Allele Imbalance, or Loss of Heterozygosity, in Normal Breast Epithelium of Sporadic Breast Cancer Cases and BRCA1 Gene Mutation Carriers Is
http://www.100md.com
《临床肿瘤学》
the Departments of Pathology and Laboratory Medicine and Medicine, Boston University School of Medicine and Boston Medical Center
Department of Biostatistics, Boston University School of Public Health
Department of Adult Oncology, Dana-Farber Cancer Institute, Boston, MA
ABSTRACT
PURPOSE: Normal-appearing breast epithelium can contain genetic abnormalities, including allele imbalance (AI), also referred to as loss of heterozygosity. Whether abnormalities are associated with cancer or cancer risk is unknown.
PATIENTS AND METHODS: We performed a miniallelotype, using 20 microsatellites, on each of 460 histologically normal, microdissected breast terminal ducto-lobular units (TDLUs) from three groups of women: sporadic breast cancer patients (SP; n = 18), BRCA1 gene mutation carriers (BRCA1; n = 16), and controls undergoing reduction mammoplasty (RM; n = 18). We analyzed the results using Fisher's exact tests, logistic regression, and generalized estimating equations.
RESULTS: AI was increased three-fold in SP and BRCA1 groups compared with RM. Both the number of TDLUs with AI increased (eight [5%] of 162 in the RM group compared with 24 [15%] of 162 in the SP and 22 [16%] of 136 in the BRCA1 groups; P = .0150), and the proportion of patients with AI increased (five [28%] of 18 in the RM group compared with 15 [83%] of 18 in the SP and 13 [81%] of 16 in the BRCA1 groups; P = .0007). The adjusted odds ratios (OR) for AI in TDLU increased in SP (OR = 15.5) and BRCA1 (OR = 13.7) patients compared with RM (P = .0025). This result was particularly evident on chromosome 17q (P = .0393), where more AI was seen in BRCA1 (OR = 12.4) than in SP (OR = 4.9) patients or RM controls.
CONCLUSION: Increased prevalence of AI in normal-appearing epithelium is associated with breast cancer and increased breast cancer risk. The increased prevalence may reflect dysregulation, even in normal-appearing epithelium, of genomic processes contributing to cancer development. The clinical significance of genetic alterations in the subset of controls remains to be determined.
INTRODUCTION
Although more than 200,000 women in the United States are diagnosed annually with breast cancer,1 knowledge of the earliest stages of breast carcinogenesis remains limited. Breast cancers are morphologically and genetically heterogeneous, and no signature abnormalities characterize all tumors, making identification of cancer precursors difficult. Cell lines and animal models are useful tools but may not mirror early stages of human disease, and investigation of human tissue in vivo is technically challenging. However, a fruitful approach has been examination of allele imbalance (AI), or loss of heterozygosity (LOH), which can be evaluated from formalin-fixed tissue specimens. AI can reflect insufficiency of several different mechanisms of DNA repair and replication (see review in Thiagalingam et al2), although the pertinent mechanism in humans is uncertain. Regardless, AI is characteristic of breast cancers3 and is hypothesized to reflect genetic instability. Recurrent sites of AI are thought to occur at genes important for breast tumorigenesis and may identify pathways that are potentially important for both prevention and treatment.
More recently, AI has been reported in normal-appearing epithelium4-7 from women with and without breast cancer and in tissue both adjacent to and distant from primary cancers. However, the prevalence of AI in normal breast tissue of women at usual cancer risk is unknown, and the central issue of whether increased AI is associated with an increased risk of cancer has not been addressed.
As far as we are aware, direct comparisons of normal tissue from different groups of patients using a consistent panel of markers and adjusting for variables that could affect rates of AI have not been reported. Previous studies have generally examined small numbers of normal epithelial samples and/or patients and used disparate markers, thus limiting comparison between groups and between studies. Identification of recurrent or frequent AI in histologically normal tissue may have important implications for understanding the evolution of breast malignancy, for predicting local disease recurrence, and for assessing risk of new cancers.
We hypothesized that normal-appearing breast epithelium of high-risk individuals might contain increased occult genetic abnormalities compared with control epithelium, indicating a predisposition or vulnerability of the high-risk epithelium to malignant transformation. To test this hypothesis, we used a genetic fingerprinting approach that we developed previously to scan numerous genomic sites for AI. We applied this technique to microdissected, histologically normal breast epithelium from three groups at varying breast cancer risk: women with sporadic breast cancer (SP), women with hereditary mutation of BRCA1, and a control group of women at usual cancer risk undergoing reduction mammoplasty (RM). The mammoplasty specimens would establish the prevalence of AI in normal breast epithelium. In addition, distinct pathways in the earliest stages of carcinogenesis would be suggested by the observation of a distinct pattern of AI in the BRCA1 group.
PATIENTS AND METHODS
Specimen Acquisition
Institutional review board approval was obtained at Boston University Medical Center and the Dana-Farber Cancer Institute. Formalin-fixed paraffin-embedded blocks of breast tissue not needed for diagnosis from 18 bilateral RM surgeries and 18 SP surgeries from patients 50 years of age were selected at random from the department of pathology archives. (Data from six SP patients had been included in a previous report comparing AI in normal epithelium and coexisting cancers8). RM cases are screened to eliminate those with a personal or strong family history of breast cancer. The available history and demographics suggest that although no testing was performed, few of the SP cases would have heritable BRCA1 mutations. In nine of 18 RM cases, terminal ducto-lobular units (TDLUs) were taken from both the left and right reduction specimens. In one of 18 SP cases, bilateral surgeries were performed, and TDLUs were taken from both sides. In the remaining 17 cases, normal epithelium was obtained from blocks at sufficient distance from the tumor that they lacked any histologically malignant cells, but their precise distance from the tumor was indeterminate. Formalin-fixed paraffin-embedded blocks of breast tissue not needed for diagnosis were selected from the breast surgeries of 16 patients 50 years of age carrying deleterious, heritable germline BRCA1 mutation. Five of these 16 patients had bilateral surgeries. Of these 21 breast operations, 10 were performed for cancer therapy, and 11 were from biopsies of benign lesions or prophylactic mastectomies in which cancers were not identified. As with the SP cases, epithelium was taken from blocks removed from the tumor. Criteria for inclusion of individual cases were five or more normal TDLUs (microdissected from any number of blocks) analyzed with five or more informative markers and surgery performed after 1990.
Microdissection and Polymerase Chain Reaction
A single pathologist (A.M.) identified histologically normal TDLUs. Each sample was isolated using laser microdissection and DNA was extracted, as previously reported.8 DNA from each individual TDLU was analyzed with a mini-allelotype consisting of 20 microsatellite markers located on nine chromosome arms: 1p: D1s468; 1q: D1s549, D1s213; 3p: D3s1283; 7q: D7s486; 11p: TH01, D11s2071; 11q: PYGM, D11s1818, D11s1819; 16q: D16s265, D16s402, D16s413, D16s512; 17p: TP53, D17s796, D17s525; 17q: D17s1290, D17s579, and D17s855. Primers were purchased from ResGen Invitrogen Corp (Huntsville, AL). Six multiplexed polymerase chain reactions were performed, electrophoresed, and exposed to film, as previously reported.8
Determination of AI
All autoradiographs were scanned by laser densitometer (Molecular Dynamics, Sunnyvale, CA) to calculate allele ratios. The normal allele ratio at each marker in each patient was determined by averaging all of the marker's allele ratios from that patient's normal tissue samples (ie, multiple separate TDLUs). AI was defined at heterozygous loci as previously described8: in other words, an imbalance of allele intensities greater than 25% (when (n1)(t2)/(n2)(t1) > than 1.33 or < 0.75, where n1 = normal samples' larger allele, n2 = normal samples' smaller allele, t1 = test sample's larger allele, t2 = test sample's smaller allele). This degree of allele imbalance indicates that a substantial proportion of the cells within this sample contains the same DNA abnormality and likely represents the presence of a clonal population. We used a less stringent cutoff than what is commonly applied to tumor DNA because the proportion of a TDLU composed of any genetically aberrant clone is unknown and may well be lower than the proportion in histologically malignant tissue. For a marker to be identified as showing AI, equivalent results had to be demonstrated in at least two independent replicate reactions, and no polymerase chain reactions could produce a normal ratio or contradictory imbalances. Using these criteria, we have estimated previously that our rate of detecting false-positive imbalances is between 2% and 4%.8 At some loci, allele imbalance may reflect increased copy number; at others, it may represent allele loss. Distinguishing between these possibilities is important conceptually, but would not alter data analysis.
Statistical Analyses
We first evaluated whether each patient had any AI. To compare the proportion of patients with any AI across the three groups, we used Fisher's exact test. To evaluate potential confounding by patient age and block age, we performed analysis of variance to examine mean patient and block ages among the three groups. To adjust for potential differences owing to these confounders (patient and block age), we performed logistic regression, weighting for the number of observations for each patient. We next evaluated whether the degree of AI differed among the three groups at each chromosome arm by summarizing information across all TDLUs for each chromosome arm and evaluating whether an arm showed any evidence of AI on any TDLU (yes or no). For each arm, Fisher's exact test was used to compare the relative prevalence of AI in each group. We also used weighted logistic regression for those arms where there was at least one patient with AI in each group (chromosomes 1q, 11p, 11q, and 17q), again weighting by the number of observations for each patient. Finally, we examined the average number and proportion of AI per duct and per informative marker. For comparisons across the groups in these analyses, we applied generalized estimating equations to account for the potential correlation among observations coming from the same person and weighted by the number of observations per duct, when appropriate.
RESULTS
Group Characteristics
The three groups (RM, SP, and BRCA1) were well-matched for patient age, number of independent cases, samples per case, samples per group, and the proportion of informative markers per group (Table 1). Differences among groups were noted for age of tissue blocks, because RM cases tended to have more recent blocks. Because increased age of blocks might increase detection of artifactual AI, we assessed whether block age was associated with presence of AI and found that it was not (P = .4, a t test), even though block age did vary across groups. Although patient age did not vary across groups, we tested whether it was associated with AI and found that it was not. Although the SP group did not have genetic testing performed, their estimated prevalence of a BRCA1 mutation based on age at diagnosis was approximately 7%,9 implying that one or two SP patients may be misclassified. Tumors in the SP group comprised a distribution of histologic and immunohistochemical features typical of an unselected series10: one lobular and 17 ductal carcinomas; two cases were grade 1/3, the others equally divided between grade 2/3 and grade 3/3; 70% of cases were positive for estrogen receptors and progesterone receptors. To the extent possible, we determined prior exposure to cancer treatment. Three of 10 BRCA1 mutation carriers with prior cancer had had chemotherapy, none had had radiation, and none of the 18 SP patients had had either chemotherapy or radiation.
Comparison of AI Across Groups
AI was seen in individual TDLUs of all groups (a representative example is shown in Fig 1). The mean allele ratio of these imbalances was approximately 0.6 in all groups, as shown in Table 2. (For ratios > 1, the reciprocal value was used for this calculation). All AIs occurred at single sites; in other words, other informative sites on the chromosome arm remained heterozygous, with one exception. (This exception was an RM case with eight AIs, five occurring in a single TDLU containing AIs of all informative markers on chromosome arms 1q and 16q).
As shown in Table 2 and Figure 2, we found differences in measurements of AI across groups. In general, a small number of abnormalities were detected in the RM group, and a larger number in the SP and BRCA1 groups. Both the number of TDLUs with AI and the proportion of patients with AI were increased nearly three-fold in the SP and BRCA1 groups compared with controls. Specifically, 24 (15%) of 162 TDLUs from the SP group, 22 (16%) of 136 TDLUs from the BRCA1 group, and eight (5%) of 162 TDLUs from the control group contained AI (P = .0150). (Eleven [18%] of 62 TDLUs from prophylactic or benign BRCA1 surgeries and 11 [15%] of 74 TDLUs from therapeutic BRCA1 surgeries had AI). Similarly, 15 (83%) of 18 SP patients, 13 (81%) of 16 BRCA1 patients, and five (28%) of 18 RM patients demonstrated AI (P = .0007). Among the five RM cases with AI, eight (20%) of 41 TDLUs had allele imbalances, a proportion that resembles what was seen in the SP and BRCA1 groups (16% and 15%, respectively).
After adjusting for block age and weighting for number of TDLUs per patient, a dramatically increased odds ratio (OR) for AI was present in both the SP (OR = 15.5; 95% CI, 2.6 to 92.4) and BRCA1 (OR = 13.7; 95% CI, 2.3 to 79.8) groups (P = .0025) compared with the RM group. The SP and BRCA1 groups were similar to each other in overall AI (P = .892, weighted logistic regression). We also noted that the mean number of AIs per TDLU and the mean number of AIs per informative marker seemed to be increased in the SP and BRCA1 groups compared with controls, although these measures did not achieve statistical significance.
To assess whether AI reflects a widespread characteristic of a patient's breast tissue or a focal abnormality of a single TDLU, we considered the distribution of AI within TDLUs and within patients. First, we noted that AI did not seem to cluster in particular TDLUs. Seventeen of the 52 patients had two or more AIs (three RM patients, six SP patients, and eight BRCA1 patients); in five of 17 patients, multiple AIs occurred in a single TDLU, and in 15 of 17 patients, they occurred in multiple TDLUs (some cases had both patterns). Next, we found that identical AIs were not commonly seen in multiple TDLUs. In only three of the 17 cases with multiple AI (one from each group) was the same AI present in two TDLUs. Third, we noted that in the RM cases, detection of AI did not seem to be influenced by whether samples from one or both sides were examined. AI was seen in three of nine cases with unilateral and two of nine cases with bilateral tissue samples available. Fourth, we found that proximity of a TDLU and cancer did not seem to influence the prevalence of TDLU with AI. In the 18 SP cases, there were 162 TDLUs with a total of 25 AIs. Fifty-six (35%) of 162 TDLUs were on the same block as the cancer, and 106 (65%) of 162 were on a different block. Six (24%) of 25 AIs were in TDLUs on the same block as the cancer, and 19 (76%) of 25 AIs were on a different block. Finally, we found that in the BRCA1 cases, detection of AI seemed unrelated to whether the tissue was from a prophylactic or therapeutic surgery. AI was seen in seven of 11 prophylactic and seven of 10 therapeutic surgeries. Fourteen AIs were detected in the prophylactic specimens, and 12 AIs were detected in the therapeutic specimens. No differences in the chromosome location of AIs were evident. Conversely, from the three BRCA1 cases with no demonstrable AI, there were two prophylactic and two therapeutic surgeries. However, prior chemotherapy may affect the detection of AI: the only three BRCA1 cases with no AIs were those that had received prior treatment. These cases are indicated in Figure 2.
AI on Specific Chromosome Arms
The particular chromosome arms with AI distinguished the three groups. These analyses, presented in Table 3, indicate differences across the three groups at 17q (P = .0393 with Fisher's exact test). Specific comparisons indicated that there were significantly more AIs in the BRCA1 (OR = 12.4; P = .0167) than in the RM group, but the excess compared with the SP group (OR = 4.9) did not reach statistical significance. The intermediate prevalence of AI at 17q in the SP group was not different statistically from the low prevalence in the RM group. Logistic regression analyses and adjustment for block age yielded similar results. The three groups seemed to have different degrees of AI at other arms (particularly 16q and 17p), but these differences did not achieve statistical significance.
Although specific genes were not assessed, the 17q markers were selected to be located from 17q21-17q23 in the vicinity of BRCA1 (www.ncbi.nlm.nih.gov, Homo sapiens genome view, build 34, version 3). The majority of the 17q AIs occurred at marker D17s1290, at 17q23, not at the markers closer to the BRCA1 gene itself (eight of 10 in the BRCA1 group, three of four in the SP group, and one of one in the RM group).
DISCUSSION
The goal of this study was to compare the prevalence of a specific clonal genetic alteration (ie, allele imbalance) in normal breast epithelium from three defined groups with different degrees of breast cancer risk: SP patients and BRCA1 germline mutation carriers (both at high risk) and patients undergoing reduction mammoplasty (at usual risk). Examining approximately 5,400 sites from 460 individual TDLUs, we found a three-fold increase in AI in high-risk compared with usual-risk patients. Three fourths of patients in the SP and BRCA1 groups, but only one fourth of patients in the RM group had AI detectable in a portion of their normal-appearing breast epithelium; approximately 15% of TDLUs from SP or BRCA1 patients, but only 5% from controls had AI. Significantly more alterations at 17q were present in the BRCA1 group.
These data provide an estimate of the prevalence of AI in normal breast epithelium from individuals at usual breast cancer risk, quantify the extent to which AIs are associated with breast cancer and high breast cancer risk, and argue that an increased prevalence of AI is a significant finding. Clonal DNA abnormalities in histologically normal tissue have been reported previously. AI,4-6 aberrant methylation,11 cytogenetic abnormalities,12,13 and telomere shortening14 have all been described. The present study advances those findings by taking a direct, comparative, and quantitative approach to evaluate AI in histologically normal epithelium from groups with different degrees of breast cancer risk and presumably varied mechanisms of breast carcinogenesis.
The present results have several implications. First, although a particular TDLU with AI may not be a cancer precursor, the increased frequency of TDLUs with AI in the high-risk groups may be a manifestation of aberrant processes ongoing in an individual's breast tissue. These processes could contribute to carcinogenesis before tissue is histologically abnormal. They may reflect abnormalities acquired by a mammary stem cell,15-17 affect a field (or patch) of uncertain size, dysregulate mechanisms governing cells' genetic or epigenetic state or stability, and perhaps predispose to cancer across the affected area. We have favored this hypothesis, based in part on our previous data that AIs in normal epithelium and coexisting cancers are usually discordant.8 Some of the present results seem to provide further support for this model. These results include the trends toward more AIs per TDLU in high-risk compared with low-risk tissue, the heterogeneous chromosome sites of AI, the lack of association between AI in the TDLU and the TDLU's proximity to the cancer, the equal proportion of altered TDLUs in prophylactic and therapeutic BRCA1 surgeries, and the apparent absence of clustering of AI in particular TDLUs.
Second, although the sporadic cancer group is similar to the BRCA1 mutation carriers when considering overall amount of AI, the chromosome sites of the abnormalities may be different. The BRCA1 mutation carriers had frequent abnormalities on 17q, where BRCA1 is located. The predominance of 17q AI in the BRCA1 mutation carriers is consistent with the relative paucity of BRCA1 gene abnormalities in sporadic breast cancers18,19 and with the prevailing hypothesis that BRCA1 -associated and sporadic breast cancers develop along different pathways.10,20-25 This particular difference in chromosomal location of AI between groups also supports the view that the clones are not random alterations, but are biologically meaningful. The 17q abnormalities could reflect inactivation of the remaining wild-type allele, a step that is believed necessary for BRCA1-associated carcinogenesis. The resulting BRCA1 null cells should have DNA repair defects and, therefore, might demonstrate more AI. However, we did not see a tendency for multiple AIs in individual TDLUs in the BRCA1 group. An alternative explanation is that germline mutation of BRCA1 may result in breakage at specific sites in the genome, including 17q23, which could lead to deletion of genetic material,26 which is then detected as AI.
Finally, the abnormalities seen in five (28%) of 18 RM cases taken from individuals at no apparent increased risk of cancer are provocative. Although the aberrant clones may reflect random or insignificant genetic variation, an alternative possibility is that controls with abnormalities in their normal epithelium constitute a subgroup of seemingly low-risk women who are actually at increased risk of developing breast cancer. Support for this hypothesis comes from finding a similar proportion of TDLUs with AI in the five RM cases with AI (20%), as in SP (15%) or BRCA1 (16%) cases; the increasing frequency of microsatellite changes in normal epithelium as breast cancer risk increases5; and the association between local recurrence and the presence of AI in breast cancer patients' normal breast epithelium.27 In addition, a recent model28 proposes that a high proportion of sporadic breast cancers arise in a susceptible minority of the population. Identifying that minority is currently impossible, but would be of great importance in allowing targeted prevention strategies to be delivered to individually selected women, likely greatly enhancing their risk/benefit ratio and efficacy. Lastly, recent data suggest that there may be more variation in the normal human genome than appreciated previously.29-31 Future studies could test whether abnormalities in normal epithelium of seemingly usual-risk women are clinically significant.
Several potential objections could be raised to the design and analysis of this study. Because the tissues were anonymous, some desirable information was unavailable (eg, established risk factors, detailed breast cancer characteristics, treatment history, and history of premalignant lesions, such as atypical ductal hyperplasia). The sample size is relatively small and, coupled with the relatively low rate of detection of AI in normal tissue, limited the types of statistical analyses that could be performed. The SP group was not genotyped, which could have introduced misclassification of BRCA1 status in a small proportion of these cases. The DNA was extracted from formalin-fixed tissue, although all samples were processed and analyzed identically, and any biases should affect all groups similarly. Limited DNA quantity from each TDLU precluded a genome-wide approach using additional microsatellites or single-nucleotide polymorphisms. Comparative genome hybridization was considered, but requires a homogeneous population of cells to detect abnormalities. Homogeneity may be typical of cancers, but it is unknown if it characterizes genetically aberrant cells in normal-appearing tissue.
Nevertheless, the present results suggest that an increased frequency of AI in normal-appearing epithelium is associated with increased cancer risk. Instead of representing a standard, background feature of breast epithelium, the presence of AI in normal-appearing breast epithelium from breast cancer patients and BRCA1 mutation-carriers may reflect ongoing, aberrant processes contributing to the development of malignancy, even while the tissue appears normal. The genetic alterations detected in a subset of usual-risk controls warrant investigation as potential markers of increased breast cancer risk.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Supported by National Institutes of Health Public Health Service Grant No. CA81078; United States Department of Defense Breast Cancer Research Program (DAMD 17-97-7191); and Massachusetts Department of Public Health Breast Cancer Research Program.
P.S.L and B.L.S. contributed equally to this work.
Presented in part at the 96th Annual Meeting of the American Association for Cancer Research, Anaheim, CA, April 16-20, 2005.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
National Cancer Institute: Surveillance, Epidemiology, and End Results (SEER). http://seer.cancer.gov/about/quality.html heterozygosity in normal tissue adjacent to breast carcinomas. Science 274: 2057-2059, 1996
Thiagalingam S, Foy RL, Cheng KH, et al: Loss of heterozygosity as a predictor to map tumor suppressor genes in cancer: Molecular basis of its occurrence. Curr Opin Oncol 14: 65-72, 2002
Kerangueven F, Noguchi T, Coulier F, et al: Genome-wide search for loss of heterozygosity shows extensive genetic diversity of human breast carcinomas. Cancer Res 57: 5469-5474, 1997
Deng G, Lu Y, Zlotnikov G, et al: Loss of heterozygosity in normal tissue adjacent to breast carcinomas. Science 274: 2057-2059, 1996
Larson PS, de las Morenas A, Cupples LA, et al: Genetically abnormal clones in histologically normal breast tissue. Am J Pathol 152: 1591-1598, 1998
Lakhani SR, Chaggar R, Davies S, et al: Genetic alterations in ‘normal’ luminal and myoepithelial cells of the breast. J Pathol 189: 496-503, 1999
Cavalli LR, Singh B, Isaacs C, et al: Loss of heterozygosity in normal breast epithelial tissue and benign breast lesions in BRCA1/2 carriers with breast cancer. Cancer Genet Cytogenet 149: 38-43, 2004
Larson PS, de las Morenas A, Bennett SR, et al: Loss of heterozygosity or allele imbalance in histologically normal breast epithelium is distinct from loss of heterozygosity or allele imbalance in co-existing carcinomas. Am J Pathol 161: 283-290, 2002
Berry DA, Iversen ES Jr, Gudbjartsson DF, et al: BRCAPRO validation, sensitivity of genetic testing of BRCA1/BRCA2, and prevalence of other breast cancer susceptibility genes. J Clin Oncol 20: 2701-2712, 2002
Lakhani SR, Van De Vijver MJ, Jacquemier J, et al: The pathology of familial breast cancer: Predictive value of immunohistochemical markers estrogen receptor, progesterone receptor, HER-2, and p53 in patients with mutations in BRCA1 and BRCA2. J Clin Oncol 20: 2310-2318, 2002
Holst CR, Nuovo GJ, Esteller M, et al: Methylation of p16(INK4a) promoters occurs in vivo in histologically normal human mammary epithelia. Cancer Res 63: 1596-1601, 2003
Cianciulli AM, Pescatore B, Bovani R, et al: Aneusomy of chromosomes 1 and 17 in normal tissue adjacent to breast carcinomas. Eur J Histochem 41: 159-160, 1997 (suppl 2)
Steinarsdottir M, Jonasson JG, Vidarsson H, et al: Cytogenetic changes in nonmalignant breast tissue. Genes Chromosomes Cancer 41: 47-55, 2004
Meeker AK, Hicks JL, Gabrielson E, et al: Telomere shortening occurs in subsets of normal breast epithelium as well as in situ and invasive carcinoma. Am J Pathol 164: 925-935, 2004
Tsai Y, Lu Y, Nichols PW, et al: Contiguous patches of normal human mammary epithelium derived from a single stem cell: Implications for breast carcinogenesis. Cancer Res 56: 402-404, 1996
Diallo R, Schaefer KL, Poremba C, et al: Monoclonality in normal epithelium and in hyperplastic and neoplastic lesions of the breast. J Pathol 193: 27-32, 2001
Novelli M, Cossu A, Oukrif D, et al: X-inactivation patch size in human female tissue confounds the assessment of tumor clonality. Proc Natl Acad Sci U S A 100: 3311-3314, 2003
Futreal AP, Liu LQ, Shattuck-Eidens D, et al: BRCA1 mutations in primary breast and ovarian carcinomas. Science 266: 120-122, 1994
Sorlie T, Andersen TI, Bukholm I, et al: Mutation screening of BRCA1 using PTT and LOH analysis at 17q21 in breast carcinomas from familial and non-familial cases. Breast Cancer Res Treat 48: 259-264, 1998
Johannsson OT, Idvall I, Anderson C, et al: Tumour biological features of BRCA1-induced breast and ovarian cancer. Eur J Cancer 33: 362-371, 1997
Crook T, Crossland S, Crompton MR, et al: p53 mutations in BRCA1-associated familial breast cancer. Lancet 350: 638-639, 1997
Crook T, Brooks LA, Crossland S, et al: p53 mutation with frequent novel codons but not a mutator phenotype in BRCA1- and BRCA2-associated breast tumors. Oncogene 17: 1681-1689, 1998
Greenblatt MS, Chappuis PO, Bond JP, et al: TP53 mutations in breast cancer associated with BRCA1 or BRCA2 germ-line mutations: Distinctive spectrum and structural distribution. Cancer Res 61: 4092-4097, 2001
Hedenfalk I, Duggan D, Chen Y, et al: Gene-expression profiles in hereditary breast cancer. N Engl J Med 344: 539-548, 2001
King TA, Gemignani ML, Li W, et al: Increased progesterone receptor expression in benign epithelium of BRCA1-related breast cancers. Cancer Res 64: 5051-5053, 2004
Adem C, Soderberg CL, Hafner K, et al: ERBB2, TBX2, RPS6KB1, and MYC alterations in breast tissues of BRCA1 and BRCA2 mutation carriers. Genes Chromosomes Cancer 41: 1-11, 2004
Li Z, Moore DH, Meng ZH, et al: Increased risk of local recurrence is associated with allelic loss in normal lobules of breast cancer patients. Cancer Res 62: 1000-1003, 2002
Peto J, Mack TM: High constant incidence in twins and other relatives of women with breast cancer. Nat Genet 26: 411-414, 2000
Iafrate AJ, Feuk L, Rivera MN, et al: Detection of large-scale variation in the human genome. Nat Genet 36: 949-951, 2004
Sebat J, Lakshmi B, Troge J, et al: Large-scale copy number polymorphism in the human genome. Science 305: 525-528, 2004
Tuzun E, Sharp AJ, Bailey JA, et al: Fine-scale structural variation of the human genome. Nat Genet 37: 727-732, 2005
Submitted March 25, 2005; accepted September 7, 2005.(Pamela S. Larson, Benjami)
Department of Biostatistics, Boston University School of Public Health
Department of Adult Oncology, Dana-Farber Cancer Institute, Boston, MA
ABSTRACT
PURPOSE: Normal-appearing breast epithelium can contain genetic abnormalities, including allele imbalance (AI), also referred to as loss of heterozygosity. Whether abnormalities are associated with cancer or cancer risk is unknown.
PATIENTS AND METHODS: We performed a miniallelotype, using 20 microsatellites, on each of 460 histologically normal, microdissected breast terminal ducto-lobular units (TDLUs) from three groups of women: sporadic breast cancer patients (SP; n = 18), BRCA1 gene mutation carriers (BRCA1; n = 16), and controls undergoing reduction mammoplasty (RM; n = 18). We analyzed the results using Fisher's exact tests, logistic regression, and generalized estimating equations.
RESULTS: AI was increased three-fold in SP and BRCA1 groups compared with RM. Both the number of TDLUs with AI increased (eight [5%] of 162 in the RM group compared with 24 [15%] of 162 in the SP and 22 [16%] of 136 in the BRCA1 groups; P = .0150), and the proportion of patients with AI increased (five [28%] of 18 in the RM group compared with 15 [83%] of 18 in the SP and 13 [81%] of 16 in the BRCA1 groups; P = .0007). The adjusted odds ratios (OR) for AI in TDLU increased in SP (OR = 15.5) and BRCA1 (OR = 13.7) patients compared with RM (P = .0025). This result was particularly evident on chromosome 17q (P = .0393), where more AI was seen in BRCA1 (OR = 12.4) than in SP (OR = 4.9) patients or RM controls.
CONCLUSION: Increased prevalence of AI in normal-appearing epithelium is associated with breast cancer and increased breast cancer risk. The increased prevalence may reflect dysregulation, even in normal-appearing epithelium, of genomic processes contributing to cancer development. The clinical significance of genetic alterations in the subset of controls remains to be determined.
INTRODUCTION
Although more than 200,000 women in the United States are diagnosed annually with breast cancer,1 knowledge of the earliest stages of breast carcinogenesis remains limited. Breast cancers are morphologically and genetically heterogeneous, and no signature abnormalities characterize all tumors, making identification of cancer precursors difficult. Cell lines and animal models are useful tools but may not mirror early stages of human disease, and investigation of human tissue in vivo is technically challenging. However, a fruitful approach has been examination of allele imbalance (AI), or loss of heterozygosity (LOH), which can be evaluated from formalin-fixed tissue specimens. AI can reflect insufficiency of several different mechanisms of DNA repair and replication (see review in Thiagalingam et al2), although the pertinent mechanism in humans is uncertain. Regardless, AI is characteristic of breast cancers3 and is hypothesized to reflect genetic instability. Recurrent sites of AI are thought to occur at genes important for breast tumorigenesis and may identify pathways that are potentially important for both prevention and treatment.
More recently, AI has been reported in normal-appearing epithelium4-7 from women with and without breast cancer and in tissue both adjacent to and distant from primary cancers. However, the prevalence of AI in normal breast tissue of women at usual cancer risk is unknown, and the central issue of whether increased AI is associated with an increased risk of cancer has not been addressed.
As far as we are aware, direct comparisons of normal tissue from different groups of patients using a consistent panel of markers and adjusting for variables that could affect rates of AI have not been reported. Previous studies have generally examined small numbers of normal epithelial samples and/or patients and used disparate markers, thus limiting comparison between groups and between studies. Identification of recurrent or frequent AI in histologically normal tissue may have important implications for understanding the evolution of breast malignancy, for predicting local disease recurrence, and for assessing risk of new cancers.
We hypothesized that normal-appearing breast epithelium of high-risk individuals might contain increased occult genetic abnormalities compared with control epithelium, indicating a predisposition or vulnerability of the high-risk epithelium to malignant transformation. To test this hypothesis, we used a genetic fingerprinting approach that we developed previously to scan numerous genomic sites for AI. We applied this technique to microdissected, histologically normal breast epithelium from three groups at varying breast cancer risk: women with sporadic breast cancer (SP), women with hereditary mutation of BRCA1, and a control group of women at usual cancer risk undergoing reduction mammoplasty (RM). The mammoplasty specimens would establish the prevalence of AI in normal breast epithelium. In addition, distinct pathways in the earliest stages of carcinogenesis would be suggested by the observation of a distinct pattern of AI in the BRCA1 group.
PATIENTS AND METHODS
Specimen Acquisition
Institutional review board approval was obtained at Boston University Medical Center and the Dana-Farber Cancer Institute. Formalin-fixed paraffin-embedded blocks of breast tissue not needed for diagnosis from 18 bilateral RM surgeries and 18 SP surgeries from patients 50 years of age were selected at random from the department of pathology archives. (Data from six SP patients had been included in a previous report comparing AI in normal epithelium and coexisting cancers8). RM cases are screened to eliminate those with a personal or strong family history of breast cancer. The available history and demographics suggest that although no testing was performed, few of the SP cases would have heritable BRCA1 mutations. In nine of 18 RM cases, terminal ducto-lobular units (TDLUs) were taken from both the left and right reduction specimens. In one of 18 SP cases, bilateral surgeries were performed, and TDLUs were taken from both sides. In the remaining 17 cases, normal epithelium was obtained from blocks at sufficient distance from the tumor that they lacked any histologically malignant cells, but their precise distance from the tumor was indeterminate. Formalin-fixed paraffin-embedded blocks of breast tissue not needed for diagnosis were selected from the breast surgeries of 16 patients 50 years of age carrying deleterious, heritable germline BRCA1 mutation. Five of these 16 patients had bilateral surgeries. Of these 21 breast operations, 10 were performed for cancer therapy, and 11 were from biopsies of benign lesions or prophylactic mastectomies in which cancers were not identified. As with the SP cases, epithelium was taken from blocks removed from the tumor. Criteria for inclusion of individual cases were five or more normal TDLUs (microdissected from any number of blocks) analyzed with five or more informative markers and surgery performed after 1990.
Microdissection and Polymerase Chain Reaction
A single pathologist (A.M.) identified histologically normal TDLUs. Each sample was isolated using laser microdissection and DNA was extracted, as previously reported.8 DNA from each individual TDLU was analyzed with a mini-allelotype consisting of 20 microsatellite markers located on nine chromosome arms: 1p: D1s468; 1q: D1s549, D1s213; 3p: D3s1283; 7q: D7s486; 11p: TH01, D11s2071; 11q: PYGM, D11s1818, D11s1819; 16q: D16s265, D16s402, D16s413, D16s512; 17p: TP53, D17s796, D17s525; 17q: D17s1290, D17s579, and D17s855. Primers were purchased from ResGen Invitrogen Corp (Huntsville, AL). Six multiplexed polymerase chain reactions were performed, electrophoresed, and exposed to film, as previously reported.8
Determination of AI
All autoradiographs were scanned by laser densitometer (Molecular Dynamics, Sunnyvale, CA) to calculate allele ratios. The normal allele ratio at each marker in each patient was determined by averaging all of the marker's allele ratios from that patient's normal tissue samples (ie, multiple separate TDLUs). AI was defined at heterozygous loci as previously described8: in other words, an imbalance of allele intensities greater than 25% (when (n1)(t2)/(n2)(t1) > than 1.33 or < 0.75, where n1 = normal samples' larger allele, n2 = normal samples' smaller allele, t1 = test sample's larger allele, t2 = test sample's smaller allele). This degree of allele imbalance indicates that a substantial proportion of the cells within this sample contains the same DNA abnormality and likely represents the presence of a clonal population. We used a less stringent cutoff than what is commonly applied to tumor DNA because the proportion of a TDLU composed of any genetically aberrant clone is unknown and may well be lower than the proportion in histologically malignant tissue. For a marker to be identified as showing AI, equivalent results had to be demonstrated in at least two independent replicate reactions, and no polymerase chain reactions could produce a normal ratio or contradictory imbalances. Using these criteria, we have estimated previously that our rate of detecting false-positive imbalances is between 2% and 4%.8 At some loci, allele imbalance may reflect increased copy number; at others, it may represent allele loss. Distinguishing between these possibilities is important conceptually, but would not alter data analysis.
Statistical Analyses
We first evaluated whether each patient had any AI. To compare the proportion of patients with any AI across the three groups, we used Fisher's exact test. To evaluate potential confounding by patient age and block age, we performed analysis of variance to examine mean patient and block ages among the three groups. To adjust for potential differences owing to these confounders (patient and block age), we performed logistic regression, weighting for the number of observations for each patient. We next evaluated whether the degree of AI differed among the three groups at each chromosome arm by summarizing information across all TDLUs for each chromosome arm and evaluating whether an arm showed any evidence of AI on any TDLU (yes or no). For each arm, Fisher's exact test was used to compare the relative prevalence of AI in each group. We also used weighted logistic regression for those arms where there was at least one patient with AI in each group (chromosomes 1q, 11p, 11q, and 17q), again weighting by the number of observations for each patient. Finally, we examined the average number and proportion of AI per duct and per informative marker. For comparisons across the groups in these analyses, we applied generalized estimating equations to account for the potential correlation among observations coming from the same person and weighted by the number of observations per duct, when appropriate.
RESULTS
Group Characteristics
The three groups (RM, SP, and BRCA1) were well-matched for patient age, number of independent cases, samples per case, samples per group, and the proportion of informative markers per group (Table 1). Differences among groups were noted for age of tissue blocks, because RM cases tended to have more recent blocks. Because increased age of blocks might increase detection of artifactual AI, we assessed whether block age was associated with presence of AI and found that it was not (P = .4, a t test), even though block age did vary across groups. Although patient age did not vary across groups, we tested whether it was associated with AI and found that it was not. Although the SP group did not have genetic testing performed, their estimated prevalence of a BRCA1 mutation based on age at diagnosis was approximately 7%,9 implying that one or two SP patients may be misclassified. Tumors in the SP group comprised a distribution of histologic and immunohistochemical features typical of an unselected series10: one lobular and 17 ductal carcinomas; two cases were grade 1/3, the others equally divided between grade 2/3 and grade 3/3; 70% of cases were positive for estrogen receptors and progesterone receptors. To the extent possible, we determined prior exposure to cancer treatment. Three of 10 BRCA1 mutation carriers with prior cancer had had chemotherapy, none had had radiation, and none of the 18 SP patients had had either chemotherapy or radiation.
Comparison of AI Across Groups
AI was seen in individual TDLUs of all groups (a representative example is shown in Fig 1). The mean allele ratio of these imbalances was approximately 0.6 in all groups, as shown in Table 2. (For ratios > 1, the reciprocal value was used for this calculation). All AIs occurred at single sites; in other words, other informative sites on the chromosome arm remained heterozygous, with one exception. (This exception was an RM case with eight AIs, five occurring in a single TDLU containing AIs of all informative markers on chromosome arms 1q and 16q).
As shown in Table 2 and Figure 2, we found differences in measurements of AI across groups. In general, a small number of abnormalities were detected in the RM group, and a larger number in the SP and BRCA1 groups. Both the number of TDLUs with AI and the proportion of patients with AI were increased nearly three-fold in the SP and BRCA1 groups compared with controls. Specifically, 24 (15%) of 162 TDLUs from the SP group, 22 (16%) of 136 TDLUs from the BRCA1 group, and eight (5%) of 162 TDLUs from the control group contained AI (P = .0150). (Eleven [18%] of 62 TDLUs from prophylactic or benign BRCA1 surgeries and 11 [15%] of 74 TDLUs from therapeutic BRCA1 surgeries had AI). Similarly, 15 (83%) of 18 SP patients, 13 (81%) of 16 BRCA1 patients, and five (28%) of 18 RM patients demonstrated AI (P = .0007). Among the five RM cases with AI, eight (20%) of 41 TDLUs had allele imbalances, a proportion that resembles what was seen in the SP and BRCA1 groups (16% and 15%, respectively).
After adjusting for block age and weighting for number of TDLUs per patient, a dramatically increased odds ratio (OR) for AI was present in both the SP (OR = 15.5; 95% CI, 2.6 to 92.4) and BRCA1 (OR = 13.7; 95% CI, 2.3 to 79.8) groups (P = .0025) compared with the RM group. The SP and BRCA1 groups were similar to each other in overall AI (P = .892, weighted logistic regression). We also noted that the mean number of AIs per TDLU and the mean number of AIs per informative marker seemed to be increased in the SP and BRCA1 groups compared with controls, although these measures did not achieve statistical significance.
To assess whether AI reflects a widespread characteristic of a patient's breast tissue or a focal abnormality of a single TDLU, we considered the distribution of AI within TDLUs and within patients. First, we noted that AI did not seem to cluster in particular TDLUs. Seventeen of the 52 patients had two or more AIs (three RM patients, six SP patients, and eight BRCA1 patients); in five of 17 patients, multiple AIs occurred in a single TDLU, and in 15 of 17 patients, they occurred in multiple TDLUs (some cases had both patterns). Next, we found that identical AIs were not commonly seen in multiple TDLUs. In only three of the 17 cases with multiple AI (one from each group) was the same AI present in two TDLUs. Third, we noted that in the RM cases, detection of AI did not seem to be influenced by whether samples from one or both sides were examined. AI was seen in three of nine cases with unilateral and two of nine cases with bilateral tissue samples available. Fourth, we found that proximity of a TDLU and cancer did not seem to influence the prevalence of TDLU with AI. In the 18 SP cases, there were 162 TDLUs with a total of 25 AIs. Fifty-six (35%) of 162 TDLUs were on the same block as the cancer, and 106 (65%) of 162 were on a different block. Six (24%) of 25 AIs were in TDLUs on the same block as the cancer, and 19 (76%) of 25 AIs were on a different block. Finally, we found that in the BRCA1 cases, detection of AI seemed unrelated to whether the tissue was from a prophylactic or therapeutic surgery. AI was seen in seven of 11 prophylactic and seven of 10 therapeutic surgeries. Fourteen AIs were detected in the prophylactic specimens, and 12 AIs were detected in the therapeutic specimens. No differences in the chromosome location of AIs were evident. Conversely, from the three BRCA1 cases with no demonstrable AI, there were two prophylactic and two therapeutic surgeries. However, prior chemotherapy may affect the detection of AI: the only three BRCA1 cases with no AIs were those that had received prior treatment. These cases are indicated in Figure 2.
AI on Specific Chromosome Arms
The particular chromosome arms with AI distinguished the three groups. These analyses, presented in Table 3, indicate differences across the three groups at 17q (P = .0393 with Fisher's exact test). Specific comparisons indicated that there were significantly more AIs in the BRCA1 (OR = 12.4; P = .0167) than in the RM group, but the excess compared with the SP group (OR = 4.9) did not reach statistical significance. The intermediate prevalence of AI at 17q in the SP group was not different statistically from the low prevalence in the RM group. Logistic regression analyses and adjustment for block age yielded similar results. The three groups seemed to have different degrees of AI at other arms (particularly 16q and 17p), but these differences did not achieve statistical significance.
Although specific genes were not assessed, the 17q markers were selected to be located from 17q21-17q23 in the vicinity of BRCA1 (www.ncbi.nlm.nih.gov, Homo sapiens genome view, build 34, version 3). The majority of the 17q AIs occurred at marker D17s1290, at 17q23, not at the markers closer to the BRCA1 gene itself (eight of 10 in the BRCA1 group, three of four in the SP group, and one of one in the RM group).
DISCUSSION
The goal of this study was to compare the prevalence of a specific clonal genetic alteration (ie, allele imbalance) in normal breast epithelium from three defined groups with different degrees of breast cancer risk: SP patients and BRCA1 germline mutation carriers (both at high risk) and patients undergoing reduction mammoplasty (at usual risk). Examining approximately 5,400 sites from 460 individual TDLUs, we found a three-fold increase in AI in high-risk compared with usual-risk patients. Three fourths of patients in the SP and BRCA1 groups, but only one fourth of patients in the RM group had AI detectable in a portion of their normal-appearing breast epithelium; approximately 15% of TDLUs from SP or BRCA1 patients, but only 5% from controls had AI. Significantly more alterations at 17q were present in the BRCA1 group.
These data provide an estimate of the prevalence of AI in normal breast epithelium from individuals at usual breast cancer risk, quantify the extent to which AIs are associated with breast cancer and high breast cancer risk, and argue that an increased prevalence of AI is a significant finding. Clonal DNA abnormalities in histologically normal tissue have been reported previously. AI,4-6 aberrant methylation,11 cytogenetic abnormalities,12,13 and telomere shortening14 have all been described. The present study advances those findings by taking a direct, comparative, and quantitative approach to evaluate AI in histologically normal epithelium from groups with different degrees of breast cancer risk and presumably varied mechanisms of breast carcinogenesis.
The present results have several implications. First, although a particular TDLU with AI may not be a cancer precursor, the increased frequency of TDLUs with AI in the high-risk groups may be a manifestation of aberrant processes ongoing in an individual's breast tissue. These processes could contribute to carcinogenesis before tissue is histologically abnormal. They may reflect abnormalities acquired by a mammary stem cell,15-17 affect a field (or patch) of uncertain size, dysregulate mechanisms governing cells' genetic or epigenetic state or stability, and perhaps predispose to cancer across the affected area. We have favored this hypothesis, based in part on our previous data that AIs in normal epithelium and coexisting cancers are usually discordant.8 Some of the present results seem to provide further support for this model. These results include the trends toward more AIs per TDLU in high-risk compared with low-risk tissue, the heterogeneous chromosome sites of AI, the lack of association between AI in the TDLU and the TDLU's proximity to the cancer, the equal proportion of altered TDLUs in prophylactic and therapeutic BRCA1 surgeries, and the apparent absence of clustering of AI in particular TDLUs.
Second, although the sporadic cancer group is similar to the BRCA1 mutation carriers when considering overall amount of AI, the chromosome sites of the abnormalities may be different. The BRCA1 mutation carriers had frequent abnormalities on 17q, where BRCA1 is located. The predominance of 17q AI in the BRCA1 mutation carriers is consistent with the relative paucity of BRCA1 gene abnormalities in sporadic breast cancers18,19 and with the prevailing hypothesis that BRCA1 -associated and sporadic breast cancers develop along different pathways.10,20-25 This particular difference in chromosomal location of AI between groups also supports the view that the clones are not random alterations, but are biologically meaningful. The 17q abnormalities could reflect inactivation of the remaining wild-type allele, a step that is believed necessary for BRCA1-associated carcinogenesis. The resulting BRCA1 null cells should have DNA repair defects and, therefore, might demonstrate more AI. However, we did not see a tendency for multiple AIs in individual TDLUs in the BRCA1 group. An alternative explanation is that germline mutation of BRCA1 may result in breakage at specific sites in the genome, including 17q23, which could lead to deletion of genetic material,26 which is then detected as AI.
Finally, the abnormalities seen in five (28%) of 18 RM cases taken from individuals at no apparent increased risk of cancer are provocative. Although the aberrant clones may reflect random or insignificant genetic variation, an alternative possibility is that controls with abnormalities in their normal epithelium constitute a subgroup of seemingly low-risk women who are actually at increased risk of developing breast cancer. Support for this hypothesis comes from finding a similar proportion of TDLUs with AI in the five RM cases with AI (20%), as in SP (15%) or BRCA1 (16%) cases; the increasing frequency of microsatellite changes in normal epithelium as breast cancer risk increases5; and the association between local recurrence and the presence of AI in breast cancer patients' normal breast epithelium.27 In addition, a recent model28 proposes that a high proportion of sporadic breast cancers arise in a susceptible minority of the population. Identifying that minority is currently impossible, but would be of great importance in allowing targeted prevention strategies to be delivered to individually selected women, likely greatly enhancing their risk/benefit ratio and efficacy. Lastly, recent data suggest that there may be more variation in the normal human genome than appreciated previously.29-31 Future studies could test whether abnormalities in normal epithelium of seemingly usual-risk women are clinically significant.
Several potential objections could be raised to the design and analysis of this study. Because the tissues were anonymous, some desirable information was unavailable (eg, established risk factors, detailed breast cancer characteristics, treatment history, and history of premalignant lesions, such as atypical ductal hyperplasia). The sample size is relatively small and, coupled with the relatively low rate of detection of AI in normal tissue, limited the types of statistical analyses that could be performed. The SP group was not genotyped, which could have introduced misclassification of BRCA1 status in a small proportion of these cases. The DNA was extracted from formalin-fixed tissue, although all samples were processed and analyzed identically, and any biases should affect all groups similarly. Limited DNA quantity from each TDLU precluded a genome-wide approach using additional microsatellites or single-nucleotide polymorphisms. Comparative genome hybridization was considered, but requires a homogeneous population of cells to detect abnormalities. Homogeneity may be typical of cancers, but it is unknown if it characterizes genetically aberrant cells in normal-appearing tissue.
Nevertheless, the present results suggest that an increased frequency of AI in normal-appearing epithelium is associated with increased cancer risk. Instead of representing a standard, background feature of breast epithelium, the presence of AI in normal-appearing breast epithelium from breast cancer patients and BRCA1 mutation-carriers may reflect ongoing, aberrant processes contributing to the development of malignancy, even while the tissue appears normal. The genetic alterations detected in a subset of usual-risk controls warrant investigation as potential markers of increased breast cancer risk.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
NOTES
Supported by National Institutes of Health Public Health Service Grant No. CA81078; United States Department of Defense Breast Cancer Research Program (DAMD 17-97-7191); and Massachusetts Department of Public Health Breast Cancer Research Program.
P.S.L and B.L.S. contributed equally to this work.
Presented in part at the 96th Annual Meeting of the American Association for Cancer Research, Anaheim, CA, April 16-20, 2005.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
National Cancer Institute: Surveillance, Epidemiology, and End Results (SEER). http://seer.cancer.gov/about/quality.html heterozygosity in normal tissue adjacent to breast carcinomas. Science 274: 2057-2059, 1996
Thiagalingam S, Foy RL, Cheng KH, et al: Loss of heterozygosity as a predictor to map tumor suppressor genes in cancer: Molecular basis of its occurrence. Curr Opin Oncol 14: 65-72, 2002
Kerangueven F, Noguchi T, Coulier F, et al: Genome-wide search for loss of heterozygosity shows extensive genetic diversity of human breast carcinomas. Cancer Res 57: 5469-5474, 1997
Deng G, Lu Y, Zlotnikov G, et al: Loss of heterozygosity in normal tissue adjacent to breast carcinomas. Science 274: 2057-2059, 1996
Larson PS, de las Morenas A, Cupples LA, et al: Genetically abnormal clones in histologically normal breast tissue. Am J Pathol 152: 1591-1598, 1998
Lakhani SR, Chaggar R, Davies S, et al: Genetic alterations in ‘normal’ luminal and myoepithelial cells of the breast. J Pathol 189: 496-503, 1999
Cavalli LR, Singh B, Isaacs C, et al: Loss of heterozygosity in normal breast epithelial tissue and benign breast lesions in BRCA1/2 carriers with breast cancer. Cancer Genet Cytogenet 149: 38-43, 2004
Larson PS, de las Morenas A, Bennett SR, et al: Loss of heterozygosity or allele imbalance in histologically normal breast epithelium is distinct from loss of heterozygosity or allele imbalance in co-existing carcinomas. Am J Pathol 161: 283-290, 2002
Berry DA, Iversen ES Jr, Gudbjartsson DF, et al: BRCAPRO validation, sensitivity of genetic testing of BRCA1/BRCA2, and prevalence of other breast cancer susceptibility genes. J Clin Oncol 20: 2701-2712, 2002
Lakhani SR, Van De Vijver MJ, Jacquemier J, et al: The pathology of familial breast cancer: Predictive value of immunohistochemical markers estrogen receptor, progesterone receptor, HER-2, and p53 in patients with mutations in BRCA1 and BRCA2. J Clin Oncol 20: 2310-2318, 2002
Holst CR, Nuovo GJ, Esteller M, et al: Methylation of p16(INK4a) promoters occurs in vivo in histologically normal human mammary epithelia. Cancer Res 63: 1596-1601, 2003
Cianciulli AM, Pescatore B, Bovani R, et al: Aneusomy of chromosomes 1 and 17 in normal tissue adjacent to breast carcinomas. Eur J Histochem 41: 159-160, 1997 (suppl 2)
Steinarsdottir M, Jonasson JG, Vidarsson H, et al: Cytogenetic changes in nonmalignant breast tissue. Genes Chromosomes Cancer 41: 47-55, 2004
Meeker AK, Hicks JL, Gabrielson E, et al: Telomere shortening occurs in subsets of normal breast epithelium as well as in situ and invasive carcinoma. Am J Pathol 164: 925-935, 2004
Tsai Y, Lu Y, Nichols PW, et al: Contiguous patches of normal human mammary epithelium derived from a single stem cell: Implications for breast carcinogenesis. Cancer Res 56: 402-404, 1996
Diallo R, Schaefer KL, Poremba C, et al: Monoclonality in normal epithelium and in hyperplastic and neoplastic lesions of the breast. J Pathol 193: 27-32, 2001
Novelli M, Cossu A, Oukrif D, et al: X-inactivation patch size in human female tissue confounds the assessment of tumor clonality. Proc Natl Acad Sci U S A 100: 3311-3314, 2003
Futreal AP, Liu LQ, Shattuck-Eidens D, et al: BRCA1 mutations in primary breast and ovarian carcinomas. Science 266: 120-122, 1994
Sorlie T, Andersen TI, Bukholm I, et al: Mutation screening of BRCA1 using PTT and LOH analysis at 17q21 in breast carcinomas from familial and non-familial cases. Breast Cancer Res Treat 48: 259-264, 1998
Johannsson OT, Idvall I, Anderson C, et al: Tumour biological features of BRCA1-induced breast and ovarian cancer. Eur J Cancer 33: 362-371, 1997
Crook T, Crossland S, Crompton MR, et al: p53 mutations in BRCA1-associated familial breast cancer. Lancet 350: 638-639, 1997
Crook T, Brooks LA, Crossland S, et al: p53 mutation with frequent novel codons but not a mutator phenotype in BRCA1- and BRCA2-associated breast tumors. Oncogene 17: 1681-1689, 1998
Greenblatt MS, Chappuis PO, Bond JP, et al: TP53 mutations in breast cancer associated with BRCA1 or BRCA2 germ-line mutations: Distinctive spectrum and structural distribution. Cancer Res 61: 4092-4097, 2001
Hedenfalk I, Duggan D, Chen Y, et al: Gene-expression profiles in hereditary breast cancer. N Engl J Med 344: 539-548, 2001
King TA, Gemignani ML, Li W, et al: Increased progesterone receptor expression in benign epithelium of BRCA1-related breast cancers. Cancer Res 64: 5051-5053, 2004
Adem C, Soderberg CL, Hafner K, et al: ERBB2, TBX2, RPS6KB1, and MYC alterations in breast tissues of BRCA1 and BRCA2 mutation carriers. Genes Chromosomes Cancer 41: 1-11, 2004
Li Z, Moore DH, Meng ZH, et al: Increased risk of local recurrence is associated with allelic loss in normal lobules of breast cancer patients. Cancer Res 62: 1000-1003, 2002
Peto J, Mack TM: High constant incidence in twins and other relatives of women with breast cancer. Nat Genet 26: 411-414, 2000
Iafrate AJ, Feuk L, Rivera MN, et al: Detection of large-scale variation in the human genome. Nat Genet 36: 949-951, 2004
Sebat J, Lakshmi B, Troge J, et al: Large-scale copy number polymorphism in the human genome. Science 305: 525-528, 2004
Tuzun E, Sharp AJ, Bailey JA, et al: Fine-scale structural variation of the human genome. Nat Genet 37: 727-732, 2005
Submitted March 25, 2005; accepted September 7, 2005.(Pamela S. Larson, Benjami)