|Home | About | Journals | Submit | Contact Us | Français|
Approximately 5% to 10% of melanoma may be hereditary in nature, and about 2% of melanoma can be specifically attributed to pathogenic germline mutations in cyclin-dependent kinase inhibitor 2A (CDKN2A). To appropriately identify the small proportion of patients who benefit most from referral to a genetics specialist for consideration of genetic testing for CDKN2A, we have reviewed available published studies of CDKN2A mutation analysis in cohorts with invasive, cutaneous melanoma and found variability in the rate of CDKN2A mutations based on geography, ethnicity, and the type of study and eligibility criteria used. Except in regions of high melanoma incidence, such as Australia, we found higher rates of CDKN2A positivity in individuals with 3 or more primary invasive melanomas and/or families with at least one invasive melanoma and two or more other diagnoses of invasive melanoma and/or pancreatic cancer among first- or second-degree relatives on the same side of the family. The work summarized in this review should help identify individuals who are appropriate candidates for referral for genetic consultation and possible testing.
It is estimated that 5% to 10% of all malignant melanomas occur in familial clusters1 and work continues to identify all the genetic factors that play a role in melanoma risk and optimal ways to use this information in treatment of the individual patient. There is increasing awareness among health care professionals and the public about the inherited basis of many cancers and the availability of genetic testing for relevant predisposing gene mutations.2 Dermatologists and other health professionals should incorporate family history and risk assessment into clinical practice to identify patients who may be at increased risk for melanoma.
Germline mutations in cyclin-dependent kinase inhibitor 2A (CDKN2A) (INK4a) are reported to be present in up to 40% of hereditary cases of melanoma, making it the most significant high-risk melanoma susceptibility gene identified to date.3,4 Mutations in CDKN2A are associated with increased risks for both melanoma and pancreatic cancer. By age 80 years, an individual ascertained from multiple-case families with a CDKN2A mutation has an increased risk of developing melanoma of 58% in Europe, 76% in the United States, and 91% in Australia.5 Gene penetrance estimated by population-ascertained mutation carriers is considerably lower, although still substantial (28% by age 80 years).6 The risk appears to vary between countries and families, and it is not yet clear whether this variation results from the type of mutation, coinheritance with other genetic variations, environmental exposures, or other not yet identified genetic variables. A correlation has been established between the presence of a CDKN2A mutation and pancreatic cancer risk in some families.7–19 Within families that demonstrated a predisposition to pancreatic cancer, the relative risk for pancreatic cancer ranges from 9.4 (95% confidence interval 2.7–33.4)19 to 47.8 (95% confidence interval 28.4–74.7),20 or up to a 25% risk of developing pancreatic cancer by age 80 years in a study of CDKN2A-positive families in the Netherlands.
The classic family history features that raise the possibility of an inherited cancer syndrome are multiple affected family members (particularly with a vertical pattern of inheritance), occurrence of cancer types known to be associated with a specific hereditary syndrome, individuals given the diagnosis of multiple primary cancers, and early age of onset. Here we provide a review of the literature from 1994 to 2007 and an assessment of the predictive value of these features to identify families that are likely to harbor a mutation in the CDKN2A gene and would benefit from referral for genetic consultation and possibly testing. Data in this review were summarized into tables and specified by country or region where the study took place. Although we have attempted to synthesize these data into general guidelines that are internationally applicable, the variability in incidence and penetrance of CDKN2A mutations in different populations is such that clinicians must consider the patient’s geographic region and genetic background when assessing individual patient risk.
There are no current relevant data on in situ melanomas and/or the lentigo maligna subtype of in situ melanoma, and these should not be counted as a melanoma for purposes of genetic risk assessment. Ocular melanomas are also not considered in this article.
The likelihood of CDKN2A mutation detection increases with the number of melanomas in the family (Table I; available online at www.eblue.org). Data from the two international melanoma consortium studies, one a combined analysis of familial studies and the other a cross-sectional survey of melanoma cases, indicate that the incidence of CDKN2A mutations in families with only one melanoma is approximately 1% whereas the likelihood of mutation detected in 2, 3, or 3 or more affected family members with melanoma are 4%, 8%, and 38%, respectively.4,6 The combined GenoMEL analysis and data in Table I (available online at www.eblue.org) show that mutation detection rates are highly variable across regions. Families with similar histories have a greater likelihood of harboring a mutation in lower incidence countries. Importantly, mutation prevalence rates do not increase above 10% in high incidence regions such as Australia until there are at least five cases of melanoma in the family. In addition, mutations are more likely to be found in familial melanoma that has been identified by clinic-based ascertainment than in similar families identified by population-based ascertainment.
Approximately 3% to 5% of all patients with melanoma will develop additional primary melanomas in their lifetime.21 As with family history, the prevalence of CDKN2A mutations increases with the number of primary melanoma diagnoses in the individual (Table II; available online at www.eblue.org). Data from the Genes Environment and Melanoma Study Group (GEM) indicate that the likelihood of a CDKN2A mutation in an individual with two or more primary melanomas is 2%, but increases to 7% if additional family history is present. The likelihood of mutation detection continues to increase with greater numbers of primary melanomas. Studies of patients having ≥4 melanomas indicate a 29% to 100% likelihood of mutation detection, at least in low incidence countries. Because of the impact of family history on the likelihood of mutation detection, the data from studies of multiple primary melanomas are subdivided in Table II (available online at www.eblue.org) by whether or not additional family history has been excluded.
GenoMEL data demonstrate that 28% of 178 families known to carry a CDKN2A mutation also had one or more pancreatic cancers in the family.4 However, further analysis of these data by geographic location shows that the CDKN2A mutation-positive families from Australia do not have a significant association with pancreatic cancer, whereas there is an association in Europe and North America.3 Although it is not a feature of all CDKN2A families, the presence of pancreatic cancer in a family with melanoma greatly increases the likelihood of mutation detection (Table III; available online at www.eblue.org). A GenoMEL analysis of families with 3 or more melanomas found CDKN2A mutations in 38%.4 However, if these families also had a pancreatic cancer diagnosis, the likelihood of a mutation went up to 72%4 (Table III; available online at www.eblue.org). Data on individuals presenting with double primaries (one melanoma, one pancreatic cancer) are limited. Review of data from 5 separate studies looking at a combined total of 21 individuals with double primaries found that 3 (15.0%) were found to have a CDKN2A mutation. Despite the association of pancreatic cancer and CDKN2A, neither sporadic nor familial pancreatic cancer appears to be a predictor of harboring a mutation. Studies of unselected patients with pancreatic cancer who have no family history of melanoma or pancreatic cancer indicate a 2% likelihood of mutation detection.13,22 These studies suggest that it is the combination of both pancreatic cancer and melanoma that increases the likelihood of a CDKN2A mutation, but that isolated pancreatic cancer and familial pancreatic cancer may not (Table III; available online at www.eblue.org).
A common feature of hereditary cancer syndromes is a younger age of diagnosis compared with the mean age of diagnosis for that particular cancer in the general population. Table IV (available online at www.eblue.org) compiles data on age of diagnosis of melanoma and pancreatic cancer in CDKN2A mutation carriers. The mean age of melanoma diagnosis in CDKN2A mutation carriers across the world is in the 30s to 40s, whereas the mean age in high-risk melanoma families without CDKN2A mutations is in the 40s to 50s. In the United States, the mean age of diagnosis of melanoma in known CDKN2A mutation carriers is 35 years (range 14–68 years)5 compared with a median age of 59 years in the general population.23 There is a wide range in age of diagnosis of sporadic melanoma, with very rare cases seen at younger than 10 years to older than 90 years.
Although younger onset is clearly a feature of CDKN2A mutations, in the absence of additional family history, young onset of melanoma alone does not predict a high likelihood of an identifiable mutation. Mutations are identified in less than 1% of individuals given the diagnosis of melanoma when they are younger than 40 years (Table V; available online at www.eblue.org).24–31 Overall, selection of patients based on young age of melanoma diagnosis alone does not result in a sufficiently high likelihood of finding a mutation to merit referral.
Some CDKN2A mutation-carrying families that exhibit numerous clinically atypical nevi (CAN) (defined based on atypical clinical features alone) and dysplastic nevi (DN) resulting in fulfillment of the formal criteria for familial atypical multiple mole melanoma syndrome32 or atypical mole syndrome.33,34 Within melanoma-prone families, whether CDKN2A-linked or not, the presence of CAN/DN is a strong risk factor for melanoma development; however, some individuals who develop melanoma in this setting do not have these markers.35 Furthermore, the association of CAN/DN with mutation carrier status in known CDKN2A mutation families is complex, and many studies have indicated the nevus phenotype to be a very unreliable indicator of CDKN2A mutation carrier status.36–40
CAN/DN is also seen outside melanoma-prone families either sporadically or genetically. There have been limited studies to date examining germline CDKN2A mutation status in patients with CAN. Celebi et al41 found no CDKN2A mutations in a study of 28 patients with CAN and Ung-Juurlink42 found 8 mutations in 251 (3.2%) patients presenting with melanoma, CAN, or both. In this study, the phenotype of the mutation carriers was not specified, but univariate analysis did not detect any relationship between CAN alone and CDKN2A status. Matsumura et al43 found no mutations in 4 patients with nonfamilial CAN, and de Snoo et al44 found 6 mutations in 167 (4%) patients with CAN, of whom 4 of the 6 turned out to have a positive family history for melanoma and one had 4 primary melanomas.
In summary, CDKN2A genetic testing in patients with CAN/DN without a positive family history of melanoma is not justified based on current data.
This article does not discuss the role of genetic testing for two other high-penetrance melanoma predisposition genes, cyclin-dependent kinase 4 CDK4 or cyclin-dependent kinase inhibitor 2A/p14 alternate reading frame CDKN2A/ARF.4 Risk estimates associated with mutations in these genes have wider confidence intervals than those estimated for CDKN2A mutations because so few have been reported. In patients who have a strong family history and are negative for CDKN2A mutation, these tests could be considered but are unlikely to be positive.
Melanocortin 1 receptor variants are associated with red hair and freckles. Melanocortin 1 receptor variants confer significant additional melanoma risk to CDKN2A mutation carriers and further refinement of this risk is ongoing in many research laboratories.45,46 Melanocortin 1 receptor testing is currently available as a research investigation.
Genetic testing is currently widely used for identifying individuals with hereditary colorectal cancer and hereditary breast/ovarian cancer, but genetic testing of CDKN2A in the context of melanoma is not part of routine practice. However, there are now at least 5 commercial laboratories in the United States offering clinical CDKN2A testing,47 and there is growing awareness by the lay public about the genetic basis of cancer and the availability of testing. The objective of this article is to help clinicians identify individuals who are at significant risk for harboring a genetic mutation and who could be referred to a genetic counseling specialist.
We have summarized the predictive value of personal and family history of melanoma and pancreatic cancer for identifying individuals who have an increased probability of harboring a mutation in the CDKN2A gene. The likelihood of detecting a CDKN2A mutation depends greatly on the population being studied, which may be a result of differences in penetrance associated with variation in melanoma predisposing phenotype (eg, fair skin, red hair) and the local amount/intensity of ultraviolet radiation exposure. In geographic areas with higher background rates of melanoma, there is greater likelihood of having multiple family members with melanoma or multiple primary melanomas caused by reasons other than a CDKN2A mutation. However, melanoma penetrance in CDKN2A mutation carriers is also higher in areas with high background rates of melanoma, indicating a potential interaction between CDKN2A and the other predisposing factors for melanoma in these areas.
The variability in the background incidence of melanoma and penetrance of CDKN2A mutations between countries is such that there is no single guideline for genetic testing that would be appropriate to apply worldwide. We, therefore, provide a framework that clinicians can use to identify appropriate candidates for genetic evaluation with regard to the specific populations they serve. For moderate to high melanoma incident areas, individuals with 3 or more primary melanomas and/or families with at least one invasive melanoma and two or more other diagnoses of melanoma and/or pancreatic cancer in aggregate among first- or second-degree relatives on the same side of the family are appropriate candidates for a genetics evaluation (Table VI; available online at www.eblue.org). For low melanoma incidence areas, two melanoma and/or pancreatic cancer events in a family may be sufficient to consider a genetics referral (Table VI; available online at www.eblue.org). There are insufficient data at this time to specifically determine the likelihood of mutation detection in individuals presenting with synchronous or metachronous diagnoses of melanoma and pancreatic cancer. However, this is another group that may warrant referral for genetics evaluation.
There are important considerations regarding the clinical use and potential implications of CDKN2A genetic testing. Before undergoing genetic testing, patients should be informed of the potential benefits and limitations of testing by a genetic counselor or other professional with expertise in melanoma genetics (Table VII; available online at www.eblue.org).48 To date there are limited data regarding the implications of CDKN2A genetic testing. Aspinwall et al49 found an increase in screening and precautionary behaviors among both mutation-positive and mutation-negative patients. After receiving test results, 55% reported adopting at least one screening behavior. Long-term follow-up data are needed to determine whether these behavioral changes are maintained, but data from testing of other hereditary cancer syndromes indicate that noncarriers in a mutation-positive family are likely to continue to undergo risk-appropriate screening.50,51 Therefore, it is important that genetic testing be done in the context of counseling and education.
Regardless of whether or not genetic testing is part of the care for families with hereditary melanoma, there is likely benefit from identifying these highest risk families and targeting them for intensive screening and education.
The higher rates of CDKN2A mutation positivity in individuals with 3 or more primary melanomas and/or families with at least one melanoma and two or more other diagnoses of melanoma and/or pancreatic cancer in aggregate among first- or second-degree relatives on the same side of the family warrant referral for a genetics evaluation. Use of these guidelines would increase the proportion of individuals identified at high risk and referred appropriately to genetic services. Patients at high risk should be allowed to weigh the pros and cons of testing and will–irrespective of actually being testing–benefit from tailored education and screening.
We would like to thank Sherri Bale at GeneDx Inc, Gaithersburg, MD, and Cindy Solomon and Jean Schaller at Myriad Genetic Laboratories, Salt Lake City, UT, for multiple discussions on genetic testing for melanoma.
Supported by numerous grants. The work of GenoMEL, including Drs Leachman, Bergman, Debniak, Newton-Bishop, Puig, Bianchi-Scarrà, Kefford, Mann, Tsao, and Elder, is supported by the National Institutes of Health (NIH) RO1 CA-83115 and GenoMEL 01872 Network of Excellence. The work of Dr Puig is partially supported by Fondo de Investigaciones Sanitarias, grant 0019/03 and 06/0265, National Cancer Institute (NCI). The work of Ms Kohlmann and Dr Leachman is supported by the Huntsman Cancer Foundation Genetic Counseling Shared Resource and core facilities supported by P30 CA042014 awarded to Huntsman Cancer Institute. The work of Dr Bressac-de Paillerets is supported by the Institut National du Cancer, Réseau Oncogénétique pour les Cancers Rares–Mélanome. The work of Dr Newton-Bishop is supported by Cancer Research-United Kingdom grant C588/A4994. The work of Drs Goldstein and Tucker are supported by the Intramural Research Program of the NIH, Division of Cancer Epidemiology and Genetics. The work of Dr Asgari is supported by the National Institute of Arthritis Musculoskeletal and Skin Diseases (K23 AR 051037). The work of Dr Tsao is supported by P50 CA-93683 (NCI), RSG MGO-112970 (American Cancer Society), NIH R01 CA-83115.
Conflicts of interest: None declared.