Screening for CDKN2A Mutations in Hereditary Melanoma

Cutaneous malignant melanoma (CMM) is a potentially fatal form of skin cancer whose etiology is heterogeneous and complex. Approximately 10% of the cases of CMM occur in persons with a familial predisposition (1), often in association with clinically dysplastic or atypical nevi (2). In 1993, a low-molecular-weight protein p16INK4A was shown to inhibit the activity of the cyclin D1-cyclin-dependent kinase 4 or 6 (CDK4 or CDK6) complex (3). This complex phosphorylates the retinoblastoma protein, allowing the cell to pass through the G1 cell-cycle checkpoint. The CDKN2A gene that encodes p16INK4A was localized to chromosome 9p21 (4,5), a region that has been implicated in melanoma by linkage, cytogenetic, and loss-of-heterozygosity studies (6–11). Somatic mutations in this gene have frequently been detected in many melanoma cell lines (4,5). Although initial reports of germline mutations in melanomaprone families (selected for genetic analyses) suggested that the p16INK4A protein might be important in approximately one half of inherited melanoma (12–14), subsequent studies have revealed lower frequencies of mutations. To date, germline CDKN2A mutations have been detected in approximately one fourth of melanoma-prone kindreds (defined as having two or more first-degree relatives with confirmed CMM) from North America, Europe, and Australia (12–21). In this issue of the Journal, Platz et al. (22) present the first report of screening for germline mutations in a population-based cohort of melanoma-prone families. The results provide further evidence that CDKN2A mutations may not be as prevalent as originally thought. Platz et al. (22) identified 177 melanoma-prone families, defined by the occurrence of two or more relatives (including first-, second-, and third-degree relatives) with histopathologically confirmed CMM, in the County of Stockholm, Sweden. A representative selection of 181 individuals with CMM, from 100 of these families, was screened for germline mutations in CDKN2A. Of the 100 families, 64 kindreds had at least two cases of melanoma in first-degree relatives; 24 kindreds had melanoma in two or more second-degree relatives; and 12 kindreds had melanomas in third-degree or more distant relatives. Both direct sequencing and single-strand conformation polymorphism (SSCP) analyses of exons 1 and 2 were conducted. Of the 64 families with two or more first-degree relatives with CMM, five (7.8%) had potentially disease-related mutations. Four of the five families with mutations had more than the mean (2.3) number of melanoma cases. Ten kindreds without detectable mutations also had more than two cases of melanoma in first-degree relatives. No mutations were detected in the 36 families with melanoma in more distant relatives (22). The frequency of CDKN2A mutations detected by Platz et al. (22) was lower than that in the initial reports of families selected for genetic analyses, yet the observations are consistent with numerous later results (13,15,17,20,23). There are several possible explanations for the divergent frequencies. First, part of the difference may be due to chance, since several studies examined small numbers of families. Second, the families that were studied likely represented a heterogeneous mix of hereditary and nonhereditary melanoma-prone kindreds. Certain investigations required three or more first-degree relatives with CMM, others required only two such relatives, and still others required only two relatives with CMM not restricted to first degree. Third, germline mutations may have been present in some families but undetected either because only one patient per family was screened or because the mutation was not revealed by direct sequencing or SSCP analysis. Mutations outside the CDKN2A coding region that interfere with RNA splicing or expression would not have been detected (24). In addition, alternative processes linked to transcriptional and/or translational regulation including methylation might also inactivate p16INK4A (25). The frequency of such undetected mutations and alternative mechanisms of inactivation is difficult to estimate. Finally, the complexity and heterogeneity of CMM likely contributed to the divergent frequencies. Previous studies (12,19,26) have demonstrated statistically significant evidence for genetic heterogeneity to the chromosome 9p21 region as well as the identification of a second melanoma gene CDK4 in two melanoma-prone families (27). All of these alternatives likely contribute to the divergent estimates of CDKN2A mutations across populations and make it difficult to determine precisely the proportion of hereditary melanoma cases associated with p16INK4A alterations. Even with the uncertainty about the frequency of CDKN2A alterations in hereditary melanoma, testing is commercially available, and suggestions about screening families and the general population for CDKN2A have been made (28). In the American Society for Clinical Oncology (ASCO) statement on genetic testing for cancer susceptibility, CDKN2A [included in Table 2 (28) as p16] is listed in Group 3 “tests for individuals without a family history of cancer, in which the significance of the detection of a germline mutation is not clear; or tests for hereditary syndromes for which germline mutations have been identified only in a small number of families; or for which the medical benefit of the identification of a heterozygote (‘carrier’) is not established” (28). The ASCO statement recommends cancer predisposition testing be offered only when “… the test can be adequately interpreted; and the results will influence the medical management of the patient or family member” (28). Neither condition is currently met for CDKN2A screening and hereditary melanoma. Given the complexity and etiologic heterogeneity of melanoma, the vast majority of patients with a positive family history will not have a CDKN2A mutation detectable by SSCP or direct sequencing analysis. Thus, a negative test in such a patient (outside a family with a CDKN2A mutation) is uninformative. Even among families with CDKN2A mutations, however, it is likely that multiple genes and environmental factors are important in determining disease risk (29–31), making test results, both negative and positive, difficult to interpret. And, at present, a positive test in a patient with melanoma will not materially influence patient care. An essential part of patient management is the early diagnosis of melanoma or lesions in transition from nevi to melanoma. Survival after melanoma is directly related to tumor thickness (32,33). Management of patients with CMM, both familial and nonfamilial, should aim to prevent melanoma by reducing risk factors that promote tumorigenesis and detect melanomas early by recognizing and performing a biopsy on nevi that are suspected of being melanoma. Knowledge about the CDKN2A (Rb) pathway may lead to the development of better diagnostic and therapeutic regimens. Only then should screening melanoma-prone family members and/or other high-risk individuals for germline CDKN2A mutations be reconsidered.