Although vitamin A and related retinoids have been widely used for the treatment and prevention of lung cancers, our review suggests that there is a lack of evidence to support this use. Preliminary evidence including preclinical and observational studies has shown promising results with respect to lung cancer pathogenesis and disease risk, respectively, however these effects have not translated to human interventional settings. Encouraging preliminary results were found by Pastorino et al. in a setting of secondary prevention of NSCLC following surgical resection as well as by de Klerk et al in primary prevention of mesothelioma, but these have not been replicated by other groups 
. The CARET trial found adverse effects on overall lung cancer risk among smokers and asbestos workers, and this seems to be supported by similar effects among current smokers in the trial by Lippman 
There is a lack of RCTs investigating retinoids for lung cancer treatment; a single trial exists showing clinical benefit when 13 CRA, a naturally occurring retinoid, was combined IL-2, and this on surrogate immune parameters and levels of VEGF only. Two RCTs investigating the synthetic rexinoid bexarotene have shown significant survival benefits in a third of patients manifesting hypertriglyceridemia as a surrogate of sensitivity, but worse outcomes in non-responders 
. Further research is required to elucidate the best genetic predictors or biochemical surrogates of responsiveness and to confirm these post hoc findings before bexarotene can be recommended for wider clinical use 
. At present these results should not be extrapolated to other retinoids due to the considerable biochemical differences between them, however, any future research investigating retinoids, classical or not, should adopt a similar strategy of identifying a surrogate of responsiveness among a subgroup of patients.
The reasons for the observed divergence and the lack of applicability between preclinical findings and the clinical setting are not clear. It is possible that the appropriate subset of patients who might benefit from retinyl palmitate or other retinoids has not been adequately identified by the clinical research to date, as has been done in the case of bexarotene. Alternately, this divergence may arise from the inherent differences between preclinical and clinical research. One theory has been suggested that preclinical lung cancer models induced by exposure to single carcinogens such as the nitrosamine 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK) do not accurately mimic the effects of a complex carcinogen such as cigarette smoke, well accepted to be the most common cause of lung cancer 
. This inconsistency has been confirmed, for instance, in studies of selenium comparing chemoprevention in both complex- or single- carcinogen lung cancer models, wherein selenium was able to prevent lung tumors induced by NNK, but not those caused by exposure to tobacco smoke 
With respect to observational evidence, it is possible that serum retinol may be a biomarker of other active anticancer substances in the diet or of an overall “healthy” dietary pattern. In this case, interventions with vitamin A alone would be expected to fail. The biomarker theory is even more applicable to beta carotene which correlates more directly than serum retinol with intake of fruits and vegetables and what is considered an overall health-promoting diet rich in a complex array of chemopreventive substances such as flavonoids and isothiocyanates 
. This biomarker hypothesis may explain in part the failure of vitamin A as an interventional agent, and is especially true insofar as beta carotene contributes to normal (but not supraphysiological) vitamin A status.
Importantly, CARET, the largest chemoprevention trial to date of vitamin A and lung cancers, found a significantly increased risk of lung cancers in the retinol and beta carotene arm 
. This trial, consisting of 18,314 current and former smokers and asbestos workers at high risk of lung cancers, was stopped early due to interim results showing increased lung cancer risk in the active arm (retinol and beta carotene), RR 1.28 (1.04–1.57) 
. When reported according to the prespecified weighted analysis, this increased to RR 1.36 (95% CI 1.07 – 1.73) 
. Follow up results found that the risk of lung cancers and overall risk of death were elevated up to six years after the intervention was discontinued, with greatest impact occurring in women 
. Adjustment for baseline beta carotene levels did not modify the elevated risks associated with active treatment, and the effect retinol and beta carotene was greatest several years after beginning consumption 
A biomarker study was conducted in a small subgroup of the CARET population to determine the effect of supplementation on target tissue levels: bronchoalveolar lavage showed that while tissue levels of beta carotene increased significantly with supplementation, retinol levels did not 
; serum levels increased for both 
. This suggests that beta carotene, independent of its provitamin A activity, may be the culprit responsible for the deleterious effect on lung cancers observed in this trial. This hypothesis is supported by similar findings from the large Finnish Alpha Tocopherol Beta Carotene (ATBC) trial, conducted in 29,133 male smokers for 5 to 8 years. Investigators found that supplementation of 20 mg beta carotene without vitamin A increased incidence of lung cancers by 16% compared to those not receiving beta carotene, RR 1.16 (95% CI 1.02 – 1.33) 
. Serum retinol levels rose only 6% compared to placebo in those receiving beta carotene supplementation (p
. This increase in risk is roughly comparable to the 28% effect seen in CARET.
While the mechanism for the detrimental effects observed have not been fully elucidated, it is hypothesized that under conditions of high oxidative stress and exposure to lung irritants such as those associated with exposure to cigarette smoke and asbestos, certain antioxidants may in fact act as conditional pro-oxidants 
. According to the conditional pro-oxidant hypothesis, the activity of an antioxidant depends on its redox potential in relation to other pro- and anti- oxidants in its microenvironment. Carotenoids are particularly vulnerable to such oxidation due to their long chains of conjugated double bonds, and are known to concentrate in the lungs 
. New evidence suggests that carotene breakdown products (CBPs) produced by high dose administration under conditions of oxidative stress may act as pro-oxidants, impairing mitochondrial function and resulting in cellular damage, thereby predisposing to carcinogenesis 
Limitations of the chemoprevention studies listed here include lack of a single-agent intervention arm in CARET, and lack of a placebo arm in the Western Perth study to distinguish the potentially differing effects of beta carotene and retinol. CARET also used a synthetic form of beta carotene. As discussed above, while beta carotene is a precursor to retinol in the body, it is nonetheless physiologically distinct in its own right, possessing redox activity independent of vitamin A 
. Also, while CBPs have been shown to impair mitochondrial function, preliminary evidence suggests that retinol may be an essential response modifier and cofactor of mitochondrial energy production 
. The limitations of these chemoprevention trials, however, impede our ability to draw any definite conclusions about the differential effects of beta carotene and retinoids on lung cancers in humans.
Notably, a 2008 systematic review and meta- analysis focused exclusively on beta carotene in relation to cancer risk. Investigators screened 22,994 records and included six RCTs plus 30 prospective observational studies . The RCTs included in this review, three of which have already been described above, were: 1) the Carotene and Retinol Efficacy Trial (CARET) 
; 2) the Alpha Tocopherol Beta Carotene (ATBC) study 
; 3) the Physicians' Health Study (PHS) 
; 4) the Western Perth, Australia study 
; 5) the Women's Health Study 
; and 6) the Linxian General Population trial 
. The pooled RR for studies comparing beta carotene supplements to placebo (n
3) was 1.10 (95% CI 0.89 – 1.36) 
. In trials conducted in high risk populations comprised of smokers and asbestos workers (n
2), there were significant increases in lung cancer risk associated with beta carotene supplementation: in the ATBC trial, RR 1.17 (95% CI 1.02 – 1.34), and in the CARET trial 1.36 (1.07 – 1.72) 
. In contrast to this interventional data but in agreement with the biomarker hypothesis, analysis of observational data found that the pooled RR for lung cancers was decreased in the highest versus the lowest category of total dietary carotenoid intake for studies reporting smoking-adjusted risk was 0.79 (95% CI 0.71 – 0.87) 
. For intake of beta carotene specifically, there was a non-significant 8% reduction in smoking-adjusted risk, RR 0.92 (95% CI 0.83 – 1.01) 
This analysis is limited by an exclusive focus on lung cancers and as such we are unable to draw any conclusions about the application of retinoids in other cancer types. It is important to note that there is possible benefits from retinoids in conditions such as head and neck cancer, mesothelioma, and certain premalignant conditions. ATRA combined with chemotherapy has been successful in treating acute promyelocytic leukemia (APL), inducing remission in what was once a highly lethal leukemia 
. Reports have shown that classical retinoids including 13CRA and ATRA may be effective for treatment of premalignant lesions including leukoplakia, actinic keratosis, and cervical dysplasia 
. In controlled trials on prevention of head and neck cancer, retinyl palmitate, 13 CRA, and fenretinide have demonstrated significant response rates 
With respect to mesothelioma, cancer of the pleural lining of the lungs, one of the studies reviewed here under primary prevention found a significantly decreased risk of mesothelioma in the retinol group, RR 0.24 (0.07-0.86) 
. A limitation of this study was the lack of a placebo comparator arm; retinol was instead compared to beta carotene. It is possible that beta carotene had an overall adverse effect on disease risk, thereby producing a falsely positive effect for retinol in comparison, however, this is impossible to determine.
Strengths of this analysis include a comprehensive and systematic survey of the literature with a clear focus on lung cancers. All levels of evidence, human, observational, and preclinical were included to achieve a broad analysis of anticancer activity of vitamin A/retinoids in lung cancers, and any possible interactions with chemotherapy and/or radiation therapy to assess safety alongside efficacy. We are not aware of another review of vitamin A for lung cancers to date that has included such an extensive analysis of the available data.