Journal of Clinical Oncology, Vol 25, No 12 (April 20), 2007: pp. 1606-1620
© 2007 American Society of Clinical Oncology.
DOI: 10.1200/JCO.2006.06.0442
Toward a Molecular Classification of Melanoma
Leslie A. Fecher,
Staci D. Cummings,
Megan J. Keefe,
Rhoda M. Alani
From the Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins; and the Departments of Oncology and Pharmacology and Molecular Sciences, Program in Cellular and Molecular Medicine, Johns Hopkins University School of Medicine, Baltimore, MD
Address reprint requests to Rhoda M. Alani, MD, Johns Hopkins University School of Medicine, 1650 Orleans St, CRB 342, Baltimore, MD 21231-1000; e-mail: ralani{at}jhmi.edu
 |
ABSTRACT
|
|---|
The incidence of melanoma is increasing at one of the highest rates of any form of cancer in the United States, with the current lifetime risk being one in 68. At present, there are limited systemic therapies to treat advanced stages of melanoma, and the key to improved survival remains early detection. Recent discoveries have allowed for a clearer picture of the molecular events leading to melanoma development and progression. Since identifying prevalent activating mutations of the BRAF kinase in melanomas, there has been a flood of additional molecular studies to further clarify the role of this pathway and others in melanomagenesis. In particular, recent genetic studies have demonstrated specific genotype-phenotype correlations that provide the first major insights into the molecular subclassification of melanoma and the heterogeneous nature of this malignancy. In this article, we review the most up-to-date molecular discoveries in melanoma biology and provide a framework for understanding their significance in melanoma development and progression. We also provide details on the development of novel therapies based on these recent molecular discoveries and insight into current and planned clinical trials. It is expected that these latest studies in melanoma will help define the critical molecular events involved in disease onset and progression and allow us to move rapidly toward a true molecular classification. We eagerly anticipate rationally designed melanoma therapies based on such a classification scheme and the associated improvements in patient outcomes.
|
INTRODUCTION
|
|---|
Traditional classification systems for primary melanoma have centered on site of origin, tumor thickness, and histologic subtype. Primary sites for melanoma include cutaneous, mucosal, uveal/ocular, and leptomeningeal, with the majority (90%) being of cutaneous origin. Cutaneous melanomas have been classified into four major histogenetic subtypes: superficial spreading, lentigo maligna, acral lentiginous, and nodular.1,2 Although these particular melanoma subtypes are clinically and histopathologically distinct, such a classification is without independent prognostic value.3 Despite a wealth of data, the most useful prognostic indicators of primary cutaneous melanoma to date remain Breslow depth and presence or absence of ulceration.4 Indeed, the translation of molecular studies in melanoma biology to useful clinical correlates or novel therapies during the past two decades has been overwhelmingly disappointing. Although molecular markers of potential clinical utility look promising in small-scale studies, the vast majority fail to prove clinically useful in larger-scale studies. Such disappointing results appear to stem from the heterogeneous nature of melanomas and the lack of a classification scheme that would allow for tailored use of molecular markers and/or therapies. Recent studies have shed new light on the molecular events associated with clinicopathologic subclasses of melanomas. This has led to an evolving paradigm for classifying melanomas according to the degree of sun exposure and associated molecular defects.5,6 We expect these studies and subsequent analogous studies will inform the melanoma research community in novel ways, which will lead to rapid advances in melanoma biology with improved translational outcomes.
Here we review several key molecular pathways implicated in melanoma pathogenesis and maintenance and provide a framework for defining the significance of these pathways in melanoma. We underscore recent studies that have provided insights into the diverse nature of melanoma and have allowed for the development of a classification scheme, which is providing insight into epidemiologic data in melanoma and their molecular counterparts. Finally, we highlight the current status of clinical studies aimed at targeting molecular pathways involved in melanoma development and progression. With the present rapid pace of discoveries in melanoma biology and genetics, and the developing molecular classification of this malignancy, we are poised for major advances in our fundamental understanding of this disease and eagerly anticipate the translation of these discoveries to improved patient management and novel therapies.
|
MITOGEN-ACTIVATED PROTEIN KINASE PATHWAY
|
|---|
The mitogen-activated protein kinase (MAPK) signal transduction pathway is the subject of intense study in oncology. It has been of particular interest in melanoma biology since the discovery of frequent activating mutations of BRAF kinase in melanomas and nevi.7,8 As this pathway normally regulates cell growth, survival, and invasion, it is not surprising that it has been implicated in the development of a broad spectrum of cancers. MAPK signaling is initiated at the cell membrane, either by receptor tyrosine kinases (RTKs) binding ligand or integrin adhesion to extracellular matrix, which transmits activation signals via the RAS GTPase on the cell membrane inner surface (extensive reviews by Giehl9 and Campbell and Der10). Active, GTP-bound RAS can bind effector proteins, leading to cell proliferation, differentiation, and survival through activation of various signaling pathways.9 The best characterized RAS effector proteins are RAF and phosphatidylinositol 3-kinase (PI3K). The RAF family of serine/threonine kinases includes three proteins—A-RAF, B-RAF, and C-RAF—translated from unique genes.11 RAF is the primary link between RAS and the MAPK pathway. Its activation is necessary and sufficient to activate the cascade.12 Activated RAF, in turn, phosphorylates and activates MAPK/ERK kinase (MEK 1/2). The cascade culminates in the activation of extracellular signal-regulated kinase (ERK 1/2), the only known substrates of MEK.13 ERKs relay proliferative or survival signals through phosphorylation of a variety of cytoplasmic targets, such as prosurvival ribosomal S6 kinase (p90rsk) or proapoptotic bcl-2 interacting mediator of cell death (BIM)13,14; cytoskeletal targets, such as microtubule-associated proteins 2 and 4 (MAP 2/4)13; and nuclear transcription factors, such as c-MYC, c-FOS, and hypoxia-inducible factor-1 alpha.13,15 A detailed illustration of the MAP kinase pathway is provided in Figure 1.
View larger version (60K):
[in this window]
[in a new window]
[PowerPoint Slide for Teaching]
|
Fig 1. The mitogen-activated protein kinase (MAPK) signaling cascade. Growth factor binding to receptor tyrosine kinases (RTKs) or integrin adhesion to extracellular matrix (ECM) activates RAS, initiating the RAF/MEK/ERK phosphorylation cascade. EGFR, epidermal growth factor receptor; VEGFR, vascular endothelial growth factor; PDGFR, platelet-derived growth factor receptor; FGFR, fibroblast growth factor receptor; FLT-3, fms-related tyrosine kinase 3; SHC, src homologous and collagen; GRB2, growth factor receptor-bound protein 2; SOS, son of sevenless; FAK, focal adhesion kinase; STAT, signal transducers and activators of transcription; HIF-1 , hypoxia-inducible factor-1 alpha; MITF, microphthalmia-associated transcription factor; BIM, bcl-2 interacting mediator of cell death.
|
|
|
RAS AND MELANOMA
|
|---|
Activating mutations of RAS occur frequently in human cancers.9 Somatic RAS mutations typically occur at codons 12, 13, or 61 and maintain RAS in a constitutively active state.16 Although K-RAS mutants are the most frequent in human malignancies, mutations of RAS isoforms other than N-RAS are rare in melanomas.17,18 Although MAPK activation is characteristic of melanomas, RAS mutations are present in only 15% of cutaneous melanomas.17,19,20 The putative transforming ability of mutant RAS has been demonstrated exhaustively in a wide variety of cell types of both murine and human origin.10 Although current mouse models of melanoma are of limited use because of the dermal location of murine melanocytes within hair follicles, such studies have been helpful in sorting out the role of RAS in melanoma development. Regrettably, these studies were performed using transgenic expression of the less relevant H-RAS isoform, as virtually all RAS mutations in human melanoma are of N-RAS. As an isolated event, RAS mutation is not sufficient to initiate melanomas in mice21,22; however, constitutive H-RAS expression targeted to melanocytes in p16ink4a/p19arf-null mice leads to cutaneous melanomas with a 60% penetrance.21 Furthermore, H-RAS is required to maintain these melanomas, as loss of activated H-RAS in established tumors leads to tumor regression.23 Interestingly, established tumors were never able to metastasize, suggesting additional genetic/epigenetic events were needed to establish this characteristic melanoma phenotype. RAS mutations in human melanoma do not correlate with the degree of sun exposure, histologic subtype, or body site. In the majority of cases, N-RAS and BRAF mutations are not present together in a single melanoma, suggesting functional redundancy.6,7 Rare coexistent BRAF (V600E) and N-RAS (Q61R) mutants have been described, but the functional significance of this remains unclear.24,25
|
BRAF AND MELANOMA
|
|---|
Although RAF has long been identified as a proto-oncogene,26 BRAF mutations in human cancers have recently garnered much attention.7 In a study aimed at identifying new cancer therapeutic targets, investigators sequenced molecular components of the MAPK pathway in a variety of cancers. Remarkably, they identified BRAF somatic missense mutations in 66% of malignant melanomas analyzed and observed all mutations to be within the kinase domain, with a single substitution (V600E) accounting for 80%.7,27,28 This V600E mutant possesses 10.7-fold kinase activity versus wild-type BRAF.7 Subsequent studies have shown the presence of activating BRAF mutations in up to 82% of benign nevi, suggesting activation of the MAPK pathway is a necessary event for melanoma development, but it is not sufficient for malignant transformation.8 It is noteworthy that neither the N-RAS nor BRAF mutations observed in melanoma demonstrate characteristic UV radiation-induced changes, and the target of UV injury leading to such mutations remains unclear.7,20 In addition to melanoma, BRAF mutations have been described in a number of other malignancies, most notably in papillary thyroid carcinoma, where rates have been reported to be as high as 69%29-31 and were demonstrated to portend a worse prognosis.32
A-RAF, B-RAF, and C-RAF (also known as raf-1) make up the RAF family of serine/threonine kinases. All RAF family members are capable of activating the MAPK pathway, although each isoform possesses a distinct expression profile with unique phosphorylation targets and signaling effects (reviewed in Beerem et al11). Mutant BRAF is able to transform NIH3T3 cells much more efficiently than wild-type–BRAF, although less so than H-RAS, suggesting the importance of additional RAS signaling events in tumor development.7 Additionally, melan-a murine melanocytes can be transformed with V600E BRAF, and they are oncogenic in nude mice.33 In human melanoma cells, V600E BRAF knockdown by small interfering RNA inhibits MAPK activation and leads to growth arrest, apoptosis, and reversal of the malignant phenotype.34 Thus, the above studies support an oncogenic role for mutant BRAF and activation of the MAPK cascade in melanoma development, and they suggest a potential therapeutic benefit of BRAF inhibition in melanoma. Since the majority of benign nevi also possess activating mutations of BRAF, it is clear that isolated BRAF mutation is not sufficient to initiate human melanoma in vivo.8 Histopathologic assessments suggest that the majority of melanomas evolve de novo, without a precursor melanocytic lesion.35 Hence, the significance of BRAF mutations in nevi and melanomas remains unclear, and it has been suggested that BRAF mutations in nevi may serve as markers of melanoma susceptibility in an individual. Interestingly, in melanomas with an underlying nevus present, the vast majority of these lesions either both possess a BRAF mutation or are both negative for the mutation,8,36 supporting a possible evolutionary event.
A recent landmark study of genome-wide alterations in DNA copy number and BRAF and N-RAS mutational status in primary human melanomas has shed light on the role of BRAF kinase in melanoma development and melanoma heterogeneity.6 Although previous assessments of BRAF status according to histologic subtype showed increased BRAF mutations in nodular and superficial spreading, compared with acral lentiginous and lentigo maligna,37-39 a mechanistic explanation for these findings remained uncertain. Additionally, mucosal melanomas were found to have infrequent BRAF mutations,40 suggesting UV-associated melanomas might evolve from a divergent set of genetic events from UV-protected sites. Previous studies have suggested that rather than a direct relationship between UV radiation and BRAF mutation, melanomas with the highest degree of BRAF mutations were those with intermittent sun exposure.37 The recent work by Curtin et al confirms and expands on this relationship to reveal a more profound correlation of UV exposure to BRAF/N-RAS mutations and additional genetic events in melanoma.6 In this elegant study, the investigators use array-based comparative genomic hybridization (CGH), DNA sequencing, and immunohistochemical (IHC) analyses to determine genome-wide changes in DNA copy number and BRAF/N-RAS mutational status in primary melanomas. Melanomas were classified into four groups based on site and UV exposure: mucosal, acral, and cutaneous melanomas with and without chronic sun-damaged (CSD) skin (defined by presence of solar elastosis). The majority of cutaneous melanomas on non-CSD skin (intermittent sun-exposed) possessed mutations in BRAF or N-RAS (59% and 22%, respectively). Melanomas without either mutation often had increased copies of CDK4 or CCND1 (cyclin D1 gene). Furthermore, no melanoma with CDK4 amplification manifested concomitant N-RAS or BRAF mutations, or CCND1 amplification, suggesting overlapping functions of the MAPK pathway and the CCND1/CDK4 pathways with independent oncogenic functions of each melanoma. The overall incidence of BRAF or N-RAS mutations was significantly lower in tumors located on CSD skin and nonexposed sites, with BRAF and N-RAS mutations being mutually exclusive. No influence of tumor thickness was noted on frequency of mutation (BRAF or N-RAS) or amplification (CCND1 or CDK4). Deletion of CDKN2A was prominent in mucosal and acral melanomas, which also had the greatest incidence of CDK4 amplifications. However, no CDK4 amplifications were present in samples with homozygous CDKN2A deletions. Of all histopathologic subtypes, acral and mucosal melanomas demonstrated the greatest number of genomic events. In CSD melanomas, BRAF mutations were rare and CCND1 copy gain predominated. Conversely in non-CSD melanomas, mutant BRAF and chromosome 10 (site of PTEN) loss were both common. Remarkably, using this genetic classification, these investigators were able to correctly distinguish melanomas with up to 89% accuracy.6
In a follow-up study, Landi et al41 identified an important link between germline mutations of the melanocortin-1 receptor (MC1R) and BRAF mutations. Although previous studies have identified MC1R variants as risk factors for melanoma,42 the precise link to sun exposure and genetic events in primary melanomas was unclear. In this recent study of nearly 200 cases of primary cutaneous melanomas, MC1R variant alleles were found to be associated with melanoma risk, specifically in patients with melanomas in non-CSD skin. This risk was associated with tumors harboring BRAF mutations suggesting that germline events can largely influence genetic events leading to tumorigenesis in response to environmental exposures (eg, UV radiation). This particular association would likely have been difficult to discern had the investigators not specifically subclassified tumors on the basis of sun exposure and associated genetic events. Thus, the heterogeneous nature of primary melanomas and their genetic and environmental basis are rapidly being clarified due to the simple subclassification described by Curtin et al.6
|
TARGETING RAS AND THE MAPK CASCADE
|
|---|
Inhibition of the MAPK signaling cascade at all levels is being keenly investigated in oncology. Multiple agents are in various stages of development (Table 1). RTKs and their stimulatory ligands are attractive targets for many cancers and have been the subject of numerous investigations.43 RAS inhibition is an attractive therapeutic strategy in cancer as it should impact multiple pathways. Since BRAF mutations, which are downstream of RAS, predominate in melanoma, the potential utility of such therapies may be limited.44 RAS-targeted therapies under investigation include farnesyltransferase inhibitors, which block association of RAS with the plasma membrane and its activation (reviewed by Adjei45). Current trials in melanoma with farnesyltransferase inhibitors as well as with agents targeting RTKs, selective ligands, and integrins are summarized in Table 1.
Sorafenib (BAY 43-9006, Nexavar; Bayer, West Haven, CT, and Onyx, Richmond, CA) is a RAF tyrosine kinase inhibitor that has been shown to inhibit the MAPK pathway in vitro and in vivo.46 Some of sorafenib's antitumor effects also are attributed to antiangiogenic activities as it targets vascular endothelial growth factor and platelet-derived growth factor receptor.46,47 Single-agent sorafenib has not shown significant activity in metastatic melanoma.48 However, in combination with chemotherapy, it has shown an improvement in objective response and progression-free survival, which was not dependent on BRAF mutational status.44,49 Sorafenib was recently approved by the US Food and Drug Administration for the treatment of advanced renal cell carcinoma based on improved progression-free survival.50,51 This activity in renal cell carcinoma, a tumor lacking BRAF mutation, supports the notion that sorafenib's antitumor activity may include, but is not limited to, inhibition of BRAF. The current focus of clinical investigation of sorafenib in melanoma is in novel combinations, both with chemotherapy and other targeted agents (Table 1). As sorafenib does not target mutant BRAF alone, other kinase inhibitors specific to mutant BRAF are in development. RAF-265 (Chir 265; Novartis, Basel, Switzerland) is currently recruiting to a phase I study in stage III/IV melanoma. PLX-4032 (Plexikkon, Berkeley, CA), a potent and specific inhibitor of mutant BRAF, also is accruing to a phase I study in advanced solid tumors.52 Additionally, two antisense oligonucleotides against CRAF are under clinical investigation: Liposome Entrapped CRAF Antisense Oligonucleotide (LErafAON; Neopharm, Lake Forest, IL) and ISIS 5132 (ISIS, Carlsbad, CA; Table 1). There are no antisense oligonucleotides against BRAF in clinical trials.
MEK is a key target for inhibition because ERK 1/2 are its only known substrates, inhibition of ERK phosphorylation should impact several oncogenic signaling pathways, and all current MEK inhibitors are specific as they bind and inhibit in a nonadenosine triphosphate (non-ATP) competitive manner.13,53 17-allylamino-17-demethoxygeldanamycin (17-AAG), an analog of Geldanamycin, indirectly targets RAF via inhibition of the ATPase activity of heat shock protein-90.54 This disrupts heat shock protein-90 complexes, and it results in the degradation of chaperoned proteins, including RAF, CDK4, Akt2, and mutated p53, among others.54 In vitro cell lines with mutant BRAF have demonstrated increased sensitivity to MEK inhibitors55 and 17-AAG.56 Two MEK inhibitors are currently in clinical trials: PD0325901 (Pfizer, New York, NY) and ARRY-142886 (AZD6244; Array, Boulder, CO, and AstraZeneca, Wilmington, DE). In addition, there are several ongoing clinical trials with 17-AAG as well as another analog, 17-dimethylaminoethylamino-17-demethoxy-geldanamycin (Table 1).
|
RETINBLASTOMA PATHWAY AND MELANOMA
|
|---|
Circumventing cell cycle control is a key feature of carcinogenesis. In a normal cell, proliferation is orchestrated by cyclin-dependent kinases (CDKs), which are positively regulated by CDK activating kinases and negatively regulated by cyclin-dependent kinase inhibitors, such as p16INK4a and p21. Each CDK paired with its cyclin facilitates a particular stage of the cell cycle. The G1/S-phase transition is mediated by cyclin E/CDK2 and cyclin D/CDK4. Cycle progression occurs with the phosphorylation and inactivation of the retinoblastoma protein (Rb). Thus, Rb is considered the gatekeeper for this transition and for the commitment of the cell to progress through the S-phase, where DNA synthesis takes place57 (Fig 2). Inactivation of Rb frees it from its association with E2F transcription family members, allowing E2F to activate transcription of S-phase genes.58 Additional targets of these particular CDKs, as well as several positive feedback loops, perpetuate this process once begun. The Rb pathway is a critical safeguard in cell cycle progression and is disrupted in virtually all human cancers.59
View larger version (30K):
[in this window]
[in a new window]
[PowerPoint Slide for Teaching]
|
Fig 2. Cell cycle regulation by cyclin-dependent kinases (CDKs): cyclins promote progression and cyclin-dependent kinase inhibitors mediate arrest. Retinoblastoma protein (Rb) controls the G1/S restriction point. CDK2/4/6 phosphorylate Rb, dissociating E2F and allowing cycle progression. p53 impacts cell cycle via differential gene expression depending on cellular stresses. 9p21 alternative splicing produces p14ARF or p16INK4a. p14ARF inhibits MDM2 and prevents p53 proteasomal degradation. p16INK4a inhibits cyclin D-CDK4/6, maintaining Rb-E2F and causing Gap 1 (G1); arrest. ARF, alternative reading frame; G2, Gap 2; S, synthesis; M, mitosis.
|
|
An important mediator of the Rb tumor suppressor pathway is the 16 kd inhibitor of cyclin-dependent kinase 4a (p16INK4a), also known as major tumor suppressor 1 and cyclin-dependent kinase inhibitor 2a (CDKN2A). p16INK4a serves as a braking force for the cell cycle by binding to and inhibiting the cyclin D/CDK4 complex, thereby preventing it from phosphorylating Rb. The human CDKN2A locus, which encodes p16INK4a, is on chromosome 9p21 and is deleted or mutated in a variety of human cancers.60-62 In addition to p16INK4a, a second tumor suppressor, p14/p19ARF is transcribed from this same 9p21 locus through alternative splicing (alternative reading frame [ARF], p19 being the mouse homolog of human p14). Animal models were critically important in confirming the dual tumor suppressor functions of the CDKN2A locus and will be described subsequently.
Mutation of p16INK4a is the most commonly known cause of inherited susceptibility in familial melanoma, but it is relatively rare in sporadic disease.63,64 The rate of p16INK4a mutation in familial disease is estimated to be between 10% and 40%.65-68 In familial atypical multiple mole melanoma syndrome, an autosomal dominant inherited syndrome characterized by multiple atypical nevi and a predisposition to melanoma, as well as pancreatic cancer,69 the reported germline p16INK4a mutation rate is 40%.70 As the vast majority of melanomas are of the sporadic type,71 it is likely that nonmutational silencing of p16INK4a or cyclinD/CDK4 activation are the predominant mechanisms for inactivation of the Rb pathway in that setting. Although the precise incidence of p16INK4a gene silencing in melanomas has been debated, a recent study found that 75% of melanomas analyzed had evidence for silencing of the p16INK4a promoter by hypermethylation.72 Of note, lower absolute p16INK4a expression correlates with progression of disease,73,74 increased cell proliferation, and poor prognosis in sporadic melanomas.75
Overexpression of CDK676 and amplification of CDK46 have been observed in melanoma. CDK4 amplification is more common in acral and mucosal melanomas than other subtypes,6 and human melanomas with homozygous p16INK4a loss do not demonstrate amplification of CDK4.6 Amplification of CCND1 also is observed in melanoma and was discussed previously. Activating CDK4 mutations also have been found in melanoma-prone families.77,78 The R24C point mutation in CDK4 abrogates binding to p16, but not to cyclin D, thereby promoting a constitutively active cyclin/CDK complex.
|
p53 PATHWAY AND MELANOMA
|
|---|
Like the Rb pathway, the p53 pathway is frequently inactivated in human cancers. The p53 pathway responds to cellular stresses that threaten the fidelity of DNA replication, genome stability, chromosome separation, and cell division.79 p53 functions primarily as a transcription factor, where cellular stresses serve to activate and target p53 to genes that initiate accelerated DNA repair, inhibit cell cycle progression, or lead to senescence or apoptosis.80 Many of these cellular stresses can be oncogenic in nature, explaining the high frequency of p53 mutations in cancer. In the skin, the most important of these DNA-damaging stresses in carcinogenesis is UV radiation. Since p53 plays a vital role in preserving the genetic integrity of a cell, it is not surprising that regulation of its activity is complex,81 integrated with several positive and negative feedback loops that either carry out or reverse the stress response. One such negative regulator, MDM2, binds tightly to p53, blocking its transcriptional activity and targeting it for ubiquitin-mediated proteasome degradation (Fig 2). Additionally, p53 positively regulates MDM2 transcription, creating a negative autoregulatory feedback loop.82 Another key player in this pathway is p14/p19ARF, a gene transcribed from the same 9p21 locus as p16INK4a. p14/p19ARF binds MDM2, decreasing its ubiquitin ligase activity, and thereby stabilizing p53.83,84 This locus is mutated in familial melanoma and has been extensively studied in this disease as it encodes two tumor suppressor genes that function within separate pathways.85 Melanoma-associated CDKN2A mutations have been documented within p16INK4a alone, p14/ARF alone, as well as in both genes confirming the independent tumor suppressor functions of each CDKN2A transcript and critical importance of the Rb and p53 pathways in melanoma.86-92
Despite the fact that p53 is one of the most commonly mutated genes in cancer, p53 itself is rarely mutated in human melanoma (0% to 25%)93-98 or in melanoma cell lines.99 Similarly, isolated p14ARF mutations in melanoma are rare.100 However, alterations of this pathway in other fashions have been linked to melanoma. MDM2 levels have been found to increase with melanoma progression,101,102 and MDM2 expression in melanomas has been demonstrated to portend a better prognosis, likely because of its indication of an intact p53 pathway.103
Mouse models for melanoma have added to our general understanding of tumor biology and provided insights into our understanding of human melanoma. Mice null for both p16ink4a and p19arf develop spontaneous tumors and display an increased sensitivity to carcinogens as do mice with either gene deleted alone, confirming that both are bona fide tumor suppressor genes. However, the majority of tumors in these animals were lymphomas and sarcomas with only rare melanomas.104-108 In a mouse model with melanocyte-targeted transgenic expression of H-RAS in a p16ink4a and p19arf-null background, mice frequently developed melanomas by age 6 months.21,108 Mice null for p19arf alone with targeted transgenic expression of H-RAS in melanocytes also developed melanomas. Tumors developed in this setting were enhanced by exposure to UV radiation and had sustained subsequent loss of p16ink4a and amplification of CDK6.109
Epidemiologically, UV exposure has been highly associated with melanoma development, but specific cellular targets besides DNA damage have only been vaguely identified and are often controversial. Patients with xeroderma pigmentosa who have defective DNA repair demonstrate high frequencies of UV-induced melanoma and an unusually high p53 mutation rate of 60%.110 Correlative studies suggest that p53 plays a role in UV-induced apoptosis in melanoma,111 and p53 stabilizers have been found to significantly increase UVB-induced apoptosis.112 Interestingly, cells null for p16ink4a or p19arf or both all showed a decreased ability to repair UV-induced DNA damage supporting a link of this pathway to UV-associated genetic mutations.113 Although UV exposure is highly linked to melanoma development, the relationship is certainly not straightforward and remains to be elucidated.
|
TARGETING THE CELL CYCLE
|
|---|
Therapeutically targeting the cell cycle involves CDK inhibitors that attempt to re-establish checkpoint integrity, cause cell cycle arrest, and induce apoptosis114 (reviewed by Schwartz and Shah115). Two agents targeting the cell cycle, flavopiridol and bryostatin-1, have shown only minimal activity as single agents in metastatic melanoma,116-118 according to traditional response criteria, and are in clinical trials for various malignancies. Agent 7-hydroxystaurosporine (UCN-01), which dephosphorylates and inactivates CDK2, leads to activation of Rb and maintains the G1/S restriction point.115 UCN-01 has broad effects as it also inhibits protein kinase C (PKC), PDK-1, and checkpoint kinase-1115 (Fig 2). Due to a partial response lasting 6 months in a melanoma patient in the phase I study,119 a phase II in stage III/IV melanoma is being pursued. Additionally, there are several ongoing phase I studies with small molecule CDK inhibitors and a unique agent (ZK 304709) that inhibits select CDKs as well as vascular endothelial growth factor and platelet-derived growth factor receptor (Table 1).120,121
Myriad cellular processes are affected by the ubiquitin proteasome pathway due to its role in regulatory protein turnover.122 Bortezomib (Velcade, Millenium Pharmaceuticals, Cambridge, MA) inhibits protein degradation through direct, reversible binding of the proteasome.122 This downregulates the transcription factor, nuclear factor- B by preventing degradation of its inhibitor, IB (Fig 4). Proteasome inhibition is thought to have several cellular effects, including: enhanced susceptibility to chemotherapy/radiotherapy-induced apoptosis, cell cycle arrest, and decreased expression of cell adhesion molecules impairing angiogenesis and metastasis.122 Single-agent bortezomib lacked significant activity in metastatic melanoma,123 but it is being evaluated in combination with various targeted and cytotoxic agents in melanoma and other cancers (Table 1).
View larger version (49K):
[in this window]
[in a new window]
[PowerPoint Slide for Teaching]
|
Fig 4. Summary of key molecular pathways in melanoma and current targeted agents in clinical trials. The agents, designated targets, and current investigational status are presented in Table 1. VEGF-TRAP, vascular endothelial growth factor-trap; RTK, receptor tyrosine kinase; SCF, stem cell factor (also known as KIT ligand); ECM, extracellular matrix; TKI, tyrosine kinase inhibitors; UCN-01, 7-hydrosystaurosporine; PDK1, phosphionositide-dependent kinase 1; AKT, also known as protein kinase B; mTOR, mammalian target of rapamycin; PKC, protein kinase C; IKK, I-B kinase; I-B, inhibitor of nuclear factor-B; NF-B, nuclear factor-B; 17-AAG, 17-allylamino-17-demethoxygeldanamycin; 17-DMAG, 17-dimethylaminoethylamino-17-demethoxy-geldanamycin; FAK, focal adhesion kinase; HSP90, heat shock protein-90; CDK, cyclin-dependent kinase.
|
|
|
PI3K PATHWAY AND MELANOMA
|
|---|
The phosphatidylinositol-3-kinase (PI3K) pathway is one of the most commonly altered pathways in human tumors.124 Class I PI3Ks are heterodimers consisting of a regulatory p85 subunit and a catalytic p100 subunit.125 PI3K activation by RTKs or G-protein coupled receptors leads to phosphorylation of phosphatidylinositol-4,5-biphosphate (PIP2) to phosphatidylinositol-3,4,5-triphosphate (PIP3)126 (Fig 3). PIP3 recruits other proteins to the plasma membrane,124 eventually resulting in activation of Akt (protein kinase B), the major downstream effector of the PI3K pathway.126 Phosphatase and tensin homolog deleted on chromosome 10 (PTEN) is a lipid and protein phosphatase that negatively regulates the PI3K pathway through dephosphorylation of PIP3.125 The PI3K pathway is activated by RAS and interacts with many other pathways in the cell, leading to diverse functions in cellular proliferation, apoptosis, cytoskeletal rearrangement, and tumor cell chemoresistance.124,125
View larger version (60K):
[in this window]
[in a new window]
[PowerPoint Slide for Teaching]
|
Fig 3. The PI3K/AKT pathway promotes cell survival through activation or inhibition of various targets. c-KIT, an RTK, promotes ERK activation and is mutated in select melanomas. Microphthalmia-associated transcription factor (MITF) transcription and translation is influenced by a variety of stimuli, including c-KIT, beta-catenin, and MC1R. MSH, melanocyte-stimulating hormone; FOXO, forkhead; PIP, phosphatidylinositol phosphate; AKT, also known as protein kinase B; BAD, bcl-associated death promoter; mTOR, mammalian target of rapamycin; CREB, cAMP responsive element-binding; PTEN, phosphatase and tensin homolog deleted on chromosome 10; IKK, I-B kinase; PDK1, phosphoinositide-dependent kinase 1; FZD, frizzled; GSK3β, glycogen synthase kinase-3 beta; SCF, stem cell factor (also known as KIT ligand); MAPK, mitogen-activated protein kinase.
|
|
Mutations of PIK3CA (p110 alpha subunit of PI3K) are rarely detected in melanoma, and are found in less than 1% of primary melanomas6 and 3% of melanoma metastases,127 with no evidence of amplification of any PI3K subunit in primary melanomas by array CGH.6 However, PI3K inhibitors Wortmannin and LY294002 have antitumor activity in vitro, inhibiting proliferation and sensitizing cell lines to chemotherapy and radiation treatment125 suggesting pathway activation independent of genetic alterations. Akt is closely related to PKA and PKC, consisting of a kinase domain, a pleckstrin homology domain, and a regulatory tail.125,126 Once active, Akt phosphorylates a number of substrates, including MDM2, nuclear factor-B, mammalian target of rapamycin (mTOR), bcl-associated death promoter, human telomerase reverse transcriptase (hTERT), and p27, which promote cell survival, proliferation, and invasion124-126 (Fig 3). Akt3 is the major isoform deregulated in melanoma.128 Although no activating mutations of Akt3 have been reported in primary melanomas, overexpression has been seen.128 Targeted decrease of Akt3, either using small interfering RNA or via increased activation of PTEN, stimulated apoptosis of melanoma cell lines, indicating a prosurvival function of Akt in melanoma.128 There have been several studies of Akt expression in melanoma. Using IHC, phospho-Akt was detected in 54% of nevi, 71.3% of primary melanomas, and 71% of melanoma metastases,129 and increased expression was seen in severe dysplastic nevi and metastatic melanomas versus benign nevi.128 Increased phospho-Akt expression in melanoma is associated with tumor progression and lower survival rate.130
PTEN is a tumor suppressor gene that has many functions. It inhibits the MAPK pathway,129 causes cell cycle arrest by upregulating p27,131 increases cell migration through dephosphorylation of focal adhesion kinase (FAK) and loss of focal adhesion formation,129 upregulates proapoptotic proteins including caspases and BH3 interacting domain death agonist, and downregulates antiapoptotic proteins such as bcl-2.131 Ectopic expression of PTEN in PTEN-deficient melanoma cells suppresses cell growth, inhibits colony formation, and reduces tumorigenicity and metastasis in mice.131 PTEN germline mutations result in Cowden disease, an autosomal dominant cancer predisposition syndrome129 in which there is not an increased risk of melanomas.132 Loss of tumor suppressor genes on chromosome 10 (including PTEN) is involved in 30% to 60% of noninherited melanomas,133 and loss of PTEN expression is seen in 30% to 50% of melanoma cell lines and 5% to 20% of primary melanomas.131 PTEN somatic mutations are seen in melanomas,131,134 where they occur in association with activating mutations in BRAF but not N-RAS.135 This is consistent with the ability of N-RAS to activate both the PI3K and MAPK cascades, abrogating the need for specific inactivation of PI3K. In the recent evaluation of genomic alterations in primary melanomas, tumors with BRAF mutations had fewer copies of PTEN than those with N-RAS mutations, suggesting that dual activation of the PI3K and MAPK pathways are important events in melanoma development.6 A recent study suggests that epigenetic silencing of PTEN through promoter methylation may also be an important means of inactivating this pathway in melanoma.136
|
TARGETING THE PI3K PATHWAY
|
|---|
Therapeutic targeting of the PI3K pathway is being approached from many angles (reviewed by Granville et al137; Fig 4). Wortmannin and LY2944002 both reliably inhibit the PI3K pathway in the laboratory, but they have not been used in humans due to pharmacologic limitations.137 PX-866 (Prolx, Tucson, AZ), a derivative of Wortmannin, is currently in preclinical development.138 Perifosine, an Akt inhibitor, is a phospholipid analog whose activity is purported to include inhibition of Akt phosphorylation and translocation to the cell membrane.137,139 Although no objective responses to single-agent Perifosine were seen in a phase II study in metastatic melanoma,140 no correlative studies were performed to evaluate baseline pathway activation or molecular response to therapy. RX-0201, an oligonucleotide complementary to Akt-1 mRNA, is being evaluated in a phase I trial141 (Table 1).
PDK-1, a kinase critical in Akt activation, is inhibited by UCN-01, which also inhibits PKC-, -β, and - and enhances apoptosis in combination with chemotherapy.137 As previously mentioned, a phase II study of UCN-01 in melanoma has completed accrual. Other agents that target individual PKC isoforms also are being evaluated (Table 1). Inhibitors of mTOR are the most mature agents in the attack on the PI3K pathway. mTOR is a serine/threonine kinase downstream of Akt that modulates protein synthesis, angiogenesis, and cell cycle progression.137 Uncontrolled signaling through the mTOR pathway commonly results from PTEN loss/inactivation,137 making mTOR inhibitors plausible agents in melanoma. Current mTOR inhibitors include Temsirolimus (CCI-779; Wyeth-Ayerst, Madison, NJ), Everolimus (Novartis) and AP-23573 (Ariad Pharmaceuticals, Cambridge, MA). A phase II study of Temsirolimus in 33 patients with metastatic melanoma yielded only one partial response lasting 2 months and a median time-to-disease progression of 10 weeks.142 However, interim analysis of a two-stage phase II trial with single-agent Everolimus in stage IV melanoma demonstrated 7 (35%) of 20 patients with stable disease at 16 weeks143 and accrual to the second stage is planned. Several trials combining mTOR inhibitors with agents aimed at other pathways are open to accrual (Table 1).
|
MICROPHTHALMIA-ASSOCIATED TRANSCRIPTION FACTOR AND MELANOMA
|
|---|
Microphthalmia-associated transcription factor (MITF) belongs to the MiT family of transcription factors.144 MITF consists of a basic domain for DNA binding, and helix-loop-helix and zipper domains for homo- and heterodimerization.145 One isoform, MITF-M, is specific for the melanocyte lineage due to the presence of a unique melanocyte-restricted promoter.144 It is the earliest known marker for the melanocyte lineage and essential for melanocyte development.146 MITF induces transcription of genes important for melanin production including tyrosinase, tyrosinase-related protein (Tyrp1), and dopachrome tautomerase (Dct).144 Expression of MITF is upregulated by the Wnt pathway through lymphoid enhancer-binding factor 1 (Lef1) and beta-catenin, alpha-melanocyte–stimulating hormone (-MSH) via cyclic-AMP (cAMP), c-KIT via MAPK, and the gp130 pathway (Fig 3).144,145,147
Although conflicting data exist regarding the role of MITF in melanoma development, evidence supporting a specific tumorigenic role is mounting. MITF is a necessary downstream effector of the proliferative beta-catenin pathway, and it also upregulates CDK2 expression, promoting cell cycle progression.144,148 In vitro studies have demonstrated that MITF overexpression in melanocytes, when combined with mutations in other oncogenic pathways, increases cellular transformation and anchorage-independence.149 The most compelling evidence for a role of MITF in melanoma development comes from studies of primary human melanoma tissues where MITF expression is usually conserved,148 and from the detection of copy gains up to 100-fold in some human melanoma samples as compared with melanocytes.150 Moreover, MITF gene amplification is seen in approximately 10% of primary cutaneous melanomas and 20% of metastatic tumors, but not in benign nevi, and disruption of MITF is lethal to melanoma cells with these amplifications.149,150 In metastatic melanoma patients, amplification of MITF was associated with decreased 5-year survival rates.150 Further, detection of circulating MITF transcripts was found to correlate with increasing American Joint Committee on Cancer staging and survival after treatment in melanoma patients.147 There is currently no data available to delineate particular melanoma subclasses that may be susceptible to genetic alterations in MITF. Further, the role of MITF in melanocyte and melanoma development is still evolving. There are no agents currently under clinical investigation that directly target MITF.
|
c-KIT AND MELANOMA
|
|---|
c-KIT (CD117) encodes a receptor tyrosine kinase whose ligand is stem cell factor (SCF or KIT ligand).151 Downstream effectors of c-KIT include MAPK and PI3K cascades, Phospholipase C, Src,151 and MITF.152 KIT-SCF signaling is essential for melanocyte development, differentiation, proliferation, survival, and migration.151 SCF is a mitogen that causes increased melanin production in normal melanocytes.153 Normal melanocytes are dependent on c-KIT signaling for survival, but this effect is lost in nevus cells.151 Inherited kinase-inactivating c-KIT mutations lead to piebaldism, characterized by loss of melanocytes and other anomalies.154
Activation of c-KIT has been noted in other cancer types,154 with various proposed mechanisms of activation.155 Data regarding c-KIT expression in melanoma is extensive and varied. However, there has been no consistent correlation between c-KIT expression and other melanoma variables (thickness, invasion, ulceration, age, site, and sex).155,156 Further, c-KIT expression has not provided prognostic value for disease-free survival as a single variable.155 Two separate studies identified c-KIT mutations in three human melanoma samples, out of a series of 153, including primary and metastatic.157,158 DNA sequencing revealed all three mutations to be somatic L576P activating mutations in exon 11, often seen in gastrointestinal stromal tumors (GISTs). None of these melanomas had mutated BRAF.157, 158</ |
TOP
ABSTRACT
INTRODUCTION