Li-Fraumeni syndrome (LFS; MIM 151623), which results from germline mutations of the p53 gene (MIM 191170), represents one of the most devastating genetic predispositions to cancer and is characterised by a wide spectrum of early onset malignancies including bone and soft tissue sarcomas, brain tumours, adrenocortical tumours, leukaemia, and pre-menopausal breast cancers (for review, see Chompret¹ and Varley²). The wide spectrum of malignancies, the range of age of tumour onset, and the incomplete penetrance in males complicate the genetic counselling and medical follow up of affected families.

A recent report indicating that the SNP309 (T→G variation)(g.2580T→G; Genbank accession number AF527840; rs2279744), within the MDM2 gene (MIM 164785), was associated with accelerated tumour formation in p53 mutation carriers^3,4 is of particular interest in this context of phenotypic variability. MDM2 is a key negative regulator of p53, which targets p53 towards proteasomal degradation, and the SNP309 T→G variation, located in the first intron of MDM2, has been found to increase Sp1 transcription factor binding and consequently MDM2 level. As the 72Arg variant of the p53 protein has been shown to have a higher affinity towards MDM2 compared with the 72Pro variant,⁵ we speculated that the p.Arg72Pro polymorphism might also influence the age of tumour onset in germline p53 mutation carriers. The impact of this polymorphism, as a risk factor in sporadic cancers,⁶ or as a modifier factor in mendelian forms of cancers such as BRCA1 related breast cancers,⁷ has already been analysed, but the results are controversial.

We therefore investigated the effect on tumorigenesis of these two polymorphisms in 61 affected or unaffected germline p53 mutation carriers collected by the French LFS network.

METHODS

Patients

In this study, 61 affected or unaffected germline p53 mutation carriers, from 41 unrelated LFS families collected by the French LFS network, were investigated. Clinical characteristics of the families and description of the mutations are available upon request. DNA analysis was performed after informed consent was obtained.

MDM2 genotyping

The first intron of MDM2 was amplified by PCR using the following primers: 5′-GAGGTCTCCGCGGGAGTTC-3′ and 5′-CTGCCCACTGAACCGGC-3′. PCR was performed in a 25 µL volume containing 100 ng of genomic DNA, 1.2 mmol/l MgCl₂, 160 µmol/l dNTPs, 0.4 µmol/l of each primer and 1 U Taq DNA polymerase (ABgene). After a denaturation step of 3 minutes at 95°C, the PCR consisted of 35 cycles of 10 seconds at 94°C, 10 seconds at 60°C, and 10 seconds at 72°C, and was followed by a final extension step of 7 minutes at 72°C. After purification of the PCR products using an extraction kit (QiaQuick Gel Extraction Kit; Qiagen), sequencing analysis of the MDM2 SNP309 was performed using a sequencing kit and automated DNA sequencer (Prism AmpliTaq FS Ready Reaction Dye Terminators sequencing kit and model 377 automated DNA sequencer; PE Applied Biosytems).

Determination of the phase of the p53 codon 72 polymorphism

The phase (cis or trans) of the p53 codon 72 polymorphism, relative to the germline mutation, was determined in the Arg/Pro heterozygotes, by cloning either PCR amplified genomic DNA in bacteria, using primers encompassing exons 4 to 5, 7, or 8, or PCR amplified cDNA in yeast, as previously described.⁸

Statistical analysis

Analysis of variance was used to compare the mean age of onset between the different groups and a χ² test was used to compare the distribution of tumour types between the different genotypes.

RESULTS AND DISCUSSION

For MDM2, the observed frequencies in our series for the SNP309 T/T, T/G, and G/G genotypes were 37%, 46%, and 17%, respectively (table 1). Among the 41 affected carriers for whom age of onset was known, comparison of the mean age of first tumour onset between the three MDM2 genotypes, using variance analysis, revealed no significant difference. When we stratified the patients according to the presence of the G allele, we observed a difference between G allele carriers (n = 27) and patients homozygous for the T allele (n = 14) (19.6 v 29.9 years, p<0.05; table 2). This result is in agreement with the results published by Bond et al.³ We detected no dosage effect for the G allele, as the mean age of tumour onset in T/G patients was 17.8 years, whereas it was 23.3 years in G/G subjects. Among the 34 G allele carriers, 50% had developed a tumour before 20 years of age, whereas only 21% of the subjects homozygous for the T allele were affected before this age (fig 1A).

For the p53 codon 72 polymorphism, the distribution of the Arg/Arg, Arg/Pro, and Pro/Pro genotypes was 41%, 46%, and 13%, respectively (table 1). We observed a difference in the mean age of tumour onset only when the patients were stratified according to the presence of the Arg allele. Indeed, the mean age of tumour onset in affected carriers of the Arg allele (n = 38) was 21.8 years and in Pro/Pro patients (n = 8) 34.4 years (p<0.05, table 2). Of the 52 Arg allele carriers, 40% had developed a tumour before the age of 20 years, whereas only 12% of the eight subjects homozygous for the Pro allele were affected before this age (fig 1B). Because the potential effect of the codon 72 polymorphism may depend on its cis or trans position relative to the germline mutation, we determined the phase in the Arg/Pro heterozygotes, but we did not detect any significant effect of the phase on the age of tumour onset.

Because both MDM2 SNP309 and the p53 codon 72 polymorphism might exert their effect through the degradation of p53, we analysed the combined effect of the MDM2G and p53Arg alleles. Despite a limited number of subjects, we observed a clear difference in the mean age of tumour onset of p53 mutation carriers according to their combined genotype for the MDM2 and p53 polymorphisms. The mean age of onset was the lowest (16.9 years) for the 23 individuals with putative at risk genotypes at both loci (T/G or G/G and Arg/Pro or Arg/Arg). The mean age of onset was the highest (43.0 years) for the two people who did not carry at risk alleles at both loci (T/T and Pro/Pro), and intermediate for the 16 with a genotype at risk at only one locus (28.5 years) (table 3). The difference between these three groups was highly significant (p<0.001), suggesting a cumulative effect of both polymorphisms.

In contrast, we could not detect any effect of these polymorphisms on the tumour type in our series. Indeed, the comparison of the tumour type, stratified into three categories (breast, sarcoma, other cancer), between individuals with at risk genotypes at both loci and individuals with other genotypes was not significant (χ² = 4.50). In each family, we obtained DNA samples only from a limited number of affected individuals (1–3 people), which hampered the analysis of the impact of the polymorphisms on the tumour type within each family. Neither MDM2 SNP309 nor the p53 codon 72 polymorphism could explain the incomplete penetrance in male patients of germline p53 mutations that we had previously estimated to be 41% at 45 years of age (compared with 84% in female patients).⁹

Therefore, our results confirm the impact of the MDM2 SNP309 G allele on the age of tumour onset in germline p53 mutation carriers, and suggest that this effect may be amplified by the p5372Arg allele, although this latter observation needs to be confirmed on a larger series of germline p53 mutation carriers. Polymorphisms affecting p53 degradation therefore represent one of the rare examples of modifier genetic factors identified so far in mendelian predispositions to cancer.