Sequence Variants in Cell Cycle Control Pathway, X-ray Exposure, and Lung Cancer Risk: A Multicenter Case-Control Study in Central Europe Free

Exposures to environmental carcinogens, such as tobacco smoke or ionizing radiation (IR), can result in various types of DNA damage and subsequently lead to cancer (1–3). Animal studies have shown that the incidence of lung cancer increases with X-ray exposures in a dose-response relationship (3). Excessive lung cancer risk was also found in patients receiving radiation therapy and among atomic bomb survivors (3–5). It has been suggested that individuals can be at enhanced risk of lung cancer caused by IR if they have certain genetic syndromes and thus can be more sensitive to radiation damage at the cellular level (3).

IR results in a variety of DNA lesions, including base damages, single- or double-strand breaks, the latter being the most difficult to repair and therefore biologically critical. The centerpiece of cellular machinery against DNA damages is cell cycle control, which can trigger cell arrest to allow for DNA lesions to be repaired before the cell continues the process of growth, mitosis, and division or induce apoptosis when the damage is unrepairable (6, 7). The critical transition from G₁ to S is regulated by several key genes, including cyclin D1 (CCND1), CDKN2A/p16, and TP53: cyclin D1 forms complexes with the cyclin-dependent kinase (CDK) 4 or 6, phosphorylates the retinoblastoma protein (pRB), and subsequently releases the cell cycle into S phase; p16 binds to the CDK4-cyclin D1 complex and inhibits the phosphorylation of pRB, thus blocking the cell cycle at the G₁-S border (8); and p53 is an archetypal checkpoint regulator, which can either negatively regulate pRB by inducing CDKN1A/p21 and CDKN1B/p27 protein and triggers cell arrest at G₁ phase or induce apoptosis in different situations (8, 9).

Although the importance of these genes in regulating cell cycle for repairing DNA damages is well established, their specific roles in the cellular responses to radiation exposures are still to be elucidated and therefore require further investigation. In vitro studies suggested that CCND1 expression level may influence the radiosensitivity in different cell lines (10, 11), and an animal study showed that p16 expression levels can be induced by X-ray exposure (12). On the other hand, the role of p53 in radiation response was extensively reviewed (7, 13), and it was shown to mediate cell arrest and apoptosis after exposures to IR (7, 14, 15). Studies showed that alterations in these cell cycle control genes may lead to unregulated pRB phosphorylation and subsequently uncontrolled cellular proliferation (16). The vast majority of epithelial tumors have a genetic or epigenetic alteration in at least one of these components (17), although the effect of sequence variants is not yet well established.

Different sequence variants have been described in these three cell cycle control genes (CCND1, CDKN2A/p16, and TP53). A G to A polymorphism in exon 4 of CCND1 (G870A, rs603965) was suggested to result in alternative splicing, generating a transcript, which encodes a protein with an altered COOH-terminal domain, termed cyclin D1b, as opposed to the cyclin D1a, the product of the common 870G allele (18–20). It has been suggested that cells with DNA damage may bypass the G₁-S checkpoint more easily in the presence of the A allele compared with the G allele, and the oncogenicity of cyclin D1b was shown in animal models (21). However, very limited epidemiologic studies have been conducted to examine the association between this CCND1 variant and cancer risk and only one on lung cancer (22). In contrast, the TP53 sequence variant, which results in an amino acid change from arginine to proline at codon 72 (dbSNP no. rs1042522), has been studied extensively. The ⁷²Pro allele was suggested to be less efficient in suppressing cell transformation and it induces apoptosis with slower kinetics (23, 24). Several studies have investigated the association between this variant and lung cancer risk; however, the results were inconsistent as reviewed in Matakidou et al. (25). A less well-studied variant of TP53 is the 16-bp duplications in intron 3 (named the A2 allele, in contrast to the A1 allele, which represents the wild-type allele; refs. 26, 27). An in vitro study suggested that the A2 allele was associated with reduced mRNA levels (28), and a U.S. study with relatively large sample size (n = 635) showed an increased risk of lung cancer for subjects who carried the A2 allele (27). It is well documented that CDKN2A/P16 is often inactivated through methylation or homozygous deletion in lung cancer patients; however, the role of CDKN2A/P16 sequence variants in lung cancer etiology is unclear (17, 29, 30). A G to A polymorphism leading to an amino substitution of alanine to threonine at codon 148 was identified (dbSNP no. rs3731249). Although information about the functional significance of this variant is limited, we deemed it as a potentially deleterious variant because this missence substitution occurred in a relatively conserved region with low intolerance index predicted from Sorting Intolerance From Tolerance (SIFT; ref. 31). No previous studies have been conducted on this variant and lung cancer risk.

Apart from specific circumstances of potentially high occupational exposure to IR, such as workers in the nuclear power industries and radiologists, X-ray examinations for diagnostic purposes or occupational surveillance represent a major source of manmade exposure to IR at the population level. We have reported previously a strong dose-response relationship between multiple X-ray examinations and lung cancer risk in a case-control study in Central Europe (32). Given the important roles of the cell cycle control genes in response to various DNA damages, we hypothesized that sequence variants in CCND1, TP53, and CDKN2A/p16 influence the predisposition to lung cancer and that subjects who carry sequence variants may have an enhanced risk of lung cancer from X-ray exposure. To address these hypotheses, large sample sizes are needed; therefore, we report on a large-scale genetic association study in the Central Europe.

Study population. The study was conducted in 15 centers in 6 countries of Central and Eastern Europe, including Czech Republic (Prague, Olomouc, and Brno), Hungary (Borsod, Heves, Szabolcs, Szolnok, and Budapest), Poland (Warsaw and Lodz), Romania (Bucharest), Russia (Moscow), and Slovakia (Banska Bystrica, Bratislava, and Nitra). The study details have been described previously (33). Briefly, each center followed an identical protocol and recruited a consecutive group of newly diagnosed cases of lung cancer and a comparable group of controls between 1998 to 2002. Eligible subjects were resident in the study area for at least 1 year. All lung cancer diagnoses had a histologic or cytologic confirmation. Consent for participation was required from both patient and patient's physician. This study was approved by the institutions at all study centers, and ethical approval was obtained from the IARC (Lyon, France), the coordinating center.

Controls in all centers, except Warsaw, were chosen among subjects admitted as in-patients or out-patients in the same hospital as the cases and frequency matched with the case group by sex, age (±3 years), and center and referral (or residence) area. The eligible diseases for controls were nontobacco-related diseases, including minor surgical conditions, benign disorders, common infections, eye conditions (except cataract or diabetic retinopathy), common orthopedic diseases (except osteoporosis), etc. In Warsaw, population controls were selected by random sampling from general practitioner records. Although controls had to be free from cancer at time of enrollment, previous history of cancer was not an exclusion criteria in either cases or controls.

Both cases and controls underwent an identical face-to-face interview during which they completed the same questionnaires. The following variables were collected with a detailed questionnaire: (a) demographic variables, such as sex, race, date of birth, and educational level, etc; (b) medical history; (c) family history of cancer; (d) tobacco exposure, including cigarette smoking and environmental tobacco exposure; (e) alcohol consumption; (f) food frequency; and (g) occupational history. A detailed history of medical and occupational X-ray examinations was collected, including routine chest X-rays at the workplace, other diagnostic examinations, and radiation treatment for a medical problem. In each center, a group of local experts assessed probability, level and frequency of exposure to 60 occupational agents, including IR, based on the information provided in the occupational questionnaire.

Laboratory techniques. Blood samples were collected at the time of interview, and 85% of cases and 79% of controls have provided blood samples. Genomic DNA was extracted from blood samples by automated Gentra Systems (Minneapolis, MN); DNA concentrations were measured by using PicoGreen dsDNA Quantification kits (Molecular Probes, Leiden, the Netherlands). All polymorphisms except the TP53 16-bp duplications in intron 3 were analyzed by the fluorescence 5′ exonuclease assay (i.e., the Taqman assay; ref. 34). Briefly, aliquots of genomic DNA (10 ng) were placed into separate wells of a 96- or 384-well PCR plate along with a PCR cocktail that included both fluorescently labeled allele specific probes. The fluorescence of the PCR products was then plotted, and genotype was determined according to the signal of the two probes. TP53 intron 3 16-bp duplication was genotyped by PCR, with the A1 and A2 alleles giving a band size on 2% agarose gel of 114 and 130 bp, respectively. The sequences of PCR primers and probes are available online.¹²

Statistical analysis. The frequency distribution of demographic variables and putative risk factors of lung cancer, including country of residence, age, sex, education, and smoking, was examined for cases and controls. Tobacco smoking included smoking of cigarettes, pipes, and/or cigars. Former smokers were defined as smokers who quit smoking at least 2 years before interview or diagnosis. Cumulative tobacco consumption was calculated as the product of smoking duration and intensity and expressed as pack-years.

The X-ray exposure was dichotomized to never versus ever and low versus high among subjects who ever had examinations, in which the median number of examinations among the control group, 20, was used as a cutoff point. Analysis related X-ray examinations was restricted to subjects who reported no previous radiation treatment and mammography to avoid residual confounding.

Hardy-Weinberg equilibrium (HWE) of allele distributions was tested in cases and in controls separately. We used unconditional multivariate logistic regression to assess the main effects of genetic polymorphisms on lung cancer risk by estimating odds ratios (OR) and associated 95% confidence intervals (95% CI). Genotypes were categorized into three groups when the allele frequencies allowed (major allele homozygote, heterozygote, and homozygous variant). In the stratified analyses and interaction analyses, genotypes were dichotomized into two categories based on the previous knowledge or frequency distribution of the single nucleotide polymorphisms as risk versus nonrisk genotypes. The designated risk genotypes are as follows: CCND1 A/A, CDKN2A/P16 Ala/Thr or Thr/Thr, TP53 Arg/Pro or Pro/Pro (for Arg⁷²Pro polymorphism), and TP53 A2/A2 (for 16-bp duplications in intron 3). We used expectation-maximization algorithms to estimate the haplotype frequency for the two polymorphisms of TP53 (35, 36).

We conducted stratified analyses by histology to investigate the effect of genetic polymorphism on each histologic subtype. We estimated the effect of sequence variants for subjects with no family history of any cancer, with family history of lung or upper aerodigestive tract cancers (UADT), or with family history of lung cancer, separately. The rationale of analyzing cases with family history separately is that one would expect genetic factor to have a stronger effect on cancer development among subjects with family history. Simulation studies also have showed that limiting analysis to familial cases can increase the study power (37). We also evaluated the modulating effects of cumulative tobacco exposure by stratifying and comparing the strata-specific risk estimates.

Given our recent finding of a strong dose-response relationship with diagnostic/surveillance X-ray exposures, we evaluated whether the effect of X-rays on lung cancer risk is modulated by sequence variants in these three cell cycle control genes. To evaluate the joint effect of two risk factors, X-ray exposures and designated risk genotypes, we used subjects who had nonrisk genotypes and never had X-ray examinations as the reference group and compared the effects of having either and both risk factors. As there are only few subjects who never had X-ray examinations, we also conducted the interaction analysis among subjects who had X-ray examinations and dichotomized the X-ray exposures using the median of control group (20 times) as the cutoff point of exposures.

Matching variables and potential confounders, such as country of residence, age (continuous), sex, and smoking pack-year (continuous), were included in all the multivariate logistic regression analyses. For all the interaction analyses, departure from multiplicativity was tested by estimating the interaction OR (an effect modification parameter), which was derived from the product term in a saturated model along with the covariates in the logistic model. Haplotype analysis was conducted with SAS/Genetic. All other analyses were conducted with STATA software version 8. We also estimated the false-positive reporting probability (FPRP) for all noteworthy findings based on the algorithm proposed by Wacholder et al. (38).

Comparing cases and controls on demographic variables, controls were slightly more educated than cases. As expected, smoking prevalence was higher among cases. Percentages of cases and controls from each country ranged from 7% to 33%. As we reported previously, cases seemed to have had a higher number of X-ray examinations than controls (Table 1; ref. 32).

The allele distributions of CCND1 G870A, CDKN2A Ala¹⁴⁸Thr, TP53 Arg⁷²Pro, and 16-bp duplications in intron 3 among controls were under HWE with P value of 0.77, 0.95, 0.73, and 0.50, respectively. It is worthwhile to mention that the allele distribution of TP53 intron 3 16-bp duplication among cases departed from HWE with excessive A2/A2 genotype (P = 0.004). The quality control of genotyping showed 98.0%, 99.6%, 99.6%, and 98.0% consistency for CCND1 G870A, CDKN2A Ala ¹⁴⁸Thr, TP53 Arg⁷²Pro, and 16-bp duplications in intron 3.

Table 2 shows the main effect of the four sequence variants overall and for cases with family history of cancer. The overall OR of TP53 intron 3 A2/A2 genotype was 1.99 (95% CI, 1.27-3.13). Restricted to cases with family history of aerodigestive cancer (lung and UADT cancers) and lung cancer alone, the OR was 2.52 (95% CI, 1.14-5.56) and 2.98 (95% CI, 1.29-6.87), respectively. The FPRP for the main effect of TP53 intron 3 was 0.049 under a prior probability of 10% and was 0.151 and 0.189 for the effect TP53 intron 3 among cases with family history of lung and UADT or lung, respectively, under the prior probability of 25%. We did not find a main effect for the variants in other cycle control genes that were studied. Stratified by histology, TP53 Pro/Pro genotype seemed to have an effect on squamous cell carcinoma with an OR of 1.44 (95% CI, 1.05-1.97) but not on other histologic subtypes. Similarly, the effect of TP53 A2/A2 genotype was mainly on squamous cell carcinoma with OR of 2.31 (95% CI, 1.34-4.00), which might simply be due to a larger sample size in squamous cell carcinoma. We did not observe any effect modification by smoking status (as never, former, and current smokers) or cumulative tobacco exposures (pack-years) on any sequence variant.

The two polymorphisms in TP53 were in linkage disequilibrium (D′ = 0.812, r = 0.529); therefore, we also conducted haplotype analyses to estimate the effect of individual haplotypes. Compared with the haplotype with the common alleles at both loci, the haplotype with both variants conferred an OR of 1.18 (95% CI, 1.02-1.37) adjusting for age, sex, country, and tobacco pack-years, whereas the haplotypes with only one variant did not seem to have an effect on lung cancer risk.

Table 3 shows the results of the interaction analysis with X-ray exposures. The joint ORs for subjects carrying the CCND1 A/A genotype who had X-ray examination or high X-ray doses were elevated compared with other strata, but no significant interactions were detected. On the other hand, there was a significant interaction between the presence of TP53 intron 3 16-bp duplication and exposures to diagnostic X-rays. The subjects who carried the TP53 A2/A2 genotype and had X-ray examinations conferred an OR of lung cancer of 2.96 (95% CI, 1.63-5.35) compared with subjects who had A1 allele and never had X-ray examinations; subjects who had high X-ray exposures (≥20 times of X-ray examinations) and carried TP53 A2/A2 genotype conferred an OR of 9.47 (95% CI, 2.59-34.6) compared with subjects carrying TP53 A1 allele and had low X-ray exposures. The interaction OR was 5.69 (95% CI, 1.33-24.3), indicating an interaction significantly greater than multiplicativity.

Figure 1 shows the X-ray dose-response relationship on lung cancer risk by genotype. We observed differential dose-response relationships when subjects carried different CCND1 genotypes. For individuals carrying the G/G genotype, X-ray exposures did not increase the risk of lung cancer, whereas for individuals carrying the 870A allele, the dose-response relationship was significant for both heterozygotes and homozygotes. The trend OR for X-ray for individuals who carried the G/G genotype, the G/A, and the A/A genotype was 1.11 (95% CI, 0.98-1.26), 1.16 (95% CI, 1.05-1.27), and 1.29 (95% CI, 1.12-1.49), respectively. The lung cancer risk of subjects with A/A genotype who had >30 X-ray examinations increased 2.8 times compared with subjects who never had any X-ray examination, whereas subjects who had same level of X-ray exposures but carried the G/G genotype did not have a significantly increased risk of lung cancer. The number of subjects who carried TP53 intron 3 16-bp duplications was not sufficient to carry out the dose-response analysis. The trend OR of the X-ray dose-response relationship among TP53 A1/A1, A1/A2, and A2/A2 carriers was 1.18 (95% CI, 1.09-1.28), 1.10 (95% CI, 0.95-1.26), and 2.12 (95% CI, 1.12-4.02), respectively.

We have conducted a large-scale association study to detect modest effects of low-penetrance genes, and as part of this endeavor, we have investigated a novel hypothesis on the effect modification of X-ray exposures by cell cycle control genotypes. In summary, we found an overall effect of the TP53 intron 3 16-bp duplications and a departure from HWE among lung cancer cases. Our results also suggested that carrying sequence variants in cell cycle control genes may modify the effect of X-ray on lung cancer risks. This was observed in two distinct cell cycle control genes, TP53 and CCND1. We found an interaction greater than multiplicativity between the TP53 intron 3 16-bp duplications and X-ray exposures. We did not find a main effect of CCND1 G870A variant; however, the dose-response relationship of X-ray examinations on lung cancer risk was modified by CCND1 genotype in a codominant manner. Our results did not support a role of the CDKN2A Ala¹⁴⁸Thr variant in lung carcinogenesis.

We found that carrying 16-bp duplications in TP53 intron 3 increased the risk of lung cancer by ∼2-fold. The departure from HWE among lung cancer patients, but not among controls, increases the likelihood of this being a genuine association. It was suggested that TP53 introns may contain important sequences for the regulation of mRNA and protein interactions (39, 40). The 16-bp duplications in intron 3 were shown to be associated with a reduced level of mRNA level (41). In combination with other TP53 polymorphisms, it was suggested to result in a lower apoptotic index and reduced DNA repair capacity (27). This is compatible with our finding that the haplotype of intron 3 16-bp duplications–⁷²Pro allele increases the risk of lung cancer.

We observed an interaction between the TP53 intron 3 16-bp duplications and X-ray exposures. Previous studies showed that the p53 modulates the cell response after radiation exposures either through p53-dependent apoptosis or p53-dependent cell arrest (7, 13). p53 has also been suggested to have a role in double-strand break process by binding nonspecifically to the DNA lesions (7). In addition, Li-Fraumeni syndrome (LFS), which is associated with inheritance of TP53 mutation, was shown to have higher risk of second primary cancers induced by irradiation (3). There have been no studies thus far about the effect of TP53 intron 3 variant on radiation response, although several studies considered p53 mutations and/or Arg⁷²Pro polymorphism (42, 43). Boyle et al. (44) found that the fibroblasts from LFS individuals with TP53 mutation showed reduced G₁ arrest after IR exposures compared with the fibroblasts without mutation. A study of workers in two Czech nuclear power plants (who had short-term doses of 0.01 and 0.12 mSV, respectively, and long-term doses of 0.46 and 5.68 mSV, respectively) showed that p53 protein levels in lymphocytes were positively correlated with the radiation exposures, and the nuclear workers who carried Arg⁷²Pro polymorphism variants had lower p53 protein levels compared with those with Arg/Arg genotype. It is difficult to evaluate whether this reduction is directly associated with ⁷²Pro allele because intron 3 16-bp duplications were not characterized (43). Our finding is compatible with the hypothesis that p53 plays an important role in the cellular response to IR, and intron 3 16-bp duplications may enhance the oncogenic effect of DNA damage by radiation either through reduced apoptosis, disturbed cell arrests, or incomplete DNA repair. Due to the small numbers of subjects with both exposures, chance finding or inflation of the OR by random error cannot be excluded.

CCND1 is frequently amplified and overexpressed in a variety of tumors, including lung cancer. Our results do not support the hypothesis of a major role of CCND1 G870A polymorphism in genetic susceptibility to lung cancer overall. However, it might be important among individuals who are exposed to high level of IR. Experimental studies showed that cyclin D1b (encoded by the 870A allele) has enhanced transforming activity relative to cyclin D1a (19). Because cyclin D1b does not possess the Thr²⁸⁶ phosphorylation sites, which are required for nuclear export and regulated degradation, the major difference between cyclin D1a and D1b is their localization: unlike cyclin D1a, which is exported to cytoplasma during S phase, cyclin D1b remains in the nucleus during the cell cycle where its constitutive expression leads to oncogenic potential (21). We found that subjects carrying the CCND1 870A allele might have an enhanced risk of lung cancer from X-ray exposures. The detailed mechanism of how the CCND1 G870A polymorphism modifies the risk from X-ray exposures is still to be elucidated. However, several studies have shown that CCND1 expression level is associated with apoptotic events induced by IR (10, 11) and the clinical response to radiotherapy (45). It is plausible that the 870A allele, which produces cyclin D1b that remains in the nucleus, might modify cellular response to the X-ray exposures and thus enhances the risk of lung cancer. Furthermore, studies have shown that, although nuclear expression of cyclin D1 does not result in cell transformation, it sensitizes cells to stochastic second hits that results in neoplasia (21). Our finding of an enhanced risk from X-ray exposures with CCND1 A/A genotype seems to be compatible with this model.

There are several limitations in this study. First, the X-ray exposures were self-reported, and apparently, the strong dose-relationship between X-ray and lung cancer risk was due to recall bias as we have pointed out in the previous report. However, the differential dose-relationship by genotype is unlikely to be entirely explained by recall bias because the study subjects were not aware of their genotype. If the dose-response relationship had been entirely due to recall bias, one would expect a similar association regardless of the genotypes. Another argument in favor of the effect modification by CCND1 being real was the codominant manner that the dose-response relationship was modified by genotype, which is unlikely to be entirely due to chance. On the other hand, because a high number of X-ray examinations can be an indicator of being used in a hazardous occupation, which requires frequent X-ray examinations, residual confounding cannot be excluded, and this may at least partially account for the higher estimates than expected. Nevertheless, further adjusting for occupational exposure did not affect the results materially. Although we have no evidence to assume that the observed effects are due to other occupational exposures that are associated with X-rays examinations, the possibilities of other unmeasured occupational carcinogens interacting with genotype, rather than the X-ray per se, cannot be excluded.

Another concern is the dosage of diagnostic X-rays, which varies between countries and with different time periods. Overall, the estimated reported doses per examination in the participating countries were 0.2 to 2.8 mSV (32). The cumulative doses of subjects who had the highest number of examinations (range, 20-30 mSV) would be comparable with the average doses among workers in the nuclear industries. Previous studies showed that p53 is responsive after exposure to clinically relevant dosage and it can induce p53-dependent apoptosis (14, 15); however, such information on CCND1 is not yet available.

In conclusion, our results suggested that sequence variants in cell cycle control pathway may increase the risk of lung cancer and modify the risk conferred by multiple X-ray exposures. The mechanism of how the variants modify the effect of X-ray need to be further elucidated through experimental studies. Given that X-rays are one of the most important exposures to IR, our results may have implication on future usage of diagnostic X-rays. However, a definite conclusion can only be reached on replications by different studies among individuals who are highly exposed to IR.

Grant support: National Cancer Institute R01 grant (contract no. CA 092039-01A2), and Association for International Cancer Research grant (contract 03-281).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Pierre Hainaut for his critical review and comments on the article.