Genotypic analysis of primary and metastatic cutaneous melanoma

Abstract

Microdissection genotyping was performed on 16 cases of melanoma, including two cutaneous and one lymph node metastases. Three benign nevi were used as controls. Where possible, tumor was microdissected at several sites. Genotyping involved assessment of loss of heterozygosity [LOH]), which was accomplished using a panel of nine polymorphic tetranucleotide microsatellites. Polymerase chain reaction was performed on the normal tissue sample to establish microsatellite heterozygous status. Informative markers were then tested on microdissected lesional tissue and scored for the presence and extent of allelic imbalance (AI). Microsatellite informativeness varied from 33% to 66%. Benign nevi were without AI. All invasive melanomas manifested acquired allelic loss, which involved 75% or 100% of the markers shown to be informative for each subject. Eleven of 13 (84%) primary melanomas demonstrated intratumoral heterogeneity of AI consistent with development of tumor subclones with differing genotypic profiles within thin as well as thick melanomas. Although a consistent pattern did not emerge among the markers, LOH of 9p21 (D9S254) occurred in 60% (9/15) of the cases followed by 40% of cases displaying LOH of 1p34, p53, 10q (MXI1), and 10q23 (D10S520) and 25% with 5q21 (D5S 592) abnormalities. A third of the cases including the metastatic foci demonstrated two different patterns of AI affecting alternative alleles of the same genomic marker within different parts of the melanoma. Two melanomas in situ did not display LOH of any markers in the informative cases although the in situ component in the invasive tumors had allelic losses that were in part similar to the invasive areas. The results of this study support the expanded use of microdissection genotyping and explore other markers to define the unique mutational profile for malignant melanoma that may complement other histologic characteristics of melanoma.

Introduction

Melanoma presents many challenges to those involved in its detection and management 1, 2, 3. In a recent analysis of 17,600 melanoma patients, tumor thickness and ulceration were found to be powerful predictors of survival. Nevertheless, within each clinicopathologic stage there is significant variability in response to treatment and overall survival [4]. Molecular characterization of the melanoma offers the possibility of incorporating mutational alterations into the histopathologic evaluation with the expectation of better diagnosis and prognostication at the individual patient level 5, 6. Development and progression of cutaneous melanoma involves multiple genetic alterations, and previous studies have shown that chromosomes 1, 6, and 9 are frequently involved. Comparative genomic hybridization of primary melanomas have revealed losses of chromosomes 10, 6q, and 8p, and the most common copy number gains were of 7,8, 6p, and 1q [7]. 6q is involved in more than 80% of metastatic melanoma. A metastases suppressor locus has been mapped to 6q16.3∼q23. Loss of heterozygosity (LOH) of 6q occurs late and appeared to play a key role in metastatic tumor progression 8, 9, 10, 11. In a statistical analysis of more than 3,016 cases of various malignancies, including melanoma, the most common chromosomes involved were 7+, −3p, −6q, −1p, −8p, −17p, −9p, −18, and −22. The majority of the tumors in this analysis displayed more than one cytogenetic route of tumor development and as karyotypic evolution continued they appeared to converge to a common pathway. Chromosome 5q involvement has been relatively uncommon in most of these studies 12, 13. Defective mismatch repair has been reported in 25% of the melanomas but is not as frequent as in colorectal cancers, and there has been no correlation with tumor thickness 14, 15. Cytogenetic and linkage studies implicated 1p and 9p as possible locations for genetic alterations predisposing to melanoma, and germline mutations in CDKN2A (9p21) have been identified in only 40% of kindred, indicating that other genes may predispose to melanoma 16, 17. The 9p21 region encodes the growth suppressor genes TP16^CDKN2A, TP15^CDKN2B, and TP19^ARF, which play key roles in the maintenance of Rb and TP53 tumor suppressor pathways. The INK4 proteins bind to and inhibit the cyclin D-dependent kinases, thereby preventing pRb phosphorylation and exit from the G1 phase. Sporadic melanoma LOH and microsatellite instability (MSI) data suggest that other melanoma susceptibility genes may be present in addition to the CDKN genes [18]. LOH of 9p, however, has also been found in melanoma and in microdissected sporadic nevi 19, 20, 21. Abnormalities of chromosome 1 are common in many cancers and melanomas, especially in the regions 1p36 and 1p22. LOH studies have implicated possible tumor suppressor genes (TSG) and CMM1 (melanoma susceptibility gene), the second CMM2 being localized to 9p21 22, 23, 24. LOH of 9q and 1p have correlated with cell proliferation [25]. Genetic alterations of chromosome 10 have been noted in a large number of carcinomas and melanomas and often involve 10q23.3∼qter. This area includes the PTEN/MMAC gene locus. Region 10q23.3 has been shown to have growth suppressor activity; this region is lost in 30–40% of sporadic melanoma cell lines and is found to be mutated in some in primary and metastatic melanomas. Chromosome 10 abnormalities are seen in ∼30–50% of melanomas, and functional and molecular cytogenetic studies suggest that PTEN genes are involved in sporadic melanoma in addition to genes located in chromosomes 1p, 6q, 7p, 11q, and 9p 9, 26, 27, 28, 29, 30, 31, 32. Immunohistochemical studies on melanocytic lesions have demonstrated a high frequency of immunopositivity in cutaneous melanomas although mutations of TP53 were not noted, but TP53 mutations have been found in melanoma cell lines 33, 34, 35.

The human genome contains mono-, di-, and trirepetitive nucleotide sequences or microsatellites interspaced between genes. The replication error phenotype is also referred to as microsatellite instability (MSI) and refers to changes in the numbers of microsatellite sequences in tumor compared with normal tissue. Microsatellite repeats can be used as polymorphic markers to determine LOH linked to loss of (TSG) located in a given locus. Previous studies using markers for 9p21∼p22, 11q23, 17q21, and 5q22 on sporadic nevi have demonstrated that while microsatellite instability was common in all melanocytic lesions, including dysplastic and banal nevi and melanomas, LOH was restricted to melanoma and some dysplastic nevi. LOH of the same loci were identified in 15% of melanomas and 8% of dysplastic nevi; therefore, LOH may be critical in the progression of a dysplastic nevus [36].

The ideal approach to molecular detection and characterization of cancer in a clinical context would be to define all inherent gene aberrations unique to a given specimen [37]. Surgically excised specimens, however, are usually are small in size and subject to standard fixation for cellular morphologic evaluation. In previous studies we have emphasized the technique of microdissection genotyping as a means to combine the advantages of microscopic and molecular analysis for simple, reliable, and cost-effective analysis of cancer. More recently, we have expanded microdissection genotyping to include sampling of tissue at a variety of sites within each neoplasm and to progressively increase the number of individual mutational tests performed on each microdissected tissue specimen 38, 39, 40, 41. Markers for loci, some of which are known to be involved in melanoma and nevi, including those located on chromosomes 1, 5, 9, 10, and 17, were analyzed in this study.