|Home | About | Journals | Submit | Contact Us | Français|
To assess telomere DNA content (TC) and the number of sites of allelic imbalance (AI) as a function of breast cancer progression.
TC and AI were determined in 54 histologically normal tissues, 10 atypical ductal hyperplasias (ADH), 122 in situ ductal carcinomas (DCIS) and 535 invasive carcinomas (Stage I–IIIA).
TC was altered in ADH lesions (20%), DCIS specimens (53%) and invasive carcinomas (51%). The mean number of sites of AI was 0.26 in histologically normal group tissue, increased to 1.00 in ADH, 2.94 in DCIS, and 3.07 in invasive carcinomas. All groups were statistically different from the histologically normal group (P < 0.001 for each); however, there was no difference between DCIS and the invasive groups.
Genomic instability increases in ADH and plateaus in DCIS without further increase in the invasive carcinomas, supporting the notion that invasive carcinomas evolve from or in parallel with DCIS.
It is widely accepted that genomic instability is a pre-requisite for the initiation and progression of virtually all cancers . Accordingly, the progression of breast cancer can be characterized by the accumulation of genetic mutations in critical genes accompanied by histological progression from normal epithelium to atypical ductal hyperplasia (ADH), to ductal carcinoma in situ (DCIS) to the development of an invasive breast carcinoma [2, 3].
A significant cause of genomic instability is telomere dysfunction [4–7]. Telomeres are nucleoprotein complexes that are comprised of 1,000–2,000 tandemly repeated copies of the hexanucleotide DNA sequence (TTAGGG) . These repeat regions are associated with numerous telomere binding proteins, such as Telomeric Repeat-binding Factor 1 (TRF1), Telomeric Repeat-binding Factor 2 (TRF2) and Protection of Telomeres 1 (POT1), which play important roles in telomere maintenance [9, 10]. Telomeres are located at and stabilize the ends of eukaryotic chromosomes, thus preventing degradation and recombination [11–13]. However, telomeres can be critically shortened, and thereby become dysfunctional, by several mechanisms, including incomplete replication of the lagging strand during DNA synthesis , loss or alterations of the telomere-binding proteins involved in telomere maintenance , and DNA damage induced by oxidative stress . Telomere loss may be compensated by the reactivation of the enzyme telomerase, as seen in 85–90% of human cancers .
Abnormalities in telomere length are early and frequent events in the malignant transformation of numerous types of carcinomas [18, 19]. In breast, telomere shortening has been observed in invasive carcinomas, in situ lesions, and histologically normal tissue proximal to breast tumors [20, 21]. Additionally, our laboratory has recently demonstrated that telomere DNA content (TC), a proxy for telomere length, in breast tumor tissues is a prognostic marker for clinical outcome [22, 23].
Genomic instability can also be manifested by the presence of allelic imbalance (AI), which is a deviation from the normal 1:1 ratio of maternal and paternal alleles. Numerous studies have shown that the presence of AI is characteristic of invasive breast carcinomas [24, 25] and is also present at the in situ stage of the disease [26, 27]. Additional studies have demonstrated that AI occurs within atypical breast hyperplasias [28, 29], histologically normal tissue proximal to breast tumors [21, 30–32], and, in some instances, breast tissue from women with benign breast disease . AI has also been found in the stromal compartment of cancer-associated breast tissues .
Numerous groups have investigated AI in the development of breast cancer. Notably, Ellsworth et al.  developed a panel of microsatellite markers specific for loci commonly lost in breast cancer. This group examined the evolution of genomic instability by characterizing AI in tissue samples representing a continuum of breast cancer development and concluded that DCIS lesions contain AI levels characteristic of advanced invasive tumors .
To evaluate the link between telomere dysfunction and the generation of allelic imbalance in the progression of breast cancer, we assessed alterations in TC and the extent of AI in a continuum of breast tissues ranging from histologically normal tissue derived from reduction mammoplasty, to ADH, DCIS and invasive carcinomas ranging from Stage I to IIIA. Here, we demonstrate that genomic instability (i.e. changes in TC or AI that exceed values typically observed in normal tissues) increases along the continuum of breast disease; however, it plateaus in DCIS without further increase in the invasive carcinomas.
A total of 721 human breast tissues were used in this study. Fifty-four normal, disease-free breast tissue samples from women undergoing reduction mammoplasty (mean age = 35.6 years; range: 17–68) were obtained from the National Cancer Institute Cooperative Human Tissue Network (Nashville, TN). Ten atypical ductal hyperplasia lesions (mean age = 56.3 years; range: 41–70) were obtained from the Department of Pathology at University of New Mexico Hospital (UNMH). Two independent cohorts of breast tumors were analyzed. The first cohort (test set) was obtained through the New Mexico Tumor Registry (NMTR) and Department of Pathology at UNMH and consisted of 163 specimens including DCIS (N = 27), and Stage I (N = 104) and IIA (N = 32) invasive breast carcinomas (mean age = 47.5 years; range: 25–77). The second cohort (validation set) was obtained through the Health, Eating, Activity and Lifestyle (HEAL) Study, an ongoing population-based, multi-center prospective cohort study , and consisted of 494 cases including DCIS (N = 95), and Stage I (N = 244), IIA (N = 112), Stage IIB (N = 39) and IIIA (N = 4) invasive breast carcinomas (mean age = 59.3 years; range: 29–89). Clinical data for the two tumor cohorts are shown in Table 1. Experiments were performed in accordance with all federal guidelines as approved by the University of New Mexico Health Science Center Human Research Review Committee.
All tissue sections were examined microscopically to confirm diagnosis. Tissue sections were not microdissected, but typically contained from 75 to 100% tumor cells. A single pathologist reviewed the histological slides for the 10 ADH lesions and cohort two (validation set); whereas, the reduction mammoplasty specimens and cohort one (test set) were reviewed by numerous pathologists. The criteria used for the ADH specimens were based on morphological characteristics of a proliferative lesion that fulfills some but not all the criteria for DCIS.
DNA was isolated from fresh, frozen or formalin-fixed, paraffin-embedded (FFPE) tissue samples using the DNeasy® silica-based spin column extraction kit (Qiagen; Valencia, CA) and the manufacturer’s suggested animal tissue protocol. FFPE samples were treated with xylene and washed with ethanol prior to DNA extraction. DNA concentrations were measured using the Picogreen® dsDNA quantitation assay (Molecular Probes, Eugene, OR) using a λ phage DNA as the standard as directed by the manufacturer’s protocol.
TC was measured in known DNA masses, typically 5–10 ng, by slot blot titration assay, as previously described [21–23]. TC is expressed as a percentage of the TC in a placental DNA standard measured in parallel, which is defined as 100%. Each measurement was repeated independently three times and the coefficient of variation for each sample was ≤10%. The content of telomere DNA sequences can easily be measured in genomic DNA obtained from fresh, frozen and paraffin-embedded tissues [22, 38]. We have previously shown that TC is (i) directly proportional to telomere length determined by Southern blotting, (ii) not affected by TTAGGG sequences outside the telomere, and (iii) not affected by DNA fragmentation less than 1 KB in length [22, 38].
The extent of AI was determined using a straight-forward, economical, and high-throughput method recently developed by our laboratory . This method evaluates AI in a panel of 16 randomly selected microsatellite markers (i.e. markers with no known relationship to breast cancer) thereby preventing measurement bias by selection of genes whose products are involved in tumorigenesis . Briefly, DNA (~1 ng) was amplified using the AmpFlSTR Identifiler PCR Amplification Kit (Applied Biosystems, Foster City, CA) using the manufacturer’s protocol. Each multiplex PCR reaction amplifies 16 short tandem repeat (STR) microsatellite loci from independent locations in the genome (Amelogenin, CSF1PO, D2S1338, D3S1358, D5S818, D7S820, D8S1179, D13S317, D16S539, D18S51, D19S433, D21S11, FGA, TH01, TPOX and vWA). Each of the PCR primers is labeled with one of four fluorescent dyes (6-FAM, PET, VIC and NED), each with a unique emission profile, allowing the simultaneous resolution of 16 amplicons of similar size. PCR products were resolved by capillary gel electrophoresis and detected using an ABI Prism 377 DNA Sequencer (Perkin Elmer, Foster City, CA). The height of each fluorescence peak in the electropherograms was quantitated using the ABI Prism GeneScan and Genotype Analysis software (Applied Biosystems, Foster City, CA) and a ratio of the peak heights of each pair of heterozygous allelic amplicons was calculated. By convention, the allele with the greater fluorescence intensity was designated the numerator. Thus, the ratio was always ≥1.0, with 1.0 representing the theoretical ratio for normal alleles. We previously defined an operational threshold of AI (i.e. ≥2 sites of AI) that could differentiate between a variety of normal and cancerous tissues independent of storage conditions (i.e. fresh, frozen or paraffin-embedded, formalin-fixed) . Of the 118 normal specimens, only 1 (0.8%) specimen demonstrated ≥2 sites of AI. In contrast, of the 239 tumor specimens, 161 (67.4%) demonstrated ≥2 sites of AI.
The mean number of sites of AI and TC distributions for histologically normal, ADH, DCIS and invasive carcinoma specimens were analyzed by non-parametric Rank Sums tests. Chi-square tests were used to determine differences for individual allelic frequencies between the DCIS and invasive groups. JMP® statistical package (SAS Institute, Cary, NC) was used for all analyses and P-values <0.05 were considered to be significant.
TC was determined in 54 histologically normal breast tissues obtained from women who underwent reduction mammoplasty. TC was tightly regulated within these histologically normal breast tissues; 95% of these normal specimens fell within the range of 75–154% (Fig. 1), nearly identical to the 75–143% range previously reported in a diverse set of 70 specimens of normal tissue from multiple organ sites, including breast . Next, TC was determined in a set of 10 ADH lesions. TC values in two specimens (20%) fell outside the 95% range found in the histologically normal specimens.
TC next was determined in a cohort of 27 DCIS, 104 Stage I and0 32 Stage IIA breast tumors. In contrast to the histologically normal group, there was a wide range of TC distribution in the tumor specimens within the test cohort (Fig. 1). Of the 27 DCIS cases, 10 (37%) fell outside the normal range. Similarly, 44 of the 104 Stage I tumors (42%) and 14 of the 32 Stage IIA tumors (44%) fell outside the normal range. However, the DCIS specimens as a group had longer telomeres than the Stage I (P = 0.0152) and Stage IIA (P = 0.0338) tumors.
The results were validated in an independent population-based breast tumor cohort comprised of 494 specimens. TC was determined in 95 DCIS, 244 Stage I, 112 Stage IIA, 39 Stage IIB and 4 Stage IIIA breast tumors. Fifty-five of the 95 DCIS cases (58%), 127 of the 244 Stage I (52%), 65 of the 112 Stage IIA (58%), 20 of the 39 Stage IIB (51%) and 3 of the 4 Stage IIIA (75%) tumors fell outside of the normal range defined by the histologically normal breast tissues. Again, the DCIS group had longer telomeres than the Stage I (P < 0.0001), Stage IIA (P = 0.0005) and Stage IIB (P = 0.0048) tumors (Fig. 1). In both the test and validation cohorts, TC did not correlate with ethnicity, nodal status, or ER and PR status.
To extend and confirm these findings, AI, another independent marker of genomic instability, was measured and compared in the same tissue cohorts. The mean number of sites of AI was 0.26 in the histologically normal and 1.00 in the ADH groups (Fig. 2). As compared to the histologically normal group, the ADH group showed a significant increase in the extent of AI (P = 0.0002), although the small number of ADH specimens must be noted.
Next, the extent of AI was analyzed in the test cohort. The mean number of sites of AI was 2.63 in DCIS, 3.24 in Stage I tumors and 2.84 in Stage IIA tumors (Fig. 2). All groups were statistically different when compared to the histologically normal group (P < 0.0001 for each). As observed for TC, there was no difference in the extent of AI in the DCIS group compared to any of the invasive groups. Additionally, there was no difference between Stage I and Stage IIA tumors.
These findings were replicated in the validation cohort. The mean number of sites of AI was 3.03 in DCIS, 3.08 in Stage I, 2.98 in Stage IIA, 2.92 in Stage IIB and 3.50 in Stage IIIA (Fig. 2). All categories were statistically different from the histologically normal group (P < 0.001 for each). There was no statistically significant difference between the DCIS group and the groups of invasive carcinoma or between any of the invasive groups. Additionally, there was no statistical difference in the mean number of sites of AI between paired groups by stage between the test and validation cohorts of breast tumors. Next, we tested our previously operationally-defined threshold for AI (i.e. ≥2 sites of AI) in these tissue cohorts . Using this threshold, 0 of the 54 (0%) histologically normal breast specimens contained ≥2 sites of AI (Table 2). In contrast, 131 of the 163 tumors in the test cohort (80.4%) and 402 of the 494 tumors in validation cohort (81.4%) contained ≥2 sites of AI (Table 2). AI did not correlate with ethnicity, nodal status, or ER and PR status in both the test and validation cohorts.
Since the mean number of sites of AI in specimens of DCIS was nearly identical to the invasive tumors in both study cohorts, we next determined whether there was a difference in the allelic frequencies at each locus as a function of stage of progression. Since the individual loci have no known involvement in the development of breast cancer, there should be no selection pressure and the frequency of AI at a particular locus should not differ as a function of progression. For this analysis, the DCIS and invasive tumors were combined from the two tumor cohorts. As shown in Table 3, there were no statistically significant differences in the allelic frequencies of 14 of the 15 markers between the DCIS and invasive groups, except at the TH01 locus which showed an increase in AI in the DCIS samples compared to the invasive tumors (P = 0.014).
The accumulation of genomic instability is characteristic of all carcinomas, including breast . It has been proposed that breast cancer progression can be modeled as a sequence of events progressing from normal epithelium to ADH, to DCIS, to finally the development of an invasive breast carcinoma [2, 3]. However, the genetic changes that underpin these histological changes still remain to be fully understood.
In this investigation we used TC and AI, two independent quantitative markers of genomic instability, to demonstrate that genomic instability increases as a function of the extent of breast disease (i.e. histologically normal tissue to ADH to DCIS). Alterations in TC and the extent of AI plateau in DCIS and do not increase further with increasing stage in invasive carcinomas. However, TC measurements show further telomere shortening between DCIS lesions and invasive carcinomas. The later finding is consistent with our previous studies demonstrating low TC compared to high TC confers an adjusted relative hazard of 4.43 (95% CI 1.4–13.6, P = 0.009)  in a cohort of 77 women. Additionally, in a population-based study of 530 women, low TC conferred an adjusted relative hazard of 2.88 (95% CI = 1.16–7.15; P = 0.022) .
Our TC findings are consistent with our previous reports that TC correlates with Stage in invasive carcinomas . Here, we show that 95% of the histologically normal breast tissues analyzed in this study fall within a range of 75–154% of the placental DNA control, nearly identical to the range previously reported , demonstrating that TC is tightly regulated regardless of inherent tissue properties that may affect TC, such as organ site or patients’ age. However, evidence of telomere dysregulation (i.e. attrition or elongation) was present in all the tumor cohorts. Speculatively, the finding of telomere elongation in tumors reflects the reactivation of telomerase, which is reactivated in 85–90% of tumors . However, the extent of reactivation varies amongst tumors as demonstrated by Hines and colleagues who showed an approximate 800-fold difference in telomerase expression among a panel of 36 breast tumors . Additionally, it has been postulated that early telomerase activation results in longer telomeres as compared to late activation, thus providing an opportunity for continued telomere shortening and accumulation of genomic instability.
Our observations confirm and extend the results of Ellsworth et al.  which demonstrated that levels of genomic instability are equivalent in DCIS lesions and advanced invasive tumors. However, that particular study utilized a panel of markers that were previously identified as important genes in the development of breast cancer. This confounds the ability to clearly interpret AI across these markers as genomic instability since these markers may be linked to oncogenes or tumor suppressor genes involved in the development of breast cancer. In contrast, the assay used in this study is based on AI at 16 random microsatellite regions that have no known involvement in the development of breast cancer, and thus reflect genomic instability independent of their linkage to genes involved with breast tumorigenesis. The differences in the extent of imbalance among the particular loci may reflect the proximity of the microsatellite region to the telomere ends. Chromosomal differences in telomere length may also contribute to the individual heterogeneity.
In conclusion, the level of genomic instability assessed by (i) dysregulation in TC (i.e. outside the 95% range found in normal breast tissue) and (ii) extent of AI assessed at 16 microsatellite loci located throughout the genome, increases along the continuum of breast disease from histologically normal, to ADH lesions to DCIS and the level of genomic instability did not differ between DCIS and invasive carcinomas. In all, these findings suggest that DCIS lesions have the same extent of genomic instability (i.e. TC alterations and increased AI) as invasive carcinomas; thus supporting the notion that invasive carcinomas evolve from or in parallel with DCIS.
This work was supported by grants DAMD17-01-1-0572, W81XWH-05-1-0226, W81XWH-05-1-0273 from the DOD Breast Cancer Research Program, NO-1-CN-65034-29 and SEER, NCI-PC-05016-20 from NCI/SEER and RR0164880 from the NIH. We thank Terry Mulcahy and Phillip Enriquez III from DNA Research Services of the University of New Mexico Health Sciences Center for gel capillary analysis.
Christopher M. Heaphy, Department of Biochemistry and Molecular Biology, MSC08 4670, 1 University of New Mexico, Albuquerque, NM 87131-0001, USA.
Marco Bisoffi, Department of Biochemistry and Molecular Biology, MSC08 4670, 1 University of New Mexico, Albuquerque, NM 87131-0001, USA.
Nancy E. Joste, Department of Pathology, University of New Mexico School of Medicine, 915 Camino de Salud, Albuquerque, NM 87131, USA.
Kathy B. Baumgartner, The New Mexico Tumor Registry, University of New Mexico School of Medicine, 915 Camino de Salud, Albuquerque, NM 87131, USA.
Richard N. Baumgartner, Department of Internal Medicine, University of New Mexico School of Medicine, 915 Camino de Salud, Albuquerque, NM 87131, USA.
Jeffrey K. Griffith, Department of Biochemistry and Molecular Biology, MSC08 4670, 1 University of New Mexico, Albuquerque, NM 87131-0001, USA, e-mail: ude.mnu.dulas@htiffirgkj.