点击显示 收起
1 From the Departments of Nutrition and Epidemiology, Harvard School of Public Health, Boston, and the Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston.
See corresponding editorial on page 901.
ABSTRACT
Background: Carotenoids may reduce lung carcinogenesis because of their antioxidant properties; however, few studies have examined the relation between intakes of individual carotenoids and lung cancer risk.
Objective: The aim of this study was to examine the relation between lung cancer risk and intakes of -carotene, ß-carotene, lutein, lycopene, and ß-cryptoxanthin in 2 large cohorts.
Design: During a 10-y follow-up period, 275 new cases of lung cancer were diagnosed in 46924 men; during a 12-y follow-up period, 519 new cases were diagnosed in 77283 women. Carotenoid intakes were derived from the reported consumption of fruit and vegetables on food-frequency questionnaires administered at baseline and during follow-up. The data were analyzed separately for each cohort and the results were pooled to compute overall relative risks (RRs).
Results: In the pooled analyses, -carotene and lycopene intakes were significantly associated with a lower risk of lung cancer; the association with ß-carotene, lutein, and ß-cryptoxanthin intakes were inverse but not significant. Lung cancer risk was significantly lower in subjects who consumed a diet high in a variety of carotenoids (RR: 0.68; 95% CI: 0.49, 0.94 for highest compared with lowest total carotenoid score category). Inverse associations were strongest after a 4–8-y lag between dietary assessment and date of diagnosis. In subjects who never smoked, a 63% lower incidence of lung cancer was observed for the top compared with the bottom quintile of -carotene intake (RR: 0.37; 95% CI: 0.18, 0.77).
Conclusion: Data from 2 cohort studies suggest that several carotenoids may reduce the risk of lung cancer.
Key Words: Carotenoids • fruit • vegetables • lung cancer • vitamins • cohort studies • epidemiology • men and women
INTRODUCTION
More men and women die of lung cancer than of any other cancer in the United States (1). Data from epidemiologic studies generally support a protective effect of fruit and vegetable consumption on lung cancer risk (2). ß-Carotene was hypothesized to account for this benefit, but recent randomized intervention trials of supplementary ß-carotene did not confirm this (3–5). Fruit and vegetables contain numerous compounds besides ß-carotene that have anticarcinogenic properties in vitro and in animal experiments (6). Many studies examined the relation between fruit and vegetable intake and the risk of lung cancer (2), but the nutrients or phytochemicals that may be responsible for these associations remain obscure. An understanding of the active components that may be beneficial would allow for more focused interventions.
Carotenoids are a group of phytochemicals that have strong singlet oxygen quenching and antioxidant properties (7, 8) and that therefore might protect biological structures from oxidative damage. Inverse associations were observed consistently in case-control and cohort studies between intake or serum concentrations of ß-carotene and the incidence of lung cancer (2, 9). However, until recently, nutrient databases did not provide information on other specific carotenoid concentrations in foods. Updated nutrient databases now provide the opportunity to examine intakes of -carotene, ß-carotene, lutein, lycopene, and ß-cryptoxanthin in relation to lung cancer risk. Several recent studies showed inverse associations for intake of -carotene, suggesting that -carotene is partly responsible for the apparent protection conferred by fruit and vegetables against lung cancer (10–13).
In a report based on the 1980–1992 follow-up of the Nurses' Health Study (NHS) cohort of women, a strong inverse association was observed for lung cancer and intake of -carotene but not 4 other carotenoids (13). We currently have data from 10 y of follow-up in the Health Professionals Follow-Up Study (HPFS), a cohort of men, and additional data from the NHS, to examine the associations between intakes of individual carotenoids and lung cancer risk in more depth. In the analysis of the NHS cohort, we used follow-up data from 1984 to 1996 because a more comprehensive measurement of carotenoid-rich foods was initiated with the 1984 food-frequency questionnaire (FFQ). The long follow-up period and repeated dietary assessments available in each cohort allowed us to perform alternative analyses to examine temporal relations between intakes of -carotene, ß-carotene, lutein, lycopene, and ß-cryptoxanthin and lung cancer risk. In addition, given the size of this study, we were able to examine the associations between intakes of individual carotenoids and lung cancer risk by smoking status and histologic subtypes.
SUBJECTS AND METHODS
Study population
The HPFS and the NHS provided data for this analysis. The HPFS was initiated in 1986 when 51529 US men aged 40–75 y responded to a mailed questionnaire. Fifty-eight percent of the men in the HPFS cohort were dentists; the remainder of the cohort included optometrists, osteopaths, podiatrists, pharmacists, and veterinarians. The NHS began in 1976 when 121700 female registered nurses aged 30–55 y responded to a mailed questionnaire. Detailed information on individual characteristics and habits, such as age, marital status, weight, height, medical history, medication use, smoking status, and physical activity, was obtained through the mailed questionnaires at baseline and subsequently on a biennial basis. The HPFS was approved by the Human Subjects Committee of the Harvard School of Public Health, and the NHS was approved by the Human Research Committee at the Brigham and Women's Hospital.
For each of the follow-up questionnaires, up to 6 mailings were sent to nonrespondents. Most of the deaths in these 2 cohorts were reported by family members or by the postal service in response to the follow-up questionnaires. In addition, the National Death Index was searched for nonrespondents; this method has been shown to have a sensitivity of >96.5% (14).
Dietary assessment
The baseline questionnaire for the HPFS participants (1986) included a 131-item semiquantitative FFQ. We mailed a 61-item FFQ to the NHS participants in 1980. In 1984 and 1986, more comprehensive FFQs (116 items) were mailed to the women; these included more fruit and vegetable items than did the 1980 questionnaire and were similar to those in the FFQ mailed to the men in 1986. Because the initial 1980 FFQ in the NHS included fewer fruit and vegetable items than did the men's questionnaires and was missing some important contributors to specific carotenoid intakes, follow-up for the NHS analysis was started in 1984, when the more comprehensive FFQ was completed. Both cohorts completed the 131-item FFQ in 1990.
On the questionnaire, participants were asked to report their average frequency of intake over the previous year of a specified serving size of each food. Individual nutrient intakes were calculated by multiplying the frequency of consumption of each food by the nutrient content of the specified portion size (obtained from the US Department of Agriculture and supplemented by other publications) and then summing the contributions from all foods. For carotenoid values, we used the new US Department of Agriculture–National Cancer Institute (USDA-NCI) database that was developed for fruit and vegetables (15) and included data from a publication on tomato-based food products (16). "Lutein intake" represents a combination of both lutein and zeaxanthin intakes because the main food contributors of these estimates contain essentially no zeaxanthin; thus, lutein and zeaxanthin values for these foods represent primarily lutein (15).
A validity study of the FFQ in 127 men from the HPFS cohort indicated that most foods are reported reasonably well; Pearson correlations between the average intake assessed by two 1-wk diet records completed 6 mo apart and one FFQ ranged from 0.25 to 0.95 for specific fruit and vegetables, and the median correlations for fruit and vegetables were 0.75 and 0.46, respectively (17). A similar validation study in the NHS showed correlations of between 0.16 and 0.74 for fruit and vegetables (18). Intakes of carotenoids (-carotene, ß-carotene, lutein, lycopene, and ß-cryptoxanthin) were reported previously to correlate with specific carotenoid plasma concentrations in a subset of the HPFS (range: 0.35–0.47) (19). For the same comparisons, the correlations were slightly lower in the NHS (range: 0.21–0.48) (19).
In addition to the carotenoid values calculated with use of the USDA-NCI database, we computed an empirical score for each carotenoid that took into account the variations in carotenoid bioavailability among foods. We calculated these scores by using the regression coefficients for the foods with the most significant associations (P 0.10) in stepwise regression models with plasma carotenoid concentration as the dependent variable, as reported in a previous study among subsamples of the 2 cohorts (19). For example, the -carotene score for each participant was calculated by multiplying the participant's reported frequency of intake of carrots and homemade soup by their respective regression coefficients and summing them together (carrots and homemade soup were the only 2 foods significantly associated with plasma -carotene).
A variable for total carotenoid intake (referred to as total carotenoid score) was created by using the quintile values of each individual carotenoid. We chose this method, rather than using intake in micrograms, because each carotenoid is consumed in varying quantities and it is untenable that any biochemical actions of diverse carotenoids would be identical on a gram basis in a biologically complex system such as a cell. The top category of total carotenoid intake was created for subjects consuming high amounts (fifth quintile) of 3 of the individual carotenoids, whereas subjects consuming low amounts (bottom quintile) of 3 of the individual carotenoids were placed in the lowest category of total carotenoid intake. The middle 3 categories were evenly divided into 3 groups. Creating categories in this fashion allowed for a more diverse intake of carotenoids than would a simple quintile analysis.
Smoking history
Smoking status and history were obtained at baseline and in all subsequent questionnaires in both cohorts. Current smokers also reported the intensity of smoking (average number of cigarettes smoked/d) on each questionnaire. Past smokers reported on the baseline questionnaire when they last smoked; time since quitting was also calculated for participants who quit smoking during follow-up. On the 1976 questionnaire, the NHS participants were asked at what age they started smoking. In the HPFS, each participant reported on the baseline questionnaire the average number of cigarettes smoked/d for each decade of life. From these data, categories were derived for age at start of smoking, based on the earliest decade in which smoking was recorded, because there was no question (on the HPFS questionnaires) asking at what specific age the participant had started smoking.
Identification of lung cancer cases
In both cohorts, the participants were asked to report specified medical conditions, including cancers that were diagnosed in the 2-y period between each follow-up questionnaire. Whenever a participant (or next of kin for decedents) reported a diagnosis of lung cancer, we asked for permission to obtain related medical records or pathology reports. When permission to obtain such records was denied, we attempted to confirm the self-reported cancer with an additional letter or telephone call to the participant. If the primary cause of death as reported by a death certificate was previously unreported lung cancer, we contacted a family member to obtain permission to retrieve medical records or at least to confirm the diagnosis of lung cancer. When medical records were available, we classified lung cancer cases by histologic type (squamous-cell, adenocarcinoma, large-cell undifferentiated, bronchioalveolar, small-cell, or other). In the participants included in these analyses, 275 new cases of lung cancer were diagnosed in the HPFS between 1986 and 1996 and 519 new lung cancer cases were diagnosed in the NHS between 1984 and 1996. Ninety-three percent of the lung cancer cases in the HPFS and 88% of the cases in the NHS were confirmed with medical records.
Statistical analysis
All participants with cancer (other than nonmelanoma skin cancer) that was diagnosed before baseline (in 1986 for the HPFS cohort and in 1984 for the NHS cohort) and all participants with implausibly high or low scores for total energy intake on the baseline questionnaires were excluded before the start of the analyses. After exclusion, 46924 men and 77283 women remained available for the analyses. For men and women, we computed person-time of follow-up for each participant from the return date of the baseline questionnaire to the date of lung cancer diagnosis, death from any cause, or the end of follow-up (31 January 1996 for men and 30 June 1996 for women), whichever came first. Other cancer cases (except nonmelanoma skin cancer) were also excluded during follow-up because, given that cancer is a systemic disease, persons with cancer are likely to change their dietary habits, and most analyses are based on cumulative dietary intakes.
Incidence rates of lung cancer were calculated by dividing the number of incident cases by the number of person-years in each category of dietary intake. We computed the relative risk (RR) for each of the upper categories of intake by dividing the rates in these categories by the rate in the lowest category of intake.
For each cohort, RRs with adjustments for potential confounders were estimated by using pooled logistic regression analyses with 2-y time increments (20). In these models, age was categorized into 5-y age groups; total energy intake was divided into quintiles; age at initiation of smoking was grouped into the categories <15, 15–19, 20–29, and 30 y; and smoking status was grouped into the categories never smokers, past smokers (with categories of time since quitting of <5, 5–14, and 15 y in the NHS cohort and <10 and 10 y in the HPFS cohort), and current smokers (with categories of 1–4, 5–14, 15–24, 25–34, 35–44, and 45 cigarettes smoked/d). We updated age, smoking status, and time since quitting smoking biennially. All carotenoid values were adjusted for total energy intake by using the residual values of each carotenoid when regressed on total energy intake (21). For the cumulative updated analysis in the NHS cohort, carotenoid data from the 1984 questionnaire was used to allocate person-time to each of the quintiles of exposure between 1984 and 1986; the average of the 1984 and 1986 intakes was used from 1986 to 1990; and the average of the 1984, 1986, and 1990 intakes was used for subsequent years (1990 through 1996). Similarly, in the HPFS cohort we used the 1986 intakes between 1986 and 1990 and the average of the 1986 and 1990 intakes for subsequent years (1990 through 1996). We ascertained trends by assigning median values to each exposure category and modeling these variables as continuous variables.
Three lagged analyses were performed to examine the temporal aspects of carotenoid intakes and diagnosis of lung cancer. Effects of a recent diet were examined by using a 0–4-y lag analysis; for example, in the HPFS cohort, 1986 diet was used for the follow-up period between 1986 through end of 1989, 1990 diet was used for follow-up period from 1990 to end of 1993, and 1994 diet was used for the last 1994–1996 period. Longer-term effects of diet were examined by using 4–8-y and 8-y lagged analyses. For the 4–8-y lagged analysis, the 1986 diet was used for follow-up between 1990 and the end of 1993, and the 1990 diet was used between 1994 and 1996. For the 8-y lagged analyses, the 1986 diet was used for follow-up between 1994 and 1996 in the HPFS cohort. Analyses were performed in a similar manner in the NHS cohort.
We pooled the data from the 2 cohorts by using a random-effects model for the log of the RRs (22). Tests of heterogeneity using the Q statistic (22) were performed for continuous variables to evaluate the overall trend before the data were pooled.
RESULTS
Although strong inverse associations were observed between the individual carotenoids and lung cancer in the age-adjusted models (Table 1), substantial attenuation was observed after adjustment for smoking status (never, past with time since quitting, current number of cigarettes smoked, and age at start). The results were similar when we used total cigarette pack-years for the smoking variable (data not shown). In the pooled multivariate analysis of the 2 cohorts, we observed significant 20–25% lower risks of lung cancer in participants consuming high compared with low amounts of -carotene and lycopene (Table 1; results of tests for heterogeneity across the 2 cohorts were not significant). High intakes of ß-carotene, lutein, and ß-cryptoxanthin were associated with nonsignificant 10–19% lower risks of lung cancer. In the individual cohorts, most carotenoids were inversely associated with lung cancer, although the associations were stronger in women than in men and only -carotene in the women was significant (Table 1). Additionally, controlling for intakes of vitamin E, vitamin C, or folate did not influence the estimates observed in the multivariate analyses (data not shown).
View this table:
TABLE 1.. Intake of specific carotenoids and incidence of lung cancer in men in the HPFS (1986–1996) and women in the NHS (1984–1996) on the basis of cumulative updated diet1
In each cohort, when we created a total carotenoid score based on the quintile of intake of each individual carotenoid (see Methods), an inverse association was observed with this total carotenoid score (Table 1). Compared with a low total intake of carotenoids, a high intake was associated with a significant 32% lower lung cancer risk in the pooled analysis (Table 1). As with the individual carotenoids, there was substantial confounding by smoking status.
When we derived a bioavailability-adjusted score for each of the 5 carotenoids (see Methods), relations between the score estimates and lung cancer risk were not substantially different from the main analyses for intakes of -carotene, ß-carotene, and ß-cryptoxanthin (data not shown). However, the associations between the lycopene scores and the risk of lung cancer were stronger in each cohort (pooled multivariate RR: 0.62; 95% CI: 0.42, 0.92, comparing top with bottom categories of lycopene score) than were those using estimates of lycopene intake based on the food-composition database. Tomato sauce was the primary contributor of the lycopene score for each cohort (19). The lutein scores were not inversely associated with risk of lung cancer in either cohort.
Of the carotenoids, only ß-carotene was generally available as a supplement during the years of this study. ß-Carotene supplements, used by 2% of the men in 1986 and <1% of the women in 1984, were not associated with risk of lung cancer (multivariate RR in the HPFS cohort: 0.82; 95% CI: 0.36, 1.85; multivariate RR in the NHS cohort: 1.23; 95% CI: 0.55, 2.76), and no associations were observed for use of multivitamin supplements, which sometimes contained ß-carotene. The association between dietary ß-carotene and lung cancer risk was similar to that for ß-carotene intakes (Table 1) after the removal of participants who took ß-carotene supplements (data not shown).
Because the various carotenoids arise from similar food groups, their intakes are highly correlated (especially - and ß-carotene; r = 0.84 in the HPFS cohort). To determine their independent effects, we modeled simultaneously each carotenoid as a continuous variable and computed RRs for incremental units chosen to be the approximate difference between the medians of the top and bottom quintiles from the NHS, for which the range of carotenoid intakes was narrower. After other carotenoids were controlled for, the specific associations were not appreciably altered, with the exception of -carotene, which became slightly more inverse in men (RR: 0.83 compared with 0.88), and of ß-carotene, which became slightly positive in both cohorts (pooled RR: 1.07 compared with 0.90).
To examine the temporal relation between exposure and disease outcome, we performed several analyses with different lag periods. The results from both cohorts were similar in these analyses and the results of tests for heterogeneity across the 2 cohorts were not significant. We observed strong inverse associations for each of the 5 carotenoids in analyses that allowed for a lag of 4–8 y between time of dietary assessment and follow-up time (pooled data are presented in Table 2). In contrast, no associations were observed for the carotenoids in the analyses using short follow-up periods (0–4-y lag). In the 8-y lag analysis, weak inverse associations were observed for intakes of - and ß-carotene but not for intakes of the other carotenoids.
View this table:
TABLE 2.. Pooled multivariate relative risks (RRs) of lung cancer and 95% CIs by carotenoid intake in different lag analyses in the HPFS and NHS cohorts1
In a separate analysis, we examined the associations between specific carotenoids and lung cancer stratified by smoking status in a separate analysis. The data from the 2 cohorts were pooled for each carotenoid because there was no evidence of heterogeneity (Table 3). Although there was no association between -carotene intake and lung cancer in current smokers and past smokers, a strong inverse association with -carotene intake was observed among never smokers when the cohorts were combined. The multivariate RR for the top compared with the bottom quintile of -carotene intake from these pooled data was 0.37 (P for trend: 0.007). Other carotenoids were also inversely associated with lung cancer in never smokers, but less than was -carotene. Current smokers in the highest quintile of lycopene intake had a significantly lower risk of lung cancer than did those in the lowest quintile, but other specific carotenoid intakes were not associated with lung cancer (Table 3). Overall, past smokers had null to weak inverse associations with the specific carotenoid intakes.
View this table:
TABLE 3.. Pooled multivariate relative risks (RRs) and 95% CIs (for top quintile) for incidence of lung cancer associated with intake of specific carotenoids, by smoking status in the HPFS (1986–1996) and NHS (1984–1996) on the basis of cumulative updated diet1
Because women in the NHS cohort were slightly younger than were men in the HPFS cohort (median age at baseline: 50 y in the NHS cohort and 54 y in the HPFS cohort), we stratified men into 2 age groups (<65 and 65 y) and reexamined the carotenoid-lung associations. With the exception of ß-cryptoxanthin, associations were inversely related to lung cancer risk in the younger age group in a comparison of the top and bottom quintiles (multivariate RRs: 0.68, 0.73, 0.76, and 0.87 for lycopene, -carotene, lutein, and ß-carotene, respectively), but not in the older age group (multivariate RRs: 0.95–1.27).
We further examined carotenoid associations by histologic subtype. Two histologic groups were created for this analysis following the Kreyberg groupings (23) (group I: squamous-cell, large-cell, and small-cell carcinomas; group II: adenocarcinoma). Results for these 2 groupings were not entirely homogeneous for the 2 cohorts; thus, the data were not pooled. In the men, associations between specific carotenoids and adenocarcinoma were either null or positive in a comparison of top and bottom quintiles, with the exception of ß-cryptoxanthin (RR: 0.57; 95% CI: 0.29, 1.10). In women, the associations with adenocarcinoma were mostly inverse in a comparison of the top and bottom quintiles (RR: 0.66, 0.77, 0.78, 0.83, and 0.78 for -carotene, ß-carotene, lutein, lycopene, and ß-cryptoxanthin, respectively). For squamous-cell, small-cell, and large-cell carcinomas (group I), both cohorts showed suggestive inverse associations for each specific carotenoid (data not shown).
DISCUSSION
-Carotene and lycopene intakes and diets high in a variety of carotenoids were significantly associated with a lower risk of lung cancer after data were pooled from 2 ongoing prospective cohort studies. The associations for ß-carotene, lutein, and ß-cryptoxanthin were inverse but not significant. The strongest associations were observed when follow-up time was started after a lag of 4–8 y from the time of dietary assessment. Among never smokers, a significant 63% lower risk of lung cancer was observed for the top compared with the bottom quintile of intake of -carotene, and nonsignificant inverse associations were observed for other carotenoid intakes. Among current smokers, a significant inverse association was observed for lycopene intake and lung cancer risk, but no associations were observed for the other carotenoids.
The marked differences between the age-adjusted and multivariate RRs of lung cancer for the various carotenoids show substantial confounding by smoking history. Intakes of all carotenoids except lycopene were significantly associated with lower lung cancer risk in the age-adjusted analyses. Additional factors that influence lung cancer risk, such as passive smoke exposure and depth of inhalation, potentially confound RR estimates. Given the strength of the association between smoking and lung cancer, residual confounding by smoking cannot easily be dismissed.
Recently, a cohort study in Finland had results similar to ours: -carotene intake had a significant inverse association with lung cancer risk and suggestive inverse associations were observed for ß-carotene, lutein, and ß-cryptoxanthin, but not for lycopene (10). All 3 case-control studies (11, 12, 24) that examined specific carotenoids from foods showed inverse associations for -carotene in a comparison of the top and bottom quintiles (RR: 0.4–0.6), although the findings were significant in only 2 studies (11, 12). Lutein and ß-carotene were inversely associated in 2 of these studies (11, 12); the associations were less consistent for lycopene and ß-cryptoxanthin.
Studies examining relations between specific carotenoids and lung cancer risk are few because, until recently, detailed data on the carotenoid compositions of foods were not available. However, a substantial number of previous studies showed inverse associations for intakes of carrots and tomatoes and the risk of lung cancer, although results were not always significant (13, 25–34). Given that carrots and tomatoes are by far the major contributors of -carotene and lycopene, respectively, the results of these studies support our findings, although other compounds in these foods may be important.
Because dietary data are obtained repeatedly over follow-up years in our cohorts, we were able to examine temporal aspects of the carotenoid-lung associations. Our findings support a relatively long latency period for carotenoids and lung cancer risk; no associations were observed when diet was assessed within 4 y of diagnosis. Intakes of lutein, lycopene, and ß-cryptoxanthin 8 y before lung cancer diagnosis were not related to lung cancer risk, and and ß-carotene intakes were only weakly associated. These findings indicate that perhaps the critical period of exposure to carotenoids, to maximize protection, is 4–8 y before disease. Our findings suggest that observational studies and clinical trials of short durations may not detect associations. However, it is also possible that smaller numbers in the long latency analyses may have limited our ability to detect an association or, alternatively, that residual confounding by smoking is stronger in the 4–8-y lagged analyses, especially because most of the effect was observed in the top quintiles.
In this study, among subjects who were never smokers, those who consumed high amounts of -carotene had a substantially lower risk of lung cancer than did those who consumed low amounts. A Finnish cohort study also showed a strong inverse association between -carotene intake and lung cancer in nonsmokers in a comparison of the top and bottom tertiles (RR: 0.33; 95% CI: 0.11, 1.02) (10). Four case-control studies that examined the associations between intake of fruit and vegetables and lung cancer risk in never smokers (29, 32, 35, 36) found an association between a high carrot intake and a lower risk of lung cancer. Therefore, persons who have never smoked may benefit from a high carrot or -carotene intake.
Antioxidant properties of carotenoids have been hypothesized to play an important role in carcinogenesis (37, 38). In vitro studies indicated that lycopene is the most efficient carotenoid scavenger of free radicals (7) and that lycopene is more efficient than is ß-carotene at preventing cell membrane damage from the nitrogen dioxide radical found in cigarette smoke (39). Lycopene, however, does not exhibit any provitamin A activity, in contrast with -carotene, ß-carotene, and ß-cryptoxanthin (37). In addition to the known biological mechanisms, it is possible that carotenoids interact with each other, or with other antioxidants, to prevent oxidant injury (40).
Although food-composition tables (41) provide detailed information on the carotenoid contents of many foods, they do not reflect the carotenoids' bioavailabilities. Because bioavailability can vary substantially depending on the cooking method used, the presence of other nutrients, and other factors (42, 43), an appropriate measure might be one that is based on bioavailability, such as blood concentrations. In this study, we observed a stronger association with the lycopene bioavailability-adjusted score than with the lycopene value obtained from the food-composition tables. Previous studies indicated that lycopene is more bioavailable in cooked products than in raw products (43). Our data suggest that lycopene is an important carotenoid for protection against lung cancer, especially in current smokers. Some studies may have missed associations with lycopene intake if the participants were eating few tomato-based products (ie, cooked tomatoes) (43) or if the dietary assessment did not separate high- and low-bioavailability sources of lycopene.
Our findings for total carotenoid intake score indicate that a high intake of a variety of carotenoids reduces the risk of lung cancer. A similar observation was noted previously in a case-control study in which a high intake of 3 carotenoids (-carotene, ß-carotene, and lutein) was associated with the lowest risk of lung cancer (11). These observations may result from either a very high total carotenoid intake or a complementary effect of the different carotenoids. Structural and polarity differences of carotenoids not only determine their reactivity (44) but also dictate where they accumulate in cell structures (45). Therefore, the combination of several carotenoids may be more beneficial than are high concentrations of one individual carotenoid.
Carotenoids are just a few of the many phytochemicals found in fruit and vegetables (6). Although the overall data suggest that carotenoids may contribute to lower the risk of lung cancer, the possibility remains that other phytochemicals found in similar foods are equally or more important in reducing carcinogenesis. It would be unwise to suggest carotenoid supplementation as an alternative to a diet plentiful in fruit and vegetables.
Results from this study suggest that overall intake of a diversity of carotenoid-rich foods is inversely related to lung cancer risk. Among never smokers, inverse associations were observed for each carotenoid, and a substantial decrease in risk was observed among those consuming high amounts of -carotene. Thus, a diet high in carotenoid-containing foods, including carrots and tomato-based products, may confer some benefits, particularly in nonsmokers. Because <10% of all lung cancers occur in nonsmokers, the public health message must continue to be that smoking cessation is by far the best way to reduce the incidence of lung cancer.
REFERENCES