Periocular invasive melanoma manifestation in a patient using bimatoprost: case report and literature review

ABSTRACT Prostaglandin F2a analogs (PGAs) are considered efficacious in the first-line treatment of glaucoma. They have however been associated with a number of periocular side effects. We present a case of periocular hyperpigmentation and progression to lentigo maligna melanoma (LMM) in a patient using bimatoprost eye drops. We conducted a literature review regarding the etiology and pathophysiology of periocular pigmentation in this setting. A 71-year-old female Caucasian patient with open-angle glaucoma using bimatoprost exclusively in her right eye noticed an ipsilateral lower eyelid/upper cheek area dark lesion after commencing treatment. Examination demonstrated a heterogeneously pigmented lesion. Excisional biopsy demonstrated extensive lentigo maligna (melanoma in situ) with superficially invasive malignant melanoma in the lesion center. The patient underwent successful staged excision and reconstruction. Literature review has demonstrated case reports supporting periocular hyperpigmentation; however, there has been no description of progression to periocular lentigo maligna and melanoma in a patient using bimatoprost.


Introduction
Prostaglandin analogs (PGAs) selectively mimic endogenous prostamides. They are thought to lower intraocular pressure (IOP) by increasing the outflow of aqueous humor through both the trabecular meshwork and uveoscleral drainage systems.
PGAs have previously been associated with hyperpigmentation of the periocular skin. These medications also result in a spectrum of periocular side effects such as trichomegaly, eyelid margin hyperemia, iris darkening, and prostaglandin-associated periorbitopathy (PAP). [1][2][3][4] These changes are usually temporary and can be reversed by stopping topical treatment. Some, however, such as PAP and iris and skin pigmentation, may be more permanent. 3,5 We present a case of periocular hyperpigmentation and progression to lentigo maligna melanoma (LMM) in a patient using bimatoprost eye drops. We discuss the possible manifestation of LMM and invasive melanoma in pre-existing hyperpigmentation through bimatoprost use. We provide a review of the current literature regarding etiology, pathophysiology, and differential diagnosis of periocular pigmentation.

Methods
This case report and review followed the Nova Scotia Health Authority Research Ethics Board guidelines and was adherent to the Declaration of Helsinki. A literature review was conducted using the PubMed advanced search features with the following keywords and their synonyms: prostaglandin analog (and specifically the various analogue medication names), periorbital, hyperpigmentation, melanoma, lentigo maligna. Please see supplemental information for search parameters used. This produced 52 articles of which all articles with novel data and at minimum an English abstract that contained sufficient information, were included.

Case report
A 71-year-old female Caucasian patient, with a 3-year history of open-angle glaucoma using bimatoprost exclusively in her right eye, noticed a right lower eyelid/upper cheek area dark lesion. She stated that this lesion had appeared during the period of eye drop use and had grown in size with gradual darkening. Medical records indicated that bimatoprost 0.03% was started in January 2011 and treatment was terminated in November 2014. In June 2014, a dark area of pigmentation on the right lower eyelid was first noted. Our patient confirmed regular use of bimatoprost, including during evolution of the unilateral pigmentation of her skin. This was confirmed with patient photography prior to the period of topical treatment. On account of eye irritation, she was told to discontinue bimatoprost and start a combination of travoprost and timolol maleate eye drops bilaterally.
The patient had a previous clinical history of hyperthyroidism treated with radioiodine with resultant hypothyroidism. She also had Meniere's disease, fibromyalgia, osteoarthritis and had an episode of transient ischemic attack many years previously. Her medication history included Eltroxin, Furosemide, Serc, and ASA. She denied any history of immunosuppression, smoking, blistering sunburn, or any other malignancy. Her past ocular history was unremarkable, and there was no history of eyelid surgery or trauma.
Examination revealed a best-corrected visual acuity of 6/9-1 OD and 6/9-2 OS, normal intraocular pressures, clear corneas, and early cataracts in both eyes. She had features of prostaglandin-associated periorbitopathy (PAP) on the right side and dermatochalasis on the left upper eyelid (Figure 1). Her retinal examination demonstrated no abnormalities. There was no clear lymphadenopathy on palpation of the preauricular, submandibular, and neck areas.
The right lower eyelid/upper cheek area demonstrated a heterogeneous pigmented lesion including a component at the medial eyelid sparing the subciliary region and the eyelid margin. The skin lesion appeared to have the shape of tear drop spread on the skin as it would run down on the eyelid. A homogeneous darkcolored lesion lateral to the inferior punctum, which clinically resembled a nevus was contiguous to this main lesion. There were no pigmented lesions on the right upper eyelid or the left upper and lower eyelids. Pan-facial examination did not demonstrate any other areas of pigmentation or similar lesion.
A 4 mm punch biopsy of the central part of the lesion was performed and submitted for histopathological examination in formalin. This exhibited melanoma in situ. The patient was planned for a staged excision and reconstruction using a modified square technique. At the following visit, an excisional biopsy with marked margins was performed, and the subsequent defect is shown in Figure 2. Histopathology demonstrated extensive lentigo maligna (melanoma in situ) and superficially invasive malignant melanoma with a Breslow thickness of 0.2 mm (Figure 3). On margin clearance, subsequent reconstruction was performed using a full thickness skin graft from the left preauricular region combined with direct closure and tissue advancement ( Figure 4).
The final histopathological diagnosis was superficially invasive malignant melanoma (stage pT1a) arising centrally within melanoma in situ, with the latter involving marked peripheral margins on the main excision specimen. Clear margins were obtained around the excision. The adjacent lesion below the punctum clinically thought to be a nevus was instead found to be an incidental simple lentigo (benign pigmented lesion) completely unrelated to the in situ and invasive melanoma on microscopy. There was no evidence of melanoma arising in a nevus. The patient was discussed at the regional multidisciplinary cancer site meeting and no further investigation or intervention was deemed necessary, although prolonged follow-up was advised and currently continues.

Discussion
Recently, the focus for first-line treatment to lower IOP has been on prostaglandin F 2a (PGF 2a ) analog eye drops such as latanoprost, bimatoprost, and travoprost for their overall low systemic side effects with high efficacy and tolerability. 6,7 Side effects of PGF 2a analogs are generally limited to ocular and periocular changes including increased iris pigmentation, ocular hyperemia, eyelash lengthening (trichomegaly)/bristling, and periocular skin discoloration. These changes reverse in a few weeks-months after stopping use, 1,2,[8][9][10][11] however iris hyperpigmentation appears to remain a permanent change. 1,11,12 These side effects, although usually mild, may cause patients to discontinue PGF 2a analog use despite their efficacy. In our patient, subtle unilateral periocular changes related to PAP provided objective evidence of patient compliance with the medication.
Studies have also looked beyond ocular changes and focused on the periocular changes from prolonged use of PGF 2a analogs, including skin changes. 1,3,5,8,11,[13][14][15][16][17][18] These changes range from hyperpigmentation and hypertrichosis to orbital changes such as deepening of the upper eyelid sulcus (DUES). Hyperpigmentation has been noted from 0% to 26% incidence in adults to as high as 100% in children. 12,[19][20][21] It has been suggested that the frequency of hyperpigmentation is related to the length of administration of the PGF 2a analog. There have been no reports to date that investigate a relationship between skin pigmentation and age, gender, race, or administration length, 3,18,22 although some preliminary work suggests a higher trend of hyperpigmentation after 6 months of use in Caucasians and females. 22 The majority of documented reports of periocular hyperpigmentation have been based on numerous case reports 23,24,25,[26][27][28][29][30][31][32][33][34] . There is a paucity of information regarding the pathophysiology of this side effect. Hyperpigmentation is most noticeable in the skin of the eyelids and the changes become apparent after approximately 3-6 months of use. 11 However, reports have described pigmentation changes as early as one week, and as late as three years after starting use. 8,11,35 Pigmentation may not be isolated to the eyelids and may also include periorbital facial skin. Indeed, our patient initially presented with hyperpigmentation of the lower eyelid and the upper cheek with a characteristic pattern following the medication path down the central eyelid onto the cheek area. Other reports include hyperpigmentation of the patient's upper cheek after being grafted from behind their ear, 14 as well as progressive hyperpigmentation starting at the eyelids and spreading to the cheeks, 16,36 and diffuse bilateral facial hyperpigmentation. 35 Fortunately, changes to the skin pigmentation typically resolve in a matter of months after stopping use. 16 However, we present a case of a patient receiving bimatoprost that later developed lentigo maligna and melanoma, suggesting that the hyperpigmentation may progress thereby making an underlying malignant entity more apparent to the clinician.
Our patient had radio-iodine treatment of hyperthyroidism. An extended 24-year follow-up of the Cooperative Thyrotoxicosis Therapy Follow-up Study demonstrated that radioactive iodine-treated patients with hyperthyroidism and greater organ-absorbed doses of iodine had a modestly positively associated risk of death from solid cancer, including breast cancer. 23 However, in a cohort study with 16,637 patients, no association was found between radioiodine treatment and increased risk of overall cancer in longterm follow-up. 37 Most notably, no association is mentioned in the literature between hyperthyroidism treatment with radioactive iodine (I-131) and periocular skin malignancies.
The mechanism(s) by which PGF 2a analogs induce skin hyperpigmentation are not yet fully understood. There are four potential drug-induced mechanisms PGF 2a analogs may activate to cause hyperpigmentation: (1) accumulation of melanin through melanin overproduction, cutaneous inflammation, or reduced melanin clearance; (2) accumulation of the drug; (3) synthesis of new pigment, such as lipofuscin; (4) iron deposition in the dermis. [38][39][40] Prostaglandins in general are potent stimulators of melanogenesis. 28,41 Melanocytes produce melanin in the epidermis, which is then transferred to neighboring keratinocytes. These keratocytes are then discarded by the epidermal cell cycle. This can account for the reversibility of hyperpigmentation once the PGF 2a analog treatment has been discontinued. PGF 2a analogs are thought to increase tyrosinase activity (the ratelimiting enzyme that oxidizes tyrosine to produce melanin). 35 As well, prostaglandins have been implicated in post-inflammatory skin hyperpigmentation. 14 Taken together, these mechanisms can drive skin hyperpigmentation in patients using a PGF 2a analog. PGF 2a analogs appear to only affect aspects of melanogenesis but not melanocyte proliferation. In one study, a skin biopsy of a patient from bimatoprost-induced hyperpigmentation was analyzed by electron microscopy. 42 This demonstrated a variable abundance of mature melanosomes in melanocytes and basal keratinocytes suggesting an increase in melanogenesis, yet there was no indication of an increased melanocyte cell number. Another study showed that in vitro treatment of melanocytes with latanoprost did not elicit a proliferative effect on the cells. 29 Lastly, latanoprost has not been shown to increase the mitotic index of human cutaneous melanoma cells. 43 Taken together, PGF 2a analogs appear to increase melanogenesis through an increased expression of tyrosinase and possibly post-inflammatory processes; however, this does not help to determine why the frequency of patients with hyperpigmentation is so variable.
To date, there have been reports of PGF 2a analoginduced hyperpigmentation in Asian, Hispanic, Caucasian, and African-American populations, 8,16,35,44 therefore suggesting multiple skin types can be affected. One study focused on Fitzpatrick skin types in relation to skin hyperpigmentation from bimatoprost and latanoprost use. 5 This suggests possibly a higher incidence of hyperpigmentation in patients with a fairer skin type, although presence of a fairer skin type may also allow earlier recognition and clinical diagnosis of hyperpigmentation.
The general incidence of melanoma of the skin for the general population in the United States and Canada are 22.8 and 20.8 per 100,000, respectively. 31,32 A review of the literature using PubMed advance search features were used for incidence of melanoma in patients using any PGF 2α analog. There does not appear to be literature available with this information. As far as the authors are aware, there have been four other documented cases of PGF 2a analog use and associated melanoma. 30,45,46 Three cases were reported by the French Health Authority which did not note any medical history. The article describes two patients developing choroidal melanoma and one patient developing lentigo maligna of the lower eyelid during the use of topical latanoprost. 30 The fourth case has been described in a patient with periocular melanoma and included some history, which demonstrated a different case presentation compared to our patient. 45 This latter case reported the development of a nevus on the bottom eyelid of the patient that was initially diagnosed as a benign nevus. The nevus continued to grow in size and thickness and was later removed through excisional biopsy and identified as a melanoma in situ with no evidence of dermal invasion. Our patient developed a heterogeneous teardrop-shaped lesion that was clinically incompatible with a nevus. This was subsequently confirmed by histology to be extensive melanoma in situ with superficially invasive malignant melanoma. The only documented similarities between these two cases are the patients' gender (female), age (71 and 77), and use of bimatoprost. The patient described by Sun et al. was using bimatoprost in both eyes for 6 years whilst the patient in our report was using the drops in the right eye only and for 3 years. Unfortunately, other patient demographics, medical history, or medications were not listed for comparison nor were specific risk factors for the development of melanoma. Specifically, our patient had few particular risk factors that may have contributed to the development of lentigo maligna melanoma, which include older age (71years-old), the use of Furosemide, which is a known photosensitive compound, 33 and Fitzpatrick skin type II. 34 Interestingly, also in our case, the size and shape of the lesion resembled that of the medication running from the eye down the eyelid and cheek. The location of the eyelid margin component in our case, confirmed as lentigo maligna (melanoma in situ), would be in keeping with initial movement of the medication via the lacrimal pump medially before spread down the lower eyelid. Tressler et al. conducted a review of the literature to determine an association between the use of latanoprost and melanoma. 46 They concluded that whilst there is not enough evidence to directly link a risk of latanoprost treatment with ocular or cutaneous melanoma, it could also not be completely excluded.
Exposure to indoor tanning before 35 years of age raises lifetime risk of melanoma by 75%. 47 Moreover, it is estimated that relative risk for melanoma increases from 1.47 for individuals with 16-40 nevi, to 6.89 in people with 101-120 nevi. 27 To our knowledge, there are no data available on incidence of melanoma in patients using PGF 2a analogs. Therefore, there have been no studies comparing incidence of melanoma in the general public compared to users of PGF 2a analogs.

Conclusion
PGF 2a analogs offer an effective IOP lowering treatment for open-angle glaucoma and are generally well tolerated. Side effects from treatment include PAP, hyperemia, trichomegaly, and hyperpigmentation. The onset of these effects remains variable and is considered largely reversible once the medication has been discontinued. Therefore, patient education should be considered before the administration of therapy. Including our patient, there are now two case reports that demonstrate the development of melanoma associated with bimatoprost use. To our knowledge, our report is the first to describe a patient using bimatoprost who demonstrated invasive melanoma on the ipsilateral treated side in a distribution similar to drop spread. It should be noted, however, that causal association is not proven without attention to epidemiological consideration. However, this may prompt further investigation looking into patients diagnosed with melanoma and what medications they may be on, including any PGF 2α analogs. Hyperpigmentation of a previously developing lesion, which came to clinical attention via bimatoprost use is also plausible. Based on our case presentation and review, we suggest that patient follow-up should specifically include examination for pre-existing periocular changes and hyperpigmentation including documentation and clinical photography. As demonstrated, in some patients using PGF 2a analogs, development of new or darkening pigmentation may herald more sinister pathology requiring early recognition and management.