Familial cutaneous malignant melanoma: Autosomal ... - Europe PMC

6 downloads 15091 Views 1MB Size Report
of CMM and DNS in families is consistent with an autosomal dominant mode of inheritance and suggest that the DNS/CMM susceptibility geneis linked to the Rh ...
Proc. Natl. Acad. Sci. USA Vol. 80, pp. 6071-6075, October 1983

Medical Sciences

Familial cutaneous malignant melanoma: Autosomal dominant trait possibly linked to the Rh locus (cancer/genetics/epidemiology/etiology)

MARK H. GREENE*, LYNN R. GOLDINt, WALLACE H. CLARK, JR.t, EVERETT LOVRIEN§, KENNETH H. KRAEMER$, MARGARET A. TUCKER*, DAVID E. ELDERt, MARY C. FRASER*, AND SHIRLEY ROWE§ *Family Studies Section, Environmental Epidemiology Branch, Landow 3C-19, National Cancer Institute, Bethesda, Maryland 20205; tSection on Psychogenetics, Biological Psychiatry Branch, National Institute of Mental Health, Bethesda, Maryland 20205; tPigmented Lesion Group and Departments of Dermatology and Pathology, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania 19104; §Linkage Laboratory, Department of Medical Genetics, University of Oregon, Portland, Oregon 97201; and ¶Laboratory of Molecular Carcinogenesis, National Cancer Institute, Bethesda, Maryland 20205

Communicated by Peter C. Nowell, June 15, 1983 Segregation and linkage analyses were underABSTRACT taken in families with multiple cases of cutaneous malignant mel-

anoma (CMM) and a recently-described melanoma precursor, the dysplastic nevus syndrome (DNS). Clinical and laboratory data, including 23 genetic markers, were collected on 401 members of 14 high-risk kindreds. Pedigree analysis was compatible with an autosomal dominant mode of inheritance for the familial CMM trait. Although a similar model probably applies to the DNS trait as well, segregation analysis could not confirm the presence of a major locus. However, linkage analysis suggested that an autosomal dominant model was appropriate for the DNS, and that a DNS/CMM susceptibility gene may be located on the short arm of chromosome 1, within 30 map units of the Rh locus [maximum logarithm of odds (lod) score = 2.00].

Writing in 1820, Sir William Norris presented a vivid description of the familial variant of cutaneous malignant melanoma (CMM) (1). He concluded, "These facts . . . would incline me to believe that this disease is hereditary." Modem investigations of familial melanoma began with the seminal work of E. P. Cawley (2), and it is now clear that familial melanoma is not uncommon (3), but to date there have been no formal genetic studies of familial CMM. The recent identification of a unique precursor to CMM in high-risk family members (4-6) has provided new research opportunities. Designated the dysplastic nevus syndrome (DNS) (7), because cytologic atypia of melanocytes is its histologic hallmark, this disorder forms the substrate from which most cases of familial CMM arise (8-15). Here we present segregation and linkage analyses of a series of 14 melanoma-prone kindreds, which document that the distribution of CMM and DNS in families is consistent with an autosomal dominant mode of inheritance and suggest that the DNS/CMM susceptibility gene is linked to the Rh locus. METHODS Clinical Survey. The Cancer-Prone Family Registry, maintained by the Family Studies Section of the Environmental Epidemiology Branch, National Cancer Institute (16), is a nonpopulation-based series of malignancy-prone kindreds referred by physicians of the National Institutes of Health, community practitioners around the United States, and concerned family members. When this study began, there were approximately 1,000 families in the Registry, from which all kindreds with at least two living, pathologically confirmed cases of CMM were The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

identified; the 25 kindreds found were listed in ascending numerical order by date of ascertainment. Families were contacted, seriatim, and offered an opportunity to participate in a combined clinical-laboratory evaluation of familial melanoma. Fourteen kindreds were studied; available data suggest that the remaining families are similar to those that were included. We examined, photographed, and obtained blood samples from all melanoma probands, from all additional living family members with melanoma, and from 97% of their first-degree relatives; in many cases, more distant relatives were studied as well. Family members were classified as affected by the DNS by using clinical and histologic criteria described in detail elsewhere (4-6, 14, 15). Patients thought to have the DNS on clinical grounds were routinely biopsied and were considered affected only if the excised nevi were histologically dysplastic. A final decision on each patient's status was made after blind review of all nevus photographs, review of examination records, and pathology reports. Four hundred and one members of these 14 families were studied, including 265 adults, 46 children (under age 20), and 90 spouses; 69 family members developed at least one melanoma, and 74 additional persons were found to have dysplastic nevi. Our histologic review of tumor tissue confirmed the CMM diagnosis in 97% of patients. Of the 48 familial CMM patients examined (the remainder had died before the study began), 44 (92%) had the DNS. Histologic confirmation of the DNS diagnosis was obtained in 90% of the 123 affected family members. Venous blood was obtained and tested for a maximum of 23 genetic markers (see footnote to Table 3) using standard techniques (17-19). Genetic Analysis. For analytic purposes, individuals were classified as "affected" in two different ways. In the first case, all individuals with either histologically confirmed DNS or CMM were considered to be affected. In a more restricted analysis, only individuals with CMM were considered affected. The Elston-Stewart likelihood method of pedigree analysis was used to test hypotheses of single-gene transmission of illness in these families (20). The genetic model assumes that there is a single two-allele locus (A,a), and therefore three genotypic groups, AA, Aa, and aa with probabilities of being affected (penetrances)fAA, fAa, and faa respectively. We defined the illness-promoting (susceptibility) allele to be A, with a population frequency qA, and assumed the frequencies of the three genotypes to be in Hardy-Weinberg proportions. A parent of genotype AA, Aa, or aa transmits allele A to his/her offspring with probability TAAA, TAaA, and Tad, respectively. An ascertainment Abbreviations: CMM, cutaneous malignant melanoma; DNS, dysplastic nevus syndrome.

6071

6072

Proc. Natl. Acad. Sci. USA 80 (1983)

Medical Sciences: Greene et al.

probability

was

incorporated. Under the Mendelian hypothe-

sis, the transmission probabilities are restricted as follows: rAAA = 1. 0, TA, = 0.5, and Ad = 0.0. Under the environmental (nongenetic) hypothesis, there is no familial transmission of the trait, and the Ts are restricted to be equal. These hypotheses were each tested against the general alternative of arbitrary transmission probabilities lying between 0 and 1, using the likelihood ratio criterion. We did not consider X-chromosome-linked hypotheses because male-to-male transmission was observed in 10 of the 14 families. Demonstration of single-locus inheritance requires both acceptance of the Mendelian hypothesis and rejection of the environmental hypothesis. Segregation analyses were performed with the computer program GENPED.I1 Logarithm of odds (lod) scores for linkage between marker traits and either CMM or CMM with DNS were calculated by the computer program LIPED (21) under various assumptions (specified in Results) with regard to the mode of inheritance of the disease traits. RESULTS only those family members with When Analysis. Segregation melanoma were considered to be affected, a dominant gene hypothesis was not rejected (x2 = 1.98, P = 0.58), a recessive gene hypothesis was rejected (x3 = 9.74, P = 0.02), and an environmental hypothesis was borderline (x = 4.48, P = 0.10) (Table 1). Therefore, a single dominant gene hypothesis is consistent with the data. The environmental hypothesis cannot be considered to be a "good" description of the data. In these families, the dominant gene has a frequency of 0.13 and a penetrance of 64% in individuals having the gene and 0% in individuals who are genotypically normal. The gene frequency is likely to be biased because of selection of families and does not reflect the true frequency in the general population. The second segregation analysis considered persons with either

CMM or DNS as affected (Table 2). Both dominant and recessive single-gene hypotheses were strongly rejected (x3 =

40.15, X3 = 45.98), as was the strictly nongenetic (i.e., environmental) hypothesis (y2 = 11.76). Thus, although the data are not consistent with the segregation of a single gene, neither are they compatible with a purely random environmental effect. The estimated transmission probability under the unrestricted model for the heferozygote individual was estimated to be 1.0 instead of 0.5, reflecting the fact that affected parents had too many affected offspring to be consistent with Mendelian segregation. In addition, the transmission of the mutant gene from the normal homozygote was much greater (0.419) than its expected value of 0, implying that many genotypically normal individuals are transmitting the trait. Linkage Analysis. Linkage analysis was done in two ways. Even though the DNS did not fit a single gene model, we tested for linkage of the markers to the DNS/CMM syndrome on the possibility that there really is a single gene and that, for some unknown reason, our data were deviant. We also tested for linkage in the more restricted case of melanoma itself because here a single-gene model is likely. In both cases, we assumed that the trait resulted from a rare dominant gene with the penetrance values as estimated by segregation analysis. Lod scores for linkage between the DNS/CMM syndrome and 23 autosomal markers are shown in Table 3. The highest lod score is 1.56 at the Rh locus, with a 30% recombination fraction. Fig. 1 shows lod scores under a range of reasonable gene frequencies and penetrances. If we assume penetrance to be complete and the gene to be rare (q = 0.001), then the max1I GENPED (a general pedigree analysis package) was designed in 1975 by E. B. Kaplan and R. C. Elston.

Table 1. Maximum likelihood estimates from segregation analysis of melanoma in 14 kindreds Hypothesis EnvironMendelian Dominant Recessive mental Unrestricted Parameter Transmission

probabilities* TAAA

TAaA TaaA

Gene frequency (qA) Penetrances

tAA fAa

faa

1.000 0.500 0.000 0.132

1.000 0.500 0.000 0.521

0.319 0.319 0.319 0.041

0.000 0.295 0.105 0.028

0.640 0.640 0.000

0.600 0.000 0.000

0.498 0.498 0.498

1.000 1.000 0.000

Difference in log likelihood

2.238 4.871 0.988 4.476 9.742 1.976 0.100 0.020 0.580 P= *Transmission probabilities for genetic models are fixed at their theoretical values; the remaining data in the table are estimated by the

x2

segregation analysis. imum lod score is 2.00. We believe these are plausible assumptions but, as the figure illustrates, varying these parameters has little effect on the estimated lod score. This lod score is strongly suggestive of linkage, but more studies would be needed to prove its existence because the accepted criterion for "proof" of a linkage hypothesis is a lod score - 3 (22) and because we do not know the correct mode of inheritance for DNS/ CMM. If linkage of a gene for DNS/CMM to the Rh locus were present, the distance between the two loci would be fairly large (O = 0.30). Heterogeneity of linkage among families was not evident but would be difficult to rule out with certainty if linkage to Rh were "loose." Lod scores between melanoma and 23 autosomal markers are shown in Table 4. There are no lod scores greater than 1, the highest being 0.74 at the Gm locus. DISCUSSION Our segregation analysis documents that the distribution of CMM in high-risk kindreds is consistent with an autosomal dominant Table 2. Maximum likelihood estimates from segregation analysis of DNS/CMM in 14 kindreds Hypothesis EnvironMendelian Dominant Recessive mental Unrestricted Parameter Transmission probabilities* TAAA

TAaA TaaA

Gene frequency (qA) Penetrances

ASA fa0 fa,

Difference in log likelihood

1.000 0.500 0.000 0.244

1.000 0.500 0.000 0.695

0.469 0.469 0.469 0.057

0.000 1.000 0.419 0.069

0.734 0.734 0.003

0.709 0.009 0.009

0.742 0.742 0.000

0.685 0.685 0.000

5.88 11.76 P