High Risk of Malignant Melanoma in Melanoma-Prone Families with Dysplastic Nevi

Transcription

1 High Risk of Malignant Melanoma in Melanoma-Prone Families with Dysplastic Nevi MARK H. GREENE, M.D.; WALLACE H. CLARK, Jr., M.D.; MARGARET A. TUCKER, M.D.; KENNETH H. KRAEMER, M.D.; DAVID E. ELDER, M.B., Ch.B.; and MARY C. FRASER, R.N., M.A.; Bethesda, Maryland; and Philadelphia, Pennsylvania The risk of hereditary cutaneous malignant melanoma was evaluated in 41 members of 14 families with an autosomal dominant form of melanoma. We documented 127 primary melanomas in 69 family members, including 39 new melanomas diagnosed in 22 study participants from the time of first examination through a maximum of 8 years of follow-up. The 39 newly diagnosed melanomas occurred only in family members with dysplastic nevi, a known precursor of familial melanoma. Of 77 patients with dysplastic nevus syndrome without prior melanomas, 4 developed their first melanoma during prospective followup, as compared with.3 cases expected. The prospective age-adjusted incidence for melanoma was 14.3/1 patients with dysplastic nevus per year, with a cumulative melanoma risk (±SE) of 7.2% (±3.6) at 8 years. The actuarial probability of melanoma developing in family members with dysplastic nevi was 56.% (±1.1) from age 2 to age 59. This study confirms that dysplastic nevi are clinical markers of high risk for, and precursors of, hereditary melanoma. FAMILIAL CUTANEOUS MALIGNANT MELANOMA was first recognized as a distinctive entity in 182 (1), but modern interest was stimulated by Cawley's seminal report in 1952 (2). Subsequently, many additional melanoma-prone kindreds were reported (3) but the basis of From the Family Studies Section, Environmental Epidemiology Branch, and the Laboratory of Molecular Carcinogenesis, Division of Cancer Etiology, National Cancer Institute, Bethesda, Maryland; and Pigmented Lesion Study Group, Departments of Dermatology and Pathology and the Cancer Center, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania. familial susceptibility remained obscure. In 1978, we reported the clinical (4) and histologic (5) features of a new melanocytic precursor to melanoma in seven highrisk families, a disorder we designated provisionally as the B-K mole syndrome (after families B and K, the first two kindreds studied). These morphologically atypical moles (nevi) were characterized by irregularities in outline and pigmentation; aberrant differentiation; a tendency to be larger and more numerous than ordinary moles; and a site distribution that (unlike common acquired nevi) frequently included sun-shielded skin such as the scalp and bathing trunk areas (4-13). With the recognition that nuclear atypia and disorderly growth patterns (dysplasia) of melanocytes were the histologic hallmarks of these nevi, we proposed that these nevi be designated "dysplastic nevi," and that affected persons be classified as having the dysplastic nevus syndrome (14). A recent National Institutes of Health Consensus Development Conference has concluded that dysplastic nevi are specifically recognizable and constitute an important melanoma precursor, especially in the familial context (15). Genetic analysis of our families has shown that the melanoma/dysplastic nevus trait segregates as an autosomal dominant disorder and that the susceptibility gene may be linked to the Rh locus, located on the short arm of chromosome 1 (16). We therefore believe it appropriate to designate this familial disorder as hereditary. In- 458 Annals of Internal Medicine. 1985;12:

2 vitro studies have documented that fibroblasts and lymphocytes from some patients with hereditary melanoma/ dysplastic nevus are both hypersensitive (17-19) and hypermutable (2-21) to ultraviolet radiation and certain chemical carcinogens, suggesting that environmental factors may contribute to the risk of melanoma in genetically susceptible persons. Various immunologic abnormalities of possible etiologic significance have been described as well (22-24). The clinical implications of these observations are significant. They suggest that persons at high risk of familial melanoma can be identified prospectively, providing an opportunity for both primary prevention and earlier detection of this potentially lethal neoplasm. We present data derived from our ongoing studies of melanomaprone families that document the magnitude of risk of hereditary melanoma and show that early diagnosis of surgically curable melanomas in family members with dysplastic nevi is feasible. Methods CLINICAL SURVEY Families with at least two cases of cutaneous melanoma were identified among all kindreds in the Cancer-Prone Family Registry maintained by the National Cancer Institute (25). Fourteen families in which at least two currently living members had had documented cases of melanoma were selected for study: The seven families originally reported (4) were re-evaluated, and seven new families were added to the series. Medical histories and causes of death of family members were documented by interviews, questionnaire, and review of hospital, physician, and vital records, as previously described (16, 25). All biopsy samples of pigmented lesions obtained elsewhere before the study or during the project were reviewed by two of the authors. To date, 114 pigmented lesions have been reviewed, including 116 primary melanomas and 412 dysplastic nevi. After giving informed consent, participating family members had a complete skin examination (including Wood's lamp evaluation), detailed clinical photography, and venipuncture or skin biopsy (or both) to obtain biologic specimens for special laboratory studies. All nevi suspicious for melanoma were excised, and, when possible, at least two clinically defined dysplastic nevi were excised for histologic diagnosis from patients clinically thought to have dysplastic nevi. Clinical and histologic criteria for dysplastic nevi have been previously described (3-9). Persons having nevus biopsy were classified as having dysplastic nevi only if at least one biopsy sample was diagnostic. Prepubertal children had biopsies taken only if they had a pigmented lesion that was suspicious for melanoma. Because our observations indicated that the usual clinical pattern of the dysplastic nevi did not appear until the onset of adolescence (26), most prepubertal children were classified as currently "indeterminate" with regard to dysplastic nevi. A final decision on each person's status was made by the study team after a review of all clinical photographs, in conjunction with a review of examination records and pathology reports. We attempted to examine all first-degree relatives of each patient with melanoma in our families. When possible, more distant relatives were studied as well. Only three first-degree relatives declined participation. Each family's clinic visit concluded with a counseling session in which we informed study participants of our findings and outlined for them a health care plan designed to minimize morbidity from melanoma (3, 7, 27, 28). RISK ASSESSMENT Person-years of observation for family members with and without dysplastic nevi were accumulated as follows. The first melanoma occurring in each of the first two patients with melanoma (the "index cases") in each family was deleted to minimize the ascertainment bias resulting from our requirement that each kindred have two cases of melanoma for entry into the study. Thus, 28 melanomas were excluded. To evaluate the risk of melanoma before the study, observation commenced in each family on the day after diagnosis of its second case of melanoma (that is, the point at which it became a melanoma-prone family) and continued for each person until" the date of first melanoma diagnosis, death, or first study examination, whichever came first; this part of the study was designated the "retrospective incidence series." To evaluate the risk of melanoma after the initiation of the study, observation began the day after the first study examination and continued to the date of first melanoma diagnosis, death, or last follow-up (December 1983); this portion was designated the "prospective incidence series." Information collected at the first study examination composed the "prevalence series." Evaluation of family members in each of these three time periods was further subdivided to permit separate assessment of persons with and without dysplastic nevi who either did or did not have a melanoma before the interval under consideration. Index cases who developed a subsequent melanoma were included in the analysis of risk of melanoma among persons with a prior melanoma. In all analyses, persons developing more than one primary melanoma during a specified interval were counted only once and were removed from observation at the time of their first melanoma diagnosis. Because no melanomas occurred in either the 91 spouses or the 43 family members whose dysplastic nevus status was indeterminate, these subgroups were excluded from the analysis. Expected numbers of melanomas were computed by applying sex-specific, 5-year age-specific, and 5-year time-specific incidence rates from the Connecticut Tumor Registry (29) to the accumulated person-years of observation, using the computer program of Monson (3). Observed numbers of melanomas were compared to expected numbers, assuming a Poisson distribution. One measure of risk was the ratio of observed to expected cases. Ninety-five percent confidence intervals were computed (31), and observed-to-expected ratios were considered significantly elevated if the lower bound of the confidence interval was greater than 1.. Given the exceedingly small expected values resulting from these computations, the data are also presented as melanoma incidence rates (the number of persons with melanoma during the specified interval per person-years of observation during the specified interval). The incidence rates were age adjusted by the direct method, using the age distribution of the entire study cohort as the reference population. The cumulative risk of melanoma as a function of age was assessed for various subgroups of family members using the method of Kaplan and Meier (32). Technically, this analysis requires prospective data on dysplastic nevus status of a type not available to us. Instead, we computed the risk of melanoma by age for all bloodline family members from birth and for family members with or without dysplastic nevi from age 2. Age 2 was chosen because our clinical data indicate that family members destined to have dysplastic nevi will express the trait by this age (26). Differences in proportions were tested by the Z statistic (33). Results The study cohort comprised 41 members of 14 white families having at least 2 living persons with melanoma and included 264 adults, 46 children under age 2, and 91 spouses. We studied 213 females and 188 males ranging in age from 1 to 91 years. We examined 38 family members and reviewed hospital and pathology data from 21 patients with melanoma who were deceased when the study began. All patients were examined initially between April 1976 and April 198, for an average follow-up of 6 years by December The ethnic backgrounds report- Greene eta/. Melanoma and Dysplastic Nevi 459

3 Table 1. Characteristics of Patients with Hereditary Cutaneous Malignant Melanomas and Those of an Unselected Population with Melanomas Current Series* Comparison n(%) Sex % Male 41(59) 5 Female 28(41) 5 Patients with > 1 melanoma 21(3) Histologic type 3t Superficial spreading 12(88) 62 Nodular 8 (7) 22 Lentigo maligna 1 (1) 16 Acral lentiginous 1 (1) 1 Unclassifiable 4 (3) Unknown 11 Melanoma site Head and neck 18(15) 19* Chest 11 (9) 5 Abdomen 6 (5) 2 Back 34(28) 24 Arm 3(24) 21 Leg 24(2) 29 Unknown 4 Clark level I 9 (8) iot II 65(56) 29 III 33(28) 3 IV 9 (8) 26 V () 6 Unknown 11 Thickness <.76 mm 72(65) 38$ mm 22(2) mm 14(13) 28 > 5. mm 3 (3) 7 Unknown 16 Histologic evidence of nevus in melanoma Nevus of any kind 94(85) 5 Dysplastic nevus 78(7) 361 * Includes the 28 index cases. Median age of patients in the current series was 32. years, compared with 51.5 years in the unselected group (unpublished data from the SEER program of the National Cancer Institute). f Data from Greene and Fraumeni (3). % Data from Beardmore (35). Estimated. Data from Elder and colleagues (7). ed by the probands revealed predominantly northern European origin: 8 were of German descent; 6, Irish; 2, Scottish; and 1 each, Norwegian, English, French, Italian, Spanish, and Russian. Three probands reported having native American ancestors. (This total exceeds 14 because each proband could report multiple ethnicities.) The study participants were distributed across the United States; most resided in the northern United States, despite higher melanoma rates in the south (34). All 14 families expressed the dysplastic nevus trait. Among bloodline family members, 121 had documented dysplastic nevi while 125 were considered definitely free of dysplastic nevi. This latter group included 3 patients with melanoma, all of whom were index cases. Four spouses with dysplastic nevi were observed, 3 with single and 1 with multiple dysplastic nevi. Dysplastic nevi were diagnosed in 74 (31%) of 241 blood relatives of patients with melanoma; 43 (18%) of these relatives were considered to have an indeterminate dysplastic nevus status because of their age. Among family members, 69 persons have had at least one melanoma (including the 28 index cases): Of 51 patients with melanoma who were evaluated by examination or review of pathology reports, 48 had dysplastic nevi. The dysplastic nevus status of 18 patients with melanoma who had died before the study was unknown. CLINICAL ASPECTS OF HEREDITARY MELANOMA A total of 127 primary melanomas were diagnosed in 69 persons from the 14 kindreds, including 45 melanomas in the 28 index cases (Table 1). Of these, 122 (96%) were verified by pathologic review of either the primary lesion (116 cases) or metastatic melanoma tissue (6 cases); the remaining 5 cases were verified from medical records. Selected characteristics of the melanomas in our series are summarized and compared in Table 1 with those of an unselected population with melanoma from the literature. Our patients included 41 males and 28 females. The median age at first melanoma diagnosis was 35 years for males, 29 for females, and 32 overall (range, 12 to 76). The difference in age between males and females was not statistically significant. Altogether, 21 patients (3%) developed more than 1 primary melanoma; males (14 of 41) were more likely to develop multiple melanomas than were females (7 of 28) (p<.1). The maximum number of primary melanomas seen in 1 patient was 11. The back was the commonest site of the origin in males (26 cases, 37%), whereas the calf (9 cases, 16%) and back (8 cases, 14%) predominated in females. Superficial spreading melanoma accounted for 12 (88%) of 116 histologically reviewed cases of primary melanoma; nodular melanoma was a distant second (7%). Most of the 116 tumors were thin: 73 (63%) were Clark level II or less and 71 (61%) were less than.76 mm thick; 67 of 116 cases (58%) were both level II and less than.76 mm in greatest thickness. Two of three patients with melanoma who had clinically normal skin had children with dysplastic nevi, suggesting incomplete penetrance in carriers of this autosomal dominant trait (16). Among the 111 primary melanomas in which an assessment could be made, 78 (7%) had histologic evidence of a dysplastic nevus at a margin of the melanoma; in 16 (14%), a residual nevus without dysplasia was identified (Table 1). The 3 patients with melanoma and clinically normal skin all had histologic evidence of a nevus in their primary melanoma; in 1, this was a dysplastic nevus. Thus, nevi were related to melanoma development in all 51 evaluable patients with melanoma, and dysplastic nevi played a role in 49 (96%). Among all persons in this cohort, 22 patients developed 1 or more new melanomas that were detected either at the initial examination for our survey (8 patients, 11 melanomas), during the course of follow-up (11 patients, 16 melanomas), or both on initial examination and during follow-up (3 patients; 7 primary melanomas at first examination; 5 during follow-up) (Table 2). Thus, a total of 39 "newly diagnosed" melanomas has occurred to date in 22 family members. For 11 family members, the lesion diagnosed in our study represented their first melanoma, whereas for the remaining 11 this was a subse- 46 April 1985 Annals of Internal Medicine Volume 12 Number 4

4 Table 2. Distribution of Patients with Melanoma by Time of Diagnosis* Time of Diagnosis Patients Melanomas Before Study First Examination During Follow-up Total Before study only First examination only 5 During follow-up only 4 Before study + first examination 3 3 Before study + during follow-up 7 17 First examination + during follow-up 2 Before + first examination + follow-up 1 6 Total n * Data include 45 melanomas in the 28 index cases: 28 index tumors, 11 melanomas during the retrospective incidence period, 1 melanoma at first examination, and 5 melanomas during prospective follow-up. quent melanoma, their first having been diagnosed before our project. Four family members developed their first melanoma and 1 patients developed a second or subsequent melanoma during follow-up that is, after their initial examination. All 39 newly diagnosed melanomas developed in family members with the dysplastic nevus syndrome; we detected no new melanomas during our study in family members with normal skin. All but 1 of these melanomas were of the superficial spreading histologic type. Nearly all 39 lesions were biologically early in their development: 35 were Clark level II or less and 36 were less than.76 mm thick; 36 had histologic evidence of a dysplastic nevus at a margin. Three of the four lesions that were beyond the verylow-risk microstage (36) were detected at the initial examination. The fourth developed during follow-up in a patient who had not had a mole evaluation for 2 years. In addition, nine patients with melanoma (31 lesions) and seven family members with only the dysplastic nevus syndrome (7 lesions) had excisional biopsy of suspicious pigmented nevi that were interpreted as histologically "borderline" because the cytologic atypia of melanocytes was so severe or extensive and the architectural pattern of melanocyte proliferation was so abnormal that early melanoma could not be completely ruled out. Such lesions were not classified as melanomas, because they failed to meet the diagnostic criteria (9, 37). RISK OF HEREDITARY MELANOMA The quantification of the risk of melanoma in the study participants is purest if one considers the "prospective incidence series" (see Methods), which excludes from observation all patients known to have had melanoma (either before the study or at first examination). Thus, 77 Table 3. Risk of Melanoma in Fourteen Melanoma-Prone Families by Period and Dysplastic Nevus Status Period* Dysplastic Nevi Present Dysplastic Nevi Absent Prospective incidence Patients, n Observed melanomas*)*, n (n) Expected melanomasj, n Observed-to-expected ratio Person-years of observation Melanoma incidence, n/1 yr Prevalence Patients, n Observed melanomas!, n (n) Melanoma prevalence, % Retrospective incidence Patients, n Observed melanomas*)*, n (n) Expected melanomas"):, n Observed-to-expected ratio Person-years of observation Melanoma incidence, n/1 yr No Prior Melanoma Prior Melanoma No Prior Melanoma (5).27 1(17) (12) 4(6) J 23 (5).53 13** 4(7) ! Prior Melanoma * See Methods for definitions. These computations exclude 11 patients with 13 melanomas whose dysplastic nevus status is unknown. t Number of persons developing melanoma (number of independent melanomas diagnosed). X Expected number of persons with melanoma, derived from Connecticut Tumor Registry (29) data (see Methods). The melanoma incidence rate as been age adjusted using the age distribution of the entire study cohort as the reference population. Of 36 patients with dysplastic nevus and melanoma before first study examination, 3 died before screening and are thus excluded. Their diagnoses of dysplastic nevus were established by review of biopsy samples of moles taken before death. II Of 121 patients with dysplastic nevi, 14 were index cases (16 melanomas) and are therefore excluded here. ** Of 14 index patients with dysplastic nevi, 1 died before the second family member developed melanoma and is therefore excluded here Greene et al. Melanoma and Dysplastic Nevi 461

5 Figure 1. Cumulative actuarial probability (± SE) of developing cutaneous malignant melanoma as a function of age. Numbers in parentheses indicate persons under observation at specified ages. The lower curve shows the melanoma risk in all 282 bloodline family members (41 - [91 spouses + 28 index cases ]) from birth through age 91. The upper curve depicts the risk of melanoma in 93 family members with dysplastic nevi (DNS) from age 2 through 76. The 28 index cases are excluded from both computations. The dysplastic nevus computation excludes all family members with dysplastic nevi who were less than 2 years of age at first examination, including 7 persons with melanoma. family members with dysplastic nevi and without prior melanoma accrued person-years of observation (Table 3). Four patients developed their first melanoma compared with.27 cases expected (observed-to-expected ratio, 148; 95% confidence interval, 4 to 379). The age-adjusted incidence of melanoma was 14.3 cases/1 persons with dysplastic nevi per year, and the cumulative risk (±SE) of first melanoma was 7.2% (±3.6) at 8 years. The risk was similar in males and females. Furthermore, among 4 family members with dysplastic nevi and a prior melanoma (either before or at first examination), 1 developed a new melanoma during person-years of prospective follow-up, for an incidence of 55.3 cases/1-yr, and a cumulative risk of subsequent melanoma of 33.% (±8.8) at 8 years. In contrast, 122 family members without dysplastic nevi accrued person-years of observation; no melanomas developed in this group (.44 cases expected). During the retrospective incidence period (see Methods), 23 of 17 persons with dysplastic nevi without prior melanoma developed melanoma while accruing person-years of observation, for a melanoma incidence of 14.8 cases/ 1-yr. Once again, no melanomas were seen in family members without dysplastic nevi during this interval. At the first study examination, 84 persons with dysplastic nevus syndrome had never developed melanoma and 33 had had at least one prior melanoma. When examined initially, 7 (8.3%) persons from the first group and 4 (12.1%) from the latter group were found to have at least one melanoma. Thus, the overall prevalence of melanoma at first examination was 9.4% in persons with dysplastic nevus syndrome and % in their clinically normal blood relatives. Considering the period from birth to age 91 in all 282 bloodline family members, the Kaplan-Meier (32) estimate of melanoma risk was 46% (±1.9) at age 76 (Figure 1). This estimate is free of retrospective classification bias and provides a minimum estimate of melanoma risk in melanoma-prone families as we have defined them. However, this curve is clearly a composite of the very large risk seen in family members with dysplastic nevi and the very low risk in family members without dysplastic nevi. The best approximation of the melanoma risk in persons with dysplastic nevus syndrome is that from age 2 to age 59, for whom the cumulative risk is 56% (±1.1) (Figure 1). Although this curve reaches 1% at age 76, the very small number of older persons under observation and the fact that the oldest patient with dysplastic nevi developed melanoma make this an imprecise estimate. Three family members without dysplastic nevi developed melanoma, but all 3 were index cases and thus excluded from this analysis. Therefore, the formal cumulative risk of melanoma in family members without dysplastic nevi was zero. Whereas the rates of melanoma in persons with dysplastic nevi and no prior melanoma were similar in the retrospective and prospective portions of the analysis (14.8 and 14.3/1, respectively), the prognostic features of the melanomas changed substantially. Thus, all first melanomas detected during the prospective followup were of the superficial spreading type, compared with 71% of index cases, 73% of retrospective cases, and 86% of prevalence cases (Table 4). Similarly, tumor depth, as measured either by level or thickness, decreased; for example, mean tumor thickness among index cases was 1.86 mm, compared with.76 mm in prospectively diagnosed melanomas. The proportion of melanomas with an identifiable dysplastic nevus at a margin of the tumor increased, while few showed evidence of ulceration or lesion regression (Table 4). Table 4. Prognostic Characteristics of First Melanomas by Period of Diagnosis* Prognostic Characteristic Index Retrospective Prevalence Prospective Cases Incidence Cases Incidence Melanomas, n Superficial spreading melanoma, % Clark level II or less, % Less than.76 mm thick, % Mean thickness, mm Dysplastic nevus present, % Regression present, % Ulceration present, % 27t * Data based on the first melanoma developing in study participants with no prior melanoma in each of the four intervals noted (see Methods for definitions of intervals). Histologic data refer to primary tumor. t Tissue from the first primary melanoma in 1 index case was not available for review. 462 April 1985 Annals of Internal Medicine Volume 12 Number 4

6 Discussion During an 8-year study of 41 members of 14 families genetically predisposed to melanoma, 22 family members developed 39 new primary cutaneous melanomas. In general, patients with hereditary melanoma were considerably younger than unselected patients with melanoma, had a striking tendency towards multiple primary lesions, and were more likely to develop thin primary lesions ( <.76 mm thick) of the superficial spreading type with an associated nevus. Newly diagnosed melanoma occurred only in persons with dysplastic nevi, and 92% of these tumors developed in histologic contiguity with a dysplastic nevus. No family member with normal skin has developed a new melanoma during the prospective phase of our study. Thus, dysplastic nevi are markers that identify the specific persons at high risk of melanoma and are the actual precursors of these melanomas. Family members with dysplastic nevi and those who have already developed at least one melanoma represent subgroups on whom efforts at melanoma prevention and early detection should be focused. The cancer susceptibility in these families was type-specific; no excess of cancers other than melanoma was observed (GREENE MH. Unpublished observations.). Family members with dysplastic nevi had a dramatic excess risk of melanoma, several hundred times that of the general population, with a melanoma rate of 1.4% to 5.4% per year depending on the presence or absence of prior melanoma. Our data show that 8% to 12% of melanoma-prone family members with dysplastic nevi will have a new melanoma detected at their first screening examination, depending on the proportion of patients with prior melanoma. In addition, the actuarial risk of developing a new melanoma during 8 years of follow-up in persons with dysplastic nevi with and without prior melanoma was 33.% and 7.2% respectively. The actuarial probability of persons with dysplastic nevus syndrome developing melanoma was 56% at age 59. That the cumulative lifetime probability of melanoma reached 46% when all bloodline family members were considered provides further indirect evidence of an autosomal dominant mode of inheritance for melanoma (a dominant genetic model with complete penetrance would predict a 5% lifetime risk) (16). However, our data indicate clearly that this 46% risk is not shared equally by all family members. Although we lack the prospective data required to address this issue properly, the risk of melanoma clearly is exceedingly high in family members with dysplastic nevi and quite low in members without dysplastic nevi. A more accurate assessment of the cumulative lifetime risk may not be possible in the current study, because our policy of excising the morphologically most atypical nevi has probably altered the natural history of melanoma in our study population. Clinically, the appearance of new black pigmentation in a dysplastic nevus was the best predictor of the presence of early melanoma (a detailed clinical description of dysplastic nevi is provided elsewhere [38]). Notably, 38 "borderline" melanocytic lesions were removed from family members with either melanoma or dysplastic nevi. Although not quite meeting the criteria for the diagnosis of malignant melanoma (9, 37), these lesions were abnormal histologically; some, in fact, were interpreted as being malignant melanoma by the referring pathologist. As Foulds (39, 4) has emphasized, preneoplastic lesions are not obligated to progress to frank malignancy; such lesions may remain stable, regress, or progress. It seems certain, however, that changing dysplastic nevi, particularly those interpreted as histologically borderline, constitute a pool of melanocytic lesions from which many melanomas are destined to arise. Their removal, therefore, represents primary prevention of melanoma. The detection and excision of histologically borderline lesions in seven persons with dysplastic nevi without prior melanoma provides an estimate of this effect and a measure of the extent to which the cumulative risk computation underestimates the true probability of melanoma in these persons. A detailed knowledge of the clinical and histologic characteristics of a melanoma in a given patient permits reliable assessment of the melanoma's probability of metastasizing (37). Thus, a melanoma that is less than.76 mm thick (41); that lacks lesional regression (42), the pattern of tumor growth known as vertical growth phase (43), and ulceration (44); and that has a demonstrable nevus in association with the tumor (45) almost never metastasizes. With rare exceptions, the complete surgical excision of such a melanoma is curative. Nearly all of the 39 newly diagnosed melanomas in our series displayed these characteristics. In particular, these newly diagnosed melanomas were significantly thinner (biologically earlier in their development): 92% of melanomas diagnosed at our first examination or during follow-up were less than.76 mm thick compared with 5% (36 of 72) of the melanomas that developed before the study began. Because the prognostic features of the melanomas steadily improved as diagnosis progressed from the index period up through the prospective phase of the survey, it is reasonable to expect that these patients have a low risk of recurrent melanoma. The progressive decrease in tumor thickness of second (78% were less than.76 mm thick) and subsequent (88%, less than.76 mm) melanomas, compared to the thickness of first melanomas (49%, less than.76 mm), documents the feasibility of detecting melanoma early in its natural history. None of the 4 patients who developed their first melanoma during followup has died, and a preliminary assessment suggests a lower mortality rate among patients whose first melanoma was diagnosed at or after the first examination (2 of 11; 18%) than among patients whose first melanoma was diagnosed before the study (28 of 58; 45%). These numbers are small, however, and a proper analysis of survival among patients with hereditary melanoma is beyond the scope of this presentation. We caution that although all newly diagnosed melanomas occurred in family members with dysplastic nevi, three index cases whose melanomas developed before the study had clinically normal skin at the time of our examination. Furthermore, we observed dysplastic nevi in eight children of family members with neither melanoma nor Greene eta/. Melanoma and Dysplastic Nevi 463

7 dysplastic nevi. Some of this apparent dissociation between dysplastic nevi and melanoma can be attributed to incomplete penetrance of the disease gene that is, a person may occasionally carry the gene but fail to express the dysplastic nevus trait. Finding dysplastic nevi in the children of two patients with melanoma who themselves lacked the dysplastic nevi at the time of their examination is compatible with this notion. In addition, whereas persons with dysplastic nevus syndrome typically have multiple precursor nevi, some have very few such lesions. One might then expect some family members to have but one dysplastic nevus; if that lesion should become malignant, there would be no remaining clinical evidence of the dysplastic nevus syndrome. In support of this possibility, one study participant had clinically normal skin on examination, yet she also had several children with dysplastic nevi. Histologic review of a mole excised 28 years before our survey documented the presence of a single dysplastic nevus. In addition, because she had normal skin when examined at age 54, some patients may retain the ability to eliminate, eventually, the aberrant differentiation in their nevi. We have clinical observations to support this notion, and it may not be coincidental that the three patients in our series with melanoma but lacking dysplastic nevi had an average age of 59 while the mean age of the patients with melanoma and dysplastic nevus syndrome was 34. Dysplastic nevi have been recently noted to occur in four clinical settings: patients with or without melanoma who either do or do not have a family history of dysplastic nevi or melanoma (46, 47). We have estimated that 32 people in the United States have familial dysplastic nevi with melanoma as described in this article (47); they may account for 5.5% of all cases of cutaneous melanoma in this country (47). Furthermore, dysplastic nevi have been described both clinically and histologically in 25% to 35% of unselected patients with melanoma (7, 9, 48, 49) and in 2% to 5% of the general population (48, 5, 51). In the current study, 4 of 91 spouses (4%) had biopsy-proven dysplastic nevi. Thus, precursor lesions originally recognized in the special setting of melanomaprone families have a much broader distribution than originally appreciated and seem likely to be associated with a substantial proportion of all melanomas. Although finding such lesions in persons who lack a family history of melanoma may permit identification of persons at higher risk of melanoma than those who have no dysplastic nevi, the magnitude of melanoma risk in these persons has not been determined. The melanoma risk in persons with nonfamilial dysplastic nevi may prove substantially different from that seen in the families reported here. Because 4.6 million persons in the United States are estimated to have at least one dysplastic nevus (47), this is a research question of utmost importance. Despite the great potential for melanoma prevention and early detection in high-risk families, physicians must exercise caution to avoid creating needless anxiety in persons with nonfamilial dysplastic nevi, pending more definitive estimates of their melanoma risk. Nonetheless, careful management of persons with dysplastic nevi (3, 7, 27, 28), especially in the familial setting, offers real promise of reducing the incidence and mortality rates for cutaneous melanoma (15). Addendum Since submission of this manuscript, four family members, all with dysplastic nevi, have developed five additional primary melanomas. One patient who developed her first melanoma is the fifth prospectively monitored patient with dysplastic nevus to develop melanoma in our study. The remaining three patients each had had a prior melanoma. This brings the total of newly diagnosed melanomas in this group of patients to 44. Four of these melanomas were less than.76 mm in maximal thickness; the fifth, 1.46 mm thick, occurred in a family member who has refused follow-up. All 5 melanomas had histologic evidence of a dysplastic nevus at a margin of the tumor. ACKNOWLEDGMENTS: The authors thank the members of our study families, whose patience and cooperation made this project possible; Richard K., the proband (now deceased) of the first family we studied, whose conviction that his family had "funny moles" led to the recognition of the B-K mole syndrome; Dr. Michael Mastrangelo, for referring Family K. to us for evaluation; Susan B., the proband of the second family, whose cooperation contributed enormously to our early histologic studies of dysplastic nevi; Drs. Robert N. Hoover, Joan Aron, and Joseph F. Fraumeni, Jr., for critical review of this manuscript; Dr. Arthur Dresdale, who performed many of the required biopsies; Deborah McGuire, our research nurse during the early stages of the project; Dr. Ann Dudgeon, who provided critical data processing support; and Gail Sheridan and Veronica McEneaney, for technical assistance. Grant support: in part by research grants CA 1652, CA 25298, and CA from the National Cancer Institute, and by contract N1-CP-154 with the National Institutes of Health. Requests for reprints should be addressed to Mark H. Greene, M.D.; Environmental Epidemiology Branch, National Cancer Institute, Landow 3C-29; Bethesda, MD 225. References 1. NORRIS W. A case of fungoid disease. Edinb Med Surg J. 182;16: CAWLEY EP. Genetic aspects of malignant melanoma. AMA Arch Dermatol 1952;65: GREENE MH, FRAUMENI JF JR. The hereditary variant of malignant melanoma. In: CLARK WH, GOLDMAN LI, MASTRANGELO MJ, eds. Human Malignant Melanoma. New York: Grune & Stratton, Inc.; 1979: REIMER RR, CLARK WH JR, GREENE MH, AINSWORTH A, FRAUME NI JF JR. Precursor lesions in familial melanoma: a new genetic preneoplastic syndrome. JAMA. 1978;239: CLARK WH JR, REIMER RR, GREENE MH, AINSWORTH A, MAS TRANGELO MJ. Origin of familial malignant melanoma from heritable melanocytic lesions: the B-K mode syndrome. Arch Dermatol. 1978;114: GREENE MH, REIMER RR, CLARK WH JR., MASTRANGELO MJ. Precursor lesions in familial melanoma. Semin Oncol. 1978;5: ELDER DE, GREENE MH, BONDI EE, CLARK WH. Acquired melanocytic nevi and melanoma the dysplastic nevus syndrome. In: ACKER- MAN AB, ed. Pathology of Malignant Melanoma. New York: Masson Publishing Co.; 1981: ELDER DE, GREENE MH, GUERRY D IV, KRAEMER KH, CLARK WH JR. The dysplastic nevus syndrome: our definition. Am J Dermatopathol. 1982;4: CLARK WH JR., ELDER DE, GUERRY D, EPSTEIN MN, GREENE MH, VAN HORN M. A study of tumor progression the precurser lesions of superficial spreading and nodular melanoma. Hum Pathol. 1984;15: LYNCH HT, FRICHOT BC III, LYNCH JF. Familial atypical multiple mole-melanoma syndrome. J Med Genet. 1978;15: BRAUN-FALCO O, LANDTHALER M, RYCKMANNS F. BK-mole syndrom. FortschrMed. 1979;97: BOVET R, BACHMANN B, DELACRETAZ J. Le BK mole syndrome est-il toujours de transmission autosomique dominante? Ann Dermatol Venereol. 198;17: MACKIE RM. Multiple melanoma and atypical melanocytic naevi-evidence of an activated and expanded melanocytic system. Br J Dermatol. 1982;17: GREENE MH, CLARK WH JR, TUCKER MA, et al. Precursor naevi in cutaneous malignant melanoma: a proposed nomenclature [Letter]. Lancet. 198;2: April 1985 Annals of Internal Medicine Volume 12 Number 4

8 15. NATIONAL INSTITUTES OF HEALTH CONSENSUS DEVELOPMENT CON FERENCE. Precursors to malignant melanoma. JAMA. 1984;251: GREENE MH, GOLDIN LR, CLARK WH JR, et al. Familial cutaneous malignant melanoma: autosomal dominant trait possibly linked to the Rh locus. Proc Natl Acad Sci USA. 1983;8: SMITH PJ, GREENE MH, DEVLIN DA, MCKEEN EA, PATERSON MC. Abnormal sensitivity to UV-radiation in cultured skin fibroblasts from patients with hereditary cutaneous malignant melanoma and dysplastic nevus syndrome. Int J Cancer. 1982;3: SMITH PJ, GREENE MH, ADAMS D, PATERSON MC. Abnormal responses to the carcinogen 4-nitroquinoline 1-oxide of cultured fibroblasts from patients with dysplastic nevus syndrome and hereditary cutaneous malignant melanoma. Carcinogenesis. 1983;4: RAMSAY RG, CHEN P, IMRAY FP, KIDSON C, LAVIN MF, HOCKEY A. Familial melanoma associated with dominant ultraviolet radiation sensitivity. Cancer Res. 1982;42: PERERA MIR, GREENE MH, KRAEMER KH. Dysplastic nevus syndrome: increased ultraviolet mutability in association with increased melanoma susceptibility [Abstract]. Clin Res. 1983;31:595A. 21. HOWELL JN, GREENE MH, CORNER RC, MAHER VM, MCCORMICK JJ. Fibroblasts from patients with hereditary cutaneous malignant melanoma are abnormally sensitive to the mutagenic effect of simulated sunlight and 4-nitroquinoline 1-oxide. Proc Natl Acad Sci USA. 1984;81: DEAN J, GREENE MH, REIMER RR, et al. Immunologic abnormalities in melanoma-prone families. JNCI. 1979;63: VANDENBARK A, GREENE MH, BURGER DR, VETTO RM, REIMER RR. Immune response to melanoma extracts in three melanoma-prone families. JNCI. 1979;63: HERSEY P, EDWARDS A, HONEYMAN M, MCCARTHY WH. Low natural-killer-cell activity in familial melanoma patients and their relatives. BrJCancer. 1979;4: BLATTNER WA. Family studies the interdisciplinary approach. In: MULVIHILL JJ, MILLER RW, FRAUMENI JF JR, eds. Genetics of Human Cancer. New York: Raven Press; 1977: TUCKER MA, GREENE MH, CLARK WH JR, KRAEMER KH, FRASER MC, ELDER DE. Dysplastic nevi on the scalp of prepubertal children from melanoma-prone families. J Pediatrics. 1983;13: FRASER MC. The role of the nurse in the prevention and early detection of malignant melanoma. Cancer Nurs. 1982;5: GREENE MH, CLARK WH JR, TUCKER MA, et al. Managing the dysplastic naevus syndrome [Letter]. Lancet. 1984;1: HESTON J, MEIGS JW, KELLY J, et al. Forty-five years of cancer in Connecticut: Natl Cancer Inst Monogr. (In press). 3. MONSON RR. Analysis of relative survival and proportional mortality. Comput Biomed Res. 1974;7: ROTHMAN KJ, BOICE JD JR. Epidemiologic Analysis with a Programmable Calculator. Bethesda, Maryland: National Institutes of Health; (DHEW publication no. (NIH) ) 32. KAPLAN EL, MEIER P. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53: REMINGTON RD, SCHORK MA. Statistics with Applications to the Biological and Health Sciences. Englewood Cliffs, New Jersey: Prentice- Hall Inc.; SCHREIBER MM, Bozzo PD, MOON TE. Malignant melanoma in southern Arizona: increasing incidence and sunlight as an etiologic factor. Arch Dermatol. 1981;117: BEARDMORE GL. The epidemiology of malignant melanoma in Australia. In: MCCARTHY WH, ed. Melanoma and Skin Cancer. Sydney: Government Printer; 1972: DAY CL JR, MIHM MC JR, LEW RA, et al. Cutaneous malignant melanoma: prognostic guidelines for physicians and patients. CA. 1982;32: ELDER DE, AINSWORTH AM, CLARK WH JR. The surgical pathology of cutaneous malignant melanoma. In: CLARK WH JR, GOLDMAN LI, MASTRANGELO MJ, eds. Human Malignant Melanoma. New York: Grune & Stratton, Inc; 1979: GREENE MH, CLARK WH JR, TUCKER MA, et al. Acquired precursors of cutaneous malignant melanoma: the familial dysplastic nevus syndrome. N Engl J Med. 1985;312: FOULDS L. Neoplastic Development. Vol. I. New York: Academic Press; FOULDS L. Neoplastic Development. Vol. II. New York: Academic Press; BRESLOW A. Thickness, cross-sectional area and depth of invasion in the prognosis of cutaneous melanoma. Ann Surg. 197;172: GROMET MA, EPSTEIN WL, BLOIS MS. The regressing thin malignant melanoma: a distinctive lesion with metastatic potential. Cancer. 1978;42: ELDER DE, GUERRY D, EPSTEIN MN, et al. Invasive melanoma lacking competence for metastasis: minimal risk melanoma. Am JDermatopathol. 1984;6? BALCH CM, WILKERSON JA, MURAD TM, SOONG S-J, INGALLS AL, MADDOX WA. The prognostic significance of ulceration of cutaneous melanoma. Cancer. 198;45: FRIEDMAN RJ, RIGEL DS, KOPF AW, et al. Favorable prognosis for malignant melanomas associated with acquired melanocytic nevi. Arch Dermatol. 1983;119: ELDER DE, GOLDMAN LI, GOLDMAN SC, GREENE MH, CLARK WH JR. Dysplastic nevus syndrome: phenotypic association of sporadic cutaneous melanoma. Cancer. 198;46: KRAEMER KH5 GREENE MH, TARONE R, ELDER DE, CLARK WH JR, GUERRY D. IV. Dysplastic naevi and cutaneous melanoma risk [Letter]. Lancet. 1983;2: RHODES AR, SOBER AJ, MIHM MC, FITZPATRICK TB. Possible risk factors for primary cutaneous melanoma [Abstract]. Clin Res. 198;28:252A. 49. RHODES AR, HARRIST TJ, DAY CL, MIHM MC JR, FITZPATRICK TB, SOBER AJ. Dysplastic melanocytic nevi in histologic association with 234 primary cutaneous melanomas. J Am Acad Dermatol. 1983;9: SCHEIBNER A, MILTON GW, MCCARTHY WH, SHAW H. Clinical features, prognosis and incidence of multiple primary malignant melanoma [Abstract]. Aust NZ J Surg. 1981;51: CRUTCHER WA, SAGEBIEL RW. Prevalence of dysplastic nevi in a community practice [Letter]. Lancet. 1984; 1:729. Greene et al. Melanoma and Dysplastic Nevi 465