Genetic variants in ATM, H2AFX and MRE11 genes and susceptibility to breast cancer in the polish population

We collected blood samples from 315 non-selected female patients diagnosed with breast cancer. A total of 515 anonymous blood samples were used as a control population. The control group consisted of individuals attending for a screening check-up in hospital or were healthy blood donors with no history of medical illness. Patients were eligible for present study if they revealed no mutations in BRCA1, BRCA2 and CHEK2 genes. The baseline characteristics of the patients are shown in Table 1. The mean age of patients was 53 years (range 26–76 years). Invasive ductal carcinoma was the most common subtype of cancer (n = 191, 60.6%). Most of the tumors were II and undetermined grade (n = 91, 28.9%, n = 75, 23.8%, respectively). ER/PgR status was available for majority of our BC patients. The study was conducted with the approval of the Central Ethical Committee of the Ministry of Health in Poland, in accordance with the tenets of the Declaration of Helsinki (Decision no. 949/16). All patients signed informed consent forms.

Genomic DNA was extracted from whole blood samples using a PureGene DNA isolation kit in accordance with the manufacturer’s protocols (Gentra Systems).

The ATM mutations analysis was done using a combination of different methods. Polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) was used to detect c.5932G > T, c.6095G > A and c.7630-2A > C mutations. PCR-RFLP analysis was performed using the restriction enzymes MseI, BfaI and AluI, respectively. The c.7630-2A > C mutation abolishes an AluI site. Digestion with AluI of the PCR product without the mutation gives four fragments (143, 70, 61 and 7 base pair (bp)), whereas the PCR product with mutation only three bands are observed (213, 61 and 7 bp). The c.5932G > T mutation creates an MseI restriction site. After digestion with MseI, a 232 bp PCR product produces three bands: 159, 40 and 33 in patients with the mutation, while c.6095G > A disrupts the BfaI site. After digestion with BfaI, a 234 bp PCR product without mutation shows two bands, 127 and 107 bp, and the PCR product of alleles with this mutation remains undigested. PCR products were digested with 10 U of restriction enzymes by overnight incubation at 37 °C. The restriction fragments were resolved on 3% agarose gel.

Furthermore, selected variants of H2AFX and MRE11 (rs7759, rs8551, rs643788, rs2509049, rs1061956 and rs2155209) were genotyped using TaqMan® SNP genotyping assays (Life Technologies, Carlsbad, California) and CFX96 BioRad instruments. Four SNPs, rs643788, rs8551, rs7759 and rs2509049, are located in the far promoter region of the H2AFX gene -1654C > T, -1420C > T, −1187A > G, and -417C > T, respectively. The PCR was performed with HOT FIREPol Probe qPCR Mix Plus (no ROX) in accordance with the manufacturer’s instructions (Solis Biodyne, Tartu, Estonia). The PCR thermal cycling was as follows: initial denaturation at 95 °C for 15 min. and next 40 cycles of 95 °C for 15 s and 60 °C for 60 s. As a quality control measure, negative controls and approximately 5% of the samples were genotyped in duplicate to check genotyping accuracy.

Nucleotide positions were determined according to the standard reference sequences for ATM NM_000051.3, whereby mutation numbering uses the ‘A’ of the ATG initiation codon as + 1. The reference sequence for H2AFX used NC_000011.10, and for MRE11 NC_000011.9.

All statistical analysis was undertaken using GraphPad Prism 5.0 software (GraphPad, La Jolla, CA, USA). The genotype frequencies of each SNP were tested for deviation from the Hardy-Weinberg equilibrium (HWE) amongst the controls. This was done by comparing the observed genotype frequencies with the expected frequencies using a Chi-squared test. The ORs and 95% CIs were calculated to assess BC risk. We considered P < 0.05 to be significant for all analyses. P values were corrected using Benjamini-Hochberg adjustment. Linkage disequilibrium (LD) measures (Lewontin’s D’ and the r² coefficient) between SNPs were calculated using Haploview 4.2 software (Daly Lab, USA). Haplotype frequencies were compared among patients and controls (using the Chi-squared test). The statistical power analyses were determined using free available Power and Sample Size Calculator.

SNPs showing significant association with BC were included in the cumulative genetic risk score (CGRS) analysis. Genotypes were coded as 0, 1 or 2, indicating the number of risk alleles in the genotype. Both unweighted (uwCGRS) and weighted (wCGRS) CGRS were calculated. In an unweighted approach, coded genotypes were counted to create a CGRS (therefore, the range of possible scores for three SNPs was 0 to 6). In a weighted approach, all the scores of the coded genotypes were multiplied by the log(OR) estimated for each risk allele in the current study. A weighted risk score is the sum of the multiplied results for each SNP and scaled by a factor of 3/∑w_i, where w_i = log(OR) (the logarithm of the odds ratio) for the ith SNP and i = 3 [13]. The effect of unweighted and weighted CGRSs on BC was calculated using logistic regression analysis. A t-test was applied to compare the average and mode values of uwCGRS and wCGRS between the BC and control groups.

In the group studied, we found six changes (five in the BC patients and two in the control group) in the functional domain of the ATM gene (c.6067G > A, c.6095G > A (twice), c.8187A > T, c.8314G > A, c.6083A > G and c.8787-55C > T). In the BC cases, three were known mutations: c.6095G > A, c.8187A > T and c.6067G > A. The patient with the c.6067G > A mutation was also a carrier of the NBN Ile171Val mutation. Moreover, c.8787-55C > T was found in a homozygous state. We detected two mutations in the control group: c.6095G > A and c.8314G > A.Deleterious consequences of all detected changes were scored using the following tools: SIFT, PolyPhen2 and MutationTaster Phylop. All these algorithms estimate the pathogenic effects of SNPs on protein in different ways. SIFT calculates score based on multiple sequence alignments. PolyPhen2 predicts the possible effects using multiple sequence alignments, 3D protein structures and residue contact information from secondary structure. Mutation taster Phylop evaluates disease-causing potential of sequence alterations comprising many aspects: evolutionary conservation, splice-site changes, loss of protein features and changes that might affect the amount of mRNA. The pathogenic effects on ATM were confirmed if more than two analysis indicated damaging consequences. The details are presented in Table 2.

SIFT algorithm ranges from 0 to 1. The amino acid substitution is predicted damaging is the score is ≤0.05, and tolerated if the score is > 0.05; PolyPhen 2 algorithm ranges from 0 to 1. possibly damaging and probably damaging (> 0.5) or benign (< 0.5); Mutation taster Phylop algorithm evaluates disease-causing potential of sequence alterations comprising evolutionary conservation, splice-site changes, loss of protein features and changes that might affect the amount of mRNA.

The genotype and allele frequencies are summarized in Tables 3 and 4. The observed genotype frequencies of the six polymorphisms referred to above were all in agreement with the HWE in the control subjects (the P values for the H2AFX HWE were as follows: 0.32, 0.82, 0.08 and 0.72; for the MRE11 HWE, the P values were: 0.84 and 0.59). For the H2AFX polymorphisms, the logistic regression analysis revealed that the rs7759 GG and rs8551 TT were significantly increased among breast cancer patients compared to the rs7759 AA and rs8551 CC genotypes, respectively (rs7759 GG versus AG: adjusted OR = 1.94, 95% CI 1.14 to 3.28; P = 0.042, after Benjamini-Hochberg correction; rs8551 CC versus TT, OR = 1.82, 95% CI 1.18 to 2.80; P = 0.034, after Benjamini-Hochberg correction). The significant association was also found between the H2AFX rs7759 polymorphism and cancer risk under the heterozygous co-dominant model (AA versus AG): OR = 1.73, 95% CI 1.28 to 2.33; P = 0.024, after Benjamini-Hochberg correction and under the dominant genetic model (AA versus AG + GG): OR = 1.76, 95% CI 1.32 to 2.34; P = 0.0016, after Benjamini-Hochberg correction. The G allele at rs7759 was significantly more prevalent in BC cases compared to the controls (OR = 1.47, 95% CI 1.19 to 1.82; P = 0.0018, after Benjamini-Hochberg correction). Otherwise, there were no differences at rs8551 under the dominant genetic model (CC versus CT + TT): OR = 1.37, 95% CI 1.021 to 1.83; P = 0.0357, or under the heterozygous co-dominant model (CC versus CT): OR = 1.26, 95% CI 0.93 to 1.71; P = 0.14. We observed that the frequency of the T allele of rs8551 was higher in BC patients than in the controls (OR = 1.33, 95% CI 1.09 to 1.64; P = 0.018, after Benjamini-Hochberg correction). There were no differences in the occurrences of the TC, CC and TC + CC genotypes at rs643788 and the CT, TT and CT + TT genotypes at rs2509049 between the BC cases and the controls. On the other hand, the T allele in rs2509049 was significantly higher in BC patients compared to controls: OR = 1.3, 95% CI 1.06 to 1.59; P = 0.024, after Benjamini-Hochberg correction. For the two MRE11 variants, there were no statistically significant differences in genotype and allele frequencies between the BC patients and the controls at rs1061956 or rs2155209.

To determine the combined effects of the four promoter H2AFX SNPs, we generated haplotypes based on the observed genotypes (Fig. 1). For H2AFX SNPs, the construction of haplotypes revealed the presence of seven haplotypes in BC patients and eight haplotypes in the control group. In both groups, the most frequent H2AFX haplotype was CACC, without any risk variant allele (54.4% in BC patients and 61.4% in the control group). We also observed haplotypes that only presented in particular groups: TGTC in BC cases and TGCT and CATC in the control group. This observation could be related to the small size of the study group. Haplotypes with a frequency lower than 1% were not considered for further analysis. Haplotypes are presented in Table 5. When the CACC haplotype was used as the reference, CACT haplotype was associated with an increased risk of breast cancer. The difference in the frequency distribution of haplotype between the BC cases and controls was statistically significant (P < 0.0001 for CACT: OR = 27.29, 95% CI 3.56 to 209.5).

All H2AFX variants that showed a significant association with BC were included in the CGRS analysis. The risk alleles were defined as G for rs7759 and T for rs8551 and rs2509049. The average (± standard deviation [SD]) of the cumulative risk scores among the two groups studied were similar (for BC 2.37 ± 2.06 and for the control group 2.01 ± 1.88), although mode values were different for the BC and control groups, at 3 and 0 respectively. Individuals were stratified into three groups according to the number of risk alleles: carrying ≤ two (reference group), three, and ≥ four alleles. The risk of BC increased with the number of alleles and was statistically significant for three alleles (P = 0.019) and ≥ four alleles (P = 0.014) compared to ≤ two risk alleles. ORs (95% CI) for carriers of three and ≥ four risk alleles were 1.46 (1.06 to 2.01) and 1.71 (1.11 to 2.62), respectively (Fig. 2). The ORs calculated in unweighted and weighted CGRS analysis were similar, which is probably connected to the low number of risk alleles and small sample size. We also compared a clinical data between patients with high cumulative genetic risk score (≥4 risk alleles) and all BC patients. The detailed clinical parameters of BC patients with high cumulative genetic risk score (≥4 risk alleles) are shown in Table 6.

The post-hoc analysis revealed that the statistical power of our study for analyses of the differences in distribution of alleles at loci rs7759, rs8551 and rs2509049 (OR: 1.3 to 1.47 for the three H2AFX SNPs with a frequency of 0.28 to 0.35) between BC patients and controls ranged between 71 and 94%. This means that the study had enough power to detect an association of the H2AFX gene in BC group, in the case-control analysis.