|Home | About | Journals | Submit | Contact Us | Français|
Genome-wide DNA hypomethylation plays has an important role in genomic instability and colorectal carcinogenesis. However, the relationship between cellular DNA methylation level and patient outcome remains uncertain. Using 643 colon cancers in two independent prospective cohorts, we quantified DNA methylation in repetitive long interspersed nucleotide element-1 (LINE-1) elements using pyrosequencing, which is a good indicator of global DNA methylation level. We used Cox proportional hazard models to calculate hazard ratios (HRs) of colon cancer–specific and overall mortality, adjusting for patient and tumoral features, including CpG island methylator phenotype (CIMP). Statistical tests were two-sided. LINE-1 hypomethylation was linearly associated with a statistically significant increase in colon cancer–specific mortality (for a 30% decrease in LINE-1 methylation: multivariable HR = 2.37, 95% confidence interval [CI] = 1.42 to 3.94; Ptrend < .001) and overall mortality (multivariable HR = 1.85, 95% CI = 1.25 to 2.75; Ptrend = .002). The association was consistent across the two independent cohorts and strata of clinical and molecular characteristics, including sex, age, tumor location, stage, and CIMP, microsatellite instability, KRAS, BRAF, p53, and chromosomal instability status. In conclusion, tumoral LINE-1 hypomethylation is independently associated with shorter survival among colon cancer patients.
DNA hypomethylation is involved in genomic instability and colorectal carcinogenesis, but whether it is associated with the survival of colon cancer patients is unknown.
Prospective study of two colon cancer patient populations to evaluate the association between DNA methylation, which was assayed in long interspersed nucleotide element-1 (LINE-1), and survival. Clinical and molecular characteristics of the tumors were included in the analyses.
LINE-1 hypomethylation was associated with higher colon cancer–specific and overall mortality, independent of other tumor characteristics.
LINE-1 hypomethylation may be a prognostic indicator of patient survival in colon cancer.
Treatment data were limited for the patients studied. The patients were all health care professionals, and it is unknown whether the findings are applicable to the general population. The mechanism involved is unknown.
From the Editors
DNA methylation is a major epigenetic mechanism in X-chromosome inactivation, imprinting, and repression of endogenous retroviruses (1,2). Repetitive nucleotide elements (eg, long interspersed nucleotide element-1 [LINE-1]) contain numerous CpG dinucleotides, and previous studies (3–5) have shown that LINE-1 methylation level has been reported to be a good indicator of cellular 5-methylcytosine level (ie, global DNA methylation level). Genome-wide DNA hypomethylation has an important role in genomic instability by reactivating transposable DNA sequences during colorectal carcinogenesis (1,6–15). However, the influence of global DNA hypomethylation on the prognosis of colon cancer patients remains uncertain.
We used two independent prospective cohort studies: the Nurses’ Health Study (121700 women followed since 1976) (16,17) and the Health Professionals Follow-up Study (51500 men followed since 1986) (17,18). Every 2 years, participants were sent follow-up questionnaires to update information on potential risk factors and to identify newly diagnosed cancer, including colon cancer (16–18). Informed consent was obtained from all study subjects. This study was approved by the Human Subjects Committees and the Institutional Review Boards of Harvard School of Public Health and Brigham and Women's Hospital. We collected paraffin-embedded tissue blocks from hospitals where patients underwent tumor resections (17). Based on availability of tissue specimens, we included 643 colon cancers that were diagnosed up to June 30, 2002. Patients were observed until death or June 30, 2006, whichever came first (19).
KRAS and BRAF sequencing and analyses for microsatellite instability (MSI) and loss of heterozygosity (chromosomal instability, CIN) were performed on DNA from paraffin-embedded tissue (20–22). Immunohistochemistry for p53 was performed using a mouse monoclonal antibody (clone DO-1, dilution 1:50; Calbiochem, San Diego, CA) as previously described (23). Bisulfite DNA treatment and real-time polymerase chain reaction [MethyLight (24)] were validated and performed (25). We quantified promoter methylation in eight CpG island methylator phenotype (CIMP)–specific promoters (CACNA1G, CDKN2A, CRABP1, IGF2, MLH1, NEUROG1, RUNX3, and SOCS1) (26–28). CIMP-high was defined as methylation in at least six of the eight promoters (26).
To accurately quantify relatively high LINE-1 methylation levels, we used pyrosequencing (29). LINE-1 methylation as measured by pyrosequencing has been reported to be a good indicator of cellular 5-methylcytosine level (ie, global DNA methylation level) (3,5,30). LINE-1 methylation levels in the 643 tumors were approximately normally distributed (mean = 61.2%; standard deviation = 9.4%) (Supplementary Figure 1, available online).
Cox proportional hazard models were used to calculate hazard ratios (HRs) of death according to LINE-1 methylation level, adjusted for age, sex, year of diagnosis, tumor location, stage, grade, and CIMP, MSI, CIN, KRAS, BRAF, and p53 status. The proportionality of hazards assumption was satisfied by evaluating time-dependent variables, which were the cross product of LINE-1 methylation level and survival time (P = .31 for colon cancer–specific mortality and P = .73 for overall mortality). The primary statistical analysis was the linear test for trend with LINE-1 methylation as a continuous variable, to avoid the possibility of selecting cut points with maximal P values. We also performed analyses using categories (Supplementary Table 1, available online) that were based on the overall distribution of LINE-1 methylation levels. We used SAS version 9.1 (SAS Institute, Cary, NC) for statistical analyses. All statistical tests were two-sided, and P values less than .05 were considered statistically significant.
LINE-1 hypomethylation was associated with a statistically significant increase in colon cancer–specific mortality (univariate analysis Ptrend < .001 and multivariable analysis Ptrend < .001) (Table 1). For a 30% decrease in LINE-1 methylation, the multivariable HR for colon cancer–specific mortality was 2.37 (95% confidence interval [CI] = 1.42 to 3.94). LINE-1 hypomethylation was also associated with a statistically significant increase in overall mortality (univariate analysis Ptrend = .008 and multivariable analysis Ptrend = .002).
To facilitate display of the results, we also grouped patients according to four categories of LINE-1 methylation (≥75%, 60%–75%, 45%–60%, and <45%). Compared with patients who had 75% or greater LINE-1 methylated tumors, those with less than 45% LINE-1 methylation had higher colon cancer–specific mortality (multivariable HR = 5.00; 95% CI = 1.92 to 13.1) and a similar increase in overall mortality (multivariable HR = 4.73; 95% CI = 2.20 to 10.2) (Table 1).
The colon cancer–specific survival probability at 5 years was progressively lower as LINE-1 methylation level declined (≥75% LINE-1 methylation: survival probability = 0.85, 95% CI = 0.69 to 0.93; 60%–75% methylation: survival probability = 0.83, 95% CI = 0.79 to 0.87; 45%–60% methylation: survival probability = 0.74, 95% CI = 0.68 to 0.80; and <45% methylation: survival probability = 0.54, 95% CI = 0.35 to 0.70; P < .001) (Figure 1, A). Similarly, overall survival at 5 years was statistically significantly lower for patients with relatively hypomethylated LINE-1 (≥75% LINE-1 methylation: survival probability = 0.80, 95% CI = 0.64 to 0.89; 60%–75% methylation: survival probability = 0.77, 95% CI = 0.72 to 0.81; 45%–60% methylation: survival probability = 0.71, 95% CI = 0.64 to 0.76; and <45% methylation: survival probability = 0.48, 95% CI = 0.30 to 0.64; P = .02) (Figure 1, B).
We examined the possibility of a nonlinear relationship between LINE-1 hypomethylation and mortality by using a nonparametric method with restricted cubic splines (31), which was independent of predetermined LINE-1 categorizations. The HRs for colon cancer–specific and overall mortality increased continuously with decreasing LINE-1 methylation (Figure 1, C and D). Notably, the lower limits of the 95% confidence intervals remained above 1.0 for all values of LINE-1 methylation, suggesting a robust association between LINE-1 hypomethylation and mortality.
We also generated a receiver operator characteristics curve for the prediction of 5-year colon cancer–specific survival using all possible cutoffs (positive vs negative) for LINE-1 hypomethylation (area under the curve = 0.56) (Supplementary Figure 2, available online). This result implies that there were limitations to using LINE-1 methylation as a binary diagnostic test to predict patient outcome at a fixed time point.
Finally, we examined the influence of LINE-1 hypomethylation on colon cancer–specific mortality across strata of other clinical and molecular characteristics (Supplementary Figure 3, available online). The association between LINE-1 hypomethylation and colon cancer–specific mortality was not modified by any of the clinical or molecular variables nor was it different between the two cohorts (the Health Professionals Follow-up Study and the Nurses’ Health Study; Pinteraction = .74).
Although prognostic factors in colon cancer have been extensively studied (32–38), little is known regarding the prognostic value of global DNA hypomethylation. The relationship between genome-wide DNA hypomethylation and clinical outcome has been examined in prostate, liver, and ovarian cancers and leukemia (39–42). Among those studies, only the study of chronic myeloid leukemia (42) showed a statistically significant association between global DNA hypomethylation and poor survival. In colon cancer, Frigola et al. (43) analyzed 93 tumors using the amplification of intermethylated sites (AIMS) method and observed a trend toward poor survival in DNA hypomethylated tumors; however, the results did not reach statistical significance.
The mechanism by which global DNA hypomethylation may confer a poor prognosis remains speculative. Genome-wide DNA hypomethylation has been associated with genomic instability (7–15), which may confer poor prognosis. Transcriptional dysregulation might be another possible mechanism, and activation of proto-oncogenes, endogenous retroviruses, or transposable elements might affect tumor aggressiveness. A third possible mechanism involves inflammatory mediators and oxidative stress; the latter has been associated with genomic DNA hypomethylation (44,45). Activation of the inflammatory pathway has been associated with poor prognosis in colon cancer (46).
The study has potential limitations. In the two cohorts, data on cancer treatment were limited. It is unlikely that chemotherapy use differed according to tumoral LINE-1 methylation level, since such data were not available to patients or treating physicians. Other potential study limitations include a nonperfect correlation (correlation coefficient < 1) between LINE-1 methylation and global DNA methylation level and that these health professionals–based cohort studies may not be completely representative of the general population.
In summary, the results from this large prospective cohort study suggest that genome-wide DNA hypomethylation as measured in LINE-1 is independently associated with poor survival among patients with colon cancer. The adverse effect of LINE-1 hypomethylation on prognosis persisted across the two independent prospective cohort studies as well as strata of other clinical or tumoral characteristics. These findings may have considerable clinical implications. Future studies are needed to confirm this association as well as examine potential mechanisms by which genome-wide DNA hypomethylation affects tumor behavior.
The US National Institute of Health (P01 CA87969, P01 CA55075, P50 CA127003, and K07 CA122826 to S.O.); the Bennett Family Fund; the Entertainment Industry Foundation. K. Nosho was supported by a fellowship grant from the Japan Society for Promotion of Science. The content is solely the responsibility of the authors and does not necessarily represent the official views of National Cancer Institute or National Institutes of Health. Funding agencies did not have any role in the design of the study; the collection, analysis, or interpretation of the data; the decision to submit the manuscript for publication; or the writing of the manuscript.
We deeply thank the Nurses' Health Study and Health Professionals Follow-up Study cohort participants who have generously agreed to provide us with biologic specimens and information through responses to questionnaires and hospitals and pathology departments throughout the United States for providing us with tumor tissue materials.