- Email: [email protected]
TP53 R72P and MDM2 SNP309 polymorphisms in modification of childhood acute lymphoblastic leukemia susceptibility
Cancer Genetics and Cytogenetics 195 (2009) 31e36
TP53 R72P and MDM2 SNP309 polymorphisms in modification of childhood acute lymphoblastic leukemia susceptibility Thuy N. Do, Esma Ucisik-Akkaya, Charronne F. Davis, Brittany A. Morrison, M. Tevfik Dorak* Genomic Immunoepidemiology Laboratory, HUMIGEN LLC, Institute for Genetic Immunology, 2439 Kuser Road, Hamilton, NJ 08690-3303 Received 5 May 2009; accepted 25 May 2009
Abstract
Genomic and immunologic surveillance mechanisms are crucial in protection from cancer. The tumor suppressor protein p53, encoded by TP53, is a major regulator of genome surveillance. Among the natural sequence variants of TP53, rs1042522 (R72P) modifies the risk for solid tumors. To investigate its relevance in childhood acute lymphoblastic leukemia (ALL) susceptibility, we genotyped 114 cases and 414 newborn controls from Wales (UK) for polymorphisms in TP53 (R72P), its negative regulator MDM2 (single-nucleotide polymorphism SNP309, rs2279744), and selected HLA complex genes whose products interact with TP53. TP53 R72P showed a risk association with gene dosage effect (P50.002) resulting in a strong association of homozygous genotype (OR52.9, 95% CI51.5e5.6) and no sex effect. SNP309 did not show any association with primary susceptibility to childhood ALL, even after stratification by sex. However, females with SNP309 minor allele had earlier onset of childhood ALL (median age at diagnosis was 36 months in females, but 60 months in males; P50.002). The HLA complex genes did not show any statistically significant interaction with R72P. We have therefore identified TP53 R72P as a possible risk modifier for childhood ALL and the association of MDM2 with age at onset with sex effect suggests prenatal hormonal programming of childhood ALL susceptibility. Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction The p53 protein (encoded by TP53), a major tumor suppressor molecule, regulates target genes that induce cell cycle arrest, apoptosis, senescence, DNA repair, autophagy, and changes in metabolism. Somatic mutations of TP53 frequently occur in tumors, and germline mutations are found in cancer-prone families with LieFraumeni syndrome. These high-penetrance mutations are rare but strong markers of cancer predisposition. TP53 also has common polymorphisms, and some have functional effects. Some of these common but low-penetrance sequence variants have been shown to modify cancer susceptibility. The most commonly examined variant is the codon 72 polymorphism Arg72Pro (R72P) [1e4], and a meta-analysis confirmed its involvement in general cancer susceptibility [5]. The p53 protein does not act in isolation and it activates downstream targets. Its own activity is regulated by other molecules, such as the human homolog of the mouse double minute 2 (MDM2) and the death-domain associated * Corresponding author. Tel.: (609) 570-1032; fax: (609) 570-1039. E-mail address: [email protected] (M.T. Dorak). 0165-4608/09/$ e see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.cancergencyto.2009.05.009
protein (death-associated protein 6) DAXX. The MDM2 promoter region single-nucleotide polymorphism (SNP) known as SNP309 modifies cancer susceptibility with sex effect due to transcriptional regulation of MDM2 expression by estrogen [6,7]. DAXX, which is the other negative regulator of TP53 activity [8,9], is encoded within the extended HLA class II region [10]. It is important to examine associations with DAXX, because it can be a confounder of the previously reported associations with HLA class II genes (DR/DQ/DP) in malignancies, in particular the HLA-DRB4 [11] and HLA-DPB1 associations [12,13] in childhood acute lymphoblastic leukemia (ALL). Most solid tumors have been examined for the role of TP53 polymorphisms in susceptibility [1e4], but leukemias have not received much attention. Only chronic lymphoid leukemia (CLL) has been studied, with negative results for an association between TP53 R72P polymorphism and primary susceptibility [14]. Likewise, MDM2 SNP309 did not show an association in CLL [15], although there is an association with clinical outcome [16]. No study has examined TP53 polymorphisms in childhood leukemia susceptibility, but one study found an effect of MDM2 SNP309 on the age at diagnosis in childhood ALL [17]. Childhood
T.N. Do et al. / Cancer Genetics and Cytogenetics 195 (2009) 31e36
32
leukemia shows sex effect in susceptibility, and some of the reported genetic associations are also sex-specific (including the HLA class II associations) [11,18]. Childhood ALL shows a consistent association with miscarriages [19], and the TP53 polymorphism R72P is implicated in implantation failure [20] and in recurrent miscarriages [21]. Findings from a study of sex-specific genotype frequencies in newborns has implicated homozygosity in the HLA class II region in sex-specific prenatal selection [22]. These features of childhood ALL make it an interesting target for the study of TP53 pathway genes, especially those encoded within the HLA complex, for their role in ALL development and in reproductive failure. Our objectives, therefore, were to study TP53 R72P, MDM2 SNP309, and selected polymorphisms of TP53 pathway and apoptosis-related genes located within the HLA complex (Table 1) in a single-center caseecontrol study. Because our controls were healthy newborns, we also examined genotype frequencies after stratification by sex to seek evidence for the involvement of these genes in sexspecific prenatal selection.
2. Subjects and methods 2.1. Case and control groups Samples were analyzed from 114 cases of childhood (<14 years) ALL consecutively diagnosed from 1988 to 1999 in South Wales (United Kingdom) and originally used to describe the HLA-DRB4 and C282Y associations [11,18]. The controls comprised 414 anonymously collected cord blood samples also from South Wales. The control samples were collected in two hospitals in Cardiff between 1997 and 1998 without any selection other than that all were from naturally born term newborns. Further
characteristics of these samples are described elsewhere [11,22]. The samples were used with the approval of the South Wales Research Ethics Committee. 2.2. Selection of polymorphism The primary aim was to examine whether the TP53 exon 4 polymorphism Arg72Pro (R72P rs1042522) has any role in modification of childhood ALL susceptibility. (Singlenucleotide polymorphisms are identified according to the SNP database, available at http://www.ncbi.nlm.nih.gov/ sites/entrez?db5snp). We included the MDM2 SNP309 (rs1042522) because of its known interaction with TP53. Because this caseecontrol group showed an association with the HLA class II genotypes, and because the extended HLA class II region contains another negative regulator of TP53 (DAXX ), three haplotypetagging SNPs were selected from HapMap (http://www. hapmap.org). The HLA complex contains other genes encoding proteins taking part in the apoptotic signaling pathway. Among these, BAT3 [23], LTA (previously TNFB) [24,25], DDR1 (CD167) [26], and IER3 [27,28] were also represented by one to four SNPs selected either for their known functional role or as haplotype tagging SNPs. The genes and SNPs evaluated are characterized in Table 1. The HLA complex or the chromosomal segments where TP53 (17p13.1) and MDM2 (12q14.3~q15) located are not known to be frequent sites of chromosomal changes in childhood ALL [29]. 2.3. Genotypings All genotypings except MDM2 rs2279744 were achieved with ABI TaqMan allelic discrimination assays (Applied Biosystems, Foster City, CA) on Stratagene (La Jolla, CA) Mx3000 real-time polymerase chain reaction
Table 1 Genes and their polymorphisms included in the present study of childhood acute lymphoblastic leukemia Minor allele freqa Gene
SNP location
dbSNP number
Chromosome:nucleotide position
Variation
Controls
HapMap
TP53 MDM2 DAXX DAXX DAXX BAT3 BAT3 BAT3 LTA DDR1 DDR1 DDR1 DDR1 IER3
R72P in exon 5 intron 1 intron 1 Y379Y in exon 4 intron 4 intron 6 intron 12 30 flanking intron 1 V599 V in exon 15 intron 3 50 flanking 50 flanking 30 UTR
rs1042522 rs2279744 rs2073524 rs1059231 rs2239839 rs805303 rs2077102 rs2736155 rs909253 rs1049623 rs1264323 rs1264327 rs1264328 rs10947089
chr17:7520197 chr12:67488847 chr6:33359235 chr6:33356957 chr6:33356761 chr6:31724345 chr6:31719819 chr6:31713178 chr6:31648292 chr6:30972808 chr6:30963886 chr6:30958561 chr6:30958121 chr6:30818114
GOC TOG TOA TOC GOT COT GOT COG TOC TOC GOA COT AOG AOG
0.249 0.346 0.497 0.299 0.317 0.386 0.181 0.449 0.377 0.396 0.395 0.349 0.391 0.033
0.233 db 0.475 0.258 0.258 0.347 0.178 0.433 0.358 0.425 0.425 0.321 0.424 0.025
Abbreviations: SNP, single-nucleotide polymorphism. a Minor allele frequency in controls of the present study (n5414) and in HapMap European sample. b No HapMap data for this SNP.
T.N. Do et al. / Cancer Genetics and Cytogenetics 195 (2009) 31e36
(PCR) instruments using standard allelic discrimination assay conditions. The ABI TaqMan assay for the MDM2 polymorphism did not yield satisfactory results (extreme deviation of HardyeWeinberg equilibrium), so a PCRerestriction fragment length polymorphism analysis was used [7]. The primer sequences were 50 -CGGGAGTTCAGGGTAAAGGT-30 and 50 -AGCAAGTCGGTGCTTACCTG-30 . MspAI, which cuts when the variant allele G is present, was used to assign genotypes. 2.4. Haplotype analysis The Haploview v4.0 software package (http://www. broad.mit.edu/mpg/haploview) was used to construct haplotypes and to evaluate allelicehaplotypic associations. 2.5. Data analysis Data analysis was performed with Stata/IC v.10 software (StataCorp, College Station, TX). We used the Stata command genhwcci, which tests HardyeWeinberg equilibrium for genotypic counts of cases, under the assumption that the genotypic counts of controls are under HardyeWeinberg equilibrium [30]. Loci deviating from HardyeWeinberg equilibrium were evaluated using a genotype-based trend test (additive model), which does not rely on HardyeWeinberg equilibrium [31] and reveals associations that depend additively upon the minor allele. The additive model was assessed by unconditional logistic regression on Stata. This test compares the heterozygotic and then homozygotic frequencies with wild-type homozygotic frequency, and yields an odds ratio (OR) per allele for stepwise change per the number of minor alleles in the genotype, and a Ptrend value. All statistically significant associations and comparisons that yielded odds ratios of !0.5 or O2.0 were assessed in a multivariate logistic regression model for independence. No conventional risk marker information other than sex was available for inclusion in the multivariate models. Interactions were assessed by logistic regression and also by analysis stratified by genotypes. KruskaleWallis test was used to assess distribution differences for age at diagnosis between male and female cases. All P-values presented are two-sided. No correction for the P-values was attempted because of the exploratory nature of the study and marginal statistical power present.
3. Results 3.1. Characteristics of the caseecontrol group We were interested in analysis after stratification by sex because of the sex effect in childhood ALL development (males are at a higher risk) and the sex-specificity of some genetic associations previously reported [11,18]. When stratified by sex, there were 62 male cases and 200 male
33
controls, and the numbers were 52 and 214 for females. All cases were under 15 years old, and all controls were newborns. Although we had information on the clinical subtypes of ALL, we did not try subtype-specific analysis because of small sample numbers. 3.2. HardyeWeinberg equilibrium With the exception of the three DAXX SNPs, genotype distributions of all SNPs in controls were in HardyeWeinberg equilibrium. We nonetheless analyzed the DAXX results, despite HardyeWeinberg disequilibrium (P<0.05) in the three controls; we found only an association as haplotypic homozygosity. Thus, DAXX results need to be interpreted with additional caution. Minor allele frequencies of the SNPs in this Welsh population was similar to those observed in the HapMap European sample (Table 1). 3.3. Univariate analysis All SNPs were assessed by additive model checking in the whole group and in males and females separately. In this analysis, only TP53 R72P and two BAT3 SNPs (rs805303 and rs2077102) showed statistically significant risk associations (Table 2). The two BAT3 SNPs were in linkage disequilibrium, and their associations were not independent. The intron 6 SNP rs805303 minor allele frequency was greater, and we used this SNP in further analysis. In multivariate analysis, TP53 and BAT3 associations were independent and retained individual statistical significance. Notably, MDM2 did not show any association with childhood ALL, even after stratification by sex. The only conventional risk marker for childhood ALL for which we had data available was sex. Adjustment for sex did not cause any change in the TP53 association (OR per allele51.66; Ptrend50.002). Likewise, the BAT3 SNP association remained the same after adjustment for sex (OR per allele50.68; Ptrend50.02). Thus, neither univariate association was confounded by sex. 3.4. Haplotype homozygosity Because we had genotyped multiple SNPs in BAT3 and DDR1 genes, we also examined haplotype frequencies in these genes. One DAXX haplotype consisting of minor alleles of rs2239839 and rs1059231 and the major allele of rs2073524 conferred increased risk in homozygous form (OR52.45; 95% CI51.22e4.91; P50.01). The DAXX haplotype homozygosity rates did not differ between the sexes, and the DAXX association was independent from the previously found DRB4 association (in the same sample [11]). 3.5. Interaction tests None of the analyses testing for interaction between TP53 and BAT3 SNPs rs805303 and rs2077102 or MDM2 SNP rs2279744 yielded a statistically significant result.
T.N. Do et al. / Cancer Genetics and Cytogenetics 195 (2009) 31e36
34
Table 2 Univariate analysis of associations between each SNP and childhood acute lymphoblastic leukemia Wild type
Heterozygotic a
Ptrend, additive model
Homozygotic a
a
Gene
SNP
Ref
Case vs. Ctrl
OR (95% CI)
Case vs. Ctrl
OR (95% CI)
Case vs. Ctrl
OR (95% CI)
P-value
TP53 MDM2 DAXX DAXX DAXX BAT3 BAT3 BAT3 LTA DDR1 DDR1 DDR1 DDR1 IER3
rs1042522 rs2279744 rs2073524 rs1059231 rs2239839 rs805303 rs2077102 rs2736155 rs909253 rs1049623 rs1264323 rs1264327 rs1264328 rs10947089
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.442 0.400 0.318 0.518 0.505 0.495 0.789 0.229 0.427 0.440 0.429 0.427 0.433 0.952
1.36 1.36 0.94 0.87 0.79 0.63 0.52 1.63 1.03 0.69 0.68 0.99 0.67 0.18
0.394 0.486 0.327 0.299 0.299 0.410 0.183 0.514 0.479 0.318 0.343 0.408 0.340 0.012
3.31 0.89 0.99 1.38 1.48 0.50 0.77 1.64 0.57 1.04 1.20 1.24 1.21 d
0.163 0.114 0.155 0.196 0.196 0.095 0.029 0.257 0.094 0.210 0.229 0.165 0.227 0.036
1.67 1.05 0.99 1.08 1.12 0.68 0.62 1.28 0.83 0.96 1.02 1.08 1.02 1.23
0.002 0.78 0.97 0.61 0.47 0.02 0.04 0.10 0.23 0.78 0.89 0.61 0.88 0.61
vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs.
0.566 0.452 0.309 0.514 0.497 0.372 0.670 0.326 0.403 0.383 0.381 0.439 0.385 0.934
(0.85e2.17) (0.86e2.16) (0.56e1.58) (0.54e1.39) (0.49e1.30) (0.40e1.00) (0.30e0.90) (0.96e2.76) (0.67e1.59) (0.42e1.14) (0.42e1.10) (0.62e1.59) (0.40e1.12) (0.02e1.34)
vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs.
0.371 0.403 0.375 0.372 0.372 0.485 0.299 0.451 0.440 0.441 0.450 0.424 0.449 0.066
(1.66e6.62) (0.44e1.80) (0.57e1.71) (0.73e6.62) (0.81e2.68) (0.24e1.05) (0.21e2.82) (0.89e3.03) (0.28e1.15) (0.57e1.89) (0.67e2.12) (0.65e2.34) (0.66e2.22)
vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs.
0.063 0.145 0.111 0.131 0.131 0.143 0.032 0.223 0.157 0.176 0.170 0.137 0.166 0
(1.21e2.30) (0.77e1.42) (0.76e1.31) (0.80e1.47) (0.83e1.50) (0.49e0.95) (0.39e0.99) (0.95e1.72) (0.62e1.13) (0.71e1.30) (0.76e1.37) (0.80e1.47) (0.75e1.40) (0.55e2.72)
Abbreviations: CI, confidence interval; OR, odds ratio; SNP, single-nucleotide polymorphism. a Frequencies for cases (n5114) vs. controls (n5414).
DAXX haplotype homozygosity was the only other statistically significant association that appeared to modify the TP53 association. The TP53 risk association was evident when it co-occurred with DAXX haplotype homozygosity (OR per allele51.82; 95% CI51.28e2.58) and not in the rest of the group (OR50.88; P50.80), although the numbers were so small for the latter comparison and statistically, the interaction did not reach significance (P50.17). When individual SNPs were examined, only DAXX 50 end htSNP rs2073524 showed a suggestive interaction with TP53 R72P between homozygous genotypes (P50.05). The risk association with TP53 R72P got stronger in the presence of the DAXX variant allele (OR per allele51.49 (95% CI51.19e2.69), P50.005). Further analysis suggested that this interaction was female-specific: OR per allele52.41 in females (P50.004) and 1.36 in males (P50.30). Given the sex-specificity of the biological interaction between MDM2 and TP53 in other cancers, we examined statistical interaction in each sex separately. The risk conferred by TP53 R72P homozygosity was greater for girls when they had the wild-type MDM2 genotype (OR57.1; P50.02), as opposed to the variant-positive genotype (OR51.4). This interaction did not reach statistical significance (P50.14). In summary, the only statistically significant interaction concerned the TP53 R72P, which also showed a significant main effect. The sample size was not sufficient for a thorough analysis of interactions, and HardyeWeinberg disequilibrium at the three DAXX SNPs also precluded obtaining a conclusive result. 3.6. MDM2 and age at diagnosis of childhood ALL MDM2 SNP309 has been reported to influence the age at diagnosis of childhood ALL [17]. We examined the same in our group of childhood ALL cases. Median ages were
similar in the whole group and in males for each genotype (Table 3). In females, there was a statistically nonsignificant trend toward earlier age at diagnosis with an increasing number of minor allele (59, 44, 32 months) (KruskaleWallis P50.13 for the difference in age at diagnosis distributions between wild-type homozygotic and minor alleleepositive female cases). In females, the median age at diagnosis was 38 months. Of the 24 cases below this age, 15 cases (62.5%) had the MDM2 variant allele, and of 24 cases above this age, 10 cases (41.7%) had the MDM2 variant allele (P50.24). The previous study did not report the results by sex [17]. The sex-specificity of the MDM2 association with age at diagnosis remains a possibility to be tested in a larger study. 3.7. Sex-specific genotype frequencies in newborns We did not observe any difference in R72P genotype frequencies between male and female newborns. Minor allele positivity was 42.0% in females and 45.0% in males. Homozygosity rates did not differ (6.3% in both).
4. Discussion The TP53-encoded molecule is called the guardian of the genome because of its crucial roles in genome surveillance. Genome and immune surveillance are the major mechanisms for protection from cancer development. Rare, high-penetrance mutations in the TP53 gene are known to cause hereditary forms of cancers. Common, low-penetrance polymorphisms also modify cancer risk, but to a lesser extent. In childhood lymphoid malignancies, TP53 mutations are rare and there is no information on the low-penetrance TP53 polymorphisms. Here, we present our findings on association of the well-known TP53 polymorphism R72P with childhood ALL, which revealed a risk
T.N. Do et al. / Cancer Genetics and Cytogenetics 195 (2009) 31e36
35
Table 3 Median age at diagnosis of childhood acute lymphoblastic leukemia by MDM2 SNP309 genotype Age, mo (sample size)
Total Male Female P-valuea a
TT, wild type
GT, heterozygosity
GG, minor allele homozygosity
GT þ GG, minor allele positivity
68 (n542) 78 (n519) 59 (n523) 1.0
56 (n551) 66 (n531) 44 (n520) 0.008
66 (n512) 77 (n57) 32 (n55) 0.16
52 (n563) 60 (n538) 36 (n525) 0.002
The P-value is for age distribution differences between male and female cases (KruskaleWallis test).
association of R72P with gene dosage effect. The association did not show sex-specificity. The R72P association was not significantly modified by polymorphisms within the MDM1 and DAXX genes (negative regulators of TP53), although some suggestive findings were present. In a prospective cohort study designed to examine the association of R72P with longevity, among all cancers, only hematopoietic cancers showed an association with R72P [32,33]. The present study is, to our knowledge, the only one to investigate an association specifically with leukemia, and our positive results further suggest the involvement of this low-penetrance common TP53 variant in leukemia development. The R72P polymorphism (rs1042522) codes for a conservative missense amino acid substitution, and also has a role in splicing regulation. R72P affects cell death via apoptosis by altering the transcription of TP53 target genes and modifying the apoptotic potential of cells [34]. The present findings suggest that leukemia may be more sensitive to alterations in the apoptotic pathway mediated by TP53 than are solid tumors. At present, there are limited data on the association of R72P with leukemia. In CLL, it does not affect primary susceptibility [14]. MDM2 SNP309 does not modify susceptibility CLL, either [15]. On the other hand, MDM2 SNP309 minor allele deteriorates survival of CLL [16]. In childhood ALL, functional TP53 mutations are rare in initial presentation, but are more common at relapse [35]. Similarly, MDM2 also plays a greater role in childhood ALL relapse [36]. Here we have presented some preliminary data that the common low-penetrance variant in the TP53 gene R72P increases the risk for childhood ALL. It has not been examined whether it increases the risk for relapse. Given its frequency (it is carried by O40% of Europeans), this polymorphism may prove to be a common marker for clinical risk stratification in childhood ALL. Children with MDM2 SNP309 minor allele homozygosity (genotype GG) develop childhood ALL at a younger age [17]. This is similar to the effect of this polymorphism on patients with LieFraumeni syndrome caused by TP53 mutations [37]. We did not observe a statistically significant association of SNP309 GG genotype with age at diagnosis in the overall group of childhood ALL cases, but median age of female cases with the GG genotype was 32 months, as opposed to 59 months for cases with the wild-type genotype (TT). In males, corresponding median ages were 77 and 78 months, respectively (Table 3). The association
suggested by this difference in female cases was confirmed by a statistically significant difference in the age at onset distributions in cases with wild-type MDM2 genotype (median age559 months) and MDM2 minor allele positive cases (median age536 months). Compared with male cases positive for the minor allele (median age560 months), the age distribution in female cases was significantly different (P50.002). A similar change in age at onset in females compared with male cases mediated by the MDM2 SNP309 minor allele has been reported also in adults with melanoma [38]. Our observation in childhood ALL raises an important question regarding the effect of estrogen on MDM2 SNP309 activity in childhood [7]. Because there is only negligible estrogen activity in childhood, the present finding, if confirmed, would provide an example of fetal programming of childhood ALL susceptibility. We have previously shown that, by exploring sex-specific genotype frequencies in newborns, evidence may be sought for markers of sex-specific prenatal selection [22]. Because the TP53 polymorphism R72P has been implicated in implantation failure [20] and recurrent miscarriages [21], we examined genotype frequencies in male and female newborns, but did not detect any differences. The HLA complex is also believed to be a modifier of reproductive success [39,40], but none of the HLA loci analyzed in the present study showed a sex-specific difference to account for the previously found male deficit in HLA class II homozygosity rates in the same group of newborns. Our results suggested that TP53 R72P polymorphism may be a modifier of risk for childhood ALL susceptibility and might interact with its regulators, MDM2 and DAXX, in females. Although biologically plausible, because of the limited sample size of the present study these results need to be replicated before considering the inclusion of R72P into a predictive genetic risk model for childhood ALL in the future. The intriguing hypothesis that fetal estrogen exposure may be involved in postnatal MDM2 activity to modify the age at diagnosis of childhood ALL may also be examined, using genetic epidemiology as a probe for disease biology.
Acknowledgments This work was funded intramurally by HUMIGEN LLC, Hamilton, NJ. The Authors are employees of HUMIGEN, The Institute for Genetic Immunology.
36
T.N. Do et al. / Cancer Genetics and Cytogenetics 195 (2009) 31e36
References [1] Dunning AM, Healey CS, Pharoah PD, Teare MD, Ponder BA, Easton DF. A systematic review of genetic polymorphisms and breast cancer risk. Cancer Epidemiol Biomarkers Prev 1999;8:843e54. [2] Fan R, Wu MT, Miller D, Wain JC, Kelsey KT, Wiencke JK, Christiani DC. The p53 codon 72 polymorphism and lung cancer risk. Cancer Epidemiol Biomarkers Prev 2000;9:1037e42. [3] Mechanic LE, Marrogi AJ, Welsh JA, Bowman ED, Khan MA, Enewold L, Zheng YL, Chanock S, Shields PG, Harris CC. Polymorphisms in XPD and TP53 and mutation in human lung cancer. Carcinogenesis 2005;26:597e604. [4] Koushik A, Tranah GJ, Ma J, Stampfer MJ, Sesso HD, Fuchs CS, Giovannucci EL, Hunter DJ. p53 Arg72Pro polymorphism and risk of colorectal adenoma and cancer. Int J Cancer 2006;119:1863e8. [5] van Heemst D, Mooijaart SP, Beekman M, Schreuder J, de Craen AJ, Brandt BW, Slagboom PE, Westendorp RG. Long Life Study Group. Variation in the human TP53 gene affects old age survival and cancer mortality. Exp Gerontol 2005;40:11e5. [6] Bond GL, Hu W, Bond EE, Robins H, Lutzker SG, Arva NC, Bargonetti J, Bartel F, Taubert H, Wuerl P, Onel K, Yip L, Hwang SJ, Strong LC, Lozano G, Levine AJ. A single nucleotide polymorphism in the MDM2 promoter attenuates the p53 tumor suppressor pathway and accelerates tumor formation in humans. Cell 2004;119:591e602. [7] Bond GL, Hirshfield KM, Kirchhoff T, Alexe G, Bond EE, Robins H, Bartel F, Taubert H, Wuerl P, Hait W, Toppmeyer D, Offit K, Levine AJ. MDM2 SNP309 accelerates tumor formation in a genderspecific and hormone-dependent manner. Cancer Res 2006;66:5104e10. [8] Zhao LY, Liu J, Sidhu GS, Niu Y, Liu Y, Wang R, Liao D. Negative regulation of p53 functions by Daxx and the involvement of MDM2. J Biol Chem 2004;279:50566e79. [9] Li Q, Wang X, Wu X, Rui Y, Liu W, Wang J, Wang X, Liou YC, Ye Z, Lin SC. Daxx cooperates with the Axin/HIPK2/p53 complex to induce cell death. Cancer Res 2007;67:66e74. [10] Herberg JA, Beck S, Trowsdale J. TAPASIN, DAXX, RGL2, HKE2 and four new genes (BING 1, 3 to 5) form a dense cluster at the centromeric end of the MHC. J Mol Biol 1998;277:839e57. [11] Dorak MT, Lawson T, Machulla HK, Darke C, Mills KI, Burnett AK. Unravelling an HLA-DR association in childhood acute lymphoblastic leukemia. Blood 1999;94:694e700. [12] Taylor GM, Dearden S, Ravetto P, Ayres M, Watson P, Hussain A, Greaves M, Alexander F, Eden OB. UKCCS Investigators (United Kingdom Childhood Cancer Study). Genetic susceptibility to childhood common acute lymphoblastic leukaemia is associated with polymorphic peptide-binding pocket profiles in HLA-DPB1*0201. Hum Mol Genet 2002;11:1585e97. [13] Taylor GM, Hussain A, Lightfoot TJ, Birch JM, Eden TO, Greaves MF. HLA-associated susceptibility to childhood B-cell precursor ALL: definition and role of HLA-DPB1 supertypes. Br J Cancer 2008;98:1125e31. [14] Lahiri O, Harris S, Packham G, Howell M. p53 pathway gene single nucleotide polymorphisms and chronic lymphocytic leukemia. Cancer Genet Cytogenet 2007;179:36e44. [15] Zenz T, Ha¨be S, Benner A, Kienle D, Do¨hner H, Stilgenbauer S. The MDM2 -309 T/G promoter single nucleotide polymorphism does not alter disease characteristics in chronic lymphocytic leukemia. Haematologica 2008;93:1111e3. [16] Gryshchenko I, Hofbauer S, Stoecher M, Daniel PT, Steurer M, Gaiger A, Eigenberger K, Greil R, Tinhofer I. MDM2 SNP309 is associated with poor outcome in B-cell chronic lymphocytic leukemia. J Clin Oncol 2008;26:2252e7. [17] Swinney RM, Hsu SC, Hirschman BA, Chen TT, Tomlinson GE. MDM2 promoter variation and age of diagnosis of acute lymphoblastic leukemia. Leukemia 2005;19:1996e8. [18] Dorak MT, Burnett AK, Worwood M, Sproul AM, Gibson BE. The C282Y mutation of HFE is another male-specific risk factor for childhood acute lymphoblastic leukemia. Blood 1999;94:3957.
[19] Dorak MT, Pearce MS, Hammal DM, McNally RJ, Parker L. Examination of gender effect in birth weight and miscarriage associations with childhood cancer (United Kingdom). Cancer Causes Control 2007;18:219e28. [20] Kay C, Jeyendran RS, Coulam CB. p53 tumour suppressor gene polymorphism is associated with recurrent implantation failure. Reprod Biomed Online 2006;13:492e6. [21] Pietrowski D, Bettendorf H, Riener EK, Keck C, Hefler LA, Huber JC, Tempfer C. Recurrent pregnancy failure is associated with a polymorphism in the p53 tumour suppressor gene. Hum Reprod 2005;20:848e51. [22] Dorak MT, Lawson T, Machulla HK, Mills KI, Burnett AK. Increased heterozygosity for MHC class II lineages in newborn males. Genes Immun 2002;3:263e9. [23] Desmots F, Russell HR, Lee Y, Boyd K, McKinnon PJ. The reaperbinding protein scythe modulates apoptosis and proliferation during mammalian development. Mol Cell Biol 2005;25:10329e37. [24] Higuchi M, Singh S, Aggarwal BB. Characterization of the apoptotic effects of human tumor necrosis factor: development of highly rapid and specific bioassay for human tumor necrosis factor and lymphotoxin using human target cells. J Immunol Methods 1995;178:173e81. [25] Rossi D, Gaidano G. Messengers of cell death: apoptotic signaling in health and disease. Haematologica 2003;88:212e8. [26] Ongusaha PP, Kim JI, Fang L, Wong TW, Yancopoulos GD, Aaronson SA, Lee SW. p53 induction and activation of DDR1 kinase counteract p53-mediated apoptosis and influence p53 regulation through a positive feedback loop. EMBO J 2003;22:1289e301. [27] Scha¨fer H, Trauzold A, Sebens T, Deppert W, Fo¨lsch UR, Schmidt WE. The proliferation-associated early response gene p22/PRG1 is a novel p53 target gene. Oncogene 1998;16:2479e87. [28] Wu MX. Roles of the stress-induced gene IEX-1 in regulation of cell death and oncogenesis. Apoptosis 2003;8:11e8. [29] Harrison CJ. The detection and significance of chromosomal abnormalities in childhood acute lymphoblastic leukaemia. Blood Rev 2001;15:49e59. [30] Cui J. HardyeWeinberg equilibrium test in caseecontrol studies. Stata Tech Bull 2000;57:17e9 [available at]. http://www.stata.com. [31] Sasieni PD. From genotypes to genes: doubling the sample size. Biometrics 1997;53:1253e61. [32] Bojesen SE, Nordestgaard BG. The common germline Arg72Pro polymorphism of p53 and increased longevity in humans. Cell Cycle 2008;7:158e63. [33] Ørsted DD, Bojesen SE, Tybjaerg-Hansen A, Nordestgaard BG. Tumor suppressor p53 Arg72Pro polymorphism and longevity, cancer survival, and risk of cancer in the general population. J Exp Med 2007;204:1295e301. [34] Imyanitov EN. Gene polymorphisms, apoptotic capacity and cancer risk. Hum Genet 2009;125:239e46. [35] Gump J, McGavran L, Wei Q, Hunger SP. Analysis of TP53 mutations in relapsed childhood acute lymphoblastic leukemia. J Pediatr Hematol Oncol 2001;23:416e9. [36] Zhou M, Gu L, Abshire TC, Homans A, Billett AL, Yeager AM, Findley HW. Incidence and prognostic significance of MDM2 oncoprotein overexpression in relapsed childhood acute lymphoblastic leukemia. Leukemia 2000;14:61e7. [37] Bougeard G, Baert-Desurmont S, Tournier I, Vasseur S, Martin C, Brugieres L, Chompret A, Bressac-de Paillerets B, StoppaLyonnet D, Bonaiti-Pellie C, Frebourg T. Impact of the MDM2 SNP309 and p53 Arg72Pro polymorphism on age of tumour onset in LieFraumeni syndrome. J Med Genet 2006;43:531e3. [38] Firoz EF, Warycha M, Zakrzewski J, Pollens D, Wang G, Shapiro R, Berman R, Pavlick A, Manga P, Ostrer H, Celebi JT, Kamino H, Darvishian F, Rolnitzky L, Goldberg JD, Osman I, Polsky D. Association of MDM2 SNP309, age of onset, and gender in cutaneous melanoma. Clin Cancer Res 2009;15:2573e80. [39] Lerner SP, Finch CE. The major histocompatibility complex and reproductive functions. Endocr Rev 1991;12:78e90. [40] Kostyu DD. HLA: fertile territory for developmental genes? Crit Rev Immunol 1994;14:29e59.
TP53 R72P and MDM2 SNP309 polymorphisms in modification of childhood acute lymphoblastic leukemia susceptibility Thuy N. Do, Esma Ucisik-Akkaya, Charronne F. Davis, Brittany A. Morrison, M. Tevfik Dorak* Genomic Immunoepidemiology Laboratory, HUMIGEN LLC, Institute for Genetic Immunology, 2439 Kuser Road, Hamilton, NJ 08690-3303 Received 5 May 2009; accepted 25 May 2009
Abstract
Genomic and immunologic surveillance mechanisms are crucial in protection from cancer. The tumor suppressor protein p53, encoded by TP53, is a major regulator of genome surveillance. Among the natural sequence variants of TP53, rs1042522 (R72P) modifies the risk for solid tumors. To investigate its relevance in childhood acute lymphoblastic leukemia (ALL) susceptibility, we genotyped 114 cases and 414 newborn controls from Wales (UK) for polymorphisms in TP53 (R72P), its negative regulator MDM2 (single-nucleotide polymorphism SNP309, rs2279744), and selected HLA complex genes whose products interact with TP53. TP53 R72P showed a risk association with gene dosage effect (P50.002) resulting in a strong association of homozygous genotype (OR52.9, 95% CI51.5e5.6) and no sex effect. SNP309 did not show any association with primary susceptibility to childhood ALL, even after stratification by sex. However, females with SNP309 minor allele had earlier onset of childhood ALL (median age at diagnosis was 36 months in females, but 60 months in males; P50.002). The HLA complex genes did not show any statistically significant interaction with R72P. We have therefore identified TP53 R72P as a possible risk modifier for childhood ALL and the association of MDM2 with age at onset with sex effect suggests prenatal hormonal programming of childhood ALL susceptibility. Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction The p53 protein (encoded by TP53), a major tumor suppressor molecule, regulates target genes that induce cell cycle arrest, apoptosis, senescence, DNA repair, autophagy, and changes in metabolism. Somatic mutations of TP53 frequently occur in tumors, and germline mutations are found in cancer-prone families with LieFraumeni syndrome. These high-penetrance mutations are rare but strong markers of cancer predisposition. TP53 also has common polymorphisms, and some have functional effects. Some of these common but low-penetrance sequence variants have been shown to modify cancer susceptibility. The most commonly examined variant is the codon 72 polymorphism Arg72Pro (R72P) [1e4], and a meta-analysis confirmed its involvement in general cancer susceptibility [5]. The p53 protein does not act in isolation and it activates downstream targets. Its own activity is regulated by other molecules, such as the human homolog of the mouse double minute 2 (MDM2) and the death-domain associated * Corresponding author. Tel.: (609) 570-1032; fax: (609) 570-1039. E-mail address: [email protected] (M.T. Dorak). 0165-4608/09/$ e see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.cancergencyto.2009.05.009
protein (death-associated protein 6) DAXX. The MDM2 promoter region single-nucleotide polymorphism (SNP) known as SNP309 modifies cancer susceptibility with sex effect due to transcriptional regulation of MDM2 expression by estrogen [6,7]. DAXX, which is the other negative regulator of TP53 activity [8,9], is encoded within the extended HLA class II region [10]. It is important to examine associations with DAXX, because it can be a confounder of the previously reported associations with HLA class II genes (DR/DQ/DP) in malignancies, in particular the HLA-DRB4 [11] and HLA-DPB1 associations [12,13] in childhood acute lymphoblastic leukemia (ALL). Most solid tumors have been examined for the role of TP53 polymorphisms in susceptibility [1e4], but leukemias have not received much attention. Only chronic lymphoid leukemia (CLL) has been studied, with negative results for an association between TP53 R72P polymorphism and primary susceptibility [14]. Likewise, MDM2 SNP309 did not show an association in CLL [15], although there is an association with clinical outcome [16]. No study has examined TP53 polymorphisms in childhood leukemia susceptibility, but one study found an effect of MDM2 SNP309 on the age at diagnosis in childhood ALL [17]. Childhood
T.N. Do et al. / Cancer Genetics and Cytogenetics 195 (2009) 31e36
32
leukemia shows sex effect in susceptibility, and some of the reported genetic associations are also sex-specific (including the HLA class II associations) [11,18]. Childhood ALL shows a consistent association with miscarriages [19], and the TP53 polymorphism R72P is implicated in implantation failure [20] and in recurrent miscarriages [21]. Findings from a study of sex-specific genotype frequencies in newborns has implicated homozygosity in the HLA class II region in sex-specific prenatal selection [22]. These features of childhood ALL make it an interesting target for the study of TP53 pathway genes, especially those encoded within the HLA complex, for their role in ALL development and in reproductive failure. Our objectives, therefore, were to study TP53 R72P, MDM2 SNP309, and selected polymorphisms of TP53 pathway and apoptosis-related genes located within the HLA complex (Table 1) in a single-center caseecontrol study. Because our controls were healthy newborns, we also examined genotype frequencies after stratification by sex to seek evidence for the involvement of these genes in sexspecific prenatal selection.
2. Subjects and methods 2.1. Case and control groups Samples were analyzed from 114 cases of childhood (<14 years) ALL consecutively diagnosed from 1988 to 1999 in South Wales (United Kingdom) and originally used to describe the HLA-DRB4 and C282Y associations [11,18]. The controls comprised 414 anonymously collected cord blood samples also from South Wales. The control samples were collected in two hospitals in Cardiff between 1997 and 1998 without any selection other than that all were from naturally born term newborns. Further
characteristics of these samples are described elsewhere [11,22]. The samples were used with the approval of the South Wales Research Ethics Committee. 2.2. Selection of polymorphism The primary aim was to examine whether the TP53 exon 4 polymorphism Arg72Pro (R72P rs1042522) has any role in modification of childhood ALL susceptibility. (Singlenucleotide polymorphisms are identified according to the SNP database, available at http://www.ncbi.nlm.nih.gov/ sites/entrez?db5snp). We included the MDM2 SNP309 (rs1042522) because of its known interaction with TP53. Because this caseecontrol group showed an association with the HLA class II genotypes, and because the extended HLA class II region contains another negative regulator of TP53 (DAXX ), three haplotypetagging SNPs were selected from HapMap (http://www. hapmap.org). The HLA complex contains other genes encoding proteins taking part in the apoptotic signaling pathway. Among these, BAT3 [23], LTA (previously TNFB) [24,25], DDR1 (CD167) [26], and IER3 [27,28] were also represented by one to four SNPs selected either for their known functional role or as haplotype tagging SNPs. The genes and SNPs evaluated are characterized in Table 1. The HLA complex or the chromosomal segments where TP53 (17p13.1) and MDM2 (12q14.3~q15) located are not known to be frequent sites of chromosomal changes in childhood ALL [29]. 2.3. Genotypings All genotypings except MDM2 rs2279744 were achieved with ABI TaqMan allelic discrimination assays (Applied Biosystems, Foster City, CA) on Stratagene (La Jolla, CA) Mx3000 real-time polymerase chain reaction
Table 1 Genes and their polymorphisms included in the present study of childhood acute lymphoblastic leukemia Minor allele freqa Gene
SNP location
dbSNP number
Chromosome:nucleotide position
Variation
Controls
HapMap
TP53 MDM2 DAXX DAXX DAXX BAT3 BAT3 BAT3 LTA DDR1 DDR1 DDR1 DDR1 IER3
R72P in exon 5 intron 1 intron 1 Y379Y in exon 4 intron 4 intron 6 intron 12 30 flanking intron 1 V599 V in exon 15 intron 3 50 flanking 50 flanking 30 UTR
rs1042522 rs2279744 rs2073524 rs1059231 rs2239839 rs805303 rs2077102 rs2736155 rs909253 rs1049623 rs1264323 rs1264327 rs1264328 rs10947089
chr17:7520197 chr12:67488847 chr6:33359235 chr6:33356957 chr6:33356761 chr6:31724345 chr6:31719819 chr6:31713178 chr6:31648292 chr6:30972808 chr6:30963886 chr6:30958561 chr6:30958121 chr6:30818114
GOC TOG TOA TOC GOT COT GOT COG TOC TOC GOA COT AOG AOG
0.249 0.346 0.497 0.299 0.317 0.386 0.181 0.449 0.377 0.396 0.395 0.349 0.391 0.033
0.233 db 0.475 0.258 0.258 0.347 0.178 0.433 0.358 0.425 0.425 0.321 0.424 0.025
Abbreviations: SNP, single-nucleotide polymorphism. a Minor allele frequency in controls of the present study (n5414) and in HapMap European sample. b No HapMap data for this SNP.
T.N. Do et al. / Cancer Genetics and Cytogenetics 195 (2009) 31e36
(PCR) instruments using standard allelic discrimination assay conditions. The ABI TaqMan assay for the MDM2 polymorphism did not yield satisfactory results (extreme deviation of HardyeWeinberg equilibrium), so a PCRerestriction fragment length polymorphism analysis was used [7]. The primer sequences were 50 -CGGGAGTTCAGGGTAAAGGT-30 and 50 -AGCAAGTCGGTGCTTACCTG-30 . MspAI, which cuts when the variant allele G is present, was used to assign genotypes. 2.4. Haplotype analysis The Haploview v4.0 software package (http://www. broad.mit.edu/mpg/haploview) was used to construct haplotypes and to evaluate allelicehaplotypic associations. 2.5. Data analysis Data analysis was performed with Stata/IC v.10 software (StataCorp, College Station, TX). We used the Stata command genhwcci, which tests HardyeWeinberg equilibrium for genotypic counts of cases, under the assumption that the genotypic counts of controls are under HardyeWeinberg equilibrium [30]. Loci deviating from HardyeWeinberg equilibrium were evaluated using a genotype-based trend test (additive model), which does not rely on HardyeWeinberg equilibrium [31] and reveals associations that depend additively upon the minor allele. The additive model was assessed by unconditional logistic regression on Stata. This test compares the heterozygotic and then homozygotic frequencies with wild-type homozygotic frequency, and yields an odds ratio (OR) per allele for stepwise change per the number of minor alleles in the genotype, and a Ptrend value. All statistically significant associations and comparisons that yielded odds ratios of !0.5 or O2.0 were assessed in a multivariate logistic regression model for independence. No conventional risk marker information other than sex was available for inclusion in the multivariate models. Interactions were assessed by logistic regression and also by analysis stratified by genotypes. KruskaleWallis test was used to assess distribution differences for age at diagnosis between male and female cases. All P-values presented are two-sided. No correction for the P-values was attempted because of the exploratory nature of the study and marginal statistical power present.
3. Results 3.1. Characteristics of the caseecontrol group We were interested in analysis after stratification by sex because of the sex effect in childhood ALL development (males are at a higher risk) and the sex-specificity of some genetic associations previously reported [11,18]. When stratified by sex, there were 62 male cases and 200 male
33
controls, and the numbers were 52 and 214 for females. All cases were under 15 years old, and all controls were newborns. Although we had information on the clinical subtypes of ALL, we did not try subtype-specific analysis because of small sample numbers. 3.2. HardyeWeinberg equilibrium With the exception of the three DAXX SNPs, genotype distributions of all SNPs in controls were in HardyeWeinberg equilibrium. We nonetheless analyzed the DAXX results, despite HardyeWeinberg disequilibrium (P<0.05) in the three controls; we found only an association as haplotypic homozygosity. Thus, DAXX results need to be interpreted with additional caution. Minor allele frequencies of the SNPs in this Welsh population was similar to those observed in the HapMap European sample (Table 1). 3.3. Univariate analysis All SNPs were assessed by additive model checking in the whole group and in males and females separately. In this analysis, only TP53 R72P and two BAT3 SNPs (rs805303 and rs2077102) showed statistically significant risk associations (Table 2). The two BAT3 SNPs were in linkage disequilibrium, and their associations were not independent. The intron 6 SNP rs805303 minor allele frequency was greater, and we used this SNP in further analysis. In multivariate analysis, TP53 and BAT3 associations were independent and retained individual statistical significance. Notably, MDM2 did not show any association with childhood ALL, even after stratification by sex. The only conventional risk marker for childhood ALL for which we had data available was sex. Adjustment for sex did not cause any change in the TP53 association (OR per allele51.66; Ptrend50.002). Likewise, the BAT3 SNP association remained the same after adjustment for sex (OR per allele50.68; Ptrend50.02). Thus, neither univariate association was confounded by sex. 3.4. Haplotype homozygosity Because we had genotyped multiple SNPs in BAT3 and DDR1 genes, we also examined haplotype frequencies in these genes. One DAXX haplotype consisting of minor alleles of rs2239839 and rs1059231 and the major allele of rs2073524 conferred increased risk in homozygous form (OR52.45; 95% CI51.22e4.91; P50.01). The DAXX haplotype homozygosity rates did not differ between the sexes, and the DAXX association was independent from the previously found DRB4 association (in the same sample [11]). 3.5. Interaction tests None of the analyses testing for interaction between TP53 and BAT3 SNPs rs805303 and rs2077102 or MDM2 SNP rs2279744 yielded a statistically significant result.
T.N. Do et al. / Cancer Genetics and Cytogenetics 195 (2009) 31e36
34
Table 2 Univariate analysis of associations between each SNP and childhood acute lymphoblastic leukemia Wild type
Heterozygotic a
Ptrend, additive model
Homozygotic a
a
Gene
SNP
Ref
Case vs. Ctrl
OR (95% CI)
Case vs. Ctrl
OR (95% CI)
Case vs. Ctrl
OR (95% CI)
P-value
TP53 MDM2 DAXX DAXX DAXX BAT3 BAT3 BAT3 LTA DDR1 DDR1 DDR1 DDR1 IER3
rs1042522 rs2279744 rs2073524 rs1059231 rs2239839 rs805303 rs2077102 rs2736155 rs909253 rs1049623 rs1264323 rs1264327 rs1264328 rs10947089
1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0
0.442 0.400 0.318 0.518 0.505 0.495 0.789 0.229 0.427 0.440 0.429 0.427 0.433 0.952
1.36 1.36 0.94 0.87 0.79 0.63 0.52 1.63 1.03 0.69 0.68 0.99 0.67 0.18
0.394 0.486 0.327 0.299 0.299 0.410 0.183 0.514 0.479 0.318 0.343 0.408 0.340 0.012
3.31 0.89 0.99 1.38 1.48 0.50 0.77 1.64 0.57 1.04 1.20 1.24 1.21 d
0.163 0.114 0.155 0.196 0.196 0.095 0.029 0.257 0.094 0.210 0.229 0.165 0.227 0.036
1.67 1.05 0.99 1.08 1.12 0.68 0.62 1.28 0.83 0.96 1.02 1.08 1.02 1.23
0.002 0.78 0.97 0.61 0.47 0.02 0.04 0.10 0.23 0.78 0.89 0.61 0.88 0.61
vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs.
0.566 0.452 0.309 0.514 0.497 0.372 0.670 0.326 0.403 0.383 0.381 0.439 0.385 0.934
(0.85e2.17) (0.86e2.16) (0.56e1.58) (0.54e1.39) (0.49e1.30) (0.40e1.00) (0.30e0.90) (0.96e2.76) (0.67e1.59) (0.42e1.14) (0.42e1.10) (0.62e1.59) (0.40e1.12) (0.02e1.34)
vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs.
0.371 0.403 0.375 0.372 0.372 0.485 0.299 0.451 0.440 0.441 0.450 0.424 0.449 0.066
(1.66e6.62) (0.44e1.80) (0.57e1.71) (0.73e6.62) (0.81e2.68) (0.24e1.05) (0.21e2.82) (0.89e3.03) (0.28e1.15) (0.57e1.89) (0.67e2.12) (0.65e2.34) (0.66e2.22)
vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs. vs.
0.063 0.145 0.111 0.131 0.131 0.143 0.032 0.223 0.157 0.176 0.170 0.137 0.166 0
(1.21e2.30) (0.77e1.42) (0.76e1.31) (0.80e1.47) (0.83e1.50) (0.49e0.95) (0.39e0.99) (0.95e1.72) (0.62e1.13) (0.71e1.30) (0.76e1.37) (0.80e1.47) (0.75e1.40) (0.55e2.72)
Abbreviations: CI, confidence interval; OR, odds ratio; SNP, single-nucleotide polymorphism. a Frequencies for cases (n5114) vs. controls (n5414).
DAXX haplotype homozygosity was the only other statistically significant association that appeared to modify the TP53 association. The TP53 risk association was evident when it co-occurred with DAXX haplotype homozygosity (OR per allele51.82; 95% CI51.28e2.58) and not in the rest of the group (OR50.88; P50.80), although the numbers were so small for the latter comparison and statistically, the interaction did not reach significance (P50.17). When individual SNPs were examined, only DAXX 50 end htSNP rs2073524 showed a suggestive interaction with TP53 R72P between homozygous genotypes (P50.05). The risk association with TP53 R72P got stronger in the presence of the DAXX variant allele (OR per allele51.49 (95% CI51.19e2.69), P50.005). Further analysis suggested that this interaction was female-specific: OR per allele52.41 in females (P50.004) and 1.36 in males (P50.30). Given the sex-specificity of the biological interaction between MDM2 and TP53 in other cancers, we examined statistical interaction in each sex separately. The risk conferred by TP53 R72P homozygosity was greater for girls when they had the wild-type MDM2 genotype (OR57.1; P50.02), as opposed to the variant-positive genotype (OR51.4). This interaction did not reach statistical significance (P50.14). In summary, the only statistically significant interaction concerned the TP53 R72P, which also showed a significant main effect. The sample size was not sufficient for a thorough analysis of interactions, and HardyeWeinberg disequilibrium at the three DAXX SNPs also precluded obtaining a conclusive result. 3.6. MDM2 and age at diagnosis of childhood ALL MDM2 SNP309 has been reported to influence the age at diagnosis of childhood ALL [17]. We examined the same in our group of childhood ALL cases. Median ages were
similar in the whole group and in males for each genotype (Table 3). In females, there was a statistically nonsignificant trend toward earlier age at diagnosis with an increasing number of minor allele (59, 44, 32 months) (KruskaleWallis P50.13 for the difference in age at diagnosis distributions between wild-type homozygotic and minor alleleepositive female cases). In females, the median age at diagnosis was 38 months. Of the 24 cases below this age, 15 cases (62.5%) had the MDM2 variant allele, and of 24 cases above this age, 10 cases (41.7%) had the MDM2 variant allele (P50.24). The previous study did not report the results by sex [17]. The sex-specificity of the MDM2 association with age at diagnosis remains a possibility to be tested in a larger study. 3.7. Sex-specific genotype frequencies in newborns We did not observe any difference in R72P genotype frequencies between male and female newborns. Minor allele positivity was 42.0% in females and 45.0% in males. Homozygosity rates did not differ (6.3% in both).
4. Discussion The TP53-encoded molecule is called the guardian of the genome because of its crucial roles in genome surveillance. Genome and immune surveillance are the major mechanisms for protection from cancer development. Rare, high-penetrance mutations in the TP53 gene are known to cause hereditary forms of cancers. Common, low-penetrance polymorphisms also modify cancer risk, but to a lesser extent. In childhood lymphoid malignancies, TP53 mutations are rare and there is no information on the low-penetrance TP53 polymorphisms. Here, we present our findings on association of the well-known TP53 polymorphism R72P with childhood ALL, which revealed a risk
T.N. Do et al. / Cancer Genetics and Cytogenetics 195 (2009) 31e36
35
Table 3 Median age at diagnosis of childhood acute lymphoblastic leukemia by MDM2 SNP309 genotype Age, mo (sample size)
Total Male Female P-valuea a
TT, wild type
GT, heterozygosity
GG, minor allele homozygosity
GT þ GG, minor allele positivity
68 (n542) 78 (n519) 59 (n523) 1.0
56 (n551) 66 (n531) 44 (n520) 0.008
66 (n512) 77 (n57) 32 (n55) 0.16
52 (n563) 60 (n538) 36 (n525) 0.002
The P-value is for age distribution differences between male and female cases (KruskaleWallis test).
association of R72P with gene dosage effect. The association did not show sex-specificity. The R72P association was not significantly modified by polymorphisms within the MDM1 and DAXX genes (negative regulators of TP53), although some suggestive findings were present. In a prospective cohort study designed to examine the association of R72P with longevity, among all cancers, only hematopoietic cancers showed an association with R72P [32,33]. The present study is, to our knowledge, the only one to investigate an association specifically with leukemia, and our positive results further suggest the involvement of this low-penetrance common TP53 variant in leukemia development. The R72P polymorphism (rs1042522) codes for a conservative missense amino acid substitution, and also has a role in splicing regulation. R72P affects cell death via apoptosis by altering the transcription of TP53 target genes and modifying the apoptotic potential of cells [34]. The present findings suggest that leukemia may be more sensitive to alterations in the apoptotic pathway mediated by TP53 than are solid tumors. At present, there are limited data on the association of R72P with leukemia. In CLL, it does not affect primary susceptibility [14]. MDM2 SNP309 does not modify susceptibility CLL, either [15]. On the other hand, MDM2 SNP309 minor allele deteriorates survival of CLL [16]. In childhood ALL, functional TP53 mutations are rare in initial presentation, but are more common at relapse [35]. Similarly, MDM2 also plays a greater role in childhood ALL relapse [36]. Here we have presented some preliminary data that the common low-penetrance variant in the TP53 gene R72P increases the risk for childhood ALL. It has not been examined whether it increases the risk for relapse. Given its frequency (it is carried by O40% of Europeans), this polymorphism may prove to be a common marker for clinical risk stratification in childhood ALL. Children with MDM2 SNP309 minor allele homozygosity (genotype GG) develop childhood ALL at a younger age [17]. This is similar to the effect of this polymorphism on patients with LieFraumeni syndrome caused by TP53 mutations [37]. We did not observe a statistically significant association of SNP309 GG genotype with age at diagnosis in the overall group of childhood ALL cases, but median age of female cases with the GG genotype was 32 months, as opposed to 59 months for cases with the wild-type genotype (TT). In males, corresponding median ages were 77 and 78 months, respectively (Table 3). The association
suggested by this difference in female cases was confirmed by a statistically significant difference in the age at onset distributions in cases with wild-type MDM2 genotype (median age559 months) and MDM2 minor allele positive cases (median age536 months). Compared with male cases positive for the minor allele (median age560 months), the age distribution in female cases was significantly different (P50.002). A similar change in age at onset in females compared with male cases mediated by the MDM2 SNP309 minor allele has been reported also in adults with melanoma [38]. Our observation in childhood ALL raises an important question regarding the effect of estrogen on MDM2 SNP309 activity in childhood [7]. Because there is only negligible estrogen activity in childhood, the present finding, if confirmed, would provide an example of fetal programming of childhood ALL susceptibility. We have previously shown that, by exploring sex-specific genotype frequencies in newborns, evidence may be sought for markers of sex-specific prenatal selection [22]. Because the TP53 polymorphism R72P has been implicated in implantation failure [20] and recurrent miscarriages [21], we examined genotype frequencies in male and female newborns, but did not detect any differences. The HLA complex is also believed to be a modifier of reproductive success [39,40], but none of the HLA loci analyzed in the present study showed a sex-specific difference to account for the previously found male deficit in HLA class II homozygosity rates in the same group of newborns. Our results suggested that TP53 R72P polymorphism may be a modifier of risk for childhood ALL susceptibility and might interact with its regulators, MDM2 and DAXX, in females. Although biologically plausible, because of the limited sample size of the present study these results need to be replicated before considering the inclusion of R72P into a predictive genetic risk model for childhood ALL in the future. The intriguing hypothesis that fetal estrogen exposure may be involved in postnatal MDM2 activity to modify the age at diagnosis of childhood ALL may also be examined, using genetic epidemiology as a probe for disease biology.
Acknowledgments This work was funded intramurally by HUMIGEN LLC, Hamilton, NJ. The Authors are employees of HUMIGEN, The Institute for Genetic Immunology.
36
T.N. Do et al. / Cancer Genetics and Cytogenetics 195 (2009) 31e36
References [1] Dunning AM, Healey CS, Pharoah PD, Teare MD, Ponder BA, Easton DF. A systematic review of genetic polymorphisms and breast cancer risk. Cancer Epidemiol Biomarkers Prev 1999;8:843e54. [2] Fan R, Wu MT, Miller D, Wain JC, Kelsey KT, Wiencke JK, Christiani DC. The p53 codon 72 polymorphism and lung cancer risk. Cancer Epidemiol Biomarkers Prev 2000;9:1037e42. [3] Mechanic LE, Marrogi AJ, Welsh JA, Bowman ED, Khan MA, Enewold L, Zheng YL, Chanock S, Shields PG, Harris CC. Polymorphisms in XPD and TP53 and mutation in human lung cancer. Carcinogenesis 2005;26:597e604. [4] Koushik A, Tranah GJ, Ma J, Stampfer MJ, Sesso HD, Fuchs CS, Giovannucci EL, Hunter DJ. p53 Arg72Pro polymorphism and risk of colorectal adenoma and cancer. Int J Cancer 2006;119:1863e8. [5] van Heemst D, Mooijaart SP, Beekman M, Schreuder J, de Craen AJ, Brandt BW, Slagboom PE, Westendorp RG. Long Life Study Group. Variation in the human TP53 gene affects old age survival and cancer mortality. Exp Gerontol 2005;40:11e5. [6] Bond GL, Hu W, Bond EE, Robins H, Lutzker SG, Arva NC, Bargonetti J, Bartel F, Taubert H, Wuerl P, Onel K, Yip L, Hwang SJ, Strong LC, Lozano G, Levine AJ. A single nucleotide polymorphism in the MDM2 promoter attenuates the p53 tumor suppressor pathway and accelerates tumor formation in humans. Cell 2004;119:591e602. [7] Bond GL, Hirshfield KM, Kirchhoff T, Alexe G, Bond EE, Robins H, Bartel F, Taubert H, Wuerl P, Hait W, Toppmeyer D, Offit K, Levine AJ. MDM2 SNP309 accelerates tumor formation in a genderspecific and hormone-dependent manner. Cancer Res 2006;66:5104e10. [8] Zhao LY, Liu J, Sidhu GS, Niu Y, Liu Y, Wang R, Liao D. Negative regulation of p53 functions by Daxx and the involvement of MDM2. J Biol Chem 2004;279:50566e79. [9] Li Q, Wang X, Wu X, Rui Y, Liu W, Wang J, Wang X, Liou YC, Ye Z, Lin SC. Daxx cooperates with the Axin/HIPK2/p53 complex to induce cell death. Cancer Res 2007;67:66e74. [10] Herberg JA, Beck S, Trowsdale J. TAPASIN, DAXX, RGL2, HKE2 and four new genes (BING 1, 3 to 5) form a dense cluster at the centromeric end of the MHC. J Mol Biol 1998;277:839e57. [11] Dorak MT, Lawson T, Machulla HK, Darke C, Mills KI, Burnett AK. Unravelling an HLA-DR association in childhood acute lymphoblastic leukemia. Blood 1999;94:694e700. [12] Taylor GM, Dearden S, Ravetto P, Ayres M, Watson P, Hussain A, Greaves M, Alexander F, Eden OB. UKCCS Investigators (United Kingdom Childhood Cancer Study). Genetic susceptibility to childhood common acute lymphoblastic leukaemia is associated with polymorphic peptide-binding pocket profiles in HLA-DPB1*0201. Hum Mol Genet 2002;11:1585e97. [13] Taylor GM, Hussain A, Lightfoot TJ, Birch JM, Eden TO, Greaves MF. HLA-associated susceptibility to childhood B-cell precursor ALL: definition and role of HLA-DPB1 supertypes. Br J Cancer 2008;98:1125e31. [14] Lahiri O, Harris S, Packham G, Howell M. p53 pathway gene single nucleotide polymorphisms and chronic lymphocytic leukemia. Cancer Genet Cytogenet 2007;179:36e44. [15] Zenz T, Ha¨be S, Benner A, Kienle D, Do¨hner H, Stilgenbauer S. The MDM2 -309 T/G promoter single nucleotide polymorphism does not alter disease characteristics in chronic lymphocytic leukemia. Haematologica 2008;93:1111e3. [16] Gryshchenko I, Hofbauer S, Stoecher M, Daniel PT, Steurer M, Gaiger A, Eigenberger K, Greil R, Tinhofer I. MDM2 SNP309 is associated with poor outcome in B-cell chronic lymphocytic leukemia. J Clin Oncol 2008;26:2252e7. [17] Swinney RM, Hsu SC, Hirschman BA, Chen TT, Tomlinson GE. MDM2 promoter variation and age of diagnosis of acute lymphoblastic leukemia. Leukemia 2005;19:1996e8. [18] Dorak MT, Burnett AK, Worwood M, Sproul AM, Gibson BE. The C282Y mutation of HFE is another male-specific risk factor for childhood acute lymphoblastic leukemia. Blood 1999;94:3957.
[19] Dorak MT, Pearce MS, Hammal DM, McNally RJ, Parker L. Examination of gender effect in birth weight and miscarriage associations with childhood cancer (United Kingdom). Cancer Causes Control 2007;18:219e28. [20] Kay C, Jeyendran RS, Coulam CB. p53 tumour suppressor gene polymorphism is associated with recurrent implantation failure. Reprod Biomed Online 2006;13:492e6. [21] Pietrowski D, Bettendorf H, Riener EK, Keck C, Hefler LA, Huber JC, Tempfer C. Recurrent pregnancy failure is associated with a polymorphism in the p53 tumour suppressor gene. Hum Reprod 2005;20:848e51. [22] Dorak MT, Lawson T, Machulla HK, Mills KI, Burnett AK. Increased heterozygosity for MHC class II lineages in newborn males. Genes Immun 2002;3:263e9. [23] Desmots F, Russell HR, Lee Y, Boyd K, McKinnon PJ. The reaperbinding protein scythe modulates apoptosis and proliferation during mammalian development. Mol Cell Biol 2005;25:10329e37. [24] Higuchi M, Singh S, Aggarwal BB. Characterization of the apoptotic effects of human tumor necrosis factor: development of highly rapid and specific bioassay for human tumor necrosis factor and lymphotoxin using human target cells. J Immunol Methods 1995;178:173e81. [25] Rossi D, Gaidano G. Messengers of cell death: apoptotic signaling in health and disease. Haematologica 2003;88:212e8. [26] Ongusaha PP, Kim JI, Fang L, Wong TW, Yancopoulos GD, Aaronson SA, Lee SW. p53 induction and activation of DDR1 kinase counteract p53-mediated apoptosis and influence p53 regulation through a positive feedback loop. EMBO J 2003;22:1289e301. [27] Scha¨fer H, Trauzold A, Sebens T, Deppert W, Fo¨lsch UR, Schmidt WE. The proliferation-associated early response gene p22/PRG1 is a novel p53 target gene. Oncogene 1998;16:2479e87. [28] Wu MX. Roles of the stress-induced gene IEX-1 in regulation of cell death and oncogenesis. Apoptosis 2003;8:11e8. [29] Harrison CJ. The detection and significance of chromosomal abnormalities in childhood acute lymphoblastic leukaemia. Blood Rev 2001;15:49e59. [30] Cui J. HardyeWeinberg equilibrium test in caseecontrol studies. Stata Tech Bull 2000;57:17e9 [available at]. http://www.stata.com. [31] Sasieni PD. From genotypes to genes: doubling the sample size. Biometrics 1997;53:1253e61. [32] Bojesen SE, Nordestgaard BG. The common germline Arg72Pro polymorphism of p53 and increased longevity in humans. Cell Cycle 2008;7:158e63. [33] Ørsted DD, Bojesen SE, Tybjaerg-Hansen A, Nordestgaard BG. Tumor suppressor p53 Arg72Pro polymorphism and longevity, cancer survival, and risk of cancer in the general population. J Exp Med 2007;204:1295e301. [34] Imyanitov EN. Gene polymorphisms, apoptotic capacity and cancer risk. Hum Genet 2009;125:239e46. [35] Gump J, McGavran L, Wei Q, Hunger SP. Analysis of TP53 mutations in relapsed childhood acute lymphoblastic leukemia. J Pediatr Hematol Oncol 2001;23:416e9. [36] Zhou M, Gu L, Abshire TC, Homans A, Billett AL, Yeager AM, Findley HW. Incidence and prognostic significance of MDM2 oncoprotein overexpression in relapsed childhood acute lymphoblastic leukemia. Leukemia 2000;14:61e7. [37] Bougeard G, Baert-Desurmont S, Tournier I, Vasseur S, Martin C, Brugieres L, Chompret A, Bressac-de Paillerets B, StoppaLyonnet D, Bonaiti-Pellie C, Frebourg T. Impact of the MDM2 SNP309 and p53 Arg72Pro polymorphism on age of tumour onset in LieFraumeni syndrome. J Med Genet 2006;43:531e3. [38] Firoz EF, Warycha M, Zakrzewski J, Pollens D, Wang G, Shapiro R, Berman R, Pavlick A, Manga P, Ostrer H, Celebi JT, Kamino H, Darvishian F, Rolnitzky L, Goldberg JD, Osman I, Polsky D. Association of MDM2 SNP309, age of onset, and gender in cutaneous melanoma. Clin Cancer Res 2009;15:2573e80. [39] Lerner SP, Finch CE. The major histocompatibility complex and reproductive functions. Endocr Rev 1991;12:78e90. [40] Kostyu DD. HLA: fertile territory for developmental genes? Crit Rev Immunol 1994;14:29e59.