Background: Li–Fraumeni and Li–Fraumeni-like syndromes (LFS/LFL), characterised by the development of multiple early onset cancers with heterogeneous tumour patterns, are associated with germline TP53 mutations. Polymorphisms in the TP53 pathway (TP53 PEX4 at codon 72, rs1042522; MDM2 SNP309, rs2279744) have modifier effects on germline TP53 mutations that may account for the individual and familial diversity of tumour patterns.
Methods and results: Four polymorphisms were analysed in a series of 135 Brazilian LFS/LFL cancer patients (32 TP53 mutation carriers and 103 wild-type subjects). We report for the first time that another polymorphism in the TP53 gene, TP53 PIN3 (rs17878362), has a strong modifier effect on germline TP53 mutations. This polymorphism, which consists of a 16 bp duplication in intron 3 (A1, non-duplicated allele; A2, duplicated allele), is associated with a difference of 19.0 years in the mean age at the first diagnosis in TP53 mutation carriers (n = 25, A1A1: 28.0 years; n = 7, A1A2: 47.0 years; p = 0.01). In addition, cancer occurrence before the age of 35 years is exclusively observed in A1A1 homozygotes. In this series, the effect of TP53 PEX4 and MDM2 SNP309 on age at diagnosis was similar to the one reported in other series and was smaller than the one of TP53 PIN3 (TP53 PIN3: difference of 19.0 years; TP53 PEX4: 8.3 years; MDM2 SNP309: 12.5 years).
Conclusion: These results suggest that TP53 PIN3 is another polymorphism in the TP53 pathway that may have a modifier effect on germline TP53 mutations and may contribute to the phenotypic diversity of germline TP53 mutations associated with LFS/LFL patients.
Statistics from Altmetric.com
The Li–Fraumeni syndrome (LFS; MIM 151623) is a rare autosomal dominant disorder of predisposition to multiple early onset cancers, caused by a germline mutation in the TP53 tumour suppressor gene (MIM 191170).1 2 LFS is associated with heterogeneous tumour patterns, the most frequent tumours being soft tissue and bone sarcomas, breast adenocarcinoma, brain glioma, and adrenocortical carcinoma.3 4 Families with incomplete features of LFS are referred to as having Li–Fraumeni-like syndrome (LFL), for which several clinical definitions have been proposed.5 6 7 The heterogeneous tumour patterns in LFS/LFL families may be explained in part by differences in TP53 mutation types and functional impacts.7 Lifestyle, genetic and environmental factors are also thought to play a major role in shaping these tumour patterns.
TP53 is highly polymorphic with 85 single nucleotide polymorphisms (SNPs) reported in the International Agency for Research on Cancer (IARC) TP53 Mutation Database (http://www-p53.iarc.fr).8 Two polymorphisms in the TP53 pathway, TP53 PEX4 and MDM2 SNP309, have been shown to have a consistent modifier effect on germline TP53 mutations.9 10 TP53 PEX4 (Polymorphism in Exon 4; rs1042522; G/C) leads to an arginine (R) to proline (P) amino acid substitution at codon 72 that induces differences in biochemical and functional properties.11 12 13 The presence of the R allele has been shown to reduce the age at cancer onset by about 12 years in germline TP53 mutation carriers.9 MDM2 SNP309 (rs2279744; T/G) is located in MDM2 promoter and affects a response element of the transcription factor SP1.10 SP1 activity is higher on promoter carrying a G allele than a T allele, leading to an increase of Mdm2 expression level and in the attenuation of the p53 pathway.10 An association of the G allele with early cancer diagnosis has been demonstrated in germline TP53 mutation carriers, as well as a cumulative effect of MDM2 SNP309 and TP53 PEX4.9 10 14
Among TP53 polymorphisms, the effects of the common TP53 PIN2 (Polymorphism in Intron 2; rs1642785; G/C),15 and TP53 PIN3 (Polymorphism in Intron 3; rs17878362; consisting of a 16 bp duplication, A1: non-duplicated, A2: duplicated),16 17 have not been investigated so far in relation with germline TP53 mutation. In this study, we have analysed whether TP53 PIN3 and TP53 PIN2, in addition to the well known TP53 PEX4 and MDM2 SNP309 polymorphisms, may have an impact on tumour patterns and age at diagnosis of first cancer in 135 Brazilian LFS/LFL subjects, who are carriers or not of a germline TP53 mutation.
Patients and methods
Patients and controls
LFS/LFL Brazilian patients and families were recruited from Cancer Risk Evaluation Programs in three Brazilian capitals: São Paulo (77 families), Porto Alegre (38 families), and Rio de Janeiro (three families). The families included were apparently unrelated and were selected because they matched at least one of the published LFS or LFL criteria.7 Of the 118 families, only six matched the strict LFS criteria of Li and Fraumeni (5.1% of the families), while the remaining 112 families matched at least one of the criteria proposed by Chompret, Birch and Eeles. Pedigree extending over five to six generations were reconstructed by clinical geneticists and annotated using the Progeny software (Progeny Software Inc, Wolfville, Nova Scotia, Canada). A total of 259 subjects were tested for germline TP53 mutations. To avoid bias due to over-representation of large pedigrees, only the proband and up to three affected relatives were included in this study, representing a total of 135 subjects. The tumour patterns and mutation status of 45 families have been previously reported.18 All subjects provided informed consent and the study was approved by the respective institutional ethics committees.
As reference, four control groups were constituted. The first one included 300 asymptomatic subjects (Brazilian reference). These subjects were part of a series of 750 participants of a community based breast cancer prevention programme in the area of Porto Alegre, Brazil, irrespective of familial history, as described elsewhere.19 The three other groups included Caucasian, African and Asian subjects of the HapMap set, each group containing 90 subjects (Garritano et al, 2009, unpublished data).
Mutation and polymorphism genotyping
DNA was isolated from peripheral blood with Qiagen DNA Extraction kit according to manufacturer instructions (QIAamp DNA blood Maxi Kit, Qiagen, Courtaboeuf, France). TP53 mutation (exons 2 to 11 with flanking splice junctions) and polymorphism analyses were carried out by direct sequencing (for detailed protocols see http://www-iarc.p53.fr). Primers and polymerase chain reaction (PCR) conditions using Taq Platinum (Invitrogen, Paisley, UK) are described in supplemental table 1. The sequencing reactions were performed using BigDye reagent (Applied Biosystems, Carlsbad, California, USA). MDM2 SNP309 was assessed using a TaqMan SNP genotyping assay (Proligo, St Louis, Missouri, USA) (probes FAM 5′-cccgcgccgcagc-3′ and HEX 5′-cccgcgccgaagc-3′; Primers 5′-ttcagggtaaaggtcacggg-3′ and 5′-tcaacctgcccactgaacc-3′).
To determine the haplotypes defined by the three TP53 polymorphisms, we developed a method based on the amplification refractory mutation system (ARMS).20 This method takes advantage of DNA mismatches that inhibits polymerase reaction under specific PCR conditions. We designed four primers (fig 1A,B): primers specific of the two TP53 PIN2 alleles (primer 1F hybridises TP53 PIN2 G allele and primer 2F the C allele); and primers specific of the two TP53 PEX4 alleles (primer 3R hybridises the TP53 PEX4 C allele (coding a P) and primer 4R the G allele (coding a R)). With this allele specific PCR, the haplotypes can be identified directly on an agarose gel (fig 1C). For example, the Y-13 sample is amplified with two combinations of primers, 1F-4R and 2F-3R, indicating a G-R haplotype and a C-P haplotype, respectively. The difference in TP53 PIN3 status can be determined by difference in the electrophoretic mobilities of the two bands, the “G-R” band migrating faster than the “C-P” band, consistent with the presence of a repeat of the 16 bp of TP53 PIN3 in the “C-P” allele. The presence of the expected haplotype was confirmed by bidirectional sequencing. Thus, the Y-13 sample contains two different alleles of TP53 gene: G-A1-R and C-A2-P.
Comparison of genotype and allelic distribution for each polymorphism between cases and controls was performed by χ2 test. Comparison of mean age and mean number of cancers were assessed by the non-parametric Mann–Whitney test. Statistical analyses (non-adjusted) were performed using the GraphPad Instat 3 Software (San Diego, California, USA).
Genetic susceptibility to cancer in relation with TP53 polymorphisms
A group of 300 Southern Brazilians from the general population was analysed as reference (termed Brazilian reference). Haplotyping using ARMS PCR showed strong linkage disequilibrium between TP53 PIN2 and TP53 PEX4 (fig 1D). These two SNPs were concordant in 95% of the subjects, making it impossible to distinguish their specific contribution within our sample set. Therefore, we have only considered TP53 PIN3 and TP53 PEX4, the latter being also a surrogate for TP53 PIN2.
The Brazilian reference was compared to Caucasian, African and Asian references (Garritano et al, 2009, unpublished data). The allelic profile of the Brazilian reference was statistically different from the three other population references (table 1). However, it was much closer to the Caucasian reference than to the others, in agreement with the notion that a large portion of the Southern Brazil population has a Caucasian origin.21 22
We first compared the distribution of TP53 polymorphisms between the Brazilian reference and LFS/LFL cancer patients, irrespective of their TP53 mutation status. The allelic frequency of each polymorphism was identical in familial cases and in the Brazilian reference (table 1). No significant difference was observed in the distribution of genotypes between the two groups for any of the three TP53 polymorphisms, TP53 PIN2, TP53 PIN3 and TP53 PEX4 (supplemental table 2). However, for TP53 PIN3, none of the familial cases was A2A2 homozygote and the Hardy–Weinberg equilibrium was not fulfilled (p<0.05).
Impact of MDM2 and TP53 polymorphisms on age at first cancer diagnosis
We next analysed the age at first cancer diagnosis in relation to TP53 mutation status and TP53 or MDM2 SNP309 polymorphisms, separately (table 2) or in combination (supplemental table 3). For MDM2 SNP309, no significant difference in genotype frequencies between mutation carriers and non-carriers was observed. Comparison between mean age at first diagnosis among TP53 mutation carriers revealed a difference of 12.5 years on average in relation with MDM2 SNP309 status, the earlier age of diagnosis being related to the presence of the G allele (age of 26.3 years vs 38.8 years; p = 0.06). This modifier effect is of borderline significance but its amplitude is compatible with earlier reports.9 10 14
For the TP53 PIN3 polymorphism, there was no significant difference in genotype frequencies between mutation carriers and non-carriers (table 2). However, the mean age at first diagnosis was significantly reduced in A1A1 TP53 mutation carriers compared to other groups (A1A2 mutation and non-mutation carriers). The average age difference between A1A1 and A1A2 subjects in mutation carriers was 19.0 years (age of 28.0 years vs 47.0 years; p = 0.01). Thus, TP53 PIN3 is a strong modifier of germline TP53 mutation phenotype.
For TP53 PEX4, a significant difference of genotype distribution was observed with an excess of the R allele in mutation carriers as compared to non-carriers, suggesting that the mutation preferentially occurs on the R allele (p = 0.01). No patient carrying a germline TP53 mutation and being homozygote for the rare allele (PP) was observed (table 2). A difference of 8.3 years in the age at cancer diagnosis was detected in mutation carriers between RR homozygotes and RP heterozygotes, but this difference was not significant (age of 30.5 years vs 38.8 years; p = 0.22).
Given the linkage disequilibrium between TP53 PIN3 and TP53 PEX4, we examined the combined effects of these two polymorphisms in LFS/LFL patients (supplemental table 3). Although no statistics could be applied due to the small sample size, we observed that TP53 PEX4 appeared to have no impact on mean (SD) age at first cancer diagnosis in A1A1 mutation carriers (age of 29.0 (18.1) years in n = 20 RR vs age of 29.8 (13.7) years in n = 4 RP subjects), suggesting that the 19.0 years difference observed between TP53 PIN3 A1A1 and A1A2 mutation carriers was mostly due to the effect of TP53 PIN3. The combined analysis of MDM2 SNP309 and TP53 PIN3 showed that there was no difference in A1A1 mutation carriers carrying either MDM2 SNP309 TT or TG/GG genotype, suggesting a non-cumulative effect of TP53 PIN3 and MDM2 SNP309 (age of 32.0 (13.6) years in n = 9 TT vs age of 25.1 (17.6) in n = 14 TG and of 30.5 (41.7) in n = 2 GG subjects).
Impact of MDM2 SNP309 and TP53 PIN3 on age at first diagnosis according to tumour type
Analysis of cancer accrual in these LFS/LFL families suggested that the effect of MDM2 SNP309 was not consistent throughout life. Indeed, fig 2A shows that the difference in cancer accrual was more important before the age of 25, and that the accrual curves tended to converge in subjects aged over 50. Since the risk of specific cancer varies with age in Li–Fraumeni subjects,7 this effect may reflect a different impact of MDM2 SNP309 on types of cancer that occur in childhood and adolescence as compared to those occurring later in life. To assess this hypothesis further, we analysed the effect of MDM2 SNP309 on the risk of two common LFS/LFL diagnoses, soft tissue sarcomas (STS), a typical childhood and adolescence tumour in LFS/LFL, and breast carcinomas, occurring at all ages from young adulthood (table 3). For breast cancer, MDM2 SNP309 had no significant effect on age at first cancer diagnosis in either TP53 mutation carriers or non-carriers. In contrast, an important and significant difference of 34.3 years was detected in TP53 mutation carriers for STS, the presence of one G allele being associated with early onset cancer (age of 9.7 years vs 44.0 years; p = 0.02). This observation indicates that the modifier effect of MDM2 SNP309 was particularly marked for STS.
The effect observed with TP53 PIN3 was different from the one of MDM2 SNP309. Among TP53 mutation carriers, only TP53 PIN3 A1A1 subjects developed cancers before the age of 35 years (fig 2B). In A1A2 TP53 mutation carriers, the earliest cancers detected were melanoma and adrenocortical carcinoma in two patients aged 35. These results suggest that only A1A1 homozygotes had predisposition to early onset cancers. Moreover, the TP53 PIN3 A1A1 non-mutation carriers developed more cancers than the A1A2 non-carriers (on average 1.6 cancers per subject in A1A1 vs 1.1 cancers in A1A2; p = 0.02), suggesting that the A1A1 genotype is associated with increased risk of multiple cancers in LFS/LFL even in the absence of a detected TP53 mutation (supplemental table 4). A1A1 homozygotes also developed their first STS earlier than A1A2 heterozygotes in both TP53 mutation carriers and non-carriers, with an average difference of 32.3 years (p = 0.02) in the age at first cancer diagnosis in mutation carriers and of 9.4 years in non-carriers (table 3). In breast cancer, A1A1 TP53 mutation carriers had a tendency to develop their first cancer at a mean age of 39.2 years, 10.8 years earlier than A1A2 mutation carriers. However, this difference was not statistically significant. Although the power of these analyses is limited by the small number of cases, these results suggest that TP53 PIN3 has a modifier effect on TP53 mutations in both STS and breast cancers, with a stronger effect on the development of STS.
Effect of TP53 PIN3 in relation to TP53 mutation type
The subjects involved in this study have been recruited in Southern Brazil and many of them are carriers of a specific germline TP53 mutation, R337H, which is present in several families due to a founder effect (Garritano et al, 2009, unpublished data).18 19 In our series, R337H was present in 18 of the 32 mutation carriers. The mutations identified in other subjects were representative of the diversity of germline TP53 mutations described in the literature.18 Due to the high representation of R337H in this series, we cannot rule out that the effect of TP53 PIN3 as a modifier may be particularly marked for this mutant. However, the association of TP53 PIN3 A2 allele with late diagnosis was not restricted to R337H: all subjects, in whom first cancer was diagnosed before 35 years, were homozygotes for TP53 PIN3 A1 (supplemental table 5). In subjects with first cancers occurring after 35 years of age, the proportion of those with A2 alleles was identical among carriers of R337H and other mutations (R337H = 36.4% vs others = 37.5%). Thus, the modifier effects of TP53 PIN3 do not appear to be restricted to a particular TP53 mutation.
The TP53 gene is highly polymorphic and some of these polymorphisms have an impact on gene expression and/or on p53 protein function.12 13 23 In this study, we have analysed the impact of four common polymorphisms, MDM2 SNP309, TP53 PIN2, TP53 PIN3, and TP53 PEX4,10 11 15 16 on the mean age at first cancer diagnosis in members of LFS/LFL families. Our results demonstrate a strong modifier effect of TP53 PIN3 in TP53 mutation carriers. The A1 non-duplicated allele was associated with an average acceleration of 19.0 years in the mean age at first cancer diagnosis in TP53 mutation carriers. In contrast, TP53 PEX4 had a small and non-significant modifier effect of 8.3 years on age at cancer diagnosis, while MDM2 SNP309 was associated with a difference of 12.5 years of borderline significance (p = 0.06). Overall, these results confirm previous results for TP53 PEX4 and MDM2 SNP3099 10 and identify TP53 PIN3 as the strongest modifier of germline TP53 mutation effects reported to date. Moreover, in non-carriers of TP53 mutation, the A1 non-duplicated allele was associated with a modest increased risk of developing multiple cancers, a result that needs to be confirmed in future studies.
TP53 PEX4 has previously been shown by Bougeard et al to have a significant effect on age at cancer onset in LFS/LFL, with the presence of the R allele reducing the age at cancer onset by about 12.6 years (p<0.05).9 In the present study, we also found that the R allele was associated with a reduction in the age at first cancer of about 8 years, although this effect was not statistically significant (p = 0.22). The fact that (1) this effect is much smaller than the one observed with TP53 PIN3 (reduction of 8 years for TP53 PEX4 vs 19.0 years for TP53 PIN3), and that (2) combining the two polymorphisms did not appear to have a cumulative effect (age at diagnosis: 29.0 years in A1A1 RR and 29.8 years in A1A1 RP), suggests that the main modifier effect is related to TP53 PIN3 and that the association of TP53 PEX4 with this effect may be due to linkage disequilibrium with TP53 PIN3. This interpretation will need to be further addressed in studies of higher statistical power.
The functional basis of the impact of TP53 PIN3 remains to be elucidated. Gemignani et al have reported that this polymorphism had an impact on levels of p53 mRNA in lymphoblastoid cells.23 Recently, we found that this polymorphism is embedded into G-quadruplex structures (Marcel et al, 2009, unpublished data). These G-quadruplexes had an effect on the splicing of intron 2, thus regulating the production of an alternatively spliced p53 mRNA. How these effects on splicing modulate p53 function remains to be determined. In contrast with TP53 PIN3, TP53 PEX4 is a non-silent polymorphism that leads to the synthesis of variant p53 proteins that differ by one residue at codon 72 and show differences in their suppressive effects in vitro.12 13 Since these two polymorphisms may affect p53 function by different mechanisms, it will be important to assess whether they may have complementary effects in modifying the activities of p53.
With respect to MDM2 SNP309, our results confirm those reported by previous studies.9 10 14 The effect of the G allele in accelerating tumour onset is compatible with the notion that SNP309, located in MDM2 promoter, is involved in the control of Mdm2 expression and in the regulation of p53 protein levels by Mdm2 mediated degradation.24 25 Consistent with previous reports, cancer accrual in relation to MDM2 SNP309 shows a greater effect for early onset cancers than for late onset cancers. Bond et al reported that TG or GG TP53 mutation carriers developed STS 38.0 years before TT mutation carriers, and the difference was 10.0 years for breast cancer.10 In our series, the effect of MDM2 SNP309 on age of breast cancer diagnosis was small and non-significant (on average 2.1 years of difference in mean age at first breast diagnosis). In contrast, in STS, a significant difference of 34.6 years was observed. A similar trend is observed with TP53 PIN3, with a non-significant difference of about 10.8 years in age at onset for breast cancer, and a highly significant difference of 32.3 years for STS. These differences suggest that these modifier effects may depend either on age, tissue type, or both.
The TP53 PIN3 polymorphism consists of a 16 bp duplication in intron 3. Study in patients with Li–Fraumeni and related syndromes has identified this polymorphism as a strong modifier of germline TP53 mutations.
In TP53 mutation carriers, the major non-duplicated allele (A1) was associated with an average acceleration of 19 years in the mean age at first cancer diagnosis.
The effect of TP53 PIN3 may contribute to the phenotypic diversity of Li–Fraumeni and related syndromes and suggests that the polymorphism of intron 3 has an impact on p53 function.
Several studies have addressed the possible impact of polymorphisms in TP53 or MDM2 in genetic susceptibility to sporadic cancer. Most of these studies have focused on TP53 PEX4. They have shown inconsistent results, depending upon tumour type and study design.26 Taking into account the modifier effects observed in germline TP53 mutation carriers, it is possible that these polymorphisms may have different impacts in tumours with or without somatic TP53 mutations. Further studies are needed to assess better the relationships between TP53 haplotypes and occurrence of somatic mutations in TP53.
In conclusion, the identification of polymorphisms acting as modifiers of germline TP53 mutations may assist in the assessment of individual cancer risk in LFS/LFL families. It may also play an important role in the delineation of cancer screening and intervention guidelines in these patients. In addition, as the A2 allele of TP53 PIN3 seems to have a protective effect against the risk of developing childhood and adolescent cancer, these observations may help to define appropriate management protocols, taking into account the risk of developing cancer either earlier or later in life.
The authors thank the physicians, nurses, and staff of the oncogenetics departments from the Division of Genetics at Instituto Nacional do Câncer (INCA), from the Medical Genetics Service at Hospital de Clínicas de Porto Alegre, and from Hospital do Câncer A.C. Camargo and Núcleo Mama Porto Alegre, Brazil for their help with recruitment of individuals included in this study.
▸ Additional tables are published online only at http://jmg.bmj.com/content/vol46/issue11
VM and EIP contributed equally to this work
Funding VM is supported by la Ligue Nationale Contre le Cancer (French association). This work is supported by EU FP6 program MUTp53, by la Ligue du Rhone Contre le Cancer and in part by grants from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) Edital MCT-CNPq/MS-SCTIE-DECIT/CT-Saúde (grant 400949/2005–9), Brazil; Susan G Komen for the Cure (grant POP0403033); FAPERGS (grant PSUS 2006); and Fundo de Incentivo à Pesquisa do Hospital de Clínicas de Porto Alegre FIPE/HCPA (grant 05–182), Brazil.
Competing interests None.
Patient consent Obtained.
Provenance and Peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.