Role of CDKN2A Mutations and Other Relevant Genes in Melanoma Predisposition

CDKN2A: Identification of the Major Melanoma Predisposition Gene

The first evidence of genetic inheritance of melanoma was produced by the observations of Dr William Norris in . Dr Norris described a male melanoma patient who developed a tumor from a mole; the patient’s father had died from a similar disease, and the patient’s sons had developed several moles. However, it was not until 1952 that studies of the hereditary aspects of melanoma were resumed, which were reinitiated by the investigations of and followed by additional descriptions of melanoma families. At that time, familial melanoma was characterized by an early age of onset and an autosomal dominant pattern of inheritance; however, no “melanoma gene” had been identified.

Succeeding studies suggested that atypical moles, which were defined in relation to color, border, and histological characteristics, are a melanoma risk factor. Thereafter, the search for a melanoma gene has included evaluating the occurrence of both melanoma and dysplastic nevus and has resulted in identifying a candidate locus at chromosome 1p ( ). However, the association of the 1p locus with hereditary melanoma was not subsequently validated in additional melanoma families ( ).

Between the 1970s and 1980s, results from experiments conducted on murine and human melanoma cell lines indicated an association between chromosome 9p and the development of melanoma ( ). A candidate 9p21 locus was tested by Cannon-Albright and colleagues using linkage analysis of 11 relatives who presented with invasive melanoma, and these authors delimitated the melanoma susceptibility locus using IFNA and D9S126 probes ( ). The identification of a melanoma patient who was carrying a germline translocation between 5p and 9p in association with a deletion of both IFNA and D9S126 markers provided further support for the importance of this genomic region ( ). After delimitation of the candidate locus, two independent groups ( ) demonstrated that this region was frequently altered in melanomas and cell lines, similarly to the previously described p16 locus (also called CDKN2 or MTS1 ), a gene that had been demonstrated to be associated with control of the cell cycle. Only a low frequency of p16 mutations was subsequently identified in melanoma families ( ). Indeed, germline mutations affecting p16 have been detected in only a portion of the melanoma families that have been studied to date ( ). A description of a homozygous individual who carried a p16 deletion but had not developed melanoma over the course of his life indicated that lack of p16 can be not enough to spur melanoma development ( ).

In 1995, two separate groups reported a new transcript for the p16 locus, p14 , which partially shares the sequence of the first identified transcript ( ). Later, two different germline mutations were identified in an exclusive region of p14 : a deletion in one family with melanoma and nervous system tumors ( ), and an insertion in one melanoma patient who had developed multiple melanomas ( ). Altogether, the collected family data demonstrated that mutations in both p16 and p14 proteins led to melanoma predisposition, supporting that CDKN2A (the official name of the p16/p14 locus) was the melanoma susceptibility gene located at 9p21.

CDKN2A Mutations

The CDKN2A gene encodes two different proteins with distinct functions: p16 and p14. The sequence codifying these two protein isoforms differs within the first exon (exon 1α and exon 1β for p16 and for p14, respectively) as a result of alternative splicing and a different reading frame ( ). The p16 protein acts as a tumor suppressor by inhibiting the CDK4 protein, which impairs the activation of pRb and results in cell arrest at the G1 phase of the cell cycle. The p14 protein also acts a tumor suppressor and inhibits the MDM2 protein, leading to p53 stabilization and G1 cell cycle arrest ( ).

Up to 40% of familial melanoma cases are caused by germline CDKN2A mutations ( ). In contrast, the detection rate of mutations in this gene is low in patients presenting with melanomas during childhood or adolescence ( ). Several characteristics of melanoma patients and their families, such as ≥2 affected family members, early age at diagnosis, and occurrence of multiple primary melanomas and pancreatic cancer, have been significantly associated with CDKN2A mutations ( ). According to the ClinVar database ( http://www.ncbi.nlm.nih.gov/clinvar ; April 2014), 58 pathogenic genetic alterations have already been reported in the CDKN2A gene, and 20 of them are germline mutations that were identified in melanoma families. In Table 7.1 , we have compiled a list of the CDKN2A mutations that have been most frequently associated with hereditary melanoma according to reports published between January 2010 and March 2015. The pathogenic mutations associated with melanoma predisposition have been primarily detected in exons 1α and 2 ( ). Mutations affecting the p16 isoform are far more frequent and are predominantly loss-of-function missense mutations, whereas the inactivating mutations that have been reported for the p14 isoform include deletions, insertions and splicing mutations ( ).

An increased CDKN2A penetrance has been identified in select regions of the world that exhibit a high melanoma incidence; however, the overall penetrance of CDKN2A has been estimated to be 30% by age of 50 and 67% by age of 80 ( ). Additionally, the relative risk of melanoma among carriers of CDKN2A mutations could diverge depending on the pathogenic variant; for instance, the odds ratio associated with the −34G>T mutation was estimated to be 15.3, whereas missense mutations that affect only p14 were determined to have odds ratios of 0.7 ( ). Other genetic variants can also influence CDKN2A penetrance. Variants of the melanocortin-1-receptor ( MC1R ) gene have been reported to increase CDKN2A penetrance because 81% of patients carrying both a CDKN2A mutation and an MC1R variant had developed melanoma before 50 years of age, whereas the estimated penetrance for these patients if they harbored only a CDKN2A mutation would be 57% ( ). The effect of MC1R on CDKN2A penetrance appears to be stronger in carriers of multiple and red hair color variants ( ). Conversely, null GSTT1 alleles were described to confer a protective effect with respect to melanoma risk even among carriers of CDKN2A mutations ( ). Regarding environmental factors, contradictory results of both positive and minor effects have been reported with regard to UV exposure ( ).

Carriers of CDKN2A mutations have been reported to be predisposed toward developing additional types of cancer, including nervous system tumors and pancreatic cancer ( ). Additionally, patients who smoke and who also carry the CDKN2A p.Arg112dup mutation have been shown to be at high-risk of developing tumors in respiratory and upper digestive tissues ( ).

Although the importance of CDKN2A mutations in melanoma susceptibility has been well established, there is currently no consensus regarding the indication of genetic screening in melanoma-prone patients, primarily because a positive result for a CDKN2A mutation does not affect clinical care, and data corroborating that such a result improves prevention is insufficient ( ). However, the detection of a CDKN2A germline mutation in a melanoma family does enable the identification of family members who are not at risk of developing melanomas and therefore can exclude them from undergoing further surveillance.

CDK4

In 1995, a CDK4 mutation (R24C) was identified that impaired the association between CDK4 and p16 and therefore contributed to melanogenesis via the disruption of cell cycle regulation ( ). The same mutation was detected later in two unrelated melanoma families without CDKN2A mutations; following this, a second type of CDK4 mutation (R24H) was discovered in a French melanoma family, strengthening the importance of this locus in melanoma predisposition.

It is noteworthy that in melanoma cells a single mutant CDK4 allele is sufficient to induce deregulation of cell cycle control, whereas both CDKN2A alleles must be mutated to produce similar results. However, no phenotypic differences have been observed when comparing melanoma patients who are carriers of mutations in these genes. Furthermore, it appears that MC1R variants also influence the penetrance of CDK4 mutations because only a low frequency of red hair color variants was detected in unaffected individuals carrying CDK4 mutations ( ). Although CDK4 can be considered a high-risk melanoma gene, its detection rate is very low; currently, only 17 melanoma families harboring CDK4 mutations have been reported ( ).