Editor—The Li-Fraumeni syndrome (LFS) is a rare familial cancer syndrome that predisposes gene carriers to the development of diverse early onset malignancies, including soft tissue sarcomas, osteosarcomas, adrenocortical carcinomas, brain tumours, breast carcinomas, and leukaemia,1-3 with other cancer types occurring less frequently.4-6 Families adhering to the classical definition of the syndrome include those in which one subject, usually the proband, is diagnosed with a sarcoma before 45 years of age, and has a first degree relative with cancer before 45 years of age, and another first or second degree relative in the same parental lineage with any cancer diagnosed under 45 years of age or with sarcoma at any age.7 Families that do not meet these strict criteria are referred to as LFS-like (LFS-L).8 9

The majority of classical LFS families harbour germline mutations of the p53 tumour suppressor gene.10-12 However, in the remaining “classical” LFS families, and in most LFS-L families, no alterations inp53 have been found.8 9 This lack of an absolute phenotype:genotype concordance could be attributed to incomplete screening of the p53 gene, inactivation of the p53 protein through interaction with other cellular proteins or viruses, or defects in other genes involved inp53 mediated cell cycle regulatory pathways.

p16^INK4a is a candidate gene that could account for the cancer predisposition inp53 wild type LFS families. It is located on chromosome 9p21 and its three exons encode a 156 amino acid, 15.8 kDa protein.13 p16^INK4a is a cyclin dependent kinase inhibitor that is frequently mutated or deleted in many human cancer cell lines14-16 and some sporadic malignancies, including sarcomas, breast cancer, leukaemia, and brain tumours, which are all component tumours of LFS.17 Germlinep16^INK4a alterations are associated with familial melanoma.18-20p16^INK4a shares functional similarities with p53 in that it blocks progression through the cell cycle at G1/S by inhibiting CDK 4/6 mediated phosphorylation of Rb.13 14 Althoughp16^INK4a is most frequently inactivated by homozygous deletion, point mutations or somatic methylation of 5′ regulatory regions are also important mechanisms of gene inactivation.17 Methylation has not been reported to date in the germline.

In view of the comparable biological and phenotypic features ofp53 andp16^INK4a inactivation, we proposed that germline inactivation of thep16^INK4a gene could account for the predisposition to cancer development in a proportion of LFS families that harboured wild type p53.

Genomic DNA was isolated from 103 subjects ascertained from 82 cancer families. These had all been previously screened for the presence ofp53 mutations in exons 2 and 4-11 by single strand conformational polymorphism (SSCP) analysis and DNA sequencing as previously described.26-29 Where available, samples from more than one family member were examined to determine whether a germline gene alteration occurred de novo or was inherited. Bothp53 wild type andp53 mutant samples were included in thep16^INK4a analysis to determine whether germline alterations of one gene precluded the occurrence of alterations of the other. Of the 103 samples, 24 (from 17 kindred) fulfilled the classical definition of LFS, 63 (from 51 kindred) met the criteria of LFS-L, while the remaining 16 (from 14 kindred) exhibited a family history of cancer but did not meet the strict criteria of either LFS or LFS-L. Excluded from the study were 12 subjects for whom complete data regarding tumour type were not available and two who were from a family in whom the proband harboured threep53 alterations.29

All samples (p53 wild type andp53 mutant) were screened forp16^INK4a mutations using SSCP analysis which has been shown to be both 90% specific and sensitive.30 Each of the three exons ofp16^INK4a was amplified using a modification of previously described primers and conditions.31 In particular, the 461 bp exon 2 fragment was digested with SmaI to yield two smaller fragments (179 bp and 282 bp). The samples were denatured for 15 minutes at 85°C and loaded immediately onto a polyacrylamide-TBE non-denaturing gel containing a range of acrylamide (4.5-9.0%) and glycerol (2.0-10.0%) conditions. Electrophoresis was conducted at room temperature for 80-400 Watt hours depending on the concentration of acrylamide and glycerol and both positive and negative controls were applied to each gel. A minimum of two gel conditions were used for each fragment in order to increase the likelihood of detection of subtle band shifts. DNA samples with detectable and reproducible band shifts on SSCP analysis were reamplified with the primers encompassing the abnormal region. Fragments were purified using a Qiaex II extraction kit and directly subcloned into a T-tailed pBSK vector. A minimum of six clones were sequenced by the Sanger dideoxynucleotide method with a Sequenase 2.0 kit (US Biochemical) to determine the precise nature of the sequence alteration. In addition, where a base pair alteration was found, duplicate PCR reactions were performed to rule out the possibility of a PCR generated artefact. For the eight samples that did not amplify on repeated attempts, Southern blot analysis was undertaken to determine whether homozygous or heterozygous deletion of the exon was present.32

Of the 103 subjects examined forp16^INK4a alterations, two unrelated subjects harboured an identical single base pair alteration in intron 1, 36 bp downstream of the intron/exon junction (fig 1A, B), and five harboured a common Ala148Thr polymorphism in exon 2 (fig 2A, B). Results from previous studies in our laboratories indicated that 26 patients had a total of 28 p53 alterations. Of these 26, 13 were within the classical LFS group and 13 were in the LFS-L group. There were a total of 17 missense mutations, two polymorphisms (Arg213Arg), three nonsense mutations, one 6 bp deletion, one 4 bp deletion, one subject with a polymorphism (Arg213Arg) and a 1 bp deletion, and one subject with a polymorphism (Arg72Pro) and a 2 bp deletion. Both the p16^INK4a andp53 mutation analyses are summarised in tables 1 and2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

p16^INK4a intron 1 alteration. (A) SSCP gel. The arrows indicate the relevant band shifts. Running conditions: 10% glycerol/7.5% acrylamide gel run at 13 W for 17 hours at room temperature. (B) Sequencing gel. WT=wild type, A=adenine, G=guanine, C=cytosine, T=thymine.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

p16^INK4a exon 2 polymorphism. (A) SSCP gel. The arrow indicates the relevant band shift. Running conditions: 10% glycerol/7.5% acrylamide gel run at 6.5 W for 18.5 hours at room temperature. (B) Sequencing gel. WT=wild type, A=adenine, G=guanine, C=cytosine, T=thymine.

All of the p16^INK4a alterations were observed in separate kindred. For p53, there were two families (denoted † and ‡ in table 2) where two subjects harboured the same mutation and one family (denoted * in table2) with three mutant gene carriers. However, one member of this third family had a completely different mutation from the other two. Only one subject (509-C) carried both thep16^INK4a Ala148Thr polymorphism in exon 2 and a p53 mutation (Arg196Pro). This subject came from a classical LFS family. Southern blot analysis did not show either homozygous or heterozygous deletions (data not shown).

Both p53 andp16^INK4a play important roles in the development of multiple types of cancers. Inactivation ofp53 has been found to occur in up to 50% of all cancers, and p16^INK4a inactivation is observed in up to 26% of all malignancies.17 Both p53 andp16^INK4a exert their growth suppressor effects at the G1/S phase of the cell cycle and inactivation of either is associated with the development of constellations of tumours with some phenotypic similarities. Bearing these features in mind, p16^INK4a seemed an ideal candidate gene to account for the development of LFS in those kindred that harboured wild type p53 in the germline.

Of the 103 germline samples analysed forp16^INK4a point mutations and deletions in this study, two harboured an intron 1 alteration and five a common exon 2 polymorphism. Only one sample showed a p53 mutation along with thep16^INK4a exon 2 polymorphism. There was no evidence forp16^INK4a deletion in any of the samples analysed.

Although we ruled out p16^INK4a point mutations and deletions, we did not examine methylation as a means of inactivation in our series. However, the likelihood ofp16^INK4a methylation defects occurring in the germline of this sample population is felt to be exceedingly low as these alterations have not been reported to occur in other studies of familial melanoma patients.p16^INK4a methylation is, however, thought to be an important mechanism of inactivation in sporadic tumours.33 34 Moreover, a zero numerator does not necessarily mean that nop16^INK4a mutations will ever be found in a similar population if enough samples are analysed. It has been shown that for a zero numerator, 3 is a good approximation of the upper boundary for a 95% CI.35 Based on this “rule of 3”, we can establish the maximum long run risk associated with the negative observations in LFS, LFS-L, and other cancer families. We can be 95% confident that the chance of a patient similar to our 103 cancer patients having ap16^INK4a mutation is at most 3/103 or 2.9%. If we consider only LFS patients, then the likelihood that one will have a p16^INK4a mutation is 3/24 or 12.5%. Similarly, LFS-L patients have a risk of 3/63 or 4.8% and miscellaneous cancer patients have a risk of 3/16 or 18.8%.

Our conclusion that sequence alterations ofp16^INK4a do not appear to account for the development of cancer in LFS or LFS-L families is based on the assumption that neitherp16^INK4a intronic alterations nor polymorphisms have any role in human tumorigenesis. This may, in fact, not be the case. Recent reports suggest that at least one polymorphism in another tumour suppressor gene, namelyp53 (Arg72Pro), may contribute to cancer development,36 although other reports refute this susceptibility.37 Moreover, bothp53 andp21^CIP1 intronic alterations that may be associated with cancer predisposition have been reported38 39 as well as a base substitution in the 3′ UTR of p16^INK4a .40

Inactivation of either p53 orp16^INK4a ultimately leads to inhibition of cell cycle regulation by preventing phosphorylation of pRB. It may not be surprising that germline alterations ofp16^INK4a were not observed in our study population, as no melanoma families were documented. On the other hand, somatic mutations of the gene had been observed in a wide spectrum of human tumours, including many associated with LFS. Our results indicate that it is not likely thatp16^INK4a plays a significant role in the predisposition to cancer in LFS which support findings of a recent smaller study.41 Further studies of genes involved in p53 mediated growth regulation may be fruitful in identifying genetic events involved in cancer predisposition in LFS families who harbour wild typep53.