|Home | About | Journals | Submit | Contact Us | Français|
β-Carotene supplementation showed neither benefit nor harm among apparently healthy male physicians in the Physicians’ Health Study (PHS) trial. The present investigation was to evaluate how long-term β-carotene supplementation affects molecular markers of lung carcinogenesis in the PHS.
The protein levels of total p53, cyclin D1, proliferating cellular nuclear antigen (PCNA), retinoic acid receptor β (RARβ), and cytochrome p450 enzyme 1A1 (CYP1A1) were measured using the immunohistochemical method in the 40 available archival lung tissue samples from patients diagnosed with lung cancer in the PHS. The protein levels of these markers were compared by category of β-carotene treatment assignment and other characteristics using unconditional logistic regression models.
The lung tumor positivity for total p53, RARβ, cyclin D1, and PCNA was nonsignificantly lower among lung cancer patients assigned to β-carotene than those assigned to β-carotene placebo. There was a borderline significant difference in the lung tumor positivity for CYP1A1 with an odds ratio of 0.2 (95% confidence interval, 0.2–1.1; P = 0.06) comparing men received β-carotene with those received β-carotene placebo.
The 50 mg of β-carotene supplementation on alternate days had no significant influence on molecular markers of lung carcinogenesis that we evaluated in the PHS. This finding provides mechanistic support for the main PHS trial results of β-carotene, which showed no benefit or harm on lung cancer risk.
β-Carotene can block certain carcinogenic processes and inhibit tumor cell growth in experimental studies.1 Observational studies have shown that higher intakes or blood levels of β-carotene are associated with reduced risk for lung cancer.2–5 However, in three large randomized trials, β-carotene supplementation showed neither benefit nor harm on the incidence of lung cancer among apparently healthy male physicians (Physicians’ Health Study [PHS], 50 mg of β-carotene on alternate days for 12 years),6 but increased risk for lung cancer among smokers (Alpha-Tocopherol, Beta-Carotene [ATBC] Cancer Prevention Study, 20 mg of β-carotene daily for 6.1 years)7 or among smokers and asbestos workers (Beta-carotene and Retinol Efficacy Trial [CARET], a combination of 30 mg of β-carotene and 25,000 IU of retinyl palmitate daily for 4 years).8
The p53 tumor suppressor gene plays a pivotal role in the balance of cell proliferation and apoptosis, in the cellular response to various cellular stresses, and in suppressing lung carcinogenesis.9 The structural changes of the p53 protein induced by p53 gene mutations, which are frequently seen in lung cancer,9 enable the mutant protein to become more stable, resulting in the p53 accumulation.10 Loss of p53 function is an early event of lung carcinogenesis.9 Retinoic acid derived from either vitamin A or β-carotene acts on normal bronchial epithelium by inducing mucous and blocking squamous differentiation.11 Because squamous metaplasia occurs during the early stages of lung carcinogenesis, perturbations in retinoid signaling may contribute to lung carcinogenesis.12, 13 Retinoic acid receptor beta (RARβ) is a retinoic acid-responsive gene. Cytochrome P450 1A1 enzyme (CYP1A1), a phase I metabolizing enzyme that is involved in the bioactivation of carcinogenic tobacco products such as polycyclic aromatic hydrocarbons, has been strongly implicated in lung cancer.14
Findings from our studies in ferrets suggest that high-dose β-carotene and cigarette smoke exposure enhance retinoic acid catabolism through the induction of CYP1A1 and CYP1A2 in the lungs.15 In addition, high-dose β-carotene and smoke exposure increased levels of total p53, cyclin D1 and PCNA and squamous metaplasia, but decreased RARβ in the lung tissue of ferrets, whereas low-dose β-carotene attenuated the smoke-induced p53 and slightly decreased squamous metaplasia, but did not significantly affect cyclin D1, PCNA, RARβ.16, 17 Because little is known on how long-term β-carotene supplementation affects molecular markers in human lungs, we measured and compared the protein levels of total p53, RARβ, cyclin D1, PCNA, and CYP1A1 in archival lung tissues from patients diagnosed with lung cancer who received β-carotene or β-carotene placebo in the PHS.
The first PHS was a randomized trial of a 2 × 2 factorial design of aspirin (350 mg) and β-carotene (50 mg) on alternate days in the primary prevention of cancer and cardiovascular disease among 22,071 U.S. male physicians aged 40–84 years at randomization.6 The trial ended on December 31, 1995. More than 7,000 of these physicians, along with 7,000 new physicians, are now taking part in the second PHS that is testing β-carotene, vitamin C, vitamin E, and a multivitamin in the primary prevention of chronic diseases.
Participants were sent yearly questionnaires to ascertain compliance to study treatment assignments and endpoints of interests. Whenever a report of cancer was made, the medical records, including the pathology report were sought. We have succeeded in obtaining records for more than 95% of the reported cases. Physicians blinded to the treatment assignment and other exposures reviewed medical records. We also extracted information on histology, presence or absence of metastases, location, size, and grade of tumors at diagnosis, which were classified according to the Manual of Tumor Nomenclature & Coding by the American Cancer Society (1968 Edition).
Deaths were ascertained through repeated questionnaire mailings, followed by telephone calls, and supplemented by searches of the National Death Index. We obtained death certificates and medical records to assign the cause of death. Follow-up for nonfatal outcomes was over 97% complete, and for mortality, 100%.
Archival formalin-fixed, paraffin-embedded lung tissue blocks were sought from the 194 surgery and biopsy cases of lung cancer diagnosed. Blocks from a total of 40 cases with available adequate specimens were processed for routine H and E slides and unstained slides. One patient was excluded from the analysis due to a prior diagnosis of prostate cancer before β-carotene randomization. The 39 cases in this analysis were similar to the total lung cancer cases in the PHS in term of patient and tumor characteristics. Written informed consent was obtained and the Human Subjects Research Committee at the Brigham and Women’s Hospital approved the study.
Unstained slides (4 μm thick) from the lung tissue blocks were immunostained for p53, RARβ, cyclin D1, PCNA, and CYP1A1 using the standard avidin–biotin complex immunoperoxidase method (Vectastain ABC-Elite; Vector, Burlingame, CA) as described previously.18 Primary antibodies had dilutions of 1:100 for p53 and CYP1A1, 1:50 for RARβ and cyclin D1, and 1:200 for PCNA. Antibodies against p53, RARβ, cyclin D1, and PCNA were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibody against CYP1A1 was purchased from ABR-Affinity BioReagents Inc (Golden, CO).
The sections were examined under light microscopy by two independent investigators who were blinded to treatment groups. Representative areas of each section were selected and 2000 tumor cells were counted under higher magnification [400×] for a total of 10 fields in these areas, immunoreactivity was quantified based on the percentage of positive tumor cells among a total of 2,000 cells.
When the tumor cell nuclei were stained with dark brown color, the cells were considered to be positive for p53, cyclin D1, and PCNA. For RARβ, when the tumor cell perinuclei were stained with defiantly brown color, the cells were considered to be positive. For CYP1A1, the cells were considered to be positive if tumor cell cytoplasm was stained with brown color above cytoplasmic background. Representative immunohistochemical staining for each molecular marker is shown in Figure 1. Immunoreactivity of molecular markers for each section was further classified as positive or negative using the cutoff points for immunohistochemical assays for lung cancer samples in previous studies. p53, cyclin D1, and PCNA were classified as positive when at least 10%,19 5%,20 and 20% (~median value),20 of the 2000 tumor cells showed nuclear staining, respectively. RARβ was classified as positive when at least 5% showed perinuclei staining,21 and CYP1A1 was classified as positive when at least 10% showed cytoplasm staining.22
We first compared the distribution of baseline risk factors for lung cancer, including age, smoking status, and randomized aspirin treatment assignment by randomized β-carotene treatment assignment to assess their potential for confounding. We also compared the distribution of age at diagnosis and lung tumor characteristics between the β-carotene vs. β-carotene placebo groups. Statistical significance between groups was tested by using Wilcoxon rank-sum test for continuous variables and Mantel-Haenszel test for categorical variables.
The odds ratios (OR) and 95% confidence intervals (CI) for the positivity of each lung molecular marker according to category of randomized β-carotene treatment assignment, age at enrollment or diagnosis, smoking status, and tumor characteristics were calculated using unconditional logistic regression models with adjustments for age at enrollment (in years) and smoking status (never, past, or current). In a separate analysis, we additionally controlled for randomized aspirin treatment assignment in multivariable models. SAS version 9.1 (SAS Institute, Cary, NC) was used for all analyses. All P values were two-sided at the significance level of α=0.05 (P value ≤ 0.05).
The median age at enrollment was 57.5 years old (Table 1). Of the 39 lung cancer patients, 20.5% were never smokers, 33.3% were past smokers, and 46.2% were current smokers. In addition, 10.3% were diagnosed with small cell lung cancers, 82% with non-small cell lung cancers, and 7.7% with others (mesothelioma). Among non-small cell lung cancers, adenocarcinoma (53.8%) was the most frequently diagnosed lung cancer, squamous cell carcinoma (20.5%) was the second, and large cell carcinoma (5.1%) and combined squamous cell carcinoma and adenocarcinoma (2.6%) were relatively rare. The average age at diagnosis was 71.1 years old. When diagnosed, more than half of lung cancers were poorly differentiated or undifferentiated (51.3%), or had distant metastasis (61.5%). Baseline age, smoking status, randomized aspirin treatment assignment, age at diagnosis, and lung tumor characteristics were similarly distributed between the β-carotene and β-carotene placebo groups (Table 1).
The lung tumor positivity for total p53 and RARβ was nonsignificantly lower among lung cancer patients assigned to β-carotene than those assigned to β-carotene placebo (Table 2). The lung tumor positivity for total p53 was also less common among those who were smokers, or younger at baseline or at diagnosis, or with undifferentiated tumors, or with distant metastasis; these differences were not statistically significant except for age at baseline (P = 0.01). On the contrary, the lung tumor positivity for RARβ tended to be more common among smokers, or among those who were younger at baseline or at diagnosis, or with poorly or undifferentiated tumors, or with distant metastasis; none of these differences were statistically significant.
The lung tumor positivity for cyclin D1 or PCNA (cell proliferation indices) was also nonsignificantly lower among lung cancer patients assigned to β-carotene than those assigned to β-carotene placebo (Table 3). The lung tumor positivity for cyclin D1 or PCNA was less common among those who were older at diagnosis, but the difference was statistically significant only for PCNA (P = 0.03). However, the lung tumor positivity for cyclin D1 or PCNA tended to be more common among those with poorly or undifferentiated tumors or with distant metastasis (Table 3). There were no meaningful differences for the lung tumor positivity for cyclin D1 or PCNA by smoking status, age at enrollment, or tumor type.
The lung tumor positivity for CYP1A1 was marginally significantly lower among lung cancer patients assigned to β-carotene than those assigned to β-carotene placebo with an OR of 0.2 (95% CI, 0.1–1.1; P = 0.06) (Table 4). The lung tumor positivity for CYP1A1 was also less common in those diagnosed at older age (P = 0.13), but was more common in those with poorly (P = 0.04) or undifferentiated (P = 0.07) tumors or with distant metastasis (P = 0.15).
As for randomized aspirin treatment, there were no differences in the lung tumor positivity for p53 (OR = 1.0; 95% CI, 0.2–4.0), PCNA (OR = 1.1; 95% CI, 0.3–4.2), and CYP1A1 (OR = 1.1; 95% CI, 0.3–4.4) between those assigned to aspirin vs. aspirin placebo, but the lung tumor positivity for RARβ (OR = 0.7; 95% CI, 0.2–2.7) and cyclin D1 (OR = 0.5; 95% CI, 0.1–1.8) was nonsignificantly lower among those assigned to aspirin. The results for p53, RARβ, cyclin D1, PCNA, and CYP1A1 according to randomized β-carotene treatment did not appreciably change after additionally controlling for randomized aspirin treatment assignment (data not shown).
In the PHS main trial, the parent study of this investigation of lung tissue molecular markers, the 50 mg of β-carotene treatment on alternate days provided no benefit or harm on lung cancer development.6 The relative risks of lung cancer by randomized β-carotene assignment in the PHS, which had 11%, 39%, and 50% of current, past, and nonsmokers at baseline, respectively, were 0.90 (95% CI, 0.58–1.40) for current smokers, 1.00 (95% CI, 0.62–1.61) for former smokers, and 0.78 (95% CI, 0.34–1.79) for nonsmokers.6 In the present study involved lung cancer patients within the PHS, the lung tumor positivity for total p53, RARβ, cyclin D1, PCNA, and CYP1A1 was nonsignificantly lower in patients assigned to β-carotene than those assigned to β-carotene placebo. The results of lung tissue molecular markers by randomized β-carotene supplementation were not affected by randomized aspirin treatment assignment. The nonsignificant results of lung tissue molecular markers are consistent with the main PHS trial result of β-carotene on lung cancer risk.
Using the ferret animal model, we previously provided the first in vivo evidence that high-dose β-carotene (equivalent to the 30 mg/day of β-carotene used in the CARET trial), cigarette smoke exposure, and their combination substantially increased protein levels of total p53, which represents p53 accumulation.17 By contrast, low-dose β-carotene (equivalent to the 6 mg/day of β-carotene attainable from a diet high in fruits and vegetables) had no influence on total p53 in nonsmoke exposed ferrets, but reduced total p53 induced by cigarette smoke exposure in ferrets.17 In addition, high-dose β-carotene and smoke exposure increased levels of cyclin D1 and PCNA and increased squamous metaplasia in the lung tissue of ferrets, whereas low-dose β-carotene had no potentially detrimental effects and even slightly decreased cell proliferation or squamous metaplasia induced by cigarette smoke in ferrets.16 When combined with α-tocopherol and ascorbic acid, both doses of β-carotene reduced cigarette smoke induced squamous metaplasia and restored retinoic acid concentrations in ferrets.23 Combined β-carotene, α-tocopherol and ascorbic acid also prevented cigarette smoke and 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone induced up-regulation of p53 and decreased both preneoplastic and neoplastic lesions in ferrets.23
Because malfunction of p53 may result in increased cell proliferation and ultimately tumor development and progression, elevated levels of p53 accumulation may serve as a marker for monitoring genotoxic effects and tumorigenesis. A nonsignificantly lower lung tumor positivity for total p53, cyclin D1, and PCNA levels in the β-carotene group observed in lung cancer patients in the PHS is consistent with the results of the low-dose β-carotene group from the in vivo ferret animal model studies.16 In the PHS, blood β-carotene levels were four times higher in the β-carotene group than those in the β-carotene placebo group (1.2 vs 0.3 mg/L).6 Similarly, blood β-carotene levels were also approximately four times higher in the low-dose β-carotene group than those in the control group in the ferret study (25 ± 5 vs 7 ± 3 nmol/L).16 In the CARET and ATBC studies, two trials that have shown a harmful effect of β-carotene supplementation on lung cancer, there were approximately 12 and 17 times differences in blood β-carotene levels between the β-carotene group and the β-carotene placebo group, respectively ([2.1 vs 0.18 mg/L] in the CARET study8 and [3.0 vs. 0.18 mg/L]) in the ATBC study.7) Interestingly, the enhancement of smoke-induced lung lesions was observed in the ferrets with blood β-carotene levels that were also 17 times higher than those in the control group.16 Among three large randomized β-carotene trials, the PHS had much higher levels of blood β-carotene in the β-carotene placebo group than those in the CARET and ATBC studies (0.3 vs 0.18 vs 0.18 mg/L, respectively), but much lower levels of blood β-carotene in the β-carotene group (1.2 vs 2.1 vs 3.0 mg/L, respectively).6–8
CYP1A1, a phase I metabolizing enzyme, is preferentially expressed in the lung24 where it is inducible and converts procarcinogens into highly reactive intermediates that bind to DNA, forming adducts.14 High-dose β-carotene with or without smoke exposure has been shown to induce CYP1A1 in the lung,15, 25 which leads to enhanced retinoic acid catabolism, resulting in decreased retinoic acid level and diminished retinoid signaling in the animal models.16, 26 Previous data in rats suggest that β-apo-8′-carotenal, an excentric cleavage product of β-carotene, but not intact β-carotene, stimulates the induction of CYP1A1.27 Our previous studies in ferrets showed that the formation of β-apo-carotenals and other oxidative excentric cleavage products of β-carotene was enhanced by smoke exposure,26 indicating that β-carotene is unstable in the free radical-rich environment of the lungs in smokers. Thus, the induction of CYP1A1 by oxidative cleavage products of β-carotene, high-dose β-carotene, or smoke exposure in the lung may bioactivate carcinogens and abolish retinoic acid, thereby enhancing lung carcinogenesis. In addition, these oxidative excentric cleavage metabolites of β-carotene themselves may be directly involved in carcinogenic process.28 In the present study, the 50 mg of β-carotene supplementation on alternate days in the PHS lowered lung CYP1A1 levels, suggesting this regimen may confer some protection against lung carcinogenesis at molecular levels. These data further suggest that the results of β-carotene trials may be related to the doses of β-carotene that were used and/or instability of the β-carotene molecule in lungs of cigarette smokers, which are rich in free radicals.
RARβ mRNA was undetectable by in situ hybridization in approximately half of non-small-cell lung cancers.29 Restoration of RARβ2 in a RARβ-negative lung cancer cell line also has been reported to inhibit tumorigenicity in nude mice.30 In addition, 9-cis-retinoic acid inhibited lung carcinogenesis in the A/J mouse model, which was accompanied by increased expression of RARβ.31 These data suggest that loss of RARβ is associated with lung carcinogenesis. Treatments with 9-cis-retinoic acid in former smokers upregulated RARβ expression in the bronchial epithelium, but had no significant effect on squamous metaplasia.32 Strong RARβ expression also has been found to be associated with a significantly worse outcome of early-stage non-small cell lung cancers.33 In the present study, the lung tumor positivity for RARβ was nonsignificantly lower among men assigned to β-carotene than those assigned to placebo. In the ferret study, RARβ level in lung tissue did not change in the low-dose β-carotene group,16 but was down regulated in the high-dose β-carotene groups (alone or with smoke exposure).16, 26 Future studies to illustrate the role of RARβ in lung carcinogenesis are warranted.
In summary, our data suggest that the 50 mg of β-carotene supplementation on alternate days had no significant influence on molecular markers of lung carcinogenesis we evaluated in the PHS. This finding provides mechanistic support for the main PHS trial results of β-carotene, which showed no benefit or harm on lung cancer risk.
Funding/Support: Supported by grants CA34944, CA40360, HL26490, HL34595, and CA097193 from the National Institutes of Health and by the investigator-initiated research grant from the BASF – The Chemical Company.
We are indebted to the participants in the Physicians’ Health Study for their dedicated and conscientious collaboration; to Eduardo Pereira for his statistical analytic support; to Melissa Aquino, Julia Fleet, and the entire staff of the Physicians’ Health Study for their assistance in the study; to Ella Litvak and Vadim Budes for the technical assistance with the manuscript; and to Drs. Ute Obermueller-Jevic and Klaus Kraemer from the BASF for their support for the study.