Fax: (502) 629-3393
The replication error phenotype is associated with the development of distant metastases in hormonally treated patients with breast carcinoma
Article first published online: 22 JAN 2004
Copyright © 2004 American Cancer Society
Volume 100, Issue 5, pages 913–919, 1 March 2004
How to Cite
Chagpar, A., Magliocco, A., Kerviche, A., Tan, L., Walley, B. and DeCoteau, J. F. (2004), The replication error phenotype is associated with the development of distant metastases in hormonally treated patients with breast carcinoma. Cancer, 100: 913–919. doi: 10.1002/cncr.20084
- Issue published online: 18 FEB 2004
- Article first published online: 22 JAN 2004
- Manuscript Accepted: 9 DEC 2003
- Manuscript Revised: 3 DEC 2003
- Manuscript Received: 6 AUG 2003
- microsatellite instability;
- replication error phenotype;
- breast carcinoma;
The positive replication error (RER+) phenotype defines a distinct subgroup of tumors with specific clinical, pathologic, and molecular features that have been documented well in hereditary nonpolyposis colon carcinoma (HNPCC). More recently, this phenotype also has been described in breast carcinoma.
To determine the effect of RER phenotype on prognosis in patients with breast carcinoma, the authors examined matched archival tumor and normal tissue from 100 women with Stage I and Stage II breast carcinoma, all of whom were treated with hormonal therapy. Patients had been followed for a minimum of 5 years or until death. Seven microsatellite loci were examined, including hMLH1 (3p22, D3S1611), hMSH2 (2p16, D2S123), NM23-H1 (17q21), TP53-Dint (17p13), TP53-Penta (17p13), APC (5q21, D5S346), and HPC1 (1q24, D1S2883). The RER+ phenotype was defined as the presence of allelic shifts at three of the seven loci examined.
Twenty-five percent of patients were classified with the RER+ phenotype based on these criteria. The two groups, women with positive RER status and women with negative RER status, were comparable in terms of other factors that may influence prognosis: age, tumor size, lymph node status, disease stage, and estrogen receptor status. The development of distant metastases to the lung, liver, or brain was correlated significantly with the positive RER phenotype, with a relative risk of 2.625 (95% confidence interval, 1.059–6.057).
The presence of high-frequency RER+ may predict for the development of distant metastatic disease in patients with early-stage breast carcinoma whoa re treated with hormonal therapy. Cancer 2004;100:913–9. © 2004 American Cancer Society.
Breast carcinoma is the most common malignancy affecting women in the Western world today.1 It is estimated that, this year alone, 211,300 women in the U.S. will be diagnosed with breast carcinoma.2 Despite the advent of mammography and better public education, which have led to a trend toward smaller tumors at the time of diagnosis,3 a subset of patients with early-stage breast carcinoma will go on to develop distant metastatic disease.4 Axillary lymph node dissection, despite its significant morbidity,5 remains a cornerstone of surgical management in breast carcinoma, because lymph node status continues to be the best prognostic marker available for patients with this disease.6 Therefore, there has been an impetus for research into the molecular biology of breast carcinoma to find potential primary tumor markers that may serve to predict prognosis.7
It has been documented that the positive replication error (RER+) phenotype occurs in breast carcinoma,8 although its significance in determining prognosis remains unclear. This phenotype is characterized by defects in the mismatch-repair system that allow frameshift mutations to accumulate at multiple sites of highly repetitive, short segments of DNA called microsatellites to go unchecked.9 Although they usually are intronic, microsatellites also have been found in exons of transforming growth factor β1 receptor type II (TGFβ1 IIR),10 the insulin-like growth factor type II receptor,11 and the BAX gene.12 Patients with the RER+ phenotype may identify a subset of carcinomas with an accumulation of mutations in a susceptible set of vulnerable genes, resulting in altered behavior and prognosis compared with patients who have the RER negative (RER−) phenotype. Furthermore, the RER+ phenotype in bacteria has been associated with chemoresistance, especially to alkylating agents.13 To our knowledge, there have been no previous studies that examined the role of the RER phenotype on outcome in patients with breast carcinoma who were treated with hormone therapy. Therefore, we sought to determine the effect of the RER+ phenotype on prognosis in patients with early-stage breast carcinoma who were treated with adjuvant antiestrogens, both in terms of the development of distant metastatic disease and survival.
MATERIALS AND METHODS
Patients with Stage I or Stage II breast carcinoma who had undergone either modified radical mastectomy or lumpectomy, axillary lymph node dissection, and radiation therapy (5000 centigrays in 25 fractions) and who had received adjuvant hormonal therapy at the Saskatoon Cancer Center were identified retrospectively for the study. All of these patients had been followed for a minimum of 5 years or until death. Patients who died of causes other than their breast carcinoma (e.g., traffic accidents, heart attacks) were excluded from the study.
Patients were treated with oral tamoxifen (20 mg daily) postoperatively for 5 years or until they developed recurrent disease. Patients who were treated with chemotherapy were excluded from this study to eliminate confounding between different chemotherapeutic regimens, dosing schedules, and modifications secondary to toxicity. In addition, any interaction between the RER phenotype and chemoresistance also was avoided. If disease recurred while patients were receiving tamoxifen, then patients who were switched to another hormonal agent, such as oral megace at a dose of 160 mg daily, still were included in the study to prevent bias in the study against patients who developed recurrences while they were receiving tamoxifen. Of the patients who met the inclusion and exclusion criteria for this study, 100 women were selected randomly for the current study.
A detailed chart review was performed. Information collected included patient demographic information, pathologic information (including tumor size, lymph node status, tumor grade, and estrogen receptor [ER]/progesterone receptor status), and follow-up information (including the date of the development of metastasis, site of metastasis, last date of follow-up, and date of death). Using this information, a clinical data base was created for these 100 women.
Archival paraffin embedded blocks of matched cancerous tumor tissue and noncancerous, benign lymph node tissue from each patient were obtained prior to any systemic therapy. Normal lymph nodes, as identified by hematoxylin and eosin staining, and immunohistochemistry (where appropriate) were used as normal controls. DNA was extracted using the QIAmp tissue kit (Qiagen, Valencia, CA) according to the manufacturer's instructions. DNA was then quantitated by optical density, and 10 ng/μL solutions of DNA were prepared.
Microsatellite Polymerase Chain Reaction
Polymerase chain reaction (PCR) was performed on the matched tumor and normal DNA using seven commercially available, premixed primers for microsatellite instability and loss of heterozygosity (Microsatellite RER/LOH Assay; Applied Biosystems, Foster City, CA). The following loci were examined: hMLH1 (3p22, D3S1611), hMSH2 (2p16, D2S123), NM23-H1 (17q21), TP53-Dint (17p13), TP53-Penta (17p13), APC (5q21, D5S346), and HPC1 (1q24, D1S2883). For each PCR reaction, 1.15 μL sterile water, 0.1 μL AmpliTaq gold, 1.25 μL of 10 ng/μL DNA, and 7.5 μL of the primer mix were added. After an initial denaturing at 95 °C for 5 minutes, 45 cycles of amplification were performed at 96 °C for 10 seconds, 55 °C for 30 seconds, and 70 °C for 3 minutes. This was followed by extension at 70 °C for 30 minutes and a final soak at 4 °C.
The PCR products were pooled (1 μL each) into 2 separate pools for analysis with NM23, TP53-dint, TP53-penta, and MSH2 in one pool and with APC, HPC1, and MLH1 in the other pool. From each pool, 2.5 μL were then mixed with 12 μL of deionized formamide (Sigma Chemical Company, St. Louis, MO) and 0.5 μL Tamra 350 size standard (Applied Biosystems). The samples were then heated at 98 °C for 2 minutes and cooled on ice for 5 minutes prior to being loaded into the semiautomated ABI 310 genetic analyzer (Applied Biosystems). Results were analyzed using GeneScan 2.1 (Applied Biosystems). The RER+ phenotype was defined by the presence of microsatellite instability (allelic shifts) at three or more of the seven loci examined.
Statistical analyses were performed using SPSS software (version 10.0; SPSS Inc., Chicago, IL). The Fischer exact test was used for discrete variables, and the Mann–Whitney U test was used for continuous variables. Log-rank tests, Breslow tests, and Tarone–Ware tests were used for univariate survival analyses, and Cox regression was used for multivariate survival analyses.
Patients were between ages 34–83 years, with a median age of 68.5 years and a standard deviation of 10.35 years. The median tumor size was 2.5 cm with a standard deviation of 1.35 cm. Twenty-nine patients had negative lymph node status at the time of diagnosis, whereas the remaining 71 patients had positive lymph node status. Of the patients with positive lymph nodes, 23 patients had > 4 positive lymph nodes at the time of diagnosis. Seventy-three patients had positive ER status; 7 patients had negative ER status; and, in 20 patients, the hormone receptor status was unknown.
Twenty-five of 100 patients were classified with the RER+ phenotype. Tumors that expressed the RER+ phenotype had significant microsatellite instability: Of 7.0 markers, an average of 3.56 markers had allelic shifts. Tumors that, conversely, were labeled as RER− had an average of only 0.84 markers with detectable instability. An example of an electropherogram demonstrating microsatellite instability is shown in Figure 1. The two groups of patients, those with RER+ tumors and those with RER− tumors, were comparable in terms of other prognostic factors, including age, tumor size, lymph node status, and ER status (Table 1).
|No. of patients (%)||Significance (P value)|
|RER negative||RER positive|
|< 2 cm||20 (26.7)||9 (36.0)||0.447b|
|≥ 2 cm||55 (73.3)||16 (64.0)|
|Lymph node status|
|No. of positive lymph nodes|
|< 4||57 (76.0)||20 (80.0)||0.789b|
|≥ 4||18 (24.0)||5 (20.0)|
|Negative||23 (29.9)||6 (26.1)||0.724b|
|Ductal histology||64 (91.4)||20 (95.2%)||0.705b|
|Positive||51 (68.0)||22 (88.0)||0.425b|
|Negative||6 (8.0)||1 (4.0)|
|Unknown||18 (24.0)||2 (8.0)||0.184b|
Of the 25 patients who had the RER+ phenotype, 7 patients (28%) developed distant metastases to the lung, liver, or brain. Conversely, of 75 patients with the RER− phenotype, 8 patients (10.7%) developed distant metastasis. This was statistically significant (P = 0.05; Fisher exact test). The RER+ phenotype was associated with the development of distant metastasis, with a relative risk of 2.625 (95% confidence interval, 1.059–6.057).
Univariate analysis was also performed to evaluate the relation between other known prognostic markers and distant metastatic disease. In this series, RER status was found to be a stronger indicator of the development of distant metastases than tumor size, lymph node status, disease stage, or ER status in univariate analysis (Table 2).
|Measure||Significance (P value)a|
|Positive RER status||0.05|
|Positive lymph node status (≥ 4 positive lymph nodes)||0.104|
|Tumor size (≥ 2 cm)||1.000|
We then performed univariate analyses for various traditional prognostic variables on overall survival. Kaplan–Meier survival analyses were performed using calculations of log-rank, Breslow, and Tarone–Ware tests for each of the following variables: RER status, patient age, tumor size, lymph node status, and ER status. The results of these tests are shown in Table 3. Age, tumor size, and lymph node status all were found to be associated significantly with survival. ER status, however, appeared to have no bearing on survival. Despite its association with the development of distant metastases, RER status did not show a statistically significant correlation with survival, although there was a 20-month median survival advantage for patients with the RER− phenotype. A Cox multiple regression survival analysis was performed to examine the effect of RER on survival, taking patient age, lymph node status, and tumor size into account (Fig. 2). Although it did not achieve statistical significance, there was a trend toward decreased survival in patients with the RER+ phenotype, with an odds ratio of 1.826 (95% confidence interval, 0.836–3.987).
|Variable||Median survival in mos (range)||P value|
|≤ 65 yrs||303.07 (244.50–361.65)|
|> 65 yrs||129.16 (94.19–164.13)||0.0019||0.0028||0.0022|
|< 2 cm||151.83 (115.35–185.95)|
|≥ 2 cm||131.00 (94.67–167.33)||0.0325||0.0295||0.0259|
|No. of positive lymph nodes|
|< 4||212.08 (112.82–301.34)|
|≥ 4||88.83 (52.53–125.13)||0.0008||0.0042||0.0021|
|Lymph node status|
Since replication errors first were discovered in colon carcinoma,14 there has been increasing research done in the area of microsatellite instability. The RER+ phenotype, although it has been studied most extensively in hereditary nonpolyposis colon carcinoma (HNPCC),15 also has been described in a variety of other malignancies, including endometrial,16 ovarian,17 cervical,18 gastric,19 lung,20 prostate,21 and skin22 carcinomas. Microsatellite instability also has been found in breast carcinoma, with reported frequencies of between 0% and 52%.23–45 To our knowledge, to date, standardized definitions for the RER+ phenotype have not been accepted by the international community for noncolonic tumors.46 With our definition, using microsatellite instability at ≥ 3 of 7 loci examined, we were able to identify 2 distinct groups, a group with a high frequency of allelic shifts (average, 3.56 loci), which we labeled RER+, and another group with a low frequency of allelic shifts (average, 0.87 loci), which we labeled RER−. Doing so, we found a 25% incidence of the RER+ phenotype in our cohort of patients with breast carcinoma, which correlates well with most previous studies.
Similar to the majority of reports in the literature, we also found no association between the RER+ phenotype and age,29–32, 38–40, 47 tumor size,23, 26, 29, 34, 36, 39, 40, 47 lymph node status,23, 26, 30, 34, 38–40, 47 or ER status24, 33, 34, 38, 39, 41 in our cohort of patients. This enabled us to compare the RER+ and RER− groups directly with respect to distant metastases. We found a statistically significant correlation between the RER+ phenotype and the development of distant metastatic disease along with a trend toward decreased survival, although this latter finding did not reach statistically significant levels.
However, in a heterogeneous population of patients with breast carcinoma, Paulson et al.28 found that the RER+ phenotype correlated with both increased distant metastatic disease and poorer survival. There are a number of possible explanations for this finding. To begin with, Paulson et al. found a correlation with positive axillary lymph nodes and larger tumor size at the time of diagnosis, which may account in part for the observed increase in mortality. Second, although all of our patients received adjuvant hormonal therapy, it is unclear which (if any) adjuvant treatment was used for Paulson and colleagues' patients. Potentially, if patients were treated with chemotherapy, then there may be an element of chemoresistance associated with patients who have RER+ tumors that added to their mortality. Finally, the statistical significance achieved in the current study between RER status and distant metastases may not have been strong enough to correlate into a significant drop in survival. Certainly, this is an area that warrants further research.
Nonetheless, the RER+ phenotype appears to be associated with a poorer prognosis compared with its RER− counterpart. This is in direct contrast to what has been observed in colon carcinoma,48, 49 in which the RER+ phenotype is correlated with increased survival. It is speculated that cells developing microsatellite instability will accumulate mutations in a set of vulnerable genes that contain polynucleotide repeat tracts in their coding regions. This set of mutated genes will result in the tumor developing the characteristics observed to occur in association with positive RER status. It is possible that genes also may be targeted in a tissue-specific manner.50, 51 For example, it is has been well established that frameshift mutations often are found in the microsatellite region of the TGFβ IIR gene in HNPCC.10 This has not been found in breast carcinoma.50, 51 Similarly, alterations in the insulin-like growth factor II receptor gene11 and the antiapoptotic BAX gene,12 which often are affected in colonic tumors with the RER+ phenotype, have not been observed in breast carcinomas.52 It is interesting to note that mutations in the microsatellite region of PTEN/MMAC1, a gene encoding a phosphatase that interacts with adhesion molecules, have been reported in RER+ breast malignancies and may be a target gene in these patients.53 The correlation between the RER+ phenotype and prognosis, thus, is one that should be investigated in a multitude of different tumors, because the genes affected by defects in the mismatch-repair system in various malignancies may be specific to the site of origin. To our knowledge, it is unknown why there is a tissue-specific distribution of gene mutations in association with the RER+ phenotype. Possible explanations include tissue-specific gene expression coupled with local selection pressures during oncogenesis. If a gene normally is not transcribed in a target tissue, then it is unlikely that mutations in its coding region will have much consequence in carcinogenesis. Conversely, mutations that convey growth advantages will be selected for in developing subclones of tumors.
However, there seems to be a correlation in breast carcinoma between the RER+ phenotype and poorer prognosis in the setting of adjuvant hormonal therapy. It is believed that RER+ tumors may be resistant to cytotoxic chemotherapeutic regimens, and our data suggest that this phenotype also is associated with the development of distant metastatic disease in patients who are treated with antiestrogens. Whether this is due to intrinsic mutations in the tumor that cause it to be more aggressive or whether there is an element of resistance to hormonal agents remains an area to be explored. The RER+ phenotype, however, is a potentially useful predictive marker in patients with early-stage breast carcinoma. The finding that this marker, which appears to be independent of patient age, tumor size, and axillary lymph node status, may be able to predict the development of distant metastatic disease in a population of patients with early-stage breast carcinoma may have significant implications for clinical management.
- 2American Cancer Society. Cancer facts and figures 2003. Available from URL: http://www.cancer.org/downloads/STT/CAFF2003PWSecured.pdf [accessed May 18, 2003].
- 16Clinical and pathological significance of microsatellite instability in sporadic endometrial carcinoma. Am J Pathol. 1997; 148: 109–112., , , et al.
- 24Microsatellite alterations on human chromosome 11 in in situ and invasive breast cancer: a microdissection microsatellite analysis and correlation with p53, ER (estrogen receptor), and PR (progesterone receptor) protein immunoreactivity. J Surg Oncol. 2000; 74: 100–107., , , et al.
- 46A National Cancer Institute Workshop on Microsatellite Instability for cancer detection and familial predisposition: development of international criteria for the determination of microsatellite instability in colorectal cancer. Cancer Res. 1998; 58: 5248–5257., , , et al.