Next Article in Journal
Guiding Efficient, Effective, and Patient-Oriented Electrolyte Replacement in Critical Care: An Artificial Intelligence Reinforcement Learning Approach
Next Article in Special Issue
Immune Cell Networks Uncover Candidate Biomarkers of Melanoma Immunotherapy Response
Previous Article in Journal
Women Caring for Husbands Living with Parkinson’s Disease: A Phenomenological Study Protocol
Previous Article in Special Issue
A Clinical Decision Support System for the Prediction of Quality of Life in ALS
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
JPM
Volume 12
Issue 5
10.3390/jpm12050660
jpm-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Obesity-Associated Differentially Methylated Regions in Colon Cancer

by
John J. Milner
1,*,
Zhao-Feng Chen
2,
James Grayson
3 and
Shyang-Yun Pamela Koong Shiao
4
1
College of Nursing, Augusta University, Augusta, GA 30912, USA
2
Bachelor Degree Program in Pet Healthcare, Yuanpei University of Medical Technology, Hsinchu 30015, Taiwan, China
3
Hull College of Business, Augusta University, Augusta, GA 30912, USA
4
Medical College of Georgia, Augusta University, Augusta, GA 30912, USA
*
Author to whom correspondence should be addressed.
J. Pers. Med. 2022, 12(5), 660; https://doi.org/10.3390/jpm12050660
Submission received: 4 March 2022 / Revised: 11 April 2022 / Accepted: 18 April 2022 / Published: 20 April 2022
(This article belongs to the Special Issue Personalized Medicine with Biomedical and Health Informatics)
Browse Figures
Versions Notes

Abstract

:
Obesity with adiposity is a common disorder in modern days, influenced by environmental factors such as eating and lifestyle habits and affecting the epigenetics of adipose-based gene regulations and metabolic pathways in colorectal cancer (CRC). We compared epigenetic changes of differentially methylated regions (DMR) of genes in colon tissues of 225 colon cancer cases (154 non-obese and 71 obese) and 15 healthy non-obese controls by accessing The Cancer Genome Atlas (TCGA) data. We applied machine-learning-based analytics including generalized regression (GR) as a confirmatory validation model to identify the factors that could contribute to DMRs impacting colon cancer to enhance prediction accuracy. We found that age was a significant predictor in obese cancer patients, both alone (p = 0.003) and interacting with hypomethylated DMRs of ZBTB46, a tumor suppressor gene (p = 0.008). DMRs of three additional genes: HIST1H3I (p = 0.001), an oncogene with a hypomethylated DMR in the promoter region; SRGAP2C (p = 0.006), a tumor suppressor gene with a hypermethylated DMR in the promoter region; and NFATC4 (p = 0.006), an adipocyte differentiating oncogene with a hypermethylated DMR in an intron region, are also significant predictors of cancer in obese patients, independent of age. The genes affected by these DMR could be potential novel biomarkers of colon cancer in obese patients for cancer prevention and progression.

1. Introduction

Obesity, or excess adiposity and having a body mass index (BMI) ≥ 30, is a preventable and treatable condition, with lifestyle and environmental modifications, which has tripled since 1975 and now affects approximately 13% of the world population [1]. Obesity has been linked to tumorigenesis and cancer progression in various organs and a reduction in life expectancy by up to 14 years [2,3,4,5,6,7]. Mortality is increased up to 10% when obesity is present with colorectal cancer (CRC), whereas the risk of CRC is reduced up to 21% with a decrease in BMI and changes in lifestyle [8,9].
The risk of developing CRC increases with age, and 50–80% of CRC can be attributed to epigenetic changes due to lifestyle and BMI [10,11,12,13,14], with increased BMI as a risk factor in the proliferation of CRC [15,16,17,18,19]. CRC is the second most common cause of cancer death in the United States, with approximately 53,000 deaths occurring in 2021 [20]. For personalized medicine, it is clinically imperative to understand the impact of obesity on epigenetic changes to prevent the progression of CRC.
DNA methylation (DNAm) is an epigenetic regulation of gene function that is implicated in the formation of CRC [21,22,23,24] and is impacted by obesity [25,26,27,28] and age [29,30,31,32]. DNAm occurs when a methyl group attaches to a cytosine nucleotide, usually within a CpG dinucleotide (CpG), inhibiting the expression of the nucleotide, thereby potentially altering the expression of the gene itself [33]. Differentially methylated regions (DMRs) represent groups of methylated cytosines within close range in various tissue types and developmental stages [34]. DMRs located in the promoter region (first exon of a gene) were linked to gene silencing [35,36,37,38,39], and the position of DMRs related to the transcription start site (TSS) could impact transcription and gene function [40].
Hypomethylation (a reduction in methylated CpG) was associated with increased expression of gene function, whereas hypermethylation with decreased expression may potentially lead to chronic diseases including cancer, especially if methylation occurs in the promoter region of a gene [37,41,42,43,44,45]. Hypermethylated tumor suppressor genes (TSG) could lead to dysfunctional gene division or apoptosis, leading to abnormal cell growth; whereas hypomethylation with oncogenes causes cells to divide abnormally faster [46]. Hypermethylation and reduction in TSG function of SRGAP2C (Slit-Robo Rho GTPase-activating protein 2C) [47,48,49] and ZBTB46 (zinc finger and born-to-bind domain containing 46) [50,51,52] were associated with tumorigenesis and cancer metastasis. Conversely, hypomethylation and upregulated oncogene functions of HIST1H3I (Histone linker 1 with Histone H3.1), HIST1H3D (Histone linker 1 with Histone H3.D) [53,54,55,56], NFATC4 (Nuclear factor of activated T-cells cytoplasmic 4) [57,58,59,60] and HOXB8 (Homeobox B8) [61,62,63] were associated with adiposity and colon cancer.
Significant research is being conducted on methylation of CpG and DMR in colon cancer [64,65,66], yet there is little consensus on what constitutes a significant methylation threshold that could potentially translate to clinical significance. Whether the methylation threshold is purely statistical using p values, or if it is a differential change measured in a percent difference, has not yet been sufficiently documented or validated. Many studies focus on single-gene associations with methylated CpG or DMR, some taking clinical data into context [67,68]. The genomic region of DMR and the impact on gene expression has also been studied, showing that DMRs on promoter regions adversely affect gene function [38,39]. With advanced sequencing technology and machine-learning-based analytics [69,70,71], we conducted this study to examine DMRs in association with obesity as a significant contributing factor for colon cancer prevention.
The United States Centers for Disease Control and Prevention (CDC) has established a need for increasing precision in cancer prevention [72]. Precision medicine takes individual differences in lifestyle, environment, and biology into account, requiring complex interactive analysis and predictive analytics, as well as standardized coding [73,74,75]. Therefore, we accessed data from the Cancer Genome Atlas (TCGA) to evaluate the association of obesity in human colon tissue to locate DMR-associated genes of interest to examine the associations of obesity with colon cancer [76]. We then applied groundbreaking machine-learning-based predictive analysis to locate DMRs, integrating BMI and age into the validation models, to enhance the accuracy of prediction.

2. Materials and Methods

2.1. Demographic and Methylation Data

We obtained methylation data files from TCGA Colon Cancer project (COAD) version 1.23.0 (https://portal.gdc.cancer.gov, accessed 16 June 2018) that were filtered to include harmonized Illumina 450 K methylation data, BMI, age and gender from normal colon (n = 15) and colon cancer (n = 225) tissues using an R package designed for data retrieval, grouping and DMR analysis, TCGAbiolinksGUI [77,78,79,80]. The Illumina 450 K methylation array provided data on 485,000 CpG sites, which covered approximately 1.6% of all CpG sites, (0.01% of the entire genome), and methylation information on 99% of all known genes [81,82,83,84]. Data included three groups of non-obese control (no cancer), non-obese cancer (BMI < 30) and obese cancer (BMI ≥ 30). At the time of data retrieval, this comprised the entire list of cases that met the inclusion criteria. BMI was used as a grouping independent variable and age as an independent variable in the regression model.

2.2. DMR Bioinformatics Analysis

CpG site analysis and DMR identification were completed using additional R packages limma (v. 3.34.9, February 2018) and bumphunter (v. 1.20.0, November 2012). Limma involves a matrix-type schema to analyze intra-sample variability (n = 240) per individual CpG site (n ≈ 485,000), and Bayesian corrected p values were provided between groups [85]. Bumphunter includes linear regression and permutation testing to determine clusters of DMR with significant CpG sites [86,87]. DMR with >2 CpG sites, excluding sex chromosomes and having ≥5% proportional change between groups of obese and non-obese cancer were annotated to protein-coding genes. A 3–10% difference in DNAm level between groups was noted as significant [88,89,90,91].
Gene annotation was conducted using data from the Catalogue of Somatic Mutations in Cancer (COSMIC) v86 August 2018 (https://cancer.sanger.ac.uk/cosmic (accessed on 16 June 2018)) [92,93] and the University of California, Santa Cruz (UCSC) genome browser (GRCh38/hg38, December 2013) (www.genome.ucsc.edu (accessed on 16 June 2018)) [94]. Gene ontology and pathway analysis were conducted using the Database for Annotation, Visualization, and Integrated Discovery (DAVID) v 6.8 (https://david.ncifcrf.gov (accessed on 16 June 2018)) [95]. To isolate genes related to obesity, only genes in both obese and non-obese comparison groups, and only those that could be further annotated to cancer and/or adipose-related functions through ontology or pathway analysis, were considered.

2.3. Predictive Analysis

Generalized regression (GR) was performed using JMP Pro v. 14 (SAS Institute, Cary, NC, USA). as a machine-learning tool to determine a predictive model from the genes identified in the obese and non-obese cancer groups. To create the predictive model for GR, variables were recoded into dichotomous values based on median distribution across the variable. The model created a prediction profile for associations between the parameters of interest and the strength of the parameters within the predictive model. Unlike logistic regression (LR), which assumed that all variables share a linear association, GR performed analysis on each independent variable to determine associations with the dependent variable and created a model that applied the nonlinear association to each variable in the prediction [96,97,98], which was then compared to the (LR) model for validation [99,100].
The final GR model was derived based on several criteria, including Akaike information criterion (AICc), misclassification rate and the area under the receiver operating characteristic (ROC) curve (AUC). For internal validity of the predictive model, an AUC as close to 1 (100%) is desired. Sensitivity refers to the number of actual cases with the finding of a positive result, whereas specificity is the number of actual cases without the finding of a negative result. AUC is a method to plot the sensitivity and specificity of test results to determine the accuracy of true positives versus false negatives [101]. Misclassification rate is precision of the model by calculating the number of errors by the total number of observations. Ideally, this number should be low. AICc is an estimate of the fitness of the model and should also be a low number.
To establish a GR model, 10 genes with the highest differential methylation, both hypomethylated and hypermethylated, were analyzed using interaction profilers and regression algorithms. The prediction validation was developed using a GR adaptive elastic net and Leave One Out (LOO) methodology with an 85% training/15% validation proportion created for machine-learning-based iterations. Interactions between factors were examined using interaction profiler plots between parameters of genes or with age, with steps of iterations to eliminate parameters from the model that had no significance or altered the significance of other parameters and prediction accuracy. Elastic net models presented a higher sensitivity and specificity than lasso [102,103,104], and LOO methodology was used to eliminate insignificant parameters in the model and is suitable for analysis with smaller sample sizes [105,106].

3. Results

Demographic factors among control and two cancer groups of obese and non-obese were compared (Table 1). BMI was different between obese cancer and two other groups (p < 0.0001), and cancer groups were younger in age (p < 0.05), with obese cancer being 9.7 years younger and non-obese cancer being 7.7 years younger than the control group on average. There were no differences among three groups on gender and racial distributions.

3.1. Significant DMRs and Associated Genes between Groups

DMR analysis was performed to determine the number of protein-coding genes of significance between groups, using a 5%, 10% and 15% methylation change between groups (Table 2), which shows a complete list of DMRs with the highest methylation differences for three between-groups pairs. DMR coordinates and gene functions are provided in the supplementary tables (Tables S1 (hypomethylated) and S2 (hypermethylated). To test the hypothetical association between obesity and CRC, gene ontology was performed with a list of 518 genes comprising a 5% methylation change in both hypermethylated (n = 178) and hypomethylated (n = 340) genes between the obese and non-obese cancer groups. A 5% methylation difference was used due to the need for a sufficient list of genes for ontological analysis between the obese and non-obese cancer groups. No novel pathways with statistical significance were discovered between the obese and non-obese cancer groups; therefore, further ontological analysis was not conducted.
Table 3 shows 10 DMRs with the highest hypomethylation difference between obese and non-obese cancer groups, and Table 4 shows the 10 DMRs with the highest hypermethylation between the two groups. Genes with functions linked to adiposity or glucose metabolism, cancer-related functions, and both adiposity and cancer-related functions were noted. Genes noted in Supplemental Tables S1 (hypomethylated) and S2 (hypermethylated) were used to derive the final GR model.

3.2. Significant Predictors

The most significant predictors associated with obesity between the two cancer groups were age ≥ 76 (p = 0.004); HIST1H3I, a hypomethylated oncogene (p = 0.002); NFATC4, a hypermethylated oncogene (p = 0.027); SRGAP2C, a hypermethylated tumor suppression gene (p = 0.025); and ZBTB46, a hypomethylated TSG interacting with age (p = 0.024), which was the only gene to have significant interaction with age in our prediction model. Variable importance analysis using independent uniform inputs shows that the order of variable importance is age (total effect (TE): 0.374), HIST1H3I (TE: 0.299), ZBTB46 (TE: 0.295), SRGAP2C (TE: 0.184), NFATC4 (TE: 0.156), HOXB8 (TE: 0.098) and HIST1H3D (TE: 0.024). Genes and interactions left in the model without significance were left to protect the integrity of the model itself, as the removal of these predictors caused model instability (see Supplemental Figure S1).
HIST1H3I is an oncogene located on chromosome 6 that encodes a nuclear protein responsible for nucleosome structure and histone modification. It has been shown to have a high affinity for tumorigenesis, has been identified as a potential biomarker for CRC, and has been isolated in adipocytes [53,54,107]. Our data (see Figure 1a) show the promoter region DMR of HIST1H3I to have 17% hypomethylation between obese and non-obese cancer, and GR analysis showed no interaction between HIST1H3I and age in the prediction model. Mean values appear to increase from control group due to outliers in the data samples, but GR is based on median values, which are not represented by outliers.
TSG such as ZBTB46, a zinc finger/BTB domain protein gene on chromosome 20, represses the oncogene PRDM1 and has similar functions to autoimmune regulators [50,51,52]. Although located on an intron, the DMR affecting ZBTB46 showed a 36% reduction in methylation between obese and non-obese cancer groups (see Figure 1b), and GR analysis showed significant interaction with age (p = 0.024).
SRGAP2C is a SLIT-ROBO GTPase-activating tumor-suppressing gene located on chromosome 1. Changes in expression may contribute to cancer metastasis, and the SRGAP2 protein is reduced or absent in many tumor samples [47,48,49]. The promoter-region DMR on SRGAP2C showed significant hypermethylation between all three groups (see Figure 1c), with the lowest at 14% proportional change between the obese and non-obese cancer group, and it revealed a marginal interaction with age in the GR model.
NFATC4, an oncogene located on chromosome 14, encodes a protein from the nuclear factor of an activated T-cell family, which is a DNA-binding complex, is expressed in many cancer tissues, and has been shown to enhance tumorigenesis. With obesity, NFATC4 is known to initiate inflammatory processes, and it is associated with increased cell death in older patients [57,58,59,60]. The significant DMR for NFATC4, located on an intron, had 15% hypermethylation between obese and non-obese cancer (see Figure 1d), and NFATC4 had a marginal interaction with age in the GR model. A large number of outliers appears to minimize the mean value differential for NFATC4, but these outliers are not factored into the GR model.
HOXB8 located on chromosome 17 is a known oncogene that is associated with colorectal cancer. It is downregulated in colon cancer, but downregulation has been associated with favorable prognosis in renal cancer [61,62,63,108]. HOXB8 has promoter-region DMR with a 20% hypomethylation between groups (see Figure 1e), and a marginal interaction with age in the prediction model.
HIST1H3D, similar to HIST1H3I, is a known oncogene also located on chromosome 6, functioning as a chromatin compactor. It is upregulated in cancer, and a reduction in its expression causes chromatin structure closing [55,56,109]. Figure 1f shows the HIST1H3D promoter-region DMR having a 14% hypomethylation between obese and non-obese cancer, with no significant age interaction in the GR model.
The prediction model shown in Table 5 presents comparable misclassification rates, with 29% for validation, 27% for LOO and 29% in the LR models, with equal precisions. AICc in the validation model was 71 compared to 76 in the LR model, revealing a fitter validation model. Figure 2 shows that the AUC for LR (a) was 74%, revealing similar internal validity to our GR model with 74% AUC for validation (b) and 76% for LOO (c). GR models provide higher quality predictions than LR models in locating possible obesity-associated colon cancer biomarkers.

4. Discussion

Generalized regression is a powerful machine learning tool to capture associations between variables, rather than assuming that one causes the other. GR eliminates variables that have no effect on the final model, allowing the remaining variables to have a greater effect on the overall model. Using GR, we reduced the pool of colon cancer obesity-impacted DMR-associated genes to six (6) and determined that age was an associated variable when considered independently, also when interacting with several genes, most notably ZBTB46 (p = 0.024). We further determined that neither gender nor ethnicity were significant factors, and this is one of only a few studies to use GR in a methylation study that also used a differential methylation value of DMR.
Of the list of top 10 genes containing the DMR with the largest methylation difference, HOXB8 was the only gene with a statistically significant hypomethylated DMR in the obese to non-obese group (p = 0.026). As an oncogene, increased expression is associated with colorectal cancer. Two genes had statistically significant DMR hypermethylation in the obese to non-obese group: ZNF426 (p = 0.049), a zinc finger protein-coding gene involved in transcriptional regulation; and TUBB3 (p = 0.043), a beta-tubulin protein family coding gene that plays a role in axon guidance. When used in the GR model, only HOXB8 had borderline significance, which caused model instability when it was removed.
One limitation of this study was the use of methylation data from solid tumor tissue, rendering it difficult to generalize for biomarker analysis; however, it provides significant information about the genomic mechanisms impacted by obesity that may be targeted by precision medicine for colon cancer patients. Further study needs to be conducted to compare methylation changes in both solid tissue and body fluid samples, to determine whether DNAm occurs systemically or is isolated to the cancer tissue, and further study needs to be conducted to determine the level that DNAm impacts the function of the gene, whether at an individual CpG site or a DMR, such that future studies can all start from a leveled analysis point. The use of a single cohort cancer database is another limitation of this study, which will be reduced in a follow-up study by using multiple cohorts as well as survival analysis to validate these findings.
Using machine-learning-based tools and grouping colon cancer cases based on BMI, we have identified genes associated with obesity, related to lifestyle that may modified, potentially impacting colon cancer through methylation. HIST1H3I, ZBTB46, SRGAP2C and HIST1H3D are all potential novel biomarkers identified through our analysis method, and using GR, multiple genes impacted by DMR can be identified with cofactors from patient lifestyle. DNAm analysis and interpretation are becoming easier and less expensive to perform and can provide insights into disease processes never considered feasible during treatment processes in the past. Combining DNAm with GR provides precision-medicine-based healthcare the tools necessary to focus on patient-centered treatment for cancer and chronic diseases.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/jpm12050660/s1, Figure S1: Interaction profile matrix for generalized regression model parameters, Table S1: Genes associated with top hypomethylated differentially methylated regions (DMR) between groups, Table S2: Genes associated with top hypermethylated differentially methylated regions (DMR) between groups. Original TCGA methylation files: demographics.csv, annotations.csv, methylation_betas.csv.

Author Contributions

Conceptualization and study design, J.J.M. and S.-Y.P.K.S.; acquisition and search of the literature, J.J.M., Z.-F.C. and S.-Y.P.K.S.; analysis and interpretation of data, J.J.M., J.G. and S.-Y.P.K.S.; writing—initial draft, J.J.M. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

The IRB of Augusta University has determined that the research conducted here is not human subjects research, on 22 August 2021 (Project 1790087-1).

Informed Consent Statement

Not applicable.

Acknowledgments

Shaoyong Su from the Georgia Prevention Institute for his support in methylation analysis. The results published here are in whole or part based upon data generated by the TCGA Research Network: https://www.cancer.gov/tcga (accessed on 16 June 2018).

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. WHO. Obesity and Overweight. Available online: http://www.who.int/en/news-room/fact-sheets/detail/obesity-and-overweight (accessed on 3 January 2021).
  2. Almen, M.S.; Nilsson, E.K.; Jacobsson, J.A.; Kalnina, I.; Klovins, J.; Fredriksson, R.; Schioth, H.B. Genome-wide analysis reveals DNA methylation markers that vary with both age and obesity. Gene 2014, 548, 61–67. [Google Scholar] [CrossRef] [PubMed]
  3. Dick, K.J.; Nelson, C.P.; Tsaprouni, L.; Sandling, J.K.; Aïssi, D.; Wahl, S.; Meduri, E.; Morange, P.-E.; Gagnon, F.; Grallert, H.; et al. DNA methylation and body-mass index: A genome-wide analysis. Lancet 2014, 383, 1990–1998. [Google Scholar] [CrossRef] [Green Version]
  4. Printz, C. Extreme obesity may shorten life expectancy up to 14 years. Cancer 2014, 120, 3591. [Google Scholar] [CrossRef] [PubMed]
  5. Shiao, S.P.K.; Grayson, J.; Yu, C.H.; Wasek, B.; Bottiglieri, T. Gene Environment Interactions and Predictors of Colorectal Cancer in Family-Based, Multi-Ethnic Groups. J. Pers. Med. 2018, 8, 10. [Google Scholar] [CrossRef] [Green Version]
  6. Jones, P.A.; Baylin, S.B. The Epigenomics of Cancer. Cell 2007, 128, 683–692. [Google Scholar] [CrossRef] [Green Version]
  7. Ayers, D.; Boughanem, H.; Macías-González, M. Epigenetic Influences in the Obesity/Colorectal Cancer Axis: A Novel Theragnostic Avenue. J. Oncol. 2019, 2019, 7406078. [Google Scholar] [CrossRef] [Green Version]
  8. Luo, J.; Hendryx, M.; Manson, J.E.; Figueiredo, J.C.; LeBlanc, E.S.; Barrington, W.; Rohan, T.E.; Howard, B.V.; Reding, K.; Ho, G.Y.; et al. Intentional Weight Loss and Obesity-Related Cancer Risk. JNCI Cancer Spectr. 2019, 3, pkz054. [Google Scholar] [CrossRef]
  9. Wu, S.; Liu, J.; Wang, X.; Li, M.; Gan, Y.; Tang, Y. Association of obesity and overweight with overall survival in colorectal cancer patients: A meta-analysis of 29 studies. Cancer Causes Control 2014, 25, 1489–1502. [Google Scholar] [CrossRef]
  10. Gao, Y.; Cao, Y.; Tan, A.; Liao, C.; Mo, Z.; Gao, F. Glutathione S-transferase M1 polymorphism and sporadic colorectal cancer risk: An updating meta-analysis and HuGE review of 36 case-control studies. Ann. Epidemiol. 2010, 20, 108–121. [Google Scholar] [CrossRef]
  11. Society, A.C. Cancer Facts and Figures 2022; American Cancer Society: Atlanta, GA, USA, 2022; Volume 500822. [Google Scholar]
  12. Yu, M.; Hazelton, W.D.; Luebeck, G.E.; Grady, W.M. Epigenetic aging: More than just a clock when it comes to cancer. Cancer Res. 2019, 80, 367–374. [Google Scholar] [CrossRef] [Green Version]
  13. Zheng, C.; Li, L.; Xu, R. Association of Epigenetic Clock with Consensus Molecular Subtypes and Overall Survival of Colorectal Cancer. Cancer Epidemiol. Biomark. Prev. 2019, 28, 1720–1724. [Google Scholar] [CrossRef] [PubMed]
  14. Islami, F.; Goding Sauer, A.; Miller, K.D.; Siegel, R.L.; Fedewa, S.A.; Jacobs, E.J.; Mc Cullough, M.L.; Patel, A.V.; Ma, J.; Soerjomataram, I.; et al. Proportion and number of cancer cases and deaths attributable to potentially modifiable risk factors in the United States. CA Cancer J. Clin. 2018, 68, 31–54. [Google Scholar] [CrossRef] [PubMed]
  15. Andreasson, A.; Hagstrom, H.; Skoldberg, F.; Onnerhag, K.; Carlsson, A.C.; Schmidt, P.T.; Forsberg, A.M. The prediction of colorectal cancer using anthropometric measures: A Swedish population-based cohort study with 22 years of follow-up. United Eur. Gastroenterol. J. 2019, 7, 1250–1260. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  16. Bjorge, T.; Haggstrom, C.; Ghaderi, S.; Nagel, G.; Manjer, J.; Tretli, S.; Ulmer, H.; Harlid, S.; Rosendahl, A.H.; Lang, A.; et al. BMI and weight changes and risk of obesity-related cancers: A pooled European cohort study. Int. J. Epidemiol. 2019, 48, 1872–1885. [Google Scholar] [CrossRef] [PubMed]
  17. Garcia, H.; Song, M. Early-life obesity and adulthood colorectal cancer risk: A meta-analysis. Rev. Panam. De Salud Publica Pan Am. J. Public Health 2019, 43, e3. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  18. Naber, S.K.; Kundu, S.; Kuntz, K.M.; Dotson, W.D.; Williams, M.S.; Zauber, A.G.; Calonge, N.; Zallen, D.T.; Ganiats, T.G.; Webber, E.M.; et al. Cost-effectiveness of risk-stratified colorectal cancer screening based on polygenic risk—current status and future potential. JNCI Cancer Spectr. 2019, 4, pkz086. [Google Scholar] [CrossRef]
  19. Anania, G.; Resta, G.; Marino, S.; Fabbri, N.; Scagliarini, L.; Marchitelli, I.; Fiorica, F.; Cavallesco, G. Treatment of Colorectal Cancer: A Multidisciplinary Approach. J. Gastrointest. Cancer 2018, 50, 458–468. [Google Scholar] [CrossRef]
  20. Siegel, R.L.; Miller, K.D.; Goding Sauer, A.; Fedewa, S.A.; Butterly, L.F.; Anderson, J.C.; Cercek, A.; Smith, R.A.; Jemal, A. Colorectal cancer statistics, 2020. CA Cancer J. Clin. 2020, 70, 145–164. [Google Scholar] [CrossRef] [Green Version]
  21. Ahn, J.B.; Chung, W.B.; Maeda, O.; Shin, S.J.; Kim, H.S.; Chung, H.C.; Kim, N.K.; Issa, J.P. DNA methylation predicts recurrence from resected stage III proximal colon cancer. Cancer 2011, 117, 1847–1854. [Google Scholar] [CrossRef] [Green Version]
  22. Beaulieu, J.-F. (Ed.) Colorectal Cancer: Methods and Protocols; Humana Press: New York, NY, USA, 2018; Volume 1765, p. 348. [Google Scholar]
  23. Berger, S.L.; Kouzarides, T.; Shiekhattar, R.; Shilatifard, A. An operational definition of epigenetics. Genes Dev. 2009, 23, 781–783. [Google Scholar] [CrossRef] [Green Version]
  24. Borchiellini, M.; Ummarino, S.; Di Ruscio, A. The Bright and Dark Side of DNA Methylation: A Matter of Balance. Cells 2019, 8, 1243. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  25. Ajonijebu, D.C.; Abboussi, O.; Russell, V.A.; Mabandla, M.V.; Daniels, W.M.U. Epigenetics: A link between addiction and social environment. Cell. Mol. Life Sci. CMLS 2017, 74, 2735–2747. [Google Scholar] [CrossRef] [PubMed]
  26. Andersen, V.; Holst, R.; Vogel, U. Systematic review: Diet-gene interactions and the risk of colorectal cancer. Aliment Pharm. 2013, 37, 383–391. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  27. Crujeiras, A.B.; Morcillo, S.; Diaz-Lagares, A.; Sandoval, J.; Castellano-Castillo, D.; Torres, E.; Hervas, D.; Moran, S.; Esteller, M.; Macias-Gonzalez, M.; et al. Identification of an episignature of human colorectal cancer associated with obesity by genome-wide DNA methylation analysis. Int. J. Obes. 2019, 43, 176–188. [Google Scholar] [CrossRef] [PubMed]
  28. Carr, P.R.; Amitay, E.L.; Jansen, L.; Alwers, E.; Roth, W.; Herpel, E.; Kloor, M.; Schneider, M.; Blaker, H.; Chang-Claude, J.; et al. Association of BMI and major molecular pathological markers of colorectal cancer in men and women. Am. J. Clin. Nutr. 2020, 111, 562–569. [Google Scholar] [CrossRef] [PubMed]
  29. Ehrlich, M. DNA hypermethylation in disease: Mechanisms and clinical relevance. Epigenetics 2019, 14, 1141–1163. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  30. Zhou, M.; Zhang, L.; Yang, Q.; Yan, C.; Jiang, P.; Lan, Y.; Wang, J.; Tang, R.; He, M.; Lei, G.; et al. Age-related gene expression and DNA methylation changes in rhesus macaque. Genomics 2020, 112, 5147–5156. [Google Scholar] [CrossRef]
  31. Tharakan, R.; Ubaida-Mohien, C.; Moore, A.Z.; Hernandez, D.; Tanaka, T.; Ferrucci, L. Blood DNA Methylation and Aging: A Cross-Sectional Analysis and Longitudinal Validation in the InCHIANTI Study. J. Gerontol. A Biol. Sci. Med. Sci. 2020, 75, 2051–2055. [Google Scholar] [CrossRef] [Green Version]
  32. Harris, C.J.; Davis, B.A.; Zweig, J.A.; Nevonen, K.A.; Quinn, J.F.; Carbone, L.; Gray, N.E. Age-Associated DNA Methylation Patterns Are Shared Between the Hippocampus and Peripheral Blood Cells. Front. Genet. 2020, 11, 111. [Google Scholar] [CrossRef] [Green Version]
  33. Cheng, X.; Hashimoto, H.; Horton, J.R.; Zhang, X. Chapter 2—Mechanisms of DNA Methylation, Methyl-CpG Recognition, and Demethylation in Mammals. In Handbook of Epigenetics; Tollefsbol, T., Ed.; Academic Press: San Diego, CA, USA, 2011; pp. 9–24. [Google Scholar] [CrossRef]
  34. Neidhart, M. DNA Methylation and Complex Human Disease; Academic Press: Oxford, MA, USA, 2016. [Google Scholar] [CrossRef]
  35. Brenet, F.; Moh, M.; Funk, P.; Feierstein, E.; Viale, A.; Socci, N.; Scandura, J. DNA Methylation of the First Exon Is Tightly Linked to Transcriptional Silencing. PLoS ONE 2011, 6, e14524. [Google Scholar] [CrossRef]
  36. Dorman, J.S.; Schmella, M.J.; Wesmiller, S.W. Primer in Genetics and Genomics, Article 1: DNA, Genes, and Chromosomes. Biol. Res. Nurs. 2016, 19, 7–17. [Google Scholar] [CrossRef] [PubMed]
  37. Du, X.; Han, L.; Guo, A.Y.; Zhao, Z. Features of methylation and gene expression in the promoter-associated CpG islands using human methylome data. Comp. Funct. Genom. 2012, 2012, 598987. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  38. Poduval, D.B.; Ognedal, E.; Sichmanova, Z.; Valen, E.; Iversen, G.T.; Minsaas, L.; Lønning, P.E.; Knappskog, S. Assessment of tumor suppressor promoter methylation in healthy individuals. Clin. Epigenet. 2020, 12, 131. [Google Scholar] [CrossRef] [PubMed]
  39. Quan, Y.; Liang, F.; Wu, D.; Yao, X.; Hu, Z.; Zhu, Y.; Chen, Y.; Wu, A.; Tang, D.; Huang, B.; et al. Blood Cell DNA Methylation of Aging-Related Ubiquitination Gene DZIP3 Can Predict the Onset of Early Stage Colorectal Cancer. Front. Oncol. 2020, 10, 544330. [Google Scholar] [CrossRef]
  40. Jones, P.A.; Takai, D. The role of DNA methylation in mammalian epigenetics. Science 2001, 293, 1068–1070. [Google Scholar] [CrossRef]
  41. Ng, J.M.; Yu, J. Promoter hypermethylation of tumour suppressor genes as potential biomarkers in colorectal cancer. Int. J. Mol. Sci. 2015, 16, 2472–2496. [Google Scholar] [CrossRef] [Green Version]
  42. Ross, J.P.; Rand, K.N.; Molloy, P.L. Hypomethylation of repeated DNA sequences in cancer. Epigenomics 2010, 2, 245–269. [Google Scholar] [CrossRef]
  43. Suzuki, M.M.; Bird, A. DNA methylation landscapes: Provocative insights from epigenomics. Nat. Rev. Genet. 2008, 9, 465–476. [Google Scholar] [CrossRef]
  44. Yin, A.; Etcheverry, A.; He, Y.; Aubry, M.; Barnholtz-Sloan, J.; Zhang, L.M.; Mao, X.; Chen, W.; Liu, B.; Zhang, W.; et al. Integrative analysis of novel hypomethylation and gene expression signatures in glioblastomas. Oncotarget 2017, 8, 89607–89619. [Google Scholar] [CrossRef] [Green Version]
  45. Costello, J.F.; Plass, C. Methylation Matters. J. Med. Genet. 2001, 38, 285–303. [Google Scholar] [CrossRef] [Green Version]
  46. Society, A.C. Oncogenes and Tumor Suppressor Genes. Available online: https://www.cancer.org/cancer/cancer-causes/genetics/genes-and-cancer/oncogenes-tumor-suppressor-genes.html (accessed on 3 January 2021).
  47. Charrier, C.; Joshi, K.; Coutinho-Budd, J.; Kim, J.E.; Lambert, N.; de Marchena, J.; Jin, W.L.; Vanderhaeghen, P.; Ghosh, A.; Sassa, T.; et al. Inhibition of SRGAP2 function by its human-specific paralogs induces neoteny during spine maturation. Cell 2012, 149, 923–935. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  48. NCBI. SRGAP2C: SLIT-ROBO Rho GTPase Activating Protein 2C [Homo sapiens]. Gene ID: 653464. Available online: https://www.ncbi.nlm.nih.gov/gene/653464 (accessed on 1 January 2020).
  49. Schmidt, E.R.E.; Kupferman, J.V.; Stackmann, M.; Polleux, F. The human-specific paralogs SRGAP2B and SRGAP2C differentially modulate SRGAP2A-dependent synaptic development. Sci. Rep. 2019, 9, 18692. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  50. Chen, W.Y.; Zeng, T.; Wen, Y.C.; Yeh, H.L.; Jiang, K.C.; Chen, W.H.; Zhang, Q.; Huang, J.; Liu, Y.N. Androgen deprivation-induced ZBTB46-PTGS1 signaling promotes neuroendocrine differentiation of prostate cancer. Cancer. Lett. 2019, 440, 35–46. [Google Scholar] [CrossRef] [PubMed]
  51. NCBI. ZBTB46: Zinc Finger and BTB Domain Containing 46 [Homo sapiens]. Gene ID: 140685. Available online: https://www.ncbi.nlm.nih.gov/gene/140685 (accessed on 1 January 2020).
  52. Satpathy, A.T.; Brown, R.A.; Gomulia, E.; Briseño, C.G.; Mumbach, M.R.; Pan, Z.; Murphy, K.M.; Natkunam, Y.; Chang, H.Y.; Kim, J. Expression of the transcription factor ZBTB46 distinguishes human histiocytic disorders of classical dendritic cell origin. Mod. Pathol. 2018, 31, 1479–1486. [Google Scholar] [CrossRef]
  53. Kim, J.; Daniel, J.; Espejo, A.; Lake, A.; Krishna, M.; Xia, L.; Zhang, Y.; Bedford, M.T. Tudor, MBT and chromo domains gauge the degree of lysine methylation. EMBO Rep. 2006, 7, 397–403. [Google Scholar] [CrossRef] [PubMed]
  54. NCBI. HIST1H3I: H3 Clustered Histone 11 [Homo sapiens]. Gene ID: 8354. Available online: https://www.ncbi.nlm.nih.gov/gene/8354 (accessed on 1 January 2020).
  55. Tagami, H.; Ray-Gallet, D.; Almouzni, G.; Nakatani, Y. Histone H3.1 and H3.3 complexes mediate nucleosome assembly pathways dependent or independent of DNA synthesis. Cell 2004, 116, 51–61. [Google Scholar] [CrossRef] [Green Version]
  56. NCBI. HIST1H3D: H3 Clustered Histone 4 [Homo sapiens]. Gene ID: 8351. Available online: https://www.ncbi.nlm.nih.gov/gene/8351 (accessed on 1 January 2020).
  57. Li, X.; Wang, W.; Wang, J.; Malovannaya, A.; Xi, Y.; Li, W.; Guerra, R.; Hawke, D.H.; Qin, J.; Chen, J. Proteomic analyses reveal distinct chromatin-associated and soluble transcription factor complexes. Mol. Syst. Biol. 2015, 11, 775. [Google Scholar] [CrossRef]
  58. NCBI. NFATC4: Nuclear Factor of Activated T Cells 4 [Homo sapiens]. Gene ID: 4776. Available online: https://www.ncbi.nlm.nih.gov/gene/4776 (accessed on 1 January 2020).
  59. Qin, X.; Wang, X.H.; Yang, Z.H.; Ding, L.H.; Xu, X.J.; Cheng, L.; Niu, C.; Sun, H.W.; Zhang, H.; Ye, Q.N. Repression of NFAT3 transcriptional activity by estrogen receptors. Cell. Mol. Life Sci. CMLS 2008, 65, 2752–2762. [Google Scholar] [CrossRef]
  60. Zhang, X.; Kang, T.; Zhang, L.; Tong, Y.; Ding, W.; Chen, S. NFATc3 mediates the sensitivity of gastric cancer cells to arsenic sulfide. Oncotarget 2017, 8, 52735–52745. [Google Scholar] [CrossRef] [Green Version]
  61. Ding, W.J.; Zhou, M.; Chen, M.M.; Qu, C.Y. HOXB8 promotes tumor metastasis and the epithelial-mesenchymal transition via ZEB2 targets in gastric cancer. J. Cancer Res. Clin. Oncol. 2017, 143, 385–397. [Google Scholar] [CrossRef]
  62. NCBI. HOXB8: Homeobox B8 [Homo sapiens]. Gene ID: 3218. Available online: https://www.ncbi.nlm.nih.gov/gene/3218 (accessed on 1 January 2020).
  63. Shen, S.; Pan, J.; Lu, X.; Chi, P. Role of miR-196 and its target gene HoxB8 in the development and proliferation of human colorectal cancer and the impact of neoadjuvant chemotherapy with FOLFOX4 on their expression. Oncol. Lett. 2016, 12, 4041–4047. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  64. Alzahrani, S.M.; Al Doghaither, H.A.; Al-Ghafari, A.B. General insight into cancer: An overview of colorectal cancer (Review). Mol. Clin. Oncol. 2021, 15, 271. [Google Scholar] [CrossRef] [PubMed]
  65. Gutierrez, A.; Demond, H.; Brebi, P.; Ili, C.G. Novel Methylation Biomarkers for Colorectal Cancer Prognosis. Biomolecules 2021, 11, 1722. [Google Scholar] [CrossRef] [PubMed]
  66. Sobhani, I. DNA Methylation Is a Main Key for Bacteria-Related Colon Carcinogenesis. Microorganisms 2021, 9, 2574. [Google Scholar] [CrossRef]
  67. Maugeri, A.; Barchitta, M.; Magnano San Lio, R.; Favara, G.; La Mastra, C.; La Rosa, M.C.; Agodi, A. The Relationship between Body Mass Index, Obesity, and LINE-1 Methylation: A Cross-Sectional Study on Women from Southern Italy. Dis. Markers 2021, 2021, 9910878. [Google Scholar] [CrossRef]
  68. Moreno-Ortiz, J.M.; Jiménez-García, J.; Gutiérrez-Angulo, M.; Ayala-Madrigal, M.L.; González-Mercado, A.; González-Villaseñor, C.O.; Flores-López, B.A.; Alvizo-Rodríguez, C.; Hernández-Sandoval, J.A.; Fernández-Galindo, M.A.; et al. High frequency of MLH1 promoter methylation mediated by gender and age in colorectal tumors from Mexican patients. Gac. Med. Mex. 2021, 157, 618–623. [Google Scholar] [CrossRef]
  69. Khodadadi, E.; Fahmideh, L.; Khodadadi, E.; Dao, S.; Yousefi, M.; Taghizadeh, S.; Asgharzadeh, M.; Yousefi, B.; Kafil, H.S. Current Advances in DNA Methylation Analysis Methods. BioMed. Res. Int. 2021, 2021, 8827516. [Google Scholar] [CrossRef]
  70. Baharudin, R.; Ishak, M.; Muhamad Yusof, A.; Saidin, S.; Syafruddin, S.E.; Wan Mohamad Nazarie, W.F.; Lee, L.H.; Ab Mutalib, N.S. Epigenome-Wide DNA Methylation Profiling in Colorectal Cancer and Normal Adjacent Colon Using Infinium Human Methylation 450K. Diagnostics 2022, 12, 198. [Google Scholar] [CrossRef]
  71. Paweł, K.; Maria Małgorzata, S. CpG Island Methylator Phenotype-A Hope for the Future or a Road to Nowhere? Int. J. Mol. Sci. 2022, 23, 830. [Google Scholar] [CrossRef]
  72. Khoury, M.J.; Bowen, M.S.; Clyne, M.; Dotson, W.D.; Gwinn, M.L.; Green, R.F.; Kolor, K.; Rodriguez, J.L.; Wulf, A.; Yu, W. From public health genomics to precision public health: A 20-year journey. Genet. Med. Off. J. Am. Coll. Med. Genet. 2018, 20, 574–582. [Google Scholar] [CrossRef] [Green Version]
  73. Khoury, M.J. No Shortcuts on the Long Road to Evidence-Based Genomic Medicine. JAMA 2017, 318, 27–28. [Google Scholar] [CrossRef] [PubMed]
  74. Khoury, M.J. Precision Medicine vs Preventive Medicine. JAMA 2019, 321, 406. [Google Scholar] [CrossRef] [PubMed]
  75. Khoury, M.J.; Engelgau, M.; Chambers, D.A.; Mensah, G.A. Beyond Public Health Genomics: Can Big Data and Predictive Analytics Deliver Precision Public Health? Public Health Genom. 2018, 21, 244–250. [Google Scholar] [CrossRef] [PubMed]
  76. Tomczak, K.; Czerwinska, P.; Wiznerowicz, M. The Cancer Genome Atlas (TCGA): An immeasurable source of knowledge. Contemp. Oncol. 2015, 19, A68–A77. [Google Scholar] [CrossRef]
  77. Colaprico, A.; Silva, T.C.; Olsen, C.; Garofano, L.; Cava, C.; Garolini, D.; Sabedot, T.S.; Malta, T.M.; Pagnotta, S.M.; Castiglioni, I.; et al. TCGAbiolinks: An R/Bioconductor package for integrative analysis of TCGA data. Nucleic Acids Res. 2016, 44, e71. [Google Scholar] [CrossRef]
  78. Liu, Y.; Qiu, P. Integrative analysis of methylation and gene expression data in TCGA. In Proceedings of the 2012 IEEE International Workshop on Genomic Signal Processing and Statistics (GENSIPS), Washington, DC, USA, 2–4 December 2012; pp. 1–4. [Google Scholar]
  79. Silva, T.C.; Colaprico, A.; Olsen, C.; D’Angelo, F.; Bontempi, G.; Ceccarelli, M.; Noushmehr, H. TCGA Workflow: Analyze cancer genomics and epigenomics data using Bioconductor packages. F1000 Res. 2016, 5, 1542. [Google Scholar] [CrossRef]
  80. Silva, T.C.; Colaprico, A.; Olsen, C.; Malta, T.M.; Bontempi, G.; Ceccarelli, M.; Berman, B.P.; Noushmehr, H. TCGAbiolinksGUI: A graphical user interface to analyze cancer molecular and clinical data. F1000 Res. 2018, 7, 439. [Google Scholar] [CrossRef] [Green Version]
  81. Babenko, V.N.; Chadaeva, I.V.; Orlov, Y.L. Genomic landscape of CpG rich elements in human. BMC Evol. Biol. 2017, 17, 19. [Google Scholar] [CrossRef] [Green Version]
  82. Chen, J.J.; Wang, A.Q.; Chen, Q.Q. DNA methylation assay for colorectal carcinoma. Cancer Biol. Med. 2017, 14, 42–49. [Google Scholar] [CrossRef] [Green Version]
  83. Morris, T.J.; Beck, S. Analysis pipelines and packages for Infinium HumanMethylation450 BeadChip (450k) data. Methods 2015, 72, 3–8. [Google Scholar] [CrossRef]
  84. Sandoval, J.; Heyn, H.A.; Moran, S.; Serra-Musach, J.; Pujana, M.A.; Bibikova, M.; Esteller, M. Validation of a DNA methylation microarray for 450,000 CpG sites in the human genome. Epigenetics 2011, 6, 692–702. [Google Scholar] [CrossRef]
  85. Ritchie, M.E.; Phipson, B.; Wu, D.; Hu, Y.; Law, C.W.; Shi, W.; Smyth, G.K. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic. Acids Res. 2015, 43, e47. [Google Scholar] [CrossRef]
  86. Jaffe, A.E.; Murakami, P.; Lee, H.; Leek, J.T.; Fallin, M.D.; Feinberg, A.P.; Irizarry, R.A. Bump hunting to identify differentially methylated regions in epigenetic epidemiology studies. Int. J. Epidemiol. 2012, 41, 200–209. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  87. Lent, S.; Xu, H.; Wang, L.; Wang, Z.; Sarnowski, C.; Hivert, M.F.; Dupuis, J. Comparison of novel and existing methods for detecting differentially methylated regions. BMC Genet. 2018, 19, 84. [Google Scholar] [CrossRef]
  88. Hibler, E.; Huang, L.; Andrade, J.; Spring, B. Impact of a diet and activity health promotion intervention on regional patterns of DNA methylation. Clin. Epigenet. 2019, 11, 133. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  89. Huang, G.; Cheng, W.; Xi, F. Integrated genomic and methylation profile analysis to identify candidate tumor marker genes in patients with colorectal cancer. Oncol. Lett. 2019, 18, 4503–4514. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  90. Slieker, R.; Bos, S.; Goeman, J.; Bovee, J.; Talens, R.; Breggen, R.; Suchiman, E.; Lameijer, E.; Putter, H.; van den Akker, E.; et al. Identification and systematic annotation of tissue-specific differentially methylated regions using Illumina 450k chips. Epigenet. Chromatin 2013, 6, 26. [Google Scholar] [CrossRef] [Green Version]
  91. Zhang, B.; Hong, X.; Ji, H.; Tang, W.Y.; Kimmel, M.; Ji, Y.; Pearson, C.; Zuckerman, B.; Surkan, P.J.; Wang, X. Maternal smoking during pregnancy and cord blood DNA methylation: New insight on sex differences and effect modification by maternal folate levels. Epigenetics 2018, 13, 505–518. [Google Scholar] [CrossRef]
  92. Bamford, S.; Dawson, E.; Forbes, S.; Clements, J.; Pettett, R.; Dogan, A.; Flanagan, A.; Teague, J.; Futreal, P.A.; Stratton, M.R.; et al. The COSMIC (Catalogue of Somatic Mutations in Cancer) database and website. Br. J. Cancer 2004, 91, 355–358. [Google Scholar] [CrossRef]
  93. Forbes, S.A.; Beare, D.; Boutselakis, H.; Bamford, S.; Bindal, N.; Tate, J.; Cole, C.G.; Ward, S.; Dawson, E.; Ponting, L.; et al. COSMIC: Somatic cancer genetics at high-resolution. Nucleic Acids Res. 2017, 45, D777–D783. [Google Scholar] [CrossRef]
  94. Kent, W.J.; Sugnet, C.W.; Furey, T.S.; Roskin, K.M.; Pringle, T.H.; Zahler, A.M.; Haussler, D. The human genome browser at UCSC. Genome Res. 2002, 12, 996–1006. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  95. Huang, D.W.; Sherman, B.T.; Lempicki, R.A. Bioinformatics enrichment tools: Paths toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 2009, 37, 1–13. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  96. Dasgupta, A.; Sun, Y.V.; König, I.R.; Bailey-Wilson, J.E.; Malley, J.D. Brief review of regression-based and machine learning methods in genetic epidemiology: The Genetic Analysis Workshop 17 experience. Genet. Epidemiol. 2011, 35, S5–S11. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  97. Grinberg, N.F.; Orhobor, O.I.; King, R.D. An evaluation of machine-learning for predicting phenotype: Studies in yeast, rice, and wheat. Mach Learn. 2020, 109, 251–277. [Google Scholar] [CrossRef] [Green Version]
  98. Song, L.; Langfelder, P.; Horvath, S. Random generalized linear model: A highly accurate and interpretable ensemble predictor. BMC Bioinform. 2013, 14, 5. [Google Scholar] [CrossRef] [Green Version]
  99. Afzali, M.H.; Sunderland, M.; Stewart, S.; Masse, B.; Seguin, J.; Newton, N.; Teesson, M.; Conrod, P. Machine-learning prediction of adolescent alcohol use: A cross-study, cross-cultural validation. Addiction 2019, 114, 662–671. [Google Scholar] [CrossRef] [PubMed]
  100. Ogutu, J.O.; Schulz-Streeck, T.; Piepho, H.P. Genomic selection using regularized linear regression models: Ridge regression, lasso, elastic net and their extensions. BMC Proc. 2012, 6, S10. [Google Scholar] [CrossRef] [Green Version]
  101. Burnham, K.P.; Anderson, D.R. Multimodel Inference:Understanding AIC and BIC in Model Selection. Sociol. Methods Res. 2004, 33, 261–304. [Google Scholar] [CrossRef]
  102. Chase, E.C.; Boonstra, P.S. Accounting for established predictors with the multistep elastic net. Stat. Med. 2019, 38, 4534–4544. [Google Scholar] [CrossRef]
  103. Waldmann, P.; Mészáros, G.; Gredler, B.; Fuerst, C.; Sölkner, J. Evaluation of the lasso and the elastic net in genome-wide association studies. Front. Genet. 2013, 4, 270. [Google Scholar] [CrossRef] [Green Version]
  104. Xiao, J.; Ding, R.; Xu, X.; Guan, H.; Feng, X.; Sun, T.; Zhu, S.; Ye, Z. Comparison and development of machine learning tools in the prediction of chronic kidney disease progression. J. Transl. Med. 2019, 17, 119. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  105. Cui, Z.; Gong, G. The effect of machine learning regression algorithms and sample size on individualized behavioral prediction with functional connectivity features. Neuroimage 2018, 178, 622–637. [Google Scholar] [CrossRef]
  106. Kirpich, A.; Ainsworth, E.A.; Wedow, J.M.; Newman, J.R.B.; Michailidis, G.; McIntyre, L.M. Variable selection in omics data: A practical evaluation of small sample sizes. PLoS ONE 2018, 13, e0197910. [Google Scholar] [CrossRef] [PubMed]
  107. Wen, H.; Li, Y.; Xi, Y.; Jiang, S.; Stratton, S.; Peng, D.; Tanaka, K.; Ren, Y.; Xia, Z.; Wu, J.; et al. ZMYND11 links histone H3.3K36me3 to transcription elongation and tumour suppression. Nature 2014, 508, 263–268. [Google Scholar] [CrossRef] [PubMed] [Green Version]
  108. Li, S.; Lu, X.; Chi, P.; Pan, J. Identification of HOXB8 and KLK11 expression levels as potential biomarkers to predict the effects of FOLFOX4 chemotherapy. Future Oncol. 2013, 9, 727–736. [Google Scholar] [CrossRef]
  109. Wang, Z.; Song, J.; Milne, T.A.; Wang, G.G.; Li, H.; Allis, C.D.; Patel, D.J. Pro isomerization in MLL1 PHD3-bromo cassette connects H3K4me readout to CyP33 and HDAC-mediated repression. Cell 2010, 141, 1183–1194. [Google Scholar] [CrossRef] [PubMed] [Green Version]
Jpm 12 00660 g001 550
Figure 1. Methylated genes from the generalized regression model. Notes. Genes with age interaction show mean values by age as trendlines; genes without interaction with age show overall mean trendline. Hypomethylation and hypermethylation percent indicates change between obese and non-obese groups. Genes are ordered (a–f) based on total effect from generalized regression model; ## indicates p < 0.05 for cancer group to control; ** indicates p < 0.05 between obese and non-obese cancer.
Figure 1. Methylated genes from the generalized regression model. Notes. Genes with age interaction show mean values by age as trendlines; genes without interaction with age show overall mean trendline. Hypomethylation and hypermethylation percent indicates change between obese and non-obese groups. Genes are ordered (a–f) based on total effect from generalized regression model; ## indicates p < 0.05 for cancer group to control; ** indicates p < 0.05 between obese and non-obese cancer.
Jpm 12 00660 g001
Jpm 12 00660 g002 550
Figure 2. Receiver operating characteristic (ROC) curve and area under the curve (AUC) for logistic and generalized regression with adaptive elastic net. (a) represents logistic regression, (b) is adaptive elastic net with validation column, and (c) is adaptive elastic net with leave-one-out validation.
Figure 2. Receiver operating characteristic (ROC) curve and area under the curve (AUC) for logistic and generalized regression with adaptive elastic net. (a) represents logistic regression, (b) is adaptive elastic net with validation column, and (c) is adaptive elastic net with leave-one-out validation.
Jpm 12 00660 g002
Table
Table 1. Demographic characteristics of the sample cases.
Table 1. Demographic characteristics of the sample cases.
Cancer
Mean ± SD
(Range)		Control
(n = 15)		Non-Obese
(n = 154)		Obese
(n = 71)
BMI, Kg/m225.5 ± 2.7
(20.1–29.8)		24.9 ± 3.1 *
(14.7–29.8)		36.1 ± 2.8 *
(30.0–54.1)
Age, years81.7 ± 13.7
(48–102)		74.0 ± 15.1 *
(41–106)		72.0 ± 12.0 *
(38–92)
Gender Female (%)9 (60)67 (43.5)37 (52)
Race White (%)13 (87)116 (75)49 (69)
Black (%)2 (13)30 (19)21 (30)
Other (%)08 (6)1 (1)
Notes. BMI: Body Mass Index; * indicates p < 0.05.
Table
Table 2. Summary of differentially methylated regions (DMR) with unique protein-coding genes per group comparison.
Table 2. Summary of differentially methylated regions (DMR) with unique protein-coding genes per group comparison.
GroupsNon-Obese Cancer/ControlObese Cancer/ControlObese Cancer/Non-Obese Cancer
Differential MethylationHyperHypoHyperHypoHyperHypo
5%4270374442034073178340
10%29671644287619092548
15%22486372173828610
Notes. Hypo refers to hypomethylated; Hyper refers to hypermethylated.
Table
Table 3. Genes associated with top hypomethylated differentially methylated regions (DMR) between obese and non-obese cancer groups.
Table 3. Genes associated with top hypomethylated differentially methylated regions (DMR) between obese and non-obese cancer groups.
DMRDis to TSSDNAmGene Function
Gene# CpGRegionNon-ObeseObese
HIST3H2A‡25Promoter8608.616.80DNA repair, MMR
HIST3H2BB‡25Promoter7018.616.80DNA repair, MMR
HOXB8†18Promoter−27914.4711.48Oncogene
HIST1H3I‡11Promoter−2417.4314.47Oncogene
TUBB2A‡3Intron−5939.938.26GTP binding
TMCO1‡13Promoter2105.494.59Calcium homeostasis
PRAC2†4Promoter10910.328.63Oncogene
AMOTL2‡4Intron−10,23515.7913.38Inhibits Wnt pathway
ARL4D^13Promoter1078.036.82Suppresses adipogenesis
HIST1H3D‡13Promoter5912.0210.26Oncogene
Notes: # CpG—number of methylated CpG sites; Dis to TSS—distance (in base pairs) to transcription start site from DMR start; DNAm—mean methylation percent; ^ represents genes that can be annotated to adiposity or glucose-related functions, † to cancer-related functions, and ‡ to both cancer and adipose/glucose-related functions.
Table
Table 4. Genes associated with top hypermethylated differentially methylated regions (DMR) between obese and non-obese cancer groups.
Table 4. Genes associated with top hypermethylated differentially methylated regions (DMR) between obese and non-obese cancer groups.
DMRDis to TSSDNAmGene Function
Gene# CpGRegionNon-ObeseObese
GNPDA2^12Promoter1077.869.74Protein metabolism
LSM14A†9Promoter5403.444.12Immune response
ZNF426†11Promoter10721.7025.80Transcription regulation
NFATC4‡8Intron−4669.9811.52Oncogene
ZNF852†3Promoter312.813.24Transcription regulation
FAM72B†9CDS302129.1933.34Oncogene
SRGAP2C‡9Promoter87929.1933.34Tumor Suppression Gene
TNFAIP2†3Promoter44514.2816.10Mediator of inflammation
ZNF747†7Promoter1887.188.07Transcription regulation
TUBB3‡3Intron24486.367.14Oncogene, immune response
Notes: # CpG—number of methylated CpG sites; CDS—coding DNA sequence; Dis to TSS—distance (in base pairs) to transcription start site from DMR start; DNAm—mean methylation percent; ^ represents genes that can be annotated to adiposity or glucose-related functions, † to cancer-related functions, and ‡ to both cancer and adipose/glucose-related functions.
Table
Table 5. Predictors of obesity-associated differentially methylated regions in colon cancer.
Table 5. Predictors of obesity-associated differentially methylated regions in colon cancer.
Generalized Regression Adaptive Elastic Net
Logistic RegressionLeave-One-OutValidation Column
ParametersEstimatep (χ2)Estimatep (χ2)Estimatep (χ2)
Intercept−0.19970.7380.01140.984−0.28690.613
Age (≥76)−2.42110.004−2.30760.003−2.29970.004
HIST1H3I (hypo, promoter) ‡1.15410.0031.20260.0011.12430.002
NFATC4 (hyper, intron) ‡−1.10340.046−1.32560.006−1.11330.027
SRGAP2C (hyper, promoter) ‡−1.28210.026−1.47790.006−1.23550.025
Age * ZBTB461.79930.0201.85450.0081.73430.024
Age * NFATC41.30510.0781.17520.0781.30650.069
Age * HOXB81.02520.1631.14100.0840.91610.081
Age * SRGAP2C1.05650.1531.15840.0881.00620.168
HIST1H3D (hypo, promoter) ‡−0.32250.401−0.61690.089−0.31190.419
ZBTB46 (hypo, intron) †−0.25510.637−0.28890.550−0.20360.702
HOXB8 (hypo, promoter) †−0.09840.8560.01150.9790.00001.000
Misclassification rate0.290-0.277-0.290-
AICc76.63---71.067-
Area under the curve0.741-0.757-0.739-
Notes—data not available; AICc—Akaike’s information criterion with corrections; * denotes interaction. † represents genes that can be annotated to cancer-related functions and ‡ to both cancer and adipose/glucose-related functions.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Milner, J.J.; Chen, Z.-F.; Grayson, J.; Shiao, S.-Y.P.K. Obesity-Associated Differentially Methylated Regions in Colon Cancer. J. Pers. Med. 2022, 12, 660. https://doi.org/10.3390/jpm12050660

AMA Style

Milner JJ, Chen Z-F, Grayson J, Shiao S-YPK. Obesity-Associated Differentially Methylated Regions in Colon Cancer. Journal of Personalized Medicine. 2022; 12(5):660. https://doi.org/10.3390/jpm12050660

Chicago/Turabian Style

Milner, John J., Zhao-Feng Chen, James Grayson, and Shyang-Yun Pamela Koong Shiao. 2022. "Obesity-Associated Differentially Methylated Regions in Colon Cancer" Journal of Personalized Medicine 12, no. 5: 660. https://doi.org/10.3390/jpm12050660

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop