In My Area Where Can I Get a Methylation Blood Trst
J Natl Cancer Inst. 2020 Jan; 112(1): 87–94.
Blood DNA Methylation and Breast Cancer: A Prospective Case-Cohort Analysis in the Sister Study
Zongli Xu
1 Epidemiology Branch, National Institute of Environmental Health Sciences, NIH, Research Triangle Park, NC
Dale P Sandler
1 Epidemiology Branch, National Institute of Environmental Health Sciences, NIH, Research Triangle Park, NC
Jack A Taylor
1 Epidemiology Branch, National Institute of Environmental Health Sciences, NIH, Research Triangle Park, NC
2 Epigenetics & Stem Cell Biology Laboratory, National Institute of Environmental Health Sciences, NIH, Research Triangle Park, NC
Received 2018 Dec 3; Revised 2019 Feb 14; Accepted 2019 Apr 9.
This article has been cited by other articles in PMC.
Abstract
Background
Peripheral blood DNA methylation may be associated with breast cancer, but studies of candidate genes and global and genome-wide DNA methylation have been inconsistent.
Methods
We performed an epigenome-wide study using Infinium HumanMethylation450 BeadChips with prospectively collected blood DNA samples from the Sister Study (1552 cases, 1224 subcohort). Differentially methylated cytosine-phosphate-guanine sites (dmCpGs) were identified using case-cohort proportional hazard models and replicated using deposited data from European Prospective Investigation into Cancer and Nutrition in Italy (EPIC-Italy) (n = 329). The correlation between methylation and time to diagnosis was examined using robust linear regression. Causal or consequential relationships of methylation to breast cancer were examined by Mendelian randomization using OncoArray 500 K single-nucleotide polymorphism data. All statistical tests were two-sided.
Results
We identified 9601 CpG markers associated with invasive breast cancer (false discovery rate = q < 0.01), with 510 meeting a strict Bonferroni correction threshold (10–7). A total of 2095 of these CpGs replicated in the independent EPIC-Italy dataset, including 144 meeting the Bonferroni threshold. Sister Study women who developed ductal carcinoma in situ had methylation similar to noncases. Most (1501, 71.6%) dmCpGs showed lower methylation in invasive cases. In case-only analysis, methylation was statistically significantly associated (false discovery rate = q < 0.05) with time to diagnosis for 892 (42.6%) of the dmCpGs. Analyses based on genetic association suggest that methylation differences are likely a consequence rather than a cause of breast cancer. Pathway analysis shows enrichment of breast cancer-related gene pathways, and dmCpGs are overrepresented in known breast cancer susceptibility genes.
Conclusions
Our findings suggest that the DNA methylation profile of blood starts to change in response to invasive breast cancer years before the tumor is clinically detected.
Identification of methylation changes in DNA from peripheral blood or serum and its association with breast cancer have been of great interest for early disease diagnosis and treatment (1–3). Several epigenome-wide studies have used prospectively collected samples with Illumina arrays or bisulfite sequencing to study the relationship between blood DNA methylation and breast cancer risk. Severi et al. reported that cases have increased DNA methylation within functional promoters across the genome and decreased DNA methylation in other regions and that the decreasing level outside promoters was inversely correlated with time since blood collection (4). van Veldhoven et al. (5) analyzed three independent datasets and reported that average methylation in gene bodies was inversely correlated with breast cancer risk in two of the three datasets and that the association was not affected by the time since blood draw. Using the Illumina 27 K array on a case-cohort study of 910 women from the Sister Study, we previously reported that of the markers associated with breast cancer, 75% had lower methylation in cases and that these differences were stronger for women diagnosed within 1 year of blood draw (6). Here, we report new results using the Illumina 450 K arrays and extend our analysis to 2776 prospectively collected blood samples from women in the Sister Study.
Methods
DNA Methylation in Sister Study Samples
The Sister Study is a prospective cohort study of 50 884 women recruited from the United States and Puerto Rico between 2003 and 2009 (7). To be eligible, women had to be age 35–75 years and could not have had breast cancer themselves, but must have had a biological sister with breast cancer. Participants provided extensive information at baseline interview. Written informed consent and blood samples were obtained during a home visit. Women are recontacted annually for information on breast cancer and other basic health information and every 2–3 years for more detailed follow-up, with approximately 95% response rates for annual updates. Women reporting breast cancer are contacted 6 months following diagnosis for additional information along with authorization to retrieve medical records. Among women for whom we obtained pathology reports, the positive predictive value of a self-reported breast cancer is 99.4% (8). The study was conducted in accordance with recognized ethical guidelines and approved by the institutional review boards of the National Institute of Environmental Health Sciences, National Institutes of Health (NIH), and the Copernicus Group. Our case-cohort design included a random sample of 1336 non-Hispanic white women (74 of whom were later diagnosed with breast cancer) drawn from the full cohort, and an additional 1542 non-Hispanic white women diagnosed with incident breast cancer during the time between enrollment and March 2015. The DNA methylation profile in blood was assessed using Illumina Human450 Methylation Arrays (see Supplementary Materials, available online). We randomly assigned case and noncase blood samples to plates and arrays. The proportions of cases were similar across plates as were distributions of follow-up time and age. We excluded 102 samples after quality control. Sister Study data may be requested through https://sisterstudy.niehs.nih.gov/English/coll-data.htm.
DNA Methylation Data from EPIC-Italy
DNA methylation raw idat files ({"type":"entrez-geo","attrs":{"text":"GSE51057","term_id":"51057"}}GSE51057) for the the European Prospective Investigation into Cancer and Nutrition in Italy (EPIC-Italy) nested case-control methylation study (9) were downloaded from the National Center for Biotechnology Information Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/) website. EPIC-Italy is a prospective study with blood samples collected at recruitment. The study used Infinium HumanMethylation450 BeadChip and includes 177 control subjects and 152 cases with time to diagnosis from 0.04 to 13.9 years. We applied the same preprocessing steps used in the Sister Study to this dataset.
Statistical Analysis
The association between breast cancer status and DNA methylation (M-value) for each cytosine-phosphate-guanine (CpG) site was tested in Sister Study data using a case-cohort proportional hazard model adjusted for cell type proportions, the top six surrogate variables (accounting for 95% of data variation) based on array nonnegative control-probes (10), and experimental plate (94 samples per plate). Age was treated as the primary time factor, with age at blood draw as the left truncation time and right censoring at age on September 2015. White blood cell type proportions were estimated with the Houseman method (11) using the R estimateCellCounts function in the minfi R package. Association between breast cancer status and DNA methylation in the EPIC-Italy dataset was tested using logistic regression with adjustment for age (in years). Association between time to diagnosis and DNA methylation level in Sister Study cases was evaluated using robust linear regression, adjusting for age, cell type proportions, the top six surrogate variables, and experiment plate. A false discovery rate (FDR) of q = 0.01 was used to assess statistical significance in differential methylation association analyses and q = 0.05 for time-to-diagnosis analyses. Bonferroni-corrected statistical significance thresholds were also calculated and reported for multiple testing with thresholds of P less than 10–7 for epigenome-wide analyses and P less than 10–8 for single-nucleotide polymorphisms (SNPs) genome-wide analyses. Correlation was evaluated using the Pearson method. All statistical tests were two-sided.
Epigenome-Wide Methylation Averages, Gene Pathway Analysis, and Functional Annotation
Mean DNA methylation for all 425 500 CpGs on the array was calculated for each woman, and the Kolmogorov-Smirnov test was used to assess case vs noncase distributions in overall mean methylation averages. CpGs on the array were mapped to nearby genes and used in gene pathway analysis with P values determined by permutation (see Supplementary Methods, available online). We used the binomial test to examine whether CpGs in 196 known breast cancer susceptibility genes (https://www.ebi.ac.uk/gwas/) were more likely to be differentially methylated than CpGs in other genes represented on the array. CpG islands, promoters, and other annotations provided by Illumina were used with χ2 tests to compare genomic context of CpGs on the array with that of differentially methylated (dm)CpGs. Similarly, ENCODE Chip-Seq data (12) were used with χ2 tests to compare histone modifications at CpG sites across the array with those at differentially methylated sites.
Genetic Association Analysis
Similar to Wahl et al. (13), we use the concept of Mendelian randomization to investigate potential causal and consequential relationships between DNA methylation and breast cancer. Genotypes for the Sister Study sample were generated using the Illumina Infinium OncoArray 500 K BeadChip as previously described (14). Genetic risk scores (GRS) based on 313 previously reported breast cancer susceptibility SNPs (15) were used for consequential analysis (see Supplementary Methods, available online).
Results
Sample Characteristics
The average age at enrollment for case subjects (n = 1552) was 58 years, whereas noncase subjects (n = 1224) had a mean age of 56 years. The average follow-up time for noncases was 6.7 years (range = 0.3–10 years), whereas the average time to diagnosis for cases was 3.8 years (range = 0.1–9.5 years). Women in the case group more frequently had a mother with a history of breast or other cancers (Table 1) but were similar to noncases for age at menarche, age at menopause, number of full-term pregnancies, and body mass index (BMI).
Table 1.
Sample characteristics*
| Characteristics | Noncases No. (%) | Cases No. (%) |
|---|---|---|
| Age at blood draw, y | ||
| 35–50 | 342 (27.9) | 279 (18.0) |
| 50–60 | 483 (39.5) | 617 (39.7) |
| 60–70 | 329 (26.9) | 526 (33.9) |
| 60–75 | 70 (5.7) | 130 (8.4) |
| Age of menarche, y | ||
| 8–11 | 90 (7.3) | 94 (6.0) |
| 11–15 | 1017 (83.2) | 1328 (85.7) |
| 15–19 | 116 (9.5) | 128 (8.3) |
| Missing | ||
| Age at menopause, y | ||
| <45 | 146 (17.3) | 124 (11.6) |
| 45–50 | 200 (23.8) | 252 (23.6) |
| 50–55 | 368 (43.7) | 477 (44.7) |
| >55 | 128 (15.2) | 215 (20.1) |
| Mother had breast cancer | ||
| Yes | 233 (19.5) | 392 (25.7) |
| No | 961 (80.5) | 1136 (74.3) |
| Mother had any cancer | ||
| Yes | 507 (42.5) | 804 (52.6) |
| No | 687 (57.5) | 724 (47.4) |
| No. of full term pregnancies | ||
| 0 | 248 (20.3) | 316 (20.4) |
| 1 | 181 (14.9) | 281 (18.2) |
| 2+ | 791 (64.8) | 951 (61.4) |
| BMI, kg/m2 | ||
| <18.5 | 10 (0.8) | 19 (1.2) |
| 18.5–25 | 481 (39.4) | 572 (36.9) |
| 25–30 | 394 (32.2) | 508 (32.7) |
| 30+ | 337 (27.6) | 453 (29.2) |
Breast Cancer-Associated CpGs
We initially evaluated associations between breast cancer status and methylation M-values in the full dataset of the 2776 Sister Study subjects, which included as "cases" both women who developed invasive breast cancer and women who developed ductal carcinoma in situ (DCIS). We identified 3452 dmCpGs at an FDR threshold q = 0.01; of these, 806 replicated at an FDR threshold of q = 0.01 in the EPIC-Italy cohort. For these 806 dmCpGs, the methylation differences between cases and noncases were largely driven by invasive cases, with DCIS cases having intermediate values (Figure 1). Within invasive cases, there was no evidence of ordering by tumor stage. Based on this finding, we excluded the 333 DCIS cases from all subsequent analyses reported below.
Averaged DNA methylation differences between breast cancer cases and noncases (n = 1224) by case clinical stage. Shown are 806 differentially methylated cytosine-phosphate-gnine sites (CpGs that are statistically significant in the Sister Study full dataset (including as cases women who developed either ductal carcinoma in situ or invasive tumors) and subsequently replicated in EPIC-Italy. CpGs are ordered by overall methylation difference between all cases and noncases. CpGs = cytosine-phosphate-gnine sites; DCIS = ductal carcinoma in situ.
After excluding women with DCIS, case-cohort analysis of 2443 women (1219 invasive cases and 1224 noncases) identified a much larger set of 9601 dmCpGs (FDR q < 0.01); 510 dmCpGs were statistically significant at a strict Bonferroni threshold of P less than 10–7 (Supplementary Table 1, available online). A total of 2095 of these dmCpGs were also statistically significant (FDR q < 0.01) and had concordant association direction in the EPIC-Italy study (Figure 2) and included 144 of 510 (28.2%) of the dmCpGs that met the strict Bonferroni threshold. A total of 72 of all replicated dmCpGs were statistically significant at a Bonferroni-corrected P value of 10–7 in the EPIC-Italy dataset. Note that CpG cg26203572 in gene LINC00525 on chromosome 7 has a much smaller association P value than all other CpGs on the array in the Sister Study (P = 2 × 10–33). This CpG is also highly statistically significant in the EPIC-Italy dataset (P = 4 × 10–8), but the association directions are discordant between the two studies. The log hazard ratio estimates for dmCpGs in the Sister Study were statistically significantly correlated with log hazard ratio estimates in EPIC-Italy (Pearson R 2 = 0.88, P < .001). Volcano plots depicting the relationship between the estimated effect sizes and association P values in the Sister Study dataset and EPIC-Italy (Supplementary Figure 1, A and B, available online) show that most of the replicated probes (1501, 71.6%) have lower methylation in cases.
Manhattan plot of P values from association analysis between DNA methylation and breast cancer status in Sister Study samples (1219 invasive cases and 1224 noncases). The 2095 cytosine-phosphate-gnine sites replicated in the EPIC-Italy dataset are marked in red. Bottom dashed line indicates the false discovery rate threshold of 0.01 and top dashed line indicates Bonferroni P value threshold of 1 × 10–7. CpG = cytosine-phosphate-gnine.
The goal of the main association analysis was to identify any CpG markers associated with breast cancer, and therefore we did not adjust for known breast cancer risk factors except for age. To explicitly examine the effect of these factors, we performed a sensitivity analysis, adjusting for BMI, alcohol consumption, smoking, and menopause status. Results with and without adjustments were very similar (Supplementary Figure 2, available online).
In case-only analysis, we examined whether methylation levels at the 2095 replicated dmCpGs were correlated with time to diagnosis (the time interval between blood draw and diagnosis of breast cancer). If altered blood methylation is a long-term marker of breast cancer susceptibility, we might expect methylation to be independent of time to diagnosis. Alternatively, if blood methylation is an early response to tumor, we might expect methylation to be correlated with time to diagnosis, becoming progressively divergent with shorter time to diagnosis. This latter pattern is true for 1955 (93.3%) of the 2095 replicated dmCpGs; 892 (42.6%) of the dmCpGs were statistically significantly associated with time to diagnosis at an FDR q threshold of 0.05 (Supplementary Table 1, available online). Figure 3 demonstrates the change in distributions of the dmCpGs by time to diagnosis.
Distributions of standardized methylation values for differentially methylated cytosine-phosphate-gnine sites (dmCpGs) that are statistically significant in the Sister Study noncases (n = 1224) and invasive cases (n = 1219) by time to diagnosis. For each dmCpG, methylation beta values were standardized across all women in the study, then median values were calculated for each dmCpG in noncases and in each case group. Violin plots show the median (white dot), interquartile range (black bar), and frequency (width) of standardized dmCpG values in different groups, with the regression line showing trend in distributions by time to diagnosis. A) Each violin plot represents the distribution of the 1501 dmCpGs that had lower methylation in cases compared with noncases. B) Each violin plot represents the distribution of the 594 dmCpGs that had higher methylation in cases.
Genetic Association Analysis
Methylation at 265 dmCpGs showed statistically significant associations (Bonferroni threshold P < 1 × 10–8) with nearby cis-SNPs. We examined the association of cis-SNPs with breast cancer and whether the predicted effects of cis-SNPs on breast cancer through methylation were consistent with the observed association. For these cis-SNPs, we found modest correlation between the predicted (through methylation) and observed effects on breast cancer (Figure 4A, Pearson R = 0.32, P < .001).
Genetic association analysis in Sister Study invasive cases (n = 1219) and noncases (n = 1224). A) Causality analysis investigating whether DNA methylation in blood influences BCa (BCa) risk. For 265 cis-single nucleotide polymorphisms (SNPs) that were statistically significantly associated with individual differentially methylated cytosine-phosphate-gnine sites (dmCpGs), we compare the predicted effect of an SNP on BCa via methylation βpred = βCpG∼SNP × βBCa∼CpG (x axis) and the directly observed effect of the SNP on BCa βBCa∼SNP (y axis) (Pearson R = 0.32, two-sided P < .001). B) Consequential analysis investigating whether dmCpG methylation in blood is the consequence of BCa. Genetic risk scores (GRS) based on BCa susceptibility SNPs were used to compare the predicted effect of GRS on dmCpG methylation through BCa βpred = βBCa∼GRS × βCpG∼BCa (x axis) and the directly observed effect βCpG∼GRS (y axis) (Pearson R = 0.70, two-sided P < .001). BCa = breast cancer; CpG = cytosine-phosphate-gnine; GRS = genetic risk score.
We also investigated whether blood DNA methylation was likely to be a consequence of breast cancer. Using published breast cancer-associated SNPs to construct GRS, we examined the association between GRS and methylation at the 2095 dmCpG sites. For GRS, we found a strong correlation between predicted effects (through breast cancer) and observed effects on dmCpG methylation (Figure 4B, Pearson R = 0.70, P < .001).
Epigenome-Wide DNA Methylation Averages
A few studies (4,5) have suggested that overall DNA methylation based on the average of all probes on the 450 K array is inversely associated with breast cancer risk, that is, that women with lower average methylation have higher risk. We calculated mean DNA methylation for all 425 500 CpGs in Sister Study subjects but found little evidence that the distributions of mean values differed between cases and noncases (Kolmogorov-Smirnov P = .5, Supplementary Figure 3, available online).
Pathway Analysis
Genes with dmCpGs are statistically significantly enriched for 21 cancer-related pathways (Supplementary Table 2, available online), including three of the four breast cancer-specific signaling pathways in the Ingenuity Pathway Analysis (IPA) database: Breast Cancer Regulation by Stathmin1, Hereditary Breast Cancer Signaling, and Estrogen-Dependent Breast Cancer Signaling. Enriched pathways also included 22 pathways related to cellular growth, proliferation, and development, three pathways related to cell cycle regulation, and 18 pathways related to cellular immune response.
Functional Annotation
Based on Illumina's annotation for CpGs on the array, the 594 dmCpGs that had higher methylation in cases are enriched in CpG islands (59.1% vs 31.6% on array, P < .001) and around gene promoter regions (18.3% vs 14.0% on array, P < .001). The 1501 dmCpGs that had lower methylation in cases are enriched in CpG non-island regions (51.2% vs 35.4% on array, P < .001), gene bodies (41.0% vs 37.0%, P < .001), and 3'untranslated region regions (9.0% vs 3.9%, P < .001).
Based on ENCODE Chip-Seq data (12), we found that the dmCpGs with higher methylation in cases coincided with H3K4me3 (64.5% vs 35.7% on array, P < .001) and H3K9ac (55.9% vs 31.1% on array, P < .001) histone marks that are often found in the promoter regions of actively transcribed genes, and with H3K27ac (52.0% vs 28.6% on array, P < .001), a mark often found at active enhancers. In contrast, dmCpGs with lower methylation in cases were much more likely than expected to coincide with H3K36me3 (18.1% vs 9.1%, P < .001), a mark found in the gene body of actively transcribed genes. Similar relationships held for these histone marks in other cell lines including the MCF-7 breast cancer mammary gland adenocarcinoma cell line, where sites with higher methylation in cases were enriched with H3K4me3 (16.0% vs 8.1%, P < .001) and H3K27ac (14.0% vs 8.9% on array, P < .001); sites with lower methylation in cases were enriched with H3K36me3 (5.9% vs 3.3% on array, P < .001).
CpGs in Genes Known to be Related to Breast Cancer
We found 42 dmCpGs were located near breast cancer susceptibility genes (Supplementary Table 3, available online), which is statistical significantly more than expected (binomial test, P = .007) and includes ATM, ESR1, FGFR2, PTEN, and MAP3K1.
Discussion
We used prospectively collected blood DNA samples from a large cohort to identify CpGs whose methylation differs between women who subsequently developed invasive breast cancer and women who did not. Although our initial case-cohort analysis included women with DCIS as cases, we found that methylation values in women with DCIS were similar to noncases. Excluding women with DCIS resulted in smaller P values and allowed the identification of a much larger set of dmCpGs associated with the development of invasive cancers, 2095 of which were replicated in an independent dataset from the EPIC-Italy study. Among women who developed invasive cancer, there was no evidence of further stratification of methylation levels by tumor stage.
Almost three-quarters of the replicated dmCpGs had lower methylation in cases compared with noncases. Based on this result, one might anticipate that relative to noncases, cases might also have lower array-wide average methylation, a finding that has been reported in other studies (4,5). However, like the findings in Ambatipudi et al. (16), we found no difference between cases and noncases for array-wide average methylation. We note that CpGs on the Illumina 450 K array are not a random sample of CpGs across the genome but are instead mainly targeted to known gene regions, which account for only a small percentage of the whole genome. Thus, our analysis result on array-wide average methylation does not exclude the possibility that genome-wide methylation differences might still exist between cases and noncases.
Our previous study (6), in a much smaller set of Sister Study women using the Illumina 27 K array, identified 250 CpGs associated with breast cancer risk at an FDR threshold of 0.05. A total of 228 of these CpGs are present on the Illumina 450 K array. Among these, 26 CpGs were statistically significant at an FDR threshold of 0.05 in the current study. Six CpGs reached FDR q less than 0.01 and five (cg03430067, cg03616357, cg07072643, cg08287471, and cg19709625) were replicated in the EPIC-Italy study dataset.
Methylation levels at dmCpGs were correlated with time between blood draw and cancer diagnosis; divergence from noncase levels was greatest among women with shorter time to diagnosis. This case-only time-dependent response is statistically independent of the case-cohort analysis. Based on this finding, we suggest that blood dmCpGs are less likely to be long-term breast cancer predisposition risk markers and more likely to reflect a response to tumors in the years preceding clinical diagnosis. This interpretation is also supported by results of genetic association analysis. Although causality analysis showed association supporting methylation as a cause of breast cancer, the results of consequential analysis were even stronger and consistent with methylation as a consequence of breast cancer.
One possible explanation for the time-to-diagnosis effects we observed in case-only analysis is that dmCpGs effects might be the result of circulating tumor DNA. To examine this hypothesis, we used the Cancer Genome Atlas Program (TCGA) breast tumor methylation data to identify 2000 CpGs with the largest differences between breast tumor and adjacent normal tissue, but only 17 of these tumor-associated CpGs were among the 2095 dmCpGs identified in our case-cohort analysis. We also identified the 2000 CpGs with the largest methylation differences between TCGA breast tumors and Sister Study non-case blood samples, but only nine of these tumor-associated CpGs were among the 2095 dmCpGs identified in our case-cohort analysis. Finally, studies of circulating cell-free DNA in cancer patients have reported a total of 59 differentially methylated genes (17–21), but only eight of these genes had dmCpGs in our analysis. Although there may be circulating tumor DNA present in cases, we believe such DNA is unlikely to be driving our observed case-cohort differences in DNA methylation.
We found that the dmCpGs with higher methylation in cases are most often in CpG islands near the promoter region of genes and coincide with active histone marks; such increased promoter methylation would be consistent with decreased expression. dmCpGs with lower methylation in cases are more likely to be located in gene bodies and are enriched for H3K36me3; a similar histone enrichment pattern in breast cancer cell lines was also reported in a previous study (22). The H3K36me3 mark is often found in actively transcribed genes; such decreased gene body methylation would also be consistent with decreased expression (23).
We used gene pathway analysis based on the IPA database to better characterize the set of genes containing dmCpGs. We found that genes with dmCpGs are enriched for 21 of the 40 canonical cancer pathways in the IPA database. Of the four breast cancer-specific pathways in IPA, we found that three are statistically significantly enriched. In a separate analysis, we explicitly examined the set of genes reported from Genome-Wide Association Studies to be associated with breast cancer and found that more than one-third contain a dmCpG, considerably more than expected by chance.
All participants in our study and the larger Sister Study from which they were drawn have a sister with breast cancer, giving them about a 2-fold increased risk of developing breast cancer (7, 24). This increased risk provides an increase in power that is important in a prospective cohort study of breast cancer (25), but, like any volunteer cohort, participants are not entirely representative of the US population, and the methylation risks that we describe here may be higher or lower in other populations. Although we have previously reported good agreement between CpG methylation measured by Illumina arrays and more precise methylation measured by pyrosequencing (26, 27), in the current study we rely on replication from the smaller EPIC-Italy study and do not provide further validation by pyrosequencing. There are some limits to relying on the EPIC data. Whereas the Sister Study used a case-cohort design allowing it to describe the hazard of breast cancer per time-unit of age, EPIC-Italy used a nested case-control design with odds ratios as the measure of risk. We expect methylation to correlate with both measures of risk but cannot exclude the possibility that the different study designs may affect the detection of some methylation marks. Further, the small size of the EPIC-Italy sample may limit its value as a replication set.
Together, our findings support the hypothesis that epigenetic differences in cancer-related genes are detectable in blood DNA from women before clinical diagnosis of invasive breast cancer. These differences do not appear to be due to circulating tumor DNA, are seen only for invasive cases, and are correlated with time to diagnosis. We suggest changes in blood DNA methylation may reflect an early response to still clinically occult tumors.
Funding
This work was supported by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences (Z01 ES049033, Z01 ES049032, Z01 ES044005).
Notes
The funders had no role in the design of the study; the collection, analysis, and interpretation of the data; the writing of the manuscript; and the decision to submit the manuscript for publication. The authors have no conflicts of interest to disclose.
Supplementary Material
djz065_Supplementary_Data
References
1. Tang Q, Cheng J, Cao X, et al. Blood-based DNA methylation as biomarker for breast cancer: a systematic review. Clin Epigenetics. 2016;8(1):115. [PMC free article] [PubMed] [Google Scholar]
2. Chopra-Tandon N, Wu H, Arcaro KF, et al. Relationships between global DNA methylation in circulating white blood cells and breast cancer risk factors. J Cancer Epidemiol. 2017;2017(1):2705860.. [PMC free article] [PubMed] [Google Scholar]
3. Basse C, Arock M.. The increasing roles of epigenetics in breast cancer: implications for pathogenicity, biomarkers, prevention and treatment. Int J Cancer. 2015;137(12):2785–2794. [PubMed] [Google Scholar]
4. Severi G, Southey MC, English DR, et al. Epigenome-wide methylation in DNA from peripheral blood as a marker of risk for breast cancer. Breast Cancer Res Treat. 2014;148(3):665–673. [PubMed] [Google Scholar]
5. van Veldhoven K, Polidoro S, Baglietto L, et al. Epigenome-wide association study reveals decreased average methylation levels years before breast cancer diagnosis. Clin Epigenetics. 2015;7(1):67.. [PMC free article] [PubMed] [Google Scholar]
6. Xu ZL, Bolick SCE, DeRoo LA, et al. Epigenome-wide association study of breast cancer using prospectively collected sister study samples. J Natl Cancer Inst. 2013;105(10):694–700. [PMC free article] [PubMed] [Google Scholar]
7. Sandler DP, Hodgson ME, Deming-Halverson SL, et al. The sister study cohort: baseline methods and participant characteristics. Environ Health Perspect. 2017;125(12):127003.. [PMC free article] [PubMed] [Google Scholar]
8. D'Aloisio AA, Nichols HB, Hodgson ME, et al. Validity of self-reported breast cancer characteristics in a nationwide cohort of women with a family history of breast cancer. BMC Cancer. 2017;17(1):692.. [PMC free article] [PubMed] [Google Scholar]
9. Demetriou CA, Chen J, Polidoro S, et al. Methylome analysis and epigenetic changes associated with menarcheal age. PLoS One. 2013;8(11):e79391.. [PMC free article] [PubMed] [Google Scholar]
10. Xu Z, Niu L, Li L, et al. ENmix: a novel background correction method for Illumina HumanMethylation450 BeadChip. Nucleic Acids Res. 2016;44(3):e20.. [PMC free article] [PubMed] [Google Scholar]
11. Houseman EA, Accomando WP, Koestler DC, et al. DNA methylation arrays as surrogate measures of cell mixture distribution. BMC Bioinformatics. 2012;13(1):86. [PMC free article] [PubMed] [Google Scholar]
12. Consortium EP. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489(7414):57–74. [PMC free article] [PubMed] [Google Scholar]
13. Wahl S, Drong A, Lehne B, et al. Epigenome-wide association study of body mass index, and the adverse outcomes of adiposity. Nature. 2017;541(7635):81.. [PMC free article] [PubMed] [Google Scholar]
14. Michailidou K, Lindstrom S, Dennis J, et al. Association analysis identifies 65 new breast cancer risk loci. Nature. 2017;551(7678):92–94. [PMC free article] [PubMed] [Google Scholar]
15. Mavaddat N, Michailidou K, Dennis J, et al. Polygenic risk scores for prediction of breast cancer and breast cancer subtypes. Am J Hum Genet. 2019;104(1):21–34. [PMC free article] [PubMed] [Google Scholar]
16. Ambatipudi S, Horvath S, Perrier F, et al. DNA methylome analysis identifies accelerated epigenetic ageing associated with postmenopausal breast cancer susceptibility. Eur J Cancer. 2017;75(7):299–307. [PMC free article] [PubMed] [Google Scholar]
17. Yang R, Stocker S, Schott S, et al. The association between breast cancer and S100P methylation in peripheral blood by multicenter case-control studies. Carcinogenesis. 2017;38(3):312–320. [PubMed] [Google Scholar]
18. Yang R, Pfutze K, Zucknick M, et al. DNA methylation array analyses identified breast cancer-associated HYAL2 methylation in peripheral blood. Int J Cancer. 2015;136(8):1845–1855. [PubMed] [Google Scholar]
19. Cheuk IWY, Shin VY, Kwong A.. Detection of methylated circulating DNA as noninvasive biomarkers for breast cancer diagnosis. J Breast Cancer. 2017;20(1):12–19. [PMC free article] [PubMed] [Google Scholar]
20. Uehiro N, Sato F, Pu F, et al. Circulating cell-free DNA-based epigenetic assay can detect early breast cancer. Breast Cancer Res. 2016;18(1):129. [PMC free article] [PubMed] [Google Scholar]
21. Legendre C, Gooden GC, Johnson K, et al. Whole-genome bisulfite sequencing of cell-free DNA identifies signature associated with metastatic breast cancer. Clin Epigenetics. 2015;7(1):100. [PMC free article] [PubMed] [Google Scholar]
22. Rivenbark AG, Coleman WB, Stahl BD.. Histone methylation patterns in human breast cancer. FASEB J. 2009;23(1_supplement):1. [Google Scholar]
23. Spies N, Nielsen CB, Padgett RA, et al. Biased chromatin signatures around polyadenylation sites and exons. Mol Cell. 2009;36(2):245–254. [PMC free article] [PubMed] [Google Scholar]
24. Collaborative Group on Hormonal Factors in Breast C. Familial breast cancer: collaborative reanalysis of individual data from 52 epidemiological studies including 58, 209 women with breast cancer and 101, 986 women without the disease. Lancet. 2001;358(9291):1389–1399. [PubMed] [Google Scholar]
25. Weinberg CR, Shore DL, Umbach DM, et al. Using risk-based sampling to enrich cohorts for endpoints, genes, and exposures. Am J Epidemiol. 2007;166(4):447–455. [PMC free article] [PubMed] [Google Scholar]
26. Harlid S, Adgent M, Jefferson WN, et al. Soy formula and epigenetic modifications: analysis of vaginal epithelial cells from infant girls in the IFED study. Environ Health Perspect. 2017;125(3):447–452. [PMC free article] [PubMed] [Google Scholar]
27. Harlid S, Xu Z, Panduri V, et al. CpG sites associated with cigarette smoking: analysis of epigenome-wide data from the Sister Study. Environ Health Perspect. 2014;122(7):673–678. [PMC free article] [PubMed] [Google Scholar]
In My Area Where Can I Get a Methylation Blood Trst
Source: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7489106/
0 Response to "In My Area Where Can I Get a Methylation Blood Trst"
Post a Comment