Tobacco Documents Online

1 REVIEW 1075 CONFIDENTIAL Subject ref 8b "Prospective study of smoking, antioxidant intake, and lung cancer in middle-aged women (USA)" F E Speizer et al Cancer Causes and Control (1999), 10, 475-482 The Nurses Health Study, established in 1976 with a mailed questionnaire to 121,700 female US registered nurses aged 30-55, has generated a large number of published reports. The current paper is of interest in that it presents results relating lung cancer to smoking (including tar level), ETS exposure and diet based on follow-up to 1992. There were a total of 593 pathologically confirmed cases of lung cancer among the 118,351 women free of cancer in 1976. Some of the analyses were based on smaller numbers of cases reflecting information first collected at a later stage. ETS In 1982 questions were asked about ETS exposure in both childhood and adulthood, including on whether the parents smoked when living with them, how many years they lived with someone smoking regularly and whether they were currently exposed to cigarette smoke from other people. There were only 356 new lung cancer cases in never smokers for whom ETS data were available and the authors note that there were only two with no adult ETS exposure, with the age- adjusted RR for ETS exposure in adulthood estimated as 1.5 with 95% CI 0.3-6.3. In the discussion the authors note that "these data provide additional evidence of an excess risk of lung cancer among women exposed to environmental tobacco smoke exposure in adult life defined as exposure both at home and at work," though they note that the "confidence interval was wide." Actually the CI is so wide that it is a pretty generous interpretation of the findings to say that they provide any real evidence at all. Since Speizer et al had data on years

3 lung cancer to tar level or to filter/plain status have not found this, with adjustment having only a minor effect. The size of the adjustment would suggest that low tar smokers smoked half as many cigarettes/day as did high tar smokers, which seems a very big difference. Diet Data are not presented on risk of lung cancer by diet specifically for nonsmokers. Instead results are presented for smokers and nonsmokers combined adjusted for amount smoked, past smoking and age of starting to smoke, as well as for age and total energy intake. Little or no relationship was seen between lung cancer risk and intake of dietary fat (see Table 3). There was also no clear relationship of risk to vitamin C, total carotene, vitamin E and folate; although a significant (p = 0.03) trend was stated to be found for vitamin C and a near significant (p = 0.10) trend was found for vitamin E, the actual pattern of risk by quintile in Table 4 seemed to be up and down to such an extent that it was difficult to see whether the claimed trend was positive or negative! The clearest relationships with diet (see Tables 5 and 6) were for carrots and alpha- carotene, to which carrots are a major contributor, with risk reducing with increasing consumption. It is interesting to note that some 25 years ago great interest in diet and lung cancer surfaced (including in the Tobacco Research Council) when Bjelke presented results of a prospective study showing a strong negative relationship between lung cancer and an index of carotene consumption which was largely actually an index of carrot consumption. This led on to particular interest in beta-carotene and the well known intervention trials, which suggested that, far from being protective, it may well increase the risk of lung cancer. The authors include some discussion on the possibility that carotenoids other than lung cancer might be important, without reaching any clear conclusions. The main conclusion of the paper is "Smoking is the most important risk factor for lung cancer in women, as it is in men. Higher vegetable consumption, particularly of carrots, may significantly reduce the risk of lung cancer, but dietary modification cannot be considered a substitute for smoking prevention and cessation." P N Lee 20.11.99.

2 of exposure to a regular smoker, why did they not use this to classify subjects into low and high ETS exposure and avoid the problem of there being only two lung cancer cases with no ETS exposure? Why also did they say they collected data on ETS exposure from parents and not report their findings? Smoking The authors report the expected strong relationship of current smoking status at baseline and amount smoked to lung cancer risk over the next 16 years, with smokers of more than 20 cigarettes a day having an age-adjusted risk 20-fold or higher than that of never smokers. Givine up smoking was associated with a reduced risk of smoking compared to continuing, with those who had not smoked for 15 years or more having a risk similar to that of never smokers. After adjustment for age and number smoked, the relative risk reduced with later age of starting to smoke, with relative risks of 1.1, 1.0 and 0.8 for ages <18, 18-19, >21. However the presentation was uninformative, with no data for age 20-21 and no test whether the trend was statistically significant. In a recent unpublished review of risk of lung cancer in relation to type of cigarette smoked, I noted that there were no studies at all relating tar level to lung cancer deaths or cases occurring in the 1980s. Here at last was one. However, the presentation of the findings was remarkably brief. They stated that tar levels were based on brands reported as being smoked in 1978 and used to divide current smokes into "tertiles." They then stated that "Although the 1978 level of tar appeared to relate to lung cancer risk with adjustment for age and age first smoked (RR for top vs. bottom quartile = 2.0 [1.5-2.8]), the effect of tar content was no longer significant after additional adjustment for current number of cigarettes smoked (RR = 1.0 [0.7-1.4])." In the discussion, they note that "We found no effect of tar content, perhaps because the large majority of these women smokers were smoking low tar cigarettes." This is really remarkably uninformative and confusing. Did they divide tar levels into three groups (tertiles) or four groups (quartiles)? What were the groupings in terms of tar levels? Why were the results for the middle groups not presented? Why did adjustment for amount smoked have such a dramatic effect on the high tar/low tar relative risk? Other studies relating