University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content BREAST CANCER IN THE MULTIETHNIC COHORT STUDY: GENETIC (PROLACTIN PATHWAY GENES) AND ENVIRONMENTAL (HORMONE THERAPY) FACTORS Copyright 2005 by Sulggi Lee 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) May 2005 Sulggi Lee Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3180308 Copyright 2005 by Lee, Sulggi All rights reserved. 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 3180308 Copyright 2005 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. ACKNOWLEDGEMENTS I would like to acknowledge the committee members of this dissertation who provided me with the guidance, insight, and support to conduct this research. Specifically, I thank Brian Henderson, Malcolm Pike, Christopher Haiman, Roberta McKean-Cowdin, Giske Ursin, and Gerhard Coetzee who have given generously of their time and thought into the development and review of this research. I thank my father who came to this country in the hope of earning a Ph.D. degree and sacrificed this pursuit to support a family, and my mother, whose unwavering faith in me encourages me to strive towards higher goals, for being my greatest sources of encouragement. I thank my siblings, Gowan, Ajean, and Earl for a lifetime of insight and advice. I am grateful for the encouragement of my friends, especially Nick, for his constant support and understanding of the effort required to complete this research and earn my doctorate. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS List of Tables vi List of Figures viii Abstract ix Preface xi Chapter 1 California Breast Cancer Research Program Grant Proposal: 1 Prolactin and Breast Cancer Risk in the Multiethnic Cohort Section 1.1 Introduction 2 Section 1.2 Background and Significance 3 1.2.1 Prolactin 3 1.2.2 Prolactin Receptor 4 1.2.3 Prolactin and Breast Cancer 5 1.2.4 Haplotype Analysis 8 Section 1.3 Preliminary Results 9 Section 1.4 Research Design and Methods 10 1.4.1 Subjects 10 1.4.2 Laboratory Methods 14 1.4.3 Statistical Analyses 17 1.4.4 Sample Size and Power Estimation 19 Section 1.5 Timeline 22 Section 1.6 References 23 Chapter 2 Haplotype-Based Analysis of Prolactin Pathway Genes in 29 Relation to Plasma Prolactin Levels and Breast Cancer Risk in the Multiethnic Cohort Section 2.1 Introduction 30 Section 2.2 Methods 32 2.2.1 Subjects 32 2.2.2 SNP Selection and Haplotype Block Determination 34 2.2.3 Haplotype Estimation and htSNP Selection 36 2.2.4 Comparison of Haplotype Frequencies among Breast Cancer 38 Cases and Controls iii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2.5 Plasma Prolactin Level Assay 40 Section 2.3 Results 42 2.3.1 LD Blocks and Haplotype Structure in the Multiethnic Panel 42 2.3.2 Breast Cancer Case-Control Analysis 48 2.3.3 Plasma Prolactin Level Analysis 56 2.3.4 Single SNP Analyses 68 Section 2.4 Discussion 68 Section 2.5 References 83 Chapter 3 An Overview of Postmenopausal Estrogen-Progestin 89 Hormone Therapy and Breast Cancer Risk Section 3.1 Introduction 90 Section 3.2 Methods 91 Section 3.3 Results 95 3.3.1 EPT and Breast Cancer Risk 96 3.3.2 EPT and Lobular versus Ductal Breast Cancer Risk 96 3.3.3 Sequential versus Continuous-Combined EPT and Breast 96 Cancer Risk 3.3.4 Current/Recent Use versus Total Lifetime Use and Breast 101 Cancer Risk Section 3.4 Discussion 103 3.4.1 EPT and Breast Cancer Risk 103 3.4.2 EPT and Lobular versus Ductal Breast Cancer Risk 105 3.4.3 Sequential versus Continuous-Combined EPT and Breast 106 Cancer Risk 3.4.4 Current/Recent Use versus Total Lifetime Use and Breast 108 Cancer Risk 3.4.5 Strengths and Limitations 109 3.4.6 Summary 110 Section 3.5 References 112 Chapter 4 Hormone Therapy and Breast Cancer Risk in the Multiethnic 116 Cohort Section 4.1 Introduction 117 iv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Section 4.2 Methods 118 4.2.1 Study Population 118 4.2.2 Surveillance 119 4.2.3 Statistical Methods 120 Section 4.3 Results 123 Section 4.4 Discussion 131 Section 4.5 References 137 Comprehensive Bibliography 140 Appendices 152 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES Chapter 2 Page 1 Descriptive characteristics among breast cancer cases and 49 controls in the Multiethnic Cohort Study 2 Common haplotypes (>5%) in blocks 1-4 of prolactin (PRL), 50 estimated using all SNPs and the htSNPs among all groups combined 3 Common haplotypes (>5%) in blocks 1 -6 of prolactin receptor 51 (PRLR), estimated using all SNPs and the htSNPs among all groups combined 4 Common haplotypes in blocks 1-4 of prolactin (PRL) among 52 African-Americans, Hawaiians, Japanese, Latinas, and Whites in the multiethnic panel 5 Common haplotypes in blocks 1-6 of prolactin receptor 54 (PRLR) among African-Americans, Hawaiians, Japanese, Latinas, and Whites in the multiethnic panel 6 Association between haplotypes in LD blocks 1-4 of prolactin 57 (PRL) and breast cancer risk 7 Association between haplotypes in LD blocks 1-6 of prolactin 61 receptor (PRLR) and breast cancer risk 8 Least squares means of plasma prolactin (PRL) levels by 67 ethnicity among postmenopausal women without breast cancer and not using hormone therapy 9 Association between haplotype dosage estimate in LD blocks 69 1-4 of prolactin (PRL) and plasma PRL levels 10 Association between haplotype dosage estimate in LD blocks 73 1-6 of prolactin receptor (PRLR) and plasma PRL levels 11a Association between htSNPs in region of low LD in PRL and 77 breast cancer risk lib Association between htSNPs in region of low LD of PRL and 77 plasma PRL levels VI Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12a Association between missense SNP IlelOOLeu (exon5) in 78 prolactin receptor (PRLR) and breast cancer risk 12b Association between missense SNP IlelOOLeu (exon5) in 78 prolactin receptor (PRLR) and plasma PRL levels Chapter 3 1 Odds Ratios per Year of Use (ORi s) of Estrogen-Progestin 97 Therapy and Breast Cancer Risk 2 Odds Ratios per Year of Use (ORi s) of Estrogen-Progestin 99 Therapy and Breast Cancer Risk by Histologic Subtype 3 Odds Ratios per year of use (ORis) of estrogen-progestin 100 therapy and breast cancer risk by progestin schedule 4 Odds Ratios per year of use (ORis) of estrogen-progestin 102 therapy and breast cancer risk by recency of use Chapter 4 1 Descriptive characteristics of women included in the study 125 2 Crude and adjusted analyses of hormone therapy (HT) 126 duration and breast cancer risk in the total population 3 Hazard Ratios (HRs) of breast cancer risk and hormone 128 therapy (HT) per 5 years of use by ethnicity 4 Hazard Ratios (HRs) of breast cancer risk and duration of 130 hormone therapy (HT) per 5 years of use, stratified by weight, stage of disease, and histologic subtype 5 Hazard Ratios (HRs) of breast cancer risk and duration of 132 hormone therapy (HT) per 5 years of use, stratified by weight using a common reference group 6 Hazard ratios (HRs) of breast cancer risk and duration of 133 hormone therapy (HT) per 5 years of use, stratified by estrogen receptor (ER) and progesterone receptor (PR) status vii Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES Chapter 2 Page 1 The genomic organization of prolactin (PRL) 43 2 The genomic organization of prolactin receptor (PRLR) 44 3 The linkage disequilibrium plot of PRL for all ethnic groups combined 45 4 Chapter 3 The linkage disequilibrium plot of PRLR for all ethnic groups combined 47 1 Studies Included in Overall Analysis of EPT and Risk of Breast Cancer: Odds Ratios with 95% Confidence Intervals per Year of Use 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT Both genetic and environmental factors may contribute to the risk of breast cancer. In this dissertation, using a novel haplotype-based approach, common genetic variation in the prolactin (PRL) and prolactin receptor (PRLR) genes is assessed in relation to plasma PRL levels and breast cancer risk among African- American, Hawaiian, Japanese, Latina, and White women in the Multiethnic Cohort Study (MEC). Animal studies and human in vitro studies suggest that PRL is involved in breast development and tumorigenesis, and the largest prospective cohort study of postmenopausal women found a 34% increase in risk of breast cancer when comparing women among the top to bottom quartiles of plasma PRL levels. However, no study has yet evaluated the possible role of common genetic variation in PRL pathway genes and breast cancer risk, or their potential contribution in determining circulating PRL levels. This dissertation also examines the role of postmenopausal hormone therapy (HT), a widely prescribed treatment for menopausal symptoms, and its association with breast cancer risk. Both observational studies and results from the Women’s Health Initiative trial indicate that HT use, specifically combined estrogen-progestin therapy (EPT) use, is associated with an overall increase in breast cancer risk. However, further data are needed on whether this association varies by factors such as stage of disease, weight, histologic subtype, schedule of progestin administration, and hormone receptor status, and no study has yet evaluated ethnic-specific differences in risk associated with HT use. The role of ix Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. HT use and breast cancer risk is evaluated, first, in a meta-analysis of the existing published literature on EPT use and breast cancer risk and second, in a prospective cohort study of HT use and breast cancer risk among postmenopausal women in MEC. These two areas of research, haplotype-based analyses of disease susceptibility genes and studies of HT use and breast cancer risk, address current issues in breast cancer research and contribute to a greater understanding of the complex etiology of breast cancer. By evaluating these factors among a multiethnic population, we address whether the association between these genetic and lifestyle factors and breast cancer risk vary by race/ethnicity. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. PREFACE The Hawaii-Los Angeles Multiethnic Cohort (MEC) Study is an ongoing, prospective study of environmental and genetic factors in relation to cancer and other chronic diseases among 215,251 African-American, Japanese, Hawaiian, Latino, and White men and women, as well as small numbers of individuals from other racial/ethnic groups. This dissertation evaluates both genetic and environmental factors related to breast cancer risk in the MEC. Chapter 1 is the research proposal for the haplotype-based study of prolactin (PRL) pathway genes and breast cancer risk in the MEC, which was funded by the California Breast Cancer Research Program (IDEA grant #9IB-0034). Preliminary results were published in the Proceedings of the American Association for Cancer Research 2004; 45:182 and presented at the American Association for Cancer Research Annual Conference, Orlando, Florida, 2004. Chapter 2 presents results from this study. Chapter 3 reviews the literature on estrogen-progestin therapy (EPT) and breast cancer risk and includes findings from all published observational studies as well as the results from the Women’s Health Initiative randomized clinical trial. Separate assessments by histologic subtype of breast cancer as well as by schedule of the progestin component of EPT are included. Chapter 4 is a prospective cohort study of postmenopausal hormone therapy (HT) and breast cancer risk in the MEC with subanalyses to assess effect modification by weight, stage of disease, histologic subtype, and hormone receptor status. xi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 1 CALIFORNIA BREAST CANCER RESEARCH PROGRAM GRANT PROPOSAL: PROLACTIN AND BREAST CANCER RISK IN THE MULTIETHNIC COHORT Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LI INTRODUCTION Hormones, of ovarian and pituitary origin, have been associated with breast cancer risk because of their ability to drive breast cell proliferation. Prolactin (PRL), while less extensively studied than steroid hormones (e.g. estradiol, estrone, progesterone) as a breast cancer risk factor, is an important regulator of mammary development during puberty and pregnancy. Experimental studies have clearly indicated a role for PRL in breast epithelial proliferation, differentiation, and 1 7 tumorigenesis, ' Studies have also shown that women with greater circulating PRL levels are at increased risk of breast cancer, suggesting that variability in PRL may also be important in determining a woman’s risk.842 We hypothesize that genetic variation in PRL and prolactin receptor (PRLR) may contribute to inter-individual differences in circulating prolactin levels and breast cancer risk. In this study we propose to comprehensively assess common genetic variation across the PRL and PRLR loci using a novel haplotype- based approach among African-American, Japanese, White, Latina, and Hawaiian women in the Multiethnic Cohort Study. We will examine common variation in these genes as predictors of plasma PRL levels and breast cancer risk in these racial/ethnic groups. Missense and putative functional variants in regulatory regions will also be assessed in relationship with circulating PRL levels and breast cancer risk. Specifically, we propose to: 1. comprehensively characterize the haplotype structure of PRL and PRLR among African-American, Japanese, White, Latina, and Hawaiian women in the MEC, 2. select a subset of SNPs that uniquely identify 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. each of the common haplotypes (haplotype “tag” SNPs, htSNPs), 3. examine the association between common haplotypes and breast cancer risk in a large multiethnic case-control study within the MEC, and 4. test the association between circulating plasma prolactin levels and common haplotypes in controls from 5 racial/ethnic groups (100 from each ethnic group). 1.2 BACKGROUND AND SIGNIFICANCE 1.2.1 Prolactin Mammary gland development occurs during puberty (ductal development) and pregnancy (ductal and alveolar development). Prolactin has been shown to be critical in the control of breast development at all stages. However, the wide variety of hormones that play a role in mammary gland development during both puberty and pregnancy make it difficult to discern the relative importance of any one factor. Rather, it is thought that these factors act synergistically to exert their effects on the breast.13,1 4 Early findings from rat and goat models show that estrogen’s ability to stimulate ductal or other mammary growth depends on the presence of PRL.13,1 5 Similarly, PRL’s ability to stimulate mammary cell proliferation and lobuloalveolar growth in rats relies on the presence of estrogen and progesterone, respectively.1 6 Estrogen is thought to regulate the number of PRLRs in mammary epithelium. Human and rat experiments show increased expression of prolactin receptors in the breast during pregnancy and after parturition,1 7 '1 9 while rat models indicate decreased numbers of PRLRs after estrogen treatment.2 0 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In humans, the PRL gene lies on chromosome 6 and is approximately 10 kilobases (kb) in length with five coding exons.2 1 Recently, an additional non coding first exon has been described that lies about 5.8 kb upstream of the pituitary promoter site.2 2 This distal promoter region has been associated with extra- pituitary expression of PRL.2 3 1.2.2 Prolactin receptor Prolactin receptor belongs to the cytokine hematopoietic superfamily of receptors and is a single transmembrane receptor.2 4 Prolactin receptors utilize a variety of kinases since they lack intrinsic kinase activity. The most well-known of these kinase pathways is the JAK2/STAT (Janus kinase and Signal Transducer and Transactivator protein) pathway. Ligand binding induces dimerization of PRLR’s extracellular domain, leading to transphosphorylation of constitutively associated JAK2 proteins. ST AT proteins are then recruited to the receptor, phosphorylated by JAK proteins, dimerize, and translocate into the nucleus to activate gene transcription via binding to gamma-interferon activation sites (GAS).2 5 Signal Transducer and Transactivator 5a (STAT5a), is thought to be the primary STAT protein of the JAK/STAT PRLR signaling pathway to be activated by PRL. Initial animal studies using STATSa knockout mice show that STATSa deficient mice exhibit incomplete mammopoeisis and failure of lactopoiesis.2 6 In vitro studies using human breast cancer cell lines demonstrate that STATSa is the main transcription factor involved in mammary cell differentiation.2 7 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The human PRLR gene is located on chromosome 5 and is approximately 180 kb in length with at least 10 exons and 9 introns. Recently, at least six alternative non-coding first exons have been described whose functions are yet unknown.29,3 0 1.2.3 Prolactin and breast cancer 1.2.3.1 Animal studies Animal studies have largely supported prolactin’s role in mammary development. Prolactin has been shown to be critical for alveolar proliferation and differentiation involving milk protein synthesis during pregnancy.7,3 1 A recent study also indicates that PRL may control ductal side branching and terminal end bud regression in non-pregnant mice.3 The most important finding from animal studies however, is the strong evidence for an effect of PRL on mammary cell mitogenesis. Mice transplanted with multiple pituitary isografts over-secreting PRL,4 given daily injections of PRL,1 or administered drugs to induce hyperprolactinemia5,6 all demonstrated increased incidence of spontaneous mammary tumors 1.2.3.2 In vivo and in vitro studies Results from human in vitro studies have demonstrated a clear association between PRL and mammary cell response. PRLR expression has been observed in over 70% of all human breast cancers studied over the past two decades.3 2 '3 5 Previous studies have shown that PRL administration stimulates increased DNA 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. synthesis,3 6 '3 9 alpha-lactalbumin synthesis,4 0 and estrogen receptor content4 1 Physiological levels of human PRL (hPRL) have been found to increase population AO doubling of primary breast cancer cells in culture. Consistently higher levels of PRLR mRNA were observed in malignant breast biopsy tissue than in normal contiguous tissue, using quantitative polymerase chain reaction (PCR). These results suggest that tumor cells may be more responsive to the stimulatory effect of PRL (pituitary or locally produced) than are normal cells. Clinical evidence supports in vitro findings. Case studies of patients with pituitary prolactinomas who develop breast cancer in women4 3 '4 5 and in men4 6 " 4 8 have been reported. Hyperprolactinemia was found to be a predictor of breast cancer progression and poor prognosis in one large, prospective study.4 9 Another study found that circadian rhythms of PRL secretion varied between high risk and low risk breast cancer groups.5 0 1.2.3.3 Epidemiologic studies There have been few epidemiologic studies investigating the direct relationship between plasma PRL level measurements and breast cancer risk. Comparisons between studies of plasma PRL and breast cancer risk are difficult due to variability in study design, sample size, study population, control group selection, menopausal status, and control of important confounding factors (time of specimen collection, fasting status, physical and mental stress level, other plasma hormone levels, other medications, lactation history, age at first full-term pregnancy, parity). In addition, plasma PRL levels follow a diurnal pattern, with 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the peak levels during the early morning hours.1 4 Different laboratory methods, such as use of standard radioimmunoassay versus bioassay with anti-lactogen antibodies, may also affect measurement. Previously documented findings indicate that storage and transport ofbiospecimens can also affect plasma PRL measurement. Many of the investigations looking at plasma PRL and breast cancer risk have been small, case-control studies that used retrospective blood samples from cases after diagnosis, making the results difficult to interpret due to possible confounding from known affects from the disease process and/or treatment.8 Additionally, a number of the studies were conducted among premenopausal women, among whom plasma PRL levels may have varied considerably depending on day of menstrual cycle.9,10,5 1 An early case-control study by Henderson and colleagues, found that premenopausal daughters of breast cancer patients had higher PRL levels than daughters of controls (p<0.05).5 1 Similarly, other studies among premenopausal9,1 0 and postmenopausal women11,1 2 observed an association between breast cancer and elevated plasma prolactin levels.9,1 0 Q S9 S3 However, other studies found no association. ’ ’ To date, there have only been two published prospective studies of plasma prolactin and breast cancer risk.8,5 4 Prospective studies address many of the issues regarding accurate PRL measurements. An earlier study by Wang and colleagues found that among 40 postmenopausal breast cancer cases, there was a 63% increased risk of breast in the top quintile of PRL levels compared to those in the bottom quintile.5 4 Though the results from this study were non-significant and 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. based on a small sample size, they are comparable to results observed in a more 8 • • • recent prospective study by Hankinson and colleagues. This recent investigation found that among 306 postmenopausal breast cancer cases and 448 matched controls, there was greater than a two-fold increase in breast cancer risk among women in the highest quartile of plasma PRL levels compared to the lowest quartile (odds ratio = 2.03, 95% confidence interval = 1.24 - 3.31). Results from this large- scale, carefully designed study provide perhaps the most convincing evidence on plasma PRL levels and breast cancer risk to date. 1.2.4 Haplotype analysis 1.2.4.1 Common diseases associated with common variants It is estimated that two randomly chosen copies of the human genome will vary no more than one every 1,250 nucleotides. Therefore, most of human genetic variation is largely attributable to a small number of common variants.5 5 This is the underlying basis for the “common variant-common disease” (CD/CV) hypothesis, the idea that genetic risk for common disease is due to common variants. Common SNPs are thought to cluster onto a more limited number of ancestral segment patterns. Assessing these ancesteral segments, or “haplotypes,” is a solution to an otherwise impractical feat of discovering and testing each of the estimated 10,000,000 common variants in the human genome for their association with disease.5 6 Recent results indicate that nearby SNPs are strongly correlated with other SNPs in the gene and may be inherited, without mutation or recombination, through generations. By identifying these ancestral segments of a 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. gene, each haplotype can then be tested against disease susceptibility using a small number of haplotype “tag” SNPs, htSNPs. Selection of htSNPs involves assessing the haplotype patterns of a gene and identifying regions of linkage disequilbrium (LD). These “blocks” of LD are discrete areas of low historical recombination across the human genome that possess a limited diversity of haplotypes (approximately 4-6 haplotypes/block). The block sizes and associated haplotypes appear to be conserved across populations, thus allowing the use of htSNPs to identify common haplotypes.5 7 Previous genetic epidemiologic studies have primarily focused on assessing putative functional variants in relation to disease, such as variants in coding exons that result in alteration of amino acid sequence. By using haplotype analysis, functional variants that lie far from coding regions that may also play an important role by influencing gene regulation, rather than protein sequence, may also be tested. Public (dbSNP) and private (Celera) databases have now begun cataloguing the growing number of discovered SNPs for use in further haplotype-based investigations. 1.3 PRELIMINARY RESULTS The Principal Investigator has extensive experience in conducting epidemiologic breast cancer research, and the associated laboratories at USC and MIT are well-established in conducting haplotype-based studies of candidate cancer susceptibility genes. In a concurrent project, MEC advanced breast cancer cases have been sequenced using primers amplifying all exons, 2 kb (kilobases) 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. upstream of the start of transcription, and up to three conserved regions (non coding mouse-human homology regions) using validated, high throughput methods currently implemented at the MIT/Whitehead Institute Center for Genome Research.55,5 8 The purpose of resequencing advanced breast cancer cases was to primarily identify missense variants, though additional polymorphisms were identified and genotyped. For PRL, three intronic variants will be selected and genotyped. SNP selection for PRLR will include one missense variant in exon 5, one synonymous variant in exon 6, and six intronic variants. Preliminary gene mapping and single nucleotide polymorphism (SNP) selection for both candidate genes have been performed by the Graduate Researcher using public (dbSNP) and private (Celera) databases, as well as MEC sequencing results. A total of 109 and 297 SNPs for PRL and PRLR, respectively, have been identified and mapped to include up to 20 kilobases (kb) upstream of the transcription initiation site and 10 kb downstream of the last exon of each gene. 1.4 RESEARCH DESIGN AND METHODS 1.4.1 Subjects 1.4.1.1 The Multiethnic Cohort The Hawaii-Los Angeles Multiethnic Cohort (MEC) Study is an ongoing, prospective NIH funded study (R01 CA54281-06: Laurence N. Kolonel, Principal Investigator). The cohort includes 215,251 men and women, enrolled between 1993 and 1996. The goal of the study was to enroll a broad representative sample of individuals from five different ethnic groups: African-Americans, Hawaiians, 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Latinos, Japanese, and Whites, to be followed up for cancer and other chronic diseases in relation to diet and environmental factors. Subjects were age 45-75 upon enrollment and were asked to fill out a questionnaire that included information on diet, demographic factors, personal behaviors, prior medical conditions, use of medications, family history of common cancers, and, for women, reproductive history and use of oral contraceptives or hormone replacement therapy. The primary source for subject recruitment was the Department of Motor Vehicle driver’s license files. To enroll an adequate number of African-American subjects, Health Care Financing Administration (HCFA) files were also used to include subjects from other California counties outside of Los Angeles. Incident cancer cases are identified quarterly using the Los Angeles County Cancer Surveillance Program (CSP) and linked annually with the State of California Cancer Registry (CCR) for subjects who have emigrated from Los Angeles. We have estimated that fewer than 3% of the Los Angeles cohort members have moved out of the state since the study began in 1993. Incident cancer cases in Hawaii are identified annually using the Hawaii Tumor Registry (HTR). All tumor registries participate in the NIH funded Surveillance, Epidemiology, and End Results (SEER) registry. Passive follow-up has already begun and includes a brief questionnaire with information on recent illnesses, current use of vitamin supplements, family history of cancer, and other disease outcomes. 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.4.1.2 Biological specimen collection Collection of blood and urine specimens from incident breast cancer cases and a random sample of cohort members began in July 1995 in Los Angeles and in February 1997 in Hawaii and California counties outside of Los Angeles. As of May 1, 2001,1421 of the 2557 (56%) invasive breast cancer patients who were contacted provided biospecimens. A sample of cohort members were contacted, based on random assignment of an “order number” that defined when subjects would be contacted for specimen collection. As of May 1, 2001, 2862 cohort controls have provided biospecimens, yielding an overall response rate of 51% for cases and random cohort controls. The Multiethnic Cohort Study is currently funded to prospectively collect biospecimens for an additional 45,000 cohort participants in Los Angeles and 40,000 in Hawaii by the end of 2006. Specimens were primarily collected by home visit in Los Angeles to avoid transportation or access issues, while in Hawaii patients were able to provide samples via satellite clinics of a large island-clinical laboratory. Trained phlebotomists collected 40 ml of blood on overnight fasting subjects and placed the samples in vacutainer tubes. The tubes were wrapped in foil to avoid exposure to sunlight and then immediately placed in sytrofoam containers with ice. The time since last meal or drink (other than water), the time of specimen collection, and the time at the end of processing was recorded for all specimens. Processing time was estimated to be within 6 hours of blood collection. An automated specimen component dispensing machine (the Cryo-Bio System), was used to aliquot serum, plasma, red blood cells, and white blood cells into “straws” of 0.5 ml volume. The 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. filled straws were then pre-ffozen to -80 °C for 30 minutes and then stored in liquid nitrogen until needed. 1.4.1.3 Subject selection For initial haplotype characterization, a multiethnic panel of 350 unaffected cohort participants (70 from each ethnic group) will be selected to test each single nucleotide polymorphism (SNP) marker. SNP frequencies greater than 5 % will then be used to determine LD block structures and construct haplotypes in each ethnic group for each candidate gene. In the nested, case control analysis, we will include all breast cancer cases identified through linkage to the Los Angeles and Hawaii cancer registries up to May 2002. Controls will be selected by frequency matching to cases on self- reported ethnicity and age. The case-control study is comprised of 1356 breast cancer cases and 2701 controls. By ethnic group (cases/controls), this includes: 278/711 African-Americans, 272/727 Latinas, 358/450 Japanese, 92/325 Hawaiians, and 356/488 Whites. Subjects selected for plasma prolactin analyses will include 500 female cohort controls with previously collected biospecimens (100 in each ethnic group). Only controls will be included in the hormone level analysis and will be a random sample from the MEC controls used in the case-control study. 13 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.4.2 Laboratory methods 1.4.2.1 DNA extraction and storage Previously extracted DNA using the Qiagen 96 DNA Blood Kit on breast cancer cases and controls was stored in labeled 96 well plates at -20 oC. Replicate 96 well dilution plates (1.5 ng/ml) will be used to make 384 well plates for genotyping. The MultiMec 96 (Beckman Coulter Inc.) and the Tomtec Quadra 384 (Tomtec Inc.) will be used for accurate DNA transfer from 96 to 96, and 96 to 384 well plates, respectively. Genotyping plates for haplotype characterization will be one 384 well plate of unaffected individuals, and genotyping plates for breast cancer case-control and haplotype-plasma hormone level analyses will be twelve 384 well plates (4,260 samples). All plates will include 5% replicates for genotyping quality control purposes. 1.4.2.2 SNP genotyping Genotyping will involve two main stages for both candidate genes: 1) testing of selected SNPs to determine haplotype blocks and common haplotypes in each ethnic group and 2) genotyping a minimal set of selected “tag” SNPs (htSNPs) for each gene to be used in the breast cancer case-control and haplotype- plasma hormone level analyses. 1.4.2.2.1 Haplotype characterization Haplotype characterization and genotyping of initially tested SNPs will first involve selection of the appropriate SNPs for genotyping. Using a comprehensive 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. SNP map that includes up to 20 kilobases (kb) upstream of the transcription start site and 10 kb downstream of the last exon of each candidate gene, we will select one SNP approximately every 4 kb. The goal for each candidate gene is to map a SNP with a minor allele frequency >10% every 3-5 kb. For PRL and PRLR, the total region covered for each gene will be 40 kb and 210 kb, respectively. This will require approximately 89 SNPs (63 SNPs selected on average every 4 kb + 26 SNPs (40%) to account for monomorphic and rare SNPs). The SNP map will include all variants found in the public database (dbSNP), private database (Celera), and from our database of previously sequenced SNPs among a subsample of MEC advanced breast cancer cases (n = 95, 19 from each ethnic group). Based on our experience, several iterations of SNP selection and genotyping, must be performed to attain adequate SNP coverage, since many SNPs are found to be rare or monomorphic in our population. Additional considerations in SNP selection include testing SNPs in highly conserved mouse-human homology regions and avoiding “repeat” regions (areas with high homology to other sequences in the human genome that may lead to high genotyping failure). All genotyping for haplotype characterization will be done at the MIT/Whitehead Institute Center for Genome Research using MALTI-TOF mass spectroscopy (Sequenom Inc.). The Sequenom system allows for multiplexing up to 5 SNPs in each well and uses single base extension and oligonucleotide mass measurement to identify alleles. Technology developed by Sequenom and MIT/WICGR will be used to automate genotype calling and data storage and analysis. Programs developed at USC will be used to assess 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. genotyping quality, allele frequency, and Hardy-Weinberg equilibrium of the tested SNPs. 1.4.2.2.2 Tag SNP genotyping Based on htSNP selection described below, a minimal set of tag SNPs will be genotyped at the USC-MEC Genotyping Laboratory using the ABI 7900 Sequence Detection System (Applied Biosystems, Foster City, CA) using the Taqman Allelic Discrimination assay. From our previous experience in haplotype tag selection, we estimate that the candidate gene PRL will contain approximately 2-3 blocks of LD and PRLR, 6-7 blocks. Based on these block numbers and our estimate of 4-5 htSNPs per block, we predict genotyping 9-14 htSNPs for PRL and 27-32 htSNPs for PRLR. The htSNPs will be genotyped among 4260 samples (1356 breast cancer cases, 2701 controls, 5% quality control repeats). 1.4.2.3 Plasma prolactin level assay Assays for plasma prolactin levels on the selected 500 cohort controls were performed by the hormone analysis laboratory at the International Agency for Research on Cancer (IARC), Hormones and Cancer group. The hormones are measured using a double-antibody, i mmunoradi ometri c assay from Diagnostic System Laboratories (Webster, Texas). The mean intra-batch coefficient of variation of the method is 5.4% using a sample volume of 25 microliters. Previously documented findings have shown plasma prolactin levels to be stable in 16 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. whole blood for 24-48 hours.5 9 hi this study, time from blood collection to processing is estimated to be no more than 6 hours. 1.4.3 Statistical analyses 1.4.3.1 Haplotype block definition, haplotype estimation, and htSNP selection Using an objective set of criteria based on confidence intervals of a measure of linkage disequilibrium, called D', we will define blocks of LD using SNPs with frequencies >10%. Using the expectation-maximization (EM) algorithm,6 0 we will estimate haplotype frequencies within defined blocks in the preliminary multiethnic panel of control subjects (70 per ethnic group). We use an objective criteria described by Stram et al.6 1 for defining the best set of haplotype-tagging SNPs (htSNPs) which will be genotyped in the case-control study. This involves the calculation of a coefficient of determination, Rh, for each of the common haplotypes that are found in the preliminary sample. This criteria measures the precision of haplotype assignment based on a given set of SNP genotype data given all the haplotypes (common and uncommon) that are found in the preliminary sample. We will pick htSNPs so that each common haplotype (those with frequency > 5 percent) within each block are identified with a precision of at least Rh = 0.9. The formal definition of our criteria of precision, Rh, is as follows: for a given set of haplotype frequencies, P, estimated by the EM algorithm, and for a given haplotype h, Rh is the squared correlation coefficient between the count of the number of copies of h, 5 h (H j), contained in the true pair of haplotypes H, in a 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. given individual and the estimated haplotype count, based on known genotype data, E .{5 h (H j) | G;}, which is computed during the course of the EM algorithm. Assuming Hardy-Weinberg equilibrium, the squared correlation coefficient can be calculated by the equation Rh 2 =Ya r [E { 6 h(H i)| Q } ] / 2 p h ( 1 - p h) , where ph is the estimated frequency of haplotype h and where the variance of E { 5 h (H j)| G j } , is computed by summation over all possible sets of genotype data G j weighted by its probability of occurrence, P ( G i) = 5 h ~ Gi P hiph 2 • (Here H~Gi refers to the set of haplotype pairs compatible with a given Gi.) 1.4.3.2 Imputation of haplotypes for breast cancer case-control analysis and for hormone level-haplotype analysis The primary method of analysis that we will use to estimate haplotype- specific risks, for a given haplotype h, is that described by Zaykin et al.6 2 In the Zaykin approach estimated haplotype counts, E{Sh(Hi)| Gi} (computed in the course of the EM algorithm when applied to the full set of cases and controls) for each subject i and each haplotype of possible interest h, are output to a file at convergence of the EM algorithm and merged with case-control status. Then standard software for case control analysis (e.g. Proc Logistic in SAS) is used to relate estimated haplotype count to case control status in the course of a logistic regression analysis. We term this method of regression analysis in which the (generally unknown) true numbers of copies of haplotype h, for each subject, are replaced by their expectation, as the “single-imputation” method. As described by 18 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Zaykin and colleagues, single-imputation provides appropriate tests of hypotheses in order to determine whether a particular haplotype is significantly associated with breast cancer risk either by itself or after adjustment for the influence of other covariates. In addition to the single-imputation method we are currently evaluating a modified EM algorithm, which simultaneously estimates disease risk parameters and haplotype frequencies in the course of a single unified analysis. In theory the expanded EM algorithm produces better estimates and more appropriate confidence intervals than does the single-imputation method. Our initial unpublished practical experience with these two algorithms (single-imputation and expanded EM) shows that they tend to produce very similar results, however, so long as haplotypes are well determined (Rh > 0.9) by the htSNPs chosen. Therefore we intend to utilize the single imputation method as our primary analysis tool for estimating haplotype- specific risks in the case control data. In addition to the analysis of case control status we will use the single imputation method to estimate the effect of estimated haplotypes on plasma PRL levels using generalized linear models adjusted for continuous (age, anthropometry) and categorical (reproductive history) variables. For all analyses, a dominant, co-dominant, and recessive model will be fitted. 1.4.4 Sample size and power estimation 1.4.4.1 Power to detect common haplotypes For haplotype characterization, we estimate that our multiethnic panel of 350 subjects (700 chromosomes), 70 subjects (140 chromosomes) in each ethnic 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. group, will be more than adequate to detect common haplotypes. We estimate having greater than 97% power to detect ethnic-specific haplotypes with frequencies as low as 5%. For common haplotypes across ethnic groups, we estimate having power greater than 99% to detect haplotypes with 5% frequency among the total population. 1.4.4.2 Power for main genotype and haplotype effects In the case-control analysis, the power to detect main effects of genotypes (e.g. missense SNPs) among 1356 cases and 2701 controls depends on the mode of inheritance. In a dominant model with an allele frequency of 5%, we estimate having 80% statistical power to detect a relative risk of 1.35, and 97% power to detect a relative risk of 1.5. In a recessive model with an allele frequency of 10%, we estimate 80% power to detect a relative risk of 2.25. By ethnic group, in a dominant model with allele frequency of 10%, we estimate having 90% power to detect a relative risk of 1.75 in all ethnic groups except Hawaiians (RR = 2.50). The power to detect the effect of a haplotype depends on the ability of the selected htSNPs to predict common haplotypes, the frequency of the haplotype, and the model penetrance. Therefore, the high degree of uncertainty associated with estimating common haplotypes results in a corresponding decrease in power. Thus, the estimated increase in sample size necessary to detect a haplotype-specific risk is 1/ Rh 2. With our large sample size of cases and controls, we estimate that haplotypes with an Rh >0.90 will allow more than adequate statistical power to detect modest risks of haplotypes of frequencies as low as 5% in the total 20 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. population. In a dominant model, we will again have 80% power to detect a relative risk of 1.35 for a haplotype of 5% frequency. Since a large number of genotypes and haplotypes will be considered, we use Bonferroni adjustment to adjust for multiple comparison testing. 1.4.4.3 Power for hormone level and haplotype effects Laboratory variation in the measurement PRL is considered to be small (~5 percent coefficient of variation, personal communication with Dr. Rudolf Kaaks, IARC) so that we neglect this issue in our analysis of power. We will have 500 subjects in the plasma hormone analysis. We estimate the mean in the population of plasma PRL to be equal to M = 8.0 and the standard deviation equal to S = 7.5. For SNP genotype analyses we anticipate that we will be able to detect (with 80 percent power) a dominant effect, which raises plasma hormone level by approximately 25 percent of a standard deviation (from 8.0 to 9.9, effect size of 0.25) for an allele with population frequency equal to 20 percent. For an allele with 5 percent frequency we will be able to detect a dominant effect only if it corresponds to an increase (effect size) size of 45 percent of a standard deviation or greater, for an allele with frequency equal to 5 percent. In haplotype analyses our choice of htSNPs so that Rh > 0.9 will mean that imprecision in haplotype estimation will have a small effect only on the detection of haplotype-specific effects on plasma PRL levels. For example we anticipate that increases in plasma hormone level of 26 percent or of 47 percent of 7.5, will be detectable for a dominant haplotype with frequency of 20 percent or 5 percent, respectively. 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.5 TIMELINE The proposed project is budgeted for an 18 month period of investigation. Though various stages of the project will be conducted at different points during the study, in general, many of the steps will overlap in time. This is due to the iterative nature of the procedures involved in haplotype characterization and final haplotype tag SNP analysis. Therefore, the following timeline gives broad estimates for when each step will be conducted: Year 1 Year 2 (6 mo.) SNP genotyping and haplotype discovery X Haplotype tag SNP selection X X Haplotype tag genotyping X X Case-control and haplotype-plasma PRL analyses X 2 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.6 REFERENCES 1. Boot LM, Muhlbock 0, Ropcke G. Prolactin and the induction of mammary tumors in mice. General and Comparative Endocrinology. 1962;2:601-602. 2. Biswas R, Vonderhaar BK. Role of serum in the prolactin responsiveness of MCF-7 human breast cancer cells in long-term tissue culture. Cancer Res. 1987;47(13):3509-3514. 3. Brisken C, Kaur S, Chavarria TE, et al. Prolactin controls mammary gland development via direct and indirect mechanisms. Dev Biol. 1999;210(1):96- 106. 4. Muhlbock O, Boot LM. Induction of mammary cancer in mice without the mammary tumor agent by isografts of hypophyses. Cancer Res. 1959;19:402-412. 5. Welsch CW, Gribler C. Prophylaxis of spontaneously developing mammary carcinoma in C3H-HeJ female mice by suppression of prolactin. Cancer Res. 1973;33(11):2939-2946. 6. Welsch CW, Nagasawa H. Prolactin and murine mammary tumorigenesis: a review. Cancer Res. 1977;37(4):951-963. 7. Wennbo H, Gebre-Medhin M, Gritli-Linde A, Ohlsson C, Isaksson OG, Tomell J. Activation of the prolactin receptor but not the growth hormone receptor is important for induction of mammary tumors in transgenic mice. J Clin Invest. 1997; 100(11):2744-2751. 8. Hankinson SE, Willett WC, Michaud DS, et al. Plasma prolactin levels and subsequent risk of breast cancer in postmenopausal women. JNatl Cancer Inst. 1999;91(7):629-634. 9. Malarkey WB, Schroeder LL, Stevens VC, James AG, Lanese RR. Disordered nocturnal prolactin regulation in women with breast cancer. Cancer Res. 1977;37(12):4650-4654. 10. Meyer F, Brisson J, Morrison AS, Brown JB. Endogenous sex hormones, prolactin, and mammographic features of breast tissue in premenopausal women. JNatl Cancer Inst. 1986;77(3):617-620. 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11. Rose DP, Pruitt BT. Plasma prolactin levels in patients with breast cancer. Cancer. 1981;48(12):2687-2691. 12. Ingram DM, Nottage EM, Roberts AN. Prolactin and breast cancer risk. MedJAust. 1990;153(8):469-473. 13. Lyons WR, Li CH, Johnson RE. The hormonal control of mammary growth and lactation. Recent Prog Horm Res. 1958;14:219-254. 14. Wilson J, Foster D, Kronenber H, Larse PR, editors. Williams textbook of endocrinology. 9th edition ed. Philadelphia, PA: W.B. Saunders Company; 1998. 15. Cowie AT, Tindal JS, Yokoyama A. The induction of mammary growth in the hypophysectomiized goat. J Endocrinol. 1966;34:185-195. 16. Talwalker PK, Meites J. Mammary lobulo-alveolar growth induced by anterior pituitary hormones in adreno-ovariectomized-hypophysectomized rats. Proc Soc Exp Biol Med. 1961 ;107(880-3). 17. Hayden TJ, Bonney RC, Forsyth IA. Ontogeny and control of prolactin receptors in the mammary gland and liver of virgin, pregnant and lactating rats. J Endocrinol. 1979;80(2):259-269. 18. Dhadly MS, Walker RA. The localization of prolactin binding sites in human breast tissue. Int J Cancer. 1983;31(4):433-437. 19. Jahn GA, Edery M, Belair L, Kelly PA, Djiane J. Prolactin receptor gene expression in rat mammary gland and liver during pregnancy and lactation. Endocrinology. 1991; 128(6):2976-2984. 20. Kledzik GS, Bradley CJ, Marshall S, Campbell GA, Meites J. Effects of high hoses of estrogen on prolactin-binding activity and growth of carcinogen-induced mammary cancers in rats. Cancer Res. 1976;36(9 pt.l):3265-3268. 21. Truong AT, Duez C, Belayew A, et al. Isolation and characterization of the human prolactin gene. Embo J. 1984;3(2):429-437. 22. Berwaer M, Martial JA, Davis JR. Characterization of an up-stream promoter directing extrapituitary expression of the human prolactin gene. Mol Endocrinol. 1994;8(5):635-642. 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23. DiMattia GE, Gellersen B, Duckworth ML, Friesen HG. Human prolactin gene expression. The use of an alternative noncoding exon in decidua and the IM-9-P3 lymphoblast cell line. J Biol Chem. 1990;265(27): 16412- 16421. 24. Kelly PA, Djiane J, Postel-Vinay MC, Edery M. The prolactin/growth hormone receptor family. Endocr Rev. 1991 ;12(3):235-251. 25. Hennighausen L, Robinson GW, Wagner KU, Liu W. Prolactin signaling in mammary gland development. J Biol Chem. 1997;272(12):7567-7569. 26. Udy GB, Towers RP, Snell RG, et al. Requirement of STAT5b for sexual dimorphism of body growth rates and liver gene expression. Proc Natl Acad Sci USA. 1997;94(14):7239-7244. 27. Wakao H, Gouilleux F, Groner B. Mammary gland factor (MGF) is a novel member of the cytokine regulated transcription factor gene family and confers the prolactin response. Embo J. 1994;13(9):2182-2191. 28. Arden KC, Boutin JM, Djiane J, Kelly PA, Cavenee WK. The receptors for prolactin and growth hormone are localized in the same region of human chromosome 5. Cytogenet Cell Genet. 1990;53(2-3): 161-165. 29. Hu ZZ, Zhuang L, Meng J, Leondires M, Dufau ML. The human prolactin receptor gene structure and alternative promoter utilization: the generic promoter hPIII and a novel human promoter hP(N). J Clin Endocrinol Metab. 1999;84(3):1153-1156. 30. Hu ZZ, Meng J, Dufau ML. Isolation and characterization of two novel forms of the human prolactin receptor generated by alternative splicing of a newly identified exon 11. J Biol Chem. 2001 ;276(44) :41086-41094. 31. Ormandy CJ, Binart N, Kelly PA. Mammary gland development in prolactin receptor knockout mice. J Mammary Gland Biol Neoplasia. 1997;2(4):355-364. 32. Peyrat JP, Dewailly D, Djiane J, et al. Total prolactin binding sites in human breast cancer biopsies. Breast Cancer Res Treat. 1981;l(4):369-373. 33. Reynolds C, Montone KT, Powell CM, Tomaszewski JE, Clevenger CV. Expression of prolactin and its receptor in human breast carcinoma. Endocrinology. 1997;138(12):5555-5560. 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34. Mertani HC, Garcia-Caballero T, Lambert A, et al. Cellular expression of growth hormone and prolactin receptors in human breast disorders. Int J Cancer. 1998;79(2):202-211. 35. Touraine P, Martini JF, Zafrani B, et al. Increased expression of prolactin receptor gene assessed by quantitative polymerase chain reaction in human breast tumors versus normal breast tissues. J Clin Endocrinol Metab. 1998;83(2):667-674. 36. Salih H, Brander W, Flax H, Hobbs JR. Prolactin dependence in human breast cancers. Lancet. 1972;2(7787):1103-1105. 37. Peyrat JP, Djiane J, Bonneterre J, et al. Stimulation of DNA synthesis by prolactin in human breast tumor explants. Relation to prolactin receptors. Anticancer Res. 1984;4(4-5):257-261. 38. Welsch CW, Iturri GC, Brennan MJ. DNA synthesis of human, mouse, and rat mammary carcinomas in vitro: influence of insulin and prolactin. Cancer. 1976;38(3):1272-1281. 39. Calaf G, Garrido F, Moyano C, Rodriguez R. Influence of hormones on DNA synthesis ofbreast tumors in culture. Breast Cancer Res Treat. 1986;8(3):223-232. 40. Wilson GD, Woods KL, Walker RA, Howell A. Effect of prolactin on lactalbumin production by normal and malignant human breast tissue in organ culture. Cancer Res. 1980;40(2):486-489. 41. Shafie SM, Grantham FH. Role of hormones in the growth and regression of human breast cancer cells (MCF-7) transplanted into athymic nude mice. JNatl Cancer Inst. 1981 ;67(1):51-56. 42. Malarkey WB, Kennedy M, Allred LE, Milo G. Physiological concentrations of prolactin can promote the growth of human breast tumor cells in culture. J Clin Endocrinol Metab. 1983;56(4):673-677. 43. Buytaert P, Viaene P. Amenorrhea, galactorrhea, hyperprolactinemia syndrome and breast carcinoma in a young woman. Eur J Obstet Gynecol Reprod Biol. 1981; 11(5):341 -346. 44. Strangs I, Gray RA, Rigby HB, Stratton G. Two case reports ofbreast carcinoma associated with prolactinoma. Pathology. 1997;29(3):320-323. 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45. Popovic V, Damjanovic S, Micic D, et al. Increased incidence of neoplasia in patients with pituitary adenomas. The Pituitary Study Group. Clin Endocrinol (Oxj). 1998;49(Oct (4)):441-445. 46. Olsson H, Aim P, Kristoffersson U, Landin-Olsson M. Hypophyseal tumor and gynecomastia preceding bilateral breast cancer development in a man. Cancer. 1984;53(9):1974-1977. 47. Haga S, Watanabe O, Shimizu T, et al. Breast cancer in a male patient with prolactinoma. Surg Today. 1993;23(3):251 -255. 48. Volm MD, Talamonti MS, Thangavelu M, Gradishar WK. Pituitary adenoma and bilateral male breast cancer: an unusual association. J Surg Oncol. 1997;64(l):74-78. 49. Holtkamp W, Nagel GA, Wander HE, Rauschecker HF, von Heyden D. Hyperprolactinemia is an indicator of progressive disease and poor prognosis in advanced breast cancer. Int J Cancer. 1984;34(3):323-328. 50. Haus E, Lakatua DJ, Halberg F, et al. Chronobiological studies of plasma prolactin in women in Kyushu, Japan, and Minnesota, USA. J Clin Endocrinol Metab. 1980;51(3):632-640. 51. Henderson BR, Gerkins V, Rosario I, Casagrande J, Pike MC. Elevated serum levels of estrogen and prolactin in daughters of patients with breast cancer. N Engl J Med. 1975;293(16):790-795. 52. Secreto G, Recchione C, Cavalleri A, Miraglia M, Dati V. Circulating levels of testosterone, 17 beta-oestradiol, luteinising hormone and prolactin in postmenopausal breast cancer patients. Br J Cancer. 1983;47(2):269-275. 53. Bernstein L, Ross RK, Pike MC, Brown JB, Henderson BE. Hormone levels in older women: a study of post-menopausal breast cancer patients and healthy population controls. Br J Cancer. 1990;61(2):298-302. 54. Wang DY, De Stavola BL, Bulbrook RD, et al. Relationship of blood prolactin levels and the risk of subsequent breast cancer. Int J Epidemiol. 1992;21 (2) :214-221. 55. Cargill M, Altshuler D, Ireland J, et al. Characterization of single nucleotide polymorphisms in coding regions of human genes. Nat Genet. 1999;22(3):231-238. 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56. Kruglyak L, Nickerson DA. Variation is the spice of life. Nat Genet. 2001 ;27(3):234-236. 57. Gabriel SB, Schaflher SF, Nguyen H, et al. The structure of haplotype blocks in the human genome. Science. 2002;296(5576):2225-2229. 58. Nickerson DA, Tobe VO, Taylor SL. PolyPhred: automating the detection and genotyping of single nucleotide substitutions using fluorescence-based resequencing. Nucleic Acids Res. 1997;25(14):2745-2751. 59. Hankinson SE, London SJ, Chute CG, et al. Effect of transport conditions on the stability of biochemical markers in blood. Clin Chem. 1989;35(12):2313-2316. 60. Excoffier L, Slatkin M. Maximum-likelihood estimation of molecular haplotype frequencies in a diploid population. Mol Biol Evol. 1995;12(5):921-927. 61. Stram DO, Haiman CA, Hirschhom JN, et al. Choosing Haplotype-tagging SNPs based on unphased genotype data using a preliminary sample of unrelated subjects with an example from the Multiethnic Cohort Study. Human Hered. 2003;55(l):227-236. 62. Zaykin DV, Westfall PH, Young SS, Kamoub MA, Wagner MJ, Ehm MG. Testing association of statistically inferred haplotypes with discrete and continuous traits in samples of unrelated individuals. Hum Hered. 2002;53(2):79-91. 63. Stram DO, Pearce CL, Bretsky P, et al. Abstract modeling and E-M estimation of haplotype specific relative risks from genotype data for a case control-study of unrelated individuals. Hum Hered. 2003;55(4):179-190. 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 2 HAPLOTYPE-BASED ANALYSIS OF PROLACTIN PATHWAY GENES IN RELATION TO PLASMA PROLACTIN LEVELS AND BREAST CANCER RISK IN THE MULTIETHNIC COHORT Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.1 INTRODUCTION Prolactin (PRL) is an essential regulator of mammary development, acting 1 2 synergistically with a wide variety of hormones during puberty and pregnancy. ’ Early studies in animals first demonstrated that prolactin could induce spontaneous mammary tumors.3 '6 Results from in vitro studies support the findings from animal studies and suggest that PRL stimulates proliferation,7 '1 0 increases cell motility and cytoskeleton alterations,1 1 and increases angiogenesis1 2 in human breast cells. Prolactin receptor (PRLR), found in both normal and malignant breast tissue, is slightly more prevalent in malignant tissue.1 3 Though early clinical studies of patients treated with bromocriptine, an inhibitor of pituitary PRL, found no association with breast cancer, recent evidence of autocrine/paracrine regulation1 4 ’ 1 5 of PRL in extra-pituitary tissue lends credence to the idea of PRL’s role in tumorigenesis. There are few prospective epidemiologic studies evaluating plasma PRL levels and breast cancer risk. The largest prospective cohort study of postmenopausal women reported a 34% increase in risk ofbreast cancer when comparing top to bottom quartiles of PRL levels;1 6 these findings were similar to results from an earlier study reporting a non-significant increase in risk of 1.34, based on a smaller sample size.1 7 Two smaller studies of postmenopausal women also reported a positive association, but these were also non-significant.1 8 ’1 9 Results from case-control studies2 0 '2 7 or prospective studies among premenopausal -JO i r t -)Q women ’ ’ give conflicting results are difficult to interpret due to the 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. retrospective nature of blood collection or large intra-individual variability due to changes in levels with the menstrual cycle. In humans, the PRL gene lies on chromosome 6 and is approximately 10 kilobases (kb) in length with five coding exons.2 9 An additional non-coding first exon has been described that lies 5.8 kb upstream of the pituitary promoter site.3 0 This distal promoter region has been associated with extra-pituitary expression of PRL, described in a variety of tissues including decidua, lymphocytes, and breast tissue. Depending on promoter usage, PRL mRNAs may differ slightly in length but encode the same mature polypeptide protein hormone.3 1 The human PRLR gene is located on chromosome 5 and is approximately 180 kb in length and is originally described as having 10 exons, of which exons 3-10 are coding exons.3 2 Recently, six alternative non-coding first exons have been described whose functions are unknown but have been found to be expressed in human ovary, testis, liver, breast tissue, and breast cells.33,3 4 In addition, an exon 11 located 15 kb downstream of exon 10 has been reported; alternative splicing of exons 10 and 11 appear to produce novel short forms of the receptor that may be involved in distinct signaling pathways than the common long form.3 5 ,3 6 This is the first study to evaluate common genetic variation in the PRL and PRLR loci in relation to breast cancer risk and plasma PRL levels. We used a haplotype-based approach among African-American, Hawaiian, Japanese, Latina, and White women in the Hawaii-Los Angeles Multiethnic Cohort (MEC) Study. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2 METHODS 2.2.1 Subjects The Hawaii-Los Angeles Multiethnic Cohort (MEC) Study is an ongoing, prospective study of dietary, environmental, and genetic factors in relation to cancer and other chronic diseases. The cohort includes 215,251 men and women, age 45 to 75 years at baseline, enrolled between 1993 and 1996 in Hawaii and California (primarily Los Angeles County). Based on self-report, 118,913 African- American, Native Hawaiian, Japanese-American, Latina, and White women were included in this study; a small number of subjects defined as of “other” ethnicity were excluded. Subjects completed a 26-page, self-administered mailed questionnaire that included information on diet, demographic factors, personal behaviors, prior medical conditions, use of medications, family history of common cancers, and, for women, reproductive history and use of oral contraceptives or hormone therapy. The primary sources for subject recruitment were the Departments of Motor Vehicles drivers’ license files for Hawaii and Los Angeles. An additional source of African-Americans in California was the Health Care Financing Administration (HCFA) files, and in Hawaii, voters’ registration files were used to identify additional persons of Native Hawaiian ancestry and Japanese-American descent. Further details regarding the study design and characteristics of the cohort have been described elsewhere.3 7 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.2.1.1 Surveillance Incident cancer cases are identified quarterly using the Los Angeles County Cancer Surveillance Program (CSP) and linked annually with the State of California Cancer Registry (CCR) for subjects who have emigrated from Los Angeles. We have estimated that fewer than 3% of the Los Angeles cohort members have moved out of the state since the study began in 1993. Incident cancer cases in Hawaii are identified annually using the Hawaii Tumor Registry (HTR) and the CCR. All tumor registries participate in the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program of cancer registration. Mortality due to cancer and other causes is determined by annual linkage to state death certificate files in California and Hawaii, and periodically to the National Death Index (NDI). Follow-up begins when the subject enters the cohort, at the time of completing the questionnaire. . 2.2.1.2 Biological specimen collection Collection of blood and urine specimens from incident breast cancer cases and a random sample of cohort members began in July 1995 in Los Angeles and in February 1997 in Hawaii and California counties outside of Los Angeles. As of May 31, 2003, response rates were 74% for invasive breast cancers and 66% for randomly selected cohort controls. The Multiethnic Cohort Study is currently funded to prospectively collect biospecimens for an additional 45,000 cohort participants in Los Angeles and 40,000 in Hawaii by the end of 2006. 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Specimens were collected primarily by home visit in Los Angeles to avoid transportation or access issues, while in Hawaii patients were able to provide samples via satellite clinics of a large island-clinical laboratory. Trained phlebotomists collected 40 ml of blood on overnight fasting subjects and placed the samples in vacutainer tubes. The tubes were wrapped in foil to avoid exposure to sunlight and then immediately placed in sytrofoam containers with ice. The time since last meal or drink (other than water), time of specimen collection, and time at the end of processing were recorded for all specimens. Processing time was estimated to be within 6 hours of blood collection. An automated specimen component dispensing machine (the Cryo-Bio System), was used to aliquot serum, plasma, red blood cells, and white blood cells into “straws” of 0.5 ml volume. The filled straws were then pre-frozen to -80 °C for 30 minutes and then stored in liquid nitrogen until needed. 2.2.2 SNP selection and haplotype block determination We constructed high density gene maps that included up to 20 kb upstream of the transcription initiation site and 10 kb downstream of the last exon of each gene using information from the National Center for Biotechnology Information SNP database (www.ncbi.nlm.nih.gov/SNP/) and the Celera database (www.celera.com), as well as SNPs discovered from a concurrent sequencing project of coding and splice site regions of PRL and PRLR among 95 advanced breast cancer cases (19 from each ethnic group) in the MEC. The goal was to have 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. at least one common SNP (>10%) every 3-5 kb across each locus to ensure a high density of markers to adequately characterize the haplotype diversity with each LD block. SNPs were selected and genotyped iteratively until 6-8 common markers per LD block were identified with approximately <10 kb between blocks. All SNPs within coding regions were included in the analysis. We used a normalized measure of allelic association, the D' statistic, to estimate the extent of linkage disequilibrium between SNPs. LD blocks were defined using the criteria described by Gabriel et al.3 8 which defines sites of historical recombination between SNPs based on the 90% confidence bounds of D'. SNPs with minor allele frequencies >10% in at least one ethnic group were used to define the blocks. Initial block boundaries were defined based on the LD structure for total population; borders were redefined to minimize inter-block distances while maintaining the number and identity of common haplotypes within the block. For African-Americans, whose extent of LD tends to be smaller than for the other groups,3 8 ’3 9 further refinement of borders was necessary to define LD blocks to best characterize this group. Genotyping plates for haplotype characterization consisted of one 384 well plate of non-diseased female subjects: African-American (n = 70), Hawaiian (n = 69), Japanese (n = 70), Latina (n = 70), and White (n = 70). Replicate quality control samples (10%) were included to assess reproducibility of the genotyping procedure. Less than 0.2% of the matched quality control pairs were discordant. DNA for the multiethnic panel were extracted from white blood cells using the 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Qiagen 96 DNA Blood Kit and stored in labeled 96 well plates at -20 °C. Replicate 96 well dilution plates (1.5 ng/ml) were used to make 384 well plates for genotyping. The MultiMec 96 (Beckman Coulter Inc.) and the Tomtec Quadra 384 (Tomtec Inc.) were used for accurate DNA transfer from 96 to 96, and 96 to 384 well plates, respectively. All genotyping for haplotype characterization was performed at the Massachusetts Institute of Technology (MIT)/Broad Institute by primer extension of multiplex products with detection by Matrix-Assisted Laser Desorption Ionization-Time of Flight (MALDI-TOF) mass spectroscopy using the Sequenom platform. 2.2.3 Haplotype estimation and htSNP selection Haplotype frequency estimates were constructed from genotype data in the multiethnic panel (one ethnicity at a time) within blocks using the expectation- maximization (E-M) algorithm of Excoffier and Slatkin.4 0 The squared correlation {Rh) between the true haplotypes (h) and their estimates from this calculation were then estimated as described by Stram et al.4 1 Briefly, for any given set of true haplotype frequencies, Ph, we can make a formal calculation (under Hardy- Weinberg equilibrium) of the squared correlation, Rh, between the estimate E{bh{Ht) | Gi}, and the true value, bh{Ht ), of the number of copies of h carried by a randomly sampled subject, [i.e. bh{Ht ) = 0 , 1, or 2]. Here Gi is the genotype data for each subject, i, and Hi is true (but generally unknown) pair of haplotypes carried 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. by that individual. The estimate is calculated as E{ 8h(Hi)| Gi} = 8h(H) P hiV h2 '£>H~GiPhlPh2 where E^-g/ indicates a summation over the haplotype pairs, H = {hlr h2), that are compatible with the observed genotype data, and p h is the frequency of haplotype h. Under the assumption of Hardy-Weinberg equilibrium, the correlation may be most easily calculated as R h - Vm\E{8h(Hi)\ Gj}] 2ph(l-ph) Where the variance of the expection is computed by averaging E{ 8h(H)| G}and E{8h(H)| G}2 over all possible genotyped G, weighting by the probability of each genotyped. This method explicitly recognizes that it is genotypes rather than haplotypes that are directly read, taking account of the resulting haplotype uncertainty. This uncertainty has not generally been accounted for in other haplotype SNP selection methods.42,4 3 Rh is a sample size inflation factor; to achieve equivalent power as having perfectly tagged the haplotypes using N samples requires approximately N/ Rh samples. We selected tag SNPs for the case-control study by finding the minimum set of SNPs within a block with an Rh >0.7 for all haplotypes with an estimated frequency of >5%. The actual Rh s for the haplotypes defined are generally higher and are shown in the Results section. All Rh calculations were performed using the tagSNPs program (www-rcf.usc.edu/~stram). 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TagSNPs were genotyped among a panel of 1,586 breast cancer cases and 2,704 age-matched controls. By ethnic group, these included (cases/controls): 329/ 712 African-Americans, 320/ 728 Latinas, 415/ 450 Japanese, 105/ 325 Hawaiians, and 397/ 489 Whites. Genotyping was conducted by the 5’ nuclease Taqman alleleic discrimination assay using ABI7900 Sequence Detection System (Applied Biosystems, Foster City, CA) and by MALDI-TOF on twelve 384 well plates with 5% replicates. Laboratory personnel were blinded to case-control status and concordance for the blinded samples was > 99%. Genotyping of case-control assays was performed in the MEC Genotyping Laboratory. In addition, we calculated the multivariate squared correlation, Rs2 , between measured SNPs and the predicted haplotypes in each block, to assess whether we had comprehensively assessed the common genetic variation in the region. This correlation is computed, as is Rh, based on the estimated haplotype frequencies under the Hardy-Weinberg equilibrium rule and suggests that unmeasured SNPs are highly correlated with the haplotypes in the block and that the htSNPs adequately predict the unmeasured SNPs. 2.2.4 Comparison of haplotype frequencies among breast cancer cases and controls Haplotype frequencies among breast cancer cases and controls were estimated using the htSNPs selected to distinguish common haplotypes (>5%) for each ethnic group in the multiethnic panel. Following the method of Zaykin et 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. al.,4 4 for each individual and each haplotype, h, the haplotype dosage estimate (i.e. an estimate of the number of copies of haplotype h) was computed using that individual’s genotype data and haplotype frequency estimates obtained from the overall dataset (cases and controls). These individual estimates were merged with all other individual data. All of the variables were used in unconditional logistic regression analyses with the estimate of haplotype dosage treated as a surrogate variable for the true haplotype. Under the null hypothesis (of no haplotype-specific effects on risk), the usual score test from the logistic regression, when haplotype is added to the mdoel, will correspond to the test described by Zaykin et al.4 4 We have found that this approach gives accurate estimates of the statistical significance, and that confidence intervals are appropriate when i?/is high.4 5 Odds ratios and 95% confidence intervals for each haplotype were estimated using all other haplotypes among the total population as the reference group, adjusted for age. Haplotype-specific risks for the total population were calculated based on data from ethnic groups where the haplotype was observed at >5% frequency. All summary odds ratios were adjusted for age and race. We first conducted a likelihood ratio test to globally test whether the common haplotype frequencies differed between cases and controls. If the global test was significant for a given block, this suggested that the individual haplotypes within the block could be associated with risk, either by ethnic group or for all groups combined. Performing the global test helps address the need to perform bonferroni corrections of the independent haplotype effects. Odds ratios and 95% confidence intervals 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. were estimated for each haplotype and the missense SNP, among all groups combined and by ethnic group. Likelihood ratio tests were performed to test for differences in effect by ethnic group by including an interaction term for the haplotype or missense SNP, and ethnicity in the multivariate model. Results were also tested to see if estimates changed based on adjustment for the following breast cancer risk factors: age at menarche, age at first birth, parity, age and type of menopause, body mass index, hormone therapy use, family history of breast cancer, and alcohol consumption. All analyses were performed using the Statistical Analysis System software (SAS Institute Inc., Cary, North Carolina). 2.2.5 Plasma prolactin level assay Subjects included in the analysis of plasma PRL levels were a random sample of the controls in the case-control panel. A total of 500 women with previously collected biospecimens (100 in each ethnic group) were included. Women reporting hormone therapy use at blood draw were excluded (n = 128), and individuals with PRL levels that were 2.5-fold outside the normal range were excluded (n = 17). Assays for plasma prolactin levels were performed by the hormone analysis laboratory at the International Agency for Research on Cancer (IARC) Hormones and Cancer group. Laboratory samples were blinded to ethnicity and case-control status of the subjects. To reduce the effect of laboratory variability, each analytic batch included equal numbers of cases and controls for each ethnic group. 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Prolactin was measured using a double-antibody, immunoradiometric assay (IRMA) from Diagnostic System Laboratories (Webster, Texas). This assay is a non-competitive assay in which prolactin is sandwiched between two antibodies, the first coated on the walls of the tubes, and the second radiolabeled for detection. The protein present in the samples (and standards) will be bound to the two antibodies, forming a sandwich complex, while the unbound material will be washed away. The amount of radioactivity left in the tubes, read by a gamma counter, is directly proportional to the amount of prolactin present in the samples. A set of standards with known amount of prolactin is used to plot a standard curve from which the unknown amount of prolactin of the samples can be calculated. The theoretic sensitivity of the assay (as stated by the manufacturer) is 0.1 ng/ml. The mean intra-batch coefficient of variation of the method was 5.4% and the mean inter-batch coefficient of variation was 12.8% using a sample volume of 25 microliters. Previously documented findings have shown plasma PRL levels to be stable in whole blood for 24-48 hours.4 6 In the MEC, time from blood collection to processing was estimated to be no more than 6 hours. We used the single-imputation method to estimate the effect of estimated haplotypes on plasma PRL levels using generalized linear models adjusted for continuous (age, anthropometry) and categorical (reproductive history) variables. The hormone measurements were log-transformed to best approximate a normal distribution. These values were transformed back to normal physiologic values for presentation. Means are presented as least-squares means (LS means). For all 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. analyses, a dominant, co-dominant, and recessive model were fitted. Data was analyzed using Statistical Analysis System software (SAS Institute Inc., Cary, North Carolina). 2.3 RESULTS 2.3.1 LD blocks and haplotype structure in the multiethnic panel We assessed genetic variation across 57.9 kb of the PRL locus; 23.6 kb upstream of PRL’s alternative first exon (la) which lies 5.8 kb upstream of pituitary promoter site to 18.7 kb downstream of exon 5. PRLR was assessed from 18.5 kb upstream of the first alternative exon (El 3) to 9.6 kb downstream of exon 11 (i.e. alternatively spliced exon 10). A total of 118 SNPs in PRL and 171 SNPs in PRLR were tested in the multiethnic panel of 349 non-diseased subjects. For PRL, 31 SNPs were monomorphic or had allele frequencies < 5% in all ethnic groups, 18 had poor genotyping results, and 7 were out of Hardy-Weinberg equilibrium in more than one ethnic group, leaving 62 polymorphic ( >5% minor allele frequency in at least one ethnic group) SNPs (Figure 1, Appendix 1). A total of 105 polymorphic SNPs were found in PRLR; 44 SNPs were monomorphic/ low frequency, 17 failed genotyping, and 5 were out of Hardy-Weinberg equilibrium (Figure 2, Appendix 2). We observed three regions of LD in PRL (blocks 1, 3, 4), and one area that showed little evidence of LD despite dense genotyping (~1 SNP every 1 kb) and extensive sequencing in this region (Figure 3). In this region of low LD, we 42 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. Figure 1. The genomic organization of prolactin (PRL) shown 3’ to 5’ (arrows pointing left).; exons are shown as vertical lines. The 62 SNPs used in the haplotype analysis are shown below the defined LD blocks. SNP location is based on the July 2003 freeze of the University of California at Santa Cruz Genome Browser. Conservation with other species and repeat regions are designated below Base P osition B lock-#' B locks « lB10Ck"g Block! | 223900001 224000001 22410000| 22420000] 22430000| 22440000] PRL blocks 62 j 61] 60 | 59 56 58 ] 57| 54] 55 I 4 7 | 53 | 52 I 51 I 50 | 49 | 48 I 46 45] 44! 43 42] 41 I 40 j 39 | 33 | 37 | 36 1 35 PRL SNPs 34 | 33 j 32 | 31 30 | 29 23 27 | 26 25 | 24 | I 23] 22 21 I 20 19] 13 12 1 1 I 13] 5] 17] 16 I 15 I 14] 13] I 6 if @ | 9 8| 7 I 5 3| 2 ] 1 4] PRL Conservation RefSeq Genes Human/Ch imp/Mouse/Rat/Ch1 cken M u ltiz f i l i gnments & Phy1 oH M M Cons chimp 1 mouse ra t i. chicken j RepeatMasker || [ | | | | | J 1 ill I i l l It Repeating II 1 li Elements I I I III Repeattg^gj ^ ■ 1 ■ 11 mi m u Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 2. The genomic organization of prolactin receptor (PRLR) shown 3’ to 5’ (arrows pointing left).; exons are shown as vertical lines. The 105 SNPs used in the haplotype analysis, including the missense (81) SNP in exon 5, are shown below the defined LD blocks. SNP location is based on the July 2003 freeze of the University of California at Santa Cruz Genome Browser. Conservation with other species and repeat regions are designated below. Base P o s itio n B lo ck s B lo cks B 1ock4 B lo ck s B lo ck s B lo c k l 1 0 5 104 1 83 1 82 9 9 j 3 9 | 9 3 | 8 6 | 9 7 | 3 5 | I 101 j 100 | 9 6 95 j 91 | 9 4 | 9 3 | 9 2 j 9 0 | 88 | 8 7 | RGXT2 PRLR 35150000! 353000031 PRLR b lo c k s 352530001 PRLR SNPs 6 4 | 8 3 j 8 £ | 6 6 | 6 3 | 5 5 | 5 0 | 4 6 | 41 | 3 6 1 31 | 2 7 | 1 9 | 9 f 6 | 4 | 8 3 1 6 5 | 5 9 | 5 3 | 4 3 | 4 4 | 4 0 | 3 5 | 2 8 | 18 | 81 51 3 [ 7 9 | 6 4 | 5 3 | 5 2 | 4 7 1 : 4 3 1 3 9 ) 3 4 | 2 6 | 1 7 | 7 ) 2 | 7 8 1 6 3 1 5 7 | 51 | 4 5 | 3 8 | 3 3 | 2 5 ] 16 | 1 7 7 ) 6 2 1 5 6 1 4 9 | 4 2 | 3 7 | 3 2 I 2 4 | 14 1 7 6 | 61 1 5 4 | 3 0 | 2 3 j 13 1 7 5 1 2 9 | 2 2 | 12 | 7 4 | 21 1 11 | 7 3 | 2 0 | 1@| 7 2 1 71 | 78 1 69 | 68 | 87 | 15 | 81 | M issen se s n p Re fS e q Genes H u m an /c n im b /M o u se /R at/C h ick en M u lt iz filig n m e n ts & P h y lohmm Cons C o n s e rv a tio n ch imp mouse r a t 1 c h i cken 1 R ep eatM asker 1 1 III III I I I R e p e a tin g E lem e n ts by R ep eatM asker iii iii i i i i m i innii iii■ will)iii mi i llinium h i ■inn in iiini Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Figure 3. The linkage disequilibrium plot of PRL for all ethnic groups combined. LD strength between the SNPs >10% in PRL, as indicated by the shading scheme, was measured using a combination of the statistic D' and LOD scores. SNP numbers are shown along vertical and diagonal axes. Block 1: SNPs 1-23, “Block” 2: SNPs 24-38, Block 3: SNPs 39-46, Block 4: SNPs 48-58. The area between the arrows represents the coding region of the gene (exons 1-5). 4 ^ U i ■ J B calculated the multivariate squared correlation, Rs , between unmeasured SNPs m relation to all selected htSNPs in PRL. Since for all ethnic groups, this R2 S value was >0.70, this suggested that the unmeasured SNPs in this region were well predicted by the other SNPs in the gene. Based on these findings and substantiated by the dense coverage in this region, we used a haplotype-based analysis to assess common variation and breast cancer risk for this region, as well as single SNP analyses. Therefore, we defined four regions in PRL: block 1 (SNPs 1-23; 14.4 kb), “block” 2 (SNPs 24-38; 19.7 kb), block 3 (SNPs 39-46; 6.6 kb), and block 4 (SNPs 48-58; 8.1 kb). We defined six blocks of LD in PRLR: block 1 (SNPs 3-9; 23.4 kb); block 2 (SNPs 11-28; 20.9 kb); block 3 (SNPs 34-40; 17.4 kb); block 4 (SNPs 42-51; 22.3 kb); block 5 (SNPs 54-74; 35.1 kb); block 6 (SNPs 90-97; 3.4 kb) (Figure 4). African-Americans had smaller block sizes for block 2 (SNPs 18-25; 9.8 kb) and block 5 (SNPs 54-66; 25.1 kb). The 25.1 kb region between block 5 and 6 was poorly characterized until the past two months; few SNPs were available from either the NCBI or Celera databases. We did include four SNPs discovered from the MEC sequencing project in this region (SNPs 80-83), including SNP 81, a missense SNP in exon 5 (IlelOOLeu). These SNPs were rare; only SNP 81 was observed at > 5% minor allele frequency in more than one ethnic group (Appendix 1). We are in the process conducting further genotyping to fully characterize this region before selecting tagSNPs for the breast caner case-control analysis. For this report, results for the missense SNP (SNP 81) in this region are shown in association with breast cancer risk and plasma PRL levels. 46 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. Figure 4. The linkage disequilibrium plot of PRLR for all ethnic groups combined. LD strength between the SNPs SlO% in PRLR, as indicated by the shading scheme, was measured using a combination of the statistic D' and LOD scores. SNP numbers are shown along vertical and diagonal axes. Block 1: SNPs 3-9, Block 2: SNPs 11-28, Block 3: SNPs 34-40, Block 4: SNPs 42-51, Block 5: SNPs 54-74, Block 6: 90-97. The area between the arrows represents the coding region of the gene (exons 3-11). 2.3.2 Breast cancer case-control analysis 2.3.2.1 Subjects The breast cancer case-control panel was made up of 1,716 breast cancer cases and 2,505 controls in the MEC (Table 1). The distribution of breast cancer risk factors among these women were consistent with the patterns observed in the overall cohort.4 7 Breast cancer risk factors were also similar for cases and controls. The mean age for cases was 64.0 years and the mean age for controls was 62.7 years. Cases were slightly more likely to be nulliparous and have had children at a later age. They were also more likely to have a family history of breast cancer and be a current hormone therapy user. 2.3.2.2 Tests of haplotype associations and breast cancer risk Twenty-one haplotype “tag” SNPs (htSNPs) that allowed for high predictability of the common haplotypes in PRL were genotyped in the case-control panel. For PRLR, 23 htSNPs were genotyped, as well as the one missense SNP, IlelOOLeu. The frequency of the common haplotypes ( >5%) predicted by the htSNPs in the multiethnic panel were nearly identical to those observed in the larger sample of cases and controls for both genes (Tables 2-5). We first performed global tests for differences in risk according to the common haplotypes observed (data not shown). Block 2 of PRLR demonstrated marginally significant haplotype-effects (p = 0.04) but this effect was not observed in the other blocks (p =.0.06 for block 1, p = 0.26 for block 3, p = 0.78 for block 4, 48 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. Descriptive characteristics among breast cancer cases (n=1716) and controls (n=2505) in the Multiethnic Cohort Study. African-Americans Hawaiians Japanese Latinas Whites Cases (n= 363) Controls (n= 659) Cases (n= 115) Controls (n= 296) Cases (n= 453) Controls (n= 422) Cases (n= 359) Controls (n= 681) Cases (n= 426) Controls (n= 447) Age (mean) 64.9 64.3 61.1 59.6 64.8 64.2 64.2 62.9 65.0 62.3 Menopausal status (%) Premenopausal 14 11 19 26 12 21 10 11 8 20 Postmenopausal3 56 56 57 53 68 62 63 61 69 61 Simple hysterectomy 18 23 12 14 9 10 16 19 16 14 Missing 12 10 12 6 10 8 11 10 8 5 HT use (%)a > b Never 57 53 52 56 37 45 48 50 32 47 Past 21 20 12 16 12 11 16 16 15 12 Current 17 21 30 27 48 41 28 25 50 40 Age at menarche (%) <12 53 46 53 59 55 51 48 48 55 49 13-14 35 39 30 29 32 35 38 39 35 43 15+ 10 14 13 11 10 14 11 11 8 8 Number of children (%) 0 14 12 10 9 16 11 10 7 19 15 1 19 16 5 9 10 10 8 6 11 9 2 or 3 37 40 43 39 56 60 36 36 50 54 4+ 27 31 42 44 17 18 46 50 19 21 Age at first birth (%)b ,c <20 38 45 39 37 8 10 35 39 19 20 21-30 39 35 45 45 62 67 46 47 52 54 31+ 6 5 2 6 11 10 7 4 8 10 First degree family history of breast cancer (%)b Yes 22 13 19 15 19 11 17 12 17 9 No 78 87 81 85 81 89 83 89 83 91 Alcohol consumption: drinks/day (%)b 0 52 53 64 56 70 75 55 52 30 40 <1 29 30 20 30 21 17 31 34 39 40 a 10 10 10 9 4 3 6 5 24 18 a Women reporting natural menopause or having had a bilateral oophorectomy. ' ’ Numbers do not add up to 100% because of missing data. c Among parous women. V O Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 2. Common haplotypes ( >5%) in blocks 1 -4 of prolactin (PRL), estimated using all SNPs and the htSNPs among all groups combined. All SNPs Block 1 (SNPs 1-23) (%) “Block” 2 (SNPs 24-38) (%) Block 3 (SNPs 39-46) (%) Block 4 (SNPs 48-58) (%) laCGTGGCCGTGCAATCCGCTCCAG 0.31 2a CCATGCGGACTTTGG 0.34 3 a AGCTTCCG 0.44 4a GCTCTTAGATC 0.32 lbCGTGGCCGCGCAATCCGCTCCAG 0.23 2b CCATGCTGACTTTGG 0.12 3b AGCTCTCA 0.13 4b GCTCTTGGATC 0.14 lc CGTGGTCGCGCAATCCGCTCCAG 0.20 2c TCATGCGGACTTTGG 0.11 3c GGCTCCTG 0.10 4c ACTCCTAAAAC 0.12 Id CGCAGCCACGCAACCCCCTCCAA 0.06 2d TTATAAGAACCATGG 0.08 3d AACCCCCA 0.08 4d ACTCTTAAAAC 0.12 2e TTATACGGAGCATGG 0.07 3eAGCTCTCG 0.07 4e ACTCTCAAATC 0.10 3f AGCTCCCA 0.05 4 f GTAGCTAAAAC 0.07 htSNPs Block 1 “Block” 2 Block 3 Block 4 (htSNPs 1,2,4, 6, 9, 23) (%) (htSNPs 24,30, 32,33,35) (%) (htSNPs 41,42, (%) (htSNPs 49, 52, (%) 43,44,46) 54, 55, 57) laCGGCTG 0.31 2a CGACT 0.34 3a CTTCG 0.44 4a CTAGT 0.33 lb CGGCCG 0.26 2b TGACT 0.18 3b CTCTA 0.13 4b CTGGT 0.14 lc CGGTCG 0.20 2c CTACT 0.12 3c CTCCG 0.12 4c CCAAA 0.12 Id CG ACC A 0.07 2d TGACA 0.10 3d CCCCA 0.08 4d CTAAA 0.12 leCTACCA 0.06 2e TGAGA 0.08 3eCTCTG 0.07 4e CTAAT 0.11 lfTTACCA 0.05 2fTGGCA 0.05 3f CTCCA 0.05 4f TCAAA 0.09 2g CGACA 0.05 4g TCAAT 0.07 e /i o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3. Common haplotypes (>5%) in blocks 1-6 of prolactin receptor (PRLR), estimated using all SNPs and the htSNPs among all groups combined. All SNPs Block 1 (SNPs 3-9) (%) Block 2 (SNPs 11-28) (%) Block 3 (SNPs 34-40) (%) Block 4 (SNPs 42-51) (%) laGTGTGGT 0.30 2a CGCTGCTTTATGCTCACG 0.43 3a TGTTGCG 0.68 4a CTTTTCTCTA 0.75 lb ATGGACC 0.24 2b TATCGCTACGCATTAAGA 0.25 3b CCCCGTT 0.14 4b GCCCCTGCCG 0.11 lc GTGGACC 0.23 2c CGCTGTTACGCACCAGCG 0.09 3c TGTCATT 0.10 4c GCCTTTGTCA 0.07 Id GCGTGGC 0.07 2d CGCTGCTTTGCACCAGCG 0.05 3d CGTCGTT 0.05 le GTGTGGC 0.05 htSNPs Block 1 Block 2 Block 3 Block 4 (htSNPs 3,4, 5, 6, 9) (%) (htSNPs 18,20, 24,26) (%) (htSNPs 35, 38,39) (%) (htSNPs 42,45,49) (%) laGTTT 0.30 2aTATA 0.44 3aGGC 0.70 4a CTC 0.77 lb ATGC 0.24 2b AGTA 0.26 3b CGT 0.14 4b GCC 0.11 lcGTGC 0.22 2c AGCG 0.10 3c GAT 0.08 4c GTT 0.07 ldGTTC 0.09 2d TGCG 0.07 3d GGT 0.07 leGCTC 0.07 2e TGCA 0.06 All SNPs Block 5 (SNPs 54-74) (%) Block 6 (SNPs 90-97) (%) 5aTCGTGGACGCGGCGATGATCA 0.46 6a CTTACCG 0.71 5b TCGTGAACGCAGCGATGACCA 0.25 6b TCCGTTA 0.19 5cCGACAGGTATGATAGCGGTTG 0.09 6c TTTACCG 0.05 5dTGGCAGACGCGGCAGCGGTTG 0.07 htSNPs Block 5 Block 6 (htSNPs 54, 57, 59, 60, 68) (%) (htSNPs 90, 92, 93, 97) (%) 5 a TTGAA 0.47 6a CTG 0.71 5b TTAAA 0.25 6b TCA 0.19 5c CCGGG 0.10 6c TTG 0.05 5d TCGAG 0.09 Table 4. Common haplotypes in blocks 1-4 of prolactin (PRL) among African-Americans, Hawaiians, Japanese, Latinas, and Whites in the multiethnic panel.3 Haplotypesb Haplotype frequencies in African- Hawaiians Americans the multiethnic panel (%) Japanese Latinas Whites Block 1 (SNPs 1-23) htSNPs: 1,2,4, 6,9, 23 la CGGCTG 16 40 33 28 38 lb CGGCCG 28 13 9 38 41 lc CGGTCG 28 23 29 12 10 Id CGACCA 6 15 10 le CTACCA 9 7 7 If TTACCA 12 6 lg CGGCCA 6 Total" 99 100 98 96 100 r \ 0.97 0.98 1.00 0.97 1.00 “Block” 2 (SNPs 24-38) htSNPs: 24, 30, 32, 33, 35 2a CGACT 20 39 37 32 43 2b TGACT 29 15 25 19 2c CTACT 17 24 10 2d TGACA 11 8 13 12 2e TGAGA 6 24 11 2f TGGCA 9 9 2g CGACA 17 2h CGGCA 9 2i CGAGA 7 2j TTACT 6 Total" 99 100 99 100 99 R \ 0.77 0.84 0.96 0.93 0.87 Block 3 (SNPs 39-46) htSNPs: 41, 42, 43, 44,46 3a CTTCG 21 47 65 39 45 3b CTCTA 24 17 12 11 3c CTCCG 16 25 18 3d CCCCA 10 5 10 17 3e CTCTG 7 8 11 3f CTCCA 21 3g CTTCA 5 7 5 3h CCCCG 7 6 3i TTCTA 5 Total" 100 99 99 99 98 R \ 0.83 0.91 1.00 0.82 0.84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4 continued. Block 4 (SNPs 48-58) htSNPs: 49, 52, 54, 55, 57 4a CTAGT 29 26 36 33 38 4b CTGGT 17 18 18 14 4c CCAAA 12 31 16 4d CTAAA 29 8 8 6 7 4e CTAAT 7 9 13 27 4f TCAAA 6 9 7 11 1 1 4g TCAAT 14 19 Total0 100 100 100 98 99 R \ 0.87 0.93 1.00 0.97 0.98 “ Haplotypes observed with >5 % frequency in at least one ethnic group in the multiethnic panel. b Haplotype order is based on the frequency as predicted by the htSNPs among all groups combined. °The percentage of all chromosomes accounted for by the common haplotypes. d The R \ that is given is the minimum R2 h of the common haplotypes in each ethnic group. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5. Common haplotypes in blocks 1-6 of prolactin receptor (PRLR) among African- Americans, Hawaiians, Japanese, Latinas, and Whites in the multiethnic panel.3 Haplotype frequencies in the multiethnic panel (%) Haplotypesb African- Hawaiians Japanese Latinas Whites Americans Block 1 (SNPs 3-9) htSNPs: 3, 4, 5, 6, 9 la GTGTT 10 49 51 22 13 lb ATGGC 23 21 23 18 34 lc GTGGC 12 18 7 40 36 Id GCGTC 11 7 7 7 le GTGTC 25 5 If GTATC 5 7 lg GTGGT 7 lh GCGTT 11 Total0 100 99 98 99 100 R \ 0.87 0.98 0.98 0.95 0.99 Block 2 (SNPs 11-28) htSNPs: 18, 20, 24, 26 2a TATA 37 38 25 56 65 2b AGTA 4 42 61 15 9 2c AGCG 26 5 6 7 8 2d TGCG 8 7 6 14 2e TGCA 8 7 10 2f AGCA 5 5 2g TATG 5 5 2h TGTA 6 Total0 100 98 99 99 99 R \ 0.83 0.99 0.98 0.94 1.00 Block 3 (SNPs 34-40) htSNPs: 35, 38, 39 3a GGC 63 61 87 71 69 3b CGT 23 14 12 11 11 3c GAT 11 10 15 3d GGT 14 14 5 Total0 99 100 99 100 100 R2 h 1.00 0.98 1.00 0.99 1.00 Block 4 (SNPs 42-51) htSNPs: 42, 45,49 4a CTC 52 72 77 91 92 4b GCC 17 10 18 5 6 4c GTT 14 9 5 4d GTC 17 4e CTT 7 Total0 100 98 100 99 99 R \ 1.00 0.98 1.00 1.00 1.00 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 5 continued. Block 5 (SNPs 54-74) htSNPs: 54, 57, 59, 60, 68 5a TTGAA 22 51 71 59 35 5b TTAAA 15 21 29 54 5c CCGGG 14 10 18 5d TCGAG 15 17 6 5e CCGAG 10 5 5f TCGGG 9 5g TTAAG 5 Total0 98 99 99 98 98 R \ 0.88 1.00 1.00 1.00 1.00 Block 6 (SNPs 90-97) htSNPs: 90, 92, 93, 97 6a CGTG 34 72 67 86 79 6b TGCA 31 22 29 6 6 6c TGCG 9 6d TATG 17 6e CATG 14 Total0 98 99 99 99 100 R \ 0.97 1.00 1.00 1.00 0.99 “ Haplotypes observed with >5 % frequency in at least one ethnic group in the multiethnic panel. b Haplotype order is based on the frequency as predicted by the htSNPs among all groups combined. T he percentage of all chromosomes accounted for by the common haplotypes. d The R \ that is given is the minimum R2 h of the common haplotypes in each ethnic group. 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. p - 0.19 for block 5, and p = 0.34 for block 6). None of the blocks in PRL showed significant haplotype-effects: block 1 (p - 0.21), “block” 2 (p = 0.45), block 3 (p = 0.12), and block 4 (p = 0.08). Within each block, some ethnic groups demonstrated a positive association with certain haplotypes and breast cancer risk (Tables 6 and 7). However, none of these associations were significant when assessed among all groups for which the haplotype was observed at >5%. (The following haplotypes were associated with risk in PRL in at least one ethnic group: 3f (African-Americans), 3a (Japanese), 4d (Japanese), and for PRLR: lb (Hawaiians, Whites), 3d (African-Americans), 5b (African-Americans), and 6e (Whites). Within block 2 of PRLR, the global test was significant (p = 0.044). Individual haplotypes in this block were not associated with risk for each ethnic group or among all groups combined. When we restricted the analysis to only those cases with advanced disease (n = 436), the global test for block 2 was no longer significant (p = 0.27). 2.3.3 Plasma PRL level analysis Among the 355 control women included in the analysis of plasma PRL levels, the median plasma PRL level was 8.1 ng/mL (10th -90th percentile range = 3.42-21.2 ng/mL). Mean prolactin levels did not vary by ethnic group (Table 8). Results were similar when we adjusted for potential confounders of PRL levels: age at first pregnancy, body mass index, family history of breast cancer, and 56 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 6. Association between haplotypes in LD blocks 1-4 of prolactin (PRL) and breast cancer risk. Haplotypes3 Block 1 Haplotype frequencies African-Americans Cases Controls (n= 363) (n= 659) ORb (95% Cl) Hawaiians Cases (n= 115) Controls (n= 296) ORb (95% Cl) Japanese Cases (n= 453) Controls (n= 422) ORb (95% Cl) laCGGCCG 29.8 30.4 0.97 (0.79,1.19) 16.7 14.5 1.25 (0.80,1.94) 9.3 9.5 0.96 (0.69, 1.35) IbCGGTCG 27.5 28.7 0.94 (0.77, 1.16) 20.3 24.5 0.75 (0.51,1.13) 30.4 32.4 0.92 (0.75,1.13) lc CTGCCA 7.7 6.1 1.31 (0.91,1.88) Id CTACCG 6.8 5.5 1.22 (0.85,1.77) leCTACCA 6.2 5.6 1.12 (0.75,1.68) 12.9 15.7 0.78 (0.49, 1.23) 21.8 18.1 1.27 (1.00,1.60) If TT ACC A 8.6 11.1 0.73 (0.53, 1.01) lgCGGCTG 12.1 11.8 1.03 (0.77, 1.38) 42.3 39.8 1.10 (0.80,1.50) 32.1 34.2 0.90 (0.73,1.01) Haplotypes3 Latinas Whites All groups with haplotypes ^5% Block 1 Cases Controls ORb (95% Cl) Cases Controls ORb (95% Cl) ORc (95% Cl) (n= 359) (n= 681) (n= 426) (n= 447) laCGGCCG 29.4 28.9 1.02 (0.83,1.24) 38.1 37.1 1.06 (0.86,1.30) 1.02 (0.92,1.14) lb CGGTCG 15.7 14.0 1.17 (0.90, 1.52) 10.6 8.8 1.23 (0.88,1.71) 0.99 (0.88,1.10) lc CTGCCA 1.31 (0.91,1.88) Id CTACCG 1.22 (0.85, 1.77) leCTACCA 17.8 20.9 0.81 (0.64,1.03) 1.00 (0.86, 1.15) If TT ACC A 6.6 6.9 0.94 (0.64, 1.38) 0.81 (0.63,1.03) lgCGGCTG 31.5 31.9 0.99 (0.82,1.20) 38.6 40.4 0.93 (0.76,1.13) 0.96 (0.87,1.07) -j T able 6 continued. u n « a - o' o « ,— & s L A o B u 0s < o Os & O C j I X o u g o c o J 2 C Q x r C O « o 0 4 r — . r - H ' '3 ' N O 0 4 t > O n O O d d d s ^ ’ 0 4 0 4 o O n 0 4 q d rJ O N _ 0 4 o o O N v o 0 4 C O O N K 0 4 v o 0 4 0 4 C O ■ s C “ ' v ^ fs os o - 00 c o ' 0 0 C O0 0 0 4 V O r - H — H T - H 0 4 0 4 > — < O N V O0 4 O Oc o 0 0 c o v o v o C " d d d d d d v — v o N O• ^ t V O O N 0 0 o o > - h o r * H q d d o 0 4 V Oo 0 4 C O d S in d V O d 0 4 N O 0 0 V O « d d N Od d 0 4 1 — . t - h © c o " V O * 0 C O ’ V t C O 0 4 0 4 t — T - H ^ X -J X ’ 't n o ' O 0 0 O O O O G O N O r- d d d, © , o, d W o o O - H 0 0 N O O ' q r- O N O N d , - H - H d d t" r-H l> O N v o d V O d d 0 0 c o o o O \o o o 04 r — 0 0 o 04 U O O ^ O o O o S n o u u o h Hh h ^ c a JP u -o < 0 U h b o js — — , NNNNNNNNNN r a Pi < O o Pi o m I I «T o 1 L O £ n o in ^ 4 ) w | | « 1 u £ u \= 0 s v o on & O p o o ■ fc V O c „ o II o 3 o & r o & ® w r vOO\'OV)^NO'toOOO O O ' O C ' l ' O O O O T j - O O O O Q O I O V O O O O O C \ « J « 0 \ 0 0 o o o o o o o o o o ^ooooomwiin'oiM' - H o o v o v o y o o o o ’t ^ ^ ^ d o d ^ ^ d ^ d r-V 0 4 v o0 4 o O C OC O T — 04 04 T - HT -H T -l - r - H 0 4 r — < r- V O O N 0 0 0 0 r- ■ * 3 * d d d d d O N o r- 04 q q o - O N d d V O q v o d V O d 04 rf V O 04 C O O N C O d t - h V O d 04 ^ r / — • N V ___ O N04 C OO ' v o O ' T - HO ' 0 0 04 04 d V OV O C OT — C O04 04 o i X T - i X X C O04 T - i C O04 c o O ' V Oo o V O O ' r- O ' N O V OOs O NO ' d d d d o d. o, d d o. Ww 'w - w O N0 0 N O0 0 0 0 C O04 O NOn 04 04 q C Oq 0 0 q 0 0 O'; On 04 T - * d d d v t v o C Ov o q r - H 0 0 04 04 v o V O I"1 T — r — C O l> y o O O fO IT ] K K d o t> c o o g u Q u o o UOO^COnUO^rJ O O O O h P H ^ i j a jp ox* « ih eo.fi ■ « (NNNMNMNNNCN 58 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 6 continued. Haplotypes2 Block 3 Haplotype frequencies African-Americans Cases Controls (n= 363) (n= 659) ORb (95% Cl) Hawaiians Cases (n= 115) Controls (n= 296) ORb (95% Cl) Japanese Cases (n= 453) Controls (n= 422) ORb (95% Cl) 3a CTCCG 18.0 16.9 1.09 (0.7, 1.68) 25.3 20.3 1.38 (1.09,1.74) 3b CTCCA 16.3 17.9 0.88 (0.68, 1.14) 3c CTCTG 5.0 5.4 0.92 (0.57,1.47) 5.7 6.9 0.78 (0.52,1.17) 3d CTCTA 27.5 26.9 1.03 (0.83,1.28) 17.8 18.5 0.99 (0.67,1.47) 3e CCCCG 6.5 5.7 1.20 (0.79, 1.83) 3f CCCCA 18.0 14.8 1.31 (1.01,1.69) 3gTTCTA 3h CTTCG 20.6 21.9 0.92 (0.73,1.16) 48.9 49.8 0.93 (0.67,1.29) 66.0 69.7 0.83 (0.67,1.03) Haplotypes3 Latinas Whites All groups with haplotypes >5% Block 3 Cases Controls ORb (95% Cl) Cases Controls ORb (95% Cl) ORc (95% Cl) (n= 359) (n= 681) (n= 426) (n= 447) 3a CTCCG 13.5 16.4 0.78 (0.59,1.02) 1.07 (0.91,1.26) 3b CTCCA 0.88 (0.68,1.14) 3c CTCTG 6.2 5.4 1.19 (0.78,1.82) 11.2 10.7 1.03 (0.74, 1.43) 0.97 (0.80,1.19) 3d CTCTA 10.6 11.0 0.94 (0.69, 1.28) 9.6 10.6 0.86 (0.62,1.21) 0.97 (0.84,1.12) 3e CCCCG 6.3 4.8 1.41 (0.89,2.22) 1.29 (0.95,1.76) 3f CCCCA 12.6 12.4 1.02 (0.76,1.35) 14.8 17.2 0.84 (0.64,1.10) 1.05 (0.90, 1.22) 3gTTCTA 4.0 5.0 0.78 (0.48,1.27) 0.78 (0.48,1.27) 3h CTTCG 42.3 42.6 1.00 (0.83,1.20) 44.2 42.6 1.09 (0.89,1.33) 0.96 (0.87,1.06) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 6 continued. Haplotypes® Block 4 Haplotype frequencies African-Americans Cases Controls (n= 363) (n= 659) ORb (95% Cl) Hawaiians Cases (n= 115) Controls (n= 296) ORb (95% Cl) Japanese Cases (n= 453) Controls (n= 422) ORb (95% Cl) 4a CTAAT 14.0 11.3 1.34 (1.00,1.80) 8.3 7.7 1.08 (0.59,1.98) 0.1 0.5 0.24 (0.03,2.13) 4b CTAAA 17.7 19.4 0.89 (0.69,1.13) 4.9 5.3 0.87 (0.39,1.94) 5.0 6.5 0.73 (0.48,1.11) 4c CTGGT 26.1 20.2 1.31 (0.92,1.87) 28.7 27.1 1.08 (0.87, 1.33) 4d CCAAA 10.6 14.6 0.74 (0.46,1.19) 33.0 28.5 1.23 (1.01,1.51) 4e TCAAT 19.8 18.1 1.14 (0.89,1.46) 12.1 16.0 0.74 (0.46,1.17) 4f TCAAA 8.4 9.9 0.80 (0.57,1.14) 9.9 11.3 0.90 (0.54, 1.52) 6.9 6.0 1.15 (0.79,1.68) 4g CTAGT 30.7 31.0 0.98 (0.81,1.20) 25.4 23.6 1.10 (0.76,1.60) 25.0 30.6 0.75 (0.61,0.93) Haplotypes® Latinas Whites All groups with haplotypes S5% Block 4 Cases Controls ORb (95% Cl) Cases Controls ORb (95% Cl) ORc (95% Cl) (n= 359) (n= 681) (n= 426) (n= 447) 4a CTAAT 16.6 16.3 1.04 (0.81,1.33) 24.3 24.7 0.98 (0.78, 1.24) 1.07 (0.93, 1.24) 4b CTAAA 5.2 5.8 0.92 (0.60,1.42) 0.86 (0.71,1.03) 4c CTGGT 17.9 14.9 1.26 (0.99,1.61) 11.6 11.6 0.98 (0.73,1.31) 1.14 (1.00, 1.29) 4d CCAAA 14.6 18.2 0.75 (0.59, 0.97) 0.98 (0.85,1.14) 4e TCAAT 1.02 (0.83, 1.27) 4f TCAAA 9.9 9.7 1.01 (0.73, 1.38) 11.0 10.6 1.03 (0.76, 1.39) 0.98 (0.83,1.14) 4g CTAGT 31.5 32.4 0.96 (0.79,1.17) 40.7 40.0 1.03 (0.85,1.26) 0.94 (0.86, 1.04) “ Haplotypes observed > 5 % frequency among controls in at least one ethnic group are shown. b ORs are estimated using unconditional logistic regression adjusted for age. C A11 groups with haplotypes >5% frequency among controls; ORs are estimated using unconditional logistic regression adjusted for age and ethnicity. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 7. Association between haplotypes in LD blocks 1-6 o f prolactin receptor (PRLR) and breast cancer risk. Haplotypes2 Block 1 Haplotype frequencies African-Americans Cases Controls (n= 363) (n= 659) ORb (95% Cl) Hawaiians Cases (n= 115) Controls (n= 296) ORb (95% Cl) Japanese Cases (n= 453) Controls (n= 422) ORb (95% Cl) laGTCTC 30.1 26.2 1.24 (1.00,1.54) lb GTCTT 8.8 10.0 0.84 (0.59, 1.19) 55.9 45.7 1.52 (1.11,2.09) 49.7 51.1 0.95 (0.79,1.14) lcGCCTC 9.5 10.7 0.85 (0.61,1.19) 4.1 5.7 0.70 (0.32, 1.52) 11.3 10.3 1.11 (0.82, 1.49) Id GCCTT 10.1 12.7 0.75 (0.55,1.02) le ATCGC 21.5 20.6 1.06 (0.84,1.33) 10.8 16.3 0.60 (0.36,0.98) 24.2 20.9 1.22 (0.97,1.54) lfGTCGC 14.1 13.8 1.03 (0.78, 1.36) 21.1 25.6 0.76 (0.52,1.11) 6.4 9.8 0.61 (0.42,0.88) Haplotypes2 Latinas Whites All groups with haplotypes >5% Block 1 Cases Controls ORb (95% Cl) Cases Controls ORb (95% Cl) ORc (95% Cl) (n= 359) (n= 681) (n= 426) (n= 447) laGTCTC 4.0 5.6 0.67 (0.42,1.07) 1.11 (0.92,1.34) lb GTCTT 26.5 26.5 1.01 (0.81,1.25) 21.0 14.8 1.56 (1.21,2.02) 1.10 (0.99,1.23) lcGCCTC 0.96 (0.78, 1.18) Id GCCTT 0.75 (0.55, 1.02) le ATCGC 25.2 24.7 1.01 (0.82,1.26) 31.6 34.2 0.89 (0.72,1.09) 1.00 (0.90,1.11) lfGTCGC 31.4 30.1 1.07 (0.87,1.32) 35.8 36.8 0.96 (0.78,1.17) 0.94 (0.83, 1.05) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 7 continued. Haplotypes8 Block 2 Haplotype frequencies African-Americans Cases Controls (n= 363) (n= 659) ORb (95% Cl) Hawaiians Cases (n= 115) Controls (n= 296) ORb (95% Cl) Japanese Cases (n= 453) Controls (n= 422) ORb (95% Cl) 2a TGTA 8.3 6.9 1.26 (0.88,1.82) 2b TGCA 7.7 8.6 0.87 (0.61,1.23) 9.3 7.2 1.42 (0.79,2.56) 2c TGCG 8.7 10.0 0.84 (0.60, 1.17) 8.5 5.0 1.79 (0.98, 3.26) 2d AGTA 5.1 5.2 0.96 (0.61,1.50) 43.9 39.3 1.20 (0.88,1.63) 55.3 55.9 0.97 (0.80,1.18) 2e AGCA 5.4 4.8 1.18 (0.74,1.87) 2f AGCG 18.6 22.1 0.78 (0.62,1.00) 3.7 5.5 0.61 (0.27,1.39) 7.2 7.3 1.00 (0.69,1.45) 2g TATA 43.5 39.5 1.19 (0.99,1.44) 33.0 41.3 0.72 (0.53,0.98) 28.5 27.0 1.08 (0.87,1.33) Haplotypes8 Latinas Whites All groups with haplotypes >5% Block 2 Cases Controls ORb (95% Cl) Cases Controls ORb (95% Cl) ORc (95% Cl) (n= 359) (n= 681) (n= 426) (n= 447) 2aTGTA 1.26 (0.88,1.82) 2b TGCA 5.7 6.6 0.87 (0.59,1.29) 0.94 (0.74,1.19) 2c TGCG 10.4 9.7 1.09 (0.80, 1.48) 15.5 12.1 1.35 (1.02,1.80) 1.14 (0.96, 1.35) 2d AGTA 19.2 19.9 0.97 (0.77, 1.22) 9.4 8.6 1.10 (0.78, 1.53) 1.02 (0.90, 1.14) 2e AGCA 1.18 (0.74,1.87) 2f AGCG 6.9 6.2 1.11 (0.77, 1.61) 7.5 8.1 0.92 (0.64,1.31) 0.89 (0.76,1.04) 2% TATA 56.9 56.2 1.02 (0.85,1.23) 64.9 67.3 0.90 (0.74,1.01) 1.01 (0.92,1.11) Table 7 continued. u n © 0 s- t o © e r f O tfl 8 2 «T o L O 3 < u t z > Q § Q . c S ■ ® ? U CO « > 3 C J U 0 s - & O C3 C /5 _ _ 5 II «J cd JL s q £ u 0 s - C r f O s s rs 3 O $ CO r<s $ l 3 < U s § 0 1 a c o o o 5 O s CO « CO T -H T — ^ OS r - CO © © OS O n © © v S T-H T -H 0 0 'O t - H ^ OS t -h 00 O rfr O S V O v o v n c o o s 0 O v o oo" o o v o t o v o © © © © OS • o © o o OS ON HH r-H © © OS v q © l > o d 2 OS © •"t r*; o s O - o s © T-H T -H T -H NO I T ) 00 i- O n O t o © ^ vo" oo" CO ■'tf N© © © © ~ v o v o t o o o VO OS © © 0 0 v q 0 0 o s v d NO © OS VO O n H H H U o < o o O O u O C I S 4 3 O T 3 CO CO CO c o O 9-U c d _ ^ £ x .2 "O .IS CS £ W C /5 & o 33 Q C C O u a o O cd U U -s® 0 s- » o O n e r f O C O '— - ’p 3 ■P v o G i| o J L u £ C /5 < U J§^ O ^ §" J a m M /-N't t^ - M O - ' v o v o O t o v o 0 0 OO ON © © © © © o s NO 0 0 © © c q © r-H © _ _ _ /~V CO •^t CO o s o c q HH T-H r— H r-H r-H o o o o " 't "St ON r - O* © , © , © , © s "“/ O ' © o s O ' CO © © © hh- © © OS T-H v q tr i o s © o s O ' © © 'd 1 t o © © O ' tr> O - O ^ O -sf CO i> o r-H o- ' f t OO h - o o © © © o r - f OO ON o o O ' © © © C O o o • v o K O n o d ^ H H H C J o < o 5 O O q O 0 3 r O cj T 3 c o c o c o c o 63 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 7 continued. Haplotypes3 Block 4 Haplotype frequencies African-Americans Cases Controls (n= 363) (n= 659) ORb (95% Cl) Hawaiians Cases (n= 115) Controls (n= 296) ORb (95% Cl) Japanese Cases (n= 453) Controls (n= 422) ORb (95% Cl) 4aCTT 7.8 7.7 0.99 (0.56, 1.76) 4b GTC 18.5 18.0 1.04 (0.82, 1.32) 4c GTT 9.5 10.4 0.91 (0.67, 1.23) 7.2 6.2 1.21 (0.65, 2.28) 4d GCC 14.7 16.2 0.89 (0.69, 1.14) 10.9 7.5 1.45 (0.87,2.41) 16.7 16.7 1.00 (0.78, 1.30) 4e CTC 57.2 55.2 1.08 (0.90,1.29) 73.0 77.3 0.81 (0.57,1.15) 78.3 78.8 0.97 (0.77,1.22) Haplo types3 Latinas Whites All groups with haplotypes >5% Block 4 Cases Controls ORb (95% Cl) Cases Controls ORb (95% Cl) OR0 (95% Cl) (n= 359) (n= 681) (n= 426) (n= 447) 4aCTT 0.99 (0.56, 1.76) 4b GTC 1.04 (0.82,1.32) 4c GTT 0.96 (0.73,1.26) 4d GCC 4.7 4.7 0.99 (0.64, 1.52) 0.99 (0.84, 1.16) 4e CTC 91.0 89.8 1.15 (0.84, 1.58) 90.7 92.5 0.82 (0.59, 1.13) 1.00 (0.89,1.12) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 7 continued. Haplotypes'1 Block 5 Haplotype frequencies African-Americans Cases Controls (n= 363) (n= 659) ORb (95% Cl) Hawaiian s Cases (n= 115) Controls (n= 296) ORb (95% Cl) Japanese Cases (n= 453) Controls (n= 422) ORb (95% Cl) 5a TTGAA 27.0 26.5 1.03 (0.83,1.26) 53.3 57.2 0.86 (0.63, 1.17) 66.1 68.6 0.89 (0.72,1.09) 5b TCGAG 21.3 17.3 1.31 (1.04, 1.66) 14.9 14.1 1.07 (0.69, 1.65) 5c TCGGG 8.0 7.9 1.02 (0.71,1.47) 5dCCGAG 10.5 10.5 0.99 (0.72,1.36) 5e CCGGG 12.6 15.2 0.79 (0.59,1.04) 10.1 7.5 1.35 (0.80,2.29) 18.6 17.5 1.07 (0.84, 1.38) 5f TTAAA 14.8 15.8 0.93 (0.72,1.19) 19.1 18.1 1.05 (0.71,1.55) Haplotypes2 Latinas Whites All groups with haplotypes >5% Block 5 Cases Controls ORb (95% Cl) Cases Controls ORb (95% Cl) OR0 (95% Cl) (n= 359) (n= 681) (n= 426) (n= 447) 5a TTGAA 57.1 56.3 1.03 (0.85,1.24) 36.4 37.5 0.96 (0.79,1.18) 0.97 (0.88,1.06) 5b TCGAG 2.7 4.9 0.52 (0.31,0.89) 1.10 (0.91,1.33) 5c TCGGG 1.02 (0.71, 1.47) 5d CCGAG 0.99 (0.72,1.36) 5e CCGGG 0.97 (0.82, 1.16) 5f TTAAA 31.1 30.8 1.02 (0.84,1.25) 52.5 50.7 1.08 (0.89,1.31) 1.03 (0.91,1.15) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 7 continued. Haplotypes3 Block 6 Haplotype frequencies African-Americans Cases Controls (n= 363) (n= 659) ORb (95% Cl) Hawaiians Cases (n= 115) Controls (n= 296) ORb (95% Cl) Japanese Cases (n= 453) Controls (n= 422) ORb (95% Cl) 6a CATG 11.9 12.7 0.92 (0.69, 1.23) 6b TGCG 6c TGCA 31.6 32.6 0.95 (0.78,1.16) 26.7 24.0 1.15 (0.80, 1.65) 26.3 25.8 1.03 (0.83,1.26) 6d TATG 12.2 12.7 0.96 (0.72,1.27) 6e CGTG 37.0 35.7 1.06 (0.88, 1.27) 68.9 71.2 0.91 (0.64, 1.29) 69.5 71.3 0.92 (0.75,1.13) Haplotypes3 Latinas Whites All groups with haplotypes >5% Block 6 Cases Controls ORb (95% Cl) Cases Controls ORb (95% Cl) OR0 (95% Cl) (n= 359) (n= 681) (n= 426) (n= 447) 6a CATG 0.92 (0.69,1.23) 6b TGCG 5.5 5.2 1.05 (0.70,1.57) 5.9 6.6 0.88 (0.59,1.32) 0.96 (0.72,1.28) 6c TGCA 6.7 6.7 0.97 (0.68,1.40) 5.6 5.6 0.94 (0.62,1.42) 1.00 (0.88,1.12) 6d TATG 0.96 (0.72,1.27) 6e CGTG 84.1 85.0 0.94 (0.73,1.20) 86.0 82.6 1.33 (1.03,1.73) 1.02 (0.92, 1.13) “ Haplotypes observed >5% frequency among controls in at least one ethnic group are shown. b ORs are estimated using unconditional logistic regression adjusted for age. C A11 groups with haplotypes >5% frequency among controls; ORs are estimated using unconditional logistic regression adjusted for age and ethnicity. O n a s Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 8. Least squares means of plasma prolactin (PRL) levels by ethnicity among postmenopausal women without breast cancer and not using hormone therapy. Least square means of plasma PRL levels African-Americans Hawaiians (n = 78) 95% Cl (n = 6 0 ) a 95%CI Japanese (n = 69) 95%CI Latinas (n = 85) 95%CI Whites (n = 63) 95%CI Pb Age, adjusted 9.1 (7.7,10.7) 8.5 (7.0, 10.3) 7.1 (6.0, 8.5) 9.3 (8.0, 10.9) 7.9 (6.6, 9.4) 0.08 Age, parity adjusted 9.3 (6.9,12.5) 8.8 (6.3, 12.2) 7.3 (5.3, 10.0) 9.6 (7.1, 12.9) 8.1 (5.8, 11.1) 0.09 Age, age at first pregnancy adjusted 9.0 (7.4, 10.9) 8.4 (6.7, 10.5) 6.9 (5.6, 8.6) 9.1 (7.5, 11.0) 7.6 (6.2, 9.3) 0.09 Age, body mass index adjusted 9.0 (7.6,10.7) 8.6 (7.1, 10.4) 7.2 (6.0, 8.7) 9.2 (7.8, 10.8) 7.9 (6.6, 9.4) 0.07 Age, estradiol adjusted 9.2 (7.8, 10.8) 8.5 (7.0, 10.3) 7.0 (5.9, 8.4) 9.3 (8.0, 10.9) 7.9 (6.6, 9.4) 0.08 Age, estrone adjusted 9.1 (7.8, 10.8) 8.5 (7.0, 10.3) 7.0 (5.9, 8.4) 9.4 (8.1, 11.0) 7.9 (6.6, 9.4) 0.16 Age, family history o f breast cancer adjusted 9.5 (7.9,11.5) 8.1 (6.6, 10.0) 7.2 (5.9, 8.7) 9.1 (7.6, 10.9) 8.0 (6.5, 9.9) 0.11 Age, menopause age and type adjusted 9.2 (7.7,11.1) 8.3 (6.7, 10.3) 6.9 (5.6, 8.4) 9.5 (8.0, 11.2) 7.8 (6.4, 9.5) 0.10 a All estimates are adjusted for assay batch. b p-value for test of heterogeneity. O S menopause age and type. Additional adjustment for parity or estradiol and estrone did not change the risk estimates. Overall, there was no evidence of an association between any of the haplotypes in PRL or PRLR and plasma PRL levels (Tables 9 and 10). Tests for trend were significant for haplotype 4e in PRL (p = 0.03) and for haplotype 2b in PRLR (p = 0.03); however, both of these estimates were based on just three individuals who had two copies of the haplotype. 2.3.4 Single SNP analyses Single SNP analyses of the htSNPs in the region of low LD of PRL demonstrated no association with breast cancer risk (Table 1 la). We observed a statistically significant association between SNP 35 and plasma PRL levels (p = 0.0005) (Table lib). The IlelOOLeu missense SNP not associated with breast cancer risk or circulating PRL levels (Table 12a and b). The SNP was observed in all ethnic groups, but at low frequency; only African-Americans and Hawaiians had a minor allele frequency >10%. 2.4 DISCUSSION We genotyped a high density of SNPs to characterize the haplotype structure of PRL and PRLR genes, using the criteria for haplotype-based studies described by Gabriel et al.3 8 and the method for selecting htSNPs by Stram et al.4 1 As described in their paper, we expected that using 6-8 common SNPs 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 9. Association between haplotype dosage estimate in LD blocks 1-4 of prolactin (PRL) and plasma PRL levels. Haplotypes2 Block 1 Controls (N) Meanb (95% Cl) Pc laCGGCCG 0 210 8.6 (7.7, 9.5) 0.54 1 122 7.9 (6.9,9) 2 23 8.5 (6.3,11.7) lb CGGTCG 0 218 8.1 (7.3, 9) 0.55 1 118 8.7 (7.6,10) 2 19 8.4 (6,11.9) lcCTGCCA 0 341 8.3 (7.6, 9) 0.15 1 13 9.7 (6.4, 14.7) 2 1 27 9 (6.5,120.7) Id CTACCG 0 337 8.5 (7.8, 9.2) 0.16 1 2 18 0 6.5 (4.6, 9.2) leCTACCA 0 273 8.1 (7.4, 8.9) 0.18 1 73 8.8 (7.4,10.4) 2 9 11.0 6(7.1,18.9) lfTTACCA 0 327 8.4 (7.7, 9.1) 0.83 1 2 28 0 8.1 (6, 10.8) lgCGGCTG 0 173 8.4 (7.4, 9.5) 0.44 1 140 8.7 (7.7, 9.9) 2 42 7.1 (5.6, 9) 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 9 continued. Haplotypes3 Block 2 Controls (N) Meanb (95% Cl) Pc 2a CCACA 0 312 8.4 (7.7, 9.1) 0.80 1 41 8.0 (6.1, 10.4) 2 2 17.1 (6.1,48.4) 2b CCAGA 0 344 8.2 (7.6, 8.9) 0.24 1 11 10.9 (6.9, 17.3) 2 0 2c CCGCA 0 341 8.3 (7.6, 8.9) 0.28 1 13 11.3 (7.4, 17.0) 2 1 6.3 (1.4, 27.6) 2d CAACT 0 271 8.6 (7.9, 9.5) 0.15 1 74 7.5 (6.2, 9.0) 2 10 7.1 (4.3,11.5) 2e TCACT 0 248 8.6 (7.8, 9.4) 0.07 1 97 8.0 (6.9, 9.4) 2 10 5.0 (3.1, 7.9) 2f TCACA 0 292 8.1 (7.4, 8.8) 0.09 1 57 9.4 (7.7, 11.4) 2 6 11.6 (6.4,21.0) 2g TCAGA 0 304 8.1 (7.5, 8.9) 0.20 1 48 9.6 (7.8, 12.0) 2 3 8.5 (3.6, 20.0) 2h TCGCA 0 323 8.3 (7.6, 9.0) 0.50 1 32 9.1 (7.0, 11.8) 2 0 2i TAACT 0 341 8.3 (7.7,9.0) 0.58 1 14 9.3 (6.2, 14.0) 2 0 2j CCACT 0 148 8.5 (7.5, 9.7) 0.30 1 163 8.6 (7.6,9.7) 2 44 7.1 (5.6, 8.9) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 9 continued. Haplotypes® Block 3 Controls (N) Meanb (95% Cl) Pc 3 a CTCCG 0 285 8.2 (7.5, 9.0) 0.58 1 65 8.9 (7.3, 10.7) 2 5 8.0 (4.1, 15.5) 3b CTCCA 0 317 8.3 (7.6, 9.0) 0.45 1 34 8.4 (6.3, 11.2) 2 3cCTCTG 4 13.5 (6.4, 28.6) 0 311 8.2 (7.5, 8.9) 0.21 1 2 44 0 9.5 (7.6,11.9) 3d CTCTA 0 264 8.6 (7.8, 9.4) 0.21 1 81 7.8 (6.6, 9.2) 2 10 6.8 (4.3,11.0) 3e CCCCG 0 331 8.3 (7.6, 9.0) 0.24 1 23 8.8 (6.4, 12.1) 2 1 40.2 (9.3, 173.1) 3f CCCCA 0 293 8.2 (7.5, 8.9) 0.17 1 56 8.7 (7.1, 10.7) 2 6 13.7 (7.5, 25.0) 3g TTCTA 0 345 8.3 (7.7, 9.0) 0.52 1 9 8.4 (5.1, 13.8) 2 1 18.4 (4.2,79.9) 3h CTTCG 0 101 9.4 (8.0,11.0) 0.07 1 179 8.2 (7.3, 9.1) 2 75 7.5 (6.3, 9.0) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 9 continued. Haplotypes8 Block 4 Controls (N) Mean” (95% Cl) Pc 4a CTAAT 0 294 8.4 (7.7, 9.2) 0.85 1 54 7.9 (6.4, 9.7) 2 7 9.3 (5.3, 16.2) 4b CTAAA 0 301 8.2 (7.5, 9.0) 0.37 1 51 8.8 (7.0, 11.0) 2 3 13.3 (5.6,31.6) 4c CTGGT 0 246 8.2 (7.4, 9.0) 0.71 1 96 8.9 (7.6, 10.4) 2 13 7.5 (5.0, 11.2) 4d CCAAA 0 285 8.4 (7.7, 9.1) 0.92 1 56 7.9 (6.5, 9.7) 2 14 9.4 (6.3,13.9) 4e TCAAT 0 288 8.6 (7.9,9.4) 0.03 1 64 7.2 (5.9, 8.8) 2 3 4.0 (1.7, 9.4) 4f TCAAA 0 291 8.6 (7.9, 9.4) 0.16 1 58 7.2 (5.9, 8.7) 2 6 8.2 (4.5, 14.9) 4g CTAGT 0 169 7.8 (7.0, 8.7) 0.14 1 141 8.8 (7.8, 10.0) 2 45 9.0 (7.2, 11.1) Mean values shown for 0, 1, or 2 copies of haplotypes. “ Haplotypes observed with >5% frequency among cases or controls in at least one ethnic group are shown. b Adjusted for age and ethnicity. c p-value for test of trend or test of difference. 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 10. Association between haplotype dosage estimate in LD blocks 1-6 of prolactin receptor (PRLR) and plasma PRL levels. Haplotypesa Block 1 Controls (N) Mean (95% Cl) Pc laGTCTC 0 301 8.4 (7.7, 9.2) 0.58 1 50 7.8 (6.2,9.8) 2 4 8.1 (3.8, 17.2) lb GTCTT 0 190 8.3 (7.4, 9.3) 0.54 1 119 8.0 (7.0, 9.2) 2 46 9.4 (7.5,11.8) lcGCCTC 0 309 8.2 (7.5, 8.9) 0.28 1 42 9.7 (7.7, 12.2) 2 4 7.9 (3.8,16.7) Id GCCTT 0 335 8.2 (7.6, 8.9) 0.07 1 17 11.4 (7.7,16.8) 2 3 12.7 (5.4, 29.9) le ATCGC 0 218 8.4 (7.6, 9.3) 0.91 1 119 8.2 (7.2, 9.4) 2 18 8.5 (6.0, 12) lfGTCGC 0 208 8.5 (7.6, 9.4) 0.23 1 125 8.6 (7.5, 9.8) 2 22 6.0 (4.3, 8.3) 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 10 continued. Haplotypes2 Block 2 Controls (N) Meanb (95% Cl) Pc 2a TGTA 0 344 8.4 (7.7, 9.1) 0.58 1 2 11 0 7.3 (4.6, 11.7) 2b TGCA 0 310 8.6 (7.9, 9.3) 0.03 1 42 7.0 (5.6, 8.9) 2 3 3.9 (1.7, 8.9) 2c TGCG 0 299 8.2 (7.5, 8.9) 0.34 1 53 9.2 (7.5,11.3) 2 3 8.7 (3.7, 20.4) 2d AGTA 0 208 7.8 (6.9, 8.7) 0.15 1 111 9.1 (7.9, 10.6) 2 36 9.2 (7,12.1) 2e AGCA 0 345 8.4 (7.7, 9.1) 0.44 1 9 6.4 (3.8,10.7) 2 1 9.0 (2.1,38.5) 2f AGCG 0 296 8.4 (7.6, 9.1) 0.89 1 50 8.1 (6.5,10) 2 9 9.7 (5.9, 16) 2g TATA 0 116 8.4 (7.3, 9.7) 0.99 1 162 8.2 (7.3, 9.2) 2 77 8.5 (7.1,10.1) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 10 continued. Haplotypes3 Block 3 Controls (N) Meanb (95% Cl) P° 3a GGT 0 324 8.5 (7.8,9.3) 0.11 1 2 31 0 6.8 (5.2, 8.8) 3b GAT 0 280 8.4 (7.7, 9.2) 0.52 1 71 8.0 (6.7,9.6) 2 4 7.1 (3.4, 14.9) 3c CGT 0 271 8.3 (7.6, 9.2) 0.78 1 73 8.1 (6.8, 9.6) 2 11 10.2 (6.6, 16.0) 3d GGC 0 40 8.4 (6.6, 10.6) 0.32 1 131 7.7 (6.8, 8.8) 2 184 8.8 (7.9,9.9) Haplotypes3 Block 4 Controls (N) Meanb (95% Cl) Pc 4a CTT 0 348 8.4 (7.7, 9.0) 0.78 1 2 7 0 7.7 (4.3,13.6) 4b GTC 0 329 8.2 (7.5, 8.9) 0.07 1 22 11.0 (7.7,15.5) 2 4 12.7 (6.1,26.7) 4c GTT 0 318 8.4 (7.7, 9.1) 0.41 1 36 8.2 (6.4, 10.5) 2 1 2.0 (0.5, 8.8) 4d GCC 0 297 8.2 (7.6, 9.0) 0.41 1 55 8.7 (7.1, 10.6) 2 3 13.1 (5.6, 30.5) 4e CTC 0 20 9.5 (6.7, 13.4) 0.41 1 97 8.9 (7.6, 10.4) 2 238 8.1 (7.3, 8.9) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 10 continued. Haplotypes® Block 5 Controls (N) Mean (95% Cl) Pc 5a TTGAA 0 96 7.7 (6.5, 9.0) 0.44 1 171 8.6 (7.7, 9.7) 2 88 8.5 (7.2,9.9) 5b TCGAG 0 300 8.2 (7.5, 8.9) 0.47 1 51 9.5 (7.7, 11.8) 2 4 6.2 (3,12.8) 5c TCGGG 0 347 8.4 (7.7, 9.1) 0.38 1 2 8 0 6.6 (3.9, 11.2) 5d CCGAG 0 329 8.3 (7.6,9.0) 0.49 1 25 9.1 (6.7, 12.4) 2 1 11.2 (2.6,48.2) 5e CCGGG 0 299 8.4 (7.7, 9.1) 0.99 1 53 8.1 (6.5, 10.0) 2 3 11.2 (4.7, 26.5) 5f TTAAA 0 206 8.3 (7.5,9.2) 0.49 1 121 8.9 (7.8,10.3) 2 28 6.4 (4.8, 8.5) Haplotypes® Block 6 Controls (N) Meanb (95% Cl) Pc 6a CATG 0 332 8.2 (7.6, 8.9) 0.25 1 20 10.8 (7.5, 15.6) 2 3 8.8 (3.8, 20.5) 6b TGCG 0 332 8.2 (7.6, 8.9) 0.20 1 22 9.8 (7.2, 13.4) 2 1 15.2 (3.5, 66.3) 6c TGCA 0 234 8.5 (7.7,9.4) 0.72 1 103 8.0 (6.9, 9.2) 2 18 8.7 (6.1, 12.5) 6c TATG 0 340 8.3 (7.7, 9.0) 0.94 1 2 15 0 8.5 (5.6,12.7) 6dCGTG 0 50 8.6 (6.8,10.8) 0.44 1 127 8.7 (7.6, 9.9) 2 178 8.0 (7.1, 9.0) Mean values shown for 0, 1, or 2 copies o f haplotypes. “ Haplotypes observed with >5% frequency among cases or controls in at least one ethnic group are shown. b Adjusted for age and ethnicity. c p-value for test of trend or test of difference. 76 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 11a. Association between htSNPs in region of low LD in PRL and breast cancer risk. Minor allele percentages among cases/controls (%) African- Hawaiians Japanese Latinas Whites All groups combined Americans SNP no. SNP ID Minor allele8 (n = 363/ 659) (n= 115/ 296) (n= 453/422) (n = 359/681) (n = 426/447) ORb (95% Cl) 24 rs767938 C 54/54 32/31 31/28 52/51 48/47 1.00 (0.82,1.22) 30 rs2744117 T 3/ 4 18/19 30/33 15/13 10/ 8 1.34 (0.95,1.90) 32 rs849876 A 13/14 2/ 4 3/ 3 6/ 5 9/ 11 0.81 (0.59,1.12) 33 rs3756824 C 1/ 1 12/12 28/ 25 11/13 3/ 3 1.00 (0.59,1.70) 35 rs2244502 T 49/48 23/ 26 31/28 34/33 29/30 0.9 (0.73,1.12) a Among all groups combined. b Odds ratios estimated using logistic regression adjusted for age and ethnicity. Table lib . Association between htSNPs in region of low LD of PRL and plasma PRL levels. SNP no. SNP ID (N) Genotype Mean8 (95% Cl) Pb 24 rs767938 120 CC 8.05 (7.03, 9.22) 0.42 147 CT 7.84 (6.95, 8.83) 56 TT 9.24 (7.53, 11.34) 30 rs2744117 249 CC 8.38 (7.60, 9.24) 0.55 89 AC 8.06 (6.83, 9.52) 11 AA 7.29 (4.57, 11.61) 32 rs849876 301 AA 8.20 (7.54, 8.93) 0.36 44 AG 9.35 (7.48, 11.68) 3 GG 8.01 (3.45, 18.60) 33 rs3756824 279 CC 7.89 (7.21, 8.64) 0.08 63 CG 9.69 (8.02, 11.71) 3 GG 8.80 (3.75, 20.63) 35 rs2244502 38 AA 11.05 (8.64, 14.12) 0.0005 151 AT 8.88 (7.91, 9.98) 159 TT 7.15 (6.34, 8.06) 8 Least square means estimated using general linearized model adjusted for age and ethnicity. b p-value for test of trend. <1 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 12a. Association between missense SNP IlelOOLeu (exon5) in prolactin receptor (PRLR) and breast cancer risk. SNP no. SNP ID Minor allele percentages among cases/controls (%) African- Hawaiians Japanese Americans Minor allele3 (n = 363/659) fn = 115/296) (n= 453/422) Latinas Whites (n = 359/ 681) (n = 426/ 447) All groups combined ORb (95% Cl) 81 seq5582 C 11/11 14/14 3/ 3 3/4 4/3 1.00 (0.83, 1.20) a Among all groups combined. b Odds ratios estimated using logistic regression adjusted for age and ethnicity. (N) Genotype Mean3 (95% Cl) pb 306 CC 8.4 (7.7, 9.2) 0.49 38 CT 8.4 (6.6, 10.8) 5 TT 5.6 (2.9,10.8) 3 Adjusted for age and ethnicity. bp-value for test of trend. -J o o ( >10% frequency) would allow us to predict more than 90% of the diversity of common haplotypes in a given block. We observed a slightly significant association in block 2 of PRLR when performed a global likelihood test (p = 0.04). However, when we assessed risk among individual haplotypes within this block, we observed no association with risk by ethnic group or among all groups combined. Therefore, it is unlikely that the global test suggests a true association, but rather reflects statistical fluctuations in estimating risk. We also observed a modest association between haplotype 4e of PRL and plasma PRL levels. However, results are inconclusive since these analyses were based on small numbers. In the single SNP analysis of the region of low LD in PRL, SNP 35 within intron 1 was found to be statistically significantly associated with plasma PRL levels. This SNP lies between two functional regions (2.2 kb downstream of exonl, 174 bases upstream of exon 2). However, whether or not this SNP is involved in affecting transcription is unlikely given that the SNP is located within a non-coding region. Further studies in larger samples are needed to definitively assess the relationship between this polymorphism and plasma PRL levels. The example of PRL raises the question of whether a haplotype-based approach is the best method to assess common genetic variation in regions of low LD. Recent studies suggest that most of the genome consists of regions of low LD of varying lengths, while other parts of the genome are characterized by “hot spots” of recombination.4 8 '5 2 Moreover, it is estimated that half of this recombination occurs in less than 10% of the sequence.5 2 Further studies have shown that as the 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. density of markers is increased, larger blocks that show little evidence of LD are found to split into smaller blocks, with an increase in the complexity of the SNP haplotypes.53,5 4 Prolactin is an example of such a region where despite a density of 1 SNP every 1 kb, there is little evidence of block structure. We used the multivariate R % to predict the common haplotypes in PRL and PRLR. This measure minimizes the bias of misclassification of rare haplotypes with common haplotypes which can lead to underestimation of haplotype effects. Assuming an R % - 0.9, we had 96% power to detect a relative risk as low as 1.35 for a dominant allele with a frequency of 10%. By ethnic group, we had between 78% and 82% power to detect a relative risk of 1.60; for Hawaiians, we only had 74% power to detect a relative risk of 2.0. Therefore, we may not have had enough power to assess ethnic-specific relative risks <1.6 nor to evaluate variants that may be differentially penetrant by ethnic group due to epistatic interactions with other genes. The purpose of this study was to assess shared common genetic variation across ethnic groups. Future studies evaluating ethnic-specific risks will need to be assessed among much larger samples. Though we used i?/to select htSNPs for the region of low LD in PRL, alternate methods that do not rely on haplotype blocks may be more appropriate for selecting htSNPs in these types of regions.5 5 However, for this analysis, we found that these SNPs were adequately predicted by the other SNPs in the gene and therefore, assessed common variation in this region using single SNP analyses as well as haplotype-based analysis. Though this region likely includes recombinant 80 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. haplotypes, the high density of SNPs suggests that we can be fairly confident that we have covered most of the common genetic variability in this region and that assessing the haplotypes in relation to breast cancer will offer additional information to the single SNP analyses. The only polymorphism previously reported in PRL or PRLR in relation to breast cancer risk is a Leul50Ile SNP in exon 6 of PRLR, observed in just 2 patients among 38 cases in a Turkish study.5 6 In our large sample, we were not able to validate the presence of a common variant in exon 6 of PRLR; however, it is possible that this variant is rare or only observed in certain populations specific to that region in which it was observed. The only other study to test for common variants in PRL pathway genes in relation to breast cancer is a small study among 30 patients in which they did not find any polymorphisms in PRLR. Two SNPs in PRL have been reported in the literature, but only in relation to other autoimmune diseases such as systemic lupus erythematosus (SLE). These variants, G-1149T in the extra-pituitary promoter region and C214T in intron 3, and were included in our analysis (rsl341239 and rs7739889) as SNP 29 in block 2 and SNP 40 in block 3 ofPRL. The strength of this study is the large sample size and the ability to assess LD block structures and haplotype across different ethnic populations. However, the ability to definitively evaluate ethnic-specific risks and the ability to assess associations with plasma PRL levels should be interpreted with caution due to the small number of subjects in these groups. 81 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This study provides the first evaluation of common genetic variation in prolactin pathway genes and breast cancer risk. Though we observed no apparent association with breast cancer risk or plasma PRL levels in relation to haplotypes in PRL and PRLR or to the missense variant, IlelOOLeu, the haplotype characterization of these genes provides a framework for future studies of PRL pathway genes and other disease outcomes (e.g. autoimmune diseases) and for larger studies of plasma PRL levels. This study provides an example of important methodological issues regarding the use of a haplotype-based approach in regions of low LD. 82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.5 REFERENCES 1. Clevenger CV, Furth PA, Hankinson SE, Schuler LA. The role of prolactin in mammary carcinoma. Endocr Rev. 2003;24(l):l-27. 2. Yen S, Jaffe RBe. Reproductive endocrinology. 4th edition ed. Philadelphia, PA: Saunders; 1999. 3. Muhlbock O, Boot LM. Induction of mammary cancer in mice without the mammary tumor agent by isografts of hypophyses. Cancer Res. 1959;19:402-412. 4. Boot LM, Muhlbock O, Ropcke G. Prolactin and the induction of mammary tumors in mice. General and Comparative Endocrinology. 1962;2:601-602. 5. Welsch CW, Gribler C. Prophylaxis of spontaneously developing mammary carcinoma in C3H-HeJ female mice by suppression of prolactin. Cancer Res. 1973;33(11):2939-2946. 6. Welsch CW, Nagasawa H. Prolactin and murine mammary tumorigenesis: a review. Cancer Res. 1977;37(4):951-963. 7. Lihy K, Neltner B, Mohamet L, Menchen L, Ben-Jonathan N. Prolactin overexpression by MDA-MB-435 human breast cancer cells accelerates tumor growth. Breast Cancer Res Treat. May 2003;79(2):241-252. 8. Schroeder MD, Symowicz J, Schuler LA. PRL modulates cell cycle regulators in mammary tumor epithelial cells. Mol Endocrinol. Jan 2002;16(l):45-57. 9. Gutzman JH, Miller KK, Schuler LA. Endogenous human prolactin and not exogenous human prolactin induces estrogen receptor alpha and prolactin receptor expression and increases estrogen responsiveness in breast cancer cells. J Steroid Biochem Mol Biol. Jan 2004;88(l):69-77. 10. Ormandy CJ, Hall RE, Manning DL, et al. Coexpression and cross regulation of the prolactin receptor and sex steroid hormone receptors in breast cancer. J Clin Endocrinol Metab. Nov 1997;82(11):3692-3699. 11. Maus MV, Reilly SC, Clevenger CV. Prolactin as a chemoattractant for human breast carcinoma. Endocrinology. Nov 1999;140(11):5447-5450. 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12. Struman I, Bentzien F, Lee H, et al. Opposing actions of intact and N- terminal fragments of the human prolactin/growth hormone family members on angiogenesis: an efficient mechanism for the regulation of angiogenesis. Proc Natl Acad Sci USA. Feb 16 1999;96(4):1246-1251. 13. Touraine P, Martini JF, Zafrani B, et al. Increased expression of prolactin receptor gene assessed by quantitative polymerase chain reaction in human breast tumors versus normal breast tissues. J Clin Endocrinol Metab. 1998;83(2):667-674. 14. Ben-Jonathan N, Liby K, McFarland M, Zinger M. Prolactin as an autocrine/paracrine growth factor in human cancer. Trends Endocrinol Metab. Aug 2002;13(6):245-250. 15. Clevenger CV, Chang WP, Ngo W, Pasha TL, Montone KT, Tomaszewski JE. Expression of prolactin and prolactin receptor in human breast carcinoma. Evidence for an autocrine/paracrine loop. Am J Pathol. Mar 1995;146(3):695-705. 16. Tworoger SS, Eliassen AH, Rosner B, Sluss P, Hankinson SE. Plasma prolactin concentrations and risk of postmenopausal breast cancer. Cancer Res. 2004;64(18):6814-6819. 17. Manjer J, Johansson R, Berglund G, et al. Postmenopausal breast cancer risk in relation to sex steroid hormones, prolactin, and SHBG (Sweden). Cancer Causes Control. 2003; 14(7):599-607. 18. Wang DY, De Stavola BL, Bulbrook RD, et al. Relationship ofblood prolactin levels and the risk of subsequent breast cancer. Int J Epidemiol. 1992;21 (2) :214-221. 19. Kabuto M, Akiba S, Stevens RG, Neriishi K, Land CE. A prospective study of estradiol and breast cancer in Japanese women. Cancer Epidemiol Biomarkers Prev. 2 0 0 0 ; 9 ( 6 ) : 5 7 5 - 5 7 9 . 20. Cole EN, England PC, Sellwood RA, Griffiths K. Serum prolactin concentrations throughout the menstrual cycle of normal women and patients with recent breast cancer. Eur J Cancer. 1977;13(7):677-684. 21. Malarkey WB, Schroeder LL, Stevens VC, James AG, Lanese RR. Disordered nocturnal prolactin regulation in women with breast cancer. Cancer Res. 1977;37(12):4650-4654. 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22. Rose DP, Pruitt BT. Plasma prolactin levels in patients with breast cancer. Cancer. 1981;48(12):2687-2691. 23. Meyer F, Brisson J, Morrison AS, Brown JB. Endogenous sex hormones, prolactin, and mammographic features of breast tissue in premenopausal women. J Natl Cancer Inst. 1986;77(3):617-620. 24. Love RR, Rose DR, Surawicz TS, Newcomb PA. Prolactin and growth hormone levels in premenopausal women with breast cancer and healthy women with a strong family history of breast cancer. Cancer. 1991 ;68(6): 1401-1405. 25. Ingram DM, Nottage EM, Roberts AN. Prolactin and breast cancer risk. MedJAust. 1990;153(8):469-473. 26. Secreto G, Recchione C, Cavalleri A, Miraglia M, Dati V. Circulating levels of testosterone, 17 beta-oestradiol, luteinising hormone and prolactin in postmenopausal breast cancer patients. Br J Cancer. 1983;47(2):269-275. 27. Bernstein L, Ross RK. Endogenous hormones and breast cancer risk. Epidemiol Rev. 1993;15(l):48-65. 28. Helzlsouer KJ, Alberg AJ, Bush TL, Longcope C, Gordon GB, Comstock GW. A prospective study of endogenous hormones and breast cancer. Cancer Detect Prev. 1994; 18(2):79-85. 29. Truong AT, Duez C, Belayew A, et al. Isolation and characterization of the human prolactin gene. Embo J. 1984;3(2):429-437. 30. Berwaer M, Martial JA, Davis JR. Characterization of an up-stream promoter directing extrapituitary expression of the human prolactin gene. Mol Endocrinol. 1994;8(5):635-642. 31. DiMattia GE, Gellersen B, Duckworth ML, Friesen HG. Human prolactin gene expression. The use of an alternative noncoding exon in decidua and the IM-9-P3 lymphoblast cell line. J Biol Chem. 1990;265(27): 16412- 16421. 32. Arden KC, Boutin JM, Djiane J, Kelly PA, Cavenee WK. The receptors for prolactin and growth hormone are localized in the same region of human chromosome 5. Cytogenet Cell Genet. 1990;53(2-3): 161-165. 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33. Hu ZZ, Zhuang L, Meng J, Tsai-Morris CH, Dufau ML. Complex 5' genomic structure of the human prolactin receptor: multiple alternative exons 1 and promoter utilization. Endocrinology. Jun 2002;143(6):2139- 2142. 34. Hu ZZ, Zhuang L, Meng J, Leondires M, Dufau ML. The human prolactin receptor gene structure and alternative promoter utilization: the generic promoter hPIII and a novel human promoter hP(N). J Clin Endocrinol Metab. 1999;84(3): 1153-1156. 35. Hu ZZ, Meng I, Dufau ML. Isolation and characterization of two novel forms of the human prolactin receptor generated by alternative splicing of a newly identified exon 11. J Biol Chem. 2001 ;276(44):41086-41094. 36. Trott JF, Hovey RC, Koduri S, Vonderhaar BK. Alternative splicing to exon 11 of human prolactin receptor gene results in multiple isoforms including a secreted prolactin-binding protein. J Mol Endocrinol. Feb 2003;30(1):31- 47. 37. Kolonel LC, Henderson BE, Hankin JH, et al. A multiethnic cohort in Hawaii and Los Angeles: Baseline Characteristics. American Journal of Epidemiology. 2000; 151(4):346-357. 38. Gabriel SB, Schaffner SF, Nguyen H, et al. The structure of haplotype blocks in the human genome. Science. 2002;296(5576):2225-2229. 39. Haiman CA, Stram DO, Pike MC, et al. A comprehensive haplotype analysis of CYP19 and breast cancer risk: the Multiethnic Cohort. Hum Mol Genet. 2003;12(20):2679-2692. 40. Excoffier L, Slatkin M. Maximum-likelihood estimation of molecular haplotype frequencies in a diploid population. Mol Biol Evol. 1995;12(5):921-927. 41. Stram DO, Haiman CA, Hirschhom JN, et al. Choosing Haplotype-tagging SNPs based on unphased genotype data using a preliminary sample of unrelated subjects with an example from the Multiethnic Cohort Study. Human Hered. 2003;55(l):227-236. 42. Zhang K, Deng M, Chen T, Waterman MS, Sun F. A dynamic programming algorithm for haplotype block partitioning. Proc Natl Acad Sci. 2002;99:7335-7339. 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43. Ke X, Cardon LR. Efficient selective screening of haplotype tag SNPs. Bioinformatics. 2003;19:287-288. 44. Zaykin DV, Westfall PH, Young SS, Kamoub MA, Wagner MJ, Eton MG. Testing association of statistically inferred haplotypes with discrete and continuous traits in samples of unrelated individuals. Hum Hered. 2002;53(2):79-91. 45. Stram DO, Pearce CL, Bretsky P, et al. Abstract modeling and E-M estimation of haplotype specific relative risks from genotype data for a case control-study of unrelated individuals. Hum Hered. 2003 ;55(4): 179-190. 46. Hankinson SE, London SJ, Chute CG, et al. Effect of transport conditions on the stability of biochemical markers in blood. Clin Chem. 1989;35(12):2313-2316. 47. Pike MC, Kolonel LN, Henderson BE, et al. 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. 2002; 11(9):795-800. 48. Kauppi L, Jeffreys AJ, Keeney S. Where the crossovers are: recombination distributions in mammals. Nat Rev Genet. Jun 2004;5(6):413-424. 49. Kauppi L, Sajantila A, Jeffreys AJ. Recombination hotspots rather than population history dominate linkage disequilibrium in the MHC class II region. Hum Mol Genet. Jan 1 2003;12(l):33-40. 50. Jeffreys AJ, Holloway JK, Kauppi L, et al. Meiotic recombination hot spots and human DNA diversity. Philos Trans R Soc Lond B Biol Sci. Jan 29 2004;359(1441):141-152. 51. Jeffreys AJ, Kauppi L, Neumann R. Intensely punctate meiotic recombination in the class II region of the major histocompatibility complex. Nat Genet. Oct 2001 ;29(2):217-222. 52. McVean GA, Myers SR, Hunt S, Deloukas P, Bentley DR, Donnelly P. The fine-scale structure of recombination rate variation in the human genome. Science. Apr 23 2004;304(5670):581-584. 53. Wall JD, Pritchard JK. Assessing the performance of the haplotype block model of linkage disequilibrium. Am J Hum Genet. Sep 2003;73(3):502- 515. 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54. Crawford DC, Carlson CS, Rieder MJ, et al. Haplotype diversity across 100 candidate genes for inflammation, lipid metabolism, and blood pressure regulation in two populations. Am J Hum Genet. Apr 2004;74(4):610-622. 55. Carlson CS, Eberle MA, Rieder MJ, Yi Q, Kruglyak L, Nickerson DA. Selecting a maximally informative set of single-nucleotide polymorphisms for association analyses using linkage disequilibrium. Am J Hum Genet. Jan 2004;74(1): 106-120. 56. Canbay E, Degerli N, Gulluoglu BM, Kaya H, Sen M, Bardakci F. Could prolactin receptor gene polymorphism play a role in pathogenesis of breast carcinoma? Curr Med Res Opin. Apr 2004;20(4):533-540. 57. Glasow A, Horn LC, Taymans SE, et al. Mutational analysis of the PRL receptor gene in human breast tumors with differential PRL receptor protein expression. J Clin Endocrinol Metab. Aug 2001 ;86(8):3826-3832. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3 AN OVERVIEW OF POSTMENOPAUSAL ESTROGEN-PROGESTIN HORMONE THERAPY AND BREAST CANCER RISK Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.1 INTRODUCTION Menopausal hormone therapy (HT) in the form of ‘unopposed’ estrogen therapy (ET) first became widely used in the U.S. in the 1960s. After epidemiologic studies found that ET was responsible for a large increase in the incidence of endometrial cancer, the use of combined estrogen-progestin therapy (EPT) increasingly replaced ET in an attempt to minimize the carcinogenic effect of the estrogen on the endometrium. Most observational studies have reported a small but positive association between ET use and breast cancer risk,1 '7 and, over the past decade, an increasing number of observational studies have suggested that EPT use incurs greater risk of breast cancer than ET use.2 '1 0 The Women’s Health Initiative (WHI)1 1 ’1 2 randomized trial results confirmed the significantly increased risk from EPT use. The WHI trial1 2 of combined 0.625 mg conjugated equine estrogen (CEE) and 2.5 mg medroxyprogesterone acetate (MPA) administered daily to healthy postmenopausal women with an intact uterus found a relative risk of invasive breast cancer of 1.24 after an average of 5.6 years of follow-up. The WHI ET arm among women with a hysterectomy actually found a non-significant reduction in breast cancer risk compared to women on placebo.1 3 The Collaborative Group on Hormonal Factors in Breast Cancer (CGHFBC)1 pooled data in 1997 from 51 epidemiologic studies to obtain an overall estimate of risk associated with menopausal HT use. The breast cancer risk estimate for ET use was based on large numbers of cases and controls, but the EPT 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. result was based on small numbers. Since then, a number of statistically powerful studies have evaluated EPT in relation to breast cancer risk. Some of these further evaluated differences in risk by schedule of progestin administration, i.e., sequential versus continuous-combined use2'4,6’7 ’9 ’1 4 4 6 and in relation to histologic A ” 7 1 7 subtype ofbreast cancer (lobular or ductal carcinoma). ’ ’ ’ We conducted a meta-analysis of the results reported by the CGHFBC1 and studies published since that overview through March 2004 to provide a more precise quantitative estimate of the risk from EPT and to summarize the current literature on risk associated with schedule of progestin administration and histologic subtype. Risks for overall EPT use were calculated including and excluding the Million Women Study (MWS), the largest observational study to date on HT and breast cancer, as some questions have been raised as to the possibility of bias in the results from this study (see Discussion). 3.2 METHODS We used the Medline database to compile a list of studies subsequent to the CGHFBC report investigating the relationship between menopausal EPT and incident breast cancer risk using the Medical Subject Headings (MeSH): postmenopausal, estrogen progestin therapy (or combined therapy), and breast cancer. For this analysis we did not include studies that presented results only for overall HT, nor did we include studies that only evaluated ET use, nor studies only evaluating breast cancer mortality. A total of 22 studies were identified for 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. possible inclusion.2 '7,9’1 0 ’1 2 ,1 4 '2 8 Because age at menopause is a critical factor in • 70 assessing HT use and breast cancer risk, we excluded those studies that did not adjust for age at menopause: this criterion excluded five studies.10,21,24,26,2 8 We also excluded three studies that did not have information on risk by duration of 1Q 70 77 IS EPT use, ’ ’ and the study of Olsson and colleagues for reasons described in the Discussion below. We further excluded two studies since the results were based on the same data incorporated into the CGHFBC report,22,2 5 and the study of 77 Jemstrom and colleagues, because it only provided results for continuous- combined EPT use, and the reason for this was the greater observed effect with such use than with sequential use. Therefore, the results from ten recent studies and from the CGHFBC pooled analysis were used to obtain an overall assessment of EPT and breast cancer risk.1 '7,9’12,1 6 In a second analysis, breast cancer risk by histologic subtype (lobular versus ductal) was evaluated in relation to EPT use. Four of the ten studies had information on histology;4,6’17,1 8 Ursin et al1 8 used data from the study of Ross et al,3 and Daling et al1 7 from the study of Weiss et al.7 No information was available by histologic subtype from the CGHFBC report. In a third analysis, we assessed breast cancer risk by progestin schedule (sequential, i.e., estrogen given alone during the first part of a monthly cycle followed by estrogen combined with a progestin for the remainder of the cycle with possibly a short hormone-free interval, versus continuous-combined, i.e., estrogen and progestin always administered together during a cycle). No information was 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. available by progestin schedule from the CGHFBC report on EPT use and breast cancer risk. Seven of the ten studies had data subdivided by progestin schedule of EPT use.2,3’6 ’7 ’9 ’14,1 6 However, two of these seven studies did not assess duration of use in relation to schedule and were omitted from this analysis.6,9 Because so few studies provided information on past HT use, we were limited in our ability to assess the difference in risk by recency of use. Of the eleven studies included in the overall analysis, four studies presented results comparing risk for past versus current HT use.2,6’7 ’1 4 Of these, three studies assessed risk by duration of past and current HT use and were included in the analysis,2,7’1 4 but only the study by Weiss et al.7 reported risk for EPT use exclusively; the other two studies2,1 4 reported risk by past combined EPT and ET use. Log odds ratios (LORs ) per year of use (LORi) and 95% confidence intervals were calculated for each study using the meta-analytic methods described by Greenland.3 0 The model fitted is log-linear in duration of EPT. (The hazard ratios calculated for the prospective and randomized trial studies closely approximate ORs and we refer to both as ORs in this paper.) For all analyses, the most fully adjusted multivariate odds ratios were used. The fixed-effects and random effects summary LORiS were calculated by standard methods.31,3 2 For all tables, we only present the fixed-effects LORis and provide two-sided p-values for heterogeneity as p het- All analyses including funnel plots33,3 4 were conducted using the meta and metabias commands in STATA (StataCorp, College Station, TX). 93 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The WHI trial1 2 found, based on an intent-to-treat analysis, an average odds ratio (AOR) of invasive breast cancer of 1.24 for EPT use after an average of 5.6 years of follow-up. To convert this 1.24 figure into an ORi, we proceeded as follows: writing the instantaneous OR at the end of d years of use as O R < j, then the AOR up to the end of year t, AORf , is the integral of the O R/s with d taking all values from 0 through t divided by the cumulative standardized risk in women not exposed to EPT, i.e., t. This can be shown to result in the following equation: AOR5 .6 = 1.24 = [(ORi56 - l)/ln(ORi)]/5.6 Solving this equation gives ORi = 1.080. For cohort studies, the true duration of EPT use is underestimated in current hormone users. This is because EPT use is assessed at baseline but continues for an unknown proportion of individuals for at least some further period until censoring time. Therefore, an additional duration of use should be added for current hormone users in the cohort studies considered.1,4’9 ’14,1 6 For example, in the cohort study of Porch et al,9 they reported OR’s of 1.11 and 1.76 for <5 and years of EPT use. We consider these categories as referring to 2.5 years and 7.5 years of EPT use. Using these duration figures, we estimate ORi as 1.079. But the OR’s of 1.11 and 1.76 do not relate to 2.5 and 7.5 years of use, but to this amount of use plus the mean duration of use after recruitment to the study until the end of follow-up. The mean length of follow-up in this study was 5.9 years and assuming that current users of EPT remained users during follow-up, this changes the values to be used in estimating ORi from 2.5 and 7.5 years to 5.45 (2.5 plus the midpoint 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. of the average follow-up, i.e., 5 .9 / 2 or the average exposure during follow-up) and 10.45 (7.5 + 5.9/2) years respectively. This changes our estimate of ORi from 1.079 to 1.052, a 34% decline in our estimate of excess risk. This is, of course, a slight exaggeration of the change since some current users at baseline will stop use during follow-up. For all cohort studies included in the analysis, we calculated risk per year of use based on this conservative method. We applied this method to all prospective studies reporting risk for current EPT use except for the study by Schairer and colleagues,4 in which this adjustment had already been applied. Risk estimates reported in the study by Magnusson et al. were converted to risks per year of use since the ORjS reported in the study excluded never users of EPT. This was done in order for these estimates to be comparable to the relative risks reported in the other studies. 3.3 RESULTS The studies included in at least one of the three analyses conducted to evaluate EPT and breast cancer risk are given in Appendix 3. A summary table of the general characteristics and overall findings for each study are presented in Appendix 4. As is apparent in the summary table, duration of EPT use in these various studies was evaluated in a wide variety of ways: categorical cutpoints, per year of use, and cumulative risk over time. 95 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.3.1 EPT and breast cancer risk The overall summary of the studies included in this meta-analysis (all histologic subtypes combined) showed a weighted average ORi of 1.076 (95% confidence interval (Cl) = 1.070, 1.082) for EPT use, with some evidence of heterogeneity, phet = 0.074 (Table 1, Figure 1). A funnel plot showed no evidence of publication bias. Excluding the MWS,1 4 the weighted average ORi was 1.070 (95% Cl = 1.057,1.083; Phet^ 0.067); the excess risk (at one year) is thus 7.9% lower (0.070 versus 0.076). Excluding only the Scandinavian studies,2 ’1 6 the weighted average ORi for the remaining nine studies was 1.074 (95% Cl = 1.068, 1.080; p ^ = 0.18), whereas the ORi for the Scandinavian studies was 1.089 (95% Cl = 1.065, 1.114; phet = 0.32). 3.3.2 EPT and lobular versus ductal breast cancer risk Two of the four studies evaluating the difference in risk between lobular versus ductal breast carcinoma found no difference by histologic subtype while the other two studies observed a slightly increased risk for lobular carcinoma (Table 2). The overall difference in risk (at one year) by histologic subtype was 0.019 (95% Cl = -0.033, 0.071), pd iff = 0.47. 3.3.3 Sequential versus continuous-combined EPT and breast cancer risk Sequential EPT use was associated with a lower ORi than continuous- combined EPT use (Table 3). The best estimate of the overall difference between 96 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. Odds ratios per year of use (ORis) of estrogen-progestin therapy and breast cancer risk. Study Case Users (N) OR, (95% Cl) Randomized Trial WHI, 2003t 199 1.080(1.004, 1.167) Prospective Studies Stahlberg, 2004'* 95 1.097(1.068, 1.127) MWS,1 4 2003** 1,891 1.077(1.071, 1.084) Porch,9 2002*n 164 1.052(1.022, 1.084) Schairer,4 2000t§ 75 1.060(0.998, 1.150) Case-Control Studies Kirsh, 2002 1 43 1.15(1.01, 1.33) Newcomb,6 2002* 215 1.04(1.01, 1.08) Weiss,7 2002 195 1.065(1.019, 1.114) Ross,3 2000* 425 1.044(1.014, 1.077) Magnusson,2 1999* 399 1.104(1.073, 1.136) Pooled Studies CGHFBC,119971 194 1.058 (0.996, 1.124) S U M M A R Y All studies Pooled Estimate 1.076(1.070, 1.082) phet= 0.074 Excluding MWS1 4 1.070(1.057, 1.083) Phet =0.067 Excluding Scandinavian studies1,3'7,9’12,1 4 1.074(1.068, 1.080) ph e t =0.18 Scandinavian studies2,1 6 1.089(1.065, 1.114) ph e t =0.32 Abbreviations: Cl, confidence interval; EPT, estrogen-progestin therapy; ORs, odd ratios; WHI, Women’s Health Initiative; MWS, Million Women Study; CGHFBC Collaborative Group on Hormonal Risk Factors in Breast Cancer. Results included (or did not specifically exclude) women with unknown age at menopause due to simple hysterectomy. *Risk is based on current and/or recent use rather than total use. Results included (or did not specifically exclude) in situ breast cancer cases. ^Calculated number of cases for women with known age at menopause: 80% of the total number of cases (n=93). '• o Figure 1. Studies included in overall analysis of EPT and risk of breast cancer: Odds Ratios with 95% confidence intervals per year of use WHI Stahlberg MWS Porch S ch airer Kirsh Newcomb Weiss Ross Magnusson CGHFBC Com bined 1.1 RR 1.2 1.3 98 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 2. Odds Ratios per year of use (ORis) of estrogen-progestin therapy and breast cancer risk by histologic subtype. Lobular Ductal Study Case Users (N) OR, (95%CI) Case Users (N ) OR, (95%CI) Difference P r o s p e c tiv e S tu d ie s Schairer,4 2000*tJ 33 1.17(1.02, 1.41) 26 1.17(1.02, 1.41) 0.000 (-0.276, 0.276) C a se , C o n tr o lS tu d ie s Baling,1 7 2002 44 1.096(1.007, 1.193) 209 1.039 (0.99,1.089) 0.057 (-0.048, 0.162) Newcomb,6 2002* 32 1.04(0.97, 1.11) 208 1.04(1.00, 1.08) 0.000 (-0.109, 0.109) Ursin,182002 46 1.060(0.996, 1.128) 291 1.049(1.016, 1.084) 0.011 (-0.063, 0.085) SUMMARY Pooled Estimate 1.067(1.026, 1.110) Phet— 0.53 Pooled Estimate 1.046(1.023, 1.069) P h et =0.56 Pooled Estimate 0.019 (-0.033, 0.071) Pdiff= 0.47 Abbreviations: Cl, confidence interval; EPT, estrogen-progestin therapy; ORs, odd ratios; WHI, Women’s Health Initiative; MWS, Million Women Study; CGHFBC Collaborative Group on Hormonal Risk Factors in Breast Cancer . Results included (or did not specifically exclude) women with unknown age at menopause due to simple hysterectomy. ' Risk was based on current and/or recent use rather than total use. *Risk among lean women, lobular/ductal versus ductal only, results included (or did not specifically exclude) in situ breast cancer. SO so Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3. Odds Ratios per year of use (ORis) of estrogen-progestin therapy and breast cancer risk by progestin schedule. Sequential Continuous Case Users Case Users Study m OR, (95%CI) (N ) OR, (95%CI) Difference P r o s p e c tiv e S tu d ie s Stahlberg et al,1 6 2004'* 29 1.063 (1.024, 1.103) 20 1.137(1.093, 1.182) -0.074 (-0.134, -0.014) MWS,1 4 2003 1,181 1.093 (1.083, 1.103) 631 1.106(1.093, 1.120) -0.013 (-0.030, 0.004) C a se , C o n tr o l S tu d ie s Weiss,7 2002 78 1.031 (0.966, 1.100) 166 1.087(1.020, 1.159) -0.056 (-0.153, 0.041) Ross,3 2000* 320 1.067(1.025, 1.109) 105 1.017(0.975, 1.062) 0.050 (-0.010, 0.110) Magnusson,2 1999* 102 1.088(1.022, 1.158) 135 1.132(1.072, 1.197) -0.044 (-0.136,0.048) SUMMARY Pooled Estimate Pooled Estimate Pooled Estimate All Studies 1.089(1.080, 1.098) 1.103(1.092, 1.115) -0.015 (-0.030, 0.000) ph e t =0.20 P h et =0.002 pdiff= 0.054 Excluding Scandinavian 1.090(1.081, 1.100) 1.099(1.087, 1.112) -0.010 (-0.026, 0.006) Studies 3 ’7 ’1 4 Phet- 0 .1 2 ph e ,= 0.001 pdiff = 0.23 Scandinavian Studies2,1 6 1.070(1.036, 1.104) 1.135(1.100, 1.172) -0.065 (-0.115, -0.015) phet - 0 .5 3 p h et =0.915 pdiff =0.010 Abbreviations: Cl, confidence interval; EPT, estrogen-progestin therapy; ORs, odd ratios; WHI, Women’s Health Initiative; MWS, Million Women Study; CGHFBC, Collaborative Group on Hormonal Risk Factors in Breast Cancer. Results included (or did not specifically exclude) women with unknown age at menopause due to simple hysterectomy. *Risk was based on current and/or recent use rather than total use. *Risk included (or did not specifically exclude) in situ breast cancer cases. the ORiS was -0.015 (95% Cl = -0.030, 0.000), pdiff= 0.054. The most obvious difference between the continuous-combined and sequential schedules was seen in 9 1 f\ the two Scandinavian studies ’ in which the difference in risk was -0.065 (95% Cl = -0.115, -0.015), p^ff = 0.010. In the remaining studies the average difference was -0.010 (95% Cl = -0.026, 0.006), pdiff = 0.23; however, this figure essentially reflects the MWS1 4 in which the difference was -0.013 whereas the remaining two studies3,7 had differences of -0.056 and 0.050. 3.3.4 Current/recent use versus total lifetime use and breast cancer risk Only three studies2,7’1 4 reported results comparing risk for past versus current HT use. Pooled estimates for these three studies showed that the difference in risk was -0.067 (95% Cl = -0.081, -0.053), pdiff < 0.001 (Table 4). In the study by Weiss et al.,7 the only study to report risk separately for past and current EPT use, the difference was -0.100 (-0.166, -0.034), Pditr = 0.003. As another way to assess potential difference in risk by recency of use, we calculated pooled estimates for those studies reporting relative risks among current/and or recent EPT use1,4,9’12,14,1 6 and compared them to pooled estimates 2 3 5 7 * for the studies reporting risk for lifetime EPT use. ’ ’ ' The pooled estimate (data not shown) for the studies assessing current/recent use was slightly higher (ORi = 1.077, 95% Cl = 1.071, 1.083) than the pooled estimate for studies reporting lifetime EPT use (ORi = 1.053, 95% Cl = 1.034, 1.072), and this difference was statistically significant, pdiff = 0.019. This difference remained 101 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. Odds Ratios per year of use (ORis) of estrogen-progestin therapy and breast cancer risk by recency of use. Past Current Case Users Case Users Study (N) OR, (95%CI) (N) OR, (95%CI) Difference Prospective S tu d ie s MWS,142003t C a s e -C o n tr o l S tu d ie s 1005 1.010(0.998, 1.023) 1891 1.077(1.071, 1.084) -0.067 (-0.081, -0.053) Weiss,7 2002 187 0.947 (0.892, 1.005) 502 1.047(1.013, 1.083) -0.100 (-0.166, -0.034) Magnusson,2 1999*f 200 1.065(1.014, 1.118) 444 1.144(1.109, 1.181) P d iff = 0.003 -0.079 (-0.142, -0.016) SUMMARY Pooled Estimate Pooled Estimate Pooled Estimate All studies 1.011 (0.999, 1.023) Phet =0.011 1.079(1.073, 1.085) Phet <0.001 -0.069 (-0.082, -0.055) P diff < 0.0001 Abbreviations: Cl, confidence interval; EPT, estrogen-progestin therapy; ORs, odd ratios; MWS, Million Women Study. Risk included (or did not specifically exclude) in situ breast cancer cases. Ttisk for ET or EPT use. 9 significant even after excluding the Scandinavian studies ’ : the weighted average ORi for current/recent use was 1.076 (95% Cl = 1.070, 1.082) and 1.049 (95% Cl = 1.028, 1.070) for lifetime use, pd iff = 0.017. 3.4 DISCUSSION 3.4.1 EPT and breast cancer risk The literature evaluating EPT and breast cancer risk is generally very consistent; all studies reported an increased risk of breast cancer with increasing duration of EPT use. The overall evidence showed a statistically significant increased risk of 7.6% per year of use. This estimate was, however, greatly influenced by the large Million Women Study (MWS),1 4 and excluding this cohort study the estimated increased risk per year of use decreased to 7.0%. The MWS reported relative risks for <1, 1-4, 5-9, and >10 years of current EPT use of 1.45, 1.74, 2.17, and 2.31, respectively, based on 1,891 invasive breast cancer cases over an average of 2.6 years of follow-up. The risk per year of use, ORi, assuming EPT use continued during follow-up was 1.077 (95% Cl = 1.071, 1.084), or assuming EPT use ceased at the time of the baseline questionnaire, was 1.086 (95% Cl = 1.079,1.093). Neither one of these figures provides an adequate fit to these data. The reported ORs for short-term use (<1 and 1-4 years) are much too high. In contrast, the OR for long-term use ( ^ 0 years) is too low but is almost perfectly fit by the overall ORi of 1.076 found in the other studies. A possible cause for the high estimates with short-term use in the MWS was the initial 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. recruitment letter that specifically stated that the intent of the study was to “find out as much as possible about [hormone replacement therapy’s] benefits and any possible side effects” and “to learn about the way different types of HRT and other lifestyle factors affect a woman’s health, particularly her breasts” (http://www.millionwomenstudy.com). By alerting potential subjects to the study hypothesis, the investigators may have oversampled HT users who were concerned about changes they had perceived in their breasts. This possible bias would have a much smaller relative effect on the risks seen in long-term users. In contrast to the possibility that the MWS may have overestimated the • 19 risks attendant on HT use, the results we used for the WHI trial are almost certainly an underestimate of the true effect in that trial. We used the result obtained from their intent-to-treat analysis, which yielded an increased risk estimate of 8.0% per year of use. The authors also analyzed their data on a drug-as- taken basis and reported that the increased risk was double that obtained from the intent-to-treat analysis. Stahlberg and colleagues1 6 presented results for EPT duration by progestin schedule rather than overall EPT use. We combined the results for sequential and continuous-combined EPT to calculate risk of overall EPT use as ORi = 1.097 (95% Cl =1.068, 1.127). In this study, the authors adjusted for age at menopause as a dichotomous variable (<55 and >55 years old) which would not adequately control for this critical factor. However, this would tend to result in an 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. underestimate of risk. The very high estimates observed in this study for continuous-combined compared to sequential EPT are discussed below. The potency of the testosterone-derived progestin, norethistosterone acetate (NETA), the most commonly prescribed progestin in Scandinavia, has been estimated to be eight times that of MPA, the progestin commonly used in the U.S. and the U.K.3 5 '3 7 Adjusting for the doses prescribed, this suggests that the progestin dose in Scandinavia is possibly 1.5-2.0 times the dose used in the U.S. and the U.K. This difference may be the reason for the observed ORis for overall EPT use being higher in Scandinavian studies (1.089, 95% Cl = 1.065,1.114) than in non-Scandinavian studies (1.074, 95% Cl = 1.068, 1.080). Based on the complete set of studies, the best estimate of EPT use and breast cancer risk is an ORi of approximately 1.076. Based on a mathematical analysis of the age-incidence curve in U.S. women this 7.6% increase per year of use is some 70% of the risk that would have occurred had they continued to ovulate.3 8 3.4.2 EPT and lobular versus ductal breast cancer risk Summary risk estimates by histologic type were higher for lobular carcinoma than for ductal carcinoma; the overall difference was 0.019 (-0.033, 0.071) but was not statistically significant. Lobular carcinomas differ in a number of ways from ductal carcinoma, including molecular profiles, likelihood of 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. multifocality and importantly, ease of detectability by mammography.1 7 Further data are needed to settle this issue. The study by Schairer and colleagues4 reported much higher estimates than the other studies for both lobular (ORi = 1.17) and ductal carcinoma (ORi = 1.17). These risks compare to their overall result (Table 1) of an ORi of 1.076. The explanation of these apparent contradictory results is that the authors only provided data by histologic subtype among lean women (body mass index <24.4 kg/m ), and in their study the effect of EPT on breast cancer risk as measured by ORi was much greater in lean women. 3.4.3 Sequential versus continuous-combined EPT and breast cancer risk The overall within-study difference between sequential versus continuous- combined EPT was -0.015 (95% Cl = -0.030, 0.000), Pdiff= 0.054. This difference was largely attributed to the two Scandinavian studies2,1 6 where the risk was -0.065 (95% Cl = -0.115, -0.015), pdiff= 0.010. In the U.S. and the U.K. the most common form of sequential EPT provides 5-10 mg of MPA per day for 10-12 days per 28-day cycle, whereas subjects assigned to receive continuous-combined EPT are typically given 2.5 mg of MPA every day (i.e., total doses of approximately 50-120 mg and 70 mg respectively per month). Thus, for EPT users as a group, there is likely to be a somewhat lower total dose of progestin, on average, with continuous-combined than with sequential EPT use. In contrast, in Scandinavia the total dose of the progestin is much higher 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with continuous-combined than with sequential EPT, at least for a commonly prescribed regimen, NET A. For this regimen, the total dose per cycle of is roughly 10 mg for sequential EPT and 28 mg for continuous-combined EPT- a 180% higher dose for continuous-combined therapy. In the two Scandinavian studies included 9 1 f \ in our analysis, ’ the pooled estimate for sequential EPT was lower for sequential than continuous-combined EPT use (pdiff= 0.010). This result is also supported by 97 the results of the study of Jemstrom et al which we excluded earlier since the authors only presented results for continuous-combined EPT use; results were reported in this manner because of the greater observed effect for continuous- combined than with sequential EPT use. In the remaining studies,3,7’1 4 there was little difference in risk between continuous-combined and sequential EPT. This latter result is in agreement with the findings of the Postmenopausal Estrogen/Progestin Interventions (PEPI) randomized trial of HT.3 9 In PEPI, increases in mammographic densities (an important predictor of breast cancer risk) were similar for sequential and continuous-combined regimens (which used total monthly doses of 120 mg and 70 mg MPA, respectively). The mean change in percent density for sequential therapy was 4.76% vs. 4.50% for continuous- combined therapy. The cohort study by Olsson and colleagues1 5 was excluded from this meta analysis. By combining the results the authors presented separately for sequential and continuous-combined EPT use, we calculated an estimate of risk for 484- months of overall EPT use at baseline as an OR of 3.595 (95% Cl = 2.113, 6.115). 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This would give an ORi of 1.229 (95% Cl = 1.136,1.331), assuming an average of 6.5 years of use and no further use after baseline (given that EPT use was reported here for lifetime use; data was not provided by current or past use). This estimate was outside the range of the other studies and also varied greatly depending on the assumption made about EPT use after the initial baseline questionnaire because of the short average use recorded at baseline and the long ten year follow-up with no information on changes in EPT use. 3.4.4 Current/recent use versus total lifetime use and breast cancer risk The study by Weiss et al.7 was the only study which compared risk for past versus current EPT use. The results from this study suggest that breast cancer risk for current EPT use is higher than for past EPT use. On the assumption that EPT use comprises a large proportion of overall past HT use, we also evaluated two studies that presented data on past HT use: the MWS study1 4 comparing past HT use and current EPT use, and the study by Magnusson et al. comparing past and current HT use. Results from these three studies suggest that breast cancer risk for current HT use is greater than for past HT use. However, since HT use in past users includes more ET use proportionately, this estimate for the most part, likely represents risk for ET use. We also found that recent use is associated with a higher risk than for overall EPT use when we compared recent versus lifetime EPT use. However, all three of these analyses are based on small numbers of studies. The observed difference may reflect a real decrease in risk with the cessation of 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. hormone therapy. However, another possible reason may be that duration of hormone use is not measured the same in current as in past hormone users in observational studies. We have shown that in cohort studies, duration of use is underestimated in current users since exposure is only assessed at baseline. Moreover, the actual duration of use within a duration category will tend to be longer in current than in past users.4 0 Therefore, despite our attempts to adjust for bias in assessing duration of EPT use, it is likely that these methods do not adequately adjust for these underlying differences in measurement of exposure between past and current users. Four of the five studies reporting risk among current/recent users attempted to address the possibility that detection bias may be a possible explanation for the observed apparent decrease in risk among past users. Porch et al9 and Schairer et al4 adjusted for mammographic screening in their final multivariate models, while both the MWS1 4 and the WHI1 1 , n’1 4 had routine screening of subjects throughout the study period. None of these studies found detection bias to be responsible for the decrease in risk among past users. 3.4.5 Strengths and limitations One difficulty in directly comparing studies is the varied methods used in defining and adjusting for age at menopause. Failure to accurately account for age at menopause has been shown to seriously bias risk estimates downwards for current users.2 9 Therefore, we only included results for women with “known” age 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. at menopause when possible. However, some studies included women with simple hysterectomy (hysterectomy without bilateral oophorectomy) and assigned them a hypothetical age at menopause or simply coded them as having an “unknown age at menopause” in the analyses.4 '6,9’1 4 Regardless of the method used to assign the hypothetical age at menopause, these studies underestimate the effect of age at menopause and the effect of HT use on breast cancer risk.2 9 Since we deemed it unreasonable to exclude all such studies, we included those studies that made at least some attempt to adjust for age at menopause. We note these studies in the tables to alert readers to potential biases in their results. Finally, there is the possibility of publication bias in our meta-analysis, since the primary source of study selection was Medline. However, we would argue that studies finding no association with combined EPT and breast cancer risk would be of equal interest to the scientific community, at least until very recently, as those finding a positive association. Additionally, funnel plots found no evidence of publication bias in the studies included in this meta-analysis. 3.4.6 Summary As difficult as it is to compare such diverse studies, in our attempt to evaluate and summarize the literature on EPT use and breast cancer risk in a meaningful way, we found a consistently positive association across studies, collectively estimated to cause a 7.6% increase in risk per year of use. We observed a greater risk with sequential than continuous-combined EPT; this 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. difference was due entirely to two Scandinavian studies2,1 6 where the total progestin dose was much greater with the sequential regimen, a presumptive explanation for this difference. We also observed a non-significant difference in risk by histologic subtype. Among studies reporting risk for past versus current EPT use, for the same recorded duration of use, we observed a lower risk in past users. The validity of this result, however, is open to questions. These findings highlight the need for further studies to explore specific EPT schedules and particular histologic subtypes of breast cancer, as well as past hormone use to more accurately define both the risks involved with EPT use and the period of vulnerability following cessation of use. I ll Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.5 REFERENCES 1. Collaborative Group on Hormonal Factors in Breast Cancer. Breast cancer and hormone replacement therapy: collaborative reanalysis of data from 51 epidemiological studies of 52,705 women with breast cancer and 108,411 women without breast cancer. Lancet. 1997;350:1047-1059. 2. Magnusson C, Baron JA, Correia N, Bergstrom R, Adami HO, Persson I. Breast-cancer risk following long-term oestrogen- and oestrogen-progestin- replacement therapy. Int J Cancer. May 5 1999;81(3):339-344. 3. Ross RK, Paganini-Hill A, Wan PC, Pike MC. Effect of hormone replacement therapy on breast cancer risk: estrogen versus estrogen plus progestin. JNatl Cancer Inst. Feb 16 2000;92(4):328-332. 4. Schairer C, Lubin J, Troisi R, Sturgeon S, Brinton L, Hoover R. Menopausal estrogen and estrogen-progestin replacement therapy and breast cancer risk. Jama. Jan 26 2000;283(4):485-491. 5. Kirsh V, Kreiger N. Estrogen and estrogen-progestin replacement therapy and risk of postmenopausal breast cancer in Canada. Cancer Causes Control. Aug 2002;13(6):583-590. 6. Newcomb PA, Titus-Emstoff L, Egan KM, et al. Postmenopausal estrogen and progestin use in relation to breast cancer risk. Cancer Epidemiol Biomarkers Prev. Jul 2002; 11(7):593-600. 7. Weiss LK, Burkman RT, Cushing-Haugen KL, et al. Hormone replacement therapy regimens and breast cancer risk(l). Obstet Gynecol. Dec 2002;100(6):1148-1158. 8. Colditz GA, Hankinson SE, Hunder DJ, et al. The use of estrogens and progestins and the risk of breast cancer in postmenopausal women. N Engl J Med. June 15 1995;332(24):1589-1593. 9. Porch JV, Lee IM, Cook NR, Rexrode KM, Burin JE. Estrogen-progestin replacement therapy and breast cancer risk: the Women's Health Study (United States). Cancer Causes Control. Nov 2002;13(9):847-854. 10. Chen CL, Weiss NS, Newcomb P, Barlow W, White E. Hormone replacement therapy in relation to breast cancer. Jama. Feb 13 2 0 0 2 ; 2 8 7 ( 6 ) : 7 3 4 - 7 4 1 . 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11. Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women's Health Initiative randomized controlled trial. Jama. Jul 17 2002;288(3):321-333. 12. Chlebowski RT, Hendrix SL, Langer RD, et al. Influence of estrogen plus progestin on breast cancer and mammography in healthy postmenopausal women: the Women's Health Initiative Randomized Trial. Jama. Jun 25 2003;289(24):3243-3253. 13. Anderson GL, Limacher M, Assaf AR, et al. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women's Health Initiative randomized controlled trial. Jama. Apr 14 2004;291(14):1701-1712. 14. Million Women Study. Breast cancer and hormone-replacement therapy in the Million Women Study. Lancet. Aug 9 2003;362(9382):419-427. 15. Olsson HL, Ingvar C, Bladstrom A. Hormone replacement therapy containing progestins and given continuously increases breast carcinoma risk in Sweden. Cancer. Mar 15 2003;97(6): 1387-1392. 16. Stahlberg C, Pedersen AT, Lynge E, et al. Increased risk of breast cancer following different regimens of hormone replacement therapy frequently used in Europe. Int J Cancer. May 1 2004;109(5):721-727. 17. Daling JR, Malone KE, Doody DR, et al. Relation of regimens of combined hormone replacement therapy to lobular, ductal, and other histologic types of breast carcinoma. Cancer. Dec 15 2002;95(12):2455-2464. 18. Ursin G, Tseng CC, Paganini-Hill A, et al. Does menopausal hormone replacement therapy interact with known factors to increase risk of breast cancer? J Clin Oncol. Feb 1 2002;20(3):699-706. 19. Newcomer LM, Newcomb PA, Potter JD, et al. Postmenopausal hormone therapy and risk of breast cancer by histologic type (United States). Cancer Causes Control. Apr 2003;14(3):225-233. 20. Persson I, Yuen J, Bergkvist L, Schairer C. Cancer incidence and mortality in women receiving estrogen and estrogen-progestin replacement therapy- long-term follow-up of a Swedish cohort. Int J Cancer. Jul 29 1996;67(3):327-332. 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21. Persson I, Thurfjell E, Bergstrom R, Holmberg L. Hormone replacement therapy and the risk of breast cancer. Nested case-control study in a cohort of Swedish women attending mammography screening. Int J Cancer. Sep 4 1997;72(5):758-761. 22. Persson I, Weiderpass E, Bergkvist L, Bergstrom R, Schairer C. Risks of breast and endometrial cancer after estrogen and estrogen-progestin replacement. Cancer Causes Control. Aug 1999; 10(4):253-260. 23. Li Cl, Weiss NS, Stanford JL, Daling JR. Hormone replacement therapy in relation to risk of lobular and ductal breast carcinoma in middle-aged women. Cancer. Jun 1 2000;88(11):2570-2577. 24. Moorman PG, Kuwabara H, Millikan RC, Newman B. Menopausal hormones and breast cancer in a biracial population. Am J Public Health. Jun 2000;90(6):966-971. 25. Rockhill B, Colditz GA, Rosner B. Bias in breast cancer analyses due to error in age at menopause. Am J Epidemiol. Feb 15 2000; 151 (4):404-408. 26. Li R, Gilliland FD, Baumgartner K, Samet J. Hormone replacement therapy and breast carcinoma risk in Hispanic and non-Hispanic women. Cancer. Sep 1 2002;95(5):960-968. 27. Jemstrom H, Bendahl PO, Lidfeldt J, Nerbrand C, Agardh CD, Samsioe G. A prospective study of different types of hormone replacement therapy use and the risk of subsequent breast cancer: the women's health in the Lund area (WHILA) study (Sweden). Cancer Causes Control. Sep 2003;14(7):673-680. 28. Li Cl, Malone KE, Porter PL, et al. Relationship between long durations and different regimens of hormone therapy and risk of breast cancer. Jama. Jun 25 2003;289(24):3254-3263. 29. Pike MC, Ross RK, Spicer DV. Problems involved in including women with simple hysterectomy in epidemiologic studies measuring the effects of hormone replacement therapy on breast cancer risk. Am J Epidemiol. April 15 1998;147(8):718-721. 30. Rothman KJ, Greenland S. Modern Epidemiology. Vol xiii. Philadelphia: Lippincott-Raven; 1998. 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31. Fleiss JL. Analysis of data from multiclinic trials. Control Clin Trials. December 7 1986;7(4):267-275. 32. DerSimonian R, Laird N. Meta-analysis in clinical trial. Control Clin Trials. September 7 1986;7(3): 177-188. 33. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics. Dec 1994;50(4): 1088-1101. 34. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. Bmj. Sep 13 1997;315(7109):629-634. 35. Dickey RP, Stone SC. Progestational potency of oral contraceptives. Obstet Gynecol. Jan 1976;47(1): 106-112. 36. Stanczyk FZ. Pharmacokinetics and potency of progestins used for hormone replacement therapy and contraception. Rev Endocr Metab Disord. Sep 2002;3(3):211-224. 37. Back DJ, Bates M, Breckenridge AM, et al. The pharmacokinetics of levonorgestrel and ethynylestradiol in women - studies with Ovran and Ovranette. Contraception. Mar 1981 ;23(3):229-239. 38. Henderson BE, Ross RK, Pike MC. Hormonal chemoprevention of cancer in women. Science. Jan 29 1993;259(5095):633-638. 39. Greendale GA, Reboussin BA, Slone S, Wasilauskas C, Pike MC, Ursin G. Postmenopausal hormone therapy and change in mammographic density. J Natl Cancer Inst. Jan 1 2003;95(l):30-37. 40. Ettinger B, Grady D, Tosteson AN, Pressman A, Macer JL. Effect of the women's Health Initiative on women's decisions to discontinue postmenopausal hormone therapy. Obstet Gynecol. December 2003;102(6): 1225-1232. 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 4 HORMONE THERAPY AND BREAST CANCER RISK IN THE MULTIETHNIC COHORT STUDY Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.1 INTRODUCTION Since the 1960s, menopausal hormone therapy (HT) initially in the form of estrogen therapy (ET) and since the mid to late 1970s, in the form of combined estrogen-progestin therapy (EPT) has been widely prescribed for women to replace ovarian hormone production. The first large analysis to review the effects of HT use and breast cancer risk was the Collaborative Group on Hormonal Factors in Breast Cancer’s study of 51 epidemiologic studies.1 In this review, the relative risk of having a breast cancer diagnosis was estimated to increase by 1.023 (95% Cl = 1.011, 1.036) per year of HT use. This study did not have adequate power to assess risk by specific hormonal formulations, i.e., ET versus EPT use, and most use was of ET. Since then, a number of statistically powerful studies,2 " 1 0 including one randomized trial,11,1 2 have reported greater increased risk with EPT than ET use. In a recent meta-analysis of all published studies that evaluated duration of EPT use and breast cancer risk, we calculated that this increase in risk is approximately 7.6% per year of use.1 3 Although EPT use is clearly associated with an overall increase in breast cancer risk, further data are needed on whether EPT use and breast cancer risk varies by specific prognostic factors (e.g., histologic subtype, hormone receptor status, stage of disease, weight) or by various forms of EPT (sequential versus continuous-combined progestin schedules). More importantly, previous studies have evaluated breast cancer risk and HT use primarily among White populations. We report here our analysis of the relationship between duration of HT use and 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. breast cancer risk in a multiethnic cohort study of African-American (AA), Native Hawaiian (NH), Japanese-American (JA), Latina (LA), and White (WH) women. We also assess EPT use and breast cancer risk by weight, stage of disease, histologic subtype, and hormone receptor status. 4.2 METHODS 4.2.1 Study population The Hawaii-Los Angeles Multiethnic Cohort (MEC) Study is an ongoing, prospective study of dietary, environmental, and genetic factors in relation to cancer and other chronic diseases. The cohort includes 215,251 men and women, age 45 to 75 years at baseline, enrolled between 1993 and 1996 in Hawaii and California (primarily Los Angeles County). Based on self-report, 118,913 African- American, Native Hawaiian, Japanese-American, Latina, and White women were included in this study; a small number of subjects defined as of “other” ethnicity were excluded. Subjects completed a 26-page, self-administered mailed questionnaire that included information on diet, demographic factors, personal behaviors, prior medical conditions, use of medications, family history of common cancers, and, for women, reproductive history and use of oral contraceptives or HT. The primary sources for subject recruitment were the Departments of Motor Vehicles drivers’ license files for Hawaii and Los Angeles. An additional source of African-Americans in California was the Health Care Financing Administration (HCFA) files, and in Hawaii, voters’ registration files were used to identify 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. additional persons of Native Hawaiian ancestry and J apanese-American descent. Further details regarding the study design and characteristics of the cohort have been described elsewhere.1 4 4.2.2 Surveillance Incident cancer cases in California were identified by matching subjects with the Los Angeles County Cancer Surveillance Program (CSP) and the State of California Cancer Registry (CCR). Incident cancers in subjects recruited in Hawaii were identified using the Hawaii Tumor Registry (HTR) and the CCR. All tumor registries participate in the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program of cancer registration. Mortality due to cancer and other causes was determined by annual linkage to state death certificate files in California and Hawaii, and periodically to the National Death Index (NDI). Follow-up began when the subject completed the questionnaire. This study includes completed linkage to the Hawaii cancer registry up to December 31, 2001 and to the Los Angeles cancer registry up to December 31, 2002. Linkage to the California registry is complete up to December 31, 2001. Subjects recruited in Los Angeles County were assumed to have been followed-up to December 31, 2002; other subjects to December 31, 2001. Subjects were censored at these linkage dates unless a diagnosis of incident cancer or death occurred prior to this date. Exclusion criteria included any prevalent cancer (including in situ breast carcinoma) either by questionnaire or from the cancer 119 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. registry at the time of enrollment, premenopausal status, missing data on HT use, and simple hysterectomy before menopause (hysterectomy without bilateral oophorectomy). Data on tumor characteristics were obtained from SEER registry information. Tumors were defined according to the following International Classification of Disease for Oncology (ICD-O) classification of histologic subtypes: ductal carcinoma (ICD-0 code 8500; n = 1,179), pure lobular carcinoma (ICD-0 code 8520; n = 129), and mixed lobular/ductal carcinoma (ICD-0 code 8522; n = 95). In the analysis, lobular carcinoma was defined as either pure lobular or mixed lobular, however, a subanalysis was performed among pure lobular carcinomas only. Other subtypes excluded from this analysis were papillary (ICD-0 codes 8050, 8260, 8503; n = 18), tubular (ICD-0 code 8211; n = 46), mucinous (ICD-0 codes 8480, 8481; n = 58), medullary (ICD-0 codes 8510, 8512; n = 6), and other (n = 93) carcinomas. Stage of breast cancer was defined according to the SEER staging system1 5 ; women with non-localized breast cancer were defined as advanced cases. Estrogen receptor (ER) and progesterone receptor (PR) status was obtained from the SEER database. 4.2.3 Statistical methods 4.2.3.1 Age at and type of menopause Age at menopause was defined as the age at last menstrual period. Natural menopause was defined as the cessation of menses with no use of HT or oral 120 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. contraceptives. For women using oral contraceptives, age at menopause was defined as the end of the period of oral contraceptive use, if no natural menstruation occurred thereafter. For women using HT, age at menopause was defined as the year of initiating HT use, with the rationale that HT use was started for menopausal symptoms. If a woman resumed normal menstruation after HT use was stopped, this period of HT use was ignored. This is the same schema to approximate age at menopause that was used in earlier studies of HT and endometrial and breast cancer.3,1 6 Women who have undergone a simple hysterectomy before menopause were excluded from these analyses since it has been demonstrated that there is no way to assign an age at menopause to these women that will lead to unbiased estimates of HT effects on breast cancer risk.17,1 8 4.2.3.2 Duration of hormone replacement therapy use calculation Subjects were asked about ET and progesterone therapy (PT) use separately in our questionnaire. EPT use was calculated based on the overlap between reported periods of ET and PT use. Age at start of HT (ET, PT) use was recorded as a categorical variable (<40, 40-44, 45-49, 50-54, 55-59, 560), as was years of HT use (<1, 1-2, 3-5, 6-9, 10-14, 15-19, ^0). After an initial check showed that age at start and years of use of HT were compatible with the subject’s age at baseline, new variables were created using the midpoints for each category of age at start and years of use of HT. Based on these midpoint values, the overlap of ET 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. and PT use was used to calculated age at start and years of use of EPT, ET, and PT as continuous variables. Current users were assumed to be continuous users if data were missing on either age at start or years of use of HT. Past users with missing information on age at start or years of use of HT were excluded, as were current users with missing information on both age at start and years of use of HT. 4.2.3.3 Relative risk modeling Hazard ratios (HRs) for breast cancer incidence were estimated using Cox regression proportional hazard models, adjusted for age at questionnaire, ethnicity, and seven known breast cancer risk factors: age at menarche, age at first full-term pregnancy, number of children, age and type of menopause, weight at questionnaire, alcohol consumption, and hormone therapy use. The underlying time variable in the regression was from the date of enrollment to the date ofbreast cancer diagnosis, death, or censoring. Current HT use at baseline continues for an unknown proportion of individuals for at least some further period until censoring time so that the true duration of HT use is underestimated. Therefore, we included analyses that assigned an additional duration of use for current HT users, based on the conservative assumption that most current users continue to use HT until the censor date. To do this, we assessed HT use as a time-dependent covariate in an age and ethnicity matched case-control analysis where each case is randomly matched to 20 controls in the cohort at the case’s age at diagnosis. Current users are then 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. reassigned a new HT duration that is equal to the old HT duration plus the difference between the case’s age at diagnosis and the current user’s age at entry. All tables, unless otherwise specified, show results for risks adjusted for this additional duration. In the HR regression model, an individual’s age-specific hazard, \(t), of breast cancer at age t is modeled as \(t)= Xo(t)f(x,0), where the relative risk function, f(x,/3), is log linear in form, i.e., f(x,/3)=exp(f(x'/3)), relates “exposures” of interest, x, to breast cancer risk relative to the age-specific baseline rates \>(t). The parameter vector /S is estimated by the analysis. The multivariate x can include categorical and continuous variables. In the tables adjusted for ethnicity, the relative risks associated with x are assumed to be the same across all five ethnic groups. We conducted stratified analyses by weight at baseline (50th percentile), stage of disease (localized versus advanced), histologic subtype (lobular versus ductal), estrogen receptor (ER) positivity, and progesterone receptor (PR) positivity. Data was analyzed using SAS (SAS Institute Inc., Cary, North Carolina) and ST AT A software (StataCorp, College Station, Texas). 4.3 RESULTS After all exclusions, 55,759 postmenopausal women were included in the analysis. The mean length of follow-up was 7.3 years and a total of 1,624 incident invasive breast cancer cases were identified. The mean age at enrollment was 61.1 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. years. Latinas were more likely to have had a natural menopause (81.7%) and African-Americans were more likely to have had a bilateral oophorectomy (29.1%) than the other ethnic groups (Table 1). Hawaiians reached menarche at an earlier age (56.5% at <12 years old), than the other ethnicities. Whites were more likely to be nulliparous (17.0%) while J apanese-Americans had the latest age at first full- term pregnancy. Japanese-Americans were also by far the leanest, with 86.0% below the median cohort weight, and African-Americans were the heaviest, with 73.6% above the median. White women were much more likely to consume alcohol, 22.2% reported having one or more drinks per day, while most Japanese- Americans never drank alcohol (78.9%). J apanese-Americans (43.5%) and Whites (45.3%) were more likely to be current HT users, and African-Americans (55.9%) and Latinas (55.4%) were more likely to be never users. Table 2 shows that current EPT use was associated with an increased risk of breast cancer for all durations of use and risk was estimated to increase 30% per five years of use (HR = 1.30, 95% Confidence Interval (Cl) = 1.24, 1.36). Current ET use was associated with a 10% increase in risk of breast cancer per five years of use (HR = 1.10, 95% Cl = 1.04, 1.15). Past use was not associated with an increase in risk. As described in the Methods section, these HR estimates are adjusted for an additional duration of current use that continues during follow-up. For reference, we show the number of cases and person-years for women who never used of any type of HT in the tables. The actual reference for the HR estimates are never users for each type of HT, where each type of HT is modeled simultaneously 124 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. Descriptive characteristics of women included in the study (N=55,759). AA HA JA LA WH Total N .(%) N (%) N (%) N (%) N (%) N All women with natural menopause or bilateral oophorectomy No. of women 9,570 3,657 16956 11885 13691 55759 No. of cases 265 133 549 258 419 1624 Type of menopause Natural 6,788 (70.9) 2,767 (75.7) 13,450 (79.3) 9,708 (81.7) 10,436 (76.2) 43149 Bilateral oophorectomy 2,782 (29.1) 890 (24.3) 3,506 (20.7) 2,177 (18.3) 3,255 (23.8) 12610 Age at menarche (years) <12 4,641 (48.5) 2,067 (56.5) 7,834 (46.2) 5,555 (46.7) 6,645 (48.5) 26742 13+ 4,929 (51.5) 1,590 (43.5) 9,122 (53.8) 6,330 (53.3) 7,046 (54.5) 29017 Age at first birth (years) Never 1,240 (13.0) 277 (7.6) 2,261 (13.3) 1,053 (8.9) 2,320 (17.0) 7151 <20 4,294 (44.9) 1,586 (43.4) 1,499 (8.8) 4,589 (38.6) 3,220 (23.5) 15188 21-30 3,563 (37.2) 1,702 (46.5) 11,592 (68.4) 5,545 (46.7) 7,221 (52.7) 29623 31+ 473 (4.9) 92 (2.5) 1,604 (9.5) 698 (5.9) 930 (6.8) 3797 Children of parous women 1 1,500 (15.7) 233 (6.4) 1,841 (10.9) 841 (7.1) 1,640 (12.0) 6055 2-3 3,450 (36.1) 1,223 (33.4) 9,392 (55.4) 3,863 (32.5) 6,524 (47.7) 24452 4+ 3,380 (35.3) 1,924 (52.6) 3,462 (20.4) 6,128 (51.6) 3,207 (23.4) 18101 Weight (pounds) <145 2,529 (26.4) 1,368 (37.4) 14,573 (86.0) 5,313 (44.7) 7,258 (53.0) 31041 >145 7,041 (73.6) 2,289 (62.6) 2,383 (14.1) 6,572 (55.3) 6,433 (47.0) 24718 Alcohol (drinks/day) Never 5,962 (62.3) 2,343 (64.1) 13,370 (78.9) 7,571 (63.7) 5,499 (40.2) 34745 <1 2,721 (28.4) 965 (26.4) 3,059 (18.0) 3,733 (31.4) 5,157 (37.7) 15635 a 887 (9.3) 349 (9.5) 527 (3.1) 581 (4.9) 3,035 (22.2) 5379 Hormone therapy* Never 5,350 (55.9) 1,786 (48.8) 7,095 (41.8) 6,584 (55.4) 4,834 (35.3) 25649 Current EPT 712 (7.4) 595 (16.3) 4,251 (25.1) 1,387 (11.7) 3,455 (25.2) 10400 Current ET 1,297 (13.6) 586 (16.0) 3,123 (18.4) 1,596 (13.4) 2,757 (20.1) 9359 Past EPT 630 (6.6) 339 (9.3) 1,312 (7.7) 918 (7.7) 1,620 (11.8) 4819 Past ET 1,942 (20.3) 535 (14.6) 2,203 (13.0) 1,883 (15.8) 2,250 (16.4) 8813 WH = Whites, AA = African-Americans, HA = Hawaiians, JA = Japanese-Americans, LA = Latinas. *Each subject may contribute to more than one category of hormone therapy (HT) use. to Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 2. Crude and adjusted analyses of hormone therapy (HT) duration and breast cancer risk in the total population (N=55,759). Hazard Ratios (HRs) of risk per 5 years of use. Cases/ Person-years HR 95% Cl HR* 95% Cl HR*t 95% Cl Never1 646/ 19,9496 1.00 1.00 1.00 Current EPT >0 to <5 years 246/ 49,827 1.84 (1.59,2.14) 1.67 (1.43,1.95) 1.49 (1.11,2.00) 5 to <10 years 112/ 16,674 2.22 (1.82,2.71) 2.04 (1.67, 2.51) 1.86 (1.56,2.23) 10+ years 112/ 12,454 2.77 (2.27, 3.39) 2.66 (2.17,3.27) 2.17 (1.85,2.55) Per 5 years of use 1.42 (1.34, 1.50) 1.38 (1.30, 1.47) 1.30 (1.24, 1.36) Current ET >0 to <5 years 88/ 27,796 1.16 (0.92,1.46) 1.27 (0.99,1.63) 0.96 (0.59,1.58) 5 to <10 years 34/ 11,468 1.04 (0.73,1.47) 1.22 (0.84,1.76) 1.29 (0.96,1.73) 10+ years 141/ 33,482 1.27 (1.06,1.53) 1.58 (1.26,1.98) 1.50 (1.21,1.87) Per 5 years of use 1.05 (1.00,1.10) 1.12 (1.05, 1.19) 1.10 (1.04,1.15) Past EPT >0 to <5 years 104/ 33,487 1.04 (0.85,1.28) 1.01 (0.82, 1.24) 0.99 (0.80,1.22) 5 to <10 years 17/ 2,357 2.21 (1.35,3.60) 2.25 (1.38,3.68) 2.12 (1.27,3.53) 10+ years 5/ 1,536 0.92 (0.38, 2.23) 0.94 (0.39,2.28) 0.89 (0.36, 2.21) Per 5 years of use 1.09 (0.89,1.34) 1.12 (0.91,1.37) 1.12 (0.90,1.38) Past ET >0 to <5 years 180/ 54,121 0.96 (0.82,1.13) 1.05 (0.89,1.24) 1.01 (0.85,1.2) 5 to <10 years 31/ 6,880 0.86 (0.60,1.23) 0.92 (0.64,1.32) 0.89 (0.61,1.29) 10+ years 26/ 7,219 0.74 (0.50, 1.09) 0.88 (0.59, 1.31) 0.83 (0.55, 1.26) Per 5 years of use 0.93 (0.83,1.04) 0.99 (0.89,1.11) 0.96 (0.85,1.08) Each subject may contribute to more than one category of use. Analyses are simultaneously adjusted for the other categories of HT use. * Adjusted for ethnicity, age at menarche, age at first foil term pregnancy, number of children, age and type of menopause, weight, and alcohol consumption. t Adjusted for current use that continues until censoring time; an additional duration of use that is equal to the duration of follow-up for the matched case was added to current EPT or current ET use. * For simplicity, never use of any type of HT use is shown here. The actual reference for these HR estimates are never users for each type of HT, where each type of HT is modeled simultaneously as a continuous variable equal to the duration of use or to zero (never use). is} as a continuous variable equal to the duration of use in years or to zero (never use). The overall frequency of PT use was very small and is not included in the tables. The increase in risk associated with current HT use was observed across all ethnic groups (Table 3). Current EPT use was associated with a statistically significant increase in risk among Japanese-American (HR = 1.37, 95% Cl = 1.27,1.47), Latina (HR = 1.33, 95% Cl =1.17, 1.51), and White (HR = 1.26, 95% Cl = 1.17, 1.37) women. There was a slight increase in risk associated with current ET use for all groups except for African-Americans; this increase was statistically significant for Hawaiians (HR = 1.25, 95% Cl = 1.02, 1.52) and J apanese-Americans (HR = 1.25, 95% Cl = 1.02,1.52). However, these differences by ethnicity for current EPT (pheterogeneity(het) = 0.272), or current ET (p het = 0.117) use were not statistically significant. We observed a significant increase in risk among African- Americans for past EPT use (HR = 1.97, 95% Cl = 1.34, 2.90); we also observed this effect when we did not adjust for an additional duration of current use (HR = 1.79, 95% Cl = 1.30, 2.46). Past ET use was not associated with an increase in risk of breast cancer for any of the ethnic groups. The association between duration of HT use and breast cancer risk varied by weight but not by stage of disease or histologic subtype (Tables 4). A subanalysis comparing pure lobular versus ductal carcinomas did not change the results. Hazard ratios for lean women were greater than for heavy women for all types of HT use, and these differences were statistically significant for current EPT use (diff = +0.14, p =0.035). However, heavier women still had a greater absolute 127 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 3. Hazard Ratios (HRs) of breast cancer risk and hormone therapy (HT) per 5 years of use by ethnicity. AA Cases/ Person-yrs H R ** (95%CI) HA Cases/ Person-yrs H R ** (95%C1) JA Cases/ Person-yrs HR** (95%CI) Never8 145/ 44,258 1.00 64/ 12,403 1.00 177/52,489 1.00 Current EPT >0 to <5 years 5+ years 15/ 4,482 10/ 1,727 1.51 1.27 (0.63, 3.61) (0.78,2.08) 18/ 3,227 10/ 1,043 0.63 1.69 (0.15, 2.71) (1.04,2.74) 113/18,876 105/ 12,347 1.52 2.33 (0.97,2.39) (1.89,2.86) Per S years use 1.18 (0.99, 1.40) 1.14 (0.92, 1.41) 1.37 (1.27, 1.47) Current ET >0 to <5 years 5+ years 16/ 4,515 19/ 6,611 1.99 1.06 (0.67, 5.95) (0.66, 1.72) 5/ 2,147 13/ 2,119 0.38 1.94 (0.05, 3.05) (0.93,4.01) 22/ 8,082 76/15,143 0.52 1.58 (0.16, 1.70) (1.09, 2.29) Per 5 years use 1.00 (0.88, 1.13) 1.25 (1.02, 1.52) 1.19 (1.08,1.31) LA Cases/ Person-yrs H R ** (95%CI) WH Cases/ Person-yrs HR** (95%CI) Never8 129/ 53,668 1.00 131/36,678 1.00 Current EPT >0 to <5 years 5+ years 25/ 8,000 21/ 3,393 1.23 2.02 (0.48,3.13) (1.40,2.92) 75/ 15,244 78/10,618 2.11 1.99 (1.22,3.63) (1.56,2.54) Per 5 years use 1.33 (1.17,1.51) 1.26 (1.17,1.37) Current ET >0 to <5 years 5+ years 21/ 6,270 17/ 6,927 0.83 1.68 (0.25, 2.73) (1.05,2.69) 24/ 6,782 50/14,150 1.87 1.31 (0.83,4.25) (0.89,1.92) Per 5 years use 1.08 (0.94, 1.23) 1.05 (0.96,1.16) to 00 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 3 continued. AA Cases/ Person-yrs HR** (95%CI) HA Cases/ Person-yrs HR** (95%CI) JA Cases/ Person-yrs HR** (95%CI) Past EPT Any past use 25/ 5,376 1.50 (0.96, 2.34) 13/ 2,423 1.16 (0.62, 2.19) 37/ 9,697 1.06 (0.74,1.52) Per 5 years use 1.97 (1.34,2.90) 1.22 (0.53, 2.81) 0.91 (0.58,1.42) Past ET Any past use 54/ 16,096 1.02 (0.72, 1.45) 22/ 3,765 1.24 (0.73,2.12) 52/ 16,200 0.82 (0.61,1.12) Per 5 years use 0.85 (0.62, 1.16) 1.02 (0.67,1.56) 0.96 (0.76,1.21) LA Cases/ Person-yrs HR** (95%CI) WH Cases/ Person-yrs HR** (95%CI) Past EPT Any past use Per 5 years use 15/ 7,545 0.83 0.33 (0.48, 1.44) (0.08,1.38) 36/ 12,339 0.93 1.11 (0.65, 1.35) (0.76,1.62) Past ET Any past use 45/ 15,399 1.15 (0.8, 1.63) 64/ 16,761 0.88 (0.65,1.18) Per 5 years use 1.13 (0.84,1.52) 0.91 (0.73,1.13) WH = Whites, AA = African-Americans, HA = Hawaiians, JA = Japanese-Americans, LA = Latinas, yr = year. Each subject may contribute to more than one category of use. Analyses are simultaneously adjusted for the other categories of HT use. * Adjusted for age at menarche, age at first full term pregnancy, number of children, age and type of menopause, weight, and alcohol consumption. * Adjusted for ethnicity, age at menarche, age at first full term pregnancy, number of children, age and type of menopause, weight, and alcohol consumption * Adjusted for current use that continues until censoring time; an additional duration of use that is equal to the duration of follow-up for the matched case was added to current EPT or current ET use. § For simplicity, never use of any type of HT use is shown here. The actual reference for these HR estimates are never users for each type of HT, where each type of HT is modeled simultaneously as a continuous variable equal to the duration o f use or to zero (never use). to VO Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4. Hazard Ratios (HRs) of breast cancer risk and duration of hormone therapy (HT) per 5 years of use, stratified by Cases/ Person-yrs HR** (95% Cl) Cases/ Person-yrs HR*f (95% Cl) Weight + Lean (^ 4 5 lbs) Heavy (>145 lbs) Never* 291/ 101,682 1.00 355/ 97,815 1.00 Current EPT 327/ 52,706 1.35 (1.27, 1.43) 143/26,250 1.21 (1.11, 1.31) Current ET 155/ 41,812 1.13 (1.05, 1.21) 108/ 30,934 1.07 (0.99, 1.15) Past EPT 71/ 20,037 1.16 (0.90, 1.51) 55/ 17,343 0.97 (0.67, 1.41) Past ET 119/ 35,282 1.01 (0.87, 1.18) 118/ 32,939 0.86 (0.70, 1.05) Stage Localized Advanced Never* 442/ 1,897 1.00 174/ 779 1.00 Current EPT 369/ 1,543 1.33 (1.26, 1.40) 94/ 417 1.25 (1.13, 1.39) Current ET 195/ 821 1.10 (1.04, 1.17) 62/ 246 1.09 (0.98, 1.21) Past EPT 95/ 393 1.12 (0.88, 1.43) 27/ 109 1.13 (0.72, 1.79) Past ET 173/ 713 0.99 (0.86, 1.13) 55/ 219 0.93 (0.71, 1.21) Histology Never* Lobular Ductal 84/ 416 1.00 470/ 1,999 1.00 Current EPT 66/ 300 1.32 (1.16, 1.50) 352/ 1,449 1.29 (1.22, 1.36) Current ET 36/ 179 1.13 (0.99, 1.28) 184/ 754 1.09 (1.02, 1.16) Past EPT 24/ 107 1.58 (1.01, 2.48) 84/ 349 1.07 (0.82, 1.39) Past ET 39/ 190 1.10 (0.83, 1.46) 169/ 660 0.98 (0.85, 1.14) Yr = year. Each subject may contribute to more than one category of use. Analyses are simultaneously adjusted for the other categories of HT use. * Adjusted for ethnicity, age at menarche, age at first full term pregnancy, number of children, age and type of menopause, and alcohol consumption. * Adjusted for current use that continues until censoring time; an additional duration of use that is equal to the duration of follow-up for the matched case was added to current EPT or current ET use. * For simplicity, never use of any type of HT use is shown here. The actual reference for these HR estimates are never users for each type of HT, where each type of HT is modeled simultaneously as a continuous variable equal to the duration of use or to zero (never use). u > o risk of breast cancer than lean women for all types of HT use (Table 5). These differences in risk by weight were observed across all ethnic groups. (Note: the fitted model included weight but not height or body mass index (BMI); after fitting weight, neither height nor BMI was statistically significant, and weight alone provided the best fit of these variables considered separately). The effect of HT on breast cancer risk was effectively confined to estrogen receptor-positive and progesterone receptor-positive (ER+/PR+) tumors (Table 6). Current EPT (HR = 1.38, 95% Cl = 1.30, 1.47) and current ET (HR = 1.11, 95% Cl = 1.03, 1.19) use were both significantly associated with ER+/PR+ tumors. We also conducted stratified analyses by women whose age at menopause was defined by their start of HT use and women whose age at menopause was defined by the cessation of menses. However, neither the differences in risk by specific type of HT use nor the overall difference in risk between the two groups were statistically significant (data not shown). 4.4 DISCUSSION Findings from this study are consistent with previous literature of an association between HT use and breast cancer, in particular, an increase in risk associated with current EPT use. We observed a 30% increase in breast cancer risk per 5 years of use of current EPT and a 10% increase in risk per 5 years of current ET use. These increases in risk were observed across all ethnic groups, and not just among White women. We observed a significant increase in risk among African- 131 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 5. Hazard Ratios (HRs) of breast cancer risk and duration of hormone therapy (HT) per 5 years of use, stratified by weight using a common reference group. Cases/ Person-yrs H R * 1 - (95% Cl) diff P? Never <145 lbs 291/ 10,1682 1.00 — > 1 4 5 lbs 355/ 97,815 1.35 (1.17, 1.55) -0.35 <0.0001 Current EPT * 5 1 4 5 lbs 327/ 52,706 1.32 (1.25, 1.40) >145 lbs 143/ 26,250 1.66 (1.34, 2.07) -0.34 0.046 Current ET < 1 4 5 lbs 155/ 41,812 1.11 (1.04, 1.18) >145 lbs 108/ 30,934 1.44 (1.17,1.78) -0.33 0.020 Past EPT *5 1 4 5 lbs 71/ 20,037 1.20 (0.93, 1.55) >145 lbs 55/ 17,343 1.33 (0.80, 2.21) -0.13 0.72 Past ET < 1 4 5 lbs 119/ 35,282 1.01 (0.87, 1.17) >145 lbs 118/ 32,939 1.19 (0.85, 1.66) -0.18 0.38 Diff = difference in risk among lean versus heavy women, yr — year. Each subject may contribute to more than one category of use. Analyses are simultaneously adjusted for the other categories of HT use. * Adjusted for ethnicity, age at menarche, age at first full term pregnancy, number of children, age and type of menopause, and alcohol consumption. * Adjusted for current use that continues until censoring time; an additional duration of use that is equal to the duration of follow-up for the matched case was added to current EPT or current ET use. *P value from likelihood ratio test. to Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 6. Hazard ratios (HRs) of breast cancer risk and duration of hormone therapy (HT) per 5 years of use, stratified by estrogen receptor (ER) and progesterone receptor (PR) status. ER+/PR+ ER+/PR- Cases/ Person-yrs HR’ d (95% Cl) Cases/ Person-yrs HR*t (95% CD Never* 379/ 1,304 1.00 — 86/ 246 1.00 — Current EPT 361/1,147 1.38 (1.30, 1.47) 50/171 1.12 (0.96, 1.30) Current ET 164/ 513 1.11 (1.03, 1.19) 34/120 1.12 (0.95, 1.32) Past EPT 87/ 263 1.13 (0.82, 1.54) 14/46 0.75 (0.31, 1.81) Past ET 167/ 495 1.10 (0.94, 1.28) 33/117 1.06 (0.76, 1.47) ER-/PR+ ER-/PR- Cases/ Person-yrs HR*f (95% CD Cases/ Person-yrs HR*f (95% CD Never* 14/ 39 1.00 — 135/473 1.00 — Current EPT 8/ 22 1.12 (0.73,1.71) 62/174 1.04 (0.89, 1.22) Current ET 12/46 0.95 (0.63, 1.43) 52/142 1.01 (0.88, 1.16) Past EPT 4/22 0.92 (0.24, 3.64) 22/64 0.82 (0.37, 1.83) Past ET 2/3 0.04 (0.00,4.01) 26/105 0.71 (0.45, 1.11) Diff = difference in risk among ER+/PR+ and other receptor status tumors, yr = year. Each subject may contribute to more than one category of use. Analyses are simultaneously adjusted for the other categories of HT use. * Adjusted for ethnicity, age at menarche, age at first full term pregnancy, number of children, age and type of menopause, weight, and alcohol consumption. Adjusted for current use that continues until censoring time; an additional duration of use that is equal to the duration of follow-up for the matched case was added to current EPT or current ET use. * For simplicity, never use of any type of HT use is shown here. The actual reference for these HR estimates are never users for each type of HT, where each type of HT is modeled simultaneously as a continuous variable equal to the duration of use or to zero (never use). Americans for past EPT use which remained significant whether or not we accounted for an additional duration of current use. This effect was based on 25 cases; we are unable to make any conclusions at this time and will investigate this further. We observed an increase in risk for short-term current HT use that was greater than would be expected for such a brief duration of exposure. We believed that this effect was influenced by the underestimation of the duration of use among current users, and thus assigned an additional duration of use for current users. This method did not significantly alter the estimates, but we included it in the model as our best attempt to address this potential bias. The relative risks associated with HT use, especially current use, were more pronounced among lean women. However, the absolute risk in heavy women was still greater than that of lean women given the same type of HT use. When we stratified this analysis by ethnicity, the effect of weight on the association between HT use and breast cancer risk was observed in all ethnic groups. There was no apparent difference in breast cancer risk by stage of disease. These findings are in contrast to the results from the WHI trial1 1 where EPT use was associated with an increase in regional (25.4%) and metastatic disease (16.0%) compared to placebo. However, other observational studies have reported an increase in localized disease due to HT use1,4 or found no difference in risk by stage.6 However, observational studies may not adequately adjust for surveillance 134 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. bias while results from a randomized clinical trial may not be representative of the overall population. We did not observe a difference in breast cancer risk by histologic subtype; the risks of both lobular and ductal carcinomas were increased with current use. These findings are consistent with earlier observational studies that did not find a difference in risk by histologic subtype associated with HT use4,6’1 9 Restricting the analysis to pure lobular subtypes only (excluding the 95 cases of mixed lobular carcinomas) did not change the results. Current EPT and ET use were statistically significantly associated only with ER+/PR+ tumors. These findings are consistent with earlier studies reporting a greater proportion of ER+ and PR+ tumors among HT users.1 9 '2 3 Stratified analyses by the definition of a woman’s age at menopause demonstrated that risks were not statistically significantly different among women whose at age menopause was defined by the start of HT and women whose age at menopause was defined by the cessation of menses. This supports our method of assigning age at menopause to women who start HT use before they are clearly menopausal and allowing them to be included in the analysis of HT use and breast cancer risk. A strength of this study is the ability to evaluate breast cancer risk and HT use among a multiethnic prospective study. Data on tumor characteristics were obtained from the SEER registry rather than from pathology reports; however, this data undergoes quality control measures to ensure accuracy of information 135 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (http ://seer. cancer .go v/about/quality.html). The results for hormone receptor status should be interpreted with caution, since there is great variability in the techniques, quantitation methods, and quality control measures in determining estrogen receptor and progesterone receptor status,2 4 ’2 5 and findings are reported to the SEER registry simply as recorded. This study provides some of the first results comparing breast cancer risk among different populations in relation to HT use. These results suggest that risk among women of other races is similar to the risk previously reported among Whites. In addition, these findings are consistent with prior findings of differences in risk by weight and hormone receptor status and encourage further studies evaluating these risks. 136 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.5 REFERENCES 1. Collaborative Group on Hormonal Factors in Breast Cancer. Breast cancer and hormone replacement therapy: collaborative reanalysis of data from 51 epidemiological studies of 52,705 women with breast cancer and 108,411 women without breast cancer. Lancet. 1997;350:1047-1059. 2. Magnusson C, Baron JA, Correia N, Bergstrom R, Adami HO, Persson I. Breast-cancer risk following long-term oestrogen- and oestrogen-progestin- replacement therapy. Int J Cancer. May 5 1999;81 (3):339-344. 3. Ross RK, Paganini-Hill A, Wan PC, Pike MC. Effect of hormone replacement therapy on breast cancer risk: estrogen versus estrogen plus progestin. J Natl Cancer Inst. Feb 16 2000;92(4):328-332. 4. Schairer C, Lubin J, Troisi R, Sturgeon S, Brinton L, Hoover R. Menopausal estrogen and estrogen-progestin replacement therapy and breast cancer risk. Jama. Jan 26 2000;283(4):485-491. 5. Kirsh V, Kreiger N. Estrogen and estrogen-progestin replacement therapy and risk of postmenopausal breast cancer in Canada. Cancer Causes Control. Aug 2002;13(6):583-590. 6. Newcomb PA, Titus-Emstoff L, Egan KM, et al. Postmenopausal estrogen and progestin use in relation to breast cancer risk. Cancer Epidemiol Biomarkers Prev. Jul 2002; 11(7):593-600. 7. Porch JV, Lee IM, Cook NR, Rexrode KM, Burin JE. Estrogen-progestin replacement therapy and breast cancer risk: the Women's Health Study (United States). Cancer Causes Control. Nov 2002;13(9):847-854. 8. Weiss LK, Burkman RT, Cushing-Haugen KL, et al. Hormone replacement therapy regimens and breast cancer risk(l). Obstet Gynecol. Dec 2002; 100(6): 1148-1158. 9. Million Women Study. Breast cancer and hormone-replacement therapy in the Million Women Study. Lancet. Aug 9 2003;362(9382):419-427. 10. Stahlberg C, Pedersen AT, Lynge E, et al. Increased risk of breast cancer following different regimens of hormone replacement therapy frequently used in Europe. Int J Cancer. May 1 2004; 109(5):721 -727. 137 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 11. Chlebowski RT, Hendrix SL, Langer RD, et al. Influence of estrogen plus progestin on breast cancer and mammography in healthy postmenopausal women: the Women's Health Initiative Randomized Trial. Jama. Jun 25 2003;289(24):3243-3253. 12. Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women's Health Initiative randomized controlled trial. Jama. Jul 17 2002;288(3):321-333. 13. Lee SA, Ross RK, Pike MC. An overview of postmenopausal estrogen- progestin hormone therapy and breast cance risk, in review. 14. Kolonel LC, Henderson BE, Hankin JH, et al. A multiethnic cohort in Hawaii and Los Angeles: Baseline Characteristics. American Journal o f Epidemiology. 2000;151(4):346-357. 15. SEER. Summary Staging Guide', April, 1977. 16. Pike MC, Peters RK, Cozen W, et al. Estrogen-progestin replacement therapy and endometrial cancer. JNatl Cancer Inst. Aug 6 1997;89(15):1110-1116. 17. Pike MC, Ross RK, Spicer DV. Problems involved in including women with simple hysterectomy in epidemiologic studies measuring the effects of hormone replacement therapy on breast cancer risk. Am J Epidemiol. April 15 1998;147(8):718-721. 18. Rockhill B, Colditz GA, Rosner B. Bias in breast cancer analyses due to error in age at menopause. Am J Epidemiol. Feb 15 2000; 151 (4) :404-408. 19. Ursin G, Tseng CC, Paganini-Hill A, et al. Does menopausal hormone replacement therapy interact with known factors to increase risk of breast cancer? J Clin Oncol. Feb 1 2002;20(3):699-706. 20. Chen CL, Weiss NS, Newcomb P, Barlow W, White E. Hormone replacement therapy in relation to breast cancer. Jama. Feb 13 2002;287(6):734-741. 21. Gajdos C, Tartter PI, Babinszki A. Breast cancer diagnosed during hormone replacement therapy. Obstet Gynecol. Apr 2000;95(4):513-518. 138 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22. Kerlikowske K, Miglioretti DL, Ballard-Barbash R, et al. Prognostic characteristics of breast cancer among postmenopausal hormone users in a screened population. J Clin Oncol. Dec 1 2003 ;21 (23):4314-4321. 23. Li Cl, Malone KE, Porter PL, et al. Relationship between long durations and different regimens of hormone therapy and risk of breast cancer. Jama. Jun25 2003;289(24):3254-3263. 24. Althuis MD, Fergenbaum JH, Garcia-Closas M, Brinton LA, Madigan MP, Sherman ME. Etiology of hormone receptor-defined breast cancer: a systematic review of the literature. Cancer Epidemiol Biomarkers Prev. Oct 2004;13(10):1558-1568. 25. Layfield LJ, Gupta D, Mooney EE. Assessment of Tissue Estrogen and Progesterone Receptor Levels: A Survey of Current Practice, Techniques, and Quantitation Methods. Breast J. May 2000;6(3):189-196. 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. COMPREHENSIVE BIBLIOGRAPHY Althuis MD, Fergenbaum JH, Garcia-Closas M, Brinton LA, Madigan MP, Sherman ME. Etiology of hormone receptor-defined breast cancer: a systematic review of the literature. Cancer Epidemiol Biomarkers Prev. Oct 2004;13(10): 1558-1568. Anderson GL, Limacher M, Assaf AR, et al. Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women's Health Initiative randomized controlled trial. Jama. Apr 14 2004;291(14):1701-1712. Arden KC, Boutin JM, Djiane J, Kelly PA, Cavenee WK. The receptors for prolactin and growth hormone are localized in the same region of human chromosome 5. Cytogenet Cell Genet. 1990;53(2-3): 161-165. Back DJ, Bates M, Breckenridge AM, et al. The pharmacokinetics of levonorgestrel and ethynylestradiol in women - studies with Ovran and Ovranette. Contraception. Mar 1981 ;23(3):229-239. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics. Dec 1994;50(4): 1088-1101. Ben-Jonathan N, Liby K, McFarland M, Zinger M. Prolactin as an autocrine/paracrine growth factor in human cancer. Trends Endocrinol Metab. Aug 2002; 13(6):245-250. Bernstein L, Ross RK, Pike MC, Brown JB, Henderson BE. Hormone levels in older women: a study of post-menopausal breast cancer patients and healthy population controls. Br J Cancer. 1990;61(2):298-302. Bernstein L, Ross RK. Endogenous hormones and breast cancer risk. Epidemiol Rev. 1993;15(l):48-65. Berwaer M, Martial JA, Davis JR. Characterization of an up-stream promoter directing extrapituitary expression of the human prolactin gene. Mol Endocrinol. 1994;8(5):635-642. Biswas R, Vonderhaar BK. Role of serum in the prolactin responsiveness of MCF- 7 human breast cancer cells in long-term tissue culture. Cancer Res. 1987;47(13):3509-3514. Boot LM, Muhlbock O, Ropcke G. Prolactin and the induction of mammary tumors in mice. General and Comparative Endocrinology. 1962;2:601-602. 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Brisken C, Kaur S, Chavarria TE, et al. Prolactin controls mammary gland development via direct and indirect mechanisms. Dev Biol. 1999;210(1):96-106. Buytaert P, Viaene P. Amenorrhea, galactorrhea, hyperprolactinemia syndrome and breast carcinoma in a young woman. Eur J Ohstet Gynecol Reprod Biol. 1981;ll(5):341-346. Calaf G, Garrido F, Moyano C, Rodriguez R. Influence of hormones on DNA synthesis of breast tumors in culture. Breast Cancer Res Treat. 1986;8(3):223-232. Canbay E, Degerli N, Gulluoglu BM, Kaya H, Sen M, Bardakci F. Could prolactin receptor gene polymorphism play a role in pathogenesis of breast carcinoma? Curr Med Res Opin. Apr 2004;20(4):533-540. Cargill M, Altshuler D, Ireland J, et al. Characterization of single-nucleotide polymorphisms in coding regions of human genes. Nat Genet. 1999;22(3):231-238. Carlson CS, Eberle MA, Rieder MJ, Yi Q, Kruglyak L, Nickerson DA. Selecting a maximally informative set of single-nucleotide polymorphisms for association analyses using linkage disequilibrium. Am J Hum Genet. Jan 2004;74(1): 106-120. Chen CL, Weiss NS, Newcomb P, Barlow W, White E. Hormone replacement therapy in relation to breast cancer. Jama. Feb 13 2002;287(6):734-741. Chlebowski RT, Hendrix SL, Langer RD, et al. Influence of estrogen plus progestin on breast cancer and mammography in healthy postmenopausal women: the Women's Health Initiative Randomized Trial. Jama. Jun 25 2003;289(24):3243- 3253. Clevenger CV, Chang WP, Ngo W, Pasha TL, Montone KT, Tomaszewski JE. Expression of prolactin and prolactin receptor in human breast carcinoma. Evidence for an autocrine/paracrine loop. Am J Pathol. Mar 1995; 146(3) :695-705. Clevenger CV, Furth PA, Hankinson SE, Schuler LA. The role of prolactin in mammary carcinoma. Endocr Rev. 2003;24(l):l-27. Colditz GA, Hankinson SE, Hunder DJ, et al. The use of estrogens and progestins and the risk of breast cancer in postmenopausal women. N Engl J Med. June 15 1995;332(24): 1589-1593. Cole EN, England PC, Sellwood RA, Griffiths K. Serum prolactin concentrations throughout the menstrual cycle of normal women and patients with recent breast cancer. Eur J Cancer. 1977;13(7):677-684. 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Collaborative Group on Hormonal Factors in Breast Cancer. Breast cancer and hormone replacement therapy: collaborative reanalysis of data from 51 epidemiological studies of 52,705 women with breast cancer and 108,411 women without breast cancer. Lancet. 1997;350:1047-1059. Cowie AT, Tindal JS, Yokoyama A. The induction of mammary growth in the hypophysectomiized goat. J Endocrinol. 1966;34:185-195. Crawford DC, Carlson CS, Rieder MJ, et al. Haplotype diversity across 100 candidate genes for inflammation, lipid metabolism, and blood pressure regulation in two populations. Am J Hum Genet. Apr 2004;74(4):610-622. Daling JR, Malone KE, Doody DR, et al. Relation of regimens of combined hormone replacement therapy to lobular, ductal, and other histologic types of breast carcinoma. Cancer. Dec 15 2002;95(12):2455-2464. DerSimonian R, Laird N. Meta-analysis in clinical trial. Control Clin Trials. September 7 1986;7(3):177-188. Dhadly MS, Walker RA. The localization of prolactin binding sites in human breast tissue. Int J Cancer. 1983;31(4):433-437. Dickey RP, Stone SC. Progestational potency of oral contraceptives. Obstet Gynecol. Jan 1976;47(1): 106-112. DiMattia GE, Gellersen B, Duckworth ML, Friesen HG. Human prolactin gene expression. The use of an alternative noncoding exon in decidua and the IM-9-P3 lymphoblast cell line. J Biol Chem. 1990;265(27): 16412-16421. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. Bmj. Sep 13 1997;315(7109):629-634. Ettinger B, Grady D, Tosteson AN, Pressman A, Macer JL. Effect of the women's Health Initiative on women's decisions to discontinue postmenopausal hormone therapy. Obstet Gynecol. December 2003; 102(6): 1225-1232. Excoffier L, Slatkin M. Maximum-likelihood estimation of molecular haplotype frequencies in a diploid population. Mol Biol Evol. 1995; 12(5):921 -927. Fleiss JL. Analysis of data from multiclinic trials. Control Clin Trials. December 7 1986;7(4):267-275. Gabriel SB, Schaffner SF, Nguyen H, et al. The structure of haplotype blocks in the human genome. Science. 2002;296(5576):2225-2229. 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Gabriel SB, Schaffner SF, Nguyen H, et al. The structure of haplotype blocks in the human genome. Science. 2002;296(5576):2225-2229. Gajdos C, Tartter PI, Babinszki A. Breast cancer diagnosed during hormone replacement therapy. Obstet Gynecol. Apr 2000;95(4):513-518. Glasow A, Horn LC, Taymans SE, et al. Mutational analysis of the PRL receptor gene in human breast tumors with differential PRL receptor protein expression. J Clin Endocrinol Metab. Aug 2001;86(8):3826-3832. Greendale GA, Reboussin BA, Slone S, Wasilauskas C, Pike MC, Ursin G. Postmenopausal hormone therapy and change in mammographic density. J Natl Cancer Inst. Jan 1 2003;95(l):30-37. Gutzman JH, Miller KK, Schuler LA. Endogenous human prolactin and not exogenous human prolactin induces estrogen receptor alpha and prolactin receptor expression and increases estrogen responsiveness in breast cancer cells. J Steroid Biochem Mol Biol. Jan 2004;88(l):69-77. Haga S, Watanabe O, Shimizu T, et al. Breast cancer in a male patient with prolactinoma. Surg Today. 1993;23(3):251-255. Haiman CA, Stram DO, Pike MC, et al. A comprehensive haplotype analysis of CYP19 and breast cancer risk: the Multiethnic Cohort. Hum Mol Genet. 2003;12(20):2679-2692. Hankinson SE, London SJ, Chute CG, et al. Effect of transport conditions on the stability of biochemical markers in blood. Clin Chem. 1989;35(12):2313-2316. Hankinson SE, Willett WC, Michaud DS, et al. Plasma prolactin levels and subsequent risk of breast cancer in postmenopausal women. J Natl Cancer Inst. 1999;91 (7):629-634. Haus E, Lakatua DJ, Halberg F, et al. Chronobiological studies of plasma prolactin in women in Kyushu, Japan, and Minnesota, USA. J Clin Endocrinol Metab. 1980;51(3):632-640. Hayden TJ, Bonney RC, Forsyth IA. Ontogeny and control of prolactin receptors in the mammary gland and liver of virgin, pregnant and lactating rats. J Endocrinol. 1979;80(2):259-269. Helzlsouer KJ, Alberg AJ, Bush TL, Longcope C, Gordon GB, Comstock GW. A prospective study of endogenous hormones and breast cancer. Cancer Detect Prev. 1994;18(2):79-85. 143 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Henderson BE, Ross RK, Pike MC. Hormonal chemoprevention of cancer in women. Science. Jan 29 1993;259(5095):633-638. Henderson BR, Gerkins V, Rosario I, Casagrande J, Pike MC. Elevated serum levels of estrogen and prolactin in daughters of patients with breast cancer. N Engl J Med. 1975;293(16):790-795. Hennighausen L, Robinson GW, Wagner KU, Liu W. Prolactin signaling in mammary gland development. J Biol Chem. 1997;272(12):7567-7569. Holtkamp W, Nagel GA, Wander HE, Rauschecker HF, von Heyden D. Hyperprolactinemia is an indicator of progressive disease and poor prognosis in advanced breast cancer. Int J Cancer. 1984;34(3):323-328. Hu ZZ, Meng J, Dufau ML. Isolation and characterization of two novel forms of the human prolactin receptor generated by alternative splicing of a newly identified exon 11. J Biol Chem. 2001;276(44):41086-41094. Hu ZZ, Zhuang L, Meng J, Leondires M, Dufau ML. The human prolactin receptor gene structure and alternative promoter utilization: the generic promoter hPIII and a novel human promoter hP(N). J Clin Endocrinol Metab. 1999;84(3): 1153-1156. Hu ZZ, Zhuang L, Meng J, Tsai-Morris CH, Dufau ML. Complex 5' genomic structure of the human prolactin receptor: multiple alternative exons 1 and promoter utilization. Endocrinology. Jun 2002; 143(6) :2139-2142. Ingram DM, Nottage EM, Roberts AN. Prolactin and breast cancer risk. Med J Aust. 1990; 153(8):469-473. Jahn GA, Edery M, Belair L, Kelly PA, Djiane J. Prolactin receptor gene expression in rat mammary gland and liver during pregnancy and lactation. Endocrinology. 1991; 128(6):2976-2984. Jeffreys AJ, Holloway JK, Kauppi L, et al. Meiotic recombination hot spots and human DNA diversity. Philos Trans R Soc Lond B Biol Sci. Jan 29 2004;359(1441): 141-152. Jeffreys AJ, Kauppi L, Neumann R. Intensely punctate meiotic recombination in the class II region of the major histocompatibility complex. Nat Genet. Oct 2001;29(2):217-222. Jemstrom H, Bendahl PO, Lidfeldt J, Nerbrand C, Agardh CD, Samsioe G. A prospective study of different types of hormone replacement therapy use and the risk of subsequent breast cancer: the women's health in the Lund area (WHILA) study (Sweden). Cancer Causes Control. Sep 2003;14(7):673-680. 144 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kabuto M, Akiba S, Stevens RG, Neriishi K, Land CE. A prospective study of estradiol and breast cancer in Japanese women. Cancer Epidemiol Biomarkers Prev. 2000;9(6):575-579. Kauppi L, Jeffreys AJ, Keeney S. Where the crossovers are: recombination distributions in mammals. Nat Rev Genet. Jun 2004;5(6):413-424. Kauppi L, Sajantila A, Jeffreys AJ. Recombination hotspots rather than population history dominate linkage disequilibrium in the MHC class II region. Hum Mol Genet. Jan 1 2003;12(l):33-40. Ke X, Cardon LR. Efficient selective screening of haplotype tag SNPs. Bioinformatics. 2003;19:287-288. Kelly PA, Djiane J, Postel-Vinay MC, Edery M. The prolactin/growth hormone receptor family. Endocr Rev. 1991;12(3):235-251. Kerlikowske K, Miglioretti DL, Ballard-Barbash R, et al. Prognostic characteristics of breast cancer among postmenopausal hormone users in a screened population. J Clin Oncol. Dec 1 2003;21 (23) :4314-4321. Kirsh V, Kreiger N. Estrogen and estrogen-progestin replacement therapy and risk of postmenopausal breast cancer in Canada. Cancer Causes Control. Aug 2002;13(6):583-590. Kledzik GS, Bradley CJ, Marshall S, Campbell GA, Meites J. Effects ofhigh hoses of estrogen on prolactin-binding activity and growth of carcinogen-induced mammary cancers in rats. Cancer Res. 1976;36(9 pt.l):3265-3268. Kolonel LC, Henderson BE, Hankin JH, et al. A multiethnic cohort in Hawaii and Los Angeles: Baseline Characteristics. American Journal o f Epidemiology. 2000;151(4):346-357. Kruglyak L, Nickerson DA. Variation is the spice of life. Nat Genet. 2001;27(3):234-236. Layfield LJ, Gupta D, Mooney EE. Assessment of Tissue Estrogen and Progesterone Receptor Levels: A Survey of Current Practice, Techniques, and Quantitation Methods. Breast J. May 2000;6(3):189-196. Lee SA, Ross RK, Pike MC. An overview of postmenopausal estrogen-progestin hormone therapy and breast cance risk, in review. Li Cl, Malone KE, Porter PL, et al. Relationship between long durations and different regimens of hormone therapy and risk of breast cancer. Jama. Jun 25 2003;289(24):3254-3263. 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Li Cl, Weiss NS, Stanford JL, Dating JR. Hormone replacement therapy in relation to risk of lobular and ductal breast carcinoma in middle-aged women. Cancer. Jun 1 2000;88(11):2570-2577. Li R, Gilliland FD, Baumgartner K, Samet J. Hormone replacement therapy and breast carcinoma risk in Hispanic and non-Hispanic women. Cancer. Sep 1 2002;95(5):960-968. Liby K, Neltner B, Mohamet L, Menchen L, Ben-Jonathan N. Prolactin overexpression by MDA-MB-435 human breast cancer cells accelerates tumor growth. Breast Cancer Res Treat. May 2003;79(2):241-252. Love RR, Rose DR, Surawicz TS, Newcomb PA. Prolactin and growth hormone levels in premenopausal women with breast cancer and healthy women with a strong family history of breast cancer. Cancer. 1991 ;68(6): 1401 -1405. Lyons WR, Li CH, Johnson RE. The hormonal control of mammary growth and lactation. Recent Prog Horm Res. 1958;14:219-254. Magnusson C, Baron JA, Correia N, Bergstrom R, Adami HO, Persson I. Breast- cancer risk following long-term oestrogen- and oestrogen-progestin-replacement therapy. Int J Cancer. May 5 1999;81(3):339-344. Malarkey WB, Kennedy M, Allred LE, Milo G. Physiological concentrations of prolactin can promote the growth of human breast tumor cells in culture. J Clin Endocrinol Metab. 1983;56(4):673-677. Malarkey WB, Schroeder LL, Stevens VC, James AG, Lanese RR. Disordered nocturnal prolactin regulation in women with breast cancer. Cancer Res. 1977;37(12):4650-4654. Manjer J, Johansson R, Berglund G, et al. Postmenopausal breast cancer risk in relation to sex steroid hormones, prolactin, and SHBG (Sweden). Cancer Causes Control. 2003;14(7):599-607. Maus MV, Reilly SC, Clevenger CV. Prolactin as a chemoattractant for human breast carcinoma. Endocrinology. Nov 1999;140(ll):5447-5450. McVean GA, Myers SR, Hunt S, Deloukas P, Bentley DR, Donnelly P. The fine- scale structure of recombination rate variation in the human genome. Science. Apr 23 2004;304(5670):581-584. Mertani HC, Garcia-Caballero T, Lambert A, et al. Cellular expression of growth hormone and prolactin receptors in human breast disorders. Int J Cancer. 1998;79(2) :202-211. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Meyer F, Brisson J, Morrison AS, Brown JB. Endogenous sex hormones, prolactin, and mammographic features of breast tissue in premenopausal women. JNatl Cancer Inst. 1986;77(3):617-620. Million Women Study. Breast cancer and hormone-replacement therapy in the Million Women Study. Lancet. Aug 9 2003 ;362(9382) :419-427. Moorman PG, Kuwabara H, Millikan RC, Newman B. Menopausal hormones and breast cancer in a biracial population. Am J Public Health. Jun 2000;90(6):966-971. Muhlbock O, Boot LM. Induction of mammary cancer in mice without the mammary tumor agent by isografts of hypophyses. Cancer Res. 1959;19:402-412. Newcomb PA, Titus-Emstoff L, Egan KM, et al. Postmenopausal estrogen and progestin use in relation to breast cancer risk. Cancer Epidemiol Biomarkers Prev. Jul 2002;ll(7):593-600. Newcomer LM, Newcomb PA, Potter JD, et al. Postmenopausal hormone therapy and risk of breast cancer by histologic type (United States). Cancer Causes Control. Apr 2003;14(3):225-233. Nickerson DA, Tobe VO, Taylor SL. PolyPhred: automating the detection and genotyping of single nucleotide substitutions using fluorescence-based resequencing. Nucleic Acids Res. 1997;25(14):2745-2751. Olsson H, Aim P, Kristoffersson U, Landin-Olsson M. Hypophyseal tumor and gynecomastia preceding bilateral breast cancer development in a man. Cancer. 1984;53(9): 1974-1977. Olsson HL, Ingvar C, Bladstrom A. Hormone replacement therapy containing progestins and given continuously increases breast carcinoma risk in Sweden. Cancer. Mar 15 2003;97(6):1387-1392. Ormandy CJ, Binart N, Kelly PA. Mammary gland development in prolactin receptor knockout mice. J Mammary Gland Biol Neoplasia. 1997;2(4):355-364. Ormandy CJ, Hall RE, Manning DL, et al. Coexpression and cross-regulation of the prolactin receptor and sex steroid hormone receptors in breast cancer. J Clin Endocrinol Metab. Nov 1997;82(ll):3692-3699. Persson I, Thurfjell E, Bergstrom R, Holmberg L. Hormone replacement therapy and the risk of breast cancer. Nested case-control study in a cohort of Swedish women attending mammography screening. Int J Cancer. Sep 4 1997;72(5):758- 761. 147 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Persson I, Weiderpass E, Bergkvist L, Bergstrom R, Schairer C. Risks of breast and endometrial cancer after estrogen and estrogen-progestin replacement. Cancer Causes Control. Aug 1999;10(4):253-260. Persson I, Yuen J, Bergkvist L, Schairer C. Cancer incidence and mortality in women receiving estrogen and estrogen-progestin replacement therapy— long-term follow-up of a Swedish cohort. Int J Cancer. Jul 29 1996;67(3):327-332. Peyrat JP, Dewailly D, Djiane J, et al. Total prolactin binding sites in human breast cancer biopsies. Breast Cancer Res Treat. 1981;l(4):369-373. Peyrat JP, Djiane J, Bonneterre J, et al. Stimulation ofDNA synthesis by prolactin in human breast tumor explants. Relation to prolactin receptors. Anticancer Res. 1984;4(4-5):257-261. Pike MC, Kolonel LN, Henderson BE, et al. 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. 2002;11(9):795-800. Pike MC, Peters RK, Cozen W, et al. Estrogen-progestin replacement therapy and endometrial cancer. JNatl Cancer Inst. Aug 6 1997;89(15):1110-1116. Pike MC, Ross RK, Spicer DV. Problems involved in including women with simple hysterectomy in epidemiologic studies measuring the effects of hormone replacement therapy on breast cancer risk. Am J Epidemiol. April 15 1998;147(8):718-721. Popovic V, Damjanovic S, Micic D, et al. Increased incidence of neoplasia in patients with pituitary adenomas. The Pituitary Study Group. Clin Endocrinol (Oxf). 1998;49(Oct (4)):441-445. Porch JV, Lee IM, Cook NR, Rexrode KM, Burin JE. Estrogen-progestin replacement therapy and breast cancer risk: the Women's Health Study (United States). Cancer Causes Control. Nov 2002;13(9):847-854. Reynolds C, Montone KT, Powell CM, Tomaszewski JE, Clevenger CV. Expression of prolactin and its receptor in human breast carcinoma. Endocrinology. 1997;138(12):5555-5560. Rockhill B, Colditz GA, Rosner B. Bias in breast cancer analyses due to error in age at menopause. Am J Epidemiol. Feb 15 2000; 151 (4):404-408. Rose DP, Pruitt BT. Plasma prolactin levels in patients with breast cancer. Cancer. 1981 ;48(12):2687-2691. 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ross RK, Paganini-Hill A, Wan PC, Pike MC. Effect of hormone replacement therapy on breast cancer risk: estrogen versus estrogen plus progestin. J Natl Cancer Inst. Feb 16 2000;92(4):328-332. Rossouw JE, Anderson GL, Prentice RL, et al. Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results From the Women's Health Initiative randomized controlled trial. Jama. Jul 17 2002;288(3):321-333. Rothman KJ, Greenland S. Modem Epidemiology. Vol xiii. Philadelphia: Lippincott-Raven; 1998. Salih H, Brander W, Flax H, Hobbs JR. Prolactin dependence in human breast cancers. Lancet. 1972;2(7787): 1103-1105. Schairer C, Lubin J, Troisi R, Sturgeon S, Brinton L, Hoover R. Menopausal estrogen and estrogen-progestin replacement therapy and breast cancer risk. Jama. Jan 26 2000;283(4):485-491. Schroeder MD, Symowicz J, Schuler LA. PRL modulates cell cycle regulators in mammary tumor epithelial cells. Mol Endocrinol. Jan 2002;16(l):45-57. Secreto G, Recchione C, Cavalleri A, Miraglia M, Dati V. Circulating levels of testosterone, 17 beta-oestradiol, luteinising hormone and prolactin in postmenopausal breast cancer patients. Br J Cancer. 1983;47(2):269-275. SEER. Summary Staging Guide; April, 1977. Shafie SM, Grantham FH. Role of hormones in the growth and regression of human breast cancer cells (MCF-7) transplanted into athymic nude mice. J Natl Cancer Inst. 1981 ;67(l):51-56. Stahlberg C, Pedersen AT, Lynge E, et al. Increased risk of breast cancer following different regimens of hormone replacement therapy frequently used in Europe. Int J Cancer. May 1 2004;109(5):721-727. Stanczyk FZ. Pharmacokinetics and potency of progestins used for hormone replacement therapy and contraception. Rev Endocr Metab Disord. Sep 2002;3(3):211-224. Stram DO, Haiman CA, Hirschhom JN, et al. Choosing Haplotype-tagging SNPs based on unphased genotype data using a preliminary sample of unrelated subjects with an example from the Multiethnic Cohort Study. Human Hered. 2003;55(l):227-236. 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Strain DO, Pearce CL, Bretsky P, et al. Abstract modeling and E-M estimation of haplotype specific relative risks from genotype data for a case control-study of unrelated individuals. Hum Hered. 2003;55(4): 179-190. Struman I, Bentzien F, Lee H, et al. Opposing actions of intact and N-terminal fragments of the human prolactin/growth hormone family members on angiogenesis: an efficient mechanism for the regulation of angiogenesis. Proc Natl AcadSci USA. Feb 16 1999;96(4): 1246-1251. Strangs I, Gray RA, Rigby HB, Stratton G. Two case reports of breast carcinoma associated with prolactinoma. Pathology. 1997;29(3):320-323. Talwalker PK, Meites J. Mammary lobulo-alveolar growth induced by anterior pituitary hormones in adreno-ovariectomized-hypophysectomized rats. Proc Soc Exp Biol Med. 1961;107(880-3). Touraine P, Martini JF, Zafrani B, et al. Increased expression of prolactin receptor gene assessed by quantitative polymerase chain reaction in human breast tumors versus normal breast tissues. J Clin Endocrinol Metab. 1998;83(2):667-674. Trott JF, Hovey RC, Koduri S, Vonderhaar BK. Alternative splicing to exon 11 of human prolactin receptor gene results in multiple isoforms including a secreted prolactin-binding protein. J Mol Endocrinol. Feb 2003;30(1):31 -47. Truong AT, Duez C, Belayew A, et al. Isolation and characterization of the human prolactin gene. Embo J. 1984;3(2):429-437. Tworoger SS, Eliassen AH, Rosner B, Sluss P, Hankinson SE. Plasma prolactin concentrations and risk of postmenopausal breast cancer. Cancer Res. 2004;64(18):6814-6819. Udy GB, Towers RP, Snell RG, et al. Requirement of STAT5b for sexual dimorphism of body growth rates and liver gene expression. Proc Natl Acad Sci U S A. 1997;94(14):7239-7244. Ursin G, Tseng CC, Paganini-Hill A, et al. Does menopausal hormone replacement therapy interact with known factors to increase risk of breast cancer? J Clin Oncol. Feb 1 2002;20(3):699-706. Volm MD, Talamonti MS, Thangavelu M, Gradishar WK. Pituitary adenoma and bilateral male breast cancer: an unusual association. JSurg Oncol. 1997;64(1):74- 78. Wakao H, Gouilleux F, Groner B. Mammary gland factor (MGF) is a novel member of the cytokine regulated transcription factor gene family and confers the prolactin response. Embo J. 1994; 13(9):2182-2191. 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Wall JD, Pritchard JK. Assessing the performance of the haplotype block model of linkage disequilibrium. Am J Hum Genet. Sep 2003;73(3):502-515. Wang DY, De Stavola BL, Bulbrook RD, et al. Relationship ofblood prolactin levels and the risk of subsequent breast cancer. Int J Epidemiol. 1992;21(2):214- 221. Weiss LK, Burkman RT, Cushing-Haugen KL, et al. Hormone replacement therapy regimens and breast cancer risk(l). Obstet Gynecol. Dec 2002;100(6):1148-1158. Welsch CW, Gribler C. Prophylaxis of spontaneously developing mammary carcinoma in C3H-HeJ female mice by suppression of prolactin. Cancer Res. 1973;33(11):2939-2946. Welsch CW, Iturri GC, Brennan MJ. DNA synthesis of human, mouse, and rat mammary carcinomas in vitro: influence of insulin and prolactin. Cancer. 1976;38(3): 1272-1281. Welsch CW, Nagasawa H. Prolactin and murine mammary tumorigenesis: a review. Cancer Res. 1977;37(4):951-963. Wennbo H, Gebre-Medhin M, Gritli-Linde A, Ohlsson C, Isaksson OG, Tomell J. Activation of the prolactin receptor but not the growth hormone receptor is important for induction of mammary tumors in transgenic mice. J Clin Invest. 1997;100(11):2744-2751. Wilson GD, Woods KL, Walker RA, Howell A. Effect of prolactin on lactalbumin production by normal and malignant human breast tissue in organ culture. Cancer Res. 1980;40(2):486-489. Wilson J, Foster D, Kronenber H, Larse PR, editors. Williams textbook of endocrinology. 9th edition ed. Philadelphia, PA: W.B. Saunders Company; 1998. Yen S, Jaffe RBe. Reproductive endocrinology. 4th edition ed. Philadelphia, PA: Saunders; 1999. Zaykin DV, Westfall PH, Young SS, Kamoub MA, Wagner MJ, Ehm MG. Testing association of statistically inferred haplotypes with discrete and continuous traits in samples of unrelated individuals. Hum Hered. 2002;53(2):79-91. Zhang K, Deng M, Chen T, Waterman MS, Sun F. A dynamic programming algorithm for haplotype block partitioning. Proc Natl Acad Sci. 2002;99:7335- 7339. 151 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDICES Appendix 1. Sixty-two SNPs used in the haplotype analysis of PRL. SNP # SNP ID Location Contig Position3 1 rs1205945 5’ 22434674 2 rs2744098 5' 22433728 3 rs2655420 5' 22431291 4 rs2655419 5' 22430714 5 rs2655418 5' 22430235 6 hCV328404 5' 22429983 7 rs2655417 5’ 22429107 8 rs2655416 5’ 22429033 9 hCV334512 5’ 22428479 10 rs2655415 5’ 22428301 11 rs2744099 5' 22428144 12 rs2655414 5' 22427227 13 rs717156 5' 22425601 14 rs715226 5' 22425542 15 rs717157 5' 22425494 16 rs2744104 5' 22425388 17 rs2744105 5' 22425298 18 rsl 569544 5' 22424921 19 rsl 883372 5' 22424437 20 rs2655448 5’ 22423583 21 rs6456484 5’ 22423005 22 rsl 807914 5' 22421224 23 rs2473122 5' 22420266 24 rs767938 5' 22419843 25 rs7752211 5' 22418624 26 rsl 935023 5' 22416724 27 rsl 209598 5’ 22416122 28 rsl341238 5' 22414679 29 rsl341239 5’ 22412183 30 rs2744117 Promoter 22410818 31 rs849875 Promoter 22407979 32 rs849876 Promoter 22407617 33 rs3756824 promoter 22406716 34 rs849877 Promoter 22406678 35 rs2244502 1 1 22402966 36 seq4633 12 22401058 37 seql139230 1 12 22400881 38 seq4915 E3 22400776 39 seq5122 13 22400569 40 rs7739889 13 22400532 41 rs849884 13 22400433 42 rs7759000 1 3 22399938 43 rs849886 13 22399346 44 rsl 205960 14 22396139 45 rs6239 E5 22395724 46 rsl205961 3' 22393991 47 rsl 205962 3’ 22393116 48 rs2744118 3' 22392518 49 rs849870 3' 22391864 50 rs849871 3' 22391794 51 rs6920781 3' 22391251 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Appendix 1 continued. 52 rs849872 3' 22391011 53 rs849873 3' 22390910 54 rs2744119 3' 22390710 55 rs2744120 3’ 22389237 56 rs2000132 3' 22386375 57 rs2153646 3’ 22385128 58 rs2655447 3’ 22384432 59 rs6940783 3’ 22381923 60 rs2066266 3' 22378099 61 rs2066265 3' 22377996 62 rsl417721 3’ 22376808 Location = gene region location, 5’ = upstream o f first coding exon, E = exon, I = intron, 3’ downstream of last coding exon, rs = NCBI database SNP, hCV = Celera database SNP, seq = MEC sequencing SNP. “ Contig Position based on the July 2003 freeze of the University of California at Santa Cruz Genome Browser. 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Appendix 2. One hundred and five SNPs used in the haplotype analysis of PRLR. SNP# SNP ID Location Contig Position3 1 hCV1438923 5' 35294510 2 hCV1438925 5’ 35291181 3 hCV 143 8926 5’ 35290964 4 hCV1438927 5’ 35286764 5 rs4703398 5' 35280380 6 hCVl 1281097 5’ 35276967 7 hCVl 1281103 UN 1 35274254 8 rsl 604427 I1N1 35269111 9 rs931740 I1N1 35267541 10 rsl 587608 I1N1 35264490 11 rsl 604422 I1N1 35262877 12 rsl604421 I1N1 35262801 13 rs924599 I1N1 35257726 14 hCVl 1281125 I1N1 35257275 15 rs47Q3511 I1N1 35256130 16 hCVl 1281126 I1N1 35255823 17 rs4703396 I1N1 35255155 18 rs4703510 I1N1 35255006 19 rs4235652 I1N1 35254840 20 rs2047740 I1N1 35248893 21 rs2047738 I1N1 35248592 22 rs875701 I1N1 35246157 23 rs875700 I1N1 35245921 24 rs4703509 I1N1 35245695 25 rs873456 I1N1 35245158 26 rsl587607 I1N1 35243963 27 hCVl 1281150 I1N1 35243825 28 rsl 5 87606 I1N1 35241931 29 rs4703508 I1N3 35234303 30 hCVl 1281167 I1N3 35233945 31 rsl610218 I1N3 35232195 32 rs4425481 I1N3 35231398 33 hCVl 155980 I1N3 35231327 34 hCVl 1281183 11N3 35228522 35 hCV 1155977 I1N3 35226590 36 rsl 039429 I1N3 35221479 37 rs4703506 I1N3 35217708 38 hCV1439028 I1N3 35213892 39 rsl 039427 I1N3 35211436 40 rsl039428 11N3 35211098 41 hCV1439032 I1N3 35208263 42 rs2047741 I1N3 35198556 43 rs954286 I1N4 35197742 44 hCVl 1281211 I1N4 35195406 45 rsl 5 87605 I1N4 35192366 46 hCV1439048 I1N4 35190763 47 rs2914111 11N4 35185426 48 rs2018170 I1N4 35182667 49 rs3797212 E1N5 35180977 50 rs2962092 I1N5 35179932 51 rs2962086 I1N5 35176249 52 rs930068 I1N5 35172879 53 hCV2935871 I1N5 35171872 54 hCVl 1278278 I1N5 35168324 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Appendix 2 continued. 55 hCVl 1278267 I1N5 35167211 56 rs4703501 I1N5 35165081 57 rsl 966571 12 35163254 58 hCV2935880 12 35161958 59 hCV2935881 12 35160112 60 hCV2935883 12 35156228 61 rs4703499 12 35149868 62 rs4703390 12 35149398 63 hCV2935892 12 35146679 64 hCV2935899 12 35144621 65 rs249534 12 35144172 66 rs249535 12 35143183 67 rs388541 12 35140325 68 rs37382 12 35138569 69 rs37383 12 35138470 70 seq646 13 35135078 71 rs249536 13 35134523 72 hCV2935908 13 35134334 73 rs37385 13 35133302 74 rs37386 13 35133225 75 rs43215 13 35132470 76 hCVl 1804735 13 35132359 77 rs37387 14 35131101 78 rs37388 14 35131063 79 rs37389 14 35130681 80 seq5331 14 35130399 81 seq5582 E5 35130148 82 seq5781 15 35129949 83 seql7599 16 35118121 84 rs37364 16 35117881 85 rs3776605 17 35115382 86 rs37367 19 35112661 87 rsl057828 110 35110016 88 rsl057829 110 35109961 89 rsl 12461 110 35108793 90 rs401694 110 35108119 91 rs3 87032 110 35107232 92 rsl 549623 110 35106343 93 rs392279 HO 35106232 94 rs249521 110 35106014 95 rsl 73627 110 35106002 96 rs249522 110 35105820 97 rs371913 no 35104675 98 hCV2935923 110 35100140 99 hCV2935925 110 35098275 100 rsl587059 110 35096009 101 rs40529 3' 35094073 102 rsl 87490 3' 35090628 103 rs37372 3' 35086821 104 rs344151 3' 35085551 105 rs3737Q 3’ 35084987 Location = gene region location, 5’ = upstream of first coding exon, E = exon, I = intron, 3’ downstream of last coding exon, E1N = alternative exon, I1N = alternative intron, rs = NCBI database SNP, hCV = Celera database SNP, seq = MEC sequencing SNP. "Contig Position based on the July 2003 freeze of the University of California at Santa Cruz Genome Browser. 155 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. Appendix 3. Studies evaluating estrogen-progestin use and breast cancer risk. Studies included in each analysis are represented by an ‘X’ and studies excluded are marked by footnotes. Study Analysis 1 Analysis 2 Analysis 3 Analysis 4 Overall Lobular Ductal Sequential Continuous- Combined Past Current R a n d o m iz e d Trials W HI,1 2 2003 X t Prospective Studies Stahlberg,1 6 2004§ X X X MWS,1 4 2003 X X X X* X Porch,9 2002 X t t Schairer,4 2000 X X X t C a s e -C o n tr o l S tu d ie s Kirsh,5 2002 X Newcomb,6 2002 X X X t t t t Weiss,7 2002/ X X X X X X X Daling,1 7 2002* Ross,3 2000/ X X X X X Ursin, 1 8 2002* Magnusson,2 1999 X X X X x« P o o le d S tu d ie s CGHFBC,1 1997 X Abbreviations: Cl, confidence interval; EPT, estrogen-progestin therapy; ORs, odd ratios; WHI, Women’s Health Initiative; MWS, Million Women Study; CGHFBC Collaborative Group on Hormonal Risk Factors in Breast Cancer. * Re suits by histologic subtype, not overall breast cancer risk (overall risk for for Baling in Weiss and for Ursin in Ross). *No results given for duration of EPT use. ^Only results for one type of progestin schedule. ^Overall risk calculated from sequential and continuous-combined use. “ Risk for ET or EPT use. O S OS Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Appendix 4. Studies evaluating estrogen-progestin use and breast cancer risk. Summary of overall characteristics and main findings. Randomized trials Cases (N) Population (N) Mean Follow-Up (Years) Adjusted Variables Results 1 C O o o C 4 ! N Healthy postmenopausal women in the Women’s Health Initiative Trial 199 treatment 150 placebo 8,506 treatment 8,102 placebo 5.6 age, dietary modification randomization group 1.24 (1.01,1.54) cumulative risk§ Prospective studies Study Population* Cases (N) Person-Years Mean Follow-Up (yrs) Adjusted Variables Results 2 Stahlberg,1 6 2004 Healthy postmenopausal women age 45+ in the Danish Nurse Cohort 139t 130f 7,572'1 6,8891 1 6.34 age, hx of BBD, age at meno (<55 vs. =55) sequential: <5 yr: 1.58 (0.79,3.17) 5-9 yr: 2.47(1.23,4.95) 10+yr: 2.18 (1.09,4.33)§ continuous-combined: <5 yr: 1.96(0.72,5.36) 5-9 yr: 4.96 (2.16,11.39) 10+yr: 6.78 (3.41, 13.48)§ 3 MWS,1 4 2003 Healthy women * in the UK age 50-64 as part of the National Health Service Breast Screening Programme (NHSBSP) in the Million Women Study 4,785 4,075 532,353 1 478,3991 1 2.6 age, time since meno, parity, AFFTP, family hx of breast cancer, BMI, region, deprivation index <1 yr: 1.45 (1.19,1.78) 1-4 yr: 1.74(1.60,1.89) 5-9 yr: 2.17(2.03,2.33) 3 0 yr: 2.31 (2.08, 2.56) § sequential: <5 yr: 1.77(1.59, 1.97) 3 yr: 2.12(1.95, 2.30){ < 1 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 4 Porch,9 2002 5 Schairer,4 2000 Healthy female postmenopausal * professionals age 45+ in the Women's Health Study Healthy postmenopausal1 women in the Breast Cancer Detection Demonstration Project (BCDDP) 3,525 1,005 -75 (known age meno) 33 26 00 441,7511 1 144,6971 1 71,438 -15,727 (known age meno) 18,948 18,948 5.9 10.2 (over 3 phases of follow-up) age, age at meno, meno type, age at menarche, nulliparity, age at first preg, abortions/ miscarriages, AFFTP, OC, hx of BBD, use of mammo screening, family hx breast cancer, race, BMI, smoking, alcohol, exercise age, edu, BMI, age at meno, mammo screening continuous-combined: <5 yr: 1.57 (1.37,1.79) 25 yr: 2.40 (2.15, 2.67)§ past HT: <1 yr: 0.94(0.84, 1.05) 1-4 yr: 1.01 (0.92, 1.12) 5-9 yr: 1.14(1.00,1.30) >10 yr: 1.05 (0.84, 1.30) current EPT: see above results for overall analysis <5yr: 1.11 (0.81, 1.52) 5+ yr: 1.76(1.29,2.39) P tr e n d =0-0004 § 1.06 (1.00,1.15) per year of use among known age meno § lobular/ductal: 1.17 (1.02,1.41) per year of use among all ductal: 1.17 (1.02, 1.41) per year of use among all women ^ ® Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 48 Case-control studies Study Population* Cases (N) 4 Q 4 | 4,142 351 2,654 849 672 730 vo 6 Kirsh, 2002 7 Newcomb,6 2002 Healthy postmenopausal1 women in Ontario, Canada Healthy postmenopausal * women from Massachusetts, New Hampshire, Wisconsin Weiss,7 2002 Healthy black and white postmenopausal women in Atlanta, Detroit, Los Angeles, Philadelphia, Seattle in the Contraceptive and Reproductive Experiences (CARE) Study 18,948 sequential: (P <15 days/mo) <1 yr: 1.1 (0.8, 1.7) >4 yr: 1.5 (1.0,2.4) among all women§ ________ Controls (N) Adjusted Variables Results 403 4,418 4.132 4.132 835 661 696 age, age at meno, type of meno, hx of BBD age at meno, type of meno, AFFTP, BMI, family hx breast cancer, edu, mammo screening hx, recent alcohol, hx of BBD, age at menarche, recent physical activity age, race, study center, type of meno, age at meno 1.15 (1.01,1.33) per year of use 1.04(1.01,1.08) per year of use lobular: 1.04 (0.97, 1.11) per year of use ductal: 1.04 (1.00,1.08) per year of use >0-<6 mo: 0.59 (0.40,0.87) 6 mo-<2 yr: 0.82(0.57, 1.18) 2-<5 yr: 1.33 (0.91,1.95) 5+ yr: 1.49 (1.05,2.12) sequential: (P <25 days/mo) >0-<6 mo: 0.70(0.38,1.30) 6 mo-<2 yr: 0.72(0.42,1.24) 2-<5 yr: 1.44 (0.79,2.61) 5+ yr: 1.18(0.70,1.98) continuous-combined: (P 315 days/mo) >0-<6 mo: 0.58 (0.36,0.94) 6 mo-<2 yr: 1.11 (0.71, 1.75) 2-<5 yr: 1.38 (0.86, 2.22) 5+ yr: 1.77(1.04,3.01) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. 859 1,174 Baling,1 7 2002 108 635 9 Ross,3 2000 Healthy postmenopausal 1,298T women in Los Angeles 1,193f 978r 899 1,042 835 835 1,108 1,012 age, race, study site, type of meno, known age at meno type of meno, age at meno, age at menarche, family hx of breast cancer, history BBD, nulliparity, AFFTP, OCs, BMI, alcohol past EPT: >0-<6 mo: 0.68 (0.47, 0.99) 6 mo-<2 yr: 0.76(0.52,1.13) 2-<5 yr: 1.24(0.79, 1.93) 5+ yr: 0.54 (0.33, 0.88) current EPT: >0-<6 mo: 0.53 (0.26,1.09) 6 mo-<2 yr: 1.11 (0.75,1.65) 2-<5 yr: 1.28 (0.95, 1.73) 5+ yr: 1.37(1.06, 1.77) lobular: <6 mo: 0.7 (0.3,1.5) 6mo-5 yr: 1.3 (0.7, 2.3) 5yr+: 1.9 (1.0,3.7) ductal: & mo: 0.6 (0.4, 0.9) 6mo-5yr: 1.0 (0.7, 1.3) 5yr+: 1.3 (0.9, 1.9) 1.24 (1.07,1.45) per 5 years of use P2-sided=0-005 sequential: (P < entire cycle) 1.38 (1.13, 1.68) P2-sided=0.0015 continuous-combined: (P = entire cycle) 1.09 (0.88,1.35) P2-sided= 0 -4 4 ______________________ Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Ursin,1 8 2002 10 Magnusson,2 1997 1,307 Healthy postmenopausal 2,137 ^ women in Sweden 1,841 f 1,874 f l,828t 0\ * 1,637 1,637 2,481 2,272 2,322 2,302 type and age at meno, age at menarche, family hx breast cancer, hx of BBD, nulliparity, AFFTP, OCs, weight, alcohol age, parity, AFFTP, age at meno, type of meno, BMI, height lobular: 1.34 (0.98,1.83) per 5 years of use Ptrend~0.06 ductal: 1.27 (1.08,1.50) per 5 years of use Ptrend=0-004 1-24 mo: 1.25 (0.96,1.63) 25-60 mo: 1.40(1.01, 1.94) 60-120 mo: 2.43 (1.72,3.44) >120 mo: 2.95(1.84,4.72) sequential: (P <16 days/mo) 1-24 mo: 1.58 (1.01,2.46) 25-60 mo: 1.34(0.71,2.54) 60-120 mo: 1.89(0.88,4.09) >120 mo: 2.45 (0.82, 7.30) continuous-combined: (P 2 9 days/mo) 1-24 mo: 0.93 (0.63,1.36) 25-60 mo: 1.26(0.76,2.09) 60-120 mo: 2.89(1.66,5.00) >120 mo: 5.36 (1.47, 19.56) past HT: (1-10 years ago): 1-60 mo: 1.09(0.76,1.55) >60 mo: 1.22(0.72, 2.08) (>10 years ago): 1-60 mo: 1.24(0.90, 1.70) >60 mo: 2.57(1.28,5.15) 2,475 current HT (<1 year ago): 1-60 mo: 1.52(1.21, 1.92) >60 mo: 2.68 (2.09, 3.42) Pooled Analysis Study Population* Cases (N) Controls (N) Adjusted Variables Results 11 CGHFBC,1 1997 Women from 22 published and 2 unpublished studies with information on type of preparation (EPT, ET, PT) 12,611 23,866 study, age at dx or pseudo dx, time since meno, BMI, parity, AFFTP <5 yr: 1.15 (SE = 0.19) =5yr: 1.53 (SE = 0.33)§ Abbreviations: WHI, Women’s Health Initiative; MWS, Million Women Study; CGHFBC, Collaborative Group on Hormonal Risk Factors in Breast Cancer. OC = oral contraceptives, hx = history, BBD = benign breast disease, BMI = body mass index, preg = pregnancy, mammo = mammography, edu = education level, meno = menopause, AFFTP = age at first full-term pregnancy, dx = diagnosis, mo = month(s), yr = year(s), P = progestin, E = estrogen, SE = standard error. Number of cases and number of starting population/person-years/controls for the results presented in final column. .' Results included (or did not specifically exclude) in situ breast cancer cases. * Results included (or did not specifically exclude) women with unknown age at menopause due to simple hysterectomy. * Risk was based on current and/or recent use rather than total use. "Risk among lean women. 1 Authors provided population rather than person-years. 2,475 current HT (<1 year ago): 1-60 mo: 1.52(1.21, 1.92) >60 mo: 2.68 (2.09, 3.42) Pooled Analysis Study Population* Cases (N) Controls (N) Adjusted Variables Results 11 CGHFBC,1 1997 Women from 22 published and 2 unpublished studies with information on type of preparation (EPT, ET, PT) 12,611 23,866 study, age at dx or pseudo dx, time since meno, BMI, parity, AFFTP <5 yr: 1.15 (SE = 0.19) =5yr: 1.53 (SE = 0.33)§ Abbreviations: WHI, Women’s Health Initiative; MWS, Million Women Study; CGHFBC, Collaborative Group on Hormonal Risk Factors in Breast Cancer. OC = oral contraceptives, hx = history, BBD = benign breast disease, BMI = body mass index, preg = pregnancy, mammo = mammography, edu = education level, meno = menopause, AFFTP = age at first full-term pregnancy, dx = diagnosis, mo = month(s), yr = year(s), P = progestin, E = estrogen, SE = standard error. Number of cases and number of starting population/person-years/controls for the results presented in final column. .' Results included (or did not specifically exclude) in situ breast cancer cases. * Results included (or did not specifically exclude) women with unknown age at menopause due to simple hysterectomy. * Risk was based on current and/or recent use rather than total use. "Risk among lean women. 1 Authors provided population rather than person-years.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Genetic risk factors in breast cancer susceptibility: The multiethnic cohort

PDF

Androgens and breast cancer

PDF

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

PDF

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

PDF

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

PDF

Genes and hormonal factors involved in the development or recurrence of breast cancer

PDF

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

PDF

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

PDF

Antioxidants and risk of myocardial infarction and cancer in a cohort of middle-aged finnish men: the kuopio ischemic heart disease risk factor study

PDF

Body size and the risk of prostate cancer in the multiethnic cohort

PDF

The role of genetic variation in insulin -like growth factors in prostate cancer risk: The multiethnic cohort

PDF

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

PDF

Genetic risk factors in multiple myeloma

PDF

beta3-adrenergic receptor gene Trp64Arg polymorphism and obesity-related characteristics among African American women with breast cancer: An analysis of USC HEAL Study

PDF

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

PDF

Effect of genetic factors in the development of childhood lymphocytic leukemia (ALL)

PDF

Hormonal and genetic risk factors of endometrial cancer and trends in incidence and survival of adult acute lymphoblastic leukemia

PDF

Determinants of the age at natural menopause: The multiethnic cohort

PDF

Examining the relationship between common genetic variation, type 2 diabetes and prostate cancer risk in the multiethnic cohort

PDF

Monoamine oxidase A: (1) As one of many candidate genes for preeclampsia and (2) as a candidate gene for tobacco addiction

Asset Metadata

Creator Lee, Sulggi (author)

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

School Graduate School

Degree Doctor of Philosophy

Degree Program Epidemiology

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

Tag biology, genetics,health sciences, medicine and surgery,health sciences, oncology,health sciences, public health,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Henderson, Brian (committee chair), Coetzee, Gerhard (committee member), Haiman, Christopher (committee member), McKean-Cowdin, Roberta (committee member), Pike, Malcolm (committee member), Ursin, Giske (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-332627

Unique identifier UC11340361

Identifier 3180308.pdf (filename),usctheses-c16-332627 (legacy record id)

Legacy Identifier 3180308.pdf

Dmrecord 332627

Document Type Dissertation

Rights Lee, Sulggi

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