Ionizing radiation is an established risk factor for breast cancers [
24‐
27]. Several epidemiologic reports have suggested that an increased risk of cancer has continued for decades after exposure, and that a higher risk of certain types of cancers still persists in A-bomb survivors [
9‐
14]. Thus, a long-lasting health effect is considered to be a contributing factor in tumorigenesis in A-bomb survivors. We have recently demonstrated an association of oncogene amplification in breast cancers among A-bomb survivors with radiation exposure [
15], which can be regarded as being the results of positive selection during breast carcinogenesis. This finding suggests that A-bomb radiation may affect the development of oncogene amplification by inducing a higher level of GIN in breast cancers found in survivors. The current study was carried out to further confirm the enhanced GIN in A-bomb radiation-associated breast cancers using the aCGH method. The aCGH method is a quite useful technique to detect the DNA CNA as an indicator of GIN, which represents chromosomal loss and gain caused by radiation-induced DNA double-strand breaks [
16]. Unger et al. [
28] found DNA CNA pattern which is characteristic of radiation-induced papillary thyroid cancer in residents living in the vicinity of Chernobyl using the aCGH method.
Tissue samples from A-bomb survivors are considered to be extremely valuable biological materials with which to analyze the radiation signature or radiation-associated human health effects, particularly in low-dose and late exposures. The molecular analyses of carcinogenesis in A-bomb survivors require clinical data of individuals and biological materials with pathologic data of tumors. Our database, which consist of two independent databases: a clinical database providing exposure distance on Nagasaki survivors registered at our institute which was established in 1972 and a pathological database by the Nagasaki Tumor Tissue Registry (NTTR) which was established in 1974, allow us to obtain FFPE archival tissue samples resected from A-bomb survivors. For the genomic analyses, we confirmed the utility of FFPE archival tissue with FISH methods to detect gene amplification despite DNA degradation caused by fixation and long storage. In the present study, we conducted an aCGH analysis using tumor DNA extracted from FFPE archival breast cancer samples from A-bomb survivors. To our knowledge, this is the first attempt to perform an aCGH analysis with solid tumors from A-bomb survivors. The samples used in this study were very old, with ranges 14 to 43 years (with a mean of 25 years) in storage. The DLRSpread obtained was 0.26 to 1.05, with a mean of 0.54, which indicated the relatively lower quality of this experiment compared with that expected with DNA from fresh frozen tissue or peripheral blood lymphocyte. However, the status of
HER2 oncogene amplification based on aCGH result was highly concordant with the results of FISH that the sensitivity, specificity and accuracy were 77.8%, 93.8% and 88%, respectively, which were comparable to the results from former aCGH studies with FFPE archival tissue [
29,
30]. By contrast, the concordance was low for the status of
C-MYC oncogene amplification between the results from aCGH and FISH, with the sensitivity, specificity and accuracy being 18%, 75% and 48%, respectively. This discordance, especially in sensitivity, may result from the use of only three probes on the
C-MYC gene and a smaller change in amplification at the region including
C-MYC than the
HER2 gene. Our results suggest that the 60K×8 CGH array is a reliable technology to identify gene copy number aberration with definite changes.
Our aCGH analysis showed a great deal of variety in its amount and pattern of genomic alterations from case to case. In comparison with previous reports on breast cancers from general population, mean number of CNA in our cases seemed to be relatively small (mean: 12.2, range: 2-28) but recurrently affected regions (8q24.3, 17q12, 19p13.11, 1q21.2-q22: Additional file
5, Figure S3) found in our cases were concordant [
4,
7,
31‐
33]. However, direct comparisons of the current results with published results in aCGH are practically difficult because the results of aCGH analyses are greatly influenced by the array design and type of samples (e.g., fresh frozen or FFPE). A previous study of an aCGH analysis of radiation-induced and spontaneous rat mammary carcinoma indicated that the frequency of carcinoma having any CNA and the number of CNA in radiation-induced carcinoma were significantly greater than that observed in the spontaneous carcinoma [
34]. Another study of an aCGH analysis of premenopausal breast cancers in the residents from a nuclear fallout-contaminated area in Belarus did not show any significant differences or tendencies in the average number of total DNA CNA compared with matched control cases from Western New York, even though breast cancer from Belarus had significantly more average number of gains [
35]. These discrepancies may result from differences in the experimental models, since the former is a study of a simplified animal cancer model and the latter is an observational study of human cancer affected by many etiological factors. But the present study endorsed the former result with a tendency for breast cancer in A-bomb survivors to have a higher number of CNA (P = 0.16, Table
1, Additional file
1, Table S1). Furthermore, mean total length of CNA were also larger, if not significantly, in the A-bomb survivors than control group (P = 0.15, Table
1, Additional file
1, Table S1). Herein, we assumed the total length of CNA as an indicator of GIN because the amount of CNA represents the consequences of double-strand breaks, abnormal DNA damage repairs and gross rearrangements of chromosomes [
1,
16], and a consecutive changes of probes is considered to be much more important than a change of only one probe in such experimental model using high density probes and relatively noisy data. Since high histological grade, ER negative expression, early age of onset and
HER2 amplification were reported to be correlated with higher incidence of genomic aberrations [
4], we examined the correlation between the total length of CNA and clinicopathological factors, followed by multivariate analysis using analysis of covariance to evaluate the impact (effect) of A-bomb exposure, age at the time of diagnosis, HER2 and C-MYC amplification, histological grade, storage time, and DLRSpread on GIN, which have shown that the status of A-bomb exposure showed a significant correlation after the exclusion of confounding factor by the multivariate analysis (Table
4). Thus, we have demonstrated that breast cancers in A-bomb survivors harbored significant GIN independently of the effect of other clinicopathological factors.