Emerging strategies to treat rare and intractable subtypes of melanoma

Abstract

Melanoma is the deadliest form of skin cancer, possessing a diverse landscape of subtypes with distinct molecular signatures and levels of aggressiveness. Although immense progress has been achieved therapeutically for patients with the most common forms of this disease, little is known of how to effectively treat patients with rarer subtypes of melanoma. These subtypes include acral lentiginous (the rarest form of cutaneous melanoma; AL), uveal, and mucosal melanomas, which display variations in distribution across (a) the world, (b) patient age-groups, and (c) anatomic sites. Unfortunately, patients with these relatively rare subtypes of melanoma typically respond worse to therapies approved for the more common, non-AL cutaneous melanoma and do not have effective alternatives, and thus consequently have worse overall survival rates. Achieving durable therapeutic responses in these high-risk melanoma subtypes represents one of the greatest challenges of the field. This review aims to collate and highlight effective preclinical and/or clinical strategies against these rare forms of melanoma.

1 |. INTRODUCTION

The melanoma field represents a paradigm for preclinical and clinical advancements in targeted and immune therapy modalities, with 13 new FDA-approved therapies since 2011 (Dummer et al., 2018a, 2018b; Long et al., 2018; Ribas & Flaherty, 2011). The catalyst for the development of targeted therapy modalities was the identification of activating NRAS mutations and BRAF mutations in 1984 (Albino, Le Strange, Oliff, Furth, & Old, 1984; Padua, Barrass, & Currie, 1984) and 2002 (Davies et al., 2002), respectively, which paved the way for molecular stratification of the melanoma patient population. Approximately 45%–50% of non-acral lentiginous (AL) cutaneous melanoma patients have tumors that harbor activating BRAF mutations (Schadendorf et al., 2017), with a single amino acid substitution of valine for glutamic acid at codon 600 (V600E) occurring in 90% of cases (Figure 1a). Activating NRAS mutations at codon 12, 13, or 61 are detectable in 15%–20% of non-AL cutaneous melanoma patients and serve as an independent predictor of worse patient overall survival (Jakob et al., 2012). Mutations of BRAF and NRAS are considered mutually exclusive (Hodis et al., 2012); however, there are rare reports where both mutations exist in different regions of the same tumor or at different metastatic sites of the same patient (Wilmott et al., 2012). To date, it remains unclear whether the same melanoma cell can harbor both a BRAF and an NRAS mutation, or at the single-cell level, these mutations are indeed mutually exclusive.

FIGURE 1.

The frequency of activating BRAF, NRAS, and GNAQ/GNA11 mutations in (a) SSM, (b) acral lentiginous, (c) mucosal, (d) uveal, (e) nodular, and (f) lentigo maligna melanoma

With discoveries revealing that ~70% of non-AL cutaneous melanomas contain mutations constitutively activating the mitogen-activated protein kinase (MAPK) pathway came intense development of inhibitors capable of targeting various nodes of the mitogen-activated protein kinase (MAPK) pathway (i.e., BRAF, MEK, and ERK inhibitors) that continues to date (English & Cobb, 2002). The first targeted therapy approved for the treatment of patients with BRAF^V600E/K mutant melanoma was the small molecule inhibitor vemurafenib, an agent designed to have high specificity against the mutant V600E, V600K, V600D, and V600R forms of BRAF (Kim et al., 2014). Vemurafenib had response rates of ~48% in phase II and III clinical trials leading to the 2011 Food Drug and Agriculture (FDA) approval. A few years later, the combination of a BRAF inhibitor and a MEK inhibitor was observed to further increase the response rate to ~76% leading to the 2014 FDA approval of dabrafenib and trametinib (Paraiso et al., 2010; Queirolo, Picasso, & Spagnolo, 2015). There are now three BRAF inhibitor plus MEK inhibitor combinations FDA approved for melanoma patients with BRAF^V600E/K mutations (dabrafenib/trametinib, vemurafenib/cobimetinib (Larkin et al., 2014; Ribas et al., 2014), encorafenib/binimetinib (Dummer et al., 2018a, 2018b).

For patients with wild-type BRAF, treatment with BRAF inhibitors that specifically target V600E/K mutant BRAF may increase melanoma aggressiveness due to the paradoxical activation of wild-type BRAF and downstream MAPK pathway signaling (Carnahan et al., 2010; Hatzivassiliou et al., 2010; Heidorn et al., 2010; Joseph et al., 2010; Poulikakos, Zhang, Bollag, Shokat, & Rosen, 2010). Preclinically, targeting downstream of BRAF with MEK inhibitors in BRAF-wild-type melanoma cells demonstrates the importance of the MAPK pathway for their survival, with significant anticancer activity (Rebecca, Alicea, et al., 2014). However, clinical trials testing multiple MEK inhibitors (i.e., binimetinib, trametinib) have concluded that although encouraging response rates and small increases in progression-free survival could be achieved in certain trials relative to dacarbazine, no significant increase in overall survival of patients with BRAF-wild-type melanoma was achieved with MEK inhibition (van Herpen et al., 2019). In an effort to increase MEK inhibitor efficacy, combination strategies with other agents (i.e., PI3K inhibitors (Greger et al., 2012), CDK4/6 inhibitors (Kwong et al., 2012)) are being clinically tested in the BRAF-wild-type (i.e., patients with or without NRAS-MT melanoma) setting after failure of immunotherapy (Dummer et al., 2017). ERK inhibitors are also being clinically investigated to see if durable efficacy can be achieved in patients with wild-type BRAF, with reports showing the first-in-class ERK1/2 inhibitor ulixertinib has an acceptable safety profile and early evidence of clinical activity (Moschos et al., 2018; Sullivan et al., 2018). Preclinical evidence suggests that concurrent inhibition of multiple nodes of the MAPK pathway in NRAS-mutant melanoma (i.e., MEK and ERK) may have synergistic activity on par with the BRAF inhibitor and MEK inhibitor combination in BRAF-mutant melanomas, and further studies evaluating this strategy are under way (Rebecca, Alicea, et al., 2014).

In parallel, large strides have been made in the development of immune checkpoint blockade strategies with the FDA approval of antibodies targeting cytotoxic T-lymphocyte antigen 4 (CTLA4, ipilimumab) in 2011 (Ledford, 2011) and programmed cell death 1 (PD1, pembrolizumab, nivolumab) in 2014 (Mahoney, Freeman, & McDermott, 2015; Schachter et al., 2017) and the combination of ipilimumab and nivolumab in 2015 (Camacho, 2015). Immune checkpoint blockade describes the use of therapeutic antibodies that overcome immunosuppressive checkpoints with the goal of unchaining antitumor immune responses (Kalbasi & Ribas, 2020). CTLA4 and PD-1 are both receptors that suppress effector T-cell activity (Blank et al., 2004; Chen et al., 1992; Dong et al., 2002). These immunotherapy-based strategies elicit long-lasting responses in a subset of patients and represent a therapeutic strategy suitable for all genotypes of non-AL cutaneous melanoma. However, the majority of patients treated with immunotherapy progress within 5 years due to poorly understood primary resistance mechanisms, and clinicians still cannot reliably discriminate which patients will respond or not respond. Both tumor intrinsic (i.e., insufficient tumor antigenicity, tumor interferon-γ signaling, tumor stemness) and extrinsic (i.e., regulatory T cells, myeloid-derived suppressor cells) resistance mechanisms have been reported, and there are intense efforts focused on overcoming these therapeutic hurdles to further increase the efficacy of immune checkpoint blockade strategies.

The promising efficacy of these new therapeutic strategies has been demonstrated largely in non-AL cutaneous melanoma patients with either superficial spreading melanoma (SSM), nodular melanoma (NM), or lentigo maligna melanoma (LMM). SSM, NM, and LMM represent the most common forms of melanoma in Caucasians (>85% of cases). It is important to appreciate that most of the recent pivotal discoveries in melanoma were performed on SSM cell lines, short-term cultures, animal models, and tumor biopsies taken from patients with SSM largely due to their greater availability (personal communication from Dr. Meenhard Herlyn). AL melanoma represents the fourth and rarest subtype of cutaneous melanoma. In addition, mucosal melanoma and uveal melanoma are other rare subtypes of melanoma that are non-cutaneous in origin. The efficacy of immune checkpoint blockade is lower in rarer subtypes of melanoma relative to patients with non-AL cutaneous melanoma, which will be discussed later. There is also little information regarding the efficacy of combination BRAF inhibitor and MEK inhibitor therapy in these subtypes. In the remainder of the review, we will flesh out details regarding mutational landscapes, etiology, and efficacy of targeted and immune therapy options for patients with rare subtypes of melanoma and draw attention to developments in each respective case in the way of emerging therapeutic strategies with promise to increase the overall survival of patients with intractable rare subtypes of melanoma.

2 |. ACRAL LENTIGINOUS: THE MOST UNCOMMON OF THE 4 MA JOR CMM SUBTYPES IN CAUCASIANS

2.1 |. Incidence/Presentation

Acral lentiginous melanoma is an uncommon yet relatively aggressive subtype of CMM that accounts for 2%–3% of all melanoma cases. AL melanoma arises on sun-protected, glabrous skin of the soles, palms, and nail beds. AL melanoma has been historically associated with worse 10-year survival rates relative to other forms of CMM (67.5% vs. 87.5%; Bradford, Goldstein, McMaster, & Tucker, 2009). Further, 10-year AL melanoma survival rates are highest in non-Hispanic Whites (69.4%), intermediate in Blacks (71.5%), and lowest in Hispanic Whites (57.3%) and Asian/Pacific Islanders (54.1%), as found by the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute evaluating data from 17 population-based cancer registries from 1986 to 2005 (Bradford et al., 2009). Another analysis of AL melanoma prognostic features in a cohort of German, Swiss, and Austrian patients suggests no significant difference exist relative to other subtypes of cutaneous melanoma; however, this conclusion may stem due to differential ethnicity landscapes between this patient cohort and that in the SEER study (Teramoto et al., 2018). There does not appear to be a gender bias, with a similar frequency between men and women and a comparable median age of diagnosis of 63.1 years for men and 62.2 years for women. The incidence of AL melanoma increases with age, and for reasons poorly understood, men are twice as likely to develop AL melanoma relative to women after the age of 80 (American Cancer Society, 2019).

The distribution of AL melanoma varies geographically among populations throughout the world (Figure 2). While AL melanoma represents only ~2%–3% of all melanoma cases in Caucasian populations, AL melanoma makes up 50%–80% of all cases in non-Caucasian individuals in the United States (i.e., those of African, Latin American, and Asian descent) (Cascinelli et al., 1994; Chang et al., 2004; Nakamura & Fujisawa, 2018; Wang, Zhao, & Ma, 2016). Furthermore, the incidence in Hispanic Whites doubles compared to non-Hispanic Whites after the aged of 70 (Bradford et al., 2009). A 2009 SEER study found the overall incidence rates of AL melanoma were similar between non-Hispanic Whites and Blacks; however, Hispanic Whites have statistically higher incidence rates relative to non-Hispanic Whites (Bradford et al., 2009). Updated epidemiological studies should be performed to continue understanding the differential incidence trends that may exist across different ethnicities. Of note, the incidence of other subtypes of cutaneous melanoma (i.e., NM, SSM) is much lower in non-Caucasians relative to Caucasians. As this subtype of melanoma is not related to ultraviolet radiation (UV), there are different theories of the cause of AL melanoma. Some reports state that trauma and pressure in the foot (a predilected area of AL) is causal (Costello, Pittelkow, & Mangold, 2017). However, the hand is also exposed to trauma but its location is less favorable. The main sites of AL melanoma metastases are the lungs, distant lymph nodes, scalp, contralateral limb, and liver (Kwon et al., 2019).

FIGURE 2.

A schematic depicting the common sites of superficial spreading, nodular, lentigo maligna, acral lentiginous, mucosal, and uveal melanomas

2.2 |. Mutational/molecular profile

Acral lentiginous melanomas possess a significantly lower mutational burden relative to the more common cutaneous melanoma subtypes, likely due to the sun-protected locations they arise from (Furney et al., 2014). BRAF mutations in are found in 1 in every 5 Al melanoma patients, leaving ~80% ineligible to receive BRAF inhibitor and combination BRAF/MEK inhibitor strategies (Curtin et al., 2005; Figure 1b). Therefore, new targets specific for AL melanoma are needed. 80% of AL melanomas display genetic aberrations of cyclin-dependent kinase 4/6 (CDK4/6) pathway-related genes (i.e., amplification of CDK4 and CCND1, and/or loss of CDK2NA), representing the most frequent copy number alteration detected (Kong et al., 2017). Additionally, activating KIT mutations are present in ~6% of cases (Zaremba et al., 2019). AL melanoma displays similar incidence of NRAS mutations as non-AL cutaneous melanoma, detectable in 15%–28% of AL melanoma patients, and NRAS mutations are an independent prognostic factor of worse overall survival (Kong et al., 2017; Mandala, Merelli, & Massi, 2014; Zaremba et al., 2019).

2.3 |. Treatment/Efficacy

Considerable barriers exist to treat patients with AL melanoma: (a) a contrasting genomic and genetic landscape relative to non-AL cutaneous melanomas, (b) unclear targetable drivers, and (3) sparse experimental models available for preclinical drug development (Turajlic et al., 2012). Unfortunately, FDA-approved targeted therapy strategies for melanoma are not available for the majority of AL melanoma patients (i.e., BRAF inhibitors since AL melanoma has a low frequency of BRAF mutations), and the efficacy of immune checkpoint blockade strategies is not well known in AL melanoma, with differing overall response rates (ORR) differing by country. For example, the ORR of anti-PD-1 in AL melanoma patients was found to be similar to that in non-AL cutaneous melanoma patients within the United States (Shoushtari et al., 2016). In contrast, the ORR was 66.7% for SSM patients and 28.6% of AL melanoma patients in a recent Japanese study, suggesting the efficacy of immune checkpoint blockade may vary with ethnicity (Yamazaki et al., 2019). The lower mutational burden observed in AL melanoma cases is thought to drive the reduced efficacy of immune checkpoint inhibitor strategies (e.g., PD-1 blockade) in patients (Furney et al., 2014). Although AL melanoma patients with Kit mutations can be treated with a KIT inhibitor per National Comprehensive Cancer Network (NCCN) guidelines, resistance mechanisms that reactivate downstream MAPK and PI3K pathway signaling have been suggested to blunt long-term durability (Table 1; Carlino, Todd, & Rizos, 2014). Due to the high percentage of AL melanoma tumors with CDK4/6-pathway aberrations, CDK4/6 inhibition represents one of the most promising targeted therapy strategies for AL melanomas clinically (NCT03454919). However, durable responses are not observed in all patients due to resistance and CDK4/6 inhibitor-based combinations will likely be needed to improve the curative rate for patients with AL melanoma (Table 1). Preclinical investigation to optimize targeted therapy strategies has not been extensively performed in AL melanoma models, but the rich body of literature that exists from studies in non-AL cutaneous melanoma models strongly suggests that single-agent approaches will not be durable due to the nearly universal onset of resistance (Rebecca, Alicea, et al., 2014; Rebecca, Wood, et al., 2014). In SSM models, treatment with a MAPK pathway inhibitor plus a CDK4/6 inhibitor has shown synergistic activity in BRAF-MT (Martin et al., 2018) and BRAF-wild-type settings (Teh et al., 2018); however, residual disease persists. Resistance mechanisms to CDK4/6 inhibitors and/or MEK inhibitors must be delineated to develop combination strategies that produce durable responses in AL melanoma patients.

TABLE 1.

Current clinical trials tailored for patients with acral lentiginous, mucosal, or uveal melanoma

Subtype	Agent(s)	Agent target(s)	ClinicalTrials.gov identifier	Phase
Nodular	RO4929097	Gamma secretase/Notch signaling pathway	NCT01120275	II
Nodular	Dinaciclib	CDK	NCT00937937	II
Nodular	RSV-TK, ganciclovir	Adenovirus, antiviral	NCT00005057	I
Lentigo maligna of head and neck	Imiquimod cream + surgery	TLR7	NCT01720407	III
Lentigo maligna of the face	Picato gel	Protein kinase C	NCT02723721	II
Lentigo maligna	5-Aminolevulinic acid nanoemulsion	Light sensitizing	NCT02685592	IV
Acral	SHR-1210, apatinib	Anti-PD-1, VEGFR2	NCT03955354	II
Mucosal, acral, vulvovaginal	Dasatinib	BCR/ABL, SRC	NCT00700882	II
Acral, mucosal	PLX3397	CSF-1	NCT02071940	II
Mucosal, acral, CSD	Nilotinib	BCR/ABL	NCT00788775	II
Acral, mucosal	Nilotinib	BCR/ABL	NCT0135121	II
Acral, mucosal	Nivolumab ± ipilimumab	Anti-PD-1, anti-CTLA-4	NCT02978443	II
Acral, mucosal	Sunitinib	VEGFR, PDGFR	NCT00577382	II
Acral, mucosal, CSD	Imatinib	ABL, c-Kit, PDGFR	NCT00424515	II
Acral	TQB2450 ± anlotinib	Anti-PD-L1, VEGFR, FGFR, PDGFR, c-Kit	NCT03991975	I, II
Acral	Palbociclib	CDK4, CDK6	NCT03454919	II
Mucosal of head and neck	Radiation	DNA	NCT03138642	II
Mucosal	Toripalimab, IFN a-2B	Anti-PD-1	NCT03178123	II
Mucosal	CM082 plus Toripalimab	Multikinase inhibitor, anti-PD-1 monoclonal	NCT03602547	II
Mucosal	Ipilimumab, nivolumab	anti-CTLA-4, anti-PD-1	NCT03241186	II
Mucosal	Apatinib, SHR01210	VEGFR2, c-Kit, c-Src, anti-PD-1	NCT03986515	II
Uveal	Vorinostat	HDAC	NCT03022565	I
Uveal	SIR-Spheres® yttrium 90, ipilimumab, nivolumab	Radiation, anti-CTLA-4, anti-PD-1	NCT02913417	I/II
Uveal	BVD-523	ERK1/2	NCT03417739	II
Uveal	Autologous dendritic cells loaded with autologous tumor RNA	Tumor vaccine	NCT01983748	III
Uveal	Selumetinib	MEK1/2	NCT02768766	I
Acral	SHR-1210, apatinib	Anti-PD-1, VEGFR2	NCT03955354	II
Mucosal, acral, vulvovaginal	Dasatinib	BCR/ABL, SRC	NCT00700882	II
Acral, mucosal	PLX3397	CSF-1	NCT02071940	II
Mucosal, acral, CSD	Nilotinib	BCR/ABL	NCT00788775	II
Acral, mucosal	Nilotinib	BCR/ABL	NCT0135121	II
Acral, mucosal	Nivolumab ± ipilimumab	Anti-PD-1, anti-CTLA-4	NCT02978443	II
Acral, mucosal	Sunitinib	VEGFR, PDGFR	NCT00577382	II
Acral, mucosal, CSD	Imatinib	ABL, c-Kit, PDGFR	NCT00424515	II
Acral	TQB2450 ± anlotinib	Anti-PD-L1, VEGFR, FGFR,	NCT03991975	I, II
Acral	Palbociclib	CDK4, CDK6	NCT03454919	II
Mucosal of head and neck	Radiation	DNA	NCT03138642	II
Mucosal	Toripalimab, IFN a-2B	Anti-PD-1	NCT03178123	II
Mucosal	CM082 plus toripalimab	Multikinase inhibitor, anti-PD-1 monoclonal	NCT03602547	II
Mucosal	Ipilimumab, nivolumab	anti-CTLA-4, anti-PD-1	NCT03241186	II
Mucosal	Apatinib, SHR01210	VEGFR2, c-Kit, c-Src, anti-PD-1	NCT03986515	II
Uveal	Vorinostat	HDAC	NCT03022565	I
Uveal	SIR-Spheres® yttrium 90, ipilimumab, nivolumab	Radiation, anti-CTLA-4, anti-PD-1	NCT02913417	I/II
Uveal	BVD-523	ERK1/2	NCT03417739	II
Uveal	Autologous dendritic cells loaded with autologous tumor RNA	Tumor vaccine	NCT01983748	III
Uveal	Selumetinib	MEK1/2	NCT02768766	I

3 |. MUCOSAL MELANOMA: ONE OF THE RAREST FORMS OF MALIGNANT MELANOMA

3.1 |. Incidence/Presentation

Mucosal melanoma (MM) is one of the rarest types of melanoma, accounting for only 1% of all cases (Dumaz et al., 2019; McLaughlin et al., 2005), and has a significantly worse prognosis relative to the other subtypes (Dumaz et al., 2019). Distinct from cutaneous melanoma, MM arises from melanocytes located in mucosal membranes inside the body (i.e., genitourinary, anorectal, nasopharyngeal; Dumaz et al., 2019). The head and neck (55), vulva (18), and anus (24) are the most common observed sites; however, MM can also occur in the gut, lungs, and urinary track (Figure 2; Chang, Karnell, & Menck, 1998). It is rarely diagnosed at early stages due to difficult visual detection, which is much more tractable for cutaneous subtypes of melanoma. The overall median age of diagnosis is 70 years, with the exception of MMs arising in the mouth that manifest more frequently in younger patients (Tyrrell & Payne, 2018). The incidence of MM has been stable for the last few years with the exception of MM in the genital tract, which is higher in females relative to males for reasons not clearly understood (McLaughlin et al., 2005).

3.2 |. Mutational/Molecular profile

Approximately 3%–15% of MMs harbor an activating mutation in BRAF, with ~63% located on the V600 codon and 37% located on a non-V600 codon (Figure 1c; Dumaz et al., 2019). This is in contrast to non-AL cutaneous melanomas where <10% of BRAF mutations are outside of the V600 codon, and more closely resembles the high prevalence of non-V600 mutations found in 48% of lung adenocarcinomas. A closer analysis of the most common non-V600 mutations reveals (a) a difference between the frequency of mutations on D594, G469, and K601 between non-AL cutaneous melanomas and MMs, and (b) convergence in the non-V600 mutational landscape between MM and lung cancers where mutations are often associated with genotoxic agents (Dumaz et al., 2019).

In regard to NRAS mutations, approximately 12% of MMs harbor activating mutations, which is lower relative to cutaneous melanomas where NRAS mutations occur in 15%–20% of cases (Tate et al., 2019). There is also a divergence in the location of NRAS mutations between MM and cutaneous melanoma, with 54% located on codon 61 in MM versus 88% in cutaneous melanoma, and 46% located on codons 12 and 13 in MM versus 12% for cutaneous melanomas. Approximately 7%–22% of MMs have v-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog (KIT) somatic mutations or amplifications (Dumaz et al., 2019; Tyrrell & Payne, 2018). MMs located in the genital area appear to be driven by mutations in SF3B1 which encodes the subunit 1 of splicing factor 3b, a component of the spliceosome that processes pre-mRNA into mature transcripts (Furney et al., 2013; Oiso, Sakai, Yanagihara, Nishio, & Kawada, 2018; Quek et al., 2019). A recent study analyzing the mutational landscape of MM identified IGF2R mutations in 31.7% of MM samples relative to 6.3% of SSM cases (Iida et al., 2018). Interestingly, a lower frequency of UV-induced DNA damage, a lower number of mutations and a link to high tobacco exposure have also been identified in MM (Iida et al., 2018).

3.3 |. Treatment/Efficacy

Unfortunately, MM is typically detected at relatively more advanced states due to difficulty in early detection (Newell et al., 2019). The main treatment for MM differs slightly on where the tumor is located; however, like any other subtype of melanoma, patients are initially treated with surgical excision. MMs arising in the head and neck are treated with complete surgical excision of the tumor when the patient is in stages III and IVA. However, this is associated with a high rate of recurrence (Patel et al., 2002). MMs that have arisen in the vulvovaginal or anorectal area also receive radiation in addition to surgical tumor excision. Therapeutic efficacy may be improved in select patients when treatment is personalized by tumor mutational status. Clinical trials targeting KIT with imatinib show no clear effect in unselected metastatic melanoma patient populations (Kim et al., 2008; Ugurel et al., 2005; Wyman et al., 2006), but encouraging clinical benefit has been observed with KIT inhibition specifically in patients with melanomas harboring KIT mutations (not in patients whose melanoma harbor KIT amplification only; Table 1). Nonetheless, disease progression ultimately occurs in the majority of cases (Cho, Kim, Kwon, Kim, & Lee, 2012; Hodi et al., 2013). These data support the practice of determining KIT mutational status for MM patients to have a higher chance of receiving additional clinical benefit (Lutzky, Bauer, & Bastian, 2008). Subsequent phase II clinical trials now require a KIT alteration for enrollment. For the relatively small number of MM patients whose tumors harbor BRAF mutations (relative to the ~50% in non-AL cutaneous melanoma patients), treatment with combination BRAF inhibitor and MEK inhibitor therapy is available. However, the efficacy of targeted therapy specifically in the MM patient population is not completely understood due to the low number available for analysis.

The efficacy of immune checkpoint inhibitor therapy also remains unclear in MM patients, with conflicting evidence of whether MM patients respond as well as non-AL cutaneous melanoma patients (Table 1). In one multi-institutional analysis of clinical trials focusing on all the subtypes of metastatic melanoma, patients with MM had similar responses compared with non-AL cutaneous melanoma patients when treated with anti-PD-1 single-agent therapy, with a progression-free survival of 3.9 months (Shoushtari et al., 2016). In another pooled analysis, MM patients treated with nivolumab as monotherapy or nivolumab in combination with ipilimumab experienced reduced clinical benefit relative to non-AL cutaneous melanoma patients (D’Angelo et al., 2017). MM patients experienced 50% shorter progression-free survival (3.0 months) relative to patients with non-AL cutaneous melanoma (6.2 months) for monotherapy (nivolumab) and for nivolumab plus ipilimumab (5.9 vs. 11.7 months; D’Angelo et al., 2017). Another recent study combining axitinib (small molecule receptor tyrosine kinase inhibitor) with toripalimab (anti-PD-1) found a median progression-free survival of 7.5 months in among 29 patients with chemotherapy-naïve mucosal melanoma (Sheng et al., 2019). Although these data suggest that MM patients may not achieve as much benefit with immune checkpoint inhibitor therapy as non-AL cutaneous melanoma patients, it should be considered that in each of the pooled analyses, the number of MM cases was only 10% of patients compared to 75% from cutaneous melanoma (D’Angelo et al., 2017). Also notable, another prospective study where 44 patients with unresectable MM were treated with immune checkpoint inhibitors concluded that the site of origin for MM (i.e., vaginal, anal) may not have a significant impact on the objective response rate, which was 8.2% for ipilimumab and 35% for pembrolizumab (Moya-Plana et al., 2019). The lower mutational burden in MM relative to non-AL cutaneous melanoma may explain the decreased efficacy of immune checkpoint blockade in MM (Newell et al., 2019).

4 |. UVEAL MELANOMA: THE MOST FREQUENT FORM OF NON-CUTANEOUS MELANOMA

4.1 |. Incidence/Presentation

Uveal melanoma (UM) is the most common form of ocular melanoma, as well as the most prevalent form of non-cutaneous melanoma, accounting for 5% of all melanomas (Figure 2) (Chang et al., 1998). It most commonly arises in non-Hispanic Whites relative to other races (i.e., African and Asian Americans), with a slight predominance for men (52.3%) relative to women (47.7%; Aronow, Topham, & Singh, 2018). The incidence of UM has remained stable over the last few decades and is diagnosed in 4–5 per million individuals in the United States each year (Aronow et al., 2018; Singh, Turell, & Topham, 2011). The median age of diagnosis is 62, and the incidence of UM increases with age. Early detection of UM provides a favorable 85% survival rate; however, this survival rate significantly decreases to 15% once UM cells have disseminated. Approximately 50% of UM patients develop metastases, and among patients with metastatic disease, 90% have liver involvement and ~70% have liver-only disease (Bedikian et al., 1995). This is a distinct metastatic pattern relative to cutaneous melanoma or mucosal melanoma.

4.2 |. Mutational/Molecular profile

Unlike non-AL cutaneous melanomas, UMs have a much lower mutational burden due to the sun-protected site they arise from within the ocular cavity (Harbour, 2012). Activating mutations in BRAF or NRAS are not detected (extremely rare) in tumor cells of UM patients (Figure 1d; Rimoldi et al., 2003). In contrast, the main drivers for UM are activating mutations of guanine nucleotide-binding protein G (GNAQ/11), splicing factor 3B subunit 1 (SF3B1), eukaryotic translation initiation factor (EIF1AX), and inactivating mutations of the tumor suppressor BRCA-associated protein-1 (BAP1; Dono et al., 2014). The GNAQ/11 genes encode specific GTP binding proteins that mediate signal transduction from the inner cell surface to the MAPK pathway through activation of the protein kinase C (PKC) enzyme. GNAQ and GNA11 mutations are mutually exclusive, and thus in total are detected in 85%–94% of UM across all stages of disease (Figure 1d; Chen et al., 2014; Field & Harbour, 2014; Goldsmith & Dhanasekaran, 2007). Due to their detection in benign uveal nevi, GNAQ/11 mutations are thought to be early mutational events (Harbour, 2012).

BAP1 (located on the short arm of chromosome 3) loss-of-function mutations are posited to serve as a predisposing factor for diverse hereditary cancers including mesothelioma, cutaneous melanoma, renal cell carcinoma, and UM (Harbour et al., 2013). A recent comprehensive review identified that among 174 patients harboring germline BAP1 mutations, 130 developed tumors that were either UM (31% of cases), cutaneous melanoma (13% of cases), renal cell carcinoma (10% of cases), or MM (22% of cases; Masoomian, Shields, & Shields, 2018). In UM, loss of BAP1 returns melanoma cells to a more stem cell-like state as BAP1 is involved in melanocyte differentiation (Matatall et al., 2013). BAP1 is frequently mutated in metastasizing uveal melanomas, which supports the growing evidence that stem-like melanoma cell states drive elements of the metastatic cascade (Harbour et al., 2010; Rambow et al., 2018).

4.3 |. Treatment/Efficacy

There has been a recent decline in UM patients treated solely with surgery due to micrometastases that develop years before primary tumor detection. The current approach for treatment of metastatic UM is radiation; however, the survival rate is not significantly improved relative to what is possible from surgery (Krantz, Dave, Komatsubara, Marr, & Carvajal, 2017). There have been an array of clinical studies trying to identify efficacious therapeutic strategies for patients with metastatic UM (Table 1). UM patients that possess GNAQ or GNA11 mutations can be treated in clinical trials with targeted therapy approaches specific for the MAPK pathway (i.e., MEK inhibitor, ERK inhibitor) as these tumors display elevated MAPK activity. Preclinical studies have shown that treatment of UM with a combination of a MAPK pathway inhibitor and a PKC inhibitor may provide synergistic efficacy relative to what is achievable by either agent alone (Chen et al., 2014). Clinical trials with selumetinib, a MEK inhibitor, reported a higher progression-free survival among UM patients (15.9 vs. 7 weeks); however, no clinically meaningful increase in overall survival was observed in comparison to the chemotherapeutic temozolomide in the metastatic setting (10.8 vs. 9.4 months). Additionally, preclinical studies identified that targeting the PI3K/AKT pathway (in GNAQ and GNA11 mutant xenograft models) in combination with a MEK inhibitor may be an effective treatment strategy for patients with GNAQ or GNA11 mutations; however, clinical trials using this combination have stopped due to low response rates and high toxicity (Khalili et al., 2012; Table 1). Inhibitors against bromodomain and extraterminal (BET) proteins have had encouraging activity preclinically in UM, which could be further increased by concurrent inhibition of escape mechanisms mediated by fibroblast growth factor receptors (Chua et al., 2019). Similarly, targeting microenvironment-derived factors including HGF can also increase MEK inhibitor efficacy against UM cells, preclinically (Cheng et al., 2017). For UM with BAP1 mutations, it has been shown preclinically that treatment with a histone deacetylase (HDAC) inhibitor could be beneficial. Because BAP1 mutations are associated with loss of melanocytic differentiation, treatment with HDAC inhibitors (valproic acid) are postulated to inhibit the growth of uveal melanoma in vivo by inducing morphological differentiation (Landreville et al., 2012).

While immune checkpoint inhibitors are the standard of care for cutaneous melanoma, UM has not yet had a phase III clinical trial for immune therapy (Rossi et al., 2019). Small studies in UM patients (10 patients) treated with pembrolizumab (anti-PD-1) after treatment with ipilimumab reported a median progression-free survival of 18 weeks; ranging from 3.14 to 49.3 weeks (Kottschade et al., 2016). Of the eight evaluable patients, four rapidly progressed, one had stable disease, two had partial responses, and one had a complete response. Although this small study resulted in comparable results seen in patients with non-AL cutaneous melanoma, other studies suggest far lower response rates to single agent anti-PD-1 and combination anti-PD-1 plus anti-CTLA-4 in UM patients. An analysis of Danish UM patients observed partial responses in 7% of patients to anti-PD-1 and 21% to concurrent anti-PD-1 plus anti-CTLA-4 (Bol et al., 2019). Metastatic UM patients treated with ipilimumab from two additional clinical studies had a median overall survival of 9 months (in contrast to 19.9 months in non-AL cutaneous melanoma; Danielli et al., 2012; Luke et al., 2013; Wolchok et al., 2017). Despite the reduced efficacy of immune checkpoint blockade in UM patients, this option may represent the most effective strategy to date.

4.3.1 |. Non-acral lentiginous cutaneous melanoma

For the remainder of the review, we herein describe relatively rarer forms of cutaneous melanoma relative to SSM (Figure 1a), as the reader can find plentiful preclinical and clinical information on SSM biology in the literature (Atay et al., 2019; Garman et al., 2017; Herlyn et al., 1985; Krepler et al., 2016, 2017; Perego et al., 2018; Roesch et al., 2010; Villanueva et al., 2010, 2013).

5 |. NODULAR: THE QUICKEST GROWING OF THE MELANOMA SUBTYPES

5.1 |. Incidence/Presentation

Nodular melanoma represents the second most common subtype of melanoma, responsible for 10%–15% of total melanomas in Caucasians. NM is the melanoma subtype most associated with increased thickness at clinical presentation, which is attributed to the relatively poorer prognosis of patients with NM (Lattanzi et al., 2019). The median age of diagnosis for NM is 53 years, with thicker tumors more common in older patients. NM is more common in women than men for reasons poorly understood and commonly presents de novo on the head, neck, or trunk of patients (Figure 2; Xiong, Charifa, & Chen, 2019).

5.2 |. Mutational/Molecular profile

Activating BRAF mutations are detected in patients with NM at a slightly lower frequency relative to SSM, with 43%–47% of patients possessing mutations mostly (88% of cases) in V600E (Gorden et al., 2003; Shinozaki, Fujimoto, Morton, & Hoon, 2004; Spathis et al., 2019; Figure 1e). A recent study identified evidence that BRAF^V600E expression may serve as a prognostic marker in primary NM associated with ulceration and reduced survival (Hugdahl, Kalvenes, Puntervoll, Ladstein, & Akslen, 2016). Preclinically, it was reported that hyperactivation of the downstream MAPK effector ribosomal protein S6 kinase (RSK1) is detectable in metastatic tumor tissues derived from NM to a higher extent relative to SSM (Salhi et al., 2015). Activating NRAS mutations are detected at a significantly elevated frequency in NM relative to SSM in 30%–33% vs. 19% of cases, respectively (Chiappetta et al., 2015; Heppt et al., 2017; Lattanzi et al., 2019). Interestingly, BRAF and NRAS mutations may not be as mutually exclusive in NM relative to SSM, with the identification of both mutations in the same tumor specimens when assessed by laser capture dissection followed by direct sequencing analysis of exons 11 and 15 of the BRAF gene and exons 1 and 2 of the NRAS gene (Chiappetta et al., 2015). Additional high-throughput sequencing of patient-derived samples of single nucleotide variations (SNVs) expected to impact protein coding reveals NOTCH4, RPSKA6, BCL2L12, TERT, ERBB3, ZNF560, SSPO, and SNX31 to be significantly under-mutated in NM relative to SSM (Lattanzi et al., 2019).

5.3 |. Treatment/Efficacy

An analysis of the most recent Surveillance, Epidemiology, and End Results (SEER) cohort and the New York University (NRU) cohort suggests that relative to patients with metastatic SSM treated with BRAF inhibitor (BRAFi) therapy, patients with metastatic NM may respond worse to BRAFi for reasons not completely understood, suggesting the potential existence of distinct clinical and biological properties between NM and SSM (Lattanzi et al., 2019). The observation of activated RSK1 via constitutive phosphorylation at the Ser-380 residue may explain the poorer efficacy of BRAFi and/or BRAFi/MEKi in patients with this melanoma subtype. In contrast, no significant difference in response rates and survival was detected in NM versus SSM among a cohort of 154 patients treated with either anti-CTLA-4, anti-PD-1, or the combination of both immune checkpoint inhibitor approaches. Immune checkpoint blockade may serve an ideal first-line therapy for patients with this subtype.

6 |. LENTIGO MALIGNA: “HUTCHINSON’S MELANOCYTIC FRECKLE”

6.1 |. Incidence/Presentation

Lentigo maligna (LM) is the third most common subtype of melanoma, comprising roughly 4%–15% of all melanoma cases (Kallini, Jain, & Khachemoune, 2013; Xiong et al., 2019), and its incidence has dramatically increased over the past few decades across the United States (Mirzoyev et al., 2014; Swetter, Boldrick, Jung, Egbert, & Harvell, 2005), the Netherlands (Greveling, Wakkee, Nijsten, van den Bos, & Hollestein, 2016), and other regions of the world. LM melanoma typically presents on chronically sun-damaged (CSD) skin of the head and neck (Figure 2), appearing as an irregular brown macule commonly on the head and neck in the elderly (DeWane, Kelsey, Oliviero, Rabinovitz, & Grant-Kels, 2019). In contrast to the mean age of diagnosis of SSM between 40 and 60 years, the mean age of diagnosis for LM melanoma is 66–72 years (Xiong et al., 2019). Credit is given to Sir John Hutchinson for the earliest description of LM melanoma in 1,890. LM melanoma was initially referred to as “Hutchinson’s melanocytic freckle” due to the prevailing thought that it was benign, non-infectious lesion owing to its slow growing nature (Xiong et al., 2019). Critical work by Ackerman and Silvers in the late 1970s–1980s finally led to wide acceptance of LM melanoma as a malignant disease worthy of clinical attention and intervention (Cohen, 1995; Silvers, 1976). Chronic ultraviolet radiation is the major risk factor for the development of LM melanoma, which differs from NM and SSM that are associated with intense intermittent ultraviolet radiation exposure (Elwood & Hislop, 1982; Holman & Armstrong, 1984). LM melanomas arise most frequently on the face and other sites of chronic sun damage which also differs from NM and SSM that arise most commonly on the trunk in men and legs in women (Figure 2). LM melanoma is thought to occur in older patients due to the increased lifetime sun and ultraviolet radiation exposure.

6.2 |. Mutational/Molecular profile

Lentigo maligna melanomas have a relatively high mutational burden compared to other melanoma subtypes due to chronic ultraviolet exposure (Xiong et al., 2019). The frequency of activating BRAF mutations in LM is unclear, with reports finding 16.7%–53.4% of LM patients harboring BRAF mutations (Saldanha, Potter, Daforno, & Pringle, 2006; Spathis et al., 2019; Figure 1f). The large variation may, in part, be attributed to the regional differences among tested patient tissue cohorts. In a Greek cohort, 16.7% of LM melanoma cases expressed BRAF mutations (Spathis et al., 2019) and 50% of LM cases in a Japanese cohort expressed BRAF mutations (Yamazaki et al., 2015). When BRAF mutations are present, the V600K substitution is frequently observed (~77%) relative to the V600E (~23%) as observed in SSM, in this small set of 13 LM patient tumor samples. This finding is consistent with V600K mutations arising on chronically sun-damaged skin (Connolly, Nehal, & Busam, 2015; Stadelmeyer et al., 2014). Activating NRAS mutations have been reported to occur in ~8.1%–16% of LM cases (Akslen et al., 2008; Saldanha et al., 2006).

6.3 |. Treatment/Efficacy

The treatment of choice for patients with localized LM melanoma consists of surgical excision as first line of therapy, followed by radiation therapy with fractionated superficial radiotherapy, or topical imiquimod cream as an alternative to surgery (Marsden et al., 2017; Mora, Karia, & Nguyen, 2015; Xiong et al., 2019). Once LM melanoma metastasizes to visceral organs, the five-year survival is similar to SSM (Kantor & Kantor, 2009). Interestingly, the efficacy of immune checkpoint blockade may be significantly higher in patients with LM melanoma relative to the other subtypes discussed. A study investigating the overall response rate (ORR) of anti-PD-1/PD-L1 in different subtypes of melanoma found patients with melanoma on CSD skin (including LM melanoma, desmoplastic melanoma, and subtype not-specified cases) exhibited an overall response rate of 70% (Kaunitz et al., 2017), which fits the theory that cancer cells with high mutational burdens may be more sensitive to immune checkpoint blockade due to the increased presence of immune-stimulatory neoepitopes (Morrison et al., 2018; Panda et al., 2017). Additional investigations on the efficacy of targeted and immune-based therapy are needed specifically for patients with LM melanoma to ensure the optimal treatment(s) is identified for this cohort and further improved through preclinical experimentation and clinical trials.

7 |. DISCUSSION

The last decade has ushered in an era of immense therapeutic possibility for patients with metastatic melanoma. What was once historically an intractable disease treated with ineffective surgery, radiation, and chemotherapy in the metastatic setting has now become a paradigm for the use of targeted and immune-based therapy. Unfortunately, the increased overall survival benefits of recently approved therapies for melanoma patients is experienced mostly by patients with the most prevalent subtype of non-AL cutaneous melanoma (SSM, NM, LMM). SSM, NM, and LMM are the most common subtypes among Caucasians in the United States and Europeans, however not among individuals of Latin American, Asian, or African descent. In these populations, other subtypes of melanoma are more predominant, and unfortunately, it remains unclear what therapeutic strategies may best cure these at-risk patient populations. Although cutaneous melanomas can be stratified into the SSM, NM, LMM, and ALM subgroups based on the work of Drs. Wallace Clark, Martin Mihm, Vincent McGovern, and Richard Reed (Scolyer, Long, & Thompson, 2011), there was debate on whether their exists justification for the use of this classification system, as early reports suggest no differences in mortality and metastases among the different subtypes. However, large-scale retrospective analyses that have been recently performed demonstrate potential disparities in the efficacy of BRAFi/MEKi and immune checkpoint inhibitor therapy among AL melanoma, as well as non-cutaneous melanoma subtypes (UM, MM) relative to non-AL cutaneous melanomas. It is critical the melanoma community address these therapeutic shortcomings in the near future and engineer adequate experimental models for preclinical investigation of resistance mechanisms to existing standard-of-care therapies to facilitate the development of optimized therapeutic strategies that increase the curative rates for patients with rarer forms of malignant melanoma. Small collections of models of rarer forms of melanoma (i.e., cell lines, patient-derived xenograft) do exist in laboratories across the world, and there has been renewed interest across the melanoma field to focus on these rarer subtypes. Only by providing the level of experimental fervor, clinical attention, and collaboration to ALM, MM, and UM that currently exists for SSM can we best heal patients with these subtypes.