Fax: (202) 806-5553
Functionally active estrogen receptor isoform profiles in the breast tumors of African American women are different from the profiles in breast tumors of Caucasian women
Article first published online: 31 JAN 2002
Copyright © 2002 American Cancer Society
Volume 94, Issue 3, pages 615–623, 1 February 2002
How to Cite
Poola, I., Clarke, R., DeWitty, R. and Leffall, L. D. (2002), Functionally active estrogen receptor isoform profiles in the breast tumors of African American women are different from the profiles in breast tumors of Caucasian women. Cancer, 94: 615–623. doi: 10.1002/cncr.10274
- Issue published online: 31 JAN 2002
- Article first published online: 31 JAN 2002
- Manuscript Accepted: 1 OCT 2001
- Manuscript Revised: 21 SEP 2001
- Manuscript Received: 3 AUG 2001
- National Cancer Institute. Grant Number: CA88347
- Susan G. Komen Breast Cancer Foundation. Grant Number: 99DC1028
- African-American women;
- aggressive breast tumors;
- breast tumorigenesis;
- ER isoform alterations;
- constitutively active estrogen receptor α;
- dominant negative estrogen receptor α;
- estrogen receptor β;
- alternate adjuvant therapy
Several cancer surveys have shown that African-American women (AAW) develop highly aggressive breast tumors and experience about three times higher mortality rates compared with other populations. Generally, breast tumors in AAW are poorly differentiated or undifferentiated and exhibit increased frequency of nuclear atypia, higher mitotic activity, higher S-phase fraction, and tumor necrosis. The molecular factors responsible for these tumor characteristics are mostly unknown.
To explore whether the aggressive tumor biology observed in AAW is related to distinct alterations in estrogen receptor (ER) isoforms, the relative expression levels of four functionally active ER isoform mRNAs, ERα wild type, ERβ wild type, ERα exon 3Δ, and ERα exon 5Δ, were measured by reverse transcriptase-polymerase chain reaction analysis in 18 immunohistochemically ERα positive tumors and in 6 ERα negative tumors and their matched normal tissues.
In the tumors of AAW, the protective ERβ isoform was decreased significantly compared with matched normal tissues (paired t test; n = 24 patients; P = 0.0018). In addition, both the constitutively active ERα exon 5Δ and the dominant negative ERα exon 3Δ mRNA levels were elevated in tumor tissues compared with matched normal tissues (paired t tests; n = 24 patients; P = 0.0002 and P = 0.024, respectively).
The data presented here show for the first time that functionally active ER isoform profiles in the breast tumors of AAW are different from those in Caucasian women. The tumors in AAW are characterized by decreased levels of the protective ERβ isoform and elevated levels of the constitutively active ERα exon 5Δ isoform. Variations in estrogen-mediated signaling because of the alterations in these two ER isoforms may account in part for differences in tumor biology between AAW and Caucasian women. Cancer 2002;94:615–23. © 2002 American Cancer Society.
A number of cancer surveys have shown that African American women (AAW) develop very aggressive breast tumors and experience about three times higher mortality rates and poorer survival compared with women in other populations.1–4 This higher mortality rate has been attributed in part to late stage of disease at the time of diagnosis, low socioeconomic status, and limited access to medical facilities and services.5–8 When these factors are controlled, the disparity in survival and mortality rates appears to be due to differences in tumor biology observed in AAW.4 Several studies have established that breast tumors in black women from both United States and South Africa are poorly differentiated or undifferentiated with increased frequency of nuclear atypia, higher mitotic activity, higher S-phase fraction, and tumor necrosis.9–11 Another characteristic of breast tumors in AAW is the frequency of expression of estrogen receptor (ER) and progesterone receptor (PR), the presence of which indicate a good prognosis and response to antiestrogen and other therapies.12–14 Several reports have shown that the frequency of ER positive/PR positive tumors in AAW is considerably lower (44%) compared with Hispanic and non-Hispanic Caucasian women (58% and 59%, respectively) after adjusting for menopausal status and age.15–17 The genetic and environmental factors that may contribute to the differences in tumor biology and the expression of hormone receptors across racial lines remain unknown.
Until recently, it was believed that estrogen action in breast tissues is mediated by only one functional ER, the classic ERα. A number of ER isoforms, including the recently discovered ERβ,18 20 splice variants of ERα,19–25 and 10 splice variants of ERβ,26–29 (Pooleetal et al., submitted, 2001) also have been detected in both normal tissues and breast tumor tissues. In vitro studies have shown that, in addition to ERα wild type, three other ER isoforms, the ERβ wild type, ERα exon 3Δ, and ERα exon 5Δ, also are active functionally.30–33 It has been shown that two of these ER isoforms (ERβ and ERα exon 3Δ) bind various ligands, and all three isoforms have distinct transactivating properties. The ERα and ERβ wild type molecules have significant differences in transactivating properties, depending on the requirement of AF-1 and AF-2 (the N-terminal and C-terminal regions, respectively, that interact with steroid receptor activators/repressors) sites for transcription of a target gene. If both are required, then the activity of ERα exceeds the activity ERβ. If only the AF-1 site is required, then the activity of ERβ seems negligible.32 In addition, ERα and ERβ wild type have opposite effects on transcription regulated by AP-1 sites. Estrogenic compounds appear to act as antagonists, and antiestrogenic compounds act as agonists if they are bound to ERβ at AP-1 sites. A recent report by Gustafsson and Warner33 showed that elimination of ERβ leads to the abnormal growth of mammary epithelial cells and increased expression of Ki-67, an indicator of proliferation rate.
Recent studies also have shown that a protein translated from ERα exon 5Δ translocates to the nucleus and possesses a constitutive transactivating property in the absence of estrogen.20, 21, 34 This protein also inhibited the wild type receptor by competing with the steroid receptor coactivator-1e (SRC-1e). The ERα exon 3Δ protein binds ligands with the same affinity as the wild type receptor, has a positive transactivating property at the AP-1 site, and inhibits the transactivation of wild type ERα by forming heterodimers and competing for SRC-1e. The two modified receptors seem to completely inhibit the wild type receptor if it is present in equimolar amounts.34 These studies with the above three receptors indicate that the response to a particular ligand may depend on the types of functionally active ERs and/or their relative amounts present in cells and tissues. Such differences in ER isoform composition may account for variations in tissue/tumor biology.
To determine whether the aggressive breast tumor biology observed in AAW is associated with variations in the ER isoform composition, we investigated the relative expression of four functionally active ER isoform mRNA levels: the ERα wild type, the ERβ wild type, ERα exon 5Δ, and ERα exon 3Δ. These mRNAs were measured by reverse transcriptase-polymerase chain reaction (RT-PCR) analysis in breast tumor tissues and in matched normal tissues. The ERα exon 5Δ and ERα exon 3Δ mRNAs were amplified using the newly developed spliced targeted primer approach that amplifies each category of splice variants as separate gene populations.23, 35 Our results described here show that the protective ERβ wild type was decreased significantly in tumor tissues compared with matched normal tissues. In addition, the constitutively active ERα exon 5Δ and the dominant negative exon 3Δ mRNAs were elevated significantly in tumor tissues compared with matched normal tissues. These changes are distinct from the alterations observed in tumors from Caucasian patients. These observations also suggest that differences in tumor biology observed between AAW and Caucasian women may be due in part to the variations in functionally active ER isoform composition.
MATERIALS AND METHODS
HotStartTaq PCR core kits, Omniscript reverse transcriptase kits, and MinElute gel extraction kits were from QIAGEN Inc. (Santa Clara, CA). All primers used in the current study were synthesized by Gibco-BRL Life Technologies (Gaithersburg, MD). The pCR®2.1-TOPO cloning vector was obtained from Invitrogen (San Diego, CA). PCR quality water and Tris-ethylenediamine tetraacetic acid (EDTA) buffer were from Biofluids (Rockville, MD).
Breast Tumor Samples
Breast tumor samples and their matched normal tissues from African-American patients were collected from Howard University Hospital immediately after surgery and stored at −80 °C until use. Tumor samples for research were routinely harvested immediately adjacent to the histologic/diagnostic section and were considered representative of the tissues used for diagnosis. Matched normal tissues were collected from a region distant from the tumor tissue of the same patient. All normal tissues used in the current study were proven devoid of any tumor tissue by histochemical staining by the attending pathologist. A total of 24 malignant tumor tissues (18 immunohistochemically ERα positive and 6 ERα negative) and their matched normal tissues were included in the current study. The tumor collection procedure was approved by the Howard University Institutional Review Board Committee. The ERα status of tumor tissues was obtained from the Tumor Registry, Howard University Cancer Center. Immunohistochemical staining for the presence of ERα was done by Oncotech Laboratories (Irvine, CA) using the monoclonal antibodies against the amino-terminal A/B region of the receptor. According to the Tumor Registry data, 19 tumors were lymph node positive, and 5 tumors were lymph node negative. One tumor was Grade 1, 9 tumors were Grade 2, 10 tumors were Grade 3, and 4 tumors were not graded. Of 24 tumors, 16 were ductual, 1 was medullary, 1 was adenocarcinoma, 1 was Paget disease mammary carcinoma in situ, 1 was lobular carcinoma, 1 was oxyphilic adenocarcinoma, 1 was intraductal papillary adenocarcinoma, and 1 was comedocarcinoma. The patients ranged in age from 38 years to 87 years (mean, 60 years).
RNA Extraction and cDNA Synthesis
Total RNA was extracted from frozen breast tissues using the Trizol reagent (Gibco-BRL Life Technologies), as described previously.36 RNA integrity was verified by both electrophoresis in 1.5% agarose gel electrophoresis and amplification of the constitutively expressed gene, glyceraldehyde-3 phosphate dehydrogenase (GAPDH). The isolated RNA was reverse transcribed using Omniscript reverse transcriptase as described previously.37 Briefly, the standard mixture contained 1 μg of total RNA, 10 U of RNAse inhibitor, 0.5 mM of each dNTP, 1 μM random hexamers, and 4 U of Omniscript reverse transcriptase in a total volume of 20 μL. For reverse transcription, tubes were incubated at 37 °C for 60 minutes, at 95 °C for 5 minutes, and finally rapidly cooled.
PCR analyses were performed in an automatic thermal cycler (MJ Research, Cambridge, MA), as described previously,38 in a total volume of 12.5 μL containing cDNA that was reverse transcribed from 125 ng of total RNA, 1 × PCR buffer, 1 × Q solution, 200 μM of each dNTP, 2 μM each of sense and antisense primers, and 0.6 U of HotStartTaq polymerase. GAPDH was amplified in parallel with cDNA that was reverse transcribed from 50 ng of total RNA. cDNA from the breast carcinoma cell line T47D was used as a positive control in all PCR analyses. The PCR conditions were initial denaturation for 1 minute at 95 °C, then at 94 °C for 1 minute, annealing for 1 minute at the specified temperature (depending on the primer pair used), extension for 2 minutes at 72 °C for 40 cycles, and final extension for 15 minutes at 72 °C. For amplification of GAPDH, transcripts were amplified for only 25 cycles. The annealing temperature for ERα wild type, ERβ wild type, and ERα exon 3Δ primers was 55 °C, and the annealing temperature for ERα exon 5Δ primer was 65 °C. Each receptor was amplified independently using specific primers. The wild type ERα sequences were amplified using a sense primer, ERαS (5′-TGCCCTACTACCTGGAGAACG-3′; position: exon 1, base pairs [bp] 615–635), and an antisense primer, ERα1A (5′-GTCCTTCTCTTCCAGAGAC-3′; position: exon 7, bp 1651–1633). ERβ wild type sequences were amplified using a sense primer, ERβS (5′-CGCTAGAACACACCTTACCTG-3′; position: exon 1, bp 433–453), and an antisense primer, ERβA (5′-CTGTGACCAGAGGGTACAT-3′; position: exon 7, bp 1344–1326). ERα exon 5Δ was amplified with ERαS and a splice-targeted, antisense primer, ERα AX4/6 (5′-ATTTTCCCTGGTTC6/4CTGGCAC-3′; positions: exon 6, bp 1481–1468; exon 4, bp 1328–1322), and ERα exon 3Δ was amplified using a splice-targeted sense primer, ERα SX2/4 (5′-AAGAGAAGTATTCAAG2/4GGATA-3′; positions: exon 2, bp 860–875; exon 4, bp 993–997), and an antisense primer, ERα2A (5′-GCACTTCATGCTGTACAGATGC-3′; position: exon 8, bp 1822–1801). The design, specificity, and applicability of splice-targeted primers in amplifying alternatively spliced ER mRNAs as separate gene populations in breast carcinoma cell lines and tumors have been well established.23, 35, 39 The GAPDH was amplified using a sense primer (5′-AAGGCTGAGAACGGGAAGCTTGTCATCAAT-3′; position: exon 3, bp 241–270) and an antisense primer (5′ TTCCCGTCTAGCTCAGGGATGACCTTGCCC-3′; position: exon 7, bp 740–711).40 The sequence and locations of ER primers were based on the full-length cDNA sequences.41, 42
Detection, Identification, and Quantitation of PCR Products
The PCR amplified products (6.5 μL) were separated by electrophoresis in 1% agarose gels in Tris-acetic acid-EDTA buffer and detected by ethidium bromide staining. To detect the PCR products of GAPDH, 1 μL was used. The identities of the PCR amplified ER products were determined by cloning the gel-purified products into pCR®2.1-TOPO vector and sequence analysis, as described previously.23 The fluorescence intensities of the PCR products were quantitated densitometrically by scanning the ethidium bromide-stained gel photographs and analyzing the images using the NIH Image software program (version 1.6; Bethesda, MD), as described previously.37 All values were first normalized to GAPDH levels then compared between the matched samples.
A paired t test was used to compare the expression of ER isoform mRNA levels in normal tissues and tumor tissues. Statistical significance was tested for the changes in the expression levels of ERα wild type, ERβ wild type, ERα exon 5Δ, and ERα exon 3Δ. All 24 samples, irrespective of ERα status, were evaluated to determine the statistical significance of changes in these ER mRNA levels. The results were considered significant if the P values were ≤ 0.05.
Wild Type ERβ Levels Were Decreased Significantly in Tumor Tissues Compared with Matched Normal Tissues
To measure the relative expression levels of wild type ERβ mRNA levels by RT PCR analysis, we used primers between exons 1 and 7, so that the wild type product obtained was exclusively from the wild type receptor mRNA and was not the combined total of products of wild type and the unaltered wild type regions of alternatively spliced ERs. If short sequences encompassing only two or three exons are amplified, then a number of splice variants that are not modified in between the primers used also will be amplified. By using primers between exons 1 and 7, any artifacts introduced due to changes in the splice variant levels were excluded.
The primers ERβS and ERβA amplified a wild type, 911-bp product along with a number of lower molecular weight products, presumably the splice variants. For the purpose of comparison between normal and tumor samples, only the 911-bp product was quantified densitometrically, as described above (see Materials and Methods). The results of calculated scanning units for each matched sample are presented in Table 1, which shows that the ERβ mRNA levels were reduced significantly (40–60%) in tumor tissues compared with matched normal tissues (paired t test; n = 24 patients; P = 0.0018) in the immunohistochemically ERα positive group. In contrast, in the ERα negative tumor group, ERβ mRNA levels were either unchanged or slightly increased. Three examples of matched samples from each group are shown in Figure 1.
|Tumor||ERα status||Grade||Changes in ERβ mRNA levels (scanning units)||Changes in ERα mRNA levels (scanning units)||ERα to ERβ ratio||Changes in ERα exon 5Δ mRNA levels (scanning units)||ERα wild type to ERα exon 5Δ ratio||Changes in ERα exon 3Δ mRNA levels (scanning units)||ERα wild type to ERα exon 3Δ ratio|
The ERα wild type mRNA was amplified between exons 1 and 7 using the primers ERαS and ERα1A. These primers amplified a 1036-bp product along with several lower molecular weight products, like the ERβ mRNA, presumably the splice variants. For measuring the wild type levels, only this 1036-bp product was evaluated. Using this approach, we did not observe any significant changes in the ERα wild type levels between tumor tissues and the matched normal tissues in the immunohistochemically ERα positive tumor group. In the immunohistochemically ERα negative tumor group, as expected, ERα mRNA levels were reduced in tumor tissues compared with matched normal tissues (Table 1). An example of matched samples from each group is shown in Figure 2.
Although ERα mRNA levels were not altered, the ERα-to-ERβ ratio was increased significantly in tumor tissues compared with matched normal tissues (paired t test; n = 24 patients; P = 0.0069) (Table 1) because of a 40–60% reduction in ERβ mRNA levels.
Constitutively Active ERα Exon 5Δ and Dominant Negative ERα Exon 3Δ Isoform mRNAs Were Elevated Significantly in Tumor Tissues Compared with Matched Normal Tissues
To determine whether the structurally altered, functionally distinct receptors described above may arise during the genesis of aggressive breast tumors in AAW, we measured their relative expression levels by RT PCR analysis using splice-targeted primers. The 5Δ splice specific antisense primer, ERα AX4/6, and a sense primer in exon 1, ERαS, amplified three products of sizes, 750 bp, 540 bp, and 420 bp (Fig. 3). They were identified as exon 5Δ; exons 5&2Δ; and exons 5, 2–3Δ, respectively, by sequence analyses. For quantitation, all three products were scanned, and the scanning units were combined. The scanning unit totals from these three products were compared between normal and tumor tissues. The results are presented in Table 1, which shows that normal tissues either were expressed at very low levels or were undetectable. In the matched tumor tissues, the exon 5Δ products were elevated significantly (paired t test; n = 24 patients; P = 0.0002). Examples of the results obtained with five matched samples are shown in Figure 3.
The relative mRNA levels of exon 3Δ were measured by RT PCR analysis using the 3Δ splice specific sense primer, ERα SX2/3, and an antisense primer in exon 8, ERα2A. These primers amplified two products of sizes, 845 bp and 661 bp. These two products were identified as exon 3Δ and exons 3&7Δ, respectively, by sequence analysis. For quantitation, both products were scanned and combined, and the scanning unit totals were compared between matched samples, Table 1 shows that normal tissues were expressed at very low or undetectable levels of these two mRNAs. Both exon 3Δ and exons 3&7Δ levels were elevated significantly in the matched tumor tissues (paired t test; n = 24 patients; P = 0.024). Examples of the results obtained with five matched samples are shown in Figure 4.
Estrogen plays an important role in the physiology of mammary tissue and is considered to be a mitogen in the genesis and progression of human breast carcinoma. The discovery of more than one functional receptor for estrogen in breast tissues with distinct ligand-binding and transcriptional properties suggests that the complex biology governed by estrogen in this tissue may be regulated by a delicate balance between the types and relative amounts of various ERs. Changes in the composition of functionally active ERs may perturb the estrogen-mediated signaling processes considerably and may contribute in part to tumorigenesis. Variations in functionally active ER isoform profiles also may alter tumor biology and characteristics. Recently, relative expression levels of four functionally active ER isoforms were investigated extensively in normal tissues and breast carcinoma tissues from Caucasian patients. Speirs et al.43 studied the relative expression levels of ERα and ERβ wild type by semiquantitative RT PCR approaches. Those authors showed that ERβ predominates in normal tissues and that ERα predominates in tumor tissues. They also observed an increase in the ERα levels in tumor tissues compared with normal tissues but did not observe a significant change in ERβ levels. Leygue et al.44 studied the relative expression levels of ERα and ERβ wild type in breast carcinoma tissues and their matched normal tissues from Caucasian women by semiquantitative RT PCR analysis. Those authors also established that ERα levels were increased significantly, but ERβ levels were unchanged in tumor tissues compared with matched normal tissues. The data from both groups indicated that the normal tissues and breast tumor tissues were heterogeneous in their expression levels of these two ERs.
In the current study, we investigated the changes in four functionally active ERs in breast tumor tissues and their matched normal tissues from AAW. We hypothesized that the differences in breast tumor biology observed between Caucasian women and AAW may be due to differences in the ER isoform composition. The data showed that, in AAW, ERβ levels were decreased significantly in tumor tissues compared with matched normal tissues (paired t test; n = 24 patients; P = 0.0018). In the tumors from Caucasian women, the decrease in ERβ mRNA levels was not significant. Tumors in AAW showed a slight increase in ERα mRNA levels, but this did not reach the level of statistical significance. The ERα-to-ERβ ratio was increased in tumor tissues compared with normal tissues in both Caucasian women44 and AAW. The increase in AAW appears to be due to decreased levels of ERβ, in Caucasian tumors due to an increase in ERα expression.
In addition to the expression levels of ERα and ERβ, we also measured the relative expression levels of ERα exon 5Δ and exon 3Δ mRNAs. Although the presence of these two ERs has been known in breast tumors for more than 10 years, only recently, Bollig and Miksicek34 have established that these two structurally altered ERs have distinct ligand-binding and transcriptional properties. The results presented here clearly show that both exon 5Δ and 3Δ mRNAs are elevated significantly in the tumor tissues (Figs. 3, 4, respectively; Table 1) compared with matched normal tissues (P = 0.0002 and P = 0.024, respectively). In a separate study, we investigated the expression of these two ERs in normal tissues and breast tumor tissues from Caucasian patients that were analyzed previously for wild type ERα and ERβ by Speirs et al.43 When we used the same primers that were used in the current study, we could not amplify any of the 750-bp, 540-bp, or 420-bp 5Δ products either in normal tissues or in tumor tissues.40 In contrast, 15 of 35 normal tissue specimens showed the presence of a single exon deletion variant, exon 3Δ. The frequency of expression of this variant was increased (26 of 38 samples) in tumor tissues, as was the expression of the double-deletion variant, exons 3&7Δ (P = 0.000032).39 Table 1 shows that most matched normal tissues from AAW express both 3Δ and 3&7Δ mRNAs, although at a lower levels compared with tumor tissues.
In summary, in this report, data are presented showing that alterations in the relative expression of ER isoforms during breast tumorigenesis in AAW are different from those of Caucasian patients. Tumors in AAW are characterized by elevated levels of constitutively active ERα exon 5Δ, dominant negative ERα exon 3Δ mRNAs and an increased ERα-to-ERβ ratio due to a decrease in ERβ mRNA levels. Although tumors from both Caucasian women and AAW show increased expression of exon 3Δ, only tumors from AAW show a significant increase in exon 5Δ and a decrease in ERβ mRNA levels. Alterations in these three ER expression levels may change considerably the estrogen-mediated signaling in a number of ways. First, because estrogenic compounds act as antagonists when bound to ERβ at AP-1 sites,30 and because this receptor also appears to protect against abnormal mammary epithelial cell growth,33 decreased levels of ERβ may reduce and/or eliminate this protective effect. Second, because exon 5Δ has constitutive transactivating properties in the absence of estrogen,20, 21, 34 elevated levels of this ER may increase the transcription of genes and inhibit the controlled transcription of wild type ERα by competing for SRC-1e. Finally, elevated levels of exon 3Δ receptor may increase transcription of genes at the AP-1 site and inhibit the controlled, transactivating property of wild type ERα by forming heterodimers and competing for SRC-1e. The overall effect due to alterations in the relative expression levels of these three ERs may be an increase in unregulated transcription and a reduction in ligand dependent transcription. These distinct changes in transcription resulting from ER isoform alterations may contribute in part to the aggressive tumor biology observed in AAW. The data presented in Table 1 also show that every normal tissue and tumor tissue sample had distinct levels of each receptor: For example, although Tumors 1 and 4 had similar ERβ levels, they had different levels of ERα exon 5Δ. The heterogeneity in various functionally active ER isoform levels may have clinical implications for the degree of response to a particular antiestrogen as well as other therapies.
According to the PubMed and MEDLINE literature searches, this is the first report to show the alterations in the constitutionally active ERα exon 5Δ during breast tumorigenesis. The data presented here also show that the relative levels of each functional ER in AAW tumors are different from the tumors in Caucasian patients. The differences in estrogen-mediated signaling through the above ER isoforms may account for the differences in tumor biology observed in AAW and Caucasian women. Finally, the difference in the ER isoform composition in AAW may be exploited to close the gap in the mortality rate between the two racial groups by designing alternate adjuvant therapies targeted at blocking ERα exon 5Δ and/or ERα exon 3Δ along with ERα wild type blockers for AAW.
The authors acknowledge Dr. Aiyi Liu of Georgetown University Medical School for statistical analyses and Ms. Sailaja Koduri for processing the tumor samples; Howard University Hospital Surgical Pathology Faculty members for their assistance in collecting fresh breast tumor samples; and Ms. Rosemary Williams, Director, Tumor Registry, Howard University Cancer Center, for her assistance in procuring pathology reports.
- 1American Cancer Society. Cancer facts and figures—2001. Atlanta: American Cancer Society, 2001.
- 3HaynesMA, SmedleyBD, editors. The unequal burden of cancer: an assessment of NIH research and programs for ethnic minorities and the medically underserved. Washington, DC: National Academic Press, 1999.
- 4Breast cancer in black women. Ann Intern Med 1996; 15: 897–905..
- 6The relationship of socioeconomic status and access to minimum expected therapy among female breast cancer patients in the National Institute Black-White Survival Study. Ethnic Dis 1999; 9: 111–25., , , .
- 7American Cancer Society. Cancer and the socioeconomically disadvantaged. Atlanta: American Cancer Society: Atlanta, GA, 1990.
- 21Expression of a constitutionally active estrogen receptor variant in the estrogen receptor negative BT-20 human breast cancer cell line. Cancer Res 1993; 53: 593–9., , , .
- 27Estrogen receptor beta mRNA variants in human and murine tissues. Mol Cell Endocrinol 1998; 7: 205–11., , , , .