrinsic properties of choroidal melanoma, as well as overlying RPE changes, drusen, and lipofuscin are best characterized by FAF, while OCT is more sensitive for the identification of subretinal and intraretinal fluid as well as atrophy, degeneration, and photoreceptor loss in the neurosensory retina.
CONCLUSIONS: Secondary retinal changes associated with choroidal melanocytic lesions can be documented by OCT and FAF. OCT-evident changes are observed more often with choroidal melanoma than choroidal nevus. OCT is better suited to identify the overlying retinal detachment and edema, even before these findings are clinically apparent. FAF is most useful in documenting the presence of lipofuscin, a finding that represents one of the important criteria in differentiating small choroidal melanoma from benign choroidal nevus.

INTRODUCTION

The relative differentiation between small choroidal melanoma and large choroidal nevus can be challenging. In such cases, quantitative factors (largest basal diameter and height) and qualitative prognostic factors (drusen, retinal pigment epithelium (RPE) changes, orange pigment, and subretinal fluid) become important criteria in estimating the probability of growth. These qualitative prognostic factors represent secondary changes at the level of neurosensory retina and RPE, some of which can be induced by the presence of choroidal melanoma. Newer imaging devices such as optical coherence tomography (OCT) and fundus autofluorescence (FAF) may help characterize these changes and also detect them at subclinical level.

The criteria to predict the behavioral pattern of pigmented choroidal lesions have been developed by Shields and associates.12 Thickness of the lesion, presence of subretinal fluid, symptoms, orange pigment, distance from the margins to the optic disc of <3 mm, ultrasound hollowness, and absence of drusen are used for early detection of small melanomas. FAF and OCT are relatively new imaging technologies that may help with early diagnosis and treatment, with the possibility of affecting survival and decrease risk for metastasis and visual loss.

FAF is based on the intrinsic property of certain molecules to show transient emission of light, or fluorescence, when illuminated by an external light source. In the eye, many tissues have autofluorescence properties such as the cornea, lens, and the RPE.

Modified imaging systems have been developed to detect autofluorescence of the fundus in both normal and pathologic states. The main source of FAF is lipofuscin. Lipofuscin accumulates in the RPE from incomplete degradation and digestion of photoreceptor outer segments. Lipofuscin is a mixture of proteins, lipids, and small chromophores. Secondary to decreased or impaired lysosomal activity, the lipofuscin accumulates in the RPE. The intensity of FAF depends mainly on the amount and concentration of lipofuscin. Depending on the intensity of autofluorescence, a lesion may be isoautofluorescent, hypoautofluorescent, or hyperautofluorescent.

The intrinsic properties of melanoma, RPE changes and the presence of drusen and lipofuscin are better characterized by FAF, while the OCT is more sensitive for the identifying the subretinal and intraretinal fluid as well as changes in the neurosensory retina [Figure 1 and Table 1].

Figure 1

(A) Choroidal nevus with juxtapapillary location with multiple drusen and retinal pigment epithelial changes (B) found on autofluorescence (C) and no signs of subretinal fluid on optical coherence tomography

Table 1

Fundus autofluorescence and optical coherence tomography findings in choroidal melanocytic lesions

Clinical findings	Fundus autofluorescence	Optical coherence tomography
Choroidal changes induced by the presence of melanoma	Iso/hyper/hypo autofluorescence	Iso/hypo/hyper reflectivity
RPE changes:	– hypoautofluorescence for RPE atrophy and hyperplasia	– thickening
– hyperplasia	– iso or mild hyperautofluorescence for fibrous metaplasia	– hyperreflectivity
– fibrous metaplasia		– fragmentation
– atropy		– RPE irregularities
RPE detachment:	Hyperautofluorescence	RPE detachment
– serous
– choroidal neovascular membrane
Subretinal fluid	Hyperautofluorescence	Hypo-reflectivity of subretinal fluid
Drusen	Central hypo-surrounded by hyper-autofluorescence	Hyper-reflectivity
Orange pigment	Intense hyper-autofluorescence	None
Neurosensory retinal changes:	None	Thinning of the photoreceptor layer
– photoreceptor atrophy		Intraretinal cystic spaces
– intraretinal fluid
Vitreoretinal changes:	None	Flattening of the foveal contour
– epiretinal membrane
– vitreous detachment

RPE: Retinal pigment epithelium

AUTOFLUORESCENCE PROPERTIES OF CHOROIDAL MELANOCYTIC LESIONS

Autofluorescence induced by choroidal melanoma

Shields et al.3 have studied the intrinsic autofluorescence properties of choroidal melanoma. Using the standard technique of 580 nm excitation and 695 nm barrier filter on a digital fundus camera, they found that choroidal melanoma has mild autofluorescent properties. In their analysis of 51 eyes with melanoma, they found choroidal melanoma displayed hypoautofluorescence (39%), isoautofluorescence (6%), and hyperautofluorescence (55%). The autofluorescence was found to be increased in larger tumors, pigmented tumors and those with disrupted overlying RPE [Figure 2]. The autofluorescent property of melanoma was of a granular quality in 92% of cases and nongranular appearance in 8%.3

Figure 2

(A) Subfoveal choroidal melanoma (B) with prominent orange pigment seen on autofluorescence (C) and subtle subretinal fluid seen on optical coherence tomography

FAF of choroidal melanoma was studied by Lavinsky et al.4 using scanning laser ophthalmoscopy with 488 nm excitation and 500 nm barrier filter, but there were no observable findings within the tumor.

Lohmann et al. studied the endogenous fluorescence of ocular malignant melanomas and found that lipofuscin granules were cleaved off and broken into small remnants in the melanoma. Several areas of increased fluorescence were seen throughout the RPE, as well as in underlying macrophages within the tumor.5

Farina et al.6 evaluated patients with pigmented cutaneous lesions using selected wavelengths between 420 and 1040 nm. They found that the ideal wavelength was 940 nm for discriminating melanoma from nevus and 578 nm for outlining tumor margins. Near-infrared wavelengths (780–820 nm) penetrate deeper into the choroid than shorter wavelengths and can excite melanin-associated fluorophores.7 This may play a future role in investigating the autofluorescence properties of choroidal melanocytic lesions.

RPE changes are better characterized with FAF than by OCT

OCT may show thickening, thinning, or irregularities at the level of RPE hyper-reflective layer.

RPE hyperplasia and metaplasia

Gunduz et al.8 noticed in six (75%) of eight amelanotic choroidal melanocytic lesions, increased FAF intensities in areas of hyperpigmentation. They postulated that this might be due to accumulation of lipofuscin in areas of hyperpigmentation, as noted before in studies on age-related macular degeneration (AMD).910

Lavinsky et al.4 observed that areas of increased pigmentation and fibrous metaplasia showed increased AF in medium and large tumors. OCT may show RPE fibrous metaplasia appearing as a focal thickening of the RPE.11

RPE atrophy

FAF findings in patients with geographic atrophy secondary to AMD result in markedly reduced signal over atrophic areas, due to decreased RPE lipofuscin, which is the dominant fluorophore for FAF imaging. The adjacent zones surrounding atrophic patches show high-intensity FAF. Similarly, choroidal nevi may present hypoautofluorescence associated with chronic RPE degenerative features and RPE atrophy.12 Choroidal melanoma usually displays more intrinsic overall autofluorescence when chronic RPE changes are noted.3 It is similar to a window defect allowing more of the intrinsic autofluorescence to be detectable.

Drusen

The presence of drusen counts as a negative correlation factor for growth in the case of choroidal melanocytic lesions. Drusen may have an increased, normal, or decreased FAF signal.13 The variable autofluorescence characteristics are related to the nature of the RPE change overlying the drusen. Some drusen may present with normal or near normal autofluorescence, and some others may have decreased autofluorescence. Large soft foveal drusen have increased autofluorescence.

Delori et al. found that drusen may present as a central area of decreased autofluorescence surrounded by a ring of increased autofluorescence.14 The authors of this study have postulated that the decreased central autofluorescence in drusen is determined by the RPE being stretched over the drusen, with a thinner layer of lipofuscin granules over the drusen.

Gunduz et al. found only partial correlation between increased FAF and drusen overlying choroidal melanocytic lesions.15 Karadimas and Bouzas,16 demonstrated that, in cases of drusenoid pigment epithelium detachment, FAF patterns vary from hyperautofluorescence to hypoautofluorescence. In these cases, the hyperautofluorescence may be caused by different fluorophores inside the fluid or drusenoid detachment, rather than by the lipofuscin.16

Lipofuscin

The lipofuscin overlying small choroidal melanomas is brightly hyperautofluorescent. Surrounding the bright lipofuscin hyperautofluorescence, there are adjacent dark hypoautofluorescent areas. This may possibly represent shifting and clumping of RPE cells with absence of RPE in between or possibly RPE atrophy. Detection of subclinical lipofuscin (orange pigment) by FAF imaging could play a role in the early detection of choroidal melanoma.17 Lavinsky et al.4 reported similar observations using the scanning laser ophthalmoscope (excitation, 488 nm; barrier filter, 500 nm).

Subretinal and intraretinal fluid

The presence of subretinal fluid represents a physical barrier between the outer segments of photoreceptors and the RPE, preventing normal phagocytosis of the shed outer segments of the photoreceptors. These subsequently accumulate on the outer retinal surface and in the subretinal space and represent a source of autofluorescence.1819

Histologically, the debris in the subretinal fluid consist of rounded or elongated densely packed multilamellar bodies that appear to result from the breakdown of outer segment disks. The retina anterior to the subretinal material shows attenuation of the photoreceptors layer. The subretinal material is bounded posteriorly by the RPE.20

Shields et al.21 examined choroidal nevi, hemangiomas, and melanomas with overlying orange pigment and subretinal fluid and they found that the pigment corresponded to lipofuscin in macrophages in the subretinal space.

Takagi et al.22 described similar findings for the retinal detachments associated with intraocular tumors. The subretinal precipitates were actually foam cells located around degenerated outer and inner segments of photoreceptor cells. These foam cells contained lysosomes with lipofuscin, and melanin.

In the analysis by Shields and colleagues on autofluorescence features of choroidal melanoma in 51 patients, 46 demonstrated active subretinal fluid, the fluid rim appeared with slightly more hyperautofluorescence than the fluid center.3

There are certain limiting factors for FAF imaging. First, media opacities, including lens opacification, decrease the FAF image contrast so that analysis of the image is not possible. Second, camera adjustment is crucial and motion artifacts should be excluded, since this may result in uneven background FAF and may lead to misinterpretation. Third, an absolute quantification of FAF images is not possible. This is probably because macular pigments vary with age, and a normative FAF imaging database is not available. The lack of reproducibility and consistency remains a major problem in many FAF studies.23

OCT FINDINGS OF CHOROIDAL MELANOCYTIC LESIONS

OCT findings associated with choroidal melanoma and nevi can be divided in three different patterns: serous retinal detachments around and overlying the tumor, intra-retinal cystic spaces in the overlying retina and loss of normal retinal architecture overlying the tumor [Figure 3].

Figure 3

(A) Small choroidal melanoma (B) with subtle orange pigment on autofluorescence (C) and subretinal fluid on optical coherence tomography

Muscat et al.24 studied the OCT findings associated with melanoma and nevi. Of 20 patients with the diagnosis of melanoma, all had serous retinal detachment on OCT, including eight cases in which no serous detachments had been noted on clinical examination. Eighteen patients also showed either intraretinal changes or distorted retinal architecture. In contrast, of 40 patients with choroidal nevi, 27 patients (67.5%) showed no retinal abnormalities on OCT. Twelve patients in this group had serous retinal detachments with only one of these noted clinically.24

Shields et al.25 studied 120 eyes with choroidal nevus using OCT. They found intraretinal changes (15%), subretinal fluid (26%), retinal thinning (22%), drusen (41%), and RPE detachment (12%). Photoreceptor loss or attenuation was noted in 51% of cases. When comparing OCT with clinical examination, these authors concluded that OCT was more sensitive in the detection of related retinal edema, subretinal fluid, retinal thinning, photoreceptor attenuation, and RPE detachment.

Choroidal melanoma, in general, is poorly imaged on OCT. However, detection of overlying subretinal fluid by OCT could be important in confirming the suspicion of melanoma in eyes with borderline small or intermediate-sized tumors. The presence of subretinal fluid is a risk factor for growth of the tumor. Another OCT finding associated with neurosensory detachment in the case of the choroidal melanoma is the presence of small reflective bodies present on the outer surface of the detached retina. These bodies corresponded to aggregates of RPE cells that are seen clinically.11

Espinoza et al.11 showed that OCT findings of subretinal fluid might have a predictive value in identifying choroidal melanocytic tumors that are likely to grow. In a study using OCT on 30 eyes with suspicious choroidal melanocytic lesions, tumor growth was found in only 8% of those with no subretinal fluid, 50% of those with active subretinal fluid, and 11% of those with retinal atrophy or edema.11

The findings on OCT of intraretinal cystic changes, RPE alterations, photoreceptor loss, and RPE detachment are related to chronic retinal degeneration and suggest a stable, chronic choroidal nevus.25 Conversely, the presence of subretinal fluid and photoreceptor preservation suggests a more acute progression and more active lesion with risk for growth into melanoma.11

Specific OCT findings of choroidal nevus and melanoma are limited to the anterior surface of the lesion with minimal information deeper within the choroidal mass. Shields and associates noted increased thickness of the RPE-choriocapillaris layer (68%) for choroidal nevi. The anterior portion of the nevus can be hyporeflective (62%), isoreflective (29%), or hyper-reflective (9%). Hyporeflectivity was observed in 62% of pigmented nevi and 18% of nonpigmented nevi.11 Because OCT is not helpful in evaluating internal tumor tissue characteristics, ultrasonography and OCT may be complementary techniques useful in evaluating a suspicious choroidal melanocytic tumor.