Carotenoids are plant pigments that are found primarily in a variety of fruits and vegetables. The most commonly ingested carotenoids in human populations include beta-carotene, alpha-carotene, lycopene, lutein and zeaxanthin, and beta-cryptoxanthin. Intake of carotenoids is highly correlated with intake of fruits and vegetables; carotenoids as measured in blood or other tissues are considered one of the best biomarkers for fruit and vegetable intake [15, 16]. Given this, it is not surprising that epidemiologic studies generally find inverse associations between consumption of carotenoids or blood levels of carotenoids and lung cancer risk, as reviewed elsewhere [4, 17].
Numerous studies have examined the association between the carotenoid beta-carotene, in diet or in serum/plasma, and lung cancer risk. A recent review of this literature by the International Agency for Research on Cancer  concluded that the association of lower serum or plasma beta-carotene with lung cancer risk was remarkably consistent. For example, three of three cohort studies, six of seven nested case-control studies, and five of five case-control studies reported an inverse association between plasma or serum beta-carotene concentrations and lung cancer risk. There was evidence of a dose–response in 12 of the 15 total studies reviewed. Considering estimated intake of carotenoids in diet, in nearly all studies reviewed, risk for lung cancer was lower among people with a high dietary intake of beta-carotene or carotenoids . Some studies have used recently available food composition databases for carotenoids to estimate consumption of the major dietary carotenoids. In these studies, no particular carotenoid has consistently emerged as being associated with reduced risk of lung cancer. Some studies have suggested that higher intake of alpha-carotene, in particular, seems to be more strongly associated with lower lung cancer risk [18, 19]. Other studies have reported comparable inverse associations for alpha-carotene, beta-carotene, and lutein . However, other studies have observed no association with alpha-carotene but inverse associations with beta-cryptoxanthin and lutein + zeaxanthin , or with lycopene . The majority of studies, however, found stronger inverse trends with vegetable and fruit intake than with estimated carotenoid intake . Le Marchand et al. , for example, concluded that their data supported greater protection afforded by consuming a variety of vegetables compared to only foods rich in a particular carotenoid.
Certain carotenoids, known as provitamin A carotenoids, can be metabolically converted into retinol. Provitamin A carotenoids include beta-carotene, alpha-carotene, and beta-cryptoxanthin. Many of the early epidemiologic studies of diet and lung cancer examined the association of total vitamin A in the diet with lung cancer risk . In 1975, for example, Bjelke  noted an association between dietary vitamin A and human lung cancer risk. As investigators began to examine associations separately for carotenoids versus retinol (preformed vitamin A), it became apparent that inverse associations with vitamin A were largely being driven by provitamin A carotenoids . That is, the evidence from observational studies linking retinol with reduced lung cancer risk is inconsistent and weak at present.
Given the consistency of the results of epidemiologic studies on beta-carotene, coupled with chemopreventive efficacy of beta-carotene in animal models of skin carcinogenesis and buccal pouch carcinogenesis , several intervention trials of beta-carotene for the prevention of lung and other cancers were implemented in the 1980s and early 1990s. The first lung cancer prevention trial involving beta-carotene to be completed was the Alpha-Tocopherol Beta-Carotene (ATBC) Study , which involved 29,133 Finnish males ages 50–69 who were heavy cigarette smokers at entry (average one pack/day for 36 years). The study design was a 2 × 2 factorial with participants randomized to receive either supplemental beta-carotene (20 mg/day), alpha-tocopherol (50 mg/day), the combination, or placebo for 5–8 years. Unexpectedly, participants receiving beta-carotene (alone or in combination with alpha-tocopherol) had a statistically significant 18% increase in lung cancer incidence and 8% increase in total mortality relative to participants receiving placebo. Supplemental beta-carotene did not appear to affect the incidence of other major cancers occurring in this population.
The Carotene and Retinol Efficacy Trial (CARET) was a multicenter lung cancer prevention trial of supplemental beta-carotene (30 mg/day) plus retinyl palmitate (25,000 IU/day) versus placebo in asbestos workers and smokers . CARET was terminated nearly 2 years early in January 1996, because interim analyses of the data indicated that, should the trial have continued for its planned duration, it is highly unlikely that the intervention would have been found to be beneficial. Furthermore, the interim results indicated that the supplemented group was developing more lung cancer, not less, consistent with the results of the ATBC trial. Overall, lung cancer incidence and total mortality were significantly increased by 28% and 17%, respectively, in the supplemented subjects. The increase in lung cancer following supplementation with beta-carotene and retinyl palmitate was observed for current, but not former, smokers.
In contrast to these findings are the results of the Physicians’ Health Study (PHS) of supplemental beta-carotene versus placebo in 22,071 U.S. male physicians . There was no effect—positive or negative—after 12 years of supplementation with beta-carotene (50 mg every other day) on total cancer, lung cancer, or cardiovascular disease. The relative risk for lung cancer was reduced by a nonsignificant 10% in current smokers randomized to beta-carotene and a nonsignificant 22% in nonsmokers randomized to beta-carotene as compared to placebo. The apparent lack of an effect of long-term supplementation of beta-carotene on lung cancer incidence, even in baseline smokers who were administered the supplements for up to 12 years, is noteworthy. A similar lack of effect of supplemental beta-carotene on overall cancer incidence was seen in the Women's Health Study , although the duration of intervention was short (median 2.1 years).
A clear mechanism to explain the apparent enhancement of lung carcinogenesis by supplemental beta-carotene, alone or in combination with retinol, in smokers has yet to emerge. As detailed elsewhere , it should be noted that the two trials that observed this enhancing effect [24, 25] had higher median plasma beta-carotene concentrations in their intervention groups relative to trials that did not observe an enhancing effect on lung cancer [26, 29]. Thus, it is possible that high tissue concentrations of beta-carotene in the presence of strongly oxidative tobacco smoke cause an interaction that promotes carcinogenesis. A recent animal study has suggested that this effect might be mediated by altered retinoid signaling .
The surprising results of the intervention trials involving beta-carotene and lung cancer prevention emphasize the value of results from randomized intervention trials prior to establishing public policy on the basis of observational data. Many have interpreted the observational data as being contradictory with the intervention trial results, but they really are not contradictory when it is recognized that the observational data that are derived from fruits and vegetables reflect relatively low doses of carotenoids in a complex matrix involving many other compounds, and generally reflect dietary patterns that may have been in existence for decades. The trials, in contrast, reflect one specific carotenoid given in a highly bioavailable preparation for a relatively short time period, and administered relatively late in the carcinogenic process to a high-risk group of subjects. The intervention trial data involving high-dose supplemental beta-carotene should not be interpreted as evidence against possible benefits of fruits and vegetables; there are currently no data to suggest that fruits and vegetables might have adverse effects with regard to lung cancer.