SEARCH

SEARCH BY CITATION

Keywords:

  • colorectal adenoma;
  • methylation;
  • smoking;
  • normal rectum;
  • field defect

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

BACKGROUND:

Colorectal cancer (CRC) is 1 of the leading causes of death in the Western world. CRC develops from premalignant lesions, chiefly colorectal adenomas. Currently, the most accurate and recommended screening method for finding colorectal adenomas is colonoscopy performed on all individuals aged >50 years. However, the costs and risks associated with this procedure are relatively high. The objectives of the current study were to correlate epigenetic alterations that occur in normal rectal mucosa, smoking status, and age with the presence or absence of concomitant colorectal adenomas and to assess the potential clinical value of methylation in normal rectal biopsies as a screening assay for the presence of polyps and, hence, the need for a full colonoscopy.

METHODS:

One hundred thirteen normal rectal mucosal biopsies from 113 patients were studied. DNA was extracted, modified with sodium bisulfite, and subjected to real-time quantitative, methylation-specific polymerase chain reaction analysis for the following genes: adenomatous polyposis coli (APC); cadherin 1, type 1, E-cadherin (epithelial) (CDH1); estrogen receptor 1 (ESR1); cytokine high in normal 1 (HIN1); hyperplastic polyposis protein 1 (HPP1); O-6 methylguanine-DNA methyltransferase (MGMT); neural epidermal growth factor-like 1 (NELL1); splicing factor 3B, 14-kDa subunit (p14); cyclin-dependent kinase (CDK) inhibitor 2B (inhibits CDK4) (p15); retinoic acid receptor beta (RARβ); somatostatin (SST); tachykinin, precursor 1 (TAC1); and tissue inhibitor of metalloproteinase (TIMP) metallopeptidase inhibitor 3 (TIMP3). Data were then analyzed using several proprietary software programs.

RESULTS:

By using several sets of genes, clinical characteristics, and multivariate analyses, the authors developed a prediction model for the presence of concomitant colorectal adenomas at the time of rectal biopsy. They also observed strong correlations between smoking status and rectal methylation pattern and between smoking status and the presence or risk of concomitant adenomas.

CONCLUSIONS:

A prediction model was developed for the presence of colorectal adenomas based on gene methylation patterns in the normal rectum. The results indicated that these genes may be involved in early stages of adenoma formation. The observed epigenetic alterations in these markers may be caused in part by the effects of smoking and/or age. Normal rectal methylation may be useful as a biomarker for narrowing the population in need of screening colonoscopy. Cancer 2010. © 2010 American Cancer Society.

Colorectal cancer (CRC) is 1 of the leading causes of death in the United States.1 CRC develops from premalignant lesions, chiefly colorectal adenomas.2 Inactivation of tumor suppressor genes because of promoter hypermethylation is 1 of the main mechanisms underlying this inactivation.3 Additional proposed mechanisms include microsatellite instability,4 point mutations,5-7 and nutritional factors.8, 9 Currently, the most accurate and widely recommended screening method for diagnosing colorectal adenomas is colonoscopy, which should be performed in all individuals aged ≥50 years with repeat procedures recommended based on initial findings. However, the cost and potential risks associated with colonoscopy are considerable, and patient compliance with these recommendations is, at best, incomplete.10 Several studies have proposed strategies based on clinical data to reduce the frequency of colonoscopic screening and surveillance.11-13 In the current study, our objective also was to develop a strategy for reducing the number of initial (screening) and follow-up (surveillance) colonoscopies from their current number. We sought to identify epigenetic marker assays in normal rectal (NR) mucosa that would be less invasive, less risky, and less costly than colonoscopy. We postulated a field defect in the colons of patients with adenomas, comprising epigenetic alterations more widespread than those in the adenoma per se.14, 15 This field defect concept implies subtle changes in a large region of a tissue or organ that is devoid of accompanying, detectable, phenotypic manifestations. We also examined several clinical risk factors that have potential impact on the development of adenomas, including smoking habits and age.16-18

A resurgent trend in epigenetics research has been studies of hypomethylation and its potential role in carcinogenesis.19, 20 This trend is of particular relevance to our study in light of publications that have linked smoking (itself a risk factor for adenoma development) to low serum folate levels21 and that have associated low folate levels with diminished methylation in various tissues, including the gastrointestinal tract.22, 23 Notwithstanding these publications, to date, few studies have connected smoking with reduced methylation levels in normal tissues.

Considering the above literature, we sought to construct a diagnostic model to predict concurrent colorectal adenomas based on methylation patterns in NR. We postulated that, because of field carcinogenesis, a difference in NR mucosa would prevail in individuals with versus individuals without a concomitant adenoma. We also theorized that altered methylation occurring in histologically NR mucosae in patients with adenoma represented an intermediate molecular step between completely normal mucosa and complete adenomagenesis. Although we relied on methylation both as a cause and as an indicator, we also included clinical factors (viz, smoking and age) in our model, and we attempted to dissect the potential impact of these clinical factors on both adenomagenesis and abnormal methylation.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

Samples

Single NR biopsies obtained from each of 113 patients were studied (for clinical and demographic data, see Table 1). Patients were selected randomly from a cohort of 300 patients who underwent serial colonoscopy by a single investigator (S.J.M.) over a 3-year period (from 2002 to 2005) in the Outpatient Endoscopy Unit at the Baltimore Veterans Affairs Medical Center; because of the particular nature of that institution, the majority of its patients are men, whereas smokers represent a much greater percentage compared with the general population. These 2 features are analyzed below (see Discussion). NR was biopsied during initial (screening) or follow-up (surveillance) colonoscopy, the biopsies were frozen on dry ice, then they were transferred to liquid nitrogen for long-term storage to preserve DNA quality. For the purpose of the study, samples were thawed and aliquotted, and DNA was extracted using a Qiagen DNeasy kit (Qiagen, Valencia, Calif); water, rather than elution buffer (normally indicated by the protocol because of the downstream quantitative polymerase chain reaction [qPCR] reactions, which may be influenced by any salt contamination), was used during the last elution step. DNA was frozen and maintained at −20°C until the next step. DNA was bisulfite-modified using a Qiagen Epitect kit, then subjected to real-time, methylation-specific qPCR (qMSP) on an ABI 7900HT sequence detector (Applied Biosystems, Inc., Carlsbad, Calif). The following genes were selected, based on their known involvement in gastrointestinal carcinogenesis: adenomatous polyposis coli (APC); cadherin 1, type 1, E-cadherin (epithelial) (CDH1); estrogen receptor 1 (ESR1); cytokine high in normal 1 (HIN1); hyperplastic polyposis protein 1 (HPP1); O-6 methylguanine-DNA methyltransferase (MGMT)19; neural epidermal growth factor-like 1 (NELL1); splicing factor 3B, 14-kDa subunit (p14); cyclin-dependent kinase (CDK) inhibitor 2B (inhibits CDK4) (p15); retinoic acid receptor beta (RARβ); somatostatin (SST); tachykinin, precursor 1 (TAC1); and tissue inhibitor of metalloproteinase (TIMP) metallopeptidase inhibitor 3 (TIMP3).20 β-Actin served as an internal control, and fully methylated DNA (Millipore, Billerica, Mass) was used as an external control. Primer-probe sets for qMSP either were derived from the literature or were custom-designed to satisfy specified MSP conditions (ie, annealing temperature as close as possible to 60°C, amplicon length <100-120 base pairs). A methylation index (MI), representing the ratio of densely methylated DNA in the sample at the target sequence relative to the fully methylated positive control, was calculated as follows: MI = (Ts/Tc)/(As/Ac) where Ts and Tc represent levels of target gene methylation in the sample DNA and in the control DNA, respectively, and As and Ac correspond to the amplified β-actin gene (ACTB) level in the sample DNA and in the control DNA, respectively.

Table 1. Clinical and Demographic Data
VariableNo. of Patients (%)
TotalSmokersNonsmokersPolyp PresentPolyp Not Present
  1. SD indicates standard deviation.

No. of patients113 (100)80 (70.8)33 (29.2)59 (52.2)54 (47.7)
Age: Mean ± SD, y65.2 ± 11.264 ± 11.368.6 ± 10.466.5 ± 11.463.8 ± 10.8
Smokers80 (70.8)  42 (71.1)38 (70.3)
Polyp present59 (52.2)43 (53.8)16 (48.5)  
Multiple polyps 19 (50)14 (87.5)33 (56%) 
Polyp size: Average ± SD, mm 7.4 ± 5.615.2 ± 16.79.3 ± 9 
Race     
 Black322661616
 White4327162815
 Unknown383081523
Sex     
 Men10576295550
 Women84444

Statistical Analysis

Raw methylation data were analyzed first by using Sequence Detector software (Applied Biosystems, Inc.) and then by using SPSS 16.0 statistical software (SPSS Inc., Chicago, Ill) and Microsoft Excel (Microsoft Corp., Redmond, Wash). Smoking status and quantity (pack-years) were assessed based on patient history. After analyzing the clinical data, we divided the patients into 3 groups: 1) nonsmokers, 2) smokers with known pack-year data, and 3) smokers without known pack-year data (Table 1). Because of the sensitive nature of any quantitative assay, all wet laboratory procedures were done in a relatively short time.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

On the basis of results from the Student t test, methylation levels of MGMT, p14, and RAR-β differed significantly between individuals with versus without adenoma at the time of NR biopsy (Table 2). In an attempt to further investigate this difference, next, we divided patients into age groups and performed the same analyses. These results are presented in Table 3.

Table 2. Mean Normalized Methylation Values in Patients With Adenomas Versus Patients Without Adenomas at the Time of Biopsy and P Values Derived From Student T Tests
VariableMean Normalized Methylation, %
APCCDH1ESR1HIN1HPP1MGMTNELL1p14p15RARβSSTTAC1TIMP3
  • APC indicates adenomatous polyposis coli; CDH1, cadherin 1, type 1, E-cadherin (epithelial); ESR1, estrogen receptor 1; HIN1, cytokine high in normal 1; HPP1, hyperplastic polyposis protein 1; MGMT, O-6-methylguanine-DNA methyltransferase; NELL1, neural epidermal growth factor-like 1; p14, splicing factor 3B, 14-kDa subunit; p15, cyclin-dependent kinase (CDK) inhibitor 2B (inhibits CDK4); RARβ, retinoic acid receptor beta; SST, somatostatin; TAC1, tachykinin, precursor 1; TIMP3, tissue inhibitor of metalloproteinase-metallopeptidase inhibitor 3.

  • a

    Significant P value (P < .05).

Patients without adenomas3.043.9319.882.384.935.775.622.641.24127.7657.0716.741.64
Patients with adenomas1.203.5414.042.253.634.224.771.390.9679.0250.7712.201.74
P (Student t test).104.663.119.760.357.015a.708.045a.193.023a.617.126.810
Table 3. P Values Derived From Student T Tests Comparing Patients With Adenomas Versus Patients Without Adenomas in Various Age Groups
Age Group, yPatients With Adenomas Versus Patients Without Adenomas: P (Student T Test)
APCCDH1ESR1HIN1HPP1MGMTNELL1p14p15RARβSSTTAC1TIMP3
  • APC indicates adenomatous polyposis coli; CDH1, cadherin 1, type 1, E-cadherin (epithelial); ESR1, estrogen receptor 1; HIN1, cytokine high in normal 1; HPP1, hyperplastic polyposis protein 1; MGMT, O-6-methylguanine-DNA methyltransferase; NELL1, neural epidermal growth factor-like 1; p14, splicing factor 3B, 14-kDa subunit; p15, cyclin-dependent kinase (CDK) inhibitor 2B (inhibits CDK4); RARβ, retinoic acid receptor beta; SST, somatostatin; TAC1, tachykinin, precursor 1; TIMP3, tissue inhibitor of metalloproteinase-metallopeptidase inhibitor 3.

  • a

    Trend toward significance.

  • b

    Significant difference.

All.0888a.7771.1163.3604.2390.0329b.2000.0651a.3806.0363b.2858.0914a.7886
≥55.1512.8763.9784.1712.0437b.1523.0437b.6734.0567a.3785.1446.7924
≥60.1676.9726.0539a.9417.2282.1812.1469.0455b.6214.0576a.4399.1313.8348
≥65.2373.8454.0905a.8378.3176.0703a.2291.0812a.5189.0819a.9819.2897.9812
≥70.2230.8012.0698a.7682.2864.1110.2013.0574a.3596.0634a.9397.1926.9711
≥75.0478b.3200.0250b.9408.5764.2688.3619.1395.0390b.0014b.6895.0385b.5897
≥80.1454.3361.0023b.8408.1129.8191.1076.4519.2096.0272b.1570.0352b.9658
50-80.1656.5272.2183.2502.2475.0142b.1804.0367b.3209.1220.1073.1198.8147
55-80.21938651.1708.7768.2190.0341b.1753.0352b.6925.2366.1045.1980.9954
60-80.2550.9576.1748.8709.3140.1341.1951.0461b.7125.3018.1171.2289.9903
65-80.3741.8387.3443.8464.4662.0269b.3334.1012.4799.4376.2760.5014.6121
70-80.3672.8184.3165.9832.4460.0472b.3173.0853a.3454.4208.3079.4036.6756
75-80.0959a.4082.2446.7926.9112.0937a.5771.1561.0509a.0487b.4419.1561.3187
50-75.2380.5271.2478.2109.2201.0397b.1648.0438b.6598.2926.1702.1605.8202
55-75.3090.9602.1965.5283.1984.0942a.1639.0439b.7829.5446.1632.2843.9385
60-75.3773.6941.2623.6795.3221.4081.2140.0714a.6082.7843.1660.4231.8883
65-75.6075.6919.6683.1913.5446.1533.4180.2108.9286.8749.5690.9838.2540
70-75.6374.6748.7677.3682.5856.2600.4603.2296.8785.6956.6408.9258.3598
50-70.0289b.5666.5686.2382.0810a.0878a.1442.0648a.7086.1951.2057.1793.9281
55-70.0847a.8974.2810.7974.0229b.2029.1082.0435b.5982.4193.2022.3634.6871
60-70.0580a.6263.3032.8980.1329.9412.1483.1820.4884.5557.2411.4539.6747
65-70.8048.8656.7957.3251.6240.2909.6790.9816.2617.4399.3358.5651.4749
50-65.0343b.4749.7193.2247.1152.0940a.2450.0506a.5217.2340.2371.1278.8889
55-65.1082.9966.3982.8433.0343b.2131.1998.0298b.7959.5055.2374.2776.6659
60-65.0580a.8907.3480.8063.1728.9641.2324.1042.8803.7067.3312.2525.7573
50-60.1008.4766.7663.2884.2876.0631a.2309.1695.2627.2468.3417.2589.6394
55-60.3129.8171.2611.8006.1075.1581.1681.1040.9671.5403.3401.5573.3301
0-55.3416.2010.6624.3121.3899.2321.9533.9293.0740.4256.9885.4631.1443

Because our objective was to develop a clinically useful prediction model for the presence of concomitant adenomas, we also constructed receiver operating characteristic (ROC) curves. For the genes that differed most significantly (MGMT and RAR-β) between individuals with versus without adenomas, this method yielded an area under the ROC curve (AUC) of 0.607 for MGMT and 0.586 for RAR-β, and the 95% confidence intervals included 0.5 as the nondiscriminatory reference (data not shown). Patients with adenomas had lower methylation levels than individuals without adenomas (see Table 2). This unexpected result is discussed below (see Discussion).

Next, using linear discriminant analysis (LDA) and leave-1-out cross-validation, we identified the following contributory variables: age, APC, NELL1, p14, and the MI, which was composed of dichotomized values for APC, ESR1, HPP1, MGMT, p14, p15, RAR-β, and TAC1. To develop this dichotomized MI, an individual threshold was established based on ROC curves for each gene. LDA yielded a more accurate prediction model than any of the individual genes (AUC = 0.6661) (Fig. 1).

thumbnail image

Figure 1. This receiver operating characteristic curve was based on a linear discriminant analysis model that combined age, and neural epidermal growth factor-like 1 (NELL1), splicing factor 3B 14-kDa subunit (p14), and a methylation index that contained the genes adenomatous polyposis coli (APC), estrogen receptor 1 (ESR1), hyperplastic polyposis protein 1 (HPP1), O-6 methylguanine-DNA methyltransferase (MGMT), p14; cyclin-dependent kinase (CDK) inhibitor 2B (inhibits CDK4) (p15), retinoic acid receptor beta (RARβ), and tachykinin, precursor 1 (TAC1) that distinguished between patients with and without an adenoma at the time of rectal biopsy. Bestpred indicates best prediction.

Download figure to PowerPoint

thumbnail image

Figure 2. Average normalized methylation values (NMVs) are illustrated for the genes O-6 methylguanine-DNA methyltransferase (MGMT), retinoic acid receptor beta (RARβ), and somatostatin (SST). This chart demonstrates a trend toward lower methylation in the following sequence: nonsmokers (NS) without adenomas > normal rectal (NR) mucosa from smokers (SM)>NS with adenomas (P) (polyp nonsmoker)>PSM (logarithmic scale). Data were analyzed from 103 NR mucosae and 113 colorectal adenomas.

Download figure to PowerPoint

It is noteworthy that the methylation values were correlated inversely not only with the presence of an adenoma but also with smoking status. In fact, methylation values differed significantly between smokers and nonsmokers (Table 4). Moreover, for all genes that we studied except APC and NELL1, the average methylation values were lower in smokers than in nonsmokers (Table 4). On the basis of data from another ongoing study of adenomas per se (data not shown), the normalized methylation values (NMVs) for MGMT, RAR-β, and SST exhibited a gradual decrease in the following sequence: nonsmokers without adenomas > smokers without adenomas > nonsmokers with adenomas > smokers with adenomas.

Table 4. Mean Normalized Methylation Values of Smokers Versus Nonsmokers With P Values Determined Using the Student T Test
VariableMean Normalized Methylation Value, %
APCCDH1ESR1HIN1HPP1MGMTNELL1p14p15RARβSSTTAC1TIMP3
  • APC indicates adenomatous polyposis coli; CDH1, cadherin 1, type 1, E-cadherin (epithelial); ESR1, estrogen receptor 1; HIN1, cytokine high in normal 1; HPP1, hyperplastic polyposis protein 1; MGMT, O-6-methylguanine-DNA methyltransferase; NELL1, neural epidermal growth factor-like 1; p14, splicing factor 3B, 14-kDa subunit; p15, cyclin-dependent kinase (CDK) inhibitor 2B (inhibits CDK4); RARβ, retinoic acid receptor beta; SST, somatostatin; TAC1, tachykinin, precursor 1; TIMP3, tissue inhibitor of metalloproteinase-metallopeptidase inhibitor 3.

  • a

    Significant (P < .05).

Nonsmokers1.914.5821.882.615.255.514.492.491.66127.1169.3321.051.96
Smokers2.083.4014.792.203.844.715.401.760.8891.6447.9011.781.60
P (Student t test).104.663.119.760.357.015a.973.045a.193.023a.617.126.810

We also explored possible explanations for methylation alterations in smokers, particularly the relation between NMVs and the amount of previous smoking. We restricted this analysis to patients whose pack-year histories were fully known (35 of 87 smokers). Although no gene exhibited a strong correlation, most genes manifested an inverse correlation between NMVs and pack-years, suggesting a potential causal relation between smoking and reduced methylation levels. Pearson correlation coefficients for each of these correlations are shown in Table 5.

Table 5. Pearson Correlation Coefficients Between Pack-Years of Smoking and Normalized Methylation Values for Each Gene
GenePearson Correlation Coefficient
  1. APC indicates adenomatous polyposis coli; CDH1, cadherin 1, type 1, E-cadherin (epithelial); ESR1, estrogen receptor 1; HIN1, cytokine high in normal 1; HPP1, hyperplastic polyposis protein 1; MGMT, O-6-methylguanine-DNA methyltransferase; NELL1, neural epidermal growth factor-like 1; p14, splicing factor 3B, 14-kDa subunit; p15, cyclin-dependent kinase (CDK) inhibitor 2B (inhibits CDK4); RARβ, retinoic acid receptor beta; SST, somatostatin; TAC1, tachykinin, precursor 1; TIMP3, tissue inhibitor of metalloproteinase-metallopeptidase inhibitor 3.

APC−0.300878
CDH1−0.39658
ESR1−0.051456
HIN1−0.385364
HPP1−0.157576
MGMT−0.399695
NELL10.0587216
p14−0.044936
p15−0.339946
RARβ−0.250353
SST0.0253795
TAC1−0.103943
TIMP3−0.136025

Another interesting finding was a clear difference between smokers and nonsmokers in their odds of having an adenoma (Fig. 3). Smokers in all age groups exhibited a greater risk of adenomas.

thumbnail image

Figure 3. The risk of developing an adenoma versus age is illustrated in smokers and nonsmokers.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES

The current results reveal significant differences in specific gene methylation levels of the normal rectum between individuals with and without a concomitant adenoma. These differences were significant for most of the genes that we studied in several age groups and for several genes in the majority of the age groups studied (Table 3). These individual gene differences exhibited insufficient discriminating power to be clinically useful. Therefore, a corrected multivariate model was constructed based on age and the methylation values of 3 genes (NELL1, APC, and p14), which performed better than single genes alone (AUC = 0.6661). No association was observed between our panel of epigenetic and clinical markers and the size of adenomas, probably because of the generally small size of specimens.

The percentages of men and smokers, as mentioned above, were higher in our study group compared with the general population. Although sex cannot be considered a confounding factor (all of the genes that we studied are located on somatic chromosomes, whereas the incidence of colon polyps varies very little between the sexes for blacks and whites), the relatively equal numbers of smokers and nonsmokers allowed us to draw more powerful conclusions compared with our prediction model for polyps.

Traditionally, in the literature published over the past 2 decades, carcinogenesis has been associated with DNA hypermethylation.24, 25 However, early studies performed in the 1980s did focused on hypomethylation,26-28 and more recent work has re-examined this type of somatic epigenetic alteration.19, 29 It is noteworthy that, in the current study, methylation values were investigated as biomarkers to distinguish between 2 phenotypically normal cohorts of rectal tissues. Because these tissues were so similar phenotypically, we did not expect to find significant differences. However, we can only speculate that the effect observed in our study 1) may be a potential cumulative effect that, in time and space (and mostly because of metabolic factors), will add to significant epigenetic changes that are able to transform normal colorectal mucosa into a premalignant structure; or 2) by its shear force (demethylation increases chromosome instability), will lead to genetic changes able to affect the integrity of genetic information.

A field defect hypothesis was supported by our finding that methylation levels of several genes lay in an intermediate range between frank adenomatous tissues and NR mucosa in individuals who had colorectal adenomas elsewhere (Fig. 2). This difference did not prevail for all genes studied, but its mere occurrence argues that individuals with adenomas already may possess epigenetic changes in their normal mucosae. Environmental factors that exert effects throughout the colon, such as cigarette smoke, may have induced this large field of lowered methylation. Alternatively, these NR changes already may have been present at birth, suggesting some type of genetic predisposition, whereas smoking occurred later and was unrelated to the field defect. The discovered methylation differences, regardless of the fold-changes, were not accompanied by obvious phenotypic changes, because all tissues studied were virtually identical (ie, NR mucosae). Thus, a methylation field defect suggests that these differences constituted a predisposition toward a pathologic state (adenomas) rather than a cause of this state.

Because methylation levels were significantly lower in smokers than in nonsmokers and in individuals with versus without adenomas (Fig. 2), we formulated the following causality chain: smoking predisposes to diminished methylation of at least several genes, which, in turn, contributes to adenoma development. One possible mechanism underlying this event sequence is the known association between smoking and low folate levels because of the interference by cigarette smoke with folate use and/or metabolism,21 especially when considered in conjunction with the known association between low folate levels and decreased methylation because of the role of folate as a methyl donor in biochemical reactions (among others).22

Further strengthening of this predictive model may be achievable by increasing the number of model parameters (genes and clinical factors), whereas increasing the number of individuals studied also could improve its performance by decreasing the confidence interval. In addition, further analyses are indicated now to explore the relation between smoking and hypomethylation. For example, 1 unbiased strategy to consider for increasing the number of methylation parameters is CpG island microarray comparisons.

The current findings suggest that potential clinical application of this or similar models could benefit colorectal screening and surveillance algorithms. Specifically, indications for screening or surveillance colonoscopy could be stratified based on NR methylation alterations, thereby reducing the number of colonoscopies while increasing patient compliance.

In summary, we have identified a panel of genes that are predictive of the presence of concurrent colorectal adenomas based on altered methylation in the normal rectum. These genes, and possibly others, may be involved in a field defect that develops during adenomagenesis and, hence, deserve further evaluation in clinical biomarker validation studies. These epigenetic alterations may be caused, at least in part, by smoking and/or age. Indications for initial and follow-up colonoscopies may be restricted to individuals who have altered normal rectal mucosal methylation.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. CONFLICT OF INTEREST DISCLOSURES
  7. REFERENCES