|Home | About | Journals | Submit | Contact Us | Français|
TP53BP1 is a key component of radiation-induced deoxyribonucleic acid damage repair. The purpose of this study was to evaluate the significance of a known common single nucleotide polymorphism in this gene (rs560191) in patients treated with breast-conserving surgery and whole-breast irradiation (BCS + RT).
The population consisted of 176 premenopausal women treated with BCS + RT (median follow-up, 12 years). Genomic deoxyribonucleic acid was processed by use of TaqMan assays. Each allele for rs560191 was either C or G, so each patient was therefore classified as CC, CG, or GG. Patients were grouped as GG if they were homozygous for the variant G allele or CC–CG if they carried at least one copy of the common C allele (CC or CG).
Of the 176 women, 124 (71%) were CC–CG and 52 (29%) were GG. The mean age was 44 years forGG vs. 38 years for CC–CG (p < 0.001). GG was more common in African-American women than white women (69% vs. 13%, p < 0.001) and more commonly estrogen receptor negative (70% vs. 49%, p = 0.02). There were no significant correlations of rs560191 with other critical variables. Despite the fact that GG patients were older, the 10-year rate of local relapses was higher (22% for GG vs. 12% for CC–CG, p = 0.04).
This novel avenue of investigation of polymorphisms in radiation repair/response genes in patients treated with BCS + RT suggests a correlation to local relapse. Additional evaluation is needed to assess the biological and functional significance of these single nucleotide polymorphisms, and larger confirmatory validation studies will be required to determine the clinical implications.
The majority of women presenting with early-stage breast cancer are treated with breast-conserving surgery and radiation. Local relapse of disease remains a significant pattern of treatment failure, particularly in younger women, which ultimately may compromise survival and can be associated with significant social and psychological consequences (1–6). Primary tumor factors, such as receptor status, margins, histologic subtype, grade, and more recently, molecular profiles, have been extensively evaluated as potential risk factors for local relapse, with some consistent but many conflicting results (1, 7–16). Host factors including age and race and, to a lesser extent, genetic factors such as BRCA1, BRCA2, and CHEK2 have also been evaluated, again with conflicting results (17–22). The most consistently reported risk factors to date for local relapse after breast-conserving surgery and radiation have been patient age, with younger age predicting for higher local relapse rates and positive margin status predicting for higher local relapse rates (1–3, 15, 22–29).
Although there are multiple studies evaluating the high-penetrance BRCA1/BRCA2 genes and local relapse, overall patient numbers have been relatively small because less than 1% of the population and fewer than 5% of breast cancer patients are carriers of deleterious mutations (30). There are even fewer data on other more rare high-penetrance genetic syndromes, where it is estimated that only 1% to 2% of familial cases are explained by mutations in other known cancer susceptibility genes, such as P53, PTEN, ATM, and CHEK2 (30–33).
Although deleterious mutations in highly penetrant breast cancer susceptibility genes, such as BRCA1/2, are relatively rare, there is a wide spectrum of much more common low-penetrance genetic changes in genes that may be clinically relevant (32–34). Many of these genetic changes are single nucleotide substitutions, which may or may not affect the function of the gene, commonly referred to as single nucleotide polymorphisms (SNPs). For some of these polymorphisms, fairly large segments of the population may be affected. A recently discovered example of a clinically relevant common polymorphism is the CYP2D6 gene, where approximately 10% of the population are homozygous carriers of a single nucleotide variation in the CYP2D6 gene (35). Patients homozygous for the polymorphism have less efficient metabolism of tamoxifen to its active metabolite endoxifen and have been reported to have inferior outcomes compared with wild-type and heterozygous patients when treated with tamoxifen. This represents a novel approach for individualizing therapy based on the genetics of the host, as opposed to the clinical, pathologic, or molecular characteristics of the primary tumor.
It is evident that evaluation of SNPs of the host as risk factors for recurrence and response, as well as for normal tissue reactions in patients undergoing radiation therapy, is an exciting and novel area of investigation that has not been extensively explored. In this regard there are a large number of candidate genes, related to radiation response and deoxyribonucleic acid (DNA) damage repair, that have common polymorphisms that may be clinically relevant. One such candidate is the gene encoding for the tumor suppressor p53 binding protein 1 (TP53BP1) (36–39). 53BP1 participates in the early DNA damage response after radiation and is recruited rapidly to sites of DNA breaks, where it is required for efficient recruitment of other critical DNA repair proteins such as BRCA1. 53BP1 and ATM interact in irradiated cells, with ATM activation leading to phosphorylation of 53BP1 and intact 53BP1 being required for optimal ATM autophosphorylation.. Polymorphic variants in TP53BP1 are therefore excellent potential candidates for predicting response to radiation and for cancer susceptibility. Several common polymorphisms in the TP53BP1 gene have been identified (36). Although the functional and biologic significance of these polymorphisms is unclear and the evidence linking polymorphisms in TP53BP1 to breast cancer risk has been mixed, 53BP1 clearly plays a major role in response to DNA damage by radiation, and modest changes in the function of the gene can have significant consequences (36, 38, 40). Evaluation of TP53BP1 polymorphisms on outcomes after radiation therapy in breast cancer patients has not evaluated. The purpose of this study was to evaluate the prognostic significance of polymorphisms in the TP53BP1 gene in premenopausal patients treated with breast-conserving surgery and radiation.
The patient population consisted of 176 premenopausal women treated with breast-conserving surgery and whole-breast irradiation between 1985 and 2003. Patients were treated by lumpectomy and whole-breast irradiation, with or without regional nodal irradiation and systemic therapy, in accordance with standard practice and as previously described (18, 41). Premenopausal early-stage patients were selected because these patients are at highest risk for local relapse and represent a relatively homogeneous cohort. Patients were recruited from the radiation therapy treatment facilities during follow-up, and after they provided informed consent, blood was drawn for genetic testing, DNA was stored, and patient data were deidentified with a unique identification code, in accordance with guidelines for this institutional review board–approved protocol. All patient data including demographic information, clinical, pathologic, treatment, and outcomes were stored in a computerized database for analysis. Because patients were recruited from follow-up clinics after radiation treatment, median follow-up from the date of original diagnosis was 12 years.
Three known polymorphisms in the TP53BP1 gene, previously described by Frank et al. (36), were genotyped by standard methods: D353E, G412S, and K1136Q. These polymorphisms are associated with amino acid changes in the coding region of TP53BP1. To test whether the allele status of the three loci in the TP53BP1 gene are highly correlated with each other, we performed a linkage disequilibrium analysis using a common methodology in population genetics described by Devlin and Risch (42). This method calculates a D′ value and an r value, which quantifies the degree of linkage between the loci based on the distribution in the population of a specific loci. This methodology allows one to create a haplotype map of the three loci and quantifies the degree of association between loci based on the genetic analysis of the population tested (42).
Genomic DNA was extracted from 1 mL of peripheral blood, obtained through venipuncture, by use of a spin column–based method according to the manufacturer’s protocol (Qiagen, Valencia, CA). Genotyping was performed with a TaqMan assay on the ABI 7000 Sequence Detection System (Applied Biosystems, Foster City, CA). In brief, reactions were performed with 5 to 10 ng of genomic DNA in a 25-µL volume. Polymerase chain reaction cycling conditions were 50°C for 2 minutes and 95°C for 10 minutes, followed by 40 cycles of 92°C for 15 seconds and 60°C for 1 minute.
For each of the three loci, alleles were classified as C, G, A, or T. In the case of D353E (rs560191), the alleles were either C or G. Genotyping for each patient was thus categorized as CC, CG, or GG; patients were classified as GG if they were homozygous for the variant G allele (GG) or CC–CG if they carried at least one copy of the common C allele (CC or CG). All clinical, pathologic, treatment, and outcomes data, along with the results of the SNP analysis, were entered into a computerized database, deidentified, and analyzed with standard statistical packages.
The linkage disequilibrium analysis of the three loci in TP53BP1 showed a tight correlation, indicating a nonrandom association between these three loci. On the basis of the analysis described previously, the D′ value between the all three loci was 1.0, indicating a tight linkage of all three loci in both white and African-American cohorts. Further linkage analysis of the correlation coefficient (r) showed a value of 1.0 between loci D353E and K1136Q among white patients and 0.98 among African Americans, whereas the r value between these loci and G412S was 0.73 in white patients and 0.82 in African Americans (42). These data confirm a very strong linkage between D353E and K1136Q and a strong but less complete linkage between these two loci and G412S. On the basis of this analysis, an estimated haplotype map of the three loci is given in Fig. 1. Correlations between the various variables and outcomes were essentially identical for K1136Q and, as expected, were not as strong for G412S, which was not as tightly linked. The linkage analysis between the three loci is shown in Table 1.
Because the three loci are so tightly linked and highly correlated, we report here on the results of the D353E locus (rs560191). Of the 176 premenopausal women, 124 (71%) were classified as CC–CG and 52 (29%) as GG. As noted in Table 2, the mean age was 44 years for the GG patients vs. 38 years for the CC–CG patients (p < 0.001), and GG was more common in African-American women than in white women (69% vs. 13%, p < 0.001). There were no significant correlations of rs560191 with other critical variables: T stage, N stage, margin status, use of adjuvant therapy, or family history. Although progesterone receptor status was evenly distributed between the populations, there was a slight predominance of estrogen receptor (ER)–negative tumors in the GG group.
In addition, BRCA1/2 mutation status was known in 149 of the 176 patients, and there was no correlation between BRCA1/ 2 mutation status and the polymorphisms in TP53BP1.
Because younger age in a majority of studies is associated with higher rates of local relapse, it was hypothesized that the GG patients, who were older, would have a lower local relapse rate. Despite the fact that GG patients were older, the 10-year rate of local relapse was higher (22% GG vs. 12% CC–CG, p = 0.04). As noted previously, there is no correlation between the TP53BP1 status and BRCA1/2 status, which would explain the difference in local control. Figure 2 shows the local relapse rate as a function of genotype over time. On multivariate analysis, when age, receptor status, margin status, and adjuvant therapy were taken into account, however, genotyping did not retain statistical significance (p = 0.12).
As noted previously, GG was more common in African-American women, and there was a predominance of ER-negative tumors in the GG group. However, in this sample there was no difference in the local relapse rate between white and African-American women (17% vs. 15% at 15 years), and there was no significant difference in local relapse as a function of ER status (17% ER negative vs. 20% ER positive at 15 years). The difference in local relapse rate between the GG and the CC–CG groups could therefore not be explained by an imbalance in race, age, or ER status.
The frequency of the homozygous variant GG genotype in our white population of patients was 13%, and it did not differ significantly from the approximate 10% frequency of the GG genotype in white breast cancer patients or white control subjects reported by Frank et al. (36).
To determine whether the higher frequency of GG genotype in our African-American breast cancer patient population was different from the frequency in control subjects, we obtained 100 age-matched control African-American samples without a diagnosis of breast cancer (Bioserve, Beltsville, MD) and conducted a 2:1 age-matched control. As shown in Table 3, the frequency of the GG genotype in our African-American patient population of 69% did not differ significantly from the GG genotype frequency of 81% in the African-American control population.
The rate of contralateral events in GG patients was slightly but not significantly higher at 10 years (13% vs. 8%, p = not significant). This is shown in Fig. 3.
Because these patients were all alive and without evidence of disease at the time of recruitment, no attempt was made to correlate genotype with distant metastasis or overall survival.
Single nucleotide polymorphisms represent the most common type of genetic variation, and with the completion of the Human Genome and HapMap project, numerous common SNPs have been identified and characterized (33, 34, 43–46). Whether many of these common SNPs are clinically or biologically significant remains to be determined. However, because some of these polymorphisms can result in amino acid changes within the protein coding portion of genes critical to cellular function, they are candidates for potential predictors of variation in biological function. One limitation of our study is the lack of understanding of how these particular polymorphisms may affect the biological function of the gene. It has been established that the 53BP1 protein is a critical component of DNA repair. Although these polymorphisms do result in amino acid changes, how those changes affect the specific structure and function of the protein has not been elucidated. Future laboratory studies will need to be performed to further explore the biological significance of these SNPs.
There are a number of genes involved in DNA repair and radiation response in which SNPs have been identified (47–50). The P53 binding protein is a critical component in response to DNA damage, and the gene encoding for this protein has been fully characterized and polymorphisms identified (36–38, 40, 51, 52). Polymorphisms in TP53BP1 have been evaluated with respect to the risk of breast cancer, with conflicting results (36–38, 40, 51, 52). With respect to breast cancer risk, although TP53BP1 may be a minor contributing factor, it is more likely that combinations of SNPs in this and other critical genes may be associated with breast cancer risk. Of note, we observed no correlation between TP53BP1 polymorphisms and deleterious mutations in BRCA1/2.
Outcomes related to SNPs in genes related to DNA damage/repair response have not been extensively evaluated. There have been a few studies that evaluated short- and long-term normal tissue reactions as a function of polymorphisms, with mixed but promising results (53–56). Polymorphisms in ATM have been shown to correlate with chronic fibrosis in breast cancer patients, and a recent study evaluating polymorphisms in transforming growth factor β showed an association with radiation pneumonitis in lung cancer patients undergoing radiation therapy (56). Krupa et al. (57) showed a correlation between polymorphisms in repair genes RAD51 and XRCC3 and nodal metastasis, but they did not evaluate correlations with local or regional relapse. Skerrett et al. (58) also noted associations of transforming growth factor β with recurrence in breast cancer but did not evaluate local control.
Data correlating a number of polymorphisms with overall or disease-free survival and response to chemotherapy and other drugs continues to rapidly develop. One of the most well-described examples of this is CYP2D6, the gene associated with metabolism of tamoxifen to its active endoxifen metabolite (35). A specific polymorphism that is found in approximately 10% of the population is associated with slow metabolism and poorer response to tamoxifen therapy. Currently, a commercially available test can be used to identify patients with this polymorphism to guide medical recommendations regarding hormonal therapy.
Evaluation of locoregional control with radiation as a function of germline polymorphisms has not been extensively evaluated. Although there are a number of studies assessing local control as a function of germline BRCA1/2 status, evaluation of local control as a function of SNPs in genes associated with radiation response is underexplored (17–21). One recent study reported a positive correlation in a functional polymorphism in the promoter of BCL2 and disease-free and overall survival in oropharyngeal cancers treated with surgery and radiation (59). This polymorphism in the BCL2 promoter has also been shown to correlate with rising prostate-specific antigen level after prostatectomy (60). In contrast to the rapidly increasing numbers of studies assessing polymorphisms associated with response to drug therapy, however, this area is largely underdeveloped (61–70).
The P53 binding protein is a key component in radiation response (51). Our laboratory has been evaluating 53BP1 expression and radiation response and hypothesized that polymorphisms in TP53BP1 may be associated with outcomes in patients undergoing breast-conserving surgery and radiation. In a cohort of patients treated with breast-conserving surgery and radiation, we found that the homozygous GG genotype was associated with a significantly higher risk of local relapse on univariate analysis. Because the GG genotype–carrying group of patients was older than the other CC and CG genotype carriers, one would expect a lower rate of local relapse, given the known association of young age and higher local relapse rates (3). Our GG genotype carriers were also more predominant among African-American women, and there were slightly more ER-negative tumors among the GG genotype carriers. However, there were no differences in the local relapse rates between the African-American and white women in this study and no differences between ER-negative and ER-positive patients in this cohort. In addition, there was no correlation between the status of TP53BP1 polymorphisms and BRCA1/2 in this cohort. Therefore the higher local relapse rate could not be explained by an imbalance in any of these factors. Although it is possible that the amino acid changes and protein structure alteration in 53BP1 result in altered radiation sensitivity, it is apparent that further studies will be required to determine the biological and functional significance of these polymorphisms. Though significant on univariate analysis, the correlation did not hold on multivariate analysis. Given that our patient numbers and number of events were relatively small, larger sample sizes and validation studies will be required to determine the clinical significance of our observations.
These results do show, however, the potential for further exploration of these types of studies, evaluating outcomes of patients undergoing radiation therapy as a function of SNPs. Genes associated with DNA response such as TP53BP1, ATM, p53, BRCA1/2, and others are all potential candidate genes with known polymorphisms that warrant further study. Although polymorphisms in any given candidate gene may show only modest effects on radiation outcomes, patterns of polymorphisms in multiple genes are likely to yield clinically relevant results. The availability of SNP chips, which allow simultaneous evaluation of thousands of genes, makes this type of research currently available and attractive (71).
Because genetic changes are present in both tumor tissues and normal tissues, polymorphisms can be associated with tumor response as well as normal tissue response. In this study we did not evaluate normal tissue reactions or cosmesis, but that is another potential endpoint that could be evaluated in future studies. In addition, we are evaluating polymorphisms in other genes associated with radiation response to assess their prognostic potential. In the era of increasing attention on personalized medicine, this area remains a unique and exciting avenue of investigation in translational research.
This work was supported by the Breast Cancer Research Foundation.
Conflict of interest: none.