Keywords
DNA Methylation, Cancer Biomarker, Epigenetics
This article is included in the Oncology gateway.
DNA Methylation, Cancer Biomarker, Epigenetics
Tumor cells have fundamentally different DNA methylation profile from normal cells of origin1–4. Some of these differences are tumor-specific, i.e. do not occur in any normal cell types, and thus could be used for tumor DNA identification. Since tumors shed DNA into bloodstream or other body fluids5–7, the detection of tumor-specific DNA methylation in these liquid biopsies could be utilized for non-invasive cancer diagnostics and monitoring8,9. This initiated a search for cancer-specific DNA methylation biomarker loci and analysis of these loci in plasma samples and other liquid biopsies10–12. We have previously described a large suite of cancer-specific DNA methylation biomarker loci discovered using TCGA and GEO data from over 10,000 tumor and normal samples13. Recently, we developed qPCR amplicons specific for a subset of these biomarker loci designed to detect common carcinoma types and tested them on clinical cfDNA samples from healthy individuals and non-small cell lung cancer (NSCLC) patients. We demonstrated that these biomarkers can distinguish between healthy subjects and NSCLC patients with high sensitivity and specificity14. Moreover, in blood samples from lung cancer patients the biomarker DNA methylation signal positively correlates with tumor size14. The purpose of the current study was to find how early during carcinogenesis the biomarker loci gain DNA methylation in order to assess their potential as detectors of early stage cancer. To this end, we analyzed DNA methylation of the biomarker loci in publically available data from several precancerous conditions. We found that the biomarker loci gain DNA methylation early in carcinogenesis since they are methylated already in majority precancerous lesions analyzed; in addition, where the data are available, the markers can distinguish lesions with malignant potential from those that stay benign.
The DNA methylation data from the Illumina HumanMethylation450 platform were downloaded from the GEO database (GEO accessions GSE60185, GSE66313, GSE53051, GSE58999, GSE48684, GSE77954, GSE72872, GSE81334, GSE108123 and GSE39279). These DNA methylation data are presented as beta values - numeric values in interval 0.0-1.0. For unmethylated CpGs the beta value approaches zero, for fully methylated CpGs beta approaches 1 and for CpGs methylated in a fraction of the sample 0<beta<1, e.g. a CpG methylated in 50% of the sample will have a beta value of approximately 0.5. All data were analyzed in the R programming environment, version 3.6.115 as follows: The beta values were normalized as described13. The normalized beta values for 10 biomarker CpGs (Table 1,14) were used in further analysis. Boxplots were created using the R function boxplot and the R library beeswarm,version 0.2.3. Multidimensional scaling (MDS) plots were constructed using the R function cmdscale on matrices of distances between samples and projected into two dimensions. The ability of the marker set to distinguish between progressive and regressive lung CIS was evaluated using receiver operating characteristic (ROC) analysis on the sums of the beta values from all 10 marker CpG Illumina probes (Table 1). The ROC analysis and AUC calculations were performed using the R library pROC16, version 1.15.3.
We have previously described a set of DNA methylation biomarker loci that are hypermethylated in 10 common carcinoma tumor types and we demonstrated that the level of DNA methylation of these loci can differentiate between plasma samples from lung cancer patients and healthy individuals. To determine the timing of the hypermethylation of these biomarker loci during human carcinogenesis and thus estimate potential of the markers to detect early disease stages we analyzed here the DNA methylation state of the biomarker loci in several premalignant conditions: breast ductal carcinoma in situ (DCIS), colorectal adenomas, Barrett’s esophagus (BE), pancreatic intraductal papillary mucinous neoplasms (IPMNs) and lung carcinoma in situ (CIS) using publically available Illumina HumanMethylation450 datasets from the GEO database.
Ductal carcinoma in situ is a precursor of invasive breast carcinoma (IBC). We analyzed DNA methylation of the biomarker loci in normal breast tissue samples, DCIS and IBC from three GEO datasets: GSE6018517, GSE6631318, GSE5305119. The results (Figure 1A) show that the biomarker loci are methylated already in DCIS at about the same level as in IBC. The multidimensional scaling (MDS) plot (Figure 1B) shows DCIS samples scattered among IBC samples, indicating comparable levels of DNA methylation of individual markers, while most of the normal samples form a small cluster on a side of the plot. Furthermore, there is no significant increase in the marker methylation during the progression to metastatic disease, as illustrated by data from a cohort (GSE5899920) of 44 pairs of primary breast tumors and lymph node metastases (Figure 1A).
Colorectal adenomas are the precursor neoplasms to colorectal cancer. We analyzed biomarker loci in normal colorectal tissue, colorectal adenomas, colorectal carcinomas and metastatic colorectal tumors from three GEO datasets: GSE4868421, GSE7795422, GSE5305119. Similar to DCIS, biomarker loci are already hypermethylated in colorectal adenomas with no further increase in methylation during the progression into invasive colorectal carcinomas or metastatic colorectal cancer (Figure 1C) and again colorectal adenomas on MDS plot are scattered among colorectal carcinomas (Figure 1D).
Barrett’s esophagus is a precancerous precursor of esophageal adenocarcinoma (EAC). We analyzed normal esophagus together with BE and EAC samples from two GEO datasets: GSE7287223, GSE8133424. Again, similar to DCIS or colorectal adenomas, biomarker loci are hypermethylated already in BE (Figure 1E, F). Similar situation was observed also in pancreatic intraductal papillary mucinous neoplasms (IPMNs), precursor lesions of pancreatic adenocarcinomas, where only one small dataset (GSE5305119) was available (Figure 1G, H).
Finally, we analyzed lung CIS. Lung CIS is a pre-invasive precursor lesion of lung squamous cell carcinoma (SCC), one of the two non-small cell lung cancers that we previously used to demonstrate the capability of the biomarkers to distinguish between clinical plasma samples from cancer patients and healthy subjects. We analyzed DNA methylation of the biomarker loci in lung CIS together with lung SCC and normal lung tissue samples from GEO datasets GSE10812325, GSE3927926. The advantage of the original lung CIS study (GSE10812325) is that the prospective follow-up information is available for CIS samples and thus the samples could be classified as either progressive (those later progressed into invasive cancer) or regressive (these later regressed to normal epithelium or low-grade disease). Our analysis revealed, similar to the other pre-invasive lesions, that the biomarker loci have increased DNA methylation already at the lung CIS stage (Figure 1I). More importantly, when we analyzed progressive and regressive lung CIS samples separately (Figure 1I), we found that the biomarker set is able to distinguish between the two types of premalignant lesions with high sensitivity and specificity (AUC = 0.92, Figure 1J). The majority of the regressive lung CIS samples on the MDS plots cluster close to normal lung controls while all progressive lung CIS samples are scattered among lung SCC samples (Figure 1K). Even when lung SCC samples are sub-grouped into the individual cancer progression stages (I-III) there is no increase in DNA methylation with the stage (Figure 1I). Together, these results show that the gain of DNA methylation of the biomarker loci is an early epigenetic event during human carcinogenesis.
The data presented here show that DNA methylation of the biomarker loci is fundamentally changed early during the malignant progression since it is already observed in precancerous lesions. The data from lung CIS further show that the DNA methylation level of the biomarkers can differentiate between potentially malignant and benign CIS. Together, these findings indicate that the biomarkers are capable, from the qualitative point of view, to detect cancer at its earliest stages. However, the detection of cancer-specific DNA methylation in blood or other body fluids is quantitative in nature and depends on the tumor size and its propensity to shed DNA into bloodstream; e.g., our previous report14 shows that the DNA methylation signal from this biomarker set in cfDNA samples depends on the NSCLC tumor size. Later disease stages are thus relatively easy to detect since larger tumors of later cancer stages shed a large amount of DNA into bloodstream resulting in high DNA methylation signal. In order to detect the early cancer stages as well, sensitive detection techniques and especially sample processing leading to minimal background DNA methylation signal will be profound to distinguish cancer from healthy samples. This report shows that the DNA methylation change of the biomarker loci is already present to its full extent in the earliest cancer stages. Thus, the combination of the sensitive detection and the timing of the release of enough tumor DNA into blood or other body fluids are the factors that will set the limit of the biomarkers to detect cancer early.
In conclusion, the biomarker loci have the potential to detect malignant disease at its earliest stage and the only limitation to the use of the biomarkers to detect cancer from liquid biopsies is the timing when the tumors start to release enough DNA into bloodstream.
Illumina HumanMethylation450 DNA methylation data used in the presented study can be downloaded from the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo/; Accession numbers: GSE60185, GSE66313, GSE53051, GSE58999, GSE48684, GSE77954, GSE72872, GSE81334, GSE108123 and GSE39279).
Views | Downloads | |
---|---|---|
F1000Research | - | - |
PubMed Central
Data from PMC are received and updated monthly.
|
- | - |
Is the work clearly and accurately presented and does it cite the current literature?
Yes
Is the study design appropriate and is the work technically sound?
Yes
Are sufficient details of methods and analysis provided to allow replication by others?
Yes
If applicable, is the statistical analysis and its interpretation appropriate?
Yes
Are all the source data underlying the results available to ensure full reproducibility?
Yes
Are the conclusions drawn adequately supported by the results?
Yes
Competing Interests: No competing interests were disclosed.
Reviewer Expertise: Epigenetics.
Is the work clearly and accurately presented and does it cite the current literature?
Yes
Is the study design appropriate and is the work technically sound?
Yes
Are sufficient details of methods and analysis provided to allow replication by others?
Yes
If applicable, is the statistical analysis and its interpretation appropriate?
Yes
Are all the source data underlying the results available to ensure full reproducibility?
Yes
Are the conclusions drawn adequately supported by the results?
Yes
Competing Interests: No competing interests were disclosed.
Alongside their report, reviewers assign a status to the article:
Invited Reviewers | ||
---|---|---|
1 | 2 | |
Version 2 (revision) 14 Feb 20 |
||
Version 1 16 Dec 19 |
read | read |
Provide sufficient details of any financial or non-financial competing interests to enable users to assess whether your comments might lead a reasonable person to question your impartiality. Consider the following examples, but note that this is not an exhaustive list:
Sign up for content alerts and receive a weekly or monthly email with all newly published articles
Already registered? Sign in
The email address should be the one you originally registered with F1000.
You registered with F1000 via Google, so we cannot reset your password.
To sign in, please click here.
If you still need help with your Google account password, please click here.
You registered with F1000 via Facebook, so we cannot reset your password.
To sign in, please click here.
If you still need help with your Facebook account password, please click here.
If your email address is registered with us, we will email you instructions to reset your password.
If you think you should have received this email but it has not arrived, please check your spam filters and/or contact for further assistance.
Comments on this article Comments (0)