University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content DIETARY FATS, FAT METABOLIZING GENES, AND THE RISK OF BREAST CANCER by Jun Wang A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (MOLECULAR EPIDEMIOLOGY) August 2004 Copyright 2004 Jun Wang Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3145309 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 3145309 Copyright 2004 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. DIEDICATION To my mother, Manwei Pang — You are my model and give me the strength of life. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGEMENTS This work was done under the direction of Dr. Sue Ingles. I am grateful to Dr. Sue Ingles for providing me with inspiration and the professional guidance that enabled me to gain valuable experience in the field of molecular epidemiology. I appreciate her kind help in the writing of this dissertation. I wish to thank Dr. Esther John for offering me the opportunity to work on her Northern California Breast Cancer Study dataset and biospecimens, and for her generous support and pertinent advice through the study. I wish to thank Dr. Louis Dubeau, Dr. Jean Shih and Dr. Mimi Yu for their valuable suggestions on this study. Thanks to my friends and lab-mates, Wei Wang, Hui-lee Wong, Melissa Wilson, and Katherine Tsai, for their help in my study. Thanks to Avi Manchandia for his help in the functional work. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS Page DEDICATION ii ACKNOWLEDGEMENTS iii LIST OF TABLES vi LIST OF FIGURES x ABSTRACT xi INTRODUCTION 1 CHAPTER 1 Dietary Fat, N-6 Polyunsaturated Fat Lipoxygenase 3 Pathway and the Risk of Breast Cancer (review) 1.1 Risk Factors for Breast Cancer 3 1.2 Dietary Fats and Breast Cancer 3 1.2.1 Composition of Dietary Fats 3 1.2.2 Dietary Fats and the Risk of Breast Cancer: Ecological Studies, 6 Migrant Studies and Secular Trends 1.2.3 Analytical Epidemiologic Studies of the Association between 9 Dietary Fats and the Risk of Breast Cancer 1.2.4 Limitations in Present Epidemiologic Studies 27 1.2.5 Effect of Types and Amounts of Fat on Breast Cancer 30 Development: Experimental Evidence 1.3 N-6 Polyunsaturated Fatty Acid Lipoxygenase Pathway and the 36 Development of Breast Cancer 1.3.1 N-6 fatty acid lipoxygenase metabolic pathways 38 1.3.2 Roles of Lipoxygenases in Breast Cancer 40 1.3.3 Interactions among Lipoxygenase Pathways 43 iv Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 2 Hypotheses and Study Aims 45 CHAPTER 3 Study Design 47 3.1 Study Population 47 3.2 Data Collection 49 3.3 Food Frequency Questionnaire Development and Dietary Assessment 49 3.4 Biospecimen Collection 51 3.5 Exposure and Confounding Variables 52 3.6 Laboratory Methods 5 3 3.7 Data Management and Statistical Analysis 59 CHAPTER 4 Results and Discussion 66 4.1 Summary of the LOX Gene Polymorphisms 66 4.2 Functional Studies of the 5-LOX and FLAP Gene 5’ Regulatory 70 Region Polymorphisms 4.3 Dietary Fat Intake and Breast Cancer Risk 75 4.4 The 5-LOX Gene Polymorphisms and Breast Cancer Risk 116 4.5 The 12-LOX Gene Polymorphisms and Breast Cancer Risk 133 4.6 Other LOX gene Polymorphisms and Breast Cancer Risk 146 CHAPTER 5 Summary 149 REFERENCES 160 v Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES Page Table 1.1 Summary of the literature on dietary fats and breast cancer risk: 11 case-control studies Table 1.2 Summary of the literature on dietary fats and breast cancer risk: 18 meta-analyses of case-control studies Table 1.3 Summary of the literature on dietary fats and breast cancer risk: 20 cohort studies Table 1.4 Summary of literature on dietary fats and breast cancer risk: meta- 24 analyses of cohort studies Table 3.1 Probes and primers for the LOX gene TaqMan genotyping 57 Table 3.2 Methods for energy adjustment at dietary data analysis 61 Table 4.1 Summary of the LOX gene polymorphisms 67 Table 4.2.1 Means and standard deviation of relative luciferase activity for the 71 5-LOX gene 5’ Spl binding site polymorphism in SK-BR3 cells Table 4.2.2 Adjusted relative luciferase activity of the 5-LOX gene 5’ Spl 72 binding site polymorphism in SK-BR3 cells Table 4.2.3 Means and standard deviation of relative luciferase activity for the 74 FLAP gene 5’ poly(A) microsatellite in SK-BR3 cells Table 4.2.4 Adjusted relative luciferase activity for FLAP gene 5’ poly(A) 74 microsatellite in SK-BR3 cells Table 4.3.1 Basic characteristics of study subjects, in all and by menopausal 76 status Table 4.3.2 Basic characteristics of study subjects, by ethnicity 79 Table 4.3.3 Basic characteristics of study population among US bom and 82 foreign bom Latinas Table 4.3.4 Pathological characteristics of breast cancer, by ethnicity 84 vi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.3.5 Comparison of the daily intake of total energy and percentage of 86 energy from dietary components Table 4.3.6 Correlations between dietary fat, as a percentage of total energy 87 intake Table 4.3.7 Dietary fat intake and breast cancer risk according to quartiles of 88 fat residual, among all cases and controls Table 4.3.8 Dietary fat intake and breast cancer risk according to quartiles of 91 fat residual, by ethnicity Table 4.3.9 Dietary fat intake and breast cancer risk according to quartiles of 92 fat intake as a percentage of total energy intake, among all cases and controls Table 4.3.10 Dietary fat intake and breast cancer risk according to quartiles of 93 fat intake as a percentage of total energy intake, by ethnicity Table 4.3.11 Dietary fat intake and breast cancer risk according to quartiles of 94 fat intake as a percentage of total energy intake, among US bom and foreign bom Latinas Table 4.3.12 Dietary fat intake and breast cancer risk according to quartiles of 95 fat intake as a percentage of total energy intake, with fatty acids mutually adjusted Table 4.3.13 Dietary fat intake and breast cancer risk according to according 97 to 5% increase of total energy intake from fatty acids, by family history of breast cancer and history of benign breast disease Table 4.3.14 Dietary fat intake and breast cancer risk according to according 99 to 5% increase of total energy intake from fatty acids, by menopausal status Table 4.3.15 Dietary fat intake and breast cancer risk according to according 99 to 5% increase of total energy intake from fatty acids, by physical activity Table 4.3.16 Dietary fat intake and breast cancer risk according to according 100 to 5% increase of total energy intake from fatty acids, by ER and PR status Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.3.17 Dietary fat intake and breast cancer risk according to according 103 to 5% increase of total energy intake from fatty acids, by stage and histological grade of disease Table 4.3.18 The most commonly used cooking fat among controls of a 104 subset of study subjects (n=1913) Table 4.3.19 Cooking fat and breast cancer risk among three ethnic groups 104 Table 4.4.1 Allele frequencies of the 5-LOX gene polymorphisms among 117 three ethnic groups Table 4.4.2 The 5-LOX gene -1286G/T and Spl binding site haplotype 118 frequencies among three ethnic groups Table 4.4.3 The 5LOX gene Spl binding site genotype distribution and crude 120 odds ratio (95% Cl) among three ethnic groups Table 4.4.4 The 5-LOX gene Spl binding site genotype and breast cancer risk 121 Table 4.4.5 The 5-LOX gene Spl binding site genotype and breast cancer risk 121 among women without family history of breast cancer Table 4.4.6 The 5-LOX gene Spl binding site polymorphism and breast 122 cancer risk, by history of benign breast disease (BBD) Table 4.4.7 The 5-LOX gene Spl binding site polymorphism and breast 122 cancer risk, by menopausal status Table 4.4.8 The 5-LOX gene Spl binding site polymorphism and breast 122 cancer risk, by progression of cancer Table 4.4.9 The 5-LOX gene Spl binding site polymorphism and breast 124 cancer risk, by linoleic acid (LA) intake Table 4.4.10 The 5-LOX gene -1286G>T polymorphism and breast cancer 125 risk Table 4.4.11 The 5-LOX gene -1286G>T polymorphism and breast cancer 126 risk among African-Americans, stratified by levels of other cofactors Table 4.4.12 The 5-LOX gene 5’ regulatory region haplotype and breast 127 cancer risk among African-Americans Vlll Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.4.13 The 5-LOX gene 760G>A polymorphism and breast cancer risk among African-Americans Table 4.4.14 The 5-LOX gene 760G>A polymorphism and breast cancer risk among African-Americans, by menopausal status Table 4.5.1 Allele and haplotype frequencies of the 12-LOX gene polymorphisms among three ethnic groups Table 4.5.2 The 12-LOX gene polymorphisms and breast cancer risk among three ethnic groups Table 4.5.3 The 12-LOX gene polymorphisms and breast cancer risk among women without family history of breast cancer Table 4.5.4 The 12-LOX gene polymorphisms and breast cancer risk, by history of benign breast disease (BBD) Table 4.5.5 The 12-LOX gene polymorphisms and breast cancer risk, by menopausal status Table 4.5.6 The 12-LOX gene polymorphisms and breast cancer risk, by tumor stages Table 4.5.7 The 12-LOX gene polymorphisms and breast cancer risk, by estrogen receptor (ER) status of tumor Table 4.5.8 The 12-LOX gene polymorphisms and breast cancer risk, by linoleic acid (LA) intake Table 4.5.9 The 12-LOX gene haplotypes and breast cancer risk among three ethnic groups Table 4.6.1 The 15-LOX-2 gene 1967G>A polymorphism and breast cancer risk Table 4.6.2 The 15-LOX-2 gene 1967G>A polymorphism and breast cancer risk among women without family history of breast cancer Table 4.6.3 The FLAP gene 5’ poly(A) microsatellite and breast cancer risk Table 4.6.4 The FLAP gene 5’ poly(A) microsatellite and breast cancer risk among women without family history of breast cancer Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF FIGURES Page Figure 1.1 Formula and structures of some important dietary fatty acids. 4 Figure 1.2 An outline of essential fatty acid metabolism 33 Figure 1.3 Metabolic pathways for biosynthesis of eicosanoids from n-6 fatty 38 acids Figure 3.1 Schematic diagram of the transfection and luciferase assay of 5- 58 LOX gene Spl binding site polymorphism Figure 4.1 The 5-LOX gene promoter region Spl binding region 69 polymorphism Figure 4.2.1 Relative luciferase activity of 5-LOX gene 5’ Spl binding site 72 polymorphism in SK-BR3 cells Figure 4.2.2 Relative luciferase activity of the FLAP gene 5 ’ poly(A) 74 microsatellite in SK-BR3 cell line Figure 4.4.1 The 5-LOX gene 5’ regulatory region and coding region 117 polymorphisms Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT Animal studies have implicated dietary fats and fatty acid metabolites in mammary carcinogenesis. However, epidemiologic studies have failed to demonstrate a convincing link, probably due to methodological issues related to dietary assessment. An alternative approach to determine the existence of a relationship between fat intake and breast cancer is to examine whether the genetic polymorphisms in fat metabolic pathways are associated with breast cancer risk. In a population-based multiethnic case-control study conducted in the San Francisco Bay area, we analyzed the association between dietary fat intake, n-6 polyunsaturated fat lipoxygenase gene polymorphisms, and breast cancer risk. Analysis of dietary fat intake among 1677 cases and 2010 controls found that high-fat intake was associated with increased risk of breast cancer (highest vs. lowest quartile, adjusted OR= 1.40, 95% Cl: 1.14-1.73). Among types of fat, oleic acid was most strongly and consistently associated with increased risk (highest vs. lowest quartile, adjusted OR= 1.49, 95% Cl: 1.21-1.84). Linoleic acid (polyunsaturated fat) was weakly associated with increased risk (highest vs. lowest quartile, adjusted OR= 1.17, 95% Cl: 0.96-1.43). Polyunsaturated fat was more strongly associated with breast cancer risk when analyzing cooking fat usage among 832 cases and 1081 controls. A higher risk of breast cancer was found for women using vegetable/com oil (rich for n- 6 polyunsaturated fat) (adjusted OR=1.34, 95% Cl: 1.07-1.69), compared to women using olive or canola oil (rich for monounsaturated fat). Analysis of lipoxyxgnease gene polymorphisms among 805 cases and 889 controls found that the 5-lipoxygenase xi Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. gene Spl binding site non-wild-type alleles were associated with increased risk of breast cancer among Latinas and African-Americans. The platelet-type 12- lipoxygenase gene polymorphisms, Arg261Gln and Asn322Ser, were also found to be associated with breast cancer risk among African-Americans. While the measurement error was still a concern in the analysis of fat intake in this study, our finding of an association between lipoxygenase gene polymorphisms and breast cancer risk implicates that dietary n-6 fats, metabolized through lipoxygenase pathways, may have a significant role in breast cancer. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. INTRODUCTION It has been hypothesized that a high-fat diet promotes the development of breast cancer and is one of the major factors that accounts for the large variation in breast cancer incidence around the world. Although this hypothesis is supported by animal experiments, epidemiologic studies in humans show weak or inconsistent associations, even though a large number of investigations have been devoted to studying this relationship. This dissertation includes a molecular epidemiologic study investigating the association between dietary fat and the risk of breast cancer in a multiethnic population. Genetic variations in the genes of n-6 polyunsaturated fatty acid lipoxygenase pathway are examined in relation to the risk of breast cancer. By examining genetic and dietary influences together, we try to identify dietary risk factors that may be obscured when diet is examined in isolation. Chapter 1 provides a brief overview of the current status of studies of dietary fat and risk of breast cancer. The evidence from ecological studies, analytical epidemiologic studies, and animal & in vitro experimental studies is summarized. Types of dietary fat were reviewed separately regarding their association with breast cancer. Limitations in present epidemiologic studies are also noted. The significance of the n-6 polyunsaturated fat lipoxygenase pathway in the development of breast cancer is reviewed based on the evidence of in vitro experimental studies and animal studies. Chapter 2 proposes our hypotheses on dietary fat, fat metabolizing genes and breast cancer risk. A molecular epidemiologic study was designed to examine these 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. hypotheses. The proposed study was funded by California Breast Cancer Research Program (CBCRP), starting in July 2001. Chapter 3 provides detailed information on study design, including selection of study subjects, data and biospecimen collection, and methods for laboratory and statistical analyses. Chapter 4 presents and interprets the results of this study. Lipoxygenase gene polymorphisms identified in this study are summarized. Two 5’ regulatory region polymorphisms in the 5-lipoxygenase gene and the 5-lipoxygenase activating protein gene are examined for transcriptional activities. Associations between dietary fat, lipoxygenase gene polymorphisms and breast cancer risk are analyzed. Results are discussed and compared with the previous literature and with our hypotheses. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 1 Dietary Fat, N-6 Polyunsaturated Fat Lipoxygenase Pathway and Risk of Breast Cancer (review) 1.1 Risk Factors for Breast Cancer The life-time risk of developing breast cancer among Western women is approximately 10%, and despite advances in therapeutic strategies, breast cancer remains the leading cause of cancer deaths in women in most developed countries. There are several well-established risk factors for breast cancer and a variety of others currently under study. Family history of breast cancer, greater hormone exposure (early menarche, late menopause, postmenopausal obesity, hormone replacement therapy, etc), history of benign breast disease, ionizing radiation, and alcohol consumption have been associated with increased risk of breast cancer, while early first-term pregnancy, lactation and active lifestyle are suggested to be protective (Henderson et al., 1996). A westem-style diet has been suggested to be partially responsible for the high incidence of breast cancer in developed countries. Among dietary factors, fat has been one of the most studied regarding its association with increased risk of breast cancer, though the epidemiologic evidence is inconsistent. 1.2 Dietary Fats and Breast Cancer 1.2.1 Composition of Dietary Fats All natural fats contain complex mixtures of three types of fatty acids: saturated, monounsaturated and polyunsaturated fatty acid, classified based on the 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. number of double bond (figure 1.1). Saturated fats have not double bond. Monounsaturated fats have only one double bone, usually at carbon-9 from methyl end, therefore are called n-9 fats. Polyunsaturated fats have more than one double bond. According to the location of the first double bond from methyl end, polyunsaturated fats are further classified into n-6 or n-3 family, which have distinct properties and biological functions. Compared to saturated and monounsaturated fats, polyunsaturated fats occur in less quantities. However, they are extremely important due to their very strong biological activity. H j C Saturated fatty acids ^ C O O H MomnMsa!«r&tc<i fatty add UarM CIfc#) Myrisfrc (0 4 :0 ) Si«*rte I Okie Polyunsaturated fatty acids Uaefck Lino teak COOH Figure 1.1 Formula and structure of some important dietary fatty acids. The number before the colon indicates the number of carbon atoms in the fatty acid chain and the number after the colon indicates the number of double bonds (Bartsch et al., 1999). Most human diets contain a variety of these fatty acids. The major saturated fatty acid in the diet is palmitic acid (C16:0), followed by stearic (C18:0), myristic (C14:0), lauric (C12:0) acids, and the short and medium chain saturated fatty acids (C4:0-10:0). Palmitic acid is found in all edible fats and oils and is particularly abundant in palm oil and in butter, milk, cheese and meats. Stearic acid is found predominantly in cocoa butter (used in chocolate) and in fats from cattle and sheep. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Myristic and lauric acids are found in coconut and palm kernel oils, which are used in confectionery, and dairy fats. Short and medium chain fatty acids occur mainly in dairy fat (Bartsch et al., 1999). The major monounsaturated fatty acid (MUFA) in human diets is oleic acid (cA-C18:l n-9). Oleic acid is present in all edible fats and oils, but olive and canola oils are particularly rich sources. The principal polyunsaturated fatty acids (PUFAs) include linoleic acid (cis, cis-C18:2 n-6) and a-linolenic acid (cis, cis, cis-C18:3 n-3). Both These n-6 and n-3 fatty acids both are essential fatty acids and as such must be provided in the diet. The primary sources of linoleic acid are the oils from seeds and grains, with sunflower, safflower and com oil being particularly rich, a-linolenic acid is found in leafy green vegetables and in flaxseed and canola oils. Linoleic acid and a- linolenic acid, after consumption, can be desaturated and elongated to 20-carbon fatty acids. Linoleic acid is converted to arachidonic acid (AA, C20:4 n-6), and a-linolenic acid to eicosapentanoic acid (EPA, C20:5 n-3) and then docosahexaenoic acid (DHA, 22:6 n-3). Meat is also a major source for AA, and fish is the source for EPA and DHA. Polyunsaturated fat contributes ~ 7% of total energy intake and ~ 20% of energy intake from fat in the diets of adults in the United States. Due to the abundance of linoleic acid, 10-fold more n-6 fats than n-3 fats are consumed, in contrast with the recommended 2~3-fold (Kris-Etherton et al., 2000). The fatty acid composition of foods can be modified in various ways, e.g., plant fats can be modified by breeding, and animal fats can be modified by feeding practices (Gurr and Harwood, 1991). It has been claimed that agricultural changes 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. have resulted in a steady increase in linoleic acid at the expense of a-linolenic acid in the human diet, and that this process has been accelerated as cattle have been fed increasingly on grains rather than grass (Kris-Etherton et al., 2000). Industrial processing also has great influence on food fatty acid composition. Industrial hydrogenation reduces the degree of unsaturation, isomerizes the cis geometric configuration to the trans form, and shifts the double bonds along the hydrocarbon chain from their original positions in the natural fat (Gurr and Harwood, 1991). Trans fat is commonly found in vegetable shortenings, some margarines, crackers, cookies, snack foods, and other foods made with or fried in partially hydrogenated oils. Consumption of fat has changed among countries during the last several decades. In the United States and western Europe, total consumption of fat as a percent of energy has actually declined, while the average linoleic acid intake has risen from -3% of energy in the 1950s to 6-7% at present, with a commensurate decrease in saturated fat (McDowell, 1994; Sanders, 2000). On the other hand, dietary fat consumption has increased in many countries with historically low intake, such as Japan, since the 1950’s (Sugano and Hirahara, 2000). 1.2.2 Dietary Fats and the Risk of Breast Cancer: Ecological Studies, Migrant Studies and Secular Trends Rates of breast cancer vary more than fivefold around the world. Although lifestyle, genetic, reproductive and many other factors have been linked to the risk for breast cancer, much of the variation in breast cancer incidence across countries and cultures remains unexplained. That high-fat diets increase the risk of breast cancer has 6 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. long been the principal dietary hypothesis in relation to breast cancer. This hypothesis was generated originally from a statistically strong correlation seen in international ecological studies, based on food disappearance data (Armstrong and Doll, 1975; Carroll and Khor, 1975). Since national fat consumption per capita correlates highly with the level of economic development and thus affluent lifestyle, one disadvantage of these ecological comparisons is that the results are potentially confounded by many known and suspected breast cancer risk factors that are more prevalent in developed countries. Although it has been reported that after adjustment for gross national product per capita and average age of menarche, the ecological relation between fat disappearance and breast cancer incidence remains statistically significant (Prentice and Sheppard, 1990), confounding from other breast cancer risk factors that were not controlled in the association between breast cancer risk and dietary fat intake, such as low parity, late age at first birth, obesity, hormone replacement therapy, alcohol consumption, low physical activity etc., still could not be excluded. The relationship between diet and breast cancer is also supported by the results of migrant studies. These migrant studies found that populations migrating from low- incidence to high-incidence countries acquire breast cancer rates similar to those in their new adopted countries within two or three generations (Shimizu et al., 1991). The increase in breast cancer risk is influenced by the age of migration and the degree of acculturation (Willett, 1999). These results suggest that the large differences in breast cancer rates among countries are attributable to factors other than genetic factors. 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The ecological association between diet and breast cancer has also been observed within some countries, in which the changes in breast cancer rates over time have been seen to be positively correlated with changes in fat disappearance (Prentice and Sheppard, 1990). A dramatic increase in breast cancer incidence occurred in Iceland over the period in which the diet of the Iceland population changed substantially, becoming higher in fat composition and more like other western countries (Bjamason et al., 1974). This increase was primarily in women 45 years of age and older, suggesting differential effects in pre- and post-menopausal women. A birth cohort in Norway who were exposed to famine as adolescents during World War II has experienced an overall 13% lower risk of breast cancer (Tretli and Gaard, 1996). Ecological studies have also suggested differential influences of individual fatty acids on breast cancer. For example, in Mediterranean countries such as Italy and Spain, where olive oil (rich in oleic acid) is a staple, the incidence of breast cancer is lower than that in North America and northern Europe (Assmann et al., 1997). Greek women, who consume -40% of their energy supply as fat, mainly from olive oil, have significantly lower rates of breast cancer than women in the United States, whose average energy intake from fat is lower (-35%) (Trichopoulou et al., 2000). Long-chain n-3 polyunsaturated fat has also been suggested to be beneficial. Eskimos, who eat fish and meat from marine mammals rich in n-3 fatty acids, and Japanese fishermen, who have the highest consumption of fish per capita in the world, have low rates of breast cancer, despite their overall high fat consumption (Bartsch et al., 1999). The ratio of n-6/n-3 polyunsaturated fats in the diet has been suggested to be 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. important. While Japan is a country with a relatively low breast cancer incidence rate, the incidence rate is increasing with the changing of dietary patterns: the proportion of n-6 fatty acids has been increasing, and long-chain n-3 fatty acids decreasing, hence decreasing the n-3/n-6 ratio (Lands et al., 1990; Shoda et al., 1996). 1.2.3 Analytical Epidemiologic Studies of the Association between Dietary Fats and the Risk of Breast Cancer While the results from ecological studies and international comparisons are vulnerable to confounding by many uncontrolled factors, there have also been many case-control and cohort studies, in which other potential risk factors can be measured and controlled in the analysis, that have also examined the relationship between dietary fat and breast cancer risk. Results from individual case-control studies are equivocal (table 1.1). While a positive association between high-fat intake and risk of breast cancer has been reported in many case-control studies (Cade et al., 1998; De Stefani et al., 1998; Ewertz and Gill, 1990; Katsouyanni et al., 1988; Qi et al., 1994; Richardson et al., 1991; Ronco et al., 1996; Sieri et al., 2002; Van't Veer et al., 1990; Wakai et al., 2000; Yu et al., 1990), there are also many other case-control studies that did not find a significant association (Goodman et al., 1992; Goodstine et al., 2003; Graham et al., 1991; Hirohata et al., 1987; Holmberg et al., 1994; Ingram et al., 1991; Lee et al., 1991; Levi et al., 1993; Mannisto et al., 1999; Martin-Moreno et al., 1994; Potischman et al., 1998; Pryor et al., 1989; Rohan et al., 1988; Yuan et al., 1995; Zaridze et al., 1991). However, meta-analyses of case-control studies generally support a positive 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. association between total fat and breast cancer risk (Boyd et al., 1993; Boyd et al., 2003; Harrison and Waterbor, 1999; Howe et al., 1990) (table 1.2). In contrast, prospective cohort studies have provided less support for the dietary fat-breast cancer hypothesis. Except one study that found breast cancer risk associated with greater total fat intake among postmenopausal women without benign breast disease (Velie et al., 2000) and another study that found that intake of animal fat during premenopausal years to be associated with an increased risk of breast cancer (Cho et al., 2003), most individual cohort studies did not show a significant positive association with total fat intake (Byrne et al., 2002; Gaard et al., 1995; Holmes et al., 1999; Howe et al., 1991; Knekt et al., 1990; Kushi et al., 1992; Toniolo et al., 1994; van den Brandt et al., 1993; Wolk et al., 1998) (table 1.3). Most meta-analyses of cohort studies also don’t support the association between dietary fat and the risk of breast cancer (Boyd et al., 1993; Boyd et al., 2003; Harrison and Waterbor, 1999; Hunter et al., 1996; Smith-Wamer et al., 2001), except for the most recent meta analysis analysis, which reported high fat intake is marginally associated with an increased risk of breast cancer (Boyd et al., 2003) (table 1.4). 10 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. Tablel.l Summary of the literature on dietary fats and breast cancer risk; case-control studies Reference Study location Cases/ Controls Dietary assessment method Energy adjustment OR (95% Cl) (Hirohata et al., 1987) Hawaii (Japanese and Caucasian) 183/183 161/161 DHQ' (43) No highest versus lowest quartile: total fat, 1.5 (0.8-2.9) (Jan) 1.3 (0.6-2.6)(Cau) saturated fat, 2.2 (1.1-4.4) (Jan) 2.0 (0.9-4.0) (Cau) (Rohan et al., 1988) Adelaide, Australia 451/451 FFQ" (179) Yes highest versus lowest tertile (premenopausal): total fat, 1.0 (0.5-1.9); 1.1 (0.6-1.9) (postmenopausal) saturated fat, 1.0 (0.5-1.9); monounsaturated fat, 0.9 (0.5-1.8); polyunsaturated fat, 0.8 (0.4-1.5); linoleic acid 0.7 (0.4-1.2); (Katsouyanni et al., 1988) Athens, Greece 120/120 FFQ (120) No 90th versus 10th centile: total fat, 1.7 (1.0-2.1); saturated fat, 0.9 (0.5-1.4); monounsaturated fat, 1.9 (1.1-3.2); polyunsaturated fat, 1.2 (0.8-2.1); (Pryor et al., 1989) Utah 172/190 (white) FFQ No highest versus lowest quartile (adolescent fat intake): total fat, 0.7 (0.2-2.1) (premenopausal) 0.7 (0.2-1.7) (postmenopausal) (Toniolo et al., 1989) Italy 250/499 DHQ Yes highest versus lowest quartile Saturated fat, 3.0 (1.9-4.7); Monounsaturated fat 1.5 (ns); Polyunsaturated fat 1.0 (ns). (Ewertz and Gill, 1990) Denmark 1486/ 1336 FFQ (21) No highest versus lowest quartile: total fat, 1.5 (1.2-1.8) ;* *trend (P value < 0.05) *DHQ: Dietary history questionnaire; **FFQ: Food frequency questionnaire Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.1 (continued) Reference Study location Cases/ Controls Dietary assessment method Energy adjustment OR (95% Cl) (Van't Veer et al., 1990) Netherlands 133/289 FFQ Yes highest versus lowest quintile: total fat, 3.5 (1.6-7.6)* *trend (P value < 0.05) significant differences in mean intake between cases and controls for total fat, saturated fat, and monounsaturated fat (Yu et al., 1990) Shanghai, China 186/372 FFQ Yes highest versus lowest quintile: total fat, 1.7(1.0-2.1); saturated fat, 0.9 (0.5-1.4); monounsaturated fat, 1.9 (1.1-3.2); polyunsaturated fat, 1.2 (0.8-2.1) (Ferraroni et al., 1991) Milano, Italy 214/215 DHQ Yes highest versus lowest quartile: animal fat, 1.1 (0.5-2.4); vegetable fat, 0.8 (0.3-1.7); saturated fat, 1.1 (0.4-2.8); monounsaturated fat, 1.1 (0.5-2.8) polyunsaturated fat, 1.3 (0.6-2.8) (Ingram et al., 1991) Western Australia 99/209 FFQ (179) No highest versus lowest consumption (based on median): total fat, 1.4 (0.8-2.5); saturated fat, 1.0 (0.6-1.8); monounsaturated fat, 1.6 (0.9-2.9): polyunsaturated fat, 0.9 (0.4-1.7) (Lee et al., 1991) Singapore 200/420 FFQ No highest versus lowest tertile: total fat, 0.8 (0.4-1.4); saturated fat, 0.9 (0.5-1.7); monounsaturated fat, 1.0 (0.5-1.8); polyunsaturated fat, 0.4 (0.2-0.7);* *trend (P value < 0.05) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.1 (continued) Reference Study location Cases/ Controls Dietary assessment method Energy adjustment OR (95% Cl) (Graham et al., 1991) Western New York state 439/439 FFQ No highest versus lowest quartile: postmenopausal, total fat, 0.9 (0.6-1.4); saturated fat, 1.0 (0.7-1.5) (Richardson et al., 1991) Montpellier, France 409/515 FFQ No highest versus lowest tertile: total fat, 1.6(1.1-2.2); * animal fat, 1.6 (1.1-2.2);* saturated fat, 1.9 (1.3-2.6); * monounsaturated fat, 1.7 (1.2-2.5);* polyunsaturated fat, 1.2 (0.9-1.7) *trend (P value < 0.05) (Zaridze et al., 1991) Moscow, Russia 139/139 FFQ Yes highest versus lowest quartile total fat, 0.5 (0.04-7.0); saturated fat, 1.7 (0.2-11.8); monounsaturated fat, 1.8 (0.2-16.7); polyunsaturated fat, 0.1 (0.03-0.7)* *trend (P value < 0.05) (Kato et al., 1992) Japan 908/908 FFQ (8) No daily consumption vs 1-2/wk: meats, 0.9 (0.7-1.2); oily foods, 1.2 (0.9-1.5) (Goodman et al., 1992) Oahu, Hawaii (Japanese and Caucasian) Jan (138/154) Cau (134/142) DHQ No highest versus lowest quartile: total fat, 1.3 (0.8-2.1). (Jan, 1.4, ns; Cau, 1.2, ns); saturated fat, 1.5 (1.0-2.5); (Jan, 1.4, ns; Cau, 1.4, ns) animal protein, 1.6 (1.0-2.6) (Levi et al., 1993) Vaul, Switzerland 107/318 FFQ (50) Yes highest versus lowest tertile: total fat 1.5 (ns) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.1 (continued) Reference Study location Cases/ Controls Dietary assessment method Energy adjustment OR (95% Cl) (Qi et al., 1994) Tianjin, China 244/244 DHQ Yes highest versus lowest quartile: total fat, 3.3 (1.3-8.2); saturated fat, 2.5 (1.3-5.1); monounsaturated fat, 3.1 (1.5-6.7); polyunsaturated fat, 2.4 (0.6-9.5) (Holmberg et al., 1994) Sweden 265/432 FFQ (60) Yes High vs low (continuous): Total fat 1.0 (ns); Saturated fat 1.2 (ns); Polyunsaturated fat 0.7 (ns) (Martin- Moreno et al., 1994) Spain 762/988 FFQ (118) Yes highest versus lowest quartile: total fat, 1.0 (0.7-1.3); saturated fat, 1.0 (0.6-1.5); monounsaturated fat, 0.9 (0.6-1.3); polyunsaturated fat, 1.3 (1.0-1.8); oleic acid, 0.8 (0.5-1.1); linoleic acid, 1.2 (0.9-1.7) (Yuan et al., 1995) China 300/300 (Tianjin) 534/534 (Shang hai) FFQ (68) Yes highest versus lowest quintile: total fat, 1.2 (0.7-2.0); saturated fat, 1.3 (0.6-2.6); monounsaturated fat, 1.2 (0.7-2.2); polyunsaturated fat, 1.3 (0.4-4.3) (Trichopoul ou et al., 1995) Athens, Greece 820/1548 FFQ (115) Yes more than once/day versus once/day: olive oil, 0.8 (0.6-1.0) OR for 4 times/month increment: butter, 1.0 (1.0-1.1); margarine 1.1 (1.0-1.1); seed oils, 1.0 (1.0-1.1) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.1 (continued) Reference Study location Cases/ Controls Dietary assessment method Energy adjustment OR (95% Cl) (Franceschi et al., 1996) Sourthem Italy 2569/ 2588 FFQ (78) Yes 100 kcal/day intake; Saturated fat, 1.2 (1.0-1.3) Monounsaturated fat, 1.0 (0.9-1.1) Polyunsaturated fat, 0.7 (0.6-0.8) (Ronco et al., 1996) Montevideo, Uruguay 169/253 FFQ Yes highest versus lowest quartile: total fat, 1.8 (1.0-3.2); saturated fat, 2.4 (1.3-4.3)* *trend (P value < 0.05) (Witte et al., 1997) LA County 140/222 FFQ Yes highest versus lowest quartile: total fat, 0.4 (0.2-0.8); * saturated fat, 0.5 (0.2-1.1); monounsaturated fat 0.5 (0.2-1.0);* polyunsaturated fat, 0.3 (0.1-0.7); * oleic acid 0.4 (0.2-0.9); linoleic acid, 0.3 (0.1-0.7); * red meat, 0.6 (0.3-1.3) *trend (P value < 0.05) (Potischman et al., 1997) Atlanta, Seattle/Puget Sound, central New Jersey 1588/ 1451 Block food frequency No highest versus lowest quartile: total fat, 1.1 (0.8-1.4); saturated fat, 1.1 (0.9-1.4) (La Vecchia et al., 1998) Italy 2569/ 2588 FFQ Yes highest versus lowest quintile: olive oil, 0.9 (0.7-1.1) OR for 10 g/day increase: saturated fat, 1.1 (1.0-1.2); monounsaturated fat, 1.0 (0.9-1.0) OR for 5 g/day increase; polyunsaturated fat, 0.9 (0.9-1.0) OR for highest vs. lowest quartile: saturated fat, 1.5(1.1-2.0); monounsaturated fat, 1.0 (0.8-1.2); polyunsaturated fat, 0.7 (0.6-0.9) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.1 (continued) Reference Study location Cases/ Controls Dietary assessment method Energy adjustment OR (95% Cl) (De Stefani et al., 1998) Montevideo, Uruguay 390/397 FFQ Yes highest versus lowest quartile: total fat, 1.5 (0.9-2.6);* saturated fat, 2.5 (1.4-4.4);* monounsaturated fat, 2.5 (1.5-4.1); * polyunsaturated fat, 1.0 (0.6-1.6); linoleic acid 0.7 (0.4-1.2); alpha-linoleic acid, 3.2 (1.9-5.6)* *trend (P value < 0.05) (Potischman et al., 1998) Atlanta, Seattle/Puget Sound, central New Jersey 1647/ 1501 Block food frequency No highest versus lowest quartile (diet during adolescence): high-fat snacks and desserts, 1.1 (0.9-1.3); animal fat, 1.0 (0.8-1.2); high-fat foods, 0.9 (0.7-1.1) (Cade et al., 1998) Southampton & Portsmouth, England 220/825 FFQ (141) No highest versus lowest quartile: saturated fat, 2.4 (1.1-5.0); monounsaturated fat 0.9 (0.4-1.8); polyunsaturated fat, 0.6 (0.3-1.3) (Mannisto et al., 1999) Finland 310/454 FFQ No highest versus lowest quintile (pre-menop): total fat, 0.7 (0.3-1.6); saturated fat, 1.0 (0.4-2.2); monounsaturated fat, 0.8 (0.3-1.9) n-3 fat, 0.7 (0.3-1.7) n-6fat, 0.7 (0.3-1.6); trans fat, 1.3 (0.5-3.1) (Wakai et al., 2000) Jakarta, Indonesia 226/452 FFQ (98) Yes highest versus lowest quartile: total fat, 8.5 (4.0-17.8) (pre-marriage); 3.5 (2.0-6.2) (post-marriage); Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.1 (continued) Reference Study location Cases/ Controls Dietary assessment method Energy adjustment OR (95% Cl) (Sieri et al., 2002) Northern Italy 56/214 FFQ (107) Yes highest versus lowest tertile (post-menop): total fat, 3.5 (1.4-8.4); animal fat, 1.8 (0.6-5.4); vegetable fat, 0.9 (0.3-2.7); saturated fat, 1.1 (0.3-4.0); monounsaturated fat, 3.0 (0.7-12.6) polyunsaturated fat, 2.0 (0.7-6.0) linoleic 1.4 (0.5-3.8) linolenic 0.7 (0.2-2.6) (Goodstine et al., 2003) Connecticut 565/554 FFQ No highest versus lowest quartile: total fat, 1.1 (0.6-1.8); saturated fat, 1.0 (0.6-1.6); monounsaturated fat, 1.2 (0.7-2.0) polyunsaturated fat, 1.1 (0.7-1.6) n-3/n-6, 0.67 (0.3-1.3)* * trend (P value < 0.05) < 1 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.2 Summary of the literature on dietary fats and breast cancer risk: meta-analyses of case-control studies Reference Study Cases/ Controls Dietary assessment method OR (95% Cl) Comments (Howe et al., 1990) meta-analysis of 12 case- control studies 4427/ 6095 FFQ OR for highest vs. lowest quintile (all subjects): total fat, 1.4 (P < 0.0001); saturated fat, 1.6 (not reported); OR for highest vs. lowest quintile (premenop): total fat, 1.1 (P~ 0.21) OR for highest vs. lowest quintile (postmenop): total fat, 1.5 (P <0.0001) Original study data used in meta-analysis; univariate summary ORs computed by conditional logistic regression, with 240 strata defined by the 12 studies and 20 age groups; significant trend (P value<0.05)) with increasing quintiles for total fat and saturated fat (Boyd et al., 1993) meta-analysis of 16 case- control studies published between 1978— 1991 6831/ 7105 DHQ/FFQ OR for highest versus lowest level of intake: total fat, 1.2 (1.1-1.3); saturated fat, 1.4 (1.2-1.6); monounsaturated fat, 1.4 (1.2-1.7); polyunsaturated fat, 0.9 (0.8-1.1) Summary natural log relative risk estimates and variance from each study used in meta-analysis; univariate summary ORs computed by random effects model to account for variability across studies; summary ORs reflect comparison between lowest and highest level of intake described in each study (e.g. tertile, quartile, etc, different from study to study) (Harrison and Waterbor, 1999) meta-analysis o f 23 case- control studies (including studies in meta-analyses performed by Howe and Boyd) DHQ 24-h recall OR for highest versus level of intake (studies in Europe): total fat, 1.5 (not significant) OR for highest versus lowest level of intake (studies in North America): total fat, 1.3 (not significant) OR for highest versus lowest level o f intake (studies in other areas): total fat, 1.2 (not significant) Summary natural log relative risk estimates and variance from each study used in meta-analysis; univariate summary ORs computed by random effects conditional model to account for variability across studies; summary ORs reflect comparison between lowest and highest level o f intake described in each study (e.g. tertile, quartile, etc, different from study to study) 00 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.2 (continued) (Boyd et al., meta-analysis 16,280/ DHQ/FFQ 2003) of 31 case- 18,966 control studies published between 1978- 2003(including studies in meta-analyses performed by Boyd et al in 1993) VO OR for highest vs. lowest level o f intake: total fat, 1.14(0.99-1.32); 1.22 (0.91-1.63), for studies with adjustment for energy and other risk factors saturated fat, 1.23 (1.03-1.46); monounsaturated fat, 1.12 (0.94— 1.32); polyunsaturated fat, 0.5 (0.39-0.63) Summary natural log relative risk estimates and variance from each study used in meta-analysis; summary ORs were computed by random effects model to account for variability across studies; summary ORs reflect comparison between lowest and highest level of intake described in each study; Additional subgroup and regression analyses to examine the potential influence of study design and execution, study population, geographical location, adjustment variables, partitioning cut points and methods o f analysis. Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.3 Summary of the literature on dietary fats and breast cancer ris Study Reference Study location Case/ Cohort Dietary assessment method Energy adjustment RR (95% Cl) NHANES (Jones et al., 1987) US 99/5485 24-h recall No highest versus lowest quartile: total fat, 0.3 (0.2-0.7); saturated fat, 0.3 (0.1-0.7); polyunsaturated fat, 0.7 (0.4-1.4) monounsaturated fat, 0.6 (0.3-1.1) Seventh-day Adventist (Mills et al., 1989) California 215/ 20,341 24-h recall No highest versus lowest quartile: animal fat, 1.2 (0.8-1.8) Finnish social insurance institute’s mobile clinic health examination survey (Knekt et al., 1990) Finland 54/3988 DHQ Yes highest versus lowest quartile: total fat, 1.7 (0.6— 4.8); saturated fat, 1.4 (0.5- 3.7); monounsaturated fat, 2.7 (1.0-7.4); * polyunsaturated fat, 1.2 (0.6-2.8) *trend (P value<0.05) Canadian national breast screening service (Howe et al., 1991) Canada 519/ 56837 DHQ (86) Yes highest versus lowest quartile: total fat, 1.3 (0.9-1.9); saturated fat, 1.1 (0.7- 1.6); monounsaturated fat, 1.2 (0.8-1.9); * polyunsaturated fat 1.3 (0.9-1.8); *trend, (P value < 0.05) Iowa Women’s Health Study (Kushi et al., 1992) Iowa 459/ 34388 FFQ (121) Yes highest versus quartile: total fat, 1.2 (0.9-1.5); saturated fat, 1.1 (0.8- 1.4); monounsaturated fat, 1.1 (0.9-1.3); polyunsaturated fat, 1.2 (0.9-1.5) c cohort studies K > O Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.3 (continued) Study Reference Study location Case/ Cohort Dietary assessment method Energy adjustment RR (95% Cl) New York Cohort (Graham et al., 1992) New York State 359/ 18586 FFQ (45) No highest versus lowest quintile: total fat, 1.0 (0.7-1.4) animal fat, 1.1 (0.8-1.6); vegetable fat, 1.1 (0.8-1.5) Nurse Health Study (Willett et al., 1992) US 1439/ 89494 FFQ (121) Yes highest versus lowest quintile total fat, 0.9 (0.8-1.1); saturated fat, 0.9 (0.7-1.0); monounsaturated fat, 0.9 (0.8-1.1); linoleic acid, 0.9 (0.8-1.1) Municipal Registries (van den Brandt et al., 1993) Netherlands 437/1598 FFQ(150) Yes highest versus lowest quintile: total fat, 1.1 (0.7-1.6) saturated fat, 1.4 (0.9-2.1)* monounsaturated fat, 0.8 (0.5-1.1); polyunsaturated fat, 1.0 (0.6-1.4) *trend (P value<0.05) NYU Women’s Health Study (Toniolo et al., 1994) New York City 180/829 FFQ(71) Yes highest versus lowest quintile: total fat, 1.5 (0.9 2.5); saturated fat, 1.5 (0.9-2.5); oleic acid: 1.6 (0.9-2.7); linoleic acid, 1.1 (0.7-2.0); meat, 1.9 (1.1-3.2) * *trend (P value < 0.05) Norwegian National Health Screening (Gaard et al., 1995) Norway 248/ 25,892 FFQ Yes highest versus lowest quartile: total fat, 1.3 (0.9-1.8); saturated fat, 1.0 (0.8-1.6); monounsaturated fat, 1.7 (1.2-2.5) * polyunsaturated fat, no associated *trend (P value < 0.05) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.3 (continued) Study Reference Study location Case/ Cohort Dietary assessment method Energy adjustment RR (95% Cl) Sweden Mammography Cohort (Wolk et al., 1998) Central Sweden 674/ 61,471 FFQ Yes highest versus lowest quartile: total fat, 1.0 (0.8-1.3); saturated fat, 1.1 (0.8- 1.4); monounsaturated fat,0.8 (0.5-1.2); polyunsaturated fat, 1.2 (0.9-1.6) Nurses’ Health Study (Holmes et al., 1999) US 2,956/ 88,795 FFQ (61) Yes RR for given increment o f fat intake per day: total fat (5% of energy), 1.0 (0.9-1.0); animal fat (5% o f energy), 1.0 (1.0-1.0); vegetable fat (5% energy), 1.0 (0.9-1.0); saturated fat (5% of energy), 0.9 (0.9-1.0); monounsaturated fat (5% of energy), 0.9 (0.9-1.0); polyunsaturated fat (5% of energy), 0.9 (0.8-1.0); trans-unsaturated fat (1% of energy), 0.9 (0.9-1.0); omega-3 fat from fish (0.1% of energy), 1.1 (0.9-1.1) RR similar in pre- and post-menopausal women BCDDP (Breast Cancer Detection Demonstration Project) (Velie et al., 2000) US 9961 40022 FFQ (60) Yes highest versus lowest quintile (total) total fat, 1.1 (0.9-1.3); saturated fat 1.1 (0.9-1.5); oleic, 0.9 (0.6-1.3); linoleic, 1.1 (0.8-1.3); highest versus lowest quintile (postmenopausal, w/o BBD) total fat, 2.2 (1.4-3.4);* saturated fat 1.2 (0.8-2.0); oleic, 1.8 (0.9-3.7); * linoleic, 1.3 (0.8-2.1); *trend (P value < 0.05) to to Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.3 (continued) Study Authors (year) Study location Case/ Cohort Dietary assessment method Energy adjustment RR(95% Cl) Nurses’ Health Study (Byrne et al., 2002) US 1071/ 44697 FFQ (61) Yes highest versus lowest quintile (Postmenopausal, w/o BBD) total fat, 0.9 (0.8-1.2); saturated fat 0.9 (0.7-1.1); oleic, 1.1 (0.9-1.5); linoleic, 0.9 (0.7-1.2); trans-unsaturated fat, 0.9 (0.7-1.1); Netherlands cohort study (Voorrips et al., 2002) Netherlands 941/ 1598 FFQ (100) Yes highest versus lowest quintile (postmenopausal) total fat, 1.2 (0.9-1.6); saturated fat 1.4 (1.0-2.0); monounsaturated fat, 0.6 (0.4-1.0);* polyunsaturated fat, 0.9 (0.7-1.2) oleic, 0.7 (0.4— 1.0); * linoleic, 1.0 (0.7-1.3); linolenic 0.7 (0.5-1.0)* *trend (P value < 0.05) Nurses’ Health Study II (Cho et al., 2003) US 714/ 90655 FFQ (133/142) Yes highest versus lowest quintile (premenopausal intake) total fat, 1.3 (1.0-1.6); saturated fat 1.1 (0.7-1.5); monounsaturated fat, 1.1 (0.8-1.6); polyunsaturated fat, 1.0 (0.7-1.3); trans-unsaturated fat, 1.0 (0.7-1.3); long-chain n-3 polyunsaturated fat, 1.1 (0.8-1.4) ro U ) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.4 Summary of literature on dietary fats and breast cancer risk: meta-analyses of cohort studies Study Reference Case/ Cohort Dietary assessment method RR (95% Cl) Comments meta-analysis of 7 studies published between 1978- 1991 (Boyd et al., 1993) 3,007/ 252,765 DHQ/FFQ 24-h recall OR for highest versus lowest level of intake: total fat, 1.0 (0.9-1.1); saturated fat, 1.0 (0.8-1.1); monounsaturated fat, 1.0 (0.8-1.1); polyunsaturated fat, 1.0 (0.9-1.1); Summary natural log relative risk estimates and variance from each study used in meta analysis; univariate summary relative risks computed by random effects model to account for variability across studies; summary relative risks reflect comparison between lowest and highest level o f intake described in each study (e.g. tertile, quartile, etc, different from study to study) multivariate meta-analysis of 8 cohort studies (Hunter et al., 1996) 4,980/ 337,819 Food frequency RR for highest vs. lowest quintile: total fat, 1.1 (0.9-1.2); saturated fat, 1.1 (1.0-1.2) monounsaturated fat, 1.0 (0.9-1.2); polyunsaturated fat, 1.1 (1.0-1.2); animal fat, 1.0 (0.9-1.1) Original study data used in meta-analysis; regression adjusted for age at menarche, menopausal status, parity, age at first birth, body mass index, height, education, history o f benign breast disease, history of breast cancer in mother or sister, oral contraceptive use, fiber intake, alcohol intake, and energy intake meta-analysis of 9 cohort studies (including studies in meta analysis performed by Hunter) (Harrison and Waterbor, 1999) Not reported Food frequency RR for highest versus lowest level of intake in studies in Europe: total fat, 1.2 (not significant); RR for highest versus lowest level of intake in studies in North America: total fat, 1.0 (not significant); RR for highest versus lowest level of intake in studies in other areas: total fat, 1.0 (not significant) Summary natural log relative risk estimates and variance from each study used in meta analysis; univariate summary RRs computed by random-effects conditional model to account for variability across studies; summary RRs reflect comparison between lowest and highest level o f intake described in each study (e.g. tertile, quartile, etc, different from study to study) N > Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 1.4 (continued) meta-analysis of 8 cohort studies (including studies in meta analysis performed by Hunter) (Smith-Wamer et al., 2001) 7,329/ 351,821 Food frequency RR for highest vs. lowest quartile: saturated fat, 1.0 (0.9-1.2); monounsaturated fat, 1.0 (0.9-1.1); polyunsaturated fat, 1.0 (1.0-1.1); animal fat, 1.0 (0.8-1.1) vegetable fat 1.0 (0.9-1.1) Original study data used in meta-analysis; Individual studies were analyzed as nested case-control studies or case-cohort studies. The random effects model was used to combine log relative risks form the multiple studies. Individual study results were weighted by the inverse of their variance. meta-analysis of 14 cohort studies published between 1978- 2003 (including studies in meta analysis performed by Boyd et al in 1993) (Boyd et al., 2003) 8,735/ 568,549 DHQ/FFQ/ 24-h recall Relative risk for highest vs. lowest level of intake: total fat, 1.11 (0.99-1.25) 1.13 (1.04-1.23), for studies with adjustment for energy and other risk factors saturated fat, 1.15 (1.02-1.30); monounsaturated fat, 1.10(0.83-1.44) polyunsaturated fat, 1.11 (1.00-1.22) Summary natural log relative risk estimates and variance from each study used in meta analysis; summary relative risks were computed by random effects model to account for variability across studies; summary relative risks reflect comparison between lowest and highest level o f intake described in each study; Additional subgroup and regression analyses to examine the potential influence of study design and execution, study population, geographical location, adjustment variables, partitioning cut points and methods of analysis. to Some studies have attempted to examine the effects of individual fatty acids or of different classes of fatty acids; however, results are still inconclusive. For the most abundant n-6 fatty acid, linoleic acid, results of a meta-analysis did not support an increased risk of breast cancer with high intake of linoleic acid, although a small increase could not be excluded (Zock and Katan, 1998). Intake of n-3 fatty acids has rarely been calculated in case-control and cohort studies of breast cancer. Fish consumption, a major dietary source of long-chain n-3 fatty acids, has been found either to be associated with decreased risk or to have no relationship with breast cancer risk (Terry et al., 2003). The effects of n-3 and n-6 polyunsaturated fats may interact with each other. In a recent report from Singapore cohort, although overall no association was found between n-6 fatty acid intake and breast cancer risk, among subjects consuming low levels of marine n-3 fatty acids, high n-6 fatty acid intake was associated with increased risk of breast cancer (Gago-Dominguez et al., 2003). Consumption of olive oil (rich in n-9 monounsaturated fatty cid, oleic acid, but low in n-6 PUFAs) has been reported to be inversely associated with the risk of breast cancer in case-control studies in Greece, Spain and Italy (La Vecchia et al., 1995; Landa et al., 1994; Martin-Moreno et al., 1994; Trichopoulou et al., 1995). It has been suggested that the observed protective effects of a high intake of olive oil might be due to relatively low intake of n-6 fatty acids and a high intake of the antioxidants present in olive oil (Bartsch et al., 1999). The fatty acid composition of serum, erythrocyte membrane phospholipids, adipose tissue or breast fat tissue has been examined in several case-control and cohort 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. studies. Most case-control studies found no significant association for linoleic acid (Chajes et al., 1999; London et al., 1993; Petrek et al., 1994; Simonsen et al., 1998). Three prospective studies (Pala et al., 2001; Rissanen et al., 2003; Vatten et al., 1993) and one case-control study (Zaridze et al., 1990) reported an inverse association between n-6 polyunsaturated fatty acids and breast cancer risk. Two case-control studies showed a positive association between n-6 polyunsaturated fatty acid levels measured in breast adipose tissue and breast cancer risk (Bagga et al., 2002; Maillard et al., 2002). Most of the above studies also found no significant association for saturated and monounsaturated fatty acids, except for one study that reported a positive association (Pala et al., 2001) and one that reported an inverse association (Maillard et al., 2002) between oleic acid and breast cancer risk. The ratio of n-3/n-6 was reported in only a few studies and most of them found an inverse association of this ratio with breast cancer risk (Bagga et al., 2002; Maillard et al., 2002; Simonsen et al., 1998). 1.2.4 Limitations in Present Epidemiologic Studies There are limitations in present epidemiologic studies of dietary fat and breast cancer risk that may be responsible for the inconsistent findings. Some of these limitations have been widely discussed, including: 1) Measurement errors, including nondifferential misclassification due to inaccurate food frequency estimates, underreport of fat intake from overweight people, etc (Hunter, 1999; Willett, 1999; Zock and Katan, 1998); 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2) The range of fat intake within each individual study is usually narrow. Therefore the variation of exposure may be insufficient to find consistent association with cancer risk (Lee and Lin, 2000; Willett, 1999; Zock and Katan, 1998); 3) Most prospective studies of breast cancer had relatively modest follow-up ( < 1 0 years), so that a longer latency between fat intake and diagnosis of breast cancer cannot be excluded (Willett, 1999). 4) The risk for breast cancer may be imprinted early in life or even in utero by a diet high in fat or in specific fatty acids, resulting in early onset of puberty and increased risk for breast cancer in adulthood. Such risk modifiers, acting early in life or during pregnancy, have not been taken into consideration in most of the studies that accessed the dietary intake of middle-aged women (Hunter, 1999). 5) Estimates of dietary fat intake all reflect sources of error related to the use of food composition tables. Therefore biomarker studies that measure fatty acid composition in plasma or tissue may provide more precise estimates of dietary intake (Kohlmeier, 1997). Biomarker studies are insufficient. Additionally, some incorrect or insufficient analyses of dietary fatty acids among epidemiologic studies also need special attention: 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1) In many earlier studies, the analysis of dietary fat intake was poorly adjusted for total energy intake. The independent effect of fat was therefore poorly evaluated. 2) The relative levels of individual fatty acids that contributed to the total amount of fat consumed were not adequately measured and analyzed in most of the published data. This may have resulted in the inconsistency among the results, since different fatty acids may have different effects on tumor development. 3) Although in some studies, saturated, monounsaturated and polyunsaturated fatty acids have been analyzed individually, polyunsaturated fatty acids were still treated as one group instead of being divided into n-3 and n-6 fatty acids, which may have opposite effects. In addition, the intake of one type of fatty acid was rarely controlled for intake of other types of fat. The n-6/n-3 fatty acid ratio has been suggested to be more important than absolute levels of each fatty acid in experimental studies (Boudreau et al., 1991). However, so far the analysis of this ratio has been only conducted in a few epidemiological studies. 4) Long-chain and short-chain n-3 fatty acids may have different effects regarding the development of breast cancer. Separate analysis of these two types of n-3 fatty acids may also help to clarify the association between fat intake and breast cancer risk. 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In addition to the above limitations, the lack of consideration of potential genetic differences in fat metabolism is also worth noting. In a population with a relatively homogenous lifestyle, dietary sources and eating habits, genetic differences may play important roles in influencing disease risk. Experimental studies have demonstrated that n-6 polyunsaturated fatty acids affect tumor development through their metabolites that have more potent biological activity (See Chapter 1.3 for details). Therefore the genetic differences in n-6 fatty acid metabolic pathways may influence the production of those active metabolites and thus the risk of breast cancer. By considering genetic susceptibility along with dietary intake of specific fatty acids, we might be able to identify dietary risk factors that are obscured when diet is examined alone. 1.2.5 Effect of Types and Amounts of Fat on Breast Cancer Development: Experimental Evidence Within the last few decades, a large number of animal experiments have shown the effects of dietary fat in the development and subsequent progression of breast cancer. The amount and type of fat can markedly influence the growth of induced breast tumors in rodents during the promotion stage, but less so during the initiation stage (Welsch, 1992). Specific fatty acids may exert different or even opposing effects, thus the net result could depend on their relative amount in the diet. A meta analysis of data on mammary tumor incidence extracted from 97 reports of experiments involving over 12,800 mice and rats has been conducted to study the effects of saturated, monounsaturated, n-6 and n-3 polyunsaturated fatty acids (Fay et 30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. al., 1997). The results indicated that overall, n-6 polyunsaturated fats have a strong, and saturated fats a weak tumor-enhancing effect, whereas n-3 fatty acids possibly have a small protective effect that was statistically non-significant, and monounsaturated fats have no consistent effect. 1.2.5.1 Tumor-Enhancing Effects of N-6 Polyunsaturated Fatty Acids In general, the n-6 fatty acids have been associated with enhancement of the experimental mammary carcinogenesis, tumor cell invasion, and expression of the metastatic phenotype. In vitro, n- 6 fatty acids stimulate the proliferation of mouse mammary epithelial cells in the presence of epidermal growth factor (EGF) (Bandyopadhyay et al., 1987). Human non-transformed breast cancer cells also exhibit a growth response to n-6 fatty acids in the presence of EGF and insulin (Balakrishnan et al., 1989). Estrogen-independent breast cancer cell lines were found to exhibit a growth response when cultured in the presence of n-6 fatty acids (Buckman et al., 1991; Rose and Connolly, 1990), whereas estrogen-dependent cell lines did not (Bardon et al., 1996; Grammatikos et al., 1994). It has been suggested that arachidonic acid may exert a stimulatory effect on breast cancer cell growth involving the induction of cyclin D1 gene expression which leads to cell cycle progression (Razanamahefa et al., 2000). Many investigators have confirmed the enhancing effects of the high-n-6 fat diet on carcinogen-induced mammary tumor development in rodents. Feeding animals with a diet rich in n- 6 fatty acids was found to stimulate growth and enhance metastasis of human breast cancer cells (Meschter et al., 1992; Rose et al., 1993; 31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Solanas et al., 2002). One issue that is frequently raised about the role of dietary fat in experimental mammary carcinogenesis is whether any observed relations are due to fat per se or to associated differences in energy intake, given that fat is the most energy- dense nutrient. While some suggested that the promotional effects of a high-fat diet on chemically induced rat mammary carcinogenesis are ascribable primarily to high- energy intake (Kritchevsky and Klurfeld, 1987; Pariza, 1987), others proposed that there are dual and distinct effects of fat or specific fatty acids and energy intake (Cohen et al., 1988; Freedman et al., 1990). One convincing argument for a direct influence of dietary fat is that the observed effects vary with the specific fatty acids. When a high-fat diet was provided in the form of com oil or sunflower oil, which is rich in n-6 fatty acids, there was a pronounced enhancement of mammary tumor development. In contrast, equivalent total fat intake in the form of oils containing saturated fatty acid produced no significant increase in tumor formation (Carroll and Hopkins, 1979). In a study that fed isocaloric diets with different types and levels of fatty acids to female nude mice with MDA-MB-231 cells injected into the thoracic mammary fat pads, the tumor growth rates were significantly lower in mice fed a 4% linoleic acid (LA) diet compared with an 8% LA diet, and the tumor growth rate was further reduced by addition of 4% DHA to the 4% LA-containing diet (Connolly et al., 1999). 1.2.5.2 Tumor-Inhibiting Effects of N-3 Polyunsaturated Fatty Acids The major n-3 polyunsaturated fatty acids include a-linolenic acid, eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA). These n-3 fatty acids 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. not only share desaturases and elongases with n-6 fatty acids for metabolism (figure 1.2 ), but also compete with n-6 fatty acids for incorporation into cell membrane phospholipids and further activation. n-6 fatty acids n-3 fatty acids Linoleic Acid (LA) 18:2n-6 18: 3n-3 a-Linolenic Acid (ALA) A6-desaturase ------► , r y-Linoleic Acid (GLA) 18:3n-6 18:4n-3 < ----- Elongase --------► Dimommo-y- r ' Linoleic Acid (DGLA) 20:3n-6 20:4n-3 1 r •#— A5-desaturase Arachidonic Acid (AA) 20:4n-6 20:5n-3 Eicosapentaenoic Acid (EPA) ^ *----- Elongase - -------- ► Adrenic Acid 22:4n-6 I 22:5n-3 1 1 22: 5n-6 22:6n-3 Docosahexaenoic Acid (DHA) Figure 1.2 An outline of essential fatty acid metabolism Since Karmali et al first reported a difference in the effects of n-3 and n-6 fatty acids on the growth of transplantable mammary tumors, in which the tumor-promoting activity of n-6 fatty acids was abrogated by competitive inhibition by n-3 fatty acids (Karmali et al., 1984), a substantial amount of evidence has shown an inhibitory effect of fish oil or the constituent n-3 fatty acids on mammary tumorigenesis. The protective effect has been observed in the carcinogen-induced tumor model (Cohen et al., 1993), transplantable tumor model (Rose et al., 1995), and spontaneous tumor model (Kamano et al., 1989). For example, feeding long-chain n-3 fatty acids to nude mice with human breast cancer cell xenografts suppressed expression of the metastatic 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. phenotype (Rose et al., 1995; Rose et al., 1993). Feeding mice with diets enriched in n-3 fatty acids from fish oil resulted in significantly smaller numbers of primary breast tumors and smaller total metastatic load (Hubbard et al., 1998). A higher intake of n-3 fatty acids alone might not necessarily produce a greater tumor inhibition. Current evidence indicates that the balance between n-3 and n-6 is an important factor in determining the outcome, since n-3 fatty acids are competitive inhibitors of the effects of n-6 fatty acids (Cohen et al., 1993). Matsuba et al. examined the effects on animal tumorigenesis of three oils that differed in their n-6/n- 3 ratio due to differences in the concentrations of linoleic acid (LA) and a-linolenic acid (ALA). In comparison with safflower oil (70% LA and 0.1% ALA) and soybean oil (50% LA and 5% ALA), perilla oil (15% LA and 65% ALA) inhibited breast tumorigenesis in rats and mice (Matsuba et al., 1998). 1.2.5.3 Other Types of Dietary Fat Monounsaturated fatty acid Oleic acid is the major form of monounsaturated fatty acid in human diet. Experimental studies on the effects of oleic acids on mammary tumorigenesis have produced inconsistent results (Fay et al., 1997; Welsch, 1992). While some studies reported that oleic acid reduced mammary tumor incidence (Cohen et al., 1986; Cohen and Wynder, 1990), some others found that oleic acid stimulates the proliferation of breast cancer cell (Chajes et al., 1995; Hardy et al., 2000; Rose and Connolly, 1990). Olive oil, which is rich in oleic acid, has been reported to have a cancer-suppressive effect (Cohen et al., 2000; Cohen et al., 1986; Solanas et al., 2002). However, it is probably not a function of increased 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. monounsaturated intake but maybe due to the corresponding reduced intake of n-6 fatty acids and increased intake of other components in olive oil, such as vitamins, flavanoids, and phenolic compounds, that may be protective against the development of breast cancer (Ip, 1997; Trichopoulou et al., 2000). Actually in an animal study, when n-6 fatty acids were added to the olive oil diet, the protective effect by olive oil was nullified (Lasekan et al., 1990). Saturated fatty acids In the previously mentioned meta-analysis of animal studies, saturated fatty acids were reported to have a tumor-enhancing effect that is much weaker than n-6 polyunsaturated fat (Fay et al., 1997). However, the effect of saturated fat is less conclusive than polyunsaturated fat. It is also suggested that saturated fats have no direct promotional effect on experimental mammary carcinogenesis (Rose, 1997). Some studies also observed that saturated fatty acids diminished the proliferation rate of breast cancer cells (Hardy et al., 2000; Rose and Connolly, 1990; Singh et al., 1995; Tinsley et al., 1981). trans fatty acids Major sources of trans unsaturated fatty acids are partially hydrogenated fats such as hard margarines and fiying fats used in industrial food preparation and in fast-food restaurants. The predominant trans fatty acid formed during hydrogenation is an 18-carbon trans monoene fatty acid. In the dimethylbenz- (a) anthracene (DMBA)-induced rat mammary tumor model (Erickson et al., 1984; Selenskas et al., 1984) and in the transplantable tumor model (Erickson et al., 1984), trans monoene fatty acids behave similarly to cis monoene fatty acids with respect to modulation of mammary tumor development. 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Conjugated linoleic acid Conjugated linoleic acid (CLA) is a collective term referring to a mixture of variants of linoleic acid that have unique structure of two double bonds being separated by a single bond. CLA is normally found as a minor constituent in the lipid fraction of many different kinds of food. Milk, cheese, and other dairy products are good sources of CLA, whereas seafood and vegetable oils are not (Ip, 1997). Meat from ruminants generally contains more CLA than does meat from non-ruminants. In contrast to linoleic acid, which has been observed consistently to enhance mammary tumorigenesis in rodents over a wide concentration range, CLA is inhibitory at amounts < 1% in the diet (Ip et al., 1994). The activity of CLA does not appear to be influenced by the amount or the type of dietary fat consumed by the host and the cancer-preventive activity of CLA is unlikely to be mediated by interfering with the disposition or metabolism of 18:2n-6 (Ip, 1997). The effect of CLA reaches a maximum at 1%, suggesting a limiting step in the capacity to convert CLA to some active product that is essential for inhibition of carcinogenesis. 1.3 N-6 Polyunsaturated Fatty Acid Lipoxygenase Pathway and the Development of Breast Cancer As tumor-enhancing effects of n-6 polyunsaturated fatty acids have been well established from experimental studies, recent studies have investigated the mechanism for the carcinogenic effects of n-6 fatty acids. It has been demonstrated that the n-6 fatty acids must undergo oxidative metabolism to produce a series of active metabolites, called eicosanoids, to elicit diverse biological activities in cancer development and progression. Blocking n-6 fatty acid metabolic pathways inhibits 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. their effects in the regulation of the metastatic phenotypes (Rose and Connolly, 1990). Human mammary epithelial cells that lose their ability to synthesize arachidonic acid from linoleic acid failed to show a growth stimulation effect in vitro when cultured in the presence of linoleic acid (Balakrishnan et al., 1989). Therefore, understanding the roles of n-6 fatty acid metabolism in cancer development is an essential topic in cancer epidemiology. The major n-6 fatty acid incorporated into cell membranes and that can be later activated and released from cell membranes is arachidonic acid (C20:4 n-6 ), a derivative of dietary linoleic acid. Arachidonic acid can also be obtained directly from food, such as meat. Arachidonic acid can be further metabolized through three major pathways, i.e. (1) the cyclooxygenase (COX) pathway, which produces prostacyclin (PGI2), thromboxane (TXA2), and various prostaglandins (PGs); (2) the lipoxygenase (LOX) pathway, which gives rise to a variety of hydroxyeicosatetraenoic acids (HETEs), leukotrienes, and lipoxins; and (3) cytochrome P-450 epoxygenase pathway, from which various epoxyeicosatrienoic acids (EETs) are derived (Tang and Honn, 1997) (figure 1.3). These n-6 fatty acid metabolites have potent and diverse biological activities and have been demonstrated to act as mediators or regulators in the cardiovascular system, the immune system, the central nervous system, the reproductive system and during the course of inflammation (Vane and Botting, 1994). They are now believed to play important roles in tumor promotion, progression, and metastasis (Steele et al., 1999). While much attention has been focused on prostaglandins and other COX-derived metabolites, mounting evidence suggests that 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LOX-catalyzed products also exert profound biological effects on the development and progression of human cancers. 1.3.1 N-6 Fatty Acid Lipoxygenase Pathway The lipoxygenases (LOXs) comprise a family of non-heme iron-containing dioxygenases, catalyzing the regio-selective and stereo-selective dioxygenation of free and esterified polyenic fatty acids to the corresponding hydroperoxy derivatives, which may be metabolized to various bioactive lipid mediators, such as hydroxyeicosatetraenoic acids (HETEs), or further into leukotrienes and lipoxins through additional sequential reactions, depending on the biosynthetic capacity of each special cell type (Brash, 1999). Diet Diet elongation & desaturation Arachidonic acid Linoleic acid Epoxygenase Prostaglandins Epoxy metabolites 13-HODE 12-Lipoxygenase 15-HETE 5-HpETE -► 5-HETE -►5-oxoETE 12-HETE Leukotrienes Cyclooxygenas Lipoxygenase 15-Lipoxygenase-2 5- Lipoxygenase 15-Lipoxygenase-l Figure 1.3 Metabolic pathways for biosynthesis of eicosanoids from n-6 fatty acids 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LOXs are categorized into subgroups according to the positional specificity of oxygen insertion into the arachidonic acid. The human LOX family mainly includes 1 2-, 15- and 5-lipoxygenases that catalyze the oxygenation of the 1 2-, 15- or 5-carbon atoms of arachidonic acid, respectively. Both 12-LOX and 15-LOX exist as families of isoforms, exhibiting tissue-specific expression and differential substrate preference. 12-LOX has three isoforms, i.e. platelet-type, leukocyte-type and epidermal-type 12- LOX, which are isolated from platelet, leukocyte and skin, respectively (Nie et al., 1999). However, their expressions are not restricted to these tissues. 15-LOX has two isoforms, 15-LOX-l and 15-LOX-2, with distinct tissue distribution (Brash et ah, 1997). Most LOXs use arachidonic acid as a substrate. 15-LOX-l also catalyzes direct oxidation of linoleic acid to generate 13-hydroxyoctadecadienoic acid (13- HODE) (Baer et ah, 1991; Daret et ah, 1989) (figure 1.3). Functions of the eicosanoids derived from n-6 fatty acids through lipoxygenase pathways have not been well characterized. They are found to be involved in a wide- spectrum of physiological and pathological activities, such as mediating inflammatory reaction, stimulating contraction of smooth muscle and constriction of bronchi, inhibiting platelet aggregation, chemoattracting leukocytes, stimulating insulin secretion and regulating cell growth and adhesion (Piper and Samhoun, 1987; Spector et ah, 1988; Tang and Honn, 1997). Human lipoxygenases are encoded by genes from a multigene family, in which each of the lipoxygenases and their isomers is encoded by a unique gene. LOX genes exhibit homology in structure and sequence. All LOX genes share a common 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. organization of exons and introns, indicating their origin from a common ancestral gene. With the exception of 5-LOX, all LOX genes have been found close together on chromosome 17pl3 (Funk, 1996). The 5-LOX gene is distinct from other lipoxygenase genes in its large size and its location on a separate chromosome, chromosome 10. Promoter regions of LOX genes show the features of housekeeping genes, however; the expression of LOX genes is highly regulated at the transcriptional and translational levels (Tang and Honn, 1999). 1.3.2 Roles of Lipoxygenases in Breast Cancer Evidence from studies with human cancer cells shows that the LOX pathways are involved in carcinogenesis in several major tissues including breast, prostate and colon, and that inhibiting these pathways may inhibit carcinogenesis (Steele et al., 1999). Understanding the roles of lipoxygenases and their metabolites in breast cancer may help to establish the LOX pathway as a target of chemopreventive intervention and gene therapy. 1.3.2.1 the Platelet-type 12-LOX Platelet-type 12-LOX is expressed in various human tumors, including prostate cancer (Gao et al., 1995), lung cancer (Hagmann et al., 1995), melanoma (Timar et al., 1999), and breast cancer (Natarajan et al., 1997). The arachidonic acid 12-LOX pathway metabolite, 12-HETE, has been demonstrated to play an important role in tumor metastasis, including (1) increasing adhesion of tumor cells to the subendothelial matrix (Liu et al., 1994; Liu et al., 1991); (2) increasing tumor cell 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. motility; (3) increasing invasive potential (Honn et al., 1994); and (4) stimulating angiogenesis (Nie et al., 2000). Linoleic acid is reported to enhance the invasive capacity of breast cancer cells and this effect can be blocked by inhibiting 12-LOX activity (Liu et al., 1996a). 12- LOX activity and expression are increased in breast cancer tissues compared to the uninvolved normal tissue, and are increased in cultured breast cancer cells compared to normal breast epithelial cells (Natarajan et al., 1997). High-metastatic tumor cells have been shown to synthesize greater amounts of 12-HETE than the low-metastatic tumor cells, suggesting that the biosynthesis of 12-HETE by tumor cells is a determinant of their metastatic potential (Liu et al., 1994). In addition, enhanced biosynthesis of 12-HETE is associated with an increase in proteolytic enzyme activity and tumor cell invasiveness. Overexpression of 12-lipoxygenase in the estrogen- dependent and poorly invasive MCF-7 breast cancer cell line is associated with both increased cell proliferation and reduced apoptosis in athymic nude mice, accompanied by high angiogenic activity (Connolly and Rose, 1998), and estrogen-dependent cells become estrogen independent (Liu et al., 1996b). 1.3.2.2 5-LOX and 5-LOX Activating Protein (FLAP) 5-LOX catalyzes the first two steps in biosynthesis of leukotrienes. The first step is the oxidation of arachidonic acid to 5-HpETE. The second step is the dehydration of 5-HpETE to form leukotriene A4 (Ford-Hutchinson et al., 1994). 5- HpETE can also be spontaneously converted to 5-HETE and then 5-oxoETE (figure 1.3). 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Arachidonic acid metabolism through both 5-lipoxygenase pathways requires 5-lipoxygenase activating protein (FLAP), which transfers arachidonic acid to 5-LOX (Mancini et al., 1993). Both 5-LOX and FLAP have been implicated in several types of cancers. The expression of 5-LOX and FLAP has been found in multiple breast cancer cell lines, and lipoxygenase or FLAP inhibitors can effectively inhibit the growth of these cancer cells (Hong et al., 1999). The major 5-LOX metabolic product, 5-HETE, is produced in response to insulin-like growth factor-1 (IGF-1) and gastrin- releasing peptide (GRP) in lung cancer cells, and it can stimulate proliferation of small-cell lung cancer cell lines (Avis et al., 1996). Blocking the 5-LOX pathway reestablished apoptotic clearance in lung cancer (Avis et al., 1996) and in breast cancer (Avis et al., 1998; Ogmundsdottir et al., 1998). Addition of 5-HETE to breast cancer cells results in growth stimulation, whereas selective inhibition of 5-LOX or FLAP reduces cell growth, increases apoptosis, and increases G1 arrest (Avis et al., 2001). A recent study found particularly high levels of 5-LOX in tumors from patients dying of breast cancer (Jiang et al., 2003). However, there was also a group reporting conflicting results about 5-LOX in breast cancer, showing that 5-LOX derivatives inhibit cellular growth in breast cancer MCF-7 cells (Przylipiak et al., 1998a; Przylipiak et al., 1998b). 1.3.2.3 15-LOXs Both 15-LOX-l and 15-LOX-2 have been found to play roles in the development of breast cancer. Expression of 15-LOX-l has been observed in normal breast tissues and was increased in some cancer tissues (Natarajan et al., 1997). In 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. breast carcinoma BT-20 cells, EGF and TGF-a dramatically stimulate metabolism of linoleic acid by 15-LOX-l to 13-HODE, which is correlated with the stimulation of DNA synthesis (Reddy et al., 1997). Treatment of BT-20 cells with a selective 15- LOX-l inhibitor results in a concomitant inhibition of 13-HODE biosynthesis and cell proliferation, which is partially restored by addition of exogenous 13-HODE. 1.3.3 Interactions among Lipoxygenase Pathways Besides the independent roles of each LOX pathway in the development of cancers, these LOX pathways also interact with each other, as well as with other arachidonic acid metabolic pathways. It has been reported that 15-HETE can inhibit activities of both 5-LOX and 12-LOX, whereas 12-HpETE, the intermediate precursor of 12-HETE, can increase 5-HETE and LTB4 production (Gurr and Harwood, 1991; Wachtershauser et al., 2000). With the mediation of FLAP, 5-LOX can oxygenate 12- HETE and 15-HETE into further products, whose functions are not clear yet (Mancini et al., 1998). FLAP inhibitors also inhibit 12-HETE production through 12-LOX pathway, suggesting that FLAP is probably involved in the 12-LOX pathway as well as 5-LOX pathway (Ozeki et al., 1999). However, so far there is not a comprehensive picture for the interaction among these lipoxygenase pathways. 1.3.4 Mechanisms Involved in the Lipoxygenase-mediated Tumor Development Mechanisms of lipoxygenase-mediated growth regulation are under investigation. It has been reported that LOX metabolites can induce the activation of oncogenes such as c-fos and ras (Haliday et al., 1991). They also mediate the growth 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. promoting effect of EGF by regulating the tyrosine kinase activity of the EGF receptor (Eling and Glasgow, 1994). 12-HETE can enhance tumor cell adhesion to extracellular matrix (ECM) by increasing integrin expression at cell surface and by inducing reestablishment of the filamentous cytoskeleton system, and this process is protein kinase C (PKC)-dependent (Tang and Honn, 1999). 12-LOX regulates cell survival and apoptosis by affecting the expression and localization of the vitronectin receptors, a(v)P(3) and a(v)P(5) (Pidgeon et al., 2003). Blockade of both 5-LOX and 12-LOX pathways induces apoptosis in breast cancer cells through the cytochrome c release and caspase-9 activation, with changes in the levels of Bcl-2 family proteins (Tong et al., 2002). A receptor-mediated action was also demonstrated, since 15- HETE and 13-HODE can act as ligands of PPARy (Kliewer et al., 1997) and a newly identified 50KD protein (Kurahashi et al., 2000). Lipoxygenase metabolites are also suggested to be related to the formation of etheno-DNA adducts under oxidative stress and lipid peroxidation (Nair et al., 2000). In addition, non-eicosanoid-related activities of lipoxygenase, such as peroxidation of membrane lipids, may contribute to the alterations in cellular functions as well (van Leyen, 1998). Since all the 5-, 12- and 15-lipoxygenase pathways may be involved in breast cancer development, and their interaction may also influence tumor cell growth, we designed a molecular epidemiologic study that examined all three pathways as candidate pathways for breast carcinogenesis. 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 2 Hypotheses and Study Aims The hypothesis that dietary fats influence breast cancer risk has been the source of much debate. While animal studies have implicated dietary fats and fatty acid metabolites in mammary carcinogenesis, epidemiologic studies have failed to demonstrate a convincing link. One weakness of previous epidemiologic studies is the failure to consider metabolic differences among individuals. We hypothesized that genetic variants in fatty acid metabolic pathways may influence the development and progression of breast cancer. Furthermore, we hypothesized that by examining genetic and dietary influences together, we may be able to identify dietary risk factors that are obscured when diet is examined in isolation. We tested whether genetic polymorphisms in fatty acid metabolic pathways are associated with breast cancer risk. Our candidate genes included the lipoxygenase (LOX) gene family (5-LOX, 12-LOX and 15-LOX) and the 5-LOX activating protein (FLAP) gene. Specifically, we (1) sequenced these LOX genes to identify all common polymorphisms in LOX gene coding regions and regulatory regions; (2) conducted laboratory studies to determine the functional significance of polymorphism in the promoter region of the 5-LOX and FLAP genes, and (3) performed epidemiologic analyses to examine the association between the LOX gene polymorphisms and breast cancer risk, and to determine whether LOX gene polymorphisms modify the effect of dietary fat intake on breast cancer risk. 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. If we find that genetic variation in specific LOX pathways contributes to breast cancer risk, this evidence will point to specific components of high fat diets (namely, n-6 fats) that may increase risk. Such a finding will provide a scientific foundation upon which to design dietary intervention. 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3 Study Design This study derives from a population-based case-control study conducted in the San Francisco Bay area by Dr. Esther M. John with funding from the National Cancer Institute (NIH/NCI R01 CA63446), the Department of Defense (DOD DAMD17- 96-1-6071) and California Breast Cancer Research Program (CBCRP 1RB0125). The study of dietary fat, fat metabolizing gene and breast cancer risk has been funded by California Breast Cancer Research Program, starting from July 2001 (PI: Ingles, CBCRP 7WB-1-0110). 3.1 Study Population Cases: Newly diagnosed breast cancer patients were identified through the population-based cancer registry that covers the San Francisco Bay area and is part of the Surveillance, Epidemiology, and End Results (SEER) program and the state-wide California Cancer Registry. Eligible cases included Hispanic, African-American, and White women aged 35-79 years, diagnosed with a first primary invasive breast cancer between April 1, 1995 and April 30, 1999, and residing in the San Francisco Bay area (i.e., San Francisco, San Mateo, Santa Clara, Alameda, and Contra Costa counties) at the time of diagnosis. Of 10,103 identified cases, 392 (3.9%) were deceased, and 157 (1.6%) could not be contacted due to physician-reported contraindications. Of 9,554 cases contacted by telephone, 8,042 (84%) completed a screening interview that established study eligibility, assessed personal and family history of breast cancer, and 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. verified race/ethnicity which, as recorded in the cancer registry records, is not always consistent with self-report (Swallen et al., 1997). Among non-participants, 667 (7.0%) were too ill or refused participation, 73 (0.8%) did not speak sufficient English or Spanish to complete the interview, 600 (6.3%) could not be located or had moved, 172 (1.8%) could not be reached despite more than ten attempts over an extended period, and 35 were not screened due to end of study. All Latina and African-American cases and a 10% random sample of Whites were eligible for the in-person interview. Of 2,054 cases (735 Latinas, 625 African-Americans, 694 Whites), 1,788 (87%) completed the in-person interview, including 648 (88%) Latinas, 543 (87%) African- Americans, and 597 (86%) Whites. Among the remaining cases, 188 declined participation, 50 were too ill, and 27 could not be located or reached after multiple attempts. One incomplete interview was excluded from the analysis. Controls: Population controls were identified through random-digit dialing. Using telephone numbers of recently diagnosed cancer patients and replacing the last two digits with random numbers, 10 phone numbers per case number or a total of 108,007 numbers were generated. Among the 65,122 numbers assessed as residential, nobody was reached (i.e., no answer or answering machine only) at 14,737 numbers despite a minimum of 10 attempts over a two to four week period. Among the remaining 50,385 numbers, a household enumeration was completed for 41,427 (82%) numbers. From the pool of potentially eligible controls, 2,999 were randomly selected according to the expected race/ethnicity and 5-year age distribution of cases, at approximate case: control ratios of 1:1 Whites, 1:1.1 for African-Americans, and 1:1.2 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. for Latinas. Fourteen control women were deceased by the time they were contacted about the study. Among the remaining 2,985 controls, 2,425 (82%) were screened, 207 (7.0%) were too ill or refused to participate, 9 did not speak sufficient English or Spanish, and 195 (6.6%) had moved or could not be located. A total of 2,523 controls were invited to participate in the study (1,021 Latinas, 726 African-Americans, 776 Whites) and 2,129 (84%) completed the in-person interview, including 885 (87%) Latinas, 598 (82%) African-Americans, and 646 (83%) Whites, 341 controls were too ill or declined participation, and 52 could not be located or reached after multiple attempts. One incomplete interview was excluded from the analysis. 3.2 Data Collection Trained bilingual, bicultural interviewers administered a structured questionnaire on demographic and cultural background, residential history, physical activity, sunlight exposure, diet, supplement use, body size, change in weight, occupational history, menstrual and reproductive history, and medical history up to the reference year, defined as the year prior to diagnosis for cases or the year prior to selection into the study for controls. In addition, the interviewers measured skin pigmentation, standing height, weight, and hip and waist circumferences. 3.3 Food Frequency Questionnaire Development and Dietary Assessment Usual dietary intake was assessed using a semiquantitative food-frequency questionnaire (FFQ) adapted from Block’s Health History and Habits Questionnaire (Block et al., 1986; Hom-Ross, 2001). Block’s Health History and Habits Questionnaire was first developed in 1986 for epidemiologic and clinical use. The 49 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. food list and the associated nutrient values were developed using dietary data from the Second National Health and Nutrition Examination Survey (NHANES II). Food items were selected on the basis of their contribution to total population intake of energy and major nutrients in the NHANES II data. The food items in the list account for at least 90% of each of the major nutrients. The quantitative values to be assigned for each food on the questionnaire were determined from the NHANES II database using frequency and portion-size information in that survey. Several potential advantages of this instrument include: (1) selection of the food items based on population data on major nutrient contributors; (2) use of population intake data in assignment of nutrient value to line items which combine several foods; (3) development of portion sizes based on population data; (4) assessment of the whole spectrum of micro- and macronutrients rather than a focus on a few nutrients; and (5) potential for self- administration. The full-length 100-item Block questionnaire has been validated in older men (Sobell et al., 1989), middle-aged women (Block et al., 1990b), and in a younger population containing black and white men and women (Block et al., 1992). It has shown to produce not only good ranking individuals but also accurate estimates of mean nutrient intake. The questionnaire has been validated in low-income black women in Atlanta using serum carotenoids and alpha-tocopherol as the reference data (Coates et al., 1991), it produced correlations as least equal to those seen in white populations. A 60-item version of the Block FFQ has been also shown to perform 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. almost as well as the 100-item version in validations against multiple days of dietary records (Block et al., 1990a). The dietary questionnaire used in this study was developed from the Block 1995 FFQ (106 items). The questionnaire was modified and more ethnic-specific food items were added for African-American and Hispanic populations. Using this modified FFQ, participants of this study were asked to estimate usual dietary intake during the reference year. Each food item had eight response options for frequency (“never or rarely” to “4+ per day”) and three options for portion size. Food models and utensils were used to estimate serving size. The food frequency questionnaires also included adjustment questions on food preparation, types of meat for cooking and doneness of meat. Consumption of low-fat food and fats used in cooking were also assessed as these variables can have large effects on estimates of fat intake. The nutrient values were obtained from the nutrient database for the Study of Women’s Health Across the Nation (SWAN) (Huang et al., 2002) and Block 1995 FFQ. The nutrient database was reviewed and updated in 1999 and 2001, based on USD A, Pennington, and databases from ESHA and Nutritionist V software program. 3.4 Biospecimen Collection Blood or mouthwash samples were collected for a subgroup of cases diagnosed between May 1, 1997 and April 30, 1999 and their frequency matched controls. In a second home visit, participants were invited by professional phlebotomists to provide a 40 ml sample of whole blood. Those who declined the blood draw were given to option to provide a mouthwash sample. Blood samples were collected before 10 AM 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. following a 12 hour fast. Mouthwash samples were collected using Scope mouth rinse, about 1 hour after the participant brushed her teeth. Of the 928 cases who completed the interview, 814 (88%) provided a biospecimen (686 blood samples and 128 mouthwash samples), including 287 (88%) Latinas, 250 (85%) African- Americans, and 277 (90%) Whites. Of the 1,046 controls who participated in the interview portion of the study, 910 (87%) provided a biospecimen (813 blood samples and 97 mouthwash samples), including 357 (85%) Latinas, 255 (84%) African- Americans, and 298 (93%) Whites. DNA was successfully extracted for 813 cases and 906 controls. For cases, DNA extraction from blood samples failed for one participant. For controls, extractions from mouthwash or blood failed for three and one participants, respectively. 3.5 Exposure and Confounding Variables Intake of specific nutrients was estimated by linking the food frequency questionnaire data to its nutrient database, including protein, carbohydrate, fiber, total fat, and types of fat (saturated fat, oleic acid, and linoleic acid). For a subgroup of cases diagnosed between 1995-1998 and their matched controls, a question was also asked for the most commonly used cooking oil. BMI as a measure of adiposity was computed as weight (kg) divided by height (m) squared based on measured height and weight or self-reported height and weight for the 11% of cases and 10% of controls who declined the body measurements. Women were considered postmenopausal if their periods had stopped more than 1 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. year prior to diagnosis/selection and they had never used HRT or used HRT only after the cessation of menses. Also included in this group were women who began using HRT prior to the cessation of menses but had attained age 55 or older at the time of diagnosis/selection, or women who reported a bilateral oophorectomy, or women who reported a hysterectomy but had attained age 55 or older at the time of diagnosis/selection. For 132 cases and 173 controls under age 55 menopausal status could not be determined as they began using HRT prior to the cessation of menses or reported a hysterectomy only. The remaining women were considered premenopausal. Average lifetime physical activity was estimated for activities (i.e., recreation, transportation, chores, and jobs) performed between age at menarche and the reference age, by summing the average weekly hours for each of the four sources of activity. A positive history of benign breast disease was defined as having a biopsy for benign breast disease at least 2 years prior to diagnosis (cases) or selection into the study (controls). 3.6 Laboratory Methods 3.6.1 General Quality Control Laboratory personnel were blinded to case/control status of DNA samples. DNA samples were labeled only by specimen ID. All genotyping assays include control samples with genotype previously confirmed by sequencing. All PCR assays include a “water blank” to guard against contamination of the PCR reaction. To assure assay reproducibility, 5% of samples were repeated with laboratory personnel 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. blinded to repeat status. Results from validation study including 75 pair of duplicates showed 100% consistency. 3.6.2 DNA Extraction Buffy coat isolated from whole blood was used for DNA extraction using a protocol developed by Bass et al (Baas et al., 1984). DNA was quantitated using VersaFluor Cuvette Fluorometer with DNA quantitation kit (Bio-Rad, Hercules, CA). 3.6.3 Polymorphism Screening For each gene of interest, 100 chromosomes were examined from 50 control subjects chosen from four ethnic groups: African-American, Hispanic, non-Hispanic white, and Asian. This will give greater than 99% power of detecting a polymorphism with a prevalence of 5% (Cargill et al., 1999). Detection of low-prevalence polymorphisms (prevalence 5-10%) is important because they are more likely to be functionally important (Cargill et al., 1999). To screen for coding region polymorphisms, RNA was extracted, reverse transcribed to cDNA, and then sequenced. For regulatory regions, genomic DNA was sequenced directly. Both forward and reverse strands were sequenced. ABI Sequencing Kit (Applied Biosystems, Foster City, CA) was used to do the sequencing reactions according to manufacturer’s instructions. Sequencing was performed on the Perkin-Elmer ABI Prism 3700 capillary sequencer. 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.6.4 Genotyping of the LOX Genes 3.6.4.1 Detection of the FLAP Gene Microsatellite The SSLP (simple sequence length polymorphism) analysis was used to identify the FLAP gene microsatellite allelic variants. The genomic region containing the FLAP gene poly(A) polymorphism (-211 to -108) was PCR-amplified using primers end-labeled with 3 3 P-y-ATP. The forward primer was 5’GGGACACACTG AACCACAGC3’. The reverse primer was 5’GAAGATCCCCGGCACAATTA3’. The 10 ul PCR mixture contained lx PCR buffer, 0.25 mM dNTP, 4pmol of each primer, 1.5 mM MgCl2, 1 unit Taq Polymerase and 15 ng genomic DNA. The PCR conditions are 3 min at 94°C, followed by 30 cycles of 94°C fori min, 61°C for lmin, and 72°C for 30s. Chain elongation was continued after the last cycle for 10 min. The PCR products were denatured at 95°C for 5 minutes and 3 ul of each sample was loaded on a 6% denaturing polyacrylamide gel and run at room temperature for 2 hours at 80w. Gels were then dried and exposed to x-ray films for 24 to 48 hours. Alleles were sized by comparison to a previously sequenced control. 3.6.4.2 Detection of the 5-LOX Gene Promoter Addition/Deletion Polymorphism GeneScan method was used to identify the 5-LOX gene promoter region 6-bp addition/deletion polymorphism. The genomic region containing the polymorphism (- 204 to -47) was PCR-amplified with fluorescently-labeled primers. The HEX-labeled forward primer was 5’CGTGAAGAGTGGGAGAGAAGTA. The non-labeled reverse primer was 5 ’ TCCAGGTATCCGCATCTAGC3 ’. The 10 ul PCR mixture contained lx PCR buffer, 0.2 mM dNTP (7’deaza-dGTP:dGTP=3:l), 5 pmol of each primer, 1.5 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. mM MgCh, 0.5 ul of DMSO, 1.25 unit Taq polymerase and 20 ng genomic DNA. The PCR conditions were 6 min at 94°C, followed by 35 cycles of 94°C for 30 s, 53°C for 30 s, and 72°C for 30 s. Chain elongation was continued after the last cycle for 5 min. The PCR products were 1:10 diluted and run on ABI 3700 Sequencer together with a ROX-labeled size standard (GeneScan 400HD Size Stand, PE ABI). Results were analyzed using GeneScan program. 3.6.4.3 Genotyping of SNPs in LOX Genes Genotyping of SNPs (single nucleotide polymorphisms) was performed by the TaqMan assay using the TaqMan Core Reagent Kit (Applied Biosystems, Foster City, CA) according to manufacturer’s instructions. Probes and primers being used were listed in the table 3.1. 15 ul TaqMan reaction contains 1.5 ul 1 O x buffer, 1.2 ul internal standard, 1.5 ul stabilizer, 3.5 mM MgCl2, 0.2 mM dNTP, 300 nM of each primer, 100 nM of each probe, 0.3 u Taq polymerase, and 20 ng genomic DNA. The conditions for real time PCR were 10 min at 95°C, followed by 50 cycles of 95°C for 25 s and annealing temperature (58°C for 5-LOX 760 A/G and 15-LOX2 1967 A/G, 63°C for 12-LOX 782 A/G, and 60°C for 12-LOX 965 A/G and 5-LOX-1285 G/T) for 60s. The fluorescent signal was measured in an Applied Biosystems Sequence Detection System model 7900HT. Multicolor analysis was used to detect both alleles of the biallelic system. Experimental samples were compared to 9 previously sequenced controls (3 of each genotype) to identify the 3 genotypes at each locus. Any samples outside of the parameters defined by the controls were identified as non-informative and were repeated. 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.1 Probes and primers for the LOX gene TaqMan genotyping Gene & SNP Probes Primers 5-LOX gene 760A/G 5’ 6FAM CTGCCCAAGAAGC 3’ 5’ VIC CTGCCCGAGAAGC 3’ F: 5’ GACCTGATGTTTGGCTACCAGTT 3’ R: 5’ CCAGGCTGCACTCTACCATCT 3’ 5-LOX gene 1286G/T 5’ 6FAM CATGTATCCGATTAGAGAC 3’ 5’ VIC TCATGTATCCTATTAGAGACT 3’ F: 5’ CGC AGA AGC AAT AAA AAT GTC TGT 3’ R: 5’ TTT GGA AAA CTG GGC TAT TTA TCT TT 3 ’ 12-LOX gene 782A/G 5’ 6FAM AAGAGCTTCAGGCTCAAC 3’ 5’ VIC AGAGCTTCGGGCTCAA 3’ F: 5’ CTCTCTGCCCTCCAGGCTAGT 3’ R: 5’ TTTGGATTCTCTCTGTCCGTTCTC 3’ 12-LOX gene 965A/G 5’ 6FAM CCTCCCAACCCCAGC 3’ 5’ VIC CCTCCCAGCCCCAG 3’ F: 5’ AATGGGAGTTCAATAATTTCTCTTTCTC 3’ R: 5’ TGAGGGCAGGAACAGTGTTG 3’ 15-LOX2 gene 1967A/G 5’ 6FAM AGGCCCTGGTTCC 3’ 5’ VIC AGGCCCCGGTTCC 3’ F: 5’ GAG CAT CGC CAC CTT CCA 3’ R: 5’ ATG TGG TCA AAG GGA TGT CAT CT 3’ 3.6.5 Luciferase Assay of the 5-LOX and FLAP Gene Promoter Region Polymorphisms To determine whether the allelic variants of the 5-LOX and FLAP gene promoters differ in their ability to drive reporter gene expression in breast cancer cell lines, each allelic variant was amplified from the genomic DNA of a subject who was homozygous for that allele or, if homozygotes were not identified, who carried one copy of the allele (identified by genotyping). PCR products from these individuals were cloned using TOPO TA cloning kit (Invitrogn Corporation, CA). The clone carrying each allele was identified by sequencing. Allelic variants were compared with respect to their ability to drive gene expression in a luciferase reporter gene system. To compare 5-LOX gene promoter region Spl binding site polymorphism, pGL3-5LO reporter plasmids for each of the four allelic variants were constructed by inserting cloned PCR products upstream of the firefly luciferase gene in the pGL3-basic vector (Promega, WI). Each of the four pGL3-5LOX vectors was transfected into the human breast cancer cell line SK-BR3, 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. using the SuperFect System (Qiagen, CA). The phGL3-SV40 control vector (Promega, WI) containing Renilla luciferase was co-transfected with each pGL-5LOX vector as an internal control to normalize luciferase activities. The Dual-Luciferase Reporter Assay System (Promega, WI) was used for the luciferase activity detection, and luciferase activity was measured in a TD-20/20 single-channel luminometer (Turner Designs, CA) (figure 3.1). Luciferase activity was expressed as relative light units, dividing total firefly luciferase by total renilla luciferase activity. Three to five clones for each allele were prepared independently and were confirmed for the polymorphism by direct sequencing. Each clone was repeated in triplicate in each assay and the assay was repeated four times. pGL3-5LOX phGL-SV40 5LOX promoter Firefly Luciferase Renilla Luciferase co-transfection O OQ Breast cancer cell 48 hrs 0 ~ o :o Cell Lysis Luciferase Assay Figure 3.1 Schematic diagram of the transfection and luciferase assay of 5-LOX gene Spl binding site polymorphism 58 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The procedure for the FLAP gene poly(A) microsatellite luciferase assay was the same as that for 5-LOX except that the pGL3 vectors cloned with FLAP poly(A) microsatellite were used for the transfection and luciferase assay. 3.7 Data Management and Statistical Analysis 3.7.1 Data Management Genotype results and specimen ID were entered into a spreadsheet and sent to Dr. John at NCCC. The NCCC data manager merged genotype results to the database that contained case/control status, ethnicity, dietary variables, and other covariates. This working database was stripped of specimen IDs for data analysis. 3.7.2 Statistical Analysis 3.7.2.1 Analysis of Dietary Data Of the 3917 women who completed the food frequency questionnaire, 129 women (3.3%) with daily total energy intake lower than 500 kcal or higher than 5000 kcal (possibly indicating unreliable data) were excluded from analysis. We also excluded another 101 women (2.6%) with major epidemiologic data missing. The final analysis was conducted among 3687 women. Odds ratios and 95% confidence intervals for breast cancer were estimated by unconditional logistic regression (Breslow and Day, 1980). All tests of significance were two-sided with p<0.05 as the significant cut-off point. To estimate the effects of nutrient intake that are independent of energy intake, all analyses were adjusted for total energy intake. We used two different energy-adjustment methods, the residual 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. method and the multivariate nutrient density method (Willett and Stampfer, 1998) (table 3.2). Both analyses are reported in the results. 1) Residual method: energy-adjusted fat intakes were computed as the residuals from the regression model with total caloric intake as the independent variable and absolute nutrient intake as the dependent variable. The fat residuals therefore provided a measure of fat intakes uncorrelated with total energy intake. By including fat residuals into logistic regression models according to quartile, odds ratios and 95% confidence intervals were calculated to estimate the association with breast cancer risk. However, this model does not provide an intuitive interpretation of actual fat intake. 2) Multivariate nutrient density method: fat density was computed by dividing the energy from fats by total energy intake and entering both density and total energy in a multiple logistic regression model. This model has a more intuitive interpretation as a measure of dietary composition. The odds ratios for fats can be interpreted as the effect of substituting calories from fat for the same percentage of calories from carbohydrate and protein. For each type of fat (saturated fat, linoleic acid and oleic acid), two models were tested: (i) with each type of fat put into the model separately according to quartile; (ii) with all three types of fat put into the model simultaneously for mutual adjustment. When testing the possible interactions between total fat (or types of fat) and certain covariates, total fat (or types of fat) were put into model linearly as 5% of total energy intake, instead of as quartiles. 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.2 Methods for energy adjustment in dietary data analysis_____________ Methods for energy adjustment_____________________________________ Model_______________________________ 1. Residual method Cancer =p0 + Pi Nutrient (total fat or types of fat) residual (quartile)* + p2 Calories + p3 _ p Covariates * Residual from regression model Nutrient= P o + Pi Calories 2. Multivariate M odel 1: Without m utual adjustm ent f o r types o ffa t nutrient density Cancer = P o + Pi Nutrient (total fat or types of fat)/Calories (quartile) + p2 method Calories + p3 .p Covariates M o d el 2: M utual adjustm ent f o r types o ffa t Cancer = P o + Pi saturated fat/Calories (quartile) + p2 linoleic/Calories (quartile) + p3 oleic/ Calories (quartile) + p4 Calories + p5 .p Covariates M odel 3: Interaction m odel Cancer = P o + Pi saturated fat (as 5% total energy) + p2 linoleic (as 5% total energy) + p3 oleic (as 5% total energy) + p4 Calories + p5 .p Covariates + P^ interaction In both the residual model and the nutrient density model, total caloric intake was included in the model with the nutrient calorie-adjusted term (residual) or the nutrient density term. This approach can reduce the random error if caloric intake has an important association with the outcome independent of nutrient intake. This approach also adjusts for the confounding that may exist between uncorrelated variables (fat residual and total calories) in nonlinear models, including logistic regression (Willett and Stampfer, 1998). Including total energy in the nutrient density model produces an ‘isocaloric’ analysis, in which the relation of the nutrient composition of the diet with disease is estimated, holding total energy intake constant. This approach is particularly advantageous when the variation in body size and total energy intake is great, because it does not assume a similar effect of nutrient intake in a very small subject (with low energy intake) and a very large subject (with high 61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. energy intake) (Willett and Stampfer, 1998). A comparative validation study of three different food frequency questionnaires showed that energy adjustment reduces measurement error in all FFQs, including the Block FFQ (Subar et al., 2001). In models using quartiles for analysis, quartiles were decided according to the distribution of total fat or types of fat intake among all controls from three ethnic groups. The lowest quartile was used as the referent. Trend tests were conducted using the median of each quartile. In addition to total energy intake, other potential confounders, selected based on their significance as main effects and on the literature, were also adjusted for in all models, including age at diagnosis for cases or at selection for controls (continuous), ethnicity (White, African-American, Latina), menopausal status (pre-, postmenopausal, undetermined), age at first full-term pregnancy (<20, 20-24, 25-29, 30+), BMI (continuous), height at interview (cm, continuous), first-degree family history of breast cancer (yes/no), age at menarche (continuous), foreign bom (yes/no), education (<12y, 12y, 13-15y, 16y and up), history of benign breast disease (yes/no), average lifetime physical activity (hrs/wk, continuous) and average alcohol intake during reference year (gm/d, continuous). To evaluate whether some factors such as menopausal status, family history of breast cancer, previous history of benign breast disease, physical activity, etc, modify the association between dietary fat and breast cancer risk, a cross-product term of each dietary fat (continuous in unit of 5% total energy intake) and these factors was included in the multivariate nutrient density model. The p value for the test of 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. interaction was obtained from the Wald test of the cross-product term. Multinomial logistic regression was used to compare the association between dietary fat and breast cancer at different stages, histological grades, and estrogen/progesterone receptor status. The p value for the test of heterogeneity was obtained from a likelihood ratio test with one degree of freedom. Spearman correlation coefficients were calculated to examine the correlations between total fat and types of fat. STATA version 7.0 (Stata Corporation, College Station, TX) was used for multichotomous logistic regression. All other analyses were performed using SAS version 8.2 (SAS Institute, NC). 3.7.3.2 Analysis of Genotyping Data Of the DNA available from 1719 subjects, 24 samples (1.4%) failed genotyping. 1 sample was excluded due to major epidemiologic data missing. The final analysis was conducted among 1694 DNA samples. For preliminary descriptive analyses, genotype frequencies were examined separately for cases and controls for each ethnic group. Chi-square tests were used to determine whether genotypic distributions among controls depart from Hardy- Weinberg equilibrium. Odds ratios and 95% confidence intervals for lipoxygenase gene genotype with respect to breast cancer were estimated by unconditional logistic regression. The most common genotype for each polymorphism was used as the referent group and other genotypes were compared to this “baseline” genotype. Tests of trend were performed by including in the logistic model a variable coded as 0, 1, and 2 for the number of “at-risk” alleles. Dominance and recessive effects were 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. explored for some polymorphisms using other coding schemes (e.g. 0, 1, 1 for dominant and 0, 0, 1 for recessive). For the 5-LOX promoter region Spl binding site polymorphism that has multiple alleles, multiple alleles were also categorized into two groups: wild-type (5) and mutated allele (non-5). Another two analyses were also conducted: 1) to categorize the allele into long (>=5) vs. short (<5) allele; 2) to score each allele by the number of Spl binding motif. Results were only reported for the method with wild-type vs. mutation. For genes with multiple polymorphisms (e.g. 12- LOX Gln261Arg and 12-LOX Ser322Asn), haplotype frequencies and haplotype specific odds ratios were estimated using the hapif command in Stata 7.0 (Mander, 2001). To estimate the odds ratios for halotypes, the probability of haplotypes given the genotype were estimated using the formula from Zaykin (Zaykin et al., 2002). For haplotype /*2 and h^, the conditional probability of pair Q 12, h^) for the ith individual with genotype G\ is Pr(fc,fo|G,)= Pr<& I ]TPr(Gi | hu,hv )PhuPhv u , v where Phu Phv denote haplotype frequencies. This probability was treated as an “importance weight” variable in the logistic regression model testing the association of haplotype with breast cancer. Potential confounders for genotypes were controlled by being included in the logistic regression model. Included factors were age (continuous), ethnicity (white, African-Americans, Latinas), menopausal status (pre-, postmenopausal, undetermined), age at first full-term pregnancy (<20, 20-24, 25-29, 30+), BMI 64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (continuous), height (continuous), family history of breast cancer (yes/no), age at menarche (continuous), foreign bom (yes/no), education (<12y, 12y, 13-15y, 16y and up), history of benign breast disease (yes/no), lifetime physical activity (continuous) and alcohol intake (continuous). The effect modifiers that have been considered included menopausal status, previous history of benign breast disease and dietary linoleic acid intake. Formal tests of effect-modification were performed by including the appropriate interaction terms in the logistic regression model. For analyses stratifying on linoleic acid intake, total caloric intake, saturated fat intake and oleic acid intake were also adjusted for in the models. 65 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER4 Results and Discussion 4.1 Summary of the LOX Gene Polymorphisms Some lipoxygenase gene polymorphisms have been reported in research publications or public genomic databases. However, these data are neither complete nor accurate, with a lot of missing or false positives. Therefore we systematically sequenced LOX family genes to detect all common coding region and regulatory region polymorphisms. For each gene, 100 chromosomes were examined from 50 control subjects chosen from four ethnic groups: African-American, Hispanic, non- Hispanic white, and Asian. To detect coding region polymorphisms, RNA was extracted, reverse transcribed to cDNA, and then sequenced. For regulatory regions, genomic DNA was sequenced directly. We identified a total of 48 common polymorphisms among LOX genes, of which 20 are synonymous coding region SNPs (single nucleotide polymorphisms) that do not change the amino acid sequence. More interestingly, we have identified 4 nonsynonymous SNPs (two in the 12-LOX gene, one in the 5-LOX, and one in the 15-LOX-2 gene), which are likely to alter protein function. In addition, we have found 20 polymorphisms among the 5'-regulatory regions, including 17 SNPs, one polymorphic microsatellite, and two deletion/addition polymorphisms. Four SNPs in the 3' untranslated regions were also identified for the 15-LOX-1 gene. These polymorphisms have the potential to alter the amount of protein product (either by influencing transcription, translation, or mRNA stability). For polymorphisms in regulatory region, “TFSEARCH (Searching Transcription 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Factor Binding Sites) (ver 1.3)” has been used to check for potential changes of transcription factor binding sites that the polymorphisms might result in. Nonsynonymous SNPs and known transcription factor binding site polymorphisms were given priority for functional and association studies in this dissertation. The LOX gene polymorphisms were summarized in table 4.1. Table 4.1 Summary of the LOX gene polymorphisms Gene Position Polymorphism Implication Platelet-type 5’ regulatory -541 A>G Unknown. Adjacent to NF-kB motif; A: USF 12-LOX region -320 T>C -217 A>G* (upstream stimulating factor) binding site Unknown. T: HSF (heat shock factor) binding site; C: GATA-1 binding site Unknown. A: ADR1 binding site; G: p54 (c-Ets binding factor) binding site exon 6 765 A>G* Synonymous exon 6 782 G>A* Arg261Gln exon 8 965 A>G* Asn322Ser exon 8 1092 G>T* Synonymous exon 11 1452 A>G* Synonymous exon 13 1789 O T * Synonymous exon 14 1875 C>T Synonymous exon 14 1902 A>G* Synonymous 5-LOX 5’ regulatory -1708 G>A Unknown. region -1368 GTTAAA deletion -1286 G>T -845 G>A -560 T>C -176 t o -147 GGGCGG deletion/addition* -59 C/T Unknown. Unknown. GATA-2 binding site Unknown. E2F binding site Unknown. T: HSF binding site Spl/Egr-1 binding region Unknown. GATA-1 binding site; C: E47 binding site Exon 1 21 O T * Synonymous Exon 2 270 A>G Synonymous Exon 6 760 A>G* Lys254Glu Exon 8 986 A>G Synonymous Exon 13 1728 A>G* Synonymous 15-LOX1 5’ regulatory region -611 A>G -295 O T -275 G>T -219 G>T -189 G>C Unknown Unknown. T: c-Ets-1 (p54) binding site Unknown. HSF binding site Unknown. G: p300 binding site; C: ADR1 binding site Unknown. G: ADR1 binding site 67 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.1 (continued) Gene Position Polymorphism Implication 15-LOX-l 5’ regulatory region -33 A>T Unknown. T: HSF binding site Exon 6 759 T>C Synonymous Exon 11 1455 A>G Synonymous Exon 13 1692 T>C Synonymous 3 ’UTR 2053 T>C* 2279 T>C 3385 T>C 3446 T>C Unknown Unknown. CU-rich region, DICE Unknown Unknown 15-LOX-2 5 ’ regulatory -1109 A>G Unknown. G: SRY binding site region -184 T>C -40 A>G Unknown. C: Spl binding site Unknown. A: CAP binding site Exon 2 258 T>C Synonymous Exon 2 345 C>T Synonymous Exon 5 705 T>C Synonymous Exon 7 1440 T>C Synonymous Exon 10 1650 A>C Synonymous Exon 11 1767 O T Synonymous Exon 13 1967 G>A Arg656Gln FLAP 5’ regulatory region -146 ~ -169 PolyA microsatellite Unknown Exon 1 57 T>C Synonymous * Polymorphisms were reported in previous literature or public genomic database. Among the 8 coding region SNPs in the platelet-type 12-LOX gene, two are nonsynonymous polymorphisms, 782G>A (Arg261Gln) and 965A>G (Asn322Ser). These two polymorphisms are in incomplete linkage disequilibrium and result in two common haplotypes, 782G/965A and 782A/965G. In the 5-LOX gene 5’ regulatory region between -176 and -147, an addition/ deletion polymorphism was identified at Spl/Egr-1 binding site (Hoshiko et al., 1990). The "wild-type" allele contains 5 tandem Spl binding motifs (figure 4.1). One or more 6 bp (-GGGCGG-) addition/deletion results in variant alleles containing 3, 4, or 6 Spl binding motifs. Very rare alleles containing 2 or 7 Spl binding motifs have also been reported (Drazen et al., 1999). Another interesting polymorphism is a 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. nonsynonymous SNP in exon 6, which changes lysine to glutamate acid. This polymorphism is only common among African-Americans and is in linkage disequilibrium with the 12 bp deletion (3 motifs) of the 5’ Spl binding site polymorphism. -180 460 .GAGAAGTACT GCGGGGGCGG G G G C G G GGGC GGGGGCGGGG GCGGGGGCAG CCGGAC... (wild) .GAGAAGTACT GCGG..................................... GGGC GGGGGCGGGG GCGGGGGCAG CCGGAC... (12 del) .GAGAAGTACT GCGG GGGCGGGGGC GGGGGCGGGG GCGGGGGCAG CCGGAC... (6 del) .TACTGCGGGG GCGGGGGCGG GGGCGGGGGC GGGGGCGGGG GCGGGGGCAG CCGGAC... (6 add) Spl binding motif -GGGCGG- Egr-1 binding motif -GCGGGGGCG-___________________________________________________________ Figure 4.1 The 5-LOX gene promoter region Spl binding region polymorphism We did not found any nonsynonymous polymorphisms in the 15-LOX-1 gene. However, the 15-LOX-1 gene 3'UTR contains a repetitive CU-rich region, called DICE (differentiation control element), which was found to be polymorphic. This element may play an important role in translational regulation of the 15-LOX-l gene (Ostareck-Lederer et al., 1994). Therefore polymorphisms found in this region may be interesting with regard to gene regulation. One nonsynonymous SNP was found in exon 13 of the 15-LOX-2 gene, changing arginine to glutamine. Whether this polymorphism changes 15-LOX-2 activity is unknown. A microsatellite poly(A) polymorphism (18A>21A) at -146 to -169 of FLAP 5’ regulatory region has been identified and found to be associated with asthma in Japanese (Koshino et al., 1999). We checked the poly(A) microsatellite and found that it is actually a continuous variation, ranging from 16A to 25A, with 19A and 23A as the common length. The only polymorphism identified in the coding region is synonymous. 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.2 Functional Studies of the 5-LOX and FLAP Gene 5’ Regulatory Region Polymorphisms 4.2.1 Luciferase Assay of the 5-LOX Gene Spl Binding Site Polymorphism in Breast Cancer Cells The 5-LOX gene 5’ regulatory region contains an addition/deletion polymorphism in a Spl/Egr-1 binding site (figure 4.1). It has been reported that this polymorphism alters promoter transcriptional activity when transfected into Hela cells (In et al., 1997) and Drosophila SL2 cells (Silverman et al., 1998). The transcriptional activity was linearly related to the number of Spl>Egr-l binding motifs in SL2 cells, whereas in HeLa cells the wild-type (5 motif) allele had the highest activity. This polymorphism also has been found to influence treatment response to certain asthma drugs. Asthma patients who do not carry a "wild-type" allele (5 Spl motifs) do not respond to 5-LOX inhibitors as well as patients carrying a “wild-type” allele (Drazen et al., 1999), suggesting that the diminished activity of variant 5-LOX alleles is clinically significant. To determine whether the allelic variants of the 5-LOX gene differ in their ability to drive reporter gene expression in breast cancer cell lines, pGL3-5LOX (pGL3-basic vectors containing 3, 4, 5 or 6 Spl motif) were transfected into the human breast cancer cell line SK-BR3 and luciferase activity was assayed. Relative luciferase activities for reporter construct carrying each variant were summarized by assay in table 4.2.1. Adjusted activities for each allele showed that allele containing “wild-type” (5 Spl motifs) had the significantly higher transcriptional activity than all 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. other variants (table 4.2.2 & figure 4.2.1). The alleles containing 6- and 12-bp deletions had similar activities, both significantly lower than the activities of the “wild-type” and 6-bp addition (p<0.001). This result is similar to the report with Hela cells (In et al., 1997), but different from the report with Drosophila SL2 cells, in which the promoter activity was proportional to the number of Spl motifs when Spl and Egr-1 were overexpressed (Silverman et al., 1998). This discrepancy may imply that other unidentified transcription factors may be interacting with this polymorphic region of the promoter. Table 4.2.1 Means and standard deviation of relative luciferase activity for the 5 LOX gene 5’ Spl binding site polymorphism in SK-BR3 cells Number of Relative luciferase activity Assay 5-LOX gene allele clone (mean+SD) 1 Wild-type (5 Spl motifs) 4 0.0088±0.0022 12-bp del (3 Spl motifs) 5 0.0071±0.0014 6-bp del (4 Spl motifs) 3 0.0057±0.0009 6-bp add (6 Spl motifs) 3 0.0070+0.0010 2 Wild-type (5 Spl motifs) 3 0.0025±0.0006 12-bp del (3 Spl motifs) 5 0.0021+0.0003 6-bp del (4 Spl motifs) 4 0.0020±0.0003 6-bp add (6 Spl motifs) 3 0.0024+0.0006 3 Wild-type (5 Spl motifs) 4 0.0086±0.0014 12-bp del (3 Spl motifs) 5 0.0060±0.0007 6-bp del (4 Spl motifs) 4 0.0071±0.0012 6-bp add (6 Spl motifs) 3 0.0082±0.0016 4 Wild-type (5 Spl motifs) 4 0.0115±0.0018 12-bp del (3 Spl motifs) 4 0.0079+0.0006 6-bp del (4 Spl motifs) 4 0.0087±0.0010 6-bp add (6 Spl motifs) 3 0.0111+0.0022 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.2.2 Adjusted relative luciferase activity of the 5-LOX gene 5’ Spl binding site polymorphism in SK-BR3 cells Adjusted relative luciferase 5-LOX gene allele______ activity (mean+SE)_______________ P___________ Wild-type (5 Spl motifs) 12-bp del (3 Spl motifs) 6-bp del (4 Spl motifs) 0.0079±0.00018 0.0058±0.00016 0.0059+0.00018 <0.001 (vs. wild-type) 0.56 (vs. 6-bp del) <0.001 (vs. 6-bp add) <0.001 (vs. wild-type) <0.001 (vs. 6-bp add) 6-bp add (6 Spl motifs) 0.0072+0.00020 0.008 (vs. wild-type) P<0.01 & > P<0.001 o < 0 ) < / > 2 a o 3 _l 0.005 - § c S wild-type 12-bp del 6-bp del 6-bp add (5 Sp1 motifs) (3 Sp1 motifs) (4 Sp1 motifs) (6 Sp1 motifs) Figure 4.2.1 Relative luciferase activity of 5-LOX gene 5’ Spl binding site polymorphism in SK-BR3 cells SK-BR3 cells were transfected with pGL3-Basic vector containing one of wide-type (5), 6-bp deletion (4), 12-bp deletion (3) or 6-bp addition (6) of the 5LOX gene promoter region. Cells were co-transfected with phGL3-SV40 control vector, containing Renilla luciferase. 4.2.2 Luciferase Assay of the FLAP Gene 5’ Poly(A) Microsatellite in Breast Cancer Cells The FLAP gene promoter region poly(A) microsatellite has been previously associated with occurrence of asthma, with the long allele (21 A) associating with 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. increased risk of asthma (Koshino et al., 1999). However, the functional significance of this polymorphism was unknown. By sequencing multiple samples, we found that the poly(A) microsatellite actually had a continuous variation ranging from 15A to 26A, with the majority of people carrying alleles of 19A or 23A. We therefore grouped this microsatellite in long (>20A) and short (<20A) alleles. To determine whether these alleles differ in their ability to drive gene expression, we cloned alleles with long or short poly(A) into pGL3-Basic vectors and transfected into SK-BR3 cell line as described for the 5-LOX functional study. We also found a third allele with a short poly(A) and a group of rare (<1%) polymorphisms upstream of the poly(A) (- 251A>G, -268A>Q -360A>T and -439G>T). We also cloned this allele to compare its activity with other common alleles. Relative luciferase activities for reporter constructs carrying each variant were summarized by assay in table 4.2.3. Adjusted relative luciferase activities showed that promoters with short or long alleles did not differ in their activities of transcription (table 4.2.4 & figure 4.2.2). The rare allele with additional SNPs showed a significantly higher transcriptional activity than both long and short alleles without those SNPs (p<0.001). However, we don’t know which of the SNPs is responsible for the increased activity. A very recent publication also reported a luciferase assay for this poly(A) microsatellite and found no significant functional consequence as well (Sayers et al., 2003). 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.2.3 Means and standard deviation of relative luciferase activity for the FLAP gene 5’ poly (A) microsatellite in SK-BR3 cells Assay FLAP gene allele Number of clone Relative luciferase activity (mean+SD) 1 Long allele (>20A) 3 0.00093±0.00017 Short allele (<=20A) 3 0.00086+0.00010 Special allele (with other SNPs) 2 0.00147+0.00024 2 Long allele (>20A) 3 0.00108±0.00032 Short allele (<=20A) 3 0.00100+0.00025 Special allele (with other SNPs) 2 0.00145±0.00023 3 Long allele (>20A) 4 0.00113±0.00035 Short allele (<=20A) 4 0.00101±0.00023 Table 4.2.4 Adjusted relative luciferase activity for FLAP gene 5’ poly(A) microsatellite in SK-BR3 cells Luciferase reporter constructs Adjusted relative luciferase activity (mean+SE) P Long allele (>20A) Short allele (<=20A) Special allele (with other SNPs) 0.0010+0.000039 0.0010+0.000039 0.0015±0.000057 0.49 (vs. long allele) <0.0001 (vs. long allele) <0.0001 (vs. short allele) & > o < 0 > (A 2 a > > a! 0.002 0.0015 - 0.001 0.0005 P<0.001 Long allele (>20A) Short allele (<=20A) Special allele Figure 4.2.2 Relative luciferase activity of the FLAP gene 5’ poly(A) microsatellite in SK-BR3 cell line Luciferase activity of SK-BR3 cells transfected with pGL3-Basic vectors containing long (>20A), short (<=20A), or rare allele of the FLAP gene promoter region. Cells were cotransfected with phGL3-SV40 control vector, containing Renilla luciferase. 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.3 Dietary Fat Intake and Breast Cancer Risk 4.3.1 Characteristics of Study Population Among a total of 1677 breast cancer cases and 2010 controls, 572 cases and 623 controls were non-Hispanic white, 495 cases and 560 controls were African- American, and 610 cases and 827 controls were Latina (table 4.3.1). Approximately 60% of the controls were post-menopausal. Cases and controls were of similar age within each ethnic group and by menopausal status. However, white participants were older than Africans-Americans, and Africans-Americans were older than Latinas (table 4.3.2). The majority of white (92%) and African-American (97%) controls were bom in the US, whereas 66% of Latina controls were foreign-born. Approximately 50% of all controls had education of college level or above. The proportion was highest among whites (73%), intermediate among Affica-Americans (56%), and lowest among Latinas (26%). Foreign-born Latinas were younger than US-bom Latinas, while ages for cases and controls within each group were similar (table 4.3.3). Only 20% of foreign-born Latina controls had education of college or above, compared with 38% among US-bom Latina controls. Overall 76% of breast cancer cases were infiltrating ductal carcinoma. African- Americans and Latinas had similar distribution for types of tumor, stage, histological grade, ER and PR status. Compared to African-Americans and Latinas, whites had slightly less ductal carcinoma, but more lobular carcinoma or with both ductal and lobular carcinoma. Whites also had more cases that were localized, or of low histological grades, or ER/PR positive (table 4.3.4). 75 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 4.3.1 Basic characteristics of study subjects, in all and by menopausal status Total Premenopausal women Postmenopausal women Cases (%) N=1677 Controls (%) N=2010 OR (95%CI)1 Cases (%) N=498 Controls (%) N=596 OR (95%CI)1 Cases (%) N=1059 Controls (%) N=1251 OR (95%CI)1 Age in reference yr. <=47 48-56 57-66 67+ mean (SD) 414(24.7) 458 (27.3) 392 (23.4) 413(24.6) 56.8(11.6) 519(25.8) 527 (26.2) 504(25.1) 460 (22.9) 56.3(11.5) 357(71.7) 138 (27.7) 3 (0.6) 0 44.3 (4.9) 436 (73.2) 153 (25.7) 7(1.2) 0 44.1 (5.1) 25 (2.4) 232(21.9) 389 (36.7) 413 (39.0) 63.6 (8.7) 31 (2.5) 263 (21.0) 497 (39.7) 460 (36.8) 63.1 (8.5) Race/ethnicity White African-American Latina 572(34.1) 495 (29.5) 610 (36.4) 623 (31.0) 560 (27.9) 827(41.1) 142(28.5) 153 (30.7) 203 (40.8) 165 (27.7) 159 (26.7) 272 (45.6) 389 (36.7) 314(29.7) 356 (33.6) 398 (31.8) 365 (29.2) 488 (39.0) Menopausal status Premenopausal Postmenopausal Undetermined 498 (29.7) 1059 (63.2) 120 (7.2) 596 (29.7) 1251 (62.2) 163 (8.1) Foreign born US born Foreign born 1309 (78.1) 368(21.9) 1398 (69.6) 612 (30.5) Ref (1.0) 0.62 (0.52-0.75) 377 (75.7) 121 (24.3) 392 (65.8) 204 (34.2) Ref (1.0) 0.58 (0.42-0.81) 841 (79.4) 218(20.6) 896(71.6) 355 (28.4) Ref (1.0) 0.66 (0.52-0.83) Education (yr) <12 12 13-15 16+ 359(21.4) 347 (20.7) 553 (33.0) 418(24.9) 599 (29.8) 423(21.0) 567 (28.2) 421 (21.0) Ref (1.0) 1.39(1.14-1.70) 1.67(1.38-2.03) 1.74(1.41-2.15) 68(13.7) 94(18.9) 177 (35.5) 159 (31.9) 149 (25.0) 117(19.6) 171 (28.7) 159 (26.7) Ref (1.0) 1.82(1.21-2.74) 2.35 (1.60-3.46) 2.38(1.60-3.54) 278 (26.3) 229 (21.6) 332 (31.4) 220 (20.8) 409 (32.7) 279 (22.3) 345 (27.6) 218(17.4) Ref (1.0) 1.17(0.92-1.49) 1.37(1.08-1.73) 1.42 (1.09-1.86) Family history of breast Cancer No Yes 1419(84.6) 258 (15.4) 1785 (88.8) 225 (11.2) Ref (1.0) 1.40(1.16-1.70) 428 (85.9) 70(14.1) 546 (91.6) 50 (8.4) Ref (1.0) 1.76(1.20-2.60) 885 (83.6) 174(16.4) 1092(87.3) 159(12.7) Ref (1.0) 1.31 (1.03-1.65) as Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.1 (continued) Total Premenopausal women Postmenopausal women Cases (%) N=1677 Controls (%) N=2010 OR (95%CI)1 Cases (%) N=498 Controls (%) N=596 OR (95%CI)1 Cases (%) N=1059 Controls (%) N=1251 OR (95%CI)1 Benign breast disease No Yes 1323 (78.9) 354 (21.1) 1696(84.4) 314(15.6) Ref (1.0) 1.41 (1.19-1.67) 417(83.7) 81 (16.3) 546(91.6) 50 (8.4) Ref (1.0) 2.09(1.43-3.05) 814(76.9) 245 (23.1) 1005(80.3) 246(19.7) Ref (1.0) 1.19(0.98-1.46) Age at menarche 8-11 12-13 13 up 389(23.2) 870 (51.9) 418(24.9) 422(21.0) Ref (1.0) 1021 (50.8) 0.91 (0.77-1.07) 567 (28.2) 0.80 (0.66-0.96) 127(25.5) 247 (49.6) 124 (24.9) 121 (20.3) 337 (56.5) 138(23.2) Ref (1.0) 0.69 (0.51-0.94) 0.87 (0.61-1.23) 231 (21.8) 554 (52.3) 274 (25.9) 253 (20.2) 607 (48.5) 391 (31.3) Ref (1.0) 0.97 (0.78-1.20) 0.76 (0.60-0.97) Age at first full term pregnancy <20 20-24 25-29 30+ nuliiparous 339 (20.2) 540 (32.2) 302(18.0) 221 (13.2) 275 (16.4) 499 (24.8) 667 (33.2) 361 (18.0) 241 (12.0) 242(12.0) Ref (1.0) 1.20(1.00-1.44) 1.24(1.00-1.53) 1.37(1.09-1.74) 1.66(1.32-2.08) 80(16.1) 143 (28.7) 91 (18.3) 79(15.9) 105(21.1) 138(23.2) 169(28.4) 100(16.8) 100(16.8) 89(14.9) Ref (1.0) 1.50(1.05-2.15) 1.64(1.10-2.45) 1.46 (0.96-2.23) 2.14(1.42-3.24) 236 (22.3) 358 (33.8) 191 (18.0) 123 (11.6) 151 (14.3) 324 (25.9) 448 (35.8) 232(18.6) 121 (9.7) 126(10.1) Ref (1.0) 1.08 (0.86-1.35) 1.10(0.84-1.43) 1.37 (1.00-1.86) 1.56(1.16-2.10) Height (quartiie, cm) Q1 (<=154.9) Q2 (154.9-160.0) Q3 (160.0-165.0) Q4 (165.0+) 332(19.8) 390 (23.3) 471 (28.1) 484 (28.9) 508 (25.3) 522 (26.0) 479 (23.8) 501 (24.9) Ref (1.0) 1.15 (0.94-1.39) 1.51 (1.24-1.85) 1.49(1.21-1.84) 69 (13.9) 101 (20.3) 155 (31.1) 173 (34.7) 123 (20.6) 158 (26.5) 142(23.8) 173 (29.0) Ref (1.0) 1.15 (0.78-1.70) 1.96(1.33-2.90) 1.80(1.21-2.68) 254 (24.0) 255 (24.1) 278 (26.3) 272 (25.7) 353 (28.2) 316(25.3) 300 (24.0) 282 (22.5) Ref (1.0) 1.09 (0.86-1.38) 1.24(0.97-1.58) 1.28 (0.99-1.67) BMI <25 25-30 30+ 488(29.1) 526 (31.4) 663 (39.5) 504(25.1) 681 (33.9) 825 (41.0) Ref (1.0) 0.82 (0.69-0.97) 0.84 (0.71-1.00) 167 (33.5) 147 (29.5) 184(37.0) 162 (27.2) 194 (32.6) 240 (40.3) Ref (1.0) 0.72 (0.53-0.99) 0.70 (0.52-0.95) 275 (26.0) 347 (32.8) 437(41.3) 302 (24.1) 435 (34.8) 514(41.1) Ref (1.0) 0.93(0.74-1.16) 1.00 (0.80-1.23) Total physical activity (quartiie, hrs/week) Q1 (<=6.7) Q2 (6.7-13.8) Q3 (13.8-24.8) Q4 (24.8+) 469 (28.0) 467 (27.9) 367 (21.9) 374 (22.3) 495 (24.6) 510(25.4) 503 (25.0) 502 (25.0) Ref (1.0) 0.97 (0.81-1.16) 0.78 (0.65-0.94) 0.81 (0.67-0.97) 132(26.5) 154(30.9) 110(22.1) 102 (20.5) 153 (25.7) 141 (23.7) 162(27.2) 140 (23.5) Ref (1.0) 1.28 (0.92-1.78) 0.80 (0.57-1.12) 0.87 (0.61-1.23) 295 (27.9) 280 (26.4) 236 (22.3) 248 (23.4) 300 (24.0) 328 (26.2) 310(24.8) 313(25.0) Ref (1.0) 0.86(0.69-1.08) 0.78 (0.62-0.98) 0.83 (0.66-1.05) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.1 (continued) Total Premenopausal women Postmenopausal women Cases (%) N=1677 Controls (%) N=2010 OR (95%CI)1 Cases (%) N=498 Controls (%) N=596 OR (95%CI)1 Cases (%) N=1059 Controls (%) N=1251 OR (95%CI)1 Total energy intake (quartiie, kcal/d) Q1 (<=1462) Q2 (1462-1970) Q3 (1970-2711) Q4 (2711) 406 (24.2) 442 (26.4) 432 (25.8) 397 (23.7) 502 (25.0) 504(25.1) 501 (24.9) 503 (25.0) Ref (1.0) 1.11 (0.92-1.34) 1.11 (0.93-1.34) 1.05 (0.87-1.27) 95(19.1) 118(23.7) 106 (21.3) 179 (35.9) 106(17.8) 137 (23.0) 167 (28.0) 186(31.2) Ref (1.0) 0.99 (0.68-1.44) 0.73 (0.50-1.06) 1.13(0.79-1.60) 292 (27.6) 289 (27.3) 289 (27.3) 189(17.9) 369 (29.5) 325 (26.0) 281 (22.5) 276 (22.1) Ref (1.0) 1.14(0.92-1.43) 1.36(1.08-1.71) 0.94 (0.73-1.20) Alcohol intake (fertile, gms/d) T1 (0) T2 (0-5) T3 (5+) 851 (50.8) 409 (24.4) 417(24.9) 1144(56.9) 448 (22.3) 410(20.8) Ref (1.0) 1.24(1.06-1.47) 1.33 (1.12-1.57) 213 (42.8) 151 (30.3) 134 (26.9) 313(52.5) 146 (24.5) 137(23.0) Ref (1.0) 1.52(1.14-2.04) 1.43(1.05-1.94) 594(56.1) 216 (20.4) 249 (23.5) 751 (60.0) 259 (20.7) 241 (19.3) Ref (1.0) 1.04(0.84-1.29) 1.25 (1.00-1.54) 1. Adjusted for age and ethnicity. oo Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.2 Basic characteristics of study subjects, by ethnicity White African-American Latina Cases (%) N=572 Controls (%) N=623 OR (95%CI)1 Cases (%) N=495 Controls (%) N=560 OR (95%CI)1 Cases (%) N=610 Controls (%) N=827 OR (95%CI)1 Age in reference yr. <=47 48-56 57-66 674- mean (SD) 106(18.5) 158 (27.6) 122(21.3) 186 (32.5) 59.0(12.0) 140 (22.5) 156(25.0) 145 (23.3) 182(29.2) 58.1(12.0) 126 (25.5) 138 (27.9) 114 (23.0) 117(23.6) 56.6(11.7) 130 (23.2) 152(27.1) 147 (26.3) 131 (23.4) 56.4(11.4) 182 (29.8) 162(26.6) 156 (25.6) 110(18.0) 55.1 (10.9) 249 (30.1) 219 (26.5) 212 (25.6) 147(17.8) 54.7(10.9) Menopausal status Premenopausal Postmenopausal Undetermined 142 (24.8) 389 (68.0) 41 (7.2) 165 (26.5) 398 (63.9) 60 (9.6) 153 (63.4) 314(63.4) 28 (5.7) 159 (28.4) 365 (65.2) 36 (6.4) 203 (33.3) 356 (58.4) 51 (8.4) 272 (32.9) 488 (59.0) 67 (8.1) Foreign born US born Foreign born 518(90.6) 54 (9.4) 571 (91.7) 52 (8.4) Ref (1.0) 1.15(0.77-1.71) 484 (97.8) 11 (2.2) 544(97.1) 16 (2.9) Ref (1.0) 0.77 (0.35-1.68) 307 (50.3) 303 (49.7) 283 (34.2) 544 (65.8) Ref (1.0) 0.51 (0.42-0.64) Education (yr) <12 12 13-15 164- 33 (5.8) 117(20.5) 196 (34.3) 226 (39.5) 39 (6.3) 129 (20.7) 209 (33.6) 246 (39.5) Ref (1.0) 1.10(0.65-1.87) 1.15(0.69-1.91) 1.16(0.70-1.92) 90(18.2) 105(21.2) 206(41.6) 94(19.0) 110(19.6) 135 (24.1) 220 (39.3) 95(17.0) Ref (1.0) 0.96(0.65-1.41) 1.16(0.81-1.66) 1.23 (0.81-1.86) 236 (38.7) 125 (20.5) 151 (24.8) 98(16.1) 450 (54.4) 159(19.2) 138(16.7) 80 (9.7) Ref (1.0) 1.53 (1.15-2.03) 2.17(1.64-2.89) 2.49(1.77-3.50) Family history of breast Cancer No Yes 459 (80.2) 113(19.8) 531 (85.2) 92(14.8) Ref (1.0) 1.41 (1.05-1.91) 428 (86.5) 67(13.5) 489 (87.3) 71 (12.7) Ref (1.0) 1.08(0.76-1.55) 532 (87.2) 78 (12.8) 765 (92.5) 62 (7.5) Ref (1.0) 1.80(1.26-2.56) Benign breast disease No Yes 418(73.1) 154 (26.9) 505 (81.1) 118(18.9) Ref (1.0) 1.55 (1.18-2.04) 396 (80.0) 99 (20.0) 466 (83.2) 94(16.8) Ref (1.0) 1.24(0.91-1.70) 509 (83.4) 101 (16.6) 725 (87.7) 102(12.3) Ref (1.0) 1.40(1.04-1.89) V O Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.2 (continued) White African-American Latina Cases (%) N=572 Controls (%) N=623 OR (95%CI)‘ Cases (%) N=495 Controls (%) N=560 OR (95%CI)1 Cases (%) N=610 Controls (%) N=827 OR (95%CI)1 Age at menarche 8-11 12-13 13 up 117(20.5) 330 (57.7) 125 (21.9) 131 (21.0) 351 (56.3) 141 (22.6) Ref (1.0) 1.05 (0.78-1.40) 0.98(0.69-1.39) 105 (21.2) 260 (52.5) 130(26.3) 117(20.9) 293 (52.3) 150 (26.8) Ref (1.0) 0.99(0.72-1.35) 0.97 (0.68-1.38) 167(27.4) 280 (45.9) 163 (26.7) 174(21.0) 377 (45.6) 276 (33.4) Ref (1.0) 0.77 (0.59-1.01) 0.61 (0.46-0.81) Age at first full term pregnancy <20 20-24 25-29 30+ nulliparous 57(10.0) 179(31.3) 135 (23.6) 89(15.6) 112(19.6) 58 (9.3) 190 (30.5) 142 (22.8) 106(17.0) 127(20.4) Ref (1.0) 0.96 (0.63-1.45) 0.96 (0.62-1.49) 0.88 (0.55-1.40) 0.92 (0.59-1.44) 165 (33.3) 148 (29.9) 52(10.5) 47 (9.5) 83 (16.8) 207 (37.0) 188(33.6) 61 (10.9) 36 (6.4) 68(12.1) Ref (1.0) 0.99 (0.73-1.33) 1.07 (0.70-1.63) 1.64(1.01-2.65) 1.53 (1.05-2.24) 117(19.2) 213 (34.9) 115(18.9) 85 (13.9) 80(13.1) 234 (28.3) 289 (35.0) 158(19.1) 99 (12.0) 47 (5.7) Ref (1.0) 1.47 (1.11-1.95) 1.44(1.04-2.00) 1.71 (1.19-2.47) 3.43 (2.25-5.24) Height (quartiie, cm) Q1 (<= 154.9) Q2 (154.9-160.0) Q3 (160.0-165.0) Q4 (165.0+) 67(11.7) 131 (22.9) 168 (29.4) 206 (36.0) 83 (13.3) 142 (22.8) 181 (29.1) 217(34.8) Ref (1.0) 1.19(0.79-1.78) 1.21 (0.82-1.79) 1.27 (0.86-1.87) 38 (7.7) 84(17.0) 161 (32.5) 212(42.8) 56 (10.0) 121 (21.6) 168 (30.0) 215 (38.4) Ref (1.0) 1.03 (0.62-1.69) 1.42 (0.89-2.26) 1.46(0.93-2.31) 227 (37.2) 175 (28.7) 142 (23.3) 66 (10.8) 369 (44.6) 259 (31.3) 130(15.7) 69 (8.3) Ref (1.0) 1.13 (0.87-1.45) 1.85 (1.38-2.49) 1.64(1.12-2.40) BMI <25 25-30 30+ 251 (43.9) 156 (27.3) 165 (28.9) 256(41.1) 188 (30.2) 179(28.7) Ref (1.0) 89(18.0) 0.83 (0.63-1.10) 155 (31.3) 0.94 (0.71-1.23) 251 (50.7) 110(19.6) 161 (28.8) 289 (51.6) Ref (1.0) 1.19(0.84-1.70) 1.08(0.78-1.49) 148 (24.3) 215 (35.3) 247 (40.5) 138(16.7) 332 (40.2) 357 (43.2) Ref (1.0) 0.60 (0.45-0.80) 0.64 (0.48-0.85) Total physical activity (quartiie, hrs/week) Q1 (<=6.7) Q2 (6.7-13.8) Q3 (13.8-24.8) Q4 (24.8+) 178 (31.1) 152(26.6) 135 (23.6) 107(18.7) 169(27.1) 180 (28.9) 153 (24.6) 121 (19.4) Ref (1.0) 0.79 (0.58-1.06) 0.83(0.61-1.14) 0.84(0.60-1.17) 143 (28.9) 149(30.1) 97(19.6) 106(21.4) 152(27.1) 147 (26.3) 140 (25.0) 121 (21.6) Ref (1.0) 1.08(0.78-1.49) 0.74(0.52-1.04) 0.93 (0.66-1.32) 148 (24.3) 166 (27.2) 135 (22.1) 161 (26.4) 174 (21.0) 183 (22.1) 210(25.4) 260(31.4) Ref (1.0) 1.07(0.79-1.45) 0.76 (0.56-1.03) 0.73 (0.54-0.98) 00 o Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.2 (continued) White African-American Latina Cases (%) N=572 Controls (%) N=623 OR (95%CI)1 Cases (%) N=495 Controls (%) N=560 OR (95%CI)‘ Cases (%) N=610 Controls (%) N=827 OR (95%CI)1 Total energy intake (quartiie, kcal/d) Q1 (<=1462) Q2 (1462-1970) Q3 (1970-2711) Q4 (2711) 166 (29.0) 178(31.1) 160(28.0) 68(11.9) 166(26.7) 197 (31.6) 153(24.6) 107 (17.2) Ref (1.0) 0.92(0.68-1.23) 1.06(0.78-1.45) 0.66 (0.45-0.96) 139(28.1) 115(23.2) 104(21.0) 137 (27.7) 193 (34.5) 119(21.3) 117(20.9) 131 (23.4) Ref (1.0) 1.35 (0.96-1.89) 1.25 (0.88-1.76) 1.47(1.06-2.06) 101 (16.6) 149 (24.5) 168 (27.5) 192 (31.5) 143 (17.3) 188(22.7) 231 (17.9) 265 (32.0) Ref (1.0) 1.13 (0.81-1.58) 1.04(0.75-1.44) 1.05 (0.76-1.44) Alcohol intake (tertile, gms/d) T1 (0) T2 (0-5) T3 (5+) 194(33.9) 164 (28.7) 214(37.4) 250(40.1) 161 (25.8) 212 (34.0) Ref (1.0) 1.34(1.01-1.79) 1.33(1.02-1.74) 308 (62.2) 89(18.0) 98(19.8) 360 (64.3) 103(18.4) 97(17.3) Ref (1.0) 1.01 (0.73-1.40) 1.18(0.86-1.63) 349 (57.2) 156 (25.6) 105 (17.2) 534 (64.6) 184(22.3) 109(13.2) Ref (1.0) 1.31 (1.02-1.69) 1.49(1.10-2.01) 1. Adjusted for age. 00 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.3 Basic characteristics of study population among US born and foreign born Latinas US-born Latina Foreign-born Latina Cases (%) N=307 Controls (%) N=283 OR (95%CI)1 Cases (%) N=303 Controls (%) N=544 OR (95%CI)1 Age in reference yr. <=47 48-56 57-66 67+ mean (SD) 86 (28.0) 76 (24.8) 86 (28.0) 59 (19.2) 55.8(11.0) 81 (28.6) 68 (24.0) 62(21.9) 72 (25.4) 56.1 (11.6) 96 (31.7) 86 (28.4) 70 (23.1) 51 (16.8) 54.3 (10.8) 168 (30.9) 151 (27.8) 150(27.6) 75 (13.8) 53.9(10.5) Menopausal status Premenopausal Postmenopausal Undetermined 97 (31.6) 186 (60.6) 24 (7.8) 86 (30.4) 176 (62.2) 21 (7.4) 106(35.0) 170 (56.1) 27 (8.9) 186 (34.2) 312(57.4) 46 (8.5) Education (yr) <12 12 13-15 16+ 89 (29.0) 88 (28.7) 86 (28.0) 44(14.3) 89(31.5) 86 (30.4) 73 (25.8) 35 (12.4) Ref (1.0) 1.02(0.67-1.57) 1.18 (0.75-1.85) 1.26 (0.72-2.20) 147 (48.5) 37(12.2) 65 (21.5) 54(17.8) 361 (66.4) 73 (13.4) 65 (12.0) 45 (8.3) Ref (1.0) 1.25 (0.80-1.94) 2.49(1.68-3.71) 3.07(1.97-4.78) Family history of breast Cancer No Yes 262 (85.3) 45 (14.7) 250 (88.3) 33(11.7) Ref (1.0) 1.32 (0.81-2.14) 270 (89.1) 33 (10.9) 515(94.7) 29 (5.3) Ref (1.0) 2.16(1.28-3.63) Benign breast disease No Yes 246 (80.1) 61 (19.9) 234 (82.7) 49(17.3) Ref (1.0) 1.19(0.79-1.81) 263 (86.8) 40(13.2) 491 (90.3) 53 (9.7) Ref (1.0) 1.40 (0.90-2.17) Age at menarche 8-11 12-13 13 up 97 (31.6) 140(45.6) 70 (22.8) 80 (28.3) 140 (49.5) 63 (22.3) Ref (1.0) 0.83 (0.57-1.21) 0.92(0.59-1.45) 70 (23.1) 140 (46.2) 93 (30.7) 94(17.3) 237 (43.6) 213 (39.2) Ref (1.0) 0.80 (0.55-1.16) 0.58(0.39-1.86) ’■ Adjusted for age o o to Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.3 (continued) US-born Latina Foreign-born Latina Cases (%) N=307 Controls (%) N=283 OR (95%CI)1 Cases (%) N=303 Controls (%) N=544 OR (95%CI)1 Age at first full term pregnancy 20 68 (22.2) 84 (29.7) Ref (1.0) 49 (16.2) 150 (27.6) Ref (1.0) 20-24 110(35.8) 106 (37.5) 1.28 (0.85-1.95) 103 (34.0) 183 (33.6) 1.71 (1.15-2.57) 25-29 57(18.6) 45 (15.9) 1.57(0.95-2.60) 58(19.1) 113(20.8) 1.56 (0.99-2.45) 30+ 35(11.4) 27 (9.5) 1.60 (0.88-2.90) 50(16.5) 72 (13.2) 2.11 (1.30-3.42) nulliparous 37(12.1) 21 (7.4) 2.17(1.16-4.05) 43 (14.2) 26 (4.8) 5.08 (2.83-9.11) Height (quartiie, cm) Q1 (<=154.9) 106 (34.5) 92 (32.5) Ref (1.0) 121 (39.9) 277 (50.9) Ref (1.0) Q2 (154.9-160.0) 78 (25.4) 107(37.8) 0.63 (0.42-0.95) 97 (32.0) 152 (27.9) 1.50(1.07-2.10) Q3 (160.0-165.0) 83 (27.0) 58 (19.8) 1.29 (0.82-2.02) 59(19.5) 74(13.6) 1.91 (1.27-2.88) Q4 (165.0+) 40(13.0) 28 (9.9) 1.24(0.70-2.21) 26 (8.6) 41 (7.5) 1.50 (0.88-2.58) BMI <25 78 (25.4) 58 (20.5) Ref (1.0) 70 (23.1) 80 (14.7) Ref (1.0) 25-30 107(34.9) 90 (31.8) 0.89(0.57-1.38) 108 (35.6) 242 (44.5) 0.51 (0.34-0.75) 30+ 122 (39.7) 135 (47.7) 0.67(0.44-1.03) 125 (41.3) 222 (40.8) 0.64 (0.43-0.94) Total physical activity (quartiie, hrs/week) Q1 (<=6.7) 61 (19.9) 57(20.1) Ref (1.0) 87 (28.7) 117(21.5) Ref (1.0) Q2 (6.7-13.8) 86 (28.0) 83 (29.3) 0.97 (0.61-1.55) 80 (26.4) 100(18.4) 1.08 (0.72-1.62) Q3 (13.8-24.8) 77 (25.1) 61 (21.6) 1.18 (0.72-1.94) 58(19.1) 149 (27.4) 0.53 (0.35-0.80) Q4 (24.8+) 83 (27.0) 82 (29.0) 0.95 (0.59-1.52) 78 (25.7) 178 (32.7) 0.59 (0.40-0.87) Total energy intake (quartiie, kcai/d) Q1 (<=1462) 55(17.9) 54(19.1) Ref (1.0) 46 (15.2) 89(16.4) Ref (1.0) Q2 (1462-1970) 88 (28.7) 74 (26.2) 1.16(0.72-1.89) 61 (20.1) 114(21.0) 1.05 (0.65-1.68) Q3 (1970-2711) 74(24.1) 75 (26.5) 0.96(0.59-1.58) 94 (31.0) 156 (28.7) 1.18 (0.76-1.83) Q4 (2711) 90 (29.3) 80 (28.3) 1.09(0.67-1.78) 102 (33.7) 185 (34.0) 1.09 (0.71-1.69) Alcohol intake (fertile, gm/d) T1 (0) 161 (52.4) 161 (56.9) Ref (1.0) 188 (62.1) 373 (68.6) Ref (1.0) T2 (0-5) 84 (27.4) 67(23.7) 1.25 (0.84-1.85) 72 (23.8) 117(21.5) 1.23 (0.87-1.74) T3 (5+) 62 (20.2) 55(19.4) 1.12(0.73-1.72) 43 (14.2) 54 (9.9) 1.58 (1.02-2.45) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.4 Pathological characteristics of breast cancer, by ethnicity Pathological characteristics All cases (%) White (%) African- American (%) Latina (%) US-born Latina (%) Foreign-born Latina (%) Histological ductal 1276 (76) 425 (74) 379 (77) 472 (77) 233 (76) 239(79) type lobular 135 (8) 59 (10) 34(7) 42(7) 29 (9) 13(4) ductal & lobular 89(5) 37 (7) 20(4) 32(5) 19(6) 13(4) other 177(11) 51(9) 62 (13) 64(11) 26 (9) 38(13) Stage localized 1111(66) 405 (71) 312(63) 394 (65) 197 (64) 197 (65) advanced 543 (33) 162 (28) 171(35) 210 (34) 109 (36) 101 (33) undetermined 24(1) 6(1) 12(2) 6(1) 1 (0.3) 5(2) Histological 1 264(19) 124 (22) 66 (13) 74 (12) 41 (13) 33 (11) grade 2 581 (41) 218 (38) 149 (30) 214(35) 105 (34) 109 (36) 3 519 (37) 131(23) 182 (37) 206 (34) 100 (33) 106 (35) 4 53(4) 16(3) 11(2) 26(4) 12(4) 14(5) undetermined 260(16) 83 (15) 87(18) 90(15) 49 (16) 41 (14) ER status positive 1060 (63) 411(72) 272 (55) 377 (72) 189(62) 188(62) negative 363 (22) 83 (15) 136 (27) 144 (28) 77 (25) 67 (22) undetermined 254(15) 78 (13) 87(18) 89 (10) 41 (13) 48 (16) PR status positive 904 (53) 358(63) 229 (46) 317 (52) 160 (52) 157(52) negative 495 (30) 130(22) 174 (35) 191 (31) 99 (32) 92 (30) undetermined 278 (17) 84 (15) 92 (19) 102 (17) 48 (16) 54 (18) 00 4^ The median total energy intake among all controls was 1970 kcal/day (table 4.3.5). The median of total energy from fat was 31%. Foreign-born Latinas had the highest total energy intake (median: 2430 kcal/day), but the lowest percentage of total energy from fat (26%). There were correlations among types of fat intake. Total fat intake was highly correlated with both oleic acid intake (r=0.93) and saturated fat intake (r=0.80) (table 4.3.6). Oleic acid intake was also highly correlated with saturated fat (r=0.75). The correlations were similar within each ethnicity. 4.3.2 Dietary Fat Intake and Breast Cancer: Residual Method Overall, total fat intake was significantly associated with increased risk of breast cancer when adjusted for age, ethnicity and total energy intake using the residual method (table 4.3.7). After further adjustment for other potential confounders, the association still held. Compared to the lowest quartiie, the odds ratio for the highest quartiie of residual was 1.41 with 95% confidence interval from 1.14 to 1.73. There was also a significant trend of increased risk across the quartiles (p trend=0.002). Analysis by types of fatty acid intake showed that both linoleic acid and oleic acid were associated with increased risk of breast cancer, while saturated fat was not significantly related. The strongest association was found between highest intake oleic acid and breast cancer risk (OR: 1.45; 95% Cl: 1.17-1.78). 85 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 4.3.5 Comparison of the daily intake of total energy and percentage of energy from dietary components All Controls Whites African- Americans Latinas US-born Latinas Foreign-born Latinas Total fat, % energy Mean (+ SD) 30.9 (± 7.5) 32.2 (+ 7.4) 33.8 (±7.7) 28.0 (±6.5) 31.7 (±5.9) 26.1 (+5.9) Median (25-75%) 30.6 (25.8-35.8) 31.6(27.2-37.3) 33.6 (28.8-38.8) 28.0 (23.3-32.3) 31.7(28.0-35.2) 25.9 (21.8-30.0) Saturated fat, % energy Mean (± SD) 10.7 (+3.1) 11.4 (+3.4) 11.0 (±3.0) 9.9 (±2.8) 10.9 (±2.5) 9.4 (± 2.8) Median (25-75%) 10.4 (8.5-12.5) 11.0(8.9-13.2) 11.0(9.1-12.9) 9.8 (8.0-11.6) 10.8 (9.3-12.6) 9.0(7.5-11.0) Linoleic, % energy Mean (± SD) 5.4 (+1.9) 5.7 (+2.1) 5.9 (± 1.9) 4.7 (±1.4) 5.2 (±1.6) 4.5 (±1.2) Median 5.1 (4.1-6.3) 5.4 (4.3-6.6) 5.7 (4.6-7.0) 4.5 (3.8-5.4) 4.9 (4.1-6.0) 4.3 (3.7-5.2) Oleic, % energy Mean (+ SD) 11.0 (+2.9) 11.4 (+2.6) 12.1 (±3.0) 9.9 (+2.7) 11.6 (±2.4) 9.0 (±2.3) Median (25-75%) 10.9 (9.0-12.8) 11.3 (9.5-13.3) 12.2(10.2-14.0) 9.9 (8.1-11.8) 11.6(10.2-13.2) 8.9 (7.5-10.5) Protein, % energy Mean (± SD) 17.5 (+3.7) 17.0 (+3.5) 17.9 (±4.3) 17.7 (±3.2) 17.4 (±3.3) 17.9 (±3.2) Median (25-75%) 17.3 (15.0-19.6) 16.8(14.5-19.2) 17.7(15.0-20.4) 17.4(15.5-19.5) 17.3 (15.2-19.3) 17.4(15.8-19.7) Carbohydrate, % energy Mean (± SD) 50.5 (+ 9.0) 49.2 (+ 8.3) 47.4 (± 9.9) 53.6 (± 7.8) 50.2 (± 7.3) 55.4 (± 7.4) Median (25-75%) 50.7 (44.2-56.7) 49.0 (43.4-54.3) 47.2 (40.5-53.5) 53.9 (48.4-58.9) 50.0 (45.0-54.8) 55.9 (50.8-60.3) Total energy, kcal/d Mean (± SD) 2149 (±923) 1990 (±804) 2024 (± 986) 2354 (±926) 2208 (± 843) 2430 (± 959) Median (25-75%) 1970(1462-2711) 1816 (1433-2436) 1789 (1279-2603) 2228 (1642-2926) 2081 (1522-2770) 2290(1703-3005) GO O n Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.6 Correlations between dietary fat, as a percentage of total energy intake Spearman correlation coefficient All Whites African- Latinas US-born Americans Latinas Foreign-born Latinas Total energy vs. total fat 0.14 0.26 0.25 0.12 0.19 0.16 vs. saturated fat 0.17 0.26 0.24 0.15 0.20 0.17 vs. linoleic 0.03 0.11 0.12 0.02 0.04 0.04 vs. oleic 0.17 0.29 0.26 0.13 0.23 0.18 Total fat vs. saturated fat 0.80 0.77 0.82 0.81 0.76 0.81 vs. Linoleic 0.61 0.56 0.56 0.57 0.52 0.54 vs. Oleic 0.93 0.91 0.94 0.93 0.90 0.92 Saturated fat vs. Linoleic 0.19 0.09 0.16 0.18 0.09 0.14 vs. Oleic 0.75 0.71 0.76 0.76 0.72 0.75 Linoleic vs. Oleic 0.52 0.46 0.47 0.49 0.35 0.48 00 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.7 Dietary fat intake and breast cancer risk according to quartiles of fat residual, among all cases and controls Dietary fat U 1 U V U & M i l V M t f V t f Adj. OR* (95% Cl) P trend Adj. OR** (95% Cl) P trend Total fat Qi Ref (1.0) Ref (1.0) Q2 1.23 (1.01-1.50) 1.17 (0.95-1.43) Q3 1.29 (1.06-1.57) 1.18(0.96-1.45) Q4 1.53 (1.26-1.86) <0.0001 1.41 (1.14-1.73) 0.002 Saturated fat Ql Ref (1.0) Ref (1.0) Q2 1.23 (1.02-1.50) 1.15 (0.95-1.41) Q3 1.23 (1.02-1.50) 1.16(0.95-1.41) Q4 1.19(0.98-1.44) 0.11 1.09(0.89-1.33) 0.48 Linoleic acid Ql Ref (1.0) Ref (1.0) Q2 1.38(1.13-1.68) 1.34 (1.10-1.64) Q3 1.40 (1.15-1.72) 1.37(1.12-1.69) Q4 1.43(1.18-1.74) 0.002 1.32(1.08-1.62) 0.03 Oleic acid Ql Ref (1.0) Ref (1.0) Q2 1.24(1.02-1.52) 1.16(0.94-1.42) Q3 1.36 (1.12-1.66) 1.25 (1.02-1.55) Q4 1.60(1.32-1.94) <0.0001 1.45(1.17-1.78) 0.0004 For all analyses in this table and following tables, * Adjusted for age (continuous), ethnicity and total energy intake (continuous). ** Besides age, ethnicity and total energy intake, further adjusted for menopausal status (pre-, postmenopausal, undetermined), age at first full-term pregnancy (<20, 20-24,25-29, 30+), bmi (continuous), height (continuous), family history of breast cancer (yes/no), age at menarche (continuous), foreign bom(yes/no), education (<12y, 12y, 13-15y, 16y and up), history of benign breast disease (yes/no), lifetime physical activity (continuous), alcohol intake (continuous). oo oo To examine whether the association between dietary fat and breast cancer is consistent across ethnicities, we conducted the same analysis among whites, African- Americans and Latinas separately. After adjustment neither total fat nor types of fatty acids were associated with breast cancer among African-Americans (table 4.3.8). Among whites and Latinas, oleic acid intake was strongly associated with increased breast cancer risk. The OR for highest versus lowest quartile of oleic acid residual was 1.49 (95% Cl: 1.00-2.21) for whites, and 1.78 (95% Cl: 1.25-2.54) for Latinas. The positive association among Latinas was mostly attributable to foreign-born rather than US-born Latinas. Linoleic acid intake also showed a positive association with breast cancer risk among whites and Latinas, but did not achieve a significant trend among white, probably due to the reduced sample size. Although the associations between total fat or types of fat with breast cancer were less consistent across ethnicity, none of the associations significantly differ by ethnicity (interaction test p>0.05). 4.3.3 Dietary Fat Intake and Breast Cancer: Multivariate Nutrient Density Model Using the multivariate nutrient density model to adjust for energy intake, results were very similar to the results from residual model (table 4.3.9). Analysis of total fat intake in multivariate analysis, where potential confounders and total energy intake were adjusted simultaneously, showed that total fat intake was significantly associated with increased risk of breast cancer (OR Q 4 vsqi= 1-40, 95% Cl: 1.14-1.73, p trend=0.0006). The effect of total fat was attributable to oleic acid (OR Q 4 Vs qi= 1.49, 89 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95% Cl: 1.21-1.84, p trend=0.0002) and linoleic acid (OR q4 v Sqi= 1.17, 95% Cl: 0.96- 1.43, p trend=0.1), but not saturated fat (OR Q 4 v s qi= 1-12, 95% Cl: 0.92-1.37, p trend=0.3). Again, when we stratified the analysis by ethnicity, we saw less consistent associations across the ethnicities (table 4.3.10 and 4.3.11). Compared with the lowest quartile, women in the highest quartile of total fat intake had increased risk of breast cancer among whites (OR: 1.47; 95% Cl: 1.02-2.12) and Latinas (OR: 1.57; 95% Cl: 1.08-2.29), but not among African-Americans. Oleic acid still had the strongest and dose-response positive association with breast cancer among whites (p trend=0.006) and Latinas (p trend=0.001). Linoleic acid also appeared to be positively associated with breast cancer among whites and Latinas, but they were weaker than the association with oleic acid. None of the types of fat were associated with breast cancer risk among African-Americans. However, the difference of the associations by ethnicity were not statistically significant (interaction test p>0.05). We further analyzed the effect of types of fats by mutually adjusting these three fatty acids (saturated, linoleic, and oleic) simultaneously in the multivariate nutrient density model. The results showed that similar effects still existed for oleic acid (p trend=0.0008) (table 4.3.12). The weak positive association between linoleic acid and breast cancer risk disappeared after mutual adjustment (p trend=0.7). Saturated fat even showed a non-significant protective trend (p trend=0.08). The changes of the estimates for linoleic acid and saturated fat before and after mutual adjustment may be due to the strong correlations among types of fat. 90 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 4.3.8 Dietary fat intake and breast cancer risk according to quartiles of fat residual, by ethnicity Whites African-Americans Latinas US-born Latinas Foreign-born-Latinas Dietary fat Adj. OR** (95% CD P trend Adj. OR** (95% CD P trend Adj. OR** (95% CD P trend Adj. OR** (95% CD P trend Adj. OR** (95% CD P trend Total fat Ql Q2 Q3 Q4 Ref (1.0) 1.07 (0.73-1.58) 1.19(0.81-1.75) 1.35 (0.92-1.97) 0.08 Ref (1.0) 0.95 (0.59-1.53) 0.84 (0.54-1.32) 1.02(0.67-1.55) 0.77 Ref (1.0) 1.27(0.95-1.70) 1.20(0.87-1.65) 1.55 (1.08-2.21) 0.02 Ref (1.0) 0.95 (0.56-1.60) 0.86 (0.51-1.45) 1.26(0.75-2.12) 0.36 Ref (1.0) 1.39(0.97-2.01) 1.53 (0.99-2.36) 1.40 (0.79-2.50) 0.04 Saturated fat Ql Q2 Q3 Q4 Ref (1.0) 1.23 (0.85-1.80) 1.17(0.80-1.69) 1.00(0.70-1.43) 0.69 Ref (1.0) 0.92 (0.62-1.38) 0.89(0.60-1.31) 0.91 (0.62-1.34) 0.65 Ref (1.0) 1.18(0.88-1.59) 1.24(0.90-1.70) 1.21 (0.88-1.68) 0.19 Ref (1.0) 0.91 (0.55-1.50) 0.84 (0.51-1.39) 1.08 (0.65-1.78) 0.76 Ref (1.0) 1.28 (0.87-1.89) 1.63 (1.07-2.51) 1.13(0.71-1.79) 0.19 Linoleic acid Ql Q2 Q3 Q4 Ref (1.0) 1.62(1.10-2.38) 1.35 (0.92-1.98) 1.46(1.01-2.11) 0.22 Ref (1.0) 1.20(0.76-1.91) 1.47 (0.94-2.29) 1.11 (0.73-1.70) 0.95 Ref (1.0) 1.16(0.87-1.55) 1.26(0.92-1.73) 1.38(0.98-1.94) 0.05 Ref (1.0) 1.21 (0.75-1.97) 1.30(0.79-2.16) 1.40 (0.86-2.27) 0.19 Ref (1.0) 1.18(0.81-1.71) 1.24(0.81-1.88) 1.33 (0.80-2.22) 0.19 Oleic acid Ql Q2 Q3 Q4 Ref (1.0) 1.08 (0.73-1.59) 1.26 (0.85-1.88) 1.49(1.00-2.21) 0.02 Ref (1.0) 0.78 (0.49-1.24) 0.71 (0.46-1.10) 0.82 (0.54-1.24) 0.54 Ref (1.0) 1.30(0.96-1.75) 1.47(1.06-2.04) 1.78(1.25-2.54) 0.008 Ref (1.0) 1.02(0.58-1.78) 1.02 (0.60-1.72) 1.19 (0.70-2.04) 0.48 Ref (1.0) 1.27 (0.88-1.83) 1.81 (1.16-2.85) 2.28(1.30-4.01) 0.0005 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.9 Dietary fat intake and breast cancer risk according to quartiles of fat intake as a percentage of total energy intake, among all cases and controls Dietary component Cases Controls Adj. OR* (95% Cl) P trend Adj. OR** (95% Cl) P trend Total energy Ql 406 502 Ref (1.0) Ref (1.0) intake Q2 442 504 1.11 (0.92-1.34) 1.12(0.93-1.36) Q3 432 501 1.11 (0.93-1.34) 1.15 (0.95-1.39) Q4 397 503 1.05 (0.87-1.27) 0.75 1.09 (0.89-1.33) 0.48 % energy from Ql 326 503 Ref (1.0) Ref (1.0) total fat Q2 395 503 1.20(0.99-1.45) 1.14(0.94-1.40) Q3 450 503 1.36(1.12-1.65) 1.26(1.03-1.54) Q4 506 501 1.51 (1.24-1.84) <0.0001 1.40 (1.14-1.73) 0.0006 % energy from Ql 346 495 Ref (1.0) Ref (1.0) saturated fat Q2 417 495 1.20 (0.99-1.45) 1.16(0.95-1.41) Q3 483 524 1.29 (1.07-1.56) 1.24 (1.02-1.50) Q4 431 496 1.20 (0.99-1.46) 0.06 1.12(0.92-1.37) 0.27 % energy from Ql 352 504 Ref (1.0) Ref (1.0) linoleic acid Q2 395 502 1.11 (0.92-1.35) 1.10(0.90-1.33) Q3 474 503 1.31 (1.09-1.58) 1.28(1.06-1.56) Q4 456 501 1.25 (1.03-1.51) 0.02 1.17(0.96-1.43) 0.12 % energy from Ql 321 504 Ref (1.0) Ref (1.0) oleic acid Q2 391 488 1.24(1.02-1.51) 1.19(0.97-1.45) Q3 430 509 1.30(1.07-1.58) 1.23 (1.00-1.50) Q4 535 409 1.62(1.33-1.97) <0.0001 1.49(1.21-1.84) 0.0002 V O K > Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission Table 4.3.10 Dietary fat intake and breast cancer risk according to quartiles of fat intake as a percentage of total energy intake, by ethnicity Dietary component Whites African-Americans Latinas Cases Controls Adj. OR** (95% CD P trend Cases Controls Adj. OR** (95% CD P trend Cases Controls Adj. OR** (95% CD P trend Total Ql 166 166 Ref (1.0) 139 193 Ref (1.0) 101 143 Ref (1.0) energy Q2 178 197 0.91 (0.67-1.23) 115 119 1.40(0.99-1.97) 149 188 1.22(0.86-1.73) Q3 160 153 1.07(0.78-1.47) 104 117 1.26(0.89-1.80) 168 231 1.23 (0.87-1.73) Q4 68 107 0.65 (0.44-0.97) 0.11 137 131 1.47(1.04-2.07) 0.05 192 265 1.28 (0.91-1.80) 0.25 % energy Ql 92 116 Ref (1.0) 67 79 Ref (1.0) 167 308 Ref (1.0) from total Q2 145 158 1.21 (0.84-1.74) 86 102 0.96 (0.62-1.49) 164 243 1.13 (0.85-1.52) fat Q3 146 166 1.21 (0.84-1.76) 132 155 0.96 (0.63-1.45) 172 182 1.34 (0.98-1.84) Q4 189 183 1.47 (1.02-2.12) 0.04 210 224 1.01 (0.68-1.51) 0.84 107 94 1.57(1.08-2.29) 0.01 % energy Ql 103 126 Ref (1.0) 92 103 Ref (1.0) 152 266 Ref (1.0) from Q2 141 138 1.31 (0.92-1.88) 99 137 0.77(0.52-1.14) 177 220 1.28 (0.95-1.73) saturated Q3 161 161 1.33(0.94-1.90) 162 165 1.04(0.72-1.51) 160 198 1.19(0.87-1.63) fat Q4 168 198 1.13(0.80-1.61) 0.71 142 155 0.96(0.66-1.41) 0.72 121 143 1.12(0.79-1.58) 0.56 % energy Ql 93 120 Ref (1.0) 76 89 Ref (1.0) 183 295 Ref (1.0) from Q2 148 141 1.41 (0.98-2.03) 78 112 0.79(0.51-1.22) 169 249 0.99 (0.75-1.32) linoleic Q3 145 174 1.13(0.79-1.61) 177 147 1.39(0.94-2.05) 152 182 1.22(0.90-1.64) acid Q4 186 188 1.35 (0.95-1.91) 0.28 164 212 0.86(0.58-1.26) 0.49 106 101 1.27 (0.89-1.81) 0.10 % energy Ql 88 114 Ref (1.0) 73 77 Ref (1.0) 160 313 Ref (1.0) from oleic Q2 136 154 1.22(0.84-1.78) 100 109 0.90(0.59-1.39) 155 225 1.24(0.92-1.67) acid Q3 159 178 1.32(0.91-1.91) 113 150 0.74(0.49-1.12) 158 181 1.43 (1.04-1.96) Q4 189 177 1.68 (1.15-2.47) 0.006 209 224 0.88 (0.60-1.31) 0.59 137 108 1.80 (1.25-2.61) 0.001 <0 u > Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission Table 4.3.11 Dietary fat intake and breast cancer risk according to quartiles of fat intake as a percentage of total energy ___________intake, among US born and foreign born Latinas__________________________________________________ Dietary fat US-born Latinas Foreign-born Latinas Cases Controls Adj. OR (95% Cl)** P trend Cases Controls Adj. OR (95% Cl)** P trend Total energy Ql 55 54 Ref (1.0) 46 89 Ref (1.0) Q2 88 74 1.23 (0.75-2.04) 61 114 1.15(0.69-1.90) Q3 74 75 1.03 (0.62-1.72) 94 156 1.37(0.86-2.20) Q4 90 80 1.26 (0.75-2.10) 0.55 102 185 1.27 (0.80-2.04) 0.35 % energy Ql 43 39 Ref (1.0) 124 269 Ref (1.0) from total fat Q2 80 82 0.89 (0.51-1.55) 84 161 1.13(0.79-1.62) Q3 103 101 0.90 (0.53-1.55) 69 81 1.77(1.17-2.67) Q4 81 61 1.22 (0.68-2.18) 0.36 26 33 1.43 (0.79-2.60) 0.02 % energy Ql 50 50 Ref (1.0) 102 216 Ref (1.0) from Q2 83 68 1.30(0.76-2.20) 94 152 1.22(0.84-1.78) saturated fat Q3 90 93 1.00(0.60-1.68) 70 105 1.37 (0.90-2.06) Q4 84 72 1.13 (0.66-1.93) 0.98 37 71 1.02(0.62-1.67) 0.53 % energy Ql 64 71 Ref (1.0) 119 224 Ref (1.0) from linoleic Q2 88 83 1.21 (0.76-1.94) 81 166 0.90 (0.63-1.31) acid Q3 79 71 1.28 (0.79-2.09) 73 111 1.18(0.80-1.75) Q4 76 58 1.44(0.87-2.37) 0.17 30 43 1.16(0.67-2.01) 0.40 % energy Ql 40 35 Ref (1.0) 120 278 Ref (1.0) from oleic Q2 73 67 1.02(0.57-1.85) 82 158 1.20 (0.83-1.73) acid Q3 91 101 0.85 (0.48-1.48) 67 80 1.89 (1.25-2.86) Q4 103 80 1.18(0.66-2.11) 0.57 34 28 2.16(1.20-3.89) 0.0007 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission Table 4.3.12 Dietary fat intake and breast cancer risk according to quartiles of fat intake as a percentage of total energy intake, with fatty acids mutually adjusted Ail Whites African-Americans Latinas US-born Latinas foreign-born Latinas Dietary fat Adj. OR*** (95% Cl) P trend Adj. OR*** (95% Cl) P trend Adj. OR*** (95% Cl) P trend Adj. OR*** (95% Cl) P trend Adj. OR*** (95% Cl) P trend Adj. OR*** (95% Cl) P trend % energy from saturated fat Ql Q2 Q3 Q4 Ref (1.0) 1.01 (0.81-1.27) 0.98(0.77-1.26) 0.81(0.61-1.07) 0.08 Ref (1.0) 1.12(0.75-1.68) 1.00(0.65-1.54) 0.74(0.46-1.19) 0.08 Ref(1.0) 0.93(0.57-1.51) 1.43(0.85-2.41) 1.33(0.75-2.37) 0.21 Ref (1.0) 1.06(0.75-1.48) 0.84(0.57-1.24) 0.68(0.43-1.08) 0.06 Ref (1.0) 1.43 (0.77-2.65) 1.08 (0.56-2.08) 1.15(0.55-2.38) 0.89 Ref (1.0) 0.91 (0.59-1.40) 0.77(0.46-1.29) 0.45 (0.23-0.86) 0.02 % energy from linoleic acid Ql Q2 Q3 Q4 Ref (1.0) 1.02(0.83-1.24) 1.13(0.91-1.38) 0.96(0.77-1.21) 0.71 Ref (1.0) 1.28(0.88-1.85) 0.95(0.65-1.38) 1.01(0.69-1.50) 0.58 Ref (1.0) 0.89(0.56-1.40) 1.67 (1.07-2.61) 1.05(0.67-1.65) 0.91 Ref (1.0) 0.88(0.66-1.19) 1.03(0.75-1.41) 0.95(0.65-1.40) 0.98 Ref (1.0) 1.21(0.74-1.96) 1.31(0.79-2.17) 1.41 (0.82-2.42) 0.24 Ref (1.0) 0.76(0.51-1.12) 0.84(0.55-1.29) 0.67(0.36-1.24) 0.22 % energy from oleic acid Ql Q2 Q3 Q4 Ref (1.0) 1.19(0.94-1.50) 1.28(0.97-1.68) 1.69(1.23-2.31) 0.0008 Ref (1.0) 1.21 (0.80-1.88) 1.45 (0.90-2.34) 2.12(1.22-3.67) 0.004 Ref (1.0) 0.78(0.46-1.31) 0.57(0.31-1.04) 0.58(0.30-1.13) 0.14 Ref (1.0) 1.31 (0.93-1.86) 1.64(1.08-2.50) 2.38(1.41-4.01) 0.001 Ref (1.0) 0.88(0.46-1.71) 0.70(0.34-1.46) 0.99(0.43-2.24) 0.87 Ref (1.0) 1.45(0.94-2.24) 2.73(1.56-4.79) 4.09(1.85-9.04) 0.000 1 *** Beside all above covariates, further adjusted for saturated fat (5% energy, continuous), linoleic acid (5% energy, continuous), oleic acid (5% energy, continuous), V O C O 4.3.4 Effect Modification We checked whether some known non-dietary risk factors might modify the association between breast cancer and dietary fat intake. Fifteen percent of cases and 11% of controls had a family history of breast cancer (FHBC) in first-degree relatives. Overall, there was no interaction between FHBC and total fat or types of fat. The odds ratios for women without a FHBC were similar to those with everybody included (table 4.3.13). Among women with a FHBC, the association with total fat appeared stronger than among those without a FHBC. However, we had insufficient power to evaluate it due to the small number of women with a FHBC. The only significant interaction was observed between linoleic acid and FHBC among US-born Latinas, probably due to chance. Sixteen percent of controls and 20% of cases had a previous history of benign breast disease (BBD). Linoleic acid was positively associated with breast cancer among women with a history of BBD, but inversely among women without a history of BBD (p=0.003 for interaction) (table 4.3.13). However, this interaction was only significant among African-Americans (p<0.0001 for interaction). 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 4.3.13 Dietary fat intake and breast cancer risk according to according to 5% increase of total energy intake from fatty acids, by family history of breast cancer and history of benign breast disease Dietary fat Adj. OR (95% CD*** for 5% increase of energy All Whites African- Americans Latinas US-born Latinas Foreign-born Latinas All Total fat 1.08 (1.02-1.13) 1.09(1.00-1.18) 0.99(0.91-1.08) 1.12(1.02-1.23) 1.04(0.90-1.20) 1.18(1.04-1.33) Saturated fat 0.91 (0.77-1.08) 0.81 (0.63-1.05) 1.19(0.84-1.68) 0.86 (0.62-1.18) 1.06 (0.64-1.75) 0.73 (0.48-1.13) Linoleic acid 0.99(0.79-1.23) 0.96 (0.68-1.35) 0.88(0.58-1.34) 1.13(0.73-1.77) 1.55 (0.86-2.79) 0.81 (0.40-1.64) Oleic acid 1.32 (1.06-1.64) 1.64(1.11-2.43) 0.85 (0.57-1.28) 1.51 (1.01-2.24) 0.86 (0.47-1.57) 2.22(1.27-3.89) Family history Total fat of breast No 1.05(1.00-1.11) 1.08 (0.98-1.18) 0.95 (0.87-1.04)d 1.10(1.00-1.21) 0.99 (0.85-1.15) 1.16(1.02-1.33) cancer Yes 1.21 (1.06-1.38) 1.16(0.95-1.43) 1.37(1.05-1.79) 1.37(0.95-1.99) 1.62(0.91-2.91) 1.51 (0.70-3.27) Saturated fat No 0.87(0.73-1.05) 0.77(0.58-1.02) 1.12(0.77-1.62) 0.83 (0.59-1.17) 0.99(0.58-1.68) 0.72(0.46-1.13) Yes 1.14(0.71-1.82) 0.96(0.50-1.87) 1.62(0.46-5.64) 0.81 (0.24-2.73) 0.62 (0.08-5.06) 0.14(0.01-2.45) Linoleic acid No 0.95 (0.75-1.21) 0.97 (0.66-1.43) 0.83 (0.53-1.30) 1.01 (0.63-1.62) 1.22(0.65-2.27)a 0.80(0.38-1.68) Yes 1.12(0.61-2.07) 0.96 (0.42-2.16) 1.55 (0.33-7.35) 1.72(0.34-8.77) 5.31 (0.44-64.2) 0.08(0.002-3.1) Oleic acid No 1.32(1.04-1.68) 1.69(1.09-2.63) 0.82 (0.53-1.27) 1.54(1.00-2.34) 0.91 (0.48-1.74) 2.22(1.23-4.02) Yes 1.36(0.75-2.47) 1.52 (0.61-3.79) 1.27 (0.30-5.30) 2.03 (0.50-8.19) 2.94(0.19-46.1) 25.9(1.15-582) History of Total fat benign breast No 1.07(1.01-1.13) 1.05 (0.96-1.16) 1.00(0.91-1.10) 1.14(1.03-1.27) 1.09(0.92-1.29) 1.15(1.01-1.32) disease Yes 1.12(1.01-1.26) 1.29(1.06-1.57) 1.00 (0.82-1.23) 1.12(0.89-1.41) 0.96 (0.70-1.32) 1.53 (1.01-2.32) Saturated fat No 0.92 (0.76-1.11) 0.81 (0.61-1.09) 1.17(0.80-1.73) 0.86(0.60-1.21) 1.02(0.57-1.81) 0.76(0.48-1.21) Yes 0.88(0.60-1.31) 0.78 (0.43-1.42) 1.03 (0.42-2.53) 1.06 (0.46-2.43) 1.85 (0.58-5.87) 0.65(0.14-2.91) Linoleic acid No 0.81 (0.63-1.05)b 0.92(0.61-1.39) 0.54 (0.33-0.88)° 1.10(0.67-1.81) 1.50(0.76-2.94) 0.79 (0.37-1.68) Yes 1.76(1.08-2.07) 0.90(0.45-1.81) 6.20(2.08-18.4) 1.38(0.44-4.34) 2.55 (0.60-10.9) 1.49(0.13-17.7) Oleic acid No 1.39(1.08-1.78) 1.51 (0.97-2.36) 1.06(0.67-1.66) 1.56(1.01-2.41) 1.02(0.51-2.02) 2.00(1.11-3.61) Yes 1.22(0.73-2.02) 2.96(1.18-7.43) 0.58(0.21-1.64) 1.22 (0.42-3.58) 0.33 (0.07-1.50) 4.77 (0.57-39.7) Interaction test: a. p=0.04; b. p=0.003; c. pO.OOOl; d. p=0.08 vo Stratifying the analysis by menopausal status, we found no significant interaction between menopausal status and total fat or types of fat intake. However, we observed a few significant interactions that were not consistent across ethnicities (table 4.3.14). Linoleic acid was found to be protective against premenopausal breast cancer among African-Americans (p=0.04 for interaction). Saturated fat was protective against premenopausal breast cancer among US-bom Latinas (p=0.04 for interaction). Total fat was positively associated with premenopausal but not postmenopausal breast cancer among foreign-born Latinas (p=0.01 for interaction). Since these interactions were not consistent across ethnicities, they might be found due to chance. Because physical activity may influence total energy intake and metabolism, we also examined whether the effects of fats are different among women with high versus low physical activity, defined as above or below the median of lifetime physical activity among the controls. No significant interaction was found between physical activity and intake of specific type of fat (table 4.3.15). Risk factors for estrogen receptor (ER)/ progesterone receptor (PR) positive breast cancer may be different from those for ER/PR negative tumors. Conducting analyses for ER/PR positive and ER/PR negative cases separately, none of the associations were found to significantly differ by ER/PR status, although the positive association between oleic acid and risk among whites and Latinas seemed stronger for ER negative breast cancer than for ER positive cancer (table 4.3.16). 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 4.3.14 Dietary fat intake and breast cancer risk according to 5% increase of total energy intake from fatty acids, ____________ by menopausal status_________________________________________________________________________ _______________________________ Adj. OR (95% Cl) *** for 5% increase of energy______________________________ All Whites African- Latinas US-born Foreign-born Dietary fat_____________________________________________ Americans___________________________Latinas___________ Latinas Total fat Premenopausal 1.08 (0.98-1.19) 1.02 (0.85-1.23) 0.98 (0.83-1.15) 1.28 (1.06-1.54) 0.92 (0.66-1.28) 1.52 (1.18-1.96)° Postmenopausal 1.08 (1.02-1.14) 1.10 (1.00-1.22) 1.04 (0.94-1.16) 1.07 (0.96-1.20) 1.06 (0.89-1.25) 1.06 (0.90-1.24) Saturated fat Premenopausal 1.04 (0.74-1.47) 0.83 (0.46-1.47) 1.80 (0.91-3.94) 0.87 (0.46-1.64) 0.30 (0.09-0.98)b 1.27 (0.56-2.86) Postmenopausal 0.90 (0.73-1.11) 0.81 (0.59-1.09) 1.17 (0.76-1.82) 0.88 (0.60-1.30) 1.63 (0.89-3.01) 0.55 (0.31-0.96) Linoleic acid Premenopausal 0.70 (0.46-1.09) 0.89 (0.46-1.71) 0.37 (0.15-0.89)a 0.75 (0.31-1.78) 0.73 (0.21-2.51) 0.75 (0.19-2.93) Postmenopausal 1.08 (0.81-1.43) 0.94 (0.61-1.46) 1.23 (0.74-2.05) 1.10 (0.62-1.96) 1.98 (0.92-4.29) 0.49 (0.18-1.31) Oleic cid Premenopausal 1.41 (0.92-2.15) 1.31 (0.59-2.92) 0.91 (0.41-2.00) 2.49 (1.16-5.36) 2.64 (0.75-9.37) 2.71 (0.92-7.95) Postmenopausal 1.27 (0.97-1.68) 1.73 (1.07-2.80) 0.84 (0.50-1.40) 1.30 (0.79-2.14) 0.55 (0.26-1.17) 2.51 (1.22-5.17) Interaction test: a. p=0.04; b. p=0.04; c. p=0.01 Table 4.3.15 Dietary fat intake and breast cancer risk according to 5% increase of total energy intake from fatty ____________ acids, by physical activity_____________________________________________________________________ ______________________________Adj. OR (95% CI)***for 5% increase of energy____________________________ Dietary fat All Whites African- Latinas US-born Latinas Foreign-born Americans Latinas Total fat PA<median 1.09 (1.02-1.16) 1.10 (0.98-1.23) 1.03 (0.92-1.16) 1.19 (1.04-1.36) 1.05 (0.85-1.30) 1.29 (1.07-1.55) PA>median 1.06 (0.99-1.14) 1.11 (0.97-1.26) 0.95 (0.83-1.09) 1.08 (0.95-1.24) 1.07 (0.87-1.30) 1.07 (0.89-1.29) Saturated fat PA<median 0.95 (0.75-1.19) 0.80 (0.57-1.13) 1.14 (0.72-1.81) 1.06 (0.67-1.67) 1.16 (0.56-2.40) 1.10 (0.59-2.05) PA>median 0.87 (0.67-1.13) 0.77 (0.52-1.15) 1.25 (0.72-2.20) 0.73 (0.46-1.15) 1.08 (0.52-2.25) 0.46 (0.24-0.88) Linoleic acid PA<median 1.05 (0.78-1.41) 1.12 (0.71-1.76) 0.75 (0.43-1.30) 1.34 (0.71-2.52) 1.80 (0.78-4.16) 1.03 (0.37-2.88) PA>median 0.91 (0.65-1.29) 0.74 (0.43-1.27) 1.13 (0.58-2.21) 0.97 (0.50-1.87) 1.44 (0.59-3.53) 0.59 (0.21-1.68) Oleic acid PA<median 1.29 (0.96-1.73) 1.50 (0.89-2.53) 1.11 (0.65-1.89) 1.33 (0.76-2.35) 0.77 (0.31-1.92) 1.69 (0.76-3.79) PA>median 1.36 (0.97-1.91) 2.17 (1.16-4.09) 0.64 (0.34-1.23) 1.67 (0.94-2.97) 0.91 (0.39-2.14) 3.21 (1.40-7.38) Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.16 Dietary fat intake and breast cancer risk, by ER and PR status Adj. OR (95% Cl)*** for 5% increase of energy Dietary fat All Whites African-Americans Latinas US-born Latinas Foreign-born Latinas ER status Total fat ER+ ER - Saturated fat ER+ ER - Linoleic add ER+ ER - Oleic add ER+ ER - 1.07(1.01-1.13) 1.10(1.01-1.20) 1.00 (0.83-1.22) 0.78 (0.58-1.07) 1.01 (0.79-1.31) 0.93 (0.63-1.36) 1.17(0.91-1.50) 1.65(1.13-2.41) 1.09(0.99-1.19) 1.19(1.00-1.42) 0.92 (0.69-1.21) 0.66 (0.38-1.16) 1.00 (0.69-1.47) 1.08 (0.55-2.09) 1.39(0.90-2.13) 2.40(1.08-5.32) 0.99 (0.90-1.10) 1.00 (0.88-1.14) 1.33 (0.88-2.01) 0.96(0.56-1.65) 0.90(0.54-1.51) 0.69 (0.35-1.36) 0.76(0.47-1.23) 1.12(0.60-2.09) 1.11 (1.00-1.24) 1.14(0.97-1.33) 0.93 (0.64-1.34) 0.80 (0.47-1.38) 1.19(0.71-1.97) 0.97(0.48-1.94) 1.30(0.83-2.06) 1.86(0.98-3.53) 0.96 (0.82-1.13) 1.06 (0.85-1.32) 1.09 (0.61-1.94) 0.92 (0.42-2.03) 1.61 (0.83-3.11) 1.17(0.48-2.82) 0.66 (0.34-1.32) 1.31 (0.51-3.39) 1.22(1.05-1.42) 1.17 (0.94-1.46) 0.82 (0.50-1.35) 0.69 (0.32-1.48) 0.80 (0.35-1.83) 0.85 (0.25-2.86) 2.14(1.12-4.07) 2.34 (0.91-6.05) PR status Total fat PR+ 1.06(1.00-1.13) 1.07(0.97-1.18) 0.98 (0.88-1.09) 1.11 (0.99-1.25) 0.98(0.82-1.16) 1.22(1.04-1.43) PR - 1.10(1.02-1.19) 1.20(1.05-1.38) 0.99(0.88-1.11) 1.13 (0.98-1.29) 1.02(0.84-1.25) 1.18 (0.97-1.43) Saturated fat PR+ 0.91 (0.75-1.12) 0.83 (0.62-1.12) 1.11 (0.72-1.73) 0.88 (0.59-1.30) 0.97(0.53-1.80) 0.84 (0.50-1.43) PR - 1.02(0.79-1.32) 1.01 (0.67-1.52) 1.28 (0.79-2.05) 0.91 (0.57-1.45) 1.18(0.59-2.37) 0.69(0.35-1.33) Linoleic add PR+ 0.97(0.74-1.26) 0.96(0.65-1.43) 0.72(0.42-1.26) 1.25(0.74-2.12) 1.62(0.82-3.21) 0.96 (0.40-2.30) PR - 1.03 (0.73-1.43) 1.12(0.65-1.93) 0.92 (0.51-1.68) 0.93 (0.49-1.76) 1.33 (0.59-2.97) 0.55 (0.18-1.63) Oleic acid PR+ 1.30(1.00-1.69) 1.53 (0.98-2.40) 0.93 (0.56-1.56) 1.39(0.86-2.24) 0.78 (0.38-1.60) 2.02(1.03-4.00) PR - 1.21 (0.87-1.68) 1.51 (0.80-2.85) 0.76 (0.43-1.34) 1.52(0.86-2.68) 0.82 (0.35-1.92) 2.57(1.12-5.86) ER and PR status Totai fat ER &PR+ 1.07(1.01-1.13) 1.06(0.96-1.17) 1.00(0.90-1.12) 1.11 (0.99-1.25) 0.97(0.81-1.16) 1.22(1.04-1.44) ER &PR - 1.10(1.01-1.21) 1.14(0.94-1.39) 1.04 (0.90-1.20) 1.13 (0.96-1.33) 1.04(0.81-1.33) 1.18 (0.93-1.48) Saturated fat ER &PR+ 0.95 (0.77-1.17) 0.82 (0.61-1.11) 1.29 (0.81-2.03) 0.95 (0.64-1.41) 1.09 (0.59-2.03) 0.88 (0.51-1.50) ER & PR- 0.84 (0.60-1.16) 0.66 (0.34-1.25) 1.15 (0.65-2.03) 0.86(0.49-1.51) 1.01 (0.44-2.33) 0.71 (0.32-1.57) Linoleic add ER &PR+ 1.01 (0.77-1.33) 0.96 (0.64-1.44) 0.85 (0.48-1.49) 1.40(0.80-2.44) 1.93 (0.94-3.98) 0.96 (0.39-2.39) ER &PR- 0.97 (0.64-1.47) 1.07(0.51-2.26) 0.82 (0.40-1.68) 1.06 (0.50-2.25) 1.33 (0.50-3.55) 0.84 (0.23-3.03) Oleic acid ER &PR+ 1.24(0.95-1.62) 1.51 (0.96-2.38) 0.82(0.48-1.40) 1.24(0.76-2.02) 0.64 (0.31-1.33) 1.97 (0.99-3.92) ER &PR - 1.54(1.03-2.31) 2.03 (0.83-4.97) 0.99 (0.51-1.92) 1.68 (0.85-3.29) 1.08 (0.39-3.03) 2.38 (0.89-6.32) We also examined whether types of fats are associated with breast cancer stage and histological grades. Saturated fats showed protective effects against localized breast cancer, but not more advanced breast cancer (p=0.004 for heterogeneity test) (table 4.3.17). This association was observed in both whites and Latinas. On the other hand, the positive association between oleic acid and breast cancer was also mostly in localized cases, especially among Latinas (p=0.06 for heterogeneity test). Histological grades of breast cancer did not influence the association between types of fat and cancer risk. Therefore, when combining the stage and grade and comparing the less progressive case (grade 1 or 2 and localized stage) and more progressive case (grade 3 or 4 and advanced stages), the results were similar to the comparison by stage only. 4.3.5 Cooking Fat Usage and Breast Cancer A question about the most commonly used cooking fat was asked among a subgroup of cases diagnosed between 1995 and 1998, and their matched controls (n=1913). We therefore could analyze the association between cooking fat usage and breast cancer among these subjects. The majority of the study participants chose either vegetable/com oil (46%) or olive/canola oil (39%) as their most commonly used cooking fat (table 4.3.18). Over 65% of whites used olive or canola oils the most, while African-Americans and Latinas used vegetable or com oils most commonly. Due to the sparse data for other categories, we summarized the cooking fat into 4 groups according to the properties of the fat. The baseline group included those choosing “pam or no oil” or “olive or canola oil”. Butter and lard were put in one 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. group as they both mainly contain saturated fat. Margarine, low-fat margarine and vegetable shortening were grouped together as hydrogenated fats. Results showed that compared with using olive or canola oils, using vegetable or com oils was associated with increased risk of breast cancer (multivariate-adjusted OR: 1.35; 95% Cl: 1.07-1.69) (table 4.3.19). Using hydrogenated fats conferred even higher risk (multivariate-adjusted OR: 1.63; 95% Cl: 1.15-2.29). These associations remained after further adjustment for types and amount of fat intake. Although the popularity of the cooking fats varied across ethnicities, similar associations were found among all three ethnicities. However, most of the associations were not statistically significant within each ethnicity due to the smaller sample size. 102 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 4.3.17 Dietary fat intake and breast cancer risk according to according to 5% increase of total energy intake from fatty acids, by stage and histological grade of disease Adj. OR (95% Cl)*** for 5% increase of energy Dietary fat All Whites African-Americans Latinas US-born Latinas Foreign-born Latinas Stage Total fat local 1.06(1.01-1.12) 1.07(0.97-1.17) 0.98(0.89-1.08) 1.13(1.02-1.26) 1.06(0.90-1.24) 1.17(1.01-1.36) advanced 1.12(1.04-1.20) 1.19(1.05-1.35) 1.02(0.90-1.15) 1.12(0.98-1.28) 1.02(0.84-1.24) 1.18 (0.98-1.43) Saturated fat local 0.80 (0.66-0.97) 0.69(0.51-0.92) 1.12(0.75-1.68) 0.75(0.52-1.08) 0.92(0.52-1.62) 0.62(0.37-1.03) advanced 1.18(0.93-1.49)“ 1.14 (0.79-1.64)b 1.21 (0.75-1.96) 1.16(0.74-1.82)' 1.36(0.69-2.67) 1.06(0.57-1.98) Linoleic acid local 0.96(0.75-1.21) 0.97(0.67-1.42) 0.86(0.53-1.40) 0.98 (0.59-1.62) 1.30(0.67-2.53) 0.76(0.34-1.72) advanced 1.05(0.76-1.45) 0.90(0.53-1.51) 0.82(0.44-1.51) 1.55(0.85-2.84) 1.99(0.93-4.27) 0.98 (0.35-2.77) Oleic acid local 1.47(1.14-1.88) 1.86(1.21-2.87) 0.87(0.54-1.38) 1.81 (1.15-2.84) 1.04(0.53-2.05) 2.67(1.40-5.09) advanced 1.11 (0.81-1.53) 1.46(0.81-2.65) 0.93(0.53-1.64) 1.01 (0.58-1.77)“ 0.64(0.28-1.46) 1.40(0.62-3.16) Histological grade Total fat Grade 1 & 2 1.06 (1.00-1.13) 1.06(0.96-1.17) 1.01 (0.91-1.13) 1.09(0.97-1.22) 1.04(0.87-1.23) 1.08(0.91-1.28) 3 & 4 1.10(1.02-1.18) 1.17(1.02-1.33) 0.94(0.84-1.06) 1.21 (1.06-1.37) 1.05(0.87-1.27) 1.34 (1.13-1.60)'' Saturated fat Grade 1 & 2 0.91 (0.74-1.12) 0.81 (0.60-1.09) 1.22(0.77-1.92) 0.81 (0.54-1.23) 0.96(0.51-1.80) 0.72(0.41-1.28) 3 & 4 0.93(0.73-1.19) 0.86(0.57-1.29) 1.14(0.72-1.81) 0.87(0.56-1.34) 0.88(0.44-1.75) 0.85(0.47-1.51) Linoleic acid Grade 1 & 2 1.09(0.83-1.43) 0.92(0.61-1.37) 1.08(0.63-1.84) 1.50(0.86-2.60) 2.11 (1.04-4.29) 0.91 (0.36-2.29) 3 & 4 0.88(0.63-1.21) 0.91 (0.53-1.56) 0.57(0.31-1.05)' 1.12(0.62-2.03) 1.29(0.58-2.85) 1.05(0.41-2.68) Oleic acid Grade 1 & 2 1.20(0.91-1.58) 1.56(0.98-2.46) 0.80(0.47-1.36) 1.22(0.74-2.03) 0.76(0.36-1.61) 1.65 (0.79-3.44) 3 & 4 1.46 (1.07-2.00) 1.90(1.03-3.50) 0.91 (0.53-1.58) 1.93(1.13-3.30) 1.28(0.56-2.92) 2.57(1.23-5.36) Stage and histological Total fat Low grade & local 1.03(0.96-1.10) 1.02(0.92-1.13) 1.00(0.88-1.14) 1.05(0.92-1.19) 1.00(0.83-1.22) 1.03(0.85-1.25) grade High grade or advanced 1.11 (1.05-1.18) 1.18(1.06-1.31) 0.97(0.88-1.07) 1.18(1.06-1.32) 1.07 (0.91-1.27) 1.26 (1.09-1.47^ Saturated fat Low grade & local 0.74 (0.58-0.94) 0.65 (0.47-0.92) 1.03 (0.61-1.76) 0.64 (0.40-1.04) 0.64(0.31-1.34) 0.63(0.32-1.22) High grade or advanced 1.08 (0.88-1.3 l)f 1.04 (0.75-1.43 y 1.22(0.81-1.84) 1.02 (0.70-1.49)h 1.21 (0.68-2.17) 0.92(0.55-1.52) Linoleic acid Low grade & local 0.99(0.73-1.34) 0.89(0.57-1.38) 0.97(0.53-1.79) 1.14(0.61-2.13) 1.65 (0.74-3.67) 0.69(0.24-2.01) High grade or advanced 1.00(0.76-1.31) 0.92(0.59-1.44) 0.75(0.44-1.26) 1.32 (0.79-2.20) 1.55(0.79-3.05) 1.13 (0.50-2.59) Oleic acid Low grade & local 1.40(1.03-1.91) 1.81 (1.09-2.99) 0.93 (0.50-1.72) 1.52(0.86-2.71) 1.08(0.46-2.53) 1.88(0.82-4.33) High grade or advanced 1.22(0.93-1.60) 1.58(0.94-2.64) 0.83 (0.51-1.35) 1.44(0.90-2.29) 0.89 (0.44-1.81) 1.95 (1.01-3.74) Heterogeneity test: a. p=0.004; b. p=0.01; c. p=0.08; d. p=0.06; e. p=0.07; f. p=0.007; g. p=0.02; h. p=0.08; i. p=0.04; j. p=0.06 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission. Table 4.3.18 The most commonly used cooking fat among controls of a subset of study subjects (n=1081) Coo Controls (%) iking oil All Whites African-Americans Latinas US-born Latinas Foreign-born Latinas Olive or canola oil 418(38.7) 208 (65.6) 74(27.8) 136(27.3) 51 (39.2) 85 (23.1) Vegetable or Corn oil 495 (45.8) 46 (14.5) 128(48.1) 321 (64.5) 64(49.2) 257 (69.8) Margarine 42 (3.9) 17(5.4) 21 (7.9) 4(0.8) 2(1.5) 2 (0.5) Low-fat margarine 19(1.8) 8 (2.5) 6(2.3) 5(1.0) 2(1.5) 3 (0.8) Vegetable oil shortening 29 (2.7) 1 (0.3) 24 (9.0) 4(0.8) 2(1.5) 2 (0.5) Butter 21 (1.9) 14 (4.4) 6(2.3) 1 (0.2) 1 (0.8) 0 Lard 21 (1.9) 0 2 (0.8) 19(3.8) 2(1.5) 17 (4.6) Pam or no oil1 36(3.3) 23 (7.3) 5(1.9) 8(1.6) 6(4.6) 2(0.5) Table 4.3.19 Cooking fat and breast cancer risk among three ethnic groups Cooking oil Cases Controls Adj. OR* Adj. OR** Adj. OR*** (95% Cl) (95% Cl) (95% Cl) All Olive or Canola oil, or no oil 327 454 Ref (1.0) Ref (1.0) Ref (1.0) Butter or Lard 31 42 1.05 (0.65-1.71) 1.16(0.70-1.92) 1.16(0.70-1.93) Vegetable or corn oil 375 495 1.11 (0.90-1.37) 1.34 (1.07-1.69) 1.32(1.05-1.65) Hydrogenated unsaturated fat 99 90 1.42(1.02-1.98) 1.61 (1.14-2.27) 1.55 (1.09-2.19) Whites Olive or Canola oil, or no oil 160 231 Ref (1.0) Ref (1.0) Ref (1.0) Butter or Lard 10 14 1.02 (0.44-2.35) 1.03 (0.43-2.45) 1.27 (0.52-3.13) Vegetable or corn oil 60 46 1.86(1.20-2.87) 1.97(1.24-3.14) 2.03 (1.26-3.28) Hydrogenated unsaturated fat 27 26 1.49 (0.84-2.65) 1.62 (0.89-2.95) 1.59 (0.86-2.94) African- Americans Olive or Canola oil, or no oil 60 79 Ref (1.0) Ref (1.0) Ref (1.0) Butter or Lard 7 8 1.24 (0.42-3.62) 0.99 (0.32-3.07) 0.93 (0.30-2.93) Vegetable or corn oil 122 128 1.24(0.82-1.90) 1.23 (0.79-1.92) 1.25 (0.79-1.96) Hydrogenated unsaturated fat 61 51 1.57 (0.95-2.60) 1.54 (0.90-2.65) 1.62 (0.93-2.81) Latinas Olive or Canola oil, or no oil 107 144 R ef (1.0) Ref (1.0) Ref (1.0) Butter or Lard 14 20 0.94(0.45-1.96) 1.28 (0.57-2.89) 1.24 (0.55-2.81) Vegetable or corn oil 193 321 0.81 (0.60-1.11) 1.29 (0.90-1.84) 1.25 (0.87-1.80) Hydrogenated unsaturated fat 11 13 1.17(0.50-2.72) 1.59 (0.66-3.84) 1.49 (0.61-3.61) 4.3.6 Discussion In this population-based multiethnic case-control study of breast cancer, we evaluated whether intakes of fat and specific types of fatty acids are associated with increased risk of breast cancer. We found a positive association between total fat and oleic acid, with breast cancer risk among all women combined. However, the associations were inconsistent across ethnicities. A large amount of evidence over decades of animal experiments supports the hypothesis that fat contributes to breast cancer initiation and promotion. Large ecologic and international comparisons also support the independent association between high-fat intake and increased risk of breast cancer (Prentice and Sheppard, 1990). Although a similar relationship has been reported in many case-control studies (De Stefani et al., 1998; Ewertz and Gill, 1990; Katsouyanni et al., 1988; Qi et al., 1994; Richardson et al., 1991; Ronco et al., 1996; Sieri et al., 2002; Van't Veer et al., 1990; Wakai et al., 2000; Yu et al., 1990) and meta-analyses of case-control studies (Boyd et al., 1993; Howe et al., 1990), these findings are always questioned because of the potential for recall bias (Giovannucci et al., 1993). Furthermore, there are also many case-control studies that did not find a significant association between high-fat intake and breast cancer risk (Goodman et al., 1992; Goodstine et al., 2003; Graham et al., 1991; Hirohata et al., 1987; Ingram et al., 1991; Lee et al., 1991; Martin-Moreno et al., 1994; Potischman et al., 1998; Pryor et al., 1989; Rohan et al., 1988; Yuan et al., 1995; Zaridze et al., 1991). 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Several large prospective studies have been conducted to investigate the association between dietary fat and breast cancer risk (Howe et al., 1991; Knekt et al., 1990; Kushi et al., 1992) (Byrne et al., 2002; Gaard et al., 1995; Graham et al., 1992; Holmes et al., 1999; Toniolo et al., 1994; van den Brandt et al., 1993; Velie et al., 2000; Voorrips et al., 2002; Willett et al., 1992; Wolk et al., 1998). Overall, these prospective studies have failed to find a significant association. One example is the Nurses’ Health Study (NHS), which has published a series of reports on dietary fat and breast cancer. Their first publication in 1987 included 601 incident cases after 4 years of follow-up and found no association between fat and increased risk of breast cancer (Willett et al., 1987). In fact, total fat intake was associated with a slightly reduced risk of breast cancer (highest vs. lowest quintile RR: 0.82, 95% Cl: 0.64 to 1.05). These data were based on a limited period of follow-up and did not exclude a possible influence at levels lower than 30 percent of calories. In later reports from the same study including 1439 cases and 8 years of follow-up (Willett et al., 1992), and even later 3000 cases and 14 years of follow-up (Holmes et al., 1999), they still found no evidence that lower intake of total fat or specific major types of fat was associated with a decreased risk of breast cancer. In response to a report from Breast Cancer Detection Demonstration Project (BCDDP) cohort, which found breast cancer risk associated with greater total, unsaturated fat, and oleic acid intake among postmenopausal women without benign breast disease (Velie et al., 2000), in 2000 the NHS published an analysis among a similar population, but finding no increase in the rate of breast cancer with greater intake of dietary fat and fat subtypes (Byrne et al., 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2002). Despite all these negative reports, the most recent publication from NHS in 2003 reported that intake of animal fat during premenopausal years was reported to be associated with an increased risk of breast cancer (Cho et al., 2003). Although studies focusing on high fat diet and breast cancer risk have covered many different populations over the world, few of them included multiple ethnicities in the same study. While most studies conducted in European and Asian countries included relatively homogenous populations, even in the more ethnically diverse US most of the studies also only focused on non-Hispanic whites. The only two case- control studies including both Caucasians and Asians in Hawaii, had small numbers in each ethnic group (Goodman et al., 1992; Hirohata et al., 1987). Other multi-ethnic studies with bigger sample size have not yet reported any results on dietary fats (Brinton et al., 1997; Kolonel et al., 2000). In this study, we included three major ethnicities of the United States, non-Hispanic whites, Latinas and African-Americans with more than 1,000 cases and controls in each ethnic group. Therefore, our study provides a unique opportunity to evaluate fat intake among a population with more diverse dietary habits. In addition, more than half of the Latinas in this study were foreign-born, providing us the chance to compare dietary difference among Latinas. The mean total fat intake among whites (32.2 + 7.4, % of energy) in this study is comparable to those reported by other studies (Byrne et al., 2002; Velie et al., 2000), while African-Americans had higher (33.8+ 7.7) and foreign-born Latinas had lower (26.1+ 5.9) mean total fat intake. Therefore, by covering multiple ethnicities that have diverse dietary habits, we could examine dietary fat over a wider range of intake. 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In our study, the positive association between total fat and breast cancer was only seen among whites and Latinas, but not among African-Americans. When we divided Latinas into US-bom and foreign-born, the effect was only seen among foreign-born Latinas, but not among US-bom Latinas. This inconsistency is hard to explain by recall bias, since there seems to be no reason that whites and foreign-born Latinas would recall fat intake differently from African-Americans and US-bom Latinas. The major dietary sources for fat differed among ethnicities. Among whites, the top 5 dietary sources for total fat were salad dressing and mayonnaise, butter, salty snacks, cheese and cheese spreads, and margarine. This is similar to other reports (Velie et al., 2000). Among American-Africans, the top source is sausage or bacon, followed by fried chicken, salty snacks, salad dressing, and French fries. Among US- bom Latinas, French fries is the top source of total fat, followed by salad dressing, salty snacks, margarine, and hamburger/cheeseburger. Among foreign-born Latnias, com bread/com tortillas are at the top, followed by cheese/cheese spread, whole/chocolate milk, fried chicken, and Mexican dishes. Although the order of the major contributors are different among ethnicities, within each ethnic group, the order for cases and controls are very similar, suggesting that the observed associations within ethnicity were not due to the different dietary source of fat, but the amount of intake. In our analysis using the multivariate nutrient density model, the effect of total fat is mostly attributable to oleic acid, some to linoleic acid, but not to saturated fat. After mutual adjustment for types of fat, similar effects still exist for oleic acid, but 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. not for linoleic acid anymore. Oleic acid is the major type of monounsaturated fat in the diet. A lack of effect of dietary oleic acid was reported in animal studies (Fay et al., 1997; Welsch, 1992). The positive association between monounsaturated fats and breast cancer, though not supported by animal studies, has been observed in some epidemiologic studies. Monounsaturated fats have been observed to be associated with increased risk of breast cancer in case-control studies conducted in France (highest vs. lowest tertile OR 1.7; 95% Cl: 1.2-2.5) (Richardson et al., 1991), Greece (90th vs. 10th centile OR 1.9; 95% Cl: 1.1-3.2) (Katsouyanni et al., 1988), Uruguay (highest vs. lowest quartile OR 2.5; 95% Cl: 1.5-4.1) (De Stefani et al., 1998), and China (highest vs. lowest quintile OR 1.9; 95% Cl: 1.1-3.2 (Yu et al., 1990); highest vs. lowest quartile OR 3.1; 95% Cl: 1.5-6.7; (Qi et al., 1994)). A meta-analysis of 16 case-control studies published between 1978-1991 also found a positive association between monounsaturated fat and increased risk of breast cancer (highest vs. lowest level OR: 1.4; 95% Cl: 1.2-1.7) (Boyd et al., 1993). Similar associations were also found in several cohort studies conducted in Finland (highest vs. lowest quartile OR 2.7; 95% Cl: 1.0-7.4) (Knekt et al., 1990), Canada (highest vs. lowest quartile OR 1.2; 95% Cl: 0.8-1.9, P trend <0.05) (Howe et al., 1991), Norway (highest vs. lowest quartile OR 1.7; 95% Cl: 1.2-2.5) (Gaard et al., 1995) and US (postmenopausal women without benign breast disease, highest vs. lowest quintile OR 1.8; 95% Cl: 0.9-3.7, P trend <0.05) (Velie et al., 2000). However, in the meta-analysis of 8 cohort studies, monounsaturated fats were not related to breast cancer risk (Hunter et al., 1996). 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. There are some case-control studies that found no significant association between monounsaturated fats and breast cancer, including studies conducted in Australia (Ingram et al., 1991; Rohan et al., 1988)); Singapore (Lee et al., 1991), Russia (Zaridze et al., 1991), Spain (Martin-Moreno et al., 1994), China (Yuan et al., 1995), Italy (Ferraroni et al., 1991; La Vecchia et al., 1998; La Vecchia et al., 1995; Sieri et al., 2002), England (Cade et al., 1998) and US (Goodstine et al., 2003). Some cohort studies including the NHANES study (Jones et al., 1987), the Iowa Women’s health study (Kushi et al., 1992), the Nurse’s health study (Willett et al., 1992) (Byrne et al., 2002; Holmes et al., 1999), the NYU women’s health study (Toniolo et al., 1994) also did not find a significant association. Monounsaturated fats were found to be protective among postmenopausal women in the Netherlands cohort study (highest vs. lowest quintile OR 0.6; 95% Cl: 0.4-1.0, p trend < 0.05) (Voorrips et al., 2002), the Swedish Mammography Screening cohort (per lOg/d increment RR: 0.5, 95% Cl: 0.2- 1.0) (Wolk et al., 1998), and a case-control study conducted at Los Angeles (highest vs. lowest quartile OR 0.5; 95% Cl: 0.2-1.0, p trend <0.05) (Witte et al., 1997). While the literature contains inconsistent findings for the relationship between monounsaturated fats and breast cancer, in our study, overall oleic acid was positively associated with the risk, although this finding was not consistent across ethnicity. We observed the positive association among whites and foreign-born Latinas, but not among African-Americans and US-bom Latinas. These discrepancies are hard to explain by bias related to data collection. It may imply that the observed association is not due to oleic acid per se. As one of the most common fats in diet, oleic acid is 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. substantially present in a variety of fats of both plant and animal origin. It composes 40-50% of animal adipose fat, 20-40% of animal structural fats, 30% of dairy fats, 20- 40% of most commonly used cooking oils, and is highest in olive oil (>70%). In the American diet, because of the high intake of animal and dairy products, oleic acid is highly correlated with saturated fats and total fat intake. In our study, the correlation coefficient was 0.75 between oleic acid and saturated fats, and 0.93 between oleic acid and total fat intake. Therefore it is difficult to separate the effect of monounsaturated fats from other types of fats, especially saturated fats. Mutual adjustment for types of fat simultaneously in the model, however, might also attenuate the precision of estimations, due to the near collinearity among types of fats, which has been a common concern of many studies (Cho et al., 2003; Smith-Wamer et al., 2001). Besides, it has been showed that the measurement errors in dietary data are usually systematic, correlated with true nutrient intakes and with each other, heteroscedastic, and non-normally distributed, leading to the distortion of the main effect of one exposure when another poorly measured adjusting covariate is added in the model (Kipnis et al., 1997). If such measurement errors also existed among types of fat intake in this study, mutual adjustment may not have improved the estimate, or even may have worsened the estimate, although there are no simple rules to judge the direction and magnitude of the bias. A better understanding of the measurement error properties of the instruments that are being used will help to correct the impact of these errors. I ll Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In this study, we also evaluated the association between the choice of cooking fat and breast cancer risk. One advantage of this analysis is that it does not suffer from the collinearity among the types of fat, since subjects were allowed to name their most commonly used cooking fat. It is also less vulnerable to recall bias, since most people habitually use a certain type of cooking fat. The results showed that, compared with using vegetable oil and com oil, or hydrogenated fats, using olive or canola oil had protective effects for breast cancer. This result was quite consistent across ethnicities, although the most often used cooking oil was very diverse among ethnicities. This result is also consistent with other studies that reported protective effects of olive oil (La Vecchia et al., 1995; Martin-Moreno et al., 1994; Trichopoulou et al., 1995). Since results from our FFQ data showed there was not a protective effect, but rather a positive association between oleic acid and breast cancer, our cooking oil results also support the idea that the protective effect of olive and canola oil is due to other specific components in olive and canola oil, rather than the monounsaturated fat. Since canola oil also contains a relatively high level (>10%) of alpha-linolenic acid (an n-3 polyunsaturated fat), and olive oil also contains vitamins, flavanoids, and phenolic compounds, these components may be protective for the development of breast cancer. However, in this study, we could not separate the effects of olive and canola oil. Frequently using olive or canola oil also reduces the use of other oils, such as vegetable and com oils that are rich in linoleic acid, which has been suggested to have mammary tumorigenic effects. 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Several non-dietary breast cancer risk factors have been suggested to modify the association between dietary fat and breast cancer. The positive association between dietary fat and risk was found to be stronger among postmenopausal women than premenopausal women in a meta-analysis (Howe et al., 1990). In this study, we did not find a consistently significant interaction between menopausal status and total fat or types of fat. The significant interactions between linoleic acid and menopausal status among African-Americans, and between saturated fat and menopausal status among US-bom Latinas were not consistent across ethnicity and were probably found by chance. Some studies also examined the association stratified on history of benign breast disease. A pooled analysis of prospective studies reported a marginally significant interaction between total fat intake and history of BBD, with fat intake being associated with risk among women with a previous history of BBD (Hunter et al., 1997), while in a recent cohort study, total fat and oleic acid were found to be associated with breast cancer among postmenopausal women with no history of BBD (Velie et al., 2000). Some other studies found no influence of history of BBD on the association between breast cancer and total fat or types of fat (Byrne et al., 2002; Cho et al., 2003; Smith-Wamer et al., 2001). In this study, we observed a significant interaction between linoleic acid and history of benign breast disease, with linoleic acid associated with increased risk of breast cancer among women with previous history of BBD. This effect was mainly attributable to African-Americans, and probably some to Latinas, but not whites. Whether this interaction was real or found by chance needs to be confirmed by other studies. 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. While some studies suggest that the association between dietary fat and breast cancer may differ by ER/PR status, with a positive association between dietary fat and breast cancer stronger among women with cancer that were positive for ER/PR (Cho et al., 2003; Kushi et al., 1995), we did not find any modifying effect for ER or PR status in this study. However, we did find that the association between saturated fat and breast cancer was different for localized disease and disease with regional or remote extension. Saturated fat showed a protective effect for localized cancer, but not for advanced cancer. This association existed in all ethnic groups except African- Americans. Saturated fat has been found to be negatively associated with breast cancer in some studies (Jones et al., 1987; Witte et al., 1997), while most other studies either reported a positive association (De Stefani et al., 1998; Goodman et al., 1992; La Vecchia et al., 1998; Qi et al., 1994; Richardson et al., 1991; Ronco et al., 1996; Toniolo et al., 1989; van den Brandt et al., 1993; Voorrips et al., 2002), or no association (Byrne et al., 2002; Ferraroni et al., 1991; Gaard et al., 1995; Goodstine et al., 2003; Graham et al., 1991; Holmes et al., 1999; Howe et al., 1991; Ingram et al., 1991; Knekt et al., 1990; Kushi et al., 1992; Lee et al., 1991; Martin-Moreno et al., 1994; Rohan et al., 1988; Sieri et al., 2002; Toniolo et al., 1994; Velie et al., 2000; Willett et al., 1992; Wolk et al., 1998; Yu et al., 1990; Yuan et al., 1995; Zaridze et al., 1991). Although our results are not supported by most of the literature, more studies are expected to investigate whether there is any interaction between breast cancer stage and saturated fat. 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. While our study could not avoid the common limitations of case-control studies, such as recall bias from the FFQ and measurement errors when estimating amounts of fatty acids from food items, these limitations are unlikely to explain the association we observed between oleic acid and the increased risk of breast cancer, or the discrepancies across ethnicities. In conclusion, in this population-based multi-ethnic case-control study, we found a positive association between total fat intake and breast cancer risk. Among types of fat, oleic acid was the most strongly associated with increased risk of breast cancer. However, these associations were not consistent across ethnicity. We also found a protective effect for olive or canola oil. While we are continuing to explore the factors responsible for the discrepancies among ethnicities, we also expect more reports that involve multiple ethnicities to further investigate the association between dietary fat and breast cancer risk. 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4 The 5-LOX Gene Polymorphisms and Breast Cancer Risk 4.4.1 Allelic Distribution and Frequencies of the 5-LOX Gene Polymorphisms Several polymorphisms in the 5-LOX gene were first reported by In et al in 1997 (In et al., 1997). Among these polymorphisms the promoter region Spl binding site 6-base pair (-GGGCGG-) addition/deletion polymorphism raised interest due to its possible significance in gene transcriptional regulation and became the focus of all later functional and association studies of 5-LOX gene. To avoid missing any other potential functionally important polymorphisms on the 5-LOX gene, we sequenced the 5-LOX gene coding region and 5’ regulatory region (as far as upstream -1750) to search for other polymorphisms. In the 5’ regulatory region, besides the Spl binding site polymorphism, we also identified 6 other common polymorphisms, i.e. - 1708G>A, -1368 (GTTAAA) deletion, -1286G>T, -845G>T, -560T>C and -59C>T. Sequencing of the 5-LOX gene from 50 subjects showed that these 6 polymorphisms are in complete linkage disequilibrium, forming two common haplotypes: (1) - 1708G/-1368normal/-1286G/-845G/-560T/-59C, and (2) -1708A/-1368(GTTAAA) deletion/-1286T/-845A/-560C/-59T (figure 4.4.1). In this study we only genotyped one of the polymorphisms, -1286G>T, among all 1694 subjects to represent the haplotype. The allelic frequencies of the -1286G>T and the Spl binding site polymorphisms among whites, African-Americans and Latinas were summarized in table 4.4.1. Genotype frequencies for both polymorphisms were in Hardy-Weinberg equilibrium among control subjects of all three ethnicities. Allele distribution of the Spl binding site polymorphism among white controls was also similar to a previous 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. report for a Caucasian population in the United Kingdom (Sayers et al., 2003). In this study, Latinas have an allele distribution similar to whites, while African-Americans had a much higher frequency for the 12-bp deletion (3 Spl motif) than whites and Latinas. Among whites and Latinas, -1286T was in nearly complete linkage disequilibrium with the 6-bp deletion (4 Spl motif), while -1286G was linked to wild- type (5 Spl motif) and all other polymorphisms except 6-bp deletion (table 4.4.2). However, among African-Americans, a small percentage of-1286G was also linked to 6-bp deletion and a small percentage of-1286T linked to 12-bp deletion. -1708G>A -1286G>T I I -560T>C 1 -59C>T 1 r / 760 G>A A w _______ A -1368 6-bp del Haplotype -1708 -1286 r . r . \ -845G >A -845 c A l Sp1 binding site 6-bp add/del -560 -59 t r ______l / Exon 1 / 1 ---- 1 / Exon 6 r T A ^ T V Figure 4.4.1 The 5-LOX gene 5’ regulatory region and coding region polymorphisms Table 4.4.1 Allele frequencies of the 5-LOX gene polymorphisms among three ethnic groups 5-LOX gene Allele frequency polymorphisms White African American Latina Case Control Case Control Case Control -1286G/T G 0.85 0.84 0.84 0.83 0.85 0.89 T 0.15 0.16 0.16 0.17 0.15 0.11 5’ Spl binding site 18-bp del (2) 0 0.002 0 0 0 0 12-bp del (3) 0.004 0.009 0.33 0.32 0.03 0.03 6-bp del (4) 0.16 0.16 0.14 0.15 0.15 0.11 wild-type (5) 0.83 0.82 0.49 0.50 0.79 0.82 6-bp add (6) 0.006 0.009 0.03 0.02 0.03 0.03 12-bp add (7) 0.006 0.002 0.01 0.004 0.004 0.007 760G/A G >0.99 >0.99 0.92 0.93 >0.99 >0.99 A <0.01 <0.01 0.08 0.07 <0.01 <0.01 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.4.2 The 5-LOX gene -1286G/T and Spl binding site haplotype frequencies among three ethnic groups Haplotype White African- American Latina -1286G/Spl(3) 0.009 0.276 0.024 -1286G/Spl(4) 0.003 0.017 0.006 -1286G/Spl(5) 0.816 0.504 0.824 -1286G/Spl(6) 0.009 0.024 0.030 -1286T/Spl(3) 0 0.041 0.004 -1286T/Spl(4) 0.160 0.134 0.104 -1286T/Spl(5) 0 0 0 The only nonsynonymous polymorphism we found in the 5-LOX gene is 760G>A (Glu254Lys) in exon 6. This polymorphism was very rare (<1%) among whites and Latinas. Among Afican-Americans, the percentage of the rare allele 760A was about 7%. 4.4.2 The 5-LOX Gene Spl Binding Site Polymorphism and Breast Cancer Risk For whites, the only common 5-LOX gene Spl binding site genotypes were 5/5 and 4/5 (table 4.4.3). For Latinas, genotypes 3/5 and 5/6 were also present at a frequency of approximately 5%. For African-Americans, genotypes containing allele “3” (3/3 and 3/5) were even more common than genotypes containing allele “4” (4/4 and 4/5). Using the wild-type (5/5) as the referent category, crude odds ratio and 95% confidence interval were calculated for each of the major genotypes and for collapsed genotypes according to the number of wild-type (5) allele (table 4.4.3). A marginally significant positive association was observed only for genotype 4/5 among Latinas (Crude OR: 1.49; 95% Cl: 1.00-2.23). Among African-Americans, genotypes with one wild-type (5) allele (3/5 and 4/5) also appeared to be positively associated with breast 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cancer, though the association was not significant. Because the genotypes with both alleles as non-wild-type were sparse for whites and Latinas, we collapsed all non-5/5 genotypes together as one group in further analyses. This way of analysis is similar to other publications studying this polymorphism. Overall there was no association between Spl binding site polymorphisms and breast cancer risk among whites, but a non-significant positive association between non-5/5 genotypes and breast cancer risk among Latinas (Adjusted OR: 1.38, 95% Cl: 0.97-1.95) and African-Americans (Adjusted OR: 1.33, 95% Cl: 0.87-2.04) (table 4.4.4). When we excluded women with family history of breast cancer, the association became stronger and statistically significant among Latinas (Adjusted OR: 1.47, 95% Cl: 1.01-2.12), and remained unchanged in other groups (table 4.4.5). When analyzing the association according to the history of benign breast disease (BBD), we found that the association between non-5/5 and increased risk appeared to be stronger among Latina and African-American women with previous history of BBD than women without BBD (table 4.4.6). However, among white women with history of BBD, the non-5/5 genotypes showed an opposite significant protective effect (Adjusted OR: 0.34, 95% Cl: 0.14-0.86). Tests for interaction were not significant because the number of women with BBD was relatively small. We also examined the association among pre- and postmenopausal women separately, but found no significant difference by menopausal status (table 4.4.7). The association with Spl binding site genotypes also did not differ by stage or histological grade of breast cancer (table 4.4.8). 119 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 4.4.3 The 5-LOX gene Spl binding site genotype distribution and odds ratio (95% Cl) among three ethnic groups Genotype White African-American Latina Cases (%) Controls (%) OR (95% Cl) Cases (%) Controls (%) OR (95% Cl) Cases (%) Controls (%) OR (95% Cl) 5/5 (WT) 185 (67.3) 193 (65.9) Ref (1.0) 57 (23.3) 70 (28.1) Ref (1.0) 176 (61.8) 238 (68.6) Ref(l.O) 2/5 0 1 (0.34) N/A 0 0 N/A 0 0 N/A 3/3 0 0 N/A 27(11.0) 29(11.7) 1.14 (0.61-2.15) 1 (0.35) 1 (0.29) N/A 3/4 0 1 (0.36) N/A 21 (8.57) 26 (10.4) 0.99 (0.51-1.95) 2 (0.70) 3 (0.86) N/A 3/5 1 (0.36) 4(1.37) N/A 79 (32.2) 70 (28.1) 1.39 (0.86-2.23) 13 (4.56) 14 (4.03) 1.26 (0.58-2.74) 3/6 1 (0.36) 0 N/A 4(1.63) 3 (1.20) N/A 0 1 (0.29) N/A 3/7 0 0 N/A 2 (0.82) 1 (0.40) N/A 0 0 N/A 4/4 1 (0.36) 5(1.71) N/A 5 (2.04) 6(2.41) N/A 8(2.81) 6(1.73) N/A 4/5 82 (28.3) 83 (28.3) 1.03 (0.72-1.49) 37(15.1) 35 (14.1) 1.30 (0.73-2.32) 65 (22.8) 59 (17.0) 1.49(1.00-2.23) 4/6 0 2 (0.68) N/A 2 (0.82) 2 (0.80) N/A 2 (0.70) 1 (0.29) N/A 4/7 3 (1.09) 0 N/A 1 (0.41) 0 N/A 0 1 (0.29) N/A 5/6 2 (0.73) 3 (1.02) N/A 6 (2.45) 5 (2.01) N/A 16(5.61) 19 (5.48) 1.14(0.57-2.28) 5/7 0 1 (0.36) N/A 3 (1.22) 1 (0.40) N/A 2 (0.70) 4(1.15) N/A 6/6 0 0 N/A 1 (0.41) 1 (0.40) N/A 0 0 N/A 5/other 85 (30.9) 92 (31.4) 0.96 (0.67-1.38) 125 (51.0) 111 (44.6) 1.38(0.90-2.13) 96 (33.7) 96 (27.7) 1.35 (0.96-1.91) other/other 5(1.8) 8 (2.7) 0.65 (0.21-2.03) 63 (25.7) 68 (27.3) 1.14(0.70-1.86) 13 (4.6) 13 (3.8) 1.35 (0.61-2.99) to o Table 4.4.4 The 5-LOX gene Spl binding site genotype and breast cancer risk Spl binding Cases (%) Controls OR Adj. OR* site genotype (%) (95% Cl) (95% Cl) White 5/5 185 (67.3) 193 Ref (1.0) Ref (1.0) non-5/5 90 (32.7) (65.9) 100 (34.1) 0.94 (0.66-1.33) 0.95 (0.66-1.37) Afr-Am 5/5 57 (23.3) 70 (28.1) Ref (1.0) Ref (1.0) non-5/5 188 (76.7) 179 (71.9) 1.29 (0.86-1.93) 1.33 (0.87-2.04) Latina 5/5 176 (61.8) 238 Ref (1.0) Ref (1.0) non-5/5 109 (38.3) (68.6) 1.35 (0.97-1.88) 1.38 (0.97-1.95) 109 _________________________ (31i)___________________________ * Adjusted for age (continuous), ethnicity, menopausal status (pre-, postmenopausal, undetermined), age at first full-term pregnancy (<20, 20-24, 25-29, 30+), BMI (continuous), height (continuous), family history o f breast cancer (yes/no), age at menarche (continuous), foreign bom (yes/no), education (<12y, 12y, 13-15y, 16y and up), history o f benign breast disease (yes/no), total physical activity (continuous), alcohol intake (continuous). (Same for all tables in section 4.4). Table 4.4.5 The 5-LOX gene Spl binding site genotype and breast cancer risk among women without family history of breast cancer Spl binding site genotype Cases (%) Controls (%) OR (95% Cl) Adj. OR* (95% Cl) White 5/5 164 (65.9) 146 (67.0) Ref (1.0) Ref (1.0) non-5/5 85 (34.1) 72 (33.0) 0.95 (0.65-1.40) 0.92 (0.61-1.38) Afr-Am 5/5 61 (28.1) 48 (23.1) Ref (1.0) Ref (1.0) non-5/5 156 (71.9) 160 (76.9) 1.30 (0.86-1.93) 1.34 (0.84-2.14) Latina 5/5 220 (69.4) 154 (61.6) Ref (1.0) Ref (1.0) non-5/5 97 (30.6) 96 (38.4) 1.41 (1.00-2.01) 1.47(1.01-2.12) 121 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.4.6 The 5-LOX gene Spl binding site polymorphism and breast cancer risk, by history of benign breast disease (BBD) Without history of BBD With history of BBD Spl binding Cases Controls Adj. OR* Cases Controls Adj. OR* P for site genotype (95% Cl) (95% Cl) Int. White 5/5 126 162 Ref (1.0) 59 31 Ref (1.0) non-5/5 61 77 1.08(0.71-1.65) 29 23 0.34 (0.14-0.86) 0.24 Afr-Am 5/5 45 55 Ref (1.0) 12 15 Ref (1.0) non-5/5 138 153 1.10(0.67-1.80) 50 25 2.19(0.71-6.71) 0.11 Latina 5/5 153 206 Ref (1.0) 23 32 Ref (1.0) non-5/5 85 95 1.21 (0.83-1.77) 24 14 3.49(1.26-9.65) 0.09 Table 4.4.7 The 5-LOX gene Spl binding site polymorphism and breast cancer risk, by menopausal status Premenopausal Postmenopausal Spl binding Cases Controls Adj. OR* Cases Controls Adj. OR* P for site genotype (95% Cl) (95% Cl) Int. White 5/5 46 62 Ref (1.0) 126 120 R ef (1.0) non-5/5 24 26 1.28 (0.59-2.77) 60 68 0.85 (0.54-1.33) 0.28 Afr-Am 5/5 12 23 Ref (1.0) 39 41 Ref (1.0) non-5/5 63 58 2.07 (0.88-4.90) 118 109 1.13 (0.66-1.96) 0.22 Latina 5/5 60 80 Ref (1.0) 98 136 R ef (1.0) non-5/5 32 45 0.84 (0.44-1.61) 67 57 1.61 (1.01-2.56) 0.24 Table 4.4.8 The 5-LOX gene Spl binding site polymorphism and breast cancer risk, by progression of cancer Spl binding Controls Cases (non Adj. OR* Cases Adj. OR* P for site genotype progressive)1 (95% Cl)i (progressive) (95% Cl) Int. White 5/5 193 88 Ref (1.0) 74 Ref (1.0) non-5/5 100 48 1.08 (0.69-1.69) 36 0.96 (0.58-1.57) 0.66 Afr-Am 5/5 70 19 Ref (1.0) 31 Ref (1.0) non-5/5 179 54 1.02(0.54-1.94) 110 1.51 (0.91-2.52) 0.34 Latina 5/5 238 55 Ref (1.0) 101 Ref (1.0) non-5/5 109 34 1.37 (0.82-2.30) 65 1.44 (0.96-2.15) 0.86 1. Non-progressive cases: localized and historical grade 1 or 2; 2. Progressive cases: non-localized cases or histological grade 3 or 4. 122 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Dietary linoleic acid is the major source for arachidonic acid, the substrate of 5-LOX. We stratified the analysis by linoleic acid intake to examine whether the association between Spl binding site genotype and breast cancer is modified by the levels of linoleic acid intake. We found that among both Latinas and African- Americans, compared to women carrying the 5/5 genotype and taking less linoleic acid (<median), those carrying non-5/5 genotypes and with high linoleic intake (>median) had a significantly increased risk of breast cancer (table 4.4.9). However, the same analyses showed an opposite association among whites, due to no main effect of linoleic acid in whites. Carrying non-5/5 genotypes and with high linoleic intake (>median) showed a negative association with breast cancer risk among whites. Interaction tests showed no significant interactions between linoleic acid intake and 5- LOX gene Spl binding site polymorphism among any of these ethnicities. Because our functional study and some other functional studies showed that the 5-LOX gene promoter carrying wild type or addition variants have stronger transcriptional activities than deletion variants (In et al., 1997; Silverman et al., 1998), we also conducted the above analyses with two other methods: (1) by categorizing the allele into long (wild-type or addition variants) and short (deletion variants), and analyzing the genotype as long-long versus long/short and short-short; (2) by scoring each genotype by averaging the number of Spl binding motif on two alleles (e.g. genotype 4/5 had score (4+5)/2=4.5), and putting this score into the logistic regression model as a continuous variable. Because the addition polymorphism is uncommon among all ethnicities, results from these two alternative methods were similar to the 123 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. results from the commonly used methods (5/5 versus others) (data not shown). Since the non-5/5 genotype group was more heterogeneous among African-Americans (table 4.4.3), we also analyzed each of the major genotypes (3/3, 3/4, 3/5, 4/5) separately with further adjustment and stratification. Results were also similar to the crude analysis (data not shown). In addition, because -15% of African-Americans also carry a nonsynonymous polymorphism (760 G>A, see below), we repeated the above analyses by excluding those carrying this polymorphism and the results did not change (data not shown). Table 4.4.9 The 5-LOX gene Spl binding site polymorphism and breast cancer ____________ risk, by linoleic acid (LA) intake______________________________ Race Spl binding site genotype Cases Controls Adj. OR** (95% Cl) Pfor interaction Whites LA<median 5/5 88 76 Ref (1.0) non-5/5 39 33 1.16(0.64-2.08) LA >median 5/5 97 117 0.63 (0.39-1.02) non-5/5 51 67 0.53 (0.31-0.91) 0.38 African- LA<median Americans 5/5 20 29 Ref (1.0) non-5/5 55 69 1.22 (0.60-2.46) LA >median 5/5 37 41 1.62 (0.72-3.62) non-5/5 133 110 2.31 (1.13-4.71) 0.73 Latinas LA<median 5/5 105 159 Ref (1.0) non-5/5 56 78 1.13(0.72-1.76) LA >median 5/5 71 79 1.09 (0.69-1.74) non-5/5 53 31 2.04(1.16-3.58) 0.18 ** Besides the covariates adjusted in *, total energy intake, saturated fat intake, oleic acid intake were also adjusted in the model. 124 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4.3 The 5-LOX Gene -1286G>T Polymorphism and Breast Cancer Risk Because the T allele of-1286G>T polymorphism was in nearly complete LD with the Spl binding site 6-bp deletion (4) among whites and Latinas, and also because the 6-bp deletion is the most common non-wild-type polymorphism at the Spl binding site among these two ethnicities, the results for -1286G>T polymorphism analysis were very close to the results we reported for Spl binding site polymorphisms. Briefly, there was no association between -1286G>T polymorphism and breast cancer among whites. However, carrying the T allele appeared to associate with increased breast cancer risk among Latinas (table 4.4.10). For African- Americans, there was less LD between these two loci, thus results differed from those observed for Spl binding site polymorphism. We did not observe an association between -1286T allele and breast cancer. Neither was an association found when we stratified analysis on family history of breast cancer, history of benign breast disease, menopausal status or linoleic acid intake (table 4.4.11). Table 4.4.10 The 5-LOX gene -1286G>T polymorphism and breast cancer risk -1286G >T genotype Cases (% ) Controls (%) OR (95% C l) Adj. OR* (95% C l) W hite GG 191 (69.5) 203 (69.3) R ef (1.0) R ef (1.0) GT 83 (30.2) 86 (29.4) Tx 1 (0.4) 4 (1.4) GT or TT 84 (30.6) 90 (30.7) 0.99 (0.69-1.42) 1.00(0.68-1.45) Afr-Am GG 169 (69.0) 171 (68.7) R ef (1.0) R ef (1.0) GT 72 (29.4) 69 (27.7) TT 4 (1.6) 9 (3.6) GT or TT 76 (31.0) 78 (31.3) 0.98 (0.67-1.44) 0.96 (0.64-1.43) Latina GG 210 (73.7) 277 (79.8) R ef (1.0) R ef (1.0) GT 67 (23.5) 65 (18.7) TT 8 (2.8) 5 (1.4) GT or TT 75 (26.3) 70 (20.2) 1.42 (0.97-2.05) 1.37(0.93-2.02) 125 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.4.4 The 5-LOX Gene Haplotype Analysis We conducted a haplotype analysis covering both -1286G>T and Spl binding site polymorphisms. Because of the nearly complete LD between these two loci among whites and Latinas, -1286G/Spl(5) and -1286T/Spl(4) were the only common haplotypes among these two ethnic groups (table 4.4.2) and the haplotype analysis gave very similar results to that from genotype analysis for each of these two loci (data not shown). Among African-Americans, besides above two common haplotypes, the haplotype -1286G/Spl(3) was also very common due to the prevalence of Spl(3). Haplotype analysis among African-Americans didn’t show significant association between any of these major haplotypes and cancer risk, though compared to carrying two -1286G/Spl(5) haplotypes, carrying one -1286G/Spl(5) with one -1286G/Spl(3) or -1286T/Spl(4) was probably associated with increased risk (table 4.4.12). Table 4.4.11 The 5-LOX gene -1286G>T polymorphism and breast cancer risk ____________ among African-Americans, stratified by levels of other cofactors -1286G>T genotype Cases Controls Adj. OR* (95% Cl) P fo r interaction Without FHBC GG 148 154 Ref (1.0) GT or TT 60 63 0.97 (0.62-1.51) Without BBD GG 129 140 R ef (1.0) GT or TT 54 68 0.87 (0.55-1.36) With BBD GG 40 31 Ref (1.0) GT or TT 22 9 1.03 (0.35-3.06) P=0.11 Premenopausal GG 49 60 Ref (1.0) GT or TT 26 21 1.60 (0.74-3.46) Postmenopausal GG 109 99 Ref (1.0) GT or TT 48 51 0.76 (0.45-1.29) P=0.15 LA <median GG 55 68 Ref (1.0)** GT or TT 20 30 0.82 (0.40-1.67) >median GG 114 103 1.73(1.01-2.96) GT or TT 56 48 1.71 (0.90-3.16) P=0.57 126 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.4.12 The 5-LOX gene 5’ regulatory region haplotype and breast cancer risk among African-Americans Major haplotypes1 OR (95% Cl) Adj. OR* (95% Cl) -1286G-Spl(5) /-1286G-Spl(5) -1286G-Spl(5) /-1286G-Spl(3) -1286G-Spl(5) /-1286T-Spl(4) -1286G-Spl(3) /-1286G-Spl(3) -1286G-Spl(3) /-1286T-Spl(4) Ref (1.0) 1.41 (0.86-2.30) 1.40 (0.76-2.57) 1.17(0.59-2.34) 1.06 (0.52-2.16) Ref (1.0) 1.56 (0.93-2.61) 1.44 (0.75-2.77) 1.22 (0.58-2.55) 0.98 (0.46-2.10) * Other rare haplotypes (<5%) include: -1286T-Spl(4) /-1286T-Spl(4), and one o f - 1286G-Spl(5), -1286G-Spl(3), or -1286T-Spl(4), with one of non-1286G -Spl(5), ono-1286G -Spl(3) or non--1286T-Spl(4). 4.4.5 The 5-LOX Gene 760G>A Polymorphism and Breast Cancer Risk among African-Americans Since 760 G>A polymorphism is only common among African-Americans (7%), we conducted an association analysis with respect to this polymorphism and breast cancer risk only among African-American women. Overall we did not find an association between this polymorphism and breast cancer (table 4.4.13). There was a hint that carrying the A allele (the rare allele) was probably associated with an increased risk for premenopausal, but not postmenopausal breast cancer among African-American women, though the interaction was not significant (table 4.4.14). Adjustment for other 5-LOX gene polymorphisms did not change the results (data not shown). 127 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.4.13 The 5-LOX gene 760G>A polymorphism and breast cancer risk ____________ among African-Americans_________________________________ 760G>A genotype Cases (%) Controls (%) OR Adj. OR* (95% Cl) (95% Cl) GG AG AA 208 (84.9) 35 (14.3) 2 (0.8) 214(85.9) 34(13.7) 1 (0.4) Ref (1.0) Ref (1.0) AG or AA 37(15.1) 35 (14.1) 1.09(0.66-1.79) 1.12(0.66-1.89) Table 4.4.14 The 5-LOX gene 760G>A polymorphism and breast cancer risk among African-Americans, by menopausal status 760G>A genotype Cases Controls Adj. OR* P for (95% Cl) interaction Premenopausal GG AG or AA 60 71 15 10 Ref (1.0) 1.78 (0.74-4.24) Postmenopausal GG AG or AA 137 128 20 22 Ref (1.0) 0.85(0.44-1.23) P=0.17 4.4.6 Discussion In this study we examined the association of the 5-LOX gene polymorphisms with breast cancer risk among three ethnic groups. We first comprehensively reported the 5-LOX gene 5’ polymorphism and haplotype with their frequencies among whites, African-Americans and Latinas. We found that the 5-LOX gene Spl binding site polymorphism was moderately associated with increased risk of breast cancer among Latina and probably among African-American women. These associations appeared to be stronger among Latinas and African-Americans with a previous history of benign breast disease. However, among whites, we either observed no association or an association opposite to those found among Latinas and African-Americans. 128 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5-LOX is a key enzyme in the arachidonic acid metabolic pathway that produces leukotrienes and 5-HETE. 5-LOX pathway is implicated in many pathological processes, including cancer. Studies of prostate, colon, lung and breast cancer all indicated that regulation of 5-LOX pathway is important for cell proliferation, differentiation and apoptosis. The 5-LOX gene promoter region Spl binding site 6-bp (-GGGCGG-) deletion or addition polymorphism changes the number of Spl binding motifs that is responsible for 5-LOX transcriptional activity by binding with transcriptional factors (Hoshiko et al., 1990). The functional consequence of this polymorphism has been demonstrated by in vitro studies, although the results did not completely agree with each other. So far, there are only a few studies that investigated the possible importance of this polymorphism among diseases. Three studies analyzed the association between this polymorphism and asthma occurrence or response to treatment. The study that first reported this polymorphism did not find a significant difference in the allele frequency between 25 normal and 31 asthmatic subjects (In et al., 1997). A recent study among Caucasians with familial asthma and normal controls also did not find a significant association between this polymorphism and asthma occurrence or phenotypes (Sayers et al., 2003). However, a clinical trial found that patients who do not carry a wild-type allele do not improve when treated with a drug whose mechanism of action is the inhibition of 5-LOX, implying its pharmaceutical effect on treatment response (Drazen et al., 1999). 129 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Although in vitro studies have suggested the role of the 5-LOX gene in many cancers including breast cancer, this is the first association study that investigated the association of 5-LOX gene polymorphisms and breast cancer risk. This is also the first study that reported and investigated polymorphisms other than the known Spl binding site polymorphism in the 5-LOX gene among multiple ethnicities. Because the overall functional studies suggest a diminished transcriptional activity of non-wild- type Spl polymorphism (In et al., 1997; Silverman et al., 1998), subjects carrying those polymorphisms should have diminished 5-LOX gene transcription and consequently reduced production of 5-LOX metabolites, such as 5-HETE and leukotrienes. Since 5-LOX and its metabolites have been implicated in promotion of breast cancer cells growth, by hypothesis, subjects that have reduced 5-LOX products by carrying a non-wild-type allele should have reduced risk of breast cancer. However, our results did not support this hypothesis. On the contrary, we found that non-wild-type alleles were associated with increased risk of breast cancer among Latinas. African-Americans also showed a similar association, although it was not significant. While these findings need to be examined by other studies, there are some possible explanations. First, we cannot rule out that these observed associations are due to the linkage disequilibrium with other polymorphisms that are also functionally significant. Second, although this Spl binding site has been shown to be important for the 5-LOX gene regulation, the mechanism of regulation is complicated and not well understood yet. For example, the 5-LOX gene promoter activity has been reported to be regulated by DNA methylation at this Spl binding site, which is involved in the 130 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cell-type specific expression of 5-LOX (Uhl et al., 2002). Whether the polymorphism at the Spl binding site influences DNA methylation is not known yet. In addition, the 5-LOX pathway not only produces 5-HETE and leukotrienes that have been found to play an important role in cell growth regulation and inflammatory process, 5-LOX is also involved in the production of endogenous anti-inflammatory mediators, such as lipoxins, from n-3 polyunsaturated fat (Serhan and Oliw, 2001). This mechanism is newly identified and has not been well studied yet. Therefore the overall function of 5-LOX is probably determined by the balance between pro- and anti-inflammatory mechanisms. The 5-LOX gene -1286G>T polymorphism represents a series of linked polymorphisms we identified in the 5-LOX gene 5’ regulatory region. The functional importance of these polymorphisms has not been investigated. These polymorphisms are spread throughout the 2 kb region upstream of the 5-LOX gene translation start site, where some positive and negative regulatory elements have been identified (Hoshiko et al., 1990). Therefore, the functional consequence of these polymorphisms may be different from the effects observed by examining the Spl binding site polymorphism alone. However, none of the published studies included these polymorphisms in their functional or association studies yet. The only nonsynonymous polymorphism in the 5-LOX gene, 760G>A, changes glutamate254 to lysine254- The rare allele A has a frequency of about 7% among African-American, 3.3% among Asians (unpublished data), and less than 1% among whites and Latinas. Functional consequences of this polymorphism have not 131 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. yet been investigated. In our association analysis, we did not find this polymorphism related to breast cancer, although an association between the A allele and an increased risk of premenopausal breast cancer cannot be excluded. In summary, our analysis suggests an association between 5-LOX gene polymorphisms and breast cancer risk. However, since the association is not consistent among all ethnicities, the possibility that the association is found by chance cannot be excluded. It is possible that this inconsistency is due to some unidentified risk factors that differ among these ethnicities. Actually we did find that white women in this study differed from African-Americans and Latinas regarding the association between breast cancer risk and some well-known risk factors, such as age at menarche, age at fist full-term pregnancy, education, etc. These factors showed associations among African-Americans and Latinas, but not among whites, suggesting that some unknown risk factors might play important roles in breast cancer for whites. Lack of adjustment for the unknown factors may be partly responsible for the inconsistency we found among the ethnicities. 132 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.5 The 12-LOX Gene Polymorphisms and Breast Cancer Risk 4.5.1 Frequencies of the 12-LOX Gene Polymorphisms and Haplotypes among Three Ethnicities Two common nonsynonymous SNPs, 782G>A (Arg261Gln) and 965A>G (Asn322Ser), have been reported by the dbSNP database (rs 1126667 and rs434473). The allele frequencies of these two polymorphisms among whites, African-Americans and Latinas in our study were summarized in table 4.5.1. African-Americans and Latinas had very similar allele frequencies for 782G>A (G:0.64/A:0.36), which were different from whites (G:0.56/A:0.44). For 965A>G polymorphism, the common allele A had a frequency of 0.56 for whites, 0.76 for African-Americans, and 0.73 for Latinas. These two loci were in nearly complete linkage disequilibrium among whites, forming two common haplotypes, 782G/965A (0.56) and 782A/965G (0.44). Among African-Americans and Latinas these two loci were in incomplete LD, therefore a third haplotype, 782A/965A also existed, with a frequency of 0.12 for African-Americans and 0.09 for Latinas. All 12-LOX genotype frequencies among controls of three ethnicities were in Hardy-Weinberg equilibrium. 4.5.2 The 12-LOX Gene Polymorphisms and Breast Cancer Risk We analyzed the associations between these two 12-LOX gene polymorphisms and breast cancer among three ethnic groups (table 4.5.2). Overall there was no association between the 782G>A polymorphism and breast cancer risk among whites and Latinas. However, we found that carrying the A allele was significantly associated with decreased risk of breast cancer among African-American women (Adjusted 133 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 0R ag: 0.77, 95% Cl: 0.52-1.14; ORaa: 0.43, 95% Cl: 0.21-0.88; P trend = 0.02). Analysis of the 965A>G polymorphism showed similar results. Carrying the G allele was associated with decreased risk of breast cancer only among African-Americans (Adjusted ORA G : 0.76, 95% Cl: 0.51-1.13; ORaa: 0.33, 95% Cl: 0.11-0.98; P trend = 0.03), but not among whites and Latinas. Results were similar when we excluded subjects with a family history of breast cancer (table 4.5.3). The association between these two polymorphisms and breast cancer did not significantly differ by previous history of benign breast disease (BBD). But carrying a 965G allele was associated with an increased risk of breast cancer among Latinas with history of BBD, which was opposite to the association we observed for African-American women (table 4.5.4). Table 4.5.1 Allele and haplotype frequencies of the 12-LOX gene ____________ polymorphisms among three ethnic groups_____________________ 12-LOX gene Allele frequency polymorphism White African-American Latina 782G>A G 0.56 0.64 0.64 A 0.44 0.36 0.36 965A>G A 0.56 0.76 0.73 G 0.44 0.24 0.27 Haplotype 782G/965A 0.56 0.64 0.64 782A/965G 0.44 0.24 0.27 782A/965A 0.002 0.12 0.09 782G/965G <0.001 <0.001 <0.001 When we analyzed the 12-LOX gene polymorphisms and breast cancer according to menopausal status, we did not find a significant difference by menopausal status. However, overall the 782A and 965G alleles were protective only 134 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. among post-menopausal women, again, only in African-Americans, whereas there was a hint of positive association with breast cancer among premenopausal white and Latinas (table 4.5.5). Table 4.5.2 The 12-LOX gene polymorphisms and breast cancer risk among three ethnic groups 12-LOX Genotype Cases Controls OR (95% Cl) P trend Adj. OR* (95% Cl) P trend 782G>A White GG AG AA 87 (31.6) 134 (48.3) 54 (19.6) 95 (32.4) 137 (46.8) 61 (20.8) Ref (1.0) 1.07 (0.73-1.56) 0.97 (0.61-1.54) 0.95 Ref (1.0) 1.07 (0.73-1.59) 1.01 (0.62-1.65) 0.93 Afr-Am GG AG AA 120 (49.0) 109 (44.5) 16(6.5) 98 (39.4) 122 (49.0) 29(11.7) Ref (1.0) 0.73 (0.50-1.06) 0.45 (0.23-0.88) 0.01 Ref (1.0) 0.77 (0.52-1.14) 0.43 (0.21-0.88) 0.02 Latina GG AG AA 120(42.1) 115(40.4) 50(17.5) 148 (42.7) 147 (42.4) 52(15.0) Ref (1.0) 0.97 (0.69-1.36) 1.19(0.75-1.87) 0.59 Ref (1.0) 0.90 (0.63-1.30) 1.10(0.68-1.78) 0.88 965A>G White AA AG GG 88 (32.0) 134(48.7) 53(19.3) 96 (32.8) 136 (46.4) 61 (20.8) Ref (1.0) 1.08 (0.74-1.56) 0.95 (0.59-1.51) 0.90 Ref (1.0) 1.08 (0.73-1.60) 0.98 (0.60-1.61) 0.99 Afr-Am AA AG GG 164 (66.9) 76 (31.0) 5 (2.0) 143 (57.4) 94 (37.8) 12 (4.8) Ref (1.0) 0.71 (0.48-1.03) 0.36 (0.13-1.06) 0.01 Ref (1.0) 0.76 (0.51-1.13) 0.33 (0.11-0.98) 0.03 Latina AA AG GG 138 (48.4) 113(39.7) 34(11.9) 189 (54.5) 125 (36.0) 33 (9.5) Ref (1.0) 1.24 (0.89-1.73) 1.41 (0.83-2.39) 0.12 Ref (1.0) 1.17(0.82-1.67) 1.24 (0.71-2.15) 0.33 135 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.5.3 The 12-LOX gene polymorphisms and breast cancer risk among women without family history of breast cancer 12-LOX Cases Controls OR P Adj. OR* P trend genotype (95% Cl) trend (95% Cl) 782G>A White GG 69 84 Ref (1.0) Ref (1.0) AG 103 113 1.11 (0.73-1.68) 1.16(0.75-1.80) AA 46 52 1.08(0.65-1.79) 0.73 1.12(0.65-1.92) 0.63 Afr-Am GG 99 85 Ref (1.0) Ref (1.0) AG 95 106 0.77 (0.52-1.15) 0.80 (0.52-1.22) AA 14 26 0.46 (0.23-0.94) 0.03 0.41 (0.19-0.87) 0.03 Latina GG 102 137 Ref (1.0) Ref (1.0) AG 103 134 1.03 (0.72-1.48) 0.95 (0.64-1.40) AA 45 46 1.31 (0.81-2.13) 0.33 1.21 (0.72-2.03) 0.60 965A>G White AA 70 85 Ref (1.0) Ref (1.0) AG 103 112 1.12(0.74-1.69) 1.17(0.75-1.80) GG 45 52 1.05(0.63-1.75) 0.79 1.09 (0.64-1.87) 0.68 Afr-Am AA 140 128 Ref (1.0) Ref (1.0) AG 64 78 0.75 (0.50-1.13) 0.82(0.53-1.26) GG 4 11 0.33 (0.10-1.07) 0.04 0.27 (0.08-0.89) 0.05 Latina AA 119 177 Ref (1.0) Ref (1.0) AG 100 111 1.34 (0.94-1.91) 1.26 (0.86-1.84) GG 31 29 1.59 (0.91-2.78) 0.04 1.41 (0.78-2.56) 0.15 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 4.5.4 Th(f 12-LOX gene polymorphisms and breast cancer risk, by history of benign breast disease (BBD) Without BBD With BBD 12-LOX Cases Controls Adj. OR* P Cases Controls Adj. OR* P p for genotype (95% Cl) trend (95% Cl) trend interaction 782G>A White GG 61 78 Ref (1.0) 26 17 Ref (1.0) AG 87 111 1.10(0.70-1.73) 47 26 0.85 (0.34-2.15) AA 39 50 1.11 (0.63-1.95) 0.69 15 1 1 0.62(0.20-1.95) 0.43 0.71 Afr-Am GG 87 80 Ref (1.0) 33 18 Ref (1.0) AG 84 105 0.75 (0.49-1.17) 25 16 0.67 (0.24-1.88) AA 12 23 0.48(0.21-1.09) 0.05 4 6 0.39(0.08-1.95) 0.22 0.94 Latina GG 101 125 Ref (1.0) 19 23 Ref (1.0) AG 95 130 0.81 (0.55-1.21) 20 17 1.71 (0.57-5.17) AA 42 46 0.98 (0.58-1.65) 0.69 8 6 3.16(0.70-14.3) 0.12 0.47 965A>G White AA 62 78 Ref (1.0) 26 18 Ref (1.0) AG 87 111 1.09(0.69-1.70) 47 25 0.89 (0.35-2.25) GG 38 50 1.06(0.60-1.87) 0.80 15 1 1 0.64(0.20-1.99) 0.46 0.72 Afr-Am AA 119 118 Ref (1.0) 45 25 Ref (1.0) AG 61 81 0.79 (0.51-1.23) 1 5 12 0.48(0.15-1.48) GG 3 9 0.28(0.07-1.15) 0.08 2 3 0.41 (0.05-3.20) 0.16 0.92 Latina AA 117 160 Ref (1.0) 21 29 Ref (1.0) AG 92 112 1.02(0.69-1.50) 21 1 3 3.21 (1.07-9.66) GG 29 29 1.16(0.64-2.11) 0.68 5 4 3.29 (0.55-19.7) 0.05 0.35 Reproduced w ith permission o f th e copyright owner. Further reproduction prohibited without permission Table 4.5.5 The 12-LOX gene polymorphisms and breast cancer risk, by menopausal status Premenopausal Postmenopausal 12-LOX genotype Cases Controls Adj. OR* (95% Cl) P trend Cases Controls Adj. OR* (95% Cl) P trend pfor interaction 782G>A White GG 17 30 Ref (1.0) 65 60 Ref (1.0) AG 35 42 1.26 (0.54-3.00) 87 87 1.00 (0.62-1.64) AA 18 16 2.55 (0.89-7.30) 0.09 43 41 0.79(0.43-1.45) 0.50 0.14 Afr-Am GG 31 33 Ref (1.0) 82 58 Ref (1.0) AG 42 41 1.24 (0.59-2.59) 63 73 0.67 (0.40-1.12) AA 2 7 0.24(0.04-1.41) 0.53 12 19 0.40 (0.17-0.96) 0.02 0.44 Latina GG 41 55 Ref (1.0) 70 78 Ref (1.0) AG 33 55 0.64(0.32-1.27) 68 80 0.92 (0.57-1.51) AA 18 15 1.70 (0.69-4.15) 0.54 27 35 0.77 (0.41-1.43) 0.42 0.22 965A>G White AA 17 30 Ref (1.0) 66 61 Ref (1.0) AG 36 42 1.29 (0.56-2.99) 86 86 1.00 (0.61-1.64) GG 17 16 2.45 (0.86-7.03) 0.11 34 41 0.79(0.43-1.45) 0.49 0.18 Afr-Am AA 46 50 Ref (1.0) 109 84 Ref (1.0) AG 28 29 1.04(0.51-2.09) 44 57 0.65 (0.38-1.09) GG 1 2 0.36 (0.03-4.71) 0.79 4 9 0.28 (0.08-1.02) 0.02 0.69 Latina AA 50 70 Ref (1.0) 77 103 Ref (1.0) AG 29 46 0.77(0.39-1.50) 71 67 1.35 (0.84-2.17) GG 1 3 9 1.85 (0.67-5.13) 0.57 17 23 0.84(0.41-1.74) 0.83 0.12 Since in vitro studies suggested that 12-LOX metabolites probably play a more important role in promoting tumor metastasis than initiation, we analyzed the 12-LOX gene polymorphisms according to the stage of tumor. We found that the observed protective association for 782A and 965G alleles among African-Americans appeared to be stronger for tumors with regional or remote extension than for localized tumors, though the difference was not significant (table 4.5.6). Table 4.5.6 The 12-LOX gene polymorphisms and breast cancer risk, by tumor stages 12-LOX Controls Cases Adj. OR* Cases (non Adj. OR* p for int. genotype (localized) (95% Cl) localized) (95% Cl) 782G>A White GG 95 64 Ref (1.0) 23 Ref (1.0) AG 137 102 1.09 (0.71-1.67) 32 1.04 (0.55-1.94) AA 61 38 0.92 (0.54-1.59) 16 1.27 (0.60-2.69) 0.60 Afr-Am GG 98 72 Ref (1.0) 43 Ref (1.0) AG 122 70 0.86 (0.55-1.35) 37 0.67 (0.39-1.16) AA 29 13 0.57 (0.26-1.23) 3 0.25 (0.07-0.88) 0.38 Latina GG 148 76 Ref (1.0) 43 Ref (1.0) AG 147 70 0.86 (0.56-1.31) 42 0.94 (0.57-1.56) AA 52 32 1.10(0.63-1.91) 18 1.12(0.57-2.17) 0.95 965A>G White AA 96 65 Ref (1.0) 23 Ref (1.0) AG 136 102 1.09 (0.71-1.67) 32 1.05 (0.56-1.97) GG 61 37 0.89 (0.52-1.53) 16 1.28 (0.60-2.70) 0.55 Afr-Am AA 143 99 Ref (1.0) 60 Ref (1.0) AG 94 52 0.88(0.56-1.37) 22 0.59 (0.33-1.05) GG 12 4 0.41 (0.12-1.35) 1 0.20 (0.02-1.65) 0.39 Latina AA 189 89 Ref (1.0) 47 Ref (1.0) AG 125 67 1.05 (0.69-1.59) 44 1.36(0.83-2.22) GG 33 22 1.20 (0.64-2.25) 12 1.34 (0.62-2.89) 0.64 139 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12-LOX has been suggested to interact with the estrogen pathway in breast cancer development. Over-expression of 12-LOX in an estrogen dependent breast cancer cell line resulted in the diminished expression of estrogen receptor and less dependence on estrogen for cancer cell growth (Liu et al., 1996b). We therefore examined whether the association between 12-LOX gene polymorphisms and breast cancer differed for estrogen receptor positive and negative tumors. Among African- American women, the associations were similar regardless of ER status (table 4.5.7). However, among whites and Latinas, the 782G and 965A alleles, which were protective among African-Americans, appeared to be associated with increased risk of ER negative breast cancer. When stratifying the analysis by linoleic acid intake, we did not find a significant interaction with 12-LOX gene among any ethnic groups. Whites and African-Americans showed an opposite association between 12-LOX genotypes and breast cancer after stratification by linoleic acid intake, with the risk lowest among Whites, but highest among African-Americans carrying 782GG or 965AA and with high LA intake (table 4.5.8). We also analyzed the association between 12-LOX gene and breast cancer risk by haplotype. Because of the LD between these two loci, there were only three haplotypes, 782G/965A, 782A/965Q and 782A/965A, among whites, African- Americans and Latinas. The 782G allele was actually equivalent to 782G/965A haplotype and the 965G allele equivalent to 782A/965G haplotype. Therefore the associations for these two haplotypes were same to the genotype analysis we reported 140 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. above. The haplotype 782A/965A was also common among African-Americans and Latinas. The protective effect against breast cancer appeared to be 782G/965A <782A/965A <782A/965Q however only among African-Americans (table 4.5.9). Table 4.5.7 The 12-LOX gene polymorphisms and breast cancer risk, by estrogen receptor (ER) status of tumor 12-LOX genotype Controls Cases (ER positive) Adj. OR* (95% Cl) Cases (ER negative) Adj. OR* (95% Cl) Pfor int. 782G>A White GG 95 64 Ref (1.0) 12 Ref (1.0) AG 137 95 0.99 (0.64-1.53) 20 1.29 (0.54-2.94) AA 61 35 0.80 (0.46-1.39) 11 1.74 (0.67-4.52) 0.31 Afr-Am GG 98 65 Ref (1.0) 33 Ref (1.0) AG 122 58 0.76 (0.47-1.23) 28 0.71 (0.39-1.30) AA 29 8 0.40 (0.17-0.98) 5 0.51 (0.17-1.63) 0.91 Latina GG 148 79 Ref (1.0) 26 Ref (1.0) AG 147 66 0.74 (0.48-1.14) 28 1.10(0.60-2.03) AA 52 25 0.77 (0.43-1.38) 18 1.98 (0.96-4.08) 0.06 965A>G White AA 96 65 Ref (1.0) 12 Ref (1.0) AG 136 95 0.99 (0.64-1.52) 20 1.30 (0.57-2.96) GG 61 34 0.76 (0.44-1.33) 11 1.79 (0.69-4.63) 0.25 Afr-Am AA 143 82 Ref (1.0) 51 Ref (1.0) AG 94 44 0.81 (0.50-1.32) 15 0.49 (0.26-0.95) GG 12 5 0.66 (0.21-2.01) 0 0.45 Latina AA 189 89 Ref (1.0) 30 Ref (1.0) AG 125 63 0.94 (0.61-1.43) 31 1.58(0.88-2.82) GG 33 18 0.93 (0.48-1.80) 11 1.97 (0.86-4.53) 0.14 141 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.5.8 The 12-LOX gene polymorphisms and breast cancer risk, by ____________ linoleic acid (LA) intake________________________________ Race Genotype Cases Controls Adj. OR** (95% Cl) Pfor interaction 782G>A Whites LA<median AG or AA 81 79 Ref GG 46 30 1.30 (0.72-2.34) >median AG or AA 107 119 0.69 (0.43-1.12) GG 41 65 0.50 (0.28-0.87) 0.08 African- LA<median Americans AG or AA 42 59 Ref GG 33 39 1.07 (0.56-2.04) >median AG or AA 83 92 1.52 (0.85-2.70) GG 87 59 2.44(1.34-4.46) 0.42 Latinas LAcmedian AG or AA 91 136 Ref GG 70 101 1.04 (0.67-1.60) >median AG or AA 74 63 1.30(0.79-2.11) GG 50 47 1.38 (0.81-2.36) 0.92 965A>G Whites LA<median AG or GG 81 78 Ref AA 46 31 1.24 (0.69-2.24) >median AG or GG 106 119 0.68 (0.42-1.09) AA 42 65 0.50 (0.29-0.88) 0.13 African- LAcmedian Americans AG or GG 28 41 Ref AA 47 57 1.13(0.58-2.13) >median AG or GG 53 65 1.39 (0.70-2.77) AA 117 86 2.39(1.26-4.54) 0.36 Latinas LAcmedian AG or GG 82 107 Ref AA 79 130 0.81 (0.52-1.24) >median AG or GG 65 51 1.21 (0.71-2.05) AA 59 59 1.16(0.70-1.93) 0.82 142 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.5.9 The 12-LOX gene haplotypes and breast cancer risk among three ethnic groups Genotype Haplotype Cases Controls Adj. OR* (95% Cl) White 782GG/965AA Two 782G-965A 87 95 Ref (1.0) 782AG/965AA 782G-965A/782A-965A 1 1 N/A 782AG/965AG 782G-965A/782A-965 G 133 136 1.07 (0.72-1.59) 782AA/965AA Two 782A-965A 0 0 N/A 782AA/965AG 782A-965A/782A-965 G 1 0 N/A 782AA/965GG Two 782A-965G 53 61 0.99 (0.60-1.62) African- 782GG/965AA Two 782G-965A 120 98 Ref (1.0) American 782AG/965AA 782G-965A/782A-965A 40 41 0.82 (0.47-1.41) 782AG/965AG 782G-965A/782A-965G 69 81 0.75 (0.49-1.17) 782AA/965AA Two 782A-965A 4 4 0.74 (0.16-3.53) 782AA/965AG 782A-965A/782A-965G 7 13 0.48 (0.17-1.30) 782AA/965GG Two 782A-965G 5 12 0.30 (0.10-0.92) Latina 782GG/965AA Two 782G-965A 120 148 Ref (1.0) 782AG/965AA 782G-965A/782A-965A 18 39 0.58(0.30-1.10) 782AG/965AG 782G-965A/782A-965G 97 108 1.01 (0.69-1.50) 782AA/965AA Two 782A-965A 0 2 N/A 782AA/965AG 782A-965G/782A-965A 16 17 1.28 (0.60-2.74) 782AA/965GG Two 782A-965G 34 33 1.11 (0.63-1.95) 4.5.3 Discussion In this study we examined the association between 12-LOX gene polymorphisms and breast cancer risk. Among African-Americans, carrying the 782A allele or the 965G allele was found associated with decreased risk of breast cancer. However, these findings were not seen among whites and Latinas. The 12-LOX gene has been reported to be overexpressed in breast cancer tissue (Natarajan et al., 1997) and the 12-LOX product 12(S)-HETE was also found to play important roles in tumor development, especially tumor metastasis (Honn et al., 1994; Liu et al., 1994; Liu et al., 1991; Liu et al., 1996a). Polymorphisms in the 12- LOX gene may influence 12-LOX expression and/or activity, and therefore may influence breast cancer risk. Both of the 12-LOX gene polymorphisms we studied here result in the change of amino acids. 782G>A polymorphism results in arginine to glutamine conversion at position 261. Glutamine is highly conserved at this position in the lipoxygenase gene family (Funk et al., 1990), implying its significance for function. How this change influences 12-LOX activity is not clear yet. We found an association between Gln261 and decreased risk of breast cancer among African- American women, but not among whites and Latinas. Because 782G>A and 965A>G are in strong linkage disequilibrium, it is hard to separate the independent effect of these two polymorphisms by the association study. It is also possible that there is a combined effect from the interaction between these two polymorphisms. Haplotype analysis also shows that the less common haplotype 782A/965A is probably associated with decreased risk of breast cancer. Since functional consequence of each 144 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. polymorphism or haplotype has not been reported, we do not know yet through what mechanisms the polymorphisms were associated with breast cancer. Although we observed that the negative association for 782G and 965A among African-Americans appeared to be stronger among postmenopausal women and for non-localized disease, the differences were not significant. The associations were not consistent across all ethnicities either. We also did not see consistent association when stratifying the analysis by history of benign breast disease and status of estrogen receptor. There was no effect modification between linoleic acid intake and 12-LOX gene genotype. While these results need to be further examined by other studies among multiple ethnicities, there are some possible reasons that might have complicated the associations observed in this study. For example, besides these two non-synonymous polymorphisms, other unknown polymorphisms at 5’ and 3’ regulatory regions, may also influence the 12-LOX gene expression. A comprehensive analysis of the 12-LOX gene variations regarding their functional consequences and the association with cancer risks needs to be done. In addition, the 12-LOX pathway also interacts with other lipoxygenase pathways, such as 5-LOX and 15-LOX pathways that have also been suggested to play important roles in various cancers. Genetic differences among other LOX pathways may also confound the association between 12-LOX and breast cancer. Understanding how these lipoxygenase pathways differ among different ethnicities may help us to clarify the discrepancies among ethnicities for the 12-LOX gene polymorphisms. 145 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4.6 Other LOX gene Polymorphisms and Breast Cancer Risk Besides the 5-LOX and 12-LOX gene polymorphisms, we also analyzed the association between polymorphisms in other LOX family genes, including the 15- LOX-2 gene and the FLAP gene, and breast cancer risk. 4.6.1 The 15-LOX-2 Gene 1967G>A Polymorphism The only nonsynonymous polymorphism in the 15-LOX-2 gene, 1967G>A, results in the change of arginine to glutamine at position 656. No significant association was found between this polymorphism and breast cancer risk among any ethnic group, though the A allele appeared to be protective among whites and Latinas (table 4.6.1). Results were similar when we performed the analysis only among subjects without family history of breast cancer (table 4.6.2). No consistent association was found when we stratified analysis by menopausal status, benign breast disease, stage of disease, or intake of linoleic acid (data not shown). Table 4.6.1 The 15-LOX-2 gene 1967G>A polymorphism and breast cancer risk 1967G>A Cases Controls OR P Adj. OR* P genotype (%) (%) (95% Cl) trend (95% Cl) trend White GG 73 (26.6) 70 (23.9) Ref (1.0) Ref (1.0) AG 136 (49.5) 140 (47.8) 0.93 (0.62-1.40) 0.84 (0.55-1.29) AA 66 (24.0) 83 (28.3) 0.76 (0.48-1.21) 0.25 0.74 (0.45-1.20) 0.22 Afr-Am GG 169 (69.0) 170 (68.3) Ref (1.0) Ref (1.0) AG 63 (25.7) 68 (27.3) 0.93 (0.62-1.40) 0.98 (0.64-1.49) AA 13 (5.3) 11 (4.4) 1.19(0.52-2.73) 0.97 0.95 (0.40-2.27) 0.88 Latina GG 111 (39.0) 117(33.7) Ref (1.0) Ref (1.0) AG 127 (44.6) 170 (49.0) 0.79 (0.56-1.11) 0.75 (0.52-1.09) AA 47 (16.5) 60(17.3) 0.83 (0.52-1.31) 0.28 0.76 (0.47-1.24) 0.18 146 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.6.2 The 15-LOX-2 gene 1967G>A polymorphism and breast cancer risk among women without family history of breast cancer 1967G>A Cases Controls OR P Adj. OR* P genotype (%) (%) (95% Cl) trend (95% Cl) trend White GG 61 (28.0) 58 (23.3) Ref (1.0) Ref (1.0) AG 105 (48.2) 121 (50.4) 0.83 (0.53-1.29) 0.73 (0.46-1.18) AA 52 (23.9) 70 (28.1) 0.71 (0.43-1.17) 0.18 0.66 (0.38-1.13) 0.13 Afr-Am GG 143 (68.8) 147 (67.7) Ref (1.0) Ref (1.0) AG 56 (26.9) 61 (28.1) 0.94 (0.61-1.45) 0.98 (0.62-1.53) AA 9(4.3) 9(4.1) 1.03 (0.40-2.66) 0.88 0.90 (0.33-2.45) 0.84 Latina GG 97 (38.3) 110(34.7) Ref (1.0) Ref (1.0) AG 112(44.8) 153 (48.3) 0.83 (0.58-1.20) 0.81 (0.55-1.19) AA 41 (16.4) 54 (17.0) 0.86(0.53-1.40) 0.43 0.81 (0.48-1.37) 0.33 4.6.2 The FLAP Gene 5’ Poly(A) Microsatellite The only common polymorphic site in the 5’ regulatory region of the FLAP gene is a poly(A) microsatellite. Overall, there was no association between the FLAP gene poly(A) microsatellite and breast cancer risk among any ethnicities (table 4.6.3). However, after excluding women with family history of breast cancer, carrying the short allele was associated with decreased risk of breast cancer among white women (table 4.6.4). No consistent association was found when we stratified analysis by menopausal status, benign breast disease, stage of disease, or intake of linoleic acid (data not shown). Because FLAP is required in the 5-lipoxygenase metabolic pathway to transfer arachidonic acid to 5-LOX, we also examined whether the associations of FLAP and 5-LOX genes were confounded by each other, or whether there is an interaction between these two genes. After adjustment for each other, the association between these two genes and breast cancer did not change (data not shown). Results 147 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. neither changed when we analyzed one gene by stratifying the analysis on the other gene, though the power of analysis was insufficient due to the small sample size. Table 4.6.3 The FLAP gene 5’ poly(A) microsatellite and breast cancer risk poly(A) Cases Controls OR P Adj. OR* P genotype (%) (%) (95% Cl) trend (95% Cl) trend White LL 206 (70.3) 208 (75.6) Ref (1.0) Ref (1.0) LS 79 (27.0) 59 (21.5) 0.74 (0.50-1.09) 0.71 (0.47-1.07) SS 8 (2.7) 8 (2.9) 0.99 (0.37-2.69) 0.24 0.96 (0.33-2.78) 0.19 Afr-Am LL 69 (27.7) 64(26.1) Ref (1.0) Ref (1.0) LS 116(46.6) 102 (41.6) 0.95 (0.62-1.46) 1.00 (0.63-1.59) SS 64 (25.7) 79 (32.2) 1.33 (0.83-2.14) 0.23 1.44 (0.87-2.38) 0.14 Latina LL 187 (53.9) 170 (59.7) Ref (1.0) Ref (1.0) LS 139 (40.1) 98 (34.4) 0.78 (0.56-1.08) 0.80 (0.57-1.14) SS 21 (6.1) 17 (6.0) 0.89 (0.46-1.74) 0.23 1.14(0.55-2.34) 0.55 Table 4.6.4 The FLAP gene 5’ poly(A) microsatellite and breast cancer risk among women without family history of breast cancer poly(A) genotype Cases Controls OR (95% Cl) P trend Adj. OR* (95% Cl) P trend White LL 171 (78.4) 173 (69.5) Ref (1.0) Ref (1.0) LS 44 (20.2) 68 (27.3) 0.65(0.42-1.01) 0.65 (0.41-1.03) SS 3 (1.4) 8 (3.2) 0.38(0.10-1.46) 0.02 0.44 (0.11-1.74) 0.03 Afr-Am LL 50 (24.0) 58 (26.7) Ref (1.0) Ref (1.0) LS 90 (43.3) 100 (46.1) 1.04(0.65-1.68) 1.08 (0.65-1.78) SS 68 (32.7) 59 (27.2) 1.34 (0.80-2.24) 0.26 1.39 (0.81-2.40) 0.22 Latina LL 149 (59.6) 168 (53.0) Ref (1.0) Ref (1.0) LS 85 (34.0) 129 (40.7) 0.74 (0.52-1.06) 0.73 (0.50-1.06) SS 16 (6.4) 20 (6.3) 0.90 (0.45-1.80) 0.21 1.08 (0.51-2.27) 0.34 148 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 5 Summary Nutritional epidemiology research has great public-health importance. Although much effort has been devoted to developing methodologies for examining the association between diet and chronic diseases, including cancer, the true relationships between diet and disease are still not clear, due to the obstacles of measurement error in dietary assessment and confounding issues. The study of the association between dietary fat and risk of breast cancer is an example of this predicament. Our study is no exception. 5.1 Major Findings of This Study In this population-based multiethnic case-control study, we analyzed the association between dietary fat intake and the risk of breast cancer using two different approaches to measure the dietary fat. Using the first approach, measuring fat intake by food frequency questionnaire, we found (1) high-fat intake is associated with increased risk of breast cancer; (2) among types of fat, oleic acid is most strongly and consistently associated with increased risk; linoleic acid is weakly associated with increased risk; saturated fat is not or is negatively associated with risk; (3) history of benign breast disease modifies the association between polyunsaturated fat and breast cancer risk, and polyunsaturated fat is associated with increased risk only among women with previous benign breast disease; (4) the positive association between high 149 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. fat diet and breast cancer risk is stronger for more progressive tumors than for less progressive tumors. Using the second approach, by asking about cooking fat usage, we found that compared to using olive or canola oil (rich in monounsaturated fat), using vegetable/com oil (rich in n-6 polyunsaturated fat), or hydrogenated fat for cooking is associated with increased risk of breast cancer. Another way to investigate the role of dietary fat is to examine the polymorphisms in fat metabolic pathways. Analysis of n-6 polyunsaturated fat lipoxygenase pathway gene polymorphisms found that (1) the 5-LOX gene Spl binding site non-wild-type alleles are associated with increased risk of breast cancer among Latinas and African-Americans; (2) the 12-LOX gene coding region polymorphisms are associated with breast cancer risk among African-Americans. 5.2 Dietary Fat Intake and Breast Cancer R isk Discrepancies among Food Frequency Data, Cooking Fat Data, and Previous Experimental Results Our finding of the significant association between total fat intake and increased risk of breast cancer is consistent with the overall finding from case-control studies. However, further analysis of types of fat revealed discrepancies among food frequency data, cooking fat data, and previous experimental results. 150 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2.1 Discrepancy between FFQ Results and Previous Experimental Results Analysis of types of fat from our food frequency data showed that oleic acid is the most strongly associated with increased risk of breast cancer. Linoleic acid, which has been consistently demonstrated to have a strong tumor-enhancing effect in animal models and in vitro studies, however, is only weakly associated with increased risk and this association disappeared after adjustment for other types of fat. This conundrum also exists in many other epidemiological studies. Although we need to be cautious in generalizing the results of animal models and in vitro studies, such discrepancies may reflect measurement errors in dietary assessment and confounding issues in epidemiological data. First, nutrient consumption is notoriously difficult to measure accurately, leading to inherent random or systematic measurement error in dietary assessment, which can have a profound impact on the results of a study. Elimination of the dietary assessment error is not readily solvable at present time. Some common measurement errors, such as random error from unreliable report of dietary intake and the systematic error from underreporting or overreporting among certain groups of subjects, may also exist in this study and have hampered the identification of the real association between dietary fat and breast cancer. Second, dietary fat is not a completely independent lifestyle factor. It is closely related to many other lifestyle risk factors, such as caloric intake, weight gain, obesity, physical activity, and other behavioral characteristics, which have a high probability of being intricately connected with the effects of dietary fat, per se (Greenwald, 1999). Therefore the independent effects of dietary fat and 151 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. these lifestyle components are very difficult to separate, leading to some inevitable degree of confounding. In addition, the types of fat, which may play completely different roles in carcinogenesis, are highly correlated with each other in human diet, making the estimate of the independent effect of types of fat even more difficult. In this study and other studies focusing on the American diet, the major contributors to oleic acid intake are animal products, such as meat and dairy products. Oleic acid intake is highly correlated with total fat (r=0.93) and saturated fat (r=0.75) intake. Therefore it is difficult to separate independent effect of oleic acid from effects from other types of fat and from other components existing in animal products. In contrast to the studies focusing on the American diet, oleic acid is reported to be protective in the Mediterranean diet, in which oleic acid is mainly from olive oil, but not from animal products. Such a discrepancy in the association between oleic acid and breast cancer, due to the difference in the source of fat, further supports that the association we observed for oleic acid may not reflect the effect of oleic acid per se. Lack of association between polyunsaturated fat and breast cancer risk has been reported by most of epidemiological studies, including this study. In addition to the issues we discussed above, lack of distinction between n-6 and n-3 fatty acids may also have attenuated the ability of finding a true relationship between polyunsaturated fat and breast cancer. Experimental studies find opposite effects of n-6 and n-3 polyunsaturated fats in mammary tumorigenesis (Fay et al., 1997). The balance between n-6 and n-3 fatty acids has also been suggested to important in breast cancer. The positive association between n-6 polyunsaturated fat and breast cancer may be 152 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. found only among people with low n-3 fatty acid intake, as suggested by a recent report (Gago-Dominguez et al., 2003). Therefore, to determine whether linoleic acid intake is associated with breast cancer risk in this study, it is necessary to further extract the data of n-3 fatty acids intake from the questionnaire for analysis. 5.2.2 Discrepancy between FFQ Results and Cooking Fat Results FFQ results were discrepant, not only with previous experimental studies, but also with the cooking fat results from our own study. While the food frequency data showed a positive association between oleic acid, but not linoleic acid, and increased risk of breast cancer, the analysis of cooking fats showed that women using cooking oils high in oleic acid (olive or canola oils) were at lower risk of breast cancer compared to those using oils high in linoleic acid (vegetable or com oils). This finding is consistent with previous studies that reported protective effects of olive oil in the Mediterranean diet (La Vecchia et al., 1995; Martin-Moreno et al., 1994; Trichopoulou et al., 1995). Although the analysis of cooking fat is less vulnerable to some of the problems existing in food frequency data, such as unreliable recall and strong correlation among types of fat, we still need to be cautious with the interpretation, due to possible confounding by other components in olive and canola oils and other life-style factors that could not be well adjusted for in the analysis. Actually, the protective effects of olive and canola oil have been suggested to be from other anti-carcinogenic components in the oils, but not from the oleic acid (Bartsch et al., 1999). 153 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.2.3 Inconsistency of FFQ Results across Ethnicity We found that the associations between fat intake and breast cancer are not always consistent across ethnicity. Because published data on multiethnic studies are very limited, we can not compare our results with other studies. Using the same FFQ among subjects from different ethnic groups could be a possible source of inconsistency. However, the Block FFQ, which was adapted for this study, has been validated among whites and African-Americans, and it was modified by adding more African-American- and Latino-specific foods for the purpose of this study. Therefore it is unlikely that the FFQ is a major source of bias responsible for the inconsistent results across ethnicity. Instead, the inconsistency may reflect important confounding issues that have not been identified or that were not easy to adjust for in data analysis. We actually found that some well-established breast cancer risk factors also did not show a consistent association across ethnicity in this study population, supporting the existence of other risk factors for breast cancer that cause diverse confounding across ethnicity. 5.3 N-6 Polyunsaturated Fat Lipoxygenase Gene Polymorphisms and Breast Cancer 5.3.1 Implication of an Association between Fat Metabolizing Gene Polymorphisms and Breast Cancer Risk With all above uncertainties that may have introduced bias and attenuated the power to identify the true association between dietary fat and breast cancer, individual 154 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. genetic susceptibility related to fat metabolism may have further clouded an existing dietary fat-breast cancer relationship. Genetic susceptibility may result in environmental factors being a more important contributor to breast cancer in some proportion of a study population than in rest of the population (Slattery et al., 1995). For example, experimental studies demonstrated that n-6 polyunsaturated fats promote the development of breast cancer, and the n-6 polyunsaturated fats metabolites produced through lipoxygenase pathways may be involved in the mechanism of the tumor-enhancing effect. Therefore the polymorphisms in lipoxygenase genes may influence the n-6 polyunsaturated fat metabolism through this pathway, and overlooking this genetic susceptibility may limit the ability to detect associations existing between dietary fat and breast cancer. Considering that both individual genetic variations and dietary intake may contribute to breast cancer risk, nutritional epidemiological data alone may not be able to provide a definitive answer to this complex relationship. Besides, while methodological issues related to dietary assessment in epidemiological studies have been a continuous concern in data analysis and interpretation, genetic data, which is largely exempt from measurement error and is a constant exposure throughout life, offer an alternative approach to determine whether there is a relationship between dietary fat and breast cancer risk. An association between lipoxygenase gene polymorphisms and breast cancer itself may imply the significance of this metabolic pathway and therefore its substrate, dietary fat, in breast cancer. 155 permission of the copyright owner. Further reproduction prohibited without permission. 5.3.2 5-LOX and FLAP Gene Polymorphisms and Breast Cancer Risk We found that the 5-LOX gene Spl binding site non-wild-type alleles are associated with increased risk of breast cancer among Latinas and African-Americans. However this result is discrepant with the findings from the functional studies. Results from both our functional study with breast cancer cell and another study with Hela cell (In et al., 1997) suggest a diminished transcriptional activity of the non-wild- type Spl binding site polymorphisms. Therefore subjects carrying non-wild-type allele should have diminished 5-LOX gene transcription and consequently reduced production of 5-LOX metabolites, such as 5-HETE and leukotrienes. Since in vitro studies support a promotional effect of 5-LOX metabolites in breast cancer cell growth, we would expect that subjects carrying the non-wild-type allele should have reduced risk of breast cancer, which is opposite to the finding of our association study. Interestingly, a recent study of the same polymorphism and atherosclerosis also found that the non-wild-type allele is associated with the enhanced intima-media thickness and chronic arterial inflammation (Dwyer et al., 2004). Such a discrepancy between functional and association studies may implicate several possibilities: (1) in vitro functional studies only covered the Spl binding site polymorphism. Since other polymorphisms have also been identified in the 5’ regulatory region, the regulation of the 5-LOX gene is probably more complicated than what has been reflected by this single polymorphism. Further functional studies covering those polymorphisms are needed to examine the combined effect of multiple polymorphisms on the 5-LOX gene regulation. (2) The association we observed for Spl binding site polymorphism 156 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. may be due to its LD with other unidentified polymorphisms that is important for the 5-LOX gene regulation. (3) methylation at the Spl binding site is involved in the cell- type specific expression of 5-LOX (Uhl et al., 2002). How the polymorphism at this site influence the gene regulation by methylation is not clear. (4) 5-LOX is also involved in the production of endogenous anti-inflammatory mediators, such as lipoxins, through several different pathways involving other lipoxygenases and cyclooxygenases (Kantarci and Van Dyke, 2003). Therefore the effect of the 5-LOX gene polymorphism is probably determined by the balance between pro- and anti inflammatory mechanisms. We did not find significant interaction between the 5-LOX gene polymorphism and dietary linoleic acid (n-6 polyunsaturated fat) intake. While we may not have sufficient power to find a significant interaction in this study, the measurement error and confounding issues in nutrient data, which may have hampered the finding of true association between diet and breast cancer, can also attenuate the ability of finding gene and diet interaction. Five-lipoxygenase activating protein (FLAP) is another important protein in the 5-LOX pathway. The expression and function of FLAP is closely related to 5- LOX. Since no functional difference was found for the short and long allele of the FLAP gene poly(A) microsatellite, it is not surprising to find no association between this polymorphism and breast cancer risk. However, considering the importance of FLAP in the 5-LOX pathway and probably also in other LOX pathways, this gene is worthy of further investigation. 157 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 5.3.3 12-LOX Gene Polymorphisms and Breast Cancer Risk For the 12-LOX gene, at the present time we only genotyped the nonsynonymous polymorphisms. The functional consequences of these polymorphisms are not clear due to the lack of functional studies. However, one of these polymorphisms, the Arg261Gln, occurs at a position that is highly conserved among LOX family genes and across species, implying its significance for LOX function. A significant association was found for this polymorphism and another polymorphism, Asn322Ser, which is in strong LD with it, with breast cancer risk among African-Americans. If Arg261Gln is a causal polymorphism that changes the activity of 12-LOX, our results suggest that carrying the 12-LOX Gin 261 allele is probably related to the decreased activity of 12-LOX and reduced production of 12- HETE, whose function is to enhance mammary tumorigenesis. Whether such a relationship is true needs to be examined by functional studies. On the other hand, since the finding is not consistent across ethnicity, it is also possible that other causal polymorphisms are in LD with this polymorphism and the LD varies across ethnicities. Therefore it is also necessary to further examine other polymorphisms in the 12-LOX gene, such as the polymorphisms we identified at 5’ regulatory region, for their functional significance, the LD among polymorphisms, and their association with breast cancer. As the first study investigating the LOX gene polymorphisms with breast cancer risk, at the present time we can not compare our results with other studies. We also eagerly expect other studies to examine the findings of our study. 158 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The hypothesis that dietary fat intake is associated with breast cancer risk remains of interest to epidemiological studies, due to its potential public health importance. The study of genetic susceptibility is becoming an important approach to illuminate the heterogeneity in diet and disease association. Findings in our study, which are among the first attempts in this approach, may provide useful information for epidemiologists and molecular biologists that are interested in solving the puzzles in the relationship between dietary fat and breast cancer risk. 159 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. References Armstrong, B., and Doll, R. (1975). Environmental factors and cancer incidence and mortality in different countries, with special reference to dietary practices. Int J Cancer 15, 617-253. Assmann, G., de Backer, G., Bagnara, S., Betteridge, J., Crepaldi, G., Femandez-Cruz, A., Godtffedsen, J., Jacotot, B., Paoletti, R., Renaud, S., et al. (1997). Olive oil and the Mediterranean diet: implications for health in Europe. (Review). British Journal of Nursing 6, 675-677. Avis, I., Hong, S. H., Martinez, A., Moody, T., Choi, Y. H., Trepel, J., Das, R., Jett, M., and Mulshine, J. L. (2001). Five-lipoxygenase inhibitors can mediate apoptosis in human breast cancer cell lines through complex eicosanoid interactions. FASEB Journal 15, 2007-2009. Avis, I., Hong, S. H., Moody, T., and Mulshine, J. L. (1998). In vitro and in vivo correlation between inhibition of the lipoxygenase (LO) pathway and up?regulation of apoptosis in breast cancer cell lines. Proc Am Assoc Cancer Res 39, 115. Avis, I. M., Jett, M., Boyle, T., Vos, M. D., Moody, T., Treston, A. M., Martinez, A., and Mulshine, J. L. (1996). Growth control of lung cancer by interruption of 5- lipoxygenase-mediated growth factor signaling. Journal of Clinical Investigation 97, 806-813. Baas, F., Bikker, H., van Ommen, G. J., and de Vijlder, J. J. (1984). Unusual scarcity of restriction site polymorphism in the human thyroglobulin gene. A linkage study suggesting autosomal dominance of a defective thyroglobulin allele. Hum Genet 67, 301-305. Baer, A. N., Costello, P. B., and Green, F. A. (1991). In vivo activation of an omega-6 oxygenase in human skin. Biochemical & Biophysical Research Communications 180, 98-104. Bagga, D., Anders, K. H., Wang, H. J., and Glaspy, J. A. (2002). Long-chain n-3-to-n- 6 polyunsaturated fatty acid ratios in breast adipose tissue from women with and without breast cancer. Nutrition & Cancer 42, 180-185. Balakrishnan, A., Cramer, S., Bandyopadhyay, G. K., Imagawa, W., Yang, J., Elias, J., Beattie, C. W., Das Gupta, T. K., and Nandi, S. (1989). Differential proliferative response to linoleate in cultures of epithelial cells from normal human breast and fibroadenomas. Cancer Research 49, 857-862. 160 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Bandyopadhyay, G. K., Imagawa, W., Wallace, D., and Nandi, S. (1987). Linoleate metabolites enhance the in vitro proliferative response of mouse mammary epithelial cells to epidermal growth factor. Journal of Biological Chemistry 262, 2750-2756. Bardon, S., Le, M. T., and Alessandri, J. M. (1996). Metabolic conversion and growth effects of n-6 and n-3 polyunsaturated fatty acids in the T47D breast cancer cell line. Cancer Letters 99, 51-58. Bartsch, H., Nair, J., and Owen, R. W. (1999). Dietary polyunsaturated fatty acids and cancers of the breast and colorectum: emerging evidence for their role as risk modifiers. Carcinogenesis 20, 2209-2218. Bjamason, O., Day, N., Snaedal, G., and Tulinius, H. (1974). The effect of year of birth on the breast cancer age-incidence curve in Iceland. International Journal of Cancer 13, 689-696. Block, G., Hartman, A. M., Dresser, C. M., Carroll, M. D., Gannon, J., and Gardner, L. (1986). A data-based approach to diet questionnaire design and testing. American Journal of Epidemiology 124, 453-469. Block, G., Hartman, A. M., and Naughton, D. (1990a). A reduced dietary questionnaire: development and validation. Epidemiology 1, 58-64. Block, G., Thompson, F. E., Hartman, A. M., Larkin, F. A., and Guire, K. E. (1992). Comparison of two dietary questionnaires validated against multiple dietary records collected during a 1-year period. J Am Diet Assoc 92, 686-693. Block, G., Woods, M., Potosky, A., and Clifford, C. (1990b). Validation of a self administered diet history questionnaire using multiple diet records. J Clin Epidemiol 43, 1327-1335. Boudreau, M. D., Chanmugam, P. S., Hart, S. B., Lee, S. H., and Hwang, D. H. (1991). Lack of dose response by dietary n-3 fatty acids at a constant ratio of n-3 to n- 6 fatty acids in suppressing eicosanoid biosynthesis from arachidonic acid. American Journal of Clinical Nutrition 54, 111-117. Boyd, N. F., Martin, L. J., Noffel, M., Lockwood, G. A., and Trichler, D. L. (1993). A meta-analysis of studies of dietary fat and breast cancer risk. British Journal of Cancer 68, 627-636. Boyd, N. F., Stone, J., Vogt, K. N., Connelly, B. S., Martin, L. J., and Minkin, S. (2003). Dietary fat and breast cancer risk revisited: a meta-analysis of the published literature. Br J Cancer 8 9 ,1672-1685. 161 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Brash, A. R. (1999). Lipoxygenases: occurrence, functions, catalysis, and acquisition of substrate. Journal of Biological Chemistry 274, 23679-23682. Brash, A. R., Boeglin, W. E., and Chang, M. S. (1997). Discovery of a second 15S- lipoxygenase in humans. Proceedings of the National Academy of Sciences of the United States of America 94, 6148-6152. Breslow, N. E., and Day, N. E. (1980). Statistical methods in cancer research. Volume I - The analysis of case-control studies. IARC Scientific Publications, 5-338. Brinton, L. A., Benichou, J., Gammon, M. D., Brogan, D. R., Coates, R., and Schoenberg, J. B. (1997). Ethnicity and variation in breast cancer incidence. Int J Cancer 73, 349-355. Buckman, D. K., Hubbard, N. E., and Erickson, K. L. (1991). Eicosanoids and linoleate-enhanced growth of mouse mammary tumor cells. Prostaglandins Leukotrienes & Essential Fatty Acids 4 4 ,177-184. Byrne, C., Rockett, H., and Holmes, M. D. (2002). Dietary fat, fat subtypes, and breast cancer risk: lack of an association among postmenopausal women with no history of benign breast disease. Cancer Epidemiology, Biomarkers & Prevention 11, 261-265. Cade, J., Thomas, E., and Vail, A. (1998). Case-control study of breast cancer in south east England: nutritional factors. Journal of Epidemiology & Community Health 52, 105-110. Cargill, M., Altshuler, D., Ireland, J., Sklar, P., Ardlie, K., Patil, N., Shaw, N., Lane, C. R., Lim, E. P., Kalyanaraman, N., et al. (1999). Characterization of single nucleotide polymorphisms in coding regions of human genes. Nature Genetics 22, 231-238. Carroll, K. K., and Hopkins, G. J. (1979). Dietary polyunsaturated fat versus saturated fat in relation to mammary carcinogenesis. Lipids 14, 155-158. Carroll, K. K., and Khor, H. T. (1975). Dietary fat in relation to tumorigenesis. Progress in Biochemical Pharmacology 10, 308-353. Chajes, V., Hulten, K., Van Kappel, A. L., Winkvist, A., Kaaks, R., Hallmans, G., Lenner, P., and Riboli, E. (1999). Fatty acid composition in serum phospholipids and risk of breast cancer: a prospective cohort study in Northern Sweden. International Journal of Cancer 83, 585-590. Chajes, V., Sattler, W., Stranzl, A., and Kostner, G. M. (1995). Influence of n-3 fatty acids on the growth of human breast cancer cells in vitro: relationship to peroxides and vitamin-E. Breast Cancer Res Treat 3 4 ,199-212. 162 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Cho, E., Spiegelman, D., Hunter, D. J., Chen, W. Y., Stampfer, M. J., Colditz, G. A., and Willett, W. C. (2003). Premenopausal fat intake and risk of breast cancer. Journal of the National Cancer Institute 9 5 ,1079-1085. Coates, R. J., Eley, J. W., Block, G., Gunter, E. W., Sowell, A. L., Grossman, C., and Greenberg, R. S. (1991). An evaluation of a food frequency questionnaire for assessing dietary intake of specific carotenoids and vitamin E among low-income black women. Am J Epidemiol 134, 658-671. Cohen, L. A., Chen-Backlund, J. Y., Sepkovic, D. W., and Sugie, S. (1993). Effect of varying proportions of dietary menhaden and com oil on experimental rat mammary tumor promotion. Lipids 28, 449-456. Cohen, L. A., Choi, K. W., and Wang, C. X. (1988). Influence of dietary fat, caloric restriction, and voluntary exercise on N-nitrosomethylurea-induced mammary tumorigenesis in rats. Cancer Research 48, 4276-4283. Cohen, L. A., Epstein, M., Pittman, B., and Rivenson, A. (2000). The influence of different varieties of olive oil on N-methylnitrosourea(NMU)-induced mammary tumorigenesis. Anticancer Res 20, 2307-2312. Cohen, L. A., Thompson, D. O., Choi, K., Karmali, R. A., and Rose, D. P. (1986). Dietary fat and mammary cancer, n. Modulation of serum and tumor lipid composition and tumor prostaglandins by different dietary fats: association with tumor incidence patterns. Journal of the National Cancer Institute 77, 43-51. Cohen, L. A., and Wynder, E. I. (1990). Do dietary monounsaturated fatty acids play a protective role in carcinogenesis and cardiovascular disease? Med Hypotheses 31, 83- 89. Connolly, J. M., Gilhooly, E. M., and Rose, D. P. (1999). Effects of reduced dietary linoleic acid intake, alone or combined with an algal source of docosahexaenoic acid, on MDA-MB-231 breast cancer cell growth and apoptosis in nude mice. Nutrition & Cancer 35, 44-49. Connolly, J. M., and Rose, D. P. (1998). Enhanced angiogenesis and growth of 12- lipoxygenase gene-transfected MCF-7 human breast cancer cells in athymic nude mice. Cancer Letters 132,107-112. Daret, D., Blin, P., and Larrue, J. (1989). Synthesis of hydroxy fatty acids from linoleic acid by human blood platelets. Prostaglandins 38,203-214. De Stefani, E., Deneo-Pellegrini, H., Mendilaharsu, M., and Ronco, A. (1998). Essential fatty acids and breast cancer: a case-control study in Uruguay. International Journal of Cancer 76, 491-494. 163 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Drazen, J. M., Yandava, C. N., Dube, L., Szczerback, N., Hippensteel, R., Pillari, A., Israel, E., Schork, N., Silverman, E. S., Katz, D. A., and Drajesk, J. (1999). Pharmacogenetic association between ALOX5 promoter genotype and the response to anti-asthma treatment. Nature Genetics 22, 168-170. Dwyer, J. H., Allayee, H., Dwyer, K. M., Fan, J., Wu, H., Mar, R., Lusis, A. J., and Mehrabian, M. (2004). Arachidonate 5-lipoxygenase promoter genotype, dietary arachidonic acid, and atherosclerosis. N Engl J Med 350, 29-37. Eling, T. E., and Glasgow, W. C. (1994). Cellular proliferation and lipid metabolism: importance of lipoxygenases in modulating epidermal growth factor-dependent mitogenesis. Cancer & Metastasis Reviews 13, 397-410. Erickson, K. L., Schlanger, D. S., Adams, D. A., Fregeau, D. R., and Stem, J. S. (1984). Influence of dietary fatty acid concentration and geometric configuration on murine mammary tumorigenesis and experimental metastasis. J Nutr 114, 1834-1842. Ewertz, M., and Gill, C. (1990). Dietary factors and breast-cancer risk in Denmark. International Journal of Cancer 46, 779-784. Fay, M. P., Freedman, L. S., Clifford, C. K., and Midthune, D. N. (1997). Effect of different types and amounts of fat on the development of mammary tumors in rodents: a review. Cancer Research 57, 3979-3988. Ferraroni, M., Decarli, A., Willett, W. C., and Marubini, E. (1991). Alcohol and breast cancer risk: a case-control study from northern Italy. International Journal of Epidemiology 20, 859-864. Ford-Hutchinson, A. W., Gresser, M., and Young, R. N. (1994). 5-Lipoxygenase. Annual Review of Biochemistry 63, 383-417. Franceschi, S., Favero, A., Decarli, A., Negri, E., La Vecchia, C., Ferraroni, M., Russo, A., Salvini, S., Amadori, D., Conti, E., and et al. (1996). Intake of macronutrients and risk of breast cancer. Lancet 347, 1351-1356. Freedman, L. S., Schatzkin, A., and Wax, Y. (1990). Analysis of dietary fat, calories, body weight, and the development of mammary tumors in rats and mice: a review. American Journal of Epidemiology 132, 1185-1195. Funk, C. D. (1996). The molecular biology of mammalian lipoxygenases and the quest for eicosanoid functions using lipoxygenase-deficient mice. Biochimica et Biophysica Acta 1304, 65-84. 164 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Funk, C. D., Furci, L., and Fitzgerald, G. A. (1990). Molecular cloning of the human platelet 12-lipoxygenase. Transactions of the Association of American Physicians 103, 180-186. Gaard, M., Tretli, S., and Loken, E. B. (1995). Dietary fat and the risk of breast cancer: a prospective study o f25,892 Norwegian women. International Journal of Cancer 63, 13-17. Gago-Dominguez, M., Yuan, J. M., Sun, C. L., Lee, H. P., and Yu, M. C. (2003). Opposing effects of dietary n-3 and n-6 fatty acids on mammary carcinogenesis: The Singapore Chinese Health Study. Br J Cancer 89, 1686-1692. Gao, X., Grignon, D. J., Chbihi, T., Zacharek, A., Chen, Y. Q., Sakr, W., Porter, A. T., Crissman, J. D., Pontes, J. E., and Powell, I. J. (1995). Elevated 12-lipoxygenase mRNA expression correlates with advanced stage and poor differentiation of human prostate cancer. Urology 46, 227-237. Giovannucci, E., Rimm, E. B., Colditz, G. A., Stampfer, M. J., Ascherio, A., Chute, C. C., and Willett, W. C. (1993). A prospective study of dietary fat and risk of prostate cancer. Journal of the National Cancer Institute 8 5 ,1571-1579. Goodman, M. T., Nomura, A. M., Wilkens, L. R., and Hankin, J. (1992). The association of diet, obesity, and breast cancer in Hawaii. Cancer Epidemiology, Biomarkers & Prevention 1, 269-275. Goodstine, S. L., Zheng, T., Holford, T. R., Ward, B. A., Carter, D., Owens, P. H., and Mayne, S. T. (2003). Dietary (n-3)/(n-6) fatty acid ratio: possible relationship to premenopausal but not postmenopausal breast cancer risk in U.S. women. Journal of Nutrition 133, 1409-1414. Graham, S., Marshall, J., Freudenheim, J., Vena, J., Swanson, M., Zielezny, M., Nemoto, T., Stubbe, N., and Raimondo, T. (1991). Nutritional epidemiology of postmenopausal breast cancer in western New York. American Journal of Epidemiology 134, 552-566. Graham, S., Zielezny, M., Marshall, J., Priore, R., Freudenheim, J., Brasure, J., Haughey, B., Nasca, P., and Zdeb, M. (1992). Diet in the epidemiology of postmenopausal breast cancer in the New York State Cohort. American Journal of Epidemiology 136, 1327-1337. Grammatikos, S. I., Subbaiah, P. V., Victor, T. A., and Miller, W. M. (1994). n-3 and n-6 fatty acid processing and growth effects in neoplastic and non-cancerous human mammary epithelial cell lines. British Journal of Cancer 70, 219-227. 165 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Greenwald, P. (1999). Role of dietary fat in the causation of breast cancer: point. Cancer Epidemiol Biomarkers Prev 8, 3-7. Gurr, M. I., and Harwood, J. L. (1991). Dietary lipids: implications for health and disease. In Lipid Biochemistry: An Introduction. Essential fatty acids and biosynthesis of eicosanoids. (Chapman & Hall), pp. 163. Hagmann, W., Gao, X., Zacharek, A., Wojciechowski, L. A., and Honn, K. V. (1995). 12-Lipoxygenase in Lewis lung carcinoma cells: molecular identity, intracellular distribution of activity and protein, and Ca(2+)-dependent translocation from cytosol to membranes. Prostaglandins 49 ,49-62. Haliday, E. M., Ramesha, C. S., and Ringold, G. (1991). TNF induces c-fos via a novel pathway requiring conversion of arachidonic acid to a lipoxygenase metabolite. EMBO Journal 10, 109-115. Hardy, S., Langelier, Y., and Prentki, M. (2000). Oleate activates phosphatidylinositol 3-kinase and promotes proliferation and reduces apoptosis of MDA-MB-231 breast cancer cells, whereas palmitate has opposite effects. Cancer Res 60, 6353-6358. Harrison, R. A., and Waterbor, J. W. (1999). Understanding meta-analysis in cancer epidemiology: dietary fat and breast cancer. Cancer Detection & Prevention 23, 97- 106. Henderson, B. E., Pike, M. C., Bernstein, L., and Ross, R. K. (1996). Breast cancer. In Cancer epidemiology and prevention, S. D. a. F. JF, ed. (NY: Oxford Univ. Press), pp. 1022-1039. Hirohata, T., Nomura, A. M., Hankin, J. H., Kolonel, L. N., and Lee, J. (1987). An epidemiologic study on the association between diet and breast cancer. Journal of the National Cancer Institute 78, 595-600. Holmberg, L., Ohlander, E. M., Byers, T., Zack, M., Wolk, A., Bergstrom, R., Bergkvist, L., Thurfjell, E., Bruce, A., and Adami, H. O. (1994). Diet and breast cancer risk. Results from a population-based, case-control study in Sweden. Archives of Internal Medicine 154,1805-1811. Holmes, M. D., Hunter, D. J., Colditz, G. A., Stampfer, M. J., Hankinson, S. E., Speizer, F. E., Rosner, B., and Willett, W. C. (1999). Association of dietary intake of fat and fatty acids with risk of breast cancer. JAMA 281, 914-920. Hong, S. H., Avis, I., Vos, M. D., Martinez, A., Treston, A. M., and Mulshine, J. L. (1999). Relationship of arachidonic acid metabolizing enzyme expression in epithelial cancer cell lines to the growth effect of selective biochemical inhibitors. Cancer Research 59, 2223-2228. 166 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Horn, K. V., Tang, D. G., Grossi, I. M., Renaud, C., Duniec, Z. M., Johnson, C. R., and Diglio, C. A. (1994). Enhanced endothelial cell retraction mediated by 12(S)- HETE: a proposed mechanism for the role of platelets in tumor cell metastasis. Experimental Cell Research 210, 1-9. Hom-Ross, P. L. (2001). Assessing phytoestrogen exposure via a food-frequency questionnaire. Cancer Causes & Control 12, 477-478. Hoshiko, S., Radmark, O., and Samuelsson, B. (1990). Characterization of the human 5-lipoxygenase gene promoter. Proceedings of the National Academy of Sciences of the United States of America 87, 9073-9077. Howe, G. R., Friedenreich, C. M., Jain, M., and Miller, A. B. (1991). A cohort study of fat intake and risk of breast cancer. Journal of the National Cancer Institute 83, 336- 340. Howe, G. R., Hirohata, T., Hislop, T. G., Iscovich, J. M., Yuan, J. M., Katsouyanni, K., Lubin, F., Marubini, E., Modan, B., and Rohan, T. (1990). Dietary factors and risk of breast cancer: combined analysis of 12 case-control studies. Journal of the National Cancer Institute 82, 561-569. Huang, M. H., Schocken, M., Block, G., Sowers, M., Gold, E., Stemfeld, B., Seeman, T., and Greendale, G. A. (2002). Variation in nutrient intakes by ethnicity: results from the Study of Women's Health Across the Nation (SWAN). Menopause 9, 309- 319. Hubbard, N. E., Lim, D., and Erickson, K. L. (1998). Alteration of murine mammary tumorigenesis by dietary enrichment with n-3 fatty acids in fish oil. Cancer Letters 124, 1-7. Hunter, D. J. (1999). Role of dietary fat in the causation of breast cancer: counterpoint. Cancer Epidemiology, Biomarkers & Prevention 8, 9-13. Hunter, D. J., Spiegelman, D., Adami, H. O., Beeson, L., van den Brandt, P. A., Folsom, A. R., Fraser, G. E., Goldbohm, R. A., Graham, S., Howe, G. R., and Willett, W. C. (1996). Cohort studies of fat intake and the risk of breast cancer— a pooled analysis. New England Journal of Medicine 334, 356-361. Hunter, D. J., Spiegelman, D., Adami, H. O., van den Brandt, P. A., Folsom, A. R., Goldbohm, R. A., Graham, S., Howe, G. R., Kushi, L. H., Marshall, J. R., et al. (1997). Non-dietary factors as risk factors for breast cancer, and as effect modifiers of the association of fat intake and risk of breast cancer. Cancer Causes & Control 8 ,49- 56. 167 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In, K. H., Asano, K., Beier, D., Grobholz, J., Finn, P. W., Silvennan, E. K., Silverman, E. S., Collins, T., Fischer, A. R., Keith, T. P., et al. (1997). Naturally occurring mutations in the human 5-lipoxygenase gene promoter that modify transcription factor binding and reporter gene transcription. Journal of Clinical Investigation 99, 1130- 1137. Ingram, D. M., Nottage, E., and Roberts, T. (1991). The role of diet in the development of breast cancer: a case-control study of patients with breast cancer, benign epithelial hyperplasia and fibrocystic disease of the breast. British Journal of Cancer 64, 187-191. Ip, C. (1997). Review of the effects of trans fatty acids, oleic acid, n-3 polyunsaturated fatty acids, and conjugated linoleic acid on mammary carcinogenesis in animals. American Journal of Clinical Nutrition 6 6 ,1523S-1529S. Ip, C., Scimeca, J. A., and Thompson, H. J. (1994). Conjugated linoleic acid. A powerful anticarcinogen from animal fat sources. Cancer 74,1050-1054. Jiang, W. G., Douglas-Jones, A., and Mansel, R. E. (2003). Levels of expression of lipoxygenases and cyclooxygenase-2 in human breast cancer. Prostaglandins Leukot Essent Fatty Acids 69, 275-281. Jones, D. Y., Schatzkin, A., Green, S. B., Block, G., Brinton, L. A., Ziegler, R. G., Hoover, R., and Taylor, P. R. (1987). Dietary fat and breast cancer in the National Health and Nutrition Examination Survey I Epidemiologic Follow-up Study. Journal of the National Cancer Institute 79,465-471. Kamano, K., Okuyama, H., Konishi, R., and Nagasawa, H. (1989). Effects of a high- linoleate and a high-alpha-linolenate diet on spontaneous mammary tumourigenesis in mice. Anticancer Research 9, 1903-1908. Kantarci, A., and Van Dyke, T. E. (2003). Lipoxins in chronic inflammation. Crit Rev Oral Biol Med 14, 4-12. Karmali, R. A., Marsh, J., and Fuchs, C. (1984). Effect of omega-3 fatty acids on growth of a rat mammary tumor. Journal of the National Cancer Institute 73, 457-461. Kato, I., Miura, S., Kasumi, F., Iwase, T., Tashiro, H., Fujita, Y., Koyama, H., Ikeda, T., Fujiwara, K., and Saotome, K. (1992). A case-control study of breast cancer among Japanese women: with special reference to family history and reproductive and dietary factors. Breast Cancer Research & Treatment 24, 51-59. Katsouyanni, K., Willett, W., Trichopoulos, D., Boyle, P., Trichopoulou, A., Vasilaros, S., Papadiamantis, J., and MacMahon, B. (1988). Risk of breast cancer among Greek women in relation to nutrient intake. Cancer 6 1 ,181-185. 168 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kipnis, V., Freedman, L. S., Brown, C. C., Hartman, A. M., Schatzkin, A., and Wacholder, S. (1997). Effect of measurement error on energy-adjustment models in nutritional epidemiology. Am J Epidemiol 146, 842-855. Kliewer, S. A., Sundseth, S. S., Jones, S. A., Brown, P. J., Wisely, G. B., Koble, C. S., Devchand, P., Wahli, W., Willson, T. M., Lenhard, J. M., and Lehmann, J. M. (1997). Fatty acids and eicosanoids regulate gene expression through direct interactions with peroxisome proliferator-activated receptors alpha and gamma. Proceedings of the National Academy of Sciences of the United States of America 94, 4318-4323. Knekt, P., Albanes, D., Seppanen, R., Aromaa, A., Jarvinen, R., Hyvonen, L., Teppo, L., and Pukkala, E. (1990). Dietary fat and risk of breast cancer. American Journal of Clinical Nutrition 52, 903-908. Kohlmeier, L. (1997). Biomarkers of fatty acid exposure and breast cancer risk. American Journal of Clinical Nutrition 66, 1548S-1556S. Kolonel, L. N., Henderson, B. E., Hankin, J. H., Nomura, A. M., Wilkens, L. R., Pike, M. C., Stram, D. O., Monroe, K. R., Earle, M. E., and Nagamine, F. S. (2000). A multiethnic cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol 151, 346-357. Koshino, T., Takano, S., Kitani, S., Ohshima, N., Sano, Y., Takaishi, T., Hirai, K., Yamamoto, K., and Morita, Y. (1999). Novel polymorphism of the 5-lipoxygenase activating protein (FLAP) promoter gene associated with asthma. Molecular Cell Biology Research Communications 2, 32-35. Kris-Etherton, P. M., Taylor, D. S., Yu-Poth, S., Huth, P., Moriarty, K., Fishell, V., Hargrove, R. L., Zhao, G., and Etherton, T. D. (2000). Polyunsaturated fatty acids in the food chain in the United States. American Journal of Clinical Nutrition 7 1 ,179S- 188S. Kritchevsky, D., and Klurfeld, D. M. (1987). Caloric effects in experimental mammary tumorigenesis. American Journal of Clinical Nutrition 45, 236-242. Kurahashi, Y., Herbertsson, H., Soderstrom, M., Rosenfeld, M. G., and Hammarstrom, S. (2000). A 12(S)-hydroxyeicosatetraenoic acid receptor interacts with steroid receptor coactivator-1. Proceedings of the National Academy of Sciences of the United States of America 97, 5779-5783. Kushi, L. H., Potter, J. D., Bostick, R. M., Drinkard, C. R., Sellers, T. A., Gapstur, S. M., Cerhan, J. R., and Folsom, A. R. (1995). Dietary fat and risk of breast cancer according to hormone receptor status. Cancer Epidemiology, Biomarkers & Prevention 4 , 11-19. 169 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Kushi, L. H., Sellers, T. A., Potter, J. D., Nelson, C. L., Munger, R. G., Kaye, S. A., and Folsom, A. R. (1992). Dietary fat and postmenopausal breast cancer. Journal of the National Cancer Institute 84, 1092-1099. La Vecchia, C., Favero, A., and Franceschi, S. (1998). Monounsaturated and other types of fat, and the risk of breast cancer. European Journal of Cancer Prevention 7, 461-464. La Vecchia, C., Negri, E., Franceschi, S., Decarli, A., Giacosa, A., and Lipworth, L. (1995). Olive oil, other dietary fats, and the risk of breast cancer (Italy). Cancer Causes & Control 6, 545-550. Landa, M. C., Frago, N., and Tres, A. (1994). Diet and the risk of breast cancer in Spain. European Journal of Cancer Prevention 3, 313-320. Lands, W. E., Hamazaki, T., Yamazaki, K., Okuyama, H., Sakai, K., Goto, Y., and Hubbard, V. S. (1990). Changing dietary patterns. American Journal of Clinical Nutrition 51, 991-993. Lasekan, J. B., Clayton, M. K., Gendron-Fitzpatrick, A., and Ney, D. M. (1990). Dietary olive and safflower oils in promotion of DMBA-induced mammary tumorigenesis in rats. Nutr Cancer 13, 153-163. Lee, H. P., Gourley, L., Duffy, S. W., Esteve, J., Lee, J., and Day, N. E. (1991). Dietary effects on breast-cancer risk in Singapore. Lancet 337,1197-1200. Lee, M. M., and Lin, S. S. (2000). Dietary fat and breast cancer. Annual Review of Nutrition 20, 221-248. Levi, F., La Vecchia, C., Gulie, C., and Negri, E. (1993). Dietary factors and breast cancer risk in Vaud, Switzerland. Nutrition & Cancer 19, 327-335. Liu, B., Mamett, L. J., Chaudhary, A., Ji, C., Blair, I. A., Johnson, C. R., Diglio, C. A., and Honn, K. V. (1994). Biosynthesis of 12(S)-hydroxyeicosatetraenoic acid by B16 amelanotic melanoma cells is a determinant of their metastatic potential. Laboratory Investigation 70, 314-323. Liu, B., Timar, J., Howlett, J., Diglio, C. A., and Honn, K. V. (1991). Lipoxygenase metabolites of arachidonic and linoleic acids modulate the adhesion of tumor cells to endothelium via regulation of protein kinase C. Cell Regulation 2, 1045-1055. Liu, X. H., Connolly, J. M., and Rose, D. P. (1996a). The 12-lipoxygenase gene- transfected MCF-7 human breast cancer cell line exhibits estrogen-independent, but estrogen and omega-6 fatty acid-stimulated proliferation in vitro, and enhanced growth in athymic nude mice. Cancer Letters 109, 223-230. 170 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Liu, X. H., Connolly, J. M., and Rose, D. P. (1996b). Eicosanoids as mediators of linoleic acid-stimulated invasion and type IV collagenase production by a metastatic human breast cancer cell line. Clinical & Experimental Metastasis 14, 145-152. London, S. J., Sacks, F. M., Stampfer, M. J., Henderson, I. C., Maclure, M., Tomita, A., Wood, W. C., Remine, S., Robert, N. J., and Dmochowski, J. R. (1993). Fatty acid composition of the subcutaneous adipose tissue and risk of proliferative benign breast disease and breast cancer. Journal of the National Cancer Institute 85, 785-793. Maillard, V., Bougnoux, P., Ferrari, P., Jourdan, M. L., Pinault, M., Lavillonniere, F., Body, G., Le Floch, O., and Chajes, V. (2002). N-3 and N-6 fatty acids in breast adipose tissue and relative risk of breast cancer in a case-control study in Tours, France. International Journal of Cancer 98, 78-83. Mancini, J. A., Abramovitz, M., Cox, M. E., Wong, E., Charleson, S., Perrier, H., Wang, Z., Prasit, P., and Vickers, P. J. (1993). 5-lipoxygenase-activating protein is an arachidonate binding protein. FEBS Letters 318, 277-281. Mancini, J. A., Waterman, H., and Riendeau, D. (1998). Cellular oxygenation of 12- hydroxyeicosatetraenoic acid and 15-hydroxyeicosatetraenoic acid by 5-lipoxygenase is stimulated by 5-lipoxygenase-activating protein. Journal of Biological Chemistry 273, 32842-32847. Mander, A. P. (2001). Haplotype analysis. The STATA journal 1, 58-75. Mannisto, S., Pietinen, P., Virtanen, M., Kataja, V., and Uusitupa, M. (1999). Diet and the risk of breast cancer in a case-control study: does the threat of disease have an influence on recall bias? J Clin Epidemiol 52,429-439. Martin-Moreno, J. M., Willett, W. C., Gorgojo, L., Banegas, J. R., Rodriguez-Artalejo, F., Femandez-Rodriguez, J. C., Maisonneuve, P., and Boyle, P. (1994). Dietary fat, olive oil intake and breast cancer risk. International Journal of Cancer 58, 774-780. Matsuba, S., Itoh, M., Joh, T., Takeyama, H., Dohi, N., Watanabe, S., and Okuyama, H. (1998). Effect of dietary linoleate/alpha-linolenate balance on experimentally induced gastric injury in rats. Prostaglandins Leukotrienes & Essential Fatty Acids 59, 317-323. McDowell, M. A. (1994). The NHANES III Supplemental Nutrition Survey of older Americans. American Journal of Clinical Nutrition 59, 224S-226S. Meschter, C. L., Connolly, J. M., and Rose, D. P. (1992). Influence of regional location of the inoculation site and dietary fat on the pathology of MDA-MB-435 human breast cancer cell-derived tumors grown in nude mice. Clinical & Experimental Metastasis 1 0 ,167-173. 171 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Mills, P. K., Beeson, W. L., Phillips, R. L., and Fraser, G. E. (1989). Dietary habits and breast cancer incidence among Seventh-day Adventists. Cancer 64, 582-590. Nair, J., Furstenberger, G., Burger, F., Marks, F., and Bartsch, H. (2000). Promutagenic etheno-DNA adducts in multistage mouse skin carcinogenesis: correlation with lipoxygenase-catalyzed arachidonic acid metabolism. Chemical Research in Toxicology 13, 703-709. Natarajan, R., Esworthy, R., Bai, W., Gu, J. L., Wilczynski, S., and Nadler, J. (1997). Increased 12-lipoxygenase expression in breast cancer tissues and cells. Regulation by epidermal growth factor. Journal of Clinical Endocrinology & Metabolism 82, 1790- 1798. Nie, D., Hillman, G. G., Geddes, T., Tang, K., Pierson, C., Grignon, D. J., and Honn, K. V. (1999). Platelet-type 12-lipoxygenase regulates angiogenesis in human prostate carcinoma. Advances in Experimental Medicine & Biology 469, 623-630. Nie, D., Tang, K., Szekeres, K., Li, L., and Honn, K. V. (2000). Eicosanoid regulation of angiogenesis in human prostate carcinoma and its therapeutic implications. Annals of the New York Academy of Sciences 905, 165-176. Ogmundsdottir, H. M., Zoega, G. M., Gissurarson, S. R., and Ingolfsdottir, K. (1998). Anti-proliferative effects of lichen-derived inhibitors of 5-lipoxygenase on malignant cell-lines and mitogen-stimulated lymphocytes. Journal of Pharmacy & Pharmacology 5 0 ,107-115. Ostareck-Lederer, A., Ostareck, D. H., Standart, N., and Thiele, B. J. (1994). Translation of 15-lipoxygenase mRNA is inhibited by a protein that binds to a repeated sequence in the 3' untranslated region. EMBO Journal 1 3 ,1476-1481. Ozeki, Y., Nagamura, Y., Ito, H., Unemi, F., Kimura, Y., Igawa, T., Kambayashi, J., Takahashi, Y., and Yoshimoto, T. (1999). An anti-platelet agent, OPC-29030, inhibits translocation of 12-lipoxygenase and 12-hydroxyeicosatetraenoic acid production in human platelets. British Journal of Pharmacology 128, 1699-1704. Pala, V., Krogh, V., Muti, P., Chajes, V., Riboli, E., Micheli, A., Saadatian, M., Sieri, S., and Berrino, F. (2001). Erythrocyte membrane fatty acids and subsequent breast cancer: a prospective Italian study. Journal of the National Cancer Institute 93, 1088- 1095. Pariza, M. W. (1987). Fat, calories, and mammary carcinogenesis: net energy effects. American Journal of Clinical Nutrition 45, 261-263. 172 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Petrek, J. A., Hudgins, L. C., Levine, B., Ho, M., and Hirsch, J. (1994). Breast cancer risk and fatty acids in the breast and abdominal adipose tissues. Journal of the National Cancer Institute 86, 53-56. Pidgeon, G. P., Tang, K., Cai, Y. L., Piasentin, E., and Honn, K. V. (2003). Overexpression of platelet-type 12-lipoxygenase promotes tumor cell survival by enhancing alpha(v)beta(3) and alpha(v)beta(5) integrin expression. Cancer Research 63, 4258-4267. Piper, P. J., and Samhoun, M. N. (1987). Leukotrienes. British Medical Bulletin 43, 297-311. Potischman, N., Swanson, C. A., Coates, R. J., Weiss, H. A., Brogan, D. R., Stanford, J. L., Schoenberg, J. B., Gammon, M. D., and Brinton, L. A. (1997). Dietary relationships with early onset (under age 45) breast cancer in a case-control study in the United States: influence of chemotherapy treatment. Cancer Causes Control 8, 713-721. Potischman, N., Weiss, H. A., Swanson, C. A., Coates, R. J., Gammon, M. D., Malone, K. E., Brogan, D., Stanford, J. L., Hoover, R. N., and Brinton, L. A. (1998). Diet during adolescence and risk of breast cancer among young women. Journal of the National Cancer Institute 90, 226-233. Prentice, R. L., and Sheppard, L. (1990). Dietary fat and cancer: consistency of the epidemiologic data, and disease prevention that may follow from a practical reduction in fat consumption. Cancer Causes & Control 1, 81-97; discussion 99-109. Pryor, M., Slattery, M. L., Robison, L. M., and Egger, M. (1989). Adolescent diet and breast cancer in Utah. Cancer Research 49, 2161-2167. Przylipiak, A., Hafner, J., Przylipiak, J., Kohn, F. M., Runnebaum, B., and Rabe, T. (1998a). Influence of 5-lipoxygenase on in vitro growth of human mammary carcinoma cell line MCF-7. Gynecologic & Obstetric Investigation 46, 61-64. Przylipiak, A., Hafner, J., Przylipiak, J., Runnebaum, B., Rabe, T., and Kohn, F. M. (1998b). Influence of leukotrienes on in vitro growth of human mammary carcinoma cell line MCF-7. European Journal of Obstetrics, Gynecology, & Reproductive Biology 77, 61-65. Qi, X. Y., Zhang, A. Y., Wu, G. L., and Pang, W. Z. (1994). The association between breast cancer and diet and other factors. Asia-Pacific Journal of Public Health 7, 98- 104. permission of the copyright owner. Further reproduction prohibited without permission. Razanamahefa, L., Prouff, S., and Bardon, S. (2000). Stimulatory effect of arachidonic acid on T-47D human breast cancer cell growth is associated with enhancement of cyclin D1 mRNA expression. Nutrition & Cancer 38, 274-280. Reddy, N., Everhart, A., Eling, T., and Glasgow, W. (1997). Characterization of a 15- lipoxygenase in human breast carcinoma BT-20 cells: stimulation of 13-HODE formation by TGF alpha/EGF. Biochemical & Biophysical Research Communications 231, 111-116. Richardson, S., Gerber, M., and Cenee, S. (1991). The role of fat, animal protein and some vitamin consumption in breast cancer: a case control study in southern France. International Journal of Cancer 4 8 ,1-9. Rissanen, H., Knekt, P., Jarvinen, R., Salminen, I., and Hakulinen, T. (2003). Serum fatty acids and breast cancer incidence. Nutrition & Cancer 45, 168-175. Rohan, T. E., McMichael, A. J., and Baghurst, P. A. (1988). A population-based case- control study of diet and breast cancer in Australia. American Journal of Epidemiology 128, 478-489. Ronco, A., De Stefani, E., Mendilaharsu, M., and Deneo-Pellegrini, H. (1996). Meat, fat and risk of breast cancer: a case-control study from Uruguay. International Journal of Cancer 65, 328-331. Rose, D. P. (1997). Effects of dietary fatty acids on breast and prostate cancers: evidence from in vitro experiments and animal studies. Am J Clin Nutr 66, 1513S- 1522S. Rose, D. P., and Connolly, J. M. (1990). Effects of fatty acids and inhibitors of eicosanoid synthesis on the growth of a human breast cancer cell line in culture. Cancer Research 50, 7139-7144. Rose, D. P., Connolly, J. M., Rayburn, J., and Coleman, M. (1995). Influence of diets containing eicosapentaenoic or docosahexaenoic acid on growth and metastasis of breast cancer cells in nude mice. Journal of the National Cancer Institute 87, 587-592. Rose, D. P., Hatala, M. A., Connolly, J. M., and Rayburn, J. (1993). Effect of diets containing different levels of linoleic acid on human breast cancer growth and lung metastasis in nude mice. Cancer Research 53, 4686-4690. Sanders, T. A. (2000). Polyunsaturated fatty acids in the food chain in Europe. American Journal of Clinical Nutrition 71,176S-178S. Sayers, I., Barton, S., Rorke, S., Sawyer, J., Peng, Q., Beghe, B., Ye, S., Keith, T., Clough, J. B., Holloway, J. W., et al. (2003). Promoter polymorphism in the 5- 174 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. lipoxygenase (ALOX5) and 5-lipoxygenase-activating protein (ALOX5AP) genes and asthma susceptibility in a Caucasian population. Clin Exp Allergy 3 3 ,1103-1110. Selenskas, S. L., Ip, M. M., and Ip, C. (1984). Similarity between trans fat and saturated fat in the modification of rat mammary carcinogenesis. Cancer Res 44, 1321 - 1326. Serhan, C. N., and Oliw, E. (2001). Unorthodox routes to prostanoid formation: new twists in cyclooxygenase-initiated pathways. Journal of Clinical Investigation 107, 1481-1489. Shimizu, H., Ross, R. K., Bernstein, L., Yatani, R., Henderson, B. E., and Mack, T. M. (1991). Cancers of the prostate and breast among Japanese and white immigrants in Los Angeles County. British Journal of Cancer 63,963-966. Shoda, R., Matsueda, K., Yamato, S., and Umeda, N. (1996). Epidemiologic analysis of Crohn disease in Japan: increased dietary intake of n-6 polyunsaturated fatty acids and animal protein relates to the increased incidence of Crohn disease in Japan. American Journal of Clinical Nutrition 63, 741-745. Sieri, S., Krogh, V., Muti, P., Micheli, A., Pala, V., Crosignani, P., and Berrino, F. (2002). Fat and protein intake and subsequent breast cancer risk in postmenopausal women. Nutrition & Cancer 42, 10-17. Silverman, E. S., Du, J., De Sanctis, G. T., Radmark, O., Samuelsson, B., Drazen, J. M., and Collins, T. (1998). Egr-1 and Spl interact functionally with the 5- lipoxygenase promoter and its naturally occurring mutants. American Journal of Respiratory Cell & Molecular Biology 19, 316-323. Simonsen, N., van't Veer, P., Strain, J. J., Martin-Moreno, J. M., Huttunen, J. K., Navajas, J. F., Martin, B. C., Thamm, M., Kardinaal, A. F., Kok, F. J., and Kohlmeier, L. (1998). Adipose tissue omega-3 and omega-6 fatty acid content and breast cancer in the EURAMIC study. European Community Multicenter Study on Antioxidants, Myocardial Infarction, and Breast Cancer. American Journal of Epidemiology 147, 342-352. Singh, R. K., Hardy, R. W., Wang, M. H., Williford, J., Gladson, C. L., McDonald, J. M., and Siegal, G. P. (1995). Stearate inhibits human tumor cell invasion. Invasion Metastasis 15, 144-155. Slattery, M. L., O'Brien, E., and Mori, M. (1995). Disease heterogeneity: does it impact our ability to detect dietary associations with breast cancer? Nutr Cancer 24, 213-220. 175 permission of the copyright owner. Further reproduction prohibited without permission. Smith-Wamer, S. A., Spiegelman, D., Adami, H. O., Beeson, W. L., van den Brandt, P. A., Folsom, A. R., Fraser, G. E., Freudenheim, J. L., Goldbohm, R. A., Graham, S., et al. (2001). Types of dietary fat and breast cancer: a pooled analysis of cohort studies. International Journal of Cancer 92,161-11 A. Sobell, J., Block, G., Koslowe, P., Tobin, J., and Andres, R. (1989). Validation of a retrospective questionnaire assessing diet 10-15 years ago. Am J Epidemiol 130, 173- 187. Solanas, M., Hurtado, A., Costa, I., Moral, R., Menendez, J. A., Colomer, R., and Escrich, E. (2002). Effects of a high olive oil diet on the clinical behavior and histopathological features of rat DMBA-induced mammary tumors compared with a high com oil diet. Int J Oncol 21, 745-753. Spector, A. A., Gordon, J. A., and Moore, S. A. (1988). Hydroxyeicosatetraenoic acids (HETEs). Progress in Lipid Research 27, 271-323. Steele, V. E., Holmes, C. A., Hawk, E. T., Kopelovich, L., Lubet, R. A., Crowell, J. A., Sigman, C. C., and Kelloff, G. J. (1999). Lipoxygenase inhibitors as potential cancer chemopreventives. Cancer Epidemiology, Biomarkers & Prevention 8 ,467- 483. Subar, A. F., Thompson, F. E., Kipnis, V., Midthune, D., Hurwitz, P., McNutt, S., McIntosh, A., and Rosenfeld, S. (2001). Comparative validation of the Block, Willett, and National Cancer Institute food frequency questionnaires : the Eating at America's Table Study. Am J Epidemiol 154, 1089-1099. Sugano, M., and Hirahara, F. (2000). Polyunsaturated fatty acids in the food chain in Japan. American Journal of Clinical Nutrition 71, 189S-196S. Swallen, K. C., West, D. W., Stewart, S. L., Glaser, S. L., and Hom-Ross, P. L. (1997). Predictors of misclassification of Hispanic ethnicity in a population-based cancer registry. Ann Epidemiol 7, 200-206. Tang, K., and Honn, K. V. (1997). Lipoxygenase metabolites and cancer metastasis. Advances in Experimental Medicine & Biology 422, 71-84. Tang, K., and Honn, K. V. (1999). 12(S)-HETE in cancer metastasis. Advances in Experimental Medicine & Biology 447, 181-191. Terry, P. D., Rohan, T. E., and Wolk, A. (2003). Intakes of fish and marine fatty acids and the risks of cancers of the breast and prostate and of other hormone-related cancers: a review of the epidemiologic evidence. Am J Clin Nutr 77, 532-543. 176 permission of the copyright owner. Further reproduction prohibited without permission. Timar, J., Raso, E., Honn, K. V., and Hagmann, W. (1999). 12-lipoxygenase expression in human melanoma cell lines. Advances in Experimental Medicine & Biology 469, 617-622. Tinsley, I. J., Schmitz, J. A., and Pierce, D. A. (1981). Influence of dietary fatty acids on the incidence of mammary tumors in the C3H mouse. Cancer Research 41, 1460- 1465. Tong, W. G., Ding, X. Z., and Adrian, T. E. (2002). The mechanisms of lipoxygenase inhibitor-induced apoptosis in human breast cancer cells. Biochem Biophys Res Commun 296, 942-948. Toniolo, P., Riboli, E., Protta, F., Charrel, M., and Cappa, A. P. (1989). Calorie- providing nutrients and risk of breast cancer. Journal of the National Cancer Institute 81, 278-286. Toniolo, P., Riboli, E., Shore, R. E., Pasternack, B. S., and Di Blasio, B. (1994). Consumption of meat, animal products, protein, and fat and risk of breast cancer: a prospective cohort study in New York. Epidemiology 5, 391-397. Tretli, S., and Gaard, M. (1996). Dietary fat and the risk of breast cancer: a prospective study of 25,892 Norwegian women. Cancer Causes & Control 7, 507-512. Trichopoulou, A., Katsouyanni, K., Stuver, S., Tzala, L., Gnardellis, C., Rimm, E., Trichopoulos, D., and Trichopoulou, A. (1995). Consumption of olive oil and specific food groups in relation to breast cancer risk in Greece. Journal of the National Cancer Institute 87, 110-116. Trichopoulou, A., Lagiou, P., Kuper, H., and Trichopoulos, D. (2000). Cancer and Mediterranean dietary traditions. Cancer Epidemiology, Biomarkers & Prevention 9, 869-873. Uhl, J., Klan, N., Rose, M., Entian, K. D., Werz, O., and Steinhilber, D. (2002). The 5- lipoxygenase promoter is regulated by DNA methylation. J Biol Chem 277, 4374- 4379. van den Brandt, P. A., van't Veer, P., Goldbohm, R. A., Dorant, E., Volovics, A., Hermus, R. J., and Sturmans, F. (1993). A prospective cohort study on dietary fat and the risk of postmenopausal breast cancer. Cancer Research 53, 75-82. van Leyen, K. (1998). Pictures in cell biology. Un-LOXing the door to organelle degradation. Trends Cell Biol 8 ,436. 177 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Vane, J. R., and Botting, R. M. (1994). Biological properties of cyclooxygenase products. In Handbook of Immunophrmacology: Lipid Mediators, C. F.M., ed. (San Diego, Academic Press), pp. 62-97. Van't Veer, P., Kok, F. J., Brants, H. A., Ockhuizen, T., Sturmans, F., and Hermus, R. J. (1990). Dietary fat and the risk of breast cancer. International Journal of Epidemiology 19, 12-18. Vatten, L. J., Bjerve, K. S., Andersen, A., and Jellum, E. (1993). Polyunsaturated fatty acids in serum phospholipids and risk of breast cancer: a case-control study from the Janus serum bank in Norway. European Journal of Cancer 29A, 532-538. Velie, E., Kulldorff, M., Schairer, C., Block, G., Albanes, D., and Schatzkin, A. (2000). Dietary fat, fat subtypes, and breast cancer in postmenopausal women: a prospective cohort study. Journal of the National Cancer Institute 92, 833-839. Voorrips, L. E., Brants, H. A., Kardinaal, A. F., Hiddink, G. J., van den Brandt, P. A., and Goldbohm, R. A. (2002). Intake of conjugated linoleic acid, fat, and other fatty acids in relation to postmenopausal breast cancer: the Netherlands Cohort Study on Diet and Cancer. American Journal of Clinical Nutrition 76, 873-882. Wachtershauser, A., Steinhilber, D., Loitsch, S. M., and Stein, J. (2000). Expression of 5-lipoxygenase by human colorectal carcinoma Caco-2 cells during butyrate-induced cell differentiation. Biochemical & Biophysical Research Communications 268, 778- 783. Wakai, K., Dillon, D. S., Ohno, Y., Prihartono, J., Budiningsih, S., Ramli, M., Darwis, I., Tjindarbumi, D., Tjahjadi, G., Soetrisno, E., et al. (2000). Fat intake and breast cancer risk in an area where fat intake is low: a case-control study in Indonesia. International Journal of Epidemiology 29, 20-28. Welsch, C. W. (1992). Relationship between dietary fat and experimental mammary tumorigenesis: a review and critique. Cancer Research 52, 2040s-2048s. Willett, W., and Stampfer, M. J. (1998). Implications of total energy intake for epidemiologic analyses. In Nutritional Epidemiology (Oxford University Press), pp. 273. Willett, W. C. (1999). Dietary fat and breast cancer. Toxicological Sciences 52, 127- 146. Willett, W. C., Hunter, D. J., Stampfer, M. J., Colditz, G., Manson, J. E., Spiegelman, D., Rosner, B., Hennekens, C. H., and Speizer, F. E. (1992). Dietary fat and fiber in relation to risk of breast cancer. An 8-year follow-up. JAMA 268, 2037-2044. 178 permission of the copyright owner. Further reproduction prohibited without permission. Willett, W. C., Stampfer, M. J., Colditz, G. A., Rosner, B. A., Hennekens, C. H., and Speizer, F. E. (1987). Dietary fat and the risk of breast cancer. N Engl J Med 316, 22- 28. Witte, J. S., Ursin, G., Siemiatycki, J., Thompson, W. D., Paganini-Hill, A., and Haile, R. W. (1997). Diet and premenopausal bilateral breast cancer: a case-control study. Breast Cancer Research & Treatment 42, 243-251. Wolk, A., Bergstrom, R., Hunter, D., Willett, W., Ljung, H., Holmberg, L., Bergkvist, L., Bruce, A., and Adami, H. O. (1998). A prospective study of association of monounsaturated fat and other types of fat with risk of breast cancer. Archives of Internal Medicine 158, 41-45. Yu, S. Z., Lu, R. F., Xu, D. D., and Howe, G. R. (1990). A case-control study of dietary and nondietary risk factors for breast cancer in Shanghai. Cancer Research 50, 5017-5021. Yuan, J. M., Wang, Q. S., Ross, R. K., Henderson, B. E., and Yu, M. C. (1995). Diet and breast cancer in Shanghai and Tianjin, China. British Journal of Cancer 71, 1353- 1358. Zaridze, D., Lifanova, Y., Maximovitch, D., Day, N. E., and Duffy, S. W. (1991). Diet, alcohol consumption and reproductive factors in a case-control study of breast cancer in Moscow. International Journal of Cancer 48, 493-501. Zaridze, D. G., Chevchenko, V. E., Levtshuk, A. A., Lifanova, Y. E., and Maximovitch, D. M. (1990). Fatty acid composition of phospholipids in erythrocyte membranes and risk of breast cancer. International Journal of Cancer 45, 807-810. Zaykin, D. V., Westfall, P. H., Young, S. S., Kamoub, M. A., Wagner, M. J., and Ehm, M. G. (2002). Testing association of statistically inferred haplotypes with discrete and continuous traits in samples of unrelated individuals. Human Heredity 53, 79-91. Zock, P. L., and Katan, M. B. (1998). Linoleic acid intake and cancer risk: a review and meta-analysis. American Journal of Clinical Nutrition 68, 142-153. 179 with permission of the copyright owner. Further reproduction prohibited without permission

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Association of vitamin D receptor gene polymorphisms with colorectal adenoma

PDF

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

PDF

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

PDF

CYP17 polymorphism and risk for colorectal adenomas

PDF

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

PDF

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

PDF

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

PDF

Androgens and breast cancer

PDF

Genetic risk factors in breast cancer susceptibility: The multiethnic cohort

PDF

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

PDF

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

PDF

Family history, hormone replacement therapy and breast cancer risk on Hispanic and non-Hispanic women, The New Mexico Women's Health Study

PDF

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

PDF

A linear model for measurement errors in oligonucleotide microarray experiment

PDF

Comparisons of metabolic factors among gestational diabetes mellitus probands, siblings and cousins

PDF

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

PDF

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

PDF

Insulin-like growth factor 1 genotype, phenotype and breast cancer risk, by racial/ethnic group

PDF

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

Asset Metadata

Creator Wang, Jun (author)

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

School Graduate School

Degree Doctor of Philosophy

Degree Program Molecular Epidemiology

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

Tag health sciences, public health,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Ingles, Sue Ann (committee chair), Dubeau, Louis (committee member), Shih, Jean Chen (committee member), Yu, Mimi (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-666799

Unique identifier UC11340834

Identifier usctheses-c16-666799 (legacy record id)

Legacy Identifier 3145309.pdf

Dmrecord 666799

Document Type Dissertation

Rights Wang, Jun

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