Background: Cruciferous vegetables, a rich source of isothiocyanates, have been reported to lower the risk of several types of cancer, including lung cancer. However, evidence from prospective observations of populations with a relatively high intake of cruciferous vegetables is sparse.
Objective: We investigated the association between cruciferous vegetable intake and lung cancer risk in a large-scale population-based prospective study in Japan.
Methods: We studied 82,330 participants (38,663 men; 43,667 women) aged 45–74 y without a past history of cancer. Participants were asked to respond to a validated questionnaire that included 138 food items. The association between cruciferous vegetable intake and lung cancer incidence was assessed with the use of Cox proportional hazard regression analysis to estimate HRs and 95% CIs (with adjustments for potential confounding factors).
Results: After 14.9 y of follow-up, a total of 1499 participants (1087 men; 412 women) were diagnosed with lung cancer. After deleting early-diagnosed cancer and adjusting for confounding factors, we observed a nonsignificant inverse trend between cruciferous vegetable intake and lung cancer risk in men in the highest compared with the lowest quartiles (multivariate HR: 0.85; 95% CI: 0.69, 1.06; P-trend = 0.13). Stratified analysis by smoking status revealed a significant inverse association between cruciferous vegetable intake and lung cancer risk among those who were never smokers and those who were past smokers after deleting lung cancer cases in the first 3 y of follow-up [multivariate HR for never smokers: 0.49 (95% CI: 0.27, 0.87; P-trend = 0.04); multivariate HR for past smokers: 0.59 (95% CI: 0.35, 0.99; P-trend = 0.10)]. No association was noted in men who were current smokers and women who were never smokers. Conclusion: This study suggests that cruciferous vegetable intake may be associated with a reduction in lung cancer risk among men who are currently nonsmokers. J Nutr doi: 10.3945/jn.117.247494.
Cruciferous vegetables, including broccoli, cabbage, and radish, are a rich source of isothiocyanates, which are known for their protective effect against cancer (1). Although the mechanisms underlying their anticancer properties are not clear, they may enhance the excretion of
1 Supported by the National Cancer Research and Development Fund, Ministry of Health, Labor and Welfare, and Ministry of Agriculture, Forestry and Fisheries. 2 Author disclosures: M Inoue is a beneficiary of a financial contribution from the AXA Research Fund. N Mori, T Shimazu, S Sasazuki, M Nozue, M Mutoh, N Sawada, M Iwasaki, T Yamaji, R Takachi, A Sunami, J Ishihara, T Sobue, and S Tsugane, no conflicts of interest.
carcinogens before they damage DNA (2, 3). Cruciferous vegetable intake is associated with lowering the risk of several types of cancer, such as bladder (4–6) and colorectal cancer (7, 8).
Several prospective epidemiologic studies have considered the association between cruciferous vegetables intake and lung cancer risk (9–17). As for all cruciferous vegetables, 2 studies to our knowledge reported a significant inverse association (12, 17), but most studies reported a nonsignificant inverse asso- ciation (9, 13–16). Among all prospective studies (9–17), 2 found a significant inverse association between individual cruciferous spices, namely Chinese greens (15) and cauliflower (13), and lung cancer risk. Both cruciferous vegetables represented the highest mean intake among the cruciferous vegetables considered in the study (13, 15).
The reason for the weak inverse association may be partly explained by the little variation in the amount of cruciferous vegetables consumed. Seven of the 9 studies were conducted in Western countries (9–14, 17), where the intake of cruciferous vegetable is lower than in Asian countries. The daily intake of total cruciferous vegetables was reported as 0.2 servings (~22.6 g/d) in the United States (18); 59.8 g cabbage, Chinese leaves, and Chinese radish intake per day in Japan (19); and 121.0 g leafy, stalk, or shoot vegetables and the brassica subgroup per day in Hong Kong (20). Therefore, a stronger association may exist in Asian populations with a relatively high intake of cruciferous vegetables. However, few such prospective investigations have been conducted in Asian populations (15, 16).
Furthermore, cigarette smoking is an established risk factor of lung cancer (21), and reports from prospective studies of the association between cruciferous vegetable intake and lung cancer stratified by smoking status are limited and inconsistent (9, 15, 17).
This study was a large-scale population-based prospective study among Japanese residents with a relatively high intake of cruciferous vegetables. We aimed to clarify the association between cruciferous vegetables intake and lung cancer risk while taking smoking status into account.
Study population. The Japan Public Health Center (JPHC)10 prospec- tive study was launched between 1990 and 1993. The details of the study design have been described previously (22). The study protocol was approved by the institutional review board of the National Cancer Center, Tokyo, Japan.
The study population was defined as registered residents aged 40– 69 y of 11 public health center (PHC) areas of Japan. Ineligible subjects were first excluded because of non-Japanese nationality (n = 51), late report of emigration before the follow-up (n = 207), incorrect birthdate (n = 7), and duplicate registration (n = 10). We then excluded 1 PHC (Tokyo area) because it lacked information on cancer incidence (n = 7078). Furthermore, we excluded 2459 subjects who had died, moved out of the study area, or with whom we lost contact before the 5-y follow-up questionnaire survey.
We used the 5-y follow-up survey as the starting point because the questionnaire used in this survey provided more comprehensive infor- mation on dietary intake than the baseline survey. We included subjects who responded to the questionnaire that contained demographic data, medical history, information on smoking, and dietary intake through an FFQ. Of the remaining 130,608 participants, 98,552 responded to the questionnaire (response rate: 75.5%).
Among the 98,552 subjects, we excluded those who reported or were diagnosed with cancer before the 5-y follow-up questionnaire survey, those who left the frequency of all cruciferous vegetable intake unanswered (n = 2271), those who reported daily energy intake at the upper or lower 2.5% ends of the range (1015 and 4204 kcal for men and 865 and 3689 kcal for women, respectively; n = 4812), and those with incomplete information on smoking status (n = 5725). Finally, 82,330 participants (38,663 men; 43,667 women) were included in the analysis.
FFQ. The FFQ used in the 5-y follow-up questionnaire survey was designed to estimate the habitual dietary intake of 138 food items (including 6 cruciferous vegetables and 3 pickled cruciferous vegetables) during the previous year in standard portion sizes and 9 frequency categories. Standard portion sizes for each food item were small (50% less than the standard serving size), medium (proportional to standard serving size), and large (50% larger than the standard serving size). The 9 frequency categories for each food items were never, 1–3 times/mo, 1–2 times/wk, 3–4 times/wk, 5–6 times/wk, 1 time/d, 2–3 times/d, 4–6 times/d, and $7 times/d. Cabbage, Chinese radish, broccoli, komatsuna, Chinese leaves, pak choi, leaf mustard, and Swiss chard were categorized as cruciferous vegetables according to the grouping rules adopted in the IARC Handbooks of Cancer Prevention (19). The 3 pickled cruciferous vegetables were pickled Chinese radish, rape and leaf mustard, and Chinese leaves. Total cruciferous vegetables did not include pickled cruciferous vegetables in this study in order to make valid comparisons with previous studies (15, 16). Isoflavone intake was also estimated with the use of the FFQ to be included as a covariate because we have previously found an inverse association between isoflavone intake and lung cancer risk (23, 24). The dietary intake of each food item was calculated by multiplying frequency by standard portion and the relative portion sizes for each food item. The food intake was log-transformed and adjusted for total energy intake with the use of the residual method (25). We examined the validity of the FFQ for assessing cruciferous vegetable intake and the reproducibility of the FFQ administered at a 1-y interval (Supplemental Tables 1 and 2).
Follow-up and identification of lung cancer incidence. We followed up participants until 31 December 2012. Participants who died or relocated to other municipalities were identified annually through residential registers in each PHC area. The cause of death was confirmed with the use of mortality data from the Ministry of Health, Labor and Welfare. Among study participants, 12,345 (15.0%) died, 2992 (3.6%) moved away, and 144 (0.2%) were lost to follow-up during the study period.
Data on lung cancer incidence were obtained from local major hospitals in the respective PHC areas and data linkage with population-based cancer registries with permission from the local governments. We also used death certificate information as supplementary data. In our cancer registry system, 6.4% of lung cancer cases were obtained through death certificates only during the study period. During the 1,195,175 person-years of follow-up (median follow-up period: 14.9 y), a total of 1499 participants (1087 men; 412 women) were diagnosed with lung cancer.
The histologic types of lung cancer were coded based on the International Classification of Diseases for Oncology, 3rd edition (C34.0–C34.9) (26). The diagnosis of lung cancer was confirmed with the use of a histo- or cytologic examination in 79% of cases (n = 1187) and was based on the clinical findings or unspecified evidence in the remaining 21% of cases. Histologic type was classified as adenocarci- noma (n = 630; 42%), squamous cell carcinoma (n = 292; 19%), small cell carcinoma (n = 146; 10%), and other histologic types according to the WHO!s histologic classification of lung tumors (27).
Statistical analysis. We prospectively counted the number of person- years of follow-up for each participant from the date of completion of the 5-y follow-up questionnaire until the date of diagnosis of lung cancer, date of death, moving out of the study area, or end of follow-up (31 December 2012), whichever came first.
Cox proportional hazards regression analysis was used to calculate the sex-specific HRs and 95% CIs of lung cancer incidence according to quartile of cruciferous vegetable intake and adjusted for potential covariates. The lowest quartile of cruciferous vegetable intake was used as a reference. P values for linear trends were calculated by assigning ordinal variables for quartiles of cruciferous vegetable intake and entering the number as a continuous variable into the model. Because cigarette smoking is the most important risk factor for lung cancer, we performed a subgroup analysis by smoking status on the association between cruciferous vegetable intake and lung cancer risk. Cox proportional hazards regression analysis was repeated to calculate the HRs and 95% CIs for each smoking status. We then repeated the same analysis after excluding 189 lung cancer cases diagnosed in the first 3 y of follow-up to diminish the potential influence of participants with subclinical cancer who had modified their dietary habits. The multivar- iate model was adjusted for age, PHC area, smoking status, fruit intake, noncruciferous vegetable intake, isoflavone intake, and menopausal status in women.
Because pickled cruciferous vegetables accounted for 26% of the total cruciferous vegetable intake in our subjects, we conducted a subanalysis by including pickled cruciferous vegetable intake in total cruciferous vegetable intake as a main exposure. Further analysis was conducted on the association between the intake of specific cruciferous vegetables and lung cancer risk by smoking status in men. Multivariate HRs and 95% CIs before and after the removal of lung cancer cases diagnosed in the first 3 y of follow-up were calculated. For those cruciferous vegetables with a Spearman!s correlation coefficient of <0.1, additional analyses were performed to investigate the association between the frequency of intake and lung cancer risk. We then calculated P values with the addition of an interaction term in the multivariate model by smoking status. The reported P values are 2-sided, and P < 0.05 was defined as significant. All statistical analyses were performed with the use of SAS version 9.3 (SAS Institute).
The baseline characteristics of participants! cruciferous vegeta- ble intake are shown in Table 1. Those with a higher intake of cruciferous vegetables were less likely to be current smokers, to have a higher BMI, and to have had a chest X-ray or sputum screening. A dietary intake of vegetables, fruit, and isoflavones was significantly higher in the highest quartile of cruciferous vegetable intake in both men and women.
Table 2 shows the association between cruciferous vegetable intake stratified by smoking status and lung cancer risk. Because of the small number of lung cancer cases among women who were past and current smokers, only the results for never smokers are shown. Although we observed trends of decreased cancer risk in the highest quartile among men who were never and past smokers, the association was not significant among never smokers (multivariate HR: 0.66; 95% CI: 0.39, 1.13; P-trend = 0.17) or past smokers (multivariate HR: 0.74; 95% CI: 0.49, 1.09; P-trend = 0.18). After excluding the lung cancer cases diagnosed in the first 3 y of follow-up, multivariate HRs of the highest quartile of cruciferous vegetable intake were signifi- cantly inversely associated with lung cancer risk among both never and past smokers: 0.49 (95% CI: 0.27, 0.87; P-trend = 0.04) and 0.59 (95% CI: 0.35, 0.99; P-trend = 0.10), respectively. When we combined never and past smokers, the association was still significant (multivariate HR: 0.54; 95% CI: 0.37, 0.79; P-trend = 0.007). However, no such association was observed in men who were current smokers (P-trend = 0.92) or in women who were never smokers (P-trend = 0.86). Associ- ations were virtually unchanged when pickled cruciferous vegetable intake was included in the total cruciferous vegetable intake (data not shown).
We also conducted an analysis of specific cruciferous vege- table intake and lung cancer risk by smoking status in men. To investigate the true effect of cigarette smoking, results of never and current smokers were compared (Table 3). After deleting lung cancer cases diagnosed in the first 3 y of follow-up, a significant inverse association was observed in the highest quartile of cabbage intake and lung cancer risk (multivariate HR: 0.57; 95% CI: 0.34, 0.97; P-trend = 0.04) among never smokers. No significant association was noted between other types of cruciferous vegetable intake and lung cancer risk in never or current smokers. Results were unchanged when we conducted additional analyses between the frequency of Chinese leaves, leaf mustard, and Swiss chard and lung cancer risk (data not shown).
We also conducted a likelihood ratio test to calculate P values for interactions by adding the potential confounding factors to the multivariate model. First, we tested the association between cruciferous vegetable intake and lung cancer risk among men only by smoking status (P-interaction = 0.65) and both sexes combined (P-interaction = 0.73). Second, we tested the associ- ation by sex (P-interaction = 0.13).
We further analyzed the association between cruciferous vegetable intake and lung cancer risk stratified by detection types to try to explain the observed null association in women who were never smokers. We found no association between cruciferous vegetable intake and symptom-detected lung cancer (multivariate HR: 0.97; 95% CI: 0.49, 1.89; P-trend = 0.84), whereas we observed a nonsignificant but increased risk of screening-detected lung cancer (multivariate HR: 1.21; 95% CI: 0.68, 2.14; P-trend = 0.36). In men, however, we observed a significant decreased risk of symptom-detected lung cancer (multivariate HR: 0.45; 95% CI: 0.23, 0.85; P-trend = 0.03).
Further investigation was conducted on the association between cruciferous vegetable intake and the risk of lung cancer by histologic type among women who were never smokers. Because of the small number of lung cancer cases (in men: adenocarcinoma, n = 65, 43%; squamous cell carcinoma, n = 24, 16%; and small cell carcinoma, n = 9, 6%; in women: adenocarcinoma, n = 272, 76%; squamous cell carcinoma, n = 12, 3%), only adenocarcinoma among women who were never smokers was analyzed. The multivariate HR among the highest quartile in reference to the lowest quartile was 0.98 (95% CI: 0.67, 1.44; P-trend = 0.85).
This prospective cohort study investigated the association between cruciferous vegetable intake and lung cancer risk in a population with a relatively high intake of cruciferous vegeta- bles. We found a significant inverse association between crucif- erous vegetable intake and lung cancer risk among men who were current nonsmokers, although no significant association was observed among men who were current smokers and women who were never smokers. Similarly, an analysis of individual cruciferous vegetable intake revealed a significant association between cabbage intake and lung cancer risk among men who were never smokers. Although our study group previously found no association between total fruit and vegeta- ble intake and lung cancer risk, the association between cruciferous vegetable intake and lung cancer risk was not considered (28).
In addition to containing several bioactive components such as folate, vitamin C, tocopherols, and carotenoids (29), crucif- erous vegetables are a rich source of glucosinolates, which are catalyzed into isothiocyanates by myrosinase. Although the mechanisms underlying the anticancer properties of isothiocy- anates are not fully understood, isothiocyanates have been reported to inhibit the development of various cancers, including lung cancer in an animal model (30).
Asians, including the Japanese, have higher cruciferous vegetable intake than non-Asian populations (19, 20). Whereas
cruciferous vegetable intake is reported to be 0.2 servings/d (~22.6 g/d) in the US population (18), the intake reported in the Japanese population is much higher (59.8 g/d) (19). However, these data only include cabbage, Chinese leaves, and Chinese radish, and total cruciferous vegetable intake is expected to be higher among the Japanese. In fact, our dietary record collected from subsamples of the JPHC study indicated a mean crucifer- ous vegetable intake of 110 g/d for men and 102 g/d for women (Supplemental Table 1).
After removing subjects with early-diagnosed cancer and adjusting for confounding factors, we observed a nonsignifi- cant inverse trend between cruciferous vegetable intake and lung cancer risk in men in the highest compared with the lowest quartiles. The prevalence of smoking reported in 2000 was 51.3% in Japanese men and 27.7% in US men (31). It is likely that because of the much higher prevalence of smoking among Japanese men, we failed to observe a significant inverse association between cruciferous vegetable intake and lung cancer risk in men.
In this study, no significant association was found between cruciferous vegetable intake and lung cancer risk among men who were current smokers. Two prospective studies investigated the association between cruciferous vegetable intake and lung cancer risk among current smokers and reported inconsistent results. Whereas Lam et al. (17) reported a significant inverse association between cruciferous vegetable intake and lung cancer risk among current smokers, Wright et al. (9) found no protective association in both sexes.
Circulating antioxidants such as ascorbic acid, a-carotene, b-carotene, and cryptoxanthin are depleted in current smokers (32, 33), which may explain our result. Likewise, Steinmetz et al. (34) found a stronger inverse association between fruit and vegetable intake and lung cancer risk among past smokers than in current smokers. They noted that the continued presence of carcinogens derived from tobacco smoke may have overwhelmed the anticarcinogenic capacity of vegetables. In addition, the residual confounding factor of smoking might have influenced the results among current smokers.
Unlike in men, we found no significant association between cruciferous vegetable intake and lung cancer risk among women who were never smokers. One possible reason for this differ- ence could be a lack of information on secondhand smoking in women who were never smokers from the 5-y follow-up questionnaire, which may have masked the association. Addition- ally, a previous study has noted that the epidermal growth factor receptor (EGFR) gene mutation is associated with adenocarcinoma histology, never smokers, East Asian ethnicity, and women (35). Matsuo et al. (36) found that the protective effect of vegetables against non-small cell lung cancer were only present for wild-type EGFRs. The EGFR gene mutation status may explain the differ- ences found in our study because no association was found between cruciferous vegetable intake and lung cancer risk in women.
In addition, we also investigated the influence of a detection bias with regard to this difference in sex. As expected from the association between healthy behavior, including screen- ing experience and cruciferous vegetable intake, we observed an increased point estimate of screening-detected lung cancer risk in the highest quartile of cruciferous vegetable intake among women who were never smokers.
No association was found in symptom-detected lung cancer risk among women. In contrast, we did not observe an increased risk of screen- or symptom- detected lung cancer in men. These results suggest that women with higher cruciferous vegetable intake were likely to be diagnosed with lung cancer via screening. However, it is not clear why the detection bias arose only among women.
The effect of cooking (37) is important because it has been reported that raw cruciferous vegetables have a higher bioavail- ability of isothiocyanates (38, 39). We identified the amount of raw intake of the 3 most frequently consumed cruciferous vegetables, namely Chinese radish, cabbage, and Chinese leaves, with the use of 28-d dietary records of 289 subjects from the JPHC validation study (40) based on the list of menus provided by the subjects. Within that period, the proportional intake of raw cabbage, Chinese radish, and Chinese leaves (grams per day) were 26.7%, 38.4%, and 0.7%, respectively, without including pickled vegetables (N Mori, T Shimazu, S Sasazuki, M Nozue, M Mutoh, N Sawada, M Iwasaki, T Yamaji, M Inoue, R Takachi, A Sunami, J Ishihara, T Sobue, S Tsugane, unpublished data). In addition, different cruciferous species have different precursors of glucosi- nolate, which can be converted into isothiocyanates with different anticarcinogenic effects (41).
We found a significant inverse association between cabbage intake and lung cancer risk among men who were never smokers. A previous case-control study in Japan found a similar inverse association of raw cabbage intake (42). Taken together, these findings suggest that raw cabbage intake has a preventive effect on lung cancer. Future prospective research is needed to assess the effect of cooking methods or cruciferous species on lung cancer.
The strengths of this study include its prospective study design, recruitment of the general Japanese population with a high response rate to the questionnaire, and low proportion of follow-up loss. However, several limitations should be noted. First, our estimation of cruciferous vegetable intake did not cover the entire intake because our FFQ only covered 6 fresh and 3 pickled cruciferous vegetables.
We examined whether the cruciferous vegetable intake estimated from the FFQ was satisfactory with the use of dietary records. Estimation from the FFQ was deemed to be satisfactory because 67% (in men) and 80% (in women) of total cruciferous vegetable was covered.
Second, dietary intake obtained with the use of an FFQ may be misclassified. Even if misclassification of cruciferous vegetable intake occurred, it may have been nondifferential and would have tended to result in an underestimation of the impact of cruciferous vegetable intake on lung cancer risk. Nevertheless, we revealed a significant inverse association between total cruciferous vegetable and cabbage intake and lung cancer risk among men who were never smokers.
Third, cruciferous vegetable intake may represent healthy eating behavior and may have cofounded the association between cruciferous vegetable intake and lung cancer risk. We therefore added fruit and other vegetable intake into the multi- variate model to investigate the independent association between cruciferous vegetable intake and lung cancer risk. We also tested whether the trend changed by adding the lung cancer screening experience into the multivariate model, but the results were virtually unchanged.
Finally, in our study there may have been a residual confounding effect of smoking (e.g., years since quitting smoking) among past smokers, as well as other remaining confounding factors such as exposure to chemical pollutants.
Further epidemiologic studies with the use of urinary isothiocyanates may be necessary because the bioavailability of isothiocyanates derived from cruciferous vegetables depend on the cooking method (38, 39). Furthermore, a prospective cohort study on the cruciferous vegetable intake and glutathione S-transferase genotype interaction may be helpful to understand the detoxification mechanism. In conclusion, this prospective cohort study conducted in Japan demonstrated a protective association between cruciferous vegetable intake and lung cancer risk among men who were current nonsmokers.