1. ^*Medical Genetics, S. Camillo-Forlanini Hospital, Rome
2. ^†IRCCS San Gallicano Dermatological Institute, Rome
3. ^‡Medical Genetics, Experimental Medicine and Pathology Department of University "La Sapienza," Rome, Italy

Correspondence: Paola Grammatico, Medical Genetics, Experimental Medicine and Pathology Department, University "La Sapienza," c/o AOS Camillo-Forlanini, Via Portuense n. 332, 00149 Rome, Italy. Email: paola.grammatico@uniroam.it

Received 13 May 2003; Revised 28 July 2003; Accepted 20 August 2003; Published online 12 February 2004.

To the Editor:

The CDKN2A gene, located in 9p21, encodes two distinct tumor suppressor proteins: p16/INK4A and p14/ARF. Exons 1, 2, and 3 give rise to protein p16/INK4A, which is part of the cyclin D1/CDK4/p16/pRB signaling pathway and plays the role of proliferation inhibitor of normal cells (^{Dracopoli and Fountain, 1996}). In contrast, exon 1, spliced to exons 2 and 3, in a different reading frame, encodes p14/ARF, which acts through the MDM2/p53 pathway (^{Pomerantz et al, 1998;Stott et al, 1998}).

CDKN2A is frequently mutated or deleted in a wide variety of tumors (^{Nobori et al, 1994;Kamb et al, 1994;Pollock et al, 1996}), whereas its germline mutations are the most common cause of inherited susceptibility to melanoma having been described worldwide in melanoma families and in individuals with multiple primary melanomas (^{Ruas and Peters, 1998;Greene, 1999;Hashemi et al, 2000}). Although it is possible to identify a CDKN2A variation associated with melanoma susceptibility in about 25% of melanoma-prone families, a significant number of kindreds linked to 9p21 do not carry alterations in the CDKN2A-coding region (^{Liu et al, 1997;Newton Bishop et al, 1999}). This phenomenon may be attributable to possible additional genes located in 9p21 involved in melanoma predisposition or to the presence of nondetected CDKN2A mutations. Mutational analyses performed worldwide in many melanoma-prone families has mainly focused on the CDKN2A-coding region and intron/exon boundaries. The first evidence of clinically relevant germline noncoding CDKN2A mutations come from the identification in Canadian subjects of a variant located 34 bases upstream the CDKN2A exon 1 initiation codon (^{Liu et al, 1999}). Subsequently,^{Harland et al (2001)} described in six English melanoma-prone families a deep CDKN2A intron 2 mutation (A to G), introducing a functional false GT splice donor site, at -105 nucleotides from exon 3. These authors were able to demonstrate that the IVS2-105A/G mutation partially abolishes the function of the wild-type CDKN2A intron 2 splice donor site, resulting in the production of three different transcripts: the wild-type one, plus two aberrantly spliced mRNAs. The predicted protein derived from the two aberrant mature transcripts diverges from the wild-type p16 only in the last four amino acids, with no apparent effect on protein function.^{Harland et al (2001)} proposed that the CDKN2A intronic mutation could act by causing a low level export of mature mRNA from the nucleus to the cytoplasm. In addition, all carriers of the IVS2-105 variant had the same haplotype for 9p21 markers, indicating the common origin of the mutation in the English population.

To determine whether the IVS2-105 variant may play a role in our Italian melanoma families and multiple melanoma cases that do not harbor CDKN2A-coding mutations, we analyzed intron 2 and identified the variant in a man aged 49 y affected by sporadic multiple melanoma. The patient with blue eyes, cutaneus phototype 2, multiple freckles and more than 50 acquired cutaneous melanocytic nevi developed 8 primary cutaneous melanomas, five of which removed at age 37.

Investigation of the patient's relatives Figure 1 showed that although no other family members are affected by melanoma, the patients maternal first cousin (son of patients mother's monozygotic twin) died at age 28 from diffuse lung metastasis, histologically defined as mucinous carcinoma. The primary tumor was never identified.

Figure 1.

Patient's family pedigree. The patient (proband) is individual III:2. His cousin (III:5) died at age 28 from lung metastasis of unknown origin. Genotypes are shown below various individuals, with CDKN2A genotype above and MC1R below.

Full figure and legend (18K)

Sequencing analysis of CDKN2A exons 1, 1, and 2; coding region of exon 3; partial 3'UTR; and terminal fragment of intron 2 showed that our patient was heterozygous for the IVS2-105A/G mutation Figure 2. The detection of this variant, never described before except in English melanoma-prone families, led us to check also the unaffected parents to verify the possible transmission. We did not detect the CDKN2A variant in either parent. To define if the deep intronic variant was a new mutation arising on a different haplotype from the English families and, at the same time, to rule out false paternity, we haplotyped CDKN2A surrounding markers in the patient and his parents. The analysis of the same 9p21 microsatellite markers utilized by^{Harland et al (2001)} demonstrated that the patient inherited from his parents two haplotypes that profoundly differ from the common one present in the English melanoma-prone families. CDKN2A analysis extended to our patient's two children detected the IVS2-105A/G variant in the 20-y-old unaffected son.

Figure 2.

Chromatogram showing the IVS2-105A/G CDKN2A variant. The heterozygous A/G peak is easily detected.

Full figure and legend (13K)

Our findings suggest that the IVS2-105 base could represent a mutational hot spot and that the IVS2–105A/G is a recurrent mutation, having arisen independently at least twice. In addition, our results confirm that the IVS2-105 variant may be a causative mutation for melanoma development, making it its linkage disequilibrium with a possible "real" mutation, in English melanoma-prone families, unlikely.

MC1R gene analysis of the proband and the family revealed that our patient is a compound heterozygote for the Val60Leu and the Arg160Trp common variant alleles that could represent a further genetic factor predisposing to melanoma (^{Box et al, 2001;van der Velden et al, 2001}). The patient's daughter is homozygous for the Arg160Trp MC1R variant, while the son, who inherited the CDKN2A IVS2-105 and MC1R Arg160Trp variants from his father, is presumably at significant risk of melanoma development.

Furthermore, the premature death in the patient's first cousin at the age of 28 in consequence of diffuse lung cancer metastases may indicate that another cancer gene may be segregating through the kindred, cooperating with the IVS2-105 and with the MC1R variants in development of melanoma in our proband.

In conclusion, our study seems to confirm the role of the IVS2-105A/G mutation in melanoma susceptibility. It is thus critical take into account the possibility of this intronic CDKN2A variant in patients with no coding sequence mutations, to improve the accuracy of genetic counseling.