Mutagenesis Advance Access originally published online on August 22, 2006
Mutagenesis 2006 21(5):313-320; doi:10.1093/mutage/gel035
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
On the difference of micronucleus frequencies in peripheral blood lymphocytes between breast cancer patients and controls
1 Department of Human Genetics, University of Ulm Germany 2 Department of Gynecology, University of Ulm Germany 3 Department of Gynecology, Technische Universitaet Muenchen Germany
Sporadic breast cancer patients as well as mutation carriers of the BRCA genes have a reduced DNA repair capacity compared to controls when assessed by the G0 micronucleus test (G0MNT) or by induced chromosomal aberrations. Since the individual MN frequencies vary widely and overlap between cases and controls it remained unclear which percentage of the cases should be considered to exhibit an increased radiosensitivity. We performed a similar case–control study and found a highly significant difference (P < 0.0001) between all breast cancer cases (N = 91) and female controls (N = 96) using descriptive statistics and ANOVA with adjustment for age. This difference also holds for baseline MN frequencies (P = 0.0006) and for subgroups of the patients similar to those without treatment (P < 0.0001). These results were confirmed in a second sample acquired at a different hospital. Since we are dealing in this analysis with two predefined groups (patients and controls), we calculated odds ratios (ORs) in order to assess the discriminative power of the G0MNT. These amounted to OR = 4.9 (P < 0.0001) for MN frequencies obtained by visual counting and ranged from OR = 11 (P < 0.0011) to OR = 22 (P < 0.0001) using automated counting. In order to overcome the problem of choosing a cut-off point inherent in ORs, receiver operating characteristic curves were calculated, which visualize specificity and sensitivity over the entire range of values and which characterize the discriminative power of a test by the area under the curve (AUC) (visual counting, baseline: AUC = 0.67; induced AUC = 0.75; automated counting: AUC > 0.88; evaluation sample: AUC > 0.73). We conclude that the G0MNT may be a useful tool to substitute the phenotype breast cancer in association and linkage studies and that it may be possible to develop a test useful in the diagnosis or risk assessment for breast cancer.
 |
Introduction |
|---|
 Top Introduction Materials and methodsResultsDiscussionReferences |
|---|
In recent years evidence from different sources has accumulated for an association of familial and sporadic breast cancer with DNA repair, suggesting that a reduced repair capacity might be genetically determined and might confer susceptibility to breast cancer (1
–3).
About two-thirds of breast cancer cases in high-risk families can be attributed to mutations in the BRCA 1 or 2 genes (4). These genes are required in the repair of DNA double-strand breaks and loss of function results in genome instability owing to multiple chromosome breaks. Surprisingly, heterozygous mutation carriers exhibit an increased radiosensitivity in their normal (non-tumor) somatic cells (5,6), an observation which may indicate that the breast cancer risk of mutation carriers can be explained directly by the reduced repair capacity without additional somatic mutation in the normal allele. Mutations in a third gene, ATM, determine the autosomal recessively inherited ataxia teleangiectasia, a chromosome breakage syndrome with high tumor susceptiblity (7). This gene has a key function in DNA repair and heterozygous mutation carriers, blood relatives of patients, have also a clearly increased risk for breast cancer (8).
Case–control studies in breast cancer patients revealed repeatedly, although not consistently, associations with DNA polymorphisms in genes relevant for different types of DNA repair, such as XRCC1, XRCC3, ATM and NBS1 (9–13). These observations conform to the assumption that breast cancer risk in sporadic cases may be due to low penetrance genes, which belong to the DNA repair pathways. Functional variants of these genes may give rise to a continuous variation of repair capacity and in consequence to variation of breast cancer risk. Furthermore a combined and additive effect of variants in different genes was seen in one study (3,14), an observation which agrees well with complex inheritance of the risk and continuous variation of repair capacity. Such continuous variation has directly been detected by measuring cell cycle delay after ionizing radiation and proved to be associated with allelic variants of XRCC1 and APE1 (10) on the one hand and breast cancer on the other hand.
There are numerous case–control studies on breast cancer in which the micronucleus test (MNT) has been used to compare repair capacity between cases and controls (15–17). In almost all of them, a proportion of 20–50% of the patients could be identified, which showed an increased MN frequency exceeding the 90th percentile of the distribution in the controls. This definition of an increased MN frequency is, however, not optimal for a variable with continuous variation and complex inheritance, because the two groups (cases and controls) have a different, although broadly overlapping distribution. Complex inheritance is generally assumed for DNA repair capacity as indicated by the numerous association studies carried out on cancer cases and genes involved in DNA repair. The data available for MN frequencies appear to be compatible with the assumption of continuous variation when cases are interpreted as a separate group with an overall slightly reduced repair capacity.
We have carried out a case–control study on breast cancer patients using the G0MNT and baseline MN frequencies to assess DNA repair capacity and analysed the data with ANOVA and receiver operating characteristic (ROC) curves. The results indicate that the MN frequencies are predictive for breast cancer and a high specificity and sensitivity can be achieved.
|
Materials and methods |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Study population
The entire study population (see Table I for further details) comprised 117 females without any cancer, and 111 breast cancer patients, including 6 known BRCA 1 mutation carriers. They were recruited in two phases. First, 96 controls, 85 cases and 6 carriers of BRCA 1 mutations, in whom the analysis was carried out by visual counting (Sample A). Aside from the mutation carriers, the patients did not fall into the criteria for hereditary breast (or ovarian) cancer and are referred to here as sporadic. We did not include patients who were actually under treatment. Hence, they were studied either in the time between diagnosis and start of the therapy or when therapy had been finished for at least 3 months, a time which ranged up to more than 5 years in a few patients. In the second round, patients and controls were all recruited in another hospital and analysis was performed by automated counting (evaluation Sample B). The female controls represented a mixture of hospital patients (for other reasons than cancer), healthy non-blood relatives of the patients and occasional controls, e.g. from self-support groups. The attempt to select the controls as age-matched pairs proved futile and, therefore, age was included in the analysis as a covariate. In our primary patient sample (Sample A), 85 sporadic breast cancer patients and 6 BRCA mutation carriers were investigated by visual counting. Later, an automated system based on image analysis became available and a subset of Sample A was re-analysed by automatic counting. The evaluation sample (Sample B) comprised 20 sporadic breast cancer patients and 21 female controls, and was collected from another hospital in order to evaluate whether the disease status could be predicted by means of individual MN frequencies of the blood donors. After coding of the probands, lymphocyte cultures were set up and micronuclei counted in at least 500 binucleated cells (BNCs) in a blinded fashion.
|
The first breast cancer sample (A) was split into different subgroups, namely BRCA 1 mutation carriers and patients with/without radiation/chemotherapy treatment. Age is well known to influence baseline MN frequencies (18
) but data concerning an age effect on induced MN frequencies are almost lacking. The influence of these potential confounding factors was determined. We did not consider habits such as smoking because these may influence baseline values only marginally (19) and are not expected to affect induced MN frequencies. The study has been approved by the Institutional Review Board of the University of Ulm and informed consent had been obtained from all participants before they were included in the MNT study.
Before starting the culture, heparinized blood samples were stored at room temperature. Cell culture, irradiation and preparation of slides were performed essentially as described by Rothfuss et al. (5). An aliquot of 0.3 ml blood was diluted in the ratio 1:9 with chromosome medium 1 A (Gibco BRL), supplemented with 2% PHA-M (Gibco BRL). The cultures, which were set up in order to determine the induced MN frequency, were exposed to gamma rays, total dose 2 Gy (source: 137CS, Gammacell 2000; Molsgaard Medical, Denmark). Thereafter, the cultures were incubated at 37°C for 44 h (after PHA stimulation), cytochalasin B (Sigma) was then added to a final concentration of 6 µg/ml. After an additional 24 h, cultures were harvested by centrifugation, treated with 0.56% KCl solution and fixed with methanol/acetic acid (5:1) mixed with an equal amount of 0.9% NaCl. Fixation with undiluted methanol/acetic acid (5:1) was then performed three times. The slides were coded and the entire analysis was carried out blindly. Air-dried slides were stained with 7% Giemsa in phosphate buffer (pH 7.0). MNs were counted in light microscope in 500–1000 BNCs according to the criteria suggested by an international consortium (20). The frequency of BNCs containing two MNs was also recorded.
For automated analysis, the slides were stained with DAPI (Sigma, Germany) 1 µg/ml in 4x SSC for 15 min, rinsed in distilled water, air dried and embedded with Vectashield (Vector Laboratories, Burlingame, CA). The slides were analysed in a fluorescence microscope (Zeiss, Axioplan2, Germany) using appropriate filter setting. BNCs were detected by image analysis and the number of MNs determined. A detailed description of the system and its validation for MN analysis has been published previously (21).
Statistical analysis
In the present study, we focused on the inter-individual variation of MN frequencies in sporadic breast cancer patients and controls. For many of the patients and controls, multiple single measurements were available which were combined into a single value (the mean of the proband) after normalization for storage time. Factors influencing the intra-individual variation of the single measurements were analysed in a previous paper (22) and there we demonstrated that the MN frequencies can be analysed assuming a normal distribution (instead of a Poisson or negative binomial). Therefore, in addition to descriptive statistics, t-tests were used to investigate the impact of status on MN frequencies and usual ANOVA with the proband's age as a covariate analysis of covariance (ANCOVA). Statistical analyses were carried out using Statistical Analysis System (SAS Institute Inc., Cary, NC).
Descriptive statistics and basic tests
In most blood donors we had more than one measurement (up to 8) and used the mean values per individual (blood donor) because we were looking for the MN frequency as an individual characteristic of the respective blood donor.
The time between drawing the blood sample and starting the blood culture influenced the radiation-induced MN rate systematically (22), and was taken out by normalization before combining the single measurements. Frequencies of spontaneous MN proved not to be influenced by blood sample storage time. Differences of means of MN frequencies between the subgroups were tested for significance with Student's t-test. Odds ratios (ORs) and their confidence intervals (CIs) were calculated. In order to determine ORs, the 75th percentile of controls was used to recode the MN frequencies into a binary variable for logistic regression. Such a cut-off value is arbitrary and was chosen in order to accommodate the broad overlap between breast cancer cases and controls in visual counting seen in descriptive statistics. For an assessment independent from a cut-off see ROC curves. For the analysis of data obtained by image analysis, only single measurements were available and were used without any correction or normalization.
ANCOVA
Both induced (corrected for storage time) and baseline MN frequencies were analysed using an ANOVA model, including the proband's disease status as the factor of interest, the proband's age as a covariate and the interaction between status and age. One aim of this analysis was to obtain mean values and a test for the difference in MN frequencies between cases and controls with adjustment for age which was not equally distributed among cases and controls. The adjusted mean values are given with 95% CI. In addition, we estimated the regression lines that predict MN frequencies from age separately for cases and controls, in order to calculate age-adjusted individual MN frequencies for each proband. The above analysis was carried out using the SAS procedure GLM.
ROC analysis
Using age-adjusted individual MN frequencies, a ROC analysis was performed to assess the potential of MN frequencies to discriminate between cases and controls. A ROC curve describes this ability by giving the sensitivity and 1 – specificity for the entire range of measurements. It is derived from a series of standard 2 x 2 tables by calculating sensitivity and specificity according to the standard definition for a cut-off, which is moved along the whole range of measurements. This procedure eliminates the problem of choosing a cut-off point. The area under the curve (AUC) is a measure for the appropriateness of a test and the more the ROC curve approaches the left upper corner of the graph, the more useful is the test for discrimination. ROC curve analysis is rather robust concerning data distribution but may, of course, be affected by confounding factors. These may operate on the level of single measurements, such as experimental conditions or scorer, and have been reported for the present dataset (22). Or else, they may be related to the individual proband, such as therapy, which we analyse here. The ROC analysis was carried out using the SAS procedure LOGISTIC.
|
Results |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
MN frequencies of patients and controls of Sample A are presented in Table II and Figure 1A. Comparing mean MN frequencies revealed a highly significant difference (P < 10–4) between controls and sporadic breast cancer patients (Table II). The frequency of cells with two MN are expected to be correlated with the MN frequency and it is not surprising that we saw similar differences (Table II).
|
|
Baseline MN frequency
Lymphocyte cultures without irradiation revealed differences between cases and controls similar to those from irradiated cultures. The breast cancer cases demonstrated a significantly higher spontaneous MN frequency compared to controls (P < 10–4; Table II), but this difference is confounded by the unequal age distribution of the controls. However, after correction for age using ANCOVA the difference (age corrected mean values: 47.8/31.3) is still significant (P = 0.007). In contrast to the induced MN frequencies, there was a difference between patients with and without therapy regarding the frequency of spontaneous MN (P = 0.01, Table II). Breast cancer patients treated by irradiation showed higher MN frequencies than those without treatment.
Separation of cases and controls
Since unselected breast cancer cases and controls had different mean MN frequencies, we asked whether vice versa the MN frequencies could be used to allocate individuals to these groups. In the literature, this type of information is usually presented as the percentage of cases with an MN frequency above the 90th percentile of the controls. Such a cut-off is arbitrary when the probands belong to two predefined groups and from experience we had chosen here a cut-off at the 75th percentile of the controls because of the broad overlap of the MN frequencies between cases and controls. These ORs are presented in Table III. Including all breast cancer patients, the OR obtained for MN frequencies by visual counting was 4.9 and was highly significant (P < 10–4). As reported previously (22), the results from visual counting were impaired probably by an unnoticed change in scoring criteria towards the end of the study. This may be the reason that the other ORs are much higher. In order to exclude any effects superimposed by patient treatment, we compared treated and untreated patients with controls. Both subgroups yielded almost identical ORs.
|
MN counts by image analysis
When automated counting by image analysis became available, fixed cells from irradiated cultures of 32 patients and 21 controls were available and re-analysed from Sample A (Table I). This reinvestigation yielded considerably lower overall MN frequencies in automated counting. In consequence, the cut-off referring to the 75th percentile was set to 120 MN/1000 BNC [for details see (21
)]. With the automated analysis the OR (= 22.9, P < 10–4; Table III) increased considerably compared with visual counting (OR = 4.0, P = 0.0089) of slides from the same patients.
Evaluation sample
In order to further evaluate the discriminative power between cases and controls we set up a new sample (B). The cut-off point for this analysis was taken from the previously analysed sample (A) (Figure 1B). The corresponding ORs amounted to 13.2 (P = 0.001; Table III).
ANCOVA
The results from ANCOVA are presented in Tables II, IV and V for induced and baseline MN frequencies. The baseline MN frequencies depended on age at least in the controls, and the over-representation of young controls could confound the impact of the disease status on MN frequencies. However, there exists obviously an age-independent effect for status.
|
|
After adjustment for age, the results (Table II) were very similar to those with descriptive statistics. The only significant effect on the induced MN frequency was found for the disease status (Table IV). Cases with chemotherapy and mutation carriers were exluded. Age did not have an effect on the induced MN frequencies of the controls and on the spontaneous MN frequencies of cases (Table V). Therefore, the age distribution of the controls did not matter for the induced MN frequencies.
ROC analysis on age-corrected measurements obtained by visual counting (Sample A) yielded the curves in Figure 2A (baseline MN frequencies) and Figure 2B (induced MN frequencies). The AUC is 0.68 for the baseline frequencies and 0.75 for the induced frequencies. An AUC has a range from 0.5 (pure chance) to 1.0 (prediction without any errors). ROC curves for automated measurements in the subsample of A and for Sample B are presented in Figure 2 (C and D). The respective AUCs amount to 0.88 (Sample A) and 0.73 (Sample B).
|
|
Discussion |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Based on a case–control study, we have demonstrated a significant difference between sporadic breast cancer patients and controls in the G0MNT. The difference of mean MN frequencies could be observed both with and without irradiation of the lymphocyte cultures. All measurements (induced MN frequency and baseline MN frequencies) in two independent subsets of breast cancer patients (Samples A and B), plus subgroups of these yielded consistent results. Our results are an extension of similar recent studies (15
,16), in that they apply to the entire group of sporadic breast cancer cases, which had not been selected according to any special criteria such as age or family history. In the earlier studies, radiosensitivity was defined by the cut-off (mean ± 2 SD) derived from the controls. These studies all identified a proportion of the patients with an increased radiosensitivity. This approach is typical for biomonitoring where exposed individuals have to be identified among the controls, and hence within one sample. Here we have, however, two predefined groups, cases and controls, and in consequence we asked how these differ using standard statistics and ANOVA. Once the difference in MN frequencies between the groups of breast cancer patients and controls had been established, we asked if the MN frequencies could be used to allocate blood donors to either of these groups. A typical way to answer this question is to calculate ORs by logistic regression after choosing a cut-off point, which in principle is arbitrary. We decided to use the 75th percentile of the controls as the cut-off. The resulting ORs were in the range of 5–10. Automated counting of some patients and controls, that had previously been counted visually, increased the OR to 22.9. The same absolute cut-off was used to study a new sample (B) in order to verify this capability independently. After decoding, an OR of 13.2 was calculated resulting in 85% of patients and 70% of the female controls being correctly classified.
Another way to circumvent the problem to pre-specify a threshold is to construct ROC curves and characterize the test by the AUC. This is a widely used procedure to assess the predictive power of quantitative clinical tests. From an inspection of the curves (Figure 2) it is immediately clear why the threshold of 90th percentile from controls (corresponding to specificity = 0.9) yields highly variable results as reported in the literature (16). The slope of the curve at these values is very steep and, hence, minor changes (differences) in mean values or standard error of MN frequencies produce a large effect on the detection rate. Although the G0MNT in our study produced a good differentiation between cases and controls one has to be aware of the fact that (i) we do not have prospective data (the test was not carried out on healthy probands and evaluated by determining those who later develop the disease, (ii) the patients were affected by breast cancer and have been diagnosed by other means, and (iii) we do not know about other conditions (including other cancers) which might also be associated with increased MN frequencies. In consequence, the G0MNT may have the potential to contribute useful information about the presence of breast cancer or the risk to develop breast cancer but, even if so, a lot of additional studies are required including prospective ones.
Effects of confounding factors?
Compared with several earlier studies (15,16,23), we were able to take a number of potentially confounding factors into consideration. In addition, we have demonstrated that the difference between patients and controls, as reported here, is modified, but neither caused nor eliminated by genotoxic treatment of the patients, one of the strongest known occupational factors.
Since treatment of breast cancer usually includes genotoxic therapies (irradiation and/or chemotherapy), it was mandatory to determine their influence on induced and baseline MN frequencies. This issue could be assessed by investigating breast cancer patients with/without treatment separately, an approach, which has rarely been used by other studies. Compared with all treated patients, regardless of the type of treatment, the untreated patients had slightly lower mean frequencies of induced MN (P = 0.4). In contrast, the baseline frequencies of untreated patients were significantly lower (P = 0.01). Nevertheless, mean MN frequencies from untreated patients are clearly different between patients and controls.
Genotoxic treatment has repeatedly been shown to increase baseline MN frequencies (24–26). Its effects are stronger than most other occupational factors, such as smoking, whose effect is usually hard to detect (19,27). Furthermore, occupational factors may be expected to influence primarily baseline, but not induced, MN frequencies, as has been demonstrated for smoking (19). Age and gender are other variables known to influence baseline MN frequencies (18). Gender has been excluded here by restricting the analysis to females and age was included in the ANOVA model. It has a major effect only on the baseline MN frequencies of the controls.
Sporadic breast cancer: a repair defect ?
There is no direct evidence for the mechanism by which the increased induced MN frequency and the increased baseline frequencies in peripheral lymphocytes of the patients could arise. Two possibilities may be considered:
- The lymphocytes grown from patients might represent a different, more sensitive subpopulation of lymphocytes compared with those from controls. Such a shift in cell population would then be expected to be correlated with the cancer. This idea might arise from the discrepancy between MN frequencies from lymphocyte cultures and lymphoblastoid cell lines which do not show an increased MN frequency when derived from BRCA 1 mutation carriers. However, this hypothetical shift in cell population would also be seen in
30% of controls and it would persist in patients after removal of the cancer (disease-free interval of 6 months to 5 years, data not shown). The speculation that another type or population of lymphocytes might be stimulated in cancer patients and not in controls is therefore unlikely.
- The second possibility would be that the same lymphocytes are studied and that these lymphocytes have different properties with respect to DNA damage and repair. This difference can be detected by the G0MNT.
With respect to this second possibility, Baeyens et al. (16) and the present study have given evidence that radiosensitivity does not differ between sporadic breast cancer patients and BRCA 1 or 2 mutation carriers who have a known repair defect (5,6). In the present study, six BRCA 1 mutation carriers were included who exhibited induced MN frequencies within the range of sporadic breast cancer cases (Table II). The baseline frequencies are elevated in the sporadic cases, not in the mutation carriers, as seen in our study and reported previously (16). This difference in spontaneous MN frequencies between sporadic cases and mutation carriers may indicate that the mechanisms leading to increased MN frequencies are similar but not identical in these two groups. Hence, it is tempting to speculate that the increased frequency of MNs observed in lymphocytes from sporadic breast cancer patients might also be due to a DNA repair defect, resulting in a cellular phenotype of increased radiosensitivity.
Evidence for potentially predisposing genes, resulting in a phenotype of reduced DNA repair capacity, has been found in several studies demonstrating deficient DNA repair in healthy relatives of breast cancer patients as measured by various end points (1,23,28). Attempts to identify a mode of inheritance were compatible with a complex two-gene model, but remained inconclusive (1). These observation agree with epidemiological data (29), which are compatible with a contribution of genetic factors involved in the etiology of breast cancer beyond BRCA 1 and 2. Modelling of these factors favoured a complex or polygenic (29) inheritance slightly more than an autosomal recessive one. It remains to be seen if family studies with the G0MNT as a quantitative trait will be able to reveal a better defined inheritance pattern. Another quantitative parameter related to DNA repair, retardation of S-phase, has recently been shown to be associated with variants of DNA repair genes (10).
Following the idea that sporadic breast cancer may be due to an impaired capacity of DNA repair, several association studies have been carried out between variants of candidate genes from the respective DNA repair pathways and breast cancer. Their effects on the occurrence of breast cancer ranged from small (30) to none (31,32). This result is not surprising when one considers that not all women with reduced repair capacity develop breast cancer, as is suggested by the incomplete penetrance of mutations in the BCRA 1 or 2 genes. Therefore, it would be interesting to see whether association studies using MN frequencies as a phenotype instead of the presence or absence of breast cancer yield more conclusive results. For example, stratification of breast cancer cases by MN frequencies in an association study with variants of DNA repair genes could reveal a correlation which otherwise remains undetected and help to identify the relevant genes. Overall, our data suggest that the G0MNT may be able to indicate a risk for breast (and may be some other) cancer and may help to identify the relevant genes.
|
Acknowledgments |
|---|
We are most grateful to all the patients and controls for their donations of blood samples and the self-support groups for active collaboration. We would also like to thank the Bethesda Geriatric Hospital, Ulm, and the Departments of Neurology and Internal Medicine at the University of Ulm for collaboration. The excellent technical assistance of Karina Eiwen and Ingrid Peter is highly appreciated. This work was supported by the Deutsche Krebshilfe, grant number 70-2680-VO2.
|
Notes |
|---|
*To whom correspondence should be addressed at: Department of Human Genetics, University of Ulm, 89069 Ulm, Germany. Tel: +49 731 50023430; Fax: +49 731 50023438; Email: walther.vogel{at}uni-ulm.de
|
References |
|---|
|
TopIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
-
1. Roberts S.A., Spreadborough A.R., Bulman B., Barber J.B., Evans D.G., Scott D. (1999) Heritability of cellular radiosensitivity: a marker of low-penetrance predisposition genes in breast cancer? Am. J. Hum. Genet. 65:784–794.[CrossRef][Web of Science][Medline]
2. Khanna K.K. and Jackson S.P. (2001) DNA double-strand breaks: signaling, repair and the cancer connection. Nat. Genet. 27:247–254.[CrossRef][Web of Science][Medline]
3. Smith T.R., Levine E.A., Perrier N.D., Miller M.S., Freimanis R.I., Lohman K., Case L.D., Xu J., Mohrenweiser H.W., Hu J.J. (2003) DNA-repair genetic polymorphisms and breast cancer risk. Cancer Epidemiol. Biomarkers Prev. 12:1200–1204.
4. Antoniou A., Pharoah P.D., Narod S., et al. (2003) Average risks of breast and ovarian cancer associated with BRCA1 or BRCA2 mutations detected in case series unselected for family history: a combined analysis of 22 studies. Am. J. Hum. Genet. 72:1117–1130.[CrossRef][Web of Science][Medline]
5. Rothfuss A., Schutz P., Bochum S., Volm T., Eberhardt E., Kreienberg R., Vogel W., Speit G. (2000) Induced micronucleus frequencies in peripheral lymphocytes as a screening test for carriers of a BRCA1 mutation in breast cancer families. Cancer Res. 60:390–394.
6. Baeyens A., Thierens H., Claes K., Poppe B., De Ridder L., Vral A. (2004) Chromosomal radiosensitivity in BRCA1 and BRCA2 mutation carriers. Int. J. Radiat. Biol. 80:745–756.[CrossRef][Web of Science][Medline]
7. Meyn M.S. (1999) Ataxia-telangiectasia, cancer and the pathobiology of the ATM gene. Clin. Genet. 55:289–304.[CrossRef][Web of Science][Medline]
8. Thompson D., Duedal S., Kirner J., McGuffog L., Last J., Reiman A., Byrd P., Taylor M., Easton D.F. (2005) Cancer risks and mortality in heterozygous ATM mutation carriers. J. Natl Cancer Inst. 97:813–822.
9. Steffen J., Varon R., Mosor M., et al. (2004) Increased cancer risk of heterozygotes with NBS1 germline mutations in Poland. Int. J. Cancer 111:67–71.[CrossRef][Web of Science][Medline]
10. Hu J.J., Smith T.R., Miller M.S., Mohrenweiser H.W., Golden A., Case L.D. (2001) Amino acid substitution variants of APE1 and XRCC1 genes associated with ionizing radiation sensitivity. Carcinogenesis 22:917–922.
11. Hu J.J., Mohrenweiser H.W., Bell D.A., Leadon S.A., Miller M.S. (2002) Symposium overview: genetic polymorphisms in DNA repair and cancer risk. Toxicol. Appl. Pharmacol. 185:64–73.[CrossRef][Web of Science][Medline]
12. Koren M., Kimmel G., Ben Asher E., Gal I., Papa M.Z., Beckmann J.S., Lancet D., Shamir R., Friedman E. (2006) ATM haplotypes and breast cancer risk in Jewish high-risk women. Br. J. Cancer 94:1537–1543.[CrossRef][Web of Science][Medline]
13. Zhang Y., Newcomb P.A., Egan K.M., et al. (2006) Genetic polymorphisms in base-excision repair pathway genes and risk of breast cancer. Cancer Epidemiol. Biomarkers Prev. 15:353–358.
14. Smith T.R., Miller M.S., Lohman K., Lange E.M., Case L.D., Mohrenweiser H.W., Hu J.J. (2003) Polymorphisms of XRCC1 and XRCC3 genes and susceptibility to breast cancer. Cancer Lett. 190:183–190.[CrossRef][Web of Science][Medline]
15. Scott D., Barber J.B., Levine E.L., Burrill W., Roberts S.A. (1998) Radiation-induced micronucleus induction in lymphocytes identifies a high frequency of radiosensitive cases among breast cancer patients: a test for predisposition? Br. J. Cancer 77:614–620.[Web of Science][Medline]
16. Baeyens A., Thierens H., Claes K., Poppe B., Messiaen L., De Ridder L., Vral A. (2002) Chromosomal radiosensitivity in breast cancer patients with a known or putative genetic predisposition. Br. J. Cancer 87:1379–1385.[CrossRef][Web of Science][Medline]
17. Baeyens A., Van Den B.R., Makar A., Thierens H., De Ridder L., Vral A. (2005) Chromosomal radiosensitivity in breast cancer patients: influence of age of onset of the disease. Oncol. Rep. 13:347–353.[Web of Science][Medline]
18. Bonassi S., Fenech M., Lando C., et al. (2001) HUman MicroNucleus project: international database comparison for results with the cytokinesis-block micronucleus assay in human lymphocytes: I. Effect of laboratory protocol, scoring criteria, and host factors on the frequency of micronuclei. Environ. Mol. Mutagen. 37:31–45.[CrossRef][Web of Science][Medline]
19. Bonassi S., Neri M., Lando C., Ceppi M., Lin Y.P., Chang W.P., Holland N., Kirsch-Volders M., Zeiger E., Fenech M. (2003) Effect of smoking habit on the frequency of micronuclei in human lymphocytes: results from the Human MicroNucleus project. Mutat. Res. 543:155–166.[CrossRef][Web of Science][Medline]
20. Fenech M., Chang W.P., Kirsch-Volders M., Holland N., Bonassi S., Zeiger E. (2003) HUMN project: detailed description of the scoring criteria for the cytokinesis-block micronucleus assay using isolated human lymphocyte cultures. Mutat. Res. 534:65–75.[Web of Science][Medline]
21. Varga D., Johannes T., Jainta S., Schuster S., Schwarz-Boeger U., Kiechle M., Garcia B.P., Vogel W. (2004) An automated scoring procedure for the micronucleus test by image analysis. Mutagenesis 19:391–397.
22. Patino-Garcia B., Hoegel J., Varga D., Hoehne M., Michel I., Jainta S., Kreienberg R., Maier C., Vogel W. (2006) Scoring variability of micronuclei in binucleated human lymphocytes in a case–control study. Mutagenesis 21:191–197.
23. Burrill W., Barber J.B., Roberts S.A., Bulman B., Scott D. (2000) Heritability of chromosomal radiosensitivity in breast cancer patients: a pilot study with the lymphocyte micronucleus assay. Int. J. Radiat. Biol. 76:1617–1619.[CrossRef][Web of Science][Medline]
24. Elsendoorn T.J., Weijl N.I., Mithoe S., Zwinderman A.H., Van Dam F., De Zwart F.A., Tates A.D., Osanto S. (2001) Chemotherapy-induced chromosomal damage in peripheral blood lymphocytes of cancer patients supplemented with antioxidants or placebo. Mutat. Res. 498:145–158.[Web of Science][Medline]
25. Trenz K., Lugowski S., Jahrsdorfer U., Jainta S., Vogel W., Speit G. (2003) Enhanced sensitivity of peripheral blood lymphocytes from women carrying a BRCA1 mutation towards the mutagenic effects of various cytostatics. Mutat. Res. 544:279–288.[CrossRef][Web of Science][Medline]
26. Lee T.K., O'Brien K.F., Naves J.L., Christie K.I., Arastu H.H., Eaves G.S., Wiley A.L. Jr, Karlsson U.L., Salehpour M.R. (2000) Micronuclei in lymphocytes of prostate cancer patients undergoing radiation therapy. Mutat. Res. 469:63–70.[Web of Science][Medline]
27. Hoffmann H., Hogel J., Speit G. (2005) The effect of smoking on DNA effects in the comet assay: a meta-analysis. Mutagenesis 20:455–466.
28. Scott D. (2000) Chromosomal radiosensitivity, cancer predisposition and response to radiotherapy. Strahlenther. Onkol. 176:229–234.[CrossRef][Web of Science][Medline]
29. Antoniou A.C., Pharoah P.D., McMullan G., Day N.E., Stratton M.R., Peto J., Ponder B.J., Easton D.F. (2002) A comprehensive model for familial breast cancer incorporating BRCA1, BRCA2 and other genes. Br. J. Cancer 86:76–83.[CrossRef][Web of Science][Medline]
30. Hu J.J., Mohrenweiser H.W., Bell D.A., Leadon S.A., Miller M.S. (2002) Symposium overview: genetic polymorphisms in DNA repair and cancer risk. Toxicol. Appl. Pharmacol. 185:64–73.[CrossRef][Web of Science][Medline]
31. Rafii S., O'Regan P., Xinarianos G., Azmy I., Stephenson T., Reed M., Meuth M., Thacker J., Cox A. (2002) A potential role for the XRCC2 R188H polymorphic site in DNA-damage repair and breast cancer. Hum. Mol. Genet. 11:1433–1438.
32. Smith T.R., Levine E.A., Perrier N.D., Miller M.S., Freimanis R.I., Lohman K., Case L.D., Xu J., Mohrenweiser H.W., Hu J.J. (2003) DNA-repair genetic polymorphisms and breast cancer risk. Cancer Epidemiol. Biomarkers Prev. 12:1200–1204.
Received on May 9, 2006; revised on June 21, 2006; accepted on July 12, 2006.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||