Replies

Cigarette smoking has been suggested as a cause of breast cancer, but many studies addressing the relationship have yielded inconsistent results. A possible explanation is that many of these studies have overlooked the potential effects of passive exposure to cigarette smoke when assessing the effects of active smoking. A further source of confusion is the hypothesised existence of a window of vulnerability to tobacco smoke carcinogens during childhood and adolescence.1 We summarise the epidemiological evidence for a causal association between smoking and breast cancer and consider recent studies on the effects of active and passive exposure to cigarette smoke in the context of this hypothesised window of vulnerability.

Conflicting results on smoking and breast cancer
A review of publications on the relationship between smoking and breast cancer to 1984 found more evidence for a protective than for a harmful effect of cigarette smoking.2 In contrast, a 1990 review reported that the summary odds ratio (OR) for breast cancer in smokers compared with non-smokers was 1.12 for case-control studies (95% CI, 1.06-1.19) and 1.14 for cohort studies (95% CI, 1.02-1.27).3 A large case-control study of about 7000 case and 9000 control participants, published in 1996, found no relationship between cigarette smoking and breast cancer.4
Conflicting results have also emerged from large, well designed cohort studies. A 1989 analysis of the United States Nurses' Cohort Study (about 120 000 women) found no relationship between cigarette smoking and breast cancer.5 In contrast, the American Cancer Society Cohort Study (about 600 000 women) reported in 1994 that women who were current smokers at the time of death had a higher breast cancer mortality than non-smokers (relative risk [RR], 1.26; 95% CI, 1.05-1.50).6 Furthermore, mortality tended to increase with the amount and duration of smoking; for example, RR was 1.74 (95% CI, 1.15-2.62) for women who smoked 40 or more cigarettes per day.

Postulated mechanisms
A range of biological and epidemiological explanations have been proposed for these varying results. On the one hand, cigarette smoking may have a protective effect, as women who smoke have an earlier menopause, with fewer total years of menstruation,7,8 and weigh less than non-smokers,7,9 and also cigarette smoking alters oestrogen metabolism.10,11 That is, the negative effect of cigarette smoking on oestrogen production may reduce the risk of oestrogen-dependent diseases.2
On the other hand, cigarette smokers may have increased risk of breast cancer, as carcinogens from tobacco smoke absorbed into the bloodstream may produce carcinogenesis of breast ductal epithelium.12 Several polycyclic aromatic hydrocarbons are produced by tobacco combustion, including the carcinogens benzo[a]pyrene, which is mutagenic for the p53 tumour suppressor gene in humans,13 and 7,12-dimethylbenz[a]anthracene, which is used to induce mammary tumours in animals.14 Data reported recently from the Carolina breast cancer study showed that archival breast cancer tissue from current cigarette smokers had a higher prevalence of p53 mutations than breast cancer tissue from never smokers: 40.4% versus 24.8%.15 Of particular interest was the finding that p53 mutations of the type found in lung cancers from smokers were detected in breast cancer tissues from 21% of current smokers but only 5% of never smokers.

In addition, Palmer and colleagues hypothesised that women's risk of breast cancer is increased by exposure to cigarette smoke during childhood and adolescence, as these are times of rapid breast growth.1 They reported two case-control studies of breast cancer risk among women who smoked 25 or more cigarettes per day compared with never active smokers. These studies found that ORs for smokers who began smoking before the age of 14 years, compared with never active smokers, were 2.6 (95% CI, 1.0-6.8) and 1.9 (95% CI, 0.9-4.1), respectively. In contrast, ORs were lower for those who began smoking later, at ages 18-21 years, compared with never active smokers: 1.1 (95% CI, 0.7-1.7) and 1.2 (95% CI, 0.7-1.8), respectively.15

Resolving the conflicts
The explanation for much of the discrepancy between the above results on active smoking and breast cancer may be that the studies did not consider passive exposure to tobacco smoke.16 This issue has been addressed by three recent case-control studies which compared smokers with women who had no exposure to tobacco smoke, either active or passive.12,17,18 They found ORs 2.0 for the overall risk of breast cancer in ever smokers (two studies12,18) or ever smokers with premenopausal breast cancer (one study17) compared with women with no active or passive exposure to tobacco smoke. Selected data from these three studies are shown in the Box.
These three studies also examined the association between breast cancer and passive exposure to cigarette smoke, as did two other case-control studies19,20 and two cohort studies.16,21 Data from the Hirayama cohort, reported in 1998 (91 540 women), showed that spouses who were passively exposed had an RR of 1.32 (95% CI, 0.83-2.09).16 A recent Korean cohort study on spouses' passive exposure to smoke and lung cancer (160 130 women) also found an increased risk of breast cancer (RR, 1.7; 95% CI, 1.0-2.8).21

Selected data from the five case-control studies on the effects of passive exposure to cigarette smoke, including effects of age of exposure, are shown in the Box. Two studies found a significant exposure-response effect in terms of the number of smokers and smoker-years to which women were passively exposed. The age of passive exposure also appeared important, as Palmer et al observed for active exposure in two previous case-control studies.1 Smith et al found childhood exposure to be almost as important as adult exposure,19 and Lash and Aschengrau found that the risk of breast cancer in females passively exposed to cigarette smoke was highest when exposure occurred before the age of 12 years, less when it occurred at 12-20 years, and least when it occurred after the age of 20.18 Furthermore, they found an OR of 7.5 (95% CI, 1.6-36) for the occurrence of breast cancer in girls exposed to passive smoking who were also active smokers before the age of 12 years. It should be noted that, as there were few case and control participants, 95% confidence intervals were wide, and the estimates should be viewed with caution.

The ORs shown in the Box suggest a causal association between active and passive exposure to cigarette smoke and breast cancer (ORs > 2.0 are considered to indicate a strong association22). Interestingly, a recent review of established and possible aetiological factors in breast cancer found that only age, strong family history, mutations in BRCA-1 or BRCA-2 genes, country of birth and atypical cells in nipple aspirates were associated with ORs over 4.0.23

Lash and Aschengrau based their study on a model of susceptibility of breast tissue to tobacco smoke which was derived from new knowledge on breast tissue development and susceptibility to chemical carcinogens.18 The physiological development of the mammary gland involves four different lobule types, representing sequential developmental stages.24 During sexual maturation (puberty), breast tissues evolve from lobule type 1 (highest doubling rate and greatest susceptibility to carcinogens) to type 2 (intermediate doubling rate and susceptibility to carcinogens). During pregnancy, or gradually with premenopausal ageing, breast tissues evolve to type 3 (low doubling rate and susceptibility to carcinogens). Type 4 lobules (maximal expression of development and differentiation) develop during lactation. Therefore, the window of maximum vulnerability of girls to tobacco-smoke-induced breast cancer should be from birth through puberty.

Where to from here?
Clearly, further studies are needed to investigate and better measure the effects of active and passive exposure to cigarette smoke in various phases of childhood, adolescence and adult life. Furthermore, the recent studies reviewed here involved women in Europe and the United States who presented with breast cancer in the 1980s and 1990s. Their exposure to cigarette smoke during the hypothesised window of vulnerability would have occurred mainly in the 1930s to 1950s, and would reflect the smoking habits of teenage and young women and their parents at that time. Therefore, Australian data are urgently needed, as exposures may have differed in this country.
Finally, if active and passive exposure to cigarette smoke in childhood and adolescence proves to cause breast cancer, then the current smoking habits of Australian teenage girls are of even greater concern. The 1996 survey on use of tobacco and alcohol among Australian secondary students revealed that 14%-20% of girls aged 12-15 years were current smokers in the period 1984-1996, and that they smoked an average of 18.7-21.2 cigarettes per week.25 By the age of 12 years, 32% of girls had experimented with cigarette smoking.25 A proportion of Australia's future burden of breast cancer may already have been initiated.