Introduction

Li–Fraumeni syndrome (LFS) is an autosomal-dominant cancer predisposition syndrome. The main tumour types of LFS include bone- and soft-tissue sarcoma (STS), breast cancer, brain tumour, adrenocortical carcinoma and leukaemia.1 The classical LFS criteria are a proband with a sarcoma aged under 45 years and a first-degree relative with any cancer aged under 45 years, plus a first- or second-degree relative in the same lineage with any cancer aged under 45 years or a sarcoma at any age.2 In addition, Li–Fraumeni-like syndrome (LFL) criteria were formulated as a proband with any childhood tumour, or a sarcoma, brain tumour or adrenocortical tumour aged under 45 years and a first or second degree in the same lineage with a typical LFS tumour at any age, plus a first- or second-degree relative in the same lineage with any cancer under the age of 60 years.3 Less stringent LFL criteria were formulated by Eeles4 as two first- or second-degree relatives with typical LFS tumours at any age. In 1990, germline mutations in the TP53 gene were found to be associated with LFS.5 At present, TP53 germline mutations are detected in approximately 75% of LFS and 40% of LFL families.6 Because still 25–60% of LFS/LFL families do not carry a germline TP53 mutation, alternative LFS susceptibility genes have been proposed.7 Moreover, single-nucleotide polymorphisms (SNP) may contribute to predisposition to cancer.8 Recently, it was suggested that an SNP in the promoter region of the MDM2 gene is associated with a significantly earlier age of onset of tumours in both TP53 mutation carriers and sporadic TP53-negative STS.9 The MDM2 gene is an important negative regulator of TP53. Bond et al9 showed that the MDM2 SNP309 G creates an improved SP1 site, leading to higher basal levels of MDM2 in cells, thereby attenuating the p53 pathway. The impact of the MDM2 SNP309 G on the age of tumour onset in germline TP53 mutation carriers was confirmed by Bougeard et al,10 by screening 41 affected TP53 mutation carriers for the presence of an SNP309 G allele. Wilkening et al11 showed that this polymorphism does not seem to play a role in familial breast cancer.

We investigated the presence of the SNP309 G allele in Dutch and Finnish TP53 mutation carriers to look at the effect on age of onset. Furthermore, we tested 72 family members of 68 TP53-negative LFS and LFS-related families from the Netherlands to determine the possible effect of SNP309 G on cancer susceptibility.

Patients and methods

For this study, 108 selected cancer patients were screened for TP53 germline mutations, 97 at the Family Cancer Clinic of the Netherlands Cancer Institute and 11 at the Helsinki University Central Hospital. These patients were divided into four groups: (1) classic LFS,2 (2) LFL according to Birch et al3 or Eeles,4 (3) LFS-suggestive, including childhood onset (under 18 years) sarcoma or brain tumours, two or more primary tumours at any age, or two first-degree relatives with a tumour at any age, of which at least one is a typical LFS tumour and (4) breast cancer before 30 years of age (without detectable BRCA1 or BRCA2 mutation) (Table 1). Groups 2–4 combined are further referred to as LFS-related families. Thirty-six TP53-positive and 72 TP53-negative family members out of 87 families (19 TP53-positive families, 68 TP53-negative families, respectively) were tested for the presence of SNP309 G in the MDM2 gene. The 68 TP53-negative families were all Dutch; in Finland, TP53-negative families were not tested for the presence of SNP309 G in the MDM2 gene. DNA from peripheral blood lymphocytes was isolated according to standard procedures. Mutation analysis of the TP53 gene was performed by sequence analysis of all coding exons (2–11) and flanking intron–exon boundaries of these exons using standard procedures. The 19 TP53 germline mutations in the TP53-positive families included 12 missense mutations, four stop mutations and three splice site mutations. DNA was sequenced for the presence of SNP309 G in the MDM2 promoter (Different primer sets were used in the Netherlands and Finland; primer 1: 5′-TGTAAAACGACGGCCAGTCGGGAGTTCAGGGTAAAGGT-3′, primer 2: 5′-CAGGAAACAGCTATGACCTCGGAACGTGTCTGAACTTG-3′ in Amsterdam, primer 1: 5′-GTTTTGTTGGACTGGGGCTA-3′, primer 2: 5′-CGGAACGTGTCTGAACTTGA-3′ in Helsinki).

Table 1 SNP309 genotype in patients and controls
Full size table

A control group of 230 Dutch healthy blood bank controls, males and females and 551 Finnish healthy female control subjects (from the same geographical region) were screened for the presence of SNP309 G. Proportions of haplotype frequencies in the combined control group were adjusted for the numbers of Dutch and Finnish patients. The Dutch controls were genotyped as described above, the Finnish controls were genotyped using restriction fragment length polymorphisms: PCR products were digested with MspA1 I (New England Biolabs) and run on a 3% agarose gel; the PCR product with a T allele is cut once and the PCR product with a G allele is cut twice.

A t-test was used to determine the statistical significance for the age of onset of cancer between the groups with and without SNP309 G. Proportions were tested by a χ2 test.

Results

The classification of the 108 LFS and LFS-related patients into four groups (see Patients and methods) is shown in Table 1a. The presence of the SNP309 G in TP53-positive patients was found in heterozygous state, T/G, in 17 out of 36 (47%) patients and in the homozygous state, G/G, in one out of 36 (3%) patients. In TP53-negative patients, T/G was found in 16 out of 72 (22%) patients, and G/G was found in 17 out of 72 patients (24%, Table 1a). The Dutch and Finnish population controls, including the ratio-adjusted Dutch/Finnish controls are listed in Table 1b. In the TP53-negative LFS and LFS-related patients, the G/G genotype was twice the percentage in the general population (24 vs 12%, respectively, P=0.02), and the T/G genotype proportion was significantly lower as compared to the general population (22 vs 47%, P<0.01).

The prevalence of the different first tumour types in both TP53-positive and TP53-negative patients is shown in Figure 1a and b, with breast cancer and STS being the most frequent cancers. The average age of onset of all first cancers was 34.3 years. When the group is divided in TP53-positive and TP53-negative patients, the average age was 37.6 and 32.6 years, respectively. When the TP53-positive group was further divided by the presence or absence of SNP309 G, the T/T group revealed a mean age of 45.5 years, as compared to the group where SNP309 G was present in heterozygous or homozygous state (T/G or G/G) where the mean age was significantly earlier, 29.7 years (P=0.005, Figure 1c). In the T/T group of TP53-negative patients, the mean age was 32.1 years, and in the T/G and G/G group combined, the mean age was 33.3 years (P=0.749; Figure 1d).

Figure 1
figure 1

Prevalence of different tumour types and mean age of tumour onset. (a) First tumours in TP53 mutation carriers (n=36). Other=stomach cancer, melanoma, mesothelioma, ovary cancer, lung cancer and thyroid cancer. (b) First tumours in TP53-negative LFS and LFS-related patients (n=72). Other=bladder cancer, Hodgkin's lymphoma, kidney cancer, thyroid cancer, basal cell carcinoma, acute myeloid leukaemia, colon cancer, non-Hodgkin's lymphoma and stomach cancer. (c) Mean age of tumour onset in TP53 mutation carriers according to their MDM2 SNP309 genotype. (d) Mean age of tumour onset in TP53-negative LFS and LFS-related patients according to their MDM2 SNP309 genotype.

Full size image

Discussion

We showed that the presence of an SNP in the MDM2 gene, SNP309 (T>G variation) in Dutch and Finnish TP53 germline mutation carriers was associated with accelerated tumour formation. Our results show a 16 years earlier age of tumour onset in TP53 mutation carriers with an SNP309 G allele as compared to the T/T SNP 309 group (Figure 1c), confirming the conclusions of Bond et al9 and Bougeard et al,10 who showed an age difference of 7 and 10 years, respectively. The SNP309 G polymorphism is proposed to act as a genetic modifying factor in TP53 mutation carriers. Because of the small numbers, our analysis has been carried out in a joint sample of Dutch and Finnish origin. Although the allele frequencies and genotype frequencies differ between the Dutch and Finnish controls, as shown in Table 1b, the genotype frequencies in the TP53-positive group did not significantly differ from the ratio-adjusted Dutch/Finnish control group, allowing a combined analysis.

Neither Bond et al9 nor Bougeard et al10 screened LFS and LFS-related patients without a TP53 germline mutation. In our TP53-negative LFS and LFS-related group, the age of tumour onset for SNP 309 G allele patients and T/T patients was not significantly different. The presence of the SNP 309 G allele was therefore not shown to have a tumour accelerating effect in our TP53-negative LFS-related patients. However, it was striking to see that the percentage of the G/G genotype in the TP53-negative LFS and LFS-related patients is twice the percentage in the general population. These data indicate an association between homozygosity for SNP309 (G/G) and the occurrence of cancer, although not related to the age of onset. The fact that these patients were considered for counselling might well be due, in part, to the presence of the homozygous G allele, under the assumption of the SNP309 G polymorphism acting as a modifier or (additional) disease-causing factor. Within the TP53-negative group, there is no difference in the age of tumour onset in the homozygous SNP309 G carriers as compared to the T/T and T/G carriers. Although three out of four classical LFS family members show a G/G genotype, the family history does not seem to be specific for the homozygous SNP309 G carriers (Table 1).

Bond et al9 found the presence of SNP309 to be associated with earlier age of onset for sporadic STS, specifically for the homozygotes (G/G). In our TP53-negative patients, 13 patients developed STS as first tumour, six patients showed a T/T genotype, five a T/G genotype and two a G/G genotype, the mean ages did not significantly differ between the G/G and T/G or T/T groups. Although both groups (our group of TP53-negative LFS and LFS-related STS patients and Bond's group of sporadic STS) are TP53 negative, the difference in proportion of G/G and in family history of patients, in addition to the small numbers in both studies, could explain the discordance in age of onset.

In conclusion, our results confirm the association of the presence of SNP309 G with a significantly earlier age of onset of the first tumour in TP53 germline mutation carriers. As such, the TP53 combined MDM2 SNP309 G germline mutation now becomes an established example of how a gene–gene interaction synergistically acts in cancer susceptibility. This association raises the question whether MDM2 SNP309 G analysis should be embedded in clinical diagnosis of TP53 mutation carriers. This remains controversial, especially because for many tumour types no early detection procedures are available.

Although the presence of an SNP309 G allele in TP53-negative LFS and LFS-related patients was not shown to be correlated with an earlier age of onset of tumours, the higher prevalence of MDM2 SNP309 homozygous G/G carriers in the TP53-negative group may suggest that this allele contributes to the LFS phenotype. Before clinical genetic implementation, these observations need to be confirmed in a larger series of TP53-negative LFS and LFS-related patients.