Introduction
Uveal melanoma is the most frequent primary intraocular malignancy in adults. It can affect any part of the uveal tract, but choroidal melanoma is predominant, while iris and ciliary body melanomas are less common [1]. Optic nerve invasion is rare, having been described in 0.6-6.9% of uveal melanoma cases treated with enucleation [2,3,4,5]. Optic nerve invasion has been associated with poor prognosis and higher melanoma-related mortality [4,6]. In rare cases, the melanoma cells may extend to the optic chiasm and cause a visual field defect in the contralateral eye [7,8].
Infiltration of the optic nerve has been seen in 5-80.8% of eyes when the uveal melanoma is located adjacent to the optic nerve head [3,4]. Loss of light perception was reported to be a sign of optic nerve invasion [7]. Infiltration of the optic nerve has also been associated with elevated intraocular pressure, non-spindle cell type, and juxtapapillary location [3,4,9]. However, the optic nerve exhibits features that may protect against invasion, such as its unique blood supply and its surrounding dense fibrous coat [10].
In the current study, we addressed morphological risk factors associated with optic nerve invasion of choroidal melanoma and identified possible mechanisms of optic nerve invasion.
Patients and Methods
The medical charts and histology slides of patients with primary choroidal melanoma who were treated by enucleation or exenteration and accessioned into the L.F. Montgomery Ophthalmic Pathology Laboratory, Emory Eye Center, between January 1997 and October 2016 and whose pathology showed invasion of the optic nerve were reviewed. Patients with the melanoma confined to the iris and/or with ciliary body melanoma were excluded. The patient data evaluated included age and sex. The clinicopathological features evaluated included intraocular pressure, pretreatment modality, tumor location, largest basal diameter (LBD), tumor thickness, histopathologic tumor cell type, scleral invasion, extrascleral extension at the time of surgery, rupture of Bruch's membrane, necrosis, mitotic activity, vitreous seeding or hemorrhage, and emissary canal, vortex vein, and retinal invasion. The study followed the tenets of the Declaration of Helsinki and was approved by the Emory University Institutional Review Board.
Histology
Enucleation/exenteration specimens were immediately fixed in formalin (10%) and embedded in paraffin. Five-micrometer-thick pupil-optic nerve sections that included the center of the melanoma were mounted on glass slides; then the sections were deparaffinized with xylene and rehydrated through a graded series of ethanol and distilled water. The sections were stained with hematoxylin and eosin (H&E) and periodic-acid Schiff, and bleached sections were stained with H&E according to routine protocols. The slides were evaluated qualitatively and quantitatively with a light microscope (Olympus BHTU; Olympus, Tokyo, Japan). The tumor was considered necrotic when >50% of the tumor was necrotic. The cell type (spindle, mixed, or epithelioid), LBD, and thickness were recorded. The growth pattern of the tumor with regard to optic nerve invasion was determined (i.e., peripapillary tumor, transvitreal tumor, retinal invasion, or a combination of these growth patterns).
Data Analysis
Statistical analysis was carried out with SPSS for Windows (version 22.0) and MedCalc (version 14.8.1) statistical software. The normality of data was tested using the Kolmogorov-Smirnov test. If normality was rejected, a nonparametric test was performed. For continuous variables, data are expressed as mean ± SD and 95% CI for the mean. Comparisons between 2 variables were performed using the Mann-Whitney U test for continuous variables and Fisher's exact test for binomial variables. A p value of <0.05 is considered statistically significant.
Results
Twenty-one patients (13 males, 8 females) with primary uveal melanoma arising from the choroid treated with enucleation (20 cases) or exenteration (1 case) with histologic optic nerve invasion were included in this analysis. The mean age was 65.67 ± 14.72 years (ranging from 39 to 90 years). The histologic cell type was mixed in 76% (16/21), epithelioid in 14% (3/21), and of the spindle type in 10% (2/21) of the cases. The mean LBD was 14.04 ± 7.11 mm (95% CI: 10.81-17.28) and the mean tumor thickness was 8.18 ± 5.00 mm (95% CI: 5.90-10.45). A peripapillary location was present in 86% (18/21) of the cases, the ciliary body was involved in 19% of the cases (4/21), and diffuse tumor growth was documented in 1 case. In 67% (14/21) of the cases, the tumor had ruptured through Bruch's membrane, and retinal invasion was seen in 43% (9/21). Vitreous seeding of melanoma cells was observed in 2 cases, and vitreous hemorrhage was also found in 10% (2/21). In 24% of the cases (5/21), the tumor was necrotic and there were 0-5 mitotic figures present in 40 high-power fields in the tumors.
The 4 types of optic nerve invasion were identified as shown in Table 1 and Figures 1, 2, 3, 4, 5, 6. Transvitreal invasion was seen in 10% (2/21) (Fig. 1, 2), retinal invasion in 23% (5/21) (Fig. 3, 4), direct peripapillary invasion in 57% (12/21) (Fig. 5, 6), and a combined mechanism in 10% (2/21) of the cases. Scleral invasion was present in 91% (19/21), emissary canal invasion was seen in 33% (7/21), and extrascleral extension was found in 29% (6/21) of the specimens. Vortex vein invasion was present in 14% (3/21). An elevated intraocular pressure was reported in 4 patients (19%). One patient had previously been treated with transpupillary thermotherapy, and 2 patients had had prior plaque brachytherapy.
Fig. 1
Mechanism of transvitreal optic nerve invasion of choroidal melanoma. This mechanism involves the melanoma breaking through Bruch's membrane, with melanoma cell invasion into the vitreous and seeding of the optic nerve head. The melanoma may also cause vitreous hemorrhage.
Fig. 2
Optic nerve invasion of melanoma cells dispersed from the surface of a peripheral melanoma arising adjacent to the ora serrata. a Gross photo. b Photomicrograph. Periodic-acid Schiff. ×5.
Fig. 3
Mechanism of retinal invasion-related optic nerve invasion of the melanoma. The choroidal melanoma (left) may directly invade the retina, and the intraretinal melanoma may extend into the optic nerve head. Alternatively (right), the melanoma may invade through the retina, and melanoma cells spread along the inner limiting membrane to the optic nerve head.
Fig. 4
a Retinal invasion (*) after breaking through Bruch's membrane. H&E; bleached. ×5. b Choroidal melanoma cells (arrowheads) spreading along the inner limiting membrane after retinal invasion. H&E. ×100.
Fig. 5
Mechanism of peripapillary invasion of choroidal melanoma into the optic nerve head. Left: the melanoma invades along the edge of Bruch's membrane into the prelaminar optic nerve. Right: the melanoma extends directly through the border tissue into the laminar level of the optic nerve.
Fig. 6
a Peripapillary choroidal melanoma spread between the end of Bruch's membrane (arrowheads) and the border tissue into the optic nerve. H&E. ×10. b Microscopic image of choroidal melanoma. Direct invasion (arrowheads) into the optic nerve through the border tissue. H&E. ×10.
Invasion Pattern
We observed direct peripapillary optic nerve invasion in 57% (12/21) of our cases. In 43% (9/21), other mechanisms played a role, including transvitreal spread and/or retinal invasion. Optic nerve invasion was prelaminar in 75% (9/12) and 56% (5/9), laminar in 8% (1/12) and 11% (1/9), and retrolaminar in 17% (2/12) and 33% (3/9) in the peripapillary and the transvitreal/retinal invasion group, respectively. Table 2 shows the clinicopathological characteristics of melanomas with peripapillary versus transvitreal and/or retinal invasion. In the peripapillary group, the cell type was mixed in 67% (8/12), epithelioid in 25% (3/12), and of the spindle type in 8% (1/12) of the cases. In melanomas with transvitreal/retinal invasion, the cell type was mixed in 89% (8/9) and of the spindle type in 11% (1/9) of the cases. A statistically significant difference was disclosed between the 2 groups in LBD (p = 0.021), tumor thickness (p = 0.017), and vortex vein (p = 0.022) and retinal invasion (p = 0.007). Elevated intraocular pressure was found in 1 case (8%) in the peripapillary group and in 3 cases (33%) in the transvitreal/retinal invasion group.
Table 2
Clinicopathological features of choroidal melanomas with peripapillary versus transvitreal and/or retinal invasion
Invasion of the Lamina Cribrosa
Optic nerve invasion was prelaminar in 67% (14/21), laminar in 10% (2/21), and retrolaminar in 23% (5/21) of the cases. In the exenteration specimen, the tumor extended through the optic nerve to the surgical margin. Table 3 summarizes the clinicopathological features of the melanomas with prelaminar and laminar/retrolaminar invasion. In the prelaminar group, the cell type was mixed in 79% (11/14), epithelioid in 14% (2/14), and of the spindle type in 7% (1/14) of the cases. In the melanomas with laminar/retrolaminar invasion, the cell type was mixed in 71% (5/7), epithelioid in 14% (1/7), and of the spindle type in 14% (1/7) of the cases. Ninety-three percent (13/14) of the prelaminar melanomas and 71% (5/7) of the laminar/retrolaminar melanomas occurred in a peripapillary location. The breakdown of optic nerve invasion according to growth pattern showed the following results: transvitreal in 7% (1/14) of the prelaminar group versus 14% (1/7) of the laminar/retrolaminar group; retinal invasion in 21% (3/14) versus 29% (2/7); peripapillary in 64% (9/14) versus 43% (3/7); and combined in 7% (1/14) versus 14% (1/7). There was a statistically significant difference between the prelaminar and the laminar/retrolaminar group associated with vortex vein invasion (p = 0.013) and tumor necrosis (p = 0.021). Elevated intraocular pressure was observed in 4 cases (29%) in the prelaminar group and in no patients in the laminar/retrolaminar invasion group.
Table 3
Clinicopathological features of choroidal melanomas with prelaminar versus laminar/retrolaminar optic nerve invasion
Discussion
Lindegaard et al. [11] evaluated 157 enucleated eyes and reported 4 different mechanisms for uveal melanomas to invade the optic nerve. Invasion of the tumor through the border tissue of Elschnig was an infrequent mechanism in that case series; extension of the tumor between the termination of Bruch's membrane and the border tissue of Elschnig into the optic nerve head was found to be a simple mechanical expansion of a growing tumor. Direct extension via this tissue was the most common mechanism in our study. Neuroretinal spread was associated with retrolaminar optic nerve invasion in their study, and tumor spread along the inner limiting membrane was found in only 1 case [11]. Invasion via the retina and vitreous was less common in our study, likely due to the rarity of uveal melanoma invading these structures.
In our study, 21 primary choroidal melanomas with histologically proven invasion of the optic nerve were identified between 1997 and 2016. Based on our findings, 4 mechanisms of optic nerve invasion of choroidal melanoma were identified. A peripapillary location was present in 86% of our cases, and direct peripapillary invasion was the most common mechanism of optic nerve invasion. A large histopathologic evaluation of uveal melanoma eyes enucleated from the Collaborative Ocular Melanoma Study (COMS) reported that 16.7% of the melanomas presented in a juxtapapillary location and 39.6% of those specimens showed optic nerve invasion, of which 1.1% extended beyond the lamina cribrosa [5].
Tumor size (LBD and thickness) is one of the most important clinical prognostic features of uveal melanoma [12]. The COMS Report No. 6 showed significantly greater invasion of the retina, vitreous, vortex veins, and tumor vessels and rupture of Bruch's membrane when the tumor was >10 mm in thickness, or >16 mm in LBD and ≥2 mm in thickness [5]. Our analysis included relatively larger choroidal melanomas with a mean LBD of 14.04 ± 7.11 mm and a mean thickness of 8.18 ± 5.00 mm. Bruch's membrane was ruptured less frequently (67% of the cases) in our series than in the COMS (88% of the cases) [5]. Cell type is generally considered to be the most potent single predictor of outcome [13]. Histologic examination showed a mixed cell type in 76%, an epithelioid type in 14%, and a spindle cell type in 10% of our cases. Microscopic scleral extension and emissary canal invasion have been reported in 10-40% of enucleated uveal melanoma eyes [14,15]. Interestingly, we found a high percentage of scleral invasion in our cases (91% of all cases), and we detected emissary canal invasion in 33% and extrascleral extension in 29% of the eyes in our study. Tumor invasion within the sclera was less frequent in the COMS report (57.3% of the medium-sized tumors and 54.8% of the large tumors) [5]. Font et al. [16] reported extraocular extension of uveal melanomas in 39% of their cases. We found the following mechanisms of scleral invasion: direct invasion (11 cases); superficial scleral invasion (7 cases); perivascular emissary canal invasion (1 case); and intrascleral vortex vein combined with superficial direct scleral invasion (1 case). The sclera was necrotic in 1 case (the exenteration specimen).
Shammas and Blodi [3] reviewed 26 cases of peripapillary melanoma, and in 80.8% the tumor extended into the optic nerve or its sheaths. They observed that tumor necrosis, retinal invasion, glaucoma, and cell type (mixed or epithelioid) influenced the extension of melanoma cells along the optic nerve [3]. Melanoma cells extended into the subarachnoidal space in 46% (12) of their cases [3]. In our study, tumor cells extended beyond the surgical margin in 1 case; no other tumor was found to spread along the optic nerve sheaths. Optic nerve invasion was prelaminar in 67% and laminar/retrolaminar in 33% of our cases. Significant differences between prelaminar and laminar/retrolaminar invasion in our study correlated with vortex vein invasion and tumor necrosis, with laminar/retrolaminar invasion being more likely if there was vortex vein or tumor necrosis.
According to histopathologic studies following enucleation of uveal melanoma, retinal invasion occurs in 23-59% of cases [3,5,17,18]. We observed retinal invasion of choroidal melanoma in 43% of our cases, and in 23% of these cases retinal invasion was the primary mechanism leading to optic nerve invasion. In enucleation specimens with retinal invasion, dispersed melanoma cells may form a malignant preretinal pigmented membrane [19]. We also observed uveal melanoma cells spreading along the inner limiting membrane and reaching the optic nerve head. We classified this growth pattern as a subclass of retinal invasion. It has been postulated that retinal invasion is more likely to occur when the tumor arises adjacent to the ora serrata or peripapillary choroid, where the retina and choroid are tightly adherent [20]. We also support the role of the unique structure of Bruch's membrane at its ora serrata and peripapillary terminations, making it predispose to retinal invasion [21,22]. Retinal invasion has prognostically been shown to increase the risk for tumor recurrence but not for metastasis or death [23].
Based on our findings, we postulate 2 mechanisms explaining transvitreal melanoma spread: (1) viable melanoma cells from the surface of the tumor might be able to disperse to the vitreous after the tumor has ruptured Bruch's membrane, or (2) extravasation of malignant cells may also occur through the retinal vessels adjacent to the tumor or melanoma cells may migrate with vitreous hemorrhage to the vitreous body [24,25]. These cells tend to invade the optic nerve head and the optic nerve beyond the lamina cribrosa. Spencer [6] reported 10 uveal melanomas (8 choroidal, 2 iris and ciliary body melanomas) with intravitreal spreading of tumor cells showing optic nerve invasion. This type of spreading is thought to be associated with more malignant melanoma cell types, particularly with those in which necrosis or hemorrhage has been found [6]. Elevated intraocular pressure is assumed to enhance dispersed viable tumor cells to adhere to and invade the optic nerve [3,6].
Kivelä and Summanen [26] defined retinoinvasive uveal melanoma as a single entity showing similarities with metastatic cancers to the retina. Interestingly, the same mechanism of both retinal invasion and transvitreal spread has been observed in cancers that metastasize to the retina and vitreous [27,28,29,30,31]. Moreover, a recent article reported a case of metastatic cutaneous melanoma to the vitreous and retina with optic nerve head involvement [27]. The explanation for this phenomenon was that the melanoma cells could pass into the retinal circulation [28] or gain access to the vitreous directly through the pars plana [29]. Jaissle et al. [30] reported isolated vitreous metastasis of a cutaneous malignant melanoma case and postulated that tumor cells might pass into the retrohyaloid space with a spontaneous epiretinal hemorrhage.
In conclusion, we observed 4 mechanisms of optic nerve invasion of primary choroidal melanoma. The peripapillary tumor location was the most prevalent in our series. However, a peripapillary location may not be the only predisposing factor for optic nerve invasion of choroidal melanoma. In 43% of the cases, other mechanisms were involved, including transvitreal spread and retinal invasion. Tumor infiltration into adjacent tissues is a characteristic of aggressive tumor behavior. Significantly higher LBD and tumor thickness measurements, as well as higher rates of vortex vein and retinal invasion, were observed in the transvitreal and retinal spread type group compared to the peripapillary invasion type group. Transvitreal and retinal spread of choroidal melanoma showed similar mechanisms of spread to those in metastatic cancers to the retina and vitreous. Choroidal melanomas exhibiting these growth patterns may have aggressive phenotypes similar to vitreoretinal metastasis from systemic cancers, having poor prognosis [11,31]. In contrast, most peripapillary melanomas with optic nerve involvement show mechanical expansion of the choroidal mass rather than an aggressive infiltrative growth pattern [11].
Acknowledgement
This study was supported by NIH grant P30EY06360 and an unrestricted departmental grant from Research to Prevent Blindness, Inc.
Statement of Ethics
The study followed the tenets of the Declaration of Helsinki and was approved by the Emory University Institutional Review Board.
Disclosure Statement
The authors declare no conflicts of interest.

Get Permission