Few long-term clinical and histologic data for melanocytic lesions have been available based on the mutation status of families at an increased risk of melanoma. In the current study, the authors describe the clinical and histologic features of dysplastic nevi and melanoma over time in families at an increased risk of melanoma with differing germline mutations in CDKN2A, CDK4, or not yet identified genes.
Thirty-three families with > 2 living members with invasive melanoma were evaluated clinically and followed prospectively for up to 25 years. All the participants were evaluated by the same study team at the Clinical Center of the National Institutes of Health or in local clinics. After informed consent was obtained, family members (n = 844) were examined and photographed. Blood was obtained for genetic studies; genotyping for CDKN2A and CDK4 was performed. Sequential photographs of melanocytic lesions were taken as part of the clinical evaluations. When melanocytic lesions were removed, the histology was reviewed. Representative photographs and photomicrographs were selected for six classes of lesions and three mutation groups.
All the families were found to have members with dysplastic nevi and melanoma; 17 had mutations in CDKN2A, 2 had mutations in CDK4, and 14 had no mutations in either gene identified. The majority of dysplastic nevi either remain stable or regress; few change in a manner that should cause concern for melanoma. With careful surveillance, melanomas can be found early.
Dysplastic nevi first were described over 20 years ago among members of melanoma-prone families.1–4 In many melanoma-prone families, dysplastic nevi are important risk factors5–8 and nonobligate precursor lesions for melanoma.9–14 Since these relations were established, the presence of dysplastic nevi has been used clinically to identify the individuals in these melanoma-prone families who are at the highest risk of developing melanoma.5, 6, 8 Based on the presence of dysplastic nevi, clinical guidelines for members of high-risk families were developed. The guidelines include surveillance of pigmented lesions, routine clinical examinations, and the use of sun protective measures. Adherence to these guidelines appears to decrease the risk of new melanoma and changing nevi and aids in the detection of melanoma at an earlier stage.15
It now is well established that the etiology of familial melanoma is heterogeneous and complex.16 Two genes have been identified: CDKN2A, a tumor suppressor, and CDK4, a protooncogene. Mutations in CDKN2A have been found in approximately 20% of families evaluated worldwide, whereas mutations in CDK4 have been reported in 3 families.17–19 Both these genes play an important role in cell cycle control in the retinoblastoma pathway. Despite one gene being a tumor suppressor and the other a protooncogene, there are no apparently significant differences in phenotype between families with CDKN2A mutations and families with CDK4 mutations.1 The estimated penetrances of these genes remain imprecise. Therefore, to make use of mutation status for melanoma risk assessment is problematic.20 Additional genes that are significant in the development of melanoma currently are being sought.
As genes associated with melanoma susceptibility have been identified, the role of dysplastic nevi in the clinical management of family members has been questioned.21 Although dysplastic nevi and melanoma early on were hypothesized to be pleiotropic effects of a single gene,22 dysplastic nevi recently have been shown to be a risk factor for melanoma independent of mutation status in these families.23 These nevi also are a substantial risk factor for melanoma outside of melanoma-prone families24–33 and are estimated to occur in 5–10% of the general population.34, 35 The natural history of dysplastic nevi in individuals who are not from melanoma-prone families has been described previously.26–30
Although some reports have found a correlation between the clinical and histologic characteristics of dysplastic nevi, the genetic status of the affected individuals has not been included. The current atlas demonstrates the natural history of dysplastic nevi and primary melanomas within melanoma-prone families with differing germline mutations. We prospectively observed and photographically documented these lesions for > 20 years.
MATERIALS AND METHODS
Thirty-three families with ≥ 2 living members with invasive melanoma were referred by health care professionals (or were self-referred) to the National Cancer Institute (NCI). The families are geographically dispersed within the U.S., but the majority come from the mid-Atlantic region. Diagnoses of melanoma were verified by histologic review of the primary tumor (n = 258) or metastatic lesions (n = 17), local pathology report (n = 19), or medical record or death certificate (n = 7).
After providing written informed consent for participation in the NCI institutional review board-approved protocol, all willing family members (n = 844) underwent a medical history review, full skin examination, photography, and phlebotomy. Information recorded from the skin examination included degree of solar damage, extent of freckling, total number of nevi, pattern of nevi, and presence of dysplastic nevi. Individuals were classified as clinically affected with dysplastic nevi if they had multiple lesions which were at least 5 mm in dimension with a flat component and at least 2 of the following 3 characteristics: variable pigmentation, irregular asymmetric outline, and indistinct borders. Routine initial photography included overviews of all skin surfaces and close-up 1:1 photographs of the most atypical nevi. During and after the skin examination, family members also were instructed in sun protective measures, conduct of self-examinations, characteristics of dysplastic nevi, and recognition of the warning signs of melanoma. Recommended routine care in these families included monthly self-examination for those with dysplastic nevi (less frequent for those without); regular (3 month–1 year) health care worker examinations, depending on the activity of nevi; and the excision of nevi observed to be changing in a manner worrisome for melanoma or of new lesions suspicious for melanoma.
After initial evaluation at the NCI, routine care was provided by local physicians for the majority of family members. Follow-up information, including new diagnoses or excision of pigmented lesions, was obtained regularly. When pigmented lesions were excised, every effort was made to obtain slides for histologic review. The criteria used for diagnoses of invasive melanoma, radial and vertical growth phase melanoma, in situ melanoma, and dysplastic nevi have been published previously.9–14 For this article, melanocytic lesions is an inclusive term for benign and malignant tumors of the melanocyte, including banal nevi. Dysplastic nevi are clinically or histologically diagnosed; melanomas are histologically diagnosed. For staging, we used the proposed American Joint Committee on Cancer (AJCC) staging system.36
Families also were reevaluated intermittently at the NCI. During these clinical examinations, if nevi were found to have changed in the interval since the last photographs, new close-up and overview photos were taken. Prospective follow-up of these families ranged from 2–25 years. The frequency of examinations at NCI varied from 3 months–10 years.
Mutation analyses were conducted as previously described.1 Mutations in CDKN2A were described in 17 families, CDK4 mutations were described in 2 families, and no mutations were found in 14 families. Although mutations in CDKN2A or CDK4 were not identified, these 14 families appeared to have quite similar inheritance patterns of melanoma compared with those with identified mutations. It is likely that additional melanoma susceptibility genes will be identified in these families. The age at the time of first diagnosis of melanoma, the number of primary melanomas per individual, the thickness of the melanomas, and the presence of dysplastic nevi did not appear to vary among the three categories.1, 37
For this atlas, melanocytic lesions with available histology suitable for photomicrographs were reviewed. Representative photomicrographs of informative lesions were taken. Clinical photographs also were reviewed to identify two classes of lesions: those with photomicrographs that had photographs prior to excision and those not excised that had sequential photographs. Among the latter group of lesions, pigmented lesions that developed into dysplastic nevi, dysplastic nevi stable for long periods, or dysplastic nevi that regressed and disappeared were selected to illustrate representative phases in the natural history. Among the lesions in individuals with CDKN2A, CDK4, or unknown mutations, the following were selected: prevalent melanoma at the time of first examination, new melanoma during follow-up, dysplastic nevus changing in a manner worrisome for melanoma, new dysplastic nevus, stable dysplastic nevus, and involuting dysplastic nevus. For the last three groups, there was no histopathology.
Prior to the families' enrollment into the current study, 194 melanomas occurred among 140 individuals. Of the 151 primary lesions for which we were able to measure thickness on review, 100 were classified as T1a lesions and 7 were classified as T1b lesions (Table 1). The majority of the T1 lesions were superficial spreading melanomas; there were three nodular melanomas, two lentigo maligna melanomas, two unclassified lesions, and one acral lentiginous melanoma. Sixty-four lesions had an identified precursor lesion; 43 of these were histologically dysplastic nevi.
Table 1. Average Thickness and Number of Melanomas Diagnosed in the Families by T Classification, Mutation Status, and Study Period
Previous to study
Prevalent Lesions Suspicious for Melanoma
At the time of the initial examination, 22 melanomas were identified on 14 people, 6 of whom had previous melanomas (Figs. 1–4). Eighteen had radial growth phase only; 4 had radial and vertical phase growth. There was 1 thick melanoma (measuring 2.85 mm, Clark Level 4). The average thickness of all other lesions was 0.45 mm among those with CDKN2A mutations, 0.32 mm among those with CDK4 mutations, and 0.47 mm among those with unknown mutations (Table 1). All but 2 (1 of which was lentigo maligna melanoma) had an identified precursor lesion; 17 of the precursor lesions were dysplastic nevi. These four lesions are representative of prevalent lesions suspicious for melanoma. Three of these lesions were the first melanoma for each individual and, typical of these familial melanomas, occurred at an early age (35 years, 30 years, and 35 years, respectively); the fourth lesion was a severely dysplastic nevus.
Prospectively Identified Melanomas
A major part of the care for these melanoma-prone families is education regarding sun and ultraviolet protection. When queried, many of the family members report that their sun exposure patterns change after enrollment in the study.15 Family members are urged repeatedly to minimize midday exposure, use protective clothing and sunscreens, and avoid burning. Clinical recommendations for these family members include regular self-examinations and routine health care provider examinations. Family members are provided with copies of their clinical photographs for their use in self-examinations, and for the use of the clinicians who examine them. The majority of family members receive their care from local physicians, and frequency of follow-up and rates of biopsies vary among providers. The clinical recommendations for these family members include biopsy of lesions changing in a manner worrisome for melanoma or new lesions suspicious for melanoma. Some health care providers are more aggressive in removing lesions. Greater than 2000 lesions have been removed from over 300 family members and subsequently reviewed by W.H.C.
In a prospective follow-up of all the family members, 86 new melanomas were found to have occurred in 37 individuals, 16 of whom did not previously have melanoma. Seventy-two melanomas were classified as T1a36 lesions (≤ 1.0 mm and ≤ Clark Level 3, no ulceration) with an average thickness of 0.31 mm; 6 melanomas were classified as T1b (≤ 1.0 mm and > Clark Level 3, no ulceration) with an average thickness of 0.62 mm; 5 melanomas were classified as T2a with an average thickness of 1.39 mm; 1 melanoma was classified as T3a, 2.10mm thick; and 2 melanomas were of unknown thickness (Table 1). The lesions that were at least T2a occurred in individuals who were not compliant with the care guidelines. Among the 16 individuals without previous melanoma, no one who followed the guidelines has developed metastatic disease to date. The average thickness among families with CDKN2A mutations was 0.44 mm, that among families with CDK4 mutations was 0.27 mm, and the average thickness among those with no known mutations was 0.38 mm. Sixty-three melanomas had a radial growth phase only, 18 had a radial and a vertical growth phase, and 5 had a vertical growth phase only. Fifty-one were found to have identified precursor lesions on histologic review, 32 of which clearly were dysplastic nevi. The distribution of precursor lesions did not vary by mutation status. In addition, new lesions that were highly suspicious for melanoma occurred in areas without previous known nevi which had no identifiable precursor on histologic review.
New Lesions Suspicious for Melanoma
The two lesions shown in Figures 5 and 6 were new lesions that were highly suspicious for melanoma and occurred in areas previously examined without identified nevi.
Changing Lesions Suspicious for Melanoma
The lesions presented herein are typical of nevi changing in a manner worrisome for melanoma; the lesions were biopsied soon after the photographs were taken (Figs. 7–10).
New Nevi to Dysplastic Nevi
When they first appear, nevi that eventually become dysplastic often are indistinguishable from nevi that will become ordinary compound or dermal nevi (Figs. 11–13). When they reach a dimension of approximately 3 mm, the first abnormality in the morphology frequently is an irregular outline or indistinct borders. These lesions were first photographed when they appeared as new nevi in comparison with previous overview photographs. Over time, they developed the clinical characteristics of dysplastic nevi. Evaluating patterns of new nevus development over time was difficult in these families because they altered their sun exposure patterns. In general, although family members may develop new clinical dysplastic nevi at any age, it certainly is more common in the second, third, and fourth decades of life.
Minimally Changed Dysplastic Nevi
Removal of all nevi or all clinically dysplastic nevi in these family members was not recommended because many individuals were found to have several hundred nevi (some exceeding 500). The vast majority of clinically dysplastic or other nevi, similar to other precursor states, either remain relatively stable over time or regress. The clinically dysplastic nevi shown in Figures 14–16 changed minimally over many years, despite their clearly abnormal morphology.
Regressing Dysplastic Nevi
The nevi shown in Figures 17–19 regressed. Often, when a clinically dysplastic nevus begins to differentiate, a central papule appears, and the flat surrounding area becomes less pigmented and less prominent over time. In contrast to lesions suspicious for melanoma, the skin markings remain normal, and there is no visual evidence of stretched or thinned epidermis or telangiectasia in the papule. The papule is soft, similar to a compound or dermal nevus. When the lesion becomes less pigmented, it becomes the color of the baseline skin, similar to a dermal nevus. It does not lose pigment like a regressing melanoma or halo nevus. When the papule flattens, it usually does so proportionally and gradually. An asymmetric involution or change should prompt biopsy. The majority of dysplastic nevi differentiate to dermal nevi or regress and completely disappear over time. Although regression of lesions can occur at any age, it is more common in individuals age > 50 years. In these families, regression also appears to be related to sun protection over time. In contrast to a regressed melanoma (Fig. 7) or halo nevus, the site of a regressed dysplastic nevus, like a regressed dermal nevus, appears similar to the surrounding skin.
The original atlas of dysplastic nevi was, of necessity, cross-sectional,4 and could not provide sequential documentation of the natural history of these important risk markers and potential precursor lesions. The current atlas provides photographic documentation of representative lesions followed for up to 24 years. Clinically, many of these lesions are relatively stable or regress over time. The photomicrographs correlate well with the clinical photographs. These photographs and clinical notes also demonstrate that the history of change, as well as the individual morphology (including the skin markings) of the lesion, are critical in the decision to perform excisional biopsy. As shown in Figures 1–4, even small lesions can be recognized as early melanoma. Similar to other initial melanomas in these high-risk families, the lesions in Figures 1, 3, and 4 occurred in relatively young individuals. Other figures demonstrated the evolution of changes worrisome for melanoma that should prompt biopsy (Figs. 8–10). Lesions that are changing in a manner suggestive of melanoma out of proportion to other nevi are particularly worrisome. However, some lesions, such as that shown in Figure 18, stabilize and then regress. Melanocytic lesions vary in activity over time. One trigger for increased activity appears to be sun exposure;15 after acute sun exposure, multiple nevi may change and new nevi develop.
Although the majority of melanomas in these families arise from nevi, particularly dysplastic nevi, some melanomas arise de novo in previously normal skin (Figs. 5 and 6). These de novo melanomas usually occur in individuals who have dysplastic nevi. The progression is consistent with other tumor progression systems such as colon carcinoma, which may arise from villous adenomas, adenomatous polyps, or adjacent clinically normal colonic epithelium. However, the majority of nevi, including clinically dysplastic nevi, evolve to a relatively stable state or regress (Figs. 11–19); only a very small percentage progress to melanoma. Because progression to melanoma is unpredictable and rare, removal of all nevi is not warranted. Even if all nevi were removed, the frequency of clinical follow-up in these individuals would not change because they still would develop new nevi or could develop de novo melanoma. Individuals still would have to check their own skin frequently and undergo routine healthcare worker examinations. Clinically dysplastic nevi are not only potential direct precursor lesions, but also phenotypic markers of individuals with skin that is at particularly increased risk of both familial and sporadic melanoma.6–8, 15, 20, 23–33
This information is descriptive, and inferences therefore are limited. To quantify the rate of change in nevi for a sufficient number of individuals to achieve statistical significance over a 25-year period obviously would be prohibitive. Despite the inevitable limitations of the information, clinical characterizations of different phases of the natural history of melanocytic lesions that are useful for health care providers can be demonstrated.
Clinical overview photographs are very helpful in identifying either new lesions in previously uninvolved skin or changing lesions, such those seen in several of the figures in the current atlas. In addition, close-up photographs are useful for identifying subtle variations over time, which may indicate that a particular nevus is changing in a manner worrisome for melanoma.4, 30, 31, 38 Prevention and early diagnosis are the keys to reducing the mortality from melanoma. As shown in these figures, changing nevi can be removed before they become melanomas, or melanomas can be recognized while they still are very early lesions (e.g., in situ or invasive radial growth phase melanoma without apparent capacity for metastasis). Unfortunately, many facilities are unable to provide the type of photographic documentation that is an important part of the clinical care for these families, and photography often is not covered by medical insurance. However, for members of high-risk families, clinical photography is an essential diagnostic and screening tool and should be available routinely.
The etiology of familial melanoma is complex. Important risk factors include not only mutations in CDKN2A, CDK4,1, 17–21 and most likely other genes, but also dysplastic nevi23 and sun exposure.39 Although it is difficult to alter mutation status or an individual's propensity to develop dysplastic nevi, sun exposure is the major environmental risk factor for melanoma40 and is potentially modifiable. Sun exposure does appear to impact on the progression of dysplastic nevi to melanoma and melanoma development. There is some evidence that minimizing sun exposure at any age, even in high-risk families, decreases the risk of melanoma.15
The clinical and histologic morphologies of dysplastic nevi and melanomas occurring in these families do not appear to vary by the melanoma susceptibility gene that is altered in the family. This is consistent with the finding of independent effects of dysplastic nevi and CDKN2A mutations, both functional and nonfunctional, in the risk of melanoma.23, 41 The prospective risks of developing melanoma also do not appear to vary by mutation status in families from similar geographic areas.1, 37 In families without CDKN2A mutations, dysplastic nevi also are associated with an increased risk of melanoma.23 It is impossible to predict mutation status on clinical examination, but the presence of dysplastic nevi is an important predictor of risk, regardless of mutation status. Nearly all melanomas in these American families occur in individuals with dysplastic nevi, similar to other melanoma-prone families in the mid-Atlantic region of the U.S.,42 in Sweden,41 and perhaps in the Netherlands.43 A somewhat lower percentage of individuals with familial melanomas have been found to have dysplastic nevi in northeast Italy44 and Australia.45
The role of genetic testing in melanoma-prone families remains controversial. Members of an international melanoma consortium recently published guidelines for considering genetic testing in melanoma-prone families20 and recommended that, although genetic testing is an important research tool for understanding the biology and the etiology of melanoma, it is not as useful for clinical decision-making. At the current time, knowledge of the mutation status of individuals does not change the clinical management of high-risk family members. As we have demonstrated in these photographs, there are no clinical or histologic differences in the nevi or melanomas that occur in these families regardless of whether mutations are found in CDKN2A, CDK4, or in neither. Even if an individual is found to have dysplastic nevi without a mutation in a family having a known mutation, that individual is at increased risk of melanoma and should follow clinical guidelines for skin care. The risk of melanoma in an individual without dysplastic nevi, but who has a mutation in CDKN2A or CDK4, is not well quantified. If these individuals follow clinical guidelines for members of high-risk families (i.e., minimizing sun exposure, intermittent self-examination, and at least annual health care worker examination), it is likely that any suspicious lesions they may have also will be detected early.
The authors thank the study participants for their continuing generosity (without which this research could not continue) and Mr. Gary Best of the Medical Arts and Photography Branch of the National Institutes of Health and Ms. Barbara Rogers for their assistance.