Rare families exist in which there is an increased risk of melanoma and in which the tendency to melanoma appears to be inherited as an autosomal dominant with incomplete penetrance, first recognized in the nineteenth century by Norris (1820). Within these families, the majority of melanomas are of the superficial spreading type, but there may be less common types such as nodular melanomas and lentigo maligna melanomas. Ocular melanomas may occur in some families, but this is very rare. In the UK families with the most common high penetrance susceptibility gene, CDKN2A so far reported, there is little evidence of increased susceptibility to cancers other than melanoma. However, other groups have reported an association in particular with pancreatic carcinoma, particularly in the Netherlands and in the USA (Bergman et al. 1990; Lynch and Fusaro 1991), and other gastrointestinal cancers.
In the families described with a founder mutation in the CDKN2A gene in the Netherlands (de Snoo et al. 2008), the increased risk of pancreatic cancer in these families was very clear in a recent study, but there was also evidence of increased risk of other cancers related to smoking such as cancers of the lip, mouth, pharynx, and respiratory system. Insufficient numbers of families from other centers have so far been studied in order to quantify the risk of nonmelanoma cancers, but it is clear that some families have an increased susceptibility to gastrointestinal cancer as well as to melanoma, which is currently being explored by the Melanoma Genetics Consortium, GenoMEL (www.genomel.org). It would seem prudent to advise family members strongly against smoking.
Some melanoma families also have the AMS, but not all. In a proportion of families, this abnormal nevus phenotype in melanoma cases may be striking. The phenotype is characterized by the presence of clinically atypical moles (by definition more than 5 mm in diameter with an irregular or blurred edge and irregularly irregular pigmentation), numerous but otherwise banal moles, and moles in unusual places such as on the buttocks, in the iris, and on the ears (Newton et al. 1993). The biological significance, even within melanoma-prone families, of this phenotype is unclear as some families with melanoma do not have abnormal nevi at all. In the UK families so far described, the members of the largest family, with nine cases of melanoma, all had normal nevi (Harland et al. 1997). Overall, it is clear that although the atypical mole syndrome is associated in some way with familial melanoma, it is a poor indicator of gene carrier status and cannot be used reliably even within these melanoma families to predict who is a gene carrier (Wachsmuth et al. 1998; Newton Bishop et al. 2000).
In deciding how atypical mole syndrome patients should be managed in terms of follow-up and risk estimation then, family history is the key.
Patients with this phenotype without a personal or family history of melanoma should be taught how to self-examine their nevi and be given advice about sun protection, but long-term follow-up is not normally appropriate. Patients with a strong family history of melanoma with or without the atypical mole syndrome should retain long-term access to the pigmented lesion service. The screening of nevi in patients with the atypical mole syndrome and a family history of melanoma is specialized, and all such patients should be referred to the local Cancer Centre Pigmented Lesion Clinic, usually run by a dermatologist. In such clinics, the emphasis is on clinical and dermoscopic examination of nevi, using photography for baseline documentation. Nevi are removed only if they appear to be changing, and therefore if malignant change is suspected, they are not excised prophylactically. Additional genetic counseling may be required.
Progress has been made in understanding the genetic basis of high-risk susceptibility. Initial reports of genetic linkage in melanoma families to chromosome 1p (Bale et al. 1989a, b) have not yet been substantiated by other groups. Strong evidence of linkage to chromosome 9p, reported by the Utah group (Cannon-Albright et al. 1992), was soon confirmed by others although genetic heterogeneity exists. The Tumours suppressor gene CDKN2A, which codes for the cyclin-dependent kinase (CDK) inhibitor p16, lies in the identified area of 9p, and all groups working on familial melanoma have now identified germline mutations in this gene so that to date the CDKN2A gene remains the major identified cause of high-risk familial melanoma. Overall, germline mutations in this gene have been identified in around 40 % of families with three or more cases of melanoma but much less frequently in families with only two cases. GenoMEL estimates an overall CDKN2A mutation penetrance for melanoma of 0.30 (95 % confidence interval (CI) = 0.12–0.62) by age 50 years and 0.67 (95 % CI = 0.31–0.96) by age 80 years, with evidence of greater penetrance where gene carriers live in sunnier climates, namely, Australia (Bishop et al. 2002). The confidence intervals for these estimations remain high so that GenoMEL will continue to improve its data. It has also been demonstrated that penetrance is higher in mutation carriers who also have certain MC1R polymorphisms (associated with a tendency to burn in the sun) (Demenais et al. 2010). Where CDKN2A mutation carriers have been ascertained from populations other than family studies, then the penetrance is predictably lower: the risk of melanoma in CDKN2A mutation carriers was approximately 14 % (95 % CI = 8–22 %) by age 50 years, 24 % (95 % CI = 15–34 %) by age 70 years, and 28 % (95 % CI = 18–40 %) by age 80 years in the GEM study (Begg et al. 2005; Soufir et al. 2000; Newton Bishop et al. 1994; Lal et al. 2000; Bartsch et al. 2002; Goldstein et al. 1994, 2004).
As above, in some families with germline mutations in CDKN2A, there is also an increased susceptibility to pancreatic carcinoma, manifest in patients over the age of 45 years (de Snoo et al. 2008). This appears to be particularly seen in families with truncating mutations such as the founder p16-Leiden mutation, but is also seen in families with the founder mutation G101W, common in Italy and France (Ghiorzo et al. 2007). Most of these G101W families in which pancreatic cancer was seen also had melanoma, but very rare mutation-positive families were reported with pancreatic cancer alone (Ghiorzo et al. 2004). In the p16-Leiden families, the incidence of pancreatic cancer was reported to be 29 times greater than population levels, and more recently a lifetime risk of 17 % was suggested. In the Italian G101W families, a 9.4-fold risk (95 % CI 0.8–5.7) was reported. The elevated risk in the p16- Leiden families is sufficiently high that MRI screening is being investigated in clinical trials as a means of early detection of pancreatic cancer.
GenoMEL has reported the predictors of CDKN2A mutation detection in melanoma families and shown that early age of onset, multiple primaries, multiple cases of melanoma, and the presence of pancreatic cancer are all associated with an increased probability of finding mutations (Goldstein et al. 2007), although there was no significant evidence of increased risk of pancreatic cancer risk in Australian families with CDKN2A mutations (Loo et al. 2003; Bishop et al. 2000; Della Torre et al. 2001; Gillanders et al. 2003; Harland et al. 2000).
There has been some suggestion of an increased risk of breast cancer in CDKN2A mutation-positive women (Borg et al. 2000), but this is unsubstantiated as yet and the risks of this and other cancers will be addressed by GenoMEL. Although families are reported in which both ocular and cutaneous melanomas occur, there is no evidence of an increased risk of ocular melanoma in CDKN2A mutation carriers.
Germline mutations in CDKN2A have been identified in around 10–15 % of melanoma patients with multiple primaries. The prevalence of germline mutations in sporadic “population” ascertained melanoma cases, however, is low: in the GEM study, an identifiable mutation was identified in 1–2 %, depending on the region of origin (Begg et al. 2005).
A second high-penetrance susceptibility gene was identified in 1996 when Dracopoli identified two families with the same single base pair substitution in another gene that coding for CDK4, producing a protein anomaly at the site at which p16, the CDKN2A product, binds (Zuo et al. 1996). Germline mutations in this gene are extremely rare. To date, only around seventeen families worldwide have been described with these mutations (Molven et al. 2005; Goldstein et al. 2007; Pjanova et al. 2007; Puntervoll et al. 2013).
However, the identification of the mutations strengthens the observation that the p16 protein is critical to melanoma carcinogenesis. Puntervoll et al. (2013) have recently reported that families with inherited CDK4 mutations have a very similar clinical picture to those with CDKN2A mutations suggesting that the CDK4 gene should be screened as well as CDKN2A in families who elect to proceed to gene testing.
Reported evidence of linkage to 9p in melanoma families in whom CDKN2A mutations could not be identified suggested that other CDKN2A mutations remained to be found. There was, however, no evidence for promoter mutations except for a mutation of the CDKN2A5′UTR which creates an aberrant initiation codon which has been identified in a number of families in North America, Australasia, and Europe (Liu et al. 1999). Splice site variants have been found (Harland et al. 2001, 2005) and 9p deletions (Mistry et al. 2005), but still a good proportion of melanoma families have no identifiable mutations.
CDKN2A is an unusual locus with an alternative reading frame coding for another protein p14ARF, whose role in melanoma carcinogenesis remains of great interest. In 1998 families with a susceptibility to melanoma and neural Tumours were described in whom there were germline deletions at 9p (Bahuau et al. 1998) and the suggestion was that there was the loss of CDKN2A and exon 1β, coding for p14ARF. More recently, a similar family was reported in which the deletion appeared to result in loss of exon 1β only, (Randerson- Moor et al. 2001) with no evidence of resultant impact on p16. This, the first evidence that p14ARF is the third high-penetrance melanoma susceptibility gene, was strengthened when a germline mutation in exon 1β, which creates an abnormal splice site, was reported in a melanoma family (Hewitt et al. 2002), and a small insertion affecting p14ARF was reported in a Spanish melanoma family (Rizos et al. 2001).
Overall then, the consensus is that there are 3 high-penetrance susceptibility genes effecting p16, CDK4, and p14ARF, and there is recent evidence for another at 1p22, with a probable role for BAP-1 in families with ocular melanoma (Harbour et al. 2010). In the UK currently, we have identified probable germline mutations in 53 and 68 % of three or more and four or more case families, respectively (Harland, Bishop and Newton-Bishop unpublished). This figure is lower in some other centers, particularly in areas of high incidence where lower-penetrance genes may be more evident (Goldstein et al. 2007). It is thought likely therefore that familial clustering in hot countries where people of Northern European origin live, such as Australia, has occurred as a result of inheritance of multiple lower- penetrance genes (such as those associated with pigmentation and nevus at-risk phenotypes described above) and intense sun exposure. The identification of BAP-1 germline mutations, however, means that remaining families might have “boutique” mutations, or mutations peculiar to their family or a very small number of families. Gene testing within melanoma families is taking place but the uptake is very variable internationally (de Snoo et al. 2003; Newton-Bishop et al. 2010; Kefford et al. 2002). Leachman et al. have argued the case for genetic counseling based upon a rule of threes (three or more cases in the family or three or more primaries in an individual) (Leachman et al. 2009), but the value of gene testing remains unclear to many. A positive CDKN2A mutation test may ultimately be of value when screening for pancreatic cancer has been shown to be effective and for families in which pancreatic cancer has occurred. There is a European screening research program called EUROPAC addressed to this issue. In the absence of pancreatic cancer, however, a positive test is unlikely to change management. As causal mutations are currently not identifiable in many families even with multiple cases of melanoma, the possibility is that there are as yet unidentified mutations, so that a negative test result is of limited utility. An online package for genetic counselors has been placed on the GenoMEL website.
The majority of the CDKN2A mutations identified to date appear to co-segregate with the Tumours, and from what is known about the structure of the p16 protein, it is to some extent possible to predict which mutations are real and which are silent. P16 (INK4A) is a member of a family of CDK inhibitors, the other members of which are p15 (INK4B), p18 (INK4C), and p19 (INK4D). The members of the group have significant sequence homology and have a structure dictated by the presence of four or five so-called ankyrin repeats. Mutations which fall outside the ankyrin repeats were thought likely to be nonsignificant having a lower likelihood of impacting on the protein, like the common CDKN2A polymorphism Ala148Thr. Some splice site variants outside these repeats are however increasingly being recognized. In order to prove that identified mutations are significant, functional tests of the p16 protein have been developed. The test most widely used is a test of the ability of the mutant protein to bind to CDK4 and CDK6. But it is not a comprehensive test of function. For mutations described in large families around the world such as Arg24Pro, Met53Ile, and 23ins24, there is confidence that the mutations are real and that gene testing would produce interpretable results. For novel mutations, an abnormal functional test result would seem to be necessary before testing. GenoMEL maintains a database of these mutations (eMelanoBase) which is now part of LOVD.