3
Lifestyle Behaviors Contributing to the Burden of Cancer1

Much of the burden of cancer in the United States can be traced to modifiable health behaviors that increase one’s risk of disease. This chapter reviews evidence on the contributions of five major risk factors to cancer incidence and mortality: tobacco use, physical inactivity, overweight and obesity, poor diet, and alcohol use. Although other lifestyle factors such as sun exposure, sexual practices, and exposure to infected blood also contribute substantially to cancer incidence, this chapter focuses only on the factors that are risks for common cancers and that also have large impacts on the incidences of other major diseases such as cardiovascular disease, stroke, diabetes, and osteoporosis. The benefits of intervention are far greater when reductions in the incidence of other chronic diseases add to reductions in the incidence of cancer achieved through lifestyle changes (Colditz and Gortmaker, 1995).

This chapter summarizes the epidemiological evidence with respect to cancer incidence for each of the five selected risk factors by using the criteria outlined by the World Cancer Research Fund and the American Institute of Cancer Research. For each epidemiological association the level of evidence is categorized as convincing, probable, or possible (Box 3.1).

Much of the literature reviewed in this chapter has been published from research conducted with data from one of the several large prospective cohort studies described in Box 3.2. Such studies involve the identification

1  

The chapter is based on the background paper prepared by Graham A. Colditz, Catherine Tomeo Ryan, Charles H. Dart III, Geetanjali Datta, Laurie Fisher, and Beverly Rockhill (www.iom.edu/ncpb).



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection 3 Lifestyle Behaviors Contributing to the Burden of Cancer1 Much of the burden of cancer in the United States can be traced to modifiable health behaviors that increase one’s risk of disease. This chapter reviews evidence on the contributions of five major risk factors to cancer incidence and mortality: tobacco use, physical inactivity, overweight and obesity, poor diet, and alcohol use. Although other lifestyle factors such as sun exposure, sexual practices, and exposure to infected blood also contribute substantially to cancer incidence, this chapter focuses only on the factors that are risks for common cancers and that also have large impacts on the incidences of other major diseases such as cardiovascular disease, stroke, diabetes, and osteoporosis. The benefits of intervention are far greater when reductions in the incidence of other chronic diseases add to reductions in the incidence of cancer achieved through lifestyle changes (Colditz and Gortmaker, 1995). This chapter summarizes the epidemiological evidence with respect to cancer incidence for each of the five selected risk factors by using the criteria outlined by the World Cancer Research Fund and the American Institute of Cancer Research. For each epidemiological association the level of evidence is categorized as convincing, probable, or possible (Box 3.1). Much of the literature reviewed in this chapter has been published from research conducted with data from one of the several large prospective cohort studies described in Box 3.2. Such studies involve the identification 1   The chapter is based on the background paper prepared by Graham A. Colditz, Catherine Tomeo Ryan, Charles H. Dart III, Geetanjali Datta, Laurie Fisher, and Beverly Rockhill (www.iom.edu/ncpb).

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection BOX 3.1 Definition of Levels of Evidence for Epidemiological Associations Level of Evidence Criteria Convincing • Epidemiological findings across a large number of well-designed studies are consistent. • A dose-response relationship has been demonstrated. • Mechanisms are biologically plausible. • Laboratory evidence is supportive. Probable • Epidemiological evidence is either inconsistent or not extensive enough to make definitive judgment. • Laboratory or mechanistic evidence is supportive. Possible • Epidemiological findings are supportive but limited in quantity, quality, or consistency. • Supportive mechanistic or laboratory evidence may or may not be available. • Supportive evidence from other disciplines may be available.   SOURCE: Adapted from (World Cancer Research Fund and American Institute for Cancer Research, 1997). of a group of individuals who share a common experience within a defined time period and monitoring of those individuals forward in time for the development of disease (Mausner and Kramer, 1985). These studies have been essential to providing an understanding of the links between health-related behaviors and health outcomes. Other evidence comes from case-control studies in which investigators identify a group of patients with a particular cancer (cases) and a group of patients without cancer (controls) and then compare the histories of the cases and controls to determine the extent to which each was exposed to the intervention of interest. TOBACCO Tobacco is the scourge of public health. In the United States alone, tobacco use, primarily in the form of cigarette smoking, causes more than 440,000 premature deaths from cancer and other causes each year and is responsible for approximately 30 percent of all cancer-related deaths (ACS, 2002b). Worldwide, the numbers are even more staggering. The World Health Organization estimated that in 1998 there would be approximately 4 million deaths linked to tobacco use worldwide (World Health Organization, 1999). By 2020, this number is expected to at least double.

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection BOX 3.2 Frequently Referenced Prospective Cohort Studies American Cancer Society’s Cancer Prevention Studies Cancer Prevention Study I (CPS I) A prospective study of approximately one million men and women. Enrollment started in 1959 and follow-up ran through 1972, with mortality being the primary outcome measure (ACS, 2001). Enrollment was household-based, and all members of a household age 30 or older were included in the study if at least one household member was age 45 or older. Upon entering the study, participants completed questionnaires assessing lifestyle factors (e.g., occupation, diet, and tobacco use), reproductive factors (in women), personal and family history of cancer, as well as anthropometric and demographic information. After this, questionnaires were sent to participants at regular intervals to update tobacco use and assess vital status. Cancer Prevention Study II (CPS II) A prospective study of approximately 1.2 million men and women (ACS, 2001). Started in 1982 and still ongoing, the primary aim of CPS II is to assess the effect that lifestyle and environmental factors have on cancer development. As with CPS I, enrollment in CPS II was household-based and included all household members age 30 or older if at least one member was age 45 or older. A range of lifestyle and other factors were assessed by questionnaire at study enrollment. The study’s main outcome measure is mortality, which is assessed biennially through links with the National Death Index. Health Professionals Follow-Up Study A prospective study of approximately 52,000 male health professionals—including dentists, optometrists, osteopaths, podiatrists, pharmacists, and veterinarians (Harvard School of Public Health, 2001). Started in 1986 in men ages 40–75, it is still ongoing. The primary aim of the Health Professionals Follow-Up Study is to assess the effect of lifestyle on the risk of chronic disease, particularly cancer and cardiovascular disease. Data are gathered largely through biennial questionnaires that ask participants detailed questions about their disease status as well as about their lifestyle and personal characteristics. Male British Doctors Study A prospective mortality study of 34,000 male British doctors, which started in 1951 and is still ongoing (Doll et al., 1994). The study’s initial aim was to build a body of evidence linking tobacco use to chronic disease. While the questionnaire at enrollment focused largely on tobacco use, subsequent follow-up questionnaires added questions related to alcohol use, aspirin use, disease status, and certain personal characteristics. Nurses’ Health Study A prospective study of approximately 120,000 female registered nurses, which is primarily designed to assess the effect of lifestyle on the risk of chronic disease, particularly cancer and cardiovascular disease (Colditz, 1995). Started in 1976 in women ages 30–55, the Nurses’ Health Study is still ongoing. Data are gathered primarily through biennial questionnaires that ask participants detailed questions about their disease status as well as their lifestyle and personal characteristics. To provide a more accurate accounting of the intake of certain minerals, participants provided toenail samples in 1982. In 1989, participants provided blood samples to allow the study of potential disease biomarkers.

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection Since the first Surgeon General’s report on smoking and health was released in 1964, a causal link has been confirmed between smoking and several cancers (U.S. Department of Health, Education, and Welfare, 1964). The data available at that time were able to support a causal link between smoking and only two cancers: lung cancer and laryngeal cancer. Over the following 35 years, however, enough evidence has accrued to now support smoking as a cause of eight cancers: lung, oral, pharyngeal, laryngeal, esophageal, bladder, kidney, and pancreatic cancer. A review of the current literature provides sufficient evidence to also implicate tobacco as a cause of additional cancers, including cancers of the colon, stomach, and cervix, and leukemia. Smoking has also been identified as a probable cause of liver cancer and has been associated with an increase in the risk of developing aggressive, more deadly forms of prostate cancer (Table 3.1). Smoking is not causally related to several types of cancer, such as breast or brain cancer. Impact of Cessation of Tobacco Use on Cancer Incidence Although the prevention of smoking is the best approach to avoiding tobacco-related diseases, there are substantial health benefits for smokers who quit. Scientific data on the benefits of smoking cessation were reviewed in detail in a 1990 Surgeon General’s report (US DHHS, 1990). In the last quarter century, half of all living Americans who have ever smoked have now quit. The 1990 Surgeon General’s report concluded that smoking cessation has major and immediate health benefits for men and women of all ages. Former smokers live longer than continuing smokers; for instance, persons who quit smoking before age 50 have, on average, half the risk of dying in the next 15 years than continuing smokers. This reduction in mortality comes from a reduction in the risk of nearly all smoking-related TABLE 3.1 Increase in Risk of Incident Cancer Associated with Smoking   Relative Risk (RR) Level of Evidence Moderate (RR 1.35–1.99) Large (RR 2.0+) Convincing Colon Stomach Leukemia Cervical Lung Oral Pharyngeal Laryngeal Esophageal Bladder Kidney Pancreatic Probable Prostate (mortality) Liver

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection diseases. Among former smokers, the decline in the risk of death compared with that for individuals who continue smoking begins shortly after quitting and continues for at least 10 to 15 years. The health benefits of smoking cessation far exceed the risk from the weight gain, which is approximately 2.3 kilograms (kg) (5 pounds [lb]), on average, that may follow quitting. Tobacco Use and Lung Cancer Voluminous and convincing epidemiological evidence links smoking to lung cancer, the leading cause of cancer death in both men and women in the United States. An estimated 169,400 new cases of lung cancer are diagnosed annually, with an estimated 154,900 deaths each year (ACS, 2002a). More than 80 percent of lung cancer cases are attributable to smoking. As outlined in detail in the Surgeon General’s reports on smoking and health, cigarette smoking is the strongest risk factor for lung cancer, increasing the risk of the disease by at least 10-fold and as much as 20-fold, depending on smoking habits and history (US DHHS, 1989; Thun, Dally-Lally et al., 1997). There is a consistent, strong, and specific link between smoking and lung cancer, with a dose-response relationship seen with the number of cigarettes smoked, the deepness of inhalation of cigarette smoke, and the duration of smoking (US DHHS, 1989). There is convincing evidence from numerous case-control and cohort studies that former smokers have a lower risk of lung cancer than current smokers (US DHHS, 1990). The benefits begin 2 to 3 years after quitting, and the risk steadily drops over the next 10 years. Although the risk of lung cancer is drastically reduced after quitting, the risk of lung cancer in former smokers never quite returns to the risk for those who have never smoked. The overall pattern seen in the study of U.S. veterans mimics that of most other large cohort studies of smoking cessation and lung cancer risk (Rogot and Murray, 1980; US DHHS, 1990). In the U.S. Veterans Study, the lung cancer mortality rate in current smokers divided by the rate in nonsmokers (i.e., the mortality rate ratio) was 11.3. In the first 4 years after quitting, this ratio increased to nearly 19 for former smokers and then slowly decreased: 7.7 after 5 to 9 years, 4.7 after 10 to 14 years, 4.8 after 15 to 19 years, and 2.1 after 20 or more years. The increase in lung cancer risk in the first few years after quitting seen in many studies likely reflects the tendency of smokers to quit because of symptoms of disease rather than a true increase in risk linked to cessation per se. Numerous studies have also found that smoking cessation benefits nearly all smokers, regardless of sex, age, and level of habit. The American Cancer Society’s Cancer Prevention Study I and Cancer Prevention Study II have demonstrated that both men and women benefit from cessation, as do both light smokers (those who smoke 1 to 20 cigarettes/day) and heavy

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection smokers (those who smoke 21 or more cigarettes/day). In addition, an analysis of national smoking trends in the United Kingdom found substantial benefits for smokers who stopped at a wide range of ages (Peto et al., 2000). Those smokers who stopped before middle age benefited the most, avoiding 90 percent of the excess risk of lung cancer linked to tobacco use. However, those who stopped at ages 50 and 60 years also benefited substantially. Tobacco Use and Oral, Pharyngeal, Laryngeal, and Esophageal Cancers Tobacco use is a well-established cause of cancer of the oral cavity, pharynx, larynx, and esophagus (US DHHS, 1989). Oropharyngeal cancer alone—including cancer of the lip, tongue, oral cavity, and oropharynx—is diagnosed in approximately 28,900 people each year in the United States and accounts for about 7,400 deaths annually (ACS, 2002a). Once oropharyngeal cancer was as much as six times more common in men than in women, but men are now only about twice as likely as women to develop the disease. This is due in large part to the disappearing disparity in smoking rates between the two sexes (CDC, 1999b). Although disparity in the incidence of oropharyngeal cancer by sex is waning, the disparity in the incidence by race or ethnicity persists. African Americans not only have a greater risk than whites of developing oropharyngeal cancer, but they also have a much less favorable prognosis after they receive a diagnosis. Only about a third of African Americans survive 5 years after diagnosis, compared to more than half of whites (Ries et al., 2000a). Numerous cohort and case-control studies have found that the risks of both the incidence of and the mortality from these types of cancers are significantly increased in smokers compared with those in nonsmokers, and many studies have demonstrated a dose-response relationship. The long-running cohort study of British male doctors found rates of mortality from esophageal cancer among the heaviest smokers to be 15 times that among nonsmokers (Doll et al., 1994). Among all current smokers combined, the increased rate of mortality was over seven times that among nonsmokers. In a study of U.S. veterans, current smokers experienced 10 times the risk of mortality from laryngeal cancer than those who never smoked, a risk similar to that seen in the American Cancer Society’s Cancer Prevention Study II (Kahn, 1966; US DHHS, 1990). In addition, a case-control study conducted by the National Cancer Institute found that the risk of oropharyngeal cancer was increased two to three times among those who ever smoked compared with that among those who never smoked, and the risk increased with the amount and duration of smoking. Women who smoked 40 or more cigarettes per day had six times the risk of oropharyngeal cancer as someone who never smoked. Men who smoked 40 or more cigarettes per

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection day had three times the risk. Overall, the increased incidence of oropharyngeal cancer in smokers compared with that in nonsmokers across studies ranges from 3 to 13 (Schottenfeld and Fraumeni, 1996). The most likely causal pathway between tobacco use and these cancers is the direct contact of the carcinogens in tobacco and its smoke with the tissues of the oral cavity, larynx, and esophagus. Yet, although smoking alone is an independent risk factor for oral, laryngeal, and esophageal cancer, alcohol consumption greatly exacerbates smoking’s effect on risk (Blot et al., 1988). It is estimated that alcohol and tobacco use together account for approximately 75 percent of oral cancers in the United States (Blot et al., 1988). There is convincing evidence that former smokers have lower risks of cancers of the oral cavity, pharynx, larynx, and esophagus than current smokers. For oropharyngeal cancer, the risk for former smokers compared with that for current smokers decreases steadily with the number of years since cessation (after the first few years), with some studies demonstrating that the risk of oropharyngeal cancer actually returns to that for never smokers 10 or more years after quitting (Blot et al., 1988). As with lung cancer and smoking cessation, there is also good evidence that the excess risk of oropharyngeal cancer decreases with younger age at cessation (US DHHS, 1990). For laryngeal and esophageal cancer, numerous studies have demonstrated significant drops in risk for former smokers compared with that for current smokers beginning 3 to 4 years after cessation (US DHHS, 1990). Tobacco Use and Bladder Cancer Approximately 56,500 cases of bladder cancer are diagnosed each year (ACS, 2002a). Whites have the highest incidence rates, about twice the rate of Hispanics and African Americans (Miller et al., 1996). Significant disparity in the incidence of bladder cancer also exists by sex, as men are about three to four times more likely to develop the disease than women (Miller et al., 1996). There is convincing evidence that smoking is a cause of bladder cancer. More than 30 case-control and 10 prospective cohort studies support a strong link between smoking and the disease (Silverman et al., 1992). Overall, moderate to heavy smokers tend to have two to five times the risk of nonsmokers, and there is a strong dose-response relationship between smoking and the risk of bladder cancer, with the risk increasing with the duration and amount of smoking. Specifically, the 40-year analysis of British male doctors found that the rate of mortality from bladder cancer was three times greater among heavy smokers (those who smoke 25 or more cigarettes a day) than among nonsmokers (Doll et al., 1994). The most likely causal pathway between tobacco smoking and bladder cancer is the expo-

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection sure of bladder tissue to the carcinogenic by-products of tobacco metabolism that are excreted in urine. It is estimated that a third of all bladder cancers in women and half of all bladder cancers in men are caused by cigarette smoking (Silverman et al., 1992). There is convincing evidence that smoking cessation can lower the risk of bladder cancer in former smokers compared with that in current smokers. A recent pooled analysis of 11 case-control studies conducted by the International Agency for Research on Cancer found that the risk of bladder cancer in former smokers compared with that in current smokers began to decrease almost immediately after smoking cessation and continued to do so with the duration of cessation (Brennan et al., 2000). The risk dropped 35 percent 1 to 4 years after quitting and dropped more than 60 percent 25 years after quitting, but it never reached that for someone who had never smoked. Tobacco Use and Kidney Cancer Although kidney cancer is less common than bladder cancer, there is convincing evidence from case-control and prospective cohort studies that smoking causes kidney cancer. Approximately 31,800 cases of kidney cancer are diagnosed each year (ACS, 2002a). Men are approximately twice as likely as women to develop the disease (Miller et al., 1996) African Americans, whites, and Hispanics each appear to develop kidney cancer at similar rates (Miller et al., 1996). As reviewed in the 1989 Surgeon General’s report, the relative risk for kidney cancer associated with smoking ranges from one to five and exhibits a dose-response relationship with the number of cigarettes smoked (U.S Department of Health and Human Services, 1989). More recent studies bolster these results. A large study with a cohort of U.S. veterans found that men who smoked more than 40 cigarettes a day had double the risk of kidney cancer compared with that for nonsmokers (McLaughlin et al. 1995a). The International Renal Cell Cancer Study, a large case-control study, also found that the risk increased with both the number of cigarettes smoked and the duration of smoking (McLaughlin et al. 1995b). The causal pathway is likely similar to that for bladder cancer: exposure of kidney tissue to the carcinogenic by-products of tobacco metabolism that end up in urine. For kidney cancer (including cancer of the renal pelvis), smoking cessation does not seem to provide as much of a benefit as it does for bladder cancer. The study of U.S. veterans found only slight differences in the risk of kidney cancer between former and current smokers (Kahn, 1966), and the International Renal Cell Cancer Study found that after more than 15 years of cessation, the risk of kidney cancer in former smokers was only 15 to 25 percent lower than among current smokers (McLaughlin et al., 1995b).

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection Tobacco Use and Pancreatic Cancer Pancreatic cancer is the most deadly of the major cancers, with 5-year survival rates after diagnosis only about 5 percent (ACS, 2002a). Approximately 30,300 cases are diagnosed each year in the United States (ACS, 2002a). The disease is more common in men than in women and is about 50 percent more common in African Americans than in whites (Miller et al., 1996). Smoking was described as a contributory factor to pancreatic cancer in the 1989 Surgeon General’s report, but the evidence now supports smoking as a cause of the disease (US DHHS, 1989). Both case-control and prospective cohort studies have not only linked smoking to an increased risk of pancreatic cancer but also demonstrated a dose-response relationship. Overall, smokers appear to have two to three times the risk of getting pancreatic cancer as nonsmokers. Although the causal pathway between smoking and pancreatic cancer is unknown, it has been proposed that tobacco carcinogens or their by-products cause mutations in pancreatic cells linked to carcinogenesis. There is convincing evidence that individuals who quit smoking experience significantly decreased risks of pancreatic cancer. Unlike many other cancers linked to smoking, the risk of pancreatic cancer in former smokers begins to drop in the first couple of years after cessation (Anderson et al., 1996). A large prospective study of men found that the risk of the disease dropped nearly 50 percent within 2 years of stopping, and after less than 10 years, the risk nearly returned to that for someone who had never smoked (Fuchs et al., 1996). Tobacco Use and Colon Cancer The American Cancer Society estimates that there were 148,300 new cases of colorectal cancer in 2002 and that 56,600 Americans died of the disease (ACS, 2002a). An association between smoking and colorectal cancer was not discussed in previous Surgeon General’s reports on smoking and health because of a lack of data, but there is now convincing new evidence for a causal association between smoking and an increased risk of colon cancer, as well as colon polyps, among both men and women (Giovannucci and Martinez, 1996). A Swedish prospective study of nearly 57,000 men and women found a 60 percent increased risk of colon cancer in smokers compared with that in nonsmokers after 11 to 20 years of follow-up (Knekt et al., 1998). The Nurses’ Health Study and the Health Professionals Follow-Up Study demonstrated significant elevations in risk in long-term smokers as well (Giovannucci et al., 1994a,b). Men who had been smoking for at least 35 years had nearly double the risk of colon cancer compared with that for nonsmokers, as did women who had smoked for 45 years. The long latency period between the initiation of smoking and

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection the elevation in risk, coupled with the consistent relationship seen between smoking and colorectal polyps, suggests that tobacco may be an initiator of colorectal carcinogenesis. There is an increasing amount of evidence that smoking cessation can lower the risk of colon cancer, although this evidence was not included in the 1990 Surgeon General’s report mentioned above. Data from the American Cancer Society’s Cancer Prevention Study II indicate that the risk of colon cancer drops with the number of years of smoking cessation as well as a younger age at the time of cessation (Chao et al., 2000). Tobacco Use and Stomach Cancer Stomach cancer is one of the most common cancers worldwide. Although it is much less common in the United States, nearly 21,600 cases of stomach cancer are diagnosed and there are approximately 12,400 deaths from stomach cancer each year (ACS, 2002a). In general, Asian populations in the United States have the highest rates of incidence of the disease; the exception, however, is Filipinos, who, along with whites, have the lowest rates of stomach cancer (Miller et al., 1996). African Americans and Hispanics tend to have higher rates than whites but lower rates than most Asian groups. As early as 1982, the Surgeon General’s reports on smoking and health have documented a positive association between smoking and stomach cancer (US DHHS, 1982). Both prospective and case-control studies have found a link between the two, with a number demonstrating a dose-response relationship (Tredaniel et al., 1997). Although the results of case-control studies have been somewhat variable, prospective cohort studies have consistently found an increased risk of stomach cancer linked to smoking. One meta-analysis of 40 studies—case-control and cohort studies combined—found that, overall, smoking increased the risk of stomach cancer by 50 to 60 percent (Tredaniel et al., 1997). When men and women were assessed separately, male smokers had a greater risk of stomach cancer from smoking than female smokers. A number of possible causal pathways exist between smoking and stomach cancer. Smoking has been linked to Helicobacter pylori infection, a major risk factor for stomach cancer worldwide, as well as decreased levels in serum of certain micronutrients such as carotenoids and vitamin C that may help protect against the disease. It is estimated that 11 percent of stomach cancers worldwide can be attributed to tobacco smoking (Tredaniel et al., 1997). A growing body of evidence supports a lower risk of stomach cancer in former smokers compared with that in current smokers. The risk of stomach cancer decreases with increasing years of cessation, with the risk nearing that for those who have never smoked after approximately 20 years of cessation (Tredaniel et al., 1997).

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection Tobacco Use and Cervical Cancer Worldwide, cervical cancer is one of the most common cancers in women. In the United States, both the incidence of cervical cancer and the rate of mortality from the disease have been declining steadily since the middle of the last century. Since the early 1970s alone, the incidence and mortality for cervical cancer have dropped by approximately 40 percent (Miller et al., 1996), a decrease most likely due to increased screening by Pap tests and the use of effective treatment. Cervical cancer is now known to be caused in large part by chronic infection of the cervix by some sub-types of human papilloma virus (HPV). Cervical cancer is more common in African-American women than white women. Between 1988 and 1992, incidence rates according to the Surveillance, Epidemiology, and End Results (SEER) Program were 13.2/100,000 for African-American women and 8.7/100,000 for white women. Vietnamese women had the highest rates in the United States (43/100,000), and Japanese women had the lowest (5.8/100,000) (Miller et al., 1996). One challenge with studying smoking and cervical cancer is potential confounding by other risk factors linked with low socioeconomic status. In particular, human papillomavirus (HPV) infection and high levels of sexual activity each increase the risk of cervical cancer and are each also more common in smokers than in nonsmokers. Separation of the effect of smoking from these other risk factors is key to uncovering the true relationship between smoking and cervical cancer. The Surgeon General’s report Women and Smoking summarizes the evidence linking smoking with cervical cancer (US DHHS, 2001a). Overall, case-control studies that have not controlled for HPV infection status demonstrate a twofold increase in the risk of cervical cancer in smokers compared with that in those who have never smoked, and the risk increases with the duration of smoking. For women who have smoked for more than 20 years, the risk of cervical cancer is threefold that for women who have never smoked (Daling et al., 1996). Results have been mixed in studies that have controlled for HPV infection status. Some have found that smoking raises the risk of cervical cancer, regardless of HPV infection status (Ylitalo et al., 1999; Daling et al., 1996), and others have found that smoking increases the risk of cervical cancer only among women who are not infected with HPV (Bosch et al., 1992; Eluf-Neto et al., 1994; Munoz et al., 1993). There is evidence of a probable inverse association between smoking cessation and cervical cancer. Taken as a whole, there are good data that former smokers experience a lower risk of cervical cancer than current smokers (US DHHS, 1990). It is unclear how the amount of time since quitting affects the risk of cervical cancer in former smokers.

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection The available evidence suggests that milk and other dairy products may raise the risk of prostate cancer, kidney, and ovarian cancer, but may lower the risk of colon cancer. Milk and Dairy Product Consumption and Prostate Cancer Case-control studies generally support a positive association between the consumption of dairy products and prostate cancer risk, whereas cohort studies are less consistent (World Cancer Research Fund and American Institute for Cancer Research 1997). A higher level of milk consumption was related to an increased risk of prostate cancer in three of the seven cohort studies on dairy products (Snowdon et al., 1984; Le Marchand et al., 1994; Schuurman et al., 1999). In addition, one study suggested an increased risk with higher levels of consumption of butter, margarine, and cheese (Severson et al., 1989). Taken collectively, the evidence suggests the possibility of a moderate positive association between the consumption of dairy products and prostate cancer risk. Although this relationship has been attributed to the high fat contents of many dairy products, recent evidence suggests that calcium may be the responsible agent (Chan et al., 1998). Calcium has been shown to cause a compensatory decrease in circulating levels of 1,25-dihydroxy vitamin D, which may in turn raise the risk of prostate cancer (Giovannucci, 1998). Milk and Dairy Product Consumption and Kidney Cancer Evidence of a possible relationship between the consumption of dairy products and kidney cancer comes entirely from case-control studies (World Cancer Research Fund and American Institute for Cancer Research, 1997). Of nine such studies, seven have shown an increased risk among those with high levels of milk consumption (Mellemgaard et al., 1996; Wolk et al., 1996; World Cancer Research Fund and American Institute for Cancer Research, 1997). Odds ratios tend to be in the range of 1.3 to 1.8, with some studies suggesting even larger effects. Proposed mechanisms involve both animal protein and animal fat (World Cancer Research Fund and American Institute for Cancer Research, 1997). Milk and Dairy Product Consumption and Ovarian Cancer Lactose is the primary sugar found in milk and other dairy products. When lactose is consumed and digested, it is broken down into two components, galactose and glucose, which are then broken down even further. If galactose is not metabolized properly, it can accumulate in the ovaries, affecting ovarian function and hormone levels. This may, in turn, contribute to the development of ovarian tumors. Two large cohort studies, the Nurses’ Health

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection Study and the Iowa Women’s Health Study, have examined the association of lactose consumption on ovarian cancer. Both studies found a modest increase in risk among women who consume large amounts of lactose. NUTRIENTS The link between whole foods and cancer is only one aspect of the diet-cancer relationship, but it is generally the most practical link when it comes to policy decisions and consumer action. Overall, people buy, prepare, and eat whole foods rather than specific nutrients. For this reason, this report focuses on the whole-food aspects of diet and their impact on cancer risk. There is a growing body of evidence that specific macronutrients or micronutrients could explain the associations between foods and cancer risk. Macronutrients Fat, carbohydrates, fiber, and protein have each been linked with cancer in some way (Table 3.8). A number of studies support a possible association between a high total fat intake and an increased risk of lung, colorectal, and prostate cancer (World Cancer Research Fund and American Institute for Cancer Research, 1997). The findings across studies are not consistent, however. Analyses based on prospective data tend to show very little, if any, association between total fat intake and cancer risk when total energy intake and other potential confounders are controlled for. A large combined analysis of breast cancer and dietary fats showed no association across a wide range of fat intakes (Hunter et al., 1996). The hypothesis that reducing intake of fats will reduce breast cancer risk is now being tested in the Women’s Health Initiative, a large intervention study in the United States (http://www.nhlbi.nih.gov/whi/). More plausible is a link between different types of fat and certain cancers. Saturated or animal fat, trans-unsaturated fat, and cholesterol have been found in some studies to increase the risk of cancer, whereas monounsaturated fat has been found in some studies to lower the risk of breast cancer (World Cancer Research Fund and American Institute for Cancer Research, 1997; Chiu et al., 1996; Willett, 1999; Zhang et al., 1999b). The consumption of large amounts of starch has been linked to a possible increase in stomach cancer risk and a possible decrease in colorectal cancer risk (World Cancer Research Fund and American Institute for Cancer Research, 1997). Micronutrients The link between micronutrients and cancer has been studied extensively, and although a number of possible associations have been identified, only a few have enough evidence to qualify as probable associations (Table 3.9).

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection TABLE 3.8 Risk of Incident Cancer Associated with Selected Macronutrients   Association of Type of Cancer for the Following Level of Evidence: Macronutrients Convincing Probable Possible Fat and cholesterol       Total fat     Positive association: • Lung cancer • Colorectal cancer • Prostate cancer Saturated/animal fat     Positive association: • Endometrial cancer • Lung cancer • Colorectal cancer • Prostate cancer • Non-Hodgkin’s lymphoma trans-Unsaturated fat     Positive association: • Non-Hodgkin’s lymphoma Monounsaturated fat     Inverse association: • Breast cancer Cholesterol     Positive association: • Lung cancer • Pancreatic cancer Carbohydrates       Starch     Inverse association: • Colorectal cancer       Positive association: • Stomach cancer Fiber     Inverse association: • Endometrial cancer • Pancreatic cancer • Colorectal cancer Protein     Inverse association: • Breast cancer (survival) Total energy intake     Positive association: • Pancreatic cancer The most well-established link is between folate and colorectal cancer. A number of studies have found that as the level of folate consumption increases, the risk of colorectal cancer (as well as polyps) decreases (Freudenheim et al., 1991; Giovannucci et al., 1993b; Giovannucci et al., 1998b; Willett, 2000). The Nurses’ Health Study found that the consumption of large amounts of folate from fruits and vegetables was sufficient to lower

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection TABLE 3.9 Risk of Incident Cancer Associated with Selected Micronutrients   Association of Type of Cancer for the Following Level of Evidence: Macronutrients Convincing Probable Possible Vitamin A and carotenoids   Inverse association: • Lung cancer Inverse association: • Breast cancer • Esophageal cancer • Stomach cancer • Colorectal cancer • Cervical cancer Folate   Inverse association: • Colorectal cancer Inverse association: • Breast cancer Vitamin D   Inverse association: • Colorectal cancer   Vitamin E     Inverse association: • Lung cancer • Cervical cancer • Colorectal cancer Calcium     Inverse association: • Colorectal cancer       Positive association: • Prostate cancer • Kidney cancer • Ovary cancer Selenium     Inverse association: • Lung cancer • Prostate cancer • Colorectal cancer the colorectal cancer risk and that supplementation with a multivitamin that contained folate offered additional reductions (Giovannucci et al., 1998b). The underlying biological role of folate and its interaction with the MTHFR gene offers a plausible biological pathway for a causal relation between low levels of folate consumption and colon cancer. The association of vitamin A and carotenoids with cancer risk has also been studied extensively, with the strongest evidence supporting a probable inverse relation with breast and lung cancers. Currently, however, it seems at most only a small reduction in the risk of breast cancer is associated with the consumption of large amounts of carotenoids from food sources (Zhang et al., 1999a). In addition, a number of observational studies support an association between the consumption of large amounts of carotenoids from food and a lower risk of lung cancer. Randomized trials of beta-carotene supplements found either no effect or an increased risk of lung cancer (Alpha-Tocopherol Beta Carotene Cancer Prevention Study Group, 1994;

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection Hennekens et al., 1996; Omenn et al., 1996), suggesting that a protective effect of fruits and vegetables is not explained by beta carotene alone. A surprise finding of the ATBC study was the lower prostate cancer risk observed in the vitamin E arm. Whether supplemental vitamin E might reduce prostate cancer risk is now being tested in the SELECT trial (http://www.cancer.gov/newscenter/SELECT). Studies with animals have suggested that the consumption of larger amounts of selenium reduces the risk of various tumors, and ecological studies show an inverse relation between selenium and cancers of the breast and colon (Clark, 1985). A randomized trial to test the ability of a selenium supplement to reduce skin cancer risk showed no effect for skin cancer, but a significant reduction in the rates of mortality from cancers of the lung, colon, and prostate (Clark et al., 1996). Evidence against a large and rapid impact of selenium on cancer risk comes from a fortification intervention that was implemented in Finland. Because of the low selenium content of the soil in Finland (leading to low food selenium levels), selenium was applied with fertilizer in the mid-1980s. Blood selenium levels rose rapidly following this ecological intervention, but there has been no apparent decline in incidence or mortality rates for prostate or colon cancer (Willett, 1999). Whether supplemental selenium might reduce prostate cancer risk is now being tested, along with vitamin E, in the SELECT trial (http://www.cancer.gov/newscenter/SELECT). Glycemic Index The relation between macronutrients in diet and the physiological response to foods is complex. During digestion, the body breaks down dietary glucose into either sugar or starch so that it can be absorbed into the bloodstream. The resulting spike in blood glucose levels leads to a compensatory increase in blood insulin levels. Because insulin allows glucose to be taken up by the tissues, the blood glucose level drops in response to the high insulin levels. Once the blood glucose levels return to normal, insulin is released into the blood at a much lower rate. The impact that a food has on these patterns of circulating blood glucose and insulin is estimated by its glycemic index. Foods with high glycemic indices are digested and absorbed more quickly, causing a rapid influx of glucose into the bloodstream, causing the insulin level to rise rapidly, while foods with a low glycemic index are converted to glucose more slowly and thus lead to a smaller spike in insulin levels. Foods with high glycemic indices may increase the risk of diabetes and coronary heart disease (Salmeron et al., 1997a; Salmeron et al., 1997b; Liu et al., 2000). Whether foods with a high glycemic index also have an impact on the risk of cancer is unclear. If hyperinsulinemia is important in carcinogenesis in the colon, however, then the glycemic load from foods may be an impor-

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection tant risk factor for colon cancer. An increasing number of epidemiological studies have found that the consumption of large amounts of sugar and foods with a high glycemic index is associated with an increased risk of colon cancer (Giovannucci, 1995b). On the basis of this growing evidence relating insulin, glycemic load, and colon cancer risk, further research is needed to determine if the insulin pathway might also contribute to other major malignancies. Disparities in Dietary Intake The proportion of adults consuming the recommended five servings of fruits and vegetables a day has been estimated to be no more than 32 percent (Krebs-Smith et al., 1995a; Krebs-Smith et al., 1995b; Thompson et al., 1999; Li et al., 2000). Although these estimates are clearly low for the entire population, there is also considerable variation across subgroups defined by sex, race or ethnicity, education, and income. The levels of fruit and vegetable consumption are lower among men than among women (Subar et al., 1995; Kamimoto et al., 1999). One study showed that the level of fruit and vegetable consumption was lowest among Hispanics (Thompson et al., 1999), another found higher levels of consumption among whites than African Americans ages 55 and older (Kamimoto et al., 1999), and other studies have found lower levels of consumption of fruits and vegetables among individuals with lower incomes and lower levels of educational attainment (Subar et al., 1995; Thompson et al., 1999). Although the level of consumption of red meat has decreased substantially in the population as a whole (U.S. Department of Agriculture, 1999), lower-income households have experienced less of a reduction than higher-income households (Interagency Board for Nutrition Monitoring and Related Research, 1993). In general, dietary trends in the last 60 years in the United States have shown a complex relationship with social class. Among those with higher socioeconomic status, diets have improved in regards to cancer and heart disease risk, whereas among those at lower socioeconomic classes, diets have worsened in the last half of the 20th century (Popkin et al., 1996). ALCOHOL Alcohol consumption has both beneficial and harmful effects on health. A U-shaped relationship between mortality and consumption of alcohol was described as early as 1926 (Pearl, 1926). A number of studies since then have shown that persons who drink heavily have higher rates of death from certain types of acute death, such as poisoning (Anderson, 1995), injuries (Andreasson et al., 1988; CDC, 1995), violence (Andreasson et al., 1988), and suicide (Andreasson et al., 1988), whereas the long-term use of alcohol increases the risk of death from cirrhosis (Norton et al., 1987),

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection hemorrhagic stroke (Donahue et al., 1986), and certain cancers (International Agency for Research on Cancer, 1988). Alcohol is listed as one of the top identifiable contributors to death in the United States (McGinnis and Foege, 1993), after tobacco and diet and activity patterns; and of all the threats to human health, alcohol probably causes the widest range of diseases and injuries (Rose, 1992). In 1990, approximately 5 percent of all deaths (about 100,000 deaths) were attributable to alcohol consumption, as were 15 percent of potential years of life lost before age 65 (Rose, 1992). The consequences of alcohol consumption extend beyond death rates. Alcohol affects nerve cells in the brain and impairs cognitive function, both temporarily and permanently, over the long term (Meyer, Terayama et al., 1998), and it contributes to the destruction of personal and social relationships (Brookoff et al., 1997). An estimated 18 million people in the United States suffer from alcohol dependence, and an estimated 76 million people are affected by the consequences of alcohol abuse through either their own habit or that of someone close to them (McGinnis and Foege, 1993). Nearly a third of all U.S. adults engage in risky drinking patterns and thus need some kind of intervention, whether it is advice to cut down or referral for further evaluation and treatment for suspected abuse or dependence (Dawson, 2002). In contrast to these harmful effects, however, there is substantial evidence for a beneficial effect of the consumption of moderate amounts of alcohol on the risk of coronary heart disease and thrombotic stroke. The risk of death from these cardiovascular causes is reduced, on average, by about 20 to 40 percent by the consumption of moderate amounts of alcohol (Doll, 1997; Thun et al., 1997). Because cardiovascular disease is the leading cause of death among those of middle and old ages, a reduction in the risk of mortality from cardiovascular disease with moderate alcohol use will translate into a reduction in the total risk of mortality in many populations. Clearly, there is a need to balance the many risks and benefits of alcohol use. The Impact of Alcohol on Cancer Incidence Because alcohol use tends to be associated with cigarette use and other high-risk behaviors, its independent effects on cancer have long been questioned. However, in 1988, on the basis of abundant epidemiological evidence, the International Agency for Research on Cancer concluded that alcohol is, in fact, a Group A carcinogen and an independent risk factor for cancers of the upper aerodigestive tract and liver (World Cancer Research Fund and American Institute for Cancer Research, 1997). Since publication of that report, a large body of evidence has confirmed that alcohol use also increases the risk of breast cancer (Longnecker et al., 1988; Willett et al., 2000) and, probably, colon cancer (Tomeo et al., 1999) (Table 3.10). Although numerous studies have compared the effects of beer, wine,

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection TABLE 3.10 Increase in Risk of Incident Cancer Associated with Alcohol Use   Relative Risk (RR) Level of Evidence Moderate (RR 1.35–1.99) Large (RR >2.0) Convincing Breast       Mouth Pharynx Larynx Esophagus Livera Probable Colon and rectum   aAlthough light and moderate levels of alcohol consumption do not increase the risk of liver cancer, heavy drinking increases the risk substantially. and liquor, the type of alcohol consumed does not appear to influence the cancer risk as much as does the amount consumed (World Cancer Research Fund and American Institute for Cancer Research, 1997). For cancers of the upper aerodigestive tract, breast, colon, and rectum, there is a dose-response relationship, such that even low or moderate levels of consumption increase risk slightly. For liver cancer, the most important factor is heavy and persistent use, such as that defined by alcoholism, which can cause chronic liver injury. Alcohol Consumption and Cancers of the Mouth, Pharynx, Larynx, and Esophagus There is convincing epidemiological evidence from both cohort and case-control studies that alcohol consumption increases the risk of cancers of the mouth, pharynx, larynx, and esophagus (World Cancer Research Fund and American Institute for Cancer Research, 1997; Kjaerheim et al., 1998). All of the cohort studies that have examined this association, including those that controlled for smoking, have demonstrated a substantially increased risk. In addition, a dose-response relationship has been observed in the majority of case-control studies, including most of those that controlled for smoking. Alcohol also appears to act synergistically with tobacco in causing these cancers (World Cancer Research Fund and American Institute for Cancer Research, 1997). Several mechanisms have been proposed to explain the carcinogenic effect of alcohol on cancers of the upper aerodigestive tracts, although none are proven (World Cancer Research Fund and American Institute for Cancer Research, 1997). Alcohol may directly influence carcinogenesis in the cells that it touches; it may alter the permeabilities of cell membranes or modify cellular metabolism, thus making these cells more vulnerable to chemical carcinogens;

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection it may contain toxic substances, such as nitrosamines and polycyclic aromatic hydrocarbons, that alter DNA; and alcohol may in other ways compromise nutritional status and increase susceptibility to cancer (World Cancer Research Fund and American Institute for Cancer Research, 1997). Alcohol Consumption and Liver Cancer More than 40 studies have evaluated the relationship between high levels of alcohol intake and liver cancer. Although these studies did not use a standard definition of heavy drinking or even the same reference groups (nondrinkers versus the entire population), a positive association has been observed across a variety of study designs and populations (World Cancer Research Fund and American Institute for Cancer Research, 1997). Unlike most other cancers, liver cancer is associated only with the heavy persistent type of drinking characteristic of alcoholism, with its related chronic liver damage. Neither light nor moderate drinking increases liver cancer risk. Alcohol Consumption and Breast Cancer More than 25 studies have shown that alcohol increases breast cancer risk, most likely by raising the level of estrogen in the bloodstream or making the breast more vulnerable to carcinogens (Smith-Warner et al., 1998). A randomized feeding trial indicated that estrogen levels are increased when women consume even low levels of alcohol on a regular basis (Reichman et al., 1993). The dose-response relationship between alcohol and breast cancer risk is best demonstrated in results from an analysis of data pooled from six prospective studies (Smith-Warner et al., 1998). In that analysis, women’s risk of breast cancer rose by 7 percent for every 10-g increase in daily alcohol consumption. The association did not differ substantially by type of alcoholic beverage consumed. Because breast tissue may be particularly vulnerable during adolescence and early adulthood, researchers have speculated that alcohol consumption during these time periods might be more harmful than alcohol consumption later in life (Colditz and Frazier, 1995), but to date, studies on this topic have been inconsistent (Harvey et al., 1987; Young, 1989; Garland et al., 1999). Alcohol Consumption and Colorectal Cancer Although the results are not entirely consistent, the majority of studies support an association between alcohol intake and an increased risk of colorectal cancer in both men and women (Giovannucci and Willett, 1994). Moreover, alcohol consumption is related to an increased risk of colorectal

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection adenomas (Giovannucci et al., 1993b; Kahn et al., 1998). A dose-response relationship has been observed across several studies, and some evidence suggests that even moderate drinkers (those who consume one drink a day) are at higher risk of colorectal cancer than nondrinkers (World Cancer Research Fund and American Institute for Cancer Research, 1997). The effects of alcohol may be exacerbated by low levels of folate (Freudenheim et al., 1991). In the Health Professionals Follow-Up Study, an increased risk of colon cancer was observed only among those current and past drinkers who consumed lower levels of methionine or folate in their diets (Giovannucci et al., 1995b). Because alcohol is an antagonist of folate-related methyl group metabolism, it may cause an imbalance in DNA methylation, which may then contribute to colorectal carcinogenesis. Alcohol Consumption and Lung Cancer Although a number of cohort studies have examined the association between alcohol consumption and lung cancer, few have adequately adjusted for the effects of smoking. Those that have adjusted adequately have yielded inconsistent findings. Four cohort studies have shown a positive association (Klatsky et al., 1981; Kvåle et al., 1983; Pollack et al., 1984; Potter et al., 1992), and four have yielded null results (Gordon and Kannel, 1984; Bandera et al., 1997; Woodson et al., 1999; Breslow et al., 2000). Disparities in Alcohol Use Data on alcohol consumption in the United States indicate that the top 20 percent of drinkers consume more than 85 percent of all alcohol consumed (Greenfield and Rogers, 1999). Men and young adults ages 18 to 29 years are disproportionately represented among those with the highest alcohol consumption levels. The National Health Interview Survey (1998) indicates that at each age, the proportion of current drinkers is highest among non-Hispanic white men, followed by African-American men, and then Hispanic men (National Center for Health Statistics, 2000). Among women, the proportion of current drinkers is also highest among non-Hispanic white women at each age. SUMMARY AND CONCLUSIONS There is overwhelming evidence that lifestyle factors affect cancer risk. As detailed throughout this chapter, current epidemiological evidence links the major behavioral risk factors with several major cancers. Tobacco use causes cancers of the lung, oropharynx, larynx, esophagus, bladder, kidney, and pancreas and contributes to the risk of leukemia as well as cancers of the

OCR for page 41
Fulfilling the Potential of Cancer Prevention and Early Detection colon, stomach, cervix, and liver. Recent declines in the rates of lung cancer in the United States among men—first among those who are younger and then among those who are older successively over time—and among women ages 40 to 59 reflect past declines in the use of cigarettes (Wingo et al., 1999). The cessation of tobacco use reduces the excess risk of nearly all of these cancers over time. Obesity increases the risk of breast, endometrial, colorectal, kidney, and esophageal cancer. Regular physical activity lowers the risk of cancers of the colon and breast cancer, and, possibly, endometrial cancer as well. Regular alcohol intake increases the risk of cancers of the oropharynx, larynx, esophagus, breast, liver, colon, and rectum. Diet has been linked to several cancers, with the most consistent evidence linking the consumption of large amounts fruits and vegetables with a lower risk cancer. Numerous researchers have also estimated that positive changes in these lifestyle factors can reduce a good proportion of the national cancer burden (see Chapter 2). Although the specific methods and the results of these analyses vary, they are all remarkably consistent in pointing to the potential benefits of reducing tobacco use, improving diet, increasing physical activity, maintaining a healthy body weight, keeping alcohol consumption at low to moderate levels, and getting screened regularly for cancer.