Part 96

[Surgeon General’s Smoking Report,  health reports Continued from Part  95 ]

2004 The Health Consequences of Smoking: A Report of the Surgeon General.  Office of the Surgeon General (US); Office on Smoking and Health (US). Atlanta (GA): Centers for Disease Control and Prevention (US);  Forty years of reports. Cancers of stomach, uterine cervix, pancreas, kidney, acute myeloid leukemia, pneumonia, aneurysm, cataract, periodontitis. Conclusion “smoking generally diminishes the  health of smokers.”

https://www.ncbi.nlm.nih.gov/books/NBK44698/

Part 96 The Health Consequences of Smoking: A Report of the Surgeon General. 2004. Cancer. Breast. A word of wisdom..

2 Cancer

Breast Cancer

Breast cancer is the most frequently diagnosed nonskin cancer among women (ACS 2003). In 2003, an estimated 212,600 new cases and 40,200 deaths attributed to breast cancer were expected to occur. From 1996–2000, the average annual age-adjusted population incidence rate of breast cancer per 100,000 in the United States was 140.8 in white women, 121.7 in black women, 97.2 in Asian/Pacific Islander women, 89.8 in Hispanic women, and 58.0 in American Indian/Alaska Native women (Ries et al. 2003). The possibility that cigarette smoking is associated with breast cancer has been a topic of substantial research, given the high prevalence of exposure to this harmful agent, the high incidence of breast cancer, and the relative difficulty of modifying many established breast cancer risk factors.

The relationship between active smoking and breast cancer has been investigated since 1960 (MacMahon and Feinleib 1960) in many large, well-designed epidemiologic studies (Palmer and Rosenberg 1993; Terry and Rohan 2002). Most of these studies have found overall associations close to the null: some RRs for the association with smoking have been modestly inverse, whereas some have been modestly positive. Investigators have hypothesized that smoking may have antiestrogenic effects as well as carcinogenic effects on breast tissue, and thus may exert countervailing influences on breast cancer risks (Palmer and Rosenberg 1993). If both of these effects have a role in breast cancer development, the increase in risk may become apparent only when women are classified according to characteristics related to their susceptibility to the antiestrogenic or carcinogenic effects. In the absence of such stratification, the hypothesized effects of cigarette smoke might be expected to lead to null findings overall in a single study and to inconsistency across studies, depending on the characteristics of the participants.

Conclusions of Previous Surgeon General’s Reports

The 2001 Surgeon General’s report on women and smoking (USDHHS 2001) reviewed the scientific data on the association between cigarette smoking and breast cancer, concluding that “Thus, active smoking does not appear to appreciably affect breast cancer risk overall. However, several issues were not entirely resolved, including whether starting to smoke at an early age increases risk, whether certain subgroups defined by genetic polymorphisms are differentially affected by smoking, and whether ETS2 exposure affects risk” (p. 217). A more detailed review of the evidence is provided in this section, including evidence on the above three points. Since the 2001 report, IARC has concluded that the evidence is indicative of no association between smoking and breast cancer (IARC 2002).

Biologic Basis

Because smokers have a higher incidence of cancers at sites that do not have direct contact with cigarette smoke, including the cervix, pancreas, and bladder (USDHHS 1982), researchers have hypothesized that constituents of cigarette smoke may reach distant tissues, including breast tissue. Biomarkers have now provided evidence supporting this hypothesis. Mutagens from cigarette smoke have been found in the nipple aspirates of nonlactating women (Petrakis et al. 1980), indicating that mutagenic tobacco smoke components do reach breast tissue. Thus, prolonged exposure to these substances may initiate and promote benign and malignant breast disease. In a small case-only study, Perera and colleagues (1995) found DNA adducts characteristic of cigarette smoke in four out of seven breast tumors from smoking women, but not in any of the tumors from eight nonsmokers. In a larger case-only study, Li and colleagues (1996) similarly found such adducts in breast tissues of all current smokers (17 out of 17) and in some (5 out of 8) former smokers, even 18 years after smoking cessation. They found the same adducts in 4 out of 52 nonsmokers. The data from former smokers suggest that smoking-induced DNA damage might persist for a long time.

Whereas the research described above suggests that breast tissue of smokers is exposed to tobacco-smoke carcinogens, some researchers (MacMahon et al. 1982) have proposed that smokers would have a reduced risk of breast cancer, based on a hypothesis that breast cancer is an estrogen-related disease and that cigarette smoking has antiestrogenic effects. However, the biologic foundations underlying both of the postulated mechanisms of this hypothesis (carcinogenic exposure and antiestrogenic effects) are not firmly established.

Empirical support for the hypothesis that cigarette smoking exerts antiestrogenic effects and therefore might lower the risk for breast cancer comes from several sources, including laboratory studies of rodents and studies of hormones in smokers and nonsmokers. Rats exposed to cigarette smoke develop fewer mammary tumors than do unexposed rats (Davis et al. 1975; Dalbey et al. 1980), although this finding may be the result of differences in weight or survival. Findings from this animal model also are interpreted in light of the uncertain relevance of the mammary tumor model in rodents for breast cancer in humans. For instance, mammary cancer in rats is prolactin-dependent (Kleinberg 1987), and the lower risk of tumors may reflect a lowering of prolactin levels from long-term exposure to tobacco smoke (Ferry et al. 1974; Andersson 1985).

Smoking has also been hypothesized as affecting estrogen levels. Researchers are uncertain about how smoking might affect the biology of estrogen-related events in women not taking oral estrogens. However, several possible mechanisms have been proposed. Polycyclic aromatic hydrocarbons in tobacco smoke may induce cytochrome P-450 enzymes that metabolize sex hormones (Conney 1967; Lu et al. 1972). Michnovicz and colleagues (1986)suggested that smoking increases the 2-hydroxylation of the estradiol metabolic pathway, thus decreasing the availability of active estrogens to tissues. Cigarette smoking leads to an early menopause, and disturbances in estrogen-dependent processes before menopause could be due to a toxic impact on the developing graafian follicle (Mattison 1980). Also, the lower body weight of smokers would result in lower estrone and estradiol levels than nonsmokers of similar age. Finally, smoking increases the levels of the adrenal androgen hormones androstenedione and dihydroepiandrosterone (Baron et al. 1990; Law et al. 1997), which could explain some (but hardly all) of the hormone effects.

Whereas initial comparisons of estrogen levels between smokers and nonsmokers documented differences, more recent studies have generally shown similar levels. Among premenopausal women, studies of urinary excretion of estrogens have tended to yield different findings from studies of plasma levels of reproductive hormones. MacMahon and colleagues (1982)were among the first to examine estrogens and smoking, and reported that premenopausal women who smoked had lower urinary excretions of estrone, estriol, and estradiol during the luteal phase of the menstrual cycle than women who had never smoked. Former smokers did not manifest this pattern, however, nor were there differences in urinary excretion during the follicular phase of the menstrual cycle. Michnovicz and colleagues (1986) found results similar to those of MacMahon and colleagues for both the luteal and follicular phases. In another study of premenopausal women, Westhoff and colleagues (1996) found that smokers had, on average, lower levels of midcycle and luteal-phase urinary estradiol levels than nonsmokers.

However, comparisons of endogenous serum estrogen levels between smokers and nonsmokers have clearly shown that among both premenopausal and postmenopausal women smokers do not have lower levels of the major estrogens than nonsmokers (Baron et al. 1990; Law et al. 1997; USDHHS 2001). Three studies of premenopausal women (Longcope and Johnston 1988; Key et al. 1991; Thomas et al. 1993) found no differences in plasma concentrations of reproductive hormones between smokers and nonsmokers. Although the study conducted by Thomas and colleagues (1993) consisted of a small number of women (26 smokers, 24 nonsmokers), it was more detailed than other similar studies. These researchers took multiple blood samples from participants over the course of a menstrual cycle, equally timed from the date of the previous cycle, and also examined the effects of smoking on luteinizing hormone pulsatility, enabling them to explore possible differences in the length of the follicular and luteal phases between smokers and nonsmokers. Thomas and colleagues (1993) concluded that smoking did not result in major alterations in cyclicity; secretion of gonadotropins, estradiol, and progesterone; metabolism of estradiol; or secretion of androgens. They noted that these data confirm those of Longcope and Johnston (1988) and Key and colleagues (1991), suggesting that the antiestrogenic properties of cigarette smoking act through mechanisms other than alterations in hormone levels.

Several studies have examined hormone levels in postmenopausal women (Friedman et al. 1987; Trichopoulos et al. 1987a; Khaw et al. 1988; Longcope and Johnston 1988; Kabat et al. 1997). Again, some studies measured hormone levels in urine; others measured levels in plasma. None found lower levels of circulating estrogens among women who smoked compared with women who did not smoke. It is possible that a failure to detect differences in estrogen levels between smoking and nonsmoking women who are postmenopausal could be due to limitations in measurement, because estrogen levels in postmenopausal women are often at the limits of detection. Differences in postmenopausal estrogen levels between smokers and nonsmokers could be due, at least in part, to body fat levels. Smokers tend to be leaner than nonsmokers, and in postmenopausal women, an important source of estrogen is the peripheral conversion of androgen precursors that occurs in fat cells.

The interpretation of differences in estrogen levels between smokers and nonsmokers, and relating them to differences in the risk of breast cancer, is complex because the effects of specific estrogens likely vary by organ site, and smoking may affect only specific estrogens (Rohan and Baron 1989). For example, Michnovicz and colleagues (1986) proposed that smoking may shift the metabolism of estrone and estradiol toward the production of catechol estrogens. This shift would leave estrogen and estradiol concentrations unchanged, but would increase catechol estrogen production at the expense of estriol. If the breast were equally sensitive to estriol and catechol estrogens, this change would not affect breast cancer risk, although it would affect organs that react differently to estriol and catechol estrogens. The estrogenic hormone dependence of breast cancer is not well defined. It is clear, however, that the estrogen dependence of breast cancer is not as marked as that of endometrial cancer, and any antiestrogenic effects of smoking might be unimportant with respect to this weaker estrogen-related disease (Rohan and Baron 1989).

Epidemiologic Evidence

This section discusses all studies of active and passive smoking in relation to breast cancer that were considered in a 1993 epidemiologic review (Palmer and Rosenberg 1993), and any additional epidemiologic studies on this topic published from September 1992 to the end of 1999, identified through a MEDLINE search. Several additional relevant reports beyond this inclusive review are also cited. A review of the observational epidemiologic literature was then used to identify articles in the fields of biology, pathology, and endocrinology that examined the biologic basis for potential positive and negative causal links between exposure to cigarette smoking and breast carcinogenesis.

Cigarette Smoking and Breast Cancer Risk

Palmer and Rosenberg (1993) reviewed all of the studies on smoking and breast cancer published in the scientific literature before September 1992 (Tables 2.36, 2.37, 2.38, and 2.39). They excluded studies of prevalent breast cancer, studies providing insufficient methodologic detail (e.g., those lacking CIs or definitions of the reference categories [all of the studies excluded for this reason had fewer than 300 cases]), and case-control studies in which patients with smoking-related diagnoses were included in the control series. These studies, with likely overestimates of the prevalence of smoking in the general population represented by the control groups, would have found spuriously reduced RR estimates if smoking truly did increase the risk for breast cancer. For each of the 19 studies deemed informative, Palmer and Rosenberg (1993) provided detailed qualitative summaries in the four tables in their review, noting where the data were available in individual studies, RR estimates for former and current smokers overall stratified by age at commencement of smoking, and for the highest categories of smoking intensity or duration.

Table 2.36

Case-control studies on the association between smoking and the risk of breast cancer that used hospital or cancer registry controls. 

Table 2.37

Case-control studies on the association between smoking and the risk of breast cancer that used healthy controls drawn from population sources. 

Table 2.38

Case-control studies on the association between smoking and the risk of breast cancer conducted among screening program participants. 

Table 2.39

Cohort studies on the association between smoking and the risk of breast cancer. 

In four case-control studies included in this review (Rosenberg et al. 1984; Baron et al. 1986; Stockwell and Lyman 1987; Palmer et al. 1991), controls were selected from among hospital patients or cancer registry patients, and only patients with conditions judged to be unrelated to cigarette smoking were included (Table 2.36). All of these studies were large (all had more than 1,700 cases; one [Stockwell and Lyman 1987] had more than 5,000 cases), and controlled for many of the known risk factors for breast cancer including age at menarche, age at birth of first child, and parity. Two of the four studies also controlled for alcohol consumption, obesity, menopausal status, and other potential confounding factors as they are risk factors for breast cancer and are associated with smoking (Rosenberg et al. 1984; Palmer et al. 1991). Relative risk estimates for the heaviest current smoking categories (i.e., one or more packs per day) were close to 1.0, ranging from 0.93 to 1.3. None of these four studies showed a dose-response gradient of risk with the number of cigarettes smoked per day.

In seven other case-control studies (O’Connell et al. 1987; Adami et al. 1988; Rohan and Baron 1989; Chu et al. 1990; Ewertz 1990, 1992; Palmer et al. 1991; Field et al. 1992), the general community was used as a source of controls (Table 2.37). All of these studies controlled for major reproductive risk factors; some also controlled for alcohol consumption and obesity. The estimated RR for heavy smoking was 0.57 in the smallest study (O’Connell et al. 1987); in the other studies, estimates ranged from 0.75 to 1.59, with no evidence of dose-response relationships.

Three studies of screened populations (Brinton et al. 1986; Meara et al. 1989; Schechter et al. 1989) compared women with incident cases (detected after the first screening) of breast cancer with women who were screened the same number of times without any detection of breast cancer (Table 2.38). All of the studies adjusted for reproductive risk factors and obesity, and one study (Meara et al. 1989) also adjusted for alcohol consumption. These studies generally found ORs between 1.2 and 1.3 for heavy smokers and long-term smokers, compared with women who had never smoked. Meara and colleagues (1989) found higher ORs but CIs were wide.

All five cohort studies (Table 2.39) (Hiatt and Fireman 1986; Hiatt et al. 1988; London et al. 1989; Schatzkin et al. 1989; Vatten and Kvinnsland 1990) controlled for obesity and alcohol consumption in addition to reproductive factors. Relative risk estimates for the heaviest current smoking categories ranged from 0.86 to 1.19. The largest study (London et al. 1989), which assessed repeated measures of smoking during follow-up, found that the RRcomparing those currently smoking 25 or more cigarettes per day with women who had never smoked was 1.02.

Palmer and Rosenberg (1993) concluded their 1993 review by stating that the existing body of epidemiologic evidence neither supported the hypothesis that cigarette smoking has a net effect of reducing the risk of breast cancer nor supported the hypothesis that cigarette smoking increases the risk of breast cancer, even among specific subgroups of women who might be assumed to be at an especially high risk from the carcinogenic effects of smoking, such as heavy smokers who began smoking as teenagers.

Since 1993, additional large, well-designed case-control studies of smoking and breast cancer (Table 2.40) have provided detailed analyses of the amount smoked, duration of smoking, and (in two of the three studies) years since smoking cessation. The largest study (Baron et al. 1996) is a population-based, case-control study with 6,888 cases and 9,529 controls from Maine, Massachusetts, New Hampshire, and Wisconsin, conducted from 1988–1991. This study investigated the effects of smoking among women at very high levels of exposure: heavy smokers, long-term smokers, and those who began smoking very early in life. The current understanding of the processes of breast cell development and differentiation has led some scientists to hypothesize that the timing of exposure to tobacco smoke relative to the stage of breast tissue development may be an important determinant of susceptibility to the carcinogenic effects of smoking. Exposure at very young ages and before a first pregnancy may more strongly increase the risk of breast cancer than exposure at older ages, because breast cells are undifferentiated before pregnancy and are therefore believed to be more susceptible to mutagenesis.

Table 2.40

Large case-control studies on the association between smoking and the risk of breast cancer published after 1993. 

In this large study, the number of cigarettes usually smoked per day was not related to risk for breast cancer. Very heavy smokers (those who smoked >2 packs per day) were not at a higher risk than lifetime nonsmokers; the OR was 1.09 (95 percent CI, 0.79–1.49). Duration of smoking was also unassociated with risk; among women who had smoked cigarettes for more than 50 years compared with women who had never smoked, the OR was 1.07 (95 percent CI, 0.84–1.37). Risk of breast cancer was also not related to the duration of smoking among heavy smokers (>2 packs per day), to the average amount smoked per day among long-term smokers (>20 years), or to pack-years of smoking. There was no overall relationship between age at initiation of smoking and risk of breast cancer. Women who began smoking at an early age (before 15 years of age) were not at an increased risk compared with women who had never smoked; the OR was 1.13 (95 percent CI, 0.97–1.31). This finding was true even among women who began smoking at an early age and who usually smoked more than 20 cigarettes per day (OR = 1.04 [95 percent CI, 0.81–1.33]). No evidence was found of an effect of smoking within subgroups of the study population. The ORs for current and former smokers within high- and low-risk strata for the various covariates, including menopausal status, family history status, history of benign breast disease, and alcohol intake, were all close to 1.0. Thus, in this large population-based study, the researchers found little evidence that cigarette smoking either increases or decreases the risk for breast cancer. Neither early age at smoking initiation, heavy smoking, nor long-term smoking demonstrated an association with an altered risk. This study had several important methodologic strengths that enhanced the validity of the findings. First, the large sample size permitted estimates of the effects of higher exposures with considerable precision. Second, the population-based design of the study, together with a high response rate (>80 percent for both cases and controls), made major response biases unlikely. Finally, substantial confounding of the findings is unlikely, because the RR estimates presented by Baron and colleagues (1996)were adjusted for the main known breast cancer risk factors, with little change over those adjusted only for the matching factors of age and geographic area.

In 1998, Gammon and colleagues published results from another large population-based, case-control study of women under the age of 55 years. This study consisted of 2,199 cases and 2,009 controls surveyed during 1990–1992 from central New Jersey; Seattle, Washington; and Atlanta, Georgia. The objective was similar to that of Baron and colleagues (1996): to examine the effects of smoking on the risk for breast cancer among women at extreme exposure levels, those who were heavy smokers as teenagers or those who were long-term smokers. Similar to Baron and colleagues, Gammon and colleagues (1998) found little evidence for increased breast cancer risk associated with smoking in their large study. Risk was significantly reduced among current smokers who reported smoking for more than 21 years (OR = 0.70 [95 percent CI, 0.52–0.94]), compared with women who had never smoked. Risk was also reduced for women who began smoking at 15 years of age and younger among both current smokers (OR = 0.59 [95 percent CI, 0.41– 0.85]) and former smokers (OR = 0.76 [95 percent CI, 0.50–1.15]). Gammon and colleagues found no significant effect modification by selected hormone-related characteristics including menopausal status, oral contraceptive use, hormone replacement therapy use, body size as an adult, and usual alcohol consumption. They also found no significant heterogeneity in breast cancer risk in relation to the age at beginning smoking.

In a national case-control study of breast cancer in the United Kingdom conducted among young women aged 35 years and younger, Smith and colleagues (1994) found no effects of cigarette smoking on the risk for breast cancer. The RR comparing women who had smoked for 10 or more years with women who had never smoked was 1.0 (95 percent CI, 0.79–1.25), whereas the RR comparing women who had started smoking at 16 years of age or younger was 1.11 (95 percent CI, 0.87–1.43).

The most recent combined analyses on smoking and breast cancer were reported in 2002 by the Collaborative Group on Hormonal Factors in Breast Cancer (2002). Data were analyzed at the individual level from 53 studies, including 58,515 cases and 95,067 controls; information on both tobacco and alcohol was included in all of these studies. The analysis of the risk associated with smoking was limited to the 22,255 cases and 40,832 controls who reported drinking no alcohol. Compared with lifetime nonsmokers, the pooled RR for breast cancer was 0.99 for current smokers and 1.03 for former smokers. Only one study found a significantly increased risk (Figure 2.7).

Figure 2.7

Results on tobacco consumption and breast cancer in women who reported drinking no alcohol. *SE = Standard error. †CI = Confidence interval.

In conclusion, hypotheses that women with higher levels of exposure to cigarette smoking (i.e., heavy smokers and those who have been smoking since an early age) would have elevated risks of breast cancer have not been supported by data from large studies. The weight of the epidemiologic evidence supports the conclusion that smoking is not associated with breast cancer risk. This null relationship is consistent with the two hypothesized mechanisms, antiestrogenic effects and carcinogenic exposures, that imply countervailing consequences of smoking that both increase and decrease the risk for breast cancer.

Genotype-Smoking Interactions

Recent advances in molecular biology and genetics, in terms of both scientific understanding of and technological applications to large populations, have enabled epidemiologists to examine the relationship between smoking and breast cancer in subgroups of women hypothesized to differ with respect to genetic susceptibility to the carcinogenic or antiestrogenic effects of cigarette smoke. Some of the genes involved in the metabolism of carcinogens play a role in the risks for various human cancers, including breast cancer, and reviews of the growing literature on these genes, known as metabolic susceptibility genes, have been published (Idle et al. 1992; Daly et al. 1994; Hirvonen 1995; Raunio et al. 1995; Rothman 1995; Vineis 1995). By definition, these genes function only in the context of interactions with the environment, because the substrates of their gene products are xenobiotic chemicals (foreign to the biologic system) or their metabolites (Garte et al. 1997).

Cigarette smoking results in exposure to aryl aromatic amine carcinogens that are metabolized and detoxified by the cytochrome P-4501A2 (CYP1A2) and NAT1 and NAT2 genes. The NAT2 gene has four major alleles (Lin et al. 1993; Hunter et al. 1997). Persons who are homozygous for any combination of the three slow acetylator alleles have a slow acetylation phenotype (slow acetylators), whereas those who have at least one copy of the rapid acetylator allele have a rapid acetylation phenotype (rapid acetylators) (Lin et al. 1993; Hunter et al. 1997). Women who are rapid acetylators are hypothesized to be less vulnerable to potential carcinogenic effects on the breast from smoking than women who are slow acetylators, because members of the former group more rapidly metabolize or “clear” the toxic agents from their tissues. Approximately 50 percent of whites and a lower proportion of African Americans inherit a polymorphism in the NAT2 gene that leads to decreased acetylator activity (i.e., NAT2 -”slow” genotype) (Bell et al. 1993; Lin 1996). The NAT1 enzyme participates in N-acetylation of a variety of carcinogenic arylamines, as does the NAT2 enzyme. However, the link between NAT1alleles and enzyme function has not been directly established, and investigations are ongoing to determine the functional importance of NAT1gene variants (Deitz et al. 1997; Grant et al. 1997; Hughes et al. 1998; Millikan et al. 1998).

In a case-control study of 304 cases and 327 controls, Ambrosone and colleagues (1996) found that among premenopausal women, being a slow acetylator did not strengthen the effect of smoking on the risk for breast cancer. In fact, risk associated with smoking increased more sharply among rapid acetylators than among slow acetylators, although all ORs were imprecise. Among postmenopausal women, Ambrosone and colleagues (1996) found an association between smoking and breast cancer risk only among women with the NAT2-slow genotype. Among women who were slow acetylators, those in the highest category of number of cigarettes smoked per day (>20) were at an increased risk for breast cancer (OR = 4.4 [95 percent CI, 1.3–14.8]), but there were only 11 cases and 5 controls in this high-exposure stratum. The response rates among cases and controls were low, raising concerns about selection biases with regard to smoking status. These methodologic problems may explain, in part, why the finding of an interaction between smoking and slow acetylator genotype has not been replicated in subsequent larger studies. Results from a case-control study nested within the Nurses Health Study cohort with 466 incident cases and 466 matched controls (Hunter et al. 1997) suggest that current smoking was associated with a slight increase in the risk for breast cancer among women with the NAT2 slow genotype, but this same slight increase was also observed among women with the rapid acetylator genotype. The OR comparing currently smoking women with the slow acetylator genotype to women with the rapid acetylator genotype who had never smoked was 1.4 (95 percent CI, 0.7–2.6); the OR comparing currently smoking women with the rapid acetylator genotype to women who had never smoked with this same “low risk” genotype was 1.2, thus providing no evidence of a genotype-smoking interaction.

To examine the specific hypothesis that smoking before a first pregnancy is an especially strong risk factor for breast cancer, Hunter and colleagues (1997)limited analyses to parous women with complete information on early-life smoking. Women with the rapid acetylator genotype who ever smoked before their first pregnancy were at an increased risk relative to women with the rapid acetylator genotype who had never smoked (OR = 1.7 [95 percent CI, 1.0–2.6]), but there was no dose-response relationship with the duration of smoking before a first pregnancy. Similarly, among women with the slow acetylator genotype, there was an increased risk for breast cancer among women who had smoked for one to five years before their first pregnancy (OR = 2.0 [95 percent CI, 1.1–3.8]), relative to the reference group of women with the rapid acetylator genotype who had never smoked, but the risk of breast cancer was not increased among women who had smoked for five or more years before their first pregnancy (OR = 0.9 [95 percent CI, 0.6–1.5]). Again, there was no evidence for a genotype-smoking interaction in this analysis.

The Carolina Breast Cancer Study, a population-based case-control study of breast cancer among white and African American women living in North Carolina, found no main effect of smoking (OR = 1.0 for current smokers [95 percent CI, 0.7–1.4], and OR = 1.3 for former smokers [95 percent CI, 0.9–1.8], both relative to lifetime nonsmokers) (Millikan et al. 1998). These results were not modified by the presence of either the NAT2 or the NAT1gene. Among postmenopausal women, those who had smoked within the past three years and had the NAT1*10 genotype had an OR of 9.0 (95 percent CI, 1.9–41.8) and those with the NAT2 rapid genotype had an OR of 2.8 (95 percent CI, 0.4– 8.0) compared with nonsmokers.

Other research into potential gene-environment interactions has considered genes related to polycyclic aromatic hydrocarbons, which are carcinogens found in cigarette smoke. The CYP1A1 gene product is involved in the metabolism of these hydrocarbons and is polymorphic, although the exact functional importance of the polymorphisms is unclear (Cosma et al. 1993; Kawajiri et al. 1993; Crofts et al. 1994; Landi et al. 1994; Wedlund et al. 1994; Jacquet et al. 1996; Zhang et al. 1996; Persson et al. 1997; Ishibe et al. 1998). Studies of potential gene-environment interactions have been small and results have been inconsistent. Ambrosone and colleagues (1995) found an interaction between smoking and the CYP1A1 genotype only among light smokers (for whom the OR comparing the high-risk to low-risk genotype was 5.22 [95 percent CI, 1.16–23.56]); however, among heavy smokers, the high-risk genotype was not associated with an increased risk (OR = 0.86 [95 percent CI, 0.24–3.09]). This somewhat contradictory finding (that no increased risk was found in the subgroup of heavy smokers, despite an increase among light smokers) was based on a small number of cases and noncases in the relevant strata; for instance, the OR of 5.22 was based on only seven cases and three controls in the high-risk genotype stratum.

To date, the largest study of the CYP1A1 genotype, smoking, and a risk for breast cancer was conducted among 900 women (cases and controls combined) nested within the Nurses Health Study cohort (Ishibe et al. 1998). In this study, current smokers with a high-risk variant at the MspI nucleotide had an OR of 7.36 (95 percent CI, 1.39–39.0) relative to lifetime nonsmokers with a low-risk variant; the corresponding OR for a variant at the exon 7 nucleotide was 1.51 (95 percent CI, 0.55–4.13). The OR of 7.36 was based on nine cases and two controls in the high-risk stratum. On the basis of the low prevalences of the high-risk genotypes in CYP1A1, Ishibe and colleagues (1998) estimated that only 2.5 percent of breast cancer cases that occurred in the Nurses Health Study cohort over a five-year period could be attributed to the combination of cigarette smoking and a high-risk genotype.

The gene GSTM1 is also involved in the metabolism of carcinogens, including polycyclic aromatic hydrocarbons (Mannervik and Danielson 1988; Nebert 1991). Ambrosone and colleagues (1995) found that the null effect of cigarette smoking was not modified by the high-risk GSTM1 genotype.

Scientists are continuing to pursue research into how genetic factors might interact with cigarette smoking to determine a risk for breast cancer, but so far few clear patterns have emerged. Currently, it is not possible to differentiate subgroups of women who are genetically “susceptible” to the carcinogenic effects of cigarette smoking from those women who are not.

Brunet and colleagues (1998) have pursued a different line of genetic research, speculating that the antiestrogenic effects of smoking might be especially potent in women at very high risk of breast cancer; that is, those who carry mutations in the BRCA1 or BRCA2 gene. It has been estimated that the risk for breast cancer associated with mutations in either gene exceeds 80 percent by the time a carrier reaches 70 years of age (Easton et al. 1995; Tonin et al. 1995), although some researchers have estimated the risk to be lower (Struewing et al. 1997). Some factors that are believed to influence penetrance (i.e., frequency of expression of a genotype) include parity (Narod et al. 1995) and, with respect to the BRCA2 gene, the position of the mutation (Gayther et al. 1997). Brunet and colleagues (1998) speculated that cigarette smoking, because of its hypothesized antiestrogenic effects, also may be associated with a lower penetrance. In their case-control study of women in Canada who were carriers of BRCA1 or BRCA2 gene mutations (186 cases, 186 controls), the risk of breast cancer in smokers was about half of that in nonsmokers. The reduction in risk associated with smoking was significant for a carrier of BRCA1 mutations who had smoked the equivalent of four or more pack-years in her life (OR = 0.47 [95 percent CI, 0.26–0.86]). For BRCA2 gene carriers the magnitude of reduction was somewhat greater (OR = 0.39 [95 percent CI, 0.10–1.49]). There was evidence of a dose-response trend: the degree of breast cancer protection associated with cigarette smoking increased with the number of pack-years smoked. The OR was 0.65 for women with four or fewer pack-years of smoking and 0.46 for those with more than four pack-years of smoking.

Contrasting findings were reported by Couch and colleagues (2001) who carried out a retrospective cohort study of women from high-risk breast cancer families. Of the sisters and daughters in the families, those who had smoked had an increased risk of breast cancer compared with those who had never smoked (RR = 2.4 [95 percent CI, 1.2–5.1]). These studies differ substantially in design, and the case-control approach of Brunet and colleagues (1998) is subject to several potential sources of bias (Baron and Haile 1998).

Passive Smoking, Active Smoking, and Breast Cancer Risk

The involuntary inhalation of tobacco smoke by nonsmokers has also been examined as a risk factor for breast cancer. Exposure to secondhand smoke and breast cancer risk has been considered relevant to understanding active smoking and breast cancer risk because passive exposure involves a lower dose of the same agents inhaled by the active smoker. The literature on passive smoking and breast cancer was reviewed in the 2001 Surgeon General’s report with the conclusion that “the totality of the evidence does not support an association between smoking and the risk for breast cancer” (USDHHS 2001, p. 13). Recently, epidemiologists have also investigated the relationship between active and passive exposures to cigarette smoke and breast cancer, and attempted to use a truly “unexposed” reference group; that is, women who have been neither active smokers nor exposed passively to another’s cigarette smoke. According to some researchers (Morabia et al. 1996), only by comparison with such a truly unexposed group will the effects of active smoking be assessed without bias.

The studies of passive smoking and breast cancer contrast somewhat with the findings of the far larger number of studies of active smoking that are consistent in showing no relationship of active smoking with breast cancer. Morabia and colleagues (1996) hypothesized that this apparent contradiction stemmed from the failure of most studies to separate passive smokers from the “unexposed” reference group when assessing the effects of active smoking. They tested this hypothesis in a population-based, case-control study conducted among women living in Geneva, Switzerland. The researchers obtained a detailed lifetime history of exposure to active and passive smoking from all participants, and defined their unexposed reference group as those women never regularly exposed to either passive or active smoking. Passive smokers were women who reported having been exposed to secondhand smoke at least one hour per day for at least 12 consecutive months during their lifetime.

The study included 244 cases and 1,032 controls, with 126 cases and 620 controls who were never active smokers. Among these never active smokers, only 28 cases and 241 controls were also never passive smokers, forming the referent “unexposed” group. The ORs comparing ever active smokers with the referent group were 2.2 for smoking an average of 1 to 9 cigarettes per day, 2.7 for 10 to 19 cigarettes per day, and 4.6 for 20 or more cigarettes per day. Among current active smokers the dose-response trend was even stronger. The ORs did not vary in magnitude when women were stratified according to whether they began smoking before or after their first pregnancy. To examine the effect of removing passive smokers from the reference group, Morabia and colleagues (1996) computed the ORs after considering all never active smokers (including those exposed to secondhand smoke) as the reference group, as in most other studies. The ORs corresponding to the three categories of active smoking given above were reduced in magnitude from 2.2, 2.7, and 4.6 to 1.2, 1.7, and 1.9, respectively. Using this same reference group, Morabia and colleagues (1996) also found an association of breast cancer risk with passive smoking.

A caution that must be raised in reference to this study relates to potential confounding. In this study of women living in Geneva, Switzerland, those with a higher formal education smoked more than women with lower educational levels, unlike the situation in the United States where the prevalence of smoking is now higher in lower socioeconomic groups. Women of a higher socioeconomic status tend to have higher risks for breast cancer because of a higher prevalence of reproductive risk factors (e.g., later age at first birth and lower parity). Thus the findings of elevated risks associated with active and passive smoking in this study of Swiss women could be confounded, in part, by the known reproductive risk factors. Although Morabia and colleagues (1996) controlled for some of these known factors (e.g., age at menarche and at first live birth), as well as for family history of breast cancer, body mass index, and alcohol consumption, there may have been residual confounding because of the control for factors in relatively crude categories and the omission of some factors from the model (e.g., parity, postmenopausal hormone use, and age at menopause). Failure to fully adjust for the higher risks associated with a higher socioeconomic status in this study could explain, in part, the relatively high ORs comparing active smokers and the unexposed control group.

Cigarette Smoking and Breast Cancer Hormone Receptor Status

It is not yet clear if breast cancers with a different hormone receptor status represent etiologically distinct forms of the disease with different risk factor profiles. Researchers have hypothesized that breast cancer tumors that have both estrogen and progesterone receptors (ER-positive/PR-positive) are most closely related to risk factors that are likely mediated by endogenous hormones, whereas tumors without these receptors (ER-negative/PR-negative) would be unrelated to these risk factors (Kelsey et al. 1993; Potter et al. 1995). Receptor status-discordant tumors might exhibit intermediate risk factor profiles. It is not clear from this hypothesis, however, whether smoking, because of its antiestrogenic properties, should decrease the risk of ER-positive/PR-positive tumors, increase the risk of ER-negative/PR-negative tumors, or do both. Findings have been inconsistent.

Several studies have examined whether smoking increases the risk of breast cancers with a particular ER status. A case-control study of Japanese women (1,154 cases, 21,714 controls) found a slightly elevated OR for all breast cancers combined associated with ever smoking (Yoo et al. 1997). This OR elevation was confined to PR-positive tumors (OR = 1.73 [95 percent CI, 1.22–2.45]) and was not observed in PR-negative tumors (OR = 1.06 [95 percent CI, 0.73–1.54]). In this study, there was no difference in estrogen receptor status (OR = 1.42 for ER-positive tumors, 1.33 for ER-negative tumors). However, estrogen receptor status was known for only 40 percent of the cases, and progesterone receptor status was known for only 39 percent of the cases.

In a cohort study reported by London and colleagues (1989), heavy smoking was associated with a small increase in the risk of ER-positive tumors (OR = 1.38 [95 percent CI, 1.04–1.84]). Smoking was not associated with either ER-positive or ER-negative tumors in a case-control analysis by McTiernan and colleagues (1986). In another study, researchers found an increased risk of ER-negative tumors among smokers (Cooper et al. 1989).

Each of the above-cited studies examined active smoking in relation to ERstatus, without removing passive smokers from the reference group (of lifetime nonsmokers). Morabia and colleagues (1998b) examined the relationship between passive smoking, active smoking, and ER status in their previously described case-control study of women in Geneva, Switzerland, again using a reference group of never active, never passive smokers. They divided smokers into three mutually exclusive categories: ever passive, ever active with fewer than 20 cigarettes per day on average, and ever active with 20 or more cigarettes per day on average. They found elevated ORs for both ER-negative and ER-positive tumors in each of the three smoking categories, relative to the reference group. The ORs were slightly higher for the ER-negative tumors, but the numbers of ER-negative cases in the various smoking strata were small, and thus the ORs were imprecise.

Cigarette Smoking and Breast Cancer Mortality

All of the previously discussed studies have examined the relationship between cigarette smoking and breast cancer incidence. Calle and colleagues (1994) examined smoking as a predictor of breast cancer mortality in CPS-II. During the six-year follow-up period, these researchers found that women who were current smokers at baseline were more likely to die of breast cancer than lifetime nonsmokers (RR = 1.26 [95 percent CI, 1.05–1.50]), whereas former smokers were slightly less likely to die of breast cancer than lifetime nonsmokers (RR = 0.85 [95 percent CI, 0.70–1.03]). The association of current smoking with risk for fatal breast cancer increased with a greater number of cigarettes smoked per day, as well as with the total number of years of smoking. The ORs for 1 to 9, 10 to 19, 20 to 29, 30 to 39, and 40 or more cigarettes smoked per day were 0.58, 1.19, 1.32, 1.44, and 1.74, respectively, all relative to lifetime nonsmokers. The ORs for breast cancer mortality for less than 10, 10 to 19, 20 to 29, 30 to 39, and 40 or more years of smoking were 1.10, 1.04, 1.10, 1.26, and 1.38, respectively, again all relative to lifetime nonsmokers.

Because the weight of the epidemiologic evidence does not support a strong etiologic relationship between smoking and breast cancer incidence, these findings on breast cancer mortality likely reflect a poorer survival experience among smokers who develop breast cancer, which might be expected for several reasons. First, smokers are more likely than nonsmokers to have comorbid conditions, such as respiratory and cardiovascular diseases, that could deleteriously affect survival. Second, smokers do not seek a screening mammography as often as nonsmokers, and therefore their disease might tend to be diagnosed at later stages. Data from the 1987 National Health Interview Survey Cancer Control Supplement indicate that current smokers are less likely than lifetime nonsmokers to receive screening mammograms and that the screening disadvantage is greatest among heavy smokers. In contrast, former smokers are more likely to receive mammograms than lifetime nonsmokers (Calle et al. 1994). These differences in screening behavior support the possibility that the results observed by Calle and colleagues (1994) are due in part to later diagnoses among current, and especially heavy, smokers and to earlier diagnoses among former smokers.

Evidence Synthesis

Since the 1960s many large, well-conducted studies of the relationship between active cigarette smoking and breast cancer have been completed, as have laboratory studies of the relationship between smoking and ovarian hormone levels. The epidemiologic evidence provides no support for an overall relationship, neither causal nor protective, between active cigarette smoking and breast cancer. The studies have been conducted in diverse populations around the world and involved thousands of participants.

Evidence for an increased susceptibility to the carcinogenic effects of cigarette smoking on the breast in subgroups of women (e.g., defined by genotype, menopausal status, age at starting smoking) has been inconsistent. The inconsistency in RRs for subgroup analyses among the various studies is not surprising given the small numbers of women in the relevant strata of many of these analyses. For some subgroups, an initial finding from one study regarding an elevated risk in a particular subgroup of women (e.g., Ambrosone and colleagues’ 1996 report of a strong positive relationship between smoking and breast cancer among women with the slow acetylator NAT genotype) has not been replicated in subsequent studies. Similarly, Brunet and colleagues (1998) observed that women with mutations in BRCA1orBRCA2 genes who smoked had a significantly lower risk of breast cancer than women with such mutations who did not smoke, but this observation was not replicated in the study conducted by Couch and colleagues (2001).

In light of the evidence showing no overall association between active smoking and breast cancer, passive smoking would also be expected not to be associated with breast cancer risks, assuming that the same mechanisms apply to both active and passive smoking. Although most studies of smoking and breast cancer did not remove passive-only smokers from the reference group of lifetime nonsmokers (Morabia and colleagues [1996] were the first to do so), one would still expect to find a dose-response gradient in analyses of active smoking because active smokers are also the most heavily exposed passive smokers. The hypothesis put forth by Morabia and colleagues (1996, 1998a) and Wells (1991, 1998), that the true (positive) relationship between active smoking and breast cancer will become apparent only when passive-only smokers are removed from the reference group, implicitly assumes that the effects of passive-only smoking are at least as great as those from active smoking. Consider a hypothetical, but realistic, study that shows a RR of 1.0 comparing current smokers who have smoked for 10 or more years and the reference group of never active smokers. If the argument is made that the “true” RR is 2.0, and that it will not become apparent unless passive-only smokers are removed from the reference group, then there is an assumption that the RR of current smokers who have smoked 10 or more years compared with passive-only smokers is 1.0, or, equivalently, that the risk conveyed by passive smoking alone is equal to that conveyed by long-term active smoking. This comparability of risks seems implausible on a biologic basis.

Conclusions

1. The evidence is suggestive of no causal relationship between active smoking and breast cancer.

2. Subgroups of women cannot yet be reliably identified who are at an increased risk of breast cancer because of smoking, compared with the general population of women.

3. Whether women who are at a very high risk of breast cancer because of mutations in BRCA1 or BRCA2 genes can lower their risks by smoking has not been established.

Implications

In contrast to evidence for many other chronic diseases, epidemiologic evidence suggests that cigarette smoking does not contribute to the burden of breast cancer. It would be false to tell women that they will prevent breast cancer if they quit smoking. Similarly, no woman should ever be advised to smoke to lower her breast cancer risk, given the lack of evidence and the extremely high health risks for other diseases known to be associated with smoking.

Summary

A systematic review of new epidemiologic evidence adds new inferences for a causal relationship between smoking and a number of cancers. This report draws several new conclusions. Specifically, it concludes that evidence is sufficient to infer a causal relationship between smoking and cancers of the cervix, kidneys, pancreas, and stomach. Also, it infers a causal relationship between smoking and acute myeloid leukemia. Although there is evidence that smoking is not related to the risk of developing prostate cancer, this report also concludes that it is probable that smoking contributes to a higher mortality rate from prostate cancer. Finally, this report concludes that active smoking is not causally related to breast cancer.

Conclusions

Lung Cancer

1. The evidence is sufficient to infer a causal relationship between smoking and lung cancer.

2. Smoking causes genetic changes in cells of the lung that ultimately lead to the development of lung cancer.

3. Although characteristics of cigarettes have changed during the last 50 years and yields of tar and nicotine have declined substantially, as assessed by the Federal Trade Commission’s test protocol, the risk of lung cancer in smokers has not declined.

4. Adenocarcinoma has now become the most common type of lung cancer in smokers. The basis for this shift is unclear but may reflect changes in the carcinogens in cigarette smoke.

5. Even after many years of not smoking, the risk of lung cancer in former smokers remains higher than in persons who have never smoked.

6. Lung cancer incidence and mortality rates in men are now declining, reflecting past patterns of cigarette use, while rates in women are still rising.

Laryngeal Cancer

7. The evidence is sufficient to infer a causal relationship between smoking and cancer of the larynx.

8. Together, smoking and alcohol cause most cases of laryngeal cancer in the United States.

Oral Cavity and Pharyngeal Cancers

9. The evidence is sufficient to infer a causal relationship between smoking and cancers of the oral cavity and pharynx.

Esophageal Cancer

10. The evidence is sufficient to infer a causal relationship between smoking and cancers of the esophagus.

11. The evidence is sufficient to infer a causal relationship between smoking and both squamous cell carcinoma and adenocarcinoma of the esophagus.

Pancreatic Cancer

12. The evidence is sufficient to infer a causal relationship between smoking and pancreatic cancer.

Bladder and Kidney Cancers

13. The evidence is sufficient to infer a causal relationship between smoking and renal cell, renal pelvis, and bladder cancers.

Cervical Cancer

14. The evidence is sufficient to infer a causal relationship between smoking and cervical cancer.

Ovarian Cancer

15. The evidence is inadequate to infer the presence or absence of a causal relationship between smoking and ovarian cancer.

Endometrial Cancer

16. The evidence is sufficient to infer that current smoking reduces the risk of endometrial cancer in postmenopausal women.

Stomach Cancer

17. The evidence is sufficient to infer a causal relationship between smoking and gastric cancers.

18. The evidence is suggestive but not sufficient to infer a causal relationship between smoking and noncardia gastric cancers, in particular by modifying the persistence and/or the pathogenicity of Helicobacter pylori infections.

Colorectal Cancer

19. The evidence is suggestive but not sufficient to infer a causal relationship between smoking and colorectal adenomatous polyps and colorectal cancer.

Prostate Cancer

20. The evidence is suggestive of no causal relationship between smoking and risk for prostate cancer.

21. The evidence for mortality, although not consistent across all studies, suggests a higher mortality rate from prostate cancer in smokers than in non-smokers.

Acute Leukemia

22. The evidence is sufficient to infer a causal relationship between smoking and acute myeloid leukemia.

23. The risk for acute myeloid leukemia increases with the number of cigarettes smoked and with duration of smoking.

Liver Cancer

24. The evidence is suggestive but not sufficient to infer a causal relationship between smoking and liver cancer.

Adult Brain Cancer

25. The evidence is suggestive of no causal relationship between smoking cigarettes and brain cancer in men and women.

Breast Cancer

26. The evidence is suggestive of no causal relationship between active smoking and breast cancer.

27. Subgroups of women cannot yet be reliably identified who are at an increased risk of breast cancer because of smoking, compared with the general population of women.

28. Whether women who are at a very high risk of breast cancer because of mutations in BRCA1 or BRCA2 genes can lower their risks by smoking has not been established.

[ 2004 Health Consequences of Smoking: the Surgeon General continues next at Part  97.]

Disclaimer: The author of each article published on this web site owns his or her own words. The opinions, beliefs and viewpoints expressed by the various authors and forum participants on this site do not necessarily reflect the opinions, beliefs and viewpoints of Utah Standard News or official policies of the USN and may actually reflect positions that USN actively opposes. No claim in public domain or fair use.    © Edmunds Tucker. UTopiAH are trade marks of the author. Utopia was written in 1515 by Sir Thomas More, Chancellor of England.