Choroidal Tumors

Ultrasonography

Choroidal nevi are often too flat and small to be accurately measured and distinguished from choroidal melanoma. Suspicious nevi should, therefore, be closely followed for changes in appearance, especially in size. The role of ultrasonography is usually to accurately measure the lesion at each follow-up examination to check for changes in size. Sufficiently elevated nevi detected on ultrasonography appear highly reflective in both A- and B-scans.

Fluorescein Angiography and Indocyanine-Green Angiography

Choroidal nevi have no particular pattern on FA. However, pigmented nevi are, in general, hypofluorescent, whereas less pigmented nevi are more likely to show relative hyperfluorescence. Nevi that produce alterations of the RPE show hyperfluorescent foci indicative of drusen and other hypofluorescent foci that correspond to pigment clumping or orange pigment accumulation. Focal detachments of the retina or the RPE overlying the choroidal nevi show characteristic late hyperfluoresecence.

On ICGA, choroidal nevi remain hypofluorescent during the entire course of the test. Intrinsic tumor vessels are visible in many nevi. Choroidal vessels around the tumor remain unchanged.

Autofluorescence

When evaluating choroidal nevi with autofluorescence, it is important to consider the intrinsic autofluorescence of the lesion itself and then the autofluorescence of the overlying RPE. Choroidal nevus has little intrinsic autofluorescence. Pigmented nevi are usually hypoautofluorescent, while nonpigmented nevi display only slight hyperautofluorescence, likely related to unmasking of deeper mild scleral autofluorescence. Autofluorescent images of overlying RPE alterations are much more dramatic and informative. Overlying RPE hyperplasia, atrophy, and fibrous metaplasia indicate chronic stable nevus and exhibit hypoautofluorescence. Overlying drusen, another sign of stable nevus, display slight ring-shaped hyperautofluorescence. Chronic or resolved subretinal fluid implies that the nevus is stable and appears hypoautofluorescent too. The presence of new-onset subretinal fluid, suggestive of nevus activity or transformation into melanoma, appears slightly hyperautofluorescent. Choroidal nevi rarely display overlying orange pigment although when visible, it is brightly hyperautofluorescent.

Optical Coherence Tomography

In the study of choroidal nevi, TD-OCT is capable of identifying retinal changes overlying the tumor, such as intraretinal edema, subretinal fluid, atrophy of the photoreceptors, and RPE alterations, but cannot display the internal characteristics of the lesions. The ability of OCT to detect subretinal fluid and secondary retinal changes can have important implications for monitoring suspicious nevi given that the presence of fluid is a predictor of growth for small melanocytic choroidal tumors, a clinical characteristic that could be easily missed in fundus exploration. Consequently, OCT is a highly valuable technique to identify suspicious nevi that need to be monitored carefully for growth. A chronic TD-OCT pattern has been described, characterized by retinal thinning, RPE alterations, photoreceptor loss, RPE detachment, and the presence of intraretinal cysts (which could be confused on clinical examination with subretinal fluid), suggesting chronic retinal degeneration and a long-standing dormant tumor. In contrast, an active TD-OCT pattern is characterized by the presence of a localized full-thickness retinal detachment overlying, or adjacent to, the choroidal tumor with normal retinal architecture, which may indicate a potentially more active tumor that is more likely to grow.

Shields et al . used TD-OCT to assess 120 choroidal nevi, finding that TD-OCT was capable of detecting retinal alterations such as retinal edema, subretinal fluid, retinal thinning, photoreceptor attenuation, and RPE detachment with high sensitivity. Moreover, TD-OCT was far superior to ophthalmoscopic or ultrasound evaluation in detecting these retinal alterations, and also permitted classification of the overlying retinal edema. In that study, OCT yielded numerous retinal findings at the site of the nevus, including overlying retinal edema (15%), subretinal fluid (26%), retinal thinning (22%), drusen (41%), and RPE detachment (12%). The thickness of the overlying retina was found to be normal (32% of eyes), thinned (22%), or thickened (45%); photoreceptor loss or attenuation was noted in 51% of cases. However, specific OCT findings for the choroidal nevus were limited to its anterior surface, with minimal penetration into the mass; these findings included increased thickness of the RPE/choriocapillaris layer (68%) and optical qualities within the anterior portion of the nevus, including hyporeflectivity (62%), isoreflectivity (29%), and hyperreflectivity (9%). Vision loss secondary to choroidal nevi can be explained by signs of chronic degeneration, including retinal edema (15%), retinal thinning (22%), retinal disorganization (42%), or attenuation of the photoreceptor layer (51%) on TD-OCT.

In contrast to TD-OCT, EDI-OCT technology not only allows for visualization of retinal changes, but also for the assessment of the choroidal component. In the first reports of this technology, amelanotic nevi had a homogenous appearance with a medium reflective band associated with visible choroidal vessels within the tumor; by contrast, melanotic nevi appeared as a highly reflective band within the choriocapillaris layer with posterior shadowing. Small choroidal nevi that are undetectable by ultrasonography can be objectively identified, imaged, and measured with this technique ( Fig. 17.1 ).

The most common imaging features of EDI-OCT registered by Shah et al . included partial (59%) or complete (35%) choroidal shadowing deep to the nevus, choriocapillaris thinning overlying the nevus (94%), RPE atrophy (43%), RPE loss (14%), RPE nodularity (8%), photoreceptor loss (43%), inner segment–outer segment junction (IS–OS) irregularity (37%), IS–OS loss (6%), external limiting membrane irregularity (18%), outer nuclear and outer plexiform layer irregularity (8%), and inner nuclear layer irregularity (6%). The presence of overlying subretinal fluid was better identified by EDI-OCT (16% of eyes) than by ophthalmoscopic examination (8%) and ultrasound evaluation (0%). Pigmented nevi showed significantly more intense choroidal shadowing versus nonpigmented nevi. Finally, measurements with EDI-OCT were more precise due to improved resolution.

Recently, SS-OCT has been shown to display more intricate details of the tumor as a homogeneous internal structure, revealing intralesional vessels (100% of eyes), intralesional cavities (20%), intralesional granularity (47%), distended bordering vessels (73%), abnormal choriocapillaris (83%), and abnormal choriocapillaris over the tumor apex (58%). Furthermore, on evaluation of the sclerochoroidal interface, SS-OCT was shown to be capable of identifying three different morphologies, designated as dome-, plateau-, and almond-shaped, which correlate with the morphologies seen on histopathologic examination. When SS-OCT and EDI-OCT were compared in terms of their ability to assess tumor shape, both techniques were found to be equivalent at determining the morphology of amelanotic nevi, but SS-OCT was significantly better at visualizing the morphology of melanotic nevi. The authors of that study found that all nevi displayed a characteristic intralesional vascular pattern (established previously by histopathology) that mimics the normal choroid and these vessels appear equidistant and equal in size. The same study found intralesional cavities in 20% of the nevi, which are probably not vascular in nature because they were not visible on ICGA. The authors of that study hypothesize that these cavities could be distinct clumps of cells with differing reflectivity or fluid-filled cysts inaccessible to the vasculature. The presence of distended bordering vessels at the border of the tumor, imaged both by ICGA and SS-OCT, may be related to the presence of previous or persistent subretinal fluid in these patients. Regardless of the tumor thickness or configuration, 71% of melanotic nevi did not have an identifiable nevus-scleral interface, indicating that even with this improved SS-OCT technology, the interface cannot be determined in most pigmented nevi. The densely packed spindle cells at the base of the nevus could explain the increased hyporeflectivity as the nevus transitions to the sclera, but this could also be related to pigment shadowing. Intralesional granularity, a previously unreported characteristic, was seen in almost half of the nevi in patients who underwent SS-OCT, and this granularity may represent an aspect of the cellular milieu in the nevus.

Treatment

In general, choroidal melanocytic nevus must be observed. In the case of presence of risk factors for growth or documented growth, the treatment should be considered as is discussed below.

Secondary changes in the surrounding retina and RPE such as subretinal fluid, RPE detachment, and the development of choroidal neovascular membrane are rare complications resulting in diagnostic confusion for malignant transformation and in vision loss depending on the location of the nevus. Various treatment modalities have been used in these cases as argon laser, photodynamic therapy (PDT), subthreshold transpupillary thermotherapy, and anti-VEGF agents.

Ultrasonography

Combined A- and B-mode ultrasonography is the most important ancillary test. On A-scan, choroidal melanoma typically demonstrates a high initial spike, low-to-medium internal reflectivity with a straight posterior climbing scleral spike. Sound attenuation (Kappa angle) is often observed in larger tumors. A-scan is also useful for evaluating internal blood flow, seen as flickering spikes within the internal tumor (fast low-amplitude peaks). On B-scan, choroidal melanoma is dome- or mushroom-shaped, or diffuse. It appears as an echo-dense mass with internal homogeneity, acoustic hollowing at the base, choroidal excavation, and orbital shadowing. Other characteristics are exudative retinal detachment and subretinal or vitreous hemorrhage. Ultrasound B-scans can reveal an extrascleral extension of the tumor that appears as nodules with low reflectivity near the base of the tumor and which contrasts with orbital fat.

Fluorescein Angiography and Indocyanine-Green Angiography

FA shows no pathognomonic pattern for choroidal melanoma; for this reason, diagnostic accuracy with this technique is limited. A small choroidal melanoma may show no appreciable abnormalities on FA. This normal pattern correlates with an intact RPE overlying a melanoma. A larger melanoma that disrupts the RPE presents a mottled hyperfluorescence. Areas of orange pigment are hypofluorescent. In the venous phase, areas of pinpoint hyperfluorescent foci, especially on the margins, become apparent. Subretinal fluid shows late staining.

In the case of a mushroom tumor, angiogenesis creates independent tumor circulation as the uveal melanoma grows. This vascularization is best visualized with ICGA, which reveals the overlying retinal and choroidal circulation, a phenomenon known as double-circulation. In these cases, extensive leakage with progressive fluorescence is also observed, together with late staining of the lesion and multiple pinpoint leaks (hot spots) at the level of the RPE.

On ICG angiography, most choroidal melanomas remain hypofluorescent/isofluorescent during the entire angiogram. As mentioned before, this technique is capable of imaging the intrinsic microcirculation of choroidal melanomas ( Figs. 17.7 and 17.8 ).

Autofluorescence

Intrinsic autofluorescence is mild in choroidal melanoma, but extrinsic autofluorescence is markedly different. Choroidal melanoma demonstrates extrinsic hyperautofluorescence from orange pigment and subretinal fluid, with the brightest hyperautofluorescence due to lipofuscin overlying choroidal melanoma.

Optical Coherence Tomography

TD-OCT has been primarily used to evaluate retinal changes over choroidal melanomas, but offers poor visualization of the details of the internal structure. It was identified as an active pattern, associated with documented tumor growth, and characterized by localized full-thickness retinal detachment overlying or adjacent to the choroidal tumor, phototoreceptor preservation, and FA hotspots, and inversely associated with drusen and atrophic RPE changes.

EDI-OCT is a recent improvement in the assessment of choroidal melanoma. The images generally show a gentle, dome-shaped, smooth-surface topography with relatively fresh subretinal fluid demonstrating shaggy photoreceptors. Detection of overlying subretinal fluid by OCT could help confirm the suspicion of melanoma in borderline tumors because the presence of fluid is a risk factor for eventual tumor growth. Moreover, EDI-OCT is better than ophthalmoscopic or ultrasonographic evaluation in identifying subretinal fluid.

In small (≤3 mm) choroidal melanomas, the estimated tumor height measured by EDI-OCT is approximately 55% less than ultrasonographic estimates (mean of 1025 µm with EDI-OCT vs 2300 µm with ultrasonography). This difference can be attributable to several factors, but primarily because gross estimation of the sclerochoroidal junction (including the sclera) with ultrasonographic calipers leads to imprecise identification of the posterior margin, and, thus, inclusion of the overlying retina in the thickness measurement. The axial resolution of EDI-OCT is 3–4 versus approximately 50–200 µm for ultrasonography; as a result, misplacement of the calipers in OCT can result in a 5–10 µm error, whereas misplacement on ultrasonography means a 200–400 µm error. Importantly, in thick choroidal tumors, lack of visualization of the sclerochoroidal junction on EDI-OCT does not allow for accurate fitting of the calipers; consequently, in these cases, ultrasonography is preferred for measurement. In summary, small tumors are more accurately measured with EDI-OCT, whereas ultrasound is better for thicker tumors.

Shields et al . reported that small choroidal melanoma visualized on EDI-OCT shows homogeneous reflectivity in the anterior surface with gradual posterior shadowing, although internal characteristics cannot be discriminated with this imaging technique. On EDI-OCT, the choroidal characteristics of melanomas can be similar to a nevus, with the presence of deep optical shadowing and overlying choriocapillaris compression in nearly all cases. These same authors found that this imaging technique was capable of identifying several of the potentially important signs most commonly present in melanomas versus nevi, such as increased tumor thickness, subretinal fluid (92% vs 16% of cases), RPE atrophy (95% vs 43%), subretinal lipofuscin deposits (95% vs 45%), shaggy photoreceptors (49% vs 0%), intraretinal edema (16% vs 0%), structural loss of photoreceptors (24%), loss of external limiting membrane (43% vs 2%), loss of ellipsoid layer (65% vs 6%), outer nuclear layer (16% vs 0%), and outer plexiform layer (11% vs 0%). Other inner retinal features included irregularity of inner plexiform layer (8% vs 0%), irregularity of inner nuclear layer (8% vs 6%), irregularity of ganglion cell layer (8% vs 0%), and abnormal nerve fiber layer (5% vs 0%). With regard to the internal features of the tumor, choroid, and Bruch membrane, Shields et al . observed no difference between choroidal nevi and melanomas.

Shaggy photoreceptors have been described as irregular, elongated, and presumed swollen photoreceptors in patients with fresh subretinal fluid. This could represent edematous photoreceptors or macrophages with lipofuscin on the posterior surface of the detached retina, as has been found in other conditions that produce subretinal fluid, such as central serous chorioretinopathy. In patients with choroidal melanoma, the outer segments of the photoreceptors were elongated from 30 to 50 μm in the presence of subretinal fluid, which contains precursors of the lipofuscin visible on FAF. The patchy increase in autofluorescence in this pathology appears to originate from the elongated photoreceptor outer segments in the subretinal space. The autofluorescent fluorophores in the photoreceptor outer segments may be concentrated in precipitates or can settle into the inferior border of the detachment, leading to hyperfluorescence of fresh subretinal fluid. The presence of shaggy photoreceptors is an important feature for differential diagnosis because these are found in nearly half (49%) of patients with a choroidal melanoma but not in eyes with a choroidal nevus.

A pilot study evaluated the value of SS-OCT to assess choroidal melanoma and small pigmented choroidal lesions that presented risk factors for growth, finding the presence of grossly homogeneous configurations, with irregularities of the internal space in all of the tumors, compared with the more homogeneous nevi. Two of the melanomas in that study showed internal hollow round spaces that were continuous between serial SS-OCT slides. Authors hypothesized that those spaces may correspond to intratumoral blood vessels. Importantly, the presence of possible intralesional vessels was noted in 40% of patients. By contrast, the thin interface identified as the choriocapillaris in the choroidal nevi was not visible in any of the melanomas (and only in one of the lesions with risk factors for growth), whereas it was visible in almost all of the nevi cases. Other features such as changes or ablation of the RPE and the outer retina, absence of drusen, presence of lipofuscin, subretinal fluid, or intraretinal edema were identified in a higher proportion of patients versus those with a nevus.

OCT is particularly useful in assessing the presence and degree of radiation-related maculopathy following radiotherapy of choroidal melanoma. Radiation-induced retinopathy is the most common cause of irreversible visual loss in patients treated with plaque or charged-particle radiotherapy for choroidal melanoma. Radiation-induced maculopathy results from the loss of perifoveal capillary endothelial cells, causing capillary occlusion with microaneurysm formation, telangiectasia, dilation of collateral capillary, hemorrhages, cotton–wool spots, hard exudates, macular edema, retinal or optic disk neovascularization, and eventually retinal atrophy. Intraretinal edema is often the first OCT manifestation of radiation maculopathy before it is visually symptomatic or clinically appreciable. OCT is also useful in monitoring resolution of radiation-induced macular edema following therapy with laser photocoagulation, anti-VEGF, or intravitreal corticosteroids, providing an objective guideline for documentation of treatment results.

Magnetic-Resonance Imaging

The majority of uveal melanomas are hyperintense to vitreous on T1-weighted images and hypointense to vitreous on T2-weighted images. The use of enhancing agents such as gadolinium improves the resolution of the choroidal melanoma and allows differentiation from subretinal fluid.

Fine-Needle Aspiration Biopsy

Currently, choroidal melanomas can be diagnosed quite accurately without the use of invasive techniques. For this reason, the role of FNAB is uncertain. Moreover, assessment of tumor aspirates is challenging and subsequent seeding has also been reported. However, FNAB has been increasingly used as a prognostic tool ever since it was demonstrated that uveal melanoma in enucleated eyes often displays chromosome 3 monosomy, the presence of which indicates a poor prognosis, with metastatic disease occurring in 50% of patients within 3 years compared with no metastatic disease in those with disomy 3. Other mutations have also been found, including chromosome 8 (53%), chromosome 6 (46%), and chromosome 1 (24%).

Gene-expression profiling (GEP) has also been employed for RNA evaluation. In 2003, two melanoma groups that correlated with monosomy 3 and disomy 3 tumors were identified by GEP. In 2004, GEP was used to confirm the presence of two classes of melanoma: Class 1 tumors (low grade) had a 95% survival rate, whereas survival in patients with class 2 (high grade) melanomas was only 31% at 8 years. Cytogenetic analysis using DNA or RNA method has emerged as a reliable prognostic test for uveal melanoma.

Treatment

The treatment options include a broad spectrum of enucleation, brachytherapy, proton beam radiotherapy, local resection, transpupillary thermotherapy, or combinations of such treatments depending on the clinical circumstances. Each case should be individualized and tumor size, tumor location, visual acuity, status of opposite eye, patient age, general health, and other factors as preferences of the patient must be considered before making a therapeutic choice. The two most frequently employed treatments for choroidal melanoma include enucleation or radiotherapy (using brachytherapy or proton beam radiotherapy).