Implications of LINE1 Methylation for Bladder Cancer Risk in Women

Bladder cancer is the ninth most incident form of cancer in the United States, with an estimated 70,980 new cases expected to occur in 2009 (1). High rates of bladder cancer are especially widespread in New England, with women in this area experiencing some of the highest bladder cancer mortality rates in the United States (2). Incidence rates of bladder cancer have been stable in men in the last 20 years but have been increasing in women at a rate of 0.2% a year; despite this, men are nearly four times more at risk of developing bladder cancer than women (1, 3). Furthermore, bladder cancer is thought to be a heterogeneous disease with different characteristics related to the histopathologic presentation of the disease primarily as noninvasive versus invasive, with the latter contributing to the vast majority of disease mortality (4). It is also believed that different molecular mechanisms may contribute to the heterogeneity of this disease, calling for more research to clarify the differences between pathologic pathways (4).

Genetics, dietary factors, occupational exposures, toxicants, and life-style factors play an important role in bladder cancer initiation and development, and the major known risk factors for bladder cancer include smoking, aromatic hydrocarbons, water chlorination byproducts, and inorganic arsenic (3, 5, 6). However, the effects of some of these risk factors on bladder cancer, particularly arsenic, has only been noted in certain populations (7). Ingestion of high levels of arsenic has been associated with bladder cancer mortality in certain regions of the world, most notably in Argentina, Chile, and Taiwan (7). Despite this, exposure to lower levels of arsenic has been found largely in susceptible subgroups (5). A recent New Hampshire study found the association between levels of arsenic higher than 0.330 μg/g and an increased risk of bladder cancer to hold true only in smokers, confirming results reported in two other studies, one conducted in Utah and the other in Finland (5, 8, 9).

Epigenetics is an emerging area of epidemiologic interest, seemingly important for understanding both disease susceptibility and etiology (10, 11). Much work has characterized promoter region hypermethylation of cancer-related genes in bladder tumors, reporting associations between these alterations and exposures causal for the disease (12–14). Past research has also shown tumor cells to be hypomethylated, on a global level, compared with normal cells (15, 16). Examination of the DNA methylation pattern in repetitive elements, such as long interspersed nuclear elements (LINE1), has been used as a surrogate of overall global methylation levels (17, 18). At the same time, epigenetic alterations of LINE1 may also be of interest in that they may initiate the clonal expansion of premalignant cells in early stages of cancer development, may contribute to genomic instability, and may lead to structural changes in chromosomes (10, 11, 19, 20). DNA hypomethylation may also lead to gene overexpression, occurring when genes become hypomethylated, and to the activation transposable elements, which may result in genetic mutations (20–22). In addition, DNA methylation may be altered by dietary availability of methyl groups, specifically with two of methylenetetrahydrofolate reductase single nucleotide polymorphisms of the enzyme, MTHFR-677T/T and 1298C/C, having been associated in previous work with lower global methylation in lymphocytic DNA (23).

There is emerging evidence that changes to the global or repetitive element methylation status in DNA derived from peripheral blood cells can serve as a biomarker of exposure and disease. Recent reports suggest that exposure to carcinogens and pollutants, specifically including particulate matter and benzene, are associated with global hypomethylation detectable in peripheral blood (24–26). Furthermore, global DNA hypomethylation in blood, measured using a radioactive methyl acceptor assay, has been associated with bladder cancer (27) and gastric cancer (28), whereas altered methylation of the LINE1 region has been linked to head and neck cancer risk (18).

In this study, we sought to determine what factors are associated with LINE1 hypomethylation, whether LINE1 hypomethylation is associated with increased risk of bladder cancer, and if any risk factors for the disease modulate or modify this association.

A description of the study design appears in earlier reports (5, 29). Briefly, cases were selected from the New Hampshire State Department of Health and Human Services' Cancer Registry and included only New Hampshire residents ages 25 to 74 y with a first diagnosis of bladder cancer ranging from July 1, 1994 to June 30, 1998. Of the 618 potential cases eligible for this study, 10 (2%) were not able to be contacted due to denial of contact by their physicians, 63 (10%) were reported as deceased by a household member or physician, 3 (<1%) did not return telephone calls about participation in the study, 75 (12%) refused to participate, and 8 (1%) were too ill to take part. Finally, 459 or 85% of the eligible cases were interviewed for this study. A standardized review was conducted by a single study pathologist (A.R.S.) to verify the diagnosis and histopathology of the cases. For the analyses presented here, the case group was restricted to cases having LINE1 methylation data and excluded cases that were diagnosed as carcinoma in situ; this included a total of 285 cases, whose characteristics are presented in Table 1. For efficiency purposes, the same control group used in a study of nonmelanoma skin cancer conducted from July 1, 1993 to June 30, 1995 was used and additional controls were selected afterwards from the sources described below (30). All controls of <65 y of age were selected from records obtained from the New Hampshire Department of Transportation and controls >65 y of age were chosen from records obtained from the Health Care Financing Administration's Medicare Program. Controls were then frequency matched to cases by age (25-34, 35-44, 45-54, 55-64, 65-69, 70-74 y) and gender and were randomly assigned a reference date comparable with the cases' diagnosis date. Of the 990 potential controls eligible for this study, 18 (2%) were reported as deceased by a member of the household, 17 (2%) did not return telephone calls about participation in the study (2%), 261 (26%) refused to participate, and 29 (3%) were mentally incompetent or too ill to take part. Finally, 665 or 70% of the eligible controls were interviewed for this study. For the analyses presented here, the control group was restricted to the 465 controls having LINE1 methylation data, whose characteristics are presented in Table 1. No significant differences in the demographic or risk factor variables examined were found between subjects with and without methylation data, and to ensure comparability of cases and controls, all subjects were asked if they currently held a driver's license or a Medicare enrollment card.

Consenting subjects underwent a detailed in-person interview, usually at their home, which assessed sociodemographic information, occupational history, detailed information about the use of tobacco products (such as history of cigarette smoking), and medical history before the reference or diagnosis date. Subjects were requested to save a toenail clipping specimen before the interview or were sent a self-addressed envelope to send in their sample after the interview. Toenail samples, which reflect arsenic burden in the body from all sources of exposure, were analyzed for arsenic using instrumental neutron activation analysis at the University of Missouri's Research Reactor Center (30). All procedures and study materials were approved by the appropriate Institutional Review Boards.

DNA was extracted from peripheral blood buffy coats using the QIAmp DNA mini kit according to the manufacturer's protocol (Qiagen). DNA (1 μg) was subjected to sodium bisulfite modification using the EZ DNA Methylation kit (Zymo Research) following the manufacturer's protocol. LINE1 region methylation extent was determined through quantitative bisulfite pyrosequencing (31), following the protocol of Bollati et al. (26), which examines the cytosine methylation status at 4 CpG sites in the LINE1 region. All PCRs, performed using the Qiagen Hot Star Taq polymerase, included a no template control, unmodified DNA control, and seven standardized percent methylation controls (0%, 15%, 25%, 45%, 65%, 75%, and 100%) derived from the Qiagen EpiTect PCR control DNA set samples. Each sequencing reaction used 10 μL of PCR product and was run according to instrument/manufacturer protocols (Qiagen) on a PyroMarkMD System (Qiagen). Three PCR amplifications were done on each sample and two pyrosequencing runs were done from each PCR reaction, resulting in six replicates for each individual to assess repeat measure variability. The SEM of the averaged individual repeats was found to be the same as the SEM for each replicate, so the average measure of LINE1 methylation across the four CpG sites for the six replicates was used for each individual. The following polymorphisms were assessed in this study: methylenetetrahydrofolate reductase (MTHFR)-677 C>T and MTHFR-1298 C>T, and both were genotyped using the 5′ nuclease Taqman allelic discrimination assay on the ABI 7900HT (Applied Biosystems) as previously described (32).

To assess any dose dependence of LINE1 methylation extent on bladder cancer risk, LINE1 methylation was coded into deciles, based on the distribution in controls, to determine more precisely the amount of methylation that would increase the risk of bladder cancer. As an increase in the odds ratio (OR) for disease was present only in the lowest decile, LINE1 methylation was coded as a binary outcome comparing the lowest decile with the reference including the remaining deciles of LINE1. Age was coded as a continuous variable with 750, mean 62.24 (SD, 10.04; minimum, 25; maximum, 74). Arsenic was classified using percentiles of the control distribution, based again on our previous research: 0.009 to 0.059 (≤25%, used as the reference group), 0.060 to 0.086 (25.1-50%), 0.087 to 0.126 (50.1-75%), 0.127 to 0.193 (75.1-90%), 0.194 to 0.277 (90.1-95%), 0.278 to 0.330 (95.1-97%), and 0.331 to 2.484 (97.1-100%; ref. 4). Smoking was evaluated using variables measuring intensity of smoking, duration of smoking and their combination, as well as the status (never, former, and current) among subjects. Smoking was used as a categorical variable defined as never, former, or current smokers as has been done previously in examinations in this and other populations (5, 27, 33). To explore the influence of MTHFR 677 and MTHFR 1298 genotypes on LINE1 methylation and bladder cancer risk, these variables were coded using wt/wt as the reference and were compared with wt/var and var/var (18). In univariate analyses, the relationship between the amount of methylation and case-control status and other factors was evaluated with χ² statistics and t tests using a P value of <0.05 as statistically significant. To investigate the association between the level of LINE1 methylation and bladder cancer, unconditional and stratified logistic models were used and included smoking, age, and gender (the latter were matching variables in the original study). To examine the association between LINE1 methylation and risk of either noninvasive or invasive bladder cancer, we performed the analysis stratified by tumor stage. Similarly, to examine the differential association of LINE1 methylation with risk of bladder cancer between men and women, we performed an analysis stratified by gender. These models were adjusted for gender and smoking status. To examine the relationship between variables including gender, age, arsenic, and smoking with the amount of LINE1 methylation, logistic models were used with low LINE1 methylation defined as being in the first decile as a binary outcome. Selenium, education, MTHFR 677, and MTHFR 1298 were examined as possible predictors of bladder cancer risk and as possible important predictors of LINE1; however, these variables were not found to significantly influence methylation or bladder cancer risk, and did not change the estimates of the effects of the other variable in the model, and thus were dropped from further models. The interaction between smoking and LINE1 methylation was examined with the likelihood ratio test. Data were analyzed by use of the SAS statistical software, version 9.1.

LINE1 methylation levels were determined for 465 controls and 285 cases, and overall ranged from 57.9% and 92.0%. The characteristics of controls and cases with LINE1 methylation data are shown in Table 1. MTHFR codon 677 and 1298 genotypes, mean toenail arsenic levels, and age were not significantly different in cases and controls, whereas smoking status and gender were significantly different in the two groups. Figure 1 shows the relationship of LINE1 methylation extent with bladder cancer risk, grouping LINE1 methylation data into deciles defined by levels in the controls. These data show that a significant estimate of disease risk exists only in the lowest decile of LINE1 methylation. The risk estimate, for being in the lowest decile was an OR of 2.08 [95% confidence interval (95% CI), 1.07-4.07] compared with the reference highest LINE1 methylation decile, and adjusted for age, gender, and smoking status. Based on this figure, we further defined LINE1 as a binary variable with the lowest decile grouping the remaining deciles as the reference.

Table 2 presents the association between LINE1 methylation extent in the lowest decile compared with all others as reference, adjusting for gender, age, and smoking status using logistic regression. A 1.80-fold increased risk of bladder cancer (95% CI, 1.12-2.90) was observed among those with LINE1 methylation extent in the lowest decile. As previous reports have suggested an interaction between global DNA methylation and smoking (27), we examined the interaction of LINE1 methylation with smoking status and found no significant association (data not shown). Because bladder cancer presents in a dichotomous fashion as noninvasive or invasive disease and these forms of bladder cancer have significant pathologic differences, we performed stratified analysis by invasive status (4). Among noninvasive cancers, being in the lowest LINE1 methylation decile confers an ~2-fold risk of bladder cancer (OR, 1.94; 95% CI, 1.17-3.22), whereas the risk is nonsignificant among invasive cancers, in models adjusted for smoking, age, and gender. We then stratified our model by gender, adjusting for age and smoking status in each stratum. Among females, being in the lowest LINE1 methylation decile confers a significant ∼2.5-fold increased risk of bladder cancer (95% CI, 1.19-5.17), whereas the risk was nonsignificant among males (OR, 1.47; 95% CI, 0.79-2.74).

Given the risk associated with low methylation, we sought to identify the predictors of hypomethylation in healthy controls. We used the binary LINE1 methylation variable as the dependent variable and examined the association between this variable and known risk factors of bladder cancer (including gender, age, smoking status, and toenail arsenic levels) stratified by case status in Table 3. Women were more likely to be in the lowest LINE1 methylation decile than men (male controls: OR, 0.49; 95% CI, 0.25-0.97; males cases: OR, 0.38; 95% CI, 0.18-0.81). The lowest decile of LINE1 methylation had a higher percentage of individuals with high toenail arsenic concentrations. A trend in ORs for low LINE1 methylation from the lowest to the highest toenail arsenic levels was statistically significant (P < 0.01), with the 75th to 90th percentile (0.13 to <0.20 μg/g arsenic) having an OR for low LINE1 methylation of 3.68 (95% CI, 1.17-11.58), and those with greater than the 90th percentile of arsenic exposure (>0.20 μg/g arsenic) having an OR for LINE1 methylation of 3.68 (95% CI, 1.04-13.02).

In models controlling for smoking status, gender, and age, we have found a significant 1.8-fold increased risk of bladder cancer among subjects with mean LINE1 methylation levels in the lowest decile of methylation, ranging from 57.89% to <74.25%, compared with subjects with >74.25% LINE1 methylation. This suggests that LINE1 hypomethylation is an independent risk factor for bladder cancer and that it may be an excellent biomarker. Our results are consistent with a previous report that also noted a significantly increased risk of bladder cancer among patients with lower global DNA methylation levels, as assessed by the radioactive methyl incorporation assay (27). Interestingly, unlike that report, we did not observe a dose-dependent association of LINE1 methylation extent with bladder cancer risk, but rather saw only an association with disease risk at the lowest level, a result consistent with our previous work examining LINE1 methylation in head and neck cancer (18).

We also observed that the risk of noninvasive bladder cancer was increased ∼2-fold for subjects in the lowest LINE1 methylation decile, whereas there was no significant association in invasive cases. This is of interest because it suggests that LINE1 methylation may be a specific biomarker of noninvasive disease. The mechanism responsible for this apparent distinction is unclear but it is consistent with the fact that there are well-known differences in this disease associated with an invasive pathology (4).

In examining the predictors of lowest tenth decile of LINE1 methylation, we found that women were >2-fold significantly more likely to have reduced LINE1 methylation extent. This finding is consistent with a study that found that females had statistically lower methylation levels than males in LINE1 and Alu repeats measured from whole blood (34) as well as with our previous work in head and neck cancer (18). Two other studies did not find a significant association between LINE1 methylation and gender in leukocytes and in adenomas; however, these studies did not use a large number of subjects and may have been underpowered to observe this association (17, 23). Our gender-stratified analyses further suggest that the risk for bladder cancer conferred by low LINE1 methylation extent is significant only among women, and is consistent with Moore et al. (27) who observed a similar increased risk of bladder cancer among women with reduced overall global DNA methylation levels.

Our finding of reduced LINE1 methylation in relation to arsenic exposure is consistent with experimental data showing that arsenic and its inorganic salt, arsenite, induce DNA hypomethylation in cultures of human lung and rat hepatocytes, as well as in rat livers through feeding studies (35–37). A study in Bangladesh, with endemic, high-level arsenic exposures, observed that reduced DNA methylation was a risk factor for arsenic-associated skin lesions (38) but found a positive relationship between urinary arsenic concentrations and global DNA methylation in peripheral blood leukocytes (using the radioactive methyl incorporation assay; ref. 39). This different finding may be related to the different measures of DNA methylation and the far greater exposures in Bangladesh compared with ours. Although the methylation extent of the LINE1 region has been correlated with overall cellular 5-methycytosine content (40, 41), and thus serve as a surrogate marker, changes in the extent of methylation of these elements may lead to specific functional consequences on their expression. Expression of these elements can lead to their retrotransposition to various regions of the genome, possibly leading to downstream effects on other genes in these cells (42). It is unclear if the changes in methylation extent being measured lead to changes in expression of specific functional LINE1 elements. Going forward, it will be important to delineate the biological nature of this association, as LINE1 measurement may be, in some cases, in the causal pathway for disease.

We did not find an interaction between LINE1 methylation and smoking status, unlike previous reports examining methylation by smoking intensity (27). The nonsignificant interaction in our model could be attributable to differential smoking patterns between the populations.

The strengths of this study included the population-based nature of the study, as well as the use of quantitative pyrosequencing for LINE1 methylation determination and detailed exposure assessment including the use of toenail analysis of arsenic concentrations. The limitations of this study included the exclusion of cases and controls without methylation data, which could lead to a selection bias if included and excluded cases and controls differ from one another. However, we did not note any differences between included and excluded cases and controls in any of our demographic information and use of tobacco products. Another limitation of this study was the low number of invasive cases available; a larger study is necessary to further understand the effect of global hypomethylation on the risk of development of noninvasive compared with invasive bladder cancer. Another limitation would be the poor age matching done in this study; however, we do not think this affected the results as age is not found to be significant in any of our analyses. Further research on hormonal influences on global methylation and risk of bladder cancer is also warranted to relate to our findings of differential risk among women.

In summary, this study adds to the growing literature on the role of epigenetics in human cancer etiology and suggests LINE1 hypomethylation as a risk factor for bladder cancer, with potential utility as a disease biomarker. In addition, this study shows an increased risk of LINE1 hypomethylation with high levels of arsenic, raising important questions about the nature of the causal pathway for alterations in DNA repeat methylation to contribute to cancer risk. These results also suggest that women show lower levels of LINE1 methylation and may suggest distinct etiologies of bladder cancer between men and women, which may also aid in explaining the disparate incidence of this disease in men and women. Further work is needed to better clarify how exposures important to the disease play a role in LINE1 region methylation and to clarify the specific subgroups in which this factor may be most predictive of disease risk.

No potential conflicts of interest were disclosed.

Grant support: Flight Attendant Medical Research Institute grant YCSA 052341 (C.J. Marsit) and the NIH grants R01CA121147, K07CA102327, P50CA097257, and P42ES007373 and R01CA57494.

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.