Consistent with some previous studies, consumption of fruits, vegetables, and cruciferous vegetables, showed statistically significant inverse associations with lung cancer risk [12, 13, 25–28]. In our study, intake of total vegetables was inversely associated with lung cancer risk in both never smokers and former smokers. However, when cruciferous vegetables were excluded from total vegetables, the associations were markedly attenuated and no longer statistically significant among former smokers, but the significant associations remained among never smokers (results not shown). Similarly, intake of fruits showed a marginally negative association with lung cancer risk only among never smokers, but no association was observed for ever-smokers, including both former and current smokers. In contrast to total vegetables and fruits, a strong inverse association between cruciferous vegetable intake and lung cancer risk was observed among smokers, in particular former smokers, although we did not detect significant interactions between cruciferous vegetable intake and smoking status. There was no statistically significant association between cruciferous vegetable intake and lung cancer risk among never smokers. Limited power due to relatively small number of never smokers in the current study could contribute to this observation, although this result is consistent with the findings from studies conducted in never smokers [16, 29–31].
Our findings indicate that cruciferous vegetables may play a preventive role in lung cancers that are smoking-related, rather than the more general effect of other vegetables and fruits that may be overshadowed by the strong effect of smoking. This hypothesis is supported by the carcinogen-modulating activity of isothiocyanates, a group of natural phytochemicals uniquely present in cruciferous vegetables. Modulation of metabolism of smoking-related carcinogens by isothiocyanates has been documented in both in vivo and in vitro studies [6, 7], as well as in humans . Isothiocyanates have also been shown to inhibit lung tumorigenesis induced by tobacco-specific carcinogens in animal models [9, 32, 33]. Interestingly, we observed a stronger inverse association between cruciferous vegetable intake and lung cancer risk in the short-term smokers (≤ 30 years of smoking) than the long-term smokers (> 30 years of smoking). Carcinogenesis is an accumulation of carcinogen-induced DNA damage over a long period of time. Although intake of raw cruciferous vegetables/isothiocyanates is able to considerably attenuate smoking-related carcinogen exposure and delay this process significantly, the capacity is not unlimited. Even in the presence of high cruciferous vegetable intake, smokers may still develop lung cancer, especially if they continuously smoke for a long enough time. This is also in line with the finding that the strongest association with intake of cruciferous vegetables was observed among former smokers, instead of current smokers. Comparison of years of smoking between them reveals a longer history of smoking in current smokers (40.3 years) than in former smokers (30.0 years). Furthermore, smoking cessation may represent a healthy life style change possibly accompanied by an increase in vegetable intake. Therefore, smoking cessation should still be emphasized for lung cancer prevention.
Not surprisingly, compared to total cruciferous vegetables, intake of raw cruciferous vegetables, in general, was more strongly inversely associated with lung cancer risk. Isothiocyanates occur naturally in cruciferous vegetables as the precursor glucosinolates (β-thioglucoside N-hydroxysulfates), and their release requires the enzyme myrosinase (thioglucoside glucohydrolase). In the plant, this enzyme is stored physically separated from glucosinolates, and released to hydrolyze glucosinolates when plant cells are damaged. Isothiocyante yield is two to nine times higher from consumption of raw cruciferous vegetables compared with consumption of their cooked counterparts, due to heat-inactivation of myrosinase, destruction of heat-labile isothiocyanates, and loss of glucosinolates during cooking procedures [19–21]. We have previously reported inverse associations between consumption of raw cruciferous vegetables and bladder cancer, another cigarette smoking-related cancer . It is noteworthy that, in addition to isothiocyanate-precursor glucosinolates, cruciferous vegetables also contain other glucosinolates such as indole-precursors. The type and total content of glucosinolates differ substantially among different cruciferous vegetables as well as within the same vegetable under different culture conditions . It is likely that indoles and other phytochemicals and nutrients in cruciferous vegetables such as carotenoids, vitamin C, folic acid, selenium, may also play a role in the chemoprevention of lung cancer, whether or not in combination with isothiocyanates.
Lung cancers were classified into five histological subtypes: small cell carcinoma, squamous cell carcinoma, adenocarcinoma, large cell carcinoma, and the others. When we examined the risk associations among these histological subtypes, significant inverse associations were apparent only among patients with squamous or small cell carcinoma. Interestingly, both squamous and small cell carcinomas tend to be more strongly associated with heavy smoking than other subtypes . Indeed, we found that patients with squamous and small cell carcinoma smoked more (30 cigarettes per day) and longer (39 years of smoking) than did patients with adenocarcinoma (23 cigarettes per day and 29 years of smoking). Additionally, in terms of smoking behavior, patients with adenocarcinoma were more similar to the controls (23.3 versus 22.6 cigarettes per day and 29.4 versus 27.1 years of smoking) (results not shown). Finally, adenocarcinoma occurs in a great proportion of lung cancers in nonsmokers, suggesting factors other than smoking may also play an important role in the etiology of this subtype . Since the potential beneficial role of cruciferous vegetable intake may at least partially rely on modulation of carcinogens in cigarette-smoking by isothiocyanates, it is not surprising to detect differential effects among subtypes of lung cancer.
Several methodological issues should be considered in the interpretation of the results from a hospital-based case-control study. Recall bias is always a concern in case-control studies, as controls may be more motivated than cases to recall cruciferous vegetable intake. However, this may be less of an issue in the current study due to the use of hospital controls who also came to hospital with a suspicion of cancer, but were diagnosed with other diseases. Selection bias may also occur. Both cases and controls were limited to individuals who came to RPCI, a large regional comprehensive cancer center, and may not fully represent the general population or the majority of lung cancer patients. Furthermore, only about 50% of the eligible cases and controls agreed to complete the questionnaire. We were unable to assess whether or not participants and non-participants differed with respect to cruciferous vegetable intake. However, daily intake of cruciferous vegetables among controls in our study (median intake 26.7 g/day, accounting for 14% of all vegetable intake) is comparable to other studies conducted in North America area (daily intake ranges from 16 to 40 g with 5.6 to 15.4% of total vegetable intake) , suggesting the detected associations might apply in general population. Misclassification of smoking status might affect our study, as our analysis was based on self-reported data. However, several related questions were employed in the questionnaire to address the same issue, which were highly correlated and would minimize the potential for this source of bias.
The strength of the current study is the use of two approaches to control the confounding effects of smoking: cases and controls were first matched on smoking status; and then smoking status, smoking intensity (indicated by number of cigarettes per day), and smoking duration (indicated by years of smoking) were further adjusted in the data analysis. The need of these approaches is supported by the results shown in Table 1. Even after matching on smoking status, cases still had significantly higher number of cigarettes per day and more years of smoking than controls. These results also indicate that the confounding effect of cigarette smoking may not be adequately controlled by adjusting only for either smoking status or pack-years of smoking, a strategy used in many previous studies. On the other hand, because we were unable to control for the type of cigarettes smoked, pipe and cigar smoking, environmental smoking exposure, residual confounding by smoking may still remain in the current study, albeit to a lesser extent.