Periocular melanoma is a rare, generally lethal, malignancy that can arise from in the eye (uveal tract), the conjunctiva, or the ocular adnexa (eyelid or orbit) (1). Uveal origin—from the iris, ciliary body or choroid—is the commonest ocular melanoma, with conjunctival melanoma being the second most frequent. The reported incidence of choroidal melanomas has increased in recent decades, possibly because of greater exposure to UV light (2), or possibly due to greater detection. Eyelid melanomas are very rare, with knowledge limited to small case series (3–5).

Orbital melanoma occurs either as primary disease, as secondary disease (local invasion from a uveal, conjunctival, or eyelid primary tumor), or as metastatic disease from distant origins such as skin. Melanoma accounts for 5–20% of metastatic and secondary orbital malignancies, but only a minute proportion of primary orbital neoplasia (6–11). Primary orbital melanoma (POM) is extremely rare, with only about 50 cases reported to date (Table 1), and is thought to arise from melanocytic cells of the leptomeninges or ciliary nerves, or from ectopic intraorbital nests of melanocytes (12). POM can occur de novo, but it is often reported in association with pigmentary changes within periocular tissues—such as nevus of Ota, blue cellular nevus, or oculo-dermal melanosis; indeed, over a half of patients with nevus of Ota have pigmentation within the orbit, including oculo-dermal melanosis (13). While POM is reported to have a very poor prognosis, there have been sporadic reports of long-survival, such as one patient who lived for almost 30 years after initial diagnosis (14).

Based on a series of patients with POM, this work aimed to extend the knowledge about clinical presentation, radiological appearance, surgical approach, and prognosis for this extremely rare condition.

All patients with a biopsy-proven diagnosis of POM, seen for diagnosis and treatment at Moorfields Eye Hospital between 1991 and 2016, were identified from a clinical orbital diagnostic database and included in the study. The database comprised all patients seen by the orbital service at Moorfield’s Eye Hospital with a diagnosis of orbital malignancy, both primary and secondary. The histology slides were reviewed by an orbital malignancy-expert pathologist, and where diagnostic uncertainty existed, a second opinion was sought by a melanoma-expert pathologist. The radiological imaging and clinical case notes were reviewed. Metastasis from a distant site was excluded by a thorough clinical, ultrasonographic, and (where available) histological examination of both uveal tracts, with a complete skin survey by consultant dermatologists at the patient’s local hospital, by review of systemic health during the follow-up interval, and by CT, MR, and, in some cases, PET imaging. This study received ethics approval from Moorfields Eye Hospital Biobank ethics board (15/SW/0104).

Thirteen patients (five men; 38%) with POM were identified, their mean age at presentation being 55.5 years (median 48; range 40–84 years) (Table 2). Twelve of the patients were white European, while one patient was West African (Table 2). The left orbit was more commonly affected (eight cases; 62%), and three patients had an underlying localized pigmentary abnormality (one nevus of Ota, one conjunctival nevus, one oculo-dermal melanosis).

We present the clinical characteristics, treatment approaches, and long-term outcomes for 13 patients with POM, this representing the largest clinical series for this disease. Previous solitary case reports are consistent with our demographic findings—namely that most individuals are of white Northern European descent and present from the 5th decade (median age 48 years; range 40–84). One patient was West African and this would appear to be unique. Review of previously published cases suggests onset at a mean age of 44 (median 42 years), with the youngest case occurring in an 8-year-old girl and the oldest patient being 79 years (Table 1) (16, 38). Interestingly, two of our patients presented in their ninth decade, this being exceptionally late for POM. There does not appear to be any gender bias—with 19 affected men (14 previously reported + 5 in this study) and 18 affected women (10 previously reported + 8 in this study).

Imaging of patients consistently showed a well-circumscribed lesion that looked typically like a benign tumor or arteriovenous malformation; this characteristic has been previously described and might lead to delay in diagnosis and treatment of disease (39). MRI signal characteristics will generally help differentiate melanoma from benign lesions, such as cavernous hemangioma and could be considered in cases where there is diagnostic uncertainty (39). Our patients all had incisional biopsy and, where possible, resection of the mass, and all were considered for high-dose fractionated orbital radiotherapy to attempt control of residual local disease. Six patients underwent exenteration—three (Cases 2, 4, and 6) to control disfiguring orbital disease, two (Cases 5 and 8) for rapid disease progression after initial debulking, and one patient (Case 9) for very late recurrence of the orbital disease.

Despite reasonably uniform management, the outcome for this cohort of POM patients was highly variable: for example, two patients died very shortly after diagnosis, while three have survived 6 years or more. It would seem that this variable course does not depend solely on the presence of metastasis or systemic progression, as one of the longest surviving patients (Case 8) had partial hepatectomy for a liver metastasis. The surviving group (eight individuals, with three of these progressing to exenteration) was insufficiently large to draw conclusions about long-term visual outcomes in this patient group. There have, however, been other larger studies of more common orbital malignancies treated with a combination of surgery and radiotherapy that have shown good visual outcomes. For example, in a study of orbital rhabdomyosarcoma, one third of patients maintained vision better than 6/9 in their treated eye, and approximately half maintaining vision of 6/9–6/60 (40).

Choroidal melanomas have been broadly classified as “type I” or “type II,” these following “aggressive” or “relatively indolent” courses, respectively. Several genetic signatures have been found for “type I” tumors—with monosomy 3, present in a half of uveal melanomas, being the most significant chromosomal aberration and strongly associated with metastasis and death (41, 42). Monosomy 3 affects prognosis due to tumor haploinsufficiency of BAP1—an important BRCA1-associated tumor suppressor gene (43, 44). Other chromosomal abnormalities, such as loss of 6q and gain of 8q, have been associated with poor prognosis in uveal melanoma (41). Point mutations in GNAQ and GNA11 have been identified in 80–90% of uveal melanomas and lead to activation of the MAPK/MEK/ERK pathway (43). It would be valuable to establish whether these genetic aberrations are present in POMs, and whether these relate to prognosis.

One patient (Case 9) is particularly interesting: despite receiving only orbital radiotherapy for widely infiltrating melanoma at the time of diagnosis (Figures 2A,B), she remained with no evidence of tumor proliferation until more than 13 years later. After tumor reactivation, however, it followed a very aggressive course (Figure 2C), requiring several palliative procedures, and the patient died 13 months later. Late metastatic or secondary melanoma to the orbit has been reported (45, 46), and similar genetic factors might possibly control the late recurrence of both primary and secondary orbital melanomas; alternatively, the orbital milieu might pre-dispose to a prolonged tumor latency before late recurrence.

In summary, POM is an extremely rare malignancy of rather variable prognosis after treatment, for which local resection with adjuvant radiotherapy remains the mainstay of therapy. Disease can remain quiescent for extended periods of time before following an aggressive course. Immunotherapy might play a role in the future, but in this cohort, genetic testing did not suggest response to currently available agents in the majority of individuals. Further genetic investigation of these rare tumors might elucidate underlying molecular mechanisms of oncogenesis, thereby improving prognostication and treatment for this patient group.

This study received ethics approval from Moorfields Eye Hospital Biobank ethics board (15/SW/0104).

Concept and design of study—AR, CJ, DV, and GR. Clinical/histological examination of patients—PL, DV, and GR. Data collection—AR. Data analysis—AR and GR. Manuscript preparation and review—AR, PL, CJ, DV, and GR.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.