University of Southern California Dissertations and Theses

Androgen receptor gene and prostate -specific antigen gene in breast cancer

(USC Thesis Other)

Androgen receptor gene and prostate -specific antigen gene in breast cancer

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ANDROGEN RECEPTOR GENE AND PROSTATE-SPECIFIC ANTIGEN GENE IN BREAST CANCER Copyright 2005 by Wei Wang A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (EPIDEMIOLOGY) December 2005 Wei Wang Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3220164 INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. ® UMI UMI Microform 3220164 Copyright 2006 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DEDICATION To my parents Tianji Wang and Heqin Deng and to my husband Yong Liu —I could not have been able to bring my dissertation work into completion without your encouragement and support. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS This work was done under the direction and supervision of Dr. Sue Ann Ingles. I am most grateful to Dr. Ingles for her professional guidance and her support throughout my Ph.D. training. I appreciate the many hours she spent working with me on my various projects, through which I have gained invaluable experience in the field o f Molecular Epidemiology. I am indebted to Dr. Esther M. John for providing me access to her Northern California Breast Cancer Study dataset and biospecimen, for her generous support, and for her invaluable comments and advice on my dissertation and manuscript. My appreciation also extends to the other members of my committee, Dr. Gerald Coetzee, Dr. Ronald Ross, and Dr. Mimi Yu for their numerous constructive comments and suggestions on my dissertation and manuscript, and great ideas that furthered my work on androgen down-stream genes. Special thanks go to Dr. David Conti for his advice on statistical analysis, Dr. Frank Stanczyk and his lab for the measurement of serum PSA and androgen levels. I am also very grateful to my labmates and friends, Jun Wang, Huilee Wong and Melissa Wilson for critical review o f my dissertation, encouraging comments and helpful discussions. iii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS DEDICATION ii ACKNOW LEDGEMENTS iii LIST OF TABLES vii LIST OF FIGURES x ABSTRACT xi INTRODUCTION 1 CHAPTER 1. ANDROGENS, ANDROGEN RECEPTO R AND THE CAG REPEATS IN BREAST CANCER (REVIEW ) 5 1.1. Androgen production in females 5 1.2. Experimental and clinical studies on androgens and androgen receptor in breast cancer 6 1.3. Possible mechanisms implicated in the growth regulatory effects of androgens in breast cancer 9 1.4. Epidemiological studies on circulating androgen levels and breast cancer risk 12 1.5. The androgen receptor gene and the CAG repeat polymorphism 18 1.5.1. AR CAG repeats and prostate cancer 20 1.5.2. AR CAG repeats and male breast cancer 21 1.5.3. Epidemiological studies on AR CAG repeats and female breast cancer in general populations 22 1.5.4. AR CAG repeats and the penetrance of BRCA1 mutations 24 CHAPTER 2. PROSTATE-SPECIFIC ANTIGEN AND BREAST CANCER (REVIEW ) 34 2.1. Prostate-specific antigen (PSA) production in females 34 2.2. PSA as a potential marker for breast cancer prognosis 35 2.3. Possible mechanisms by which PSA influences the development and progression of breast cancer 38 2.4. Hormonal regulation of PSA production in females 41 2.5. Polymorphisms in the PSA gene 43 iv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3. A CASE-CONTROL STUDY OF ANDROGEN RECEPTOR AND PROSTATE-SPECIFIC ANTIGEN GENE POLYMORPHISMS AND BREAST CANCER IN AFRICAN- AMERICAN WOMEN 45 3.1. Introduction 45 3.2. Study design 48 3.2.1. Study population 48 3.2.2. Data and biospecimen collection 48 3.2.3. Variable definitions 49 3.2.4. Laboratory methods 50 3.2.5. Statistical analysis 52 3.3. Results 55 3.4. Discussion 72 3.5. Conclusion 83 CHAPTER 4. GENETIC AND HORMONAL DETERMINANTS OF SERUM PSA LEVELS 85 4.1. Background 85 4.2. Study design and methods 94 4.2.1. Study subjects 94 4.2.2. Serum PSA and hormone levels 95 4.2.3. Resequencing of the ECR 97 4.2.4. Genotyping methods 97 4.2.5. Statistical analysis 100 4.3. Results 103 4.3.1. Characteristic of the study subjects 103 4.3.2. Racial differences in serum PSA, androgen and SHBG levels 105 4.3.3. Serum PSA levels and self-reported history of BPH or prostatitis 106 4.3.4. Age-related change in serum PSA levels 107 4.3.5. Serum PSA levels and serum androgen and SHBG levels 107 4.3.6. Serum PSA levels and the PSA gene polymorphisms 110 4.3.7. Serum PSA levels and the AR CAG polymorphism 126 4.3.8. Effect modifications in the associations between serum PSA levels and the PSA gene and AR gene Polymorphisms 13 0 4.4. Discussion 138 4.5. Conclusion 149 CHAPTER 5. SUMMARY AND FUTURE STUDY PLANS 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REFERENCES APPENDICES Appendix I. GRANT PROPOSAL: ANDROGEN RECEPTOR GENE, PSA GENE IN BREAST CANCER RISK (FUNDED BY THE CBCRP DISSERTATION AWARD) Appendix II. MANUSCRIPT: ANDROGEN RECEPTOR AND PROSTATE-SPECIFIC ANTIGEN GENE POLYMORPHISMS AND BREAST CANCER IN AFRICAN-AMERICAN WOMEN (IN PRESS) Appendix III. GRANT PROPOSAL: ANDROGEN RECEPTOR GENE, P21 GENE IN BREAST CANCER RISK (FUNDED BY THE CBCRP POSTDOCTORAL AWARD) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES Table 1.1. Summary of selected studies on circulating testosterone and breast cancer 17 Table 1.2. Summary of studies on AR CAG repeats and breast cancer risk 31 Table 1.3. Studies on AR CAG repeats and hereditary breast cancer 33 Table 3.1. Characteristics of the study participants, by case-control status 57 Table 3.2. AR CAG genotype frequencies in cases and controls 58 Table 3.3. The comparison of the AR allele lengths between cases and controls 59 Table 3.4. AR CAG polymorphism and breast cancer risk among all subjects 61 Table 3.5. AR CAG polymorphism and breast cancer risk, stratified by age 64 Table 3.6. AR CAG polymorphism and breast cancer risk, stratified by menopausal status 64 Table 3.7. AR CAG polymorphism and breast cancer risk among postmenopausal women, stratified by HRT use 65 Table 3.8. AR CAG polymorphism and breast cancer risk, stratified by BMI 65 Table 3.9. AR CAG polymorphism and breast cancer risk among postmenopausal women, stratified by BMI 66 Table 3.10. AR CAG polymorphism and breast cancer risk, stratified by 1 S t-degress family history of breast cancer 66 Table 3.11. AR CAG polymorphism and breast cancer risk, by tumor stage, ER status, or PR status 68 Table 3.12. The PSA gene ARE-I -158G/A polymorphism and breast cancer risk 69 Table 3.13. The PSA gene ARE-I -158G/A polymorphism and breast cancer risk, by tumor stage, ER or PR status 71 vii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.14. Gene-gene interaction between the AR CAG polymorphism and the PSA gene ARE-I -158G/A polymorphism in breast cancer 72 Table 3.15. Linear effect of per CAG repeat increment in breast cancer risk by the PSA gene ARE-I -158G/A polymorphism 72 Table 4.1. Basic characteristic, serum androgen and PSA levels, by race and study center 104 Table 4.2. Race-specific allele frequencies of the PSA gene polymorphisms, by study center 105 Table 4.3. General characteristics of the study subjects 105 Table 4.4. Serum PSA, androgen and SHBG levels by race 106 Table 4.5. Serum PSA levels by self-reported BPH and prostatitis history 107 Table 4.6. Age-related changes in serum PSA levels 108 Table 4.7. Serum PSA levels and serum androgen, SHBG levels 109 Table 4.8. Allele frequencies of the common polymorphisms in the PSA gene 111 Table 4.9. Allele frequencies of some rare polymorphisms in the PSA gene 111 Table 4.10. Serum PSA levels and common polymorphisms in the PSA gene among whites 112 Table 4.11. Serum PSA levels and common polymorphisms in the PSA gene among African-Americans 113 Table 4.12. Pairwise LD (Levwontin’s D ’ and R2) among whites 115 Table 4.13. Pairwise LD (Levwontin’s D ’ and R2) among African- Americans 115 Table 4.14. Serum PSA levels and genotype combinations o f the 3 functional SNPs among whites 125 Table 4.15. Serum PSA and the haplotypes o f SNPs -5307 G/A and -5217 T/A among African-Americans 125 viii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.16. Race-specific distributions of the AR CAG repeat lengths 128 Table 4.17. Serum PSA levels and the AR CAG polymorphism 129 Table 4.18. Serum PSA levels and the PSA gene polymorphisms (-5413 T/C and -3647 T/A) among whites, by age 132 Table 4.19. Serum PSA levels and the PSA gene polymorphisms (-5307 G/A and -5217 T/A) among African-Americans, by age 132 Table 4.20. Serum PSA levels and the AR CAG polymorphism, by age 133 Table 4.21. Serum PSA levels and the PSA gene polymorphisms (-5413 T/C and -3647 T/A) among whites, by BPH 133 Table 4.22. Serum PSA levels and the PSA gene polymorphisms (-5307 G/A and -5217 T/A) among African-Americans, by BPH 134 Table 4.23. Serum PSA levels and the AR CAG polymorphisms, by BPH 134 Table 4.24. Serum PSA levels and the PSA gene polymorphisms (-5413 T/C and -3647 T/A) among whites, by 3a-diol G 135 Table 4.25. Serum PSA levels and the PSA gene polymorphisms (-5307 G/A and -5217 T/A) among African-Americans, by 3a-diol G 135 Table 4.26. Serum PSA levels and the AR CAG polymorphism, by 3a-diol G 136 Table 4.27. Serum PSA levels and the PSA gene polymorphisms (-5413 T/C an d -3647 T/A) among whites, by SHBG 136 Table 4.28. Serum PSA levels and the PSA gene polymorphisms (-5307 G/A and -5217 T/A) among African-Americans, by SHBG 137 Table 4.29. Serum PSA levels and the AR CAG polymorphism, by SHBG 137 Table 4.30. Serum PSA levels and the AR CAG polymorphism, by PSA genotypes 138 Table 4.31. Serum PSA levels and the PSA polymorphisms, by the AR CAG Genotypes 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES Figure 3.1. The distribution of the CAG repeat lengths in cases and controls 59 Figure 3.2. The cumulative distribution curves of the smaller and the larger AR CAG alleles in all cases and controls 60 Figure 3.3. The cumulative distribution curves of the smaller and the larger AR CAG alleles in cases and controls without a positive l st-family history of breast cancer 67 Figure 3.4. The cumulative distribution curves of the smaller and the larger AR CAG allele in cases and controls with a positive 1st-family history o f breast cancer 67 Figure 4.1. AREs in the distal enhancer and proximal promoter regions of the PSA gene 87 Figure 4.2. Common polymorphisms in the PSA gene enhancer and promoter regions 90 Figure 4.3. The distribution o f the AR CAG repeat lengths by race 127 Figure 4.4. Lowess smoothing average plot of serum PSA levels vs. AR CAG repeat lengths 129 x Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT The roles of androgens in breast cancer are unclear. Looking at genetic variations in the androgen receptor gene which alter the receptor function and thus affect androgen activities provides an alternative way to investigate the effects of androgens in breast carcinogenesis. In addition, studying target genes of the androgen signaling pathway may also help to understand how androgens exert their effects. Prostate-specific antigen (PSA) is a serine protease that is directly induced by androgens. Previous studies have suggested that PSA in the breast tissue was a favorable prognostic marker of breast cancer progression. In a case-control study of 501 African-American women, we examined a CAG repeat polymorphism in exon 1 of the AR. gene, which encodes a polyglutamine tract in the transactivation domain of the protein whose length is inversely correlated to AR activity. We also looked at a polymorphism (-158 G/A) at the androgen response element I (ARE-I) of the PSA gene, which had been associated with PSA levels in serum or in breast tissue in some studies. Overall, there was no significant association between the CAG repeat polymorphism and breast cancer risk. However, among women with a first-degree family history of breast cancer, shorter CAG repeats were associated with a significantly reduced breast cancer risk. We did not find significant association between the ARE-1 polymorphism and breast cancer risk. This lack of association could be due to the fact that the ARE-I polymorphism is not functional. xi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. We then conducted a genotype-phenotype correlation study to examine other common polymorphisms in the enhancer/promoter region of the PSA gene. Among 556 male controls from two case-control studies of advanced prostate cancer, we observed significant association between serum PSA levels and several of these polymorphisms including three which have been previously shown to affect the gene promoter activities in vitro, and a new polymorphism we identified in an evolutionarily conserved region in the 5’ regulatory region the PSA gene. Furthermore, serum levels of an androgen metabolite, 3a-androstanediol glucuronide, and sex hormone binding-globulin were also significant predictors of serum PSA levels. There was some evidence support a hormone-gene interaction. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. INTRODUCTION Breast cancer is a hormone-related cancer. Exposure to endogenous and exogenous hormones is known to influence breast cancer risk. Certain menstrual and reproductive factors leading to a greater estrogen and/or progesterone exposure (such as early menarche, late menopause, postmenopausal obesity, and hormone replacement therapy) are among the well-established risk factors for breast cancer. The role of progesterone in breast carcinogenesis is also strongly supported by recently published results form the Women’s Health Initiative randomized trial, in which adding progesterone to menopausal estrogen therapy was found to substantially increase the risk of breast cancer (Chlebowski et al., 2003). On the other hand, the role of androgen, another important hormone in the female body, is much less clear. With recent rapid-growing interest in the possible usage of androgen in postmenopausal hormone replacement therapy, this topic has gained much attention from the scientific community (Enserink, 2005). Originally androgens, because they are precursors of estrogens, were postulated as risk factors for breast cancer. This hypothesis is supported by epidemiological studies comparing circulating androgens in patients and healthy controls, especially among postmenopausal women. However, accumulating evidence from experimental studies suggest that androgens may also act directly (through binding to androgen receptor) to protect against breast cancer. One reason for the paradox may be that the effect of androgens independent of estrogens can not be easily assessed in epidemiological 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. studies. An alternative approach to studying the effects of androgens on breast cancer, which we utilized in this work, is to examine androgen receptor (AR) gene variants that alter the receptor function. This approach can provide a way to specifically examine the estrogen-independent effects of androgens. In the first exon of the AR gene there is a CAG repeat polymorphism which encodes a variable-length polyglutamine tract in the amino-terminal domain (the transactivation domain) of the protein. Several in vitro studies have shown that the polyglutamine tract length is inversely associated with the transactivation activity of the AR. Therefore, I have examined a possible association between this CAG repeat polymorphism and breast cancer risk in a population-based case-control study. To explore possible mechanisms of androgen action in breast cancer development and progression, I have also studied a down-stream gene of the androgen signaling pathway, the PSA gene, which encodes the prostate-specific antigen (PSA) protein. PSA protein has been implicated as a favorable prognostic factor in breast cancer (Yu et al., 1995b; Yu et al., 1998). A single nucleotide polymorphism in the androgen responsive element-I in the PSA gene promoter has been associated with PSA levels and with prostate cancer risk. My study hypothesis was that this polymorphism in the PSA gene might be associated with breast cancer risk by affecting hormone-regulated PSA production. Chapter 1 provides a review o f present studies (experimental, clinical and epidemiological) on the significance of androgens, the androgen receptor, and the CAG polymorphism in the development of breast cancer, with special focus on this 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CAG repeat and breast cancer risk in epidemiological studies. Chapter 2 summarizes the studies on the significance of prostate-specific antigen in the development/progression of breast cancer. The rational for studying a single nucleotide polymorphism (-158G/A) in the PSA gene and breast cancer is also presented. Based on evidence provided in the literature, a molecular epidemiological study on these two above-mentioned polymorphisms in the AR gene and the PSA gene was carried out (supported by a California Breast Cancer Research Program Dissertation Grant, the grant proposal and reviewers’ comments are attached in Appendix I). The study design, the results and the interpretation of the study findings are presented in Chapter 3. The manuscript (attached in Appendix II) has been accepted by Cancer Epidemiology, Biomarkers and Prevention and is currently in press. Chapter 4 provides the background of a study on the hormonal and genetic determinants of serum PSA level. The study methods, results, and discussion of this genotype-phenotype study are also included. Based on the interesting results of our case-control study o f AR and PSA gene polymorphisms in breast cancer, I submitted a grant proposal to re-examine our finding in a larger population. In addition, I proposed to look at another important gene downstream of the androgen receptor, the p21 gene (also known as the cyclin- dependent kinase inhibitor 1 A, CDKN1A). This proposal (attached in Appendix III) 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. has been funded by California Breast Cancer Research Program and will support my postdoctoral training. 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 1. ANDROGENS, ANDROGEN RECEPTOR AND THE CAG REPEATS IN BREAST CANCER (REVIEW) 1.1. Androgen production in females Androgens are a group of 19-carbon steroids including dehydroepiandrosterone (DHEA), dehydroepiandrosterone-sulfate (DHEAS), androstenedione (A4-dione), testosterone (T) and dihydrotestosterone (DHT). DHEA, DHEAS, and A4-dione have weak androgenic activities and are generally considered androgen precursors. T and DHT are the active androgens. In females, there are two sources o f androgens: the adrenal glands and the ovaries. The adrenal glands contribute to the production and secretion of the majority of DHEA and DHEAS, about 50% of A4-dione and 25% of testosterone. The other 50% of A4-dione and 25% of testosterone (varying during normal menstrual cycle) are produced and secreted by the ovaries. The remaining 50% o f testosterone is from the conversion of androstenedione in peripheral target tissues of androgens including the breast (Longcope, 1986). In postmenopausal women, the ovarian contribution of the circulating A4-dione decreases to about 20%. However, the decrease in testosterone production in the ovaries is less marked and the equal contribution of the ovaries and adrenal glands to circulating T seems to remain unchanged (Labrie et al., 2003). Dihydrotestosterone (DHT) is the most potent form of androgens in the body (i.e. with the highest affinity and specificity to the androgen receptor) and is primarily 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. produced in the peripheral tissues from the conversion of A4-dione, testosterone, and the other precursors (Toscano, 1986). The secretion of androgens in the adrenal glands is stimulated by adrenocorticotropic hormone (ACTH), while luteinizing hormone (LH) is involved in controlling androgen synthesis and release in the ovaries. In premenopausal women, there is a notable variation in serum A4-dione and T levels during the menstrual cycle, i.e. the levels of A4-dione and T are the highest in the mid-cycle around the LH surge (Judd and Yen, 1973; Massafra et al., 1999). The production of androgens declines with aging. The decline occurs early in the life span (rather than during or after menopause). In a study of 33 healthy women, it was found that the plasma testosterone levels in women in their 40s were only about half those of women in their 20s (Zumoff et al., 1995). Studies found either no significant change or a small decline in the circulating testosterone level during the menopause transition (reviewed in (Lillie et al., 2003)). 1.2. Experimental and clinical studies on androgens and androgen receptor in breast cancer Evidence from most experimental studies suggests that androgens might act as a growth inhibitor of breast cancer. For example, many studies have reported that testosterone and/or DHT inhibit the growth of human breast cancer cell lines MCF-7 (Ando et al., 2002; Ortmann et al., 2002), ZR-75-1 (Birrell et al., 1995a; Hackenberg and Schulz, 1996; Lapointe et al., 1999; Poulin et al., 1988; Poulin et al., 1989), T47D (Birrell et al., 1995a; Ortmann et al., 2002), CAMA-1 (Lapointe and Labrie, 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2001) and MFM-223 (Hackenberg and Schulz, 1996). Only in a few other studies a growth-stimulating effect of androgens was observed on cell lines MCF-7, MDA- MB-453 and EFM-19 (Birrell et al., 1995a; Boccuzzi et al., 1994; Hackenberg and Schulz, 1996). The exact reasons for these divergent proliferative responses are not known. Possible explanations may include clonal variation in the cell lines (which may relate to AR content), differences in media conditions, plating densities and cellular metabolism of the androgens (Ando et al., 2002). The doses of androgens used in the experiments might be another explanation. Biphasic effects of DHT on the growth of the MCF-7 cell line (inhibitory at low concentrations and stimulatory at very high concentrations) have been observed (Boccuzzi et al., 1994). High doses (1 uM) of DHT induce a growth stimulation in the human breast cancer cell lines MFC-7 and EFM-19 (Hackenberg et al., 1988). However, in another study DHT remained inhibitory to cell line MFC-7 at similar concentrations (Ando et al., 2002). In animal studies, androgens have also been repeatedly observed to inhibit breast tumor development and growth. Testosterone inhibits estrogen-induced mammary epithelial proliferation in ovariectomized female rhesus monkeys (Zhou et al., 2000); DHT represses estrogen-stimulated tumor growth in a dimethylbenz(a)anthracene (DMBA)-induced mammary tumor model in ovariectomized rats (Dauvois et al., 1989); DHT also exhibits similar growth- suppressing effects on ZR-75-1 xenograft in athymic mice (Dauvois et al., 1991). The binding to androgen receptor (AR) is necessary for androgenic effects on breast cancer cell proliferation. The cell lines that respond to androgens are all AR- 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. positive cell lines. Blocking androgen receptor with its antagonist resulted in the abolishment of the effect both in vitro (Ando et al., 2002; Birrell et al., 1995a; Lapointe et al., 1999; Lapointe and Labrie, 2001; Poulin et al., 1988; Poulin et al., 1989) and in vivo (Dauvois et al., 1989). The importance of AR in mediating the growth-inhibiting effect of endogenous androgens is also supported by the observation that treatment with flutamide, an androgen receptor blocker, results in an enhanced mammary epithelial proliferation in normally cycling monkeys (Dimitrakakis et al., 2003). In humans, AR expression has been detected in some normal mammary tissues (Ruizeveld de Winter et al., 1991; Wilson and McPhaul, 1996). We are not yet clear about the AR-mediated physiological roles of androgens in normal human mammary gland development, although AR knock-out has been observed to result in retarded mammary gland development in mice (Yeh et al., 2003). AR expression is detected in about 75% to 90% of breast tumor tissues (Kuenen-Boumeester et al., 1992; Lea et al., 1989; Soreide et al., 1992). Some (Birrell et al., 1995b; Bryan et al., 1984; Langer et al., 1990) but not all (Soreide et al., 1992) studies have suggested that the expression of AR in breast tumor, in addition to ER and PR, is a marker for good response to hormone therapy and survival. Although side effects preclude their wide use, androgens have been shown to be effective in treating women with advanced breast cancer. A higher response rate, a longer time to disease progression and better survival have been reported when fluoxymesterone (a synthetic 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. testosterone derivative) is combined with anti-estrogen tamoxifen compared to treatment with tamoxifen alone (Ingle et al., 1991; Tormey et al., 1983). 1.3. Possible mechanisms implicated in the growth regulatory effects of androgens in breast cancer It has been suggested that at least part of the growth inhibitory effect of androgens may be attributed to androgen-induced down-regulation of the estrogen receptor, because DHT suppresses ER expression and opposes estrogen-induced proliferation in ZR-75-1 breast cancer cells(Labrie et al., 1990; Poulin et al., 1989). There are also studies suggesting a possible functional crosstalk between AR and ER signaling transduction (Lanzino et al., 2005). The over-expression of AR markedly decreases estradiol signaling in a reporter gene assay (with an estrogen-responsive- element-luciferase reporter plasmid), which appears to be further inhibited by simultaneous exposure to lOOnM o f DHT but reversed by addition of androgen receptor blocker hydroxyflutamide. The inhibitory effects were not observed when mutant AR was expressed (Ando et al., 2002). On the other hand, it has been observed that the inhibitory effect of androgens on breast cancer cell growth is additive to that induced by antiestrogen and could occur in the absence o f estrogenic stimulation, implying that androgens may exert a direct inhibitory effect on breast cancer cell proliferation that could be independent o f estrogen receptor levels (Poulin et al., 1988). Other lines o f evidence suggest that androgen-induced growth inhibition is through modulating apoptosis. Bcl-2 is a protein encoded by the proto-oncogene bcl- 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 that inhibits programmed cell death by preventing the activation of several proteins participating in cell death regulation (reviewed in (Kroemer, 1997)). In the breast cancer cell line ZR-75-1 which is growth inhibited by DHT, it has been observed that DHT at physiological concentrations (InM) reduces the mRNA and protein levels of Bcl-2 by up to 50% and this effect could be reversed by hydroxyflutamide (Lapointe et al., 1999). Similar results were reported in a study by Kandouz et al, in which DHT was shown to be pro-apoptotic with maximal effects at 10 nM concentration in the AR-positive human breast cancer cell lines, T47D and ZR-75-1 (Kandouz et al., 1999). This modulation of apoptosis by down-regulating Bcl-2 via an AR-mediated pathway provides a potential explanation for androgen-induced growth inhibition in breast cancer. Several studies have also reported an androgen-induced prolonged duration of the cell cycle (de Launoit et al., 1991) and increased proportion of cells at G1 phase (Greeve et al., 2004; Szelei et al., 1997), leading to the postulation that androgen may specifically regulate the expression of one or more genes involved in cell cycle control. Progression from G1 to S phase of the cell cycle involves a complicated network of cyclins, cyclin-dependent kinases (CDKs), CDK inhibitors, tumor suppressor retinoblastoma protein (Rb), and transcription factor E2F. The simplified model involves the sequential activation of CDKs (reviewed in (Massague, 2004; Sherr, 1994)). There are two classes o f CDKs involved in the G1 to S transition, CDK4/CDK 6 and CDK2, whose activities require their association with specific cyclin partners (cyclin D for CDK4/6 and cyclin E or cyclin A for 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CDK2). The inhibition of CDK activity by cyclin-dependent kinase inhibitors (CDKIs) constitutes one of the key mechanisms of the negative regulation of cell cycle progression. There are two families of CDKIs, namely the INK4 (inhibitors of CDK4 and CDK6) family and the CEP/KIP (cyclin-dependent kinase-interacting proteins/kinase inhibitor protein) family (reviewed in (Sherr and Roberts, 1995)). p2 1 w a fi/c ip i a n ( j p 2 7 k >p b e l o n g to the CIP/KIP family of CDKIs and have been speculated to mediate androgen-induced growth inhibition. Changes in these CDKIs are often seen in breast tumors. p27 is frequently under-expressed in breast carcinomas and such under-expression seems to be associated with poor prognosis and a more aggressive phenotype (Catzavelos et al., 1997; Chu et al., 1999; Tan et al., 1997). In breast cancer cell lines, DHT treatment can lead to an increase in the amount of p27K ipl protein, a decrease in cyclin E-CDK2 kinase activity, together with an increase in the proportion of cells in the G1 phase of the cell cycle, indicating that the anti-proliferative effect o f DHT may be associated with altered expression and activity o f p27 (Lapointe and Labrie, 2001). Although the detailed mechanism of androgen-induced p27 up-regulation is not clear, post-transcriptional regulation has been suggested. p21 is a promising candidate gene for mediating androgen-regulated cell cycle control. p21 is thought to be an important regulator of breast cancer cell proliferation and is responsible for the inhibition of MCF-7 cell proliferation by anti estrogens (Skildum et al., 2002). It is also one of the very few well-characterized AR-regulated genes. Sequence analysis has demonstrated a putative consensus 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. androgen responsive element (ARE) at position -200 within the promoter of the p21 gene (AGCACGCGAGGTTCC) (Lu et al., 1999). p21 gene transcription in MCF-7 cells was enhanced upon androgen receptor activation and reversed by hydroxyflutamide (Yeh et al., 2000), confirming the direct transcriptional regulation. However, studies on the role of p21 in breast cancer have been inconclusive. First, p21 down-regulation does not always correlate with increased cell proliferation. Secondly, both high and low levels of this protein have been correlated with prolonged disease-free survival for breast cancer patients (reviewed in (Fernandez et al., 1998)). This is probably due to the fact the p21 has a wide spectrum of CDK substrates as well as multiple roles in cell cycle control, DNA repair, and antiapoptosis (summarized in (Lu et al., 1999)). Thus, the role of p21 in mediating the growth regulation by androgens in breast cancer is still unclear at present and warrants further study (see details in the attached grant proposal “Androgen receptor gene, p21 gene in breast cancer”). Another interesting down-stream gene of the AR signaling is the PSA gene, which encodes the prostate-specific antigen (PSA). Studies which suggested that PSA pathway is also a possible mechanism involved in androgen and breast cancer. A detailed review o f PSA in breast cancer is provided in chapter 2. 1.4. Epidemiological studies on circulating androgen levels and breast cancer risk In contrast to experimental and clinical studies, epidemiological studies examining circulating levels of active androgens generally agree that elevated levels 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of testosterone are associated with higher risk of breast cancer (reviewed in Secreto et al, 1994; in Zumoff et al, 1994 and in Lillie et al, 2003). The findings are more consistent in studies among postmenopausal women (reviewed in (Lillie et al., 2003; Somboonpom and Davis, 2004). In fact, for premenopausal women, the positive association between circulating androgens and breast cancer risk was found in some case-control studies (Secreto et al., 1984a; Secreto et al., 1984b) (Secreto et al., 1989), but only in one out of three cohort studies (Kaaks et al., 2005; Thomas et al., 1997b; Wysowski et al., 1987). There are some potential limitations in these epidemiological studies of circulating hormones: 1) A potential limitation of case-control studies is that the temporal relationship between elevated hormone levels and the risk of breast cancer cannot be established. Even in cohort studies, if the hormone levels are not measured well before cancer occurs, it is also possible that elevated levels of hormones in the cases may be a result of an increased production by the tumor. However, a prospective study examining changes in circulating hormone levels over time did not seem to suggest that the presence of a growing tumor has a major effect on circulating androgen levels(Zeleniuch-Jacquotte et al., 2004). If this is proven to be true, the temporal issue probably would not be a cause o f a biased result. 2) A single measurement of circulating hormone was used in these studies, which may not reflect accurately an individual’s long-term average hormone 13 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. level. Serum testosterone levels vary due to diurnal variation(Vierhapper et al., 1997), diet, stress, and physical activity (reviewed in (Zitzmann and Nieschlag, 2001). The intra-class correlation coefficient (assessment of within-person variability over time) for testosterone was estimated to be 0.64 or 0.87 in two studies of postmenopausal women (Hankinson et al., 1998b; Zeleniuch-Jacquotte et al., 2004). After correcting for this variability, the relative risks associated with elevated androgen levels were strengthened considerably (Hankinson et al., 1998b), suggesting that measurement inaccuracy due to a single measurement may be non-differential in cases and controls and thus may cause a bias toward the null. This bias does not appear to be an explanation for the inconsistency between the experimental and epidemiological studies. 3) Because of the inherent close relationship between androgens and estrogens, it is difficult to assess the independent effects of androgens (see below). 4) Very importantly, the measurement of circulating androgen levels does not reflect the real androgen exposure in breast tissue. Breast tissue is able to produce androgens locally (Labrie et al., 2003). 5a-reductase, the enzyme responsible for the local production o f DHT from testosterone, is associated with significantly low cell proliferation rate, tumor size, and histological grade of breast cancer (Suzuki et al., 2001), actually indicating an inhibitory role for androgen in tumor growth. 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5) Another important factor, the individual difference in the efficiency of androgen signaling (androgen receptor function) was not taken into account in these epidemiological studies. The positive association between androgens (especially testosterone) and breast cancer in epidemiological studies could be explained, at least in part, by their close relationship with estrogens, which have been shown to be breast cancer promoting factors in many studies. Androgens can be converted into estrogens by aromatase. In fact, the conversion from androstenedione to estrone is the main source of estrogen production in postmenopausal women. Androgens also compete with estrogens for sex hormone-binding globulin (SHBG) with higher binding affinity (Dunn et al., 1981), resulting in an increased fraction of free estrogens and thus increased bioavailability of estrogens. Results from four prospective studies on postmenopausal women appear to support the hypothesis that androgens act indirectly in breast cancer as substrates of estrogen production (Table 1.1). For example, in New York University Women’s Health Study Cohort, a significantly increased risk associated with an elevated serum level of testosterone was found when testosterone was analyzed alone. However, when serum level of total estradiol and the percentage of estradiol bound to SHBG were adjusted for in the analysis, the risk estimate for testosterone was substantially attenuated and the association was no longer statistically significant (Zeleniuch- Jacquotte et al., 1997; Zeleniuch-Jacquotte et al., 2004). Similar results were obtained from the Guernsey cohort in the United Kingdom (Thomas et al., 1997a), 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the Nurses’ Health Study (Hankinson et al., 1998b), and the Study of Osteoporotic Fractures in the United States (Cauley et al., 1999). However, in two nested case- control studies conducted in Italy and Sweden, adjusting for serum estradiol level did not weaken the association between breast cancer risk and testosterone level (Berrino et al., 1996; Manjer et al., 2003). Dorgan et al also reported that the association between testosterone and breast cancer was only attenuated slightly by the adjustment for free estradiol level (Dorgan et al., 1999). In a pooled analysis of eight published prospective studies on circulating level of androgens and breast cancer risk, the risk estimate associated with elevated androgen levels was reduced slightly after adjusting for estradiol concentration and remained significant. The relative risk associated with a doubling of testosterone concentration was 1.42 (95% Cl: 1.25- 1.61) before the adjustment for estradiol and 1.32 (1.15 to 1.51) after the adjustment. The mixed results from these studies suggest that it is difficult to assess the independent role of androgens in epidemiological studies due to the high correlation between androgens and estrogens. More importantly, because substantial conversion of androgens to estrogens occurs in breast adipose tissue (Thijssen, 2004), adjustment for circulating estrogen may not adequately adjust for local estrogen levels in the breast. This may be especially true among postmenopausal women when aromatase activity in the breast is high because of the low circulating estrogen levels. This might partly explain the observation that the positive association between higher levels of androgens and increased risk is more consistently found 16 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. among postmenopausal women, as compared to premenopausal women (Somboonpom and Davis, 2004). Table 1.1. Summary of selected studies on circulating testosterone and breast cancer Study First author (year) Exposure (unit) Levels RRs (95% Cl) before adjusting for estradiol RRs (95% Cl) after adjusting for estradiol Guernsey Cohort, UK Thomas (1997) Total testosterone (nmol/L) <0.73 0.73-1.25 >1.25 1.00 1.83 (0.82-4.12) 2.39 (1.01-5.65) p for trend=0.045 1.00 1.10 (0.43-2 77) 0.82 (0.28-2.42) p for trend=0.657 NYU women’s Health Study, USA Zeleniuch- Jocquotte (1997) Total testosterone (nmol/L) <0.73 0.73-1.01 1.02-1.45 >1.45 1.00 2.4(1.0-5.6) 3.5(1.4-8.4) 2.7(1.1-6.8) p for trend <0.05 1.00 1.4 (0.6-3.5) 1.8 (0.7-5.0) 1.2 (0.4-3.5) p for trend NS^ Nurses’ Health Study, USA Hankinson (1998a) Total testosterone (ng/dL) <15 16-22 23-31 >31 1.00 1.12(0.6-2.10) 1.07 (0.57-2.00) 1.40(0.73-2.70) p for trend=0.04 1.00 NA NA 1.08 (0.52-2.25) Study of Osteoporotic Fractures, USA Cauley (1999) Free testosterone (pmol/L)* <5.54 5.54-8.31 8.32-13.16 >13.17 1.0 2.2 (0.6-7.5) 6.4 (2.1-19.6) 3.3(1,1-10.3) p for trend=0.009 1.0 RR=2.1 (0.9-4.7) Study of hormones and diet in the etiology of breast tumors, Italy Berrino (1996) Free testosterone (pg/mL)* <0.57 0.57-0.86 >0.86 1.0 1.8 (0.4-9.3) 5.7(1.5-22.2) p for trend=0.005 1.0 NA 5.9(1.2-29.3) Breast Cancer Serum Bank cohort, USA Dorgan (1999) Total testosterone (ng/dL) <10 11-17 18-26 >27 1.0 2.9 (0.9-9.4) 2.9(1.0-8.6) 6.2(2.0-19.0) p for trend=0.002 1.0 2.0 (0.6-7.6) 2.1 (0.6-7.3) 4.6(1.3-16.6) p for trend=0.01 Malmo Diet and Cancer study, Northern Sweden Health and Disease Study, Sweden Manjer (2003) Total testosterone Highest vs. lowest quartile 1.9(1.1-3.3) 1.9 (1.1-3.3) Nested case- control NYU women’s Health Study Zeleniuch- Jacquotte (2004) Total testosterone a doubling in total T levels 1.2(1.1-1.4) 1.1 (0.9-1.3) (adjusting for estrone) similar results were obtained when total testosterone level was analyzed. T'IS: not significant 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.5. The androgen receptor gene and the CAG repeat polymorphism Androgens exert their functions through binding to the androgen receptor. Androgen receptor (AR) is a ligand-dependent nuclear transcription factor. AR remains in an inactive state (in a complex with heat shock proteins) in the cytoplasm of target cells until it binds to its ligand, T or DHT. The ligand binding induces the dissociation of AR from heat shock proteins. The ligand-receptor complex then undergoes homodimerization, translocates to the nucleus and binds to specific androgen responsive elements on the target genes, resulting in the regulation of their transcriptional activity (review in (Gnanapragasam et al., 2000)). The androgen receptor is a protein of approximately 919 amino acids, with a molecular size of about 110 kDa (reviewed in (Gnanapragasam et al., 2000)). It has three major domains: the carboxyl-terminal ligand-binding domain, the DNA- binding domain, and the amino-terminal domain (NTD) (Brinkmann et al., 1989a; Brinkmann et al., 1989b; Chang et al., 1988). The AR protein is encoded by the AR gene, which is located on the X chromosome (Xql 1-12) (Brown et al., 1989) and is comprised of eight exons. Exon 1 encodes the entire NTD of the protein, also known as the transactivation domain, which functions to regulate the expression level of the target genes, through its interaction with other cofactors (Faber et al., 1989). Exons 2 and 3 encode a DNA-binding domain, Exons 4 through 8 encode the ligand binding domain (Lubahn et al., 1989). There are two polymorphic trinucleotide repeats in Exon 1, namely a CAG repeat and a GGC repeat, encoding a polyglutamine and a polyglycine amino acid 18 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. stretches respectively. Previous studies of breast cancer have been mainly focused on the CAG repeat polymorphism. The normal physiological size range of polyglutamine tract is 7-37, as estimated in population-based studies (Edwards et al., 1992; Irvine et al., 1995; Xue et al., 2001). In vitro studies suggested that polyglutamine tract expansion interferes with normal transcriptional activation function of AR. A study conducted by Chamberlian et al revealed that there was a linear decrease of transactivation function of human AR with progressive expansion of the polyglutamine tract (from 25 to 35, 49 and 77), despite normal ligand-binding function (Chamberlain et al., 1994). This inverse and linear relation between the polyglutamine tract lengths (CAG repeat lengths) and AR transactivation activity was later confirmed by other studies (Irvine et al., 2000; Kazemi-Esfaijani et al., 1995; Tut et al., 1997). Completely deleting this polyglutamine tract resulted in 4- fold increased activity to activate an androgen-responsive reporter gene compared to the wild type AR (Callewaert et al., 2003). It has been postulated that the decrease in AR transactivation activity with increasing polyglutamine lengths may be attributable to the inhibition of the functional interaction between AR and its coactivators, p i 60 (Irvine et al., 2000). Shorter polyglutamine tract may also give AR a more accessible or stable surface for the interaction between its amino- and carboxyl- terminals, which is believed to be important to its function (Callewaert et al., 2003). The effect of polyglutamine tract lengths on AR transcriptional activity has been suggested to be cell specific by one study. In this study, the inverse correlation between polyglutamine tract lengths and transactivation activity of AR 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. was only observed in COS-1 (monkey kidney cell line) and LNCaP (prostate cancer cell line), but not in PC-3 (prostate cancer cell line) and MCF-7 (breast cancer cell line) (Beilin et al., 2000). However, the study by Irvine et al did confirm the negative association between polyglutamine length and AR transactivation activity in PC-3 cell line (Irvine et al., 2000). The available data are so sparse and conflicting that this cell line specificity is yet to be confirmed. The importance of the CAG repeat length on normal AR function is supported by evidence from human studies. The expansion of the CAG repeats to ~40 or more causes the X-linked spinal and bulbar muscular atrophy (SBMA) or Kennedy’s disease (La Spada et al., 1991). In a study among Singapore Chinese, moderate expansion of the CAG repeats was significantly associated with an increased risk o f defective spermatogenesis, an exquisitely androgen-dependent process (Tut et al., 1997). This was later confirmed in Australian (Dowsing et al., 1999), and U.S. populations (Mifsud et al., 2001b). 1.5.1. AR CAG repeats and prostate cancer A role of the AR gene in cancer predisposition is indicated by repeatedly observed associations between increased prostate cancer risk and shorter CAG repeat lengths (Giovannucci et al., 1997; Hsing et al., 2000; Ingles et al., 1997; Stanford et al., 1997; Xue et al., 2000). However, more recent studies tend to report null results (Beilin et al., 2001; Bratt et al., 1999; Correa-Cerro et al., 1999; Gsur et al., 2002; Latil et al., 2001; Miller et al., 2001). To explain these discordant findings, it has been suggested by Giovannucci et al that in recent studies, because of the widely- Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. used prostate-specific antigen (PSA) screening for prostate cancer, the case pool is changed by the inclusion of those with more indolent tumors (organ-confined, small volume) who may never have been detected if were not for the PSA screening. These cases may represent a different group etiologically, compared to those detected-by clinical manifestations. More specifically, the effect o f high androgenicity may be more prominent among a subgroup of prostate cancer patients (with younger age of onset, relatively aggressive phenotype), thus the association between the CAG repeat length and the risk of prostate cancer only manifests in this group (Giovannucci, 2002). A recent meta-analysis not only confirmed that men with shorter CAG repeats were at a significantly increased risk of prostate cancer (OR=1.19, 95% CI=1.07- 1.31, comparing CAG< vs.>21), but also corroborated the hypothesis of Giovannucci et al by showing that the association between CAG repeats and prostate cancer was stronger when controls were compared to cases with more aggressive disease or diagnosed at an early age (Zeegers et al., 2004). 1.5.2. AR CAG repeats and male breast cancer The involvement of the androgen receptor in breast cancer development in men was recognized in the early 1990’s, by the finding o f two rare germline AR mutations (which result in weakened DNA binding ability) in male breast cancer subjects (Wooster et al, 1992; Lobaccaro et al, 1993). Later studies on the CAG repeats and male breast cancer have consistently shown an increased risk of male breast cancer associated with longer CAG repeat lengths (weaker androgen receptor activity). A study o f 59 male breast cancer cases and 79 controls found that male 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. breast cancer patients tended to have alleles with longer than 28 CAG repeats, although the difference in the median CAG numbers between case group and control group did not reach statistical significance (Young et al., 2000). This was in agreement with another study of 29 cases and 30 controls in which alleles with 30 or more repeats were also found only in cases (Haraldsson et al., 1998). The small sample sizes could be an explanation for the lack of significant results in these two studies. In a recent study with larger sample size (40 cases and 456 controls), it was found that the frequency of longer CAG repeats ((CAG)n>24) was significantly higher in the cases than in the controls (MacLean et al., 2004). 1.5.3. Epidemiological studies on AR CAG repeats and female breast cancer in general populations Epidemiological studies on AR CAG repeat and female breast cancer have provided somewhat consistent data in that shorter CAG repeats (stronger AR transactivation activity) appear to be protective against breast cancer, at least in certain subgroups of the study population, although some studies have reported null results (Table 1.2). Population-based case-control studies of AR CAG and breast cancer were first published in 1999. The study of Spurdle et al of 652 Australians focused on premenopausal women (younger than 40) (Spurdle et al., 1999), while Dunning et al looked at 934 British women within a wider age range (25 to 75 years of age) (Dunning et al., 1999). No statistically significant association between CAG repeat size and breast cancer risk was observed in either study. Later, Giguere et al in a study o f 716 French-Canadians (using both hospital-based and population-based 22 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. controls) suggested a protective role of androgen in breast cancer by showing that shorter CAG (both alleles with (CAG)n < 20) repeats was associated with a significantly lower breast cancer risk (Giguere et al., 2001). In this study, menopausal status was found to be a significant effect modifier in this association. The decreased risk associated with the short-allele AR genotypes was present mostly among postmenopausal women, predominantly among women with surgical menopause. In a study by Ehaji et al of 470 Canadian women over age 40, longer CAG repeats (>26) were also found to be more common in breast cancer patients, compared to controls (Elhaji et al., 2001). However, it is noteworthy that in this study, genotyping of the cases was conducted using DNA from tumor tissue rather than blood, therefore, it is impossible to determine whether the difference in CAG repeat length between cases and controls was due to somatic changes. More recently, data from a case-control study nested within the Nurses’ Health Study Cohort, which was mostly comprised of postmenopausal Caucasian women, failed to show an association between AR genotype and breast cancer risk in the whole study population, before or after stratifying by menopausal status. However, the authors found that among women with a first degree family history of breast cancer, longer CAG repeat alleles (one or two alleles with (CAG)n > 22 compared to baseline both alleles <22) were associated with a significantly increased risk of breast cancer (Haiman et al., 2002). Two subsequent case-control studies, one population-based of 985 Americans (about 95% Caucasian) by Suter et al (Suter et al., 2003) and one hospital-based of 598 Philippinas by Leide et al (Liede et al., 2003) also reported a 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. significant association between longer CAG repeat alleles (sum of CAG repeats in two alleles > 43 vs. <43 in the former; and the average number of CAG repeats of two alleles > 25 vs. <25 in the latter) and increased breast cancer risk. This association did not differ by menopausal status (Liede et al., 2003) or by family history o f breast cancer (Suter et al., 2003). Contradicting the negative association between shorter CAG repeats and reduced risk of breast cancer, shorter CAG repeats seemed to be associated with more aggressive disease (a higher histological grade, positive lymph nodes and worse overall survival) (Yu et al., 2000). The association was no longer significant when other clinical and pathological variables (e.g. age, nodal status, tumor size, etc) were adjusted in the analysis, but the author argued that the reduction of sample size could be the reason for the loss of significance in the multivariate analysis (Yu et al., 2000). The study by Elhaji et al also found that shorter CAG repeats were more common in poorly differentiated tumors (Elhaji et al., 2001). These findings are in contrast to those suggesting shorter CAG repeats to be associated with lower breast cancer risk. One speculation may be that androgens might have different effects on breast tumor initiation and progression. 1.5.4. AR CAG repeats and the penetrance of BRCA1 mutations Studies on AR CAG repeat length and breast cancer risk in a special group of women (with inherited germline BRCA1 mutations) were summarized in Table 1.3. A study by Rebbeck et al on 304 American women carrying BRCA1 mutations (165 with breast cancer and 139 unaffected) was the first to suggest a protective role of 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. short CAG repeat length in breast cancer. In this study, BRCA1 mutation carriers were at significantly lower risk of developing breast cancer if they also carried an AR gene with shorter CAG repeats, compared to BRCA1 carriers with longer CAG repeats (Rebbeck et al., 1999). But the modifying effect of AR CAG repeats on BRCA1 penetrance was not confirmed by three subsequent studies (Kadouri et al., 2001; Menin et al., 2001; Spurdle et al., 2005). In the study conducted in Israel by Kadouri et al (most of the BRCA1 or BRCA2 carriers were of Ashkenazi origin), there was no significant association of the CAG repeats and breast cancer risk when comparing the CAG repeat numbers in 122 affected carrier with 66 unaffected carrier. The estimated relative risk o f breast cancer in carriers associated with more than 27 CAG repeats was 1.08. Restricting the analyses to only the Ashkenazi carriers or to only BRCA1 or BRCA2 carriers did not change the conclusion (Kadouri et al., 2001). In the study by Menin et al, among 101 breast cancer patients from breast/ovary cancer high-risk families, there was a marginally significant association between age of onset and the CAG repeat numbers o f the shorter allele, but not longer allele. However, when dichotomous variables were used (using the median CAG repeat numbers for short and long alleles as the cutoff points), no association between CAG repeat length and age o f disease onset was found. No association was found when analysis was limited to a sub-group of people with only hereditary breast cancer (including 11 cases with BRCA1 or BRCA2 mutations) (Menin et al., 2001). A very recent study o f 604 Australian and British BRCA1/2 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. carriers (including 376 BRCA1 carriers) also failed to detect association between the CAG repeat lengths and breast cancer risk (Spurdle et al., 2005). The modifying effect of AR CAG repeat length on BRCA1 mutation penetrance observed in Rebbeck’s study was indirectly supported by in vitro studies suggesting BRCA1 as an AR coactivator (in both prostate cell lines PC-3 and DU- 145, and breast epithelial cell line MCF-7) (Park et al., 2000; Yeh et al., 2000). In the study by Park et al, it was observed that BRCA1 interacted synergically with other AR coactivators (i.e. GRIP1, SRC-la and AIB1) to potentiate AR transactivation activity. Based on the finding that BRCA1 had direct physical contacts with AR and with the p i 60 coactivators from GST-pull down experiments, it was suggested that BRCA1 participated in the formation and/or stabilization of the androgen receptor coactivator complex, and thus increased the efficacy and accuracy o f AR-mediated transcriptional activity. Moreover, when the BRCA1 gene and AR gene encoding variable sizes of poly glutamine tract (poly-Q) were co-expressed in PC-3 cell line, the reduced transactivation activity association with increased poly-Q lengths was no longer observed (Park et al., 2000). This in vitro study was consistent with the finding in Rebbeck’s study and probably with the finding in Haiman’s study (because we expect to have certain proportion o f women with l st-degree family history to be BRCA1 carriers), in that women with less efficient ARs (encoded by longer CAG repeats) may be at higher risk of breast cancer development, only when they also carrying germline BRCA1 mutations. 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In summary, although not replicable in all studies (Kadouri et al., 2001; Spurdle et al., 2005), shorter CAG repeat lengths appear to be associated reduced breast cancer risk (Giguere et al., 2001; Haiman et al., 2002; Liede et al., 2003; Rebbeck et al., 1999; Suter et al., 2003), or at least in some specific groups of women, for example, postmenopausal women (Giguere et al., 2001); women with a family history of breast cancer (Haiman et al., 2002), or BRCA1 carriers (Rebbeck et al., 1999). Here are some noteworthy points in these studies: 1) It is difficult to model AR CAG repeats with regard to their association with breast cancer. AR is located on the X chromosome and one of the two X chromosomes in female is inactivated at random in early embryonic life. However, there is an increased frequency of non-random (skewed) X chromosome inactivation with aging, and maybe in women with certain pathophysiological conditions such as hirsutism (Vottero et al., 1999). Skewed X chromosome inactivation was also frequently seen in breast cancer patients (both in the peripheral blood cells and in breast tumor tissues) (Kristiansen et al., 2002). The non-random inactivation may cause some measurement error and add to some variations in exposure estimate in people classified as Long/Short heterozygotes. However, it is not obvious that this measurement error would account for the discordance between studies. 2) The big difference in subjects’ ages across studies may be a reason for the disparate results, since different age at diagnosis might suggest different etiological factors. For example, genetic factors such as mutations in breast cancer predisposing genes (BRCA1 and BRCA2) may play a more important role in early onset breast 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cancer, while environmental factors may be more important in cancer with later age of onset. More importantly, the hormonal setting in a female body changes with aging (especially before and after menopause), therefore it is possible that the impact of CAG repeat numbers on breast cancer risk may be more evident in certain age/menopausal status group. This is seen in the study by Liede et al, which found that the reduced risk associated with shorter CAG repeats was most prominent among women aged 50 years or older (Liede et al., 2003). But it has to be taken cautiously as this finding was based on an extremely small sample size. Menopausal status also appeared to be an important effect modifier in the study by Giguere et al (Giguere et al., 2001). However, two later studies (Haiman et al., 2002; Liede et al., 2003) showed that the effect of CAG repeats did not differ by menopausal status. 3) The study of Giguere et al and Liede et al were different from the other studies in that hospital controls were used. In the study by Giguere et al, shorter CAG repeats were found to be protective against breast cancer mainly among women having a surgical menopause (accounting for 42% of postmenopausal cases and 53% of postmenopausal controls in that study population). This result is difficult to explain biologically. It is not clear whether it is a biased result due to the use of hospital controls. It could also be due to co-morbidities that may be a cause or a result of changed endogenous hormonal environment or due to possible hormonal replacement therapy after surgical menopause. A study has suggested that hormone replacement therapy (especially estrogen plus progestin therapy) was an effect 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. modifier in the association between CAG repeats and mammographic density (Lillie et al., 2004). 4) Among the studies of BRCA mutation carriers, small sample size is a common problem. For example, even in the two studies with more than 300 subjects (Rebbeck et al., 1999; Spurdle et al., 2005), the use of cutoff points at the extreme end of the CAG repeat distribution resulted in small numbers of subjects in some groups. It might be possible that the studies with null results did not have enough power to detect the effect of CAG repeat length on breast cancer risk, if the effect was weak. Small sample size also made it impossible to evaluate this effect by specific BRCA1 or BRCA2 mutations (although Rebbeck et al did sub-group analyses and found the results remained the same within the unique BRCA1 mutation sub-group). Differences in racial distributions in these studies might also account for the inconsistency among studies. This is hard to assess because no race compositions of the study populations were clearly stated, except for the study conducted in Israel (Kadouri et al., 2001). In summary, although a protective role o f shorter AR CAG repeats in breast cancer has been suggested by several studies, at least in certain subgroups of women, more well-designed studies with large sample size are still needed in order to provide a clearer picture. In addition, if a protective role of a more active androgen receptor in breast cancer is proven to be true, studies of androgen-regulated genes will help to understand the mechanisms how this protective effect is exerted. However, our knowledge of down-stream genes in androgen-signaling pathway (especially in the 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. mammary cells) is very limited. Among the few well-characterized genes, PSA has been suggested as a promising candidate in the study of breast cancer by some recent studies. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.2. Summary of studies on AR CAG repeats and breast cancer risk Author (yr) Study subjects (CAG)n comparison Stratify by family history Stratify by menopausal status Finding Spurdle et al (1999) Australian; Caucasian; mostly premenopausal; 368 incident cases diagnosed before age 40; 284 controls frequency- matched to cases by age; population-based study Cutoff point=22 SS:OR=1.0 SL: OR=l.40 (0.92-2.15) LL: OR=l.40 (0.87-2.26) Also tested other cutoff points: 19,20,21,27,29 Without family history: SS:OR=1.0 SL+LL: OR=1.35 (0.84-2.15) With family history: SS:OR=1.0 SL+LL: OR=1.36 (0.92-2.15) Combined: SS:OR=1.0 SL+LL: OR=1.40 (0.94-2.09) NA No association Dunning et al (1999) Caucasians from the UK; both pre- and post menopausal; 234 incident cases and 155 controls in Series 1; 274 prevalent cases and 271 controls in Series 2. population- based study Cutoff point=23 SS*: OR=1.0 SL: OR=l.22 (0.92-1.61) LL: OR=0.76 (0.51-1.15) Also tested cutoff point: 29 NA NA No association Giguere et al (2001) Canadian; >95% French Canadians; Both pre- and post menopausal; 255 incident cases; 272 population- based controls and 189 hospital-based controls; matched to cases on age and residency area. Cutoff point=20 SS*:OR=1.0 SL+LL: OR=2.13 (1.22-3.70) NA Premenopausal: SS*: OR=1.0 SL+LL: OR=1.03 (0.43-2.5) Postmenopausal: SS*: OR=1.0 SL+LL: OR=3.22 (1.54-6.67) Short CAG associated with reduced risk Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.2. Summary of studies on AR CAG repeats and breast cancer risk (continued) Author (yr) Study subjects (CAG)n comparison Stratify by family history Stratify by menopausal status finding Haiman et al (2002) Nurses’ Health Study; 727 incident cases and 969 controls matched to cases on year of birth, menopausal status, HRT use, time of day, month, and fasting status at blood draw; population- based study Cutoff point=22 SS:OR=1.0 SL+ LL: OR=1.06 (0.83-1.35) Also tested other cutoff points: 23, 25, 27,29 vs.22; Without family history : SS:OR=1.0 SL+LL: OR=0.91 (0.71-1.17) With family history: SS:OR5 =1.01 (0.60-1.70) SL+LL5: OR=1.70 (1.20-2.40) P for interaction=0.04 Premenopausal: SS: OR=1.0 SL+LL5: OR=1.35 (0.64-2.84) Postmenopausal: SS: OR=1.0 SL+LL5: OR=0.98 (0.76-1.28) Shorter CAG repeats associated with reduced risk among women with a family history Suter et al (2003) Population-based; mostly Caucasain; mostly premenopausal; 524 incident cases (<45 years old); 461 controls frequence- matched to cases on age and reference year Cutoff point=22 SS: OR=1.0 SL: OR=1.3 (0.9-1.8) LL: OR=1.2 (0.8-1.7) Cumulative number o f repeats o f two alleles <42: OR=1.0 >43: OR=1.3 (1.0-1.7) Family history did not modify the effects o f genotypes on the risk NA Shorter CAG repeats associated with reduced risk; an increased risk associated with OC use was only found among women with SS genotype Liede et al (2003) Hospital-based; 299 incident cases and 229 control from Philippin Genera Hospital; both pre- and post menopausal Cutoff point= 25 (Average repeats numbers in two alleles) S: OR=l .0 L: OR=2.13 (1.06-4.17) Without family history: S: OR=1.0 L: OR=2.38 (1.12-5) With family history: NA due to small number Premenopausal: S: OR=1.0 L: OR=2.56 (1.10-5.88) Postmenopausal S: OR=1.0 L: OR=2.56 (0.57-12.5) Shorter CAG repeats associated with reduced risk. SS: both alleles have CAG repeat length below the cutoff point; SL: one allele has CAG repeat length above the cutoff point; LL: both alleles have CAG repeat lengths above the cutoff point; OR: odd ratio; numbers in parenthesis represent the 95% confidence interval; OC: oral contraceptive. * ORs were recalculated because in the original paper, LL genotype was the baseline group. 5 ORs were calculated for (CAG)n>23 vs. (CAG)n<22 * Compared to the baseline group with no family history and (CAG)n<22 U ) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.3. Studies on AR CAG repeats and hereditary breast cancer Author (yr) Study population Modeling o f CAG repeats Risk Age o f onset Rebbeck et al (1999) 304 American women with BRCA1 mutation; 165 affected (21-73 years o f age); 139 unaffected (19- 89 years o f age) One CAG increment (size o f short allele; long allele; or average size) At least one allele: >28 vs. both allele<28 >29 vs. both allele<29 >30 vs. both allele<30 N o significant results HR=1.81 (1.06-3.08) HR=2.66 (1.51-4.69) HR=4.45 (1.31-15.16) No significant results Diagnosed earlier by 0.8 years 1.8 years 6.3 years Kadouri et al (2001) BRCA1 or BRCA2 carriers M ostly Ashkenazi origin 122 affected; 66 unaffected (aged >56) One CAG increment At least one allele >28 vs. both alleles <28 RR=1.05 (0.97-1.17) RR=1.08 (0.45-2.61) NA Menin et al (2001) 101 Italian BC cases from high-risk families (hereditary or familiar cases) 11 with BRCA mutation Short allele: <19 >19 Longer allele: <23 >22 NA Median age at diagnosis 44 48 (p=0.07) 46 45 (p=0.8) Spurdle et al (2005) 376 BRCA1 and 219 BRCA2 carriers from 4 studies in Australia and Britain At least one allele >28 vs. both alleles <28 RR=0.74 (0.42-1.29) among BRCA1 carriers RR=1.12 (0.55-2.25) among BRCA2 carriers CHAPTER 2. PROSTATE-SPECIFIC ANTIGEN AND BREAST CANCER (REVIEW) 2.1. Prostate-specific antigen (PSA) production in females The KLK-3 (also named hK3, PSA) gene is the most extensively studied androgen-regulated gene. It encodes a 33kDa protein, prostate-specific antigen (PSA). PSA has been a widely-used tumor marker for prostate cancer screening. It has also been used to monitor the progression of prostate cancer after treatment, because of its apparently exclusive expression in the prostate. However, with the recent development of an ultrasensitive immunoflurometric PSA assay, it has been demonstrated that PSA expression is actually not prostate-specific. It is produced at a much lower level in many forms of female tissues and body fluids. PSA has been detected in normal breast tissue (Yu et al., 1996), in normal endometrial tissue (Clements and Mukhtar, 1994), in milk of lactating women (Yu and Diamandis, 1995c) and in amniotic fluid of pregnant women (Yu and Diamandis, 1995b). PSA levels in the sera of females are approximately 1000 times lower than that of males (Black et al., 2000). Depending on the sensitivity of PSA tests in different studies, PSA was detected in about 15% (Yu and Diamandis, 1995a) to 50% (Diamandis et al., 1996) of the serum samples from healthy women. In the study by Yu et al, 7 out o f 674 (>1%) healthy women had serum PSA levels of 0.1 ng/mL or higher (Yu and Diamandis, 1995a), which is in the normal range for men. 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2. PSA as a potential marker for breast cancer prognosis PSA is expressed in breast tumors and its expression appears to be associated with the clinical characteristics of the tumors. In a cohort of more than 1200 female breast cancer patients, 30% of them had detectable PSA (about 0.03ng/mg of total protein) in their tumor cytosolic extracts and about 5% of the tumor samples had PSA levels of 0.6 ng/mg protein (which was equivalent to 0.4 ng/mL) (Yu et al., 1994a). Higher PSA levels in breast tumor were associated with early clinical stages (Diamandis et al., 1994; Yu et al., 1996; Yu et al., 1995b). PSA expression in tumor tissues was also found be an independent predictor for survival in breast cancer patients. In a study by Yu et al of 174 female breast cancer patients (Italians) with primary breast cancer, the survival after surgery was followed up for about 3 years (ranged between 7 and 67 months). After adjusting for other significant breast cancer prognostic factors (i.e. age, clinical stage, tumor size, histological grade, nodal status, estrogen receptor and progesterone receptor status), patients with PSA- positive tumors still had a significantly reduced risk for relapse when compared with patients with PSA-negative tumors (Yu et al., 1995b). In a later study by the same group, which had a much larger sample size (more than 900 American patients) and a longer follow-up (an average follow-up time of 5 years), it was confirmed that PSA presence was associated with smaller tumor size and a 30%-40% reduction in risk of disease relapse or death. In this study, PSA immunoreactivity was also found to be inversely associated with markers o f active cell proliferation in breast tissue (Yu et al., 1998). Some other studies also suggest that PSA expression in tumor tissue is an 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. independent favorable indicator for survival, but only among subgroups of women (postmenopausal women with node-positive tumors in a study of 219 breast cancer patients in Greece (Griniatsos et al., 1998), or premenopausal women with ER- negative tumors in a study of 171 breast cancer patients in Finland (Alanen et al., 1999)). To explore the possible usefulness of PSA as a tumor marker for breast cancer, the same group of researchers studied PSA level in nipple aspirate fluid (NAF). PSA was detected in more than 90% of the NAF samples analyzed. In addition, PSA levels in the NAF obtained from women with breast cancer were significantly lower than those from women without cancer (Sauter et al., 1996; Sauter et al., 2004). However, other studies were not able to confirm the negative association between PSA expression and breast cancer incidence or disease aggressiveness. In a study by Black et al of 118 breast cancer patients and 99 healthy female blood donors in Italy in 2000, patients with breast cancer or benign breast disease had significantly higher serum PSA levels compare to disease-free female controls. Among breast cancer patients, elevated PSA levels were significantly associated with larger breast tumor size and with higher histological grade, while no association between serum PSA and disease relapse was found (Black et al., 2000). In the other study by Heyl et al of 100 breast cancer patients from Germany, PSA level in tumor tissue was not significantly correlated with any other clinical characteristic of the 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. tumors (such as stage, grade, lymph nodes involvement, ER or PR status) or survival time (Heyl et al., 1999). The reason for the inconsistent results is not obvious. There was a lack of detailed information about patient recruitment, their characteristics (e.g. menopausal status, exogenous hormone usage) or treatment patients received in the studies by Black et al (Black et al., 2000) and Heyl et al (Heyl et al., 1999), which made it difficult to compare studies. Potential confounding factors including age, receptor status, hormonal treatment were not addressed in survival analyses in the studies by Black et al (Black et al., 2000) and Heyl et al (Heyl et al., 1999). Therefore, results from these two studies need to be taken cautiously. In addition, some differences between these studies and the others are noteworthy. 1) In the study by Black et al, PSA level in serum (Black et al., 2000), not in breast tissue (Alanen et al., 1999; Griniatsos et al., 1998; Yu et al., 1995b; Yu et al., 1998)or in NAF (Sauter et al., 1996; Sauter et al., 2004), was studied as in other studies. Because the correlation between PSA levels in breast tissue and in serum has not been studied, it raises a question whether serum PSA level is a good indicator of local PSA production in the breast. A previous study of prostate cancer found that there was a lack of correlation between the immunohistochemical assessment of tissue PS A in prostate cancer and serum PSA level (Weir et al., 2000). The inconsistent finding by Black et al from the others is probably, at least in part, due to the fact the serum PSA does not reflect the PSA production in the breast. 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2) Different methods were used to determine the presence of PSA in breast tissue. In the study by Heyl et al (Heyl et al., 1999), qualitative immunohistochemical analysis was used on paraffin-embedded tissue, while in other studies (except for the study by Alanen et al) (Alanen et al., 1999), cytosolic extracts from frozen tumor specimens were tested quantitatively by immunofluorometric assay. Lack o f correlation between immunofluorometric and immunohistochemical detection of PSA has been reported (Howarth et al., 1997). This difference in PSA measurement could partly explain the reason for a lack of significant findings the study by Heyl (Heyl et al., 1999). 3) No details about the treatments patients received before or after surgery were reported by Heyl et al (Heyl et al., 1999) or by Black et al (Black et al., 2000). In the study by Yu et al (Yu et al., 1995b), patients with PSA-positive tumors appeared to respond better to adjuvant therapy including tamoxifen, although the difference was not statistically significant. This study result, together with the finding that tamoxifen induced PSA expression in vitro (Yu et al., 1994b), suggests that stratified analysis by adjuvant treatment should be performed. 2.3. Possible mechanisms by which PSA influences the development and progression of breast cancer Since very little is known about the physiological role o f PSA in breast tissue, it is difficult to explain the mechanism by which PSA confers a favorable prognostic outcome in breast cancer as seen in some studies. Some studies have 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. suggested that PSA may have some functional significance in cancer development and/or progression, in addition to its role as a marker of hormone (androgens, in particular) levels and activities (see below: hormone regulation of PSA production in females). Several lines o f evidence have led to the speculation that PSA might act as a pathogenic factor in breast cancer. 1) PSA is a protease for the major insulin-like growth factor binding protein, IGFBP3. The PSA-cleaved IGFBP3 has a lower affinity binding to insulin-like growth factors (IGFs) and thus helps in the release of free, bioactive IGFs (Cohen et al., 1994), which have been implicated in the carcinogenesis of the breast because o f their mitogenic and anti-apoptotic effects (Sachdev and Yee, 2001). High plasma level of IGF-1 has been associated with a significantly increased risk for the subsequent development of breast cancer (Hankinson et al., 1998a). IGFBP3 can also work in an IGF-independent fashion in modulating the growth of both normal and malignant mammary epithelial (Sachdev and Yee, 2001). There is also evidence supporting the role of PSA as an anticarcinogenic molecule. It has been observed in vitro that PSA inhibits the proliferation, migration, and invasion of endothelial cells, blocks the response of endothelial cells to angiogenic stimulators, suggesting an antiangiogenic activity of PSA (Fortier et al., 1999). PSA is also able to activate latent transforming growth factor-P (TGF- P) (Dallas et al., 2005; Killian et al., 1993). However, the role of TGF- p in breast cancer is very complex. Several researchers have suggested a biphasic action of TGF- P, which acts as an antipromotor at the early stages of breast cancer development but later acts as a promoter enhancing the malignant conversion 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and invasion (reviewed in (Walker, 2000)). Another target of PSA proteolytic modulation is parathyroid hormone-related protein (PTHrP). The cleavages of PTHrP by PSA results in its inactivation (Cramer et al, 1996). Although PTHrP signaling has been shown to be important in both normal mammary gland development (Dunbar and Wysolmerski, 1999) and in the bone metastasis of breast cancer (Kakonen and Mundy, 2003), evidence is so sparse that the involvement of the PTHrP cleavage by PSA in breast cancer is largely unknown. It is very important to note that all the above-mentioned studies investigated the effects of PSA at high concentrations. In fact, these concentrations were much higher than that has been found in female breast tissue or in female serum. Therefore, whether or not these activities of PSA have any biological relevance in breast cancer development and progression is not known. Probably the most relevant in vitro evidence for a role of PSA in breast cancer came from a study done by Lai et al in 1996. In this study, it was found that at very low concentrations (from 0.001 ng/mL up to 0.1 ng/mL), PSA significantly inhibited MCF-7 breast cancer cell growth and the degree of inhibition increased with decreasing concentrations of PSA. At concentrations greater than 0.1 ng/mL, the inhibitory effect disappeared. It was also found that within the same concentration range, PSA showed a stimulatory effect on the oxidative activity of estrogen 17-oxidoreductase (alias, 17 hydroxysteroid dehydroxynase), which catalyzes the conversion o f the potent estrogen estradiol to the less potent estrogen estrone (Lai et al., 1996). If confirmed, this would not only provide a potential 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. explanation for the observed association between high PSA level and good prognosis in patients with breast cancer found in some epidemiological studies, but also would it suggest a role of PSA in breast cancer susceptibility. 2.4. Hormonal regulation of PSA production in females The expression of PSA is under hormonal regulation, hi males, androgen is the major regulator of PSA expression (Henttu et al., 1992; Luke and Coffey, 1994; Young et al., 1992). In females, it seems that the regulation of PSA production is under control of both progesterone and androgen. In a tissue culture system of breast cancer cell line T-47D, it was observed that androgen and progestin were two independent potent stimulators of PSA production (Zarghami et al., 1997a). Other lines of evidence supporting progesterone as a regulator of PSA gene expression includes: 1) Women who were taking oral contraceptive containing progestin had elevated PSA level in the breast tissues, compared to women not taking oral contraceptives (Yu et al., 1995a). 2) In premenopausal women, there was a change in serum PSA levels during the menstrual cycle and this change correlated with change in serum progesterone level (Zarghami et al., 1997b). In the same study, both progesterone and sera obtained from study participants during the mid- to late luteal phase were able to up-regulate the PSA mRNA and protein level in a T-47D breast cancer cell line (Zarghami et al., 1997b). Studies on hirsute women provided strong in vivo evidence that in females PSA production is under the influence o f androgen as well. Hirsutism is one of the manifestations of androgen excess in women. In a study by Melegos et al, it was 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. found that serum PSA levels were significantly higher in hirsute women than in normal women. Furthermore, PSA values were significantly positively correlated with serum level of 3a-androstanediol glucuronide, a major metabolite of the potent androgen dihydrotestosterone (Melegos et al., 1997). This androgen-regulated PSA production in women was later supported by a double-blinded, placebo-controlled study assessing the effects o f three antiandrogen drugs (spironolactone, flutamide and finasteride) in 40 hirsute women. After 6 months of treatment, there was a significant reduction of serum PSA level in women in all three treatment groups, but not in women who took the placebo. The authors therefore concluded that serum PSA may be a biochemical marker o f androgen action in females (Negri et al., 2000). Androgen-regulated PSA expression is thought to be mediated by the binding of AR-ligand (androgen) complex to certain DNA sequences within the regulatory region o f the PSA gene. These sequences are called androgen responsive elements (AREs). At least three such response elements in PSA gene have been identified (see details in Chapter 4). It is known that the consensus sequence of the response element for the androgen receptor (AR), GG(A/T)ACAnnnTGTTCT, also binds to progesterone receptor (PR) and glucocorticoid receptor (GR) and mediate the transcription by PR and GR (Cleutjens et al., 1997a; Lieberman et al., 1993; Roche et al., 1992). Therefore, it would be reasonable to speculate that in females, PR mediates the effect of progesterone through the AREs in the PSA gene. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.5. Polymorphisms in the PSA gene Recent studies have reported that a polymorphism in the ARE I of the PSA gene (G-> A at position -158) may contribute to the variations in serum PSA levels. In a study conducted by our group, it was found that the AR CAG repeat polymorphism and this polymorphism in ARE-I of PSA gene contributed interactively to the variation in serum PSA level in 420 healthy male subjects (Xue et al., 2001). Carriers of the A allele had higher serum PSA levels. In addition, longer AR CAG repeats were associated with lower serum PSA levels among subjects with PSA gene AA and AG genotypes, but not among subjects with GG genotype. Given the polymorphism’s position in a high-afflnity androgen response element, it has been speculated by Xue et al, that the two alleles probably encode ARE-Is with different binding affinity to androgen receptor (Xue et al., 2001). The higher serum PSA levels associated with AA genotype were also found in two other studies of prostate cancer patients (Medeiros et al., 2002) and men without prostate cancer (limited to men with serum PSA<4 ng/ml) (Salinas et al., 2005). However, three other studies failed to confirm the association between this polymorphism with serum PSA level in men without prostate cancer (Gsur et al., 2002; Rao et al., 2003; Xu et al., 2002). The same PSA promoter polymorphism was also studied in females. DNA from the tissue of 143 breast cancer cases (women who underwent surgical treatment, median age 54 with a range o f 25-93 yrs, from Italy) and DNA from blood samples of 48 controls (female outpatients with no malignancy, median age 57 with 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. a range o f 50-70 yrs, from Canada) were compared this study (Bharaj et al., 2000). No difference in genotype distribution was found between cases and controls. However, PSA concentration in tumor tissues of the cases was significantly associated with PSA genotypes, with highest PSA concentration in tissues of cases with the GG genotype, followed by cases with the GA genotype, and lowest PSA levels were found in tissues o f cases with AA genotype (which is opposite to some of the above findings in males). The difference in PSA level between cases with GG and AA genotypes was statistically significant. In the survival analysis, cases with AA genotype had significantly worse overall and disease-free-survival compared to cases with GG or GA genotypes. In the only in vitro study using PSA gene promoter constructs differing only by the -158 G/A polymorphism, no difference in PSA gene promoter activity in response to androgen was found (Rao et al., 2003). The association between the ARE I polymorphism and serum PSA found in some studies has been speculated to be a result of linkage disequilibrium with other functional polymorphisms in the PSA gene regulatory region (see details in Chapter 4). In summary, although still disputable, new evidence from experimental, clinical and epidemiological studies tends to point to the idea that PSA plays a role in breast cancer development and/or progression, either as a marker o f hormone (androgen and progesterone) activity or directly through its protease activity. Studies to assess the impact o f PSA gene polymorphisms on breast cancer are warranted, especially in conjunction with the AR CAG repeat polymorphism. 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3. A CASE-CONTROL STUDY OF ANDROGEN RECEPTOR AND PROSTATE-SPECIFIC ANTIGEN GENE POLYMORPHISMS AND BREAST CANCER IN AFRICAN-AMERICAN WOMEN 3.1. Introduction The role of androgens in breast carcinogenesis is unclear. Higher circulating androgen levels observed in breast cancer patients compared to healthy controls, especially among postmenopausal women, suggest that androgens may increase breast cancer risk (reviewed in (Lillie et al., 2003; Secreto and Zumoff, 1994; Somboonpom and Davis, 2004; Zumoff, 1994)). In contrast, many recent experimental studies found that androgens inhibit growth of cultured breast cancer cell lines (Hackenberg and Schulz, 1996; Lapointe and Labrie, 2001; Poulin et al., 1988; Szelei et al., 1997) and tumor xenografts in athymic mice (Dauvois et al., 1991). Androgens have also been observed to suppress chemically-induced mammary tumor development in rats (Dauvois et al., 1989) and inhibit estrogen- induced mammary epithelial proliferation in rhesus monkeys (Zhou et al., 2000). Whether androgens can directly suppress breast cancer development in humans has been difficult to assess in epidemiologic studies. Androgen levels are highly correlated with levels of their metabolites, estrogens. Furthermore, androgens (compared to estrogens) have a higher affinity for the steroid hormone binding globulin (SHBG) (Dunn et al., 1981) and thus act to increase the fraction of free (bioavailable) estrogens. Therefore, a positive association between androgen levels 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and breast cancer risk may not indicate a direct carcinogenic effect of androgens, but rather may simply reflect the effects of high concomitant estrogen activity. An alternative approach is to examine androgen receptor (AR) gene variants that alter the receptor function. An association of variant alleles with risk would suggest that androgens act directly (by binding AR protein) to influence cancer development. A common functionally significant variant in the AR gene is a CAG repeat polymorphism which encodes a variable-length polyglutamine tract in the transactivation domain of the protein. Several in vitro studies have shown that long polyglutamine tract length reduces AR transactivation activity (Chamberlain et al., 1994; Irvine et al., 2000; Kazemi-Esfarjani et al., 1995; Tut et al., 1997). Recent epidemiological studies have lent some support for a protective role of androgens in breast cancer in that shorter CAG repeats (with stronger AR transactivation activity) have been associated with reduced breast cancer risk (Giguere et al., 2001; Haiman et al., 2002; Liede et al., 2003; Rebbeck et al., 1999; Suter et al., 2003). However, not all studies found this association (Dunning et al., 1999; Kadouri et al., 2001; Menin et al., 2001; Spurdle et al., 2005; Spurdle et al., 1999). The mechanisms of androgen action and especially of the androgen-induced growth inhibition of breast cancer cells remain unclear, due to our limited knowledge of down-stream genes in the androgen signaling pathway. The PSA gene (also named APS, hK3, or KLK3), which encodes the prostate-specific antigen (PSA) protein, is one of the few well-characterized /fi?-regulated genes. Recent studies not only found PSA protein to be present in many female tissues including breast tissue (Clements 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and Mukhtar, 1994; Yu and Berkel, 1999; Yu and Diamandis, 1995b; Yu and Diamandis, 1995c; Yu et al., 1996), but also implicated PSA as a useful prognostic marker in breast cancer (Yu et al., 1995b; Yu et al., 1998). Higher PSA levels in breast tumors have been associated with a significantly reduced risk of relapse and with better survival after surgery (Yu et al., 1995b; Yu et al., 1998). In the PSA gene promoter, there is a guanine to adenine substitution (-158 G/A) in an androgen- responsive element (ARE-I). An association between this polymorphism and PSA levels in serum or in breast tissue has been reported in some studies (Bharaj et al., 2000; Medeiros et al., 2002; Xue et al., 2001), but not in others (Rao et al., 2003; Xu et al., 2002). In addition, this polymorphism was associated with breast cancer prognosis in one previous study (Bharaj et al., 2000). In this chapter, I present results from a case-control study examining the AR CAG length polymorphism and the -158 G/A polymorphism in the ARE-I of the PSA gene in breast cancer. This study is built on a population-based case-control study conducted in the San Francisco Bay Area (San Francisco, San Mateo, Santa Clara, Alameda, and Contra Costa counties) by Dr. Esther M. John, with funding from the National Cancer Institute and the Department of Defense (R01 CA63446, DAMD 17-96-1-6071, R01 CA77305). The parent study recruited women from three ethnic groups: Non-Hispanic Whites, Hispanics and African-Americans. Because of funding limit (supported by a BCRP dissertation grant) and because almost all previous studies of the AR polymorphism were conducted in non-Hispanic Whites (except for one in Asians), we focused here on African-Americans, who have 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. notably shorter CAG repeat lengths and different age-specific breast cancer incidence rates, compared to non-Hispanic Whites. 3.2. Study design 3.2.1. Study population Eligible cases for this study included African-American women newly diagnosed with primary invasive breast cancer between M ayl, 1997 and April 30, 1999, aged 35-79 and residing in the San Francisco Bay area at the time o f diagnosis. They were identified through the population-based Greater Bay Area cancer registry which ascertains all incident cases as part of the Surveillance, Epidemiology, and End Results (SEER) cancer registry and the California cancer registry programs. Population controls were identified through random digit dialing (using telephone numbers of recently diagnosed cancer patients and replacing the last two digits with random numbers (10 phone numbers per case number were generated). Controls were frequency-matched to cases by race and five-year age group. 3.2.2. Data and biospecimen collection Cases and controls were first contacted by phone for a brief screening to establish study eligibility, assess personal and family history of breast cancer, and verify race/ethnicity (response rate o f 84% and 86% among cases and controls, respectively). All eligible cases and controls were then invited to complete an in- person interview. Trained interviewers administered a structured questionnaire which inquired about a broad array of established and suspected breast cancer risk factors, including demographic background, family history of breast cancer in 1st degree 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. relatives, menstrual and reproductive histories, hormone use, and body size. In addition, the interviewers measured standing height, weight, hip and waist circumferences. Interviews were completed for a total of 292 African-American cases and 305 controls (response rate of 87% and 82% among cases and controls, respectively). In a second home visit, participants were invited by a professional phlebotomist to provide 40 ml of whole blood. Those who declined the blood draw were given the option to provide a mouthwash sample (collected using Scope mouth wash). A total of 249 cases and 255 controls provided a biospecimen sample. 3.2.3. Variable definitions Menopausal status: Women were considered postmenopausal if their periods had stopped or if they had a bilateral oophorectomy more than 1 year prior to diagnosis (cases) or selection into the study (controls). Women who reported a simple hysterectomy or a hysterectomy with one-sided oophorectomy were also classified as postmenopausal if they were aged 55 or older at the time of diagnosis/selection. Also included in this group were women who began using hormone replacement therapy (HRT) prior to the cessation o f menses but had attained age 55 or older at the time of diagnosis/selection. For women under age 55 at the time of diagnosis/selection, their menopausal status was undetermined if they reported a simple hysterectomy or a hysterectomy with one-sided oophorectomy, or they began using HRT prior to the cessation of menses. The remaining women were considered premenopausal. 49 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Body Mass Index (BMI) (kg/m2 ) was calculated based on measured height and weight. For subjects who declined the measurement (16% of cases and 12% of controls), self-reported height and weight were used. Family history referred to l st-degree family history (breast cancer in mother, sister(s), or daughter(s)). A dichotomous variable was used for breast cancer stage representing localized vs. advanced diseases. The advanced stages included: regional extension only; regional lymph nodes involvement only; regional extension and lymph node involvement; and remote metastasis. 3.2.4. Laboratory methods General quality control DNA samples were labeled only by specimen ID and laboratory personnel were blinded to case/control status of DNA samples when genotyping work was carried out.. All genotyping assays included control samples that had genotypes previously confirmed by sequencing. All PCR assays included a “water blank” to guard against contamination of the PCR reaction. To assure assay reproducibility, 5% of samples were included as duplicates with laboratory personnel blinded to duplicate status. Results from these duplicates showed a concordance of 100%. DNA extraction DNA extraction from buffy coat separated from whole blood followed a protocol used by Bass et al (Baas et al., 1984). DNA extraction from mouthwash followed a protocol used by Giacia-Closas (Garcia-Closas et al., 2001). DNA 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. extraction failed for 3 samples, leaving 246 cases and 255 controls for genotype analyses. Genotyping Simple sequence length polymorphism (SSLP) analysis was used to identify the AR exon 1 CAG repeat variant. The genomic region containing the CAG repeat was PCR amplified using the forward primer 5 ’-CGCGAAGTGATCCAGAAC-3’ and the reverse primer 5’-CAGGACCAGGTAGCCTGTG-3’. The PCR reaction was performed in a 20 pi volume consisting of 30 ng of genomic DNA. Taqman Core Reagent Kit (Applied Biosystems, Foster City, CA) was used according to the manufacturer’s instructions, in addition to 6 pmoles of unlabeled forward and reverse PCR primer, plus 0.4 pmol of FAM-labeled reverse primer. Touchdown thermal cycling was performed. The resulting PCR product was mixed with size standard HD-400 (Applied Biosystems, Foster City, CA) and was run on the ABI 3700 capillary sequencer. The allele sizes were scored using GeneScan software (version 3.5) (Applied Biosystems, Foster City, CA). DNA samples from 12 male subjects with various CAG repeat lengths (determined from direct sequencing) were included in each run as controls. A standard curve was drawn based on these 12 control samples and was used to calculate CAG repeat number for study subjects. Genotyping of the single nucleotide polymorphism in the PSA gene was performed by the Taqman assay. The two labeled oligonucleotide probes were 5’- FAM-CAGAACAGC AAGTACTAGCTCTCCCTC-3' and 5’-CY3- AGAACAGCAAGTGCTAGCTCTCCC-3'. In both probes the thymidines were 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. replaced with Propyne-dU to increase the Tm of the probe ~1°C for every addition (Biosearch Technologies, INC, Novato, CA). The forward Primer was 5'- GGTGCATCCAGGGTGATCTAG-3' and reverse primer was 5'-CACACCCAGA GCTGTGGAAG-3'. The PCR reaction was performed in a 15 pi volume consisting of 25 ng of genomic DNA. Taqman Core Reagent Kit (Applied Biosystems, Foster City, CA) was used according to the manufacturer’s instructions, in addition to 1.5 pmoles of the FAM probe, 3 pmoles of the CY3 probe, and 4.5 pmoles of each PCR primer in each reaction. The thermal cycling was performed first for 10 min at 95°C, followed by 50 cycles at 95°C for 15 sec and at 64°C for 1 min in MicroAmp Optical plates (Applied Biosystems, Foster City, CA). The fluorescence signal was detected using an ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Foster City, CA). Genotypes were determined using the graphical view from ABI Sequence Detection software (version 2.1). Nine previously sequenced DNA samples (3 of each genotype) were included as genotype controls. 3.2.5. Statistical analysis We referred to the two AR CAG alleles carried by each woman as the smaller allele (the shorter o f the two alleles) and the larger allele (the longer of the two alleles). Hardy-Weinberg equilibrium (HWE) of the CAG length distribution among the controls was tested by the method of Guo and Thompson (Guo and Thompson, 1992) (SAS, PROC ALLELE). The distributions of allele lengths for cases and controls were compared using the Wilcoxon rank sum test. This was done separately for the smaller, the larger allele (assuming preferential X chromosome inactivation) 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and the average length of the two alleles (assuming random X chromosome inactivation). Logistic regression was used to estimate odds ratio (OR) and 95% confidence interval (Cl) for the effect of AR CAG repeat length on breast cancer risk. The linear effect of AR CAG repeat length was estimated by including the smaller and the larger allele lengths as continuous variables in a logistic regression model. CAG repeat lengths were also dichotomized as short and long using the median CAG length as the cut point. ORs were estimated by comparing genotypes long/long and short/long to the “reference” genotype short/short. We also dichotomized CAG length using the same cut points as those in previously published papers in order to allow comparison. For the PSA gene, HWE of the genotype distribution among the controls was tested using the Pearson goodness-of-fit statistic with a x2 distribution with ldf. ORs and 95% CIs were estimated comparing genotypes A/A and G/A to G/G. This coding scheme modeled a co-dominant effect. A test of trend was also performed by including in the logistic model a variable coded as 0, 1,2 for the number o f “at-risk” alleles. All models were adjusted for age (continuous). Other known or suspected breast cancer risk factors that were examined for their potential confounding effects were: age at menarche, age at first full-term pregnancy, parity, body mass index (BMI), menopausal status, age at menopause, previous and current use of oral contraceptives, previous and current use of hormone replacement therapy (HRT), family history o f breast cancer, benign breast disease and education. None of these 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. factors, either individually or jointly, changed the OR estimates by more than 10% and therefore were not included in the final logistic regression models. We considered age, menopausal status, BMI, HRT use, and family history of breast cancer as potential effect modifiers. ORs for “risk” genotypes were estimated within strata defined by these factors. Formal tests of effect modification were performed by including the appropriate interaction terms in the logistic model. Cases were also stratified by disease stage (localized vs. regional extension or distant spread) and estrogen receptor (ER) or progesterone receptor (PR) status. Polytomous logistic regression was used to assess associations of AR and PSA genotypes with cancer subtypes and to test for homogeneity across subtypes. To assess possible interactions between the AR and PSA genes, ORs were estimated for each AR (as dichotomous variable)/PSA genotype combination. ORs were also estimated for the linear effect o f CAG repeat length within each of the three PSA genotype categories. Power for detecting a shift of two CAG repeats between the distributions of CAG repeat lengths in cases and controls with a positive family history was estimated by bootstrap (Collings and Hamilton, 1988). Each bootstrap sample consisted of 67 observations (32 cases and 35 controls). One thousand samples were drawn with replacement from the empirical distribution of controls in our dataset, with two CAG repeats being added to the sampled value for each case. Wilcox on rank-sum tests were performed on each sample and the percentage of significant results was calculated. 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. All analyses were performed with the use of STATA software (unless otherwise specified). All reported p-values were 2-sided and statistical significance is taken as p value <0.05. 3.3. Results Genotyping results were missing for 13 subjects (7 cases and 6 controls) due to PCR failure (7 failed the CAG assay and 8 failed the PSA assay), leaving 488 subjects (239 cases and 249 controls) in final data analyses. Characteristics of cases and controls are shown in Table 3.1. The mean ages o f cases at diagnosis and controls at selection into the study were 55.6 (SD, 11.5) and 55.3 (SD, 11.6), respectively. Among premenopausal women, the mean BMI was significantly lower in cases than in controls (30.1 vs. 32.4, p=0.05), whereas postmenopausal cases and controls had similar mean BMI (31.2 vs. 31.5, p=0.71). A history of biopsy- confirmed benign breast disease was significantly more common among cases compared to controls (p=0.006). Cases were more likely to be nulliparous, more likely to have higher education, a positive family history, a history o f oral contraceptive usage, and to have a first full-term pregnancy at later ages but less likely to have previous or current HRT usage, although the differences were not statistically significant. Age at menarche or age at menopause did not significantly differ between cases and controls. The CAG genotype frequencies of the cases and controls are displayed in Table 3.2. The genoytpe frequenices of the controls did not show significant departure from Hardy-Weinberg Equilibrium (p=0.14, exact test). Among all 488 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. study subjects, the CAG repeat lengths ranged from 8 to 30, with a median of 19 (Figure 3.1). The comparisons of the means and the medians of the CAG repeat lengths (for the smaller allele, the longer allele, and the average of the two alleles carried by one woman, seperately) between the cases and controls are shown in Table 3.3. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.1. Characteristics of the study participants, by case-control status Cases Controls P value (n=239) t____________ (n=249) j_________________ Age Mean (SD)* Median (IQR)1 55.6(11.5) 54 (47-64) 55.3 (11.6) 54 (46-65) 0.79 0.78 Menopausal status Premenopausal 72(30.1% ) 82 (32.9%) Postmenopausal 146 (61.1%) 146 (58.6%) 0.80 Undetermined 21 (8.8%) 21 (8.4%) Education (years) <12 42(17.6% ) 46(18.5% ) 12 53 (22.2%) 64 (25.7%) 13-16 94 (39.3%) 101 (40.6%) >17 50 (20.9%) 38 (15.3%) 0.41 Family history Yes 35 (14.6%) 32 (12.8%) No 204 (85.4%) 217(87.2% ) 0.57 Benign breast disease Yes 61 (25.5%) 38(15.4%) No 178 (74.5%) 209 (84.6%) 0.006 Age at menarche <12 52(21.9% ) 55 (22.4%) 12-13 133 (55.9%) 130(52.9%) >14 53 (22.3%) 61 (24.8%) 0.76 Age at menopause Mean (SD)’ 46.4 (7.02) 46.4 (8.08) 0.97 Median(IQR)f 48(44-51) 48(43-51) 0.78 Parity Nulliparous 44(18.4% ) 34 (13.7%) 1 42(17.6% ) 47(18.9% ) 2 54 (22.6%) 55 (22.1%) 3 43 (18.0%) 48 (19.3%) >4 56 (23.4%) 65 (26.1%) 0.68 Age at 1st full-term pregnancy <20 83 (42.6%) 93 (43.3%) 20-24 59 (30.3%) 79 (36.7%) 25-29 27 (13.9%) 27(12.6% ) >30 26(13.3 %) 16(7.4%) 0.18 History of oral contraceptive use Yes 160 (67.5%) 163 (66.0%) No 77 (32.5%) 84 (34.0%) 0.72 History of HRT use (postmenopausal women) Yes 70 (48.0%) 83 (58.0%) No 76 (52.0%) 60 (42.0%) 0.09 BMI Premenopausal Mean (SD) 30.1±6.21 32.4±8.20 0.05 Median(IQR)t 29.3 (25.0-34.5) 30.5 (25.5-39.9) 0.11 <25 18(25.0%) 16(19.5%) 25-29 20 (27.8%) 23 (28.1%) >30 34 (47.2%) 43 (52.4%) 0.69 Postmenopausal Mean (SD) 31.2±6.34 31.5±7.17 0.71 Median(IQR)+ 30.5 (26.6-35.2) 30.8 (26.5-34.6) 0.97 <25 24 (16.6%) 26 (17.8%) 25-29 44 (30.3%) 40 (27.4%) >30 77 (53.1%) 80 (54.8%) 0.85 * SD, standard deviation;TIQR, inter-quartile range; * The numbers in the table do not add up due to missing values. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.2. AR CAG genotype frequencies in cases and controls (CAG )n genotype # o f cases # o f controls (CAG)n genotype # o f cases # o f controls (CAG)n genotype # o f cases # o f controls c ----------- U ----------- 8,16 1 1 15,20 5 1 18,28 0 1 9,14 0 1 15,21 0 4 19,19 3 1 10,10 0 1 15,22 1 1 19,20 1 4 10,15 0 1 15,23 2 1 19,21 3 5 10,17 1 0 15,24 0 1 19,22 2 4 11,18 0 1 16,16 3 5 19,23 5 2 11,21 2 1 16,17 6 9 19,24 1 3 12,17 3 0 16,18 7 1 19,25 0 3 12,18 1 0 16,19 2 3 19,27 1 3 12,20 1 0 16,20 8 7 19,28 1 0 12,22 1 0 16,21 7 5 19,29 1 1 12,23 1 0 16,22 5 3 19,30 1 0 13,13 0 1 16,24 2 2 20,20 4 5 13,14 0 2 16,25 2 2 20,21 8 7 13,16 1 0 16,26 1 1 20,22 2 5 13,17 1 6 16,27 0 2 20,23 6 3 13,18 0 1 16,30 1 0 20,24 5 5 13,19 2 2 17,17 12 7 20,25 2 1 13,20 1 0 17,18 5 10 20,26 1 2 13,21 1 0 17,19 11 8 20,27 0 1 13,22 0 1 17,20 6 9 20,28 1 0 13,23 0 2 17,21 6 9 21,21 1 4 13,24 1 0 17,22 4 3 21,22 2 3 14,15 0 1 17,23 5 5 21,23 1 2 14,16 0 1 17,24 3 3 21,24 3 2 14,17 2 1 17,25 3 1 22,22 1 2 14,18 2 2 17,27 1 0 22,23 2 1 14,19 2 3 18,18 4 3 22,25 1 0 14,20 1 1 18,19 4 1 22,26 0 1 14,21 0 1 18,20 5 7 22,27 1 0 14,22 1 3 18,21 3 2 23,25 0 1 14,23 1 1 18,22 8 6 24,24 0 1 14,24 0 1 18,23 3 2 24,26 1 0 15,16 4 2 18,24 2 1 24,27 1 0 15,17 3 4 18,25 2 2 25,25 0 1 15,18 2 5 18,26 2 0 25,26 0 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. o <N > £ > 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 AR[CAG]n ■ Cases H Controls Figure 3.1. The distribution of the CAG repeat lengths in cases and controls Table 3.3. The comparison of the AR allele lengths between cases and controls Overall 19.0 18.9 0.60 19 19 0.75 (3.3) (3.3) (17-21) (17-21) Smaller allele 17.1 17.1 0.99 17 17 0.94 (2.4) (2.6) (8-24) (8-25) Larger allele 20.8 20.6 0.41 21 21 0.64 (2.9) (3.0) (16-30) (10-29) Average o f the 19.0 18.9 0.61 19 19 0.86 two allele (2.3) (2.4) (12-25.5) (10-25.5) *t-test ^ wilcoxon test Mean (SD) Median (range) cases controls P value* cases controls P valueT The cumulative distribution curves for the smaller and the larger allele of the two alleles carried by one woman were almost identical between the cases and the controls (p=0.94 for the smaller allele; p=0.64 for the larger allele, Wilcoxon tests) (Figure 3.2). The cumulative distributions for the average lengths o f the two alleles 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. were also almost identical between the cases and the controls (p=0.89, not shown in the figure). o o - 00 ~ o CM ' — S r - o - 22 26 30 18 20 24 28 10 12 14 16 8 Number of CAG repeats ------• ------ ca se smaller — — ctrl_smaller ------■------ ca seja rg er — *■— ctrljarger Figure 3.2. The cumulative distribution curves of the smaller and the larger alleles in all cases and controls The ORs associated with a one repeat increment were not significantly different from 1 (OR=1.0, 95% Cl =0.93-1.07 for the smaller allele and OR=1.03, 95% Cl =0.97-1.09 for the larger allele). ). Neither were significant associations observed when CAG repeat lengths were dichotomized. Compared with women having 0 alleles with (CAG)n > 22 (corresponding to the SS genotype in the literature), women who carried 1 or 2 alleles with (CAG)n > 22 (SL and LL genotypes, respectively) had an OR=1.09 (95% 0=0.75-1.57). This cutoff point had been used in several reports previously. Similar results were obtained when other cutoff points were used. For example, women who carried 1 or 2 alleles with 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (CAG)n >19 (the median repeat length among the study subjects) had an OR=1.13 (95% Cl 0.75-1.69), compared to women with two shorter alleles (CAG)n<19 (Table 3.4). Table 3.4. AR CAG polymorphism and breast cancer risk among all subjects Cases Controls Age-adjusted OR* N (%) N (%) (95% Cl) No. o f alleles with CAGn>23 0 172 (72.0) 187 (75.1) R ef (1.0) 1 65 (27.2) 58 (23.3) 1.22 (0.81-1.83) 2 2 (0.8) 4 (1 .6 ) 0.55 (0.10-3.05) 1 or 2 67 (28.0) 62 (24.9) 1.17(0.78-1.76) No. o f alleles with CAGn>22 0 145 (60.7) 156(62.7) R ef (1.0) 1 87 (36.4) 85 (34.1) 1.10(0.76-1.60) 2 7 (2.9) 8 (3.2) 0.94 (0.33-2.66) 1 or 2 94 (39.3) 93 (37.3) 1.09 (0.75-1.57) No. o f alleles with CAGn>21 0 114(47.7) 118(47.4) R ef (1.0) 1 111 (46.4) 112(45.0) 1.02 (0.71-1.48) 2 14 (5.9) 19 (7.6) 0.76 (0.36-1.59) 1 or 2 125 (52.3) 131 (52.6) 0.99 (0.69-1.41) No. o f alleles with CAGn>20 0 82 (34.3) 84 (33.7) R ef (1.0) 1 114(47.7) 117(47.0) 1.00 (0.67-1.49) 2 43 (18.0) 4 8(19.3) 0.92 (0.55-1.53) 1 or 2 157 (65.7) 165 (66.3) 0.97 (0.67-1.42) No. o f alleles with CAGn>19 0 58 (24.3) 66 (26.5) R ef (1.0) 1 119(49.8) 109 (43.8) 1.24 (0.80-1.93) 2 62 (25.9) 74 (29.7) 0.95 (0.59-1.55) 1 or 2 181 (75.7) 183 (73.5) 1.13 (0.75-1.69) No. o f alleles with CAGn>18 0 37(15.5) 43 (17.3) R ef (1.0) 1 107(44.8) 107(43.0) 1.16(0.69-1.94) 2 95 (39.7) 99 (39.8) 1.12 (0.66-1.88) 1 or 2 202 (84.5) 206 (82.7) 1.14 (0.71-1.84) No. o f alleles with CAGn>l 7 0 9 (3.8) 16(6.4) R ef (1.0) 1 79 (33.1) 79 (31.7) 1.78 (0.74-4.26) 2 151 (63.2) 154 (61.8) 1.74 (0.75-4.07) 1 or 2 * " " - _ 230 (96.2) 233 (93.6) 1.76 (0.76-4.05) f Further adjusting for family history o f breast cancer, benign breast disease, age at menarche, parity, age at first full-term pregnancy, menopausal status, HRT use, OC use, education, and BMI did not change the risk estimate appreciably (no more than 10%). These adjusted ORs and 95% CIs were not shown here. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The association between the CAG repeat genotypes and breast cancer risk did not differ significantly by subjects’ ages (Table 3.5) or by menopausal status (Table 3.6). Among postmenopausal women, the genotype and cancer association did not differ significantly by HRT use (Table 3.7). We did not observe significant effect modification by BMI overall (Table 3.8), or among postmenopausal women only (Table 3.9) Family history of breast cancer, however, significantly modified the genotype effect. While cases and controls without a family history had distributions (for both the smaller and the larger alleles) that were nearly identical (p=0.93 for the smaller allele, p=0.62 for the larger allele) (Figures 3.3), the distribution for the larger allele was noticeably shifted to the right among cases with a family history compared to controls with a family history (p=0.008). For the smaller allele, there was no significant shift (p=0.62) (Figure 3.4). The distributions o f the average lengths of the two alleles did not differ significantly between cases and controls without a family history (p=0.63), or with a family history (p=0.09) (not shown in Figure 3.3 or Figure 3.4). Among women with a family history, the OR associated with a one CAG repeat increment in the larger allele was 1.33 (95% CI= 1.07-1.64), the OR for the smaller allele was 1.01 (95% CI= 0.80-1.28). Among women without a family history, the ORs were 1.00 (95% CI= 0.93-1.06) for the larger allele and 1.00 (95% CI= 0.92-1.07) for the smaller allele, respectively. For the larger allele, the interaction was significant (p=0.01). In analyses using dichotomous variables, similar effect modification was observed. Among women with a positive family history, having one or two alleles with 22 or more CAG repeats was associated with 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. a significantly increased risk (OR=3.18, 95% CI=1.08-9.36), while no association was observed among women without a family history (OR=0.92, 95% CI=0.62- 1.37). The interaction was statistically significant (p=0.03). Similar results were obtained when other cutoff points were used (Table 3.10). There was no evidence that CAG repeats had a different influence on breast tumors with different stages or with different ER or PR expression levels (the respective p values for heterogeneity tests were 0.94, 0.15 and 0.25 when 22 was used as the cutoff point) (Table 3.11). Although the data appeared to suggest a difference in genotype effect by PR status when a cutoff point of 19 was used (p for the test of heterogeneity = 0.04), the genotype effect was not significant in either the PR+ or the PR- stratum, and this significant heterogeneity was not observed when other cutoff points were used (e. g. 20, 21 in addition to 22, data not shown). Therefore, this significant heterogeneity result might be due to chance alone. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.5. AR CAG polymorphism and breast cancer risk, stratified by age Age>55 Age<55 Case N (%) Controls N (%) Age-adjusted OR (95% Cl) Case N (%) Controls N (%) Age-adjusted OR (95% Cl) Both allele <19 23(21.7) 32 (28.3) Ref (1.0) 35 (26.3) 34 (25) Ref (1.0) 1 or 2 >19 83 (78.3) 81 (71.7) 1.43 (0.77-2.64) 98 (73.7) 102 (75) 0.94(0.55-1.63) p inter =0.31 Both allele <20 36 (34) 38 (33.6) Ref (1.0) 46 (34.6) 46 (33.8) Ref (1.0) 1 or 2 >20 70 (66) 75 (66.4) 0.98 (0.56-1.72) 87 (65.4) 90 (66.2) 0.97 (0.59-1.61) p inter =0.96 Both allele <21 47 (44.3) 52 (46) Ref (1.0) 67 (50.4) 66 (48.5) Ref (1.0) 1 or 2 >21 59 (55.7) 61 (54) 1.07(0.63-1.83) 66 (49.6) 70 (51.5) 0.93 (0.57-1.50) p inter =0.70 Both allele <22 62 (58.5) 71 (62.8) Ref (1.0) 83 (62.4) 85 (62.5) Ref (1.0) 1 or 2 >22 44(41.5) 42 (37.2) 1.20(0.70-2.06) 50 (37.6) 51 (37.5) 1.00 (0.61-1.64) p inter =0.63 Table 3.6. AR CAG polymorphism and breast cancer risk, stratified by menopausal status Premenopausal Postmenopausal Cases N (%) Controls N (%) Age-adjusted OR (95% Cl) Cases N (%) Controls N (%) Age-adjusted OR (95% Cl) Both allele <19 16(22.2) 22 (26.8) Ref (1.0) 36 (24.7) 37(25.3) Ref (1.0) 1 or 2 >19 56 (77.8) 60 (73.2) 1.27 (0.60-2.68) 110(75.3) 109 (74.7) 1.04(0.61-1.76) p inter =0.65 Both allele <20 21 (29.2) 28 (34.1) Ref (1.0) 53 (36.3) 45 (30.8) Ref (1.0) 1 or 2 >20 51 (70.8) 54 (65.9) 1.25 (0.63-2.48) 93 (63.7) 101 (69.2) 0.78 (0.48-1.27) p inter =0.27 Both allele <21 36 (50.0) 38 (46.3) Ref (1.0) 68 (46.6) 67 (45.9) Ref (1.0) 1 or 2 >21 36 (50.0) 44 (53.7) 0.86 (0.46-1.62) 78 (53.4) 79(54.1) 0.97 (0.62-1.54) p inter =0.76 Both allele <22 43 (59.7) 51 (62.2) Ref (1.0) 90 (61.6) 90 (61.6) Ref (1.0) 1 or 2 >22 29 (40.3) 31 (37.8) 1.11 (0.58-2.12) 56 (38.4) 56 (38.4) 1.01 (0.63-1.61) p inter =0.82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.7. AR CAG polymorphism and breast cancer risk among _______postmenopausal women, stratified by HRT use________ With HRT use Without HRT use Cases N (%) Controls N (%) Age-adjuested OR (95% Cl) Cases N (%) Controls N (%) Age-adjusted OR (95% Cl) Both allele <19 14 (20.0) 16(19.3) Ref (1.0) 22 (28.9) 20 (33.3) Ref (1.0) 1 or 2 >19 56 (80.0) 67 (80.7) 0.96 (0.43-2.13) 54 (71.1) 40 (66.7) 1.23 (0.59-2.56) p inter =0.65 Both allele <20 23 (32.9) 22 (26.5) R ef (1.0) 30(39.5) 22 (36.7) Ref (1.0) 1 or 2 >20 47(67.1) 61 (73.5) 0.74 (037-1.48) 46 (60.5) 38 (63.3) 0.89 (0.44-1.78) pinter =0.71 Both allele <21 27(38.6) 36 (43.4) Ref (1.0) 41 (53.9) 30 (50.0) R ef (1.0) 1 or 2 >21 43 (61.4) 47 (56.6) 1.22(0.64-2.33) 35 (46.1) 30 (50.0) 0.85 (0.43-1.68) p inter =0.46 Both allele <22 38 (54.3) 49 (59.0) Ref (1.0) 52 (68.4) 39 (65.0) Ref (1.0) 1 or 2 >22 32 (45.7) 34(41.0) 1.21 (0.64-2.31) 24(31.6) 21 (35.0) 0.86 (0.42-1.77) p inter =0.48 Table 3.8. AR CAG polymorphism and breast cancer risk, stratified by BMI BMI>30 BMI<30 Case N (%) Controls N (%) Age-adjusted OR (95% Cl) Case N (%) Controls N (%) Age-adjusted OR (95% Cl) Both allele <19 30 (25) 39 (28.7) R ef (1.0) 27 (22.9) 27 (23.9) Ref (1.0) 1 or 2 >19 90 (75) 97 (71.3) 1.21 (0.69-2.12) 91 (77.1) 86 (76.1) 1.05 (0.57-1.94) p inter =0.76 Both allele <20 44 (36.7) 49 (36) R ef (1.0) 37 (31.4) 35 (31) Ref (1.0) 1 or 2 >20 76 (63.3) 87 (64) 0.97 (0.58-1.63) 81 (68.6) 78 (69) 0.97(0.56-1.70) pinter =0.98 Both allele <21 57 (47.5) 65 (47.8) R ef (1.0) 56 (47.5) 53 (46.9) Ref (1.0) 1 or 2 >21 63 (52.5) 71 (52.2) 1.01 (0.62-1.65) 62 (52.5) 60 (53.1) 0.97 (0.58-1.63) p inter =0.93 Both allele <22 75 (62.5) 81 (59.6) R ef (1.0) 69 (58.5) 75 (66.4) Ref (1.0) 1 or 2 >22 45 (37.5) 55 (40.4) 0.88 (0.53-1.46) 49(41.5) 38 (33.6) 1.40 (0.82-2.39) p inter =0.23 65 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.9. AR CAG polymorphism and breast cancer risk among postmenopausal women, ______________________________ stratified by BMI______________________________ BMI>30 BMI<30 Case N (%) Controls N (%) Age-adjusted OR (95% Cl) Case N (%) Controls N (%) Age-adjusted OR (95% Cl) Both allele <19 20 (26.0) 24 (30.0) Ref (1.0) 15(22.1) 13 (19.7) Ref (1.0) 1 or 2>19 57 (74.0) 56 (70.0) 1.22 (0.60-2.45) 53 (77.9) 53 (80.3) 0.87 (0.38-2.00) p inter =0.54 Both allele <20 31 (40.3) 29 (36.3) Ref (1.0) 21 (30.9) 16(24.2) Ref (1.0) 1 or 2 >20 46 (59.7) 51 (63.8) 0.84 (0.44-1.60) 47(69.1) 50 (75.8) 0.72(0.33-1.54) p inter =0.75 Both allele <21 34 (44.2) 40 (50.0) R ef (1.0) 33 (48.5) 27 (40.9) Ref (1.0) 1 or 2 >21 43 (55.8) 40 (50.0) 1.27(0.67-2.37) 35 (51.5) 39(59.1) 0.73 (0.27-1.45) p inter =0.25 Both allele <22 46 (59.7) 49 (61.3) R ef (1.0) 43 (63.2) 41 (62.1) R ef (1.0) 1 or 2 >22 31 (40.3) 31 (38.8) 1.08 (0.57-2.04) 25 (36.8) 25 (37.9) 0.95 (0.47-1.92) p inter =0.82 Table 3.10. AR CAG polymorphism and breast cancer risk, stratified by l s t-degree _____________________family history of breast cancer_____________________ With a family history Without a family history Case N (%) Controls N (%) Age-adjusted OR (95% Cl) Case N (%) Controls N (%) Age-adjusted OR (95% Cl) Both allele <19 3 (8.6) 13 (40.6) Ref (1.0) 55 (27.0) 53 (24.4) Ref (1.0) 1 or 2 >19 32(91.4) 19 (59.4) 7.81 (1.92-31.9) 149 (73.0) 164 (75.6) 0.87(0.56-1.36) p inter <0.01 Both allele <20 7 (20.0) 15(46.9) Ref (1.0) 75 (36.8) 69 (31.8) R ef (1.0) 1 or 2 >20 28 (80.0) 17(53.1) 3.80(1.26-11.5) 129(63.2) 148 (68.2) 0.80(0.54-1.20) p inter =0.01 Both allele <21 13 (37.1) 18(56.3) Ref (1.0) 101 (49.5) 100(46.1) Ref (1.0) 1 or 2 >21 22 (62.9) 14 (43.8) 2.05 (0.76-5.52) 103 (50.5) 117(53.9) 0.87 (0.59-1.30) p inter =0.09 Both allele <22 18(51.4) 25 (78.1) R ef (1.0) 127 (62.3) 131 (60.4) Ref (1.0) 1 or 2 >22 17(48.6) 7(21.9) 3.18 (1.08-9.36) 77 (37.7) 86 (39.6) 0.92 (0.62-1.37) p inter =0.03 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 8 10 12 14 16 18 20 22 24 26 28 30 Num ber of CAG rep eats ----- • — - case_ sm a ller — — - ctrl sm aller ----- ■— - c a s e ja r g e r — a - — - ctrljarger Figure 3.3. The cumulative distribution curves of the smaller and the larger alleles in cases and controls without a positive 1s t degree family history of breast cancer O o - o 00 0 o _ Q.^ O CM O - 8 10 12 14 16 18 20 22 24 26 28 30 Num ber of CAG rep eats ----- • — - c a s e sm aller — 4 - — - ctrl sm aller ----- ■— - c a s e ja r g e r — j t r - - ctrlja rg er Figure 3.4. The cumulative distribution curves of the smaller and the larger alleles in cases and controls with a positive 1s t degree family history of breast cancer 67 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.11. AR CAG polymorphism and breast cancer risk, by tumor stage, ER _________________________ status, or PR status_________________________ Cutoff point=19 Cutoff point=22 ss N (%) SL/ LL Age-adjusted N (%) OR (95% Cl) SS N (%) SS/SL Age-adjusted N (%) OR (95% Cl) Control 66 (26.5) 183 (73.5) Ref (1.0) 156 (62.7) 93 (37.3) Ref (1.0) s Localized cases 38(16.2) 115(49.1) 1.10(0.69-1.74) 93 (60.8) 60 (39.2) 1.07(0.71-1.63) Advanced cases 19(24.1) 60 (75.9) 1.13 (0.62-2.03) 48 (60.8) 31 (39.2) 1.10(0.65-1.85) p* =0.93 p* =0.94 ER+ cases 28(17.1) 100 (61) 1.30(0.78-2.15) 70 (54.7) 58 (45.3) 1.38 (0.90-2.13) ER- cases 19(29.7) 45 (70.3) 0.85 (0.46-1.56) 42 (65.6) 22 (34.4) 0.88 (0.50-1.57) p* =0.23 p* =0.15 PR+ cases 21 (19.1) 89 (80.9) 1.53 (0.88-2.67) 60 (54.5) 50 (45.5) 1.39 (0.88-2.19) PR- cases 26 (32.1) 55 (67.9) 0.76(0.44-1.31) 51 (63) 30 (37) 0.99 (0.59-1.66) p*=0.04 p’ =0.25 ER+PR+ cases 19(18.3) 83 (79.8) 1.57 (0.89-2.79) 54 (52.9) 48 (47.1) 1.49 (0.94-2.38) ER- PR- cases 17(30.4) 39 (69.6) 0.83 (0.44-1.56) 36 (64.3) 20 (35.7) 0.93 (0.51-1.70) p* =0.10 p* =0.17 p values for the test o f heterogeneity (polytomous analysis). The distribution of the PSA -158A/G genotype frequencies among controls did not show significant departure from Hardy-Weinberg equilibrium (p=0.29). There was no significant association between breast cancer risk and PSA genotype overall (OR=0.98, 95% CI=0.65-1.50 for AG vs. AA and OR=1.04, 95% CI=0.62- 1.76 for GG vs. AA). Nor were there significant associations after the subjects were stratified by their age or menopausal status (respective p values for interaction tests were 0.74 and 0.10), HRT use (p=0.44), or family history of breast cancer (p=0.82) (Table 3.12). ORs did not vary by stage of disease, or ER or PR status (the respective p values for the heterogeneity tests were 0.36, 0.96 and 0.35). (Table 3.13). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.12. The PSA gene ARE-I -158 G/A polymorphism and breast cancer risk ARE-I Cases N (%) Controls N (%) Age-adjusted OR (95% Cl) A ll women AA 65 (27.2) 68 (27.3) 1.0 AG 125 (52.3) 132 (53.0) 0.98 (0.65-1.50) GG 49 (20.5) 49 (19.7) 1.04 (0.62-1.76) P*=0.90 Menopausal Pre- AA 26 (36.1) 28 (34.1) 1.0 status AG 31 (43.1) 44 (53.7) 0.75 (0.37-1.52) GG 15 (20.8) 10(12.2) 1.81 (0.68-4.84) p trd * =0.46 Post- AA 32(21.9) 37 (25.3) 1.0 AG 86 (58.9) 74 (50.7) 1.33 (0.76-2.35) GG 28 (19.2) 35 (24.0) 0.92 (0.47-1.84) p trd * =0.86 p inter f =0.10 HRT use among post menopausal women Yes AA AG GG 13 (18.6) 41 (58.6) 16(22.9) 23 (27.7) 39 (47.0) 21 (25.3) 1.0 1.90 (0.84-4.33) 1.36(0.53-3.51) p trd * =0.55 No AA 19 (25.0) 13 (21.7) 1.0 AG 45 (59.2) 35 (58.3) 0.88 (0.38-2.02) GG 12(15.8) 12 (20.0) 0.69 (0.24-2.01) p trd ’=0.51 p inter * =0.44 Family Yes AA 10 (28.6) 8 (25.0) 1.0 history AG 18(51.4) 19(59.4) 0.80 (0.25-2.51) GG 7 (20.0) 5 (15.6) 1.05 (0.24-4.68) p trd ’ =0.99 No AA 55 (27.0) 60 (27.6) 1.0 AG 107 (52.5) 113 (52.1) 1.04(0.66-1.63) GG 42 (20.6) 44 (20.3) 1.05 (0.60-1.84) p trd * =0.86 p inter + =0.82 p value for the test of trend. ^ p value for the test o f interaction (AG vs. AA and GG vs. AA). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.12. The PSA gene ARE-I -158 G/A polymorphism and breast cancer risk _____________________________ (continued)_____________________________ ARE-I Cases N (%) Controls N (%) Age-adjusted OR (95% CD Age <55 AA 41 (30.8) 39 (28.7) 1.0 AG 67 (50.4) 73 (53.7) 0.86 (0.50-1.50) GG 25(18.8) 24 (17.7) 0.94 (0.46-1.94) p trd * =0.80 >55 AA 24 (22.6) 29 (25.7) 1.0 AG 58 (54.7) 59 (52.2) 1.20 (0.63-2.32) GG 24 (22.6) 25 (22.1) 1.17 (0.54-2.55) p trd * =0.69 p interf =0.74 >30 AA 30(25.1) 40 (29.4) 1.0 BMI AG GG 64 (53.3) 26 (21.7) 72 (52.9) 24(17.7) 1.19(0.66-2.12) 1.44(0.70-2.99) p trd * =0.32 <30 AA 35 (29.7) 28 (24.8) 1.0 AG 61 (51.7) 60 (53.1) 0.81 (0.44-1.49) GG 22 (18.6) 25 (22.1) 0.69 (0.32-1.48) p trd * =0.33 p interf =0.39 BMI >30 AA 16(26.8) 23 (28.8) 1.0 in postmenopausl women AG GG 46 (59.7) 15(19.5) 43 (53.8) 14(17.5) 1.48 (0.68-3.22) 1.53 (0.58-4.04) p trd * =0.32 <30 AA 16(23.5) 14(21.2) 1.0 AG 40 (58.8) 31 (47.0) 0.81 (0.44-1.49) GG 12(17.7) 21 (31.8) 0.69 (0.32-1.48) p trd * =0.33 p interf =0.39 p value for the test o f trend. * p value for the test o f interaction (AG vs. AA and GG vs. AA). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.13. The PSA gene ARE-I -158 G/A polymorphism and breast cancer risk, by tumor stage, ER or PR status P S A genotype A A A G GG Controls 68 (27.3) 132(53.0) 4 9 (1 9 .7 ) Cases (localized) 36 (23.5) 85 (55.6) 32 (20.9) OR (95% Cl) 1.0 (ref) 1.25 (0.75-1.99) 1.19(0.65-2.17) Ptrd‘=0.56 Cases (advanced) 26 (32.9) 3 8 (48.1) 15 (19.0) OR (95% Cl) 1.0 (ref) 0.76 (0.43-1.36) 0.89(0.41-1.82) Ptrd * = 0 .6 0 P h etero b=0.36 Controls 68 (27.3) 132(53.0) 49 (19.7) Cases (ER+) 35 (27.3) 66 (51.6) 27 (21.1) OR (95% Cl) 1.0 (ref) 0.96 (0.58-1.58) 1.03 (0.55-1.93) Ptrd*=0.94 Cases (ER-) 19(29.7) 33 (51.6) 12(18.8) OR (95% Cl) 1.0 (ref) 0.91 (0.48-1.72) 0.91 (0.40-2.05) ptrd*=0.80 P hetero* — 0.96 Controls 68 (27.3) 132(53.0) 49 (19.7) Cases (PR+) 26 (23.6) 59 (53.6) 25 (22.7) OR (95% Cl) 1.0 (ref) 1.15 (0.67-1.99) 1.29 (0.67-2.51) Ptrd* = 0.45 Cases (PR-) 27 (33.3) 40 (49.4) 14(17.3) OR (95% Cl) 1.0 (ref) 0.77(0.43-1.36) 0.73 (0.34-1.53) Ptrd * = 0 .3 5 P h etero — 0.35 Controls 68 (27.3) 132(53.0) 4 9 (1 9 .7 ) Cases (ER+ PR+) 2 2 (2 1 .6 ) 56 (54.9) 24 (23.5) OR (95% Cl) 1.0 (ref) 1.30(0.73-2.30) 1.46 (0.74-2.92) ptr d * =0.27 Cases (ER- PR-) 15(26.8) 3 0 (53.6) 11 (19.6) OR (95% Cl) * . _ 1.0 (ref) 1.04(0.53-2.08) 1.05 (0.44-2.50) Ptrd* = 0 9 0 Pheterot = 0.79 p value for the test o f trend. f p value for the test o f heterogeneity in polytomous analysis (AG vs. AA and GG vs. AA). * The numbers in parentheses are percentage, unless otherwise specified To explore possible gene-gene interaction, the effect o f the PSA gene polymorphism was estimated within stratum defined by the AR genotypes (both alleles <22 vs. 1 or 2 alleles > 22). There was no evidence for a significant gene-gene interaction (p for in te ra c tio n a l5) (Table 3.14). 71 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.14. Gene-gene interaction between the AR CAG polymorphism and the PSA gene ARE-I -158 G/A polymorphism in breast cancer CAG repeat PSA genotype cases controls OR and 95%CI Both AA 43 (29.7) 46 (29.5) 1.0 alleles <22 AG 75 (51.7) 74 (47.4) 1.08 (0.64-1.83) GG 27 (18.6) 36(23.1) 0.80 (0.42-1.54) P trd=0.58 1 or 2 AA 22 (23.4) 22 (23.7) 1.0 alleles >22 AG 50 (53.2) 58 (62.4) 0.85 (0.42-1.72) GG 22 (23.4) 13 (14) 1.64(0.66-4.09) P trd=0.34 P inter=0.15 We also failed to observe any significant interaction when a linear effect of per CAG repeat increment in breast cancer risk was examined by the PSA gene polymorphism (Table 3.15). Table 3.15. Linear effect of per CAG repeat increment in breast cancer risk ___________ by the PSA gene ARE-I -158 G/A polymorphism___________ Among AA genotype Among AG genotype Among GG genotype P for interaction Effect per repeat o f the shorter allele 1.10(0.96-1.27) 0.94 (0.86-1.04) 1.02 (0.85-1.22) 0.17 Effect per repeat o f the longer allele 1.01 (0.90-1.13) 1.01 (0.93-1.10) 1.08 (0.94-1.24) 0.72 Effect per repeat o f the average o f the two alleles 1.07 (0.92-1.23) 0.98 (0.88-1.08) 1.08 (0.90-1.29) 0.51 3.4. Discussion In this population-based case-control study, we found that among African- American women with a first-degree family history of breast cancer, a significant increase in risk was associated with carrying one or two AR CAG long alleles. This Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. is consistent with several reports that shorter AR CAG repeat lengths are associated with reduced risk (Liede et al., 2003; Suter et al., 2003), at least in certain groups of women, for example, postmenopausal women (Giguere et al., 2001), BRCA1 mutation carriers (Rebbeck et al., 1999) and women with a family history of breast cancer (Haiman et al., 2002). Our study is the first to examine the AR CAG length polymorphism and breast cancer risk among African-American women. There is a notable racial difference in the distribution of CAG repeat lengths, with short lengths found most commonly in African-Americans (Edwards et al., 1992). The consistency o f our finding with earlier studies of other racial groups (whites in (Giguere et al., 2001; Haiman et al., 2002; Suter et al., 2003) and Asians in (Liede et al., 2003)) provides evidence against population stratification/confounding by ethnicity as an explanation for our result. Moreover, for population stratification to cause substantial bias, there must be substantial variations across ethnicities in both allele frequencies and disease rates (after adjustment for risk factors other than the genotype of interest), and the allele frequencies must also track with disease rates across ethnicities (Wacholder et al., 2000). Although African-Americans have fewer long (high-risk) CAG repeats, their breast cancer incidence rate does not differ greatly from U.S. whites after adjustment for differences in the distribution of known risk factors (Pike et al., 2002). In addition, the differences in the AR gene allele frequencies between racial/ethnic groups do not always track with the racial/ethnic differences in breast cancer incidence rates. Compared to non-Hispanic whites, African-Americans have 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. lower age-adjusted incidence rates postmenopausally, but higher rates among those aged 40 and younger (Parkin et al., 2002). Therefore, if the association observed in our study was solely due to population stratification, we would expect it to be stronger in postmenopausal women, among whom the difference in AR CAG lengths and in cancer incidence rates track together. The fact that we did not find a significant association between CAG genotype and breast cancer risk when analysis was limited to postmenopausal women further indicates that our finding was unlikely to be due to population stratification. Our results are in complete agreement with a report of the Nurses’ Health Study that an inverse association with short CAG repeats was only found among women with a positive family history o f breast cancer (Haiman et al., 2002). Our finding is also somewhat in accordance with a study by Rebbeck et al, in which women carrying BRCA1 mutations were found to be at significantly lower risk of developing breast cancer and were diagnosed at a later age if they also carried an AR gene with shorter CAG repeats, compared to BRCA1 carriers with longer CAG repeats (Rebbeck et al., 1999). A statistical interaction between AR genotype and BRCA1 mutation status is strongly supported by in vitro studies suggesting that BRCA1 protein is an AR coactivator (Park et al., 2000; Yeh et al., 2000) . However, the modifying effect of AR CAG repeats on breast cancer risk in BRCA1 mutation carriers observed by Rebbeck et al (Rebbeck et al., 1999) was not confirmed by three subsequent studies in mostly Ashkenazi Jewish carriers (Kadouri et al., 2001), Australian and British carriers (Spurdle et al., 2005), and Italian women from high- 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. risk families (Menin et al., 2001). Small sample size could be one reason for the inconsistent findings. For example, even in the two studies with more than 300 subjects (Rebbeck et al., 1999; Spurdle et al., 2005), the use of cutoff points at the extreme end of the CAG repeat distribution resulted in small numbers of subjects in some groups. Our lack o f information on BRCA1 mutation status precluded the investigation of the interaction between AR and BRCA1 in the present study. While we expect that some women with a family history of breast cancer will carry BRCA1 mutations, most women probably will not, because of very low frequency of BRCA1 mutation carriers in African-American women (Newman et al., 1998; Olopade et al., 2003). It is possible that the interaction between family history and AR CAG genotype observed in our study and in the Nurse’s Health Study is due, at least in part, to variant(s) in other gene(s) or to some other familial risk factor(s). We realize that our sample size also became very small in the family history-positive stratum. However, by examining the entire distribution curve rather than relying solely on cutoff points, we had reasonable power (74%) to detect a shift of two CAG repeats between cases and controls with a positive family history. Nevertheless, studies with larger numbers of BRCA1 mutation carriers and/or family history positive subjects are needed to confirm our findings and those observed in other studies (Haiman et al., 2002; Rebbeck et al., 1999). Because of the complexity of the interplay between androgens and estrogens (androgens can serve as estrogen precursors and can increase bioavailability of estrogens by competing for sex-hormone binding globulin, while at the same time, 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. androgens can act directly through binding to AR to counteract estrogen growth- stimulating effects on breast cancer cells— see details in chapter 1), it is reasonable to speculate that the net manifestation of androgen effect (or AR activity) could depend upon the hormonal environment, especially the balance between estrogens and androgens in a giving individual. In a previous study by Giguere et al, a significantly decreased breast cancer risk associated with short AR CAG repeats was found to be limited to postmenopausal women. This could indicate that this protective effect of stronger AR activity is specific to late-onset breast cancer cases (i.e. after menopause) as the authors suggested, or it could also indicate the effect is specific to a group of women with low estrogen-androgen ratio or stronger aromatase activity. Therefore, in the present study, we examined possible effect modification in the AR CAG genotype-breast cancer association by age, menopausal status, postmenopausal hormone replacement therapy, and factors associated with overall aromatase activity (such BMI as a measure of adiposity, and more specifically, BMI among postmenopausal women when adipose tissue is the most important source of estrogen production). We did not observe any modifying effect of these factors, consistent with two other studies that did not observe significant effect modification by menopausal status (Haiman et al., 2002; Liede et al., 2003). O f note, in the study by Giguere et al, shorter CAG repeats were found to be protective against breast cancer mainly among women having a surgical menopause. This result is difficult to explain biologically. It could be due to co-morbidities that may be a cause or a result of changed endogenous hormonal environment or due to possible hormonal 76 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. replacement therapy after surgical menopause. A study has suggested that hormone replacement therapy (especially estrogen plus progestin therapy) was an effect modifier in the association between CAG repeats and mammographic density (Lillie et al., 2004). In the present study, we did not observe a significant association between AR CAG genotype and breast cancer when data analysis was limited to women with surgical menopause (36 cases and 33 controls) (data not shown), nor did we observe the difference in the association by HRT use. However, we were not able to examine different types of HRT use specifically, due to the lack of detailed information from questionnaire. In summary, if there is any difference in the AR genotype-breast cancer association among women with different hormonal profiles, the magnitude of the difference is probably small and may not have been detectable with our limited sample size. Another reason for our inability to replicate the finding by Giguere et al could be that the finding was due to chance alone. Further investigation of the phenomenon observed by Giguere is required to confirm this finding and to identify its possible biological relevance. Since a role of CAG repeat in breast cancer progression has been suggested by Yu et al, with shorter CAG repeats being associated with more aggressive disease (a higher histological grade, positive lymph nodes and worse overall survival) (Yu, 2000), we looked at the effect of CAG genotype separately for localized disease and advanced disease. We did not find significant heterogeneity in the CAG genotype effect in different disease stages. 77 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Another very important pathological characteristic o f breast tumors is ER and PR status. It has been suggested that breast tumors with different estrogen and progesterone receptor expression levels represent distinct subtypes of the disease and hormonal risk factors may be different in these different subtypes (Colditz et al., 2004). Modulating the expression of ER or ER signaling has also been suggested as one of the possible mechanisms for the AR-mediated anti-proliferative effect of androgens (Ando et al., 2002; Labrie et al., 1990; Poulin et al., 1989; Zhou et al., 2000). Therefore, we investigated possible heterogeneity o f the CAG genotype effect across different ER or PR categories. We found no significant difference in the genotype and disease association for ER+ and ER- tumors. Similar results were observed when we looked at PR+ vs. PR- tumors, and when we looked at the two receptor status together (ER+ and PR+ vs. ER- and PR-). Our results are consistent with two previous studies which examined possible differences in AR genotype and breast cancer association by stage, and by ER and PR status (Haiman et al., 2002; Spurdle et al., 1999). Our study differs from previous studies in that we explored the possible mechanism of the androgen effect by examining a genetic variation in the PSA gene, a gene which is known to be transcriptionally regulated by AR, together with the AR CAG polymorphism. In a series of studies by the group of Diamandis et al, PSA expression in the breast appears to be associated with the clinical characteristics of breast tumors. In a study of 174 breast cancer patients, PSA expression in tumor was significantly negatively associated with disease stage and tumor size. Patients with 78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PSA-positive tumors had a significantly reduced risk for relapse after surgery when compared with patients with PSA-negative tumors (Yu et al., 1995b). A later, much larger (more than 900 American patients) study with longer follow-up (an average follow-up time of 5 years), confirmed that PSA presence was associated with smaller tumor size and a 30%-40% reduction in risk of disease relapse or death. PSA immunoreactivity was also found to be inversely associated with markers of active cell proliferation in the breast tissue (Yu et al., 1998). PSA levels in nipple aspirate fluid (NAF) obtained from women with breast cancer were significantly lower than those from women without cancer (Sauter et al., 1996; Sauter et al., 2004). However, two other studies were not able to confirm this negative association between PSA levels and breast cancer progression or risk (Black et al., 2000; Heyl et al., 1999). Although very little is known about possible role(s) o f PSA in breast cancer pathogenesis, one study has suggested that PSA at physiologically relevant concentrations in women may modulate the conversion of the potent estrogen estradiol to the less potent estrogen estrone (Lai et al., 1996). In women, androgen and progesterone are major regulators of PSA production in breast tissue (Yu et al., 1994b; Zarghami et al., 1997a) (Melegos et al., 1997; Negri et al., 2000; Yu et al., 1995a; Zarghami et al., 1997b). Androgen- regulated PSA gene expression is mediated by the binding of the AR-androgen complex to the androgen response elements (AREs) in the PSA gene. At least eight androgen response elements (AREs) have been identified in the PSA gene (Cleutjens et al., 1997b; Cleutjens et al., 1996; Huang et al., 1999; Riegman et al., 1991; Schuur 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. et al., 1996), with the ARE-I having the highest affinity to ligand (Huang et al., 1999). Since the consensus sequence o f the response element for the AR is known to be able to bind to progesterone receptor (PR) and glucocorticoid receptor (GR) and mediate the transcription by PR and GR (Cleutjens et al., 1997a; Lieberman et al., 1993; Roche et al., 1992), it is reasonable to believe that AREs in the PSA gene are also able to mediate the effect of the PR- progesterone complex, Several studies have suggested that the PSA -158A/G polymorphism in the ARE-I may contribute to inter-individual variation in serum PSA levels. In an earlier study, we found that this polymorphism contributed interactively with the AR CAG polymorphism to the variation in serum PSA level in 420 healthy male subjects from a multiethnic cohort. Carriers of the PSA A allele had higher mean serum PSA levels. Among subjects with AA and AG genotypes, longer CAG repeats were associated with lower serum PSA level, but not among subjects with the GG genotype (Xue et al., 2001). In a later study by Medeiros et al, it was also observed that the A allele was associated with elevated pretreatment PSA levels among prostate cancer patients (Medeiros et al., 2002). Given the position of the polymorphism in a high-affmity androgen response element, it has been speculated by Xue et al, that the two alleles probably encode ARE-I sites with different binding affinity to androgen receptor (Xue et al., 2000). However, two recent studies failed to confirm the association between this polymorphism and serum PSA levels in two groups of men without prostate cancer (Rao et al., 2003; Xu et al., 2002), and the only study in females found the A allele to be associated with lower, not higher, PSA 80 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. concentration in breast tumor tissues (Bharaj 2000). Cases with the AA genotype (lower PSA) also had significantly worse overall and disease-free-survival compared to cases with GG or GA genotypes (Bharaj et al., 2000). In the present study, we found no significant association between this polymorphism and breast cancer risk, either overall or in strata defined by family history of breast cancer, age/menopausal status, postmenopausal HRT use, or BMI. One possible explanation for our negative results is that the polymorphism we examined in the PSA gene is not functional, despite the fact that it is located in the ARE-I site. In an in vitro study using PSA gene promoter constructs that differed only by the -158 G/A polymorphism, no difference in PSA gene promoter activity in response to androgen was found (Rao et al., 2003). Although more in vitro studies are needed before any conclusion can be drawn, we speculate that the association of the ARE I polymorphism with serum PSA found in some studies may be a result of other functional polymorphisms in linkage disequilibrium with this -158A/G polymorphism. In a study by Cramer et al in 2003, a group of single nucleotide polymorphisms (SNPs) in the 5’ enhancer region of the PSA gene were associated with serum PSA. Three of them were also found to be functional in reporter gene assays (Cramer et al., 2003). However, the region containing these SNPs did not seem to be critical in stimulating PSA gene transcription in breast cancer cell lines (Schuur et al., 1996), suggesting that this enhancer region might be tissue-specific. More functional studies with breast cancer cell lines may be needed to rule out the 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. importance of this enhancer region (and the three functional SNPs) in AR-regulated PSA transcription. In summary, the strengths of our study include: 1) We focused on a group of women which has not been studied previously for association between the AR CAG repeat polymorphism and breast cancer. As a study of breast cancer in African-American women, our study has a reasonable sample size. The fact that our finding replicated that of the Nurses’ Health Study (of Caucasian women) provides a credible basis for future studies examining this group of women specifically and looking for biological explanations. 2) We explored the possible mechanism of the androgen effect by examining a genetic variation in the PSA gene, a gene which is known to be transcriptionally regulated by AR, together with the AR CAG polymorphism. Although our study does not seem to provide support for the involvement of PSA in breast cancer development, further studies looking at known functional polymorphisms in the PSA gene are still needed to rule out a possible role of PSA in breast cancer. The limitations of our study include: 1) Our sample sizes become very small in stratified analysis, especially in the stratum o f women with a l st-degree family history o f breast cancer, where a positive association between AR CAG genotype and breast cancer was found. We realize that our finding was based on 67 subjects 82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and therefore may need to be taken with caution. The small sample size also precluded us from examining possible AR-PS A gene-gene interaction in this stratum. 2) Due to the fact that AR gene is located on the X chromosome and one X chromosome may be preferentially inactivated in breast cancer cases (Kristiansen et al., 2002), we expect a misclassification of the heterozygotes, maybe especially among cases. Taken together, our finding that there was a significant shift among the cases toward longer CAG repeat lengths in the larger o f the two alleles carried by a woman, but not in the smaller of the two might suggest that the larger allele is preferentially expressed during breast carcinogenesis. However, it is beyond the scope o f our study to address this hypothesis. 3.5. Conclusion Our finding that AR CAG length is associated with breast cancer risk among African-American women with a family history of breast cancer adds to the literature suggesting that androgen protects against breast cancer in certain groups of women. While we could not rule out involvement of the PSA pathway in the subgroup of women with a positive family history o f breast cancer (as we could not examine any possible gene-gene interaction between AR and PSA in this small subgroup), we were unable to provide any evidence that PSA is the pathway through which the 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. protective effect of androgen operates. Other androgen target genes need to investigated. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 4 GENETIC AND HORMONAL DETERMINANTS OF SERUM PSA LEVELS 4.1. Background Prostate-specific antigen (PSA) is a serine protease which in men is predominantly produced by prostatic epithelial cells (Wang et al., 1979). Serum PSA has been widely used as a marker for prostate cancer screening, diagnosis and monitoring of progression. However, prostate cancer does not always result in elevated serum PSA levels. Meanwhile, many conditions other than prostate cancer can also contribute to elevated serum PSA levels, whcih include: 1) age and/or benign prostate hyperplasia (through increased prostate size); 2) prostatitis, prostatic ischemia and infarction (by causing the disruption of normal prostate architecture) (reviewed in (Brawer, 2000)); and 3) hormal and genetic factors involved in PSA gene expression regulation, such as elevated androgen levels, polymorphisms in the androgen receptor gene and the PSA gene. Some o f these factors may influence PSA levels through more than one mechanism. For example, higher androgen levels and stronger androgen receptor activities may contribute to elevated serum PSA levels by both stimulating prostate epithelial cell proliferation (Janulis et al., 2000) and by increasing PSA production in individual prostatic epithelial cell. In addition to its role as a tumor marker, accumulating evidence suggests that PSA may have functional significance in prostate cancer development and progression. PSA itself is able to stimulate the proliferation o f BPH-derived prostatic 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. stromal cells (Sutkowski et al., 1999). PSA is a protease for the major insulin-like growth factor binding protein, IGFBP3. The PSA-cleaved IGFBP3 has a lower affinity binding to insulin-like growth factor-1 and results in an increased bioavailability of mitogenic IGF-1 (Cohen et al., 1994; Sutkowski et al., 1999). PSA is also able to activate latent transforming growth factor-p (TGF- P) (Dallas et al., 2005; Killian et al., 1993) and inactivate parathyroid hormone-related protein (PTFIrP) (Cramer et al., 1996; Iwamura et al., 1996), both of which are thought to be potentially important in bone metastasis o f prostate cancer. Moreover, PSA may promote prostate tumor invasion and metastasis by proteolysis of basement membrane (Webber et al., 1995). On the other hand, PSA can inhibit the proliferation, migration, and invasion of endothelial cells, and block the response of endothelial cells to angiogenic stimulators, suggesting an antiangiogenic activity of PSA (Fortier et al., 1999). Although the exact significance of all these mechanisms in prostate carcinogenesis remains to be determined, these potential biological functions of PSA imply that understanding the regulation of PSA production in normal prostate may be important etiologically. The active PSA protein has 237 amino acids with a molecular weight of approximately 33 kilodaltons (kDa) (Armbruster, 1993). It is encoded by the PSA gene which is located at chromosome 19ql3 (Sutherland et al., 1988). The PSA gene spans about 6 kb and contains 5 exons (Riegman et al., 1989). It is a well- characterized gene which is regulated by androgen at the transcriptional level. Within a region approximately 6 kb upstream from the transcription start site of the 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PSA gene, there are eight androgen response elements (AREs), the cognate sequence that binds to the androgen-androgen receptor (AR) complex and mediates the transcriptional regulatory effect of androgens (Figure 4.1). Two AREs are within several hundred base pairs (bp) from the transcription start site (ARE-I from -170 to -156 and ARE-II, also named androgen response region (ARR), from -394 to -380) (Cleutjens et al., 1996; Riegman et al., 1991) and the third (ARE-III) is located at positions from -4148 to -4134 (Cleutjens et al., 1997b; Schuur et al., 1996). Within a 5’ upstream enhancer region (located between about -3.7 kb and -5.4 kb from the transcription start site) (Schuur et al., 1996), there are five additional AREs (from - 3870 to -4366), namely ARE-IIIB, ARE-IIIA, ARE-IV, ARE-V, and ARE-VI. These five AREs are centered at -3955, -4079, -4179, -4225, and -4298 respectively (Huang et al., 1999). The ARE-I and III are high-affmity binding sites and the ARE II is a low-affmity site (Cleutjens et al., 1997b; Cleutjens et al., 1996); the affinities of other AREs follow the order of: ARE V, ARE IIIA or ARE IV, ARE VI or ARE IIIB (Huang et al., 1999). Distal enhancer (-5322 to -3738) Proximal promoter -4148 t o -4134 -395 t o -376 -170 t o -156 -28 to -23 A R E III A R E II A R E I T A T A +1 / GGAACAtatTGTATC \ GGATCAgggAGTCTC AGAACAgcaAGTGCT - 4 3 6 6 HHHHHHP- 3 8 7 4 AREs VI V IV III IIIA IIIB 'igure 4.1. AREs in the distal enhancer and proximal promoter regions of the PSA gene *The figure is not drawn to scale 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Polymorphisms in the PSA gene promoter region have been reported and studied for possible associations with serum PSA levels. A single nucleotide polymorphism (SNP) in the ARE-I region, a guanine to adenine substitution at the - 158 position (-158 G/A), was first reported by Rao et al (Rao and Cramer, 1999). Because this SNP resides in the most potent androgen receptor binding site, it has been postulated that the two alleles may be functionally different in mediating androgen-regulated PSA gene expression. Our group first studied the possible influence of this SNP on serum PSA levels in 420 healthy male participants of the Multiethnic Cohort Study (men with prostate cancer or with PSA >4 ng/mL were excluded) (Xue et al., 2001). We found that the A allele o f this SNP was associated with a significantly increased serum PSA level. Given the role o f ARE-I in mediating AR-regulated PSA gene expression, we also investigated a possible interaction between this SNP and a functional polymorphism in the AR gene, the CAG repeat length polymorphism in Exon 1. We observed that the AR CAG repeat lengths were negatively associated with serum PSA levels and this association only existed in men carrying at least one A allele (Xue et al., 2001). In a later study by Medeiros et al, pretreatment serum PSA levels in 151 Portuguese men with prostate cancer were also found to be significantly higher among AA carriers (AA vs. AG/GG) (Medeiros et al., 2002). However, three subsequent studies did not observe significant associations between serum PSA levels and either the -158 G/A (Gsur et al., 2002; Rao et al., 2003; Xu et al., 2002) or the AR CAG polymorphism. The study by Xu et al 88 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. included 483 asbestos-exposed workers with no prostate cancer history (91% whites and 9% blacks). They found that the PSA G allele carriers had slightly lower serum PSA levels, but the difference was not significant. CAG lengths were not associated with serum PSA level. However, when limiting the subjects to men with PSA<4 ng/mL, the difference between men with shorter and longer CAG repeats became more evident and was marginally significant (p=0.05) (Xu et al., 2002). In the study by Gsur et al, the subjects were 190 Austrian men with lower urinary tract symptoms due to BPH, but not prostate cancer based on negative digital rectal examination (DRE), negative serum PSA according to age-specific reference values, or based on histological results from ultrasound-guided transrectal prostate biopsies or transurethral resection of the prostate. They found that serum PSA levels were not significantly related to either PSA or CAG repeat genotypes (Gsur et al., 2002). A study by Rao et al included 109 white participants of an annual prostate screening who had normal PSA level (PSA<4 ng/mL), normal DRE, no BPH or prostate cancer history. The mean age-adjusted PSA did not differ by ARE-I genotype. No association between the AR CAG repeat lengths with serum PSA was found, either alone or when stratified by ARE I genotype (Rao et al., 2003). In the only in vitro study comparing the promoter activities of the two alleles at the -158 position, no significant difference was found (Rao et al., 2003). One possible explanation of the observed association between this SNP and serum PSA levels in some studies may be that it is in linkage disequilibrium (LD) with some 89 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. other functional polymorphism(s) in this gene. The variation in the strength of LD in different populations may be a potential reason for the inconsistent results. In an effort to search for the functional polymorphism(s) in the PSA gene, Cramer et al sequenced about a 1.9 kilo-basepair (kb) region (from -3873 to -5749 with respect to the transcription start site) of the 5’ upstream region of the PSA gene (the enhancer region). Eight newly-identified polymorphisms with frequencies greater than 5% (at positions -5567, -5429, -5412, -5307, -5217, -4643, -4289 and - 158) were examined for their possible associations with serum PSA levels in 409 Caucasian men (Cramer et al., 2003) (Figure 4.2). -158 G/A - 3 6 4 7 T / A - 2 9 0 5 C / T -5567 G/A -5307G/A -5217T/A -4643A/G ARE-I ARE-II ARE-III 4-1 -5429 T/G -5412 T/C AREs VI V IV III IIIA IIIB -4330 -4289 C/A CCCCCCCC ccccccccc Figure 4.2. Common polymorphisms in the PSA gene enhancer and promoter regions * The SNPs in bold are the functional ones based on luciferase reporter gene assays (Cramer et al., 2003). The SNPs in italics are the ones in an evolutionarily conserved region (described below). The transcription start site is indicated as +1. The figure is not drawn to scale. In this study, SNPs -4289 A/C (located in the ARE-VI region), -4643 A/G, - 5307G/A, -5412 T/C, -5429 T/G were significantly associated with serum PSA levels. Luciferase reporter gene assays were performed to examine whether the sequence variations at -4289, -4330, -4643, -5429 and -5412 resulted in differences 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in promoter activities. Because -5429 and -5412 SNPs were in very high LD, the two common haplotypes were examined in the reporter gene assay, instead of individual SNPs. Both the -4643 and the -5412/-5429 variants significantly impacted promoter activities in response to androgen, while the -4289 and -4330 variants did not. The - 158 G/A SNP was in LD with the three functional SNPs (with the A allele exclusively linked to the alleles with lower activities at the -5429, -5412 and -4643 loci) and also appeared to be associated with serum PSA levels (p=0.06 when comparing the heterozygotes and homozygotes of the A allele to homozygotes of the G allele). Very interestingly, for almost all the above mentioned SNPs (with the exception of -5307 G/A), it was always the heterozygotes who had the highest PSA levels, which is difficult to explain biologically. Despite the long-recognized importance of androgens in regulating PSA gene expression, there is a lack of studies on the correlation between serum androgens and serum PSA levels. Studies have demonstrated that exogenous androgen usage increases serum PSA levels in older men (57-75 years old) (Tenover, 1992), transsexual women (Obiezu et al., 2000) or hypogonadal men (Gerstenbluth et al., 2002) and conversely, anti-androgen therapy o f BPH results in decreased PSA concentration in men (Gormley et al., 1994; McConnell et al., 1992; Stone and Clejan, 1991; Weber et al., 1989). However, it has also been observed that administration o f testosterone to healthy young males did not increase the concentrations of serum PSA (Cooper et al., 1998). In a study o f 150 men aged 41- 79 from an impotency clinic, serum testosterone levels within the normal range were 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. not associated with serum PSA levels (Monath et al., 1995). Another study has suggested that the correlation between serum testosterone levels and PSA levels existed among subfertile men, but not among normal men (Mifsud et al., 2001a). Similarly, studies on serum PSA levels and the functional CAG repeat polymorphism in the AR gene have not been consistent. As mentioned earlier, out of 4 studies examining the AR CAG polymorphism and the PSA -158 G/A polymorphism together, only one found a significant negative association between the CAG repeat lengths and PSA levels (Xue et al., 2001). However, this association was found by two other studies examining PSA levels and the AR CAG repeats alone. In the study conducted in Singapore, a negative correlation between serum PSA and CAG repeats was found in a group o f 34-36 years old subfertile men (112 subjects with azoospermia or oligozoospermia, recruited from infertility clinics), but not in a group of fertile men (90 subjects with proven fertility) (Mifsud et al., 2001a). In the other study of 274 Swedish males aged 18-21, it was also found that longer CAG repeat lengths were significantly associated with decreased PSA levels in semen samples (Giwercman et al., 2004). The inconsistency is difficult to explain. In addition to the facts that the association was only found among subfertile men in the study by Mifsud and in the study by Giwercman, semen samples rather than serum samples were used, it is also noteworthy that subjects in these two studies were much younger than most subjects in the other four studies. In summary, although previous literature has suggested that androgens may work interactively with the polymorphisms in the AR and the PSA gene to affect 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PSA levels, the results have been conflicting and there is a lack o f studies with good design that incorporate all these factors and look at them systematically. Some of the previous studies looked at very special groups of men (asbestos-exposed workers, men with lower urinary tract symptoms, men with impotency or infertility). Most of the previous studies had relatively small sample sizes and had only one racial/ethnic group. It is noteworthy that in the study by Rao et al, the genotypes o f -158 G/A were not in Hardy-Weinberg Equilibrium and the AR CAG genotyping failed in a large proportion of study subjects, hence measurement error could not be ruled out. In the present study, we included men from a general population. More specifically, the subjects were 643 participants of two existing population-based case-control studies. It has a relatively large sample size and two racial groups. We investigated possible effects o f the newly identified polymorphisms in the distal enhancer region o f the PSA gene (some o f demonstrated functional significance based on in vitro study results) on serum PSA levels. In addition, a close examination of the 5’ regulatory region revealed that in the gap between the enhancer region and the proximal promoter region, there is about 1.1 kb o f sequence highly conserved across species (i.e. the evolutionarily conserved region, ECR). Because of the potential functional importance of ECRs, we screened this region for new polymorphism(s). Common polymorphisms in this region were also studied for their possible association with serum PSA levels. We examined the polymorphisms in the PSA gene systematically with serum androgen levels and with a functional polymorphism in a PSA upstream gene, the CAG repeats in the AR gene. Our study 93 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. objective was to identify potentially important polymorphisms in the PSA gene (and to investigate how they interact with androgens and the AR polymorphism) for future studies of this gene and prostate cancer risk. 4.2. Study design and methods 4.2.1. Study subjects The study subjects were: 1) 256 controls from a population-based case- control study of advanced prostate cancer conducted in the San Francisco Bay Area conducted by Dr. Esther M. John. 2) 387 controls from a population-based case- control study of advanced prostate cancer conducted in the Los Angeles Area by Dr. Sue Ingles. The study subjects from San Francisco Area In the study conducted in the San Francisco Bay Area, controls aged 40 to 79 were identified through random-digit dialing (RDD). In addition, the identification of men age 65-79 was supplemented by random selections from the rosters of beneficiaries of the Health Care Financing Administration (HCFA). The eligibility criteria for the controls included: 1) aged 40-79 years; 2) with no prior history of prostate cancer; 3) residing in the San Francisco Bay area at the time of selection into the study and at the time of contact for in-person interview; 4) African-American or non-Hispanic whites based on self-identification; 5) English-speaking; 6) mentally competent to participate. Through RDD and HCFA rosters 1,081 controls were selected into the study (717 through RDD, 364 HCFA). O f these, 123 (11.4%) did not meet the eligibility 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. criteria. For 90 HCFA men, phone numbers were not located. O f the remaining 868 eligible controls with phone numbers, 545 (63%) completed a detailed in-person interview. Among these men, 525 (96%) provided a biospecimen, including 256 fasting blood samples and 269 mouthwash samples. These 256 men (219 non- Hispanic whites and 37 African-Americans) with blood samples available were included in the phenotype-genotype correlation study. The study subjects from Los Angeles Area In the study conducted in the Los Angeles Area, neighborhood controls were identified using an algorithm developed by a centralized unit for identifying neighborhood controls at the University o f Southern California. Because the parent study was still ongoing at the time when the present study was carried out, we only included a representative sample of controls from the parent study, i.e. 387 controls (including 238 non-Hispanic whites, 107 African-Americans and 42 Hispanics) with questionnaire and blood sample available by January 2004. 4.2.2. Serum PSA and hormone levels Fasting blood samples were drawn in the morning between 7am and 12pm. Blood samples were stored at 4°C until the serum was separated, which was within at most 6 hours. The serum was promptly frozen and stored at -70°C until it was thawed for the hormone analyses. Serum levels of testosterone (T) and 3a- androstanediol glucuronide (3a-diol G), sex hormone binding globulin (SHBG) and PSA were quantified by immunoassay methods in the Reproductive Endocrine Research Laboratory of the Department o f Obstetrics and Gynecology at University 95 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of Southern California. T was extracted from serum using ethyl acetate: hexane (3:2) and, following evaporation of the organic solvents, the extract was re-dissolved in isooctane and applied on a Celite partition chromatography column; the stationary phase was ethylene glycol. T was eluted with 40% (v/v) toluene in isooctane. After evaporating the organic solvents, the residue was reconstituted in assay buffer and the analyte was quantified by radioimmunoassay (RIA) (Goebelsmann et al., 1979). The T RIA utilized an antiserum against T in conjunction with an iodinated T derivative as the radioligand. After an appropriate incubation period, separation of antibody-bound T was achieved by the use of a second antibody. After centrifugation, the antibody-bound fraction was counted in the gamma counter. 3 a- Diol G was quantified by direct RIA using a commercial kit (Diagnostic Systems Laboratories, Webster, TX) (Narang et al., 1991). A highly specific antiserum was used in the assay in conjunction with iodinated 3a-diol G. After the appropriate reagents were mixed and incubated, the antibody-bound fraction was separated from the unbound fraction by a second antibody procedure. The bound fraction was counted in a gamma counter. SHBG and PSA were analyzed by direct immunochemiluminometric assay methods using the hnmulite analyzer (Diagnostic Products Corporation, Inglewood, CA). Free T was calculated using the measured T and SHBG concentrations, as well as an assumed average concentration for albumin (Goebelsmann et al., 1974; Probst-Hensch et al., 1999). This method was found to have high validity compared with direct measurements for free T (Sodergard et al., 1982). Interassay coefficients of variation (CV) for measurements of the different 96 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. analytes ranged from 5.0% to 11.0%.In addition to the internal quality controls provided with the kits, 27 aliquots of a pooled human serum were distributed randomly among the study samples (with the lab technicians blind to the identity of these pool serum samples). For these 27 samples, the interassay CV for the measurements o f all analytes ranged from 3.7% to 8.8%. 4.2.3. Resequencing of the ECR The sequences in a region from about -3.7 kb to -2.6 kb with respect to the transcription start site were highly conserved across species (compared to rats and mice, using 50 bp, 80% criteria). I screened this region for polymorphisms by direct sequencing using 60 control samples in our lab. The primer sequences were: 5’- CTGCCAAACCCAGAATAAGG-3 ’ (forward) and 5’-TCCAATCTGATCCTCCA TCC-3’ (reverse). DNA sequencing was performed with the use of BigDye Terminator sequencing kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions. Sequencing products were ethanol precipitated, air dried, resuspended in 10 uls of formamide and run on a 3700 DNA analyzer (Applied Biosystems, Foster City, CA). DNA sequence data were analyzed using the ABI Sequence Analysis software (Version 3.7). Two common polymorphisms were identified in this region, -3647T/A and - 2905C/T, and were genotyped by Taqman assay. 4.2.4. Genotyping methods Direct sequencing 97 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The polymorphisms in the region from nucleotide positions -5081 to -5723 were genotyped by direct sequencing. The primer sequences were: 5’-CAACTG AGCCT TGATGTCCA-3’ (forward) and 5 ’-CAAAGACCTCAGG CGTTTCC-3’. An approximately 330 bp region surrounding the -4289 A/C and -4336 8C/9C polymorphisms was also sequenced. The primers were: 5’-AGCAGACAGATGAG GTTCA-3’ (forward) and 5’-TCTCAGATCCAGGCTTGCTT-3’ (reverse). DNA sequencing was performed as described above. Taqman assay The SNPs -158 G/A, -2905 C/T, -3647 T/A, and -4643 A/G were genotyped using the 5’ nuclease assay (Taqman assay). For the -158 G/A, the primer sequence were 5'-GGTGCATCCAGGGTGATCTAG-3' (forward) and 5'-CACACCCAGAG CTGTGGAAG-3' (reverse). The two probes were: 5'-FAM-CAGAACAGCAAG TACTAGCTCTCCCTC-3' (A allele) and 5'- CY3- AGAACAGCAAGTGCTAG CTCTCCC-3' (G allele). For the -2905 C/T, the two primers were: 5’-GTCTATTG CTCTCCCAAGTGAGTCT-3’ (forward) and 5’-CCCCAGTTCCTTCAC-3’ (reverse). The two probes were: 5’-FAM-CCAGATACGAGGCAC-3’ (C allele) and 5’-VIC-CCAGATATGAGGCACT-3 ’ (T allele). For the -3647 T/A, the primers were: 5’-GGTCCACCCTCTTGAATTTCAA-3’ (forward) and 5’-CCTAAGAGT TCGGAACTGACACTTT-3’ (reverse). The probes were: 5’-FAM-CCCCAGTTC CTTCAC-3’ (A allele) and 5’-VIC-CCCCAGTACCTTCAC-3’ (T allele). For the - 4643 A/G, the primer sequences were: 5’-CCTCTCTTTTAGGGCTCTTTCTGA-3’ (forward) and 5 ’-CATAGAGTCAAGAGGGTACAGAATACAATG-3 ’ (reverse). 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The probes sequences were: 5’-6FAM-CCACCATGATACTAGGAC-3’ (A allele) and 5 ’-VIC-CACCATGGTACTAGGA-3’ (G allele). The polymorphic sites are underlined. The PCR reactions were performed in a 15 pi volume consisting of 25 ng of genomic DNA. Taqman Core Reagent Kit (Applied Biosystems, Foster City, CA) was used according to the manufacturer’s instructions. The thermal cyclings were performed in MicroAmp Optical plates (Applied Biosystems, Foster City, CA). The fluorescence signal was detected using an ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Foster City, CA). Genotypes were determined using the graphical view from ABI Sequence Detection software (version 2.1). Genescan Simple sequence length polymorphism (SSLP) analysis was used to identify the AR exon 1 CAG repeat variant. The genomic region containing the CAG repeat was PCR amplified using the forward primer 5’-CGCGAAGTGATCCAGAAC-3’ and the reverse primer 5 ’ -CAGGACCAGGTAGCCTGTG-3 ’. The PCR reaction was performed in a 20 pi volume consisting o f 30 ng o f genomic DNA. Taqman Core Reagent Kit (Applied Biosystems, Foster City, CA) was used according to the manufacturer’s instructions, in addition to 6 pmoles of unlabeled forward and reverse PCR primer, plus 0.4 pmol of FAM-labeled reverse primer. Touchdown thermal cycling was performed. The resulting PCR product was mixed with size standard HD-400 (Applied Biosystems, Foster City, CA) and was run on the ABI 3700 capillary sequencer. The allele sizes were scored using GeneScan software (version 3.5) (Applied Biosystems, Foster City, CA). DNA samples from 16 samples with 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. various known CAG repeat lengths (determined from direct sequencing) were included in each run as controls. A standard curve was drawn based on these 12 control samples and was used to calculate CAG repeat number for study subjects. General quality control All PCR assays included a “water blank” to guard against contamination of the PCR reaction. Genotyping was repeated for 3%-5% randomly-picked samples. Results from these repeats showed a concordance of 100%. Direct sequencing was performed if the genotyping results were uncertain. 4.2.5. Statistical analysis Because the distribution o f serum PSA levels significantly deviated from a normal distribution, a natural logarithm transformation was performed on the PSA levels and was used in linear regression models. The transformation made the distribution approach normality, although it remained significantly deviated from a normal distribution (p value for Skewness test=0.99, p value for Kurtosis test=0.004). Student t tests were used to compare the two racial groups (non-Hispanic whites and African-Americans) with respect to mean age, BMI and waist circumference. Wilcoxon rank sum tests were used to compare the racial groups and the two study centers with respect to PSA, serum total T, free T, 3a-diol G, and SHBG (variables that are not normally distributed). A chi-square test was used to compare frequencies of BPH, prostatitis or the PSA polymorphisms between two 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. centers or two racial groups. In the cases where 20% of the cells had expected frequencies less than or equal to 5, an exact test was used instead. General linear regression models were fit to estimate the effects o f age, serum total T, free T, 3a-diol G and SHBG on serum PSA levels. Age, serum total T, free T, 3a-diol G and SHBG were categorized by quartiles (based on the distributions in non-Hispanics whites and African-Americans combined) and least squares means of PSA on the log scale were estimated for each quartile group, after adjusting for age, center and other potential confounders. Trend tests were performed with the quartile numbers 1, 2, 3, 4 in the regression model as continuous variables. For all the common polymorphisms in the PSA gene, Hardy-Weinberg • 2 equilibrium was tested using the Pearson goodness-of-fit statistic with a % distribution with ldf. General linear regression models were used to examine the effect of the genotypes by comparing the heterozygotes and homozygotes o f the less common allele respectively to baseline group, the homozygotes o f the more common allele (co-dominant effect models). Trend tests were also performed to examine the additive effect o f the sequence variants by coding the three genotypes as 0,1,2 variables (representing the copies of the less common allele) in the regression model. In the cases where homozygotes of the less common allele were very rare, this group was collapsed with the heterozygote group. The maximum likelihood estimates of the pairwise linkage disequilibrium (LD) measurements, Levwontin’s D ’ and R2, were obtained by Haploview software (Barrett et al., 2005). The EM (expectation-maximization) algorithm was used to 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. estimate haplotype phase and population frequencies from genotype data within our study population, as implemented in Haploview (Barrett et al., 2005). In haplotype analyses, an estimation of the number of copies of each haplotype carried by each individual was obtained from the TagSNP software and used in regression models as a continuous variable (Stram et al., 2003). In the analyses of the AR CAG repeat polymorphism, because a few observations with extremely long or extremely short CAG repeat lengths were highly influential in regression analyses, CAG repeats were grouped into approximate quintiles. Quintile 1 corresponded tol2-17 repeats, 2 corresponded to 18-19 repeats, 3 corresponded to 20 repeats, 4 corresponded to 21-22 repeats, and 5 corresponded to 23-34 repeats. Trend tests were performed by using the medians of the quintile groups: 17, 19, 20, 21, 24 as scores for coding CAG repeat lengths in the regression models. Analyses using deciles were also performed. Age and study center were adjusted by including these variables in regression models. We considered age and BPH as potential effect modifiers. Interactions between serum hormone levels and the PSA and AR genotypes were also examined. Formal tests of interaction were performed by including the appropriate interaction terms in the regression model.F tests were used to compare models that were nested, Akaike information criterion (AIC) was used to compare non-nested models. Analyses were performed with the use of ST AT A and SAS software. All reported p-values were 2-sided and statistical significance was taken as p value <0.05. 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3. Results 4.3.1. Characteristics of the study subjects Because this study was to examine factors that may be important in determining serum PSA levels in normal men, we excluded twenty-seven subjects with serum PSA level greater than or equal to lOng/mL. In addition, two subjects were excluded because they had testosterone levels greatly above the normal range (repeating the assay ruled out laboratory error, and thus these values could be a result of exogenous androgen usage). One subject was excluded because of missing hormone levels. To allow possible comparison between genotype analyses and haplotype analyses, eight subjects were exclude due to missing genotyping results for one or more common PSA SNPs. Due to the small number o f Hispanics (42), they were excluded from data analyses. Therefore, subjects included in final dataset were 428 non-Hispanic whites (whites hereinafter) and 128 African-Americans. Age, frequencies of benign prostate hyperplasia (BPH) and prostatitis, serum hormone and serum PSA levels in the study subjects are reported in Table 4.1. Non- Hispanic whites and African-Americans from the Northern California study center and the Southern California study center were similar with respect to all these factors except for serum SHBG levels (and as a result free testosterone levels). The reason for this center-difference is sill under investigation. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.1. Basic characteristics, serum androgen and PSA levels, by race and study center AA from SF AA from LA * P Whites from SF White from LA P n 30 98 202 226 Age Mean (SD) 60.9 (6.5) 61.8(9.6) 64.7 (8.0) 63.5 (10.7) Median (IQR) 62 (57-65) 62.5 (55-68) 0.33 65 (59-70) 63 (55-71) 0.11 History of BPH yes 2(7.1%) 18(18.6%) 61 (30.5%) 58 (26.0%) no 26 (92.9%) 79 (81.4%) 0.24 139 (69.5%) 165 (74.0%) 0.31 History o f prostatitis yes 3 (10.0%) 1 (1.0%) 23 (11.4%) 22 (9.8%) no 27 (90.0%) 96 (99.0%) 0.04f 179 (88.6%) 203 (90.2%) 0.59 PSA (ng/mL) Median (IQR) 1.1 (0.6-2.3) 1.1 (0.7-2.0) 0.67 1.5 (0.8-2.8) 1.1 (0.6-2.4) 0.18 Total T (ng/dL) Median (IQR) 471.5 (359.0-627.0) 524.5 (411.0-649.0) 0.41 470.0 (360.0-581.0) 478.0 (366.0-591.0) 0.36 Free T (pg/mL) Median (IQR) 91.0 (61.0-130.0) 114.0 (87.5-142.4) <0.001 89.0 (66.0-111.0) 105.6 (82.5-136.1) 0.04 3 a-AG (ng/mL) Median (IQR) 4.9 (4.1-7.4) 5.1 (3.7-7.3) 0.27 5.3 (3.6-7.3) 5.5 (3.9-7.6) 0.98 SHBG (nmol/L) Median (IQR) 39.2 (34.4-60.0) 35.5 (28.1-47.7) <0.001 43.3 (33.2-59.9) 35.2 (28.0-46.2) 0.06 AA: African-Americans; SF: San-Francisco Bay Area; LA: Los-Angeles Area *P values were from non-parametric tests (wilcoxon rank-sum test) for continuous variables, and from Chi- square tests for dichotomous variables. f p value from exact tests The race-specific frequencies of the PSA gene polymorphisms (Table 4.2) and the distributions of the CAG repeat lengths (not shown) were very similar between subjects from the two study centers. Study subjects from these two centers were therefore combined in later data analyses. To control for possible confounding by study centers, a “center” variable was included in all regression models. The general characteristics of the combined study subjects are presented in Table 4.3. Whites were approximately equally distributed between the two study centers, while most African-Americans (76.5%) were from the Southern California Center. Similar proportions of whites and African-Americans had elevated PSA levels (>4ng/mL) 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (p=0.83). Whites and African-Americans had similar distributions of BMI and waist circumferences (p=0.85 and p=0.95). However, whites were significantly older than African-Americans (p=0.01) and they had significantly higher frequencies of self- reported BPH and prostatitis (p=0.01 and p=0.01). T able 4 .2 . R ace-sp ecific allele frequencies o f the P S A gen e p olym orphism , b y study center A A from A A from W hites from W hites from SF LA SF LA N 30 98 p value 202 226 p value -5567 0% 2.6% 0.59' 9.2% 6.4% 0.13 -5429 10.0% 9.7% 0.94 25.0% 26.5 % 0.61 -5413 10.0% 9.7% 0.94 24.8% 27.5% 0.70 -5307 43.3% 40.8% 0.73 30.7% 26.5% 0.18 -5217 16.7% 10.7% 0.22 5.9% 8.0% 0.25 -4643 8.3% 8.7% 0.93 21.5% 24.1% 0.37 -4336 25% 32.1% 0.29 42.8% 42.0% 0.82 -4289 8.3% 8.7% 0.93 21.8% 24.3% 0.38 -3647 21.7% 19.4% 0.70 12.9% 16.2% 0.18 -2905 36.7% 37.8% 0.88 33.0% 29.0% 0.16 -158 40.0% 53.1% 0.08 51.7% 45.4% 0.06 Fisher’s exact test T able 4.3. General characteristics o f the study subjects W hites A frican-A m ericans (n = 428) (n = 128) P S A > 4n g/m l 5 7 (1 3 .3 % ) 1 8 (1 4 .1 % ) P =0.83 A g e 64.0 (9.6)* 6 1 .6 (9 .0 )* P=0.01 B M I (kg/m 2) 28.5 (5.31)* 2 8 .6 (5.38)* P =0.85 W aist circum ference (cm ) 99.7 (1 2 .0 ) * 9 9 .7 (1 2 .6 )* P =0.95 H istory o f B P H 1 1 9 (2 8 .1 % ) 20(16.0% ) P=0.01 H istory o f prostatitis 4 5 (1 0 .5 % ) 4 (3.2% ) P=0.01 Study center u s e 2 2 6 (52.8% ) 98 (76.6% ) N C C C * __L \ 202 (47.2% ) 30 (23.4% ) mean (SD) 4.3.2. Racial differences in serum PSA, androgen, and SHBG levels African-Americans had higher total T and free T levels than whites (p=0.02 and p<0.01). However, the differences were no longer significant after adjusting for age and study center. Further adjustment for BPH, prostatitis did not change the 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. results. There was no significant difference in serum PSA, 3a-diol G or SHBG levels between the two ethnic groups before or after adjusting for age, study center, BPH and prostatitis (Table 4.4). BMI was not a confounder in the comparison of serum PSA and androgen measurements between two racial/ethnic groups (data not shown). T able 4.4. Serum P SA , androgen and SH B G le v els b y race W hites African-Am ericans PSA (ng/mL) Crude means (95% Cl) Adjusted mean* (95% Cl) Adjusted mean+ (95% Cl) 1.31 (1.20-1.43) 1.28 (1.18-1.39) 1.38 (1.21-1.59) 1.22 (1.04-1.42) 1.30(1.11-1.52) 1.48 (1.20-1.82) P=0.44 P=0.88 P=0.46 Total T (ng/dL) Crude means (95% Cl) Adjusted mean* (95% Cl) Adjusted mean+ (95% C l) 478.4 (462.5-494.6) 480.3 (464.4-496.5) 4 8 1 .6 (4 5 5 .9 -5 0 8 .2 ) 520.1 (486.8-554.7) 507.3 (476.5-539.2) 502.3 (469.8-550.5) P=0.02 P=0.13 P=0.13 Free T (pg/mL) Crude means (95% Cl) Adjusted mean* (95% Cl) Adjusted mean+ (95% Cl) 97.3 (93.4-101.2) 9 7 .7 (9 4 .4 -1 0 1 .2 ) 100.1 (94.5-105.9) 109 .2 (1 0 1 .7 -1 1 7 .1 ) 100.2 (93.8-106.9) 103.2 (94.8-112.1) P O .0 1 P=0.52 P=0.43 3 a-A G (ng/mL) Crude means (95% Cl) Adjusted mean* (95% Cl) Adjusted m ean+ (95% C l) 5.33 (5.05-5.62) 5.35 (5.08-5.63) 5 .1 0 (4 .6 9 -5 .5 5 ) 5.21 (4.75-5.72) 5.02 (4.55-5.53) 4.78 (4.22-5.42) P=0.69 P=0.29 P=0.26 SH BG (nmol/L) Crude means (95% Cl) Adjusted mean* (95% Cl) Adjusted mean^ (95% Cl) 39.4 (37.7-41.2) 39.4(37.8-41.1) 3 8 .6 (3 6 .1 -4 1 .3 ) 38.3 (35.4-41.4) 41.4 (38.2-44.7) 40.3 (36.4-44.6) P=0.52 P=0.29 P=0.36 f------------ ; ------------------------------------------------------------------------------------------------------ geometric means, after adjusting for age (continuous) and study center. ^ further adjusting for history o f BPH (yes or no) and history o f prostatitis (yes or no) 4.3.3. Serum PSA levels and self-reported history of BPH or prostatitis Men with self-reported BPH had significantly higher serum PSA levels compared to those without (p=0.005, after adjusting for age, center and race). PSA levels were slightly higher in men with self-reported prostatitis than in men without, but the difference was not statistically significant (the adjusted p=0.30) (Table 4.5). Serum androgen (total T, free T and 3a-diol G) and SHBG were not significantly 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. associated with BPH or prostatitis (data not shown). Adjusting for BMI did not change the results appreciably (data not shown). T able 4.5 Serum P S A le v e ls b y self-reported B PH and prostatitis history N* M edian (IQR) Crude geometric means (95% Cl) Adjusted geometric m eans (95% CI)+ BPH yes 139 1.8 (1.0-3.9) 1.81 (1.55-2.11) 1.58 (1.35-1.87) N o 409 1.1 (0.6-2.0) 1.14(1.05-1.24) 1.23 (1.11-1.35) P 0 .0 0 1 PO.OOl P=0.005 prostatitis Y es 49 1.6 (0.6-3.1) 1.47(1.08-2.00) 1.48 (1.15-1.91) N o 505 1.2 (0.7-2.4) 1.27 (1.18-1.38) 1 .2 9 (1 .1 8 -1 .4 1 ) P=0.29 P=0.29 P =0.30 * 8 subjects with BPH information missing, 2 subjects with prostatitis information missing * adjusting for age, study center and race. 4.3.4. Age-related change in serum PSA levels There was a significant increase in serum PSA with increasing age (Table 4.6). In the highest age quartile, the geometric mean PSA was 1.14 times higher (1.93 ng/mL) than in the lowest quartile (0.90 ng/mL). The age-related increase in serum PSA levels did not differ significantly by race (p value for interaction was 0.51 when comparing the slopes for the two races). 4.3.5. Serum PSA levels and serum androgen and SHBG levels There was no statistically significant association between serum PSA levels with total T levels or free T levels. However, we observed a significant increase in PSA levels associated with increasing serum 3a-diol G levels (p=0.01). We also observed a significant negative association between PSA levels and serum SHBG levels (p=0.04) (Table 4.7). Neither the positive association between PSA and 3a- diol G nor the negative association between PSA and SHBG differed by race (p for interaction= 0.43 for 3a-diol G; p for interaction= 0.89 for SHBG). 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. T able 4.6. A ge-related ch an ges in serum P S A le v els Overall W hites African-Am erican A ge N InPSA Geometric N InPSA Geometric N InPSA Geometric (yr) (556) lsm ean (SE) m ean (428) lsm ean (SE)* mean (128) lsm ean (SE)* mean <58 58-64 65-70 71+ P (SE) 156 143 124 133 -0 .1 5 (0 .0 7 ) 0.11 (0.08) 0.36 (0.08) 0.64 (0.08) 0.26 (0.03) ptrdO .001* 0.90 1.16 1.47 1.93 118 107 90 113 -0.16 (0.08) 0.14 (0.09) 0.34 (0.10) 0.67 (0.09) 0.27 (0.04) p trdO.OOl* 0.89 1.19 1.44 1.99 38 36 34 20 -0.11 (0.16) 0.01 (0.15) 0 .3 9 (0 .1 6 ) 0.45 (0.21) 0.21 (0.07) P trd=0.005* 0.94 1.05 1.52 1.60 lsmean (SE): least-square mean (standard error) adjusting for center and race. Further adjusting for BMI and BPH did not change the results. + adjusting for center * trend cross age groups o oo Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.7. Serum PSA levels and serum androgen, SHBG levels all subjects whites African-Americans N InPSA lsmean (SE)* Geometric mean N InPSA lsmean (SE)+ Geometric mean N InPSA lsmean (SE)+ Geometric mean • Total T Q1 <372.0 140 0.21 (0.08) 1.27 112 0.24 (0.09) 1.31 28 0.04(0.17) 1.08 (ng/dL) Q2 372.0-482.5 138 0.30 (0.08) 1.39 113 0.32 (0.09) 1.42 25 0.16(0.18) 1.21 Q3 482.6-608.0 140 0.17(0.08) 1.22 107 0.20 (0.09) 1.26 33 0.03 (0.17) 1.06 Q4 608.0+ P (SE) 138 0.23 (0.08) -0.007 (0.03) p trd*=0.83 1.30 96 0.18(0.09) -0.032 (0.04) p trdf=0.44 1.23 42 0.28(0. 15) 0.065 (0.07) P trd*=0.34 1.36 FreeT Q1 <76.0 141 0.16(0.08) 1.21 117 0.20 (0.09) 1.25 24 -0.03 (0.18) 1.01 (pg/mL) Q2 76.0-100.0 138 0.24 (0.08) 1.30 113 0.26 (0.09) 1.34 25 0.08 (0.19) 1.12 Q3 100.1-128.2 138 0.24 (0.08) 1.31 99 0.25 (0.09) 1.33 39 0.14(0.16) 1.18 Q4 >128.2 P (SE) 139 0.28 (0.08) 01037 (0.04) p trd*=0.33 1.36 99 0.25 (0.10) 0.014(0.04) p trd{=0.75 1.32 40 0.32(0.15) 0.113(0.08) Ptrd*=0.14 1.41 3a-diol G Q1 <3.80 141 0.07 (0.08) 1.11 107 0.09 (0.09) 1.13 34 -0.06(0.16) 0.98 (ng/mL) Q2 3.80-5.35 137 0.18(0.08) 1.24 104 0.22 (0.09) 1.29 33 0.01 (0.16) 1.05 Q3 5.36-7.45 139 0.35 (0.08) 1.45 108 0.32 (0.09) 1.42 31 0.37 (0.17) 1.48 Q4 >7.45 P (SE) 139 0.32 (0.08) 0.092 (0.03) p trd}=0.007 1.42 109 0.32 (0.09) 0.079 (0.04) p trdt=0.04 1.41 30 0.28 (0.17) 0.139(0.07) P trd*=0.06 1.37 SHBG Q1 <29.0 142 0.31 (0.08) 1.41 110 0.33 (0.09) 1.43 32 0.21 (0.17) 1.27 (nmol/L) Q2 29.0-38.4 138 0.30 (0.08) 1.38 99 0.29 (0.09) 1.38 39 0.25(0.15) 1.32 Q3 38.5-52.5 138 0.21 (0.08) 1.27 110 0.25 (0.09) 1.32 28 0.03(0.17) 1.06 Q4 >52.5 P (SE) 138 0.10(0.08) -0.071 (0.04) p trd*=0.05 1.15 109 0.10 (0.09) -0.073 (0.04) p trd*=0.07 1.14 29 0.06(0.17) -0.065 (0.07) P trd*=0.38 1.10 adjusting for age, center and race. Further adjusting for BMI and BPH did not change the results. * adjusting for age and center * trend cross the quartile groups 4.3.6. Serum PSA levels and the PSA gene polymorphisms Common polymorphisms (with minor allele frequency greater than 5% in Whites or greater than 8% in African-Americans) in the PSA enhancer/promoter region are listed in Table 4.8, and rare polymorphisms are listed in Table 4.9. New polymorphisms identified in the present study are highlighted, including the two SNPs (-3647 T/A and -2905 C/T) located in the ECR. We noticed that some SNP locations were incorrectly reported in the paper by Cramer et al, for example, polymorphism -5413 T to C was reported as -5412 T to C, and -4336 (8 or 9 cytosines) was reported as -4330 (8 or 9 cytosine) previously. Allele frequencies in whites in this study were comparable to those reported by Cramer et al, except for the SNP -4289 C/A (the minor allele was the C allele in our study, not the A allele as reported in the previous study). The rare SNPs listed in Table 4.9 are not included in the analyses for the possible associations with serum PSA levels. For all the common SNPs, the genotype distributions did not statistically significantly deviate from HWE in either racial group (data not shown). The associations between serum PSA levels and the common polymorphisms in the PSA gene are reported in Table 4.10 for whites and Table 4.11 for African- Americans respectively. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.8. Allele frequencies of the common polymorphisms in the PSA gene Polymorphisms’ whites (n=428) African-American (n=128) -5567 G to A 7.7% 2.0% -5429 T to G 25.8% 9.8% -5413 T to C 25.4% 9.8% -5307 G to A 28.5% 41.4% -5217 T to A 7.0% 12.1% -4643 A to G 22.9% 8.6% -4336 8C to 9C 42.4% 30.5% -4289 C to A 23.1% 8.6% -3647 T to A 14.6% 19.9% -2905 C to T 31.1% 37.5% -158 G to A * 48.4% 50.0% Polymorphisms in bold italics are the new polymorphisms identified in the present study Table 4.9. Allele frequencies of some rare polymorphisms in the PSA gene ~ " “ ; 5 " j " ~ " . Polymorphisms’ whites n=428 African-American n=128 -5557A to G 0.7% 5.5% -5546 C to T 0% 0.4% -5512 T to C 0.1% 0.4% -5501 T to C 1.3% 0.4% -5500 GtoC 0% 5.9 % -5498 G to A 0 % 0.4% -5485 G to A 0.8% 0% -5480 G to A 0.1% 0% -5479 G to A 0.1% 0.8% -5466 G to A 2.7% 0% -5457 C to T 0.1% 0% -5411 CtoT 0.2% 0% -5380 G to T 0% 5.5% -5336 T to C * 0.1% 0% Polymorphisms in bold italics are the new polymorphisms identified in the present study Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.10. Serum PSA levels and common polymorphisms in the PSA gene among whites Genotypes n InPSA (lsmean) InPSA (SE) Geometric mean (ng/mL) P t P p for trend -5567 GG 365 0.23 0.05 1.30 AG 60 0.27 0.12 1.35 0.72 AA 3 1.12 0.52 3.10 0.22 0.09 0.31 -5429 TT 232 0.18 0.06 1.24 TG 171 0.23 0.07 1.30 0.57 GG 25 0.85 0.18 2.38 0.002 <0.001 0.01 -5413 TT 236 0.17 0.06 1.22 TC 167 0.24 0.07 1.31 0.47 CC 25 0.85 0.18 2.38 0.002 <0.001 0.007 -5307 GG 217 0.26 0.06 1.33 AG 178 0.18 0.07 1.24 0.37 AA 33 0.36 0.16 1.47 0.49 0.57 0.89 -5217 TT 369 0.25 0.05 1.32 AT 58 0.19 0.12 1.25 0.67 AA 1 -0.05 0.91 0.99 0.87 0.74 0.62 AT or AA 59 0.19 0.12 1.25 0.64 -4643 AA 252 0.18 0.06 1.24 AG 156 0.25 0.07 1.32 0.46 GG 20 0.89 0.20 2.47 0.003 <0.001 0.01 -4336 88 149 0.37 0.07 1.49 89 195 0.14 0.06 1.19 0.02 99 84 0.24 0.10 1.31 0.05 0.27 0.13 -4289 CC 250 0.18 0.06 1.24 (ARE IV) AC 158 0.25 0.07 1.32 0.43 AA 20 0.89 0.20 2.47 0.003 <0.001 0.01 -3647 TT 312 0.31 0.05 1.40 AT 107 0.06 0.09 1.10 0.01 AA 9 -0.24 0.30 0.82 0.01 0.07 0.003 -2905 CC 201 0.25 0.06 1.32 CT 188 0.19 0.06 1.25 0.53 TT 39 0.38 0.14 1.50 0.48 0.41 0.80 -158 GG 115 0.26 0.08 1.33 (ARE I) AG 212 0.21 0.06 1.27 0.64 * AA 101 0.28 0.09 1.36 + 0.76 0.82 0.85 global ANOVA test;+ compared to the homozygotes o f the common allele;1 p value for the trend test; all after adjusting for age and center Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T able 4 .11. Serum P SA levels and com m on polym orphism s in the P SA gene __________________________ am ong A frican-A m ericans__________________________ Genotypes n InPSA InPSA Geometric . t p for (lsmean) (SE) mean (ng/mL) P P trend1 -5567 GG 123 0.14 0.09 1.19 AG 5 0.15 0.41 1.20 0.98 NA AA 0 -5429 I T 104 0.10 0.10 1.14 TG 23 0.36 0.19 1.47 0.19 GG 1 -1.01 0.88 0.40 0.19 0.21 0.49 TG or GG 24 0.31 0.19 1.40 0.31 -5413 TT 104 0.10 0.10 1.14 TC 23 0.36 0.19 1.47 0.19 CC 1 -1.01 0.88 0.40 0.19 0.21 0.49 TC or CC 24 0.31 0.31 1.40 0.31 -5307 GG 39 -0.14 0.15 0.91 AG 72 0.18 0.11 1.24 0.06 AA 17 0.57 0.21 1.81 0.02 0.005 0.004 -5217 TT 99 0.25 0.10 1.32 AT 27 -0.13 0.17 0.92 0.05 AA 2 -0.83 0.61 0.47 0.04 0.09 0.01 AT or AA 29 -0.18 0.16 0.87 0.02 -4643 AA 106 0.10 0.10 1.14 AG 22 0.32 0.20 1.42 0.29 NA GG 0 -4336 88 60 0.08 0.12 1.12 89 58 0.17 0.13 1.22 0.58 99 10 0.43 0.29 1.58 0.50 0.25 0.27 -4289 CC 106 0.10 0.10 1.14 (ARE VI) AC AA 22 0 0.32 0.20 1.42 0.29 NA -3647 TT 81 0.15 0.11 1.20 AT 43 0.16 0.15 1.21 0.98 AA 4 -0.13 0.45 0.92 0.82 0.54 -2905 CC 48 0.02 0.14 1.06 CT 64 0.19 0.12 1.25 0.31 TT 16 0.28 0.23 1.36 0.48 0.31 -158 GG 33 0.11 0.16 1.15 (ARE-1) AG 62 0.13 0.12 1.18 0.92 * AA t 33 0.19 0.17 1.25 0.94 0.74 0.74 T ---------------------------- T -------------------------------------------------------------------- global ANOVA test; compared to the homozygotes of the common allele; 1 p value for the trend test; all after adjusting for age and center Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. SNPs -5429T/G, -5413T/C, -4643A/G Report gene assays have demonstrated that these three SNPs significantly impact PSA gene promoter activities (Cramer et al., 2003). In single SNP analyses among whites, they were all significantly associated with serum PSA levels (Table 4.10). For each SNP, men with two copies of the “higher-activity” allele (the allele with higher promoter activity) had significantly higher PSA levels, compared to those with two copies of the “lower activity” alleles. Among African-Americans, higher serum PSA levels were also associated with the “higher-activity” alleles at these three loci when comparing the heterozygotes to the homozygotes of the “lower-activity” alleles, but the differences were not statistically significant (Table 4.11). We were not able to examine PSA levels in the homozygotes of the “higher- activity” alleles in African-Americans, because of sparse data. There was very strong LD among these three SNPs. The pair-wise LD and R2 are shown in Table 4.12 for whites and Table 4.13 for African-Americans. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.12. Pairwise LD (Levwontin’s D’ and R2 ) among whites -5567 -5429 -5413 -5307 -5217 -4643 -4336 -4289 -3647 -2905 -158 -5567 0.03 0.03 0.21 0.01 0.03 0.11 0.03 0.01 0.18 0.09 -5429 1 0.98 0.13 0.03 0.79 0.24 0.79 0.05 0.08 0.17 -5413 1 1 0.14 0.03 0.82 0.24 0.81 0.05 0.09 0.17 -5307 1 0.97 1 0.02 0.12 0.48 0.12 0.05 0.33 0.08 -5217 1 1 1 0.84 0.02 0.06 0.02 0.07 0.03 0.07 -4643 1 0.96 0.96 1 1 0.22 0.99 0.05 0.13 0.25 -4336 1 0.96 0.96 0.94 1 1 0.22 0.08 0.10 0.00 -4289 1 0.95 0.95 1 1 1 1 0.05 0.14 0.24 -3647 1 0.91 0.91 0.81 0.38 1 0.58 1 0.08 0.16 -2905 0.97 0.72 0.75 0.61 1 1 0.4 1 1 0.46 -158 1 0.71 0.73 0.43 1 0.94 0.01 0.92 1 0.97 Numbers in the lower left cells below the diagonal line show the value o f D ’ and the upper right cells above the diagonal line represent the value o f R2. Table 4.13. Pairwise LD (Levwontin’s D’ and R2 ) among African-Americans -5429 -5413 -5307 -5217 -4643 -4336 -4289 -3647 -2905 -158 -5429 1.0 0.08 0.02 0.87 0.03 0.87 0.03 0.02 0.05 -5413 1 0.08 0.02 0.87 0.03 0.87 0.03 0.02 0.05 -5307 1 1 0.03 0.07 0.18 0.07 0.02 0.06 0.001 -5217 1 1 0.55 0.01 0.04 0.01 0.05 0.01 0.02 -4643 1 1 1 1 0.04 1.0 0.02 0.06 0.07 -4336 0.76 0.76 0.53 0.81 1 0.04 0.06 0.02 0.003 -4289 1 1 1 1 1 1 0.02 0.06 0.07 -3647 1 1 0.33 0.29 1 0.32 1 0.02 0.05 -2905 0.61 0.61 0.25 0.39 1 0.14 1 0.35 0.52 -158 0.64 0.64 0.04 0.39 0.86 0.08 0.86 0.42 0.92 Numbers in the lower left cells below the diagonal line show the value o f D ’ and the upper right cells above the diagonal line represent the value o f R2. 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In whites, we observed only 5 haplotypes, with the two major haplotypes (TTA and GCG in the order of -5429, -5413, -4643) having a combined frequency of 95.9%: TTA (73.6%), GCG (22.3%), GCA (3.1%), TTG (0.6%), and GTA (0.5%). In African-Americans, there were only 3 haplotypes: TTA (90.2%), GCG (8.6%), and GCA (1.2%) and the two major haplotypes had a combined frequency of 98.8%. The strong LD among the three SNPs explained why single locus analyses of the three SNPs gave essentially the same results (Table 4.10 and 4.11). For example, in whites, P associated with an increase of one copy of the G allele at -5429 was 0.184, P associated with an increase of one copy of the C allele at -5413 was 0.194, and p associated with an increase of one copy of the G allele at -4643 was 0.189 (we call these models the single SNP additive models— e.g. Model 1 below). Also because of the strong LD, we were not able to examine the independent effect of each SNP by adjusting for the other two in regression analyses. Model 1: lnpsal= Po+PiPSA5429nG*+pa g e age+ pcenterCenter *PSA5429nG: # o f copies o f the G allele at the -5429 locus, as 0,1,2 We further attempted to examine whether some special combinations o f the sequence variants at all three loci were more strongly associated with serum PSA levels than other combinations. Because of the strong LD and a resultant rareness of minor haplotypes, there was little, if any, phase ambiguity in the double or triple heterozygotes. For example, among whites, the estimated possibility of heterozygotes having a combination o f two rare haplotypes was as low as 0.0011; and the possibility among African-American was 0 (as estimated by TagSNP). Thus 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the phase ambiguity was ignored and the haplotype configurations for each individual corresponding to his genotypes at the three loci were easily obtained. For whites, the haplotype configurations are listed in Table 4.14. These haplotype configurations can be simply viewed as the overall numbers of “higher-activity” alleles at three loci (-5429G, -5413C and -4643 G)— the “allelic dosage” (Table 4.14). Assuming equal effect of the three alleles, there was an increase of 0.066 in PSA levels on the log scale associated with an increase in the copies of “higher- activity” allele (p=0.008, we call this model the multiple loci additive model— Model 2). Model 2: lnpsal=p0 + Piallelic_dose* + page age+ pcenterCenter *alle!ic dose: the allelic dosage as a continuous variable When comparing the multiple loci additive model (model 2) to the single SNP additive model (model 2), the multiple loci additive model (the model with the combined allelic dosage at three loci) did not seem to fit better than the model with SNP -5429 alone (Model 2, AIC =1124.6 vs. Model 1: AIC =1125.2). A likely interpretation is that the homozygotes o f the “lower-activity” allele, the heterozygotes and the homozygotes o f the “higher-activity” allele at -5429 (TT, TG and GG) were mostly the same subjects carrying 0, 3 and 6 copies of the “higher- activity” alleles overall (haplotype configurations TTA/TTA, TTA/GCG and GCG/GCG, in the order of -5429, -5413 and -4643). Very similar results were obtained when we compared model 2 to the models with the single SNP additive effect o f -5413 (AIC=1124.5) and -4643 (AIC=1125.3), respectively. 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. We also relaxed the additivity assumption and fit models using dummy variables to code combined genotype categories. The model of the effect of genotype combinations at the three loci (with different allelic dosages) (AIC=1125.2, the multiple loci codominant model— model 3) did not improve predictive information when compared to models of co-dominant effect of each SNP only (single SNP codominant model— model 4, AIC=1121.0) (Table 4.14) Model 3: lnpsal=p0 + PiCom_geno2+ p2 com_geno3+ p3com _geno4+ p4 com_geno5 +p5 com_geno6 + pa g e age+pc e n te rcenter *com_geno2: dummy variable for genotype combination with allelic dosage=l *com_geno3: dummy variable for genotype combination with allelic dosage=2 *com_geno4: dummy variable for genotype combination with allelic dosage=3 *com_geno5: dummy variable for genotype combination with allelic dosage=4 *com_geno6: dummy variable for genotype combination with allelic dosage=5 Model 4: lnpsal= p0 + Pi PSA5429AG+ p2PSA5429GG+pa g e age+ p centerC enter, * PSA 5429AG: dummy variable for heterozygotes AG at -5429 * PSA 5429GG: dummy variable for homozygotes GG at -5429 We were not able to examine the effect of the combinations of the three sequence variants in African-Americans because the 3 variants always tracked together except in only 2 subjects. In summary, the three functional SNPs were all associated with serum PSA levels, at least among whites. Due to the very strong LD among them, we were not able to examine the independent effect of each SNP, and a model with any one of the three SNPs provided predictive information of PSA levels as good as a model with the genotype combinations of three SNPs. Therefore, in the following analyses, -5413 was used to represent the 3 functional SNPs. 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. SNP -4289 C/A SNP -4289 C/A is located in ARE-VI and is therefore potentially functionally important, although the sequence variation at -4289 did not cause significant change in promoter activity in vitro (Cramer et al., 2003). Among whites, men with two copies o f the A allele had significantly higher PSA levels compared to those with two copies of the C alleles (p<0.001). Among African-American men, we were not able to examine the homozygotes of the A allele because there were no subjects with this genotype. However, we did observe that heterozygotes had higher serum PSA levels than the homozygotes of the C alleles, although the difference was not statistically significant. We further examined whether the observed significant association could be explained by LD between this SNPs with the 3 functional SNPs (as represented by - 5413). SNP -4289A/C was in strong LD with -5413. For example, the pairwise D ’ and R2 between -4289A/C and -5413C/T were 0.95 and 0.81. 96.2% of the A alleles of the -4289 SNP (which associated with higher PSA level) were linked to the “higher-activity” C allele of the -5413 SNP; and 97.0% of the C alleles of the -4289 SNP were linked to the “lower-activity” T allele o f the -5413 SNP. The model with the genotype combinations of -4289 and -5413 did not provide more predictive information than the model with the -5413 SNP alone (AIC=1123.9 for the multiple loci codominant model, adjusting for age and center vs. AIC=1121.0 for the single SNP codominant model) (data not shown). Our data suggested that the observed 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. association between the -4289 was likely to be due to its high LD with the three functional SNPs. Polymorphism -4336 8C/9C Among whites, SNP -4336 was associated with serum PSA levels. However, it was only the heterozygotes who had significantly lower PSA levels compared to the baseline group, homozygotes of the 8 cytosine allele (-4289 8C) (p=0.02), i.e. a dose-response was not observed. Among African-Americans, there was no significant association between this SNP and serum PSA levels. This polymorphism had no effect on promoter activities in vitro (Cramer et al., 2003). The LD between the -4336 polymorphism and -5413 SNP was less strong (D’=0.76, R2 =0.03), therefore its independent effect on serum PSA was examined by adjusting for SNP -5413. The association between the -4336 and serum PSA levels was no longer significant after adjusting for SNP -5413, suggesting that the observed associated m aybe due to its LD with SNP -5413 (F(2 , 4 2 i) =1.19, p=0.31 comparing the model with the main effects of the two polymorphisms to the model with 5413 alone, adjusting for age and center) SNP -3647 T/A SNP -3647 T/A is located in an ECR. It was a newly-identified polymorphism in the present study and no functional data are available yet. Among whites, there was a significant decrease in serum PSA levels with an increase in the copies of the A allele (p=0.03). Among African-Americans, we did not observe a 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. statistically significant association between SNP -3647 T/A and serum PSA levels, although the homozygotes of the A allele appeared to have lower PSA levels. The LD between this SNP and the 3 functional SNPs was not perfect. For example, the pairwise D ’ and R2 between -3647A/T and -5413C/T were 0.91 and 0.05. 70.7% of the T alleles of the -3647 SNP were linked to the C allele of the -5413 SNP and 29.3% were linked to the T allele; 97.9% of the A alleles of the -3647 SNP were linked to the T allele of the -5413 SNP and 2.1% of the A alleles were linked to the C allele. The association between the -3647 SNP and PSA levels remained significant after adjusting for the -5413 SNP, suggesting the association o f the -3647 SNP with PSA levels was independent from its LD with the functional SNPs (additive effect of -3647: F(i> 4 2 4 ) =4.99, p=0.03, comparing model 5 to model 6). Model 5: lnpsal= p0+ P i PSA5413TC* + p2PSA5413CC* + p3PSA3647nA+ +Page age + PcenterCenter *dummy variables for genotypes at -5413 * copies o f the A allele at -3647 Model 6: lnpsal= p0 +piPSA5413TC* + p2PSA5413CC* +pa g e age P centerC enter *dummy variables for genotypes at -5413 Because the minor haplotype -5413C/-3647A was very rare (frequency = 0.003), the estimated possibility of double heterozygotes carrying this haplotype was low (5%), i.e. there was little phase ambiguity. We therefore were not able to assess any possible additional “phase” effect between these two SNPs. Although the effect of the -3647 SNP was more evident among men carrying at least one C allele at the - 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5413 SNP, we did not observe a statistically significant interaction between the two SNP (among men carrying two T alleles at -5413, those with at least one A allele at - 3647 had an average PSA levels 23% higher than those with two T alleles at -3647; among men carrying at least one C allele at -5413, those with at least one A allele at -3647 had an average serum PSA levels 46% higher than those with two T alleles at - 3647; p for interaction=0.46). SNPs -5307 G/A and -5217 T/A There are no functional data on these two SNPs. They were significantly associated with serum PSA levels among African-Americans, but not among whites. Among African-Americans, these two SNPs were not in strong LD (D’=0.55 and R2 =0.03) and their associations with serum PSA level appeared to be independent, because they both remained significant after adjustment for each other (P5307ga=0.32, SE=0.17, p=0.06; P 5307a a = 0 .6 4 , SE=0.25, p=0.01; and p 52i 7T T =0 .3 6 , SE=0.18, p=0.05 in the model with the linear combination of the two SNPs, adjusting for age and center). To further investigate whether specific combinations of the sequence variants at the two loci (interaction and phase) were more strongly associated with PSA levels, we compared the haplotype model (Model 7) (Table 4.15) to the model with the linear combination of additive effects of the two SNPs (we call it the two loci additive effect model— Model 8). Because in the case of 2 SNPs, the haplotype model (Model 7) was equivalent to adding one parameter (capturing both interaction and phase information) to Model 8 (i.e. these two models were nested) (Conti and 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gauderman, 2004), an F test statistics was used to compare the two models. The haplotype model did not provide significantly more information than the two loci additive effect model (F(i;i2 2 ) =1.11, p=0.29). However, we have to note that we had very few subjects carrying haplotype -5307A/-5217A (estimated haplotype frequency=0.023) and data were sparse in haplotype analysis. Therefore, inference about possible interactions and phase effect between these two SNPs must be made cautiously from our limited data. Model 7: lnpsal= ( 3 0 +Pi h a p l o t y p e . 53 0 7 G /-5 2 i7 A * + P 2 h a p l o t y p e . 530 7 A /-5 2 i7 T * + p3 h ap lotyp e.5307A/-52i7A* +Pa ge age+ pcen tercenter) * haplotype_5307G/-52i7A, haplotype_5307A /-52i7T , haplotype.5 3 07A /-52i7A : number o f copies o f the specified haplotypes (for double heterozygotes, the probability o f carrying this haplotype) Model 8: lnpsal= Po+ Pipsa5307nG*+ P2psa5217nA* +pag e age+ pcen tercenter *PSA5307nG: # o f copies o f the G allele at the -5307 locus, as 0,1,2) *PSA5217nA: # o f copies o f the A allele at the -5217 locus, as 0,1,2) SNPs -5567G/A, -2905 C/T, and -158 G/A At locus -5567, white males carrying two copies of the A allele appeared to have higher PSA levels, but the result was not statistically significant (p=0.09) and we have to note that there were only 3 subjects in this genotype group. This SNP was not associated with serum PSA in African-Americans. SNP -2905 C/T was a newly- identified SNP in the ECR. There was no significant association between this SNP and serum PSA in either whites or African-Americans. We did not observe Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. significant association between the SNP in the ARE-I region (-158 G/A) and serum PSA levels in either whites or African-Americans. Summary: The results of the analyses on single polymorphisms in the PSA gene and the joint effects of these polymorphisms suggested that SNPs -5413 T/C and -3647 T/A were the only two important and independent predictors of serum PSA levels among non-Hispanic whites, and the SNPs -5307G/A and -5217T/A were the important predictors of serum PSA levels among African-Americans. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.14. Serum PSA levels and genotype combinations of the 3 functional SNPs among whites Genotype combinations at three loci -5429 -5413 4643 Corresponding haplotype configuration n InPSA lsmean (SE)* Allelic dosage* Dummy variables InPSA lsmean (SE) TT TT AA TTA /TTA 227 0.19(0.06) 0 ref 0.19(0.06) TT TT AG TTA/TTG 5 -0.19(0.40) 1 com geno2 -0.17(0.30) GT TT AA TTA/GTA 4 -0.14(0.45) 1 GT CT AA TTA/GCA 21 0.17(0.19) 2 com geno3 0.17(0.19) GT CT AG TTA / GCG 146 0.25 (0.07) 3 com geno4 0.25 (0.07) GG CC AG GCG / GCA 5 0.65 (0.40) 5 com geno5 0.65 (0.40) GG CC GG GCG / GCG 20 0.89 (0.20) 6 com_geno6 0.89 (0.20) adjusted for age and center * the overall number of “higher-activity” alleles (the allele with higher promoter activity) at three loci Table 4.15. Serum PSA and the haplotypes of SNPs -5307 G/A and -5217 T/A ___________________ among African-Americans________ _________ haplotypes frequencies P’ (SE) P -5307G/-5217T 0.487 ref -5307G/-5217A 0.392 0.21(0.22) 0.35 -5307A/-5217T 0.099 0.57 (0.13) 0.01 -5307A/-5217A 0.023 -0.37 (0.69) 0.60 compared to homozygotes of haplotype -5307G/-5217T, an increase in mean serum PSA levels on the log scale with an increase in the copies o f the haplotype of interest K > or. 4.3.7. Serum PSA levels and the AR CAG polymorphism The race-specific distributions of the CAG repeat lengths were reported on Figure 4.3 and in Table 4.16. As reported previously, African-Americans had shorter CAG repeat lengths, compared to non-Hispanic whites (Edwards et al., 1992; Irvine et ah, 1995; Xue et ah, 2001). The mean CAG repeat lengths in non-Hispanic whites and African-Americans were 21.1 and 18.9 and the medians were 21 and 19, respectively (p<0.001). CAG repeat lengths were not significantly associated with serum PSA levels fl3= -0.008, p=0.63, after adjusting for age, center and race) (Table 4.17). We did not observe significant difference in slopes between the two racial groups (p=0.81). Deciles were also checked and the results were similar (data not shown). When moving average (smoothing) method was applied to re-examine the data more closely (Figure 4.4), we noticed that there appeared to be an inverse association between serum PSA level and the CAG repeat length within the range from 16 to 28 CAG repeats. This observation corroborated with a recent study by Buchanan et al, which suggested that the linear negative association between CAG repeat lengths and AR transactivation activity only existed within this range o f CAG lengths from 16 to 29 (Buchanan et al., 2004). Since there were only a few subjects with CAG repeat lengths at the two extreme ends, we were not able to study PSA levels in men with extremely short or long CAG lengths. After excluding men with extremely short or extremely long CAG repeats from the analysis, we still did not find any significant association between serum PSA levels and CAG repeat lengths 126 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (P for an increment of one CAG repeat = -0.009, SE=0.017, p=0.59 in whites, and P = 0.002, SE=0.032 and p=0.9384 in African-Americans, after adjusting for age and study center). Interestingly, we observed that men with CAG repeat length equal to 20 appeared to have significantly lower PSA levels compared to men with any other CAG repeat lengths (p<0.01). This group of men did not differ significantly from the other in age, study center, history of BPH, history of prostatitis, or PSA genotypes (data not shown). o 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 32 34 AR [CAG]n ■ non-Hispanic Whites □ African-Americans Figure 4.3. The distribution of the AR CAG repeat lengths by race 127 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.16. Race-specific distributions of the AR CAG repeat lengths Whites African-Americans (n=414)_________________(n=124) No of repeats No Percentage (%) No Percentage (%) 12 0 0.0 1 0.8 13 3 0.7 4 3.2 14 2 0.5 2 1.6 15 2 0.5 4 3.2 16 8 1.9 15 12.1 17 24 5.8 18 14.5 18 37 8.9 13 10.5 19 41 9.9 16 12.9 20 74 17.9 14 11.3 21 51 12.3 13 10.5 22 47 11.4 10 8.1 23 40 9.7 9 7.3 24 40 9.7 0 0.0 25 18 4.4 1 0.8 26 6 1.5 3 2.4 27 9 2.2 1 0.8 28 5 1.2 0 0.0 29 4 1.0 0 0.0 30 1 0.2 0 0.0 32 1 0.2 0 0.0 34 1 0.2 0 0.0 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.17. Serum PSA levels and the AR CAG polymorphism CAG repeats all subjects whites African-Americans N lnpsa lsmean (SE)* Geometric mean N lnpsa lsmean (SE)t Geometric mean N lnpsa lsmean (SE) * Geometric mean 12-17 83 0.34 (0.10) 1.45 39 0.46 (0.14) 1.63 44 0.15(0.14) 1.20 18-19 107 0.32 (0.09) 1.41 78 0.35(0.10) 1.45 29 0.21 (0.17) 1.27 20 88 -0.06 (0.10) 0.98 74 -0.01 (0.10) 1.03 14 -0.25 (0.25) 0.81 21-22 121 0.22 (0.09) 1.28 98 0.20 (0.09) 1.25 23 0.35(0.19) 1.46 23-34 Trend test 139 0.24 (0.09) p = -0.01 (SE=0.02) p=0.60 1.31 125 0.29 (0.08) p= -0.01 (SE=0.02) p=0.73 1.38 14 -0.01 (0.23) p = -0.01 (SE=0.03) p=0.73 1.03 Adjusting for age, center and race;+ adjusting for age and center; p for interaction =0.91 Lowess smoother CN Q_ C 10 15 20 25 30 35 arcag b a n d w id th = .8 Figure 4.4. Lowess smoothing average plot of serum PSA levels vs. AR CAG repeat lengths K > V O 4.3.8. Effect modifications in the associations between serum PSA levels and the PSA gene and AR gene polymorphisms Age Our stratified analyses by age (using the 75th percentile) suggested that the associations between the -5413 T/C SNP, the -5307 G/A SNP and PSA levels were more evident among men younger than 70 (Table 4.18 and Table 4.19) (p for interaction=0.09 for -5413 SNP in non-Hispanic whites and p for interaction=0.05 for -5307 SNP in African-American). The associations between other SNPs in the PSA gene and serum PSA did not appear to differ in the two age groups (Table 4.18 and Table 4.19). Nor did we observe any significant heterogeneity in the association between the CAG repeat polymorphism and serum PSA levels by age (Table 4.20). Self-reported BPH Although the association between the -5413 T/C SNP and PSA levels was more evident among men without BPH, the interaction was not significant (Table 4.21). We did not observe any significant heterogeneity in the associations between other SNPs in the PSA gene and serum PSA levels by BPH status (Table 4.21 and Table 4.22). Nor did we observe any significant heterogeneity in the association between the CAG repeat polymorphism and serum PSA levels by BPH (Table 4.23). Serum androgen and SHBG levels We did not observe any significant difference in the associations between the SNPs in the PSA gene and serum PSA levels by 3a-diol G levels or SHBG levels (Table 4.24, Table 4.25, Table 4.27 and Table 4.28). We did not observe any 130 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. significant heterogeneity in the association between the CAG repeat polymorphism and serum PSA levels by 3a-diol G levels or SHBG levels. Although the data appeared to suggest that there was a negative association between the CAG repeat lengths and serum PSA levels among men with lower bioactive androgen levels (i.e. lower 3a-diol G levels or higher SHBG levels), the trend was not significant in either group (Table 4.26 and Table4.29). Interaction between the SNPs in the PSA gene and AR CAG repeats We did not observe significant interactions between the SNPs in the PSA gene (-5429 T/G and -5413 T/C in whites, -5307 G/A and -5217 T/A in African- American) and AR CAG repeats in relation to serum PSA levels (Table 4.30 and Table 4.31). However, our data did suggest that the effects of the PSA SNPs were in general more evident among men with longer CAG repeats (CAG >20), expect for the -5217 SNP (Table 4.31). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. T able 4 .1 8 . Serum P S A le v els and the P S A gen e p olym orph ism s (-5413 T /C and -3 6 4 7 T /A ) am ong w h ites, b y age Age<70 A ge>=70 genotypes n lnpsa lsmean (SE) Geometric mean P Trend test n lnpsa lsmean (SE) Geometric mean P Trend test p inter -5413 TT TC CC 175 114 17 -0.04 (0.07) 0.13 (0.08) 0.77 (0.22) 0.99 1.18 2.20 0.10 <0.001 P=0.28 (SE=0.08) p<0.001 61 53 8 0.74(0.11) 0,54 (0.12) 1.09 (0.32) 2.13 1.76 3.00 0.25 0.30 p= -0.01 (SE=0.13) p=0.93 0.09 -3647 TT AT A A 226 78 2 0.14 (0.06) -0.14(0.10) -0.18 (0.64) 1.19 0.91 0.87 0.02 0.61 P= -0.27 (SE=0.11) P=0.02 86 29 7 0.75 (0.10) 0.58(0.17) 0.17(0.34) 2.16 1.82 1.22 0.36 0.10 P= -0.24 (SE=0.14) P=0.09 0.94 adjusting for age and center T able 4.19. Serum P S A le v els and the P S A gen e p olym orphism s (-5 3 0 7 G /A and -5 2 1 7 T /A ) am ong A frican-A m ericans, b y age Age<70 Age>=70 genotypes n lnpsa lsmean (SE) Geometric mean P Trend test n lnpsa lsmean (SE) Geometric mean P Trend test p inter -5307 GG GA AA 31 62 13 -0.29 (0.15) 0.07 (0.11) 0.73 (0.22) 0.78 1.11 2.12 0.04 <0.001 P=0.47 (SE=0.13) p<0.001 8 10 4 0.59 (0.48) 1.00 (0.6) 0.53 (0.75) 1.84 2.76 1.73 0.52 0.93 p= 0.004 (SE=0.37) p=0.99 0.05 -5217 TT AT or AA 81 25 0.18(0.1) -0.31 (0.17) 1.24 0.77 0.01 NA 18 4 0.71 (0.48) 0.65 (0.65) 2.07 1.96 0.94 NA 0.34 adjusting for age and center Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.20. Serum PSA levels and the AR CAG polymorphism, by age age <70 age >70 CAG repeats N lnpsa lsmean (SE) PSA Geometric mean N lnpsa lsmean (SE)* PSA Geometric mean 12-17 66 0.12(0.11) 1.17 17 1.00 (0.23) 2.75 18-19 82 0.20 (0.10) 1.25 25 0.67 (0.21) 1.99 20 58 -0.21 (0.12) 0.85 30 0.41 (0.19) 1.54 21-22 88 0.12(0.10) 1.17 33 0.52 (0.18) 1.72 23-34 107 0.08 (0.10) P = -0.004 (SE=0.02) Ptrd=0.82 1.12 32 0.67 (0.2) P = -0.03 (SE=0.04) p trd=0.47 2.00 * adjusting for age, center and race p for interaction=0.63 Table 4.21. Serum PSA levels and the PSA gene polymorphisms (-5413 T/C and -3647 T/A) among whites, by BPH Without BPH With BPH genotypes n lnpsa lsmean (SE) Geometric mean P Trend test n lnpsa lsmean (SE) Geometric mean P Trend test p inter -5413 TT 162 -0.02 (0.07) 1.02 p=0.28 70 0.59 (0.11) 1.83 P= 0.07 TC 124 0.13 (0.08) 1.18 0.14 (SE=0.08) 42 0.54 (0.14) 1.75 0.78 (SE=0.14) CC 18 0.81 (0.21) 2.28 <0.001 p<0.001 7 0.96 (0.34) 2.66 0.30 p=0.64 0.17 -3647 TT 222 0.19(0.06) 1.25 P= -0.33 87 0.62 (0.09) 1.89 P= -0.16 AT 78 -0.18(0.10) 0.88 0.02 (SE=0.11) 28 0.61 (0.17) 1.88 0.96 (SE=0.15) AA 4 -0.2 (0.45) 0.86 0.40 P=0.002 4 -0.16(0.44) 0.89 0.09 P=0.29 0.37 adjusting for age and center Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.22. Serum PSA levels and the PSA gene polymorphisms (-5307 G/A and -5217 T/A) among African-Americans, by BPH Without BPH With BPH genotypes n lnpsa lsmean (SE)* Geometric mean P Trend test n lnpsa lsmean (SE)* Geometric mean P Trend test p inter -5 3 0 7 GG 35 -0.13 (0.15) 0.91 (3=0.33 3 -0.59 (0.59) 0.59 P= 0.66 GA 58 0.16(0.12) 1.21 0.04 (SE=0.13) 13 0.35 (0.42) 1.46 0.52 (SE=0.39) AA 12 0.57 (0.25) 1.81 <0.001 p=0.01 4 0.78 (0.62) 2.22 0.93 p=0.12 0.28 -5 2 1 7 TT 81 0.21 (0.11) 1.27 16 0.24 (0.42) 1.31 ' AT or AA 24 -0.15(0.18) 0.90 0.07 NA 4 -0.08 (0.57) 0.96 0.60 NA 0.34 adjusting for age and center Table 4.23. Serum PSA levels and the AR CAG polymorphism, by BPH Without BPH With BPH CAG repeats N lnpsa lsmean (SE) PSA Geometric mean N lnpsa lsmean (SE) PSA Geometric mean 12-17 64 0.2 (0.11) 1.26 17 0.77 (0.22) 2.20 18-19 77 0.23 (0.10) 1.30 29 0.66 (0.19) 1.97 20 64 -0.21 (0.12) 0.85 23 0.36 (0.22) 1.46 21-22 91 0.20 (0.10) 1.26 29 0.32 (0.19) 1.42 23-34 101 0.13(0.10) 1.18 36 0.62 (0.19) 1.90 P = -0.004 p =- 0.01 (SE=0.02) (SE=0.03) Ptrd=0.83 p trd=0.72 adjusting for age, center and race p for interaction=0.90 4 ^ Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.24. Serum PSA levels and the PSA gene polymorphisms (-5413 T/C and -3647 T/A) among whites, by 3a-diol G 3a-diol G <median 3a-diol G >median genotypes n lnpsa Geometric P Trend test n lnpsa Geometric P Trend test p inter lsmean (SE) mean lsmean (SE) mean -5 4 1 3 TT 115 0.06 (0.09) 1.10 P=0.28 121 0.28 (0.08) 1.36 p= 0.11 TC 84 0.18(0.10) 1.24 0.39 (SE=0.11) 83 0.30 (0.09) 1.38 0.87 (SE=0.09) CC 12 1.01 (0.27) 2.78 0.001 p<0.01 13 0.68 (0.23) 2.02 0.10 p=0.25 0.24 -3 6 4 7 TT 150 0.24 (0.08) 1.31 P= -0.25 162 0.38 (0.07) 1.50 P= -0.25 AT 56 0.0 (0.13) 1.04 0.12 (SE=0.13) 51 0.12(0.12) 1.17 0.06 (SE=0.12) AA 5 -0.3 (0.44) 0.78 0.22 P=0.06 4 -0.10(0.42) 0.94 0.26 P=0.03 0.96 adjusting for age and center Table 4.25. Serum PSA levels and the PSA gene polymorphisms (-5307 G/A and -5217 T/A) among African-Americans, ___________________________________________ by 3a-diol G.___________________________________________ 3a-diol G <median 3a-diol G >median genotypes n lnpsa lsmean (SE) Geometric mean P Trend test n lnpsa lsmean (SE) Geometric mean P Trend test P inter -5 3 0 7 GG 23 -0.22 (0.21) 0.84 P=0.27 16 -0.02(0.18) 1.02 P= 0.39 GA 36 0.24 (0.17) 1.31 0.07 (SE=0.18) 36 0.10(0.13) 1.14 0.58 (SE=0.15) AA 8 0.15(0.35) 1.20 0.35 p=0.15 9 0.89 (0.24) 2.47 0.003 p=0.01 0.54 -5 2 1 7 TT 49 0.22 (0.16) 1.29 50 0.25 (0.12) 1.32 AT or AA 18 -0.25 (0.23) 0.81 0.07 NA 11 -0.02 (0.23) 1.02 0.60 NA 0.28 adjusting for age and center L*J Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.26. Serum PSA levels and the AR CAG polymorphism, by 3a-diol G 3a-diol G <median 3a-diol G >median CAG repeats N lnpsa lsmean (SE) PSA Geometric mean N lnpsa lsmean (SE)* PSA Geometric mean 12-17 42 0.33 (0.15) 1.43 41 0.35 (0.13) 1.46 18-19 52 0.17(0.14) 1.22 55 0.46 (0.12) 1.62 20 42 -0.14(0.15) 0.91 46 0.04 (0.13) 1.08 21-22 72 0.17(0.12) 1.22 49 0.31 (0.13) 1.40 23-34 62 0.04 (0.13) P = -0.03 (SE=0.03) Ptrd=0.23 1.08 77 0.41 (0.11) p = 0.01 (SE=0.03) p trd=0.63 1.54 Adjust for age, center and race p for interaction =0.39 Table 4.27. Serum PSA levels and the PSA gene polymorphisms (-5413 T/C and -3647 T/A) among whites, by SHBG SHBG <median SHBG >median genotypes n lnpsa lsmean (SE)* Geometric mean P Trend test n lnpsa lsmean (SE) Geometric mean P Trend test p inter -5413 TT 118 0.15(0.08) 1.20 P=0.24 118 0.18(0.08) 1.24 p= 0.16 TC 78 0.25 (0.10) 1.32 0.46 (SE=0.10) 89 0.23(0.10) 1.29 0.73 (SE=0.10) CC 13 0.90 (0.25) 2.50 0.004 p=0.02 12 0.8 (0.26) 2.25 0.03 p=0.11 0.58 -3647 TT 162 0.30 (0.07) 1.39 p= -0.25 150 0.32 (0.07) 1.41 p= -0.25 AT 44 0.03(0.14) 1.07 0.08 (SE=0.13) 63 0.10(0.11) 1.14 0.11 (SE=0.12) AA 3 -0.07 (0.52) 0.97 0.49 P=0.06 6 -0.29 (0.37) 0.79 0.11 P=0.04 1.00 adjusting for age and center Os Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.28. Serum PSA levels and the PSA gene polymorphism (-5307 G/A and -5217 T/A) among African-Americans, by SHBG SHBG <median SHBG >median genotypes n lnpsa lsmean (SE) Geometric mean P Trend test n lnpsa lsmean (SE)* Geometric mean P Trend test p inter -5307 GG 16 -0.25 (0.26) 0.82 p=0.40 23 -0.06(0.18) 0.98 P= 0.30 GA 45 0.21 (0.16) 1.27 0.09 (SE=0.19) 27 0.11 (0.17) 1.15 0.49 (SE=0.17) AA 10 0.52 (0.3) 1.72 0.05 p=0.04 7 0.65 (0.33) 1.96 0.06 p=0.08 0.72 -5217 TT 54 0.30 (0.15) 1.39 45 0.20 (0.14) 1.26 AT or AA 17 -0.14 (0.24) 0.91 0.08 NA 12 -0.22 (0.24) 0.84 0.14 N A 0.83 adjusting for age and center Table 4.29. Serum PSA levels and the AR CAG polymorphism, by SHBG SHBG <median SHBG >median CAG repeats N InPSA lsmean (SE) Geometric mean N InPSA lsmean (SE)* Geometric mean 12-17 45 0.27(0.13) 1.35 38 0.41 (0.15) 1.55 18-19 54 0.32 (0.13) 1.41 53 0.31 (0.13) 1.40 20 43 -0.12(0.15) 0.92 45 -0.01 (0.14) 1.03 21-22 63 0.36(0.12) 1.47 58 0.05 (0.13) 1.09 23-34 69 0.35 (0.12) P = 0.02 (SE=0.02) p trd=0.38 1.46 70 0.12(0.12) P = -0.03 (SE=0.02) p trd=0.17 1.17 ‘adjust for age, center and race p for interaction^). 18 OJ Table 4.30. Serum PSA levels and the AR CAG polymorphism, by PSA genotypes PSA genotypes n beta (SE)f P p for interaction -5413* TT 230 -0.01 (0.03) 0.69 TC or CC 184 -0.01 (0.04) 0.70 0.88 -3647* TT 300 0.01 (0.02) 0.25 -5307* TA or AA 114 -0.05 (0.04) 0.24 0.25 GG 37 -0.07 (0.06) 0.25 -5217* AG or AA 87 0.02 (0.04) 0.68 0.24 TT 95 -0.03 (0.04) 0.45 TA or AA 22 0.005 (0.06) 0.92 0.54 ♦examined in whites (results were similar if included all subjects, adjusting for race) + examined in African-Americans only I Change in serum PSA levels on log scale associated with per increment in CAG repeat length. Table 4.31. Serum PSA and the PSA polymorphisms, by the AR CAG genotypes Allele of PSA SNPs AR genotypes n beta (SE)J P p for interaction -5413 C* short (CAG<20) 205 0.09 (0.11) 0.39 long (CAG >20) 223 0.27 (0.09) 0.004 0.22 -3647 A* short (CAG<20) 205 -0.18(0.13) 0.25 long (CAG >20) 223 -0.34(0.12) 0.005 0.38 -5307 A* short (CAG<20) 91 0.25 (0.14) 0.08 long (CAG >20) 37 0.72 (0.21) 0.002 0.12 -5217 Af short (CAGK20) 91 -0.41 (0.19) 0.03 long (CAG >20) 37 -0.31 (0.38) 0.43 0.81 ♦examined in whites (results were similar if included all subjects, adjusting for race) * examined in African-Americans only t Change in serum PSA levels on log scale associated with an increase in one copy o f the specified allele o f the PSA SNPs Excluding men with CAG=20 did not change the results appreciably. 4.4. Discussion This is the first study to systematically examine possible interactions among polymorphisms in the PSA and AR genes and serum androgen levels in determining serum levels of PSA. In the PSA gene, we found three SNPs (-5429 T/G, -5413 T/C, and -4643 A/G) in the distal enhancer region to be associated with serum PSA levels among non-Hispanic white men. At each of the three loci, the “higher-activity” allele, as identified in a previous in vitro study (Cramer et al., 2003)was associated with 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. significantly increased PSA levels. These three functional SNPs were in tight LD and therefore models using any one of the SNPs were able to capture as much variation in serum PSA levels as were models using all three SNPs. Two additional polymorphisms in the distal enhancer region (SNP -4289C/A and polymorphism - 4336 8C or 9C) were also significantly associated with serum PSA levels in whites. However, sequence variation at these two loci did not cause significant changes in PSA promoter activity in a previous functional study (Cramer et al., 2003), and our results indicated that their associations with serum PSA levels were likely to be due to high LD between them and the three functional SNPs. In African-Americans, due to smaller sample size and lower frequencies of these SNPs, we were not able to find significant associations. Specifically, at the three functional loci we had almost no homozygotes o f the less common alleles, while these were the only genotype groups that showed significantly higher PSA levels among whites. However, consistent with what was found in whites, we did observe higher serum PSA levels associated with the “higher-activity” alleles at these three loci when comparing the heterozygotes to the homozygotes of the “lower-activity” alleles, although the results were not significant. In contrast to what was found in whites, two other SNPs, -5307 G/A and -5217 T/A, were significantly associated with serum PSA among African-Americans. Interestingly, for the -5307 G/A SNP, the allele that associated with significantly higher PSA levels in our study was associated with a significantly lower PSA levels in a previous study that included whites only (Cramer et al., 2003). These significant but inconsistent results, 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. if not due to chance, may be explained by the different LD patterns of this SNP with other unknown functional polymorphism in different racial groups. Similarly, the significant association between the -5217 T/A SNP and serum PSA levels could be due to chance or due to its LD with other unknown functional polymorphism(s). In addition to the previously studied distal enhancer region, we also identified two evolutionarily conserved regions (ECRs) lying within 10 kb o f the transcription start site of the PSA gene. The first ECR spans a 400-bp region upstream from the transcription start site (which also covers the ARE-1 site). In a previous study conducted by our group (Xue et al., 2001), we sequenced this region and found two common polymorphisms (-252 G/A and -232 delA). Since neither o f these two polymorphisms was significantly associated with serum PSA levels, and haplotypes of these two polymorphism with the -158G/A ARE-I SNP did not predict serum PSA better than the -158 SNP alone, they were not included in this study. In the second ECR (from -3.7 kb to -2.6 kb), we identified two common SNPs, -3647T/A and - 2905 C/T. The -3647 T/A SNP predicted serum PSA levels independently from the three functional SNPs in non-Hispanic whites. In African-Americans, although homozygotes with two A alleles had lower average PSA levels, the difference was not statistically significant. Lack of power could be one explanation. Alternatively, the -3647 SNP may not be functional; rather its association with serum PSA levels in whites could be due to its LD with other unknown functional polymorphism(s) in the PSA gene. Further functional studies on this SNP are needed to clarify its role in PSA gene regulation. 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Consistent with what has been reported by 3 previous publications (Gsur et al., 2002; Rao et al., 2003; Xu et al., 2002), we were not able to confirm the association between the ARE-1 SNP (-158 G/A) and serum PSA levels, which was reported in one o f our previous papers (Xue et al., 2001). Although it has been suggested that our previous significant result could be due to LD between this SNP and the three functional SNPs in the enhancer region (namely, the -5429 T/G, -5413 T/C, -4643 A/G) (Cramer et al., 2003), available data do not support this notion. In both the study by Cramer et al and the present one, the A allele of the -158 G/A SNP (which was significantly associated with increased PSA levels in our previous paper) was actually almost exclusively linked to the “lower-activity” alleles at the three functional loci; whereas the G allele of the -158 SNP could be linked to either the “higher-activity” alleles or the “lower-activity”. Therefore, if the -158 G/A SNP is associated with PSA levels due to its linkage with these three SNPs, we would expect to see the G, not the A, allele being associated with higher levels. On the other hand, our data suggest that LD between this SNP and the -3647T/A SNP could explain our previous finding. In the present study, the A allele of the -158 G/A SNP was exclusively linked to the T allele at -3647, which was associated with significantly higher PSA levels. In summary, the three functional SNPs in the PSA gene regulatory region (- 5429 T/G, -5413 T/C, -4643 A/G) are significant predictors of serum PSA levels. In addition, a SNP in an ECR, -3647 T/A, also appears be an independent predictor of serum PSA levels. Its functional significance needs to be examined in in vitro 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. functional studies. Similarly, our results suggest that the SNPs -5307 G/A and -5217 T/A may be also be important, especially among African-Americans, but their functional significance remains to be examined. Moreover, our data suggest that there may still be some unknown functional polymorphism(s) in other regulatory regions of the gene, including the unscreened region from -2600 to -500-bp upstream of the transcription start site. Our present study was not able to confirm a significant linear negative association between CAG repeat lengths and PSA levels in serum or in seminal fluid which was observed in three previous studies including one o f our own (Giwercman et al., 2004; Mifsud et al., 2001a; Xue et al., 2001). However, three other studies (Gsur et al., 2002; Rao et al., 2003; Xu et al., 2002) also failed to find an association between the AR CAG repeats and serum PSA. The inconsistency among studies is difficult to explain, although there are some noteworthy differences in the participants of these studies. In two o f the three studies that found a significant association, the subjects were young men in their 30s (Mifsud et al., 2001a) or from age 18-21 (Giwercman et al., 2004), while the other 4 studies covered a wider age range and had generally older subjects (with median ages in the 50s and 60s). Young adult males are known to have higher androgen levels (Gapstur et al., 2002; Svartberg et al., 2003) and at the same time, lower serum PSA levels (Collins et al., 1993; Morgan et al., 1996; Oesterling et al., 1993), presumably due to lower incidence of age-related prostate enlargement or prostate cancer (Collins et al., 1993). Therefore, it is reasonable to postulate that perhaps the effect of CAG repeat 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. lengths on PSA production only becomes manifest in men with high androgen levels and/or when it is less obscured by undiagnosed BPH or prostatic neoplasia. However, the study by Mifsud actually suggested that the association between serum PSA and CAG repeats was only found among men with low androgen levels (men with azoospermia or oligospermia) (Mifsud et al., 2001a). Data from the present study also appeared to suggest that the CAG repeat lengths was negatively associated with serum PSA levels among men with relatively lower androgen levels. However, the association was not significant. In all the studies with older subjects, effort has been taken to try to reduce the interference from undiagnosed prostate cancer or BPH by limiting the study subjects to men with PSA<4 ng/mL or with normal DRE. In the study by Xu et al (Xu et al., 2002), it was observed that after limiting the analysis to men with PSA<4 ng/mL, the difference between men with shorter and larger CAG became more evident and marginally significant (p=0.05). However, in the present study, the results did not change after we limited our analysis to men with PSA < 4ng/mL. We studied androgen levels in relation to serum'PSA levels using three different measurements: total T, free T (as calculated from total T and SHBG) and 3a-diol G. Total T is the most commonly measured androgen in epidemiological studies, but it does not yield specifically meaningful information about actual tissue androgen exposure. In normal adult men approximately 44% o f circulating T is specifically bound to SHBG, 50% is nonspecifically bound to albumin, 3.5% is bound to corticosteroid binding globulin (CBG) and only about 2.5% is unbound 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (free T) (Dunn et al., 1981). Testosterone that is bound to SHBG cannot exit blood capillaries and thus is not bioavailable. Although there is still controversy about whether or not albumin-bound T is bioactive, non-SHBG bound and non-SHBG- non-albumin bound T (free T) are highly correlated and used interchangeably as a better measurement of bioactive T (de Ronde et al., 2005). The gold-standard method for free T measurements is equilibrium dialysis. However, this method is costly, time-consuming and not well suited for epidemiological studies. In this study, free T was estimated by calculation from total T and SHBG concentrations. Previous studies have suggested that the calculated free T value appears to be a rapid, simple, and reliable index of bioactive T (Vermeulen et al., 1999). In addition, we measured 3a-diol G as a marker o f peripheral androgen exposure. In peripheral androgen target tissue such as the prostate, testosterone can be further metabolized into dihydrotestosterone (DHT) by 5a-reductase. DHT is the major intracellular androgenic metabolite within the prostate (Janulis et al., 2000) and is two or three times more potent than T (i.e. binds to androgen receptors with higher affinity and specificity) (Toscano, 1986). However, serum DHT itself is a poor circulating marker of androgenicity because it acts in a paracrine or intracrine fashion and is quickly metabolized into other products before entering the blood circulation. 3a-diol G is among the major metabolites of DHT and has been widely used as an excellent serum marker of 5a-reductase activity and peripheral androgen action (Horton and Lobo, 1986). However, the evidence mainly came from studies on hirsutism and there are no data at present to correlate serum 3a-diol AG with 144 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. intraprostatic DHT levels. In this study, we observed a significant positive association between serum 3a-diol G levels and PSA levels. On the other hand, consistent with an earlier study (Monath et al., 1995), we did not find a significant association between serum total T and PSA levels, nor did we find significant association between free T and PSA levels. These results lend support to the use of 3a-diol G as a marker of peripheral androgen action in the prostate and it may also explain the lack of correlation between T and seminal PSA levels in an earlier study (Giwercman et al., 2004). We also found a significant negative association between serum SHBG and PSA. It is not clear why SHBG, but not free T correlates with serum PSA levels. One possible explanation could be the difference in the reliability of the two measurements. Another possible explanation is that although serum SHBG is an indirect measurement of free T, it is also an indirect measurement of active estrogens. Moreover, both serum androgen and estrogen levels may regulate serum SHBG levels by influencing the hepatic production o f this protein (Edmunds et al., 1990). Therefore, serum SHBG may reflect other aspects o f the hormonal milieu, in addition to free T levels. We also noticed that in our study population, SHBG levels differed significantly between the two study centers. We are currently investigating the reason for the center-difference. We are not sure how it affects our results. Meanwhile, we performed center-specific analysis o f the association between serum PSA and SHBG levels, the results were also most identical in the two strata, suggesting the center difference in SHBG levels did not have an effect on the ordinal relationship between serum PSA and SHBG. 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In this study, we did not find significant effect modification by serum androgen levels in the associations between PSA levels and the polymorphisms in the PSA. However, as mentioned above, there appeared to be a negative association between CAG repeat lengths and PSA levels among men with relatively lower androgen levels, although the association was not significant. Similarly, although not statistically significant, our data showed that the effects of the PSA SNPs (-5413T/C, -3647T/A and -5307 G/A) were generally more evident (and statistically significant) among men with longer CAG repeats (weaker AR activities), except for SNP -5217 T/A. We have to note that our sample size is small to study gene-gene interaction, especially for African-Americans. If it is confirmed by future studies that the effect of polymorphisms in the AR gene and the PSA gene on PSA production only manifests in a low androgen hormonal environment (or low androgen receptor activity in the case of the PSA SNPs), a possible explanation would be that when the androgen response mechanism is saturated in a high androgenic environment additionally higher AR transactivation activity or higher PSA promoter activity does not lead to an additionally higher PSA production (Mifsud et al., 2001a). Our data also suggested that age might be an effect modifier in the association between the PSA gene polymorphisms and serum PSA levels as also shown by Cramer et al (Cramer et al., 2003), although the interactions were not significant (for -5413T/A) or marginally significant (for -5307 G/A). One interpretation is that the effect of the polymorphisms on PSA gene expression is masked by commonly occurring BPH and latent prostate cancer in elderly men. 146 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. However, stratified analysis by self-reported BPH did not appear to support this hypothesis and further limiting men to those with PSA < 4 ng/mL actually attenuated the association between these SNPs and serum PSA. We have to note that our sample size became small and the results need to be confirmed by other studies. One strength of this study is we tried to systematically investigate three important components of androgen-regulated PSA production. Another strength is that subjects were from a general population, unlike most other previous studies. However, at the same time, our population-based study has its intrinsic limitation in studying PSA production in normal men, that is we do not have accurate assessment of prostate cancer status of these subjects. We therefore excluded men with elevated PSA levels (>10 ng/mL) as an effort to exclude or decrease the number of undiagnosed prostate cancer cases from our study population (to preclude the effect of prostatic structural damage). Although the generally accepted upper limit of normal cutoff point is 4 ng/mL for prostate cancer screening, studies have also found that the predictive value for prostate cancer is low for cutoff points below 10 ng/mL (Stamey, 2001; Stamey et al., 2002) and therefore the generally accepted cutoff point for “highly suspicious” of prostate cancer is 10 ng/mL. Further limiting our study subjects to those with PSA <4 ng/mL resulted in attenuated associations between serum PSA and serum 3a-diol G, SHBG, and the polymorphisms in the PSA gene. Moreover, the association between PSA and SHBG, SNP -4336 8C/9C (in whites), and -5217 T/A (in African-Americans) became no longer significant. Yet, it is difficult to draw conclusion that this is simply because we had less men with 1 4 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. probable prostate cancer in the group. It might as well be because we have less men with BPH after excluding those with PSA >4 ng/mL. In addition, a reduced power could also explain why some associations became no longer significant. We know that BPH is a very important contributor to elevated serum PSA levels. To understand whether the association between higher PSA and higher androgen levels (3a-diol G) is due to elevated PSA production in individual prostatic cell or to prostate hyperplasia, we stratified our analysis by self-report BPH status and found the results were similar in men with and without BPH. However, we also realize self-reported BPH is not a reliable variable. It relates to both the availability of medical care and the severity of symptoms (which does not necessarily correlate with prostate size (Partin, 2000)). Excluding men with PSA>4 ng/mL is another way to exclude men with BPH (about 30% of men with BPH have PSA levels greater than 4 ng/mL). The fact that we observed an attenuated association between PSA levels and serum 3a-diol G, SHBG, and the polymorphisms in the PSA gene after excluding men with PSA >4 ng/mL perhaps also suggests prostate hyperplasia as part of the reason for elevated serum PSA levels caused by these hormonal and genetic factors. Another limitation of our study is that serum PSA levels were used as the surrogate marker o f PSA production in the prostate. There is a lack of data on the correlation between serum PSA levels and PSA in the prostate tissue. However, to our knowledge, at present there is no good method to study intraprostatic PSA expression and androgen levels or the genetic variants in PSA gene in vivo. 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.5. Conclusion In our study population, genetic variants in the PSA gene enhancer region and androgen levels (as measured by 3a-diol G) contribute significantly to variation in serum PSA levels. Future studies on the PSA gene polymorphisms and prostate cancer risk should include one of the three tightly linked functional SNPs (-5429 T/G, -5429T/C and -4643 A/G) and, at least until more functional data become available, the-3647 T/A SNP. The SNPs -5307 G/A and -5217 T/A may also be important if African-Americans are studied. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 5. SUM M ARY AND FUTURE STUDY PLANS My dissertation is mainly comprised of two pieces of work. The first part is a population-based case-control study which examines possible associations between genetic variations in the androgen receptor gene and one of its down-stream genes, the PSA gene, and breast cancer risk. When this case-control study started, our knowledge of sequence variation in the PSA gene was very limited. The commonly studied polymorphism was the -158 G/A, which had been postulated to be of functional significance in the androgen-regulated PSA expression because it is located in the most potent androgen response element, ARE-1. However, this polymorphism is not significantly associated with breast cancer risk in this study. What could be the possible explanations for the negative finding? New evidence has pointed to one very likely explanation: this polymorphism may not be functional. In a later study, new polymorphisms in a 5’ enhancer region were identified, with some of them being shown to be functional and associated with serum PSA levels. These new findings led us to carry out a genotype-phenotype correlation study of PSA polymorphisms and serum PSA levels, which is the second part of my dissertation. In this genotype-phenotype study, we confirmed that these functional polymorphisms in the enhancer region, but not the ARE-I polymorphism, are important predictors of serum PSA levels. In addition, three other polymorphisms may be important as well (two in African-Americans specifically), although their functional significance is unknown at present. Now the question is: should we 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. examine these new polymorphisms in breast cancer? This question is complicated by the reported tissue specificity of the enhancer region in PSA gene expression, i.e. it probably does not play an important role in PSA expression in mammary cells. In our opinion, until more is known about this enhancer region, we are not ready to perform studies o f the functional polymorphisms in this region in breast cancer. Although a role of PSA gene in breast cancer development and progression can not be ruled out based on our negative finding o f the ARE-I polymorphism and breast cancer, I also realize that extending our studies to other genes in the androgen- signaling pathway will be o f great importance in our understanding o f the role of androgen in breast cancer. The next gene on our list is the p21 gene, which is a critical component in the cell cycle control network. One task in my postdoctoral training (funded by a dissertation grant from California Breast Cancer Research Program) is to examine possible associations between polymorphisms in the p21 gene, together with the androgen receptor gene CAG polymorphism, and breast cancer risk. At present, our understanding of androgens in breast cancer is greatly hampered by that fact that very little is known about the down-stream genes in the androgen-signaling pathway, especially in mammary cells. Therefore, I hope to collaborate with researchers in Molecular Biology Department at USC to search for these genes and to understand more about androgen-signaling in mammary cells. My hope is that this way we would be able to identify some good candidate genes to investigate in epidemiological studies. My future research work would then be 151 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. expanded to studies of polymorphisms in these candidate genes in breast cancer. I believe that confirmed associations between functional polymorphisms in these genes will not only promote our understanding of the role of androgens in breast cancer, but also point to mechanisms and possible intervention. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REFERENCES Alanen, K. A., Kuopio, T., Collan, Y. U., Kronqvist, P., Juntti, L., and Nevalainen, T. J. (1999). Immunohistochemical labelling for prostate-specific antigen in breast carcinomas. Breast Cancer Res Treat 56, 169-176. Ando, S., De Amicis, F., Rago, V., Carpino, A., Maggiolini, M., Panno, M. L., and Lanzino, M. (2002). Breast cancer: from estrogen to androgen receptor. Mol Cell Endocrinol 193, 121-128. Armbruster, D. A. (1993). Prostate-specific antigen: biochemistry, analytical methods, and clinical application. Clin Chem 39, 181-195. Baas, F., Bikker, H., van Ommen, G. J., and de Vijlder, J. J. (1984). Unusual scarcity of restriction site polymorphism in the human thyroglobulin gene. A linkage study suggesting autosomal dominance of a defective thyroglobulin allele. Hum Genet 67, 301-305. Balk, S. P., Ko, Y. J., and Bubley, G. J. (2003). Biology of prostate-specific antigen. J Clin Oncol 27,383-391. Barrett, J. C., Fry, B., Mailer, J., and Daly, M. J. (2005). Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 21, 263-265. Beilin, J., Ball, E. M., Favaloro, J. M., and Zajac, J. D. (2000). Effect o f the androgen receptor CAG repeat polymorphism on transcriptional activity: specificity in prostate and non-prostate cell lines. Journal of Molecular Endocrinology 25, 85- 96. Beilin, J., Harewood, L., Frydenberg, M., Mameghan, H., Martyres, R. F., Farish, S. J., Yue, C., Deam, D. R., Byron, K. A., and Zajac, J. D. (2001). A case-control study of the androgen receptor gene CAG repeat polymorphism in Australian prostate carcinoma subjects. Cancer 92, 941-949. Berrino, F., Muti, P., Micheli, A., Bolelli, G., Krogh, V., Sciajno, R., Pisani, P., Panico, S., and Secreto, G. (1996). Serum sex hormone levels after menopause and subsequent breast cancer. J Natl Cancer Inst 88, 291-296. Bharaj, B., Scorilas, A., Diamandis, E. P., Giai, M., Levesque, M. A., Sutherland, D. J., and Hoffman, B. R. (2000). Breast cancer prognostic significance of a single nucleotide polymorphism in the proximal androgen response element of the prostate specific antigen gene promoter. Breast Cancer Res Treat 61, 111-119. 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Birrell, S. N., Bentel, J. M., Hickey, T. E., Ricciardelli, C., Weger, M. A., Horsfall, D. J., and Tilley, W. D. (1995a). Androgens induce divergent proliferative responses in human breast cancer cell lines. J Steroid Biochem Mol Biol 52, 459-467. Birrell, S. N., Roder, D. M., Horsfall, D. J., Bentel, J. M., and Tilley, W. D. (1995b). Medroxyprogesterone acetate therapy in advanced breast cancer: the predictive value of androgen receptor expression. J Clin Oncol 1 3 ,1572-1577. Black, M. H., Giai, M., Ponzone, R., Sismondi, P., Yu, H., and Diamandis, E. P. (2000). Serum total and free prostate-specific antigen for breast cancer diagnosis in women. Clin Cancer Res 6, 461-413. Boccuzzi, G., Brignardello, E., Di Monaco, M., Gatto, V., Leonardi, L., Pizzini, A., and Gallo, M. (1994). 5-En-androstene-3 beta, 17 beta-diol inhibits the growth of MCF-7 breast cancer cells when oestrogen receptors are blocked by oestradiol. Br J Cancer 70, 1035-1039. Bratt, O., Borg, A., Kristoffersson, U., Lundgren, R., Zhang, Q. X., and Olsson, H. (1999). CAG repeat length in the androgen receptor gene is related to age at diagnosis of prostate cancer and response to endocrine therapy, but not to prostate cancer risk. Br J Cancer 81, 672-676. Brawer, M. K. (2000). Prostate-Specific Antigen. In Prostatic diseases, H. Lepor, ed. (W.B. Saunders Company), pp. 381-390. Brinkmann, A. O., Faber, P. W., van Rooij, H. C., Kuiper, G. G., Ris, C., Klaassen, P., van der Korput, J. A., Voorhorst, M. M., van Laar, J. H., Mulder, E., and et al. (1989a). The human androgen receptor: domain structure, genomic organization and regulation of expression. J Steroid Biochem 34, 307-310. Brinkmann, A. O., Klaasen, P., Kuiper, G. G., van der Korput, J. A., Bolt, J., de Boer, W., Smit, A., Faber, P. W., van Rooij, H. C., Geurts van Kessel, A., and et al. (1989b). Structure and function of the androgen receptor. Urol Res 17, 87-93. Brown, C. J., Goss, S. J., Lubahn, D. B., Joseph, D. R., Wilson, E. M., French, F. S., and Willard, H. F. (1989). Androgen receptor locus on the human X chromosome: regional localization to Xql 1-12 and description of a DNA polymorphism. Am J Hum Genet 44, 264-269. Bryan, R. M., Mercer, R. J., Bennett, R. C., Rennie, G. C., Lie, T. H., and Morgan, F. J. (1984). Androgen receptors in breast cancer. Cancer 54, 2436-2440. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Buchanan, G., Yang, M., Cheong, A., Harris, J. M., Irvine, R. A., Lambert, P. F., Moore, N. L., Raynor, M., Neufing, P. J., Coetzee, G. A., and Tilley, W. D. (2004). Structural and functional consequences of glutamine tract variation in the androgen receptor. Hum Mol Genet 13, 1677-1692. Callewaert, L., Christiaens, V., Haelens, A., Verrijdt, G., Verhoeven, G., and Claessens, F. (2003). Implications of a polyglutamine tract in the function of the human androgen receptor. Biochemical & Biophysical Research Communications 306, 46-52. Catzavelos, C., Bhattacharya, N., Ung, Y. C., Wilson, J. A., Roncari, L., Sandhu, C., Shaw, P., Yeger, H., Morava-Protzner, I., Kapusta, L., et al. (1997). Decreased levels of the cell-cycle inhibitor p27Kipl protein: prognostic implications in primary breast cancer. Nat Med 3, 227-230. Cauley, J. A., Lucas, F. L., Kuller, L. H., Stone, K., Browner, W., and Cummings, S. R. (1999). Elevated serum estradiol and testosterone concentrations are associated with a high risk for breast cancer. Study of Osteoporotic Fractures Research Group. Ann Intern Med 130, 270-277. Chamberlain, N. L., Driver, E. D., and Miesfeld, R. L. (1994). The length and location o f CAG trinucleotide repeats in the androgen receptor N-terminal domain affect transactivation function. Nucleic Acids Res 22, 3181-3186. Chang, C. S., Kokontis, J., and Liao, S. T. (1988). Structural analysis of complementary DNA and amino acid sequences of human and rat androgen receptors. Proc Natl Acad Sci U S A 85, 7211-7215. Chlebowski, R. T., Hendrix, S. L., Langer, R. D., Stefanick, M. L., Gass, M., Lane, D., Rodabough, R. J., Gilligan, M. A., Cyr, M. G., Thomson, C. A., et al. (2003). Influence of estrogen plus progestin on breast cancer and mammography in healthy postmenopausal women: the Women's Health Initiative Randomized Trial. Jama 289, 3243-3253. Chu, J. S., Huang, C. S., and Chang, K. J. (1999). p27 expression as a prognostic factor of breast cancer in Taiwan. Cancer Lett 141, 123-130. Clements, J., and Mukhtar, A. (1994). Glandular kallikreins and prostate-specific antigen are expressed in the human endometrium. J Clin Endocrinol Metab 78, 1536- 1539. Cleutjens, C. B., Steketee, K., van Eekelen, C. C., van der Korput, J. A., Brinkmann, A. O., and Trapman, J. (1997a). Both androgen receptor and glucocorticoid receptor are able to induce prostate-specific antigen expression, but differ in their growth- stimulating properties of LNCaP cells. Endocrinology 138, 5293-5300. 155 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cleutjens, K. B., van der Korput, H. A., van Eekelen, C. C., van Rooij, H. C., Faber, P. W., and Trapman, J. (1997b). An androgen response element in a far upstream enhancer region is essential for high, androgen-regulated activity o f the prostate- specific antigen promoter. Mol Endocrinol 11, 148-161. Cleutjens, K. B., van Eekelen, C. C., van der Korput, H. A., Brinkmann, A. O., and Trapman, J. (1996). Two androgen response regions cooperate in steroid hormone regulated activity of the prostate-specific antigen promoter. J Biol Chem 271, 6379- 6388. Cohen, P., Peehl, D. M., Graves, H. C., and Rosenfeld, R. G. (1994). Biological effects of prostate specific antigen as an insulin-like growth factor binding protein-3 protease. J Endocrinol 142, 407-415. Colditz, G. A., Rosner, B. A., Chen, W. Y., Holmes, M. D., and Hankinson, S. E. (2004). Risk factors for breast cancer according to estrogen and progesterone receptor status. J Natl Cancer Inst 96, 218-228. Collings, B. J., and Hamilton, M. A. (1988). Estimating the power o f the two-sample Wilcoxon test for location shift. Biometrics 44, 847-860. Collins, G. N., Lee, R. J., McKelvie, G. B., Rogers, A. C., and Hehir, M. (1993). Relationship between prostate specific antigen, prostate volume and age in the benign prostate. Br J Urol 71, 445-450. Conti, D. V., and Gauderman, W. J. (2004). SNPs, haplotypes, and model selection in a candidate gene region: the SIMPle analysis for multilocus data. Genet Epidemiol 27, 429-441. Cooper, C. S., Perry, P. J., Sparks, A. E., Maclndoe, J. H., Yates, W. R., and Williams, R. D. (1998). Effect of exogenous testosterone on prostate volume, serum and semen prostate specific antigen levels in healthy young men. J Urol 159, 441- 443. Correa-Cerro, L., Wohr, G., Haussler, J., Berthon, P., Drelon, E., Mangin, P., Fournier, G., Cussenot, O., Kraus, P., Just, W., et al. (1999). (CAG)nCAA and GGN repeats in the human androgen receptor gene are not associated with prostate cancer in a French-German population. Eur J Hum Genet 7, 357-362. Cramer, S. D., Chang, B. L., Rao, A., Hawkins, G. A., Zheng, S. L., Wade, W. N., Cooke, R. T., Thomas, L. N., Bleecker, E. R., Catalona, W. J., et al. (2003). Association between genetic polymorphisms in the prostate-specific antigen gene promoter and serum prostate-specific antigen levels. J Natl Cancer Inst 95, 1044- 1053. 156 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cramer, S. D., Chen, Z., and Peehl, D. M. (1996). Prostate specific antigen cleaves parathyroid hormone-related protein in the PTH-like domain: inactivation of PTHrP- stimulated cAMP accumulation in mouse osteoblasts. J Urol 156, 526-531. Dallas, S. L., Zhao, S., Cramer, S. D., Chen, Z., Peehl, D. M., and Bonewald, L. F. (2005). Preferential production of latent transforming growth factor beta-2 by primary prostatic epithelial cells and its activation by prostate-specific antigen. J Cell Physiol 202, 361-370. Dauvois, S., Geng, C. S., Levesque, C., Merand, Y., and Labrie, F. (1991). Additive inhibitory effects o f an androgen and the antiestrogen EM -170 on estradiol- stimulated growth of human ZR-75-1 breast tumors in athymic mice. Cancer Res 51, 3131-3135. Dauvois, S., Li, S. M., Martel, C., and Labrie, F. (1989). Inhibitory effect of androgens on DMBA-induced mammary carcinoma in the rat. Breast Cancer Res Treat 14, 299-306. de Launoit, Y., Dauvois, S., Dufour, M., Simard, J., and Labrie, F. (1991). Inhibition of cell cycle kinetics and proliferation by the androgen 5 alpha-dihydrotestosterone and antiestrogen N,n-butyl-N-methyl-11-[16' alpha-chloro-3',17 beta-dihydroxy- estra-r,3',5'-(10')triene-7' alpha-yl] undecanamide in human breast cancer ZR-75-1 cells. Cancer Res 51, 2797-2802. de Ronde, W., van der Schouw, Y. T., Pierik, F. H., Pols, H. A., Muller, M., Grobbee, D. E., Gooren, L. J., Weber, R. F., and de Jong, F. H. (2005). Serum levels of sex hormone-binding globulin (SHBG) are not associated with lower levels of non-SHBG-bound testosterone in male newborns and healthy adult men. Clin Endocrinol (Oxf) 62, 498-503. Diamandis, E. P., Yu, H., and Melegos, D. N. (1996). Ultrasensitive prostate-specific antigen assays and their clinical application. Clin Chem 42, 853-857. Diamandis, E. P., Yu, H., and Sutherland, D. J. (1994). Detection o f prostate-specific antigen immunoreactivity in breast tumors. Breast Cancer Res Treat 22, 301-310. Dimitrakakis, C., Zhou, J., Wang, J., Belanger, A., LaBrie, F., Cheng, C., Powell, D., and Bondy, C. (2003). A physiologic role for testosterone in limiting estrogenic stimulation of the breast. Menopause 10, 292-298. Dorgan, J. F., Longcope, C., Stanczyk, F. Z., Stephenson, H. E., Jr., and Hoover, R. N. (1999). Re: Plasma sex steroid hormone levels and risk of breast cancer in postmenopausal women. J Natl Cancer Inst 91, 380-381. 157 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Dowsing, A. T., Yong, E. L., Clark, M., McLachlan, R. I., de Kretser, D. M., and Trounson, A. O. (1999). Linkage between male infertility and trinucleotide repeat expansion in the androgen-receptor gene. Lancet 354, 640-643. Dunbar, M. E., and Wysolmerski, J. J. (1999). Parathyroid hormone-related protein: a developmental regulatory molecule necessary for mammary gland development. J Mammary Gland Biol Neoplasia 4, 21-34. Dunn, J. F., Nisula, B. C., and Rodbard, D. (1981). Transport of steroid hormones: binding of 21 endogenous steroids to both testosterone-binding globulin and corticosteroid-binding globulin in human plasma. J Clin Endocrinol Metab 53, 58- 68 . Dunning, A. M., McBride, S., Gregory, J., Durocher, F., Foster, N. A., Healey, C. S., Smith, N., Pharoah, P. D., Luben, R. N., Easton, D. F., and Ponder, B. A. (1999). No association between androgen or vitamin D receptor gene polymorphisms and risk of breast cancer. Carcinogenesis 20, 2131-2135. Edmunds, S. E., Stubbs, A. P., Santos, A. A., and Wilkinson, M. L. (1990). Estrogen and androgen regulation o f sex hormone binding globulin secretion by a human liver cell line. J Steroid Biochem Mol Biol 37, 733-739. Edwards, A., Hammond, H. A., Jin, L., Caskey, C. T., and Chakraborty, R. (1992). Genetic variation at five trimeric and tetrameric tandem repeat loci in four human population groups. Genomics 12, 241-253. Elhaji, Y. A., Gottlieb, B., Lumbroso, R., Beitel, L. K., Foulkes, W. D., Pinsky, L., and Trifiro, M. A. (2001). The polymorphic CAG repeat of the androgen receptor gene: a potential role in breast cancer in women over 40. Breast Cancer Res Treat 70, 109-116. Enserink, M. (2005). Let's talk about sex— and drugs. Science 308, 1578. Faber, P. W., Kuiper, G. G., van Rooij, H. C., van der Korput, J. A., Brinkmann, A. O., and Trapman, J. (1989). The N-terminal domain of the human androgen receptor is encoded by one, large exon. Mol Cell Endocrinol 61, 257-262. Fernandez, P. L., Jares, P., Rey, M. J., Campo, E., and Cardesa, A. (1998). Cell cycle regulators and their abnormalities in breast cancer. Mol Pathol 51, 305-309. Fortier, A. H., Nelson, B. J., Grella, D. K., and Holaday, J. W. (1999). Antiangiogenic activity of prostate-specific antigen. J Natl Cancer Inst 91, 1635- 1640. 158 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gapstur, S. M., Gann, P. H., Kopp, P., Colangelo, L., Longcope, C., and Liu, K. (2002). Serum androgen concentrations in young men: a longitudinal analysis of associations with age, obesity, and race. The CARDIA male hormone study. Cancer Epidemiol Biomarkers Prev 11, 1041-1047. Garcia-Closas, M., Egan, K. M., Abruzzo, J., Newcomb, P. A., Titus-Emstoff, L., Franklin, T., Bender, P. K., Beck, J. C., Le Marchand, L., Lum, A., et al. (2001). Collection of genomic DNA from adults in epidemiological studies by buccal cytobrush and mouthwash. Cancer Epidemiol Biomarkers Prev 10, 687-696. Gerstenbluth, R. E., Maniam, P. N., Corty, E. W., and Seftel, A. D. (2002). Prostate- specific antigen changes in hypogonadal men treated with testosterone replacement. J Androl 23, 922-926. Giguere, Y., Dewailly, E., Brisson, J., Ayotte, P., Laflamme, N., Demers, A., Forest, V. I., Dodin, S., Robert, J., and Rousseau, F. (2001). Short polyglutamine tracts in the androgen receptor are protective against breast cancer in the general population. Cancer Res 61, 5869-5874. Giovannucci, E. (2002). Is the androgen receptor CAG repeat length significant for prostate cancer? Cancer Epidemiol Biomarkers Prev 11, 985-986. Giovannucci, E., Stampfer, M. J., Krithivas, K., Brown, M., Dahl, D., Brufsky, A., Talcott, J., Hennekens, C. H., and Kantoff, P. W. (1997). The CAG repeat within the androgen receptor gene and its relationship to prostate cancer. Proc Natl Acad Sci U S A 94, 3320-3323. Giwercman, Y. L., Richthoff, J., Lilja, H., Anderberg, C., Abrahamsson, P. A., and Giwercman, A. (2004). Androgen receptor CAG repeat length correlates with semen PSA levels in adolescence. Prostate 59, 227-233. Gnanapragasam, V. J., Robson, C. N., Leung, H. Y., and Neal, D. E. (2000). Androgen receptor signalling in the prostate. BJU Int 86, 1001-1013. Goebelsmann, U., Arce, J. J., Thomeycroft, I. H., and Mishell, D. R., Jr. (1974). Serum testosterone concentrations in women throughout the menstrual cycle and following HCG administration. Am J Obstet Gynecol 119, 445-452. Goebelsmann, U., Bernstein, G. S., Gale, J. A., Kletzky, O. A., Nakamura, R. M., A.H., C., and Korelitz, J., J, (1979). Serum gonadotropin, testosterone, estradiol and estrone levels prior to and following bilateral vasectomy. In Vasectomy: Immunologic and pathophysiologic effects in animals and man, L. I.H., and R. Crozier, eds. (New York, Academic Press), pp. 165-175. 159 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gormley, G. J., Ng, J., Cook, T., Stoner, E., Guess, H., and Walsh, P. (1994). Effect of finasteride on prostate-specific antigen density. Urology 43, 53-58; discussion SB- 59. Greeve, M. A., Allan, R. K., Harvey, J. M., and Bentel, J. M. (2004). Inhibition of MCF-7 breast cancer cell proliferation by 5alpha-dihydrotestosterone; a role for p21(Cipl/W afl). J Mol Endocrinol 32, 793-810. Griniatsos, J., Diamantis, E., Gioti, J., Karyda, I., Vassilopoulos, P. P., and Agnanti, N. (1998). Correlation of prostate specific antigen immunoactivity (IR-PSA) to other prognostic factors in female breast cancer. Anticancer Res 18, 683-688. Gsur, A., Preyer, M., Haidinger, G., Zidek, T., Madersbacher, S., Schatzl, G., Marberger, M., Vutuc, C., and Micksche, M. (2002). Polymorphic CAG repeats in the androgen receptor gene, prostate-specific antigen polymorphism and prostate cancer risk. Carcinogenesis 23, 1647-1651. Guo, S. W., and Thompson, E. A. (1992). Performing the exact test of Hardy- Weinberg proportion for multiple alleles. Biometrics 48, 361-372. Hackenberg, R., Hofmann, J., Holzel, F., and Schulz, K. D. (1988). Stimulatory effects of androgen and antiandrogen on the in vitro proliferation of human mammary carcinoma cells. J Cancer Res Clin Oncol 114, 593-601. Hackenberg, R., and Schulz, K. D. (1996). Androgen receptor mediated growth control o f breast cancer and endometrial cancer modulated by antiandrogen- and androgen-like steroids. Journal of Steroid Biochemistry & Molecular Biology 56, 113-117. Haiman, C. A., Brown, M., Hankinson, S. E., Spiegelman, D., Colditz, G. A., Willett, W. C., Kantoff, P. W., and Hunter, D. J. (2002). The androgen receptor CAG repeat polymorphism and risk of breast cancer in the Nurses' Health Study. Cancer Res 62, 1045-1049. Hankinson, S. E., Willett, W. C., Colditz, G. A., Hunter, D. J., Michaud, D. S., Deroo, B., Rosner, B., Speizer, F. E., and Poliak, M. (1998a). Circulating concentrations of insulin-like growth factor-I and risk o f breast cancer. Lancet 351, 1393-1396. Hankinson, S. E., Willett, W. C., Manson, J. E., Colditz, G. A., Hunter, D. J., Spiegelman, D., Barbieri, R. L., and Speizer, F. E. (1998b). Plasma sex steroid hormone levels and risk of breast cancer in postmenopausal women. J Natl Cancer Inst 90, 1292-1299. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Haraldsson, K., Loman, N., Zhang, Q. X., Johannsson, O., Olsson, H., and Borg, A. (1998). BRCA2 germ-line mutations are frequent in male breast cancer patients without a family history of the disease. Cancer Res 58, 1367-1371. Henttu, P., Liao, S. S., and Vihko, P. (1992). Androgens up-regulate the human prostate-specific antigen messenger ribonucleic acid (mRNA), but down-regulate the prostatic acid phosphatase mRNA in the LNCaP cell line. Endocrinology 130, 766- 772. Heyl, W., Wolff, J. M., Biesterfeld, S., Schroder, W., Zitzelsberger, D., Jakse, G., and Rath, W. (1999). Immunohistochemical analysis of prostate-specific antigen does not correlate to other prognostic factors in breast cancer. Anticancer Res 19, 2563-2565. Horton, R., and Lobo, R. (1986). Peripheral androgens and the role of androstanediol glucuronide. Clin Endocrinol Metab 15, 293-306. Howarth, D. J., Aronson, I. B., and Diamandis, E. P. (1997). Immunohistochemical localization of prostate-specific antigen in benign and malignant breast tissues. Br J Cancer 75, 1646-1651. Hsing, A. W., Gao, Y. T., Wu, G., Wang, X., Deng, J., Chen, Y. L., Sesterhenn, I. A., Mostofi, F. K., Benichou, J., and Chang, C. (2000). Polymorphic CAG and GGN repeat lengths in the androgen receptor gene and prostate cancer risk: a population- based case-control study in China. Cancer Res 60, 5111-5116. Huang, W., Shostak, Y., Tarr, P., Sawyers, C., and Carey, M. (1999). Cooperative assembly of androgen receptor into a nucleoprotein complex that regulates the prostate-specific antigen enhancer. J Biol Chem 274, 25756-25768. Ingle, J. N., Twito, D. I., Schaid, D. J., Cullinan, S. A., Krook, J. E., Mailliard, J. A., Tschetter, L. K., Long, H. J., Gerstner, J. G., Windschitl, H. E., and et al. (1991). Combination hormonal therapy with tamoxifen plus fluoxymesterone versus tamoxifen alone in postmenopausal women with metastatic breast cancer. An updated analysis. Cancer 67, 886-891. Ingles, S. A., Ross, R. K., Yu, M. C., Irvine, R. A., La Pera, G., Haile, R. W., and Coetzee, G. A. (1997). Association o f prostate cancer risk with genetic polymorphisms in vitamin D receptor and androgen receptor. J Natl Cancer Inst 89, 166-170. Irvine, R. A., Ma, H., Yu, M. C., Ross, R. K., Stallcup, M. R., and Coetzee, G. A. (2000). Inhibition of pl60-mediated coactivation with increasing androgen receptor polyglutamine length. Hum Mol Genet 9, 267-274. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Irvine, R. A., Yu, M. C., Ross, R. K., and Coetzee, G. A. (1995). The CAG and GGC microsatellites of the androgen receptor gene are in linkage disequilibrium in men with prostate cancer. Cancer Res 55,1937-1940. Iwamura, M., Heilman, J., Cockett, A. T., Lilja, H., and Gershagen, S. (1996). Alteration of the hormonal bioactivity of parathyroid hormone-related protein (PTHrP) as a result of limited proteolysis by prostate-specific antigen. Urology 48, 317-325. Janulis, L., Grayhack, J. T., and Lee, C. (2000). Endocrinology of the prostate. In Prostatic diseases, H. Lepor, ed. (W.B. Saunders Company), pp. 58-74. Judd, H. L., and Yen, S. S. (1973). Serum androstenedione and testosterone levels during the menstrual cycle. J Clin Endocrinol Metab 36, 475-481. Kaaks, R., Berrino, F., Key, T., Rinaldi, S., Dossus, L., Biessy, C., Secreto, G., Amiano, P., Bingham, S., Boeing, H., et al. (2005). Serum sex steroids in premenopausal women and breast cancer risk within the European Prospective Investigation into Cancer and Nutrition (EPIC). J Natl Cancer Inst 97, 755-765. Kadouri, L., Easton, D. F., Edwards, S., Hubert, A., Kote-Jarai, Z., Glaser, B., Durocher, F., Abeliovich, D., Peretz, T., and Eeles, R. A. (2001). CAG and GGC repeat polymorphisms in the androgen receptor gene and breast cancer susceptibility in BRCA1/2 carriers and non-carriers. Br J Cancer 85, 36-40. Kakonen, S. M., and Mundy, G. R. (2003). Mechanisms o f osteolytic bone metastases in breast carcinoma. Cancer 97, 834-839. Kandouz, M., Lombet, A., Perrot, J. Y., Jacob, D., Carvajal, S., Kazem, A., Rostene, W., Therwath, A., and Gompel, A. (1999). Proapoptotic effects of antiestrogens, progestins and androgen in breast cancer cells. J Steroid Biochem Mol Biol 69, 463- 471. Kazemi-Esfaijani, P., Trifiro, M. A., and Pinsky, L. (1995). Evidence for a repressive function o f the long polyglutamine tract in the human androgen receptor: possible pathogenetic relevance for the (CAG)n-expanded neuronopathies. Hum Mol Genet 4, 523-527. Killian, C. S., Corral, D. A., Kawinski, E., and Constantine, R. I. (1993). Mitogenic response of osteoblast cells to prostate-specific antigen suggests an activation of latent TGF-beta and a proteolytic modulation of cell adhesion receptors. Biochem Biophys Res Commun 192, 940-947. 162 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kristiansen, M., Langerod, A., Knudsen, G. P., Weber, B. L., Borresen-Dale, A. L., and Orstavik, K. H. (2002). High frequency of skewed X inactivation in young breast cancer patients. J Med Genet 39, 30-33. Kroemer, G. (1997). The proto-oncogene Bcl-2 and its role in regulating apoptosis. Nat Med 3,614-620. Kuenen-Boumeester, V., Van der Kwast, T. H., van Putten, W. L., Claassen, C., van Ooijen, B., and Henzen-Logmans, S. C. (1992). Immunohistochemical determination of androgen receptors in relation to oestrogen and progesterone receptors in female breast cancer. Int J Cancer 52, 581-584. La Spada, A. R., Wilson, E. M., Lubahn, D. B., Harding, A. E., and Fischbeck, K. H. (1991). Androgen receptor gene mutations in X-linked spinal and bulbar muscular atrophy. Nature 352, 77-79. Labrie, F., Luu-The, V., Labrie, C., Belanger, A., Simard, J., Lin, S. X., and Pelletier, G. (2003). Endocrine and intracrine sources of androgens in women: inhibition o f breast cancer and other roles of androgens and their precursor dehydroepiandrosterone. Endocr Rev 24, 152-182. Labrie, F., Poulin, R., Simard, J., Zhao, H. F., Labrie, C., Dauvois, S., Dumont, M., Hatton, A. C., Poirier, D., and Merand, Y. (1990). Interactions between estrogens, androgens, progestins, and glucocorticoids in ZR-75-1 human breast cancer cells. Ann N Y Acad Sci 595, 130-148. Lai, L. C., Erbas, H., Lennard, T. W., and Peaston, R. T. (1996). Prostate-specific antigen in breast cyst fluid: possible role o f prostate-specific antigen in hormone- dependent breast cancer. Int J Cancer 66, 743-746. Langer, M., Kubista, E., Schemper, M., and Spona, J. (1990). Androgen receptors, serum androgen levels and survival of breast cancer patients. Arch Gynecol Obstet 247, 203-209. Lanzino, M., De Amicis, F., McPhaul, M. J., Marsico, S., Panno, M. L., and Ando, S. (2005). Endogenous coactivator ARA70 interacts with estrogen receptor alpha (ERalpha) and modulates the functional ERalpha/androgen receptor interplay in MCF-7 cells. J Biol Chem 280, 20421-20430. Lapointe, J., Fournier, A., Richard, V., and Labrie, C. (1999). Androgens down- regulate bcl-2 protooncogene expression in ZR-75-1 human breast cancer cells. Endocrinology 140, 416-421. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Lapointe, J., and Labrie, C. (2001). Role of the cyclin-dependent kinase inhibitor p27(Kipl) in androgen-induced inhibition of CAMA-1 breast cancer cell proliferation. Endocrinology 142, 4331-4338. Latil, A. G., Azzouzi, R., Cancel, G. S., Guillaume, E. C., Cochan-Priollet, B., Berthon, P. L., and Cussenot, O. (2001). Prostate carcinoma risk and allelic variants o f genes involved in androgen biosynthesis and metabolism pathways. Cancer 92, 1130-1137. Lea, O. A., Kvinnsland, S., and Thorsen, T. (1989). Improved measurement of androgen receptors in human breast cancer. Cancer Res 49, 7162-7167. Lieberman, B. A., Bona, B. J., Edwards, D. P., andNordeen, S. K. (1993). The constitution of a progesterone response element. Mol Endocrinol 7, 515-527. Liede, A., Zhang, W., De Leon Matsuda, M. L., Tan, A., and Narod, S. A. (2003). Androgen receptor gene polymorphism and breast cancer susceptibility in The Philippines. Cancer Epidemiol Biomarkers Prev 12, 848-852. Lillie, E. O., Bernstein, L., Ingles, S. A., Gauderman, W. J., Rivas, G. E., Gagalang, V., Krontiris, T., and Ursin, G. (2004). Polymorphism in the androgen receptor and mammographic density in women taking and not taking estrogen and progestin therapy. Cancer Res 64, 1237-1241. Lillie, E. O., Bernstein, L., and Ursin, G. (2003). The role of androgens and polymorphisms in the androgen receptor in the epidemiology of breast cancer. Breast Cancer Res 5, 164-173. Longcope, C. (1986). Adrenal and gonadal androgen secretion in normal females. Clin Endocrinol Metab 15, 213-228. Lu, S., Liu, M., Epner, D. E., Tsai, S. Y., and Tsai, M. J. (1999). Androgen regulation of the cyclin-dependent kinase inhibitor p21 gene through an androgen response element in the proximal promoter. Mol Endocrinol 13, 376-384. Lubahn, D. B., Brown, T. R., Simental, J. A., Higgs, H. N., Migeon, C. J., Wilson, E. M., and French, F. S. (1989). Sequence o f the intron/exon junctions of the coding region of the human androgen receptor gene and identification of a point mutation in a family with complete androgen insensitivity. Proc Natl Acad Sci U S A 86, 9534- 9538. Luke, M. C., and Coffey, D. S. (1994). Human androgen receptor binding to the androgen response element of prostate specific antigen. J Androl 15, 41-51. 164 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. MacLean, H. E., Brown, R. W., Beilin, J., Wame, G. L., and Zajac, J. D. (2004). Increased frequency of long androgen receptor CAG repeats in male breast cancers. Breast Cancer Res Treat 88, 239-246. Manjer, J., Johansson, R., Berglund, G., Janzon, L., Kaaks, R., Agren, A., and Lenner, P. (2003). Postmenopausal breast cancer risk in relation to sex steroid hormones, prolactin and SHBG (Sweden). Cancer Causes Control 14, 599-607. Massafra, C., De Felice, C., Agnusdei, D. P., Gioia, D., and Bagnoli, F. (1999). Androgens and osteocalcin during the menstrual cycle. J Clin Endocrinol Metab 84, 971-974. Massague, J. (2004). G1 cell-cycle control and cancer. Nature 432, 298-306. McConnell, J. D., Wilson, J. D., George, F. W., Geller, J., Pappas, F., and Stoner, E. (1992). Finasteride, an inhibitor of 5 alpha-reductase, suppresses prostatic dihydrotestosterone in men with benign prostatic hyperplasia. J Clin Endocrinol Metab 74, 505-508. Medeiros, R., Morais, A., Vasconcelos, A., Costa, S., Pinto, D., Oliveira, J., Carvalho, R., and Lopes, C. (2002). Linkage between polymorphisms in the prostate specific antigen ARE1 gene region, prostate cancer risk, and circulating tumor cells. Prostate 53, 88-94. Melegos, D. N., Yu, H., Ashok, M., Wang, C., Stanczyk, F., and Diamandis, E. P. (1997). Prostate-specific antigen in female serum, a potential new marker of androgen excess. J Clin Endocrinol Metab 82, 777-780. Menin, C., Banna, G. L., De Salvo, G., Lazzarotto, V., De Nicolo, A., Agata, S., Montagna, M., Sordi, G., Nicoletto, O., Chieco-Bianchi, L., and D'Andrea, E. (2001). Lack o f association between androgen receptor CAG polymorphism and familial breast/ovarian cancer. Cancer Lett 168, 31-36. Mifsud, A., Choon, A. T., Fang, D., and Yong, E. L. (2001a). Prostate-specific antigen, testosterone, sex-hormone binding globulin and androgen receptor CAG repeat polymorphisms in subfertile and normal men. Mol Hum Reprod 7, 1007-1013. Mifsud, A., Sim, C. K., Boettger-Tong, H., Moreira, S., Lamb, D. J., Lipshultz, L. I., and Yong, E. L. (2001b). Trinucleotide (CAG) repeat polymorphisms in the androgen receptor gene: molecular markers of risk for male infertility. Fertil Steril 75, 275-281. Miller, E. A., Stanford, J. L., Hsu, L., Noonan, E., and Ostrander, E. A. (2001). Polymorphic repeats in the androgen receptor gene in high-risk sibships. Prostate 48, 200-205. 165 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Monath, J. R., McCullough, D. L., Hart, L. J., and Jarow, J. P. (1995). Physiologic variations o f serum testosterone within the normal range do not affect serum prostate-specific antigen. Urology 46, 58-61. Morgan, T. O., Jacobsen, S. J., McCarthy, W. F., Jacobson, D. J., McLeod, D. G., and Moul, J. W. (1996). Age-specific reference ranges for prostate-specific antigen in black men. N Engl J Med 335, 304-310. Narang, R., Rao, J., Savjani, G., Peterson, J., Gentzschein, E., and Stanczyk, F. Z. (1991). Radioimmunoassay kit for the quantitative measurement of androstanediol glucuronide in unextracted serum. Paper presented at: 17th National Meeting of the Clinical Ligand Assay Society (Chicago, IL). Negri, C., Tosi, F., Dorizzi, R., Fortunato, A., Spiazzi, G. G., Muggeo, M., Castello, R., and Moghetti, P. (2000). Antiandrogen drugs lower serum prostate-specific antigen (PSA) levels in hirsute subjects: evidence that serum PSA is a marker of androgen action in women. J Clin Endocrinol Metab 85, 81-84. Newman, B., Mu, H., Butler, L. M., Millikan, R. C., Moorman, P. G., and King, M. C. (1998). Frequency of breast cancer attributable to BRCA1 in a population-based series of American women. Jama 279, 915-921. Obiezu, C. V., Giltay, E. J., Magklara, A., Scorilas, A., Gooren, L. J., Yu, H., Howarth, D. J., and Diamandis, E. P. (2000). Serum and urinary prostate-specific antigen and urinary human glandular kallikrein concentrations are significantly increased after testosterone administration in female-to-male transsexuals. Clin Chem 46, 859-862. Oesterling, J. E., Jacobsen, S. J., Chute, C. G., Guess, H. A., Girman, C. J., Panser, L. A., and Lieber, M. M. (1993). Serum prostate-specific antigen in a community- based population of healthy men. Establishment o f age-specific reference ranges. Jama 270, 860-864. Olopade, O. I., Fackenthal, J. D., Dunston, G., Tainsky, M. A., Collins, F., and Whitfield-Broome, C. (2003). Breast cancer genetics in African Americans. Cancer 97, 236-245. Ortmann, J., Priflti, S., Bohlmann, M. K., Rehberger-Schneider, S., Strowitzki, T., and Rabe, T. (2002). Testosterone and 5 alpha-dihydrotestosterone inhibit in vitro growth o f human breast cancer cell lines. Gynecol Endocrinol 16, 113-120. Park, J. J., Irvine, R. A., Buchanan, G., Koh, S. S., Park, J. M., Tilley, W. D., Stallcup, M. R., Press, M. F., and Coetzee, G. A. (2000). Breast cancer susceptibility gene 1 (BRCAI) is a coactivator of the androgen receptor. Cancer Res 60, 5946- 5949. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Parkin, D. M., Whelan, S. L., Ferlay, J., Teppo, L., and Thomas, D. B. (2002). Cancer incidence in five continents. Volume VIII. IARC Scientific Publications, 1- 781. Partin, A. W. (2000). Benign Prostatic Hyperplasia: Etiology of Benign Prostatic Hyperplasia. In Prostatic Diseases, H. Lepor, ed. (W.B. Saunders Company), pp. 95- 105. Pike, M. C., Kolonel, L. N., Henderson, B. E., Wilkens, L. R., Hankin, J. H., Feigelson, H. S., Wan, P. C., Stram, D. O., and Nomura, A. M. (2002). Breast cancer in a multiethnic cohort in Hawaii and Los Angeles: risk factor-adjusted incidence in Japanese equals and in Hawaiians exceeds that in whites. Cancer Epidemiol Biomarkers Prev 11, 795-800. Poulin, R., Baker, D., and Labrie, F. (1988). Androgens inhibit basal and estrogen- induced cell proliferation in the ZR-75-1 human breast cancer cell line. Breast Cancer Res Treat 12, 213-225. Poulin, R., Simard, J., Labrie, C., Petitclerc, L., Dumont, M., Lagace, L., and Labrie, F. (1989). Down-regulation of estrogen receptors by androgens in the ZR-75-1 human breast cancer cell line. Endocrinology 125, 392-399. Probst-Hensch, N. M., Ingles, S. A., Diep, A. T., Haile, R. W., Stanczyk, F. Z., Kolonel, L. N., and Henderson, B. E. (1999). Aromatase and breast cancer susceptibility. EndocrRelat Cancer 6, 165-173. Rao, A., Chang, B. L., Hawkins, G., Hu, J. J., Rosser, C. J., Hall, M. C., Meyers, D. A., Xu, J., and Cramer, S. D. (2003). Analysis of G/A polymorphism in the androgen response element I o f the PSA gene and its interactions with the androgen receptor polymorphisms. Urology 61, 864-869. Rao, A., and Cramer, S. D. (1999). Identification of a polymorphism in the ARE 1 region of the PSA promotor. Paper presented at: Proc. Am. Assoc. Cancer R es.,. Rebbeck, T. R., Kantoff, P. W., Krithivas, K., Neuhausen, S., Blackwood, M. A., Godwin, A. K., Daly, M. B., Narod, S. A., Garber, J. E., Lynch, H. T., et al. (1999). Modification of BRC A 1-associated breast cancer risk by the polymorphic androgen- receptor CAG repeat. Am J Hum Genet 64, 1371-1377. Riegman, P. H., Vlietstra, R. J., van der Korput, J. A., Brinkmann, A. O., and Trapman, J. (1991). The promoter of the prostate-specific antigen gene contains a functional androgen responsive element. Mol Endocrinol 5, 1921-1930. 167 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Riegman, P. H., Vlietstra, R. J., van der Korput, J. A., Romijn, J. C., and Trapman, J. (1989). Characterization of the prostate-specific antigen gene: a novel human kallikrein-like gene. Biochem Biophys Res Commun 159, 95-102. Roche, P. J., Hoare, S. A., and Parker, M. G. (1992). A consensus DNA-binding site for the androgen receptor. Mol Endocrinol 6, 2229-2235. Ruizeveld de Winter, J. A., Trapman, J., Vermey, M., Mulder, E., Zegers, N. D., and van der Kwast, T. H. (1991). Androgen receptor expression in human tissues: an immunohistochemical study. J Histochem Cytochem 39, 927-936. Sachdev, D., and Yee, D. (2001). The IGF system and breast cancer. Endocr Relat Cancer 8, 197-209. Salinas, C. A., Austin, M. A., Ostrander, E. O., and Stanford, J. L. (2005). Polymorphisms in the androgen receptor and the prostate-specific antigen genes and prostate cancer risk. Prostate. Sauter, E. R., Daly, M., Linahan, K., Ehya, H., Engstrom, P. F., Bonney, G., Ross, E. A., Yu, H., and Diamandis, E. (1996). Prostate-specific antigen levels in nipple aspirate fluid correlate with breast cancer risk. Cancer Epidemiol Biomarkers Prev 5, 967-970. Sauter, E. R., Klein, G., Wagner-Mann, C., and Diamandis, E. P. (2004). Prostate- specific antigen expression in nipple aspirate fluid is associated with advanced breast cancer. Cancer Detect Prev 28, 27-31. Schuur, E. R., Henderson, G. A., Kmetec, L. A., Miller, J. D., Lamparski, H. G., and Henderson, D. R. (1996). Prostate-specific antigen expression is regulated by an upstream enhancer. J Biol Chem 271, 7043-7051. Secreto, G., Recchione, C., Fariselli, G., and Di Pietro, S. (1984a). High testosterone and low progesterone circulating levels in premenopausal patients with hyperplasia and cancer o f the breast. Cancer Res 44, 841-844. Secreto, G., Toniolo, P., Berrino, F., Recchione, C., Di Pietro, S., Fariselli, G., and Decarli, A. (1984b). Increased androgenic activity and breast cancer risk in premenopausal women. Cancer Res 44, 5902-5905. Secreto, G., Toniolo, P., Pisani, P., Recchione, C., Cavalleri, A., Fariselli, G., Totis, A., Di Pietro, S., and Berrino, F. (1989). Androgens and breast cancer in premenopausal women. Cancer Res 49, 471-476. Secreto, G., and Zumoff, B. (1994). Abnormal production of androgens in women with breast cancer. Anticancer Res 14, 2113-2117. 168 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Sherr, C. J. (1994). G1 phase progression: cycling on cue. Cell 79, 551-555. Sherr, C. J., and Roberts, J. M. (1995). Inhibitors o f mammalian G1 cyclin- dependent kinases. Genes Dev 9, 1149-1163. Skildum, A. J., Mukherjee, S., and Conrad, S. E. (2002). The cyclin-dependent kinase inhibitor p21 W AFl/Cipl is an antiestrogen-regulated inhibitor of Cdk4 in human breast cancer cells. J Biol Chem 277, 5145-5152. Sodergard, R., Backstrom, T., Shanbhag, V., and Carstensen, H. (1982). Calculation of free and bound fractions of testosterone and estradiol-17 beta to human plasma proteins at body temperature. J Steroid Biochem 16, 801-810. Somboonpom, W., and Davis, S. R. (2004). Postmenopausal testosterone therapy and breast cancer risk. Maturitas 49, 267-275. Soreide, J. A., Lea, O. A., Varhaug, J. E., Skarstein, A., and Kvinnsland, S. (1992). Androgen receptors in operable breast cancer: relation to other steroid hormone receptors, correlations to prognostic factors and predictive value for effect of adjuvant tamoxifen treatment. Eur J Surg Oncol 18, 112-118. Spurdle, A. B., Antoniou, A. C., Duffy, D. L., Pandeya, N., Kelemen, L., Chen, X., Peock, S., Cook, M. R., Smith, P. L., Purdie, D. M., et al. (2005). The androgen recetor CAG repeat polymorphism and modification o f breast cancer risk in BRCA1 and BRCA2 mutation carriers. Breast Cancer Res 7, R176-R183. Spurdle, A. B., Dite, G. S., Chen, X., Mayne, C. J., Southey, M. C., Batten, L. E., Chy, H., Trute, L., McCredie, M. R., Giles, G. G., et al. (1999). Androgen receptor exon 1 CAG repeat length and breast cancer in women before age forty years. J Natl Cancer Inst 91, 961-966. Stamey, T. A. (2001). Preoperative serum prostate-specific antigen (PSA) below 10 microg/1 predicts neither the presence o f prostate cancer nor the rate of postoperative PSA failure. Clin Chem 47, 631-634. Stamey, T. A., Johnstone, I. M., McNeal, J. E., Lu, A. Y., and Yemoto, C. M. (2002). Preoperative serum prostate specific antigen levels between 2 and 22 ng./ml. correlate poorly with post-radical prostatectomy cancer morphology: prostate specific antigen cure rates appear constant between 2 and 9 ng./ml. J Urol 167, 103- 111. Stanford, J. L., Just, J. J., Gibbs, M., Wicklund, K. G., Neal, C. L., Blumenstein, B. A., and Ostrander, E. A. (1997). Polymorphic repeats in the androgen receptor gene: molecular markers of prostate cancer risk. Cancer Res 57, 1194-1198. 169 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Stone, N. N., and Clejan, S. J. (1991). Response of prostate volume, prostate-specific antigen, and testosterone to flutamide in men with benign prostatic hyperplasia. J Androl 12, 376-380. Strain, D. O., Haiman, C. A., Hirschhom, J. N., Altshuler, D., Kolonel, L. N., Henderson, B. E., and Pike, M. C. (2003). Choosing haplotype-tagging SNPS based on unphased genotype data using a preliminary sample of unrelated subjects with an example from the Multiethnic Cohort Study. Hum Hered 55, 27-36. Suter, N. M., Malone, K. E., Daling, J. R., Doody, D. R., and Ostrander, E. A. (2003). Androgen receptor (CAG)n and (GGC)n polymorphisms and breast cancer risk in a population-based case-control study of young women. Cancer Epidemiol Biomarkers Prev 12, 127-135. Sutherland, G. R., Baker, E., Hyland, V. J., Callen, D. F., Close, J. A., Tregear, G. W., Evans, B. A., and Richards, R. I. (1988). Human prostate-specific antigen (APS) is a member o f the glandular kallikrein gene family at 19ql3. Cytogenet Cell Genet 48, 205-207. Sutkowski, D. M., Goode, R. L., Baniel, J., Teater, C., Cohen, P., McNulty, A. M., Hsiung, H. M., Becker, G. W., and Neubauer, B. L. (1999). Growth regulation of prostatic stromal cells by prostate-specific antigen. J Natl Cancer Inst 91, 1663-1669. Suzuki, T., Darnel, A. D., Akahira, J. I., Ariga, N., Ogawa, S., Kaneko, C., Takeyama, J., Moriya, T., and Sasano, H. (2001). 5alpha-reductases in human breast carcinoma: possible modulator of in situ androgenic actions. J Clin Endocrinol Metab 86, 2250-2257. Svartberg, J., Midtby, M., Bonaa, K. H., Sundsfjord, J., Joakimsen, R. M., and Jorde, R. (2003). The associations of age, lifestyle factors and chronic disease with testosterone in men: the Tromso Study. Eur J Endocrinol 149, 145-152. Szelei, J., Jimenez, J., Soto, A. M., Luizzi, M. F., and Sonnenschein, C. (1997). Androgen-induced inhibition of proliferation in human breast cancer MCF7 cells transfected with androgen receptor. Endocrinology 138, 1406-1412. Tan, P., Cady, B., Wanner, M., Worland, P., Cukor, B., Magi-Galluzzi, C., Lavin, P., Draetta, G., Pagano, M., and Loda, M. (1997). The cell cycle inhibitor p27 is an independent prognostic marker in small (Tla,b) invasive breast carcinomas. Cancer Res 57, 1259-1263. Tenover, J. S. (1992). Effects o f testosterone supplementation in the aging male. J Clin Endocrinol Metab 75, 1092-1098. 170 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Thijssen, J. H. (2004). Local biosynthesis and metabolism o f oestrogens in the human breast. Maturitas 49, 25-33. Thomas, H. V., Key, T. J., Allen, D. S., Moore, J. W., Dowsett, M., Fentiman, I. S., and Wang, D. Y. (1997a). A prospective study o f endogenous serum hormone concentrations and breast cancer risk in post-menopausal women on the island of Guernsey. Br J Cancer 76, 401-405. Thomas, H. V., Key, T. J., Allen, D. S., Moore, J. W., Dowsett, M., Fentiman, I. S., and Wang, D. Y. (1997b). A prospective study of endogenous serum hormone concentrations and breast cancer risk in premenopausal women on the island of Guernsey. Br J Cancer 75, 1075-1079. Tormey, D. C., Lippman, M. E., Edwards, B. K., and Cassidy, J. G. (1983). Evaluation of tamoxifen doses with and without fluoxymesterone in advanced breast cancer. Ann Intern Med 98, 139-144. Toscano, V. (1986). Dihydrotestosterone metabolism. Clin Endocrinol Metab 15, 279-292. Tut, T. G., Ghadessy, F. J., Trifiro, M. A., Pinsky, L., and Yong, E. L. (1997). Long polyglutamine tracts in the androgen receptor are associated with reduced trans activation, impaired sperm production, and male infertility. J Clin Endocrinol Metab 82, 3777-3782. Vermeulen, A., Verdonck, L., and Kaufman, J. M. (1999). A critical evaluation of simple methods for the estimation of free testosterone in serum. J Clin Endocrinol Metab 84, 3666-3672. Vierhapper, H., Nowotny, P., and Waldhausl, W. (1997). Determination of testosterone production rates in men and women using stable isotope/dilution and mass spectrometry. J Clin Endocrinol Metab 82, 1492-1496. Vottero, A., Stratakis, C. A., Ghizzoni, L., Longui, C. A., Karl, M., and Chrousos, G. P. (1999). Androgen receptor-mediated hypersensitivity to androgens in women with nonhyperandrogenic hirsutism: skewing of X-chromosome inactivation. J Clin Endocrinol Metab 84, 1091-1095. Wacholder, S., Rothman, N., and Caporaso, N. (2000). Population stratification in epidemiologic studies of common genetic variants and cancer: quantification of bias. JN atl Cancer Inst 92, 1151-1158. Walker, R. A. (2000). Transforming growth factor beta and its receptors: their role in breast cancer. Histopathology 36, 178-180. 171 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Wang, M. C., Valenzuela, L. A., Murphy, G. P., and Chu, T. M. (1979). Purification o f a human prostate specific antigen. Invest Urol 17, 159-163. Webber, M. M., Waghray, A., and Bello, D. (1995). Prostate-specific antigen, a serine pro tease, facilitates human prostate cancer cell invasion. Clin Cancer Res 1, 1089-1094. Weber, J. P., Oesterling, J. E., Peters, C. A., Partin, A. W., Chan, D. W., and Walsh, P. C. (1989). The influence o f reversible androgen deprivation on serum prostate- specific antigen levels in men with benign prostatic hyperplasia. J Urol 141, 987- 992. Weir, E. G., Partin, A. W., and Epstein, J. I. (2000). Correlation o f serum prostate specific antigen and quantitative immunohistochemistry. J Urol 1 6 3 ,1739-1742. Wilson, C. M., and McPhaul, M. J. (1996). A and B forms of the androgen receptor are expressed in a variety of human tissues. Mol Cell Endocrinol 120, 51-57. Wysowski, D. K., Comstock, G. W., Helsing, K. J., and Lau, H. L. (1987). Sex hormone levels in serum in relation to the development of breast cancer. Am J Epidemiol 125, 791-799. Xu, J., Meyers, D. A., Sterling, D. A., Zheng, S. L., Catalona, W. J., Cramer, S. D., Bleecker, E. R., and Ohar, J. (2002). Association studies of serum prostate-specific antigen levels and the genetic polymorphisms at the androgen receptor and prostate- specific antigen genes. Cancer Epidemiol Biomarkers Prev 11, 664-669. Xue, W., Irvine, R. A., Yu, M. C., Ross, R. K., Coetzee, G. A., and Ingles, S. A. (2000). Susceptibility to prostate cancer: interaction between genotypes at the androgen receptor and prostate-specific antigen loci. Cancer Res 60, 839-841. Xue, W. M., Coetzee, G. A., Ross, R. K., Irvine, R., Kolonel, L., Henderson, B. E., and Ingles, S. A. (2001). Genetic determinants o f serum prostate-specific antigen levels in healthy men from a multiethnic cohort. Cancer Epidemiol Biomarkers Prev 10, 575-579. Yeh, S., Hu, Y. C., Rahman, M., Lin, H. K., Hsu, C. L., Ting, H. J., Kang, H. Y., and Chang, C. (2000). Increase of androgen-induced cell death and androgen receptor transactivation by BRCA1 in prostate cancer cells. Proc Natl Acad Sci U S A 97, 11256-11261. Yeh, S., Hu, Y. C„ Wang, P. H., Xie, C., Xu, Q„ Tsai, M. Y., Dong, Z., Wang, R. S., Lee, T. H., and Chang, C. (2003). Abnormal mammary gland development and growth retardation in female mice and MCF7 breast cancer cells lacking androgen receptor. J Exp Med 198, 1899-1908. 172 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Young, C. Y., Andrews, P. E., Montgomery, B. T., and Tindall, D. J. (1992). Tissue- specific and hormonal regulation of human prostate-specific glandular kallikrein. Biochemistry 37, 818-824. Young, I. E., Kurian, K. M., Mackenzie, M. A., Kunkler, I. H., Cohen, B. B., Hooper, M. L., Wyllie, A. H., and Steel, C. M. (2000). The CAG repeat within the androgen receptor gene in male breast cancer patients. J Med Genet 37, 139-140. Yu, H. (2000). Clinical implications of prostate-specific antigen in men and women. J Gend Specif Med 3, 45-48, 53. Yu, H., and Berkel, H. (1999). Prostate-specific antigen (PSA) in women. J La State Med Soc 151, 209-213. Yu, H., Bharaj, B., Vassilikos, E. J., Giai, M., and Diamandis, E. P. (2000). Shorter CAG repeat length in the androgen receptor gene is associated with more aggressive forms of breast cancer. Breast Cancer Res Treat 59, 153-161. Yu, H., and Diamandis, E. P. (1995a). Measurement of serum prostate specific antigen levels in women and in prostatectomized men with an ultrasensitive immunoassay technique. Journal o f Urology 153, 1004-1008. Yu, H., and Diamandis, E. P. (1995b). Prostate-specific antigen immunoreactivity in amniotic fluid. Clin Chem 41, 204-210. Yu, H., and Diamandis, E. P. (1995c). Prostate-specific antigen in milk o f lactating women. Clin Chem 41, 54-58. Yu, H., Diamandis, E. P., Levesque, M., Giai, M., Roagna, R., Ponzone, R., Sismondi, P., Monne, M., and Croce, C. M. (1996). Prostate specific antigen in breast cancer, benign breast disease and normal breast tissue. Breast Cancer Res Treat 40, 171-178. Yu, H., Diamandis, E. P., Monne, M., and Croce, C. M. (1995a). Oral contraceptive- induced expression o f prostate-specific antigen in the female breast. J Biol Chem 270, 6615-6618. Yu, H., Diamandis, E. P., and Sutherland, D. J. (1994a). Immunoreactive prostate- specific antigen levels in female and male breast tumors and its association with steroid hormone receptors and patient age. Clin Biochem 27, 75-79. Yu, H., Diamandis, E. P., Zarghami, N., and Grass, L. (1994b). Induction of prostate specific antigen production by steroids and tamoxifen in breast cancer cell lines. Breast Cancer Res Treat 32, 291-300. 173 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Yu, H., Giai, M., Diamandis, E. P., Katsaros, D., Sutherland, D. J., Levesque, M. A., Roagna, R., Ponzone, R., and Sismondi, P. (1995b). Prostate-specific antigen is a new favorable prognostic indicator for women with breast cancer. Cancer Res 55, 2104-2110. Yu, H., Levesque, M. A., Clark, G. M., and Diamandis, E. P. (1998). Prognostic value o f prostate-specific antigen for women with breast cancer: a large United States cohort study. Clin Cancer Res 4, 1489-1497. Zarghami, N., Grass, L., and Diamandis, E. P. (1997a). Steroid hormone regulation of prostate-specific antigen gene expression in breast cancer. Br J Cancer 75, 579- 588. Zarghami, N., Grass, L., Sauter, E. R., and Diamandis, E. P. (1997b). Prostate- specific antigen in serum during the menstrual cycle. Clin Chem 43, 1862-1867. Zeegers, M. P., Kiemeney, L. A., Nieder, A. M., and Ostrer, H. (2004). How strong is the association between CAG and GGN repeat length polymorphisms in the androgen receptor gene and prostate cancer risk? Cancer Epidemiol Biomarkers Prev 13, 1765-1771. Zeleniuch-Jacquotte, A., Bruning, P. F., Bonfrer, J. M., Koenig, K. L., Shore, R. E., Kim, M. Y., Pasternack, B. S., and Toniolo, P. (1997). Relation of serum levels of testosterone and dehydroepiandrosterone sulfate to risk o f breast cancer in postmenopausal women. Am J Epidemiol 145, 1030-1038. Zeleniuch-Jacquotte, A., Shore, R. E., Koenig, K. L., Akhmedkhanov, A., Afanasyeva, Y., Kato, I., Kim,.M. Y., Rinaldi, S., Kaaks, R., and Toniolo, P. (2004). Postmenopausal levels of oestrogen, androgen, and SHBG and breast cancer: long term results of a prospective study. Br J Cancer 90, 153-159. Zhou, J., Ng, S., Adesanya-Famuiya, O., Anderson, K., and Bondy, C. A. (2000). Testosterone inhibits estrogen-induced mammary epithelial proliferation and suppresses estrogen receptor expression. Faseb J 14, 1725-1730. Zitzmann, M., and Nieschlag, E. (2001). Testosterone levels in healthy men and the relation to behavioural and physical characteristics: facts and constructs. Eur J Endocrinol 144, 183-197. Zumoff, B. (1994). Hormonal profiles in women with breast cancer. Obstet Gynecol Clin North Am 21, 751-772. Zumoff, B., Strain, G. W., Miller, L. K., and Rosner, W. (1995). Twenty-four-hour mean plasma testosterone concentration declines with age in normal premenopausal women. J Clin Endocrinol Metab 80, 1429-1430. 174 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX I. GRANT PROPOSAL: ANDROGEN RECEPTOR GENE, PSA GENE IN BREAST CANCER RISK (FUNDED BY THE CBCRP DISSERTATION AWARD) 1. INTRODUCTION and HYPOTHESIS Breast cancer is a hormone-related cancer. While estrogens are well-established risk factors for human breast cancer, the roles of androgens are unclear. Originally androgens, because they are precursors of estrogens, were postulated as risk factors for breast cancer. This hypothesis was supported by epidemiological studies comparing plasma androgens or their urinary metabolites in patients and healthy controls. However, recent experimental studies suggest that androgens may act directly as protective factors by binding to androgen receptors (AR). Furthermore, an AR-regulated protein, PSA, has been implicated as a favorable prognostic factor in breast cancer. One reason for the paradox may be that in epidemiologic studies it is often difficult to assess hormone exposure of the target tissue. In particular, many studies did not assess the effect of androgens independent of estrogens. A better way to address the question regarding the independent role of androgens in breast cancer development is to look at genetic variations in genes that regulate the exposure of breast tissue to endogenous androgens. In this study, we hypothesize that functional polymorphisms in the androgen receptor and the PSA genes may influence the development and progression of breast cancer by modifying the efficiency of the androgen signaling pathway. We propose an association study of genetic variations in the AR and PSA genes utilizing an existing case-control study of breast cancer carried out at the Northern California Cancer Center. (PI: John; NIH/NCI OOl CA 77305). We will focus on African-American women, a group with a notably higher frequency of more androgenic genotypes. This group also seems to be protected from breast cancer relative to whites. Specific Aims 1 . To determine whether androgen receptor CAG repeat size is associated with breast cancer risk among African-Americans in an ongoing case-control study. 2. To determine whether the polymorphism in ARE-I of the PSA gene is associated breast cancer risk among African-Americans in this case-control study. 3. To test for possible interaction between the AR gene and the PSA gene with respect to breast cancer risk. This study will help us gain a better understanding of the role of endogenous androgens in breast cancer etiology. Furthermore, with the increased interest in androgen replacement therapy in symptomatic women undergoing menopause, knowledge of the effects of androgens on the breast will be critical. 2. BACKGROUND and SIGNIFICANCE 2.1 Androgen, the androgen receptor and breast cancer Testosterone and dihydrotestosterone (DHT) are the two active androgens in the female body. There is an extensive body of information from epidemiological studies concerning 175 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. plasma levels of testosterone, DHT, or urinary metabolites of these androgens in both pre- and postmenopausal women with breast cancer. Although results from these studies are mixed, there seems to be a general agreement: elevated levels of these androgens are associated with higher risk of breast cancer (e.g. Secreto et al, 1984 and 1991; Berrino et al, 1996; Dorgan et al, 1996.) In contrast to epidemiological studies, experimental studies have suggested that androgens might act as a protective factor for breast cancer. This effect is mediated through the binding to the androgen receptor (AR). For example, the growth of human breast cancer cell lines expressing AR or transfected with AR is inhibited by DHT. (Poulin et al, 1998; Hackenberg et al, 1996; Birrell et al, 1995a; Szelei et al, 1997). Testosterone can inhibit estrogen-induced mammary epithelial proliferation and suppress estrogen expression in female rhesus monkeys (Zhou et al, 2000). The growth of chemically induced mammary tumors in rats was also inhibited by DHT (Dauvois et al, 1989). In addition, some clinical studies found that the presence of AR in breast tumor was a marker for good response to hormone therapy and survival (Birrell et al, 1995b; Bryan et al 1984). One reason for the inconsistency between epidemiological studies and experimental studies may be the failure of epidemiological studies to assess the independent effect of androgens. Androgens and estrogens are closely related. Androgens are converted into estrogens by aromatase in breast tissue. Androgens also compete with estrogens for sex hormone-binding globulin (SHBG), resulting in an increased fraction of free estrogens and thus higher bio-availability of estrogens. In two recent cohort studies on postmenopausal women, the observed positive association between elevated levels of plasma testosterone and breast cancer risk disappeared when estrogen level was adjusted (Hankinson et al, 1998; Zeleniuch-Jacquotte et al, 1997). Another problem with the epidemiological studies is that the measurement of plasma or urinary androgen levels do not reflect the true androgen exposure in breast tissue. Local production of the active androgen DHT in breast tissue has been documented and was found to be associated with a low cell proliferation rate in breast tumor (Suzuki et al, 2001). To better measure the local action of androgens, which is to a great extent genetically regulated, we will measure androgen receptor and PSA genotypes. 2.2 The androgen receptor CAG repeat and female breast cancer Androgen receptor (AR) mediates the proliferative effects of androgens in hormonally responsive tissue, such as breast. Functional allelic variation in the AR gene has been well documented. A polymorphic CAG repeat in exon 1 of the AR gene encodes a polyglutamine tract, the length of which is inversely and linearly related to AR activity (Chamberlian et al, 1994; Kazemi-Esfarjani et al, 1995; Tut et al, 1997; Irvine et al, 2000). The role of CAG repeat size in cancer predisposition is indicated by the well-established association between shorter CAG repeat size (i.e. higher AR transcriptional activity) and increased prostate cancer risk. (Irvine et al, 1995; Ingles et al, 1997; Stanford et al, 1997; Giovannucci et al, 1997; Hakimi et al, 1997; Ekman et al, 1999; Hsing et al, 2000). Epidemiological studies on AR CAG repeat and female breast cancer have provided inconsistent data. In 1999, results from two large population-based case-control studies were published. The study of Spurdle et al focused on premenopausal women (younger than 40 for both cases and controls) (Spurdle et al, 1999), while Dunning et al looked at subjects within a wider age range (25 to 75 years of age) (Dunning et al, 1999). No significant association between CAG repeat size and breast cancer risk was observed in either study. However, a more recent large scale case-control study suggested a protective role of 176 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. androgen in breast cancer by showing that shorter CAG repeats were associated with a significantly lower breast cancer risk (Giguere et al, 2001). Menopausal status was found to be a significant effect modifier in this association. The decreased risk associated with the short-allele AR genotypes was present mostly among postmenopausal women (predominantly among women with surgical menopause). A study of AR CAG repeat size and breast cancer risk in a special group of women (with inherited germline BRCA1 mutations) lent support to this protective effect of androgen (Rebbeck et al, 1999). Women carrying a BRCA1 mutation were at significantly lower risk of developing breast cancer and had later age-of-diagnosis if they also carried an AR gene with shorter CAG repeats, compared to BRCA1 carriers with longer CAG repeats. However, the modifying effect of AR CAG repeats on BRCA1 penetrance was not confirmed by two subsequent studies (Kadouri et al, 2001; Menin et al, 2001). The role of CAG repeat on tumor progression was also studied in a more general population. CAG repeat size did not appear to modify the age of disease onset in a study of 178 breast cancer cases aged 65 or younger (Given et al, 2000). But in another study, shorter CAG repeats seemed to be associated with more aggressive disease (Yu et al, 2000). In this study, the shorter AR-CAG repeats were associated with higher histological grade, positive lymph nodes and worse overall survival. Some limitations in the above mentioned studies may explain the lack of consistent results. The big difference in subjects’ ages across studies may be a reason for the disparate results. Since age at diagnosis might be influenced by CAG repeat size, it is possible that the impact of CAG repeat size on breast cancer risk may be more evident in certain age ranges. Furthermore, according to the finding of Giguere et al, menopausal status is an important effect modifier; however, it is difficult to explain biologically why shorter CAG repeats were protective only among surgical menopausal women. The results of this study should be interpreted with caution because of the use of hospital controls. In most of the other studies, menopausal status was not stated or not clearly defined, making it impossible to compare studies with regard to this point. Among the studies on BRCA1 mutations, small sample size seems to be a common problem. Lack of power might explain the failure to detect the effect of CAG repeat length on breast cancer risk in some studies. In summary, although a protective role of a functional androgen receptor in breast cancer was suggested based on data from clinical studies, experimental studies and some epidemiological studies, more well-designed epidemiological studies are needed in order to provide a clearer picture. Our proposed study will focus on African-Americans, a group for whom no association studies on CAG repeat and breast cancer have been conducted. After age 40, African-American women in the United States have a lower risk of developing breast cancer compared to non-Hispanic Whites. Given the fact that they have a higher incidence of obesity, which is a well-established risk factor for breast cancer in postmenopausal women (Kuczmarski, et al, 1994), the reason for this lower risk remains unclear. There is also a notable racial difference in the distribution of the sizes of AR-CAG repeats, with smaller sizes in African-Americans, compared to non-Hispanic Whites (Edwards et al, 1992). If smaller CAG repeat size is proven to be a protective factor for breast cancer development, the difference in the breast cancer incidence rate between African-Americans and non- Hispanic Whites may be partly explained by this difference in CAG repeat size (and thus different AR function). 177 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.3 PSA (prostate-specific antigen) and breast cancer Originally, it was believed that PSA was produced exclusively in men by the epithelial cells of the prostate gland. However, with the development of an ultrasensitive PSA immunoassay, it has been demonstrated that PSA is also expressed at a lower level in many female tissues and fluids, such as breast and endometrial tissue, milk and amniotic fluid (summarized in Magklara et al, 1999). PSA production appears to be a good prognostic factor for breast cancer. PSA expression has been detected in about 30% of breast tumors (Yu et al, 1994), and PSA immunoreactivity was inversely associated with markers of active cell proliferation in the breast tissue. High expression of PSA protein in breast tumors was significantly negatively associated with disease stage and tumor size. After adjusting for other significant breast cancer prognostic factors (i.e. age, clinical stage, tumor size, histological grade, nodal status, estrogen receptor and progesterone receptor status), PSA-positive patients still had a significantly reduced risk for relapse when compared with PSA-negative patients. (Yu et al, 1995a andl996; Diamandis et al, 1994, Sauter et al, 1998a). The role of PSA in cancer development and progression is largely unknown. Many recent findings suggest that PSA has some functional significance other than just as a tumor marker. For example, antiproliferative, antiangiogenic, and apoptosis-inducing activities of PSA have been observed in a series of in vitro and in vivo studies, suggesting that PSA can be protective in cancer development and progression. On the other hand, a deleterious role of PSA in cancer was also suggested by other studies (reviewed by Diamandis, 2000). In females, it seems that the regulation of PSA production is under control of both progestin and androgens (Zarghami et al, 1997; Yu et al, 1995b; Sauter et al, 1998b; Negri et al, 2000; Mifsud et al, 2001). Both androgen and progestin-regulated PSA expression is thought to be mediated by binding of AR-ligand complex to certain DNA sequences, androgen responsive elements (ARE), within the regulatory region of the PSA gene. At least three such responsive elements have been identified, two within several hundred bases of the transcription start site (ARE-I and ARE-II from -170 to -156 and -394 to -380) (Riegman et al, 1991; Cleutjens et al, 1996) and a third approximately 4200 bases further upstream (ARE-III) (Cleutjens et al, 1997). In a previous study conducted by our group, it was found that genetic variations in the ARE-I of the PSA gene (a nucleotide variation, G-> A at position -158) and the AR gene (CAG repeat polymorphism) contributed interactively to the variation is serum PSA level in 420 healthy male subjects (Xue et al, 2001). This PSA promoter polymorphism was also related to PSA concentration in breast tumor tissues. Furthermore, in the survival analysis, genotype was significantly related to disease-free and overall survival (Bharaj et al, 2000). The functional significance of this polymorphism is unknown. One possible explanation could be, as suggested by Xue, et al (Xue et al, 2000), that the two alleles encode ARE-I sites with different binding affinity to androgen receptor. In summary, although still disputable, new evidence from epidemiological studies points to the idea that PSA plays an important role in breast cancer development and/or progression. Given the fact that PSA is a down-stream gene of the androgen signaling pathway, studies looking at the interaction between the AR and PSA gene polymorphisms in breast cancer are highly warranted. 3. PRELIMINARY RESULTS 178 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cases from this study were obtained from the population-based cancer registry covering the San Francisco Bay Area. This cancer registry is part of the NCI SEER program. Data from the SEER registry show that after age 40, African-American women have lower breast cancer incidence rates, compared to non-Hispanic White women (Figure 1). Figure 1. B reast ca n cer a g e-sp ecific in cid en ce ra tes (per 100,000) by race/ethnicity, SEER 1992-97 600 t n 0) ■ if ■ - W hites - • — African-Americans r e • 500 u | f 400 o ° - ■- o 300 k- T * 0 ) u c r e o q . 200 i n r e o 100 C O 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85+ Age Work for this proposal will be performed in the laboratory of Dr. Sue Ingles. She has been PI of grants to study the effects of vitamin D (BCRP 1IB0-353 & 3IB-0089), dietary fat (BCRP 7WB-0110), hormone metabolism genes (BCRP 6IB-0102), and aromatase gene regulation (DOD 981193) in breast cancer etiology. Her group previously found that genetic variation in the AR and PSA genes contributes interactively to variation in serum PSA levels in healthy men. The allele frequencies were significantly different between African-Americans and Whites (see appendix for article by Xue et al.). 4. RESEARCH DESIGN AND METHODS 4.1 Study Population The proposed study will be built on a population-based case-control study conducted in the San Francisco Bay Area by Dr. Esther M. John, with funding from the National Cancer Institute and the Department of Defense (PI: E. John, R01 CA63446, DAMD 17-96- 1-6071, R01 CA77305). Eligible cases for the proposed study include African-American women newly diagnosed with primary invasive breast cancer between Mayl, 1997 and April 30, 1999, aged 35-79 and residing in the San Francisco Bay area at the time of diagnosis, and without a prior a history of breast cancer. They were identified through the population- based cancer registry covering the San Francisco Bay Area. The cancer registry is part of the NCI SEER program and the California Cancer Registry. Population controls matched to cases by ethnicity and age were identified through random digit dialing. Cases and controls were first contacted by phone to administer a brief screening. All eligible cases and controls were then invited to complete an in-person interview and anthropometric measurements and to provide a blood sample (or mouthwash sample, if blood draw was declined). The interview was completed for a total of 292 African-American cases and 305 controls, with a response rate of 90% and 85% respectively. Of the subjects who completed the interview, 249 cases and 255 controls provided a biospecimen (blood or mouthwash), with response rates of 85% and 84%, respectively. 179 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.2 Data collection Trained professional bilingual/bicultural interviewers administered a structured questionnaire (in English or Spanish) which inquired about a broad array of established and suspected breast cancer risk factors, including family history, mentrual and reproductive histories. Anthropometric measurements were taken on all of these study participants. In a second home visit, participants were invited by professional phlebotomists to provide a 40 ml sample of whole blood. Those who declined the blood draw were given the option to provide a mouthwash sample (collected using Scope mouth rinse). To date we have collected 40 ml of blood or mouthwash samples for 249 cases and 255 controls. 4.3 Laboratory methods 4.3.1 General quality control. All genotyping assays will include control samples that have had genotype previously confirmed by sequencing. All PCR assays include a “water blank” to guard against contamination of the PCR reaction. To assure assay reproducibility, 5% of samples will be repeated with laboratory personnel blinded as to repeat status. 4.3.2 Genotyping. Simple sequence length polymorphism (SSLP) analysis will be used to identify the AR exon 1 CAG repeat variant. The genomic region containing the CAG repeat is PCR amplified with one fluorescently-labeled and one unlabeled primer. The resulting PCR product is run on the ABI 3700 capillary sequencer and the allele sizes are scored using GeneScan software. Genotyping of the single nucleotide polymorphism in the PSA gene will be performed by the Taqman assay. Two oligonucleotide probes, one specific for each allele, are labeled with different fluorophores and included during the PCR amplification. Hybridization of the reporter probe to its target sequence leads to nucleolytic cleavage of the probe during PCR amplification as a result of the 5’>3’ nuclease activity of Taq Polymerase, causing the release of the fluorescence signals. Fluorescence is detected using an ABI 7700 Taqman Sequence Detection System and the alleles are scored using ABI Sequence Detector software. 4.4 Data management and statistical analysis 4.4.1 Data management. Laboratory personnel will remain blinded to case-control status. DNA samples are labeled only by specimen ID. Genotype results and specimen ID will be entered into a spreadsheet and sent to Dr. John at NCCC. The NCCC data manager will merge genotype results to the database that contains case-control status and other covariates, such as family history, menstrual and reproductive history, etc. This working database will be stripped of IDs and returned to the investigator for data analysis. 4.4.2 Statistical analysis. Standard analyses for case-control studies will be conducted (Breslow & Day, 1980). The linear effect of AR CAG repeats will be estimated by including a continuous variable in the logistic model. This will be done separately for the short allele (the allele with shorter CAG repeat size), the long allele, and for the sum of CAG repeats over the two alleles. The CAG repeat sizes will also be analyzed as categorical variables using the median CAG size as the cutpoint. Odds ratios will be estimated to compare genotypes short/short and short/long to the “baseline” genotype long/long. We will also try to model the effect of CAG repeats using the same cutpoints as those in previous published papers in order to allow comparison. 180 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. For the polymorphism in the PSA gene, odds ratios (with 95% confidence intervals) will be estimated to compare genotype A/A and G/A to G/G. A test of trend will be performed by including in the logistic model a variable coded as 0,1,2 for the number of “at- risk” alleles. This coding scheme represents the co-dominant model. Dominant and recessive effects can be explored using other coding schemes (eg. 0,1,1 for dominant and 0,0,1 for recessive effect). 4.4.3 Confounders and effect-modifiers. Potential confounders, such as age, will be controlled by being included in the logistic regression model. Other known risk factors for breast cancer, such as family history, age at menarche, parity, and BMI will be examined to see if they appear to be confounders. The primary effect modifiers to be considered are menopausal status and family history. Odds ratios for “risk” genotypes will be estimated within strata defined by menopausal status or family history. Formal test of effect-modification will be performed by including the appropriate interaction terms in the logistic model. Case subjects will also be stratified by tumor grade, disease stage and estrogen receptor/progesterone receptor status to assess whether genotype exhibits an association with cancers of a particular subtype. 4.4.4 Gene-gene interaction. Because the PSA gene is an androgen-regulated gene and the polymorphism we are studying is located in the androgen responsive element, we would expect that certain AR/PSA genotype combinations will confer increased risk. To assess possible gene-gene interaction, odds ratios will be estimated for each AR/PSA genotype combination. Odds ratios will be also estimated for the effect of CAG repeat size within each of the three PSA genotype categories. 4.4.5 Power analysis. 4.4.5.1 Main effect of genotype: We will have approximately 250 cases and 250 controls. For power analyses, we made the following simplifications. For the PSA gene, the AG and GG genotypes are grouped together since men with these genotypes had similar PSA levels in our previous study (Xue et al, 2001). The frequency of the AA genotype in African- Americans was 25%. For CAG, women will be categorized as having two short alleles (<20 CAG repeats) vs. at least one long allele, similar to the paper of Giguere (Giguere et al, 2001). In African-Americans, 44% have two short alleles (Irvine et al, 1995). The following table gives the smallest main effect that can be detected with 80% power. Table 1. Smallest detectable odds ratios with 80% power Odds ratio PSA AA 1.00 AG/GG 1.68 CAG SS 1.00 SL/LL 1.90 4.4.5.2 Gene-gene interaction: Assuming the main effects in the above table, we will 80% power to detect an interaction odds ratio as low as 1.96. 4.4.6 Timeline. Months 1-2: Genotyping condition optimizing. Months 3-20: Genotyping (AR CAG and PSA). Months 21-24: Statistical analysis, manuscript preparation. 181 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REFRENCES: Berrino F, Muti P, Micheli A, Bolelli G, Krogh V, Sciajno R, Pisani P, Panico S, and Secreto G. Serum seex hormone levels after menopause and subsequenct breast cancer. J Natl Cancer Inst 88:291-296, 1996. Bharaj B, Scorilas A, Diamandis EP, Giai M, Levesque MA, Sutherland DJA, Hoffman BR. Breast cancer prognostic significance of a single nucleotide polymorphism in the proximal androgen response element of the prostate specific antigen gene promoter. Breast Cancer Res Treat. 61(2): 111-119, 2000. Birrell SN, Bentel JM, Hickey TE, Ricciardelli C, Weger MA, Horsfall DJ, Tilley WD. Androgens induce divergent proliferative responses in human breast cancer cell lines. J Steroid Biochem Molec Biol 52(5):459-467, 1995a. Birrell SN, Roder DM, Horsfall DJ, Bentel JM, and Tilley WD. Medroxy-progesterone acetate therapy in advanced breast cancer: the predictive value of androgen receptor expression. J Clin Oncol 13(7): 1572-1577, 1995b. Breslow NE and Day NE. Statistical methods in cancer research v.l Lyon, France:IARC 1980. Bryan RM, Mercer RJ, Bennett RC, Rennie GC, Lie TH, and Morgan FJ. Androgen receptors in breast cancer. Cancer 54:2436-2440, 1984. Chamberlian NL, Driver ED and Miesfeld RL. The length and location of CAG trinucleotide repeats in the androgen receptor N-terminal domain affect transactivation function. Nucleic Acid Res. 22(15):3181-3186, 1994. Cleutjens KBJM, van Eekelen CCEM, van der Korput HAGM, Brinkmann AO, and Trapman J. Two androgen response regions cooperate in steroid hormone regulated activity of the prostate-specific antigen promoter. J Biol Chem 271:6379-6388, 1996. Cleutjens KBJM, van der Korput HAGM, van Eekelen CCEM, van Riooji HCJ, Faber PW, and Trapman J. An androgen responsive element in a far upstream enhancer region is essential for high androgen-regulated activity of the prostate specific antigen promoters. Mol Endocrinol 11:1458-1461, 1997. Dauvois S, Li SM, Martel C, and Labrie F. Inhibitory effect of androgens on DMBA- induced mammary carcinoma in rat. Breast Cancer Res. Treat 14:299-306, 1989. Diamandis EP, Yu H, and Sutherland DJA. Detection of prostate-specific antigen immunoreactivity in breast tumors. Breast Cancer Res Treat 33:301-310, 1994. Diamandis EP. Prostate-specific antigen: a cancer fighter and a valuable messenger? Clin Chem 46(7):896-900, 2000. Dorgan JF, Longcope C, Stephenson Jr. HE, Falk RT, Miller R, Franze C, Kahle L, Campbell WS, Tangrea JA, and Schatzkin A. Relation of prediagnostic serum estrogen and androgen levels to breast cancer risk. Cancer Epidemiol. Biomarker Prevent. 5(7):533-539, 1996. Dunning AM, McBride S, Gregory J, Durocher F, Foster NA, Healey CS, Smith N, Pharoah PDP, Luben RN, Easton DF, Ponder BAJ. No association between androgen or vitamin D receptor gene polymorphisms and risk of breast cancer. Carcinogenesis 20(11):2121 -2135, 1999. Edwards A, Hammond HA, Jin L, Caskey CT and Chakraborty R. Genetic variation at five trimeric and tetrameric tandem repeat loci in four human population groups. Genomics 12: 241-253, 1992. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ekman P, Gronberg H, Matsuyama H, Kivineva M, Bergerheim US, and Li C. Links between genetic and environmental factors and prostate cancer risk. Prostate 39:262- 268, 1999. Giguere Y, Dewailly E, Brisson J, Ayotte P, Laflamme N, Demers A, Lorest V, Dodin S, Robert J, and Rousseau P. Short polyglutamine tracts in the androgen receptor are protective against breast cancer in the general population. Cancer Res 61:5869-5874, 2001 . Giovannucci E, Stampfer MJ, Krithivas K, Brown M, Dahl, D, Brufsky A, Talcott J, Hennekens CH, and Kantoff PW. The CAG repeat within the androgen receptor gene and its relationship to prostate cancer. Proc Natl Acad Sci USA. 94:3320-3323, 1997. Given HF, Radboume R, Oag H, Merritt S, Barclay E, Hanby AM, Lamlum H, McGrath J, Curran C, and Tomlinson IPM. The androgen receptor exon 1 trinucleotide repeat does not act as a modifier of the age of presentation in breast cancer. European Journal of Cancer 36:533-534, 2000. Hackenberg R, and Schulz KD. Androgen receptor mediated growth control of breast cancer and endometrial cancer modulated by antiandrogen-and androgen-like steroids. J Steroid Biochem Molec Biol. 56( 1 -6): 113-117, 1996. Hakimi JM, Schoenberg, MP, Rondinelli RH, Piantadosi S, and Barrack ER. Androgen receptor variants with short glutamine or glycine repeats may identify unique subpopulations of men with prostate cancer. Clin Cancer Res 3:1599-1608, 1997. Hankinson SE, Willett WC, Manson JE, Colditz GA, Hunter DJ, Spiegelman D, Barbieri RL, and Speizer FE. Plasma sex steroid hormone levels and risk of breast cancer in postmenopausal women. JNatl Cancer Inst. 90(17): 1292-1299, 1998. Hsing AW, Gao YT, Wu G, Wang X, Deng J, Chen YL, Sesterhenn IA, Mostofi FK, Benichou J, and Chang C. Polymorphic CAG and GGN repeat lengths in the androgen receptor gene and prostate cancer risk: a population-based case-control study in China. Cancer Res 60:5111-5116, 2000. Ingles SA, Ross RK, Yu MC, Irvine RA, La-Pera G, Haile RW, and Coetzee GA. Association of prostate cancer risk with genetic polymorphisms in vitamin D receptor and androgen receptor. J Natl Cancer Inst 89:166-170, 1997. Irvine RA, Yu MC, Ross RK, Coetzee GA. The CAG and GGC microsatellites of the androgen receptor gene are in linkage disequilibrium in men with prostate cancer. Cancer Res 55:1937-1940, 1995. Irvine RA, Ma H, Yu MC, Ross RK, Stallcup MR and Coetzee GA. Inhibition of pi 60- mediated coactivation with increasing androgen receptor polyglutamine length. Hum. Mol. Genet. 9(2):267-274, 2000. Kadouri L, Easton DF, Edwards S, Hubert A, Kote-Jarai Z, Glaser B, Durocher F, Abeliovich D, Peretz T, and Eeles RA. CAG and GGC repeat polymorphisms in the androgen receptor gene and breast cancer susceptibility in BRCA1/2 carriers and non-carriers. Br J Cancer, 85(l):36-40, 2001. Kazemi-Esfarjani P, Trifiro MA, and Pinsky L. Evidence for a repressive function of the long polyglutamine tract in the human androgen receptor possible pathogenetic relevance for the (CAG)n-expanded neuropathies. Human Molecular Genet. 4(4):523-527, 1995. Kuczmarski RJ, Flegal KM, Campbell SM, and Johnson CL. Increasing prevalence of overweight among US adults: The national health and nutrition examination surveys, 1960-1991. JAMA 272(3):205-211, 1994. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Magklara A, Scorilas A, Lopez-Otin C, Vizoso F, Ruibal A, and Diamandis EP. Human glandular kallikrein in breast milk, amniotic fluid, and breast cyst fluid, clinical chemistry 45(10): 1774-1780, 1999. Menin C, Banna GL, De Salvo G, Lazzarotto V, De Nicolo A, Agata S, Montagna M, Sordi G, Nicoletto O, Chieco-Bianchi L, D’Andrea E. Lack of association between androgen receptor CAG polymorphism and familial breast/ovarian cancer. Mifsud A, Choon AT, Fang D, and Yong EL. Prostate-specific antigen, testosterone, sex- hormone binding globulin and androgen receptor CAG repeat polymorphisms in subfertile and normal men. Mol Human reproduction 7(11): 1007-1013, 2001. Negri C, Tosi F, Dorizzi R, Fortunato A, Spiazzi GG, Muggeo M, Castello R, and Moghetti P. Antiandrogen drugs lower serum prostate-specific antigen (PSA) levels in hirsute subjects: evidence that serum PSA is a marker of androgen action in women. J Clin Endocrinol Metabol 85(l):81-84, 2000. Poulin R. Baker D, and Labrie F. Androgens inhibit basal and estrogen-induced cell proliferation in the ZR-75-1 human breast cancer cell line. Breast Cancer Res Treat 12:213-225, 1998. Rebbeck TR, Kantoff PW, Krithivas K, Neuhausen S, Blackwood MA, Godwin AK, Daly MB, Narod SA, Garber JE, Lynch HT, Weber BL, and Brown M. Modification of BRCA1-associated breast cancer risk by the polymorphic androgen-receptor CAG repeat. Am J Hum Genet 64:1371-1377, 1999. Riegman PHJ, Vlietstra RJ, van der Korput JAGM, Brinkman AO, and Trapman J. The promoter of the prostate specific antigen contains a functional androgen responsive element. Mol Endocrinol 5:1921-1930, 1991. Sauter ER, Daly M, Linahan K, Ehya H, Engstrom PF, Bonney G, Ross EA, Yu H, and Diamandis E. Prognostic value of Prostate-specific Antigen for women with breast cancer: A large United States cohort study. Clin Cancer Res 4:1489-1497, 1998a. Sauter ER, Babb J, Daly M, Engstrom PF, Ehya H, Malick J, and Diamandis E. Prostate- specific Antigen production in the female breast: association with progesterone. Cancer Epidemiol. Biomark. Prev. 7:315-320, 1998b. Secreto G, Toniolo P, Berrino F, Recchione C, Pietro SD, Fariselli G, and Decarli A. Increased androgen activity and breast cancer risk in premenopausal women. Cancer Res. 44:5902-5905,1984. Secreto G, Toniolo P, Berrino F, Recchione C, Cavalleri A, Pisani P, Totis A, Fariselli G, and Pietro SD. Serum and urinary androgens and risk of breast cancer in postmenopausal women. Cancer Res. 51:2572-2576,1991. Spurdle AB, Dite GS, Chen X, Mayne CJ, Southey MC, Batten LE, Chy H, Trute L, McCredie MRE, Giles GG, Armes J, Venter DJ, and Hopper JL, Chenevix-Trench G. Androgen receptor exon 1 CAG repeat length and breast cancer in women before age forty years. J Natl Cancer Inst 91:961-966, 1999. Stanford JL, Just JJ, Gibbs M, Wicklund KG, Neal CL, Blumenstein BA, and Ostrander EA. Polymorphic repeats in the androgen receptor gene: molecular markers of prostate cancer risk. Cancer Res 57:1194-1198, 1997. Suzuki T, Darnel AD, Akahira J, Ariga N, Ogawa S, Kaneko C, Takeyama J, Moriya T, and Sasano H. 5a-reductases in human breast carcinoma: possible modulator of in situ androgenic actions. J. Clinical Endocrinology & Metabolism 86(5):2250-2257, 2000. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Szelei J, Jimenez J, Soto AM, Luizzi MF, and Sonnenschein C. Androgen-induced inhibition of proliferation in human breast cancer MCF7 cells transfected with androgen receptor. Endocrionology 138(4): 1406-1412. Tut TG, Ghadessy FJ, Trifiro MA, Pinsky L and Yong, EL. Long polyglutamine tracts in the androgen receptor are associated with reduced trans-activation, impaired sperm production, and male infertility. J. Clin. Endocrinol. Metab., 82:3777-3782, 1997. von Eckardstein S, Syska Am Gromoll J, Kamischke A, Simoni M, and Mieschlag E. Inverse correlation between sperm concentration and number of androgen receptor CAG repeats in normal men. J Clin Endocrinol Metab 86:2585-2590, 2001. Xue W, Coetzee GA, Ross RK, Irvine R, Kolonel L, Henderson BE, Ingles SA. Genetic determinants of serum prostate-specific antigen levels in healthy men from a multiethnic cohort. Cancer Epidemiol Biomark Prev. 10:575-579, 2001. Xue W, Irvine RA, Yu MC, Ross RK, Coetzee GA, Ingles SA. Susceptibility to prostate cancer: interaction between genotypes at the androgen receptor and prostate-specific antigen loci. Cancer Res. 60:839-841, 2000. Yu H, Diamandis EP, and Sutherland DJA. Immunoreactive prostate-specific antigen levels in female and male breast tumros and its association with steroid hormone receptors and patients age. Clin Biochem 27:75-79, 1994. Yu H, Giai M, Diamandis EP, Katsaros D, Sutherland DJA, Levesque MA, Roagna R, Ponzone R, and sismondi P. Prostate-specific Antigen is a new favorable prognostic indicator for women with breast cancer. Cancer Res. 55:2104-2111, 1995a. Yu H, Dianandis EP, Monne M, Croce CM. Oral contraceptive-induced expression of prostate-specific antigen in the female breast. J. Bio Chem 270(12):6615-6618, 1995b. Yu H, Diamandis EP, Levesque M, Giai M, Roagna R, Ponzone R, Sismondi P, Monne M, and Croce CM. Prostae specific antigen in breast cancer, benign breast disease and normal breast tissue. Breast Cancer Res Treat 40:171-178, 1996. Yu H, Bharaj B, Vassilikos EJK, Giai M, and Diamandis EP. Shorter CAG repeat length in the androgen receptor gene is associated with more aggressive forms of breast cancer. Breast cancer research and treatment 59(2): 153-161, 2000. Zarghami N, Grass L and Diamandis EP. Steroid hormone regulation of prostate-specific antigen gene expression in breast cancer. Br J Cancer 75(4):579-588, 1997. Zeleniuch-Jacquotte A, Bruning PF, Bonfrer JMG, Koenig KL. Shore RE, Kim MY, Pasternack BS,Toniolo P. Relation of serum levels of testosterone and dehydroepiandrosterone sulfate to risk of breast cancer in postmenopausal women. Am J Epidemiol. 145(11) 1030-1038, 1997. Zhou J, NG S, Adesanya-Famuiya P, Anderson K, and Bondy CA. Testosterone inhibits estrogen-induced mammary epithelial proliferation and suppresses estrogen receptorexpression. FASEB J. 14:1725-1730, 2000. 185 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. California Breast Cancer Research Program A p p lica tio n E valu ation — D iss e r ta tio n A w ard, P o s td o c to r a l F e llo w sh ip , N ew In v estig a to r R e v ie w C o m m ittee: E tio lo g y a n d P rev en tio n A p p lic a tio n #: 8G B -0103 PI N am e: W ei W ang T itle: Proaptotic Androgen R eceptor G en e and P SA G en e in B reast C an cer Risk S c i e n t i f i c M e r i t C o m p o n e n t S c o r e s : (1 = w o r s t t o i o = b e s t ) In n o v a tiv e n e s s: 8 .4 7 Im pact: 8 .6 0 A p p ro a ch : 8 .6 0 F ea sib ility : 9 .2 0 C areer D ev elo p m en t: 9 .0 7 S c ie n tific M erit S c o r e: 8.79 P e r c e n tile R ank (T op R a n k = 100% ): 100 (Within award type in this com m ittee)_______________________________________________________________ I. S u m m a r y o f A im s a n d M e t h o d s : REVIEWER A Although the role o f PSA (prostate specific antigen) and androgens has been clearly defined as a biomarker in prostate cancer, recently there has been a considerable body of work investigating the role o f these markers in breast cancer. PSA is a serine protease that is directly induced by androgens such as DHT and testosterone in the prostate and is used as a screening marker in prostate cancer. In the breast low levels o f PSA can also be measured in breast tissue, nipple aspirates and milk. From previous studies PSA has been shown to be a good prognostic factor for breast cancer and is associated with good outcome. Androgen receptors (AR) which mediate the responsiveness to androgens and regulate the induction o f PSA, have polymorphic variation in a CAG repeat which encodes a polyglutamine amino acid tract in the N terminal end o f the protein whose length is inversely correlated to AR activity. Among ethnic groups it has been shown that African-Americans have shorter CAG repeats and thus more active A R receptors. Complicating this factor there are polymorphisms in the promoter o f the PSA gene at the androgen receptor response element, which have been previously shown in males to influence the ability o f AR to induce PSA. Since SEER data indicates that African-American women have lower age related breast cancer incidence rates than Caucasians, the goal of this study is to examine the role o f AR and PSA polymorphisms in contributing to this ethnic differences in this breast cancer incidence. These studies will utilize a existing study population o f African American women approximately 250 breast cancer cases and 250 controls. The aims of the study are to: 1) determine whether AR CAG size is associated with risk, 2) to determine where the polymorphisms in PSA are associated with risk, 3) to test for the interactions in risk between AR and PSA. The applicant is an excellent student with a 4.0GPA and has superb references from her mentor and from one o f the collaborators on the project. The applicant has an. MS and has put together the majority o f the 1 8 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PI Name: Wei Wang background and rationale for this project through some earlier coursework and projects. The applicant has several publications including a first authored peer reviewed. The applicant received the highest score for her PhD qualifying exam. The mentor clearly states that the application was written entirely by the student. R E V IE W E R B The applicant hypothesizes that known polymorphisms, in the androgen receptor gene and the androgen- regulated PSA gene, together influence production o f PSA in breast tissue, and may thereby influence the development and progression o f breast cancer. There is precedence for this idea from prostate cancer, and breast tissue PSA levels have been found associated with prognosis. This is a molecular epidemiologic study, using data and blood samples collected in an existing case-control study o f African-American women in northern California, diagnoses with breast cancer between May 1,1997 and April 30, 1999. The specific aims o f the proposed project are: 1. To determine whether the androgen receptor CAG repeat size is associated with breast cancer risk among African-American women in an ongoing case-control study. 2. To determine whether the polymorphism in the androgen responsive element-1 o f the PSA gene is associated with breast cancer risk among African-American women in this study. 3. To test for possible interactions between the androgen receptor gene and PSA gene on breast cancer risk. The applicant has a background in epidemiology and laboratory molecular techniques, entirely suitable for this project. B y using existing epidemiologic data she w ill have experience with the vagaries of such data, and the challenges of analysis. The project appears suitable for a dissertation. M y only concern is whether her training suffers for lack o f the experience o f primary epidemiologic data collection or even sample collection, although she may have already had this experience. However, this is for her PhD committee to decide. Judging by her publications, the applicant clearly has an established professional relationship with her mentor. II. M a t c h t o B C R P P r i o r i t y I s s u e a n d A w a r d T y p e : R E V IE W E R A There is good correspondence between the award type and the priority issue selected for the project. R E V IE W E R B Priority Issue: Etiology. The influences o f androgens in breast cancer etiology are relevant and unclear, as are common polymorphisms in low-penetrance genes, so this is a very appropriate priority. Restriction o f the proposal to African-American women only is unfortunate, because it precludes comparison of her findings with women o f other races/ethnicities, but is reasonable in order to contain the project to a manageable size and use existing data and samples. Award Type: Dissertation Award: The aims and research plan are appropriate to the expertise needed and time available for completion o f a dissertation project. R E V IE W E R C This dissertation proposal is responsive to the Priority Issue ‘Etiology’, and is a dissertation proposal. 187 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PI Name: Wei Wang A D V O C A T E R E V IE W E R The project is a dissertation in the etiology priority issue. It could also fit into the racial/ethnic differences category. 1 1 1 . S c ie n t if ic M e r it S c o r e C o m p o n e n t s - S t r e n g t h s a n d W e a k n e s s e s : R E V IE W E R A In n o v a tiv e n e s s The center where the research will be done has produced other works that similarly follow genetic differences in other markers such as Vitamin D and breast cancer. The department has an active epidemiology program and collaborators to assist in the project. The proposed studies will use an established cohort o f women that has been used and continues to be used to evaluate other markers in African-American women. The methods for evaluating the genetic composition o f AR and PSA are standard. This project is innovative in that it will try to link PSA for the first time with polymorphisms in the PSA gene and also in the AR gene. Although some of these studies have successfully been done in males they have never been fully investigated in females with breast cancer. Im p act This project could have a significant impact in further defining the use o f PSA as a marker in breast cancer. Further, these studies will help to increase our understanding o f the effects o f AR and PSA polymorphisms and their contribution to breast cancer risk. The study is being done in an underserved population and will help to shed light on differences in ethnic risk to breast cancer A p p ro a ch The research plan is well organized and presented. The hypothesis and specific aims are w ell described. Although important data will be gleamed from the African-American cohort between breast cancer cases and controls, it is also reasonable that these studies could continue to larger cohorts and to comparisons of other ethnic groups. In the confounding factors it is hoped that use o f exogenous hormones such as birth control pills with progestins will also be controlled in these studies. PSA can also be induced by progestins. F ea sib ility The training received in this environment will be sufficient to complete this project, A similar study was completed by the mentor in males (see appendix) studying the same types o f polymorphisms in prostate cancer patients. It is highly likely that this study in this environment has a low level of risk from a technical point o f view. C areer D e v e lo p m e n t The mentor’s commitment to the candidate is clear from her letter o f support. There is also adequate support indicated in the letters by the collaborators on this project. This training w ill enable this student to complete project and give her a strong foundation to look at expanded or other markers o f risk for breast cancer. Both mentor and applicant are highly motivated to continue work in breast cancer. These studies could form the basis for future studies in the career o f the applicant. Prim ary r e a s o n s for r e c o m m e n d e d ratin g This project’s success rests on the excellent qualifications and record of the applicant. The mentor o f the project is capable and has past and present experience to assist the student on this type of project. The 188 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PI Name: Wei Wang cohorts to be used in this project are established and in place. The applicant has written a thoughtful and feasible project which focuses on an underserved population. This project could lead to future studies and a starting point for other projects in the career of the applicant. The issues and aims are timely to finding new non-invasive markers to assess risk in breast cancer patients. REVIEWER B I n n o v a tiv e n e s s The applicant clearly already has some considerable experience in molecular biology and epidemiology, and shows good appreciation o f the complexities o f this subject, and need for further illumination o f the influences o f androgens and other hormones on breast cancer etiology. With this project she will gain additional training in the issues of gene-environment interaction. However, as stated above, she will not receive experience with primary data and sample collection, although she may have in her past work. Im p act This project w ill contribute to understanding o f hormones and breast cancer etiology, and possibly to the role o f additional low penetrance genes in the hormonal pathways. A p p ro a ch In general, this appears to be a logical and reasonable research plan. A few concerns are: • it is unclear how rapidly cases were accrued in the parent study, and precisely what “newly diagnosed” means; • it would have been helpful to have the questionnaire used in the parent study available for review, although it is expected that all questions relevant to the hormone hypotheses will have been asked. However, since the androgen receptor and PSA are both associated with risk or prognosis of prostate cancer, inclusion o f fa m ily h isto ry o fp r o s ta te ca n cer w o u ld stre n g th en this p ro p o sa l, • h er d iscu ssio n o f c o va ria tes d o es n o t inclu de m ea su res o f so c io -ec o n o m ic status, releva n t to an y stu d y o f c a n c er a n d A frican-A m ericans, o r u se o f exogen ou s hormones', • she says variously that the PSA polymorphism is and is not functional; • she refers to “laboratory personnel” remaining blinded to case-control status, suggesting she may- not be doing all the work herself. F ea sib ility The year and one half for which funding is requested should be more than adequate for lab work, although it may be a bit short for analysis and writing. If any data cleaning duties are required in order to use the epidemiologic data, more time m aybe required. Judging by the papers co-authored by the applicant and her mentor, the research environment should be very conducive to completion o f the project. The N onis Cancer Center is well regarded for research in this area. Her collaborator, Dr. John, is essential to the project in providing data and DNA samples, and expresses great enthusiasm for her project. C areer D e v e lo p m e n t The applicant appears to conceptualize associations as evidence of causal relationships, although it is possible this simply reflects her usage o f English as a second language. The hypothesis as stated on page 17 is that “ functional polymorphisms in the androgen receptor gene and the PSA genes may together influence the development and progression o f breast cancer by modifying the efficiency o f the androgen signaling pathway.” This is well beyond the scope o f this proposal and method. Similarly, on p. 18 she 189 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PI Name: Wei Wang says that in order “to better measure the local action of androgens ... w e will measure androgen receptor and PSA genotypes”. The applicant has a reasonable career development plan, and her mentor indicates an excellent on-going mentorship which will continue. P rim a ry r e a s o n s fo r r e c o m m e n d e d ratin g The strengths of this application lie especially in the collaborations she has established for this project, and the use of data pertaining to African-American women. I do not see any glaring weaknesses. R E V IE W E R C I n n o v a tiv e n e s s In this dissertation work, the applicant will use state of the art genotyping methodology to investigate the role of androgen pathway genes, AR and PSA in breast cancer risk in African-American women. The mentoring team uses cutting edge methods for analysis o f molecular epidemiologic results, and the proposed work will provide excellent training for the applicant. The resources are rich and her mentor is one o f the rising stars in cancer molecular epidemiology, with broad and extensive background in biochemistry, molecular biology, epidemiology and statistics. Im p a ct The role o f androgens in breast cancer etiology is a new and important area o f research, and application o f the hypotheses to breast cancer in African-American women is novel and important A p p r o a ch T h is app lica tio n is a m o d el f o r a d isserta tio n a w a r d p ro p o s a l. The applicant gives an extensive, thoughtful and thorough review o f the literature, and builds a convincing case for her logical hypotheses. Preliminary data, the parent study, and the proposed study methodology are well-described in detail. There are some concerns about study power, particularly for the necessary stratified analyses by menopausal status, but this is acceptable for a dissertation project. There is also some concern that the applicant did not discuss the proposed aims in relation to increased risk for African-American women for breast cancer before age 40, and the more aggressive disease and poorer prognosis for black wom en o f all ages. H o w ever, it is h o p ed th a t these conun dru m s w ill b e e x p lo red in the d isserta tio n p r o c e s s . F e a sib ility There is no doubt that the research project will be successfully completed. C a r ee r D e v elo p m en t Dr. Ingles has written the strongest possible letter o f support for the candidate, attesting to her intelligence, perseverance and abilities. She has further noted her commitment to working with the applicant and guiding her training. Dr. Wang already has several publications, and it is clear that this further training in genotyping and data analysis and interpretation w ill enhance her capabilities to become an independent breast cancer researcher. P rim ary r e a s o n s fo r r e c o m m e n d e d rating This is a well-written and well-designed proposal submitted by a candidate with a very strong academic record and outstanding recommendations from her mentors. The project is innovative and important, 190 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PI Name: Wei Wang performed in a rich research environment, and the work will likely yield meaningful information. There is some concern that there was not a more thorough review of breast cancer in African- American women (early age at diagnosis, worse prognosis) in relation to androgen metabolism, but it is likely that these issues will be addressed as the research project matures. A D V O C A T E R E V IE W E R S u m m a ry o f th e a p p lic a tio n Using DNA from African American women from a previous case control study researchers propose to study whether or not androgens protect against breast cancer. Previous studies have yielded mixed results. The PI proposes to study the question from a genetic perspective, since shorter CAG repeats seem to be associated with lower breast cancer risk. The PSA pathway is proposed as a way in which androgens may operate in breast tissue. In n o v a tiv e n e s s The applicant has her M.D. from Beijing University and is working towards a Ph.D. in Epidemiology at USC. She has a Masters in Applied Biometry and Epidemiology. Im p a ct There are reportedly no previous association studies on genetic variation o f the PSA gene and the risk o f breast cancer, or on the androgen receptor gene in African American women. A p p ro a ch This presentation was exceptionally well organized. D N A samples from 500 African American women (250 breast cancer patients and 250 cancer free women) are available from a previous case control study for use in this project. It appears to be controlled for confounders. Consents have been signed. F ea sib ility Background and training are appropriate and I have no concerns about the research being able to be completed. C a reer D e v e lo p m e n t The mentor has expressed a high level o f commitment to the candidate and to her ability to be competitive in her next career phase. IV. O t h e r C o n s i d e r a t i o n s B u d g e t, PI tim e c o m m itm e n t, p r o je ct d u ra tio n , a n d overlap: No overlap in projects and appropriate timeline/duration for the project In order to keep within the budgetary constraints, the applicant has only budgeted $250 to attend the annual BCRP meeting, and nothing to attend a scientific meeting. Attending a meeting, such as the American Association for Cancer Research, or other meeting would contribute to her educational experience and training. 191 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PI Name: Wei Wang If she will be taxed on her tuition remission, she may wish to consider increasing the amount allotted there. H um an a n d a n im a l s u b j e c t s risk s: As stated above these studies focus on an underserved population o f breast cancer patients. IRB is pending approval but use o f existing samples has already been collected and agreed to be used in future studies. Confidentiality will be maintained for DNA samples, questionnaires and other identifying information. No risks, and no concerns at all. V . P r o g r a m m a t i c R e v ie w The application received a high programmatic review score. Although it did not, in the end, affect the recommendation for funding, a better priority choice would have been "Racial/Ethnic Differences in Breast Cancer", a higher priority at the programmatic level than Etiology. VI. K e y P o i n t s f r o m R e v i e w C o m m i t t e e D i s c u s s i o n : A very impressive candidate. Excellent overall project dealing with an underserved population; good cohort to work with. Excellent mentor. 192 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX II. MANUSCRIPT: ANDROGEN RECEPTOR AND PROSTATE-SPECIFIC ANTIGEN GENE POLYMORPHISMS AND BREAST CANCER IN AFRICAN-AMERICAN WOMEN (IN PRESS) Wei Wang1 , Esther M. John2 , Sue Ann Ingles1 1 Department of Preventive Medicine, University of Southern California, Los Angeles, CA 90089 2 Northem California Cancer Center, Fremont, CA 94538 Running Title: AR gene, PSA gene polymorphisms and breast cancer Key Words: androgen receptor, CAG repeat, prostate-specific antigen, genetic polymorphisms, breast cancer Footnotes: 1 Supported by California Breast Cancer Research Program Grant: 8GB-0103 2 Requests for reprints should be addressed to: Dr. Sue Ann Ingles, University of Southern California, USC/Norris Comprehensive Cancer Center, Room 6419,1441 Eastlake Avenue, Los Angeles, CA 90089-9175. 3 The abbreviations used are: AR, androgen receptor; PSA, prostate-specific antigen; ARE-I: androgen response element I; SNP, singles nucleotide polymorphism; BMI, body mass index; OR, odds ratio; Cl, confidence interval; SD, standard deviation; IQR, inter-quartile range. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Abstract: Several previous studies have found the CAG repeat polymorphism in Exon 1 of the androgen receptor (AR) gene to be associated with breast cancer risk among some groups of Caucasian and Asian women. In a population-based case- control study of 488 African-American women (239 cases and 249 controls), we examined this polymorphism along with a polymorphism (-158 G/A) in an androgen-regulated gene (PSA) whose expression has been correlated with breast cancer prognosis. Overall we did not observe any significant association between the CAG repeat polymorphism and breast cancer risk. However, among women with a first-degree family history of breast cancer, longer CAG repeats were associated with a significantly increased risk. Women carrying at least one longer allele (CAG)n>22)) had a three-fold increased risk compared to those with two shorter alleles (OR=3.18, 95% Cl: 1.08-9.36). There was no significant association between the PSA gene polymorphism and breast cancer risk, nor was there significant gene-gene interaction. In summary, our results further support that shorter CAG repeats (stronger AR transactivation activity) may reduce the risk of breast cancer, at least among some groups of women. Our data, however, are unable to provide evidence that PSA is the pathway through which the protective effect of androgens operates. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Introduction: The role of androgens in the etiology of breast cancer remains unclear. Tumor growth-inhibitory effects of testosterone and dihydrotestosterone have been observed in breast cancer cell lines and in animal models (1), whereas higher circulating androgen levels have been noted in breast cancer patients compared to controls (2, 3). Androgen and estrogen levels are highly correlated. Therefore, a positive association between androgen levels and breast cancer risk may reflect the effects of high concomitant estrogen levels and activities. Adjustment for circulating estrogen has attenuated the association with testosterone levels in some prospective studies (4-7), but not in others (8-10). However, because substantial conversion of androgens to estrogens occurs in breast adipose tissue (11), adjustment for circulating estrogen may not adequately adjust for local estrogen levels in the breast. An alternative approach is to examine androgen receptor (AR) gene variants that alter the receptor function. A CAG repeat polymorphism in Exon 1 encodes a variable-length polyglutamine tract in the transactivation domain of the protein. Long polyglutamine tract length reduces AR transactivation activity in vitro (12-14). Some (15-19), but not all (20-24) epidemiological studies found longer CAG repeats associated with increased breast cancer risk. 195 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The interaction with down-stream genes in the androgen signaling pathway, such as the PSA gene (also named KLK3), may also be important. Prostate-specific antigen (PSA) protein, encoded by the PSA gene is present in breast tissue (25), and may be a useful prognostic marker in breast cancer (26, 27). A guanine to adenine substitution (-158 G/A) in an androgen-responsive element (ARE-I) of the PSA gene promoter, has been associated with PSA levels in serum or in breast tissue in some (28-30), but not all studies (31, 32) and with prognosis (30). We examined the AR CAG length polymorphism and the single nucleotide polymorphism (SNP) in the ARE-I of the PSA gene in relation to breast cancer risk in African-American women, a population with a notably shorter mean CAG repeat length and a wider CAG repeat length distribution compared to Whites and Asians (33). Material and Methods Study Population Study subjects were participants in a population-based case-control study conducted in the San Francisco Bay area (34). In brief, cases aged 35-79 years and newly diagnosed with invasive breast cancer between 1997 and 1999 were identified through the regional cancer registry. Controls were identified through random digit 196 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. dialing (81% response to household enumeration). 249 cases and 255 controls completed a telephone screening interview (84% of cases, 86% of controls), and an in-person interview (87% of cases, 82% of controls), and provided a blood or mouthwash sample (85% of cases, 84% of controls). DNA was available for 246 cases and 255 controls. The study was approved by the Institutional Review Boards of both the University of Southern California and the Northern California Cancer Center. Laboratory methods The AR exon 1 CAG repeat variant was genotyped by simple sequence length polymorphism (SSLP) analysis. The genomic region containing the CAG repeat was PCR amplified using the forward primer 5'-CGCGAAGTGATCCAGAAC-3’ and the reverse primer 5 ’-CAGGACCAGGTAGCCTGTG-3 ’ ((FAM-labeled) (Applied Biosystems, Foster City, CA). Touchdown thermal cycling was performed. The resulting PCR product was run on the A B I3700 capillary sequencer and allele sizes were scored using GeneScan software (version 3.5) (Applied Biosystems, Foster City, CA). DNA samples from 12 male subjects with various CAG repeat lengths (determined from direct sequencing) were included in each run as controls. A standard curve was drawn based on these 12 control samples and was used to calculate CAG repeat number for study subjects. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Genotyping of the SNP in the PSA gene was performed by the Taqman assay (Applied Biosystems, Foster City, CA). The two labeled oligonucleotide probes were 5’-FAM-CAGAACAGC AAGTACTAGCTCTCCCTC-3'and 5’-CY3- AGAACAGCAAGTGCTAGCTCTCCC-3In both probes the thymidines were replaced with Propyne-dU to increase the Tm of the probe ~1°C for every addition (Biosearch Technologies, INC, Novato, CA). The forward Primer was 5'- GGTGCATCCAGGGTGATCTAG-3' and reverse primer was 5'- CACACCCAGAGCTGTGGAAG-3'. Nine previously sequenced DNA samples (3 of each genotype) were included as genotyping controls. Ambiguous genotyping results were confirmed by sequencing. Concordance for duplicates (5% random sample of all blood specimens) was 100%. Statistical analysis: We refer to the two AR CAG alleles carried by each woman as the smaller allele (the shorter of the two) and the larger allele (the longer of the two). SAS PROC ALLELE was used to assess Hardy-Weinberg equilibrium (HWE) of the CAG length distribution among controls (SAS Institute Inc., Cary, NC). Wilcoxon rank sum test was used to compare the distributions of the repeat lengths (for the smaller and the larger allele separately). Logistic regression was used to estimate odds ratio (OR) and 95% confidence interval (Cl) for the effect of AR CAG repeat 198 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. length on breast cancer risk. CAG repeat lengths were dichotomized as short and long using a cutoff point (CAG=22) commonly used in previous publications (17, 18, 20). Other cutoff points including the median were also examined. Women with 1 or 2 alleles > 22 were combined given the few women with 2 alleles > 22. For the PSA gene, ORs and 95% CIs were estimated comparing genotypes A/A and G/A to G/G. A test of trend was performed by including in the logistic model a variable coded as 0, 1,2 for the number of “at-risk” alleles. All models were adjusted for age (continuous). Adjustment for other known breast cancer risk factors (see Table 1) did not, either individually or jointly, change the OR estimates by more than 10%. We considered menopausal status (defined as in (17)), hormone replacement therapy (HRT) use, and first-degree family history of breast cancer as potential effect modifiers. Formal tests of effect modification were performed by including the appropriate interaction terms in the logistic model. To assess possible interactions between the AR and PSA genes, ORs were estimated for each AR/PSA genotype combination. Power for detecting a shift of two CAG repeats between the distributions of CAG repeat lengths in cases and controls with a positive family history was estimated by bootstrap (35). With each bootstrap sample comprising 67 observations (32 cases, 35 controls), 1000 samples were drawn with replacement from the 199 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. empirical distribution of controls in our dataset, with two CAG repeats being added to the sampled value for each case. Wilcoxon rank-sum tests were performed on each sample and the percentage of significant results was calculated. Results Genotyping results were missing for 13 subjects (7 cases and 6 controls) due to PCR failure (Table 1). The CAG repeat lengths ranged from 8 to 30 ( median=19) and the distribution in controls did not show significant departure from HWE (p=0.14). Overall, the distributions of the CAG repeat lengths in cases and controls were almost identical (Figure la) (p=0.94 for the smaller allele; p=0.64 for the larger allele). Compared with women having no allele with (CAG)n > 22 (corresponding to the SS genotype in the literature), women who carried one or two alleles with (CAG)n > 22 (SL and LL genotypes, respectively) had an OR=1.09 (95% CI-0.75- 1.57) (Table 2). Similar results were obtained when other cutoff points were used. Mean age at diagnosis of cases did not differ by their CAG repeat genotypes (data not shown). The association between CAG repeat genotypes and breast cancer risk did not differ significantly by menopausal status (Table 2), and, among postmenopausal women, did not differ significantly by HRT use (p for interactional).48, data not shown). 200 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cases and controls without a first-degree family history of breast cancer had nearly identical distributions (for both the smaller and the larger alleles), whereas among subjects with a family history, the distribution of the larger allele was noticeably shifted to the right among cases compared to controls (p=0.008) (Figures lb and lc). Among women with a family history, having one or two alleles with 22 or more CAG repeats was associated with a significantly increased risk (OR=3.18, 95% 0=1.08-9.36), while no association was observed among women without a family history (OR=0.92, 95% 0=0.62-1.37) (p for interaction^.03) (Table 2). Similar results were obtained when other cutoff points were used (data not shown). The distribution of the PSA -158A/G genotype frequencies did not show significant departure from HWE among controls (p=0.29). There was no significant association between breast cancer risk and PSA genotype overall. Nor was there any significant interaction with menopausal status (p=0.10), or family history (p=0.82) (Table 2). There was no evidence for a significant gene-gene interaction when the effects of PSA genotype were estimated within strata defined by the AR genotypes (OR=1.08, 95% 0=0.64-1.83 for AG vs. AA and OR=0.80, 95% 0=0.42-1.54 for GG vs. AA among women with SS genotype and OR=0.85, 95% 0=0.42-1.72 for AG vs. AA and OR=1.64, 95% 0=0.66-4.09 for GG vs. AA among women with SL 201 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. or LL genotypes (p for interaction^. 15, data not shown). We were not able to test for interaction between the AR and the PSA genotypes among women with a positive family history due to small sample size. Discussion Among African-American women with a first-degree family history of breast cancer, a significant increase in breast cancer risk was associated with carrying one or two AR CAG long alleles. Our results agree with the Nurses’ Health Study that reported a reduced risk with shorter CAG repeats only among women with a positive family history of breast cancer (17). and with a study by Rebbeck et al in BRCA1 mutation carriers that found a significant risk reduction associated with shorter CAG repeats (15). A statistical interaction between AR genotype andBRCAl mutation status is strongly supported by in vitro studies suggesting that BRCA1 protein is an AR coactivator (36, 37). We had no information available on BRCA1 mutation status to investigate this interaction. However, the number of BRCA1 mutation carriers is likely to be small (38). It is possible that the interaction between family history and AR CAG genotype observed in our study and in the Nurse’s Health Study is due, at least in part, to variant(s) in other gene(s) or to some other familial risk factor(s). 202 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Three other studies (22-24), however, did not confirm the finding by Rebbeck et al. (15), possible due to small sample size. Our family history-positive stratum was also small. But by examining the entire distribution curve rather than relying solely on outpoints, we had reasonable power (74%) to detect a shift of two CAG repeats between cases and controls with a family history. Nevertheless, confirmation by studies with larger numbers of BRCA1 mutation carriers and/or family history positive subjects is needed Unlike previous studies, we explored the possible mechanism of the androgen effect by examining a genetic variation in the PSA gene, an AR down stream gene, together with the AR CAG polymorphism. Some (28, 29) but not all (31, 32) studies have suggested that a polymorphism in the ARE-I of the PSA gene (-158 G/A) may contribute to inter-individual variations in serum PSA levels in men. The only study in females found the A allele to be associated with lower PSA concentration in breast tumor tissue (30). Cases with the AA genotype (lower PSA) also had worse survival than cases with GG or GA genotypes (30). We found no association with this PSA polymorphism, possibly because it is not functional. In an in vitro study the two alleles showed no differences in PSA gene promoter activity (32). We speculate that the association of the ARE-I polymorphism with serum PSA found in some studies may be a result of its linkage disequilibrium 203 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with other functional polymorphisms in the PSA gene. Several polymorphisms in the 5’ enhancer region of the PSA gene have recently been associated with serum PSA levels (39). Three of them were also found to be functional in reporter gene assays (39). However, the region containing these polymorphisms did not seem to be critical in stimulating PSA gene transcription in breast cancer cells (40), suggesting that this enhancer region might be tissue-specific. Our finding adds to the literature suggesting that androgen protects against breast cancer in some groups of women. While we could not rule out involvement of the PSA pathway in family history-positive women (as we could not examine any possible gene-gene interaction between AR and PSA in this small subgroup), we were unable to provide any evidence that PSA is the pathway through which the protective effect of androgen operates. Other androgen target genes need to be investigated. We realize that our study has small sample size, especially when examining interactions. Our results (both positive and negative) need to be taken with caution. This is the first study to examine the AR CAG polymorphism in African- American women. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References: 1. Somboonpom W, Davis SR. Testosterone effects on the breast: implications for testosterone therapy for women. Endocr Rev 2004;25:374-88. 2. Lillie EO, Bernstein L, Ursin G. The role of androgens and polymorphisms in the androgen receptor in the epidemiology of breast cancer. Breast Cancer Res 2003;5:164-73. 3. Somboonpom W, Davis SR. Postmenopausal testosterone therapy and breast cancer risk. Maturitas 2004;49:267-75. 4. Zeleniuch-Jacquotte A, Bmning PF, Bonffer JM, et al. Relation of serum levels of testosterone and dehydroepiandrosterone sulfate to risk of breast cancer in postmenopausal women. Am J Epidemiol 1997;145:1030-8. 5. Thomas HV, Key TJ, Allen DS, et al. A prospective study of endogenous serum hormone concentrations and breast cancer risk in post-menopausal women on the island of Guernsey. Br J Cancer 1997;76:401-5. 6. Hankinson SE, Willett WC, Manson JE, et al. Plasma sex steroid hormone levels and risk of breast cancer in postmenopausal women. J Natl Cancer Inst 1998;90:1292-9. 7. Cauley JA, Lucas FL, Kuller LH, Stone K, Browner W, Cummings SR. Elevated semm estradiol and testosterone concentrations are associated with a high risk for breast cancer. Study of Osteoporotic Fractures Research Group. Ann Intern Med 1999;130:270-7. 8. Berrino F, Muti P, Micheli A, et al. Semm sex hormone levels after menopause and subsequent breast cancer. J Natl Cancer Inst 1996;88:291-6. 9. Dorgan JF, Longcope C, Stanczyk FZ, Stephenson HE, Jr., Hoover RN. Re: Plasma sex steroid hormone levels and risk of breast cancer in postmenopausal women. J Natl Cancer Inst 1999;91:380-1. 10. Key T, Appleby P, Barnes I, Reeves G. Endogenous sex hormones and breast cancer in postmenopausal women: reanalysis of nine prospective studies. J Natl Cancer Inst 2002;94:606-16. 11. Thijssen JH. Local biosynthesis and metabolism of oestrogens in the human breast. Maturitas 2004;49:25-33. 205 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12. Chamberlain NL, Driver ED, Miesfeld RL. The length and location of CAG trinucleotide repeats in the androgen receptor N-terminal domain affect transactivation function. Nucleic Acids Res 1994;22:3181-6. 13. Kazemi-Esfarjani P, Trifiro MA, Pinsky L. Evidence for a repressive function of the long polyglutamine tract in the human androgen receptor: possible pathogenetic relevance for the (CAG)n-expanded neuronopathies. Hum Mol Genet 1995;4:523-7. 14. Tut TG, Ghadessy FJ, Trifiro MA, Pinsky L, Yong EL. Long polyglutamine tracts in the androgen receptor are associated with reduced trans-activation, impaired sperm production, and male infertility. J Clin Endocrinol Metab 1997;82:3777-82. 15. Rebbeck TR, Kantoff PW, Krithivas K, et al. Modification of BRCA1- associated breast cancer risk by the polymorphic androgen-receptor CAG repeat. Am J Hum Genet 1999;64:1371-7. 16. Giguere Y, Dewailly E, Brisson J, et al. Short polyglutamine tracts in the androgen receptor are protective against breast cancer in the general population. Cancer Res 2001;61:5869-74. 17. Haiman CA, Brown M, Hankinson SE, et al. The androgen receptor CAG repeat polymorphism and risk of breast cancer in the Nurses' Health Study. Cancer Res 2002;62:1045-9. 18. Suter NM, Malone KE, Daling JR, Doody DR, Ostrander EA. Androgen receptor (CAG)n and (GGC)n polymorphisms and breast cancer risk in a population-based case-control study of young women. Cancer Epidemiol Biomarkers Prev 2003;12:127-35. 19. Liede A, Zhang W, De Leon Matsuda ML, Tan A, Narod SA. Androgen receptor gene polymorphism and breast cancer susceptibility in The Philippines. Cancer Epidemiol Biomarkers Prev 2003;12:848-52. 20. Spurdle AB, Dite GS, Chen X, et al. Androgen receptor exon 1 CAG repeat length and breast cancer in women before age forty years. J Natl Cancer Inst 1999;91:961-6. 21. Dunning AM, McBride S, Gregory J, et al. No association between androgen or vitamin D receptor gene polymorphisms and risk of breast cancer. Carcinogenesis 1999;20:2131-5. 206 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22. Menin C, Banna GL, De Salvo G, et al. Lack of association between androgen receptor CAG polymorphism and familial breast/ovarian cancer. Cancer Lett 2001;168:31-6. 23. Kadouri L, Easton DF, Edwards S, et al. CAG and GGC repeat polymorphisms in the androgen receptor gene and breast cancer susceptibility in BRCA1/2 carriers and non-carriers. Br J Cancer 2001;85:36-40. 24. Spurdle AB, Antoniou AC, Duffy DL, et al. The androgen recetor CAG repeat polymorphism and modification of breast cancer risk in BRCA1 and BRCA2 mutation carriers. Breast Cancer Res 2005;7:R176-R183. 25. Yousef GM, Diamandis EP. The new human tissue kallikrein gene family: structure, function, and association to disease. Endocr Rev 2001;22:184-204. 26. Yu H, Giai M, Diamandis EP, et al. Prostate-specific antigen is a new favorable prognostic indicator for women with breast cancer. Cancer Res 1995;55:2104-10. 27. Yu H, Levesque MA, Clark GM, Diamandis EP. Prognostic value of prostate-specific antigen for women with breast cancer: a large United States cohort study. Clin Cancer Res 1998;4:1489-97. 28. Xue WM, Coetzee GA, Ross RK, et al. Genetic determinants of serum prostate-specific antigen levels in healthy men from a multiethnic cohort. Cancer Epidemiol Biomarkers Prev 2001;10:575-9. 29. Medeiros R, Morais A, Vasconcelos A, et al. Linkage between polymorphisms in the prostate specific antigen ARE1 gene region, prostate cancer risk, and circulating tumor cells. Prostate 2002;53:88-94. 30. Bharaj B, Scorilas A, Diamandis EP, et al. Breast cancer prognostic significance of a single nucleotide polymorphism in the proximal androgen response element of the prostate specific antigen gene promoter. Breast Cancer Res Treat 2000;61:111-9. 31. Xu J, Meyers DA, Sterling DA, et al. Association studies of serum prostate- specific antigen levels and the genetic polymorphisms at the androgen receptor and prostate-specific antigen genes. Cancer Epidemiol Biomarkers Prev 2002;11:664-9. 207 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32. Rao A, Chang BL, Hawkins G, et al. Analysis of G/A polymorphism in the androgen response element I of the PSA gene and its interactions with the androgen receptor polymorphisms. Urology 2003 ;61:864-9. 33. Edwards A, Hammond HA, Jin L, Caskey CT, Chakraborty R. Genetic variation at five trimeric and tetrameric tandem repeat loci in four human population groups. Genomics 1992;12:241-53. 34. John EM, Hom-Ross PL, Koo J. Lifetime physical activity and breast cancer risk in a multiethnic population: the San Francisco Bay area breast cancer study. Cancer Epidemiol Biomarkers Prev 2003;12:1143-52. 35. Codings BJ, Hamilton MA. Estimating the power of the two-sample Wilcoxon test for location shift. Biometrics 1988;44:847-60. 36. Park JJ, Irvine RA, Buchanan G, et al. Breast cancer susceptibility gene 1 (BRCAI) is a coactivator of the androgen receptor. Cancer Res 2000;60:5946-9. 37. Yeh S, Hu YC, Rahman M, et al. Increase of androgen-induced cell death and androgen receptor transactivation by BRCAI in prostate cancer cells. Proc Natl Acad Sci U S A 2000;97:11256-61. 38. Newman B, Mu H, Butler LM, Millikan RC, Moorman PG, King MC. Frequency of breast cancer attributable to BRCAI in a population-based series of American women. Jama 1998;279:915-21. 39. Cramer SD, Chang BL, Rao A, et al. Association between genetic polymorphisms in the prostate-specific antigen gene promoter and serum prostate-specific antigen levels. J Natl Cancer Inst 2003;95:1044-53. 40. Schuur ER, Henderson GA, Kmetec LA, Miller JD, Lamparski HG, Henderson DR. Prostate-specific antigen expression is regulated by an upstream enhancer. J Biol Chem 1996;271:7043-51. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 1 . Characteristics of African-American study participants, by case-control status Cases Controls P value (n=239) j (n=249) j________________ Age Mean (SD)* 55.6(11.5) 55.3 (11.6) 0.79 Median (IQR)+ 54 (47-64) 54 (46-65) 0.78 Menopausal status Premenopausal 72(30.1%) 82 (32.9%) Postmenopausal 146 (61.1%) 146 (58.6%) 0.80 Undetermined 21 (8.8%) 21 (8.4%) Education (years) <12 42(17.6% ) 46(18.5% ) 12 53 (22.2%) 64 (25.7%) 13-16 94 (39.3%) 101 (40.6%) >17 50 (20.9%) 38(15.3%) 0.41 First-degree family history Yes 35 (14.6%) 32 (12.8%) No 204 (85.4%) 217(87.2%) 0.57 Benign breast disease Yes 61 (25.5%) 38(15.4%) No 178 (74.5%) 209 (84.6%) 0.006 Age at menarche <12 52(21.9% ) 55 (22.4%) 12-13 133 (55.9%) 130(52.9%) >14 53 (22.3%) 61 (24.8%) 0.76 Age at menopause Mean (SD) 46.4 (7.02) 46.4 (8.08) 0.97 Median(IQR)1 ' 48(44-51) 48 (43-51) 0.78 Number of full-term pregnancies Nulliparous 44(18.4% ) 34(13.7% ) 1 42(17.6% ) 47(18.9% ) 2 54 (22.6%) 55 (22.1%) 3 43 (18.0%) 48(19.3% ) >4 56 (23.4%) 65 (26.1%) 0.68 Age at 1 full-term pregnancy <20 83 (42.6%) 93 (43.3%) 20-24 59 (30.3%) 79 (36.7%) 25-29 27(13.9% ) 27(12.6%) >30 26(13.3 %) 16(7.4%) 0.18 209 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 1. Characteristics of African-American study participants, by case-control status (cont.) History o f oral contraceptive use Yes 160 (67.5%) 163 (66.0%) No 77 (32.5%) 84 (34.0%) 0.72 History o f HRT use (postmenopausal women) Yes 70 (48.0%) 83 (58.0%) No 76 (52.0%) 60 (42.0%) 0.09 BMI in premenopausal women Mean (SD)* 30.1 (6.21) 32.4 (8.20) 0.05 Median(IQR)1 ' 29.3 (25.0-34.5) 30.5 (25.5-39.9) 0.11 <25 18(25.0%) 16(19.5%) 25-29 20 (27.8%) 23 (28.1%) >30 34 (47.2%) 43 (52.4%) 0.69 BMI in postmenopausal women Mean (SD)* 31.2 (6.34) 31.5 (7.17) 0.71 Median(IQR)t 30.5 (26.6-35.2) 30.8 (26.5-34.6) 0.97 <25 24(16.6%) 26(17.8% ) 25-29 44 (30.3%) 40 (27.4%) >30 77 (53.1%) 80 (54.8%) 0.85 * SD , standard deviation f IQR, inter-quartile range { The numbers in the table do not add up due to m issing values Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 2. The association between the AR CAG and PSA ARE-I -158 G/A polymorphisms and breast cancer in African-American women CAG length Cases Controls OR (S:<22; L:>22) N (%) N (%) (95% Cl) A ll women SS 145 (60.7) 156 (62.7) 1.0 SL or LL 94 (39.3) 93 (37.3) 1.09 (0.75-1.57) Menopausal status Pre- SS 43 (59.7) 51 (62.2) 1.0 SL or LL 29 (40.3) 31 (37.8) 1.11 (0.58-2.12) Post- SS 90 (61.6) 90 (61.6) 1.0 SL or LL 56 (38.4) 56 (38.4) 1.01 (0.63-1.61) p inter*=0.82 First-degree family history Yes SS 18(51.4) 25 (78.1) 1.0 SL or LL 17(48.6) 7(21.9) 3.18(1.08-9.36) No SS 127 (62.3) 131 (60.4) 1.0 SL or LL 77 (37.7) 86 (39.6) 0.92 (0.62-1.37) p inter*=0.03 Cases Controls OR ARE-I genotypes N (%) N (%) (95% Cl) All women AA 65 (27.2) 68 (27.3) 1.0 AG 125 (52.3) 132 (53.0) 0.98 (0.65-1.50) GG 49 (20.5) 49(19.7) 1.04 (0.62-1.76) p trendt= 0.90 Menopausal status Pre- AA 26 (36.1) 28 (34.1) 1.0 AG 31 (43.1) 44 (53.7) 0.75 (0.37-1.52) GG 15 (20.8) 10(12.2) 1.81 (0.68-4.84) p trendt=0.46 Post- AA 32(21.9) 37 (25.3) 1.0 AG 86 (58.9) 74 (50.7) 1.33 (0.76-2.35) GG 28(19.2) 35 (24.0) 0.92 (0.47-1.84) p trend' =0.86 p inter* =0.10 First-degree family history Yes AA 10 (28.6) 8 (25.0) 1.0 AG 18 (51.4) 19 (59.4) 0.80 (0.25-2.51) GG 7 (20.0) 5 (15.6) 1.05 (0.24-4.68) p trendt =0.99 No AA 55 (27.0) 60 (27.6) 1.0 AG 107 (52.5) 113 (52.1) 1.04 (0.66-1.63) GG 42 (20.6) 44 (20.3) 1.05 (0.60-1.84) p trend1 =0.86 p inter* =0.82 * p value for test of interaction + p value for test of trend Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 1. The cumulative distribution curves o f the CAG repeat lengths for the smaller and the larger allele o f the 2 alleles carried by each woman, (a) in all cases (n=239) and controls (n=249). p*=0.94 (smaller allele), p=0.64 (larger allele); (b) in cases (n=35) and controls (n=32) with a lst-degree family history o f breast cancer p=0.62 (smaller allele), p=0.008 (larger allele); and (c) in cases (n=204) and controls (n=217) without a lst-degree family history o f breast cancer p=0.93 (smaller allele), p=0.62 (larger allele). (*p values are for W ilcoxon tests). — *— case_smaller — ■— c a se ja rg e r — *— ctrl_smaller ctrljarger 8 o 28 30 12 14 16 18 20 22 24 26 8 10 Nurrtier o f CAG repeats 8 8 c 0 ) o 10 12 14 16 18 20 22 24 26 28 30 8 Nurrtier of CAG repeats Figure 1c 8 ,8 8 O 10 12 14 16 18 20 22 24 26 Nurrber of CAG repeats 28 30 8 Figure 1b 212 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX III. GRANT PROPOSAL: ANDROGEN RECEPTOR GENE, P21 GENE IN BREAST CANCER RISK (FUNDED BY THE CBCRP POSTDOCTORAL AWARD) 1. INTRODUCTION and HYPOTHESIS Breast cancer is a hormone-related cancer. While estrogens are generally accepted as mitogenic and cancer-promoting for human breast cancer, the roles of androgens remain unclear. Androgens, because they are precursors of estrogens, were originally postulated as risk factors for breast cancer. This hypothesis was supported by epidemiological studies comparing circulating androgens in patients and healthy controls, especially among postmenopausal women (reviewed in (1-4)). However, accumulating new evidence from experimental studies has suggested that androgens may also act directly (through binding to androgen receptor) to protect against the development and progression of breast cancer. If this direct protective effect of androgens is verified, the overall effect of androgens would be a result of the balance between their direct action and indirect action (through their conversion to estrogens by aromatase). It has been suggested by Somboonporn et al that in situations where the conversion of androgens to estrogens is low (when aromatase activity is suppressed), the protective effect of androgens may manifest. Moreover, based on the fact that aromatase activity is apparently suppressed in women who take hormone replacement therapy (HRT), they proposed that adding androgens to HRT may oppose the cancer-promoting effect of estrogens and progesterone in the breast (5). If this hypothesis is proven to be true, it could have tremendous health implications. Furthermore, with the increasing public interest in the use of androgen replacement therapy (alone or in addition to HRT) in symptomatic women undergoing menopause, understanding the roles of androgens in breast cancer becomes critical. The epidemiological studies on circulating androgen levels and breast cancer have a major limitation in that they are not able to look at the effect of androgens independent of estrogens, because the circulating levels of these hormones are highly correlated. Furthermore, circulating androgens are converted to estrogen in the breast tissue (especially in menopausal women), and circulating levels of estrogens may not reflect estrogen levels in the breast. We therefore would not able to estimate the independent effect of androgens by simply adjusting for circulating estrogen levels in our statistical analysis. An alternative approach to studying the effects of androgens on breast cancer is to examine androgen receptor (AR) gene variants that alter the receptor function. This approach provides a way to specifically examine the independent direct effects of androgens in breast cancer development. In the first exon of the AR 213 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. gene there is a CAG repeat polymorphism which encodes a variable-length polyglutamine tract in the amino-terminal domain (the transactivation domain) of the protein. It has been shown in several in vitro studies that the polyglutamine tract length is inversely associated with the transactivation activity of the AR (6-9). Recent epidemiological studies on this AR CAG polymorphism have lent some support for a protective role of androgens in breast cancer in that shorter CAG repeats (with stronger transactivation activity of the AR) have been associated with reduced breast cancer risk (10-14), supporting what has been found in most experimental studies. But not all studies have found this association (15, 16). In our recently completed case-control study of this polymorphism and breast cancer among 501 African-American women (manuscript in preparation, supported by BCRP grant 8GB-0103), we also found a reduced risk in women carrying two alleles with shorter CAG repeats, compared to women carrying at least one allele with longer CAG repeats. Consistent with results of the Nurses’ Health Study by Haiman et al (12), the association between the CAG repeat length and breast cancer risk was only found among women with a positive 1s t-degree family history of breast cancer. Our finding is also in accordance with Rebbeck’s study which reported an association between the CAG repeat polymorphism and breast cancer risk in female BRCA1 carriers (10). However, because of the small number of subjects (in Rebbeck’s study and in the family history positive group in the studies by Haiman et al and by us), no firm conclusions can be drawn without the replication of these findings in larger studies (preferably among women carrying BRCA1 mutations). Therefore, in this study, we propose to re-examine our finding in two large population-based studies (in Northern and Southern California) with three different ethnic groups. In addition, we propose to look at a gene downstream of the androgen receptor, the p21 gene (also known as the cyclin-dependent kinase inhibitor 1A, CDKN1A). The p21 gene is one of the very few well-characterized genes that are transcriptionally regulated by androgen (through the androgen- responsive element in the promoter region of the gene). The p21 protein is a major component of the cell cycle control system, and plays a critical role in blocking cell cycle transition from G1 to S phase by inhibiting cyclin and cyclin-dependent kinase complexes(17). It is reasonable to postulate that p21 may play an important role in mediating the growth inhibitory effect of androgens observed in both in vitro and in vivo studies (see below). Three relatively common polymorphisms in the p21 gene have been reported in the literature. Associations between two of these polymorphisms and the risk of breast cancer have been suggested by a very limited number of small studies, although there is a lack of evidence on the functional significance of these polymorphisms. Primary Aims: 1. To determine whether the androgen receptor (AR) CAG repeat polymorphism is associated with breast cancer risk, especially among women with a 1s t-degree family history of breast cancer. 214 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2. To identify all common polymorphisms in the evolutionarily conserved regions in the 5’ upstream regulatory region of the p21 gene. 3. To determine whether common polymorphisms in the p21 gene are associated with breast cancer risk, especially among women with a positive 1s t-degree family history of breast cancer. Secondary Aims: 1. To test for possible interaction between polymorphisms in the AR gene and in the p21 gene with respect to breast cancer risk. 2. To determine whether the association between the AR CAG polymorphism and breast cancer differs by BRCA1 mutation status in a case-case study (using a subgroup of cases with known BRCA1 mutation status). 3. To determine whether the associations between p21 polymorphisms and breast cancer differ by BRCA1 mutation status in a case-case study (using a subgroup of cases with known BRCA1 mutation status). 2. BACKGROUND and SIGNIFICANCE 2.1 Androgens and breast cancer in experimental and clinical studies: Evidence from most experimental studies suggests that androgens may protect against breast cancer. For example, many studies have found that testosterone and/or its metabolite, dihydrotestosterone (DHT), at physiological concentrations inhibited the growth of human breast cancer cell lines MCF-7, ZR-75-1, T47D, CAMA-1 and MFM-223, though a few other studies observed a growth-stimulating effect of androgens, the difference in cell lines (and their culture conditions), and the doses of androgens used in these experiments may be the most likely explanations of the divergent results (reviewed in (2)). Biphasic effects of androgens on MCF-7 cell line (growth-inhibitory at low concentrations and stimulatory at very high concentrations) have been reported by two studies (18, 19). A growth-inhibitory effect of androgen has been repeatedly shown in animal studies. In some studies, it appears that this is an antiestrogen effect of androgens. For example, testosterone was able to suppress mammary epithelial estrogen receptor expression and inhibit estrogen-induced mammary epithelial proliferation in ovariectomized female rhesus monkeys (20). Supporting this view, another study found that a low-dose oral contraceptive induced robust mammary epithelial proliferation in rats, but the addition of methyltestosterone to the therapy significantly suppressed the proliferation (21). On the other hand, other studies observed that the inhibitory effect of androgens could occur in the absence of estrogenic stimulation and thus suggested a direct effect of androgens that is independent of the modulation on estrogen receptor levels. For example, DHT exhibited tumor- suppressing effects on ZR-75-1 xenograft in athymic mice, either in the presence or 215 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. absence of exogenous estradiol (22). DHT was also shown to inhibit the growth of chemically (DMBA)-induced mammary tumors in rats (23). The growth inhibitory effects of androgens are mediated by androgen receptor (AR), as blocking androgen receptor with its antagonist resulted in the abolishment of the effect both in vitro (24, 25) and in vivo (20-23). 2.2 Epidemiological studies on androgen receptor polymorphism in breast cancer: Recent epidemiological studies on the functional polymorphism in the exon 1 of the AR gene, the CAG repeats, have lent some support for a protective role of androgens in breast cancer. Shorter CAG repeats (with stronger transactivation activity of the AR) have been associated with a reduced breast cancer risk in several studies (10-14), although others failed to find this association (15,16, 26). Interestingly, in most of the studies where a significant association between the CAG repeat length and breast cancer risk was reported, the association was limited to some specific groups of women, for example, postmenopausal women (11); women with a family history of breast cancer (12), or BRCA1 carriers (10). In a recent study of 501 African-American women conducted by our group (paper in preparation, supported by BCRP grant 8GB-0103), we also found an increased risk in women who carry at least one allele with a longer CAG repeat, compared to women carrying two alleles with a shorter CAG repeat. Consistent with results of the Nurses’ Health Study by Haiman et al (12), the association between the CAG repeat length and breast cancer risk was only found among women with a 1s t-degree family history of breast cancer. Our finding is also in accordance with Rebbeck’s study which reported an association between the CAG repeat polymorphism and breast cancer risk in female BRCA1 carriers (10), because we expect to have a certain percentage of women with family history of breast cancer to carry BRCA1 mutations (12, 27). The statistical interaction between the AR CAG repeat polymorphism and BRCA1 status (or family history of breast cancer) suggested by ours and two previous epidemiological studies (10, 12), is strongly supported by two in vitro studies, showing that the BRCA1 protein is an AR coactivator (28, 29). It was demonstrated in these two studies that wild-type BRCA1 gene product could enhance the ligand-dependent AR transactivation of reporter genes in breast cancer cell lines MCF-7 and T47D. This was accompanied by a direct physical contact observed in GST pull-down assays between BRCA1 and AR (and also AR coactivators p160) in (28). What is even more compelling as an explanation for our finding is the observation made by Park et al that when the BRCA1 and AR genes were co-expressed, the reduced transactivation activity associated with increasing poly-Q (glutamine) length was no longer observed in the prostate cancer cell line PC-3. A hypothesis was therefore proposed by the authors that among women with germline BRCA1 mutations (having reduced BRCA1 protein function, and thus less activation of AR signaling), those with less efficient genotypes (i.e. ARs with longer polyQ (glutamine) tracts, encoded by longer CAG repeats) may be at a higher risk of cancer development, compared to those with more efficient AR genotypes (encoded by shorter CAG repeats), while in women with no BRCA1 germline mutation, the 216 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. reduced AR signaling associated with longer CAG repeats may not be manifested. However, reproducibility in more in vitro studies, especially ones conducted in breast cancer cell lines are still need to further support this hypothesis. In our previous study, the lack of information on BRCA1 mutation status precludes us from directly investigating this interaction between AR and BRCA1. We also realize that sample size became very small in the family-positive stratum in both of the studies by Haiman et al and by us. The small sample size is also an issue in the study by Rebbeck et al. More studies with larger samples sizes (preferably among BRCA1 carriers) are still needed to confirm the findings by us and by the other two groups. 2.3 Androgen signaling and cell cycle control: The mechanisms of the androgen-induced growth inhibition of breast cancer cells remain unclear, due to our limited knowledge of down-stream genes in the androgen signaling pathway. In vitro, androgen treatment has been shown to lead to an increased proportion of cells in the G1 phase of the cell cycle (30, 31). p21 (also referred to as cyclin-dependent kinase inhibitor-1 A (CDKN1A); wildtype p53-activated fragment 1 (WAF1); or cyclin- dependent kinase-interacting proteins (CIP1)) is a critical negative regulator of cell cycle progression (see below) and is also one of the few well-characterized molecules whose gene transcription is directly regulated by androgens (with an androgen-responsive element at position -200 in the promoter region of the gene) (32), which makes it a very attractive down-stream effector to study to elucidate the role of androgens in breast cancer. 2.4 p21 in cell cycle control: Progression from G1 to S phase of the cell cycle involves a complicated network of cyclins, cyclin-dependent kinases (CDKs), CDK inhibitors, tumor suppressor retinoblastoma protein (Rb), and transcription factor E2F. The simplified model involves the sequential activation of CDKs (reviewed in (33, 34)). There are two classes of CDKs involved in the G1 to S transition, CDK4/CDK 6 and CDK2, whose activities require their association with specific cyclin partners (cyclin D for CDK4/6 and cyclin E or cyclin A for CDK2). The inhibition of CDK activity by cyclin-dependent kinase inhibitors (CDKIs) constitutes one of the key mechanisms of the negative regulation of cell cycle progression. There are two families of CDKIs, namely the INK4 (inhibitors of CDK4 and CDK6) family and the Cl P/KIP (cyclin-dependent kinase-jnteracting proteins/kinase inhibitor grotein) family (reviewed in (35)). P21 is a member of the CIP/KIP family. It is best known as the tumor suppressor p53-target molecule. It is transcriptionally regulated by p53 and is essential in p53-meidated cell cycle arrest in response to DNA damage and cellular stress (36, 37). Later studies found that its direct inducers also include TGF-B (38), Vitamin D (39) and androgen (32), which have all been shown to have inhibiting effects on cell-cycle progression. However, accumulating new evidence has suggested that the role of p21 and other CIP/KIP family members is very complex in proliferating cells, since they are also essential components of active CDK4/6-cyclin D complexes. This apparent paradox arises from their dual functionality (acting as Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. assembly factors necessary for the association and catalytic activity of D-type cyclins with CDK4/6 at lower concentration and as CDK inhibitors at higher concentrations), which is the stoichiometric effects of p21 observed by Zhang et al (40) and LaBaer et al (41). 2.5 p21 in breast cancer: Despite the known roles of p21 as G1 blocker and very potent inhibitor of cell proliferation in various cell culture models (17, 36), studies on its possible role in breast carcinogenesis have not reached consensus. In one study, high p21 expression was observed to predict better response to adjuvant hormonal treatment with anti-estrogens (42), in accordance with the study which found that p21 was responsible for the inhibition of MCF-7 human breast cancer cell proliferation by anti-estrogens (43). However, as reviewed by Fernandez et al (44) and by Tsihlias et al (45), the expected negative association between p21 expression levels and cell proliferation rate was not observed in breast tumors. Moreover, both low and high levels of this protein have been correlated with higher histological grade (poorly differentiated tumors) and/or shortened disease-free survival for breast cancer patients. The studies on p21 knockout animals or cell lines also failed to provide a clear picture of the role of p21 in tumorigenesis. It was observed that the mouse p21 -/- embryonic fibroblasts exhibit an enhanced growth rate and higher saturation density in cell culture. In addition, these p21 -/- cells were significantly deficient in their ability to arrest in G1 in response to DNA damage (46, 47), supporting the importance of p21 in mediating cell cycle control. However, p21 -/- mice did not have increased rate of tumorigenesis compared to normal mice (47). Later studies suggested that with the presence of other oncogenic stimuli (e.g. activated ras), p21 knockout or deficiency could result in an increased tumor formation and accelerated tumor growth (48, 49). Very interestingly, in a study by Jones et al, among transgenic mice carrying a Wnt-1 oncogene, it was those that were also heterozygotes with p21-deficiency (p21+/-) that showed the fastest growth of mammary tumors, compared to p21+/+ and p21 -/- mic, despite of similar mammary tumor incidence among these three groups. It was postulated by the authors that the p21 stoichiometric effects might be the explanation. The authors also suggested that in certain in vivo contexts, reduced p21 dosage can promote tumor progression. Moreover, low to moderate levels of p21, but not its absence, may have growth stimulatory effects in mammary tumors (50). 2.6 p21 gene and studies on p21 gene polymorphisms: El-Deiry et al in 1993 (36) mapped the p21 gene to 6p21.2. The gene spans a region of 8.62 kb and has 3 exons, encoding a protein with 165 amino acids. Because of the importance of p21 in regulating cell cycle and its possible role as a tumor suppressor, there have been quite a number of studies seeking mutations in this gene in human tumors. Somatic mutations of p21 were either not found or occurred at very low frequencies in breast tumors (less than 3%), as summarized in (51), suggesting that cell cycle dysregulation caused by mutations in the p21 gene may not contribute to breast cancer initiation and progression. However, polymorphisms in p21 have been well documented in the literature and may act as predisposing factor for cancer 218 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. development in the presence of mutations in other genes (for example, oncogenes or tumor suppressor genes) or environmental risk factors (which may result in DNA damage). There are three common polymorphisms in the p21 gene reported in the literature. They are all single nucleotide polymorphisms (SNPs). One of them is a C to A transversion at codon 31 (AGC to AGA), resulting in a change of amino acid from a serine to an arginine. The reported frequencies of the arginine allele are from 10% to 14% in Caucasians, 24% in Hispanics, and 26% to 29% in African- Americans (52-54). The second polymorphism is a C to T transition 20 basepairs (bps) downstream of the stop codon in the 3’ untranslated region (3’URT), which was shown to be in complete linkage disequilibrium (LD) with the codon 31 polymorphism in two studies conducted in Canada and in the U.S., respectively (55) (56). The third polymorphism is a C to G substitution in intron 2,17 bps downstream of the splice donor site. In a study from Australia by Powell et al, 36% of 143 individuals without cancer (mostly Caucasians) carried the minor allele of this polymorphism (51). In addition, a codon 149 polymorphism (GAT->GGT, with aspartate changed to glycine) was observed in 8 of 50 normal individuals in a study from India (57) and a codon 84 polymorphism (a CGG->CAG substitution, with arginine changed to glutamine) was carried by 2 of 10 patients (in genomic DNA) with locally advanced breast cancer (58). But the fact that these last two polymorphisms were not reported in other studies which screened the entire coding region of the gene suggests that they only exist in certain ethnic groups or are very rare in normal populations. The functional significance of the codon 31 SNP is not clear. In a study of this polymorphism and p21 protein expression in several types of tumor tissues, it was found that unlike endometrial carcinomas in which the arginine allele was associated with an increased protein expression level, there was no correlation between the presence of this allele with p21 protein expression in breast tumor tissues (59). In an in vitro transfection study conducted in a lung adenocarcinoma cell line H1299, it was found that the arginine allele of this polymorphism showed similar functional activity as a tumor suppressor to the wild-type serine allele (52). Although there has been no functional study of the intron 2 polymorphism, it was postulated that, given its close proximity to the 5’ splice donor site of intron 2, it might be important in mRNA splicing (51). The codon 31 SNP has been associated with risk of a few cancer types (e.g. lung, prostate, endometrial and cervical cancers) (56, 60-62). It was first linked to breast cancer in a study conducted in Canada by Mousses et al. When comparing 100 breast tumor samples to 103 blood samples from normal individuals, it was found that the frequency of the arginine allele was significantly higher in the breast tumors without p53 mutations (but not those with p53 mutations) (55). In a later study in the US of 160 breast cancer cases and 327 controls from three ethnic groups, there was some suggestion that the minor allele of the codon 31 SNP was associated with an increased risk of breast cancer among African-Americans and Latinas, but not among Caucasians, although the test statistics did not reach 219 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. significance level in any of the ethnic groups (54). Very interestingly, the minor allele in this study was the serine allele, which was the opposite one to what has been reported in all other studies listed above. It is likely to be an error. It is also noteworthy that the controls in this study were hospital-based controls with more than 50% of them with breast diseases. If the polymorphism of the p21 gene also plays a role in other breast diseases, the study results may be biased toward the null with the inclusion of an abnormally high percentage of breast disease patients in the control group. In addition, the high genotyping failure rate (9.2%) and the departure from Hardy-Weinberg equilibrium of the genotype frequencies in Caucasians are also limitations of the study. Another study of 286 breast cancer patients and 81 controls conducted in the Netherlands failed to find any significant association between this codon 31 SNP and breast cancer risk. However, there was a weak but significant association between the polymorphism in intron 2 of the gene and the risk of breast cancer (51). In summary, although some studies have observed associations between the polymorphisms in the p21 gene and breast cancer risk, there is a lack of well-conducted studies systematically looking at the polymorphisms in this gene in breast cancer. Additional large studies with diverse ethnicities are needed to confirm these associations. 2.7 p21 and BRCA1: There is evidence suggesting possible interactions between the p21 gene and the BRCA1 gene in cell cycle control. The p21 gene can be activated by BRCA1 indirectly or directly: 1) BRCA1 binds to p53 resulting in the indirect activation of p21; 2) BRCA1 also binds to the ZBRK1 binding sequence motif (GGGxxxCAGxxxTTT) present in the promoter regions of the p21 gene and activate its transcription (reviewed in (63)). The activation of the p21 gene may play an role in mediating the BRCA1-induced cell cycle arrest, because it was observed that the arrest in G1 phase induced by the overexpression of BRCA1 is dependent on the presence of p21 (64, 65). Furthermore, AR and BRCA1 can act synergistically to transcriptionally activate the p21 promoter, as observed in MCF-7 cell lines (29). All these complex connections between AR, p21 and BRCA1 make it intriguing to study the AR gene and the p21 gene in breast cancer within BRCA1 mutation carriers. 3. PRELIMINARY RESULTS We recently finished a case-control study of the AR CAG repeat polymorphism and breast cancer among 501 African-American women (246 cases and 255 controls), which was supported by the BCRP dissertation grant 8GB-0103. The manuscript is attached in the appendix. In brief, we found that overall, there was no significant difference in the distribution of the CAG repeat lengths between the cases and the controls (p=0.94 for short allele, p=0.64 for long allele) (In this text, we refer to the allele with shorter CAG repeats of the two alleles carried by an individual woman as the “short allele” and the longer of the two as the “long allele”). However, family history of breast cancer was a significant modifier of the genotype effect. Among women with a positive 1s t-degree family history of breast cancer, the distribution of 220 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the long allele was shifted to the right in cases, compared to controls (p<0.01). Having 1 or 2 alleles longer than 22 was associated with a significantly increased risk (OR=3.18 with 95% Cl=1.08-9.36), while no significant association was observed among women with a negative family history (OR=0.92, 95% Cl=0.62- 1.37). The interaction was statistically significant (p=0.03). 4. RESEARCH DESIGN and METHODS 4.1 Study population: The proposed study will use interview data and biospecimens collected for two existing population-based case-control studies of breast cancer. One was conducted in the San Francisco Bay Area (PI: E. John, supported by the National Cancer Institute (NCI) grants R01CA63446 and R01CA77305, and the Department of Defense (DOD) grant DAMD17-96-1-6071). The other one is a study of BRCA1, oral contraceptives and hormonal risk factors conducted in the Los Angeles Area (PI: G. Ursin, supported by NCI grant R01CA74847). 4.1.1 Study population from the San Francisco Bay Area: Eligible cases for this study were non-Hispanic white, Hispanic, or African-American women newly diagnosed with invasive breast cancer between May 1997 and April 1999, aged 35- 79, residing in the San Francisco Bay Area at the time of diagnosis, English or Spanish-speaking, and without a prior history of breast cancer. They were identified through the population-based cancer registry covering the San Francisco Bay Area, which is part of the Surveillance, Epidemiology, and End Results (SEER) program and the state-wide California Cancer Registry. Population controls, frequency- matched to cases by ethnicity and age (5-year age group), were identified through random digit dialing. Identified cases and controls were first contacted by phone to administer a brief screening to establish study eligibility, assess personal and family history of breast cancer, and verify race/ethnicity. Eligible cases (all Hispanics and African-Americans and a 10% random sample of Non-Hispanic Whites) and controls were then invited to complete an in-person interview. Trained professional bilingual/bicultural interviewers administered a structured questionnaire (in English or Spanish) which inquired about a broad array of established and suspected breast cancer risk factors, including family history, menstrual and reproductive histories. The interview was completed for 929 cases and 1046 controls. In a second home visit, participants were invited by professional phlebotomists to provide a 40 ml sample of whole blood. Those who declined the blood draw were given the option to provide a mouthwash sample. Of the 929 cases who completed the interview, 814 provided a biospecimen, including 277 (90%) non-Hispanic Whites, 287 (88%) Hispanics, and 250 (85%) African-Americans. Of the 1046 controls who completed the interview, 910 (87%) provided a biospecimen, including 298 (93%) non-Hispanic Whites, 357 (85%) Hispanics, 255 (84%) African-Americans. The subjects with a biospecimen available will be included in our proposed study. 4.1.2 Study population from the Los Angeles Area: Eligible cases for this study were non-Hispanic white, Hispanic, or African-American women newly diagnosed 221 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with invasive breast cancer between February 1998 and May 2003, aged 20-49, residing in the Los Angeles County Area at the time of diagnosis, U.S-born English- speaking, and without a prior history of breast cancer. They were identified through the Los Angeles County Cancer Surveillance Program (CSP), the Los Angeles SEER registry. This study was originally designed as a case-case study to examine genetic risk factors for breast cancer, but population controls were later added for about 25% of the cases. The controls were identified through neighborhood walks and were frequency-matched to cases by ethnicity and age (5-year age group). Controls were recruited between July 2000 and May 2003. Cases and controls were interviewed in-person using a structured questionnaire designed to obtain information on contraceptive and reproductive experiences. The cases and controls were invited by professional phlebotomists to provide a 30ml sample of whole blood. The interview was completed for 1788 cases (including 1302 non-Hispanic whites, 276 Hispanics and 210 African-Americans) and 444 controls (including 362 non- Hispanic whites, 47 Hispanics and 35 African-Americans). All cases and about 374 controls (304 non-Hispanic white, 40 Hispanics and 30 African-Americans) provided a blood sample. In addition, 1188 cases have been screened for mutations in the BRCA1 gene to date. The mutation-screening work is still on-going. It is projected that the BRCA1 mutations status will be available for about 1550 cases by July 2005. To date, Dr. Ursin’s group has identified 46 cases carrying definite disease causing mutations in the BRCA1 gene. Based on this frequency (-3.8%), we project to have about 60 mutation carriers by July 2005. 4.2 Laboratory methods 4.2.1 Polymorphism discovery in p21 gene (Specific Aim 2). In the literature, the entire coding region has been screened for mutations and polymorphisms in several studies. Three relatively common ones have been identified (codon 31, intron2 and 3’ UTR). In addition, there are five relatively common SNPs (with allele frequencies>10%) reported in the NCBI SNP database. Four are located within 1.5kb upstream of the transcription start site, one of which falls in a region that is conserved across species (evolutionarily conserved region, ECR). The last one is located within the 400 basepairs downstream of the gene. These SNPs have all been validated by frequency or genotype data. There are two ECRs in the 5’ upstream region (from -5 kb to -3.5kb and from -2.5kb to -1 kb with respect to the transcription start site) of the gene, where the putative p53 binding sites are located (36, 66, 67). There are no SNPs reported in these regions. Because of the potential functional importance of these regions, we will sequence these two regions to screen for possible common polymorphisms. We expect to detect 1-2 common polymorphisms in each of the two regions. 100 chromosomes from 50 control subjects will be chosen from four ethnic groups: non-Hispanic white, Hispanic, African-American, and Asian. Direct sequencing will be used for polymorphism detection. Six sequencing assays will be able to cover the screening of these two ECRs. 4.2.1 General quality control. All genotyping assays will include control samples that have had genotype previously confirmed by sequencing. All PCR assays 222 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. include a “water blank” to guard against contamination of the PCR reaction. To assure assay reproducibility, 5% of samples will be repeated (as the duplicates) with laboratory personnel blinded as to repeat status. 4,2.2 Genotyping. Simple sequence length polymorphism (SSLP) analysis will be used to identify the AR exon 1 CAG repeat variant. The genomic region containing the CAG repeat is PCR amplified with one fluorescently-labeled and one unlabeled primer. The resulting PCR product is run on the ABI 3700 capillary sequencer and the allele sizes are scored using the GeneMapper software. This genotyping will not be done for African-American subjects in the Northern California study because the data have been available from our previous study. Genotyping of the SNPs in the p21 gene will be performed by the Taqman assay. Two oligonucleotide probes, one specific for each allele, are labeled with different fluorophores and included during the PCR amplification. Hybridization of the reporter probe to its target sequence leads to nucleolytic cleavage of the probe during PCR amplification as a result of the 5’>3’ nuclease activity of Taq Polymerase, causing the release of the fluorescence signals. Fluorescence is detected using an ABI 7700 Taqman Sequence Detection System and the alleles are scored using ABI Sequence Detector software. 4.3 Data management and statistical analysis 4.3.1 Data management. Laboratory personnel will remain blinded to case-control status. DNA samples are labeled only by specimen ID. Genotype results and specimen ID will be entered into a spreadsheet and sent to Dr. John and to Dr. Ursin. The data managers of these two studies will merge genotype results to the database that contains case-control status and other covariates, such as family history, menstrual and reproductive history, etc. This working database will be stripped of IDs and returned to the investigator for data analysis. 4.3.2 Statistical analysis. 4.3.2.1 Pooling of data from the two studies from Northern and Southern California: There are some differences in the two parent studies of our proposed study: 1) these two studies were originally designed to focus on different risk factors (lifestyle factors in Northern California and genetic factors in Southern California); 2) the subjects in these two studies had different age structure (the southern California study only focuses on breast cancer patients younger than 50); 3) the controls in the two studies were selected differently (random-digit dialing in Northern California and neighborhood walk in Southern California), and the cases at the Southern California site were able to be ascertained sooner after diagnosis by making use of the efficient rapid case ascertainment system 3) the questionnaires used in two studies are different. Therefore, data from the two study sites will initially be analyzed separately. If odds ratios do not differ greatly by site, common odds ratio will be estimated by pooling the data and adjusting for sites in the statistical analyses. 4.3.2.2 Analysis of the main effects of genes (Specific Aim 1 and 3): Standard analyses for case-control studies will be conducted (Breslow & Day, 1980). The linear effect of AR CAG repeats will be estimated by including a continuous variable 223 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. in the logistic model. This will be done separately for the short allele (the allele with shorter CAG repeat size), the long allele, and for the sum of CAG repeats over the two alleles. The CAG repeat sizes will also be analyzed as categorical variables using the median CAG size as the cutpoint. Odds ratios will be estimated to compare genotypes short/short and short/long to the “baseline” genotype long/long. We will also try to model the effect of CAG repeats using the same outpoints as those in previous published papers in order to allow comparison. For the SNPs in the p21 gene, odds ratios (with 95% confidence intervals) will be estimated to compare the heterozygotes and the homozygotes of the minor allele(s) to the baseline group, the homozygotes of the major allele(s). A test of trend will be performed by including in the logistic model a variable coded as 0,1,2 for the number of the minor alleles. This coding scheme represents the co-dominant model. Dominant and recessive effects can be explored using other coding schemes (eg. 0,1,1 for dominant and 0,0,1 for recessive effect). Haplotype analysis will be performed for the SNPs in the p21 gene. The matching variable age will be included in all logistic models. Other factors known to be related to breast cancer, such as family history, age at menarche, parity and age at first full-term pregnancy, breast-feeding, education, BMI, oral contraceptive use, postmenopausal HRT use will be examined to see if they appear to be confounders (adding to or deleting from the model results in a change in the risk estimates greater than 10%). We will perform analyses stratified by race/ethnicity to evaluate whether associations vary by race. If the odds ratios do not differ greatly, common odds ratio will be estimated by adjusting for race in the logistic models. The other primary effect modifiers to be considered are family history and menopausal status/age. Odds ratios for “risk” genotypes will be estimated within strata defined by family history or menopausal status/age. Formal tests of effect-modification will be performed by including the appropriate interaction terms in the logistic model. 4.3.2.3 Gene-gene interaction (Secondary Specific Aim 1). Because the p21 gene is an androgen-regulated gene, certain AR/p21 genotype combinations may confer increased risk. To assess possible gene-gene interaction, odds ratios will be estimated for each AR/p21 genotype combination. Odds ratios will also be estimated for the linear effect of CAG repeat size within each of the three p21 genotype categories. 4.3.2.4 Case-case analysis (Secondary Specific Aim 2 and 3). Within the subset of cases who have known BRCA1 mutation status, we will treat the BRCA1 mutation carrying women as the ‘cases’ and the women without BRCA1 mutations as ‘controls'. We will use traditional multivariate methods to estimate the ORs for the association between BRCA1 mutation status and the genotypes of the AR or the p21 gene. Any OR above 1 will indicate that the genotypes of the AR or the p21 gene are associated with a higher relative risk of breast cancer in BRCA1 mutation carriers than in non-carriers, i.e. interaction between BRCA1 and the gene of interest (AR and p21) in breast cancer. This case-case study design assumes 224 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. independence between BRCA1 mutation status and the genotypes of interest (in the AR or p21 gene) in the population. This is a reasonable assumption, because the AR gene and the p21 gene are located on chromosome X and chromosome 6, while BRCA1 gene is located on chromosome 17, and they are segregated independently. Although the small number of mutations carriers precludes stratified- analyses by race, we will adjust for race in logistic model. 4.4 Power Calculation 4.4.1 Main effect of genotype (Specific Aim 1 and 3): The sample sizes of the Northern California Study and of the Southern California study are listed below in Table 1. Source status Non Hispanic- Whites Hispanics African- Americans Total Northern California Case 277 287 250 814 Control 298 357 255 910 Southern California Case 1302 276 210 1788 Control 304 40 30 374 combined Case 1579 563 460 2602 control 602 397 285 1284 Because of the small number of Hispanic and African-American controls in the Southern California study, we will not perform formal statistical tests in these two groups. However, the sample sizes in all other groups will provide us with excellent power to detect an odds ratio of 1.8 and above, for a minor allele frequency of 10% and higher (based on 250 cases and 250 controls, for the dominant model of genotypes) (figure not shown). Based on the proportion of subjects with a positive 1s t-degree family history in these two studies (about 15% among cases and 11% among controls), we estimate that we will have 390 cases and 141 controls with a family history in the combined dataset (assuming the association between the genotypes and breast cancer risk does not differ by race or by study sites). The power calculation is shown below (Figure 2). Under the dominant model, the proportion of subjects in the “at-risk” group is given in Table 2, for a range of allele frequencies of interest in this study (minor allele frequencies of at least 10%). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 0.7 0.5 0.4 0.3 0.2 0.2! 0.4 P ro p o rtio n ca rry in g m in o r a lle le 0.6 Figure 2: Powers for dominant model (Case=390/CO=141 Table 2. Percent of subjects carrying at least one minor allele Minor allele frequency Minor allele carriers 10% 20% 30% 40% 50% 19% 36% 51% 64% 75% For the dominant model, we will have excellent power to detect odds ratios of 2.0 and above. For a minor allele frequency of 10% (19% minor allele carrier rate), power to detect an odds ratio of 2.0 is 84%. 4.4.2 gene-gene interaction (Secondary Specific Aim 1): Our sample size is not big enough for us to perform a formal test on gene-gene interaction. 4.4.3 case-case study on interaction between BRCA1 mutation and genotypes of the AR gene and the p21 gene (Secondary Specific Aim 2 and 3): 226 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 0.9 )R = 2 .2 !=2.0 0.7 o 0.6 5 S . 0.5 0.4 0.3 0.2 0.6 0.8 1 0.2 0.4 0 p ro p o rtio n carry in g m inor allele Figure 3: Powers (case-case study, dominant model) When minor allele frequency is 10% or larger, we will have adequate power (>80%) to find ORs of approximately 2.2 for the association between the presence of BRCA1 definite-disease-causing mutations and the “at-risk” genotype among cases (power to detect an interaction odds ratio of 2.2 between the presence of BRCA1 definite-disease-causing mutations and carrying the “at-risk” genotype). 4.5 Timelines: 1-4 months: mutation screening 5-8 months: optimizing genotyping conditions 9-26 months: genotyping 27-36 months: statistical analysis, manuscript writing Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Reference: 1. Lillie, E. O., Bernstein, L. & Ursin, G. (2003) Breast Cancer Res 5 , 164-73. 2. Somboonporn, W. & Davis, S. R. (2004) Endocr Rev 25, 374-88. 3. Zumoff, B. (1994) Obstet Gynecol Clin North Am 21, 751-72. 4. Secreto, G. & Zumoff, B. (1994) Anticancer Res 14, 2113-7. 5. Somboonporn, W. & Davis, S. R. (2004) Maturitas 49, 267-75. 6. Chamberlain, N. L., Driver, E. D. & Miesfeld, R. L. (1994) Nucleic Acids Res 22,3181-6. 7. Kazemi-Esfarjani, P., Trifiro, M. A. & Pinsky, L. (1995) Hum Mol Genet 4, 523-7. 8. Tut, T. G„ Ghadessy, F. J„ Trifiro, M. A., Pinsky, L. & Yong, E. L. (1997) J Clin Endocrinol Metab 82, 3777-82. 9. Irvine, R. A., Ma, H., Yu, M. C., Ross, R. K., Stallcup, M. R. & Coetzee, G. A. (2000) Hum Mol Genet 9, 267-74. 10. Rebbeck, T. R., Kantoff, P. W., Krithivas, K., Neuhausen, S., Blackwood, M. A., Godwin, A. K., Daly, M. B., Narod, S. A., Garber, J. E., Lynch, H. T., Weber, B. L. & Brown, M. (1999)AmJHum Genet 64, 1371-7. 11. Giguere, Y., Dewailly, E., Brisson, J., Ayotte, P., Laflamme, N., Demers, A., Forest, V. I., Dodin, S., Robert, J. & Rousseau, F. (2001) Cancer Res 61, 5869-74. 12. Haiman, C. A., Brown, M., Hankinson, S. E., Spiegelman, D., Colditz, G. A., Willett, W. C„ Kantoff, P. W. & Hunter, D. J. (2002) Cancer Res 6 2 ,1045-9. 13. Suter, N. M., Malone, K. E., Daling, J. R., Doody, D. R. & Ostrander, E. A. (2003) Cancer Epidemiol Biomarkers Prev 1 2 ,127-35. 14. Liede, A., Zhang, W., De Leon Matsuda, M. L., Tan, A. & Narod, S. A. (2003) Cancer Epidemiol Biomarkers Prev 12, 848-52. 15. Spurdle, A. B., Dite, G. S., Chen, X., Mayne, C. J., Southey, M. C., Batten, L. E., Chy, H., Trute, L., McCredie, M. R., Giles, G. G., Armes, J., Venter, D. J., Hopper, J. L. & Chenevix-Trench, G. (1999) J Natl Cancer Inst 91, 961-6. 16. Dunning, A. M., McBride, S., Gregory, J., Durocher, F., Foster, N. A., Healey, C. S., Smith, N., Pharoah, P. D., Luben, R. N., Easton, D. F. & Ponder, B. A. (1999) Carcinogenesis 20, 2131-5. 17. Xiong, Y., Hannon, G. J., Zhang, H., Casso, D., Kobayashi, R. & Beach, D. (1993) Nature 366, 701-4. 18. Boccuzzi, G., Brignardello, E., Di Monaco, M., Gatto, V., Leonardi, L., Pizzini, A. & Gallo, M. (1994) B rJ Cancer 70, 1035-9. 19. Hackenberg, R. & Schulz, K. D. (1996) Journal of Steroid Biochemistry & Molecular Biology 56, 113-7. 20. Zhou, J., Ng, S., Adesanya-Famuiya, O., Anderson, K. & Bondy, C. A. (2000) Faseb J 14, 1725-30. 21. Jayo, M. J., Register, T. C., Hughes, C. L., Blas-Machado, U., Sulistiawati, E., Borgerink, H. & Johnson, C. S. (2000) J Soc Gynecol Investig 7, 257-65. 22. Dauvois, S., Geng, C. S., Levesque, C., Merand, Y. & Labrie, F. (1991) Cancer Res 51, 3131-5. 228 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23. Dauvois, S., Li, S. M., Martel, C. & Labrie, F. (1989) Breast Cancer Res Treat 14, 299-306. 24. Birrell, S. N., Bentel, J. M., Hickey, T. E., Ricciardelli, C., Weger, M. A., Horsfall, D. J. & Tilley, W. D. (1995) J Steroid Biochem Mol Biol 5 2 ,459-67. 25. Lapointe, J., Fournier, A., Richard, V. & Labrie, C. (1999) Endocrinology 140, 416-21. 26. Kadouri, L., Easton, D. F., Edwards, S., Hubert, A., Kote-Jarai, Z., Glaser, B., Durocher, F., Abeliovich, D., Peretz, T. & Eeles, R. A. (2001) B rJ Cancer 85, 36-40. 27. Olopade, O. I., Fackenthal, J. D., Dunston, G., Tainsky, M. A., Collins, F. & Whitfield-Broome, C. (2003) Cancer 97, 236-45. 28. Park, J. J., Irvine, R. A., Buchanan, G., Koh, S. S., Park, J. M., Tilley, W. D., Stallcup, M. R., Press, M. F. & Coetzee, G. A. (2000) Cancer Res 60, 5946- 9. 29. Yeh, S., Hu, Y. C., Rahman, M., Lin, H. K., Hsu, C. L., Ting, H. J., Kang, H. Y. & Chang, C. (2000) Proc Natl Acad Sci U S A 9 7 , 11256-61. 30. Lapointe, J. & Labrie, C. (2001) Endocrinology 142, 4331-8. 31. Szelei, J., Jimenez, J., Soto, A. M., Luizzi, M. F. & Sonnenschein, C. (1997) Endocrinology 138, 1406-12. 32. Lu, S., Liu, M., Epner, D. E., Tsai, S. Y. & Tsai, M. J. (1999) Mol Endocrinol 13, 376-84. 33. Sherr, C. J. (1994) Cell 79, 551-5. 34. Massague, J. (2004) Nature 432, 298-306. 35. Sherr, C. J. & Roberts, J. M. (1995) Genes Dev 9 , 1149-63. 36. el-Deiry, W. S., Tokino, T., Velculescu, V. E., Levy, D. B., Parsons, R., Trent, J. M., Lin, D., Mercer, W. E., Kinzler, K. W. & Vogelstein, B. (1993) C ell75, 817-25. 37. el-Deiry, W. S., Harper, J. W., O'Connor, P. M., Velculescu, V. E., Canman, C. E., Jackman, J., Pietenpol, J. A., Burrell, M., Hill, D. E., Wang, Y. & et al. (1994) Cancer Res 54, 1169-74. 38. Datto, M. B„ Yu, Y. & Wang, X. F. (1995) J Biol Chem 270, 28623-8. 39. Liu, M., Lee, M. H., Cohen, M., Bommakanti, M. & Freedman, L. P. (1996) Genes Dev 10, 142-53. 40. Zhang, H., Hannon, G. J. & Beach, D. (1994) Genes Dev 8, 1750-8. 41. LaBaer, J., Garrett, M. D., Stevenson, L. F., Slingerland, J. M., Sandhu, C., Chou, H. S., Fattaey, A. & Harlow, E. (1997) Genes Dev 11, 847-62. 42. Pellikainen, M. J., Pekola, T. T., Ropponen, K. M., Kataja, V. V., Kellokoski, J. K., Eskelinen, M. J. & Kosma, V. M. (2003) J Clin Pathol 5 6 ,214-20. 43. Skildum, A. J., Mukherjee, S. & Conrad, S. E. (2002) J Biol Chem 277, 5145- 52. 44. Fernandez, P. L., Jares, P., Rey, M. J., Campo, E. & Cardesa, A. (1998) Mol Pathol 51,305-9. 45. Tsihlias, J., Kapusta, L. & Slingerland, J. (1999) Annu Rev Med 50, 401-23. 46. Brugarolas, J., Chandrasekaran, C., Gordon, J. I., Beach, D., Jacks, T. & Hannon, G. J. (1995) Nature 377, 552-7. 229 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47. Deng, C., Zhang, P., Harper, J. W „ Elledge, S. J. & Leder, P. (1995) Cell 82, 675-84. 48. Missero, C., Di Cunto, F., Kiyokawa, H., Koff, A. & Dotto, G. P. (1996) Genes Dev 10, 3065-75. 49. Bearss, D. J., Lee, R. J., Troyer, D. A., Pestell, R. G. & Windle, J. J. (2002) Cancer Res 62, 2077-84. 50. Jones, J. M., Cui, X. S., Medina, D. & Donehower, L. A. (1999) Cell Growth Differ 10,213-22. 51. Powell, B. L., van Staveren, I. L., Roosken, P., Grieu, F., Berns, E. M. & lacopetta, B. (2002) Carcinogenesis 23, 311-5. 52. Chedid, M., Michieli, P., Lengel, C., Huppi, K. & Givol, D. (1994) Oncogene 9, 3021-4. 53. Birgander, R., Sjalander, A., Saha, N., Spitsyn, V., Beckman, L. & Beckman, G. (1996) Hum Hered 46, 148-54. 54. Keshava, C., Frye, B. L., Wolff, M. S., McCanlies, E. C. & Weston, A. (2002) Cancer Epidemiol Biomarkers Prev 11,127-30. 55. Mousses, S., Ozcelik, H., Lee, P. D., Malkin, D., Bull, S. B. & Andrulis, I. L. (1995) Hum Mol Genet 4, 1089-92. 56. Facher, E. A., Becich, M. J., Deka, A. & Law, J. C. (1997) Cancer 79, 2424- 9. 57. Ralhan, R., Agarwal, S., Mathur, M., Wasylyk, B. & Srivastava, A. (2000) Clin Cancer Res 6, 2440-7. 58. Staalesen, V., Leirvaag, B., Lillehaug, J. R. & Lonning, P. E. (2004) Clin Cancer Res 10, 3438-43. 59. Lukas, J., Groshen, S., Saffari, B., Niu, N., Reles, A., Wen, W. H., Felix, J., Jones, L. A., Hall, F. L. & Press, M. F. (1997) Am J Pathol 150, 167-75. 60. Sjalander, A., Birgander, R., Rannug, A., Alexandrie, A. K., Tornling, G. & Beckman, G. (1996) Hum Hered 46, 221-5. 61. Roh, J. W„ Kim, J. W„ Park, N. H., Song, Y. S., Park, I. A., Park, S. Y., Kang, S. B. & Lee, H. P. (2004) Gynecol Oncol 93, 499-505. 62. Roh, J., Kim, M., Kim, J., Park, N., Song, Y., Kang, S. & Lee, H. (2001) Cancer Lett 165, 59-62. 63. Somasundaram, K. (2003) J Cell Biochem 88, 1084-91. 64. MacLachlan, T. K., Somasundaram, K., Sgagias, M., Shifman, Y., Muschel, R. J., Cowan, K. H. & El-Deiry, W. S. (2000) J Biol Chem 275, 2777-85. 65. MacLachlan, T. K. & El-Deiry, W. S. (2000) Nat Med 6, 1318-9. 66. el-Deiry, W. S., Kern, S. E., Pietenpol, J. A., Kinzler, K. W. & Vogelstein, B. (1992) Nat Genet 1,45-9. 67. Nozell, S. & Chen, X. (2002) Oncogene 21,1285-94. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Dietary fats, fat metabolizing genes, and the risk of breast cancer

PDF

A large-scale genetic association study of prostate cancer in a multi-ethnic population

PDF

Association of vitamin D receptor gene polymorphisms with colorectal adenoma

PDF

Androgens and breast cancer

PDF

Genetic risk factors in breast cancer susceptibility: The multiethnic cohort

PDF

Breast cancer in the multiethnic cohort study: Genetic (prolactin pathway genes) and environmental (hormone therapy) factors

PDF

Determinants of mammographic density in African-American, non-Hispanic white and Hispanic white women before and after the diagnosis with breast cancer

PDF

BRCA1 mutations and polymorphisms in African American women with a family history of breast cancer identified through high throughput sequencing

PDF

Dietary fiber intake and atherosclerosis progression: The Los Angeles Atherosclerosis Study

PDF

CYP17 polymorphism and risk for colorectal adenomas

PDF

Colorectal cancer risks in Singapore Chinese: Polymorphisms in the insulin-like growth factor-1 and the vitamin D receptor

PDF

A linear model for measurement errors in oligonucleotide microarray experiment

PDF

An intervention and program evaluation to determine the effectiveness of public health reforms on primary prevention practices by chiropractic interns

PDF

Biopsychosocial factors in major depressive disorder

PDF

Association between body mass and benign prostatic hyperplasia in Hispanics: Role of steroid 5-alpha reductase type 2 (SRD5A2) gene

PDF

Evaluation of the accuracy and reliability of self-reported breast, cervical, and ovarian cancer incidence in a large population-based cohort of native California twins

PDF

Assessment of fatigue as a late effect of therapy among survivors of childhood leukemia

PDF

A case/parental/sibling control study of Ewing's sarcoma/peripheral primitive neuroectodermal tumor (pPNET)

PDF

A prospective cohort study of morbidity and mortality among middle-aged men in Shanghai, China

PDF

Descriptive epidemiology of thyroid cancer in Los Angeles County, 1972-1995

Asset Metadata

Creator Wang, Wei (author)

Core Title Androgen receptor gene and prostate -specific antigen gene in breast cancer

School Graduate School

Degree Doctor of Philosophy

Degree Program Epidemiology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag health sciences, public health,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Ingles, Sue Ann (committee chair), Coetzee, Gerhard (committee member), Ross, Ronald (committee member), Yu, Mimi (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-618264

Unique identifier UC11336424

Identifier 3220164.pdf (filename),usctheses-c16-618264 (legacy record id)

Legacy Identifier 3220164.pdf

Dmrecord 618264

Document Type Dissertation

Rights Wang, Wei

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA