Fluid intake, micturition habits, associated medications and conditions as potential risk factors for bladder cancer

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Request accessible transcript

Transcript (if available)

Content FLUID INTAKE, MICTURITION HABITS, ASSOCIATED MEDICATIONS AND CONDITIONS AS POTENTIAL RISK FACTORS FOR BLADDER CANCER by Xuejuan Jiang A Dissertation Presented to the FACULTY OF THE GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfillment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (EPIDEMIOLOGY) August 2007 Copyright 2007 Xuejuan Jiang ii DEDICATION To my parents, Jiang, Cheng-Qi & Zhang, Mei-Xian iii AKNOWLEDGEMENTS I would like to acknowledge the committee members of this dissertation who provided me with the guidance, insight, and support to conduct this research. Specifically, I thank Dr. Manuela Gago-Dominguez, David Conti, Susan Groshen, Victoria Cortessis, and Darryl Shibata who have given generously of their time and thought into the development and review of this research. My undiluted appreciation also extends to Dr. J. Esteban Castelao for his invaluable input and support. iv TABLE OF CONTENTS DEDICATION ii AKNOWLEDGEMENTS iii LIST OF TABLES vi ABSTRACT viii CHAPTER 1 INTRODUCTION 1 1.1 BLADDER CANCER 1 1.1.1 Age, Sex, and Race 1 1.1.2 Histopathology 2 1.1.3 Risk Factors 3 1.2 UROGENOUS-CONTACT HYPOTHESIS 5 CHAPTER 2 FLUID INTAKE AND BLADDER CANCER RISK 7 2.1 TOTAL FLUID INTAKE 7 2.1.1 Introduction 7 2.1.2 Materials and Methods 8 2.1.3 Results 11 2.2 WATER INTAKE 15 2.2.1 Introduction 15 2.2.2 Materials and Methods 15 2.2.3 Results 16 2.2.4 Discussion 26 2.3 ALCOHOLIC BEVERAGES 33 2.3.1 Introduction 33 2.3.2 Materials and Methods 34 2.3.3 Results 35 2.3.4 Discussion 45 2.4 COFFEE 51 2.4.1 Introduction 51 2.4.2 Materials and Methods 51 v 2.4.3 Results 52 2.4.4 Discussion 57 2.5 GENE-ENVIRONMENT INTERACTIONS 60 2.5.1 Introduction 60 2.5.2 Materials and Methods 62 2.5.3 Results 64 2.5.4 Discussion 72 CHAPTER 3 MEDICAL HISTORY, MEDICATION AND BLADDER CANCER RISK 75 3.1 URINARY TRACT INFECTIONS 75 3.1.1 Introduction 75 3.1.2 Materials and Methods 76 3.1.3 Results 77 3.1.4 Discussion 84 3.2 HYPERTENSION, DIURETICS, AND ANTIHYPERTENSIVES 94 3.2.1 Introduction 94 3.2.2 Materials and Methods 95 3.2.3 Results 97 3.2.4 Discussion 105 3.3 OTHER MEDICAL CONDITIONS 108 3.3.1 Introduction 108 3.3.2 Materials and Methods 108 3.3.3 Results 108 3.3.4 Discussion 113 CHAPTER 4 MICTURITION AND BLADDER CANCER RISK 117 4.1 INTRODUCTION 117 4.2 MATERIALS AND METHODS 118 4.3 RESULTS 119 4.4 DISCUSSION 125 BIBLIOGRAPHY 127 vi LIST OF TABLES Table 1. Fluid intake and risk of bladder cancer the Los Angeles Bladder Cancer Case-Control Study 13 Table 2. Water intake and bladder cancer risk by sex 19 Table 3. Water intake and bladder cancer risk by smoking status and sex 20 Table 4. Water intake and bladder cancer risk by frequency of urination 22 Table 5. Water intake and bladder cancer risk by nocturia among subjects with high ( ≥6 times/day) frequency of daytime urination 24 Table 6. Water intake and bladder cancer risk by BMI 25 Table 7. Basic characteristics of all the subjects included in the analyses 39 Table 8. Total alcohol consumption and risk of bladder cancer 41 Table 9. Effect of total alcohol consumption on risk of bladder cancer by smoking status and sex 42 Table 10. Total alcohol consumption and bladder cancer risk by daytime urination among subject who had smoked for less than 30 years 43 Table 11. Beverage-specific consumption and risk of bladder cancer 44 Table 12. Demographic characteristics of controls by coffee consumption. 54 Table 13. Effect of coffee consumption on risk of bladder cancer by smoking status and sex 55 Table 14. Effect of coffee consumption on risk of bladder cancer by age 56 Table 15. Alcohol and bladder cancer risk by GSTM1 genotype 66 Table 16. Alcohol and bladder cancer risk by GSTT1 genotype 67 Table 17. Alcohol and bladder cancer risk by GSTM1 and GSTT1 genotypes 68 Table 18. Alcohol and bladder cancer risk by CYP1A2 phenotype 69 Table 19. Coffee and bladder cancer risk by CYP1A2 phenotype 70 Table 20. Coffee and bladder cancer risk by NAT2 phenotype 71 vii Table 21. Urinary tract infections and risk of bladder cancer by sex 80 Table 22. Bladder infections and bladder cancer 81 Table 23. Bladder infections and bladder cancer by cigarette smoking and grade 82 Table 24. Diuretics, antihypertensive drugs, and hypertension in relation to bladder cancer risk 99 Table 25. Diuretics, antihypertensives, & hypertension in relation to the risk of bladder cancer 100 Table 26. Hypertension, diuretics/antihypertensives and risk of bladder cancer by sex 101 Table 27.Hypertension, diuretics/antihypertensives and risk of bladder cancer by cigarette smoking in men 102 Table 28. Hypertension, diuretics/antihypertensives and risk of bladder cancer by BMI in men 103 Table 29. Hypertension, diuretics/antihypertensives and risk of bladder cancer by GSTM1 genotype in men 104 Table 30. Selected medical conditions and risk of bladder cancer by sex. 110 Table 31. Diabetes and bladder cancer risk 111 Table 32. Angina, heart attack and bladder cancer risk by age in men 112 Table 33. Spearman correlation coefficients between frequency of urination and other risk/protective factors among controls 121 Table 34. Frequency of urination and bladder cancer 122 Table 35. Frequency of urination and risk of bladder cancer by sex 123 Table 36. Frequency of urination and risk of bladder cancer by use of diuretics and antihypertensives 124 viii ABSTRACT The urogenous-contact hypothesis proposes that the development of bladder cancer is associated with prolonged exposure to carcinogens in urine. To fully examine this hypothesis, this dissertation systematically addressed fluid intake, micturition habits, associated medical conditions, and medications in relation to risk of bladder cancer in the Los Angeles Bladder Cancer Case-Control Study. Total fluid intake was not associated with risk of bladder cancer, but specific types of beverages had different influence on bladder cancer. Consumption of water and alcoholic beverages were associated with a reduced risk of bladder cancer, whereas consumption of coffee was associated with a slightly increased risk, especially among heavy drinkers. The alcohol-bladder cancer association was modified by genetic variations in glutathione S-transferases, and the coffee-bladder cancer association was modified by variations in N-acetyltransferase 2 and cytochrome P450 1A2. This dissertation also examined the role of selected medical conditions and medications which have been shown to affect the intake of fluid and the frequency of urination. Women with a history of bladder infection had a lower risk of bladder cancer, while subjects with diabetes, heart attack, or gout exhibited a higher risk. A history of hypertension, and use of diuretics or antihypertensive drugs were not associated with bladder cancer; however, untreated hypertensive patients had a significantly reduction in bladder cancer risk compared to normotensive subjects, especially among men. ix This dissertation also addressed the role of urination frequency on bladder carcinogenesis. The role of urination frequency was also addressed in this dissertation. Frequency of daytime urination seemed to exert its effect on bladder cancer depending on the substrate. The inverse associations between water intake, alcohol intake, and bladder cancer risk were modified by frequency of daytime urination with the reduction in risk confined to those who urinated more frequently. However, frequency of urination seemed not to play a role in the bladder infection- bladder cancer association. Nocturia was associated with an increased risk of bladder cancer among women. 1 CHAPTER 1 INTRODUCTION 1.1 BLADDER CANCER In the US, an estimated 67,160 bladder cancer cases will occur in 2007, among which 50,040 will be male and 17,120 will be female (Jemal et al., 2007), making this malignancy the fourth most common cancer among men and the twelfth most common cancer among women (Jemal et al., 2005). Rates are higher in many European countries such as Italy where cigarette smoking is highly prevalent (Parkin et al., 2002). Bladder cancer, particularly squamous cell carcinoma, is also highly prevalent in some areas of Africa and the Middle East such as Egypt, where chronic infection due to Schistosoma haematobium is common (IARC, 1994). 1.1.1 Age, Sex, and Race Bladder cancer is rare before age 35 (Kirkali et al., 2005). The median age at diagnosis of bladder cancer is 65-70 years. Men are 3-4 times more likely to have bladder cancer than women (Ries et al., 2006), partially due to sex-specific differences in exposure to known risk factors, such as cigarette smoking and occupation. In addition, hormonal differences are suggested as possible reasons, based on the observations that pregnancy may protect women against bladder cancer and this protection may increase with increasing parity (Miller et al., 1980; Plesko et al., 1985; Green et al., 1988; Cantor et al., 1992). Bladder cancer is more often found at a higher stage at initial diagnosis in women than men (Ries et al., 2006). 2 There is a marked ethnic variation in bladder cancer incidence. It is least common among Native Americans and Asians, and most common among non- Hispanic Whites, with the age-adjusted rates varying from 7.2 to 24.2 per 100,000 persons during 2000-2003 (SEER) (Ries et al., 2006). The pattern is similar in men and women. The difference is unlikely to be caused by differential exposure to known risk factors, but may be due to racial variations in susceptibility to carcinogens, such as the capacity to activate and detoxify carcinogens, and/or to repair DNA. Genetic polymorphisms of metabolic enzymes, such as N-acetyl- transferases (NATs) and glutathione S-transferases (GSTs) have been proposed as possible explanations (Yu et al., 1994; Yu et al., 1995). When first diagnosed, bladder cancer is more likely to be localized in whites than in blacks (SEER 1996- 2002, 75% vs. 59%), followed also by better 5-year survival rates in whites than in blacks (81.4% vs. 63.8%) (Ries et al., 2006). This suggests that under-detection of low-stage tumors might be more common in blacks. 1.1.2 Histopathology Roughly 90% of bladder tumors in the United States (US) are of the transitional cell type, and 7% are squamous carcinomas (Yu & Ross, 2002). About 70% of bladder cancers occur in the posterior and lateral walls, near the trigone (a triangular region of the internal bladder formed by the two ureteral orifices and the internal urethral orifice) of the bladder, 20% arise in the trigone, and the remaining 10% arise in the dome (roof of the bladder) (Melicow, 1974). In the geographic areas 3 with endemic infections of Schistosoma haematobium, 55-80% of bladder tumors are of squamous type, and the trigone is rarely affected (Oyasu & Hopp, 1974). More than 70% of transitional cell carcinomas of the bladder are superficial at diagnosis, and the rest of cancers are muscle invasive (T2-T4). Among the superficial cancers, there are three morphological forms with distinctive etiologies and prognoses: a. 70% are papillary non-infiltrating tumors (Ta); b. 10% are carcinoma in situ (Tis or CIS); c. and 20% are T1 lesions (Yu & Ross, 2002; Cote & Datar, 2003; Kirkali et al., 2005; Knowles, 2006). Tumors at the Ta stage, which may arise via simple hyperplasia and minimal dysplasia, are usually well- differentiated (low-grade) and recur over a prolonged clinical course; however, most of them do not invade the underlying tissues. CIS is characterized by flat, poorly differentiated lesions that arise from dysplasia. CIS usually progresses rapidly and is considered a precursor of invasive cancers (T1-T4) (Cordon-Cardo et al., 2000). It has been proposed that two distinct pathways of molecular alterations may be involved in the carcinogenesis of Ta versus CIS/invasive bladder cancers (Spruck et al., 1994; Knowles, 2006). 1.1.3 Risk Factors Cigarette smoking and occupational exposure to arylamines are the only two well-established risk factors for bladder cancer. Many other possible etiologic factors such as artificial sweeteners or caffeine-containing beverages have been extensively explored, but none of them have been definitively established as causative agents. 4 Occupational Exposures Occupation is the first known risk factor of bladder cancer. Exposure to 2-naphthylamine, 4- aminobiphenyl (ABP), and benzidine among workers in the textile dye and rubber tire industries have been found to be strongly associated with bladder cancer (Yu & Ross, 2002). Most of these suspected carcinogens have been banned or placed under strict regulations since the late 1950s. An excessive risk of bladder cancer has also been observed among truck/bus/taxi drivers, leather workers, painters, aluminum smelters, and hair dressers. Reasons for the increased risk of bladder cancer among these workers are unknown, and it has been proposed that exposure to incomplete combustion from kerosene heater emissions and diesel engine exhaust, hair dye and fumes from heated cooking oils may result in the formation of 4-ABP hemoglobin (Hb) adducts, established human bladder carcinogens (Skipper et al., 2003). Cigarette Smoking Cigarette smoking is the most important risk factor for bladder cancer in the US, accounting for about 50% of the total cases diagnosed in this country (Shi et al., 2002). It is the most prominent known source of exposure to aromatic amines, including-naphthylamine (IARC Group 1 carcinogen) (IARC, 1987b), and 3- , and 4-ABP (IARC Group 1 carcinogen) (IARC, 1987a). Both 3- and 4-ABP are recognized human bladder carcinogens, capable of forming DNA adducts and inducing mutations in DNA (Ward et al., 1996; Yu & Ross, 1998, 2002). The risk of bladder cancer associated with smoking varies by the population studied. In Los Angeles County, non-Hispanic white men who have comparable smoking habits as African-Americans are 2-2.5 times more likely to have bladder cancer (Yu & Ross, 5 2002). In our study, the risk of bladder cancer in women who smoked was found to be statistically significantly higher than that in men who smoked a comparable amount of cigarettes (Castelao et al., 2001). In some studies, carriers of NAT2 slow- acetylator phenotype have been found to be more susceptible to smoking-induced bladder cancer (Garcia-Closas et al., 2005). Differences in metabolism of smoking- related carcinogens, other exogenous agents, such as dietary factors, and other unknown risk factors may have contributed to the differences in smoking-related bladder cancer risk. 1.2 UROGENOUS-CONTACT HYPOTHESIS The urogenous-contact hypothesis, which associates the development of bladder cancer with prolonged exposure to carcinogens in urine, was first proposed by Oyasu, Hupp and Melicow in 1974 (Oyasu & Hopp, 1974). It was supported by the following observations that: (a) Most bladder cancers develop near the trigone of the bladder, where urogenous carcinogens enter the bladder lumen with the urine, and (b) In experimental studies, no tumors were induced by beta-naphthlamine, a bladder carcinogen, in the part of beagle’s bladder which had been isolated from urine, while tumors were induced in the part of the bladder which had not been isolated from urine. Urine is an aqueous solution of waste excreted by the kidneys, transported through the ureters into the bladder for storage, until it is eventually expelled from the body by micturition, i.e. urination. It is a mixture of chemically diverse 6 endogenous and exogenous substances collected from blood or interstitial fluid, including urea, dissolved salts, and metabolic byproducts (Pradella et al., 1988). The composition of urine is adjusted by reabsorption when essential molecules, such as glucose, are needed by the body. The amount of urine is affected by the state of hydration, physical activities, environmental factors (such as weather), body size, and health status. The duration of contact between the bladder urothelium and carcinogens in the urine is most likely responsible, at least in part, for bladder carcinogenesis. In other words, the hypothesis indicates that risk of bladder cancer is affected by not only the composition of urine but also the duration of storage of urine in the bladder. Thus, the urogenous-contact associated risk of bladder cancer is highly dependent on the intake of fluids and carcinogens, metabolic reactions, fluid- electrolyte balance, renal function, urination habits, and other factors that may influence the composition and flow of urine. An analysis that systematically addresses these factors is needed to fully examine the urogenous-contact hypothesis. 7 CHAPTER 2 FLUID INTAKE AND BLADDER CANCER RISK 2.1 TOTAL FLUID INTAKE 2.1.1 Introduction Occupational exposure to arylamines is a well-established risk factor for bladder cancer, but arylamines have been under strict regulation in the US and largely removed from occupational settings for more than 50 years (Yu & Ross, 2002). Currently, cigarette smoking is the most important risk factor for bladder cancer. However, rates of bladder cancer are low in some countries where smoking rates are very high (Jones & Ross, 1999), indicating other unidentified factors may also play a role in bladder carcinogenesis. The Health Professionals Follow-up Study found that drinking more fluids, especially water, is associated with a significantly lower risk of bladder cancer (Michaud et al., 1999). Theoretically, high intake of fluid could reduce the risk of bladder cancer either by lowering the urinary concentration of carcinogens or by increasing the frequency of urination. However, high intake of fluids also could increase the risk of bladder cancer if fluids contain contaminants that are bladder carcinogens, such as chlorination byproducts found in chlorinated drinking water or arsenic in drinking water. Thus, a recent meta-analysis indicated an 8% increase of bladder cancer risk in men for an increase of 1 L/day in fluid intake (Villanueva et al., 2006). This increased risk was associated with consumption of tap water, but not with consumption of non-tap water fluids (Villanueva et al., 2006). However, the overall evidence of an association between fluid intake and bladder cancer is not entirely consistent. 8 2.1.2 Materials and Methods Study Population From 1987 to 1999, we conducted a population-based case-control study of bladder cancer in Los Angeles County using in-person interviews (Castelao et al., 2001). Eligibility criteria included histologically confirmed bladder cancers diagnosed between January 1, 1987 and April 30, 1996 among non-Asians between the ages of 25 and 64 years. Cases were identified through the Los Angeles County Cancer Surveillance Program (SEER registry) (Bernstein & Ross, 1991). In total, 2,384 cases were identified. Two hundred and ten (9%) died before we could contact them or were too ill to be interviewed. Permission to contact 99 (4%) patients was denied by their physicians. Four hundred and four patients (17%) declined to participate in the study. For each enrolled case, a standard procedure was followed to recruit a control subject from the neighborhood of residence of the case at the time of cancer diagnosis, with control matched to the case by age (within five years), sex and race (non-Hispanic white, Hispanic white, or African American/others) (Castelao et al., 2001). We attempted to identify the sex, age, and race of all inhabitants of each housing unit; “not at home” units were systematically revisited to complete the census. The first resident along this defined route who satisfied all eligibility criteria for controls was asked to participate in this study (i.e., first eligible control). If that individual refused, the next eligible control (i.e., second eligible control) in the sequence was asked and so on until we located an eligible control who agreed to be interviewed. When we failed to find any resident who met our matching criteria after canvassing 150 housing units, we excluded race 9 from the matching criteria. If a matched control subject based on this relaxed criterion could not be found within a maximum of 300 housing units, the case was dropped from the study. We interviewed 1,671 (70%) eligible cases, and a matched control could not be found for 5% (85) of interviewed cases. Twenty-one control subjects were not matched to the index case by race. One thousand and ninety (69 %) enrolled control subjects were first eligible controls, 325 (20%) were second eligible controls, and 111 (7%) were third eligible controls. The remaining 60 (4%) control subjects were fourth or higher order eligible controls. Therefore, a total of 1,586 pairs of cases and controls were included in the analyses. All study subjects signed informed consent forms, approved by the Human Subjects Committee at the University of Southern California, Keck School of Medicine. Data Gathering and Exposure Definitions All study cases and controls were interviewed at home by trained interviewers using a structured questionnaire. The questionnaire requested information up to two years prior to the diagnosis of cancer for cases and two years prior to the diagnosis of cancer of the index case for matched controls (i.e., reference year). Each subject was asked to report information on demographic characteristics, height, weight, lifetime use of tobacco products and alcohol, usual adult dietary habits, lifetime occupational history, prior medical conditions, and prior use of medications. All subjects were asked about consumption of water, coffee, tea, alcohol, milk, and juice. Questions on consumption of hot chocolate and soda (such as Coca Cola, 7-up, Root Beer, etc.), and specific-alcoholic beverages (beer, wine, and hard liquor) were not included in the original 10 questionnaire, but were added to interviews conducted after January 1992. Regular drinkers were defined as those who drank the specified beverage (except water) at least once a week for six months or longer. For regular drinkers, information was requested on the age at starting to drink regularly and frequency and duration of drinking. Consumption of juice was estimated based on individual questions on frequency of intakes of orange juice, grapefruit juice, and fortified fruit drinks / orange juice substitutes. Intake of milk was estimated as the summarized intake of whole milk, 2% milk, and skim milk. For juice, milk, and water, one glass (8oz) was considered as one serving. For coffee, tea, alcohol, hot chocolate, and soda, each subject was asked if they ever drank a specific beverage (one 8 oz-cup) on a regular basis. Subjects were also asked the specific type of coffee (regular or decaffeinated), tea (black, green, herbal), soda (regular or diet, with caffeine or caffeine-free) they regularly drank. Statistical Analysis The association of bladder cancer with fluid intake was estimated by odds ratios (ORs) and corresponding 95% confidence intervals (CIs) using conditional logistic regression. Total fluid intake was calculated as the sum of water, coffee, tea, juice and milk intake. For subjects interviewed after 1992, intake of hot chocolate, soda, beer, wine, and hard liquor were also added to total fluid intake. When the case or the control subject of a pair failed to answer the relevant questions, we eliminated that case-control pair from the corresponding conditional analysis. We first fitted models with no covariates, thus the analyses were adjusted only for matching factors (age, sex, and race). We then repeated the analyses with 11 further adjustment for the following identified risk or protective factors for bladder cancer: average number of cigarettes smoked per day, number of years of smoking, smoking status in reference year (smoker or nonsmoker) (Castelao et al., 2001), level of education (high school or below, 1-3 years of college, college graduate), lifetime use of non-steroidal anti-inflammatory drugs (NSAIDs) (non/irregular user, <1,441 pills,1,441 pills over lifetime) (Castelao et al., 2000), intake of carotenoids (quintiles among controls) (Castelao et al., 2004), and duration of employment as a hairdresser or barber (years) (Castelao et al., 2001). Results of analyses were similar with and without adjustment for level of education. There were no material changes in the results with further adjustment for hair dye use, and thus this variable was not retained. ORs with 2-sided p values less than 0.05 were considered statistically significant. All p-values reported are 2-sided. 2.1.3 Results A total of 1,586 case-control pairs were included in the current study. Higher level of total fluid intake was associated with higher intake level of each type of fluids: water, alcoholic beverages, coffee, tea, milk, juice, soda, and hot chocolate. Among these fluids, intake of water, alcoholic beverages, and coffee contributed the most to the total volume of reported fluid intake. Controls with higher level of water intake were more likely to be female and nonsmokers, and to have higher intake of NSAIDs. 12 Total fluid intake was not associated with bladder cancer in our study (multivariate OR for an increase of 240 ml in total daily fluid intake, 1.00; 95% CI, 0.98-1.01; Table 1). The OR was 0.98 (95% CI, 0.77-1.26) for the highest quartile of total daily fluid intake (>2,983 ml per day) as compared with the lowest quartile (<1,526 ml per day) (Table 1). There was no material difference in these associations between subjects who were interviewed before 1992 using the original questionnaire and those who were interviewed after 1992 using the revised questionnaire. Daily water intake seemed to be associated with a slightly decreased risk of bladder cancer (P for trend=0.21), and this association remained materially unchanged after additional adjustment for consumption of other beverages, such as alcohol, coffee, and soda. Alcoholic beverages were found to be associated with reduced risk of bladder cancer in a dose- and duration-dependent manner. Subjects who drank soda were 1.30 (95% CI, 1.04-1.62) times as likely to develop bladder cancer as those who did not drink soda. However, there was no clear trend associated with increased intake of this beverage. Drinking coffee was associated with a slightly increased risk of bladder cancer (P trend=0.052), especially among heavy drinkers who drank at least seven cups of coffee per day (OR, 1.38; 95% CI, 0.95-2.00). I will next present the major findings associated with the intake of the three most common types of beverages, i.e., water, alcohol, and coffee, and bladder cancer risk. 13 Table 1. Fluid intake and risk of bladder cancer the Los Angeles Bladder Cancer Case-Control Study Fluid intake Frequency of intake P trend Total fluid(quartile) 1st 2nd 3rd 4th Case/Control 371/390 355/391 373/389 439/391 OR 1 (95% CI) 1.00 0.89(0.70-1.13) 0.94(0.75-1.19) 0.98(0.77-1.26) 0.96 Water(glasses/day) <1 1 2 3 4-5 6 Case/Control 293/214 249/230 335/332 190/230 257/330 220/229 OR 1 (95% CI) 1.00 0.89(0.67-1.18) 0.80(0.62-1.04) 0.77(0.57-1.03) 0.74(0.56-0.98) 0.90(0.67-1.21) 0.21 Alcohol(drinks/day) 0 <1 1-4 4 Case/Control 432/453 364/385 484/481 293/258 OR 1 (95% CI) 1.00 0.85(0.68-1.06) 0.77(0.61-0.96) 0.69(0.53-0.90) 0.003 Coffee(cups/day) 0 <1 1-2 3-4 5-6 7 Case/Control 129/190 49/64 501/588 467/414 226/193 210/137 OR 1 (95% CI) 1.00 1.15(0.71-1.85) 1.04(0.78-1.38) 1.21(0.89-1.64) 1.19(0.85-1.68) 1.38(0.95-2.00) 0.052 Tea(cups/day) 0 <1 1-2 3-4 5 Case/Control 1103/1022 154/191 210/257 74/73 43/38 OR 1 (95% CI) 1.00 0.96(0.74-1.26) 0.95(0.76-1.20) 1.16(0.80-1.69) 0.88(0.54-1.45) 0.89 Milk(glasses/day) 0 <1 1 2 Case/Control 189/162 628/612 497/532 272/280 OR 1 (95% CI) 1.00 1.06(0.81-1.38) 0.97(0.74-1.28) 0.97(0.72-1.32) 0.53 Juice(glasses/day) 0 <1 1 2 Case/Control 80/72 1154/1115 319/350 33/49 OR 1 (95% CI) 1.00 1.15(0.79-1.67) 1.12(0.75-1.67) 0.87(0.47-1.60) 0.66 14 Table 1 continued. Fluid intake Frequency of intake P trend Soda (cups/day) 0 <1 1-2 3-4 5 Case/Control 283/324 277/265 256/248 80/89 71/44 OR 1 (95% CI) 1.00 1.42(1.08-1.86) 1.21(0.92-1.59) 1.15(0.77-1.70) 1.39(0.87-2.21) 0.19 Hot chocolate (cups/day) 0 <1 1 Case/Control 869/854 55/67 47/48 OR 1 (95% CI) 1.00 0.93(0.60-1.43) 1.06(0.67-1.67) 0.80 1: Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 15 2.2 WATER INTAKE 2.2.1 Introduction The Health Professionals Follow-up Study found that drinking more water, which presumably increases micturition rate, is associated with a significantly lower risk of bladder cancer (Michaud et al., 1999). On the other hand, a recent meta- analysis of six case-control studies indicated an increased risk in men with consumption of tap water, but not with other non-tap water fluids (Villanueva et al., 2006). Based on this finding, the authors suggested that the increased risk for tap water may be related to the presence of carcinogens such as disinfection by-products in tap water. The overall evidence of an association between water intake and bladder cancer is not entirely consistent. 2.2.2 Materials and Methods Study Population Refer to Chapter 2.1.2. Data Gathering and Exposure Definitions Refer to Chapter 2.1.2. After January 1992, frequency of daytime urination and nighttime urination during adult life before reference year was requested. Statistical Analysis Refer to Chapter 2.12. We also examined the possibility that the water intake-bladder cancer association was modified by sex, cigarette smoking, and frequency of urination. Unconditional logistic regression was used in analyses of subsets stratified by smoking status (lifelong nonsmokers, smokers of <30 years, smokers of30 years). We broke the case–control matching to analyze 16 data on all informative cases and control subjects. In addition to all the covariates mentioned above, 10 age-sex strata (age groups of <46, 46–50, 51–55, 56–60 and >60 years for each sex), and race (non-Hispanic white, Hispanic white, or African American/other) were included in the regression models. To test a possible difference in water intake-associated risk of bladder cancer by smoking status (i.e. an interaction effect), a product term of frequency of water intake and duration of cigarette smoking (0, <10, 10 to <20, 20 to<30, 30 to <40,40 years) was used in a conditional logistic regression. For subset analyses with small sample size, covariates were excluded from models in the following order: number of years employed as hair dresser, race. ORs with 2-sided p values less than 0.05 were considered statistically significant. All p-values reported are 2-sided. 2.2.3 Results Daily water intake seemed to be associated with a slightly decreased risk of bladder cancer (P for trend=0.21; Table 2), and this association remained materially unchanged after additional adjustment for consumption of other beverages, such as alcohol, coffee, and soda. This water intake-bladder cancer association seemed to be modified by sex (P for effect modification by sex =0.072) with protection being more pronounced in women (P trend=0.036) than men (P trend=0.59). Among women, drinking six or more glasses of water per day was associated with 0.69 times (95% CI, 0.36-1.34) the risk of bladder cancer compared to drinking less than one 17 glass of water per day. Since the magnitude of the protective effect of water intake on bladder cancer risk was comparable between the highest five categories, subjects in 1 to6 categories of water intakes were grouped together. Compared to drinking less than one glass of water per day, drinking at least one glass of water per day was associated with 0.82 (95% CI, 0.65-1.02) times the risk of bladder cancer among total subjects, 0.88 (95% CI, 0.68-1.12) times the risk among men, and 0.61 (95% CI, 0.36-1.05) times the risk among women. As the most important established non-occupational source of bladder carcinogens, cigarette smoking was examined as a potential modifier of the water intake-bladder cancer association (Table 3). In men, the effect of water intake did not differ significantly by smoking status, whereas in women, the reduction in bladder cancer risk associated with consumption of water was seen mostly among lifetime nonsmokers and shorter-term smokers (smokers of <30 years), but not among longer-term smokers (smokers of30 years). Compared to drinking <1 glass of water per day, OR associated with drinking1 glasses of water per day was 0.26 (95% CI, 0.095-0.70) among female nonsmokers, 0.45 (95% CI, 0.20-1.04) among female shorter-term smokers, and 1.12 (95% CI, 0.48-2.60) among female longer- term smokers. However, the test for effect modification did not reach statistical significance (p for interaction =0.20). The association between water intake and bladder cancer seemed to be modified by frequency of urination (P for interaction=0.13; Table 4). A significant inverse association was observed among subjects who urinated at least six times per 18 day (P trend=0.015), but not among those who urinated less frequently. Among those subjects who urinated at least six times per day, drinking1 glasses of water per day was associated with a 0.36 (95% CI, 0.19-0.66) times the risk of bladder cancer compared to drinking <1 glass per day. Similar protection from water intake was obsevered in both men and women who urinated at least six times per day. Nocturia (van Kerrebroeck et al., 2002), defined as waking at night to void on one or more occasions, has been associated with clinically significant bladder lesions (cystitis glandularis or malignancy) (Mommsen et al., 1982, 1983; Wu et al., 2006). In the present study, nocturia was a risk factor for bladder cancer among women (OR, 1.67; 95% CI, 0.99-2.83). We therefore examined the water intake-bladder cancer association by nocturia, i.e. we separated subjects who usually urinated at least once during the night from those who did not urinate during the night (Table 5). The protection from water intake among those who urinated6 times per day was confined to women who did not report night-time urination (OR = 0.003, 95% CI <0.001-0.38) and to men who did (OR = 0.15, 95% CI 0.046-0.47). We further examined whether the water intake-bladder cancer association was modified by additional factors, such as level of education, use of hair dye, use of NSAIDs, consumption of vitamin C and carotenoids, and body mass index (BMI). Of these variables, only BMI was found to modify the water intake-bladder cancer relationship. Subjects were categorized asmedian BMI (25.13 for men and22.52 for women) or >median BMI (>25.13 for men and >22.52 for women). A significant reduction in risk of bladder cancer was seen mostly among women withmedian 19 Table 2. Water intake and bladder cancer risk by sex Total Men Women Water intake (glass/day) Case/Control OR 1 (95% CI) Case/Control OR 1 (95% CI) Case/Control OR 1 (95% CI) Total 1 <1 293/214 1.00 223/176 1.00 70/38 1.00 1 249/230 0.89 (0.67-1.18) 194/175 0.98 (0.71-1.35) 55/55 0.68 (0.35-1.34) 2 335/332 0.80 (0.62-1.04) 255/262 0.80 (0.59-1.07) 80/70 0.76 (0.40-1.42) 3 190/230 0.77 (0.57-1.03) 151/182 0.83 (0.59-1.16) 39/48 0.58 (0.28-1.18) 4-5 257/330 0.74 (0.56-0.98) 212/251 0.90 (0.66-1.22) 45/79 0.31 (0.16-0.62) 6 220/229 0.90 (0.67-1.21) 167/172 0.93 (0.66-1.30) 53/57 0.69 (0.36-1.34) P trend 0.21 0.59 0.036 P for interaction with sex =0.072 1 1251/1351 0.82 (0.65-1.02) 979/1042 0.88 (0.68-1.12) 272/309 0.61 (0.36-1.05) 1: Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking; 20 Table 3. Water intake and bladder cancer risk by smoking status and sex Nonsmokers Smokers: <30 years Smokers:30 years Water (glass/day) Ca/Co OR 1 (95% CI) Ca/Co OR 2 (95% CI) Ca/Co OR 2 (95% CI) Both sexes <1 44/66 1.00 98/78 1.00 151/70 1.00 1 30/83 0.57 (0.32-1.03) 86/84 0.99 (0.63-1.55) 133/63 1.02 (0.66-1.57) 2 63/122 0.76 (0.46-1.25) 113/126 0.81 (0.53-1.22) 159/84 0.97 (0.64-1.45) 3 39/81 0.77 (0.44-1.34) 76/86 0.84 (0.54-1.32) 75/63 0.66 (0.42-1.05) 4-5 64/121 0.79 (0.48-1.32) 86/126 0.67 (0.44-1.04) 107/83 0.74 (0.48-1.13) 6 40/96 0.62 (0.36-1.09) 81/81 1.01 (0.63-1.61) 99/52 1.04 (0.65-1.66) P trend 0.43 0.35 0.33 P for interaction=0.24 1 236/503 0.71 (0.46-1.09) 442/503 0.84 (0.59-1.19) 573/345 0.88 (0.63-1.23) Men <1 32/56 1.00 74/63 1.00 117/57 1.00 1 21/55 0.75 (0.38-1.49) 70/66 1.11 (0.67-1.85) 103/54 1.05 (0.65-1.70) 2 44/92 0.84 (0.47-1.51) 87/10 0.88 (0.55-1.41) 124/70 0.95 (0.60-1.50) 3 29/58 0.91 (0.48-1.74) 66/70 1.01 (0.61-1.68) 56/54 0.60 (0.36-1.00) 4-5 52/85 1.07 (0.60-1.92) 72/97 0.81 (0.50-1.32) 88/69 0.80 (0.50-1.30) 6 28/64 0.74 (0.39-1.43) 63/67 1.05 (0.62-1.77) 76/41 1.11 (0.65-1.89) P trend 0.96 0.66 0.52 P for interaction=0.086 1 174/354 0.88 (0.54-1.43) 358/400 0.95 (0.64-1.41) 447/288 0.89 (0.61-1.30) 21 Table 3 continued. Nonsmokers Smokers: <30 years Smokers:30 years Water (glass/day) Ca/Co OR 1 (95% CI) Ca/Co OR 2 (95% CI) Ca/Co OR 2 (95% CI) Women <1 12/10 1.00 24/15 1.00 34/13 1.00 1 9/28 0.20 (0.060-0.68) 16/18 0.59 (0.20-1.71) 30/9 1.09 (0.36-3.33) 2 19/30 0.39 (0.13-1.21) 26/26 0.56 (0.22-1.45) 35/14 1.37 (0.49-3.85) 3 10/23 0.28 (0.084-0.96) 10/16 0.24 (0.072-0.81) 19/9 1.03 (0.33-3.20) 4-5 12/36 0.19 (0.057-0.63) 14/29 0.24 (0.081-0.71) 19/14 0.94 (0.31-2.87) 6 12/32 0.24 (0.072-0.78) 18/14 0.68 (0.22-2.15) 23/11 1.08 (0.35-3.33) P trend 0.062 0.097 0.93 P for interaction=0.66 1 62/149 0.26 (0.095-0.70) 84/103 0.45 (0.20-1.04) 126/57 1.12 (0.48-2.60) 1: Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser; 2: Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 22 Table 4. Water intake and bladder cancer risk by frequency of urination Urination: <4 times/day Urination: 4-5 times/day Urination:6 times/day Water intake (glass/day) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Both sex <1 64/57 1.00 64/56 1.00 56/21 1.00 1 46/57 0.85 (0.47-1.52) 57/54 0.88 (0.51-1.54) 30/22 0.44 (0.19-1.02) 2 66/64 1.10 (0.64-1.92) 77/88 0.83 (0.50-1.39) 38/43 0.36 (0.17-0.78) 3 36/37 1.03 (0.54-1.97) 45/73 0.65 (0.37-1.14) 27/33 0.31 (0.14-0.71) 4-5 37/68 0.62 (0.34-1.15) 72/80 1.02 (0.61-1.73) 42/59 0.29 (0.14-0.60) 6 30/21 1.66 (0.80-3.46) 53/63 0.98 (0.55-1.73) 55/47 0.41 (0.20-0.85) P trend 0.89 0.91 0.015 P for interaction=0.13 1 215/247 0.95 (0.60-1.51) 304/358 0.87 (0.57-1.32) 192/204 0.36 (0.19-0.66) Men <1 49/48 1.00 53/46 1.00 39/17 1.00 1 39/46 0.91 (0.47-1.74) 47/36 1.15 (0.61-2.19) 23/16 0.68 (0.26-1.76) 2 52/54 1.10 (0.59-2.04) 53/62 0.81 (0.45-1.47) 27/32 0.41 (0.17-0.98) 3 32/29 1.22 (0.60-2.48) 37/58 0.72 (0.38-1.34) 11/25 0.23 (0.085-0.64) 4-5 34/58 0.73 (0.37-1.44) 61/62 1.22 (0.68-2.20) 35/37 0.47 (0.21-1.08) 6 21/18 1.41 (0.61-3.24) 41/52 0.93 (0.49-1.77) 39/31 0.47 (0.20-1.10) P trend 0.90 0.96 0.068 P for interaction=0.15 1 178/205 1.01 (0.61-1.69) 239/270 0.96 (0.59-1.55) 135/141 0.44 (0.22-0.88) 23 Table 4 continued. Urination: <4 times/day Urination: 4-5 times/day Urination:6 times/day Water intake (glass/day) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Women 2 <1 15/9 1.00 11/10 1.00 17/4 1.00 1 7/11 0.37 (0.069-1.96) 10/18 0.60 (0.16-2.24) 7/6 0.12 (0.015-0.97) 2 14/10 1.07 (0.23-4.95) 24/26 0.94 (0.28-3.15) 11/11 0.23 (0.038-1.44) 3 4/8 0.19 (0.030-1.26) 8/15 0.50 (0.12-2.13) 16/8 0.33 (0.049-2.19) 4-5 3/10 0.15 (0.021-1.08) 11/18 0.68 (0.17-2.74) 7/22 0.061 (0.010-0.39) 6 9/3 1.54 (0.23-10.55) 12/11 1.09 (0.26-4.66) 16/16 0.20 (0.033-1.14) P trend 0.63 0.95 0.074 P for interaction=0.15 1 37/42 0.53 (0.16-1.83) 65/88 0.75 (0.25-2.27) 57/63 0.16 (0.033-0.77) 1: Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 2: number of years as hairdresser was excluded. 24 Table 5. Water intake and bladder cancer risk by nocturia among subjects with high (6 times/day) frequency of daytime urination Without nocturia With nocturia Water intake (glass/day) Case/Control OR 1 (95% CI) Case/Control OR 1 (95% CI) Men <1 17/10 1.00 22/6 1.00 1 79/71 0.72 (0.27-1.92) 56/68 0.15 (0.046-0.47) P for interaction=0.097 Women <1 14/1 1.00 2/3 1.00 1 23/28 0.003 (<0.001-0.38) 34/35 10.1 (0.83-124.2) P for interaction=0.014 1: Unconditional logistic regression, adjusted for age, level of education, use of NSAIDs, intake of carotenoids, cigarette smoking status, duration of smoking, and intensity of smoking. Race and number of years as hair dresser were excluded considering the small sample size in the analysis and estimates were materially unchanged. 25 Table 6. Water intake and bladder cancer risk by BMI 1: Unconditional logistic regression, adjusted for age, race, level of education, use of NSAIDs, carotenoids intake, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. BMI median 2 BMI > median 2 Water intake (glass/day) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Men <1 121/92 1.00 102/84 1.00 1+ 509/539 0.84 (0.61-1.17) 469/503 0.93 (0.66-1.31) Women <1 45/17 1.00 25/21 1.00 1+ 143/141 0.42 (0.22-0.82) 128/167 0.85 (0.41-1.74) 26 BMI (OR associated with one glass of water per day, 0.42; 95% CI, 0.22-0.82), but not among women with >median BMI or among men (Table 6). 2.2.4 Discussion In this case-control study, a high intake of water was associated with a reduced risk of bladder cancer among female nonsmokers/shorter-term smokers, and among those who urinated more frequently. The present study found that drinking more water, which presumably increases the rate of micturition, is associated with a significantly decreased rate of bladder cancer. The results are consistent with the urogenous-contact hypothesis (Braver et al., 1987), first proposed by Oyasu, Hupp and Melicow in 1974, which attributes the development of bladder cancer to prolonged exposure to carcinogens in urine (Oyasu & Hopp, 1974). A high intake of water can dilute the urine, increase the frequency of urination, and thereby reduce the contact of carcinogens with the urothelium. Frequency of urination is directly related to the intensity and duration of urothelium distension. It has been proposed that measuring total fluid intake while accounting for the daily number of urinary voiding events is more reliable than measuring fluid intake alone (Radosavljevic, 2004). While most studies have been simply focused on fluid intake, frequency of urination was rarely considered epidemiologically because of the difficulties in assessing this behavior retrospectively as well as prospectively. Only a few studies have investigated the role of frequency of urination in the etiology of bladder cancer (Braver et al., 1987; 27 Radosavljevic et al., 2003). In one study, healthy controls were found to have more frequent urination compared to cancer patients (Radosavljevic et al., 2003). Another study found greater urine concentration and less frequent micturition in high-risk groups for bladder cancer (Braver et al., 1987). Experimental data from dogs given 4-ABP indicated that exposure to 4-ABP and subsequent ABP-DNA adduct formation in the bladder are directly dependent on voiding frequency, and more adducts were formed in dogs with less frequent urination (Kadlubar et al., 1991). It has been shown that frequency of urination is an important factor contributing to interindividual differences in DNA-binding of ABP in human bladder (Bois et al., 1995). In our study, the association between water intake and bladder cancer risk was modified by frequency of urination with the reduction in risk confined to those who urinated more frequently. This observation is predicted by the urogenous-contact hypothesis. Our study found that the water intake-bladder cancer association was more pronounced among women than men. In experimental studies, female rats consistently consumed significantly more water (61.0 ml/kg) than the age-matched male rats (33.2 ml/kg) (Yoshimura et al., 2000), when normalized by weight. In our study, healthy women drank slightly more water (average 3.24 glasses/day) than healthy men (average 3.12 glasses/day), despite their smaller body size (average weight: women: 64.2 kg vs. men: 82.4kg). It has been found that the diuretic response to a modest water load is greater in women than men (Claybaugh et al., 2000), which is consistent with our observation that women urinated more frequently 28 (5.0 times/day) than men (4.5 times/day) (P t-test =0.0075). In line with these sex differences in both water intake and micturition, our study found that water intake was associated with a reduced risk of bladder cancer in women, with a less consistent association in men. A similar pattern was observed in some other studies (Wilkens et al., 1996; Pohlabeln et al., 1999). The influence of hormonal factors on micturition may underlie the difference in the associations between fluid intake and urination in men versus women. Estradiol administration has been shown to attenuate the antidiuretic action of vasopressin in the ovariectomized rat (Longhurst et al., 1992).s In addition, in the present study, the water intake-bladder cancer inverse association was confined to women who did not experience nocturia and to men who did. One previous study found that men with nocturia had a significantly higher urine output both during the day (difference, 131ml) and the night (difference, 384ml), and consequently over 24-h (difference, 462ml), than men without nocturia (Rembratt et al., 2003), consistent with another previous study (Matthiesen et al., 1999). However, women with nocturia appeared to have had the same level of 24-h urine as non- nocturics, but differently distributed over day- and night-time (a significantly lower diurnal urine volume (difference, -147ml) and higher nocturnal urine volume (difference 227ml)) (Rembratt et al., 2003). This sex difference in voiding patterns indicates a possible difference in pathophysiology of nocturia in men and women, or at least a different distribution of bladder problems underlying this condition. In women, nocturia has been associated with significant bladder lesions (cystitits glandularis or malignancy)(Wu et al., 2006) and has been found to result from 29 irritative bladder symptoms caused by bladder cancer (Lundgren, 2004; Wu et al., 2006). In men, however, nocturia has long been considered a common symptom of benign prostatic hyperplasia, a condition which affects more than 50% of men aged over 60 years and 90% of men aged over 80 years (Thorpe & Neal, 2003). In our study, nocturia was found to be associated with an increased risk of bladder cancer in women (OR, 1.67; 95% CI, 0.99-2.83), but not in men (OR, 0.95; 95% CI, 0.74- 1.22). Thus, nocturia may serve as a surrogate for different factors among men versus women. Increasing BMI has been associated with increasing rate of lower urinary tract symptoms including urinary incontinence(Holroyd-Leduc & Straus, 2004) and overactive bladder (Tubaro & Palleschi, 2005), mostly among women. An experimental study found that diuresis was markedly attenuated in obese hypertensive dogs (West et al., 1992). In our study, BMI was also found to modify the water intake-bladder cancer association, especially in women. The reduction in bladder cancer risk was most pronounced among women with lower BMI who also had the highest frequency of urination (5.1 times/day), compared to women with higher BMI (4.8 times/day) when a similar amount of water was consumed. These findings give additional support to the urogenous-contact hypothesis by showing the highest protection among those with the least contact-time with urine. Smoking was also found to modify the water intake-bladder cancer association, especially in women. Among women, the reduction in risk was seen mostly among non-smokers and shorter-term smokers (smokers of < 30 years), but 30 not among long-term smokers (smokers of30 years). One could hypothesize that smokers would have benefited more from increased water intake than nonsmokers, if the mechanism of protection involves reducing exposure to cigarette-derived carcinogens in urine. Accordingly, we would expect the much lower level of carcinogens to preclude our observing the beneficial effects of water intake among lifetime nonsmokers. On the other hand, the protective effect of water intake may be overwhelmed in longer-term smokers, in which the levels of carcinogens may be too high to be eliminated by the carcinogen-dilluting properties of water. In addition, cigarette smoke may not be the only source of carcinogens in the bladder. Use of hair dye is another substantial source of arylamine exposure in humans, especially in women. A dose- and duration-dependent increase in bladder cancer risk among women who use permanent hair dyes has been recently reported from our study, and the increase in risk was mainly confined to women who are deficient in arylamine detoxification (Gago-Dominguez et al., 2001; Gago-Dominguez et al., 2003). Other sources including passive smoking of tobacco, kerosene heater emissions, diesel engine exhaust, and fumes from heated cooking oils have also been proposed (Skipper et al., 2003; Gan et al., 2004). Our study has several limitations. First, we measured only frequency of urination but not volume of urination, which would be a more desirable marker. Women who drank more water and smoked less tobacco may have been more health conscious and therefore adopted healthier life styles that may have contributed to the lower risk of bladder cancer. For example, health-conscious women may have 31 consumed more water and had higher physical activity, which has been associated with a decrease in bladder cancer risk in some studies (Wannamethee et al., 2001; Tripathi et al., 2002b). However, we did not collect information on physical activity in our study, and therefore we were unable to control for this potential confounder. Polydipsia and poyluria are common symptoms among diabetic patients(Bankir et al., 2001) and diabetes has been associated with increased risk of bladder cancer (Risch et al., 1988; Kravchick et al., 2001; Tripathi et al., 2002a; Ng et al., 2003; Coughlin et al., 2004; Larsson et al., 2006). Exclusion of diabetic patients from our analysis did not materially change our results. The source of drinking water may also be important. Villanueva et al.(Villanueva et al., 2006) pooled six case-control studies and found an increased risk of bladder cancer associated with intake of tap water, but not with non-tap water fluids, which indicates that carcinogenic chemicals in tap water may explain the increased risk. Associations have also been found between bladder cancer and high levels of arsenic (Zierler et al., 1988; Wu et al., 1989) and nitrates (Morales Suarez-Varela et al., 1993) in drinking water. Some investigators propose that the quality of drinking water, with or without carcinogens, rather than the quantity of water should be measured to assess the risk of bladder cancer (Allam, 2005). Our study did not collect information on the source of drinking water, so we were not able to compare effects of tap and non-tap water. The sex difference we observed could also be due to differences in unidentified environmental and dietary exposures or differences in metabolic genes and hormonal factors. As in any other epidemiological study, our findings may have been due to chance, although we 32 believe this is not likely given the consistent observation of the negative association with water intake across many subgroups of the population. In conclusion, our data showed that water intake was associated with a slightly reduced risk of bladder cancer. This inverse association was modified by sex, cigarette smoking, BMI, day-time frequency of urination, and presence or absence of nocturia. 33 2.3 ALCOHOLIC BEVERAGES 2.3.1 Introduction The role of alcoholic beverages in bladder carcinogenesis has been investigated in several epidemiologic studies (1988). No significant association was found in most studies including the recent publication from the Framingham Heart Study (Djousse et al., 2004). However, a meta-analysis of urinary tract cancers suggested a slightly increased risk of 1.3 for current alcohol-drinkers, when compared with nondrinkers (Zeegers et al., 1999). Results of beverage-specific analyses are also inconclusive. Some studies indicated higher risks from spirit/liquor consumption (Risch et al., 1988; Chyou et al., 1993; Zeegers et al., 2001b; Djousse et al., 2004), and others found an inverse association with beer consumption (Pelucchi et al., 2002; Djousse et al., 2004). Despite the large number of published studies, questions remain about the definite role of alcohol on bladder carcinogenesis, the effect of the different alcoholic beverages, drinking pattern, and modification by factors such as cigarette smoking and frequency of urination. We collected detailed information on alcoholic beverage intake in a large population-based case-control study of bladder cancer in Los Angeles County, where we have identified risk/protective factors, such as cigarette smoking (Castelao et al., 2001), use of nonsteroidal anti-inflammatory drugs (NSAIDs) (Castelao et al., 2000), consumption of carotenoids and vitamin C (Castelao et al., 2004) and use of permanent hair dyes among women (Gago-Dominguez et al., 2001; Gago-Dominguez et al., 2003). Results from this study will help clarify the role of alcohol intake in bladder cancer. 34 2.3.2 Materials and Methods Study Population Refer to Chapter 2.1.2. Data Gathering and Exposure Definitions Refer to Chapter 2.1.2. One alcoholic drink is defined as one 12-ounce can of beer (360 ml), one 4-ounce glass of wine (120 ml) or one shot of hard liquor (45 ml) whose alcoholic contents are similar (about 13 g of ethanol per drink) (Agriculture, 1992, 1994; Yuan et al., 1998b). A total of 1,133 bladder cancer cases and 1,148 controls were asked about their frequency of beverage-specific drinking. Statistical Analysis Refer to Chapter 2.1.2. We also examined the possibility that the alcohol-bladder cancer association could be modified by the status of cigarette-smoking, sex and frequency of urination during the day. Unconditional logistic regression was used in analyses of subsets defined by smoking status (lifetime nonsmokers, smokers of <30 years, smokers of 30 years) and frequency of urination during the day (<4 times, 4 times/day). In addition to all the covariates mentioned above, 10 age-sex strata (age groups of <46, 46-50, 51-55, 56-60 and >60 years for each sex), and race (non-Hispanic white, Hispanic white or African American/others) were included in the regression models to account for the matched sampling. Modification by sex was addressed by conditional logistic regression, since the cases and controls were designed to match on sex. For analyses of beverage-specific consumptions of alcohol, all three types of alcoholic beverage were included in the conditional analyses, which were also adjusted for other covariates mentioned above. 35 ORs with 2-sided p values less than 0.05 were considered statistically significant. All p values presented are 2-sided. 2.3.3 Results Table 2.7 summarizes the baseline characteristics of the study subjects. Regardless of their disease status, subjects who drank more alcohol, especially those who drank more than four drinks per day, were more likely to be male, less educated, current heavy smokers, regular users of NSAIDs and drinkers of other fluids. Among controls, intake of alcoholic beverages was associated with intake of coffee, soda and hot chocolate, inversely associated with intake of juice and not associated with intake of water, tea and milk. However, alcohol drinkers and nondrinkers were similar in age (mean age 55-57 years), race distributions (90-94% non-Hispanic white), intake of carotenoids and occupation. There was strong evidence of an inverse association between consumption of alcoholic beverages and bladder cancer risk after adjustment for smoking (Table 8). Additional adjustment for high-risk occupation, level of education, use of NSAIDs and intake of carotenoids did not change the association. The risk of bladder cancer decreased significantly with the frequency (p for trend = 0.003) and duration (p for trend = 0.017) of alcohol consumption. Subjects who drank more than four drinks of alcoholic beverages per day had a 32% lower risk of bladder cancer compared to subjects who never drank any alcoholic beverage (OR, 0.68; 95% CI, 0.52-0.90). This association persisted after we restricted the analyses to subjects interviewed 36 with the new questionnaire only. Further adjustment for total fluid intake (excluding alcoholic beverages) did not materially change our results. We also examined the alcohol-bladder cancer association by volume (data are only available for subjects interviewed with new questionnaire). An increase of 240 ml in alcoholic beverage intake was associated with 4% reduction in bladder cancer risk (OR, 0.96; 95% CI, 0.93-0.99). To test a possible difference in alcohol-related risk of bladder cancer by smoking status (i.e. an interaction effect), a product term of frequency of alcohol consumption (0, <1, 1-4, >4 drinks/day) and duration of cigarette smoking (0, <10, 10 to <20, 20 to <30, 30 to <40, 40 years) was used in a conditional logistic regression. We found the association to be significantly different by smoking status (p for interaction = 0.030; df = 15). No significant associations between consumption of alcoholic beverages and bladder cancer risk were found among nonsmokers (Table 9). Most of the reduction in bladder cancer risk from alcohol consumption was seen among smokers, and this was primarily confined to shorter-term smokers, i.e. smokers of less than 30 years. The risk of bladder cancer was 0.46 (95% CI, 0.30-0.70) times lower among shorter-term smokers who drank at least four drinks of alcoholic beverages per day compared to shorter-term smokers who did not drink any alcoholic beverages. There was no significant association between consumption of alcoholic beverages and bladder cancer risk among subjects who had been smoking for more than 30 years (p for trend = 0.46). Consumption of alcoholic beverages was also a modifier of the smoking-bladder cancer association. Alcohol 37 drinkers had lower risk from regular cigarette smoking (OR, 1.95; 95% CI, 1.57- 2.43), compared to nondrinkers (OR, 4.34; 95% CI, 3.19-5.90). These associations were similar when the analyses were restricted to men only. The association between consumption of alcoholic beverages and bladder cancer was not significant among women (p for trend = 0.27, data not shown), and it was not modified by smoking. However, consumption of alcoholic beverages among women was not as common as among men. There were only 35 women who reported drinking more than four drinks of alcohol per day. Because alcohol is a strong diuretic, we next examined the alcohol-bladder cancer association by frequency of urination, which was requested only in the new questionnaire used in interviews conducted after January 1992. The effect of alcohol consumption was marginally modified by frequency of urination during daytime (p for interaction = 0.057; Table 10). Among shorter-term smokers who urinated less frequently (<4 times per day, median urination frequency among controls), no association was found between consumption of alcoholic beverages and bladder cancer risk (p for trend = 0.91). However, among shorter-term smokers who urinated more frequently ( 4 times per day), consumption of alcoholic beverages was inversely associated with bladder cancer risk (p for trend <0.0001); drinking at least four drinks per day significantly reduced the risk of bladder cancer by 76% (OR, 0.24; 95% CI, 0.12-0.47). Beverage-specific data was requested only by the new questionnaire used in interviews conducted after January 1992. Each beverage had a different impact on 38 the risk of bladder cancer (Table 11). There was a significantly decreased risk associated with beer drinking (p for trend = 0.002), after adjustment for other types of alcoholic beverages, i.e. wine and hard liquor. Subjects who drank more than four cans of beer per day had a 0.54 (OR, 0.54; 95% CI, 0.35-0.83) times the risk of bladder cancer compared to subjects who did not drink any beer. Wine consumption was also associated with a slightly lower risk of bladder cancer (p for trend = 0.054), although the numbers were very small (only 15 cases and 16 controls drank more than four glasses of wine per day). Drinking at least one glass of wine per day was associated with 0.59 (OR, 0.59; 95% CI, 0.44-0.81) times the risk of bladder cancer. No association was found between hard liquor consumption and bladder cancer (p for trend = 0.85). We also examined the alcoholic beverage-specific associations by volume (in 240 ml). Results were materially unchanged, and beer was the only alcoholic beverage statistically significantly associated with a lower risk of bladder cancer. More than 80% of our bladder cancer cases are Ta and T1 (i.e., superficial or noninvasive bladder cancers). Only less than 20% are T2+ (i.e., invasive bladder cancers). Nonetheless, we examined the association between alcohol and bladder cancer separately for superficial and invasive bladder cancers and the alcohol-related bladder cancer protection was mostly confined to superficial bladder cancers (p for trend = 0.003 and 0.88, for superficial and invasive bladder cancers, respectively). 39 Table 7. Basic characteristics of all the subjects included in the analyses Controls (N=1577 1 ) Cases (N=1573 1 ) Category of total alcohol consumption (drinks/day) 0 <1 1-4 >4 0 <1 1-4 >4 Number of subjects 453 385 505 234 432 364 512 265 All other beverages (ml/day) 2 2043 2057 2214 2689 2300 2114 2269 2721 Age (years) 56.2 56.0 56.8 56.7 55.6 55.8 57.2 56.5 Male (%) 61 77 86 95 61 76 87 91 White (%) 92 94 93 90 90 94 93 91 Years of education (%) 12 (High school or below) 33 25 27 46 44 31 42 51 13-15 (1-3 years of college) 27 29 31 28 28 31 28 28 16 (College graduate) 40 46 43 27 29 38 31 21 Alcohol consumption Mean frequency (drinks/day) 0 0.47 2.03 9.33 0 0.50 2.13 9.73 Mean duration (years) 0 26.4 29.2 29.8 0 25.9 29.3 29.4 Age at 1st drinking / 23.5 22.5 20.1 / 23.2 22.3 20.4 Smoking status (%) Never 59 37 26 14 25 24 13 6 Former 23 46 49 44 30 42 40 32 Current 17 17 25 43 45 34 47 61 1pack/day 15 22 25 14 19 15 17 8 >1 pack/day 25 42 49 73 56 60 70 86 <30 yrs 22 41 45 42 32 42 35 29 30 yrs 19 23 29 44 43 34 51 65 40 Table 7 continued. Controls (N=1577 1 ) Cases (N=1573 1 ) Category of total alcohol consumption (drinks/day) 0 <1 1-4 >4 0 <1 1-4 >4 NSAIDs intake over lifetime (%) Never/irregular 69 69 63 54 71 68 72 61 1-1441 pills 16 16 17 20 16 16 16 16 1441+ pills 15 15 20 26 12 16 13 23 Intake of carotenoids (%) 1st quintile 24 23 18 13 28 29 25 28 2nd quintile 17 19 21 21 19 20 20 15 3rd quintile 21 22 19 18 21 20 18 19 4th quintile 21 17 20 19 17 19 20 15 5th quintile 18 18 22 28 15 12 16 23 Hair dresser (%) 1 1 1 1 2 1 1 2 1 Frequency of alcohol consumption is missing for 13 cases and 9 controls.- 2 Intake of all other beverages (water, coffee, tea, milk, juice, hot chocolate, and soda). 41 Table 8. Total alcohol consumption and risk of bladder cancer OR (95% CI) Alcohol consumption Ca/Co Crude 1 Smoking-adjusted 2 Multivariate 3 Adjusted for total fluid 4 Nondrinkers 432/453 1.00 (reference) 1.00 (reference) 1.00 (reference) 1.00 (reference) Frequency (drinks/day) <1 364/385 0.99 (0.82-1.21) 0.85 (0.68-1.05) 0.85 (0.68-1.06) 0.85 (0.68-1.07) 1-4 512/505 1.07 (0.89-1.30) 0.75 (0.60-0.93) 0.77 (0.62-0.96) 0.78 (0.62-0.98) >4 265/234 1.19 (0.95-1.50) 0.66 (0.51-0.86) 0.68 (0.52-0.90) 0.67 (0.51-0.89) P trend 0.11 0.0008 0.003 0.003 Duration (years) 1-20 311/303 1.08 (0.87-1.33) 0.81 (0.64-1.03) 0.83 (0.65-1.05) 0.82 (0.64-1.06) 21-30 275/292 0.99 (0.79-1.23) 0.71 (0.55-0.90) 0.72 (0.56-0.92) 0.71 (0.55-0.93) 31-40 376/338 1.19 (0.97-1.47) 0.85 (0.67-1.08) 0.87 (0.68-1.10) 0.88 (0.69-1.13) 41+ 188/199 1.00 (0.77-1.30) 0.64 (0.47-0.86) 0.66 (0.48-0.89) 0.67 (0.49-0.92) P trend 0.47 0.007 0.017 0.032 Age at 1st use (years) 25+ 296/294 1.06 (0.86-1.31) 0.84 (0.67-1.07) 0.86 (0.68-1.09) 0.82 (0.64-1.04) 18-24 607/605 1.07 (0.89-1.29) 0.76 (0.62-0.94) 0.78 (0.63-0.96) 0.80 (0.64-1.00) <18 247/232 1.13 (0.90-1.42) 0.69 (0.53-0.90) 0.70 (0.54-0.92) 0.74 (0.56-0.98) P trend 0.31 0.002 0.005 0.025 1 Conditional logistic regression, without adjustment.- 2 Conditional logistic regression, adjusted for cigarette smoking status, duration of smoking, and intensity of smoking.- 3 Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber, cigarette smoking status, duration of smoking, and intensity of smoking.- 4 Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber, cigarette smoking status, duration of smoking, intensity of smoking, and intake of all other beverages (water, coffee, tea, milk, juice, hot chocolate, soda). 42 Table 9. Effect of total alcohol consumption on risk of bladder cancer by smoking status and sex Nonsmokers (N=856) 1 Smokers: <30 years (N=1131) 2 Smokers:30 years (N=1163) 2 Alcohol (drinks/day) Ca/Co OR (95% CI) Ca/Co OR (95% CI) Ca/Co OR (95% CI) Both sexes 0 108/268 1.00 140/100 1.00 184/85 1.00 <1 89/141 1.67 (1.17-2.40) 152/157 0.75 (0.52-1.08) 123/87 0.69 (0.47-1.03) 1-4 68/133 1.32 (0.89-1.95) 181/227 0.58 (0.41-0.82) 263/145 0.92 (0.65-1.31) >4 17/32 1.22 (0.62-2.40) 76/98 0.46 (0.30-0.70) 172/104 0.77 (0.52-1.15) P trend 0.16 <0.0001 0.46 Women 0 44/106 1.00 46/42 1.00 80/30 1.00 <1 22/36 1.63 (0.83-3.22) 32/33 1.29 (0.62-2.71) 32/19 0.52 (0.24-1.15) 1 8/18 1.14 (0.44-2.99) 31/43 0.89 (0.43-1.85) 53/21 0.75 (0.36-1.57) P trend 0.39 0.80 0.36 Men 0 64/162 1.00 94/58 1.00 104/55 1.00 <1 67/105 1.74 (1.13-2.68) 120/124 0.64 (0.41-0.99) 91/68 0.73 (0.45-1.17) 1-4 60/116 1.35 (0.87-2.09) 155/191 0.51 (0.34-0.77) 228/128 0.96 (0.63-1.44) >4 17/31 1.28 (0.64-2.57) 71/91 0.38 (0.24-0.63) 154/100 0.74 (0.48-1.16) P trend 0.23 <0.0001 0.43 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber. - 2 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber, cigarette smoking status, duration of smoking, and intensity of smoking. 43 Table 10. Total alcohol consumption and bladder cancer risk by daytime urination among subject who had smoked for less than 30 years Daytime urination: <4 times/day Daytime urination: 4 times/day Alcohol (drinks/day) Case/Control OR (95% CI) 1 Case/Control OR (95% CI) 1 0 26/29 1.00 61/32 1.00 <1 31/29 1.46 (0.61-3.47) 64/60 0.58 (0.32-1.06) 1-4 31/35 1.04 (0.45-2.44) 71/94 0.38 (0.21-0.67) >4 16/17 1.24 (0.44-3.50) 34/60 0.24 (0.12-0.47) P trend 0.91 <0.0001 P for interaction between alcohol intake and daytime urination = 0.057 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber, cigarette smoking status, duration of smoking, and intensity of smoking. 44 Table 11. Beverage-specific consumption and risk of bladder cancer Alcohol (drinks/day) Number of cases/controls OR (95% CI) 1 OR (95% CI) 2 Beer 0 660/647 1.00 1.00 <1 243/266 0.83 (0.65-1.06) 0.81 (0.61-1.07) 1-4 141/152 0.81 (0.61-1.08) 0.68 (0.49-0.95) >4 82/78 0.88 (0.61-1.28) 0.54 (0.35-0.83) P trend 0.26 0.002 Each 240 ml/day 1.00 (0.98-1.03) 0.96 (0.93-0.99) Wine 0 798/748 1.00 1.00 <1 231/266 0.72 (0.57-0.91) 0.84 (0.64-1.09) 1-4 86/115 0.57 (0.41-0.79) 0.65 (0.44-0.95) >4 15/16 0.77 (0.37-1.60) 0.91 (0.41-2.02) P trend 0.0003 0.054 Each 240 ml/day 0.88 (0.76-1.02) 0.92 (0.80-1.05) Liquor 0 648/709 1.00 1.00 <1 251/245 1.33 (1.06-1.69) 1.18 (0.90-1.55) 1-4 156/135 1.57 (1.18-2.10) 1.01 (0.72-1.41) >4 71/54 1.74 (1.16-2.63) 1.01 (0.63-1.62) P trend 0.0003 0.85 Each 240 ml/day 1.36 (1.09-1.70) 1.02 (0.81-1.28) 1 Conditional logistic regression, controlled for other types of alcoholic beverages. - 3 Conditional logistic regression, additionally adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, smoking status, duration and intensity of smoking. 45 2.3.4 Discussion Most previous studies reported either null associations (Najem et al., 1982; Mommsen et al., 1983; Thomas et al., 1983; Brownson et al., 1987; Iscovich et al., 1987; Mills et al., 1991; Chyou et al., 1993; Bruemmer et al., 1997; Michaud et al., 1999; Djousse et al., 2004) or slightly positive associations (Claude et al., 1986; Risch et al., 1988; Slattery et al., 1988; Nomura et al., 1989; Murata et al., 1996; Donato et al., 1997; Zeegers et al., 2001b) between consumption of alcoholic beverages and risk of bladder cancer. Our results regarding beverage-specific consumption are in agreement with some, but not all, previous studies. Beer was found to be significantly associated with lower risk of bladder cancer in both the Framingham Heart Study (>4 drinks/day vs. nondrinkers, OR, 0.50; 95% CI, 0.2-0.8; p for trend = 0.03) (Djousse et al., 2004) and an Italian study (drinkers vs. nondrinkers, OR, 0.69; 95% CI, 0.52-0.92) (Pelucchi et al., 2002), each reporting a similar magnitude of reduction in risk. Other studies found that liquor or spirit was associated with elevated risk of bladder cancer (Risch et al., 1988; Chyou et al., 1993; Zeegers et al., 2001b; Djousse et al., 2004), which is consistent with our findings among female subjects (OR associated with 1 shot of hard liquor, 2.39; 95% CI, 0.97-5.89), although the number of female subjects in our study is too small to be conclusive. Alcohol consumption is known to be associated with an increase in risk of cancers of the mouth, oropharynx, esophagus, liver and breast, with the risk of these diseases increasing with higher volume of consumption (Smith-Warner et al., 1998; 46 Gunzerath et al., 2004; Boffetta & Hashibe, 2006; Cho et al., 2006). The formation of acetaldehyde-DNA adducts, reactive oxygen and nitrogen species, and the impairment of methionine-folate metabolism and DNA methylation by alcohol, have been proposed as possible mechanisms for alcohol-related carcinogenesis (Boffetta & Hashibe, 2006; Cho et al., 2006). On the other hand, moderate consumption of alcoholic beverages have been found to protect against coronary heart disease, diabetes mellitus (Gunzerath et al., 2004; Beulens et al., 2005), renal cell carcinoma (Asal et al., 1988; Wolk et al., 1996; Parker et al., 2002; Hu et al., 2003; Nicodemus et al., 2004; Rashidkhani et al., 2005), prostate cancer (Baglietto et al., 2006), and non-Hodgkin’s lymphoma (Morton et al., 2005), perhaps through mechanisms such as improved immune response and increased insulin sensitivity. One possible mechanism that may be responsible for the alcohol-mediated bladder cancer protection is the urogenous-contact hypothesis (Melicow, 1974; Oyasu & Hopp, 1974; Jones & Ross, 1999; Michaud et al., 1999), which we will discuss below. Other mechanisms, perhaps not as well supported by the data, are the anti- inflammatory properties of alcohol (Imhof et al., 2001; Sierksma et al., 2002), and the anti-carcinogenic effects of polyphenols in red wine (Luceri et al., 2002) and beer (Gerhauser et al., 2002; Zhao et al., 2003a; Stevens & Page, 2004). The following findings are consistent with the urogenous-contact hypothesis, which associates bladder cancer risk with prolonged exposure to carcinogens in the urine (Melicow, 1974; Oyasu & Hopp, 1974; Jones & Ross, 1999; Michaud et al., 1999), may play a role in the alcohol-mediated bladder cancer protection. First, the 47 fact that beer was the beverage associated with the highest reduction in risk suggests that the amount of fluid may play a role in this association. Beer is consumed in greater volume than the other types of alcoholic beverages (1 can of beer 360 ml; one glass of wine 120 ml; one shot of hard liquor 45 ml). Second, the analyses stratified by frequency of urination, in which alcohol appeared to be most protective among individuals who had a high frequency of urination, also point to the urogenous- contact hypothesis playing a role in this association. Third, previous reports document the diuretic properties of alcohol in both experimental animals and human, with acute alcohol consumption increasing flow of urine (Bruger, 1940; Eggleton, 1941), possibly by inhibiting antidiuretic hormone (Epstein, 1997). Studies have shown that even in chronic alcohol abusers, an acute alcohol load can induce a diuretic response similar to that observed in non-alcoholic controls (Ogata, 1963; Kissin et al., 1964). These three lines of evidence suggest that increased fluid intake and frequency of urination may play a role in the alcohol-mediated bladder cancer protection by decreasing the time the bladder is exposed to carcinogens in the urine. Coffee drinking, which has also been postulated to have a diuretic effect, was not associated with bladder cancer risk in our study. The available literature has shown that doses of more than about 250-300 mg caffeine (equivalent to 2-3 cups of coffee) have a mild diuretic effect, while doses of less than 250 mg do not show any effect (Maughan & Griffin, 2003; Armstrong et al., 2005). Thus, doses of caffeine equivalent to the amount normally found in standard servings of tea, coffee and carbonated soft drinks appear to have no diuretic action (Maughan & Griffin, 2003). 48 In contrast, alcohol has a more potent diuretic effect and one gram is sufficient to increase urine output by approximately 10ml (Eggleton, 1941). Cigarette smoking was a modifier of the alcohol-bladder cancer association. Consumption of alcoholic beverages appeared most protective among shorter-term smokers, less protective among longer-term smokers and had no effect among nonsmokers. One could hypothesize that smokers would have benefited more from alcohol consumption than nonsmokers, if the mechanism of protection involves reducing exposure to cigarette-derived carcinogens in urine. Accordingly, we would expect the much lower level of carcinogens to preclude our observing the beneficial effects of alcoholic beverages among nonsmokers. On the other hand, the protective effect of alcohol by increasing flow of urine and reducing exposure in urine to carcinogens may be overwhelmed in longer-term smokers, in which the levels of carcinogens are too high to be eliminated by these anticarcinogenic properties of alcohol. The present study is a population-based case-control study, therefore designed to minimize selection bias. However, nonresponse rate was 31% among first eligible controls and only 17% among cases, leading to a potential selection bias problem. It is possible that nonresponse is related to a less altruistic, less healthy and thus drinking behavior, which would mean that alcohol consumption could be underestimated among our participating cases and controls, or more so among our controls (as one may expect because the cases have disease which may make them all more cooperative), but that underestimation would upwardly bias our estimates, 49 and thus become a force moving them toward the null, given that the associations we observed are inverse. We also believe that recall bias has probably not played an important role in our findings. Even though consumption of alcoholic beverages has been related to increased illnesses, it is not an established risk/protective factor for bladder cancer. One limitation of our study is the lack of data on drinking pattern, such as consuming the same total weekly amount of alcohol over a period of one or two days as compared with several days, which is correlated with other lifestyle factors and could have impacted on the effect of alcoholic beverages, for example, by modifying the diuretic effect of alcohol. To our knowledge, no studies that looked at the possible association between alcohol consumption and bladder cancer have examined the effect of drinking pattern. In addition, because we did not collect information about the type of wine consumed, we were unable to test the hypothesis that a protective effect of red wine may have been due to its high polyphenol content. Even though residual confounding from cigarette smoking has been a serious concern for most of the studies of alcohol that reported positive associations, it is not likely to explain the inverse associations we found between alcohol consumption and bladder cancer risk. However, confounding may have resulted from unmeasured socioeconomic and other lifestyle factors that differentiate drinkers and nondrinkers of beer, wine and hard liquor. As in any other epidemiological study, our findings may have been due to chance, although we believe this is not likely considering the consistency of our findings. For our large well-established population-based case- control study, we were able to match our cases and controls on age, sex, race and 50 neighborhood of residence. Because we had also available data describing consumption of fruit and vegetable, use of NSAIDs and use of hair dye, we were previously able to identify these factors as risk or protective factors for bladder cancer and subsequently control for them in our current analyses. In conclusion, our data showed that consumption of alcoholic beverages was associated with a decreased risk of bladder cancer. Beer was the beverage offering the highest protection. There was also some suggestive evidence of a moderate association between wine consumption and reduced risk of bladder cancer. The urogenous-contact hypothesis and the anti-inflammatory properties of alcohol are proposed as possible mechanisms for the alcohol-mediated bladder cancer protection. More well-designed epidemiologic studies are needed to confirm these findings 51 2.4 COFFEE 2.4.1 Introduction Coffee is one of the most widely consumed beverages in the world. In 1989, 52.5% of the US population over 10 years of age drank coffee. Coffee is rich in chemical compounds that may be carcinogenic such as nitrosamines, polycyclic aromatic hydrocarcons, heterocyclic amines, pesticide residues, and mycotoxins, and also compounds that may be protective such as water, kahweol/cafestol palmitates, and chlorogenic acids (IARC, 1991). The impact of coffee on bladder cancer risk has been investigated in numerous epidemiological studies (IARC, 1991; Sala et al., 2000; Tavani & La Vecchia, 2000; Zeegers et al., 2001a; Zeegers et al., 2004). In 1991, the International Agency for Research on Cancer (IARC) concluded that human studies are consistent with a weak positive relationship between coffee consumption and bladder cancer, but it is possible that this effect may be due to bias or confounding (IARC, 1991). Considering the complexity of the chemical composition of coffee, and the rich biological effects of those compounds, it is also likely that coffee may have a different impact in different populations. 2.4.2 Materials and Methods Study Population Refer to Chapter 2.1.2 Data Gathering and Exposure Definitions Refer to Chapter 2.1.2. Statistical Analysis Refer to Chapter 2.1.2. We also examined the possibility that the coffee-bladder cancer association could be modified by sex, age, cigarette- 52 smoking status, and frequency of urination during the day. Unconditional logistic regression was used in analyses of subsets defined by smoking status (lifetime nonsmokers, smokers of <30 years, smokers of 30 years) and frequency of urination during the day (<4 times, 4 times/day). In addition to all the covariates mentioned above, 10 age-sex strata (age groups of <46, 46-50, 51-55, 56-60 and >60 years for each sex), and race (non-Hispanic white, Hispanic white or African American/others) were included in the regression models to account for the matched sampling. Modification by sex or age was addressed by conditional logistic regression, since the cases and controls were designed to match on sex and age. ORs with 2-sided p values less than 0.05 were considered statistically significant. All p values presented are 2-sided. 2.4.3 Results Eighty four percent of our study controls reported drinking at least one cup of coffee per day (Table 12). Among them, 89% reported drinking regular coffee only, five percent reported drinking decaffeinated coffee only, and six percent reported drinking both regular and decaffeinated coffee. The average frequency of coffee drinking was 3.6 cups per day, and the average duration was 34 years. Subjects who drank at least one cup per day were older, more likely to be male and non-Hispanic white, compared to those who drank <1 cup per day. As expected, coffee drinking was also associated with cigarette smoking. Subjects who drank more coffee were more likely to be current smokers with a longer history of heavy smoking. 53 Coffee consumption was associated with a slightly increased risk of bladder cancer (p trend=0.062; Table 13). Drinking5 cups per day was associated with a 22 percent increase in risk (OR, 1.22; 95% CI, 0.91-1.63) among total subjects. A two- fold increase in bladder cancer risk was observed (OR, 2.06; 95% CI, 1.06-4.03) among women who drank5 cups per day, but no significant increase in risk was observed among men (OR, 1.12; 95% CI, 0.81-1.56). Cigarette smoking may be a potential confounder of the coffee-bladder cancer association (Table 13). However, a small increase in bladder cancer risk was observed among both smokers and lifetime nonsmokers in our study. The coffee- bladder cancer association did not significantly differ by cigarette smoking, i.e., drinking at least five cups of coffee per day was associated with a similar increase in bladder cancer risk among both female smokers (OR, 2.39; 95% CI, 0.84-6.83) and female nonsmokers (OR, 2.05; 95% CI, 0.91-4.61). The effect of coffee was strongly modified by age (Table 14). Among subjects who were 45 years or younger, drinking coffee was associated a significant increase in risk of bladder cancer in a dose-dependent manner (p=0.002). Compared to drinking less than one cup per day, drinking at least one cup per day was associated with 2.38 (95% CI, 1.30-4.36) times the risk of bladder cancer, and drinking at least five cups of coffee per day was associated with 4.6 (95% CI, 1.80- 11.76) times the risk. A similar effect was observed for both men and women. Drinking at least one cup of coffee per day was associated with 1.91 (95% CI, 0.94- 54 Table 12. Demographic characteristics of controls by coffee consumption. Frequency of coffee consumption (cups/day) <1 1-<3 3-<5 5 Number of subjects 254 588 414 330 All other beverages (ml/day) 2 1798 1653 1669 1881 Age (years) 50.6 55.0 55.1 55.6 Male (%) 72 76 78 87 Non-Hispanic white (%) 90 89 96 97 Years of education (%) 12 (High school or below) 28 31 29 36 13-15 (1-3 years of college) 27 30 29 28 16 (College graduate) 45 39 43 36 Coffee consumption Mean frequency (drinks/day) 0.5 1.6 3.4 7.8 Mean duration (years) 23.8 32.6 34.6 37.2 Age at 1st drinking 25.7 21.1 19.7 17.6 Regular (%) 63 85 93 94 Decaffeinated (%) 22 7 2 2 Both regular & decaffeinated (%) 16 8 4 4 Smoking status (%) Never 69 39 25 19 Former 18 40 49 46 Current 13 20 25 35 1 pack/day 15 23 23 13 >1 pack/day 16 37 52 68 <30 yrs 22 38 44 38 30 yrs 8 22 31 43 NSAIDs intake over lifetime (%) Never/irregular 71 67 63 59 1-1441 pills 14 18 17 17 1441+ pills 15 15 20 24 Intake of carotenoids (%) 1st quintile 29 19 21 16 2nd quintile 15 22 19 20 3rd quintile 24 22 16 19 4th quintile 16 20 20 21 5th quintile 15 18 24 24 1 Frequency of alcohol consumption is missing for 13 cases and 9 controls. 2 Intake of all other beverages (water, coffee, tea, milk, juice, hot chocolate, and soda). 55 Table 13. Effect of coffee consumption on risk of bladder cancer by smoking status and sex All 1 Nonsmokers (N=860) 2 Smokers (N=2312) 3 Coffee (cups/day) Case/Control OR (95% CI) Case/Control OR (95% CI) Case/Control OR (95% CI) Both sexes <1 178/254 1.00 (reference) 85/176 1.00 93/78 1.00 1-<3 501/588 1.00 (0.78-1.30) 106/231 1.01 (0.70-1.46) 395/357 0.99 (0.69-1.41) 3-<5 467/414 1.16 (0.88-1.53) 58/105 1.28 (0.83-1.99) 409/309 1.12 (0.78-1.61) 5 436/330 1.22 (0.91-1.63) 36/62 1.26 (0.75-2.11) 400/268 1.09 (0.75-1.58) P trend 0.062 0.22 0.37 1 1404/1332 1.09 (0.86-1.39) 200/398 1.11 (0.80-1.56) 1204/934 1.06 (0.75-1.48) Men <1 133/184 1.00 (reference) 60/126 1.00 73/58 1.00 1-<3 384/446 1.01 (0.75-1.36) 80/163 1.10 (0.71-1.69) 304/283 0.88 (0.58-1.33) 3-<5 366/321 1.17 (0.86-1.61) 46/75 1.44 (0.87-2.40) 320/246 1.00 (0.66-1.52) 5 350/286 1.12 (0.81-1.56) 24/50 1.12 (0.61-2.06) 326/236 0.93 (0.61-1.42) P trend 0.29 0.36 0.86 1 1100/1053 1.08 (0.82-1.43) 150/288 1.19 (0.80-1.76) 950/765 0.93 (0.63-1.38) Women <1 45/70 1.00 (reference) 25/50 1.00 20/20 1.00 1-<3 117/142 0.90 (0.52-1.54) 26/68 0.94 (0.45-1.96) 91/74 1.42 (0.68-3.00) 3-<5 101/93 1.05 (0.59-1.90) 12/30 1.11 (0.44-2.80) 89/63 1.55 (0.73-3.27) 5 86/44 2.06 (1.06-4.03) 12/12 2.39 (0.84-6.83) 74/32 2.05 (0.91-4.61) P trend 0.021 0.16 0.087 1 304/279 1.09 (0.66-1.81) 50/110 1.13 (0.58-2.22) 254/169 1.59 (0.79-3.21) 1 Conditional logistic regression. - 2 Unconditional logistic. - 3 Unconditional logistic regression, adjusted for smoking. 56 Table 14. Effect of coffee consumption on risk of bladder cancer by age Age:45 Age: 46-60 Age: >60 Coffee (cups/day) Ca/Co OR (95% CI) 1 Ca/Co OR (95% CI) 1 Ca/Co OR (95% CI) 1 Both sexes <1 34/63 1.00 98/127 1.00 46/64 1.00 1-<3 64/58 2.04 (1.06-3.90) 273/318 0.77 (0.54-1.11) 164/212 0.95 (0.56-1.61) 3-<5 38/34 2.16 (0.97-4.79) 237/227 0.75 (0.51-1.10) 192/153 1.50 (0.88-2.56) 5 36/17 4.60 (1.80-11.76) 250/189 0.83 (0.55-1.24) 150/124 1.25 (0.72-2.19) P trend 0.002 0.56 0.081 1 138/109 2.38 (1.30-4.36) 760/734 0.77 (0.55-1.09) 506/489 1.19 (0.73-1.94) Women <1 9/21 1.00 27/38 1.00 9/11 1.00 1-<3 18/12 17.80 (1.91-166.1) 62/72 0.70 (0.33-1.50) 37/58 0.29 (0.073-1.16) 3-<5 9/11 2.13 (0.29-15.39) 55/59 0.76 (0.33-1.74) 37/23 0.74 (0.18-2.96) 5 8/0 / 50/25 1.36 (0.53-3.48) 28/19 0.76 (0.18-3.20) P trend 0.026 0.42 0.27 1 35/23 8.82 (1.74-44.83) 167/156 0.80 (0.40-1.63) 102/100 0.50 (0.14-1.76) Men <1 25/42 1.00 71/89 1.00 37/53 1.00 1-<3 46/46 1.61 (0.76-3.42) 211/246 0.76 (0.50-1.15) 127/154 1.12 (0.63-2.01) 3-<5 29/23 2.03 (0.75-5.45) 182/168 0.74 (0.47-1.15) 155/130 1.60 (0.88-2.90) 5 28/17 3.21 (1.13-9.14) 200/164 0.73 (0.46-1.16) 122/105 1.38 (0.74-2.57) P trend 0.029 0.29 0.14 1 103/86 1.91 (0.94-3.86) 593/578 0.75 (0.50-1.11) 404/389 1.33 (0.77-2.30) 1 Conditional logistic regression. 57 3.86) times the risk of bladder cancer among men and 8.82 (95% CI, 1.74-44.83) times the risk among women. 2.4.4 Discussion In our study, a slight increase in bladder cancer risk was observed among heavy coffee drinkers, especially among women and younger subjects. Overall, our results are consistent with previous studies. The IARC has concluded that there is a weak positive association between coffee and bladder cancer (IARC, 1991). A meta-analysis of case-control studies published between 1971 and 1992 suggested a small but statistically significant increased risk among men (Viscoli et al., 1993). A recent review concluded that consumption of coffee increases the risk of bladder cancer at high levels of intake, but not at levels lower than five cups per day (1999). Tavani et al. reviewed the epidemiological studies published between 1990 and 1999 and concluded that a strong association between coffee drinking and bladder cancer could be excluded, and that it was still not clear whether the weak association identified was causal. Residual confounding from cigarette smoking has been proposed as a possible explanation; however, in a recent pooled analysis, lifetime nonsmokers were found to have small excess risks of bladder cancer if they drank10 cups per day (Sala et al., 2000). The authors proposed that confounding by cigarette smoking may not be the only explanation for the excess risk observed in previous studies. Heavy coffee drinkers may differ from other subjects in exposure to environmental tobacco smoke, intake of other fluids, 58 diet, occupation, and other behavioral or social aspects. Our study found a reduced risk associated with consumption of alcoholic beverages (Jiang et al., 2007a) and a slightly increased risk associated with environmental tobacco smoking (Jiang et al., 2007b). Adjustment for these factors did not change our observed coffee-bladder cancer association. One possible mechanism to explain this association relates to 1, 2, 4-benzenetriol, which is present in roasted coffee beans and is a major hydrogen peroxide generating component able to induce DNA single strand breaks. Studies conducted in human volunteers (Long & Halliwell, 2000; Hiramoto et al., 2002) have also found that drinking coffee increases hydrogen peroxide level in urine in a time-dependent fashion. We have proposed that alcohol could decrease the risk of bladder cancer by its diuretic properties (Jiang et al., 2007a). Caffeine has been found to have a mild diuretic effect mediated by alterations in the hormonal control of renal function, possible by blocking the anti-diuretic hormone (Spindel, 1984; Maughan & Griffin, 2003; Armstrong et al., 2005). The available literature has shown that doses of more than about 250-300 mg (equivalent to 2-3 cups of coffee) have a mild diuretic effect, while doses of less than 250 mg do not show any effect (Maughan & Griffin, 2003). Thus, doses of caffeine equivalent to the amount normally found in standard servings of tea, coffee and carbonated soft drinks appear to have no diuretic action (Maughan & Griffin, 2003; Armstrong et al., 2005). In our study, the coffee-bladder cancer association differed by age with the highest increase in risk confined to subjects 45 years or younger. In contrast, a study 59 conducted in New York from 1979 to 1985 found that the increased risk associated with coffee was higher among those 65 years and older compared to those under age 65 (Vena et al., 1993). Another study conducted on coffee and myocardial infarction found a higher risk among individuals younger than median age of 59 years (Cornelis et al., 2006). Genetic variations may explain the susceptibility to coffee among this young population. For example, the coffee-myocardial infarction association previously reported was found only among individuals with slow Cytochrome P450 1A2 (CYP1A2)*1F allele (Cornelis et al., 2006), which impairs caffeine metabolism, suggesting that caffeine plays a role in the association. In that study, the CYP1A2-cofffee interaction was observed only among individuals younger than the median age, suggesting that caffeine may have a greater relative effect on younger individuals. We will explore this potential CYP1A2-coffee interaction in Chapter 2.5. 60 2.5 GENE-ENVIRONMENT INTERACTIONS 2.5.1 Introduction Glutathione S-transferases (GSTs) are a family of Phase II detoxification enzymes that catalyse the conjugation of reduced glutathione to a variety of electrophilic xenobiotics (McIlwain et al., 2006). This activity of this family is important in detoxification of endogenous compounds as well as the metabolism of xenobiotics. Epidemiological studies have linked functional polymorphic variations in GSTs with cancer susceptibility (McIlwain et al., 2006). A recent meta-analysis of 28 studies found a summary OR of 1.5 (95% CI, 1.3-1.6) for GSTM1 null versus present genotype with no evidence of heterogeneity by cigarette smoking (Garcia- Closas et al., 2005). In the Los Angeles Bladder Cancer Case-Control Study, polymorphisms in GSTM1, GSTT1, and GSTP1 did not modify the association between cigarette smoking and bladder cancer (Castelao et al., 2001). Another route of detoxification of xenobiotics is N-acetylation, which is catalyzed by the N-acetyltransferases (NATs), preferentially by NAT2 in the liver and by NAT1 in skin cells (Gago-Dominguez et al., 2003). N-acetylation by hepatic NAT2 is an important detoxification pathway in metabolism of aromatic amines, the most important class of bladder carcinogens in cigarette smoke. Cigarette smoking was associated with a higher risk of bladder cancer among subjects with NAT2 slow acetylation (Garcia-Closas et al., 2005). In the Los Angeles Bladder Cancer Case- Control Study, NAT2 phenotype but not NAT1*10 genotype was found to affect 4- ABP-Hb adduct levels (Probst-Hensch et al., 2000). NAT2 slow acetylators had 61 significantly higher mean levels of 3- and 4-ABP Hb adducts than rapid acetylators. Women with slow NAT2 and rapid NAT1 were found to have a higher risk of bladder cancer from long-term use of permanent hair (Gago-Dominguez et al., 2001; Gago-Dominguez et al., 2003). The protective effect of carotenoids on bladder cancer also seemed to be affected by variations in NAT1 and NAT2 with the association confined to subjects having rapid NAT1 and rapid NAT2 (Castelao et al., 2004). N-oxidation has long been regarded as a necessary activation step for carcinogenic arylamines, while ring-oxidation has been generally considered to be an important detoxification mechanism for arylamines (Gago-Dominguez et al., 2003). Both enzymatic reactions are carried out in the liver and usually involve the CYP1A2. In the Los Angeles Bladder Cancer Case-Control Study, female carriers of ‘slow’ CYP1A2 were found to have a higher risk of bladder cancer from long-term use of permanent hair dyes (Gago-Dominguez et al., 2003). Carriers of ‘rapid’ CYP1A2 were more likely to benefit from intake of carotenoids (Castelao et al., 2004). Here we described our findings on the potential modifying effect of the different genotypes/phenotypes of enzymes involved in arylamine activation and/or detocification, including NAT1, NAT2, GSTM1, GSTT1, GSTP1, and CYP1A2 on the fluid intake-bladder cancer association. 62 2.5.2 Materials and Methods Blood and Urine Collection In January 1992, we expanded our original aims of this study to include the possible effects of tobacco-carcinogen metabolizing genes on risk of bladder cancer. Therefore, at the end of the in-person interview beginning in January 1992, each study subject was asked to provide a blood specimen and an overnight urine specimen following ingestion of a standard amount of caffeine in the afternoon of the previous day. Of the 1,044 cases asked, 724 (69%) provided an overnight urine sample, and 757 (73%) provided a blood sample. Similarly, among 979 controls, 689 (70%) and 770 (79%) provided urine and blood samples, respectively. Four 30 ml aliquots of urine for each study subject were stored at -20 o C. Blood specimens were collected in heparinized and non-heparinized tubes, respectively, each with 10 ml. Heparinized blood specimens were fractioned into plasma, buffy coat, and erythrocytes on the same day of the sample collection, and were continuously stored at –80 o C. DNA for genotyping has been extracted from non-heparinized blood. Genomic DNA was isolated from blood lymphocytes. Genotyping of GSTM1 (null versus non-null), GSTT 1(null versus non-null), GSTP1 (105 Ile/Val) were conducted as described previously (Gago-Dominguez et al., 2003). NAT1 genotyping was conducted for variations in the 3' region of NAT1 near the putative polyadenylation signal; namely, NAT1*10 (T1088A, C1095A), NAT1*3 (C1095A), and NAT1*11 (9 bp deletion between 1065–1090) (Bell et al., 1995). Genotypes containing at least one NAT1*10 allele were compared with other, non-NAT1*10 63 genotypes (NAT1*10 versus non-NAT1*10 genotypes), considering the NAT1*10 allele has been associated with higher risk of bladder cancer (Badawi et al., 1995). Methods for determination of CYP1A2 and NAT2 phenotypes using the caffeine-containing urine samples have been described before (Yu et al., 1994; Gago-Dominguez et al., 2003). Each subject was given two packets of instant coffee (about 70 mg of caffeine) to be drunk between 3 and 6 p.m. The subject collected an overnight urine sample (ending with the first morning void) into a 1-liter plastic bottle that was picked up and processed the same day. On the day of collection, the subject was briefly interviewed about caffeine intake (in addition to the prescribed packets of instant coffee) and use of acetaminophen on the previous day. Excessive caffeine consumption (>300 mg or >4 cups of coffee) has been shown to affect the validity of the phenotyping assay, and acetaminophen is used as the internal standard for the methylxanthine and urate assays so the presence of this compound may affect assay validity (Kalow & Tang, 1991). The urine specimens were acidified (400 mg of ascorbic acid per 20 ml of urine) prior to storage at -20°C. The NAT2 acetylator phenotype (slow, rapid) and CYP1A2 phenotype (slow, rapid) determinations were performed according to the method of Kalow and Tang (Kalow & Tang, 1991). Statistical Analysis The algorithm for classifying study subjects into CYP1A2-slow and CYP1A2-rapid individuals has been described before (Gago- Dominguez et al., 2003; Castelao et al., 2004). CYP1A2 is an inducible enzyme with cigarette smoking being a known inducer. Study subjects were first stratified by their smoking status around the time of urine collection (smokers and lifetime 64 nonsmokers). Among nonsmokers, subjects with CYP1A2 index levels below the subgroup median were labeled as CYP1A2-slow, while those with above median values were classified as CYP1A2-rapid. Among smokers, the range of smoking intensity (i.e., number of cigarettes smoked per day) was relatively narrow and the slope of the regression line between CYP1A2 index and number of cigarettes smoked per day around the time of urine collection was not statistically different from zero. Therefore, among smokers, the value of CYP1A2 index was not further adjusted by the number of cigarettes smoked per day. Smokers whose levels were above the subgroup median were labeled as CYP1A2-rapid while those with below median values were classified as CYP1A2-slow. The interaction between genes and fluid consumption was measured by ORs and corresponding 95% CIs from unconditional logistic regression. ORs with 2-sided p values less than 0.05 were considered statistically significant. All p-values quoted are 2-sided. 2.5.3 Results The association between specific alcoholic beverages (beer, wine, and liquor) and bladder cancer was modified by genetic variations in GSTM1 (Table 15) and GSTT1 (Table 16). Among subjects with the GSTM1 null genotype, no significant associations were detected between any of the alcoholic beverages and bladder cancer. However, among subjects with the GSTM1 non-null genotype, both beer (p trend=0.002) and wine (p trend=0.038) were associated with a significantly reduced 65 risk of bladder cancer, and liquor was associated with a significantly increased risk of bladder cancer (p trend=0.007). Similar differences in the alcoholic beverage- bladder cancer associations were also observed between subjects with the GSTT1 null and non-null genotypes. When GSTM1 and GSTT1 were examined in combination (Table 17), the alcoholic beverages-bladder cancer associations were more pronounced among individuals with high GST activity (GSTM1 non-null and GSTT1 non-null). The OR (95% CI) associated with drinking more than four cans of beer per day was 1.70 (0.32-8.89), 0.54 (0.28-1.05), and 0.35 (0.16-0.75) among subjects with low, intermediate, and high GST activities, respectively. The corresponding figures for drinking one or more glass of wine per day were 1.25 (0.37-4.25), 0.81 (0.47-1.40), and 0.66 (0.35-1.25), whereas the corresponding figures for hard liquor were 0.44 (0.09-2.10), 1.24 (0.54-2.82), and 2.21 (0.99-4.91). Compared to individuals with slow CYP1A2 phenotype, individuals with CYP1A2 rapid phenotype displayed a stronger alcohol–bladder cancer association, especially for wine (Table 18). OR (95% CI) associated with drinking one or more glasses of wine was 0.61 (0.36-1.05) among subjects with the rapid CYP1A2 phenotype and 1.30 (0.73-2.31) among subjects with slow CYP1A2 phenotype. Regarding coffee, subjects with the rapid CYP1A2 phenotype also displayed a stronger coffee-bladder cancer association compared with subjects with slow CYP1A2 phenotype (Table 19). A statistically significant association between coffee intake and bladder cancer risk was only observed among subjects exhibiting the rapid CYP1A2 phenotype (P for trend=0.035). The coffee-bladder cancer association 66 Table 15. Alcohol and bladder cancer risk by GSTM1 genotype GSTM1: null GSTM1: non-null Alcohol (drinks/day) Case/control OR 1 (95% CI) Case/control OR 1 (95% CI) Beer 0 189/171 1.00 166/187 1.00 <1 114/111 0.84 (0.56-1.27) 67/110 0.72 (0.46-1.13) 1-4 65/57 0.78 (0.48-1.27) 44/62 0.61 (0.36-1.03) >4 31/22 0.71 (0.36-1.38) 26/38 0.36 (0.19-0.70) P trend 0.27 0.0018 P for interaction=0.27 Wine 0 238/208 1.00 221/233 1.00 <1 112/100 1.09 (0.73-1.63) 58/117 0.60 (0.38-0.94) 1 51/54 0.92 (0.56-1.52) 24/48 0.66 (0.36-1.20) P trend 0.90 0.038 P for interaction=0.080 Liquor 0 193/202 1.00 151/218 1.00 <1 117/96 1.39 (0.93-2.08) 73/109 1.53 (0.97-2.40) 1-4 66/46 1.12 (0.68-1.86) 52/54 1.66 (0.98-2.80) >4 24/19 0.73 (0.35-1.50) 26/16 2.45 (1.16-5.20) P trend 1.00 0.0068 P for interaction=0.24 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber, cigarette smoking status, duration of smoking, and intensity of smoking. 67 Table 16. Alcohol and bladder cancer risk by GSTT1 genotype GSTT1: null GSTT1: non-null Alcohol (drinks/day) Case/control OR 1 (95% CI) Case/control OR 1 (95% CI) Beer 0 74/74 1.00 278/280 1.00 <1 32/33 0.75 (0.36-1.56) 149/185 0.79 (0.56-1.10) 1-4 22/20 0.89 (0.37-2.17) 86/98 0.67 (0.45-0.99) >4 13/13 0.70 (0.24-2.03) 44/48 0.44 (0.26-0.73) P trend 0.51 0.0013 P for interaction=0.98 Wine 0 90/92 1.00 365/348 1.00 <1 32/34 0.91 (0.42-1.95) 138/178 0.89 (0.64-1.24) 1 19/15 0.96 (0.38-2.47) 56/86 0.74 (0.49-1.13) P trend 0.82 0.19 P for interaction=0.60 Liquor 0 71/86 1.00 270/333 1.00 <1 37/29 2.31 (1.08-4.94) 152/172 1.28 (0.92-1.78) 1-4 26/17 1.84 (0.73-4.63) 92/81 1.23 (0.83-1.83) >4 7/8 0.84 (0.23-3.02) 43/27 1.42 (0.80-2.53) P trend 0.39 0.12 P for interaction=0.44 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber, cigarette smoking status, duration of smoking, and intensity of smoking. 68 Table 17. Alcohol and bladder cancer risk by GSTM1 and GSTT1 genotypes GSTM1 / GSTT1: both null Either null Both present Alcohol (drinks/day) Ca/co OR 1 (95% CI) Ca/co OR 1 (95% CI) Ca/co OR 1 (95% CI) Beer 0 46/36 1.00 170/172 1.00 138/148 1.00 <1 22/20 0.71 (0.27-1.87) 102/103 0.91 (0.59-1.42) 57/95 0.68 (0.41-1.11) 1-4 17/13 1.26 (0.42-3.81) 53/51 0.74 (0.43-1.25) 38/54 0.57 (0.32-1.03) >4 6/4 1.70 (0.32-8.89) 32/27 0.54 (0.28-1.05) 19/29 0.35 (0.16-0.75) P trend 0.58 0.078 0.003 Wine 0 49/47 1.00 229/205 1.00 179/188 1.00 <1 26/16 1.91 (0.70-5.24) 92/101 0.80 (0.52-1.24) 52/95 0.72 (0.44-1.18) 1 16/10 1.25 (0.37-4.25) 38/49 0.81 (0.47-1.40) 21/43 0.66 (0.35-1.25) P trend 0.37 0.50 0.081 Liquor 0 46/42 1.00 170/204 1.00 126/174 1.00 <1 25/17 1.51 (0.57-4.06) 104/89 1.76 (1.15-2.70) 60/94 1.24 (0.76-2.05) 1-4 14/7 1.39 (0.37-5.24) 64/48 1.40 (0.83-2.37) 40/43 1.42 (0.79-2.53) >4 6/7 0.44 (0.094-2.10) 19/13 1.24 (0.54-2.82) 25/15 2.21 (0.99-4.91) P trend 0.71 0.14 0.049 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber, cigarette smoking status, duration of smoking, and intensity of smoking. 69 Table 18. Alcohol and bladder cancer risk by CYP1A2 phenotype CYP1A2: slow CYP1A2: rapid Alcohol (drinks/day) Case/control OR 1 (95% CI) Case/control OR 1 (95% CI) Beer 0 178/162 1.00 161/147 1.00 <1 92/108 0.75 (0.49-1.16) 79/96 0.66 (0.42-1.05) 1-4 49/43 0.88 (0.50-1.53) 53/62 0.54 (0.32-0.90) >4 24/27 0.47 (0.24-0.95) 29/26 0.42 (0.21-0.83) P trend 0.097 0.0035 P for interaction=0.72 Wine 0 212/208 1.00 220/190 1.00 <1 88/97 1.02 (0.66-1.58) 69/88 0.77 (0.49-1.21) 1 42/34 1.30 (0.73-2.31) 35/54 0.61 (0.36-1.05) P trend 0.30 0.078 P for interaction=0.14 Liquor 0 166/185 1.00 163/172 1.00 <1 94/93 1.37 (0.88-2.15) 83/86 1.41 (0.90-2.22) 1-4 60/44 1.15 (0.68-1.94) 52/51 1.09 (0.65-1.84) >4 22/16 0.96 (0.43-2.12) 24/22 0.69 (0.33-1.43) P trend 0.69 0.74 P for interaction=0.90 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber, cigarette smoking status, duration of smoking, and intensity of smoking. 70 Table 19. Coffee and bladder cancer risk by CYP1A2 phenotype CYP1A2: slow CYP1A2: rapid Coffee (cups/day) Case/Control OR (95% CI) 1 Case/Control OR (95% CI) 1 Both sexes <1 38/62 1.00 35/53 1.00 1-<3 114/130 1.18 (0.69-2.03) 87/129 0.95 (0.54-1.68) 3-<5 110/81 1.51 (0.85-2.67) 101/84 1.37 (0.75-2.49) 5 80/67 0.96 (0.52-1.77) 100/64 1.62 (0.86-3.06) P trend 0.99 0.035 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber, cigarette smoking status, duration of smoking, and intensity of smoking. 71 Table 20. Coffee and bladder cancer risk by NAT2 phenotype NAT2: slow NAT2: rapid Coffee (cups/day) Case/Control OR (95% CI) 1 Case/Control OR (95% CI) 1 Both sexes <1 40/54 1.00 34/61 1.00 1-<3 115/125 1.06 (0.61-1.82) 88/136 1.06 (0.61-1.84) 3-<5 116/92 1.15 (0.65-2.05) 99/77 1.93 (1.07-3.47) 5 106/76 0.99 (0.54-1.82) 74/57 1.52 (0.80-2.89) P trend 0.98 0.037 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser/barber, cigarette smoking status, duration of smoking, and intensity of smoking. 72 was also modified by NAT2 phenotype (Table 20). Rapid NAT2 acetylators displayed a stronger coffee-bladder cancer association compared with slow NAT2 acetylators. In fact, statistically significant associations between coffee intake and bladder cancer risk were observed only among subjects exhibiting the NAT2 rapid acetylation phenotype (P for trend = 0.037). When CYP1A2 and NAT2 phenotypes were examined in combination, a significant association between coffee and bladder cancer was observed only among subjects with both rapid NAT2 and rapid CYP1A2. There were no discernible differences in the associations between fluid intake and bladder cancer by any other genes examined. 2.5.4 Discussion Interactions between alcohol and GSTs have been previously reported. Zheng et al. (2003) (Zheng et al., 2003) found that breast cancer risk was increased 6.8-fold for postmenopausal women with the GSTT1-null genotype who consume more than 250 kg of spirit equivalents. An 8.2-fold significantly increased risk was observed among heavier drinkers who had GSTM1A and GSTT1-null genotypes. In another study (Rundle et al., 2003), alcohol drinkers with the GSTM1 null genotypes had increased levels of PAH-DNA adducts in breast tissue of cases. These studies found an increased risk of breast cancer among subjects with the low activity GST genotypes, whereas our study found an increased bladder cancer protection among subjects with the high-activity GST genotypes. It is unclear how GSTM1 and GSTT1 participate in the inverse association between alcohol and bladder cancer. It is 73 possible that GSTs may affect the metabolism of alcohol and therefore affect bladder cancer development. In addition, it has been shown that ethanol exposure can affect GST activity (Hetu et al., 1982; Munoz et al., 1987; Gonzalez et al., 1988; Yang & Carlson, 1991; Van de Wiel et al., 1993; Carlson et al., 1995; Vanhaecke et al., 2000a; Vanhaecke et al., 2000b). Some experimental studies have found that GST activity was increased after chronic administration of ethanol (Hetu et al., 1982; Munoz et al., 1987; Yang & Carlson, 1991; Vanhaecke et al., 2000a; Vanhaecke et al., 2000b); however, other studies found that chonic administration of ethanol reduced (Van de Wiel et al., 1993) or had no effect on GST activity (Carlson et al., 1995). The mechanism by which CYP1A2 modifies the alcohol-bladder cancer association is not clear. Alcohol is predominantly metabolized by CYP2E1 (Havrda et al., 2005), but CYP1A2 also contributes to alcohol metabolism when only a small amount of alcohol is consumed. Prenylflavonoids found in beer have been found to inhibit CYP1A2 in vitro studies (Havrda et al., 2005). Caffeine is an aromatic amine, which is metabolized in the liver by CYP1A2 (Butler et al., 1989) and NAT2 (Straka et al., 2006). Given the same level of caffeine intake, rapid metabolizers will be exposed to lower caffeine levels than slow metabolizers. If caffeine played a role in the coffee-related increase in bladder cancer risk, subjects with slow CYP1A2 and slow NAT2 phenotype would exhibit a stronger coffee-bladder cancer association than their high-activity counterparts. This is consistent with a study of coffee and myocardial infarction in which coffee was 74 associated with an increased risk of myocardial infarction only among individuals with slow caffeine metabolizer (Cornelis et al., 2006). However, this is contrary to our observation that a significant coffee-bladder cancer association was only observed among rapid metabolizers. Thus, caffeine is unlikely to have played a role in the association we observed between coffee and bladder cancer. In addition to the roles of CYP1A2 and NAT2 in detoxification of carcinogens such as aromatic amines, both of these enzymes also participate in activation of precarcinogens such as heterocyclic amines (Boobis et al., 1994; Straka et al., 2006). Carcinogens such as nitrosamines (IARC, 1991) and possibly heterocyclic amines (Kato et al., 1991) have been found in coffee. Thus, the observed effect modification by CYP1A2 and NAT2 suggests a possible involvement of carcinogens such as nitrosamines and heterocyclic amines in the coffee-bladder cancer association. 75 CHAPTER 3 MEDICAL HISTORY, MEDICATION AND BLADDER CANCER RISK 3.1 URINARY TRACT INFECTIONS 3.1.1 Introduction Urinary tract infections (UTIs) are the most common kidney and urologic diseases in the United States, especially among women. Approximately 50% of women have at least one symptomatic UTI during their lifetime, and many have recurrent episodes (Stamm, 2002). However, UTIs in men are uncommon until after age 50, and are usually indicative of an underlying urologic abnormality (Stamm & Hooton, 1993). In the US, the majority of uncomplicated UTIs are caused by Escherichia coli (80%) or Staphylococcus saprophyticus (10-15%) (Stamm & Hooton, 1993). Bladder cancer, particularly the squamous cell carcinoma, is also highly prevalent in some areas of Africa and the Middle East such as Egypt, where chronic infection with Schistosoma haematobium is common (IARC, 1994). A experimental study in rats has shown associations between chronic infections with Escherichia coli and early bladder neoplasia (Davis et al., 1991). Associations between a history of UTIs and bladder cancer have been examined in several population studies conducted in Western countries (Howe et al., 1980; Kantor et al., 1984; Piper et al., 1986; Kjaer et al., 1989; Gonzalez et al., 1991; La Vecchia et al., 1991; Sturgeon et al., 1994; Parker et al., 2004), and results were not entirely consistent. 76 3.1.2 Materials and Methods Study Subjects Refer to Chapter 2.1.2. Data Collection A structured questionnaire was used during in-person interviews to request general and exposure information up to two years prior to diagnosis of cancer for the cases, and two years prior to diagnosis of cancer of the index case for the matched controls. Each subject was asked to report information on demographic characteristics, lifetime use of tobacco products and alcohol, usual adult dietary habits, lifetime occupational history, prior medical conditions, and prior use of medications. The questionnaire included questions on histories of physician- diagnosed hypertension, diabetes, angina, heart attack, stroke, analgesic nephropathy, renal papillary necrosis, kidney/renal stones, bladder stones, polycystic kidney disease, hereditary kidney disease, and injury to kidney. After January 1992, history of tuberculosis, thyroid disease, polycythemia vera, and gout was also asked. For each medical condition, we also asked the year of first diagnosis by a doctor. In addition, participants were asked if they were told by a doctor that they had kidney infections, bladder infections, or other type of UTIs. If so, the year of first and first diagnosis, and total number of diagnoses were also requested. Statistical Analysis Data were analyzed by standard matched-pair methods (Breslow and Day, 1980). The associations of bladder cancer with medical histories were measured by ORs and their corresponding 95% CIs and P-values. Conditional logistic regression models were used to examine the relationship between any medical history and bladder cancer risk with adjustment for other risk factors for 77 bladder cancer: average number of cigarettes smoked per day, number of years of smoking, smoking status in reference year (smoker or nonsmoker) (Castelao et al., 2001), level of education (high school or less, some college, college or above), lifetime use of NSAIDs (non/irregular user, < 1441 pills,1441 pills over lifetime) (Castelao et al., 2000), intake of carotenoids (quintiles) (Castelao et al., 2004), and duration of employment as a hairdresser/barber (years) (Gago-Dominguez et al., 2001). Pairs in which either the case or the control failed to answer a question were eliminated from the corresponding analysis. ORs with 2-sided P-values less than 0.05 were considered statistically significant. All P-values quoted are 2-sided. 3.1.3 Results History of one or more UTIs was reported by 26% of cases and 28% of controls (Table 21), occuring more frequently in women (46% of cases and 57% of controls) than in men (20% of cases and 20% of controls). In women, the most frequent UTIs were bladder infections; in men, however, UTIs occurred more frequently in urinary tract organs other than kidney and bladder, presumably urethra and ureter. Thus, there were 200 control women reporting a history of UTI; among them, 169 (85%) reported a history of bladder infections, 43 (22%) reported kidney infections, and only 18 (9%) reported other UTIs (presumably urethra & ureter). The corresponding figures among men are 100 (41%), 52 (21%), and 115 (48%), respectively. 78 A history of any UTI was not associated with bladder cancer risk among all subjects, men and women combined (OR, 1.00; 95% CI, 0.83-1.20; Table 21). A marginally significant inverse association was observed among women, but not among men (P for interaction with sex=0.039). Women reporting a history of UTI had a 26% lower risk of bladder cancer than women reporting no history of any UTI (OR, 0.74; 95% CI, 0.53-1.03). A history of kidney infection or other UTIs (ureteral, urethral) was not significantly associated with bladder cancer. However, risk was slightly elevated among men reporting a history of other UTIs (OR, 1.35; 95% CI, 0.99-1.83), especially among those men reporting infection ocurring within five years of cancer diagnosis (OR, 2.85; 95% CI, 1.34-6.06). Bladder infections were much more frequent among women than men in our study (48% vs. 8%, respectively). A history of bladder infection was associated with a reduced risk of bladder cancer (OR, 0.79; 95% CI, 0.62-1.01). This inverse association was observed in women (OR, 0.65, 95% CI, 0.45-0.94), but not in men (OR, 0.95; 95% CI, 0.68-1.34). Women who were diagnosed with a bladder infection five or more years before cancer diagnosis still experienced a significant reduction in risk (OR, 0.52; 95% CI, 0.34-0.81), suggesting that the detected inverse association between history of bladder infection and bladder cancer is unlikely to be a consequence of the cancer (Table 22). Frequency of bladder infection was statistically significantly associated with reduced risk of bladder cancer (Table 22, P for trend=0.015), with the highest reduction in risk associated with multiple infections. ORs (95% CI) for women with 1, 2-3, and >3 diagnoses of bladder 79 infection were 0.88 (0.50-1.58), 0.60 (0.35, 1.04), and 0.60 (0.37-0.98), respectively, compared to women who were never diagnosed with a bladder infection. Lower risk of bladder cancer was mostly associated with infections that occurred before age 50. Compared to women who were never diagnosed with a bladder infection, women who were 50 years of age or older at the time of first bladder infection did not experience a reduction in bladder cancer risk (OR, 1.19; 95% CI, 0.48-2.96), while a 42% (OR, 0.58; 95% CI, 0.39-0.87) lower risk of bladder cancer was observed among women who were between 20 and 49 years of age at the time of first diagnosis of bladder infection. A significant reduction in bladder cancer risk was only observed among women who had at least two infections occuring before age 50 (Table 23). In some studies in which UTIs were associated with increased risk of bladder cancer, cigarette smoking has been found to potentiate the detrimental effect of UTIs (Kantor et al., 1988; Dolin et al., 1994). In our study, the bladder infection-bladder cancer inverse association was also modified by smoking status. The decrease in bladder cancer risk appeared to be confined to female smokers (Table 23), although the test for interaction between cigarette smoking and bladder infections was not statistically significant (p for interaction=0.45). Among female smokers, a history of multiple bladder infections (>3) was associated with a 0.46 (95% CI, 026-0.82) times the risk of bladder cancer. No association was found among female nonsmokers. Regarding grade and stage of disease, a history of bladder infection was associated with reduced risk of only high-grade bladder cancer, particularly grade 3- 80 Table 21. Urinary tract infections and risk of bladder cancer by sex All Men Women Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Any UTI No 1172/1144 1.00 988/995 1.00 184/149 1.00 Yes 414/442 1.00 (0.83-1.20) 249/242 1.15 (0.92-1.44) 165/200 0.74 (0.53-1.03) Bladder infection 2 No 1365/1317 1.00 1148/1137 1.00 217/180 1.00 Yes 221/269 0.79 (0.62-1.01) 89/100 0.95 (0.68-1.34) 132/169 0.65 (0.45-0.94) Kidney infection 2 No 1480/1491 1.00 1179/1185 1.00 301/306 1.00 Yes 106/95 1.13 (0.82-1.56) 58/52 1.01 (0.65-1.55) 48/43 1.33 (0.80-2.21) Other urinary infection 2 No 1443/1453 1.00 1108/1122 1.00 335/331 1.00 Yes 143/133 1.28 (0.97-1.70) 129/115 1.35 (0.99-1.83) 14/18 0.77 (0.34-1.76) 1: Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 2: Additionally adjusted for infections at the other two sites. 81 Table 22. Bladder infections and bladder cancer Men Women History of bladder infection Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) No 1148/1137 1.00 (reference) 217/180 1.00 (reference) Yes 89/100 0.96 (0.69-1.35) 132/169 0.66 (0.46-0.96) Duration between first infection and cancer diagnosis (years) <5 24/19 1.40 (0.72-2.75) 39/39 1.03 (0.55-1.93) 5 53/73 0.82 (0.54-1.24) 70/110 0.52 (0.34-0.81) Frequency of infection (times) 1 49/75 0.66 (0.43-1.01) 44/46 0.88 (0.50-1.58) 2-3 20/14 1.75 (0.79-3.88) 43/55 0.60 (0.35-1.04) >3 20/11 2.10 (0.92-4.78) 45/68 0.60 (0.37-0.98) p trend 0.25 0.015 Age at first infection (years) <20 2/5 0.40 (0.072-2.20) 14/17 0.68 (0.28-1.66) 20-49 56/70 0.88 (0.58-1.35) 88/131 0.58 (0.39-0.87) 50 20/19 1.17 (0.58-2.35) 15/12 1.19 (0.48-2.96) Age at last infection (years) <40 18/45 0.40 (0.21-0.77) 47/69 0.52 (0.31-0.89 40-49 26/21 1.28 (0.66-2.49) 27/37 0.57 (0.29-1.11) 50 34/27 1.69 (0.95-3.02) 36/45 0.80 (0.44-1.46) 1 Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 82 Table 23. Bladder infections and bladder cancer by cigarette smoking and grade Men Women Frequency of infection Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) All 0 1148/1137 1.00 217/180 1.00 1 49/73 0.66 (0.43-1.01) 44/46 0.86 (0.53-1.41) 2-3 20/14 1.75 (0.79-3.88) 43/55 0.66 (0.40-1.06) >3 20/11 2.10 (0.92-4.78) 45/68 0.64 (0.40-1.01) p trend 0.25 0.022 Smoking status Nonsmokers 0 193/383 1.00 43/88 1.00 1 10/24 1.02 (0.47-2.24) 9/21 0.82 (0.33-2.06) 2-3 5/6 1.75 (0.52-5.92) 9/27 0.68 (0.28-1.66) >3 3/1 6.10 (0.62-60.19) 14/24 1.18 (0.53-2.64) p trend 0.12 1.00 Smokers 0 955/754 1.00 174/92 1.00 1 39/51 0.68 (0.43-1.06) 35/25 0.84 (0.45-1.54) 2-3 15/8 1.79 (0.74-4.35) 34/28 0.65 (0.36-1.19) >3 17/10 1.64 (0.72-3.72) 31/44 0.46 (0.26-0.82) p trend 0.43 0.005 83 Table 23 continued. Men Women Frequency of infection Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Histological grading Grade 1 0 221/214 1.00 46/51 1.00 1 3/14 0.094 (0.011-0.76) 10/11 0.97 (0.25-3.80) 2-3 4/14 1.37 (0.10-19.09) 14/6 2.69 (0.73-9.93) >3 3/1 1.38 (0.22-8.82) 15/17 1.35 (0.53-3.44) p trend 0.76 0.38 Grade 2 0 446/454 1.00 95/74 1.00 1 26/27 1.07 (0.56-2.07) 18/24 0.87 (0.35-2.14) 2-3 9/3 3.34 (0.67-16.57) 21/25 0.71 (0.31-1.63) >3 7/4 2.60 (0.61-11.01) 18/29 0.62 (0.26-1.47) p trend 0.074 0.21 Grade 3-4 0 386/376 1.00 64/42 1.00 1 17/27 0.59 (0.28-1.22) 9/10 0.28 (0.047-1.66) 2-3 4/8 0.89 (0.22-3.60) 6/16 0.13 (0.020-0.81) >3 8/4 2.44 (0.59-10.10) 7/18 0.18 (0.047-0.68) p trend 0.83 0.002 1 Unconditional logistic regression, adjusted for age, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 84 4 (Table 23). Women reporting a history of multiple bladder infections experienced a 0.18 (95% CI, 0.05-0.68) times the risk of grade 3-4 bladder cancer. No differences in the bladder infection-bladder cancer association were detected by stage or invasiveness of disease. Diabetes and bladder or kidney stones, which are associated with an increased susceptibility to UTIs, may have confounded the bladder infection-bladder cancer association. In our study, a history of diabetes and a history of kidney stones were each independently associated with an increased risk of bladder cancer (OR, 1.66; 95%, 1.15-2.39; OR, 1.25; 95% CI, 0.94-1.65, respectively). Very few subjects reported a history of bladder stones. The inverse association between bladder infections and bladder cancer remained unchanged after excluding subjects reporting a history of diabetes, kidney and/or bladder stones. 3.1.4 Discussion Our study found a significant association between bladder infections and reduced risk of bladder cancer among women, with the highest protection found among women who experienced multiple infections. This inverse association was modified by cigarette smoking and was limited to high grade bladder cancer. Seven case-control studies have examined the association between UTIs and bladder cancer risk in developed countries, where about 90% of bladder cancers are transitional cell carcinoma (Howe et al., 1980; Kantor et al., 1984; Claude et al., 1986; Piper et al., 1986; Kjaer et al., 1989; Gonzalez et al., 1991; La Vecchia et al., 85 1991; Sturgeon et al., 1994). The overall evidence supports an increased risk from UTIs. However, much higher risk was observed when infections occurred approximately within five years of cancer diagnosis, and risk was substantially reduced or disappeared when infections occurred many years prior to cancer diagnosis (Howe et al., 1980; Gonzalez et al., 1991; La Vecchia et al., 1991). Since the dates of UTIs were not obtained in all these studies, it is also possible that the occurrence of UTIs may have been the consequence of early bladder cancer, rather than a cause of the disease. In addition, three out of the seven studies (Claude et al., 1986; Kantor et al., 1988; Gonzalez et al., 1991), all of which found increased risks from UTIs, did not discriminate between the different types of UTI. This is an important consideration, since the different types of UTI may have different impact on bladder cancer risk. In our study, a decreased risk of bladder cancer was found for bladder infections in women but an increased risk was observed with infections occurring in urinary tract organs other than kidney and bladder in men, most likely urethra. Majority urethritis in men is secondary to sexually transmitted disease Gonorrhea and Chlamydia (Wong & Stamm, 1983). Consistent with this notion, a history of gonorrhea was associated with significantly increased risk of bladder cancer in two studies (Mommsen & Sell, 1983; La Vecchia et al., 1991). UTIs are the most common kidney and urologic disease in the United States, especially among women. Approximately 50% of women have at least one symptomatic UTI during their lifetime (Stamm, 2002), while UTIs in men are uncommon until after age 50 years (Stamm & Hooton, 1993). If a history of UTI 86 indeed increased the risk of bladder cancer, one would expect a higher incidence of bladder cancer among women than men, as well as a higher incidence of squamous cell carcinoma than transitional cell carcinoma (TCC) (Burin et al., 1995). However, bladder cancer is 3-4 times more common in men than in women, and approximately 90% of bladder cancers in the US are TCC and only 7% are squamous carcinomas (Yu & Ross, 2002). Our study found a significant association between bladder infections, especially multiple infections, and reduced risk of bladder cancer. There are several possible reasons to explain these findings. One of them relates to the antibiotics used to treat bladder infections, which have been found in experimental studies to inhibit the growth of bladder cancer (Seay et al., 1996; Ebisuno et al., 1997; Aranha et al., 2000; Aranha et al., 2002; Kamat & Lamm, 2004). The traditional treatment of uncomplicated UTIs has been a short-course of antibiotics such as amoxicillin, trimethoprim (TMP) alone or with sulfamethoxazole (TMP-SMX), nitrofurantoin, cephalosporins, and fosfomycin (Stamm, 2002). Fluoroquinolones, which first became available in the mid-1980s, are also highly effective for therapy (Stamm & Hooton, 1993). Antibiotic agents preferred for UTIs are agents that are primarily excreted through urinary tract and produce prolonged and high concentrations both in the urine and in vaginal secretions (Stamm, 2002). About 70-90% of TMP and 30% of SMX are excreted unchanged in the urine (Masters et al., 2003). Nitrofurantoin is well concentrated in the urine achieving levels of 200µg/ml or more (Cunha, 1988). After four days of continuous oral dosing of 250 mg of 87 ciprofloxacin twice a day, peak urinary concentrations exceeds 250µg/ml with levels remaining near 100µg/ml for at least eight hours (Gonzalez et al., 1984). Besides their bactericidal effect, the commonly used antibiotics TMP-SMX, cefazolin (cephalosporin), and nitrofurantoin significantly inhibited cell proliferation in human TCC cell line HT89 (grade 2), T24 (grade 3), and TccSUP (grade 4) (Seay et al., 1996), producing a proliferation inhibition rate as high as 95.4%. Inhibition occurred in a dose-dependent manner at concentrations attainable in the urine after oral administration and the maximal effect did not differ among the different antibiotics (Kamat & Lamm, 2004). Ciprofloxacin was also found to significantly inhibit cell proliferation in human TCC cell line TccSUP (Seay et al., 1996; Kamat & Lamm, 2004), T24 (Seay et al., 1996; Ebisuno et al., 1997; Kamat & Lamm, 2004), HTB9 (Kamat & Lamm, 2004), and J82 (Seay et al., 1996) (grade 3) in a dose- and time-dependent manner. Significant cytoxicity was seen starting at 12.5 µg/mL and reached maximal inhibition rate at around 400µg/mL with the inhibition rate as high as 98.2%. Similar cytotoxicities were also seen from other fluroqquinolones, such as ofloxacin (Seay et al., 1996) and fleroxacin (Ebisuno et al., 1997). The mechanism for antibiotics’ cytotoxicity against bladder cancer cells remains unclear. Ciprofloxacin, a fluroquinolone antibiotic, was shown to cause cell cycle arrest, disruption of calcium homeostasis, mitochondrial swelling, and redistribution of Bax (a pro-apoptotic protein) to the mitochondrial membrane and eventually lead to apoptosis in human TCC lines (Aranha et al., 2000; Aranha et al., 2002). It has been shown that antibiotic treatment for UTIs reduced urinary 88 prostaglandin E2 levels, which was increased in bladder cancer (Wheeler et al., 2002; Pruthi et al., 2004). Antibiotics used to treat UTIs can be ideal candidates as chemopreventive agents. More than 70% of the patients with bladder cancer present initially with superficial tumors (Raghavan et al., 1990). The standard initial treatment of superficial bladder tumors is transurethral resection (TUR) (Brunicardi et al., 2004). However, disease will recur in 60-70% of patients depending on cancer grade, tumor stage, and number of tumors (Amling, 2001). In an attempt to prevent recurrence, intravesical instillation of agents such as thiotepa, mitomycin, doxorubicin, and Bacillus Calmette-Guérin (BCG) (Raghavan et al., 1990) has been widely used. Tumor recurrence rate can be reduced to 30-40% when intravesical therapies are used in conjunction with TUR (Amling, 2001). However, these treatments may have side effects such as leucopenia from thiotepa treatment, genital rash from mitomycin treatment (Thrasher & Crawford, 1992; Koya et al., 2006), and require intravesical instillation. Because of these side effects and the unacceptable recurrence rates after TUR, alternative treatment modalities are needed to improve the disease-free interval, as well as the overall survival for bladder cancer (Aranha et al., 2000). Fluroquinolones and other antibiotics used to treat bladder infections are administered orally, are highly concentrated in the urine, and are relatively nontoxic. A considerably higher level can be achieved in urine than in serum with increasing oral intake of fluroquinolones, such as viprofloxacin and ofloxacin (Hoffken et al., 1985; Lode et al., 1987; Seay et al., 1996). Thus, other tissues are protected from 89 their cytotoxicity, but the malignant or pre-malignant urothelial cells are heavily exposed to induce apoptotic cell death (Aranha et al., 2000). A clinical observation study found a recurrence rate of 22.5% in bladder cancer patients treated with pefloxacin (a fluoroquinolone) after TUR and adjuvant intravesical treatments (Lovisolo et al., 1997), lower than the 30-40% recurrence rates from TUR and intravesical treatments alone (Lamm & Griffith, 1992; Lamm et al., 1994; Alexandroff et al., 1999; Amling, 2001; Busby & Kamat, 2006). However, well- designed randomized clinical trials are still needed to prove the efficacy of antibiotics. We are not surprised by the observation that the risk reduction effect of bladder infections was confined to women. We believe that the lack of exposure- disease association among our male subjects is likely the result of low exposure rate leading to low study power to detect the anticipated association. In fact, with a hypothesized 40% reduction in bladder cancer risk associated with a history of multiple (2) bladder infections, the present study possesses 86% statistical power to detect an association in women but only 35% power to observe the same association in men, under the assumption of a 5% significance level (Gauderman & Morrison, 2006; Jemal et al., 2007). Furthermore, the highest reduction in risk was associated with multiple infections and infections occurred before age 50, which are rare among men (Stamm & Hooton, 1993). The bladder infection-bladder cancer association was modified by frequency of bladder infection, with the highest reduction in risk confined to women with two 90 or more bladder infections. An estimated 20-30% of women with a first UTI will have recurrent episodes, and 5% will have chronic recurring infections (Foxman et al., 2000). For women who have had two or more symptomatic UTIs during one 6- month period or three or more such infections over a 12-month period, a low-dose antimicrobial prophylaxis taken daily or thrice weekly for 3-6 months is recommended to prevent recurrent infections (Stamm & Hooton, 1993; Fihn, 2003). Dosage seems to play a role in the anticancer effect of antibiotics used to treat UTIs. It has been reported that bladder cancer cell growth was inhibited by ciprofloxacin in a dose- and time-dependent fashion, but the inhibition was reversible at lower doses, and the rate of reversal was dependent on the initial drug concentration (Zehavi- Willner & Shalit, 1992). However, exposure of bladder cancer cells to a higher dose, i.e. 250µg/mL, ciprofloxacin resulted in a lytic effect on the cells, and the antiproliferative effect was irreversible (Zehavi-Willner & Shalit, 1992). Given the long-term use of antibiotics among women with recurrent bladder infections, it is possible that higher concentrations of antibiotics are reached in their urine for longer periods of time, at levels high enough to exert irreversible antiproliferative effect on bladder cancer cells. It is unclear why the reduction in bladder cancer risk was limited to bladder infections that occurred before age 50. Among younger women, susceptibility to recurrence can be affected by genetic, biologic, anatomic, as well as behavioral factors, particularly sexual intercourse and use of diaphragms and spermicides (Stamm & Hooton, 1993; Stamm, 2002). Among postmenopausal women, 91 recurrence is also common, possibly attributable to changes in the vaginal microflora caused by estrogen deficiency (Stamm & Hooton, 1993). Estrogen has been shown to inhibit the growth of bladder cancer in experimental studies (Bertram & Craig, 1972; Okajima et al., 1975; Tanahashi et al., 1977; Reid et al., 1984; Shirai et al., 1987). It is also possible that bladder infections that occurred after age 50 could constitute an early sign of bladder cancer, since the mean duration between the infection and diagnosis of cancer was only 6.1 years in our study. Cigarette smoking is one of the major risk factors for TCC of the bladder (Albanes et al., 1996). We found that the inverse association between bladder infections and bladder cancer was confined to smokers. One could hypothesize that smokers would benefit more than nonsmokers from bladder infections if the mechanism of protection is killing pre-cancer cells, which are more common among smokers. The lack of study power among female nonsmokers may also explain this difference by smoking status. We proposed that cytotoxicity from antibiotic therapy may explain the bladder infection-associated reduction in risk of bladder cancer. In addition, immune response triggered by infection may also play a role. BCG immunotherapy for bladder cancer has been widely accepted for its effect in reducing recurrent rate and increasing median time to recurrence (Alexandroff et al., 1999). BCG infection induces release of chemokines/cytokines, attract and activate neutrophils, macrophages and T-cells, resulting in improved recognition and killing of tumor cells (Alexandroff et al., 1999). It is possible that bladder infection, which is mostly 92 caused by E. coli, can trigger similar immune responses as BCG treatment, killing hyperplastic cells before they transform into cancerous cells. The present study is a population-based case-control study, therefore designed to minimize selection bias. We believe that recall bias has probably not played an important role in our findings, since the observed association is inverse. Compared to early studies, this study was able to obtain the age at first and last infection and were therefore able to identify that the infection was not likely a result of bladder cancer. However, this study did not request information on antibiotic therapy associated with UTIs; therefore we were unable to directly examine the potential positive effect of antibiotics. We were also able to exclude subjects reporting a history of diabetes, kidney stones, and bladder stones and therefore eliminate the possibility that the effect of bladder infections could be explained by these related diseases. However, confounding may have resulted from unmeasured socioeconomic, genetic, and other lifestyle factors that differentiate women who had infection from women who did not. As in any other epidemiological study, our findings may have been due to chance, although we believe this is not likely considering the consistency of our findings. For our large well-established population-based case-control study, we were able to match our cases and controls on age, sex, race and neighborhood of residence. Because this study had also available data describing consumption of fruits and vegetables, use of NSAIDs and use of hair dye, we were previously able to identify these factors as risk/protective factors for bladder cancer and subsequently control for them in our current analyses. 93 Our results suggest that a history of bladder infection is associated with a reduced risk of bladder cancer. Cytotoxicity against bladder cancer cells from antibiotics commonly used to treat bladder infections is proposed as a possible mechanism to explain this association. 94 3.2 HYPERTENSION, DIURETICS, AND ANTIHYPERTENSIVES 3.2.1 Introduction The intake of fluid and initiation of urination are affected by several medical conditions and medications. Diuretics are a group of drugs that block normal re- absorption along the nephron, inducing bodily urine excretion. For decades, physician have been using diuretics to treat hypertension, heart failure, kidney stones and other kidney diseases (Shah et al., 2004a, b; Krakoff, 2005). By increasing sodium and water loss throught the urine, diuretics can correct the unfavorable alterations in body fluid homeostasis/edema related to those diseases. In some studies, antihypertensives, especially diuretics, have been associated with increased risk of cancer, including kidney cancer (Felmeden & Lip, 2001; Grossman et al., 2002b). Diuretics, in particular low dose thiazide and thiazide-like diuretics, are in the forefront of therapy for hypertensive patients (Reyes, 2002; Shah et al., 2004a). As antihypertensive therapy, they can be used alone or in combination with other antihypetensive agents. Given that both anti-hypertensive drugs and diuretics are extremely heavily prescribed medications, it is important to establish if these drugs and/or the conditions that call for their use (hypertension and/or obesity) are related to the risk of bladder cancer, dependently or independently. 95 3.2.2 Materials and Methods Exposure definitions Refer to Chapter 3.1.2. We explicitly listed 32 brand names of diuretics and antihypertensive drugs in the questionnaire, drugs representing all the common prescription medications in these respective categories marketed in the US since the 1950s (Yuan et al., 1998a). Diuretics included are aldactazide, aldactone, diuril, dyazide, enduron, esidrix, furosemide, hydrochlorothiazide, hydrodiuril, hygroton, lasix, metahydrin, oretic, and zaroxolyn. Antihypertensives included are aldomet, apresoline, capoten, catapres, corgrad, hydralazine, inderal, lopressor, nimipress, procardia, rau-sed, reserpine, serpasil and tenormin. Adoril, esimil, hydropres and serapes are considered as diuretic/antihypertensive combination drugs. We first asked the subject whether they had ever taken the drug two or more times over their lifetime. If the answer was no, the subject was defined as a ‘non-user’. Otherwise, the subject was further asked if they had ever taken the drug two or more times a week for one month or longer. If the answer was no, the subject was classified as an ‘irregular user’. Otherwise, the subject was defined as a ‘regular user’ and was further asked about the ages at first and last use, duration of use, usual frequency and dosage of use, and the primary reason for such use. Aside from the 21 brand-name analgesics listed, the subject was asked if they had taken any other drug regularly. If the answer was yes, the names of the drugs were recorded, and ages at first and last use, duration of use, usual frequency and dosage of use, and the primary reason for use were similarly asked. 96 The formulations of each of the listed drugs as well as those volunteered by study subjects were established through numerous pharmaceutical sources, including the annually updated Physician’s Desk Reference. Each class of drugs was then placed into major formulation categories. For example, diuretics were grouped as thiazides, furosemides or potassium-sparing diuretics. Antihypertensives were classified as beta blockers, central anti-adrenergic agents, neuronal depleting agents, angiotensin- converting enzyme inhibitors or vasodilators. Age-specific exposure to a given drug was estimated from the subject’s reported dose and duration of use at that age. Lifetime cumulative exposure to a specific class of compounds (in grams) was computed by summing age-specific exposures across all brand name drugs belonging to that class of compounds. Cumulative exposures were grouped into tertiles according to their distributions among control subjects. Statistical Analysis Data were analyzed by standard matched-pair methods (Breslow and Day, 1980). The associations of bladder cancer with medical history, diuretics use, and antihypertensive drug use, were measured by ORs and their corresponding 95% CIs and P-values. The effects of medical history reported by2% of participants were examined. Conditional logistic regression models were used to examine the relationship between any medical history, any medication use and bladder cancer risk with adjustment for other risk factors for bladder cancer: average number of cigarettes smoked per day, number of years of smoking, smoking status in reference year (smoker or nonsmoker) (Castelao et al., 2001), level of education (high school or less, some college, college or above), lifetime use of NSAIDs (non/irregular 97 user, < 1441 pills,1441 pills over lifetime) (Castelao et al., 2000), intake of carotenoids (quintiles) (Castelao et al., 2004), and duration of employment as a hairdresser/barber (years) (Gago-Dominguez et al., 2001). Pairs in which either the case or the control failed to answer the relevant questions were eliminated from the corresponding analysis. ORs with 2-sided P-values less than 0.05 were considered statistically significant. All P-values quoted are 2-sided. 3.2.3 Results Fifteen percent of our subjects had regularly taken diuretics and 16% had regularly taken anti-hypertensive drugs (Table 24). In total, 22% of the subjects had used either diuretics or anti-hypertensive drugs regularly. Use of diuretics was not statistically significantly associated with bladder cancer in our study. Regular use of antihypertensive drugs seemed to increase the risk of bladder cancer by 20% (OR, 1.20; 95% CI, 0.96-1.48), but the association was not statistically significant. A history of hypertension was not significantly associated with risk of bladder cancer (p=0.33; Table 25). However, there was a marginally significant difference in the risk of bladder cancer between medically treated and untreated hypertensive patients (p =0.065). Untreated hypertensive patients had 0.64 (95% CI, 0.47-0.87) times the risk compared to subjects who were not hypertensive and did not take diuretics and/or antihypertensive drugs (Table 25). There was no difference in risk of bladder cancer among treated patients with or without a history of hypertension. 98 The inverse association between hypertension and bladder cancer was observed only among men (Table 26). Untreated hypertensive men exhibited a 42% (OR, 0.58; 95% CI, 0.42-0.80) lower risk of bladder cancer compared to normotensive men. However, no significant difference in risk was observed between hypertensive and normotensive women. We also examined the hypertension-bladder cancer association by number of years between diagnosis of hypertension and diagnosis of bladder cancer. No significant differences were observed. A reduced risk was observed among untreated hypertensive subjects, regardless of whether hypertension was diagnosed within five years of cancer diagnosis or more than 15 years before cancer diagnosis. As the most important established non-occupational source of bladder carcinogens, cigarette smoking was examined as a potential effect modifier of the hypertension-bladder cancer association in men (Table 27). The protection from bladder cancer associated with untreated hypertension seemed to be stronger among smokers. Among male smokers, untreated hypertension was associated with 0.54 (95% CI, 0.38-0.75) times the risk of bladder cancer compared to normotension. However, the corresponding figure was 0.82 (95% CI, 0.43-1.57) among male nonsmokers. Age, alcohol intake, and urination frequency during daytime did not modify the hypertension-bladder cancer association. However, BMI seemed to modify the association between untreated hypertension and bladder cancer (Table 28). Among obese men (BMI>30), untreated hypertension was not associated with bladder cancer. 99 Table 24. Diuretics, antihypertensive drugs, and hypertension in relation to bladder cancer risk Ca/Co OR 1 (95% CI) Diuretics No 1354/1357 1.00 yes 232/229 1.03 (0.82-1.29) Antihypertensives No 1326/1346 1.00 yes 260/240 1.20 (0.96-1.48) Diuretics/antihypertensives No 1225/1241 1.00 yes 361/345 1.09 (0.91-1.33) Hypertension No 1184/1157 1.00 Yes 401/429 0.91 (0.76-1.10) 1 Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 100 Table 25. Diuretics, antihypertensives, & hypertension in relation to the risk of bladder cancer No hypertension Hypertension Medications Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Diuretics No 1153/1120 1.00 (reference) 200/237 0.81 (0.64-1.02) yes 31/37 0.79 (0.45-1.38) 201/192 1.03 (0.81-1.32) Antihypertensive drugs No 1148/1125 1.00 (reference) 177/221 0.74 (0.58-0.95) yes 36/32 1.31 (0.76-2.26) 224/208 1.12 (0.89-1.41) Diuretics/antihypertensive drugs No 1120/1091 1.00 (reference) 104/150 0.64 (0.47-0.87) yes 64/66 1.02 (0.68-1.52) 297/279 1.06 (0.86-1.30) 1 Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 101 Table 26. Hypertension, diuretics/antihypertensives and risk of bladder cancer by sex 1: Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 2 Additionally adjusted for infections at the other two sites. All Men Women Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Normotensive 1184/1157 1.00 (reference) 914/881 1.00 (reference) 270/276 1.00 (reference) Treated hypertensive 297/279 1.05 (0.86-1.29) 231/220 1.00 (0.79-1.26) 66/59 1.40 (0.87-2.26) Untreated Hypertensive 104/150 0.64 (0.47-0.87) 91/136 0.58 (0.42-0.80) 13/14 1.08 (0.43-2.73) Duration of hypertension Treated hypertensive <5 34/27 1.02 (0.58-1.81) 26/21 0.96 (0.50-1.84) 8/6 1.15 (0.30-4.43) 5-14 142/138 0.97 (0.74-1.29) 110/110 0.89 (0.65-1.23) 32/28 1.72 (0.88-3.36) 15+ 114/109 1.13 (0.83-1.54) 92/85 1.16 (0.82-1.65) 22/24 1.07 (0.52-2.21) Untreated Hypertensive <5 14/25 0.45 (0.22-0.93) 12/22 0.44 (0.20-0.95) 2/3 0.52 (0.06-4.34) 5-14 48/60 0.83 (0.53-1.30) 42/57 0.71 (0.44-1.14) 6/3 3.00 (0.58-14.43) 15+ 42/60 0.60 (0.37-0.97) 37/53 0.57 (0.34-0.96) 5/7 0.61 (0.15-2.45) 102 Table 27.Hypertension, diuretics/antihypertensives and risk of bladder cancer by cigarette smoking in men 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, and number of years as hairdresser. 2 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. Nonsmokers Smokers Ca/Co OR 2 (95% CI) Ca/Co OR 2 (95% CI) Normotensive 174/313 1.00 740/568 1.00 Treated hypertensive 22/66 0.57 (0.33-0.98) 209/154 1.19 (0.92-1.54) Untreated hypertensive 15/35 0.82 (0.43-1.57) 76/101 0.54 (0.38-0.75) 103 Table 28. Hypertension, diuretics/antihypertensives and risk of bladder cancer by BMI in men BMI: <25 BMI: 25-<30 BMI:30 Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Normotensive 431/427 1.00 (reference) 408/379 1.00 (reference) 75/75 1.00 (reference) Treated hypertensive 75/69 1.05 (0.72-1.56) 107/106 0.92 (0.66-1.28) 49/45 1.19 (0.66-2.14) Untreated hypertensive 28/44 0.62 (0.36-1.06) 45/75 0.50 (0.33-0.77) 18/17 1.01 (0.45-2.29) 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 104 Table 29. Hypertension, diuretics/antihypertensives and risk of bladder cancer by GSTM1 genotype in men 1 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, and number of years as hairdresser. 2 Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. GSTM1: non-null GSTM1: null Ca/Co OR 2 (95% CI) Ca/Co OR 2 (95% CI) Normotensive 169/232 1.00 254/201 1.00 Treated hypertensive 43/58 0.96 (0.58-1.58) 52/45 1.04 (0.64-1.70) Untreated hypertensive 17/24 0.99 (0.48-2.05) 18/39 0.27 (0.14-0.51) 105 The OR (95% CI) associated with untreated hypertension was 0.62 (0.36-1.06) among normal-weight men and 0.50 (0.33-0.77) among overweight men. We also examined the hypertension-bladder cancer association by variations in genes involved in bladder carcinogenesis (Table 29). The hypertension-bladder cancer association was only observed among men with the GSTM1 null genotype. Among these subjects, untreated hypertension was associated with a lower risk of bladder cancer (OR, 0.27; 95% CI, 0.14-0.51). No modification of effect of hypertension was observed for any other genotype or phenotype studied. 3.2.4 Discussion Our study found a reduced risk of bladder cancer among hypertensive subjects who were not treated by diuretics or antihypertensives. This reduction in risk was mostly confined to men with the GSTM1 null genotype. Hypertension has been found to be associated with an increased risk of cancer incidence and cancer mortality (Grossman et al., 2002a). It is not known if hypertension itself or its related antihypertensive treatments are responsible for the increased risk. Several antihypertensive drugs have been associated with increased risk of cancer (Grossman et al., 2002a). An increased incidence of cancer among hypertensive patients could also be due to detection bias, since patients treated for hypertension are under closer medical supervision than are healthy subjects leading to an increased probability of detecting cancer. A possible link between carcinogenesis and the pathway leading to proliferative abnormalities in vascular 106 smooth muscle cells associated with hypertension has also been suggested (Lindgren et al., 2003). The possible biological mechanism that could explain the association between untreated hypertension and reduced risk of bladder cancer remains unclear. Since the reduction in bladder cancer risk is limited to subjects who were not treated by diuretics or antihypertensives, treatments for hypertension are not likely to explain this inverse association. Our observed association between untreated hypertension and bladder cancer did not differ by duration of hypertension. However, the fact that the association was not restricted to hypertension diagnosed closer to cancer diagnosis supports the view that hypertension is not a consequence of cancer. The association between hypertension and bladder cancer might be confounded by other risk factors that affect both diseases, such as cigarette smoking, diabetes, obesity, and alcohol consumption. However, the hypertension-bladder cancer association remained unchanged after we stratified our analyses by these potential confounding factors. In most studies, the hypertension-associated increase in cancer risk was higher among men than women (Grossman et al., 2002a). In our study, the untreated hypertension related decrease in bladder cancer risk was limited to men. Given the small sample size among women, it is possible that this difference by sex can be attributed to chance. However, this sex-difference could also be due to differences in environmental and dietary exposures not yet identified or innate sexual characteristics such as hormonal factors. 107 Genetic variance in GSTM1 seemed to play a role in the hypertension- bladder cancer association. GSTM1 is a member of phase II detoxification enzymes responsible for metabolizing electrophilic substrates through conjugation with GSH, therefore plays an important role in protecting DNA from damage (Hayes & Pulford, 1995; Yu et al., 1995). Loss of the GSTM1 gene results in absence of its enzyme activity, which has been found to be involved in the pathology of hypertension (Marinho et al., 2007). However, it is still unclear why the hypertension-bladder cancer association only exists among subjects with the GSTM1 null genotype. Our study has its own strengths and limitations. We did not collect data on levels of blood pressure and hence were unable to measure the effect of hypertension quantitatively. It is unknown why some hypertensive patients were not medically treated in our study. Other factors that affect the decision of treatment selection for hypertensive patients might explain the association between untreated hypertension and bladder cancer risk. It is unlikely that our observation could be due to chance alone, considering the consistency of our findings among different subsets. 108 3.3 OTHER MEDICAL CONDITIONS 3.3.1 Introduction Some diseases, such as diabetes mellitus, can alter the quantity and consistency of the urine (Lifford et al., 2005). Several studies have found an association between diabetes and bladder cancer risk (Tripathi et al., 2002b; Ng et al., 2003). Case-control studies have also shown that conditions associated with urinary stasis, such as bladder stones, nephrolithiasis, and prostate diseases, increase the risk of bladder cancer (Braver et al., 1987; Kjaer et al., 1989; Gonzalez et al., 1991). Patients diagnosed with conditions like gout are usually recommended by their doctors to increase their fluid intake for prevention purposes. 3.3.2 Materials and Methods Refer to Chapter 3.1.2. 3.3.3 Results The conditions more frequently reported by our study subjects (by at least two percent of them) were diabetes, angina, heart attack, kidney/renal stone, thyroid disease, gout, as well as UTIs and hypertension. Six percent of cases (93 out of 1,586) and four percent of controls (62 out of 1,586) reported having diabetes (Table 30). Diabetic patients had 1.7 (95% CI, 1.18- 2.45) times the risk of bladder cancer than non-diabetics, and the increase in risk from diabetes was similar between men and women. Eighty four percent of diabetic 109 patients were diagnosed at least five years before cancer diagnosis, and only 15 cases and six controls were diagnosed within five years of cancer diagnosis (Table 31). The diabetes-bladder cancer association remained unchanged when diabetic patients diagnosed within five years of cancer were excluded. The diabetes-bladder cancer association was modified by age. The OR (95% CI) associated with diabetes was 2.25 (1.42-3.55) among subjects of at least the median age (58 years) and 0.83 (0.43- 1.59) among younger subjects. The diabetes-bladder cancer association may be confounded by factors such as obesity, hypertension, and cigarette smoking. Therefore, we further examined the association between diabetes and bladder cancer by these factors (Table 31). The diabetes-bladder cancer association was not significantly affected by BMI (<25 vs. 25), cigarette smoking (lifetime nonsmokers vs. smokers), or blood pressure (normotensive vs. hypertension), even though slightly nonsignificant higher risks were observed among older subjects who had lower BMI, never smoked, and reported having hypertension. Six percent of cases (93 out of 1,586) and four percent of controls (68 out of 1,586) reported having had a heart attack, and a very similar number of subjects reported having had an angina (Table 32). Both heart attack and angina were found to be associated with an increased risk of bladder cancer in our study, especially among men. Among them, OR (95% CI) associated with a history of heart attack was 1.72 (1.65-2.55) and the corresponding figure for angina was 1.44 (0.96-2.18). The increased risk of bladder cancer in men was observed regardless of whether 110 Table 30. Selected medical conditions and risk of bladder cancer by sex. All Men Women Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Diabetes No 1493/1523 1.00 1162/1187 1.00 331/336 1.00 Yes 93/62 1.70 (1.18-2.45) 75/49 1.71 (1.14-2.56) 18/13 2.07 (0.88-4.91) Angina No 1494/1518 1.00 1155/1178 1.00 339/340 1.00 Yes 92/68 1.35 (0.92-1.99) 82/59 1.44 (0.96-2.18) 10/9 0.96 (0.29-3.20) Heart attack No 1493/1518 1.00 1150/1175 1.00 343/343 1.00 Yes 93/68 1.52 (1.05-2.22) 87/62 1.72 (1.16-2.55) 6//6 1.00 (0.28-3.65) Kidney/renal stones No 1432/1450 1.00 1105/1115 1.00 327/335 1.00 Yes 151/133 1.24 (0.94-1.64) 129/120 1.20 (0.89-1.62) 22/13 1.91 (0.80-4.58) Thyroid diseases No 902/893 1.00 717/717 1.00 185/176 1.00 Yes 64/75 0.86 (0.57-1.29) 32/32 0.88 (0.49-1.58) 32/43 0.71 (0.38-1.32) Gout No 897/918 1.00 692/702 1.00 205/216 1.00 Yes 66/55 1.53 (0.99-2.36) 58/51 1.35 (0.86-2.11) 8/4 12.08 (1.27-114.89) 1 Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 111 Table 31. Diabetes and bladder cancer risk All Age<median (58 years) Agemedian Diabetes Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) No 1493/1523 1.00 744/743 1.00 749/780 1.00 Yes 93/62 1.70 (1.18-2.45) 24/24 0.83 (0.43-1.59) 69/38 2.25 (1.42-3.55) Duration between diagnosis of diabetes & diagnosis of cancer(years) <5 15/6 1.83 (0.66-5.08) 7/3 1.56 (0.35-7.06) 8/3 1.70 (0.42-6.90) 5 78/56 1.68 (1.14-2.48) 17/21 0.72 (0.35-1.48) 61/35 2.31 (1.42-3.76) BMI<25 2 No 768/773 1.00 373/372 1.00 395/401 1.00 Yes 31/14 2.57 (1.29-5.14) 7/3 2.21 (0.52-9.48) 24/11 2.67 (1.21-5.90) BMI25 2 No 725/750 1.00 371/371 1.00 354/379 1.00 Yes 62/48 1.23 (0.82-1.87) 17/21 0.70 (0.35-1.41) 45/27 1.64 (0.97-2.78) Nonsmokers 2 No 269/553 1.00 155/300 1.00 114/253 1.00 Yes 17/20 2.02 (1.02-4.02) 3/8 0.85 (0.22-3.35) 14/12 2.79 (1.20-6.55) Smokers 2 No 1224/970 1.00 589/443 1.00 635/527 1.00 Yes 76/42 1.42 (0.95-2.12) 21/16 0.94 (0.47-1.89) 55/26 1.71 (1.03-2.84) Normotensive 2 No 1143/1126 1.00 598/594 1.00 545/532 1.00 Yes 41/30 1.33 (0.80-2.21) 13/12 1.04 (0.44-2.45) 28/18 1.49 (0.77-2.85) Hypertensive 2 No 349/397 1.00 14149 1.00 203/248 1.00 Yes 52/32 1.80 (1.10-2.96) 11/12 0.75 (0.30-1.87) 41/20 2.51 (1.38-4.60) 1 Conditional logistic regression. 2 Unconditional logistic regression, additionally adjusted for age, sex, and race. 112 Table 32. Angina, heart attack and bladder cancer risk by age in men All Age<median Agemedian Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Angina No 1155/1178 1.00 562/576 1.00 593/602 1.00 Yes 82/59 1.44 (0.96-2.18) 26/12 1.98 (0.88-4.44) 56/47 1.20 (0.74-1.96) Duration between angina & diagnosis of cancer(years) <5 19/11 1.74 (0.74-4.07) 10/4 1.55 (0.42-5.76) 9/7 1.89 (0.61-5.83) 5 62/47 1.39 (0.88-2.19) 16/8 2.23 (0.84-5.95) 46/39 1.12 (0.66-1.88) Heart attack No 1150/1175 1.00 553/571 1.00 597/604 1.00 Yes 87/62 1.72 (1.16-2.55) 35/17 2.04 (1.03-4.08) 52/45 1.53 (0.93-2.50) Duration between heart attack & diagnosis of cancer(years) <5 23/12 2.40 (1.08-5.34) 10/7 1.11 (0.38-3.25) 13/5 6.21 (1.80-21.41) 5 61/47 1.60 (1.02-2.51) 24/10 2.73 (1.16-6.43) 37/37 1.19 (0.69-2.05) 1 Conditional logistic regression, adjusted for level of education, use of NSAIDs, carotenoids intake, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 113 angina or heart attack occurred within five years cancer. Age seemed to modify the association between heart attack, angina, and bladder cancer. The OR (95% CI) associated with recent (within five years of cancer diagnosis) heart attacks was 1.11 (0.38-3.25) among younger men and 6.21 (1.80-21.41) among older men. The corresponding figures associated with older (at least five years prior to diagnosis of cancer) heart attacks were 2.73 (1.16-6.43) and 1.19 (0.69-2.05). A similar age pattern was also found for angina. A history of kidney/renal stone or thyroid diseases was not associated with bladder cancer risk (Table 30). A history of gout was associated with an increased risk of bladder cancer (OR, 1.53; 95% CI, 0.99-2.36), especially among women, although the numbers were small. 3.3.4 Discussion In our study, diabetes was found to increase the risk of bladder cancer, especially among older subjects, and a history of both angina and heart attack were also found to increase the risk of bladder cancer, but mostly among younger subjects. Diabetes is one of the major causes of morbidity and mortality in the US (Mokdad et al., 2003). Diabetics have been found to have increased risks of cancers of the pancreas (Everhart & Wright, 1995), breast (Wolf et al., 2005), endometrium (Anderson et al., 2001), liver (El-Serag et al., 2006), kidney (Washio et al., 2007), colon and rectum (Larsson et al., 2005). A meta-analysis of diabetes and bladder cancer was conducted recently and suggested 1.24 (95% CI, 1.08-1.42) times the risk 114 of bladder cancer among diabetics compared with non-diabetics (Larsson et al., 2006). This meta-analysis included seven case-control studies, three cohort studies and six cohort studies of diabetic patients conducted from 1966 to 2006. Diabetes was associated with an increased risk of bladder cancer in case-control studies and cohort studies, but not in cohort studies of diabetic patients. It is not clear how diabetes leads to bladder cancer. In diabetic patients, insulin resistance leads to hyperinsulinaemia, which can increase IGF-1. IGF-1 has been found to stimulate cell proliferation and inhibit apoptosis and has been implicated in the carcinogenesis of prostate, breast, colon and rectum (Larsson et al., 2006). A higher plasma concentration of IGF-1 has been observed in bladder cancer patients than controls (Zhao et al., 2003b). Obesity, cigarette smoking, and hypertension could be potential confounders of the diabetes-bladder cancer association. Cigarette smoking (Eliasson, 2003)and hypertension (Jandeleit-Dahm & Cooper, 2006), often associated with diabetes (Eliasson, 2003; Jandeleit-Dahm & Cooper, 2006), have been found to increase the risk of bladder cancer in our study. Findings on the relationship between obesity and bladder cancer risk are limited and inconsistent (Larsson et al., 2006), but obesity is an important risk factor for diabetes (Kahn et al., 2006). Therefore, we stratified our analysis by these three potential confounding factors, but did not observe any significant difference in the diabetes-bladder cancer association according to these potential confounders. Thus, it is unlikely that obesity, cigarette smoking, or 115 hypertension could explain the increased risk of bladder cancer found among diabetic patients. A significant increase in oxidative stress has been observed among subjects with unstable angina or acute myocardial infarction (Dubois-Rande et al., 1994), and both experimental studies and epidemiological studies have supported a role of reactive oxygen species in bladder carcinogenesis (Hung et al., 2004). Consistent with prior studies, we observed an increased risk from a history of both angina and heart attack. However, it is possible that treatments for these diseases rather than the diseases themselves may be responsible for the observed associations. It is also possible that the association between a history of angina and heart attack and risk of bladder cancer could be due, at least in part, to detection bias. Among older subjects, an increased risk was observed only if angina or heart attack occurred within five years of cancer diagnosis. Thus, it is possible that the diagnosis of angina or heart attack have lead to more medical examinations which helped detecting bladder cancer earlier. Our study has its own strengths and limitations. We did not collect information of the type of diabetes; therefore we were unable to distinguish the effect of Type I, Type II, and gestational diabetes. However, only 15 diabetic patients (7 cases and eight controls) reported being diagnosed before age 30 and most were male, therefore we believe that the majority of the reported diabetes was probably Type II. Recall bias might have played a role in our observed association. However, other investigators have reported a high concordance between self reported diabetes and 116 medical record review (Midthjell et al., 1992). If recall bias played a role in the association between diabetes and bladder cancer, one would expect a similar effect among younger and older subjects. Diabetic patients may have also had more medical examinations leading to early diagnosis resulting in detection bias. As a means to explore this issue, we examined the diabetes-bladder cancer association separately for superficial and invasive bladder cancers. Similar associations were observed for superficial (OR, 1.61; 95% CI, 1.08-2.41) and for invasive bladder cancer (OR, 2.14; 95% CI, 0.87-5.26), therefore detection bias is unlikely the explanation for our observed association. 117 CHAPTER 4 MICTURITION AND BLADDER CANCER RISK 4.1 INTRODUCTION Frequency of urination is directly related to the intensity and duration of urothelium distension. While most of studies have been simply focused on the total or individual fluid intake, frequency of urination was rarely considered. It has been proposed that measuring total fluid intake and taking into account the daily number of urinary voiding events is more reliable, since total fluid intake depends on body size, diet, physical activity, health status, and environments (Radosavljevic, 2004). Only a few studies have investigated the role of frequency of urination in the etiology of bladder cancer. Healthy controls were found to have more frequent urination compared to cancer patients (Braver et al., 1987; Radosavljevic et al., 2003). Experimental data from dogs given 4-ABP indicated that exposure to 4-ABP and subsequent ABP-DNA adduct formation in the bladder are directly dependent on voiding frequency (Kadlubar et al., 1991). It has been shown that urination frequency is an important factor contributing to interindividual differences in DNA-binding of ABP in human bladder (Bois et al., 1995). Urination, also known as micturition, is usually under voluntary control. It is initiated by the distension of the bladder wall when the micturition reflex is activated (Andersson, 2002). However, other mechanisms also contribute to the initiation of the micturition reflex, both in normal and dysfunctional bladders. Frequent urination can result from large volume of fluid intake, radiation therapy, and medications such as diuretics, and it is also a common symptom of conditions such as infections, 118 bladder maglignancy, benign prostatic hyperplasia (Brunicardi et al., 2004), diabetes, bladder dysfunction, and pregnancy. Consumption of alcoholic beverages or those containing caffeine may irritate the lining of the bladder, initiating the need to urinate. Nocturia, defined as waking at night to void on one or more ooccasions, can occur at any age and affect both men and women equally (Lundgren, 2004). It is particularly common in the elderly, affecting up to 70% of people older than 65 years (Lundgren, 2004). Nocturia has a multifactorial etiology, i.e., it can occur as a symptom of another underlying disease or as a condition in its own right. 4.2 MATERIALS AND METHODS Exposure Definitions Refer to Chapter 2.1.2. After January, 1992, we requested information on urinary frequency. Each subject was asked how many times they usually urinated during their waking hours and during the night separately as an adult. Statistical Analysis Refer to Chapter 2.1.2. Spearman correlation coefficient among control subjects was used to measure the relationship between urination frequency and sex, age, level of education, intake of carotenoids, use of NSAIDs, and cigarette smoking, water intake, alcohol intake, coffee intake, bladder infection, use of diuretics/antihypertensives, history of diabetes, angina, and heart attack. Total pack- years of cigarette smoking over the lifetime were used to measure cigarette smoking. Continuous variables were used for all the variables except sex. Male gender was 119 coded as one and female as two. Positive history of bladder infection, diabetes, angina, and heart attack history was coded as one and negative history as 0. Users of diuretics or antihypertensives were coded as one and never users as 0. 4.3 RESULTS A total of 973 pairs of cases and controls provided information on urination frequency in our study. Overall, subjects who urinated more frequently during daytime also urinated more frequently during the night (Table 33). Women and younger subjects urinated more frequently during daytime. Use of NSAIDs, bladder infection, and consumption of fluids such as water, alcohol, and coffee increased urination frequency during both daytime and nighttime. Hypertension and diabetes increased urination frequency during nighttime but not daytime. Use of diuretics or antihypertensive drugs had no significant impact on daytime or nighttime urination. Cigarette smoking, level of education, intake of carotenoids, angina, and heart attack were not associated with urination during the day or night. Eight percent, 80%, and 12% of our control subjects reported urinating 1-2, 3-6, and 6+ times during daytime, respectively (Table 34). The mean number of daytime urination among controls was 4.6. Neither daytime nor nighttime urination was associated with bladder cancer risk, with and without additional adjustment for other potential risk/protective factors. However, compared to subjects who urinated 1-2 times per day, subjects who urinated 3-6 times per day had a marginally significantly lower risk of bladder cancer. The OR (95% CI) comparing urinating 3-6 120 times per day to urinating 1-2 times per day was 0.73 (0.50-1.05) among all subjects, and it was similar between men and women (Table 35). Majority of controls (66%) did not urinate during the night, while 24% controls urinated once every night, and 11% urinated more than once (Table 34). Nocturia was associated with a significantly increased risk of bladder cancer among women (p trend=0.032), but not among men. Compared to no voiding at night, voiding more than once at night was associated with 2.27 (95% CI, 1.00-5.15) times the risk of bladder cancer among women and 0.88 (95% CI, 0.59-1.32) times the risk among men (Table 35). Use of diuretics or antihypertensives seemed to modify the relationship between urination frequency and bladder cancer (Table 36). Among subjects who used diuretics or antihypertensives, increased risk of bladder cancer was observed with increasing frequency of daytime (p trend=0.079) and nighttime urination (p trend=0.025). The increased risk was observed both men and women, but was more pronounced in the latter. No association was found for nonusers of diuretics or antihypertensives. There were no discernible differences in the association between frequency of urination and bladder cancer by any other factor examined. 121 Table 33. Spearman correlation coefficients between frequency of urination and other risk/protective factors among controls Daytime urination Nighttime urination Daytime urination / 0.229 1 Nighttime urination 0.229 1 / Sex 2 0.121 1 0.022 Age 2 -0.138 1 0.044 Education 2 0.012 -0.026 Carotenoids 2 0.023 0.025 NSAIDs 2 0.080 1 0.133 1 Cigarette smoking 2 0.006 -0.008 Water 2 0.211 1 0.102 1 Alcohol 2 0.125 1 0.102 1 Coffee 2 0.101 1 -0.020 Bladder Infection 2 0.064 0.061 Diuretics or antihypertensives 2 0.024 0.038 Hypertension 2 -0.023 0.082 1 Diabetes 2 -0.020 0.065 1 Angina 2 -0.017 -0.004 Heart attack 2 -0.020 -0.004 1 Statistically significant with p<0.05; 2 Spearman partial correlation coefficients, adjusted for age and sex. 122 Table 34. Frequency of urination and bladder cancer (times/day) Ca/Co OR 1 (95% CI) OR 2 (95% CI) Daytime urination 1-2 91/81 1.00 1.00 3-6 682/762 0.81 (0.59-1.12) 0.73 (0.50-1.05) >6 150/115 1.14 (0.78-1.68) 0.92 (0.59-1.44) Trend 0.25 0.95 Nighttime urination 0 627/626 1.00 1.00 1 225/234 0.95 (0.77-1.18) 1.08 (0.84-1.38) >1 101/100 0.99 (0.73-1.34) 1.05 (0.74-1.49) Trend 0.78 0.61 1: Conditional logistic regression. 2: Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 123 Table 35. Frequency of urination and risk of bladder cancer by sex Men Women (times/day) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Daytime urination 1-2 79/69 1.00 12/12 1.00 3-6 537/592 0.73 (0.50-1.09) 145/170 0.65 (0.22-1.89) >6 100/80 0.82 (0.50-1.35) 50/35 1.31 (0.40-4.31) Trend 0.51 0.12 Nighttime urination 0 501/487 1.00 126/139 1.00 1 168/178 0.99 (0.74-1.32) 57/56 1.49 (0.83-2.67) >1 72/76 0.88 (0.59-1.32) 29/24 2.27 (1.00-5.15) Trend 0.60 0.032 1: Conditional logistic regression, adjusted for level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 124 Table 36. Frequency of urination and risk of bladder cancer by use of diuretics and antihypertensives Diuretics or anti-hypertensive Nonusers Users (times/day) Ca/Co OR 1 (95% CI) Ca/Co OR 1 (95% CI) Daytime urination 1-2 76/60 1.00 15/21 1.00 3-6 512/600 0.66 (0.45-0.96) 170/162 1.50 (0.69-3.27) >6 113/90 0.89 (0.56-1.44) 37/25 2.30 (0.89-5.97) Trend 1.00 0.079 Nighttime urination 0 504/497 1.00 123/129 1.00 1 150/176 0.98 (0.75-1.29) 75/58 1.72 (1.07-2.77) >1 63/75 0.86 (0.58-1.27) 38/25 1.71 (0.91-3.23) Trend 0.51 0.025 1: Unconditional logistic regression, adjusted for age, sex, race, level of education, use of NSAIDs, intake of carotenoids, number of years as hairdresser, cigarette smoking status, duration of smoking, and intensity of smoking. 125 4.4 DISCUSSION Frequency of daytime urination seems to exert its effect on bladder cancer depending on the substrate. When enough substrate such as alcohol and water is present, frequency of urination seems to play a role modifying the potentially protective associations between water, alcohol, and bladder cancer. In our study, the inverse associations between water intake, alcohol intake, and bladder cancer risk were modified by frequency of daytime urination with the reduction in risk confined to those who urinated more frequently. However, frequency of urination seems not to play a role in the bladder infection-bladder cancer association. One limitation of our study is that we measured only frequency of urination, not volume of urination. Frequency is an indirect measure of urine flow and may not correlate with the actual amount of urine excreted. Also, frequency of urination can be affected by many other factors such as use of diuretics, diabetes, benign prostatic hyperplasia (BPH), etc, which are very common among elderly people. Thus, it is possible that the reported urination habit is more likely a reflection of recent experience rather than the whole adulthood. In this study, nocturia was associated with an increased risk of bladder cancer among women and among subjects who used diuretics/antihypertensives. The etiology of nocturia is multifactorial. It can occur if excessive amounts of fluid are consumed in the late evening (Tanagho & McAninch, 2003). In addition, nocturia can result from diseases of the peripheral urinary tract, diabetes, cardiac disorders, sleep-disorders and aging (Asplund, 2002; Tanagho & McAninch, 2003). In men, 126 nocturia has long been considered a common symptom of BPH, a condition which affects more than 50% of men aged over 60 years and 90% of men aged over 80 years (Thorpe & Neal, 2003). However, factors other than BPH may also play a role in nocturia among elderly men. It has been found that among BPH patients who were treated by transurethral resection of the prostate, one third of them still complained of an unchanged frequency of nocturia. In women, nocturia has been associated with significant bladder lesions (cystitits glandularis or malignancy) (Wu et al., 2006) and has been found to result from irritative bladder symptoms caused by bladder cancer (Lundgren, 2004; Wu et al., 2006). In our study, nocturia was also found to be associated with an increased risk of bladder cancer in women. Thus, nocturia reported by women is more likely to result from early symptoms caused by bladder cancer. It is not clear how use of diuretics or antihypertensive drugs modifies the association between frequency of urination and bladder cancer. Among subjects who use these drugs, frequency of daytime and nighttime urination is possibly an indication of the individual’s reaction to the drugs or the severity of the conditions that called for their use. Also, diuretics increase the frequency of urination through pharmacologic mechanisms, but this increase is not accompanied by a concomitant increase in intake of fluids. 127 BIBLIOGRAPHY Agriculture USDo. 1992. Nutrient database for standard reference, release 10. Bethesda, MD: Human Nutrition Information Service, USDA. Agriculture USDo. 1994. USDA nutrient database for individula food intake surveys and data sets, release 7. Springfield, VA: National Technical Informaion Service. Albanes D, Heinonen OP, Taylor PR et al. 1996. Alpha-Tocopherol and beta- carotene supplements and lung cancer incidence in the alpha-tocopherol, beta- carotene cancer prevention study: effects of base-line characteristics and study compliance. J Natl Cancer Inst 88:1560-1570. Alexandroff AB, Jackson AM, O'Donnell MA, James K. 1999. BCG immunotherapy of bladder cancer: 20 years on. Lancet 353:1689-1694. Allam MF. 2005. Fluid intake and urinary bladder cancer. European Journal of Cancer Prevention 14:77-78. Amling CL. 2001. Diagnosis and management of superficial bladder cancer. Curr Probl Cancer 25:219-278. Anderson KE, Anderson E, Mink PJ et al. 2001. Diabetes and endometrial cancer in the Iowa women's health study. Cancer Epidemiol Biomarkers Prev 10:611-616. Andersson KE. 2002. Overactive bladder--pharmacological aspects. Scandinavian Journal of Urology & Nephrology Supplementum 210:72-81. Aranha O, Wood DP, Jr., Sarkar FH. 2000. Ciprofloxacin mediated cell growth inhibition, S/G2-M cell cycle arrest, and apoptosis in a human transitional cell carcinoma of the bladder cell line. Clin Cancer Res 6:891-900. Aranha O, Zhu L, Alhasan S, Wood DP, Jr., Kuo TH, Sarkar FH. 2002. Role of mitochondria in ciprofloxacin induced apoptosis in bladder cancer cells. J Urol 167:1288-1294. Armstrong LE, Pumerantz AC, Roti MW et al. 2005. Fluid, electrolyte, and renal indices of hydration during 11 days of controlled caffeine consumption. Int J Sport Nutr Exerc Metab 15:252-265. Asal NR, Risser DR, Kadamani S, Geyer JR, Lee ET, Cherng N. 1988. Risk factors in renal cell carcinoma: I. Methodology, demographics, tobacco, beverage use, and obesity. Cancer Detect Prev 11:359-377. 128 Asplund R. 2002. Nocturia in relation to sleep, somatic diseases and medical treatment in the elderly. BJU Int 90:533-536. Badawi AF, Hirvonen A, Bell DA, Lang NP, Kadlubar FF. 1995. Role of aromatic amine acetyltransferases, NAT1 and NAT2, in carcinogen-DNA adduct formation in the human urinary bladder. Cancer Research 55:5230-5237. Baglietto L, Severi G, English DR, Hopper JL, Giles GG. 2006. Alcohol consumption and prostate cancer risk: results from the Melbourne collaborative cohort study. Int J Cancer 119:1501-1504. Bankir L, Bardoux P, Ahloulay M. 2001. Vasopressin and diabetes mellitus. Nephron 87:8-18. Bell DA, Stephens EA, Castranio T et al. 1995. Polyadenylation polymorphism in the acetyltransferase 1 gene (NAT1) increases risk of colorectal cancer. Cancer Research 55:3537-3542. Bernstein L, Ross RK. 1991. Cancer in Los Angeles County. Los Angeles, CA: University of Southern California. Bertram JS, Craig AW. 1972. Specific induction of bladder cancer in mice by butyl- (4-hydroxybutyl)-nitrosamine and the effects of hormonal modifications on the sex difference in response. Eur J Cancer 8:587-594. Beulens JWJ, Stolky RP, van der Schouw YT, Grobbee DE, Hendriks HFJ, Bots ML. 2005. Alcohol Consumption and Risk of Type 2 Diabetes Among Older Women. Diabetes Care 28:2933-2938. Boffetta P, Hashibe M. 2006. Alcohol and cancer. The Lancet Oncology 7:149-156. Bois FY, Krowech G, Zeise L. 1995. Modeling human interindividual variability in metabolism and risk: the example of 4-aminobiphenyl. Risk Analysis 15:205-213. Boobis AR, Lynch AM, Murray S et al. 1994. CYP1A2-catalyzed conversion of dietary heterocyclic amines to their proximate carcinogens is their major route of metabolism in humans. Cancer Res 54:89-94. Braver DJ, Modan M, Chetrit A, Lusky A, Braf Z. 1987. Drinking, micturition habits, and urine concentration as potential risk factors in urinary bladder cancer. Journal of the National Cancer Institute 78:437-440. Brownson RC, Chang JC, Davis JR. 1987. Occupation, smoking, and alcohol in the epidemiology of bladder cancer. Am J Public Health 77:1298-1300. 129 Bruemmer B, White E, Vaughan TL, Cheney CL. 1997. Fluid intake and the incidence of bladder cancer among middle-aged men and women in a three-county area of western Washington. Nutr Cancer 29:163-168. Bruger M. 1940. The effects of alcohol on the normal and pathologic kidney: A review. Q J Stud Alcohol 1:85-94. Brunicardi FC, Andersen DK, Billiar TR et al. 2004. Schwartz's Principles of Surgery: McGraw-Hill. Burin GJ, Gibb HJ, Hill RN. 1995. Human bladder cancer: evidence for a potential irritation-induced mechanism. Food Chem Toxicol 33:785-795. Busby JE, Kamat AM. 2006. Chemoprevention for bladder cancer. J Urol 176:1914- 1920. Butler MA, Iwasaki M, Guengerich FP, Kadlubar FF. 1989. Human cytochrome P- 450PA (P-450IA2), the phenacetin O-deethylase, is primarily responsible for the hepatic 3-demethylation of caffeine and N-oxidation of carcinogenic arylamines. Proc Natl Acad Sci U S A 86:7696-7700. Cantor KP, Lynch CF, Johnson D. 1992. Bladder cancer, parity, and age at first birth. Cancer Causes & Control 3:57-62. Carlson GP, Chaney AM, Jiang F. 1995. Comparison of the induction of rat glutathione S-transferase and fatty acid ethyl ester synthase activities. Toxicology 96:127-132. Castelao JE, Yuan JM, Gago-Dominguez M et al. 2004. Carotenoids/vitamin C and smoking-related bladder cancer. Int J Cancer 110:417-423. Castelao JE, Yuan JM, Gago-Dominguez M, Yu MC, Ross RK. 2000. Non-steroidal anti-inflammatory drugs and bladder cancer prevention. Br J Cancer 82:1364-1369. Castelao JE, Yuan JM, Skipper PL et al. 2001. Gender- and smoking-related bladder cancer risk. J Natl Cancer Inst 93:538-545. Cho CH, Purohit V, eds. 2006. Alcohol, tobacco and cancer. Basel, Switzerland: Karger. Chyou PH, Nomura AM, Stemmermann GN. 1993. A prospective study of diet, smoking, and lower urinary tract cancer. Ann Epidemiol 3:211-216. 130 Claude J, Kunze E, Frentzel-Beyme R, Paczkowski K, Schneider J, Schubert H. 1986. Life-style and occupational risk factors in cancer of the lower urinary tract. Am J Epidemiol 124:578-589. Claybaugh JR, Sato AK, Crosswhite LK, Hassell LH. 2000. Effects of time of day, gender, and menstrual cycle phase on the human response to a water load. Am J Physiol Regul Integr Comp Physiol 279:R966-973. Cordon-Cardo C, Cote RJ, Sauter G. 2000. Genetic and molecular markers of urothelial premalignancy and malignancy. Scandinavian Journal of Urology & Nephrology Supplementum 205:82-93. Cornelis MC, El-Sohemy A, Kabagambe EK, Campos H. 2006. Coffee, CYP1A2 genotype, and risk of myocardial infarction. Jama 295:1135-1141. Cote RJ, Datar RH. 2003. Therapeutic approaches to bladder cancer: identifying targets and mechanisms. Critical Reviews in Oncology/Hematology Targeted Therapies in the Treatment of Genitourinary Cancers 46:67-83. Coughlin SS, Calle EE, Teras LR, Petrelli J, Thun MJ. 2004. Diabetes mellitus as a predictor of cancer mortality in a large cohort of US adults. Am J Epidemiol 159:1160-1167. Cunha BA. 1988. Nitrofurantoin--current concepts. Urology 32:67-71. Davis CP, Cohen MS, Hackett RL, Anderson MD, Warren MM. 1991. Urothelial hyperplasia and neoplasia. III. Detection of nitrosamine production with different bacterial genera in chronic urinary tract infections of rats. J Urol 145:875-880. Djousse L, Schatzkin A, Chibnik LB, D'Agostino RB, Kreger BE, Ellison RC. 2004. Alcohol consumption and the risk of bladder cancer in the Framingham Heart Study. J Natl Cancer Inst 96:1397-1400. Dolin PJ, Darby SC, Beral V. 1994. Paraplegia and squamous cell carcinoma of the bladder in young women: findings from a case-control study. British Journal of Cancer 70:167-168. Donato F, Boffetta P, Fazioli R, Aulenti V, Gelatti U, Porru S. 1997. Bladder cancer, tobacco smoking, coffee and alcohol drinking in Brescia, northern Italy. Eur J Epidemiol 13:795-800. Dubois-Rande JL, Artigou JY, Darmon JY et al. 1994. Oxidative stress in patients with unstable angina. Eur Heart J 15:179-183. 131 Ebisuno S, Inagaki T, Kohjimoto Y, Ohkawa T. 1997. The cytotoxic effect of fleroxacin and ciprofloxacin on transitional cell carcinoma in vitro. Cancer 80:2263- 2267. Eggleton MG. 1941. The diuretic action of alcohol in man. J Physiol 101:172-191. Eliasson B. 2003. Cigarette smoking and diabetes. Prog Cardiovasc Dis 45:405-413. El-Serag HB, Hampel H, Javadi F. 2006. The association between diabetes and hepatocellular carcinoma: a systematic review of epidemiologic evidence. Clin Gastroenterol Hepatol 4:369-380. Epstein M. 1997. Alcohol's impact on kidney function. Alcohol Health Res World 21:84-92. Everhart J, Wright D. 1995. Diabetes mellitus as a risk factor for pancreatic cancer. A meta-analysis. Jama 273:1605-1609. Felmeden DC, Lip GY. 2001. Antihypertensive therapy and cancer risk. Drug Safety 24:727-739. Fihn SD. 2003. Clinical practice. Acute uncomplicated urinary tract infection in women. N Engl J Med 349:259-266. Foxman B, Gillespie B, Koopman J et al. 2000. Risk factors for second urinary tract infection among college women. Am J Epidemiol 151:1194-1205. Gago-Dominguez M, Bell DA, Watson MA et al. 2003. Permanent hair dyes and bladder cancer: risk modification by cytochrome P4501A2 and N-acetyltransferases 1 and 2. Carcinogenesis 24:483-489. Gago-Dominguez M, Castelao JE, Yuan JM, Yu MC, Ross RK. 2001. Use of permanent hair dyes and bladder-cancer risk. Int J Cancer 91:575-579. Gan J, Skipper PL, Gago-Dominguez M, Arakawa K, Ross RK, Yu MC, Tannenbaum SR. 2004. Alkylaniline-Hemoglobin Adducts and Risk of Non- Smoking-Related Bladder Cancer. J Natl Cancer Inst 96:1425-1431. Garcia-Closas M, Malats N, Silverman D et al. 2005. NAT2 slow acetylation, GSTM1 null genotype, and risk of bladder cancer: results from the Spanish Bladder Cancer Study and meta-analyses. The Lancet 366:649-659. Gauderman W, Morrison J. 2006. QUANTO 1.1: A computer program for power and sample size calculations for genetic-epidemiology studies, http://hydra.usc.edu /gxe. 132 Gerhauser C, Alt A, Heiss E et al. 2002. Cancer chemopreventive activity of Xanthohumol, a natural product derived from hop. Mol Cancer Ther 1:959-969. Glade MJ. 1999. Food, nutrition, and the prevention of cancer: a global perspective. American Institute for Cancer Research/World Cancer Research Fund, American Institute for Cancer Research, 1997. Nutrition 15:523-526. Gonzalez CA, Errezola M, Izarzugaza I, Lopez-Abente G, Escolar A, Nebot M, Riboli E. 1991. Urinary infection, renal lithiasis and bladder cancer in Spain. European Journal of Cancer 27:498-500. Gonzalez J, Munoz ME, Martin MI, Collado PS, Fermoso J, Esteller A. 1988. Influence of acute ethanol administration on hepatic glutathione metabolism in the rat. Alcohol 5:103-106. Gonzalez MA, Uribe F, Moisen SD, Fuster AP, Selen A, Welling PG, Painter B. 1984. Multiple-dose pharmacokinetics and safety of ciprofloxacin in normal volunteers. Antimicrob Agents Chemother 26:741-744. Green A, Beral V, Moser K. 1988. Mortality in women in relation to their childbearing history. BMJ 297:391-395. Grossman E, Messerli FH, Boyko V, Goldbourt U. 2002a. Is there an association between hypertension and cancer mortality? Am J Med 112:479-486. Grossman E, Messerli FH, Goldbourt U. 2002b. Carcinogenicity of antihypertensive therapy. Current Hypertension Reports 4:195-201. Gunzerath L, Faden V, Zakhari S, Warren K. 2004. National Institute on Alcohol Abuse and Alcoholism report on moderate drinking. Alcohol Clin Exp Res 28:829- 847. Havrda DE, Mai T, Chonlahan J. 2005. Enhanced antithrombotic effect of warfarin associated with low-dose alcohol consumption. Pharmacotherapy 25:303-307. Hayes JD, Pulford DJ. 1995. The glutathione S-transferase supergene family: regulation of GST and the contribution of the isoenzymes to cancer chemoprotection and drug resistance. Critical Reviews in Biochemistry & Molecular Biology 30:445- 600. Hetu C, Yelle L, Joly JG. 1982. Influence of ethanol on hepatic glutathione content and on the activity of glutathione S-transferases and epoxide hydrase in the rat. Drug Metab Dispos 10:246-250. 133 Hiramoto K, Kida T, Kikugawa K. 2002. Increased urinary hydrogen peroxide levels caused by coffee drinking. Biol Pharm Bull 25:1467-1471. Hoffken G, Lode H, Prinzing C, Borner K, Koeppe P. 1985. Pharmacokinetics of ciprofloxacin after oral and parenteral administration. Antimicrob Agents Chemother 27:375-379. Holroyd-Leduc JM, Straus SE. 2004. Management of Urinary Incontinence in Women: Scientific Review. JAMA 291:986-995. Howe GR, Burch JD, Miller AB et al. 1980. Tobacco use, occupation, coffee, various nutrients, and bladder cancer. J Natl Cancer Inst 64:701-713. Hu J, Mao Y, White K, Canadian Cancer Registries Epidemiology Research G. 2003. Diet and vitamin or mineral supplements and risk of renal cell carcinoma in Canada. Cancer Causes Control 14:705-714. Hung RJ, Boffetta P, Brennan P et al. 2004. Genetic polymorphisms of MPO, COMT, MnSOD, NQO1, interactions with environmental exposures and bladder cancer risk. Carcinogenesis 25:973-978. IARC. 1987a. Overall Evaluations of Carcinogenicity: An Updating of IARC Monographs. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 1. IARC. 1987b. Overall Evaluations of Carcinogenicity: An Updating of IARC Monographs. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans 4. IARC. 1988. Alcohol drinking. Biological data relevant to the evaluation of carcinogenic risk to humans. IARC Monogr Eval Carcinog Risks Hum 44:101-152. IARC. 1991. Coffee, Tea, Mate, Methylxanthines and Methyglyoxal. IARC Monographs: Evaluation of Carcinogenic Risks to Humans: International Agency for Research on Cancer, Lyon. IARC. 1994. Schistosomes, Liver Flukes and Helicobacter pylori. Monograph on the Evaluation of Carcinogenic Risks to Humans. Lyon, France: International Agency for Research on Cancer. Imhof A, Froehlich M, Brenner H, Boeing H, Pepys MB, Koenig W. 2001. Effect of alcohol consumption on systemic markers of inflammation. Lancet 357:763-767. Iscovich J, Castelletto R, Esteve J et al. 1987. Tobacco smoking, occupational exposure and bladder cancer in Argentina. Int J Cancer 40:734-740. 134 Jandeleit-Dahm K, Cooper ME. 2006. Hypertension and diabetes: role of the renin- angiotensin system. Endocrinol Metab Clin North Am 35:469-490, vii. Jemal A, Murray T, Ward E et al. 2005. Cancer Statistics, 2005. CA Cancer J Clin 55:10-30. Jemal A, Siegel R, Ward E, Murray T, Xu J, Thun MJ. 2007. Cancer statistics, 2007. CA Cancer J Clin 57:43-66. Jiang X, Castelao JE, Groshen S, Cortessis VK, Ross RK, Conti DV, Gago- Dominguez M. 2007a. Alcohol consumption and risk of bladder cancer in Los Angeles County. Int J Cancer 17:17. Jiang X, Yuan JM, Skipper PL, Tannenbaum SR, Yu MC. 2007b. Environmental tobaco smoke and bladder cancer risk in never smokers of Los Angeles County. Cancer Res In Press. Jones PA, Ross RK. 1999. Prevention of Bladder Cancer. N Engl J Med 340:1424- 1426. Kadlubar FF, Dooley KL, Teitel CH et al. 1991. Frequency of urination and its effects on metabolism, pharmacokinetics, blood hemoglobin adduct formation, and liver and urinary bladder DNA adduct levels in beagle dogs given the carcinogen 4- aminobiphenyl. Cancer Research 51:4371-4377. Kahn SE, Hull RL, Utzschneider KM. 2006. Mechanisms linking obesity to insulin resistance and type 2 diabetes. Nature 444:840-846. Kalow W, Tang BK. 1991. Use of caffeine metabolite ratios to explore CYP1A2 and xanthine oxidase activities. Clinical Pharmacology & Therapeutics 50:508-519. Kamat AM, Lamm DL. 2004. Antitumor activity of common antibiotics against superficial bladder cancer. Urology 63:457-460. Kantor AF, Hartge P, Hoover RN, Fraumeni JF, Jr. 1988. Epidemiological characteristics of squamous cell carcinoma and adenocarcinoma of the bladder. Cancer Research 48:3853-3855. Kantor AF, Hartge P, Hoover RN, Narayana AS, Sullivan JW, Fraumeni JF, Jr. 1984. Urinary tract infection and risk of bladder cancer. Am J Epidemiol 119:510-515. Kato T, Takahashi S, Kikugawa K. 1991. Loss of heterocyclic amine mutagens by insoluble hemicellulose fiber and high-molecular-weight soluble polyphenolics of coffee. Mutat Res 246:169-178. 135 Kirkali Z, Chan T, Manoharan M et al. 2005. Bladder cancer: Epidemiology, staging and grading, and diagnosis. Urology International Consultation on Bladder Tumors 66:4-34. Kissin B, Schenker VJ, Schenker AC, Belanger IR, Stucker M, Woerner R. 1964. Hyperdiuresis after Ethanol in Chronic Alcoholics. Am J Med Sci 248:660-669. Kjaer SK, Knudsen JB, Sorensen BL, Moller Jensen O. 1989. The Copenhagen case- control study of bladder cancer. V. Review of the role of urinary-tract infection. Acta Oncologica 28:631-636. Knowles MA. 2006. Molecular subtypes of bladder cancer: Jekyll and Hyde or chalk and cheese? Carcinogenesis 27:361-373. Koya MP, Simon MA, Soloway MS. 2006. Complications of intravesical therapy for urothelial cancer of the bladder. J Urol 175:2004-2010. Krakoff LR. 2005. Diuretics for hypertension. Circulation 112:6. Kravchick S, Gal R, Cytron S, Peled R, Weissman Y, Mukamel E, Koren R. 2001. Increased incidence of diabetes mellitus in the patients with transitional cell carcinoma of urinary bladder. Pathol Oncol Res 7:56-59. La Vecchia C, Negri E, D'Avanzo B, Savoldelli R, Franceschi S. 1991. Genital and urinary tract diseases and bladder cancer. Cancer Res 51:629-631. Lamm DL, Griffith JG. 1992. Intravesical therapy: does it affect the natural history of superficial bladder cancer? Semin Urol 10:39-44. Lamm DL, Riggs DR, Shriver JS, vanGilder PF, Rach JF, DeHaven JI. 1994. Megadose vitamins in bladder cancer: a double-blind clinical trial. J Urol 151:21-26. Larsson SC, Orsini N, Brismar K, Wolk A. 2006. Diabetes mellitus and risk of bladder cancer: a meta-analysis. Diabetologia 49:2819-2823. Larsson SC, Orsini N, Wolk A. 2005. Diabetes mellitus and risk of colorectal cancer: a meta-analysis. J Natl Cancer Inst 97:1679-1687. Lifford KL, Curhan GC, Hu FB, Barbieri RL, Grodstein F. 2005. Type 2 diabetes mellitus and risk of developing urinary incontinence. Journal of the American Geriatrics Society 53:1851-1857. Lindgren A, Pukkala E, Nissinen A, Tuomilehto J. 2003. Blood pressure, smoking, and the incidence of lung cancer in hypertensive men in North Karelia, Finland. Am J Epidemiol 158:442-447. 136 Lode H, Hoffken G, Olschewski P, Sievers B, Kirch A, Borner K, Koeppe P. 1987. Pharmacokinetics of ofloxacin after parenteral and oral administration. Antimicrob Agents Chemother 31:1338-1342. Long LH, Halliwell B. 2000. Coffee drinking increases levels of urinary hydrogen peroxide detected in healthy human volunteers. Free Radic Res 32:463-467. Longhurst PA, Kauer J, Leggett RE, Levin RM. 1992. The influence of ovariectomy and estradiol replacement on urinary bladder function in rats. J Urol 148:915-919. Lovisolo JAJ, Vio P, Benvenuti C, Bono AV. 1997. Possible effect of intravenous perioperative pefloxacin on recurrence rate and disease-free interval in patients with superficial bladder cancer. Current Therapeutic Research 58:266-271. Luceri C, Caderni G, Sanna A, Dolara P. 2002. Red wine and black tea polyphenols modulate the expression of cycloxygenase-2, inducible nitric oxide synthase and glutathione-related enzymes in azoxymethane-induced f344 rat colon tumors. J Nutr 132:1376-1379. Lundgren R. 2004. Nocturia: a new perspective on an old symptom. Scand J Urol Nephrol 38:112-116. Marinho C, Alho I, Arduino D, Falcao LM, Bras-Nogueira J, Bicho M. 2007. GST M1/T1 and MTHFR polymorphisms as risk factors for hypertension. Biochem Biophys Res Commun 353:344-350. Masters PA, O'Bryan TA, Zurlo J, Miller DQ, Joshi N. 2003. Trimethoprim- sulfamethoxazole revisited. Arch Intern Med 163:402-410. Matthiesen TB, Rittig S, Mortensen JT, Djurhuus JC. 1999. Nocturia and polyuria in men referred with lower urinary tract symptoms, assessed using a 7-day frequency- volume chart. BJU Int 83:1017-1022. Maughan RJ, Griffin J. 2003. Caffeine ingestion and fluid balance: a review. Journal of Human Nutrition and Dietetics 16:411-420. McIlwain CC, Townsend DM, Tew KD. 2006. Glutathione S-transferase polymorphisms: cancer incidence and therapy. Oncogene 25:1639-1648. Melicow MM. 1974. Tumors of the bladder: a multifaceted problem. J Urol 112:467-478. Michaud DS, Spiegelman D, Clinton SK, Rimm EB, Curhan GC, Willett WC, Giovannucci EL. 1999. Fluid intake and the risk of bladder cancer in men. N Engl J Med 340:1390-1397. 137 Midthjell K, Holmen J, Bjorndal A, Lund-Larsen G. 1992. Is questionnaire information valid in the study of a chronic disease such as diabetes? The Nord- Trondelag diabetes study. J Epidemiol Community Health 46:537-542. Miller AB, Barclay TH, Choi NW et al. 1980. A study of cancer, parity and age at first pregnancy. Journal of Chronic Diseases 33:595-605. Mills PK, Beeson WL, Phillips RL, Fraser GE. 1991. Bladder cancer in a low risk population: results from the Adventist Health Study. Am J Epidemiol 133:230-239. Mokdad AH, Ford ES, Bowman BA, Dietz WH, Vinicor F, Bales VS, Marks JS. 2003. Prevalence of obesity, diabetes, and obesity-related health risk factors, 2001. Jama 289:76-79. Mommsen S, Aagaard J, Sell A. 1982. An epidemiological case-control study of bladder cancer in males from a predominantly rural district. Eur J Cancer Clin Oncol 18:1205-1210. Mommsen S, Aagaard J, Sell A. 1983. An epidemiological study of bladder cancer in a predominantly rural district. Scand J Urol Nephrol 17:307-312. Mommsen S, Sell A. 1983. Prostatic hypertrophy and venereal disease as possible risk factors in the development of bladder cancer. Urol Res 11:49-52. Morales Suarez-Varela M, Llopis Gonzalez A, Tejerizo Perez ML, Ferrandiz Ferragud J. 1993. Concentration of nitrates in drinking water and its relationship with bladder cancer. Journal of Environmental Pathology, Toxicology & Oncology 12:229-236. Morton LM, Zheng T, Holford TR et al. 2005. Alcohol consumption and risk of non- Hodgkin lymphoma: a pooled analysis. Lancet Oncol 6:469-476. Munoz ME, Martin MI, Fermoso J, Gonzalez J, Esteller A. 1987. Effect of chronic ethanol feeding on glutathione and glutathione-related enzyme activities in rat liver. Drug Alcohol Depend 20:221-226. Murata M, Takayama K, Choi BC, Pak AW. 1996. A nested case-control study on alcohol drinking, tobacco smoking, and cancer. Cancer Detect Prev 20:557-565. Najem GR, Louria DB, Seebode JJ, Thind IS, Prusakowski JM, Ambrose RB, Fernicola AR. 1982. Life time occupation, smoking, caffeine, saccharine, hair dyes and bladder carcinogenesis. Int J Epidemiol 11:212-217. Ng Y, Husain I, Waterfall N. 2003. Diabetes mellitus and bladder cancer--an epidemiological relationship? Pathology Oncology Research 9:30-31. 138 Nicodemus KK, Sweeney C, Folsom AR. 2004. Evaluation of dietary, medical and lifestyle risk factors for incident kidney cancer in postmenopausal women. Int J Cancer 108:115-121. Nomura A, Kolonel LN, Yoshizawa CN. 1989. Smoking, alcohol, occupation, and hair dye use in cancer of the lower urinary tract. Am J Epidemiol 130:1159-1163. Ogata M. 1963. Clinical and Experimental Studies on Water Metabolism in Alcoholism. Q J Stud Alcohol 24:398-411. Okajima E, Hiramatsu T, Iriya K, Ijuin M, Matsushima S. 1975. Effects of sex hormones on development of urinary bladder tumours in rats induced by N-butyl-N- (4-hydroxybutyl) nitrosamine. Urol Res 3:73-79. Oyasu R, Hopp ML. 1974. The etiology of cancer of the bladder. Surg Gynecol Obstet 138:97-108. Parker AS, Cerhan JR, Lynch CF, Ershow AG, Cantor KP. 2002. Gender, alcohol consumption, and renal cell carcinoma. Am J Epidemiol 155:455-462. Parker AS, Cerhan JR, Lynch CF, Leibovich BC, Cantor KP. 2004. History of urinary tract infection and risk of renal cell carcinoma. Am J Epidemiol 159:42-48. Parkin DM, Whelan SL, Felay J, Teppo L, Thomas DB. 2002. Cancer Incidence in Five Continents. Lyon, France: IARC Publications. Pelucchi C, Negri E, Franceschi S, Talamini R, La Vecchia C. 2002. Alcohol drinking and bladder cancer. J Clin Epidemiol 55:637-641. Piper JM, Matanoski GM, Tonascia J. 1986. Bladder cancer in young women. Am J Epidemiol 123:1033-1042. Plesko I, Preston-Martin S, Day NE, Tzonou A, Dimitrova E, Somogyi J. 1985. Parity and cancer risk in Slovakia. International Journal of Cancer 36:529-533. Pohlabeln H, Jockel KH, Bolm-Audorff U. 1999. Non-occupational risk factors for cancer of the lower urinary tract in Germany. Eur J Epidemiol 15:411-419. Pradella M, Dorizzi RM, Rigolin F. 1988. Relative density of urine: methods and clinical significance. Critical Reviews in Clinical Laboratory Sciences 26:195-242. Probst-Hensch NM, Bell DA, Watson MA et al. 2000. N-acetyltransferase 2 phenotype but not NAT1*10 genotype affects aminobiphenyl-hemoglobin adduct levels. Cancer Epidemiology, Biomarkers & Prevention 9:619-623. 139 Pruthi RS, Derksen E, Gaston K, Wallen EM. 2004. Rationale for use of cyclooxygenase-2 inhibitors in prevention and treatment of bladder cancer. Urology 64:637-642. Radosavljevic V. 2004. Fluid balance, diet and bladder cancer occurrence. European Journal of Cancer Prevention 13. Radosavljevic V, Jankovic S, Marinkovic J, Djokic M. 2003. Fluid intake and bladder cancer. A case control study. Neoplasma 50:234-238. Raghavan D, Shipley WU, Garnick MB, Russell PJ, Richie JP. 1990. Biology and management of bladder cancer. New England Journal of Medicine 322:1129-1138. Rashidkhani B, Akesson A, Lindblad P, Wolk A. 2005. Alcohol consumption and risk of renal cell carcinoma: a prospective study of Swedish women. Int J Cancer 117:848-853. Reid LM, Leav I, Kwan PW, Russell P, Merk FB. 1984. Characterization of a human, sex steroid-responsive transitional cell carcinoma maintained as a tumor line (R198) in athymic nude mice. Cancer Res 44:4560-4573. Rembratt A, Norgaard JP, Andersson KE. 2003. Differences between nocturics and non-nocturics in voiding patterns: an analysis of frequency-volume charts from community-dwelling elderly. BJU Int 91:45-50. Reyes AJ. 2002. Diuretics in the therapy of hypertension. Journal of Human Hypertension 16. Ries L, Harkins D, Krapcho M et al. 2006. SEER Cancer Statistics Review, 1975- 2003, National Cancer Institute. Bethesda, MD: SEER web site: seer.cancer.gov/csr/1975_2003/ based on November 2005 SEER data submission. Risch HA, Burch JD, Miller AB, Hill GB, Steele R, Howe GR. 1988. Dietary factors and the incidence of cancer of the urinary bladder. Am J Epidemiol 127:1179-1191. Rundle A, Tang D, Mooney L, Grumet S, Perera F. 2003. The interaction between alcohol consumption and GSTM1 genotype on polycyclic aromatic hydrocarbon- DNA adduct levels in breast tissue. Cancer Epidemiol Biomarkers Prev 12:911-914. Sala M, Cordier S, Chang-Claude J et al. 2000. Coffee consumption and bladder cancer in nonsmokers: a pooled analysis of case-control studies in European countries. Cancer Causes & Control 11:925-931. Seay TM, Peretsman SJ, Dixon PS. 1996. Inhibition of human transitional cell carcinoma in vitro proliferation by fluoroquinolone antibiotics. J Urol 155:757-762. 140 Shah SU, Anjum S, Littler WA. 2004a. Use of diuretics in cardiovascular disease: (2) hypertension. Postgraduate Medical Journal 80:271-276. Shah SU, Anjum S, Littler WA. 2004b. Use of diuretics in cardiovascular diseases: (1) heart failure. Postgraduate Medical Journal 80:201-205. Shi SR, Cote RJ, Liu C, Yu MC, Castelao JE, Ross RK, Taylor CR. 2002. A modified reduced-temperature antigen retrieval protocol effective for use with a polyclonal antibody to cyclooxygenase-2 (PG 27). Appl Immunohistochem Mol Morphol 10:368-373. Shirai T, Tsuda H, Ogiso T, Hirose M, Ito N. 1987. Organ specific modifying potential of ethinyl estradiol on carcinogenesis initiated with different carcinogens. Carcinogenesis 8:115-119. Sierksma A, van der Gaag MS, van Tol A, James RW, Hendriks HF. 2002. Kinetics of HDL cholesterol and paraoxonase activity in moderate alcohol consumers. Alcohol Clin Exp Res 26:1430-1435. Skipper PL, Tannenbaum SR, Ross RK, Yu MC. 2003. Nonsmoking-related arylamine exposure and bladder cancer risk. Cancer Epidemiology, Biomarkers & Prevention 12:503-507. Slattery ML, West DW, Robison LM. 1988. Fluid intake and bladder cancer in Utah. Int J Cancer 42:17-22. Smith-Warner SA, Spiegelman D, Yaun SS et al. 1998. Alcohol and breast cancer in women: a pooled analysis of cohort studies. Jama 279:535-540. Spindel E. 1984. Action of the methylxanthines on the pituitary and pituitary- dependent hormones. Progress in Clinical & Biological Research 158:355-363. Spruck C, 3rd, Ohneseit P, Gonzalez-Zulueta M et al. 1994. Two molecular pathways to transitional cell carcinoma of the bladder. Cancer Res 54:784-788. Stamm WE. 2002. Scientific and clinical challenges in the management of urinary tract infections. Am J Med 113 Suppl 1A:1S-4S. Stamm WE, Hooton TM. 1993. Management of urinary tract infections in adults. N Engl J Med 329:1328-1334. Stevens JF, Page JE. 2004. Xanthohumol and related prenylflavonoids from hops and beer: to your good health!. Phytochemistry 65:1317-1330. 141 Straka RJ, Burkhardt RT, Lang NP, Hadsall KZ, Tsai MY. 2006. Discordance between N-acetyltransferase 2 phenotype and genotype in a population of Hmong subjects. J Clin Pharmacol 46:802-811. Sturgeon SR, Hartge P, Silverman DT, Kantor AF, Linehan WM, Lynch C, Hoover RN. 1994. Associations between bladder cancer risk factors and tumor stage and grade at diagnosis. Epidemiology 5:218-225. Tanagho EA, McAninch JW. 2003. Smith's general urology: McGraw-Hill. Tanahashi NK, Suzawa N, Azuma C. 1977. Effects of sex hormones on oncogenesis in rat urinary bladder by N-butyl-N-(4-hydroxybutyl)-nitrosamine. Int J Clin Pharmacol Biopharm 15:101-105. Tavani A, La Vecchia C. 2000. Coffee and cancer: a review of epidemiological studies, 1990-1999. European Journal of Cancer Prevention 9:241-256. Thomas DB, Uhl CN, Hartge P. 1983. Bladder cancer and alcoholic beverage consumption. Am J Epidemiol 118:720-727. Thorpe A, Neal D. 2003. Benign prostatic hyperplasia. Lancet 361:1359-1367. Thrasher JB, Crawford ED. 1992. Complications of intravesical chemotherapy. Urol Clin North Am 19:529-539. Tripathi A, Folsom AR, Anderson KE. 2002a. Risk factors for urinary bladder carcinoma in postmenopausal women. The Iowa Women's Health Study. Cancer 95:2316-2323. Tripathi A, Folsom AR, Anderson KE, Iowa Women's Health S. 2002b. Risk factors for urinary bladder carcinoma in postmenopausal women. The Iowa Women's Health Study. Cancer 95:2316-2323. Tubaro A, Palleschi G. 2005. Overactive bladder: epidemiology and social impact. Curr Opin Obstet Gynecol 17:507-511. Van de Wiel JA, Fijneman PH, Teeuw KB, Van Ommen B, Noordhoek J, Bos RP. 1993. Influence of long-term ethanol treatment on rat liver biotransformation enzymes. Alcohol 10:397-402. van Kerrebroeck P, Abrams P, Chaikin D et al. 2002. The standardisation of terminology in nocturia: report from the Standardisation Sub-committee of the International Continence Society. Neurourol Urodyn 21:179-183. 142 Vanhaecke T, Lindros KO, Oinonen T et al. 2000a. Effect of ethanol on the expression of hepatic glutathione S-transferase: an in vivo/in vitro study. Biochem Pharmacol 60:1491-1496. Vanhaecke T, Lindros KO, Oinonen T, Vercruysse A, Rogiers V. 2000b. Effect of long-term ethanol exposure on the acinar distribution of hepatic glutathione S- transferase. Drug Metab Dispos 28:1470-1474. Vena JE, Freudenheim J, Graham S, Marshall J, Zielezny M, Swanson M, Sufrin G. 1993. Coffee, cigarette smoking, and bladder cancer in western New York. Annals of Epidemiology 3:586-591. Villanueva CM, Cantor KP, King WD et al. 2006. Total and specific fluid consumption as determinants of bladder cancer risk. International Journal of Cancer 118:2040-2047. Viscoli CM, Lachs MS, Horwitz RI. 1993. Bladder cancer and coffee drinking: a summary of case-control research. Lancet 341:1432-1437. Wannamethee SG, Shaper AG, Walker M. 2001. Physical activity and risk of cancer in middle-aged men. Br J Cancer 85:1311-1316. Ward EM, Sabbioni G, DeBord DG et al. 1996. Monitoring of Aromatic Amine Exposures in Workers at a Chemical Plant With a Known Bladder Cancer Excess 10.1093/jnci/88.15.1046. J Natl Cancer Inst 88:1046-1053. Washio M, Mori M, Khan M et al. 2007. Diabetes mellitus and kidney cancer risk: The results of Japan Collaborative Cohort Study for Evaluation of Cancer Risk (JACC Study). Int J Urol 14:393-397. West DB, Wehberg KE, Kieswetter K, Granger JP. 1992. Blunted natriuretic response to an acute sodium load in obese hypertensive dogs. Hypertension 19:I96- 100. Wheeler MA, Hausladen DA, Yoon JH, Weiss RM. 2002. Prostaglandin E2 production and cyclooxygenase-2 induction in human urinary tract infections and bladder cancer. J Urol 168:1568-1573. Wilkens LR, Kadir MM, Kolonel LN, Nomura AM, Hankin JH. 1996. Risk factors for lower urinary tract cancer: the role of total fluid consumption, nitrites and nitrosamines, and selected foods. Cancer Epidemiol Biomarkers Prev 5:161-166. Wolf I, Sadetzki S, Catane R, Karasik A, Kaufman B. 2005. Diabetes mellitus and breast cancer. Lancet Oncol 6:103-111. 143 Wolk A, Gridley G, Niwa S et al. 1996. International renal cell cancer study. VII. Role of diet. Int J Cancer 65:67-73. Wong ES, Stamm WE. 1983. Urethral infections in men and women. Annu Rev Med 34:337-358. Wu JM, Williams KS, Hundley AF, Jannelli ML, Visco AG. 2006. Microscopic hematuria as a predictive factor for detecting bladder cancer at cystoscopy in women with irritative voiding symptoms. Am J Obstet Gynecol 194:1423-1426. Wu MM, Kuo TL, Hwang YH, Chen CJ. 1989. Dose-response relation between arsenic concentration in well water and mortality from cancers and vascular diseases. American Journal of Epidemiology 130:1123-1132. Yang CM, Carlson GP. 1991. Effects of ethanol on glutathione conjugation in rat liver and lung. Biochem Pharmacol 41:923-929. Yoshimura Y, Schmidt F, Constantinou CE. 2000. Gender specificity of tolterodine on micturition and the diurnal variation of urine production of the conscious rat. BJU Int 86:879-885. Yu MC, Ross RK. 1998. Epidemiology of Bladder Cancer. Berlin: Springer-Verlag. Yu MC, Ross RK. 2002. Bladder Cancer: Epidemiology. In: Bertino JR, ed. Encyclopedia of Cancer. San Diego: Academic Press. pp 215-221. Yu MC, Ross RK, Chan KK, Henderson BE, Skipper PL, Tannenbaum SR, Coetzee GA. 1995. Glutathione S-transferase M1 genotype affects aminobiphenyl- hemoglobin adduct levels in white, black and Asian smokers and nonsmokers. Cancer Epidemiology, Biomarkers & Prevention 4:861-864. Yu MC, Skipper PL, Taghizadeh K, Tannenbaum SR, Chan KK, Henderson BE, Ross RK. 1994. Acetylator phenotype, aminobiphenyl-hemoglobin adduct levels, and bladder cancer risk in white, black, and Asian men in Los Angeles, California. Journal of the National Cancer Institute 86:712-716. Yuan JM, Castelao JE, Gago-Dominguez M, Ross RK, Yu MC. 1998a. Hypertension, obesity and their medications in relation to renal cell carcinoma. British Journal of Cancer 77:1508-1513. Yuan JM, Gago-Dominguez M, Castelao JE, Hankin JH, Ross RK, Yu MC. 1998b. Cruciferous vegetables in relation to renal cell carcinoma. Int J Cancer 77:211-216. 144 Zeegers MP, Dorant E, Goldbohm RA, van den Brandt PA. 2001a. Are coffee, tea, and total fluid consumption associated with bladder cancer risk? Results from the Netherlands Cohort Study. Cancer Causes Control 12:231-238. Zeegers MP, Kellen E, Buntinx F, van den Brandt PA. 2004. The association between smoking, beverage consumption, diet and bladder cancer: a systematic literature review. World Journal of Urology 21:392-401. Zeegers MP, Tan FE, Verhagen AP, Weijenberg MP, van den Brandt PA. 1999. Elevated risk of cancer of the urinary tract for alcohol drinkers: a meta-analysis. Cancer Causes Control 10:445-451. Zeegers MPA, Volovics A, Dorant E, Alexandra Goldbohm R, van den Brandt PA. 2001b. Alcohol Consumption and Bladder Cancer Risk: Results from the Netherlands Cohort Study. Am J Epidemiol 153:38-41. Zehavi-Willner T, Shalit I. 1992. The inhibitory effect of ciprofloxacin on proliferation of a murine bladder carcinoma cell line. J Antimicrob Chemother 29:323-328. Zhao F, Nozawa H, Daikonnya A, Kondo K, Kitanaka S. 2003a. Inhibitors of nitric oxide production from hops (Humulus lupulus L.). Biol Pharm Bull 26:61-65. Zhao H, Grossman HB, Spitz MR, Lerner SP, Zhang K, Wu X. 2003b. Plasma levels of insulin-like growth factor-1 and binding protein-3, and their association with bladder cancer risk. J Urol 169:714-717. Zheng T, Holford TR, Zahm SH et al. 2003. Glutathione S-transferase M1 and T1 genetic polymorphisms, alcohol consumption and breast cancer risk. Br J Cancer 88:58-62. Zierler S, Feingold L, Danley RA, Craun G. 1988. Bladder cancer in Massachusetts related to chlorinated and chloraminated drinking water: a case-control study. Archives of Environmental Health 43:195-200.

Asset Metadata

Creator Jiang, Xuejuan (author)

Core Title Fluid intake, micturition habits, associated medications and conditions as potential risk factors for bladder cancer

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Epidemiology

Publication Date 06/27/2007

Defense Date 06/14/2007

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

Tag Alcohol,bladder cancer,fluid intake,OAI-PMH Harvest,urinary tract infection,Water

Language English

Advisor Gago-Dominquez, Manuela (committee chair), Conti, David V. (committee member), Cortessis, Victoria Kristence. (committee member), Groshen, Susan (committee member), Shibata, Darryl K. (committee member)

Creator Email xuejuanj@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m559

Unique identifier UC1161135

Identifier etd-Jiang-20070627 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-508618 (legacy record id),usctheses-m559 (legacy record id)

Legacy Identifier etd-Jiang-20070627.pdf

Dmrecord 508618

Document Type Dissertation

Rights Jiang, Xuejuan

Type texts

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

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email uscdl@usc.edu

Abstract (if available)

Abstract The urogenous-contact hypothesis proposes that the development of bladder cancer is associated with prolonged exposure to carcinogens in urine. To fully examine this hypothesis, this dissertation systematically addressed fluid intake, micturition habits, associated medical conditions, and medications in relation to risk of bladder cancer in the Los Angeles Bladder Cancer Case-Control Study.