This investigation is nested within one of a small number of primary lung cancer prevention trials that can examine the biological mechanisms responsible for the adverse effects of long-term β-carotene supplementation in smokers. Consistent with results from prior animal studies, we found some evidence for more aberrant cell growth in the respiratory epithelium of smokers supplemented with β-carotene compared to those randomized to placebo. These changes were indicated by modest differences in the expression of cyclin D1 in normal lung tissues between the intervention groups. We also observed somewhat suggestive, yet statistically non-significant, increases in Ki67 expression in lung tumors among cases randomized to the β-carotene versus placebo group. By contrast, β-carotene supplementation had no effect on RAR-β, CYP1A1, or AP-1 expression in our study.
In the only other study to date that has examined similar hypotheses within the context of a randomized supplementation trial, Liu et al. found no significant differences in cyclin D1, proliferating cellular nuclear antigen (PCNA), RAR-β, or CYP1A1 staining in lung tumor tissues from a small subset of β-carotene versus placebo group cases within the Physicians’ Health Study (PHS) (15
). It is important to note that, in contrast to ATBC and CARET, PHS showed no effect of β-carotene supplementation on lung cancer incidence (5
), and it is therefore not surprising that the immunohistochemical results of Liu et al. were consistent with the main trial finding.
Cyclin D1 is an important regulator of cell cycle progression in normal cells, and its overexpression leads to shortened transition through the G1 phase and uncontrolled growth and proliferation (16
). Ki67 is only expressed by proliferating cells and is therefore an established indicator of rapid cell growth (17
). Several reports have shown that Ki67 can be detected in the normal appearing lung tissue of smokers (18
), and a recent review concluded that Ki67 is an independent predictor of prognosis in NSCLC (21
). Since proliferation is a direct result of more rapid progression through the cell cycle, we expected concordant staining between the two markers, which has been previously reported (22
). Instead, we found that compared to the placebo group, men receiving β-carotene had higher cyclin D1, but not Ki67, expression in normal appearing lung epithelium and slightly higher Ki67, but not cyclin D1, expression in malignant tissue. Of note was our finding that only 6% of cases exhibited positive Ki67 immunostaining in the normal bronchiolar lung tissue, which is low when compared to several previous reports (18
). Although this discrepancy may be attributed to our threshold for positive Ki67 immunostaining – greater than 5% of counted cells in a particular target tissue – lowering the cutpoint to >3% of cells staining positive yielded nearly identical results. The discordance between cyclin D1 and Ki67 staining suggests that negative regulators of cyclin D1, such as cyclin-dependent kinase (CDK) inhibitors (including the INK4 and Cip/Kip families of proteins (23
)), may be active in lung tumor tissue, and that malignant cells are proliferating in a cyclin D1-independent manner. Additional evaluation of CDK inhibitors such as p21Cip1
would be highly informative in this regard. Alternatively, overexpression of cyclin D1 may be influencing processes unrelated to proliferation, or our findings could be due to chance.
In contrast to animal studies that showed diminished retinoid signaling – confirmed by decreases in the expression of retinoic acid receptor β (RARβ) and up-regulation of downstream targets, including activator protein-1 (AP-1) – and increased expression of CYP1A1 and CYP1A2 in ferrets supplemented with β-carotene and simultaneously exposed to tobacco smoke (10
), we did not observe an effect of β-carotene supplementation on RAR-β, AP-1, or CYP1A1 expression in the lungs of male smokers. Although we found that expression of CYP1A2 in lung tumor tissues was suggestively lower in β-carotene supplemented versus placebo cases, these differences were not statistically significant. These discrepancies could be due to subtle biological and anatomical differences between the two species and / or differences in study procedures, including specific antibodies used and route of administration of the carcinogenic agents. For example, while the entire ferret lung was exposed to tobacco smoke – yielding peripheral alveolar changes – cigarette smoke concentrates in the central airways in humans. Importantly, none of the ferrets in any exposure group developed lung tumors, although some developed squamous metaplasia (11
) – a lesion common in the bronchi of smokers. Differences could also be due to the fact that all ATBC Study participants were smokers, whereas smoke-exposed and non-smoke-exposed ferrets with or without β-carotene supplementation were evaluated in aforementioned experiments.
Loss of expression of RAR-β – a nuclear receptor that mediates the effects of retinoic acid on normal epithelial cell growth and differentiation – is a common event in lung cancer precursor lesions and overt tumors (24
), and is often ascribed to promoter methylation (28
). Contrary to expectation, we observed that RAR-β was upregulated in lung tumors as compared to normal adjacent tissue. This may be explained by failure of our antibody to distinguish between the different RAR-β isoforms; this could be important since the β4 form appears to be oncogenic whereas the β2 variant acts as a tumor suppressor (26
Our study benefited from the availability and examination of both lung tumor and adjacent normal tissue in many of the cases, although histologically normal appearing epithelial cells may not actually be truly normal due to field cancerization (29
). The small sample size may have restricted our ability to detect modest, yet significant, differences in protein expression between the β-carotene and placebo groups; in a similar vein, all stratified analyses were considered completely exploratory. Also, a comparison group supplemented with β-carotene but not having lung cancer (e.g., from a bronchoscopy series) could have been informative but was not available for study. The semi-quantitative nature of the immunohistochemical techniques employed in our study is a known limitation (30
). Finally, paraffin-embedded tissue sections may show diminished immunoreactivity over time, which could lead to false-negative staining (32
). In our study, decreased antigenicity due to extended storage could, for example, have masked subtle differences in Ki67 staining in the normal bronchial epithelium between trial intervention groups.
In summary, our findings indicate that the higher lung cancer rates observed in smokers randomized to receive supplemental β-carotene in two separate clinical trials may have been due to aberrant cell growth in the respiratory epithelium of these individuals, although our findings certainly require confirmation in CARET or other completed trials of β-carotene supplementation. Importantly, most of the molecular markers dysregulated by the combination of exposure to tobacco smoke and β-carotene supplementation in animal experiments, including RAR-β, were not similarly altered in our investigation. Identification of additional molecular markers of the adverse β-carotene effect on lung carcinogenesis is warranted.