Skip Navigation


Carcinogenesis Advance Access originally published online on May 23, 2007
Carcinogenesis 2007 28(8):1745-1751; doi:10.1093/carcin/bgm116
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Data
Right arrowOA All Versions of this Article:
28/8/1745    most recent
bgm116v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (1)
Google Scholar
Right arrow Articles by Marsit, C. J.
Right arrow Articles by Kelsey, K. T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Marsit, C. J.
Right arrow Articles by Kelsey, K. T.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

© The Author 2007. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org
The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press are attributed as the original place of publication with the correct citation details given; if an article is subsequently reproduced or disseminated not in its entirety but only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact journals.permissions@oxfordjournals.org

Promoter hypermethylation is associated with current smoking, age, gender and survival in bladder cancer

Carmen J. Marsit, E. Andres Houseman1, Alan R. Schned2, Margaret R. Karagas3 and Karl T. Kelsey4,*

Department of Pathology and Laboratory Medicine, Brown University, Providence, RI 02912, USA
1 Department of Work Environment, University of Massachusetts-Lowell, Lowell, MA 01854, USA
2 Department of Pathology, Dartmouth Medical School, Lebanon, NH, USA
3 Department of Community and Family Medicine, Dartmouth Medical School, Lebanon, NH 03756, USA
4 Department of Genetics and Complex Diseases, Harvard School of Public Health, 665 Huntington Avenue, Boston, MA 02115, USA

* To whom correspondence should be addressed. Tel: 617 432 3313; Fax: 617 432 0107; Email: kelsey{at}hsph.harvard.edu


   Abstract
Top
 Abstract
Introduction
 Subjects and methods
 Results
 Discussion
 Supplementary material
 Funding
 References
 
Hypermethylation of tumor suppressor genes is central to the pathogenesis of human malignancy, yet this alteration's etiology remains unclear, and its clinical impact has not yet been studied using an approach that can yield directly generalizable results. Therefore, we sought to examine both of these issues in bladder cancer, seeking to understand the characteristics of epidemiologically well-defined patient tumors with differing levels of methylation silencing. We analyzed epigenetic silencing of 16 genes in a population-based incident case series of 331 bladder transitional cell carcinomas. We utilized a novel item response theory (IRT) model, to examine, in an unbiased fashion, the relationship of patient characteristics, carcinogen exposure history and tumor characteristics with the underlying propensity for gene hypermethylation. Age, male gender and current cigarette smoking were significantly positively associated with the methylation latent trait. Promoter hypermethylation as a latent trait significantly predicted both non-invasive/high-grade and invasive stage disease and was also significantly associated with survival, with each unit increase in the latent trait resulting in a 30% increase in the risk of death. This work, studying all stages and grades of incident bladder cancer, provides definitive evidence that carcinogen exposures play a critical role in selecting these alterations in tumorous clones and that epigenetic silencing is a strong and significant predictor of tumor stage and overall patient survival. Finally, our novel approach provides insight into the etiology of promoter hypermethylation, suggesting that selected, carcinogen-exposed individuals have a greater propensity for hypermethylation that is associated with more aggressive, fatal disease.

Abbreviations: BcA, bias corrected and accelerated; CI, confidence interval; IHC, Immunohistochemical; IRT, item response theory; PCR, polymerase chain reaction


   Introduction
Top
 Abstract
 Introduction
Subjects and methods
 Results
 Discussion
 Supplementary material
 Funding
 References
 
Bladder cancer is the fourth most incident cancer in males and ninth most incident in females, with over 60 000 new cases expected to be diagnosed in 2006 (1). Tobacco smoking (2) and occupational exposures (particularly including aromatic amines and polycyclic aromatic hydrocarbons) (3) are well known to contribute to the occurrence of bladder cancer. Drinking water arsenic exposure in highly exposed populations, exposure to hair dyes, chlorination by-products, individual fluid intake and diet also may play a role in genesis of this cancer (4). Bladder cancer in the USA arises chiefly from the transitional cells of the bladder epithelium and most commonly presents as a non-invasive, papillary tumor that is readily treated through resection. However, ~20% of incident cancers present as invasive disease (5) which have a 5-year survival rate of 30–50% (6,7). This pattern of disease incidence suggests that non-invasive and muscle-invasive disease arise from separate pathways, and that these pathways may be characterized by a distinct pattern of molecular alterations critical in the genesis of bladder tumors (5,8). Identification of the particular alterations and pathways that characterize these different forms of the disease holds great promise for application as clinical biomarkers of disease prognosis and patient survival. Additionally, molecular characterization of these distinct forms of the disease should offer a better understanding of the etiology of the disease; examining the patterns of molecular alterations associated with risk factors (specifically including exposure to bladder carcinogens) will illuminate their modes of action and role in disease occurrence.

The critical and causal role for epigenetic alterations in cancer is becoming increasingly understood. The most studied epigenetic alteration, DNA hypermethylation of the promoters of specific tumor suppressor genes, has been well established as an indicator of specific gene silencing, playing a crucial role in a variety of human cancers, including bladder cancer (9). Indeed, a phenotype arising in a subset of tumors characterized by an increased propensity for methylation has been described in a variety of tumors; this is most clear in colorectal cancer (the CpG island methylator phenotype) (10,11). Yet, the etiologies of gene-specific hypermethylation and the hypermethylator phenotype have been less well explored, and the clinical relevance of the phenotype (particularly across tumor types) remains unclear. In bladder cancer, we and others have linked DNA hypermethylation and inactivation of specific genes to poorer patient survival (12–14) and specific gene methylation to particular carcinogen exposures such as tobacco smoking and drinking water arsenic (15). There is evidence that an increasing number of genes are subjected to hypermethylation silencing with increasing disease stage in bladder cancer (12,16), providing initial evidence for the existence of an underlying propensity for promoter hypermethylation, similar to that observed in colon cancer, that drives the clonal expansion of tumors and inactivation of critical pathways, potentially inducing the divergent forms of bladder cancer. We have previously reported that in bladder cancer, as well as lung cancer, head and neck squamous cell carcionoma and malignant pleural mesothelioma, that although there may be a propensity towards promoter hypermethylation in some tumors, it is not as clearly evident as in colorectal cancer (17). Thus, appropriate statistical methodologies, as we have described (17,18), must be used to examine this data, particularly in order to examine the relationship between this propensity for methylation and its etiology or significance. This study is aimed at applying our IRT model methodologies, to examine 16 previously characterized (17) tumor suppressor gene promoters for hypermethylation, in a population-based case series of incident bladder cancers from the state of New Hampshire, which also includes detailed patient characteristics and exposure histories obtained from in-person interviews, expert pathologist reviewed clinical data on the tumors, and additional molecular characterization including TP53 status, and follow-up information on patient survival. This approach will allow for a detailed understanding of the etiology and clinical significance of these alterations that are considered central to the pathogenesis of bladder and other solid cancers.


   Subjects and methods
Top
 Abstract
 Introduction
 Subjects and methods
Results
 Discussion
 Supplementary material
 Funding
 References
 
Study population
Residents of New Hampshire aged 25–74 years, diagnosed from 1 July 1994 to 30 June 1998 were identified by the rapid reporting system of the New Hampshire State Cancer Registry and asked to participate in the study (19). All study participants were consented under the appropriate institutional review board protocols. Study participants underwent a personal interview to obtain information on demographic traits and lifestyle factors, such as tobacco use (including frequency, duration and intensity of cigarette smoking). Follow-up data on overall patient survival were obtained through national death index query. Pathology material was complete from a total of 355 patients. Pathology reports and paraffin-embedded tumor specimens were requested from the treating physician/pathology laboratories. Bladder tumors were reviewed by the study pathologist (A.R.S.) and classified according to the 1973 and 2004 World Health Organization guidelines for bladder tumors. Additionally, the proportion of malignant cells in each sample was recorded for each sample, averaging 69% (median 80%). The final statistical analysis was performed on 331 tumors which had complete data on exposures, demographics and all laboratory analyses, including methylation at all loci and TP53 immunohistochemical (IHC) staining.

DNA extraction and sodium bisulfite modification
The study pathologist selected tumor sections with the greatest proportion of malignant tissue. Three 20 µM sections were cut from each formalin-fixed, paraffin-embedded tumor sample and the sections were transferred into microcentrifuge tubes. The paraffin was dissolved using Histochoice Clearing Agent (Sigma-Aldrich, St Louis, MO) followed by two washes with 100% ethanol and one wash with phosphate-buffered saline. The samples were then incubated in sodium dodecyl sulfate lysis solution (50 mM Tris–HCl, pH 8.1, 10 mM ethylenediaminetetraacetic acid, 1% sodium dodecyl sulfate) with proteinase K (Qiagen, Valencia, CA) overnight at 55°C. De-cross linking was performed by adding NaCl (final concentration 0.7 M) and incubating at 65°C for 4 h. DNA was recovered using the Wizard DNA clean-up kit (Promega, Madison, WI) according to the manufacturer's protocols. Sodium bisulfite modification of the DNA was performed using the EZ DNA Methylation Kit (Zymo Research, Orange, CA) following the manufacturer's protocol, with the addition of a 5 min initial incubation at 95°C prior to addition of the denaturation reagent. The de-cross-linking steps in the extraction as well as the 95°C incubation ensure more complete melting of the DNA and thus more complete sodium bisulfite conversion for these highly cross-linked formalin-fixed specimens.

Methylation-specific polymerase chain reaction
We have specifically chosen to utilize traditional methylation-specific polymerase chain reaction (PCR) for the analysis of promoter hypermethylation in these studies, as we have performed matched analysis between fresh-frozen and formalin-fixed, paraffin-embedded samples, and find the greatest concordance (>95%) for methylation detection using this method (data not shown). We have also previously examined potential biases in the sensitivity of using this assay against the relative quantitative Taqman-based methods (12), and have seen no evidence for potential bias based on tumor quantity or tumor stage in the samples analyzed. Finally, this method allows for detection of a large number of genes from the limited DNA samples available on many of these tumors, with the advantage that it is standardized and potentially more widely available to investigators than the quantitative technologies which require real-time PCR systems.

Sodium bisulfite-modified DNA was used as the template for methylation-specific PCR as previously described (20) using primers specific for the methylated promoters of CDKN2A (20), RASSF1A (21), APC (22), PYCARD (23), LAMC2 (24), SFRP1, SFRP2, SFRP4, SFRP5 (25), MGMT, DAPK, RARB, CDH1 (26), CDH13 (27), MLH1(28) and PRSS3 (29). All methylation-specific PCRs are optimized to detect >5% methylated substrate in each sample. To control for the presence of modified DNA, primers specific to a modified region of the ACTB gene containing no CpG sites were used (30). Modified circulating blood lymphocyte DNA (obtained from a control subject) and the same lymphocyte DNA completely methylated using SssI DNA methylase and modified by treatment with sodium bisulfite were used as the negative and positive controls, respectively, in each run.

This group of genes examined for promoter hypermethylation using this method have been previously shown to be correlated to transcriptional silencing of these genes. We also wished to examine the silencing of tumor suppressor genes involved in a variety of cellular processes and pathways, thereby not limiting the analysis to genes involved in a single pathway targeted for inactivation. The genes selected are known to be involved in processes including cell cycle control (CDKN2A, RASSF1A, APC), apoptosis (DAPK, PYCARD), extracellular interactions (LAMC2, PRSS3), transcriptional regulation (RARB), WNT signaling (SFRP family), cell–cell signaling (CDH1, CDH13) and DNA repair (MGMT, MLH1).

TP53 immunohistochemistry
The immunohistochemical detection of TP53 has been previously described (31).

Statistical analysis
Since gene promoter hypermethylation does not occur as an independent event targeting individual genes, we applied a latent variable model. This is based upon the notion that fundamental relationships of interest can be characterized in terms of a small number of variables (conceptually perhaps viewed as biological parameters such as occult DNA polymorphisms or gene mutation/amplification events that could drive a methylator phenotype), some of those being unobserved or latent (32), and allows for the examination of the underlying propensity for promoter hypermethylation (33). We then employed the IRT model (18,34), where binary responses on each of J items (gene promoters methylated) are related to a single continuous latent variable U, which can have a differential relationship on each item. Observed values, including those assessing exposure, demographics and clinical characteristics, can then be related to the latent variable through regression modeling. Using this methodology, we fit a model with 16 item responses (1 = methylated, 0 = not methylated) at the 16 gene loci. This model has been described in detail previously (17).

We examined whether the underlying propensity for hypermethylation, characterized by U, would also be a predictor of the tumor's clinical phenotype. Our conceptual framework is depicted in Figure 1, which shows a directed acyclic graph (35) with observed and latent variables depicted as in Skrondal et al. (36); here, s represents a phenotype, y a vector of methylation outcomes, m an underlying latent methylation variable and z a predictor of methylation. To this end, we used the predictions Û in a second stage analysis where we employed a polytomous (multinomial) logistic regression (37) to model the association of tumor stage on hypermethylation and additional covariates. This approach is characterized as follows:

Formula (1)


Figure 1
View larger version (12K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		Fig. 1. Causal diagram illustrating use of two-stage analysis method. In the diagram, s is survival or tumor stage, m is latent methylation, y is a vector of methylation item responses and z is a covariate such as smoking. If m were directly observed, it would not be necessary to include s in the model if one is interested in examining the relationship between m and z (35) This relationship, represented by E(m/z), depends only on p(m/z) (i.e. s and y could be ignored). In the case of propensity for promoter hypermethylation, m is not directly observed and must be measured via observables. In theory, y and s can (and should) both be used to measure m. However, s is distinct from y in that it includes multiple other sources of variation that add substantial noise relative to y when s serves as a surrogate for m. Additionally, as the probability calculation shows, the relationship between s and z must be correctly modeled if s and z are associated independently of m.

 
where S represents tumor stage, ranging across s = 1 (non-invasive low-grade), s = 2 (non-invasive high-grade) and s = 3 (invasive); W is a vector of covariates, including an intercept term and the latent variable Û, and {xi}s is a vector of corresponding regression coefficients with {xi}1 equal to the vector of zeros for the reference stage, non-invasive low-grade. Tumors classified as CIS were excluded from analysis due to the small number of tumor samples available.

Finally, we examined the associations between propensity for hypermethylation, again characterized by U, and patient survival. In a second stage analysis, we used the predictions, Û, and employed a Cox proportional hazards model to examine the logarithm of the hazard of death as a linear function of covariates, including Û. To account for the error introduced by predicting U via Û, we employed a bootstrap approach for estimating standard errors and bias-corrected and accelerated (BcA) confidence intervals (CIs) (38). All analyses were conducted in R version 2.1.1 (39), including custom software for the IRT model, available upon request.

To construct our IRT model, we used a stepwise selection combined with domain knowledge examining the effect of exposures on the latent methylation variable. Age and gender are included in the model as there have been reports of age-related methylation (40) and there is a well-established difference in the prevalence of bladder cancer by gender. We examined the effect of exposures which have been demonstrated to be associated with bladder cancer incidence including cigarette smoking and inorganic arsenic exposure (measured as toenail arsenic), both of which have been associated with gene-specific promoter hypermethylation in bladder cancer (12,15), as well as the possible association of occupational exposures [measured through work history questionnaires and classified dichotomously as high or low risk for occupational exposures (3)] and use of hair dyes with methylation silencing. Cigarette smoking was examined, as a variable comparing never, former and current smokers, as our previous work has suggested significant associations between current smoking and gene-specific hypermethylation (41), and there is evidence that current smoking and longer smoking duration are associated with increased risk for bladder cancer (42–44). As never and former smokers showed similar effect estimates for the latent trait, those groups we combined in the final model. We also examined models with smoking included as a continuous term representing duration (years smoked), intensity (cigarettes smoked per day) or pack-years smoked, but found that models using the smoking status variable to better fit the data. We also examined in the model, the effect of alteration of TP53, both as mutation of the TP53 gene or as aberrant IHC staining intensity [scored on a four-point scale, and dichotomized in the model as 0 and 1 versus 2 and 3 (31)], as alterations of TP53 measured as altered IHC staining have been associated with specific gene-promoter hypermethylation (15). The final model included age, gender, current smoking status and TP53 IHC staining as covariates in addition to the promoter hypermethylation status of the 16 genes examined.


   Results
Top
 Abstract
 Introduction
 Subjects and methods
 Results
Discussion
 Supplementary material
 Funding
 References
 
The demographics of the study population are shown in Table I and the prevalence of promoter hypermethylation of each of the genes examined is provided in Table II. Consistent with prior reports, we observed a correlation between promoter hypermethylation of various loci, prompting the use of a latent variable model to examine the characteristics of the propensity for methylation (supplementary Figure 1 is available at Carcinogenesis Online). We fit latent class models with 2–6 latent classes, as well as an IRT model [which essentially models an infinite number of latent classes as a continuous predictor (45)], and found the most parsimonious to be an IRT model that had a Bayesian Information Criterion value lower than the best (three class) latent class model (supplementary Table I is available at Carcinogenesis Online). Hence, we fit the IRT model to the described above to the 16 methylation item responses and included those covariates (age, gender, smoking status and TP53 IHC staining) that had a significant effect on the distribution of the latent methylation trait.


View this table:
[in this window]
[in a new window]

  		Table I. Characteristics of the bladder cancer cases 1994–1998 in New Hampshire

 


View this table:
[in this window]
[in a new window]

  		Table II. Prevalence of promoter hypermethylation in bladder cancer

 
Results of the IRT model, including estimates of the slope and intercept coefficients for each of the genes examined, as well as the estimates of the slope coefficient for the covariate terms and their bootstrap-derived standard errors and BcA 95% CIs are given in Table III. The distribution of the modeled latent propensity for methylation is depicted (supplementary Figure 2 is available at Carcinogenesis Online) in order to better illustrate the spread of the latent variable for interpretation in later analyses. Overall, the model shows statistically significant differences among the IRT slopes (Wald P < 0.0001), suggesting this approach models the underlying propensity for methylation better than simply summing the number of methylated loci in each sample. All but two loci show significant sensitivity to the underlying latent methylation variable, with item response slopes ranging from 0.30 (CDKN2A) to 2.47 (SFRP2), implying that hypermethylation events do not occur independently. Promoter hypermethylation of MLH1 and CDH1 showed only borderline statistically significant sensitivity to the underlying latent variable. This may be related to their relative low prevalences of ~3% each.


View this table:
[in this window]
[in a new window]

  		Table III. Item response model for bladder cancer promoter hypermethylation and covariates

 
This model suggests that age is related to the propensity for hypermethylation (overall Wald P value = 0.002), with those patients in the highest quartile of age at diagnosis (>70 years) having a significant, positive association with a propensity for hypermethylation (Table III). This association appears to be a local trend, as there was no appreciable trend in age among the three lower age groups; those patients in the lower quartiles had no significant or increasing estimates of their mean effect on the underlying latent variable. Patients of the male gender also showed a significant positive association with the hypermethylation latent trait (estimate 0.41, 95% BcA CI 0.03, 0.75). Tumors with altered TP53 protein, defined as intense IHC staining (2 or 3+) versus low or no staining, also showed a significant positive relationship with promoter hypermethylation. Consistent with previous reports about gene-specific hypermethylation events associated with tobacco smoking, this model suggests that current smokers, compared with never or former smokers, also have an increased propensity for promoter hypermethylation. If modeled separately, never and former smokers had similar estimates and were collapsed into a single category. Current smoking had a latent trait mean effect of 0.43 (95% BcA CI 0.10, 0.70) and performed better in the model than when we used duration (years of smoking), intensity (packs of cigarettes smoked per day) or pack-years smoked.

Our group and others have previously reported associations between gene-specific promoter hypermethylation as well as TP53 alteration and clinical features of the disease, particularly tumor stage. Thus, we asked whether the underlying propensity for hypermethylation also would be a predictor of the tumor's clinical phenotype. This was done by a polytomous logistic regression model (2), modeling disease stage (low stage/low grade, low stage/high grade and high stage) with the latent trait term as well as patient age, gender and TP53 IHC status (known to be important in prediction of disease stage or as possible confounders). Overall, neither age nor gender demonstrated a statistically significant association with stage. As has been previously reported, moderate to intense TP53 IHC staining was significantly associated with invasive stage disease, with an odds ratio of 4.3 (95% BcA CI 2.2, 9.2) for invasive stage disease compared with non-invasive/low-grade disease assuming the tumors were not of other disease stages. Interestingly, TP53 IHC was not correlated to non-invasive/high-grade status after adjusting for other factors. The hypermethylation latent trait was statistically significantly associated with both non-invasive/high-grade and invasive stages of disease (Table IV). For each one unit increase in the methylation latent trait, the odds of being a non-invasive/high-grade tumor increased by 2.1-fold (95% BcA CI 1.1, 3.5) compared with non-invasive/low-grade tumors, assuming the tumor was in either of those categories. The odds of being an invasive stage tumor increased by 2.9-fold (95% BcA CI 2.0, 4.4) compared with non-invasive/low-grade tumors, assuming the tumor was in either of those categories.


View this table:
[in this window]
[in a new window]

  		Table IV. Multinomial logistic regression model of tumor stage with methylation latent variable and covariatesa

 
We also used a Cox's proportional hazards model, predicting the outcome of death, with the hypermethylation latent variable Û as a predictor, and controlling again for patient age and gender, as well as TP53 IHC staining [which we have shown to be previously independently associated with patient survival (12)]. The results of this model are presented in Table V. We have specifically examined the relationship between the methylation latent trait and survival without tumor stage in the model, as our previous polytomous regression analysis suggests that these variables are highly correlated and biologically we believe that the propensity for promoter hypermethylation and stage exist in the same causal pathway. Thus, including both in the model would reduce power to see the associations of each with patient survival. In the model, every unit increase in the methylation latent trait increased the instantaneous hazard of death by 1.3 (95% BcA CI 1.0, 1.6, P < 0.02). Interestingly, TP53 IHC staining defined as 2 or 3+ compared with low or no IHC positivity no longer, had a significant relationship with patient survival time when controlling for promoter hypermethylation. As expected, age and patient gender remains significant predictors of poor prognosis.


View this table:
[in this window]
[in a new window]

  		Table V. Cox proportional hazards models of survival using methylation latent trait

 

   Discussion
Top
 Abstract
 Introduction
 Subjects and methods
 Results
 Discussion
Supplementary material
 Funding
 References
 
Epigenetic alterations, particularly, promoter hypermethylation-associated gene silencing, are known to be critically important in carcinogenesis, although the etiology and mechanism of gene-specific targeting for silencing remain unclear. Further, the silenced genes and pathways in bladder cancers span the range of cellular processes, interact in many cellular networks driving development and growth, and do not occur as independent events. Hence, we have studied the etiology and significance of promoter hypermethylation in bladder cancer by employing a large, population-based case series design, capturing the entire range of incident disease. We have taken cues from the field of systems biology in the analysis of these results, utilizing a statistical approach that specifically does not require assumptions of independence for the epigenetic events or the existence of a defined dichotomous phenotype and which allows for the study of the impact of methylation as a latent variable in a regression framework, consistent with the biological notion of effect modification.

In examining methylation as a latent variable, we observed a significant relationship with patient age at diagnosis, with those patients in the highest quartile of age at diagnosis showing the largest association with the methylation latent trait. A similar relationship has been observed in specific genes in aging colon tissue, although this has been thought to be limited to specific gene promoters (10,46,47). Our data suggest that older bladder cancer patients have an increased propensity for more general hypermethylation, a result consistent with reports examining specific genes in bladder cancer (40). This propensity for overall hypermethylation may be related to the aging process, to increased DNA damage and genomic instability related to aging or to an increased duration of carcinogen exposure, although we saw no specific effects, for example, of a relationship between the methylation latent trait and duration of cigarette smoking. We also observe a significant relationship with altered TP53 protein. TP53 pathway inactivation may be contributing to increased promoter hypermethylation as a response to the increasing genomic instability and recombination events leading to alterations in chromatin structure and thus DNA methylation related.

Similar to previous reports in bladder cancer linking current cigarette smoking at the time of diagnosis and gene-specific promoter hypermethylation events (12,15), we observed a significantly increased propensity for promoter hypermethylation in current smokers compared with never or former smokers. This adds to a growing body of evidence in a number of malignancies that cigarette smoke is not only a mutagenic carcinogen but also is associated with the genesis of epigenetic alterations in cancer (48–51). Our data, showing current smoking, rather than duration or intensity of smoking, to most alter promoter methylation suggest that the recency of exposure or the continuous exposure of the developing tumor to tobacco-related carcinogens may be critical in driving or selecting clones harboring epigenetic alterations. Indeed, tumors in never or former smoking patients will be characterized by different molecular alterations driven by different selective pressures. Importantly, these data suggest that the propensity for methylation requires continual selection pressure at the cellular level and might be positively impacted by changes in lifestyle.

The initial IRT model also suggests that men have an increased propensity for promoter hypermethylation compared with women. Men have a 3- to 6-fold higher incidence of bladder cancer than women which cannot be accounted for by differences in carcinogen exposure (52,53). Our results raise the possibility of a biological difference between tumors of men and women, similar to that posited in a study of lung cancer. Lai et al. (54) observed an association of gene hypermethylation with gender and their data suggested that exposure of lung cancer cell lines to ß-estradiol could reverse promoter hypermethylation-induced gene silencing. However, additional evidence connecting hormones and protection or promotion of gene promoter hypermethylation is lacking. In the case of bladder cancer, the anatomical differences between men and women might also play a role. Consistent with the concept that continued exposure and cell turnover are crucial for selection of clones with silenced genes (e.g. current smoking and age), it is tempting to speculate that chronic exposure of the epithelium to urinary carcinogens due to an inability of older men to completely void (consequent to prostatic hyperplasia or other urinary tract conditions) provides the needed selection pressure. Frequency of urination is associated with bladder cancer, with greater frequency and greater fluid intake being protective for the disease (55–57). More specific examination of voiding frequency and volume would be necessary to more thoroughly examine this potential link.

Compared with low-stage/low-grade tumors, we found an increased methylation propensity to be associated with greater odds of being either higher grade or higher stage, while TP53 inactivation predicted only higher stage (invasive) disease. Promoter hypermethylation, therefore, may be a critical factor in defining both higher grade, non-invasive disease, as well as predicting invasive disease. Like Blaveri et al. (58) who showed that gross genomic instability was lower in invasive stage disease than in non-invasive/high-grade disease, our data suggest that, in invasive disease, epigenetic alteration may be the dominant mode of inactivation of tumor suppressor genes. In non-invasive disease, genetic alterations (resultant of instability) characterized by chromosomal loss, transition and amplification may predominate.

Consistent with the forgoing, promoter hypermethylation modeled as a latent trait was associated with patient survival. As we believe the propensity for methylation and stage to be in the same causal pathway for survival, we employed a model examining the effect of the methylation latent trait controlled for age, gender and TP53 IHC staining, but not stage. In this model, the methylation latent trait was significantly associated with patient survival. If stage is included in the model, the hazard ratio (HR) for both stage and the modelled latent trait would be non-significant, and the HR for both reduced. Taken together with the multivariate model linking methylation propensity to tumor stage, these results imply that methylation is causally related to tumor stage and thus part of the causal pathway of patient survival.

We have previously shown that TP53 alteration, as demonstrated by intense IHC staining was significantly associated with invasive stage tumors and patient survival (31). In the current analysis, TP53 IHC staining was not a significant predictor of survival, but remained a significant predictor of invasive stage tumors, whereas the propensity for hypermethylation is associated with both non-invasive/high-grade and invasive stage tumors as well as with patient survival. This suggests distinct clinical phenotypes associated with these alterations, in that TP53 staining likely strongly predicts all invasive stage disease, similar to the way in which mutation of the FGFR3 gene predicts non-invasive disease, specifically (59,60). At the same time, the propensity for promoter hypermethylation may specifically predict the most aggressive and deadly types of bladder cancer, regardless of their morphology and staging at diagnosis. This very strongly suggests that clinical prospective studies closely examine the utility of the methylation propensity (rather than single gene silencing) as a marker of patient outcome.

Overall, our results suggest an etiology for both promoter hypermethylation as well as for the genesis of the crucial epigenetic alterations that create a tumor phenotype and alter patient outcome. Our model indicates that continuous tobacco carcinogen exposure, in combination with increasing age, and male gender, drives and enhances the selection of epigenetically altered cells. Additionally, these models show that the invasive form of this disease is characterized by a predominance of epigenetic events, whereas the non-invasive forms, consistent with previous reports, are characterized by gross genomic instability. Thus, promoter hypermethylation profiling of tumor specimens may help to define those more probably to progress and lead to increased mortality. It may also delineate patients who would benefit from more radical treatment or more careful follow-up.


   Supplementary material
Top
 Abstract
 Introduction
 Subjects and methods
 Results
 Discussion
 Supplementary material
Funding
 References
 
Supplementary Figures 1 and 2 and Table I can be found at http://carcin.oxfordjournals.org/


   Funding
Top
 Abstract
 Introduction
 Subjects and methods
 Results
 Discussion
 Supplementary material
 Funding
References
 
National Institutes of Health (ES05974, ES007373 and CA100679), Flight Attendants Medical Research Institute Young Clinical Scientist Award (to C.J.M.).


   Acknowledgments
 
Conflict of Interest statement: None declared.


   References
Top
 Abstract
 Introduction
 Subjects and methods
 Results
 Discussion
 Supplementary material
 Funding
 References
 

  1. Jemal A, et al. Cancer statistics. CA Cancer J. Clin. (2006) 56:106–130.[Abstract/Free Full Text]
  2. Vineis P, et al. Levelling-off of the risk of lung and bladder cancer in heavy smokers: an analysis based on multicentric case-control studies and a metabolic interpretation. Mutat. Res. (2000) 463:103–110.[CrossRef][ISI][Medline]
  3. Colt JS, et al. Occupation and bladder cancer risk in a population-based case-control study in New Hampshire. Cancer Causes Control (2004) 15:759–769.[CrossRef][ISI][Medline]
  4. Pelucchi C, et al. Mechanisms of disease: the epidemiology of bladder cancer. Nat. Clin. Pract. Urol. (2006) 3:327–340.[CrossRef][ISI][Medline]
  5. Knowles MA. The genetics of transitional cell carcinoma: progress and potential clinical application. BJU Int. (1999) 84:412–427.[CrossRef][ISI][Medline]
  6. Stein JP, et al. Radical cystectomy for invasive bladder cancer: long-term results of a standard procedure. World J. Urol. (2006) 24:296–304.[CrossRef][ISI][Medline]
  7. Hermann GG, et al. The influence of the level of lamina propria invasion and the prevalence of p53 nuclear accumulation on survival in stage T1 transitional cell bladder cancer. J. Urol. (1998) 159:91–94.[CrossRef][ISI][Medline]
  8. Dinney CP, et al. Focus on bladder cancer. Cancer Cell (2004) 6:111–116.[CrossRef][ISI][Medline]
  9. Jones PA, et al. DNA methylation in bladder cancer. Eur. Urol. (1998) 33(Suppl. 4):7–8.
  10. Toyota M, et al. CpG island methylator phenotypes in aging and cancer. Semin. Cancer Biol. (1999) 9:349–357.[CrossRef][ISI][Medline]
  11. Toyota M, et al. CpG island methylator phenotype in colorectal cancer. Proc. Natl Acad. Sci. USA (1999) 96:8681–8686.[Abstract/Free Full Text]
  12. Marsit CJ, et al. Epigenetic inactivation of SFRP genes and TP53 alteration act jointly as markers of invasive bladder cancer. Cancer Res. (2005) 65:7081–7085.[Abstract/Free Full Text]
  13. Maruyama R, et al. Aberrant promoter methylation profile of bladder cancer and its relationship to clinicopathological features. Cancer Res. (2001) 61:8659–8663.[Abstract/Free Full Text]
  14. Stoehr R, et al. Deletions of chromosome 8p and loss of sFRP1 expression are progression markers of papillary bladder cancer. Lab. Invest. (2004) 84:465–478.[CrossRef][ISI][Medline]
  15. Marsit CJ, et al. Carcinogen exposure and gene promoter hypermethylation in bladder cancer. Carcinogenesis (2006) 27:112–116.[Abstract/Free Full Text]
  16. Salem C, et al. Progressive increases in de novo methylation of CpG islands in bladder cancer. Cancer Res. (2000) 60:2473–2476.[Abstract/Free Full Text]
  17. Marsit CJ, et al. Examination of a CpG island methylator phenotype and implications of methylation profiles in solid tumors. Cancer Res. (2006) 66:10621–10629.[Abstract/Free Full Text]
  18. Houseman EA, et al. Feature-specific penalized latent class analysis for genomic data. Biometrics (2006) 62:1062–1070.[CrossRef][ISI][Medline]
  19. Karagas MR, et al. Design of an epidemiologic study of drinking water arsenic exposure and skin and bladder cancer risk in a U.S. population. Environ. Health Perspect. (1998) 106(Suppl. 4):1047–1050.[CrossRef][ISI][Medline]
  20. Herman JG, et al. Methylation-specific PCR: a novel PCR assay for methylation status of CpG islands. Proc. Natl Acad. Sci. USA (1996) 93:9821–9826.[Abstract/Free Full Text]
  21. van Engeland M, et al. K-ras mutations and RASSF1A promoter methylation in colorectal cancer. Oncogene (2002) 21:3792–3795.[CrossRef][ISI][Medline]
  22. Virmani AK, et al. Aberrant methylation of the adenomatous polyposis coli (APC) gene promoter 1A in breast and lung carcinomas. Clin. Cancer Res. (2001) 7:1998–2004.[Abstract/Free Full Text]
  23. Virmani A, et al. Aberrant methylation of TMS1 in small cell, non small cell lung cancer and breast cancer. Int. J. Cancer (2003) 106:198–204.[CrossRef][ISI][Medline]
  24. Sathyanarayana UG, et al. Molecular detection of noninvasive and invasive bladder tumor tissues and exfoliated cells by aberrant promoter methylation of laminin-5 encoding genes. Cancer Res. (2004) 64:1425–1430.[Abstract/Free Full Text]
  25. Suzuki H, et al. Epigenetic inactivation of SFRP genes allows constitutive WNT signaling in colorectal cancer. Nat. Genet. (2004) 36:417–422.[CrossRef][ISI][Medline]
  26. Zochbauer-Muller S, et al. Aberrant promoter methylation of multiple genes in non-small cell lung cancers. Cancer Res. (2001) 61:249–255.[Abstract/Free Full Text]
  27. Toyooka S, et al. Aberrant methylation and simian virus 40 tag sequences in malignant mesothelioma. Cancer Res. (2001) 61:5727–5730.[Abstract/Free Full Text]
  28. Chan AO, et al. CpG island methylation in aberrant crypt foci of the colorectum. Am. J. Pathol. (2002) 160:1823–1830.[Abstract/Free Full Text]
  29. Yamashita K, et al. A tumor-suppressive role for trypsin in human cancer progression. Cancer Res. (2003) 63:6575–6578.[Abstract/Free Full Text]
  30. Eads CA, et al. MethyLight: a high-throughput assay to measure DNA methylation. Nucleic Acids Res. (2000) 28:E32.[CrossRef][Medline]
  31. Kelsey KT, et al. A population-based study of immunohistochemical detection of p53 alteration in bladder cancer. Br. J. Cancer (2004) 90:1572–1576.[CrossRef][ISI][Medline]
  32. Sanchez B, et al. Structural equation models—a review with applications to environmental epidemiology. J.Am.Stat.Assoc. (2005) 100:1443–1455.[CrossRef][ISI]
  33. Bartholomew DJ, et al. Latent Variable Models and Factor Analysis (1999) New York: Oxford University Press.
  34. Hambleton RK, et al. Latent trait models and their use in analysis of educational testing data. J. Educ. Meas. (1977) 14:75–96.[CrossRef]
  35. Pearl J. Causality: Models, Reasoning, and Inference (2000) New York: Cambridge University Press.
  36. Skrondal A, et al. Generalized Latent Variable Modeling: Multilevel, Longitudinal, and Structural Equation Models (2004) Boca Raton, FL: Chapman and Hall.
  37. Agresti A. 7.1 Nominal responses: baseline-category logit models. In: Categorical Data Analysis, 2nd Edition (2002) Hoboken, NJ: John Wiley & Sons. 267–272.
  38. DiCiccio T, et al. Bootstrap confidence intervals. Stat. Sci. (1996) 11:189–228.[CrossRef][ISI]
  39. R Development Core Team. R: a language and environment for statistical computing (2005) Vienna, Austria: R Foundation for Statistical Computing.
  40. Yates DR, et al. Methylational urinalysis: a prospective study of bladder cancer patients and age stratified benign controls. Oncogene (2006) 25:1984–1988.[CrossRef][ISI][Medline]
  41. Marsit CJ, et al. Carcinogen exposure and gene promoter hypermethylation in bladder cancer. Carcinogenesis (2005).
  42. Jee SH, et al. Smoking and cancer risk in Korean men and women. Cancer Causes Control (2004) 15:341–348.[CrossRef][ISI][Medline]
  43. Gritz ER, et al. Smoking, the missing drug interaction in clinical trials: ignoring the obvious. Cancer Epidemiol. Biomarkers Prev. (2005) 14:2287–2293.[Abstract/Free Full Text]
  44. Zeegers MP, et al. A prospective study on active and environmental tobacco smoking and bladder cancer risk (The Netherlands). Cancer Causes Control (2002) 13:83–90.[CrossRef][ISI][Medline]
  45. Lindsay B, et al. Semiparametric estimation in the Rasch model and related exponential response models, including simple latent class model for item analysis. J. Am. Stat. Assoc. (1991) 86:96–107.[CrossRef][ISI]
  46. Baylin SB, et al. Alterations in DNA methylation: a fundamental aspect of neoplasia. Adv. Cancer Res. (1998) 72:141–196.[ISI][Medline]
  47. Ahuja N, et al. Aging, methylation and cancer. Histol. Histopathol. (2000) 15:835–842.[ISI][Medline]
  48. Belinsky SA. Silencing of genes by promoter hypermethylation: key event in rodent and human lung cancer. Carcinogenesis (2005) 26:1481–1487.[Abstract/Free Full Text]
  49. Kim DH, et al. p16(INK4a) and histology-specific methylation of CpG islands by exposure to tobacco smoke in non-small cell lung cancer. Cancer Res. (2001) 61:3419–3424.[Abstract/Free Full Text]
  50. Marsit CJ, et al. Hypermethylation of RASSF1A and BLU tumor suppressor genes in non-small cell lung cancer: implications for tobacco smoking during adolescence. Int. J. Cancer (2005) 114:219–223.[CrossRef][ISI][Medline]
  51. Toyooka S, et al. Smoke exposure, histologic type and geography-related differences in the methylation profiles of non-small cell lung cancer. Int. J. Cancer (2003) 103:153–160.[CrossRef][ISI][Medline]
  52. Puente D, et al. Gender-related differences in clinical and pathological characteristics and therapy of bladder cancer. Eur. Urol. (2003) 43:53–62.[CrossRef][ISI][Medline]
  53. Hartge P, et al. Unexplained excess risk of bladder cancer in men. J. Natl Cancer Inst. (1990) 82:1636–1640.[Abstract/Free Full Text]
  54. Lai JC, et al. Gender difference in estrogen receptor alpha promoter hypermethylation and its prognostic value in non-small cell lung cancer. Int. J. Cancer (2005) 117:974–980.[CrossRef][ISI][Medline]
  55. Radosavljevic V, et al. Non-occupational risk factors for bladder cancer: a case-control study. Tumori (2004) 90:175–180.[ISI][Medline]
  56. Michaud DS, et al. Fluid intake and the risk of bladder cancer in men. N. Engl. J. Med. (1999) 340:1390–1397.[Abstract/Free Full Text]
  57. Villanueva CM, et al. Total and specific fluid consumption as determinants of bladder cancer risk. Int. J. Cancer (2006) 118:2040–2047.[CrossRef][ISI][Medline]
  58. Blaveri E, et al. Bladder cancer stage and outcome by array-based comparative genomic hybridization. Clin. Cancer Res. (2005) 11:7012–7022.[Abstract/Free Full Text]
  59. Sibley K, et al. Frequency of fibroblast growth factor receptor 3 mutations in sporadic tumours. Oncogene (2001) 20:4416–4418.[CrossRef][ISI][Medline]
  60. van Rhijn BW, et al. The fibroblast growth factor receptor 3 (FGFR3) mutation is a strong indicator of superficial bladder cancer with low recurrence rate. Cancer Res. (2001) 61:1265–1268.[Abstract/Free Full Text]
Received March 13, 2007; revised April 27, 2007; accepted May 9, 2007.


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 Cancer Res.Home page
E. M. Wolff, G. Liang, C. C. Cortez, Y. C. Tsai, J. E. Castelao, V. K. Cortessis, D. D. Tsao-Wei, S. Groshen, and P. A. Jones
RUNX3 Methylation Reveals that Bladder Tumors Are Older in Patients with a History of Smoking
Cancer Res., August 1, 2008; 68(15): 6208 - 6214.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow Supplementary Data
Right arrowOA All Versions of this Article:
28/8/1745    most recent
bgm116v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (1)
Google Scholar
Right arrow Articles by Marsit, C. J.
Right arrow Articles by Kelsey, K. T.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Marsit, C. J.
Right arrow Articles by Kelsey, K. T.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?