Genetics of Nonmelanoma Skin Cancer FREE

There are 2 broad classes of genes that become mutated and contribute to cancer: oncogenes and tumor suppressor genes. Cancer-promoting genes, or oncogenes, were originally identified as viral genes that "transform" a normal cell into a malignant cell. Subsequent molecular studies have detected normal counterparts to these viral oncogenes in the human genome (proto-oncogene). Many proto-oncogenes are growth-signaling molecules that become mutated and are perpetually "turned on." Cellular growth signals are then amplified and overwhelm the normal restraints imposed by cellular homeostasis. Oncogenes are, in general, genetically dominant in that a mutation of 1 copy of the proto-oncogene is sufficient to produce the phenotype. RAS is an example of an oncogene that can be mutated in cutaneous melanoma.¹

A second class of genes termed tumor suppressor genes negatively regulates cell growth or promotes cell death. Unlike oncogenes, both copies of the tumor suppressor gene must be inactivated for complete loss of function. One group of tumor suppressor genes (ie, the "gatekeepers") restricts cellular growth by inhibiting the cell cycle and cell division, down-regulating growth signals, or promoting cell death. The patched (PTC) gene is an example of a gatekeeper tumor suppressor gene inactivated in both sporadic and hereditary basal cell carcinomas (BCCs).^{2 - 3} A second group of tumor suppressor genes (ie, the "caretakers") does not directly participate in growth regulation, but rather maintains the integrity of the human genome. When the genetic caretaker function is disrupted, mutations accumulate in gatekeepers, thus allowing for accelerating tumorigenesis. Xeroderma pigmentosum (XP) is an example of a disorder in which genes responsible for repair of UV-induced genetic lesions are deficient, and cutaneous malignancies result at a tremendous rate.⁴

The most common cancer in the United States is BCC, with almost 1 million estimated cases per year.⁵ Basal cell carcinomas can develop in both a hereditary and sporadic fashion.

Nevoid BBC syndrome (Online Mendelian Inheritance in Man [OMIM] entry 109400 [http://www.ncbi.nlm.nih.gov/Omim/]) is an autosomal dominant disorder characterized by the rapid development of numerous BCCs early in life. Most white individuals with nevoid BCC syndrome develop BCCs by a median age of 20 years,^{6 - 7} and individuals with the disorder may develop anywhere from 1 to more than 100 BCCs (median of 8 BCCs⁷ ).

Early analyses of nevoid BCC syndrome kindreds suggested a putative BCC tumor suppressor gene on chromosome 9q22-31.^{8 - 12} Several years later, 2 groups^{3 ,13} were able to demonstrate germline mutations of the PTC gene from patients with nevoid BCC syndrome and somatic mutations in tumor DNA from sporadic BCCs. In recent studies, 15% to 39% of the affected individuals from families with a history of nevoid BCC syndrome were found to harbor mutations in the PTC gene.^{2 ,14} Furthermore, PTC mutations have also been found in sporadic medulloblastomas, breast carcinomas, meningiomas, and one colon cancer cell line.^{15 - 16} More recently, Smyth et al¹⁷ have demonstrated a mutation in PTCH-2 (a homologue of PTC located on chromosome 1p32) in a sporadic BCC.

The PTC gene is involved in the development of various organisms from fruit flies (Drosophila) to mammals. In Drosophila, mutations in the PTC gene cause segmental patterning defects¹⁸ ; hence, the appellation, "patched." The PTC protein binds and inhibits a transmembrane protein, smoothened (SMO) (Figure 1).¹⁹ However, this inhibition can be relieved when the soluble protein sonic hedgehog (SHH) binds PTC. Restriction of SMO signaling is apparently critical for tumor suppression, ie, SMO signaling is growth promoting (Figure 1). Consequently, changes that increase SMO signaling, such as loss of PTC^{2 - 3 ,13 ,15 - 16} or activating mutations of SMO,^{20 - 21} are both associated with human cancer. Cyclopamine inhibits the SMO signaling pathway and may represent a novel approach to targeted cancer treatment.²²

This rare disorder (OMIM entry 301845) is characterized by pitting or "ice pick scars" of the skin (follicular atrophoderma) and development of BCCs by age 30 years.²³ Because male-to-male transmission has been only rarely observed in the reported kindreds, an X-linked dominant mode of inheritance has been suggested. Consonant with these observations, recent studies have linked markers on chromosome Xq24-27 to this syndrome.²⁴

Michaelsson et al²⁵ described a 4-generation pedigree (OMIM entry 180730) with members exhibiting cyanosis, vermiculate atrophoderma of the cheeks, and multiple facial papules on the face. In this pedigree, BCCs along with 1 trichoepithelioma developed around age 35 years. Unlike Bazex-Dupre-Christol syndrome, male-to-male transmission occurs.²⁵

Oley et al²⁶ described a 4-generation family with multiple BCCs. Other findings include sparse scalp and body hair and facial milia (OMIM entry 109390).²⁶

The gene TP53 encodes for the protein p53, which has been termed guardian of the genome. The function of p53 is to sense genotoxic injury and arrest cell division to allow for DNA repairs. However, if the genetic insult is severe, p53 can also induce an apoptotic response in an effort to eliminate defective and potentially malignant cells. TP53 was originally designated as an oncogene given its capacity to transform normal cells into malignant cells.²⁷ This apparent paradox was later resolved with the recognition that mutations of p53 can contribute to cancer formation by acting as either a classic tumor suppressor gene or a dominant-negative oncogene (Figure 2). Certain transdominant mutations (ie, "oncogenic p53") create altered versions of p53 that bind other normal p53 molecules and disrupt their function as well.

Mutations of TP53 have been described in BCCs, with reported rates ranging from 0% to 60%.^{28 - 32} Many of the mutations are CC→TT or C→T changes at dipyrimidine sites suggestive of UV damage.

Activating RAS mutations are among the most common oncogenic lesions in human cancer. The RAS proteins are small G-proteins that transduce intracellular signal. Because RAS is active only when guanosine triphosphate (GTP) is bound, the RAS signal is attenuated by the hydrolysis of this GTP to guanosine diphosphate (GDP). Mutations in RAS frequently alter the rate of hydrolysis, leading to an "activated" protein that inappropriately promotes cell growth and survival. A class of proteins shuts off RAS signaling by increasing the GTP hydrolysis to GDP (GTPase-activating proteins, or GAP proteins). There are inconsistent reports of RAS involvement in BCCs, although some studies suggest that up to 30% of BCCs harbor RAS mutations.^{33 - 37} There is an isolated report of rare nonsense mutations of the GAP gene in BCCs.³⁸

Approximately 200 000 cases of squamous cell carcinoma (SCC) develop per year, causing about 2000 deaths. Unlike BCCs, which have no known precursor lesions, SCCs can emerge from actinic keratoses. Because most actinic keratoses and SCCs occur on chronically sun-exposed sites, UV damage is probably the major cause of genetic injury.

There are several rare syndromes that predispose individuals to squamous neoplasias of the skin; however, there are no monogenic disorders that feature SCCs exclusively. Keratoacanthomas, which are clinically defined as self-resolving keratinocytic neoplasms, histologically overlap with SCCs and have also been described in genetic disorders. Xeroderma pigmentosum (which is discussed below) increases the risk of BCCs, SCCs, keratoacanthomas, and cutaneous melanomas.

Multiple self-healing squamous epithelioma (OMIM entry 132800) is an autosomal dominant condition characterized by multiple self-resolving epithelial tumors that occur as early as the first decade and as late as the fifth decade of life. Although the skin of the face, ears, arms, and legs is most commonly affected, the skin of the anus, scrotum, and anterior abdomen can also become involved.^{39 - 40} Blair et al⁴¹ recently linked this disorder to markers on chromosome 9q22, although the target of mutation in this region is still unknown.

Genetic studies of actinic keratoses and sporadic SCCs have suggested possible tumor suppressor genes on chromosomes 9p, 13q, 17p, 17q, and 3p.^{42 - 43} Except for TP53, which lies on chromosome 17p, the targets for mutations at the other chromosomal sites are unknown.

Inactivation of the tumor suppressor gene TP53 seems to play a central role in the development of actinic keratoses and SCCs. Brash et al⁴⁴ found that 14 (58%) of 24 invasive SCCs of the skin contain mutations in TP53. Many of the alterations were CC→TT mutations that are UV-signature mutations.^{44 - 45} Other studies have also found similar mutations of TP53, albeit at lower rates in different series.^{30 ,46}

How does loss of p53 lead to SCC? First, p53 inactivation can occur in early precancerous actinic keratoses and SCC in situ lesions,^{47 - 49} suggesting that early loss of p53 function contributes to later malignant degeneration, but mutation of p53 alone is not sufficient to fully induce malignancy. Second, UV irradiation of mice with deficient p53 activity leads to a lower apoptotic response in the epidermis than in mice with normal p53 function. This finding suggests that the skin possesses a p53-dependent response to DNA damage that includes a program to abort precancerous cells. Taken together, one model suggests that UV irradiation can select for clonal expansion of p53-mutated cells by acting as both tumor initiator and promoter.⁴⁷

Malkin et al⁵⁰ found that patients with Li-Fraumeni syndrome harbor germline mutations in the TP53 gene. Individuals with Li-Fraumeni syndrome have an increased risk of sarcomas along with other hematological and solid malignancies. However, individuals with Li-Fraumeni syndrome have not been reported to be at increased risk for SCCs.

The reported frequency of RAS mutations in SCCs ranges from approximately 10% to almost 50%.^{37 ,51} Premalignant actinic keratoses also harbor RAS mutations.⁵¹ Variations in the rate of RAS mutations in both SCCs and BCCs may reflect technological differences that have evolved over the past decade.

Although numerous studies on p16 have been performed on cutaneous melanomas, several recent studies have also reported mutations of p16 in up to 24% of SCCs.^{52 - 53} Mutations of p16 may, in part, account for the loss of heterozygosity observed on chromosome 9p21 in SCCs.

Muir-Torre syndrome (MTS) (OMIM entry158320) is an autosomal dominant disorder characterized by sebaceous skin tumors, with or without keratoacanthomas, and internal malignancies. The sebaceous tumors can range histologically from sebaceous adenomas to epitheliomas to carcinomas.⁵⁴ For internal malignancies in MTS, gastrointestinal cancers seem to be the most common, followed by genitourinary cancers.

A genetic interaction between MTS and other familial cancer syndromes had been long postulated on clinical grounds.⁵⁵ With better phenotyping, it became clear that MTS shares features with the colon cancer syndrome hereditary nonpolyposis colon cancer (HNPCC).^{56 - 57} More recently, with the demonstration of inactivating germline mutations of the mismatch repair gene hMSH2 in a subset of HNPCC kindreds, Kolodner et al⁵⁸ confirmed and extended the observed genetic overlap between MTS and HNPCC by demonstrating that the cancer susceptibility in 2 families with MTS was also due to inherited mutations in the hMSH2 gene. A molecular signature of HNPCC tumors is the presence of varying lengths of repetitive DNA sequences ("microsatellite instability") within the tumors.⁵⁹ This microsatellite instability occurs in keratoacanthomas^{60 - 61} and sebaceous tumors associated with MTS.⁶² Thus, it seems that the loss of the DNA mismatch repair machinery in colonic, sebaceous, and keratinocytic epithelia can manifest 2 distinct, but related, syndromes.

Merkel cell carcinoma (MCC), or trabecular carcinoma of the skin, is a rare neuroendocrine malignancy that originates on sun-exposed sites of the skin. The anatomic localization and the increased risk of MCC after treatment with psoralen–UV-A (PUVA)⁶³ suggest that UV damage contributes to the pathogenesis of this tumor. Allelic deletion studies have revealed significant loss of heterozygosity on chromosome 1p.⁶⁴ More recently, similar analysis with more markers pointed to a possible tumor suppressor locus for MCC on chromosome 1p32-1p36, a region that has also been implicated in melanoma tumorigenesis. No specific targets for mutations in MCC have been convincingly demonstrated, although a low rate of TP53 mutations has been reported.⁶⁵ No known familial cases of MCC have been documented.

Pilomatricoma (or calcifying epithelioma of Malherbe) is a benign tumor of follicular structures. Familial cases have been reported without other stigmata⁶⁶ or in conjunction with myotonic dystrophy^{67 - 69} and Rubenstein-Taybi syndrome.⁷⁰ Gat et al⁷¹ first reported that mice expressing a stabilized oncogenic β-catenin targeted to the epidermis developed hair tumors resembling pilomatricomas.⁷¹ Subsequently, Chan et al⁷² analyzed 16 human pilomatricoma specimens and found a high frequency (75%) of activating mutations in the human β-catenin gene. Although the exact details of β-catenin's ability to promote cancer growth are unknown, the gene participates in both intercellular adhesion and transcriptional regulation. β-Catenin is also down-regulated by adenomatous polyposis coli,^{73 - 74} the product of the tumor suppressor gene that is responsible for familial adenomatous polyposis and Gardner syndrome.⁷⁵ This may explain the clinical observation that multiple pilomatricomas occur in Gardner syndrome.^{76 - 78}

Familial cylindromatosis (OMIM entry 132700) is an autosomal dominant condition that is also occasionally associated with trichoepitheliomas. Analysis of familial cylindromatosis kindreds and evaluation of cylindromas both identified a putative tumor suppressor gene on chromosome 16q12.^{79 - 80} Recently, germline mutations of a chromosome 16q12 gene, CYLD, were described in familial cylindromatosis families, while somatic mutations of CYLD were also found in cylindroma specimens.⁸¹ Although the function of the CYLD gene is unknown, the CYLD protein resembles other proteins involved in the attachment of organelles to microtubules.

Cytogenetic analyses of dermatofibrosarcoma protuberans show recurrent lesions such as reciprocal translocations t(17;22)(q22;q13) and supernumerary ring chromosomes derived from the t(17;22). The translocation fuses the collagen type I alpha 1 (COL1A1) gene to the platelet-derived growth factor B (PDGFB) gene.⁸² This chimeric sequence has functionally been shown to be an oncogene in classic cellular transformation assays.⁸³ More recent analyses have confirmed that this fusion seems to be relatively common in dermatofibrosarcoma protuberans and may represent a future molecular marker for this tumor.^{84 - 85}

Xeroderma pigmentosum is a complex of autosomal recessive disorders characterized by intense photosensitivity and early onset of cutaneous malignancies. Actinic keratoses, BCCs, SCCs, and cutaneous melanomas (CMs) usually develop in the first decade of life (median age, 8 years). The anatomic distribution of SCCs, BCCs, and CMs are similar to that of the general population except for an increased risk of CM on the face, head, and neck. However, nearly 90% of SCCs occur on the chronically irradiated head and neck region while only 34% of CMs develop in this area.⁸⁶ Patients with XP experience a 2000-fold increased risk for BCCs, SCCs, and CMs and a 10 000-fold increased risk of SCC on the tip of the tongue compared with patients of similar age.⁴

Through classic complementation studies, 7 genes responsible for the XP phenotype have been identified (XPA to XPG; Table 1). Mutations in different XP genes lead to different phenotypes. For instance, XPA mutations cause the most severe variant with skin cancers and frequent neurological decay. On the other hand, XPC mutations are the most common in the European population and are associated with skin cancers but are rare in neurological findings.

The XP genes are all components of a UV-responsive DNA repair process known as nucleotide excision repair. The conservation of nucleotide excision repair genes from yeast to humans suggests that the nucleotide excision repair apparatus is critical for the survival of all cells in response to UV damage. Briefly, UV irradiation produces specific types of DNA damage (cyclobutane pyrimidine dimers and [6-4] pyrimidine-pyrimidone product), which, if uncorrected, can lead to carcinogenic mutations. The various XP genes encode for proteins that recognize injured DNA (XPA and XPE), unwind the coiled DNA structure to expose the lesion (XPB and XPD), and repair the damaged DNA strand (XPF and XPG).

Because these nucleotide excision repair proteins function as genomic caretakers, the formation of tumors in XP results from the mutagenic inactivation of tumor suppressors and activation of oncogenes. Given the rarity of XP, a comprehensive list of mutated genes (specific or nonspecific for XP) has not been produced. However, molecular analyses of limited XP tumors have shown alterations in TP53,^{87 - 88} RAS,⁸⁹ and p16/CDKN2A.⁹⁰

Molecular genetic studies of tumors so far have largely focused on specific chromosomal regions followed by labor-intensive mutational analyses of candidate genes. With the completion of the Human Genome Project, a more global approach to cancer will be made possible as advances in both genomic mapping and genetic technology will undoubtedly accelerate exponentially. Given its ease of access, skin cancers represent an appropriate group of tumors for analysis in the postgenomic era. How will the Human Genome Project specifically have an impact on skin cancer? One possible scenario is diagrammed in Figure 3.

The DNA from normal individuals can be obtained from the peripheral blood and analyzed for sequence variants. These variants may be useful in determining risk for disease. Mutations in genes that confer a high risk for skin malignancies, such as PTC and nevoid BCC syndrome, are probably uncommon in the general population; however, if mutations in these genes are detected, the treatment for these high-risk individuals may be intensive. Mutations in low- to moderate-risk genes, such as the α-melanocyte stimulating hormone receptor gene MC1R and pigmentation, affect a much larger proportion of the population, although the relative risk for developing cancer may be lower than for hereditary cancer disorders. At the population level, more cancers probably develop in genetically low- to moderate-risk individuals than in high-risk persons.

The human genome is also covered with single nucleotide variations (or polymorphisms, ie, single nucleotide polymorphisms [SNPs]) that may or may not affect the expression or function of any genes. These SNPs, however, are markers that allow for genetic fingerprinting of individuals. Just as blood groups have uncovered associations with diseases unrelated to hematology, these SNPs may provide even more refined clues to disease association. Analysis of SNPs reveals patterns that do not imply mechanism. With enough genetic patterns to generate statistical power, clinical management decisions need not be based solely on an understanding of disease.

Finally, the metabolism of certain chemotherapeutic drugs may also depend on genetic variants of enzymes. Pharmacogenomics has emerged as a potentially powerful ally in defining treatment and adverse effect profiles.

Once a tumor develops, the cancerous cells themselves offer a source of genetic material for analyses. In hematological malignancies, the presence or absence of recurrent translocations has provided critical diagnostic information for years. With the emergence of new genomic technologies, molecular diagnoses of solid tumors have gained prominence. One such example is the recurrent COL1A1-PDGFB fusion that may be used as a diagnostic feature in morphologically questionable dermatofibrosarcoma protuberans.

Because cancers are by definition tissue specific, the pattern of genes expressed in the tumor tissue has been a major focus of genomic research. Although the various approaches are beyond the scope of this article, expression profiling can allow for a fingerprint of the tumor itself. Once again, even without any mechanistic information, the pattern itself may provide more precise prognostic data on survival or response to therapy. Because the process to better understand disease mechanisms is labor intensive, a longer-term goal of the postgenomic era is to define targeted therapies. For instance, both T4 endonuclease⁹¹ for patients with XP and cyclopamine²² for PTC-mutated tumors represent novel targeted therapies derived from our current understanding of nucleotide excision repair and the SMO signaling pathways, respectively.

In summary, technological advances in cancer have largely enhanced discriminatory capabilities. The light microscope and electron microscope allowed for cellular and subcellular discrimination, respectively. Genomic technologies represent a departure from morphologically based criteria and usher in a new era of genetically based criteria for neoplastic behavior.

Accepted for publication July 24, 2001.

Corresponding author: Hensin Tsao, MD, PhD, Department of Dermatology, Massachusetts General Hospital, Barlett 622, 48 Blossom St, Boston, MA 02114 (e-mail: tsao.hensin@mgh.harvard.edu).